i UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
' WASHINGTON, D-C. 20460
EPA-SAB-RAC-LTR-93-004 December 9, 1992
Honorable William K, Reilly
OFFICE OF
THE ADMINISTRATOR
U.S. Environmental Protection Agency
401 M Street, S.W,
Washington, D. C, 20460
Be: Evaluation of EPA's Proposed Methodology for Estimating Radiogenic
Cancer Risks
Dear Mr. Reilly:
In a memorandum dated January 13, 1992, Margo T. Oge, Director, Office
of Radiation Programs, asked the Science Advisory Board to review EPA's revised
methodology for estimating human cancer risks from exposures to ionizing
radiation. The charge for this review requested the SAB to respond to the
following four questions:
1, Has the Agency analysis considered the most relevant risk estimates
of low-LET radiation?1
2. Does the Agency analysis accurately compare the most relevant
features and assumptions of the various models?
3. Is the Agency's analysis technically sound?
4. Are the recommended methods for estimating the cancer risks
appropriate and supportable in light of the current scientific evidence?
In addition to the charge, the ORP initially provided the SAB with extensive
background material as listed in Appendix 1. On May 1» 1992, ORP provided the
SAB with a follow-up document titled "Proposed Methodology for Estimating
Radiogenic Cancer Risk."
LET — Linear energy transfer, a measure of the rate at which radiations deposit energy in matter and create
toftixatton. Gamma and beta radiation are wiwidered to be low-LET radiation, while alpha radiation is high-LET and
leads to moi« densely distributed jonizatiem along its track. Adjustments must be zaadfi to make a dose of high-LET
radiation biplagiealy "equivalent" to a larger ctose of low-LET radiation, by the introduction of a "quality factor" (Q)
which -represents aa approximated ave^p "Relative Biological Effectiveness11
-------
In the opinion of the Radiation Advisory Committee EPA has reviewed and
considered all major new data sets and current risk estimates of low-LET ionizing
radiation. Although no single data set and model for predicting radiogenic cancer
risk is ideal, the method of analysis chosen by EPA is adequately supported by
present scientific evidence. A few areas of uncertainty exist that eventually may
require modification of the Agency's analysis when further data become available.
Among these Is the method for utilizing ("transporting") risk estimates from, the
atomic bomb survivor study in Japan where the base-line risks for several cancers
differ significantly from those in the U.S. Another is the question of whether to
apply a "Dose Rate Effectiveness Factor" (DREF) for solid tumors at low dose
rates or at low doses of low-LET radiation; the Agency's choice of a DREF of 2 Is
in accord with the current choice of other radiation protection groups world-wide.
An additional concern is the continuing uncertainty in the dosimetry for the
Japanese atomic bomb survivors including the magnitudes of the neutron
components. Further discussion of these and other issues is contained in the
subsequent parts of this Letter Report.
The Radiation Advisory Committee addressed the charge and the
background materials at its meetings on February 12 and May 21, 1992 and
approved this report August 5, At the first of these meetings, the Committee
orally informed the ORP that its initial analysis was sufficient to guide further
work and accepted QRP's proposal to limit further consideration to the cancer risk
models developed for the Nuclear Regulatory Commission (NEC) by Ethel Gilbert
and for the International Commission on Radiological Protection (ICRP) by
Charles Land and Warren Sinclair.
On May 21, 1992, ORP described the procedure and risk coefficients2 it
proposed to use henceforth for estimating cancer risk from ionizing radiation. In
brief, EPA proposed largely to adopt the ICRP model and use risk coefficients that
are the geometric means of the two projections presented by Land and Sinclair for
most cancer sites. Risk coefficients for other sites (e.g., liver, bone, and thyroid)
are based on a variety of underlying data sets thought by EPA to be most
appropriate. The risk coefficients presented by Land and Sinclair in ICRP
Publication 60 (1991) were derived from the observations of cancers in the
Japanese survivors of the atom bombs and then "transported" to the U.S.
population by means of two different methods3. One method assumes that the
excess relative risk in the U.S. would be the same as in Japan regardless of
The risk eoeSieimtg a*« expressed as the ftxcess probability of developing fatal cancer over a lifetime of erposore per
unit dose-equivalent (rsm or Sv).
Transport1' of radiation risk estimates across populations generally refers to th« method(s) by which a cancer risk
estimate obtained for out population is made applicable to (or may be compared to) that of another population wheti
the underlying background risks for the two populations differ.
-------
differences in baseline cancer rates (the "multiplicative" model) and the other
assumes that the excess absolute risk over the period of observation will be the
same in the two populations, and then projects a constant relative risk forward in
time for the U.S. population (the NIH model). Both methods thus assume a
constant relative risk above the baseline cancer risk in an unexposed population
independent of time after exposure, except for leukemias. Furthermore, a
minimum latent period of 10 years is assumed to apply for all solid tumors. The
risk of radiogenic leukemia is subject to a different analysis that includes a shorter
latency and a relative risk that increases with time after exposure and then
declines. For cancers other than those identified by sites, EPA proposes to use
the risks for exposure during ages 10-19 as an approximation of the risks for
persons exposed under 10 years of age rather than the Land and Sinclair
estimates. EPA's justification is that risks estimated by these authors for
exposures at less than 10 years are inexplicably low for males and high for females
and are based on few observations with likely high sampling errors.
With this description as background, the Committee offers the following
responses to the charge:
1, Has the Agency analysis considered the most relevant risk
estimates of low-LET radiation?
Yes. The Committee commends ORP for considering all the major analyses
of the Japanese epidemiology as well as other studies of radiogenic cancer risks.
2. Does the Agency analysis accurately compare the most relevant
features and assumptions of the various models?
For the most part, yes. ORP has presented a thorough and unbiased
description of the strengths and limitations of the various data sets and analyses
of radiogenic cancer risk. The Agency could have noted that several of the risk
estimates presented in the NRG proposal are not directly dependent on the
dosimetry in the atom bomb survivors and are therefore more robust, in a limited
sense, than are the estimates in the ICEP model
3, Is the Agency analysis technically sound?
For the most part, yes. While arguments could be raised that the National
Academy of Sciences BEIR V report contains results that might have been given
greater consideration, the risk estimates for all cancer sites combined are relatively
consistent among all analyses of the Japanese experience, and this estimate is the
one primarily needed by EPA because most regulations will be based on overall
-------
cancer risk. The following remarks respond to four specific important aspects of
the Agency's analysis; the geometric mean of transport models, dose rate
effectiveness factor(s), relative biological effectiveness (RBE) for alpha radiation,
and uncertainty analysis.
a) Geometric Mean of Transport Models
The Committee notes that EPA did not offer any scientific rationale for
using the geometric mean rather than consistently using one or the other of the
transport models or using a different weighting procedure (e.g., the arithmetic
mean). The Agency did effectively argue that neither of the transport models gave
results that seemed reasonable for all individual organs, and supported the idea
that the geometric mean for each organ provides a measure of central tendency
that reflects the results of each of the models without allowing the result of one to
dominate the mean value. At the same time, the geometric mean seems the more
consistent with the limited data available on the radiogenic cancer risks in the
U.S. The risks thus calculated are also reasonably consistent with the NRG risk
coefficients, which were derived judgmentally from both the Japanese data
transported to the U.S. and other considerations. Thus the EPA procedure is as
supportable as any other for estimating organ-specific and total cancer risks from
low-LET radiation. The Committee notes, however, that the site-specific risk
estimates are far from firm and should be used with caution; it is likely that the
diagnosis of cancer in the Japanese was significantly in error for some sites,
particularly in earlier years. This limitation was one of the reasons why the NAS
BEIR V committee provided risk estimates for certain organ systems but not for
individual organs within those systems. While the Committee recognizes the
necessity to make organ-specific risk estimates for situations involving internally
deposited radionudides that are not distributed uniformly in the body, the
Agency's documentation should make it clear that there are relatively larger
uncertainties for organ-specific risk estimates than for the estimated whole-body
radiation risks,
b) Dose Rate Effectiveness Factor(s)
The overwhelming majority of studies with both experimental animals and
with cells in culture have shown that in terms of lethality, mutagenesis, and
tumorigenesis, a given dose of low-LET radiation is significantly more effective
when administered at a high dose rate than when administered at a low dose rate
or when, administered as multiple small fractions over a longer period. This
observation is incorporated in most current radiation risk estimation procedures
through use of a Dose Rate Effectiveness Factor (DREF). Most DREFs
determined in experimental systems fall in the range 2 to 10 (NCEP report 64),
-------
Epidemiological studies have not produced data in support of a DREF for
carcinogenesis in humans except in tlie case of leukemia, for which the dose-
response is best described by a "linear-quadratic" model that in itself produces a
DEEP because the quadratic term becomes insignificant at low doses and at low
dose rates of low-LET radiation. The dose-response relationships for solid tumors
are best described by a linear model; it should be kept in mind, however, that
excess cancer risks at low radiation doses (e.g., less than 0.10 Gy) are generally
too small to be observed directly.
EPA has proposed to use a DREF of 2 for radiation-induced solid tumors;
that is, EPA will assume that low-LET radiation accrued at high doses and dose
rates (e.g., in the Japanese atomic bomb studies) is twice as likely to cause cancer
as the same dose accrued at low dose rates. The Committee agrees that this
choice is reasonable; it is consistent with current scientific judgment, and it is in
the lower range of DREPs observed in experimental systems.
EPA questions whether it should use a DREF of 1 (Le.t no dose rate
adjustment) for the observed risks of radiogenic breast cancer and thyroid cancer,
arguing that dose fractionation did not show much reduction in risk for breast
cancer in women exposed to repeated fluoroscopic x-rays. The Committee believes
that this observation does not rule out a dose-rate effect, because each fraction
was delivered at a high dose rate; it is not aware of any good date set for
induction of human breast cancer at low dose rates.' Similarly, the data on thyroid
cancer do not include exposures at truly low dose rates. Thus, the breast and
thyroid need not be treated differently from other organs with respect to their
response to low dose rates. Some scientists believe that it is more parsimonious to
argue for a DREF of 1 in humans lacking epidemiologic evidence for a dose rate
effect. It should be noted that a DREF of 1.0 (i.e., the absence of a dose rate
effect) represents the more conservative approach to low-dose and low dose-rate
risk estimation which might be viewed as a more prudent stance for a regulatory
agency in setting radiation protection standards. However, the Committee is not
prepared to reject a DREF of 2 or greater in all tissues, whether animal or
human, in view of the considerable evidence for a dose rate effect in experimental
systems. Also, by applying a DREF of 2 for most if not all organ-specific risk
estimates, EPA will be in harmony with other radiation protection groups
worldwide.
The Committee recognizes the potentially large policy implications of the
choice and application of a DREF since the great majority of environmental and
occupational iow-LET radiation exposures will occur at low dose rates. If the
Agency decides to apply a DREF, the Committee strongly suggests that EPA
define a boundary between low and high dose rates so that users of EPA risk
-------
coefficients will not inadvertently apply a DREF in situations for which it is
Inappropriate. *
c) Relative Biological Effectiveness (RBE) for Alpha Radiation
As EPA points out, the Relative Biological Effectiveness (RBE) to be used
for alpha radiation depends on the choice made for the DREF for low-LET
radiation, because alpha radiation risks per unit dose do not appear to vary
significantly with dose rate. Sufficient direct epidemiological data are available to
estimate the risk per unit exposure of high-LET alpha-particle-emitting
radionuclides for both leukemias and respiratory cancer. Hence, for the most
common cancers related to alpha-particle irradiation, no need exists to use low-
LET risk coefficients in combination with an RBE or Q factor (see footnote 1),
The Committee commends EPA for proposing to use high-LET risk information
directly (e.g., for the risks of exposures to radium and radon).
Where direct epidemiological data are not available to support such risk
estimation procedures, the Committee agrees that an RBE of 20 be used for alpha
radiation in conjunction with a DREF of 2 for low dose rates of low-LET
radiation. When comparing risk of alpha-particle radiation with risk from acute,
high-dose low-LET radiation, however, an RBE of 10 should be used.
Although not commonly required for EPA risk estimation tasks, it may be
useful for the Agency to provide a Q factor for neutron radiation, if only to assure
that the use of the Japanese epidemiology in the development of its risk
coefficients remains valid in the event of further adjustments to the atomic bomb
dosimetry (see below),
d) Uncertainty Analysis
The Committee is gratified by EPA's intent to estimate the cumulative
uncertainties in its calculated risk coefficients. This undertaking is crucial to
informed use of the risk estimates.
The uncertainty analysis should include the uncertainty in the dosimetry for
the Japanese atom bomb survivors which for individual doses could be as great as
30-45%. Furthermore, the whole set of dose estimates might be biased towards
the low side because the 1986 dosimetry may have discounted the neutron
In the <&m of low-LET radiation, NCEP Report No. B4 suggested upper boundaries for tow doa« of 20 rad
(0.2 Gy) and for low dose rate 5 rad (0,05 Gy) per y«ar. EPA may decide to select other bonntaxififi-
-------
component too much, A small increase in neutron doses would produce a larger
increase in dose equivalents (physical doses x Q factor) and reduce the risk
estimates for low-LET radiation correspondingly. Moreover, uncertainty analyses
should be conducted for the risk estimates for individual organs which, as pointed
out above, are much less robust than the risk estimate for total cancers.
In assigning an uncertainty to the DREF, the Committee recommends that
EPA also consider observations that the DREF might be larger than 2 as well as
those suggesting it might be as small as 1.
4. Are the recommended methods for estimating the cancer risks
appropriate and supportable in light of the current scientific
evidence?
Yes, with the cautions noted in the response to Question 3, above. The
Committee commends the Agency for its preparation of material to support its
proposed methodology and notes that the subject is not easy to cover m a few
pages of text,
The Science Advisory Board appreciates the opportunity to comment on
Agency's proposed methodology for estimating radiogenic cancer risks and looks
forward to receiving a summary of EPA's responses to the comments provide
above.
Sincerely,
L .
lond C. Loehr, phalrfian
Executive Committee
Science Advisory Board
)ddvar E. Nygaard, Clja|j»nan
Radiation Advisory Committee
Science Advisory Board
Enclosures: Committee Roster
Appendix I: List of Background Material Provided by the
Office of Radiation Programs
-------
U.S. ENVIRONMENTAL PROTECTION AGENCY
SCIENCE ADVISORY BOARD
RADIATION ADVISORY COMMITTEE
ROSTER
CHAIRMAN
Dr. Oddvar F. Nygaard, Case Western Reserve University, Cleveland, OH
MEMBERS
Dr. Stephen L. Brown, ENSR Consulting & Engineering,
Alameda, CA
Dr. Kelly H. Cliftoo, University of Wisconsin Clinical Cancer Center
Madison, WI
Dr. James E. Martin, University of Michigan, Ann Arbor, MI
Dr. Genevieve M, Matanoski, The Johns Hopkins University,
Baltimore, MD
Dr. H. Robert Meyer, C.N.SX, Harrisburg, PA
Dr. Richard G, Sextro, Lawrence Berkeley Laboratory, Berkeley, CA
Mr. Paul G. Voilieque, MJP Risk Assessment, Inc., Idaho Falls, ID
Dr. James E. Watson, Jr., University of North Carolina at
Chapel Hill, NC
CONSULTANTS AND LIAISON MEMBERS
Dr. Paul Deisler-Science Advisory Board's Executive Committee
Mr. Seymour Jablon—National Cancer Institute
Dr. Jan Stolwijk-Yale University and the Science Advisory Board's Indoor Air and
Total Human Exposure Committee
-------
Science Adviaory Board Staff
Dr. Donald G, Barnes, Drieetor, Science Advisory Board, U.S. EPA,
Washington, D.C.
Mrs. Kathleen W. Conway, Designated Federal Official, Science Advisory Board,
U.S. EPA, Washington, D.C.
Mrs. Dorothy M, Clark, Secretary, Science Advisory Board
U.S. EPA, Washington, D.C.
-------
APPENDIX I
Background Material Provided by the
Office of Radiation Programs
1. "Revaluation of EPA's Methodology for Estimating Radiogenic Cancer
Risks", background document presented to RAG in January 1992,
2. "Proposed Methodology for Estimating Radiogenic Cancer Risk", background
document presented to the RAC in May 1992.
3. Portions of 1990 Recommendations of the International Commission on
Radiological Protection. ICRP Publication 60. Ann. ICRP 21, 1991,
4, Land, C.E. and W.K. Sinclair, "The Relative Contributions of Different
Organ Sites to the Total Cancer Mortality Associated with Low-Dose
Radiation Exposure" Ann. ICRP 22, 1991.
5. "Health Effects Models Developed from the 1988 UNSCEAR Report", NRPB-
R226.
6. Gilbert, E.S., in Health Effects Models for Nuclear Power Accident
Consequence Analysis. NUREG/CR-4214, Rev. 1, Part II, Addendum 1}
Chapter 3, LMF-132,
-------
U.S. ENVIRONMENTAL PROTECTION AGENCY
NOTICE
This report has been written as a part of the activities of the Science
Advisory Board, a public advisory group providing extramural scientific
information and advice to the Administrator and other officials of the
Environmental Protection Agency, The Board is structured to provide a balanced
expert assessment of scientific matters related to problems facing the Agency,
This report has not been reviewed for approval by the Agency; and hence, the
contents of this report do not necessarily represent the views and policies of the
Environmental Protection Agency or other agencies in Federal government.
Mention of trade names or commercial products does not constitute a
recommendation for use.
-------
ABSTEACT
In a memorandum dated January 13, 1992, Margo T. Oge, Director, Office
of Radiation Programs, asked the Science Advisory Board to review EPA'a revised
methodology for estimating human cancer risks from exposures to ionizing
radiation. The charge for this review requested the SAB to respond to the
following four questions;
L Has the Agency analysis considered the most relevant risk estimates
of low-LET radiation?
2. Does the Agency analysis accurately compare the most relevant
features and assumptions of the various models?
3, Is the Agency's analysis technically sound?
4. Are the recommended methods for estimating the cancer risks
appropriate and supportable in light of the current scientific evidence?
In addition to the charge, the OEP initially provided the SAB with extensive
background material On May 1, 1992, ORP provided the SAB with a follow-up
document titled "Proposed Methodology for Estimating Radiogenic Cancer Risk."
In the opinion of the Radiation Advisory Committee EPA has reviewed and
considered all major new data sets and current risk estimates of low-LET ionizing
radiation. Although no single data set and model for predicting radiogenic cancer
risk is ideal, the method of analysis chosen by EPA is adequately supported by
present scientific evidence. A few areas of uncertainty exist that eventually may
require modification of the Agency's analysis when further data become available.
Among these is the method for utilizing ("transporting") risk estimates from the
atomic bomb survivor study in Japan where the base-line risks for several cancers
differ significantly from those in the U.S. Another is the question of whether to
apply a "Dose Rate Effectiveness Factor" (DREF) for solid tumors at low dose
rates or at low doses of low-LET radiation; the Agency's choice of a DREF of 2 is
in accord with the current choice of other radiation protection groups world-wide.
An additional concern is the continuing uncertainty in the dosimetry for the
Japanese atomic bomb survivors including the magnitudes of the neutron
components. Further discussion of these and other issues is contained in the
subsequent parts of this Letter Report.
KEYWORDS; Radiation Risk Assessment, Low-LET, Radiogenic Cancer Risk
-------
r
DISTRIBUTION LIST
Administrator
Deputy Administrator
Regional Administrators, Regions 1-10
Assistant Administrators
Director, Office of Radiation and Indoor Air
Director, Office of Science and Technology
Director, Office of Ground Water and Drinking Water
EPA Headquarters Library
NTIS Distribution System
13
------- |