AN ASSESSMENT OF THE !R1SKS OF
STRATOSPHERIC MODIFICATION
Volume IV: Appendix A
Review of the Literature Relevant to
Assessing the Role of Ultraviolet^B Radiation
in Cutaneous Malignant Melanoma
Submission to the
Science Advisory,-Board
U.S. Environmental Protection Agency
By
Office of Air and Radiation
U.S. Environmental Protection Agency
October 1986
Comments should be addressed to:
John S.
U.S. Environmental; Protection Agency > PM 22l
401 M Street, S.W,
Washington, D.C. 20460
USA
The following report is being submitted to the Science. Advisory Board and
to the Public for review and comment. Until the Science Advisory Board
review has been completed and the document is. revised, this assessment
does not represent EPA's official position on the risks, associated with
Stratospheric Modification. This report has been written as part of the
activities of the EPA's congressionally-established Science Advisory Board,
a public group providing extramural advice on scientific issues. The Board
is structured to provide a balanced independent expert assessment of scientific
issues it reviews, and hence, the contents of this report do not necessarily
represent the views and policies of the EPA nor of other agencies in the
Executive Branch of the Federal Government. Until the final report is
available, EPA requests that none of the information contained in this
draft be cited or quoted. Written comments should be sent to: John S. Hoffman
at the EPA by November 14, 1986.
-------
ACKNOWLEDGEMENTS
Many people contributed to this document. Principal authors are:
Ralph Buncher. University of Cincinnati Medical Center
Institute for Environmental Health, Cinciannati, OH
Edward DeFabo. Department of Dermatology The George
Washington University School of Medicine, Washington, D.C.
Sara Foster. ICF Incorporated, Washington, D.C.
John Hoffman. Office of Air and Radiation, U.S. Environmental
Protection Agency, Washington, D.C. (Project Director)
Kathleen Knox. Statistical Policy Branch. Office of Policy,
Planning and Evaluation, U.S. Environmental Protection Agency,
Washington, D.C.
Patsy Lill. Department of Pathology, University of South
Carolina Medical School, Columbia, SC
Janice Longstreth. ICF Incorporated, Washington, D.C. (Project
Manager)
Hugh Pitcher. Office of Policy Analysis, Office of Policy,
Planning and Evaluation, U.S. Environmental Protection Agency,
Washington, D.C.
Audrey Saftlas. ICF Incorporated and The Johns Hopkins University,
Washington, D.C.
Bill Ward. ICF Incorporated, Washington, D.C.
David Warshawsky. University of Cincinnati Medical Center, Institute
for Environmental Health, Cincinnati, OH
The following individuals reviewed earlier drafts of this document and
submitted valuable comments:
Dr. Edward A. Emmett
Department of Environmental Health Sciences
Division of Occupatinal Medicine
The Johns Hopkins School of Hygiene and Public Health
3100 Wyman Park drive, Bldg. No. 6
Baltimore, MD 21211
* * DRAFT FINAL
-------
Dr. Aparna Koppikar
Carcinogen Assessment Group
U.S. Environmental Protection Agency
401 M Street
Washington, B.C.
Dr. David Elder
The Pigmented Lesion Group
University of Pennsylvania
3600 Spruce Street
Philadelphia, PA 19104
Dr. Victoria Hitchins
Radiation Biology Branch
Office of Science and Technology
Center for Devices and Radiological Health (HFZ-114)
Food and Drug Administration
Rockville, MD 20857
Dr. Robert Stern
Department of Dermatology
Harvard Universtiy
Beth Israel Hospital
330 Brookline Avenue
Boston, MA 02215
Dr. Justin McCormick
Carcinogenesis Laboratory
Fee Hall
Michigan State University
East Lansing, MI 48824
* DRAFT FINAL
-------
TABLE OF CONTENTS
SECTION I:
Introduction:
Chapter 1:
Chapter 2:
Chapter 3:
SECTION II:
Chapter 4:
Chapter 5:
Chapter 6:
Chapter 7:
Chapter 8:
Chapter 9:
Chapter 10:
Chapter 11:
Chapter 12:
Chapter 13:
Chapter 14:
SECTION III
Chapter 15:
Chapter 16:
Chapter 17:
INTRODUCTORY MATERIAL
Why Modification of the Ozone Layer is an Issue
Goals and Approach of this Risk Assessment
Solar Radiation and Its Potential Biological Effectiveness
Background Information on Cutaneous Malignant Melanoma
REVIEW OF EPIDEMIOLOGIC INFORMATION
Time-Related Factors in the Incidence and Mortality:
Age, Period, and Birth Cohort Effects
Variations in the Anatomical Distribution
Geographic Distribution
Studies of Migrants
Correlations with Measures of Intermittent or Severe Sun
Exposure
Correlations with Measures of Cumulative Sun Exposure
Correlations With Degree of Skin Pigmentation
Correlations with Socio-Economic Status and Occupational Factors
Other Factors
Predisposing Conditions/Lesions for Melanoma
A Comparison of Melanoma and Nonmelanoma Skin Tumors
REVIEW OF EXPERIMENTAL EVIDENCE
Adverse Effects of Solar Radiation: Evidence From Cellular/
Molecular Studies
UV Radiation Can Cause Skin Cancer in Animals
Effects of Ultraviolet Radiation on the Immune Response and Its
Relationship to Melanomas
DRAFT FINAL
-------
TABLE OF CONTENTS
(Continued)
DOSE-RESPONSE RELATIONSHIPS AND CONCLUSIONS
Chapter 18: Plausible Dose-Response Relationships
Chapter 19: Conclusions
Appendix A: Preparation of the Document
Appendix B: Review of Critical Epidemiological Studies
* * * DRAFT FINAL *
-------
INTRODUCTION
WHY MODIFICATION OF THE OZONE LAYER IS AN ISSUE
In the early 1970's, researchers hypothesized that proposed fleets of
supersonic transports could unintentionally deposit nitrogen oxides directly
into the stratosphere, depleting the ozone layer and allowing more
ultraviolet-B radiation (UV-B or UVB) to reach earth's surface (Crutzen 1970;
Johnston 1971). Because of long known effects of UV-B on biological
organisms, this possibility led to immediate concern about the possible
effects of ozone depletion on health, ecological, agricultural and aquatic
systems.
Depletion of the ozone layer is a possibility because stratospheric ozone
levels depend on physical and chemical processes that are constantly producing
and destroying ozone molecules. In the stratosphere, high energy ultraviolet
radiation that does not reach the earth's surface photodissociates bimolecular
oxygen (0 ) . The single oxygen atom that results chemically recombines with
09 to form ozone (0 ). Ozone does not accumulate endlessly. Some ozone
is lost when it is physically transported to the troposphere, where it is more
easily destroyed. Other ozone is lost when 0. combines with single oxygen
atoms to form 0 or in other chemical processes. Generally the various
physical and chemical processes that produce and remove ozone in the
stratosphere are in approximate balance, although shifts in climate and solar
activity do cause seasonal and yearly variations around long-term mean ozone
values.
OZONE ABUNDANCE AND ULTRAVIOLET RADIATION
The ozone layer acts as a shield for the earth, protecting its surface
from all UVC radiation (UV below 290 nanometers) and some but not all of the
.UVB (UVB in the range of 295 to 320 nanometers). Figure 1 (adapted from NAS,
1982), shows how the ozone layer blocks radiation differently for various
wavelengths of UV-B.
The quantity of ultraviolet radiation reaching the earth's surface varies
with location because ozone density and the zenith angle describing the path
of solar photons to the earth varies with latitude, creating a strong gradient
for UVB radiation. For example, models predict that the yearly dose of
radiation from 295 to 299 nanometers at Nairobi, just south of the equator is
5.2 times the energy received at Washington, D.C. (latitude = 38°N) and 34.5
times the radiation at Oslo (latitude = 60°N). Because ozone does not
effectively filter other solar radiation, latitudinal variations in UVA and
visible light are much smaller as shown in Figure 2. In addition to
latitude, clouds can reduce the UVB radiation reaching the earth's surface and
altitude can increase it. UVB also varies with time of day. If the ozone
layer depletes, ultraviolet radiation would increase everywhere on the earth's
surface, with the largest percentage increases occurring at 290 to 300
nanometers and the smallest at 310 to 320 nanometers.
DRAFT FINAL * »
-------
-11-
FIGURE 1
UVR-WITH AND WITHOUT THE OZONE LAYER
RELATIVE
SUNLIGHT
INTENSITY
O.OO01
Elfactivenasa of Olona Layar
in aioeking UV
UV raaching ground with no Ozone Layar
' UV faaching ground with Ozona Layar
-UV-8-
290
Source: Adopted from NAS (1982).
30O 3 !O. 320
WAVELENGTH (nm)
This figure shows that the effectiveness of the ozone layer in blocking UV is
poor at 320 nm and improves to complete effectiveness at 290 nm.
CONCERN ABOUT CHLOROFLUOROCARBONS
In the early 1970's (Lovelock, 1973) discovered that chlorofluorocarbons
(CFCs) were not destroyed in the lower atmosphere (the troposphere). Instead,
they were observed to accumulate, so that concentrations increase with time.
Chlorofluorocarbons are used as aerosol propellants (in many countries other
than the United States), refrigerants, solvents, foam blowing agents, and in
many other industrial processes. In 1974, Molina and Rowland (1974)
hypothesized that CFCs would not accumulate endlessly, but would, in small
quantities, be transported to the stratosphere, where they would be
photodissociated by high energy radiation, releasing their chlorine. The
chlorine would then enter into a catalytic cycle that would destroy ozone
molecules, reducing mean ozone levels. Particular concern was expressed about
this threat, because CFCs have long lifetimes: 75 years for CFC-11, 110 years
for CFC-12 (NASA 1986), and 90 years for CFC-113 (NAS 1984). Because of these
long lifetimes, ozone depletion due to CFCs would be almost irreversible.
Public reaction to the potential threat of chlorofluorocarbons to the
ozone layer was swift. Between 1974 and 1978, the year in which the U.S.
banned the use of CFCs in non-essential aerosals, the quantity of CFCs used in
aerosals declined 60 percent, as consumers and producers alike abandoned CFC
propelled aerosols for hydrocarbon propelled aerosols. After the U.S. ban of
non-essential uses in 1978, CFC use in aerosols fell to negligible quantities
in this country. (Kavanaugh 1986). Some but not all nations also took action
to ban or reduce CFC use in aerosols. Until recently, the decline in aerosol
use of CFCs worldwide had offset the steady increase in non-aerosol uses
(Figure 3).
* * * DRAFT FINAL *
-------
-iii-
FIGURE 2
VARIATIONS IN SOLAR RADIATION
WITH LATITUDE
Relative Variation of Radiation at 600,
375, and 295 nm
i i i i i i i i i i i i
Latitude
This figures shows that radiation varies with latitude much more for UV-B the
UV-A or visible.
* * * DRAFT FINAL
-------
-iv-
FIGURE 3
HISTORICAL PRODUCTION OF CFCs
END USE FOR CFC 11 + CFC 12 PRODUCED
u.
o
o
_l
2
900
800 -
700 -
600 -
500 -
400 -
300. -
200 -
100
0 -
BY CMA COMPANIES
I960
1965
1970
.1973
1980
1985
This figure demonstrates that the flattening of the emission curve for CFC use
in the last decade consists of two opposing trends: a decrease in aerosol
uses and a steady increase in non-aerosol uses.
Source: Chemical Manufacturers Association (1985).
* DRAFT FINAL
-------
-v-
Throughout the 1970s, scientific research on ozone modification
intensified. Many new studies were done on various aspects of stratospheric
chemistry and dynamics, including more sophisticated analyses of the rate at
which chemical reactions took place and the types of chemical reactions that
could influence the balance of ozone in stratosphere (NAS 1976, 1979, 1982,
1984). During this period, the convention arose of using constant emissions
for comparing stratospheric depletion models and for estimating future ozone
depletion in assessments (NAS 1982). Not only did this standard assumption
allow easy intercomparison of models, it seemed like a "reasonable" assumption
of future CFG emissions, since growth did not seem to be occurring in that
period. As stratospheric science matured, estimates of ozone depletion for
constant emissions went up and down (Figure 4) because the predictions are
sensitive to the accuracy of a large number of factors that are subject to
uncertainty.
In 1983, the assumption that CFG use would not grow in the future began to
be questioned. Examination of the total trend in CFG use revealed that, while
aerosol use had been falling, the use of CFCs in non-aerosols had been growing
in a manner that seemed highly related to economic growth (Figure 5). As a
result of this "discovery," scientists began to examine what would happen to
ozone if there was sustained growth in chlorofluorocarbons. In addition, as
global measurements revealed that many other potential stratospheric
perturbants were also changing on a global scale, the atmospheric science
community began to consider the possible influence of other gases on the
stratosphere (Figure 6). Preliminary results indicated that significant
global depletion might occur with higher CFG growth and at least one model
claimed that depletion may even be non-linear, once some threshold value of
CFCs is exceeded (Prather et al. 1984).
Since 1983 a number of scientific studies have analyzed the risks of ozone
depletion, including a large international assessment (WMO 1986). That
assessment examined the state of knowledge about the upper atmosphere and
concluded that with sustained growth in CFCs serious risk of ozone depletion
still exists. Recent modeling efforts by Isaksen indicate that stratospheric
ozone depletion is a real risk (Figure 7). This time-dependent model show
significant depletion at different altitudes. Furthermore the recent
discovery that ozone has been depleted over 50% in the last six years in the
Antarctic region raises new concerns about the vulnerability of the ozone
layer (Farman et al. 1985). This has spurred intensive scientific efforts to
ascertain the causes and potential implications of this unexpected
phenomenon. Efforts are being made, worldwide, to assess the risks posed by
various substances to the stratosphere and the health and welfare risks that
stratospheric modification might pose. This report assesses the risk that
stratospheric modification will cause an increase in the severity or incidence
of melanoma.
* * * DRAFT FINAL
-------
-VI-
PREVIOUS REVIEWS OF THE RELATIONSHIP OF MELANOMA, UVB AND
OZONE DEPLETION
The relationship between skin cancer and exposure to solar radiation has
been a subject of interest for many years. The information relevant to
assessing the role of UV radiation as a causal factor in skin cancer in man
has been the subject of many reviews (Urbach 1969; CIAP 1975; Scott and Straf,
1977; NAS 1979; NAS 1982) and it has generally been accepted that the
likelihood of basal and squamous cell cancers increases with accumulated UV
exposures (Fears at al. 1977; Fitzpatrick, 1986).
The relationship of cutaneous malignant melanoma to solar radiation
exposure has been considered to be less clear. A latitude gradient for CMM
was reported as early as 1975 (Scott and Straf, 1977), however, conclusions
about the relationship of CMM to solar exposure have tended to ascribe
different degrees of certainty on the relationship from review to review.
Thus, in 1979, the NAS concluded "...UVB exposure [however]...the
dose-response relationship between skin melanoma and UVB appears to be more
complicated than that observed for non-melanomas of the skin" (NAS, 1979).
By 1982, the NAS concluded that "Since 1976, the case for an association
between UVB and melanoma has been weakned rather than strengthened by the
results of additional clinical pathological and epidemiological studies.
Furthermore, (with the exception of a single animal), it has not been possible
to use UVB alone to.induce melanomas in experimental animals." (Nas, 1982).
Even within the same review, there has been uncertainty. For example, NAS
(1982) stated:
"The incidence of skin melanoma appears to depend on
latitude, an indication that sunlight is a contributing
factor. Circumstantial evidence such as occupational
differences and location of cancers on the body suggests,
however, that exposure to sunlight is only one of several
factors."
Still more recent reviews (Fitzpatrick and Sober, 1985; and Clark et al.,
1986; Holman et al, 1986), conclude that the relation between melanoma and
solar radiation is different from that observed for non-melanoma in that the
exposure parameter of concern is probably related to intense intermittent
exposures and there is a greater impact of precursor lesions.
Since the NAS review in 1982, a wealth of new information has been added
to the data base. In assessing whether or not a decrease in column ozone with
the resultant increase in UVB will be associated with an increase in cutaneous
melanoma, this review draws on the past literature to examine what was the
data base which led to such diverse conclusions as those presented in the NAS
reports and adds to the examination an analysis of the new information and its
impact on assessing the role of UVB in melanoma development.
* * * DRAFT FINAL
-------
-vii-
FIGURE 4
DEPLETION ESTIMATES FOR ONE
ATMOSPHERIC MODEL OVER TIME*
#
UJ
O
U
2
O
U
o
o
Q
UJ
>—
<
_J
U
10
5
0
-5
- 10
- 15
-20
CALCULATED OZONE-COLUMN CHANGE TO STEADY STATE
LLNL Calculaticns
I
1974 1978 1978 1980 " 1982 1934
YEAS IN WHICH CALCULATION WAS MADE
C?C STEADY
197*
EMISSION
RATE
19SS
1988
This figure shows how "best case" depletion estimates for steady emissions at
1974 level of CFCs have shifted over time with changes in kinetics, cross
sections, and chemistry.
* Model was the Lawrence Livermore Model.
Source: Connell and Wuebbles (1986).
* * * DRAFT FINAL
-------
-viii-
FIGURE 5
U.S.: NON-REROSOL CFC
PRODUCTION VERSUS
TOTRL INDUSTRIRL OUTPUT
240
CFC 11 and CFC 12
ilon-Aerosol)
Industrial Output
100
1965
1970 1975
TE
source:
1960
985
ii999j. us I re ri9au-
KCSCKVC BOHKO 119931
correlation coefficiant=92%
This figure shows that non-aerosol CFC production has been closely associated
with aggregate industrial output, growing at approximately twice its rate.
Source: Based on Council of Economic Advisors (1984), Palmer et al. (1980),
and USITC 1968-1984.
* * * DRAFT FINAL * * *
-------
FIGURE 6
GASES INCREASING GLOBALLY THAT INFLUENCE
THE STRATOSPHERE AND GREENHOUSE EFFECT
Compounds
Rate of Increase
Stratospheric Ozone
Global Temperature
Ch lorof 1 uoroca rbons
Ha Ions (Bromine)
Methane (CH )
4
Nitrous Oxide (N 0)
2
Carbon Dioxide (CO )
2
about 5%
about 22%
about 1%
about 0.25%
about 0.6%
Depletes
Depletes
Counters Depletion: Adds Ozone
(Troposphere)
Counters Depletion in High
Chlorine Cases
Adds Ozone
(by cooling stratosphere)
1 ncreases
Increases
Increases
1 ncreases
1 ncreases
I
H-
X
I
-------
-x-
FIGURE 7
DEPLETION ESTIMATES AT VARIOUS LATITUDES
Ozone
Depletion
-14
30° N
Global & 40°N
50° N
60° N
Spring 60 N
1960
1970
1980 1990 2000 2010 2020 2030
Scenario of Growth Rates
CFCs:
CH, :
3%
0.6%
0.25%
Source: Isaksen (1986) "Ozone Perturbations Studies in a Two-Dimensional
Model with Temperature Feedback in the Stratosphere Included," UNEP
Workshop.
* * * DRAFT FINAL * * *
-------
CHAPTER 1
GOALS AND APPROACH OF THIS RISK ASSESSMENT
Under Part B of the Clean Air Act, the Administrator of EPA "shall propose
regulations for the control of any substance, practice, process, or activity
(or any combination thereof) which in his judgment may reasonably be
anticipated to affect the stratosphere, especially ozone in the stratosphere,
if such effect in the stratosphere may reasonably be anticipated to endanger
public health or welfare" (42 U.S.C. 7457). This report seeks to provide the
Administrator of EPA with a basis for making a judgment about whether
cutaneous malignant melanoma (CMM) can be reasonably anticipated to increase
in incidence or severity if there is a modification of the column density of
ozone and consequent alteration in the flux of ultraviolet-B radiation (UVB)
reaching earth's surface. This report seeks to determine what dose-response
relationships are consistent with current epidemiological evidence on melanoma
and UV, so that quantitative estimates can be made of how an alteration in the
flux of UV would change the incidence and mortality of various populations for
melanoma.
LEGAL BACKGROUND AND FRAMEWORK
In January, 1986, the EPA issued a Federal Register Notice (51 FR
1257: January 10, 1986) that announced a program of domestic and
international activities aimed at.reaching domestic and international
decisions on the need to take additional action to control stratospheric
perturbants. EPA announced that it will issue a Federal Register Notice in
May 1987, that either proposes regulations or proposes that no regulations are
needed. Under the auspices of the United Nations Environmental Programme an
international diplomatic conference is also scheduled in April 1987, at which
time an attempt will be made to achieve agreement on an international protocol
on CFCs.
As part of the preparation for those decisions, the staff of EPA is
submitting this risk assessment to the Science Advisory Board (SAB) in October
1986. The review analyzes all aspects of the stratospheric protection issue.
The SAB will review this document and after revision it will be used by the
Administrator of EPA as the scientific foundation for analysis of various risk
management options. This report is an appendix of that review that assesses
the likelihood that ozone could alter the incidence or severity of melanoma.
The Administrator will use the risk assessment document and later risk
management analyses to decide what is to be done.
THE FOCUS OF THIS REVIEW
Cutaneous malignant melanoma (CMM) is a complex disease, with a complex
etiology. For several decades, concern has been expressed about the
relationship of sunlight, UVB, and melanoma, with a large body of literature
hypothesizing a strong relationship. The etiology of melanoma, however, is
far from being completely understood. There is no irrefutable proof that
* * * DRAFT FINAL * » *
-------
1-2
solar radiation (including ultraviolet radiation) is among the causes of the
disease. Some evidence seems to support the hypothesis that solar radiation
(and UVB in particular) is one of many factors that cause melanoma. Some
evidence seems to undermine the hypothesis that solar radiation (and UVB in
particular) are among the causes of melanoma. This report weighs and balances
the evidence that melanoma incidence could increase if there is a change in
UV-B radiation, applying the standard that the Congress legally
mandated--whether "it may be reasonably anticipated" that the particular
effect will occur if there is a change in the ozone layer. In making such a
judgment, one must decide whether available information supports the view that
melanoma incidence or mortality will vary if ozone density varies, not whether
the evidence provides certainty or a complete understanding of the etiology of
the disease. Consequently, this report focuses on the weight of evidence, not
the specific etiology or mechanisms by which people get melanoma. From the
perspective of the public health decision mandated by Congress, the issue is
not whether we understand the exact mechanisms by which UV might indicate or
promote melanoma or whether we are certain or almost certain of the
relationship, but rather, whether the weight of all evidence appears to
support a judgement that there is a reasonable probability that variations in
UV-B can cause variations in melanoma incidence or severity.
APPROACH
To address whether a change in UVB may be reasonably anticipated to alter
melanoma incidence or mortality, we will try to answer three distinct
questions:
1. Does the evidence support the hypothesis that for
susceptible populations, solar radiation is a cause of
melanoma?
2. Does the evidence support the hypothesis that UVB is a
major component of solar radiation which causes
melanoma?
3. What dose-response relationships between melanoma and
UVB are consistent with the epidemiological and
experimental data?
To address these questions the literature review has been organized around
a variety of commonly held assertions or "findings" about melanoma. For
example, many studies have "found" that melanoma varies with latitude; other
studies have claimed differences between males and females; other studies have
"found" that migrants differ from natives in their likelihood of getting
melanoma. Our report first analyses such "findings," to determine the
scientific rigor of studies that support or deny each finding. Later in the
conclusions chapter we consider the implications of these findings (to the
extent they are supported) for each question.
* DRAFT FINAL * * *
-------
1-3
REPORT ORGANIZATION
This review starts with three background chapters, then goes on to eleven
epidemiological chapters, three chapters on animal, cell, and molecular
evidence and a chapter on plausible dose-response relationships. Finally, the
last chapter presents our conclusions with regard to the questions raised in
Chapter 1, i.e., is solar radiation probably one cause of CMM in sensitive
populations, is UVB a likely component of solar radiation responsible for this
disease; what dose response relationship(s) is (are) appropriate for
predicting future incidence and mortality. This chapter has discussed the
legislative framework and focus for this review. The second chapter discusses
how the energy delivered at different wavelengths of UV varies by season, time
of day, latitude and altitude and how the energy delivered relates to what is
biologically effective. The third background chapter provides information
necessary for understanding the nature and magnitude of melanoma as a disease,
along with key aspects of skin biology and photobiology useful for
interpreting other chapters.
Chapter 4 reviews information on time trends in melanoma incidence and
their relationship to age, period and birth cohort effects. Chapter 5
examines the anatomic site distribution of melanoma, analyzing potential
implications of this information for understanding the character of the solar
radiation-melanoma relationship. Chapter 6 reviews the geographic
distribution (latitude and altitude of CMM). Since latitude and altitude are
surrogate for variations in solar radiation, with UVB varying far more than
UVA or visible light, this chapter will have relevance both to the issue of
the relationship of solar radiation and melanoma and to the issue of whether
UVB is a component of sunlight that may be an agent. Chapter 7 examines the
trends in melanoma incidence among persons migrating from lower to higher
sunlight areas. Chapter 8 examines the relationship of intermittent or severe
exposure to solar radiation and melanoma incidence. Chapter 9 summarizes
information from studies which examined measures of cumulative solar
exposure. Chapter 10 examines the relationship between pigmentation and the
incidence of melanoma. This information has bearing on how to interpret other
epidemiologic evidence as well as providing information on an interesting
"natural" experiment in how the skin transmission qualities of different
populations modify their reaction to solar radiation. Chapter 11 evaluates
information on the correlation between melanoma incidence rates and
socioeconomic status and occupational factors. Chapter 12 examines the
potential relationships between variations in melanoma incidence rates in
certain populations and a variety of environmental factors including steroid
hormones and immunosuppression regimens. This information is also useful to
interpreting the data in other chapters (including ones before it,
unfortunately). Chapter 13 examines several predisposing lesions for
melanoma, providing important information to interpreting the data from
epidemiologic studies. Chapter 14 compares and contrasts the behavior of
nonmelanoma and melanoma skin cancer in order to investigate what the
differences in behavior mean relative to an etiologic agent or mechanism.
* * * DRAFT FINAL * » *
-------
1-4
The chapters on animal, cellular and molecular, and immunologic studies
then focus on solar radiation effects as determined in experimental studies.
Chapter 15 reviews information on the effect of UVB as interpreted through
studies of cells and also reviews studies of the effects UVB at the molecular
level, particularly with regard to impacts on DNA and urocanic acid. Chapter
16 reviews the information on the role of UVB as a causal agent in animal
tumors and indicates that animals can be induced to develop melanomas but that
as yet there is no animal model for melanoma induction by UVB. Chapter 17
reviews information that UVB can have immunosuppressive effects and how this
may impact on melanoma. Chapter 18 reviews the epidemiological literature to
determine what relationship(s) can best be used to determine how much increase
in melanoma can be expected if the density of column ozone is altered.
Chapter 19 summarizes our conclusions about solar radiation and UV-B as an
agent for melanoma in susceptible populations, bringing together all the
evidence, examining what conclusions can be made, and the uncertainties
remaining.
Appendix A reviews the process used to develop this report, available
databases developed (including a description of the keyword system developed),
and a bibliography of articles reviewed. Appendix B reviews, in detail,
several major epidemiologic studies considered crucial to these analyses, for
their scientific rigor and applicability as well as presenting detailed
reviews of many other epidemiologic studies. While these studies are
discussed in earlier chapters, these reviews go into much greater depth as to
methodology and design of these studies. There are also reviews of other
epidemiologic studies that were used in developing the analysis presented in
subsequent chapters.
DRAFT FINAL
-------
1-5
REFERENCES
(FOR INTRODUCTION AND CHAPTER 1)
Chemical Manufacturers Association. Production and Release of Chlorofluoro-
carbons 11 and 12 Based on Reported Data through 1984. Chemical
Manufacturers Association, Washington, D.C. (1985).
CIAP. Impacts of climatic change on the biosphere. (ed) Grobecker, A.J.,
CIAP Monograph 5 Department of Transportation, Washington, D.C. DOT TST-75-55
(1975).
Clark, W.H. Jr., Elder, D.E., and van Horn, M. The biologic forms of
malignant melanoma. Human Pathology 17:443-450 (1986).
Connell, P.S. and D.J. Wuebbles. Ozone Perturbations in the LLNL One-
Dimensional Model -- Calculated Effects of Projected Trends in CFC's, CH.,
NO and Halons Over the Next 90 Years, (DRAFT REPORT). Lawrence Livermore
National Laboratory, Livermore CA (1986) .
Council of Economic Advisers. Economic Report of the President. United
States Government Printing Office, Washington, D.C. (1984).
Crutzen, P.J. The influence of nitrogen oxides on the atmospheric ozone
content. Quart J Roy Meteorol Soc 96:320-325 (1970).
Cunnold, D.M., R.G. Prinn, R.A. Rassmussen, P.G. Simmonds, F.N. Alyea, C.A.
Cardelino, A.J. Crawford, P.J. Fraser and R.D. Rosen. The atmospheric
lifetime experiment. 3. Lifetime methodology and application to three years
of CFC13 data. J Geophys Res 88:8379-8400 (1983a)..
Cunnold, D.M., R.G. Prinn, R.A. Rassmussen, P.G. Simmonds, F.N. Alyea, C.A.
Cardelino and A.J. Crawford. The atmospheric lifetime experiment. 4. Results
for CF Cl based on three years data. J Geophys Res 88:8379-8400 (1983b).
Farman, J.C., E.G. Gardiner and J.D. Shanklin. Large losses of total ozone in
Antarctica reveal seasonal CIO /NO interaction. Nature 315:207-210 (1985).
X X
Fears, T.R., Scotto, J., Schneiderman, M.H. Mathematical models of age and
ultraviolet effects on the incidence of skin cancer among whites in the United
States. Am J Epidemiol 105:420-427 (1977).
Fitzpatrick, T.B. Ultraviolet-induced pigmentary changes: benefits and
hazards. Curr Probl Dermatol 15:25-38 (1986).
Fitzpatrick, T.B. and Sober, A.J. Sunlight and skin cancer. New Eng J Med
313:818-820 (1985).
* DRAFT FINAL » * »
-------
1-6
Holman, C.D.J., Armstrong, B.K., Heenan, P.J., Blackwell, J.B., Gumming, F.J.,
English, D.R., Holland, S., Kelsall, G.R.H., Matz, L.R., Rouse, I.L., Singh,
A., Ten Seldam, R.E.J., Watt, J.D. and Xu, Z. The causes of malignant
melanoma: Results from the West Australia Lions melanoma research project.
Rec Results in Cancer Res 102:18-37 (1986).
Isaksen. Ozone Perturbation Studies in a Two-Dimensional Model with
Temperature Feedback in the Stratosphere Included. UNEP Workshop.
Johnston, H.R. Reduction of stratospheric ozone by nitrogen oxide catalysts
from SST exhaust. Science 173:517-522 (1971).
Kavanaugh, M. Eliminating CFCs from Aerosol Uses: The U.S. Experience and
Its Applicability to Other Nations. Prepared for the U.S. Environmental
Protection Agency by ICF Incorporated (1986).
Keeling, C.D. and GMCC/NOAA. Monthly Concentrations of Carbon Dioxide at
Mauna Loa, Hawaii (1985) unpublished.
Lovelock, J.E. Halogenated hydrocarbons in and over the Atlantic. Nature
241:194-196 (1973).
Molina, M.J. and F.S. Rowland. Stratospheric sink for chlorofluoromethanes:
Chlorine atom-catalysed destruction of ozone. Nature (1974).
National Academy of Sciences (NAS). Causes and Effects of Changes in
Stratospheric Ozone: Update 1983. National Academy Press, Washington, D.C.
(1984).
National Academy of Sciences (NAS), Causes and Effects of Stratospheric Ozone
Reduction: An Update. National Academy Press, Washington, D.C. (1982).
National Academy of Sciences (NAS) Halocarbons: Effects on Stratospheric
Ozone. National Academy Press, Washington, D.C. (1976).
National Academy of Sciences (NAS). Protection Against Depletion of
Stratospheric Ozone by Chlorofluorocarbons. National Academy Press,
Washington, D.C. (1979).
National Aeronautics and Space Administration. Present State of Knowledge of
the Upper Atmosphere: Processes that Control Ozone and Other Climatically
Important Trace gases; An Assessment Report. National Aeronautics and Space
Administration (1986).
Palmer, A., W.E. Mooz, T.H. Quinn and K.A. Wolf, Economic Implications of
Regulating Chlorofluorocarbon Emissions from Nonaerosol Applications,
R-2524-EPA, The RAND Corporation, Santa Monica, CA (1980).
Prather, M.J., M.B. McElroy and S.C. Wofsy. Reductions in ozone at high
concentrations of stratospheric halogens. Nature 312:227-231 (1984).
DRAFT FINAL *
-------
1-7
Rassmussen, R.A. and M.A.K. Khalil. Atmospheric methane in the recent and
ancient atmospheres: Concentrations, trends, and interhemispheric gradient.
J Geophys Res 89:11599-11605 (1984).
Rassmussen, R.A. and M.A.K. Khalil. Atmospheric methane in the recent and
ancient atmospheres: Concentrations, trends, and interhemispheric gradient.
J of Geophys Res 89(D7): 11599-11605 (1984).
Scott, E.L. and Straf, M.L. Ultraviolet radiation as a cause of cancer, in
Origins of Human Cancer, Book A, Incidence of Cancer in Humans edited by H.H.
Hiatt, J.D. Watson and J.A. Vinsten. Cold Spring Harbor Conferences on Cell
Proliferation 4:529-546 (1977).
Stordal, F. and I.S.A. Isaksen. Ozone Perturbations Due to Increases' in N20,
CH,, and Chlorocarbons: Two-Dimensional Time Dependent Calculations.
Universitetet I Oslo, Norway (1986).
United States International Trade Commission. Report on U.S. Production of
Selected Synthetic Organic Chemicals (Including Plastics and Resin Materials),
United States International Trade Commission, Washington, D.C. For years
1968-1984.
Urbach, F (ed) The Biologic Effects of Ultraviolet Radiation: With Emphasis
on the Skin, Pergamon Press, New York, 1969.
Weiss, R.F. The temporal and spatial distribution of tropospheric nitrous
oxide. J Geophys Res 86:7185-7195 (1981).
World Meteorological Organization (WMO). Atmospheric Ozone: Assessment of
Our Understanding of the Processes Controlling Its Present Distribution and
Change. WMO, Washington, D.C. (1986).
* * * DRAFT FINAL » » *
-------
CHAPTER 2
SOLAR RADIATION AND ITS POTENTIAL BIOLOGICAL EFFECTIVENESS
In order to understand the potential of solar radiation, and ultraviolet B
(UVB) radiation in particular, as etiologic agents in the development of
cutaneous melanoma, the characteristics of solar radiation in the environment
and its effect on biological targets must be understood. This chapter
examines the physical and biological concepts relevant to understanding how
potential exposure and biologically effective doses of solar radiation might
be measured and might differ. An analysis of the role of UVB in CMM based on
the epidemological and experimental evidence presented in this document
requires an understanding of the material covered in this chapter.
The chapter is divided into three parts:
• An overview of key concepts:
an explanation of the spectrum (wavelengths and
energy),
the various units of energy used in studies and
their equivalencies,
-- the relationship of spectral, characteristics of
light to biological effectiveness (the action
spectrum), and
an explanation of the difference between exposure
dose and biologically effective dose.
• The key action spectra, with emphasis on potential
targets in the skin:
the key targets in skin, and
the relative effectiveness of different wavelengths
in inducing cell mutation, lethality and
transformation.
• Estimates of variation in solar radiation received at
the earth's surface'-:
variations in solar radiation with different times,
locations, and conditions on the earth, and
variations expected with ozone depletion.
"Information on variation in solar radiation by wavelength comes from a model
developed using NASA satellite data, detailed information for which is given
on Page 2-16 and Appendix C.
* * * DRAFT FINAL » * *
-------
2-2
AN OVERVIEW OF BASIC CONCEPTS OF SOLAR RADIATION
AND BIOLOGICAL EFFECTIVENESS
The sun produces energy by the process of nuclear fusion. Energy is
transferred from the sun to the earth by radiative processes in the form of
electromagnetic energy. Electromagnetic energy can be divided into a
spectrum, in which photons that transfer energy have both a wavelength and
energy level. Figure 2-1 shows the electromagentic spectrum; it is divided
into ultraviolet, visible and infrared regions. Ultraviolet radiation (UVR)
has been further divided into three parts: (JVC, UVB, and UVA. With time, the
definition of UVB has been shifting; in this analysis we will adopt the
convention that UVC is all UV radiation below 295 nm, that UVB is radiation
between 295 nm and 320 nm, and that UVA is radiation between 320 nm and 400
nm. (Early researchers used 280 nm to 320 nm as UVB; however the absorption
of all radiation below 295 nm by the ozone layer makes the definition (adopted
by many later researchers) of 295-320 nm as UVB much more biologically useful
as a cutoff point.)
The quantity of ultraviolet radiation can be measured in a variety of units.
Table 2-1 provides a table of units used and their definitions.
Most of the energy received by the earth is in the visible part of the
electromagnetic spectrum. Figure 2-2 shows the energy received by the earth
above the atmosphere as compared to the amount received at the surface.
Clearly some filter acts to limit the amount of radiation below 320 nanometers
that reaches the 'earth's surface; that filter is primarily the ozone layer.
The ozone layer is and will continue to be, even if ozone depletion occurs, an
effective shield for UVC, preventing almost all of it from reaching the
earth's surface.
Biological Effectiveness of Different Wavelengths
Some molecules have the capacity to absorb ultraviolet radiation and in so
doing undergo changes which manifest as adverse biological effects. For any
given molecule the probability of photon absorption will vary with the
wavelength of radiation, thus the relative effectiveness of radiation for
producing a specified biological response is strongly dependent upon its
wavelength. For example, radiation at 295 nm may be ten, a hundred, or even a
thousand times more effective in producing a given effect on a target molecule
than energy at 315 nm. In contrast, some wavelengths of radiation are totally
ineffective at producing given biological effects. Figure 2-3 shows the
relative effectiveness of three wavelengths in inducing pyrimidine dimers in
DNA (that is, in facilitating the covalent bonding of two adjacent pyrimidine
molecules) as compared to their effectiveneess in inducing cellular
transformation.
A plot of the relative effectiveness of different wavelengths of UVR in
inducing a given endpoint is termed the action spectrum of that endpoint. It
DRAFT FINAL
-------
2-3
x-rays and gamma rays
Infrared and radio
100
200
300
400
50O
6OO
700
800
9OO
1.OOO
Wavelength in Nanometers
FIGURE 2-1
THE ELECTROMAGNETIC SPECTRUM
* * DRAFT FINAL * * *
-------
2-4
TABLE 2-1
UNITS COMMONLY USED TO IDENTIFY QUANTITIES OF RADIATION
Quantity Commonly Used Units Symbol
Radiant energy joule 'J
kilowatt-hour kWh
.Radiant density joule per cubic meter J/m3
Radiant flux watt W
joule per second J/s
Radiant flux density watt per square centimeter W/cm2
at a surface joule per second per J/s/cm2
Radiant existance square centimeter
Source: Handbook of Chemistry and Physics, 1978.
* * * DRAFT FINAL
-------
2-5
no 200
300 400 500 600 TOO 8OO
Nanometers
FIGURE 2-2
COMPARISON BY WAVELENGTH OF THE RADIATION RECEIVED AT
THE EARTH'S ATMOSPHERE (~ KM) AND AT ITS SURFACE (0 KM)
Source: Heicklen (1976).
* * * DRAFT FINAL * *
-------
2-6
30 60 90
60 120 180
Exposure, J/m2
4000 8000 12,000
FIGURE 2-3
EFFECTIVENESS OF 297, 302 AND 313+ NM UVR
AT INDUCING PYRIMIDINE DIMERS ( d i AND TRANSFORMATION .(•)
Source: Adapted from Doniger et al. 1981.
* - * DRAFT FINAL * * *
-------
2-7
provides a clear picture of the relative efficiency of various wavelengths of
radiation in producing a specified effect in a molecule, cell, organ, or
entire organism. The action spectrum plots the dose of radiant exposure
necessary to produce an effect versus the wavelength at which that dose
occurs; it will differ for different cells or organs because radiation
absorption is determined by the chemical composition of the photoreceptors in
the exposed tissue. The presence of absorbing molecules or chromophores
determines the relative effectiveness of the incident radiation (Pitts et al.
1977).
Several general approaches have been used by researchers to derive action
spectra for different biological effects. One common approach relies on
monochromatic radiation sources. The relative effectiveness of individual
wavelengths (or bands of wavelengths) are determined by measuring the'dose of
radiation needed to produce a specified biological response at each
wavelength. The response generally is defined as a threshold value. For
example, the response defining the effective dose at each wavelength often is
set at the dose which causes a 50 percent or greater occurrence of some
effect, e.g., mortality, in the biological system under investigation. Thus,
the dose causing a 50 percent or greater response rate describes the relative
effectiveness of each wavelength. The most "effective" wavelength (which is
defined as that wavelength which requires the least amount of radiant energy
to cause the specified effect) is then chosen as a reference point and its
corresponding dose measurement assigned a unit value. The other dose levels
at the different wavelengths are then normalized to correspond with the
reference point. The compilation of dose levels at the various wavelengths
thus forms a spectrum over which the relative effectiveness of each wavelength
for producing a particular effect is compared.
The Difference Between Exposure Dose and Biologically Effective Dose
There are certain concepts that are critical to understanding the
dosimetry associated with exposure to solar radiation and its possible role in
cutaneous malignant melanoma. Table 2-2 presents four dose concepts designed
to address the difference between the energy which is delivered and that which
is effective. The total amount of energy across all wavelengths that an
individual could possibly receive — equivalent to the energy delivered during
the sunlit hours — is defined as the potential exposure dose. The actual
exposure dose is defined as that fraction of the energy present in the sunlit
hours which is actually delivered to skin--the potential exposure dose
modified by clothing and sun exposure habits. The potential biologically
effective dose will be defined as the amount of effective energy present in
the actual exposure dose. The potential biologically effective dose is
determined by using a weighting factor for each waveband which is based on the
action spectrum for the effect of concern. The weighting factor for a
waveband times the amount of energy delivered in the actual exposure in that
waveband is equal to the amount of effective energy delivered per waveband.
The sum of the products across all wavebands present in the actual exposure
dose is equal to the potential biologically effective dose. The actual
biologically effective dose--the amount of energy actually delivered to the
target cell or molecule — is the potential biologicially effective dose
DRAFT FINAL » - *
-------
2-8
TABLE 2-2
DOSE CONCEPTS IN PHOTOBIOLOGY
USED IN THIS DOCUMENT
Potential exposure dose -- the total amount of solar energy that an
individual could receive--equivalent to the amount
of solar energy delivered in the sunlit hours at
place(s) of work or residence.
Actual exposure dose -- the actual amount of total solar energy delivered
to the skin surface—the potential exposure dose
modified by clothing and behavior patterns
Potential biologically -- the amount of biologically effective energy-
effective dose present in the actual exposure dose--determined by
applying a weighting function based on an action
spectrum to the actual exposure dose.
Actual biologically -- the actual amount of biologically effective
effective dose energy delivered to the target in the skin—the
energy in the potential biologically effective
dose modified by the absorptions of energy by
competing chromophores, e.g., keratin and melanin,
prior to its reaching the target.
DRAFT FINAL * *
-------
2-9
modified by additional factors such as an individual's pigmentation
characteristics and the amount of other competing photoreceptors.
ACTION SPECTRA OF CONCERN IN HUMAN SKIN
UVR induces a number of effects in the skin which could influence the
induction and growth of cutaneous melanoma. Each effect has its own action
spectrum; thus there are a number of weighing functions which potentially
could be important to estimating the biologically effective dose. Presented
below are action spectra from a number of different effects that could be
important to the etiology of sunlight in CMM. Detailed discussions of these
effects are presented in subsequent chapters.
Figure 2-4 shows the average action spectrum for DNA damage proposed by
Setlow (1974); its importance lies in the role DNA damage plays in
carcinogenesis (see Chapter 15 for detailed discussion). Figure 2-5 shows the
erythema action spectrum (which is related to those wavelengths most effective
in causing sunburn) as compared to that for melanogenesis (tanning); the
weighing function derived from the erythema action spectrum would be similar
to that for the Robertson-Berger (R-B) meter. Figure 2-6 shows the action
spectrum for edema in mice as compared to the Setlow DNA damage action and
that of the R-B meter. Cole found the edema action spectrum to provide the
most appropriate weighing function for assessing the carcinogenic impact of
UVR in mice. Figures 2-7 shows the absorption spectrum-'" for urocanic acid and
the action spectrum for the induction of systemic suppression of CHS, two
possible mediators of the effect of solar radiation on the immune system
(DeFabo and Noonan 1983). A key characteristic of the action spectra
presented here is that energy in the UVB (295-320 nm) wavelengths tends to be
much more biologically active than UVA or visible light.
VARIATIONS IN AMBIENT SOLAR RADIATION
The total amount of energy from ultraviolet radiation that any target
receives in a given amount of time (the actual exposure dose) will depend, in
part, on the variations in ambient radiation that occur in the natural
environment. One effect of the earth's rotation and revolution is the
modification of potential exposure doses from place to place over time.
Ambient solar radiation incident on various sites on the earth's surface
varies significantly with the latitude, altitude, season (day of the year),
time of day, cloudiness, reflectiveness of surfaces (albedo), and atmospheric
aerosols. More important, these variations differ in intensity for different
wavelengths of UVR. Understanding these variations in UVR is critical to
predicting the effectiveness of solar radiation and its role in cutaneous
melanoma induction. In general, ambient UVB varies much more than UVA or
visible light. Because the potential biologically effective dose a person
receives will depend not only on the action spectrum of concern for potential
* An absorption spectrum only measures the amount of incident radiation
absorbed by a molecule at individual wavelengths, whereas an action spectrum
relates the amount of energy absorbed at each wavelength to an effect.
DRAFT FINAL * * *
-------
2-10
UJ
Q.
<
u
8
Q1O'
ui
10'
10'
it!
3
-------
2-11
10
10
-3
10
-4
10
-5
10
-6
10
-7
26O 280 30O 320 340 36O 380 400
1 I I I I I I I I I I I I I I
SKIN TYPES I 8 E
SKIN TYPES in a EC
I I I I I I I I I I I I I I I I
FIGURE 2-5
ACTION SPECTRA FOR ERYTHEMA AND MELANOGENESIS
Source: Gange et al. 1986.
* DRAFT FINAL
-------
2-12
100.0
10.0
o
£ 1.0
UJ
0.1
260 280 300 320 340
WAVELENGTH
FIGURE 2-6
ACTION SPECTRUM OF MOUSE EDEMA (MEE48)
AS COMPARED TO THAT OF DMA DAMAGE AND THE ROBERTSON BERGER METER
Source: Cole et al. 1986.
* * DRAFT FINAL * *
-------
2-13
10'F
100
10-1
o
o
CD
uj 10-2
P
<
UJ
-------
2-14
skin targets, but also on the distribution of energy in the various wavebands,
differential variation is very important to understanding or computing this
dose. For example, if total sunshine hours were the only consideration, that
is if photons in the UVB, UVA and visible wavelengths, were equally effective,
then outdoor workers would clearly receive a larger dose of effective energy
than indoor workers. However, if UVB is more important biologically, then the
situation could be different; since the amount of energy delivered in the UVB
range is highest at noon, an office worker who intentionally exposes himself
to sunlight on the beach from 11:00 a.m. until 2:00 p.m. on weekends may
receive greater radiation at 295 nm than an outdoor worker who eats in the
shade every noonday and wears protective clothing. Thus, in order to
calculate the actual biologically effective dose of solar radiation an
individual receives, one must consider not just sunlit hours, but also the
exposure to particular wavelengths as they are related to particular effects
of concern.
Modulators of Actual Exposure and Biologically Effective Doses
The actual biologically effective dose of solar radiation a person or
target molecule receives at a given wavelength depends on several factors
other than the potential exposure dose at a certain location. One important
factor is the amount of time a person spends outside during certain periods of
the day. The total amount of energy delivered at a given location provides an
upper bound of exposure, not the actual exposure. Few individuals are out in
the sun during all daylight hours; therefore actual exposure is correspondingly
reduced. As indicated in a subsequent section, seasonal and hourly variations
in incident solar radiation are quite significant, particularly in those
wavelengths that are most biologically effective (295-299 nm), so all hours in
the sun cannot be considered equal. People living in areas having the same
number of sunlit hours may have additional behavioral differences that modify
the amount of radiation reaching the skin (the actual exposure dose). For
example, some people wear lots of clothing, others do not. Some people wear
sunscreens, while others use sun reflectors to gather more solar radiation.
In addition to these difficulties in determining the actual exposure dose
for an individual, the transition from an actual exposure dose to a potential
biologically effective dose also involves technical problems. The
determination of the potential biologically effective dose is influenced
principally by the choice of weighting function. Using different weighting
functions (based on different action spectra) will result in differing
estimates of the potential biologically efffective'dose. Estimating a
biologically effective dose via the use of a weighting function based on the
action spectrum for erythema would probably overestimate the role of UVA if
the effect of concern was related to DNA damage because the DNA damage action
spectrum lies more in the UVB range than does that of erythema.
The actual biologically effective dose is the result of still additional
factors which modify the potential biologically effective dose. Probably the
most important such factor is the degree of pigmentation. As discussed in
detail in Chapter 3, melanin, the major pigment in the skin, absorbs UV
radiation; the more energy absorbed by melanin, the less energy there is
DRAFT FINAL * * *
-------
2-15
available to be absorbed by target molecules. The quantity of melanin in
black skin is much more effective in preventing solar radiation from reaching
targets than the lower quantity of melanin present in an olive-skinned white
person and the quantity of melanin in the skin of a fair-skinned red head is
probably the least effective. A person who is tanned receives a lower dose of
solar radiation to the basal layer than someone who has not tanned. Since
tanning can take place in different seasons, when UVB varies, the actual dose
received is not invariant with the time of year when the first sun exposure
started. Wavelengths of solar radiation vary differentially with season, so
that someone who starts getting significant sun exposure in March or April
(outdoor workers, for example) may receive relatively high quantities of UVA
and low quantities of UVB while tanning. Conversely, someone whose
significant exposure starts in June or July (some indoor workers, for example)
receive high UVB with the quantities of UVA, allowing penetration of UVB
before tanning occurs.
In addition to pigmentation there are additional properties of the skin
which can influence the transmission of UVR to the basal layer. One such
important property is the thickness of the epidermis, particularly the stratum
corneum. As discussed in detail in Chapter 3, keratin, the major protein of
the keratinocyte, absorbs strongly in the UV region. A thicker stratum
corneum thus provides more keratin to act as a UVR absorber. Other optical
properties of the skin, for example, scatter, can also vary with its
condition, and may also impact the delivery of radiation to targets within the
skin.
It is clear from the above information that an understanding of the
factors that can lead to variations in biological effective doses are
important to the interpretation of epidemological data. Information on
ambient variations in UV flux will be discussed in the remainder of this
chapter; Chapter 3 will provide information on how the tanning process and
skin pigmentation can influence the transmission of UVC through the skin to
the target cells (melanocytes) in the basal layer.
Background on the NASA Model
The radiation numbers used in this chapter are estimates taken from a
satellite-based model developed by John Frederick and George Serafino while
both were at the National Aeronautics and Space Administration (NASA).
(Detailed explanation of how the model works are found in the report by
Seraphino and Frederick (1986) which accompanies this document; information
which demonstrates its validations with the ground-based Robertson-Berger
meters is presented in the report by Pitcher 1986.
At any particular location it is difficult to predict exactly what the
levels of solar radiation have been because appropriate data to model cloud
cover are not yet available in a form which would allow the UV data in the
model to be fully exploited in estimating radiation under cloudy skies. Work
is progressing in this regard, however, and should be facilitated by the use
of data from the International Cloud Climatology project.
* * * DRAFT FINAL
-------
2-16
Latitude as a Cause of Variation
The angle between a location and the sun determines the amount of the
atmosphere that photons must pass through before reaching that part of the
earth. The lower the angle, the longer the path the photons must pass through
in the stratosphere, where ozone differentially absorbs solar radiation of
various wavelengths. Furthermore, a longer passage through the troposphere,
also provides a longer exposure to aerosols which results in greater scatter
of radiation thereby reducing delivery of the energy to the earth's surface.
Since latitude determines the average angle between a location and the sun,
the latitude of any location has a strong effect on the amount of radiation
received at that location, producing differential gradients for various
wavelengths that vary quite systematically across the earth's surface.
Figure 2-8 shows estimates from the NASA model which predicts that at 12
noon on March 21, the flux of UVB at 295-299 nm can be expected to vary from
the equator to 70°N by a factor of over 100, while UVA at 335-339 nm can be
expected to vary by a factor of 5. Clearly, the flux of UVB varies by
latitude more than that of UVA. Figure 2-9a and 2-9b present a comparison for
UVB (295-299 nm) and UVA (375-379 nm) of how variations in UV flux by latitude
change by season. It indicates that north of the equator the variation in UVB
by latitude in June is much greater than that seen for UVA, but that the rate
of change for each waveband stays relatively constant.
Another point worth noting from Figure 2-9a is that the model, estimates of
peak instanteous flux of UVB (295-299 nm) does not vary much from the equator
to 30°N to 60°N. At latitudes between the equator and 30°N, the highest peak
value of UVB flux is 2 x 10-2 mw/cm2 (occurring in June at the equator)
and the lowest peak is about 1.7 x 10-2 raw/cm2 (oc'curring in March and
June at about 14°N). From 30°N to 60°N the highest peak goes from about 1.8 x
10-2 raw/cm2 to about 0.7 x 10-2 mw/cm2 (June at 60°N)--
about a 60 percent drop. This becomes an important consideration if as some
researchers have suggested, it is the high intensity, peak exposures to solar
radiation which are important because the information present above suggests
that there will be little variability in peak exposures to UVB below 30°N and
a lot of variability from 30°N to 60°N.
Variation by Season
There are tremendous variations in the flux of solar radiation by season.
Figure 2-10 reinforces the conclusions from Figure 2-9, showing model
estimates of how the various wavelengths vary by month for Washington, B.C.
Energy at 295 nm increases by about a factor of 10 from winter (December) to
spring (March) and by about another factor of 10 by mid-summer (July). For
UVB at 305 nm, there is an increase by a factor of 2 from winter to spring and
another factor of 4 by mid-summer. Over the same period, UVA does not vary
much at all.
Variation by Time of Day
The relative amount of UVB at 295 nm also varies tremendously with time of
day. Figure 2-11 shows NASA model predictions the variation in UV flux by
* * * DRAFT FINAL * * *
-------
2-17
100.0
0)
tn
c
03
A
U
-P
C
0)
U
n
0)
10.Q _
20 40
Degrees North Latitude
60
295-299
1 I I I I I I I I I I I I I I I I I I I
nm
nm
nm
80
FIGURE 2-8
VARIATION IN UV RADIATION BY LATITUDE AS PERCENT
OF LEVELS AT THE EQUATOR ON MARCH 21 AT NOON
This graph shows the variation estimated by the NASA model for five wavebands
of ultraviolet radiation for a single day, high noon, in March. The variation
would be different for other days.
DRAFT FINAL * *
-------
2-18
20 40 60
Degrees North Latitude
30
FIGURE 2-9a
SEASONAL PATTERN OF UV BY LATITUDE
295-299 nm, CLEAR INSTANTANEOUS FLUX
This graph shows that the NASA model estimates the peak values of 295-299 nm
radiation are reached at different times of the year at different locations.
Note that the peak value falls very little from 0° (the equator) to 30°N, but
then falls precipitiously.
* * * DRAFT FINAL * * *
-------
2-19
U
•
cr
en
\
6
I I I i i i I i
20 40 60
Degrees North Latitude
80
FIGURE 2-9b
SEASONAL PATTERN OF UV BY LATITUDE
375-379 nm, CLEAR INSTANTANEOUS FLUX
This graph shows that the model estimates the peak values of UVA at 375-379 nm
are reached at different times of the year at different latitudes and that the
peak varies very little between 0° and 40°N and not much more at 70 N.
* * * DRAFT FINAL * * *
-------
2-20
(0
JS
u
0)
•H
-P
(fl
I I I I I I/M I I I I I I I I I
FIGURE 2-10
UV RADIATION BY MONTH IN WASHINGTON, D.C.
This graph shows that the NASA model estimates that radiation at 295 nm has a
much larger proportional gain than at 305 or higher wavelengths. Note that
the 335 nm line is almost coincident with the x axis, indicating low month
variation in Washington, D.C.
* * * DRAFT FINAL - * »
-------
2-21
(X 100)
FIGURE 2-11
RELATIVE CHANGE IN UV FLUX BY HOUR
IN WASHINGTON, D.C. ON JUNE 21
* * * DRAFT FINAL
-------
2-22
hour in Washington, D.C. on June 21. From 8:36 a.m. to 12:00 noon radiation
at 295 nm increases by a factor of five. A large proportion of total daily
radiation at 295 nm arrives between 11:00 a.m. and 1:00 p.m. UVA (e.g.,
335-339 nm) hardly varies with time of day, so that the quantity of radiation
delivered between 11:00 a.m. and 1:00 p.m. is a small fraction of total daily
radiation in those wavelengths. For example, the model predicts that from 5
a.m. to noon UVA (335 to 339 nm) will vary about 20-fold (data not shown)
whereas UVB (295-299 nm) will vary about 2500 fold (Figure 2-12). Thus the
time of day of exposure is a very important factor in determining the
biologically effective dose.
Variation in Ambient UV as a Function of Cloud Cover or Surface Albedo
Solar radiation varies tremendously with cloud cover (Figure 2-12'). In
this case, however, there is not much differential variation with wavelength.
Increasing cloud cover decreases solar radiation fairly equally at all
wavebands, shorter wavelengths of UVB are affected slightly more than UVA or
longer wavelengths of UVB. An increase in percentage cloud cover from 0 to 40
percent can reduce solar radiation 20 percent.
Another factor which can cause variations in ambient UV is albedo. Albedo
can be explained as that property of a surface which is defined as the ratio
of the light reflected from it to the total amount incident on it. Water,
sand, and snow have much higher albedos than asphalt and grass. Exhibit 2-13
examines the effects of changing, the albedo of UVR incident on Washington
D.C.; it indicates that although there is variation in ambient UV at different
albedos, there is little difference by wavelength.
Variation by Altitude
Because higher altitudes have less atmosphere to pass through, scattering
due to aerosols is reduced and the amount of UV incident on the surface
increase. Figure 2-14 presents results based on the hypothetical assumption
that San Francisco's altitude was raised from sea level to 4,000 meters. In
this instance, San Francisco would receive 1.35 and 1.25 times the UVB at 295
nm and 315 nm respectively and UVA at 375 nm would only increase by 1.07.
Differences in Peak Versus Cumulative Potential Exposure
Diverse locations have differential variation in peak and cumulative
exposure. Figure 2-15 demonstrates this point using El Paso as the reference
point for 3 wavelengths. For 295 nm, the difference between Minneapolis and
El Paso for cumulative exposure is about 3.5 fold. For peak exposure, the
difference is only about two fold. For a longer wavelength of UVB, 315 nm,
the variation in peak is exposures between these two cities is only about 1.7
(about 50% lower than that seen at 295 nm). For cumulative exposure at 315 nm
the ratio between the cities drops to 1.3. For UVA at 375 nm the variation is
even less; 68:100 for cumulative and 100:100 or no variation for peak.
An important implication of this pattern of variation is that if peak
exposure were the critical factor in the etiology of a disease, the expected
variation in incidence would be less than if cumulative dose were the
* * * DRAFT FINAL * * *
-------
2-23
0)
&>
C
-------
2-24
(X 1000)
16
w
M
d
o
g
0
M
\
•n
g
12
i- i . i
0.2
i
t_ i i
0.4
albedo
0.6
315
l l i
305
295
0.8
FIGURE 2-13
CHANGES IN UV RADIATION AS THE SURFACE
ALBEDO IS INCREASED AT WASHINGTON, D.C.
This graph shows that as the albedo (that is, reflectiveness) of surfaces
increase, ambient UV increases. The lower the wavelength the greater the
relative increase.
* * * DRAFT FINAL * - *
-------
2-25
Clear *ky tot*l daily flux v«. altitude
for sit* 47 in Jvr*
295
OJ
tr>
C
fO
X!
U
0)
•H
-P
0)
i i i i I i i i i I i i i i
(X 1000)
FIGURE 2-14
RELATIVE INCREASE IN UV FLUX BY WAVELENGTH WITH CHANGES
IN ALTITUDE HOLDING LATITUDE (=SAN FRANCISCO) CONSTANT
This graph shows that higher attitudes can increase the radiation at any
location, with the proportion of change universely related to decreasing
wavelength.
* * » DRAFT FINAL * * *
-------
2-26
(X 0.01)
108
F
a
c
t
0
r
I
n
c
r
e
a
s
e
28
34
37 40 43
Degrees North Latitude
FIGURE 2-15
PERCENT CHANGE IN PEAK AND CUMULATIVE ENERGY FOR
EL PASO, SAN FRANCISCO, AND MINNEAPOLIS
This graph shows that there is a greater proportional decrease in cumulative
exposure than peak exposure, with variation decreasing from 295 nm to 375 nm
to the point that peak exposures at 375 nm do not vary with latitude.
* DRAFT FINAL
-------
2-27
important factor. Furthermore, if peak exposures were the key etiologic
factor, one would expect no variation in incidences from place to place if UVA
radiation is the biologically effective portion of the UV spectrum.
IMPLICATIONS OF VARIATIONS IN AMBIENT SOLAR RADIATION
FOR THE INTERPRETATION OF EPIDEMOLOGICAL STUDIES
Information on ambient variation in UV can have important implications in
interpreting epidemological studies. Epidemological studies can be done on
matched groups (i.e., case control studies) or ecologically. Ordinarily,
matched groups are preferred because it is believed that better control of
variables can be achieved. In ecological studies three approaches to
estimating doses have been used to attempt to evaluate the relationship
between solar radiation and melanoma: a) those in which exposure was
estimated based on consideration of predicted or measured wavelengths; b)those
in which sunlit hours at place of residence was used; or c) studies in which
latitude as a surrogate for UV exposure was used. In most case-control
studies, hours of sun exposure as assessed by questionnaire have generally
been used to estimate dose.
The different approaches have important implications for evaluating the
epidemological studies and their utility. In particular, the substitution of
hours of sun exposure for actual information on time of day and year of
exposure, creates real problems in ascertaining either the actual exposure
dose or the biologically effective dose. At any given latitude, merely
counting the hours of sunshine will not allow one to know whether a person has
received a higher exposure or biologically effective dose than another. " In
fact, to the extent that the case and the control populations differ in their
expected time/day of exposure and in the temporary pigmentation they are
likely to have at times of high UVB flux, sunshine hours of exposure might be
negatively correlated with biologically effective dose. Thus one group of
people, outdoor workers, for example, might spend more total hours in the sun,
but less time with untanned skin at periods of high UVB than a group of office
workers. Data on this point is lacking, but an understanding of variation in
ambient UV should lead to real concern about the issue.
The systematic variation of UVB and other solar radiation with latitude
also has implications for interpreting ecological studies that use latitudinal
variation as a surrogate for effective dose. If the populations being
evaluated are large enough that they show little systematic variation by
latitude then the real difference between populations at two different
latitudes would be in the actual biological effective dose. As such,
ecological studies might actually control better for exposure than matched
studies. Of course, it would not be possible without additional data to
distinguish whether the effective dose stems from peak or cumulative
exposure. And the danger exists that there could be other factors, e.g.,
dietary or cultural habits, that also vary by latitude, and thus could
confound the association of latitude (as a surrogate for dose) with CMM
incidence. Clearly in ecological studies, one would expect much variation to
go unexplained simply because the determination of actual exposures in a
DRAFT FINAL » * *
-------
2-28
population will be affected by noisiness in potential exposure variables,
e.g., altitude, cloudiness, albedo, as well as differences in behavioral
(recreation, clothing) and pigmentary characteristics in the population.
As a consequence of the systematic differential variation of solar
radiation in the environment, the possibility needs to be considered that well
designed ecological studies may provide a reasonably well controlled natural
experiment in countries where there are few apparent biases in the ways humans
with different skin colors and behavior have arranged themselves. Matched
studies that fail to consider the details of hours of exposure and rely on
small groups that differ in critical ways may actually provide weaker data
than ecological studies.
OZONE DEPLETION AND CHANGES IN SOLAR RADIATION
Clearly, the best epideimologic study and possibly the only adequate way
to estimate an accurate and precise UVB dose response relationship would be a
study in which much effort had been expended to estimate the biologically
effective dose received by both cases and controls. This would require
careful attention to details about when, where and how an individual acquired
his/her sun exposure, as well as information on tanning and clothing habits
and normal pigmentation.
Understanding the effect of depletion of the ozone layer on the peak flux
of UVB, the total potential dose of UVB and the relative amount of various
wavelengths of UVB will be critical to estimating future incidence and
mortality of UV induced malignancy. If ozone is depleted there will be an
increase in the amount of UVB that reaches the earth's surface, but not UVA or
visible light. Figure 2-16 shows NASA model predictions for how ozone
depletion would affect UV flux in Minneapolis. The figure indicates that a 10
percent depletion would result in a 50 percent increase in UVB at 295 nm, a 20
percent increase at 305 nm, and a 10 percent increase at 315 nm. At 335 nm
there is no increase.
SUMMARY
There are several critical summary points:
2.1 Ozone differentially blocks various wavelengths of UVB
and does not block UVA or visible light.
2.2 Wavelengths between 295 nm and 300 nm are generally
much more effective at inducing adverse biological
effects, e.g., mutation, transformation and cell death,
in target cells in the skin than the longer wavelengths
of UVB and even more so than radiation in the UVA.
2.3 The biologically effective dose depends on the number
of photons of appropriate energy that are actually
absorbed by target molecules. This is influenced by
* * * DRAFT FINAL » * *
-------
2-29
i I i i i i I i I I
295
10 15 20
Percent Ozone Depletion
305
315
335
30 375
FIGURE 2-16
MINNEAPOLIS: TOTAL YEARLY FLUX VS. OZONE DEPLETION
* * * DRAFT FINAL * * *
-------
2-30
2.4 The flux of radiation at 295-299 run varies more than
other UVB wavelengths and much more than UVA by
latitude, altitude, time of day, and time of year.
2.5 The differential variation between UVB and UVA is
greater for cumulative potential exposure than
potential peak exposure.
2.6 Cloudiness and albedo can cause large variations in UVA
and UVB flux but they affect all wavelengths relatively
equally.
2.7 Ozone depletion would cause the largest increases in
UVB in the 295-299 nm range and little change in UVA.
» » » DRAFT FINAL » » *
-------
CHAPTER 3
BACKGROUND INFORMATION ON
CUTANEOUS MALIGNANT MELANOMA
This chapter is designed to provide an introduction to the concepts in
dermatology, photobiology and carcinogenesis necessary to evaluate the role of
solar radiation, and in particular the wavelengths between 290 and 320 nm
(ultraviolet B; UVB), in the induction and progression of the dermatologic
cancer known as cutaneous malignant melanoma (CMM). Particular attention has
been paid to the presentation of information which may help explain variations
in the incidence or mortality of the disease among various populations.. The
first section of the chapter outlines the extent of the public health problem
presented by CMM: its incidence, and mortality and, in the U.S. It also
briefly presents information on the time trends observed in incidence and
mortality of the disease. The second section presents a summary of normal
skin biology; reviewing skin structure, component cells, and macromolecules.
This is then followed by a review of what is known about the interaction of
solar radiation with the skin and the skin's mechanisms for the reduction or
repair of solar damage. This information is necessary to an understanding
and interpretation of the data presented on racial and skin color differences
as well as that derived from cellular and molecular studies. The final
section is a discussion of the biology of CMM, including information on the
differences in biology that might help explain the behavior of the disease in
various populations. This information may help resolve some of the apparent
contradictions and complexities in the relationship of melanoma and sunlight.
In particular, it is clear that there are several forms of melanoma which have
different natural histories including different relationships to sunlight.
EXTENT OF THE PUBLIC HEALTH PROBLEM
Malignant melanoma is a form of cancer whose normal precursor cell is the
melanocyte. Most melanomas arise in the skin, although they may also arise in
other sites, e.g., the eye. In the U.S., melanoma accounts for one percent of
all cancers excluding non-melanoma skin cancer and about the same proportion
of cancer deaths. Although melanoma represents only three percent of
cutaneous neoplasms, it is responsible for 65 percent of all skin cancer
deaths (Mastrangelo et al. 1982). In 1940, the 5 year overall survival was 40
percent (Sober et al. 1979). NCI (1985), based on SEER data, found 5 year
survival rates of 60 percent (1960-1963), 68 percent (1970-1973), 78 percent
(1973 -1976) and 80 percent (1977-1982).
Melanoma incidence has been rising steadily in the U.S. In the
mid-1930's, the incidence was one per 100,000; however, by 1980 the incidence
had risen to 6 per 100,000 (Kopf et al. 1984). This equates to a probability
that in 1980, one individual in every 350 will develop malignant melanoma in
the average lifespan, and, if the current rise in incidence continues
unabated, one out of every 150 individuals born in the year 2000 and living a
full life span will develop malignant melanoma (Kopf et al. 1984). Similar
* * * DRAFT FINAL * *
-------
3-2
calculations based on incidence data from 1974 in New South Wales Australia
indicate that by the age of 74, one out of every 66 residents will develop
melanoma (Holman et al. 1980). Scotto (1985) predicted that the current trend
would produce 24,000 cases of cutaneous malignant melanoma in the U.S. in
1985, and that by 1989, this number would rise to 30,000.
THE BIOLOGY OF SKIN
CMM is one of several skin cancers, the other predominant ones are basal
cell (BCC) and squamous cell carcinomas (SCC). Comparative incidence rates
for these three tumors are given in Table 3-1. There are some interesting
differences in the apparent relationship of solar radiation to these various
skin cancers (for details see Chapter 14) which may be related to differences
in the way the various normal epidermal cell types (and/or their principal
molecular products) function in the epidermis. In order to understand and
assess the role of solar radiation (and in particular UVB) in CMM, it is
necessary to have some knowledge of the normal structure and function of the
skin and the comparative relationship of melanocytes to other epidermal cells,
in particular basal and squamous cells. It is also important to have an
understanding of the impacts of solar radiation on the skin, the skin's
response mechanisms to solar radiation, and the role played by each of the
different cell types in these processes. The next sections present a brief
review of information in these areas.
Skin structure and function: The skin is the largest organ of the body.
Weighing 3 to 4 kg which is 2 to 3 times the weight of the liver, it
constitutes 6 percent of the body weight (Fitzpatrick and Soter 1985). There
are three principal layers of skin: the epidermis, the dermis and the
panniculus adiposus (Figure 3-1). The layers vary in thickness depending on
their location. The epidermis is typically 70 to 170 ym thick (max: 1560
ym) ; the thinner epidermis is found on the head, trunk and upper limbs and
the thicker epidermis is found on the lower limbs. The epidermis on the palms
and soles is as much as ten times thicker than that on the head and trunk.
The stratum corneum generally comprises between 8 to 15 ym of the epidermis'
thickness. (Pearl 1984; Fitzpatrick and Soter 1985) The dermis is typically
between 1700-2000 ym thick (min: 600 ym; max: 3000) and the subcutaneous
layer between 4000 and 9000 ym (min: 600; max: 30,000 ym) (Fitzpatrick and
Soter 1985).
The epidermis is populated by a mixture of three cell types of different
embryonic origin and function; these are (in order of percent composition) the
keratinocyte (80 percent), the melanocyte (5-10 percent) and the Langerhans
cell (5-10 percent). The dermis is composed mainly of connective tissue
fibers. These fibers are secreted by cells called fibroblasts and are
responsible for the skin's resilience and elasticity. Many of the skin
changes associated with aging are due to the impact of solar radiation on
these dermal fibers although there are also radiation-independent, age-related
changes in the chemistry of these fibers. The subcutaneous layer, or
panniculus adiposus, is a specialized layer of connective tissue which
functions as a cushion between the bone and the epidermis and dermis. It
consists primarily of fat cells; as such, it is a reservoir for caloric energy
(Fitzpatrick and Soter 1985).
* - - DRAFT FINAL * * *
-------
3-3
TABLE 3-1
AGE ADJUSTED INCIDENCE RATES3 OF BASAL CELL CARCINOMA (BCC)
SQUAMOUS CELL CARCINOMA (SCC) AND CUTANEOUS MALIGNANT MELANOMA
(CMM) AMONG WHITE POPULATIONS IN THE UNITED STATES'
Males Females
BCCb 246.6 150.1
SCCb 65.4 23.6
CMMb 9.2 7.7
Rates per 100,000 per year
Source: Scotto and Fraumini 1982
C Source: NCI 1985 - Based on 1978 data.
DRAFT FINAL * * *
-------
3-4
r y '
» <-*^J^_^L^
Stratum com«um
Granulgr ctll lay«r°
Squamous call loytr
8o»al call loytr -J
PAPILLARY
D£RMIS
RETICULAR
OERMIS
h- SUB CUT is
Figure 3-1 Structure of the Skin
Adapted from: Fitzpatrick and Soter 1985
DRAFT FINAL
*• * •*•
-------
3-5
SKIN CELL TYPES AND THEIR MAJOR PRODUCTS
The major cell type of the epidermis is the keratinocyte.
Embryologically these cells trace their lineage back to the ectoderm which
differentiates into the neural crest, the neural plate and the epidermis at
the time of neurulation (Balinsky 1965). As indicated below, melanocytes are
derived from the neural crest so that although both melanocyte and
keratinocytes are of embryonic ectodermal original -- additional
differentiation has occurred before precursor melanoblasts are generated.
In the epidermis, keratinocytes are organized vertically by various stages
of differentiation. In its first differentiation stage, the keratinocyte is
termed a basal cell. Basal cells are organized in a basal cell layer which is
oriented along a basement membrane that demarcates the interface of the dermis
and the epidermis. There are structurally and functionally distinct
populations of basal cells; some may represent the epidermal stem cell
population whereas other may serve an anchoring function (Fitzpatrick and
Soter 1985). Those basal cells that serve as epidermal stem cells, divide to
produce daughter cells which migrate outward, differentiating into squamous
cells which lose their nucleus to finally become corneocytes in the stratum
corneum (Anderson 1983). In normal skin the transition from basal cell layer
to stratum corneum takes about two weeks. The cells then pass through the
stratum corneum and are sloughed off in an additional two weeks (Pearl 1984).
Both basal cell carcinomas and squamous cell carcinomas are of.
keratinocyte origin. It is currently believed that basal cell carcinomas are
derived from undifferentiated pluripotent stem cells; ultrastructural studies
indicate that BCC cells are very similar to primitive ectodermal cells and
that they can simulate both epidermal and adnexal element (Laerum and Iverson
1981). In contrast, squamous cell carcinomas consist of cells which are
differentiated to the point that they tend to produce keratin. Individual SCC
may show different degrees of keratinization (Laerum and Iverson 1981).
Growth rate of the two types of nonmelanoma skin tumors varies tremendously.
Generally basal cell tumors are slow growing, with a cell cycle time that has
been estimated at 82 and 217 hours in two studies. For SCC of the head, the
cell cycle time has been estimated to be between 51 and 88 hours (Laerum and
Iverson 1981).
Keratinocytes produce at least three biologically important proteins:
keratin, histidine-rich proteins, and interleukin I (IL-1, also known as
epidermal-derived thymocyte activation factor - ETAF). The major substance
produced is keratin which is the primary structural protein of the epidermis.
Keratins consist of a family of helically configured, disulfide-rich,
polypeptides which strongly absorb photons in the UVC and UVB range (Harber
and Bickers 1981); their molecular weights range from 40 to 70 kilodaltons.
In the basal layer, where keratinization begins, keratin is organized into
intracellular intermediate filaments. As cells ascend the epidermis,
synthesis of new fibers stops, and the bundles of fibers become oriented in
alignment with the surface of the skin. The keratinocytes at this point have
a thick deposit of keratin underneath a much thickened cell membrane. The
* * * DRAFT FINAL * » *
-------
3-6
final transition is to a corneocyte in the stratum corneum layer -- an
enucleated dehydrated keratinocyte whose water content drops from 70 percent
to 10 percent (Fitzpatrick and Soter 1985).
Another group of proteins produced by keratinocytes are called histidine-
rich proteins or HRPs. In keratinocytes, HRPs are sequestered with keratin
into keratohyalin granules. As keratinocytes differentiate into corneocytes,
the HRPs break down, releasing free histidine which is subsequently converted
to urocanic acid (Scott et al. 1982). The degradation of HRPs is a multistep
process involving at least three species of HRPs. The initial molecule is
about 94K molecular weight (MW); it is subsequently broken down first to
HRP-II (32K MW) then to HRP-I (15-20 K MW) (Horie 1983). Ultimately the HRPs
are degraded to histidine.
Apparently it is only the enucleated, keratinized corneocyte which has the
ability to convert the free histidine released by the HRP degradation into
urocanic acid. Experiments have localized 93 percent of epidermal histidenase
to a cell compartment which contains the smallest percentage of epidermal DNA
(Scott 1981; Scott et al. 1982). The rate of conversion of histidine to
urocanic acid is pH dependent - the further, towards the surface of the skin a
corneocyte moves the more acid its environment becomes, and the more slowly
the urocanic acid is produced.
IL1/ETAF is a protein which functions as a major mediator in both immune
and inflammatory responses. Originally thought to be a solely
macrophage-produced lymphokine, characterization of its biological properties
indicate that it is a potent inducer of lymphocyte activation and chemotaxis,
that it enhances production of acute phase proteins by the liver, increases
the number of circulating macrophages and can result in elevated temperatures
(fevers) (Daynes et al. 1986). Production of IL1/ETAF by UVR-irradiated
keratinocytes is only depressed at irradiation levels which are cytotoxic. It
has also been recently noted that human and murine stratum corneum contain
substantial amounts of IL1/ETAF (Daynes et al. 1986).
There is some evidence that different populations within the keratinocyte
lineage may have differential sensitivity to solar radiation. This conclusion
is based on the observation that fairly early after irradiation, a population
of so called "sunburn cells" appears. These cells are hyperatotic cells which
when stained with hemotoxylin and eosin, exhibit dark, pyknotic nucleic and
shrunken homogenous eosinophilic cytoplasm (Gschnait and Brenner 1982; Ley and
Applegate 1985). The pathogenesis of sunburn cell development was originally
though to involve sensitization of keratinocytes by melanin. This was based
on the work of Johnson et al. (1972) who observed that sunburn cells contain
more melanin than their neighboring keratinocytes, that vitiligenous skin
produces fewer sunburn cells with the same relative doses of UVR and that
macrophages laden with squid melanin exhibit greater sensitivty to UVR.
Subsequent work suggested that DNA might be the target molecule as an
action spectrum for sunburn cell formation was consistent with DNA as a target
(Woodcock and Magnus 1976) and autoradiographic studies showed that sunburn
cells exhibited less DNA repair activity than normal keratinocytes
* DRAFT FINAL * * *
-------
3-7
(Brenner and Gschnait 1979). In addition studies performed in the marsupial
Monodelphis domestica (which has the ability to photoreactivate pyrimidine
dimers in its epidermal DNA) indicated that pyrimidine dimers are the major
photoproduct involved in the induction of sunburn cells and hyperplasia in
this species (Ley and Applegate 1985).
The melanocyte is the second cell type in the epidermis. Figure 3-2
presents three increasingly magnified versions of the structure of the skin
which focus down on the melanocyte and its positional relationship in the
epidermis (Fitzpatrick and Soter 1985; Pearl 1984; Nordlund 1981).
Melanocytes are pigment-producing cells derived from the embryonic neural
crest which, in the case of skin, have migrated to the basal layer of the
epidermis or the matrix of hair follicles (Romsdahl and Cox 1976).
Melanocytes may also migrate to the oral and nasal cavities, the upper third
of the esophagus, the vaginal mucosa, and the eye (Briele and Das Gupta 1979;
Nordlund 1981). In the skin, melanocytes tend to reside at the interface of
the dermis and the epidermis where they interact with keratinocytes via long
extensions of their cell bodies called dendritic processes (Romsdahl and Cox
1976).
Melanocytes in the basement membrane at the interface of the dermis and
the epidermis rarely divide although there is some evidence that UV
irradiation may cause melanocyte proliferation (Pathak et. al. 1978.) One
report in the literature indicated that there is about a 10 percent reduction
in melanocyte numbers per decade of life (Snell and Bischitz 1963). This
reduction probably parallels that seen in hair follicles which is expressed as
graying of the hair (Fitzpatrick and Soter 1985).
The major function of the melanocyte is the production and distribution of
melanin to the keratinocytes. Melanocytes which migrate to the hair follicles
produce the melanin which gives hair its characteristic color, and it is
melanin which has been purified from hair which is used most often in
analytical studies (Menon et al. 1983a).
In the skin, each melanocyte distributes melanin, packaged in an organelle
termed a melanosome, to about three dozen keratinocytes. Figure 3-3 shows the
close association between a melanocyte and its keratinocytes; this grouping is
frequently referred to as an "epidermal melanin unit" (Fitzpatrick and Soter
1985).
Melanin is a pigment that absorbs light in the broad range of 250 to 1200
nm; its absorption increases steadily towards the shorter, more biologically
active wavelengths (Anderson 1983). As such, it has been suggested that
melanin's major role-in the skin is to protect against the adverse effects of
UVR (Briele and das Gupta 1979). Other hypotheses (reviewed in Morison 1985)
that have been advanced for the function of melanin include camouflage and
heat absorption.
There are qualitative and quantitative differences in the melanin present
in human skin. These differences may be important to an understanding of the
difference in melanoma incidence among different races and among Caucasians of
* * * DRAFT FINAL * * *
-------
3-8
Cornifled cells
Melanin dus
Melanosomes
Keratinocyte<^
Stratum corneum
Qranular layer
Keratlnocyte layer.
Melanocytea
Baaamant membran
Baaal call layei
EPI-
DERMIS
•Stratum eomeum
Granular cell layer
Squamoua call layer
•Baaal cell layer
^EPIDERMIS
^-PAPILLARY DERMIS
-RETICULAR DERMIS
.SUBCUTIS
Figure 3-2 Three Increasingly Magnified Versions of the
Structure of the Skin Showing the
Relationship of the Melanocyte.
Adapted from: Fitzpatrick and Soter 1985, Pearl 1984 and Nordlund 1981.
* * * DRAFT FINAL * * *
-------
3-9
Figure 3-3 The Epidermal Melanin Unit
LC= Langerhans Cell
Me = Melanocyte
Ke = Keratinocyte
Adapted from: Stanbury, J.B., Wyngaarden, J.B., and Fredrickson, D.S.: The
Metabolic Basis of Inherited Disease. 4th ed. New York, McGraw-Hill Book Co.,
1978.
DRAFT FINAL
-------
3-10
different genetic heritages. For this reason a somewhat detailed review of
melanin biology and chemistry is presented here.
In melanocytes, melanin is synthesized from tyrosine and becomes complexed
with a protein matrix (Romsdahl and Cox 1976) inside membrane-bound structures
termed melanosomes. The chromophoric backbone of the melanin complex is very
stable and can withstand attack from proteases, acids and bases (Anderson
1983).
In humans and other mammals there are two predominant forms of melanin --
eumelanin which is brown or black, and pheomelanin which is yellow or auburn
and is the pigment responsible for red hair. In normal skin pigmentation,
tyrosine is the starting point for both types of pigment (Nordlund 1981).
Tyrosinase oxidizes tyrosine to dopa (3,4-dehydroxyphenylalanine) and' dopa to
dopaquinone (Nordlund 1981; Romsdahl and Cox 1976). The dopaquinone then
undergoes a series of spontaneous oxidative changes forming a variety of
indoles and quinones which polymerize and are deposited on the lamellar
protein matrix of the melanosome. To form pheomelanin, dopaquinone is shunted
by some unknown control mechanism into a pathway in which it is covalently
bonded to cysteine; thus pheomelanin is a polymer of indole cysteine (Nordlund
1981).
The melanosomes containing these two types of melanin differ
structurally. Those with pheomelanin are round and have a protein matrix with
a "tangled" appearance whereas eumelanin-containing melanosomes are round and
have a characteristic lamellar structure (Nordlund 1981).
The two melanins have different UV and visible absortion spectra (Figure
3-4) with.the eumelanin looking much like the dopa melanin and the pheomelanin
absorption decreasing more rapidly in the 280-370 nm range. (Dopa melanin is
a synthetic melanin generally produced in the laboratory by allowing a
solution of dopa quinone to spontaneously polymerize.) Pheomelanin also has a
quite different infrared absorption spectrum showing an increased
transmittance at almost every wavelength (Figure 3-5) (Menon et al. 1983a).
The two melanins also show different behaviors when subjected to UV-
visible irradiation in vitro. Irradiation of pheomelanin produces
considerable amounts of superoxide (a free radical) under conditions in which
the irradiation of eumelanin does not. It has also been shown that while
irradiation of cells in the presence of either melanin produced cell damage,
pheomelanin was much more effective than eumelanin in producing this effect
(Menon et al. 1983b).
Extensive studies have indicated that the number of melanocytes varies
very little from one racial group to another. There are variations however,
by anatomic region, with the highest concentration of melanocytes occurring in
2
the cheek (2,310 + 150 melanocytes/mm ) and the lowest occurring in the
2
back and thigh (890 + 70 and 1,000 + 70 per mm , respectively) (Briele and
das Gupta 1979). Instead, the pigmentary difference between races is
apparently due to the quantity and quality of melanin deposited in melanosomes
and the number of melanosomes transferred to keratinocytes. Table 3-2 shows
* * * DRAFT FINAL * * *
-------
8
I
3-11
o
I
5
a
280 340 400 460 520 580 640 700
280 340 400 460 520 580 640 700
Wavelength nm
Wavelength nm
• Ultraviolet and visible spectra of black hair melanin and dopa
melanin. Black hair melanin was irradiated for 30 and 60 min. Nonir-
radiated samples of black hair melanin and dopa melanin were used as
controls. The melanins were precipitated by adding 3 M HC1 to pH I;
the precipitates were dissolved in Solune-100 to give a final concentra-
tion of 200 jig/ml : Dopa melanin: : nonirradiated black hair
meianui: : hair melanin irradiated for 30 min; : hair melanin
irradiated for 60 min.
-0.02
-0.04
- 0.06
-0.08
-0.10
- 012
- \
RED HAIR MELANIN
BLACK HAIR MELANIN
OOPA MELANIN
Ultraviolet and visible spectra of red hair melanin and dopa
melanin. Red hair melanin was irradiated for 30 and 60 min. Nonirra-
diated samples of red hair melanin and dopa melanin were used as
controls. The melanins were precipitated by adding 3 M HC1 to pH I:
the precipitates were dissolved in Soluene-100 to give a final concen-
tration of 200 jig/ml : Dopa melanin: : nonirradiated red hair
melanin; : hair melanin irradiated for 30 min; : hair melanin
irradiated for 60 min.
. First derivatives of absorption spectra of dopa melanin, black
hair melanin, and red hair melanin : Dopa melanin: : black
hair melanin; : red hair melanin.
310 350 390 430 470
Wavelength nm
510
Figure 3-4. Differences in the UV-Visible Absorption
Spectra for Black Hair Melanin (eumelanin) and Red Hair
Melanin (pheomelanin) (From Menon et at. 1983a)
DRAFT FINAL
-------
3-12
Frequency (cm-1)
4000 3600 3200 2800 2400 2000 1800 1600 1400 1200 1000 800 600 400
7 8 9 10 12 14 18 25
25
Figure 3-5. Infrared Absorption Spectra of Dopa Melanin,
Black Hair Melanin, and Red Hair Melanin. Dopa Melanin;
Black Hair Melanin; Red Hair Melanin.
(From Menon et al. 1983a)
* * * DRAFT FINAL * *
-------
3-13
the relationship between skin color melanosome size, and organization and skin
type classification. Note that the table indicates that melanosomes in
keratinocytes from lighter-colored skin are characteristically found as
aggregates whereas melanosomes in keratinocytes from dark-skinned individuals,
which are larger, generally occur singly. It also indicates a greater
tyrosinase activity in melanocytes from dark-skinned individuals, probably
because there is more melanin per melanosome (Pathak et al. 1976). It has
been suggested that the size of the individual melanosome may determine
whether they are taken up singly or as aggregates by the keratinocytes (Hu
1981).
There are instances in which melanocytes and/or their embryonic precursors
fail to reach the epidermis but come to reside in the dermis instead. In many
such instances -- as a result of unknown stimuli -- they may lose their
melanogenic properties and may acquire the ability to contribute to the
fibrous matrix of the dermis or to its neuronal network (Reed 1983; Elder et
al. 1981). In so doing they become nevus cells. The factors that govern this
change are unknown. One classification scheme considers the melanocytic nevus
to be an evolving lesion in which melanocytes ("nevus cells") proliferate in
the epidermis, drop into the dermis and undergo further maturation there
(Elder et al. 1981). Another scheme differentiates between acquired and
congenital melanocytic nevi by whether or not the melanocyte precursor goes to
the epidermis and then returns to the dermis or fails to achieve the epidermis
because of arrested development in the dermis (Reed 1983) . It has been
suggested (Holman et. al. 1983) that differences in behavior between different
histogenetic types of melanoma may be related to their having as precursor
cells different differentiation states of the melanocytes.
Langerhans cells represent the third significant cell population
present in the epidermis. It is generally accepted that Langerhans cells are
bone marrow-derived cells which are functionally and immunologically related
to the monocyte-macrophage series (Katz et al. 1979). These cells are present
in the epidermis at a concentration similar to that of melanocytes--between
460/mm2 and 1000/mm2 (Lever and Schaumburg-Lever 1979). In contrast to
melanocytes, the number of Langerhans cells does not increase with repeated
exposure to ultraviolet light (Scheibner et al. 1986); however, these cells
are very sensitive to UV light and lose their functional ability when exposed
to very low doses. Their primary function appears to be one of antigen
presentation (Stingl et al. 1978). A defect in this function is thought to be
the cause of the immunosuppressive effect of UVR noted in studies of
UV-induced cutaneous tumors (Kripke 1974) and the response of individuals to
cutaneously applied antigens (Toews et al. 1980).
Skin photobiology
The interaction of sunlight with the skin is a complex process involving
the transfer of energy from sunlight to various molecules in the various skin
layers. As shown in Figure 3-6, the solar radiation reaching the earth
contains wavelengths from about 290 to 4000 nm. This radiation is described
as UV, visible or infrared depending on the wavelength. The UV portion of the
electromagnetic spectrum covers the range from 200 to 400 nm. Solar radiation
* DRAFT FINAL
.'- JL. .A.
-------
TABLE 3-2
RELATIONSHIP AMONG SKIN COLOR, SIZE AND DISTRIBUTION PATTERN OF
MELANOSOMES AND SKIN TYPE CLASSIFICATION
Skin Color
Size of Melanosomes
Tyros inase
Activity in
Melanocytes
Distribution of
Melanosomes in
Epidermal Keratinocytes
Approx ima te
Number of
Melanosomes per Skin
Basal Keratinocyte* Type**
Heavily piymentod skin of Africans, 0.7-0.8 urn x 0.3-O.U urn
American Negro s a rut Australian
Abor ig i nes
Moderately pigmented skin of
Mongoloids (American Indians,
OricntaIs)
Moderately pigmented skin of
Caucasoids (East Indians,
I ta I ians, Egypt i ans)
Lightly pigmented skin of
Caueaso ids (fai r-skinned
Americans, British, French,
Germans, etc.)
0.5-0.7 urn x 0.3-O.U urn
0.5-0.7 urn x 0.3-O.U urn
0.4-0.6 urn x 0.3-0.4 urn
Marked Single, non-aggregated 400+35
Moderate Mixed non-aggregated
as well as aggregated
Moderate Predominantly aggregated
Weak Predominantly aggregated 100+50
VI
250+50
200+5
V
VI
1 1
1 1 1
11
After Pathak MA, et a I.: Sunlight and melanin pigmentation, in Smith KG (ed): Photochemical and PhotobioloqicaI Reviews.
New York: Plenum, 1976, pp. 211-239.
* Based on random counts of 50 keratinocytes in basal layer.
** Estimate of sensitivity to DVB erythema based upon history of sunburn and ability to tan.
-------
3-15
-ULTRAVIOLET-
-VISIBLE-
- INFRARED-
100
80
ki 60
40
20
I
uvel OVA
— TERRESTRIAL
SOLAR SPECTRUM
2OO 3OO «OO 500 6.0 TOO
WAVELENGTH , am
1OOO
3OOO
50OO
Figure 3-6 Spectrum of electromagnetic radiation that
reaches the earth's surface from the sun. Wavelengths
shorter than about 290 nm are absorbed by ozone
in the stratosphere
DRAFT FINAL
-------
3-16
shorter than about 290 is absorbed by ozone in the atmosphere and does not
reach the earth's surface. The range from 200 to 400 nm is often divided into
three categories: UVA, UVB and UVC. The UVA portion (320 to 400 nm) is longer
wavelength, less energetic UV radiation which is not strongly absorbed by
proteins and nucleic acids and does not cause erythema in normal skin at
moderate doses in the absence of photosensitizing chemicals (Anderson 1983).
This range is also called black light and near UV radiation. UVB (290-320 nm)
is that portion of the spectrum responsible for solar erythema. It is also
known as midrange UV and sunburn radiation. UVC (200 to 290 nm) is
biologically active but does not reach the earth's surface. However,
radiation at 254 nm is frequently used in experimental biology to study the
effects of UV on various systems.
About 95 percent of incident radiation penetrates the stratum corneura, the
other 5 percent is reflected by the stratum corneum. Two processes, scatter
and absorption, determine the penetration of radiation into the skin. Table
3-3 shows estimates (from two sources) of the epidermal transmission of
various UV wavelengths through Caucasian skin as well as a comparison from one
source of transmission through Black skin. Note that the relative
transmission of White to Black is about 5 fold, i.e., Black epidermis is 5
times more effective than White epidermis at preventing radiation from
reaching the basal layer. Figure 3-7 graphically shows these differences and
indicates that between 340 and 400 nm Black skin may be ten times better than
White skin at preventing radiation from reaching the basal layer (Kaidbey et
al. 1979). These authors also estimate that in Blacks, the minimal erythemal
dose (MED) is about 13 times that in fair-skinned humans. Figure 3-8
indicates that there are also great differences in the transmission properties
of Black versus White skin for wavelengths up to 800 nm (Wan et al. 1981).
Data on a parameter related to transmission --diffuse spectral absorbance --
indicates as well that "...above 1200 nm, the optics of the skin are
essentially unaffected by melanin pigment" (Anderson 1983; Jacquez et al.
1955a, b).
The amount of radiation reaching the basal cells is a function of the
thickness of the epidermis and stratum corneum as well as their content of
radiation-absorbing molecules (chromophores). Although there are a large
number of such substances in the skin, some of them as yet uncharacterized,
most of the optical absorbance within the skin is attributable to melanin,
proteins containing aromatic amino acids, urocanic acid, carotenoids (in the
stratum corneum only) and nucleic acids (Anderson 1983). Figure 3-9 shows the
ultraviolet absorption spectra of the major epidermal chromophores.
Only certain of the wavelengths present in sunlight are capable of
stimulating or altering the functional state of melanocytes. The UVA (320
-400) and UVB (290-320) wavelengths are the most active at stimulating the
production of melanin as well as being the most effective at activating
melanocyte function (Pathak 1985). Visible radiation (400-760 nm) and
infrared radiation have some ability to stimulate or induce melanin pigment
production; however, this response appears to be secondary to the damaging
effect of the heat that such light transmits to the skin. The mechanism of
the UVB effects is unknown, although one author (Pathak 1985) has suggested
* * * DRAFT FINAL * * *
-------
3-17
TABLE 3-3
ESTIMATES OF THE PERCENT EPIDERMAL TRANSMISSION a/
OF VARIOUS WAVELENGTHS OF UV-RADIATION
X, mn
Bruls et al. (1984) b/
White
Kaidbey et al.
(1979) c/
White Black
Relative
Transmission (%)
Black Compared to
White d/
290
295
300
305
310
315
320
325
330
335
340
0.003
0.014
0.045
0.074
0.084
0.111
0.123
0.134
0.143
0.152
0.160
0.083
0.171
0.270
0.348
0.407
0.453
0.488
0.515
0.537
0.550
0.562
0.011
0.025
0.041
0.056
0.068
0.078
0.086
0.092
0.097
0.100
0.103
13
15
15
16
17
17
18
18
18
18
18
a/ Adapted from Kubitschek et al. (1986).
b/ Values interpolated from Table 1 assuming a value of 67 ym for
epidermal thickness. Fluorescence radiation was excluded.
c/ Average values from Figure 3. Fluorescence radiation was not excluded.
d/ Based on values from Kaidbey et al. (1979).
* * * DRAFT FINAL
-------
3-18
790
ITS
329 390
WAVELENOTM
ITS
400 MM
Figure 3-7. Average absorption spectrum for black and white epidermis
Source: Adapted from Kaidbey et al. (1979)
* DRAFT FINAL * * *
-------
3-19
IOO -
u
c
a
C
a
I I 4 I I I I I I I I I I I I I I I I I I I I I 4 I I I I
250
400
60O
80O
Wavelength ( nm)
Figure 3-8. Comparison of measured epidermal transmittance
A, Caucasian; B, Dark Black.
Source: Adapted from Wan et al. 1981.
DRAFT FINAL * * *
-------
3-20
200 220 240 260 2SO
WAVELENGTH, NM
300 320 3;o
Figure 3-9. Ultraviolet absorption spectra of major
epidermal chromophores
DOPA-melanin, 1.5 mg% in H-O);
urocanic acid, 10"4 in H-O;
calf thymus DNA, 10 mgr0 in H2O (pH 4.5);
tryptophan (TRP), 2 x 10"4 M (pH 7);
tyrosine (TYR), 2 x 10"10 (pH 7)
(From Anderson (1983) pg 65)
DRAFT FINAL
-------
3-21
that it involves direct damage to the cell nuclei of melanocytes following
direct absorption of photons by DNA that results in DNA base photoproducts and
subsequent mitoses and proliferation of melanocytes. In contrast, the effects
of the longer wavelength UVA apparently involve the absorption of photons by
non-DNA chromophores followed by the reaction of these excited species with
molecular oxygen and the generation of reactive oxygen species that then
interact with melanin, the cell membrane or DNA thereby exerting damage
(Pathak 1985).
Skin response mechanisms: The skin has 3 response mechanisms for
dealing with solar radiation: 1) dose-reduction through the production of
increased melanin (tanning) or through keratinocyte hyperplasia, 2) damage
repair in which the cell's DNA repair mechanisms remove photoinduced damage
and 3) cell removal.
(1) Dose-reduction via tanning may be either an immediate or a delayed
response. The immediate response, sometimes referred to as immediate pigment
darkening or IPD (Pathak 1985) is not due to an increase in epidermal
melanoctyes or melanin but rather is due to changes in melanin already
present. Part of the change is evidently the oxidation of melanin to
semiquinone-like free radicals which occurs both in melanocytes and
keratinocytes. In addition, the distribution of melanosomes within the
melanocyte is altered; they move from a perinuclear position into the
dendritic processes and are then transferred into keratinocytes (Gange and
Parrish 1983). The most effective wavelengths at inducing IPD are in the 320
to 400 nm range. This process of melanocyte transfer may begin within one
minute following exposure, is maximal at about 1 hour and is gone in about 4
hours.
The delayed response is the result of the de novo synthesis of new
melanosomes followed by their transport to keratinocytes to produce a skin
darkening about 10 hours post exposure. This process may continue several
days with the maximal skin darkening (tan) being achieved in about a week.
The most active wavelengths at inducing delayed tanning are in the UVB range;
however, UVC, UVA and visible light can also cause delayed tanning. Along
with the increased production of melanosomes, there is an increase in the
number of active melanocytes, some of which are probably due to increased
mitoses and others the result of the recruitment of dormant melanocytes. Mice
repeatedly irradiated with UVB showed a four to sixfold increase in their
melanocyte populations in 11 days; a smaller increase in non-irradiated sites
was also observed, suggesting that a systemic growth factor might be
involved. With time the tan wears off as the keratinocytes containing the
extra pigment are sloughed off.
There is a wide variation in the tanning as well as the burning (erythema)
response of individuals. This has been characterized through the
classification of skin responses into six skin types as outlined below (Pathak
1985):
» * DRAFT FINAL » -
-------
3-22
Skin type I: These individuals are very sensitive, they tan
little or not at all even with repeated exposure and may
develop a moderate to severe sunburn reaction after exposure
for an hour to solar radiation in the summer months. They
are very fair, have red or blond hair, blue eyes, often have
freckled skin and have a low minimal erythemal dose (MED) -
in the range of 15 to 30 mJ/cm2. They do not show any IPD
response. «
Skin type II. These individuals are also very sensitive and
burn frequently. They do not tan well, but upon repeated
exposures may acquire a slight tan. They have a low MED in
the range of 20-35 mJ/cm2, and may show a weak IPD
response.
Skin type III. These individuals generally burn first then
tan later. They exhibit an IPD reaction upon exposure to
light, and can generally acquire an average to a good tan
with 2 or 3 moderate exposures.
Skin type IV. These individuals generally have dark eyes
and hair, and normally have lightly tanned skin. They burn
minimally and have a strong IPD reaction. In the summer
months, the facial skin color of these individuals changes
from light brown to olive or medium brown. This group
includes pigmented Caucasoids, American Indians, Orientals
and people from the Mediterranean region.
Skin type V. These individuals generallly acquire a deep
tan and demonstrate an intense IPD reaction . Their eyes and
hair are deep brown or black. This group includes Mexicans,
Indians, Egyptians, Malaysians, Puerto Ricans and other
Spanish speaking peoples.
Skin type VI. These individuals are darkly pigmented
without exposure (e.g., American and African Blacks and
Australian aborigines) and become even more deeply pigmented
upon exposure.
Gange et al. (1986) have compared the action spectra for tanning to that
for erythema for skin types I through IV. Figure 3-10 presents the
information provided by this group. The shorter UV wavelengths (250-296) were
more erythemogenic than melanogenic. Wavelengths of 296 to about 330 are
equally erythmogenic and melanogenic to all skin types. In contrast at
wavelengths above 330 there is a divergence in response, for instance,
individuals who tan well have a melanogenic dose at 365 nm which is
approximately one quarter of that required to induce erythema.
These authors also examined the photoprotective effects of visually
identical UVA and UVB tans induced in the same individuals (Gange et al.
1986). They found that UVB tans were much more protective in that a UVB tan
DRAFT FINAL
-------
3-23
10
,-*
260 28O 300 320 340 S6O 380 400
10
M2/J
10
10
10
-9
I I I I I I I I I I
SKIN TYPES I 8 E
SKIN TYPES man
I I I I I
Figure 3-10. Action spectra for Erythema and Pigmentation
Source: Gange et al. 1986.
DRAFT FINAL
-------
3-24
could protect the skin from up to 3 times the normal MED whereas a UVA tan
could not.
An additional way by which the epidermis may reduce the dose delivered to
the basal layer is by increased production of keratinocytes and a subsequent
thickening of the epidermis. Studies in humans have shown that after a single
exposure to UVB, there is a sustained increase in epidermal mitoses which
leads to increases in thickness of the epidermis and stratum corneum of
between 1.5 and 3 fold over the course of 1 to 3 weeks. The dose first
induces a transient depression in macromolecular synthesis in which DNA, RNA
and protein synthesis are markedly reduced and then elevated. The elevation
of synthesis is maximal in 24 to 48 hours post-irradiation but may continue
for as long as a week. UVB and UVC are the most effective at inducing this
response but UVA will also induce it. This additional skin thickness'offers
some measure of protection to individuals with a poor tanning ability because
the disulfide rich keratin synthesized by the keratinocytes absorbs photons in
the UVC and UVB ranges. It is likely that, in lightly pigmented individuals,
this skin thickening is the most important photoprotective or dose-reducing
response whereas with dark-skinned individuals, tanning is the more important
response (Gange and Parrish 1983).
(2) The second mechanism by which the skin can respond to the damage due
to solar radiation is DNA repair. (A brief discussion of this subject is
presented here; it is covered in much greater detail in Chapter 15.) The
direct or indirect DNA damage inflicted on cells by UVR can be very
detrimental. The most studied alteration in DNA structure by biologically
relevant doses of UV is the cyclobutyl pyrimidine dimer (Spikes 1983). These
dimers are formed between adjacent pyrimidines on the same DNA strand and
their presence renders the phosphodiester bond joining the deoxyribose
moieties resistant to nuclease digestion. The cell has three repair processes
by which to correct damage to its DNA: photoreactivation, excision repair and
post-replication gap repair.
Photoreactivation involves UVA-dependent, enzyme-mediated repair of
pyrimidine dimers in which the enzyme binds to the dimer forming an
enzyme-substrate complex which absorbs photons of UVA light and the dimer is
then monomerized (Spikes 1983). Photoreactivation is an error free,
nonmutagenic repair pathway which offers a number of advantages to the cell;
it uses an outside energy source (UVA photons), there is no incision into the
DNA phosphodiester backbone (and therefore no risk of DNA degradation) and
there is no polymerization (and thus no chance for the introduction of coding
errors) (Spikes 1983).
Another efficient DNA repair process which works more slowly but may
repairs not only dimers but other kinds of damage, e.g., bulky carcinogens
linked to DNA, is called excision repair. In the case of dimers, excision
repair works via the activity of three enzymes, an endonuclease specific for
dimers, a polymerase and then a ligase. The endonuclease attacks the DNA at
or adjacent to the dimer and introduces a nick in the DNA. The DNA polymerase
removes the damaged DNA segments while utilizing the opposite strand as a
template for new synthesis, and finally a ligase joins the newly synthesized
DRAFT FINAL
-------
3-25
DNA to the preexisting strand. Other kinds of damage such as bulky
carcinogens linked to DNA require different endonucleases but the activities
of the polymerase and the ligase remain the same (Spikes 1983).
There is a final repair process which may be available to the cell should
the previous two mechanisms fail to repair a pyrimidine dimer. Termed
post-replication gap repair, this mechanism is fairly well understood in
bacteria such as E coli where it is invoked when, following DNA replication,
there are gaps left in the DNA opposite the dimer. (The DNA polymerase which
faithfully transcribes one daughter strand to produce a new strand cannot
interpret a dimer and consequently leaves a gap in the newly synthesized
DNA.) Not much is known about post replication repair in mammals, however.
If mammalian cells use a mechanism similar to that observed in E. coli then
repair would proceed via a DNA recombinational mechanism which provides a good
copy of the DNA needed to repair the damage; using this information, the cell
can repair the damage with an excision repair mechanism (Spikes 1983).
However, this mechanism is fairly error prone.
There are three possible outcomes as a result of the combined efforts of
these repair systems: (1) the excision repair or photoreactivation process
has worked properly such that all photolesions are removed from the cell
before DNA replication takes place so that the cell suffers no mutations or
other UV-induced DNA changes, (2) the photlesions in the DNA are not removed
prior to DNA replications such that mutations occur or (3) misrepair or lack
of repair of the DNA results in cell death. From the perspective of melanoma
development it is the second category which is probably most important because
a non-lethal mutation could cause the cell or its progeny to take on the
characteristics of a cancer cell.
(3) The third protective mechanism available to the skin is the removal
of potentially harmful cells via either immune surveillance or, in the case of
keratinocytes, programmed senescence and the shedding of dead cells from the
stratum corneum. Immune surveillance may be compromised in chronically
UV-irradiated individuals because UVB experimentally induces a systemic
immunosuppression to UV-induced tumor antigens (Kripke 1984). The skin is
also protected in part from the development of mutant cells because the normal
life history of keratinocytes is to move upward in the epidermis gradually
being removed from the nutrients provided by the basal layer and eventually
losing their ability to divide neoplastic an essential property of cells.
This latter mechanism does not apply to melanocytes, however.
THE BIOLOGY OF CUTANEOUS MALIGNANT MELANOMA
As mentioned in the first part of this chapter, malignant melanoma can be
considered to be the neoplastic endpoint of melanocyte transformation.
Transformation represents a disturbance in cell behavior; it is acquired over
a finite time period through the slow and cumulative modification of different
abilities in progressively growing groups of cells. Once a critical mass of
cells with the appropriate properties is present they can then be designated a
neoplasm. The characteristics of a neoplasm are that "it is an abnormal new
growth of tissue that is uncontrolled, has no expected endpoint and is
DRAFT FINAL *
-------
3-26
aggressive to the host" (Perez-Tamayo 1984). An additional characteristic of
malignancy that is acquired is the ability of a neoplasm to spread from its
site of origin to other sites, i.e., to metastasize.
The biology of cutaneous malignant melanoma is complex in that there are
several morphologic types of melanoma which may have different pathways of
histogenesis; they tend to behave differently in terms of age at appearance
and characteristic location, and yet have common elements in their tumor
progression. What follows is a description of the various morphologic types,
information on their unique and similar characteristics and finally a
discussion of their similar method of tumor progression. This information is
important to the interpretation of the epidemiologic and experimental studies
discussed later in this document because seeming incongruities in the
relationship of sunlight (or UVB) and the disease entity known as CMM'may be
explained by the different characteristics of these tumor types.
Morphology: Terminology for the various morphologic types was proposed
by McGovern et al. (1973) and has been subsequently modified (Smith 1976;
Elder et al. 1980). The descriptions that follow are composites drawn from
the above sources. There are four principal classes of melanoma:
(1) melanoma arising in Hutchinson's melanotic freckle
(HMFM)
(2) superficial spreading melanoma (SSM)
(3) nodular melanoma (NM)
(4) unclassifiable melanoma (UCM).
Melanoma arising in Hutchinson's melanotic freckle begins with a
characteristic preinvasive lesion, the melanotic freckle, which is a linear
proliferation of atypical melanocytes in the basal layer of the epidermis.
The freckle is often marked by a profuse production of melanin, some of which
may be taken up by cells in the dermis. A lymphocytic infiltrate may also be
present. Such freckles generally occur in skin marred by solar degeneration,
having atrophy of the epidermis and elastosis of the dermis. The freckle
progresses to malignant melanoma via a characteristic course including
gradually increasing numbers of melanocytes which are initially
morphogenically normal in shape. As proliferation increases, the arrays of
melanocytes begin to take on a palisade appearance and the cells begin to
assume a spindle shape. Clusters of cells begin to form and the abnormal
melanocyte development can be observed in the external root sheath of the hair
follicle and sometimes even in the sweat ducts. The cell nests generally
remain localized to the dermo-epidermal junction with a relative lack of
invasion into epidermal areas outside the circumference of the freckle.
Superficial spreading melanoma develops as a flat, irregularly expanding,
slightly palpable lesion which, although more circumscribed than Hutchinson's
freckle, often has an irregular outline with prominent indentations. The
color may vary considerably within the lesion, initially consisting of varying
DRAFT FINAL
-------
3-27
shades of brown, black and bluish-grey. As the lesion grows, there may also
be areas of grey-white, blue-grey, or whitish-pink indicative of complete or
partial depigmentation following spontaneous regression.
The proliferating melanocytes of SSM are not, like those of HMFM, confined
to the dermoepidermal junction but rather invade the epidermis and then the
dermis. Melanocytes of SSM in the dermoepidermal region do not display the
pleomorphism seen in HMFM but rather are generally uniform in size and shape.
Once they invade the dermis, however, they may retain their characteristic
shape or may become epithelioid, spindle cell or nevus cell-like. This flat,
radially spreading stage of the evolution of a melanoma (either SSM or HMFM)
is termed the "radial growth phases" (RGP). It is believed that most, if not
all melanomas which achieve the radial growth phase will progress to the next
stage, the vertical growth phase characterized by the appearance of a-nodule
within the pre-existing plaque of the RGP (Elder 1986) .
Nodular melanoma is the third category of melanoma; it is used for those
tumors which are never observed in the radial growth phase but are first
observed in the vertical growth phase. Such "pure" vertical growth phase
tumors are characterized by a much more rapid rate of evolution than SSM or
HMFM. These aggressive tumors first appear as palpable nodules with no
antecedent radial growth phase. The nodularity is evidently the result of a
rapid focal proliferation of melanocytes in which cells at the dermo-epidermal
junction invade the dermis from the onset of tumor development.
The fourth category of malignant melanoma is unclassifiable melanoma.
Tumors are placed in this category when they do not fit readily into the other
three categories. This may arise because of some distinctive characteristic,
e.g., in hyperkeratotic papillary melanoma, or because tumors have components
of more than one of the three classes, e.g., the pleomorphism of the cells
seen in HMFM accompanied by the dermo-epidermal location normally observed in
SSM.
In addition to the four histogenic types of melanoma reviewed above,
several groups have suggested the need to distinguish a fifth variant: acral
lentigenous melanoma (ALM). These melanomas occur in the palms, soles,
subungual regions and the mucocutaneous junction of the oral and nasal
cavities and the anus (Briele and das Gupta 1979; Mastrangelo 1982; Clark et
al. 1976; Arrington et al. 1977). Melanomas at these sites comprise 7 to 9
percent of all cutaneous melanomas although in certain populations, e.g., the
Black, this type of melanoma, and in particular melanoma of the soles, is the
most common form of melanoma (Arrington et al. 1977). According to Clark et
al. (1976) this type of melanoma is similar in its developmental pathway to
SSM although, histolbgically, Arrington et al. (1977) indicate its
characteristics to be more similar to HMFM. This discrepancy might be
important in treatment because HMFM is considered by some to be a less
aggressive tumor, which may be treated conservatively whereas ALM is
certainly, when fully evolved, an aggressive variant with a poor prognosis.
ALM is of epidemiologic interest because although it occurrs with
approximately equal frequency in all races itis virtually the only type of
melanoma observed in blacks and individuals who have never lived south of the
* DRAFT FINAL * * *
-------
3-28
Artie Circle (Mark et al 1986), and because of its location in sites protected
from light by thick keratin layers, e.g., the palms and the soles.
COMPARATIVE INCIDENCE, SITE, SEX, AGE, AND RACIAL DISTRIBUTION
In the U.S., by far the most common type of melanoma is SSM; estimates of
its contribution to the total number of cases range between 54 and 70 percent
(Adler and Gaeta 1979; Briele and das Gupta 1979). The mean age at diagnosis
is 45 years and the tumor occurs about equally in both sexes. Estimates of
the proportion of melanoma cases represented by nodular melanoma range from 12
to 32 percent (Adler and Gaeta 1979; Briele and das Gupta 1979) and the mean
age at diagnosis is 50 years (Briele and das Gupta 1979). The least common
form of melanoma is HMFM. It constitutes from 10 to 14 percent of the
melanoma cases (Adler and Gaeta 1979; Briele and das Gupta 1979) and the
median age at diagnosis is 70 years (Briele and das Gupta 1979). Figure 3-11
shows the age distribution by type of melanoma for data from the Melanoma
Clinical Cooperative Group, 1978 (Sober et al. 1979, p. 630).
Table 3.5 indicates the site and sex distribution of primary malignant
melanoma observed in 4868 cases from NCI's SEER data for the period 1978-1981
(Scotto and Fears 1986). These data agree reasonably well with those
presented by needs Smith (1976) (Table 3-6) which though from a much smaller
sample, are also presented in order to show the relationship to other analyses
presented in Tables 3.6 and 3.8 to make point criteria are not uniformly that
appreciation of classification. Table 2.6 gives the age, sex and racial
distribution observed by Smith (1976) in the.M.D. Anderson population.
HMFM comprises almost one-half of the melanomas of the head and neck
(Adler and Gaeta 1979) and is about twice as common in women as men (Briele
and das Gupta 1979).. Superficial spreading melanoma accounts for about 70
percent of the melanomas of the lower extremities of women (Adler and Gaeta
1979) but occurs about equally among the two sexes (Briele and das Gupta
1979). Nodular melanoma is said to occur most commonly on the sole of the
foot by one group (Adler and Gaeta 1979) and on the head, back, and neck by
another (Clark et al. 1975). The latter group also indicates that nodular
melanoma occurs about twice as often in men as in women.
Table 3-7 presents the histogenic distribution of melanomas observed by
Smith (1976).
These data are not in accord with those of Adler and Gaeta (1979) who
found that HMFM on head and neck represented about 50 percent of all cases
rather than the 25 percent of the cases found here. SSM does represent about
70 percent of the lower extremities' tumors in this sample and nodular
melanoma is about equally distributed between head, neck, trunk and the lower
extremities.
Although, as detailed above, the various histogenic forms of melanoma have
different incidences, site, sex, and racial distributions, these tumors have
in common their mode of progression (Elder et al. 1980) Each (except NM) is
thought to begin in a radial growth phase and proceed to a vertical growth
DRAFT FINAL
-------
3-29
§§20
10
Figure 3-11. Age Distribution by Type of Melanoma:
Superficial Spreading (SSM). Nodular (NM) and
Lentigo Maligna (LLM) Melanomas
Source: Sober et al. (1979).
DRAFT FINAL
-------
3-30
TABLE 3-4
PRIMARY MELANOMA-DISTRIBUTION BY SITE AND SEX (SEER)
Males
Females
Head
and Neck
573
386
Upper
Extremities
503
647
Trunk
1,072
616
Lower
Extremities
11
33
Total
2,402
2,464
TABLE 3-5
PRIMARY MELANOMA-DISTRIBUTION BY SITE AND SEX
Males
Females
Head
and Neck
87
43
Upper
Extremities
46
61
Trunk
59
25
Lower
Extremities
61
128
Total
253
257
TABLE 3-6
AGE, SEX AND RACIAL DISTRIBUTION OF PRIMARY
MELANOMAS (M.D. ANDERSON)
Age
Sex
Male
Female
00-
Race 09
White
Black
Latin American
White 1
Black
Latin American 1
10- 20-
19 29
6 24
-
-
8 24
-
1
30-
39
47
-
1
45
-
~
40-
49
48
-
-
54
1
1
50-
59
43
2
2
51
1
1
60-
69
32
1
3
41
1
""
70-
79
29
-
1
18
2
~
80-
89 90+ Total
14 - 243
3
- — 7
5 1 248
5
4
DRAFT FINAL * *
-------
3-31
TABLE 3-7
HISTOGENIC DISTRIBUTION OF PRIMARY MELANOMAS
(M.D. ANDERSON)
Flat
Site (SSM)
Nodular
on Flat
(SSM)
Nodular
(Pure)
HMFM
HMFM (nodular)
UCM
Head and Neck 34 35 29 17 12 3
Trunk 33 33 13 2 3 0
Upper Extremities 52 23 24 6 1 ' 1
Lower Extremities 102 38 38 209
Total 221 129 104 27 16 13
phase. NM develops directly as a vertical growth phase nodule. In the radial
phase, tumors "form a spreading patch or plaque that expands radially with
time but becomes only slightly elevated." The neoplastic melanocytes may be
found in the basal layer of the epidermis and, relatively early in tumor
development, also in the papillary dermis. A competent host immune response
possibly limits proliferation in the dermis at least initially. At some
point, however, the tumor escapes this control and begins to develop vertical
growth. At that time, the neoplastic cells - possibly a clone - have acquired
the ability to grow in groups or sheets and a nodule appears. This phase may
also be initially limited to an expansion within the papillary dermis until
such time as the neoplastic cells acquire additional characteristics which
allow them to survive in the more hostile (less vascular) environment of the
reticular dermis. Further changes may then occur which confer on these cells
the ability to metastasize (Elder et al. 1980).
The rate at which these developments occur may vary tremendously among the
various morphologic types of melanoma. The rate of development of HMFM is the
slowest: between 5 and 20 years. SSM progresses somewhat faster - 1 to 7
years, and nodular melanoma is very rapid - on the order of months
(Fitzpatrick and Soter 1985). Indeed nodular melanoma may skip the radial
growth phase, or so shorten it as to make it appear missing. It should be
noted that once SSM and HMFM enter the vertical phase then they may be
indistinguishable (prognostically though not clinically) from NM, although
some believe that even in the vertical phase HMFM progresses much more slowly
than SSM. This may be because HMFM and other lentigenous melanomas (e.g.,
ALM), are more frequently associated with spindle cells or desmoplastic
vertical growth phases (Elder et al. 1980), an observation which has led to
the suggestion (discussed below) that HMFM and SSM have different precursor
cells.
Several theories have been suggested to explain the observed histogenic
differences in malignant melanoma particularly with regard to SSM and HMFM.
* » * DRAFT FINAL - » *
-------
3-32
Perhaps the best known one is that of Mishima (1967) in which, on the basis of
electron raicrosopic observations, he proposed that melanoma be divided into
two separate histogenic lineages -- one derived from melanocytes and one
derived from nevus cells. In his report, Mishima (1967) suggests that HMFM
develops from melanocytes and is characterized by radiosensitivity and a
slower rate of growth, metastasis, and invasiveness than that observed in
melanomas developing from junctional nevi. The latter type of melanoma would
be equivalent to SSM or pure NM. Tumors from junctional nevi are more rapidly
growing, invasive and metastatic. Mishima also distinguished between these
two lineages on the basis of their melanosomes, their degree of spontaneous
regression, and the presence of sun-damaged skin. HMFM regresses more
completely, shows ellipsoid melanosomes, and appears more frequently in the
presence of sun-damaged skin than does SSM and rarely shows more than partial
regression.
Mishima's theory has been accepted by some researchers (Holman et al.
1983) but not by others (Paul and Illig 1976). Holman et al. (1983) use the
Mishima hypothesis as the starting point for a further hypothesis of their own
in which they postulate that HMFM arises from sun-damaged melanocytes whereas
UCM and SSM arise from initiated nevus cells. The initiation process can be
effected by a variety of agents, e.g., viruses, chemicals or UV. Once a nevus
cell is generated, however, its subsequent development can be "promoted" by
UV, estrogens/other hormones, or trauma.
Paul and Illig (1976), however, do not .feel that a dualistic theory of
melanoma histogenesis is appropriate. In their work, they studied the
presence of dendritic-branched tumor cells in all types of malignant
melanoma. They found that such cells occur in all tumors, but that this is to
some extent controlled by the environment. This led them to hypothesize that
the difference between HMFM on the one hand and NM and SSM on the other was
that the properties of HMFM were attributable to the sun-damaged skin in which
it develops. These authors proposed in addition that the nevus cell is an end
stage in differentiation, i.e., that it is not the precursor to SSM and UCM
proposed by Mishima. In this position, Paul and Illig (1976) are seconded by
Ackerman and Mihara (1985) who challenge the contention that the dualistic
hypothesis is supported by the available data. These authors view the
dysplastic nevus as one of the many variants of melanocytic nevi and, in their
experience "rarely see evidence of malignant melanoma in continuity or
contiguity with a dysplastic nevus." They contend that those researchers who
assert that 20 to 40 percent or more of all malignant melanomas begin in
association with pre-existing dysplastic nevi have misinterpreted the
histology of melanomas in situ to conclude that they are dysplastic nevi.
Clark and his colleagues (Clark et al. 1986) believe that ALM, HMFM and
SSM differ etiologically. Light is not considered to be an inducer of ALM
because of its site preference (soles, palms and subungual surfaces), racial
distribution (it is virtually the only form in blacks) and "occurrence in
persons who have never lived south of the Artie circle." In contrast, HMFM
and SSM, are thought by these authors to be induced by light, but in different
fashion. The favored sites for HMFM are the face, neck and back of the hands
-- sites irradiated by light in a relatively continuous fashion. "if
irradiation of this type is to induce melanoma, it does so, as a rule on the
* * * DRAFT FINAL * * *
-------
3-33
face of a fair, freckled person who has few or no melanocytic nevi." The
melanomas which develop do so via an indolent hyperplastic process which
progresses to HMFM uncommonly.
SSM is thought to derive from a process which begins with the induction of
melanocytic nevi by discontinuous or sporadic light exposure. The susceptible
individual has "a cutaneous phenotype quite different from that of the
fair-freckler - a phenotype that tans better and is darker." In some
individuals the distributions of freckles and nevi appear to be mutually
exclusive with freckles and no nevi on areas of continuous exposure such as
the face and with nevi and no freckles on areas receiving discontinuous
irradiation. The nevi may progress through hyperplasia to dysplasia and
thence (rarely) to SSM and (even more rarely) to NM (Clark et al. 1986).
SUMMARY AND CONCLUSIONS
The information reviewed in this chapter is drawn from a wide variety of
disciplines and indicates, to some degree, the complexity of the problem
addressed by this document. The chapter is designed to provide background
information important to understanding the ultimate question that this
document must answer: If the UVB component of sunlight is increased will there
be an increase in the incidence, mortality and/or morbidity of melanoma?
The summary points of this chapter are as follows:
3.1 Melanoma is a cutaneous neoplasm whose incidence in
white Caucasian populations is increasing. It has been
estimated that by the year 2000 one out of every 150
individuals will develop melanoma in their lifespan.
3.2 Melanocytes are the precursor cell for melanoma. Basal
cells and squamous cells are differentiation states of
keratinocytes; when transformed these cells become
basal cell and squamous cell carcinomas, respectively.
3.3 Melanocytes may have 2 differentiation states:
melanocytes and nevus cells. There is some discussion
whether a nevus cell is a differentiated state of the
melanocyte or a premalignant precursor cell for
melanoma.
3.4 Melanocytes produce melanin and distribute it to the
keratinocytes via organelles termed melanosomes.
Blacks and whites differ in the amount of melanin
produced by their melanocytes, the quality and quantity
of melanosomes delivered to the keratinocytes, but not
in the total number of melanocytes per site. Black
(pigmented) skin is between five and ten times more
effective at protecting the basal layer from UVB and
other wavebonds of UVR than white (non-pigmented) skin.
* DRAFT FINAL
-------
3-34
3.5 The highest concentration (~2300/mm3) of
melanocytes is found on the cheek while the back and
the thigh have the lowest number (approximated 900 and
1000/mm3, respectively). This distribution pattern
differs from those of all forms of melanoma except
possibly HMFM.
3.6 Melanin absorbs in a broad range of UV wavelengths
although evidently not much beyond 1200 nm. There are
several forms of melanin; red hair melanin
(pheomelanin) has a greater ability to produce
superoxide (free radicals) when irradiated with UV and
visible radiation light than does black hair melanin
(eumelanin).
3.7 The wavelengths of sunlight that reach the earth's
surface lie above 290 nm. UVB (290-320) comprises a
very small amount of the total energy from sunlight
reaching the earth but is considered to contain much of
the biological activity.
3.8 There are five different types of melanoma: 1) melanoma
in Hutchinson's melanotic freckle (HMFM) or lentigo
maligna melanoma, 2) superficial spreading melanoma
(SSM), 3) nodular melanoma (NM) , 4) unclassified
melanoma and 5) acral lentiginous melanoma (ALM). They
behave differently in site preference", in their
relationship to cumulative sun exposure, and possibly
in their precursor cells. These differences in
behavior may be important to the question of whether
sunlight is an agent for melanoma in that the answer
may have to differ qualitatively or quantitatively for
each of the different types of melanoma.
3.9 Certain authorities on melanoma believe that the
relatively continuous solar irradiation received on
face, neck and back of hands, when it induces melanoma
does so on the face of the fair freckled individual
with few or no melanocytic nevi via an indolent tumor
progression pathway involving increasing degrees of
atypical melanocytic hyperplasia (lentigo maligna)
until HMFM occurs after years of growth.
3.10 The same authorities also believe that discontinuous or
sporadic light exposure such as that received by the
back induces melanocytic nevi which may progress to
abnormal melanocytic dysplasia, an followed, on rare
occasions, to SSM and even more rarely to NM.
* * DRAFT FINAL « *
-------
3-35
3.11 UVB radiation is known to cause DNA damage via the
formation of pyrimidine dimers. The most active
wavelengths are in the UVB range, however, UVA may also
cause DNA damage. The latter damage is probably not
via pyrimidine dimers, but is thought to involve
reactive oxygen species.
•• * * DRAFT FINAL * » *
-------
3-36
REFERENCES
Ackerman, A.B. and Mihara, I. Dysplasia, dysplastic melanocytes, dysplastic
nevi, the dysplastic nevus syndrome and the relation between dysplastic nevi
and malignant melanomas. Human Pathology 16:87-91. (1985).
Adler, S. and Gaeta, J.F. Malignant melanoma. In: Cancer Dermatology Helm,
F. (ed). Philadelphia: Lea and Febiger pp 141-157 (1979).
Allen, E.P. Malignant melanoma spontaneous regression after pregnancy. Brit
Med J 2:1067 (1955).
Anderson, D.E., Smith, J.L. and McBride, C.M. Hereditary aspects of malignant
melanoma. J Am Med Assoc 200:741-746 (1967).
Anderson, R.R. Tissue optics and photoimmunology. Chapt 4, In:
Photoimmunology Parrish, J.A., Kripke, M.L. and Morison, W.L. (eds). New
York: Plenum Medical Book Co. pp 61-76 (1983).
Armstrong, B.K., Holman, C.D.J., Ford, J.M., and Wooding, T.L. Trends in
melanoma incidence and mortality in Australia. In: Trends in Cancer
Incidence: Causes and Practical Implications Magnus, K. (ed). New York:
Hemisphere Publishing Corp pp 399-417. (1980).
Arrington, J.H., Reedj R.J., Ichinose, H., and Krementz, E.T. Plantar
lentiginous melanoma: a distinctive variant of human cutaneous malignant
melanoma. Amer J Surg Path 1:131-143 (1977).
Balinsky, B.I. An Introduction to Embryology. Philadelphia: W.B. Saunders
Co. (1965).
Beral, V., Ramcharan, S., and Faris, R. Malignant melanoma and oral
contraceptive use among women in California. Brit J Cancer 36:804-809 (1977).
Brenner, W., and Gschnait, F. Decrease DNA repair activity in sunburn cells.
Arch Dermatol Res 266:11-16 (1979).
Briele, H.A. and Das Gupta, T.K. Natural history of cutaneous malignant
melanoma. World J Surg 3:255-270 (1979).
Bruls, W.A.G., Slaper, H., van der Leun, J.C. and Berrens, L. Transmission of
human epidermis and stratum corneum as a function of thickness in the
ultraviolet and visible wavelengths. Photochem and Photobiol 40(4):485-494
(1984).
Clark, W.H. Jr., Ainsworth, A.M., Bernadino, E.A., Yang, C.H., Miln, M.C. and
Reed, R.J. The developmental biology of primary human malignant melanoma.
Semin Oncol 2:83-103 (1975).
* DRAFT FINAL
-------
3-37
Clark, W.H. Jr., Elder, D.E., Guerry, D., Epstein, M.N., Greene, M.H. and van
Horn, M. A study of tumor progression: The precursor lesions of superficial
spreading and nodular melanoma. Human Pathology 15(12):114-1165 (1984).
Clark, W.H. Jr., Elder, D.E., and van Horn, M. The biologic forms of
malignant melanoma. Human Pathology 17:443-450 (1986).
Clark, W.H. Jr., From, L., A.M., Bernadino, E.A., and Mihn, M.C. The
histogenesis and biologic behavior of primary human malignant melanoma of the
skin. Cancer Res 29:705-726 (1969).
Clark, W.H. Jr., Mastrangelo, M.J., Ainsworth, S.M., Berd, D., Bellet, R.E.
and Bernadino, G.A. Current concepts of the biology of human cutaneous
melanoma Adv Cancer Res 24:267-338 (1977).
Clark, W.H. Jr., Min, B.H., and Kligman, L.H. The developmental biology of
induced malignant melanoma in guinea pigs and a comparison with other
neoplastic systems. Cancer Res 36:4079-4091 (1976).
Davis, N., McLeod, R., Beardmore, G. et al.. The Henry Joseph Windsor
Lecture: Melanoma is a word, not a sentence. Aust NZJ Surg 46:188-196
(1976).
Doynes, R.A., Samlowski, W.E., Burnham, D.K., Gahring, L.C. and Roberts, L.K.
Immunobiological consequences of acute, and chronic UV exposure. Curr Prob
Derm 15:176-194 (1986).
Elder, D.E., Jucovy, P.M., Tuthill, R.J. and Clark, W.H. Jr. The
classification of malignant melanoma. Am J Dermatopath 2:315-320 (1980).
Elder, D.E., Greene, M.H., Bondi, E.E., Clark, W.H., Ackerman, A.B. (eds).
Acquired Melanocytic Nevi and Melanoma. The dysplatic nevus syndrome, Chapt.
11, Pathology of Malignant Melanoma pp. 185-215, New York, Masson Publishing,
USA (1981).
Elder, D.E. Personal communication (1986).
Fitzpatrick, T.B. and Soter, N.A. Pathophysiology of the skin, in
Pathophysiology: The Biological Principles of Disease Smith, L.H., Jr. and
Thier, S.O. (eds). pp. 1297-1330 (1985).
Gange, R.W., Blackett, A.D., Matzinger, E.A., and Sutherland, B.M.
Differential effects of UVA- and UVB-induced pyrimidine dimer formation and
erythema in human skin. J Invest Dermatol 82:419-420 (1984).
Gange, R.W., Park, Y.K., Auletta, M., Kagetsu, N. , Blackett, A.D., and Parish
F.A. Action spectra for cutaneous responses to ultraviolet radiation In: The
Biological Effects of UVA Radiation Urbach, F. and Gange, R.W. (eds). New
York: Praeger Publishers pp 57-65 (1986).
* - » DRAFT FINAL * » *
-------
3-38
Gange, R.W. and Parrish, J.A. Acute effects of ultraviolet radiation upon the
skin. Chapt 5, In: Photoimmunology Parrish, J.A., Kripke, M.L. and Morison,
W.L. (eds). New York: Plenum Medical Book Co. pp 77-94 (1983).
Green, A., MacLennan, R. and Siskind, V. Common acquired naevi and the risk
of malignant melanoma. Int J Cancer 35:297-300 (1985).
Greene, M.H., Golden L.R., Clark, W.H.,Jr., Lovrian, E., Kraemer, K.H.,
Tucker, M.A., Elder, D.E., Fraser, M.C. amd Rowen, S. Familial cutaneous
malignant melanoma: Autosomal dominant trait possibly linked to the Rh
locus. Proc Natl Acad Sci USA 80:6071-6075 (1983).
Gschnait, F. and Brenner, W. Patterns of UV injury to the epidermis, in Marks
R and Christophers E (ed) The Epidermis in Disease; Proceedings of the
European Society for Dermatological Research, Philadelphia, Lippincott pp.
423-429 (1982).
Harber, L, Bickers, D. and Kochevar, I. Introduction to ultraviolet and
visible radiation. Chapt 2, In: Photosensitivity Diseases: Principles of
Diagnosis and Treatment. Harber, L.C., and Bickers, D.R. (eds).
Philidelphia: W.B. Saunders. pp 13-23 (1981).
Harber, L.C. and Bickers, D.R. (eds). Photosensitivity Diseases: Principles
of Diagnosis and Treatment. Philidelphia: W.B. Saunders (1981).
Holman, C.D.J. and Armstrong, B.K. Pigmentary traits, ethnic origin, benign
nevi and family history as risk factors for cutaneous malignant melanoma. JNCI
72:257-266. (1984).
Holman C.D.J., Armstrong, B.K. and Heenan, P.J. A theory of the etiology and
the pathogenesis of human cutaneous malignant melanoma. JNCI 71:651-656'
(1983).
Holman, C.D.J., James, I.R., Gattey, P.H. and Armstrong, B.K. An analysis of
trends in mortality from malignant melanoma of the skin in Australia. Intl J
Cancer 26:703-309 (1980).
Horii, I., Kawasaki, K., Koyama, J. Histidine-rich protein as a possible
origin of free amino acids of the stratum corneum. J Dermatol 10:25-33 (1983)
Hu, F. Melanocyte cytology in normal skin, melanocytic nevi and malignant
melanomas: A review. In: The Pathology of Malignant Melanoma Ackerman, A.B.
(ed). New York: Masson Publishing USA. pp 1-22 (1981).
Jacquez, J.A., Huss, J., McKeehan, W., Dimitroff, J.M. and Kuppenheim, H.F.
Appl Physiol 8:297-299 (1955a).
» » - DRAFT FINAL * »
-------
3-39
Jacquez, J.A., Kuppenheim, H.F., Dimitroff, J.M.', McKeehan, W. and Huss, J.
Spectral reflectance of human skin in the region 235-700 ran. J Appl Physiol
8:213-214 (1955b).
Johnson, B.E., Mandell, B., Daniels, F. Melanin and cellular reactions to
ultraviolet radiation. Nature 235:147-149 (1972).
Kaidbey, K.H., Agin, P.P., Sayre, M.M., and Kligman A.H. Photoprotection by
melanin -- A comparison of Black and Caucasian skin. J Amer Acad Dermatol
1:249-260 (1979).
Katz, S.I., Tamaki, K., and Sachs, D.H. Epidermal Langerhans cells are
derived from cells originating in bone marrow. Nature 252:324-326 (1979).
Konishi, T. Resistance to ultraviolet radiation in human melanoma cells. Mie
Med J 31(l):15-27 (1981).
Kopf, A.W., Kripke, M.L. and Stern, R. S. Sun and malignant melanoma. Am Acad
Dermatol 11:674-684 (1984).
Kraemer, K.H., Greene, M.H., Tarone, R., Elder, D.E., Clark, W.H. Jr., and
Guerry, D. Dysplastic naevi and cutaneous melanoma risk. Lancet ii:
1076-77 (1983).
Krementz, E.T., Sutherland, C.M., Carter, R.D. Malignant.melanoma in the
American Black. Ann Surg 183:533-542 (1967).
Kripke, M.L. Antigenicity of murine skin tumors induced by ultraviolet
light. J Natil Cancer Inst 53:1333-1336 (1974).
Kripke, M.L. Effects of methoxsalen plus near-ultraviolet radiation or
mid-ultraviolet radiation on immunologic mechanisms. Natl Cancer Inst Monogr
66:247-251 (1984).
Kubitschek, H.E., Baker, K.S. and Peak, M.J. Enhancement of mutagenesis and
human skin cancer rates resulting from increased fluences of solar ultraviolet
radiation. Photochem Photobiol (In press 1986).
Laerum, O.D. and Iversen, O.H. Biology of Skin Cancer (Excluding Melanoma).
UICC Technical Report Series Vol. 63 Geneva (1981).
Lever, W.F. and Schaumburg-Lever, G. Histology of the Skin. In :
Histopathology of the skin. Philadelphia: J.B. Lippincott Co. pp 681-687
(1979).
Ley, R.D., and Applegate, L.A. Ultraviolet radiation-induced histopathologic
changes in the skin of the marsupial Monodelphis domesticus II.
Quantitative studies of the photoreactivation of induced hyperplasia and
sunburn cell formation. J Invest Dermatol 85:365-367 (1985).
* * * DRAFT FINAL
-------
3-40
Mastrangelo, M.J., Rosenberg, S.A., Baker, A.R. and Katz, H.R. Cutaneous
melanoma. Chapter 31, In: Melanoma Cancer: Principles and Practice of
Oncology DeVita, V.T., Hellman, S., Rosenberg, S.A. (eds). Philadelphia:
J.P. Lippincott Company pp 1124-1170 (1982).
McGovern, J.J., Mihm, M.C. Jr., Bailly, C., Booth, J.C., Clark W.H. Jr.,
Cochran, A.J., Hardy, E.G., Hicks, J.D., Levene, A., Lewis, M.G., Little, J.H.
and Milton, G.W. The classification of malignant melanoma and its histologic
reporting. Cancer 32:1446-1457 (1973).
-i
Menon, I.A., Persad, S. , Haberman, H.F., and Kurian, C.J. A comparative
study of the physical and chemical properties of melanins isolated from human
black and red hair. J Invest Dermatol 80:203-206 (1983a).
Menon, I.A., Persad, S., Ranadive, N.S., and Haberman, H.F. Effects of
ultraviolet-visible irradiation in the presence of melanoma isolated from
human black or red hair upon Ehrlich ascites carcinoma cells. Cancer Res
43:3165-3169 (1983b).
Mishima, Y. Melanocytic and nevocytic malignant melanomas: cellular and
subcellular differentiation. Cancer 20(5):633-649 (1967).
Moore, G.E. and Moore, M.B. Auto-inoculation of cultured human lymphocytes in
malignant melanoma. NY State J Med 69:400-462 (1969).
Morison, W.L. What is the function of melanin? Arch Dermatol 131:1160-1163
(1985).
Nathanson, L., Hall, T.C., Vawtor, G. and Farber, S. Melanoma as a medical
problem. Arch Int Med 119:479-492 (1967).
National Cancer Institute. eds. 1985 Annual Cancer Statistics Review,
presented to National Cancer Advisory Board (1985).
Nordlund, J.J. Genetic basis of pigmentation and the disorders of
pigmentation In : Pathology of Malignant Melanoma Ackerman, A.B. (ed). New
York: Masson Publishing USA pp 23-45 (1981).
Pathak, M.A. Activation of the melanocyte system by ultraviolet radiation and
cell transformation. Ann N Y Acad Sci 453:328-339. (1985).
Paul, E. and Illig, L. Malignant Melanoma and Nevocellular Nevi Doerr, W.
and Leonhart, H. (eds). New York : Georg Theme Verlag (1984).
Paul, E. and Illig, L. Melanin-producing dendritic cells and histogenesis of
malignant melanoma. Arch Derm Res 257:163-177 (1976).
Pearl, DH. A stocastic cancer model involving the environment of transforming
cells with application to skin cancer. Ph.D. Thesis University of California
at Berkeley (1984).
* * DRAFT FINAL * * »
-------
3-41
Perez-Tamayo, R. Mechanisms of Disease: An Introduction to Pathology, Volume
Two. Year Book Medical Publishers, Inc. Chicago pp. 173-174, 560-569,
613-615 (1984).
Reed, R.J. Neuromesenchyme. The concept of a neurocrestic effector cell for
dermal mesenchyme. Am J Dermatopath. 5:385-395 (1983).
Romsdahl, M.M. and Cox, I.S. Biological aspects of pigment cells and
malignant melanoma In : Neoplasms of the Skin and Malignant Melanoma. Year
Book of Medical Publishers, Inc. pp 251-277 (1976).
Scheibner, A., Hollis, D.E., McCarthy, W.H., and Milton, G.W. Effects of
sunlight exposure on Langerhans cells and melanocytes in human epidermis.
Photoderm 3:15-25 (1986).
Scott, I.R. Factors controlling the expressed activity of histidine
ammonia-lyase in the epidermis and the resulting accumulation of urocanic
acid. Biochem J 194:829-838 (1981).
Scott, I.R., Harding, C.R. and Barrett, J.G. Histidine-rich protein of the
kertokyalin granules: source of the free amino acids urocanic acid and
pysolidone carboxylic acid in the stratum corneum. Biochem and Biophys Acta
719:110-117 (1982).
Scotto, J. Melanoma among Caucasians in the United States. Increases
predicted for the 1980s. The Skin Cancer Found J (1985).
Soctto, J., and Fears, T.R. The association of solar ultraviolet radiation
and skin melanoma among Caucasians in the United States. Cancer
Investigations (1986 in press).
Scotto, J. and Fraumeni, J.F. Skin (Other than Melanoma), pp. 996-1011,
Chapter 60. In: Cancer Epidemiology and Prevention. Schottenfeld and
Fraumeni (eds.) Philadelphia: W.B. Saunders Co. (1982).
Smith, J.L. Jr. Histopathology and -biologic behavior of malignant melanoma.
In: Neoplasms of the Skin and Malignant Melanoma..- Year Book of Medical
Publishers, Inc. pp 293-330 (1976).
Snell, R.S. and Bischitz, P.G. The melanocytes and melanin in human abdominal
wall skin: A survey made at different ages, in both sexes and during
pregnancy. J Anat 97:361-376 (1963).
Sober, A.J., Mihm, M.C., Fitzpatrick, T.B., Clark, W.H., Jr. Malignant
melanoma of the skin, and benign neoplasms and hyperplasias of melanocytes in
the skin In: Dermatology in General Medicine. 2nd ed New York: McGraw-Hill
Book Company pp 629-654 (1979).
Spikes, J.D. UV-damage in biological systems In: Experimental and Clinical
Photoimmunology Daynes, R.A. and Spikes, J.D. (eds). CRC Press Inc. (1983)
DRAFT FINAL
-------
3-42
Stingl, G. , Katz, S.I., Clement, I., Green, I., and Shevach, E.M. Immunologic
functions of la bearing epidermal Langerhans cells. J Immunol 121:2005-2013
(1978).
Summer, W.C. Spontaneous regression of melanoma: Case report Cancer
6:1040-1043 (1953).
Toews, G.B., Bergstresser, P.R., Streilein, J.W., and Sullivan, S. Epidermal
Langerhans cell density determines whether contact hypersensitivity or
unresponsiveness follow skin painting with DNFB. J Immunol 124:445-455
(1980).
Wan, S., Anderson, R.R. and Parrish, J.A. Analytical modeling for the optical
properties of the skin with in vitro and in vivo applications. Photochem
Photobiol 34:493-499 (1981).
* * DRAFT FINAL
-------
CHAPTER 4
TIME-RELATED FACTORS IN THE INCIDENCE AND
MORTALITY: AGE, PERIOD, AND BIRTH COHORT EFFECTS
The increasing incidence and mortality due to cutaneous malignant melanoma
have been the subject of numerous international studies and reports over the
past two decades. Most publications have been based upon tumor registry and
vital statistics data from westernized countries where recordkeeping systems
are generally very good. Research has primarily focused on: definition of
secular trends; verification of trends; description of age-specific curves;
and cohort analysis of CMM rates to determine the respective contributions of
time-related effects. This chapter summarizes the literature addressing
time-related factors in the incidence and mortality of CMM.
SECULAR TRENDS IN INCIDENCE AND MORTALITY
Sharp rises in incidence and mortality due to cutaneous malignant melanoma
have been reported in nearly all Caucasian populations worldwide (refer to
Table 4-1). Magnus (1982) analyzed trends in incidence in Norway, Sweden,
Denmark, Finland, and Iceland for the period 1943-1976 and found that,
although the absolute levels of incidence varied from country to country, an
approximate annual increase of 6 percent was observed in all five countries
with no sign of leveling off. Likewise, Connecticut incidence rates increased
five-fold from 1935 to 1974, with an annual average increase of 5 percent
(Roush et al. 1985b). Osterlind and Jensen (1986) noted that increases in
Danish incidence were most pronounced since 1955, and showed no signs of
leveling off as late as 1982. Danish mortality, however, did not begin to
increase until 1965, about 10 years after incidence rates began to rise
markedly (Osterlind and Jensen 1986). Incidence rates in the United States
increased steadily until 1983 when a 5 percent decrease was observed; however,
this drop is thought to be an artifact due to the advent of DRG (disease
related group) legislation, which is resulting in decreased hospitalization of
individuals with melanoma in the United States (NCI 1985).
In general, increases in mortality due to CMM have been less steep than
those observed for incidence. From 1943-1982, Danish incidence rose five-fold
while corresponding mortality rates doubled, with signs of leveling off in
women since 1975, but not in men (Osterlind and Jensen 1986). Leveling off of
Australian mortality has been evident at least since 1965-1969 in both sexes,
although incidence continues to increase (Armstrong 1981). In Sweden, overall
incidence rose by 7 percent per year from 1959-1968, while only negligible
changes in mortality occurred in both sexes (Malec and Eklund 1978).
Similarly, data from the New Mexico Tumor Registry indicate that incidence
rose considerably from 1969-1976, but mortality did not change over this
nine-year time period (Pathak et al. 1982). Lee (1982a) reported steady
increases in mortality of 3 percent per year over the period 1951-1975 in
England and Wales, Canada, and the United States. Lee (1982a) noted that if
diagnosis or treatment had improved over this time period, their influences
were sufficiently constant that no change in the mortality trends were
observed. More recent data from the United States (1974-1983) indicate that
DRAFT FINAL
-------
TABLE 4-1
INCREASES IN INCIDENCE AND MORTALITY FROM
MALIGNANT MELANOMA FROM DIFFERENT COUNTRIES
First Period
of Observation
I nc idence
I nc idence
Morta I
Morta I
Morta I
Morta I
Morta I
Morta I
"ty
ity
ity
ity
ity
ity
Inc idence
I nc idence
Morta I
ity
Country
New York State
Norway
Norway
Canada
United Kingdom
Austra 1 ia
Denma rk
Sweden
Connect icut
U.S.A.
U.S.A.
Sex
M
F
M
F
M
F
M
F
Both
M
F
M
F
M
F
M
F
WM
WF
WM
WF
Time
1941-1943
1941-1943
1955
1955
1956-1960
1956-1960
1951-1955
1951-1955
1950
1931-1940
1931-1940
1956-1960
1956-1960
1956-1960
1956-1960
1935-1939
1935-1939
1974
1974
1950
1950
Rate
6
per 10
1.2
1.8
1.8
2.6
1.6
1.3
0.7
0.6
0.5
1,0
0.8
1.6
1.6
1.7
1 . 1
1.1
0.9
6.7
6.0
1.0
0.8
Second Period
of Observation
Time
1967
1967
1970
1970
1966-1970
1966-1970
1966-1970
1966-1970
1967
1961-1970
1961-1970
1966-1969
1966-1969
1966-1968
1966-1968
1975-1979
1975-1979
1983
1983
1977
1977
Rate
6
per 10
3.4
2.9
6.3
6.8
2.7
1.8
1.4
1.2
1.0
3.6
2.5
2.4
2.1
2.1
1.5
8.2
6.8
9.6
8.3
2.6
1.6
Tota I
Percent
Increase
176
65
264
195
69
36
93
107
100
267
227
49
32
30
40
645
656
43
38
160
200
Number of
Years
25
25
15
15
10
10
15
15
16
30
30
10
10
9
9
40
40
10
10
27
27
Annua 1
Percent
Increase
7.0
2.6
17.6
13.0
6.9
3.6
6.2
7.1
6.3
8.9
7.6
4.9
3.2
3.3
4.4
16.1
16.4
4.3
3.8
5.9
7.4
Adapted from: Elwood and Lee (1976); NCI 198 ); NCI (1985a); NCI (1985b).
-------
4-3
mortality due to CMM has continued to increase significantly both in white
males and females (NCI 1985).
Secular trends in CMM mortality are less steep than those for incidence
probably due to earlier diagnosis rather than increasing diagnosis and
registration of benign and semi-malignant cases in recent years (Magnus 1977,
Elwood and Lee 1975). A study of cancer registries over the period 1955-1980
in Alabama and New South Wales, Australia (Balch et al. 1983) showed that
melanoma skin tumors tended to be detected at progressively earlier stages of
development; tumors were thinner, less invasive, and less likely to present
with ulcerations. In addition, nodular tumors became less frequent while the
incidence of superficial spreading melanomas, which have a better prognosis,
increased over the period. Incidence rates of nodular and superficial
spreading melanoma in Finland, however, increased to the same extent 'over the
period 1963-1976 (Teppo 1981).
WORLDWIDE INCREASES IN CUTANEOUS MELANOMA: ARTIFACTUAL OR REAL?
Whenever significant changes in incidence or mortality rates are observed,
it is necessary to determine whether the trends are genuine. In 1977,
Ressuguie suggested that rising trends in CMM incidence were largely a result
of improved diagnosis and registration (as cited in Lee 1982a). Histo-
pathological studies, however, indicate that criteria for diagnosis of CMM
have been consistent over time in Norway (Magnus 1975), and that the quality
of diagnoses in well-run population-based cancer registries has been excellent
(Pakkanen 1977, Malec et al. 1977). In addition, reasonable consistency in
histologic diagnosis has. been demonstrated between pathologists working in the
same city (McCarthy 1980) and in different countries (Larsen 1980).
Considerable observational data have also led most investigators to con-
clude that actual increases in the incidence and mortality of CMM have
occurred. Osterlind and Jensen (1986) point out that if substantial changes
in histopathological criteria and registration efficiency have occurred, higher
rates of diagnosis would most likely have resulted in stepwise increments in
incidence, rather than the gradual rise observed. Furthermore, Danish
mortality tends to mirror incidence rates with a ten-year lag period,
suggesting that the rise in incidence is real (Osterlind and Jensen, 1986).
Lee (1982) has noted that improved diagnosis or registration is not likely to
impact mortality since it would mean that the lower mortality rates observed
earlier were due to the incorrect registration of large numbers of deaths in
countries having very good systems of vital statistics. The fact that propor-
tional increases in incidence have been similar in populations from both high
and low incidence areas, such as Queensland and Hawaii (high rates), and Canada
and the United Kingdom (low rates) further suggest that the increases are real
(Elwood and Hislop 1982) and associated with a universal factor(s). Incidence
rates also demonstrate distinct patterns over time by sex, age, and anatomic
site, further suggesting that increases in rates are consistent with changes
in sunlight exposure habits by sex and birth cohort (refer to Chapter 5).
There appears to be general agreement that the sharp rise in the incidence
of CMM is associated with increasing exposure to ultraviolet radiation through
sun exposure as a consequence of changing clothing and leisure habits (Magnus
* * * DRAFT FINAL * * *
-------
4-4
1981). The etiologic mechanism of solar radiation in the causation of CMM,
however, remains controversial (Magnus 1982).
AGE-SPECIFIC TRENDS IN INCIDENCE AND MORTALITY
The age-specific incidence and mortality curves for CMM are unlike those
for most other forms of cancer, which tend to increase linearly with
increasing age (Cook et al. 1969, Elwood and Lee 1978, Magnus 1982). In
contrast, steep increases in CMM incidence begin in adolescence, leveling off
through middle age, followed by less steep increases in the older age groups.
Distinct changes in the shape of the age-specific curves occur when rates are
stratified by sex and anatomical site (refer to Chapter 5).
Age-specific incidence curves also change in shape according to the decade
of diagnosis, as shown in the Danish population in Figure 4-1 for the years
1943-1982 (Osterlind and Jensen 1986). Among individuals diagnosed from
1943-1952, incidence increased gradually with increasing age. Over subsequent
decades, the age-specific curves showed progressively steeper increases from
ages 20 to 50, followed by a more gentle slope or plateau in the older age
groups (Osterlind and Jensen 1986). These changes over time in the shape of
the cross-sectional age-specific curves suggest the potential influence of
birth cohort effects (Lilienfeld and Lilienfeld 1980). Based on Figure 4-1,
it is not surprising that the mean age at diagnosis of CMM has tended to
decrease over time in Denmark.
Using cancer registry data from Norway, Magnus (1981) compared the
age-specific incidence curves for the two periods, 1955-1970 and 1971-1977,
and found that the difference between the two curves was greatest for the age
groups 30 to 70. Magnus (1981) concluded that this finding was due to birth
cohort effects operating primarily on the individuals born between 1900 and
1930. This conclusion will be discussed further in the following section oh
cohort analyses of CMM incidence and mortality.
COHORT ANALYSES OF CMM INCIDENCE AND MORTALITY
The technique of cohort analysis involves careful study of incidence,
mortality, or prevalence rates (i.e., the proportion of individuals with the
disease at a specified time period) in individuals born in the same period of
time, usually within the same decade. Age-specific rates in cohorts are
compared, the major objective being to distinguish the three time-related
effects - age, period of diagnosis (i.e., calendar time), and birth cohort -
that might explain the changing trends (Kleinbaum et al. 1982). An age effect
is present when the disease rate varies by age, regardless of birth cohort; a
period effect is present when the disease rate varies by time, regardless of
age or birth cohort; a cohort effect is present when the disease rate varies
by year of birth, regardless of age (Kleinbaum et al. 1982). Cohort analyses
may be conducted graphically, and using statistical modelling techniques which
attempt to separate the respective contributions of age, period, and birth
cohort effects.
As early as 1961, Haenszel suggested on the basis of United States data
that, "persons born after 1885 have been exposed with increasing intensity to
* * * DRAFT FINAL * * *
-------
4-5
Males
300 _
20.0 _
10 0_
e.c_
5fl_
o
030
O
a.
Oi
2.c.
"5 a:.
§60
Females
.• •*•-._.'971-1977 •" .
2.0 _
01:
20 3C 40 50 6C
80 2C X
Age
. j ;..
5C 6C
~ e:
FIGURE 4-1
AVERAGE ANNUAL AGE-SPECIFIC INCIDENCE RATES OF
CUTANEOUS MALIGNANT MELANOMA
IN NORWAY 1955-1970 AND 1970-1977
Source: Magnus (1981).
* DRAFT FINAL
-------
4-6
100.0 -i
8
S
a
D
u
C
u
10.0 -
1.0 -
0.1
•— lfU-73
• • • U7S-M
15 20 25 30 35 40 45 SO 55 60 65 70 75 8085+
AGE
100.0 n
10.0 -
1
1.0 -
0.1
1MJ-73
15 20 25 30 35 40 45 SO 55 60 65 70 75 8085+
AGE
FIGURE 4-2
AGE-SPECIFIC MALE AND FEMALE INCIDENCE RATES OF
MALIGNANT MELANOMA OF THE SKIN IN DENMARK,
FOR ALL SITES COMBINED AND FOR
SUCCESSIVE TIME PERIODS
Source: Osterlind and Jensen (1986).
DRAFT FINAL
-------
4-7
some factor(s) associated with high skin cancer mortality" (Gordon et al.
1961). These increases in skin cancer mortality in successive cohorts were
apparently due to malignant melanoma, although CMM was not separately
classified until the sixth revision of the International Classification of
Diseases in 1950. In 1970, Lee and Carter first associated the long-term
trends in total skin cancer mortality with the effects of CMM, and concluded
that year-of-birth effects were most likely responsible for the secular
increases. Other birth cohort analyses of overall CMM trends have since been
conducted by investigators using data from several different countries, all
reporting similar findings.
Graphical Analyses of Birth Cohort Effects
Magnus (1981; 1982) plotted age-specific incidence rates for separate
birth cohorts in Norway over the period 1955-1977 as shown in Figure 4-3.
Here it can be seen that the risk of malignant melanoma within each cohort
rises consistently throughout life, as is true for most other cancers.
Graphing CMM rates by birth cohort substantially changed the age-specific
curve from that observed in the cross-sectional data; the stable rates in
middle age seen cross-sectionally disappeared. Shifting of the birth cohort
curves to the left as seen in Figure 4-3 implies that there are consistent
increases in incidence for each successive cohort. The cohort effect in
Norway is most marked for individuals born from 1900 to 1930, where distance
between the curves is greatest. For example, at ages 45-49, individuals born
1920-1929 had incidence rates four times higher than those of individuals born
1900-1909 (refer to Figure 4-3). Cohort curves after 1930 are closer
together, suggesting the differences in incidence rates by cohort are being
reduced. One factor postulated as being responsible for differences in
incidence rates in successive birth cohorts is solar radiation.
Cohort effects are most evident for sites which have shown the greatest
increase in incidence over time, such as the trunk in males and lower limbs in
females, and are minimal for sites which increased less dramatically, such as
the face and neck (Magnus 1981; Houghton et al. 1980; Boyle et al. 1984;
Stevens and Moolgavkar 1984).
For CMM of the trunk in Norwegian males 50-54 years old, the incidence
rate for the 1920-1929 birth cohort was six times that of the 1900-1909 birth
cohort (Magnus 1981). As shown in Figure 4-4, among males and females born
between 1890 and 1909, the incidence of CMM of the face was greater than that
of the trunk among males and lower limbs among females. For cohorts born
between 1930 and 1949, however, the highest incidence (per skin surface area)
was of the trunk of males and the lower limbs of females. Magnus (1981)
concluded that the ratio of carcinogenic exposure to the trunk-lower limbs and
to the face-neck varied according to year of birth. He suggested that the
shift in melanoma distribution by cohort was consistent with changes in
clothing and suntanning habits during the first half of this century.
Magnus (1982) notes a slight tendency for the cohort curves to level off
with age, particularly in generations born after World War I. The leveling
off of the cohort curves is best seen in incidence rates for CMM of the trunk
and lower limbs, particularly among Norwegian cohorts born from 1900-1929.
* » * DRAFT FINAL - * *
-------
4-8
200 r-
100
80
SO
40
030
o
o
Q.
a, !0 —
o C8 .
•5 06 .
03 _
C2 _
Males
1890
Females
193C
1940
•— -49
1930
-39.-*
,*,S2C
1890
'900 -99
188°
./ q
Cl
20 30 40 50 60 7C 80 SO 2: 3C 4C SC 60 70 30 90
Age
FIGURE 4-3
AGE-SPECIFIC INCIDENCE RATE OF TOTAL SKIN MELANOMA
BY COHORT IN NORWAY 1955-1977
Source: Magnus (1982).
DRAFT FINAL * * *
-------
4-9
19JO- 49
< 001
Males
IbOin
1910 -29
20 )C 1C SO JC
1850-I9C9
J- . i -J . »__ Jl._. I (_
tC bC 60 70 50 5C 70
1930-49
; s
Females
(bO"M
I9!0- 29
I850-H09
Age
30 -C SO X. "to" 'X"5;">0 ~5C""5C
FIGURE 4-4
INCIDENCE OF MALIGNANT MELANOMA IN NORWAY
1955-1977 PER AREA UNIT OF THE PRIMARY SITE3
AGE SPECIFIC RATES FOR COHORTS
The area unit corresponds to 1 percent of the total skin surface.
Source: Magnus (1981).
* * - DRAFT FINAL •• * *
-------
4-10
Magnus (1981) notes that this leveling off with age is "rather atypical for
most types of cancer", and postulates that it may be due to decreases in
sunning with increasing age. This statement implies that sunlight might have
a promoting, as well as an initiating role in the etiology of CMM.
Birth cohort analyses of incidence rates in Denmark (1943-1972),
Connecticut (1935-1974) (Houghton et al. 1980), and Finland (1953-1973) (Teppo
et al. 1978) yielded overall findings consistent with those of Magnus (1981,
1982), i.e., increasing incidence rates were observed in successive birth
cohorts. The first sign of changing incidence for all sites was observed
among those born 1892-1895 in Connecticut and Denmark (Houghton et al. 1980).
Rising incidence began at different times for different anatomic sites in
Denmark, with changes in incidence for those born as early as 1882 for the
trunk-neck in males and lower limbs in females (Houghton et al. 1980)'. The
incidence curves for facial CMM changed little in successive cohorts.
Muir and Nectoux (1982) conducted cohort analyses with CMM incidence data
from Australia, Czechoslovakia, England and Wales, France, Japan, and the
Netherlands. Cohort effects were apparent in all six nations, but were most
discernable in individuals born from 1910-1930. Even in Japan where incidence
rates are 15 times lower than those in Australia, an upward trend in incidence
of comparable magnitude was observed in successive cohorts (Muir and Nectoux
1982). Later birth cohorts born from 1940-1950 did not differ with respect to
incidence. This lack of increase in rates among the youngest cohorts may be
due to a small number of cases, or might indicate that later cohorts had
similar exposure to the carcinogenic factor(s) (Muir and Nectoux 1982). Muir
and Nectoux (1982) concluded that the universality of a cohort effect strongly
implicates an environmental factor that is widespread, and affects
light-skinned people more and both sexes equally, though on somewhat different
parts of the body.
Utilizing recent CMM incidence data from the Connecticut Tumor Registry
(1940-1979) Roush et al. (1985a) plotted age-specific incidence rates for
seven age groups by birth cohort (Figure 4-5). The age-specific incidence
curves were generally parallel on a semi-log scale, showing that rates within
age groups increased similarly with each successive cohort. The slopes of the
curves, however, showed a tendency to be less steep for individuals in age
groups born 1925 and later. If rates are actually leveling off in the younger
cohorts, this could be interpreted to mean that sunlight exposure has peaked
in these individuals, i.e., that the maximum change in behavior patterns with
regard to leisure time has occurred. This finding is consistent with those of
Muir and Nectoux (1982).
Birth cohort analyses of New Zealand's non-Maori population were conducted
for the period 1948-1977 revealing different trends for CMM incidence and
mortality (Cooke et al. 1983). While incidence rates continued to increase in
recent cohorts, mortality rates stabilized in both sexes by the 1924 birth
cohort. These mortality changes are similar to those observed in a number of
other countries (Elwood and Lee 1974, Lee et al. 1979, Holman et al. 1980).
The New Zealand incidence patterns are also similar to those described in
Norway, Finland, Denmark, and Connecticut (Magnus 1981, 1982; Teppo et al.
1978, Houghton et al. 1980). Cooke et al. (1983) concluded that both the
* * * DRAFT FINAL « * *
-------
4-11
MALES
1875 1885 1895 1905 1915 1925 1935 1945 1955
MIDPOINT OF BIRTH COHORT
1865 1875 1885 1895 1905 1915 1925 1935 1945 1955
MIDPOINT Of BIRTH COHORT
FIGURE 4-5
AGE-SPECIFIC INCIDENCE RATES IN CONNECTICUT FOR CUTANEOUS
MALIGNANT MELANOMA BY SEX AND COHORT
Source: Roush et al. (1985a).
* * * DRAFT FINAL *
-------
4-12
incidence and mortality data were correct, and that the stabilization of
mortality rates was most likely due to improvements in prognosis.
Modelling Approaches in the Evaluation of Cohort Effects
Lee et al. (1979) studied mortality rates by cohort in the United States
(white population), England and Wales, and Canada for the period 1951-1975.
They calculated age-specific cohort slopes for each sex within each of the
three populations, finding large and consistent differences in mortality rates
with each successive cohort. These authors concluded that secular increases
in mortality over this time period (approximately 3 percent per year) were
caused by cohort effects. Although case-fatality (measured as the proportion
of CMM cases that died among CMM cases diagnosed over a specified time period)
decreased over this period, slopes of the age-specific cohort curves did not
appear changed. This observation suggests that any effects of earlier
diagnosis or improved treatment occurred evenly over the study period, thereby
failing to alter the trend in the slopes (Lee et al. 1979)
Holman et al. (1980) examined Australian mortality rates due to malignant
melanoma over the period 1931-1977, when rates more than quadrupled in both
sexes. Estimates for the independent effects of calendar year, birth cohort,
and age on mortality were determined using statistical modelling techniques.
On the basis of their analysis, Holman et al. (1980) concluded that virtually
all of the secular trends in mortality rates could be attributed to increases
in successive cohorts, beginning with those born from 1865-1885. Increases by
birth cohort, however, stabilized by the 1925 (women only) and 1935 (men only)
birth cohorts. Slowing of mortality rates has also occurred in cohorts born
around this period (1926) in England and Wales (Lee and Carter 1970) and
Finland (1930-1940) (Teppo et al. 1978). Holman et al. (1980) emphasized that
the stabilization of rates was probably not due to the immigration of persons
with lower rates of CMM because migrants were not overrepresented among the
cohorts in which the rates leveled off (i.e., 1925 and after). Instead,
cohort trends in CMM mortality were more likely associated with lifestyle
changes which led to more recreational exposure to the sun over the
generations (Holman et al. 1980). The authors, however, were unable to state
whether Australian sun exposure habits have stabilized, and if so, whether
those born in 1925 and after would have been the first cohorts affected.
Improvements in prognosis would probably affect all cohorts equally, and
therefore are not likely to account for the stabilization of mortality rates
in the later cohorts.
The individual effects of age, birth cohort, and calendar year derived
from CMM mortality rates by Holman et al. (1980) are shown on graphs in
Figures 4-6, 4-7, and 4-8. The age factor (Figure 4-6) rose sharply between
the age groups 10-14 and 30-34, followed by less rapid increases in the
subsequent age groups. The time factor, after correction for age and birth
cohort, demonstrated very little change with year of death (Figure 4-7). It
can be seen that the birth cohort factor (Figure 4-8) increased with
successive cohorts from about 1865 to 1935 in males, and 1865 to 1925 in
females. Holman et al. (1980) concluded that secular increases in CMM
mortality will continue for approximately 30-40 years, until the cohorts born
before the stabilization of rates (i.e., 1925-1935 cohorts) die.
* * * DRAFT FINAL * « *
-------
4-13
i-
-2-
-J
-S
-7
!•- t§- M- I
-14 -II -24 -
X- •»- 4»- M-
ACE
H -M -m
—! 1 1 1—
Jt- 71- M- tt>
•M -n - M
FIGURE 4-6
STATISTICAL MODELLING RESULTS FOR THE EFFECT
OF AGE ON CMM MORTALITY RATES IN AUSTRALIA
(1931-1977), ADJUSTING FOR EFFECTS OF
CALENDAR YEAR AND BIRTH COHORTS
Source: Holman et al. (1980).
* Standard errors associated with point estimates are indicated by I for
males and t for females and are conditional on elimination of linear trend
in the time factors.
DRAFT FINAL
-------
4-14
B -<
3
-10-
KALtJ
'MS IISI ItSS IUO 111$ 1171 II7S IW« lilt in« l|« 1M« lj« tilt 1I1S tin IMS 1131 I IB l»*4 1MI 1H« 1IU 1M«
MEDIAN YEAR OF IIMTN
FIGURE 4-7
STATISTICAL MODELLING RESULTS FOR THE EFFECT OF
BIRTH COHORT ON CMM MORTALITY RATES
IN AUSTRALIA (1931-1977), ADJUSTING FOR EFFECTS
OF AGE AND CALENDAR YEAR
Source: Holman et al. (1980).
* Standard errors associated with point estimates are indicated by I for
males and t for females and are conditional on the elimination of linear
trend on the time factors.
* * * DRAFT FINAL * * *
-------
4-15
i .
§
i
FEMAIES
MALES
1131-34 1I3$-3I tM«-44 1t4S-4t
1 1 1
1IM-M 19M-H 1IM-M
VEAR OF DEATH
IMt-M 1171-74 1175-77
FIGURE 4-8
STATISTICAL MODELLING RESULTS FOR THE EFFECT OF
CALENDAR YEAR ON CMM MORTALITY RATES
IN AUSTRALIA (1931-1977), ADJUSTING FOR EFFECTS OF
AGE AND BIRTH COHORT
Source: Holman et al. (1980).
* Standard errors associated with point estimates are indicated by I for
males and t for females and are conditional on the elimination of linear
trend on the time factors.
DRAFT FINAL
J- J. .t.
-------
4-16
Utilizing a modelling approach similar to that of Holman et al. (1980),
Venzon and Moolgavkar (1984) conducted cohort analyses of CMM mortality in
five countries: Australia, New Zealand, United States, Canada, and England and
Wales. These countries were selected to represent populations with high,
intermediate, and low rates of CMM in order to determine whether similar time-
related effects were present in populations varying in CMM risk. Under all
models tested, cohort effects were seen to "drive up mortality" within each
country. In addition, the relative increases in mortality due to cohort
effects were approximately the same in the five countries. Thus, Venzon and
Moolgavkar (1984) were able to derive an age-specific mortality curve,
corrected for birth cohort effects, using combined data from all five
populations. A nearly straight-line relationship of CMM mortality and age was
observed, the slope being somewhat less in women .than men. The authors stated
that the lower slope in females might reflect a larger proportion of female
deaths in the younger age groups; it could also be interpreted as reflecting
an excess of male deaths in the older age groups. Lee and Storer (1978)
discuss a hormonal risk factor in premenopausal women that could be
responsible for higher mortality in young women.
When cohort effects derived from statistical modelling were plotted for
the five countries, cohort effects appeared to be leveling off in recent
cohorts (Venzon and Moolgavkar 1984). The authors suggested that this may
represent a slowing down of the increase in incidence for melanomas of the
trunk and lower limbs (i.e., sites of greatest cohort effects), or possibly
improving prognosis for these sites.
Both Stevens and Moolgavkar (1984) and Boyle et al. (1984) modelled the
independent effects of age (i.e., correcting for birth cohort effects) on
site-specific CMM incidence rates using data from Denmark and Connecticut, and
Norway, respectively. Both studies noted rapidly increasing risk by birth
cohort for all sites. Stevens and Moolgavkar (1984) concluded that the fit of
their model showed a similar age-dependence for all common subsites of CMM;
while Boyle et al. (1984) found that age-dependent relationships differed by
site (refer to Chapter 5). Discrepancies between the findings and conclusions
of Stevens and Moolgavkar (1984) and Boyle et al. (1984) may be associated
with the application of different statistical models to different sets of
data. In addition, these authors grouped the anatomical sites somewhat
differently.
Roush et al. (1985a,b) conducted cohort analyses of CMM incidence data
from Connecticut, 1940-79, using statistical modelling techniques similar to
those used by Holman et al. (1980) and Venzon and Moolgavkar (1984). As in
similar studies, modelling demonstrated the importance of cohort effects on
CMM incidence rates, while period effects were not detected. Roush et al.
(1985b) suggested, however, that period effects (i.e., time) could
theoretically be present since CMM rates by year of diagnosis show marked
fluctuations annually with sunspot activity (Houghton et al. 1978), or
seasonally with changes in sunlight exposure (Swerdlow 1979; Holman and
Armstrong 1981). The period effects, present as cross-sectional fluctuations
in the rates, could be superimposed on the underlying cohort patterns
preventing their detection (Roush et al. 1986b). Holman et al. (1983) have
suggested that annual or monthly fluctuations in diagnosis of CMM would be
* * * DRAFT FINAL * - *
-------
4-17
consistent with promotional effects of sun on transformed cells in the
development of melanoma. Roush et al. (1985b) suggested that period effects
and cohort effects may reflect different stages of neoplastic transformation
(i.e., promotional and initiating, respectively) in the etiology of CMM.
The dramatically changing public health importance of CMM was emphasized
in another recent analysis of Connecticut Tumor Registry data by Roush et al.
(1986a). Modelled summary incidence rates (age-adjusted) for cutaneous
malignant melanoma were compared with rates of colon cancer within the
youngest birth cohorts. The analysis revealed that incidence rates for CMM in
the 1955 cohort will rival those for colon cancer, presently the third most
common cancer site in Connecticut. Thus, in the coming decades, CMM could
easily become one of the most common malignancies in the absence of preventive
measures (Roush et al. 1986a).
FINDINGS
The following findings can be drawn based on the studies reviewed in this
chapter:
4.1 Sharp increases in incidence and mortality have been
reported in White but not non-White populations
worldwide. Based on observational and analytical
evidence, most experts agree that the trends are
. genuine, and not due to increases in the registration
and diagnosis.
4.2 Steeper increases have been reported for incidence
versus mortality rates. In addition, there are
indications that mortality rates are leveling off in
some areas where incidence rates continue to rise
annually, such as Australia, Denmark, and New Mexico.
Diagnosis at earlier stages of the disease leading to
increased survival is thought to be a major cause for
the leveling off of mortality rates.
4.3 The age-specific incidence curve for CMM is unlike that
for most other forms of cancer which tend to increase
linearly with increasing age. Steep increases in CMM
incidence begin in adolescence, level off in middle
age, and show low rates of increase, if any, in the
older age groups. This low slope of age-specific
incidence is due to the high lifetime risk of melanoma
in younger individuals. The slope of the age-specific
incidence curve increases substantially when rates are
plotted on a semi-log scale and stratified by birth
cohort.
6.4 Most authors who have conducted cohort analyses of CMM
incidence and mortality rates conclude that virtually
all secular increases in CMM are due to cohort
effects. In most countries, the first signs of
* * * DRAFT FINAL * * *
-------
4-18
increasing rates are seen in cohorts born around 1900,
although increases in cohorts born as early as 1865 are
observed in Australia and New Zealand. In Norway and
several other countries, there is a tendency for a
slowing of the increase in incidence in cohorts born
around and after 1930. Stabilization of mortality
rates is also occurring in cohorts born 1925-1935 and
later in countries such as Australia, New Zealand,
England and Wales, and Finland.
4.5 On the basis of the Connecticut Tumor Registry data,
statistical modelling indicated that the incidence
rates of CMM in the 1955 birth cohort will rival those
for colon cancer, currently the third most common
cancer site in Connecticut.
* * * DRAFT FINAL * »
-------
4-19
REFERENCES
Armstrong, B.K., Holman, C.D.J., Ford, J.M. and Wooding, T.L. Trends in
melanoma incidence and mortality in Australia. In: Trends in Cancer
Incidence, Causes and Practical Implications. Magnus, K. (ed.), Hemisphere
Publications Corporation, pp. 399-417 (1982).
Balch, C.M., Soong, S., Milton, G.W., Shaw, H.M., McGovern, V.J., McCarthy,
W.H., Murad, T.M. and Maddox, W.A. Changing trends in cutaneous melanoma
over a quarter century in Alabama, USA, and New South Wales, Australia.
Cancer 52: 1748-1753 (1983).
Boyle, P., Day, N.E. and Magnus, K. Mathematical modeling of malignant
melanoma trends in Norway, 1953-1978. Am J Epi 118:887-896 (1983).
Cook P.J., Doll, R. and Fellingham S.A. A mathematical model for the age
distribution of cancer in man. Int J Cancer 4:93-112 (1969).
Cooke, K.R., Skegg, D.C.G. and Fraser, J. Trends in malignant melanoma of
skin in New Zealand. Int J Cancer 31:715-718 (1983).
Elwood, J.M. and Lee, J.A.H. Recent data on the epidemiology of malignant
melanoma. Sem Oncol 2:149-154 (1975).
Elwood, J.M. and Hislop. Solar radiation in the etiology of cutaneous
malignant melanoma in Caucasians. Nat Cancer Inst Mono 62:167-171 (1982).
Gordon, T., Crittenden, M. and Haenszel, W. End results and mortality trends
in cancer. II. Cancer mortality trends in the United States. Nat Cancer
Inst Mono 6:133-350 (1961).
Holman, C.D.J. and Armstrong, B. Skin melanoma and seasonal patterns. Am J
Epi 113:202 (1981).
Holman C.D.J., Heenan P.J., Caruso V., Clancy, R.J. and Armstrong, B.K.
Seasonal variation in the junctional component of pigmented neavi. Int J
Cancer 31:213-215 (1983).
Holman C.D.J., Armstrong B.K., and Heenan P.H. Guest editorial: A theory of
the etiology and pathogenesis of human cutaneous malignant melanoma. Nat
Cancer Inst 71:651-656 (1983).
Holman, C.D.J., James, I.R., Gattey, P.H. and Armstrong, B.K. An Analysis of
Trends in Mortality from Malignant Melanoma of the Skin in Australia. Int
J Cancer 26:703-709 (1980).
Houghton, A., Munster, E.W. and Viola, M.V. Increased incidence of malignant
melanoma after peaks of sunspot activity. Lancet 1:759-760 (1978).
* DRAFT FINAL *
-------
4-20
Houghton, A., Flannery, J., and Viola, M.V. Malignant melanoma in Connecticut
and Denmark. Am J Cancer 25:95-104 (1980).
Kleinbaum, D.G., Kupper, L.L. and Morgenstern, H. Epidemiologic Research:
Principles and Quantitative Methods. New York: Van Nostrand Reinhold
Company (1982).
Larsen, T.E., Little, J.H., Orell, S.R., and Prade, M. International
pathologists' congruence survey on quantitation of malignant melanoma.
Pathology 12:245-253 (1980).
Lee, J.A.H. and Carter, A.P. Secular trends in mortality from malignant
melanoma. J Nat Cancer Inst 45:19-97 (1970).
Lee, J.A.H., Petersen, G.R., Stevens R.G. and Vesanen, K. The influence of
age, year of birth, and date on mortality from malignant melanoma in the
populations of England and Wales, Canada, and the white population of the
United States. Am J Epi 110:734-739 (1979).
Lee, J.A.H. Melanoma. In: Cancer Epidemiology and Prevention, Shottenfeld,
D. and Fraumeni, J.F. Jr. (eds.). W.B. Saunders Company, Philadelphia
(1982a).
Lee, J.A.H. Melanoma and exposure to sunlight. Epi Rev 4:110-136 (1982b).
Lee, J.A.H. and Storer, B.E. Further .studies on skin melanomas apparently
dependent on female sex hormones. Int J Epi 11:127-131 (1982).
Lee, J.A.H. Epidemiology of malignant melanoma: 10 years' progress. Pigment
Cell 6:1-21 (1983).
Lilienfeld, A.M. and Lilienfeld, D.E. Foundations of Epidemiology. New
York: Oxford University Press (1980).
Magnus, K. Incidence trends in the Nordic countries: Effects of sun
exposure. In: Trends in Cancer Incidence, Causes and Practical
Implications,
Magnus, K. (ed.), Hemisphere Publication Corporation, pp. 387-392 (1982).
Magnus, K. Habits of sun exposure and risk of malignant melanoma: An analysis
of incidence rates in Norway 1955-1977 by cohort, sex, age, and primary
tumor site. Cancer 48:2329-2335 (1981).
Magnus, K. Incidence of malignant melanoma of the skin in the five nordic
countries: Significance of solar radiation. Int J Cancer 20:477-485
(1977).
Magnus, K. Epidemiology of malignant melanoma of the skin in Norway with
special reference to the effect of solar radiation. In: Biological
Characterization of Human Tumors, pp. 249-259, Exerpta Medica
International Congress Series No. 375, Amsterdam (1975).
» * DRAFT FINAL * * *
-------
4-21
Magnus, K. Incidence of malignant melanoma of the skin in Norway, 1955-1970.
Variations in time and space and solar radiation. Cancer 32:1275-1286
(1973).
Malec, E., Eklund, G., and Langerlof, B. Reappraisal of malignant melanoma
diagnosis in the Swedish Cancer Registry. Acta Path Microbiol Scand 85:
707-712 (1977).
Malec, E. and Eklund, G. The changing incidence of malignant melanoma of the
skin in Sweden, 1959-1968. Scand J Plas Reconstr Surg 12:19-27 (1978).
McCarthy, W.H., Blach, A.L., and Milton, G.W. Melanoma in New South Wales: An
epidemiologic survey 1970-76. Cancer 46:427-432 (1980).
Muir, C.S. and Nectoux, J. Time trends: Malignant melanoma of the skin. In:
Trends in Cancer Incidence, Causes and Practical Implications, Magnus, K.
(ed.) Hemisphere Publishing Corp., pp. 365-385 (1982).
National Cancer Institute (NCI), Sondek, E., Young, J.L., Horn J.W. and Ries
L.A.G. (eds), 1985 Annual Cancer Statistics Review (1985).
Osterlind, A. and Jensen, O.M. Trends in incidence of malignant melanoma of
the skin in Denmark 1943-1982. Recent Results in Cancer Research 102:8-17
(1986).
Pakkanen, M. Clinical appearance and treatment of malignant melanoma of the
skin. Ann. Chir. Gynaecol. 66:21-30 (1977).
Pathak, D.R., Samet, J.M., Howard, C.A. and Key, C.R. Malignant melanoma of
the skin in New Mexico 1969-1977. Cancer 50:1440-46 (1982).
Pondes, S., Hunter, J.A.A., White, H, Mclntyre, M.A. and Prescott, R.J.
Cutaneous malignant melanoma in South-East Scotland. Quart J Med, New
Series L. 197:103-121 (1981).
Roush, G.C., Schymura, M.J., and Holford, T.R. Risk for cutaneous melanoma in
recent Connecticut birth cohorts. Am J Pub Hlth 75:679-682 (1985a).
Roush, G.C., Schymura, M.J., Holford, T.R., White, C. and Flannery, J.T. Time
period compared to birth cohort in Connecticut incidence rates for
twenty-five malignant neoplasms. J Nat Cancer Inst 74:779-788 (1985b).
Stevens, R.G. and Moolgavkar, S.H. Malignant Melanoma: Dependence of
site-specific risk on age. Am J Epi 119:890-895 (1984).
Swerdlow, A.J. Incidence of malignant melanoma of the skin in England and
Wales and its relationship to sunshine. Brit Med J 2:1324-1327 (1979).
Teppo, L., Pakkanen, M., and Hakulinen, T. Sunlight as a risk factor of
malignant melanoma of the skin. Cancer 41:2018-2027 (1978).
DRAFT FINAL * * *
-------
4-22
Teppo, L. Incidence trends of melanoma of the skin in Finland by histological
type. In: Trends in Cancer Incidence, Causes and Practical
Implications, Magnus, K. (ed.) Hemisphere Publishing Corp., pp. 393-397
(1982).
Venzon, D.J. and Moolgavkar, S.H. Cohort analysis of malignant melanoma in
five countries. Am J Epi 119:62-70 (1984).
Viola, M.V. and Houghton, A.N. Solar radiation and cutaneous reviews. Hosp
Prac 4:110-136 (1982).
* » * DRAFT FINAL - *
-------
CHAPTER 5
VARIATIONS IN THE ANATOMICAL DISTRIBUTION
OF CUTANEOUS MALIGNANT MELANOMA
The anatomical distribution of cutaneous malignant melanoma (CMM) has been
the subject of numerous epidemiological studies. Research efforts have
focused on site-specific trends in CMM incidence related to sex, age, race,
histogenic type, birth cohort, and season. The most pronounced differences
have been associated with gender, race, and birth cohort. Some researchers
have also investigated the effect of primary CMM site on patient prognosis.
Several factors which could explain the observed trends in CMM site
distribution have been presented in the literature, including effects of
occupation, recreational activities, clothing habits, pigmentation, trauma,
and exposure to sunlight. This chapter summarizes the epidemiological
information addressing the anatomical distribution of CMM.
OVERALL SITE DISTRIBUTION
The overall site distribution of melanomas among several study populations
is presented in Table 5-1. The relative proportion of CMMs at each site
listed in Table 5-1 varies considerably as a result of many potentially
influential factors, such as the sex and racial distributions of the study
populations. To gain a better understanding of CMM distribution by site, .it
is necessary to examine the epidemiological data by these and other
subcategories.
Gender Differences
Gender is associated with pronounced differences in CMM site distri-
bution. Table 5-2 lists data from several studies reporting CMM site
distribution among males and females. The percentage of total CMMs occurring
on the head/neck, the trunk, the upper extremities and the lower extremities
for each sex is presented based on data from different countries. This table
indicates that most epidemiological studies which have examined the effect of
sex on site distribution observe higher incidences of CMM on the lower
extremities among females and on the trunk among males than on other parts of
the body. Although these sites (the lower extremities and the trunk) are
relatively less exposed than the head, neck and upper extremities, most
researchers have concluded that the observed anatomical differences by sex are
not incompatible with the hypothesis that sunlight exposure is involved in the
development of CMM among Caucasians (Pathak et al. 1982, Hinds and Kolonel
1980, MacDonald 1976, and Movshovitz and Modan 1973).
The first 13 studies listed in Table 5-2 used data obtained primarily from
Caucasians. These studies indicate that melanomas of the trunk occurred about
one-and-a-half times more frequently in males than in females for the same
site. Of the total CMMs among males, the percentage occurring on the trunk
ranged from 25 percent (Denmark) to 53 percent (United States), in contrast to
the occurrence of melanomas of the trunk among females which ranged from 14
percent (Scotland) to 31 percent (United States).
* * * DRAFT FINAL » - »
-------
TABLE 5-1
ANATOMIC SITE DISTRIBUTION OF CUTANEOUS MALIGNANT MELANOMA
(Percentage of Total Tumors)
Locat ion
Years
Sample Size Head/Neck Trunk
Extremi t ies
Upper Lower
Source
a
United States
(calicos ian)
Texas
Texas
Alabama
New Mexico
New South Wa Ies
Queens land
New Zea land
( Maori /Po lynes ian)
I srae I
Norway
F i n land
Japan
Hong Kong
Uganda
1978-1981
1944-1966
1954-1970
1955-1980
1966-1977
1955-1980
1977
1963-1981
1960-1972
1955-1970
1953-1973
1961-1982
1964-1982
1963-1966
4,864
911
510
537
103
1,110
690
24
966
2,541
2,501
546
43
152
20
22
25
27
27
14
21
13
16
b
22
19
15
7
8
35
25
16
28
29
37
34
13
25
c
43
37
20
7
f
11
23
19
21
19
19
14
20
0
21
8
10
13
d
21
5
22
13
37
23
25
33
24
54
38
18
26
46
e
63
9
72
Scotto 1986
MacDonald 1976
Smith 1976
Ba Ich et a 1 . 1982
Pathak et a 1 . 1982
Ba Ich et a 1 . 1982
Little et a 1 . 1980
Moss 1984
Ana i se et a 1 . 1978
Magnus 1973
Teppo et a 1 . 1978
Takahashi 1983
Col I ins 1984
Kiryabwire et a 1 . 1968
l/l
N)
Based on data from Seattle, Detroit, Iowa, Utah, San Francisco/Oakland, Atlanta, and New Mexico.
3
Face.
Neck/truck.
i
17 percent of the total on the hands.
2
56 percent of the total on the feet.
r
"Skin" and genitals.
3
Feet and legs.
-------
TABLE 5-2
ANATOMIC SITE DISTRIBUTION OF CUTANEOUS
MALIGNANT MELANOMA BY GENDER
(Percentage of total tumors by gender )
Loca t ion
(Ethnic Group)
Extremities
Years Sample Size
Sex
Head/Neck
Trunk
Upper
Lower
Source
Predominantly Caucasian Study Populations
United States
a
United States
North America
( Caucas i an )
Europe
(Caucas ian)
Texas (White)
Hawa i i ( Caucasian)
New Mexico (Anglos)
1 srae 1
Scot 1 and
Fin land
Denma rk
Queens 1 and
Texas
1980 4,545
1978-1981 4,864
-
-
1944-1966 1,252
1960-1977 262
1966-1977 403
1960-1972 966
1961-1976 477
1953-1973 2,501
1943-1957 1,204
1977 713
1954-1970 510
F
M
F
M
F
M
F
M
F
M
F
M
F
M
F
M
F
M
F
M
F
M
F
M
F
M
15
23
16
24
16
27
16
23
17
23
9
24
22
33
16
16
25
32
18
19
24
32
19
23
17
34
31
53
25 '
45
29
44
23
42
18
25
28
37
20
41
19
32
14
27
28
48
19
25
21
45
10
23
19
13
26
21
18
17
13
13
25
20
28
24
21
16
15
21
14
11
11
9
14
9
22
17
24
18
35
11
33
11
36
12
47
23
28
11
35
15
37
10
50
31
47
30
36
17
33
.22
35
13
50
24
Ba Ich et a 1 . 1984
Scotto 1986
c
Crombie 1981
c
Crombie 1981
MacDonald 1976
Hinds and Kolonel 1980
Pathak et a 1 . 1982
Ana i se et a I . 1978
Pondes et a 1 . 1981
Teppo et a I . 1978
Clemmesen 1965 (cited
in Lee 1982)
Little et a 1 . 1980
Smith 1976
Ui
-------
TABLE 5-2 (Continued)
Loca t ion
(Ethnic Group)
Extremi t ies
Years Sample Size Sex
Head/Neck
Mixed Study Populat
Israel (Mixed Race)
Japan
Texas (Spanish
surname )
Texas (Non-White)
New Mexico
( H i span ics )
Hawa i i
( Non-Caucas ian)
Hawa i i
( Non-Caucas ian)
Uganda
1961-1967
1961-1982
1944-1966
1944-19146
1969-1977
1960-1970
1960-1977
1963-1966
368 F
M
546 F
M
206 F
M
30 F
M
35 F
. M
66 F
M
64 F
M
152 F
M
15
17
20
13
22
18
6
31
23
23
26
16
26
16
. 6
10
Trunk
ions
21
35
19
23
16
22
6
--
36
46
17
.19
17
20
h
11
h
11
Upper
15
5
18
11
18
11
18
8
32
23
13
19
13
20
4
5
Lower
42
25
d
42
e
51
30
30
53
54
9
8
44
47
f
44
9
44
i
78
j
74
Source
Movshovitz and Modan
1973
Takahashi 1983
MacDonald 1976
MacDonald 1976
Pathak et a 1 . 1982
Hinds and Kolonel 1980
Hinds 1979
Lewis 1967
Percentages may not total 100 percent because of rounding errors and exclusion of "other sites" of MM.
D
Based on data from Seattle, Detroit, Iowa, Utah, San Francisco/Oakland, Atlanta, and New Mexico.
Truck includes scrotum and "unspecified" melanomas.
J
29 percent of the total were on the feet.
:»
39 percent of the total were on the feet.
r
22 percent of the total were on the feet.
3
42 percent of the total were on the feet.
T
"Skin" and genitals.
63 percent of the total were on the feet.
j
64 percent of the total were on the feet.
-------
5-5
Similarly, the occurrence of CMM of the lower extremity in females was
generally two times greater than that for males. The percentage of total CMMs
on the lower extremities among females (for the same 13 studies) ranged from
28 percent (Texas) to 50 percent (Israel), whereas for males the range was 11
percent (Texas) to 31 percent (Israel).
Pathak et al. (1982) concluded that the distribution of melanomas among
New Mexican Anglos was compatible with the sunlight hypothesis. They observed
that the most common sites, the trunk in males and the lower extremities in
females, were affected by styles in dress and recreational activities, which
in turn influenced site-specific levels of exposure to the sun. Based on
cancer incidence data from Europe and North America, Crombie (1981) observed
that the lower limbs were the major site in females, whereas the trunk was the
most common among males. Crombie concluded that the sex differences in site
distribution corresponded in direction and magnitude to differences in
exposure associated with accepted dress styles among males and females and
that the observed melanoma pattern was not incompatible with the role of
sunlight as a major etiologic factor.
Some epidemiological studies have indicated an excess of upper extremity
lesions in females and an excess of head and neck lesions in males, but the
trends are not as consistent nor are the differences as significant as those
observed for the lower extremities in females and the trunk in males. Crombie
(1981) noted that head lesions were slightly more common among males than
females, while there was little difference for CMMs of the upper extremities.
Hinds and Ko.lonel (1980) observed that the incidence of head and neck
melanomas among 262 Caucasian Hawaiians was seven times lower among females
than among males. They concluded that the differences in site distribution by
sex were consistent with the sunlight hypothesis, suggesting that the lower
incidence of head and neck tumors among women was due to protection of the
scalp, ears, and neck by longer hair. There was a less than two-fold
difference in incidence of facial lesions, which was ascribed to the
potentially higher occupational exposure to the sun by males.
Race and Ethnic Background
Race and ethnic background also show pronounced effects on the
distribution of melanomas. The predominant difference is the higher
proportion of melanomas occurring on the feet and in some cases the hands
among darker-skinned populations compared to lighter-skinned populations.
Hinds (1979) noted that the anatomical distribution among blacks differed
from that of Caucasians with the largest proportion of melanoma lesions
occurring on the feet among blacks while among Caucasians there was a more
even distribution of CMMs over the entire body. Hinds pointed out that in two
separate studies, less than 5 percent of the lesions were observed on the feet
of Caucasians in Australia (both sexes) and less than 10 percent on the feet
of Caucasians in Norway. In a study of 31 blacks and an unspecified number of
Caucasians from the United States, Reintgen et al. (1983) observed that 60
percent of the lesions among blacks occurred on the feet whereas the dominant
site among whites was the trunk.
* * DRAFT FINAL » » *
-------
5-6
A similar predominance of melanoma of the feet has been observed in blacks
from both Africa and America. Kiryabwire et al. (1968) in a study of 152
melanoma cases from Uganda, observed that 64 percent occurred on the feet
(Table 5-1). Lewis (1967) noted that there were tribal and ethnic differences
in distribution even within the Ugandan study population. He suggested that
the high incidence of melanomas on the feet of Ugandans was closely related to
the high incidence of discrete pigmented areas on the sole, which he
characterized as genetically determined, potentially unstable melanocytes.
Lewis indicated that trauma (e.g., heat and wood smoke) could have played a
role in inducing malignant changes in these pigmented areas.
Collins (1984) found that among 43 cutaneous melanomas in Chinese
patients, 74 percent occurred at volar (palm) and subungual (under the nail)
sites. Fifty-six percent of the melanomas occurred on the foot with 83
percent of these on the plantar surfaces (Table 5-1). Collins observed that
the preponderance of melanomas on the feet and the pigmentation of the skin
among Chinese, were intermediate between those for whites and blacks. He also
noted that the annual incidence of melanoma occurring on plantar surfaces may
be similar in all racial groups although in some groups (i.e., blacks) the
percentage of lesions occurring on the feet may be higher.
In a study of 24 melanoma cases among Maoris and Polynesians, Moss (1984)
observed that over half of the melanomas were on the lower limbs (13 of 24),
with six of these occurring on the soles of the feet. Among 546 melanoma
patients from Japan, Takahashi (1983) observed that 45 percent of the
melanomas occurred on the hands and feet with particularly elevated incidence
on the thumb and big toe (Table 5-2).
Hinds (1982) observed that the high risk of melanoma on the sole of the
foot among blacks from Africa had led to speculation that trauma may be
involved. However, he noted that although incidence appeared to be higher
among blacks of African than American origin, the sole of the foot remained
predominant for both groups. In a review of the role of trauma in the
etiology of malignant melanoma, Briggs (1984) concluded that there was no
unequivocal evidence indicating that trauma played a role in the development
of the vast majority of melanomas. He cited information indicating no
difference in incidence on the foot between shoe-wearing and non-shoe-wearing
Africans. In addition, he pointed out that although the hand was one of the
most traumatized parts of the body, melanoma was not excessive on this site
compared to other sites.
Hinds (1979) suggested that trauma could still be a factor among
non-Caucasians other than blacks that developed melanoma on the feet due to
use of open-toed shoes and sandals. In addition, Hinds (1979) concluded that
exposure to sunlight was probably not an important risk factor for any site
among non-Caucasians. In support of this hypothesis, Magnus (1973) observed
that the foot was the only site not showing a north-south gradient in melanoma
incidence.
In a study of the anatomic site distribution of CMM among 262 Caucasians
and 66 non-Caucasians from Hawaii, Hinds and Kolonel (1980) observed a higher
proportion of melanomas of the lower extremities and a lower proportion of
* * * DRAFT FINAL * * *
-------
5-7
melanoma of the trunk and upper extremities among both male and female
non-Caucasians (Table 5-2). The highly significant differences between
Caucasian and non-Caucasian males were mostly due to the higher proportion of
melanomas on the lower extremities among non-Caucasian males. The difference
for females was not quite as large. The authors suggested that factors other
than solar exposure were responsible for the anatomic distribution of melanoma
among non-Caucasians and that trauma could play a role among certain ethnic
groups in Hawaii.
Other differences in site distribution have been observed between
Hispanics and Anglos. Based on a study of 495 melanoma cases in New Mexico,
Pathak et al. (1982) concluded that site-specific incidence rates differed
significantly between Anglos and Hispanics (Table 5-2). The site-specific,
age-adjusted incidence rates for Anglos from New Mexico were significantly
higher than for Hispanics from New Mexico at each site. Melanomas of the head
and neck were 10 times more frequent among Anglos than among Hispanics,
melanomas of the trunk 6.2 times more frequent, upper extremities 5.6 times
more frequent, and lower extremities 24.9 times more frequent. Lower
extremities and the trunk were the most common sites in Anglo females and
males, respectively, whereas the trunk was the most common site for
Hispanics. The authors noted that protection by skin pigmentation and
cultural factors were the most obvious explanations for the observed
differences.
In a study of 2,252 white, 30 non-white, and 206 Spanish-surnamed Texans
with CMM, MacDonald (1976) observed a higher proportion of lower extremity
melanomas among non-white males (54%), followed by Spanish-surnamed males
(30%) and then white males (24%) (Table 5-2). Only among whites were there
notable sex differences for the lower extremities. In contrast, melanomas on
the upper extremities were more frequent among male and female whites (18 and
24%, respectively) than among the other groups (11 and 18%, respectively among
Spanish-surnamed and 8 and 18%, respectively among non-whites). Among
non-whites, lesions on the head and neck were more frequent among males and
less frequent among females compared to the other two groups.
Anatomical Distribution and Age
Age differences appear to be associated with two general patterns of
melanoma incidence, one for melanomas of the face and the other for melanomas
of the neck-trunk and extremities. Some researchers have cautioned, however,
that such age-specific effects may be confounded by birth cohort effects.
Since both age and birth cohort may affect individual risk of melanoma,
analyses of changing incidence over time without consideration of potential
cohort effects may obscure real trends in the data. The discussion below
focuses on age-specific trends, particularly those observed after controlling
for birth cohort effects.
In a study of 5,108 melanoma cases in Norway, Magnus (1981) observed an
age-specific incidence pattern by 10-year birth cohorts for the face-neck,
with rates increasing rather slowly up to about 50 years and more rapidly
thereafter. For the trunk and lower extremities, however, age-specific
* * DRAFT FINAL
-------
5-8
incidence rates were observed to increase steeply up to about 40 years and
gradually level off or decrease thereafter. In an analysis of 2,501 melanoma
patients from Finland, Teppo et al. (1978) also observed different
age-specific incidence curves (by time of diagnosis) for different sites as
shown in Figure 5-1. .Melanomas of the trunk in both sexes and melanomas of
the lower extremities among women showed a relatively high incidence rate at
30-49 years and a leveling off of incidence thereafter. For head and neck
lesions in both sexes and lower extremity lesions in males, rates were lower
in middle-age and increased continuously with age. The age-specific curve for
the upper limbs was intermediate between the two other site groups. The
shapes of the curves were similar in three successive time periods (1953-1959,
1961-1970, 1971-1973). Excess trunk lesions among males and lower limb
lesions among females were apparent in almost all age groups.
Holman et al. (1980) showed that the incidence rate of invasive head and
neck melanomas among 542 Australians progressively increased after about age
40. Melanoma incidence rates for the lower extremities increased at 20-29
years among females and 30-39 years among males, peaked at about 50-59 years
in both sexes, and declined thereafter. The authors noted that the
age-specific pattern for upper and lower extremities and trunk would be
consistent with a birth cohort effect beginning with those born around 1915
[even though other researchers (e.g., Magnus 1973) had observed cohort effects
beginning as early as 1900].
In a study of incidence rates in Denmark, Houghton et al. (1980) also
observed differing age-specific melanoma incidence rates in 10-year birth
cohorts for melanomas of the face versus melanomas of the trunk-neck, and
upper and lower extremities. Figures 5-2 and 5-3 present the age-specific
incidence curves by anatomical location for Danish males and females,
respectively. For cases of facial melanoma, incidence was relatively low for
both males and females until age 60 when it rose rapidly. This age-specific
pattern was not seen for melanomas of the trunk-neck and upper and lower
extremities for males and females. For these sites, incidence rose from
adolescence to middle-age and generally plateaued thereafter. As a result,
the mean age for these sites for both sexes was lower than for melanomas of
the face. A comparison of the Danish age-specific melanoma incidence curves
for the lower extremities for males and females also reveals striking
differences. Not only do the curves for females indicate higher incidence
rates than for males for all three 10-year birth cohorts, but they also show
the rapid increase in age-specific incidence rates among the 1963-1972 birth
cohort group for melanomas of the lower extremities. The authors postulated
that the distinct age-specific patterns suggested two different mechanisms of
carcinogenesis with continued, prolonged effects of solar radiation implicated
in a majority of facial melanomas. In contrast, melanomas of the trunk and
lower extremities, including nodular and superficial types, were considered to
be related to short-term non-cumulative effects of solar radiation.
Lee (1982) similarly observed that the mean age at diagnosis of melanoma
was lower for the trunk and limbs than for the head and neck, but cautioned
that birth cohort effects could have accounted for differences in the
site-specific age distribution. Based on an analysis of incidence data from
DRAFT FINAL * * *
-------
5-9
FIGURE 5-1
TRUNCATED AGE-ADJUSTED INCIDENCE RATES (per 10s) OF
CUTANEOUS MELANOMA IN FOUR AGE GROUPS IN FINLAND
BY SEX, ANATOMICAL LOCATION, AND TIME OF DIAGNOSIS
0.01
0-29 »-« SO-H 70-
0-2* 30-0 S0-e» 70-
(WMO*
10:
0.1
LOWER LIMBS
p«MOs
10
0.1
UPPER LIMBS
0-2* X-4* U-«S 7O-
0-2* X-tt M-M 70-
p«f 10«
1:
0.1:
0.01
TRUNK
FEMALES
10J
0.1
10 q
0-2* 10-4* 50-6* 70-
• io5
LOWER LIMBS
0.01
HEAD AND NECK
10
0.1
0-J5 30-49 M-l* 70-
UPPER LIM8S
0-2* M-AI M-M 7O-
o-i* »-4» w-a 7o-
NOTE: Time of Diagnosis: 1=1953-1959, 11=1961-1970, 111=1971-1973
Source: Teppo et al. (1978).
- * * DRAFT FINAL * * *
-------
5-10
FIGURE 5-2
AGE-SPECIFIC INCIDENCE OF MELANOMA OF THE SKIN IN
DANISH MALES BY SITE (GROUPED BY YEAR OF DIAGNOSIS)
1*63-72
I0203040M4070W
AGE
Source: Houghton et al. (1980).
* * DRAFT FINAL
-------
5-11
FIGURE 5-3
AGE-SPECIFIC INCIDENCE OF MELANOMA OF THE SKIN IN
DANISH FEMALES BY SITE (GROUPED BY YEAR OF DIAGNOSIS)
30-
20-
z
UJ
40-
30-
20-
10-
UPPER EXTREMITY
20-
FACE
1963-72 ,
so-
40-
10-
TRUNK-NECK
LOWER
EXTREMITY
10 20 X 40 JO 40 70 10
10 20 » 40 » 60 70
AGE
Source: Houghton et al. (1980).
DRAFT FINAL
-------
5-12
five Nordic countries, Magnus (1977) observed two sets of age-specific
incidence curves by primary melanoma site. The curve for the face indicated
an exponential rise with age whereas for the neck-trunk and lower limbs, the
maximum incidence was reached at middle age. Magnus (1977) noted that the
neck-trunk classification was not uniform across countries. Sweden grouped
CMMs of the neck with those of the face and scalp, and Finland included CMMs
of the scalp in the neck classification.
In a study of 5,741 melanoma cases from Norway, Boyle et al. (1983)
observed a smooth rise (approximately linear on a log-log scale) in the
incidence of head and neck lesions with age for both sexes (assuming equal
cohort effects for each site). In contrast, the incidence rate for the trunk
and lower limbs declined in older age groups, particularly in females for whom
lower limb melanoma incidence remained virtually constant after age 40. The
authors suggested that the decline in incidence may have been associated with
reduced sunlight exposure with the passage of youth.
Histologic Type
Differences in the anatomical distribution of melanoma have also been
observed on the basis of histologic type. In a study of 542 Western
Australians with pre-invasive or invasive melanoma, Holman et al. (1980)
showed a predominance of Hutchinson's melanotic freckle (HMF) among
pre-invasive melanomas of the head and neck, sites which are usually exposed,
and a predominance of superficial spreading (SSM) and invasive malignant
melanomas (ICMM) among melanomas of the trunk in males and of the lower
extremities in females (Table 5-3).* These latter sites are usually not
exposed. In an analysis of 1973-1981 Surveillance Epidemology End Results
(SEER) data, Scotto (1985) also observed that HMF occurred predominantly on
the face, head and neck and that SSM occurred predominately on the trunk among
males and the lower extremities among females (Table 5-4). In a study of 510
melanoma patients in Texas, Smith (1976) noted that SSM was most common on the
lower extremities, and HMFM was most common on the head and neck.
Adler and Gaeta (1979) noted that SSM was more common on non-exposed skin,
and HMFM was more common on the face and other exposed sites. Among 477
melanoma patients from southeast Scotland, SSM and NM occurred most frequently
on the lower limb, with the distribution by histogenic type being similar in
each sex (Pondes et al. 1981). Among 29 volar and subungual melanomas in
Chinese patients, Collins (1984) observed that five of the six SSMs and the
two NMs occurred on the sole, while the sixth SSM occurred on the palm.
Geographical Area
Geographical area of residence and latitude have been noted in trends of
overall melanoma incidence, but only a few studies have investigated the
effects of those considerations on melanoma site distribution.
* HMF is the pre-invasive lesion of HMF melanoma (HMFM). It is also called
lentigo maligna.
» * * DRAFT FINAL * * *
-------
5-13
TABLE 5-3
PERCENTAGE DISTRIBUTION OF PRE-INVASIVE MELANOMA AND INVASIVE
MALIGNANT MELANOMA IN WESTERN AUSTRALIA BY BODY SITE3
Pre-Invasive Melanoma
b
HMF
Body Site
Head and Neck
Trunk
Upper Extremities
Lower Extremities
Cryptogenic Metastases
d
Males
(14)
79
7
14
0
-
Females
(14)
79
7
0
1
-
c
SSM
Males
(36)
17
39
27
17
-
Females
(54)
17
26
20
37
-
Malignant
Melanoma
Males
(208)
24
41
14
17
4
Females
(210)
18
18
26
37
1
Data on site were missing in 4 patients.
b
Hutchinson's melanotic freckle.
c
Superficial spreading melanoma, non-invasive.
d
Numbers in parentheses refer to numbers of patients.
Source: Holman et al. (1980).
* * * DRAFT FINAL
-------
TABLE 5-4
PERCENTAGE DISTRIBUTION OF MALIGNANT MELANOMA IN UNITED STATES
CAUCASIANS BY HISTOGENIC TYPE, SEX AND BODY SITE a/
Face, Head and Neck
Trunk
Upper Extremities
Lower Extremities
Other Sites
Ma les
62
18
11
8
1
b
HMF
Fema les
58
8
18
16
-
Males
10
27
26
36
1
c
SSM
Fema les'
17
26
20
37
-
Ma les
22
45
22
10
1
d
NM
Fema les
15
24
26
35
-
Ma les
20
38
18
11
13
e
NOS
Fema les
14
20
24
34
8
Other Ce 1
Ma les
31
25
17
14
13
f
1 Types
Fema les
34
20
20
26
10
Based on SEER data from Seattle, Detroit, Iowa, Utah, San Francisco/Okland, Atlanta, and New Mexico.
3
Hutchinson's melanotic freckle.
Superficial spreading melanoma, non-invasive.
J
Nodular melanoma.
Not otherwise specified.
f
For example, spindle cell and a melanotiz melanoma.
Source: Scotto (1985).
-------
5-15
Magnus (1973) observed some geographic variations by site in 2,541
melanoma patients from Norway. Figures 5-4 and 5-5 show that incidence by
site noticeably decreased from southern to northern parts of the country and
was clearly higher in the capital and provincial towns than in rural areas for
the neck-trunk and lower extremities. For facial melanomas, only the
northernmost part of the country exhibited somewhat lower incidence rates
compared to those in the southernmost part of the country. The urban-rural
difference for facial melanomas was less conclusive in that the face was a
relatively rare site in the capital. The foot was the only site which lacked
a definite north-south gradient in melanoma incidence. It was proposed that
the north-south gradient in incidence was due to the increasing amount and
intensity of sunlight, as well as the increasing temperature with decreasing
latitude; both of which would promote sunbathing to a greater extent in the
south than the north. Since the foot would rarely be exposed to the sun, the
absence of a gradient for that site also fits the sun exposure hypothesis.
Magnus noted that the high incidence of melanomas of the feet indicated the
possible importance of other etiological factors.
Comparison of sites among melanoma patients from Queensland living in
tropical and subtropical areas (Little et al. 1980) generally showed similar
site distributions in both males and females. For example, among males the
proportion of back lesions did not significantly vary from the tropics to the
subtropics. There were a few notable exceptions, however, as 19 percent of
the lesions among males and females in tropical areas (12 patients) were of
the arm in contrast to 9 percent in subtropics (26 patients). The leg was the
site of 16 percent of the tumors in the tropics (10 patients) and 27 percent
in the subtropics (78 patients).
Balch et al. (1982), in a comparative study of 676 melanoma patients from
Alabama and 1,110 from Australia, observed that although melanomas of the
trunk were the most common in both patient -series, they occurred in higher •
proportions among Australians (37 percent vs. 28 percent) (Table 5-1). The
occurrence of melanomas on t-he lower extremities was also more common among
Australians (33 percent vs. 23 percent) whereas head and neck lesions occurred
more frequently among the Alabama patients (27 percent vs. 14 percent). In
contrast, Magnus (1977) did not observe dissimilar melanoma site distributions
across five Nordic countries.
Scotto (1986) analyzed National Cancer Institute age-adjusted 1978-1981
melanoma incidence data for Caucasians from seven locations in the United
States (Seattle, Detroit, Iowa, Utah, San Francisco/Oakland, Atlanta, and New
Mexico). As shown in Figures 5-6 and 5-7, latitudinal trends (based on UV-B
measurements) were observed for the upper extremities and for the head and
neck among white males and females.
A common hypothesis has been forwarded by many researchers to explain the
observed trends in anatomical location. The hypothesis proposes that
differential exposure to sunlight due to patterns of attire and recreational
activities may play a role in determining the predominant sites of melanoma in
both males and females. This hypothesis has been cited to explain the higher
proportion of melanomas on the lower extremities among females and the trunk
* * DRAFT FINAL * * *
-------
5-16
FIGURE 5-4
TOTAL AGE-ADJUSTED INCIDENCE RATE OF MALIGNANT MELANOMA
OF THE SKIN IN NORWAY 1955-1970 BY REGION
AND ANATOMICAL SITE
Oslo
Eastern Southern Western
Norway Norway Norway la
- Northern
Norway
1 -
Lower limb (exd. fool)
Other and unspecified
MalesFemaiesM F M F
M F
M F M F
Source: Magnus (1973).
* * * DRAFT FINAL
-------
5-17
FIGURE 5-5
TOTAL AGE-ADJUSTED INCIDENCE RATE OF MALIGNANT MELANOMA OF
THE SKIN IN NORWAY 1955-1970 IN THE CAPITAL, PROVINCIAL
TOWNSa, AND RURAL AREAS BY ANATOMICAL SITE
Provincial Rural
towns areas
Other and unspecified
Males Females M F M F
Areas administratively classified as'urban excluding the capital,
Source: Magnus (1973).
* * DRAFT FINAL
-------
5-18
FIGURE 5-6
SKIN MELANOMA INCIDENCE BY UV RADIATION INDEX AMONG WHITE MALES
ACCORDING TO ANATOMICAL SITE (1978-1981 1
IN THE UNITED STATES
O T
O J
o
o
° 2
O
5 0.1
D
-e
95
OIT — 101
EATTLE
ETR
80
o
—I—
100
D
D
o
a
i
i
i
a
Jo 'S-
o
«
L
«n
f> T
SAN R
ATLANTA
150
D
m
7
o
3.
o
o
Legend
O
—I
200
THUHK o
UO/TDOT ^
0.1
SEATTLE — 95
DETROIT — 10
o
a
i
5
i
I/I
I
00
I
3
O
I
o
o
2 btgend
g • HtAP/ntg
z D
80
100
150
200
SOLAR ULTRAVIOLET (UVB) INDEX
SOLAR ULTRAVIOLET (UVB) INDEX
Source: Scotto and Fears (1986).
-------
5-19
FIGURE 5-7
SKIN MELANOMA INCIDENCE BY UV RADIATION INDEX AMONG WHITE FEMALES IN THE UNITED STATES
ACCORDING TO ANATOMICAL SITE (1978-19811
Q.
O
Q_
O
O
o
o"
o
a:
o
CD
t/>
|
Ul
o
<
2
<
d
$
o.i
D
D
Q.
O
Q.
O
O
O
o:
:D
o
t
E
f
Ol
T
i
3
o
z
0
0
Legend
I TRUNK
If UO/TOOT
UJ
.o
2
BO 100 150 200
SOLAR ULTRAVIOLET (UVB) INDEX
o
m
F
i
u> a
i!
D
i
3
o
z
O
0
H-
i
L«g«nd
• HtAP/MCCK
D MtK/HMO
80 100 150 200
SOLAR ULTRAVIOLET (UVB) INDEX
Source: Scotto and Fears (1986).
-------
5-20
among males. In addition, significant increases in melanoma incidence on the
trunk of males, and often on the trunk and lower extremities of females, have
been observed for birth cohorts born between 1900-1935 (see Chapter 6). These
trends have been attributed to the sunlight hypothesis.
The high proportion of melanomas on the feet of blacks and other
dark-skinned ethnic groups represents the only major deviation from this
postulated explanation. The potential damaging effect of trauma has been
cited as a possible explanation for this occurrence, although many have noted
that a predominance of melanomas on the feet has also been observed among
shoe-wearing blacks and an excess has not been noted on the hands, one of the
most traumatized parts of the body. Nevertheless, most agree that exposure to
sunlight is not involved in the development of melanomas on the feet.
TEMPORAL TRENDS IN SITE DISTRIBUTION
In a review of epidemiologica-l evidence, Elwood and Hislop (1982) noted
that incidence rates in Connecticut from 1955-1970 increased the most. for _the.
lower limb among females, the trunk among males, and the upper limbs in both
sexes, a pattern which was consistent with changes in general clothing habits
during that time.
In a study of melanoma incidence data from New South Wales, Australia from
1970 to 1976, McCarthy et al. (1980) observed that melanomas of the trunk
among males and of the lower limbs among females more than doubled from 1971
to 1976 (Figures 5-8 and 5-9). Melanoma incidence rates of the upper limbs
increased from 1971 to 1972-*, but from 1972 to 1976 no marked increases above
the 1972 rates were observed. Incidence of melanoma of the face and ear
increased markedly in males but only a slight increase was observed for these
sites in females. The authors argued that the trends in incidence supported
the hypothesis that sunlight exposure is the dominant cause of melanoma. They
cited trends in attire among the male and female populations as causes of
increased melanomas of the trunk among men and melanomas of the leg among
women. The presence of more hair and use of cosmetics among females were
cited as causes of lower incidence increases of lesions of the head and neck
for females than males. The authors also proposed that the steeper incidence
rise from 1971-1972 than for later years could have been related to major
sunspot activity in 1970. This hypothesis may be questionable since the
latency period of melanoma is generally considered to be longer than two
years. However, it is possible that sunlight, acting via a promotional
mechanism, could show such a short lag period.
A comparison of site-specific incidence among 3,289 melanoma patients from
Sweden between 1959 and 1968 indicated an overall annual increase of about 7
percent, with increases greater for females than for males (Malec and Eklund
1978). The greatest annual increase in percentages for both sexes was for
melanomas of the trunk (e.g., 11 percent for males) whereas the smallest was
for the arm in males (4 percent) and the head-neck in females (4 percent).
The correlation between annual incidence rate and year (from 1959-1968) was
highly significant (p§0.001) for the trunk and head-neck in males and the
trunk in females, and moderately significant (p§0.05) for the lower limbs in
females.
* * * DRAFT FINAL
-------
5-21
FIGURE 5-8
ANNUAL MELANOMA INCIDENCE RATES FOR
MALES BY ANATOMICAL SITE
(New South Wales, Australia: 1970-1976)
to -
s
i.
Source: McCarthy et al. (1980).
* * * DRAFT FINAL
-------
5-22
FIGURE 5-9
ANNUAL MELANOMA INCIDENCE RATES FOR
FEMALES BY ANATOMICAL SITE
(New South Wales, Australia: 1970-1976)
3
i
I .0
30-
Source: McCarthy et al. (1980).
DRAFT FINAL * » *
-------
5-23
Balch et al. (1983) observed a significant increase in melanomas of the
trunk and a significant decrease in melanomas of the head and neck among males
between 1955 and 1980 for 1,647 Stage I patients from Alabama and New South
Wales. No significant changes in site distribution were observed for the
extremities in males or any site in females.
Hinds and Kolonel (1980) conducted a study on 333 melanoma cases from
Hawaii from 1960-1977. Among the 262 Hawaiian whites, site-specific
age-adjusted incidence rates increased in both sexes except for fairly stable
rates for lower extremities in females. By 1972-1977 as compared to
1960-1971, the incidence rate for lower extremities in males had increased to
equal that in females. For CMM of the head-neck and the trunk, incidence
rates for males were consistently higher than rates for females. The authors
concluded that the apparently stable incidence rate of lower extremity lesions
among females from 1960-1977 could not be ascribed to the lack of a trend
towards more revealing clothing. They postulated that the stable rate in
females suggested that a maximum rate of melanoma induced by sunlight had been
reached and that the similar rates between the sexes by 1977 was due to
attainment of a similar maximum in males.
Magnus (1981) investigated trends in 5,108 melanoma cases from Norway
between 1955-1977. Overall incidence increased by about seven percent
annually for both sexes, with a doubling of melanomas of the face-neck and a
five-fold increase of melanomas of the trunk and lower limbs. An anlysis of
the data in two time periods (1955-1970 and 1970-1977) indicated a reduction
in annual percentage incidence increases for melanomas of the trunk among
males (from 13.5 percent to 8.4 percent) and an increase in annual percentage
incidence increases .for females (4.5 percent to 12.5 percent). This resulted
in a more similar pattern of annual incidence trends for the two sexes during
1970-1977 than 1955-1970. Magnus suggested that this equalizing trend between
the sexes may have reflected the increased similiarity of sunlight exposure to
various parts of the body.
Teppo et al. (1978) analyzed 2,501 melanoma cases in Finland between 1953
and 1973, and observed that the incidence of melanomas of the trunk in both
sexes and melanomas of the lower limbs in females increased markedly over this
period. Over 80 percent of the total incidence increase in males was
attributed to lesions of the trunk, while 44 percent and 28 percent of the
total increase in females was attributed to the lower limbs and the trunk,
respectively. Only minor incidence changes were observed for the head and
neck from 1953-1973. the authors concluded that the site-specific incidence
increases could be interpreted as a cohort effect and accounted for by changes
in clothing habits which would have increased sunlight exposure to these sites.
SEASONAL VARIATIONS
Seasonal variations in site-specific CMM have also been investigated.
Some seasonal trends have been observed for melanomas of the extremities,
particularly among females, but the scarcity of available epidemiologic
evidence limits the ability to draw broad conclusions from these data.
* * * DRAFT FINAL * * *
-------
5-24
Malec and Eklund (1978) examined the association between site-specific
melanoma incidence and season (summer versus winter) for 3,289 melanoma cases
from Sweden. They observed a highly significant increase (p§0.001) of
melanomas on the lower extremities of females during summer months
(June-August) compared to winter months (December-February). No corresponding
seasonal increase was observed for males. Given the apparent predominance of
SSM of the leg among females, and the possibility that the vertical growth
phase may become clinically apparent in a few weeks or months, the authors
suggested that the seasonal rise in melanoma of the lower extremities among
females could have been due to an increase in SSM and a result of short-term
ultraviolet radiation exposure.
Scotto and Nam (1980) tested for monthly patterns of melanoma incidence
among 2,168 melanoma cases from nine locations in the United States during
1969-1971. They applied a first-order sine wave model to the data that
estimated the frequency of cases diagnosed in 1-month intervals. As shown in
Table 5-5, they observed a strong seasonal pattern of incidence with
summertime peak for upper and lower extremities among females (p=0.0001 and
0.0007 for lower and upper extremities, respectively) and a nonsignificant
summertime peak for upper extremities among males (p=0.11). When incidence
trends for melanomas of the upper extremities were analyzed by two-month
periods, statistical significance was greater for females (p=0.003) than for
males (p=0.07). The lower significance for males was possibly due to the
small number of cases. Analysis of seasonal pattern by age (^55, >55
years) indicated significant trends among females less than 55 years of age
for lower extremities, and upper and lower extremities pooled, but not for the
trunk or face and head lesions. The authors concluded that it was difficult
to determine if seasonal melanoma patterns resulted from the potential
promoting effects of UVB exposure or from enhanced recognition during summer
months.
Hinds et al. (1981) analyzed the seasonal pattern of site-specific
melanoma in 351 Hawaiians during 1960-1978 also using a sine wave model.
Since significant differences were not observed between male and female
site-specific diagnosis of melanoma (divided into two-month periods), the
authors applied the model to male and female cases combined. The sine wave
model was consistent with (i.e., not rejected) a peak in summer months for all
sites and for the four site subgroups (head-neck, upper extremities, lower
extremities, and trunk). The amplitude of the sine wave model was, however,
significantly different from zero (p<0.02) only for all cutaneous melanomas
and melanoma of the lower extremities. The authors concluded that their
findings supported the hypothesis that solar radiation may have been a factor
in promoting the short-term growth of cutaneous malignant melanoma.
The anatomical distribution of melanomas among lighter-skinned populations
appears to be influenced by variations in exposure to sunlight. Furthermore,
interpretation of epidemiological evidence from ecological studies to a
certain extent indicates that two mechanisms of carcinogenesis may be
involved, in which prolonged cumulative exposure to the sun may be associated
with facial melanomas and short-term acute exposure may be associated with
melanomas of the trunk and lower extremities.
» * DRAFT FINAL
-------
TABLE 5-5
SEASONAL PATTERNS OF SKIN MELANOMAS AND EDWARDS' TEST RESULTS,
THIRD NATIONAL CANCER SURVEY, 1969-1971
Number
Sex and
Ma les
Face,
Upper
Trunk
Lower
Fema les
Face,
Upper
Trunk
Lower
Anatomic
Head and
Extremi t
Extremi t
Head and
Ext rem i t
Extremi t
Si te
Neck
ies
ies
Neck
i es
i es
Jan
39
17
32
9
15
18
20
26
Feb
19
11
27
14
13
20
17
29
Mar
20
24
37
11
18
13
22
49
Apr
27
14
32
16
20
22
18
37
of Cases by
May
31
20
27
8
23
31
20
37
Jun
18
22
35
12
20
33
23
46
Month
Jul
12
23
34
15
13
39
26
40
of Diagnosis
Aug
18
19
34
9
16
34
18
40
Sept
23
22
32
9
18
21
26
33
Oct
23
16
27
8
19
19
26
25
Nov
21
14
34
9
13
32
17
27
Dec
16
15
32
11
12
22
22
21
2
Edwards' test X (2.d.f)
*
4.50 (p
0.12 (p
2.06 (p
3.02 (p
14.42 (p
1.53 (p
18.19 (p
*
= 0.11)
= 0.94)
= 0.36)
= 0.22)
= 0.0007]
= 0.47)
= 0.0001 ]
(p-va lue)
Cn
1 '
Ln
The data for male face, head and neck depart significantly from the sine curve model
thus Edwards' method for testing amplitude is considered inappropriate.
Source: Scotto and Nam (1980).
by chi-square goodness-of-fit test (p<0.05)
-------
5-26
RELATION OF SITE DISTRIBUTION TO PROGNOSIS
The results of several epidemiological studies indicate that the primary
melanoma site may be related to prognosis. Significantly poorer survival
rates were observed among those melanoma patients from south-east Scotland
(both sexes combined) with lesions of the trunk, with 5-year survival rates
(based on deaths due to melanoma) of 75 percent for upper extremities, 74
percent for the head and neck, 72 percent for lower extremities, and only 49
percent for the trunk (Pondes et al. 1981). Among females, 5-year survival
rates were 86 percent for melanoma of the head and neck, 76 percent for lower
extremities, 79 percent for upper extremities, and only 45 percent for the
trunk. Among males site-specific prognosis was not as favorable with 5-year
survival rates of 67 percent for the upper extremities, 59 percent for the
lower extremities, 54 percent for the head and neck, and 54 percent for the
trunk. The sex difference in survival- was due mainly to the poorer survival
of males older than 50 years.
Lemish et al. (1983) also observed that 5-year survival rates among 401
Australians with invasive malignant melanoma from 1975-1976 were affected by
the primary site of melanoma. However, lower survival rates were not
associated with melanomas of the trunk among these patients. Table 5-6
indicates that five-year relative survival rates for both sexes combined were
80 percent for the head and neck, 88 percent for upper extremities, 90 percent
for the trunk, and 91 percent for lower extremities. Among females, the
lowest survival rates were for the head and neck (77 percent) followed by
upper extremities (86 percent), the trunk (88 percent), and lower extremities
(97 percent). Among males, 5-year survival rates were the lowest for the
lower extremities (77 percent) and highest for the upper extremities and trunk
(93 percent and 90 percent, respectively).
In an investigation of the influence of 11 clinical and prognostic factors
on survival rates among 776 melanoma patients from Australia and 293 melanoma
patients from Alabama, Balch et al. (1982) observed in a multivariate analysis
(stage of disease and other tumor characteristics were controlled for) that
anatomic location of the lesion was a significant factor associated with
survival (p<0.00001) among the Australian patients (p<0.09 for Alabama
patients). Survival rates for the patients from Australia and from Alabama
were, however, not significantly different. The distribution of melanomas
differed between the two patient series, with melanomas of the lower
extremities and the trunk present in higher proportions among the Australians
and melanomas of the head and neck higher among the Alabama patients.
Cascinelli et al. (1985) also observed that primary lesion site was
significantly related to prognosis (p<0.05) based on 686 melanoma patients in
the World Health Organization melanoma register from 1967-1974.
FINDINGS
The most pronounced differences in site distribution of melanomas are
associated with sex, race, and birth cohort. Other potentially influential
factors include age, histogenic tumor type, geographical location, and season
of diagnosis. The following major findings can be drawn from the
epidemiological data presented in this chapter:
» DRAFT FINAL * * *
-------
5-27
TABLE- 5-6
CRUDE AND RELATIVE FIVE-YEAR SURVIVAL RATES BY
PRIMARY SITE OF INVASIVE MALIGNANT MELANOMA IN
WESTERN AUSTRALIA, 1975-1976
Site
Head and Neck
Trunk
Upper Extremities
.Lower Extremities
Five-Year Survival Rate (%)
Men
a
(87) 66
(120) 83
(83) 86
(111) 72
Crude
Women
72
83
81
93
b
Relative
Total
69
83
83
87
Men
79
90
93
77
Women
77
88
86
97
Total
80
90
88
91.
Numbers of cases in parentheses.
Relative to the survival rate expected on the basis on sex-
and age-specific mortality rates in the normal Western Australian
population.
Source: Lemish et al. (1983).
DRAFT FINAL * * *
-------
5-28
5.1 Cutaneous malignant melanomas of the lower extremities
occur about two times more frequently among females
than among males. Melanomas of the trunk occur about
1.5 times more frequently among males than among
females. Although these sites (the lower extremities
and the trunk) are relatively less exposed than the
head, face and neck and the upper extremities, most
researchers have concluded that the observed
differences by sex are still compatible with the
hypothesis that sunlight exposure is involved in the
development of CMM among Caucasians.
5.2 The major difference in melanoma site distribution
between different racial and ethnic groups is the
higher proportion of melanomas occurring on the feet
and in some cases the hands among darker-skinned
populations (blacks and orientals) compared to
lighter-skinned populations (whites). The available
epidemiological evidence suggests that factors other
than sunlight exposure are associated with the anatomic
distribution of melanomas among dark-skinned
populations. The annual incidence of melanoma
occurring on plantar surfaces may, however, be similar
in many racial and ethnic groups (Collins 1984).
5.3 The anatomical distribution of CMM has been observed to
vary according to histologic type, with Hutchinson's
melanotic freckle melanoma (HMFM) more commonly
recurring on regularly exposed sites such as the head
and neck and superficial spreading melanoma (SSM)
occurring more frequently on regularly unexposed sites
such as the trunk and the lower extremities.
* * * DRAFT FINAL *
-------
5-29
REFERENCES
Adler, S. and Gaeta, J.F. Malignant melanoma. Chapter 12, In: Cancer
Dermatology F. Helm (ed). Philadelphia: Lea and Febiger pp 141-157 (1979).
Anaise, D., Steinitz, R., and Ben Hur, N. Solar radiation: A possible
etiological factor in malignant melanoma in Israel; A retrospective study
(1960-1972). Cancer 42:299-304 (1978).
Balch, C.M., Soong, S., Milton, G.W., Shaw, H.M., McGovern, V.J., McCarthy,
W.H., Murad, T.M. and Maddox, W.A. Changing trends in cutaneous melanoma over
a quarter century in Alabama, USA, and New South Wales, Australia. Cancer
52:1748-1753 (1983).
Balch, C.M., Soong, S., Milton, G.W., Shaw, H.M., McGovern, V.J., McCarthy,
W.H., Murad, T.M. and Maddox, W.A. A comparison of prognostic factors and
surgical results in 1,786 patients with localized (Stage I) melanoma treated
in Alabama, USA, and New South Wales, Australia. Ann Surg 6:677-684 (1982).
Balch, C.M., Karakousis, C., Mettlin, C., Natarajan, N., Donegan, W.L., Smart,
C.R., and Murphy, G.P. Management of cutaneous melanoma in the United
States. Surg Gyn Obstetrics 158:311-318 (1984).
Boyle, P., Day, N.E. and Magnus, K. Mathematical modelling of malignant
melanoma trends in Norway, 1953-1978. Am J Epidemiol 118:887-896 (1983).
Briggs, J.C. The role of trauma in the aetiology of malignant melanoma: A
review article. Brit J Plastic Surg 37:514-516 (1984).
Cascinelli, N., Marabini, E., Morabito, A., and Bufalino, R. Prognostic
factors for stage I melanoma of the skin: A review. Statistics in Medicine
4:265-278 (1985).
Collins, R.J. Melanoma in the Chinese of Hong Kong: Emphasis on volar and '
subungual sites. Cancer 54:1482-1488 (1984).
Crombie, I.K. Distribution of malignant melanoma on the body surface.
British Journal of Cancer 43:842-849 (1981).
Elwood, J.M., and Hislop, T.G. Solar radiation in the etiology of cutaneous
malignant melanoma in Caucasians. Natl Cancer Inst Monogr 62:167-71 (1982).
Hinds, M.W. Anatomic distribution of malignant melanoma of the skin among
non-Caucasians in Hawaii. Brit J Cancer 40:497-499 (1979).
Hinds, M.W. Nonsolar factors in the etiology of malignant melanoma. Natl
Cancer Inst Monogr 62:173-178 (1982).
Hinds, M.W. and Kolonel, L.N. Malignant melanoma of the skin in Hawaii,
1960-1977. Cancer 45:811-817 (1980).
* * * DRAFT FINAL
-------
5-30
Hinds, M.W., Lee, J. and Kolonel, L.N. Seasonal patterns of skin melanoma
incidence in Hawaii. Am J Publ Hlth 71:496-499 (1981).
Holman, C.D.J., Mulroney, C.D., and Armstrong, B.K. Epidemiology of
pre-invasive and invasive malignant melanoma in Western Australia. Brit J
Cancer 25:317-323 (1980).
Houghton, A., Flannery, J., and Viola, M.V. Malignant melanoma in Connecticut
and Denmark. Am J Cancer 25:95-104 (1980).
Kiryabwire J.W., Lewis, M.G., Ziegler, J.L., and Loefler, I. Malignant
melanoma in Uganda. East Afr Med J 45:498-507 (1968).
Lee, J.A.H. Melanoma and exposure to sunlight. Epidemiologic Reviews
4:110-136 (1982).
Lemish, W.M., Heenan, P.J., Holman, C.D.J., and Armstrong, B.K. Survival from
preinvasive and invasive malignant melanoma in Western Australia. Cancer
52:580-85 (1983).
Lewis, M.G. Malignant melanoma in Uganda: The relationship between
ligmentation and malignant melanoma on the soles of the feet. Brit J Cancer
21:483-495 (1967).
Little, J.H., Holt, J., and Davis, N. Changing epidemiology of malignant
melanoma in Queensland. Med J Aust 1:66-69 (1980).
MacDonald, E.J. Incidence and epidemiology of melanoma in Texas. In:
Neoplasms and Malignant Melanoma. Chicago: Year Book Medical Publishers,
Inc. pp 279 (1976).
Magnus, K. Incidence of malignant melanoma of the skin in Norway, 1955-1970.
Variations in time and space and solar radiation. Cancer 32:1275-1286 (1973).
Magnus, K. Incidence of malignant melanoma of the skin in the five Nordic
countries: Significance of solar radiation. Int J Cancer 20:477-485 (1977).
Magnus, K. Habits of sun exposure and risk of malignant melanoma: An
analysis of incidence rates in Norway 1955-1977 by cohort, sex, age, and
primary tumor site. Cancer 48:2329-2335 (1981).
Malec, E. and Eklund, G. The changing incidence of malignant melanoma of the
skin in Sweden, 1959-1968. Scand J Plast Reconstr Surg 12:19-27 (1978).
McCarthy, W.H., Black, A.L., and Milton, G.W. Melanoma in New South Wales:
An epidemiologic survey 1970-76. Cancer 46:427-432 (1980).
Moss, A.L.H. Malignant melanoma in Maoris and Polynesians in New Zealand.
Brit J Plastic Surg 37:73-75 (1984).
DRAFT FINAL * * *
-------
5-31
Movshovitz, M. and Modan, B. Role of sun exposure in the etiology of
malignant melanoma: Epidemiologic inference. J Natl Cancer Inst 51:777-779
(1973).
Pathak, D.R., Samet, J.M., Howard, C.A., and Key, C.R. Malignant melanoma of
the skin in New Mexico 1969-1977. Cancer 50:1440-46 (1982).
Pondes, S., Hunter, J.A.A., White, H., Mclntyre, M.A., and Prescott, R.J.
Cutaneous malignant melanoma in South-East Scotland. Quart J Med L
197:103-21 (1981).
Reintgen, D.S., McCarty, K.S., Cox, E., and Seigler, H.F. Malignant melanoma
in the American Black. Current Surgery:215-217 May-June (1983).
Scotto, J. and Nam, J. Skin melanoma and seasonal patterns. Am J Epidemiol
111:309-314 (1980).
Scotto, J. Melanoma among Caucasians in the United States. Skin Cancer Found
J 1:38-39 (1985).
Scotto, J., and Fears, T.R. The association of solar untraviolet radiation and
skin melanoma among Caucasians in the United States. Cancer Investigations (in
press 1986)
Smith, J.L. Histopathology and biologic behavior of malignant melanoma. In:
Neoplasms of the Skin and Malignant Melanoma. Chicago:Yearbook Medical
Publishers, Inc. pp 293-330 (1976).
Stevens, R.G., and Moolgavkar, S.H. Malignant melanoma: Dependence of
site-specific risk on age. Am J Epidemiol 119:890-895 (1984).
Takahashi, M. and Seiji, M. Acral melanoma in Japan. In: Pigment cell.
Vol. 6. Malignant melanoma. MacKie, R.M. (ed). Karger, Basel pp 150-165
(1983).
Teppo, L., Pakkanen, M., and Hakulinen, T. Sunlight as a risk factor of
malignant melanoma of the skin. Cancer 41:2018-2027 (1978).
* * * DRAFT FINAL * *
-------
CHAPTER 6
GEOGRAPHIC DISTRIBUTION
The hypothesis that ultraviolet radiation is causally associated with the
development of cutaneous malignant melanoma has been explored by examining
potential exposure to sunlight. Most of the studies are ecological studies
which sought to find correlations between incidence or mortality rates of CMM
and the potential for sun exposure estimated by proximity to the equator
(latitude) or by other surrogates for average exposures to sunlight or UV
radiation, such as number of sunlight hours at residence.
This chapter presents results from analyses of the geographic distribution
of CMM and includes both incidence and mortality studies. These ecological or
geographic correlation studies are subject to many assumptions, thus results
must be interpreted in light of the many limitations inherent in the study
design. For example, the failure to measure individual exposure to sunlight
in the study population makes it impossible to establish a causal association
between CMM and solar radiation; yet significant associations between
surrogate estimates of UV exposure and CMM have provided indications of causal
associations which in turn led to studies in which individual exposure has
been measured (these studies are discussed in Chapter 8).
The earliest epidemiological evidence that solar radiation might play a
role in the etiology of cutaneous malignant melanoma resulted from ecological
studies which showed a negative correlation between latitude of residence and
incidence and mortality rates for CMM in white populations. Using Australian
mortality data, Lancaster (1956) showed that the distribution of CMM among
"relatively fair-skinned types" was associated with proximity to the equator.
Among the states of Australia, he found a north to south decreasing gradient
in death rates (1951-53) due to melanoma with the crude death rates from
melanoma in Queensland almost three times higher than those in Tasmania and
Victoria, and a decreasing gradient of rates in the intervening states with
increasing latitude.
In the same study, Lancaster (1956) examined melanoma mortality rates in
other predominantly Caucasian populations and found that rates tended to be
higher in populations living closer to the equator. Mortality from melanoma
was higher on the northern island of New Zealand than on the southern; rates
in the British Isles and Europe were generally lower than those in Australia
and New Zealand; and rates in Canada were below those in the United States.
Within Europe, however, no latitude gradient was found when CMM mortality
rates from the same years were compared. In fact, Norway and Sweden reported
higher rates than any other European countries. Lancaster (1956) attributed
this absence of a gradient to probable differences in medical certification of
death, differences in statistical coding practices between countries, and the
fact that melanoma had only recently been separated from carcinomas of the
skin. However, studies cited in Chapters 7 and 8 indicate that this lack of
gradient may be due to other factors such as differences in ethnicity or
pigmentation characteristics. Within the United States (grouped into eight
areas), mortality from CMM (1949-52) was somewhat higher in southern than in
* * * DRAFT FINAL * * *
-------
6-2
northern areas and the coastal areas had higher CMM mortality rates than the
mountain or central areas. Lancaster (1956) linked the general latitude
association to CMM mortality with the variations in sunlight and UVR, and
concluded that the distribution of melanoma among fair-skinned populations was
consistent with a hypothesis of excess sunlight as an important predisposing
cause of CMM.
Holman et al. (1980a) examined Australian mortality rates from cutaneous
malignant melanoma by state in successive 5-year periods from 1931-34 through
1975-77 and determined that melanoma mortality rates were highest in
Queensland (the northern part of Australia) and diminished on a gradient from
north to south with the lowest rates in Victoria and Tasmania. This
relationship of increasing melanoma mortality with decreasing latitude was
generally maintained for the six states of Australia over roughly 45 'years,
even though CMM mortality rates increased steadily over this time period.
An analysis of melanoma incidence rates (1975-1976) in Western Australia
(Holman et al. 1980b) showed that the highest rates of CMM occurred in areas-
which received the fewest hours of bright sunlight per day, although each area
received a considerable amount of sunlight each day. Stratifying on the five
statistical divisions in the state of Western Australia, the highest rates
were found in Perth and the southwestern region which tend to be coastal and
more urbanized. Although tropical, the northern areas are less populated and
residents may be more likely to avoid sun exposure in the middle of the day
and wear protective clothing than in the more urbanized southwestern coastal
areas where residents are more likely to sunbath and engage in outdoor
recreational activities.
Herron (1969) analyzed data on 1,100 melanoma patients in Queensland,
Australia (1963-69) to investigate the geographical distribution of CMM in
this state and found little difference between northern, central and southern
Queensland (rates per 1,000 = 0.63, 0.51 and 0.71, respectively). There were,
however, higher rates in residents of coastal areas than in those from inland
areas, suggesting that this may be related to sun exposure.
CMM incidence rates in Queensland, Australia, were also examined for
geographical differences by Green and Siskind (1983). All incident CMM cases
over a 12-month period (7/79-6/80) were analyzed by statistical division and
no association with latitude was found. Like Herron (1969) and Holman et al.
(1980b), they found a significantly increased incidence of melanoma in the
coastal areas compared to the inland regions. Green and Siskind (1983) stated
that a latitudinal relationship with CMM incidence may have been absent due to
increased summer cloud cover in North Queensland. In addition, the mapped
contours of erythemal UV doses during the summer months deviate substantially
from latitude circles and, in summer, the critical recreational portion of the
year, there is no association of UV radiation with latitude.
Fears et al. (1976) plotted U.S. melanoma and non-melanoma skin cancer
incidence rates (Third National Cancer Survey 1969-1971) and U.S. CMM
mortality rates (1950-69 by county) for white males and females against
latitude and found that incidence, and mortality rates due to melanoma and
non-melanoma skin cancers increased with decreasing latitude. Regression
* * * DRAFT FINAL * * *
-------
6-3
analyses produced strong negative correlation coefficients for each parameter
(see Table 6-1).
Haenszel (1963) found a similar increasing north-south gradient in the
incidence of malignant melanoma in four northern and four southern U.S. cities
based on a 1947 survey of hospitals and physicians' offices. The ratio of
north/south incidence rates showed higher risks of CMM in the southern cities
than in the northern cities, although the ratio was somewhat lower for CMM of
the trunk or lower extremities in males. The north to south incidence ratios
were generally similar to those presented in a report for basal cell carcinoma
and squamous cell carcinomas.
When the 1973-1976 incidence data from the National Cancer Institutes
(NCI) Surveillance Epidemiology and End-Results (SEER) program were plotted
against the 1977-1978 NCI R-B meter measurements of accumulated dose in eight
locations in the United States, a strong positive association was noted
(Scotto et al. 1982) as shown in Figure 6-1.
Elwood and Lee (1974) examined age-standardized (1960 U.S. population)
mortality rates of cutaneous malignant melanoma and other (non-melanoma) skin
tumors separately over the period 1950-1967 for each Canadian province and
U.S. state. In both analyses (melanoma and other skin tumors), the mortality
rates showed a strong negative correlation with geographic latitude based on
the latitude of the largest city in each province or state. Estimates of
annual ultraviolet radiation in the erythema-producing wavelengths
were made for each province and state. These estimates showed a strong
negative association with latitude and with CMM mortality rates. The results
were similar for analyses of melanoma mortality and for other skin cancer
mortality; in addition, the estimates of UV flux in the erythema producing
wavelengths showed a similar association with CMM mortality as with latitude
alone. The authors concluded that the similar relationships of mortality
rates due to CMM and other skin cancers to estimates of annual UVR suggest the
involvement of ultraviolet radiation as. the causal agent in both diseases, and
that latitude is a major factor affecting CMM mortality rates.
Baker-Blocker (1980) failed to find a significant correlation between CMM
mortality in U.S. white males and females and amount of ultraviolet radiation
received in the area. Correlation of average annual UV radiation estimates
from 18 counties with CMM mortality rates by county for 1950-1969 (Mason and
McKay 1974) showed no association, although the estimates of ultraviolet
radiation were significantly correlated with latitude (p=0.01). El Paso
County (Texas) had the lowest melanoma mortality rates (1.2 and 1.1. per
100,000 for white males and females, respectively) even though it received the
highest amount of ultraviolet radiation. The counties with the highest
melanoma mortality rates (Leon County, Florida -- 2.9/100,000 for females,
Tarrant County, Texas -- 1.9/100,000 for females and 2.5/100,000 for males and
Nassau County, New York -- 2.3/100,000 for males) received considerably less
UV radiation than El Paso County. Bernalillo County (New Mexico) also had low
CMM mortality (1.2/100,000 for both sexes) but received more ultraviolet
radiation than those counties with the highest melanoma mortality rates.
These results were interpreted as evidence that CMM mortality may be due to
factors other than ultraviolet radiation, and that the high rates in Nassau
* *'* DRAFT FINAL * * *
-------
6-4
TABLE 6-1
SUMMARY STATISTICS FOR REGRESSIONS OF U.S.
SKIN CANCER INCIDENCE AND MORTALITY ON LATITUDE:
1969-71 INCIDENCE RATES (THIRD NATIONAL CANCER SURVEY)
AND 1950-69 MORTALITY RATES (MASON AND McKAY, 1974)
Males Females
Correlation Regression Correlation Regression
Coefficient" Slope + SD Coefficient* Slope + SD
Non-Melanoma Incidence -0.89 -0.037+0.013 -0.83 -0.033+0.016
Melanoma Incidence -0.86 -0.031+0.007 -0.83 -0.028+0.007
Melanoma Mortality -0.81 -0.017+0.002 -0.71 -0.014+0.002
'v Simple correlation coefficient of log rate and latitude.
Source: Fears et al. (1976).
* DRAFT FINAL - » *
-------
6-5
FIGURE 6-1
ANNUAL AGE-ADJUSTED INCIDENCE RATES FOR CMM
(SEER DATA 1973-1976) AMONG WHITE FEMALES
(OPEN SYMBOLS) AND MALES (CLOSED SYMBOLS),
ACCORDING TO ONE-YEAR'S UV MEASUREMENTS
IN SELECTED AREAS OF U.S.1
< cc
I- 01
a.
t/j Z
o 5
01 o.
§8
3 °.
a 01
< o.
100
Skin Melanoma
——— White Females
White Males
Detroit
_L
J_
£ "• - i
« = m ~
5 $8 <
I i l_
o
v
2
120 140 160
SOLAR UV RADIATION INDEX
180
200
1 The UV radiation index is the total Robertson-Berger meter counts over
-4
a one-year period multiplied by 10 . The meters read UV-B between 290 nm
and 320 nm, as well as some UV-A.
Source: NAS 1982.
DRAFT FINAL
-------
6-6
and Ulster Counties (New York) and Montgomery County (Maryland) suggest that
urbanization may play a role. However, the ethnicity and pigmentation
characteristics of the populations were not considered in this analysis. The
counties with the lowest rates have high proportions of Hispanics and Indians
among their white populations in contrast to counties such as Nassau, Ulster,
and Montgomery; in addition, the differences in CMM rates could reflect the
varying susceptibility of the populations.
Anaise et al. (1978) found similar results of high CMM incidence in
coastal versus inland areas in Israel. An analysis of all incident CMM cases
in the total Israeli Jewish population during 1960-1972 showed higher
age-adjusted rates of melanoma (3.5 and 3.2 per 100,000) for two coastal
cities (Haifa and Tel Aviv) than for Jerusalem (2.0 per 100,000) which is
situated inland in the mountains. This difference may be consistent 'with the
sunlight exposure hypothesis of etiology, since those in the coastal regions
are likely to spend more leisure time in outdoor activities and on the beach,
and wear clothes which expose more skin. No analysis of latitude was done in
this study.
Melanoma incidence data from 14 health regions in England and Wales
(1962-1970) showed a significant negative correlation with latitude for both
males (p<0.0001) and females (p<0.05) (Swerdlow 1979). In addition,
Swerdlow (1979) analyzed mean daily hours of sunshine (1960-1968) for each
region and found a positive correlation between melanoma incidence and mean
hours of sunshine, although the correlation was statistically significant
(p<0.05) in women only. The author concluded that these findings suggest
that exposure to sunshine is an important causal factor for cutaneous
malignant melanoma. He also postulated that the higher melanoma incidence
rates and stronger correlation with sunshine for females may be due to greater
skin exposure to sunlight or sunburn due to sunbathing and the style of
clothing.
An examination of geographical variation in Nowegian CMM incidence rates
during 1955-1970 (Magnus 1973) also showed a marked north-south increasing
gradient, with the age-adjusted incidence in southern Norway being two to
three times higher than the rate in the northern part of the country.
An analysis of melanoma incidence from 1953-1973 in Finland (Teppo et al.
1978) also showed a distinct north-south increasing gradient with the
age-adjusted (1950 world population) rates being higher in the south.
However, when the rates were adjusted for the urban/rural differences in
population residence, the north-south gradient almost disappeared because of
the high rates of melanoma in the urban areas of southern Finland. While this
observation suggests that factors other than latitude may be causally
associated with malignant melanoma, this finding does not negate a sunlight
hypothesis because people in urban areas in Finland are thought to experience
more exposure to the sun through leisure activities and holidays than those
living in rural areas where skin has traditionally been more protected from
direct sunlight by clothing (Teppo et al. 1978).
Eklund and Malec (1978) used data from the Swedish Cancer Registry during
the period 1959-1968 to investigate association of latitude with CMM incidence
* * * DRAFT FINAL - •• »
-------
6-7
in Sweden. They found a negative correlation between incidence rates and
latitude (r=-0.74) signifying decreasing incidence with increasing latitude.
A similar relationship was found when CMM incidence was correlated with annual
estimates of ultraviolet radiation (in the erythema-producing wavelengths),
supporting the hypothesis that melanoma is influenced by environmental factors
such as ultraviolet radiation. These analyses, however, showed considerable
variations with some counties falling well above the regression line.
Population density was also associated with melanoma incidence rates, with
higher rates in urban areas regardless of latitude. Further analyses tested
the hypothesis that these variations were related to frequency of foreign
travel to sunny locations (discussed in Chapter 8).
The International Agency for Research on Cancer (IARC 1976) collected and
published data on cancer incidence in five continents, including both
age-specific and age-standardized incidence rates from 59 population-based
cancer registries in 27 countries. Crombie (1979) analyzed the IARC data on
malignant melanoma from the 27 cancer registries in Europe and 16 North
America to investigate the relationship -between melanoma incidence and
latitude in areas with predominantly Caucasian populations. Melanoma
incidence in North America showed significantly increasing trends with
decreasing latitude for both males (p<0.01) and females (p<0.05). A
similar trend of increasing melanoma incidence with decreasing latitude was
also found for both males (p<0.001) and females (p<0.05) in England.
Analyses of data for the European cancer registries showed significant trends
in the opposite direction, i.e., increasing melanoma incidence with increasing
latitude. The analyses showed particularly high melanoma incidence and .
mortality rates in Sweden and Norway. These European results contradict
findings of ecological studies in North American and England, but are
consistent with'those of Lee and Isenberg (1972) who found higher incidence
and mortality rates due to CMM in Swedish versus. English populations, as shown
in Table 6-2. IARC (1976) and Armstrong (1984) have attributed the lack of a
latitude gradient in Europe to differences in complexion between Mediterranean
and Scandinavian populations. Lee and Isenberg (1972) postulated that
differences between English and Swedish rates could be due to genetic or
occupational factors.
Armstrong (1984) also considered the inconsistencies seen in the
relationship between geographic area and melanoma, mainly in Europe, and
concluded that the south-to-north gradient associated with melanoma risk may
be explained by a gradient in the opposite direction for skin pigmentation
(i.e., more highly pigmented populations reside in southern Europe), and by a
more intermittent pattern of sun exposure for populations living in northern
Europe.
FINDINGS
From the information reviewed above, the following findings are evident:
6.1 Within nations with predominantly White populations,
most ecological studies of melanoma and latitude show
increasing melanoma incidence and/or melanoma mortality
with decreasing latitude, leading to the hypothesis
DRAFT FINAL * » ••
-------
6-8
TABLE 6-2
AGE-ADJUSTED* INCIDENCE AND DEATH RATES PER 100,000
FOR MALIGNANT MELANOMA BY SEX: ENGLAND AND WALES
(1962-67) AND SWEDEN (1962-65)
England
and Wales Sweden
Incidence Per 100,000
Male 1.43 3.94
Female 2.40 4.23
Deaths Per 100,000
Male 0.92 1.92
Female 1.01 1.41
* Adjusted to UICC standard European
population distribution.
.Source: Lee and Issenberg (1972)-.
DRAFT FINAL
-------
6-9
that melanoma is associated with sunlight, particularly
its UV-B component, because of UV's strong correlation
with latitude. The ecological studies which failed to
show this association may not have accounted adequately
for pigmentation differences.
6.2 Although further north, Sweden has higher incidence and
mortality rates due to CMM than England and Wales. As
discussed in Chapter 7, pigmentation differences may be
responsible since the Swedish are a more homogeneous
fair-skinned population than the English who are a
mixture of several European races.
6.3 In general, CMM incidence and mortality rates tended to
be higher in populations living closer to the equator,
in coastal compared with inland areas, and in urban
versus rural areas within various nations.
DRAFT FINAL
-------
6-10
REFERENCES
Anaise, D., Steinitz, R., and Ben Hur, N. Solar radiation: A possible
etiological factor in malignant melanoma in Israel; A retrospective study
(1960-1972). Cancer 42:299-304 (1978).
Armstrong B.K. Melanoma of the skin. Brit Med Bull 40:346-350 (1984).
Baker-Blocker, A. Ultraviolet radiation and melanoma mortality in the United
States. Env Res 23:24-28 (1980).
Crombie, I.K. Variation of melanoma incidence with latitude in North America
and Europe. Brit J Cancer 40:774-781 (1979).
Eklund, G. and Malee, E. Sunlight and incidence of cutaneous malignant
melanoma. Scand J Plas Reconstr Surg 12:231-241 (1978).
Elwood, J.M. and Lee, J.A.H. Trends in mortality from primary tumours of skin
in Canada. CMA Journal 110:913-915 (1974).
Fears, T.R., Scotto, J., and Schneiderman, M.A. Skin cancer, melanoma and
sunlight. Amer J Pub Hlth 66:461-464 (1976).
Green, A. and Siskind, V. Geographical distribution of cutaneous melanoma in
Queensland. The Med J Aust 1:407-410 (1983).
Haenszel, W. Variations in skin cancer merdena within the United States. NCI
Monograph 10:225-243 (1963).
Herron, J. The geographical distribution of malignant melanoma in
Queensland. The Med J Aust 1:892-894 (1969).
Holman, C.D.J., James, I.R., Gattey, P.M., and Armstrong, B.K. An analysis of
trends in mortality from malignant melanoma of the skin in Australia.
Intl J Cancer 26:703-709 (1980).
Holraan, C.D.J., Mulroney, C.D., and Armstrong, B.K. Epidemiology of
pre-invasive and invasive malignant melanoma in Western Australia. Brit J
Cancer 25:317-323 (1980).
IARC monographs. 3. Biological data relevant to the evaluation of
carcinogenic risk to humans. Appendix to vol 40, in press. (1986)
Lancaster, K.O. and Nelson, J. Sunlight as a cause of melanoma: A clinical
survey. The Med J Aust 6:452-456 (1957).
Lee, J.A.H. and Issenberg, H.J. A comparison between England and Wales and
Sweden in the incidence and mortality of malignant skin tumors. Brit J
Cancer 26:59-66 (1972).
DRAFT FINAL
-------
6-11
Magnus, K. Incidence of malignant melanoma of the skin in Norway, 1955-1970:
Variations in time and space and solar radiation. Cancer 32:1275-1286
(1973).
Scotto, J., Fears, T.R. and Fraumeni, J.F., Jr. Solar Radiation. In: Cancer
Epidemiology and Prevention. Shottenfeld, D and Fraumeni, J.F., Jr.
^eds.) W.B. 'Saunders Company, Philadelphia (1982).
Swerdlow, A.J. Incidence of malignant melanoma of the skin in England and
Wales and its relationship to sunshine. Brit Med J 2:1324-1327 (1979).
Teppo, L., Pakkanen, M. and Hakulinen, T. Sunlight as a risk factor of
malignant melanoma of the skin. Cancer 41:2018-2027 (1978).
DRAFT FINAL
-------
CHAPTER 7
MIGRANT STUDIES
Several epidemiological studies have compared cutaneous malignant melanoma
incidence and mortality rates among migrants, among natives of the country
adopted by these migrants, and among natives in the countries from which the
migrants originated. These investigations have generally indicated that CMM
risks among migrants who have moved to sunnier climates are lower than those
among the native-born population from the adopted country (Houghton and Viola
1981; Lee 1982). All of the epidemiological results showed that increasing
duration of residence and earlier age at arrival were associated with higher
risks of CMM among European-born immigrants to Israel, Australia, and New
Zealand (Movshovitz and Modan 1973; Anaise et al. 1978; Katz et al. 1982;
Holman et al. 1980; Dobson and Leeder 1982; Holman and Armstrong 1984; Cooke
and Fraser 1985). There are no published data on the CMM risk of migrants to
less sunny locales. This chapter reviews the available epidemiological
information relevant to risk of CMM among immigrants.
The earliest studies which identified migrants as a unique population with
respect to CMM were conducted in Israel. The Israeli population was largely
developed over the last century as a result of substantial immigration from
Europe, Asia, and Africa, and thus provided a valuable population for various
studies. Three studies using data from Israel's Central Cancer Registry have
been conducted (Movshovitz and Modan, 1973; Anais-e et. al. 1978; Katz et. al.
1982) with cases overlapping among the studies; therefore, only the most
recent study is described.
The most recent Israeli study examined 1,050 CMM cases diagnosed from
1960-1974 and reported in the Central Cancer Registry (Katz et al. 1982). The
authors analyzed incidence rates (based on Central Bureau of Statistics data)
by place of birth and length of stay. As shown in Table 7-1 incidence rates
among the Israeli-born or the European- and American-born Jews were higher
than the rates among Asian- or African-born Jews for all age groups and each
average length of residence. The age-categorized incidence rates (see Table
7-1) among the European- and American-born were generally lower than the rates
among the Israeli-born except for those immigrants residing in Israel for at
least 17 years and in the 15-29 or 65+ age groups, and those living in Israel
for an average of four years and 65-1- years of age. The incidence rates among
European- and American-born slews who had lived in Israel for 17+ years
consistently exceeded the rates for those who lived in Israel for an average
of 13 years but not for those in Israel for an average of four years. A
clearer difference was observed when incidence rates were compared for the
period of immigration (before 1947 and after 1948) for European- and
American-born Jews. For those diagnosed between 1965 and 1974, the range of
average annual incidence rates within 15-year age groups for those immigrating
before 1947 was 10.28-12.36/105 compared with the range for those
immigrating after 1948 (1.09-4.05/105). For cases diagnosed between 1960
arid 1964, the differences between those arriving in Israel before 1947 and
* DRAFT FINAL
-------
7-2
TABLE 7-1
INCIDENCE OF MALIGNANT MELANOMA IN ISRAEL
(1965-1974) AMONG JEWS BY PLACE OF BIRTH,
AGE, AND AVERAGE LENGTH OF RESIDENCE IN ISRAEL*
Average
Length of
Residence
Place of Birth (yr.)
Israel
Asia or Africa 17+
13
4
Europe or America 17+
13
4
Number
of
Cases
248
25
30
10
334
121
66
Age
All
0-14
0.18
b
-
c
-
c
-
b
3.08
c
~
15-29
2
0
0
5
1
2
.82
c
-
.38
.50
.70
.68
.08
30-44
11.72
0.68
1.04
c
-
8.36
4.14
6.66
45-64
13
2
1
2
9
6
8
.02
.12
.64
.22
.42
.20
.44
65+
7
3
1
6
10
7
17
.84
.12
.84
.44
.52
.26
.38
Ages
2
1
0
0
9
4
5
.16
.24
.88
.86
.26
.72
.96
Average annual incidence rate per 100,000.
b
No population.
c
No cases.
Source: Katz et al. (1982).
* * » DRAFT FINAL » •• *
-------
7-3
after 1948 were not nearly as large. Katz et al. (1982) concluded that the
higher incidence rates among European- and American-born immigrants compared
to the more pigmented immigrants from Middle Eastern countries, and the
increasing incidence with length of exposure among the European- and
American-born were "consistent with the cumulative effect of solar radiation
as a causative factor."
Beginning in 1980, four detailed studies based on CMM data from Australia
and New Zealand were published. Holman et al. (1980) analyzed data on 120
pre-invasive malignant melanoma (PIM) and 422 invasive malignant melanoma
(CMM) cases identified in Western Australia from hospital discharge records
and pathology lab reports for 1975-1976. As shown in Table 7-2, incidence
rates (age-standardized to the total population of Western Australia in 1976)
were greater among native born Australian males with PIM and CMM and females
with CMM than among British immigrants. Table 7-2 also indicates that rates
of CMM among male and female British-born immigrants were about two times
greater than all other immigrants combined. The differences in incidence
rates were unchanged after adjustment for social class and age. Although
their data sources did not provide information on duration of residence, the
authors noted that the observed differences in incidence rates would be
expected if CMM risks "were proportional to duration of residence in an area
of high sun exposure".
Dobson and Leeder (1982) examined data on 2,243 CMM deaths obtained from
the Australia Bureau of Statistics for 1968-1977. Standardized mortality
ratios adjusted for age and country of birth were about two times higher among
native-born Australians (124.3 and 118.6 for males and females, respectively)
than among immigrants (except immigrants from New Zealand). The standardized
mortality ratios among male immigrants ranged from 30.1 for the Netherlands to
71.0 for "elsewhere" (i.e., non-European). Among female immigrants the
standardized mortality ratios ranged from 28.4 for Germany to 85.2 for
Poland. Among immigrants from England and Ireland, those living in Australia
for 24 years or longer had higher mortality rates than those living in
Australia for less than 24 years. This finding was consistent in all sex and
age groups considered, except for females in age group 20-39. The CMM
mortality rates among these immigrants by age, sex, and duration of residence
were all less than the comparable mortality rates among native-born
Australians. The mortality rates among Australian immigrants of at least 24
years exceeded those rates shown for England and Wales (Lee and Yongchaiyudha
1971, as cited in Dobson and Leeder 1982).
Dobson and Leeder (1982) also compared CMM deaths for English and Irish
immigrants (for 1968-1977) with deaths from all causes (for 1971) in Australia
according to age at arrival, age at death, and sex. For CMM deaths occurring
after age 40, the ratio of CMM deaths to deaths from all causes among
immigrants generally had an inverse relationship with age (i.e., the earlier
the age of arrival the higher the ratio). For example, males arriving before
10 years of age were about three times more likely to die from melanoma than
those who arrived after age 40 and about two times more likely to die from CMM
than those arriving between 20 and 39 years of age. Dobson and Leeder (1982)
concluded that the higher CMM mortality among immigrants arriving during
* ••• DRAFT FINAL
-------
7-4
TABLE 7-2
AGE-STANDARDIZED INCIDENCE RATES OF PREINVASIVE AND
INVASIVE MELANOMA IN WESTERN AUSTRALIA BY PLACE OF BIRTH
Males
Females
Place of Birth Number Incidence Rate
Number Incidence Rate
Australia
British Isles
Elsewhere
Pre-invasive Melanoma
38 5.6 49
7 2.8 11
5 3.0 6
6.7
4.8
3.3
Invasive Malignant Melanoma
Australia
British Isles
Elsewhere
170
23
13
26.1
10.0
. 6.1
170
29
8
23.7
12.7
4.3
Rate per 100,000 standardized to the age-distribution of the total
population of Western Australia in 1976.
Source: Holman et al. (1980).
- - DRAFT FINAL * -> *
-------
7-5
childhood and adolescence suggested that young people may be especially
susceptible to a melanoma initiating agent, such as sunlight, which is more
prevalent in Australia than in their countries of birth.
Holman and Armstrong (1984) examined incident melanoma patterns among
Western Australians in a case-control study. The 511 cases (23 percent
migrants), were classified according to histologic type. The controls were
matched by sex, 5-year birth period, and area of residence from the Australian
Commonwealth Electoral Roll or, for 10 cases younger than 18 years, from
public school student rolls. Sixty-five percent of all immigrants in the
study were born in the U.K. The authors did not, however, provide a numerical
breakdown of the places of birth of the cases and controls. As shown in Table
7-3, the odds ratios calculated for each histogenic type increased with
increasing duration of residence (0-24, 25-39, 40-59, and >60 years),
especially for nodular melanomas (NM) and Hutchinson's Melanomic Freckle
(HMF). The trend of increasing melanoma incidence with increasing duration of
residence was statistically significant (P<0.003) for all melanomas combined
and each histogenic type except for unclassified melanoma (UCM).
Holman and Armstrong (1984) also evaluated the CMM MSK associated with age
at arrival and discovered that age at arrival was a better predictor than
duration of residence for risk of all melanomas combined and superficial
spreading melanoma (SSM). For NM and HMF, the variables age at arrival and
duration of residences were too highly interrelated to permit separation of
their effects. The data on 267 SSM case-control pairs were further analyzed
by age at arrival after controlling for ethnicity (i.e., numbers of European,
African, and Asian grandparents) and the results are presented in Table 7-4.
The risk of SSM in immigrants arriving between 0-4 and 5-9 years were somewhat
though not significantly greater than that for native-born Australians. The
odds ratios were less than one, however, for those immigrants arriving between
10-14 and 15-19 years of age and then generally stabilized around 0.25 for
subsequent ages of arrival. The trend in odds ratios by age at arrival was
significant (P=0.0001). Holman and Armstrong (1984) concluded that the
results for SSM suggested a crucial age at arrival somewhere between 10 and 15
years of age, before which exposure to sunlight in early childhood may play a
role in the production of benign nevi. They hypothesized that the benign nevi
in turn may act as precursor lesions for SSM.
Holman and Armstrong (1984) also observed an increased proportion of
palpable nevi on the arms of controls of English, Celtic, or Australian
heritage who were born or arrived in Austrialia before 10 years of age
compared with those who were 10 years or older at arrival. They hypothesized
that if the production of "initiated nevus cells" was a step in the
pathogenesis of SSM, then the potential to develop SSM would be determined by
the number of initiated nevus cells induced in childhood or young adulthood.
This, they concluded, could explain the overriding effect of age at arrival
versus duration of residence in Australian immigrants with SSM and the
uniformly low rate of SSM in immigrants arriving after 10-14 years of age.
Holman and Armstrong (1984) further concluded that it would be difficult to
propose a factor other than sun exposure that could account for the lower CMM
incidence rate in British migrants compared with native-born Australians, the
majority of whom were of British descent.
* * * DRAFT FINAL * * *
-------
7-6
TABLE 7-3
RELATIONSHIP OF HISTOGENIC TYPES OF MALIGNANT
MELANOMA TO DURATION OF RESIDENCE IN AUSTRALIA
b
Duration of Residence, yr.
a
Parameter
All melanomas (507)
Odds ratio
95% CI
Histogentic type
HMF (86)
Odds ratio
95% CI
SSM (267)
Odds ratio
95% CI
UCM (89) -
Odds ratio
95% CI
NM (51)
Odds ratio
95% CI
0-24 25-39
(215) (293)
1.00 1.47
0.92-2.35
1.00 0.90
0.21-3.87
1.00 2.30
1.21-4.39
1.00 0.71
0.24-2.08
1.00 1.05
0.20-5.44
40-59
(275)
3.24
1.93-5.44
3.12
0.65-15.03
4.13
1.98-8.63
1.18
0. 39-. 358
4.85
0.85-27.55
>60
(231)
4.87
2.41-9.85
6.35
1.11-36.45
3.46
1.30-9.20
3.91
0.36-42.02
14.72
1.16-186.16
P- value
of Trend
•
0.000001
0.003
0.00008
0.545
0.009
Numbers in parentheses are number of case-control pairs.
b
Numbers in parentheses are total number of subjects in each exposure category.
Source: Holman and Armstrong (1984).
DRAFT FINAL * * *
-------
7-7
TABLE 7-4
RELATIONSHIP OF SSM TO AGE AT ARRIVAL
IN AUSTRALIA WITH CONTROL FOR NUMBERS OF
EUROPEAN, AFRICAN, AND ASIAN GRANDPARENTS
(Based on 267 case-control pairs)
Age at Arrival
(yr.)
Birth
0-4
5-9
10-14
15
-19
20-24
25
-29
>30
Odds
1
1
1
0
0
0
0
0
Ratio
.00
.17
.65
.74
.25
.25
.23
.38
95%
0
0
0
0
0
0
0
P-value
CI of Trend
.25-5
.34-
.17-
.05-
.08-
.07-
.19-
7
3
1
0
0
0
.45
.97
.28
.43
.83
.73
.78 0.0001
Source: Holman and Armstrong (1984).
* * •••- DRAFT FINAL
-------
7-8
A recent migrant study (Cooke and Fraser 1985) focused on 893 melanoma
cases who died between 1972 and 1980 in New Zealand. The data were obtained
from the National Health Statistics Center and, for immigrant cases, were
restricted to those with at least five years residence in New Zealand.
Mortality rates for the New Zealand-born were calculated using 1976 census
data while for immigrants, an unpublished census table of "usually resident"
populations was used. Cooke and Fraser (19S5) observed that CMM mortality
rates were consistently lower for European immigrants than for those born in
New Zealand although the number of deaths in some immigrant groups (e.g., the
Netherlands) was small. The authors compared mortality rates by age and sex
for New Zealand-born cases from 1972-1980, for UK immigrants from 1972-1980,
and for CMM cases in England and Wales in 1976. The immigrant mortality rates
were intermediate between rates for England and Wales and rates for New
Zealanders (except for 15-54 year old female immigrants with only four CMM
deaths from 1972-1980). The age-standardized mortality rate for 35-64 year
olds was higher for those arriving in New Zealand before 30 years of age
(7.1/105; 95% CI 4.6-10.5) than at 30 years or older (2.8/105).
Alternatively, the age-standardized mortality rate for 35-64 year olds was
higher for those living in New Zealand for at least 20 years (3.9/105) than
for those residing in New Zealand for 5-19 years (2.9/105). The authors
concluded that an early age at migration appeared to be associated with
increased risk of CMM death among immigrants, a risk similar to that in their
adopted country. The authors postulated that some factor acting in the first
few decades of life, possibly patterns of sun exposure, were important in
determining CMM risk.
An Hawaiian study (Hinds and Kolonel 1980) provided results which
contradicted those from most other studies. The investigators of this study
found that among 265 Caucasian CMM cases, age-adjusted incidence rates among
immigrants to Hawaii were much greater than rates among Hawaiian-born
Caucasians. The study did not include information on country of birth, ethnic
background, duration of residence, or age at arrival. Without more detailed
data on these variables, the explanation for the differences in incidence
rates cannot be meaningfully explained. The authors indicated, however, that
the immigrating Caucasian population may have been more susceptible to
melanoma than the primarily Portuguese native Caucasian population.
FINDINGS
Evidence from the studies reviewed in this chapter supports the following
findings:
7.1 Immigrants moving to sunnier climates in which the
native CMM incidence rates exceed those of the
immigrant's country of origin tend to have lower CMM
risks than the native population. These risks
increase, however, with increasing duration of
residence or earlier age of arrival in the adopted
homeland.
* ••• * DRAFT FINAL *
-------
7-9
7.2 In an Australian study, age at arrival in Austrialia
was more important than duration of residence with
respect to the odds of SSM, with arrival before age 10
associated with an odds ratio near to or greater than
the estimated risk of SSM for those born in Australia.
Odds ratios decreased in association with age at
arrival 'at 10-14 years relative to those born in
Australia; odds ratios stablized at a significantly
lower level for those who arrived at or after age 15.
* DRAFT FINAL
-------
7-10
REFERENCES
Anaise, D., Steinitz, R., and Ben Hur, N. Solar Radiation: A possible
etiological factor in malignant melanoma in Israel; A retrospective study
(1960-1972). Cancer 42:299-304 (1978).
Cooke, K.R. and Fraser, J. Migration and death from malignant melanoma. Int
J Cancer 36:175-178 (1985).
Dobson, A.J., and Leeder, S.R. Mortality from malignant melanoma in
Australia: Effects due to country of birth. Int J Epidemiol 11(3):207-211
(1982).
Hinds, M.W., Kolonel, L.N. Malignant melanoma of the skin in Hawaii,
1960-1977. Cancer 45:811-817 (1980).
Holman, C.D.J. and Armstrong, B.K. Cutaneous malignant melanoma and
indicators of total accumulated exposure to the sun: An analysis separating
histogentic types. JNCI 73:75-82 (1984).
Holman, C.D.J.; Mulroney, C.D., and Armstrong, B.K. Epidemiology of
pre-invasive and invasive malignant melanoma in Western Australia. Brit J
Cancer 25:317-323 (1980).
Houghton, A.M. and Viola, M.V. Solar radiation and malignant melanoma of the
skin. J Am Acad Dermatol 5:477-483 (1981).
Katz, L., Ben-Turia, S., and Steinitz, R. Malignant melanoma of the skin in
Israel: Effect of migration. In: Trends in cancer incidence: Causes and
practical implications. Magnus, K. .(ed). (1982).
Lee, J.A.H. Melanoma and exposure to sunlight. Epidemiologic Reviews
4:110-136 (1982).
Movshovitz, M., and Modan, B. Role of sun exposure in the etiology of
malignant melanoma: Epidemiologic enference. JNCI 51(3):777-779 (1973).
* * DRAFT FINAL * * *
-------
CHAPTER 8
CORRELATIONS WITH INDICATORS OF INTERMITTENT OR SEVERE SUN EXPOSURE
Results from epidemiological studies of cutaneous malignant melanoma (CMM)
have led to conclusions that ranged from those which implicate sunlight as a
causal factor, [e.g., most of the latitude studies (Chapter 8), migration
studies (Chapter 7)], to those which do not support sunlight as a causal
factor [e.g., no overall elevated risk of melanoma observed among outdoor
workers compared to office workers (Chapter 11)]. In attempt to reconcile
these differences, it has been suggested that the risk of CMM on sites not
usually exposed to sun is increased by intermittent exposure to more intense
sunlight, while chronic sun exposure at a relatively constant dose may have
little effect or be protective due to preventive tanning of the skin. This
hypothesis can be examined using different measures of intermittent exposure
as surrogates for actual exposure. This chapter reviews studies of CMM which
investigated intermittent exposures to sunlight as estimated by a history of
sunny vacations, recreational activities, and history of severe sunburn.
(particularly in early life). Also included is a section which reviews
studies of sunspots and seasonal differences in the incidence of CMM.
Sunspots result in higher levels of UV-radiation which reach the earth's
surface in a cyclical pattern and may result in more severe UV exposures.
HISTORY OF SUNNY VACATIONS OR RECREATIONAL ACTIVITIES
A vacation in a sunny location may result in intermittent exposure to UVB
radiation at higher than usual levels. The levels of exposure can be variable
but, in general, sunny vacations are chosen simply because of the increased
sunlight in the area. Several studies have examined past history of sunny
vacations using various methods to explore the role of sun exposure in the
etiology of cutaneous malignant melanoma.
Using data from the Swedish Cancer Registry for 1959-1968, Eklund and
Malec (1978) found that CMM incidence increased with population density. In
an attempt to explain this finding, the authors hypothesized that an increase
in foreign travel as estimated by passport issuance, might explain the over
representation of CMM in the large cities because "foreign travel in Sweden
generally means sunshine trips." A 3.7 percent increase in the mean frequency
of annual passport issue between 1959 and 1968 corresponded to an increase in
CMM incidence over the same time period. Both increases were especially high
in the densely populated cities and counties. Regression analysis using an
exposure index based on annual UV radiation in the erythema-producing
wavelengths and latitude. Results showed variations in CMM incidence between
Sweden's major cities and some counties, but these differences were reduced
when frequency of passport issuance was considered in the analysis. Thus,
foreign travel as represented by passport issuance, was seen as a possible
explanation for the higher CMM incidence in the most populated areas of Sweden.
A Norwegian study (Klepp and Magnus, 1979) investigated sunny vacations
through self-administered questionnaires from 78 CMM cases and 131 unmatched
* * - DRAFT FINAL * * *
-------
8-2
controls who were other cancer patients. The study showed no differences
between cases and controls in pigmentation characteristics (hair and eye
color), in estimated time spent outdoors during leisure activities, or in
degree of exposure to sunshine during occupational or leisure activities.
There was, however, a borderline significant difference (p=0.05) between the
proportion of cases (19.2 percent) and controls (9.2 percent) who had traveled
to Southern Europe for sunbathing during the previous five years (estimated
relative risk of 2.4). This study did not account, however, for socio-
economic status, a variable often related to CMM and the tendency to go on
vacations.
A study of oral contraceptive use and CMM in England (Adam et al. 1981)
included sun exposure factors to explore potential confounding of study
results. The study included 169 women aged 15-49 with CMM registered during
1971-1976 at the Oxford and South Western cancer registries and 507 female
controls (3 controls per case) matched by 5-year age group and marital status,
drawn from the physicians' practice lists. The authors found no significant
differences between cases and controls who "tanned themselves while on holiday
abroad" (for legs, 78 percent cases and 73 percent controls, and for trunk, 70
percent cases and 67 percent controls). A higher proportion of cases than
controls spent outdoor leisure time deliberately tanning their legs (77
percent versus 69 percent) and trunk (64 percent vs. 53 percent), although
these differences were not statistically significant.
A study of 595 melanoma,case-control pairs in Western Canada by Elwood et
al. (1985a) showed that substantial intermittent sun exposure (as assessed by
vacation and recreation histories) was strongly associated with CMM. There
was a significant (p<0.001) trend of increasing risk of CMM occurrence with
the number of "sunny vacations" (defined as severe or more intense sun
exposure than normal) per decade. A relative risk of 1.8 for CMM was also
associated with a history of four or more sunny vacations per decade. This
relative risk remained significant (RR=1.7, 95% C.I. 1.2-2.3) even after
adjustment for pigmentation factors (hair color, skin color, history of
freckles) and ethnic origin. Further analyses were conducted on the risk
associated with certain activities and practices on sunny vacations. Sun
exposure through beach and other light-clothing activities during sunny
vacations was associated with a relative risk of 1.9 (95% C.I. 1.3-3.0) for
20-39 exposure hours per summer season compared with no summer vacation sun
exposure after adjustment for pigmentary factors and ethnic origin. This risk
level could be reached by 4-8 hours per day of sun exposure during a 1-week
vacation. The relative risk for 40 or more hours of vacation sun exposure per
summer season decreased slightly to 1.5 (95% C.I. 1.0-2.3), and was of
borderline statistical significance. Socio-economic factors were not
controlled for in the above analyses, and may have led to biased results.
Holman et al. (1986) analyzed data from their matched case-control study
of CMM in Western Australia to investigate the relationship of different
histologic types of CMM with intermittent sun exposure, as measured by history
of summer sun exposure (both recreational and total which includes
occupational) and clothing habits. Analyses were also performed to
-.v -:.- * DRAFT FINAL
-------
8-3
investigate variations in relationships of sun exposure and CMM by primary
tumor site.
With the exception of HMFM, all histologic types of CMM appeared to be
inversely associated with total summer outdoor exposure (none of the
associations were statistically significant). Further analysis measured
recreational exposure as a proportion of total outdoor time in summer to
evaluate the concentration of sun exposure during days off work. This
variable -- recreational outdoor exposure proportion in summer (ROEP) --
provided an index in which the range from 30 percent to 100 percent indicated
an increasing concentration of outdoor (sun exposure) time during leisure
days. The higher proportions in this index imply a "burst" of exposure over a
relatively short part of the week as contrasted with low proportions (ROEP
near 0 percent) for farmers who worked outdoors 7 days a week. The effects of
ROEP were examined separately by age group (10-24 years, 25-39 years, 40 years
and over) and by different time intervals prior to diagnosis of the case (0-4,
5-9, 10-19 and >20 years pre-diagnosis). Results showed little evidence of
any association of CMM with ROEP after control for potential confounders
(pigmentation characteristics, ethnic origin, and age at arrival in
Australia). For SSM and NM, there was a suggestion of a dose-response
gradient in ROEP at ages 10-24 years, but the trends were not statistically
significant (p=0.148 for SSM and p=0.258 for NM) . Additional analyses using
the absolute average number of hours exposed per week or using the difference
between average hours of recreational and work exposure as measurements of
recreational exposure to summer sun showed no stronger evidence of an .
association between incidence of CMM and recreational sun exposure than that
mentioned previously.
Analyses of specific outdoor recreational pasttimes were conducted based
on the 276 SSM case-control pairs. Both boating and fishing at least once per
week during summer showed significantly increased risks for SSM when compared
with those who never participated (boating: OR=2.43, 95% C.I. 1.10-5.39;
fishing: OR=2.72, 95% C.I. 1.15-6.43). There was little evidence, however,
for a relationship of SSM with swimming. Frequency of summer sunbathing at
ages 15-24 years showed only a weak association with SSM (OR=1.26, 95% C.I.
0.78-2.05 for "less than one/week" and OR=1.32, 0.80-2.17 for "once or
more/week"); however, when the analysis was confined to SSM occurring on the
trunk with never sunbathing as the reference group, the odds ratio for the
higher frequency sunbathing group attained statistical significance with an
p-value for trend 0.044, OR=1.20, 0.51-2.81 for less than once per week and OR
of 2.55 (95% C.I. 1.05-6.19) for once or more per week. Again, socio-economic
factors were not controlled for in the analysis and may have led to biased
results.
The types of bathing suit worn by women at ages 15-24 years (usually the
period of most frequent sunbathing) and in the 10 years prior to case
diagnosis were also examined. Results for the type of bathing suit worn
showed a strong increase in risk of melanoma of the trunk with decreasing
DRAFT FINAL
-------
8-4
bathing suit coverage, even after control for potential confounders. Not all
results reached statistically significant levels as shown on Table 8-1, but
increased OR's were associated with type of bathing suit' style for CMM of the
trunk (p-value for trend = 0.005 for women at ages 15-24 years and 0.006 for
0-9 years per diagnosis).
SUNBURN IN EARLY LIFE
Elwood et al. (1984, 1985) in a Canadian, case-control study (595 matched
pairs), assessed both the tendency to sunburn and history of sunburn. A
specific question was asked about sunburn in childhood using gradations of 1,
2, and 3 for "rare, very mild, or no burn," "moderate or infrequent," and
"severe or frequent burn", respectively. The authors found a significantly
increased risk for sunburn in childhood (RR=1.9) although the risk be'came
smaller and statistically nonsignificant after adjustment for pigmentation
factors (hair, skin, and eye color). A history of frequent sunburn in
childhood remained a significant risk factor along with pigmentation even
after control for ethnic origin, but the authors concluded that "the risk is
due to characteristics of pigmentation associated with poor sun tolerance"
(Elwood et al. 1984).
Lew et al. (1983) found similar results in a Massachusetts case-control
study (111 melanoma cases and 107 unmatched controls): risk factors with
elevated OR's included "blistering from sunburns in adolescence" (OR=2.05, 95?&
C.I., 1.18-3.56) and "painful sunburn as a child" for both those who tanned
well (OR=2.8, 1.3-6.3) and those who tanned poorly (OR=3.0, 0.9-9.8). A
history of extended sunny vacations (30 days or more as a child) was also
found to elevate CMM odds significantly (OR=2.5, 1.1-5.8). The authors
concluded that the same etiology underlies each of these risk factors, i.e.,
"the degree of response to sun exposure," and that the nature of these
traumatic exposures and sunny vacations in early life suggest that traumatic
doses of sun may outweigh lifetime cumulative doses as a risk factor for
melanoma. These findings should be interpreted with caution because there was
no control for pigmentation factors or socio-economic status, and the control
selection was flawed (described in Chapter 7). In addition, the potential for
recall bias, whereby cases are more likely to remember early sunburn episodes
than controls, should be considered.
Holman and Armstrong (1986) analyzed sunburn histories of CMM cases and
controls in Western Australia (507 pairs matched on age, sex and area of
residence) for age groups under 10 years and 15-24 years. No relationship
between sunburn in childhood or early adulthood and any histologic type of
melanoma was found after control for potential confounders such as chronic and
acute skin reaction to sunlight, hair color, ethnic origin, and age at arrival
in Australia. Compared with persons who reported no painful sunburns in
childhood (under 10 years of age), the odds ratios for SSM were 1.06 (95% C.I.
0.65-1.75) for those who experienced painful sunburn up to four times and 1.11
(0.51-2.41) for those who reported five or more painful sunburns. For sunburn
during ages 15-24 years, the corresponding odds ratios for SSM were 1.04
(0.66-1.66) for four or fewer sunburns and 0.98 (0.53-1.82) for five or more
sunburns.
* DRAFT FINAL
-------
8-5
TABLE 8-1
RELATIONSHIP OF CUTANEOUS MALIGNANT MELANOMA IN WOMEN TO TYPE
OF BATHING SUIT WORN IN SUMMER, CONTROLLED FOR
POTENTIAL CONFOUNDERS*
Melanoma Site
and Period of Life
Trunk (60 case-control
Ages 15-24 years
Type of
One Piece,
High Back
pairs)
1.0
0-9 years prediagnosis 1.0
Bathing Suit Worn
One Piece,
Low Back
4.04
(0.65-25.23)
1.12
(0.23-5.47)
in Summer
Bikini
or Nude
12.97
(1.95-83.94)
8.94
(1.45-55.07)
Null
p-value
of
' Trend
0.005
0.006
SSM Trunk (30 pairs)
Ages 15-24 years 1.0
0-9 years prediagnosis 1.0
Non-SSM Trunk (30 pairs)
Ages 15-24 years 1.0
0-9 years prediagnosis
0.19
(0.01-7.19)
0.61
(0.03-10.50)
3.22
(0.10-103.48)
4.06 0.054
(0.14-117.09)
2.52 0.175
(0.18-34.58)
374.14 0.119
(0.36-389.40)
* Chronic and acute skin reaction to sunlight, hair color, ethnic origin, and
age at arrival in Australia.
- Too few exposed for analysis.
Source: Holman et al. (1986).
* •'• DRAFT FINAL
-------
8-6
LIFETIME HISTORY OF SEVERE SUNBURN
The following section includes only those studies which reported sunburn
history, i.e., remembered severe sunburn, whether in childhood, adolescence,
or adulthood. The possibility of recall bias among CMM cases with respect to
sunburn exposures should be considered in the interpretation of these studies.
Mackie and Atchison (1982) conducted a case-control study of 113
age/sex-matched pairs (hospital-based controls) in the West of Scotland which
also considered social class and skin type in the analysis. There was a
significant difference (p<0.05) between the cases and controls in the
history of severe or prolonged sunburn five years before diagnosis of CMM
(i.e., blistering sunburn or erythema persisting seven or more days after sun
exposure). Overall, 56 percent of the melanoma patients had a history of
severe sunburn compared with 22 percent of the controls (RR=2.8, 95% C.I.
1.1-7.4 referent group not specified by authors). When separated by sex,
there were still significant differences (p<0.05) between the cases and
controls with respect to history of severe or prolonged sunburn. The presence
of recall bias among the melanoma cases could have biased the findings if
cases were more likely to remember or overstate their history of sunburns.
More extensive analyses of Western Canada study data on 595 age-, sex-,
and residence-matched case-control pairs, by Elwood et al. (1985b), evaluated
the relationship of CMM incidence to sunburn history at any age and vacation
sunburn history for those with recorded vacations. There was an increasing
trend for risk of CxMM with increasing'number and severity of sunburn episodes
(p<0.01) using a vacation sunburn score, and significantly raised odds
ratios of CMM (p<0.05) for each of the sunburn history categories considered
separately (sunburn on vacations, sunburn in childhood, and history of severe
sunburn causing pain or blistering for over 2 days) relative to those with no
or mild sunburn. Vacation sunburn scores were analyzed with the variable
"usual degree of suntan" in order to consider the separate effects of these
variables, but each remained statistically significant after adjustment for
the other, indicating that these two factors acted independently.
As in their earlier analysis of the sunburn in childhood data (Elwood et
al. 1984), the authors evaluated whether the tendency to sunburn and sunburn
history, as measured by vacation sunburn score, were independent risk factors
for melanoma. The results of this analysis showed a weak association between
CMM risk and vacation sunburn score after adjustment for the usual reaction to
sun, i.e., ranging from "tan to burn" to "burn only." Multivariate analyses
controlling for usual reaction to sun and other pigmentary factors (hair
color, skin color, and freckles in adolescence) further weakened the
association between CMM and vacation sunburns. The authors concluded that,
while history of severe or frequent sunburn was associated with an increased
risk of cutaneous malignant melanoma, the tendency to burn easily and tan
poorly was more strongly associated with CMM risk; thus, it appeared that it
was factors associated with the tendency to burn rather than a positive
history of sunburn which determined CMM risk. The authors concluded, however,
that in contrast to sunburn history and vacation sunburn score, melanoma risk
was increased by heavy vacation or recreational exposure to sunlight and that
* DRAFT FINAL
-------
8-7
the increased CMM risk from these substantial intermittent sun exposures was
independent of constitutional factors (pigmentation characteristics, reaction
to sun, and number of nevi). This association is probably not due to the
trauma of sunburn itself but due to another characteristic of individual skin
reaction, "related presumably to variations in melanocyte function,
distribution or prevalence."
A case-control study (183 pairs matched by age, sex, and place of
residence) in Queensland, Australia (Green et al. 1985) showed a positive
trend of increasing CMM risk with an increasing number of severe sunburns
(p<0.001). When the number of sunburns was grouped (0-1, 2-5, 6+) and an
adjustment was made for the presence of pigmented nevi on the arms (the
strongest risk factor determined in this study) and age, the positive trend
for CMM risk with number of sunburns remained significant (p<0.05); the
adjusted relative risk of melanoma for six or more severe sunburns was 2.4
(95% C.I. 1.0-6.1). The risk estimates were essentially unchanged by further
control for other risk factors such as presence of other skin cancers, migrant
status, and social class. The authors did not control for pigmentation
variables such as hair or skin color or the tendency to sunburn in the
analysis; therefore the results should be cautiously interpreted. Green et
al. (1985) stated that "an experience of painful erythema indicates that acute
high-dose UV has been delivered to the level of the melanocyte" and, because
of this, the above-mentioned variables should not confound an association
between CMM and severe sunburn (defined as sunburn with pain persisting longer
than 48 hours, with or without blistering). The authors also concluded that
the dose-response relationship obtained from their analysis supports a causal
interpretation of an observed sunburn-cutaneous melanoma association. The
authors believe that their results support an "intermittent episodes of acute
UV exposure" theory, but may also fit a "cumulative" or dose-related theory in
which the high UV dose has accumulated from multiple severe sunburn episodes.
Holman et al. (1986) analyzed sunburn histories from 507 CMM matched cases
and controls (matched on sex, age and residence) from Western Australia,
ranking them according to increasing sunburn severity: "peeling sunburn",'
"painful sunburn (pain for 2 days or more)", and "blistering sunburn". After
control for potential confounding factors (chronic and acute skin reaction to
sunlight, hair color, ethnic origin, and age at arrival in Australia), only
HMFM showed some association with the occurrence of severe sunburn (p-value
for trend = 0.059). For nodular melanoma (NM), there was a significant trend
in the opposite direction (p=0.010), giving the appearance of a protective
effect. This result was based on small numbers (51 NM cases) and attributed
to chance. For SSM and UCM, there was no association with past sunburn
severity after control for confoundering variables.
SUNSPOTS AND SEASONAL INCIDENCE DIFFERENCES
Scotto and Nam (1980) analyzed monthly incidence of CMM from the Third
National Cancer Survey (1969-71) to test for seasonal patterns in incidence.
There are monthly patterns in the amounts of solar radiation reaching the
earth's surface with the highest intensity UV-B radiation occurring during the
* -v * DRAFT FINAL
-------
8-8
summer months. The authors examined seasonal patterns of CMM incidence for
2,167 white patients (998 males and 1,169 females) by sex, age, geographic
region, and tumor site.
A highly significant seasonal pattern (p=0.00003) with a summertime peak
was found for the total female CMM cases -- over 20 percent of all cases were
diagnosed during-June and July, while less than 14 percent were diagnosed
during the winter months of December and January. For males, no seasonal
pattern of sustained peak or dip was found. Among males, the trunk was the
most common tumor site (38 percent), while among females, it was the lower
extremities (35 percent). Analysis of seasonal patterns of solar radiation by
anatomic site showed two highly significant sites for females -- upper
extremities, and lower extremities (p=0.0007 and p=0.0001, respectively), but
none for males -- upper extremities showed a tendency toward the seasonal
pattern with summertime peak but this was not statistically significant
(p=0.11). An attempt to reduce single-month disturbances in the patterns by
grouping into 2-month periods still showed a significant pattern for females
(p=0.003) but not for males (p=0.07).
The observations of seasonal patterns with summertime peaks in CMM
incidence for females with tumors on upper or lower extremities may be due to
the promotional effects of UV-B exposure or it may be a result of greater
awareness of skin changes and/or problems during the summer months when less
clothing is worn.
Hinds et al. (1981) conducted similar analyses .using 1960-1978 CMM
incidence data for Caucasians, in Hawaii. Based on 353 incident cases (males
and females combined), the authors found significant seasonal patterns with
summertime peaks for melanomas of all sites (p=0.018), and for those of the
lower extremities (n=79, p=0.017). For head and neck melanomas, there seemed
to be a similar seasonal pattern but it was of borderline statistical
significance (p=0.069). Hinds et al. (1981) stated that, because there is
little variation in the types of clothing worn throughout the year in Hawaii,
it is unlikely that a seasonal pattern would be due to increased observation
of the skin during the summer months. These authors concluded instead that
their findings supported the hypothesis that solar (UV) radiation may be a
short-term promoter of some malignant melanomas of the skin.
Houghton et al. (1978) used Connecticut Tumor Registry data on 2,983 CMM
cases diagnosed during 1935-1974 to analyze CMM incidence rates. The rate per
100,000 rose from 1.1 to 6.2 over this 40-year time period with evidence of
cyclical patterns with 3-5 year peaks in incidence every 8-11 years. The
secular increase in melanoma incidence was highly correlated (correlation
coefficient = 0.9327, p<0.01) with three sunspot cycles of 8-11 years over a
33 year period. CMM incidence rose sharply at the peak of each sunspot cycle
and the high rates persisted for 3-5 years before returning to a stable but
increased rate. Controlling for the time effect on increasing rates did not
alter the significant association between annual sunspot activity and CMM
incidence rates in each of the subsequent three years.
* * * DRAFT FINAL
-------
8-9
Analyses of melanoma incidence from New York State and Finland also showed
significant correlation with sunspot activity although data from Norway showed
no significant results. Analysis of data from New York State (1950-1971)
showed significant association between sunspots and CMM incidence in the 1-2
years subsequent to sunspot activity. In Finland, however, incidence rates
were significantly correlated only with the years of sunspot activity and the
first subsequent year; this decrease in lag period is postulated to be a
result of Finland's higher latitude. Wigle (1978) reported similar cyclical
variations in CMM incidence in Saskatchewan and Alberta, Canada; when CMM
incidence rates were correlated with periods of sunspot activity, incidence
rates were found to increase 0-2 years after the peak sunspot activity (0-1 in
Saskatchewan and 2 in Alberta). After a review of these data, Houghton and
Viola (1981) concluded that "rises in CMM incidence occur 0-3 years after
sunspot peaks, suggesting that heavier exposures to UV radiation trigger the
clinical appearance of melanoma."
FINDINGS
The following findings can be drawn from the review presented above:
8.1 A case-control study in Western Canada found evidence
of increasing CMM risk with increasing number of
"sunny vacations" taken, even after adjustment for
pigmentation factors and ethnic origin. This study
did not, however, control for socio-economic factors
which are likely associated both with CMM incidence
and number of sunny vacations.
8.2 Using a variable for recreational sun exposure,
defined as the ratio of summer recreational sun
exposure to total summer sun exposure, a Western
Australia study found no significant association of
this factor in any of the age groups examined.
However, an increased SSM risk for some summer sun
activities at early ages was observed for boating,
fishing, and female sunbathing (on trunk only). For
CMM of the trunk in women, particularly for SSM, risks
increased with decrease in coverage by the bathing
suit style worn at 15-24 years of age.
8.3 CMM risk is associated with a history of childhood
sunburn and/or lifetime history of sunburn but this
appears only to reflect an individual's pigmentary
characteristics particularly as they relate to poor
sun tolerance.
8.4 A seasonal pattern with a summertime peak was found
for CMM (female in U.S., both sexes in Hawaii); this
may be related to a greater awareness of skin changes
in the summer months.
DRAFT FINAL
-------
8-10
8.5 Most studies which examined the relationship between
CMM incidence rates and sunspot cycles found high
correlations. Different studies have observed
slightly different lag periods between peak sunspot
activity and increased CMM rates.
* - DRAFT FINAL
-------
8-11
REFERENCES
Adam, S.A., Sheaves, J.K., Wright, N.H., Mosser, G., Harris, R.W. and Vessey,
M.P. A case-control study of the possible association between oral
contraceptives and malignant melanoma. Brit J Cancer 44:45-50 (1981).
Eklund, 6. and Malec, E. Sunlight and incidence of cutaneous malignant
melanoma. Scand J of Plastic Reconstr Surg 12:231-241 (1978).
Elwood, J.M., Gallagher, R.P., Hill, G.B., Pearson, J.C.G. Cutaneous melanoma
in relation to intermittent and constant sun exposure - The Western Canada
melanoma study. Int J Cancer 35:427-433 (1985a).
Elwood, J.M., Gallagher, R.P., Davison, J., and Hill, G.B. Sunburn, suntan
and the risk of cutaneous malignant melanoma - The Western Canada melanoma
study. Br J Cancer 51:543-549 (1985b).
Elwood, J.M., Gallagher, R.P., Hill, G.B., Spinelli, J.J., Pearson, J.C.G.,
and Threlfall, W. Pigmentation and skin reaction to sun as risk factors
for cutaneous melanoma: Western Canada melanoma study. Brit Med J
288:99-102 (1984).
Green, A., Siskind, V., Bain, C., and Alexander, J. Sunburn and malignant
melanoma. Br J Cancer 51:393-397 (1985).
Hinds, M.W., Lee, J. and Kolonel, L.N. Seasonal patterns of skin melanoma
incidence in Hawaii. Am J of Publ Hlth 71:496-499 (1981).
Holman, C.D.J., Armstrong, B.K., Heenan, P.J. Relationship of cutaneous
malignant melanoma to individual sunlight-exposure habits. JNCI
76:403-414 (1986).
Houghton, A.N. and Viola, M.V. Solar Radiation and Malignant Melanoma of the
Skin. J Am Acad Dermatol 5:477-483 (1981).
Houghton, A., Munster, E.W., and Viola, M.V. Increased incidence of malignant
melanoma after peaks of sunspot activity. The Lancet April 8:759-760
(1978).
Klepp 0. and Magnus, K. Some environmental and bodily characteristics of
melanoma patients. A case-control study. Int J Cancer 23:482-486 (1979).
Lew, R.A., Sober, A.J., Cook, N., Marvell, R., and Fitzpatrick, T.B. Sun
exposure habits in patients with cutaneous melanoma: A case control
study. Journal of Dematol Surg Oncol 9:981-986 (1983).
MacKie, R.M., and Aitchinson. T. Severe sunburn and subsequent risk of
primary cutaneous malignant melanoma in Scotland. Br J Cancer 46:955-960
(1982)'.
•- DRAFT FINAL
-------
8-12
Scotto, J. and Nam, J. Skin melanoma and seasonal patterns. Am J Epi
111(3):309-314 (1980).
Wigle, D.T. Malignant melanoma of skin and sunspot activity. Lancet 2:38
(1978).
•-•- --•- * DRAFT FINAL
-------
CHAPTER 9
CORRELATIONS WITH INDICATORS OF
CUMULATIVE SUN EXPOSURE
The results from the ecological studies of melanoma and latitude led to
case-control studies which further investigated the hypothesis that UV
radiation is causally associated with cutaneous malignant melanoma (CMM) using
estimates of cumulative sun exposure. The estimates of sun exposure were
defined somewhat differently from study to study, but all were based on
individual interviews or questionnaires and, as such, provided an assessment
of individual exposure. Two early studies (Lancaster and Nelson 1957; Gellin
et al. 1969) might be considered crude by current epidemiological standards,
but they are historically important to the development and exploration of the
hypothesis. Three other case-control studies of CMM investigated total or
cumulative sun exposure in Western Australia (Holman and Armstrong 1984a,
1984b, Holman et al. 1986), in Queensland, Australia (Green 1984), and in
Western Canada (Elwood et al. 1985). The Western Australia and Western Canada
studies are the largest case-control studies of CMM to date, and unlike most
other studies, they controlled for important confounding variables, such as
pigmentary factors.
An early case-control study of melanoma in Sydney, Australia (Lancaster
and Nelson 1957) estimated total sun exposure by means of a scoring index
based on the factors: length of life in Australia, occupational exposure,
industrial hazards, war service and outdoor sports. This study is of
historical importance because it was the first epidemiological study of CMM to
measure sun exposure in individuals. Interview results from 173 CMM cases,
173 non-melanoma skin cancer controls (age- and sex-matched to the cases) and
173 non-skin cancer controls (also age- and sex-matched to the melanoma cases)
showed a higher proportion of melanoma cases with "excessive" sun exposure
(26.6 percent) as compared with the other skin-cancer controls (21.4 percent)
and the non-skin cancer controls (16.2 percent). The authors concluded that
this finding in addition to others supported the hypothesis that sun exposure
is important factor in the occurrence of CMM.
Gellin et al. (1969) analyzed data from interviews of 79 cutaneous
melanoma cases and 1,037 unmatched controls (non-tumor skin conditions) during
1955-1967 in New York. Results showed that 68 percent of the melanoma
patients reported spending three or more hours per day outdoors as compared
with only 37 percent of the controls (p<0.01), a finding which authors
stated "raises the question anew of the role of sunlight in the pathogenesis
of this .form of cutaneous malignancy."
Green (1984) analyzed interview data from 183 melanoma patients and 183
matched population controls (age +5 years, sex, residence) in Queensland,
Australia, to investigate the "relationship between cumulative hours of solar
ultraviolet B (UV-B) radiation and melanoma (excluding lentigo maligna
melanoma)." Total hours of sun exposure (as a surrogate for UV-B) were
estimated by summation of reported occupational and recreational sun hours
from 10 years of age onward. Results showed the cases to be more heavily
DRAFT FINAL * * *
-------
9-2
exposed to the sun than controls -- the relative risks increased from 3.2 (95%
C.I., 0.9-12.4) for intermediate exposure (2,000-49,999 hours) to 5.3 (95%
C.I., 0.9-30.8) for 50,000 or more hours of sun exposure when compared with
less than 2,000 hours after adjustment for exact age, presence of nevi on
arms, hair color, and sunburn propensity. The intermediate exposure category
was, however, extremely broad and it is not known how this may have affected
the results or how many cases and controls were in the highest group.
Eliminating exposures before 10 years of age could also have affected the
results if, in fact, early sun exposures are important to the development of
melanoma (see Chapter 9). Further analyses using actinic skin damage
(keratoses or other skin cancers) as indicators of heavy lifetime exposure to
solar UV radiation showed that the CMM patients had significantly more actinic
lesions on their faces in comparison with controls (p<0.0001) resulting in
an increased CMM relative risk of 2.8 (95% C.I. 1.1-7.2) after adjustment for
exact age and presence of nevi on the arms. Socio-economic factors were not
controlled for and may have biased study findings.
Holman and Armstrong (1984a) analyzed data on cumulative sun exposure at
residence from 511 melanoma cases and 511 matched (age, sex, residence)
controls in Western Australia during 1980-1981. The measure of cumulative sun
exposure was based on location and duration at each residence and mean annual
hours of bright sunshine at each location, resulting in an estimate for
lifetime exposure at home rather than an actual estimates of time spent in the
sun. Analysis showed a significantly decreased odds ratios for migrants to
Australia relative to native Australians, therefore, most analyses were
restricted to native-born Australians. Results (Table 9-1) show significant
positive trends for all CMM (p=0.003) and for SSM (p=0.020). A strong
positive gradient with increasing sun exposure was seen for HMFM with an odds
ratio of 3.78 for the highest exposure group; however, the p-value for trend
(0.101) failed to reach statistical significance probably due to the small
number of pairs with HMF in the analysis.
Data from this study on sun exposure were-also examined by age for high
levels (>2,800 mean annual hours of bright sunlight) of exposure during any
age period (0-9, 10-24, 25-39, >40 years). The greatest risks for SSM were
observed for high levels of sunlight in the 10-24 year age group (OR = 11.31,
95% C.I. 1.40-91.11) and in the 25-39 year age groups (OR = 3.40, 1.41-8.20).
Elevated risks were seen for HMFM in each of the age groups but were not
significant, probably due to small numbers.
While the measure of sun exposure in these analyses was basically
ecological in nature, i.e., annual hours of bright sunlight at residence, an
analysis of actinic skin damage by cutaneous microtopography showed an
increasing risk for all CMMs with worsening actinic skin damage (Table 9-2).
Analysis by histogenetic type showed significantly increased risks with
increasing grade of actinic damage for both HMFM and SSM. History of
non-melanotic skin cancer also resulted in an increased risk of melanoma, as
shown in Table 9-3. The increased risk of melanoma with history of
non-melanoma skin cancer remained even after control for pigmentary factors
(OR = 2.S7, 95% C.I. 1.64-5.04, p=0.0002). The findings of increased CMM risk
with increasing actinic skin damage and with previous non-melanoma skin cancer
* * * DRAFT FINAL * »
-------
9-3
TABLE 9-1
ODDS RATIOS AND 95% CONFIDENCE INTERVALS FOR CMM
BY MEAN ANNUAL HOURS OF BRIGHT SUNLIGHT AT RESIDENCE,
RESTRICTED TO NATIVE-BORN AUSTRALIANS
Mean Annual Hours of
a Bright Sunlight at Residence p-Value
Histogenetic Type <2,600 2,600-2,799 >2,800 of Trend
All Melanomas (494) 1.00 1.34 CO.96-1.86) 1.92 (1.16-3.18) 0.003
.HMFM (82) 1.00 1.54 (0.57-4.15) 3.78 (0.54-26.35) 0.101
SSM (259) 1.00 1.17 (0.68-2.00) 2.12 (0.90-4.98) 0.020
UCM (89) 1.00 0.75 (0.31-1.83) 1.41 (0.46-4.33) 0.922
NM (50) 1.00 0.32 (0.04-2.93) 0.32 (0.04-2.59) 0.193
a
Number in parentheses is number of case-control pairs.
Source: Holman and Armstrong (1984a).
DRAFT FINAL * * *
-------
TABLE 9-2
ODDS RATIOS AND 95% CONFIDENCE INTERVALS FOR HISTOGENETIC TYPES OF
CMM BY SKIN CONDITION GRADED BY CUTANEOUS MICROTOPOGRAPHY
a
II i s togcriu t i c Type
Al 1 Melanomas (389)
IIMI' (72)
SSM (198)
UCM (71)
NM ( 34 )
CMT Grade
1-3
1.00 1.64 (0
b
1.00
1.00 1.53 (0
1.00 2.69 (0
1.00 0.64 (0
4
.97-2.78)
.81-2.91)
.70-10.41)
.07-5.55)
1
4
2
0
0
.76
.05
.25
.62
.94
5
(0.97-3.
(0.93-17
(1.04-4.
(0.17-2.
(0.06-14
19)
.67)
88)
20)
.10)
2.68
4.37
2.63
1.19
5.17
(1
d
(1
(0
(0
6
.44-4.
.21-15
.14-6.
.32-4.
.27-98
98)
.74)
06)
45)
.69)
P-Va I ue
of Trend
0.003
0.048
0.021
0.652
0.092
a
Number in parentheses is number of case-control pairs.
b
Grades 1-4 combined for baseline because no case of HMF had CMT graded less than 4.
Source: Ho I man and Armstrong (1984b).
-------
9-5
TABLE 9-3
RELATIONSHIP OF HISTOGENETIC TYPES OF MALIGNANT
MELANOMA TO A HISTORY OF NONMELANOTIC SKIN CANCER
Melanoma Type
Odds
Ratio 95% C.I. p-Value
All Melanomas (507 case-control pairs) 3.71 2.11-6.57 0
HMFM (86 pairs)
SSM (267 pairs)
UCM (89 pairs)
NM (51 pairs)
. 5.25 1.71-18.03 0
3.33 1.27-9.26 0
3.00 1.02-9.42 0
5.00 0.57-113.13 0
.000001
.001
.011
.044
.221
Source: Holman and Armstrong (1984a).
» * -- DRAFT FINAL -'•
-------
9-6
was seen to support an association between cutaneous melanoma and sun exposure
with the strongest associations for SSM and HMFM (Holman and Armstrong 1984b).
In contrast with these results, a more recent analysis of total outdoor
exposure from the same study population (Holman et al. 1986) showed that, with
the exception of HMFM, all histologic types of melanoma appear to be inversely
associated with total outdoor exposure (estimated by mean weekly total
occupational and recreational outdoor sun exposure averaged over a working
life) after control pigmentation factors. HMFM showed slightly elevated risks
in the two highest exposure categories (OR = 1.40 for 16-22 hours/week; OR =
1.32 for >23 hours/week) but neither odds ratio was statistically significant
nor was there a significant trend. The inverse associations for the remaining
histogenetic types (SSM, UCM, and NM) were also not significantly different
from the baseline exposure group (0-10 hours/week). Analysis of the
recreational sun exposure variable in this study (Holman et al. 1986) is
discussed in Chapter 10 as a measure of intermittent sun exposure.
A case-control study of CMM in Western Canada (Elwood et al. 1985)
examined histories of sun exposure from occupational, recreational and
vacation activities for 595 melanoma patients and 595 matched population
controls (matched on age, sex, and province of residence). A significant
increase in risk with increasing sun exposure from recreational and vacation
activities was'found even after adjusting for hair color, skin color, history
of freckles and ethnic origin (p<0.01). For occupational exposure, no trend
was observed and the only elevated risk was in the mild exposure group
(approximately 8 hours per week) (RR = 1.8, 95% C.I. 1.2, 2.5). Analysis of
sun exposure from all sources combined showed some elevated risks in the
higher groups when compared with the lowest, but none were statistically
significant, nor was there a significant trend of increase.
FINDINGS
Results of case-control studies of CMM and total sun exposure seem to vary
by the measure used to estimate the exposure, and may be affected by
adjustment for pigmentation factors such as hair and skin color, propensity to
sunburn and ethnic origin; adjustment for these factors tends to lower the
risk estimates. Two early studies showed significantly higher proportions of
melanoma cases than controls with high sun exposure, although pigmentary
factors were not considered in the analyses.
Findings based on the information from the three most recent case-control
studies are presented below:
9.1 A study from Western Australia which controlled for
the potentially confounding effects of pigmentary
factors found significantly elevated odds ratios for
total CMM, SSM and HMFM associated with increased
annual hours of bright sunlight at residence,
increasing actinic skin damage and previous
non-melanoma skin cancer. The same study found no
* DRAFT FINAL *
-------
9-7
increased risk for CMM or any histogenetic type of CMM
with increasing total outdoor exposure in summer as
measured by mean weekly total occupational or
recreational sun exposure averaged over working life.
9.2 In Queensland, Australia, elevated CMM risks were
associated with increasing estimated total hours of
sun exposure after 10 years of age, while controlling
for exact age, presence of nevi on arms, hair color,
and sunburn propensity. The confidence intervals for
intermediate and higher levels of exposure included
unity.
9.3 In Western Canada, an analysis of total sun exposure
showed some increased risks in higher exposure groups
compared with the lowest exposure group but none were
statistically significant nor was there a significant
trend of increasing risk.
9.4 Studies which have evaluted the association of CMM
with a measure of delivered dose of UV radiation,
(presumably modified by an individual's susceptability
to solar radiation), have shown an increased CMM risk
associated with increased sun damage to the skin, even
when a consistent association with cumulative exposure
(as assessed by questionnaire) was not found.
•- * ••- DRAFT FINAL * * *
-------
9-8
REFERENCES
Elwood, J.M., Gallagher, R.P., Hill, G.B., Pearson, J.C.G. Cutaneous melanoma
in relation to intermittent and constant sun exposure - The Western Canada
melanoma study. Int J Cancer 35:427-433 (1985).
Gellin, G.A., Kopf, A.W., and Garfinkel, L. Malignant melanoma: A controlled
study of possibly associated factors. Arch Derm 99:43-48 (1969).
Green, A. Sun exposure and the risk of melanoma. Aust J Dermatol
25(3):99-102 (1984).
Holman, C.D.J. and Armstrong, B.K. Cutaneous malignant melanoma and
indicators of total accumulated exposure to the sun: An analysis separating
histogenetic types. JNCI 73:75-82 (1984a).
Holman, C.D.J. and Armstrong, B.K. Pigmentary traits, ethnic origin, benign
nevi, and family history as risk factors for cutaneous malignant melanoma.
JNCI 72:257-266 (1984b).
Holman, C.D.J., Armstrong, B.K., Heenan, P.J. Relationship of cutaneous
malignant melanoma to individual sunlight-exposure habits. JNCI 76:403-414
(1986).
Lancaster, H.O. and Nelson, J. Sunlight as a cause of melanoma: A clinical
survey. Med J Aust. April 6:452-456 (1957).
--• * -•'• DRAFT FINAL
-------
CHAPTER 10
SKIN PIGMENTATION AS A RISK FACTOR
Skin color plays an important role in the determination of ultraviolet
radiation (UVR) effects such as erythema or sunburn. Resistance to these
effects is conferred by racially determined pigmentation of the skin or by
temporary pigmentation from preventive tanning which increases melanin and
thickens the stratum corneum and epidermis in areas of sun-exposed skin.
Fair-skinned people require 3-5 times less UVR to induce erythema than do
those with moderately pigmented skin and up to 30 times less than darkly
pigmented people (Parrish et al. 1983). Melanin also plays a protective role
in the development of basal and squamous cell carcinomas of the skin. This is
seen in the consistent negative association between these effects and skin
pigmentation, although the precise mechanisms of protection are not known.
There are also marked differences in cutaneous malignant melanoma (CMM)
incidence by skin color, with the disease rates varying by the degree of
pigmentation. Epidemiologic evidence from several countries is consistent and
shows a clear-cut difference between white and non-white races in the
incidence of melanoma. The study of pigmentation differences and melanoma
incidence within the Caucasian race provides an opportunity to investigate
whether the protective effects of increased pigmentation also moderate the
risk of malignant melanoma. Increased pigmentation is known to protect
individuals from acute effects on skin exposed to UV radiation, and results
showing similar protective effects against melanoma may provide indirect
evidence towards the role of sunlight or UV radiation in the etiology of
melanoma.
This chapter reviews studies of racial differences in melanoma incidence
from many countries as well as investigations of differences within the
Caucasian or white population. The latter studies go beyond the basic
white/non-white differences in pigmentation and do not assume that all white
skin has the same amount (or lack) of protective melanin. Several
epidemiologic measures have been used to define skin differences within
Caucasian populations which may alter the susceptibility to melanoma in the
presence of a causal factor such as sunlight or UV radiation exposure. These
measures include skin color, hair color, eye color, freckling or a tendency to
freckle upon sun exposure, skin reaction to sun (tendency to sunburn, ability
to tan), and ethnicity (used to estimate skin color because of the genetic
dominance of certain pigmentation characteristics).
RACIAL DIFFERENCES
The International Agency for Research on Cancer (IARC 1976) collected and
published data on cancer incidence in five continents, including both age-
specific and age-standardized incidence rates from 59 population-based cancer
registries in 27 countries. Crombie (1979) analyzed the IARC data on
malignant melanoma using incidence rates standardized to the 1950 world
population (Segi 1960) and found a statistically significant 3-fold increase
in the mean CMM incidence in whites over that of nonwhites (Table 10-1).
» DRAFT FINAL * »
-------
10-2
TABLE 10-1
INCIDENCE OF MALIGNANT MELANOMA FROM 59 POPULATION-BASED
CANCER REGISTRIES ON FIVE CONTINENTS BY WHITE/NON-WHITE
STATUS* STANDARDIZED TO 1950 WORLD POPULATION
White Non-White
Incidence per 100,000
Both Sexes 2.9 0.8
Male . 2.6 0.8
Female 3.2 0.8
Number of Registry Populations 48 26
* The years of reporting varied by registry but were
within the range 1960-73, with the most common
reporting period being 5 years.
Source: Segi (1960).
DRAFT FINAL * * »
-------
10-3
Crombie's analysis assigns populations from the IARC data into white and
nonwhite categories in two ways: seven of the registries are separated by
racial group (Crombie treats each of these as an independent population) and
the other registries are assigned the white/nonwhite status presumably by
using the information on racial and ethnic background provided to IARC by each
registry (IARC 1976). Crombie recognized that the category, "nonwhite", is
not completely satisfactory because of the heterogeneous nature of the racial
groups included, but the small numbers of melanoma cases for some of the
nonwhite populations made it impossible to analyze them as separate racial
entities. For further analysis by race, the registries were treated as
samples from the same population and were grouped to obtain age-specific
incidence rates for the composite population. The lowest melanoma incidence
rates were found among Asian populations while rates among blacks showed much
variation (lower in North America, higher in Africa) (Crombie 1979). •
Malignant melanoma in Africans was mostly on the lower limb, frequently on the
foot (Crombie 1979, Kiryabwire 1968, Malik 1974) and may account for the
excess incidence in Africans over the Asian groups (see Chapters 5 and 13 for
more detail on CMM variation by site).
A study of hospital records in South Africa (Rippey and Rippey 1984) for
the years 1959-1970 found that CMM was almost six times more common in whites
than in blacks. For both sexes combined, the incidence per 100,000 was 6.2
for whites and 1.1 for blacks. By sex, the difference between races remained,
but the ratio of white to black was about 13 to 1 for males (5.3/0.4), whereas
for females it was only about 4 to 1 (7.0/1.7). These racial differences are
consistent with patterns- elsewhere in the world, but the incidence rates may
not be representative of the true South African rates, particularly for blacks
since there are no reliable population estimates for this racial group.
In 1968, Kiryabwire reported an incidence rate of 1.5/100,000 for melanoma
in Uganda based on 152 cases diagnosed during 1963-1966. These cases were
likely to be black because the author stated in the introduction that CMM is
not just "a condition of the white races" and that he would refute this
"impression on world-wide statistical analysis."
Another African study reported on a population of mixed races in Sudan
(Malik et al. 1974) based on a review of records from two laboratories
providing histopathology services for the country. The population was
reported to be Arabic, Hamitic and Semitic with a variable admixture of
Negroid blood. The southern Sudan, where the population is almost exclusively
Negroid, had a lower proportion of CMM diagnoses than in other parts of the
Sudan. Fifty-two percent of the melanomas occurred in patients from two
provinces in the western part of northern Sudan, an area which ethnically is
"a mixture of Arabs and Negroids with some preponderance of the latter." CMM
incidence rates were not calculated due to factors prevalent in many third
world countries, e.g., disparity in availability of medical services and
diagnostic facilities in different parts of the country, inadequate health
certification, ana failure to seek medical attention due to ignorance or
shyness in segments of the population.
In each of the three previously mentioned studies in Africa, i.e., South
Africa (Rippey and Rippey 1984), Uganda (Kiryabwire 1968), and Sudan (Malik et
al. 1984), over 60 percent of all malignant melanomas were on the foot,
-:• * * DRAFT FINAL * * *
-------
10-4
usually on the sole, a less pigmented area of the body in blacks. This factor
is discussed in more detail in Chapter 5, but the high occurrence of CMM on
the foot in blacks (including American blacks) may indicate the presence of
other risk factors in blacks, which may or may not be UVR- or pigment-related.
Data from the U.S. National Cancer Institute (NCI) on incidence of
malignant melanoma by white/nonwhite racial groups show more striking
differences than results from the IARC analyses. The NCI's Surveillance,
Epidemiology, and End Results (SEER) Program reported cancer incidence for
1973-1983 from 10 cancer registries (5 entire states and 28 counties
representing 5 major metropolitan areas within 5 other states) (NCI 1985). In
1983, the age-adjusted incidence rates (1970 U.S. standard population) were
9.6/100,000 in white males and 0.5/100,000 in black males, with a white:black
ratio of 19; in white women, the age-adjusted incidence rate was 8.0/100,000
compared to 0.81/100,000 in black women, with a white:black ratio of 10. Over
the period 1974-1983, incidence rates remained stable in blacks while rates
increased by 40.5 percent in white males, and by 27.9 percent in white females.
Other U.S. data also show a much higher proportion of malignant melanoma
in whites than in other racial groups. A 1980 survey of 614 hospitals studied
4,545 melanoma patients (representing one-third of the total estimated 14,100
cases diagnosed in 1980) and found that 98 percent of the patients were white
and less than 1 percent (37 patients) were black (Balch et al. 1984). A
retrospective review of melanoma cases at the Duke University Comprehensive
Cancer Center during 1972-81 showed similar results. Of the 2,612 patients
with melanoma during this period,.only 31 (1 percent) were black (Reintgen et
al. 1982, 1983). This does not represent a true proportion of black melanoma'
cases in this geographic area since the Duke Center receives many referral
patients from various geographic areas. However, the authors adjusted the
melanoma case ratio (83 whites:! black) by the white to black patient ratio at
Duke (4:1) and found that the estimated ratio of white to black melanoma cases
remained high: 20 to 1.
Both New Mexico and Hawaii have particularly high CMM incidence rates in
their white populations, showing a significant excess of cases in whites over
other races. A study of malignant melanoma from the New Mexico Tumor Registry
during 1969-1977 showed incidence rates for non-Hispanic whites of 8.7 and
9.0/100,000 in non-Hispanic white males and females, respectively (Pathak et
al. 1982). These rates exceeded the total U.S. white rates from comparable
years based on data from the Third National Cancer Survey (TNCS) and SEER, and
were approximately 6 times higher than the rates for other ethnic groups in
New Mexico (Hispanic, American Indian, and black). The TNCS and SEER results
are not fully comparable to the New Mexico results because of the inclusion of
Hispanics in the "white" racial group.
Hawaiian Caucasians also have a high incidence of malignant melanoma. For
1968-1972, IARC (1976) shows age-adjusted (1950 world standard) incidence of
6.8 per 100,000 for Caucasian males and 5.7 for Caucasian females while the
rates for the remaining population of Hawaii were less than 1.0. Hinds and
Kolonel (1980, 1983) analyzed data on malignant melanoma from the Hawaii Tumor
Registry and showed an excess of cases in the Caucasian population relative to
other Hawaiians as well as a steady increase in the rate of melanoma diagnosed
among whites during the study period, 1960-1980. Over this time period, SO
'- DRAFT FINAL •• - -
-------
10-5
percent of the diagnosed malignant melanoma patients were white, even though
the white population constituted only 34 percent of the state's population in
1970. Also during this time period, the incidence rates for whites more than
tripled (for 1978-80, 24.0 and 19.5/100,00 for males and females,
respectively), while the incidence among nonwhites remained fairly stable.
A similar situation exists in New Zealand, the country with the second
highest incidence of melanoma in the world. Moss (1984) reviewed records from
the National Cancer Registry of New Zealand for the years 1963-1981. During
this time, the non-Caucasian population of New Zealand (Maoris and
Polynesians) had a CMM incidence rate of 2.9/100,000, while the incidence rate
in the Caucasian population was 16.9/100,000, a nearly six-fold difference.
Israel has maintained a central cancer registry since 1960 and reports
differences in CMM incidence between European-born (including American-born)
immigrants and African- or Asian-born immigrants. Movshovitz and Modan (1973)
reported that among all foreign-born residents of Israel, CMM incidence was
much higher for European-born than for Asian-African-born based on data from
1961-1967. Anaise et al. (1978) analyzed all melanoma cases in the total
Israeli Jewish population during 1960-1972 and found incidence rates of 3.4
among European-born, 0.44 among Asian-born, and 0.27 among African-born (rates
per 100,000). These differences by country of origin may relate more to
ethnicity and skin color variation within Caucasians than to actual racial
(Caucasian/Negroid) differences due to the racially heterogeneous population
immigrating from Africa.
DIFFERENCES WITHIN CAUCASIANS
While it has been shown that malignant melanoma is more prevalent in the
white or Caucasian population than in nonwhite populations, there are
differences in melanoma incidence within the white population which have led
researchers to search for more definitive constitutional risk factors than
white race. The assumption that all white populations have the same risk of
melanoma may lead to errors in the extrapolation of study results to other
white populations. For example, Lee and Isenberg (1972) compared CMM
incidence and mortality rates from England and Wales with those from Sweden to
determine whether the different latitudes of two countries (and indirectly,
the intensity of solar radiation) might be associated with differences in
melanoma as found in earlier studies (e.g., Lancaster 1956). The authors
assumed that two white populations with similar latitude might be expected to
have similar skin cancer incidence rates if latitude was related to skin
cancer, and that a lower latitude would result in an increased incidence
rate. In fact, Sweden's rates were higher than those for England and Wales
although Sweden is situated at a higher latitude. This might occur, however,
if the white populations were not comparable with respect to CMM risk factors,
such as tanning ability or sun exposure habits. The authors stated that their
results suggested genetic or occupational effects. This genetic contribution
to CMM risk may be further investigated by studying skin color differences
within the Caucasian population.
Many epidemiological studies have been conducted to further identify risk
factors associated with increased incidence of CMM in Caucasian populations.
* * * DRAFT FINAL * * *
-------
10-6
The variables most frequently investigated for pigmentation differences which
may alter susceptibility to CMM within white populations have been:
• Skin color - self assessed or assessed by an
interviewer, either by observation or by comparison with
a color chart
• Hair color - in childhood or adolescence, or at study
time
• Eye color
• Freckles or tendency to freckle upon sun exposure
• Reaction to sun - tendency to sunburn, ability to tan
• Ethnicity
Most of the above risk factors are interrelated requiring controlled analyses
to evaluate their independent effects on risk of CMM. Table 10-2 summarizes
the above-mentioned measures of skin pigmentation according to the studies
which investigated each of these factors. The study findings associated with
these pigmentation variables are discussed in the following sections.
Skin Color
As seen in Table 10-2, skin color was a significant risk factor in all of
the studies that considered it. Using all major hospitals in Sydney,
Australia, Lancaster and Nelson (1957) age- and sex- matched each of 173
melanoma patients with two control patients (one non-melanoma skin cancer and
one non-skin cancer). All patients were of European descent. Skin color was
determined "subjectively" by the researchers and classified as fair, medium,
or olive. The authors reported a larger proportion of fair-skinned persons in
the melanoma and non-melanoma skin cancer groups than in the other cancer
controls (77 percent, 72 percent, and 63 percent, respectively).
Gellin, Kopf, and Garfinkel (1969) in an age- and sex-matched case-control
study of 79 cases and 1,037 controls at the New York University Medical Center
determined that a significantly higher proportion of cases than controls (50
percent vs. 37 percent) had a fair complexion (p<0.05) based on
self-assessment. When analyzed by sex, the greatest difference in the
proportion of those with fair complexions was between female cases and
controls (61 percent vs. 41 percent, p<0.05). Although slightly more male
cases than controls had fair complexions (36 percent vs. 32 percent), the
difference was not statistically significant.
Beral et al. (1933) age-matched 257 white female melanoma cases (18-54
years old) treated in Sydney's Melanoma Clinic during 1975-1980 with'574 white
female controls and found fair skin to be associated with a two-fold risk of
CMM (RR=1.9, 95°o C.I. 1.3S-2.50) relative to medium or olive skin, after
adjusting for hair coJor. Fair skin was a risk factor independent of hair and
eye color, although its importance depended to a certain extent on the
associated hair color. For those with red hair, fair skin had only a small
••> * •'•- DRAFT FINAL * - -
-------
10-7
TABLE 10-2
MALIGNANT MELANOMA RISK FACTORS BY MEASURES OF
SKIN PIGMENTATION WITHIN THE CAUCASIAN POPULATION
Measures of Skin Pigmentation
Within Caucasian Population
Reaction
to Sun
Skin Hair Eye (Tanning/
Study Reference Color Color Color Freckling Sunburn) ' Ethnicity
Lancaster and Nelson, 1957 + + + +
Gellin et al., 1969 + + + +/-
Lane Brown et al., 1971 +
IARC, 1976 +
MacDonald, 1976 +
Klepp and Magnus, 1979 - • - + +
Mackie and Atchinson, 1982
Beral et al., 1983 + + - +. +
Hinds and Kolonel, 1982 +
Lew et al., 1983 - +
Elwood et al., 1984 + + + + +
Holman and Armstrong, 1984 + + + + + +
Graham et al., 19S5 + + + + +
NOTE: + = Significant risk factor
- = Not significant risk factor
Blank = Not included in studv
* -• - DRAFT FINAL •'• •• *
-------
10-8
additional effect, increasing the risk by 39 percent; for those with blonde
hair, the risk was increased by 80 percent; but for those with black or brown
hair, the risk was increased by 108 percent.
A case-control study by Elwood et al. (1984) in Western Canada found a
relative risk of 2.4 (p<0.05) for light inner arm skin color as opposed to
dark based on data from 595 melanoma cases and 595 population-based controls
(matched on age, sex and province of residence).
A three-fold greater melanoma risk- was observed for those with the
lightest skin relative to those with darkest pigmentation (OR=3.07, 95% C.I.
1.47-6.39), as well as a significant (p<0.1) linear trend in odds ratios
among the four skin pigmentation groups.
In a study based on 499 case-control pairs (matched on sex, 5-year birth
period and area of residence) in Western Australia, Holman and Armstrong
(1984) measured skin color reflectance of the left upper inner arm as an
indicator of skin color. For each histogenetic type (HMF, SSM, UCM, or NM),
elevated odds ratios were associated with fair skin color of the upper inner
arm relative to dark skin.
A case-control study (404 cases, 521 unmatched controls) of CMM patients
at Roswell Park Institute (NY) during 1974-1980 (Graham et al. 1985)
consistently found increased risk with increasing fairness in skin, hair, and
eye color. .Using self-assessment of fair, medium, and dark, both male and
female cases had elevated .risks for fair as opposed to those with dark or
medium complexions and found significantly elevated odd ratios for fair
complexions in both males and females (p<0.01).
Hair and Eye Color
Six of eight studies found hair and eye color to be significant risk
factors for melanoma within Caucasian populations in six. Regardless of eye
color, Beral et al. (1983) found that red hair was associated with a three- or
four-fold increase in melanoma risk in women. Individuals with fair skin and
red hair had a slightly higher risk (RR=4.4) than those with dark skin and red
hair (RR=3.2). All risk analyses were conducted with reference to women with
brown or black hair and medium or olive complexions. Blonde hair was
associated with a relative risk of 2.7 in fair-skinned women and 1.5 in
dark-skinned women. Significantly elevated risks for red hair in childhood
(RR=3.0, 95°0 CI 1.95-4.73) and for blonde hair in childhood (RR=1.6, 95% CI
1.15-2.14) were also noted. Results from analyses of combinations of hair,
eye, and skin color show significant relative risks for all women with red,
blonde, or fair skin (Table 10-3). Eye color had no independent effect on
risk.
Klepp and Magnus (1979) studied 78 Norwegian CMM cases and 131 unmatched
"other" cancer controls in 1974-75 and found no difference between cases and
controls with respect to hair and eye color. As mentioned by the authors,
these results may be explained by Norway's very homogeneous population of
fair-skinned individuals; what may be recorded in other more mixed populations
as fair skin or light brown hair may be assessed as "dark" by the
* * -• DRAFT FINAL * •- »
-------
10-9
TABLE 10-3
REPORTED HAIR COLOUR (AT AGE 5), SKIN COLOUR AND
EYE COLOUR IN CASES AND CONTROLS, AND RELATIVE RISK
ASSOCIATED WITH EACH COMBINATION
(Data missing for three cases and six controls)
Hair colour
Skin colour
Eye colour
Cases
Relative
Controls risk+
Red
Red
Red
Red
Blonde
Blonde
Blonde
Blonde
Brown or black
Brown or black
Brown or black
Brown or black
Fair
Fair
Medium or olive
Medium or olive
Fair
Fair
Medium or olive
Medium or olive
Fair
Fair
Medium or olive
Medium or olive
Green or brown
Blue
Green or brown
Blue
Green or brown
Blue
Green or brown
Blue
Green or brown
Blue
Green or brown
Blue
22
17
10
2
33
48
23
27
23
20
41.
13
18
15
11
3
48
63
72
51
56
30
151
50
4 . 5**
4.2**
3 . 4**
2.5
2.6**
2 . 8**
1.2**
2 . 0*
1 . 9*
2 . 5**
1.0
1.0
* Differs significantly from 1.0, p<0.05.
** p<0.01.
+ Relative to those with brown or black hair and medium or olive skin.
Source: Beral et al. (1983).
* * * DRAFT FINAL *
-------
10-10
predominantly blue-eyed blonde-haired Norwegians. It is also unlikely that
there was sufficient power to detect small differences in a homogeneous
population based on the small numbers of cases and controls.
Lew et al. (1983) also found no difference in hair and eye color between
111 CMM cases and 107 controls in Massachusetts. This may have resulted from
a bias in the method of control selection. Each patient was to provide 2-3
friends of the same age (+5 years) and sex for interview, however, 46 cases
provided 0 controls, 30 provided 1 each, 28 provided 2 each, and 7 provided 3
each. With no more information on these populations (the authors state that
sex and median age were comparable between cases and controls), it is
impossible to know what biases may have been introduced.
Lancaster and Nelson (1957), in their study of 173 melanoma cases' with
age- and sex-matched non-melanoma skin cancer controls and other cancer
controls (173 of each), found an excess of red-haired and fair-haired melanoma
patients (42 percent) when compared with the non-melanoma skin cancer (36
percent) and other cancer (29 percent) patients. They also found more
light-eyed (blue or green-gray) patients among the melanoma and non-melanoma
skin cancer patients than among the other cancer controls, 61 percent, 66
percent, and 46 percent, respectively.
Gellin et al. (1969) found similar significant differences between their
79 melanoma cases and 1,037 unmatched controls in eye color (55 percent of
cases vs. 35 percent of controls had blue or green/gray eyes) and in hair
color (26 'percent of cases vs. 9 percent of controls had white, blond, or red
hair.
In a Canadian population of 595 white melanoma cases and 595 matched (age,
sex, and province of residence) controls, Elwood et al. (1984) found the
highest relative risks associated with hair color: 7.1 (95% C.I. 2.6-19.2)
for blond hair in childhood and 3.7 (95% C.I. 1.8-7.7) for red hair as
compared with black hair in childhood, adjusting for skin and eye color. Eye
color was not independently associated with risk of CMM (adjusting for hair
and skin color); this finding is consistent with that of Beral et al. (1983)
and Klepp and Magnus (1979). The authors stated that of all the pigmentation
variables examined (hair color, skin color of upper inner arm, eye color, and
freckles in adolescence), hair color showed the strongest association with an
increased risk of malignant melanoma.
Results in an Australian study of 511 melanoma patients and 511 controls
matched on sex, 5-year birth period and area of residence (Holman and
Armstrong 1984) also showed positive results for hair color while eye color
did not contribute to CMM risk after controlling for other pigmentary
characteristics. Persons with red hair had nearly twice the risk of those
with dark hair (OR=1.89, 95?0 C.I. 1.26-4.30), while persons with blond or
light brown hair showed intermediate, but significant, levels of risk (OR=1.56
and 1.24, respectively); these findings were of borderline significance. When
risk factors were examined by histogenetic type of melanoma, hair color was
found to be related to each of the four types (HMFM, SSM, UCM, or KM), while
eye color was related to only SSM and NM.
* ••- •• DRAFT FINAL * •'• »
-------
10-11
The results of Graham et al. (1985) using data on 404 melanoma patients
and 521 unmatched hospital controls in New York State indicated there were
excesses of CMM for both males and females who had blonde or red hair in
childhood with blue eyes and fair complexions. When hair, skin, and eye color
were analyzed together, the odds of melanoma increased as the "lightness in
tone" increased. Red hair showed a significantly elevated odds for males
(OR=2.45, p<0.05) and both red and blond hair showed positive associations
with CMM in females (OR=3.99, p<0.10 and 2.14, p<0.01, respectively). Eye
color showed similar results; males with blue eyes and females with blue or
blue-green/gray eyes all showed significantly elevated odds ratios relative to
persons with brown eyes.
Freckling
Of the epidemiologic studies of melanoma reviewed, only four investigated
freckling as a risk factor and all found a positive association of CMM
incidence with freckling. Although these studies defined this risk factor
somewhat differently, their results were consistent regardless of how
freckling was ascertained.
Using data on 78 cases and 131 unmatched, non-skin cancer controls, Klepp
and Magnus (1979) found significant odds ratios for persons who responded
positively to the question: "Do you have freckles, or do you freckle
easily?" The risks were particularly high in those aged 20-49 years (RR=3.94
for males and 4.88 for females). In the 50 years and older group, only
females maintained an elevated risk (RR=2.06).
In a study of CMM in white Australian women (287 cases and 574 age-matched
controls), Beral and associates (Beral et al. 1983) found a crude odds ratio
of 1.9 for those who reported that they usually freckled after a 30-minute
exposure to midday summer sun relative to those who sometimes or never
freckled. After adjustment for hair and skin color, however, the odds ratio
associated with freckling was of borderline significance (OR=1.4, 95% C.I.
1.00-1.95).
Elwood et al. (1984) identified freckles as an important host factor in
the development of malignant melanoma, based on analysis of data on 595 age-,
sex-, and residence-matched case-control pairs. The risk associated with
heavy freckling in childhood and adolescence (RR=2.6) remained significant
(RR=2.1, p<0.001) even after adjustment for other pigmentation factors such
as hair, skin and eye color, sun reaction, and ethnic origin.
Holman and Armstrong (1984) questioned study subjects (511 age-, sex-, and
residence-matched case-control pairs) regarding their reactions to chronic sun
exposure and found a significantly elevated risk for those who only freckled
or never tanned relative to those who tanned deeply (OR=3.53, 95% C.I.
1.82-6.84). After controlling for other pigmentary characteristics (acute
reaction to sun, hair, skin and eye color), the risk associated with inability
to tan remained significant (OR=2.44, 95% C.I. 1.19-5.02).
DRAFT FINAL * » *
-------
10-12
Reaction to Sun Exposure
Nine of studies reviewed assessed reaction to sun exposure. Most studies
reported positive associations, however, one (Mackie and Atchison 1982)
reported no significant elevation in risk for melanoma in the sun-sensitive
groups and another (Gellin et al. 1969) reported no association of CMM and
sun-sensitivity in males.
In Mackie and Atchison's study, the skin-types included four categories
ranging from "always burns, never tans" (Type 1) to "always tans, never burns"
(Type IV). No significant differences between CMM patients and controls were
found, either as a group or separately by sex. However, this study had a
small sample size (113 cases and 113 controls) and was conducted within a
relatively homogeneous population (Western Scotland) where 75 percent of the
cases and 70 percent of the controls were skin types I or II. These facts
should be considered in the interpretation of Mackie and Aitchison's results;
the lack of a significant finding does not preclude a potential relationship
between severe reaction to sun exposure and CMM.
In a U.S. study of 79 CMM cases and 1,037 unmatched controls, Gellin et
al. (1969) found a significant difference only for females. Fifty-six percent
of the cases and 38 percent of controls said they sunburned easily and
conversely, 9 percent of cases and 25 percent of controls said they tanned
easily. Gellin et al. (1969) reported a higher proportion of male controls
(33 percent) than male cases (21 percent) who said they sunburn easily and a
lower proportion of controls who tan easily (25 percent vs. 41 percent of the
cases). No explanation was provided for this result based on the analysis of
data on 34 male cases and 405 male controls with nontumor skin conditions,
although the result may be due to control selection of only patients with
other (nontumor) skin conditions if those conditions were related to the
factors under study, such as reaction to sun exposure and skin type.
In a hospital-based individually matched case-control study in Sydney,
Australia, Lancaster and Nelson (1957) found that 62 percent of 173 melanoma
patients reported that they burn easily with sun exposure, as compared with 54
percent of the non-melanoma skin cancer controls and only 36 percent of the
other cancer controls.
Responses to the question "How much sun do you tolerate?" in a Norwegian
case-control study of CMM (78 cases and 131 unmatched controls) showed
elevated risks for those who answered "very little" or "not very much" versus
those who replied in three of the four age-sex groups (Klepp and Magnus
1979). Relative risks for cases were 6.08 for males 20-49 years, 2.42 for
females 20-49 years, and 2.00 for females 50 years and over. Only males 50
and over did not have an increased risk associated with low sun tolerance.
Eeral et al. ("1983) found a slightly elevated but significant relative
risk for cases (RR=1.4) who reported "blistering or peeling" after a 30-minute
exposure to midday summer sun with no tan versus those reporting milder
reactions. After adjustment for skin and hair color, however, the relative
risk for this factor was not significant (RR=1.1).
* * DRAFT FINAL
-------
10-13
Results from Lew et al. (1983), based on data from 111 melanoma cases and
107 unmatched controls, should be interpreted with caution due to the
potential biases from their method of control selection described earlier;
their findings, however, are in agreement with results from several other
studies -- the risk of melanoma was significantly increased among those who
had difficulty tanning as an adolescent relative to those who did not.
Elwood's study in Canada (1984) also showed a significant risk (RR=2.3) for
those who sunburned and rarely tanned. The risk remained significantly
elevated even after adjustment for other significant pigmentation
characteristics such as skin and hair color, freckling and ethnicity (RR=1.7,
p<0.01).
Holman and Armstrong (1984) separated burning and tanning into acute
(blister) and chronic (freckle no tan) reactions to sunlight and found these
two sun-sensitive reactions to be the most significant of the pigmentary risk
factors for melanoma in their Australian case-control study (511 matched
pairs), even after adjustment for hair, skin, and eye color. Burning and
tanning were also analyzed separately by Graham et al. (1985) who showed a
two-fold significant odds for those who burned or freckled versus those who
did not (OR=1.97 for males, 2.02 for females) and a two-fold risk (OR=1.95)
for females who answered "no" to tanning versus those who responded
affirmatively (for males, the increased risk of 1.65 was not statistically
significant).
Hereditary Differences
Ethnic background within the Caucasian population has a major role in skin
pigmentation and, as such, has been analyzed to identify the presence of a
relationship with melanoma incidence. Some early studies analyzed ethnicity
using descriptive techniques on incidence data finding lower rates of CMM for
ethnic groups having darker skin tones than other Caucasian ethnic groups,
e.g., Spanish or Portugese.
IARC (1976) published melanoma incidence rates from five continents.
These include data on populations from New Mexico (1969-1972) and El Paso,
Texas (1968-1971), separated into Spanish-origin whites and other whites. The
following age-standardized (1950 world population) CMM incidence rates were
considerably higher for non-Spanish whites than for Spanish whites in each
area:
Age-Standardized Incidence/100,OOP
Male Female
New Mexico: Spanish 0.8 0.9
Other White 4.8 5.3
El Paso, Texas: Spanish 0.0 1.0
Other White 3.8 4.8
* * DRAFT FINAL * * *
-------
10-14
MacDonald (1976) found similar results when she analyzed 23 years of
melanoma incidence data from six major regions in Texas (56 counties). Of the
2,328 cases of melanoma, 91 percent occurred among non-Spanish whites, and
only 8 percent among whites with Spanish-surnames.
In Hawaii, a state with very high CMM incidence in whites, Hinds and
Kolonel (1983) attempted to examine differences in melanoma between the
non-Portugese white and Portugese white sectors of the population using
1960-1980 data from the Hawaii Tumor Registry. An estimated 10 percent of
Hawaiian whites were of Portugese ancestry, but no exact population figures
were available. Analysis of proportional cancer incidence during the study
period showed that melanoma accounted for only 0.5 percent of all cancer cases
in Portugese men and 0.2 percent in Portugese women. In contrast, CMM
accounted for 4.7 percent of cancer cases in non-Portugese white men and 3.1
percent in non-Portugese white women.
In a hospital interview study of total skin cancers in Sydney, Australia,
Lane Brown et al. (1971) used surnames and interview questions to identify
ethnic background (Irish as well as Scottish and Welsh Celtic names). The
study compared proportions of persons half or more Celtic (Irish, Scottish,
and Welsh) between different hospital populations. The melanoma and other
skin-cancer groups each showed a higher proportion of persons with Celtic
heritage than those groups without skin cancer. A random sample of 2,607
names drawn from the Sydney telephone directory resulted in only 26 percent
Celtic names, a proportion similar to those in the non-skin cancer hospital
groups. In summary, this showed that Celtic heritage was associated with
higher proportions of patients diagnosed with basal and squamous cell
carcinomas and malignant melanomas than patients hospitalized for reasons
other than skin cancer.
Three recent case-control studies (Canada, Western Australia, and New York
State) found that ethnicity was a significant risk factor for CMM. In Canada,
Elwood et al. (1984) showed a significantly low risk (OR=0.5) among study
subjects from East or South European background compared with those of English
origin. This odds ratio remained significant (p<0.05) even after adjustment
for other risk factors such as pigmentation characteristics and freckling.
Likewise, Holman and Armstrong (1984) found that having two or more Southern
European grandparents resulted in a significantly lower risk of melanoma.
When the analysis was done controlling for age at arrival in Australia and for
pigmentary characteristics, the strength of this protective effect was reduced
to a barely significant level. Graham et al. (1985) found a significantly
increased risk only for females (OR=2.26, 95% C.I. 1.5-3.5) with an ethnic
derivation from Northern European countries (Scandinavia, Poland, Germany,
France, British Isles) when compared to women from other ethnic backgrounds.
FINDINGS
The findings presented below are based on the review of epidemiological
studies in this chapter.
DRAFT FINAL * »
-------
10-15
7.1 Epidemiologic studies have shown that Caucasian populations have
much higher rates of CMM incidence and mortality than black
populations. Based on 1983 SEER data, white:black ratios were
19:1 and 8:1 in males and females, respectively.
7.2 Differences in CMM incidence and mortality have also been observed
between Caucasians and other races. For example, whites in New
Zealand experience much higher incidence rates than new Zealand
Maoris and Polynesians. Likewise, American Indians experience
much lower rates of CMM than American whites.
7.3 Within the Caucasian race, differences in rates of CMM occur
according to country of origin. CMM incidence rates for Hispanic
whites in New Mexico, for example, are much lower than those for
non-Hispanic whites; individuals from the Mediterranean countries
in southern Europe tend to have lower rates than Caucasians from
northern Europe; individuals of Celtic origin in Australia tend to
have higher rates than non-Celtic individuals. Variation in the
incidence of CMM within the Caucasian race is commonly thought to
be a function of variation in genetically-determined pigmentary
traits across ethnic groups.
7.4 Numerous epidemiologic studies have focused on identification of
important pigmentary characteristics in the etiology of CMM. The
following associations were identified in this chapter:
a) Skin color -- fair complexions relative to dark complexions
were associated with elevated risks of CMM in all studies
reviewed.
b) Hair and eye color -- red and blonde hair in childhood
relative to dark hair were associated with increased risk of
CMM in most studies. Blue eyes were an independent risk
factor in only one of four well controlled epidemiologic
studies; however, this could be due to the homogeneous nature
of most of the study populations.
c) Freckling -- those who freckled readily were at consistently
elevated CMM risk relative to other individuals.
d) Reaction to sun exposure -- individuals who usually burned and
were unable to tan were at significantly higher risk of CMM
than those who tanned well in most studies reviewed.
7.5 Over the period 1974-19S3, incidence rates remained stable in
blacks while rates increased by 40.5 percent in white males and by
27.9 percent in white females in the U.S.
-•• DRAFT FINAL * * *
-------
CHAPTER 11
CORRELATIONS WITH SOCIO-ECONOMIC STATUS
AND OCCUPATIONAL FACTORS
INTRODUCTION
In this chapter, three potential risk factors for cutaneous malignant
melanoma (CMM) are discussed. First, the epidemiological studies which have
examined trends in CMM according to socio-economic status and general
occupational classifications are reviewed. Second, studies on the occurrence
of CMM among workers exposed to chemicals or radiation are described. Third,
the epidemiological data on the development of CMM among indoor workers
exposed to fluorescent lighting are reviewed. Although these areas are
addressed separately, they all focus on the potential links between CMM and
occupation.
MELANOMA TRENDS ACCORDING TO SOCIO-ECONOMIC STATUS
Potential relationships between socio-economic status and CMM incidence
and mortality have been analyzed in several epidemiological studies.
Variables used to reflect socio-economic status have included occupational
groups (e.g., professional versus laborer), types of work (e.g., indoor office
versus outdoor), and general indicators such as education and income. The
results of these studies have not produced a clear understanding of the
relationship of CMM to socio-economic status. Several epidemiological studies
have indicated that CMM incidence and mortality are positively related to
socio-economic status (Holman et al. 1980; Lee and Strickland 1980; Mackie and
Aitchison 1982; Cooke et al. 1984; Aquavella et al. 1983; Teppo et al. 1982).
While some studies have shown that outdoor workers do not have an elevated
risk of melanoma compared to office workers (Lee and Strickland 1980; Cooke et
al. 1984), other studies have indicated that outdoor workers have slightly
elevated CMM risks for normally uncovered parts of the body such as the face
and neck (Beral and Robinson 1981; Vagero et al. 1986). Professional and
administrative type office workers, but not other indoor workers, have been
shown to be at elevated CMM risk compared to outdoor workers (Lee and
Strickland 1980; Holman et al. 1980) and to have an elevated risk of CMM on
normally co\7ered parts of the body (Beral and Robinson 1981; Vagero et al.
1986) .
One epidemiological study which specifically examined trends in melanoma
by socio-economic status was conducted by Lee and Strickland (1980). Data on
CMM incidence by occupation were obtained for 1968-1970 from the Supplement on
Cancer to the Registrar General's Statistical Review of England and Wales.
Mortality data for England and Wales were obtained from the Occupational
Mortality Decennial Supplements for the periods 1949-1953, 1959-1963,. and
1970-1972. Standardized mortality ratios (SMRs) were based on census
population statistics corresponding to 1951, 1961, and 1971. Because there
were no population data for the British cancer registration data, the
incidence data were analyzed by proportional ratios which compared the
DRAFT FINAL
-------
11-2
proportion of incident melanomas to all cancers for each occupational group.
The authors noted that the ratios were susceptible to distortion by large risk
differences between occupational groups (e.g., lung cancer), but stressed that
they may be useful in conjunction with other information. Social class
categories consisted of I (professional), II (intermediate), III (skilled
workers including manual, HIM, and non-manual, IIIN), IV (semi-skilled), and
V (unskilled).
Lee and Strickland (1980) observed a general trend of increasing SMRs for
CMM with increasing social class for males from 1949-1972 (Table 11-1). A
comparison of SMRs according to finer occupational groupings (as outlined in
Table 11-2) showed that outdoor workers (e.g., farmers and construction
workers) did not have higher SMRs than indoor workers such as warehousemen,
shopkeepers, or engineers and had lower SMRs than professional, technical,
administrative, and managerial workers. A similar comparison using the CMM
incidence data also indicated higher standardized proportional registration
ratios among professional and administrative workers than among construction
workers, engineers, and warehousemen. Lee and Strickland (1980) concluded
that their results suggested a relationship between CMM incidence and some
feature of life associated with education or economic status. Lee (1982) also
noted that these results showed the lack of a marked effect of outdoor
occupation on CMM mortality.
A study conducted by Holman et al. (1980) also identified differences in
melanoma incidence rat.es according to social class and occupation. Holman et
al. (1980) analyzed melanoma incidence data -for 1975-1976 obtained from
hospital and pathology records in Western Australia. Information on the 120
pre-invasive melanoma (PIM) and 422 invasive malignant melanoma (IMM) cases
included occupation and location of usual residence. Social classes were
assigned from 1 to 4 based on socio-economic data for each residential area.
As shown in Table 11-3, for IMM cases, the highest incidence rates among males
and females occurred in social class 1. Among females, the incidence rates
declined with lower social class] the relationship was more complex among
males. For PIM, a pattern was not evident. Controlling for country of birth
and proximity to sea did not alter the apparent relationship between social
class and IMM incidence. Table 11-4 indicates that the highest incidence
rates occurred among professional, clerical, sales, administrative, and
managerial workers whereas the lowest rates occurred among laborers,
tradesmen, farmers, and fishermen. Holman et al. (1980) observed that
although these results were consistent with an association between melanoma
incidence and social class, they were not what would have been expected if
total exposure to the sun were a predominant causal variable. They
hypothesized that the results could be explained if intermittent (e.g.,
recreational) sun exposure were more likely to induce malignant melanoma than
continuous exposure. In addition, differences in host factors and ethnic
background by social class could also partially explain these observations.
Mackie and Aitchison (1982) conducted a case-control study on 113 CMM
patients presenting with primary CMM in West Scotland from 197S-1980 and 113
age- and sex-matched controls. Matched case-control comparisons were analyzed
using conditional multiple logistic regression. Information on each case
DRAFT FINAL
-------
11-3
TABLE 11-1
STANDARDIZED MORTALITY RATIOS (AND NUMBERS OF DEATHS)
FOR MALIGNANT MELANOMA BY SOCIAL CLASS.
REGISTRAR GENERAL'S OCCUPATIONAL MORTALITY REPORTS
1949-72
Socio-Economic Class
V
IV
HIM
IIIN
II
I
Unskilled
Partly Skilled
Skilled Manual
Skilled Non-Manual
Intermediate
Professional
Male
90 (121)
85 (217)
92 (485)
123 (192)
120 (290)
143 (80)
Female
88 (99)
82 (198)
103 (524)
116 (177)
. 118 (293)
140 (76)
Source: Lee and Strickland (1980).
DRAFT FIN7AL
-------
11-4
TABLE 11-2
STANDARDIZED MORTALITY RATIOS (AND NUMBER OF DEATHS)
FOR MALIGNANT MELANOMA 1959-63 AND 1970-72,
ENGLAND AND WALES BY SELECTED OCCUPATIONAL ORDERS
Occupation Order*
Farmers, Foresters, Fishermen
Construction Workers
Engineering Trades
Warehousemen, Storekeepers, Packers
Clerical Workers
Sales Workers
Administrators and Managers
Professional and Technical
1959-63
90
95
87
85
122
123
115
117
(26)
(19)
(68)
(17)
(49)
(58)
(30)
(49)
1970-72
103
67
87
120
112
127
121
142
(20)
(12)
(64)
(21)
(38)
(49)
(39)
(72)
* Selected occupation order for England and Wales.
Source: Lee and Strickland (1980).
* * * DRAFT FINAL
-------
11-5
TABLE 11-3
AGE-STANDARDIZED INCIDENCE RATES OF PRE-INVASIVE
AND INVASIVE MELANOMA IN THE PERTH STATISTICAL
DIVISION DISTRIBUTED BY SOCIAL CLASS
Males Females
Social Class Number Incidence" Number Incidence
Pre-Invasive Melanoma
1 12 5.7 22 9.2
2 8 4.5 16 9.1
3 8 4.8 8 4.6
4 12 9.3 8 5.8
Invasive Malignant Melanoma
1
2
3
4
63
32
42
26
29.5
18.8
23.7
19.9
58
34
34
24
24.5
19.5
17.8
17.0
"' Rates per 100,000 per year standardized to the age
distribution of the total population in Western Australia in
1976.
Source: Holman et al. (1980).
DRAFT FINAL
-------
11-6
TABLE 11-4
AGE-STANDARDIZED INCIDENCE RATES OF INVASIVE
MELANOMA IN MEN AGED 15 TO 64 YEARS IN
THE WORK-FORCE DISTRIBUTED BY OCCUPATION
Occupation
Professional Workers
Clerical and Sales Workers
Administrators and Managers
Sport and Recreation Workers
Transport and Communication Workers
Labourers and Tradesmen
Farmers and Fisherman
Number*
26
31
26
11
12
40
16
Incidence'"*
39.0
37.3
35.8
32.1
21.9
18.8
18.5
* Excludes 3 men aged 15-64 years whose occupation could not
be ascertained.
** Rates per 100,000 per year standardized to the age
distribution of all men aged 15 to 64 in Western Australia in
1976.
Source: Holman et al. (1980).
DRAFT FINAL
-------
11-7
included positive history of recreational or occupational sun exposure
(classified as 16 or more hours outdoors per week), social class (V, unskilled
to I, professional), and history of severe sunburn. When the data were
analyzed for males and females combined melanoma patients were of higher
social class and had lower recreational sun exposures than controls
(p<0.05). Among males only, socio-economic status and history of severe
sunburn were significantly higher for cases compared to controls (p<0.05).
Occupational exposure was significantly less in males cases than controls
(p<0.05). Among females, the only significant difference was the higher
incidence of severe sunburn among the CMM patient group. Mackie and Aitchison
(1982) concluded that melanoma appeared to be more common among higher
socio-economic classes and for male professional or administrative workers.
The authors questioned the accuracy of the socio-economic status data on
females since these were classified as that of their husbands. They'also
claimed that their results confirmed the hypothesis that isolated episodes of
intense burning sun exposure may be an important factor in melanoma.
In a more recent study, Cooke et al. (1984) examined CMM incidence and
mortality data for 1972-1976 and 1973-1976, respectively for New Zealand
non-Maori males aged 25-64. The 501 incident cases and 142 melanoma mortality
cases were classified according to occupation and then reclassified by
socio-economic status (based on income and education) and average outdoor
occupational exposure (10 or more, 2-10, or 2 or less hours outdoors per
week). Standardized incidence and mortality ratios were calculated for four
10-year age groups by indirect standardization based on 1971 and 1976 census
data. When analyzed by major occupational group (Table ll-5), the observed.
number of incident melanoma cases significantly exceeded the expected number
(p<0.001) for professional/technical and administrative/managerial workers.
The observed number of incident melanoma cases was significantly lower than
expected (p<0.001) for production, transportation, and labor occupational
categories. Smaller differences were noted when similar comparisons were made
for the mortality data.
Table 11-6 displays the Cooke et al. (1984) age-standardized incidence
data which were reclassified according to anatomical site, socio-economic
status and outdoor versus indoor exposure. Trends in the data according to
socio-economic status were apparent for melanomas of each site (e.g., the head
and neck, trunk, and upper and lower limbs) among indoor workers. Among
outdoor workers, socio-economic trends were observed only for melanomas of the
trunk. This analysis was, however, limited by the small number of
registrations in some groups and the fact that 37 of 501 incident cases did
not have site information (these cases were spread across all age groups).
The authors concluded that the elevated melanoma incidence rates among
professional, technical, administrative, and managerial workers appeared to be
due to differences in socio-economic status. They observed that there was no
evidence of differences in risk between indoor and outdoor workers of similar
socio-economic classes. Similar results were observed for the mortality data
when analyzed according to age, socio-economic status, and outdoor exposure.
DRAFT FINAL
-------
11-8
TABLE 11-5
INCIDENT MELANOMA CASES AND MELANOMA DEATHS IN MAJOR GROUPS
OF OCCUPATIONS, NON-MAORI MEN AGED 25-64 YEARS,
NEW ZEALAND (1972-1976 INCIDENCE, 1973-1976 MORTALITY)
Number of Registrations Number of Deaths
a a
Major Occupational Group Observed Expected 0/E Observed Expected 0/E
Professional, Technical
Administrative, Managerial
Clerical
Sales
Service
Agricultural
Production, Transport
and Labouring
103
47
53
51
24
64
159
61
29
47
55
27
68
214
1.7*
1.6**
1.1
0.9
0.9
0.9
0.7*
20
13
14
19
6
18
52
17
9
14
15
8
19
60
1.2
1.5
1.0
1.2
0.8
0.9
0.9
All Specified Occupations 501 501 1.0 142 142 1.0
a
Observed/Expected.
* Significantly different from 1.0 (p<0.001; two-tailed tests).
** Significantly different from 1.0 (p<0.01).
Source: Cooke et al. (1984).
* " * DRAFT FINAL - * *
-------
11-9
TABLE 11-6
MELANOMA INCIDENCE RATES ACCORDING TO SITE,
SOCIO-ECONOMIC STATUS, AND OUTDOOR EXPOSURE:
MEN AGED 25-64 YEARS, NEW ZEALAND
(1972-1976 INCIDENCE DATA)
Age-Standardized Incidence Rate
(per 100,000 persons)
Site
Head, Neck
Trunk
Upper Limbs -
Lower Limbs
Exposure
Group
Indoor
Outdoor
Indoor'
Outdoor
Indoor
Outdoor
Indoor
Outdoor
Number of
Registrations
43
22
123
48
42
27
68
37
Adjusted for
Socio-Economic Status Socio-Economic
1,2 3
3.8 2.2
-» 2.2
11 7.7
7.3
4.9 2.7
3.3
6.4 6.2
6.8
4
2.3
3.3
4.1
4.8
1.3
5.4
2.6
4.4
5,6
1.7
2.0
5.7
3.9
-
2.1
5.1
Status
2.5
2.6
7.0
6.2
2.2
3.7
4.0
4.5
* Fewer than five registrations.
Source: Cooke et al. (1984).
- * DRAFT FINAL
-------
11-10
Cooke et al. (1984) concluded that their results did not support the
hypothesis that recreational sunbathing of inadequately tanned skin is
important in the etiology of melanoma. The similarity between outdoor and
indoor workers implied that solar exposure was unlikely to have been
important. The authors noted, however, that different patterns of
recreational sun exposure (e.g., sunny winter holidays or use of sun lotions)
may have sufficiently varied within social classes to have overcome the
tendency of greater exposure of outdoor workers.
Several case-control studies have also examined trends in melanoma with
respect to outdoor occupational exposure patterns. A Norwegian case-control
study by Klepp and Magnus (1979) found outdoor work (3-4 hours per day in
fresh air at work) to be more prevalent among their 35 male melanoma patients
(40 percent) than among their 92 other non-skin cancer male non-matched
controls from the same hospital (32 percent), but this difference was not
statistically significant. Results from another case-control S-tudy, conducted
by Mackie and Aitchison (1982) in West Scotland, are the reverse of the
findings by Klepp and Magnus in Norway (1979). Mackie and Aitchison (1982)
showed significantly lower levels of occupational sun exposure (p<0.05)
among 52 male melanoma patients compared with 52 age-matched male controls.
Twenty-three percent of the male cases had positive occupational exposure (16
or more hours outdoors each week) as compared with 48 percent of the controls.
In a much larger case-control study conducted in Western Canada (595 age-,
sex-, and residence-matched pairs), Elwood et al. (1985) assessed sun exposure
using a life-time occupational history with information on each job, industry,
and usual numbers of outdoor hours per week on the job during the summer and
winter seasons. Results of a multiple logistic regression analysis showed a
significantly increased relative risk of 1.6 for those with "mild"
occupational sun exposure during summer (approximately 1-8 hours/week)
compared to those with no occupational sun exposure. After adjustment for
host factors (hair color, skin color, history of freckles) and ethnic origin,
the relative risk increased slightly to 1.8 (95% C.I. 1.2-2.5) at the same
mild exposure level. No increased risk was seen, however, at higher
occupational sun exposure levels (8-16 hours/week, 16-32 hours/week, or 32+
hours/week).
The effect of clothing habits during outdoor work on the risk of melanoma
was examined by Holman et al. (1986) in a case-control study of CMM cases in
Western Australia (507 age-, sex-, and residence-matched pairs). This is one
of the few studies which investigated histogenic types of CMM and sunlight
exposure patterns. For all melanomas combined and all histogenic types except
SSM, the risks were higher if the primary melanoma site was sometimes exposed
rather than usually -exposed or usually covered while working outdoors. SSM,
in contrast, showed a significant increasing linear trend (p=0.008) for site
exposure as follows: OR=1.0 for "usually covered", OR=2.16 (95% C.I.
1.14-4.10) for "sometimes exposed" and OR=2.43 (95?0 C.I. 1.18-4.97) for
"usually exposed." This result, which contradicts the hypothesis that
intermittent exposure is important in the development of SSM, was not
specifically discussed by the authors.
DRAFT FINAL
-------
11-11
In a cohort study using college health records from 50,000 male alumni of
Harvard University and the University of Pennsylvania to identify predictive
risk factors for fatal skin, blood, and lymphatic cancers, Paffenbarger et al.
(1978) found an increased relative risk of melanoma (RR=3.9, p=0.01) for
outdoor work prior to college. This was based on 45 deaths from malignant
melanoma during the 35-year observation period (1.71 million person-years of
observation) compared with 180 surviving controls (4 controls per case were
chosen from classmates born in the same year and known to survive the
decedent). There was no information on outdoor work after university
admittance, but previous outdoor work was the only significant risk factor
identified for melanoma in this study.
A study of CMM incidence data for England and Wales by Beral and Robinson
(1981) revealed site-specific trends according to occupation. They examined
melanoma and basal- and squamous-cell cancer incidence data from 1970-1975 for
England and Wales obtained from the Office of Population Censuses and
Surveys. Information for each case included occupation and, for each melanoma
case, anatomical location. The anatomical location data were grouped into
either exposed (head, face, and neck) or unexposed site categories.
Occupational groups were classified as indoor office workers, other indoor
workers, or outdoor workers. Age-specific standardized cancer registration
ratios were calculated by indirect standardization and based on 1971 census
data.
Table 11-7 shows the age-standardized registration ratios for melanomas of
exposed and unexposed sites, and other skin cancers, by place of work for •
males aged 15-64 years. These results indicate that outdoor work was
associated with a 10 percent excess of basal- and squamous-cell carcinomas, a
9 percent nonsignificant excess of melanomas of the head, face and neck, and a
22 percent deficit of melanomas of unexposed sites. In contrast, office work
was associated with a 31 percent excess of melanomas of unexposed sites.
These differences persisted when the data were analyzed for social class III
(skilled workers) only as shown in Table 11-8. There was, however, one main
difference -- office work was also significantly associated with an excess of
squamous cell and basal cell carcinomas (p<0.05).
Beral and Robinson (1981) concluded that the similarity of melanomas of
"exposed" sites and squamous- and basal-cell carcinomas by occupational group
suggested that prolonged sun exposure may be important in the etiology of
melanomas of regularly exposed parts of the body. Furthermore, the low
incidence of melanomas of unexposed sites in outdoor workers indicated that
occupational exposure was not associated with increased melanoma incidence on
normally covered parts of the body. The authors noted that the reasons for
differences between office workers and other indoor workers (who had lower
melanoma and other skin cancer incidence rates) were not clear. They
observed, however, that whatever the cause (e.g., a greater tendency to expose
normally covered parts of the body to sunlight), it was unlikely that
prolonged sun exposure was an important etiological factor in office workers.
A recent study by Vagero et al. (1986) confirms some of the findings of
Beral and Robinson (1981). Vagero et al. (1986) examined incidence data on
DRAFT FINAL
-------
11-12
TABLE 11-7
STANDARDIZED REGISTRATION RATIOS FOR MALIGNANT MELANOMA
OF EXPOSED AND UNEXPOSED SITES AND OTHER SKIN CANCERS,
BY PLACE OF WORK, MALES AGED 15-64, ENGLAND AND WALES 1970-75
Outdoor Work Office Work
Other
Indoor
Work
Melanoma of Face, Head, 109
and Neck
All
Occupations
Squamous and Basal Cell 110- (1,194) 97 (1,221) 92* (813) 100 (3,228)
Carcinoma
(94) 102 (104) 87 (66) 100 (264)
Melanoma of Other Sites 78* (285) . 131* (573) 85* (281) 100 (1,139)
* The numbers marked with an asterisk differ significantly from 100 (p<0.05).
Source: Beral and Robinson (1981).
DRAFT FINAL
-------
11-13
TABLE 11-8
STANDARDIZED REGISTRATION RATIOS (AND NUMBER OF CASES)
FOR MALIGNANT MELANOMA OF HEAD, FACE, AND NECK AND OTHER
SITES AND OTHER SKIN CANCERS BY PLACE OF WORK, MEN AGED
15-64 YEARS, ENGLAND AND WALES, 1970-75
SOCIAL CLASS III ONLY
Outdoor Work Office Work
Other
Indoor
Work
All
Occupations
Squamous Cell and Basal 112* (487) 111* (391) 85* (568) 100 (1,446)
Cell Carcinoma
Melanoma of Head, Face 105 (38)
and Neck
Melanoma of Other Sites 71* (111)
106 (31) 81 (47) 100 (116)
143* (178) 75* (189) 100 (478)
* Significantly different from 100 (p<0.05).
Source: Beral and Robinson (1981).
DRAFT FIN'AL
-------
11-14
4,706 CMM cases and 4,244 basal- and squaraous-cell carcinoma cases for
1961-1979 from the Swedish Cancer Environment Registry. The data were
classified by occupation (office, other indoor, and outdoor workers), and
anatomical location for CMM cases (covered and uncovered parts of the body).
Standardized Morbidity Ratios (SMBRs) were calculated based on 1960 population
census data and adjusted for age, sex, residence, and social class.
A comparison of the calculated SMBRs for males and females combined is
shown in Table 11-9. The results were consistent with a slightly higher risk
of melanoma of the face and neck for outdoor workers. For office workers,
however, the observed number of melanomas were lower than expected for
normally uncovered parts of the body and were higher than expected for
normally covered parts of the body. Vagero et al. (1986) concluded that the
elevated risk of melanoma of covered parts of the body among office workers
was not entirely due to differences in social class. They estimated that
indoor office workers, as compared to other indoor workers, may have a 10
percent greater CMM incidence after taking into account differences in age,
residence, and social class distribution. The authors hypothesized that the
observed differences in CMM incidence did not merely reflect risk differences
between social classes such as those assumed to be caused by different
patterns of sun exposure. Such differences would not explain the contrasts
between office and other indoor workers within the same social class which
have been observed in this and other studies (Lee and Strickland 1980; Beral
and Robinson 1981). However, the authors could not rule out the possibility
that within each social class, patterns of sunlight exposure and sunburn
experience were different among office, other indoor, and outdoor workers.
MELANOMA IN WORKERS EXPOSED TO CHEMICALS OR RADIATION
There have been a number of studies in which an increased incidence of
cutaneous melanoma has been reported in cohorts occupationally exposed to
chemicals and/or radiation. Rushton and Alderson (1981) evaluated mortality
records for workers in eight oil refineries in Britain. To be included in the
study, workers had to have worked for at least one year between January 1,
1950 and December 31, 1975. The study population consisted of 34,701 white
males with 575,982 person years of observation and a mean follow up of 16.6
years. Comparison populations were males in England and Wales for the English
and Welsh refineries and in Scotland for the Scottish refineries. Data were
analyzed across all refineries or by individual refineries but there was very
little ancillary information on the differences between refineries with regard
to location, size of work force, type of product produced or length of service.
Two refineries showed a significant excess of observed melanoma deaths
versus expected deaths: p=0.0037 for refinery B and p=0.0003 for refinery H.
At refinery B, four out of five individuals who died from melanoma were
"operators", whereas at refinery H, the six deaths included an operator, two
boilermakers, a pipefitter, a laborer and a clerk. There was no definition of
the exposure pattern normally encountered by these various positions, nor was
it possible to determine what either refinery produced and thus to what these
individuals may have been exposed.
DRAFT FINAL
-------
11-15
TABLE 11-9
MORBIDITY RATIOS STANDARDIZED FOR AGE, GENDER,
COUNTY OF RESIDENCE AND SOCIAL CLASS
SWEDISH CANCER CASES, 1961-79
Type of Work
Number of Cases
Gender Observed Expected
Standardized
Morbidity
Ratio
95% Confidence
Limits
Malignant Melanoma of Uncovered Parts
Office
Indoor, Non-Office
Outdoor
m+f
m+f
m+f
142
352
186
156.0
347.5
170.0
91
101
109
77-107
91-112
94-126
Malignant Melanoma of Covered Parts
Office
Indoor, Non-Office
Outdoor
m+f
m+f
m+f
1,062
1,821
620
980.3
1,816.2
690.3
108
100
90
102-115
96-105
83-97
Squamous and Basal Cell Cancers
Office
Indoor, Non-Office
Outdoor
m+f
m+f
m+f
890
1,875
1,479
867.5
1,970.2
1,394.1
103
95
106
96-110
91-100
101-112
Source: Vagaro et al. (1986).
DRAFT FINAL
-------
11-16
Hoar and Pell (1981) in a retrospective cohort study of chemists working
for E.I. DuPont Company evaluated records from 3,713 white males and 75 white
females who were employed in 1959 as chemists as well as 19,262 white males
and 673 white females who were "non-chemists". The authors indicated that the
job title chemist at DuPont was difficult to define, that most chemists were
exposed to a number of chemicals that changed in quantity and quality and that
different chemists rarely had the same exposures. Thus it was not possible to
even estimate the kinds of chemicals to which "chemists" were exposed. The
title non-chemist was applied only to people who had never been chemists. In
addition, their exposures were also not stated. Death certificates were
obtained on 105 (93 percent) of the male and 6 (100 percent) of the female
deceased chemists and on 1,863 (92 percent) of the male and 15 (79 percent) of
the female deceased non-chemists.
When compared to the non-chemist cohort, melanoma incidence in the male
chemists was not significantly different. However, when compared to the Third
National Cancer Survey, the incidence of melanoma among male chemists was more
than expected (0/E=8/3.3; standardized incidence ratio of 239, 95% C.I.=111-454)
A subsequent comparison of the non-chemist cohort to the Third National Cancer
Survey revealed that this group too showed a higher than expected incidence
(0/E=38/17.1; standardized incidence ratio of 223, C.I. not provided).
In discussing the above finding, the authors indicated that a similar
observation was made for all DuPont employees in the period 1956-1974 and that
one possible explanation may have been occupational exposure to chemicals
suspected of being skin carcinogens. Another possible explanation suggested
by the authors was exposure to solar radiation. The majority of DuPont plants
are located in the Southeastern United States where solar exposure is
greatest. Sixty-five percent of the DuPont salaried employees resided in the
15 southern states which account for 37 percent of U.S. melanoma mortality.
Holmbert et al. (1983) studied a cohort of 13,114 persons who had worked
at two plants in the Swedish rubber industry for at least 12 months between
January 1, 1951 and December 31, 1975. Workers were placed into one of three
exposure categories. Category 1, work in the weighing and mixing department
consisted of 739 individuals. Category 2, other production work (e.g.,
calendaring, vulcanization, pressing, tire building, inspection, service work,
floor cleaning, storage work), consisted of 9,883 individuals, and Category 3
white collar work (office personnel, department heads) consisted of 2,492
individuals. An increased occurrence of malignant melanomas was found in
exposure class 2 resulting in a risk ratio for this group of 2.50.
There is one brief report in the literature linking exposure to a specific
class of chemicals with an increased incidence of melanoma. NIOSH (1976) and
Bahn et al. (1976) reported increased mortality from melanoma in a cohort of
workers exposed to Aroclor 1254 (Monsanto's Tradename for PCBs) during a nine
year period in the late 1950's at a petrochemical plant in the northeast
United States. The study evaluated information from two small cohorts; one of
51 workers at a research and development facility exposed from 1949 to 1957,
and one of 41 workers in a refinery exposed from 1953 to 1958. Two melanomas
occurred in the first cohort (versus 0.04 expected, p<0.001). In the second
* * » DRAFT FINAL * » *
-------
11-17
cohort of 51 workers, excess melanomas were observed, however, no detailed
information was provided on exposure or method of analysis. There was no
attempt to quantify or even describe exposures in this workplace and it is
possible that these workers were exposed to chemicals other than PCBs.
In the period of 1972-1977, 19 cases of melanoma were reported among
approximately 5,100 employees of the Lawrence Livermore National Laboratory
(LLNL) (Austin et al. 1981). This was approximately three times the rate
expected in a comparable age/race/sex-adjusted geographical segment of the
population from the San Francisco Bay Area. To investigate this finding, each
case was matched with four controls drawn from the laboratory population and
analyzed to examine the relationship of risk to occupational variables such as
length of employment, cumulative radiation exposure and job classification
("scientist" versus "non-scientist"). Cases and controls were matched on the
basis of 5-year age group, race, sex and census tract. No relationship was
evident between melanoma and any of these parameters; however, "chemists" had
a relative risk of 6.97 (p=0.011).
In an attempt to control for socio-economic differences between the study
group and reference groups, cases and controls were matched by census tract of
residence in the incidence analysis. It was not possible to evaluate the
efficacy of this technique, but since the difference in melanoma incidence
between the highest and lowest quartiles of SES for all census tracts in the
San Francisco SMSA was only twofold, it seems unlikely that SES differences
accounted for the finding at LLNL. Beyond the conclusion that this.was a real
increase in the incidence of melanoma that could not be acc'ounted for by other
factors such as socioeconomic status, the authors were unable to identify a
work related factor (other than job title) which showed an association with
melanoma incidence in this population.
As a result of the LLNL report, Acquavella et al. (1983) conducted a
case-control study of melanoma at the Los Alamos National Laboratory (LANL).
Twenty cases were identified and, for each case, four controls were selected
and matched on the basis of sex, ethnicity, date of birth and date of first
employment. Controls were selected from employees hired immediately before
and after each case. Most controls were selected from a pool of 100 employees
but occasionally this was expanded to as large as 500 in order to obtain
adequate matches. The data obtained for cases and controls included a number
of occupational variables such as length of employment, cumulative external
radiation exposure two years prior to case's occurrence of melanoma, job
title, and educational status. The authors concluded that there was no
indication of an association between melanoma occurrence and any particular
form of radiation. With regard to educational attainment, however,
individuals with a college or graduate level degree had elevated risks of
developing melanoma. College graduates had a standardized rate ratio (SRR) of
2.11 and those with a graduate degree had an SRR of 3.17.
MELANOMA AND EXPOSURE TO FLUORESCENT LIGHTING
Several studies have examined the potential link between exposure to
fluorescent lighting and CMM. Interest in this area developed in response to
DRAFT FINAL
-------
11-18
information indicating apparent differences in melanoma incidence between
outdoor.workers and indoor office workers who are regularly exposed to
fluorescent light. Emissions from fluorescent lights often extend into the
ultraviolet range, although the emitted wavelength distribution varies with
the type of lamp, glass envelope, and other covers.
Beral et al. (1982) analyzed data on 274 female cases of CMM 18-54 years
of age and 549 age- and residence-matched female controls from a study
originally designed to investigate the association between melanoma and oral
contraceptive use in New South Wales, Australia. They found exposure to
fluorescent light at work to be associated with a 2.1 relative risk of
melanoma (95% C.I. 1.32-3.32) as compared to no exposure at work. This risk
increased with increasing duration of exposure (p<0.001 for trend) and was
higher for women who had worked mainly in offices (RR=2.6) than for those who
had worked mainly indoors but not in offices (RR=1.8). The increase in risk
associated with fluorescent light exposure at work was further examined to
determine whether other factors might be indirectly affecting the risk.
Neither long-term nor intense short-term recreational exposure to sunlight
showed a consistent relationship to increased melanoma risk, nor did
stratification by the following factors diminish the overall association:
amount of time spent outdoors, main outdoor activity and amount of clothing
worn in childhood and at ages 20 and 30, sunburn history on various parts of
the body, place of birth, hair color, skin color, use of oral contraceptives,
and frequency of naevi on the body. Some of these factors, however, seemed to
modify the risk of melanoma associated with exposure-to fluorescent light
slightly, e.g., the relative risks tended to be lower for women who had been
most heavily sun-exposed as estimated by amount of time spent outdoors in
childhood and main outdoor activity at age 20 and higher for women who
reported having more than an average number of naevi. In contrast with these
results, there was no increase in melanoma risk for fluorescent lights in the
home (RR=0.9, 95% C.I. 0.6-1.6) even when analysis was restricted to women who
had never been exposed to fluorescent lights at work and who had never worked
outdoors.
A small series of 27 male melanoma cases 18-56 years of age and 35 male
controls of similar ages was available from the same melanoma clinic (Beral et
al. 1982) and showed a similar significant increase in melanoma risk with
exposure to fluorescent light among those who always worked indoors. The
relative risk of CMM for males with 10 or more years of fluorescent light
exposure (RR=4.4, 95?0 C.I. 1.1-17.5) was higher than the relative risks for
women with more than 10 years of exposures, although confidence limits
overlapped (10-19 years, RR=2.5, 95% C.I. 1.5-4.2; 20 or more years, RR=2.6,
95% C.I. 1.2-5.9). The males also showed slightly higher melanoma risk for
ever having worked outdoors compared to those who had always worked indoors
(RR=2.2, 95% C.I. 0.6-8.0). Although results are based on small numbers of
male CMM cases and controls, the findings are in agreement with those from the
larger study of females.
Dubin et al. (1986) also examined the association of CMM to fluorescent
light exposure in an interview study of 1,103 CMM cases and 585 controls
randomly chosen among new patients, ages 20 and older at the New York
DRAFT FINAL
-------
11-19
University Skin and Cancer Unit general skin clinic. A preliminary analysis
of a subset of these data (Pasternack et al. 1983) yielded a significant
positive association between CMM and fluorescent light exposure. A
reliability study was later conducted to confirm the interview assessment of
fluorescent light exposure by means of a mailed questionnaire. The
reliability study data did not support the interview data that formed the
basis of their preliminary report (Pasternack et al. 1983) of a positive
association of CMM to fluorescent light exposure. Dubin et al. (1986) believe
that interview bias may have affected the fluorescent light data, leading to
overestimates of exposure only among cases but not controls. For this reason,
no conclusions regarding the association of CMM to fluorescent light exposure
can be drawn from this study.
Elwood et al. (1986) conducted a matched case-control study of 83 CMM
patients and 83 age-, sex-, and residence-matched controls which evaluated
exposure to both diffused and undiffused fluorescent lighting. No significant
trends in relative risk were associated with level of exposure to fluorescent
lighting through occupational or home exposure. The relative risk for
individuals in the highest total occupational fluorescent light exposure
category (50,000 •+• hours) compared to those with no occupational exposure was
1.4 (95% C.I.=0.4-5.1). Corresponding relative risks for those exposed to the
highest categories of diffused (25,001-50,000 hours) and undiffused (50,000+)
flourescent lighting were 1.5 (95% C.I.=0.5-4.4 and 4.0 (95% C.I.=0.8-19.2),
respectively. Associations of CMM with fluorescent lighting based on a
subsequent postal questionnaire were weaker than those based on the personal
interviews cited above. This also occurred in the study of Dubin et al.
(1986), and may reflect either recall bias in personal interviews or that
mailed questionnaires are less reliable than personal interviews (Elwood et
al. 1986).
Rigel et al. (1983) found no increased risk of melanoma associated with
fluorescent-light exposure in a preliminary analysis of 114 melanoma patients
and 228 matched (5-year age groups) controls from the New York University
Medical Center. There was no significant difference in the proportion of
indoor office workers between the cases (57 percent) and the controls (60
percent), nor was there a difference in average daily exposure to fluorescent
lights (4.9 hours for cases and 5.4 hours for controls). For indoor office
work, the average daily amounts of fluorescent light exposure was 5.93 hours
for cases vs. 5.99 hours for controls. The authors found increased risks for
several risk factors, e.g., recreation activities (RR=2.4, p=0.01 for outdoor
vs. indoor) and sun exposure 2 hours/day (11-20 years ago RR=2.5, p=0.0005 and
6-10 years ago RR=1.6, p=0.05), and postulate that the increased melanoma risk
for indoor workers may be explained by their recreational habits and not by
fluorescent-light exposure.
DRAFT FINAL
-------
11-20
FINDINGS
A number of findings are noteworthy regarding patterns of CMM with respect
to socio-economic status and other occupational factors:
11.1 CMM incidence and mortality are positively related to
increasing socio-economic status. Furthermore, total
CMM incidence has been observed to higher among
"professional" and "administrative" indoor office
workers, but not other indoor workers, compared to
outdoor workers. Evidence indicating that outdoor
workers do not have an elevated risk of melanoma
•compared to office workers may be confounded by
differences in socio-economic status, host factors,
ethnic background, melanoma site, and histologic type.
11.2 For usually uncovered parts of the body (e.g., the
face), the incidence and risk of CMM is higher among
outdoor workers than among indoor office workers. For
usually covered parts of the body, the incidence of
CMM among indoor office workers is higher than for
outdoor workers.
11.3 The incidence and risk of CMM among indoor office
workers is higher for sites, that are usually covered
(e.g., the trunk) than for sites that are usually
exposed (e.g., the face). Among outdoor workers, CMM
risks are higher for usually exposed sites than for
usually covered sites.
11.4 A number of studies investigating the melanoma risk of
workers in refineries, or chemical or pharmaceutical
plants have failed to find a significant association
between melanoma and potential exposure to chemicals
although in at least one study an increased risk of
melanoma was found for male DuPont workers (both
chemists and non-chemists) when compared to males from
the Third National Cancer Survey. In one study from
Lawrence Livermore National Laboratories, an increased
risk of melanoma was observed in individuals with the
title "chemist"; however, no other work-related factor
demonstrated an association.
11.5 It has been suggested that the risk of developing CMM
may be elevated among individuals exposed to
fluorescent lighting at work. However, several
studies have failed to find a significant and
consistent association between CMM and exposure to
fluorescent lighting at work. In addition, although
two of these studies initially found an association
based on personal interview data, an attempt by one
study to validate its findings using a self-
administered postal survey was not successful.
••- DRAFT FINAL * * *
-------
11-21
REFERENCES
Acquavella, J.F., Wilkinson, G.S., Tietjen, G.L., Key, C.R., Stebbings, J.H.
and Voelz, G.L. A melanoma case-control study at the Los Alamos National
Laboratory. Hlth Phys 45:587-592 (1983).
Austin, D.F., Reynolds, P.J., Snyder, M.A., Biggs, M.W. and Stubbs, H.A.
Malignant melanoma among employees of Lawrence Livermore National Laboratory.
Lancet 2:712-716 (1981).
Bahn, A.K., Rosewaike, I., Hermann, N., Grover, P., Stellman, J. and O'Leary,
K. Letter to the editor: Melanoma after exposures to PCB's. New Eng 'J Med p
450 (August, 19 1976).
Beral, V., Evans, S. Shaw, H. and Milton G. Malignant melanoma and exposure
to fluorescent lighting at work. The Lancet 290-293 (August 7, 1982).
Beral, V. Ramcharan, S. and Faris, R. Malignant melanoma and oral
Contraceptive use among women in California. Brit J Cancer 36:804-809 (1977).
Beral, V. and Robinson, N. The relationship of malignant melanoma, basal and
squamous skin cancers to indoor and outdoor work. Br J Cancer 44:886-891
(1981).
Cooke, K.R., Skegg, D.C.G. and Fraser, J. Socio-economic status, indoor and
outdoor work and malignant melanoma. Intl J Cancer 34:57-62 (1984).
Dubin, N., Mosemon, M. and Pastefnack, B.S. Epidemiology of malignant
melanoma: Pigmentary traits, ultraviolet radiation and the identification of
high-risk populations. Rec Results Can Res 102:56-76 (1986).
Elwood, J.M., Gallagher, R.P., Hill, G.B. and Pearson, J.C.G. Cutaneous
melanoma in relation to intermittent and constant sun exposure - The Western
Canada melanoma study. Int J Cancer 35:427-433 (1985).
Elwood, J.M., Williamson, C. and Stapleton, P.J. 1986. Malignant melanoma in
relation to moles, pigmentation, and exposure to fluorescent and other
lighting sources. Brit J Cancer 53:65-74.
Hoar, S.K. and Pell, S. A retrospective cohort study of mortality and cancer
incidence among chemists. J Occup Med 23:483-494 (1981).
Holman, C.D.J., Armstrong, B.K. and Heenan, P.J. Relationship of cutaneous
malignant melanoma to individual sunlight-exposure habits. JNCI 76:403-414
(1986).
Holman, C.D.J., Mulroney, C.D. and Armstrong, B.K. Epidemiology of
Pre-invasive and Invasive Malignant Melanoma in Western Austria. Brit J
Cancer 25:317-323 (1980).
DRAFT FINAL
-A- •>? -.'r
-------
11-22
Holmbert, B., Westerholm, P., Massing, R., Kestrup, L. , Gumaelius, H.,
Holmlund, L.and England, A. Retrospective cohort study of two plants in the
Swedish rubber industry. Scand J Work Environ Hlth 9(Suppl 2):59-68 (1983).
Klepp, 0. and Magnus, K. Some environmental and bodily characteristics of
melanoma patients. A case control study. Intl J Cancer 23:482-486 (1979).
Lee, J.A.H. and Strickland, D. Malignant melanoma: Social status and outdoor
work. Brit J Cancer 41:757-763 (1980).
Lee, J.A.H. Melanoma and exposure to sunlight. Epi Rev 4:110-136 (1982).
MacKie, R.M. and Aitchinson, T. Severe sunburn and subsequent risk of primary
cutaneous malignant melanoma in Scotland. Brit J Cancer 46:955-960 01982).
NIOSH Unpublished data: melanoma after exposure to PCBs (1976).
Paffenbarger, R.S., Wing, A.L. and Hyde, R.T. Characteristics in youth
predictive of adult-onset malignant lymphomas, melanomas, and leukemias:
Brief communication. J Natl Cancer Inst 60:89-92 (1978).
Pasternack, R.S., Daba, N. and Moseson, M. Malignant melanoma and exposure to
fluorescent lighting at work. Lancet:704 (March 26, 1983).
Rigel, D.S., Friedman, R.J., Bernstein, M. and Greenwald, D.J. Letter.to the
editor. Malignant melanoma and exposure to fluorescent lighting at work. The
Lancet:704 (March 26, 1983).
Rushton, L. and Alderson, M.R. An epidemiological survey of eight oil
refineries in Britain. Brit J Indust Med 38:225-234 (1981).
Teppo, L., Pukkala, E., Kahama, M., et al. Way of life and cancer incidence
in Finland. Scand J Soc Med (Suppl) 19:50-54 (1980). As cited in Lee (1982).
Vagero, D., Ringback, G. and Kiviranta, H. Melanoma and other tumours of the
skin among office, other indoor and outdoor workers in Sweden 1961-1979. Brit
J Cancer 53:507-512 (1986).
DRAFT FINAL
-------
CHAPTER 12
OTHER FACTORS
STEROID HORMONES AND MALIGNANT MELANOMA
Evidence for a relationship between the biological behavior of melanoma
and steroid hormone action has been observed in several areas of research.
These observations include differing survival prognosis favoring females over
males, the rarity of the tumor in prepubescent children, the improved survival
for postmenopausal and multiparous women, and the increased melanoma incidence
among 30 to 50-year-old women. Effects of pregnancy and exogenous hormones on
growth and development of melanomas have also been shown. Estrogen receptors
have also been observed in human melanomas (McCarty et al. 1980).
Effects of Endogenous Hormones
Several studies indicate that sex hormones may influence melanocyte
activity and the natural history of malignant melanoma. Based on the observed
rapid increase in mole counts in both sexes during puberty, Mackie et al.
(1985) suggested the presence of an hormonal influence on the
pigment-producing activity of nevi. The authors suggested that this could
result from either a new appearance and proliferation of pigment-producing
nevi or the activation of the melanin-producing enzyme pathway of.
pre-existing, inactive, non-pigment-producing nevi.
Hodgins (1983) noted that the pigmentation changes of genital and areolar
skin at puberty and in pregnancy suggest that gonadal hormones influence at
least some populations of melanocytes. Greene et al. (1985) stated that they
counsel high-risk family members (those with dysplastic nevi) to pay
particular attention to nevi during periods of hormonal flux (i.e., puberty
and pregnancy).
Several epidemiological studies have indicated that the observed sex
differences in melanoma incidence and mortality could be related to hormonal
differences. Hodgins (1983) concluded, however, that there is no clear
evidence linking these differences to levels of steroid hormones. For
example, the prognosis for men with malignant melanoma is worse than for
women. Hodgins cited Shaw et al. (1978) who concluded that better survival
resulted at least in part from the earlier stage at presentation and
prognostically more favorable sites among women. However, when male patients
were matched to female patients by age, and size and thickness of lesion, the
female survival advantage among premenopausal stage I patients compared to
matched male patients persisted. The survival advantage was much less for
postmenopausal women compared to matched male patients. The results, Hodgins
noted, supported the concept of a barrier to tumor metastasis in premenopausal
women.
The observation of higher melanoma incidence and mortality rates among
reproductive and menopausal aged women than among men of the same age in the
* * * DRAFT FINAL * * *
-------
12-2
British Isles led Lee and Storer (1980) to suggest that a hormone-dependent
variant of melanoma may account for the difference. The authors conducted a
descriptive comparison of WHO mortality data from eight European countries and
Office of Population Censuses and Surveys incidence data from England and
Wales. Age-specific female-to-male mortality ratios indicated higher female
mortality rates relative to male rates in the British Isles from 1955-1974,
whereas the reverse was observed for Australia, North America (the U.S. and
Canada), Japan, and Scandinavia (Denmark, Norway, Sweden, and Finland). Lee
and Storer (1980) did not offer an explanation for the higher British Isles
mortality ratios. They noted, however, that a similar pattern of age-specific
sex ratios (risks for females compared with males peaked in the latter half of
reproductive life, and decreased or leveled off in middle age) was observed in
these different populations.
Excess female mortality and incidence rates relative to those in males in
the British Isles was greatest from ages 30-44. The elevated sex ratio from
ages 30-49 did not change from 1950 to 1974. Lee and Storer (1980) suggested
that the low rates for malignant melanoma in the British Isles, compared to
those for Australia and New Zealand, permitted the observation of a
hormone-dependent variant of melanoma in the British Isles.
Lee and Storer (1982) analyzed age-specific changes in the sex ratio of
malignant melanoma for several countries of Europe, North America, Australia,
and Japan using WHO mortality data. The female/male mortality ratio was less
than 1.0 for all countries examined in contrast to the female excess in the
British Isles (Lee and Storer 1980). In each of these populations, however,
female/male sex ratios peaked during the reproductive years and declined in
middle age. The authors tested whether an interaction of sex and age on CMM
mortality rates could have been produced by birth cohort effects and sex
differences in incidence and mortality. An examination of the data by 5-year
birth cohort intervals indicated a persistent increased female risk in the
reproductive years. Using a mathematical model to separate age and cohort
effects, Lee and Storer observed that interaction terms for age and sex and
year of birth and sex were both significant. Lee and Storer concluded that
the specific variation in female-to-male ratios may reflect the same biologic
progress that underlies changes in melanoma survival in relationship to
childbearing (e.g., decreased survival among pregnant women).
Holman et al. (1984) conducted a case-control study on 276 female melanoma
patients identified in the West Australia Lions Melanoma Research Project from
1980 to 1981. Two hundred and seventy-six age- and electoral-subdivision-
matched controls were selected from the Australian Commonwealth Electoral Roll
and a few from public school student rolls. The authors observed no
consistent evidence of a relationship of incidence rates of different
histogenic types with age at menarche, duration of menstural life, or number
of pregnancies of over 20 weeks duration.
Effects of Pregnancy
In a brief communication in Lancet, an anonymous writer (Anon 1971) stated
that while the relationship of pregnancy and melanoma used to be a matter of
* * * DRAFT FINAL * * *
-------
12-3
debate, pregnant women run the same risk of developing melanoma as
non-pregnant women, and tumor behavior is similar in the two groups.
Scattered evidence has indicated, however, that pregnancy can activate
metastatic disease in a previously treated melanoma, or increase its growth
rate (Lee and Storer 1982; McCarty et al. 1980).
In a study by Foucar et al. (1985),. 86 pregnant white patients visiting
the obstetrics clinic of the University of Iowa Hospitals and Clinics between
June and August 1982 permitted the removal of 128 nevi for study. Fifty-one
non-pregnant female controls and fifty male controls were obtained from the
Department of Pathology files of the University of Iowa Dermatology Clinic.
Controls were excluded if they were not between the ages of 16-39. In
addition, controls with evidence of atypical nevi were not included. .Case and
control nevi were compared using a graduated scoring system for atypia ranging
from 1 to 16. The 128 nevi from pregnant patients did not include any lesions
with sufficient atypia to suggest malignancy. The histopathologic features
among cases' nevi were identical to those seen in the male controls' nevi.
However, based on small but noticeable differences among histopathologic
"activation" measures between cases' nevi and nevi from female controls,
Foucar et al. (1985) suggested that mild changes in some nevocellular nevi may
occur during pregnancy. The authors noted, however, that a potential bias in
selecting controls may have resulted from elimination of controls with
clinical features potentially associated with histopathologic atypia.
Most epidemiologic investigations on pregnancy and melanoma have focused
on differences in survival of melanoma patients by pregnancy status at or near
the time of diagnosis. In an analysis of survival data for female melanoma
patients of childbearing age, White et al. (1961) observed that pregnancy did
not have an adverse effect on survival even after stratifying by age and"stage
of disease. The study population consisted of 18 women seen at Stanford
Hospital and 53 women from the California Tumor Registry aged 15 to 39 years
at CMM diagnosis. The patients were divided into a pregnant group (N=30)
(those pregnant within one year before and five years after diagnosis), a
nonpregnant group (N=31), and a pregnancy-undetermined group (N=10). Five-
year survival rates were examined for the three groups and indicated a higher
survival rate for pregnant (73 percent) than nonpregnant patients (55 percent)
although the difference was not significant due to the small sample size. To
test the robustness of the observed differences between the pregnant and
nonpregnant groups, a range of extreme assumptions concerning the makeup of
the pregnancy-undetermined group was applied. The results indicated that
within the limits of the data, 5-year survival rates among pregnant women were
similar to, or greater than, rates among nonpregnant women. Within three age
groups (under 20, 20-29, 30-39), survival rates were still equal to or more
favorable than rates for nonpregnant women. When analyzed by stage of
disease, White et al. (1961) observed that among those with localized disease
(25 cases pregnant, 21 nonpregnant), 5-year survival rates were slightly
higher among pregnant patients (88 percent) than nonpregnant patients (81
percent). It should be emphasized that the number of women in this study was
small and may not have been adequate for the detection of differences by
pregnancy status. The authors concluded that, based on their data, pregnancy
did not appear to have a deleterious effect on survival.
* * * DRAFT FINAL
-------
12-4
Shiu et al. (1976) found lower 5-year survival rates in pregnant women who
had experienced activation of a lesion during a previous pregnancy based on a
survival analysis of 251 female cases ages 15-45 who received treatment at the
Memorial Sloan-Kettering Cancer Center from 1950 to 1969. Cases were selected
if accurate recorded data on pregnancy at the time of admission were
available. The authors observed no significant difference in 5-year survival
rates for Stage I patients (n=165) between nulliparous, parous nonpregnant,
and pregnant women. Among 86 Stage II patients, however, significantly lower
survival rates (p<0.05) were observed for pregnant patients (29 percent) and
parous patients who had lesion activation in a previous pregnancy (22 percent)
as compared with nulliparous patients (55 percent) and other parous patients
(51 percent) combined. Age differences did not account for the differences in
5-year survival rates. Shiu et al. (1976) concluded that the differences in
survival rates and frequency of symptoms in Stage II patients (e.g., bleeding,
ulceration, and elevation of lesion) "strongly suggest an adverse influence of
pregnancy on women with stage II melanoma."
Hodgins (1983) has noted, however, that suggestions of adverse effects of
pregnancy upon malignant melanoma have not been supported by more recent
epidemiological studies. Elwood and Goldman (1978) observed that 5-year
survival rates did not differ among 254 ever-pregnant and 51 never-pregnant
melanoma patients. The study population was comprised of 305 consecutive
melanoma patients seen in Vancouver and diagnosed between 1960 and 1976. The
authors noted that their results differed from those of Hersey et al. (1977)
who studied 443 consecutive female patients seen in Sydney from 1961-1971.
Substantially better survival rates for ever-pregnant women were reported; the
largest difference in survival was for women over fifty with survival rates of
73 percent (ever-pregnant) and 53 percent (never-pregnant). After restricting
their series to cutaneous lesions (89 percent of the total) and adjusting for
stage at diagnosis, Elwood and Goldman (1978) still did not observe any
association between survival and pregnancy history. Elwood and Goldman
concluded that the inconsistency between their results and those of Hersey et
al. argued against the hypothesis of improved survival among ever-pregnant
melanoma patients.
Houghton et al. (1981) compared 3- and 5-year survival rates among female
melanoma patients aged 15 to 40 years of age using data obtained from the
Connecticut Tumor Registry from 1950-1954, 1960-1964, and 1970-1974. The
study included 12 patients diagnosed during pregnancy (cases) and 175 patients
not pregnant at the time of diagnosis. Each case was matched with two
nonpregnant patients according to age, anatomic site, and stage of disease at
diagnosis. No differences in survival between the two case groups were
noted. The three-year survival rate was 65 percent among pregnant patients
versus 67 percent among matched nonpregnant patients; the five-year survival
rate was 55 percent among pregnant versus 58 percent among matched nonpregnant
cases. The expected number of pregnant women among the 187 patients reviewed,
estimated from Connecticut livebirth rates, was 13.3, compared to the observed
12 pregnancies, suggesting that melanoma incidence did not substantially
increase during pregnancy. Houghton et al. (1981) noted that survival rates
were significantly lower among pregnant cases when age, anatomic site of
primary lesion, and stage at diagnosis were not considered.
* * * DRAFT FINAL * * *
-------
12-5
Effects of Exogenous Hormones
Several studies have examined the possible relationship between oral
contraceptive (OC) and other hormone use and melanoma. Beral et al. (1984)
concluded based on available epidemiologic evidence that "while oral
contraceptives and other exogenous sex hormones are clearly not major
determinants of melanoma, the accumulating evidence suggests that they may
increase the risk of disease."
Jelinek (1970) observed that pigmentary changes (melasma) caused by oral
contraceptives suggest that estrogen or progesterone could control skin
melanin. According to Jelinek, estrogens stimulate melanocytes and
progesterone causes the pigmentation to spread, indicating that the total
amount of both hormones could increase melasma incidence. Jelinek (1970)
stated, however, that the putative relationship between malignant melanoma and
OC use could be adequately explained by chance. Mackie et al. (1985) observed
that neither pregnancy nor OC use stimulate development of new moles although
some pre-existing moles have been found to temporarily darken during
pregnancy. Hodgins (1983) noted that in spite of the observed pigmentary
changes associated with OC use (attributable mainly to estrogens), there is
little evidence to suggest any link between CMM and the contraceptive pill.
Lederman et al. (1985), in a prospective study of 289 Caucasian female
Stage I melanoma patients, conducted a multivariate analysis to examine the
effects of prior estrogen and progesterone use on tumor characteristics and
survival. The cases were consecutively evaluated and prospectively entered
into a natural history study at the Massachusetts General Hospital and New
York University. Hormone users presented with thinner tumors than nonusers;
76 percent of OC users and 64 percent of menopausal estrogen (MPE) users had
primary tumors less than 1.69 mm, as compared with 58 percent of nonusers.
Users of OC's in the year prior to CMM diagnosis had significantly thinner
tumors than nonusers (p<0.01) in the year before diagnosis. Univariate
analysis showed that exogenous hormone use was not associated with increased
risk of death from CMM. Life table analysis revealed slightly greater five-
and nine-year survival rates in hormone users. Nine-year survival rates were
90 percent for OC users, 87 percent for MPE users, and 81 percent for
nonusers. The finding of thinner tumors in hormone users may have explained
their apparently more favorable survival. The fact that estrogen users had
thinner tumors may have been due to a direct effect of estrogens, the tendency
of hormone users to seek medical attention sooner than nonusers, or that
hormone users tend to be under closer medical surveillance than nonusers.
Lee and Storer (1982) noted that in the British Isles population, the
elevated female-to-male melanoma incidence ratio did not change over
successive five-year periods during a 25-year span from 1951 to 1975. These
data suggested that the large scale introduction of OCs did not produce the
elevated ratio.
Using the same methodological approach as Lee and Storer (1982), Stevens
and Lee (1980) examined incidence data from Connecticut (1935-1974), Denmark
(1943-1972) and Finland (1953-1974), and mortality data from the United States
(1951-1975), Canada (1951-1975) and England and Wales (1951-1975). The
* * * DRAFT FINAL * * *
-------
12-6
authors observed that there was no perturbation in female melanoma incidence
rates relative to male incidence rates at or after the introduction of oral
contraceptives. On the basis of these descriptive data, Stevens and Lee
(1980) argued that there is no association between OC use and malignant
melanoma since if there were an association, male incidence and mortality
rates would not be a constant multiple of female rates over an extended
period of study.
Beral et al. (1984) conducted a case-control analysis of the effect of OC
and hormone use, and pregnancy history on risk of melanoma. The cases
consisted of 287 white women aged 15-84 years who were seen at the Sydney
Hospital. General population controls were matched by age and area of
residence to cases diagnosed between 1974-1978 ("old" cases). Controls
selected from hospital inpatients were matched to "new" cases (those
interviewed between 1978 and 1980) by age only. Case-control comparisons
indicated that women with melanoma were more likely to have taken oral
contraceptives for long periods of time. A consistently increased risk of
melanoma was only observed in those who had begun OC use at least-10 years
before diagnosis and whose use continued for five or more years relative to
non-users (relative risk (RR)=1.5; 95% CI 1.03-2.14). This elevated risk
persisted after controlling for reported hair and skin color, frequency of
moles on body, place of birth and measures of sunlight and fluorescent light
exposure. Socio-economic status, which has been associated with CMM (see
Chapter 11), was not controlled for in this study. If there were an
association between socio-economic status and OC use, the failure to control
for socio-economic status may confound the observed results. Beral et al.
(1984) also observed that cases were more likely than controls (but not
significantly) to have used hormones to regulate periods (RR=1.9), hormonal
replacement therapy (RR=1.4), and been given hormone injections to suppress
lactation (RR=1.4).
Based on conclusions from several studies, Beral et al. (1984) stated that
while most studies reported weak or no associations of CMM to ever-use of
OC's, the five studies examining data on prolonged OC use found increased
risks (not always statistically significant) associated with long-term pill
use. As shown in Table 12-1, relative risk estimates for long duration of OC
use are in the range of 1.4 to 4.4. These relative risks are of the same
order of magnitude as those for recognized pigmentary risk factors such as red
or blonde hair and fair skin. It should be kept in mind, however, that some
of these results may be confounded by socio-economic status.
Endocrine Therapy
In breast cancer, the presence of specific estrogen receptors in a tumor
has provided an indication of responsiveness to endocrine therapy. Several
studies have searched for hormone receptors in malignant melanomas in the hope
of identifying patients who might respond to hormone treatment. Studies have
shown that steroid receptors are a necessary, if not sufficient, requirement
for steroid-hormone responsiveness in target tissues (Fisher et al. 1976).
Occurrence of Hormone Receptors. Based on a review of 14 published
studies, Hodgins (1983) concluded that melanomas generally contain low
* * * DRAFT FINAL * * *
-------
12-7
TABLE 12-1
SUMMARY OF FINDINGS FROM DIFFERENT STUDIES ON
MELANOMA AND ORAL CONTRACEPTIVE USE
Relative Risk
Ever Use of Oral
Contraceptives
Versus Never Use
Long Term Use of
Oral Contraceptives
Versus Shorter Term
Use or Never Use-*
Case Control Studies
Beral et al.
Adam et al.
Adam et al.
Bain et al.
Holly et al.
Beral et al.
(1977)
(1981a)
(1981b)
(1982)
(1983)
(1984)
1.9
1.1
1.34
0.93
1.15
1.0
No data
1.6
1.4
3.0*
4.4*
1.5*
Cohort Studies
Beral et al. (1977)
Adam et al. (1981)
Kay (1981)
Ramcharan et al. (1981)
1.4
0.3
1.5
3.5*
1.7
No data
No data
No data
* Differs significantly from 1.0 (P<0.05).
** Definitions for long term use were as follows:
Beral et al. (1977): total duration of use of 4+ years.
Adam et al. (1981): total duration of use of 5+ years;
a = data from postal survey;
b = data from GP records.
Bain et al. (1982): total duration of use of 2+ years
beginning 10+ years before diagnosis.
Holly et al. (1983)
total duration of use of 5+ years
beginning 12+ years before diagnosis
(superficial spreading melanoma only)
Beral et al. (1984): total duration of use of 5+ years,
beginning 10+ years before diagnosis.
Source: Beral et al. (1984).
* * * DRAFT FINAL * * *
-------
12-8
estrogen receptor concentrations, below those considered significant for
predicting hormone responsiveness for breast cancer. Ellis et al. (1985)
noted, however, that several studies describing hormonal receptors in human
malignant melanoma have found that 0-78 percent of melanomas have detectable
levels of estrogen receptors and from 0-100 percent of melanomas have
detectable levels of progesterone receptors.
McCarty et al. (1980) examined tumors from 20 patients aged 23 to 80 years
hospitalized at the Duke Comprehensive Cancer Center. No restriction was made
concerning age, sex, race, or menstrual status of the patients included in the
study. Seven of the 20 tumors showed high affinity estrogen binding of more
than 3 fmol/mg cytosol tissue protein by dextran-coated charcoal analysis
(DCCA). No relationship was observed between estrogen binding and age, sex,
menstrual status, or parity. No evidence of high affinity progesterone
binding was observed in any of the 20 tumors. The authors cautioned, however,
that their results also supported the hypothesis that the enzyme tyrosinase
may mimic estrogen receptor binding. The possibility that steroids may
interact with nonreceptor proteins such as tyrosinase gives less credence to
the specificity of estrogen binding to receptors in melanoma skin tumors.
Rumke et al. (1980) conducted hormone-receptor assays on 21 metastatic
tumors from 17 male patients, and 22 metastic tumors from 17 female melanoma
patients. Estrogen and androgen receptors were detected in 7 out of 31
cutaneous metastases. No relationship was observed between estrogen receptor
and sex, age, androgen receptor, or prognosis. The study showed that estrogen
and androgen receptors can be present in some melanoma metastases but at .
levels generally too low to be considered of relevance to endocrine treatment.
Creagan et al. (1980) assayed 38 tumor specimens from 34 melanoma patients
for cytoplasmic estrogen receptors (ER) by the dextran-coated charcoal
method. Only 4 of the 34 patients (12 percent) had detectable ERs, leading
the authors to conclude that chemical usefulness of the ER assay in melanoma
is probably limited.
Fisher et al. (1976) analyzed biopsies from 35 malignant melanoma patients
and found 16 (46 percent) with cytoplasmic receptors for estrogen. Equal per-
centages were observed for males and females. Using the Scatchard technique,
the authors observed a straight line plot suggesting that estradiol was
binding to a single class of high affinity, limited capacity receptor sites.
Ellis et al. (1985) investigated a related topic: whether
hormone-receptor binding in melanocytic lesions could be indicative of a
potential for malignancy. Estrogen and progesterone binding was examined in
22 melanocytic lesions from 14 patients with the dysplastic nevus syndrome and
in 21 patients with acquired intradermal nevi using a fluorescent hormone
binding technique. Large amounts of both estrogen and progesterone binding
were seen in nevi from patients with dysplastic nevus syndrome, while most of
the acquired intradermal control nevi were negative for binding. The authors
concluded that positive estrogen and progesterone binding in melanocytes from
patients with dysplastic nevi may correlate with clinical lesion changes
DRAFT FINAL * * *
-------
12-9
during times of normal hormonal change (e.g., puberty, pregnancy, and OC
use). The authors noted the predictive value of estrogen and progesterone
receptors in melanocytic lesions for malignancy remains unproven.
Effects of Endocrine Treatment. In a 1983 review article, Hodgins
(1983) concluded that estrogens and anti-estrogens appeared to be useful in
treatment of some melanomas, but more work was needed to identify patients
likely to respond. There was little evidence concerning the effects of
castration, androgen, or antiandrogens in the treatment of melanoma. Hodgins
cautioned that gluco-corticoids and retinoids, while able to inhibit the
growth of melanoma cells, have widespread negative health effects. Finally,
Hodgins stated that assays of steroid sex hormone receptors in melanomas
appeared to offer little help in selecting patients for endocrine treatment.
While premalignant melanocytic lesions and early stage tumors might be
estrogen-dependent, Hodgins suggested that as tumors spread, hormone-resistant
variants become established. Tumors could then contain a mixed population of
hormone-responsive (staining) and hormone-insensitive (nonstaining) cells.
Most hormonal treatments for CMM involve administration of the
anti-estrogen, tamoxifen (Hodgins 1983). Hodgins noted that in 10 reported
trials on 154 patients with advanced disease, partial or complete remissions
lasting from three weeks to over one year were observed in 9.7 percent of
patients. Hodgkins (1983) assessed the relationship of estrogen receptor
content of melanoma and response to tamoxifen in 58 of the 154 patients; 31
percent (N=18) contained receptors. The response rate among the 58 patients
was 10.3 percent, similar to the overall rate of remission (9.7 percent). The
majority of the 18 receptor-containing tumors were not responsive to
tamoxifen. Hodgins concluded that although 10 percent of melanoma patients
responded to anti-estrogen treatment, response was not predicted by presence
of estrogen receptors.
Hodgins also reported that Fisher et al. (1976) observed partial responses
to diethylstilbestrol in 2 of 18 patients. Hodgins noted the paradox of a
response to anti-estrogen treatment when prognosis appeared better for women
than men. Since diethylstilbestrol was effective in treating some patients,
Hodgins questioned whether tamoxifen was acting primarily as an anti-estrogen
or as a weak estrogen.
Rumke et al. (1976) commented that Fisher et al. (1976) similarly showed
that the presence of estrogen receptors did not correlate with a response to
diethylstilbestrol treatment. The two responsive patients did not indicate
estrogen-binding activity while 4 of the 18 patients with estrogen-receptor in
cells from metastases did not respond. Rumke et al. (1980) noted that
although hormonal dependence of melanoma growth rate has been shown, it occurs
in so few patients with advanced disease that endocrine therapy for CMM is not
a customary practice as it is for mammary carcinoma. After observing
measurable androgen binding activity in 11 of 43 melanoma metastases, two
young male patients were given an anti-testosterone treatment and one also
received ethinylestradiol in high doses. These treatments had no effect on
disease progression. Rumke et al. (1980) tentatively concluded that receptor
determinations were not useful in the management of patients with advanced
disease.
* * DRAFT FINAL
-------
12-10
Adler and Gaeta (1979) caution against use of stribestrol, estrogen, or
estrogen-progesterone combinations among by females with a diagnosis or past
history of melanoma since, in some instances, reactivation of tumor growth has
been observed after hormonal treatments.
OCCURRENCE OF MELANOMA IN IMMUNOSUPPRESSED PATIENTS
Immunosuppression can occur through a variety of mechanisms. Some
patients are born with diseases that have immunosuppressive components. By
far the most common cause of immunosuppression, however, is iatrogenic; drugs
given to transplant recipients and patients with autoimmune diseases suppress
the immune response in an effort to promote graft survival or decrease the
autoimmune disease process. Cytotoxic anti-cancer drugs frequently have an
immunosuppressive side effect, and cancers themselves have been shown to exert
a suppressive effect on the immune system, especially malignancies of the
reticulo-endothelial system.
Patients who are immunosuppressed for whatever reason, have an increased
susceptibility to certain malignancies. For transplant patients, there is an
increased risk of skin cancers in particular (Penn 1980; Maize 1977; Sloan et
al. 1977; Hardie et al. 1980; Penn 1978). Because malignant melanomas carry
antigens on their surface and because the cellular inflammatory infiltrate is
said to correlate with prognosis (Balch et al. 1978), the question has been
raised as to whether the incidence of malignant melanoma is increased in
immunosuppressed patients.
Numerous case studies which report CMM in immunosuppressed patients have
been published. Comparisons with expected numbers of CMM based on incidence
rates or control groups were not conducted. Bencini et al. (1983) reported
two melanomas in a group of 105 renal transplant patients. In another report
(Penn 1980), 906 organ transplant recipients developed 399 skin cancers,
fourteen of which were CMM. Hardie et al. (1980) reported two melanomas in a
group of 301 organ recipients with fatal results. Among 50 patients on
immunosuppressive therapy for glomerulonephritis or collagen diseases, one
patient developed melanoma (Walker et al. 1976). In the same report, none of
the 135 kidney allograft recipients developed melanoma. In a series of 1,884
renal allograft recipients, three developed melanoma (Sheil 1977). Brody et
al. (1977) reported 21 second malignancies among 1,028 patients originally
treated for Hodgkin's disease; one one of the 21 cancers was diagnosed as CMM
(Brody et al. 1977). In another case report, one patient developed a melanoma
in a preexisting mole five years after renal transplantation (Younis et al.
1980). Chaudhuri et al. (1980) reported six cases of melanoma in patients who
received immunosuppressive therapy; Hill (1976) reported on five patients who
developed skin malignancies after immunosuppressive drug therapy for lymphoma,
one of which was CMM.
Greene et al. (1981) reported clinical and histological data on 13
patients who developed 14 cutaneous malignant melanomas after renal
transplantation. The primary CMM's were histologically reviewed for 13 of the
14 tumors. Ten of the melanomas arose from a precursor nevus. There was also
an abnormal host response to the tumor in 10 of the 13 patients indicated by
the absence of the normal lymphocyte/macrophage infiltrate. Greene et al.
* * * DRAFT FINAL * * *
-------
12-11
(1978), in their earlier study of 4,869 patients with chronic lymphocytic
leukemia, observed that CMM developed in nine patients, compared to the 1.34
expected (based on inciderice rates for the general population from the
Connecticut Tumor Registry) for an increased relative ratio (RR) of 6.7
(p<0.05). When treatment modalities were compared, there was no significant
increase in risk of CMM for untreated patients (RR=3.2, 95% C.I.=0.4-12.0).
However, patients treated with chemotherapy (RR=12.0, 95% C.I.=3.0-43.8) or
radiation (RR=16.8, 95% C.I.=5.4-51.2), were at significantly elevated risk of
CMM relative to expected numbers of cases based on age, sex, and time specific
incidence rates from the NCI End Results Program. Thus, increased risk of CMM
was associated with immunosuppressive treatment regimes, although it should be
noted that these risk estimates were based upon very small numbers of observed
cases of CMM as second primary tumors.
Hoover (1977) reported on a series of 16,290 renal transplant recipients,
six of whom developed CMM. The relative risk based on a comparison with
expected CMM rates for the Connecticut Cancer Registry (1966 and 1971) was
calculated as 3.9 (95% C.I. =-1.4-8.5). The degree of immunosuppression of
these patients was not addressed. Birkeland et al. (1975) reported a
significant (p<0.001) increase in CMM only in female patients among 418
renal transplant recipients. However, the percentage of females in the
recipient group was not given. Kinlen et al. (1979) evaluated tumor incidence
in both immunosuppressed non-transplant and immunosuppressed renal transplant
patients. The expected numbers of CMM cases were derived from population
incidence rates in an area whose incidence of melanoma was thought to be
similar to that of the study population. Kinlen et al. (1979) .found an
observed/expected ratio of 5.0 for renal transplant recipients and 9.0 for
non-transplant immunosuppressed patients. Spector and Filipovich (1980)
studied cancers arising in patients with naturally occurring immunodeficiency
diseases. Two melanomas occurred in a. registry of 298 patients for a relative
risk of 2.9; however the 95% C.I. = 0.8-9.0. These data are from Greene et
al. (1981) who discussed this study and received additional data from the
authors so that the observed/expected ratio could be calculated based on age-
and sex-specific rates for melanomas in the U.S.
More recently, a report appeared which specifically examined the incidence
of CMM in patients who had been treated for Hodgkin's disease (Tucker et al.
1985). Eight cutaneous malignant melanomas were diagnosed in six of 1,405
patients with Hodgkin's disease. The relative risk was 8.0 (95% C.I.=3-17).
Of the six melanomas histologically reviewed, all had a sparse inflammatory
cell infiltrate, as did those from renal transplant patients (Greene et al.
1981). Precursor nevi were identified in five of the six CMM tumors, a
finding also in agreement with Greene et al. (1981).
FINDINGS
Two different topics potentially related to melanoma have been discussed
in this chapter: steroid hormones and CMM, and melanoma among
immunosuppressed patients. A few general findings can be drawn from the
epidemiological data for these areas:
* * * DRAFT FINAL * * *
-------
12-12
12.1 With the possible exception of long-term oral
contraceptive (OC) use, hormonal status does not
appear to impact the risk of CMM. The potential
effects of short-term OC use are not known.
Epidemiological results relating OC use to melanoma
may, however, be confounded by several factors such as
socio-economic status.
12.2 Available epidemiological evidence indicates that
pregnant females have similar risks of developing CMM
as non-pregnant females. In addition, survival rates
have not been observed to significantly differ between
pregnant females with melanoma and non-pregnant
females with melanoma after controlling for age,
anatomic site and stage at diagnosis. Limited
evidence has indicated that pregnancy may, however,
activate metastatic disease in a person with a
previously treated melanoma, or increase the growth
rate of a previously untreated primary CMM.
12.3 CMMs have been reported to occur at an increased rate
in immunosuppressed patients. The tumors which appear
in immunosuppressed patients may have a worse
prognosis (Greene et al. 1981; Tucker et al. 1985);
they may lack the normal macrophage/lymphocyte
inflammatory infiltrate (Balch et al. 1978) associated
with a good prognosis.
DRAFT FINAL *
-------
12-13
REFERENCES
Adam, S.A., Sheaves, J.K., Wright, N.H., Mosser, G., Harris, R.W. and Vessey,
M.P. A case control study of the possible association between oral
contraceptives and malignant melanoma. Brit J Cancer 44:45 (1981).
Adler, S. and Gaeta, J.F. Malignant Melanoma. Chapter 12, In: Cancer
Dermatology Helm, F. (ed). Philadelphia: Lea and Febiger pp 141-157
(1979).
Anonymous. Sunlight and Melanomas. The Lancet January 23:172-173 (1971).
Bain, C., Hennekens, C.H., Speizer, F.E., Rosner, B., Willett, W. and'
Belanger, C. Oral contraceptive use and malignant melanoma. JNCI 68:537
(1982).
Balch, C.M., Mured, T.M., Soong, S.J., Ingalls, A.L., Halpern, W.B., Maddox,
W.A. A multifactorial analysis of melanoma: Prognostic histopathological
features comparing Clark's and Breslow's staging methods. Ann Surg
188:737-742 (1978).
Bencini, P.L., Montagnino, G., DeVecchi, A., Taratino, A., Crosti, C.,
Caputo, R., Ponticelli, C. Cutaneous manifestations in renal transplant
recipients. Nephron 34:79-83 (1983).
Beral, V., Ramcharan, S. and Faris, R. Malignant melanoma and oral
contraceptive use among women in California. Brit J Cancer 36:804 (1977).
Birkeland, S.A., Kemp, E., Hauge, M. Renal transplantation and cancer. The
Scandia transplant material. Tiss Antigens 6:28-36 (1975).
Brody, R.S., Schottenfeld, D., Reid, A. Multiple primary cancer risk after
therapy for Hodgkin's disease. Cancer 40:1917-1926 (1977).
Chaudhuri, P.K., Walker, M.J. and Das Gupta, T.K. Cutaneous malignant
melanoma after immunosuppression therapy. Arch Surg 115:322-323 (1980).
Creagan, E.T., Ingle, J.N., Woods, J.E., Pritchard, D.J., and Jiang, N.S.
Estrogen receptors in patients with malignant melanoma. Cancer 46:1785-1786
(1980).
Ellis, D.L., Wheeland, R.G., and Solomon, H. Estrogen and progesterone
receptors in melanocyte lesions: Occurrence in patients with Dysplastic Nevus
Syndrome. Arch Dermatol 121:1282-1285 (1985).
Elwood, J.M. and Goldman, A.J. Previous pregnancy and melanoma prognosis.
Lancet 2:1000-1001 (1978).
* DRAFT FINAL * * *
-------
12-14
Fisher, R.I., Neifeld, J.P., and Lippman, M.E. Oestrogen receptors in human
malignant melanoma. Lancet 2:337-338 (1976).
Foucar, E., Bentley, T.J., Laube, D.W., and Rosai, J. A histopathologic
evaluation of nevocellular nevi in pregnancy. Arch Dermatol 121:350-354
(1985).
Greene, M.H., Clark, W.H., Tucker, M.A., Elder, D.E., Kraemer, K.H., Guerry,
D., Witmer, W.K., Thompson, J., Matozzo, I., and Fraser, M.C. Medical
intelligence current concepts: Acquired precursors of cutaneous malignant
melanoma, the Familial Dysplastic Nevus Syndrome. New Engl J Med 312(2):91-97
(1985).
Greene, M.H., Hoover, R.N., Fraumen Jr., J.F. Subsequent cancer in patients
with chronic lymphocytic leukemia -- A possible immunologic mechanism. JNCI
61:337-340 (1978).
Greene, M.H. Young, T.I., Clark Jr., W.H. Malignant melanoma in renal-
transplant patients. Lancet 1:1196-1199 (1981).
Hardie, I.R., Strong, R.W., Hartley, L.C.J., Woodruff, P.W.H., Clunie, G.J.A.
Skin cancer in Caucasian renal allograft recipients living in a subtropical
climate. Surg 87:177-183 (1980).
Hersey, P., Morgan, G., Stone, D.E., McCarthy, W.H., and Milton, G.W. 1977.
Lancet 451i. As cited in Elwood and Goldman (1978).
Hill, B.H.R. Immunosuppressive drug therapy as a potentiator of skin tumors
in five patients with lymphoma. Aust J Derm 17:46-48 (1976).
Hodgins, M.B. Steroid hormones, receptors and malignant melanoma. Pigment
Cell 6:116-126 (1983).
Holly, E.A., Weiss, N.S. and Liff, J.M. Cutaneous melanoma in relation to
exogenous hormones and reproductive factors. JNCI 70:827 (1983).
Holman, C.D.J., Armstong, B.K. and Heenan, P.J. Cutaneous malignant melanoma
in women: Exogenous sex hormones and reproductive factors. Brit J Cancer
50:673-680 (1984).
Hoover, R. Effects of Drugs -- Immunosuppression. In: Origins of human
cancer, Book A. Incidence of cancer in humans. Hiatt, H.H., Watson, J.D.,
Winston, J.A. (eds). pp 369-379 (1977).
Houghton, A.N., Flannery, J., and Viola, M.V. Malignant melanoma of the skin
occuring during pregnancy. Cancer 48:407-410 (1981).
Jelinek, J.E. Cutaneous side effects of oral contraceptives. Arch Derm
101:181-186 (1970).
* * DRAFT FINAL * * *
-------
CHAPTER 13
PREDISPOSING CONDITIONS/LESIONS FOR MELANOMA
There are certain genetic conditions which are known to be associated with
an increased incidence of melanoma. This chapter reviews information about
two such conditions, dysplastic nevus syndrome and xeroderma pigmentosum
(XP). It examines how knowledge gained from investigation of the dysplastic
nevus syndrome may explain the development of melanoma in the normal
population and also reviews information on both congenital and acquired nevi
as risk factors in the development of malignant melanoma. The chapter also
summarizes what is known about the role of solar radiation in the development
of normal acquired nevi as well as in the progression of dysplastic nevi to
cutaneous malignant melanoma (CMM). Information about XP is reviewed in order
to gain insight into the factors that contribute to a vastly increased rate of
melanoma in patients with this disorder. In addition, information about the
action spectrum of the molecular defect in XP is reviewed in order to evaluate
the role of UVB in the cutaneous cancers which these patients develop.
Finally, a third syndrome, albinism, is evaluated in order to understand what
the reported normal incidence of melanoma in albinos means in light of the
apparently protective role of melanin in the development of CMM.
DYSPLASTIC NEVUS SYNDROME
According to Greene (1984), "dysplastic nevus syndrome" (DNS) was first
identified in 1976. It was originally characterized as familial melanoma
associated with a distinctive cutaneous pattern of unusual (dysplastic) nevi
(Elder et al. 1983). Subsequent investigations have identified these
dysplastic nevi in familial and nonfamilial settings and in individuals with
or without melanoma leading to the classification of DNS kindred into four
types (A through D) with the risk of melanoma increasing from one type to the
next as indicated in Figure 13-1 (Kraemer and Greene 1985; Greene et al.
1985b). Type A kindred comprise nonfarailial (sporadic) and type B familial
dysplastic nevus syndrome. Type C kindred are comprised of a single
individual with dysplastic nevi and melanoma but no family history of either
(this is one form of so-called "sporadic" melanoma). The last group, type D,
are kindred with either one (type Dl) or at least two (type D2) family members
with both melanoma and dysplastic nevi. As a familial condition, DNS is
apparently inherited in an autosomal dominant fashion with a highly penetrant
gene possibly linked to the Rh locus (Greene et al. 1983; Bale et al. 1986).
The frequency of familial DNS is not known; however, as of 1985, 150
melanoma prone families had been identified by the National Cancer Institute
(NCI) and it has been estimated that there may be as many as 32,000 people
with familial DNS in the US. These individuals could account for as much as
5.5 percent of all cases of CMM in the US. Furthermore, dysplastic nevi (DN)
have been described in between 25 to 35 percent of patients with melanoma and
in as much as five percent of the general population, potentially making DN
very important as precursor and marker lesions for this disease. It should be
noted however, that as many as 4.6 million people living in the US are
DRAFT FINAL
-------
13-2
DYSF
S
TYPE
A
B
C
D-1
D-2
•LASTIC NEVUS
YNDROME
DESCRIPTION
Sporadic
Dysplastic Nevi
Familial
Dysplastic Nevi
Sporadic
Dysplastic Nevi
with Melanoma
Familial
Dysplastic Nevi
with Melanoma
MELANOMA DYSPLASTIC NEVI
IN IN TWO OR MORE
KINDRED? BLOOD RELATIVES?
-
4-
•+
+ +
4- +* +
NUMBER OF
MELANOMA PEOPLE
RISK AFFECTED
LOW MANY
II
HIGH FEW
"AT LEAST TWO BLOOD RELATIVES WITH CUTANEOUS MELANOMA
FIGURE 13-1
CLASSIFICATION OF KINDREDS WITH
DYSPLASTIC NEVUS SYNDROME
Source: Kraemer and Greene (1985a) and Greene et al. (1985b)
DRAFT FINAL
-------
13-3
estimated to have at least one dysplastic nevus. The risk of individuals in
this population is under investigation; it is clearly substantially less than
that of individuals with the familial disease (Kraemer and Greene 1985; Greene
et al. 1985a).
The worldwide distribution of DNS is also not known, but families have
also been identified in Germany, France, Great Britain, the Netherlands,
Japan, Australia and New Zealand (Kraemer and Greene 1985; Elder 1986).
Relationship to common acquired nevi: The characteristics of
dysplastic nevus syndrome are best considered in comparison with ordinary
acquired nevi (moles). Typically acquired nevi are absent at birth, first
appear in early childhood, increase in number through middle adult life and
decrease in number thereafter. The average white middle-aged adult may have
10 to 40 moles, usually on sun-exposed areas of the body and especially above
the waist (Kraemer and Greene 1985; Greene et al. 1985a). As a general rule,
common acquired nevi are smooth round or oval, pigmented lesions which are
sharply demarcated from the surrounding skin (Greene et al. 1985a; Kraemer and
Greene 1985). Their pigment is most often a uniform brown or tan, however,
some nevi show mottled or stippled variations in these colors.
The characteristic cell type of the common acquired mole is the nevus
cell. Nevus cells, although basically a subset of melanocytes, differ from
melanocytes by generally being somewhat larger and by lacking dendritic
processes (Elder et al. 1983; Lever and Schaumburg-Lever 1979). Another
distinguishing characteristic is the aggregation of nevus cells into nests or
clusters. This is not an obligate characteristic, however, because many nevi
show unusual concentrations of melanocytes located basally without the nested
organization (Elder et al. 1982).
Nevus cells in the epidermis and dermis show considerable variation in
their appearance. In the lower epidermis and upper dermis, nevus cells are
usually cuboidal or oval in shape, possess a well demarcated homogeneous
cytoplasm, a large, round or oval nucleus and frequently contain melanin
(Lever and Schaumburg-Lever 1979). Such cells are sometimes referred to as
"epithelioid" or "type A" cells (Elder et al. 1981; Lever and Schaumburg-Lever
1985). Nevus cells in the mid dermis (type B cells) are smaller than type A
cells; they rarely contain melanin and thus may resemble lymphocytes, which
has led them to also be characterized as small or "lymphocytoid" (Elder et al.
1981). Nevus cells in the lower dermis (type C) resemble fibroblasts or
Schwann cells, since they are usually elongated and possess a spindle-shaped
nucleus (Lever and Schaurnburg-Lever 1979). Sometimes also referred to as
"neuroid" cells, these nevus cells do not contain melanosomes and are
tyrosinase negative. Type B cells may have both cholinesterase and the
enzymes of melanogenesis and are considered to be transitional between type A
and type C cells (Elder et al. 1981) leading to the suggestion by these
authors "that the r.eurotized cells of dermal nevi arise from the epidermal
nevus cells by the process of 'abtropfung' and of subsequent maturation..."
This concept of nevus cell maturation views the acquired nevus as "...an
evolving lesion in which melanocytes ('nevus cells') proliferate in the
epidermis, drop into the dermis and undergo maturation there"(Elder et al.
1981).
* * * DRAFT FINAL
-------
13-4
Nevi in which the nevus cells lie primarily on the dermal-epidermal
junction are termed junctional nevi. This is the initial state of most nevi
which are generally first observed as tiny (1 to 2 mm) pigmented "dots" on the
surface of the skin (Clark et al. 1984). These small lesions gradually
enlarge to a maximal diameter of 4 to 6 mm (Elder et al. 1985; Kraemer and
Greene 1985). After this initial lateral development, most nevi develop
further along one of two paths. One group becomes arrested and remains a
stable population of uniformly brown flat lesions. In the second population,
while lateral growth ceases, vertical growth begins with cells also descending
into the dermis (Clark et al. 1984). The resulting elevated pigmented papule
is commonly known as a compound nevus. In the next stage in nevus
development, nevus cells continue their descent into the dermis and disappear
from the dermo-epidermal junction; the resulting lesion is a non-pigmented
(pink or flesh-colored) papule (dermal nevus). The natural history of the
common acquired nevus ends with the return to normal of the skin, either
through the sloughing off of a pedunculated papular remnant or by
differentiation of nevus tissue along neural (Schwannian) lines (Elder et al.
1981). This process of evolution occurs over many decades and may stop at any
point along the way thus giving rise to the continuum of nevi seen on a given
individual.
Characteristics of dysplastic nevi: In dysplastic nevus syndrome,
affected individuals tend to have larger and more numerous nevi. Such
individuals may have between 25-75 abnormal nevi although even 100 would not
be unusual (Elder et al. 1981). Typically, dysplastic nevi range in size
between 6 and 15 mm in diameter; however, this cannot be taken as
pathognomonic as some common acquired nevi may be as large and patients with
DNS may have both dysplastic as well as normal nevi.
Dysplastic nevi follow the same initial time course as ordinary acquired
nevi - absent at birth, then appearing in the first years of life. The first
indication that a patient may have DNS is an abnormally large number (20 to
40) of small, uniform, deeply pigmented nevi generally noticed between the
ages of five and eight. With the onset of puberty, some of the nevi may take
on an aberrant morphology, such that by the late teens or early twenties the
syndrome is fully manifested. Another characteristic of DNS is that nevus
development does not cease with middle age as is seen with normal acquired
nevi, but rather continues, though at a slower pace, throughout life (Greene
et al. 1985a; Kraemer and Greene 1981).
Dysplastic nevi have a characteristic color and morphology that
distinguishes them from ordinary acquired nevi. DN are generally variegated
in color showing random mixtures of tan, brown, dark brown and pink. Their
margins are diffuse, often fading into the adjacent skin. DN are
predominantly flat (i.e., macular), colored lesions but may have a complex
topology ranging from a "pebbled" appearance to that of a central elevated
papule surrounded by a colored area which is not elevated (Kraemer and Greene
1985; Greene et al. 1985a).
Histologically, there are two characteristics that must be present in
order to make the diagnosis of dysplastic nevus: a few or many (but almost
DRAFT FINAL
-------
13-5
always less than 50 percent) of the component melanocytes display nuclear
pleomorphisra and hyperchromatisra, and there is a dermal lymphocytic and
fibroblastic host response. In addition, these are associated patterns of
nevus cell arrangement including variations in the size and shape of the
nests, bridging of the nests between the rete ridges and, in the dermis, a
tendency of the nevus cells to be of the small type without neurotization
(Elder et al. 1981; Elder 1986).
Relationship of DNS to melanoma: Progression of dysplastic nevi into
frank superficial spreading melanoma has been documented in familial DNS and,
even in patients with the sporadic form of melanoma, dysplastic nevi have been
demonstrated contiguous to melanoma in 36% of 300 superficial spreading
melanoma (Elder et al. 1981). In a prospective study of 14 families with
familial melanoma (2 family members with melanoma), of 51 evaluable patients
with melanoma, nevi played a role in melanoma development in all and in 49 of
the 51, the nevi were dysplastic (Greene et al. 1985b)
Almost 90 percent of the melanomas observed in patients with DNS are SSM
with the bulk of the remainder (8 percent) being nodular. HMFM (lentigo
maligna) which comprises about 16 percent of melanomas in the normal
population only accounts for 1 percent of melanomas in DNS patients. The
trunk is the predominant location and the median age is 32 - considerably
younger than the 51 years observed for the general population (Kraemer and
Greene 1985).
Kraemer et al. (1983) in attempting to quantify the risk of melanoma in
DNS patients as compared to the general public, estimated that the type D2
individual had a melanoma risk 395 times that of the general population and a
lifetime melanoma risk of 100 percent. Other DNS patients are estimated to
have a relative risk 26 times greater than the general population and a
lifetime melanoma risk of 20 percent. If these estimates are recalculated
using current estimates of DNS prevalence (5%), the estimated risk falls to
about 10 times greater than the normal population, which is consistent with
unpublished observations from the Pigmented Lesions Group at the University of
Pennsylvania, where patients with dysplastic nevi are followed (Elder 1986).
Abnormal sensitivity to UV in DNS: The finding in xeroderma
pigmentosum (XP) patients that their sensitivity to sunlight was associated
with a decreased ability to repair UV induced DNA damage (discussed in detail
in a subsequent section of this chapter) led to similiar investigations in
patients with DNS. The first report of such an investigation was that of Smith
et al. (1982) in which fibroblast strains from five FM/DNS families were
examined for their survival, and DNA synthesis and repair characteristics
after irradiation with either 254 nm UV or gamma-radiation. Fibroblasts from
FM/DNS patients showed an enhanced UV-sensitivity ranging in severity from low
normal to a two-fold increase comparable to that seen in the form of XP termed
XP variant. A similiar sensitivity to gamma-radiation was not demonstrable.,
nor was it possible to demonstrate a defect in DNA repair synthesis. The
efficiency of pyrimidine dimer repair in FM/DNS strains was normal and the
recovery of DNA repair after irradiation was also essentially normal.
DRAFT FINAL * * *
-------
13-6
In additional studies of the same fibroblast strains, this group (Smith et
al. 1983) investigated the response of these cell lines to the carcinogen,
4-nitroquinoline 1-oxide (4NQO). This compound induces a type of DNA damage
which is not repaired by XP cells; as a consequence it has been termed
"UV-like" (Cleaver 1983). Smith et al. (1983) found that three of the six
FM/DN7S showed increased cell killing as compared to normal controls when
exposed to 4NQO. Investigations of the inhibition and recovery of DNA
synthesis in the affected cell lines indicated that this was not the mechanism-
of their increased sensitivity to 4NQO nor was DNA repair synthesis
deficient. As an alternative hypothesis, these authors suggest that some
class of DNA damage other than pyrimidine dimers may constitute the
potentially lethal lesions induced by 4NQO in the sensitive FM/DNS
cell-lines. This hypothesis was supported further by their preliminary
observation that one of the FM/DNS cell lines showed increased cell killing
compared to normal cells following exposure to long wavelength (365 nm) UV
whose main cytotoxic effects are not related to pyrimidine dimer formation.
Although the UV- and carcinogen-induced repair synthesis experiments
indicate that excision repair of pyrimidine dimers is not compromised in DNS
patients, the fidelity of the repair process was not addressed by these
experiments. More recent work has investigated this question by examining
mutagenesis at the hypoxanthine-guanine phosphoribosyltransferase (HGPRT)
locus in lymphoblastoid cells from FM/DNS patients as compared to those from
normal or XP individuals (Perera et al. 1986). Four parameters were examined
in these studies: survival following 254 nm irradiation, frequency of mutants
per clonable cell following increasing doses of UV, recovery of DNA synthesis
following irradiation, and strand breakage and repair as measured by alkaline
elution. The only abnormality demonstrable in DNS cells was a dose-related
increase in HGPRT mutants with increasing doses of 254 nm radiation.
Survival, recovery of DNA synthesis and DNA strand breakage and repair were
all normal following irradiation.
One possible interpretation of the finding that DNS is associated with
hypermutability but normal excision repair is that this syndome is associated
with an error-prone DNA repair process. As indicated by Perera et al. (1986)
this hypothesis, that genetic instability may contribute to the development of
melanoma in susceptible kindreds, finds additional support in the finding that
karyotypic analysis of DNS individuals with or without melanoma shows
excessive numbers of apparently random chromosomal changes in number and
structure (Caporaso et al. 1986).
Work by Richmond et al. (1986) suggests that the random chromosomal
changes may be related to the acquisition by nevus cells of the ability to
produce endogenous mitogenic growth factors. This ability in turn may be due
to mutation in a regulatory gene controlling production of such factors.
The finding by Perera et al. (1986) of apparently normal survival in
lymphoblastoid lines from DNS patients conflicts with the observations of
Smith et al. (1982) that fibroblasts from DNS patients showed abnormal
survival characteristics. The two groups used cell lines derived from the
same individuals, yet came to very different conclusions with regard to the
••'- * » DRAFT FINAL
-------
13-7
survival characteristics of these cells. This may be due to great differences
in approaches, however, it surely warrants further investigation.
Other information which may be important to the analysis of the role of
UVR in melanoma development comes from work by Smith and Paterson (1981) in
which they compared the responses to 254 or 313 nm of cells from patients with
disorders featuring either sensitivity to sunlight or sensitivity to ionizing
radiation. The first group of patients contained one patient with
photosensitive myositis (PM) and basal cell carcinoma (BCC), two XP patients
and one patient with Bloom's syndrome(BS). The second group of patients
contained four with ataxia telangietasia (AT), one AT heterozygote, five with
Franconi's anemia (FA), two with Rothmund Thomson Syndrome (RTS) and one with
hereditary retinoflastoraa.
A very interesting pattern of responses to these two wavelengths was
found. Cells from the patient with PM and BCC were sensitive (i.e., showed a
reduced colony forming ability) to 313 nm alone, whereas the XP and BS
patients were sensitive to both UV wavelengths. In the second group of
patients, 2 of 4 AT strains showed hypersensitivity to 313 nm only, one FA
strain was sensitive to both 254 nm and 313 nm and normal sensitivity to both
wavelengths was shown by the remaining 4 FA and 2 AT strains as well as the
RTS, AT heterozygote and hereditary retinoblastoma strains.
Biochemical studies of the strains sensitive to 313 nm UV radiation
suggested that this sensitivity was not due to pyrimidine dimers. In
investigating the yield of pyrimidine dimers induced at the two wavelengths,
it was discovered that the cell line which was sensitive to 313 nm on the
basis of colony forming ability showed no more pyrimidine dimers than cells
without such sensitivity. Furthermore, the repair of pyrimidine dimers in the
313 nm-sensitive strain was equal to that in a normal strain.
Smith and Paterson (1981) also draw attention to the fact that the three
strains with preferential sensitivity to 313 nm are also cosensitive to
ionizing radiation (in particular JT-rays). There is genetic heterogeneity
in the response of AT strains with respect to their responses to DNA damaging
agents, however, and this may thus explain the heterogenicity of AT
cell-responsiveness observed here. One possible explanation for the observed
differences in cosensitivity to ionizing radiation and mid UV-observed in
these studies is that If irradiation induces a spectrum of lesions, one of
which is identical to that induced by 313 nm. Thus the cosensitivity seen in
the PM/BCC patient is due to an identical defect in the same repair pathway.
The difference in the different AT strains could then be explained on the
basis that those that are not sensitive to irradiation at 313 nm, develop
other lesions after 2f irradiation that they are not able to repair because
of a defect in a repair pathway separate from the one responsible for the
sensitivity to 313 nm.
These authors conclude their report by suggesting that "(a) mid UV appears
to be a more appropriate probe than far UV for cellular sensitivity to
sunlight; (b) non-dimer coproducts may have biologically important
DRAFT FINAL
-------
13-8
consequences if their repair is singularly inefficient and (c) mid UV has a
partially radiomimetic biologically damaging action on human cells..."(Smith
and Paterson 1981).
Other observations which appear to support the conclusions of Smith and
Paterson are those by Tyre 11 (1984) and Colella et al. (1986) with regard to
mid UV induction of mutation to ouabain resistance. In contrast to mutation
to 6 thioguanine resistance, which can be induced by many different DNA
damaging events, mutation to resistance to ouabain seems to involve base-pair
substitution events only (Colella et al. 1986). Tyrell found that irradiation
of human lymphoblasts with 313 run light induced 10 times more ouabain
resistant mutants than irradiation with either 302 or 254 nm UV. Colella et
al. (1986) extend the Tyrell (1984) observation to 308 nm. Taken together
these findings suggest that certain UVB wavelengths, particularly those
bordering UVA, may induce a different kind of lesion from UVC.
Relationship of congenital and acquired melanocytic nevi to melanoma:
In addition to the association of dysplastic nevi with melanoma described
above, there is also evidence that both congenital and acquired melanocytic
nevi may be associated with an increased risk of melanoma. Congenital
melanocytic nevi differ from the acquired melanocytic nevi in that they appear
as pigmented lesions at birth, are-generally relatively large in size, are
composed of nevus cells that penetrate into the lower two-thirds or deeper of
the reticular dermis and often involve dermal appendages and neurovascular
structures (Mark et al. 1973; Rhodes et al. 1985). The majority of
congenital nevi (>90 percent) fall in the "small" category by virtue of
being less than 4 mm in size. The- designation "large" is generally applied to
lesions which cannot be excised easily, with "giant" being a subset of "large"
and gigantic or garment being applied to nevi which cover a major anatomic
area (Rhodes 1983). Congenital nevi 10 cm or larger in size occur at a
frequency of less than 1 in 20,000 newborns (Mark et al. 1973), yet it is this
class of congenital nevi for which the best documentation of an increased risk
of melanoma exists; it is estimated that such nevi are associated with at
least a 6.3 percent increase in lifetime risk (Rhodes 1983). The so-called
"small" lesions were once estimated to be associated with an increased risk to
age 60 of as high as 4.9 percent (Rhodes and Melski 1982); however a recent
re-evaluation of the diagnostic criteria by which to identify small congenital
nevi has led this group to urge "...caution when interpreting the histologic
association [they] reported previously for small congenital nevi and cutaneous
melanoma." (Rhodes et al. 1985).
It has also been noted that on the basis of clinical history, primary
melanoma has been associated with a pre-existing nevus in from 18 to 85
percent of such tumors, whereas on the basis of histopathology, nevi have been
identified contiguous to melanoma with a frequency between 18 to 72 percent
(Elder et al. 1981). Very few of the studies summarized above indicated the
type of nevus reported.
Clark et al. (1984) believe that the common acquired melanocytic nevus is
a precursor lesion to melanoma. These authors have concluded that there are
six lesional steps in tumor progression, the first one being a common acquired
DRAFT FINAL
-------
13-9
melanocytic nevus. If the nevus does not follow a normal differentiation
pathway but acquires the characteristic of lentiginous hyperplasia, then it
has started down the pathway to melanoma. Subsequent steps include the
acquisition of melanocytic nuclear atypia (i.e., melanocytic dysplasia)
followed by progression to the radial, then the vertical growth phases of
primary melanoma, and finally metastatic melanoma. In this scheme the
dysplastic nevus is the second step down an aberrant pathway of
differentiation.
What leads to this aberrant differentiation process is not clear.
Conceivably, one possible stimulus is a promotional effect of sunlight.
Holman et al. (1983) found that there was a seasonal variation in the
functional component of pigmented nevi that corresponded with heavy exposure
of sun. The junctional component observed by Holman et al. (1983) may well
equate to the lentiginous melanocytic hyperplasia described by Clark et al.
(1984). Such a promotional effect may allow the nevus cell to express the
endogenous mitogenic factor described by Richmond et al. (1986) so that cell
division once promoted by sunlight now becomes permanently "turned on" by an
endogenous raitogen. This in turn would lead to the cytogenetic abnormalities
observed in tissue specimens of nevi and melanomas (Caporaso et al. 1986) and
in cultured cells from nevi and melanomas (Richmond et al. 1986). It would be
interesting to know if other cells, e.g., fibroblasts and lymphocytes from
dysplastic nevus syndrome patients, acquire the ability to produce endogenous
mitogens following UV irradiation. If such were the case, this might explain
the hypermutability (Perera et al. 1986) and other abnormal responses (Smith
et al. 1982; Smith et al. 1983) observed in these cells following UV radiation.
Frequency and distribution of nevi in normal populations: The
frequency of nevi in the normal population has been the subject of several
studies (Pack et al. 1952; Nicholls 1973; Cooke et al. 1985; Mackie et al.
1985). The study by Pack et al. (1952) looked at all moles and determined an
average value of 14.6 moles per white adult. Nicholls (1973) evaluated only
moles 2 mm in diameter or larger and did a complete age, sex and site
distribution evaluating 15 age categories and 10 sites per sex. The
population evaluated was from Sydney, Australia. No details were given on how
his population was selected but more than 85 percent of the study population
were in the first and second decades. His data indicate that, in this
population, males have their peak number of moles by age 15 while the peak age
for females is 20-29. Furthermore the peak number in males was higher than
that in females, and the distribution was different; males had more moles on
their trunk and females showed the highest concentration on their legs.
Nicholls (1973) concluded that sunlight is important to the development of
pigmented moles because the number of moles reaches the peak frequency on the
more sun-exposed areas sooner than on the less sun-exposed areas.
A study by Mackie et al. (1985) used a population selected from healthy
non-hospitalized volunteers with no family history of melanoma and (in the
case of the older age groups which were drawn from long-stay geriatric
patients) without a history of drug therapy that might have affected the
melanocyte system. Total body mole counts were performed on 432 individuals
(204 males and 228 females) whose ages ranged from 4 days to 96 years.
DRAFT FINAL * *
-------
13-10
Information was gathered on each subject's phenotype with regard to skin type,
hair and eye color and tendency to freckle. A mole was defined as any brown
pigmented lesion £3 mm in diameter which was present throughout the year
even without solar stimulation. The study subjects were presumably drawn
from the population of Glasgow although this was never stated and the
population was identified only as "Scottish" and "a Caucasian population in a
temperate climate."
Mean and median mole counts were calculated for males and females by
decade intervals with the last interval being "80+." Table 13-1 presents the
data from this analysis organized by sex and decade. In the first decade (0-9
years), the respective mean mole counts for males and females were 3 and 2 and
the median counts were 2 and 0. Counts begin to rise sharply in the second
decade, and peaked in the third decade with the respective peak means for
males and females being 22 and 33 (respective medians: 24 and 16). After the
third decade counts slowly decline. The rapid rise was associated with
puberty in both sexes leading the authors to suggest that hormonal activity
may either stimulate the pigment production of existing but non-pigment
producing nevus cells or may lead to the expansion of the pigment producing
nevus cell system.
Although this study indicated that total body maps were made of mole
location, the analysis by site was limited to moles on palms, soles, upper arm
(shoulder to arm), trunk and lower leg (knee to ankle). Moles on soles and
palms were found in 6 men and 10 women (3.9 percent of the population); 75
percent of these lesions occurred on palms. In evaluating the site
distribution in the group of 226 subjects aged 20-59, the 122 women had a mean
total body mole count of 26 with the distribution being upper arm: 30 percent,
trunk: 20.5 percent, and lower leg: 17.5 percent. For the 104 males, the mean
total body mole count was 16 and the distribution was upper arm: 20 percent,
trunk: 36 percent, and lower leg: 17.5 percent. Male and female values were
significantly different for total counts and for arm and leg but not for
trunk. On a per surface area of skin the highest number of moles was found on
the female upperarm.
Individuals in this study were asked to identify any moles which had been
present on their skin since the age of five or earlier. Fifty two reponded
positively but more than half of the lesions identified had characteristics of
acquired nevi. Nevertheless, individuals who identified moles as having been
present from birth were in the upper 10 percent of their age group for total
mole counts possibly indicating a "group of individuals who have an active
melanocyte system and who may be at greater risk of developing malignant
melanoma at any body site in later life." An evaluation was also made of the
association of mole counts with pigmentary factors such as skin type, hair and
eye color and tendency to freckle. No positive associations were found.
Mackie et al. (1985) also compared the site distribution information to a
parallel case-control study of melanoma patients in the same geographic region
in which they observed a significant association of increased mole counts with
melanoma risk (Swerdlow et al. 1984) as compared to age and sex matched
DRAFT FINAL
-------
13-11
TABLE 13-1
TOTAL BODY MOLE COUNTS IN 432 HEALTHY CAUCASIAN VOLUNTEERS
Age
(Years)
0-9
10-19
20-29
30-39
40-49
50-59
60-68
70-78
80+
Mean
Female
3
23
33
25
22
16
10
6
6
Number
Male
2
18
22
11
20
7
6
6
4
of Moles
MalerFemale
Ratio
1:5
1:3
1:5
2:3
1:1
2:3
1:7
1:0
1:5
Median
Female
0
16
24
19
11.5
12
3.5
3.5
3
Number
Male
2
10
16
10
15
4
4
2
2
of Moles
Male:Female
Ratio
1:6
1:5
1:9
' 0:8
3:0
0:8
1:8
1:5
Source: Mackie et al. (1985).
* - * DRAFT FINAL
-------
13-12
controls. These authors conclude that the prevalence by site observed for
nevi is not the same as that seen for CMM thus rejecting the hypothesis that
melanoma occurs on the sites with the greatest numbers of benign nevocytic
nevi.
At about the same time as the study reported above, Cooke et al. (1985)
reported on a community based evaluation of the distribution of moles in
adults in New Zealand. The survey was performed in the winter of 1981 and
included 78 percent of the adult population of the town of Milton, Otago, New
Zealand. Nevi were counted on all parts of the body except the abdomen and
buttocks for all subjects and the anterior chest for women. In order to allow
comparison between sexes, counts for the anterior chest for men were excluded
from the reported values. Counts were made under two sets of criteria. Both
excluded freckles, lentigines and "any arrays of similarly and uniformly
pigmented macules"; one count (hereafter termed type I) included only those
nevi ^2 mm for which there was no degree of diagnostic uncertainty whereas
the second count (type II) included all moles regardless of size and even with
a small degree of uncertainty. These two different types of counts were
performed in order to allow comparison with the counts performed by Nicholls
(1973) and Pack et al. (1952).
There were 872 people on whom a type I count was performed; 436 men and
436 women. Type II counts were performed on 105 men and 73 women. Table
13-2 presents the summary data from this study. Since this study is on adults
only, the first age group evaluated is equivalent to that of the third decade
in the Mackie et al. (1985) study. As a consequence, the large increase in
nevi number with the onset of puberty observed by Mackie and her co-workers is
not shown by these data. For the type I count, there is a decrease in the
number of nevi with increasing age; however, the decrease is not as sharp as
that seen in the Mackie et al. study. In addition the data from Cooke et al.
(1985) show no significant difference between the sexes.
Table 13-3 shows a comparison of summary information from all four of
these studies. On the basis of a comparison of their findings to those of
Nicholls (1973) and Pack et al. (1952), Cooke et al. (1985) suggest that the
prevalence of moles has increased as malignant melanoma has become more
common. However, as these authors point out this conclusion is based on
observations from three groups whose study populations are separated by time
and space. When the comparison is made between two studies done at
approximately the same time but on different continents the conclusion is no
longer clear cut, for the mean number of moles is very similar, and yet the
melanoma incidence rate is 3 times higher in New Zealand than in Scotland.
Indeed, if Mackie et al. (1985) had counted nevi >2 mm instead of >3 mm,
it is very likely that their counts would have been much greater than those
observed by Cooke et al. (1985). This comparison may too have its flaws
however, because the population studied by Mackie et al. would probably have
been very uniform in its Celticness whereas the New Zealand population is
likely to have been much more heterogeneous in its generic makeup. It is
conceivable that one characteristic of the population studied by Mackie et al.
was a greater tendency to produce nevi.
DRAFT FINAL -
-------
13-13
TABLE 13-2
MEAN NUMBER OF MOLES IN NEW ZEALAND ADULTS*
I II
Mean Number of Moles -- >2 mm** Mean Number of Moles of All Sizes**
Age
(Years) Males Females Males Females
20-29 17 (115) 16 (108) 47 (29) 53 (22)
30-39 17 (113) 16 (107) 43 (26) 52 (16)
40-49 13 (83) 16 (74) 28 (24) 24 (15)
50-59 14 (75) 13 (85) 27 (18) 26 (15)
60-69 14 (50) 9 (62) 48 (8) 21 (5)
20-69 15 (436) 14 (436) 38 (105) 39 (73)
* Moles on the anterior chest, abdomen, and buttocks are not included.
** Numbers of subjects shown in parentheses.
Source: Gooke et al. (1985).
DRAFT FINAL
-------
13-14
TABLE 13-3
COMPARISON WITH PREVIOUS SURVEYS
Place and Time
of Survey
Incidence of
Malignant Melanoma
a
per 100,000
Mean Number of Moles
in Comparable Body Areas
Diameter >2 mm
All Sizes
Males Females (both sexes)
New York
c 1950
Sydney
c 1970
New Zealand
1981
Glasgow
1982
2 -- -- 12b
11 9 12
C C
15 15° 14° . 39
5 16d 26d
The sources of these estimates, which are standardized for age
using a world standard population, are given in the discussion.
Adjusted for omission of certain body areas, using data on site
distribution from the paper by Pack et al. (1952).
Directly standardized to the age distribution in the Sydney survey.
Diameter ^3 mm.
Source: Mackie et al. (1985).
* * * DRAFT FINAL
-------
13-15
Relationship of nevi to sunlight: In a study designed to investigate
whether the distribution and prevalence of nevi was associated with exposure
to sun, Kopf et al. (1978) evaluated the distribution of nevocytic nevi on the
lateral and medial aspects of arms in 1000 individuals. Only those nevi that
were "smooth-surfaced, pigmented (tan-brown-black)..., 2 mm or more in
diameter, either visibly or palpably raised..." were counted. Macular lesions
were excluded to avoid the problem of having to differentiate between
lentigines, ephelides and junctional nevi. The population was drawn
principally from patients of the Skin and Cancer Unit who were being treated
for "...various unrelated dermatological problems", some were friends or
relatives of these patients. There were 607 women and 393 men; 93 percent of
the population was white, 5.5 percent was black and 1.5 percent was oriental.
Information on the age, sex, race, amount of sun exposure and tanning ability
of each individual was determined via a history and physical examination.
In the population of 1000 individuals, there were 349 elevated nevi on the
lateral aspects of the arms and 116 on the medial aspects. Of the 1000
subjects, 234 had one or more nevi on the lateral aspect of their arms while
only 101 had nevi on the medial aspect. Figure 13-2 shows the age
distribution of subjects with nevi on the lateral and/or medial aspect of the
arms. In all decades of life, there was a greater percentage of individuals
with nevi on the lateral aspects of their arms than on the medial aspects. In
most instances, these differences were statistically significant. This
observation only held true for the white individuals; the percentage of black
individuals with nevi on the lateral aspect was equal to the percentage of
those with nevi on the medial aspect.
When these data were analyzed with respect to sun exposure, individuals
were divided among four categories: practically none, little, moderate and
much. In all groups, the number of subjects with nevi on the lateral aspects
significantly exceeded the number with nevi on the medial aspects. There also
appeared to be an inverse relationship between the amount of exposure to the
sun and the prevalence of nevi on the lateral aspects of the arm. The authors
suggest that one possible interpretation of these findings is "that exposure
to sun not only elicits nevocytic nevi, but that excessive or cumulative
exposure leads to their disappearance." The authors also evaluated whether
the ability to tan was related to the number of nevi on the lateral or medial
aspects of the arms, but no association was found.
A subsequent paper by this same group (Kopf et al. 1985) investigated the
relationship of nevi to sun exposure in individuals with dysplastic nevus
syndrome. The population examined was a group of 80 patients, all of whom had
had one or more nevi histologically confirmed as dysplastic. Thirty-one
percent of the patients had prior or concurrent CMM at the time of examination
and 23 percent had family histories of CMM in at least one first degree
relative. Thirty-two of the patients were men and forty-eight were women.
Nevi on various aspects of the thorax (lateral, anterior, or posterior)
were counted if they were 4 mm or larger and were black, tan, dark or light
brown, or dark tan. Nevi were not characterized histologically, thus they
were referred to as nevocellular nevi even through some or many of them may
DRAFT FINAL
-------
13-16
LATERAL VS. MEDIAL NEVI
§40-*
!
b »-"
»-
e
DCCAOC
s
^ «
„ ,
£ <
0-
S
it
»7T
t )
D^
20
1
177
1 "
»'
—
12
MM
* O-rJ 20-21 1O-J» 4C-41
)
2-« »«
03 < OOi 005 <00l 00} < 00 '» 0)
r> 2i' <2t "*•
is: .44 rr !t
1000
FIGURE 13-2
AGE DISTRIBUTION OF SUBJECTS WITH NEVI ON THE LATERAL
AND/OR MEDIAL ASPECTS OF THE ARMS*
•'"In this histogram the portion of the bars below the horizontal
lines represents the percent of subjects who have nevi on both
lateral and medial aspects of the arras.
Source: Kopf et al. (1978).
DRAFT FINAL
-------
13-17
have been dysplastic nevi. Table 13-4 shows the frequency distribution by sex
of nevi across the three locations as well as the cumulative total. The data
in the table clearly indicate that in DNS patients there are relatively fewer
nevi on sun protected sites compared to sun exposed sites, suggesting the
sunlight may plan a role in the induction of nevocellular nevi in patients
with DNS (Kopf et al. 1985).
Holman and Armstrong (19S4b) in one of a series of reports on a case-
control study which examined the relationship of melanoma to a number of risk
factors, present data addressing the relationship of nevi to sun exposure.
Table 13-6 shows the percent distribution of number of palpable nevi on the
arms as a function of the age at arrival in Australia. There was a
significant trend (p=0.009) towards a greater proportion of migrants having
increasing numbers of nevi if they arrived in Australia before 10 years of
age. These authors use these data in support of their suggestion that
exposure to sunlight in childhood may be a factor in the production of benign
nevi. These authors also found that the prevalence of nevi was directly
related to the frequency of painful sunburn (Holman 1982, Armstrong et al.
1986). In a subsequent report in which the response of controls was
specifically examined, a number of interesting associations were observed.
Among pigmentary characteristics the best association was with skin color of
the upper arm and the highest prevalence of nevi was in those of intermediate
darkness. Among sun exposure variables the highest prevalence of odds ratio
(POR=2.18) was observed for 3-6 painful sunburns as a child, however the
confidence interval included one and 7 or more sunburns gave a FOR of 0.97.
The only factor which gave a statistically significant FOR was usual summer
suntan over arms (POR=2.13, C.I.: 1.20-6.38).
Epidemiologic evidence relating nevi to melanoma: A number of case
control studies have specifically examined the relationship of nevi to
melanoma incidence (Holman and Armstrong 1984a, 1984b; Swerdlow et al. 1984;
Green et al. 1985; Elwood et al. 1986). The study by Holman and Armstrong
(1984a) reported on a comparison of 511 cases diagnosed with melanoma in
Western Australia in 1980-1981 with 511 population-based controls who were
chosen to match the cases by sex, 5 year birth interval and residence in
electoral subdivision. A number of constitutional risk factors were assessed
by interview, e.g., ability to tan, and other parameters were objectively
measured e.g., counts of palpable nevi, measurement of actinic skin damage,
skin color. Nevi were counted only on the arms and only below the axillae.
In order to be sure not to count freckles, only nevi that were palpably raised
above the surrounding skin were counted. (Such a procedure would skew the
count in favor of compound and intradermal nevi and against junctional nevi.)
Two parameters related to nevi were examined in this study and analyzed with
respect to their relationship with the various histogenetic types of
melanoma: number of raised nevi on the arm and number of excised benign
nevi. Table 13-5 shows the result of that analysis. It indicates that both a
history of excised moles and increasing number of raised nevi on the arms are
strong risk factors for melanoma. In a stepwise analysis of the roles of
nevi, pigmentary characteristics and family history, all were important
factors which apparently acted independently of one another. For example,
- * * DRAFT FINAL
-------
13-18
TABLE 13-4
FREQUENCY DISTRIBUTION OF NEVOCYTIC NEVI IN THREE
THORACIC LOCATIONS IN PATIENTS WITH THE
DYSPLASTIC NEVUS SYNDROME
(both sexes combined)
Thoracic Area
Lateral
Anterior
Posterior
Number of
Patients
80
80
80
Number of
Nevi
177
361
506
Average Number
of Nevi
2.2a
4.5b
6.3°
Total
80
1,044
13.0
p values: a vs b, p<0.001; b vs c, p=0.04; a vs c, p<0.001.
Source: Kopf et al. (1985).
DRAFT FINAL * * *
-------
13-19
TABLE 13-5
RELATIONSHIPS OF MALIGNANT MELANOMA TO HISTORY OF
EXCISION OF BENIGN MOLES AND NUMBER OF RAISED
NEVI ON THE ARMS
Number of
Factor Case-Control Pairs OR 95% CI
Number of Excised Benign Moles
None
1
>2
£1 Excised Benign Moles With
Histogenetic type:
HMF
SSM
UCM
NM
Number of Raised Nevi on the Arms
None
1-4
5-9
>10
>Raised Nevi on the Arms With
Histogenetic Type:
HMF
SSM
UCM
NM
507
86
267
89
51
507
86
267
89
51
1.00
1.38
5.09a
1.50
2.35b
1.83
2.00
i:00
1.96
4.03
11.31d
1.54
3.00°
1.60
3.00d
0.86-2. 21
0.26-11.46
0.57-4.01
1.29-4.32
0.63-5.55
0.63-6.70
1.44-2.65
2.37-6.85
4.92-25'.98
0.73-3.27
1.98-4.57
0.81-3.20
1.13-8.43
a
p<0.0001 for linear trends in OR.
b
p<0.01 for difference of OR from 1.0.
c
p<0.0001 for difference of OR from 1.0.
d
p<0.05 for difference of OR from 1.0.
Source: Holman and Armstrong (1984).
* -•• » DRAFT FINAL
-------
13-20
TABLE 13-6
PERCENTAGE DISTRIBUTION OF NUMBER OF PALPABLE NEVI
ON THE ARMS ACCORDING TO AGE AT ARRIVAL IN AUSTRALIA IN
CONTROLS OF CELTIC, ENGLISH, OR AUSTRALIAN ETHNICITY
Distribution, Percent According to
Number of Palpable Nevi
Age at arrival in Australia
Birth or Before 10 Years (334)
At or After 10 Years (94)
None
54.7
68.1
1-4
36.2
28.7
5
9.0b
3.2
a
Numbers in-parentheses are number of control subjects.
b
P=0.009 for the trend to increasing proportions of those born in or
arrived in Australia before 10 years of age in each nevus count
category.
Source: Holman and Armstrong (1984b).
DRAFT FINAL
-------
13-21
number of nevi £10 had an original odds ratio (OR) of 11.31; after
controlling for pigmentary characteristics and family history, the OR for this
parameter was 10.35.
Swerdlow et al. (1984) compared 131 melanoma cases with 108 controls,
assessing the association of melanoma with benign nevi as ascertained by
dermatological examination. The cases came from patients presenting in
Glasgow and Edinburgh between 1977-1984 and ranging in age between 15-84 years.
Analysis was by stratum-matched logistic regression with stratum-matching for
age, sex, and city of treatment. These authors evaluated a number of nevi- of
nevi-related factors for their association with melanoma risk: presence of
color variation or an irregular edge in nevi, number of all nevi and number of
large nevi >7 mm. Relative risks were calculated without and with an
adjustment that controlled for hair and eye color, skin type and amount of sun
exposure and excluded individuals with dysplastic nevi. (Dysplastic nevi were
found in 21 cases but no controls.) The highest unadjusted relative risks
(RR) were observed for color variation (RR=36.41, 95% C.I. = 4.67-256.34,
p<0.01) and irregular edge (RR=43.49, 95% C.I. = 5.85-323.11, p<0.01)
however, these were no longer significant once adjusted for pigmentary and
exposure factors and the dysplastic nevus trait. The presence of large
numbers of nevi remained a significant risk factor even after adjustment.
With 10-24 nevi, the unadjusted RR was 6.61 (95% C.I. = 2.61-16.78, p<0.01)
and the adjusted RR was 6.39 (95% C.I. = 2.31-17.68, p<0.01). With 25 or
more nevi the unadjusted RR was 24.83 (95% C.I. = 8.42-73.20, p<0.01) and
the adjusted RR was 19.63 (95% C.I. = 4.75-81.18, p
-------
13-22
TABLE 13-7
DISTRIBUTION OF 183 CASES OF MELANOMA AND 183
CONTROLS IN RELATIONS TO NUMBER OF NAEVI ON THE ARMS,
AND ASSOCIATED CRUDE RISK OF MELANOMA
None
Number
2-4
of Naevi on Arms
5-10 >10
Any
Cases
Controls
Relative Risk3
29
137
1.0
93
28
15.7
44
14
14.9
17
4
20.1
154
46
15,
.8
Unmatched relative risk. There were too few discordant
pairs in which the control had naevi to do the matched
analyses of separate categories of number of naevi.
Source: Green et al. (1985).
DRAFT FINAL
-------
13-23
the major risk factors were the presence of nevi (RR=30.1, 95% C.I. =
10.3-87.5), red (RR=5.9, 95% C.I. = 1.5-23.1) or blond (RR=3.5, 95% C.I. =
1.3-9.5) hair, and propensity to sunburn (RR=3.5, 95% C.I. = 0.9-13.2).
Green et al. (1985) suggest that the "mole-proneness" described in their
studies is the most important risk factor for melanoma and that the trait
represents a tendency of melanocytes to proliferate. They do not feel that
the evidence is strong enough to conclude that nevi are necessary precursors
to melanoma, however. One alternate interpretation that these authors offer
for their finding that there is not an increased risk with increasing numbers
of nevi, is that normal nevi could be phenotypic markers that indicate a
genetic predisposition to melanocyte proliferation.
The most recent study that directly examined the relationship of rievi to
melanoma was a case-control study of 83 patients from urban and suburban
Nottingham (U.K.) who first presented with CMM between 1 July 1981 and 31
March 1984 (Elwood et al. 1986). Controls were selected at random from all
eligible comparison individuals who had been inpatients or outpatients at a
Nottingham hospital in the same time period. Selection of controls was made
via a computerized system which identified individuals independently of the
number of visits. Data were gathered by home interviews using a structured
questionnaire. Because this study was also used to assess the potential role
of exposure to fluorescent lighting, a complete occupational history was
included in the questionaire as well as information on pigmentary
characteristics, and normal reaction to sun. Skin and hair color were
assessed using comparison charts developed for the Western Canada study
(Elwood et al. 1984). Only raised nevi were counted and the count was limited
to lesions on the upper arm to the shoulder.
The pigmentary characteristics evaluated in this study included numbers of
nevi on the upper arm, estimated number of moles (by subject), freckles as a
child and as an adult, hair color as a child and as an adult, eye color,
reaction to sun and history of sunburn. The greatest risk factor was 3 or
more moles on the upper arm (RR=17.0, 95% C.I. = 6.6-43.8, p<0.001); an
estimate by the patient of 15 or more moles on the body was associated with a
relative risk of 6.7 (95% C.I. = 2.7-17.0, p<0.001). Freckles as an adult
were associated with an RR of 7.0 (95% C.I. = 3.3-14.5, p<0.001) whereas
remembered freckles in childhood showed a lesser RR of 4.3 (95% C.I. =
1.7-11.1, p=0.002) for those individuals with many obvious freckles. Red or
blond hair as an adult was a slightly stronger risk factor than the same hair
color as a child; the relative risks were 2.5 and 2.2, respectively, (p=0.02
and 0.03, respectively). Skin reaction to sunlight was divided into four
responses: tan, no burn; tan, no burn if protected; burn and tan; burn easily,
tan rarely. All but the first showed significant relative risks (RR=4.7, 3.6
and 4.6). History of a painful sunburn (pain for 2 days or more) was also
associated with a significant RR of 3.2 ( p<0.001). In order to evaluate
the interrelationships of these factors, Elwood et al. (1986) performed a
multivariate analysis. The analysis indicated that the major risk factors in
this study were 3 or more nevi on the upper arm, freckles as an adult and
reaction to sunlight. Three or more nevi showed the greatest RR of 13.3 (95%
C.I. = 4.0-43.9); freckles as an adult after multivariate analysis had an RR
DRAFT FINAL
-------
13-24
of 6.0, and the RR of the three reactions to sun ranged from 3.9 for tan, no
burn if protected, to 1.8 for burn easily, rarely tan. An interesting outcome
of this analysis is the independence of the two risk factors, number of moles
and extent of freckling on face and arms. The authors conclude that "...these
two simple measures are not merely aspects of the same host characteristic,
but [may be] related to melanoma in different ways." This epidemeologic
distinction is supported by histologic evidence as well: nevi are
proliferative lesions of melanocytes whereas freckles while characterized by
hyperpigmentation are not associated with melanocytic proliferation except in
slight degree (Rhodes 1983).
XERODERMA PIGMENTOSUM
Xeroderma pigmentosum (XP) is a hereditary disease which has been' found
worldwide in individuals of Caucasian, Oriental or Negroid racial background.
In North America, XP occurs at a frequency of about 1 in 250,000; in Japan and
Israel the frequency is higher (1 in 40,000 and 1 in 100,000,
respectively)(Kraemer and Slor 1984). XP is generally inherited in an
autosomal recessive fashion although one family with a dominant form of the
disease has been described. The disease affects both males and females in
approximately equal numbers (Kraemer and Slor 1984).
In homozygotes, XP is characterized by a variety of cutaneous
manifestations including a high incidence of non-melanoma and melanoma skin
cancer (Cleaver 1983). Data summarized in Lynch et al. (1967) indicate a
melanoma incidence rate in XP patients of three percent (300 per 100,000). A
later review of the literature found a 5 percent melanoma incidence in 830
patients worldwide. However, as high as a 50 percent CMM incidence has been
observed in several small studies in Europe and the US (Kraemer and Slor
1984). This contrasts with the 3 percent incidence observed in 141 patients
from Japan (Takebe et al. 1985). Kraemer et al. (1982) in evaluating the
increased risk of XP patients below the age of 20, found a 2000-fold increase
in melanoma and a 4800-fold increase in basal and squamous cell carcinoma.
Studies using fibroblasts from XP patients indicate a defect in DNA repair
which is accompanied by a hypersensitivity to the cell killing and mutagenic
effects of UVB. These observations have led to the suggestion that DNA repair
plays a role in protection against UV-induced neoplasia (Kraemer et al.
1983). By virtue of the increased risk of both melanoma and nonmelanoma in
XP patients, these findings suggest as well that failure to repair UV-induced
damage to DNA is one possible route by which melanomas develop. This section
explores characteristics of the disease XP, how the disease differs among
various populations as well as what is known about the molecular mechanism of
XP and how this information relates to a role for UVB in melanoma development.
XP exists in two clinical forms. The major form principally involves the
skin and eyes, the other has neurological manifestations as well. In the
extreme case, XP with neurological symptoms is known as deSanctis-Cacchione
syndrome. In a recent literature review of 800 patients, the frequency of
patients with neurologic abnormalities was 20 percent; however, it was
indicated to be higher in Japan (Kraemer and Slor 1984).
DRAFT FINAL
-------
13-25
The clinical symptoms of XP generally begin very early with many patients
displaying an acute sun sensitivity in early infancy. Minimal sun exposure
may result in prolonged erythema, edema and blistering (Robbins et al. 1974).
A lower than normal minimal erythema dose may be the first diagnostic
indication of the disease; however, other skin changes such as the development
of freckles and the characteristic dryness and scaliness which gave the
disease its name (i.e., xeroderma: parchment skin), are also pathognomonic
(Cleaver 1983).
The two clinical forms of XP can be further divided into subgroups on the
basis of complementation analysis. Complementation analysis is performed by
fusing fibroblasts (to make a heterokaryon) from affected individuals and
determining if the ability to repair DNA damage is restored in the
heterokaryon. Restoration of repair indicates that the two cells have
different but complementary defects such that each provides what the other
lacked in order to repair UV damage. On the basis of complementation
analysis, nine different types of XP have been characterized. Groups A, B, D,
G and H are the neurologic forms whereas the non-neurologic forms are groups
C, E, F and "variant" (Cleaver 1983). Patients in the various complementation
groups appear with different frequency among different populations (Table
13-8) with Japan having far more group A and far fewer group C patients than
observed in other nations (Takebe et al. 1985).
The DNA repair defect in groups A through H appears to be an inability to
perform the initial step of pyrimidine dimer excision. As discussed in detail
in Chapter 18, pyrimidine dimers are the principle photoproduct of the
interaction of 254 nm UV with DNA. The repair process which handles the
repair of these dimers is termed nucleotide excision repair; it involves the
sequential function of an UV-endonuclease, an exonuclease, a DNA polymerase
and a ligase. The endonuclease nicks the DNA next to the dimer, the
exonuclease excises the damaged region plus up to 100 adjacent nucleotides,
the polymerase fills in the gap using the intact opposite strand as a template
and the strand is resealed by the ligase (Kraemer and Slor 1985). The defect
in XP appears to be in the first step of this process -- the endonucleolytic
strand breakage. This was once thought to be due to a defect in the UV
endonuclease; however, the existence of 8 complementation groups suggests that
there are at least 8 different genes whose normal function is required for
this process. Indeed other information (presented in Paterson et al. 1984)
suggests that there are a variety of different defects in the various XP
cells. Cell extracts from Groups A, C and D excise dimers from UV-damaged
"naked" DNA with normal kinetics, yet group A cell extracts are deficient in
the removal of dimers from UV-damaged chromatin, possibly because they are
missing some factor which allows access to the DNA in order to effect repair
of thymine dimers. There is also data suggesting that some of complementation
groups A-H have mutations in regulatory genes mediating multiple repair
mechanisms. Thus, strains in groups A, B, C, D, G, and H have partial defects
in post replication repair and extracts of cells from groups A, B, C, D, and E
possess reduced photolygase activity which may jeopardize their ability to
perform the ligation step in DNA repair.
* -••- * DRAFT FINAL
-------
13-26
TABLE 13-8
GENETIC GROUPS OF XP PATIENTS
Complementation Group
Area
Japan
USA
Europe
Egypt
A
27
3
10
7
B
0
1
0
0
c
3
5
14
12
D
3
5
8
0
E
2
0
2
0
F
4
0
0
0
G
0
0
2
0
H
0
1
0
0
Variant
21
2
5
5
Compiled at the 16th International Congress of Dermatology, Tokyo,
May, 1982, with additional cases (Modhell et al. 1983, and personal
communication from Y. Fujiwara).
Source: Takebe et al. (1985).
DRAFT FINAL
-------
13-27
Paterson and his colleagues (Paterson et al. 1984), intrigued by the
paradoxical observation that one group of XP cells, group D, while appearing
totally unable to recognize dimer containing sites, still performed a
substantial amount of UV-induced unscheduled DNA synthesis (UDS). In
investigating this discrepancy, these workers gathered data which indicated
that during post-UV incubation of strains from groups A and D, the
phosphodiester bond between the two dimer-forming pyrimidines can be cleaved
leaving the DNA chains held together at the dimer site by a cyclobutane
bridge. In group A cells, DNA repair aborts at this stage, however, in Group
D cells there is an apparent abortive attempt to insert a normal DNA patch.
These processes are compared in Figure 13-3.
As indicated in the figure legend the proposed mechanism requires the
hypothesis that the first step in mammalian excision repair is the action of a
dimer phosphodiesterase rather than either the dimer glycosylase or the
exonuclease complex seen in bacterial systems. Confirmation that this process
also occurs in normal excision repair and is not just a characteristic of XP
cells was gained in subsequent studies (Gentner et al. 1984) and confirms the
observation by LaBelle and Linn (1982) that the putative UV endonucleolytic
activity of humans proceeds first by hydrolysis of the intradimer
phosphodiester bond.
The variant form of XP has a different defect -- cells from these
individuals lack a gene product necessary for accurate post replication repair
(Cleaver 1983). This is reflected in an increased sensitivity of XP variant
cells to caffeine inhibition of DNA synthesis after UV treatment (Kraemer and
Slor 1985).
ALBINISM
Albinism is the designation given to a variety of genetic disorders which
have in common a hypomelanosis derived from metabolic defects in the
melanocyte systems of the eye and skin. There are many different forms of
albinism; however, in all but one (the BADS syndrome), melanocytes are normal
but fail to synthesize adequate amounts of melanin (Witkop et al. 1983).
There are three broad classes of albinism: oculocutaneous -albinism (OCA)
which involves decreased pigmentation in hair, skin and eyes; ocular albinism
(OA); and oculocutaneous albinoidism. OCA exists in 10 distinct forms, OA in
two forms and oculocutaneous albinism in one. Table 13-9 gives a comparison
of the distinguishing pigmentary characteristics of the 10 forms of OCA.
As indicated in Table 13-10, the most common form of OCA in the U.S. is
the ty-pos type with Caucasians in the U.S. having an incidence of 1:37,000
and blacks an incidence of 1:15,000 (Witcop et al. 1983). Other populations
show much higher incidences of ty-pos albinism: 1:85-1:240 among South
Western Amerindian populations; 1:5,000 in Nigerians in Laos and 1:1,100 among
Ibos (Witcop 1981). It has been estimated that the incidence of all types of
albinism in the world population is slightly less than that seen in the U.S.:
1:20,000 while the Irish have been estimated to have a slightly greater
prevalence of all types -- 1:10,000. (Witcop et al. 1983).
DRAFT FINAL
-------
13-28
— 3'
OIMER PHOSPHOOIESTERASE
asia •
W'M'M"''
.OIMER ENOONUCLEASE
REPAIR PROCESS^
...3- ABORTS
DNA POLYMERASE
t
I
ONA LIGASE
t
1 i i i i i
K K >. r* K i>
-] P>J P-J P-J P-J P-J'
NORMAL
EXONUCLEASE
--31
ONA POLYMERASE
t
I
ONA LIGASE
t
— 3'
XP GROUP A
XP GROUP 0
FIGURE 13-3
MODEL OF THE NUCLEOTIDE MODE OF EXCISION REPAIR
IN NORMAL HUMAN CELLS (LEFT), XP GROUP A CELLS (MIDDLE),
AND XP GROUP D CELLS (RIGHT)
Source: Paterscn ec al. (1984).
* •• * DRAFT FINAL
-------
TABLE 1.3-9
COMPARISON OF THE PIGMENTARY CHARACTERISTICS OF HYPOMELANOTIC DISEASES WITH
FEATURES OF OCULOCUTANEOUS ALBINISM
Character-is tic
Ty-Neg
Ty-Pos
Ym
HPS
CHS
Hair color
Skin color
White throughout life
Pink to red
White, yellor-tan;
darkens with age
Pink-white to cream
White at birth; yellow-
red by 6 months
White at birth; cream,
slight tan on exposed
skin
White, red, brown
Cream-gray to light
normal
Blond to dark brown;
steel gray tint
Pink to pink-white
Pigraented nevi and Absent
freckles
Susceptibility to skin ++++
neoplas ia
Eye color
Gray to blue
Serum tyrosine levels Normal
-Melanocyte-simulating Normal
hormone levels
May be present and
numerous
Blue, yellow-brown;
age-arid race-dependent
Low normal to normal
Normal
Present
Unknown
Present
Present
Blue in infancy; darkens
with age
Blue-gray to brown; age- Blue to dark brown
and race-dependent
CO
i
ro
Melanosome in hair
bulbs
Incubation of hair
bulbs in tyros ine
Other
Stages I and II only
No pigmentation
lleterozygotes have less
than half normal
tyrosinase activity
To early stage III,
polyphagosomes
Pigmentation
^HOIl test suggests
heterogeneity in ty-pos
albinos
To stage III
polyphagosomes
None to questionable
increase
Hair bulb test shows
increased red or yellow
with tyrosine-cysteine
incubation.
To stage III,
polyphagosomes ,
pheomelanosomes
Pigmentation
Platelet defect; ceroid
storage; cytoplasmic
bodies in monocytes
Giant to normal
stage IV
Pigmentation
Susceptibility to
infection; giant
lysosomal-like
granules ;
lymphoreticular like
malignancy
Source: Adapted from Witkop et al. 1983.
-------
TABLE 13-9 (Continued)
COMPARISON OF THE PIGMENTARY CHARACTERISTICS OF HYPOMELANOTIC DISEASES WITH
FEATURES OF OCULOCUTANEOUS ALBINISM
Cross Syndrome
Brown OCA
Rufous OCA
Autosomal Dominant OCA
Black-Locks-Alb in ism-
Deafness Syndrome
White to light blond
Pink to pink-white
Beige to light brown in
Africans
Mahogany red to deep red
Cream to liglit tan on exposed Reddish brown
skia
White to cream with reddish Snow white with pigmented
tint locks
White to cream
White with melanized macules
Present
Unknown
Gray-blue
May be present
Similar to Caucasians in
Africa +
Hazel to light brown
May be present
Low
Reddish brown to brown
May be present
Unknown
Gray to blue
May be present in macular
areas
Unobserved, but probably
Gray-blue
i
u>
o
Normal
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Scanty; stage III;
some stage IV
Pigmentation
Stage I to stage III, some
lightly pigmented stage IV,
polyphagosomes
Pigmentation
Oligophrenia; microplithalmia; This defect recognized to
gingival fibromatosis; date only in Africans and
a tlietos is
New Guineans
Unknown
Pigmentation
Seen in New Guineans and
Africans
Stage I to early stage III;
no structural abnormality
Pigmentation. Increased
tyrosine activity in Golgi
Melanocytes present in
normal numbers
No melanocytes in white hair
and skin; normal melanocytes
and melanosomes in pigmented
hair and skin;
White hair — 0; pigmented
ha.ir — pigment increases
Profound sensorineural deaf-
ness; probably due to
failure of embryonic neural
elements to migrate from
crest to ear
Source: Adapted from Witkop et al. 1983.
-------
13-31
TABLE 13-10
ESTIMATES OF PREVALENCE OF TY-POS AND TY-NEG
ALBINOS IN THE GENERAL POPULATION OF THE
UNITED STATES BY RACE*
Albinism Prevalence
Population Ty-Neg Ty-Pos Combined
Caucasian 1:39,000 1:37,000 1:19,000
Afro-American 1:28,000 1:15,000 1:10,000
Total United States 1:37,000 1:31,000 1:16,800
* Corrected for 88 percent Caucasian and 12 percent
Negro, disregarding other racial components.
Source: Revised from Witkop (1983).
DRAFT FINAL
-------
13-32
Given the reduced or absent ability to form melanin in albinism it would
be expected that individuals with albinism would be particularly susceptible
to solar induced neoplasia. This is indeed the case for non-melanoma skin
cancer.
FINDINGS
Information presented above reviews the relationship of dysplastic nevus
syndrome to CMM, evidence relating ordinary acquired and congenital
melanocytic nevi to melanoma and the role of sunlight to the induction of
nevi. The following findings can be drawn from this information:
13.1 Dysplastic nevi are a risk factor for melanoma
independent of freckling and pigment.
13.2 The possession of melanocytic nevi (congenital or
acquired) also may be a risk factor.
13.3 Exposure to sunlight appears to encourage the
appearance and maybe disappearance of nevi and
dysplastic nevi from the skin.
13.7 Information from XP patients indicates that
individuals who have an inability to repair solar
radiation-induced DNA damage also have a high
incidence of melanoma relative to the normal
population.
13.8 The best characterized defect in XP patients is an
inability to excise pyrimidine dimers suggesting that
the repair of such lesions can be important to the
prevention of melanomas.
DRAFT FINAL
-------
13-33
REFERENCES
Armstrong, B.K., deKlerk, N.H. and Holman, C.D.J. Etiology of common acquired
melanocytic nevi: constitutional variables sun exposure and diet. JNCI
77:329-335 (1986).
Bale, S.J., Chakravarti, A. and Greene, M.H. Cutaneous malignant melanoma and
familial dysplastic nevi: evidence for autosomal dominance and pleiotropy.
Am. J. Hum. Genet. 38:188-196 (1986).
Caporaso, N., Greene, M.H., Tsai, S., Pickle, L. and Mulvihill, J.J.
Cytogenetics in familial malignant melanoma and dysplastic nevus syndrome --
Is dysplastic nevus syndrome a chromosomal instability disorder? Cancer Genet
Cytogenet in press (1986).
Clark, W.H. Jr., Elder, D.E., Guerry, D.,IV., Epstein, M.N., Green, M.H., and
Van Horn, M. A study in tumor progression: The precursor lesions of
superficial spreading and nodular melanoma Human Path 15:1147-1165 (1984).
Cleaver, J.E. Xeroderma pigmentosum. Chapter 57, In: The metabolic basis of
inherited disease Stanbury, J.B., Wyngaarden, J.B., Frederickson, D.S.,
Goldstein, J.L. and Brown, M.S. (eds). 5th edition New York: McGraw-Hill Book
Company pp 1227-1248 (1983).
Colella, C.M., Bogani, P., Agati, G. and Fusi, F. Genetic effects of UV-B:
Mutagenicity of 308 nm light in Chinese hamster V79 cells. Photochem
Photobiol 43:437-442 (1986).
Cooke, K.R., Spears, G.F.S., and Skegg, D.C.G. Frequency of moles in a defined
population. J Epidemiol Commun Hlth 39:48-52 (1985).
Elder, D.E., Greene, M.H., Bondi, E.E., Clark, W.H. Acquired melanocytic nevi
and melanoma. The dysplastic nevus syndrome. Chapter 11, In: Pathology of
malignant melanoma Ackerman, A.B. (ed). New York: Masson Publishing pp
185-215 (1981).
Elder, D.E., personal communication, 1986.
Elwood, J.M., Gallagher, R.P., Hill, G.B., Spinelli, J.J., Pearson, J.C.G.,
and Threlfall, W. Pigmentation and skin reaction to sun as risk factors for
cutaneous melanoma: Western Canada melanoma study. Br Med J 288:99-107 (1984).
Elwood, J.M., Williamson, C. and Stapleton, P.J. Malignant melanoma in
relation to moles, pigmentation, and exposure to fluorescent and other
lighting sources. Br J Cancer 53:65-74 (1986).
Centner, N.E., Weinfeld, M., Johnson, L.D. and Paterson, M.C. Incision of the
phosphodiester bond internal to the pyrimidine dimer-forming bases may occur
during excision repair of UV-induced damage in human fibroblasts. 15th Ann
Meet Environ Mut Soc (Abstr) p 113 (1984).
* DRAFT FINAL
-------
13-34
Green, A., MacLennan, R. and Siskind, V. Common acquired nevi and the risk of
malignant melanoma. Br J Cancer 35:297-300 (1985).
Greene, M.H. Dysplastic Nevus Syndrome. Hosp Prac:91-108, January (1984).
Greene, M.H., Clark, W.H. Jr., Tucker, M.A., Elder, D.E., Kraemer, K.H.,
Guerry, D., Witmer, W.K., Thompson!, Matozzo, I., and Fraser, M.C. Acquired
precursors of cutaneous malignant melanoma. The familial dysplastic nevus
syndrome. N Eng J Med 312:91-97 (1985a).
Greene, M.H., Clark, W.H.,Jr, Tucker, M.A., Kraemer, K.H., Elder, D.E., and
Fraser, M.C. High risk of malignant melanoma in melanoma-prone families with
dysplastic nevi. Annals Int Med 102:458-465 (1985b).
Greene, M.H., Goldin L.R., Clark, W.H., Jr., Lovrian, E., Kraemer, K.H.,
Tucker, M.A., Elder, D.E., Fraser, M.C. and Rowen, S. Familial cutaneous
malignant melanoma: autosomal dominant trait possibly linked to the Rh
locus. Proc Natl Acad Sci USA. 80:6071-6075 (1983).
Holman, D.J. and Armstrong, B.K. Pigmentary traits, ethnic origin, benign
nevi and family history as risk factors for cutaneous malignant melanoma. J
Natl Cancer Inst 72:257-266 (1984a).
Holman, C.D.J. and Armstrong, B.K. Cutaneous malignant melanoma and
indicators of total accumulated exposure to the sun: An analysis separating
histogenetic' types. J Natl Cancer Inst 73:75-81 (1984b).
Holman, C.D.J., Heenan, P.J., Caruso, V., Glancy, R.J. and Armstrong, B.K.
Seasonal variation in the functional component of pigmented nevi. Int J
Cancer 31:213-215 (1983).
Kopf, A.W., Lazar, M., Bart, R.S., Dubin, N. and Bromberg, J. Prevalence of
nevocytic nevi on lateral and medial aspects of arms. J Dermatol Surg Oncol
4:153-158 (1978).
Kopf, A.W., Lindsay, A.C., Rogers, G.S., Friedman, R.J., Rigel, D.S., and
Levenstein, M. Relationship of nevocytic nevi to sun exposure in dysplastic
nevus syndrome. J Amer Acad Derm 12:656-662 (1985).
Kraemer, K.H., and Slor, H. Xeroderma pigmentosum. Clinics in Dermatology
2:33-69 (1984).
Kraemer, K.H., and Greene, M.H. Dysplastic nevus syndrome - Familial and
sporadic precursors to cutaneous melanoma Derm Clin 3:225-237 (1985).
Kraemer, K.H., Greene, M.H., Tarone, R., Elder, D.E., Clark, W.H., Jr, and
Guerry, D. Dysplastic nevi and cutaneous melanoma risk. Lancet 2:1067-8
(1983).
* * * DRAFT FINAL * *
-------
13-35
Kraemer, K.H., Lee, M.M., and Scotto, J. DNA repair protects against
cutaneous and internal neoplasia: Evidence from xeroderma pigmentosum.
Carcinogenesis 5:511-514 (1984).
LaBelle, M. and Linn, S. In vivo excision of pyrimidine dimers is mediated by
a DNA N-glycosylase in Micrococcus Luteus but not in human fibroblasts.
Photochem Photobiol 36:319-324 (1982).
Lever, W.F. and Schaumburg-Lever, G. Melanocytic nevi and malignant melanoma,
Chapter 33, In: Histopathology of the Skin. Philidelphia: J.B. Lippincott
Company pp 681-687 (1979).
Lynch, H.T., Anderson, D.E., Smith, J.L., Howell, J.B., and Brush, A.J.
Xeroderma pigmentosum, malignant melanoma and congenital ichthyosis. Arch Derm
96:625-635 (1967).
Mackie, R.M., English, J., Aitchison, T.C., Fitzsimons, C.P. and Wilson, P.
The number and distribution of benign pigmented moles (melanocytic naevi) in a
healthy British population. Brit J Derm 113:167-174 (1985).
Mark, G.J., Mihm, M.G., Liteplo, M.A., Reed, R.J. and Clark, W.H. Congenital
melanocytic nevi of the small and garment type. Clinical, histologic and
ultrastructural studies. Hum Pathol 4:395-418 (1973).
Nicholls, E.M.. Development and elimination of pigmented moles and the
anatomical distribution of primary malignant melanoma. Cancer 32:191-195
(1973).
Pack, G.T., Lenson, N. and Gerber, D.M. Regional distribution of moles and
melanomas. Arch Surg 65:862-70 (1952).
Paterson, M.C., Centner, N.E., Middlestadt, M.V. and Weinfeld, M. Cancer
predisposition, carcinogen hypersensitivity and aberrant DNA metabolism. J
Cell Phys (suppl) 3:45-62 (1984).
Pehamberger, H., Honigsmann H., Wolff, K. Dysplastic nevus syndrome with
multiple primary amelanotic melanomas in oculocutaneous albinism. J Amer Acad
Derm 11:731-735 (1984).
Perera, M.I.R., Kyung, I.U., Greene, M.H., Waters, H.L., Bredberg, A. and
Kraemer, K.H. Hereditary dysplastic nevus syndrome: Lymphoid cell ultraviolet
hypermutability in association with increased melanoma susceptibility. Cancer
Res 46:1005-1009 (1986).
Rhodes, A.R. Pigmented birthmarks and precursor melanocytic lesions of
cutaneous melanoma identifiable in childhood. Ped Clinics North America
30(3):435-463 (1983).
Rhodes, A.R. and Melski, J.W. Small congenital nevocellular nevi and the risk
of cutaneous melanoma. J Pediatr 100:219-224 (1982).
* * * DRAFT FINAL * * *
-------
13-36
Rhodes, A.R., Melski, J.W., Sober, A.J., Harrist, T.J., Mihm, M.C.,
Fitzpatrick, T.B. Increased intraepidermal melanocyte frequency and size in
dysplastic melanocyte nevi and cutaneous melanoma: a comparative quantitative
study of dysplastic melanocytic nevi, superficial spreading melanoma,
nevocellular nevi, and solar lentigines. J Invest Dermatol 80(5):452-459
(1983).
Rhodes, A.R., Silverman, R.A., Harrist, T.J., and Melski, J.W. A histologic
comparison of congenital and acquired nevomeloanocytic nevi. Arch Dermatol
121:1266-1273 (1985).
Richmond, A., Fine R, Murray D, Lawson, D.H., and Priest, J.H. Growth factor
and cytogenetic abnormalities in cultured nevi and malignant melanomas. J
Invest Dermatol 86:295-302 (1986).
Robbins, J.H., Kraemer, K.H., Lutzner, M.A., Festoff, B.W., and Coon, H.G.
Xeroderma pigmentosum. An inherited disease with sun sensitivity, multiple
cutaneous neoplasms, and abnormal DNA repair. Ann Intern Med 80:221-248
(1974).
Smith, P.J., Greene, M.H., Devlin, D.A., McKeen E.A. and Paterson, M.C.
Abnormal sensitivity to UV-radiation in cultured skin fibroblasts with
hereditary dysplastic nevus syndrome. Int J Cancer 30:39-45 (1982).
Rhodes, A.R., Melski, J.W., Sober, A.J., Harrist, T.J., Mihm, M.C.,
Fitzpatrick, T.B. Increased intraepidermal melanocyte frequency and size in
dysplastic melanocyte nevi and cutaneous melanoma. A comparative quantitative
study of dysplastic melanocytic nevi, superficial spreading melanoma,
.nevocellular nevi, and solar lentigines. The J Invest Dermatol 80(5):452-459
(1983).
Smith, P.J., Greene M.H., Adams D., and Paterson M.C. Abnormal response to
the carcinogen 4-nitroquinoline 1-oxide of cultured fibroblases from patients
with dysplastic nevus syndrome and hereditary cutaneous malignant melanoma.
Carcinogenesis 4:911-917 (1983).
Smith, P.J., Greene, M.H., Devlin, D.A., McKeen, E.A., and Paterson, M.C.
Abnormal sensitivity to UV-radiation in cultured skin fibroblasts from
patients with hereditary cutaneous malignant melanoma and dysplastic nevus
syndrome. Int J Cancer 30:39-45 (1982).
Smith, P.J. and Paterson M.C. Abnormal responses to mid-ultraviolet light of
cultured fibroblasts from patients with disorders featuring sunlight
sensitivity. Cane Res 41:511-518 (1981).
Swerdlow, A.J., English, J., Mackie, R.M., O'Doherty, C.J., Hunter, J.A.A.,
and Clark, J. Benign naevi associated with high risk of melanoma. Lancet
2:168 (1984).
* * * DRAFT FINAL * »
-------
13-37
Takebe, H., Tatsumi, K., and Satoh, Y. DNA repair and its possible
involvement in the origin of multiple cancer. J Clin Oncol (suppl 1)
15:299-305 (1985).
Tyrell, R.M. Mutagenic action of monochromatic UV-radiation in the solar
range on human cells. Mutat Res 129:103-110 (1984).
Witcop, C.J., Jr. Epidemiology of skin cancer in mongenetic factors In:
Biology of skin cancer Laerum, D., Iverson, O.H.(eds). Geneva International
Union Against Cancer Technical Report Series 60:58-86 (1981).
Witkop, C.J.,Jr., Quevedo, W.C., Jr., and Fitzpatrick, T.B. Albinism and
other disorders of pigment metabolism Chapter 15, In: The Metabolic Basis of
Inherited Disease Stanbury, J.B., Wyngaarden, J.B., Frederickson, D.S.,
Goldstein, J.L. and Brown, M.S. (eds). 5th edition New York: McGraw-Hill Book
Company pp 301-346 (1983).
* * * DRAFT FINAL * * *
-------
CHAPTER 14
A COMPARISON OF. MELANOMA AND NONMELANOMA SKIN TUMORS
A comparison of information on cutaneous malignant melanoma (CMM) and
nonmelanoma skin tumors may provide insight into the role of solar radiation
and other etiological risk factors in the development of different types of
skin cancer. Prolonged sun exposure is considered to be the dominant risk
factor for nonmelanoma skin tumors. Through an examination of the
similarities and dissimilarities between CMM and nonmelanoma skin tumors, the
potential role of solar radiation in the development of CMM can be evaluated.
The study of nonmelanoma skin tumors (i.e., basal and squamous cell
carcinomas) is difficult because most patients are seen and treated in offices
and are rarely hospitalized. In addition, the study of nonmelanoma skin
tumors is made more difficult by the perception that they are relatively
trivial problems because most are successfully treated (e.g. ,' by surgical
excision). Several population-based nonmelanoma skin tumor data bases have
however, been developed although they generally require special surveys for
data collection. This chapter will briefly review the studies that have
specifically addressed the differences and similarities between CMM and
nonmelanoma skin tumors and other relevant epidemiological information.
NONMELANOMA SKIN TUMORS
There are two major forms of nonmelanoma skin tumors: basal cell
carcinoma (BCC) and squamous cell carcinoma (SCC). Although most skin cancer
statistics combine the two cell types, the limited data available from the
U.S. and several other countries indicate differences between these tumors
with respect to a number of characteristics.
Basal cell carcinomas are neoplasms of the germinal layers of the
epidermis and the appendages which differentiate toward glandular structures
(Scotto and Fraumeni 1982). As a rule these tumors are slow growing and
follow a relatively benign course, although on rare occasions they may result
in extreme morbidity, mutilation, or if they metastasize, death (Pollack et
al. 1982).
Basal cell carcinomas are believed to arise from a pluripotent epithelial
cell present in the epidermis. It has been hypothesized that basal cell
carcinomas arise because an abnormal interaction between these pluripotent
stem cells and the surrounding connective tissue induces the cells to
differentiate neoplastically (Pollack et al. 1982; Kent 1976). These tumors
appear to be very stromal-dependent, however, and it has been suggested that
they will rarely metastasize to a foreign tissue bed unless they take along a
portion of their stroma (Pinkus 1953). This hypothesis has been confirmed in
part by studies which have shown that basal cell tumor cells cultured in
vitro in the absence of accompanying connective tissue convert to a
keratinizing epithelium (Flaxman 1972; Kubilus et al. 1980). It has also been
* * * DRAFT FINAL * * *
-------
14-2
shown that autotransplants fail if the cells are transferred without stroma
(Epstein et al. 1984).
Squamous cell carcinomas are neoplasms of the epidermis that differentiate
towards keratin formation (Scotto and Fraumeni 1982). SCCs are far less
fastidious in their growth requirements In vitro and will grow under a
variety of conditions. They are generally more aggressive than BCCs and
account for about three-fourths to four-fifths of the deaths attributable to
rionmelanoma skin cancer (Dunn et al. 1965).
As described in detail in Chapter 2, melanomas are neoplasms of pigment
forming cells that are derived from the embryonic neural crest. Although
melanomas are far less common than BCCs or SCCs, they are generally more
lethal. As a result, the raw mortality figures for nonmelanoraa are similar to
those for CMM (Scotto and Fraumeni 1982).
INCIDENCE AND AGE
Nonmelanoma skin tumors are among the most common malignant neoplasms
occurring in white populations. Based on a one-year survey (1977-1978)
conducted by the National Cancer Institute (NCI), it was determined that
nonmelanoma skin cancers developed in approximately 400,000 white Americans
each year (Scotto and Fraumeni 1982). The annual age-adjusted incidence rate
for this survey period was estimated to be 232.6/105 among whites. For
comparison, the estimated incidence rate for all other cancers among whites in
the United States based on. 1973-1976 data from the Surveillance, Epidemiology
and End Results (SEER) Program was 318.9/105. Among blacks, the annual
age-adjusted incidence rates for nonmelanoma skin tumors and all other cancers
were 3.4/105 and 347.3/105, respectively.
The incidence of BCC is generally several times greater than the incidence
of SCC. Similarly, the incidence of SCC exceeds that of CMM (Lee 1982;
Epstein et al; 1984, Eastcott 1963). For example, based on data collected for
62 skin cancer cases registered from 1956-1960 in three public hospitals in
New Zealand, Eascott (1963) observed that 73 percent of the cases had BCC, 15
percent had SCC, and 7 percent had CMM. Lee (1982) presented data on 2,019
skin cancer cases in Switzerland for 1974-1978 and showed that approximately
69 percent of the cases had BCC, 20 percent had SCC, and 11 percent had CMM.
These relative differences in BCC, SCC and CMM persist when males and
females are examined separately. Based on a comparison of BCC and SCC
incidence data from Scotto and Fraumeni (1982) and CMM incidence data from NCI
(1985) for white American males and females, the incidence of BCC was
approximately four to six times greater than the incidence of SCC which in
turn was approximately three to seven times greater than the incidence of CMM
(see Table 3-1).
As has been described in earlier chapters, the incidence of CMM has been
increasing over the past several decades. Lee (1982) has estimated that the
incidence of BCC as well as SCC is increasing as fast as or faster than the
incidence of CMM. Epstein et al. (1984) has pointed out that the rate of
increase in SCC is greater than that in BCC. Scotto and Fraumeni (1982),
- * * DRAFT FINAL * * *
-------
14-3
however, noted when comparing the NCI 1971-1972 and 1977-1978 survey data that
the observed incidence increases applied mainly to BCC. The authors noted
that the incidence of BCC among United States whites increased by
approximately 15-20 percent over the six-year period between surveys.
In general, older individuals (e.g., over 60 years of age) are at higher
risk of developing nonmelanoma skin tumors than are younger individuals.
Vitaliano and Urbach (1980) observed that only three percent of a total 424
BCC and SCC cases from the Tumor Clinic of the Skin and Cancer Hospital in
Philadelphia were under 40 years of age. However, Emmett (1982) and Harris
(1982) have observed that SCC and BCC are no longer only diseases of old age
since an increasing number of younger individuals have been presenting with
nonmelanoma skin tumors. Harris (1982) suggested that the occurrence of
nonmelanoma skin tumors among younger individuals was consistent with'
increased sunlight exposure among these age groups.
The age-specific incidence patterns for SCC and BCC are not completely
identical, suggesting that different etiological mechanisms may exist.
Although incidence rates for SCC and BCC have been reported to rise with age
and level off at the oldest age groups (Scotto and Fraumeni 1982), the
increase with age was sharper for SCC than BCC. Laerum and Iversen (1981)
summarized study results indicating that among a group of BCC cases, 15 •
percent were less than 50 years of age and 65 percent were less than 70 years
of age. Among a group of SCC cases, in contrast, Laerum and Iversen (1981)
observed that. 70 percent were over 70 years of age.
ANATOMICAL DISTRIBUTION
The anatomical distribution of nonmelanoma skin tumors differs from that
of melanomas. Nonmelanoma skin tumors predominantly occur on regularly
exposed sites whereas melanomas occur on both regularly exposed and nonexposed
sites depending upon the histologic type and sex of the individual. Scotto
and Fraumeni (1982) noted that the tendency for nonmelanoma skin tumors to
develop in exposed areas was consistent with the belief that exposure to solar
radiation is a dominant risk factor. As already described in Chapter 5, the
observed anatomical distribution of melanomas has also been attributed in part
to patterns of sunlight exposure.
Table 14-1 compares the distribution of CMM with the distributions of BCC
and SCC by sex for tumors occurring in whites in the United States (Scotto and
Fraumeni 1982). The data for SCC and BCC were collected as part of the
1977-1978 NCI survey whereas the CMM data were collected from 1973-1976
(Scotto and Nam 1980). The predominant sites for both SCC and BCC and for
both males and females were the face, head and neck. These areas accounted
for 60 percent or more of the total nonmelanoma skin tumors. Most of the
remaining BCCs occurred on the trunk whereas most of the remaining SCCs
occurred on the upper extremities. This distribution of tumors differs from
that for CMM which follows a more even distribution among sites and occurs
predominantly on the trunk among males and the lower extremities among
females. A comparison of 1974-1978 skin tumor data from Switzerland in Table
14-2 shows similar tumor site distributions (Lee 1982).
* * DRAFT FINAL * * *
-------
14-4
TABLE 14-1
PERCENTAGE OF TUMORS BY ANATOMIC SITE FOR
NONMELANOMA SKIN CANCER AND MELANOMA
AMONG WHITE MALES AND FEMALES IN THE UNITED STATES
a
BCC
Anatomic Site Male Female
Face, Head and Neck 81.2 84.1
Trunk 12.0 8.9
Upper Extremities 4.9 3.4
Lower Extremities 1.3 2.9
Other Sites 0.5 0.7
a b
SCC Melanoma
Male Female Male Female
74.8 60.1 27.0 17.0
4.5 5.3 38.0 22.0
18.1 25.8 22.0 26.0
1.3 5.7 13.0 35.0
1.4 3.2 NA NA
a • ' .
Nonmelanoma skin tumor data were for 1977-1978.
b
Based on Scotto and Nam (1980) as cited in Scotto and Fraumeni (1982)
Source: Scotto and Fraumeni (1982).
* * * DRAFT FINAL * *
-------
14-5
TABLE 14-2
DISTRIBUTION BY SEX AND ANATOMIC SITE OF SKIN TUMORS:
CANTON OF VAUD, SWITZERLAND 1974-1978
BCC
Anatomic Site
Head and Neck
Trunk
Upper Limbs
Lower Limbs
Other
Total
Male
540
(71.2)
130
(17.1)
14
(1.8)
15
(2.1)
58 '
(7.6)
758
(100.0)
Female
505
(78.7)
76
(11.9)
17
(2.6)
18
(2.8)
26
(4.0)
642
(100.0)
sec
Male
202
(80.5)
8
(3.2)
26
(10.3)
5
(2.0)
10
(4.0)
251
(100.0)
Female
103 .
(72.5)
11
(7.7)
19
(13.4)
7
(4.9)
2
(1.4)
142
(100.0)
Melanoma
Male
27
(28.7)
34
(36.2)
14
(14.9)
11
(11-7)
8
(8.5)
94
(100.0)
Female
' 22
(16.7)
24
(18.2)
27
(20.4)
51
(38.6)
8
(6.1)
132
(100.0)
Source: Levi and Chapallaz (1981) as cited in Lee (1982).
* * DRAFT FINAL
-------
14-6
There are some differences in the observed distributions of SCC and BCC
which suggest that these nonmelanoma skin tumors may respond differently to
different dosages of sun exposure. The available epidemiological data
generally show that BCC is more likely to develop on regularly unexposed sites
compared with SCC. Pollack et al. (1982) noted that although it is
well-established that sun-exposed areas are more prone to BCC, in at least one
study (Urbach et al. 1972), approximately one-third of the BCCs occurred in
light-protected regions. Laerum and Iversen (1981) observed that among a
group of SCC and BCC cases, 90 percent of the BCCs and approximately 50
percent of the SCCs occurred on the head and neck. They estimated that of
these tumors, two-thirds of the BCCs occurred on sites of the head and neck
receiving the highest UV radiation doses (e.g., the nose) whereas all the SCCs
occurred at these sites. Hilstrom and Swanbeck (1970) examined only SCC cases
and observed that about 80 percent of the SCCs occurred on the head. 'Facial
SCCs occurred relatively equally between males and females, but of the SCCs.on
the external ear, 90 percent occurred among males and only 10 percent among
females. Emmett (1982) examined only BCC cases and observed that 75.5 percent
occurred on the head and neck, 16 percent on the limbs, and 8.4 percent on the
trunk. Lee (1982) noted that although BCC and SCC tend to be more
concentrated on exposed sites than superficial spreading melanoma (SSM) or
nodular melanoma (NM), the distribution of BCC (somewhat similar to CMM) did
not precisely correspond with sun-exposed areas.
RISK FACTORS
Several features common to both CMM and nonmelanoma skin tumors have been
identified in epidemiological studies including susceptibility to burn, a
latitudinal gradient in incidence, and predisposing host factors such as
red/blond hair and blue/light eyes. The applicability of these risk factors,
all of which are somehow related to sunlight exposure, to both nonmelanoma
skin tumors and CMM suggests that solar radiation is indeed involved in the
development of both types of skin cancer. However, differences between CMM
and nonmelanoma skin tumors, for example with respect to anatomical
distribution and age-specific patterns of incidence, also suggest that a more
complex set of risk factors may be involved in the development of CMM that
nonmelanoma skin tumors. In this section, features believed to be
associated with these different skin tumors and the hypotheses that have been
forwarded to explain these findings will be reviewed.
As already noted, prolonged sun exposure is considered to be the dominant
risk factor for nonmelanoma skin tumors (NRC 1982; Scotto and Fraumeni 1982;
Greene and O'Rourke 1985; Lee 1982; Beral and Robinson 1981). Scotto and
Fraumeni (1982) and others (Scotto et al. 1981; Laerum and Iversen 1982;
Emmett 1982) have cited several observations supporting this hypothesis:
• the tendency for nonmelanoma skin tumors to develop in
sun-exposed sites,
• the higher incidence rates among occupational groups
with outdoor exposures compared to those with indoor
exposures,
- - - DRAFT FINAL - * *
-------
14-7
• the latitudinal and UV radiation gradient showing
highest incidence rates in geographic areas of
relatively high UV radiation exposure,
• the inverse correlation between nonmelanoma skin tumor
incidence and degree of skin pigmentation,
• the high risk among genetically predisposed
individuals (e.g., those with xeroderma pigmentosura),
• the predisposition for nonmelanoma skin tumors to
develop among light-skinned individuals who are
susceptible to sunburn and who have red/blond hair,
blue/light eyes, and a Celtic heritage, and
• the capacity of UV radiation to induce nonmelanoma
skin tumors in experimental animals.
The observed variations in nonmelanoma incidence by latitude in particular
support an association with sun exposure. For example, Scotto and Fraumeni
(1982) observed, based on the 1977-1978 NCI survey data, that the incidence of
SCC and BCC in the United States showed a latitudinal gradient with higher
rates in the south. These results are displayed in Figures 14-1 and 14-2 for
white males and white females, respectively. Also shown in these figures are
1973-1976 CMM data from the SEER program. The latitudinal gradients for CMM
were least pronounced. However, only a small number of data points (as shown
in Figures 14-1 and 14-2) were used to examined the latitudinal gradients of
CMM, BCC, and SCC.
Prolonged sun exposure may also be a risk factor for specific types of
CMM. Greene and O'Rourke (1985) noted that chronic sun exposure has been
implicated in the causation of nonmelanoma skin tumors, but that a different
disease-exposure relationship may predominate for CMM, which often occurs on
irregularly exposed areas. The authors hypothesized that a positive
association between CMM and nonmelanoma skin tumors would suggest an etiologic
role for cumulative sun exposure in CMM.
To test this hypothesis, Greene and O'Rourke (1985) conducted a
case-control study of 232 CMM patients and 232 age-, sex-, and
residence-matched controls in Queensland. The case data were collected from
patients presenting with a first primary cutaneous melanoma between 1 July
1979 and 30 June 1980. A random sample of cases was drawn from a population
of 871 potential cases and was stratified by geographic location of residence
in order to ensure proportional representation from the less densely populated
interior part of the state. The controls were randomly selected from the
Queensland electoral rolls. Case-control information included history of
lifetime sun exposure (based on all outdoor occupations held for more than 6
months and all recreations ever pursued on a regular basis after 10 years of
age), complexion, hair color, propensity to sunburn, social class, country of
birth, and ethnicity. The face and left forearm were also examined for
actinic tumors (i.e., solar keratosis, BCC and SCC).
* * * DRAFT FINAL * * *
-------
14-8
500 -
300
200
100 -
<
hU
>
14O ISO
MOM ULTRA VIOLET IUV.«I RADIATION INOCX
FIGURE 14-1
ANNUAL AGE-ADJUSTED INCIDENCE RATES FOR BASAL AND
SQUAMOUS CELL CARCINOMAS (1977-78)
AND MELANOMA (SEER DATA, 1973-76) AMONG WHITE MALES*
* According to annual UVB measurements at selected areas of the
United States, with regression lines based on exponential model.
Source: Scotto and Fraumeni (1982).
* * DRAFT FINAL
-------
14-9
GOO
SOO
too
300
-------
14-10
As shown in Table 14-3, Greene and O'Rourke (1985) observed a greater
percentage of actinic tumors on the face and forearms of CMM cases than of
controls after adjustment for age and the presence of nevi. The results
remained basically unchanged after adjustment for social class, ethnicity, and
migrant status. The major risk factor appeared to be the presence of any
nonmelanoma tumor since there was no trend of increasing risk with increasing
number of tumors. The relative risk (RR) associated with the presence of
actinic tumors was highest among CMM patients with lentiginous malignant
melanoma (LMM) (RR=5.2, 95% C.I. = 1.7-15.8). Lee (1982) has, in fact,
suggested that the pathogenesis of LMM is closely analogous to that of the
nonmelanoma skin tumor SCC.
Cumulative hours of sun exposure were also higher among cases than
controls. The relative risks of CMM after adjustment for age, presence of
nevi, hair color, and sunburn propensity were 3.2 (95% C.I. = 0.9-12.4) for
2,000-50,000 hours exposure and 5.3 (95% C.I. = 0.9-30.8) for 50,000 or more
hours of exposure. Among the control population alone, total hours of sun
exposure and hair color were also associated with risk of actinic tumors.
Based on the established link between chronic sun exposure and actinic
tumors, and the results showing an increased occurrence of actinic tumors
among CMM patients, Greene and O'Rourke (1985) concluded that large cumulative
UV radiation exposures were associated with an increased risk of CMM.
Vitaliano and Urbach (1980) examined several risk factors for SCC and BCC
in a case-control study. The study included 366 BCC and 58 SCC cases seen at
the Tumor Clinic of the Skin and Cancer Hospital in Philadelphia (dates were
not specified). A group of 294 white controls without carcinoma were selected
from the skin and cancer outpatient department. Information compiled on each
case and control included cumulative solar exposure based on vocational and
military history, time spent sunbathing, and participation in outdoor sports
(as a spectator or participant). Exposure was divided into four categories:
£1=0-9,999 hours, E2=10,000-19,999 hours, £3=20,000-29,999 hours, and
£4=30,000 or more hours. The host factors that were examined included
complexion (pale or mild-dark), age (0-59 years or 60 and over), and ability
to tan (tans or burn-sensitive).
Based on a logistic regression of the case-control data, Vitaliano and
Urbach (1980) identified the most important risk factors for BCC and SCC as
follows:
BCC: cumulative solar exposure (p<0.001) » ability to
tan (p<0.001) » age (p<0.005) » complexion
(p<0.025)
SCC: cumulative solar exposure (p<0.001) » age
(p<0.001) » ability to tan (p<0.005).
Cumulative solar exposure was the most important risk factor for both SCC and
BCC. Ability to tan was also important even at low levels of exposure.
Complexion was a less important risk factor for SCC than for BCC.
* * * DRAFT FINAL
-------
14-11
TABLE 14-3
DISTRIBUTION OF 232 CASES OF CUTANEOUS MALIGNANT MELANOMA
AND 232 CONTROLS ACCORDING TO PRESENCE OF ACTINIC TUMORS;
ESTIMATED RISK IN RELATION TO PRESENCE OF ACTINIC TUMORS
ON THE FACE, AFTER ADJUSTING FOR PRESENCE OF NEVI
ON THE ARMS AND'FOR AGE
Class of
Melanoma (No.)
All classes (232)
LMM (49)
SSM, nodular,
IND (183)
Site of
Actinic
Tumors
Face
Arm
Face
Arm
Face
Arm
No.
of
Cases
(01
v°.
95
109
31
34
63
75
)
(41)
(47)
(63)
(69)
(34)
(41)
No.
of
Controls
f/
I/O.
34
52
13
19
21
33
)
(15)
(22)
(27)
(39)
(11)
(18)
RR* (95%
Confidence
Interval)
3.6 (1.8-7.3)
5.2 (1.7-15.8)
2.8 (1.1-7.2)
* RR = relative risk: 1.0 is taken as a base-line category
representing absence of actinic tumors on the face.
Source: Greene and O'Rourke (1985).
DRAFT FINAL » * »
-------
14-12
Table 14-5 presents the estimated relative risks (RRs) of BCC and SCC for
the combinations of risk factors considered in the Vitaliano and Urbach (1980)
study. Vitaliano and Urbach (1980) concluded that the most important
difference between SCC and BCC was their relationship with cumulative
exposure. As shown in Table 14-5, a higher exposure level was required for
BCC than for SCC to reach similar RRs. The authors noted that the maximum
response of BCC to solar exposure occurred in exposure category E4 (30,000 or
more hours) whereas for SCC it occurred in exposure category E3 (20,000-29,999
hours). They observed that the results.were consistent with the belief that
exposure to UV radiation has a greater effect on the development of SCC than
on BCC although an association between BCC and sunlight does exist.
Variations in the incidence ratio of BCC to SCC by latitude again suggest
that the two forms of nonmelanoma skin tumors respond differently to 'solar
exposure. The results of MacDonald and Bubendorf (1964, as cited in Vitaliano
and Urbach 1980) showed that the BCC/SCC ratio decreased from approximately
ten-to-one in northern United States cities to approximately two- or
three-to-one in southern rural areas. Similarly, Scotto and Fraumeni (1982)
noted that the ratio of SCC incidence to BCC incidence increased with
decreasing latitude and increasing sunlight exposure (see Figures 14-1 and
14-2). Pollack et. al. (1982) commented that dosimetry studies revealed a
poor correlation between BCC density in a site and UV radiation dose. They
suggested that etiological factors for BCC other than UV radiation, such as
the presence of areas of scarring and epidermal nevi, may exist.
Beral and Robinson (1981) examined the similarities and dissimilarities
between the anatomic distributions of CMM, BCC, and SCC to determine whether
different sun exposure patterns were likely to be involved in different types
of skin tumors. As described in Chapter 12, Beral and Robinson (1981)
observed that BCC, SCC and melanomas of the head, face, and neck had similar
distributions among outdoor workers suggesting that prolonged exposure to
sunlight was important in the etiology of melanomas of exposed parts of the
body as well as BCC and SCC. The results also suggested that prolonged sun
exposure was not involved in the etiology of melanomas of unexposed parts of
the body (e.g., trunk and limbs) among office workers.
FINDINGS
The observed similarities between CMM and nonmelanoma skin tumors strongly
suggests that exposure to solar radiation, known to be the predominant risk
factor for nonmelanoma skin tumors, is also involved in the etiology of CMM.
However, the available epidemiological evidence indicates that these tumors
may respond somewhat differently to different etiological factors. The
following findings regarding the relationships of CMM and nonmelanoma skin
tumors to sunlight exposure and other risk factors can be drawn:
16.1 Prolonged sun exposure is considered to be the dominant risk fator
for nonmelanoma skin tumors. Similarities in the observed
patterns of nonmelanoraa skin tumors and CMM suggest that prolonged
sun exposure may also play a role in the development of CMM.
These similarities include an elevated risk among light-skinned
- * DRAFT FINAL *
-------
14-13
TABLE 14-5
ESTIMATED RELATIVE RISKS OF BASAL AND SQUAMOUS
CELL CARCINOMA FOR 32 COMBINATIONS OF FACTORS
Tans
Burns-Sensitive
Exposure Grade
(Total Exposure, hr)
0-59 yrs
60+ yrs
0-59 yrs
60+ yrs
Dark Pale Dark
Pale Dark
Pale Dark
Pale
Basal Cell Carcinoma
E4 (30,000 or more)
E3 (20,000-29,999)
E2 (10,000-19,999)
El (0-9,999)
3.19 4.94
2.86 4.43
1.77 2.75
1.00 1.55
4.99
4.49
2.79
1.57
7.76
6.95
4.32
2.43
6.10
5.47
3.39
1.91
9.43
8.47
5.26
2.96
9.57
8.58
5.32
3.00
Squamous Cell Carcinoma
Source: Vitaliano and Urbach (1980).
14.80
13.29
8.25
4.65
E4
E3
E2
El
7.
7.
4.
1.
09
09
42
92
22.79
22.79
5.72
1.00
28.
28.
17.
7.
61
61
94
76
90.12
90.12
23.08
4.03-
26.61
26.61
16.60
7.19
84.66
84.66
21.41
3.74
107
107
66
29
.70
.17
.99
.19
347 . 08
347.08
86.52
15.06
DRAFT FINAL
-------
14-14
individuals who have a susceptibility to sunburn, blue/green
eyes, red/blond hair, and a Celtic heritage and among
predisposed susceptible individuals (e.g., xeroderma pigmentosum
patients). A latitudinal gradient and notable incidence
increases over the past several decades have also been observed
for both CMM and nonmelanoma skin tumors.
16.2 Differences between nonmelanoma skin tumors and CMM indicate,
however, that prolonged sun exposure is only one of a complex
set of risk factors that may be involved in the etiology of
CMM. The differences between CMM and nonmelanoma skin tumors
include their overall anatomical distributions by sex, the
concentrations of CMM on usually unexposed sites and
nonmelanomas on usually exposed sites, and the potential
importance of nevi in the development of CMM.
* * * DRAFT FINAL
-------
14-15
REFERENCES
Beral, V. and Robinson, N. The relationship of malignant melanoma, basal and
squamous skin cancers to indoor and outdoor work. Br. J. Cancer 44:886-891
(1981).
Dunn, J.E. Jr., Levin, C.A., Linden G. et al. Skin cancer as a cause of
death. Calif Med 102: 361-363 (1965).
Eastcott D.F. Epidemiology of Skin Cancer in New Zealand NCI Monograph No.
10: 141-151. (1963).
Emmett, A.J.J. Basal cell carcinoma. Chapter 4. Malignant Skin TumOrs.
Emmett, A.J. and O'Rourke, M.G.E., Eds. Churchill Livingston, New York (1982),
Epstein, W.L., Bystryn, J.C., Edelson, R., Elias, P.M., Lowy, D.R. and Yuspa
S. Nonmelanoma skin cancer melanomas, warts and viral oncogenesis. J Amer
Acad Dermatol 5(2)=960-970 (1984).
Flaxman, B.A. Growth in vitro and induction of differentiation in cells of
basal cell cancer. Cancer Res 32:462-469 (1972).
Green, A.C. and O'Rourke, M.G.E. Cutaneous malignant melanoma in association
with other skin cancers. J.N.C.I. 74:977-980 (1985).
Green, A.C., and O'Rourke, M.G.E. Cutaneous malignant melanoma in association
with other skin cancers. JNCI 74(5) .-977-980 (1985).
Harris, T.J. Squamous cell carcinoma. Chapter 5. Malignant Skin Tumors.
Emmett, A.J. and O'Rourke, M.G.E., Eds. Churchill Livingston, New York (1982)
Kent, A. Pathology of basal cell carcinoma in Andrade R, Gumport SL, Popkin
GL et. (eds) Cancer of the Skin Philadelphia. W.B. Sanders Co. pp. 845-882
(1976).
Kubilus, J, Baden H.P. and McGilvray N. Filamentous protein of basal cell
epithelioma: Characteristics in vivo and in vitro. J Natl Cancer Inst
65:869-875 (1980).
Laerum, O.D. and Iversen, O.K., Eds. Biology of Skin Cancer (Excluding
Melanomas). A series of workshops on the biology of human cancer. Report No.
15. International Union Against Cancer. UICC Technical Report Series.
Volume 63. Geneva (1981).
Lee, J.H. Melanoma and Exposure to Sunlight Epi Rev 4:110-136. (1982).
National Cancer Institute (NCI). 1985 Annual Cancer Statistics Review.
Presented to National Cancer Advisory Board. National Institutes of Health.
Bethesda, Maryland (1985).
DRAFT FINAL
-------
14-16
National Research Council (NRG). Causes and Effects of Stratospheric Ozone
Reduction: An Update. National Academy Press, Washington, D.C. (1982).
Pinkus, H. Premalignant fibroepithebal tumors of skin. Arch Dermatol.
Syphelol 67:98-615 (1953).
Pollack, S.V., Gaslen, J.B., Sherertz, E.F., and Jegasothy, B.V. The biology
of basal cell carcinoma: A review J Amer Acad Dermatol 7(5):569-577 (1982).
Scotto, J. and Fraumeni, J.F. Jr. Skin (other than melanoma) In: Cancer .
Epidemiology and Prevention. eds. D. Schottenfeld and JF Fraumeni
Philadelphia WB Saunders Co pp 996-1011 (1982).
Scotto, J., Fears, T.R. and Fraumeni, J.F. Jr. Incidence of Nonmelarioma Skin
Cancer in the United States. National Cancer Institute. U.S. Department of
Health and.Human Services. December 1981. Publ. No. (NIH) 82-2433 (1981).
Vitaliano, P.P. and Urbach, F. The Relative Importance of Risk Factors in
Nonmelanoma Carcinoma. Arch Dermatol 116:454-456 (1980).
Urbach, F., Rose, D.B. and Bonnem, M. 1972. Genetic and environmental
interactions in skin carcinogenesis. In Environment and Cancer. The Williams
and Wilkins Co., Baltimore (1972). pp 355-371 as cited in Pollack et al.
(1982).
DRAFT FINAL
-------
CHAPTER 15
ADVERSE EFFECTS OF SOLAR RADIATION:
EVIDENCE FROM CELLULAR/MOLECULAR STUDIES
INTRODUCTION
Although it may be possible on the basis of epidemiologic evidence to draw
conclusions as to whether sunlight is a causal agent for cutaneous malignant
melanoma, the determination of whether UVB is an active component of such a
process will require a thorough understanding of the molecular and cellular
effects induced by UVB. This chapter is designed to present a review of the
information relevant to assessing the role of UVB in cellular and molecular
events. The information is treated generically in separate sections detailing
effects at the cellular or effects at the molecular level followed by a
section integrating observations and relating them to melanoma and melanocytes.
As indicated in Figure 3-6, the energy from solar radiation reaching the
earth is principally derived from wavelengths in the visible range (400-800
nm), although significant amounts of energy in the ultraviolet (200-290 run:
UVC; 290-320 nm: UVB; 320-400 nm: UVA) and infrared (800-17,000 nm) ranges are
also received. The photobiology of sun-induced lesions is primarily concerned
with the wavelengths that reach the earth's surface and have biologic
effects. A significant amount of biological research has been done using
wavelengths in the UVC (200-290) range, however, so that the biologic effects
of this band of radiation are particularly well characterized. As a
consequence, wherever possible, information on UVB, UVA and visible light are
compared and contrasted with information on UVC.
EFFECTS AT THE CELLULAR LEVEL
At the cellular level, solar radiation and its component ultraviolet and
visible wavelengths have been observed to cause such changes in cells as the
induction of cell division and/or differentiation, the loss of specialized
functions, mutation, transformation and death. Subsequent subsections will
discuss these changes and in the next section, those events occurring at the
molecular level that are thought to be responsible for the changes observed at
the cellular level will be reviewed. Wherever possible the information
reviewed will be prioritized to first present information on mammalian
epidermal cells, followed by information on other animal cells; in some
instances, information on bacterial cells, where informative or instructive,
will also be cited.
Cell Division/Differentiation
A single in vivo exposure of human skin to UVB radiation can cause the
epidermis to thicken, thereby increasing the tolerance of skin to subsequent
radiation. This response is due to a sustained increase in keratinocyte
mitosis and is associated with similar changes in the rates of DNA, RNA and
protein synthesis. (Gange and Parrish 1983).
* * * DRAFT FINAL
-------
15-2
Most of the work elucidating this mitotic response has used single
exposures of human or animal skin, although similar studies using repeated
exposures have found similar responses. Following irradiation, the first
change observed (1 to 6 hrs after irradiation) is a reduction in
macromolecular synthesis, accompanied by a considerable increase in DNA repair
(which peaks immediately after irradiation and is diminished in 5 hours).
(Epstein et al. 1970). This change in DNA repair is followed by an increase
in macromolecular synthesis and mitosis that generally persists for several
days and may last as long as a week. UVB and UVC are the most effective
wavelengths for this response although it has also been seen with UVA.
The pathway by which this thickening proceeds is not known. However,
concomitant with the stimulation of proliferation, UV irradiation induces a
large increase in epidermal ornithine decarboxylase. The increase is' first
seen at 2 hours post irradiation, reaches a first plateau at 4 hours and a
more sustained plateau at 24 to 30 hours then declines to normal after 48 to
72 hours. At peak concentration the level of ornithine decarboxylase may be
200 times greater than that seen initially. What makes this an interesting
observation is that ornithine decarboxylase is the rate-limiting enzyme in the
synthesis of the polyamines putresine, spermidine and spermine, which are
increased during proliferative states. Tumor promoters such as phorbol esters
also induce ornithine decarboxylase although with a somewhat more rapid
timecourse (O'Brien 1976) - one that is similar to that seen with multiple UVB
exposures (Lowe 1981)
UVB radiation exposure also induces melanocyte division. Iri a study of
the mitiotic activity of epidermal melanocytes from C57BL mice, Rosdahl and
Szabo (1978) observed that a 5-6 fold increase in the epidermal melanocyte
population of the ear was associated with tritiated thymidine (3HTdr)
labeling of 65 to 80 percent of the melanocytes. The administration of the
3HTdr was timed in such a way as to ensure that the labeling was the result
of DNA duplication and not repair. The authors suggest that the high
percentage of labeled cells is indicative that the UVB-induced increase in
melanocyte population is primarily the result of mitosis, and that the
activation of tyrosinase negative "precursor cells" or the invasion of dermal
melanocytes probably contributes little to the increase in melanocyte numbers
observed. Interestingly, these authors also observed that 26-55 percent of
melanocytes in the control ear (which was shielded from irradiation) were also
labeled. There was no increase in the number of melanocytes in the control
ear suggesting that the labeling was not being stimulated by a systemic effect
of irradiation. This conclusion contrasts, however, with that arrived at by
Scheibner et al.(1986) who evaluated the kinetics of melanocyte density in
human skin before and after sunbathing. These authors felt that "Sunlight
appeared to stimulate melanocytes, both directly in sun-exposed skin and,
indirectly, to a lesser extent, in non-sun-exposed skin, 6-8 weeks after
cessation of exposure.
Other investigators, studying the response of melanocytes in the mouse
trunk epidermis, have concluded that the increase in the number of active
melanocytes after exposure of black mice to UVR involves both proliferation
and recruitment of amelanogenic melanocytes (Miyazaki et al. 1974). This work
* * * DRAFT FINAL *
-------
15-3
finds support in the observation by Uesugi et al. (1979) that indeterminate
cells containing 100 angstrom (intermediate) filaments are precursors of the
melanotic melanocytes that appear in UVR-irradiated trunk skin. In human
skin, there is a great variation in dopa reactivity among nonirradiated
melanocytes, leading to the suggestion that there may be large numbers of
amelanotic or weakly melanogenic melanocytes which escape detection in sheets
of nonirradiated epidermis but which upon irradiation with UV develop
uniformly intense dopa reactivity (Quevedo et al. 1969; Quevedo and Fleishmann
1980).
Loss of Specialized Function
Exposure to UVR has been found to result in impairment of antigen
presenting cell function in both the mouse and man (Greene et al. 1979). The
original observation resulting in this discovery was that of Kripke (1974)
that most UV-induced tumors are rejected by normal syngeneic hosts but not by
UVR treated hosts. Investigation of this phenomenon led to the discovery that
UVR treated mice were immunosuppressed due to the presence of
UV-tumor-antigen specific suppressor lymphocytes (T ) (Kripke et al. 1977;
Daynes et al. 1977; and an investigation of the reasons for the generation of
such suppressor cells led to the discovery of a defect in antigen presenting
cell function (a more detailed discussion of this system is presented in
Chapter 18, The Effect of UVR on the Immune System . . .) About the same time
it was determined that the cell responsible for antigen presentation in the
skin was the Langerhans cell (LC) (Stingl et al. 1978; Toews et al. 1980) and
that these cells are very sensitive to ultraviolet radiation (Toews et al.
1980; Aberer et al. 1981). In the work by Toews et al. (1980), C57BL/6 mice
were irradiated with 100 J/m2 once a day for 4 days (using the output from
an unfiltered FS20 sunlamp) on a shaved 2.5 cm2 area of abdominal wall skin
and then tested in a standard immunizing regimen for development of delayed
type hypersens itivity (DHS) to dinitrofluorobenzene (DNF). The radiation
regimen produced moderate thickening of the epidermis but no significant
cellular infiltrate, yet normal immunization did not occur. Furthermore,
there was a direct correlation between the number of ATPase-staining LC in the
skin and the ability of the skin to permit sensitization during the time
period following radiation when LC were repopulating the irradiated skin.
Subsequent experiments showed that the specific unresponsiveness of UV
irradiated skin was local - irradiation of abdominal wall skin did not alter
the DHS response of dorsal skin.
The loss of ATPase positive cells observed by Toews et al. (1980) was
first interpreted to be the result of cell death or emigration. However,
subsequent work by Aberer et al. (1981) confirmed the work of Toews et al. and
extended it to show that after UV radiation, LC lose their ability to react
with antibody to the la surface antigen. These experiments also showed that
despite their loss of ATPase staining, many LC's remained in the epidermis and
could be detected by electron microscopy. Furthermore, additional work by
Streilein et al. (1980) showed that UVR delivered at 100 J/m2 per day for
four days, did not affect the immunogenicity of skin grafts in donor-host
pairs differing only by the I region of the major histocompatibility complex.
Since Langerhans cells are the main source of I region antigens in the skin,
* * * DRAFT FINAL *
-------
15-4
this suggested that UVR does not destroy LC. An action spectrum for the
effect of UV on LC (Noonan et al. 1984) indicates that while very low doses of
270 and 290 nm UV radiation produced significant alterations in the morphology
and ATPase staining of LC, similar doses of 270 nm radiation did not.
However, these authors also showed that at 320 nm, where there was no effect
on the numbers or morphology of LC, there was a significant effect on DHS,
thus demonstrating that UV effects on LC and UV-induced suppression of DHS
could be separated through the use of different wavelengths of UV radiation
(Noonan et al. 1984).
Mutation
Ultraviolet light is mutagenic for both bacteria and mammalian cells. In
bacteria, UVR treated cells if subsequently treated with photoreactivating
light, show a substantially decreased frequency of mutants per survivor
leading to the conclusion that pyrimidine dimers are responsible for much of
the observed mutagenesis (Hall and Mount 1981). Peak et al. (1984) derived
action spectra for DNA dimer induction, lethality and mutagenesis in E. coli
over wavelengths between 254-405 (Figure 15-1) and found that all three
end-points decreased in efficiency in a similar fashion as the wavelengths of
radiation increased. Between 300 and 320 nm, all characteristics showed
differences of about 2.5 orders of magnitude. Furthermore between about 250
and 320 nm, the values for the three end-points either coincide with or
closely parallel Setlow's (1974) proposed average DNA action spectrum. In the
UVA range (above 325 nm), the spectra for the three end-points diverge sharply
with lethalities at. the UVA wavelengths being approximately ten times greater
relative to mutagenicity than at the shorter wavelengths.
The mechanism by which UV induced mutations are produced in bacteria is
still under investigation. There are efficient repair processes available to
bacterial cells for the removal of pyrimidine dimers. Nevertheless, the close
association between the action spectra for UV induced mutations and pyrimidine
dimers at wavelengths below 320 suggest that pyrimidine dimers are somehow
involved. Two possibilities are suggested by the evidence. First, analysis
of UV induced mutations shows that many of them are produced opposite pyrimi-
dine pairs in the DNA strand which serve as the template for replication --
suggesting that the mutations may have occurred because of some type of DNA
misrepair past pyrimidine dimers. Since pyrimidine dimers disrupt normal base
pairing, what may have occurred is the insertion of one or two incorrect bases
opposite the damaged sites (Hall and Mount 1981).
Studies of irradiated bacteriaphage lambda suggest an alternative
hypothesis, for mutations in the phage DNA occur even when the host bacterial
cell is irradiated and the phage added subsequently. One explanation for this
observation is that host cell functions, induced by UV irradiation, may relax
the fidelity of replication thereby producing mutations during the replication
of undamaged DNA (Hall and Mount 1981).
Experiments performed using fibroblasts from xeroderma pigmentosum (XP)
patients and contrasting their responses to normal human fibroblasts (Mahler
et al. 1979) suggest that unexcised pyrimidine dimers present in cellular DNA
* * * DRAFT FINAL * * *
-------
15-5
Figure 15-1
Action Spectra for DMA Dimer Induction,
Lethality and Mutagenesis
a, • Relative dimer yield per quantum
A—A Relative lethality per quantum
Relative mutagenicity per quantum
Average DNA spectrum
(Setlow, 1974)
A, • Xenon lamp
A, o Hg lines
(Tyrrell, 1973)
250 _. 300 350 400
WAVELENGTH (nml
Source: Peak et al. 1984.
* * * DRAFT FINAL * *
-------
15-6
at the time of cell division are also responsible for the production of
mutations in mammalian cells. These authors found that normal human fibro-
blasts which were UVR irradiated, then stimulated to undergo cell division by
immediate replating, showed much higher numbers of mutations than similarly
irradiated cells which were maintained for seven days as nondividing monolayer
cultures, thus indicating that normal human cells can repair the damage
(presumably by excision repair) induced by UVR. XP cells, by comparison, gave
comparable number of mutants whether replated immediately or given 7 days of
incubation. Since XP cells are unable to excise pyrimidine dimers, these
experiments suggest that unexcised dimers present at the time of cell division
may have been responsible for the production of mutations.
The early experiments in XP cells were performed principally at 254 nm.
Subsequent work provides an action spectrum for lethality in XP fibroblasts
which indicates that between 254 and 313 nm, pyrimidine dimers are the major
lethal lesions. Above 313, however, different lethal mechanisms of an unknown
mechanism are evoked (Keyse et al. 1983).
The importance of mutation to an assessment of the role of UVB in melanoma
development lies in a theory of carcinogenesis which suggests that somatic
mutation in mammalian cells is the first step down a pathway which includes
malignant transformation and ends in neoplasia and metastasis (Trosko and Chu
1975). Peak and his colleagues (Kubitschek et al. 1986) have built upon that
hypothesis to derive estimates of the increase in mutagenesis and basal and
squamous cell skin cancer which might be expected from increased fluences of
solar UVR resulting from ozone depletion. Their conclusions which are based
on the action spectrum for mutagenesis in E. coli, unpublished information
in similar studies in mammalian cells, information on the epidermal
transmission rate of the various UV wavelengths, and consideration of a
biological amplification factor taken from van der Leun (1984), were that a
3-5 percent stratospheric ozone depletion rate would lead to an increase in
nonmelanoma skin cancer rates of about 10 to 20 percent. If UVB is mutagenic
for melanocytes then similar predictions for melanoma are reasonable.
Transformation
In vitro transformation is thought to be correlated, albeit imperfectly,
to in vivo tumorigenesis (Heidelberg 1977). As such it has been used
extensively to characterize the potential carcinogenicity of a wide variety of
chemicals and physical agents. Cells which have been induced in vitro to
undergo loss of certain normally seen growth controls are frequently, although
not always, tumorigenic in mice. A hierarchy of transformational changes is
recognized and the ability of cells to grow without attachment to a solid
substrate (loss of anchorage dependence) is generally accepted as the best
correlate to tumorigenicity (Freedman and Shin 1974).
While most of the in vitro experimentation with UVR has used lamps that
emit most of their radiation at 254 nm, there is some information from studies
using both monochromatic and polychromatic light sources producing wavelengths
between 290 and 400. Withrow et al. (1980) compared the transformation of
murine BALB/c 3T3 cells with a germicidal and a GE 275 watt sunlamp and found
* * DRAFT FINAL * * *
-------
15-7
both were capable of transforming these cells in vitro to anchorage
independent growth. Although studies on the in vitro transformation of
epidermal cells are rare, work by Ananthaswamy and Kripke (1981) indicates
that transformation of primary cultures of BALB/c epidermal cells is possible
with a UVB-emitting FS40 sunlamp. Six of the seven resultant transformed
lines produced tumors when injected into immunosuppressed mice. These tumor
cell lines lacked histologic characteristics of epidermal cells but were shown
by electron microscopy to possess the intermediate cell junctions that are
characteristic of epidermal cells.
In the same laboratory, Fisher et al. (1984) transformed the murine
fibroblast cell line 10T1/2 with a germicidal (primarily emitting 254 nm) lamp
and showed that like their in vivo transformed counterparts (Kripke 1977), the
in vitro transformed cell lines grew preferentially in UVB-irradiated'mice.
This was not true of cell lines transformed with the carcinogen
3-methylcholantrene or with X-irradiation, indicating that there was an
antigenic similarity between cells transformed in vitro and in vivo with UVR.
It could also be shown that FS40 UV irradiation in vitro of murine
fibrosarcomas which were induced in vivo with UVB, increased their metastatic
potential (Fisher and Cifone 1981). In an additional experiment, Ananthaswamy
(1984a) showed that fibroblasts removed from mouse skin irradiated with UVR in
vivo, showed transformation when they were grown in vitro. The important
point of these studies is that not only do they demonstrate that UVR is
capable of transforming cells in vitro but they allow the comparison of in
vitro and in vivo studies and demonstrate the validity of investigating in
vitro transformation in order to understand the in vivo effects of UVR.
It is possible to use inexpensive filters to remove the shorter
wavelengths from polychromatic UVR sources such as the FS40 sunlamp. Such
experiments show that as one removes the shorter wavelengths, the dose of UV
required to transform and mutate cells increases (calculated as per surviving
cell) (Suzuki et al. 1981; Ananthaswamy 1984b).
Thus, whereas the dose of polychromatic UVB required to give a transforma-
tion frequency of about 10-3 per surviving cell is 500 J/m2 for an FS40
lamp, it is about 4000 J/m2 for an FS40 lamp filtered by polystyrene which
removed the wavelengths from 280-294 (Ananthaswamy 1984b). Filtration by
Mylar, which removes wavelengths below about 315 nm, further reduces both
mutation and transformation frequency (Suzuki et al. 1981). Since the FS40
lamp emits less radiation at the shorter UVB wavelengths, the inference from
these studies is that the shorter UVB wavelengths have greater transforming
effectiveness. This is actually borne out by experiments with'monochromatic
light. Doniger et al. (1981) developed action spectra for transformation,
lethality and thymine dimer formation using a monochromatic light source. In
dose response studies comparing pyrimidine dimer formation (indicative of DMA
damage; discussed in detail in later sections of this chapter) and transforma-
tion of Syrian hamster embryo cells, the slopes of the dose-response curves
were not always parallel. The discordance was greatest at 290 nm. The lowest
exposure required for equivalent cell transformation, lethality and pyrimidine
dimer formation was at 270 nm. Comparing results at 290 and 297 nm, the
* * * DRAFT FINAL * * *
-------
15-8
relative effectiveness was 2.3 to 8.7 for dimer formation and 7.4 to 47 for
transformation. Therefore, as the wavelength decreases, the effectiveness of
UVR in transforming cells increases more rapidly than one would predict on the
basis that the pyrimidine dimer is the sole lesion responsible for tranforma-
tion. A similar spectrum was found for induction of anchorage independent
growth in human fibroblasts (Sutherland et al. 1981). These authors also
found a maximum effectiveness at 265 nm and that transformation at 290 nm was
six times more effective per photon than that at 297 nm.
The above data are important for three reasons. The first is the clear
indication that UVR can cause transformation in cells in the absence of any
confounding immunological, hormonal, or physiological effects encountered in
vivo; the conclusion to be drawn is that UVR is directly transforming in
hamsters, mice and human cells of both epidermal and mesenchymal origin. The
second is the confirmation that experiments conducted in vitro may be used to
extrapolate effects in vivo, and the third is that the shorter UVB wavelengths
are more effective in transforming mammalian cells.
As yet, there are no reports of in vitro transformation of melanocytes by
either UV or carcinogens. Only recently, however, have techniques been
developed that allow the cultivation of melanocytes in vitro (Eisinger and
Marko 1982). As noted above, UVB irradiation in vivo does increase the
mitotic activity of murine epidermal melanocytes (Rosdahl and Szabo 1978). In
an attempt to gain insight into the possible role of UVR in the induction of
human melanoma, the sensitivity of human malignant melanoma cell lines to UV
in vitro has been investigated. The finding that some lines are sensitive and
some resistant to UVR makes drawing conclusions difficult (Chalmers et al.
1976; Lavin et al. 1981; Howell et al. 1984).
Lethality
In vitro studies in bacterial and mammalian studies have demonstrated that
one important effect of irradiation with UV is cytotoxicity. Studies of the
action spectrum for this effect suggest that in mammalian cell lines, the
action spectrum for cytotoxicity and neoplastic transformation are the same
(Doniger et al. 1981; Ananthaswamy 1984b) and correlate well with the spectrum
for pyrimidine dimer formation (Doniger et al. 1981). The efficiency of both
cytotoxicity and transformation goes down with increasing wavelength, however,
and one group found that dimer induction alone was not sufficient to account
for killing effectiveness at the upper wavelength ranges (Elkind and Han
1978). A possible answer to this apparently discrepant conclusion may lie in
the observation that at the higher wavelengths e.g. 313 nm, lethality in
bacteria can be in part attributed to an oxygen dependent DNA-strand breakage
(Miguel and Tyrrell 1983). Studies using fibroblasts from XP patients as well
as normal controls would tend to confirm the hypothesis that at 313 nm, a
mechanism other than pyrimidine dimer formation may be important for at 313 nm
the ratio of thymine dimers to single strand breaks is 9 to 1 whereas at 254
nm it is 5700 to 1 (Cerutti and Netrawali 1979). However, in investigating
the repair capabilities of the cells after these radiation treatments, these
investigators found that repair mechanisms were much better able to deal with
the strand breaks and that it was the unexcised dimers which represented the
inhibitory lesions at both wavelengths.
* * * DRAFT FINAL * * *
-------
15-9
EFFECTS AT THE MOLECULAR LEVEL
Chemical changes and biological damage induced by ultraviolet light
require the absorption of light energy (photons) by molecules within the
target. The absorption event is specific and unique, because each type of
molecule is capable of absorbing radiation only in specific wavelength
ranges. Examination of the absorption spectra of molecules in biological
systems indicates that a number of biomolecules can absorb radiation in the
220 to 400 nm region and thus may be critical targets for detrimental UV
effects (Spikes 1979). These molecules include polynucleotides like DNA and
RNA, highly polymerized proteins like keratin and melanin, other proteins
which contain chromophoric cofactors or amino acids and small molecular weight
compounds like urocanic acid.
The following subsections will review what is known about the interaction
of solar radiation with these molecules, tying the information back to events
at the cellular level wherever possible, and will also review the mechanisms
available by which the cell can repair effects on DNA.
Keratins
As indicated in Chapter 3, keratins are a family of polymeric proteins
which constitute a major protein product of the keratinocyte and are the
principal structural proteins of the epidermis (Fitzpatrick and Soter 1985)
Keratins are disulfide-rich proteins which strongly absorb photons in the UVC
and UVB range (Harber and Bickers 1981). .When the epidermis thickens
following UV radiation, most of the thickening is due to increased production
of keratinocytes. The additional keratin produced by these cells results in a
shift in the absorption maximum of the epidermis from approximately 260 nm to
275-280 nm (Agin et al. 1981).
Melanin
The ultraviolet absorption spectra of major epidermal chromophores
including dopa-melanin (a synthetic eumelanin) are given in Figure 3.8.
Keratins would absorb principally in the 250 to 300 nm areas of the
ultraviolet spectrum because of their content of the amino acids tryptophane
(trp) and tyrosine (tyr) with some contribution made by the disulfide bonds
contributed by cysteine residues. In contrast, melanin absorbs across a broad
range of wavelengths although more strongly in the shorter wavelengths than
the longer (it has twice the absorbance at 200 nm that it has at 340 nm.)
Urocanic Acid
Urocanic acid is a naturally occurring substance found in the stratum
corneum of the skin of humans and other mammals and is the major UV-radiation
absorbing compound found there. Its UV absorption spectrum lies between 250
and 320 nm and there are indications that it may act as a photoreceptor,
albeit not in the usual sense.
* * DRAFT FINAL * * *
-------
15-10
Urocanic acid is formed in the stratum corneum by a single-step
deamination of histidine catalyzed by the enzyme histidine-ammonia lyase
(histidase). It has been proposed that UCA acts as a natural sunscreen
protecting the skin from ultraviolet radiation. However, there is evidence
that patients with no histidase activity, and thus no UCA in the epidermis,
are not unusually sensitive to solar radiation (Zannoni and LaDu 1963). Thus
UCA may be involved in a UV-induced process, but other than that of prevention
of erythema.
Based on the close fit of the absorption spectrum of UCA to the action
spectrum for contact hypersensitivity, its location in the stratum corneum,
and its photochemical properties, UCA has been proposed to have an effect in
UV-induced local immune suppression. This is supported by experiments
involving the removal of the stratum corneum with subsequent prevention of the
suppression of DHS (DeFabo and Nponan 1983). However, it is now known that
Langerhans cells can be destroyed by high fluences of UVA, but this does not
prevent the development of UVB-induced suppression of DHS. In addition,
UV-induced suppression of DHS can be produced by certain wavelengths that do
not affect Langerhans cells (Morison 1984). These observations indicate that
a factor other than injury to Langerhans cells is probably responsible for
UV-induced local immune suppression. The possibility still exists that
Langerhans cells may be involved in the subsequent development of a systemic
immune suppression of antigen presentation which play a role which could in
turn promote tumor development.
One of these photoproducts, or a secondary product (formed by interaction
between the photoreceptor and the epidermis), might enter the systemic
circulation and initiate immunologic suppression. Alteration in the antigen-
presenting ability of cells or the distribution of these cells may direct the
production of T suppressor cells specific for a particular antigen (e.g., a
chemical contact sensitizer) (Daynes et al. 1977) instead of effector immune
cells (DeFabo and Noonan 1983).
For many years the physiological role of UCA has remained obscure, despite
the fact that a relatively large amount of this substance (in the trans
configuration) accumulates in mammalian epidermis. A physiological role was
postulated for trans-UCA by DeFabo and Noonan (1983) as a UVB-absorbing skin
photoreceptor necessary to regulate against autoimmune attack on sun-damaged
skin. Under their hypothesis, trans-UCA is converted to the cis isomer, which
is then able to initiate the production of antigen-specific suppressor cells
via the induction of an antigen-processing defect. These suppressor cells
would be specific for the photoantigens induced on sun-damaged skin cells.
UCA might therefore be involved in the outgrowth of UV-induced skin tumors by
inadvertently protecting the tumor cells via the production of tumor-specific
suppressor cells. These suppressor cells would be formed along with those
needed to protect against autoimmune attack on sun-exposed skin.
Alternately, it has been suggested that UCA plays a role in the
photoprotection of DNA. The absorption spectrum of UCA significantly overlaps
with that of DNA, but UCA absorbs UVR much more efficiently. At wavelengths
longer than 290 nm, UCA can be up to 300 times more likely to absorb photons
* * * DRAFT FINAL * * *
-------
15-11
than DNA (DeFabo and Noonan 1983). So, at wavelengths that are involved in
photocarcinogenesis, UCA could be expected to protect DNA from some of the
adverse effects of ultraviolet radiation by reducing doses which reach
sensitive tissues. UCA can directly and efficiently absorb excitation energy
in the trans-cis photoisomerization reaction and it is readily triplet
sensitized, acting as a low triplet energy sink. Due to its low triplet
energy level, UCA is also a very efficient scavenger of singlet oxygen
(Morrison 1985).
DNA
As indicated in Figure 3-6, DNA has an absorption maximum at 265 nm (UVC
range). The shorter wavelengths of UVB radiation are also sufficient to
initiate a certain number of direct photoexcitations of DNA; however,-as the
wavelengths in the UVB range get longer, the probability that direct
photoexcitation events will occur is reduced. Thus, at 320 nm, the
probability of damage to DNA via a direct mechanism is significantly less than
at 290 nm.
A number of different lesions are induced in DNA by UV irradiation. These
include 1) pyrimidine dimers, 2) pyrimidine adducts, 3) single strand breaks,
4) double strand breaks and 5) protein-DNA crosslinks.. Different wavelengths
have different efficiencies for the production of these lesions and there is
also evidence of two possible types of mechanisms - a direct mechanism
resulting from absorption of energy by DNA and an indirect mechanism involving
reactive oxygen species.
The modifications to DNA by UVB are thought to be principally by the
direct mechanism whereas those induced by UVA are thought to principally be
indirect and involve the photoactivation of natural endogenous chromophores
(i.e., bilirubin, porphyrins, and urocanic acid) or photosensitive agents from
cosmetics, tanning oils, or medical preparations. Most of these
photosensitizers have absorption peaks in the UVA range and become excited by
UVA photons. The excited state sensitizers then transfer energy to molecular
oxygen, which in turn becomes photoexcited. Photoexcited oxygen can exist in
one of several excited states at any given time and interconverts between
excited states (singlets, triplets) in a poorly understood fashion (Peak and
Peak 1986c). As the reactive species of oxygen decay, they can transfer their
energy to cellular components, such as DNA, resulting in damage by an indirect
route. The products generated by this indirect route are much the same as
those generated directly, thus, the subsequent sections of this chapter will
discuss these products indicating the irradiation conditions under which they
have been generated.
(1) Pyrimidine dimers may be formed by direct absorption of photons in
the UVB wavelength range. Adjacent pyrimidine molecules on the same strand of
DNA become linked together by a cyclobutane ring between the 5 and 6 carbon
atoms of each residue. Figure 15-2 (taken from Robbins et al. 1974)
illustrates a dimer between adjacent thymines. The 5-6 double bond of one
pyrimidine molecule absorbs a photon of energy and a radical reaction
involving the 5-6 double bond of the adjacent pyrimidine molecule takes
* * * DRAFT FINAL
-------
15-12
FIGURE 15-2
Cyclobutane Pyrimidine Dimer Formed By
UV Light in DNA*
.Suocf
— J,
—-^
f* •! "*
thymine dimer
Sugar
HN
SUGAR
Source: Cleaver (1983).
*Top: Adjacent thymines form 5-5 and 5-6 bonds after absorption of UV
photons.
Bottom: Approximate structure of diraer in DNA with pyrimidine rings stacked
above one another.
* * DRAFT FINAL * * *
-------
15-13
place. The excited state of thymine is a much lower energy state than that of
cytosine and so thymine dimers are the most likely pyrimidine dimers.
Cytosine co-dimers are also possible, as are heterodimers between thymine and
cytosine or uracil (Spikes 1982) .
The action spectrum for pyrimidine dimer formation in E. coli cells is
presented as part of Figure 15-1. As indicated earlier, the action spectra
for pyrimidine dimer formation and mutagenesis are very similar. Furthermore,
in vitro studies of UVR-treated mammalian cells indicate a similar action
spectrum for mutagenesis (Peak et al. 1984) and dimer induction (Doniger et
al. 1981) as do experiments performed in vivo in hairless mice (Ley et al.
1983). These action spectra are also in concordance with that of Freeman
(1975) for ultraviolet carcinogenesis. This information in conjunction with
the observation that in fish the induction of 254 nm UVR-induced thyroid
tumors can be markedly reduced by following the 254 nm treatment with
photoreactivating UVR (>320 nm) (Hart et al. 1977) suggests that pyrimidine
dimers may well be important in the carcinogenesis associated with exposure to
ultraviolet irradiation below 313 nm.
The induction of pyrimidine dimers of the cyclobutane type by UVA has been
demonstrated in E. coli for 334 and 365 nm radiation (Tyrrell 1973b, Peak et
al. 1984) and in mammalian cell DNA for 365 nm radiation (Han et al. 1983).
However, pyrimidine dimers have not been detected in response to fluences at
405 nm radiation at doses well above the biologically effective dose for other
lesions (Han et al. 1983). Endonuclease sensitive sites in DNA have been
observed in UVA irradiated human skin fibroblasts and keratinocytes
(Schothorst et al. 1985) and in intact human skin (Freeman et al. 1985, Gange
et al. 1985).
(2) Action spectra for four-membered ring pyrimidine adduct formation
between two consecutive bases on the same strand of DNA resemble the action
spectra for pyrimidine dimer formation (Patrick and Rahn 1976). The
efficiency of formation of these photoproducts is about 2-10 times lower than
for cytosine-cytosine and thymine-thymine dimers. The proportion of these
adducts varies according to the base content of DNA and becomes much higher
when the ratio of guanine-cytosine/adenine-thymine is greater than or equal to
one. The cycloaddition photoproduct that is an adduct of cytosine and thymine
is quickly deaminated to form a thyminylpyrimidinone product.
Thymine glycols, another form of pyrimidine adduct, are defined as a group
of ring saturated lesions of the 5,6-dihydroxydihydrothymine type and have
been detected in the DNA of human cells after irradiation at 254 nm and 313 nm
(Hariharan and Cerruti 1977). The formation of these lesions is probably
caused by the action of hydroxyl radicals on the 5-6 double bond of thymine.
In both the UVB and UVA range, thymine glycols may result from the action of
reactive oxygen species produced by endogenous sensitizers.
These saturated lesions occur with almost the same frequency as thymine
dimers at 313 nm, indicating their possible significance in the UVB range
(Cerruti and Netrawali 1979). Glycol lesions may undergo spontaneous decay to
form apyrimidinic sites in fashion similar to that described for gamma-
* * * DRAFT FINAL * * *
-------
15-14
radiation produced saturated thyraine glycols (Dunlap and Cerrutti 1975). Only
about one third of the thymine glycols are released from the DNA backbone, but
there is little evidence to suggest that either the ring saturated thymine or
the apyrimidinic decay products are lethal lesions in UV-irradiated DNA.
Pyrimidine hydrates of cytosine result from the addition of a water
molecule across the 5-6 double bond. This water molecule is quickly and
easily lost (the half-life is 58 minutes in native DNA and 51 minutes in
denatured DNA). Cytosine photohydrate may be deaminated during this process
forming uracil, the naturally occurring thymine analog in RNA, and this may
lead to a mutation in newly replicated DNA (Helene 1983).
Brash and Haseltine (1982) have shown that there is a linear relationship
between base damage incidence and mutation incidence. For shorter wavelengths
(UVB), it seems clear that pyrimidine dimers and 6-4 photoadducts are involved
in mutagenesis, but at 365 nm the correlation between dimers/adducts and
mutagenesis is not known (Peak et al. 1984).
(3) DNA single strand breaks can be induced directly by UVB radiation
or indirectly by UVA. Analysis of the relative efficiencies for the induction
of single-strand breaks (SSBs) reveals an action spectrum that corresponds
with nucleic acid absorption below 313 nm (Peak and Peak 1986a). Relative to
thymine dimers, single-strand breaks are induced only to a small extent by 254
nm radiation, but as the wavelength is increased the proportion of single-
strand breaks to thymine dimers increases. At 313 nm, one single-strand break
is induced for every nine thymine dimers (Cerrutti and Netrawalli 1979). Some
single-strand breaks induced at 313 nm may be due to indirect effects from
photosensitizers and oxygen dependent mechanisms (Miguel and Tyrrell 1983),
but the spectral analysis indicates that the majority of single-strand breaks
induced at this wavelength are due to direct effects (Peak and Peak 1986a).
Single strand breaks in the backbone of DNA caused by UVA radiation
deviate significantly from predictions based on a mechanism of direct DNA
photoexcitation. Studies utilizing anoxic and anaerobic irradiation of human
cells (Peak and Peak 1986c) indicate that oxygen is required to induce most of
the DNA damage observed following irradiation at 405 nm. The enhancement of
induction of single-strand breaks by 365 and 405 nm radiation in the presence
of deuterated water (DO) shows the involvement of those reactive oxygen
species that are longer-lived in D90 than in HO. The involvement of
various reactive oxygen species in the induction of single-strand breaks at
UVA wavelengths is complicated and several pathways have been implicated.
Singlet oxygen, peroxide and hydroxyl radicals, and other reactive species
have been identified as being involved in UVA- and visible light-induced
single-strand breaks (Peak and Peak 1982).
Oxygen-dependent single-strand breaks apparently require the participation
of non-DNA photosensitizers that initially can absorb the photons of UVA
wavelength radiation, because neither DNA nor molecular oxygen absorb
significant energy at these wavelengths. Various endogenous photosensitizers
have been suggested, including bilirubin, porphyrins, nicotinamide coenzymes,
and riboflavin (Peak and Peak 1986c). For DNA single-strand breaks in human
* * * DRAFT FINAL * * *
-------
15-15
fibroblasts, an absorption maximum near 460 nm is observed. This maximum
corresponds well with the absorption spectrum for bilirubin, suggesting that
this endogenous chromophore may act as a photosensitive agent in the
production of these lesions in human cells (Peak et al. 1985b).
(4) DMA double strand breaks occur about 80 times less frequently than
single strand breaks at 313 nm in bacteria (Tyrrell 1984). Coupled with data
indicating that 90 percent of x-ray induced double strand breaks in mouse
leukemia cells are reannealed within two hours (Bradley and Kohn 1979), this
indicates that double-strand are likely to have little effect on cell
lethality or transformation in mammalian cells. There appears to be a similar
fast repair mechanism for double-strand breaks induced by non-ionizing
ultraviolet radiation (Ley et al. 1978).
(5) DNA-protein crosslinks can be induced by borderline UVB radiation
(290 nm) via a direct photon-absorbing mechanism in human cellular DNA.
Evidence that the mechanism of DNA-protein cross linking does not involve
photosensitizers and reactive oxygen species, can be seen from
aerobic/anaerobic experiments in DO and water in which no difference was
observed in the amount of DNA-protein cross linking for each of the conditions
(Peak et al. 1985a).
DNA-protein cross linking that results from irradiation at wavelengths
above 290 nm is probably due to photodynamic effects involving excited
photosensitizers and reactive oxygen species (Peak et al. 1985b). The action
spectrum for DNA-protein crosslinking deviates significantly from the DNA
spectrum (Figure 15-3) suggesting that crosslinks are induced by an indirect
mechanism. Below 320 nm, there are approximately 40 DNA-protein crosslinks
per lethal event. As cells can survive 2xl03 DNA-protein crosslinks induced
at longer wavelengths (405 nm), it appears that such a small number of
DNA-protein crosslinks is not important in UV-induced cell lethality, assuming
that there are no interactions between dimers and DNA-protein crosslinks (Peak
et al. 1985b).
DNA-protein crosslinking is demonstrable in normal human fibroblasts
immediately after ultraviolet light irradiation, but this crosslinking is
partially reversed after about 12 hours. In fibroblasts from XP patients,
crosslinking after UV-exposure was not reversed and actually progressed with
time leading to the suggestion that crosslinks might be important to the
mechanism of the toxicity of UVR for XP patients. It has also been suggested
however, that the observed in ability to repair cross links is a secondary
change due to severe cell damage (Fornace and Kohn 1976). The mechanism of
the abnormal sensitivity of XP cells to UV radiation has previously been
considered to be due to defective capacities to repair cyclobutylpyrimidine
dimers in cellular DNA (Smith and Paterson 1981).
One study (Sugiyama et al. 1984) has shown crosslinking between
photoexcited thymine residues on DNA and the lysine e-amino groups of
histones (proteins in the helical interstices of DNA). Although this
experiment was conducted using cellular extracts in buffered solution, it
would seem reasonable to infer that photocrosslinking of DNA to histones is
likely in intact cellular systems, given their proximity in the nucleosome.
* * * DRAFT FINAL * * *
-------
15-16
FIGURE 15-3
Action spectrum for the induction of
DNA-to-protein crosslinks
compared with the spectrum of DMA*
IOC
CO
CO
o
cr
rr. 10'
o
cr
o.
-------
15-17
DNA protein crosslinks can also be induced by ultraviolet radiation in the
UVA range via an indirect mechanism that is largely oxygen dependent (Peak et
al. 1985a). A minor peak at 405 run indicates that porphyrins (a class of
photosensitizers) may be involved. The action spectrum for DNA-protein
cross linking in the UVA range indicates an aerobic dependence and a DO
enhancement that is consistent with an indirect photodynamic mechanism
involving singlet oxygen. The spectral dependence of the formation of
DNA-protein crosslinks in the UVA region indicates that crosslinks may play a
significant role in cell lethality and mutagenicity. It should be noted that
at 405 nm the only lesions induced in DNA by irradiation are single-strand
breaks and DNA-protein crosslinks. Because single-strand breaks are
considered non-mutagenic and their numbers do not correspond well with
lethality, it appears that DNA-protein crosslinks may by more likely .to play a
role in rnutagenic and lethal events due to UVA exposure.
Although the biological significance of DNA-protein crosslinks are not
clear, it would seem that these lesions are not lethal to the cell. In normal
cells the number of DNA-protein crosslinks per genome per lethal hit is
greater than 900 (Peak and Peak 1986a). The conclusion that normal cells have
the ability to repair these lesions seems reasonable, as it is unlikely that
DNA could be properly replicated with significant amounts of protein
covalently bonded to DNA.
RNA
Since the vast majority of organisms carry their genetic information in
DNA, photochemically-induced RNA damage is much less biologically significant
than damage to DNA. Because RNA is replaceable, photoinduced modification may
not be a factor in mutagenesis and lethality. Modification of the properties
of messenger RNA could be important, but the occurrence of such changes has
not been demonstrated (Tyrrell 1984).
DNA-DNA
Little is known about intramolecular DNA crosslinks, but they are formed
in low yields by far ultraviolet radiation (Patrick and Rahn 1976). DNA-DNA
crosslinks have not been detected at longer wavelengths, but some chemical
compounds are known to induce these lesions (Bradley et al. 1979). The
presence of these lesions even in small amounts could lead to serious problems
in DNA replication.
Purine damage
Quantum yields for purine damage are typically ten times lower than those
for pyrimidines, but this type of damage does exist (Patrick and Rahn 1976),
and presumably must be removed from DNA. It seems likely that cellular
endonucleases and glycosylases would remove this type of damage (Tyrrell
1984). The low incidence of purine damage also suggests that these are not
significant lesions in DNA, thus not major contributors to mutagenesis or cell
death.
* * DRAFT FINAL * *
-------
15-18
DNA REPAIR MECHANISMS
When cellular DNA is damaged, the lesions either may have the potential to
cause cell death, or may be less severe and merely disturb DNA transcription
and replication. Changes in DNA that adversely affect cellular functions and
survival need to be repaired in order to assure continuation of a species.
Any process that removes lesions from DNA and/or restores a functional DNA
molecule is generally called a repair mechanism. Often the repair of a lethal
DNA lesion may result in a functional DNA, but with an accompanying
modification of the base sequence leading to altered genetic characteristics
in the surviving cells. Based on the somatic mutation theory of carcino-
genesis discussed earlier, it is these lesions that result in mutations that
subsequently presumably result in neoplastic transformation.
Modes of repair that generate altered segments of DNA are generally called
"error-prone" mechanisms, whereas repair mechanisms that result in unchanged
DNA are called "error-free". The occurrence of misrepair may be due to a
malfunction in the repair process or it may be due to the presence of certain
physiological conditions or cellular enzymatic processes that play a role in
the repair of DNA lesions. The inaccuracy of any repair mechanism is likely
to result in secondary structural changes in DNA, some of which may lead to
mutations.
Many modes of repair are considered constitutive, that is, they are always
"turned on." Excision repair and post-replication repair are always
operational,, although at very low rates, even in the absence of UV-light or
any other insulting agent. Other mechanisms are activated only in the
presence of DNA damage (SOS Repair) or are triggered by the incidence of light
at specific wavelengths (photoreactivation).
Photoreactivation
Repair of DNA lesions mediated by light-induced enzymatic process is term
photoreactivation. The usual photoreactivation results from the action of a
single cellular enzyme, deoxyribodipyrimidine photolyase (Hall and Mount
1981). This enzyme is present in both bacteria and mammalian cells, but
probably constitutes a major pathway for the removal of pyrimidine dimers from
DNA only in bacteria. In bacteria, the photoreactivating enzyme (PRE) is
activated by light in the range 310-440 nm, with a peak activity around 385 nm
(Ikenaga et al. 1970). In mammals, PRE has an action spectrum which extends
into the yellow wavelengths (up to 600 nm) and operates at a lower ionic
strength than the prokaryotic PRE, leading to the suggestion that the human
enzyme may play a different role in the cell (Cleaver 1983).
The photoreactivating enzyme binds very tightly to DNA containing
pyrimidine dimers but does not bind to native DNA. Upon absorption of light
in the appropriate nm range, the PRE photocatalyzes the splitting of the
cyclobutyl ring of pyrimidine dimers, regenerating the original pyrimidine
bases (Helene 1983).
* * * DRAFT FINAL * * *
-------
15-19
A second type of photoreactivation (Type II PR) has been demonstrated in
E. coli. Type II PR has a much narrower wavelength range (310-370 nm) than
the type I enzyme, with a peak of activity around 340 nm. Type II PR does not
split thymine dimers and apparently has the same mechanism as photoprotection
(Ikenaga et al. 1970). Type II PR removes pyrimidine adducts from DNA, but
not directly, and has therefore been called indirect photoreactivation.
A third type of photoreactivation in bacteria, Type III PR, has an even
narrower wavelength range (310-340 nm) with a peak at about 315 nm (Patrick
1977). This type of photoreactivation removes cytosine-thymine heteroadducts
from DNA but the mechanism is unknown.
Studies involving the induction of pyrimidine dimers by 365 nm radiation
indicate that these lesions may contribute to lethal damage at this wavelength
(Tyrrell 1973). Following the observation that pyrimidine dimers are induced
by 365 nm radiation, it was noted that photoreactivation of these lesions was
not observed. As pyrimidine dimers induced at 365 nm are normally photo
reactivatable, the data suggest that photoreactivating enzyme may be destroyed
by high doses of 365 nm radiation (Tyrrell et al. 1973).
Excision Repair
Perhaps the most common mechanism for repairing UV-induced damage to DNA
is excision repair. The process has been extensively studied for bacteria.
In bacteria, an endonuclease recognizes lesions specific for UV-induced damage
and splits the phosphodiester bond near the dimer, usually on the 51 side.
Following this "cut" the DNA fragment containing the lesion peels away, and
another enzyme, DNA polymerase, uses the complementary undamaged strand as a
template to resynthesize the damaged strand, in the 5* to 31 direction. The
dimer region is then excised by the exonuclease activity of DNA polymerase.
Finally, the newly synthesized DNA and the original DNA strand are joined by
DNA ligase. This process, at least in E. coli, is under the control of a
complex system of genes designated UVR A, B, and C. Their gene products
(proteins) associate at pyrimidine dimers on UV-irradiated DNA and cleave the
DNA chain as indicated above. Excision repair systems have been well
characterized in bacterial systems but the comparable mammalian enzymes have
not been well studied (Hall and Mount 1981).
The process of excision repair of thymidine dimers in eukaryotic cells
does appear to differ significantly from excision repair in bacteria. It had
been proposed that the mode of excision repair for mammalian and animal cells
is a nick and cut method similar to that for bacteria but evidence from
experiments involving human cells indicate that this is not the case.
Instead, excision repair in eukaryotic cells involves extensive exonuclease
action to make a wide single-strand region before polymerization, removal of
an oligonucleotide considerably larger than the pyrimidine dimer, or a
combination of both (Cleaver 1984). The distinctive action of eukaryotic
polymerases (especially polymerase a, that cannot act on nicked substrates)
indicate that excision repair involves a sequence of between ten and twenty
bases. Thus, the mode of excision repair for UV-induced damage in human cells
is probably significantly different from the mechanism utilized by E. coli.
* * * DRAFT FINAL * * *
-------
15-20
Excision repair has also been implicated in the repair of other types of
UV-induced DMA damage. Any DNA lesions involving crosslinking of adjacent
moieties on the same strand of DNA are subject to excision repair. The
excision of cytosine-thymine heteroadducts has been noted in M. radiodurans
(Varghese and Day 1970), which has a high concentration of cytosine. This
high concentration of cytosine results in more UV-induced cytosine-thymine
heteroadducts (and a noticeable rate of repair) than in other organisms that
have a lower relative concentration of cytosine in their DNA.
The rate of cytosine-thymine heteroadduct formation in human and animal
cells is probably around five percent of all thymine containing lesions. In
eukaryotic cells, these lesions are also removed by an excision repair
mechanism.
Excision repair is involved in the removal of ring saturated thymine
products of the 5,6-dihydroxydihydrothymine type (thymine glycols) from
UV-irradiated mammalian cells (Hariharan and Cerrutti 1976). DNA-protein
photoinduced crosslinks are also removed by excision repair (Helene 1983).
At least one other type of ultraviolet radiation-induced lesion is removed
from DNA by excision repair. Photoinduced complexes between furocoumarins and
DNA have been shown to be excised from E. coli by the action of the UVR A,
B, C complex. Psoralens, a type of furocoumarin, are used in phototherapy of
skin diseases (in conjunction with UVA for the prophyloxis of psoriasis), in
some tanning preparations, and in cosmetics. These psoralen compounds complex
with DNA, and upon absorption of UVA light are covalently bound to DNA (De Mol
et al. 1981). These photoinduced covalent bonds between DNA and psoralens can
then be removed by the action of an excision repair mechanism.
The removal of psoralens covalently linked to DNA is important because not
only are the psoralen-DNA adducts likely -to represent premutagenic or
precytotoxic lesions, the furocoumarins can also act as photosensitizing
agents, producing singlet oxygen in the presence of UVA. De Mol and
colleagues (1984) have shown that while psoralens which are not complexed to
DNA produce some singlet oxygen, covalently bound psoralens produce more than
three times as many singlet oxygen molecules. Thus excision of psoralen
covalently bound to DNA is important in the prevention of mutagenicity and
cell lethality both because of the direct effect of the adduct and because of
possible from the indirect effects of ultraviolet radiation mediated by
singlet oxygen formation.
Post-Replication Repair
When DNA synthesis occurs in UV-irradiated cells, the newly synthesized
DNA has been found to have a lower molecular weight than newly synthesized DNA
from unirradiated cells. The low molecular weight is due to gaps in the DNA
daughter strand that arise because replication is blocked at pyrimidine dimers
and other lesions, and then resumes at some site past the lesion. Repair of
these gaps has been called post-replication repair or daughter-strand gap
repair. It should be noted that post-replication repair is not a repair
process removing damage from DNA, such as occurs in excision repair and
* * * DRAFT FINAL * *
-------
15-21
photoreactivation, but rather a process enabling cells to replicate damaged
chromosomes (Hall and Mount 1981). It may not accurately repair the damaged
DNA and is thus sometimes referred to as error-prone repair.
During this type of repair the gaps are filled and the discontinuous
strands are joined into molecules of high molecular weight. The mechanism by
which this occurs involves recombinational strand exchange resulting in
stretches of parental DNA covalently bound to daughter strands (Walker et al.
1985). After the transfer of DNA, via a series of complex rearrangements, any
new gaps created in the second daughter strand can be filled by DNA polymerase
using the correct information on the original undamaged parent strand.
In E. coli, this process is under control of the recA gene, whose
product plays a fundamental role in the recombination step. The final fate of
the damage may be removal by excision repair or it may be "diluted out" by
being transmitted to only one of the daughter cells (Helene 1983).
Post-replication repair in bacterial systems results in 50 percent of thymine
dimers being transferred from parental to newly synthesized strands of DNA.
In UV-irradiated human fibroblasts there are conflicting data concerning
recombination as a mechanism for gap-filling. The action of UV-specific
endonucleases indicate that pyriraidine transfer may occur at a low rate (5-15
percent of total pyrimidine dimers) but some pulse-labeling studies indicate
that this may not be the case. Other pulse-labeling studies (label introduced
two hours after UV treatment) indicate that pyrimidine dimers can be detected
in newly synthesized DNA (Hall and Mount 1981).
Although the possibility of pyrimidine dimer transfer from parental to
newly synthesized DNA may not be great, there are other mechanisms that
indicate genetic recombination could still be involved in gap-filling. It is
conceivable that only short fragments of DNA (compared to the average distance
between dimers) might be exchanged, thereby reducing the chance of dimer
transfer to newly synthesized DNA. Additionally, SV-40 virus probably
utilizes host-cell functions to undergo UV induced recombination, and so these
same processes might also act on UV-damaged cellular DNA (Hall and Mount 1981).
Convincing evidence that either supports or denies a recombinational
repair method in UV-irradiated mammalian cells is lacking. Even though DNA
replication is temporarily blocked at the sites of pyrimidine dimers, it is
difficult to say how different mammalian cell types react subsequently. There
are conflicting data on gap formation, DNA synthesis re-initiation, and
genetic recombination in the repair of gaps opposite UV-induced lesions; thus,
a clear picture of post-replication repair in mammalian cells is not yet
available.
Constitutive expression of DNA polymerase may account for repair of some
UV-induced lesions. Peak and Peak (1982) have shown that an oxygen dependent
mechanism is involved in the production of single-strand breaks at longer
wavelengths in both wild type and mutant strains of bacteria. Single strand
breaks induced in E. coli by 365 run UVR can be repaired by a fast mechanism
* * * DRAFT FINAL * * *
-------
15-22
in wild-type cells, up to eighty percent of lesions being repaired in ten
minutes (Ley et al. 1978). In DNA-polymerase-I-deficient mutant strains of
E. coli, repair of 365 nm-induced single strand breaks is not observed.
The rate of induction of single-strand breaks in DNA polymerase
I-proficient cells may be lower in vivo than in vitro due to the immediate
action of pre-existing polymerase. Because DNA polymerase apparently is
responsible for the resealing of single-strand breaks in DNA, intact cells may
repair single-strand breaks during the course of irradiation, thereby
producing fewer breaks than would be expected based on studies of cell-free
isolates of DNA.
DNA polymerases also play a role in the elongation of daughter strand
DNA. Lesions on parental strands of DNA are known to at least temporarily
inhibit elongation of DNA daughter strands (Tyrrell 1984). One such lesion is
the apyrimidinic site that is formed by the spontaneous release of damaged
bases from UV-irradiated DNA. Thymine dimers and thymine glycols are examples
of lesions induced in UV-irradiated DNA that can be spontaneously released by
the action of glycosylases from the DNA backbone leaving apyrimidinic sites
(Dunlap and Cerrutti 1975). DNA polymerases have been shown to incorporate
nucleotides into apyrimidinic sites (Walker 1985) using the opposite strand as
a template.
The repair of pyrimidine hydrates is extremely difficult to measure both
in vivo and in vitro. Although there are no known photoenzymatic or direct
photolytic removal processes for pyrimidine hydrates (Tyrrell 1984), a
mechanism for repair has been postulated. Cytosine hydrates, their deaminated
products (uracil and uracil hydrates), and rare thymine hydrates are
susceptible to recognition and removal by glycolytic activity, leaving an
apyrimidinic site. This would be followed by the usual response to
apyrimidinic sites.
CELLULAR AND MOLECULAR MECHANISMS IN MELANOMA
The relevance of the foregoing discussion to melanoma induction is
dependent on the demonstration that UVR can cause mutagenesis and
transformation of melanocytes. Unfortunately, the conditions required to
obtain proliferating cultures of melanocytes have just recently been defined
and experiments in which UVR has been used to transform melanocytes have not
yet been published. There is some information on melanoma cells maintained in
vitro, however, which indicates that some cell lines are particularly
resistant to cell killing following irradiation with 254 nm UVR whereas others
are not (Chalmers et al. 1976; Lavin et al. 1981). The differences were
thought by these authors to not involve the pyrimidine dimer excision repair
process.
Research from another laboratory (Konishi 1981) examined not only excision
repair but also caffeine-sensitive (post-replication) repair in melanoma cell
lines. This report also concluded that the difference lay not in the excision
repair process and noted that the data suggest that melanoma cell lines
demonstrate a more rapid post-replication repair process which is also more
* * DRAFT FINAL * * *
-------
15-23
caffeine sensitive than that seen in the comparison (HeLa) cells. This report
has not been confirmed, so must be viewed with caution. It does, however,
suggest an interesting hypothesis which is presented below.
Given that UV is clearly mutagenic for both fibroblasts and epidermal
cells, with the most active wavelengths for this effect being in the UVB
range, it is reasonable to conclude that UVB will be mutagenic for
melanocytes. Differences in the relative efficiency of mutagenesis as
compared to keratinocytes may occur because the melanin produced by
melanocytes may protect the nuclear material in a manner which is dose
dependent but has a threshold such that small doses never reach the DNA but
large doses either saturate the ability of the melanin to absorb them or
saturate the ability of the raelanocyte repair processes to repair them.
Alternatively, or perhaps in conjunction with the protection to the DNA
provided by melanin, a more effective post replication repair mechanism may
contribute to the apparently different responsiveness of melanoma and the
non-melanoma skin cancers to solar radiation.
FINDINGS
The material reviewed above leads to the following three findings:
15.1 UVR is the most active portion of solar radiation in the
induction of adverse effects; it has been shown in vivo and in
vitro to induce transformation of mamalian epidermal cells. It
is also mutagenic. All of these effects are thought to occur
via a mechanism that involves DNA damage.
15.2 UVB is the most active waveband for these effects and for the
induction of pyrimidine dimers which are thought to be
important to skin cancer development e.g., in XP patients.
15.3 UVR and in particular UVB is active in the induction of
melanogensis and beratinocyte proliferation - two mechanisms
which provide protection from solar radiation to the basal
layer.
* * DRAFT FINAL *
-------
15-24
REFERENCES
Aberer, W., Schuler, G. , Stingl, G., Honigsman, H., Wolf, K. Ultraviolet
light depletes surface markers of Langerhans cells. J Invest Dermatol
76:202-210 (1981).
Agin, P.P., Descrochers, D.L. and Sayre, R.M. The relationship of immediate
pigment darkening to minimal erythemal dose, skin type and eye color.
Photodermatology 2:288-294 (1985).
Ananthaswamy, H.N. Neoplastic transformation of neonatal mouse skin
fibroblasts in culture after a single exposure to ultraviolet radiation in
vivo. Photodermatol 1:106-113 (1984a).
Ananthaswamy, H.N. Lethality and transformation of 10T 1/2 mouse embryo
fibroblast cell line by various wavelengths of ultraviolet radiation.
Photodermatol 1:265-276 (1984b).
Ananthaswamy, H.N., Kripke, M.L. In vitro transformation of primary cultures
of neonatal BALB/c mouse epidermal cells with ultraviolet radiation. Cancer
Res 41:2882-2990 (1981).
Anderson, R.R. Tissue optics and photoimmunology. Chapt 4 in Parrish, J.A.,
Kripke, M.L. and Morison, W.L. (eds). Photoimmunology. Plenum Medical Book
Co. New York. pp. 61-76 (1983).
Bradley, M.O., Hsu, I.C., Harris, C.C. Relationship between sister chromatid
exchange and mutagenicity, toxicity and DNA damage. Nature 282:318-320 (1979).
Bradley, M.O. and Kohn, K.W. X-Ray induced DNA double strand break production
and repair in mammalian cells as measured by neutral filter elution. Nucleic
Acid Res 7:793-804 (1979).
Brash, D.E. and Haseltine, W.A. UV-Induced mutation hotspots occur at DNA
damage hotspots. Nature 298:189-192 (1982).
Cerutti, P.A. and Netrawali, M. Formation and repair of DNA damage induced by
indirect action of ultraviolet light in normal and xeroderma pigmentosum skin
fibroblasts. In: Radiation research: Proceedings of the Sixth
International Congress of Radiation Research Okada, S., Imamura, M.,
Terashima, T. and Yamaguchi, H.(eds). Tokyo pp 424-432 (1979).
Chalmers, A.H., Lavin, M., Atisoontornkul, S., Mansbridge, J., Kidson, C.
Resistance of human melanoma cells to ultraviolet radiation. Cancer Research
36:1930-1934 (1976).
Chan, G.L., Peak, M.J., Peak, J.G., Haseltine, W.A. Action spectrum for the
formation of (6-4) photoproducts induced in DNA by ultraviolet radiation.
Photochem Photobiol (1984).
* * * DRAFT FINAL * *
-------
15-25
Cleaver, J.E. Completion of excision repair patches in human cell
preparations: Identification of a probable mode of excision and resynthesis.
Carcinogenesis 5(3):325-330 (1984).
Cleaver, J.E. Xeroderma pigmentosum. Chapt 57 in Stanbury, J.B.,
Wyangaarden, J.B., Frederickson, D.S., Golstein, J.L., and Brown, M.S. (eds).
The Metabolic Basis of Inherited Disease. McGraw-Hill Book Company. New
York. pp. 1227-1248 (1983).
Daynes, R.A., Spellman, C.W., Woodward, J.C., and Stewart, D.A.
Transplantation 23:343-348 (1977).
De Mol, N.J., Beijersbergen van Henegouwen, G.M.J. and van Beele, B. Singlet
oxygen formation by sensitization of furocoumarins complexed with, or bound
covalently to DMA. Photochem Photobiol 34:661-666 (1984).
DeFabo, E.G. and Noonan, F.P. Mechanism of immune suppression by ultraviolet
radiation in vivo: I. evidence for the existence of a unique photoreceptor in
skin and its role in photoimmunology. J Exp Med 158:84-98 (1983).
Doniger, J., Jacobson, E.D., Krell, K., Di Paolo, J.A. Ultraviolet light
action spectra for neoplastic transformation and lethality of Syrian hamster
embryo cells correlate with spectrum for pyrimidine dimer formation in
cellular DMA. Proc Nat Acad Sci (USA) 78(4):2378-2382 (1981).
Dunlap, B. and Cerrutti, P.A. Apyrimidinic sites in gamma-irradiated DNA.
Febs Lett 51(1):188-190 (1975).
Eisinger, M., Marko, 0. Selective proliferation of normal human melaocytes in
vitro in the presence of phorbol ester and cholera toxin. Proc Natl Acad Sci
(USA) • 79:2018-2022 (1982).
Elkind, M.M., and Han, A. DNA single-strand lesions due to "sunlight" and UV
light: A comparison of their induction in Chinese hamster and human cells,
and their rate in Chinese hamster cells. Photochem Photobiol 27:717-724
(1978).
Epstein, J.H., Fukuyama, K., and Fye, K. Effects of Ultraviolet Radiation on
the Mitotic Cycle and DNA, RNA and Protein Synthesis in Mammalian Epidermis in
Vivo. Photochemistry and Photobiology 12:57-65 (1970).
Fisher, M.S., Cifone, M.A. Enhanced metastatic potential of murine fibrosar
comas treated in vitro with ultraviolet radiation. Cancer Res. 41:3018-3023
(1981).
Fisher, M.S., Kripke, M.L., Chan, G.L. Antigenic similarity between cells
transformed by ultraviolet radiation in vitro and in vivo. Science
223:593-594 (1984).
* * * DRAFT FINAL * * *
-------
15-26
Fitzpatrick, T.B. and Soter, N.A. Pathophysiology of the skin in
pathophysiology: The Biological Principles of Disease. Smith, L.H. Jr. and
Thier, S.O. (eds), pp. 1297-1330, 1985.
Fornace, A.J. and Kohn, K.W. DNA-protein crosslinking by ultraviolet
radiation in normal human and Xeroderma Pigmentosum fibroblasts. Biochem
Biophys Acta 435(1):95-107 (1976).
Freedman, V.H., Shin S. Cellular tumorigenicity in nude mice: Correlation
with cell growth in semi-solid medium. Cell 3:355-359 (1974).
Freeman, R.G. Data on the action spectrum for ultraviolet carcinogenesis. J
Natl Cancer INst 55:1119-1121 (1975).
Gange, R.W., and Parrish, J.A. Acute effects of ultraviolet radiation upon
the skin. In: Photoimmunology. Parrish, J.A., Kripke, M.L., and Morison,
W.L. (eds). New York: Plenun Publ Co. pp 77-93 (1983).
Greene, M.I., Sy, M.S., Kripke, M.L., Benacerrof, B. Impairment of
antigen-presenting cell function by ultraviolet radiation. Proc Natl Acad Sci
(USA) 76:6592-6595 (1979).
Hall, J.D. and Mount, D.W. Mechanisms fo DNA replication and mutagenesis in
ultraviolet irradiated bacterial and mammalian cells. Proc Nucl. Acid Res Mol
Biol 25:53-126 (1981).
Han, A., Peak, M.J. Peak, J.G. Induction of DNA-Protein crosslinking in
Chinese hamster cells by monochromatic 365 nm and 405 nm ultraviolet light.
Photochem Photobiol 39:343-348 (1983).
Harber, L.C. and Bickers, D.B. Photoprotection ('mad dogs and Englishmen go
out in the midday sun1 -- Noel Coward). Chapt. 9 in Harber, L.C. and Bickers,
D.R. (eds). Photosensitivity Diseases: Principles of Diagnosis and
Treatment. W.B. Saunders Company. Philadelphia, pp. 85-96 (1981).
Hariharan, P.V. and Cerrutti, P.A. Formation of products of the 5,6-
dihydroxydihydrothymine type by ultraviolet light in HeLa cells. Biochemistry
16:2391-2395 (1977).
Hariharan, P.V. and Cerrutti, P.A. Excision of ultraviolet and gamma ray
products of the 5,6-dihydroxydihydrothymine type by nuclear preparations of
Xeroderma Pigmentosum cells. Biochem Biophys Acta 447:375-378 (1976).
Hart, R.W., Setlow, R.B., Woodhead, A.D. Evidence that pyrimidine dimers in
DNA can give rise to tumors. Proc Nat Acad Sci (USA) 74(12):5574-5578 (1977).
Heidelberg, C. Oncogene transformation of rodent cell lines by chemical
carcinogens. In: Origins of human cancer Hiatl, H.H., Watson, J.D.,
Winster, J.A. (eds). Cold Spring Harbor Laboratory pp 1513-1520 (1977).
* * DRAFT FINAL * *
-------
15-27
Helena, C. Photoinduced responses in UV irradiated cells. In: Molecular
models of photoresponsiveness Montagnoli, G. and Erlanger, B.F.(eds). New
York: Plenum Press Co. pp 57-73 (1983).
Howell, J.N., Greene, M.H., Corner, R.C., Maher, V.M., McCormick, J.J.
Fibroblasts from Patients with Hereditary Cutaneous Malignant Melanoma are
Abnormally Sensitive to the Mutagenic Effect of Simulated Sunlight and
4-nitroquinoline 1-oxide. Proc. Natl. Acad. Sci. USA 81:1179-1183 (1984).
Ikenaga, M., Patrick, M.H., Jagger, J. Action of photoreactivating light on
pyrimidine heteroadduct in bacteria. (1970).
Keyse, S.M., Moss, S.H., Davies, D.J.G. Action spectra for inactivation of
normal .and Xeroderma Pigmentosum human skin fibroblasts by ultraviolet
radiations. Photochem Photobiol 37(3):307-312 (1983).
Konishi, T. Resistance to ultraviolet radiation in human melanoma cells. Mil
Medical Journal 31(1):15-27 (1981).
Kripke, M.L. Antigenicity of murine skim tumors induced by ultraviolet light.
J Natl Cancer Inst 57:211-215 (1974).
Kripke, M.L. Latency, histology, and antigenicity of tumors induced by
ultraviolet light in three inbred mouse strains. Cancer Research 37:1395-1400
(1977). .
Kriple, M.L., Lofgreen, J.S., Beard, J., Jessup, J.M., Fisher, M.S. In vivo
immune responses of mice during carcinogenesis by ultraviolet irradiation. J.
Natl. Cancer Inst. 59:1227-1230 (1977).
Kubitschek, H.E., Baker, K.S., Peak, M.J. Enhancement of mutagenesis and
human skin cancer rates resulting from increased fluences of solar ultraviolet
radiation. Photochem Photobio in press (1986).
Lavin, M.F., Good, M., Counsilman, C., Kidson, C. Sensitivity and resistance
of human melanoma cells to ultraviolet radiation. Austral J Exp Biol Med Sci
59:515-520 (1981).
Ley, R.D., Peak, M.J., and Lyon L.L. Induction of pyrimidine dimers in
epidermal DNA of hairless mice by UBG: an action spectrum. J Invest Derm
80:188-191 (1983).
Ley, R.D., Sedita, B.A., and Boye, E. DNA polymerase I-mediated repair of 365
nm-induced single-strand breaks in the DNA of Escherichia Coli. Photochem
Photobiol 27:323-327 (1978).
Miguel, A.G. and Tyrrell, R.M. Induction of oxygen-dependent lethal damage by
monochromatic UVB (313 mm) radiation: Strand breakage, repair, and cell
death. Carcinogenesis 4(4):375-380 (1983).
* * * DRAFT FINAL * * *
-------
15-28
Mahler, V.M., Dorney, D.J., Mendrala, A.L., Konz-Thomas, B., and McCormick
J.J. DNA Excision-repair processes in human cells can eliminate the cytotoxic
and mutagenic consequences of ultraviolet irradiation. Mutation Research
62:311-323 (1979).
Miyazaki, H., Kawada, A., Tahaki, Y., Sato, T. and Matsutani. Effects of
ultraviolet light on epidermal dendritic cells of hairless mice in sunlight
and man Fitzpatrick, T.B., Pathok, M.A., Harber, L.C., Seiji, M. and
Kukita, A. (eds). Tokyo Unio Tokyo Press pp 217-229 (1974).
Morison, W.L. Immunology. Yearly Review. Photochem Photobiol 40(6):781-787
(1984).
Morrison, H. Photochemistry and photobiology of urocanic acid.
Photodermatology 2:158-165 (1985).
Noonan, F.P., Bucana, C., Sauder, D.N. and DeFabo, E.G. Mechanism of systemic
immune suppression by UV irradiation in vivo: II. The UV effects on number
and morphology of epidermal Langerhans cells and the UV-induced suppression of
contact hypersensitively have different wavelength dependence. Journ of
Immunol 132:2408-2416 (1984).
Noonan, F.P., Kripke, M.L., Pedersen, G.M. and Greene, M.I. Suppression of
contact hypersensitivity in mice by ultraviolet irradiation is associated with
defective antigen-presentation. Immunology 43:527 (1981).
Patrick, M.H. Studies of thymine-derived UV photoproducts in DNA-polymerase -
I. Formation and biological role of pyrimidine adducts in DNA. Photochem
Photobiol 25:357-372 (1977).
Patrick, M.H. and Rahn, R.O. Photochemistry of DNA and polynucleodimers:
Photoproducts. In: Photochemistry and photobiology of nucleic acids Wang,
S.Y. (ed). Vol. 2 New York: Academic Press pp 35-95 (1976).
Peak, M.J., and Peak, J.G. Induction of single strand breaks in human cell
DNA by far-ultraviolet, near ultraviolet, and visible radiations: Action
spectrum and mechanisms. Submitted to Photochem Photobio for review. Not in
press (1986a).
Peak, M.J., and Peak, J.G. Molecular Photobiology of UVA. In press (1986b).
Peak, M.J. and Peak, J.G.. DNA to protein crosslinks and backbone breaks
caused by far- and near-ultraviolet, and visible light radiations in mammalian
cells. Photchem Photobiol in press (1986c).
Peak, M.J., Peak, J.G. and Jones, C.A. Different (direct and indirect)
mechanisms for the induction of DNA-protein crosslinks in human cells by far-
and near-ultraviolet radiations (290 and 405 nm). Photochem Photobiol
42(2):141-146 (1985a).
* DRAFT FINAL * * *
-------
15-29
Peak, J.G.j Peak, M.J., Sikorski, R.S. and Jones, C.A. Induction of
DNA-protein crosslinks in human cells by ultraviolet and visible radiations:
Action spectrum (1985b).
Peak, M.J., Peak, J.G., Moehring, M.P. and Webb, R.B. Ultraviolet action
spectra for DNA dimer induction, lethality, and mutagenesis in Escherichia
coli with emphasis on the UVB region. Photochem Photobiol 40:613-620 (1984).
Peak, M.J., and Peak, J.G. Single-strand breaks induced in Bacillus Subtilis
DNA by ultraviolet light: Action spectrum and properties. Photochem
Photobiol 35:675-682 (1982).
Quevedo, W.C. Jr., Szabo, G., Virks, J. Influence of age and UV on the
population of tyrosine positive melanocytes in human skin. J Invest -Derm
52:287-290 (1969).
Quevedo, W.C. and Fleishmann, R.D. Developmental biology of mammalian
melanocytes. The Journal of Investigative Dermatology 75(1):116-120 (1980).
Robbins, J.H., Kraemer, K.H., Lutzner, M.A., Festoff, B.W., and Coon, H.G.
Xeroderma Pigmentosum: An Inherited Disease with Sun Sensitivity, Multiple
Cutaneous Neoplasms, and Abnormal DNA Repair. Annals of Internal medicine
80:221-248 (1974).
Rosdahl, I.K. Melanocyte mitosis in UVB irradiated mouse skin. Acta Dev Mato
.Venereol (Stockh) 58: 217-221 (1978).
Rosdahl, I.K. and Szabo, G. Mitotic activity of epidermal melanocytes in
UV-irradiated mouse skin. JID 70:143-148 (1978).
Scheibner, A., Holbs, D;E., McCarthy, W.H. and Multon, G.W. Effects of
sunlight exposure on Langerheins cells and melanocytes in human epidermis.
Photoderm 3:15-25 (1986).
Setlow, R.B. The wavelengths in sunlight effective in producing skin
cancer: A theoretical analysis. Proc Nat Acad Sci (USA) 71(9):3363-3366
(1974).
Setlow, R.B., Regan, J.D., German, J. and Carrier, W.L. Evidence that
xeroderma pigmentosum cells do not perform the first step in the repair of
ultraviolet damage to their DNA. Proc Natl Acad Sci USA 64:1035-1041 (1969).
Smith, P.J. and Paterson, M.C. Abnormal responses to mid-ultraviolet light
of cultured fibroblasts from patients with disorders featuring sunlight
sensitivity. Cancer Research 41:511-518 (1981).
Spikes, J.D. UV-damage in biological systems in Daynes, R.A. and Spikes, J.D.
(eds). Experimental and Clinical Photoimmunology, Volume I. CRC Press Inc.
(1979).
* * DRAFT FINAL * * *
-------
15-30
Stingl, G., Kathy, S.I., Clement, L., Green, I. and Shevach. Immunological
functions of la bearing epidermal Langerhans cells J Immunol 121:2005-2013
(1978).
Streilein, J.W., Toews, G.B., and Bergstresser, P.R. Langerhans cells:
Functional aspects revealed by in vitro grafting studies. J. Invest. Derm.
75:17-21 (1980).
Siigiyama, S., Matsuura, T. and Saito, I.. The role of thymine-lysine
photoaddition in photo-crosslinking of proteins to DNA. Photochem Photobiol
39:Suppl(91S) (1984).
Sutherland, B.M., Delihas, N.C., Oliver, R.P., and Sutherland, J.C. Action
spectra for ultraviolet light-induced transformation of human cells to
anchorage independent growth. Cancer Research 41:2211-2214 (1981).
Suzuki, F., Han, A., Lankas, G.R., Utsumi, H. and Elkind, M.M. Spectral
dependance of killing, mutation, and transformation in mammalian cells and
their relevance to hazards caused by solar ultraviolet radiation. Cancer
Research 41:4916-4924 (1981).
Toews, G.B., Bergstresser, P.R., Streilein, G.W. and Sullivan, S. Epidermal
Langerhans cell density determines whether contact hypersensitivity or
unresponsiveness follow skin painting with DNFB J. Immunol 124(1):445-453
(1980). . .
Trosko, J.E. and Chu, E.H.Y. The role of DNA repair and somatic mutation in
carcinogenesis. Adv. Cancer Res. 21:391-419 (1975).
Tyrrell, R. Induction of pyrimidine dimers in bacterial DNA by 365 nm
radiation. Photochem Photobiol 17:69-73 (1973).
Tyrrell, R.M., Webb, R.B., Brown, M.S. Destruction of photoreactivating
enzyme by 365 nm radiation. Photochem Photobiol 18:249-254 (1973).
Tyrrell, R. Damage and repair from non-ionizing radiations. In: Reparable
lesions in microorganisms Hurst, A. and Nasin, A. (eds). New York:
Academic Press (1984).
Uesugi, T., Mitsuko, K.I., HGorikoshi, T., Sadao, S., and Jimbrew, K. Mode of
activation and differentiation of dormant melanocytes after UV exposure of
mouse skin. In: Pigment cell 101 4 biologic basis of pigmentation. Based
Harger, S., Klaus, S.N. (eds). pp 337-344 (1979).
Urbach, F. Ultraviolet radiation and skin cancer. In: Topics in
photomedicine. Smith, K.C. (ed). New York: Plenum Press pp 39-142 (1984).
van der Leun, J.C. Yearly Review: UV-Carcinogenesis. Photochemistry and
Photobiology 39:861-868 (1984).
Varghese, A.J. and Day III, R.S. Excision of cytosine-thymine adduct from
the DNA of ultraviolet-irradiated micrococcus radiodurans. Photochem
Photobiol 11:511-517 (1970).
* * * DRAFT FINAL * * *
-------
15-31
Walker, G.C., Marsh, L. and Dodson, L. Cellular responses to DNA damage.
Environmental Health Perspectives 62:115-117 (1985).
Withrow, T.J., Lugo, M., Dempsey, M.J. Transformation of BALB/c 3T3 cells
exposed to a germicidal UV lamp and a sunlamp. Photochem Photobiol
31:135-141 (1980).
Zannoni, V. and LaDu. Determination of Histidine a-Deaminase in Human
Stratum Corneum and Its Absence in Histidinaemia. Biochem. J. 88:160-162
(1963).
* * * DRAFT FINAL * *
-------
CHAPTER 16
UV RADIATION CAN CAUSE SKIN CANCER IN ANIMALS
The presence of melanoma and other skin cancers in animals indicates the
theoretical possibility that these animals can be used as models for
discovering the role of sunlight and ultraviolet B radiation (UVB) as an agent
in the development of melanoma. This chapter reviews information on the
natural occurrence of melanoma in animals, potential animal models of
melanoma, attempts to induce melanoma in animals in the laboratory, and
induction of non-melanoma skin cancer in laboratory animals. There are two
important lines of analysis that are pursued: the direct evidence for a
UVB/melanoma relationship by the induction of melanoma under experimental
conditions, and an analysis of the potential of ultraviolet radiation (UVR)
and in particular UVB to cause melanomas by examining its capacity to directly
induce other epiderimal cell neoplasms.
ANIMAL MODELS OF MELANOMA
The investigation of animal models of melanoma has proceeded along two
lines: (1) evaluation of the development of spontaneous melanomas in domestic
animals and (2) evaluation of melanomas induced in laboratory animals. The
following chapter discusses the occurrence of melanomas in domestic animals in
brief and then discusses the induction of melanocytic tumors in laboratory
animals in greater detail.
Although melanomas do occur spontaneously in domestic animals, the
incidence is too low to serve as a potential model for the effect of UV on
cutaneous melanoma with the exception of the high melanoma incidence strains
of swine, Duroc and Sinclair swine.
The histopathologic correlation of the biological behavior of melanoma in
domestic animals has been studied and has some similarities to melanoma in
man. Benign tumors with junctional activity and dermal melanomas ("blue
nevi") have been identified. In dogs, malignant melanomas composed of
epithelioid cells or spindle cells or both have been identified. The
epithelioid type of malignant melanoma carries the worst prognosis, especially
if there are areas with a high mitotic index. These melanomas have been
reported to metastasize widely, mainly to lymph nodes and lung.
Occurence of cutaneous melanomas in domestic animals has been reported
in a wide variety of species including dogs, cats, horses, mules, donkeys,
cows, pigs, sheep, Indian water buffalos, gerbils, hamsters, rabbits, and
chickens (rev. in Garma-Avina et al. 1981). With the exception of dogs, the
incidence of malanoma is rare in domestic animals. The reported frequency of
melanomas in dogs ranges from 2.3 percent of all tumors studied to 19.6
percent of cutaneous tumors studied. Purebred dogs, including boxers and
cocker spaniels were overrepresented in more than one study. Larger tumors
which appear late in a dog's life (mean 9.5 years) were more apt to be
malignant than ones that occur at an earlier age.
* * * DRAFT FINAL * * *
-------
16-2
In grey horses, incidence of melanoma may reach 80 percent if animals
survive to old age (Garma-Avina et al. 1981) but in other horses, only 1.6 to
4.8 percent of all tumors have been reported to be melanomas. The different
types of melanomas identified in dogs have not been recognized in horses.
Melanomas are rare in cattle (Garma-Avina et al. 1981), constituting less than
one percent of tumors. Unlike dogs, melanomas in cattle are found in nearly
equal numbers in young and old animals and most are considered benign.
Melanomas are also rare in cats, with an overall reported frequency of less
than one percent.
With the exception of Duroc-Jersey and Sinclair swine, melanomas are also
rare in pigs (Garma-Avina et al. 1981). In Sinclair swine, selective breeding
has increased the prevalence of cutaneous melanoma from the 11 percent
reported in 1968 to as high as 54 percent at birth in the progeny of two
affected pigs (Hook et al. 1979). In the Sinclair swine, the pigs we're
frequently born with melanoma, although most tumors appeared in the first year
of life. This is an interesting system in that melanomas, including any
metastases present, usually regress (Hook et al. 1982). Only rarely does the
melanoma kill the host. Although direct proof is lacking, the tumor
regression is probably immunologically mediated since the host inflammatory
response in regressing tumors is characteristic of cell-mediated immunity.
This is also supported by the observation that lymphocytes from swine with
melanoma are highly toxic to melanoma cells in vitro with patterns of
leukocyte reactivity that for the most part paralleled the patterns of in vivo
tumor growth and regression (Berkelhammer et al. 1982). The histopathology of
these tumors is similar to human tumors and, likewise, metastatic disease, is
correlated with deeply invasive tumors. Melanocytic tumors in Duroc swine
also demonstrate spontaneous regression and rare metastases and the occurence
is as high as 50 percent of offspring of two affected pigs (Hordinsky 1985).
However, no studies have been reported examining the effects of UVR or of
sunlight on the growth of these tumors.
The experimental induction of melanomas was first reported in the late
1930's and early 1940's when a great many studies were done examining the
effects of carcinogenic tars and purified polycyclic aromatic hydrocarbons in
laboratory animals. In 1938, Passey reported the induction of three tumors,
all melanomas (one was histologically malignant), in the skin of dogs after
six to seven years of tarring. However, the latent period of the tumors was
extremely long and this did not prove to be a valuable animal model for the
induction of melanomas. Further work concentrated on three animal species,
all of which are susceptible to induction of melanoma: the hamster, the guinea
pig, and the mouse. The following section will discuss melanoma induction in
these three species, beginning with the hamster.
Experimental induction of melanomas in hamsters was reported in 1956 by
Delia Porta et al. who investigated the production of melanotic lesions by a
single dose of 7,12-dimethylbenzanthracene (DMBA) painted on the skin of
Syrian golden hamsters. These tumors were located in the dermis and
subcutaneous tissues and were sharply demarcated from surrounding lesions.
These tumors were not considered malignant because there were no signs of
invasion or metastasis even when tumors reached a diameter of 4 cm and even
* * DRAFT FINAL * * *
-------
16-3
though efforts to transplant the tumor as fragments were successful in two
cases. Repeated application of DMBA, rather than a single dose, gave rise to
papillomas and carcinomas rather than melanomas.
Although the original studies were done in brown hamsters, Rappaport et
al. (1961) and Nakai and Rappaport (1963), produced tumors in white
hamsters. Tumors in white hamsters were sometimes amelanotic, were reported
to contain neural elements, and were clinically benign, in contrast to
spontaneous hamster melanomas which are malignant and do not contain neural
elements.
Bernfeld and Hamburger (1983) published a comprehensive study detailing
the kinetics of the appearance of DMBA-induced melanocytic tumors in
genetically-defined hamsters. Most of the work was done in F10 Alexander
hybrids. They reported that hamsters were susceptible to the induction of
melanocytic tumors with DMBA but that benzo(a)pyrene and 3-methylcholanthrene
did not induce melanocytic tumors. Promotion of DMBA induced melanocytic
lesions with TPA was not reported to increase the number or decrease the
latent period of melanocytic tumors induced in hamsters. It was stated that
DMBA-induced tumors greater than 2 mm in diameter were usually malignant
melanomas which "occasionally" metastasized, although no systematic autopsies
were reported. The number of weeks required for 95-100 percent of the
hamsters to develop melanomas was dose dependent, reaching a plateau of 9-10
weeks at 100 mg/hamster. At 10 mg, the time required to develop melanomas was
14 weeks, and at 3.33 mg, 26 weeks. The average number of melanomas per
hamster was also dose dependent, reaching a plateau at 3.33 mg DMBA per
hamster. There.were wide variations in susceptibility of various strains of
hamsters, to DMBA induction of melanomas; however, differences were not related
to coat color. Different strains of white hamsters and agouti hamsters were
among both the most and the least sensitive.
Melanocytic tumors can also be produced by a single intragastric dose of
DMBA (Goerttler et al. 1982). The yield of melanocytic tumors could be
approximately doubled by painting with TPA. Histopathology of the tumors in
this study was not reported.
Another carcinogen, urethan (ethyl carbamate), which is not a
polycyclicaromatic hydrocarbon (PAH), also induces melanocytic tumors in
hamsters (Pietra and Shubik 1960; Toth et al. 1961) when introduced into the
drinking water but not following cutaneous application. Non-melanoma tumors
also appeared in the forestomach of animals treated both topically and
systemically with urethan.
Vesselinovitch et al. (1970) induced malignant melanomas in Syrian white
hamsters with urethan injected intraperitoneally in neonates. Treated animals
developed melanomas in 62 percent of males and 40 percent of females. Unlike
the melanocytic tumors induced by skin painting with DMBA, these tumors were
much more metastatic. Fifty percent of tumor-bearing animals of both sexes
had lymph node metastases and 82 percent of males and 25 percent of females
had distant metastases. No systematic description is given of the
histopathology of the primary tumors but it is evident that those which
metastasized were clearly malignant. No information is given on the number of
melanomas per animal.
* * DRAFT FINAL * *
-------
16-4
Therefore, melanocytic tumors induced in hamsters by topical application
of DMBA are usually not malignant by the criteria of metastasis. However, the
tumors are usually multiple and can sometimes be transplanted into suitable
recipients. Therefore, as a model for human malignant melanoma, this system
has severe limitations. Malignant melanomas induced with urethan in hamsters
seem to be biologically more relevant to human studies in that the tumors are
malignant and do metastasize. However, there are no reported studies on the
developmental biology of these tumors and thus the similarity to human
melanomas is not known.
Melanocytic tumors produced by DMBA in guinea pigs have been reported to
be more frequently malignant than those similarly induced in hamsters.
Berenblum (1949) reported one of ten spotted guinea pigs painted with DMBA
developed a melanoma on pigmented (brown) skin which invaded extensively and
had lymph node metastases. Animals injected, rather than painted, with DMBA
did not develop melanoma, a finding which is identical in all the species
which are susceptible to induction of melanocytic tumors via cutaneous
application of DMBA. Edgcomb and Mitchelich (1963) also painted guinea pigs
of various colors and reported that four of 20 female guinea pigs developed
invasive melanomas, three of which metastasized widely. The melanomas were
said to be surrounded by nevi but no histological description is given of the
nevi.
Pawlowski et al. (1976) produced junctional and compound nevi in albino
guinea pigs by painting with DMBA. Pigmented spots appeared in 40 of 70
animals and the number of spots per animal increased over time. None
regressed but no signs of malignancy such as invasion or metastases were
reported. Focal incontinuities were seen in the dermoepidermal junction in
nevi excised six months after their appearance (at the end of the study) and
it is possible that if the experiment had been carried further, invasion
and/or metastasis would have occurred.
In another sequential study, Clark et al. (1976) chose the Weiser-Maple
guinea pig, a pigmented animal with a uniform coat color. Lesions were
excised or biopsied periodically from the fifth to the 130th week of
painting. The results showed that the lesions progressed from focal, small,
well-demarcated melanocytic tumors, through increasing size and cellular
atypias, to a stage Clark et al. call a malignant melanoma with "intralesional
transformation", deep invasion, and frequent and progressive lymph node
metastasis. Intralesional transformation was described as showing clusters of
cells which were nearly pigment free and which upon occasion showed
pleomorphism and mitotic figures. Stages having small, non-progressively
growing lymph node metastases were not termed malignant. Only four to five
percent of animals bearing melanotic lesions demonstrated intralesional
transformation. These authors stated that the developmental biology of the
early stages of human and guinea pig melanomas is quite different but both go
through well defined histological stages.
Thus skin painting of guinea pigs with DMBA does lead to the induction of
malignant melanomas, albeit with low incidence. However, although sequential
painting of hamsters resulted in the induction of carcinomas and papillomas
rather than melanocytic tumors, melanoma induction in guinea pigs is
reportedly increased by repeated skin painting. There are no reports in the
* * * DRAFT FINAL * *
-------
16-5
literature reporting on investigations of the interaction of UVB.on'
melanocytic tumors in the guinea pig.
One of the first laboratory animals in which melanomas were induced was
mice. As early as 1940, Badger et al. reported the production of melanomas in
an undefined strain of mice by the repeated application of 5,9,10-trimethyl-
1,2-benzanthracene (TMBA) in benzene. Hartwell and Stewart (1942) painted
DBA, C57 Black, I, C57BLxDBA, and DBAxC57BL mice with TMBA in benzene twice
weekly for ten months. There were more non-melanoma tumors than melanomas in
all groups and none of the albino mice (I) developed melanomas. However, the
pigmented mice developed melanomas and the susceptibility was strain dependent
with C57BL x DBA and DBA x C57BL mice the most susceptible followed by C57
and then DBA. The pigmented foci induced by the TMBA appeared to be benign
collections of melanocytes in the subcutaneous tissue near the epidermis and
were not considered malignant. Burgoyne et al. (1949) also painted C57BL,
DBA, and their Fl-hybrids with TMBA twice weekly up to 16 months. Although
the incidence of pigmented foci was somewhat less than observed by Hartwell
and Stewart (1942), the same relative susceptibility of the strains was
observed. Only one invasive malignant melanoma was found in a female DBA
after 221 days of painting.
Klaus and Winkelmann (1965) painted Mayo pigmented hairless mice with
DMBA. Those hairless mice which were more pigmented, called "dark" in the
study, developed numerous pigmented lesions in seven of ten animals, beginning
eight to ten weeks after the initial application of carcinogen. The tumors
were in mid-dermis, no metastases were seen, and there was no cytological or
clinical evidence of malignancy.
Recently two groups have repeated some of the work of Hartwell and Stewart
(1942) and have induced pigmented lesions on C57BL mice with DMBA.
Berkelhammer et al. (1982) used a single application of DMBA to the scapular
area of newborn C57BL/6 mice followed by applications of croton oil. Two
melanomas appeared in female littermates, JB/MS and JB/RH. Both have
exhibited metastases upon transplantation. Takizawa et al. (1985) painted
seven week old female C57BL/6, DBA/2, BALB/c x DBA (BDF ) and C57BL x DBA
(BDF ) mice once with DMBA and thereafter with croton oil thrice weekly for
two years. Small black macules appeared in C57BL, BDF , and CDF. mice
after about 25 weeks. Macules over 2 mm were called malignant melanomas and
in addition 125 of 206 lesions were examined histologically. Again, the
BDF mice were more susceptible to induction of melanocytic tumors than
CDF. and C57 mice respectively. Pulmonary metastases were noted from one
BDF tumor, although lymph node metastases were not mentioned.
Thus melanocytic tumors and malignant melanomas can be induced by skin
painting with DMBA in mice, although the susceptibility is highly strain
dependent and appears not to occur in albino mice.
Herlyn et al. (1986) transplanted human cutaneous nevi onto nude mice.
Over a period of two to three months the nevi were found to change such that
the number of nevomelanocytes and the tendency of the nevic cells to form
nests decreased. When transplanted human nevi were treated with DMBA, most
* * * DRAFT FINAL * *
-------
16-6
nevomelanocytes showed signs of hypertrophy and in four of the nine specimens
treated with DMBA for greater than 80 days, there were atypical enlarged
nuclei with mitotic figures. Therefore, DMBA is capable of affecting the
growth and differentiation of human melanocytes as well as melanocytes of
laboratory animals.
Occurrence of ultraviolet light induced melanomas. The interaction of
UVR and melanomas has only rarely been investigated in animal studies. UVR of
mice in the absence of previous chemical induction results in papillomas,
squamous cell carcinomas, and spindle cell sarcomas rather than melanocytic
tumors. Melanomas are extremely rare if they occur at all.
However, there is some evidence from animals studies that UVR can
influence the growth and progression of melanocytic tumors in mice. Benign
blue nevi were produced on the backs of hairless mice by Epstein et a'l. (1967)
with a single application of DMBA. Thirteen months after DMBA application,
mice were irradiated with a lamp most of whose UVR was in the UVB range. In
five of eleven mice surviving to five months after irradiation, invasive
melanomas developed. There were apparent lymph node metastases but no
evidence of distant metastases. Transplantation was unsuccessful; however,
the mice were not inbred. Mice receiving DMBA alone, UV alone, or no
treatment did not develop melanomas. The conclusion reached was that UVB
resulted in the production of melanomas from the benign melanocytic tumors.
Ultraviolet light was also implicated in the production of a single
melanoma in one of 40 C3H mice given ten UVR treatments over a two week period
followed by twice weekly painting with croton oil (Kripke 1979). The tumor
arose in the 92nd week and was named K1735. Bilateral metastases were seen in
the lymph nodes although no distant metastases were found. The tumor was
transplantable and a lung metastasis was found in the third transplant
generation in an immunosuppressed mouse. The original tumor arose in an area
of hyperpigmentation and is very interesting since there are no. reports in the
literature of spontaneous melanomas in C3H mice. Tumors induced in haired
mice by UVR are usually squamous cell carcinomas and fibrosarcomas.
The relationship of spontaneous and induced melanocytic tumors in animals
to melanoma in humans is unclear. In most animal models, the rate at which
metastasizing tumors of melanocyte origin are induced is quite low. For
example one mouse in 90 (Takizawa et al. 1985), two mice in 20 (Berkelhammer
et al. 1982), and four to five percent of treated guinea pigs (Clark et al.
1976) developed metastatic tumors of melanocytic orgin. The rate of
metastatic melanomas produced by urethan in white Syrian hamsters is much
higher but no attempt has been made to examine the developmental biology of
these tumors. Additional difficulty is caused by the lack of agreement on
terminology in the animal tumors. For example, authors frequently neglect to
clearly state by what criteria a tumor is deemed malignant. Some authors
designate well-circumscribed groups of melanocytes melanomas, whereas some
others consider tumors to be malignant if they are invasive, and others only
call tumors with a relatively heavy metastatic load malignant. This adds to
the confusion and makes comparison of studies difficult, but it is clear that
melanomas can be induced by chemicals (DMBA, TMBA, urethane) in hamsters,
guinea pigs, and mice both by skin painting (DMBA, TMBA) and by systemic
administration (urethane, DMBA).
* * * DRAFT FINAL * *
-------
16-7
The relationship of UV light to the development of melanomas has
apparently been addressed in two studies. Epstein et al. (1967) clearly
induced growth and progression of chemically-induced nevi by treating those
nevi with UVR. In addition, UVR is also strongly implicated in the induction
of the K1735 melanoma in C3H mice.
Therefore, although the effect of UVR on the majority of animal models of
melanoma has not been measured, two studies Suggest that UVR may play a role
in melanoma induction under the right experimental circumstances. Because
they focus primarily on chemical carcinogens, most studies on melanoma in
animals neither support nor refute the hypothesis that UVR is at least
partially responsible for induction of human melanoma. Studies showing that
UVR can cause progression of chemically-induced melanocytic tumors lead to
speculation that a similar process may occur in humans. However, we have no
direct evidence that this occurs in humans and only limited data from'animal
studies.
UV-INDUCED CARCINOGENESIS IN RODENTS
The association of sunlight with human skin cancer has been discussed
since the beginning of the 20th century. The observations that men who had
outdoor occupations seemed to suffer more skin cancer prompted Findlay to look
experimentally for evidence that UVR was carcinogenic. In 1928, Findlay
reported that not only did UVR alone induce skin tumors in mice, but tumors
induced by tar appeared more rapidly if those mice were subsequently exposed
to UVR.. The tumors induced in mice by UVR are primarily squamous cell
carcinomas and fibrosarcomas (Epstein and Epstein 1962; Hsu et al. 1975;
Kligman and Kligman 1981; Kripke 1977; Spikes et al. 1977; Stenback 1975;
Strickland et al. 1979; Winkelman et al. 1963) and are mostly monoclonal in
origin (Burnham et al. 1986). The following section will'discuss the action
spectrum of UVR in tumorigenesis in laboratory animals., strain and genetic
differences in susceptibility, and some other factors which have been found to
affect the induction of tumors in laboratory animals by UVR.
Subsequent studies expressed interest in determining which wavelengths
were responsible for the carcinogenicity of UVR. Unfortunately there is a
lack of agreement between many of the studies. Early studies used filters to
remove shorter wavelengths from broadband UVR so that the effect of different
wavelengths could be studied. In one of the original studies on the
carcinogenic action spectrum of UVR, Rusch et al. (1943) reported that the
carcinogenic wavelengths lay between 290 nm and 334 nm; wavelengths greater
than 334 nm and those at 254 nm were not found to be carcinogenic. However, a
germicidal lamp, which emits at least 90 percent of its UVR at 254 nm, was
found to be carcinogenic in mice (Lill 1983) and rats (Strickland et al.
1979). A potential explanation for this difference in results is that the
dose of UVR to mice in the second experiment was 3 x 10* J/m2/wk whereas
in that of Rusch et al. (1943) it was approximately 2 x 103 J/m2/wk, a
15-fold difference.
Blum (1943) reported that nearly comparable doses of broadband UVR were
less effective in producing tumors when wavelengths of less than 297 nm were
removed by filters. A dose of 8.8 x 10" J/m2 from the unfiltered lamp
resulted in a median tumor latent period of 167 days, whereas filtration to
* * * DRAFT FINAL * * *
-------
16-8
remove wavelengths of less than 297 nm reduced the dose to 8.0 x 10* J/m2
and increased the median latent period to 300 days. Upon calculation of the
dose which penetrated below the epidermis, the discrepency is even greater.
However, removal of wavelengths below 265 nm, which reduced the incident dose
to 4.1 x 10" J/m2 but only reduced the UVR which reached below the
epidermis from 1.3 x 10* J/m2 to 1.0 x 10* J/m2, resulted in a median
latent period of 187 days, not appreciably different than the unfiltered
lamp. Therefore, he concluded that the most effective wavelengths in
producing tumors in mice lay in the region between 260 nm and 300 nm.
Strickland et al. (1979) also compared.the tumor yield from UVR at 254 nm to
that of a lamp emitting UVA+UVB and found that the UVA+B lamp was much more
effective in producing tumors. However, if the dose of UVR was corrected for
penetration of the statum corneum, then the two lamps were very similar.
Additional experimentation was done using monochromatic UVR by Freeman
(1975). He compared doses of UVR with an equivalent minimal erythemal dose
(MED), using data from human skin as a guide. Therefore, the ears of albino
mice were exposed to a weekly dose of 420 J/m2 at 290 nm, 600 J/m2 at 300
nm, 7500 J/m2 at 310 nm, and 49,500 J/m2 at 320 nm. Using these doses,
the mice receiving UVR at 300 and 310 nm developed tumors with the same median
latent period. Only 2 of 5 mice receiving radiation at 320 nm developed
tumors and no mice irradiated at 290 nm developed tumors. When mice were
given the same incident radiation at 300 and 310 nm, no mice given radiation
at 310 developed tumors. Therefore, the carcinogenic effectiveness of UVR at
300 nm is greater than at 310 nm. This finding is also paralleled by in vitro
studies which show that above a peak at about 260-265 nm, the shorter the
wavelength, the greater the effectiveness of cell transformation (Suzuki et
al. 1981). However, the relative lack of effectiveness at 290 nm is somewhat
surprising and does not correlate with in vitro findings. But, Cole (1981)
has published extensive studies on the action spectrum of acute skin damage to
mouse skin and has found that UVR at 290 nm is less effective in producing
acute skin damage than UVR at 300 nm. In other words, there is a peak in the
erythema action spectrum at 300 nm. Given this information, to use an
equivalent MED in these experiments, Freeman would have had to use UVR at 290
nm at a higher dose than at 300 nm, rather than the lower dose which he did
use. This may explain why, if the hypothesis is correct that tumorigenicity
is proportional to erythemal effectiveness, no tumors were produced at 290 nm
and would explain the discrepancy between this report and that of Blum
discussed above. This is substantiated by a recent report from the same
laboratory discussed below (Cole et al. 1986).
Forbes et al. (1982) addressed the question of the carcinogenic action
spectrum with a study designed specifically to test the effect of simulated
stratospheric ozone depletion on photocarcinogenesis in hairless mice. They
used filters to remove increasing amounts of UVR so that the effects of
varying the dose-rate on photocarcinogenesis could be determined. However,
this also has the effect of incrementally removing shorter wavelengths of
UVR. They also tested the ability of the R-B sunburn meter to predict
tumorigenicity: Results of the experiments showed that the tumorigenicity of
UVR could be correlated with its erythemal effectiveness. More important for
this discussion is the finding that the R-B sunburn meter underestimated the
carcinogenic effectiveness of the shorter wavelengths of UVR. This is
particularly important for the decision to regulate CFs since it is the
* * DRAFT FINAL * * *
-------
16-9
shorter wavelengths of UVR which will increase most with decreasing
stratospheric ozone. Cole et al. (1986) reported on determination of the
carcinogenic action spectrum using three sources of UVR and a series of
filters to further produce source spectra. The tumorigenic effects of these
sources were then analyzed mathematically using an equation which incorporated
the spectral source description and a weighting function dependent on the
action spectrum for acute skin edema in hairless mice. The equation assumed
no effectiveness from wavelengths greater than 330 nm and the results
supported the conclusion than the relative carcinogenic effectiveness of
radiation greater than 330 nm was less than 0.02 percent of that at 297 nm.
There was no evidence for wavelength interaction in the spectral range of from
260 to 400 nm. The group examined their data using several weighting
functions. Using an averaged DNA action spectrum in the plotting of data of
tumorigenesis resulted in overweighting of the importance of the shorter
wavelengths. However, use of the R-B sunburn meter to weight the relative
importance of the wavelengths underestimated the contribution of the shorter
wavelengths. However, the relative effectiveness of UVR in inducing edema 48
hours after a single acute dose of UV (MEE,„) proved to be the best of the
three tested predictors of source effectiveness in tumorigenesis. The
investigators state that their results are consistent with many previously
reported papers but are specific for certain parameters, such as the strain of
animal used, irradiation geometry, and time-dose reciprocity. In addition,
the mathematical relationship does not hold in linear fashion for low doses of
UVR. It would predict a 50 percent tumor incidence at 0 dose in 70 weeks.
The relationship for predicting 50 percent tumor incidence fails at
approximately 110 J/ra2 (MEE,., weighted, dose). Thus, the mathematical
relationships described here will not serve to predict tumor yield in all
cases but supply very valuable insights into the relative donation of efficacy
of the various wavelengths found in UV sources.
UVA has also been reported to cause tumors in mice, albeit in very large
doses. For example, van Weelden et al. (1983) reported that to result in
tumors with approximately the same median latent period, it required
approximately 3000 times greater incident energy for UVA compared to UVB.
More important for this discussion, there has been one report in the
literature that UVA augments the carcinogenic effects of UVB (Willis et al.
1981) both when a constant daily dose of UVR is given and when doses which
escalate weekly were used. In addition, Staberg et al. (1983) reported that
mice irradiated with UVB for 3 months followed by irradiation with UVA had
greatly increased tumor incidence. Therefore, although UVA is in and of
itself a poor carcinogen, it is reported here to interact with UVB in an as
yet undefined manner to increase the carcinogenic effects of UVB, although
these results are not in agreement with those of Cole et al. (1986).
There are significant genetic differences in susceptibility to UV-induced
carcinogenesis, both among different strains of the same species and among
different species. Of the experimental animals tested, mice are the most
sensitive to the carcinogenic effects of UVB. Stenback (1975) compared the
effects of UVR in mice, rats, guinea pigs, and hamsters. He found that, using
the same protocol of repeated UV exposure, 40 percent of rats and 50 percent
of mice developed tumors whereas 35 percent of hamsters but only 12.5 percent
of guinea pigs developed tumors. However, in the mouse, 70 percent of the
tumors were malignant whereas in the rat only 24 percent were malignant, in
* * * DRAFT FINAL * * *
-------
16-10
the hamster only three percent, and in the guinea pig'none were malignant.
Therefore, although mice, hamsters, and rats are similarly sensitive to the
tumorigenic effects of UVB, mice develop more malignant tumors than the other
species. Tumors can be induced in both mice (Hsu et al. 1975) and rats
(Strickland et al. 1979) following a single dose of UVB, although most of the
tumors induced were benign. In addition, in mice, 21 percent of the tumors
spontaneously regressed whereas in rats only about five percent of the tumors
regressed.
It is difficult to compare the induction of tumors from one laboratory to
another since lamps vary and the method of measuring the UVR may vary.
However, from doses published in the literature, the hairless mouse (Kligman
and Kligman 1981; Winkelmann et al. 1963) seems to be equally sensitive to the
carcinogenic effects of UV light as haired mice (Kripke 1977; Spikes et al.
1977). There are strain differences; among three inbred.strains of mice,
BALB/c mice were more sensitive to UVR that either C3H or C57BL (Kripke 1977)
but the median tumor latent period is not influenced by the major
histocompatibility complex (Roberts et al. 1984). Sencar mice, which are
extremely susceptible to two-stage skin carcinogenesis by chemical
carcinogens, are also hypersensitive to UVR (Strickland 1982). Huepner (1941)
reported that hairless rats were less susceptible to UV carcinogenesis than
their haired littermates.
Although the yield of tumors is directly related to the cumulative dose,
(de Gruijl et al. 1983) there has been a great deal of experimentation to
determine if there is dose-rate reciprocity in UV carcinogenesis. Blum et al.
(1942) studied the effect of the intensity of the dose on the tumor latent
period and reported that for a 10-fold dose range of between 4.3 J/m2/sec
and 0.42 J/m2/sec, no significant differences were found. In a second study
using a wider range of intensities and greater numbers of animals he reported
that above 0.4 J/m2/sec the effectiveness of UVR in .inducing tumors .is
almost independent of dose-rate, but below that the effectiveness falls off
rapidly with intensity. Strickland et al. (1979) reported that the UVR
dose-rate dependence of tumor yield in hairless rats was linear for 254 run.
By comparison, exposure from a lamp emitting both UVA and UVB, resulted in
more tumors per unit exposure at lower intensities than at higher
intensities. These authors found that the total dose of UVR to produce a
tumor in hairless rats was less at low doses than at high doses of UVR and
suggested that oncogenesis and cell lethality might be competing events at
high doses. When the tumor yield per dose of UVR was corrected for lethality
by measuring survival of follicles, the dose-response curve for tumorigenesis
for the UVA+B lamps increased its linearity. Additional correction for
penetration of the epidermis resulted in the dose-response curves for UVC and
UVA+B becoming nearly coincident.
Forbes et al. (1981) examined the effect of dose fractionation on tumor
induction and reported that increasing the fractionation of the dose (more
exposures to yield the same total dose) increased the effectiveness of the
protocol. He reported that the effectiveness was related directly to the
number of exposures per week, assuming that the total delivered dose per week
is kept constant (Forbes 1981).
* DRAFT FINAL *
-------
16-11
One can also calculate the amount of energy to produce tumors in 50
percent of animals to determine if the dose rate affects tumor development.
De Gruijl et al. (1981) stated that the total dose delivered to hairless mice
to induce tumors must be greater if a high daily dose is given than if a low
daily dose is given. The greatest dose given was 9.4 x 103 J/m2/wk, total
output of lamps. The same trend was reported by Spikes et al. (1977) in C3H
mice. They calculated the average tumor latency as dose of UVR received by
the mice at the time that 50 percent of the irradiated mice had developed
tumors. For mice receiving UV doses three times per week of 60 sec (2220
J/m2) the average tumor latency was reached at 29 x 10* J/m2 whereas for
mice receiving doses three times per week of 2 sec (74 J/m2) the average
tumor latency was obtained at 2.1 x 10** J/m2. Therefore the amount of
energy required to produce tumors was much less at low doses than at high
doses.
There are many factors which serve to modify the carcinogenic response to
UVR. One factor is the immune status of the irradiated animal. Unfortunately
there are no simple trends and one cannot state that immunosuppressed animals
will be more susceptible to photocarcinogenesis than animals with normal
immune functions. Immunsuppression by administration of rabbit anti-mouse
lymphocyte serum enhanced photocarcinogenesis where administration of
6-mecatopurine appeared to inhibit photocarcinogenesis (Nathanson et al.
1976). Treatment of mice with silica, known to decrease the number of
peritoneal macrophages, increased the susceptibility of mice to UV-induced
carcinogenesis when the dose of UV was abbreviated (Norbury and Kripke 1979)
but not when a. full course of UV treatment was given. Conversely, pyran
copolymer, which increased the number of peritoneal macrophages and increased
resistance to a transplated tumor, also afforded protection against
photocarcinogenesis using an abbreviated dose of UVR. Norbury and Kripke
(1978) also studied the effect of T-cell depletion on UVR carcinogenesis.
Although mice which were T-cell depleted by adult thymectomy, lethal
X-irradiation, and reconstitution with neonatal liver cells were more
susceptible to UVR than control mice, the picture is complicated by the
finding that the T-cell depleted mice, when reconstituted with thymic grafts,
were even more susceptible than T-cell depleted mice to photocarcinogenesis.
In addition, there were differences in the proportions of squamous cell
carcinomas and fibrosarcomas in the two groups. Their finding of increased
numbers of squamous cell carcinomas following whole-body radiation is
interesting in the light of the reports of increased risk for skin carcinomas
in psoriasis patients undergoing PUVA (psoralen-UVA) treatment only in those
patients who had received whole body radiation (Stern et al. 1979). The
ability of UVR itself to alter the immune response to a tumor, particulary
UV-induced tumors, has been discussed in a separate section in this chapter.
The age of animals has also been investigated as a factor in ultraviolet
carcinogenesis, particularly since sunlight-induced tumors in man usually
appear in older persons. However, when both young and old mice are treated
with a constant dose of UVR, young mice are more sensitive to the effects of
UVR than are old mice (Blum et al. 1942; Forbes et al. 1981). Ebbesen and
Kripke (1982), using skin grafting as a method for separating the effects of
the age of the skin and the age of the host, found that when skin was grafted
and then UV irradiated, tumors developed on skin when it was grafted onto
younger mice more readily than onto older mice, regardless of the age of the
* * * DRAFT FINAL * * *
-------
16-12
skin donor mice. Therefore the finding that sunlight-induced tumors in humans
usually occur in older persons may be due to accumulated dose rather than
increased susceptibility of the aged.
The actual dose of UVR required to reduce tumors in 50 percent of
irradiated mice varies tremendously from one report to the next. Obviously
there are differences in genetic backgrounds of the animals, dose-rates of UVR
employed; schedules of UVR used (daily versus three times per week), and the
spectral power distribution of the lamps used. We have seen that all these
factors play a role in determining the tumorigenicity of UVR in laboratory
animals. Therefore, making comparisons in an attempt to arrive at a
generalized dose-response is not productive. For example, Swiss albino mice
developed no back tumors after 67 weeks of irradiation with a total dose of
1.3 x 101* J/m2 (Epstein and Epstein 1962) but albino mice developed ear
tumors in 50 percent of the group in 46 weeks after 1.76 x 10s J/m2 of UVR
(Blum et al. 1942) and Kripke (1977) reported a 50 percent tumor incidence
after 20 weeks of UVR with a cumulative UV dose of 6 x 105 J/m2 in BALB/c
mice.
FINDINGS
None of the extensive information available on photocarcinogenesis in
laboratory animals has any direct bearing on the question of the role of UVR
in human melanoma. However, there are certain findings which may be used to
develop predictions and hypotheses.
16.1 It is clear that'UVR is carcinogenic and that UVB wavelengths are -
most effective. Although the tumors induced by UVB in laboratory
animals are not melanomas, it is quite clear that UVB is capable
of causing malignant change in both fibroblasts and epithelial
cells .in vivo.
16.2 It is also clear that the shorter wavelengths of UVB, those that
would be increased by decreasing levels of ozone in the
atmosphere, are more carcinogenic than the longer wavelengths.
16.3 There are great genetic differences in the susceptibility to
photocarcinogenesis, not all of which can be ascribed to
differences in pigmentation.
16.4 There are a wide variety of factors which can influence UV
carcinogens is including dose rate, humidity and temperature.
16.5 There is as yet no appropriate animal model in which UVR has
consistently produced melanoma.
* * * DRAFT FINAL * * *
-------
16-13
REFERENCES
Berenblum, I. The carcinogenic action of 9,10-dimethyl-1,2-benzanthracene on
the skin and subcutaneous tissues of the mouse, rabbit, rat, and guinea pig.
JNCI 10:167-74 (1949).
Berkelhammer, J., Ensign, B.M., Hook, R.R., Hecker, C.J., Smith, G.D.,
Oxenhandler, R.W. Growth and spontaneous regression of swine melanoma:
Relationship o.f in vitro leukocyte reactivity. JNCI 68:461-486 (1982).
Berkelhammer, J., Oxenhandler R.W., Hook, R.R., Hennessy, J.M. Development of
a new melanoma model in C57BL/6 mice. Cancer Res 42:3157-3163 (1982).
Bernfeld, P., Hamburger F. Skin painting studies in Syrian hamsters. Prog
Exp Tumor Res 26:128-153 (1983).
Blum, H.F. Wavelength dependence of tumor induction by ultraviolet
radiation. 3:533-537 (1943).
Blum, H.F., Grady, H.G., Kirby-Smith, J.S. Relationships between dosage and
rate of tumor induction by ultraviolet radiation. JNCI 3:91-97 (1942).
Blum, H.F. and Lippincott, S.W. Carcinogenic effectiveness of ultraviolet
radiation of wavelength 2537A. JNCI 3:211-216 (1942).
Burgoyne, F.HI, Heston W.E., Hartwell J.L., Stewart, H.L. Cutaneous melanin
production in mice following application of the carcinogen 5,9,10-
trimethy1-1,2-benzanthracene. JNCI 10:665-687 (1949).
Burnham, D.K., Gahring, L.C., Daynes, R.A. Clonal origin of tumors induced by
ultraviolet radiation. JNCI 76:151-158 (1986).
Clark, W.H., Jr., Min, B.H., and Kligman, L.H. The developmental biology of
induced malignant melanoma in guinea pigs and a comparison with other
neoplastic systems. Cancer Res 36:4079-4091 (1976).
Cole, C.A. Interspecies comparison of action spectra for acute skin responses
following ultraviolet radiation: Mouse vs. human. Thesis The Graduate
School of Temple University Philadelphia, PA (1981).
Cole, C.A., Forbes, D., Davies, R.E. An action spectrum for UV
photocarcinogenesis. Photochemistry and Photobiology 43:275-284 (1986).
De Gruijl, F.R., van der Meer, J.B., van der Leun, J.C. Dose-time dependency
of tumor formation by chronic UV exposure. Photochem Photobiol 37:53-62
(1983).
Ebbesen, P., Kripke, M.L. Influence of age and anatomical site on ultraviolet
carcinogenesis in BALB/c mice. JNCI 68:691-694 (1982).
* * * DRAFT FINAL * *
-------
16-14
Edgcomb, J.H., Mitchelich, H. Melanomas of the skin of guinea pigs following
the application of a solution of 9,10-dimethyl-l,2-benzanthracene in benzene.
Acta Unio Int Contra Canerum 19:706-707 (1963).
Epstein, J.H., Epstein, W.L., and Nakai, T. Production of melanomas from
DMBA-induced "blue nevi" in hairless mice with ultraviolet light. JNCI
38:19-30 (1967).
Epstein, J.H., Epstein, W.L. Cocarcinogenic effects of ultraviolet light on
DMBA tumor initiation in albino mice. J Invest Derm 39:455-460 (1962).
Findlay, G.M. Ultra-violet light and skin cancer. Lancet ii:1070-1073 (1928).
Forbes, P.D. Photocarcinogenesis: An overview. J Invest Derm 77:139-143
(1981).
Forbes, P.D., Blum, H.F., Davies, R.E. Photocarcinogenesis in hairless mice:
dose-response and the influence of dose-delivery. Photochemistry and
Photobiology 34:361-365 (1981).
Forbes, P.O., Davies, R.E., Urbach, F., Berger, D., Cole, C.. Simulated
stratospheric ozone depletion and increased ultraviolet radiation: Effects of
Photocarcinogenesis in hairless mice. Cancer Res 42:2796-2803 (1982).
Freeman, R.G. Data on the action spectrum for ultraviolet carcinogenesis.
JNCI 55:1119-1122 (1975).
Garma-Avina, A., Vallie E., Lumsden J.H. Cutaneous melanomas in domestic
animals. J Cutan Pathol 8:3-24 (1981).
Goerttler, K., Hecker, E., Loehrke, H., Seip, H., Hesse, B., and Schweitzer,
J. Effect of the tumor promoter 12-0-tetradecanoylphorbol-13-acetate and its
nonpromoting analogue 4-0-methyl-TPA on dorsal dermal melanocytes of the
Syrian golden hamster (Mesocricetus auratus). J Cancer Res Clin Oncol
103:305-11 (1982).
Hartwell, J.L., Stewart, H.L. Action of 5,9,10-trimethyl-l,2-benzanthracene
on the skin of the mouse. JNCI 3:277-289 (1942).
Herlyn, D., Elder, D.E., Bondi, E., Atkinson, B., Guerry IV, D., Koprowski,
H., Clark Jr., W.H. Human cutaneous nevi transplanted onto nude mice: A
model for the study of the lesional steps in tumor progression. Cancer Res
46:1339-1343 (1986).
Hook, R.R., Aultman, M.D., Adelstein, E.H., Oxenhandler, R.W., Millikan, L.E.,
Middleton, C.C. Influence of breeding on the incidence of melanomas in
Sinclair miniature swine. Int J Cancer 24:668-672 (1979).
Hook, R.R., Berkelhammer J., Oxenhandler, R.W. Animal model of disease --
melanoma. Sinclair swine melanoma. Amer J Path 108:130-133 (1982).
* * * DRAFT FINAL * *
-------
16-15
Hordinsky, M.K., Ruth, G., King, R. Inheritance of melanocytic tumors in
Duroc swine. J of Heredity 76:385-386 (1985).
Hsu, J., Forbes, P.D., Harber, L.C., Lakow, E. Induction of skin tumors in
hairless mice by a single exposure to UV radiation. Photochem Photobiol
21:185-288 (1975).
Huepner, W.C. Cutaneous neoplastic responses elicited by ultraviolet rays in
hairless rats and in their haired litter mates. Cancer Res 1:402-406 (1941).
Klaus, S.N. and Winkelmann, R.K. Pigment changes induced in hairless mice by
dimethylbenzanthracene. J Invest Dermatol 45:160-167 (1965).
Kligman, L.H. and Kligman, A.M. Histogenesis and progression of ultraviolet
light-induced tumors in hairless mice. JNCI 67:1289-1297 (1981).
Kripke, M.L. Speculations on the role of ultraviolet radiation in the
development of malignant melanoma. JNCI 63:541-545 (1979).
Kripke, M.L. Latency, histology, and antigenicity of tumors induced by
ultraviolet in three inbred mouse strains. Cancer Res 37:1395-1400 (1977).
Lill, P.H. Latent period and antigenicity of murine tumors induced in C3H
mice by short-wavelength ultraviolet radiation. J Invest Derm 81:342-346
(1983).
Nakai, T. and Rappaport, H. Carcinogen-induced melanotic tumors in animals.
Natl Cancer Inst Monograph 10:297-322 (1963).
Nathanson, R.B., Forbes, P.D. and Urbach, F. Modification of
photocarcinogenesis by two immosuppressive agents. Cancer Let 1:243-247
(1976).
t
Norbury, K.C. and Kripke, M.L. Ultraviolet carcinogenesis in T-cell depleted
mice. J Natl Cancer Inst 61:917-921 (1978).
Norbury, K.C. and Kripke, M.L. Ultraviolet-induced carcinogenesis in mice
treated with silica, trypan blue or pyran copolymer. J Reticuloendo Soc
26:827-837 (1979).
Passey, R.D. Experimental tar tumors in dogs. J Pathol Bacteriol 47:349-351
(1938).
Pawlowski, A., Haberman, H.F., and Menon, I.A. Junctional and compound
pigmented nevi induced by 9,10-dimethyl-l,2-benzanthracene in skin of albino
guinea pigs. Cancer Research 36:2813-2821 (1976).
Pietra, G. and Shubik, P. Induction of melanotic tumors in the Syrian golden
hamster after administration of ethyl carbamate. JNCI 25:627-630 (1960).
* * DRAFT FINAL * *
-------
16-16
Rappaport, H., Pietra, G., and Shubik, P. The induction of melanocytic tumors
resembling cellular blue nevi in the Syrian white hamster by cutaneous
application of 7,12-dimethylbenz[a]anthracene. Cancer Res 21:661-666 (1961).
Roberts, L.K., Bernhard, E.J., and Daynes, R.A. Experimental ultraviolet
radiation carcinogenesis: I. Relationship of the major histocompatibility
complex to tumor latency and immunogenicity. Photoderm 1:57-64 (1984).
Rusch, H.P. and Baumann, C.A. Tumor production in mice with ultraviolet
irradiation. Amer J Cancer 35:55-62 (1939).
Rusch, H.P., Kline, B.E., and Baumann, C.A. Carcinogenesis by ultraviolet
rays with reference to wavelength and energy. Arch Path 31:135-146 (1943).
Shubik, P., Porta, G.D., Rappaport, H., and Spencer, K. A transplantable
induced melanotic tumor of the Syrian golden hamster. Cancer Res 46:1031-32
(1956).
Spikes, J.D., Kripke, M.L., Conner, R.J., and Eichwald, E.J. Time of
appearance and histology of tumors induced in the corsal skin of C3Hf mice by
ultraviolet radiation from a mercury arc lamp. JNCI 59:1637-1643 (1977).
Staberg, B., Wulf, H.C., Poulsen, T., Lemp, P., and Brodthagen, H.
Carcinogenic effect of sequential artificial sunlight and UV-A irradiation in
hairless mice. Arch Derm 119:641-643 (1983).
Stenback, F. Species-specific neoplastic progression by ultraviolet light on
the skin of rats, guinea pigs, hamsters and mice. Oncology 31:309-225 (1975).
Stern, R.S., Thibodeau, L.A., Kleinerman, R.A., Parrish, J.A., Fitzpatrick,
T.B. Risk of cutaneous carcinoma in patients treated with oral methoxsalen
photochemotherapy for psoriasis. New Eng J Med. 300:809-313 (1979).
Strickland, P.T. Tumor induction in Sencar mice in response to ultraviolet
radiation. Carcinogenesis 3:1487-1489 (1982).
Strickland, P.T., Burns, F.J., and Albert, R.E. Induction of skin tumors in
the rat by single exposure to ultraviolet radiation. Photochem Photobiol
30:683-688 (1979).
Suzuki, F., Han, A., Lankas, G.R., Utsumi, H., and Elkind, M.M. Spectral
dependencies of killing, mutation, and transformation in mammalian cells and
their relevance to hazards caused by solar ultraviolet radiation. Cancer Res
41:4916-4924 (1981).
Takizawa, H., Sato, S., Kitajima, H., Konishi, S., Iwata, K., and Hayashi, P.
Mouse skin melanoma induced in two stage chemical carcinogens is with
7,12-dimethylbenz[a]anthracene and croton oil. Carcinogenesis 6:921-923
(1985).
* * * DRAFT FINAL *
-------
16-17
Toth, B., Tomatis, L., and Shubik, P. Multipotential carcinogenesis with
urethan in the Syrian golden hamster. Cancer Res 21: 1536-1541 (1961).
van Weelden, H., de Gruijl, F.R., and van der Leun, J.C. Tumors induced by
UV-A in mice. Photochem Photobio 375:795 (1983).
Vesselinovitch, S.D., Mihailovich, N., and Richter, W.R. The induction of
malignant melanomas in Syrian white hamster by neonatal exposure to urethan.
Cancer Res 30:2543-2547 (1970).
Willis, I., Menter, J.M., and Whyte, H.J. The rapid induction of cancers in
the hairless mouse utilizing the principle of photoaugmentation. J Invest
Derm 76:404-408 (1981).
Winkelman, R.K., Zollman, P.E., and Baldes, E.J. Squamous cell carcinoma
produced by ultraviolet light in hairless mice. J Invest Derm 40:217-224
(1963).
* * * DRAFT FINAL * *
-------
CHAPTER 17
EFFECT OF ULTRAVIOLET RADIATION ON THE IMMUNE RESPONSE
AND ITS RELATIONSHIP TO CUTANEOUS MELANOMA
Ultraviolet radiation (UVR) has a variety of effects on the immune
system, many of which could affect the incidence or morbidity of melanoma
This chapter reviews information of the relevant UVR-induced effects on the
immune response (IMR) and explores the impact that UVR might have on melanoma
incidence and mortality mediated via the immune response. For a more general
discussion of the Immune Effects of UVR, refer to the appropriate chapter of
the Risk Assessment Document
The chapter is organized to first review the theory by which effects on
the immune system could have an impact on the carcinogenic process, then
briefly discusses the cell types relevant to the theory. Subsequent sections
present information on the experimental systems used to explore the role of
UVR on the IMR followed by information from studies in humans. A last section
discusses the relevance of this information to melanoma development.
Most studies have investigated the relationship between the effects of
UVR in the induction and growth of squamous cell carcinomas and fibrosarcomas
in mice, since these tumor types, are the ones most frequently induced in mice
by UVR. However, there is information available which relates the impact of
UVR on the immune system to its potential effects on the development of
melanomas.
IMMUNE SURVEILLANCE - A PROPOSED ROLE FOR THE IMMUNE SYSTEM IN
CARCINOGENESIS
The most widely quoted theory which discusses the impact that the immune
system may have on carcinogenesis is that of immune surveillance (Burnet,
1970). Burnet proposed that throughout life cells are transformed to a
malignant potential which is manifested via the expression of new antigens.
Furthermore, Burnet proposed that there exists a certain population of
lymphocytes which routinely "surveys" the body and kills those cells with
newly-expressed tumor antigens. This theory was the outgrowth of two
experimental findings. The first was the finding that tumors bear antigens on
their surfaces which were not normally expressed on the cells of an adult.
The second was that there existed a population of lymphocytes which, when
properly immunized, were capable of killing tumor cells. These lymphocytes
were found to mature in the thymus and thus were named thymus-derived
lymphocytes or T lymphocytes, frequently shortened to T cells. The immune
surveillance theory provided the stimulus for a great deal of research
directed at proving or disproving its validity. What has emerged is an
understanding that the interaction of the immune system with neoplastic cells,
although having elements of Burnet's idea, is not as simple as the originally
proposed.
* * * DRAFT FINAL * * *
-------
17-2
CYTOTOXIC LYMPHOCYTES - T CELLS AND OTHERS
Although there are T cells which can kill tumors both in vivo and in
vitro, the relationship between depression of T cell function and resistance
to tumorigenesis is complicated. For example, "nude" mice, which are
genetically deficient in T cells, are no more susceptible to chemical
induction of tumors and have no more spontaneous tumors than littermates with
normal levels of T cell activity (Stutman, 1975). Moreover, there are cells
of the immune system other than T cells which are also capable of killing
tumor cells. Two such other cell types are natural killer cells (NK cells)
and natural cytotoxic cells (NC cells). Both are lymphocytes which lack
either B or T cell markers and can kill tumor cells nonspecifically in the
absence of antibody. Killing by these cells is not immunologically specific
since tumor cells can be killed without prior immune sensitization of the
effector cells. It is sometimes difficult to distinguish the activities of
these two cells; as a consequence some authors refer to natural cell-mediated
cellular cytotoxicity (NCMC) which includes activities of both cell types.
Also, macrophages, if their intracellular killing mechanisms have been
activated by any of a variety of agents, can non-specifically kill tumor cells
but do not kill normal cells. To complicate matters, it has been found that
the precise mechanism of tumor cell control, (i.e., the cell or cells which
are responsible for control of tumor growth in vivo) differs widely from one
tumor system to another.
EXPERIMENTAL SYSTEMS
There are practical problems in directly assessing the effects of immune
alterations on tumor growth, especially in humans since one does not
transplant tumors into humans to see if they grow. However, researchers have
found that there is some relationship between the ability of an animal to
develop cell mediated immunity (contact hypersensitivity) such as is involved
in poison ivy reactions, and the ability of the animal to develop cellular
immunity to a tumor. The correlations are not perfect and there are
exceptions, but the concept has allowed immunolegists to examine closely many
of the mechanisms of the immune response which are necessary to the rejection
of a tumor.
A note should be made on the sources of UV radiation used for the
experiments to be discussed. Much of the research reported here on the
effects of UVB on the immune response has been performed with fluorescent sun
lamps such as the Westinghouse FS40 or FS20. These lamps emit about 60
percent of their energy in the UVB range but also emit small amounts of UVC
and significant amounts of UVA and visible light. Therefore, unless control
experiments are done with filtered light or a monochromatic source, one cannot
be certain that the effects are due specifically to UVB (Spikes, 1983). In
fact, most experiments are performed with polychromatic light sources and one
should assume, unless specifically noted otherwise in the discussion, that the
results may not be due solely to the effects of UVB irradiation.
* * DRAFT FINAL * * *
-------
17-3
CHARACTERISTICS OF IMMUNOSUPPRESSION IN UVR TREATED ANIMALS
One of the first reports which discussed the potential effects of UVR on
the immune response to UVR-induced primary tumors was that of Kripke (1974)
who reported that most murine tumors induced by UVR were highly antigenic and
were rejected when transplanted into normal syngeneic recipients, whereas the
same tumors grew progressively in syngeneic mice immunosuppressed by X-ray or
UVR (Fisher and Kripke 1979). This was found to hold true in 3 inbred strains
of mice (Kripke 1977) and was thus not a strain-specific phenomenon.
Therefore the question was raised: why, if the tumors are antigenic and are
rejected when transplanted into normal mice, do they develop and grow
progressively in the autochthonous (original) hosts, which is opposite to what
is predicted by the immune surveillance theory? One of the first hypotheses
examined was that UV irradiation of mice leads to a generalized
immunosuppression. Tests of immune function performed by Kripke et al. (1977)
included allogeneic tumor rejection, rejection of H-2 compatible skin grafts,
antibody production, ability to a mount a graft-versus-host reaction, ability
to function as a recipient in graft-versus-host reaction, and ability to
exhibit delayed type hypersensitivity (DTK) to dinitrochlorobenzene (DNCB).
However, after 3 to 4 months of UVR, at the time when UV-irradiated mice were
uniformly susceptible to challenge with UV-induced tumors and well before any
primary tumors develop, all other immune functions had returned to normal
(Kripke et al. 1977). In a different laboratory, Spellman et al. (1977a)
obtained similar results. Mice were UV-irradiated for five weeks and it was
found that after such treatment none of the following measures of immunity
differed between normal and UV-irradiated mice 1) the mitogenic response of
spleen cells to Concanavalin A (Con A) and lipopolysaccharide (LPS), 2) the
plaque forming cell response to sheep red blood cells (SRBC) and
polyyinylpyrrolidone, 3) the percent of B cells, T cells, and macrophages in
spleens, 4) the in vitro proliferative response by spleen cells to allogeneic
spleen cells, and 5) the in vitro cytotoxic response to allogeneic spleen
cells or trinitrophenyl-modified syngeneic spleen cells. Naturally occurring
cell-mediated cytotoxic activity in UV-irradiated mice was found to be
transiently suppressed six days after UV radiation but has returned to normal
after five to ten weeks of UVR (Lynch and Daynes 1983, Noonan et al. 1981a).
Norbury et al. (1977) examined further immune responses of UV irradiated mice
from two weeks to six months of UVR treatment. The in vitro mitogenic
responses of spleen cells and lymph node cells to Con A, LPS, and
phytohemagglutinin (PHA) did not differ; neither did the number of peritoneal
exudate cells elicited by thioglycollate, macrophage phagocytosis, or
activation of macrophages by xenogeneic lymphokines or endotoxins.
In summary, mice which have been chronically irradiated with UV
radiation, although they are unable to reject UV-induced tumors, are not
generally immunosuppressed but have normal immune functions when tested by a
very wide number of assays. Thus one would expect that they should have the
capability to respond immunologically to tumors. In fact, it was found that
UV-irradiated mice respond normally to methylcholanthrene-induced tumors,
spontaneous tumors, and virus-induced tumors (Kripke et al. 1979).
Interestingly, the only tumor not thought to originate in UV-irradiated mice
* * DRAFT FINAL * * *
-------
17-4
which was found to grow preferentially in UV-irradiated mice was the B16
melanoma. B16, a melanoma which arose as a spontaneous tumor in C57BL/6 mice,
was found to grow in a greater percentage of UV-irradiated mice than normal
control mice (Kripke et al 1979). These results were confirmed by Bowen and
Brody (1983) who reported increased percentage of successful transplants and
decreased latency in mice which were UV-irradiated. However, the immune
response of UV-irradiated mice to UV-induced tumors is clearly deficient since
UV-induced tumors which are immunologically rejected in normal mice grow
progressively in UV-irradiated mice (Fortner and Kripke 1977, Fisher and
Kripke 1977, Lill and Fortner 1978). Those tumors which are rejected in
normal, non-UV-irradiated mice are termed regressor tumors, although they do
grow progressively in UV-irradiated mice. The ability of a transplanted
UV-induced regressor tumor to grow progressively in UV-irradiated mice is
paralleled by a depressed inflammatory infiltrate of T cells in and around the
tumor, compared to that which develops when a UV-induced tumor is transplanted
into normal mice (Lill and Fortner 1978). Also, UV-irradiated mice do not
develop significant cytotoxicity to transplanted UV-induced regressor tumors
although non-UV-irradiated mice do (Fisher and Kripke 1978, Fortner and Kripke
1978). Thus UV-irradiation renders a mouse susceptible to the growth of a
UV-induced tumor without significantly affecting its ability to reject
transplanted tumors induced by most other agents. This effectly circumvents
immune surveillance to those tumors which are induced by UVR, and also
apparently upon occasion to other tumors such as the B16 melanoma.
IDENTIFICATION OF T SUPPRESSOR CELLS SPECIFIC FOR UVR-INDUCED TUMORS
Passive transfer experiments showed that the mechanism for the lack of
tumor rejection of UV-induced tumor by UV irradiated mice is a radiosensitive,
la positive, Lyt-1 2 (Ullrich and Kripke 1984) T suppressor cell
(Spellman and Daynes 1977, 1978; Fisher and Kripke 1977, 1978, Daynes et al.
1979) which is specific for UV-induced tumors. This suppressor cell prevents
the mouse from rejecting a syngeneic UV-induced tumor. The T suppressor cell
apparently acts at the induction phase of immunity and not on the action of
cytotoxic effector cells since if the mouse is sensitized to a UV-induced
tumor prior to UV irradiation, the mouse can reject that specific tumor to
which it was immunized but is susceptible to all other UV-induced tumors
(Kripke and Fisher 1976). There is preliminary evidence that UV-irradiated
mice develop helper cells which are sensitized to specific UV-induced tumors
but that the suppressor cell acts to prevent the helper cell from inducing
cytotoxic effector cells (Romerdahl and Kripke 1986). This is consistent with
the data of Roberts et al. (1980) who cloned a T cell line which could
suppress antitumor responses in vivo and which could also suppress the
differentiation of cytotoxic T cells from the lymph nodes of mice which had
been sensitized to a UV-induced tumor. Therefore, the suppressor cell
apparently does not prevent sensitization to a UV-induced tumor nor does it
prevent the effector action of cytotoxic T cells. Its mode of interfering
with immune surveillance is apparently to prevent the effective development of
cytotoxic T cells capable of killing the emerging tumor cells in vivo.
* * * DRAFT FINAL * * *
-------
17-5
All of the above experiments were performed with artificial polychromatic
UV sources which do not provide an exact simulation of sunlight. However,
Morison and Kelley (1985) reported that irradiation of mice with sunlight also
rendered them susceptible to challenge with UV-induced tumors and that
filtration of sunlight through filters to remove UVB abrogated this effect.
The susceptibility was also transferable to immunosuppressed recipients. Thus
one can achieve the same effects on the transplantation immunity to UV-induced
tumors with sunlight as with the artificial sources of UV radiation used in
experimentation and in both cases it is the UVB which has the most detrimental
effects. This finding is very important when attempting to relate the
experimental evidence on the effects of UV-irradiation to potential harmful
effects on human tumor growth.
ROLE OF IMMUNOSUPPRESSION IN TUMOR DEVELOPMENT
A second, and more important, question about the impact of UVR on immune
surveillance is whether the presence of UVR-induced suppressor cells alters
the latent period for appearance of tumors or the number of tumors which
develop? There is evidence that in certain circumstances UVR may enhance the
appearance of autochthonous non-UV-induced tumors. Roberts and Daynes (1980)
found that when ventral skin painting with benzo(a)pyrene (BaP) was preceded 3
weeks by dorsal UVR, the latent period of BaP-induced tumors was significantly
shortened. Although in a similar experiment the latent period of
methylcholanthrene(MCA)-induced tumors was not significantly shortened in
animals pretreated with UVR, MCA-induced tumors from UVR-irradiated mice more
frequently showed enhanced growth when transplanted into UV-irradiated mice
compared to normal mice. Ebbesen (1981) also reported an enhanced incidence
of lymphomas in BALB/c mice which had been UV-irradiated. Thus under these
two experimental conditions, UV-irradiation of mice rendered those mice more
susceptible to the induction of tumors induced by a carcinogen other than
UVR. The exact mechanism is not understood, but it is entirely possible that
it may be due to in interference with normal immune surveillance mechanisms.
De Gruijl and Van Der Leun (1983) reported that ultraviolet irradiation
of ventral skin of mice enhanced subsequent UV tumorigenesis in previously
unexposed dorsal skin. The mechanism of this effect was investigated by
Fisher and Kripke (1982). Mice were lethally x-irradiated and reconstituted
with lymphoid cells from either normal or UV-irradiated mice. These mice were
then skin grafted with UV-irradiated skin to provide a source of transformed
cells. The latent period was decreased and the percent of animals developing
tumors in the transplanted skin was enhanced in those animals which had been
reconstituted with lymphocytes from UV-irradiated mice. In a second
experiment normal mice received an injection of T-cell-enriched lymphoid cells
from UV-irradiated or normal mice and were then UV-irradiated. Transfer of
T-cell-enriched lymphoid cells from UV-irradiated mice rendered the recipients
more susceptible to UV tumorigenesis than transfer of T cells from normal
mice. Thus it was concluded that the presence of the T-suppressor cell is of
great importance in determining if a tumor will appear. Additional evidence
for this conclusion comes from the work of Strickland et al. (1985) who showed
that ventral UVB irradiation greatly enhanced the susceptibility of mice to
*•* * DRAFT FINAL * *
-------
17-6
dorsal two-stage carcinogenesis consisting of initiation of dorsal skin with
UVR followed by promotion of dorsal skin with TPA. One may then postulate
that the development of the T suppressor cell is the mechanism by which immune
surveillance is circumvented. Once mice have developed cells which suppress
an effective immune response against UV-induced tumors, then those transformed
cells which bear appropriate antigens will be able to grow progressively into
grossly visible tumors without interference from the immune response.
Thus, mice that are given UVR are not generally immunosuppressed but do
develop T suppressor cells which are generally specific for UV-induced tumors
and whose presence has been shown to be directly related to tumor
development. Although it is clear that tumor development in UVR-treated mice
is at least in part due to the development of UV-tumor-specific suppressor
cells, the mechanism by which UVR interacts with the immune response to induce
the generation of such suppressor cells is far from clear. Although trauma
such as mild thermal burns and phototoxic damage to skin is also transiently
immunosuppressive, suppressor cells to UV-induced tumors do not develop
(Morison et al 1985).
ANTIGEN PRESENTATION
An additional effect of UV-irradiation, that of suppression of antigen
presentation, may help to explain how UV-irradiation of mice may lead to the
generation of suppressor cells. It had been discovered earlier (Shevach and
Rosenthal 1973) that certain cells in the spleen and lymph nodes were required
to present antigen to T cells in order to elicit an immune response. When T
cells encounter antigen without the aid of antigen presentation by these
adherent cells, they respond poorly or not at all. An additional concept
which is important to the following discussion is the idea that immunization
leads to an immune response, but that the response may be the induction of
suppressor cells rather than the induction of effector cells. Therefore,
although one cannot directly measure a suppressed immune response since the
animal did not generate effector cells, there was a negative response to the
antigen. This is not the same as the lack of a response in which the animal
fails to recognize a particular antigen and results from a switching on of T
suppressor cells which regulate the lack of measurable response. Thus an
animal may not respond to an immunization procedure for a variety of reasons
but if suppressor cells/develop, that is an active immune response.
The antigen presenting capability of cells from UV-irradiated mice is
depressed when tested in UV-irradiated mice under certain circumstances. Much
of the recent work in the field assesses antigen presentation by measuring the
immune response to syngeneic cells to which a hapten has been covalently
linked. A hapten is a molecule which is not by itself immunogenic, but when
linked to another and larger molecule (or to a cell) can induce an immune
response to itself. Thus the cells "present" the hapten to responder cells and
the immune response to the hapten is subsequently measured. A second method
is to measure the ability of an animal to be sensitized to a chemical which is
painted on the skin, contact hypersensitivity.
* * * DRAFT FINAL * * *
-------
17-7
CONTACT HYPERSENSITIVITY AND LOCAL IMMUNITY
The effect of UV radiation on cell mediated immunity can be separated
into two effects, suppression of local immunity and systemic suppression.
Suppression of local immunity occurs when antigen is administered through
UV/irradiated skin shortly after treatment. Relatively low doses of UVR are
required and the effect is not long lasting. However, systemic suppression of
contact hypersens itivity also occurs so that animals are hyporesponsive even
when immunized through skin which was not exposed to UVR. Systemic
hyporesponsiveness requires larger doses of UVR and is longer lasting. In
both cases the type of immune response affected is delayed-type
hypersens itivity (antibody production is not affected) and an antigen specific
T suppressor cell develops.
In an example of a systemic effect, splenic adherent cells or peritoneal
exudate cells from UV-treated donor mice which were
trinitrophenyl(TNP)-derivatized could not induce hapten-specific delayed/type
hypersensitivity when injected into UV irradiated mice; whereas
TNP/derivatized adherent cells from normal mice were able to induce DTK in
UV/treated mice (Greene et al. 1979; Noonan et al. 1981b). Thus the cells
from UV-treated mice were unable to present the antigen TNP appropriately to
induce the production of cytotoxic effector cells although the UV-irradiated
mice were capable of responding to TNP when it was presented by cells from
normal mice. The hyposens itivity could be passively transferred by
lymphocytes into a second recipient and a T suppressor cell was shown to be
responsible. Thus presentation of an antigen by cells from UV-irradiated mice
induced suppressor cells specific for that antigen. UV radiation was also
found to suppress contact hypersensitivity in a second species, guinea pigs
(Morison and Kripke (1984); thus the observation is not species specific.
One can also measure the response to a contact allergen such as DNCB
which, when painted on skin, normally induces an immune response similar to
that which develops in response to poison ivy. If the mice are exposed to an
antigen such as DNCB through UVB-irradiated body wall skin a state of
unresponsiveness develops that is maintained at least in part by an Lyt-l+ T
suppressor cell. This cell was found to act on the induction phase of/immunity
(Elmets 1983). The suppression of the immune response to antigens presented
through UV-irradiated skin is short-lived. The animals have normal or
elevated responses as soon as three days after irradiation of the skin (Lynch
et al 1983). It was suggested that UV impairs the antigen/presenting
potential of Langerhans cells and in the absence of that ability
hapten-derivatized keratinocytes are able to deliver a tolerogenic signal.
The great majority of the experimentation has been done with lamps which
simulate sunlight. However, it is important to note that is was also reported
that sunlight itself could suppress contact hypersensitivity in both mice and
guinea pigs and that the wavelengths responsible were in the UVB range
(Morison et al 1985).
Granstein (1985a, 1985b) has reported that when mice are immunized
subcutaneously with hapten-coupled UV-irradiated epidermal cells, they are
* * * DRAFT FINAL * * *
-------
17-8
hyporesponsive when contact hypersensitivity to the hapten is measured and
that hapten-specific T suppressor cells are present in the immunized mice.
However, depletion of which bear the antigen I-J from the epidermal cell
suspension prior to UVR and haptenation prevents the appearance of these
suppressor cells. Non-UVR treated epidermal cells which are I-J depleted and
haptenated can induce contact hypersensitivity. Therefore, there seems to be a
separate cell in the epidermis which is UV resistant and which apparently
presents antigen in a tolerogenic fashion and thus preferentially induces
suppressor cells.
ACTION SPECTRA OF IMMUNE SUPPRESSION
Experimentation on the photobiology of the phenomenon has provided a
great deal of information on the relative effectiveness of various wavelengths
and on reciprocity. The wavelengths which most effectively suppress contact
hypersensitivity and which also lead to the production of tumor specific
suppressor cells lie in the UVB range. It was found that the ability of UVR
to induce susceptibility to transplanted UV induced tumors in mice lies in
wavelengths below 315 nm, i.e. UVB, and appears to show reciprocity (dose rate
independence)(De Fabo and Kripke, 1979, 1980). This is in contrast to
tumorigenesis by UVR, which does not appear to show dose/rate reciprocity
(Forbes et al 1978) but which also is primarily caused by UVB.
In a study of the photobiology of the suppression of contact
hypersensitivity by UVR (Noonan et al 198la) it was reported that a Mylar
filter, which removes wavelengths less than 315 nm, abrogated the suppressive
effect. Thus it was concluded that depression of contact hypersensitivity was
due to UVB. Elmets et al (1983) also studied the wavelength dependence of the
local suppressive effect of UVR on contact hypersensitivity and determined
that the greatest effect was at 297 nra. Wavelengths of 290 nm or greater than
315 nm were less effective. De Fabo and Noonan (1983) constructed a detail in
vivo action spectrum using 10 different narrow wavebands (2-3 nm) and found
peak suppression of contact hypersensitivity between 260 nm and 270 nm. They
found a shoulder at 280 to 285 nm and then a steady decline in effectiveness
to about 3 percent of maximum at 320 nm.
The great majority of the experimentation has been done with lamps which
simulate sunlight but it has also shown that sunlight itself could suppress
contact hypersensitivity in both mice and guinea pigs (Morison et al 1985). In
the latter experiment, the wavelengths responsible were reported to be in the
UVB range. The data for all experiments described is quite consistent in
showing that UVA wavelengths are ineffective in suppressing contact
hypersensitivity. The wavelengths reported to cause the greatest depression
of contact hypersensitivity depend somewhat on the UVR source used. For
example, in sunlight, UVB is apparently the most effective. However, when one
compares a broad range of wavelengths, there is a general trend that the
shorter the UVB, the greater the suppression. However, the most effective
wavelengths lie in the longer wavelengths of UVC. Since reduction of ozone in
the stratosphere would increase the penetration of the shorter wavelengths of
UVB more than that of the longer UVB, this is an important consideration.
* * * DRAFT FINAL * * *
-------
17-9
POSSIBLE ROLE OF UROCANIC ACID IN IMMUNE SUPPRESSION BY UVR
Noonan e't al (1981a) found a clear differential in wavelength
effectiveness in suppression of contact hypersensitivity and they suggest that
this indicates the existence of a unique photoreceptor mediating UV/induced
immunosuppression. However, the action spectra which they constructed for
suppression of contact hypersensitivity is not congruent with the absorption
spectrum of DNA proposed by Setlow (1974). Because their action spectrum
closely matches the absorption spectrum of urocanic acid, and urocanic acid is
found in the epidermis, they conclude that there is a strong possibility that
this compound is the photoreceptor. Evidence supporting this hypothesis is
given by studies of Noonan et al (1984). Trans urocanic acid, a derivative of
histidine, accumulates in the stratum corneum to as much as 0.5 percent of the
wet weight of the epidermis (De Fabo et al. 1984) and undergoes isomerization
to the cis form when exposed to UVR (Morison 1981). Removal of the stratum
corneum by tape stripping prior to UVR abrogated the induction of suppression
of contact hypersensitivity by UVR (Noonan 1981b). In addition, UV-irradiated
urocanic acid was reported to suppress delayed type hypersensitivity in mice
(Ross et al 1986). Thus these results suggest that absorption of UVR by DNA,
although generally considered of greatest importance for carcinogenesis, might
not mediate the suppression of the immune response.
UVR-INDUCED ANTIGENS ON UV-IRRADIATED SKIN
Spe.llman and Daynes (.1984) have reported that in mice there are antigens
present on UV-irradiated skin that are cross-reactive with UV/induced tumors.
Mice were irradiated 30 minutes per day for 6 weeks, which is well urider the
time of radiation required to produce tumors in that system. Skin from these
animals was then transplanted as 1 cm grafts to normal mice. Twenty days
later a second graft of UV-irradiated skin was made. When the mice that had
received the skin grafts of UV-irradiated skin were challenged with a
UV-induced tumor 15 days later, the animals did not succumb to the tumor,
i.e., they were protected from the challenge. Since the grafting of
UV-irradiated skin protected against a challenge with UV induced tumors, the
conclusion was drawn that antigens existed on UV/irradiated skin that were not
normally present on skin and which were also present on UV-induced tumors.
Therefore, if UVR causes the expression of neoantigens on skin and at the same
time facilitates the induction of suppressor cells to antigens, then the
immune response against those tumors might take the form of induction of
suppression rather than induction of cytotoxic cells. This would effectively
circumvent immune surveillance.
HUMAN STUDIES
There have been very few studies of the effects of in vivo UVR on the
immune response in humans. Morison et al (1979) UV-irradiated human
volunteers with 1.5 minimal erythemal doses (MED) to produce what they termed
asymptomatic erythema or 3 MED which they termed symptomatic erythema. Post
irradiation, all subjects showed a significant increase in the proportion of
circulating polymorphonuclear leukocytes and a corresponding decrease in the
proportion of circulating lymphocytes. Subjects who received three MED also
had a significant decrease in the proportion of circulating E-rosette forming
* * * DRAFT FINAL * * *
-------
17-10
cells, i.e. T cells, and an increase in the proportion of null cells (cells
which lack surface markers for T or B cells). The changes in the absolute
numbers of these cells were not significant. Thirty minutes and four hours
after UV-irradiation, the response of lymphocytes to PHA was increased, it
then decreased to a minimum 12 hours after UVR, and returned to a normal range
72 hours post exposure. This experiment demonstrates that in vivo UV
irradiation of humans can affect the function of lymphocytes as measured in
vitro, although the relationship of th'ese experiments to those in mice is not
quite clear.
Hersey et al. studied the effects of UVR on humans after solarium
exposure (1983a) and after exposure to ten days of sunlight for one hour per
day (1983b). In both studies there was a marked increase in the ratio of
suppressor cells to cytotoxic/helper cells which had not returned to normal
two weeks after UV-irradiation. Also, there was an increase in the in vitro
activity of a gamma radiation-sensitive suppressor T cell activity against
nonspecific (pokeweed mitogen-induced) IgG and IgM production. Although
immunoglobulin production in vitro had too wide a range for statistical
analysis of these small numbers of patients to be of value, Hersey et al.
(1983a,.b) found that gamma irradiation of T cells prior to culture increased
the amount of immunoglobulin produced in patients exposed to sunlight but not
in controls. . O'Dell et al. (1980) reported that there was a diminished immune
response in sun-damaged skin. The concentration of DNCB required to elicit a
positive patch test was greater in sun-damaged skin than in skin which was
normal. In addition, the delayed-type hypersens itivity to intradermal
injection of Candida, mumps, and PPD antigens was decreased in sun-damaged
skin so that the differences were not due to a difference in percutaneous
absorption of antigen through sun-damaged skin. The inflammatory response to a
primary irritant was the same in both sun-damaged skin and normal skin and
there was no difference in the two tested sites (back of the neck and the
back) in volunteers without sun-damaged skin at the back of the neck.
Therefore, there is apparently a local suppression of contact hypersensitivity
in sun-damaged skin. These reports suggest strongly that UVR can suppress
certain parameters of the immune response in humans as well as in laboratory
animals and further suggests that UV-irradiation of humans could interfere
with immune surveillance of UV-induced tumors in a fashion similar to that
observed in mice.
CONCLUSIONS
Very little of the research done is related directly to melanoma.
However, there is evidence in mice that, although UVR does not seem to affect
the ability of a mouse to reject most transplanted tumors (those not induced
by UVR) (Kripke et al. 1979), there are reports that UVR of mice does enhance
the development of spontaneous (Ebbesen 1981) and chemically induced (Roberts
and Daynes 1980) autochthonous tumors. There is, then, circumstantial
evidence that UVR may potentially facilitate the growth of melanomas in
humans. First, there are cross-reacting antigens on melanomas among mouse
(Gersten and Marchalonis 1982) and among human melanomas (Houghton et al.
1982) and there are similarities in antigens found on both human and mouse
* * * DRAFT FINAL * * *
-------
17-11
melanomas (Tomecki et al. 1980). Humans bearing melanomas make an immune
response to that tumor but the response is ineffective in at least those
patients who develop progressive melanomas and the tumor is able to grow and
kill the host. Any effect on the immune response which interferes with that
immune response can potentially increase the growth rate of that tumor or
allow it to escape from immune surveillance. The experimentation with animals
has shown us that UVR induces the production of suppressor cells which not
only facilitate the growth of UV-induced tumors but also the B16 melanoma.
Similar experimentation with other melanomas has not been reported and no such
experimentation is possible with human melanomas. However, the potential
exists that the same events may take place in humans. If so, then UVR might
facilitate the growth of melanomas in a fashion similar to what is seen in
mice.
Finally, the experiments of Mersey et al. (1983 a,b) and O'Dell (1980)
demonstrate that the immune system of humans can be adversely affected by
UVR. All of this information in toto implies that UVR may permit the growth
of human melanomas. There is certainly no direct evidence that this does
occur but there is certainly experimental evidence that it is a possibility.
It is clear that there is a real and pressing need for more research in this
area to as best as possible directly address the question of the effect of UVR
on the growth of melanomas as well as the induction of melanomas.
FINDINGS
In conclusion, ultraviolet radiation has been shown to have a variety of
effects on the functioning of the immune system.
17.1 UVR, specifically UVB, can result in the generation of T suppressor
cells which specifically suppress the immune response to UV-induced
tumors. The suppressor cells have been shown to be capable of
shortening the latent period of tumors induced by UVR. UVB-treated
animals also appear to have increased susceptibility to one form of
melanoma in experiments with transplanted tumors.
17.2 The wavelengths most effective in causing the production of tumor
specific suppressor cells in mice appear to be in the UVB range.
* * DRAFT FINAL * * *
-------
CHAPTER 18
MELANOMA DOSE-RESPONSE
This chapter reviews the recent literature relevant to assessing
dose-response relationships between ultraviolet radiation (particularly
ultraviolet radiation-B -- UVB) and cutaneous melanoma skin cancer (CMM).
Dose-response relationships for CMM have been estimated in several
epidemiological studies; at present, there is no animal model for CMM
dose-response relationships. The focus of this chapter is on evaluating the
epidemiological studies for their usefulness in estimating how future melanoma
incidence and mortality rates would respond to potential future modifications
in total column ozone that, in turn, alter the flux of UVB reaching the
earth's surface.
THE DATABASES AVAILABLE FOR DOSE-RESPONSE MODELING
Estimation of the changes in melanoma incidence or mortality that would be
associated with variations in UV flux could be performed in a variety of
ways. Information from animal models could be used, in which the responses of
animals to large doses are assumed to be relevant to the responses of humans
to small doses. Human epidemiological data could be used, in which UVB
measurements taken at different locations are related to melanoma incidence
and mortality at these locations. Each approach has strengths and weaknesses.
. The use of animal data to extrapolate to human risk has several generic
weaknesses, the largest one being that extrapolation is required from high to
low doses and from one species to another. In addition, generally animal data
come from a very homogeneous population both genetically and in terms of
exposure, whereas humans are genetically very heterogeneous and have very
different exposure patterns. In the case of CMM there is one additional large
weakness, and that is the availability of an appropriate animal model. As a
general rule, the type of tumor developed in an animal experiment has not been
considered relevant to the use of those data in risk estimation. In the case
of melanoma, however, given the great differences in the biologic behavior of
CMM and non-melanoma skin cancer it does not seem appropriate to use the
animal data on NMSC to estimate human risk from CMM.
Use of epidemiological studies avoids some of the problems associated with
the use of animal studies but faces other, different weaknesses. The
advantages of epidemiological studies include: (1) the ability to control for
environmental factors that could influence the effectiveness of dose; (2) the
analysis of UVB doses that are often within the range of predictive concern;
and (3) obviously, a focus on human beings -- the species of predictive
concern. The cost of this "realism," however, is high. Many genetic and
environmental factors that are believed to influence melanoma incidence and
mortality cannot be measured accurately in epidemiological studies.
Furthermore, human exposure to UVB rarely varies systematically, making it
difficult to find appropriate test and control populations. Nonetheless, a
* * * DRAFT FINAL * *
-------
18-2
number of ecologically-oriented epidemiological studies have been performed
(Fears, Scotto, and Schneiderman 1976, 1977; Scotto and Fears 1986; and
Pitcher 1986). These are the studies that will be evaluated in this chapter.
Sources of Unexplained Variation
In using epidemiological databases of melanoma incidence and mortality in
order to derive dose-response relationships, researchers must make a number of
assumptions about the -variables that potentially influence both UVB dose and
response. Ambient UVB varies substantially in the environment, with gradients
of peak and cumulative UVB existing for latitude. UVB also varies with
altitude, cloudiness, albedo, and the time of day. Another factor causing UVB
variations is the climatology of ozone transport and abundance (Angell 1983).
Available measurements of UVB show that these factors cause variations1 of UVB
(Scotto and Fears 1986; Reinsel et al. 1983).. Consequently, whether one is
using modeled UVB data (Greene 1983), a combination of modeled and measured1
data (Serafino and Frederick 1986), or measured data alone (Scotto and Fears
1986), the exposure data used in ecological studies will only approximate the
actual potential exposure of individuals over a long period of time. Some
amount of unexplained variation in UVB affects any ecological analysis,
decreasing the ability of any model to explain variations in CMM incidence or
mortality.
In addition, a number of factors affect the actual dose a person
receives. Skin color in the absence of tanning and the ability to tan or to
have skin thicken can affect the amount of UVB that reaches the basal layer of
the skin.. Individual behavior can vary across locations. People have
different patterns of dress, work exposure, recreation, eating, and even
medical care and intervention. All of these factors can be expected, in
varying degrees, to introduce variations in CMM incidence or mortality that
are unexplained in- epidemiological studies. For example., Crombie (1981)
concludes that differences in ethnicity and skin color make the observation of
a latitudinal gradient of incidence across European countries impossible.
Within England, however, where variation in skin color and tanning capability
is relatively small, changes in solar radiation appear to be highly related to
variations in melanoma incidence rates (Crombie 1981). In other countries,
such as Italy where skin color varies north to south, the influence of skin
color variability and behavior on incidence and mortality may overwhelm the
impact of variations in UVB (Crombie 1981). Finally, little is known about
genetic fac;ors that affect melanoma incidence and mortality but are unrelated
to solar exposure. Chapters 2 and 3 of this review discuss the many factors
that influence potential exposure and effective doses.
1 Measured UVR data usually comprise a few locational sample points
collected over a short time period.
DRAFT FINAL * * *
-------
18-3
EVALUATION OF EPIDEMIOLOGICAL STUDIES USEFUL
FOR EVALUATING DOSE-RESPONSE
A variety of case-control epidemiological studies have identified risk
factors for CMM. Unfortunately, none of the case-control studies focused
specifically on UVB, so these efforts cannot be used to estimate a
dose-response relationship between melanoma and UVB. For example, studies of
the relationship between "sunshine hours" and CMM were not evaluated in this
chapter because "sunshine hours" and UVB may be unrelated. The ecological
studies evaluated in this chapter do not fully control individual UVB dose but
do focus specifically on how variations in UVB exposures explain variations in
melanoma incidence and mortality.
The number of countries in which ecological studies appropriate to
assessing the relationship between UVB and CMM can be performed is limited.
Ideally, such studies should use data from a country that spans a wide range
of latitudes, thus allowing the use of UVB data that reflect a significant
range of exposures. However, the studies must control for differences in skin
coloration across latitudes, requiring either: (1) data from a country that
has a population with a predominant single skin color (e.g., Norway, Sweden,
or England) or countries in which skin color does not vary with latitude, such
as the so-called immigrant countries (the U.S., Australia, or New Zealand); or
(2) data on population characteristics that would enable researchers to
control for differences in skin coloration. (Fears and Scotto 1983). Ruling
out ecological studies that do not meet these requirements, only a few
epidemiological studies remain.
One study which is potentially useful for estimating dose-response for CMM
in the United States was that performed by Fears, Scotto, and Schneiderman
(1976). These authors use two types of data to represent UVB exposure: (1)
latitude -- latitude and UVB radiation weighted for erythema effectiveness
correlate at 0.97, and (2) monthly totals of erythema-producing UV radiation
-- expressed as Biologically Effective Units (BEUs) and derived from Schultze
(1974). Both types of data were correlated with CMM mortality and incidence
data. The incidence data were for four cities included in the Third National
Cancer Survey (TNCS) (1975). Mortality data were from the U.S. Cancer
Mortality by County database (Mason and McKay.1973).
Exhibit 18-1 shows the results of the simple correlation model based on
latitude:
log R = a + PL
where R is the age-adjusted rate of incidence or mortality, a and 3 are
constants, and L is latitude.
Exhibit 18-2 shows the results estimated using BEUs. These estimates were
based on an exponential model and represent the percentage changes in
incidence and mortality estimated to occur with increases in UV radiation of
between 10 and 30 percent. Note that the use of an exponential model leads to
higher dose-response relationships for higher base exposures.
* * * DRAFT FINAL * * *
-------
EXHIBIT 18-1
Summary Statistics for Regressions of
Skin Cancer Incidence and Mortality on Latitude
MALES
Correlation Regression Doubling
Coefficient* Slope + S.D. Latitude
(degrees)
Melanoma Incidence -.86 -.037 ± .007 -9.8
Melanoma Mortality -.81 -.017 ± .002 -19.9
FEMALES
Correlation Regression Doubling
Coefficient* Slope + S.D. Latitude
(degrees)
-.83 -.038 ± .007 -10.7
-.71 -.014 ± .002 -22.2
* Simple correlation coefficient between log of incidence or mortality rate and latitude. Model
coefficients were statistically significant at p<0.01.
Source: Fears, Scotto, and Schneiderman, 1976.
00
i
* * * DRAFT FINAL * * *
-------
EXHIBIT 18-2
Estimated Percentage Increases in Melanoma Skin Cancer Incidence and
Mortality Associated With Changes in Erythema Dose a./
(Figures in parentheses are 95 Percent Confidence Intervals)
BASE
BEU
MALES
Increase
in Total
20%
Dose
FEMALES
Increase in Total Dose
20% 30%
Melanoma
Inc idence
Melanoma
MortaIity
650
850
1050
15%(7-24)
20
25
32%
44(
57
18-75)
52%
72
96(37-180) b/
13%
18
22
39
50
64
84 b/
650
850
1050
8(6-10)
10
13
16
22(16-28)
28
26
35
45(33-58)
6
9
11
13
18
22
21
28
36
00
i
Ln
ay A sample computation is outlined:
A 10% increase in total dose at 48.25'N, where exposure is 650 BEU, equals 65 BEU.
A change of 65 BEU ii equivalent to a reduction in latitude of 1.93'.
A change of 1.93' at 48.25'N is associated with an 18% increase in nonmelanoma incidence.
b/ These estimates require extrapolation well beyond the range of the data.
Source: Fears, Scotto, and Schneiderman, 1976.
* * * DRAFT FINAL * * *
-------
18-6
There are several limitations in this work. In particular, only four
cities were used for melanoma incidence. With a sample this small, variations
in behavior, .skin pigmentation, and cloud cover may bias estimates of the
coefficients. As a result, caution must be used in evaluating the error of
the estimate.
In another study, the same authors estimate a dose-response relationship
for melanoma incidence using a power model (Fears, Scotto, and Schneiderman
1977):
P.. = b (U.A.)
where P.. = probability of developing melanoma for jth age group at
location i;
U. = annual UV count at ith location;
A. = midrange of jth age. group; and
b, c, k are constants.
They assume that melanoma skin cancer incidence for the jth age group at .
location is binomally distributed. They argue, that because R. . a:re small
and population size is large, then ln(R..) can be regarded as normally
distributed with variance equal to the inverse of the expected number of cases
(W.. ). Using least squares regression, parameters were estimated by
fitting a log form of the model (for lnR..=a + clnU. 4- kLnA . +
E..) after weighting by the observed number of cases. Annual counts from
Robertson-Berger meters were used for UVB . Incidence data also were from the
TNCS (1975). This measure differs from the erythema spectrum by weighting
more heavily towards longer wavelengths.
Exhibit 18-3 presents the regression coefficients and statistics.
Applying the results in this exhibit, a one percent relative increase in UV
Q
radiation results in an estimated 100*(1.01 -1) percentage increase in
melanoma incidence, or 2.47 for females and 2.24 for males. The authors
stress, however, that these estimates may be biased by the omission of
location-specific demographic and environmental variables.
In applying the authors' 1976 and 1977 studies to a risk assessment of
ozone depletion, notice should be paid to how two factors, the biological
amplification factor (BAF) and radiation amplification factor (RAF),
interact. A BAF is a number that indicates the percentage change in incidence
or mortality associated with a given percentage change in UVB. The RAF
relates percentage changes in ozone levels to percentage changes in UVB.
Fears, Scotto, and Schneiderman (1976), use the erythema action spectrum to
* * * DRAFT FINAL * * *
-------
18-7
EXHIBIT 18-3
Summary of Fears, Scotto, and Schneiderman (1977)
Regression Analysis of Melanoma Incidence
Dose-Response
Percent t-statistic c + SD
Variation (UV a + SD K + SD (UV
Sex Explained Coefficient) (Constant) (Age Coefficent) Coefficient)
Males
Females
74
62
5.4
4.6
-35.6 + 6.5
-30.5 + 6.9
.80 + .31
.29 + .34
2.45 + .45
2.23 + .48
- In Rii = o.+ cln U. + < InA. + E..
i J ij
p<.01
Source: Fears, Scotto, and Schneiderman (1977).
* * * DRAFT FINAL * * *
-------
18-8
weight wavelengths in the UVB. Because erythema varies more with latitude
compared to other action spectra, a lower BAF was estimated. At the same
time, erythema is more sensitive to ozone depletion; it has a higher RAF.
Scotto and Fears (1986) estimated a dose-response relationship for
melanoma incidence using an expanded database about populations in a greater
number of cities. Scotto and Fears (1986) also include a new term (VAR) to
adjust for the presence of some host or environmental characteristics:
Ln R.. = a + b In (Age.) + c In UVB. + dVAR.. + e. .
ij i J ij ij
Seven areas are included in the database: Detroit, Seattle, Iowa, Utah, San
Francisco, Atlanta, and New Mexico. Again, counts obtained from Robertson-
Berger meters were used for UVB.
Separate estimates were made for: (1) different anatomical sites -- FHN
(face, hands, neck); UE (upper extremity); LE (truck and lower extremity); and
(2) age groups -- 20-39; 40-54; 55-64; 65-74; Non-whites and Hispanics in New
Mexico were excluded. For most categories,. Scotto and Fears found that after
controlling for confounding variables each 1 percent increase in UVB increases
-incidence by less than 1 percent. Exhibit 18-4 shows the results for
different anatomical sites. The dose-response estimates are statistically
significant (p<0.1). Exhibit 18-5 shows the effects of introducing
different constitutional and environmental variables. These factors were
analyzed using stepwise multiple regressions. The analysis showed
statistically significant biological amplification factors (p<0.5) for
ethnic groups that spent some time outside, with particular risk associated
with weekend sun exposure.
Pitcher (1986) used data on melanoma death rates by county for a 30-year
period and exposure data from a National Air and Space Administration model
(the UV Model) to estimate the relationship between CMM death rates and
several alternative measures of UVB exposure. The alternative measures
included UV radiation doses estimated with different action spectra and
assuming different exposure scenarios, e.g., evaluating an annual mean amount
of UV or that delivered during a peak one day exposure (clear day in June).
Data on melanoma mortality by county were obtained from the EPA/NCI data
base (U.S. EPA 1983) for the period 1950 to 1979. The UV Model estimated UV
exposures in these counties, by using modeling relationships based on
latitude, longitude, altitude, surface albedo, total column ozone, and cloud
cover to estimate the amount of UV delivered and weight that information with
various biological action spectra. Pitcher.made an extensive effort to
validate the exposure data obtained from the UV model.
* * * DRAFT FINAL * * *
-------
18-9
EXHIBIT 18-4
Biological Amplification Factors for Skin Melanoma
By Sex and Anatomical Site Groups, Adjusting for
Age and Selected Constitutional and
Exposure Variables*
MALE
VARIABLE
Age
Sunburn
Freckles
Scandinavian Ancestry
Lt. hair color
Scot/Irish Ancestry
Moles
Lt. eye color
Fair Skin
Sunscreen use
Suntan lotion use
Radiation protection
Protective clothes
Hours outdoors on weekdays
Hours outdoors on weekends
Trunk/ LE
6%
4
5
5
5
6
6
6
6
4
5
5
8
7
5
FHN/UE
8%
8
7
9
8
6
7
8
8
7
8
8
10
9
7
FEMALE
Trunk/ LE
5% .
5
5
5
5
4
5
4
4
4
4
6
5
5
.8
FHN/UE
' 10%
10
10
10
10
8
11
9
9
10
10
10
10
10
13
* The biological amplification factor indicates the percentage change in
melanoma incidence associated with a 10 percent relative increase in UVB.
Source: Scotto and Fears (1986).
* * DRAFT FINAL * *
-------
18-10
EXHIBIT 18-5
Biological Amplification Factors for Skin Melanoma
By Sex and Anatomical Site Groups, Adjusting for
Age and Combinations of Selected Constitutional
and Exposure Variables*
ANATOMICAL SITE
MALE
FEMALE
Trunk/LE
Variables
included
in model
FHN/UE
Variables
included
in model
3 Percent
suntan lotion use
Scandinavian
It hair color
UVB INDEX
hours outdoors w/days
5 Percent
Scot/Irish Ancestry
suntan lotion use
fair skin
UVB INDEX
hours outdoors w/ends
4 Percent
suntan lotion use
Scandinavian
It hair color
UVB INDEX
hours outdoors w/ends
6 Percent
Scot/Irish Ancestry
suntan lotion use
fair skin
UVB INDEX
* A biological amplification factor is a number that indicates the percentage
change in melanoma incidence for a 10 percent percentage increase in UVB.
Source: Scotto and Fears (1986).
* * * DRAFT FINAL * * *
-------
18-11
An examination of Pitcher's data in Exhibits 18-6 and 18-7 shows that the
relationship between age and melanoma death rates is approximately exponen-
tial. Further, the rates for males and females are different. These
observations led Pitcher to fit two different models for each sex group.
Defining DRM... as the death rate for the ith cohort in the jth location in
the kth time period, he estimated:
DRM... = exp(b. + b.AGE., + b.EXP.) + e
ijk ^ 0 ,1 ik j j
where AGE., is the age of ith cohort in the kth time period and EXP . is
IK j
the exposure in the jth location (SMA) .
In this model, the percentage change in melanoma mortality associated with
variations in UV flux is higher the greater the baseline exposure. Pitcher
also developed a second model, which differs from the first only in the use of
the log of the exposure variable. In that model, a 1 percent increase in
exposure generates the same percentage increase in melanoma death rates
regardless of the baseline death rate.
Pitcher found that many of the birth cohorts had zero deaths in any given
5-year period, especially in smaller cities. Further, because rates differed
significantly across cohorts, the variances of the cohort death rates varied
under the assumption that the probability of death had a binomial
distribution. The different population sizes within cohorts also caused
•variation in the variance of 'the cohort death rate. These issues made the use.
of normal linear regression techniques infeasible. Normal weighting
techniques could not be used because zero rates for some cohorts prevented
computation of weights. Therefore, weights were computed using expected fates
rather than actual rates.
Using UVB radiation data from the UV Model for a clear day in June and
measured with the DNA action spectrum, Pitcher estimated constants on the
exposure term in the second formulation to be 0.85 for males (standard error
equals 0.067) and 0.58 for females (standard error 0.078). The clear day
values of UV radiation can be considered estimates of peak daily UVB
radiation. With these constants, a 1 percent increase in UV radiation results
in an estimated increase in melanoma mortality of about 0.85 percent for males
and about 0.58 percent for females. Exhibit 18-8 extends the estimates to
larger relative changes in UVB.
In the analyses done by Pitcher, the question was raised as to which
measure of UVB dose would be most appropriate for estimating mortality risks
in the event of ozone depletion. The regression analysis showed a more
significant statistical relationship using the DNA action spectrum for peak
values (June 15th) as the exposure variable, as compared to using the other
action spectra (e.g., erythema and RB) and/or other time periods (e.g., annual
cumulative values or cumulative values for June) which would be another way to
* * * DRAFT FINAL *
-------
18-12
EXHIBIT 18-6
Melanoma Death Rate Per 100,000 Population
As a Function of Age
Males -- 1882-1975
OJ
jj
ro
cr
-------
18-13
EXHIBIT 18-7
Melanoma Death Rate per 100,000 Population
As a Function of Age
Females -- 1882-1975
01
•4-1
tO
cr
01
QJ
a
i i i i i i i i i i i i i
4 =•
* * * DRAFT FINAL *
-------
18-14
EXHIBIT 18-8
Estimated Percentage Change in Melanoma Mortality
For Different Percentage Changes in UVB
(DMA Action Spectrum for a Clear Day in June)
PERCENTAGE
INCREASE IN
UVB
1
2
5
10
20
MALES
0.85
1.7
4.2
8.4
17
FEMALES
0.58
1.2
2.9
5.7
11
Source: derived from Pitcher (1986).
* * * DRAFT FINAL » *
-------
18-15
measure peak values. It must be emphasized, however, that the statistical
differences due to the different exposure measures were extremely small.
Therefore, the decision regarding which dose measure to use is primarily
judgmental, and should be based inter alia on hypotheses about, disease
etiology.
Although peak values were used in Pitcher's analysis, the question exists
about whether this approach overestimates the possible risks of ozone
depletion. Using peak UVB values as the exposure measure may overstate these
risks because the variability of peak values across locations is less than the
variability of UVB measured over different time periods. The lower
variability of the peak values results in larger estimates of the
responsiveness of melanoma mortality to changes in UVB for this exposure
measure. Preliminary analysis indicates that using alternative exposure
measures could reduce the sensitivity of melanoma mortality to changes in UVB
radiation by as much as 60 percent. Consequently, the appropriate exposure
measure for estimating dose-response for melanoma mortality is an area
requiring future research, and sensitivity analyses should be included in
assessments of melanoma mortality.
PROBLEMS IN USING RESULTS OF EPIDEMIOLOGICAL STUDIES
A variety of problems exist in applying the results of the epidemiological
studies to estimate the changes in incidence and mortality rates that would
occur with ozone depletion. Ethnic shares in the population may not be
maintained over time. Depending on the future population mix, this issue
could under or overestimate future rates. Migration was not considered in the
studies; if southern cities have large future population increases the
sensitivity of the overall U.S. population to UVB will be underestimated.
Individuals might also change their personal behaviors that influence UV dose,
such as the use of sunscreens or the propensity to work outdoors.
The dose-response relationships developed in epidemiological studies also
omit other important variables that could influence melanoma incidence and
mortality. For example, to the extent that the dose of UVB relates to income
and sunny vacations, risks could be biased. In addition, some seasonal
components of UVB are more important than others, so that the independent
variables used in the dose-responses estimates could over or underestimate
responses. Because of these and othor unresolved uncertainties, dose-response
relationships continue to be developed.
FINDINGS
18.1 At the current time there are, no animal data which are appropriate for
evaluating dose-response relationships between CMM and UVB; there is no
animal model for CMM induction by UVB and animal data on the relation-
ship between nonmelanoma skin cancer (NMSC) and UVB are probably not
appropriate given the large differences in biologic behavior between CMM
and NMSC in human populations.
* * * DRAFT FINAL * *
-------
18-16
18.2 Studies of the relationship between changes in UVB radiation and
melanoma incidence and mortality continue to be developed. Neverthe-
less, recent efforts suggest a significant correlation.
18.2a Scotto and Fears (1986) found that each 10 percent increase in
UVB would be associated with a 3 to 5 percent increase in
melanoma incidence after controlling for confounding host or
environmental characteristics. This range reflects differences
in: (1) anatomical site -- trunk and lower extremities were
found to respond less to changes in UV radiation than face, head,
neck, and lower extremities; (2) sex; and (3) the constitutional
and exposure variables included in the estimates.
18.2b Pitcher (1986) estimated that a 10 percent increase in UVB
radiation would be associated with an 8.4 percent increase in
melanoma mortality for males and a 5.7 percent increase in
females. The strongest association between mortality and UV
radiation was found when the dose parameter was UVB energy
weighted by the DNA action spectrum. Preliminary analysis
indicates, however, that using, alternative measures of UVB
radiation could reduce the estimated sensitivity of melanoma
mortality to changes in UVB by as much as 60 percent.
18.3 Because the strength of the associations between UV radiation and
melanoma incidence and mortality appears sensitive to the choice of
action spectrum, assessing the risks of melanoma skin cancer due to
ozone modification requires using an appropriate action spectrum when
estimating: (1) the relationship between ozone depletion and UV
radiation, and (2) how melanoma incidence and mortality respond to
changes in UV radiation.
* * * DRAFT FINAL *
-------
18-17
REFERENCES
Angell, J.K., and Kofshover, J. Global Variation in Total Ozone and
Layer-Mean Ozone: An Update through 1981. J. of Climatology and Applied
Meteorology. Vol 22, pp. 1611-1627 (1983).
Crombie, I.K. The Limitations of Case-Control Studies in the Detection of
Environmental Carcinogens. J. Epidemiol Community Health 35(4):281-287 (1981).
Epstein, J.H., Epstein, W.L., Nakai, T. Production of Melanomas from
DMBA-induced "Blue Nevi" in Hairless Mice with Ultraviolet Light. J. Natl.
Cancer Inst. 38:19-30 (1967).
Fears, T.R., Scotto, J., and Schneiderman, M.H. Skin Cancer, Melanoma and
sunlight. Amer J Pub health 66:461-464 (1976).
Fears, T.R., Scotto, J. , Schneiderman, M.H. Mathematical Models of Age and
Ultraviolet Effects on the Incidence of Skin Cancer Among Whites in the United
States. Am. J. of Epi 105:420-427 (1977).
Fears, J.R., and Scotto, J. Estimating Increases in Skin Cancer Morbidity Due
to Increases in Ultraviolet Radiation Exposure. Cancer Invest. 1:119-126
(1983).
Green, A', and V. Siskind. . Geographical Distribution of Cutaneous Melanoma in.
Queensland. MJ of Aust 1:407-410 (1983).
Kripe, M.L. Speculations on the Role of Ultraviolet Radiation in the
Development of Malignant Melanoma. JNCI 63:541-548 (1979).
Mason, J.J. and McKay, F.W. 1974 U.S. Cancer Mortality by County:
1950-1969. (NIH) 74-615. Department of Health, Education, and Welfare,
Washington, D.C. (1973).
Pitcher, H. Melanoma Death Rates and Ultraviolet Radiation in the United
States 1950-1979. Unpublished manuscript (1986 in press).
Reinsel, G., Tiao, G.C., Lewis, R., and Bobkoski, M. Analysis of Upper
Stratosphere Ozone Profile Data from the Ground-Based Dankehr Method and the
Nimbus-4 BUY Satellite Experiment. J. of Geophysical Research Vol. 88 pp.
5383-5403. (1983).
Schulze, R. Increase of Carcinogenic Ultraviolet Radiation due to Reduction
in Ozone Concentration in the Atmosphere. Composition and General Circulation
of the Upper and Lower Atmosphere, and Possible Anthropogenic Perturbations.
Atmosphere Environment Service, Ontario, Canada, Vol. 1:427-493. (1974).
Scotto, J., and Fears, T.R. The Association of Solar Ultraviolet Radiation
and Skin Melanoma Among Caucasians in the United States. Cancer
Investigations (1986 in press).
DRAFT FINAL * * *
-------
18-18
Serafino, G. and Frederick, J. Global Modeling of the Ultraviolet Solar Flux
Incident on the Biosphere, prepared for U.S. Environmental Protection Agency,
Washington, B.C. (1986).
Third National Cancer Survey: Incidence Data. Monograph 41. DHEW Publ. No.
(NIH) 75-787. National Cancer Institute. U.S..DHEW, Bethesda, MD. (1975).
* * DRAFT FINAL * * *
-------
CHAPTER 19
CONCLUSIONS
This chapter seeks to determine whether or not the weight of evidence
presented in this document supports the conclusion that it may be reasonably
anticipated that a change in ambient ultraviolet radiation caused by
modification of column ozone would increase the incidence or mortality of
cutaneous malignant melanoma (CMM). An affirmative judgement must be made on
the basis of weighing and balancing all the evidence, not on whether the
evidence provides absolute proof or certainty that increases in UVB will be
associated with increases in CMM.
In order to address the question of whether ozone depletion can be
reasonably anticipated to increase CMM, this review has been designed to
address three sub-questions:
1. Does the evidence support the hypothesis that for
susceptible populations, solar radiation is a cause of
melanoma?
2. Does the evidence support the hypothesis that UVB is a
major component of solar radiation which causes
melanoma?
3. What dose-response relationships between melanoma and
UVB are consistent with the epidemiological and
experimental data?
There are certain points which must be recognized before addressing the
questions presented above. In our review, we found no perfect studies nor any
single overwhelming piece of evidence that answered any of these questions.
In fact, as indicated in Table 19-1, we found a number of serious limitations
to the current database.
TABLE 19-1
LIMITATIONS TO THE DATABASE
1. There is no experimental animal model.
2. There is no experimental in vitro model for malignant
transformation of melanocytes.
3. There are no epidemiologic studies of CMM where individual
human UVB exposures have been adequately addressed.
* * * DRAFT FINAL * * *
-------
19-2
Consequently, this review had to build on many diverse pieces of evidence
from many different disciplines. Most of the evidence appears to support a
clear response to these questions, but some appears contradictory.
Consequently, in doing this analysis we have had to focus on the weight of the
evidence, the likelihood of various answers and the best approximations of the
dose-response relationship, not on trying to definitively prove an answer to
any of the questions.
BACKGROUND
Chapters 2 and 3 summarized important information about ambient solar
radiation, the skin and the nature of CMM relevant to understanding the
epidemiologic and experimental evidence reviewed in this document. Table 19-2
summarizes key points from these chapters. The central implication of all
these summary points is that uncertainty exists about how to determine both
the appropriate dose of solar radiation and the appropriate response -- dose
will be one thing if the important wavelengths are UVB, another if UVA.
Responses of one histological type of melanoma may vary from another. Studies
that focus on sun-exposed hours and lump all melanoma types together suffer
from aggregating what may be very different things.
IS SOLAR RADIATION A CAUSE OF MELANOMA?
The first question addressed is:
Does the evidence support the hypotheses that, for susceptible
populations, solar radiation is a cause of melanoma?
Table 19-3 presents a list of findings that can be interpreted as
supporting the conclusion .that solar radiation is one of the causes of
cutaneous melanoma.
The first three points in Table 19-3 all suggest that the risk of
developing CMM is somehow inversely associated with degree of pigmentation.
Thus blacks1 have less risk than whites, and whites with darker pigmentation,
either constitutional or acquired via tanning, are at less risk than
light-skinned whites. This information, in view of the fact that melanin, the
major skin pigment, is a strong absorber of solar radiation, lends support to
belief that solar radiation plays a role in CMM development in that it
suggests that the presence of melanin by reducing the effective dose of solar
radiation penetrating through the skin reduces the likelihood of CMM
induction. Alternative explanations are possible; for example melanin in the
skin could act as an anticarcinogen in general, but such explanations are less
straightforward and have a weaker factual basis.
* * * DRAFT FINAL * * *
-------
19-3
TABLE 19-2
SUMMARY POINTS RELEVANT TO ASSESSING
THE EPIDEMIOLOGICAL AND EXPERIMENTAL EVIDENCE
Ozone differentially removes wavelengths of UVB (295-320 nm) and does
not remove UVA (320-400 nm) or visible light (400-900 nm).
Wavelengths between 295 nm and 300 nm are generally much more
biologically effective (damage target molecules in the skin including
DMA) than other wavelengths in UVB and even more so than UVA radiation.
The effective dose that actually reaches target molecules depends on
the duration of exposure at particular locations, time of day, and
time of year, on behavior in terms of clothes and sunscreens,- and on
pigmentation and other characteristics of the skin including temporal
variations (e.g., in tanning and in skin thickening).
Latitudinal variations exist in solar radiation; model predictions
indicate that the greatest variability is seen in cumulative UVB
(e.g., monthly doses), followed by cumulative UVA and then peak UVB
(highest one-day doses). Peak UVA varies very little across latitude
up to 60°N.
Cloudiness and albedo, although causing large variations in the amount
of UVB and.UVA, do not greatly change the ratio of UVB to UVA.
Ozone depletion is predicted to cause the largest increases in the
295-300 nm UVB range, less in 300-320 nm UVB range and none in UVA.
Cutaneous malignant melanoma has a number of different histologic
. types which vary in. their relationship to sunlight, their site and
racial preference and possibly in their precursor lesions; assessment
of these types is not consistent among registries, confounding
attempts to evaluate associations between CMM and solar radiation.
Melanin is the principal pigment in skin that gives it color; melanin
absorbs UV radiation very effectively, thus the darker the skin the
more the basal layer is protected from UV.
* * * DRAFT FINAL *
-------
19-4
TABLE 19-3
INFORMATION THAT HAS BEEN INTERPRETED AS SUPPORTING
THE CONCLUSION THAT SOLAR RADIATION IS ONE OF
CAUSES OF CUTANEOUS MALIGNANT MELANOMA (CMM)
• Whites have higher CMM incidence and mortality rates than blacks.
• Light-skinned whites including those who are unable to tan or who tan
poorly, get more CMM than darker-skinned whites.
• Sun exposure leading to sunburn apparently induces melanocytic nevi.
• Individuals who have more melanocytic nevi, develop more CMM; the greatest
risk is associated with a particular type of nevus--the dysplastic nevus.
• Sunlight induces freckling, and freckling is an important risk factor.
• Incidence has been increasing in cohorts in a manner consistent with
changes in patterns of sun exposure, particularly with respect to
increasing intermittent exposure of certain anatomical sites.
• Immigrants who move to sunnier climates have higher rates of CMM than
populations in their country of origin and develop rates approaching those
of the adopted country; this increase, in risk is particularly accentuated
in individuals arriving before the age 'of puberty (10-14 years).
• CMM risk is associated with childhood sunburn; this association may
reflect an individual's pigmentary characteristics or may be related to
nevus development.
• Most studies that have used latitude as a surrogate for sunlight or UVB
exposure have found an increase in the incidence or mortality of CMM as
one approaches the equator.
• Patients with xeroderma pigmentosum who cannot repair UVB-induced lesions
in skin DNA have a 2000-fold increased risk of CMM by the age of 20.
• One form of CMM, Hutchinson's melanotic freckle melanoma, appears almost
invariably on the chronically sun damaged skin of older people.
* * * DRAFT FINAL *
-------
19-5
Further reinforcing the solar radiation hypothesis is the fact that the
development of nevi and freckles appears to be a response to solar radiation
and that there is apositive association between these conditions and the risk .
of CMM. Some argue that the development of nevi, possibly dysplastic nevi,
may be a step in CMM related to solar exposure.
The rise in CMM rates during this century has been cited as both
supporting an involvement of solar radiation in CMM and as not supporting it.
Those who argue it does not support an association of CMM with solar radiation
point out that ozone depletion has not occurred, to our knowledge, in this
century. Those who argue the rising rates are positive evidence argue that
shifts have occurred in patterns of dress and recreational sun exposure, thus
increasing intermittent solar exposure to those sites showing the greatest
increase.
Immigrant studies provide further support to the solar radiation .
hypothesis because they show that CMM rates rise in immigrants who move to
countries with more solar radiation than in their lands of origin.
Furthermore, the longer an immigrant resides in such a high exposure country
the more his/her risk increases. Part of this increased risk may be due to
exposure during some critical event in childhood; there is some indication
that arrival before the age of 10-14 is associated with increased risk. An
alternative explanation, however, is that childhood is a better time to
influence behavior patterns of high sun exposure.
Many ecologic studies show that increasing incidence or mortality of CMM
is inversely correlated with latitude. Latitude is closely correlated with
UVB flux. . While these, studies do not control for differences in skin
pigmentation, for the ratio of indoor and outdoor workers, or for differences
in behavior, these characteristics (at least in the U.S.) are probably
randomly distributed with respect to latitude. Thus, the differences in
behavior, skin pigmentation, and indoor/outdoor worker ratios could be
expected to introduce only a smalll amount of noise into the analyses,
reducing the amount of variation in CMM incidence or mortality that can be
explained by UVB dose, but not enough to obscure the fundamental
relationship. The fact that a number of studies have found a relationship and
that the best studies have done so, lends support to the solar hypothesis.
The presence of high rates of CMM in xeroderma pigmentosum patients
indicates that a component of solar radiation, UVB, could be important in CMM
etiology, and again adds another different kind of evidence to the array of
evidence supporting the conclusion that solar radiation is etiologically
related to CMM.
* * * DRAFT FINAL * *
-------
19-6
Table 19-4 presents information that has been interpreted as not
supporting, being in conflict with, or refuting the solar radiation
hypothesis. The points discussed here are, in fact, the fundamental conundrum
of this issue.
The failure to find latitudinal gradients of CMM incidence or mortality in
some ecological studies has been interpreted by some to introduce uncertainty
into the relationship between solar radiation and CMM. In contrast, many of
the original researchers and others have suggested that in the failure to find
a correlation between latitude and CMM was a product of an inverse gradient of
skin pigmentation (i.e., dark-skinned whites in the south, and fair-skinned
whites in the north) or a lack of a UVB gradient north to south in the study
area.
Although it would have reinforced the solar hypothesis greatly to have all
latitudinal studies in agreement, the main conclusion to be drawn is the need
to control for skin pigmentation, behavior, and actual doses of solar
radiation by wavelength in such studies.
The difference in CMM rates between indoor and outdoor workers is probably
the most difficult information with which to reconcile the solar radiation
hypothesis. If solar radiation is a key factor, how is it possible that
outdoor workers, who presumably get a large total dose of solar radiation than
indoor workers, have lower incidence rates? This may be a question of
definition. For example, one study, which looked only at highly exposed
sites, found that outdoor workers had higher CMM rates than indoor workers.
Another study, which separated indoor workers into professionals and "others"
showed no difference between the indoor "other" classification and outdoor
workers, suggesting the hypothesis that higher exposure due to opportunities
associated with higher socio-economic status may explain this issue.
Unfortunately, no studies have computed the biologically effective dose of
solar radiation, and it is uncertain how much UVB indoor and outdoor workers
receive and how their skin pigmentation differs when they are exposed.
Consequently, the only conclusion that can be reached with certainty about the
differences between indoor and outdoor workers is that an estimate of UVB dose
based on total hours of solar exposure does not correlate well with CMM
incidence. Until epidemiologic studies are done which compute biologically
effective dose, not solar hours of exposure, no more can be said with
certainty.
Because in comparison to basal cell (BCC) and squamous cell cancers (SCC),
CMM occurs more frequently on intermittently-exposed than continuously-exposed
sites clearly indicates that CMM is related to solar radiation exposure in a
way different from BCC and SCC. Many hypotheses have been put forward to
explain why CMM behaves differently ranging from a requirement for early
childhood exposures, or intermittent or excess exposures, e.g., large doses
prior to tanning or sunburns. Unfortunately, at this time, the epidemiologic
and experimental databases do not allow us to determine which, if any, of
these hypotheses is true. Thus, this observation weakens the case for solar
exposure at least as it is now being measured (sunlit hours).
* * * DRAFT FINAL * * *
-------
19-7
TABLE 19-4
INFORMATION THAT HAS BEEN INTERPRETED AS NOT SUPPORTING
THE CONCLUSION THAT SOLAR RADIATION IS ONE OF
CAUSES OF CUTANEOUS MELANOMA
• Some ecologic epidemiology studies have failed to find a latitudinal
gradient for CMM.
• Outdoor workers generally have lower incidence and mortality rates for CMM
than indoor workers.
• Unlike basal cell and squamous cell carcinomas, most CMM occurs on sites
that are not habitually exposed to sunlight.
* * * DRAFT FINAL * *
-------
19-8
This finding indicates that CMM, if it is etiologically linked to solar
radiation, must be influenced not by total hours exposed to solar radiation
but by some other dose parameter. In essence, the evidence that adds
uncertainty to the role of solar radiation in the induction of CMM suggests a
relationship between induction and a biologically effective dose rather than
between induction and a simple measure of cumulative exposure. Thus, it might
be possible to resolve this inconsistency if one could compare the UVB
exposures of indoor and outdoor workers through time. Research must focus on
determining the UVB energy actually received at the surface of the skin and,
if possible, the penetration of that radiation as tanning takes place.
BALANCING THE EVIDENCE
The supporting evidence that solar radiation is a causal factor in CMM is
not indisputable. One must admit that a possibility exists that CMM is not
related to solar radiation. On balance, however, a web of evidence taken as a
whole supports solar radiation as at least one of the causes of CMM. One can
find a number of alternative explanations for each piece of evidence, but, in
general these do not fit together to provide a meaningful alternative
explanation to solar radiation. The likelihood that the array of evidence
pointing to solar radiation is either all wrong or is completely explained by
a series of other unknown factors is low.
The evidence that fails to support or apparently contradicts the issue
does not refute the solar radiation hypothesis. Rather, it supports the view
that CMM is a complex disease. On balance, the evidence seems to support
solar radiation as a cause of CMM, but does not support the hypothesis that •
the appropriate measure of dose is sunlit hours of exposure.
IS UVB A WAVEBAND OF SOLAR RADIATION THAT CAUSES CMM?
The second question addressed is:
Does the evidence support the hypothesis that UVB is a major component
of sunlight which causes cutaneous melanoma?
~ In Table 19-5, we present summary points which we believe support the
hypothesis that UVB is most likely to be the major component of solar
radiation which causes cutaneous melanoma, whereas information which has been
considered in contradiction to this hypothesis, e.g., the lack of an animal
model, is discussed in the text principally as flaws in the database.
Information on patients with xeroderma pigmentosum is perhaps the
strongest evidence that we have on a role for solar radiation and, in
particular, UVB, in the etiology of CMM. These individuals comprise a
susceptible population that is at considerably increased risk for melanoma.
The major defect associated with this condition is an inability to repair
pyrimidine dimers--a lesion in DNA whose action spectrum principally spans the
UVB region. There is little information which' addresses the issue of whether
it is appropriate to extrapolate from this susceptible population to the
general population, nevertheless we believe that the fact that there is a
human population known to be extremely sensitive to UVB which is also
extremely susceptible to melanoma strongly supports a role for UVB in CMM.
* * * DRAFT FINAL * * *
-------
19-9
The second and third points of Table 19-5 are similar; one states the
evidence which supports the conclusion on the basis of in vitro experiments,
the other is based on information derived from in vivo experiments. The point
to be made is that UVB is the most active portion of the solar spectrum in
inducing adverse effects that are related to carcinogenesis. It is true that
other portions of the spectrum, in particular UVA, can cause adverse effects,
but these wavelengths are several orders of magnitude less effective at
inducing most types of damage. The final point is that UVB is the most active
wavelength at inducing UVB-induced tumor-specific immunosuppression thus
further adding credence to the hypothesis that UVB is the most likely
candidate waveband in solar radiation to be the cause of CMM.
There are several points derived from the experimental literature which
are often cited as evidence which does not support a role for solar radiation
in CMM, which we have called weaknesses in the database. The two key ones
relevant to this issue are the lack of an animal model and the lack of an in
vitro correlate. The lack of positive evidence certainly is a weakness in the
database. Efforts have been expended to develop an animal model in mice—but
have only been partially successful; those which have partially succeeded have
required treatment of animals with a carcinogen followed by UVR treatment. In
several experiments such treatments induced nevi that later became melanomas.
This suggests that a possible reason for the lack of animal model is that we
do not yet fully.understand the role of precursor lesions in the disease and
thus we cannot develop an adequate animal model.
The lack of an in vitro correlate may be due to a similar sort of
technical problem. Melanocytes are difficult cells to grow in culture and it
is only recently that the growth in vitro has become routine. It is likely
that this difficulty has impeded the development of an in vitro system for
melanocyte transformation by UVB. Of course, most data from negative
experiments is never published so it is not possible to judge if adequate
experiments have been carried out and not reported or if technical
difficulties have precluded the design and performance of adequate experiments.
BALANCING THE EVIDENCE
The evidence that UVB is a major component of solar radiation which could
cause cutaneous melanoma is- not absolute. Clearly the evidence has
limitations as indicated by the lack of either in vivo or in vitro evidence
that UVB can cause transformation of melanocytes. However, there is no
evidence which would suggest any alternative hypothesis, and the available
experimental evidence indicates quite strongly that UVB is the waveband most
likely to be carcinogenic, not only in animals but also (as evidenced by what
is known about non-melanoma skin cancer in general and melanoma in xeroderma
pigmentosum patients) in humans. Given this conclusion, it is reasonable and
prudent to examine possible dose-response relationships for UVB and melanoma.
* * * DRAFT FINAL * *
-------
19-10
TABLE 19-5
INFORMATION THAT .SUGGESTS THAT UVB
IS THE MAJOR COMPONENT OF SOLAR RADIATION
WHICH CAUSES MELANOMA
• Xeroderma pigmentosum patients who fail to repair UVB-induced DNA damage
(pyrimidine dimers) have a 2000 fold excess rate of CMM by the time they
are 20.
• UVB is the most active part of the solar spectrum in the induction of
mutagenesis and transformation in vitro.
• UVB is the most active part of the solar spectrum in the induction of
carcinogenesis in experimental animals and is considered by most to be a
causative agent of non-melanoma skin cancer in humans.
• UVB is the most active portion of the solar spectrum in inducing
immunosuppression which may have a role in melanoma development.
* * * DRAFT FINAL * * *
-------
19-11
WHAT DOSE RESPONSE RELATIONSHIPS ARE CONSISTENT WITH THE DATA?
The last question to be addressed is:
What dose-response relationships between melanoma and DVB are
consistent with the epidemiological and experimental data?
Table 19-6 summarizes information relevant to evaluating which
dose-response relationships are consistent the epidemiologic and experimental
data.
In evaluating dose-response relationships for risk assessment - two types
of data might be used: that which comes from animal experiments or that which
comes from epidemiologic studies. In the case of CMM, there are no
appropriate animal data. There is no animal model for the induction of CMM by
UVB, and the data linking UVB exposure to non-melanoma skin cancer (NMSC) in
animals is probably inappropriate given the differences in biologic behavior
between NMSC and CMM in humans.
Among the available epidemiologic studies, there is no perfect study. The
available case-control studies which assessed exposure to solar radiation via
questionaire do not provide information with which to assess UVB exposures and
so are not useful for estimating dose-response relationships. There are,
however, several ecologic studies in which measured or estimated exposures to
UVB at various locations were correlated to CMM incidence or mortality at
those locations. Two of the most recent studies took slightly different
approaches. In one, actual annual UVB measurements from Robertson-Berger
meters located in seven cities were converted to biologically effective units
using a DNA action spectrum and these estimated doses were correlated to
incidence data from those seven cities using a power model to relate incidence
to UV units. Also incorporated into this model was a term which provided some
adjustment for host or environmental characteristics. The analysis was
segregated by anatomic site so that separate estimates were made for trunk and
lower extremities (trunk/LE) versus face, head and neck and upper extremities
(FHN/UE). This analysis predicted that a 10 percent increase in UVB would
result in a 6 and 5 percent increase in CMM of trunk/LE for males and females,
respectively, and 8 and 10 percent increase in CMM of FHN/UE in males and
females, respectively.
A second study derived UVB estimates by county based on a NASA satellite
model and weighted by a variety of different action spectra. These were then
correlated with county-based mortality data using two different model forms.
When the estimate of UVB was based on clear day peak flux weighted by the DNA
action spectrum, this study predicts that a ten percent change in this measure
of UVB will be associated with an 8.4 percent increase in CMM mortality in
males and a 5.7 percent increase in females. Preliminary analysis suggests,
however, that these estimates are very sensitive to the choice of UV flux
measure, i.e., peak or annual estimates, as well as the choice of action
spectrum. Thus for example, the estimate given above may overstate the
responsiveness of CMM mortality to changes in UVB flux by as much as 60
percent as compared to estimates using different action spectra or an annual
estimate of flux.
* DRAFT FINAL *
-------
19-12
TABLE 19.6
KEY POINTS RELEVANT TO EVALUATING DOSE-RESPONSE RELATIONSHIPS
There are no animal data appropriate for developing dose-response
estimates for CMM; there is no animal model for CMM and animal
non-melanoma skin cancer (NMSC) data are not appropriate given the
differences in biologic behavior between CMM and NMSC in humans.
Among epidemiologic studies, the most promising for estimating
dose-respone relationships are ecologic studies in which measured or
estimated exposures to UVB at various geographic locations are correlated
with CMM incidence or mortality at those locations.
One such study evaluated the relationship between measured levels of UVB
(annual R-B meter counts) in seven cities and CMM incidence rates in those
seven cities; this study predicts that a 10 percent change in UVB will be
assciated with about a 7 percent increase in CMM incidence in males and
7.5 percent increase in females.
A second study evaluated the relationship of model-based estimates of
peak day UVB flux (weighted for DNA damage) with CMM mortality using
county-based CMM mortality data; this study predicts that a ten percent
change in UVB will be associated wih an 8.4 percent increase in CMM
mortality in males and 5.7 percent increase in females.
* * * DRAFT FINAL *
-------
19-13
There are still many uncertainties in these estimates of dose-response
relationships particularly as they apply to future populations. As a result,
estimates of the appropriate dose-response relationship between UVB exposure
and CMM (incidence and mortality) continue to be developed. In particular,
the epidemiologic studies conducted to date do not account for population
shifts, particularly with regard to ethnicity. Also, these studies cannot
adequately control for host factors such as pigmentation because of a lack of
the appropriate data. For example, if pigmentation varies systematically with
latitude then the inability to control for it would bias the resulting
estimates. There are other variables that might show systematic variation by
latitude or geographic location (e.g., socioeconomic status). Again should
such variation occur it could bias the estimates. Other problems yet to be
solved in epidemiological studies include: 1) selection of the correct
functional form to describe dose-response, 2)controlling fully for genetic and
host characteristics such as mode of dress and sunscreen use, and 3) the
development of an action spectrum specific for CMM which could be used to
weight solar radiation in order to derive a better estimate of UVB dose.
* DRAFT FINAL *
-------
APPENDIX A
PREPARATION OF THE DOCUMENT
INTRODUCTION
The goal of this appendix is to provide information as to the manner in
which this document has been prepared. Included in it is a description of the
process by which the literature in this document was identified, reviewed,
organized into topics and incorporated into chapters.
As indicated in the overview of the document, the goal of this document
was to critically review the available experimental and epidemiologic'evidence
pertinent to determining if it can be reasonably anticipated that a change in
stratospheric ozone and a resulting change in the amount of UVB delivered to
the earth's surface would result in a change in the incidence or morbidity of
cutaneous malignant melanoma (CMM).
The review was designed to be comprehensive but places particular emphasis
on information entering the literature after the latest National Academy of
Sciences report (NAS 1983). One very recent publication, The Epidemiology of
Malignant Melanoma, Volume 102 of Recent Results in Cancer Research, was not
available in time to be included systematically, although, where possible, we
have included that information in the relevant chapters. The articles of
Gallagher et al., Holman et al., Green et al., and Osterlind and Jensen are
particularly relevant to the issues discussed in this review; their findings
and analyses tend to support the conclusions reached in this document.
* * DRAFT FINAL * *
-------
A-2
LITERATURE RETRIEVAL
The retrieval of literature for this project was performed principally on
the National Library of Medicine's Elhill databases: MEDLINE, CANCERLINE,
SDILINE and TOXLINE, although some searches on the DIALOG databases were also
made. Initially the search strategy consisted of printout of a search based
on the keywords melanoma cross-referenced with solar or ultraviolet or light
or sun or sunlight. Abstracts from this search were selected and the litera-
ture for them was retrieved. Subsequent searches were much more focussed; in
all there were more than 100 such searches run. An example - the search
strategy used to research the question of animal models for primary (as
opposed to transplanted) cutaneous malignant melanoma is presented in
Exhibit Al.
The search strategies generally were designed with one of two objectives,
either to retrieve as much as was known about a particular area of interest,
e.g., animal models for primary melanoma, or to pinpoint the key information
or the most current information on a particular point, e.g., the generation of
superoxides following irradiation of melanin with UVB. In some instances,
author searches were performed. For example, in reviewing the epidemiologic
evidence it was clear that certain authors published heavily in a given area.
Some of the authors for whom such searches were made include: C.D.J. Holraan,
A. Green, M.H. Greene, J.A. Lee, J. Scotto, V. Beral, A. Swerdlow,
P. Armstrong, A. Houghton, C.M. Balch, I. Crombie, K. Magnus, J. Elwood, D.E.
Elder, and W.H. Clark.
With the exception of the very first more or less global searches, all of
the searches on the ELHILL databases were performed by the information
specialist/project manager in response to questions raised during the review
or chapter-drafting processes. In addition to the online searches, new
journals in the relevant areas were also checked on a weekly or biweekly
basis. A list of the journals that were routinely searched in this manner is
presented in Exhibit A2. Also as a way of ensuring that the information
reviewed was kept current, ELHILL*S most current month file - SDILINE - was
checked on a regular basis for anything relevant to melanoma, ultraviolet
radiation and carcinogenesis.
SEARCHABLE BIBLIOGRAPHIC DATA BASE
About midway into the literature identification and retrieval for this
project, it became apparent that identifying the relevant reports among the
vast amount of literature retrieved might be facilitated by a relational data
base of the articles, searchable by keywords. Accordingly, a key wording
process was instituted. Exhibit A3 presents the coding sheet used and Exhibit
A4, the definitions of each keyword. The data base was designed in
DBaselll® and is searchable principally by keywords but also, to a limited
extent, by first author. Keywords can be "ored" or "anded". Originally, the
file required DBaselll® in order to be used, it was subsequently compiled
using the program Clipper® and now runs as a "stand-alone". The program
requires an IBM-PC/XT or AT but is fairly slow on the XT.
* * * DRAFT FINAL * * *
-------
A-3
The sytem proved very useful at identifying the literature already
in-house. One problem encountered, however, was that authors in writing their
chapters would tree-search to identify new information and those new articles
were often delayed in being put on the data base.
THE REVIEW PROCESS
The review process was carried out by the team of contributors indicated
on the title page. Initially the review process involved in-depth evaluation
of the same key studies by the group as a whole, followed by meetings in which
the reviews were discussed and the key findings identified, and verified. The
purpose of these reviews and meetings was to ensure that the reviewers who
were later to be authors of various chapters in the document would start with
the same background information and knowledge of the area, and would approach
the analysis in a similar fashion.
Information to be reviewed was divided into chapters and assignments were
made to the various contributors. The division into chapters was based on the
findings to be examined; as the review process proceeded, the material was
reorganized several times in order to develop the final organization presented
here.
Much of the new information dealt with a number of new epidemiological
studies. Because of the importance of this information, detailed critical
reviews of these studies were provided by Dr. Ralph Buncher and David
Warschowsky fo the Unviersity of Cincinnati Medical School, Department of
Epidemiology. Those reviews, as well as critical reviews of other important
epidemiologic references done by Dr. Buncher and Ms. Sarah Foster are
presented in the following Appendix B.
CHAPTER DRAFTING AND REVIEW
First drafts of each chapter were prepared and submitted to the other
contributors for review. Chapters reviewing the epidemiologic evidence were
written by either Dr. Saftlas, Ms. Knox or Ms. Foster and were critically
reviewed principally by Dr. Aparna Koppikar,but also by Dr. Saftlas, Ms. Knox
and Ms. Foster. Chapters reviewing the experimental evidence were principally
written by Drs. Lill, Longstreth and DeFabo, with internal reviews being
provided by contributors other than the author. The introductory chapter on
melanoma and the chapter on predisposing lesions which were combinations of
experimental and epidemiolgic evidence were written by Dr. Longstreth and were
kindly reviewed by Dr. David Elder of the Pigmented Lesion Group in
Philadelphia, as well as by the other contributors. The chapters on
variations in UV by latitude and altitude and on the dose-response
relationships for melanoma were written by Mr. Hugh Pitcher and reviewed by
Dr. Frederick and Mr. Scotto respectively. The document was given a detailed
review by Dr. E.A. Emmett of the Johns Hopkins School of Public Health and Mr.
J. Scotto of the National Cancer Institute.
* * * DRAFT FINAL * * *
-------
A-4
EXHIBIT A1
SEARCH STRATEGY FOR ANIMAL MODELS OF PRIMARY CUTANEOUS
MALIGNANT MELANOMA
SEARCH FORMULATION SE-j INNING AT 33 : :
CALL MELANOMA AND ALL ANIMAL* ' — 37'^S COSTINGS
SEARCH FORMULATION BEGINNING AT S3 2 :
:.3S 1 AND NOT ALL SIS > — 2B27 POSTINGS
SEARCH FORML'LATTGN BEGINNING AT 33 3 :
(SS 2 AND NOT ALL CULTURE* ) — 1-377 POSTINGS
SEARCH "SMljLATtON BEGINNING AT S3 4 :
CSS 3 AND NOT ALL VITRO > — 1623 POSTINGS
SEARCH'FORMULATION BEGINNING AT SS 3 : '
CSS 4 AND NOT ALL TRANSPLANT: > — 1261 POSTINGS
SEARCH FORMULATION BEGINNING AT 53 6 .-
CSS 5 AND NOT ALL DRUG* J — 1123 POSTINGS
SEARCH FORMULATION BEGINNING AT-SS 7 :
.CSS 6 AND NOT HUMAN CTUi) 1 — 733 POSTINGS
SEARCH FORMULATION BEGINNING AT SS 8 :
CSS 7 AND NOT PATIENT CTU) J — 710 POSTINGS
SEARCH FORMULATION BEGINNING AT SS '3 :
CSS 8 AND ALL MODEL* ') — 59 POSTINGS
SEARCH FORMULATION BEGINNING AT SS 1O :
CSS 8 AND NOT ALL UVEAL OR SS 8 AND NOT ALL EYE* :>
70S POSTINGS
SEARCH FORMULATION BEGINNING AT SS 11 :
CSS 10 AND NOT ALL B-IS ) — 632 POSTINGS
SEARCH FORMULATION BEGINNING AT SS 12 -•
CSS 11 AND NOT ALL CLOUDMAN ) — 510 POSTINGS
SEARCH FORMULATION BEGINNING AT SS 13 :
>:SS 12 AND NOT ALL UVEAL > — 605 POSTINGS
SEARCH FORMULATION BEGINNING AT S3 14 :
CSS 13 AN9 ALL ENG CLA.1 :> — 523 POSTINGS
SEARCH FORMULATION BEGINNING AT SS 15 :
<.SS 14 AND NOT ALL DF.'JfS: ) — 4SO POSTI.VGS
* * * DRAFT FINAL * * *
-------
A-5
EXHIBIT A1 (cont)
SEARCh FOPr-v.'LATION BEGINNING AT S3 17 :
"rS 15 AND :;QT ALL PHARMACEU7: > — -30 POSTINGS
SEARCH FC~r!_."_ATION BEGINNING AT S3 18 :
<12 1~ AND NOT ALL HORSE* OF: S3 17 AND NOT ALL EQUINE* :>
473 POSTINGS
SEARCH FORMULATION BEGINNING AT S3 1'3 :
',£3 18 A.\'D NOT ALL COW* !> — 475 POSTINGS
SEARCH FORMULATION BEGINNING AT S3 20 :
(S3 1-3 AND NOT ALL OPTHAL: '< — 475 POSTINGS
SEARCH FORMULATION BEGINNING AT S3 21 :
(S3 19 AND NOT ALL OPHTHAL: ) — 470 POSTINGS
SEARCH FORMULATION BEGINNING AT SS 22 :
(S3 21 AND NOT ALL NEWT: !> — 463 POSTINGS
SEARCH FORMULATION BEGINNING AT S3 23 :
CSS 22 AND NOT ALL PROSTAGLANDIN: AND SS 22 AND NOT ALL PLASMINOGEN:
457 POSTINGS
SEARCH FORMULATION BEGINNING AT S3 24 :
CSS 23 AND NOT ALL CHEMO: ) — 423 POSTINGS
SEARCH FORMULATION BEGINNING AT S3 25 :
CSS 24 AND NOT ALL INTRAOC: ) — 413 POSTINGS
NO SUBHEADINGS APPLIED TO ANY SEARCH STATEMENT.
SF,C
USER:
RESTACK
PROG:
S3 2 /C?
USER :
1 AND NOT ALL COMPLEX:
PROG:
33 i:2> PSTG (401J
ss 2 />:•••
USER:
9 AND NOT ALL MONOCLONAL*
r-ROG:
NF -:3.-.
t.NONE-
2 AND NOT ALL MONOCLONAL
»?£!•••
* DRAFT FINAL * * *
-------
A-6
EXHIBIT A2
CURRENT JOURNALS FOLLOWED
Carcinogenesis
Journal of Investigative Dermatology
Photodermatology
Journal of the National Cancer Institute
Photochemistry and Photobiology
Human Pathology
International Journal of Cancer
American Journal of Epidemiology
International Journal of Epidemiology
Cancer Research
Cancer
British Journal of Dermatology
International Journal of Dermatology
Archives of Dermatology
British Journal of Cancer
Journal of Experimental Medicine
The American Journal of Dermatopathology
Journal of American Academy of Dermatology
* * * DRAFT FINAL * * *
-------
A-7
EXHIBIT A3
KEYWORD CODING FORM
Coded by:
QC:
KEYWORD
Citation:
10 melanoma
20 non-melanoma
30 economic impact
40 letter/abstract
SO skin/cutaneous (nos)
60 review/editorial
70 action spectrum
80 dose-response
90 neoplasia
91 initiator
92 promoter
100 HUMAN STUDY
110 histopathologic type
120 case report
130 treatment technique
140 pregnancy/estrogen
ISO site/racial differences
1S5 racial/ethnic
160 clinical diagnostic factors
170 genecic factors
17S skin characteristics
180 environmental .factors
181 smoking
182 diet/nutrition
200 EPIDEMIOLOGICAL
205 Descriptive
210 case-control
220 cohort study
22S time trends
230 cross-sectional
240 retrospective
250 prospective
260 migrant
26S occupational
270 socioeconomic status
280 age/sex
290 incidence
291 mortality
292 survival
300 ANIMAL STUDY
310 mouse
320 guinea pigs
330 hamster
340 tumor transplant
3SO chemical carcinogen
400 CELLULAR (in vitro)
410 transformation
411 differentiation/development
420 raelanocyte/raelanoblast
430 keratinocyte
440 squaoous cell
450 epidermis/epidermal cell
460 nevus/nevus cell/mole
470 basal cell
480 stratum comeum
500 MOLECULAR
510 urocanic acid/histidine
S20 DNA damage
530 melanin/tyros ine
540 pigment
S50 mechanism/hypothesis-
600 IMMUNOLOGIC/lymphocyte (nos)
610 la antigens (class II MHC)
620 Langerhans cells
630 T-lymphocytes
640 B-lymphocytes
650 autoimmunity
660 antibody
670 immunosuppression
700 CATARACT/eye (nos)
710 cornea
720 lens
730 retina
800 SUNLIGHT (nos)
810 ultraviolet light
811 UV-A
812 UV-B
820 latitude gradient
330 fluorescent lights
340 erythema/sunburn
350 ionizing radiation
860 exposure characteristics
361 intensity/duration
900 VIRUSES
910 Herpes
* * * DRAFT FINAL
* *
-------
A-8
EXHIBIT A4
KEYWORD DEFINITIONS
10 - Melanoma: use for anything that references melanoma.
20 - Non-melanoma: use for any study that refers to non-melanoma
skin cancer, either basal or squamous cell.
30 - Economic impact: Use for studies which address costs of
treatment, costs of disease (hospitalization) loss of work time, etc.
40 - Letter/abstract: Any article which is a meeting abstract or a
letter to the editor.
50 - Skin/cutaneous (nos): Use for studies that mention the skin
or cutaneous diseases/eruptions
60 - Review/editorial: Any article that identifies itself as a
review article or any article which provides no new data, e.g., an
editorial.
70 - Action spectrum: Any studies that evaluate several.
wavelengths of UV light to determine which is the most active at causing
an effect.
80 - Dose-response: Studies which report a relationship between
dose and response, e.g., by investigating the different level of effects
of several doses.
90 - Neoplasiacarcinogenesis: Any studies having to do with
benign or malignant tumors or growth. Use also for theories of
carcinogenesis.
91 - Initiator: Any cancer causing substance which can initiate a
neoplastic event, i.e., which will induce a tumor without subsequent
application of a promoter.
92 - Promoter: A substance which is not itself carcinogenic but
which increase the carcinogenic effect of an initiator.
100 - HUMAN (study): Any study dealing with humans or human cells
except for those studies based on populations, i.e., epidemiological
studies.
110 - Histopathologic type: Use this if the study differentiates
among different types of tumors or cells, e.g., draws different
conclusions for superficial spreading melanoma vs nodular melanoma.
* * * DRAFT FINAL * * *
-------
A-9
EXHIBIT A4 (cont)
120 - Case report: Use for human studies that report results of
medical cases, e.g., physicians report of patients presenting with
different kinds of melanoma.
130 - Treatment technique: Use for reports evaluating various
treatments, e.g., cryosurgery, surgery, chemotherapeutics radiation.
140 - Pregnancy/hormone: Use for the evaluation of impact of
pregnancy, birth control pills or post-menopausal hormones. Should also
be used for homones (nos).
150 - Anatomical differences: Use if report compares or constrasts
different rates among anatomical sites
155 - Racial/ethnic: Use if reprot compares or contrasts
differneces among races or ethnic groups.
160 - Clinical/diagnostic factors: Use if a report evaluates
disease states with regard to staging of disease, or
prognostic/diagnostic factors or clinical parameters which help predict
disease onset or progression .
170 - Genetic factors: Use if report evaluates familial history or
genetic factors/markers such as HLA antigens.
180 - Environmental factors: Use if report evaluates other
environmental factors such as smoking, nutrition, chemicals
190 - Skin characteristics: use for a study that evaluates various
skin characteristics e.g., types I,II,III or freckles or degree of
pigmentation. May also be used for studies that deal with abnormalities
of the skin such as vitiligo.
200 - EPIDEMIOLOGICAL: Use if report is an epidemiologic study or
if a review summarizes epidemiologic information.
205 - DESCRIPTIVE: Use if study summarizes trends in a data set,
such as trends in incidence or mortality with respect to age, site, sex,
or occupation. Do not use for a study which analyzes and evaluates
potential cause/effective relationships.
210 - Case-control: The individuals in the study are selected
based upon whether they have the disease of interest (cases) or not
(controls). In general, it is preferable to choose incident rather than
prevalent cases within a specified time period. Case-control studies may
be retrospective (if exposure or cause information is for a time period
proceeding the disease occurrence) or nondirectional (if disease and
* * * DRAFT FINAL
-------
A-10
EXHIBIT A4 (cent)
exposure information are from the same time period). Case-control
studies can be used for relatively infrequent diseases, although a
problem often arises in remembering or documenting exposure (i.e., cause)
information.
220 - Cohort: The individuals to be studied (the cohort) are
defined based upon characteristics manifest before the appearance of the
disease being studied. The cohort is then followed up to determine the
frequency of disease in it. Cohort studies usually measure incidence or
mortality. A prospective cohort study is one in which the cases of
disease have not occurred at the time the study has begun, but the causes
may or may not have occurred. A retrospective cohort study is one in
which all causes and effects have already occurred at the time the study
is initiated. (Note that retrospective and prospective area, in this
terminology, used to describe the time of occurrence of the events being
studied relative to the study investigator's place in time.) Prospective
cohort studies are only economical when the disease is relatively
frequent. For rare diseases, very large cohorts are needed. In a cohort
study, the number of cases in the cohort may be compared with the
expected number of cases in a reference population. The reference
population should be comparable to the cohort with respect to factors
other than the exposure of interest.
225 - Time trends: Use if a report evaluates trends such as
increasing incidence with time or increasing mortality with age of
onset. This keyword should also be used if reference is made to
relationship of incidence to sunspot activity or other periodic
phenomena. Also use for birth cohort effect.
230 - Cross-sectional: In a cross-sectional study, the
measurements of cause and effect have been made at the same point in
time. Thus, it is a nondirectional (i.e., neither retrospective nor
perspective) study. Cross-sectional studies are usually based on disease
prevalence information rather than incidence. In contrast to a cohort
study, a cross-sectional study that assesses prevalence data cannot
ascertain the direction of the relationship between the study factor and
the disease (i.e., cannot determine whether the hypothesized cause was an
antecedant or a consequent of the disease). It differs from a
case-control study in that the study population is selected from a single
target population.
240 - Retrospective: An epidemiologic study that selects its
population to be watched on the basis of a known characteristic and then
tracks backwards in time to identify any other characteristics held in
common.
* * * DRAFT FINAL * * *
-------
A-ll
EXHIBIT A4 (cont)
250 - Prospective: Any epidemiological study that takes a
population and watches it for the subsequent appearance of a
characteristic, e.g., a disease.
260 - Migrant: Epidemiologic studies that evaluate the response of
individuals that migrate from a geographic location of relatively low
exposure to a location of relatively high exposure.
265 - Occupational: Use for studies that evaluate the impact of
occupation on incidence, or survival of a disease. This word should be
used if a study mentions difference between indoor and outdoor work, or
occupational exposure to carcinogens or differences between office
workers and factory workers.
270 - Socio-economic status: Use if report talks about
relationship of disease to socioeconomic status, e.g., years of
education, blue-collar/white-coliar.
280 - Age/sex differences: Use if report deals with differences in
reponse of different age groups or between males and females.
290 - Incidence: Use if a study gives incidence rates or ratios or
evaluates differences in incidence between different parameters e.g.,
anatomical site.
291 - Mortality: Use if a report evaluates a mortality parameter,
e.g., SMA (standard mortality ratio), or compares and contrasts mortality
in different populations based on various characteristics.
292 - Survival: Use if a study evaluates the survival of a
population after various treatments or based on different
characteristics, e.g., smoking habits.
300 - ANIMAL (study): Experimental studies in which whole animals
are used to determine the impact of an agent at the animal, tissue or
cellular level.
310 - Mouse: Use for experimental studies employing mice (or mouse
cells).
320 - Guinea pig: Use for experimental studies employing guinea
pigs (or their cells).
330 -.Hamster: Use for experimental studies employing hamsters (or
their cells).
* * * DRAFT FINAL * *
-------
A-12
EXHIBIT A4 (cont)
340 - Tumor transplant: Use if study mentions transplantation of
tumors.
350 - Chemical carcinogen: Use if study mentions the use of a
chemical carcinogen to induce a tumor.
400 - CELLULAR (in vitro): Experimental studies dealing with the
impact of an agent on cells. May be in vivo or in vitro. There may be
cross-over with animal studies.
410 - Transformation: Any study that evaluates the preneoplastic
or neoplastic development of a cell. May be applied to in vivo or in
vitro studies.
411 - Differentiation/development: Use for any study that
evaluates cellular differentiation or growth states, or for a study that
looks at developmental changes or stages.
420 - Melanocyte/melanoblast: Studies of cells (or their
precursors) that synthesize melanin.
430 - Keratinocyte: Any information that mentions keratinocytes or
epidermal cells that synthesize keratin. These cells eventually become
the corneocytes that form the stratum corneum.
440 - Squamous cell: Use for any information on squamous cell
carcinoma or for any studies that deal with the precursor cell for this
lesion.
450 - Epidermis/epidermal cell: Use for studies of the epidermis
or for cells not otherwise specified (nos) which populate the epidermis.
This should track with keratinocyte, melanocyte, basal cell and
Langerhans cells.
460 - Nevus/nevus cell/mole: Use for any study mentioning nevi,
nevus or moles.
470 - Basal cell: Use for basal cell carcinoma and any studies
dealing with basal cells in the dermis or epidermis.
480 - Stratum corneum: Use for studies which mention or examine
the stratum corneum layer of the skin.
500 - MOLECULAR: Use for any study designed to investigate the
problem at the molecular level, i.e. could be at the biochemical or
histochemical level. Could involve urocanic acid or DNA damage.
* * * DRAFT FINAL
-------
A-13
EXHIBIT A4 (cont)
510 - Urocanic acid/histidine: Use for any studies dealing with
urocanic acid or the metabolism of histidine. •
520 - DNA: Any studies dealing with damage to the genetic
material, measured biochemically. This should also include oncogene
research.
530 - Melanin/tyrosine: Use for studies of or tyrosine melanin
biochemistry, characterization, synthesis, etc.
540 - Pigment: Use for studies evaluating pigment differences not
otherwise specified, i.e., could be used for epi studies where darkness
of pigmentation was an evaluated parameter.
550 - Mechanism/hypothesis: Use for studies which put forth a
mechanism or a hypothesis.
600 - IMMUNOLOGIC/lymphocyte (nos): Use for anything having to do
with the immune system and its responses. Use also for any reference to
lymphocytes whi^h does not identify a specific type.
610. - la antigfens/Class II MHC: Use for studies evaluating the
role of la-or Class II MHC antigens in the immune response or as a marker
for Langerhans cells.
620 - Langerhans cells: Use for studies which mention these cells
of the macrophage lineage that normally reside in the skin..
630 - T-lymphocyte: Use for studies that mention "thymus derived"
lymphocytes. These eels may be "helper" cells, "inducer" cells,
"suppressor" cells or "killer" cells.
640 - B-lymphocyte: Use for studies which deal with antibody
forming cells and their precursor surface-antibody positive lymphocytes.
650 - Autoimmunity: Use for studies evaluating aspects of immune
reactions against self. Use for any studies of systemic lupus
erythematosus.
660 - Antibody: Use for any studies relating to the production of
antibodies during disease states. Do not use if antibodies are being
used as a diagnostic tool.
670 - Immunosuppression: Use for reports indicating lack or
depression of normal immune function.
* * * DRAFT FINAL * * *
-------
A-14
EXHIBIT A4 (cont)
700 - CATARACT/eye (nos): Use for studies relating to the etiology
of cataract development and for any general studies of the eye.
710 - Cornea: Use for studies evaluating the impact of an agent on
the cornea.
720 - Lens: Use for studies relating to the impact on the lens.
All cataract studies are also lens studies but not vice versa.
730 - Retina: Use for studies relating to retinal behavior or
damage.
800 - SUNLIGHT: Use for studies where sunlight is evaluated as a
risk factor but where no wavelength is specified.
810 - Ultraviolet light: This should be used when UV is identified
but the wavelength is not specified..
811 - UV-A: Use for studies that have evaluated the impact of UV
in the 320-400 nm range.
812 - UV-B: Use for studies that have evaluated the impact of UV
in the 280-320 nm range.
820 - Latitude gradient: Use for studies which evaluate impact
correlated to latitude changes.
830 - Fluorescent light: Use for studies which evaluate the role
of fluorescent or indoor light on disease development (could be melanoma,
cataract, etc.).
840 - Erythema/sunburn: Use for studies that evaluate role or
amount of sunburn that an individual received, or studies that deal with
the erythemal response to ultraviolet irradiation.
850 - Ionizing radiation: Use for studies that evaluate the impact
of exposure to. x-irradiation.
860 - Exposure characteristics: Use for studies that evaluate the
characteristics of exposure to radiation.
861 - Intensity/duration: Use for studies that evaluate the
intensity or duration of exposure, e.g., differences between acute and
chronic exposures, or sunburn versus gradual tanning.
900 - VIRUSES: Use for studies which deal with viruses (nos).
* * * DRAFT FINAL * * *
-------
A-15
EXHIBIT A4 (cont)
910 - Herpes: Use for studies of herpes viruses and the dieases
they induce.
* * * DRAFT FINAL * * *
-------
APPENDIX B
REVIEW OF CRITICAL EPIDEMIC-LOGIC STUDIES
* * * DRAFT FINAL * * *
-------
GREEN AND COLLEAGUES -- QUEENSLAND MELANOMA DATA
This is a case-control study from Queensland in northeast Australia, an
area which has the highest incidence of melanoma in the world. Queensland is
unusual in that a predominantly Caucasian population inhabits both tropical
and subtropical latitudes.
Cases were those persons first diagnosed with a primary cutaneous melanoma
during the year from July 1, 1979 to June 30, 1980. Cases were ascertained
from the records of the 24 Queensland pathology laboratories. There were 871
cases diagnosed of which 201 were reported to have an in situ component of
Hutchinson's melanotic freckle. No histological classification was given for
10 cases and age was not stated for an additional 27 leaving 633 cases' for the
analysis of non-Hutchinson's melanoma. Address at the time of diagnosis was
used for geographic correlations.
In a series of seven publications to date, Green and her colleagues have
examined several subsets of the Queensland patient series. Three of the
articles analyze randomly selected patient subsets. The four other articles
present case-control comparisons for randomly selected patient subsets and
age-, sex-, and residence-matched controls randomly selected from Queensland
electoral rolls.
Three studies deal with the incidence and reporting of CMM in .Queensland,
the relationship between CMM incidence and latitude, and the diagnosis of
Hutchinson's melanotic freckle (HMF). The case-control studies investigate
the effects of cumulative sun exposure, episodes of sunburn, the number of
naevi on the left arm, and the presence of nonmelanotic skin tumors on risk of
CMM.
For the case-control studies, information obtained on each subject
included lifetime sun exposure (occupational or recreational), acute and
chronic response to sun exposure, episodes of severe sunburn, complexion
(skin, eye and hair color), number of naevi (2 mm or more in diameter) on the
left arm, family melanoma history, social class (based on occupation),
ethnicity, and nonmelanotic facial skin cancers.
* * DRAFT FINAL * * *
-------
B-2
Reference:
Green, A. "incidence and reporting of cutaneous melanoma in Queensland."
Aust. J. Derm. 23:105-109 (1982).
Investigator Results:
1. The incidence rate was still rising, especially of thin tumors. The
1979-1980 crude annual incidence rate was 39.6/105 compared to
32.7/105 in 1977.
2. Of the 871 cases, 455 (52 percent) were in women and 416 (48 percent)
were in men. More women than men noticed their own tumors. Among the
236 patients, in females, 68 percent of the lesions were first
detected by the patient, 12 percent by other non-medical individuals,
and 20 percent by the doctor; in males the numbers were 46 percent, 28
percent, and 26 percent, respectively.
3. Cell types included malignant and nonmalignant tumors; 57 percent were
superficial spreading melanoma (SSM). Lentigo maligna (melanoma)
(LMM) accounted for 23 percent of all lesions compared to 15 percent
in 1977 and 7 percent in 1967. Nodular melanoma comprised 15 percent
of the lesions compared to 15 percent in 1977 and 28 percent in 1967.
A distribution of anatomic location by cell type was given showing a
majority of tumors on the head to be LMM. Most of the SSM and LMM
tumors were thin while the nodular tumors tended towards greater
thickness.
4. The author commented that "... it is difficult to distinguish between
a true rise in disease which may have occurred, e.g. as the result of
sun exposure received at increasingly popular beach resorts, and a
rise resulting from early reporting in a community which is aware of
the disease."
Methodology:
All 871 patients with a first cutaneous melanoma diagnosed in Queensland
from 1 July 1979 to 30 June 1980 reported by one of the 24 pathology
laboratories were studied. The author interviewed a random sample of 236
patients with a standard questionnaire concerning how they first became
aware of the lesion.
Experiemental Design and Analysis Issues:
A descriptive analysis of MM information from Queensland, Australia.
* * * DRAFT FINAL * * *
-------
B-3
Reference:
Green, A. and Siskind, V. "Geographical Distribution of Cutaneous
Melanoma in Queensland." Med. J. Australia 1:407-410 (1983).
Investigator Results:
1. A latitude gradient for melanoma incidence in Queensland from
1979-1980 was found. The reported incidence rate for Queensland had,
however, doubled from 16.4/105 in 1965 to 32.7/105 in 1977.
2. A significantly increased incidence of melanoma was observed in
coastal, compared to inland regions.
3. Authors concluded that one possible explanation of the observed
geographical distribution lies in the ready accessibility to beaches
which encourages sunbathing in the coastal population.
Methodology:
A cross-sectional analysis of the relationship between melanoma incidence
rates and latitude in Queensland. Incidence rates were calculated based
on 633 cases of first primary cutaneous melanoma diagnosed and reported at
the 24 Queensland pathology laboratories between 1 July 1979 - 30 June
1980., directly age-standardized to 1979 Australian Bureau of Statistics
population estimates. Patient information included age, sex, residential
address at time of diagnosis, and histologically classified lesion
according to Clark method. Patients with Hutchinson's melanotic freckle
and melanotic freckle melanoma were treat separately.
Queensland was partitioned into four regions (tropical, subtropical,
inland, coastal) as well as into 11 statistical divisions defined by the
Australian Bureau of Statistics.
Experiemental Design and Analysis Issues:
A cross sectional analysis of the relationship between melanoma incidence
and latitude in Queensland.
For Result 1:
Age-standardized incidence rates ranged from 9.4/105 - 41.6/105 in
Queensland by statistical division. Differences between divisions and
by latitude were not significant.
* * * DRAFT FINAL * * *
-------
B-4
For Result 2:
Age-standardized incidence rates differed significantly by region (p
less than 0.05) with the source of the difference due to inland vs.
coastal incidence rates (p less than 0.01) rather than tropical vs.
subtropical rates (p greater than 0.5). There was no evidence that
the major histological types differed by geographic area. The
erythemal UV dose decreases with latitude in July (mid-winter) but
this pattern is modified in January (mid-summer) when there is also an
increasing gradient as one moves inland.
Comment:
It is uncertain whether the case ascertainment was complete since'it is
possible that some cases were treated in an individual physician's office
from which no material was submitted to one of the pathology laboratories
surveyed or from which the material might have been submitted to another
laboratory. The address at the time of diagnosis was used in this study;
the classic problem of the address when the cancer was "caused" then
becomes relevant since we are not assured that the patient resided at the
diagnosis address for any particular period of time.
* * * DRAFT FINAL * * *
-------
B-5
Reference:
Green, A., Little, J.H. and Weedon D., "The diagnosis of Hutchinson's
melanotic freckle (lentigo maligna) in Queensland." Pathology 15:33-35
(1983).
Investigator Results:
1. Among a sample of 99 melanoma lesions, 76 had an in-situ component of
Hutchinson's melanotic freckle (HMF) type, 12 had in-situ component of
superficial spreading melanoma (SSM) type, 4 indeterminant/lack of
agreement in-situ component, and 7 were lentigines with insufficient
atypical features to warrant a diagnosis of HMF.
2. The authors comment ... "Our review highlights the potential for
misclassification as well as overdiagnosis of HMF". In Queensland the
number of reported cases of primary cutaneous melanoma "rose from 106
in 1977 (15 percent of the year's total of 705) to 224 for the 12
month period from July 1979 (26 percent of 878)".
Methodology:
A descriptive analysis of a sample of 99 lesions from 97 patients (44
percent) out of 224 cases diagnosed in Queensland, Australia from 1 July
1979 to 30 June 1980 which were reviewed by two pathologists.
Comment:
With a strict reproduceability rate of 76/99 = 77 percent,
misclassification and overdiagnosis could explain some of the increase in
the disease. On the other hand, it is difficult to explain a doubling of
the disease by this mechanism.
* * * DRAFT FINAL * * *
-------
B-6
Reference:
Green, A. "Sun exposure and the risk of melanoma." Br. J. Derm.
25:99-102 (1984).
Investigator Results:
1. In a case-control study the risk of melanoma increased with
accumulated solar exposure through life including lifetime
occupational and recreational exposures adjusted for residential
mobility.
2. Among 183 matched case-control pairs, risk of melanoma increased with
number of cumulative hours of sun exposure through life even after
adjusting for the effects of exact age, presence of naevi on the arms,
hair color, and sunburn propensity in a multivariate model. For less
than 2,000 hours, the relative risk (RR) was 1.0. For 2000 to 50,000
hours, the RR was 3.2 (95% C.I. 0.9-12.4) and for more than 50,000
hours, the RR was 5.3 (95% C.I. 0.9-30.8). The unadjusted RRs were
1.0, 2.0, and 3.3, respectively.
3. Cases had significantly more actinic lesions (a three-fold increase)
on their faces than controls.
4. The author commented that "These data strongly suggest that melanoma
does have an association with high doses of solar UV radiation."
Methodology:
A case-control study of MM patients who reported their first primary MM
between 1 July 1979 - 30 June 1980 in Queensland and for whom histological
diagnosis and tumor thickness were provided by Statewide pathology
libraries. Of 871 total cases diagnosed in the year, 243 were randomly
selected and 236 (97 percent) contacted and interviewed. Controls,
randomly selected from electoral rolls, were matched by age, sex, and
place of residence. Information on cases and controls, obtained by
interviewer questionnaire, included all episodes of severe sunburn (48+
hours duration), number of sunburn experiences by age group (0-9, 10-19,
20-29, 30+ with virtually all burns occurring before age 40), lifetime sun
exposure (occupational and recreational), eye and hair color, acute and
chronic response to sun exposure, nonmelanotic facial skin cancers, and
number of naevi (dark brown lesions 2 mm or more in diameter) on left
arm. After lentigo maligna and acral lentiginous melanomas were excluded,
183 case-control study pairs from 14 to 81 years of age remained.
Crude unmatched RRs were calculated, as well as matched from RRs
unadjusted and using conditional logistic regression (CLR).
* * * DRAFT FINAL * *
-------
B-7
Comment:
The relative risks in the multivariate model have the possibility that
hours of sun serves as a pseudonym for age and/or presence of nevi
especially after these have been included in the multivariate model. It
would be important to know the correlation among these predictor variables
* * * DRAFT FINAL * * *
-------
B-8
Reference:
Green, A., MacLennan, R., and Siskind, V. "Common Acquired Naevi and the
Risk of Malignant Melanoma." Int. J. Cancer 35:297-300 (1985).
Investigator Results:
1. Among 183 malignant melanoma (MM) patients and 183 sex-, age-, and
area of residence-matched controls, there was a strong association
between presence of pigmented naevi on arms and MM, with a crude
relative risk (RR) of 28.0. After adjusting for hair color,
propensity to sunburn, and lifetime sun exposure, the RR was 30.1.
2. Family history did not appear to be a determinant of MM independent of
above-mentioned risk factors.
3. The authors commented that "Similar findings regarding the risk of
melanoma in sun-sensitive persons ... provide excellent circumstantial
evidence that sun exposure has a causal association with disease."
Methodology:
A case-control study of MM patients who reported their first primary MM
between 1 July 1979 - 30 June 1980 in Queensland and for whom histological
diagnosis and tumor thickness were provided by Statewide pathology
libraries. Of 871 total cases diagnosed in the year, 243 were randomly
selected and 236 (97 percent) contacted and interviewed. Controls,
randomly selected from electoral rolls, were matched by age, sex, and
place of residence. Information on cases and controls, obtained by
interviewer questionnaire, included all episodes of severe sunburn (48+
hours duration), number of sunburn experiences by age group (0-9, 10-19,
20-29, 30+ with virtually all burns occurring before age 40), lifetime sun
exposure (occupational and recreational), eye and hair color, acute and
chronic response to sun exposure, nonmelanotic facial skin cancers, and
number of naevi (dark brown lesions 2 mm or more in diameter) on left
arm. The final 183 case-control pairs included cases with intra-epidermal
components of superficial spreading or indeterminate melanoma types, and
those with no intra-epidermal component (nodular). Lentigo maligna and
acral lentiginous melanomas were excluded.
Crude associations were estimated with simple matched and unmatched
univariate screening; other factors were assessed with Mantel-Haenszel RR
estimates and conditional logistic regression (CLR) for matched pairs.
Experimental Design and Analysis Issues:
A well-designed case-control study of the association of a variety of
factors and MM.
* * * DRAFT FINAL * * *
-------
B-9
For Result 1:
The strongest association was between MM and naevi on arms (2-4 naevi
vs. none, RR=15.7; 5-10 naevi, RR=14.9; 10+ naevi, RR=20.1). The
major effect was presence of any naevi with RR=15.8 (95% confidence
interval 9.4-26.5) with unmatched data, RR=28.0 (10.6-106.8) with
matched data. Several other factors were crude predictors of MM risk,
including propensity to burn then tan, or burn then peel (crude
RR=3.6, adjusted for naevi RR=2.5), moderate or no tan tendencies
(crude RR=4.5, adjusted RR=3.0), propensity to freckle, light
brown/blonde hair color, and red hair color (crude RR=3.6, adjusted
RR=2.4).
In a stepwise CLR with all phenotypic characteristics, the major MM
determinants were naevi on arms, propensity to sunburn upon acute
exposure, and hair color. After adjusting for lifetime sun exposure,
propensity to burn and hair color, the RR for presence of naevi was
30.1, for red hair 5.9, for light brown/blond hair 3.5, for black/dark
brown hair 1.0, and for the propensity to sunburn 3.5.
For Result 2:
There was no evidence of an association between MM and positive family
history of skin and noncutaneous cancer. Significantly more cases
than controls had a family, history of skin cancer (18% vs. 8%) with .
crude RR=2.5. When naevi, sunburn and hair color were considered, the
RR for family MM history was 0.96.
* * * DRAFT FINAL * * *
-------
B-10
Reference:
Green, A., Siskind, V., Bain, C., Alexander, J. "Sunburn and Malignant
Melanoma." Br. J. Cancer 51:393-397 (1985).
Investigator Results:
1. Among 183 case-control pairs, 99 percent of sunburn experiences
occurred prior to age 40 (median age at diagnosis was 46). Of the 183
cases, 141 (77 percent) had superficial spreading melanoma, 36 (20
percent) had nodular melanoma, and 6 (3 percent) had interdeterminate
melanoma.
2. There was an association between multiple sunburns and melanoma
(excluding lentigo maligna) among the case-control pairs. After
controlling for other risk factors, a significant dose response
association (p less than 0.05) was evident (relative risk (RR)=1.5 for
2-5 sunburns in life, RR=2.4 for 6+ sunburns).
3. The authors commented that "... sunburn exposure factor is a
consequence of the amount of UV received at the skin surface and the
degree of pigment protection provided by melanin against UV
transmission through the epidermis. Thus regardless of an
individual's innate colouring or tanning from previous sun exposure,
an experience of painful erythema indicates that acute high-dose UV
has been delivered to the level of the melanocyte."
Methodology:
A case-control study of MM patients who reported their first primary MM
between 1 July 1979 - 30 June 1980 in Queensland and for whom histological
diagnosis and tumor thickness were provided by Statewide pathology
libraries. Of 871 total cases diagnosed in the year, 243 were randomly
selected and 236 (97 percent) contacted and interviewed. Controls,
randomly selected from electoral rolls, were matched by age, sex, and
place of residence. Information on cases and controls, obtained by
interviewer questionnaire, included all episodes of severe sunburn (48+
hours duration), number of sunburn experiences by age group (0-9, 10-19,
20-29, 30+ with virtually all burns occurring before age 40), lifetime sun
exposure (occupational and recreational), eye and hair color, acute and
chronic response to sun exposure, nonmelanotic facial skin cancers, and
number of naevi (dark brown lesions 2 mm or more in diameter) on left
arm. After lentigo maligna and acral lentiginous melanomas were excluded,
183 case-control study pairs from 14 to 81 years of age remained.
Crude unmatched RRs were calculated, as well as matched from RRs
unadjusted and using conditional logistic regression (CLR).
* * * DRAFT FINAL * * *
-------
B-ll
Experimental Design and Analysis Issues:
A case-control study among 183 age-, sex-, residence-matched pairs.
Cases generally had more sunburns than controls (p less than 0.01), and
when number of sunburns were considered (0-1 controls, 0-1 cases, 2-5
cases, 6+ cases), there were significantly more cases than controls (p
less than 0.001) with 2-5 burns, crude RR=2.4 and with 6+ burns, crude
RR=3.3.
When number of naevi on arms and age were included in a multivariate
model, the adjusted RR was 1.5 (95% C.I. 0.7-3.2) for 2-5 burns and 2.4
(95% 'C.I. 1.0-6.1) for 6+ burns (p less than 0.05). When presence of skin
cancers, migrant status, and social class were included, the RRs remained
essentially unchanged. Although small samples limited analysis by
histogenic type, a tendency of increasing RRs with number of sunburns,
especially for superficial spreading was observed.
* * * DRAFT FINAL * * *
-------
B-12
Reference:
Green, A. and O'Rourke, M.G.E. "Cutaneous Malignant Melanoma in
Association with Other Skin Cancers." JNCI 74(5):977-980 (1985).
Investigator Results:
1. In a case-control study of 232 malignant melanoma (MM) cases and 232
matched controls, a fourfold increase in risk was associated with
presence of facial actinic tumors. The risk persisted for melanoma
groups: lentigo maligna melanoma, superficial spreading melanoma, and
nodular melanoma.
2. Cases with heavy lifetime sun exposure had higher relative risks than
comparable controls, even after adjusting for age, nevi, hair color,
and propensity to sunburn.
Methodology:
A case-control study of 232 malignant melanoma cases (14-86 years old) who
had primary melanoma reported between 1 July 1979 - 30 June 1980 in
Queensland (randomly selected and stratified by geographic location from
871 eligible cases). Controls randomly selected from the Electoral Roll
were matched by age, sex, and area of residence. Interviewer
questionnaire (by one of the authors) obtained information on history of
lifetime sun exposure (all outdoor occupations of 6+ months and all
regular outdoor recreations since 10 years old), complexion (e.g., hair
color), sun sensitivity (e.g., propensity to burn), social class based on
occupation, country of birth of patient and 2 previous generations,
actinic tumors on face and left forearm, and nevi (2 or more mm in
diameter). Tumors were classified as superficial spreading (60.8%, SSM),
nodular (15.5%, NM), lentigo maligna (21.1%, LMM), and indeterminate
(2.6%, IND).
Experimental Design and Analysis Issues:
A case-control study of 232 melanoma patients and 232 age-, sex-, and
residence-matched controls in Queensland.
For Result 1:
A significantly greater percentage of facial actinic tumors occurred
among cases (41%) than controls (15%, p less than 0.0001), with a
crude relative risk (RR) of 4.4. The relative risk when adjusted by
age and nevi was 3.6. When other possible risk factors (e.g., social
class, ethnic origin) were taken into account, the risks were
relatively unchanged. This case-control difference was also observed
among LLM, and combined SSM, NM and IND group (p less than 0.001).
Results for forearm tumors were similar. The data showed no strong
trend of increasing risk with increasing number of facial tumors.
Presence of nevi on arms was associated with a 30-fold increase in
melanoma risk.
* * * DRAFT FINAL * * *
-------
B-13
For Result 2:
More cases than controls were in high sun exposure categories
(accumulated hours of sun during work and recreation). The crude
relative risk (RR) was 2.3 for 50,000+ hours of lifetime exposure
compared to less than 2,000 hours exposure. Risks were higher after
adjustment for age, nevi, hair color, and sunburn propensity. The
adjusted RR was 3.2 for 2,000-50,000 hours exposure and 5.3 for
50,000+ hours exposure.
* * * DRAFT FINAL * * *
-------
B-14
HOLMAN AND COLLEAGUES -- WESTERN AUSTRALIA MELANOMA STUDY
Holman and his colleagues have presented several studies on cutaneous
malignant melanoma incidence and mortality in Australia. In this section,
nine of these studies are reviewed.
Of particular importance among these studies are five case-control
comparisons based on 1980-1981 data from Western Australia. The cases
consisted of 511 melanoma patients under 80 years of age (233M, 278F) with
histologically diagnosed pre-invasive and invasive melanoma, possible
melanoma, and Spitz nevus occurring in Western Australia for the first time
from 1 January 1980 - 5 November 1981. The cases were identified from
histopathology reports issued by public and private medical laboratories. Of
the 820 preinvasive and invasive melanomas ascertained, 766 were
histologically reviewed by six pathologists for confirmation or rejection of
diagnosis, histogenic type (McGovern classification), level of invasion, and
tumor thickness. After review and including acceptance of the original
diagnosis in 39 primary cases unavailable for review, acceptance of 4 of 9
possible melanoma cases and 1 of 14 Spitz nevi cases, the study series
consisted of 815 melanomas arising in 798 patients. Of the 798 patients, 670
were eligible for interview (interviews were not attempted in some rural and
remote areas), and permission to interview was granted for 582 cases (87%).
The number of individuals approached for interview was 565 (15 were
untraceable or had migrated and 2 were in distant rural areas) from which 511
(90%) actually responded. The cases ranged ,from 10-79 years of age.
Sections from the tumors (except for 14 patients) were classified by a
panel of six pathologists into the following histological types: Hutchinson's
melanotic freckle (HMF), superficial spreading melanoma (SSM), unclassifiable
melanoma (UCM), or nodular melanoma (NM). Clinical details were obtained from
the general practitioner and surgeons who attended the patients.
Of 824 potential controls randomly identified for the study, 511 were
selected from the Australian Commonwealth Electoral Roll, and for 10 cases
under 18 years, the student rolls of a public school in the case's area of
residence. Controls were matched by age (+5 years), sex, and area of
residence. The number of controls actually approached was 740 from which 511
(69%) were interviewed. A selection bias may have been introduced because
only 69% of controls approached were interviewed in contrast to 90% of
approached cases. The methods used to contact potential participants and to
obtain their cooperation were identical for cases and controls.
Subjects were interviewed in their homes (occasionally at their workplace)
by trained nurse interviewers. To the extent possible, interviewers were
blinded with respect to who were cases and who controls. A highly structured
questionnaire was administered. Information collected on each subject by
interview consisted of skin, hair, and eye color, number of palpable nevi on
the arms, history of sun exposure, harmone use, diet, measurements of weight,
height, hairiness, and extent of actinic damage in skin (based on cutaneous
microtopographs), acute and chronic skin reactions to sunlight, ethnicity of
* * * DRAFT FINAL * * *
-------
B-15
grandparents, family history of melanoma or xeroderma pigmentosum, history of
mole excisions, and treatment of nonskin cancers. Skin color was measured at
the dorsum of the left hand (continuous sun exposure), tip of left shoulder
(intermittently exposed), and left upper inner arm (not usually exposed) and
was graded according to ranges of reflectance values (%). A voluntary venous
blood sample was collected for retinol and cholesterol assays.
In 1983, following a report that melanoma was associated with exposure to
fluorescent lighting at work, 337 of the cases and 349 controls were
reinterviewed regarding fluorescent light exposure.
The matched case-control data were analyzed according to methods described
by Breslow and Day. Empirical odds ratios (ORs) were calculated by
conditional maximum likelihood estimation. The significance of trends in the
ORs were assessed by the chi-square formula for matched data. Conditional
logistic regression was used to analyze two or more risk factors. Four
controls reporting history of mole excision and their corresponding cases were
excluded from all analyses except that f6r the history of melanoma.
* * * DRAFT FINAL * * *
-------
B-16
Reference:
Holman, C.D.J., Mulroney, C.D. and Armstrong, B.K. "Epidemiology of
Pre-Invasive and Invasive Malignant Melanoma in Western Australia." Br.
J. Cancer 25:317-323 (1980).
Investigator Results:
1. Annual incidence-rates of melanoma in western Australian in 1975-1976
were 4.4/105 (males) and 6.2/105 (females) for pre-invasive
lesions and 18.6/10s and 18.8/105 for invasive lesions.
2. Incidence patterns with age indicated a progressive increase beginning
about age 40 for Hutchinson's melanotic freckle and head and neck
invasive melanoma, and an early rise at about age 20 with mid-life
peak and subsequent stabilization or decline for superficial
spreading, non-invasive and invasive lower limb and less frequently
trunk and upper limb melanomas.
3. Incidence rates were highest in native-born Australians, followed by
British immigrants. Rates among British immigrants were over 2 times
higher than rates in the U.K.
4. Incidence was highest in high social class residential areas, and
higher in indoor (vs. outdoor) workers.
5. Incidence was highest in the capital city (Perth) and southwest corner
of State (vs. north part).
6. Patterns for invasive and pre-invasive lesions were similar.
7. Aspects of the data were inconsistent with the solar hypothesis, but
some inconsistencies might be explained if intermittent, intense sun
exposure were more relevant to melanoma than continuous exposure.
Methodology:
A descriptive analysis of pre-invasive (PIM) and invasive malignant (IMM)
melanoma in western Australia determined retrospectively for 1975-1976
from discharge records for all western Australia hospitals in 1975-1976,
and histopathology reports in 1975-1976 from all pathology labs in the
State. Each case (120 PIM, 422 IMM) had to be a usual resident of western
Australia, have histological verification of diagnosis, have first biopsy
between January 1975-December 1976, and have skin as primary lesion site.
Information sought on each case included sex, birth date, country of
birth, occupation, usual residence, diagnosis date, anatomical site and
description of histopathology (categorized only as PIM or IMM, and for
PIM, Hutchinson's melanotic freckle (HMF) and "other including superficial
and spreading, non-invasive (SSM), or in situ melanomas). Age- and
sex-specific incidence rates were calculated based on Australian Bureau of
Statistics 1976 population estimates.
* * * DRAFT FINAL * * *
-------
B-17
Experimental Design and Analysis Issues:
A descriptive analysis of 542 PIM and IMM patients in western Australia
from 1975-1976.
For Result 1:
No additional information.
For Result 2:
For both PIM and IMM, incidence rates increased progressively with age
in males but less regularly in females, with initial peak at 40-49
and, after further rise, apparent decline after 80+ years. The
differences between sexes in age-incidence curves were explained by
pathologic lesion characteristics. Among PIM, equal number of HMF in
men and women 1.3/105) but over 50% more SSM in women than men
(incidences 3.1/105 and 4.9/105, respectively). HMF incidence
increased progressively with age except for early peak at 60-69,
whereas SSM reached peak at 50-59 years and declined thereafter (both
sexes similar).
For IMM and SSM, a male trunk and female lower limb predominance was
observed. For HMF, 79% occurred on head and neck in both sexes. Head
and neck lesions showed progressive incidence rise from about age 40,
whereas lower limb lesions (and to lesser extent trunk and upper limb
lesions) showed increased from age 20-29 (females) and 30-39 (males),
peak at 50-59, and decline thereafter. These different patterns
produce irregular IMM pattern as a whole.
No incidence changes were seen in women (or men) 40-49 and 50-59. The
pronounced irregularity of PIM at all sites in women between 40-60 was
due to high proportion of limb lesions. The female age incidence
pattern does not support hypothesis that hormones may be involved in
melanoma etiology.
For Result 3:
Incidence rates among PIM and IMM were over 2 times greater among
native-born Australians (5.6/105 males, 6.7/105 females for PIM;
26.1/105 males, 23.7/105 females for IMM) than among immigrants.
Rates of IMM in British born (10/105 males, 13/105 females) were
about 2 times greater than all other immigrants combined, and 2-5
times higher than British rates from 1968-1972 (2-5/105
age-standardized). Differences in rates were unchanged after
adjustment by social class and age.
* * * DRAFT FINAL * •* *
-------
B-18
For Result .4:
An analysis by social class (based on a socio-economic index
referencing occupation and education divided into 4 "classes")
indicated highest IMM incidence in highest social class with
progressive incidence decline in women with decreasing social class.
For PIM the pattern was less clear. After controlling by country of
birth, proximity to sea, and age, no appreciable changes were observed.
IMM incidence rates were highest among professional workers
(39/105), clerical and sales workers, and administrators and
managers, and lowest among farmers and fishermen (18.5/105),
laborers and tradesmen, and transport and communication workers.
For Result 5:
Highes PIM and IMM incidence rates were in Perth or Southwest region
(e.g., 5.6/105 PIM and 2.55/105 IMM in males) whereas rates in
Central, Pilbara, and Kimberley regions were -lower (e.g., 1.9/105
PIM and 18.0/105 IMM in males) in spite of their northerly latitude
and higher sun exposure.
For Result 6:
Similarities between PIM and IMM included more common occurrence among
females, similar age-incidence curves, similar curve irregularities
due to two patterns of incidence change with age (progressive increase
beginning in mid-life for HMF and head and neck IMM and rapid increase
in early adult life with peak at about 50 and decline thereafter for
SSM and lower limb IMM), more common occurrence in native-born
Australians, in higher social classes, and in Perth and Southwest of
State.
* * * DRAFT FINAL * * *
-------
B-19
Reference:
Holman, C.D.J., James, I.R., Gattey, P.H., and Armstrong, B.K. "An
Analysis of Trends in Mortality from Malignant Melanoma of the Skin in
Australia." Int. J. Cancer 26:703-709 (1980).
Investigator Results:
1. Age-standardized Australian malignant melanoma (MM) mortality rates
more than quadrupled from 0.8/105 (males) and 0.6/105 (females) in
1931-1934 to 4.2/105 (males) and 2.5/105 (females) in 1975-1977.
Approximately parallel increases were found for each of the six
Australian states.
2. MM mortality rates decreased from north to south.
3. Increasing mortality rates could be explained by increases in
successive birth cohorts, beginning in 1865, and stabilizing around
1925 (females) and 1935 (males). The authors suggested that cohort
increases resulted from lifestyle changes in successive generations.
They predicted that the trend towards increasing total MM mortality
would stabilize over the next 40 years.
Methodology:
A descriptive analysis of MM mortality data for 1931-1977 (in 5-year age
groups) obtained from the Australian Bureau of Statistics (subdivided by
the 6 Australian States after 1950). Annual mortality rates for each
5-year age group and for 5-year time periods were based on Australian
Bureau of Statistic's population estimates. The data were analyzed to
separate birth cohort, age, and calendar year of death effects in an
additive three-factor model which estimated expected mortality rates.
Experimental Design and Analysis Issues:
A descriptive analysis of MM mortality data and the fit to a birth cohort,
age, and calendar year of death dependent additive model.
For Result 1:
Crude and age-standardized MM mortality rates from 1931-1934 to
1975-1977 more than quadrupled in both sexes, although the increase
was greater in males (429%) than females (302%).
For Result 2:
Highest mortality rates were in Queensland (latitude 11°-29°S),
intermediate rates in Western Australia (14°-35°S), New South Wales
(28°-37S), and South Australia (26°-38°S), and lowest rates in
Victoria (34°-39°S) and Tasmania (40°-43°S).
For Result 3:
The pattern of mortality rate changes from 1931-1934 to 1975-1977 was
* * * DRAFT FINAL * * *
-------
B-20
more consistent with cohort-based rather than cross-sectional based
mortality increases due to a slowing mortality rate increase among
1920-1935 cohorts. The cohort pattern was also evident for each
Australian State (except Tasmania and Queensland).
Age, cohort, and time factors from the mortality data fit the
three-factor model well. The age factor rose quickly among 10-14 to
30-34 year-olds, and less steeply thereafter. The cohort factor
showed increasing trends from 1865-1935 (males) and 1865-1925
(females). Non-linear fluctuations in the time factor were small in
comparison to those in age and cohort factors.
When only age and cohort factors were considered in a two-factor
modeL, the fit was poorer than for the three-factor model. However,
the age and cohort values were similar for both models.
* * * DRAFT FINAL * * *
-------
B-21
Reference:
Holman, C.D.J and Armstrong, B.K. RE: "Skin Melanoma and Seasonal
Patters." Am. J. Epi. 113:202 (1981).
Investigator Results:
A seasonal pattern of melanoma incidence with summertime peak was observed
for 541 cases in western Australia from 1975-1976. Largest number of
diagnosed cases tfas in early summer, November (females), and December
(males), and cyclic trends were significant in males (p=0.04) and females
(p=0.02).
Methodology:
A short letter to the editor summarizing seasonal trends in incident
melanoma cases among 541 patients in western Australia from 1975-1976.
Cyclic trends were tested by Edward's test.
* * * DRAFT FINAL * * *
-------
B-22
Reference:
Holman, C.D.J. and Armstrong, B.K. "Hutchinson's melanotic freckle
Melanoma Associated with Non-permanent Hair Dyes." Aust. J. Cancer,
48:599-601 (1983).
Investigator Results:
1. There was no evidence of a relationship between melanoma of the head
and neck nor any histologic type and ever use of permanent hair dyes.
2. Hutchinson's melanotic freckle (HMF) was associated with use of
nonpermanent dyes with an odds ratio (OR) of 3.4 (95% C.I. 1.1-10.2)
for subjects exposed on 10 or more occasions. There was a linear
dose-response relationship for HMF with increasing frequency of use of
nonpermanent dyes (p=0.02 for trend).
3. The results suggested that nonpermanent hair dyes increase the risk of
HMF. The authors commented that "It was postulated that HMF results
from an accumulation of damage induced by UV radiation in the genome
of melanocytes, whereas superficial spreading melanoma may develop
from initiated cells in pigmented naevi which undergo promotion by
intermittent sun exposure and other agents. The results of this study
if confirmed by further research would suggest that initiating
carcinogens other than ultraviolet radiation, such as one or.more of
the aromatic compounds present in nonpermanent hair dyes, may also
contribute to the causation of HMF." This study's results, however,
require further confirmation in other studies.
Methodology:
A case-control study of 511 melanoma patients and 511 age-, sex-, and
residence-matched controls in western Australia (see introduction to this
section). Information obtained from cases and controls regarding past
exposure to permanent and nonpermanent (i.e., temporary and semipermanent)
hair dyes was analyzed in this study. Four case-control pairs were
excluded from analysis because of a past history of melanoma in the
controls.
Experimental Design and Analysis Issues:
For Result 1:
Permanent hair dyes had been used by 22% of the cases and 21% of
controls. Except for an OR of 3.5 (95% C.I. 0.7-24.3) for nodular
melanoma (NM) based on 9 discordant case-control pairs, there was no
evidence of a relationship between any histologic subtype nor melanoma
of the head and neck with ever-use of permanent hair dyes.
* * * DRAFT FINAL * * *
-------
B-23
For Result 2:
Semipermanent or temporary hair dyes had been used by 34% of cases and
33% of controls. For HMF, in addition to an OR of 3.4 associated with
10 or more uses of nonpermanent dyes, there was a linear dose-response
relationship with ORs of 1.5 and 3.4 for 1-9 uses and 10 or more uses.,
respectively, of nonpermanent hair dyes (p=0.02 for trend). There was
no evidence of an association between head and neck melanoma other
than HMF and use of nonpermanent hair dyes. The ORs for 10 or more
uses of nonpermanent hair dyes for HMF of the head and neck and HMF
elsewhere on the body were 3.1 (95% C.I. 0.9-10.7) and 4.8 (95% C.I.
0.4-52.2), respectively.
* * * DRAFT FINAL * * *
-------
B-24
Reference:
Holman, C.D.J., Evans, P.R., Lumsden, G.J. and Armstrong, B.K. "The
determinants of actinic skin damage: Problems of confounding among
environmental and constitutional variables." Amer. J. Epi. 120:414-422
(1984).
Investigator Results:
1. Of the total 1,216 individuals studied, 46 percent were male, 53
percent were older than 50 years, 24 percent had fair or red hair, and
39 percent were blue eyed. Of those with actinic skin damage, 59
percent were male, 80 percent were older than 50, 64 percent had fair
or red hair, and 61 percent had blue eyes.
2. Individuals aged 50 or more had 3.44 times as much actinic skin damage
as those under 50 (95% C.I. 3.05-3.89). Males had 1.21 times the
damage of females (95% C.I. 1.09-1.34). Individuals with fair or red
hair had 1.27 times the damage of those with black or brown hair (95%
C.I. 1.13-1.43) and blue eyes had 1.23 times the damage of other eye
colors (95% C.I. 1.11-1.37). Other factors that were statistically
significant and their prevalence ratios were longest held outdoor
occupation (1.28), outdoor leisure activity once or more per week
(0.85), swimming as a preferred activity (0.52), boating as a
preferred activity (0.69 -- borderline significance based on only 24
yes answers), maintenance of a suntan (0.67), and use of sunscreens
sometimes versus never (0.73). Studied but not statistically
significant were number of Celtic grandparents, skin reaction to
sunlight, and skin color of forearm.
3. When all of the above-mentioned factors were combined in a multiple
logistic regression, male sex, age, the square of age, burn reaction
to sunlight, and outdoor occupation were the only statistically
significant factors.
4. The authors concluded that "... more attention should be paid to the
various possibilities for confounding .... detailed information on
potential confounders should be sought and appropriate analyses
performed."
Methodology:
A cross sectional survey of 1,216 persons (560 M, 656 F aged 16 to 86)
attending a triennial health survey from 23 November to 3 December 1981 in
Busselton (200 km south of Perth) in western Australia. The subjects
representing 35 percent of those attending the survey, were chosen
randomly and represented 17 percent of the residents of Busselton. The
health survey was based on information obtained by a questionnaire and on
semi-objective measures by the investigators.
* * * DRAFT FINAL * * *
-------
B-25
Comment:
This is a useful survey of the interaction of many of the key variables in
this area of research. There was, however, no discussion of the selection
process involved in comparing those who volunteered for the survey
compared to those other citizens of Busselton who did not.
* * * DRAFT FINAL * * *
-------
B-26
Reference:
Holman, C.D.J., Armstrong, B.K., and Heenan, P.J. "Cutaneous Malignant
Melanoma in Women: Exogenous Sex Hormones and Reproductive Factors." Br.
J. Cancer 50:673-680 (1984).
Investigator Results:
1. In a western Australia case-control study on women, there was no
consistent evidence of a relationship between incidence rates of
different melanoma histogenic types and age at menarche, duration of
menstrual life, degree of obesity, number of pregnancies of more than
20 weeks duration, or use of oral contraceptives (OC).
2. No consistent trend was observed when OC was examined by age periods
(10-19, 20-29, and 30+ years) and time intervals before diagnosis
(10+, 5-9, less than 5 years).
3. Borderline evidence was shown of an association between superficial
spreading melanoma and duration of estrogen use.
4. On the basis of seven studies on the relationship of OC use and
melanoma, the authors estimated that the total melanoma incidence rate
of an OC ever-user was unlikely to be increased by over one-third the
rate in never-users. Only one of the seven studies produced a
statistically significant relation and the combined estimate from the
studies was 1.12 (95% C.I. 0.94-1.33). -
5. The authors noted that data from Australia may not generalize to
populations with a low incidence of melanoma. "With this proviso, the
results of this study give no support for a role of endogenous sex
hormones or related phenomena in the aetiology of melanoma in women."
Methodology:
A case-control analysis (see introduction to this section) of 276 female
melanoma patients under 80 years old (out of 373 total) identified in the
West Australia Lions Melanoma Research Project from 1980 to 1981. The
mean age was 44.9 years (range 10-79 years). All but 7 tumors were
histologically classified by six pathologists into one of four categories;
Hutchinson's melanotic freckle (HMF), superficial spreading (SSM),
unclassifiable (UCM), or nodular (NM). Sixty-two percent of the lesions
were classified as SSM. 276 Age- and electoral subdivision-matched
controls were selected from the Australian Commonwealth Electoral Roll and
a few from a public school student roll. Information on cases and
controls, obtained by interviewer questionnaire, included constitutional
and hereditary factors, sun exposure, diet, exposures to known or
suspected carcinogens, menstrual and obstetric histories, weight and
height, and history of OC use or other estrogenic preparations.
Conditional maximum likelihood estimation was used to calculate odds
ratios (ORs).
* * * DRAFT FINAL * * *
-------
B-27
Experimental Design and Analysis Issues:
An analysis of 278 female melanoma cases and age- and residence-matched
controls.
For Result 3:
When only SSM cases and controls were analyzed by OC use, while
controlling for skin reaction to sunlight, hair color, number of
raised nevi on arms, age at arrival of migrants, level of residential
sun exposure, weekend recreational sun exposure (from 10-24 years
old), and frequency of summer outdoor activities, odds ratios were
0.78 in under 2-year OC users, 2.24 for 2-4 year users, and 1.62 for
5+ year users.
There was borderline evidence of an association between SSM and total
duration of estrogen use (OR=2.26 for 13+ months use, p=0.082). When
controlling for the same potentially confounding factors (above), the
ORs were 2.51 in under 12-month users and 2.15 in 12+ month users.
* * * DRAFT FINAL * *
-------
B-28
Reference:
Holman, C.D.J. and Armstrong, B.K. "Pigmentary traits, ethnic origin,
benign nevi, and family history as risk factors for cutaneous malignant
melanoma." JNCI 72:257-266 (1984).
Investigator Results:
1. In a case-control study of 511 patients and 511 age-, sex-, and
residence-matched controls in western Australia, the strongest risk
factor was the number of palpable benign nevi on a subject's arms.
The crude relative risks (RRs) of melanoma compared to persons having
no nevi were 2.0 for 1-4 nevi, 4.0 for 5-9 nevi, and 11.3 for 10 or
more nevi (p=<0.0001). The authors noted "the probable importance
of nevi either as an early stage in the pathogenesis of non-HMF
melanomas or an indicators of increased risk."
2. Inability to tan was the most important pigmentary trait associated
with risk of melanoma. Susceptibility to sunburn and hair color were
also significantly associated with risk of melanoma independent of
tanning ability. After controlling for these traits, skin color and
eye color had no additional effects. The authors concluded that the
results suggested "that the ability to tan quickly in response to
sunlight is of prime importance in reducing risk of skin cancers and
is more important than the base-line skin color... That acute
reaction to sunlight and hair color had significant effects" after
taking chronic reaction into account "does not necessarily mean that
they operate through different mechanisms."
3. Persons with two or more Southern European grandparents had a reduced
risk of melanoma (odds ratio (OR) = 0.39, p=0.025). Persons of Celtic
origin did not have a significantly increased risk of melanoma
(OR=1.18).
4. Persons with one or more affected blood relatives were at higher risk
(OR=2.69, p=<0.0001).
5. The effects of pigmentary traits, benign nevi, ethnic origin, and
family history as risk factors were largely independent of one another
based on a stepwise logistic regression.
Methodology:
A case-control study of 511 melanoma patients and 511 age-, sex-, and
residence-matched controls in western Australia (see introduction to this
section).
Experimental Design and Analysis Issues:
A case-control study of 511 cases diagnosed from 1980-1981 and 511 age-,
sex-, and residence-matched controls in western Australia.
* * * DRAFT FINAL * * *
-------
B-29
For Result 1:
Both the number of nevi on the arms and history of mole excision were
strong risk factors for melanoma . The OR for two or more excised
benign nevi was 5.09 (95% C.I. 0.26-11.46) and was 2.35 (95% C.I.
1.29-4.32) for one or more excised benign moles for the 267
superficial spreading melanoma (SSM) case-control'pairs only. The
association with nevi (one or more raised) was strongest from SSM
(OR=3.00, 95% C.I. 1.98-4.57) and weakest for Hutchinson's melanotic
freckle (HMF) (OR=1.54, 95% C.I. 0.73-3.27).
For Result 2:
No consistent relationship was seen between risk of melanoma 'and skin
color of the dorsum of the hand or of the shoulder tip although there
was a tendency toward an increased risk in the fairest skin color
group: for dorsum lightest skin group (reflectance >52%) OR=1.81
(95% C.I. 1.04-3.14) and for shoulder lightest skin group (reflectance
>56%) OR=1.46 (95% C.I. 0.88-2.40). A stronger, significant
association was observed for skin color of upper inner arm with a
threefold greater risk for the fairest skin group (reflectance >65%,
OR=3.07, 95% C.I. 1.47-6.39) compared to the darkest skin group
(reflectance <47%). Persons with red hair were at higher risk
(OR=2.33, 95% C.I. 1.26-4.30) as were those with blue eyes (OR=1.61,
95% C.I. 1.16-2.24).
Melanoma risk was strongly and significantly associated with acute and
chronic skin reactions to sunlight with the highest risks among those
whose acute reaction was to blister (OR=3.39, 95% C.I. 1.90-6.03) and
whose chronic reaction was to freckle rather than tan (OR=3.53, 95%
C.I. 1.82-6.84). In a stepwise logistic regression, chronic reaction
to sunlight was the most important risk factor (OR for no tan = 2.44,
p=0.000002 for contribution of step), followed by acute reaction to
sunlight (OR for blistering = 2.08, p=0.008) and then hair color (OR
for red hair = 1.89, p=0.039). After taking these three factors into
account, skin color of upper inner arm and eye color were associated
with much lower ORs compared to the crude ORs presented previously.
The effects of skin color were apparently explained almost entirely by
acute and chronic reaction to sunlight. An increasing severity of
burn as an acute reaction was not associated with an increased risk.
Risks of all histogenic types increased with decreasing ability to tan
(a chronic reaction). The association was much stronger for HMF (OR
for no tanning ability = 10.01, 95% C.I. 0.97-103.1, p=0.0002 for
linear trend) than for SSM, unclassifiable melanoma (UCM), or nodular
melanoma (NM). Each histogenic type was also associated with acute
reaction to sunlight, and hair, skin (upper inner arm), and eye color,
except that HMF and UCM were not associated with eye color.
* * DRAFT FINAL * * *
-------
B-30
For Result 3:
Possession of two or more Southern European grandparents was
significantly associated with reduced risk in a logistic regression
designed to separate the independent effects of having two or more
grandparents from various ethnic groups and to control for confounding
by age at arrival in Australia (OR=0.39, 95% C.I. 0.17-0.89,
p<0.05). Low risks were also seen for African or Asian and Northern
European grandparents. The strength of the protective effect of
Southern European grandparents was reduced when confounding by
pigmentary traits was controlled.
For Result 4:
A history of melanoma in a blood relative was a significant risk
factor (OR for two or more affected relatives=5.00, 95% C.I.
1.45-17.27), especially for HMF (OR for one or more=5.50, 95% C.I.
1.44-20.97.)
* * * DRAFT FINAL * * *
-------
B-31
Reference:
Holman, C.D.J. and Armstrong, B.K. "Cutaneous malignant melanoma and
indicators of total accumulated exposure to the sun: An analysis
separating histogenic types." JNCI 73:75-82 (1984).
Investigator Results:
1. In a case-control study of 511 melanoma patients diagnosed from
1980-1981 and 511 age-, sex-, and residence-matched controls in
western Australia, duration of residence in Australia of migrants was
positively associated with melanoma risk.
2. Mean annual hours of bright sunlight averaged over a lifetime'was
positively associated with risk of melanoma .
3. Risk of melanoma increased with worsening skin condition (e.g.,
actinic damage).
4. Persons with a history of nonmelanptic skin cancer were at higher risk
of developing melanoma.
5. The authors concluded that the hypothesis that melanoma is related to
sun exposure was supported by the observed associations with actinic
skin damage, history of nonmelanotic skin cancer, duration of
residence in Australia of migrants, and mean annual hours of sunshine
at subject's area of residence.
6. The findings regarding HMF in this and the previous Holman and
Armstrong (1984) study "are all consistent with the assertion that the
causal relationship of sunlight with HMF is more direct than with the
other histogenic types.... HMF ... may be related to the total
accumulated dose of sunlight received on exposed body sites." The
authors suggested that the results generally supported the hypothesis
that "an individual's maximum potential to develop SSM would be fixed
by the number of initiated nevus cells induced by UV radiation and
other agents in-childhood and young adulthood ... Evidence
associating UCM with sun exposure was much weaker than for other
histogenic types... A role of sunlight in the causation of NM is
supported by these results." The risk factors for NM were a mixture
of those associated with HMF and SSM which was consistent with the
authors suggestion that "NM represents a common end stage of the other
histogenic types."
Methodology:
A case-control study of 511 melanoma patients and 511 age-, sex-, and
residence-matched controls in western Australia (see introduction to this
section).
* * * DRAFT FINAL * * *
-------
B-32
Experimental Design and Analysis Issues:
For Result 1:
For migrants moving to Australia (23% of all subjects), duration of
residence was based on year of arrival. For native-born Australians,
duration was taken as age at interview. The odds ratios (ORs) for all
melanomas and each histogenic type increased with increasing duration
of residence. This trend was strongest for Hutchinson's melanotic
freckle (HMF) (OR for 60 or more years resident = 6.35, 95% C.I.
1.11-36.45) and nodular melanoma (NM) (OR for 60 or more years
residence = 14.72, 95% C.I. 1.16-186.16).
For all melanomas combined, age at arrival was a better predictor of
melanoma risk (OR for arrival at 30 years or later = 0.30, 95% C.I.
0.08-1.13) than was duration of residence. The same result was
observed when SSM and unclassifiable melanoma (UCM) were analyzed
separately. It was impossible to separate the effects of age at
arrival and duration of residence for HMF and NM. For SSM subjects,
the incidence of SMM among migrants arriving before 9 years was near
to or greater .than that among persons born in Australia (OR for SSM
arriving at 5-9 years = 1.65, 95% C.I. 0.34-7.97), whereas for
migrants arriving after 10-15 years of age, the risks were lower than
for native born Australians (ORs <0.38 for ages 15-19, 20-24, 25-29,
>30). The authors suggested that "it is possible that exposure to
sunlight in childhood is a factor in the production of benign nevi,
which have their strongest relationship with SSM, probably as
precursor lesions." They cited results showing a trend toward higher
proportions of nevi in those arriving in Australia before 10 years of
age compared with those arriving later in age (p=0.009).
For Result 2:
Among native-born Australians, ORs significantly increased with
increasing mean annual hours of bright sunlight for all melanomas
combined (p=0.003) and SSM (p=0.02). The OR gradient was steepest for
HMF. An analysis of the effects of more than 2,800 hours of sunlight
exposure annually at different ages showed elevated ORs in all age
periods for HMF (ORs =1.33 and 3.55 for exposure at >40 years and
0-9 years, respectively). High exposure at ages 10-24 years was a
strong risk factor for SSM (OR=11.31, 95% C.I. 1.40-91.11). Exposure
at 25-39 years was also associated with an increased risk of SSM
whereas exposure at 0-9 years and >40 years had no effect. An
elevated risk was also observed for those with SSM who had high-sun
exposure 10-19 years prior to diagnosis.
* * * DRAFT FINAL * * *
-------
B-33
For Result 3:
The ORs increased progressively with worsening skin condition (graded
by CMT) for all melanomas combined (p=0.003), HMF (p=0.048), and SSM
(p=0.021). HMF had the strongest association with CMT (ORs for grades
5 and 6 were 4.05 and 4.37, respectively). No case had a CMT graded
less than 4.
For Result 4:
Persons treated for at least one nonmelanotic skin cancer had more
than a threefold increase in melanoma incidence. The OR for all
melanomas combined was 3.71 (p=0.000001). For HMF and SSM, the ORs
were 5.25 (p=0.001) and 3.33 (p=0.011), respectively, After
controlling for the effects of chronic and acute skin reactions to
sunlight, hair color, and number of European, African, and Asian
grandparents, the OR for all melanomas combined dropped to 2.87
(p=0.0002) suggesting that the association with nonmelanotic skin
cancer was explained only partly by constitutional factors.
* * * DRAFT FINAL •* * *
-------
B-34
Reference:
Holman, C.D.J., Armstrong, B.K., and Heenan, P.J. "Relationship of
cutaneous malignant melanoma to individual sunlight-exposure habits."
JNCI 76:403-414 (1986).
Investigator Results:
1. An increased incidence rate of superficial spreading melanoma (SSM)
was associated with low total outdoor exposure in early adulthood and
frequent participation in boating and fishing. SSM of the trunk was
related to frequency of sunbathing at 15-24 years of age and to
exposure of the trunk while working outdoors.
2. In females, the rate ratio for all types of melanoma occurring on the
trunk was 12.97 (95% C.I. 2.0-83.9) in those who wore a bikini or
bathed nude at 15-24 years of age compared to those who wore a more
conservative one-piece bathing suit. There was little evidence that
sunbathing or wearing a bikini within 10 years of case diagnosis were
risk factors for melanoma of the trunk.
3. After control of confounding due to constitutional factors, only
Hutchinson's melanotic freckle (HMF) showed a relationship to severe
sunburn. For nodular melanoma (NM), sunburn appeared to be protective.
4. Although many of the results supported the hypothesis that melanomas
other than HMF are related to occasional bursts of recreational sun
exposure during early adult life, little support for the hypothesis
was obtained when recreational sun exposure was expressed as a
proportion of total outdoor exposure (a proportion which had been
considered a priori to be an index of intermittent sunlight
exposure). The authors noted, however, that "some but not all of our
results" supported the hypothesis that intermittent exposure to
sunlight plays an important role in etiology of SSM.
5. The authors concluded that their results suggested that "nevi and
occasional or recreational sun exposure interact to produce an effect
on rate of SSM greater than the addition of the two independent
effects, but ... less than expected Cbased on! multiplication of
effects."
Methodology:
A case-control study of 507 melanoma patients and 507 age-, sex-, and
residence-matched controls in western Australia (see introduction to this
section). Thirteen case-control pairs were excluded from the analysis of
recreational and occupational outdoor exposure because they were rated by
the interviewers as providing poor information.
* * * DRAFT FINAL * * *
-------
B-35
Experimental Design and Analysis Issues:
For Result 1:
With the exception of HMF, all histologic types of melanoma were
inversely associated with mean total outdoor exposure (i.e.,
occupational and recreational) in summer (after controlling for
constitutional factors). The strongest inverse association was for
SSM with odds ratios (ORs) of 0.82, 0.72, and 0.57 for mean total
outdoor exposures of 11-15, 16-22, and >23 hours/week, respectively
(p=0.092 for trend). In an analysis of the relationships of
histologic types to high levels of total outdoor exposure (>23
hours/week) at different ages, the inverse associations were strongest
at 10-24 years of age for SSM (OR=0.45, 95% C.I. 0.21-0.98) and for NM
(OR=0.08, 95% C.I. 0.01-1.33).
SSM was associated with frequent participation (once or more per week)
in boating and fishing (OR for boating = 2.43, 95% C.I. 1.10-5.39; OR
for fishing = 2.72, 95% C.I. 1.15-6.43). There was little evidence of
a relationship between SSM and either swimming or sunbathing (either
at 15-24 years of age or 0-9 years prediagnosis). The evidence was
stronger when SSM on the trunk was considered separately. For
sunbathing at 15-24 years of age, ORs (with never sunbathing as
reference) were 1.20 (95% C.I. 0.51-2.81) and 2.55 (95% C.I.
1.05-6.19) for sunbathing less than.once per week and once or more per.
week, respectively (p=0.044 for trend). . For sunbathing within 10
years of diagnosis, the equivalent ORs were 1.50 (95% C.I. 0.53-4.20)
and 1.56 (95% C.I. 0.62-3.93) (p=0.354 for trend). For other
histogenic types, the results did not indicate an association with
water sports and sunbathing except that regular swimmers had a reduced
rate of HMF (ORs =0.98 and 0.26 for participation less than once per
week and once or more per week, respectively; p=0.005 for trend).
In an analysis of habits of dress regarding the primary site of the
case and the same site on the control during outdoor work in summer,
the ORs (except for SSM) were higher in persons if the body site was
sometimes exposed rather than usually exposed or usually covered. For
SSM there was a linear trend of increasing risk from usually covered
to sometimes exposed to usually exposed. The ORs for SSM of the trunk
were 2.93 and 5.96 for sometimes exposed and usually exposed,
respectively (p=0.032 for trend).
* * * DRAFT FINAL * * *
-------
B-36
For Result 2:
The type of bathing suit worn by females while at the beach in summer
was a strong risk factor for melanoma. In- females who wore a bikini
or bathed nude at 15-24 years of age (compared to those who wore a
one-piece bathing suit with high back) the OR was 12.97 (95% C.I.
1.95-83.94). Similar results were seen for those who wore a bikini
0-9 years prediagnosis (OR=8.94, 95% C.I. 1.45-55.07). An additional
analysis suggested that exposure at 15-24 years was more relevant than
exposure 0-9 years prediagnosis. No relationships were observed
between type of bathing suit and melanomas occurring at sites other
than the trunk. The association between SSM of the trunk and exposure
0-9 years prediagnosis or at 15-24 years of age was weaker than for
the other histologic types combined.
For Result 3:
In an analysis of the relationship of histologic types to the highest
severity of past sunburn (no sunburn > peeling sunburn > painful
sunburn > blistering sunburn), HMF was related to the occurrence of
severe sunburn even when confounding effects (eg., hair color, skin
reaction to sunlight, ethnic origin, age at arrival in Australia) were
taken into account. The ORs for HMF were 0.64, 2.45, and 2.78 for
peeling sunburn, painful sunburn, and blistering sunburn, respectively
(p=0.059 for trend). For MM there was a protective effect of severe
sunburn and for SSM and UCM there was no association.
After controlling for confounders, no relationship was observed
between childhood sunburn or sunburn in early adulthood and any
histologic type of melanoma. After controlling for confounders,
reported use of sunscreen appeared to provide no protection against
any histologic type of melanoma.
For Result 4:
In an analysis of intermittent recreational exposure, exposure was
expressed as a percentage of total outdoor time in summer and the
variable was called ROEP. For example, ROEP would equal 29% for 2
days recreational exposure and 5 days occupational outdoor exposure
per week (2 out of 7 days). ROEPs of 30-100% are indicative of
increasing concentration of outdoor time during leisure days. The
effects of ROEP were examined for each histologic type, for four age
groups (never, 10-24, 25-39, and >40 years), and for four intervals
before diagnosis (0-4, 5-9, 10-19, and >20 year prediagnosis). The
results provided little evidence of an association between melanoma
and ROEP. Reanalysis using two different measurement approaches for
recreational exposure to sun also provided no stronger evidence of an
association.
* * * DRAFT FINAL * * *
-------
B-37
ELWOOD AND COLLEAGUES -- WESTERN CANADA MELANOMA STUDY
This major case-control study of melanoma was conducted in Western Canada
consisting of the provinces of British Columbia, Alberta, Saskatchewan, and
Manitoba.
Cases of newly diagnosed, histologically confirmed primary cutaneous
malignant melanomas were obtained through the cancer registries of each
province during the two year period from 1 April 1979 to 31 March 1981. There
were 801 patients who were age eligible out of the 904 diagnosed. Forty
patients could not be located, 21 were dead, for 21 the physician felt that an
interview would not be in the patients' best interests, and 54 persons
declined participation in the study. The remaining 665 (83 percent) were
interviewed along with matched controls.
Each patient was matched by gender and age within two years with a control
subject selected at random from medical insurance plan lists of subscribers,
which cover virtually the entire adult population of each province.
Fourteen cases had acral lentiginous melanoma and 56 had lentigo maligna
so most analyses refer to the 595 cases with other diagnoses, i.e., 415
superficial spreading melanoma, 128 nodular melanoma, 23 unclassified
melanoma, and 29 borderline melanomas. Of the 595 cases, 361 were females and
234 males. . ..'•..
A standardized abstract of the medical record was made for each patient
including data on symptoms, treatment, and recurrences. Pathological slides
were reviewed in a standardized manner by one of two pathologists; for 20
percent of the patients, slides were unavailable resulting in use of the
original pathology report.
Patients and controls were interviewed (1.5 to 2.0 hours) in their homes
by trained interviewers using a standardized questionnaire. Information was
obtained on pigmentation, skin freckling in childhood, sensitivity to
sunlight, tanning, and sunbathing both as an adult and as a child, residence,
occupational history, recreational activities with specific reference to
sunlight exposure, medical history, chronic drug use, family history, diet,
smoking and alcohol consumption, and for women reproductive history and use of
oral contraceptives and menopausal estrogens.
Skin and hair color were determined by direct comparison with prosthesis
and wigmaker samples made specifically for the project. Eye color was
recorded based on direct observation/ Natural hair color in childhood was
asked for those whose hair had greyed with age.
* * * DRAFT FINAL * * *
-------
B-38
Reference:
Elwood, J.M. and Gallagher, R.P. "Site distribution of malignant
melanoma." Can. Med. Assoc. J. 128:1400-1404 (1983).
Investigator Results:
1. The anatomic site for a series of 300 cutaneous malignant melanoma
cases was studied. Sites usually covered by clothing had lower rates
than those usually exposed. Rates of melanoma were higher for the
face and upper arms in both sexes; rates were low for the abdomen and
buttocks and for the forearm and hands in both sexes. Rates for the
back, especially below the scapula, were higher in males than females;
rates for the leg were greater than expected for women and lower than
expected for men.
2. The mean age at diagnosis was 44.9 years. For tumors of the face the
mean age was 52 years, and for tumors of the trunk and limb it was
43-44 years.
3. No seasonal variation of incidence was found.
4. The authors commented that "if melanoma occurrence were related
directly to the total amount of solar exposure, as is the case for
squamous cell carcinoma of the skin, the site-specific increase in
melanoma could be explained by the argument that the exposure of those
sites (the lower limbs in women and the back in men) has increased in
recent years owing to changes in recreational and clothing habits,
whereas the solar exposure of sites such as the face has presumably
not changed." Prior studies have not validly compared tumor
distribution with the body surface area.
Methodology:
A case series of 300 melanomas diagnosed from 1 January 1976 to 31
December 1979 in Vancouver, British Columbia, Canada were examined. Of these,
281 cases of superficial spreading or nodular melanoma were available for
analysis. The site of the primary lesions was recorded on standardized forms
for each case. The distribution was compared to the surface area of
well-defined body surface divisions.
* * * DRAFT FINAL * * *
-------
B-39
Reference:
Elwood, J.M., Williamson, C. and Stapleton, P.J. "Malignant melanoma in
relation to moles, pigmentation, and exposure to fluorescent and other
lighting sources." Br. J. Cancer 53:64-74 (1986).
Investigator Results:
1. In a case-control study of 83 melanoma patients and 83 age-, sex-, and
residence-matched controls from England, the presence of three or more
moles on the upper arm was significantly associated with an increased
risk of melanoma (relative risk (RR)=13.3).
2. The presence of many freckles on adults was significantly associated
with an increased risk of melanoma (RR=6.0).
3. Risk factors which were not statistically significant included
reaction to sun .(burn easily/tan rarely), adult hair color (red,
blond), and history of severe sunburn.
4. Exposure to undiffused or diffused fluorescent lighting did not
produce consistent results. The authors concluded that "The current
results on fluorescent lighting are equivocal. ... The current
results are consistent with a real situation of no association or a
weak positive association leading to an apparent stronger positive
• association because of bias."
Methodology:
A small case-control study of 83 out of 112 National Health Service
patients in Nottingham, England who had a first primary cutaneous melanoma
between 1 July 1981 and 31 March 1984. Controls were matched "precisely"
for age, sex, and residence (in same area) and were selected from all
persons who had an in- or outpatient attendance at a Nottingham hospital
during the same time period. Seven controls were replaced by a second
choice. Home interviews were conducted to obtain information on full
occupational history, including in particular the lighting in the
workplace. A count of palpable moles on the upper arm was also taken.
Experimental Design and Analysis Issues:
For Result 1:
Forty-two percent of melanoma patients and five percent of controls
had three or more raised moles on the upper arm. Three or more moles
gave a relative risk of 17.0 when compared to no moles (95% C.I.
6.6-43.8). In a multiple logistic model, moles on the upper arm had
an RR of 13.3.
* * * DRAFT FINAL * * *
-------
B-40
For Result 2:
Fifty-five percent of melanoma cases compared to 16 percent of
controls had many freckles on the face and arms, giving an RR of 7.0
(95% C.I. 3.3-14.5) compared to those with no freckles. Freckles in
childhood were associated with a similar but less strong
relationship. In the multiple logistic model, the RR for many adult
freckles was 6.0.
For Result 3:
The RR associated with red or blond hair was 2.5 (95% C.I. 1.2-5.3)
for adulthood and 2.2 (95% C.I. 1.1-4.8) for childhood. Blue/grey
eyes were not significant but the RR was 1.3 (95% C.I. 0.6-2.9). The
tendency to burn easily/tan rarely gave an RR of 4.6 (95% C.I.
1.9-11.1) compared to the tendency to tan/no burn. A history of
sunburn causing pain for two days or more gave an RR of 3.2 (95% C.I.
1.7-5.9). No significant risk was associated with social class. In
the multiple logistic model, risk factors that were not statistically
significant included reaction to sun (burn easily/tan rarely), adult
hair color (red, blond), and history of severe sunburn. Ten cases and
six controls spent a year or more living in a tropical or subtropical
climate (RR=1.8, 95% C.I. 0.6-5.1), 14 due to military service
overseas.
For Result 4:
No consistent effects were found with undiffused or diffused
fluorescent lighting. Control for other variables did not change the
results. Some of the occupational exposures to light sources were
interesting, e.g., welding, cinema projection (carbon arc lamp),
printing/dyeline copying, and UV lights, but the numbers of cases are
too small for a definitive study.
* * * DRAFT FINAL * * *
-------
B-41
DUBIN AND COLLEAGUES -- NYU MEDICAL CENTER MELANOMA STUDY
This large case-control study is based on patients seen at New York
University Medical Center between 1972 and 1982 although the study was begun
in 1979.
A case series of patients entering one of three New York City hospitals
with a newly diagnosed primary malignant melanoma was constructed based on an
interview and physical examination. Two-thirds of the cases were diagnosed in
1976 to 1980 because of changes in the number of cases presented at these
hospitals, and from 1980 onward the speed with which patients were treated at
and released from Day Surgery precluded interview for the study.
Potential controls were randomly chosen from patients 20 years of age and
older with a first visit to the New York University Skin and Cancer Unit
general skin clinic or a reregistration after two years absence. A total of
748 controls were interviewed between October 1979 and January 1982, about
twice as many as concurrent cases. An additional 426 skin clinic patients
refused to participate as controls. Based on a random sample of 100, those
who refused participation differed only negligibly with respect to age, sex,
marital status, race, year of visit, or dermatologic diagnosis.
Of the 1,132 potential cases, 29 were excluded for age less than 20 or
unknown age (10 cases), non-white race (6), or .previous melanoma (13).. Of the
748 potential controls, 163 were.excluded for non-white race (79 controls),
age (1), previous melanoma (8), other prior skin cancer (35), other prior
malignancy (18), or current diagnosis of cancer (22). For the study, this
left 1,103 valid cases and 585 valid controls.
Supplementary questions were added to the original interview in 1979.
Thus distribution of moles was asked of 289 cases and all 585 controls, skin
color for 472 cases and 585 controls, and freckling diagram for 208 cases and
457 controls. In addition, moles and freckles were counted during physical
examination for the patients.
There were 566 female and 537 male cases and 320 female and 265 male
controls. Histology of the cases was as follows: superficial spreading
melanoma (813 cases), nodular melanoma (102), lentigo maligna melanoma (52),
acral lentiginous melanoma (35), unclassified radial growth phase (50), other
(34), and unknown (17).
* * * DRAFT FINAL * * *
-------
B-42
Reference:
Dubin, N., Moseson, M., and Pasternack, B.S. "Epidemiology of malignant
melanoma: Pigmentary traits, ultraviolet radiation, and the
identification of high-risk populations." Recent Results in Cancer
Research 102:56-75 (1986). A partial report was also included in a Lancet
letter to the editor (26 March 1974, 1:704) re: Fluorescent Lights.
Investigator Results:
1. In a multiple logistic regression of data on 1,103 melanoma cases and
585 (non-matched) controls, relative risks (RRs) were 2.39 for no
ability to tan relative to average and 3.90 for no tendency to burn
relative to tendency for painful burn. (Note that the latter' RR is
opposite of the usual finding.)
2. The RR for a history of freckling relative to no freckling was 3.61.
The RR for more than 100 moles compared to 0-25 moles was 3.67
3. Red hair color and blue eyes were associated with higher risks of
melanoma.
4. Mostly outdoor work was associated with an RR of 2.43 relative to
mostly outdoor work.
5. Subjects reporting that free time was spent mostly outdoors had 1.65
times the risk of those spending free time mostly indoors. In the
multiple logistic regression the RR dropped to 1.03.
6. A previous medical history of solar keratosis had an RR of 4.69, and
included ten percent of the cases and 1.6 percent of the controls.
7. The interview data produced a relative risk of 2 or 3 for 7-8 or 9+
hours of fluorescent light exposure per day. The reliability
questionnaire data showed relative risks of 0.95 and 0.61 for the same
exposure parameters, respectively (i.e., a protective effect). The
authors commented that the "results for fluorescent light exposure are
disturbing. The reliability study data clearly do not support either
the interview data that formed the basis of our preliminary report or
the findings of Beral et al. (1982). Despite all our efforts to the
contrary, interviewer bias may have affected our fluorescent light
data."
8. An increased risk was found for persons 72 inches (183 cm) or more in
diameter, for those with the largest body surface area, and for those
who were unmarried (never married, separated, and divorced).
* * * DRAFT FINAL * * *
-------
B-43
Methodology:
A case-control study of 1,132 cases with newly diagnosed primary malignant
melanoma given an interview and a physical examination from three
hospitals in New York between November 1972 and January 1982. From 1980
about one -third were lost because of one-day care and lack of interviews
(note potential bias in ascertainment). There were 1,103 valid cases
remaining after exclusions.
The controls were randomly chosen from patients age 20 and over at one of
the hospitals for a first visit or reregistration after 2 years absence.
Different interviewers were used. Of the 748 controls interviewed during
the period October 1979 to January 1982, 585 remained after exclusions.
An additonal 426 potential controls refused to participate.
A reliability questionnaire was mailed to all cases and controls in 1983.
The questionnaire asked for information on sun exposure, fluorescent
lighting exposure, and skin color (a repeat of some information from the
interview). The physical exam included a count of moles and freckles.
Comment:
Since cases and controls were not concurrent, interpretations are made
more difficult.
* * * DRAFT FINAL * * *
-------
REVIEWS OF OTHER EPIDEMIOLOGICAL STUDIES
-------
References
Acquavella, J.F., Wilkinson, G.S., Tietjen, G.L., Key, C.R., Stebbings,
J.H., and Voelz, G.L. "A Melanoma Case-Control Study at the Los Alamos
National Laboratory." Health Physics 45:587-592 (1983).
Investigator Results:
1. The case-control study focusing on Los Alamos National Laboratory
(LANL) employees did not uncover an association between melanoma
and plutonium (Pu) body burden, or cumulative external radiation
exposure.
2. Melanoma cases were more educated than controls, with melanoma
risks of 2.11 among college-educated and 3.17 among those with
graduate degrees, indicating importance of personal characteristics,
especially higher education, as risk factors for melanoma. No
significant association was apparent for chemists or physicists
of both sexes.
Methodology:
A case-control analysis of 20 malignant melanoma (MM) cases (15M, 5F)
employed at LANL for at least one year and diagnosed after original
employment date reported in New Mexico Tumor Registry (1969-1982),
in Los Alamos Medical Center (1951-1982), and in state records (1950-
1976).
Four LANL controls randomly selected for each case were matched for
sex, ethnicity (Anglo, Hispanic), birth date (_+2.5 yr.), and date of
first LANL employment. Data on each subject consisted of primary job
title, education, dates of initial employment and termination, and
radiation exposure history (Pu body burden, external radiation exposure
2 years prior to case's date of diagnosis and corrected for background
radiation exposure).
Statistical analyses were based on 1-sample, 2-tailed t-test, contingency
tables for matched and unmatched case-control studies, and 95% confidence
intervals for odds ratios. Mantel-extension trend test was used to
test for dose-response relationships.
Experimental Design and Analysis Issues:
A well-designed case-control study of LANL employees, but limited by
the small number of cases.
For Result 1:
No differences between male cases and controls were observed for any
type of radiation exposure (B, neutron) and Pu body burden. There
was no indication of an association between melanoma and any particular
form of radiation exposure. Note that the number of cases in each
B-45
-------
exposure category ranged from 0 (neutron, Pu) to 6 (total external
radiation).
For Result 2:
Contingency analysis suggested that male cases achieved a college edu-
cation or graduate degree more often than controls, and were more likely
to be employed in a professional capacity. For both sexes, no signifi-
cant association was apparent for chemists or physicists. Standardized
rate ratios (SRR) increased with increasing educational attainment
(2.11 for college graduates, 3.17 for graduate level) and the Mantel-
extension test for trend indicated a significant association with in-
creased education (p = 0.04).
B-46
-------
References
Allyn, B., Kopf, A.W., Kahn, M. and Witten, V.H.. "Incidence of Pigmented
Nevi." JAMA 186:890-893 (1963)
Investigator Results:
1. Incidence of nevi was evaluated among 1,000 randomly selected subjects;
greater than or equal to 1 plantar nevi (9%), 1 palmar nevi (5.8%),
1 conjunctival nevi (0.8%), 1 nailbed nevi in whites (0%). Greatest
concentration of plantar nevi was in arch region. Highest nevi
incidence was in dark-skinned ethnic groups and male subjects.
Incidence of nevi increased in first three decades of life and
declined thereafter.
2. Routine excision of pigmented nevi on palms and soles was concluded
to be infeasible and unwarranted because of their frequency.
Methodology:
A descriptive analysis of the incidence of pigmented nevi on 1,000
randomly selected individuals from inpatient and outpatient services
of New York University Schools of Medicine Dermatology Departments
and departmental personnel. Nevi were classified by site (palms, soles,
conjunctival, nailbeds), size, surface characteristics and duration.
Individuals were classified by ethnic background or color (Caucasian,
Latin-American, Negro, Oriental), age, sex, and skin complexion.
Experimental Design and Analysis Issues:
Straightforward descriptive analysis of the data set.
For Result 1:
Incidence of pigmented nevi was provided by site, with 14.3% of overall
nevi incidence on palms and soles. The data generally indicated
increasing nevi incidence with increasing skin color groups and increasing
incidence in the 1st and 2nd decades, a peak in the 3rd decade and
a progressive decline thereafter. Plantar pigmented nevi were predomi-
nantly on the arch.
For Result 2:
Given a potential 10%-25% incidence of pigmented nevi on palms and
soles, which on average indicated a possible 1 in 6 persons having
greater than or equal to 1 nevi on a palm or sole, approximately
30 million pigraented nevi would have to be removed to catch the rare
ones.
B-47
-------
Reference:
Anaise, D., Steinitz, R. and Ben Hur, N.. "Solar Radiation: A Possible
Etiological Factor in Malignant Melonoma in Israel: A Retrospective
Study (1960-1972)." Cancer 42:299-304 (1978).
Investigator Results:
1. Incidence of malignant melanoma (MM) was higher among European-born
Jews (34/10 1 than among African-born Jews (2.7/10 ) or Asian-born
Jews (4.4/10 ).
2. Among European-born Jews of the same age and ethnic background,
MM incidence was higher among earlier migrants (58/1Q , 20-30 years
prior to diagnosis) than more recent migrants (17/10 , 2-5 years).
3. Higher incidence was found among Kibbutz agricultural workers (54/10 )
compared to city residents (17/10 ) and among coastal residents
(35/10 ) compared to mountain residents (20/10 ).
4. The predominant melanoma sites were lower extremity in females
(50%) and trunk in males (30%). Lower extremity melanomas among
females were higher in 50-79 year-olds (55%) than in 0-19 year
olds (41%) and higher in less recent migrants (61%) than recent
migrants (49%) .
Methodology:
A descriptive analysis of all 966 new MM cases reported in the
cancer registry of Israel from 1960-1972. All cases were histolog-
ically confirmed. Cases were classified by sex, age, ethnicity
(African, Asian, European), and tumor site.
Experimental Design and Analysis Issues:
A straightforward descriptive analysis of age-adjusted MM incidence
rates in Israel from 1960-1972.
3-48
-------
Reference:
Austin, D.F., Reynolds, P.J., Snyder, M.A., Biggs, M.W., and Stubbs,
H.A. "Malignant Melanoma Among Employees of Lawrence Livermore National
Laboratory." Lancet 2:712-716 (1981).
Investigator Results:
1. The number of malignant melanoma (MM) cases observed from 1972-1977
among 5,100 Lawrence Livermore National Laboratory (LLNL) employees
was significantly higher (p less than 2xlO~ ) than the expected
number of cases in a comparable age/sex/race/geographical segment
of the San Francisco Bay Area population, sugesting a role of an
occupational factor.
2. Case-control comparisons indicated that MM risk was not associated
with length of employment at LLNL nor with type of monitored radia-
tion exposure.
3. The data did not show an association between MM incidence and all
scientific job classifications combined, but an excess relative
risk was observed among chemists.
Methodology:
Observed MM incidence was calculated from 5,100 full- or part-time
white LLNL employees who lived in Alameda and Contra Costa counties
from 1972-1977. Members were grouped by 5-year age group, sex, year
of study, and concurrent census tract of residence. The 19 MM cases,
identified from the California Tumor Registy (CTR) for the San Fran-
cisco-Oakland Standard Metropolitan Statistical Area, were included
only if diagnosis was made while employed at LLNL. Expected MM inci-
dence for the LLNL study group was estimated based upon age, race,
sex, and census-tract-specific incidence rates among all new MM cases
reported to the CTR for the same counties and time period. The Mantel-
Haenszel procedure was used to test for differences between observed
and expected cases.
Each of the 19 LLNL cases were matched with 4 control LLNL employees
by age group (5-year), race, sex, and area of residence. Information
on cases and controls consisted of duration of employment (referenced
to diagnosis date for each case), cumulative radiation exposure above
background (gamma, neutron, tritium), beta radiation of the skin and
hand, and job classification. Differences between cases and controls
were tested by the 1-sample, 2-tailed t-test. Approximate relative
risks for different job classifications (scientists vs. non-scientists,
chemists vs. non-chemists) were assessed using a method described by
Miettinen.
B-49
-------
Experimented. Design and Analysis Issuese
A comparison of observed-to-expected incident MM cases among LLNL employees
followed by a case-control study.
For Result 1:
From 1972-1977, 19 MM cases (17Mf 2F) were identified. For two types
of melanoma (all or invasive only), the observed number of cases among
males (17 all melanoma, 14 invasive only) was approximately three times
higher than expected (5.6 all melanoma, 4.7 invasive only) (p less
than 6 x 10~ all melanoma, 2 x 10~ invasive only). Similar results
were observed among the entire study group but not among females (only
2 cases). The MM incidence rate for 20-64 year-old white males in
the study group (48.8/10 ) was higher than that for a similar, non
LLNL group of Alameda County residents (11.7/10 ).
For Result 2:
Case-control comparison did not indicate an association between MM
incidence and length of LLNL employment (p=0.804). The mean duration
of employment was 148.8 months for cases, 152.9 months for controls.
Case-control data did not indicate a relation between MM incidence
and any type of radiation (p=0.31 for gamma).
For Result 3:
Case-control data did not show a relation between scientist job classi-
fication and MM incidence (relative risk = 1.64, p=0.348). MM incidence
may be higher than expected among chemists (4/19 cases were chemists
vs. 3/76 controls, relative risk = 6.97, p=0.011).
B-50
-------
Reference:
Bako, G., Hill, G.B., and Hendin, M. "Correlation of Incidence Rates
for Selected Cancers in 29 Census Sub-Divisions of Alberta, Canada,
1961-1981." Ecol. Dis. 2:129-131 (1983).
Investigator Results:
In an analysis of correlations between age-adjusted male and female
cancer incidence rates for 36 cancer sites, two of the 17 highest cor-
relations were between melanoma and cancer of bone and connective tis-
sue (r=0.62) and oesophagus cancer (r=0.61) in females.
Methodology:
A correlation analysis of age-adjusted male and female cancer incidence
rates for 1961-1981 for 36 sites and 29 census sub-divisions of Alberta
calculated from the Alberta Cancer Registry. Only correlations with
r greater than or equal to 0.55 were considered.
Experimental Design and Analysis Issues:
A straightforward pairwise correlation analysis of cancer incidence
rates in Alberta from 1961-1981.
B-51
-------
Reference:
Baker-Blocker, A. "Ultraviolet Radiation and Melanoma Mortality in
the United States." Env. Res. 23:24-28 (1980).
Investigator Results:
Melanoma mortality rates among white males and females in 18 U.S. coun-
:ies for 1950-1969 were correlated with latitude and received ultraviolet
(UV) radiation (p^O.Ol). No significant correlation was found between
melanoma mortality in white males or females separately and UV radiation,
suggesting that factors other than UV radiation may play a role in
melanoma mortality.
Methodology:
Ultraviolet radiation measurements were obtained for 18 U.S. counties—
8 counties from the federal network used Robertson-Berger meters (counts/day),
the other counties used Eppley instruments (received UV radiation).
Robertson-Berger counts/day were converted to UV radiation units/ with
the knowledge that the two instruments' measurements had a correlation
coefficient of 0.88.
Melanoma mortality data was determined for the 18 counties (thus nonrandomly).
A comparison of melanoma mortality rates for each county to those for
all U.S. counties revealed that the range of mortality rates among
the 18 counties was slightly narrower than for all U.S. counties (e.g.,
for white males, 0.2-3.8 per 10 for all U.'S. counties vs. 0.5-2.5
per 10 for 18 counties). Source of melanoma mortality data was not
indicated.
Experimental Design and Analysis Issues:
A cross-sectional analysis of melanoma mortality rates, and latitude
and received UV radiation for 1950-1969 in 18 U.S. counties.
Graphical analysis of received UV radiation and melanoma mortality
rate showed no readily apparent correlation for either sex. Correlation
coefficient for white male melanoma mortality and UV was -0.13, and
for white females was 0.27. A similar analysis of male/female melanoma
mortality ratio vs. latitude for the 18 counties yielded a correlation
coefficient of 0.44 (significant at 10% level), indicating that the
ratio decreases with decreasing latitude. For other skin cancers,
the ratio increased significantly with decreasing latitude (coefficient =
-0.49), significant at 5% level. The author stated that this difference
in the sex ratio strengthened the hypothesis that UV radiation, while
important in basal and squamous skin cancers, was not an important
factor in melanoma. Increasing white female mortality was correlated
with decreasing latitude (coefficient = , significant at 2% level),
but a similar relationship was not observed for white males. There
was a strong correlation between received UV and latitude (coefficient =
0.69, significant at 1% level).
B-52
-------
Reference:
Balch, C.M., Karakousis, C., Mettlin, C., Natarajan, N., Donegan, W.L.,
Smart, C.R., and Murphy, G.P. "Management of Cutaneous Melanoma in
the United States." Surg.,Gyn. and Obst. 158:311-318 (1984).
Investigator Results:
1. Among 4,545 cases of invasive melanoma, a change in mole was the
most common presenting symptom of melanoma.
2. The typical melanoma was relatively thin (less than 1.5 mm), not
ulcerated (except in 9%) and did not reach Stage IV or V.
3. Melanomas occurred equally in both sexes, however the predominant
sites were the trunk for males (53%) and upper and lower extremities
for females (50%) .
Methodology:
A descriptive analysis of 4,545 histologically confirmed invasive melanoma
cases reported from 614 hospitals from throughout the U.S. in 1980.
Patients with melanoma from an unkown site, melanomas of the eye or
mucous membrane, or melanoma diagnosed at autopsy were excluded.
Experimental Design and Analysis Issues:
A descriptive analysis of 4,545 invasive melanoma, cases reported in
1980. Simple comparisons among cases which were described included
symptoms of melanoma (change in mole most common), clinical features
of melanoma (87% of cases with Stages I and II), sex (occurrence was
equal in both sexes), site (male trunk predominance, female upper and
lower extremity predominance), age, and race (98% of cases were white).
Pathologic features which were examined included level of invasion
(35% Level IV, about 25% Level II), tumor thickness (over 33% "thin"
lesion—less than 0.76 ram), ulceration (occurred in 9% of cases, usually
in thicker lesions), and growth pattern (56% superficial spreading,
30% nodular, 14% lentigo maligna). Surgical and nonsurgical treatments
were summarized. Noninvasive melanomas in 225 patients occurred in
a slightly older population and more commonly on the head and neck
than invasive melanomas.
B-53
-------
Reference:
Balch, C.M., Soong, S., Milton, G.W., Shaw, H.M., McGovern, V.J., McCarthy,
W.H., Murad, T.M., and Maddox, W.A. "Changing Trends in Cutaneous
Melanoma Over a Quarter Century in Alabama, USA, and New South Wales,
Australia." Cancer 52:1748-1753 (1983).
Investigator Results:
1. Prom 1955-1980, there was a steady increase in proportion of pa-
tients presenting with clinical Stage I (localized) melanoma among
1,110 patients from the University of Sydney.
2. Melanomas of the trunk in males significantly increased while those
of the head and neck decreased as a proportion of the total. No
significant change in site distribution was observed for females
from 1955-1980.
3. Melanomas became thinner, less invasive, and less ulcerative, and
exhibited more of a radial growth phase. Clinical and pathological
parameters among Alabama and New South Wales patients differed
minimally even when accounting for year of diagnosis.
4. Stage I melanomas became more curable from 1955-1980 with long-
term survival rates increasing slightly for the study population.
5. The authors concluded that the changes that occurred were probably
due to earlier diagnosis and changes in the biological nature of
melanoma.
Methodology:
A descriptive analysis of 1,647 clinical Stage I patients treated at
the University of Alabama in Birmingham and the University of Sydney
(Australia) between 1955 and 1980. The University of Alabama melanoma
Registry provided information on 537 patients, prospectively or retro-
spectively followed-up since 1975, who were treated since 1955. The
Melanoma Clinic at the Sydney Hospital provided information on 1,110
patients prospectively followed-up since 1955, Clinical evaluation,
surgical treatment, and pathologic interpretation were carried out
or supervised by the authors. All melanomas were histologically con-
firmed by one of the two pathology authors. Survival curves were cal-
culated based on the Kaplan-Meier method, the log-rank test was used
to evaluate differences, and the median and chi-square tests were also
used.
Experimental Design and Analysis Issues:
A descriptive analysis of clinical and pathological parameters for
1,647 clinical Stage I melanoma patients.
B-54
-------
For Result 1:
Overall incidence of patients presenting with localized (Stage I) mela-
noma increased in Australia from 73% of all patients before 1960 to
81% for 1976-1980. The upward trend was not as consistent in Alabama
(83% in 1955, 91% in 1975, 86% in 1980).
For Result 2:
An examination of melanoma cases by site and gender indicated a sig-
nificant increase in trunk melanomas among males (40% to 56%, p=0.0004)
and a significant decrease in male head and neck melanomas (36% to
17%, p=0.001). No significant changes in site distribution were ob-
served for male extremities or any site on females.
For Result 3:
Median tumor thickness decreased from predominantly thick lesions (3+ mm)
prior to 1960 to an average 1.3 mm thickness after 1975 (p less than
0.0001). The proportion of thin melanomas increased from 11% before
1960 to 26% during 1976-1980 period. Level of invasion also changed,
with increasing level II melanomas (13% to 25%) and decreasing level IV
melanomas (54% to 42%, p less than 0.0001). Incidence of ulceration
significantly decreased (p less than 0.0001) in Alabama (54% to 35%)
and New South Wales (47% to 19%). The incidence of nodular melanomas
decreased while incidence of superficial spreading melanomas increased
significantly (p less than 0.0001).
For Result 4:
The 8-year survival rate for Stage I patients increased by 5% over
2 decades for both Alabama and New South Wales.
B-55
-------
Reference:
Balch, C.M., Soong, S., Milton, G.W., Shaw, H.M., McGovern, V.J., Murad,
T.M., McCarthy, W.H., and Maddox, W.A. "A Comparison of Prognostic
Factors and Surgical Results in 1,786 Patients with Localized (Stage I)
Melanoma Treated in Alabama, USA, and New South Wales, Australia."
Annals Surg. 6:677-684 (1982)
Investigator Results:
1. Similarities among two series of Stage I melanoma patients (from
the Universities of Alabama and Sydney) were observed with respect
to actuarial survival rates, tumor thickness, level of invasion,
surgical results, and age and sex distributions. The greatest
differences between the two series were observed for anatomic distribu-
tion, growth pattern, and incidence of ulceration. The trunk was
the most common site, occurring more frequently among Australian
patients (37% vs. 28%). The biologic behavor of melanoma in Sidney,
Australia and Birmingham,- Alabama was "virtually the same, with
only minor differences that did not significantly influence survival
rates."
2. In a multifactoral analysis, the dominant prognostic factors (p less
than 0.001) were ulceration, tumor thickness, initial surgical
management, anatomic location, pathologic stage, and level of invasion.
Methodology:
A descriptive and raultifactorial analysis of two series of Stage I
melanoma patients treated since 1955 at the University of Alabama (676)
and the University of Sydney (1,110) . The median follow-up period
of observation was 7 years. Clinical and pathological information
on the patients were provided and all cases were histologically confirmed.
Experimental Design and Analysis Issues;
For Result 1:
A simple comparison of the clinical features of the two patient series
revealed similarities with respect to age- and sex-distributions
and hair and eye color, and differences with respect to anatomic
distribution (lower extremity and trunk melanomas higher in New South
Wales, head and neck melanomas higher in Alabama).
A comparison of prognostic features indicated similarities with respect
to tumor thickness, level of invasion, degree of lymphocyte invasion,
and pigmentation. The two major prognostic differences were incidence
of ulceration (37% in Alabama vs. 22% in New South Wales) and growth
patterns (nodular growth patterns predominated in Alabama and superfical
lateral spreading predominated in New South Wales). Correlations
were observed between tumor thickness and level of invasion, and
between ulceration and gender.
B-56
-------
Overall.survival rates were "virtually the same over a 25-year period .
of time (p=0.14)." Comparison of survival rates for wide local excision
(WLE) vs. WLE plus elective regional lymph node dissection (RLND)
showed improved survival in patients having WLE and RLND for melanomas
of intermediate thickness (0.76-3.99 mm).
No major differences among melanoma patients in Alabama and New South
Wales were found in terms of survival rates and major prognostic
parameters. The dominant prognostic factors, tumor thickness and
ulceration, were virtually the same for the two patient series and
were the most predictive factors for metastatic disease.
For Result 2:
A multiple regression analysis of eleven clinical and prognostic
factors indicated that tumor thickness, ulceration, level of invasion,
initial surgical treatment, and lesion location were significant
(p<0.001 determinants of melanoma incidence among Australian patients
(n=776). Pathological state (I vs. II) was probably not significant
due to the small sampl-e with clinical Stage I, pathologic Stage II
melanoma (n=24). in the 293 Alabama patients, tumor thickness, ulcera-
tion, initial treatment, and pathological stage were significant
factors (p less than 0.01). A final multiple regression model for
the combined data in which all the factors were significant (p less
than or equal to 0.001) was log X(t)/X (t) = 0.3216 (tumor thickness)
+ 0.6455 (lesion location) - 0.9408 (sSrgical treatment) + 0.9013
(ulceration) + 0.4399 (level of invasion) + 1.137 (pathologic/stage),
where X(t)/X (t) was relative risk.
o
B-57
-------
References
Barger, B.D., Acton, R.T., Soong, S., Roseman, J., and Balch, C. "Increase
of HIA-DR4 in Melanoma Patients from Alabama." Cancer Res. 42:4276-4279
(1982) .
Investigator Results:
1. There was a significant increase in frequency of HLA-DR4 phenotypes
(p=0.0003) among 91 melanoma patients (38.5%) compared to 106 controls
(16.0%) producing a relative risk of 3.3. The difference remained
significant after correcting for number of antigens (p=0.0018),
suggesting that DR4 may be associated with melanoma development.
2. Clinically assessed "high risk" patients had significantly lower
DR3 (p=0.02) than "low risk" patients suggesting that DR3-may represent
a marker for long-terra survival
Methodology:
A case-control study of 91 Caucasian melanoma patients treated at the
University of Alabama Melanoma Clinic whose diagnoses were histologically
confirmed. "Low risk" patients (decreased risk for metastases) were
characterized by Stage I melanoma/ tumor thickness less than or equal
to 4.0 mm, and no ulceration. Patients at "high risk" for raetastases
were those with Stage I tumor thickness greater than 4.0 mm or ulceration,
or those with Stage II or III (regional or distant metastases). All
patients were long-term Alabama residents, 93% of whom were born in
southeastern U.S.
The 106 Caucasian controls were (apparently non-randomly) selected
from hospital personnel, paternities, and acquaintances from the Birmingham
metropolitan area. Over 76% of controls were born in southeastern
U.S. There was no restriction on place of residence for cases or controls.
No attempt was made to match controls for age, sex, or ethnic background.
All cases and controls were HLA typed (DRl-5,7) with microdroplet lympho-
cytotoxicity test using B-lyrapocytes. Antigen frequencies were estimated
by maximum likelihood, and odds ratio (relative risk) was estimated
by Woolf method. The likelihood ratio asymptotic chi-square was used
to test for HLA antigen associations.
Experimental Design and Analysis Issues:
An unmatched case-control analysis (91 cases, 106 controls) of HLA-
DR determinants among Caucasians in Alabama, followed by a comparison
among high- and low-risk patients.
For Result 1:
The only HLA-DR determinant that differed significantly between cases
and controls (from DRl-5,7) was HLA-DR4 (38.5% of cases vs. 16.0% of
B-58
-------
controls, p=0.0003) with a relative risk of 3.3. The attributable
risk was 69.4% indicating that the genetic component was an important
factor. The difference was still significant after correction for
number of DR antigens (p=0.0018) .
For Result 2: "
THe DR4 antigen was higher in low-risk patients (n=67) than in controls
(p=0.0009) with a relative risk of 4.0. There was no significant DR4
difference between high- and low-risk patients, when patients were
grouped according to time interval between presentation and HLA typing
(1-4+ years), DR4 remained significantly elevated in each group and
there were no significant DR4 differences among the groups. The DR3
antigen was significantly lower in high-risk patients (n=24) than in
controls (p=0.01) and in low-risk patients (p=0.02), but when corrected
for number of DR antigens the differences were no longer significant.
B-59
-------
Reference;
Bellet, R.E., Vaisman, I., Mastrangelo, M.J., and Lustbader, E. "Multiple
Primary Malignancies in Patients with Cutaneous Melanoma." Cancer
40:1974-1981 (1977).
Investigator Results:
1. The observed number of non-melanocytic, non-cutaneous malignancies
among 295 melanoma patients (23/281) did not differ significantly
from expected based on patient-years at risk.
2. Patients with primary non-melanocytic, non-cutaneous tumors were
significantly older at time of diagnosis, than patients with melano-
ma alone. Patients with multiple primary tumors were at risk for
a significantly longer period of time. With the exception of breast
cancer, the association of cutaneous melanoma with additional non-
melanocytic, non-cutaneous malignancies appeared to be random.
3. The observed number of additional primary melanomas among melanoma
patients (9/267) was significantly greater than expected. Patients
with multiple primary melanomas were not at risk significantly
longer than patients with one melanoma. The authors concluded
that development of additional primary melanomas in patients with
an initial melanoma was not a random event, but probably represents
greater susceptibility to malignant transformation of melanocytes.
Methodology:
A descriptive analysis of 295 histologically confirmed Caucasian mela-
noma patients (126F, 169M) evaluated at the Fox Chase Cancer Center
Melanoma Unit (no dates provided). Where possible, the primary melanoma
was classified according to Clark's system. The study population was
divided into four groups: single primary cutaneous malignant melanoma
(MM) (Group 1, n=259); non-melanocytic, non-cutaneous malignancies
and single primary MM (Group 2, n=22); multiple primary cutaneous MM
(Group 3, n=8); and primary cutaneous non-melanocytic malignancy and
single primary cutaneous MM (Group 4, n=6) (Group 4 not analyzed).
Patients underwent extensive initial evaluation including detailed
history, physical exam, and routine lab studies. Statistical analyses
included use of Fisher's Exact Test, 2-sample t-test, Wilcoxon Signed
Rank test, or the Mantel-Haenszel chi-square test. The expected number
of additional primary cancers was based upon the number of years at
risk for developing cancer for which input age-specific incidence rates
for 1970 were extracted from the Third National Cancer Survey for all
cancers (excluding skin cancer) as well as for melanoma and breast
cancer in Caucasians.
B-60
-------
Experimental Design and Analysis Issues:
For Result 1:
Observed number of patients in Group 2 (23/281, prevalence 8.2%) was
not significantly different from the expected number for each sex (14.71
expected vs. 12 in males, 12.72 vs. 11 in females).
For Result 2:
A comparison of Groups 1 and 2 by sex, age at diagnosis, years at risk
for developing additional malignancy, or histologic type, revealed
no significant difference in male/female ratio. Group 1 patients were
significantly older at time of diagnosis than Group 2 patients (p less
than 0.005 males, p less than 0.1 females), and were at risk a signifi-
cantly greater period of time (p less than 0.005 males, p less than
0.02 females).
For Result 3:
The observed number of cases in Group 3 (9/267, prevalence 3.4%) was
significantly greater than expected (0.31 expected in males vs. 5,
0.22 vs. 3 in females, p less than 0.001 for both sexes). A comparison
of Groups 1 and 3 indicated no significant differences with respect
to sex composition, age at diagnosis, or years at risk of developing
an additional primary melanoma.
B-61
-------
References
Letters to Lancet re: Beral et al. (7 August 1982)
Investigator Results:
T.S. Davies (23 October 1982, p. 935) — polychlorinated biphenyls
(PCBs) are found in office atmospheres and could explain the finding
reported, i.e., it is not the fluorescent light but the PCBs emitted
by fluorescent lighting fixtures in offices.
R.S. Stern (27 November 1982, p. 1227) — findings could be explained
by biases in control selection such as higher socioeconomic -status.
V. Beral and S. Evans (27 November 1982, p. 1227) — a reply presenting
additional data showed that biases in selection of controls did not
occur or did not effect the reported finding.
B.S. Pasternak, N. Dubin, and M. Moseson (26 March 1983, p. 704) -In
an ongoing study of melanoma with 136 cases and 282 skin clinic controls
in New York, fluorescent light exposure was recorded as average hours
per day at home or at work during three time periods: up to five years,
five to ten years, and ten to twenty years previously. Exposure during
the last five years, after adjusting for other risk factors, was insignifi-
cant.
D.S. Rigel, R.J. Friedman, M. Levenstein, D.I. Greenwald (26 March
1983, p. 704)— UV-A exposure from fluorescent lights was l/3000th
that of autumn sun in New York. In a preliminary study of 114 patients
and 228 age-matched controls, no increased risk for melanoma for persons
with fluorescent light exposure was found. "These results suggest
that the factor that puts these indoor office workers at risk is not
their fluorescent light exposure, but their weekend and holidays at
the beach which give them intense sun exposure to normally covered
body sites."
K.J. Maxwell and J.M. Elwood (3 September 1983, p. 579) — spectral
power irradiance calculations show that at a wavelength of 290 nm the
dose received from fluorescent lights may considerably exceed that
from sunlight.
B-62
-------
Reference:
Beral, V., Evans, S./ Shaw, E., and Milton, G. "Cutaneous factors
related to the risk of malignant melanoma." Brit. J. Dermatology 109:165-
172 (1983) .
Investigator Results:
Red hair at age five had a relative risk (RR) of 3.0 (95% C.I. 1.95-4.73) .
The RR was 4.4 for both red hair and fair skin and 3.2 for red hair
and dark skin. Eighteen percent (51 women) of cases and 8% (47 women)
of controls had red hair. Graying of the hair was found to be protective
(RR = 0.60, 95% C.I. 0.46-0,83). After adjusting for hair color, fair
skin had an RR of 1.9 (95% C.I. 1.38-2.50). Fifty-nine percent of
cases and 40% of controls were faired skinned. Eye color had no indepen-
dent effect on risk. Forty-five percent of cases and 37 percent of
controls had blue eyes. An above average number of nevi increased
the risk (RR = 3.5) .
Methodology:
A case-control analysis of 287 white females who attended the Melanoma
Clinic at Sydney Hospital aged 15-84 years and 574 age-matched controls.
Controls from the general population for 213 "old cases" (diagnosed
between 1974-1978) were also matched by area of residence. Controls
for "new cases" (diagnosed between 1978-1980) were selected from hospital
inpatients (excluding those with vascular disease, gyn. disorder, diabetes,
gallbladder or breast disease, or chronic disease of 2+ years duration).
B-63
-------
References
Beral, V., Evans, S., Shaw, H. and Milton, G. "Malignant melanoma
and exposure to fluorescent lighting at work." Lancet 2:290-293 (7
August 1982) .
Investigator Results:
1. In a case-control study of 274 females with melanoma and 549 controls,
exposure to fluorescent light at work was associated with a relative
risk of 2.1 (95% C.I. 1.32-3.32) compared to the 25 women who were
never exposed to fluorescent lights. Relative risks tended to
increase with increasing years of exposure. Melanomas occurred
on the trunk among 24% of the ever exposed compared to the 9% of
others with melanoma. Although other factors were shown to increase
the risk, stratification on these factors did not change the relation-
ship to fluorescent lights. Fluorescent lights in the home were
not associated with melanoma (RR = 0.9, 95% C.I. 0.6-1.6). Previously
collected information on 27 males with melanoma and 35 controls
yielded an RR of 4.4 (95% C.I. 1.1.-17.5).
2. Among cases, a significant excess of lesions (p<.05) was observed
among those exposed to fluorescent lights (24%) compared to those
never exposed (99%).
3. The authors commented that "It is curious, however, that fluorescent
lights appear to be so important, especially since they emit much
smaller amounts of 'erythemal1 ultraviolet (UV-B wavelength 280-
315 nm) than solar radiation."
Methodology:
A case-control study of 274 female melanoma patients obtained from
the melanoma clinic in Sydney Hospital. There were 213 "old" cases
(diagnosed after June 1978 in a study of the relation of oral contracep-
tives and melanoma) out of 300 potential cases and 74 "new cases" who
were 18-54 year old females diagnosed between June 1978 and December
1980.
Two controls were chosen for each case and were matched by 5-year age
groups. Old cases were also matched by area of residence. Controls
for new cases were Sydney Hospital inpatients who did not have a long
term disease.
Trained interviewers asked the subjects about demographic factors,
occupation, exposure to sunlight and other factors from a standard
questionnaire. Occupational information was not available for 13 cases
and 25 controls leaving 274 cases and 549 controls. The age range
in both groups was 18-54 years.
B-64
-------
Reference:
Beral, V., Evans, S., Shaw, E., and Milton, G. "Cutaneous factors
related to the risk of malignant melanoma." Brit. J. Dermatology 109:165-
172 (1983) .
Investigator Results:
Red hair at age five had a relative risk (RR) of 3.0 (95% C.I. 1.95-4.73).
The RR was 4.4 for both red hair and fair skin and 3.2 for red hair
and dark skin. Eighteen percent (51 women) of cases and 8% (47 women)
of controls had red hair. Graying of the hair was found to be protective
(RR = 0.60, 95% C.I. 0.46-0,83). After adjusting for hair color, fair
skin had an RR of 1.9 (95% C.I. 1.38-2.50). Fifty-nine percent of
cases and 40% of controls were faired skinned. Eye color had no indepen-
dent effect on risk. Forty-five percent of cases and 37 percent of
controls had blue eyes. An above average number of nevi increased
the risk (RR = 3.5) .
Methodology:
A case-control analysis of 287 white females who attended the Melanoma
Clinic at Sydney Hospital aged 15-84 years and 574 age-matched controls.
Controls from the general population for 213 "old cases" (diagnosed
between 1974-1978) were also matched by area of residence. Controls
for "new cases" (diagnosed between 1978-1980) were selected from hospital
inpatients (excluding those with vascular disease, gyn. disorder, diabetes,
gallbladder or breast disease, or chronic disease of 2+ years duration).
B-63
-------
References
Beral, V., Evans, S., Shaw, H. and Milton, G. "Malignant melanoma
and exposure to fluorescent lighting at work." Lancet 2:290-293 (7
August 1982) .
Investigator Results:
1. In a case-control study of 274 females with melanoma and 549 controls,
exposure to fluorescent light at work was associated with a relative
risk of 2.1 (95% C.I. 1.32-3.32) compared to the 25 women who were
never exposed to fluorescent lights. Relative risks tended to
increase with increasing years of exposure. Melanomas occurred
on the trunk among 24% of the ever exposed compared to the 9% of
others with melanoma. Although other factors were shown to increase
the risk, stratification on these factors did not change the relation-
ship to fluorescent lights. Fluorescent lights in the home were
not associated with melanoma (RR = 0.9, 95% C.I. 0.6-1.6). Previously
collected information on 27 males with melanoma and 35 controls
yielded an RR of 4.4 (95% C.I. 1.1.-17.5).
2. Among cases, a significant excess of lesions (p<.05) was observed
among those exposed to fluorescent lights (24%) compared to those
never exposed (99%) .
3. The authors commented that "It is curious, however, that fluorescent
lights appear to be so important, especially since they emit much
smaller amounts of 'erythemal1 ultraviolet (UV-B wavelength 280-
315 ran) than solar radiation."
Methodology:
A case-control study of 274 female melanoma patients obtained from
the melanoma clinic in Sydney Hospital. There were 213 "old" cases
(diagnosed after June 1978 in a study of the relation of oral contracep-
tives and melanoma) out of 300 potential cases and 74 "new cases" who
were 18-54 year old females diagnosed between June 1978 and December
1980.
Two controls were chosen for each case and were matched by 5-year age
groups. Old cases were also matched by area of residence. Controls
for new cases were Sydney Hospital inpatients who did not have a long
term disease.
Trained interviewers asked the subjects about demographic factors,
occupation, exposure to sunlight and other factors from a standard
questionnaire. Occupational information was not available for 13 cases
and 25 controls leaving 274 cases and 549 controls. The age range
in both groups was 18-54 years.
B-64
-------
Experimental Design and Analysis Issues:
The association with fluorescent lights had not been reported before
and although it is possible, it must be viewed cautiously pending the
results of additional studies.
B-65
-------
Reference:
Beral, V., Evans, S., Shaw, H. and Milton, G. "Oral Contraceptive
Use and Malignant Melanoma in Australia." Br. J. Cancer 50:681-685
(1984) .
Investigator Results:
1. Women with melanoma were more likely to have taken oral contracep-
tives for long periods of time, the relative risk (RR) associated
with 5+ years use begun 10+ years before melanoma diagnosis was
1.5 (95% confidence interval 1.03-2.14). This elevated risk per-
sisted after controlling for reported hair and skin color, frequency
of moles on body, place of birth, and measures of sunlight and
fluorescent light exposure. Results suggested that prolonged contracep-
tive use after lag of 10 or more years may increase risk of MM.
2. Cases were more likely than controls (but not significantly) to
have used hormones to regulate periods (RR=1.9), used hormonal
replacement therapy (RR=1.4), and been given hormone injections
to suppress lactation (RR=1.4).
Methodology:
A case-control analysis of 287 white females who attended the Melanoma
Clinic at Sydney Hospital aged 15-84 years and 574 age-matched controls.
Controls from the general population for 213 "old cases" (diagnosed
between 1974-1978) were also matched by area of residence. Controls
for "new cases" (diagnosed between 1978-1980) were selected from hospital
inpatients (excluding those with vascular disease, gyn. disorder, diabetes,
gallbladder or breast disease, or chronic disease of 2+ years duration).
Information obtained from interviewer questionnaire included pregnancy
history and use of oral contraceptives and other hormones.
Experimental Design and Analysis Issues:
A case-control analysis of the effect of oral contraceptive and hormone
use, as well as pregnancy history on risk of melanoma among 287 female
cases and 574 age-matched controls.
For Result 1:
A nonsignificant difference was observed between cases who had taken
oral contraceptives for 5+ years (29.3%) and controls (24.3%). A sig-
nificantly greater percentage of cases than controls had begun taking
the pill at least 10 years earlier (46.6% vs. 38.5%, p=0.05). A con-
sistently increased risk of melanoma was only observed in those who
had begun pill use 10+ years before and whose use had continued for
5+ years (RR=1.5, 95% confidence interval 1.03-2.14).
After adjusting for marital status, hair, eye and skin color, country
of birth, number of moles on body, educational status, exposure to
B-66
-------
fluorescent light, history of cholasma, extent of outdoor activity
at 10, 20, 30, and 40 years, and sunburn history, the RR varied between
1.43-1.58. There was no significant difference in site of lesion,
tumor thickness, or tumor type between those who had and had not used
the pill.
For Result 2:
The RRs and 95% confidence intervals for use of hormones to regulate
periods was 1.9 (0.85-4.12), for hormone replacement therapy 1.4 (0.78-
2.61), and for hormones to suppress lactation 1.3 (0.92-1.82).
B-67
-------
References
Beral, V., Ramcharan, S. and Paris, R. "Malignant melanoma and oral
contraceptive use among women in California. "Br. J. Cancer 36:804-809
(1977) .
Investigator Results:
1. In the Group A case-control comparison, rates of melanoma were
higher in oral contraceptive (OC) users than nonusers but the differ-
ence was not statistically significant. Only eye color (light-
colored versus brown) was a significant risk factor for melanoma.
A history of skin cancer was more common in OC users than never-
users.
2. In the Group B case-control comparison, ever versus never 'use of
oral contraceptives was 1.8 times as high in the cases as controls.
No relation with type of OC or oestrogen/progesterone content was
found.
3. Both groups combined showed an excess of lesions in OC/oestrogen
users of lower limb melanoma.
Methodology:
A case-control study conducted for two groups of cases. Group A was
based on a prospective study of 17,942 females aged 17-59 years for
whom OC use was recorded between December 1968 and February 1972.
The cases were predominantly white middle class (see Ramcharan 1974).
OC users included those who reported estrogen use. Skin cancer among
Group A cases was determined from Kaiser-Permanente records. Case
data were recorded by questionnaire. Follow-up was about 5 years per
woman. Group B cases were 37 females who were not in Group A and were
diagnosed for melanoma between 1 January 1968 and 30 June 1976. Age
was 20-59 years at diagnosis. Two controls were matched to each case
with respect to date of birth (within one year) and were chosen from
Kaiser Health Plan files. Outpatient records were searched for informa-
tion on OC and estrogen use during comparable periods for cases and
controls.
B-68
-------
References
Beral, V., and Robinson, N. "The Relationship of Malignant Melanoma,
Basal and Squamons Skin Cancers to Indoor and Outdoor Work." 3r. J.
Cancer 44:886-891 (1981).
Investigator Results:
In England and Wales from 1970-1975, office work was associated with
excess trunk and limb melanomas, whereas outdoor work in which prolonged
occupational exposure to sunlight would occur was associated with excess
head, face, and neck melanomas, as well as basal- and squamous-cell
skin carcinomas. The high rate of trunk and limb lesions in office
workers may have reflected sunbathing or other recreational habits
but contrasted with observed low nonmelanoma skin cancer rates among
indoor workers.
Methodology:
A descriptive analysis of melanoma incidence and mortality data in
England and Wales from 1970-1975 obtained from the Office of Popula-
tion Censuses and Surveys, which included information on occupation,
type of cancer, and anatomical site. On the basis of occupation, cases
were assigned to one of three groups: outdoor workers, indoor office
workers, and other indoor workers (mainly factory). Standardized cancer
registration ratios (SRR) based on age-specific rates in all occupa-
tional groups combined, and standardized mortality ratios (SMR) based
on 1971 population by occupational group were calculated by indirect
standard!zation.
Experimental Design and Analysis Issues:
A descriptive analysis of incidence of basal- and squamous-cell skin
cancers and melanoma including calculation of SRRs and SMRs.
Based on SRRs for males 15-64 years, outdoor work was associated with
a 10% excess of squamous- and basal-cell carcinoma, a 9% excess of
head, face, and neck melanomas, and a 22% deficit of melanomas at other
sites compared to the national 1971 averages. Office work was associ-
ated with a 31% melanoma excess of "other" sites. Other indoor workers
had a deficit of all tumor types. When reanalyzed only for social
class III the findings were similar except for moderate increases of
squamous- and basal-cell carcinomas and head, face and neck melanomas
among office workers.
B-69
-------
References
Blair, A. and Hayes, H.M., Jr. "Mortality Patterns Among U.S. Veterina-
rians, 1947-1977: An Expanded Study." Int. J. Epi. 11:391-397 (1982).
Investigator Results:
Proportions of death among 5,016 white male veterinarians were sig-
nificantly elevated for cancers of the skin as compared to a distri-
bution based on the general U.S. population. Although economic and
methodological factors may have been involved, the pattern suggested
that sunlight exposure was responsible for the excess among vets whose
practices were not exclusively limited to small animals.
Methodology:
A descriptive analysis of mortality data for 5,016 white male veteri-
narians who died between 1947-1977 obtained from the Journal of the
American Veterinary Medical Association. Each case was grouped accord-
ing to reported professional specialty into one of two large categories:
practitioners (2,846) and non-practitioners (2,170), and into one of
five subcategories: small-animal practitioners (331), other practi-
tioners (2,515), meat inspectors (178), lab specialists (110), and
regulatory vets (713). Expected numbers of deaths were calculated
based on 5-year age and calendar period groups for white U.S. males;
proportionate mortality ratios (PMRs) were calculated and tested by
a chi-square test.
Experimental Design and Analysis Issues:
A descriptive analysis of vet mortality data by professional specialty.
The observed number of deaths due to skin cancers was 24 compared to
14.9 expected deaths (p<0.025). Among the 24 skin cancers, 18 were
malignant melanomas. Most of the skin cancers occurred in the other
than small animal practitioner group for which the number of observed
cases was not significantly different than expected (12 vs. 7.5).
The seven observed skin cancers among more recent vets (began practice
between 1908-1957 and ended between 1965-1974) who died at or before
64 years of age was significantly higher than the 2.1 expected p<0.005).
B-70
-------
References
Boyle, P., Day, N.E., and Magnus, K. "Mathematical Modelling of Malignant
Melonoraa Trends in Norway, 1953-1978." Am. J. Epi. 118:887-896 (1983).
Investigator Results:
Malignant melanoma (MM) time trends in Norwegian Cancer Registry data
from 1953-1978 (by subsite and sex) could be explained by a common
age effect and a separate birth cohort effect for each subsite in both
sexes, or by a common cohort effect but different age effects for each
subsite and sex. In neither case was the time period of registration
an important factor.
Methodology:
A descriptive analysis of the fit of various melanoma risk models to
data on 5,741 MM cases reported in the Norwegian Cancer Registry between
1953-1978 and classified by sex, year of birth, year of diagnosis and
subsite. The time period of registration was divided into six periods:
1953-1954, 1955-1959, 1960-1964, 1965-1969, 1970-1974, and 1975-1978.
Nine 10-year birth cohorts (1870-1879 to 1950-1959) and 7, 9-year age
categories (10-19 to 70-79) were considered. Ninety-nine percent of
cases in the registry had histological verification. Cases of lentigo
maligna melanoma were excluded. The model used estimated melanoma
risk as a function of sex, age, years of birth and diagnosis, and subsite,
factors included as single and interaction terms in the models.
Experimental Design and Analysis Issues:
A descriptive analysis of the fit of various age, birth cohort, and
time period dependent melanoma risk models to Norwegian melanoma data
from 1953-1978.
Among four additive models (age, age with cohort, age with time trend,
and age with both), the fit to the data by sex and subsite did not
vary by much except for male trunk tumors for which a cohort effect
gave superior fit. Two additional models including interaction terms
for age with cohort and age with time indicated an improved fit for
the age-time model over the additive age and time model.
Among a series of models variously including terms for age, sex, subsite,
time trend, and cohort (but not cohort and time trend simultaneously),
two models provided alternative but equally good pictures of the data:
1) a model with a single cohort effect for all subsites and both sexes,
but varying by an age-sex-subsite interaction factor; and 2) a model
with an age effect for each sex-subsite-cohort group.
All subsites in both sexes showed rapidly increasing risk by year of
birth cohort, especially for the trunk and lower limbs. Incidence
of tumors in both sexes of head, neck or other sites rose approximately
linearly on the log-log scale with age, i.e., an exponential rate.
B-71
-------
For trunk and lower limb melanomas, incidence increases fell off in
older age groups, especially for lower limbs in females for whom incidence
remained constant after 40 years.
B-72
-------
Reference:
Boyle, J., MacKie, R.M., Briggs, J.D., Junor, B.J.R., and Aitchison,
T.C. "Cancer, Warts, and Sunshine in Renal Transplant Patients."
Lancet, 1:702-705 (1984).
Investigator Results:
Melanoma was not addressed in this study. Among 94 renal transplant
patients and 94 age, sex, and sun exposure matched control, 17 patients
with high exposure to sunshine had 2 squamous cell carcinomas, 3 basal
cell carcinomas, and 7 actinic keratoses, lesions which did not appear
in the other patients or the controls. The immunosuppressive effect
of UV radiation (290-320 nm) may be related to increased incidence
of cutaneous malignancy, actinic keratoses, and warts among renal trans-
plant patients (already immunosuppressed by drugs).
Methodology:
A case-control study of 94 renal transplant patients (62M, 32F) from
the Western Infirmary, Glasgow and 94 sex-, age-, and sun exposure-
matched controls treated at the infirmary's emergency department.
Cases and controls were questioned and grouped according to history
of sun exposure before reference transplantation date (more than 3
months in tropical or sub-tropical climate or worked outdoors for 5+ years
were considered high exposure). Details of previous herpes simplex
and zoster infections and presence of warts, fungal infections, and
malignant and premalignant cutaneous lesions were also noted. Data
were analyzed by McNemar and chi-square tests.
Experimental Design and Analysis Issues:
Seventeen patients and matched controls had history of high sun ex-
posure, and among these 7 patients had actinic keratosis, 3 had basal
cell carcinoma, and 2 had squamous cell carcinoma; no controls had
these lesions.
B-73
-------
References
Brown, J., Kopf, A.W., Rigel, D.S., and Friedman, R.J. "Malignant
Melanoma in World War II Veterans." Int. J. Dermatol. 23(10):661-663
(1984) .
Investigator Results:
1. In a case-control study of 89 World War II (WWII) veterans with
melanoma and 65 age-matched controls, 83% of the melanoma group
compared to 76% of the control group had served in armed forces
during WWII; moreover, a significantly greater percentage (p=0.0002)
of the cases (34%) served in tropics than did controls (6%).
2. A greater percentage of melanoma patients among those who had served
in the tropics had malignant melanomas (MM) originating in nevocytic
nevi compared to melanoma patients who had served in nontropical
areas.
3. The authors suggested that Caucasians heavily exposed to sunlight
in the tropics for several years during early life might be at
higher risk of developing MM. They also suggested a two-step pheno-
menon: first step solar induction of nevocytic nevi, second malignant
transformation within them.
Methodology i
A case-control analysis of 89 (out of 120) patients entered into the
Melanoma Cooperative Group at New York University School of Medicine
during 1972-1980 who were 18-31 years-old during WWII (out of 1,067
consecutive patients total) and 65 age-matched controls who visited
NYU Department of Dermatology for cutaneous ailments other than MM.
Questionnaire information determined whether cases and controls had
served in WWII and if so, in which theater(s) of operation.
Experimental Designs and Analysis Issues:
A case-control analysis of 89 male melanoma patients and 65 age-matched
controls.
For Result 1:
A greater percentage of cases (83%) had served in Armed Forces during
WWII than controls (76%) and a significantly greater percentage had
served in tropics (34% cases vs. 6% controls, p=0.0002).
For Result 2:
Among the 89 melanoma patients, those who had served in the tropics
had their tumors arise more often in nevocytic nevi (53%) than those
who had served in U.S./Europe (24%). There were no significant dif-
ferences between these two groups in thickness, level of penetration,
histologic type, or anatomic site.
B-74
-------
References
Crorabie, I.K. "Distribution of malignant melanoma on the body surface."
Br. J. Cancer 43:842-849 (1981).
Investigator Results:
1. From a review of 37 cancer registries which published detailed
site-incidence data on malignant melanoma among whites, a higher
incidence on the female lower limb and the male trunk was found.
For all 37 registries, the proportion of melanoma by site in the
order head, upper limb, lower limb, and the remainder (mostly trunk)
for males was 24.4, 14.4, 18.4, and 42.8% and the for females was
16.1, 15.4, 42.4, and 26.1%. The melanocyte density was quoted
as being 13, 31, 19, and 37% in the same order.
2. Females had a significantly higher incidence of melanoma than males.
Differences in the rates by sex were statistically significant
for all sites combined for all 37 registries combined, for the
15 North American registries, but not for the 20 European registries.
By body site all differences in incidence rates were significant
except for the upper limb in the 15 North American registries.
3. The median incidence per 100,000 and the male to female sex ratio
by site were as follows: head 0.526 and 0.425 giving 1.24; upper
limb 0.261 and 0.528 giving 0.494; lower limb 0.378 and 1.079 giving
0.350; and remainder (mostly trunk) 0.927 and 0.674 giving 1.37.
Methodology:
The distribution of melanoma on the body surface was examined based
on data from Cancer Incidence in Five Continents using a four digit
ICD classification.
B-75
-------
Reference:
Crombie, I.K. "Variation of melanoma incidence with latitude in North
America and Europe." Br. J. Cancer 40:774-781 (1979).
Investigator Results:
In North America and in England, melanoma incidence increased with
decreasing latitude; in Europe the relationship was in the opposite
direction.
A regression of cancer incidence rates for 16 North American registries
and for 14 regions in England (1967-1973) on latitude showed a statisti-
cally significant relationship for melanoma but not for all sites,
for both male and females. In 27 European registries, the relationship
was reversed with a significant increase in melanoma as the latitude
was increased for males and females and a significant increase in all
sites for females but not males.
Methodology:
The relationship between melanoma incidence rates and latitude was
studied for 43 population-based cancer registries in North America
and Europe. From Cancer Incidence in Five Continents, data on rates
between 1967 and 1973 from 43 registries in Europe and North America
which record cancer incidence among white populations were studied.
Also 14 hospital regions in England were studied.
Comments:
The effect of the very high incidence in Norway and Sweden on this
regression needs to be investigated. It is not known if the regression
would still be significant if Norway and Sweden were omitted.
B-76
-------
Reference:
Fears, T.R., Scotto, J., and Schneiderman, M.A. "Skin Cancer, Melanoma,
and Sunlight." AJPH 66:461-464 (1976).
Investigator Results:
The authors applied a modeling approach which estimated increases in
nonmelanoma and melanoma incidence and mortality with increases in
ultraviolet (UV) radiation. The results indicated that total UV dose
increased with decreasing latitude, latitude decreased with incidence
increases, and latitude decreased with increasing mortality. The model
predicted an additional 49.7/10 cases of nonmelanoma cancer, 0.69/10
cases of melanoma, and 0.13 additional deaths from melanoma associated
with a 10% increase in radiation dose.
Methodology:
A cross-sectional analysis of the fit of melanoma and nonmelanoma inci-
dence and mortality rate data to a single-variable (latitude) linear
regression model.
Incidence rate data, obtained from the Third National Cancer Survey
(TNCS) for the 1969-1971 period included nonmelanoma skin cancer data
for four U.S. regions and melanoma data for nine U.S. regions. Regions
were classified by latitude. Age-adjusted melanoma mortality rates
were obtained from Mason and McKay. The data were fit to a single-
variable (latitude) linear regression model: log R= x+BL where R=age-
adjusted rate and L=latitude.
Experimental Design and Analysis Issues:
A cross-sectional linear regression analysis investigating the relation-
ship of latitude to melanoma incidence and mortality rate data and
nonmelanoma incidence data.
The regression coefficient (slope) for melanoma incidence (9 data points)
was significant ( B=-0.03 males, p less than 0.01) whereas for nonmela-
noma incidence (4 data points) the negative slope was steeper but not
significant. For melanoma mortality data, the less steep slope (B=-0.017
males) was not significantly different from that for melanoma incidence.
Risk of developing melanoma doubled with every 9.8 (males) and 10.7
(females) decrease in degrees latitude.
For 33 northern hemisphere locations, latitude was converted into Bio-
logic Effective Units (BEU) for 12 months representing the sum of UV
radiation from 295 nm-325 run weighted by erythema effectiveness. Cor-
relation between degrees north latitude and BEU was 0.97 (p less than
0.005).
The effects of 10%-30% increases in radiation dose on incidence and
mortality, based on 650, 850, and 1,050 BEUS (for North Dakota, Iowa,
B-77
-------
and Oklahoma latitudes, respectively), indicated increases for a 10%
BEU increase of 49.7/10 cases of nonmelanoma, 0.69/10 cases of mela-
nmona, and 0.13/10 deaths from melanoma. The increases were larger
for nonmelanoma than melanoma, men than women, and incidence than mortality.
B-78
-------
Reference:
Hinds, M.W. "Anatomic Distribution of Malignant Melanoma of the Skin
Among Non-Caucasians in Hawaii." Br. J. Cancer 40:497-499 (1979).
Investigator Results:
Among 64 non-Caucasian malignant melanoma cases in the Hawaii Tumor
Registry during 1960-1977, predominant lesion sites were the feet in
males (n=41, 41.5%) and in females (n=18 27.8%).
Methodology:
A descriptive analysis of all 64 non-Caucasian invasive malignant mel-
anoma (MM) cases reported in the Hawaii Tumor Registry during 1960-1977.
Over 94% of all cancer cases in the registry had been histologically
confirmed.
Experimental Design and Analysis Issues:
In addition to the information cited above, site distributions were
summarized for several ethnic subgroups. Among the 20 Japanese cases,
30% occurred on the feet. Among the 9 Filipinos, 78% on feet and among
the 5 Chinese, 40% on feet. Among the 16 Hawaiians and part-Hawaiians
(4 part-Caucasians), 25% of lesions were on feet. Among the 14 mixed
racial background cases (2 part-Caucasian), 21% on feet. Of the 22
foot lesions, 12 were on plantar surface and 7 on toe.
B-79
-------
Reference:
Houghton, A.N./ Flannery, J., and Viola, M.V. "Malignant Melanoma
of the Skin Occurring During Pregnancy." Am. J. Cancer 25:95-103 (1980).
Investigator Results:
1. In a non-matched case-control study of 12 patients with melanoma
during pregnancy (cases) and 175 nonpregnant melanoma patients
(controls)/ 3- and 5-year survival rates were significantly lower
for cases (65% and 55%, respectively) than controls (86% and 83%,
respectively, p less than 0.05). When controls were matched to
cases by age, anatomic site, and stage at diagnosis, 3- and 5-year
survival rates were not significantly different.
2. Melanoma occurred more often on the trunk and at more advanced
stages in cases than in controls.
3. MM incidence did not substantially increase during pregnancy when
compared to the expected number of melanoma cases among pregnant
women.
4. Melanoma occurring during pregnancy usually carried poor prognosis
but once diagnosed, disease course was not worse considering stage
of disease and primary site.
Methodology:
A case-control study based on review of female MM patients between
15-40 years of age reported in the Connecticut Tumor Registry from
1950-1954, 1960-1964, and 1970-1974. Patients diagnosed with MM dur-
ing pregnancy were selected as cases (n = 12) and 175 female patients
diagnosed during the same calendar periods not during pregnancy were
selected as controls. Each case was matched with 2 controls by age,
anatomic site, and stage of disease at diagnosis. Number of expected
live births per year for female melanoma patients was calculated from
Connecticut live birth rates. Survival rates were compared by 2-sided
Fisher's exact and logrank tests.
Experimental Design and Analysis Issues:
A case-control study of survival rates among female MM patients pregnant
during diagnosis (cases) vs. nonpregnant (controls).
For Result 1:
Among cases, 50% died of melanoma compared to 14.8% among controls.
Three- and five-year survival rates were higher among all 175 con-
trols (86% and 83%, respectively) compared to the 12 cases (65% and
55%, respectively, p less than 0.05). However, 3- and 5-year survival
rates did not differ significantly when compared for 12 cases and
24 matched controls.
B-80
-------
For Result 2:
A comparison of the 12 pregnant cases and 175 nonpregnant controls
indicated occurrence of melanomas among cases more often on trunk
(50% vs. 32% controls), and less often on lower extremities (33%
vs. 43% controls). Regional and distant metastases were more common
among cases than controls (16.7% vs. 9.7% for regional and 16.7%
vs. 2.3% for distancts).
For Result 3:
The expected number of pregnant women among the 187 MM patients reviewed,
estimated from Connecticut live birth rates, was 13.3, compared to the
observed 12 pregnancies, suggesting MM incidence did not substantially
increase during pregnancy.
B-81
-------
Reference:
Houghton, A.N., Munster, E.W., and Viola, M.V. "Increased Incidence
of Malignant Melanoma After Peaks of Sunspot Activity." Lancet April:759-
760 (1978) .
Investigator Results:
1. Age-adjusted malignant melanoma (MM) incidence rates in Connecticut
(CT) rose from 1.1/105 in 1935 to 6.2/105 in 1975. Superimposed
on linear incidence increases were 3-5 year periods of cyclic inci-
dence rises occurring every 8-11 years.
2. Significant partial correlations (p less than 0.01) were observed,
after adjustment for time, between CT melanoma incidence and sunspot
numbers from 1 to 3 years following sunspots (closest association
for 2 years with r = 0.695 (males and females) and r = 0.717 (males)).
Methodology:
A descriptive and cross-sectional anaysis of 2,983 histologically con-
firmed incident melanoma cases registered in the CT Tumor Registry
from 1935-1974. Relative sunspot numbers were obtained from Waldmeir
and Eddy.
Experimental Design and Analysis Issues:
A descriptive analysis of melanoma incidence rates in CT and a cross-
sectional analysis of the association between melanoma incidence and
sunspot activity.
For Result 1:
Age-adjusted incidence rate was 1.1/105 (1935) and 6.2/105 (1975).
Superimposed on the 40-year steady incidence increases were 3-5 year
periods of more acute incidence. Between 1935 and 1975 this cycle
occurred 4 times in 8-11 year intervals.
For Result 2:
The correlation using a linear regression equation between melanoma
incidence and time over three sunspot cycles (33 years) was signifi-
cant (r = 0.9327, p less than 0.01). Examinations of deviations
from a time-adjusted regression model indicated cyclic correlations
between sunspot cycles and MM incidence rates. Partial correlations
between annual sunspot numbers and melanoma incidence, controlling
for time effect, found statistically significant correlations in
each of the subsequent 3 years, the closest association being 2 years
later. Correlation coefficients with p less than 0.01 were observed
for males and females 1 and 2 years following sunspots and for males
1, 2, and 3 years following sunspots.
B-82
-------
Data for New York, Norway, and Finland cancer registries for 1950-1971,
similarly analyzed, indicated significant partial correlations (p
less than 0.01) 1-2 years following sunspot activity for New York.
Nonsignificant correlations were observed for Norway 0-4 years subsequent
to sunspots. In Finland, a significant partial correlation (p less
than 0.05) was observed for the year of sunspot activity.
B-83
-------
Reference:
Houghton, A.N., and Viola, M.V. "Solar Radiation and Malignant Melanoma
of the Skin." J. Am. Acad. Dermatol. 5:477-483 (1981).
Investigator Results:
A review of clinical and epidemiologic evidence supporting the role
of solar (UV) radiation in pathogenesis of malignant melanoma (MM).
Discussion includes apparent correlation between sunspot cycles, UV
maxima, and increased MM incidence, and apparent promoter effect of
UV radiation in melanoma pathogenesis.
Methodology:
A general review of epidemiological and clinical findings supporting
role of solar (UV) radiation in pathogenesis of MM.
Experimental Design and Analysis Issues:
The evidence for a role of sun exposure in MM comes from epidemiological
and clinical observations indicating that MM incidence among whites is
inversely proportional to latitude, and ultraviolet radiation is more
intense at lower latitudes; MM tends to occur on more heavily exposed
sites during recreation such as trunk in men and lower legs in women;
and MM is more likely to develop in lightly pigmented persons. Other
less compelling observations include: susceptible persons living in
sunny climate for their lifetimes have higher MM incidence than recent
migrants from less sunny areas; MM rates are relatively low for those
with outdoor occupations and higher for higher income groups, suggest-
ing recreational but not chronic sun exposure is one pathogenic factor.
The correlation between sunspot cycles and MM incidence increases (in
Connecticut, Denmark, Finland, and New York) further relates solar
radiation to MM. Similar fluctuations were not observed for non-melanoma
malignancies. Sunspot numbers are clearest indicator of solar activity
and amount of UV radiation reaching the Earth's surface. Rises in MM
incidence occur 0-3 years after sunspot peaks suggesting importance
of heavier exposures.
For most non-melanoma skin cancers (e.g., basal- and squamous-cell
skin carcinomas), solar radiation behaves as dose-dependent (initiator)
carcinogen usually on most chronically exposed body parts. Excluding
lentigo maligna melanoma, solar radiation does not act as a dose-depen-
dent carcinogen in most melanoma cases, but appears to act proximally
in melanoma pathogenesis usually on intermittently and intensively
exposed body parts (e.g., during recreation). Evidence suggests that
UV radiation may act as promoter in melanoma pathogenesis.
Other potentially important factors in melanoma pathogenesis may include
genetic susceptibility, recreational behavior, cultural and socioeconomic
factors, altitude, cloud cover, and time of day during exposure.
B-84
-------
Reference:
Kripke, M.L. "Speculations on the role of ultraviolet radiation in
the development of malignant melanoma." (guest editorial) JNCI 63:541-548
(1979).
Investigator Results:
1. In this guest editorial, the author stated that the incidence of
and mortality from melanoma were doubling every 10-17 years.
2. The author discussed animal models for melanoma. A possible mechanism
of melanoma is one in which UV lights "affect tumor growth but
not tumor production." Another possibility was that UV produced "a
systemic alteration that was conducive to tumor growth. ...This
systemic alteration could be immunologic, like the one we have
described for nonmelonoma skin tumors in mice; alternatively, it
could be a biochemical alteration in which skin photoproducts provide
a nutritive advantage for proliferating tumor cells."
3. The authors stated that it was "not unreasonable to suppose that
at least some human melanomas might also arise for reasons that
do not include UV light. Human melanomas might have in common
only the cell affected in the neoplastic process and need not share
an etiology in every case."
B-85
-------
Reference:
Lew, R.A., Sober, A.J., Cook, N., Marvell, R., and Fitzpatrick, T.B.
"Sun exposure habits in patients with cutaneous melanoma: a case-control
study." J. Dermatol. Surg. Oncol. 9:981-986 (1983).
Investigator Results:
1. Telephone interviews to 111 patients and 107 controls revealed
that painful or blistering sunburns during either childhood or
adolescence were associated with increased risk of developing cutaneous
melanoma.
Risk factors were blistering sunburns during adolescence (odds
ratio (OR) = 2.05, 95%, C.I. 1.18-3.56), poor vs. good tanning
ability (OR = 1.93, 95% C.I. 1.13-3.3), and 30 days of vacation
in sunny warm places during childhood vs. fewer days (OR = 2.5,
95% C.I. 1.18-5.8). The same set of risk factors was found for
those less than 50 years old and those 50 and over. Among those
with no painful sunburn, cases were more frequently encouraged
to be outdoors (OR = 3.32, 95% C.I. 1.2-5.7).
2. No differences were found between cases and controls regarding
use of hormones and oral contraceptives.
3. The authors commented that "Sunlight appears to play a role in
the etiology of melanoma....the present study suggests that traumatic
dosage may outweigh lifetime cumulative dosage as a factor."
Methodology:
A case-control study of 111 patients (cases) in the Melanoma Clinical
Cooperative Group of the Massachusetts General Hospital in Boston.
Controls were friends of the patients (age within 5 years) and of the
same sex. Matching was not used because 46 patients provided no controls,
30 provided 1, 28 provided 2, and 7 provided 3. Telephone interviews
were conducted to determine the history of sun exposure and related
behavior for childhood into the adult years.
B-86
-------
Reference:
MacKie, R.M. and Aitchinson, T. "Severe Sunburn and Subsequent Risk
of Primary Cutaneous Malignant Melanoma in Scotland." Br. J. Cancer
46:955-960 (1982)
Investigator Results:
1. Melanoma patients had increased incidence of severe sunburn, signifi-
cantly less recreational sun exposure, and were of higher social
class than non-melanoma controls. Among males alone, social class,
occupational sun exposure and severe sunburn were significantly
different from controls, while recreational sun exposure was not
significantly different. Among females, only severe sunburn differed
significantly between patients and controls. Skin type was not
a significant factor.
2. Individuals with histories of multiplicative severe sunburns were
2.8 times more likely to have melanoma (95% relative risk confidence
interval 1.1-7.4) than individuals without histories of severe
sunburns.
3. The authors commented that "Questions about severe sunburn were
confined to the 5-year period before the development of the primary
tumor, as it was felt that distant memory might well be inaccurate.
It is likely, however, that patients with a tendency to severely
sunburn will have had more than one such episode in their lifetime
and this was in many cases confirmed by the patients." At the
end of the interview, only 24 percent of cases reported that they
thought sunlight exposure and/or sunburn might be related to their
disease thus in the minds of the authors eliminating concern with
bias in response.
4. The authors concluded that "This study thus provides evidence to
suggest that short intense episodes of UV exposure resulting in
burning may be one of the aetiological factors involved in subsequent
development of melanoma."
Methodology:
A case-control study of 113 patients with superficial spreading melanoma
(SSM) and nodular melanoma (NM) in the west of Scotland between 1978
and 1980 and 113 age- and sex-matched control patients. Patients with
lentigo malignant melanoma were excluded. Interviewer questionnaire
included information on skin, eye, and hair color, skin responses to
sunlight exposure, total hours of occupational and recreational sun
exposure in winter and summer, weeks in warm climate and history of
severe and prolonged sunburn (in 5-year period before development of
primary tumor), and social class. Thirty-three percent of the cases
and 27 percent of the controls had never left the UK.
B-87
-------
Experimental Design and Analysis Issues:
The matched case-control data was analyzed using conditional multiple
logistic regression, which yielded estimates of relative risk.
For Result 1:
In the male group, three factors were significant at the 5% level
in predicting increased melanoma risk—social class (negative associa-
tion) , occupational sun exposure (negative asociation), and severe
sunburn (positive association). Among females, only one factor,
severe sunburn, was significant at the 5% level. For the combined
data set, social class (negative association), recreational sun exposure
(negative association), and severe sunburn (positive asociation)
were significant predictors of excess melanoma risk. Skin type was
not a significant factor in the group as a whole or in males' and
females separately. The number of holidays and days spent in summer
climates were similar between cases and controls. Overall, 56% of
the melanoma patients had histories of severe burning (vs. 22% of
controls). A history of severe burning was given by 26 (50 percent)
of male cases and 12 (23 percent) of controls; and by 37 (61 percent)
of female cases and 12 (20 percent) of controls.
For Results 2:
The relative risk of melanoma for an individual with a history of
severe sunburn (in the 5-year period before develoment of primary
tumor) was 2.8 with a 95% confidence interval of 1.1-7.4 (Exp(B.)=2.8) .
"This may be associated with enhanced photosensitivity which is not
correlated to skin type."
B-88
-------
Reference:
Morton, W., and Starr, G.F. "Epideraiologic Clues to the Cause of Melanoma."
West. J. Med. 131:263-269 (1979).
Investigator Results:
1. Risk of melanoma was highest in a moist, flat residential area
in Oregon's Lane County urban portion and in its agricultural area.
Overall melanoma risk was higher in urban areas. The incidence
pattern strongly suggested local cycles of etiologic agents in
subcounty units.
2. An apparent widespread rural epidemic was identified beginning
in 1965 and lasting several years.
Methodology:
A descriptive analysis of 146 new melanoma cases from 1958-1972 ob-
tained from pathology files representing all six Lane County hospitals,
coastal portions of Lane County, the Veterans Administration Hospital
tumor registry in Portland and the University of Oregon Medical School
tumor registry in Portland. Ocular melanomas were added to the data.
Mortality from cutaneous or ocular melanoma from 1958-1972 was also
reviewed for Lane County residents.
Census tracts, to which each case was assigned, were grouped into geo-
graphic regions or socioeconomic strata. Incidence rates were directly
age-standardized to the 1970 U.S. population.
Experimental Design and Analysis Issues:
A descriptive analysis of melanoma incidence and mortality data.
For Result 1:
5 5
Lane County incidence rate (5.5/10 /year) and mortality rate (1.4/10 /year)
increased from 1958-1972 (except for male mortality from 1968-1972).
Male mortality was consistently greater than female due mostly to high
death rates among male 75-84 years-olds. Total incidence for 1968-1972
exceeded Third National Cancer Survey incidence for 1961-1971 by 67.4%
(males) and 54.5%(females). Age-distributions for incidence show maximum
incidence rates for 75-84 year-olds (males) and 65-74 (females) with
secondary peaks from 35-44.
Geographic analysis by census tract of incidence and mortality rates
indicated that place of residence influenced melanoma risk. "Very
high" incidence rates (10.5/10 +) occurred in urban areas of north
and northwest Eugene, whereas "high" rates (7.0/10 to 10.4/10 ) oc-
curred in flat, moist bottom land adjacent to "very high" risk areas
in urban tracts (with history of vigorous insect control measures)
and in the rural agricultural area. Effect of socioeconomic status
B-89
-------
on melanoma distribution was weaker and more inconsistent than geo-
graphic pattern.
For Result 2:
An analysis of rates for four groups (1958-1961, 1962-1964, 1965-1968,
and 1969-1972) by region (rural with 5 subdivisions and urban with
7 subdivisions) indicated the onset of a rural epidemic in 1965, two
similar incidence peaks (decade apart) in rural north, and a consis-
tently greater incidence in north Eugene than in county as a whole.
Otherwise, incidence data displayed significant local deviations from
respective county-wide rates. Investigation of potential seasonal
patterns of occurrence (by diagnosis date) showed some grouping of
cases during summer months from 1958-1963 but a general year-round
pattern in subsequent years. There were differences in occurrence
by sex but no seasonal pattern for either.
B-90
-------
References
Moss, A.L.H. "Malignant Melanoma in Maoris and Polynesians in New
Zealand." Br. J. Plast. Surg. 37:73-75 (1984)
Investigator Results:
1. The lower limb was the predominant site for primary malignant melanoma
(MM) tumors (13 out of 24 Maori and Polynesian MM patients), with
almost half of these occurring in the sole. Half of the six head
and neck MM primary tumors occurred in the oral mucosa.
2. Based on the medical history of 24 patients, MM in Maoris and Polyne-
sians appeared to have poor prognoses.
Methodology:
Collection over a 19-year period from the National Cancer Registry
of the National Health Statistics Centre, the New Zealand Health Statistics,
and New Zealand Hospital Boards of information on 21 Maoris and Polynesians
with 24 primary MMs. Histologic material, where available, was reviewed
and autopsy reports were used to confirm the diagnoses on the other
patients. Patient information included race, anatomical distribution,
stage at presentation, depth of tumor invasion, and follow-up mortality.
Experimental Design and Analysis Issues:
A simple descriptive summary of a small, probably non-randomly selected
populaton of 24 Maoris and Polynesians presenting with MM.
For Result 1:
The predominant primary tumor site was the lower limb (13 of 24 patients)
for which the sole was most common (6 of 13). Three of the 6 head
and neck tumors were of the oral cavity. There were 11 nodular and
5 superficial spreading MMs, and 8 unclassified lesions.
For Result 2:
Only 2 patients survived longer than five years. One patient was
lost to follow-up at 6 years. Of the patients known to have died,
13 died of their malignancy within 52 months. It was concluded that
Maoris and Polynesians, perhaps because of their general reluctance
to seek medical advice, have poor prognoses for MM.
B-91
-------
Reference:
Malec, E., and Eklund, G. "The Changing Incidence of Malignant Melanoma
of the Skin in Sweden, 1959-1968." Scand. J. Plast. Reconstr. Surg. 12:19-
27 (1978).
Investigator Results:
1. The age- and sex-adjusted incidence of malignant melanoma (MM)
of the skin increased from 1959 to 1968 by 7% per year, with most
pronounced increases for the trunk and legs in both sexes.
2. The ages for which incidence increases were greatest were 30-39
for females and 40-64 for males.
3. The incidence in females on the legs was higher during the summer
months.
4. Negligible changes in MM mortality were observed from 1959 to 1968.
Methodology:
A descriptive analysis of 3,289 MM patients (1,534 M, 1,755 F) registered
in the Swedish Cancer Registry between 1959 and 1968. Notification
of all cancer cases to the Registry is required under a special decree.
All cases were histologically confirmed. The clinical reports provided
information on primary tumor site, metastases, and sex. Additional
information on 1,409 patients (808 M, 601 F) whose deaths were from
melanoma was obtained from the Swedish Bureau of Statistics.
Experimental Design and Analysis Issues:
A straightforward, descriptive analysis of MM incidence rates between
1959 and 1968.
For Result 1:
The annual MM age- and sex-adjusted incidence rate (per 100,000)
rose by 7.0% for both males and females between 1959 and 1968. The
rise in incidence rate was highly significant (r = 0.94 for males,
r = 0.90 for females). The rate was higher for women. The trunk
was the most frequent site in males and the leg in females. The
correlation between incidence rate and year was highly significant
for the trunk and head and neck in males and the trunk in females
(p less than 0.001) and moderately significant for the legs in females
(p less than 0.01) .
For Result 2:
The increase in incidence with age was significant in many age groups:
it was highly significant in males between 50 and 54, 55 and 59,
B-92
-------
and 60 and 64 and highly significant in females between 30 and 34
and 50 and 54. The most pronounced increases were for 40-64 year-old
males and 30-39 year-old females.
For Result 3:
Based on month of tumor diagnosis information, a highly significant
increase in MM incidence on the lower extremity and totally for females
was observed when summer months (June-August) were compared to winter
months (December-February). For leg melanoma, the seasonal difference
was highly significant (p less than 0.001). No seasonal increase
was observed in males.
For Result 4:
Age-standard!zed mortality rates for MM were fairly constant for
both sexes and were substantially lower than corresponding incidence
rates.
B-93
-------
Reference:
Magnus, K. "Incidence of Malignant Melanoma of the Skin in the Five
Nordic Countries: Significance of Solar Radiation." Int. J. Cancer
20:477-485 (1977).
Investigator Results:
1. A descriptive comparison of malignant melanoma (MM) incidence in
five Nordic countries showed similar incidence rates in Denmark,
Norway and Sweden, and lower incidence rates in Finland and Iceland.
2. Increasing incidence rates were clearly observed for successive
birth cohorts, with a doubling of the rate occurring in a period
of 10-17 years.
3. The anatomic distribution of MM tumors differed between males and
females: the predominant tumor site was the neck/trunk for males
and the lower limb for females. Increasing MM incidence was observed
for the neck/trunk among males and the lower limbs among females.
Only a slight increase was seen in MM of the face. Authors concluded
that there had been a real increase in MM incidence in all Nordic
countries which was in accordance with the hypothesized association
between solar radiation and risk of MM.
Methodology:
Analysis of time trends in MM incidence data based on 13,101 cases
recorded in cancer registries of Denmark, Finland, and Norway (1957-197_) ,
Iceland (1955-1974), and Sweden (1959-1971). Lentigo maligna was excluded
except in Sweden's data. Tumors were classified by anatomical site.
Indirect age-standard!zed incidence rates (based on age-specific rates
pooled for all countries) were calculated.
Experimental Design and Analysis Issues:
A descriptive analysis of MM incidence rates in five Nordic countries.
For Result 1:
Annual age-adjusted MM incidence rates (both sexes) were similar in
Denmark, Norway, and Sweden (4.5-5.0 per 10 ) from late 1950's to early
1970's. The rate in Finland was 3.4/10 and in Iceland about 2/10 .
Only in Denmark and Iceland was there a marked sex difference (female
incidence exceeded male incidence).
For Result 2:
The time trend of incidence in Iceland was not statistically significant.
For the four other countries, annual percentage increases ranged from
4.1% to 7.0%, corresponding to a doubling of the MM incidence rate in
17 and 10 years, respectively. There was no significant difference in
B-94
-------
rates by sex. Incidence rates generally increased steeply in adolescence,
levelled off in middle age, and rose again in later years. This pat-
tern was most evident in high incidence countries and for females.
For Result 3:
When grouped by tumor site, differences were observed between sexes.
Excess incidence was observed for the neck/trunk in males, and the
lower and upper limbs in females. There was no systematic difference
for facial tumors. The distribution of cases by sex and site is simi-
lar for the five countries, with neck/trunk and extremities accounting
for about 75% of all cases, and the face 15-20%. A slight or unnotic-
able increase in facial MM was observed in all countries except Sweden
where incidence increased (possibly due to classification of neck tumors
as face tumors in the data set for Sweden). For neck/trunk tumors
there was a consistent increase, more prominent for males. A-doubling
of incidence rates was observed over a 7-9 year period for male neck/trunk
tumors and female lower limb tumors (Denmark, Norway, and Sweden).
B-95
-------
Reference:
Magnus, K. "Habits of Sun Exposure and Risk of Malignant Melanoma:
An Analysis of Incidence Rates in Norway 1955-1977 by Cohort, Sex,
Age, and Primary Tumor Site." Cancer 48:2329-2335 (1981).
Investigator Results:
1. Malignant melanoma (MM) incidence increased in Norway from 1955-
1977 by approximately 7% per year for both sexes. Incidence in-
creases and birth cohort variations were greater for trunk and
lower limb melanomas than for face and neck melanomas.
2. The shape of the age-specific incidence rate curves for cohorts
also differed for these sites, indicating that carcinogenic expo-
sure through life differed for the face-neck and the trunk-lower
limb. For generations born 1930-1949, MM incidence per unit skin
area was greater for trunk-lower limb than for face-neck. The
authors suggested that both cumulative dose and solar radiation
intensity may be significant factors in MM.
Methodology:
A descriptive analysis of 5,108 new MM cases (99.5% histologically
confirmed) reported in the Cancer Registry of Norway from 1955-1977
by primary site, sex, age, and birth cohort.
Experimental Design and Analysis Issues:
A clear descriptive analysis of MM incidence data from Norway by age,
site, sex, and birth cohort.
For Result 1:
The distribution of cases by sex and primary site showed male excess
of trunk melanomas and female excess of lower limb melanomas. Age-
adjusted incidence rates increased at a fairly constant, similar rate
for both sexes (approximately 7% per year), more than quadrupling from
1955-1977. Increasing age-adjusted incidence rates by site indicated
a doubling of face-neck melanoma, and a five-fold increase for trunk
and lower limb melanomas. Incidence increases were not as linear by
melanoma site subgroups due to smaller sample sizes and systematic
deviations from linearity. For trunk and lower limb melanomas, the
annual increases in incidence rates by sex were more similar from 1970-
1977 than from 1955-1970.
For Result 2:
Age-specific incidence curves showed increasing incidence in adoles-
cence, a leveling off in middle-age and further increases in older
ages. A comparison of incidence rates for birth cohorts born before
1900, from 1900-1930, and after 1930, showed increasing risk of MM
B-96
-------
in later cohorts with greatest increases occurring from 1900-1930.
Cohort variations were minimal for face-neck melanomas, but were clearly
evident for trunk and lower limb melanomas. Incidence rates for trunk
melanomas among those born 1920-1929 were over six times greater than
among those born 1900-1929.
The ratio of age-specific trunk-lower limb to face-neck incidence rates
by birth cohort indicated that the ratio of carcinogenic exposure to
the two site areas varied by year of birth. A comparison of incidence
per area of primary site for face-neck melanomas and trunk melanomas
(males) or lower limb melanomas (females) indicated a common cancer
pattern for face-neck (gradual increase to age 50, steep rise there-
after) but a unique pattern for the 1930-1949 trunk and lower limb
cohorts (steep increases to age 40, gradual increases thereafter).
Face-neck melanomas were higher among the 1890-1909 cohort than the
1930-1949 cohort.
B-97
-------
References
Moss, A.L.H. "Malignant Melanoma in Maoris and Polynesians in New
Zealand." Br. J. Plast. Surg. 37:73-75 (1984)
Investigator Results:
1. The lower limb was the predominant site for primary malignant melanoma
(MM) tumors (13 out of 24 Maori and Polynesian MM patients), with
almost half of these occurring in the sole. Half of the six head
and neck MM primary tumors occurred in the oral mucosa.
2. Based on the medical history of 24 patients, MM in Maoris and Polyne-
sians appeared to have poor prognoses.
Methodology:
Collection over a 19-year period from the National Cancer Registry
of the National Health Statistics Centre, the New Zealand Health Statistics,
and New Zealand Hospital Boards of information on 21 Maoris and Polynesians
with 24 primary MMs. Histologic material, where available, was reviewed
and autopsy reports were used to confirm the diagnoses on the other
patients. Patient information included race, anatomical distribution,
stage at presentation, depth of tumor invasion, and follow-up mortality.
Experimental Design and Analysis Issues:
A simple descriptive summary of a small, probably non-randomly selected
populaton of 24 Maoris and Polynesians presenting with MM.
For Result 1:
The predominant primary tumor site was the lower limb (13 of 24 patients)
for which the sole was most common (6 of 13). Three of the 6 head
and neck tumors were of the oral cavity. There were 11 nodular and
5 superficial spreading MMs, and 8 unclassified lesions.
For Result 2:
Only 2 patients survived longer than five years. One patient was
lost to follow-up at 6 years. Of the patients known to have died,
13 died of their malignancy within 52 months. It was concluded that
Maoris and Polynesians, perhaps because of their general reluctance
to seek medical advice, have poor prognoses for MM.
B-98
-------
Reference:
Nicholls, E.M. "Development and Elimination of Pigmented Moles, and
the Anatomical Distribution of Primary Malignant Melanoma." Cancer
32:191-195 (1973).
Investigator Results:
The number of pigmented moles per person increased to age 15 (males)
and age 20-29 (females), with peak values reached soonest on most exposed
parts of body. The number then decreased to almost no moles in 80 year-
olds. Depigmented spots and nevi with definite/faint halos were more
common soon after peak nevi values had been reached.
Methodology:
A study population of 1,518 individuals (570F, 948M) was selected and
the number of moles, site of moles, age and sex were recorded. The
source of the study population and the selection method were not identified.
Experimental Design and Analysis Issues:
A simple descriptive analysis of data. The analysis summarized the
age at which the number of moles was greatest and the body sites where
moles were most prevalent.
B-99
-------
Reference:
Paffenbarger, R.S., Wing, A.L., and Hyde, R.T. "Characteristics in
Youth Predictive of Adult-Onset Malignant Lymphomas, Melanomas, and
Leukemias: Brief Communication." JNCI 60:89-92 (1978)
Investigator Results:
Outdoor environmental exposure was associated with increased relative
risks of malignant melanoma (MM) based on a comparison of data from
45 Harvard University and University of Pennsylvania male alumni who
died from MM and four times as many surviving classmates.
Methodology:
A cohort of 50,000 male alumni of Harvard University (entering classes
1916-1950) and the University of Pennsylvania (entering classes 1931-1940)
was retrospectively followed-up with less than one percent lost to
follow-up. Forty-five alumni were identified who had died from MM.
Using this data, relative risks were estimated in a case-control study
using the 45 MM deaths as cases and the 180 surviving classmates born
in the same year as controls.
Potential predictive factors obtained from university records included
history of contagious disease, familial tendencies, physical attributes,
personal traits, and social influences. Causes of death were identified
and classified from official death certificates.
Experimental Design and Analysis:
From the cohort, 45 men who had died from malignant melanoma were identi-
fied (incidence rate 2.6/10 person-years). The potential predictive
characteristics examined for the cases and controls were history of
eight contagious diseases (e.g., measles, mumps), parent dead, only-
child, two or more siblings, tonsillectomy, ponderal index, cigarette
smoker, coffee drinker, exercise, outside work, and New England origin.
Of these, only outdoor employment before college was significantly
associated with malignant melanoma, indicating a relative risk 3.9 times
greater than that of men not reporting outdoor work.
B-100
-------
Reference:
Schreiber, M.M., Bozzo, P., and Moon, T. "Malignant Melanoma in Southern
Arizona." Arch. Dermatol. 117:6-11 (1981)'
Investigator Results:
1. The crude incidence rate of malignant melanoma (MM) in southern
Arizona increased from 6.5/10 in 1969 to 28.6/10 in 1978, repre-
senting an average annual increase of about 18% (note: the authors
state an annual increase of 35%, but 18% is the actual annual rate
of increase).
2. The highest tumor incidence was in 50-59 and 60-69 year-olds.
The most common site was the back, especially in males. The occurrence
of tumors on legs (13% of total) was eight times higher in females.
3. The high melanoma incidence in southern Arizona was "probably due
to meteorologic and geographic factors" allowing penetration of
UV radiation to earth's surface.
Methodology:
The study included information for 533 MMs in whites from all 8 Tucson
hospitals, all 17 practicing Tuscon dermatologists, the 3 private Tucson
pathology laboratories and 8 small southern Arizona hospitals from
January 1969 to December 1978. Patient information consisted of date
of tumor removal or biopsy, age, sex, race, occurrence of metastases
and type of treatment used. Tumor information consisted of location,
size, color, Clark level of invasion, tumor type, and histologic depth
(where available) as well as histopathological confirmation (60% reviewed
by authors). Incidence rates were calculated using 1969-1978 annual
population figures for six southern Arizona counties from the Arizona
Statistical Review.
Experimental Design and Analysis:
A simple descriptive analysis of MM tumor incidence rates in southern
Arizona.
For Result 1:
From 1969-1978, 533 MMs were removed from 533 patients (277 M, 256
F) . The number of MMs increased from 20 (1969) to 120 (1978) , an
increasing crude incidence rate of 6.49 to 78.57 (27.20 standardized)
per 10 , respectively.
For Result 2:
The highest percentage of melanomas for the study period occurred
in 50-59 (24%) and 60-69 (18%) year-olds. The most common location
was the back (31%) with twice as many in males as in females. Other
B-101
-------
sites with similar sex distribution were the face (16%) and arms
(8%). Leg tumors (13%) were eight times as common in females as
in males. "Not specified" melanomas were most common (44%) followed
by superficial spreading melanomas (35%), which were common on backs
of males and females (13%) .
For Result 3:
The annual average increases in southern Arizona melanoma incidence
rates (34-37%) were significantly higher than comparable U.S. rates
(5%, p less than 0.05) based on Third National Cancer Survey data
for 1973-1976. The melanoma distribution by age or sex in southern
Arizona did not substantially change from 1969-1978, nor was it substan-
tially different from the total U.S. distribution. "There are certain
meteorologic and geographic factors in southern Arizona that allow
a greater quantity of UV radiation to reach the earth's surface,
thus increasing the incidence of malignant melanoma..." The authors
noted that Tucson has more sunlight, more clear days, and less daytime
cloudiness than any populated site in North America, and also has
low average humidity, high average temperature, relatively high altitude,
low latitude and low atmospheric ozone.
B-102
-------
Reference:
Pathak, D.R., Samet, J.M., Howard, C.A., and Key, C.R. "Malignant
Melanoma of the Skin in New Mexico 1969-1977." Cancer 50:1440-1446
(1982)
Investigator Results:
1. Malignant melanoma (MM) incidence rates varied with ethnicity.
Rates for non-Hispanic whites (Anglos) exceeded comparison U.S.
rates, and were approximately six times higher than for other ethnic
groups. Annual incidence rates for Hispanics, American Indians,
and blacks (both sexes) ranged from 0.0-1.8/10 .
2. Lower extremities were the most common MM sites in Anglo women,
while the trunk was most common in Anglo men. For Hispanic men and
women, the trunk was the most common site.
3. Statistically significant increasing incidence was observed only
for Anglo women.
4. MM mortality rates varied widely during the study period and did
not correlate with incidence rates.
Methodology:
A descriptive analysis of incidence and mortality rate data was conducted
for 495 New Mexico MM cases reported in the New Mexico Tumor Registry
(NMTR) from 1969-1977. Factors analyzed included ethnicity, sex, age,
and site distribution. Ethnicity of MM cases was determined from reporting
facilities or hospital charts (American Indians), NMTR records or surnames
(Hispanics), and medical records (Anglos). The 1970 Census and 1975
University of New Mexico Bureau of Business and Economic Research popula-
tion estimates were used to estimate age-standardized (to 1970 U.S. popu-
lation) MM incidence rates.
Mortality data was obtained from the New Mexico Bureau of Vital Statistics,
which included information on ethnicity. Mortality rates were calculated
in the same manner as incidence rates. The ratios of New Mexico MM
incidence rates to white rates from the Third National Cancer Survey
and from the SEER program were tested by a method based on the Poisson
distribution. Cart's exact test was used to assess incidence ratios
adjusted for age, sex, and ethnicity. Time trends in incidence and
mortality were assessed with a log-transformed linear regression model.
An exact test described by Zelen was also used.
Experimental Design and Analysis Issues:
A descriptive analysis of incidence and mortality rate data for New
Mexico MM cases.
B-103
-------
For Result 1:
5 5
Anglo incidence rates (mean 8.7/10 male; 9/10 female) were much
higher than for other ethnic groups (e.g., mean Hispanic 1.2/10
male; 1.8/10 female). The overall Anglo-to-Hispanic incidence ratio
(Gart's exact method), age- and sex-adjusted, was 6.0 (p less than
0.01). The differences in incidence were consistent with a protective
effect of skin pigmentation. Except for American Indians, female
incidence rates exceeded male rates in each group. Age distribu-
tions of MM diagnosis were similar for Anglos and Hispanics. Anglo
rates exceeded comparison U.S. rates (for 1969-1971 and 1973-1976)
whereas Hispanic rates were below comparable rates.
For Result 2:
For Anglos and Hispanics, site distribution varied by ethnicity
and sex. Site-specific, sex-adjusted ratios of Anglo-to-Hispanic
incidence (calculated with Gart's exact method) indicated significantly
higher (p less than 0.01) Anglo incidence at each site. Incidence
ratios were 10.0 (head and neck), 6.2 (trunk), 5.6 (upper extremities),
and 24.9 (lower extremities). Male-to-female, ethnicity-adjusted
incidence ratios showed similar risk for head and neck MM, higher
male risk for trunk MM (ratio = 1.6, p less than 0.01), higher female
risk for upper extremities (ratio = 0.61, p less than 0.05) and lower
extremities (ratio = 0.23, p less than 0.01). The overall male-to-
female ratio, age- and ethnicity-adjusted, was not significantly
different from one.
For Result 3:
Age-specific incidence rates, calculated for Anglos only, increased
with age for all sites combined. Age-specific rates for head and
neck MM increased slowly to age 50 and rose steeply thereafter in
males and females. Incidence rates for male trunk rose until 50-59
and then declined. In women, rates for lower extremities increased
up to 60-69 years. In both sexes, the age-rate relationship for
head and neck MM was significantly different than for trunk MM.
A logarithmic regression model showed a statistically significant
incidence rate increase from 1969-1977 for Anglo women (5.5%/year,
p less than 0.05) but not for men (6.8%/year, p = 0.10). Zelen's
exact method did show a statistically significant (p less than 0.05)
incidence rate increase for Anglo men.
For Result 4:
Annual age-adjusted MM mortality rates did not exhibit trends from
1969-1979 and were not consistent with age-adjusted incidence rates
for Anglos and Hispanics.
B-104
-------
Reference:
Pondes, S., Hunter, J.A.A., White, H., Mclntyre, M.A., and Prescott,
R.J. "Cutaneous Malignant Melanoma in South-East Scotland." Quart.
J. Med. Winter 103-121 (1981)
Investigator Results:
1. In a retrospective clinical follow-up of 477 cutaneous malignant
melanoma (MM) patients in south-east Scotland, mean annual incidence
during 1971-1976 was 4.6/100,000 with female incidence almost twice
as high as males.
2. Median age at presentation was in 6th decade (both sexes), with
88% having clinical Stage I disease. One-third of all primary
lesions were on female lower leg. Superficial spreading melanoma
(SSM) was the most common growth pattern.
3. Overall 5-year survival rate for males (48%) was significantly
less than for females (67%). Other factors with effects on prognosis
were thickness of primary lesion (best index of prognosis), age,
tumor site, and mitotic rate. Improved survival may be achieved
better by earlier diagnosis and treatment than by change in management.
Methodology:
A descriptive analysis of a retrospective clinical follow-up of 477 patients
(315 F, 162 M) presenting with MM from 1961-1976 in south-east Scotland.
Cases were identified primarily from 3 pathology departments, the regional
cancer registry and the Lothian Health Board's list of hospital admissions.
Incidence rates were calculated for 1971-1976 patients. Case information
included sex, age, address, site of primary lesion, clinical stage
of disease, operative treatment, date and site of first recurrence.
Follow-up information was from general practitioners, case records,
regional cancer records, and death certificates. When available, specimens
were histologically reviewed and used to determine histogenic type,
primary tumor depth, level of penetration, and mitotic activity.
Experimental Design and Analysis Issues:
For Result 1:
Mean annual incidence for males and females from 1971-1976 was 4.6/10 ,
3.2/10 for males and 5.8/10 for females. Fluctuations in incidence
rates were due to the small number of cases.
For Result 2:
Clinical stage of disease at presentation (in 385 of 404 case records)
was Stage IA and II local disease (88%), Stage II regional node disease
(10%), and Stage III disease (2%). The age/sex specific incidence
rates showed little incidence change between 30-70 years (3.1-6.0/10
B-105
-------
males, 7.7-9.3/10 females) and higher incidence over 70 years
(11.4-16.1/10 males and females). Nearly a third of all tumors
were on the female leg. Trunk melanomas represented 27% of male
tumors, 14% of female tumors. Head and neck melanomas occurred signifi-
cantly later than tumors at other sites and this was the only site
where mean age of sexes differed (p less than 0.05).
Distribution of tumor size was similar for males and females, but
larger tumors were more frequent in older patients. SSM was most
common (51%), followed by nodular melanoma (35%), and lentigo maligna
melanoma (9%). The distribution of histogenic type was similar in
each sex. There was no significant association between sex and lesion
thickness and no apparent relationship between size and depth (less
than 3.5 mm) of tumors. Deep tumors were most common on the trunk,
thin tumors were most common on head, neck, and upper limbs. Deeper
tumors and nodular melanomas had higher mitotic indexes.
For Result 3:
Five-year survival rates in men (48%) and women (67%) were significantly
different (p less than 0.001). The survival prognosis worsened in
women after 70 years, and in men after 50 years. Survival varied
significantly (p less than 0.01) by site, with decreased survival
for trunk melanoma patients (in females, 39% trunk vs. 76% head and
neck) .
Depth of primary tumor was most important prognostic factor, with
97% 5-year survival rate for tumors less than 0.5 mm deep vs. 38%
for tumors greater than 3.5 mm deep.
High mitotic index was also associated with poor prognosis, even
when depth was taken into account (p less than 0.01). There were
no significant differences in survival for three patient cohorts
(1961-65, 1966-70, 1971-76).
B-106
-------
Reference:
Reintgen, D.S., McCarty, K.S., Cox, E., and Seigler, H.F. "Malignant
Melanoma in the American Black." Curr. Surg. 40:215-217 (1983)
Investigator Results:
1. The median survival (MS) time for 31 blacks with melanoma was 31.4
months, with a 5-year survival rate of 22%. Black men (MS = 26.3 months)
and Stage II patients (MS = 10.5 months) had worse prognoses than
black women (MS = 40.0 months) and Stage I patients (MS = 41.4 months).
2. Blacks had statistically significant lower actuarial survival rates
than whites (p = 0.000001) which could not be accounted for by
differences in age, sex, stage of disease at diagnosis, Clark's
level or primary site. The most common site for blacks was the
foot vs. the trunk for whites.
Methodology:
A descriptive analysis of survival rates among 31 black melanoma patients
(15M, 16P) registered at the Duke University Comprehensive Cancer Center
since 1972 and 100% followed-up for periods ranging from 6 months to
10 years. Information included age at diagnosis, primary site, sex,
disease stage, and Clark's level. The number of Caucasians registered
at the Cancer Center and used in the study were not identified, nor
were their follow-up periods indicated.
Experimental Design and Analysis Issues:
Acturial survival curves were constructed for black and white populations
and the Cox-Mantel rank test was used to test for statistical significance.
For Result 1:
No additional information provided.
For Result 2:
Black actuarial survival rates were lower than white survival rates
(p = 0.000001) even when adjusted for age (p = 0.0007), sex (p = 0.0002),
Clark's level (p = 0.0003), or primary site (p = 0.001). When
controlled for Clark's level, the mean survival time for whites
was 86.1 months compared to 26.8 months for blacks.
The most common site for blacks was the foot (60% plantar or subungual),
whereas for Caucasions the predominant site was truncal. Blacks
presented with more advanced disease stage (64% Stage I vs. 85.3%
whites) and showed more invasive disease (88% Clark's level 4 or
5 vs. 60% for whites).
B-107
-------
Reference:
Reynolds, P. and Austin, D.F. "Epideraiologic-Based Screening Strategies
for Malignant Melanoma of the Skin." Advances in Cancer Control:
Epidemiology and Research. Alan R. Liss, New York (1984). Pp. 245-254
Investigator Results:
Individuals with 12 or more large moles, previous diagnosis of basal
or squamous cell skin cancer, parental history of skin cancer and propen-
sity to burn with sun exposure were at high risk for melanoma. A screening
program for individuals at high risk for developing melanoma was proposed.
Methodology:
Relative risk and odds ratios were calculated in a case-control study
of 31 malignant melanoma (MM) employees of the Lawrence Livermore National
Laboratory (LLNL) diagnosed between 1969 and mid-1980 (cases) and 110
LLNL controls matched by age, race, and sex. The data which was collected
by mailed questionnaire and personal interview included presence of
moles, previous diagnosis of basal or squamous cell cancer, parental
history of skin cancer, propensity to sunburn/tan, eye, hair and skin
color, and freckling or Celtic heritage.
Experimental Design and Analysis Issues:
Four familial risk factors were significantly associated with MM:
presence of moles greater than 1/2 cm diameter (p less than 0.0005)
previous diagnosis of basal or squamous cell skin cancer, parental
history of skin cancer, and propensity to burn rather than tan with
sun exposure (all p less than 0.05). With increasing number of moles,
risk of having MM increased. Individuals with 12 or more moles were
41 times more likely to have MM (p less than 0.005).
Small subgroups with 12 or more moles and the identified familial
risk factors may experience over a 200-fold risk. Comparison of
cases with controls and another random sample of white Contra Costa
County residents (CA) indicated that about 2% or less of general
white population may be at high risk by having 12 or more large moles.
The risk factors represented criteria which could easily be used
for melanoma screening. The proposed screening program would consist
of dermatological examination, health education, collection of question-
naire/interview data, and follow-up.
B-108
-------
References
Scotto, J., and Nam, J. "Skin Melanoma and Seasonal Patterns." Am. J.
Epi. 111(3): 309-314 (1980)
Investigator Results:
A strong seasonal pattern of melanoma incidence with summertime peak
was observed for females, particularly those under 55 years and those
of all ages for melanomas of the upper and lower extremities. For
males, a seasonal pattern with summertime peak was observed only for
melanomas of the upper extremities.
The authors concluded that it was difficult to determine if seasonal
melanoma patterns resulted from promoting effects of UV-B exposure
or from enhanced recognition during summer months.
Methodology:
Data were obtained from the Third National Cancer Survey for nine U.S.
locations during 1969-1971 of 2,168 skin melanomas in whites for which
month of first diagnosis, age, sex, and anatomic site were reported.
A sine curve first-order harmonics model was used to test for seasonal
patterns. The model was:
y = £ + b sin (P + 2L- x)
where y = proportional distribution of skin melanomas for interval
x (e.g., month), k = no. intervals in full cycle, b and P are amplitude
and phase angle, respectively (estimated by maximum likelihood method).
A chi-square goodness-of-fit test was used to verify the fit of the
model to the data, and if not rejected was followed by Edwards' test
of b.
Experimental Design and Analysis:
A statistical analysis of the fit of the first-order harmonics model
to the data (using chi-square or Edwards' test) was conducted by sex
and anatomic site.
Among females, over 20% of all cases were diagnosed during summer months
(June and July) while less than 14% were diagnosed during winter months
(December and January). The seasonal pattern among females was statisti-
cally significant (p = 3xlO~ Edwards' test). No sustained peak or
trough was observed for males.
When monthly seasonal trends by anatomic site were examined, female
seasonal patterns for upper and lower extremities were statistically
significant (p less than 0.001) while for males the upper extremity
pattern did not quite achieve statistical significance (p = 0.11).
When analyzed for bi-monthly seasonal patterns, both male and female
trends for upper extremities were statistically significant (p = 0.07
and p = 0.003, respectively).
B-109
-------
Analysis of seasonal patterns by age (under 55 or 55 and over), indicated
significant trends for females only (p = 0.002 for less than 55> p = 0.004
for 55 and over). Among females less than 55, significant peaks in
summer were observed for lower extremities (p = 0.001) and upper and
lower extremities pooled (p = 0.0003), but not for trunk or face and
head.
By geographic region, female seasonal patterns were most significant
in the south (p = 0.008), followed by the north (p = 0.03), but were
not significant in mid-laditude region.
B-110
-------
Reference:
Shaw, H.M., McGovern, V.J., Milton, G.W., Farago, G.A., and McCarthy,
W.H. "Histologic Features of Tumors and the Female Superiority in
Survival from Malignant Melanoma" Cancer 45:1604-1608 (1980)
Investigator Results:
1. In 780 Stage I malignant melanoma (MM) patients, a direct correlation
between 5-year survival rate and tumor thickness was observed in
males and females, but females had higher survival rate at each
thickness level. Average tumor thickness was significantly less
in females due to preponderance of thin lesions in females and
thick lesions in males.
2. No significant sex differences in survival were apparent When evidence
of regression, histogenic type, and mitotic activity were examined.
Prognosis for regressing and nonregressing lesions was markedly
different only for very thin lesions.
Methodology;
A descriptive analysis of the survival rates of 780 patients (372M,
408F) who presented with Stage I MM at the Sydney Hospital Melanoma
Clinic between January 1950 and March 1978. Patient information consisted
of tumor thickness (0.1-0.7 mm (196 patients), 0.8-1.5 mm(225), 1.6-3.0 mm
(220), and 3.1+ mm (139)), evidence of regression, histogenic type
and grade of mitotic activity. Cumulative survival rates were calculated
(life table method) and differences were statistically analyzed (logrank
method).
Experimental Design and Analysis Issues:
For Result 1:
The 5-year survival rate was significantly higher for females (82.7%)
than for males (66.6%) (p less than 0.001). Five-year survival
rates also differed for males and females at all tumor thickness
levels. Average tumor thickness was less in females due to preponderance
of thin lesions in females (62% vs. 37.2% in males, p less than
0.001) and thick lesions in males (56.8% vs, 43.2% in females,
p less than 0.02). The largest difference in 5-year survival rates
was for lesions 3.1+ mm thick (63.2% males, 42.9% females).
For Result 2:
Significantly more lesions in males than females had evidence of
regression (42.9% vs. 28.9%, respectively, p less than 0.001),
and this occurred for all thicknesses. Evidence of regression
was slightly, but not significantly, associated with improved survival
rates. There was no difference between males and females 5-year
survival rates when histogenic type or grades of mitotic activity
were examined.
B-lll
-------
References
Shiu, M.H., Schottenfeld, D., Maclean, B., and Fortner, J.G. "Adverse
Effect of Pregnancy on Melanoma." Cancer 37:181-187 (1976)
Investigator Results:
No significant difference in 5-year survival rate for Stage I melanoma
patients was observed between nulliparous, parous nonpregnant, and
pregnant females. For Stage II patients, a significantly lower 5-year
survival rate (p less than 0.05) was observed for pregnant patients
(29%) and parous patients who had lesion activation in a previous pregnancy
(22%) as compared with that for nulliparous patients (55%) and other
parous patients (51%) .
Differences in survival rates and symptoms in Stage II patients (e.g.,
bleeding, ulceration, irritation, and elevation of lesion) "strongly
suggest an adverse influence of pregnancy among females with Stage II
melanoma."
Methodology:
A descriptive analysis of survival rates in 251 15-45 year-old female
cutaneous melanoma patients (165 Stage I, 86 Stage II) who received
treatment at Memorial Sloan-Kettering Cancer Center from 1950 to 1969
and for which accurate recorded data on pregnancy at time of admission
was available. The effect of pregnancy on prognosis was examined through
four study groups: nulliparous women, parous women with no activation
of lesion during previous pregnancy, parous women with definite activation
of lesion during previous pregnancy and women with melanoma admitted
and treated during pregnancy. Statistically significant differences
were evaluated using the chi-square test with Yates correction.
Experimental Design and Analysis Issues:
A descriptive analysis of 5-year survival rates among female melanoma
patients.
The overall 5-year survival rate was 84% for Stage I patients with
little difference between study groups. For Stage II patients, overall
survival rate was 42%, and was higher for nulliparous women (55%) and
parous women with no lesion activation in previous pregnancy (51%).
Significantly lower 5-year rates (p less than 0.05) were observed for
parous women with lesion activation in previous pregnancy (22%) and
for pregnant women (29%) combined when compared with the two other
study groups combined.
Age differences for the study groups did not contribute towards differences
in 5-year survival rates. Lesions on the lower extremity accounted
for 49% of all melanomas. Trunk lesions were more frequent for parous
women with prevous lesion activation and pregnant women, especially
for Stage II patients. Parous patients with previous lesion activation
B-112
-------
and pregnant patients had more frequent symptoms of melanoma (bleeding,
irritation, itching, scaling, ulceration, and/or elevation) but did
not have statistically significantly poorer survival rates than asympto-
matic patients.
B-113
-------
Reference:
Stevens, R.G. and Moolgavkar, S.H. "Malignant Melanoma: Dependence
of Site-Specific Risk on Age." Am. J. Epi. 119:890-895 (1984)
Investigator Results:
1. Analysis of melanoma incidence data from Connecticut and Denmark
showed steadily increasing rates in successive birth cohorts, which
accounted for secular trends in melanoma of the trunk and limbs
and for differences seen among cross-sectional age curves of various
sites.
2. Incidence data did not support hypothesis that melanoma of the
face and melanoma of the trunk and limb involve distinct pathogenic
mechani sms.
Methodology:
Melanoma incidence data specified by site from Connecticut (1935-
1981) and Denmark (1943-1972) were fit to two models that estimated
the expected number of melanoma cases at each site (Model I) and
the expected number of cases at all sites (Model II) based on age,
year of birth, and person-years at risk.
Experimental Design and Analysis Issues:
A study of the fit of incidence data to two age, birth cohort models.
For Result 1:
Age-adjusted incidence rates of melanoma of the trunk increased the
most in males, and melanoma of the legs increased the most in fe-
males. Model I and Model II fit the data for each population by
sex and by site well. The fit of the models were consistent with
an age dependence that was similar for each site. The slope of estimated
age effects was not close to zero for any individual site or all
the subsites fitted simultaneously. The slope of the regression
line of log(age effect) on log(age) was 4.3 for males and 3.5 for
females in Connecticut and 4.5 for males and 4.2 for females in Denmark.
In both Connecticut and Denmark, the male slope was greater than
the female slope, indicating more rapid incidence increases with
age in males than females.
For Result 2:
The results showed that increases in melanoma of trunk and limbs
could be described by cohort effects. The relationship of age to
incidence was the same in all sites, thus different pathogenic mechanisms
did not necessarily explain site-specific melanoma incidence data.
B-114
-------
Reference:
Swerdlow, A.J. "Incidence of Malignant Melanoma of the Skin in England
and Wales and its Relationship to Sunshine." Br. Med. J. 2:1324-1327
(1979)
Investigator Results:
1. Mean skin malignant melanoma (MM) incidence in 14 English health
regions and Wales correlated negatively with latitude and posi-
tively with hours of sunshine, suggesting that exposure to sun-
shine was an important causal factor.
2. Male and female MM incidence within a region tended to show similar
yearly fluctuations implying a common factor affecting incidence
in both sexes with a short latent period of action.
3. Incidence of MM in females correlated positively with hours of
sunshine 2 years earlier, indicating that exposure to sunshine
may cause melanoma after about a 2-year induction period.
Methodology:
For Oxford Region residents, age-standardized sex-specific annual
MM incidence rates for 1952-1975 were calculated using 1961 Oxford
Region population, incidence data from the Oxford Cancer Registry
and age-specific 1955-1974 population data from published sources.
For other regions of England and Wales, crude incidence rates for
1955-1969 in the Southwestern Region and for 1962-1970 in the re-
maining regions and in England and Wales overall were used. Mean
incidence, the rate of increase in incidence, and the expected in-
cidence in each year were calculated for each region. Mean daily
hours of bright sunshine were estimated for each region from the
meteorological office "District values station" nearest to main pop-
ulation centers. For England and Wales, overall sunshine hours were
obtained from the Registrar General's Statistical Review of England
and Wales.
For each region for males and females separately, mean incidence
and rate of incidence increase were correlated with latitude of main
regional population center and with hours of sunshine in the region
2 years earlier. Correlations for deviations of male and female
rates from expected rates (based on linear regression) were also
calculated. Correlations were also calculated between annual deviations
from expected melanoma incidence and deviation from expected annual
hours of sunshine 1, 2, 3, 4, and 5 years earlier.
Experimental Design and Analysis Issues:
A cross-sectional analysis of the relationship of incidence data and
hours of sunshine and latitude in England and Wales.
B-115
-------
For Result 1:
Age-standard!zed MM incidence rates rose for females in all but three
regions and in all but four regions for males, the increases generally
being greater in females. Both the mean incidence and rate of secular
increase in incidence for males and females were negatively correlated
with latitude and positively correlated with hours of sunshine in
the 15 English regions and Wales combined (several of the correlations
were statistically significant).
For Result 2:
The annual female MM incidence correlated positively with annual
male incidence in 12 of 15 regions after discounting for long-term
trends. The positive correlation was statistically significant in
three regions (p less than 0.05) and for all regions combined (p
less than 0.01).
For Result 3:
After discounting for long-term trends, correlations between female
incidence and hours of sunshine 2 years earlier were positive in
11 of 15 regions (p less than 0.01 in one region) and for all regions
combined (p less than 0.05). No other time interval, from 1 to 5
years, indicated a strong pattern of positive correlation for females,
with the 1-year period showing negative correlations for all but
three regions (two were significant with p less than 0.05).
For males, no association between yearly MM incidence and prior sunshine
was apparent for any lag interval. The 2-year period gave positive
correlations in 6 of 15 regions and for all region-years combined.
B-116
-------
Reference:
Teppo, L., Pakkanen, M., and Hakulinen, T. "Sunlight as a Risk Factor
of Malignant Melanoma of the Skin." Cancer 41:2018-2027 (1978)
Investigator Results:
1. The age-adjusted incidence of malignant melanoma (MM) in Finland
(1953-1973) was equal in males and females, and more than doubled
during the study period. The most common locations were the trunk
(48% in males, 28% in females) and lower limbs (17% in males, 36%
in females).
2. The incidence of trunk melanomas in both sexes and lower limbs
in females (Group I) increased markedly with time, with age-specific
incidence rates for these sites increasing sharply in middle age
and levelling off thereafter. Melanomas on head and neck in both
sexes and lower limbs in males (Group II) did not increase with
time, and risk was low in middle age but increased throughout life.
3. Age-adjusted incidence rates were higher in urban areas than in
rural areas and higher in southern parts of the country. After
adjusting for urban/rural differences, north/south differences
almost disappeared, implying that the north/south gradient was
attributable to degree of urbanization and not necessarily to the
effect of latitude itself.
4. The increased MM incidence with time could be accounted for by
a cohort effect. Recognition of sunlight as the only important
MM risk factor may be an oversimplification.
Methodology t
The study analyzed all 2,501 (1,108 M, 1,393 F) cases of MM (malignant
lentigo excluded) reported to the Finnish Cancer Registry in 1953-1973.
Histological confirmation was provided for 98% of cases. Also obtained
from the Registry were 1966-1970 data on basal cell carcinomas and
"other" skin cancers, as well as death certificates for cutaneous melano-
mas used to calculate mortality rates. All rates were age-adjusted
to "world standard population" and urban/rural-adjusted. The country
was geographically divided into four regions.
Experimental Design and Analysis Issues:
A descriptive and birth cohort analysis of Finnish MM incidence and
mortality data.
For Result 1:
The most common tumor sites were the trunk (48% in males, 28% in
females) and lower limbs (17% in males, 36% in females) , whereas
only 14% of tumors were on the face. The annual age-adjusted incidence
B-117
-------
rates more than doubled for both sexes from 1953-1973 and was about
3.5/10 in early 1970's. The increase in time was most marked for
trunk tumors in males (80% of increase) and females (28% of increase),
and on upper and lower limbs (44% of increase) in females. The timing
of the increase in melanoma was similar to that observed in many
other countries and the same birth cohorts were involved. Since
mortality from melanoma increased at the same time as the increase
in incidence, changes in diagnosis do not explain the increases.
For Result 2:
The increase in risk mainly concerned age groups over 30, with melanomas
on the trunk in both sexes and on lower limbs in females showing
relatively high rates in the 30-49 age group and a levelling off
in older age groups. For head and neck tumors, and lower limb tumors
in males, middle- age rates were lower and increased continuously
with age. The shape of the curves were similar for successive time
periods (1953-59, 1961-70, 1971-73). Male MM excess on the trunk
and female excess on lower limbs were apparent in almost all age
groups.
For Result 3:
Age-adjusted MM incidence among males increased more in urban than
rural areas. In both urban and rural areas, increasing incidence
was observed for the male trunk and the female lower limbs. Age-adjusted
incidence rates were also higher in the south than the north. When
adjusted for urban/rural population ratios, the north/south gradient
decreased for 1953-1959 and almost disappeared for 1961-1970. The
authors suggested that "... people in urban areas in Finland probably
experience more exposure to the sun (open air leisure, holidays)
than those living in rural districts where skin has traditionally
been more protected from direct sunlight. The association with urbani-
zation is consistent with the findings from England and Wales that
the risk of melanoma in males is highest in professional and managerial
workers and administrators, particularly at younger ages.
For Result 4:
Each successive male birth cohort born before 1940 and each female
cohort born before 1930 experienced a higher risk of melanoma than
the previous cohort. The increase in melanoma incidence in Group I
can be interpreted in terms of a cohort effect accounted for by changing
clothing habits which increased sunlight exposure to Group I sites.
In Group II, exposure didn't change with time and no changes in melanoma
risk were observed. However, the anatomical distribution of melanomas
does not correspond to degree of sunlight exposure, and incidence
rates in the face of both sexes equaled that of lower limbs in males
although there is a marked difference in sunlight exposure for these
sites.
B-118
-------
Reference:
Venzon, D.J., and Moolgavkar, S.H. "Cohort Analysis of Malignant
Melanoma in Five Countries." Am. J. Epi. 119:62-70 (1984)
Investigator Results:
1. A cohort analysis of malignant melanoma (MM) mortality data, after
adjusting for geographic and temporal trends, indicated a higher
proportion of deaths in females than in males among younger age
groups.
2. Increases in mortality rates by birth cohort were approximately
equal in different regions and appeared to be slowing down in more
recent years.
Methodology:
Mortality data for MM in England and Wales, Canada, New Zealand, and
the U.S. were obtained from the WHO which included five five-year
cross-sections from 1951-1975 except for New Zealand with four cross-sections
from 1955-1974 and a fifth from 1975-1977. Australian data were provided
for nine five-year cross-sections from 1931-1975. For all countries,
14 five-year age groups (covering 15-84 year-olds) were used.
The data were fit to several different models (tested with chi-square
goodness-of-fit) which estimated expected MM deaths. Model 1 estimated
the expected number of deaths based on age group, birth cohort, sex
and country:
E. . = N.. a. b. .
13cs 13cs ics ijcs
where E.. = expected deaths in age group i, birth cohort j, country
c, and sex s, N^. = size of population (person-years), a- = effect
of being in age group i, c, and s, and b.. = effect of belonging
to birth cohort j and i,c, and s. Modelz was a simplified version
of model I, but added a parameter r to adjust for relative rates
in each sex- and country-specific population. Model 1 represented
complete dependence of age and cohort parameters on population, whereas
Model 2 represented complete independence. Model 3 assumed that mortality
was similar in all populations but that cohort effects differed by
more than a multiplicative constant (E.. = N.. a. b. ). Model 4
tested for age effects which differed by sex but were constant over
countries (E.. „ = N.. a. b. ).
1}CS 1JCS IS JCS
Experimental Design and Analysis Issues:
A descriptive analysis of the fit of mortality data to a range of age,
birth cohort, country and sex dependent models.
B-119
-------
For Result 1:
Model 1 fit the data well (p less than 0.05) except for Australian
males. Model 2 did not fit Australian males, U.S^ white females,
nor England and Wales females well. The total fit for Model 3 was
poor, primarily due to the U.S. white female population. The data
on U.S. females differed in that mortality increased less rapidly
with age than for all other populations. Two age curves, one for
males and one for females, were believed to suffice for all five
countries. The curve for males increased faster with age than for
females, implying that a larger proportion of female deaths occurred
in younger age groups.
For Result 2:
The fit of the data to Model 4 indicated that relative increases
in mortality by birth cohort have been approximately the same (2-4%/year)
in different regions. This trend appeared to be slowing down or
leveling off in more recent birth cohorts.
B-120
-------
For Result 2:
The mean age of facial melanoma development was higher than for other
sites, with incidence rising rapidly after age 60. Incidence of
facial melanoma changed little over study decades. Incidence of
nonfacial melanoma rose steadily until middle age and then plateaued.
Increasing male melanoma rates in CT were accounted for by 10-fold
increases in trunk and neck lesions between 1970-1974 and 1935-1944.
Increased female melanoma rates in DK were largely attributable to
rapid increase in lower extremity lesions particularly among 30-39
year-olds.
For Result 3:
When the DK data were analyzed according to birth cohorts, little
change in facial melanoma incidence was observed, but rates 'for the
trunk-neck and upper and lower extremities tended to increase with
younger cohorts (beginning with 1882 cohorts).
For Result 4:
The ratio of observed to expected annual melanoma incidence relative
to melanocyte density indicated excessive rates of facial melanoma
(both sexes) and trunk-neck lesions (males) but lower than expected
rates of upper extremity melanoma (both sexes) and lower extremity
melanomas (males) in CT and DK. Declining ratios of facial melanoma
were observed during the study period with greatest decreases in
middle ages.
For Result 5:
Differences observed in facial versus nonfacial melanoma suggested an
acute and chronic effect of sun exposure. Of all sites, the face is
most chronically exposed. Facial melanomas are strongly age-related
(incidence rises sharply after age 60) and are more frequent than
would be predicted from facial surface area and melanocyte density.
Sites less exposed to sun, such as male trunk and female leg, show
excessive melanoma incidence strongly associated with middle, not
old, age. Effects of continued prolonged solar radiation appear
to be implicated in facial melanoma, whereas short-term, noncumulative
effects of intense solar radiation appear to be related to melanomas
of the trunk or leg.
For Result 6:
The increase in age-adjusted incidence rates in CT from 1935-1975
was significantly correlated in a linear regression equation with
time over three sunspot cycles that occurred in the 33-year study
period (r = 0.9327) . Deviations in incidence from the regression
line were also cyclic. Statistically significant partial correla-
tions were observed between annual sunspot numbers and annual mel-
anoma incidence in each of the 3 years following sunspot maxima,
B-122
-------
with closest correlations 2 years after maxima. In additional New
York data, the timing of increased melanoma rates was also significantly
correlated with annual sunspot numbers (with 1-2-year lag).
B-123
-------
References
Viola, M.V./ and Houghton, A.N. "Solar Radiation and Cutaneous Melanoma."
Hosp. Prac. 17:97-106 (1982).
Investigator Results:
1. Melanoma incidence rates in Connecticut (CT) and Denmark (DK) were
similar over approximately 3 decades of study.
2. Rising incidence of melanoma has occurred in non-random patterns
identifiable in terms of site, age, sex, and country.
3. Birth cohort effects in DK melanoma incidence were observed for
the trunk-neck and upper and lower extremities.
4. Greater than expected number of facial melanomas (both sexes) and
trunk-neck lesions (males) were obserbed in both CT and DK. Lower
than expected rates were observed for upper extremities (both sexes)
and lower extremities (males).
5. Differences observed in facial versus nonfacial melanoma suggested
both an acute and a chronic effect of sun exposure.
6. Age-adjusted incidence rates in CT were significantly correlated
with time over three sunspot cycles (r = 0.9327), and significant
partial correlations were observed between annual sunspot numbers
and annual melanoma incidence.
Methodology:
Over 7,500 melanoma cases registered in the CT Tumor Registry (2,966
cases from 1935-1974) and the Danish Cancer Registry (4,547 cases
from 1943-1974) with information on age, sex, and tumor site were
analyzed. Age-adjusted incidence rates were calculated using relevant
U.S. and Eurpoean data.
Experimental Design and Analysis Issues:
A descriptive analysis of melanoma incidence in CT and DK by age, sex,
and tumor site, and an analysis of correlations between incidence and
sunspot cycles.
For Result 1:
Melanoma incidence was 1.1/10 in CT (1935) and in DK (1943). By
1974, incidence had risen to 5.7/10 in DK and 5.8/10 in CT. Age-
specific data indicated increasing rates after age 20, leveling off
between 40 and 60, and increasing after age 60. When analyzed by
sex, DK incidence data in females increased more rapidly, whereas
the CT incidence in males increased by more.
B-121
------- |