&EPA
             United States
             Environmental Protection
             Agency
            Office of Health and
            Environmental Assessment
            Washington, DC 20460
EPA/600/9-90/031b
August 1989
            Research and Development
Presentation of Risk
Assessments of
Carcinogens
            Appendix B:
            List of Case Studies

-------

-------
  CONTENTS
APPENDIX B
                Introduction


                List of Case Studies and Ordering Information
                                                             9

                Excerpts from the Case Studies Illustrating the Attributes:


                     B. 1 General Attributes


                     B.2 Hazard Identification


                     B.3 Dose-Response Evaluation


                     B.4 Exposure Assessment


                     B.5 Risk Characterization

-------
INTRODUCTION
                 T
his appendix is intended to convey important background for risk
analysts. The report of the Study Group includes a discussion of the
desirable attributes of a risk assessment and refers to examples of
these attributes in the case studies. This appendix consists of excerpts
taken from the case studies to illustrate the attributes identified. The
reader may wish to refer to the complete case study for a more
comprehensive view of the context from which a particular example
was taken. Ordering information for the case studies is contained in
this appendix.
The appendix is organized according to the sequence of the report
discussion: i.e., general attributes and specific attributes for hazard
identification, dose-response evaluation, exposure assessment, and
risk characterization.
In selecting the case studies and the excerpts contained here, the
Study Group considered the nature of presentation but did not form
judgments on the conclusions reached by the authors of the assess-
ments. The reader should also note that the assessments were pre-
pared prior to mid-1987; thus, the information excerpted is not
necessarily current.  .

-------
                                                                                                          I
         RISK ASSESSMENT

               CASE STUDIES

ORDERING INFORMATION
                                  To obtain complete copies of the risk assessments analyzed by the'
                                  Study Group, the following organizations should be contacted:


                              A. DEHP (Di- [2-ethylhexyl] phthalate)

                                  Turnbull, D. and Rodricks, J.V. (1985). Assessment of Possible Car-
                                  cinogenic Risk to Humans Resulting from Exposure to Di (2-ethylhex.yl)
                                  phthalate (DEHP). J. Amer. Coll. Toxicol. 4, 111-145.

                                  Contact:   Joseph Rodricks
                                            Environ Corporation
                                            The Flour Mill
                                            1000 Potomac Street, N.W.
                                            Washington, D.C. 20007
                                            (202) 337-7444
                              B.  ("Dioxin") TCDD (2,3,7,8 - Tetrachlorodibeniodioxin I

                                  Environ Corporation (May 1987). Dioxin and Its Human Health
                                 .Significance. Prepared for the National Council of the Paper Industry
                                  for Air and Stream Improvement, Inc.

                                  Contact:    William J. Gillespie
                                            National Council of the Paper Industry for Air and
                                            Stream Improvement, Inc.
                                            260 Madison Avenue
                                            New York, NY 10016
                                            (212)532-9349
                              C. Ethylene Oxide

                                 Sielken, R. L. (1987). A Time-to-Response Perspective on Ethylene
                                 Oxide's Carcinogenicity. In Risk Assessment of Environmental and
                                 Human Health Hazards: A Textbook of Case Studies. Dennis
                                 Paustenbach, Ed., Wiley & Sons, New York (1988).
                                 Contact:   Robert L. Sielken
                                           Sielken Incorporated
                                           Suite 210
                                           3833 Texas Avenue
                                           Bryan, Texas 77802
                                           (409) 846-5175

-------
         RISK ASSESSMENT
              CASE STUDIES
ORDERING INFORMATION
                              D.  Formaldehyde
                               •   EPA (April 1987). Assessment of Health Risk to Garment Workers and
                                  Certain Home Residents from Exposure to Formaldehyde. Office of Pesti-
                                  cides and Toxic Substances, Washington, D.C.
                                  Contact:   TSCA Assistance Office (TS-799)
                                           U.S. Environmental Protection Agency
                                           401 M Street, S.W.
                                           Washington, D.C. 20460
                                           (202) 382-3790
                              E.  Formaldehyde
                                  OSHA (December 4, 1987). Occupational Exposure to Formaldehyde;
                                  Final Rule. Federal 'Register Volume 52, No. 233, pp. 46173-46237.

                                  Contact:    Lisa Odoms
                                            American Industrial Health Council
                                            Suite 300
                                            1330 Connecticut Avenue, N.W.
                                            Washington, D.C.  20036
                                            (202) 659-0060
                               I. Lead
                                  Wallsten, T.S., and Whitfield, R.G. (December 1986). Assessing the
                                  Risks to Young Children of Three Effects Associated With Ekvated Blood-
                                  LeadLevels. Publication of ANL/AA-32, Argonne National Labora-
                                  tory. Prepared for EPA Office of Air Quality Planning and Stan- "
                                  dards, Washington, D.C.
                                  Contact:   Les Grant
                                            Environmental Criteria and Assessment Office
                                            (MD52)
                                            U.S. Environmental Protection Agency
                                            3200 Highway 54
                                            Research Triangle Park, NC 27711
                                            (919) 541-4173

-------
         RISK ASSESSMENT

               CASE STUDIES

ORDERING INFORMATION
                         G/H.  Methylene Chloride
                                  EPA (July 1987). Technical Analysis of New Methods and Data Regarding
                                  Dichloromethane Hazard Assessment. EPA/600/8-87/029A.

                                                           — and—
                                  Update to the Health-Assessment Document and Addendum for Dichlorom-
                                  ethane (Methylene Chloride): Pharmacokinetics, Mechanism of Action, and
                                  Epidemiokgy. EPA 600/8-87/030A. Office of Health and Environ-
                                  mental Assessment, Washington, D.C.

                                  Contact:    Marie Pfaff
                                            U.S. Environmental Protection Agency
                                            401 M Street, S.W., (RD 689)
                                            Washington, D.C. 20460
                                            (202)382-7345
                              I.  Para-dichlorobenzene

                                  EPA (February 13, 1987). Exposure Assessment. In Assessment of
                                  Human Cancer Risks from Para-dichlorobenzene B-l-16. Office of Toxic
                                  Substances, Washington, D.C.
                                  Contact:   Risk Analysis Branch
                                           Office of Toxic Substances
                                           U.S. EPA                                      .
                                           401 M Street, S.W.
                                           Washington, D.C. 20460
                                           (202)382-3832
                              J. Red Dye No. 3 (FD & C Red No. 3; erythrosine)
                                 FDA (July 1987). An Inquiry Into the Mechanisms of Carcinogenic Action
                                 ofPD&C Red No. 3 and Its Significance for Risk Assessment. FD&C Red.
                                 No. 3 Peer Review Panel, Washington, D.C.
                                 Contact:   Office of the Director
                                          . National Center for Toxicological Research
                                         1 NCTR Drive
                                           Jefferson, Arkansas 72079-9502
                                           (501)541-4517

-------
         RISK ASSESSMENT
              CASE STUDIES
ORDERING INFORMATION
                            K. Tetrachloroethylene
                               Bogen, K.T., Hall, L.C., McKone, T.E., Layton, D.W., and Patton,
                               S.E. (April 1987). Health Risk Assessment of Tetrachloroethylene (PCE)
                               in California Drinking Water. Publication UORL-1583, Lawrence
                               Livermore National Laboratory. Prepared for the California Public
                               Health Foundation, Berkeley, California.
                               Contact:    Joseph Brown   .     • •
                                         California Department of Health Services
                                         2151 Berkeley Way
                                         Berkeley, CA 94704
                                         (415) 540-3191  .

-------
                      * B.I

GENERAL ATTRIBUTES
             Attribute \   The.'scope and objectives of the report are explicitly stated.
                 SOURCE  Case Study J. Red Dye No. 3 (Pages 1-3).
                     Note  After presenting contextual background, the authors clearly outlined
                            the objectives and scope on Pages 2 and 3 of the excerpt.

-------
           A REPORT BY THE FD&C  RED  NO.  3  PEER REVIEW FANEL
AH INQUIRY INTO THE MECHANISM OF CARCINOGENIC ACTION  OF  FD&C  RED  NO.  3
               AND ITS SIGNIFICANCE FOR RISK ASSESSMENT
                                     Prepared  by:

                                     Dr.  Ronald  W.  Hart,  NCTR/FDA  (Chairman)
                                     Dr.  Thomas  Burka,  NIEHS/NIH
                                     Dr.  Scan  C.  Freni,  CEH/CDC
                                     Dr.  Robert  Furrow,  CVM/FDA-
                                     Dr.  David W. Gaylor,  NCTR/FDA
                                     Dr.  Theodore Meinhardt,  NIOSH/CDC
                                     Dr.  Bernard  Sass,  NCI/NIH,
                                     Dr.  Elizabeth  K. Weisburger,  NCI/NIH
                                     Executive  Secretaries -
                                       Dr.  Paul  Lepore,  ORA/FDA
                                       Dr.  Angelo Turturro, NCTR/FDA
                                July,  1987

-------
                                                                   July,  1987
                            CHAPTER 1 - INTRODUCTION
 A.   The  Recent Use and Certification of FD&C Red No. 3 (R-3)
      FD&C Red  No.  3 (R-3)  is  a color  additive that is  permanently listed
 
-------
                                                                  July,  1987

of the Certified,Color Manufacturers' Association  (CCMA),  proposed  that the
FDA postpone the closing date for  the provisional  listings of  R-3,  and  take
no action  with respect  to  the  permanent listing  of  R-3 (9),  based on the
assertion  that  R-3 should.be  regulated using  the general safety  require-
ments, rather than the anticancer  clauses,  of  the  Act.
    The  main  points  of  the .petitioner's  submission can  be   summarized  as
follows:
    1)  The weight  of scientific  information  indicated  that  R-3 was not  a
        primary carcinogen  or tumorigen which  acted directly on  the genetic
        material to initiate a  tumorigenic  process;
    2)  R-3 appeared  to  be  a non-direct or  secondary  carcinogen  which acted
        indirectly to increase  the incidence of tumors;
    3)  Additional studies  were then currently underway  which  would further
        support  these conclusions;  and
    4)  If  the weight  of   scientific information  supported  the first  two
        points,  then  R-3 should  be considered  to act  through  a secondary
        mechanism  and not be regulated  similarly to  an  agent  acting through
        a  primary mechanism of  carcinogenesis.
    The  recent  White House Office of  Science  and Technology  Policy (OSTP)
consensus  document on  chemical  carcinogens   (10)  suggested   that  relevant
biological and  biochemical data,  including  information  from   studies  of
mechanism,  be  incorporated into evaluations of cafcinogenicity.  In accord
with  this,  FDA has  considered  that  it  may  need   to  use a  spectrum  of
approaches based on mechanism  to  determine the potential  health impact of
the use  of regulated  substances.

C.  FDA  Commissioner's  Charge  to the Peer Review Panel,
    Therefore,  the FDA  Commissioner directed  that  a Peer Review  Panel of
scientists with pertinent  training and expertise  be  constituted to provide
scientifically valid  answers  to the following  three questions:
     1)   Do the data  indicate a secondary mechanism  of  action with respect
         to R-3 and allow determination of any  potential risk  posed by human
         exposure to this color?
    2)   If not,  then 'what further studies  or analyses  are, necessary to
         resolve these issues  and provide an adequate basis for  risk assess-
         ment?

-------
                                                                   July, 1987

     3)  In  the  interim,  what  concerns  relative  to human  health, if  any,
         would be posed by continued use  of  this additive while such studies
         and analyses are conducted and evaluated?
     To accomplish the Panel's charge, pertinent scientific  information con-
 tained in the following  was  reviewed  and evaluated including:   1) the sub-
 missions by the various product sponsors  (FDA Color.Additive  Petition [CAP]
 96); 2)  the recent  peer review  of  the  R-3  risk assessments (4); 3)  the
 relevant  published  literature;  4) interviews  of  some  'of  the  principal
 researchers involved in  specific  scientific investigations; and 5) the  re-
 sults of additional studies presented to  the  Panel  by Drs.  S.  Schwartz  and
 S.  Irigbar.  It should be understood that most of these studies  were not  de-
 signed to examine direct effects  by R-3  on  the  target organ,  but,  instead,
 were specifically aimed at demonstrating an indirect mechanism of  interest.

 D.   Some Major Considerations in the Panel's Evaluation
              *
     It  is  worthwhile to discuss, briefly, five major issues which  impact on
 providing  the  answers to the Commissioner's  questions.
     First,  a  point  discussed   in  the FDA's  1973  Proposed  Regulation on
 Selenium in Animal Feed  (11),  and a precedent  set  in the  subsequent 1974
 Finalized  Regulation  (12)  concerning  carcinogenesis,  are of importance.
 The  preamble to  the  Proposed Regulation (11) states  that -
                                                                  i?
     "The various  anticancer  clauses contained  in the Act [the specific
     clauses  of the act  are noted in Footnote #1  below]  were predicated
     on  the  theory that,  since we do not know the mechanisms  of carcin-
     ogenesis,  even  one  molecule of a carcinogen should not  be allowed
     into the food supply."
 The  discussion continues  -
     "The anticancer clauses do not apply  in  the  case  of  an agent which
     [1]  occurs  naturally in  practically  all foods,  [2]  is used in a
    manner such  that  the natural  level in the food  is not  increased,
                                                                      '-
     [3] has a definite hepatotoxic  effect/no effect  level, and [4]  has
    a possible carcinogenic  effect which  is associated only  with  the
    hepatotoxic effect."
Footnote  #  1:  Sees.  409[c][3][A],   512[d][1][H],  706[b][5][B],  72  Stat.
   1786, 82  Stat  345,  74 Stat. 400;  21  U.S.C. 348[c] [3] [A],  360[d] [1] [H],
   376[b][5][B] .

-------

-------
                         B.I

GENERAL ATTRIBUTES
             Attribute 2  The report's content is laid out impartially, with a balanced
                             treatment of the evidence bearing on the conclusions.
                  SOURCE   Case Study  H.  Methylene  Chloride (Pages 71-87).
                      Note   This case study focused, to a considerable extent, on two alternative
                             approaches to dose-response evaluation. The attached excerpts are
                             illustrative of balanced treatment of the evidence. This balance was
                             generally apparent throughout the report.

-------
                   United Slates
                   Environmental Protection
                   Agency
Office of Health and
Environmental Assessment
Washington DC 20460
EPA/600/8-87/030A
July 1987
External Review Draft
                   Research and Development
EPA
                    Ujpdate to the Health Assessment Document

                    and Addendum for Dic±0.oromethane (Methylene

                    Chloride):  Pharmacokinetics,  Mechanism of

                    Action, and Epidemiology
                          Review

                          Draft

                          (Do Not
                          Cite or Quote)
                          (Permission  to
                          reproduce granted
                          by EPA)
                                          Notice
                    This document is a preliminary draft. It has"not been formally
                    released by EPA and should not at this stage be construed to
                    represent Agency policy. It is being circulated for comment on its
                    technical accuracy and policy implications.

-------
 8.3.   COMPARISON OF METHODS 1 AND 2
     .As  with any risk estimation that  involves extrapolation,
 many  assumptions affect  the actual risk  numbers, and Methods 1
 and 2  are  certainly no exception.  Both  use pharmacokinetic
 information  in  an.attempt  to reduce uncertainties  inherent in
                                *•                      .
 risk  assessments.   It is commonly, although erroneously,
 conceived "that  by incorporating  pharmacokinetic information into
 a risk assessment,  magical reductions  of uncertainity are
 achieved.  Actually,  the examination and application of
 pharmacokinetic data and models  for this compound  have revealed
 something  quite different.   As one becomes more familiar with
 developing and  using this  type of information, new and often
 more complex questions arise.  The utility of pharmacokinetics
 is, in fact,  this very point.  It allows for a  systematic
 analysis of  a chemical's disposition in  the body,  an important
 component  of the risk assessment.  In  applied-dose approaches,
 assumptions  are frequently made which, although sometimes based
 on empirical evidence, are often  inflexible and thus in error at
 some conditions of  the human exposure.   For example,.absorption
 fraction is.  frequently set at  some arbitrary value determined
 from some  empirical  evidence or  from assuming the  "worst case" of
 100% absorption.  Pharmacokinetic  modeling, when properly
performed, is able to  account  more logically and realistically
 for amounts  absorped on  a  time basis.  Pharmacokinetic models
seek to account  for  instantaneous  concentrations and changes in
those concentrations that  are  due not only to changes in exposure
                               71

-------
conditions, but changes in the physiologic responses as well.
     As observed in earlier discussions and in the HRAC (1987)
report, while many uncertainties are reduced,  several of the "old
problems" remain, and in fact, new challenges  arise.  Only with
continued work and trial applications will the science continue
to mature.  Only two possible methods have been applied here,  but
given the body of evidence and the development of the science,  at
this time these two possibilities are considered the most
reasonable.                                    «                  '
     Methods 1 and 2 employ many of the same assumptions, and yet
vary in some very significant ways.  Although the actual
calculated numbers are almost identical (within a factor of 4 for
liver and much less for lung), the methodologies are quite
different.  When in error even some of the common assumptions
have different implications depending on the method chosen.
However, the differences are mainly in the "last step" of the
risk assessment process, that is, how to actually use delivered
dose to calculate a risk number.
     A major and fundamental  assumption that EPA has made for
both methods is that the physiologically based pharmacokinetic
model used by Andersen et al. (1986, 1987) is a reasonable method
for describing and predicting the disposition of DCM and its
metabolites in human tissues.  This would include acceptance of
          9
the model's structure and input parameters.  The HRAC  (1987)
report raises several important questions that are  deemed
important, and future elucidation for purposes of methodology

                                72

-------
development will be necessary.  However/ for the present,  EPA has
applied the model with some minor changes.  It was felt that the
uncertainity raised by questions regarding model structure were
no greater than those raised by a conventional applied-dose risk
assessment.  In fact, because the model is able to quantitatively
describe numerous physiologic and biochemical processes, it is
highly probable that model structure questions pose less
uncertainty than the traditional approach.  The HRAC is less
certain about some of the input parameters, such as the metabolic
rate constants.  The consequences of errors in these could be
great, and the impact may be somewhat different depending on
whether Method 1 or Method 2 is employed.
     There are three major sources of uncertainity with the
metabolic scheme and parameters in the model.  First, the model
assumes that any carbon dioxide observed at low doses is being
produced from the MFO pathway.  The implication is that- the
carbon dioxide observed by several investigators at low doses is
still compatible with the assumption built into the model that
the GST pathway is virtually  inactive at low doses.  If incorrect
it would mean that the GST pathway is active at low doses, where
the model is predicting that  it is not.  Both methods would be  in
significant error in predicting risk at low doses
(underprediction).
     A second uncertainity common to both Methods  1 and 2  are the
values of the input parameters which apportion metabolism  by  both
pathways between the liver and lung.  As  discussed in  the  HRAC

                                73

-------
 (1987)  report,  there is  great concern-over the values estimated
•for these parameters.  The pharmacokinetic model is quite
 sensitive to these parameters,  and thus any error would be
 reflected in model predictions.   Such error would be significant
 in both methods (over- or underprediction).
      The third  uncertainty regarding metabolism is with the
 values  of the metabolic  rate constants.  As discussed in the HRAC
 (1987)  report,  most questions remain with  the value determined
 for the first-order rate constant for the  GST-mediated pathway.
 Determined  by allometeric scaling and "curve fitting" of the
 model to exposure chamber data,  the value  of this parameter is
 uncertain.   In  fact, data from  CEFIC  (1987b) indicate that the
 value selected  by Andersen et al. (1986, 1987) is in error.
 Significant questions also remain regarding the methodology and
 results of  the  CEFIC experiments. It is reported that
 experiments are being conducted which may  reduce some of the
 uncertainty with regard  to this rate  constant.  Method  1 is more
 sensitive to this parameter  than is Method 2;  thus, any error
 would result in greater  error  in the  risk  number calculated using
 Method  1.  However, because  of  this  sensitivity,  if the value  of
 this parameter  is established more  accurately, Method  1 would
 better  reflect  the impact of such findings.
      As described previously,  numerous  uncertainties  are
 associated  with the assumptions that have been made to develop a
 risk assessment using pharmacokinetic information for DCM.   Apart
 from the major generic difference between Methods 1 and 2,  there

                                 74

-------
is also the uncertainty associated with the metabolic rate
constant of the GST-mediated metabolic pathway.  The risk
calculation resulting from Method 1 would be greatly lowered by
significant reductions in the estimate of this parameter.
     Also, one might want to allow for some possible minor role
of metabolism by the MFO-mediated pathway in the carcinogenic
process.  No current evidence suggests a contribution by this
                                                         «
pathway, and a large role is ruled out by the very low tumor
response in the National Coffee Association drinking water  *
bioassay (NCA, 1982a, b; 1983), in which the MFO metabolism was
saturated at a level similar to those associated with the highly
tumorogenic exposures in the NTP inhalation bioassay (NTP, 1985,
1986).  However, even a small contibution to DCM's
carcinogenicity by MFO metabolism at high doses might have a
meaningful impact on low-dose risks, since the proportion of the
dose metabolized by this pathway increases at low exposure
levels.  One might, for example, hypothesize that 5% of the
carcinogenic risk to mice at the bioassay doses resulted from MFO
metabolism.  (Any higher contribution begins to conflict with the
observed lack of correlation of MFO metabolism—and clear
correlation of GST metabolism—with tumor incidences.)  Such a
small contribution to tumor production by MFO metabolites would
not have a major impact on the human risk estimates as they have
currently been calculated using Method 1.  If, however, the human
GST metabolic rate constant were greatly reduced from the present
estimate, resulting.in much lower predicted human risk from this

                                75

-------
pathway, that same hypothetical 5% contribution from MFO
metabolism would have a far greater impact on the total human
risk estimate at low doses.  Neglecting a contribution of as
                                     t
little as 5% by the MFO-mediated pathway towards carcinogenesis
would, under those circumstances, greatly .underestimate the risk.
If new data indicate that human GST activity towards DCM is much
less than the estimate used here, then a reevaluation of the
     «
assumptions would be necessary.  More confidence in the
assumption that the MFO path does not contribute to
carcinogenicity and greater certainty in the values of the
appropriate metabolic parameters will be required before the
concomitant reduction in the risk estimates would be accepted as
appropriate.
     Another question that arises, regardless of method, is upon
which organs are risk estimates to be based?  The pharmacokinetic
approach gives information regarding specific organs.  Site
concordance of tumor production between animals and humans  is not
normally assumed in performing risk estimates.  It is not clear
how to extrapolate for the entire human  (all organs) when risks
have been calculated for specific organs by using pharmacokinetic
knowledge.  One possible solution is to select the organ with the
highest risk number and apply  this to the  whole body.  This could
result in an overestimation of the risk for many organs  but would
ensure that no underestimation would occur due to a  lack of
knowledge about an oversensitive or a highly metabolic tissue.
Alternately, if individual organ risks occur  independently,  they

                                76

-------
could be mathematically summed.  Both Methods 1 and 2 could do
either of these.  However, in the case of DCM, because of the
comparative insensitivity of Method 2 to the GST metabolic
                                              *
parameter, even organs with several fold greater metabolic
activity than the lung would not be expected to have a risk far
different from that calculated here.  The results of Method 1,
however, are more difficult to apply to other organs.  A several
fold change in the GST level (as might be observed in other
organs) would result in a different value for the risk number.
without knowing, specific GST activities towards DCM in other
tissues, it is difficult to ascertain the impact of such an
uncertainty.  Although there is no clear evidence of
carcinogenicity in organs other than the lung or liver, there
are some findings that raise concern about this issue.  Benign
mammary gland adenomas and salivary gland tumors developed in
rats (NTP, 1985, 1986).  The HRAC (1987, Chapter 6) discussed
pancreatic tumors in workers exposed to DCM.  Although these
tumors may not be significant,  some note should be taken.
     It is clear that, once estimates or measurements of internal
dose at the sites of toxic action are obtained, many difficult
issues must be resolved as to how to use such data in the
extrapolation of risk from experimental animals to humans.  The
problem is not confined to DCM, nor does it result from any
shortcoming in the information on the pharmacokinetics of this
compound.  It is a general problem, reflecting the lack of •  >-
understanding of the pharmacodynamics of carcinogenesis.

                                77

-------
     As discussed earlier, there are many.difficulties in using
metabolic differences in species to modify a carcinogenic risk
assessment.  Extrapolation between species involves many factors,
including metabolism and pharmacokinetics.   The ability to
elucidate a species difference in one contributing component does
not necessarily indicate what, if any, adjustments should be made
to the overall extrapolation.  It does not necessarily provide
more certainty than the empirical process currently used; in
fact,: making the necessary assumption may introduce new
questions.
     Method 1 advocates the adjustment of the applied-dose risk
extrapolation by the degree to which humans  (at lower doses) and
the bioassay rodents metabolize different proportions of their
applied doses at the internal site of carcinogenic action.  In
the present case, this method leads to a risk reduction of 8.8-
fold from  the level estimated in EPA's previous applied-dose  risk
assessment,  in Method 1, the observed pharmacokinetic ,
differences between species  are to be compared with those
expected to emerge as a result of differences  in  physical.size
and the rates of physiological processes  in rodents and humans.
     Method 2, which leads to a risk  reduction of 2.1-fold,
advocates  the adjustment  of  the applied-dose risk extrapolation
only by the degree to which  the proportion of  the applied"dose
that is metabolized differs  from high human doses to  low human
doses; any species difference in the  proportion  of the dose  that
is metabolized  is  ignored as a basis  for determining  human

                                78

-------
carcinogenic potency.  Instead, the interspecies component of
extrapolation"" is carried out as would be done if using applied
dose.  The reasoning is that, in addition'to the effect of
species differences in metabolism, there are expected (but
unknown) differences in the carcinogenic responsiveness of the
tissues to a given delivered dose.  Pharmacokinetic data
illuminate only the metabolic differences.  It may be, for
example, that greater sensitivity to carcinogens in humans
"compensates" for lower metabolic activation of the applied dose.
The justification for using the surface area correction on the
applied dose during the species-to-species extrapolation rests on
tradition.  Empirical comparison of carcinogenic potencies in
humans  (determined directly from epidemiologic data) with those
from experimental animals shows the surface .area scaling  .       •
relationship to be a reasonable estimator for many compounds,
although other chemicals show potencies that differ from the
expectation based on this relationship by orders of magnitude.
     Because of the problem of specifying interspecies
differences in tissue responsiveness to carcinogens, both Methods
1 and 2 can only give a relative adjustment to the applied-dose
calculation of human risk.  That is, incorporation of
pharmacokinetic information can only raise or lower the "dose
delivery" .component of interspecies extrapolation relative to its
appearance in the applied-dose procedure, while the component.
representing "responsiveness" or pharmacodynamic differences
between experimental™ animals and humans remains problematic and
                                79        '             »

-------
continues to be based on assumptions retained from the former
         «*                        •                       ' '
applied-dose procedure.  Methods 1 and 2 differ*, chiefly in the
way that assumptions from the applied-dose extrapolation
procedure are retained when data on the pharmacokinetic component
are available.
     Method 1 is based on the conclusion that, given the most
reasonable scaling of key physiological variables across species,
delivered dose is expected, a priori, to be the same proportion
of applied dose in rodents and in humans.  That is, differences
in body size and physiological rates between rodents and humans
do not, in themselves, lead to an expectation of differences in
the delivered doses of metabolically activated carcinogens.  If
the proportion of a dose that is metabolized is in fact the same
across species, then the applied dose serves as a good surrogate
measure for the delivered or internal dose of a carcinogen, and
both dose measures will result in the same risk extrapolation.
Thus, the surface area correction, as traditionally used in the
applied-dose procedure at EPA, corresponds to an assumption about
(and correction for) interspecies sensitivity differences rather
than about metabolic differences.  It is the  factor by which
human risk is assumed to exceed mouse risk for a given dose
(applied or internal).  This same assumption  about relative
sensitivity is retained when a pharmacokinetic analysis of the
proportion of a dose that is metabolized replaces the prior
assumption of equality of dose delivery across species, implicit
in the use of applied dose in extrapolation.

                                80

-------
      In the present case,  for example,  the surface area
 correction between mouse and human doses is a factor of 12.7.
 According to the model used by Andersen and Reitz,  the species
 difference in metabolism is such that humans (at low exposure
 levels)  metabolize about one-ninth as much of their applied dose
'via the GST pathway as do mice at 2000  or 4000 ppm.   (Most of
 this difference is due to high- to low-dose differences that
 result from the saturation of the competing MFO pathway at the
 high bioassay exposures experienced by  mice—the interspecies
 difference at the same applied dose is  quite small.)   According
 to  Method 1,  the lower metabolic activation of DCM  in humans
                     e
 implies  that the carcinogenic potency difference between humans
 and mice is only one-ninth as large as  it was previously thought
 to  be, before the metabolism data were  available.  The
 carcinogenic potency in humans (expressed in units of applied
 dose)  is only one-ninth as large as the value based  on applied
 dose.
     Method 2,  in contrast,  suggests  that no reasonable
 assumption  can be made about the effect of allometric scale on
metabolic differences  among species.  Under this view,  any
magnitude of  species difference  in metabolism seems  equally
probable a  priori, and so  there  is  no prior assumption against
which to compare  empirical  data  on the  actual  difference.
Instead, it  is presumed  that,  for a given dose level,  the
combined effect of metabolic  and  sensitivity differences is given
by the surface area correction on applied dose.  No  explicit

                                81

-------
assumption about the species difference in sensitivity is made;
in fixing the magnitude of the combined effect,  however,  a value -
of the sensitivity component is assumed implicitly.   For example,
in the present case the pharmacokinetic model estimates that
humans at high doses metabolize 4.5-fold less of their delivered
dose in the lung and 1.5-fold less in the liver than do mice at
equally high doses.  By assuming.that the overall interspecies
factor is 12.7, Method 2 implicitly assumes that these metabolic
deficits are compensated for by greater human sensitivity of
57.2-fold in lung  (1/4.5 x  57.2 = 12.7) and 19.1-fold in liver
(1/1.5 x 19.1 * 12;7).  Low-dose human risks are adjusted by the
                                        i *
degree to which the proportion of the applied dose that  is
metabolized via the GST pathway is different than at these high
human doses.  That is, delivered dose is  used only for the
extrapolation within species, where the question of  interspecies
difference in sensitivity does not arise.
     Thus, the  crux of the  difference between the two methods  is
whether  or not  a  reasonable prior assumption about the -expected
species  differences  in metabolism can  be  made before
pharmacokinetic data  are available.   If a prior expectation can
be specified, when pharmacokinetic  data become  available, one may
replace that 'assumption with data (which may show the assumption
to have been inappropriate for that compound).   The same
 assumption about species differences in sensitivity is applied in
 all cases.   If, on the other hand,  no prior expectation about
 pharmacokinetic differences between species can be specified,
                                 82

-------
 there is no way to know whether the observed differences are
 bigger or smaller than usual.  The applied dose is therefore
 used to extrapolate across species, and the sensitivity
 assumption is adjusted to make its combined effect with the
 observed metabolic differences come out to be equal to the
 surface area correction, since it is assumed that the combined
 effect scales in this way.
      The choice between Method 1 and Method 2 has not been an
 easy one,  and has been made only after considerable debate and
 discussion both within EPA and with representatives of other
 federal  regulatory agencies.   The attributes of each method that
 have been  considered include  their relative conservatism in the
 face of  uncertainty,  their sensitivity to  errors in the
 underlying assumptions and estimates  of the pharmacokinetic model
 used by  Andersen  and Reitz, their correspondence to previous
 practice,  their ability to incorporate current  understanding of
 metabolism, however  imperfect,  into the risk extrapolation
 process, and/ of course, the plausibility  of the assumptions upon
 which they are founded.
     EPA concludes that Method  1, which extrapolates risk across
 species and from high  to low doses based on  the  amount of
metabolism of DCM by the GST pathway,  is the most advisable  basis
 for use of current pharmacokinetic information.  The evident
 importance of differences in metabolism among rats, mice, and
hamsters to DCM's carcinogenic potency  in these species makes the
use of metabolic differences desirable in the estimation of human

                                83

-------
risk.  While acknowledging that many factors in addition to
pharmacokinetics influence species differences in carcinogenic
potency, EPA concludes that it is reasonable to modify risk
extrapolation from experimental animals to humans by the degree
to which the species manifest different degrees of metabolic
activation of their applied doses at the site of carcinogenic
action.  The absolute  levels of human risk that.are estimated
remain uncertain, as always, owing  to the lack of knowledge about
the  contributions of the other, non-pharmacokinetic factors to
the  relative  carcinogenicity of DCM in  rodents  and human's.  The
need to retain assumptions about  the role  of such factors should
not, in EPA's opinion, dissuade us from examining the potential
 contribution of'such factors as can be experimentally examined.
 The choice of Method 1, the choice of 'the GST pathway as the sole
 route to carcinogenic activation, and the choice of. the model
 used by Andersen .and Reitz as a means of its estimation have been
 made because, in EPA's  judgment, they represent the most likely
 and plausible interpretation of the data at hand..  Each  choice  is
 made  in the face of some uncertainty, and the interpretation of
 the resulting estimate of the  carcinogenic  potency of DCM in
 humans must be tempered with the knowledge  that further data may
 lead  to other choices and different risk  estimates becoming  more
 defensible.
       The  unit risk for continuous inhalation of 1 ug/m3 of  DCM is
 thus estimated as  4.7 x 1CT7.   For comparison,  the applied-dose
  extrapolation leads to a value of 4.1 x ICT* (which  is 8.8 times

                                  84  ,

-------
 higher), and the use of the same metabolic data, but
 extrapolating to human risk using Method 2, results in a value of
 1.8 x 10-6 (which is 2.1-fold lower than the applied-dose method
 and 3.8-fold higher than Method 1).
      It should also be noted that both Methods 1 and 2 are
 presented as modifications of the method of risk extrapolation to
 humans commonly employed by the EPA and CPSC,  that is, with the
 surface area correction used as an interspecies correction
 factor.   If one instead uses^the body weight basis for defining
 equally risky applied doses in animals and humans (as is done by
 the FDA),  then the estimated human risk by all methods would be
 12.7 times lower.   That is, the applied dose procedure would lead
 to a unit  risk estimate of 3.2 x lO'7,  while modifications of .
 this unit  risk by  accounting for metabolism would yield unit
 risks of 3.7  x 10~8  for Method 1 and 1.4 x 10~7 for Method 2.
 These numbers are  12.7-fold,  8.8 x 12.7 = Ill-fold,  and 2.1 x
 12.7 = 26.7-fold,  respectively,  below the published EPA unit risk
 of 4.1 x 1CT6 based  on  applied doses scaled by surface area.
 Andersen et al.  (1986,  1987)  and Reitz  et al.  (1986)  argue that,
 because interspecies differences in metabolic  and physiologic
 parameters have been accounted for by the pharmacokinetic  model,
 there  is no longer a need  for  any interspecies  correlation
 factor.  Implicit in this  view is  the assumption  that  the
 interspecies correction  factor on  applied dose  is used solely to
 account for species  differences  in metabolism,  and that metabolic
differences completely account  for differences  in carcinogenic
                                85

-------
potency of a compound in animals and humans.   The EPA takes  the
view that this is not the case, since it ignores the contribution
of non-pharmacokinetic factors that influence a species'
responsiveness to a given internal dose.
     Lifetime extra risks over background from continuous and
constant low-level exposure to DCM may be estimated by
multiplying the vapor concentration by the internal unit risk
value.  However, the EPA's analyses of the model used by Andersen
and Reitz indicate that," if vapor concentrations exceed 100 ppm
or so  for any part of an exposure/ substantial  nonlinearities
begin  to appear that tend to  invalidate the  assumptions allowing
the unit risk to be used.  Under such conditions the MFO pathway
begins to show  saturation, resulting in disproportionally more
DCM being available to  GST metabolism,  which results in
disproportional increases, in internal dose.   Exposures involving
high  vapor  concentrations can have estimated risks that  are
 several fold above the levels implied by the "equivalent"  time-
 weighted average exposure.   The reader is also reminded that the
 unit risk assumes a breathing rate of 20 m3/day.  Occupational
 exposures,  or other exposures occurring during more-strenuous-
 than-average activity, will consequently have  risks somewhat
 underestimated.
      Although EPA feels that  it is warranted to use species-to-
 species pharmacokinetic and metabolic  information to  adjust
 estimates of human risk based on animal data,  the absolute levels
 of estimated human risk remain uncertain, owing to the unknown
                                 86

-------
 contribution of species differences in sensitivity to a given
 internal  dose of carcinogen.   EPA recommends  that  intensive
 efforts be  made to develop information' on  the pharmacodynamics of
 carcinogenesis that could be  used in the risk assessment process
 in the future.   One approach,  which may elucidate  the magnitude
 and variability of. the  pharmacodynamic  factor for  various species
 comparisons,  is to  obtain pharmacokinetic  information in both
 animals and humans  for  known human  carcinogens.  This would allow
 an implicit determination of the pharmacodynamics  for humans
 relative to various rodent species,  since the  contribution of
pharmacokinetics and the  relative potencies of applied doses
could be estimated from available data.
                               87

-------

-------
                        B.I
GENERAL ATTRIBUTES
             Attribute 3  The risk assessment presentation includes a description of any
                            review process that was employed, acknowledging specific
                            review commentary.
                 SOURCE   Case Study E. Formaldehyde (Pages 46237-46247). .
                     Note   These excerpts summarize public review of regulatory alternatives
                            followed by response by the regulatory agency (OSHA). The review
                            comments reflect a mixture of scientific review and commentary on
                            regulatory impacts.

-------
Friday
December 4, 1987
Part  I!
Department of  Labor

Occupational Safety and Health
Administration
29 CFR Parts 1910 and 1926

Occupational Exposure to Formaldehyde;
Final Rule

-------
          Federal Register  /  Vol. 52.  &o. 233 / Friday. December 4. 1987  / Rule* aad Regulations
                                                                      48237
the apparel industry an average of 30
workdays were lost in 1985 for each
case of skin disease reported to the
Bureau of Labor Statistics.
  The incidence rates of reportable
occupational illness related to skin
disease or disorders for 1985 was 10.9/
10,000 full-time workers in textile
finishing (except wool), SIC 2260, and
4.4/10.000.in the apparel industry, SIC
23. These rates are substantially above
low risk industries, such as finance,
insurance, and real estate, where the
rates is 1.0/10,000 full-time workers.
  OSHA has no estimate on the
incidence  rate of formaldehyde-induced
skin disease in the workplace. However,
about 4 percent of alfpatients tested in
allergy clinics were found to be  .
sensitized {Exs. 42-75; 42-93]. If the
percent of employees.who would
become sensitized to formaldehyde
without the use of personal protection is
similar to  the incidence seen in allergy
clinics, then provision of protective
clothing to 269,700 employees exposed
to formaldehyde would prevent 10,790
cases of allergic dermatitis annually.
Moreover, numerous cases of non-
allergic skin irritation, which would not
have consequences as serious as  allergic
dermatitis, would  also be prevented (see
OSHA's Regulatory Impact Analysis).
OSHA believes that the risk of skin
diseases and disorders among workers
who come into contact with
formaldehyde solutions and
formaldehyde-bearing solids is
significant and that this risk will be
substantially reduced by the new
standard for formaldehyde.
  In a few cases, information was
sufficient to quantify  the risks of skin
diseases and disorders in specific
industry sectors. These cases could be
used to determine the likely benefits of
full compliance with the revised
formaldehyde standard since improved
work practices and housekeeping along
with the use of adequate protective
clothing and equipment should virtually
eliminate skin diseases and disorders.
  Available literature on dermatitis
(both irritation and sensitization) was
generally derived  from "problem areas"
where NIOSH HHEs have been
requested or where investigators  were
examining known hazards. The
information available is generally on
skin disease of a serious nature and
does not distinguish between irritation
and sensitization.  There are five studies
of textile dermatitis [Exs. 77-11; 78-84:
85-20; 85-23; 85-24). Incidences ranged
from 3  to 58 percent of those
administered queetionnaries with some
correlatioa between incidence rates, the
amount of releaseafale formaldehyde ia
the fabric, and airborne ct>ncerrtratkrn»
of formaldehyde. (Heat and humidity
also influence release of formaldehyde,
as does the age of the doth and whether
or not it has been washed.) Twenty-six
percent of 874 workers were affected.
Airborne levels ia all bat one case were
below QJj ppm indicating a need to
provide protection from dermal contaST
regardless of the airborne level of
formaldehyde.
  There are also reports of adverse
health effects in persons who handled
tissue-preserving solutions {Exs. 42-ltJl;
42-123; 73-86D; 78-53]. Dermatitis, either
irritant or sensitizing, was  found in 28 to
37 percent of these workers. (The effects
described were more severe than those
generally seen in the apparel workers,
even though most wore gloves some of
the time.) A total of 47 cases in 140
individuals was reported, for an overall
incidence rate of 33.8 percent, well
above figures reported in the overall
occupational setting and well above the
incidences observed In dermatology
clinics [Ex. 42-75]. Available literature
indicates that the incidence of asthma
and bronchitis, as well as the incidence
of dermal Irritation and sensitrzation,
may be quite high in this group (see
Health Effects section).
  Summary of the Significance of the
Risk: OSHA has determined that the
existing standard for formaldehyde
poses significant risk to employees of
cancer, sensory irritation, dermatitis,
and asthma. Full compliance with afl of
the provisions of the revised standard (1
ppm TWA, 2 ppm STEL, and aaciHary
provisions) will substantially reduce
these risks to a level that can be viewed
as safe for the worker. These findings
have been made using the approach
which OSHA has used in setting other
standards for'toxic substances siace the
Benzene decision and are consistent
with the formaldehyde ruleraaking
record.

VIII. Summary of Regulatory Impact and
Regulatory flexibility Analysis

  Executive Order 12291 (48 FR 13197,
February 19,1981) requires that a
regulatory analysis be conducted for
any ijile having major economic
consequences on the national economy,
individual industries, geographical
regions, or levels of government. In
addition, the Regulatory Flexibility Act
of 1980, 5 U.S.C. 601 et seq., requires
OSHA to determine whether a
regulation will have a significant impact

-------
46238    .   Federal Register / Vol. 52. No. 233 / Friday. December 4,1987 / Rules and Regulations
on a substantial number of small
entities.
  Consistent with these requirements.
OSHA has prepared a Regulatory
Impact and Regulatory Flexibility
Analysis (RIA) for the formaldehyde
standard. This analysis includes a
profile of the industries that are covered
by the standard, an estimate of the
number of exposed workers, a review of
the nonregulatory alternatives, and
assessments of the technological
feasibility, costs, benefits, and overall
economic impacts of the final standard.
This RIA is available at the OSHA
Docket Office.
  Based upon an analysis of the record,
OSHA has determined that the
industries affected by the revised
standard can be grouped into three
classes, according to the potential
exposure levels. Tier One, which covers
approximately 36,000 affected
establishments and approximately
412,000 exposed workers, consists of the
industries where some firms have
workers who are currently exposed
above either the 1 ppm PEL or 2 ppm
STEL. This group is comprised of
foundries, laboratories, funeral homes,
and industry sectors engaged in the
manufacture of: (1) Hardwood plywood.
(2) particleboard, (3) fiberboard, (4)
furniture, and (5) formaldehyde resins.
Tier Two, which includes approximately
29,000 affected establishments and
approximately 1.1 million exposed
workers, consists of the industries
where some firms have workers who are
currently exposed between the 0.5 ppm
action level and the 1 ppm PEL and
where no firms have employees exposed
above either the 1 ppm PEL or 2 ppm
STEL This group is comprised of textile
finishing and industry sectors engaged
in the manufacture of: (1) Apparel, (2)
formaldehyde, and (3) plastic molding.
Tier Three consists of 24 industries
where some workers are currently
exposed above 0.1 ppm and where no
employees are exposed above the 0.5
ppm action level. This group covers
approximately 47,000 establishments
and approximately 676.000 workers.
Table 5 presents OSHA's estimate of the
number of affected establishments and
employees for each of the affected
industry sectors. Establishments are
grouped by the highest exposure found
within the establishment.
   Based on the rulemaking record,
 OSHA has determined that compliance
with  the revised standard is
 technologically feasible. Exposures in
 the Tier One industries can be reduced
 to below the PEL and STEL through the
 increased use of ventilation (either local
 or general), and the subsitution of low
 emitting urea formaldehyde resins
 (LEUF). No industry representatives
 contended that such controls could not
 achieve these limits. Personal protective
 equipment (e.g., gloves, goggles,
 respirators, etc.), monitoring badges, and
 hygiene equipment (e.g., eye wash and
 emergency shower) are also readily
 available. Medical resources will not be
 a problem because the medical
 surveillance questionnaire, of which an
 example is provided in nonmandatory
 Appendix D to the standard, may be
• administered under the supervision of,
 and not necessarily by, a physician.
 Also, the medical exams, when required,
 consist of standard tests. Finally, the
 requirements of several provisions (e.g.,
 recordkeeping, the development of an
 emergency plan, and employee training)
 can be met using in-house personnel.
   OSHA estimates that the annual cost
 of the revised standard will be'
 approximately $64.2 million. Medical
 surveillance ($12.9 million), and the
 installation, maintenance and operation
 of the required engineering controls in
 Tier One ($27.6 million) are the two most
 expensive provisions, accounting for 20
 percent and 43 percent of total annual
 costs, respectively. Table 6 provides a
 breakdown of industry compliance costs
 by exposure level of individual
 establishments.
   OSHA has determined that
 compliance with the revised standard is
 economically feasible for firms in each
 of,the industry sectors based upon  an
 industry-specific analysis of revenue
 and profit ratios. Table 7 displays total
 annualized compliance costs for each
 industry as a percentage of annual
 industry sales and profits. The annual
 compliance costs are less than one-half
 of one percent of revenues in all sectors
 expect for the fiberboard sector, where
 this ratio is 1.65 percent. Impacts of this
 magnitude will not adversely impact the
 viability of most firms in these
 industries. Where data^allowed. OSHA
 also examined profit levels of the
 industries affected by the standard and
 determined that in all but a few of  the
 wood product sectors, costs would not
 exceed 3 percent of profits. For those
 industries, improved growth spurred by
 the reduced value of the U.S. dollar will
 permit the industry to pass through
 some of the costs of compliance to the
 purchasers of these products.

-------
        Federal Register / VaJ.  52.  No. 233 / Friday. December 4,  1987 / Rules *aA Regulations
                         TABLE 5.—NUMBER OF AFFECTED ESTABLISHMENTS .AND EMPLOYEES
SIC and industry
4Wr3M WWi CxpoMrn From 0.1 1o Above 1:0 ppm)
ifV\ — -Hardwood otvwood 	 ...,......-„.,.--•-,,—„.,„..,.-, 	
7nr(A 	 	 	 , 	 ,
2BI9— Qy Jostruction
822— Veterinary anatomy 	 _ 	 : 	
Subtotal 	 	 	 	 	
Total. 	 	 _ 	 _„ 	 	 	 	

Establishments
Total
30S
54
14
10.003
440
4J004
14.000
15.128
43JB49
723
24.391
868
11,653
37,455
81,404
250
43
299
222
296
1.461
1«9
1,441
152
330
683
1.439
374
474
179
263
415
226
-433
1,635
3,589
10,50O
22,575-
19
47,537
128,941
Above 1
PPM
40
14
9
184
35
1.047
2,137
0
3,466
0
0
0
0
0
3,486
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
. 0
t)
fl
t)
o
0'
0
0
0
3.466.
Between
0.5-1
PPM
€£
1fi
5
2£4£
31
1.43S
.1,861
0
6,060
6«5:
5,73?
1«
i.ooe
7.43ft
13^08.
e
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
13,408
Between
0.1-0.5
PPM
94
16
0
2,645
31
520
8,1«7
15,000
26,473
0
17,211
33
4,000
21,244
47.717,
290
43,
299
222
296
1,491
189
1,441
152
330
. 683
4,439
37*
474
179
263
415.
22S.
433
1.655.
3.389-
10,300
22,575
19
47,537
95,254
Total
affected
200
46
14
5/474
97
3,002
12.1«5
15,006
35.998
«84
22.948
49
5,000
28«H£
44:660
250
43
299
222
296
1,491
«9
1.441
152
330
663
1.439
374
474
179
263
415
226
433
1,655
3,589
tO.500
22.575
19
47,537
Employees
Produc-
tion
Employ-
ees
• 16,000
6,100
1.537
462,014
'32.800
171,800
'42,000
>7«,02»
828,277
42,900
960,000
99,500
490,000
1,532.400
2.349,677
31.100
12.800
HX3.106
43,000
19,000
67,400
•K.OOO
27,600
•6.300
8,700
W.900
88,100
17,000
21 .SCO
15.500
20,300
34.900
14,100
32.300
11.200
71.742
31.500
28,950
38
676,330
112.217J3.017.007
Above 1
PPM
455
381
230
14)31
385
4,509
8,269
0
13,180
0
0
0
0
0
14^80
0
0
9
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
13,180
Between
0.5-1
PPM
1374
1.440
588
23.223
980
12.169
5.951
0
45,925
19,125
117,663
480
10,000
147,268.
183,193
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
193.193
Between
0.1-0.5
PPM
8.699
2436
335
235.095
8,335
43322.
24,441
so.eae
368.OC3
10,298
823,637
3,401
90,000
927,336
1.280.39*
31,180
12,800,
100,100
43,080"
19,009
67,4081
16.000:
27.600
6.300
9.700
10,900
23,100,
17.000-
21,800,
15,500
29*300.
31,900-
ia.too,
32,300"
11,200
71,742
31,500
28,350
38
676,330
1,956,72*
ToW
«KpOS3d
10.728
4.577
1,153
259.349
9.700
60
tsaaee
ajteo
&3flO
9.MS
to^oe
23,»«)
WiQflO
2i.eco
15^00
saaao
J1*00
18.100
aomin
IJ^QO
7U42
3t50U
28^50
38
676,330
4,1S3,«M
1 Figure* represent total employees.
* Pathology, nistology and gross anatomy labs.
» Some eettMsnments -exceed -STEL.
Source: U.S. Department of Labor. OSHA, Office of Reflulatory Analysis.

-------
46240
                                    "                                                       J*

Federal Register / Vol. 52. No. 233 / Friday,  December 4. 1987 / Rules and Regulations



                     TABLE 6.—COMPLIANCE COSTS BY INDUSTRY AND PROVISION
InduUry
With Exposure* Abovt 1
PPM
Hardwood plywood . 	 ..........
Parfcleboard ...„._ 	 ........ 	 ..
Ffcerboard. 	 . 	 . 	
Furniture ™..«™..™™.™........™..
Rwir* ,„„„.„..„„.... 	 . 	
Foundries:.«w».,.....»..,,,..,..,,...,.
Labor atories .™....™.^._.._™....
Funeral services... 	
SUbtOUJ. .„...:......„... 	
Wttti Exposure* Between
03 and 1 PPM
Hardwood plywood 	
PH\&QtXWd....,~ ....................
Fiberboard 	 . 	
Furnfojre Mn.«.HMMH...H.......u,m.
RO$k^H«MM»,
-------
           Federal Register /  Vol.  52,  No. 233 ./  Friday.  December 4.  1987 J Rules  and RqgiAaitiena     4S241
                       TABLE 7.—COMPLIANCE COSTS AS A PERCENTAGE OF REVENUES AND PROFITS
.SIC and industry
H»i<**ulS*»aat





006. flQ7 LnborntoafiS ... 	 «M...............,..,_........_.,.,T..-t,,,-T[, T 	 	 - 	 - 	

226— 3«KU« filling 	 ~ 	 	 - 	 	 .-. 	 	 -..- 	 -.-..-. 	 	 	 	 	
*rj— .^uppral -r - . 	 ...... ..„_, 	

3O7JB"~PI0fttic moWBnfl 	 	
2*38— Softwood ptywxxJ

CZI Pipit milli .. i. ,., ,,.....„.,.,.
2531 — Papftrboard nulls -.. 	 	 	
2642~Envotopos. j.-r.i... .. ... .........U..... 	 ,,,...^..,, ... ^ ! ^ _. _a . _. _m .
2653— Corrugated and wlid "fiber boxoo . 4... ......„,, ...... ... . ..
2866— Cyclic crudes, dyos and pigments 	 	 	 	 	
2051 Paints piornonts 	 •• 	 ...


2891 Adhosivos and Malantfl .„.„.« 	 .„......._.«... 	 *.....« 	 -. 	 ,..- 	 .'. 	 	 	

3291 Abraaiv® products 	 - 	 -. 	 « 	 '.

TTM MinttiJ wool JnauWian .„,».„ , . . ..„,),,„.,,.. ,
1t634 fl^rtric homowartu ind tent •

38M f
-------
46242     Federal Register /  Vol. 52.  No. 233  /  Friday.  December 4. 1987 / Rules and Regulations
approximately S5.7 million. Further, the
use of personal protective equipment by
about 270.000 workers, in conjunction
with other regulatory requirements, such
as medical surveillance and training, is
estimated to prevent about 11,000 cases
of dermatitis, which will provide
approximately $35.5 million in annual
savings. Table 8 presents OSHA's
estimates of the number of employees
exposed to various levels of
formaldehyde, the reduced risk of
cancer provided by the standard, and an
enumeration of respiratory and
dermatitis benefits derived from the
standard.

IX. Environmental Impact

  The National Environmental Policy
 Act of 1969, 42 U.S.C. 4321 et. seq.,
 requires OSHA to determine whether
 this standard will have a significant
 impact on the environment.

   Formaldehyde, by nature, is volatile
 and does not remain soluble in water for
 any length of time. Consequently,
 formaldehyde is not considered to be a
 potentially hazardous wastewater
 effluent. Similarly, formaldehyde in
 waste products such as sludge would
 vaporize rapidly, thereby having the
 potential to  affect air quality more than
 waste disposal. In any event, OSHA
 does not expect that the proposed action
 would significantly increase the amount
 of waste containing formaldehyde.

   Although more emissions will

              TABLE 8
 theoretically be removed from the
 workplace to the outside environment,
 because of the nature of formaldehyde
 and its ability to dissipate rapidly, no
 incre'ase in the amount of formaldehyde
 in the ambient air is anticipated. It is not
 expected that compliance with the'
 revised standard .will add significantly
 to the levels  of airborne formaldehyde
 that already exist in the environment as
 a result of automobile exhaust, cigarette
 smoking, and emissions from fabrics
 treated with formaldehyde-rbased resins.
 pressed wood products and other
 sources. Similarly, it is believed that the
 levels of ozone (a formaldehyde
 decomposition product) in the ambient
 atmosphere,  will not increase
 significantly as a result of this action.
                               A. Cancers Avoided by Reducing the PEL to 1.0 ppm

                            [Maximum Likelihood Estimate (MLE) and Upper Confidence Level (UCL)]
Industries
Hardwood Ptywood 	 „ 	 .„ 	 	 	
Pofticteboard
Rbwtooi/d ..,.....,„..........._.,...

KOSiO&.*.«*iM*<.H* <«.««•.«•*»••.•*•»...•..«•...•.•...».....•........«.««.
Fooodrioa............... ._.„.._,...„....... 	
Labor* too«» .....!..._....._...-..„....„......._.,................
Total
Cancws Avoided:
3-sUg» (MLE)......... .„ 	 	 	 .....
-.5-slifl9 (UCL)..........™..,...., 	 "i™"!.!"™.
Employees exposed at different levels
Parts Per Million
Total
455
301
230
1.031
385
4.509
_6,289
13,180
1.25
455
255
164
•873
192
2,116
2,070
6.125
2.16
0.08
8.11
1.75
0
46.
66
158
193
1.058
1,437
2,958
3.36
0.18
7.89
2.25
0
0
0
0
0
463
690
1,153
2.91
0.22
4.68
2.75
0
0
0
0
0
232
288
520
2.44
0.25
2.88
3.50
0'
0
0
0
0
,320
230
550
5.37
0.79
4.49
4.00
0
. 0
0
0
0
320
1,554
1.874
27.40
4.98
19.45.
Estimated cancers avoided
3-stage
(ML6)
. 0.16
0.14
0.13
0,49
0.29
12.00
30.41
43.62
43.62
'8.48
47.51
5-stage
(MLE)
0.01
0.01
0.01
. 0.02
0.01
1.60
4.83
6.48
5-stage
(UC*>
0.60
0.46
0.39
1.58"
0.77
14.72
28.98
47.51
     B. OTHER ANNUAL BENEFITS

   (Relited to PPE, Medical Surveillance.
            TraWng, etc.]

Rttpinitocy — „..
DtrmattJs 	
Total — ;„,.
cases
avoided
5,911
10,780
16,70t
vahMOf
benefits
5.650.844
35,469.359
41,120,203
  OSHA concludes that, as a result of
this action, there will be no significant
impact on the general quality of the
human environment outside the
workplace, particularly in terms of
ambient air quality, water quality, or
'solid waste disposal. No comments
made at the public hearing or submitted
to the record contradict this conclusion.
X. Summary and Explanation of the
Standard
  The following sections of the
 preamble discuss-the individual
 provisions of the final standard for
 formaldehyde. Each section includes an
 analysis of the record evidence and the
 reasons underlying OSHA's adoption of
 these individual regulatory '
 requirements.
 Overview.
   The final standard applies to all
 Occupational exposures to
.formaldehyde, including those resulting

-------
             Federal Register  /  Vol. 52. No. 233 / Friday. December  4. 1987 / Rules  and
                                                                                                                 46243
  from formaldehyde gas, solutions, and
  materials releasing formaldehyde. The
  standard contains two permissible
  exposure limits-(PELs) for formaldehyde:
  an 8-hour time-weigh ted average (TWA)
  exposure limit of 1 ppm and a 15-minute
  short-term exposure limit (STEL) of 2
  ppm.
    Engineering and work practice
  controls are the preferred methods of
  compliance, respirators are required
  where engineering and work practice
  controls are not feasible and respirators
  are necessary to achieve compliance
  with the PELs. In addition, employers
  must perform employee exposure
  monitoring, unless they can demonstrate
  by means of objective data that such
  monitoring is not necessary in their
  workplaces. Other requirements in the
  standard include provisions to establish
  regulated areas and medical
  surveillance programs. The final rule
  specifies how affected employers can
  comply with OSHA's generic hazard
  communication standard (29 CFR
  1910.1200). The final rule also specifies
  how affected employers can comply
  with other existing OSHA standards (29
  QER 1910.132, Protective equipment; 29
 CFR 1910.133, Eye and face protection;
 and 29 CFR 1910.134, Respiratory
 protection) that may be relevant to
 workplaces where employees are
 exposed to formaldehyde.
   Several provisions have been
 modified from- the proposed standard in
 response to testimony and comments
 submitted to the record, the availability
 of new data and information, or the
 need to clarify language. The reasons for
 these changes are described below. The
 requirements of the final rule are
 considered both necessary and  •
 appropriate to provide adequate
 protection to employees exposed to
 formaldehyde.

 Proposed Regulatory Alternatives A and"
 B

  In the. preamble to the proposed rule
 [50PR 50468-69], OSHA noted that the
 regulatory options before the Agency
 were broad, ranging from a change in
 the PELs. for formaldehyde in 29 CFR
 1910.1000, Table Z-2, to adoption of a
 full health standard with the provisions
 found in standards for carcinogens. The
 preamble indicated that OSHA's choice
 of regulatory strategy would depend on
 the Agency's  decision to regulate
 formaldehyde as an irritant, a
 carcinogen, or as both. OSHA invited
 the public to comment on "the entire
spectrum of regulatory possibilities"
before the Agency.
  If OSHA adopted the Alternative A
strategy—regulation of formaldehyde as
an irritant—the preamble stated that "it
  might be sufficient to lower the
  permissible exposure limits for
  formaldehyde contained in Table Z-2
      * and to rely on existing sections of
  the general industry standards, such as
  §§1910.132,1910.133,1910.134 and
  1910.1200, to provide the supplementary
  coverage needed for employee
  protection" [50 FR 50468].
    OSHA also stated that if available
  evidence was sufficient to conclude that
  "formaldehyde is an animal carcinogen
  and should be treated for regulatory
  purposes as a potential occupational
  carcinogen," Alternative B would be
  appropriate [50 FR 50469]. Proposed
  Alternative fr recognized
  formaldehyde's role as a potential
  human carcinogen, skin sensitizer and
  irritant, and eye and respiratory system
  irritant. Proposed Alternative B was a
  comprehensive OSHA health standard,
  containing PELs and provisions for
  employee monitoring, medical
  surveillance, protective equipment and
  clothing, hygiene facilities,  and hazard
  communication.
   The final rule regulates occupational
  exposures to formaldehyde on the, basis
  of formaldehyde's potential
  careinogenicity, as well as its ability to
  aet as a strong irritant and sensitizer of
  the skin, eyes, and respiratory system
  (see the Health Effects section, above).
   Many industry participants favored
 lowering the PELs for formaldehyde
 through adoption of Alternative A [Exs
 77-2; 77-15; 77-33; 77-34; 80-1; 80-4; 80-
 5: 80-6; 80-15; 80-18; 80-23; 80-24; 80-26;
 80-30; 80-38; 80-40; 80-43; 80-47; 80-55;
 80-58; 80-59; 80-61; 80-63; 80-64; 80-67;
 80-68; 80-69; 80-71; 80-72; 80-76; 80-77;
 80-78; 80-79; 80-80; 80-83; 80-85; 80-86;
 80-89; 80-100; 80-132; 80-261; 80-303; 86-
 7; 140; 150; 171; 201-9; Tr. 5/13/86, p. 62:
 Tr. 5/13/86, p. 137]. The reason most
 often presented to explain why these
 rulemaking participants preferred
 Alternative A to Alternative B was that,
 in their .opinion, occupational exposure
 to formaldehyde presented no risk  of
 cancer [Exs. 77-2; 80-1; 80-5; 80-6; 80-
 30; 80-58; 80-61; 80-68; 80^-71; 80-72; 80-
 76; 80-77; 80-83; 80-85; .80-86; 80-89].
 Typical of these comment* was the
 statement submitted by John T. Barr,
 Manager of Air Products and Chemicals:
  We support regulatory alternative A:  We
 do not betjeve that current occupational
 exposures to formaldehyde present a risk of
 cancer to humans, but we believe that proper
 steps need to be taken to assure that this safe
 condition continues (Ex. 80-1, p. 3].

  Some participants felt occupational
exposures to formaldehyde at current
leVels present only a risk of "minor"
irritant effects [Exs. 77-33; 80-43; 80-55:
  80-59; 80-67; 80-80; 80-303; Tr. 5/13/86.
  p. 66].
     Some participants believed
  Alternative A was attractive because it
  would be easy to implement [Exs. 77-34,
  80-5; 80-23; 80-26). For example, Leon J
  Mamch, Technical Director of the
  Louisiana-Pacific Corporation, a
  manufacturer of medium-density
  fiberboard, endorsed the adoption of
  Alternative A because, in his view, it-
  would "protect the public and serve as a
  workable approach for industry" fEx
  80-5, p. 1].                     l
    The costs associated with Alternative
  B's ancillary provisions, i.e.,
  requirements such as emergency plans,
  employee monitoring, and hazard
  communication, were emphasized in the
  submittal of Norman Newhouse. an
  Illinois lumber dealer. Mr. Newhouse
  favored adoption of Alternative A
  because thecosts of Alternative B were
  likely to be "considerable" [Ex. 80-132].
    Other commenters favored a simple
  revision of the Z-2 table limits for
  formaldehyde because they felt that
  other OSHA standards, such as the
  generic hazard communication standard
  and §§1910.132.1910.133, and 1910.134.
  already provided formaldehyde-exposed
  workers with adequate protection {Exs.
  80-47; 80-71; 80-79; 140].
   Michael Farrar, Vice President of the
  American Paper Institute and the
  National Forest Products Association.
  trade  associations representing the pulp,.
 paper, paperboard. and solid wood
 industries, urged the adoption of
 Alternative A [Ex. 80-63, p. 6] so that
 formaldehyde would not have to be
 treated as a carcinogen under OSHA's
 generic Hazard Communication
 standard (29 CFR 1910.1200). However.
 as  is discussed elsewhere in more detail.
 formaldehyde clearly should be
 regarded as a potential human
 carcinogen (see Health Effects and
 Significance of Risk discussion above).
 Employers will have to treat
 formaldehyde in accordance with the
 Hazard Communication standard's  '
 requirements for carcinogens since it
 meets  the criteria set out in that
 standard. Regulating formaldehyde only
 as an irritant  would be grossly
 inconsistent with the record and with
 the intent of the Hazard Communication
 standard.
  Four principal beliefs appear to
 underlie the preference of several
rulemaking participants for the adoption
of Alternative A: (1) Employers should
be allowed the maximum amount of
discretion in determining conditions in
their workplaces: (2) formaldehyde is
not a carcinogen; (3) formaldehyde's
transient irritant effects do not pose a

-------
46244     Federal  Register I Vol. 52.  No. 233  /  Friday. December 4.  1987 / Rules  and
significant risk of material health
impairment to«Kpt»eand Signtficanoe-of Risk portions
of this preamble, fcowever, (he evidence
in OSR/Vs rococd clearly .establishes
that formaldehyde fx»ses a significant
riskof •uUeristl topainaent of health at
the old PELs of 3 ppw 48-honr TWA), 5
Mxn (oeiJing), -and M ppni (pea?k). In
addition, evidence in the record
indicates that merely reducing -airborne
crpttmr. as proposed in Alternative A,
would not address the rick of skin
diseases aari c*har irri trtion .and
scnsitizatioo caused by dermai contact
with -fotatridfihyde.
   The evidence in Ae record as a whote
does cot «Mpport Ae adoption of
Alternative A. fa .short, the Agency has
food AafcJSj Some esnfdoyer* have not
exercised Appropriate discretion to
protect iaeir-eHyiloyee*; {2}
formaldekjwb is a potential human
carciw^erc 43) &rn»WeJ^de-HKteced
irritation aad aeasiiiralioH can
comtitute material ianwirmest of feeahh;
andfd) acoBBjireberrtwe OSHA
 standard i» .needed to jpnotect workeis.
 OSHA kwodacted Alteaatrws A
 because the Agenqr beKeses inat a
 chfiaga JH 4be PELs alone will not
 adequately protect the iseal&»f
 forroakieb yde-«xpased waiters,
                    .a certified
  Several rulemaking participants
pointed out lhal formaldehyde meets
established criteria for classification as
a potential carcinogen. For example. Dr.
Michael Silverstein, Assistant Director
of the Health and Safety Department,
UAW. asserted that "formaldehyde
musttre regulated as a presumptive
human carcinogen" because
"indisputable animal carcrrmgemcity
data * * * [are] sufficient to «nder
such a ludfmenT [Tr. 5/M/86, pp. Wl-
200). Ms. Jamie Osihen, Project
Coordinator -af the Ihrited Fwniture
Workers of America, also urged OSHA
to promulgate a comprehensive
standard because formaldehyde is a
carcinogen [Tr. 5/1S/98, p. 2j. Margaret
Seminario, Associate Oirectw of the
AFL-ClO's Department of.Occupa«0Kal
Safety, Health, and Social Security,
reported fca* ihe APL-CJO ooaeiders
formaldeiiyde a potential occupational
carcinogen aoader terms of OSHA"*
cancer policy [29 . « tNIOSH); Tr. 5/14f«6. p. 196
view that formaldehyde «kauUl he
handled as a.carcinc|gefl in occupational
envirosmenU. ftHOSH reoonaoejided
that formaldehyde expawwe "be
controlled to the lowest feasible luait"
on the basis of its rJ
  '•urnittire Workers -of Araerrca)].
                           *" as a
 ,carcin«gen £Ex. 77-U..p, 3j. JSOOSH
 submitted these comments in fulfillment
 of its atalutoxy jespojosibilities «nder the
 OSH Act, Which xeijuire the Institute .to:
   dea/eJsp criteria deoihifwith to«jc
 matariak aod han»iul phyaical afento ^nd
 substances which will describe exposure
 levels that are safe for various periods of
 emjftoyment, indlodhigljut not limited to the
 exponire levels at Which no -employee -wffl
 suSer .imptntsi be riifc «r innctional
 capaottiM -
50.jip.a-y.
   In -addition, OSHA's review erf tbe
record .Eodicatei that seduction of (be
exposure limits in Table £-<2 Jilaae
weiM mot provide protection «gaimt
dermal sensiiizatwm ao4 •tteer nan-
canQBT.tfects. OSMA bases tfai«
conckiBfcm, in part, «n 'the itafltiknony -of
expemb at flie xulemalnngaieariiig. fw
example, Ei>eard A. &Mnc4t, MJD.,
DiMctor^f the Center *w Occupational
and Enraonmentad Heal* «ft ? ofais
Hopkins University, stated tfeafc
   Regulatory Ahernalive A *  * * would
ignore or discount Oil joiner -effects «ff
 formaldehyde terin«ng-burns to fl>e eye,
 senaHiartJan-df *e fldn, tH'ttatSwofthe skm,
 and indeed tite carcS»ogen5cfty fif
 formaldehyde in experimental •ni
 has  led to a presampttonift* ff« a
 h«amn«cra«p«tiomd -carcinogen.
                                       does JMt oratnm «iy,ynorai»i«H»to decieaM
                                       skin ej^potwae, ani J»ri»cwe«rcartHjil»f«ir
                                       levels of -fesnaMehyde wiH jiave 4»Hlei«r m
                                       effect on protecting ibe AiB.an-direot-ooBtocI
                                       with sources of fQJTOaldeijrde [TX.5J&/M. ff.
                                       68-69].
                                           t&flf\ fJWTD *»OTC»t«»J »,»»**»*»•. — •*» —
                                        comments wgBrtling the prapos«d
                                        regulatory alternatives, andfcas, 4n
                                       * addition, weighed ihe «vi4asee aa Jbe
                                        health effects associated with
                                        workplace exposures to formaldehyde.
                                        The Agency concludes, .that only-a
                                        comprehensive standard can decrease
                                        the risks associated with aH of fhese
                                        adverse iieatth effects; a-regulatory
                                        alternative that simply lowers fhe PEL
                                        cannot protect workers against the
                                        broad range of formaldehyde-induced
                                        health effects.
                                          The ancillary provisions contained in
                                        a Mllieatth standard ace eapeciaHy
                                        appropriate in the regulation of An
                                        irritant and sensUizer such as
                                        formaldehyde. For example, training and
                                        personal protective equipment pro vide
                                        important worker protection when
                                        irritants, which by definUiox ane
                                        corrosive in action, inflamiqg the meiet
                                        mucous BurJaces .of the body JEx. 7J-a7a,

-------
              Federal Register /  Vol.  52.  No. 23* / Friday. December 4. 1987 / Rules  and Regulations
                                                                          46245
   P- 20], are being handled in the
   workplace. The inclusion of such
   requirements is.consiftent with section
   6f b)(7} of the OSH Act, which specifies
   that:                  .

    Any standard promulgated under this
   subsection shall prescribe tht uae of labels or
   other appropriate forms of warning as are
   necessary to insure that employees are
   apprised of all hazards to which they are
   exposed, relevant symptoms and appropriate
   emergency treatment, and proper conditions
   and precautions of safe use or exposure.
   Where appropriate, such standard shall also
   prescribe suitable protective equipment and
   control or technological procedures to be
   used in connection with such hazards and
   shall provide for monitoring or measuring
  employee exposure al such locations and
  intervals, and in such manner as may be
  necessaiy for the protection of employees. In
  addition, where appropriate, any such
  standard shall prescribe the type and
  frequency of medical examinations or other
  tests which shall be made available, by the
  employer or at his cost, to employees
  exposed to such hazards in order to most
  effectively determine whether the health of
  such employees is adversely affected by such
  exposure.

   Based on the record as a whole, the
  Agency concludes that regulation of
  formaldehyde as a potential human
  carcinogen, an irritant, and a sensitiaer
  is both necessary and appropriate to
  protect workers' health and functional
  capacity. The best available evidence,
  discussed above in the health effects
  section, clearly dictates promulgation of
  a final rule that will addresa
 formaldehyde's potential adverse
 effects, including skin, eye, and
 respiratory irritation: dermal and
 pulmonary sensitization; and cancer.

 Paragraph (a)—Scope and Application
   Like the proposal, the final standard
 applies to all occupational exposures to
 formaldehyde, including those in general
 industry, maritime, and construction.
 The language in paragraph (a)(l) has
 been altered from that of the proposal to
 clarify that the standard applies to a
 single, specific chemical entity,
 formaldehyde, with a Chemical
 Abstracts Service Registry number of
 5O-00-O.
   Employee exposure to formaldehyde
 can occur in a number of ways. For
 example, formaldehyde production
 workers-are potentially exposed'to
 formaldehyde gas; workers who handle
 formaldehyde solutions.
 paraformaldehyde. or materials
 containing resins or glues that have
 releasable formaldehyde may also be
 exposed to gaseous formaldehyde.
These solid and liquid products may
also present a risk from dermal contact
with formaldehyde. The final standard
   is designed to protect against all
   hazards from formaldehyde exposure,
   regardless of. the source. (Of couse, the
   standard does not necessarily protect
   against all of the hazardous ingredients
   in a mixture).                        -
    The Agency's determination that all
   occupational exposures should be
   covered by the final rule is" consistent
   with evidence in the record that the risk
   of formaldehyde exposure is related to
   the degree of exposure rather than to the
   operation, workplace, or segment of
   industry in which such exposure occurs.
   OSHA's position on the appropriate
   scope of the standard is unchanged from
   the proposal, and the Agency received
  only a few comments on this subject.
    The proposed rule explicitly stated
  that OSHA intended the standard to
  apply to the construction industry. The
  proposal also noted that OSHA's
  Advisory Committee on Construction
  Safety and Health had requested that
  OSHA develop a separate standard for
  formaldehyde for the construction
  industry or that, as an alternative
  measure, the Agency adopt "the most
  protective standard available" [50 FR
  50414],
   Despite OSHA's request for
  information on how the standard should
  be modified to consider the unique
  characteristics of the construction
  industry [50 FR 50413], OSHA received
  very little information on this industry
  and has reached and has reached the
  conclusion, based on the Agency's
  evaluation of the types of construction
  jobs known or suspected of having some
  potential for formaldehyde exposure,
  that the impact on this industry is small.
 For.example. Scott Schneider, an
 industrial hygienist with the United
 Brotherhood of Carpenters and Joiners
 of America testified that there are
 several construction jobs where there is
 potential formaldehyde exposure,
 including laying floors in a confined
 space when glues containing
 formaldehyde are used, blowing of urea-
 formaldehyde foam insulation into
 walls, and cutting and sanding wood
 products containing  formaldehyde [Tr.
 May 14.1988, pp. 137-138]. It is apparent
 that in the past construction workers
 could have received  substantial
 exposures to formaldehyde. However,
 negative publicity surrounding the use of
 UFFI has virtually eliminated this use of
 formaldehyde. Recent evidence on the
 generation of "particleboard aerosol" by
 a sanding process conducted under
 laboratory conditions indicated all
 airborne formaldehyde concentrations
below 1 ppm [Ex. 201-5A], showing that
under the more typical intermittent
exposure situation during actual
carpentry, exposures should be well
   below the action level of 0.5 ppm. With
   the exception of industrial construction.
   whch can also be hazardous because of
   the presence of formaldehyde from the
   operation, there appears to be little
:'  impact that the formaldehyde standard
   will have on the construction industry.
   OSHA's analysis of the data available
   suggests that most construction
   activities result in worker exposure well
   below 0.5 ppm. To the extent that there
  •are any unique operations, such as
   construction-related maintenance, or an
   increase in the use of formaldehyde-
   releasing resins in a confined area, the
   general industry standard is being
   applied to construction {see 1910.19).
    A representative of the maritime
  industry, Hal Draper, observed that:
  it would be impractical if not impossible for
  me marine cargo handling industry to meet
  the requirements of the proposed
  formaldehyde standard due to the uae of
  .casual labor, and mobile work sites with
  remote possibility of employee exposure to
  formaldehyde [Ex. 80-53, p. 2J.

    OSHA has chosen not to write a
  separate standard for the construction
  or maritime industries, both because
  commenters did not suggest any specific
  modifications to the standard to adapt it
  to the mobile worksite environment
  typical of these sectors and because the
  Agency believes that the final standard
  is flexible enough to present few
  compliance burdens for employers in
  these sectors.
   The standard has been tailored so
  that certain provisions become
  inapplicable or have only limited
  applicability when employee exposure
  is low. For example, routine medical
  surveillance is triggered by employee
  exposures above the action leVel. Thus,
  the standard is more stringent where
  employee exposures to formaldehyde
  present higher risk and becomes more
  flexible in situations where exposure
  and  risk decrease. The final rule has
  been structured so that compliance
  burden imposed by the standard is
 directly to the potential hazard posed by
 occupational exposure to. formaldehyde
 in each particular employment setting.
 The Agency believes  that, because of
 this approach, no significant compliance
 burden will be imposed on construction
 or martime employers whose employees,
 in general, are exposed to formaldehyde
 only  at concentrations believed by
 OSHA to be well below the PELs.
   In the proposal. OSHA expressed the
 intent to fully cover laboratory uses of
 formaldehyde under the standard [50 FR
 50470]. Further, it was proposed that the
 limited coverage of laboratories under
 Hazard Communication be enlarged in
 the formaldehyde standard so that

-------
482*6     Federal Rogigtec / Vci. 52.  No. 233  /  Friday.  December 4. 1987
                                                                                       and
laboratories using focmaldebyde would
be subjected to all -of the Hazard
Communication proviitons rather -than
Ihc limited provisions foand ia 28 CFR
more fully in ihe Hazard •Corarausicatioa
section of Ihe txamnsy and explanation,
OSHA has .decided not to enlarge the
hazard commuckatian coverage of
laboratories ID this final rule.
  The Standard Oil Company, which
has several production and 'laboratory
facilities where formaldehyde is used,
requested «pecia4 •consideration of
laboratories. According -to Standard Q&
  The StawlardOtt Company does not
believe ikai laboratory uecekplBces should be
subjected to Ihe same req»pemen)a Mother
workplaces within the scope of the proposed
standard.
  In t»ntr«9t t« typical manufacturing
locations, wo* practices, quantities of
formaldehyde handled and exposure controls
are vasUy-fiffferent for a laboratory. Small  *
quanUtkrt'of formaldehyde ere-u«d irnnost
quality oonlrd an3 research laboratories; 1he
formaldehyde is handled ty highly trained
technicians *od chemists: .the .exposure to
formaldehyde, is usually below sensory
irritation levels: the duration ol (he work tasV
Is short. wrasUy fastmg only a matter of
minutest and tome Ufcs require handling
fornwldcfcs, de in « 3«&»tsttty -exhaust hood.
Based cat MtsamsJl jonsuntf«f fenoaMehyde
usad.She concentration of-fQaxeMebydeiin
Iheiabiiocd odtwiat is mUirMl^Ex-ao-SB.
pp. 2-3 of attached caaunentsj.
   In its ieaaripticrL Staatdszd Oil
presents the case of a typical iatotatory
where 4he rets of any oae cheaacal, im
this case formaldehyde, is-very
incidental and a minor part iof ihe
overall exposure potential Xheoe are
undoukJ*dly many tach chcutnatances
where fscmaldehyde mofariiocre. in SMOOT
quaoiifieB.jrre used as one of masjy
reagents .or wfar^re very sas^l amaxrds of
 formsiWefcyde JBIC peesent at
prcsernttives. l&ese are psecisah/ the
 circuctatances that tke Texic Substances
 in Laboratories proposal (SI f£ 26860,
 July 24, t0a£} «±leinpiad te address;
 namely the aoacmtiEte use of small
 amounts cif amccj»as toxic .substances.
   OSHA is Trmdtni «f tirepotBntial for
 overlap 'between rrmcediEss leqnired in
 laboratories by the focmaldehyde
 standard scd the yrocedtngs imrtar the
 laboratory «tandard. la feaHring ihe
 laboratory -staadaud, OSHA will make
 every effort to assure lhat Acre ace cot
 conflicts or dtmlicartive resnurementB.
   However, OSHA has Identified one
 laboratory use of foraaaidebyde where
 the seventy  of the exposures to
 employees have tended to be even
 greater ifcan -the typicaJ ex^nMures tbtct
 occur withia §eaer*l industry. This is
 the tue of aoltkisas otataming
 formaldehyde tie preserve tisane and ihe
subsequent handbag of «och (tissues.
Exposed employees are laboratory
workers and teachers in histology,
pathnksgy, and anatomy Laboratories.
  Evidence submitted to OSItA's record
[Exs. 78-20; 78-54; 85-29; 91; 128J dearly
confirm ihat work in Bach laboratories
may result iniamthoe exposure to
formaideByde, boA by inhalation and
by dermal contact. Furthesmane,
numecous examples of exposure to
extremely high airborne concentrations
of formaldehyde were found {Exs. 42-85;
42-9ft 1E5-17: ITS-IB], as was •evidence
of formaldefayde-inducari irritation and
skin diaocdexs «nd senBhizaition
                       «; 85-28].
                                       rea
                                                    .
                                       These jtrafalems were Strand '.despite the
                                       very high levdl jrf training and education
                                       of some of the indmdu«l< who were
                                       being expoeefl.
                                         While many tidk^ laboratories have
                                       installed adequate engmeeruig ccmtrok
                                       and leqDtDe -their employees to observe
                                       good work practices, trthers have not
                                       done so. Dr. Melvin First, of the Harvard
                                       School of ftibiic Healtii *nd an expert
                                       witness in iatdustniail iiygMne and
                                       engineerkig contasls, testified "tiist 17
                                       percent of a group of 637 personal
                                       samples from hospital iabaralories were
                                       above&e .8-iour TWA oft ppm. in Or.
                                       First's opinion, these results showed
                                       that a "a^niScant nainber -of hospital
                                       laboratoides are using poe shelter accommodated
                                        under Ibis final Tale wn Occapationai
                                        Expcmre io Formalrieiryde .or the more
                                        general Toxic .Substances in
                                        LabGrstories mie wown it is
                                        promtdgaded. listsetavett, any category «f
                                       * laboratory thai as e»enu«ally exempted
                                        from the Toxic Sdsstaaces io
                                        LaberstarifiB «tBsndand wail
                                        automatically be covered by tfei«
                                        standard to the caokeat -there axe
                                        occupational expcaares tto
                                        formaldehyde. T4w ««aie wifl te
                                                                              considered f urflier as part -of fee
                                                                              promulgation sf the Toxic Substances fa
                                                                              Laboratories regulation. OSHA believes
                                                                              that most laboratories'{except histology,
                                                                              pathology, and anatomy) are already in
                                                                              compliance with the provisions al 'the
                                                                              formaldehyde standard. However, ito
                                                                              avoid imposing start-up cos** {or other
                                                                              laboratories wider this -standard which _
                                                                              may be unnecessatry if-fhe formaWBhyde
                                                                              standard is sirsperseded by fhe .general
                                                                              standard for laboratories. OSHA is
                                                                              extending the compliance dote lor .other
                                                                              laboratories to September 1,1088, «t
                                                                              w hich .time -the L*bar«tory standard is
                                                                              expected to -be dn effect •Qhrea the
                                                                              seventy of the health effects projected
                                                                              to occur at the existing 3 ppm TWA, 3
                                                                              ppm ceiling, and 10 ppm peak, "however,
                                                                              OSHA is requiring aH laboratories to lie
                                                                              in compliance with .the newPELi at 1
                                                                              ppm as a TWA and .2 ppmyfts a.STEL
                                                                              Clearly, formaldehyde is A very toxic
                                                                              substance which must be handled
                                                                              extremely carefully.. White many other
                                                                              labs ace presently m OoiBptianoe «rith
                                                                              this standard, -some are «ot, fmdOSHA
                                                                              is hesitant to create an open-ended
                                                                              period where •some jrflbe "rtlier"
                                                                              laboratory employees -wail have .110
                                                                              protection  from some of ihe advecse
                                                                              effects of formaldehyde, •MpaeiaHir
                                                                              those .proteotioBa in this ade toigganed
                                                                              by concern *bout tba d*amal •effects
                                                                              associated with fonnaktefejaie exposure.
                                                                              Therefore, should the Toxic Substtscaa
                                                                              in 'Laboratories final mde »«rt be is .effeot
                                                                              by September 1, IBSa this ntle (2BOFS
                                                                              1910.104&J will become (eifactisre far such
                                                                              other laboratories-so (that tisear
                                                                              employees will .be appnsfHimiely
                                                                              protected.
                                                                                 The Society of the Plastics Industry.
                                                                              Inc. (SPI) and the E.I. duPont de
                                                                              Nemours 'Company requested an
                                                                              exemption for the thermoplastic acetal
                                                                              molding industry from the *cope «(f beteves It
                                                                               would be mapppapriate forOSHA 4e
                                                                               grant fhe Tequested e»eeinpfien. OSHA
                                                                               notes, hovrevw, Ifeert 
-------
            Federal Register /  Vol. 52.  No. 233  /  Friday.. December 4,  1987 / Rules  and Regulations
                                                                      46247
 below the action level, the compliance
 burden will be minimal.
 .  The Scott Paper Company urged
 OSHA.to. exempt paper product* from
 the scope of the standard's hazard
 communication requirements because
 /'the use of paper products under normal
 conditions will not result in exposure of
 workers to formaldehyde at levels even
 remotely approaching" the action level
 or PELS proposed by OSHA [Ex. 80-62,
 pp. 1-2]. Where paper products emit
 only trivial amounts of formaldehyde,
 the final rule's hazard communication
 provisions will not be triggered.
 Furthermore, the final standard exempts
 employers whose workplaces contain
 only mixtures or solutions composed of
 less than 0.1 percent formaldehyde or
 materials incapable of releasing
 formaldehyde at concentrations at or
 above 0.1 ppm from compliance with the
 hazard communication provisions. Use
 of this regulatory approach means, in
 effect, that the great majority of
 downstream uses of paper products will
 effectively be exempted from the
 standard. Evidence in  the record,
 however, indicated that certain workers
 involved in the production of paper
 products may be exposed to significant
 levels of formaldehyde [Ex. 149,
 Appendix A]. In addition, the United
 Paperworkers International Union
 (UPIU) testified that many industrial
 hygiene practices and  procedures used
 in this sector-are inadequate [Ex. 149,'
 Appendix B]. Accordingly, it would be
 inappropriate for OSHA to exempt this
 industry from coverage.
  The American Furniture
 Manufacturers Association (AFMA)
 believed that the occupational
 exposures resulting from downstream
 uses of formaldehyde-bearing products
 are sufficiently different from those of
 formaldehyde producers to warrant
 separate ancillary requirements [i.e.,
 other  than the PELs). The AFMA felt
 that the proposed standard's protective
 clothing, emergency, and waste disposal
 provisions were inappropriate for
 furniture manufacturers [Ex. 80-68, pp.
 5-6]. In general, OSHA agrees with
AFMA that these provisions are much
 less important in a plant that assembles
 furniture that they are  where large
 quantities of formalin solution are being
handled. This does not mean that it is
necessary to design industry-specific
standards. The general standard's
provisions require only that the
employer provide the protection needed
given  the individual circumstances.
  A representative of the maritime
industry. Hal Draper of the West Gulf
Maritime Association,  asked .that the
final r-'e sontain an exemption for "the
 storage, transportation, distribution, or
 sales of formaldehyde in intact
 containters" from all of the provisions of
 the final rule except the hazard
 communication and emergency
 requirements [Ex. 80-53, p. 2]. Mr.
 Draper noted that such a "partial
 exemption" would be in keeping with
 similar language in OSHA's proposed
 benzene standard [50 FR 50512]  and in
 other OSHA standards [Ex. 80-53, p. 2].
 With the exception of the need to make
 an objective determination that  the
 containers are, in fact, intact, the
 formaldehyde standard already
 provides the relief requested by  Mr.
 Draper. Employers whose only use of
 formaldehyde involves intact containers
 would have minimal obligations under
 the standard.
   On the basis of OSHA's analysis of
 these comments, the Agency believes
 that, at the present time, a single
 formaldehyde standard applicable to all
 occupational exposures to formaldehyde
 is most appropriate and protective of
 employee safety and health! OSHA
 believes that all exposed employees,
 including those exposed only
 infrequently, should be provided with
 some basic protection because some of
 the adverse effects of exposure to
 formaldehyde are acute and arise after
 only a few minutes of exposure.
  The major change to the scope and
 application section of the standard since
 the proposal is the deletion of the
 proposed exemption for (i) Liquid
 formaldehyde solutions containing less
 than 0.1 percent formaldehyde, and (ii)
 solid materials made from or containing
 formaldehyde that are incapable of
 releasing formaldehyde into the
 workplace air.
  OSHA received many comments on
 this proposed exemption [Exs. 80-21; 80-
 34; 80-35; 80-37; 80-54; 80-56; 80-59; 80-
63; 80-64; 80-69; 80-72; 80^-303; Tr. 5/8/
86, p. 18] from participants requesting
that OSHA extend the proposed  0.1
percent exclusion for liquid-
formaldehyde solutions to solids
containing a similar percentage of
formaldehyde [Exs. 80-54; 80-56; 80-59;
80-64; 80-69; 80-72; 80-303]. The  major
concern expressed was that, without an
exclusion, employers would be required
to label articles such as textiles,  apparel,
envelopes, and other common items
routinely used by consumers [Exs. 80-56;
80-59; 80-63; 80-72; 8O-303]. Evidence in
the record indicates, however, that a 0.1
percent exemption for solids which
continue to release formaldehyde over
long periods of time is inappropriate and
potentially dangerous. For example, all
textiles would be exempt under such a
definition according to representatives
 of the National Cotton Council [Tr. 5/
 14/86, p. 76], but the record contains
 numerous reports of formaldehyde-
 induced illnesses among garment
 workers (see, for example, the NIOSH
 HHEs (Exs. 78 and 85) cited in Health
 Effects). OSHA's generic Hazard
 Communication standard clearly cover-
 textiles used in apparel manufacture,
 and the record in this rulemaking clepr'
 supports such coverage.1
   OSHA's response to these
 commenters has involved two changes
 to the final rule. First,  the Agency has
 refined the definition of formaldehvde
 so that it is clear that the standard
 applies only to formaldehyde: it does
 not coyer all of the other substances that
 may be present in a mixture. Even
 though OSHA's generic Hazard
 Communication standard exempts
 components of mixtures present in
 concentrations of less than 0.1 percent
 by weight, an employer is still obligated
 to recognize the minor component as
 hazardous if employees are exposed at
 airborne concentrations over either PEL.
 The best method available to assure that
 such hazardous exposures are not
 occurring  is the use of employee
 exposure monitoring, or at the least,
 objective data, which the formaldehyde
 standard assures will be collected for
 formaldehyde. Accordingly, the
 exemption has been moved to paragraph
 (m](l)(i), which defines, for the purposes
 of hazard  communication,'
 "formaldehyde gas, all mixtures or  ,
 solutions composed of greater than 0.1
 percent formaldehyde, and materials •
 capable of releasing formaldehyde into
 the air under any normal condition of
 use at concentrations reaching or
 exceeding 0.1 ppm" as a health hazard.
  Although use of this definition of a
 formaldehyde health hazard will not
 ensure that every employer is alerted to
 the presence of each and every product
 that contains even a trace of
 formaldehyde, OSHA believes that a
 manufacturer, importer, or distributor of
 a formaldehyde-bearing product would
 be aware if a product is capable of
 emitting sufficient formaldehyde to pose
 a health hazard.

Paragraph (b)—Definitions .

  In the final standard, the definitions of
 "Assistant Secretary". "Authorized
persons", and "Director" remain
unchanged from the proposal.
  The definition of an "action level" as
half the PEL, calculated as an eight-hour
time-weighted average (TWA), is also
essentially unchanged from the
proposal. An action level is an exposure
limit above which the monitoring and
annual training provisions of the

-------

-------
GENERAL ATTRIBUTES
             Attribute 4
The key findings of the report are highlighted in a concise
executive summary.
                 SOURCE   Case Study J. Red Dye No. 3 (Pages vi, vii).
                     Note  The excerpt illustrates a very concise executive summary for an
                           assessment of complex information relating to a somewhat limited
                           objective and scope.                         <°

-------
           A REPORT BY THE FD&C RED NO. 3 PEER REVIEW PANEL
AN INQUIRY INTO THE MECHANISM OF CARCINOGENIC  ACTION  OF  FD&C  RED  NO.  3
               AND ITS SIGNIFICANCE  FOR RISK ASSESSMENT
                                     Prepared, by:                    .

                                     Dr. Ronald W. Hart, NCTR/FDA (Chairman)
                                     Dr. Thomas Burka, NIEHS/NIH   .
                                     Dr.. Stan-G.  Freni, CEH/CDC
                                     Dr. Robert Furrow, CVM/FDA
                                     Dr. David W.  Gaylor,  NCTR/FDA
                                     Dr. Theodore  Meinhardt,  NIOSH/CDC
                                     Dr. Bernard  Sass,  NCI/NIH
                                     Dr. Elizabeth K.  Weisburger,  NCI/NIH
                                       Executive  Secretaries -  ,
                                        Dr. -Paul  Lepore,  ORA/FDA
                                      '  Dri  Angelo Turturro, NCTR/FDA
                                  July, 1987

-------
                                                                  July, 1987
                              EXECUTIVE SUMMARY               '

    A  Peer Review  Panel of  scientists drawn  from a  number  of  different
 agencies  was  directed  to  make an  inquiry  into whether  the  data  indicated
 that  FD&C Red  No.  3  (R-3)  had  a secondary  mechanism  of  carcinogenesis,
 whether  the  potential  human  risk of  the  color  could be determined,  and
 whether' additional  studies  to address  important  questions  in determining
 risk should be  performed.
    In  spite  of  the difficulties, of  the  task,  the  Panel  made  a  "best"
 effort  to  address the  charge  and  make a reasonable estimate of human risk
 based on this effort.   Because of  the  complexity of the task, consideration
 of  the  conclusions of the  report  without  an appreciation  of   their
 scientific context invites misunderstanding.
    The Panel  developed a working  definition  for a secondary  mechanism, of
 carcinogenesis  which  was specifically directed to R-3.   Examination  of  the
 pharmacokinetic studies using the color in rats and humans led to a  series
 of conclusions, including: absorption  is similar in ra.t and man, and  is  low
 (less than 2%).   Analysis of  short-term tests  indicate  no  evidence for  any
 DNA interaction relevant to  mammalian systems, and give no  suggestion of
 potential  direct  mechanisms   of  genetic toxicity.    However, there  is  a
 light-activated toxic activity.  Evaluation of long-term toxicity. studies
 indicated  that  R-3  induces   thyroid  follicular  tumors  in rats  in  a  two-
 generation study, with  adequate negative studies in mice.
    Because  the  toxic  effect  was  on  the  thyroid, the effect of  R-3  on
 thyroid  economy was  evaluated.    R-3  had  a  number  of effects  in  vivo,
 including an  elevation of Thyroid  Stimulating Hormone  (TSH)  and  thyroxine
 levels.  A number of  possible  mechanisms of this action were  considered, as
 well as other mechanisms for  inducing  tumors.   As  part  of understanding  .the
 biological relevance  of the  results in  the animal tests to man,  the  rela-
 tionship of thyroid  tumors  in rat  to  those in  man was  explored, as was  the
 special  role  that  iodide,  both  as contaminant and  as a product  of  R-3
metabolism, plays in  the toxic effect   of R-3.               .
    Using a weight-of-evidence approach, the Panel found sufficent evidence
 for animal oncogenicity and limited evidence of animal  carcinogenicity.
                                       vi

-------
                                                                 .July,  1987

    Characterizing  the  practical situation , in  humans, an average  exposure
of 1.41 mg/d was estimated  in  adults,  with analysis of the limits  on  human
                                                              5:
exposure.    The use  of  these  exposure  estimates  in  defining .risk  was
explored  in a  very comprehensive  fashion,  exploring basic  questions  in
assessing risk.
    In summary, the Panel  could not come  to  any conclusion concerning  the
exact  mechanism by which  R-3  induced  thyroid  tumors in  rats.    However,
there was little evidence that R-3 operates through any mechanism which  was
               ?                  -                •
inconsistent with the working definition of a secondary mechanism of carci-
nogenesis,  qualified  by the acknowledged  limitations  that  the definition
has in describing  carcinogenic mechanisms.   A  variety of ways of  deriving
risk estimates using this information  are  discussed and analysed.
                                     vii

-------
                         B.I

GENERAL ATTRIBUTES
              Attribute 5   77^ report explains clearly how and why its findings differ
                              from other risk assessment reports on the same topic.
                  SOURCE  Attribute J. Red Dye No. 3 (Pages 56-63).
                       Note  The excerpts illustrate how the authors' conclusions differ from
                             those of other studies. Section C lists important differences in expo-
                             sure estimations based on more recent data. Section D discusses
                             differences in tumor incidences reported by various pathologists.
                             Section F discusses conclusions on a no-observable effect level that can
                             be drawn from various studies.

-------
           A REPORT BY THE FD&C RED NO.  3  PEER  REVIEW  FANEL
AN INQUIRY INTO THE MECHANISM OF CARCINOGENIC  ACTION OF  FD&C  RED  NO.  3
               AND ITS SIGNIFICANCE  FOR RISK ASSESSMENT
                                     Prepared by:

                                     Dr. Ronald W. Hart, NCTR/FDA (Chairman)
                                     Dr. Thomas Burka, NIEHS/NIH
                                     Dr. Scan C. Freni, CEH/CDC
                                     Dr. Robert Furrow, CVM/FDA -
                                     Dr. David W. Gaylor, NCTR/FDA
                                     Dr. Theodore Meinhardt, NIOSH/CDC
                                     Dr. Bernard Sass, NCI/NIH
                                     Dr. Elizabeth K. Weisburger, NCI/NIH
                                      Executive  Secretaries -
                                        Dr. Paul  Lepore, ORA/FDA
                                        Dr. Angelo Turturro, NCTR/FDA
                                 July,  1987

-------
                                                                  July, 1987

cancer.   Further,  the above definition may  not  be applicable to all possi-
ble mechanisms  that the Panel may  consider  to be  secondary.   However, the
above  is  a working definition  that facilitates further  discussion in this
chapter.                                                             ,

B.  Pharmacokinetics                                ,
    The animal  and human  studies on  the  bioavailability of  ingested R-3,
described in Chapter  2, confirm  older  observations of poor absorption (131-
133).  As  shown in these  absorption  studies,  the  intestinal  absorption  of
orally given  R-3  is  less  than  2%  for" rats  and humans.   It is worthy  to
mention that the definitive rat  data were only  recently  accessible.   Thus,
these  studies were not  available to the Color Additive  Review Panel in its
assessment of the  risk  from 6 dyes  for external use  (4).   If  the risk from
external uses of  R-3 is re-evaluated  based  on  these  studies,  it is  likely
to be  increased and,  thus,  may need to be recalculated in  light  of the new
information.

C.  Hormonal Effects  of R-3                              '    . •        .    •
    As discussed  in the previous- chapter,  T~ and  T, are  thyroid  hormones
subject to  a  feedback  regulatory system that  includes  the  pituitary,  the
hypothalamus, and  the hormones TSH  and'TRH.   A decrease in Ta  or T^  levels
evokes an increase  in  TSH.   A  TSH-increasing  effect has  also been  seen
after  administration  of iodide and  a number  of  antithyroid drugs, which are
known  to  suppress the  formation  and  release  of  thyroid  hormones.    The
studies discussed  in  Chapter  5 have 'shown that R-3 induces  an  elevation  of
both T^ and TSH in rats.    Evidence  for an  action of R-3  through TSH  in
rats  would  gain in  strength if  a  decrease  in T, is observed.   Such  an
effect was seen in the  2 semichronic studies  (109,113).   The  observation  of
an increased, total T- and  a  decreased percentage  of  free  T~ in one  study
(123)  is so complicated by the  factors discussed  in  Chapter  5,  Section G.3.
that  it   can  be  considered  to  demonstrate  only  .that  TRH  provocation
increases following R-3 exposure.
    For humans,  more  detail is provided below.
    1.  Human Studies
        a) Ingbar _et_ _al_. (27) administered a  single dose  of  75-80 mg R-3  to
5 persons,  who  were  pretreated  with a  large  dose of  iodide to block  the.
                                     56

-------
                                                                  July, 1987

 uptake  of iodide by  the thyroid.  Within  a few days,  TSH  levels  showed a
 statistically significant increase.  It is not possible, however, to attri-
 bute  with certainty this  effect  to R-3, since  the  concomitant dosing with
 iodide  may have caused the TSH increase as well.   No  changes  were  seen in
 the T,  and T, levels.  This  study  is  discussed more extensively in Chapter
 2 in  terms of R-3 absorption  and  disposition.
        b) Witorsch ^t _al.  (134)  (Study Wl) exposed  males orally to 20, 60,
 and 200 mg R-3/d for  2  weeks (10 males per dose  group).   The differences
 between the means of  Day 15 and Day 1  TSH levels were -0.03, 0.29,  and 0.51
 UU/ml at  doses of  20, 60,  and 200  mg/d of  R-3, respectively.   The  average
 increase  in the TSH level for the 200  mg/day group was statistically signi-
 ficant.   No statistically significant  changes were seen in the levels of T~
 and T^  in this dose group, or of TS,  T^ and TSH in  the lower  dose  groups.
 In the  200 mg/d dosesgroup, the daily  urinary excretion of iodide was about
 one mg.   Provocation with intravenously  administered TRH  resulted  in  a
 large and statistically  significant increased TSH response  in  the  200 mg/d
 dose group, but not at lower  dose levels.
        c)  In a second  study, Witorsch et_ al.  (135)  (Study W2) exposed  9
 males and 9 females to  an  oral dose of 0.75 mg iodide-, twice  a  day,  for 2
 weeks.  In addition,  on  day 1 (dosing  of iodide started day  2),  and on day
 15, a  dose of  0.5  mg TRH  was administered intravenously to study  the TSH
 response.  A  significant decrease was  observed in the  levels of  T~  and T,,
 but no  change was  seen  in  free T_.  TSH did increase  with  the iodide dose,
 level,  but  this  increase  was  not  statistically  significant..   The  TSH
 response  on the TRH provocation was significantly  larger on  day 15  (after 2
 weeks of  iodide exposure)  than on  day 1.   The  total urinary  excretion  of
 iodide  was 1  to  1.1  mg  per  day, and  thus the same  as  seen  in Study Wl.
 There was no  difference  in  the results  between  males  and females.
    2.  Analysis
        From  the above described  studies,  it could be  inferred that 200  mg
 R-3 significantly increases the human TSH levels,  but, there  were no  statis-
 tically significant increases  in  average TSH levels  for the 20 and  60 mg/d
 groups in Study Wl.  Interpretation of  the  outcomes  of this  study is diffi-
 cult,  however.   The mean baseline  TSH  value is different for  each group,
and the baseline for  the 200  mg dose group  is  significantly  lower  than for
20 and  60 mg  doses.   In addition  to  the   difference  in  baseline  values,
                                     57

-------
                                                                    July, 1987

  heterogeneity  is  also evident  from the  large  variance observed  in the 20
  and  60 mg  groups.   Moreover,  the increased  TSH in  the 60  mg  group  is
  entirely attributable to  the  response  of  2 individuals, showing an increase
  2 to 3 times higher than  the  other subjects in this or other  groups.  This
  may indicate either measurement variation or an  abriormal  thyroid function.
  One should -be aware that  any  differences  between Day 1  and Day 15, even in
  the  200  mg/d group,  are still  small  in  view of  the  TSH assay  variance.
  Also,  since  the baseline TSH level at  the 200  mg is  close  to  the assay
  detection  limit,   it  is  probable  that  errors  in results  based   on  the
  difference between the  baseline and another value is  likely to  be  higher
  than in  the  other dose  levels  where the  baseline  level is* well  above  the
  assay detection level.
      Given the small number .and apparent heterogeneity of the  test  subjects,
  and in view of  the small,  if  any,  increase  in  the  R-3  level, (which, is well
  within the range  of  normal values), the  Panel feels  that this study does
  not allow an  assessment of whether or not  a  dose-response effect  is pre-
  sent.  Since even  the 200 mg/d  dose-group had  final TSH values (at  day  15
  after dosing)  not  different from those of  the 20  and 60 mg/d groups,  it may
  be even questioned whether the study has shown  an effect on TSH at all.   It
             .                          .                            «  '
  is quite possible  that  the TSH increase  seen  in this  group is  an artifact
,  caused by the abnormally low starting values.   No information was  available
  to exclude this possibility.  There is a lack of  quality assurance and con-
  trol,  which  could have  included  information  on the  health staus   of .the
  volunteers,  the time  of  day the blood samples  were taken, the  basal meta-
  bolic rate,  the use of medication with potential  effects on this rate or  on
  TSH,  and  on laboratory  performance.   The  study  design  could have been
  improved  by randomizing  the  treatment  over  all study  participants,  and  by
  measures  that  ensure  equal study and laboratory conditions.   Therefore,  it
  appears that  the most  that can be concluded  from  this  study is  that  a NOEL
  seems  to be present at  20 and 60 mg R-3/d,  but  that the  study design did
  not. exclude higher or  lower NOEL '"s.                              '..'-•
       
-------
                                                                  July, 1987

    A dose  of 1.5 mg  iodide,  effectively equal to  that  available from 200
mg  R-3  (judged  on  the  equal  amounts  of  iodide  excreted  in  the  urine),
appears sufficient to reduce  TS and T^  in humans  if  the iodide  is  given
alone, without  concomitant exposure to R-3.   Howevert> the increase  in TSH
with  exposure  to  that  amount  of  iodide  is  small  and not  statistically
significant.  In the PRI  study (109), male rats  showed increased TSH in all
treatment groups,  although statistical  significance  was not reached for the
purified R-3-only  group.    The Panel does  not  share PRI's conclusion that
the significant  effect in the Nal-orily group is due to chance.  The obser-
vation that this  group showed  the  largest increase is  consistent with the
demonstration that the purified R-3-only group showed  the least increase.
Since adding  more  than 80 ppm of Nal in the diet did not result in a larger
effect,  that level may  reflect  some-limit,  such as  saturation.   However,
inferences  from non-significant  differences between  groups,  albeit consis-
tent, should be judged with  caution.   Female rats  exhibited a significant
TSH increase in  the  R-3  + Nal  dose group  only.   Strangely  enough,  the
"serial assays" of both  male  and female  rats did not  show significant TSH
changes in  any of the treatment  groups.   This sampling  group was designed
to  evaluate  longitudinal  effects  by avoiding  inter-assay  variation (all
samples  were frozen until .the  end  of  the study,  and testing  for  TSH was
done in one run).  When  this  is done,  there was no effect with time at all.
    It  is  difficult  for  the  Panel .to  understand  the  results of  the PRI
study.   It seems  that R-3 affects  the T3 and  T^ balance, but not the TSH
level, while the  reverse is  true for Nal.   The TSH outcomes  of the study
can be explained  by  the  effect of the iodine (as impurity or as part of the
molecule)  component  of R-3.   Statistically significant decreases in T3 and
increases  in T, levels were  observed in  male and female rats treated with
R-3 fortified with Nal,  not in- the groups treated with Nal or  R-3 only.
    An  aspect not discussed  by  the researchers  conducting the semichronic
rat studies  is the  difficulty in interpreting  the' longitudinal results  in
light of  the role of  TSH in  thyroid hormone economy.  Although the studies
done by PRI and Ingbar _et_ al. (109,121,122) did show changes  in TSH, T3 and
T,  levels,  it appears that the  changes  occurred within the first months  of
observation.   It is  not  explained why  the hormone levels disclosed a  tend-
ency  to  go back to  normal at the  end  of the six  to  seven  month studies,
rather  than maintaining  their  peak  level in a  state  of equilibrium.   With
                                      59

-------
                                                                    July,  1987

  this  in  mind,  it. is unknown  what to  expect of  the  hormone  levels  of a
  second  generation after, two  years  in  the bioassays.   Moreover,  because  of
  the  rapid  response,  the  first  month  rat  values  are  likely  to  be  an
 'inadequate  reference  for  hormonal changes  in   studies  involving  humans
  exposed for only  15  days  (Study Wl  and Study W2).
     At present, it appears  that the peripheral conversion of T, to T3 has a
  central role in rat  thyroid  economy.   This conversion has been confirmed in
  brain, pituitary,  liver,  and kidney  (91).   R-3  inhibited  this  process _in
 vitro in  liver cells  of  the same  cohort  that showed the increased  T  and
                                                                        4
 TSH (124).  The inconsistency in the  effects of R-3 in vivo on the hormone
 levels may  be  attributable  to  a  dual  effect of  R-3.    On one  hand,  R-3
 lowers the TS  level  by inhibiting  (or reducing)  its peripheral  production
 from  T
4.
     As
                     deiodination  is  reduced,  the  T4  level  is.  likely  to
 increase.  A decreased. T3  will induce 'an elevation  of  TSH through  a  feed-
 back mechanism.   This  elevation will stimulate  the thyroid  to produce  and
 release more TA ,  further increasing T4  blood  levels.   On the other  hand,
 R-3 has  iodide  as a metabolite  and  as  impurity.   This will  tend to  lower
 the release of T4 and T3 by the thyroid, and . to  increase  TSH.   The actual
 hormonal effect  of R-3  should  then be  thought of  as a  function  of  the
 equilibrium state  of  the  different effects  of  R-3  proper  and  iodide.
 Whether TSH and/or  the  thyroid  hormones  will increase,  decrease, or remain
 unchanged is,  according  to this hypothesis,  a matter of  the balance  between
 the amount  of  free  iodide,  the  deiodination rate  of   R-3,  the  metabolic
 conversion  rate  of T4  to T3 in the liver, the conversion rate in the pitui-
 tary,  and the  susceptibility of the thyroid  to  stimulation by iodide.  This
 balance  is  complicated  by   the  clinical  .observation that   the   thyroid
 suppression  by  iodide  may be  only  temporary  and  an  escape from suppression
 may occur (136).
     In  summary,  in the intact animal, R-3 appears to act on thyroid economy
 through an  inhibition   of. peripheral metabolic   conversion  of  T,   to  T~,
 resulting in a reactive  TSH increase.  The latter response is increased by
 the  effect of the  iodide  impurity  and the iodine  component in  the R-3 mole-
 cule.   Because there  is an  increased TSH-response  on  the 'TRH-provocation
with high doses  of R-3,   it  is possible   that R-3  increases the  sensitivity
of  the  pituitary TSH response to thyroid hormone levels  as  well.   Since
 there is  evidence  of an  effect  on  thyroid economy, and since  the effect (an
                                     60

-------
                                                                  July,  1987

increase in  TSH elaboration)  appears  to be  important  in the  induction  of
thyroid follicular  neoplasia, and  since the. target  organ is  the  thyroid,
therefore, available  evidence supports the conclusion  that  the tumorigenic
effect of R-3 is the  result  of an indirect effect on the  thyroid,  probably
involving TSH.  This  is further supported by  finding  little  evidence of R-3
genetic toxicity (Chapter 3).

D.  Pathology                                   .    .
    The microscopic  distinction  between  adenomas and carcinomas  of  the
thyroid is  often  difficult,  which  may lead to  differences  in  diagnosis
among pathologists.   The  slides of  the  IRDC  studies  have been reviewed  by
several pathologists  with differing  results  (see Table  3).    Although the
original reports of  the IRDC studies indicated only  an oncogenic effect of
R-3 in  male  rat thyroid follicular  cells,  and only at the 4%  dose,  a  more
comprehensive  evaluation  of  the  available  data  has  presented  another
picture.  As discussed in Chapter 7,  there  is very weak evidence that there
is a carcinogenic effect of  R-3 in male rats.
    The Panel  has  evaluated the information  on ultramicroscopy of  the  thy-
roid  and  agrees that there is some evidence  that  the increased , abnormal-
lysosomal 'bodies may represent  a compound-related  effect.    However, the
Panel's evaluation, as well as that of pathologists  from  FDA and the  study
contractors, is  based on a  non-quantitative judgment.  Confirmation of  this
feature  requires  a  quantitative  method  of  evaluation,   such  as   image
analysis or  cytophotometry.

E.  Maximum  Tolerated Dose  in a Chronic Toxicity Bioassay
    The Panel  has  considered the  issue of  the effect  of.exceeding the Maxi-
mum Tolerated Dose (MTD) on the  validity  of the tumor response data  with
regard  to  their relevance  to  risk estimation.   The  OSTP  document (10) has
defined  the  MTD as "the highest  dose ....  which,  given for  the duration ,of
the chronic  study,  is just  high  enough to  elicit signs of minimal toxicity
without  significantly altering the animal's  normal lifespan due to effects
other  than carcinogenicity."   The problem is how  to define "minimal  toxi-
city."   If  this definition  is broadened to include any histomorphologic or
pathophysiologic effect, many would  find  the definition unacceptable for
chronic toxicity  testing.    Using  the broadened  definition,   the  4%  dose
                                      61

-------
                                                                     July,  1987

   level of the IRDC study  (74)  did exceed the MTD, as shown by  the electrons-
   microscopic lesions  of  the  thyroid and  in the weight  loss  in  R-3 dosed
   animals  (Chapter 5).   Exceeding  the  MTD could  have  resulted in  metabolic
   mechanisms  ultimately responsible, for  tumor formation  not occurring at  the
   lower doses.   Using  the  OSTP definition, bioassay  results  at a  dose which
   may  be above  the  MTD  might be acceptable as  long as  these toxic  effects do
   not  interfere with the  tumorigenic  mechanism, e.g., by  inducing  premature
   mortality .or  triggering a metabolic mechanism  that would not have occurred
   at a  lower  dose  level.  From  this viewpoint,  the 4%  dose level in the IRDC
   study  did  not exceed  the MTD.   The Panel's opinion  is  that  the possible
   exceeding of the MTD  in this  experiment does not  alter  the positive finding
  of oncogenicity since  excess  tumors  have also been observed;below the MTD.
  Risk estimates will  be presented  for  data  both  inclusive and  exclusive of
  the 4% test results.
      Another complication  in  the  interpretation of the  MTD for R-3 is  that
  the data concerning -oncogenicity  result  from multigeneration studies.   In
  addition to the possible  intrauterine exposure of the  fetuses to  R-3 from
  the moment  of conception  until the fetus regulates  its own hormone levels,
  there are also the consequences  of exposure  to  altered maternal  levels  of
  T4,  T3 and  TSH.  The  fetus is also likely to be  exposed  to elevated  levels
  of-iodide.   Even slightly  increased  TSH levels  may  disturb the development
  of  the fetal growth  regulation system.  It is, thus,  conceivable that doses
  of  R-3 not  exceeding  the *MTD in  the mother,  may grossly exceed a  threshold
  beyond which the growth  regulation system of the fetus is affected.   This
  effect  may be  the  cause of tumor  formation.   That the R-3 doses in  the IRDC
  studies were fetotoxic is  beyond doubt.  Doses lower  than the lowest IRDC
  dose  have caused  fetal  absorption and  decreased litter size In rabbits
  (80).  Rats  fed 4% R-3  had offspring  with fewer  and smaller pups (73).
     However, it should also be recognized  that  exposure  of .the  fetus from  '
  conception  through maternal  exposure to  R-3 mimics  human exposure.    The
 Panel is of  the opinion that  the  issue  of an  MTD in multigeneration studies
-needs further investigation.  jSome  arguments  have  been presented above per-
  taining to  fetal  intoxication  as a triggering  mechanism.  Although  fetal
 toxicity is   likely to  have occurred  even at  doses lower  than 4%,  the lack
 of factual data on hormone  levels in  fetuses  or newborn rats  renders  these
 arguments  speculative,  albeit' based on scientific  reasoning.
                                      62

-------
                                                                  July,  1987

F.  No-Effect Levels of R-3
    No  observed effect  levels  (NOELs)  are  often  used  to  set  limits  to
exposure levels for non-neoplastic  health effects.   Because R-3 may  work
through  secondary mechanisms  which  are  not  by themselves  directly  neo-
plastic, and  which may  have thresholds,  it  is useful  to  try to define  a
NOEL for its effects.  It  should be clear that the value of  a NOEL  depends
on a number of  factors,  such as  the  toxic potency of  the chemical,  the size
of the study population, the number  and  spacing of  doses,  etc.
    None of  the rat  studies discussed in  this report  demonstrated  a NOEL
for the effects on  T3> T4  or TSH,  not even at the  lowest subchronic  dietary
R-3 dose  level  of  0.5%  (approximately 250  mg/d),  for  which hormone esti-
mates  were  available.    In humans,  as  noted above,   Study Wl  found  a
statistically  significant  increase  in TSH from, a  dose of 200  mg/day, but
not at  20  or 60 mg.  However,  as stated  earlier, analysis  of this study is
complicated by'apparent heterogeneity among  the 3  dose  groups.   Because of
the  lack of significant differences, the  20 and  60  mg/d groups  could be
taken  as the NOEL.   Ingbar et  al.. (27) found no effects in humans at any of
the dose levels from 5  to 25  mg/d administered for up  to  4  weeks,  but the
same  authors  reported a TSH-increasing effect  of  a single, oral  dose of 75-
80 mg  R-3  (28).
    From the  above, the Panel concludes  that a human NOEL for R-3 hormonal
effects,   i.e.,  the highest level ,of R-3 that does  not  cause  changes in
hormone levels that may possibly lead to tumorigenesis, can  be set at  20 to
60 mg/d.  Alternatively,  one  could set  the NOEL at a higher level,  perhaps
200 mg/d, and  use a safety factor  to account for the  poor  quality of  the
 study which  "defines" -the  NOEL.-   Because of  this  uncertainty, the  Panel
will give results based on  several  possible  NOELs.

 G.  Relevance  of Animal Toxicity  to Humans
     As  stated in  the  OSTP document (10),  when  evaluating   human  carcino-
 genicity,  in   the  absence  of  adequate  human  data  it  is  reasonable,  for
 practical  purposes,  to  regard  chemicals   for' which  there  is  sufficenf
 evidence  of  carcinogenicity  in animals  as  if  they  present  a  carcinogenic
 risk  to humans.   However, there  are   some  human data  that suggest  that
 thyroid tumors in humans  may have  a  pathogenefic  basis different from that
 in rats.  • First,  the annual age-adjusted  incidence rate of all  types  of
                                       63

-------
                        B.I

GENERAL ATTRIBUTES
             Attribute 6   The report explicitly and fairly conveys scientific uncertainty,
                             including a discussion of research that might clarify the degree
                             of uncertainty.
                 SOURCE  Case Study D. Formaldehyde (Pages 7-14 to 7-23).
                      Note  This report illustrates a treatment of uncertainty. Potential research
                            to clarify uncertainty is not discussed.

-------
         Assessment of  Health  Risks

to Garment Workers and Certain Home Residents

        from Exposure to Formaldehyde
                  April 1987
  Office of Pesticides and Toxic Substances
     U.S.  Environmental Protection Agency

-------
 7-3.   Uncertainty in Riafc Estimates

      Model-derived risk estimates should be viewed m the prcoec

 context.   The upper bound estimate should not be viewed as -a

 point estimate of risk.   As the Guidelines state (s?A,  1936):

 "the linearized multistage procedure leads to a plausible upper

 limit to  che risk that is consistent with some proposed

 mechanisms of carcinogenesis.   Such  an estimate,  however,  does

 not necessarily give  a realistic prediction of the risk.   The

 true value of the risk is unknown, and may be as  low  as  zero."

 Other factors are also important.

      As Table 7-2 illustrates,  there is  a wide range  between the

 MLE  and upper bound estimates,  approximately  4 or 5 orders of

 magnitude.   This  illustrates the  statistical  uncertainty  of.the

 estimates  generated due to  the  input data from the study  used,

which in this  case is highly non-linear.   For  instance, the

 individual  risks  for apparel .workers  range  from 1 X 10"3  [31] zo

6 X  10"  CBl].  In addition, it has  been  shown that the MLE  _s  -

sensitive  to  small changes  in response data when  the  response is

very nonlinear in the experimental range.  For instance, the dose

giving a risk of 1 X 10'6 (MLE) varies significantly  due to  small

changes in the response data of the Kerns et ai.  (L983) study

(Cohn, 1985b).  The following illustrates this:
        Response at 2 ppm
          .(malignant)

          1.    0 (actual)
          2.    1/1,000
         .3.    1
Dose for Risk of
 1 X 10"6 (MLE)-

   0.67 ppm
   0.0022 ppm
   0.0006 ppm
                              7-14

-------
     Ten perturbations of the squamous ceil carcinoma data for
the Fischer 344 rats were selected by slight alteration in one  of
the dose-response proportions or the,elimination of a dose level
from the study in an attempt to show sensitivity to these
perturbations was examined by modeling.  These estimates appear
in Appendix 5.  It was found that, in general, slight
perturbations of the data do not significantly disturb the
predictive power of the model for upper bound estimates.  This is
not  the  case  for MLEs.  Only extreme perturbations significantly
affect upper  bound  risk estimates.  Consequently, when modeling
data that  are very  non-linear,  one  should  not place  great
certainty  on  MLE estimates.  In addition,  model  choice can  lead
to uncertainty.  As Appendix 3  illustrates,  there  is a wide
divergence in risk  estimates obtained using the  CUT rat data.
Independent background,  tolerance distribution  models such as,
the probit,  logit,  and Weibull, produce estimates  indicating
 virtually zero risk (probit predicts zero risk).  The independent
 and additive background gamma-multihit models produce similar
 results.  However,  when additive background models are used risk
 estimates are much higher, with the multistage model giving the
 highest riafca.  As discussed in section 7.1, the linearized
 multistage procedure was used  for primary  risk  estimation.
      As discussed  above, the major contributor  to the uncertainty
 seen in the  risk estimates  using the  multistage model is the    .
 steep dose-response  seen  in the Kerns et  al (1983)  study.   There
 were no carcinomas at  2 ppm,  2 at  5.6 ppm,  and  103  at ,14.5 PPm,
                                7-15

-------
which is a 50-fold increase  for only a 2.5 times increase in



dose.  If changes in respiratory rate are taken into account (the



rats at 14.3 ppm are receiving the equivalent of a 12 ppm



exposure—use of this data leads to no significant change in



estimated risks at exposures of concern) (Grinstaff, 1985),  there



is a 50-fold increase for 'only a doubling of the dose.



     HCHO's ability -to cause rapid cell proliferation, cell



killing and subsequent restorative cell proliferation, its



ability to interact with single-strand DNA (during replication),



interfere with DNA repair, its demonstrated mutagenicity, and the



fact that the dose was delivered to a finite area may help



explain the abrupt increase in the response.   However, none of



these factors demonstrate the presence of a threshold or minimal



risk at exposures below those that cause significant



nonneoplastic responses such as cell proliferation,  restorative



cell growth,  etc.  For instance,  although HCHO causes varying



degrees of cell proliferation in the nasal mucosa of rats due to



HCHO exposure,  it must be remembered that there is a natural rate



of cell turnover in this tissue.   While it is low in comparison



to HCHO induced increases, it does provide the opportunity for



HCHO to react with single-strand DNA during cell replication,



possibly resulting in a' mutant cell which,  if proper conditions



are met,  could result in a neoplasm.   While an event such as this



may be rare,  it is not unreasonable when one considers that the
         •>


opportunities for this event to occur are great due to the



immense number of cell-turnovers which may lead, to defects ;in
                               7-US

-------
some cells of the population of the individuals exposed.   Even
so, the marked nonlinearity of the response introduces
considerable uncertainty into any discussion of the possible
mechanism of HCHO induced carcinogenicity at exposures below the
experimental range.                                     ;~
     The .different responses seen in the animals tested also
leads to a degree of uncertainty.  Although rats, mice, and
hamsters have been tested in long-term bioassays, only in rats
have statistically significant numbers of neoplasms been
observed.  Only two carcinomas were seen in mice at the highest
dose in the CUT study, but the nature 'of this response is
complicated by the fact that mice are able to reduce their
breathing rate to a greater extent than rats.  If this effect is
accounted for, the "dose" mice received at 14.3 ppm is
approximately that which the rats received at 5.6 ppm, where two
carcinomas were observed.  Consequently, on a "dose" received
basis, rats and mice may be equally sensitive to HCHO.  Although
no neoplasms were seen in the hamster study, a number  of  factors
may be responsible.  Firs't, there was poor survival.   About 40%
of the 88 hamster died before eighty weeks, and only 20 hamsters
survived ninety weeks or more.   If a response comparable  to that
of the CUT-study were expected, 25% or five of  the hamsters
surviving ninety weeks or more would have had tumors.  However,
the duration of the study may not have permitted them  to  be
grossly visible.  Second, the limited pathology  protocol  may  not
have been able to detect small tumors.  And  third,  the dosing
                               7-17

-------
regimen and physiologic factors*  (changes  in breathing  rate) may

have been factors (see section 4.1).

     Although the foregoing helps explain  some of  the  species

differences observed, there remains the possibility that other,

unknown, factors may be important.  However,  in  any event,  no

data have been developed to show that humans  would respond

differently to ECHO than rats and data exist  showing that rats

and monkeys respond similarly to HCHO when nasal irritation and'

squamous metaplasia are used an  endpoints.

     It. is often useful to compare lifetime excess risks

estimated from the epidemiologic studies  to "those  risks estimated

from animal data.  Tables 7-4 and 7-5 and  Figure 7-1 present such

a comparison.  Estimated lifetime excess  risk can  be determined

for either occupational or domestic exposure  to  HCHO.  This

comparison assumes that exposure- to HCHO  is associated with an

increase in neoplasms at one site only and that  the site-specifie

excess risk observed in the epidemiological study  is the excess

above a risk of one for the study population  relative  to the U.S.

population (Margosches and Springer, 1983).   Hence, lifetime

excess risks based on the epidemiological  studies,  are  calculated

by multiplying the excess risk observed in the epidemiologi.c

study by the site-speed fie mortality ratio.   1980  mortality data
                                              t,
are used in this calculation.

     The estimated lifetime excess risks  were based on-

significant associations observed in the  Blair et  al.  (1986),

Vaughan et al. (1986a,b), Hayes  et al. (1986), Stroup, Harrington

and Oakes (1982), and Harrington


                               7-L8

-------
                              Table 7-4.
                 Upper Bound Risk Estimates Based on
             the CUT Data for Given Exposures to HCHO
Exoosure
Level (ppm)
Animal Based
Upper Bounda
Resin Worker
  0.24
  1.4
Furniture Worker
Pathologists

Mobile Home
Residents
(10 years)
  0.1
  1.3

  3.2

  0.19
  5 X 10
                                         X  10
  1 X 10

  2 X 10

  6 X 10

  3 X 10
-4

-3

-4

-3
                                              -3
a Based on the linearized multistage model  and  the  rat  data
from Kerns et al.  (1983).
                                7-19

-------
                               Table 7-5

                    Estimated Lifetime Excess Risks
               Calculated from the Epidemic-logic Studies
Exposure Author
Resins Blair et al.
Site
Lung
Risk
Ratio
1.32b
Nasopharynx 2.0C
Resin, Glue Vaughan et.al.
HCHO & Wood . Hayes et al .
Pathologists Harrington &
Shannon
Harrington &
Oakes°
Anatomists Stroup
a Estimated lifetime excess risk
Nasal
Cavity &
Sinuses
Nasal
Cavity &
Sinuses
Leukemia
Brain
Brain
= (RR-1) *
3'. 8
1.9C
2.0
3.31.
2.7
Estimated Lifetime
Excess Riska
2 X 10"2
8 x icr4
7 X ID'4
2 X 10~3
2 X 10~2
1 X 10"2
8 X 10~3
*•=•
# of site-specific deaths
proprotion
— c
of site specific
^aths __
Mortality proportion based on  1980 deaths.


Analysis -of white male wage workers with greater -than  20  years  latencv
and HCHO exposure above Oppm-year.
                          -                                       .    '
Analysis o.f white male wage worker with HCHO exposure  greater  than  OD
year .             ,.,'••
                                 7-20

-------
Pathologists
                  10
                                                         3.2 ppm
1 	 	 	 I
-.5

.-4


«-3 J ,

,-2


10
                                                 10
                              10
                                                  Stroup,    Harringtons  Harrir.o-
                                                  Brain"     Oakes,       sShannc:
                                                            Brain        Leukeni
Kesin Workers
                  10
                    -5
                                       0.24  ppm    .1. 4 ppm














,


10
  -4
                                     Vaughan
                                     et al. ,
                                     SNC*
ao
               Blair  et al. ,
               Nasopharynx
                                                                10
                   Blair et al,
                   •Lung- 7.0 yr
                   • latency
 Pttrniture Workers
                   10
                     -5
 Mobile Hone
 Residents
                   10
                     -5
                                 0.1 pom
                 1.3  ppm
 10
   -4 '
 10
                                                   -3
                               10'
                                                Kayes et al.
                                                SMC, Controlled
                                                for high wood dust
                                                exposure
                                    0.19 ppm
 10'
                                                 10
                                                   -3
                10'
                                                Traurrhan et  al. ,
                                                Masoioharvnx
       Fiqure 7-1.  Comparison of the upper bound  risks  based,on the  animal
       data to estimated lifetime excess  risks  based  on  the  epidemiological
       studies.  Animal-based upper bound risks for the  identified exposure
       lavel to HCHO are above -the line.   The estimated  excess lifetime
       risks based on the observed excesses in  site-specific neoplasms are

-------
and Shannon  (1975) studies.   For  example, when  one  examines  lifetime


risks from exposure to resins, the estimated  lifetime  excess risk


associated with the 35% increase  in  lung cancer among  white  males


with a greater than 20 years  latency reported by Blair et  al.  (1986)


would be 2 X 10~2 and the estimated  lifetime  excess  risk associated


with their reported 200% increase in nasopharyngeal  cancers would be


8 X 10~. ,* The 280% increase  observed by Vaughan et  al., (as


reported in SAIC, 1986) for nasal sinus and cavity neoplasms in


conjunction with exposure greater than 10,000 hours  to resins,


glues,  and adhesives gives an estimated lifetime  excess risk of 7 X


10~ .  The upper bound risk for an exposure of  0.24  ppm HCHO based


on the animal data is 5 X 10~4, and for an exposure  of 1.4 ppm


HCHO, would be 3 X 10~3.


     Comparing the results reported by Hayes et  al.  (1986) is


more complicated since Hayes et al. do not delineate the exposed


population.  However, if one chooses an exposure  group, such as


furniture workers who may be exposed to both wood dust and HCHO,.


one can make some observations.  The reported exposure for


furniture workers ranges from 0.1 ppra to 1.3' ppm  HCHO  as an     /


8-hour,  time-weighted-average.  Upper bound risks based on the


animal data associated with these exposures  are  1 X  10"4 and


2 X 10~3, respectively.  Using the 90% increase  in nasal cavity


and sinus risk observed in analyses which controlled for high


wood dust exposure,  the estimated lifetime excess risk based on


the Hayes et al.  study would be 2 X 10~3.


     Thus,  when individual tumor types are examined, one can see
                         '.                         /

that the upper bounds are not indicating larger  excesses than
                               7-22

-------
seen in certain studies given uncertainties about exposure.
Although HCHO's potential carcinogenic effects are not expected
to be limited to one site in humans because humans do not
necessarily breathe through their noses as rats do,  the analysis
described above provides a check of the risks derived from animal
data and those seen in human studies.
     Finally, a factor that can have a major bearing on
population risk estimates is the quality of the available
exposure data.  Assumptions made in reporting exposure levels can
have a major impact.'  For instance, it is not uncommon during a
monitoring exercise to find a number of samples that are below
the detection limit of the analytical technique used.  Thus,  when
a mean exposure level is calculated it should be realized that if,
the nondetectable  (ND) samples are counted as 0 the calculated
mean will understate the actual situation.  Conversely,  if the MD
samples are counted as the limit of detection, the mean win
overstate the true situation.  Another factor that can skew
exposure estimates are changes in  non-governmental exposure  limit
recommendations and the  number of  years over  which the data  are
collected.   Since  a number of years of exposure  data  are  often
used  to calculate  means, it  is possible that  the mean will be  ,
weighted by  samples taken prior  to changes in voluntary  exposure
limits.  Thus,  the reported  mean could be substantially
overestimating  the true  situation.  For  instance,  in the garment
industry, HCHO  levels  have  apparently been falling since the late
70's  and early  80's as a result  of increased concern and a
                               7-23

-------
                             B.2

HAZARD  IDENTIFICATION
                  Attribute  1
All relevant information is presented and reviewed.
                      SOURCE  Case Study B. TCDD (Pages v-xvi)
                          Note  These excerpts are' from the executive summary of a very compre-
                                 hensive assessment. This illustrates the many types of information
                                 given detailed consideration in the full report.
                      SOURCE  Case Study H. Methylene Chloride (Pages 5-9).
                      SOURCE  Case Study J. Red Dye No. 3 (Pages iii-v).
                         • Note   This report considers a large body of information. The Table of
                                 Contents is included in this appendix to illustrate the types of data
                                 considered.

-------
ncasl
special report
NATIONAL COUNCIL OF THE PAPER INDUSTRY FOR AIR AND STREAM IMPROVEMENT. INC, 260 MADISON AVENUE. NEW YORK. N.Y. 1001.
               EXECUTIVE SUMMARY



      DIOXIN: A CRITICAL REVIEW OF ITS DISTRIBUTION,



       MECHANISM OF ACTION, IMPACTS ON HUMAN HEALTH,



       AND THE SETTING OF ACCEPTABLE EXPOSURE LIMITS
             SPECIAL REPORT NO, 87-07
                  MAY 1987

-------
                             — y, -
 C.    The  Synthesis  and  Discovery of  "Dioxin"

      Unlike many  environmental  chemicals which have  value  in
 manufacturing  or  commerce,  no use has  been discovered  for
 dioxin.*  Although  it was  first  synthesized in the heyday  of
 German  organic  chemistry near the turn of the century, mention
 of  it practically disappeared from the scientific literature
 until 1957.  In that year,  an outbreak of chloracne, an
 occupational skin disease  caused by  exposure to certain
 chlorinatediorganic chemicals, prompted a dermatologist and  a
 chemist to investigate  the  chemicals in a plant in Germany
 After a number  of false starts,  they identified dioxin as  a
 contaminant of  trichlorophenol,  a chemical used in the
 manufacture of  various pesticides, the most important of which
 was 2,4,5-T (2,4,5-trichlorophenoxy acetic acid), an herbicide.

      The chemists at the German  plant  devised new production
 methods^ that reduced dioxin contamination of trichlorophenol
 and eliminated  chloracne from,the plant's workforce.   Within a
 decade or so,  similar changes were made in other plants
worldwide, and chloracne as a consequence of exposure during
trichlorophenol manufacture was  largely controlled.

D-   Human.Effects of Dioxin                      •
     1.
Effects .from Occupational
Exposures to High Levels of Dioxin
          The chloracne discovered in the German plant was
     associated with routine exposures from leaks and spills
     during the production process:   Greater exposures resulted
     from industrial accidents.   At  one time or another in the
     United States,. Denmark,  England,  France,  and Germany,
     large pressurized kettles (autoclaves)  blew out their
     seals spraying trichlorophenol  and,  we  now know,  dioxin
     across workrooms and workers.   Chloracne  was common in men
     directly exposed and in men who cleaned up the messes.   In
     addition,  some liver pathology,  dizziness,  disabling aches
     and  pains  (nervous system damage),  reduced sex drive,  and
    The word  "dioxin"  is used to  refer to
                                                   als°  called

-------
                        - vi -
other adverse effects were seen in some of those workers.
With the exception of chloracne, which still persists in
some men exposed more than 35 years ago, the other
symptoms abated over time.  The presence of chloracne
among these men is convincing evidence of high-level
exposure.

     Numerous studies of more than 1,000 workers who were
exposed to high levels of dioxin during trichlorophenol
manufacture and accidents have failed to show elevated
cancer, heart disease, reproductive difficulties, or
premature mortality.  An excess of stomach cancers in one
occupationally exposed group is statistically significant,
but no excess of that cancer has been seen in other
exposed groups.  These overall negative findings are
especially important in judging dioxin's effects on humans
because there is no question that those workers were
exposed.  Many other studies and proposed studies of the
possible effects of dioxin on humans are hobbled or
forestalled by the  impossibility of deciding who was and
who was not exposed, and  if they were, to how much.^

2.   Dioxin Exposure at Seveso, Italy

     On June  10, 1976, an industrial accident spewed
dioxin and other chemicals into the air, and the resulting
chemical cloud drifted over the town of Seveso,  Italy.
Some effects  were  immediately apparent; vegetation  browned
and died, and small wild  animals sickened and died.

     The suddenness of the accident compounded  with
incomplete plans for handling- such a disaster resulted  in
many residents  remaining  for up to two weeks in the area
contaminated  by  "fallout" from  the cloud.  Before
evacuation was  completed, residents could have  been
exposed directly to the cloud,  to  chemical  fallout  from
it,  and through ingestion of contaminated garden
vegetables.   The exposures were sufficiently high to cause
chloracne  in  more  than  100 children.   This  is the only
exposure  situation, except for  the high-level occupational
exposures,  in which dioxin caused  chloracne.

     During the next year,  reports appeared stating that
the exposures had caused  abortions and an increase  in
birth defects.   In addition,  some  alterations  in human
biochemistry were associated with  the  exposures.   In the

-------
                          -  vii  -
  ten years since the accident,  the importance attached to
  the reports of abortions and birth defects has decreased.
  In 1976,  therapeutic abortions were illegal in Italy,'and
  some 30 presumably exposed pregnant women went to
  Switzerland for abortions rather than risk delivering a
  child with birth defects.  Suggestions,have been made that
  secret induced abortions may have been reported as
  "spontaneous"  abortions, contributing to an apparent
  increase.   The .study of birth defects was hampered by poor
  historical data in that region of Italy making comparisons
  with "normal"  rates dependent  on guesswork.   After
  Studying  all the reports from the exposed population,  the
  Government of  Italy and an advisory group of international
  experts have decided that chloracne was  the  only human
  health effect.

       Seveso was characteristic  of environmental  exposures
  in  that a  cross-section of  the  human population—children,
  the elderly, women and  men—were all  exposed.  It was
  unusual in that it was  a single high-level exposure.  The
  question of a  possible  increase in cancer  rates  as  a
  result  of  the  accident  cannot be eliminated;  10 years may
  not  be  sufficiently long for the development  of cancer,
  but  the Seveso  population continues  to be carefully
  monitored.  If  there  is  some health  effect that is  not yet
  manifest,  it should be  detected in the future.

  3.    Herbicide  Sprayers

      Chemical  industry workers  and the children of  Seveso,
• both of whom developed chloracne were clearly exposed.
 Although their  exposure  levels  are unknown, it can be
 assumed that their  one-time or  repeated exposures were
 greater, in general, than those of people who did not
 develop chloracne.  Epidemiologic studies in those highly
 exposed people have failed to produce convincing evidence
 of  diseases other than chloracne being related to dioxin
 exposure.

      The evidence for an association between dioxin
 exposure and human cancer that has attracted the most
 attention is an excess of cancers in herbicide sprayers
 The associations were made in two "case control studiesi"
 in  which the histories of men who had specific cancers
 (the cases) were compared to the histories of men who did-
 not  have the specific cancers  (the controls).   The cases
 more often reported exposures  to herbicides,  and  those

-------
                       - Vlll -
studies showed associations between cancers classified as
soft tissue sarcomas and lymphomas and herbicide use in
Swedish lumberjacks.  Similar case control studies in New
Zealand, Finland, and another one in Sweden faiTed to
confirm the associations.  Furthermore, a "cohort study,"
in which the incidences of those cancers in Swedish
agricultural and forestry workers were compared to the
incidences in workers in other industries failed to detect
an excess.  The strength of the association in the Swedish
case control studies makes it impossible, to disregard
completely those results.  At the same time, absence of
confirmatory studies makes them less than convincing.
Studies of another group of herbicide sprayers revealed an
excess of stomach cancers.  Although that finding
parallels a report from one chemical plant population, it
is not complemented by other studies of herbicide sprayers
or chemical workers where this effect has not been found.

     The most famous group of herbicide sprayers is
probably the Ranch Hands, the Air Force unit that sprayed
Agent Orange in Vietnam.  All of those 1,200 or so men
have had thorough medical and psychological examinations,
which are to be repeated at 3-year intervals for 20
years.  As of late 1985, the death rate among Ranch Hands
was lower, but not statistically so, than among a
comparison group of Air Force personnel who were not
exposed to herbicides.  There is no difference in cancer
rates between the two groups.  There are no soft tissue
sarcomas among Ranch Hands, and one in the comparison
population.  Sixteen years have passed since the last
Agent Orange spraying and about 20 since the peak spray
years.  The absence of excess disease or early mortality
among Ranch Hands so far argues that no excesses will be
found.

     Herbicide sprayers, whether lumberjacks or Ranch
Hands, are probably an intermediate exposure group.  They
are, judging from the absence of chloracne among sprayers,
less exposed than the chemical plant workers, and they are
almost certainly more exposed than people who may have
been  "environmentally" exposed through being around one or
a few application areas.

-------
                         - ix -
4.    Environmental Exposures to Dioxin

     a.    Environmental Exposure and Claims.of Miscarriages

          In 1978, EPA announced that it was  considering
     suspending all uses of 2,4,5-T as well as other
     pesticides made from trichloroghenol.  At the time of
     the announcement,  available data about the toxicity
     of  2,4,5-T and dioxin were limited to  results from
     animal  studies and fragmentary reports from studies
  .   of  industrially exposed workers.   That changed
     significantly when an Oregon high school  teacher
     wrote letters to Federal agencies reporting her
     investigation of a possible association between
     2,4,5-T spraying and miscarriages in women living
     near  Alsea,;Oregon.   The EPA commissioned a quickly
     done  epidemiologic study that appeared to confirm  the
     association.   Acting on those results, EPA in
     February 1979 declared most 2,4,5-T uses  an "imminent
     hazard"  and  issued an emergency suspension of those
     uses.

          Scientists  critiqued and criticized  the  EPA's
     epidemiologic study,  and it is  now  accorded scant
     credibility.   The  surest test of  a  scientific  study's
     value is whether or  not  scientists  cite it  in
     subsequent discussions.   The  Oregon miscarriage study-
     has almost disappeared except when  referenced' as the
     cause of the  emergency suspension of uses  of  2,4,5-T.

     b.   Agent Orange                           -

         In the 1960's,  a  new market opened up  for
     2,4,5-T.  The United States military settled on a
     50:50 mixture of 2,4,5-T and  2,4-D
     (2,4-dichlorophenoxy acetic acid) as the best
    herbicide to denude the  jungles and destroy certain
     food crops of the enemy  in Vietnam.  That mixture,
    called Agent Orange because of the color-coded band
    on the drums  in which  it was  shipped, became the best
    known herbicide in history.

        . From the initiation of the chemical  spray
    program in Vietnam, some Vietnamese, both North and
    South, claimed that the chemicals had caused birth
    defects and miscarriages.  The laboratory results
    showing that dioxin caused birth defects  in animals

-------
                    - x -
 created  enough concern that  a  series  of  Congressional
 hearings culminated in suspension of  Agent  Orange
 spraying in 1970.

      A team of American scientists that  visited
 Vietnam  during the war years found it impossible to
 verify that Agent  Orange had caused birth defects  or
 miscarriages.   Those health  effects are  notoriously
 difficult to study—they are not always  recorded and
 diagnostic criteria differ according  to  time and
 place.  Furthermore, record  keeping in the  midst of  a
 war is necessarily poor.  The team concluded that  the
 evidence was not persuasive, but that the many
 problems with records and information precluded
 coming to a definite decision.
                                                     %
      Ground troops in Vietnam could have been exposed
 to Agent Orange, and some veterans claim their
 illnesses as well as birth defects among their
 children stem from those exposures.  A study by the
 Centers  for Disease Control  (CDC) failed to find any
 excess of birth defects among children of Vietnam
 veterans.  The same study found associations between
 3 birth defects and "opportunities for exposure to
•Agent Orange," but the  authors of the study as well   •
 as many reviewers discount the reported
 associations.  The link between "opportunities for
 exposure" and the likelihood of exposure is so
 tenuous that the associations are suspect.   An
 equally good explanation for the  apparent
 associations is-that they are due to chance.

       Congress,  in 1979, mandated  a study of the
 health  of veterans who  were exposed to Agent Orange.
 The  study has not yet begun because of the to-date
 impossibility of deciding which veterans may have
 been exposed and which  were not.  However, the recent
 development of  techniques to measure dioxin in about
 100  milliliters of blood possibly provides a method
 to verify  exposure.   If the method proves out, the
 study of possible effects -of Agent Orange exposure on
 veterans may get underway in  1987.

-------
                    - xi -
 c.   Exposures in Missouri

      In Missouri, oily wastes from trichlorophenol
 manufacture were sprayed on unpaved roadways and in
 trailer parks and in horse arenas.  The confirmation
 of soil concentrations greater than 1 part per
 billion (ppb) in Times Beach was the stated reason
 the Federal government purchased the town.

      Epidemiologic studies have found no increased
 disease incidence among former residents of Times
 Beach nor among other Missouri residents that lived
 ,at or near sites with high dioxin levels.   However, a
 recent CDC study reported that there may be   '
 dioxin-related impairment of the immune system in
 people who lived in a trailer park where
 trichlorophenol wastes were used for dust
 suppression.   Soil concentrations of dioxin at that
 site ranged up to 1,100 ppb,  which is consistent with
;possible high exposures.   The CDC study depended on
 measurements  of skin reddening as an indication of
 immune system competence.   Problems in
 interpretations of those skin tests resulted in the
 investigators discarding data from 61 percent of the
 "exposed"  population and 32 percent of the
 "unexposed" control population.   The missing data,  of
 course,  might contribute to or account for tlie
 reported differences in the immunologic
 characteristics of the two populations.  Furthermore,
 the reported  immune deficits  are of such magnitude
 that clinically detectable diseases would  be expected
 to be more common in the exposed people.   No disease
 excesses have been seen which casts doubt  on the
 validity of the reported observations.

      Studies  of the exposed populations  in Missouri
 continue,  and additional  information will  be
 forthcoming.   Whatever  the results  of  those studies,
 they will  likely be the primary  source of  data about
 environmental exposures in the United  States.   It is
 unlikely that people anywhere else  in  the  United
 States were environmentally (as  opposed  to
 occupationally)  exposed to  more  dioxin.

-------
                            - xii -
E.   Animal Studies of Dioxin Toxicity

     Almost all our knowledge of the toxicity of dioxin comes
from animal studies.  Although much has been learned, much
research remains to be done to understand dioxin's effects in
animals.  For instance, the mechanism by which it kills, remains
unclear.  Neither is it understood why the lethal dose varies
5,000-fold between guinea pigs '(0.6 ng/kg) and hamsters
(>3000 ng/kg).   What is known is that lethally exposed
animals "waste away," gradually losing weight, and die after
two weeks or so.  One model for dioxin's lethal action is that
it alters a biological set point' causing the animal to eat so
little that it cannot sustain itself.  In addition, alterations
in Vitamin A levels and thyroid functioning have been suggested
as important in the wasting syndrome and death.  The .decreased
food consumption is sufficient to cause death, but the
mechanism that causes the decrease remains obscure.

     The wasting syndrome, in contrast to effects on specific
organs, is the only observed,effect of dioxin intoxication in
some species.  In other species, various organ systems are
affected, but the toxicity seen in those organ systems does not
appear to be 'sufficient to cause death.            •  .

     The human exposures most likely to have been immediately
life threatening were those in industrial accidents.  However,
there have been no reports of dioxin-related deaths".  In at
least one dioxin-contaminated workplace, rabbits were used as
biomonitors.  Even though workers had been in the same area
with no ill effects except chloracne, the animals died.  These
observations may mean that humans are less sensitive to the
lethal effects of dioxin, but that must be a guarded
conclusion.  Workers were clothed, reducing their contact with
contaminated surfaces and atmosphere, and the animals licked
and cleaned their fur.  Those differences in opportunities for
exposure complicate drawing conclusions about relative
sensitivity.

     Possible chronic effects from dioxin exposure, especially
cancer and reproductive health effects, are of greater concern
than acute effects.  Investigations of exposed human
populations have failed to produce convincing, consistent
evidence that dioxin has caused those effects, but they have
been seen in laboratory animals.      -

     In the 1960's, Congress and the Executive Branch directed
attention at possible health risks from pesticides.  In
response to those concerns, the National Cancer 'Institute (NCI)

-------
 contracted for  the testing of some 30 pesticides in laboratory
 animals.   The pesticides  were tested for carcinogenicity and
 teratogehicity  (the properties of  causing cancer and birth
 defects,  respectively).   In 1969,  the herbicide" 2,4,5-T emerged
,as a potent animal teratogen, causing cleft  palates in
 laboratory mice exposed  in the uterus.   Within a short time,
 further  studies showed that the teratogenic  activity- resided in
 the dioxin contaminant in 2,4,5-T,  not in the herbicide
 itself.   Even especially  cleaned-up 2,4,5-T,  with dioxin
 contamination reduced to  1/30 normal,  remained a potent animal
 teratogen.

      The  most important finding for assessing the risk of
 dioxin was  the  discovery .that it causes  cancer in laboratory
 animals.   Federal  agencies  assess  cancer  risks using  methods
 different from  those used for other diseases.   As a result,  at
 low exposure  levels, the  predicted  risk  of cancer is  greater
 than the  risk of other diseases.

 F.    The  Carcinogenicity  of  Dioxin

      There  is no doubt that  dioxin  causes cancer  in laboratory
 animals.  The most  sensitive organ  is the liver  of  female  rats;
 the  liver of  male  rats is  less  sensitive toward  the
 carcinogenicity of  dioxin.   This pattern of sexual  difference
 in  carcinogenicity  is commonly  seen in rats, but  the pattern
 also  shows  that despite its  potency, dioxin is not  equally
potent in all animals.  That  sex hormones are probably  involved
 in carcinogenicity  in the rat Liver has been shown  by
experiments with females from which the ovaries have been
removed.  They do not develop cancer following exposure to
dioxin.

     A growing body of scientific data supports the idea that  a
mutation  (or  at least a "mutation-like event") is the first
step  in carcinogenesis.   That step  is also called
 "initiation."  Whether or not an initiated cell progresses to  a
cancer cell, each of its  daughter cells inherits the initiated
DNA.  These initiated cells, although harboring a change in
their DNA, are not recognizably different from other cells.
However, given appropriate conditions, such as exposure to
agents called "promoters," these cells ccn become cancerous.

     Promoters,  unlike initiators,  are not believed to interact
directly with DNA,  and they have little effect on non-initiated
cells.  They can, however, hasten progress from the initiated
state to the cancerous state.  Initiators, because they
interact with DNA,  are "genotoxic;" promoters are not.

-------
                            - xiv -
     Many experiments have examined the potential of dioxin to
cause mutations.   Almost all of those experiments have been
negative, and the few positive reports have not been replicated
despite repeated tries.   Committees of experts in various
countries, including advisors to the EPA, have concluded that
there is no convincing evidence for dioxin being mutagenic.
Therefore, there is no evidence for it being an initiator of
carcinogenesis.

     If dioxin is not an initiator, what role does it play in
carcinogenesis?   Despite-intensive efforts directed at that
guestion, only partial answers are available,  The most
productive research for generating a testable hypothesis about
dioxin's mechanism has focused on interactions between the
chemical and certain "receptor" molecules in the cell.

     Animals and humans are exposed to many foreign
("exogeneous") chemicals.   Many plants synthesize toxic
chemicals that inhibit growth of parasites or cause sickness in
animals that eat them; spoiled meat and vegetables can be
contaminated with putrification products (less of a problem in
this age of refrigeration and food preservatives, but certainly
a problem for our ancestors), cheeses and other fermentation
products contain large numbers of complex organic chemicals,
and smoke from wood fires and other sources are loaded with
chemicals.  Animal and human cells have the genetic information
to produce enzymes that metabolize those foreign chemicals and
eliminate them from the body.  A group of enzymes called the  -
"P-450 enzymes"  or "mixed function oxidases" (MFOK or., "aryl
hydrocarbon hydroxylases" (AHH) are synthesized in response to
the entry of foreign chemicals into the cell.  They are
generally studied in the liver or in liver cells because that
organ is the site of much of the metabolism of foreign
chemicals.

     There are several classes of P-450 enzymes.  One class is
responsive to pqlyaromatic hydrocarbons (PAHs).  Exposure of an
animal to such a chemical causes increased synthesis of the
P-450 enzymes.  The mechanism of the induction of synthesis is
binding of a PAH or related chemical to a protein called the Ah
receptor.  The chemical-Ah receptor complex then interacts with
a specific location on the DNA to "turn on" the P-450 enzyme
genes.  The products of the genes are P-450 enzymes which
degrade or otherwise metabolize the foreign compounds.

-------
                              -  xv  -
      Dioxin  is  a  potent dnducer of the PAH-indueible P-450
 enzymes.  Metabolism of dioxin by P-450 enzymes  produces
 products  that are, less toxic  than the parent  compound,  and  the
 current consensus holds that  dioxin itself, and  not  a
 metabolite,  is  the active  toxic agent.
                     j» •                      v' ,
      The  induction of P-450 enzymes is  a reversible  event.
 Once  the  level  of the inducer,  whether  a PAH  or  dioxin, falls
 below a critical  concentration,  the Ah-receptor  protein is  not
 bound and does  not interact with DNA.
                                         s

      A current  model for the  carcinogenicity  of  dioxin  is that
 the dioxin-Ah receptor complex acts to  "turn  on"  a second
 battery of genes,  different from the P-450  enzyme genes, in
 some  cells or under  certain conditions.   The.toxic effects  of
 dioxin appear to  be'limited to  certain  cells.  For instance,
 dioxin causes cancer in initiated liver  cells  and skin  cells of
 hairless  mice,  but not in  uninitiated liver cells or  in skin
 cells  of  normally haired mice.

      The  exact  nature of the  second battery of genes  is not
 known.. Neither is the action  of  the gene products.   Currently,
 some  scientists are  investigating the possibility that dioxin
 disrupts  the normal  regulation  of cell growth.  Dioxin also
 interferes with the  immune system of laboratory  animals, and
 certain immune deficiencies, could facilitate cancer
 development.  The  possibility that  the promotion  activity of
 dioxin involves inhibition of the immune  system  is also a focus
 of active investigation.

     Whatever- the  nature of the genes controlled  by the
 dioxin-receptor. complexes, what  is  known about the seguence of
 events is consistent with the idea  that there is  a threshold
 for dioxin's role  in carcinogenesis.  A sufficient
 concentration of the complex is necessary to turn on  and keep
 turned on the genes to produce enough gene products to cause  -
 some critical biochemical reactions.  Those reactions probably
 have a threshold;  until a certain number of reactions
 accumulate,  damage should be reversible.  Although not all
 aspects of this mechanism are known, there is  substantial
 experimental support for it.

     Classification of carcinogens  as initiators or promoters
 is based on observations  from many different experiments.
Dioxin has been found to  cause cancer in laboratory animals at
 0.01 micrograms/kilogram body weight of the test animal/day
 (hereafter vig/kg/day) administered over a lifetime. ,  However,

-------
                             -XVI -
it has not been found to interact with DNA or to cause
mutations.  It is not an initiator and therefore, can be
classified as a promoter.  The mechanism of dioxin promotion
activity is not completely understood.

     The promotion model for the action of' dioxin can be used
to explain cancer caused in laboratory animals.  Liver tumors
arise "spontaneously" in rats that are not"treated with any
chemical.  Therefore, the presence of initiated cells does not
depend on a specific treatment.  The excess of tumors seen in
dioxin-treated animals could result from the chemical hastening
the process from initiated cell to cancer cell so that more
tumors developed in the exposed animals'  lifetimes.

G.   The Importance3 of the Decision as To
     Whether or Not a Chemical is a Carcinogen

     The decision about whether or not a chemical is a
carcinogen is the most important step in estimating what level
of exposure is associated with human risk and what level of,
exposure is permissible.

     It is assumed that thresholds exist for most toxic effects
and that doses below the threshold will not cause harm.  For
instance, a chemical that causes liver damage in laboratory
animals is a potential cause of liver damage in humans.
However, the laboratory studies can also establish that there
are doses below which the .chemical does not cause liver
damage.   At those doses, the chemical might be metabolized and
excreted from the body before toxic effects are manifest or the
toxic effects may be so few that fio damage is detectable.
There is, of course,  some possibility that some humans would be
more sensitive to the chemical than are the test animals.

     Allowances are made for those possible differences:
First, a No Observed Adverse Effects Level (NOAEL)  is
determined in laboratory animals.   In practice,  that is
generally the highest dose that does not  cause a detectable
adverse effect.   Then the NOAEL is divided by a "safety factor"
of 100 or 1,000 to set an Acceptable Daily Intake (ADI) level
for humans.   The saf.ety factor of  100 incorporates  a factor of
10 to allow for possible differences in sensitivity between
animals  and humans and a factor of 10 to  allow for  differences .
in sensitivity in the human population.   Depending  on the
quality of the experiment,  how long the animals  were exposed
and observed for instance,  an additional  factor  of  10,  based on°
judgement,  may be incorporated.   The NOAEL plus  safety factor

-------
EPA
                   United States
                   Environmental Protection
                   Ag«ncv
•Off
-------
    2.  OVERVIEW OF DICHLOROMETHANE CANCER HAZARD/RISK ISSUES







     In May of 1985, EPA's Office of Toxic Substances found that



DCM met the criteria for priority review under section 4(f)  of

              *                        v

the Toxic Substances Control Act (TSCA).   Underpinning the 4(f)



decision was the conclusion that DCM should be considered a



probable human carcinogen, as defined by EPA's Guidelines for



Carcinogen Risk Assessment.

                          \.

     In assessing the cancer hazard posed by exposure to DCM,



primary consideration was given to the evidence of



carcinogenicity from the NTP's animal studies (1985).  The NTP



carcinogenesis bioassays clearly demonstrate that DCM is



oncogenic in two species of laboratory animals, rats and mice,



exposed at different dose levels via the primary route of human



exposure to DCM, inhalation.



     In the mouse bioassay, DCM induced a dose-dependent,



statistically significant  increase  in liver and lung adenomas and
                                   t*to                             • *


carcinomas in male  and  female mice  exposed through  inhalation for



a lifetime at concentrations of 2000 or 4000 ppra.   Tumor



incidences were as  follows:  at 2000 ppm,' 30/48 female mice and



27/50 male mice developed  lung tumors; 16/48  female mice  and



24/49 male mice developed  liver tumors.   At 4000 ppm,  41/48



female mice and 40/50 male mice developed lung tumors;  40/48



female mice and 33/49 male mice developed liver tumors.



      In the rat bioassay,  DCM  induced a  statistically significant



increase  in benign mammary gland  tumors,  of a type not expected

-------
                       V
                        ,0^
             ,.c°
                           ^
       tf   #
   <*\
      ^'^
     ~ <*
U°  O ^


 U0° j
          •cP'


   "A
   • »''
   ?
      ^. CO



            ^v^;


        0-°
            ?e^  ^A
^V^  Xk.»*


 ^^  e
*    A.O
   «?t
               6-*
              t.»
O*

-------
to progress to malignant tumors (McConnell et al.,  1986),  at the
two highest doses in female rats exposed, at 1000,  2000,  or
4000 ppm.  Male rats developed mammary gland fibroadenomas at
4000 ppm, but only at a marginally significant rate.  The NTP
interpreted their study as showing clear evidence of animal
carcinogenicity, and data from the NTP bioassay on mice are the
basis of the regulatory agencies' estimates of human risks at
expected human exposures.
     A study of Syrian golden hamsters exposed to DCM at inhaled
doses of 500.to 3500 ppm was negative, but several chronic
studies of mice and rats, including inhalation studies by Dow
Chemical Company (1980, 1982) and a drinking water study by the
National Coffee Association  (NCA) (1982 a, b; 1983), reported an
increase in tumors in rats and mice at sites corresponding to the
sites observed in the NTP bioassay.  One of the Dow studies
                 *
(1980) (inhalation at 1500 to 3500 ppm) reported an increase in
salivary gland sarcomas in male rats.  These tumors have hot been
repeated in other studies.  Results of the DOw and NCA studies,
conducted at doses below those used in the NTP bioassays, were
not statistically significant, with the exception of"the salivary
gland tumors in male rats.
     Based on an estimated risk comparison with the NTP bioassay
data, EPA concluded that despite the lack of statistical
significance, the results of the Dow and NCA studies were not
clearly inconsistent with those of the NTP bioassays.  For
example, comparing the NCA and NTP unit risk numbers  (estimated

-------
using the multistage model) for liver tumors in male mice,  the
95% upper confidence limit (UCL)  for the NCA study was estimated
to be 0.78 x 10~3; the UCL derived from data on male mice in the
NTP Study was 0.195 X 10~3 (U.S.  EPA, 1985b).
     At the time of the 4(f)  decision, data on humans exposed to
DCM in the workplace were considered to be inadequate for judging
carcinogenic potential.  Data from two epidemiologic studies did
not show evidence of a significant increase in deaths from lung
or liver cancer in exposed workers, but these studies had
insufficient statistical power to detect increased risks as
predicted using the upper-bound estimate derived from the NTP
bioassay on mice.
     Based on the evidence, EPA concluded that DCM should be
classified as a probable human carcinogen, group B2.  This
classification signifies that evidence of animal carcinogenicity
as provided by the NTP bioassays is sufficient, but data from
human studies are inadequate.  CPSC, FDA, and OSHA, after
reviewing the DCM database, came to similar conclusions.
     In response to EPA's 4(f) announcement in 1985 and the
initiation of investigations by CPSC, OSHA, and FDA, a number of
comments and studies were submitted to the  federal agencies
advancing reasons why the results of the NTP bioassay on DCM in
rats and mice should not lead to the conclusion that DCM presents
a high risk to humans.  The major criticisms of the preliminary
assessments suggest that  (1) current DCM risk  estimates
overestimate risks to humans because they  ignore  species

-------
 differences in metabolism and pharmacokinetics;  or (2)  the
 carcinogenic response shown by mice is unique to that species,
 i.e.,  the mechanism by which DCM causes cancer  in mice is not
 expected in humans.                                            .
     Addressing these criticisms calls for a brief review of DCM
 metabolism.   DCM is metabolized in mice, .the species  which  showed
 a clear carcinogenic response,  by two routes; one mediated  by the
 cytochrome P-450 oxidative system [often referred to  as the mixed
 function oxidase (MFO)  pathway],  and the.other  by the
 glutathione-S-transferase system (also known as  the GST pathway).
 Both pathways may be active in mice at low doses,  but at higher
 doses  the MFO pathway becomes saturated and the  metabolic load is
 increasingly shifted to the alternative GST pathway.   Recent
 studies (CEFIC,  I986e)  indicate that the GST pathway  is less
 active in rats,  hamsters,  and humans than in mice.
     Arguments against the. conclusion that DCM presents a risk to
 humans take  the position,  in general,  that the carcinogenicity of
 DCM  is due to reactive metabolites produced by the GST metabolic
 pathway,  and that this pathway  is significantly  active only
 following saturation of the MFO pathway,  i.e., only at high
 doses.   Further,  the GST pathway  is assumed to be  the sole
 carcinogenic pathway,and to be  far less  active in  humans than in
mice,  the  test species  in which malignant  tumors have been
observed.  Finally,  some hypothesize that  the metabolites of the
GST pathway  are not  reactive with DNA, but initiate cancer  in
mice through  some alternative mechanism such  as  specialized cell

                                8

-------
toxicity or increased, cell turnover, events unlikely to occur at
low doses and .possibly irrelevant to humans.  One might conclude
from these assumptions that the human risk for developing tumors
from exposure to DCM is very low, that it may not exist below
some threshold level, or that there may be no risk to humans
whatsoever.

-------
           A  REPORT  BY THE FD&C RED NO. 3 PEER REVIEW PANEL
AN INQUIRY INTO THE MECHANISM.OF  CARCINOGENIC ACTION OF FD&C RED NO. 3
               AND ITS SIGNIFICANCE  FOR RISK ASSESSMENT
                                     Prepared  by:

                                     Dr.  Ronald  W.  Hart,„ NCTR/FDA (Chairman)
                                     Dr.  Thomas  Burka, NIEHS/NIH
                                     Dr.  Scan  C. Freni,  CEH/CDC
                                     Dr.  Robert  Furrow,  CVM/FDA
                                     Dr.  David W. Gaylor, NCTR/FDA
                                     Dr.  Theodore Meinhardt, NIOSH/CDC
                                     Dr..  Bernard Sass, NCI/NIH
                                     Dr.  Elizabeth K. Weisburger,  NCI/NIH
                                     Executive  Secretaries -
                                      Dr. Paul Lepore, ORA/FDA
                                      Dr.. Angelo Turturro , • NCTR/FDA
                                July,  1987

-------
                                                                 July,  1987
                             TABLE OF CONTENTS
SECTION
PAGE
ACKNOWLEDGMENTS         ,                                                 ii

TABLE OF CONTENTS                                                       iii

EXECUTIVE SUMMARY                                                        vi

CHAPTER 1 - INTRODUCTION       •                                            1

  A.  The Recent Use and Certification of FD&C Red No. 3 (R-3)             1
  B.  Issues Confronting the FDA on R-3                                    1
  C.  FDA Commissioner's Charge to the Peer Review Panel                   2
  D.  Some Major Considerations in the Panel's Evaluation                  3
  E.  Conclusion                                                           6

CHAPTER 2 - CHEMICAL DISPOSITION AND METABOLISM                            7

  A.  Chemistry                                               "             7
   Figure 1                                                                7
  B.  General Considerations for Chemical Disposition  Studies              8
  C.  Studies of the Disposition of R-3 in Rats                            9
      1.  Study 1: Oral Administration of R-3  to Male  and Female Rats      9
          la.  Study 1 Phase I - Excretion Balance of  Orally
               Administered R-3 in Rats.                                 10
          Ib.  'Study 1 Phase II - Absorption and Distribution of
               Orally Administered R-3.        ,       '                  10
      2. Study 2: Intravenous  Administration of R-3 to Male Rats         12
          2a.  Study 2 Phase I - Excretion Balance of  Intravenously
               Administered R-3 in Male Rats.       \                    12
          2b.  Study 2 Phase II - Distribution and Blood Clearance of
               Intravenously Administered Rr3  in male  Rats.            .  13
      3.  Discussion of Studies 1 and 2 in Rat                    .       13
  D.  Studies of Disposition in Humans                                   15
      1.  Study 3: Oral Administration of R-3  to'Humans .                 15
      2.  Study 4: Oral Administration of 131I-R-3 to  Humans  ._           15
  E.  Summary                                                      .      16

CHAPTER 3 - SHORT-TERM TESTS FOR GENETIC TOXICITY                        18
                   C                                                 '
  A.  Mutagenicity       .              .                                  18
  B.  DNA/Damage Repair                                                  19
  C.  Chromosomal Damage                                      .           19
  D.  Transformation                                                     20
  E.  Discussion of Test Results                                      ,20
  F.  Activity of Metabolites  and Contaminants                           21
  G.  Significance of Results                                            22
                                      iii

-------
                                                                  July, 1987
 CHAPTER 4 - ANIMAL TOXICITY STUDIES
A. Introduction
B. Long-term Carcinogenicity Studies
1.
2. *
3.
4.
Rats
Mice
Gerbil
Dog
   C.   Reproductive Effects
       1.   Rat
       2.   Rabbits
 .  D.   Behavioral Effects
   E.   Analysis
   Tables  1-3            '

 CHAPTER 5  -  THYROID ANATOMY,  FUNCTION,  AND  PATHOLOGY

   A.   Introduction
   B.   Thyroid Anatomy
   C.   Thyroid Function                  .                      ,
   D.   Thyroid Hormone Production
   Figure  2
   E.   Pathology of Alterations of Growth  of  Follicular Epithelium
   F.   TSH  and Oncogenesis
   G.   Effects of R-3
       1.   Pathology Findings  in Studies with R-3 in  Rats
       2.   Discussion of the Pathology and Toxiciology Findings
       3.   Changes  in Thyroid  Economy
  H.   Other Possible Mechanisms
       1.   Binding  to Brain
       2.   Binding  to the Pituitary
       3.   Direct Effect on Follicular Cells        '.
      4.   Increase in Sensitivity to Iodide
      5.   Binding  to TBG and TBPA in Humans
  I.   Summary
   Tables 4-11

CHAPTER 6-- EXPOSURE AND TOXICITY CONSIDERATIONS RELEVANT TO
            RISK ESTIMATION

  A.   Introduction
  B.   Pharmacokinetics
  C.   .Hormonal Effects  of R-3
      1.   Human Studies          =
      2.   Analysis
      Pathology
      Maximum Tolerated Dose in a Chronic Toxicity Bioassay
      No-effect  Levels  of  R-3
      Relevance  of  Animal  Toxicity  to Humans
      Human Exposure to- R-3
   Table  12         .       .
  I.   Weight-of-Evidence for Carcinogenicity
  J.   Summary
 24

 24
 24
 24
 25
 26
 26
 26
 26
 27
 27
 27
 28

 33

 33
 33
 33
 34
 35
 36
 37
 39
 39
 40
 41
 44
 44
 44
 45
 45
 45
 46
 47
55

55
56
56
56
57
61
61
63
63
66
67
70
72
                                    .iv

-------
                                                                 July, 1987
CHAPTER 7 - ISSUES IN QUANTITATIVE RISK ESTIMATION

  A.  Introduction
  B.  Animal Tumor Data
      1. Follicular Cell Tumors
   Tables 13-15
      2. C-Cell Tumors
      3. Summary
  C.  Dose-Response Models                                        /AriTv
  D.  No-Observed-Effect-Level (NOEL) and Acceptable Daily Intake (ADI)
  E.  R-3 Risk Estimates                               ,
  F.  Discussion
   Tables 16,17
   Figures 3 and 4

CHAPTER 8 - RISK CHARACTERIZATION

  A.  Introduction
  B.  'Characterization  of Assumptions
  C.  Analysis
  D.  Conclusions

ADDENDUM - SUGGESTED  STUDIES

REFERENCES

APPENDICES    '         .       '          '

  APPENDIX A -  Charter, Agenda,  and. Minutes  of  the Meetings
                held by  the  Panel
73

73
74
74
75
78
78
78
79
80
82"
84
86

87

87
88
94
96

98

102
 Al
  APPENDIX B - Administrative Record of Panel materials
                                                                          Bl

-------
                           B.2
HAZARD IDENTIFICATION
                 Attribute 2  The report highlights critical aspects of data quality.
                     SOURCE  Case Study D. Formaldehyde (Pages 1-7 to 1-14).
                         Note  Contextually, this excerpt was preceded by a treatment of non-
                                cancerous effects and followed by a discussion of additional cancer-
                                related information.

-------
          Assessment  of  Health Risks

to Garment Workers and Certain Home Residents

        from Exposure to Formaldehyde
                 April 1987
 Office of  Pesticides  and Toxic Substances
    U.S,  Environmental Protection Agency

-------
1.2.1.  Studies of Humans

     The EPA has examined 28 epidemiologic studies relevant to

formaldehyde.  Three of these studies, two cohort* (Blair et al.,

1986; 1987 in press; .Stayner et al., 1986) and one case-control2

(Vaughan et al., in press),  were well conducted and specifically

designed to detect small to moderate increases in formaldehyde-

associated human risks.  Each of these three studies observed

statistically significant associations between respiratory site-

specific cancers and exposure to formaldehyde or formaldehyde-

containing products.  These associations are noteworthy since

during, inhalation, tissues in the nose, nasal sinuses, buccal

cavity (mouth), pharynx,3 and lungs come into direct contact with

formaldehyde.   In each of the above three studies, the

populations studied were also undoubtedly exposed to other

chemicals and these exposures may have contributed to the

observed increases in cancer risk.  Only the study by Vaughan

et al. (1986a,b) controlled for smoking and alcohol consumption.
  A cohort study follows a group of exposed individuals  for a
specified time period and measures the incidence of site-specific
deaths.  The observed number of site-specific deaths which
occurred in the time period are compared to the number of site-
specific deaths which would be expected based on jnortality rates
of a standard population.              -                 ,

2 A case-control study  identifies cases with the .disease of
interest and controls who do not have the disease.  The  cases and
controls are compared with respect to past exposure.
 ^ The pharynx  is  the passage between  the nasal  cavity  and  the
 larynx.  The nasopharynx, hypopharyrtx, oropharynx,  and
 laryngopharynx comprise the pharyngeal region.
                                1-7

-------
     The Blair et al. (1986; 1987 in press) cohort study observed

significant excesses in lung and nasopharyngeal cancers among U.S.

workers occupationally exposed to formaldehyde at 10 industrial

sites.  Blair et -al. (1986), however, argued that the lung cancer

excesses provided little evidence of an association with

formaldehyde exposure since the lung cancer risk did not increase

consistently with either increasing intensity or cumulative

formaldehyde exposure.  EPA, after reviewing the data, has

concluded that the,  significant excesses in total lung cancer

mortality, in analyses either with or without a latency period

equal to or greater than 20 years, and together with nasopharyngeal

cancer mortality among formaldehyde-exposed workers are meaningful

despite the lack of significant trends with exposure.

Misclassification of exposure (or lack of specificity between

exposure categories) and categorization of deaths into four

exposure levels which lowers the power to detect small increases in

risk, may have accounted for the observed lack of a significant

dose-response relationship.  -The significance of these findings is
                     *
reinforced by the fact that the site of the tumors seen in humans

(the nasopharyngeal region) is similar to that seen in animals.

Blair.et al.  (1987) performed further analyses of the

nasopharyngeal cancers regarding exposure to  formaldehyde and

particulates.  For  those workers with particuiate exposure,•the

trend between increasing nasopharyngeal risk  and increasing

cumulative formaldehyde exposure was not  statistically significant,

however, the authors concluded that  formaldehyde and particulates

appeared  to be a  risk  factor  for nasopharyngeal cancer.

-------
     The Stayner et al. (1986) cohort study reported statistically

significant excesses in mortality from buccal cavity tumors among

formaldehyde-exposed garment workers.  The standardized mortality

ratio (SMR), a ratio of the observed number of deaths to an age-

adjusted number of deaths expected in the group, was highest among

workers with a long duration of employment (exposure) and follow-

up period (latency).  A significant excess in deaths from cancer

of the tonsils was also reported, but there were too few-deaths to

examine any trends with exposure.

     Results from the case-control study by Vaughan et al.

(19S6a,b) showed a significant association between nasopharyngeal

cancer and having lived 10 or more years in a "mobile home".

Persons for whom this association was drawn had  lived in mobile

homes that were built in the 1950s to 1970s.  This study also

reported significant associations between sinonasal cancer  and

orohypopharyngeal cancer and exposure to resins, glues, and

adhesives (SAIC, 1986).4  No significant trends  were found  in

cancer incidence at any of these sites with respect to

occupational formaldehyde exposure; however, the risk estimates
                         >                .               '
for the highest exposure level and cancers of the orohypo-  and

naso-pharynx appeared elevated.  As stated earlier, however, this

population, like the two previously discussed, .was also

undoubtedly exposed to other chemicals which may have contributed

to the observed increases  in cancer risk.
^Several  residential and occupational  characteristics  were _a
priori selected  as  likely  surrogates  for  formaldehyde  exposure,
Among these were mobile home residency and occupational  resins,
glue, and adhesive  exposure.
                                1-9

-------
     'EPA previously had reviewed 25 other epidemiologic studies.


These studies had limited ability (lower power) to detect small to


moderate increases in formaldehyde-related risks due to (1) small


sample sizes; (2) small numbers of observed site-specific deaths;


and (3) insufficient follow-up.  Even with these potential


limitations, six of the 25 studies (Acheson et al., 1984a; Hardell


et al., 1982,; Hayes et al. , 1985; Liebling et.al.', 1984; Olsen  et


al.,  1984; Stayner et al., 1985) reported significant associations


between excess site-specific respiratory (lung, buccal cavity,  and


pharyngeal) cancers and exposure to formaldehyde.'


     The Olsen et al. (1984), Hayes et al. (1986), 'and Hardell  et
          i

al. (1982) studies reported significant excesses of sinonasal


cancer in individuals who were  exposed to both formaldehyde and


wood-dust, or who were employed in particleboard manufacturing


where formaldehyde is a component of the resins used to make


particleboard.  Only the.  Hayes  et al. (1986) and Olsen et'al.


(1984) studies controlled for wood-dust exposure; the detection


limits in both studies,.however, exceeded corresponding expected


excesses in the incidence of sinonasal tumors and, therefore, no


significant excesses were likely to have been observed.


     The Acheson et al. (1984a) study conducted in the United


Kingdom supports the results of Blair et al. in that, when


compared to mortality rates of  the general population, significant


excesses in mortality from lung cancer were observed in one of  six


formaldehyde resin producing plants in England.  A trend of


borderline significance with dose'was observed for this one


plant.  Acheson et al. concluded that the increases in mortality
                               1-10

-------
from lung cancer were not related to formaldehyde exposure since



the elevation and trend were not statistically significant when
                                                 *»


compared with local lung cancer rates.  EPA believes that the



risks and trends from analyses using local lung cancer rates as



the comparison risks appeared sufficiently increased for



corroborative use.           *



     The remaining two studies reported-significant excesses- of



buccal cavity cancer among garment workers in 3 plants (Stayner e.t



al., 1985) and excesses of buccal cavity and pharyngeal cancer



among formaldehyde resin workers in 1 plant (Liebling et al.-/



1984).  Portions of the Liebling et al. (1984) and Blair et al.



(1986, 1987) studies overlapped as did portions of the two Stayner



et al.* (1985; 1986) studies.  However, the non-overlapping



portions and improved design of "the more recent studies (i.e.,



Blair et al. 1986,  1987; Stayner et al. 1986) reinforce the



conclusions of the earlier studies.



     Analyses of the remaining 19 epidemiologic studies have



indicated the possibility that observed leukemia and- neoplasms of



the brain and colon may be associated.with formaldehyde



exposure.  The biological support for such postulates, however,



has not yet been demonstrated.
                               l-ll

-------
     Based on a review of these studies, EPA has concluded that

there is "limited" evidence to indicate that formaldehyde may be a

carcinogen in humans.5  Nine studies reported statistically

significant associations between site-specific respiratory

neoplasms and exposure to formaldehyde or formaldehyde-containing

products.  This is of interest 'since inhalation is the primary

route of exposure in humans.  Although the common exposure in all

of these studies was formaldehyde, the epidemioiogic evidence is

categorized a's "limited" primarily due to possible exposures to

other agents which may have confounded the findings of excess

cancers.

1.2.2.  Studies in Animals
                                                     . **
     The principal evidence indicating that formaldehyde causes

cancer in animals comes from studies conducted by the Chemical

Industry Institute of Toxicology (CUT)  (Kerns et al., 1983) and'

those by Albert et al. (1982) and Tobe et al. (1985).  The CUT

study was a well, conducted, multidose inhalation study in rats and

mice.   In this study, a statistically significant increase in

malignant tumors  (i.e.,*squamous cell carcinomas) was seen in the

nasal cavities of male and  female rats dosed at 15 ppm.  In

addition, a small increased incidence of squamous cell carcinoma,

while not statistically significant, was seen in male mice.
* EPA's Guidelines  for Carcinogen  Risk  Assessment define  limited
evidence of carcinogenicity  in humans as  indicating  that  "...a
causal interpretation is credible, but  that alternative
explanations, such  as chance, bias, or  confounding,  could not
adequately be excluded."
                               1-12

-------
Because  this  type of  nasal  lesion  is  rare  in  mice,  these data can

be considered to have biological importance.   Benign  tumors  (i.e.-,

polypoid adenomas) were  seen  in male  rats .in  the  CUT study  at all


dose  levels and in female rats, exposed  to  2 ppm of  formaldehyde.

Notably, the  dose-response  curve for  the benign tumors  in  this

study was not linear; the tumor incidence was highest at 2.0 ppm

and decreased at higher  doses.


     Tobe et  al. also observed a statistically significant

increase in the numbers  of  squamous cell carcinomas in  the same

strain of male rats as was  used in the  CUT study.  Albert et al.

reported a statistically significant  elevation of the same
         •    '    •                                      „        .
malignant tumor type in  male  rats of  a  different  strain.   In both

the Tobe et al. and Albert  et al. studies benign  squamous cell

papillomas were seen.  This observation was in contrast to the

CUT study in which polypoid  adenomas were the only benign tumors

observed.  Hamsters have been tested  in long-term inhalation

studies (Dalbey,  1982) but  no increased incidence of  tumors was

seen in formaldehyde-treated animals.   However, deficiencies in

the study design and poor survival.limit the interpretation of the

results from these, studies.


     Additional studies  in  animals that indicate an association

between exposures to formaldehyde and cancer are those by Dalbey

(1982) in which formaldehyde enhanced the production  of tumors

induced by  a known animal carcinogen  (i.e., diethylnitrosamine);

Mueller et  al. (1978) in which formalin (a water solution of

formaldehyde)  produced lesions in the oral mucosa of  rabbits which

showed histological features of carcinoma in situ; and studies by
                               1-13

-------
Watanabe et al'.' (1954; 1955) in which injections of formalin and

hexamethylenetetramine (from which formaldehyde is.-liberated in

vivo) produced sarcomas  (malignant tumors) and one adenoma  (benign

tumor) at the site of injection.

     Based upon a review of these studies, EPA has concluded that

there is "sufficient" evide'nce of carcinogenicity of  formaldehyde

in animals by the inhalation route.6  This finding is based on  the

induction by formaldehyde of an increased incidence of  a  rare, type

of malignant tumor  (i.e., nasal squamous-cell carcinoma)  in both

sexes of rats, in multiple  inhalation- experiments, and  in multiple

species  (i.e., rats  and  mice).  In these  long-term  laboratory

studies, tumors were'not observed .beyond  the  initial  site of nasal

contact  nor have other.mammalian  in  vivo  tests  shown  effects at

distant  sites.     •      •

1.2.3.   Additional  Supportive  Evidence                    .  .

      Other  relevant  information which is  considered  in  carcinogen

assessments  include  results from  short-term  tests  designed  to

measure effects  of  a chemical  on  DNA.  Tests for point  mutations,

numerical  and  structural chromosome  aberrations,  DNA

damage/repair,  and  in vitro cell  transformation provide evidence

 for the potential  mechanisms of carcinogenicity.   A battery of
 6 EPA'3 Guidelines for Carcinogen Risk Assessment define
 sufficient evidence of carcinogenicity from studies in
 experimental animals as indicating that ".. .there is ^increased
 incidence of malignant and benign tumors:  (a)  In multiple
 species or strains; or (b)  in multiple experiments (preferably
 with different routes of administration or using different dose
 levels); or (c)  to an unusual degree with regard to incidence,
 site or type of tumor, dose-response effects, as well as
 information from short-term tests or on chemical structure..
                                1-14

-------
                          B.2

HAZARD IDENTIFICATION
                Attribute 3  A weight-of-the-evidence approach is presented for judgment
                              as to the likelihood of human carcinogenic hazard and
                              includes a clear articulation of the rationale for the position
                              taken.
                   SOURCE   Case Study J. Red Dye'No. 3 (Pages 70-72).

-------
           A REPORT BY THE FD&C RED NO. 3 PEER REVIEW PANEL
AN INQUIRY INTO THE MECHANISM OF CARCINOGENIC ACTION OF FD&C RED'NO. 3
               AND ITS SIGNIFICANCE FOR RISK ASSESSMENT
                                    Prepared by:

                                    Dr. Ronald W. Hart,  NCTR/FDA  (Chairman)
                                    Dr. Thomas Burka, NIEHS/NIH'
                                    Dr. Stan C. Freni,  CEH/CDC
                                    Dr. Robert Furrow,  CVM/FDA
                                    Dr. David W. Gaylor,  NCTR/FDA
                                    Dr. Theodore Meinhardt, NIOSH/CDC
                                    Dr. Bernard Sass, NCI/NIH
                                    Dr. Elizabeth K. Weisburger,  NCI/NIH
                                     Executive Secretaries -
                                       Dr. Paul Lepore,  ORA/FDA
                                       Dr. Angelo Turturro, NCTR/FDA
                                 July,  1987

-------
                                                                  July,  1987

I.  Weight of Evidence for Carcinogenicity
    Relative  to  many chemicals,  there  is  substantial  information on  the
exposure and  toxicity of  R-3.   Interpretation of these  data  depends  on the
assumptions used and the weight given  to various factors  in  their  analysis.
The Panel has come to the following conclusions:
    a.  There is  some  evidence that certified  R-3  has an oncogenic  effect
        on rats, i.e., it increases  the incidence  of benign  thyroid  folli-
        cular adenomas.
    b.
There  is  equivocal evidence that  certified  R-3 is carcinogenic  in
male rats.  The effect  is only  seen  by  one reviewer,  and  it  is  only
marginally  significant.  In  addition,  interpretation  of   the
important  response 4%  dose  level is complicated  by evidence  that
the  MTD  may  have  been  exceeded.    This  may  have  triggered  a
mechanism  not  occurring  at  lower  doses and,  thus, not  at human
exposure levels.
Further, the above effects  of  certified  R-3 have been observed  in
two-generation  rat studies  and have not been  reproduced in other
studies in rat and other species.  While  some of  these  studies  were
inadequate in  design,, at  least two  negative mouse studies .were  of
sufficient size and duration to be adequate  one-generation studies.
There is some suggestion that R-3 is not a tumorgen in  a one-gener-
ation  test.    Lack of  reproducibility  in   the  tumorigenic   effect
weakens confidence that R-3 is a human  tumorgen.
    c.   Short-term tests  of R-3 have been negative for mutagenicity in bac-
        teria  or mammals.  Thus, there is  evidence  that  tumor formation by
        R-3  is not through a mechanism directly affecting the genome.

    d.   There  is some  evidence that  the tumorigenic  and hormonal effects of
        R-3  are exerted  by  its iodine component, either  molecularly bound
        or present as  an  impurity.   The reported R-3 induced alterations in
        thyroid  economy,  with  the exception  of the effect  on the T, - T,
        balance,  and  s.ome of  the  results in short-term  tests,  are  well
        known  properties   of  iodide.   There  is  insufficient  evidence  that
        the  tumorigenic effect of R-3  is  attributable  to  R-3 proper".
                                     70

-------
                                                             July, 1987

e.  There is  some evidence  that,  in rats,  the tumorigenic  effect  of
    certified R-3 is  mediated  through  the hypothalamus-pituitary-
    thyroid axis in which TSH has a central role.

f.  There is  insufficent  evidence  that the  mechanisms  controlling the
    hypothalamus-pituitary-thyroid axis in  man are  qualitatively
    different  from  those"  in rats.   Although  the human  baseline TSH
    level is  about 1/10  of that  in rats  and  although TSH  levels  in
    males are not higher  than  in females, (unlike rats), available data
    on  quantitative  differences in  the  response  to R-3  exposure are
    insufficent to allow  conclusions be drawn  pertaining  to  a differ-
    ence between humans and rats in  the sensitivity to R-3 and iodine.

g.  There is  some evidence  that thyroid   cancer  in humans may  not  be
    related  to  TSH levels.   The  higher  incidence of  tumors in women
    compared  to man, different  than  in rats, also suggests a different
    mechanism than that in rat.

h.  There have been no epidemiologic studies on R-3'.  However, the low
    incidence of  thyroid  cancer  in combination  with the low tumorigenic
    potency  of  R-3,  as predicted  by the   rat studies,  would require  a
    descriptive study  size far  beyond what is  reasonably feasible.  It
    is  unknown' whether there is a  sufficient number of people 'exposed
    to  high  dose  levels  to  conduct an  occupational  follow-up  study.
    Because  of  the low potency,  more cancer cases  would be .required  for
    case-control  studies  than are  usually  available.

i.  There  is some evidence  that the human  exposure  level  from  R-3  in
    food and drugs,  estimated  to be 1.41  mg/d,  is  below  the  NOEL of  20
    to  60  mg/d for an increase  in TSH levels  in  humans.   However,  if
    TSH does  not  play  a role in  the  genesis of  human  thyroid  tumors,  it
    can be  questioned what  the  value for  tumorigenesis  is  of a human
    NOEL for a  TSH effect.

j.    In summary,  there  is  limited  evidence  for certified  R-3   to  be
                                        *
    oncogenic in  humans•
                                  71

-------
                                                   • •   •           July,  1987

 J.   Summary
     The Panel concludes  from the preponderance of  available evidence  that
 R-3  probably acts on the hormone economy by processes occurring outside the
 thyroid,  with no  evidence for a direct mechanism.   Evidence  for an indirect
 or   secondary mechanism  includes   the  demonstrated  absence  of  relevant
 genetic toxicity, the association  between elevated TSH  and tumorigenesis,
 and  the association between R-3 exposure  and  TSH  elevation.  Following the
 Panel's ad-hoc definition,  R-3  is, thus,  considered  to  act predominantly
 through a secondary  mechanism.   For the  estimation of  the risk,  it  has been
 shown that  the intestinal absorption of R-3  is  poor,  and that  the absorp-
 tion rate in humans  approximately equals  that in  rats.   For humans,  a NOEL
 is taken  to  be 20 to 60 mg/day, although  the  data for these values  are of
 poor quality.   The  NOEL is not necessarily equivalent to a  threshold dose.
For the U.S.  population  as  a whole', the average per capita  availability of
R-3 in  food  products and  drugs is estimated to be  1.4 mg/day.
                                    72

-------

-------
                           B.2
HAZARD IDENTIFICATION
                 AttMuUte *l  The report identifies research that would permit a more
                               confident statement"about human hazard.
                     SOURCE  Case Study J. Red Dye No. 3 (Pages 98-101).
                       '  Note  The excerpt from the report outlines suggested studies on mechanism
                               of action, which would be useful in clarifying both hazard and dose- •
                               response extrapolation.

-------
           A REPORT  BY THE  FD&C  RED  NO.  3  PEER REVIEW PANEL
AN INQUIRY INTO THE MECHANISM OF .CARCINOGENIC ACTION OF  FD&C  RED  NO.  3
               AND ITS SIGNIFICANCE FOR RISK ASSESSMENT
                                     Prepared  by:

                                     Dr.  Ronald  W.  Hart,  NCTR/FDA (Chairman)
                                     Dr.  Thomas  Burka, NIEHS/NIH
                                     Dr.  Scan  C.  Freni,  CEH/CDC
                                     Dr.  Robert  Furrow,  CVM/FDA
                                     Dr.  David W. Gaylor,  NCTR/FDA
                                     Dr.  Theodore Meinhardt, NIOSH/CDC
                                     Dr.  Bernard  Sass, NCI/NIH
                                     Dr.  Elizabeth  K. Weisburger,  NCI/NIH
                                     Executive  Secretaries -
                                      Dr.  Paul  Lepore,  ORA/FDA
                                      Dr.  Angelo Turturro, NCTR/FDA
                                July,  L987

-------
                                                                  July, 1987
                        ADDENDUM  -  SUGGESTED  STUDIES
    Since the exact mechanism(s)  of  action of R-3 on  the  rat  thyroid  which
could  result  in activation  of  secondary mechanisms  such as  increased  TSH
and T^  levels and decreased Tg  levels  are not  known,  additional  studies
which would be  useful  to  understand  .the mechanisms of R-3  oncogenicity  and
help in extrapolation are outlined here.
     1.
     2.
                         131
 A radioactive  scan  of     I  labeled  R-3  to  study  the  possible
 localization in  thyroid  should  be  carried out  at a  higher  dose
 level  than  in  .the  studies  performed  by  the  petitioner.    The
 material should  be. pure,  however,  with  no contaminating  radio-
 active  iodine.    Since  there  is  dehalogenation,  appropriate
 controls should  be used  to  distinguish R-3 localization and iodine
 localization.   More advanced techniques,  such  as nuclear  magnetic
 resonance and  deuterium-labeled compounds,  should be considered.

 The  effect  of  the  sex hormones on the thyroid  should  be  assessed
 since there are' differences  in  the  two  sexes.   For  instance,  in
 males the  incidence' of  thyroid neoplasms  was  significantly  in-
 creased.   Animal  experiments  provide  support  for the  concept  of
 sex-influenced  development  of  thyroid  tumors.    In  studies  of
 irradiation-induced thyroid tumors  in rats,  it  was  reported  that
 more  than twice  as many male  rats developed thyroid  adenomas 'and
 carcinomas  as  females  (161).    In  other studies where only  male
 rats  were used, X-ray or fadioiodine induced, tumor incidences  were
 enhanced  when serum TSH  .levels  were  elevated  by  treatment  of '
 animals  with.methylthiouracil  or propylthiouracil  (162).   Hoffman
 et al.  (163) found that  the incidence of irradiation  induced  thy-
 roid  tumors  in  rats could be significantly reduced by  castration
 prior to  irradiation  and  the  effect  of castration  can be  reversed
 by administration of testosterone.    Although  the results varied
with  the  age at   administration,  testosterone  increased the inci-
dence of tumors   up to  100%.   The  authors also  showed  that TSH
levels  in  both  irradiated intact^ and  castrated rats treated with
 testosterone were  higher  than   in  untreated intact  males.   The
                                    98

-------
                                                             July,  1987,

    effect of testosterone on  the  induction  and/or  progression  of  rad-
    iation-associated  thyroid  tumors  may  be  through  a  direct  or
    indirect mechanism.   It has been  shown  that ,TSH may  enhance  thy-"
    roid  tumor  development  by  growth-promoting  actions,  therefore,
    androgens  may  indirectly  modulate  growth  of  radiation-induced
    thyroid tumors through TSH.  Since,  in the  R-3  studies,  TSH levels
    were  suspected  of being elevated,  it would  be important  to  Jcnow
    whether or  not  there was  a testosterone-mediated  effect  for  the
    sex-specific  carcinogenic  action.   There  are  similar  considera-
    tions for estrogen effects.

3.  Further  studies  using  hypophysectomized   animals  could  be  very
    useful in elucidating the mechanism.  If hypophysectomized  animals
    treated with  a tumorigenic  dose of  R-3 did not develop  tumors,
    this  would  be further support for an indirect action using  TSH.
    Recent studies are exploring aspects  of  this.

4.  A chronic bioassay where R-3 is given at a  tumorigenic dose and in
    which some  agency  restores T-  levels would be  definitive proof of
    mediation through R-3.  It is  our understanding a study  similar to
    this is underway.

5.  The obvious occurrence  of  abnormal  lysosomes in  R-3  treated  ani-
    mals needs to be evaluated in  a quantitative  fashion,  since it may
    form  the  basis  of an  inferred dose-response relationship.   Such
    information 'would  not only be  useful in understanding the  mechan-
    ism of action of  R-3, but  it  may also  provide an alternative to
    tumor incidence in the calculation of the risk.

6.  One area totally ignored is the  effect of  R-3 on the  fetus.   In a
    multi-generation  study,  the critical effect  may  be on  the fetus,
    especially since R-3  influences  reproductive endpoints.   A series
    of studies  should  characterize this important  area, e.g.,  studies
    of fetal pathology and physiology.

7.  A useful  area would  be to  investigate  the  longitudinal  effect of
                                99

-------
                                                           .  July,  1987

    R-3  on  human  hormone  levels,  with  particular  emphasis  on
    randomizing  dose levels,  sampling  blood at  the  same time of  the
    day,  and  prescreening  for  pre-existing thyroid  pathology.    The
    dose  levels  of interest  are 50, 100, 250, and 1000 mg/d  for up to-
    one  month.    Hormone  levels should  be  determined  at  weekly
    intervals.   The actual assays  should be deferred  until they can be
    done  in  one  batch.   Compounds  to be  tested are commercial  R-3 and
    iodide  at  dose  levels comparable  to what would  be available  from
    the doses  of commercial  R-3.

 8.  It  is of crucial importance to assess  interspecies differences in
    sensitivity  to R-3  in a quantitative fashion.  Tests  of  two weeks
    should  be  done  in humans  and  rats in  a  strictly standardized way
    with  regard  to time  of  the day, day  of blood  sampling, randomizing
    of  dosing, exclusion or control  of  iodide  in the diet  from  other
    sources.  For  humans, compounds  to  be  tested and dose  levels are
    as  in proposal 7.  For  rats,  dose  levels should  be the  same  on a
    mg/kg dose basis, but should  also include .an extra  dose level of
    70  and 350 mg/kg-d, which  is  comparable to 5000 and  2500  mg/d in
    humans,  to account  for  possible greater sensitivity  of the  rat.

 9.    It  would   provide  important  information  to  conduct  a  study
    evaluating   the  predictive  ability of various procedures  for
    extrapolating  risk,  viz.,  biologically-based mathematical models
    and  the NOEL approach.    There is  currently a  substantially im-
    proved data base on human  carcinogens  and  human  risk of cancer at
    high  doses in animals and  low "and moderate  doses  in  humans.
                                    *
10.  Conduct  a  large scale  epidemiologic  study  among people  occupa-
   *  tionally exposed to high levels of iodine or  R-3.  Exposure to R-3
    may include  skin contact,  since  the  absorption of R-3 through the
     skin  is  of  the  same  order  of  magnitude as  through the gut  mucosa.
    Preferably,   the study  design should  be that  of a  case-control
     study nested in a retrospective cohort study.

11.   Conduct a 'new survey of food" consumption pattern, taking care  that
                                 100

-------
                                                      ...  July,  1987

representative  samples  are taken  of  the food products  (including
iodine-containing  commodities),  named  by  the  interviewees,   for
testing on  the content  of R-3  and iodine.   The analysis .should
cover amount of food  consumed,  variety of products within  a given
category,  and amount of R-3 and  iodine ingested.
                             101

-------
                  B.3

DOSE-RESPONSE
      EVALUATION
       AtfriDUIe I   Valid data sets and plausible models for high-to-low dose and
                       interspecies extrapolation are presented in dose-response
                       modeling.                    •
           SOURCE   Case  Study  H. Methylene  Chloride (Pages  71-87).
                Note   See General Attribute 2 in this Appendix./
           SOURCE   Case Study D. Formaldehyde (Pages 1-23 to 1-28).
                Note  The excerpts illustrate the consideration of several data sets. In this
                       case, only one model was selected for extrapolation.
           SOURCE   Case Study A. DEHP (Pages 111-145).
                Note  This report illustrates all three of the attributes for dose-response
                       evaluation. The report is included in its entirety to. provide contextual
                       background. The report is especially illustrative of consideration of
                       data sets of different types (bioassays, genotoxicity, species vari-
                       ations, and possible mechanisms). Several modeling approaches are
                       presented, along with strengths and weaknesses. The authors present
                       a range of risk estimates. The authors were unable to indicate a
                       quantitative central tendency of the potency estimates but indicated
                       qualitative considerations in weighing the range of risks.
                       Some other pertinent portions are found on Pages 121-127, 129-130,
                       and 138.

-------
DOSE-RESPONSE

     EVALUATION
      Attribute 1   (continued)
           SOURCE  Case Study B. TCDD (Pages v to xvi).


               Note  See Hazard Identification Attribute 1 in this Appendix.
           SOURCE  Case Study C. Ethylene Oxide (Pages'9-23).


               Note  This report illustrates the evaluation of several data sets and models
                     with particular emphasis on time-to-response modeling.

-------
         Assessment of Health  Risks

to Garment Workers and Certain Home Residents

       - from Exposure to Formaldehyde
                  April 1987
  Office  of  Pesticides  and Toxic  Substances
     U.S.  Environmental Protection Agency

-------
less intense exposure.  In addition,  the cellular effects are
expected to be reversible once formaldehyde exposure is
eliminated.
1.4.2.  Cancer Dose-Response Assessment
     In principal, data from studies of humans are preferred for
making numerical risk estimates.  However, as is" often the case,
the available epidemiologic data on formaldehyde were not suitable
for low dose quantitative cancer risk estimation, mainly because
of a lack of adequate exposure information in the studies.
Accordingly, results from studies in animals were used to estimate
low-dose human cancer risk.  In addition, even though the
epidemiologic studies were not suitable for quantifying a dose-
response curve, those studies with observed statistically elevated
cancer risks provided some support for the animal-based predicted
upper bound risk.  This comparison, while yielding valuable
information to the assessment, should be viewed with caution since
exposure levels in these epidemiologic studies were subject to
some variation.                                                  .
                                                             •
1.4.2.1.  Selection of Data
     Of the carcinogenicity studies with  formaldehyde  in  animals,
EPA has selected the  CUT study in rats as the best study for
cancer risk extrapolation.  This  study was well,designed, well
conducted, included multiple doses, and .used a  large number of
animals par dose.                   °    •   _
     Each of  the  remaining  inhalation  studies suffered from
various limitations which precluded their  use in quantitative risk
assessment.   The  CUT study in  mice showed a  limited  tumor
                               1-23

-------
 response oaly at. the highest dose of formaldehyde,  while the



 Albert et al.  (1982)  study had only a single formaldehyde-exposed



 group.  -Although the Tobe et al.  (1985)  study contained multiple



 dose  groups,  a tumor  response was seen only at the  highest dose,



 and  the number of animals per group was  relatively  small.   Lower



 cancer  risks  than those  estimated from the  CUT study  in rats



 would have been predicted had the Agency been able  to  use  the  CUT



 study in  mice  for risk extrapolation,  while higher  cancer  risks



 would have been estimated had the results from the  Tobe et al.



 (higher by a  factor of ten)  or Albert  et al.  studies been  used.



      Two  types  of nasal  tumors were obs-erved in the CUT study in



 rats, squamous  cell carcinomas (malignant tumor)  and polypoid



 adenomas  (benign tumor).   EPA's risk assessment relied only on the



 malignant  tumor  data of  the.CUT  study to predict human cancer



 risks because: -(1) the malignant  tumor response in  formaldehyde-



 exposed rats was  definite  and  unequivocal in both .males  and



 females, whereas  the frequency of benign  tumors  reached
                                                  .>>


 statistical significance only  when  the incidences, in males  and .



 females were pooled; (2) the  malignant tumor response  in the CUT



 study in rats showed an increasing  dose-related  trend,  while-the



 benign tumor response showed  a decreasing trend; (3) unlike the



benign tumor response which was not confirmed by the other  rat



 inhalation studies, similar malignant tumor types were  found both



in all rat and mouse inhalation studies with formaldehyde and in a



study of acetaldehyde, a close structural analogue  of  formaldehyde.



     The appearance of benign  nasal tumors in rats  following



inhalational exposure .to formaldehyde in  the CUT study
                               1-24

-------
contributes to the qualitative weight-of-the-evidence that
formaldehyde may pose a carcinogenic hazard,  but because of the
attendant uncertainties they were not included in the
quantitative estimate of human cancer risk.   Had the Agency
chosen to use the benign tumor response in the quantitative
estimation of human cancer risk,  the predicted values would have
been about ten-fold greater than those reported in Section 1.4.3
using the malignant tumor response alone.
1.4.2.2.  Choice of Mathematical Extrapolation Model
     Since risks at low exposure levels cannot be measured
directly either by experiments in animals or by epidemiologic
studies, a number of mathematical models have been developed to
                 it
extrapolate from results at high doses to expected responses at
low doses.  The Office of Science and Technology Policy (OSTP)
published principles on model selection which states that:
     "No single mathematical procedure is recognized as the most
     appropriate for low dose extrapolation in carcinogenesis.
     When relevant biological evidence on mechanism of action _.
     exists, the models or procedures employed should be
     consistent with the evidence.  When data and information  are
     limited, however, and when much uncertainty exists regarding
     the'rwchanisra of carcinogenic action, models or procedures
     which incorporate low dose linearity are preferred when
     compatible with the limited information."
     Data relevant to selecting a model  for extrapolation of
cancer risk associated with  exposure to  formaldehyde were
reviewed? some of the biological information  support a direct
                               1-25

-------
relationship between exposure and carcinogenicity while other

data are consistent with a non-linearrresponse.  The Agency,

however, did not conclude that enough information was available

to propose an extrapolation model for formaldehyde that was

different from the one recommended by the OSTP and EPA's

Guidelines for Carcinogen Risk Assessment (i.e., linearized

multistage procedure).  The Agency has presented various other

models for comparative purposes.

     Biologic evidence on mechanism of action, which can,aid in

model selection,  largely is inferred from a variety of types of
                          J"
studies.  These are limited'and suggestive of several mechanisms
         i
for formaldehyde.  Mutagenicity studies suggest a direct

relationship (i.e., a linear one) between exposure to

formaldehyde and carcinogenicity.  Thus, the ability of

formaldehyde to cause point mutations, chromosome aberrations and

DNA damage is consistent with the chemical's ability to initiate

the carcinogenic reaction.

     The steep curvilinearity of the rat nasal carcinoma dose-

response data in the CUT study in rats suggests, however,  that

cancer development is greatly accentuated above ce'rtain con-

centrations.  In keeping with this observation are the results of

experiments.on DNA synthesis and cell proliferation following *

short-term formaldehyde exposures and the conversion of normal

mucosal cells to squamous cell epithelium (squamous metaplasia)

following longer exposures which indicate that certain toxic

effects are only noted above certain formaldehyde

concentrations.  Any relationship between cell proliferation
                               1-26

-------
following formaldehyde exposures and the carcinogenic process is

currently unknown.  Likewise, although squamous metaplasia may

represent a step in the formation of squamous cell carcinoma, its

specific role is uncertain.  No lesions that,may represent stages

in a continuum between the-squamous metaplasia and carcinoma were

identified in the CUT study.              ,

     The CUT also conducted molecular dosimetry experiments

attempting to relate ambient exposures to formaldehyde with

tissue-specific levels of  formaldehyde-DNA adducts.  Use of  the

data generated by these experiments in risk extrapolation models

yields lower estimates of  risk, sometimes significantly'lower

than use of the experimental doses.  The CUT data have been

reviewed by EPA scientists and a review panel of non-government

scientists to determine whether or not they should be used in the

quantitative risk assessment.  Both groups concluded that the

study had several shortcomings which preclude its use in

modifying the doses used in  quantitative risk assessment, and

they provided three reasons  for their conclusion.  First, the

experimental methodologies must be validated to assure that  the

experimental assumptions were  scientifically sound and that  the

formaldehyde-DNA-protein complexes were  identified properly;

second/ the single intracellular target used in the  study may be

inadequate; and third, and perhaps most  important, the use of an

acute exposure model in the  CUT study may not be  appropriate

because chronic,  not acute exposure is most relevant to  risk
                       i       '        ?
assessment.              •
                               1-27

-------
     Different"extrapolation models fit the observed data


reasonably well but there are large differences, among them in the


risks calculated at low doses.  EPA's Guidelines  for Carcinogen


Risk Assessment state, however, that goodness of  fit to the


observed tumor data,by a given model is not an effective means of


discriminating among models.  In the absence of compelling


biological evidence on the mechanism of action, as in the case
                                              3.   —.

for formaldehyde, EPA's guidelines specify that the linearized


multistage procedure will be used, with the possible presentation


of various other models for comparative purposes.  The analysis


showed that of the models examined, only -the one-hit model


produced higher risk estimates (about ten fold higher).   .


     Studies show that non-human primates and rats respond


similarly to formaldehyde exposure,.  Accordingly, an interspecies


scaling factor was not used in the risk extrapolation.  This


position was supported by the Consensus Workshop  on


Formaldehyde.  Consequently, the response of rats and humans was


judged to be the same at equivalent exposure levels and


durations.  However, if a conversion factor, such as nasal


surface area, had been used the estimated human cancer risks


would have been about an order of magnitude higher.
                               1-28

-------
JOURNAL OF THE AMERICAN COLLEGE OF TOXICOLOGY
Volume 4, Number 2, 1985
Maiy Ann Llebert, Inc., Publishers
     Assessment  of Possible  Carcinogenic  Risk
       to Humans Resulting from Exposure  to
           Di(2-ethylhexyl)phthalate  (DEHP)

                     D. TURNBULL and J.V. RODRICKS
                                ABSTRACT                          .

The purpose of this work was to estimate the degree of risk that might be associated with
human exposure to low levels of the plasticizer di(2-ethylhexyl)phthalate (DEHP). DEHP
is a common component, sometimes at high concentrations, of poiyvinyl chloride (PVC)
plastics and was recently reported by the National Toxicology Program (NTP) to be carci-
nogenic in rats and mice, inducing hepatocellular tumors in both species. This work was
also designed to illustrate an approach to risk assessment that attempts to incorporate all
available biological data. Based on the dose-response data generated by the NTP bioas-
says, we have performed extrapolations of risk to low dose levels using several proce-
dures, including some that incorporate inferences from the available data that shed light
on the likely relationship between dose level and risk at low dose levels. In drawing these
inferences, consideration was given to such factors as genotoxicity, metabolism and phar-
macokinetics, and physiological and biochemical effects of DEHP that might reveal its
mechanism of action. The relative merits of each of the various risk estimates are de-
scribed, based on current understanding of DEHP's mode of biological action. It is con-
cluded  that DEHP's mechanism of carcinogenicity in rodents most likely involves its abil-
ity to induce  peroxisome proliferation and  related enzymatic changes, although other
mechanisms cannot be excluded. If humans and rodents are assumed to be at the same
risk at the same daily dose level of DEHP, application of the various low dose extrapola-
tion models leads to the prediction that the  daily dose resulting in a lifetime risk of no
more than 1 in 1 million would be between  1.5 and  791 mg/kg per day, with the most
likely figure being 116 mg/kg per day. If the  carcinogenicity of DEHP is dependent upon
its pattern of metabolism, however, it would be inappropriate to extrapolate from rodents
to man without qualification because of the major quantitative differences in metabolism
in rats, mice, and primates, including man. One of the major differences in metabolism of
DEHP between rats and mice and primates is in production of a metabolite whose level
may be an indicator of the level of peroxisomal activity and, hence, if the peroxisome pro-
liferation theory of DEHP carcinogenicity is correct, of carcinogenic risk. However,  the
substantial doubt that exists regarding the applicability of rodent carcinogenicity data to
humans must be expressed in qualitative terms.
  Environ Corporation, Washington, D.C.
                                     Ill

-------
                          TURNBULL AND RODRICKS

                               INTRODUCTION

    Di(2-ETHYLHEXYL)pHTHALATE (DEHP), the structure of which is shown in Figure 1 along with
    those of its two primary hydrolysis products, is a widely used plasticizer for polyvinyl chloride
(PVC). It has been estimated that 1188 million pounds of plasticizers were used in PVC in 1980, and
of this total, 30% (about 356 million pounds) was DEHP.'" PVC plastics may contain up to 40%
DEHP by weight and are widely used in consumer products, such as imitation leather bags and fur-
                                               CH2CH3
               Di(2-ethylhexyl)phthalate  (DEHP)
                                      0
                                      C-0-CH0CHCH0CH-_CH.CH,
                                              /      2.   2.   Z
               Mono(2-ethylhexyl)phtha.late (MEHP)
                     CH2CH3
               2-Ethylhexanol  (2-EH)

               FIG. 1.  Chemical structures of DEHP, MEHP, and 2-EH.

                                      112

-------
                               RISK ASSESSMENT OF DEHP

nishings, wallpaper, lawn furniture, rainwear, swimming pool liners, flooring, footwear, children's
toys, containers and tubing for transfusions of blood and blood products.(1>  Since DEHP is not
chemically bound within the PVC, there is at least the potential for widespread human exposure to
DEHP as a result of migration out of the plastic. Human exposure to DEHP is of concern particu-
larly in light of the recent report by the National Toxicology Program that DEHP at high dietary lev-
els is carcinogenic in rats and mice.U)
  Estimation of the magnitude of human exposure to DEHP is, however, extremely complex be-
cause of the wide range of items that contain DEHP and the uncertainty regarding how much of the
DEHP content of a PVC item to which someone was  exposed would  reach that individual and how
much would be absorbed. No attempt has been made to estimate human  exposure in this paper.
Rather, our purpose is to describe the available data pertinent to the assessment of carcinogenic risk
resulting from exposure to DEHP and to present estimates of risk per unit of dose (or no-observed-
effect levels), which can subsequently be combined with estimates of human exposure to estimate
human risk.                                                                              :
  The conduct of this risk assessment follows the recommendations of the National Academy of Scir
ences.'31 It starts with a critical review and evaluation of the literature pertaining to the carcinogenic
properties of DEHP. This is followed by a review of data that might shed light  on the underlying
mechanism(s) of tumor induction in animals.  This  may assist in dealing with the next two compo-
nents of risk assessment: dose-response evaluation and interspecies extrapolation. The approach to
risk assessment used here is one in which several estimates Tare presented, along with a discussion of
their relative degrees of support based on current understanding of DEHP's biological behavior. We
avoid presenting only worst-case estimates but also attempt  to avoid overstating the degree of cer-,
tainty associated with the other estimates presented.
  The broad outline of this risk assessment process is presented in Figure 2. Each component of this
process is discussed in relation to available data on DEHP.
                                         REVIIW DATA ON
                                        CARCINOGfNICITY

                                      STRfNCTHS & WEAKNESSES
                                      , .  OF ANIMAL DATA
                                       Ml CHAN ISM .OF CANCtR
                                    INDUCTION  IN EXPERIMENTAL
                                           ANIMALS

                                     (includes all underlying
                                           processes)
                                                     METABOLIC SHIFTS
                                                     HIGH-TO-LOW DOSE
       Lifetime risk per
       unit or avoraqe.
       daily lifetime)
         exDOSuro
Li Tetimo  risk per
unit or average,
da ily Iifettme
   exposure
Lifetime  risk per
unit of average,
da ily Ii fetime
   exposure
                                          THRESHOLD DOSE FOR
                                          CANCER INDUCTION
                                                                       THRESHOLD MODEL
                                                                      - EXPERIMENTAL OR
                                                                        ESTIMATED NOEL
                    FIG. 2.   Broad outline of carcinogenesis risk assessment.

                                             113

-------
                              TURNBULL AND RODRICKS

                      * CARCINOGENICITY EVALUATION

  Few studies are available that evaluate possible adverse effects of chronic human exposure to
DEHP, and none of these specifically address the possibility of an association between DEHP expo-
sure and cancer. (4"S) The only evidence of earcinogenicity of DEHP comes from the recent NTP bio-
assay.<2) Earlier negative studies involving exposure of smaller numbers of animals to lower doses re-
vealed no significant carcinogenic effect. <7~10) These studies do not conform to current standards for
carcinogenicity bioassays,'11' however, and although they demonstrate that DEHP is not a potent
carcinogen, they were probably insufficiently sensitive to have detected an effect of DEHP of the
small magnitude seen in the NTP bioassay.
  In the NTP bioassay,(2> DEHP was fed in the diet for 103 weeks to groups of 50 B6C3F1 mice and*
50 F-344 rats  of each  sex at each of 2 dose levels, 3000 and 6000 ppm in mice and 6000 and 12,000
ppm in rats. Using the food consumption data presented by NTP, these feeding levels approximately
correspond to the following average doses:  320 and  670 mg'/kg per day in male rats, 390 and 770
mg/kg per day in female rats, 670 and 1300 mg/kg per day in male mice, and 800 and 1800 mg/kg
per day in female mice. In rats, mortality was not affected by treatment, but there was a dose-related
decrease in body weight throughout the study for males at both doses  and for females at the high
dose. Food consumption was reduced slightly in all treatment groups.
  The incidence of liver tumors increased in treated rats, as shown in Table 1. Using the Cochran-
Armitage test, there was a significant dose-related trend in the incidence of combined carcinoma or
neoplastic nodules in male rats (P = 0.002) and female rats (P < 0.001) and in carcinomas alone (P
= 0.002) and neoplastic nodules alone (P = 0.03) in female rats. Also, by direct comparison using
the Fisher's exacktest, there was a significant increase in the incidence of hepatocellular carcinoma
(P = 0.003) and neoplastic nodules (P = 0.028) in high dose female rats and in combined carcinoma
and neoplastic nodules in low dose (P = 0.012) and high dose (P < 0.001) female rats and in high
dose male rats (P  = 0.01). No other tumor type was significantly increased in incidence in rats.
  In mice, there was a dose-related decrease in body weight gain in female mice from Week 25 to the
end of the study, but  food consumption was within 4% of the control level in both treated groups.
No positive trends in mortaility were noted, but the low dose female mice had significantly shortened
survival compared to controls. Overall, survival to the end of the study varied between 50% and
78% in the various groups.
  The incidence of liver tumors was increased in treated mice, as shown in Table 2. By the Cochran-
Armitage test, there was a significant dose-related trend in the incidence of hepatocellular carcinoma
(P = 0.018) and combined carcinoma and adenoma (P = 0.002) in male mice and of the same tumor
types in females (P < 0.001 for both). By direct comparison using Fisher's exact test there were sig-
nificant increases in hepatocellular carcinoma in high dose male mice (P = 0.022) and low dose (P =
0.006) and high dose (P < 0.001) females. Combined carcinoma and adenoma were significantly in-
creased in all  groups of mice fed DEHP (P = 0.013 in low dose males, P = 0.002 in high dose males,
P = 0.001 in low dose females, and P < 0.001 in high dose females). Of the 57 treated mice of both
           TABLE 1.  INCIDENCE OF LIVER TUMORS IN RATS IN NTP BIOASSAY OF DEHP
                           Control
                   Low Dose
                            High Dose
                       Male
Female
Male
Female
Male
   "Significantly greater than corresponding control by Fisher's exact test (P < 0.03).

                                            114
Female
Hepatocellular
carcinoma
Neoplastic
nodules
Combined
1/50

2/50
3/50
0/50

0/50
0/50
1/49

5/49
6/49
2/49

4/49
6/49a
5/49

7/49
12/49*
8/50*

5/50*
13/50a

-------
                             RISK ASSESSMENT OF DEHP
          TABLE 2.  INCIDENCE OF LIVER TUMORS IN MICE IN NTP BIOASSAY OF DEHP
                          Control
                   Low Dose
                            High Dose
                      Male
Female
Male
Female
Male
Female
Hepatocellular
carcinoma
Neoplastic
nodules
Combined
9/50

6/50

14/50
0/50

1/50

1/50
14/48

11/48

25/48*
7/50*
•
5/50

12/50*
19/50*

10/50

29/50*
17/50*

1/50*

18/50*
  Significantly greater than corresponding control by Fisher's exact test (P < 0.05).
sexes having hepatocellular carcinoma, 20 (12 males and 8 females) had pulmonary metastases. None
of the controls had pulmonary metastases. No other type of neoplasm was increased in incidence in
treated mice of either sex.
  Northup et al.(l2) have suggested that the  maximum tolerated dose (MTD) was exceeded in all
treatment groups except low and high dose male mice and low dose female rats because body weight
gain was depressed by more than 10% in all other treated groups. NTP,<2) however, have pointed
out that the 10% weight differential is only a guideline and that the primary reason for not exceeding
the theoretical MTD is to avoid excessive early deaths, which might prevent tumor development, and
to avoid pathological changes other than neoplasia that might be involved^in secondary mechanisms
of carcinogenesis. NTP concluded that both of these goals  were fulfilled in the case of the DEHP
bioassay. However, as will be discussed in more detail later, there is support for the hypothesis that
peroxisome proliferation, which occurs in rats and mice at the doses of DEHP used in the NTP bio-
assay, "3il4) is involved  in a secondary mechanism of cancer induction. At low doses where peroxi-
some proliferation does not occur,  one would not expect cancer to develop.
  In both mice and rats, only liver tumors were increased in incidence. The relevance of liver tumors
in rodents to humans has been questioned, particularly because of the high and variable spontaneous
incidence of liver tumors in various strains of mice'15-16' and the high  spontaneous incidence in_
the livers of rats  of preneoplastic  cells  that can be stimulated by promoting agents to produce
tumors. <17M9> Such high incidences of preneoplastic cells are not known to exist in the human
liver,  which, therefore, would not be susceptible to enhancement of tumor incidence by such a
mechanism.
  Taking these  factors  into account is is concluded that under the conditions of the NTP bioassay,
DEHP at very high dietary levels was carcinogenic to mice and rats of both sexes. There does, how-
ever, remain some question of the relevance of these findings to lower dose levels and to humans.
This topic is discussed  in detail in a later section.
Strength of evidence of carcinogenicity

  As described above, the only evidence that DEHP is carcinogenic comes from the NTP bioassay in
which tumors at a single histogenic site (hepatocytes) were increased in incidence in rats and mice.
Earlier, less sensitive studies found no carcinogenic effects, and there is no evidence from epidemio-
logical studies that DEHP is carcinogenic. Several schemes have been developed for assessing the
strength of evidence that a particular chemical is a human carcinogen. <20~22)  In all 3 of these classifi-
cation schemes, DEHP is assigned to a low'category because the evidence of its carcinogenicity is rel-
atively weak, since it comes only from experimental animals,  based on neoplasms occurring at a
single histogenic site, induced at a relatively high dose level, with no supporting evidence of genotox-
icity (see following section). This  fact should be taken into account when the significance of the risk
is considered.

                                            115

-------
                               IURNBULL AND ROUR1CKS

                                    GENOTOXICITY

  The overwhelming preponderance of evidence indicates that DEHP and its metabolites, MEHP
and 2-ethylhexanol, are not genotoxic (Table 3), though a few studies have reported positive results,
mostly with MEHP. These are discussed briefly below.
DEHP     »

  DEHP has been found nonmutagenic .in at least 8 separate Ames assays. Only Tomita et al.<30)
have reported positive results, and their data are not convincing: results of only a single dose were re-
ported, and the increase was less than 2-fold. Tomita et al.<30) also reported increased chromosomal
aberrations and transformation in embryonic  cells  from Syrian hamsters treated trarisplacentally
with DEHP at 3.75-15 g/kg administered by gavage. It might be noted that the proportion of nor-
mal diploid cells in all cultures including the controls was'low (40-70%), suggesting chromosomal in-
stability or technical deficiencies in experimental procedures.
  The significance of the apparent positive results in the transformation assay is unclear,  since the
increase is small, and it would be desirable to repeat the study to show that the results were not due
simply to variations in background transformation-frequency in different animals. This test system is
not directly comparable to the 3T3-system used by Barber et al.,U9) since the latter is entirely in vitro
and the former involves treatment in vivo, permitting full metabolic activity. The possibility of a dif-
ference between in vivo and in vitro systems is also raised by the weakly positive results in dominant
lethal assays in mice reported by Singh et al.(44) and Autian.(4S> Interpretation of effects in the study
by Singh et al.(44) is difficult because mice treated at the high.dos&(25.56 ml/kg) had reduced fertil-
ity, which might be expected, since high doses of DEHP cause testicular degeneration  in rats'50-5"
and mice.'2' Hence, increases in early fetal deaths may be an indication of testicular toxicity rather
than mutagenicity. Effects at lower doses (1-10 ml/kg) reported by Autian(45) are not statistically
significant.
  Albro et al.<47) reported "association" of radioactivity with DNA when  14C-DEHP with label in
the ethylhexyl moiety was fed to rats.  No association occurred when the label was in the phthalate
group or when DEHP was saponified to phthalic acid and free  14C-ethylhexanol before being fed.
The lack of an effect with 14C-ethylhexanol suggests that labeling is not due only to incorporation of
label into a general metabolic pool. However, Von  Daniken et al.,(48) while finding similar associa-
tion of radiolabel from DEHP with DNA  under similar circumstances, also found  label in DNA
when 14C-ethylhexanol was given orally. They presented evidence that the results with DEHP were
caused by metabolism and incorporation of label into nucleotides and not due to covalent DNA
binding of the type seen with genotoxic carcinogens. More recent work by Albro et al.<52) also indi-
cates that association of label from 14C-DEHP  is  due to catabolism of the ethylhexyl moiety and in-
corporation de novo into normal DNA nucleotides.
MEHP

  Once again most of the reported positive results come from Tomita et al.<30) They report a dose-
related increase in toxicity to rec" compared to rec* Bacillus subtilis. This seems to be a genuine ef-
fect, since consistent results were obtained over a range of doses (100-500 rng/disc). These authors
also report an apparent dose-related increase in revertants in Salmonella (TA.100) and Escherichia
coli (WP2 B/r) treated with MEHP in suspension. The results are reported in terms of revertants/
survivor. This method of reporting results can be misleading unless the protocol used is appropriate.
For example, the standard Ames assay with Salmonella involves plating the bacteria in the presence
of a small amount of histidine. If histidine is incorporated following suspension treatment,  a treat-
ment  that is simply toxic can appear to be mutagenic, since the number of spontaneous revertants
appearing is governed by the number of cells that the histidine can sustain not by the number of

                                             116

-------
                             RISK ASSESSMENT OF DEHP
         TABLE 3. SUMMARY OF GENOTOXICITY TESTS ON DEHP AND ITS METABOLITES
Type of Test
Mammalian cell
  mutagenicity
DNA damage
In vitro cytoge-
  netics
                                           Results
Reference
                                       DEHP

Bacterial muta-       DEHP negative in standard Ames assay ± S9                    23
  genicity            DEHP reported "nonmutagenic in bacterial test"                  24
                    DEHP at up to 1000 ^g/plate negative in Ames assay with  *       25
                      strains TA98 and TA100
                    DEHP at up to 10 /xl/plate negative in Ames assay ± S9          26
                    DEHP at 0.1-10.0 mg/plate negative in 4 strains in Ames           4
                      assay ± S9
                    DEHP at lO'MO'2 M negative in Ames strains TA98 and          28
                      TA100 ± S9
                    DEHP at 0.15-150/tg/plate negative in 5 strains in Ames          29
                      assay ± S9
                    DEHP reported positive at 5 mg/plate in Salmonella strain         30
                      TA100 with S9; no data on other doses reported
                    DEHP at 50-2000 /tg/plate negative in Ames strains TA98         31
                      and TA100
                    Urine from rats given DEHP by gavage at 2 g/kg per day          32
                      for 15 days was negative in Ames assay ± S9 and  ± j8-
                      glucuronidase and arylsulfatase
                    DEHP at 25-10,000 pg/ml negative in V79/HGPRT assay         33
                       ± S9
                    DEHP at 7.8-250. nl/ml negative in mouse lymphoma TK          29
                      assay ± S9
                    DEHP at 0.016-1.0 /tl/ml without S9 and 0.067-5.0 /d/ml         26
                      with S9 negative in mouse lymphoma TK assay
                    No induction of unscheduled DNA synthesis (UDS) in pri-         34
                      mary rat hepatocytes in vitro with DEHP up to 10
                      mg/ml
                    DEHP negative in rat hepatocyte UDS assay                     29
                    No increase in alkaline elution of DNA from hepatocytes •         35
                      of rats given DEHP at 500 mg/kg per day for 14 days
                      and no induction of UDS in rat or human hepatocytes
                      treated in vitro at 0.1-10 mM or in hepatocytes from,
                      rats fed DEHP at 1200 ppm for 30 days
                    No significant effect of DEHP at up to 75 /tg/ml on chro-         36
                      mosome aberration frequency in human lymphocytes
                    No significant effect of DEHP at up to 60 /tg/ml on chro-         37
                      mosome aberration frequency in G0 human lymphocytes
                      or human fetal lung cells
                    No chromosome damage in human lymphocytes treated            38,
                      with DEHP at 0.16 mg/ml
                    No significant increase in chromosome breaks but slight in-         39
                      crease in SCE (not dose related) in Chinese hamster cells
                      at  up to 1 mM
                    No increase in chromosome aberration assay in CHO cells         40
                      at  up to 2 mM  (limit of solubility)
                    No increase in chromosome aberrations in Chinese hamster         41
                      cells
                                          117

-------
                              TURNBULL AND RODRICKS
                                  TABLE 3.  (CONTINUED)
Type of Test
                        Results
Reference
In vivo cytoge-
  netics
Dominant lethal
  assay
Cell transforma-
   tion
 In vivo DNA
  binding
 Bacterial muta-
   genicity
No increase in chromosomal aberrations in peripheral lym-         42
  phocytes of workers exposed to DEHP for average of 22
  years
DEHP at 5, 1.7, and 0.5 g/kg per day for 5 days by gavage         43
  caused no increase in chromosome aberrations in bone
  marrow cells in male rats
DEHP negative in micronucleus test in mice  at 5 g/kg-             29
  single dose  and 5 g/kg per day for 5 days
Apparent increase in chromosomal aberrations in cells cul-         30
  tured from  hamster embryos treated with DEHP trans-
  placentally  at 7.5 and  15 g/kg
DEHP at 25.56 mg/ml (2/3 IP LD50)  in male mice caused          44
  reduced fertility, reduced litter size, and increased early
  fetal deaths in first 3 weeks after treatment; slight effects
  at 1/2 and  1/3 LD50
Slight but not significant increases in early fetal deaths at          45
  1-10 ml/kg
DEHP at MTD (9.86 g/kg) and 1/2 and 1/4 MTD by ga-          46
  vage daily for 5 days "was negative in dominant lethal
  assay in mice
No increase in transformation in mouse 3T3 cells in vitro          29
  at 0.875-21.0 nl/ml without metabolic activation or at
  6.25-100 /d/ml with rat hepatocytes to provide activation
Apparent increase in frequency of transformation in cells          30
  of hamster embryos treated transplacentally at 7.5 and
   15 g/kg
Some incorporation of radiolabel into liver DNA in rats           47
  fed "C-DEHP  with label in ethylhexyl moiety but not if
  labeled in phthalate moiety; nature of attachment to
  DNA unclear
Similar results to  those of Albro et al.,(47) but found to be         48
  caused by incorporation into nucleotides via intermedi-
  ary metabolism

                    MEHP

MEHP "showed DNA-damage-provoking activity in B.            24
  subtilis and mutagenicity in E. coir" no details
MEHP negative in Ames test in TA98 and TA100 at up to         25
   1  mg/plate ± S9
MEHP negative at 0.002-0.2 /J/plate in Ames test                26
MEHP negative at 1.03-1030 /tg/plate ± S9 in Ames test.         29
   with 5 strains of Salmonella
Apparent dose-related increase in revertants  in E. coli             30
   (WP2) and slight increase in Salmonella (TA100) without
   S9
MEHP negative at 50-2000 j^g/plate in Ames strains TA98         31
   and TA100 ± S9
                                           118

-------
                              RISK ASSESSMENT OF DEHP
                                  TABLE 3. (CONTINUED)
Type of Test
                        Results
Reference
Mammalian cell
  mutagenicity
DNA damage
In vitro cytoge-
  netics
In vivo cytoge-
  netics
Dominant lethal
  assay
Cell transforma-
  tion
Bacterial muta-
  genicity
Mammalian cell
  mutagenicity
DNA damage

In vitro cytoge-
  netics
MEHP negative at 0.013-0.32 /J/ml in mouse lymphoma          26
  TK assay ± S9                      ,
MEHP negative at 0.081-1.25 mM in CHO/HGPRT             40
  mutagenicity assay
Apparent dose-related increase in differential inhibition in         30
  repair-deficient bacteria (Rec assay)
MEHP at "toxic and nontoxic levels" negative in primary          34
  rat hepatocyte UDS assay
MEHP induced significant increases in chromosome aber-         40
  rations in CHO cells at 0.8-1.75 mM but no increase in
  SCEsatO.7-1.3 mM
Apparent increase in SCEs in V79 cells at 25-50 /tg/ml            30
MEHP at 0.14, 0.05, and 0.01 g/kg per day for 5 days            43
  caused no significant increase in chromosome aberra-
  tions in bone marrow cells
Apparent reproducible increase in the frequency of micro-         29
  nucleated cells in bone marrow of female mice given IP
  doses of MEHP at 125 mg/kg per day on 2 successive
  days; no effect with a single dose or in males
Apparent increase in chromosome abnormalities in cells           30
  from hamster embryos treated transplacentally at 375-
  1500 mg/kg
MEHP at 50, 100, and 200 mg/kg per day for 5 days by         •  46
  gavage was negative in assay in mice
MEHP negative in mouse 3T3 transformation test at 25-           29
  120 nl/ml without metabolic activation and at 5-125
  nl/ml with rat  hepatocytes to provide activation

                2-Ethylhexanol  .

Negative in Ames assay ±  S9 at 0.01-1.0 /ul/plate                 26
Negative in 5 strains in Ames assay ± S9 at 0.002-1.8            29
  jd/plate
Small (less than 3-fold) increase in mutation frequency            49 •
  matched by reduction in survival with 2-EH at 0.5-
  1.5 mM
Negative in Ames strains TA98 and TA100  ± S9 at 0.1-           28
  10 mM
Urine from rats given 2-EH by gavage at 2 g/kg per day           32
  for 15  days was negative in Ames assay ± S9 and ±
  /3-glucuronidase/arylsulfatase
Negative in mouse lymphoma assay at 0.018-0.24 ul/ml           26
  ± S9
2-EH negative "at nontoxic and toxic levels" in primary rat „      34
  hepatocyte  UDS assay
Negative  in CHO chromosome aberration assay at 1.5-            40
  2.8 mM
                                           119

-------
                               TURNBULL AND RODRICKS
                                    TABLE 3. (CONTINUED)
 Type of Test
                         Results
Reference
 In vivo cytoge-
   netics
 Cell transforma-
   tion
2-EH negative in micronucleus test in mice treated once or         29
  twice with 456 mg/kg per day IP
2-EH negative in bone marrow chromosome aberration            43
  assay in rats at 0.21, 0.07, and 0.02 g/kg per day for
  5 days
Negative in mouse 3T3 cells at 48-225 nl/ml without meta-        29
.  bolic activation and at 0.011-0.162/il/ml with  rat hepa-
  tocytes for activation
viable cells plated. Unfortunately, details of the procedures used were not reported; therefore, this
finding cannot be evaluated.
  A dose-related increase in chromosome aberrations and transformation in hamster embryo cells
treated transplacentally at 375-1500 mg/kg of MEHP was also reported by Tdmita et al.<30) Our
comments on these findings are the same as those we made in connection with the work of these in-
vestigators on DEHP. They also report a small increase in SCEs in Chinese hamster V79 cells treated
in vitro with MEHP at 25-50 mg/ml for 24 hours and mention —but give no details of—an  increase
in mutations in the azaguanine/thioguanine resistance system with V79 cells in vitro and Syrian ham-
ster embryo cells treated transplacentally.
-  Overall, the report of Tomita et al.(30) suffers from a lack of details about the procedures used.
Without this and some of the supporting data that are'missing from the report, a detailed evaluation
is impossible. However, some of the reported positive results are in systems sufficiently different in
endpoint (B. subtitis rec assay) or method of treatment (transplacental assays) from other published
negative results that they can not simply be considered inconsistent (e.g., if intact mammalian me-
tabolism is needed for expression of activity, Tomita's positive results with transformation do not
necessarily conflict with the negative results found by Barber et al.(29) in 3T3 cells in vitro).
  Yagi et al.<24)  report in .an abstract that MEHP "showed DNA-damage provoking activity in B.
subtilis and mutagenicity in E. coli," but  no details are given. Some  of these authors are from the
same laboratory as Tomita et al.,<30) and they may be reporting data from the same studies.
  Phillips et al. <40> reported an increased frequency of chromosomal aberrations in Chinese  hamster
(CHO) cells in vitro treated for 2 hours with MEHP at 0.8-1.75 mM. In particular, there was a dose-
related increase in chromatid interchanges in addition to  increases in chromatid and chromosome
breaks. These does levels were in the toxic range and reduced cell survival to between about 65% to
less than  10% of control values. Unlike Tomita et al.,<30) Phillips et al.(40) found no increase in SCE
in CHO cells treated with MEHP, but Phillips et  al. used CHO cells and treated for 2 hours,  whereas
the positive results reported by Tomita et al.(30) involved  V79 cells treated for 24 hours.
  Barber et al.<29) observed a slight, reproducible increase in the frequency of cells with micrdnuclei
among polychromatic erythrocytes in the bone marrow of female mice given IP doses of MEHP at
125 mg/kg per dose on 2 successive days. No effects were seen in males given this treatment or in
either sex given a single dose of 125 mg/kg. Although the increase was-significant when compared to
a concurrent solvent control group, the frequency, observed was within the range of historical con-
trol values.      '         ,      *
  Albro et al.(47) found "association" of .label with DNA when 14C-MEHP labeled in the ethylhex-
anol moiety was  fed to rats.  Labeling was less than resulted with an equivalent amount of 14C-
DEHP. The nature of this "association" is not known for certain, but the study by Von Daniken et
al.(48) mentioned above in relation to DEHP suggests that incorporation of  label via intermediary
metabolism may be responsible.

                                             120

-------
                              RISK ASSESSMENT OF DEHP
2-Ethylhexanol (2-EH)
  The only suggestion of positive results with 2-ethylfiexanol comes from an azaguanine-resistance
assay in Salmonella.<49) There was a slight, apparently dose-related increase in azaguanine-resistant
mutants in bacteria treated with 0.5, 1.0, and 1.5 mM EH. The increase in frequency of mutants/
survivor was small (less than 3-fold) and was matched by a reduction in survival, however, so that
there was no increase in the absolute number of mutants at any dose. Such a response is not gener-
ally considered conclusive  evidence of mutagenicity.


Discussion

  As we have noted and as is evident from' Table 3,  the overwhelming preponderance of evidence
suggests that  DEHP, MEHP, and 2-EH have little or no propensity for direct interaction with and
alteration of  DNA. A few studies suggest genotoxic activity, but, as described above, in most of
these cases serious questions exist regarding methodology or the significance of the  results. How-
ever, there are a few findings that cannot be dismissed, and it appears that there remains some uncer-
tainty regarding the capacity of at least MEHP to display some degree of genotoxic potential in some
systems.
  If the proposed mechanism of DEHP carcinogenesis is correct (see next section), one would expect
genetic damage to be induced in vivo (or in in vitro systems with peroxisomes present) at doses capa-
ble of« inducing peroxisome proliferation.  In this sense DEHP/MEHP may be threshold genotoxic
agents, which would not be expected to display genotoxic activity if the test systems used did not
provide for the presence of peroxisomes or if the DEHP/MEHP doses were insufficiently high to
cause peroxisome proliferation.
               MECHANISM OF TUMOR INDUCTION IN RODENTS

   Recent investigations into possible mechanisms of chemical carcinogenesis have emphasized that
 more than one biological process may be involved in cancer development.<53) The influential work of
 the Millers'54-"' emphasized the fact that a large proportion of known carcinogens either are them-
 selves, or are metabolized to, electrophiles. These react with and damage nucleophilic sites in the .
 cell, particularly DNA, and are, therefore, genotoxic. Damage to DNA provides a basis for explain-
 ing the permanent nature of the neoplastic state on the basis of a heritable (at the cellular level) alter-
 ation in the genome.
   Although the ability to damage DNA seems to be an important property of many carcinogens —
 the so-called genotoxic carcinogens —some substances appear to increase the incidence of tumors in
 experimental animals without interacting directly with DNA.  Examples include tumor promoters
 and cocarcinogens, such as phorbol esters, that  increase the incidence of tumors when applied in
 conjunction with a genotoxic carcinogen, and some hormones, particularly estrogens.'53' Also in-
 cluded in this group  of epigenetic or nongenotoxic carcinogens are such chemicals as saccharin,
 DDT, and perhaps carbon tetrachloride.'53'
   As discussed in the previous section, DEHP has no substantive genotoxic activity and, therefore,
 belongs in the general class of nongenotoxic carcinogens. Several hypotheses have been proposed to
 explain the mechanism of action of various nongenotoxic carcinogens.(S3) Thus, immunosuppressive
 drugs may permit tumors to develop from transformed cells that would normally be detected and de-
 stroyed  by the body's immunological system. Some chemicals may alter the activation or detoxificaT
 lion of other carcinogens so that higher levels of an active metabolite are present.
   There is  considerable support for the hypothesis  that DEHP  belongs  to a diverse class of
 nongenotoxic carcinogens whose mechanisms of action involve induction of peroxisome prolifera-
 tion. '"•"' In this section, the support for this hypothesis and its implications concerning the shape
 of the dose-response curve for tumors in rodents and, hence, its implications for assessing the risk of
 tumor development at low doses are discussed.

                                             121            ,

-------
                               TURNBULL AND RODRICKS

   There is one other aspect of the behavior of DEHP that influences the dose-response*relation, and
 this is the relationship between administered dose and target-site dose over a range of dose levels
 (i.e., dose-related pharmacokinetics). The term "target-site dose" is used here to mean the dose in the
 target organ (liver) of the substance (DEHP or one or more of its metabolites or by-products of its
 metabolism) that is ultimately responsible for causing (directly or indirectly) the observed increase in
 the incidence of liver tumors. The term has been used in the past to refer to the dose of a genotoxic,
 active metabolite of a proximate carcinogen. We use the term here  in a more general sense, with no
 implication  of genotoxicity. A study of the dose-related pharmacokinetics of DEHP in rats has re-
 cently been  completed, and the results of the study appear to be useful for defining the likely dose-
 response relation.
   In addition to catalase, which breaks down' hydrogen peroxide, peroxisomes contain several en-
zymes that catalyze reactions that  generate hydrogen peroxide. These include several oxidase en-
zymes, such as L-a-hydroxyacid oxidase, D-amino acid oxidase, and urate oxidase.1581 Recent studies
have demonstrated that peroxisomes contain a system of enzymes involved in the /3-oxidation of
fatty acids, which also generates hydrogen peroxide.(71)
   The hypothesis that there is a relationship between peroxisome proliferation and liver carcinogen-
esis in rodents was proposed by Reddy et al.<"> on the basis of their findings with a structurally di-
verse group of chemicals, including some drugs used in the treatment of hyperlipidemia (clofibrate,
nafenopin, Wy-14,643, BR-931, and tibric acid). All 5 of these chemicals caused hypolipidemia (re-
duction in serum lipid levels, particularly triglyceride levels), liver enlargement (hepatomegaly) with-
out necrosis, proliferation of liver peroxisomes, and hepatocellular carcinoma in mice or rats, but
none were mutagenic in the Ames assay or caused DNA damage in the lymphocyte 3H-thymi.dine in-
corporation assay. Since then, additional chemicals have been found  to display this same set of
effects.'72'
   Although an association between peroxisome proliferation and development of hepatic  tumors
seems clear for several chemicals, there may be some exceptions. Acetylsalicylic acid (aspirin) is a
weak inducer of peroxisome proliferation, but there is no evidence that it is carcinogenic.'72' Fenofi-
brate, a hypolipidemic drug related to clofibrate, caused hypolipidema, heptatomegaly and peroxi-
some proliferation in Sprague-Dawley rats at dose levels of 50-1000 mg/kg per day for 3 months,
but no significant increase in liver tumors was seen in Sprague-Dawley rats fed fenofibrate at 10 or
60 mg/kg per  day for 117 weeks or in Swiss  mice fed fenofibrate  at 50 mg/kg per day for 92
weeks.(73) However-, the dose levels  used in the carcinogenicity studies were fairly low and probably
                                           122  .

-------
                               RISK ASSESSMENT OF DEHP

did not represent maximum tolerated doses as used by NTP with DEHP. It is, therefore, unclear
whether fenofibrate is a true exception to the correlation pattern or whether testing at higher dose
levels would reveal carcinogenic effects. This is highlighted by studies with bezafibrate, another hy-
polipidemic drug related to clofibrate. Although bezafibrate caused hypolipidemia, hepatomegaly,
and peroxisome proliferation in rats at dose levels of 10-500  mg/kg per day for 1 week, it caused no
increase in liver tumors in Sprague-Dawley rats when fed at 300, 750,  or  1500 ppm for  24-26
months.(74) However, when the dose level was increased to 6000 ppm in the diet, liver tumors  devel-
oped in the female rats.'"'
  In considering how chemicals that cause peroxisome proliferation might induce cancer, one may
start by examining data on the physiological and biochemical effects of peroxisome proliferation
that might be relevant to carcinogenicity. Reddy and co-workers'28'75' have proposed that increased
production of hydrogen peroxide by the peroxisomes is responsible for the carcinogenic effects. Evi-
dence for this mechanism is incomplete, but it appears to be the best  available explanation for the
carcinogenic activity of DEHP in rodents.
  Although peroxisomes contain several enzymes that catalyze reactions in which hydrogen peroxide „
is produced,<58-60-7 most of these enzymes are little affected by. chemicals that cause  peroxisome
proliferation.(7I) However, peroxisomal enzymes involved in /3-oxidation of fatty acids (which form
hydrogen peroxide as a by-product) are substantially increased in activity in liver cells from rats fed
hypolipidemic drugs that cause peroxisome proliferation. <71'77' Similarly, induction of peroxisomal
^-oxidation has since been demonstrated with DEHP.'78'82'  The pathway for peroxisomal /3-oxida-
tion of fatty acids is illustrated in Figure 3. As indicated there, the reduced FAD produced by fatty
acyl-CoA oxidase is coupled directly to oxygen, producing hydrogen peroxide, rather than being
coupled to the respiratory chain, as it is in the mitochondria. (60-7U Because it is not coupled  to the
respiratory chain, the peroxisomal /3-oxidation pathway is insensitive to cyanide; thus, an indepen-
dent measure of peroxisomal /3-oxidation can be made by using cyanide to inhibit selectively mito-
chondrial /3-oxidation.t78-79-82'  Rats fed high dose levels of DEHP show substantial increases in per-
oxisomal /3-oxidation activity per mg protein or per gram of liver.(83"8S1 Canning et al.(82) found a
slightly lesser increase in j3-oxidation in rats fed the same concentration of MEHP but found no ef-
fect with phthalic acid or 2-EH. Similar results with- MEHP- and 2-EH were found by  Morton.(80)
 Recent work by Mitchell et al.(86) indicates that 2 metabolites of MEHP (metabolites VI and XI in
 Figure 5) also induce /3-oxidation and may be the proximate inducers of peroxisome proliferation.
   Of particular importance to our consideration of mechanism of carcinogenesis, these increases in
 /3-oxidation activity were not accompanied by correspondingly large  increases in catalase activ-
 Ity^s.s^ai) -phis difference in extents of increase in /3-oxidation and catalase activity is important be-
 cause, as mentioned before,  catalase normally inactivates the potentially hazardous hydrogen per-
 oxide generated by the /3-oxidation system, although glutathione peroxidase also plays  a  role in
 disposing of hydrogen peroxide. Since the increase in /3-oxidation activity induced by DEHP is much
 greater than the increase in catalase, it is possible that this creates an imbalance, resulting in excess
 levels of hydrogen peroxide in the cell. The consequence of excess intracellular levels  of hydrogen
 peroxide could be detrimental to the cell. (61-87' Addition of hydrogen peroxide to mammalian cells ih
 culture causes chromosome damage*64-"' and induces sister chromatid exchange.(66'67-69'70' Hydro-
 gen peroxide can react with DNA, causing alteration and  liberation of DNA bases and backbone
 breakage.("' Also, recent Japanese studies, although not entirely conclusive, suggest that hydrogen
 peroxide may have tumor-promoting or cocarcinogenic activity, <88-89) and it is mutagenic in a newly
 developed strain of Salmonella bacteria.'90'
    In addition to these effects of hydrogen peroxide itself,  it can form the highly reactive hydroxyl
 radical either by spontaneously splitting to  form 2 hydroxyl radicals'"' or by reacting with  ferrous
 iron:'*7'

      Fe II + HiO, - Fe HI  + OH' + OH'

 or by reacting  in the Haber-Weiss reaction with superoxide ion,  which may be generated by the
 microsomes:'61'
      HaO2 + O; - O»  + OH" + OH'
                                             123

-------
                        TURNBULL AND RODRICKS
               R-CH -CH -COOH
                       Fatty acyl-CoA' synthetase


           R-CH2-CH9-CO~S-CoA
                                FAD
                          Fatty acyl CoA
                              oxidase
                              FADH
H2°2
            R  -  CH  = CH-CO~S-CoA
                           Enoyl hvdratase
             R-CHOH-CH-CO~S-CoA
          .Catalase
                                   NAD         •

                        L-3-Hydroxy fatty acyl CoA  dehydrogenase

                                   \ADH + H+
              R-CO-CH-CO~S-CoA
             CoA-SH / \ Thiolase
,R-CO~S-CoA
              FIG. 3. Pathway of peroxisomal 5-oxidation of fatty acids.

                                 124           .

-------
                               RISK ASSESSMENT OF DEHP

  The hydroxyl radical is highly reactive and can cause DNA damage directly, <61-621 or it can initiate
lipid peroxidation, which in turn yields several chemicals, such as fatty acid hydroperoxides, choles-
terol hydroperoxide. endoperoxides, fatty acid and cholesterol epoxides, enals and other aldehydes,
and alkoxy and hydroperoxy radicals that have been shown to exert rriutagenic, promoting, or carci-
nogenic effects."11
  It is recognized that there are limitations to this hypothesis. The most important of these limita-
tions is that there is no evidence that excess levels of hydrogen peroxide are generated in the livers of
rats fed high levels of DEHP. Nor is there evidence of toxic effects, such as lipid peroxidation, that
might be expected to accompany excess levels of intracellular hydrogen peroxide. However, it is not
clear to what extent such effects have been looked for,  and there is some evidence of increased
steady-state levels of hydrogen peroxide and increased levels ofjipofuscin (indicative of lipid peroxi-
dation) in the livers of rats treated with other peroxisome proliferators.'72'.
  The proposed mechanism of carcinogenicity is summarized in Figure 4, which shows only the es-
sential outline of the peroxisome proliferation hypothesis. It is not intended as a comprehensive de-
scription of the fate and biological effects of DEHP.
               DOSE-DEPENDENT PHARMACOKINETICS OF DEHP

   Before discussing the available data on the pharmacokinetics of DEHP in rodents, it would be
useful to review briefly the major pathways of metabolism of DEHP. Our understanding of the me-
tabolism of DEHP is due largely to the work of Albro and co-workers, on whose work the following
summary is based. M7-"-92) The initial step in metabolism of orally administered DEHP in all species
studies is hydrolysis by a nonspecific lipase in the gut to yield MEHP and 2-EH. In the liver, MEHP
undergoes «- and (u-l)-oxidation to yield the metabolites shown in Figure 5. The metabolite num-
bering system is that of Albro et al.'47-"-92' Several other minor metabolites have recently been iden-
lified."11 In rats and mice, metabolite V, a product of co-oxidation, may then undergo /3-oxidation
with loss of a 2-carbon unit to yield metabolite I. This /3-oxidation is important because, as discussed
above, peroxisomes contain the enzymes involved in fatty acid /3-oxidation and may, thus, be in-
                           Enters Intermediary Metabolism
              TCHP  ""e"lntm-HFH11 <• l-Ethyliiexdiiul
                                                                    motivation {Strand Breaks)
                                                                                •Carcinomas 7
                                                                    React ive
                                                                    Nodular  —
                                                                    Hyperplasia
-Carcinomas
                                             02 *• H2<3
 FIG. 4.  Schematic of the peroxisome proliferation hypothesis. At high doses of DEHP, it is pro-
 posed that excess H2O2 or other oxygen species are produced in excessive amounts because catalase
 production does not increase as rapidly as peroxide production.
                                              125

-------
                                TURNBULL AND RODRICKS
                             :o
                     w-i oxidation
                     R-CH-CH_-CH,-CH-CH,
                      CH,

                      1  "•
                      CH,
                     R-CH-CH,-CH,-C-CH,
                      I    -   'I   '
                      CH,      0
                      !  '
                      CH,
                                         CH,


                                         CH,
                                       R-CH-CH,-CH,-CH,-CH.
                                       R-CH-CH,-CH,-CK,-CH_
CH,
                                                          R-CH-CH,-CH,-CH,-CH,OH

                                                            CH,
                  R-CH-CH, -CH, -CH, -COOH


                   CH,
      a-oxidacion


   R-CH-CH,-CH,-CH,-CH3

    COOH
                                                                    8-oxidat ion
                                                               R-CH-CH -COOH
                                                                    2
FIG. 5.  Major pathways of metabolism of MEHP in rats. The pathways illustrated are inferred
from knowledge of the structures of the urinary metabolites shown, based on the work of Albro et
aj (47.83.92) Metabolites are numbered according to Albro's convention.


volved in this step of the metabolism of DEHP in rodents, generating hydrogen peroxide. However,
metabolite I has not been identified in humans and is only a minor metabolite in primates, (47-94> ap-
parently because humans achieve water solubility of DEHP metabolites by glucuronide formation
rather than by extensive oxidative metabolism. Peroxisome proliferation is much less extensive in
primates and presumably in humans than in rodents given DEHP.""" Hence, if peroxisomal /3-oxi-
dation is important  in forming metabolite I in rodents, lower levels of metabolite I would be ex-
pected in primates, in agreement with the experimental findings.<47-*"
   The influence of dose level and previous exposure on the metabolism of DEHP in rats has been in-
vestigated. <"' The most important finding in the context of this risk assessment was that in rats with
prior exposure to DEHP, the percentage of the dose excreted in urine as metabolite I increased dis-
proportionately with increased dose. The percentage of the dose excreted in the urine as metabolite I
doubled from about 1 l°7o of the dose at 1000 ppm to about 25% of the dose at 6000 ppm, and more
than doubled to 31 % of the dose at 12,000 ppm. This increase in the percentage of the dose excreted
in urine was offset by a decrease in the percentage of the dose excreted in feces, partially by a de-
crease in the percentage excreted as metabolite IX.
   Metabolite I is believed to be formed by /3-oxidation of metabolite V,(92) and the urinary excretion
patterns observed  for these metabolites were consistent with this^pathway. Therefore, repeated ad-
ministration of DEHP at 6000 and 12,000 ppm in the diet to rats apparently results in a nonlinear in-
crease in metabolism by /3-oxidatiori.
   It thus appears  that there is a nonlinear  relationship between administered dose and the active
dose of DEHP metabolites. By "active,"  we refer to that which produces peroxisome  proliferation,
an indirect measure of which is probably production of metabolite I by /3-oxidation of metabolite V.
                                             126

-------
                              RISK ASSESSMENT OF DEHP

It is possible, however, that some of the enhanced conversion of metabolite V to metabolite I is at-
tributable to mitochrondrial oxidation as well as to peroxisome proliferation.
  The 6000 ppm and 12,000 ppm exposure levels used in this study were identical to those used in the
NTP bioassay, which lead to the appearance of an excess of liver tumors. This study reveals that a
marked shift in the production of metabolite  I (and in the pattern of other metabolites as well) oc-
curs between the 1000 ppm level and the higher levels. This observation strongly  suggests that the
dose-response relation for peroxisome induction (and subsequent tumor induction) may decline in a
nonlinear fashion below the region of observed tumorigenic  responses.
            DOSE-RESPONSE EXTRAPOLATION OF RODENT DATA

  In order to estimate the risk to humans of exposure to DEHP in the environment, it is necessary to
estimate the relationship between risk and exposure levels in the low range to which humans may be
exposed. As a first step, we need to extrapolate the observed relationship between DEHP dose and
tumor incidence in rodents to low dose levels. Several mathematical models have been developed and
proposed for extrapolating from observed data on tumor incidence at high experimental doses to
risk at low doses.t85-95-961 Which of these models is most appropriate to use in any particular case is
an  important and  controversial aspect of risk assessment methodology. Some of the controversy
arises because in some cases the various models predict risks at low exposure levels that differ by sev-
eral orders of magnitude,' even though they may all fit the experimental data in the high exposure
range almost equally well. For example, in the preamble to OSHAs Cancer Policy (45FR 5200), esti-
mates of lifetime risks to humans from vinyl chloride and saccharin at likely exposure levels are pre-
sented. These estimates vary 1 million-fold depending on which extrapolation model is used.
   In selecting a model for carcinogenic risk assessment, the functioning of the model should be con-
sistent with the scientific information on the phenomenon being modeled. In the present case, it is
desirable that at least some of the models used should be consistent with our limited understanding
of  the possible mechanism(s) of carcinogenicity of DEHP.
   In conducting dose-response extrapolation, the dose in question is generally the applied dose of
the chemical, usually expressed in terms of mg/kg body weight per day. However, probably of more
importance in determining how the risk will change with dose is the dose of the ultimate carcinogen
at the target site. This latter dose will be a function of the pharmacokinetics of the chemical (its ab-
sorption, distribution,  biotransformation, and elimination). As long as all of the pharmacokinetic
parameters of a chemical are linear with dose, any increase in applied dose will cause a proportional
increase in what we may call the "target-site dose," but at high doses processes such as absorption,
biotransformation, and excretion may become saturated,  and there will no longer be a linear rela-
tionship between applied dose and target-site dose.'97-98' As an example of the application of this
phenomenon, Gehring et al.(99) examined data on the carcinogenicity and pharmacokinetics of vinyl
chloride. They demonstrated that metabolism of vinyl chloride was not linearly related to applied
dose but followed Michaelis-Menten kinetics.     .
   The situation with DEHP is different from that with vinyl chloride, which is a genotoxic carcino-
gen whose biotransformation to an active carcinogenic form is apparently saturated at high dose lev-
els. In the case of DEHP, our proposed mechanism involves peroxisome proliferation and, perhaps,
subsequent excess production of peroxide and other active oxygen species at high dose levels. If this
model is correct, the target-site dose in which we are  interested is that of the various active oxygen
species. Unfortunately, no data are currently available to  measure this target-site dose  directly.
 Instead, an indirect  measure is proposed, recognizing the many uncertainties in the use of  such a
 measure.
   This indirect measure is the level of urinary excretion of DEHP metabolite I. This is assumed to
 provide a surrogate  for the true target-site dose, which is a function of the effects of DEHP on the
 peroxisomes,' perhaps involving a metabolic disturbance or the generation of active oxygen species.
There is circumstantial evidence that production of metabolite I may serve as an indicator of peroxi-
 somal activity and perhaps of active oxygen generation. First, metabolite I is probably produced as a
                                                                f
                                             127

-------
                               TURNBULL AND RODRICKS

 result of /3-oxidation; (92> second,  production of metabolite I is increased at high dose levels of
 DEHP;'57'  and  third,  peroxisomal /3-oxidation activity increases substantially  at  similar dose
 levels.'78"82'
   There is uncertainty in using metabolite I as a surrogate for target-site dose. We do not know
 whether metabolite I is formed in the peroxisomes or in the mitochondria (or other cellular compart-
 ment). Obviously, production of  metabolite I in the mitochondria would not be an indicator of
 peroxisomal activity and would not involve peroxide generation. Thus,  the use of metabolite I
 production would overestimate peroxisomal activity and peroxide generation to the extent that mito-
 chondrial /3-oxidation is  involved in metabolite I production. However, the involvement of the per-
 oxisomes in the production of metabolite I is supported by the results of recent studies by Lhuguenot
 et al.,<10°-102) who examined the effects of dose and time of the metabolism of MEHP using Wistar
 rats in vivo and rat hepatocytes in vitro. They found that both in vivo and in vitro production of me-
. tabolite I increased in parallel with increases in cyanide-insensitive peroxisomal /3-oxidation activity.
 Also the change in rate of production of metabolite I from metabolite V with time of the hepatocytes
 in culture mirrored'the change in the activity of cyanide-insensitive palmitoyl Co A oxidation in the
 hepatocytes.(102)
   That mitochondrial /3-oxidation may also be involved in metabolite I production is suggested by
 reports of increases in the activity of enzymes involved in ,3-oxidation in both peroxisomes and mito-
 chondria after treatment of rats with DEHP.'79-80' However, provided the  mitochondrial contribu-
 tion to metabolite I production does not increase with dose of DEHP to a greater extent than the
 peroxisomal contribution, the use of level of metabolite I production as a surrogate for peroxisomal
 activity and, peroxide level will at least not underestimate the true values.
   In the past, pharmacokinetic data have been used to estimate target-site doses of genotoxic carcin-
 ogens, and these dose estimates have been used in place of measures of applied dose in risk extrapo-
lation.'99-103' In assessing the risks  of DEHP, we may use our knowledge of the nonlinear relation-
 ship between applied dose of DEHP and  urinary excretion of metabolite I to provide a surrogate to
 estimate the presumed nonlinear relationship between applied dose of DEHP and peroxisome activ-
 ity, which is in turn an indicator of the target-site dose of peroxide or other active oxygen species. To
 do this, it is necessary to  examine the quantitative relationship between daily dose of DEHP and ex-
 cretion of metabolite "I under conditions approximating those of the NTP bioassay. Data for this are
 derived from the Little study  of metabolism of DEHP in male rats fed DEHP at 1000, 6000, and
 12,000 ppm in  the diet for 20 days  before receiving '4C-labeled DEHP at the same dietary  levels.'571
 These data are presented in Table  4.
   The relationship between daily dose of DEHP and daily excretion  of metabolite I was fit to a
 power curve with equation:

     I = 0.0187 D1-4287

 where I = daily excretion of metabolite I, and D = daily dose of DEHP. This curve was chosen as
            TABLE 4. RELATIONSHIP BETWEEN DAILY DOSE OF DEHP AND EXCRETION OF
                      METABOLITE I IN MALE RATS FED DEHP FOR 20 DAYS
        Dietary
        Concentration
        of DEHP (ppm)
   Equivalent
   Daily Dose
(mg/kg per day)*
Fraction of Dose
Excreted in Urine
 as Metabolite I
   Amount of
•"'  Metabolite I
Excreted per Day
(mg/kg per day)b
1,000
6,000
12,000
64.97
370.81
769.64
0.11
0.25
0.31
7.15
92.76
238.59
          "Based on consumption of food containing '4C-labeled DEHP and terminal body weight.15"
         bAmount excreted = daily dose of DEHP x fraction excretion as metabolite I.
                                            128

-------
                               RISK ASSESSMENT OF DEHP

best fitting the data (r = 0.9996). This equation is used to adjust the doses in the NTP bioassay to
give doses that are surrogates for the target-site dose before fitting extrapolation models to the bio-
assay dose-response data. Because of the uncertainties outlined above, this adjustment must be con-
sidered only semiquantitative. However, the uncertainties are such that, the adjustment is likely to
overestimate risk at low dose levels and, hence, be conservative.


Choice of extrapolation methods

  A review of the various extrapolation models that have been used for risk assessment reveal that
no single model is clearly the most appropriate for use with DEHP. We have, therefore, used several
procedures for low dose extrapolation:

  1. Multistage model using applied doses of DEHP
  2. Multistage model using a surrogate (metabolite I level) for target-site doses ,.,
  3. Mantel-Bryan probit model using applied doses of DEHP
  4. Mantel Bryan model using a surrogate for target-site doses
  5. Threshold model

  The multistage model is used because it is the model most widely used by regulatory agencies and
has a substantial basis in theories of cancer causation."04"107' This model has a biological basis in
that it is  derived from a widely accepted view of the carcinogenic process.  Under this view, cancer
arises in a single cell and is expressed after the cell or its progeny have passed through k transitions,
the rates  of one or more of which are proportional to the concentration of the carcinogen at the tar-
get site."081 Some carcinogens may act at the first stage of the process (so-called initiators or early-
stage carcinogens), others may act only at later stages (promoters or late  stage carcinogens), and still
others may act at both early and late stages of the process of tumor development (so-called complete
carcinogens). In the general case, where mechanisms are unknown,  it  has become the practice to
adopt the linearized multistage model.
  The Mantel-Bryan probit model is another commonly used model.
-------
                              TURNBULL AND RODRICKS

  Although we have not used them for this study, other extrapolation models, such as the Weibull
or multihit models, may also be appropriate for low dose extrapolation. <106-107)


Modeling of rodent carcinogenicity data

  In estimating low dose risks to rodents, data derived from the NTP bioassay were used. To pro-
vide a conservative estimate of risk, the combined incidences of hepatocellular carcinoma and ade-
noma/neoplastic nodules were used. Also, to provide a better estimate of the background tumor in-
cidence,, pooled controls were used, combining data from the bioassays that were performed in the
same room of the same laboratory at the time as the DEHP bioassay. Thus, control data from bioas-
says of DEHP, butylbenzylphthalate, guar gum, and di(2-ethylhexyl) adipate were used. The dose
levels in mg/kg per day estimated by NTP<2) on the basis of food consumption were used. The 4 data
sets used for risk extrapolation are presented in Table 5. Included in Table 5 are the estimates of the
surrogate target-site doses based on the expected urinary levels of metabolite I for rats. These are de-
rived as described above using data from the multidose metabolism study.(S7) Since this metabolism
study involved only male rats, the adjustment, strictly, is applicable to male rats. However, we have
also conservatively applied the adjustment to the female rat data, since these  data predict slightly
higher risks.  Corresponding  data are not available  for mice. Therefore, no such adjustments are
made for the doses administered to mice; although a similar nonlinear relationship between applied
dose of DEHP and "target-site dose" probably also occurs in mice.
  The computer program Global 82 written by Howe and Crump of Science Research Systems, Inc.,
Ruston, Louisiana, was used  to perform low dose extrapolation using the multistage model. A com-
puter program was also used to perform low dose extrapolation by the Mantel-Bryan probit proce-
dure. As discussed earlier in the description of low dose extrapolation models, the Mantel-Bryan
probit procedure was not intended by its authors to provide valid estimates of low dose risk. Hence,
the program used generates estimates of "virtually safe dose" (VSD). The VSDs we report here corre-
spond to lifetime risks of less than lO'6 (1 in 1 million) and lO'8 (1 in 100 million).
  The results of applying the multistage model to all four data sets in Table 5 are presented in Table
6, which presents the dose coefficients ^ and 
-------
                               RISK ASSESSMENT OF DEHP


  TABLE 6.= MULTISTAGE MODEL DOSE COEFFICIENTS BASED ON DEHP DOSES IN NTP
                   qt (mg/kg per day)'1
q2 (mg/kg per day)-
qt* (mg/kg per day)'
Male rats
Female rats
Male mice
Female mice
1.42 x 10-4
2.13 x 10-4
4.59 x 10-"
2.35 x ID"1
2.86 x 10-'
2.07 x lO'7
0 ,
0
4.88 x 10-"
4.96 x 10-4
6.70 X 10-4
3.29 X 10-"
  "The figures tabulated are the dose coefficients ql and q2 in the multistage model: P(d) =  1 - exp[-(?o +
qtd + qid1)] where P(d) is the probability of developing a tumor after lifetime exposure to a dose of d mg/kg •
per day. For small values of d, the excess risk of developing a tumor closely approximates (q,d + qd1)- Also
tabulated is ^i*, the upper 95th percentile confidence limit on 
-------
                              TURNBULL AND RODRICKS

  Table 8 lists the values of "virtually safe dose" corresponding to a risk of 10"6 and 10"8 predicted by
the Mantel-Bryan probit procedure when applied to the data on each species for the sex that showed
the higher risk (female rats and male mice).
    TABLE 8. "VIRTUALLY SAFE DOSES" OF DEHP'CORRESPONDING TO RISKS OF 10'6 AND 10~8
             FROM THE MANTEL-BRYAN MODEL FOR FEMALE RATS AND MALE MICE
                      BASED ON DEHP DOSE AND ON SURROGATE DOSE

                                    Virtually Safe DEHP Dose fug/kg per day)
                          Based on DEHP Dose
                        Based on Surrogate Dosea
                      Risk =  ID'6
Risk =  IO-*
Risk =
Risk =  10-*
Female rats
Male mice
46.8
. H-9
6.5
1.6
791
198
  aSee Footnote b to Table 7.
           INTERSPECIES EXTRAPOLATION: RODENTS TO HUMANS

  The risks per unit of exposure (or NOELs) derived by application of the various models listed in
the previous section pertain strictly to mice and rats. The next stage of analysis involves extrapola-
tion of these risks to humans. Extrapolation between species adds considerably to the uncertainty of
risk assessment, as has been discussed by many individuals from such groups as the National Acad-
emy of Sciences/3-11" the Interagency Regulatory Liaison Group,(85) and the Scientific Committee
of the Food Safety Council.(9S) In the preamble to its Cancer Policy (45 FR 5200), OSHA discussed
many of the complications of interspecies risk extrapolation and concluded:

           Extrapolation from  animal data  to predict risks in  humans  introduces
           many additional uncertainties. These include selection of appropriate scal-
           ing factors for size, lifespan, and  metabolic rate; differences in routes of
           exposure, duration and schedule of exposure, absorption, "metabolism, and
           pharmacokinetics; differences in intrinsic susceptibility and repair capabili-
           ties; intra-population variation and susceptibility; and exposure to other
           carcinogens and intrinsic and extrinsic  modifying factors.  At least theo-
           retically, these factors can affect the relative response of humans and ani-
           mals by many orders of magnitude.   .                   '

There follows a discussion of  some of these factors, as they apply to interspecies  extrapolation for
DEHP, and the major  approaches that have been proposed to take them into consideration.
 Unit of dosage measurement

  Consideration of an appropriate dosage unit encompasses consideration of animal size, lifespan,
 and, to some extent, metabolic rate. In specifying a unit of dose there are generally three factors
 involved: a measure of the amount of the substance administered (mg, ml, mmole, and so on), an
 indication of the size of the organism (kg body weight, m2 body surface area, blood volume, and
 so on), and some indicator of time (day, lifetime, and so on). Among the most commonly used units
 of dose are mg/kg body weight per day, mg/kg body weight per lifetime, and mg/m2 body surface

                                           132

-------
                              RISK ASSESSMENT OF DEHP

area per day. Debates over the choice of dosage unit have centered on the appropriate measure for
body size (kg body weight or m2 body surface area) and on the temporal descriptor (per day or per
lifetime).'9'-105-"2-"4'
  Hoel et al.("2> proposed the use of dosage units in mg/m2 body surface area per day on the basis
of studies of the acute toxicity of anticancer drugs in humans and animals.<115) In these studies the
acutely toxic level was similar in mouse, rat, hamster, dog,  monkey, and man when dosage was ex-
pressed as mg/m2 per day. This procedure has also been adopted by the Environmental Protection
Agency(IOS> and is the procedure used by the Consumer Products Safety Commission"071 in its risk
assessment of DEHP. However, the relationship between dose and body surface area, determined
on the basis of acute drug toxicity, a priori means very little when considering chronic effects, such
as cancer. Given the uncertainty about the factors contributing to carcinogenicity, the most appro-
priate basis for judging the suitability of a species-to-species conversion factor is empirical  data on
the relative susceptibilities  of different species.
  What few relevant data exist have been analyzed by Crump et al.,'105) who concluded that the dos-
age measurement giving the closest correlation between species was mg/kg body weight per day. In
the absence of good evidence for the use of a more complex procedure, we propose the use of mg/kg
per day as a generally acceptable basis for interspecies dosage comparison. In addition to its relative
simplicity, this procedure appears to have the best empirical support.(10S)
  All of the foregoing discussion and almost all of the discussion in the scientific literature concern
the relative merits of the possible measures of body size and temporal factors. Little attention has
been paid to the appropriateness of the measurement of the amount of substance applied (usually
measured in milligrams). In cases where the administered dose is linearly related to the dose of the
active carcinogen at the target site, a direct measure "of the amount applied is appropriate. However,
as alluded to earlier, in cases where nonlinear pharmacokinetics apply, measurement of the dosage in
terms of the material applied may  be inappropriate, and,  where available,  pharmacokinetic data
should be used to modify the dosage measurement so that the units of dosage are an indication of the
level of the ultimate carcinogen at the target  site. Alternatively, one might actually measure the
amount of an active metabolite of a precarcinogen or measure DNA interactions to estimate the tar-
get site dose. This would not be necessary if the pharmacokinetic parameters were the same in both
the experimental species and in humans. However, in many cases, the rates of production and inacti-
vation of active metabolites are likely to have species differences. For example, the rate of produc-
tion of active metabolites of chloroform, perchloroethylene, and 2-acetylaminofluorene (AAF) dif-
fers in different species and is correlated with the relative.carcinogenic potency of the particular
chemical in  the different species.(98-116-"7'


Route,  duration, and schedule of exposure

  In the NTP Bioassay of DEHP, animals were fed DEHP at a uniform concentration in  the diet
continuously from the age of about 5-6 weeks (i.e., after about  1/20 of their lifespan had elapsed)
for the subsequent 103 weeks. In contrast, although some dietary exposure may occur, human expo-
sure to DEHP is likely  also to involve inhalation of low levels of DEHP vapor released from PVC
products, dermal contact with products containing DEHP, some  of which may be absorbed through
the skin, and, in some cases, parenteral  exposure from DEHP  in medical devices, such as blood
bags. It is unclear if DEHP absorbed from these different routes is  equivalent in carcinogenicity.
Certainly, although DEHP will be absorbed from blood bags intact, the action of nonspecific lipases
in the gut results in absorption of the hydrolysis products 2-EH and MEHP rather than DEHP itself.
For the purpose of this  risk assessment, however,.we will assume that DEHP absorbed by all routes
is equivalent.
  In addition to differences in routes of exposure, human  exposure to DEHP differs from that of
the rodents in the NTP  bioassay in its temporal pattern; humans are not exposed to DEHP at a con-
stant dietary concentration for their lifetime starting shortly after weaning. The actual temporal pat-
tern has not been evaluated in detail, but it is likely to be nonuniform. Infants  may be exposed to
                                            133

-------
                              TURNBULLANDRODRICKS       .

higher than average amounts of DEHP as a result of chewing and sucking PVC teethers, pacifiers,
and toys and from skin contact with vinyl playpen liners, baby pants, and so on."07' Individuals re-
ceiving multiple transfusions of blood or blood products (e.g., hemophiliacs) are also likely to be ex-
posed to higher than average levels of DEHP, since the plasticizer is known to leach from the PVC
blood bags into the blood during storage. (118-119)
   It is common practice in conducting carcinogenic risk assessments to equate risks in humans  and
animals receiving the same average lifetime daily dose (mg/kg per day). However, if the multistage
model of carcinogenesis is correct, Day and Brown'I08) have demonstrated that for anyparticular
temporal pattern of exposure, the lifetime risk of cancer will depend on whether the carcinogen is an
early  stage carcinogen (an initiator) or a late stage carcinogen (a promoter).
   Unfortunately, we do not know at what stage or stages DEHP acts to  produce carcinogenic ef-
fects. There is some indirect evidence that it may be a promoter. Reddy and Rao'120' have shown that
the hypolipidemic drugs Wy-16,634 and clofibrate promote the appearance of hepatocellular carci-
noma in the liver of rats given an initiating dose of the'liver carcinogen diethylnitrosamine (DEN).
,As discussed  earlier, these drugs produce a similar spectrum of effects in the liver of rodents as does
DEHP. Hence, DEHP may also be a promoter. Additional support for this hypothesis comes from
the work of Ward et al.,(121> who found an increase in preneoplastic foci and adenomas in the liver in
mice given an initiating dose of DEN followed by treatment with DEHP compared to those receiving
DEN without DEHP. However, this effect has not been confirmed by others.(122) As discussed ear-
lier, Hirota and Yokoyama(88> and Ito et al.(89) have reported data suggesting that hydrogen perox-
ide is  a tumor promoter in the gut. However, Levin et al.(90) have reported that hydrogen peroxide is
mutagenic in a specially constructed tester strain of Salmonella, and if the mechanism of action pro-
posed earlier involving peroxisome proliferation is correct, the DNA damage that would be caused
by the active  oxygen species generated by the peroxisomes might initiate tumor development.(99)  It is
possible that DEHP has both initiating and promoting activity.
   Since we do not know whether DEHP acts as an initiator, a promoter, or both, we will use the
simple assumption that risk is a function of lifetime average daily dose.


Interspecies differences in target-site susceptibility

   The liver in rodents appears to be particularly susceptible to the induction of tumors, as evidenced
by the high and variable incidence of spontaneous liver tumors in various strains of mice,'15-16' the
high spontaneous incidence in the livers of rats of preneoplastic cells that can be stimulated by pro-
moting agents to produce tumors, <17"19) and the high proportion of animal carcinogens whose site of
action is the  rodent liver.'123' By contrast, few chemicals are known to cause liver tumors in  hu-
mans,'20' and humans seem to be less susceptible to peroxisome proliferation than are  rats.'72'
Hence, if the peroxisome proliferation hypothesis of carcinogenesis is correct, humans would be ex-
pected to be less susceptible than rodents. Unfortunately, no data are available that would allow  this
difference in  sensitivity to be taken into consideration in a quantitative way. We are limited to noting
qualitatively that this is another reason why use of the available animal data is likely to result in over-
estimation of the risk to humans of DEHP.
 Interspecies differences in metabolism

   Albro and co-workers have conducted and reported on numerous studies of the metabolism of
 DEHP in various species of rodents, the Green monkey, and man.  Data available from studies in
 various species are summarized in Table 9,<47) which reveals striking differences in the pattern of uri-
 nary metabolites between man and rat and the similarity between man and monkey. Another differ-
 ence between the rat and man is, that,the major fraction of metabolites formed in man and the green
 monkey is excreted in conjugated form, whereas little such conjugation takes place in the rat.'47'
   Albro et al.'47' have suggested that rats compensate for not making glucuronides by carrying oxi-
                                            134

-------
                              RISK ASSESSMENT OF DEHP
        TABLE 9. DISTRIBUTION OF URINARY METABOLITES OF DEHP IN VARIOUS SPECIES

                                         Percentage of Total Metabolites
Metabolite*
Residual DEHP
MEPH
I
II
III
IV
V
VI
VII
VIII
IX
X
A, B, C
Phthalic acid
Rat

Trace
17.2
2.0
1.2
3.3
51.3
2.6
2.6

13.3
0.6
4.1
1.8
Mouse
0.5
18.6
16.8
1.0
0.4
0.8
1.1
14.9
7.2s

. 12.3
2.2
8.1
12.4
Guinea
Pig

71.2
2.4'
0.4
0.5
0.4
6.9
1.1
0.8

3.4
1.3
6.2
5.4
Green
Monkey
2.2
28.9-
0.1



4.2
5.9
7.0
5.7
38.2 '
0.1
7.6
0.1
Man

18.3

1.8

1.2
5.3 <*,
12.1
11.9
8.1 .
36.2
0.1
4.9
0.1
Hamster
0.3
4.5
13.0
0.1
0.3
0.4
14.0
10.2
4.9

32.7
1.9
6.1
9.5
  From Albro et al.14"
  aSee Figure 5.
dative metabolism all the way to the highly water-soluble diacids. These authors also noted that if
formation of hydroxyl side chains involves by analogy with fatty acid u-oxidation, a mixed function
oxidase reaction, one would expect a net conversion of NAD(P)H to NAD(P). The additional steps
from alcohol to aldehyde (or ketone) and from aldehyde to acid, as well as the apparent a- and j3-ox-
idations needed to produce metabolites I, II, and III, would all be associated with net conversion of
NAD(P) to NAD(P)H. Thus, the overall demand on the oxidation potential of the  liver when high
doses of DEHP are given would be in opposite directions for rat and primate. Albro et al.(47) con-
cluded that to the extent that metabolism of DEHP is involved in its biological activity, one must
question seriously whether rats are an appropriate  model for man.
  Although the metabolic differences between man and rats are striking, the differences are not so
marked when  one compares mouse and  man (Table 9). Because the NTP bioassay revealed that
DEHP is carcinogenic in mice, it may appear that metabolic differences are not important for carci-
nogenesis. However, Table 9 shows that rats and mice produce similar percentages of metabolite I,
and both differ from humans.  If metabolite I is an indicator of peroxisome proliferation and en-
hanced /3-oxidation as has been hypothesized, then in keeping with the hypothesis  relating peroxi-
some proliferation to hepatic tumors, the major differences between rodents and primates in metab-
olite I production imply equally major differences  in susceptibility to cancer from  DEHP.
  As with all the  other data we have described, there are gaps and uncertainties in  the metabolism
data. Most important is the fact that the data shown in Table 9 were obtained under different experi-
mental conditions for the different animal species and are therefore not strictly comparable. To ad-
dress this uncertainty, a study has been conducted to compare the. metabolic fate of DEHP when
administered under identical conditions to monkeys, rats, and mice.(94) In this comparative metabo-
lism study, a single dose (100 mg/kg) of l4C-labeled DEHP in corn oil was administered by gastric in-
tubation to 3 male cynomolgous monkeys, 5 male F-344 rats, and 25 male B6C3F1 mice. All 3 spe-
cies excreted an average of 30-40% of the dose in the urine, all but 5% or less in the first 12 hours
after dosing in the mouse and rat and in the first 24 hours in the monkey. All'species excreted about
50% of the dose in the feces, all but 3% or  less during the first 24 hours in the mouse and rat and
during the first 48 hours in the monkey.
  The metabolites excreted in the urine were identified as shown in Table 10. In general, the pattern
                                           135

-------
                               TURNBULL AND RODRICKS

                     TABLE 10.  URINARY METABOLITES OF DEHP EXPRESSED
                       AS PERCENTAGES OF TOTAL URINARY RADIOACTIVITY
                        EXCRETED IN THE FIRST 24 HOURS AFTER DOSING
                     Metabolite*
Monkey
Rat
Mouse b
MEHP
Phthalic acid
I
II
III
IV
V
VI
VII
IX
X
XII
XIII
Uncertain
XIV
Uncertain
11
2
b.s

0.5
0.5
25
1
7d
18
9
2
6
15
2
1
_LC
2
11
0.9

4
,29
. 11
3
18
4
6
6
1
1
1
17
13
13
0.8
0.8
3
1
12
6
11
2
5
7
5
2
1
                       From Moran et ai.|94)
                       aMetabolites are numbered according to the convention of
                     Albro.
                       bThe mouse urine extract analyzed by HPLC contained only
                     79% of the radioactivity excreted in 0-24 hours. The remainder
                     of the radioactivity was eluted from the SAC-2 resin in the
                     acidic aqueous wash and probably contained some of the more
                     polar metabolites, perhaps including glucuronides.
                       c Radioactivity in the sample was less than twice background
                     for the system.
                       dThis fraction may include metabolite VIII, which was iden-
                     tified in monkey urine after IV administration of DEHP.1"'
of urinary metabolites identified in this study is similar to that described by Albro et al.(47) in rats,
mice, and monkeys. In particular, metabolite I constitutes a high proportion of the total urinary me-
tabolites in the mouse (13%) and the rat (11%) but only a very low proportion of the total urinary
metabolites in the monkey (0.5%). If the hypothesis described earlier regarding the mechanism of
carcinogenicity of DEHP is correct, and production of metabolite I is an indicator of target-site
dose, the carcinogenicity of DEHP in monkeys (and presumably in humans, who also excrete very
small amounts of metabolite I) is likely to be much lower than in rats or mice at the same dose level.
Since it is not known what proportion, if any,  of the metabolite I produced in monkeys is formed by
^-oxidation in the peroxisomes, and hence would produce peroxide, it is not possible to quantify the
likely difference in susceptibility to DEHP-induced carcinogenicity between monkeys and rodents.
However, if comparative urinary excretion of metabolite I can be used as an indicator, monkeys
would be 20- to 25-fold less  susceptible than rodents (Table 10).
  If the metabolic and biological data are keys to carcinogenic activity, they strongly suggest that
risk predicted under all of our various models simply does not apply to man or, more likely, that the
magnitude of human risk is, at a given (low) level of exposure, very much less than that predicted for
rodents. Although direct evidence supporting  this assumption is lacking, it is supported by the cur-
rently most reasonable hypothesis about the tumorigenic action of perbxisome proliferators in ro-
dents. The actual human risk per unit of exposure cannot be quantified  but is likely to be less than
that  predicted for rodents and may even be zero.

                                            136

-------
                              RISK ASSESSMENT OF DEHP

  In performing rodent to human extrapolation for this risk assessment we have used 2 procedures,
both of which probably overestimate the risk to humans. Both assume that humans are at equal risk
as rodents at the same dose level in mg/kg per day.(10S) The first, and most conservative, procedure
assumes that the relevant dose level is the applied dose of DEHP and, thus, takes into account none
of the information available on the likely mechanism of action of DEHP in rodents and the data sug-
gesting that the effects seen in rodents are not likely to occur in monkeys or humans.
  The second procedure uses the dose adjustment discussed in the previous section. This adjustment
uses the relationship between applied dose of DEHP and urinary excretion of metabolite I in rats as
a surrogate for the likely target-site dose of active oxygen species generated by the peroxisomes. This
second procedure assumes the same relationship in rats and in humans between the applied dose of
DEHP and the dose of metabolite I, the latter being a surrogate for the dose of the ultimate carcin-
ogen. The dose of metabolite I at low doses of DEHP is calculated using the relationship determined
empirically in rats from data generated by Robinson et al.(57):

    / =  O.OnSD1-4287

where / = dose of metabolite I, and D = daily dose of DEHP.

  Under these assumptions, the data presented in Tables 6 and 7 may be applied directly to humans
to estimate the risk at low dose  levels predicted by application of the multistage model without
(Table 6) or with (Table 7) the surrogate target-site dose adjustment described above. Similarly,
Table 8 shows the virtually safe doses predicted by the Mantel-Bryan model without .and with  the
same adjustment.
NOEL/Safety factor approach to risk assessment

  If the mechanism of carcinogenicity of DEHP that has been proposed above is correct, no in-
creased risk of cancer would occur at exposure levels that do not cause peroxisome proliferation and
subsequent excess peroxide production. Such pathological effects are of the type normally protected
against by the classic toxicological procedure of identification of a no-observed-effect level (NOEL)
and application of a safety factor to determine an acceptable daily intake (ADI). To use this proce-
dure, it is of course necessary to identify a NOEL. Unfortunately, it is not clear if a NOEL has been
identified. In the Phase I validation study of the CMA Voluntary TSCA testing program,(14) groups
of 12 male and 12 female F-344 rats were fed diets containing DEHP at 0, 1000, 6000,  and 12,000
ppm for 3 weeks. The activity of the enzyme carnitine acetyltransferase, which occurs in the peroxi-
somes and the mitochondria, showed a dose-related increase in activity at all dose levels after as little
as 1 week of treatment. Dose-related effects were also noted on liver weight (increased 20, 66, and
98% in male rats at 3 weeks),  serum triglycerides (decreased to 56, 31, and 18% of control in males
at 3 weeks), and a cytochemical test (dose-related increase in peroxisome proliferation).  Males were
affected more than females. These effects were not evident, however, in the animals allowed 2 weeks
of recovery after  DEHP treatment, indicating that the effects are reversible.
  The European Chemical Industry Trade Association (CEFIC) sponsored a similar study in which
groups of male and female Alderly Park SPF-derived albino rats  (number unspecified)  were  fed
DEHP at 0, 50, 200, and 1000 mg/kg per day (about 1000 ppm, 4000 ppm, and 20,000 ppm) for 28
days.'"4' Liver weight was increased in all treated  groups in a dose-related manner. There was a
dose-related proliferation of peroxisomes starting at the lowest dose level and a similar proliferation
of smooth  endoplasmic reticulum.
  Morton'801 fed DEHP at various dose levels to groups of 5-12 male Sprague-Dawley rats for 7
days and measured such parameters as liver weight, serum triglyceride level, liver catalase, carnitine
acetyltransferase  (CAT), carnitine palmitoyltransferase (CPT), and  /3-oxidation  activity. Liver
weight was significantly increased in a dose-related manner at DEHP dietary levels of  iOOO, 2500,
and 5000 ppm but not at 50, loo,  or 500 ppm. Serum triglyceride levels were significantly reduced in
                                            137

-------
                              TURNBULL AND RODRICKS

                                                                                f
a dose-related manner at all levels tested (50, 500, and 2500 ppm). Catalase activity was increased sig-
nificantly at 5000 ppm but not at  100 or 1000 ppm. CAT and CPT activities were significantly in-
creased in a dose-related manner at 100, 500,  1000, and 2500 ppm but not at 50 ppm.
  Of perhaps most importance to the present discussion is the liver /3-oxidation activity. When total
liver /3-oxidation activity was measured, significant dose-related increases were seen at 500, 1000,
and 5000 ppm, and slight but not significant increases were seen at 50 and 100 ppm. Morton*80' also
examined /3-oxidation activity in isolated mitochrondria and peroxisomes after feeding DEHP, at 0,
100,  1000, and 5000 ppm. In peroxisomes, /3-oxidation activity  was increased significantly only at
5000 ppm, but slight, nonsignificant increases were seen at 100 and  1000 ppm. In isolated mitochon-
dria, ^-oxidation activity was increased significantly at 5000 and 1000 ppm and slightly but not sig-
nificantly at 100 ppm.
  In a study recently conducted  by the British Industrial Biological Research Association,<125)
DEHP was fed to groups of 5 Fischer 344 rats of each sex at dietary levels of 0, 100, 1000, 6000,
12,000, and 25,000 ppm  for 21 days. Males and females fed 6000 ppm or more showed significantly
increased liver weights and significantly increased peroxisomal (8-oxidation activity.  Serum triglyc-
eride levels were significantly reduced at the same levels in males. Microsomal lauric acid 11- and 12-
hydroxylase activity was  significantly increased in males at 1000 ppm and above, as was the number
of peroxisomes in the liver. These  latter effects were seen in females only at 6000 ppm or more. No
significant changes in any of the parameters monitored occurred at 100 ppm, with the exception of
an increase in serum  triglyceride level in males.
  In attempting to identify a NOEL from these data, several choices are possible. Based on the dose
level causing a significant increase in peroxisomai jS-oxidation activity in the studies by Morton<80>
and BIBRA,11"' a NOEL of 1000 ppm could be identified (about 70 mg/kg per day based on Mor-
ton's food consumption and  body  weight data  or 106 mg/kg per day based on BIBRA data). Based
on total liver /3-oxidation activity, the NOEL would be set lower,  at 100 ppm (about 7 mg/kg per
day). A NOEL for all effects of 100 ppm (about 11 mg/kg per day)  was identified in the BIBRA
study.<12S) However, a NOEL for all effects cannot be identified from the Morton study,<80) since a
significant reduction in serum triglyceride level was seen even at the lowest dose of 50 ppm (about 3.5
mg/kg per day), though  such an effect was not seen in the BIBRA study at levels below 6000 ppm.
Since these possible NOELs are derived from studies of short-term exposure (7-21 days), estimation
of a chronic ADI for humans would typically involve application of a safety factor of 1000.(" " This
would lead to a chronic ADI of between 70 to less than 3.5 ^g/kg per day, with the most likely value
being 11  /ig/kg per day, which is derived from the NOEL for peroxisomal proliferation in the
BIBRA study.11"'
                                      DISCUSSION

  To provide a comparison of the implications for risk at low dose levels of these 5 models (multi-
stage and Mantel-Bryan models, both with and without target-site dose adjustment, plus threshold
model) their predictions of virtually safe dose (risk less than 10'6 on lifetime exposure) or ADI are
listed in Table 11. For the multistage* model, both maximum likelihood estimates and 95th percentile
upper confidence estimates are listed. For each model, results for the data set (species and sex) pre-
dicting the highest risk are presented.
  Under the most conservative procedure (upper confidence limit on multistage model with applied
dose levels), exposure would need to be less than 1.5 ^g/kg per day to ensure a lifetime risk of less
than 1 in 1 million (10'6).  At the other extreme, the Mantel-Bryan model using the surrogate target-
site dose adjustment predicts a risk of less than lO'6 at a daily dose of 791 pg/kg per day. The model
that is most consistent with our general understanding of cancer development (the multistage model)
when combined with our  attempt to make the best use of data to provide inferences regarding the
likely shape of the dose-response curve at low doses (the surrogate target site dose adjustment) pre-
dicts a lifetime risk of 10"6 or less at DEHP4dose levels as  high as 116 ^g/kg per day.
  These estimates of "virtually safe dose" assume equal risk to humans and rodents at the same dose
                                           138

-------
                             RISK ASSESSMENT OF DEHP
               TABLE 11. ESTIMATES OF SAFE LEVELS OF EXPOSURE TO DEHPa

                                            Virtually Safe Dose (risk < lO'6) or
                                                  ADI f/jLg/kg per day)
      Extrapolation Model
     Maximum
Likelihood Estimate .
    Lower 95th
Percentile Estimate
      Multistage with applied
        doses
      Multistage with surrogate
        doses
      Mantel-Bryan with applied
        doses
      Mantel-Bryan with surrogate
        doses
      Threshold
       2.2

     116
         1.5

       86.3

       11.9

      791
      < 3.5-70
        "The safe dose levels listed represent the dose levels associated with a lifetime risk of 10'6 or
      less predicted by the multistage or Mantel-Bryan model or the ADI predicted by application of
      a safety factor to the NOEL as described in the text. In each case, the value for the species and
      sex predicting the lowest safe level (highest risk) is listed.
level of DEHP. However, as we have discussed above, there is information on interspecies differ-
ences in target organ susceptibility and in physiological and biochemical responses to DEHP that
suggest that humans are likely to be less susceptible to  the carcinogenic effects of DEHP than
rodents. That is, all of the estimates in Table 11 probably overestimate the risk to humans from
DEHP exposure.
  In conclusion, there is a substantial body of data to indicate that the simple application of a low
dose extrapolation model to the available data on the carcinogenicity of DEHP to.estimate human
risks likely overestimates these risks. Factors contributing to such overestimation are (1) the likely
mechanism of carcinogenicity of DEHP in rodents and the likely nonlinear relationship between the
administered dose of DEHP  and 'the dose of the proximate carcinogenic species, (2) differences in
target-site sensitivity between humans and rodents for liver tumors in general, and (3) differences in
the response of monkeys and probably humans to DEHP, which indicate that the  hypothesized
mechanism of carcinogenicity of DEHP in rodents does not occur or occurs to a lesser extent in hu-
mans than in rodents.
                                ACKNOWLEDGMENTS

  We thank the members of the Chemical Manufacturers Association Phthalate Esters Panel, par-
ticularly Dr. Elizabeth J. Moran, for advice and support in the conduct of this study. We also thank
the members of CMA's Toxicology Research Test Group, particularly Drs. Bernard Astill, Arthur
W. Lington, and Bernard F. Schneider for useful discussions on the technical content of this paper.
We are indebted to Ms. Karen McCrary, Mr. James Woldahl,  and Ms. Claudia Barber for typing
and Dr. Kenny S. Crump for providing computer runs of the multistage and Mantel-Bryan models.
                                     REFERENCES

  1. LITTLE, A.D. (1982). Phthalates'in Consumer Products. Contract No. CPSC-C-80-1001. Washington,
     D.C.: Consumer Product Safety Commission.     t
  2. NATIONAL TOXICOLOGY PROGRAM (NTP). (1982). NTP Technical Report on the Carcinogenesis
                                            139

-------
                                TURNBULL AND RODRICKS
  3.

  4.



  5.


  6.


  7.

  8.

  9.
 10.

 11.

 12.


 13.


 14.
 15.
 16.
 17.


 18.


 19.


20.


21.


22.

23.
 Bioassay of.Di(2-Ethylhexyl)Phthalate (CAS No. 117-81-7) in F344 Rats and B6C3F Mice (Feed Study).
 NIH Pub. No. 82-1773.
 NATIONAL ACADEMY OF SCIENCES (NAS). (1983). Risk Assessment in the Federal Government:
 Managing the Process. Washington, D.C.: National Academy Press.
 MILKOV, L.E., ALDYREVA, M.V., POPOVA, T.B., LOPUKHOVA, K.A., MAKARENKO, YU.L.,
 MALYAR, L.M., and SHAKHOVA, T.K. (1973). Health status of workers exposed to phthalate plasticiz-
 ers in the manufacture of artificial leather and films based on PVC resins. Environ. Health Perspect 3,
 175-178.
 GILIOLI, R., GULGHERONI, C., TERRANCE, T., FILLIPINI, G. MASSETO,  N., and BOERI, R.
 (1978). A neurological electromyographic and electroneurographic study in subjects working at the pro-
 duction of phthalate plasticizers: preliminary results. Med. Lav.  69, 631.
 THIESS,  A.M., KORTE, A., and FLEIG, I. (1978). Study of morbidity in BASF workers exposed to di-
 (2-ethylhexyl)phthlate (DOP). 18th Annual  Meeting of the Human Society of Industrial Medicine  DD
 137-151.                                           '                                       '
 CARPENTER, C.P., WEIL, C.S., and SMYTH, H.F.  (1953).  Chronic oral toxicity of di(2-ethyl-
 hexyl)phthlate for rats, guinea pigs and dogs. Arch. Ind. Hgy. Occup. Med. 8, 219-226.
 HARRIS, R.S., HODGE, H.C., MAYNARD, E.A., and BLANCHET, H.J., JR. (1956). Chronic oral
 toxicity of 2-ethylhexyl phthalate in rats and dogs. Arch. Ind. Health 13, 259-264.
 LEFAUX, R. (1968). Practical Toxicology of Plastics. Cleveland: CRC Press.
 W.R. GRACE AND CO. (1948). Dioctyl Phthalate. Technical Brochure. As cited in:  Krauskopf, L.G.
 (1973). Studies on the toxicity of phthalates via ingestion. Environ. Health Perspect. 3, 61-72.
 SONTAG, J.M., PAGE, N.P., and SAFFIOTTI, V. (1976). Guidelines for Carcinogen  Bioassay in Small
 Rodents. NCI Technical Report  Series No. 1. DHEW Publication No. (NIH) 76-801.
 NORTHUP, S., MARTIS, L., ULBRICHT, R., GARBER, J., MIRIPOL, J., and SCHMITZ, T. (1982).
 Comment on the carcinogenic potential of di(2-ethylhexyl)phthalate. J.  Toxicol. Environ. Health  10
 493-518.           '                                                                        '
 REDDY, J.K., MOODY, D.E., AZARNOFF, D.L., and RAO, M.S. (1976). Di-(2-ethylhexyl)phthalate:
 An industrial plasticizer induces hypolipidemia and enhances hepatic catalase  and carnitine acetyltrans-
 ferase activities in rats and mice. Life Sci. 18, 941-946.
 MIDWEST RESEARCH INSTITUTE (MRI). (1982). Toxicological effects of diethylhexyl phthalate. Fi-
 nal Report. MRI Project No. 7343-B. Phthalate esters program panel, Voluntary Test  Program, Health
 Effects Testing, Phase I: Validation results. Volume II. Washington, D.C.: Chemical Manufacturers As-
 sociation.
 BUTLER, W.H., and NEWBERNE, P.M. (eds.). (1975). Mouse Hepatic Neoplasia. Amsterdam: Elsevier.
 CLAYSON, D.B. (1981). International Commission for Protection against Environmental Mutagensand
 Carcinogens. ICPEMC Working Paper 2/3: Carcinogens and Carcinogenesis Enhancers. Mutat  Res  86
 217-229.
 OGAWA, K., ONOE, T., andTAKEUCHI, M. (1981). Spontaneous occurrence of q-glutamyl transpep-
 tidase-positive hepatocytic foci in 105-week-old Wistar and 72-week-old Fischer 344 male rats  JNCI  67
 407-412.                         ,                                                  .        '
 WARD, J.M. (1983). Increased susceptibility of livers of aged F344/NCr rats to the effects of phenobarbi-
 tal on the incidence, morphology, arid histochemistry of hepatocellular foci and neoplasms. JNCI  71,
 815-823.
 SCHULTE-HERMANN, R., TIMMERMANN-TROSIENER, I., and SCHUPPLER, J. (1983).  Promo-
 tion of spontaneous preneoplastic cells in rat liver as a possible explanation of tumor  production by non-
 mutagenic compounds. Cancer Res. 43, 839-844.
 INTERNATIONAL AGENCY FOR RESEARCH ON CANCER (IARC). (1979). Chemicals and indus-
 trial processes associated with cancer in humans. IARC Monographs, Volumes 1 to 20. Lyon, France:  In-
 ternational Agency for Research on Cancer.
 GRIESEMER, R.A., and CUETO, C., JR. (1980). Toward a classification scheme for degrees of experi-
 mental evidence for the carcinogenicity of chemicals for animals. IARC Scientific Publications No  27
 259-281.                                                                                '    '
 SQUIRE,  R.A. (1981). Ranking animal carcinogens: A proposed regulatory approach. Science 214,
 877—880.
SIMMON, V.F., KAUHANEN, K., and TARDIFF, R.G. (1977).  Mutagenic activity of  chemicals identi-
 fied m drinking water. In: Progress in Genetic Toxicology. D. Scott, B.A. Bridges, and F.H. Sobels (eds.).
Amsterdam: North-Holland, pp.  249-258.
                                              140

-------
                               RISK ASSESSMENT OF DEHP

24. YAGI, Y., TUTIKAWA, K., and SHIMOI, W. (1976). Teratogenicity and mutagenicity of a phthalate es-
    ter. Teratology 14, 259-260.
25. RUBIN, R.J., KOZUMBO, W., and KROLL, R. (1979). Ames mutagenic assay of a series of phthalic acid
    esters: positive response of the dimethyl and diethyl esters in TA100. In: Abstracts of Papers, Eighteenth
    Annual Meeting, Society of Toxicology, Inc.,  New Orleans, Louisiana. New York: Academic Press.
26. KIRBY, P.E., PIZARELLO, R.F., LAWLOR, T.E., HAWORTH, S.R., and HODGSON, LR. (1982).
    Evaluation of di-(2-ethylhexyl)phthalate and its major metabolites in the Ames test and L5178Y mouse
    lymphoma mutagenicity assay. Environ. Mutagen. 5, 657-663.  •
27. ZEIGER, E.,  HAWORTH, S., SPECK, W., and MORTELMANS, K. (1982). Phthalate estertesting in
    the National Toxicology Program's Environmental  Mutagenesis Test  Development Program. Environ.
    Health Perspect. 45, 99-101.
28. WARREN, J.R., LALWANI, N.D., and REDDY, J.K. (1982). Phthalate esters as peroxisome prolifera-
    tor carcinogens. Environ. Health Perspect. 45, 35-40.
29. BARBER, E.D. MLJLHOLLAND, A., JAGANNATH, D.R., CIFONE, M., CIMINO, M., MYHR, B.,
    and RUNDELL, J. (1985). The testing of di(2-ethylhexyl)phthalate (DEHP), mono(2-ethylhexyi)phthalate
    (MEHP), di(2-ethylexyl)adipate (DEHA), and 2-ethylhexanol (2EH) in a battery of genotoxicity assays.
    Paper presented at Society of Toxicology meeting, March, 1985.
30. TOMITA, I., NAKAMURA, Y., AOKI, N., and INUI, N. (1982). Mutagenic/carcinogenic potential of
    DEHP and MEHP. Environ. Health Perspect. 45, 119-125.
31. YOSHIKAWA, K., TANAKA, A., YAMAHA, T., and KURATA, H. (1983). Mutagenicity study of nine
    monoalkyl phthalates and a dialkyl phthalate using  Salmonella typhimurium and Escherichia coii. Food
    Chem. Toxicol. 21, 221-223.
32. DIVINCENZO, G.D.,  DONISH, W.H., MUELLER, K.R., HAMILTON, M.L.,  and BARBER,  E.D.
    (1983). Mutagenicity testing of urine from rats dosed with 2-ethylhexanol derived plasticizers (abstf.). En-
    viron. Mutagen. 5, 471.
33. MALCOLM, A.R.  (1982). Mutagenicity and tumor-promoting potential of di-(2-ethylhexyl)phthalate in
    Chinese hamster cells. United States Environmental Protection Agency. Environmental Research Labora-
    tory. Narragansett,  R.I. Draft.
34. HODGSON, J.R., MYHR, B.C., McKEON, M., and BRUSICK, D.J. (1982). Evaluation of di-(2-ethyl-
    hexyl)phthalate and its major metabolites in the primary rat hepatocyte unscheduled DNA synthesis assay
    (abstr.). Environ. Mutagen. 4, 388.
35. BUTTERWORTH, B. (1984). The genetic toxicology of di(2-ethylhexyl)phthalate (DEHP). CUT Activi-
    ties 4, 1-8.
36. TURNER, J.H., PETRICCIANI, J.C., CROUCH, M.L., and WENGER, S. (1974). An evaluation of the
    effects of diethylhexyl  phthalate (DEHP) on mitotically capable cells in blood packs. Transfusion 14,
    560-566.
37. STENCHEVER, M.A., ALLEN,  M.A., JEROMINSKI, L., and PETERSEN, R.V.  (1976). Effects of
    bis(2-ethylhexyl)phthalate on chromosomes  of human leukocytes and human  fetal lung cells. J. Pharma-
    ceut. Sci. 65, 1648-1651.
38. TSUCHIYA, K., and HATTORI,  K. (1976). A chromosomal study of cultured human leukocytes treated
    with phthalic acid esters. Rep. Hokkaido Inst. Public Health 26, 114.
39. ABE, S., and SASAKI, M. (1977). Chromosome aberrations'and sister chromatid exchanges in Chinese
    hamster cells exposed to various chemicals. JNCI 58, 1635-1641.
40. PHILLIPS,  B.J., JAMES, T.E.B., and GANGOLLI, S.D. (1982). Genotoxicity  studies of di(2-ethyl-
    hexyl)phthalate and its metabolites in CHO cells. Mutat. Res. 102, 297-304.
41. ISHIDATE, M., JR., and ODASHIMA, S. (1977). Chromosome tests with 134 compounds on Chinese
    hamster cells  in vitro—A screening for chemical carcinogens. Mutat. Res. 48, 337-354.
42. THIESS, A.M., and FLEIG, I. (1978). Chromosomal studies of employees after exposure to di-2-ethyl-
    hexyl phthalate (OOP) Zentralbl. Arbietsmed. Arbietsschutz Prophyl. 28, 351-355.
43. PUTMAN, D.L., MOORE, W.A., SCHECHTMAN, L.M., and HODGSON, J.R. (1983). Cytogenetic
    evaluation of di-(2-ethylhexyl)phthalate and its major metabolites in Fischer 344 rats. Environ. Mutagen.
    5,,227-231.
44. SINGH, A.R., LAWRENCE, W.H., and AUTIAN, J. (1974). Mutagenic and antifertility sensitivities of
    mice to  di-2-ethylhexyl phthalate (DEHP) and dimethoxy-ethyl phthalate (DMEP). Toxicol. Appl. Phar-
    macoi. 29, 35-46.
45. AUTIAN, J.  (1982). Antifertility effects and dominant lethal assays for mutagenic effects of DEHP. En-
    viron. Health Perspect. 45, 115-118.
                                               141'

-------
                                 TURNBULL AND RODRICKS
  46

  47


  48,


  49.

  50.

  51.

  52.


  53.

  54.

  55.

 56.

 57.


 58.

 59.

 60.

 61.

 62.

 63.


 64.

 65.

 66.

 67.

 68.


 69.

70.

71.
  RUSHBROOK, C.J., JORGENSON, T.A., and HODGSON, J.R. (1982). Dominant lethal study of di-(2-
  ethylhexyl phthalate and its major metabolites in ICR/SIM mice (abstr.). Environ. Mutagen 5, 387
  ALBRO, P.W., CORBETT, J.T., SCHROEDER, J.L., JORDAN, S., and MATTHEWS, H.B. (1982).
  Pharmacokinetics, interactions with macromolecules and species differences in metabolism of DEHP En-
  viron. Health Perspect. 45, 19-25.
  VON DANIKEN, A., LUTZ, W.K., JACKH, R., and SCHLATTER, C. (1984). Investigation of the po-
  tential for binding of di(2-ethylhexyl)phthaiate (DEHP) and di(2-ethyihexyl)adipate (DEHA) to liver DNA
  in vivo. Toxicol. Appl. Pharmacol. 73, 373-387.
  SEED, J.L. (1982). Mutagenic activity of phthalate esters in bacterial liquid suspension assays  Environ
  Health Perspect. 45, 111-114.
  OISHI, S., and HIRAGA, K. (1979). Effect of phthalic acid esters on gonadal function in male rats Bull
  Environ. Contain. Toxicol. 21, 65-67.
  OISHI, S., and HIRAGA, K. (1980). Effect of phthalic acid esters on mouse testes. Toxicol  Lett  5
  413-416.
  ALBRO, P.W., CORBETT, J.T., SCHROEDER, J.L., and JORDAN, S.T. (1983). Incorporation of ra-
  dioactivity from labeled di-(2-ethylhexyl)phthalate into DNA of rat liver in vivo. Chem. Biol. Interact. 44,
  1—16.
  WEISBURGER, J.H., and WILLIAMS, G.M. (1980). Chemical carcinogens. In: Casarett and Doull's
  Toxicology, 2nd ed. J. Doull, C.D. Klaassen, and M.O. Amdur (eds.). New York: Macmillan, pp. 84-138.
  MILLER, B.C. (1978). Some current perspectives on chemical carcinogenesis in humans and experimental
  animals. Cancer Res. 38, 1479-1496.                .
  MILLER, E.G., and MILLER, J.A. (1966). Mechanisms of chemical carcinogenesis: Nature of proximate
 carcinogens and interactions with macromolecules. Pharmacol. Rev. 18, 805-838.
  REDDY, J.K., AZARNOFF, D.L., and HIGNITE, C.E. (1980). Hypolipidaemic hepatic peroxisome pro-
 liferators form a novel class of chemical carcinogens. Nature 283, 397-398
 ROBINSON, E., LINGTON, A., DIVINCENZO, G., CHADWICK, M., BRANFMAN, A., SILVEIRA
 D., and McCOMISH, M. (1985). Nonlinearity of metabolism of DEHP with dose and prior exposure!
 Paper presented at Society of Toxicology Meeting, March 1985.
 MASTERS, C., and HOLMES, R. (1977). Peroxisomes: New aspects of cell physiology and biochemistry
 Physiol. Rev. 57, 816-882.
 LORD, J.M. (1980). Biogenesis of peroxisomes and glyoxysomes.  In:  Subcellular Biochemistry D B
 Roodyn (ed.). New York: Plenum Press, Vol. 7, pp. 171-211.
 TOLBERT, N.E. (1981). Metabolic pathways in peroxisomes and glyoxisomes. Annu. Rev. Biochem. 50,

 CHANCE, B., SIES, H., and BOVERIS, A. (1979). Hydroperoxide metabolism in mammalian organs
 Physiol. Rev. 59, 527-605.
 FREESE, E. (1971). Molecular mechanisms, of mutation. In: ChemicalMutagens. Principles andMethods
for Their Detection. A. Hollaender (ed.). New York: Plenum Press, Vol. 1, pp  1-56
 WANG, R.J., ANANTHASWAMY, H.N., NIXON, B.T., HARTMAN, P.S., and EISENSTARK, A.
 (1980). Induction of single-strand DNA breaks in human cells by H2O, formed in near UV (black light)-
 irradiated medium. Radial. Res. 82, 269-276.
SCHONEICH, J. (1967). The induction of chromosomal aberrations by  hydrogen peroxide in strains of
ascites tumors in mice. Mutat. Res. 4, 384-388.-
STICK, H.F., WEI, L., and LAM, P. (1978). The need for a mammalian test system for  mutagens: Action
of some reducing agents. Cancer Lett. 5,  199-204.
BRADLEY,  M.O., HSU, I.C., and HARRIS, C.C. (1979). Relationship between sister chromatid ex-
change and mutagenicity, toxicity and DNA damage. Nature 282, 318-320.
MACRAE, W.D., and STICH, H.F. (1979). Induction of sister chromatid exchanges in Chinese hamster
ovary cells by thibl and hydrazine compounds. Mutat. Res. 68, 351-365
PARSHAD,  R., TAYLOR, W.G., SANFORD, K.K., CAMALIER, R.F., GANTT, R., and TARONE,
R.E. (1980). Fluorescent light-induced chromosome damage in human IMR-90 fibroblasts. Role of hydro-
gen peroxide and related free radicals. Mutat. Res. 73,  115-124.
SPEIT, G., and VOGEL, W. (1982). The effect of sulfhydryl compounds on sister chromatid exchanges.
II. The question of cell specificity and the role of H2O2. Mutat. Res. 93, 175-183.
SPEIT, G., VOGEL, W., and WOLF, M. (1982). Characterization of sister chromatid exchange induced
by hydrogen peroxide. Environ. Mutagen. 4, 135-142.
LAZAROW, P.B., and DeDUVE, C. (1976). A fatty acyl-CoA oxidizing system in rat liver peroxisomes;
enhancement by clofibrate, a hypolipidemic drug. Proc. Natl. Acad. Sci. USA 73, 2043-2046.
                                             142

-------
                                RISK ASSESSMENT OF DEHP


72. REDDY, J.K., and LALWANI, N.D. (1983). Carcinogenesis by hepatic peroxisome proliferators: Evalu-
    ation of the risk of hypolipidemic drugs and industrial plasticizers to humans. CRC Crit. Rev. Toxicol. 12,
    1-58.
73. BLANE, G.F., and PINAROLI,  F. (1980). Fenofibrate: animal toxicology in relation to side effects in
    man. Nouv. Presse Med. 9, 3737-3746.  .
74. FAHIMI, H.D., REINICKE, A., SUJATTA, M., YOKATA, S., OZEL M., HARTIG, F., and STEG-
    MEIER,  K. (1982). The short-  and long-term effects of bezafibrate in the rat. Ann NY Acad. Sci. 386,
    111-135.
75. REDDY, J.K., WARREN, J.R., REDDY, M.K., and LALWANI, N.D. (1982). Hepatic and renal effects
    of peroxisome proliferators. Ann. NY Acad. Sci. 386, 81-110.
76. LAZAROW,  P.B. (1978). Rat  liver peroxisomes catalyze the 0-oxidation of fatty acids. J. Biol. Chem.
    253, 1522-1528.
77. LAZAROW,  P.B. (1977). Three hypolipidemic drugs increase hepatic palmitoylcoenzyme A oxidation in
    the rat. Science 197, 580-581.
78. OSUMI, T., and HASHIMOTO, T. (1978). Enhancement of fatty acyl-CoA oxidizing activity in rat liver
    peroxisomes by di-(2-ethylhexyl)phthalate. J. Biochem. 83, 1361-1365.
79. OSUMI, T., and HASHIMOTO, T. (1979). Subcellular distribution of the enzymes of the fatty acyl-CoA
    (3-oxidation system and their induction by di(2-ethylhexyl)phthalate in rat liver. J, Biochem. 85, 131-139.
80. MORTON, S.J. (1979). The hepatic effects of dietary di-2-ethylhexyl phthalate. Ph.D. Thesis, The Johns
    Hopkins University, Baltimore., MD.
81. CANNING, A.E., and DALLNER, G. (1981). Induction of peroxisomes and mitochondria by di(2-.ethyl-
    hexyl)phthalate. FEES Lett. 130, 77-79.
' 82. CANNING, A.E., KLASSON, E., BERGMAN, A., BRUNK, U., and DALLNER, G. (1982). Effect of
    phthalate ester metabolites on  rat liver. Acta Chem. Scand.  Sec. B. B36, 563-565.
83. ALBRO, P.W., HAAS, J.R.,  PECK, C.C., ODAM, D.G., CORBETT, J.T., BAILEY, F.J., BLATT,
    H.E., and BARRETT, B.B. (1981). Identification of the metabolites of di-(2-.ethylhexyl)phthalate in urine
    from the African green monkey.  Drug Metab. Dispos. 9, 223-225.
 84. NIXON, J., and JACKSON, S.J. (1982). Bis(2-Ethylhexyl)Phthalate: A Comparative Subacute Toxicity
    Study in the Rat and Marmoset. Report No. CTL/P/690. Prepared for CEFIC (Council European of the
    Federations of the Industry Chemical).
 85. INTERAGENCY  REGULATORY LIAISON GROUP (IRLG). (1979). Scientific bases for identification
    of potential carcinogens and estimation of risks. JNCI 63, 244-268.
 86.  MITCHELL, A.M., LHUGUENOT, J.C., BRIDGES, J.W., and  ELCOMBE, C.R. (1984). Identifica-
    tion of the proximate peroxisome proliferator(s) derived from di-(2-ethylhexyl)phthalate. Toxicol. Appl.
     Pharmacol. In press.
 87.  HILL, H.A.O. (1979). The chemistry of dioxygen and its reduction products. In:  Oxygen Free Radicals
    and Tissue Damage. Ciba Foundation Symposium 65, 5-17.
 88.  HIROTA, N., and YOKOYAMA, T. (1981). Enhancing effect of hydrogen peroxide upon duodenal and
     upper jejunal carcinogensis in  rats. Gann 72, 811-812.
 89.  ITO, A., NAITO, M., NAITO, Y., and WATANABE, H. (1982). Induction and characterization of gas-
     tro-duodenal lesions in mice given continuous oral  administration of hydrogen peroxide. Gann 73, 315-
     322.
 90.  LEVIN, D.E., HOLLSTEIN, M., CHRISTMAN, M.F., SCHWIERS, E.A., and AMES, B.N. (1982). A
     new Salmonella tester strain (TA 102) with A"T base pairs at the site of mutation detects oxidative muta-
     gens. Proc. Natl.  Acad. Sci. USA 79, 7445-7449.
 91.  AMES,  B.N. (1983). Dietary carcinogens and anticarcinogens. Science 221, 1256-1264.
 92.  ALBRO, P.W., THOMAS, R., and FISHBEIN, L. (1973). Metabolism of diethylhexyl phthalate by rats.
     Isolation and characterization of the urinary metabolites. J. Chromatography 76, 321-330.
 93.  ALBRO, P.W., TONDEUR, I., MARBURY, D., JORDAN, S., SCHROEDER, J., and CORBETT, J.T.
     (1983). Polar metabolites of di-(2-ethylhexyl)phthalate in the rat. Biochem. Biophys. Acta 760, 283-292.
 94.  MORAN, E., LINGTON, A., DIVINCENZO, G., ROBINSON, E., CHADWICK, M., BRANFMAN,
     A., McCOMISH, M., and SILVEIRA, D. (1985). Species differences in the metabolism of DEHP in mon-
     keys, rats, and mice. Paper presented at Society of Toxicology meeting, March 1985.
 95. FOOD SAFETY COUNCIL. (1980). Proposed System for Food Safety Assessment. Final Report of the
     Scientific Committee of the Food Safety Council, Washington, D.C., June, 1980.
 96. BROWN, C.C. (1984). H5gh-to  low-dose extrapolation in animals. In: Assessment and Management of
     Chemical Risks. J.V. Rodricks and R.G. Tardiff (eds.). ACS Symposium Series 293. Washington, D.C.:
     American Chemical Society, pp. 57-79.
                                               143

-------
                               TURNBULL AND RODRIGKS


 97.  GEHRING, P.J., and BLAU, G.E. (1977). Mechanisms of carcinogenesis: Dose response. J. Environ.
     Pathol. Toxicoi. 1, 163-179.
 98.  REITZ, R.H., GEHRING, P.J., and PARK, C.N. (1978). Carcinogenic risk estimation for chloroform:
     An alternative to EPA's procedures. Food Cosmet. Toxicoi. 16, 511-514.
 99.  GEHRING, P.J., WATANABE, P.O., and PARK, C.N. (1978). Resolution of dose-response toxicity
     data for chemicals requiring metabolic activation: Example—vinyl chloride. Toxicoi. Appl. Pharmacol.
     44, 581-59L
100.  LHUGUENOT, J.C., MITCHELL, A.M., and ELCOMBE, C.R. (1984). Dose- and time-dependency of,
     mono(2-ethylhexyl)phthalate (MEHP) metabolism in primary rat hepatocyte cultures. Presented at Society
     of Toxicology Conference, Atlanta, GA, March 1984. Abstract No. 383.
101.  LHUGUENOT, J.C., LOCK, E.A., and ELCOMBE, C.R. (1984). Dose- and time-dependency of mono-
     and di-(2-ethylhexyl)phthalate (MEHP and DEHP) metabolism in rats. Presented at Society of Toxicology
     Conference, Atlanta, GA, March 1984. Abstract No. 382.
102.  LHUGUENOT, J.C., MITCHELL, A.M., and ELCOMBE, C.R. (1984). Dose- and time-dependent me-
     tabolism of mono(2-ethylhexyl)phthalate and its major metabolites in rat hepatocyte cultures. Toxicoi.
     Appl. Pharmacol. In press.
103.  ANDERSON, M.W., HOEL, D.G., and KAPLAN, H.L. (1980). A general scheme for the incorporation
   , of pharmacokinetics in low dose risk estimation for chemical carcinogenesis: Example —vinyl chloride.
     Toxicoi. Appl. Pharmacol. 55, 154-161.
104.  CRUMP, K.S., HOEL, D.G., and LANGLEY, C.H. (1976). Fundamental carcinogenic processes and
     their implications for low dose risk assessment. Cancer Res. 36, 2973-2979.
105.  CRUMP, K.S., HOWE, R., and FIERING, M.B. (1980). Approaches to carcinogenic, mutagenic and
     teratogenic risk assessment. Task A, Subtask No. 5, Summary Report, contract No. 68-01-5975, U.S. En-
     vironmental Protection Agency.
106.  U.S. ENVIRONMENTAL PROTECTION AGENCY (EPA). (1980). Water Quality Criteria Document;
     Availability. Appendix C—Guidelines and Methodology Used in the Preparation of Health Effect Assess-
     ment Chapters of the Consent Decree Water Critera Documents. Fed. Reg. 45, 79347-79357.
107.  CONSUMER PRODUCTS SAFETY COMMISSION (CPSC). (1983). Children's Chemical Hazards. Risk
     Assessment on Di-(2-Ethylhexyl)Phthalate in Children's Products. Chemical Hazards Program, Director-
     ate for Health Sciences, CPSC. August 1983.
108.  DAY,  N.E., and BROWN, C.C. (1980). Multistage models and primary prevention of cancer. JNCI 64,
     977-989.
109.  MANTEL, N., and BRYAN, W. (1961). "Safety" testing of carcinogenic agents. JNCI 27, 455-470.
110.  MANTEL, N., BOHIDAR, N.R., BROWN, C.C., CIMINERA, J.L., and TUKEY, J.W. (1975). An im-
     proved Mantel-Bryan procedure for "safety" testing of carcinogens. Cancer Res. 35, 865-872.
111.  NATIONAL ACADEMY OF SCIENCES. (1977). Drinking Water and Health. Washington, D.C.: NAS.
112.  HOEL, D.G., GAYLOR, D., KIRSCHSTEIN, R., SAFFIOTTI, U., and SCHNEIDERMAN, M. (1975).
     Estimation of risks of irreversible delayed toxicity. J. Toxicoi. Environ.  Health 1, 133-151.
113.  NATIONAL ACADEMY OF SCIENCES. (1975). Pest Control: An Assessment of Present and Alterna-
     tive Technologies. Vol. 1. Contemporary Pest Control Practices and Prospects: The Report of the Execu-
     tive Committee. Washington, D.C.: NAS.
114.  CROUCH, E., and WILSON, R. (1979). Interspecies comparison of carcinogenic potency. J. Toxicoi. En-
     viron.  Health 5,  1095-1118.
115.  FREIREICH, E.J., GEHAN, E.A., RALL, D.P., SCHMIDT, L.H., and SKIPPER, H.E. (1966). Quan-
     titative comparison of toxicity of anticancer agents in mouse, rat, hamster, dog, monkey, and man. Can-
     cer Chemother. Rep. 50, 219-244.
116.  SCHUMANN, A.M.,  QUAST, J.E., and WATANABE, P.O. (1980). The pharmacokinetics and macro-
     molecular interactions of perchloroethylene in mice and rats as related to oncogenicity. Toxicoi. Appl.
     Pharmacol. 55, 207-219.
117.  MILLER,  E.G.,  MILLER, J.A., AND ENOMOTO, M. (1964). The comparative carcinogenicities of 2-
     acetylaminofluorene and its N-hydroxy metabolite in mice, hamsters, and guinea pigs. Cancer Res. 24,
     2018-2031.
118.  JAEGER,  R.J., and RUBIN, R.J. (1970). Plasticizers from plastic devices. Extraction, metabolism, and
     accumulation by biological systems. Science 170, 460-462.
119.  JAEGER,  R.J., and RUBIN, R.J. (1973). Extraction, localization, and metabolism of di-2-ethylhexyl
     phthalate from PVC plastic medical devices. Environ. Health Perspect. 3, 95-102.
120. REDDY, J.K., and RAO, M.S. (1978). Enhancement by Wy-14,643, a hepatic peroxisome proliferator, of
     diethylnitrosamine-initiated hepatic tumorigenesis in the rat. Br. J.  Cancer 38, 537-563.
                                              144

-------
                                RISK ASSESSMENT OF DEHP

 121. WARD, J.M., RICE, J.M., CREASIA, D., LYNCH, P., andRIGGS, C. (1983). Dissimilar patterns of
     promotion by di(2-ethylhexyl)phthalate and phenobarbital of hepatocellular neoplasia initiated by diethyl-
     nitrosamine in B6C3F1  mice. Carcinogenesis 4,, 1022-1029.
 122. DEANGELO, A.B., and GARRETT, C.T. (1983). Inhibition of development of preneoplastic lesions in
     the livers of rats fed a weakly carcinogenic environmental contaminant. Cancer Lett. 20, 199-205.
 123. PURCHASE, I.F.H. (1980). Interspecies comparisons of carcinogenicity. Br. J.  Cancer 41, 454-468.
 124. HINTON, R.H., PRICE, S.C., HALL, D., GRASSO, P., and MITCHELL, F. (1982). Report to CEFIC
     on a 28 Day Dose and Time Response Study on Di-ethylhexyl Phthalate in Rats. Study 5/81/TX. Robens
     Institute of Industrial and  Environmental Health and Safety, University of Surrey, Guildford, Surrey,
     England.
 125. BRITISH INDUSTRIAL BIOLOGICAL RESEARCH ASSOCIATION (BIBRA). (1984). A 21-day dose-
     response study of di(2-ethylhexyl)phthalate in rats. Report No. 0512/1/84. Draft. July 1984.

                                                                   Address reprint requests to:
                                                                             Duncan Turnbull
                                                                         Environ Corporation
                                                                   WOO Potomac Street, N. W.
                                                                        Washington,  DC 20007

Submitted July 26, 1984
Accepted February 11, 1985
                                              145

-------
A Time-to-Response Perspective on
 Ethylene Oxide's Carcinogenicity
       Robert LSielken Jr.
           President
          Sielken, Inc
  Suite 410,3833 Texas Avenue
       Bryan,Texas 77802
        September 1987

-------
   (b) the secondary data set, equaling the primary data set plus the histologically
     examined female rats which were sacrificed at the end of the experiment,
   (c) the EPA data set, corresponding to the secondary data set minus the rats
     which died or were sacrificed prior to 18 months plus the 18-month
     scheduled sacrifices in the 0 ppm and 100 ppm groups but not in the 10 ppm
     and 33 ppm groups. (49)
More detailed time-to-response data was also available. Time was measured in
terms of months from the onset of exposure. The time of death was recorded for all
rats which died. No further classification of the cause of death for these animals
was reported.
   The carcinogenic response analyzed in this chapter is defined as death with MCL
in a female rat.
   Any female rat which had MCL at the time it died or was sacrificed was
considered to have had the carcinogenic response, death with MCL

Risk Characterizations Emphasizing Time-to-Response but Not Requiring Dose-
Response Modeling
   It is appropriate to characterize risk in terms of the time that a response occurs or
the amount of time proceeding a particular consequence (29,36,39). Some such
characterizations can be made without having to do dose-response modeling. As
shown below, these largely assumption free, straightforward representations of the
experimental data suggest a high degree of similarity between the real risks at
0 ppm and 10 ppm.
   Table 2 indicates the number of female rats which died with mononuclear cell
leukemia (MCL) during each of the exposure months, the number which died
without MCL, and the number which died but were lost to follow-up (that is, not

-------
histologically examined for MCI). There were 120 female rats in each of the two
identical control groups (denoted as Control I and Control II or Cl and Cll herein) and
in each of the 10 ppm, 33 ppm, and 100 ppm exposure groups. The control level is
referred to herein as the 0 dose level.
   There were, 1,10,10, and 20 rats randomly selected at 3,6,12, and 18 months
respectively for interim sacrifice at each dose level (0,10.33, and 100 ppm). There
was also a terminal sacrifice at approximately 24 months for all female rats.   All
rats are accounted for in Table 2.' The animals which were randomly selected for
sacrifice at scheduled times can be distinquished in Table 2 from the other rats; in
that, a table entry is a ratio like x/y if there were y > 0 rats which were scheduled
sacrifices. For example. Table 2 implies the following:
        (a) there were no female rats which died with  MCL (among those not lost
           to follow-up) before 18 months of exposure;
       1$
        (b) in Control II and at 33 and 100 ppm one. rat died with MCL during the
           18-th exposure month;
        (c) the one rat which died with MCL at 100 ppm, during the 18-th
           exposure month, was a scheduled sacrifice; and,
        (d) at 100 ppm there were a total of 16 female rats which died with MCL in
           scheduled sacrifices while 12 died with* MCL by deaths that were not
                                                                         *
           scheduled sacrifices.
   The life table analysis estimates of the probabilities of a female rat dying with
MCL following various numbers of exposure months are shown in Table 3.  These
estimates are straightforward nonparametric estimates in that they do not presume
any model or mathematical form for how these probabilities change with dose or
time.  The estimates are computed using the fairly standard technique described in
(47). This method  incorporates all of the rats and the length of time they were at
risk.

-------
                                                                       11
   There are at least two conclusions which can be drawn from Tables 2 and 3. The
 first is that, if a response occurs at all, it occurs very late relative to an average rat
 lifetime -- mostly in the 24-th month and no sooner than the 18-th month in a 24-
 month study. The second conclusion is that the number and timing of the responses
 atO ppm and 10 ppm are very similar. The latter conclusion is also evident in Figure
 1 which indicates the plots of the life table analysis estimates versus time for the 10
 ppm exposure group and the two control groups. The plot for the 10 ppm group
 lies between those of the two control groups except for the very last month.
   There are several other ways to characterize
        (a)  the lateness of the MCL response in female rats,
        (b)  the similarity between the response times for the rats exposed to 10
            ppm and the response times for the control rats, and
        (c)  the changes in the response times at 33 ppm and 100 ppm relative to
            those at 0 ppm and 10 ppm.
 For example, Table 4 indicates the number of exposure months until the proportion
 of animals which die with MCL reaches a prescribed percentage (1%, 5%, 10%, or
 20%). In particular, it took the same number of months (23 months) for both the 10
 ppm group and the combined controls to reach a 5% incidence rate whereas it took
 a noticeably shorter time attKe 33 ppm and 100 ppm levels (namely, 20 and 21
 months respectively). Throughout Table 3 the rats at 0 ppm and 10 ppm took similar
times to reach the same incidence rate (risk level); in fact, it never took less time to
 reach a specified incidence rate at 10 ppm than it did for the combined controls.
   The average amount of time until a response occurs cannot be calculated when
some individuals do not develop the response. However, the average amount of
time in an experiment during which an individual has not had the response can be
calculated. This latter average, called the mean response free period,  is not the same
as the average time until the response occurs.  For instance, if no rats died with MCL

-------
during a 24 month experiment, the average response free period would be 24
months; however, the average time to death with MCL would certainly be greater
than 24 months but how much greater is unknown..
   Table 5 shows the average number of exposure months in the experiment that
the female rats were free from the specified response, death with MCL. The
maximum possible response free period is 24 months (the length of the
experiment). The female rats in the 10 ppm group averaged 23.55 months of
response free time (98.1 % of the maximum response free period). The two control
groups averaged 23.45 months (97.7%) and 23.70 months (98.7%) without
mortality with MCL. Hence, the average for the 10 ppm group is between the
averages for the two control groups. The average for the 33 ppm and 100 ppm
groups were slightly lower (namely, 23.10 months (96.2%) and 22.88 months
(95.3%) respectively) but noticeably different from that for the 10 ppm group. The
similarity between the control groups and the 10 ppm group is clear. The
dissimilarity between the 10 ppm group and both the 33 ppm and TOO ppm groups
is also clear.

Simple Statistical Analysis of Time-to-Response Data

   Several two-sided 95% confidence intervals on the lifetime probability of a
female rat dying with MCL are shown in Table 6. Regardless of which subset of the
                                         a
experimental data is being considered, there is substantial overlap between the
confidence intervals for the MCL response rate at 0 ppm and at 10 ppm (see also
Figure 2). There is no statistically significant difference between the lifetime MCL
response rate for the 10 ppm group and the combined controls (even at the 20%
significance level). This suggests that there  is no statistically detectable increase in

-------
                                                                       13
the lifetime probability of dying with MCL from being exposed to 10ppmEO
instead of 0 ppm.
   Not only is the lifetime probability of dying with MCL similar at 0 ppm and 10
ppm, but also there is no difference between 0 ppm and 10 ppm in when an
individual is likely to die with MCL  In particular, being at 10 ppm instead of 0 ppm
does not appear to increase the likelihood of dying with MCL earlier. The
hypothesis that there is no shifting of the likelihood of dying with MCL from one
time to another (especially to an earlier time) is the hypothesis that the distribution
of the time to death with MCL is the same for different dose levels. This hypothesis
                                                        i
was tested using the available information on the length of time without a female
rat dying with MCL. The statistical procedures decribed in (48) are appropriate for
this analysis and were applied to the data on the female rats which were
histologically examined for MCL and which died or were sacrificed at the end of the
experiment.  (The procedures considered the frequency of death with MCL in three
intervals: 0-18,19-21, and 22-24exposure months.) The hypothesis that the 0,10,
33, and 100 ppm groups had the same time-to-response distribution was rejected at
the 5% significance level. The hypothesis that the 0,10, and 33 ppm groups"had the
same time-to-response distribution was also rejected. However, the hypothesis that
the controls and the 10 ppm group had the same time-to-response distribution was
not rejected. These tests, which do not require any specific dose-response modeling
assumptions, imply that the time-to-response distributions for the controls and the
rats at 10 ppm are similar, and that the time-to-response distributions at 33 ppm
and 100 ppm are different from those at 0 ppm and 10 ppm. These results are
consistent with the implications of Table 5 where the estimated mean times without
a female rat dying with MCL are nearly identical at 0 ppm and 10 ppm but smaller at
33 ppm and 100 ppm. In short, even though the same cannot be said for the 33 ppm

-------
                                                                         u
 and 100 ppm groups, the 10 ppm group is similar to the controls with respect to
 when and how often death with MCL occurs.

 EPA's Position on Time-to-Response Information

   The final report (49) published by the U.S. EPA in June, 1985, contains a summary
 of their quantitative risk assessment for EO. Their approach to estimating the risk at
 very low doses like many of the past attempts by regulatory agencies, fails to reflect
 the available time-to-response information in either its dose-response modeling or
 its risk characterizations. The methodology to incorporate this additional
 information has been available for almost 10 years and, as discussed previously, it
 should be a part of most assessments if the objectives of the EPA's Cancer Risk
 Guidelines (46) and those of the NAS(1) are to be met.
   There have frequently been several methodogical shortcomings in the past
 quantitative risk assessments prepared by regulatory agencies for ethylene oxide,
 and other chemicals, in addition to the failure to reflect the available time-to-
 response information. Many of these shortcomings are discussed briefly in
 Appendix B.                                      *
EPA's Interpretation of the Ethvlene Oxide Bioassav Data

    The EPA's quantitative risk assessment failed to incorporate a consideration of
the time-to-response information in its summary risk characterizations (49).  Since
1982, the EPA has summarized the results of their risk modeling efforts in terms of
what EPA calls "unit risk estimates" (although these are really bounds on risk and
not estimates). These risk characterizations ignore the time-to-response
information and are based solely on the proportions of animals developing the

-------
                                                                       15
specified response at any time. Such tumor/no tumor data or response/no response
data are quantal response data and do not reflect the time at which the response of
concern occurs.
   In EPA's description of the BRRC study, the cumulative mortality percentages
versus exposure months were tabulated, and the agency noted that
      "At no time were significant increases in mortality observed in the 10 ppm
      exposure group of either sex." (49, page 9-100)
Regretably they did not present the corresponding tables and implications for death
with MCL or other potential carcinogenic responses. Had they done so, they would
have discovered that the similarity between 0 ppm and 10 ppm observed for
mortality from all causes combined extends to the specific carcinogenic responses
they eventually focused on (MCL in particular).  Furthermore, they would have
realized that there was more to the risk characterization of EO than that suggested
by the quantal response data alone.
   The only time-to-response information for the carcinogenic responses which the
EPA reported was a table of "Time in months to First Tumor" and "Time in months
to Median Tumor". EPA summarized their interpretation of this data in the
following  two sentences (49, page 9-105):
        "The time to first tumor for some neoplasms (but not for mononuclear cell
        leukemias) was decreased in the high-dose group as compared to controls,
        as shown in Table 9-23. Median time-to-tumor was not reduced."
This information deserves more emphasis and, since they pursued no further time-
to-response analyses, more discussion and follow-up. The similarity of the time-to-
response behavior at 10 ppm to that at the control level was not explicitly noted by
the EPA, although, as discussed previously, it woulcl appear to be very relevant data
to be considered by risk managers. Further, the omission of most of the time-to-

-------
                                                                          16
 response information and implications should be included in the EPA's discussion of
 the uncertainties in their risk assessment.
 Effect of Evaluating Only a Particular Portion of the Quantal Response Data

    The inclusion or exclusion of subgroups of female rats on the basis of different
 factors such as sacrificed versus non-sacrificed or gross examination versus
 histological examination and inconsistency across the dose levels in the
 determination of the number of animals at risk caused the EPA risk characterization
 to differ markedly from the risk characterizations corresponding to alternative (and
 probably more reasonable) representations of the quanta! response data. In most
 experiments there are several ways to detemine which animals are included in the
 quantal response data and which animals are excluded (not counted). The different
 determinations may not be equally appropriate or equally reflective of the
 underlying dose-response relationship. Furthermore, the same risk characteristic
 can differ by several orders of magitude for different determinations of the quantal
 response data. The most informative determination may depend on the
 experimental protocol and the chemical's time-to-response behavior. The
 sensitivity of the risk characterization to the determination of the representation of
 the quantal response data should be evaluated and made clear to the risk manager.
   In the primary data set the observed proportion dying with  MCL did not increase
 as the dose increased. As indicated in Table 7, the observed percentages were
 30.6%, 13.6%, 32.3%, and 25.5% for the combined controls, 10 ppm, 33 ppm, and
 100 ppm groups respectively. In fact, the percentages at 10 ppm, 33 ppm and 100
ppm are between the 22.2% and 38.9% observed in the two  individual control
groups. Among the female rats which died there was no increased risk of dying

-------
                                                                         17
with mononuclear cell leukemia from a lifetime of inhalation exposure to ethylene
oxide at or below 100 ppm.
   In their analysis (49), the EPA evaluated a mixture of sacrificed and non-sacrificed
animals. The usefulness of this analysis is questionable since there was not an '
increasing dose-response relationship observed among the non-sacrificed animals,
but there was one observed in the female rats sacrificed at 24 months.
   The observed percentages of female rats with MCL in the 24-month sacrificed
group were 9.5%, 20.4%, 29.2%, and 57.7% in the combined controls, 10 ppm, 33
ppm, and  100 ppm groups respectively. The percentage for the 24-month sacrificed
group at 100 ppm is significantly greater than the percentage for the non-sacrificed
group at 100 ppm (at the 1% significance level). The corresponding differences in
the percentages for the 10 ppm and 33 ppm are not significant (even at the 20%
significance level). However, the percentage for the 24-month sacrificed controls is
significantly decreased relative to the percentage for the non-sacrificed controls (at
the 1% significance level).
   The reasons for the observed differences between the non-sacrificed and
24-month sacrificed rats are not evident from the numbers themselves. However,
the numbers are consistent with the hypothesis that the dose-related occurrences of
MCL were "incidental" as opposed to "lethal"; that is, consistent with the dose-
related occurrences not leading to an increased threat to life. The numbers are also
consistent with the claim that the dose-related occurrences of MCL happen very late
in the lifetime of female rats. In either case, the EPA's risk characterizations do not
reflect health impact since they either are reflecting an incidental carcinogenic
response and/or they do not reflect the amount of a lifetime adversely affected by
exposure.
   The specific criteria used by EPA to select their single representation of the
quantal response data did not have a consistent impact on the different dose

-------
                                                                        18
 groups. In fact, the criterion of including only histologically examined rats appears
 reasonable at first glance; however, it meant that the 20 female rats sacrificed at 18
 months in each of the 10 ppm and 33 ppm groups (i.e., about 17% of these dose
 groups) were excluded from the quantal response data (because they were only
 grossly examined) while the 40 female control rats and the 20 female rats at 100
 ppm which were sacrificed at 18 months were included (because they were
 histologically examined). It is true that the rats in the 10 ppm and 33 ppm groups
 were only grossly examined for MCLand not histologically examined; however,
 none of the 101 female rats which were histologically examined before 18 months
 had MCL, and only 1 of the 20 histologically examined female rats which were
 exposed to 100 ppm and sacrificed at 18 months had MCL Thus, the female rats
 which were at 10 ppm and 33 ppm and sacrificed at 18 months probably did not
 have MCL Therefore, by including the 40 control rats sacrificed at 18 months and
 excluding the 20 rats at each of 10 ppm and 33 ppm which were sacrificed at 18
 months, EPA's choice decreased the proportion with MCL among the controls and,
 most likely, increased the proportion with MCL at 10 ppm and 33 ppm. This choice
 exaggerates any increase in the proportion with MCL at 10 ppm and 33 ppm relative
to the controls. In short, the  EPA's non-uniform treatment of the female rats
sacrificed at 18 months results in biased dose-response information which inflates
the risk characterization -- in fact, more than doubled the implied risk.
   Several alternatives to the EPA's representation of the quantal response.data on
female rats with MCL are indicated in Table 7. They include
       (a) the primary data set,
       (b) the secondary data set -- the primary data set plus the 24 month
           sacrifices,
       (c) EPA's choice plus the rest of the 18-month sacrifices,

-------
                                                                        19
        (d)  all histologically examined female rats which survived the first year of
            the experiment, and
        (e)  the female rats in (d) plus the rest of the 18-month sacrifices.
In all of these alternatives, the apparent increase in the percentage of female rats
with MCI between the controls and the 10 ppm group is less than the implied.
increase under EPA's choice. For alternative (a), the primary data set, there is no
implied increase. For alternatives (b), (c), and (e) the implied increases are only
roughly one-half of the implied increase under the EPA's choice. Alternative (d)
suffers from the same disparate treatment of the 18-month sacrifice data as EPA's
choice and implies risks twice those implied by (e).
   All of the representations of the quanta! response data on MCL which combine
the sacrificed and non-sacrificed female rats hide the fact that there was no
observed increasing dose-response relationship among the non-sacrificed rats.  Of
course, these representations also failed to indicate the very long period of time
without evidence of a carcinogenic effect.

Risk Characterizations can Emphasize Time-to-Response
   As suggested previously, a practical way to characterize the effects of a particular
exposure is to describe the corresponding average amount of time in a specified
observation period during which the subject is free from a specified response. This
characterization has been called the mean response free period or mean free period
(36,39). If the observation period is 24 months (as it was in the BRRCstudy), then
the mean response free period could be as long as is 24 months. As indicated in
Table 5, when the response is death with MCL, the observed mean response free
periods (observed mean numbers of months without dying with MCL) are 23.58,
23.55,23.10, and 22.88 months for a female rat at 0,10,33, and 100 ppm

-------
                                                                        20
 respectively. Thus, the risk for female rats at 10 ppm can be characterized as a
 decrease of 0.03 months in the mean response free period (0.03 = 23.58 - 23.55).
    For female rats exposed between 0 ppm and 10 ppm, the overall observed rate
 of decrease in the observed mean respqnse free periods for dying with MCL was

                     23.58 months - 23.55 months = 0.003 months/ppm
                          10 ppm-0 ppm
 If a linear interpolation between 0 ppm and 10 ppm is applied, this rate of decrease
 implies that a female rat's timespan without dying with MCL during a 24 month
 exposure decreases approximately 0.003 months (approximately 0.09 days, 2.2
 hours, or 0.0125% of a 24 month lifetime) with each 1 ppm increase in exposure
 from 0 ppm to 10 ppm. The linearity assumption is almost certain to overstate the
 actual risks. If the mean response free period decreased linearly with the dose level
 at a rate of 0.003 months/ppm, then  dose levels of 2.2 ppm, 0.31 ppm, and 0.013
 ppm would correspond to average reductions of approximately 0.00391%,
 0.00016%, 0.0000027%, of a 24 month expected lifetime. These percentages
 correspond to approximately one week, one day, and one hour in a 70 year
 expected lifetime.
   The observed mean response free periods and estimates of dose levels
 corresponding to specified reductions in the mean response free period described in
 the two preceding paragraphs do not take advantage of the experimental
 information on the shape of the dose-response relationship obtained by combining
the animal information from several different dose levels together. Dose-response
 modeling allows the information from different experimental dose levels to be
combined. In addition, instead of arbitrarily assuming how risk characteristics
should be interpolated between dose levels, dose-response modeling allows the

-------
                                                                       21
experimental data themselves to imply how interpolations or extrapolations over
doses and over times are to be done.
   The family of multistage-Weibull time-to-response models can be used to
reasonably model the dose-response relationships for ethylene oxide inhalation in
rats. These models are generalizations of the quantal multistage model typically
used by the EPA. The multistage-Weibull models can be used to extrapolate over
time as well as over dose. The multistage-Weibull models fit to the available dose-
response information has the probability of a specified response occurring by time t
at dose d in the absence of competing risks equal to
             P(t;d) = 1 -exp{-[oo +  aid + a2d2 + a3d3] x [t-pilp2>
where ao. ai, 02, as, pi. and 02 are the unknown parameters whose values are to be
estimated from the experimental data (12).
   In such analyses, the dose scale used for modeling the dose-response relationship
should be the same as the dose scale that is going to be used for species
extrapolation. Table 8 indicates the dose levels to which the rats in the BRRC study
were exposed on the three dose scales corresponding to EO concentration in air
(ppm), intake relative to body weight (mg/kg/day), and intake relative to surface
area (mg/kg2/3/day).
   The rats in the BRRC study were exposed for 6 hours/day, 5/days/week for 2 years
which is equivalent to approximately 18% of a complete 2-year lifetime.  Hence, the
 lifetime average exposure concentrations are 18% of the experimental
 concentrations (0,10,33, and 100 ppm). Tyler and McKelvey (50) reported that the
 average cumulative daily intake for male Fischer 344 rats inhaling ethylene oxide
 vapor for 6 hours/day was
                                            *   "  • ' -
         (a)  2.7 mg/kg/day when the concentration of ethylene oxide was 10 ppm,
             and

-------
                                                                       22
        (b)  20.24 mg/kg/day when the concentration of ethylene oxide was 100
            ppm.
Since the BRRC rats were exposed 5 days/week, the corresponding lifetime average
                                                *'
exposures were 5/7 times these numbers; that is 1.93 mg/kg/day at 10 ppm and
14.46 mg/kg/day at 100 ppm. Linear interpolation between these two numbers
gives 5.13 mg/kg/day for 33 ppm. These numbers were also assumed to apply to
female rats. The exposures to male and female rats can be converted from a body
weight scale (mg/kg/day) to a body surface area scale (mg/kg2/3/day) by multiplying
the mg/kg/day exposures by (0.42)1/3 and (0.22)1/3 for males and females
respectively where 0.42 kg and 0.22 kg were the average weights of male and
female rats respectively in the BRRC study (42,43).
   The estimated dose-response relationships for female raits dying with MCL were
affected by the dose scale used during the model fitting. Such affects were due
solely to the differences in dose scales and occur even without extrapolating across
species. These effects are illustrated in Table 9 using the experimental data for the
controls, the 10 ppm group, and the 33 ppm group. As shown, for a female rat
                               f
inhaling EO at 1 ppm for 6 hours/day, 5 days/week for a lifetime, the fitted model
values for the added probability of dying with MCL were 0.0056,0.0028, and 0.0028
when the modeling dose scales were lifetime average concentration (ppm), intake
relative to body weight (mg/kg/day), and intake relative to surface area
(mg/kg2/3/day). The corresponding decreases in the mean response free period
during the 24 month experiment were 0.0065,0.0033, and 0.0033 respectively.
Thus, for this data set the choice of the dose scale used during the model fitting
made about a two-fold difference in the estimated risks for female rats
extrapolated from the experimental levels of 10 ppm and 33 ppm to 1 ppm.

-------
                                                                        23
   The reason for these differences in estimated risk is that, when the dose scale is
changed from ppm to mg/kg/day or mg/kg2/3/day, the relative distance between the
doses corresponding to 0,10, and 33 ppm changes.  On the lifetime average ppm
scale the value corresponding to 33 ppm is 3.3 times greater than that for 10 ppm
                          JP
whereas on either the mg/kg/day or the mg/kg2/3/day scales the value
corresponding to 33 ppm is only about 2.7 times greater than that for 10 ppm. Thus,
the increases in the proportion dying with MCL observed between 0 ppm and 33
ppm are shallower (more linear and more rapidly increasing) on the ppm scale than
the other scales, and consequently the low dose risks are estimated to be greater
when the modeling is done on the ppm scale. There is no difference between the
estimates for female rats obtained using the mg/kg/day and the mg/kg2/3/day scales
since these two scales are linearly related - here one scale is a simple constant
multiple of the other.

The Best Available Risk Characterization
   The multistage-Weibull time-to-response model, when applied to the primary
                                           *
data on female rats dying with MCL, indicated that risks do not increase as the dose
level increases. That is, based on the primary data set, the fitted time-to-response
model implied that, as the dose increases between 0 ppm and 100 ppm, the
probability of a female rat dying with MCL does not increase and the mean response
free period does not decrease! These implications based on the time-to-response
information are consistent with the implication of no increases in risk with increases
in dose based on the quanta! response data for female rats which died.
   The best available risk characterization of EO based on the female rats dying
with MCL is that there is no increased  risk at low levels of exposure, particularly
below 10 ppm. The remainder of this section considers what the risk characteristics

-------
                 B.3
DOSE-RESPONSE
      EVALUATION
       Attribute 2  The presentation of dose-response evaluation includes both an
                     , upper and lower bound of potency estimates and, wherever
                      possible,- some measure of the central tendency.
           SOURCE  Case Study D. Formaldehyde (Pages 1-29 to 1-31).
                Note  This excerpt shows a tabulation of upper bound estimates of risk for
                      several exposure scenarios and MLE estimates as a central tendency.
                      A lower bound of zero is stated. Uncertainty is also discussed.
           SOURCE  Case Study A. DEHP (Pages 138439),
                Note  See Dose-Response Attribute 1 in this Appendix. This report displays
                      upper and lower estimates for a "virtually, safe dose" or ADI derived
                      from several models and data sets. Uncertainty- is discussed, but no
                      central tendency is indicated.

-------
         Assessment, of  Health  Risks

to Garment Workers and Certain Home Residents

        from Exposure to Formaldehyde
                  April 1987
   Office of Pesticides  and Toxic Substances
     U.S. Environmental Protection Agency

-------
 1.4,3.  Numerical  Risk Estimates

     The  risk  estimates for the linearized multistage procedure,

 upper bound  (UB) and  maximum likelihood estimates (MLE)7 at

 various exposure levels are presented  in Table 1-2.   Risks  at  any

 exposure  level range  from  the upper  bound to zero.   An

 established  procedure does  not yet exist for making  "most likely"

 or  "best",estimates of risk within the 'range of uncertainty

 defined by the upper  bound  and zero.   The upper bound estimate

 for excess lifetime risk of developing cancer is 3 x 10~"4

 [Group Bl]a  for apparel workers  exposed to formaldehyde  at  the

 0.17 ppm  level, 2  x 10"4 [Group  Bl]  for residents of mobile homes

 who are exposed for 10 years  to  an average level of  0.10 ppm; and

 1 x 10"   [Group Bl] for residents of some conventional homes who

 are exposed  for 10 years to an average  level  of 0.07-ppm.   The

 upper bound  unit risk estimate for an  ambient exposure of 1 ug/m3
  The shapes of most models' upper bound estimates .tend to
parallel the shapes of the models themselves, unless a procedure
has been devised to provide otherwise.  This is the case  for the
linearized multistage procedure, which provides a  linear  upper
bound estimate at low dose.  The maximum likelihood estimate
(MLE), which is the estimate given by a fitted model, takes only
the experiment to which the model has been fitted  into account.
The upper bound estimate, on the other hand, is intended  to
account for experiment to -experiment variability as well  as
extrapolation uncertainties. •
g                                              •
  EPA's Guidelines for Carcinogen Risk Assessment recommend
categorizing chemicals in Group B (Probable Human Carcinogen)
when "the evidence of human carcinogenicity from epidemiologic
studies ranges from almost 'sufficient'  to  'inadequate.'  To
reflect this range, the category is divided into higher and lower
degrees of evidence. •  Usually, category Bl  is reserved  for agents
for which there is at least limited evidence of carcinogenicity
to humans from epidemiologic studies."
                               1-29

-------
(0.00082 ppra) for 70 years -is 1.3 x 10"5 [Group Bl].  The fitted



model gives the maximum likelihood estimate curve and, specific

                                           **-
to the CUT study/ it has a pronounced S-shape.  By contrast/ as



the linearized multistage procedure's upper bound estimate is



traced toward lower doses, its linear nature accomodates



increasing variability and extrapolation uncertainty.  Both



estimates are shown in Table 1-2 to illustrate how the



perspectives they give on risk differ.  Thus at 3 ppm (which is



in the experimental range), the difference between the MLE and



the UB is ten-fold, whereas at about one-tenth of that exposure,



a 100,000 fold difference is generated.


     The lower bound on risk is always recognized to be as low as



zero.  The upper bound estimate is ordinarily shown to allow for



extrapolation uncertainty.  It is for this reason, along with



adherence to EPA's Guidelines for Carcinogen Risk Assessment,



that the upper bound was selected,to represent potential human



risk.  While some of the existing information on formaldehyde is
                      3

consistent with non-linear interpretations, some support for a
             *


linearized upper bound comes from the epidemiologic studies.  The



excess cancer incidences observed in the epidemiologic studies



are about the same as the upper bound on lifetime risk based on



the rat nasal carcinoma data.
                                _ in

-------
                               TABLE 1-2


    SUMMARY OP CANCER  RISKS  ASSOCIATED WITH FORMALDEHYDE EXPOSURE
    Population Segement
     {Exposure Level)
         Lifetime
      Individual Risk
  Curre.nt OS'HA std. (3 ppm)
    UBb   6 x 10~;j [B1J
    MLEC  6 x 10"4 [Bl]
  Garment Workers

    NIOSH  •
   (0.17 ppm)
    UB  3 x 10~4 [Bl]
    MLE 4 x 10
              -9
[Bl]
       Mobile Home
       Residents
  (0.10 ppm 10-yr average)
    UB-   2  x 10~/J   [Bi:
    MLE 2 x 10"10 (Bl]
  Conventional Home*
     Residents
  (0.07 10-year average)
    UB  1 X 10"4 [Bl]
    MLE 6 X IQ"IL [Bl]
  Home/Environment
    Background Upper Limit
    (0.0 5 ppm)

       10.yr.
       70 yr,
    UB  7.0 x 10"5 [Bl
    MLE 1.0 X .10"11 [B
                                                    [Bl]
                                               ,-4
    UB  5.0 x 1.0"?n[Bl] •
    MLE 1.0 x 10"iu [Bl]
*  For hom«s containing substantial  amounts  of  urea-formaldehyde
   pressed wood (e.g./ floor  underlayment and/or paneling)
   upp«r Bound

c  Maximum Likelihood Estimate

d 'Airborne Unit Risk, 1 ug/m3
   UB - 1.3 x 10"5  [Bl].
- 70 yrs; Lifetime individual risk,
                              1-31

-------

-------
                 B.3

DOSE-RESPONSE
      EVALUATION
      Attributed
                      The report offers an explicit rationale for any preferred data
                      set(s) and model(s) used in dose-response evaluation;
                      strengths and weaknesses of the preferred data sets and
                      models are discussed, and scientific consensus or lack thereof
                      is indicated for critical issues or assumptions.
          SOURCE  Case Study H. Methylene Chloride (Pages 71-87).
               Note  See General Attribute 2 in this Appendix.
          SOURCE  Case Study A. DEHP (Pages 111 - 145) .
               Note  See Dose-Response Attribute 1 in this Appendix.
          SOURCE  Case Study D. Formaldehyde (Pages 1-23 to 1-28).
               Note  See Dose-Response Attribute 1 in this Appendix. This excerpt presents
                     the rationale for selection of data sets and models. Scientific consen-
                     sus that supports use of generic assumptions is discussed.
          SOURCE  Case-Study J. Red Dye No. 3 (Page? 73-86).

-------
           A  REPORT  BY  THE  FD&C  RED  NO.  3  PEER REVIEW PANEL
AN INQUIRY INTO THE MECHANISM OF CARCINOGENIC ACTION OF  FD&C RED NO. 3
               AND ITS SIGNIFICANCE FOR RISK ASSESSMENT
                                    Prepared by:

                                    Dr. Ronald W. Hart, NCTR/FDA (Chairman)
                                    Dr. Thomas Burka, NIEHS/NIH
                                    Dr. Stan C. Freni, CEH/CDC
                                    Dr. Robert Furrow, CVM/FDA
                                    Dr. David W. Gaylor, NCTR/FDA
                                    Dr. Theodore Meinhardt, NIOSH/CDC
                                    •Dr. Bernard Sass, NCI/NIH
                                    Dr. Elizabeth K. Weisburger, NCI/NIH
                                     Executive Secretaries -       ,
                                      Dr. Paul Lepore, ORA/FDA
                                      Dr. Angelo Turturro, NGTR/FDA
                                July, 1987

-------
                                                                   July,  1987
              CHAPTER 7 - ISSUES IN QUANTITATIVE RISK  ESTIMATION

 A.   Introduction
     Since the  thyroid  tumors in rats  produced  by R-3  appear  likely to  be
 the result of a secondary mechanism, the question of  the appropriate method
 of  low-dose  extrapolation  arises.    Because thyroid  tumors occur  in  the
 control animals,  it  can not  be ruled  out  that  R-3  may be  affecting a
 tumorigenic  process already  operating  in  control  animals in the absence  of
 R-3.   It  has  been shown by  several  authors  (147-149) that if a background
 tumor  risk exists which is not  totally independent  of the mechanism of  the
 production of tumors by administration of an agent, regardless of the mech-
 anism,  that linearity  is  the appropriate model at  low dose  levels  of  the
 actual  oncogen.  If an agent produces tumors by the same biological mechan-
 ism that  produces  some spontaneous  tumors  in  the  controls,  then those
      a
 effects  are  additive.   If tumors are produced in the control animals due  to
 the presence of  an endogenous  or  environmental  agent,  then the effective
 dose  in  the  control   has  already  surpassed  the  threshold  dose,   if   one
 exists.   If  the addition of a small  amount of  an  agent increases the level
 of  the  active  oncogen,  either directly or indirectly,  this will result in a
 proportionately small  increase in tumors.  From a  standpoint of not under-
 estimating the risk, these  arguments support the use  of  linear extrapola-
 tion in  the  low-dose region.
    Since  thyroid tumors in  rats  produced by R-3  appear likely  to  be  the
 result  of a secondary  mechanism,  arguments could  be  presented for  the
 existence  of a  threshold dose  of R-3  below which  no increase  in  thyroid
 tumors occur.   For example,  a threshold dose  would  occur if there is  a dose
 of  R-3 for  which no  increase of  the active  oncogen  at  the  target  site
 occurs in  any  individual.
    Two approaches to  risk assessment are considered  in- this chapter.   One
 approach  is  based upon a no-observed-effect-level  (NOEL) of R-3  in  humans
which  does not appear  to  produce  an increase in TSH.   Safety  factors  are
applied  to this NOEL  to arrive at an  acceptable  daily intake  (ADI).   The
second  approach is  based  upon estimates  not likely  to  underestimate  risk
using linear  low dose  extrapolation of thyrdid tumor  incidence  from  animal
bioassay data.   The  animal"thyroid tumor  data are  examined below.
                                      73

-------
                                                                  July,  1987

B.  Animal Tumor Data
    Detailed statistical analyses  of  age-adjusted thyroid tumor  rates  were
performed by the Panel for the two IRDC chronic bioassays  (72,74) conducted
in rats, described in Chapter  4.   Age-adjusted analyses (150) were  used  to
correct tumor rates for differences in mortality  across  dose  groups.  Anal-
yses were conducted  for follicular cell and  C-cell 'adenomas  and carcinomas
of the thyroid.  It was assumed that  these  tumors did  not  contribute to the
death  of  any animal.   A  summary of  the  number of  animals with  thyroid
tumors is  given in Tables  13  and 14  for  males and females,  respectively,
derived from the detailed reports by  the various  pathologists.   Differences
were considered significant if they existed at  the P<0.05  level.
    1.  Follicular Cell Tumors      .
    Statistical tests for dose-response trends  on age-adjusted tumor preva-
lence  rates  were  conducted according  to  procedures  given by Kodell et_ al.
(151)  based  upon  Peto et al.  (151).   Separate  analyses were conducted <• on
each study.  Where one group provided  diagnoses  for  both studies, the trend
test statistics were  combined  to  give an over-all significance  level.   The
results of the  trend  tests are given  in Table 15.
    The numbers of animals developing  follicular adenomas in females  were
lower  in  the 4.0% dose  group  in Study 410-011  (74)  than in  the 1.0%  dose
group  from Study 410-002  (72).   Since these  tumors tend  to develop  late in
life,  the  lower rates in Study 410-011 may be  due,  in part,  to  the  shorter
length of 122 weeks compared to the  length of 128 weeks for Study 4LO-002.
    It is  noted from Table'15 for males  that there is a  highly statisti-
cally  significant  increase  in  follicular  adenomas at the 4.0% dose level in
Study  410-011.   For  the  diagnoses of follicular  adenoma  in  males by IRDC,
there  is a marginally significant dose-response trend in Study 410-002.
    Based  on the  FDA diagnoses, there is  a marginally statistically signi-
ficant (P<0.048)  increase in follicular carcinomas in male rats.
    For  adenomas  and carcinomas  combined in  male  rats,  the  results are
dominated  by. the  larger prevalence of adenomas.   For all pathology groups,
a highly statistically significant increase  in  follicular tumors was found
in male  rats at the 4.0% dose  level in Study 410-011.
    For  female  rats,  there were not enough follicular carcinomas to perform
statistical  tests *for  dose-response  trends.  For  follicular  adenomas or
follicular  tumors  in  female  rats,  there  were  statistically   significant
                                      74

-------
                                                                            July, 1987
                                      TABLE 13
                     Numbers  of  male  rats  with thyroid tumors
           Cla       C2     0.1%      0.5%      1.0%     ,     C3B              4.0%




           °F     I   F    I    F     IF     IF     IRFG    I   R   F    G
Number
Diagnosed •
Follicular
Adenoma
Follicular
Carcinoma
C-Cell
Adenoma
C-Cell
Carcinoma
69 64 69 61-64 - 66 69 57

00 1. 0-3 -8 31

00 01-3 -0 03

05 07-7 -2 03

00 00-0 -0 00
-•
70'

0
. -
1

3

1

67 68

1 1

2 1

3 4

1. 1

69 70 68 69

0 16d 15d 14d

0 3d 3d 5d

88 6

244

69

8

2





Cl - 0, C2 - 0, 0.1, 0.5, and 1.0% dose levels in Study 410-002  (72).



C3 - 0 and 4.0% dose levels in Study 410-011 (74).




Group: I- IRDC (81), F- FDA (82), R- CCMA consultant (83), G- A  Canadian group  (84)



One animal had both a follicular adenoma and carcinoma.
                                        75

-------
                                                              July, 1987
                               TABLE 14
              Numbers of female rats with thyroid tumors
cia
Ifc Rf
Number 70 69
Diagnosed
Follicular 0 1
Adenoma
Follicular 1 0
Carcinoma
C-Cell 0 2
Adenoma
C-Cell 0 1
Carcinoma
C2 0.1% 0.5% 1.0% C3b 4.0%
If Rf ' If Rf • If Rf If Rf If Rf If Rf
'70 68 - 67 - 69 68 68 67 66 70 65

00 -1 -3 65 0022

00 -0 -0 01 00 10
-
03 -10-7 01 24 35

01 -1 -3 0102 01

Cl - 0, C2 - 0, 0.1, 0.5, and 1.0% dose levels in Study 410-002 (72).




C3 - 0 and 4.0% dose levels in Study 410-011 (74).




If- IRDC (86) and R- consultant for CCMA (87).
                                  76

-------
                              TABLE  15






   Significance Levels  (P-values) of Dose-Response  Trend Tests




          of Age-adjusted Prevalence of Follicular  Tumors
Sex Type
Male Follicular
Adenoma
™

Follicular
Carcinoma


Combined



Female Follicular
Adenoma
Combined

Group
I
R
F
6
I
R
F
G
I
R
F
G
I
R
I.
R
410-002
0.043
0.108
—
a
-
0.154
—
0.043
—
0.054
^
0.0002
0.0022
0.0017
0.0006
410-011
0.00005
0 .00040
0.00100
0.00390
0.188
0.380
0.068
0.094
0.00005
0.00089
0.00029
0.00118
0 . 120
0.109
0.085
0.109
Combined
0.00002
0.00054
. -


0.048
-
0.00002
_
0.00011
-
0.0075
0.012
0.011
0.007
No carcinoma reported.
                              77

-------
                                                                 July, 1987

dose-response trends in Study 410-002. '  Failure to achieve significance in
Study 410-011 may  be  due, in part,  to  the shorter length  of  122  weeks of
that study compared to 128 weeks for Study 410-002.
    2.  C-Cell Tumors
    There was  a marginally  statistically  significant increase  in  C-cell
tumors (adenomas and/or  carcinoma)  in  male rats  administered  4.0% R-3 in
Study  410-011.    The  statistical  significance  levels  for   age   adjusted
prevalence of C-cell  tumors  in male rats  were 0.056* 0.025,  and 0.120  for
diagnoses provided by groups I,R, and F, respectively.
    3.
    There was  a  highly statistically  significant  increase  in  follicular
adenomas in male rats administered 4.0% R-3.  Female rats  showed  a  statist-
ically  significant  increase, in  follicular  adenomas in  Study 410-002,  but
not in the shorter Study 410-011.

C.  Dose-Response Models
    Let P » g(d)  represent  the relationship between the proportion of ani-
mals above background with  tumors,  P, and the daily dose,  d,  of  the active
oncogen.   Let d  -  f (D) represent  the relationship  between  the  effective
dose and  the  daily dose of  the administered agent, D.-  This  relationship,
f(D), represents  the  combined effects of absorption,  distribution, physio-
logic,  activation, and  detoxification processes  which  produce the effective.
oncogenic dose.   Then,  P - g[f(D)] - h(D).   That is, the  chronic  bioassay
data provide  a measure  of  the  relationship  between risk  and administered
dose without  knowledge of the  relationship  between the effective  dose and
the  administered  dose.   If  the dose  response, P  -  h(D),  is  sublinear
(curving upward)  in the low  dose region,  then low-do"se linear extrapolation
will overestimate risk;,  see,  e.g..  Gay lor  and Kodell (152).   If an oncogen
at  least partially  augments an oncogenic  process  already in  progress in
control animals,  then low-dose linearity is  expected.   Since this possibil-
ity cannot be  ruled  out, an approach which  overestimates  risk is to  assume
low dose linearity with the  effective dose,  P - bd> b*f(D), where b  is the
slope.  Whittemore jet al.  (153) proposed a similar approach for incorporat-
ing  pharmacokinetic data  into  the risk  assessment process  to  utilize the
effective  dose.   Whether  or not this model  can be extrapolated below the
experimental  dose range  for estimating ' risk depends upon  the  validity of
                                      78

-------
                                                                  July, 1987
low-dose  linearity and the accuracy  of the form of  the model  used for d -
f(D).  If  the  true model for effective dose is  sublinear  (curving upward),
low-dose  linear  extrapolation  overestimates   risk.    Even  if tumors ^are
produced  by  a  secondary mechanism, this  does  not necessarily  imply that a
threshold  dose  for  tumors exists.   If  the  effective  dose function, d »
f(D),  is  a threshold  function  of the  administered  dose,   then  doses  below
this dose  threshold would  not  result  in an increase  in tumors .
D.  No-Observed-Effect-Level  (NOEL)  and Acceptable Daily Intake (ADI)
    Because  the  Panel has concluded  that  the " mechanism in  R-3 oncogenesis
is probably  indirect,  an important  question  is:   what  is  the mathematical
relationship  between the administered  dose of  R-3 and  the  active oncogen?
If a particular  secondary mechanism  is  assumed  (e.g., "TSH elaboration) ,  the
effective agent  becomes TSH, not  R-3.   The background  tumor incidence  in-
dicates  that it  is  possible that normal TSH  levels in  some  animals  are
above  any tumor  threshold,   if  one  exists.   For  TSH  the  relationship  of
effective dose to administered  dose  of  R-3, d » f(D),  cannot be established
from the  available  animal data.  Thus,  an estimate of  the tumor risk based
on  the dose-response  in animals  for  this  secondary  mechanism  cannot  be
computed directly by P - g(d).
    Without  this  information, one is  forced  to use  assumptions.   One  may
assume a  threshold  function  for the dose-response of R-3 and TSH based  on
the premise  that these reactions  have   thresholds.   Thus,  one  could esta-
blish  what  appears  to be  an  experimental  dose  level  with  little  or  no
biological effect, and  divide this level by a  safety factor to account  fpr
the uncertainty in the data in  order to arrive  at  an ADI
    It appears,  from human studies  (Chapter  6) that  an exposure  of 20-60
mg/d of R-3  may  be  a NOEL for humans.   Using a traditional  adult male body
weight of 70 kg, an  exposure  level of 20 mg/d  is  equivalent to an exposure
of 20/70 - 0.29 mg/kg-d or 290  Ug/kg-body weight-day.   Since a NOEL depends
on the -number of  subjects examined,  the biological  endpoints measured,  and
the variability  in  the measurements, every study  is limited in its ability
to detect small effects.  Therefore, a NOEL is not  necessarily an exposure
level  without  adverse  biological  effects,  particularly  when  short-term
tests are used to model lifetime chronic exposure.
    In using  a  NOEL to  establish  an ADI,  a  key  question is, what  are  the
                                      79

-------
                                                                  July, 1987

 appropriate safety factors to use?   How some safety factors can be used is
 illustrated below.  For instance, since  only 10 males in  the  study of in-
 terest  were  used,  employing  a  safety  factor  of  10  would  be  standard
 practice  to account for  variable sensitivities to  chemicals  among indivi-
 duals.   Since only  a  14-day exposure was used  to  model  a lifetime chronic
 exposure,  an additional safety  factor  would likely be used.   For example,
 if  10  is  used for the additional  safety factor,  the ADI would be the NOEL/
 100 - 290/100 -  2.9  Ug/kg-d.   Also,  if  60 mg/d is used as the NOEL, the ADI
 would  be  8.6  Hg/kg-d.   Alternatively,  one might also  use a higher  dose
 level,  e.g.,  200 mg/d, and use  an additional safety factor  to account for
 the  apparent effect at  this  level.   Other safety  factors could  be  used
 resulting  in different ADI's.   The choice of safety  factors is a risk man-
 agement  function  arid  they  are  only  employed  here  to  illustrate  the
 procedure.

E.  R-3 Risk Estimates
    Since  the data are insufficent to establish a threshold function of R-3
 and  TSH,   either on experimental or mechanistic grounds,  there may  be  a
 tumor risk at the ADI.   Hence,  it  is  necessary  to estimate  the potential
 tumor risk at  low doses.   If  one does  not make the  assumption of a thres-
hold, the  estimate of  the risk using the multistage model can be  calculated
 using  GLOBAL82   (154)  using the 'data from  the  two IRDC  chronic bioassays
 discussed  in Section B for rats.  The procedure of Kodell jsit  al. (155) was
used to calculate  lifetime tumor incidence for  each dose  group standardized
 to  the  mortality  experience of  the control animals.   These  standardized
lifetime tumor incidence  rates were  then used in  the multistage model.  The
point estimates  of risk are dependent upon  the  choice of the model.   How-
 ever, the  estimates  of upper limits on  low-dose risks  obtained from linear
low-dose  extrapolation are rather insensitive to  the  choice of  the  model
used in the experimental  dose  range.  Choices of  other models in  the exper-
imental data  range used  in  conjunction with linear  low-dose  extrapolation
would have little influence on  the  upper limits of  risk.   In the  case  of
R-3, the  upper  confidence limits are  not  far above  the  point  estimates
because of  the linear  component  exhibited in the'chronic  bioassay data for
thyroid follicular cell  tumors  in  rats.    The  nearly linear  relationship
between TSH and  R-3  reported in humans  (134), although in a limited study,
                                      80

-------
                                                                  July, 198,

tends  to  support use of  a  low-dose linear extrapolation to estimate upper
limits on  tumor  risk at low doses.  This  does not necessarily  mean that a
straight line is fit  to  the  data.   Rather, the curvilinear multistage model
is fit to  the data  and  linear extrapolation is  used  below the experimental
data range to estimate upper limits on low-dose risk.
    Low-dose risk estimates  were based  on the results of  the FDA patholo-
gists  for  male  rats and  by  the  results  of  the  CCMA consultant  for  the
female rats  because these were  the most  complete pathology  data  sets  for
each study.   In Chapter  6,  the  average lifetime  exposure  to  R-3 was calc-
ulated to  be 1.41  mg/day,  based  upon  production for the past  10  .years.
Using the average body weight  of  the U.S.  population  for  ages 1-74 of 61 kg
(145), the average  exposure  to R-3  is  23 ng/kg-day.  Point estimates of the
risk for lifetime exposure to  23  ug/kg-day of  R-3  based upon the multistage
model are shown in  Tables 16 and  17 for male and female rats,  respectively,
as are the point estimates of  the exposure corresponding  to a lifetime risk
of one in a million.  Upper  limits  of  risk and lower  limits on the exposure
corresponding to a  risk  of  one in a million based upon the linearized mul-
tistage formula are also  given.
    If, as is suggested  in the exposure information,  the  exposure to R-3 is
higher at  young  ages, and if  R-3 effects the early  stages in carcinogene-
sis, the multistage model would  predict  that  the risk estimates  based  on
extrapolation of an average  adult dose will be too low.
    A  plot of  the  standardized  lifetime thyroid  follicular  cell  tumor
(adenoma and/or  carcinoma)  incidence  for male and  females  are given  in
Figures  3  and  4,   respectively.    The  fitted  multistage  models are  also
plotted.    Considerable  variability  of  the data from the  fitted multistage
model is noted.  If the  high dose level (4%)   is omitted,  the  slopes become
somewhat steeper  for males  and  considerably  steeper for  females.   Hence,
discarding the 4.0% dose  group tends to result in higher  estimates of risk
(Tables 16 and  17).  It  appears  that  the absorption of R-3 is  similar for
rats and humans.    Further,  it is  assumed in. the  -absence  of  data that the
lifetime incidence  of tumors  in  humans is equivalent  to  the incidence  of
tumors in  rats  for  equal daily doses  on a body  weight  basis.   Even if  the
same mechanism produces  tumors in rats  and humans, exposure to  equal doses
of  R-3, could have  different  quantitative tumor  rate  effects  because  the
dose-response slopes  may be different  at the different   spontaneous  tumor
                                      81

-------
                                                          .       . July, 1987

 rates  for rats  and  humans..  Generally,  it is expected  that  the  slope and
 hence  the additional risk would be lower for exposure above the human back-
 ground  level  than  at the  higher background  level in  rats.    If  the dose
 conversion between  rats  and  humans is  made on a  surface area basis,  the
 estimated risks would be  approximately  a factor of  five higher.   Alterna-
 tively,  if R-3 produces tumors by a different  mechanism than spontaneously.
 occurring tumors and  if  a threshold  of  R-3 exists for  each  individual  in
 the population which is above  his/her exposure level (i.e.,  no active on-
 cogen is  produced by R-3  for  any  individual),  the  tumor  risk would be zero.
F.  Discussion
    The ADI and  risk estimation are two different approaches used by regul-
atory  agencies  in  the  management  of  health hazards.   For R-3  the  ADI is
baaed on no observed short-term changes in TSH in adult males  and risk es-
timation  is based  on  tumor  incidence  in rats.   Two  different kinds  of
safety  considerations  are  utilized  in the  two  approaches.    The   ADI  is
derived  from  applying  safety factors  to  a dose level  considered to  be  a
NOEL.  An  upper  limit on the risk  estimate is derived  from a  mathematical
extrapolation model, which  is  believed  to overestimate  the risk.   Since
there may not be a  threshold  dose of the oncogen or  since  the  ADI might be
                                  c
above a  threshold  dose, if one exists, there is a potential for tumors at
the ADI.  Hence, it  is  useful to estimate the magnitude of the risk at the
ADI.  Applying the multistage model,  and using the  information  in Tables 16
or 17, the risk at  an ADI of  0.0029 mg/kg-d is 0.0029/0.007  x  lo"6  - 0.4  x
10" .   Similarly,   the  risk  at an  ADI  of  0.0089  mg/kg-d is  0.0086/0.007  x
10~6 - 1.2 x 10"6.
    The multistage  model may be  used to  estimate  the potential  risk from
oral  exposure  to R-3 through food  (or  food and drugs). " This  requires  a
choice be made from the options  for  selecting  the  tumor data  base  for the
extrapolation model.  The options have  been listed  and discussed in Chapter
6.1.  The slope  of  the  extrapolation model, and thus  the  point estimate of
the risk, depends heavily on  the  chosen data base.  With the information in
Tables 16 and 17, risks can  be calculated for different estimates of daily
dose levels by direct proportionality.   For example,  reduction  of the daily
exposure  to 1.2 mg/d  -  0.020 mg/kg-d (based  upon  the.  annual  certified
poundage of R— 3  averaged over  the most  recent, five  years  rather than the
                                     82

-------
                                                                  July, 1987

past  ten  years) would  result  in a risk of  20/23 times the  risk  listed  in
Tables 16 and 17.
     For  comparison,  a recent  effort by  a  joint FAO/WHO  expert  committee
defined a temporary ADI for R-3  at  0-0.6 mg/kg body  weight, or 0-600 ug/kg-
d  (155).   If finalized,  this  would be approximately  one  to  two  orders  of
magnitude greater  than  the possible ADI's  used for illustration here. .How-
ever,  it  appeared  from the discussion  in  the summary  document  that  the
expert committee  did  not have the  advantage of some  of the  studies in rat
and man that the Panel  had.  WHO considered  its estimate only tentative.
    Low—dose estimates  of oncogenic  risk  could  be  improved  if the active
oncogenic dose  were estimated  as a  function of  the  administered  dose  of
R-3, i.e.,  establish  d - f(D).   Or,  if  it were possible to  at least iden-
tify an intermediate  agent whose dose level  is  proportional  to  the tumor
incidence,  then the dose  level of that intermediate agent  could be  used  as
a  surrogate for the oncogenic  dose." Under  these circumstances,  it would
only  be  necessary to  establish a relationship  between   the  dose   of  the
                    ."
intermediate agent and  the dose  of  R-3 administered.
                                     83

-------
                                                                 July,  1987
                                .  TABLE 16
Estimates of Lifetime Risk of Thyroid Follicular Cell Tumors  in  Male  Rats
         Based on Results of FDA Pathology Analysis  (See Table 13)
Tumor Type    Data
              Sets*
          Estimates of Risk at
               23 ug/kg-d
           Point        Upper
                             Estimates of Dose for
                                  Risk of 10~6
                               Point         Lower
Adenoma
All
2.5 X 10~6   4.0 X 10~6
9.1 ug/kg-d   5.8 ug/kg-d
 alone        -4%     5.4 X 10~6   10.4 X  10~6
                                      4.2 ug/kg-d   2.2 ug/kg-d
Carcinoma     All
         0.7 X 10~6   1.6 X 10~6     31.8 ug/kg-d  14.5 ug/kg-d
 alone
           ca. 0
             4.3 X 10
                                            -6
              5.4 ug/kg-d
Combined      All       3.2 X  10~6    4.8 X  10~6       7.2  ug/kg-d   4.8  ug/kg-d

             - 4Z       6.4 X  10~6   12.3 X  10~6       3.6  ug/kg-d   1.9  ug/kg-d
  - 4% means  that  the  data used exclude the 4% group.
x - The  point estimate gives  practically no risk,  therefore,  the dose for a
    risk of 10~  is  high,  approximately 14.0 mg/kg-d.
                                      84

-------
                                                                 July,  1987
                                  TABLE 17
Estimates of Lifetime Risk of Thyroid Follicular Cell Tumors  In  Female Rats
       Based on Results of CCMA Consultant's Analysis (See Table 14)
Tumor Type     Data     Estimates of Risk at
               Sets8         23 ug/kg-d
                         Point  .      Upper
                                     Estimates of Dose for
                                          Risk of l(f6
                                       Point          Lower
Adenoma
All     0.5 X 10~6   1.3 X 10~6
48.1 lig/kg-d  17.5 Hg/kg-d
 alone
 4%     3.1.X 10"6   5.5 X 10"6
 7.4 ug/kg-d   4.2 ng/kg-d
Combined
All     0.6 X 10~6   1.4 X 10"6     41.7 ng/kg-d   15.9 ug/kg-d
              -4%     3.6 X  10"6    6.1 X  10~6       6.5 ug/kg-d     3.8  Ug/kg-d
  - 4% means that  the data  used  exclude  the  4%  group.
                                      85

-------

-------
                   B.3

DOSE-RESPONSE
      EVALUATION
      At mDUte «l  T^e report reveals how dose-response relationships change
                       with alternate data sets, assumptions, and models.
           SOURCE   Case Study D. Formaldehyde (Pages 7-12 to 7-13).
                Note   Excerpts from the EPA report on formaldehyde illustrate an analysis
                       of benign tumor incidence and a resultant unit risk estimate. This
                       analysis was subsequently compared to the results from analysis of
                       malignant tumor incidence.
           SOURCE   Case Study F. Lead (Pages iii, iv, ix, 1, 49, 50).


                Note   This illustration is taken from a very comprehensive analysis of the
                       effects of blood lead levels on young children. The authors elicited
                       judgments from experts on the effects at various blood levels of lead as
                       a basis for their estimates. Sensitivity analysis was used to determine
                       the effects of two critical aspects of the study. The authors' presen-
                       tation is closely tied to the technical context of the discussion. The
                       illustration is included in the appendix to highlight how a sensitivity
                       analysis can be clearly presented. To assist the reader in understand-
                       ing the context of the sensitivity analysis, the authors' abstract, the
                       Table of Contents, and the list of acronyms preface the discussion of
                *      sensitivity analysis.

-------
         Assessment of Health  Risks

to Garment Workers and Certain Home Residents

        from Exposure to Formaldehyde
                  April 1987
   Office of Pesticides and Toxic Substances
      U.S.  Environmental  Protection  Agency

-------
 7.2.   Riate Batiaatea Baaed on Polypoid Adenoma Data
      There appears to be little credible evidence that polypoid
 adenomas progress to any of the malignant tumors seen in the Kern
 et al.  (L983) study.  However, while the adenomas should not be
 combined statistically with the squamous carcinomas for hazard
 identification purposes/ they represent an endpoint that can be
 quantified separately for analysis pruposes.
      Because it is beyond the capability of•the various
 extrapolation models to fit data with a negative slope, an
 alternative extrapolation procedure is to drop the two highest-
 doses and use the data from the 2.0 ppm rat exposure,grouo
 (straight line to zero).  However, since the true slope of the
 dose-response curve is unknown below 2.0 ppm, this approach may
 vastly overestimate the true risk if the curve is convex, and
 underestimate it if it- -is concave. .- The reason the occurrence of
. polypoid adenomas has a negative slope probably lies with the
 fact that the cell type in the respiratory epithelium from which '
 these tumors arise is lost sooner and to a greater extent with
 increasing dose due,, to squamous metaplasia.  The less, respiratory
 epithelium available the smaller the chance for adenomas to
 develop.  0-ther explanations are also possible as discussed in
 section 7.4.1.
      Risk estimates using polypoid adenomas appear, in Table
 7-3.  For polypoid adenoma as the endpoint instead of squaraous
 cell carcinoma there is no difference between the two procedures
 described earlier to adjus,t for animals at risk.  The first
                                7-12

-------
   observation of  a polypoid  adenoma ,was  in a rat  sacrificed  at 10

   months.   Eliminating all rats dead of  any cause prior to, that

   time  and applying the method used  for  the carcinoma data.leads to

   7/159 for the response at  2 ppm with  1/156 at control, the same

   as  if all rats  dead prior  to an including the 18 month sacrifice

   were  excluded.'
                                      Table 7-3.
                    RISK EXTIMATES USING POLYPOID ADENOMA DATA
  Category

tobile Hone
 Residents
 Based on HUD
  Target Level
     Dose
0.15
(112 hrs/wk
 for 10 yrs)
Manufacturers
of 
-------
                       ARGONNE NATIONAL LABORATORY
                   9700 South Cass Avenue, Argonne, Illinois 60439
                                   ANL/AA-32
                   ASSESSING THE RISKS TO YOUNG CHILDREN
                      OF THREE EFFECTS ASSOCIATED WITH
                         ELEVATED BLOOD-LEAD LEVELS
                                       by

                    Thomas S. Wallsten* and Ronald G. Whitfield

                     Energy and Environmental Systems Division
                  Decision Analysis and Systems Evaluation Section
                                 December 1986
                                work sponsored by

                  U.S. ENVIRONMENTAL PROTECTION AGENCY
                    Office of Air Quality Planning and Standards
*L.L. Thurstone Psychometric Laboratory, University of North Carolina, Chapel Hill

-------
                                  CONTENTS
ACRONYMS	   lx

SYMBOLS	    x

ACKNOWLEDGMENTS	   xi

ABSTRACT	•		•	   ' 1

1  INTRODUCTION	    l

   1.1 Report Organization	 ?	    2
   1.2 Motivation	    «
   1.3 Judgmental Probability Encoding	    5
   1.4 Dose-Response Uncertainty	•	    6
   1.5 Risk Assessment Strategy	    6

2  PROBABILISTIC DOSE-RESPONSE FUNCTIONS FOR LEAD-INDUCED
   ELEVATED EP LEVELS	    7

3  PROBABILISTIC DOSE-RESPONSE FUNCTIONS FOR LEAD-INDUCED
   Hb DECREMENTS	• •	   n

   3.1 Protocol Development	•  11
   3.2 Protocol Outline	•	  *2
   3.3 Conduct of the Sessions	  13
   3.4 Encoding the Judgments	-	  13
   3.5 Representing the Judgments	:	  ijj
   3.6 The Experts	• • •	  JJ
   3.7 Results	  "
       3.7.1   Hb Level < 11 g/dL, Ages 0-3	•	  18
       3.7.2   Hb Level < 11 g/dL, Ages 4-6	  20
       3.7.3   Hb Level < 9.5 g/dL, Ages 0-3	  21
       3.7.4   Hb Level < 9.5 g/dL, Ages 4-6	  22
   3.8 Discussion	  23

 4  PROBABILISTIC DOSE-EFFECT AND DOSE-RESPONSE FUNCTIONS FOR
   LEAD-INDUCED IQ DECREMENTS	•	• •	  27

   4.1  Protocol Development	•	• • •	  28
   4.2  Protocol Outline	  30
   4.3  Conduct of the Sessions	• • • •	•	  30
   4.4  Encoding the Judgments	   31
   4.5  Representing the Judgments	   32
   4.6  The Experts	•	   32
   4.7  Results	•	   33
        4.7.1  Control-Group Mean  IQ  .. *	•	• •   34
        4.7.2  Within-Group IQ Standard Deviation	•   35
        4.7.3  Mean IQ Decrements for the Low SES Group	   36
        4.7.4  Mean IQ Decrements for the High SES Group	   36

-------
                                CONTENTS (Cont'd)
         4.7.5  Change in Percentage of Low SES Group with IQ < 85  ..............   37
         4.7.6  Change in Percentage of High SES Group with IQ  < 85 ., ..... !!!!!.'!   40
    4.8  Discussion ............................... . .................     « • • • •

 5  ESTIMATED RISKS OF ADVERSE HEALTH EFFECTS VERSUS GEOMETRIC
    MEAN PbB LEVEL ...... ,. ...... . ......... ...............................   43

    5.1  Estimated PbB Distributions ____ .......................... .              43
    5.2  Overview of the Risk Results f or  EP ........ .'.'.".*!.'!.'.'."!!!.'!!""""""   44
    5.3  Overview of the Risk Results for  Hb ............ ...'.......... ..........   45
    5.4  Overview of the Risk Results for  IQ  . ........... •.' 1 !.'.*.'!!!.'!.'-!!.*!.'!!!!!!   45
         5,4.1   Risk Distributions over Mean IQ Decrement ..... . . ...... !!!!...!     47
         5.4.2   Increased Probability of Lead-Induced IQ Levels Seine ......
               ^IQ* ....... .................................. ;...: ....... ...  48
    5.5 Sensitivity Analysis ................................ 4 ......              4q

 6  CONCLUDING REMARKS ....... . ..... ... .............. . ____ .............   51

 REFERENCES .........................                                        co
                                     ****•******•"*••,*••••••»••••••••*••••.»•»   O /

 APPENDIX A:  Fitting Functions to Data on Lead-Induced Elevated EP
               Levels..... ............................................ ......  55

 APPENDIX B:  Fitting Functions to Encoded Judgments Relating to
               Lead-Induced Hb Decrement ............ ... ...... . .............  gj

 APPENDIX C:  Fitting Functions to Encoded Judgments Relating to
               Lead-Induced IQ Effects ...................................... 101

 APPENDIX D:  Risk Distributions ..... .......... ..... ...... . . .. ............... 141


                                     TABLES


 1   Probability of Suffering a Specified Health Effect under Alternative
    NAAQS, Given Complete Information ...................... ...............    4

 2   Probabilities  of Suffering a Specified Health Effect under Alternative
    NAAQS, Given Incomplete Information ......... , ....... . ..................    5

3   Sample Sizes  for the EP Data ........ . ..................  ..........           9

4   Consultants for the Hb  Protocol ........................ ..  ...........         12

5   Experts Participating in the Hb Encodings ......... ...... ........ '. .........   17

6   Consultants for the IQ Protocol .. ..... . ..... .......... ............ .         29

7   Experts Participating in the IQ Encodings ..................................   33

-------
                                  ACRONYMS

ALAD        6-aminolevulinic acid dehydrase
ALAS        6-aminolevulinic acid synthase
CD          criteria document
CDF         cumulative distribution function  ,
CI           credible interval
CNS         central nervous system
ECAO        Environmental Criteria Assessment Office
EDTA        ethylenediaminetetraacetate
EEG         electroencephalogram
EP          erythrocyte protoporphyrin
EPA         U.S. Environmental Protection Agency
FEP         free erythrocyte protoporphyrin
GM          geometric mean
GSD         geometric standard deviation
Hb          hemoglobin
HERL        Health and Environmental Research Laboratory
IQ          intelligence quotient
NAAQS      National Ambient Air Quality Standard(s)
NHANES II   second National Health and Nutrition Survey
NOLO       normal-on-log-odds
OAQPS       Office of Air Quality Planning and Standards
PbB          blood lead
PDF          probability density function
PMF         probability mass function
SES          socioeconomic status
ZPP          zinc protoporphyrin

                                        iz

-------
                   ASSESSING THE RISKS TO YOUNG CHILDREN
                      OF THREE EFFECTS ASSOCIATED WITH
                          ELEVATED BLOOD-LEAD LEVELS
                                        by
                     Thomas S. Wallsten and Ronald G. Whitfield
                                    ABSTRACT
               -  Formal risk assessments were  conducted as part of the U.S.
         Environmental  Protection  Agency's  current  review of the primary
         National Ambient Air Quality Standard for  lead.  The assessments
         focused on three potentially adverse effects of exposure to lead in
         children from  birth  through  the  seventh  birthday:   erythrocyte
         protoporphyrin  (EP)  elevation,  hemoglobin  (Hb)  decrement, and
         intelligence quotient  (IQ)  effect.  The same general strategy was
         followed in all three cases:  for two levels of each effect, probability
         distributions over population response rate were estimated at a series
         of blood-lead (PbB) levels.  These distributions were estimated from
         data in the case  of EP elevation and  from expert judgments  in the
         cases of Hb decrement and IQ effect.  Although of interest in their
         own right,  these estimates were combined with PbB distributions  to
         yield probability  distributions over the  estimated percentages  of
         children experiencing the particular health effects.
                                1  INTRODUCTION
       The Clean Air Act charges  the  U.S. Environmental Protection Agency (EPA)*
wjth setting  and reviewing both primary and secondary National Ambient Air  Quality
Standards (NAAQS) for selected pollutants. Each primary standard must be set at a level
sufficient to protect public health  with an adequate margin  of safety.  This report
presents the results of a risk assessment performed to assist in the review of the primary
NAAQS for lead.
                           **
       For each review,  the  scientific  basis for revising the  primary lead  NAAQS is
presented in  an updated document entitled Air Quality Criteria for Lead (EPA, 1986a),
hereafter  referred to as  the  criteria  document  (CD).   It  summarizes and analyzes
available scientific evidence about the adverse health effects of lead.  After  evaluating
and interpreting  the information in the CD,  a draft  EPA  staff paper (EPA, 1986b)
identifies the critical elements that EPA staff believe should be considered in  the review
and possible revision of the lead NAAQS.  Particular attention is paid to those subject
*A11 acronyms used in this report are listed alphabetically on pp. ix and x.

-------
                                         49
among low SES children sheltered from  lead to be quite  low (around 85). No estimates
are given for Expert I because that individual did not provide the needed judgments about
mean IQ levels for children sheltered from lead exposure or for IQ standard deviation.
5.5 SENSITIVITY ANALYSIS

        Two sensitivity analyses were conducted to study .the effects of (1) having dose-
response distributions on intervals smaller than  10 ug/dL and (2) changing the GSD value
assumed for the PbB distributions.  We considered the Hb risk assessment for the first
sensitivity analysis and the EP, Hb,  and IQ risk assessments for the second sensitivity
analysis.

        We chose the Hb judgments for the first sensitivity' analysis because the log-odds
transformation resulted in functions  with approximately equal  slopes.  This "common
slope" allowed interpolation  with a high degree of confidence between the  probability
distributions encoded for Experts  C,  D, and E.  The IQ judgments were less suitable for
this type of analysis because the slopes of the transformed distributions were not equal.
Furthermore, we chose to consider children aged 0-3 because they are the most sensitive
to lead exposure  and their dose-response distributions display the largest  variations,
which tends to accentuate any sensitivities that  may be present.

        Twenty-one dose-response distributions  were specified for Experts C and E (six
distributions on 10-yg/dL intervals were encoded)  by plotting the median values of the
transformed distributions versus the six PbB levels and drawing a smooth curve  through
the points. The smooth curve allowed estimation of median values in steps of 2.5 pg/dL
from  5  yg/dL to 55 pg/dL.  (Eighteen distributions were specified for Expert E  because
only five distributions, beginning  at PbB = 15 pg/dL,  were encoded in his case.) These
values,  along  with the  common  slope value,  completely specified  the  dose-response
distributions.  These distributions were then combined with PbB distributions in a fashion
identical to that used to produce the results described in Sees. 5.2-5.4.

        For Experts C, D, and E, the risk distributions based on the larger number of PbB
levels are virtually identical to those based on the smaller number of PbB levels.  The
differences between the two sets of ..calculations are very small, but systematic, for each
expert. For Experts C and E, the risk distributions based on fewer PbBlevels are about
0.1%  closer to the origin than are the corresponding risk distributions based on more PbB
levels.  In  other words, the risk  estimates we  reported are slightly smaller than those
that would have been produced by a finer-grained analysis. The opposite is the case for
Expert D:  the risk distributions based on more PbB  levels are about 0.2% closer to the
origin.  These results strongly indicate that encoding at only five or six PbB levels was
adequate, at least  for Hb effects.

        The second sensitivity analysis was simpler than the first. We repeated  the risk
calculations for EP, Hb, and IQ effects (performed assuming a GSD of 1.42 pg/dL for the
PbB distribution) for two additional GSD values:  1.3  pg/dL and 1.5  yg/dL.  The chosen
values bound those reported in the literature and summarized in  the  CD for lead (EPA,
1986a).  Geometric standard deviation values  for PbB  distributions  in populations  of
children are extensively discussed in the EPA staff paper that reviews NAAQS  for lead
(EPA, 1986b).  The 1.42-g/dL value is reported in NHANES II (Annest et al., 1982).

-------
                                         50
        For EP level > 53 yg/dL, risk results are virtually unchanged at low (< 7.5 yg/dL)
GM PbB values. Differences are greatest at GM = 15 yg/dL:  results at GSD = 1.3 yg/dL
are about  28% lower  (i.e., a response rate of 4% versus 5.5%), and results at GSD =
1.5 yg/dL are  about  22% higher  than those* assuming GSD  = 1.42 yg/dL.   At  GM =
27.5 yg/dL, results for GSD values of 1.3 yg/dL and 1.5 yg/dL are within 11% of those
assuming a GSD of 1.42 yg/dL.  The  threshold for lead-induced EP effects, which was
calculated  by Piomelli et  al.  (1982) to be at a PbB level of about 16.5 yg/dL, probably
explains the observation  that results are most sensitive to GM PbB  values  around
15 yg/dL. '

        Results for Experts C and E are virtually unaffected by the GSD value for Hb
level < 9.5  yg/dL and children aged 0-3.  For Expert D, differences in mean values of the
risk distributions are about 10% for GM values  > 15 yg/dL (e.g., median response rates at
GM  =  27.5 yg/dL are  4.6%, 5.3%,  and  5.7%  for GSD  =  1.3, 1.42, and  1.5  yg/dL,
respectively).  The difference beginning at 15 yg/dL can probably be attributed to Expert
D's judgment that a threshold exists for lead-induced Hb effects  in the 15-25 yg/dL PbB
range.

        For IQ  decrement among low  SES children, the  different GSD values  have
essentially no effect on  the risk distributions  for any of the six IQ experts (Experts F
through K). The differences are generally less than 0.1 mean IQ points. The same is true
for the  IQ  response rate at IQ* =  70 and low SES children.  Differences between, GSD =
1.42 yg/dL and the other two GSD levels are less than 0.2% (in terms of response rate)
for Experts H, J, and K,  and less than 0.1% for Expert G.  Risk distributions for Expert F
are unaffected.

-------

-------
                       B.4
EXPOSURE ASSESSMENT
              Attribute 1  The purpose and scope of the exposure assessment and the
                           underlying methodologies are clearly described.
                  SOURCE  Case Study K. Tetrachloroethylene (Pages 1-77).

-------
                                 UCRL-15831
   Health Risk Assessment
of Tetrachloroethylene (PCE)
in California Drinking Water
  K. T. Bogeh, L. C Hall, T. E. McKone,
     D. W. Layton, and S. E. Fatten
   Environmental Sciences Division
Lawrence Livennore National Laboratory
       University of California
           P.O. Box 5507
        Livennore, CA 94550
           April 10, 1987
            Prepared for
  California Public Health Foundation
            P.O. Box 520
         Berkeley, CA 94701

-------
                                 1.  INTRODUCTION

      This document presents an assessment of the potential health risks
 associated with exposure to tetrachloroethylene (also known as
 perchloroethylene or PCE) dissolved in California drinking waters.  .This
 assessment is being provided to the California Department of Health Services
 (CDHS) for the development of drinking-water standards to manage the health
 risks of PCE exposures.   Other assessments required in the risk-management
 process include analyses of the technical  and economic feasibilities of
 treating water supplies  contaminated with  PCE.   A primary goal  of this
 health-risk assessment is to evaluate dose-response relationships for observed
 and potential toxic end  points of PCE in  order to define dose rates that can
 be used to establish standards that will  protect members of the general  public
 from .adverse health effects resulting solely from water-based exposures  to
 PCE.  We also analyze the extent of human  exposures attributable to
 PCE-contaminated ground  water in California.
      The document consists of eight sections,  plus supporting appendices.
 Each section provides information that risk  managers  at  the CDHS can use to
 develop PCE drinking-water standards that  will  safeguard human  health.   Our
 assessment begins in Section 2 with a review of the uses of PCE,  its
 environmental chemistry,  and concentrations  measured  in  different
 environmental media.   The next section provides an overview of  published
 studies on the absorption,  distribution, metabolism,  and elimination of  PCE,
 emphasizing those studies that have defined  the rate  and extent of  these
.processes  in rodents  and  humans.   In Section  4,  we review studies of the
 acute,  subchronic,  and chronic toxicity of PCE  to  animals,  including a  summary
 of data from bioassays conducted  to evaluate  PCE  carcinogenicity.   In
 Section 5,  we provide an  overview of PCE health effects  in  humans,  review
 epidemiological  studies  involving PCE,  and examine human  data on  PCE's toxic
 effects on  specific  organs  and systems.
      In Section  6,  we describe our procedure  for  calculating  human  PCE
 exposures  attributable to contaminated  groundwater supplies.  This  section
 takes  an  integrated  approach  to  the exposure  assessment.  A household
 consisting  of two adults  and  two  children  uses  approximately  1000 L/d of water
 from wells  or surface water.   Our  approach considers  how  PCE contained in  this

-------
amount of water can result in human exposure through ingestion,"inhalation,
and dermal absorption.  For each pathway we develop pathway-dose  factors  that
translate a unit concentration in mg/L in tap water into a lifetime equivalent
dose rate in mg/kg-d.   We use the pathway-dose factors and data from A81803
surveys (CDHS, 1986) toadetermine the magnitude and distribution  of
human-lifetime dose rates attributable to PCE in California groundwater
supplies.
     We do a quantitative dose-response assessment for PCE cardnogenicity in
Section 7, using a "linearized" multistage dose-response extrapolation  model
along with four sets of animal cancer-bioassay data as input to that model.
In this quantitative carcinogenic potency assessment,  a relationship between
the doses applied in the animal bioassays and the corresponding effective or
metabolized doses is derived using a simple pharmacokinetic model  and
available data on PCE metabolism in rodents.   The results of our  metabolic
analysis are compared to and shown to be consistent with results  based  on
other analytic methods; the method we use is  also shown to provide a good
prediction of available data on human PCE metabolism.   Our calculated
carcinogenic potencies of PCE to animals based on different sets  of bioassay
data are then extrapolated to humans using two different methods  of
inter-species extrapolation.  This yields a set of 112 alternative potency
values based on different assumptions that might be applied to humans exposed
to PCE in the context,of regulatory risk assessment.  Finally, we discuss
methods applicable to calculating PCE concentrations in water associated  with
given, predicted cancer risk levels using information  provided in this  section
and in Section 6.
     The last section addresses some of the key uncertainties associated  with
the health-risk assessment and also presents  some research recommendations for
reducing those uncertainties.

-------
          .   2.  CHEMICAL AND  PHYSICAL  PROPERTIES  AND  ENVIRONMENTAL
                          TRANSPORT AND TRANSFORMATION

     Tetrachloroethylene is commonly referred to as PCE, or perchloroethylene,
This compound is a volatile, chlorinated hydrocarbon that is widely used as a
degreasing solvent.  In this section we provide an overview of its ,use in the
U.S. and, importantly, its  transport and fate in the environment.

CHEMICAL AND PHYSICAL* PROPERTIES

     PCE is a colorless, nonflammable liquid with  a chloroform-like  odor. .It
is slightly soluble in water and has a vapor pressure of 15.8 mm Hg  at 209C
and'a boiling point at 12TC.   Table 2-1 lists the chemical structure,
alternative names, and identifiers of PCE.   Its physical and chemical
properties are listed in Table 2-2.

PRODUCTION AND USES                                                 :       ,
     In the U.S., PCE is primarily produced via chlorination of hydrocarbons
(based on ethane or propane) and from processes based on ethylene d1chloride
(i.e., oxychlorination of 1,2-dichloroethane).   Additives, amines or esters,
are added in small amounts to stabilize the product.   PCE is the most stable
of the chlorinated ethanes and ethylenes, requiring only small  amounts of
stabilizers (Keil, 1979).  PCE production plants vary in size from 20 million
to 90 million kg annual  production (Lowenheim and Moran, 1975).  In 1983, the
United States production totaled 308,076 metric tons  (308 million kg) (CARB,
1984).  The only producer of PCE in California is Dow Chemical  Company,
Plttsburg, with an annual capacity of 22.7 million kg (Chemical Marketing
Reporter, 1983).  There are no known natural sources  of tetrachloroethylene.
     PCE is a widely used solvent with applications as  a dry-cleaning agent, a
metal degreaser, and a chemical  intermediate in the manufacturing of
fluorocarbons.  It 1s also used as a fumlgant,  in the extraction of caffeine
from coffee, in removal  of soot 'from industrial boilers, and as a heat-transfer
medium.  Of the total PCE used 1n 1983, 59! was as a  dry-cleaning agent and in,
textile processing, 2U as a metal degreaser, 111 for export, 6t as a chemical
intermediate, and 31 in miscellaneous uses (Chemical  Marketing Reporter, 'l983).

-------
Table 2-1.  Chemical  structure of tetrach'loroethylene,-alternative names,
identification numbers, empirical formula, and molecular weight.
Chemical structure:
Cl - C - C - Cl
     Cl  Cl
Empirical formula:  CaCU             Molecular weight:   165.85
Chemical Abstracts Service registry number:   127-18-1
NIOSH Registry of Toxic Effects of Chemical  Substances number:  KX3850000
                                                       «
Alternative names:  PCE, Perc, tetrachloroethene,  perchloroethylene,  ethylene
                    tetrachlorlde, carbon dichloride
Common trade names: Antisol, Dee Solv, Per Sec, and Texranec.
ENVIRONMENTAL TRANSPORT AND TRANSFORMATION

     Tetrachloroethylene tends to partition primarily to the atmosphere.  It
has been estimated that 85 to 901 of the PCE produced is eventually released
to the atmosphere (U.S. EPA, 1985a; WHO, 1984).  The key properties of PCE
that affect its movement in the environment are its high vapor pressure and
low solubility in water.
     GEOTOX (McKone and Layton, 1986) was used to estimate the equilibrium
distribution of PCE in air, soil, and water.  GEOTOX is a multimedia
compartment model that simulates the environmental transport and transformation
of a chemical, based on its physical and chemical characteristics and the
properties of the landscape into which it 1s released.  A simulation*of the
environmental partitioning of PCE was run using California landscape data,
properties of PCE (see Table 2-2), and PCE source-emission data from the
California Air Resources Board (CARS).  The PCE source  term was represented by
an annual release of 1.83 x 107 kg/y over an area of 411,000 km  (CARS,
1984); of this source, 101  1s assumed released into soil, 11 to surface water,
and the  remainder directly  to *he atmosphere.  The simulated equilibrium
distribution of PCE is shown  in Fig. 2-1.  Most of the  PCE released  to  the
environment is found in the atmosphere.  However, the  equilibrium
distributions, 861 in  the air and  111 in surface  water,  reflect both the

-------
      Atmospheric
                                 86i
             Biomeu
                     0.02%
      Upper toil
             0.4%
                        i
      Lower soil
               0.5%
                        i
                                  Atmospheric per tides

                                        4 x  10"8%
I	
      Ground weter
     8
                             1%
     t
Surface wctar

         11%
Sediments
         1%
                                                      8
Flgurs 2-1.  Environmental  distribution  of PCE undtr steady-state  conditions,
Partitioning between  compartments  Is  predicted by the computer  model  GEOTOX
(McKone an4 Layton,  1986).

-------
Table 2-2.   Chemical,  physical,  and organoleptic properties of
tetrachloroethylene.
Property
Boiling point at 760 mm Hg
Freezing/melting point
Density at 20aC
Vapor pressure at 20aC
Henry's law constant at 20aC
Conversion factor
Units .
ac
9C
g/cm3
mm Hg
atm-nH/mol
mg/m^-ppmv
Value
121
-22.4
1.65
15.8
0.0227
6.89
Reference
Hawley (1981)
Hawley (1981)'
Hawley (1981)
Sittig (1985)
Mackay and Shiu (1981)
Verschueren (1983) .
m
 2/s
Diffusion constants
  at 1 atm, 20aC
    Air
    Water
Solubility in water at 25'C    mg/L
Log octanol/water
  partition coefficient
Odor threshold in water
Unitless
mg/L
7.4 x 10~6
7.6 x 10-10
 150
              3.14
              2.46
Lyman et aU  (1982).
                          MacMnson e_t aj..
                          (1981)
               Leo (1983)
               Callahan et al.  (1979)
           3.0 x 10-1     Zoeteman et a_L (1974)
relative magnitude of the source (891 to air and 11 to water) and the
effective residence times.  The loss rate of PCE in air is an order of
magnitude greater than that in surface water.  This accounts for the apparent
"enrichment" of PCE in surface water.
Air
      PCE  in the atmosphere is subject to relatively rapid chemical or
photochemical degradation.  In the troposphere, it photodegrades, ultimately
leading to the formation of hydrochloric acid, trichloroacetic acid, and carbor
dioxide in the presence of atmospheric water  (U.S. EPA,  1985a).   PCE can also
be  removed by scavenging mechanisms, primarily through hydroxyl radicals
(Dimltriades et a]..,  1983).   Estimates of  its atmospheric residence time are
on  the order of one year or less  (see U.S.  EPA, 1985a).

-------
     Singh et a_L (1981) compiled monitoring data  for  the concentrations of
 several volatile organics in ambient air and found that  for  the. western half
                                                       •3
 of the U.S. the average PCE concentration was 4.3 ng/m  and  the overall range
 was 0.23 to 51.6 pg/m  .  The U.S. EPA (1985a) reported ambient air PCE
 concentrations in California (1972-1980) ranging from  0.2 to  19.0 ng/m3.  The
 California Air Resources Board (Nystrom, 1986), based  on preliminary data,
 found average ambient air PCE concentrations for several California locations:
 Los Angeles 1180 ± 900 parts-per-trillion by volume (pptv) (8.1 ± 1.2 jig/m3);
 San lose 490 ± 330 pptv (3.4 ± 2.3 iag/m3); Long. Beach  1030 ±  560 pptv
 (7.1 ± 3.9 yg/m3); Stockton 450 ± 170 pptv (3.1 ±1.2  ng/m3), and Simi
 Valley 330 ± 250 pptv (2.3 ± 1.7 jjg/m3).  These data indicate that PCE
 concentrations in the ambient air of urban areas are higher  than those in
 rural areas (or less densely populated areas).
Water
     In surface waters, PCE rapidly volatilizes into the atmosphere.  Wind
speed,, agitation of the water, and water and air temperatures affect
evaporation rates.  Photodegradation, in contrast, is a slow decay process and
does not appear to be an important transformation mechanism in water.  The
half-life of PCE in shallow water due to volatilization has been estimated at
24 to 28 min in laboratory experiments (Oilling e_t a_L, 1975).  Zoeteman
et a].. (1980) measured PCE persistence in surface waters of the Netherlands
from 3 to 30 days (half-life), while in lakes and ground waters, the half-life
was estimated to be 10-fold higher.
     In ground water, PCE is relatively persistent, with degradation occurring
through hydrolysis and biotransformation.  It is denser than water as an
undissolved liquid, consequently it tends to sink in ground water.  Vogel and
McCarty (1985) have shown that PCE biotransforms to trichloroethylene (TCE),
dichloroethylene, and vinyl chloride via reductive dehalogenation under
anaerobic conditions.  They further suggest that the potential exists for the
complete mineralization of PCE to carbon dioxide in aquifer systems.  The
half-life of PCE due to aqueous hydrolysis in natural waters can be on the
order of months (Dilling et a_L,  1975) to several  years (Pearson and
McConnell,  1975).

-------
     The U.S. Environmental Protection Agency (1985a) reported a mean PCE
concentration of 1 vg/L from 1102 surface water measurements in 45 states
(from August 1975 to September 1984).  An important source of data on the
concentrations of PCE in drinking water supplies is a survey of large water
utilities in California (i.e., utilities with more than 200 service
connections) that was conducted by the California Department of Health Services
(1986).  From January 1984 through December 1985, the wells in 819 water
systems were sampled for contamination by organic chemicals.  The water systems
considered included a total of 5650 wells, 2947 of which were sampled.  The
wells sampled were selected based on the likelihood of contamination.  PCE was
found in 199 wells in concentrations up to 166 jig/L, with a median
concentration of 1.9 n9/L.  Generally, the highest fraction of contaminated
wells and the wells with the highest concentrations were found in the heavily
urbanized areas of the state.  Contamination was state-wide.  Los Angeles
County registered the greatest number of contaminated weMs (i.e., 140).
Soil
     There is limited information on the behavior of PCE in soil.   The solvent
can be adsorbed to soil or leached through soil when dissolved in  water or as
a separate organic phase (as in large spills).  PCE associated with soil air
or soil water is more mobile than the absorbed portion (Schwarzenbach and
Westall, 1981).
     The adsorption of PCE to soils appears to be correlated to its octanol/
water partition coefficient, the organic carbon content-of the soil, and the
concentration of PCE in the liquid phase.  PCE appears to leach rapidly
through soils of low (<0.1%) organic carbon content (U.S. EPA, 1985a;
Schwarzenbach and Westall, 1981).  .
     Several studies have documented the mobility of tetrachloroethylene in
soi 1/groundwater systems (Piet et a].., 1981; Schneider et aj.., 1981;
Schwarzenbach and Westall, 1981).  Wilson et a_]_. (1981) showed that most of
the chemical was lost from the soil via leaching or volatilization to the
atmosphere.  Persistence in soil ranges from months to years.

-------
                          3.   PHARMACOLOGY AND METABOLISM

      PCE 1s readily absorbed through the lungs and gastrointestinal tract and
 may, to a lesser extent, be absorbed through the skin.  Once in the body, PCE
 distributes into all tissues.  Steady-state tissue concentrations are a
 function of the absorbed dose, partitioning factors, and pharmacokinetic
 properties, such as rate of metabolic conversion and elimination.
     8 The primary metabolic pathway of PCE 1s thought to, involve oxidation to
 an epoxide as the first step, although this epoxide intermediate has never been
 isolated in vivo (Bonse et aj.. 1975; Greimeta_[., 1975).   The epoxide
 undergoes rearrangement to form trichloroacetyl  chloride,  and ultimately
 trichloroacetic acid (Yllner, 1961; Daniel, 1963; Moslen et aT., 1977;  Costa
 and Ivanetich,  1980).   Studies In which radiolabeled PCE was administered to
 anjmals have occasionally recovered oxalic acid  as a significant urinary
 metabolite miner, 1961; Dimltrieva, 1967;' Pegg et aj..,  1979)., Carbon
 dioxide is also commonly produced (Pegg et a]..,  1979;  Schumann  et aj..,  1980).
      In this section,  we present an overview of  published  studies on the
^absorption,  distribution, metabolism, and elimination  of PCE.  Our emphasis  is
 on studies that have defined the rate and extent of each  of these processes  in
 humans  and tn rodents.   Proposed metabolic  pathways are  discussed in some depth
 because metabolism is  responsible for the transformation  of PCE to one  or more
 reactive species.

 ABSORPTION
      In  the  following  paragraphs  we  review relevant  data  on  PCE  uptake  through
 ingestlon, dermal  absorption,  and inhalation.

 Inqestion

      Absorption of PCE from  the gastrointestinal  tract  has been  measured
 indirectly as  percent  of  dose  recovered.   The  percentage  of  dose recovered
 after administration of PCE  is similar  in  mice and rats,  varying between 80 to
 1001.  PCE is  absorbed rapidly; peak  blood concentrations were measured in
 rats  one hour  after a  500 mg/kg dose  (Pegg et  aj.., 1979),

-------
     Little information exists regarding amount or rate of oral absorption of
PCE in humans.  Koppel et a_]_. (1985) reported that the blood concentration of
PCE following an oral dose of 400 mg was described by a two-compartment model
with half-lives of 160 min and 33 h, respectively (unpublished data cited in
Koppel et aj... 1985).  The same authors (Koppel et a_L , 1985) measured the
concentration of PCE in blood at 21.5 yg/mL within one hour of ingestion of
12 to 16 g.  Although far from definitive, these reports suggest fairly rapid
and complete oral absorption.

Dermal absorption

     Jakobsen et a].. (1982) measured the absorption of PCE through guinea pig
skin.  Animals were in contact with liquid PCE for 6 h; during the exposure,
blood concentrations rose rapidly, and peaked within 30 min.  Tsuruta (1975)"
estimated the rate of absorption of PCE through mouse skin to be
              2
24 nmol/min-cm  of skin.
     Percutaneous penetration of PCE vapor in humans exposed to ambient air    %
concentrations of 600 ppmv is approximately one percent of pulmonary absorption
(Riihimaki and Pfaffli, 1978).  In studies in which volunteers immersed their
thumbs in liquid PCE, a peak concentration of 0.3 ppmv in expired air within
40 min was measured, which decreased thereafter.  Four other chlorinated
solvents were tested by this method; trichloroethylene, carbon tetrachloride,
rnethylene chloride, and 1,1,1-trichloroethane.  Peak alveolar concentrations
were 0.5, 0.6, 3.0, and 0.7  ppmv, respectively.  PCE produced the lowest
maximum  concentration in  alveolar air, and it had the slowest rate of
elimination  in breath.  The  relatively low concentration of PCE in exhaled air
suggests that dermal absorption is  Mmi ted (Hake and Stewart, 1977; Stewart
and Oodd,  1964).  However, the work of Brown, et a].. (1984)  suggests that
dermal absorption of dilute  aqueous solutions may contribute a significant
amount to  the overall absorption of PCE  (see Section  6).

Pulmonary Uptake

      During  inhalation, PCE  diffuses across  the  lungs  and  dissolves  into  the
bloodstream.   The rate  of transfer  is  dependent  on  the  blood/gas  partition
coefficient  for  PCE.  One estimate  (Monster  et  a]..,  1979)  placed  the  human.
                                        10

-------
 blood/gas  coefficient of PCE  at  16,  while  Gargas  et  a_K  (1986)  reports  a value
 of  10.3.   Both values reflect  the'fact  that  PCE  is  lipophilic and  readily
 diffuses into the blood.  The  uptake of PCE  by the  lungs  is  also determined by
 the  alveolar ventilation rate  (i.e., that  fraction of  total  respiratory
 ventilation from which volatile  organic compounds, such  as PCE, may  be  cleared
 by absorption into alveolar capillary blood), ambient  concentration, exposure
 duration,  and metabolism.
     At steady state, the amount of  PCE taken up  or  retained will  be equal to
 the  amount of PCE metabolized.   However, as  pointed  out by Guberan and
 Fernandez  (1974), the time it  takes  for humans to approximate a steady-state
 equilibrium upon inhalation of a constant  concentration of PCE  is  one to two
 weeks.  Since no human studies for PCE  involve constant exposure for this
 length of  time, steady-state conditions  were not  approximated in any of these
 studies.   Therefore, the uptake or retention rate varies as a function of time
 for  all of these" studies, and no generalizations  can be made that are
 independent of the duration of exposure.   The value of these observations for
 purposes of risk assessment are  therefore  limited, since generalizations
 relating uptake to metabolism cannot be  confirmed by data from any of these
 reports (see Section 7).
     Several experimental human  studies  have quantified PCE absorption in
 terms related to what we here refer  to  as  "percent absorption" or "percent
 uptake" (operationally defined as one minus the ratio of alveolar to ambient
 air  concentration, ^multiplied by 100X)  and we review such values here.
 However, as noted above these values are of limited use in predicting the net
 quantity of PCE retained over extended  periods of time following environmental
 exposure.
     Yllner (1961) reported the average pulmonary absorption  of  mice exposed
 to 1.3 mg/g of PCE in .air to be 701.  In these animals, absorption  varied from
 42 to 87%.   In a study by Pegg et aj.. (1979), the peak blood  concentration  of
 PCE  in rats during a 6-h exposure to 600 ppmv was approximately  10  ng/mL.
     Monster et aJL  (1979)  observed an  inverse relationship between uptake  and
 exposure duration in humans  over the course of a  4-h  inhalation  exposure to 72
or 144 ppmv of PCE.   The net uptake at  the end of 4 h was approximately  60t of
 that during the first hour.   This observation indicates that  net uptake
decreases  as blood and tissue concentrations of PCE equilibrate  with PCE in
 the air space  of the lungs.   Net uptake is affected by differences  in
                                       11.

-------
ventilation rate.   When volunteers were exposed to 142 ppmv while under a work
load (i.e., an increased ventilation rate), the uptake of PCE increased to
over two times what it was at rest (Monster et a_L,  1979).
     Fernandez et  aJL  (1976) exposed humans to 100 ppmv PCE for 8 h and
measured the concentration in alveolar air.  Alveolar air concentration rose
rapidly in the first half hour and then continued to increase throughout the
experiment, although at a slower rate, reflecting a sustained decline in
percent uptake observed.
     The percentage of PCE absorbed in humans through the lungs has been
estimated at 50t (Ohtsuki et a].., 1983).  After a 6-hr exposure of five
volunteers to 0.39 mg/L PCE, retention of PCE became stabilized after 1.5 h to
an average of 62% of respired PCE (Bolanowska and Golacka, 1972).  The report
of Bolanowska and  Golacka (1972) is contradict,  by the data of Fernandez
et ah (1976) and  Monster et a].. (1979).  These investigators found that
retention of PCE showed no signs of approaching equilibrium after exposures of
6 to 8 h.
     Table 3-1 summarizes the absorption and recovery of experimentally
administered PCE in animals.  Table 3-2 lists parameters of absorption,
metabolism, and disposition of PCE in humans.

DISTRIBUTION AND BIOACOJMULATION

     PCE diffuses into the bloodstream and distributes to tissues, primarily
organs and fat.  The binding of metabolites of PCE to hepatic protein has been
measured by Pegg et aj..  (1980), and Mitoma et aj.. (1985).  Savolainen et a_K
(1977) has also observed substantial  levels of PCE metabolites in the
perirenal  fat of rats 17 h after the  end of exposure (200 ppmv, 6 h/d for
4 d).  Seventy-two hours after oral or inhalation exposure to
tetrachloroC14C]ethylene, measurable  radioactivity was found in the liver,
kidneys, fat, lungs, heart, and adrenals of rats.  The major part of  the
radioactivity was concentrated  in  the liver, kidneys, and fat  (Pegg et  a]..,
1979).  A  similar distribution was observed by Frantz and Watanabe (1983)
after PCE  was administered  to  rats in a saturated drinking-water  solution
(containing approximately 150  ppmv).
                                        12

-------



















,
VI
f**
anima
e
41
e
V
-C
2
o
u '
13
L.
41

O
U
41
0
U
41
t.
-o
S
f»
e
o
.u
^L !
a
VI
J3
41


41
U
C
41
41
ec


4i e
1- 0
• 3 •<-
VI **
o /a
a. i.
X 3
uj -a
c
a
3 41
IV VI
S. 0
e
41 U
U O
O


41
.4-1
3





VI 4rf
— en
, u —
4) 41






41
3
'"ifl
>







Parameter



^—
**Q
en
u
41
C
>

.C
CM


en
"X

CO
*

0
13
*9
^»
Ji
e
*•»





41
VI
i



_
CO
o
CM

01
o en
• e
g £





•^
41
' J3
U
O
VI
Percent at
(average)
' ' 00
*™
*
^j
' jW
3


x
-
x
•o .
in


en
en
s

1





0

^



41
VI
i








m
\o
an



41 .

3
VI
41 0
VI Q.
O X
•o.a 4»
o -a ' vi
4i a
*j u a.
C 41
41 > JZ
u o
<- u oo
41  -C
u a
i. U CM
41 41 r-
a. u ^
i
~
^
^™f
1^1
«i-
,5
a
'£
^
•a
1/1


en
jt
i1
- ^p
S3
0





1





•*J
ec









•w
CO



41
U
3
VI
41 O
vi a.
O X
•o  JS
u o
t. u oo
OL 2 3

en
^™'
„
•
'"^
•a .
en
en
41
^
41 3
r— VI
f |
 41



"
^S.
-





2
o




t*^
en
- S









a
CO
o



41
k. .
3
VI
31 §.
O X
-a 4>
a -a «i
41 O
— u a.
e 4i
41 > £
u o
U. U CM
4) 4i r-
Q. U ~.

en.
!«•
^

^J
43
at
en
-£

41 3
a, S
e a.
— x
t/t 4)


9
•s.
at

0
o





2
o





a
*j in
<« CM
OK —>









CM
cn



•J
L.
3
VI
4* O
VI CL
a x
•Q (I)
•^ *4
o -a vi
4i a
« v. a.
C 41
41 > j;
u «
<- U CM
4» 41 I-.
a. t. ^
v
e
*^
3
j
•a
e
'Q
N
_ fO
u oo
t. en

f
CM


en — .
J£ •
GA O
5 ••
— a.
oo -2.
41
S.
e en
•*• c
•'^
•— c
O -a

"en

03
CM
O "
m
m CM









S3
O
, . - O



 £
w a
U u CN
41 41 r-
a. i. —
13

-------
i
8































01
3

C
0
O

PI
i

4)
(J
C
*•
VM
QC
4) e
1. 0
VI ^J
a 19
a. u
X 3
uj -a
e
o
^j 4)

^ O
_• -a
S u
u a
0


41
O
cc

VI «J
.2 f
ftl 41
ex 2
10 —



§
2







Parameter


1
^
ti
"3
Ol
Ol
41
Q.




.c

VO





Q.
a.
o

Q
•5
i

• Ol
•** vrt
<9 04
oc •—



en



£
3
VI
4> a
vi a.
a x
•a 4i
^B ^t
O T3 VI
« 0
** U Q.
C 09
41 > .C
w a
u u oj
cu 4) PH
a. u ~~

' «r>
1
J
y iq
m
Ol
Ol
41
a.




^

vO





i.
a.
o
VO

a
"S
.c >
c

"S
o
* O4
OC — •



0
0
a



41
3
1 Of O
0 X
•0 41
^^ ^J
o -a vi
41 a
.u u a.
C 4)
vj O
U U CM
« 41 r*»
a. u —
' §
. 2
3
tl
e
e
1 .
i/i




.c

vO





a.
a.
o

a
19
j=
e

41
i



o
§



£
'3
VI
4i a
vi a.
e x
•a 4i
*^ " ^
o -a vi
4i a
*j u a.
e «
41 > -C
u a
u u oj
4i 4) r»
a. u «—


to
2
_
19
L.
3
VI
- 1—
















_
i
4)
a

41
VI .
i
e

Q
O
•x
c
CM




41
19
VI 1-
3
o e
Percutane
absorptio
CM
01 ~
~ ^
J 3
C
^^1 4)
OJ( VI
Ol O
Ol -*
41 - - 15
a. o
41
U
41 3
r~ VI
Ol O
CO. JS
••• X




01

s
1


-
2 1
p a
Ol
a.
01 ^
41
o e
,19 04 3
a ~- o
3


X
o
^ a. v/)
^- <»»• a


•o -a
1 J
"^ i ^ §'
19 •— ifl —
41 *> 41 *J
a. 
-------

















































































•a
41
3
e
• «•
^ .
e
a

i^

,
^B
|
4)

^3
1—


4)
U
e
4)
L.
41
41
oc










4i e
i. a
VI -*
a fl
o» ^
X 3
LbJ •^


e
a
Z 4)

41 -C
— O)
U •—
41 41
& S
(SI ^

V;







41
3




















U
41
4)
g

U
2.
*•
&

*
*
^•J
I9

"Si

0)
o>
41
O.










^g

*O







1


i




e
a

%
"^
^^
e










^^
^
oe





^^
1 _

'c "3
* *

O) 9)

IT) O

K C3











U
fQ
e
1

3
a.
4)


^ ^j
a a.
h- 3
VI
41
><-
t~
O
^3
(^

4J ~

•
• -c


41
i.
A
a

VI
*J
e
i
4)
1.
3
VI
tQ
i
>,
jg

a


•o
41

e
3
a
u
u
UJ
°- .
a

41
VI
a
•o
4)

41

VI
'c
1

<4-
a

e
a

^^
u
a
o.

41

a

VI
u
41

2

•a
41
41

a

«
u

41 41
VI 3
a vi
•a vi

*4B ^
a'
•a
•^ e
4)
u
u iq
41 -w
& 41
1.
|Q U
X
41
N
3
a
_
o
^m*

^~ •

U

e
41
•a


•W
e
41
1.
ifl
VI
U
41




1.
Q.
a •

•"
41


^

41
41
a
.c
e
e
o.
i.
a
VI
13-
^
e
41
U
U
4)
a*
a
4)
'£'

U

^g :
e


41
J3

a
^
41
41

.^
VI

u

VI •
"* "n
T9 1^1
4t go

41 <—

a
U <£
41 Q.
I. UJ


e vi
41
u 3

a. iJ-T
4» "
J5 *••
^^

^3 •
"SI
15.

-------
              fvl
              r»
              en









u
c
4)
U
41
St.
41
ae
e
o
*j
2

a
o

'U 41

c u a
§ 4) U
-c u , o
c-
C 0
— VJ
UJ
*•«
o
s
•*» A)

'M 0
a.
V)
'•5
•0
C

B* 41
VI 3
**• f"»
•"• ^
o >
*
1
e
a
^J
a.
u
o
VI

<

•
CM I.
1 41

4)
41 §
r- g

fQ fQ
H- a.

on
CO
ai


•
<3
•S3
vt
••»
Jrf
3
VI
*J
;§

















e
a
m
*fl
•§
*"*








ei
vn


f
£
^j
*J


CJ
JQ
U
O
W1
^3
r^

^j
e
41 VI
u en
u e
41 3
a. <—
vO *fl ^r • • • •
ececk.
41 e 414) e u /o i) ra "o 41

VI e VIM |Q J 41 41 41 4) in
e ^ ce *— 4>.jQ.o.a.dc
a 41 55 O •••'3333Q
x u. xx 09  O *> *J U1
U1 Irt Irt , •
(\4 {V) t/3*T U) I/1COCOCOCN4O
vo <— v _ cn vn »• •—

e
a
f4l 41 41 —
"3 "3 "3 • •«
g«> *j w i.
Q. a. Q. -u
j: 3 33 e u 4)
-u eu — 3 >.
>> >i >i e u 19 * •&
•o u u u a evia
41 fd l«IQ — O •« •— J3

u a a o e — 41 c TS
a s as 41 j« ifl cn — fl o
I/I F— r— r— *•> r—
<9 0. Q. O. U a. 41 — J 0 ^>- j=

^| flj 3J QJ ^J ^ *
e en encn e *j i i i i i i
4) VI nj  > > 41 —
cv -- < < < a. t-
16

-------




















.













^
41
3
e
••••

e
Q
LJ

s
|S
& o
ON
m ~
e
o
ro
2
' C
• '




m
CM





VI
•o 01

•<-> 3

1 .2

41 "O
41
^i O)
e c
41 ro
u j:
u u
41 C
a. 3 ,
>T
•^
N
41
•a
e
ra
c
u
41
14.
•a
c
e
41
3
U





-e
00



a.
•a.
0
*^;
C
*•
<•
c

Ol
E
o
o
0











c
41
• rs
3
.a

• |
T
••^

41
•a
e
e
41
• U.
•o
s
e
u
41
J3
3 *






^
CO



a.
a.
o
^~
c
o
2
_g
c
OJ

Ol ifl
E .S
o x
o o
m u
. a.









c '
41
•a

3
41
VI
O
a.
3
r*
^-*
N
41
C
C
1.
41
U.
•a
c
rO
e
u
41
3
0





-c
00



a.
o
^
c
o
*
J?
c


-C

•
r^

CNi
—
*^

'JZ
^

ai
Q
o
-O
•59
41

(j

1
a.
•^
C ^
ON

*"
rO 4J
U U
I 1
1,2
•a e
C 0
•a
(fl Q)
•O VI
41 ro
1—4 •«!*
41
e? "iT
•^ t
VI 3
— VI
-c a.
r^ 4)



a.
a.
0
*"~
e
a
"3
2
c




^^
in
M3








CM

-IT

a
rO
^
f— „
a.
VI
41
ae

2

^_

u
3
i
js
C
•a
.41
"

•a
Lrt
, m
•^^
' J=
00
s
a.
0
o
o

*"~
c
o .
rO
•5
C



•JC
1
2


eg
>s»
^**
*^

-------

















































•o
01
3
'*•
C
0
o

B
CM

41
J3

>—

41
u
e
41
0)

"S
ee










e
o

IS
u
°
e
0
'H 41
||
4) U
U O
e
o
u





^j
3
O
ce



at
3
IS











41

1
1
IS
Q.
f".
2
j


ta

u
41

VI
c











5
JC
CO



1 -


o
a
e
a
'£
is

fQ
•C
c



^

uv
CM





41 41
IS O
*^ OU
VI ••-
1 -9
T3
IS •
51
VI •-
a .a
«J — 41
— 3
4) — VI
g 3 VI
.S er —
t— 4) **
i
j


ta


^
3
VI,
jC
0











f
CO



i
a.

1
e
a
'2
IS

IS

e





u


8
.12

o
J3

1

a
e
o

i.
3
IS
v>
in
CO
en
-


^j


/.•
41
a.
o

4)
k.
3
VI
e
a.
X
41

41
'01
. e
1/1




2

|





^**
/Q
u
O

e
'i
JS
o
va n
.• ro







CM
~

Q


e
J
PMB
UJ
































c
'I
41

IS

a.
3

41
«J
" .1
VI
41
41
a>
41
§
U
**"
•a
41
*
3
U
Is
u
VI
41
3

>




























i





















41
<*.
*«-
T 4)

>— J3
IS IS
JS U
I/I 1—
, 4i a.
w a.
O IS
e
41 ->J
•a o
e

-- i'
J3 U


18

-------
      Information on  the  distribution  of  PCE  in  humans  comes  largely  from
 reports  of  accidental  exposures  (Stewart et  a_L,  1961a;  Stewart,  1969; Hake
 and  Stewart,  1*977; Koppel  et  a]..,  1985).  Overexposure can  result  in central
 nervous  system  (CNS) depression,  cardiac arrhythmias,  alteration  of kidney
 function, and liver  injury.   These observations provide  indirect  evidence that
 PCE  distributes to the nervous system, liver, and'kidneys.   Additionally,
 tissue concentrations  in the  liver and brain have  been measured at levels 10
 to 50 times greater  than those in blood  following  fatal  intoxication
 (Lukaszewski, .1979).                       ,     "
     PCE is a relatively stable  molecule and is metabolized  slowly.  PCE is
 soluble  in  lipids, and this factor and its slow rate of  metabolism lead to its
 accumulation  in tissue following  repeated exposure (Filser and Bolt,  1979;
 Loew et  ah,  1983).
     On  the basis of a model'  of  PCE uptake, distribution, and elimination,
 Guberan  and Fernandez  (1974)  predicted that the solvent  woulS distribute
 primarily to  three body compartments; adipose tissue,  muscle, and tissues rich
 in blood vessels (some PCE will  also  enter poorly perfused tissues such as
 bone and cartilage).  This pattern of uptake and distribution is dependent on
 the  blood flow  to a given  tissue, the volume of each tissue, and on the
 solubility of PCE.  The model of Guberan and Fernandez (1974) shows that the
 accumulation of PCE in all body  compartments will  increase rapidly during the
 course of an 8 h exposure.  At the end of this time, the greatest
 concentration of PCE will be  in muscle and fat (these  tissues wiH have
 approximately equal amounts of PCE).
     Because the blood supply to adipose tissue is less  than to other .tissues,
 and  because PCE is more soluble  1n fat than in blood,  levels of PCE will
 continue to increase in fat for many  hours after the end of exposure.   (This
 is in contrast to muscle and other tissues,  where concentrations of PCE begin
 to decrease as soon as exposure is terminated.)
     Once whole-body, steady-state concentrations  are  reached in relation to
 concentrations in air,  the amount of PCE in  each tissue depends on'the tissue/
 blood partition  coefficient.   PCE is   lipophilic  and the adipose/blood partition
 coefficient is the highest of any tissue type (at  37°C the adipose/blood
 partition coefficient is  about 107)  (Guberan  and Fernandez',  1974).  The model
of Guberan and Fernandez  (1974)  predicts  that dai.ly occupational  exposure to

-------
PCE at TOO ppmv, 8 h/d would lead to accumulation of PCE in fat .(at
equilibrium, more than 90% of the body burden of PCE will be in the fat).
     Savolainen et a_K (1977) noted accumulation of PCE in the blood, liver,
fat, kidneys, and brain of rats following 5 d of inhalation exposure at
200 ppmv.  PCE levels in perirenal fat, brain, and lungs rose continuously
during the experiment.  Concentrations in blood and liver also increased,  but
the rate of accumulation slowed by the third day.
     Evidence of PCE accumulation is also available from examinations of the
concentration of PCE in exhaled air.  Nhen volunteers were exposed to
100 ppmv, 7 h/d for 5 d, the concentration of PCE in expired air increased
with each exposure (Hake and Stewart, 1977).  This suggests that the uptake .
capacity of tissues had not been reached by 5 d of 7-h exposures and that
consequently, body burden was still increasing with each exposure.

METABOLISM AND ELIMINATION

     Among the most important enzyme systems for metabolism of toxic substances
are the mixed-function oxygenases (MFO).  These enzymes are concentrated in
the liver, kidneys, lungs, and skin, and are present in other tissues as well.
Mixed-function oxygenases catalyze the addition of oxygen to compounds, which
facilitates their excretion from the body.  Oxidation of a compound can
function as a mechanism of detoxification or can transform it to a reactive
(toxic) substance.
     Evidence that MFO's are directly involved in the metabolism of PCE comes
from the work of several investigators.  Moslen et a].. (1977) demonstrated tha-
a number of substances, including phenobarbital (PBT) and Aroclor 1254, induce'
hepatic MFO.  Pretreatment of rats with either of these compounds, followed by
administration of PCE, increased the metabolism of PCE five to seven times ove.
controls.  Costa and Ivanetich (1980) showed that substances that inhibit
cytochrome P450, a component of MFO's, also inhibit metabolism of PCE in rats.
Induction of cytochrome P450 by PBT or pregneno1one-16a-carbonitrile
increased the metabolism of PCE.  PCE was also shown to bind to the active
site of P450 in rat hepatic microsomes.
     The first step in metabolism of PCE  is thought to be transformation to ar
epoxide by the MFO, although this  epoxide has never been isolated in vivo
(Bonse et §1.,  1975; Greim  et a]..,  1975).   The epoxide of PCE rearranges
                                        20

-------
  spontaneously with migration  of  a  chlorine  to  form trichlorqacetyl  chloride
  and,  ultimately,  trichloroacetic acid  (Moslen  et  a_L,  1977;  Leibman and Ortiz,
  1977;  Reichert,  1983).   Trichloroethanol  has also been  reported  as  a
  metabolite; however, a pathway has  not  been proposed  that  explains  its
  formation  (Ikeda  and Ohtsuji, 1972;  Ikeda et a_L,  1972; Monster  et  a_L, 1933;
  Koppel et  a]...  1985).
      As yet, it  is not clear  if conjugation (phase  II)  reactions are involved
  in the metabolism of PCE (Pegg et §1.,  1979; Lafuente and Mallol, 1986).
  Although not consistently reported,  the production  of oxalic acid and CO  as
  metabolites of PCE has been documented  in rodents (Yllner, 1961; Pegg et a_L,
  1979; Schumann et §1., 1980).  Their formation argues for a second  and   ~~  '
  possibly minor pathway of oxidative metabolism.
      It has been proposed that oxalic acid and C02 are formed as end products
  of a metabolic path that also includes ep.oxide formation as the first step
  (Pegg et a].. 1979).  In this scheme, chloroethylene glycol is formed from the
  epoxide by the action of epoxide hydrolase/  This reaction is followed by
 hydrolysis to oxalic acid and/or decarboxylation to CO- and possibly formic
 acid.
      Chloride,  dichloroacetic acid,and ethylene glycol have also been reported
 as urinary metabolites of PCE in rodents (Daniel,  1963;  Yllner,  1961;
 Oimitrieva, 1967).  It appears that these substances are minor metabolites,
 and little is  known about their formation.   Figure 3-1 shows  the structure of
 metabolites as  well  as the  proposed metabolic  pathways.
      There are  some similarities  in the urinary metabolites produced by humans
.and rodents following exposure to PCE.   Trichloroacetic  acid  appears to be the
 principal  metabolite  formed,  regardless  of species.  Tables  3-3  and  3-4  .list  the
 metabolites identified in humans,  mice,  and  rats,  as well  as  the  conditions  of
 exposure.
      Yllner (196V)  was  the  first  to analyze  urinary metabolites  of  PCE  in  mice.
 Although  18% of  the  radiolabeled  compound  was not  accounted for,  52% of the
 portion metabolized was recovered as  trichloroacetic acid  (TCA),  11% as oxalic
 acid,  and  a trace  amount  as dichloroacetic acid.   Daniel  fed  labeled PCE to
 rats  and found TCA  (0.6%) and  inorganic  chloride as  the  only metabolites
 (Daniel, 1963).   Total urinary metabolite production in  rats was  measured  by
 Moslen  et a].. (1977).  They did not  identify the actual  metabolites,' with  the
 exception of TCA  (which was the major metabolite produced).  Pegg et ah (1979)
                                       21

-------
          Cl
              c=c.
                     ci
          Tetrach loroethy lena
                             Mixed-function
                               oxidases
                              (02, NADPH)
CL
                   Cl
                                                           Cl
                                              Tatrachloroethylerw oxide
                                         Epoxide
                                        hydrolu*
                  Hydrolysis
  Hydrolysis
            CO,   -r-
          Carbon
          dioxide
                    H-
                           OH
                                     HO
                                                OH
                    Formic Kid
Oxalic acid
                      Chloride
                      migration
                                                                   Cl
                                                               Cl
                                                                         \
                                                                           Cl
                                                                    Cl
                                                             Trichlofoaeetyl chloride
                                                                 Hydrolysis
                                                               Cl
                                                                    Cl
                            c—c
                                                                           OH
                                                                    Cl
                                                              Trichloroacatic acid
Figure  3-1.  Metabolic pathways of PCE (brackets denote compounds that have  not
been  Isolated  In  vivo) (Daniel, 1963; Pegg  et'alL ,  1979; Costa and  Ivanetich,
1980).   Although  trlchloroethanol, dlchloroacetic  add, ethylene glycol, and a
thioether have  been Identified as  urinary  metabolites, the  route(s)  by which
each  Is formed  has not been characterized.                   ,
                                            22

-------





































VI
C
CD
O
^

C
"~
C
o
4-1
u
•a
•o
^
0>
4->
•f~
*™-
O
.0
ra
CD
UJ
a.


•
i
m
OJ
2
re
t—






a;
u
c .
CD
w
OJ
CD
Q£







O
CU
c u
O 3

re a.
U X
3 CU
a


<*- a>
0 u
3
CU vi
-M O
3 a.
0 x

c
0
4n» CU
rO VI
i- O
4-» -a
c
CD !_
U O
o •
CJ




OJ Q i—
*•* ^rf •
..- re
U • £
CU O •«-
a. s c
oo --/ /a


0)
4-1
•^
»••
O
ra
cu
3E








_
WO
Ol
*—

t 	

c
»—
^•>
*~








-C

CM
c
o

-M
ITS
*—
ITS
J=
c





01

1

*—




in

3)
VI
3
i

u
£
cu
u
re
0
o
,—
u -a
•j_ u
I— re




WO
CJ1
m
WO
O^ re
1— >
CD
• •*-•
CU 4-1
•^ »f-»
C £
^ •*••
O O
4>
^
3 T3
VI
O rn
a.
x u
 4-1
<£ 2












CM
2

— "
3
-M
i—
O

•o
c
ITS
re
•a
CU
^^
, 1— 1








^^

co
c
o

4->
rO
*—
re
fm
• C




f
^
£
a.
a.
o
o





CO



4-1
a:












<^j
2

— "
3
^J

O

•a
c
ra
ra
•o
CU
^£
1—1
cu
i.
3
vt^
o
0.
X
cu

CU
"01
c
•^*
00








Q.


Ol
^
f^



CO
r—
CM




p—



4^
•a












CM
2

— "
3
• -M
r-
O

•o
c
ra
re
•o
cu
J^
*—4








^^

CO
c
o

4-1

O ra
z ce













-
-M
cu
re
—
_
o

"O
c
re
c
cu in
J2 OO
3 cy>
03 •—

'"^3

vO
o

^
4!
•a

"i




cu
Ol
ITS

re



Ol

•^^
Ol
os
-MO
o
o o
CM CM


O cu
4-> VI
o
T -a
w <-»
O) CU •—
vi a. cu
3 >
O ^- cu
Z CM —












CM
2

•*•••
•f"^
3
VI
^.
O

•b
c

rO
•a
cu
^*
^~"








.c

CO
e
.2

•^
fV>
!•"»
rtS
.e
c





^
£
Q.
Q.
O
o
CM




CO
>~'


4-1
re
cu

'o
c
re
JS

cu
o
o
.£:
u
•f^
H—
CM
2

.,-*
3
VI
.C
0

•^
c
re
re
•a
cu
•^y
H-l








^

03
e
o

4-1
ITS

ITS

C


*


^
£
a.
a.
o
o
CM




^
*^" '
CU
(/I
3
i












CM CM
P** P^
O^ Crt
"~ "~
•r- — "
3 3
VI VI
4^ 4-*
UZ <~
0 0

•O T3
C C
re re
re re

CD CU

I— 1 1— 1
CU CD

3 3
*x >— '
CU
VI
4-> 3
rO O












23

-------


















































s"%
•a
ai
c
c
o
0


•
en
i
en

at

«— \
itj
H-



















(*-
O

C
O

Id
u
3
a


o

ai
4-1
3
O
OS

o
4-i
rd
+j
C
ai
u
e

t/i U-
O) O
°u •
0) O
a. c
CO >-»




















Ol
u
c
ai
i^
01
<+-
0)
ce








ai
t-
23
i/i
a
X
Ol



u
3
VI
O
a.
X




•^•J
•f—
^"
o
.a
rd
4*1
d)
3E








,_
VO
cn
»••—

«k
U
0)
c

^H»
*"










f

CM
c

!J3
rd
i^—
fa
_£J
C
I—I



en

C3J
£
en


^
ai
v>
3





•a

'u


u

P»
rd
X
O



VO
o»
^^
•
rd
>
OJ
*^>
•M
•f"
E

5

•a

en

u
0
<*_

•a
^2

in
c
O
^
rd
^^
rd
j^
C






_J
*«*,,
i1

0
^^


4^
ra
ae


















S S
•— <—
. .
, •
f«>l P^
rdl rd

4->| 4-1
ail at
O) CT
O) O)
03 
C fl
« (J
> en
t ^
<=L f
8 o
vo o
in
^
^D I*
^3
o
en en



4-1 4->
rd rd


















r-
•~
.
•
*—
rd

4-1
Ol
en
en
Ol
a.










^^

VO
c
o
'^3
rd
i—
rd
JZ
c

>
t
a.
o
O
VO

j 	
O
o
en



^j
a:


Ol
•a
X
O

•5

c
o
J3
t»
rd
O



O^
•~
.
•
^— I
rdl

4>j I
0)|
en
en
o>
a.
Ol
=
i/l
0
a.
X
OI

, ai
en
c

CO



Ol
en
rd

rd

en

i1
O
o


o

en'
^


4-1
rd
ae















0 0
so co
O"l OA
-
"rdl "rd

4J| 4J
Oil O)

C C
c c
i i

.C -C
u u

01
e 3
I/i
O
a.
x
Ol

0)

J= C

vo J5
c
o
4^
rd 01

rd rd
JC >
C rd
N»l (^
> en
E ^
a. -^.
a. ' en
8 o
VO O
IT)
^
0 s-
o
en en
^ «-*
O) Ol
 
3 3





















_
VO
O^
»—

•
s_
Ol
c

im
*" : ;










^gj

CM
=
o
*4j
rd
^M
r«
^^
c
1— 1



en
. ,^^
m
P5
en
•
IT)
^ .
01
en
3
• .2



u
rd
O
Lm
Q
f^
.s -a
• U .t-
1 *^ CJ
O rd



VO
cn

«
rd
^
Ol
•^
u
•f«
E

a

•a

en

VJB
0
*^»

•a
iG

m
e
0
4-1
rd
fmm
• rd
^c
C
MM





.J
tk ^^
O)
ID

^^


4^
rd
cc

^
o
u
,_*
en

01
c
O)
1^—
>•)
^^
4-1
LU





en
vo
•CTl
— —

*
Ol

C
rd
O
-.*
3
t/1
o
o.
X
Ol

Ol
en
c
•^
00





*••» .
rt3
^
o


en
^
en
E

0
2
vo
***


4-1
rd
ae







Ol
T3
•i^
^
O
»— "
jC»
O
24

-------


t





c
cu
ai
^4i—
cu
ce.


(4-
O
0)
C i-
O 3
4J O
IQ a.
S_ X
3 ai
a

'o-u
3
O) vi
•M O
3 ex
0 x
Q£  . .,—
•<-• o >
"o Z;o
ra • .r- -
«u
LU
<_>'
Q.  v» C C CU
CU 3 (9 t9 ^3
O C
•a 19 «M -o ~o c
-^ O en _^ ^ ai
>— I CO •— l— i I— I U.
•
Si>*
O)
^" co ^o co
.
CN CS4
r^s ^
wi C^
§ao GO ao in
^^ c^ ^n GO ^j * .,
£ - ._ _ ^ ^ ._
3 ...
o ...
X — j — —

a
S- V. i_ U
cu cu a> cu •—
2 ? '"
* * M aC
; « ° S
J • * c 1
cu «
» '
4J 4J4-I4JCU  c c c a. cucu _i _ ,_,
cu
^^ ^^ ' '-• *

• — cu cu cu
•a . — « 4-> 4-1

vo _= cu .- 'i ^
v» 3 i_ i. GO
j= ^o.^c ^c4->
GO **O QJ C3 ^i— ' O s**1' <~~
c c c c e
o o o o o
4-1 4-1 4-1 4-1 4J
J2 ^2 ^2 j2 ^2
^ rtS /4 rti /ti
•C -C .C J= .C
= C C C C
•— ' •— • . H- 1 P— 1 1— 1
i > i i-
a. E a. a.
1 ^ 'I i J
o .S o 2 2
_ .0 o o
O en o O O
>— m «M csj —






m to ^ vo «g-
QO (43 CM



U

^.4
cu .
o
O
o
(—
u -a
•^ »^™
^ ^ ^: 3
-^ 4-1 j_i
a, ^ ^ -J S ' ^ = =
— j=^!ca. "" * *
Cjl<-s - x ij **
C CU ^ *J* ^ CU *O ^, ^
"- u to en un s ' ' 's •
v» 3 « CU . £ £
s_»v» o O O— "O >>eu >»cu
O -M 4-< 4-> CT "^ —4-1 —4-J
-C ex u c -C •*- e «— cz
^j P^^ -^ • _l| ^^ ^^ "^ ?^ •• •
^rcu m m m  > o
O a. ex ex o
0. 4->
O r» m
r^ r^. ^ . (D oo






vo M en en -












c , c c
o o o
'•*J 4J 4J
^S ^0 ~ ^fl
^^ ^i» ^_
ro n3 TO
c c c
S > CL
Si. . °-
o * 8
5 S ^
o o ' 5
4-> 4-1
00 0
— CM CM






IT) «5J- VO
<^ (43



—
O
c
ra
£
cu
O -
0
•—
u
b_
25

-------































































•o
O)
3
C

C
O
u


>
^»
1
m

O)

.0
* flj
O 1-
3
CD v>
4-» O
3 a.
O X
cr ai



c
O
•*••
4>j ^
id v>
U 0
4-> -a
c
oj v.
u o
1

vt
"id
<*- 3
o ~o

O *>
•5
C
•*••





O)
4-1
'""
f^
J=]

•a  3 »
>t OJ ^— ' t/1 O
~- 4-> O 4J
••- C -CO.

o >-* ^ d> fi




e c c
o O O

4V 4-> , 4-1
(& ^O ^0
^MB ^H» *H*
^9 ^S ^9
-C J= -C
c c c
1— 1 1— 1 I-H


e >
Q. m
a. , a.
a.
0
0 O

4J i_ e
o a
8 o »





«T <43 ~ LO
CM




















^

mm
03 CO —
O1O% O
^d
^l •—! ^
rdl id| "O
c
4J| 4->| id
^1 w|
O)
W ^ 4^
O)  3 VO
e c «- ao
3E X — 1 •—

22 2

4= , 4= -a

^T ^T W^
un un
*
O O T3
4-> 4-> •»«
4=

m m 03




e c c
O O O

4-14-1 4-1
id id id
^" f«^ ^"»
• & rG . ^
^W ^M -C
c e c
1— 1 1— 1 1—4



>
g
a.
a.
rd id O
> > m
I. i. o
a. a. 4-1
r-» m LO
^r un -—





01 01 ^



a>
^
•»-
id
^
•»i—
V.
O)
"^
u
ai
4=
^•rf

CD
id
u
O)
^
id
I
•o
O)
4-1
^**
•£•*
OJ
2
1

f-
4-1

4-»
C
0)
at
^
a.
a;
^
VI
3)
3
^™
^a
>

rd

26

-------
and QimHrieva (1967) Identified oxalic acid as the primary product of PCS
metabolism in rats.  Carbon dioxide has been recovered as a metabolite of PCE
in mice (Schumann et ah  (1980) and in rats (Pegg et ah, 1979).
     In contrast to rodents, TCA and trichloroethanol are the only metabolites
of PCE that have repeatedly been Identified In humans.  In some instances,
identification'of trichloroethanol  has been by the Fujiwara reaction (Ikeda
et a_L, 1972; Ikfda and Ohtsujl, 1972).  The results of this test are
qualitative and the accuracy is questionable.   However, Monster et a_K (1983)
and Koppel et a],. (1985)  determined the presence of trichloroethanol by gas
chromatograpny.  Still  other studies have failed to detect trichloroethanol;
therefore, it 1s far from clear under what circumstances this substance is
formed (Monster el: aK, 1979; Fernandez et a_K, 1976).  Measurements of human-
metabolite production are confounded by the fact that carbon dioxide, oxalic
acid, and chlorine are normal products of metabolism.  Their presence as
products of PCE metabolism could be quantified only by the administration of
radiolabeled PCE—a procedure which has not been undertaken with  humans.
     Lafuente and Mallol  (1986) have reported  the presence of a thioether
derivative of PCE in the urine of women occupationally exposed to PCE.  A
gradual Increase in thioether production was observed over the course of a
week.  This Increase appears to have been associated with continued exposure
to PCE (with concomitant accumulation) over the work week.  However, measured
amounts of thioether in exposed women were reported not to be statistically
different from levels found in non-exposed individuals.
     The recent Identification of thioether derivatives in humans exposed to
PCE, coupled with .problems associated with accurate and complete  Identification
of metabolites, indicates that not all human metabolites of PCE have been
accounted for.  It is possible that characterization of PCE metabolites in
rodents is incomplete as  well, since no complete mass-balance studies have been
conducted.  Although the major metabolites recovered in all  species are
presumably produced by the same enzymes, many  uncertainties remain.  Based on
present Information, humans and rodents appear to metabolize PCE  in
qualitatively similar ways.
     PCE is eliminated from the body by two major processes; metabolism
followed by excretion of urinary metabolites,  and pulmonary elimination of
unchanged PCE.  Although some PCE may be eliminated through the skin,.
preliminary measurements  indicate that this is a minor route, at  least in
                                       27

-------
humans (Bolanowska and Golacka, 1972).  Most of the PCE systemically absorbed
under experimental conditions (e.g., in mice, rats and humans) was eliminated
unchanged In expired air, so that metabolic degradation and subsequent
elimination are thought to account for only a fraction of absorbed PCE.
Experimental data reviewed below indicate that metabolism is dose-dependent
and saturable, and that the amount metabolized appears to be species-dependent
as wel1.
     When the production of urinary metabolites was measured 1n mice given
1.3 mg/g  of PCE for two hours, Yllner (1961) found that only 21 of an Inhaled
dose was  excreted by this route.  Seventy, percent of the parent compound was
recovered in expired air.  Pegg et a_K (1979) compared the metabolism of PCE
in Sprague-Dawley rats at different doses and routes of exposure.  PCE was
administered by gavage (1 or 500 mg/kg) or by inhalation (10 or 600 ppmv).  The
primary route of elimination of PCE was through the lungs as the unmetabolized
parent compound.  Urinary excretion of metabolites accounted for the majority
of the remaining PCE.  The percentage of PCE metabolized was dose dependent:
elimination of unmetabolized PCE in expired air following oral exposure
Increased from 721 after 1 mg/kg, to 901 after a dose of 500 mg/kg.  there was
a corresponding decrease in metabolites from 28 to 101.  An analogous pattern
was seen after inhalation exposure.  At 10 ppmv, 681 of the dose was
eliminated  unchanged through the lungs and 321 was recovered as metabolites.
Treatment at 600 ppmv caused an increase in pulmonary elimination to 881 of
the dose, with a concomitant decrease in metabolites to 121.  Pulmonary
elimination of PCE was linear  and had a half-life of approximately 7 h.  The
half-life was Independent of'dose or route of administration.  The half-life
of PCE in blood was  6 or 7 h,  after oral or  inhalation exposure,  respectively.
      The pharmacpkinetlcs of PCE metabolism  in Sprague-Oawley rats and B6C3F1
mice  were studied by Schumann  et aj..  (1980).  Rats were given a  single oral
dose  of 500 mg/kg or were exposed to  10 ppmv  14C-tetrachloroethylene by
inhalation. The  authors  reported that at  10 ppmv  the major route of
elimination was excretion of  unmetabolized  PCE  1n  expired  air, although
supporting  data were not provided.  Some PCE was  metabolized,  and metabolites
were  recovered  in the  urine.   After an oral  dose  of  500 mg/kg of labeled^PCE,
radioactivity  was measured  in  the expired  air,  urine,  feces,  and carcass.   The
relative  importance of each  route in  the elmination  of  PCE was not  discussed.
                                        28

-------
     The elimination of PCE by B6C3F1 mice differed, depending on the dose, and
possibly on the route of administration.  Metabolism, with urinary excretion
of PCE, was the primary route of elimination after inhalation exposure at
10 ppm; 62.51 of the dose was recovered as urinary metabolites and only 121
was excreted through the lungs.  Eighty-three percent of a single oral dose
(500 ing/kg) was eliminated through the lungs, while 10.31 appeared as urinary
metabolites (Schumann et aj.., 1980).
     The fate of PCE 1n Sprague-Dawley rats fed PCE in their drinking water
was reported by Frantz and Watanabe (1983).  Animals were given PCE in a
saturated solution over a 12-h period, resulting 1n an average dose of
8.1 mg/kg.  Treatment was followed by a 72-h observation period prior to
sacrifice.  Most (87.91) of the PCE was eliminated unmetabolized via the lungs.
Although urinary excretion was the second largest route of elimination, the
amount metabolized was .relatively small (7.21).  The half-life of pulmonary
elimination was 7.1 h; this 1s nearly identical to the value determined by
Pegg et a].. (1979) after oral doses of PCE were given to rats.
     Mitoma et aj.. (1985) studied the metabolic disposition of PCE in Osborne-
Mendel rats and 86C3F1 mice using a follow-up period of onl^y 48 h.
Substantial differences between species in the amount of PCE eliminated
unchanged in expired air were noted.  There were also differences in the
percentage-of dose metabolized.  Rats eliminated 791 through
»the1r  lungs and metabolized 51 of a 1000 mg/kg dose.  Pulmonary elimination by
mice (of a 900 mg/kg dose) was 57.51, while 221 was metabolized.  The actual
amount of PCE metabolized (measured as mmol/kg) also differed by species.  Mice
metabolized approximately four times as much of the total dose of PCE as rats.
However, in both species, as the dose was quadrupled, the amount metabolized
only increased about 2.5-fold.  These data suggest that metabolism approaches
saturation at high doses in both species.
     Buben and 0*Flaherty (1985) also demonstrated that oxidative metabolism
of PCE in Swiss mice is a saturable process.  Animals received 0, 20, 100, 200,
500, 1000, 1500, or 2000 mg/kg-d of PCE by gavage and were followed for 72 h
post exposure.  TCA was measured as an index of PCE-metabol1te production; the
estimated maximum  rate of urinary metabolite excretion was 136 mg/kg-d.  As
the dose Increased, the percentage of the dose metabolized decreased.  At  the
lowest doses, approximately 251 was metabolized; this decreased to 51 at
                                       29

-------
the highest doses.   Tables.3-5 and 3-6 summarize data on the metabolism of PCE
in rats and mice, respectively.  Table 3-7 lists some pharmacologlc constants
of PCE 1n rodents.
     The data reviewed above Indicate that metabolism of PCE in both rats and
mice displays saturable kinetics.  The percentage of dose metabolized decreases
as the dose 1s Increased until the amount metabolized is no longer a function
of the dose (a zero-order reaction).  The extent of metabolism of PCE is
apparently species-dependent.  Rats consistently metabolize a relatively small
amount of PCE regardless of the route of administration.  Mice typically
metabolize a greater percentage of a dose than rats.
     Measurements of PCE metabolism 1n humans have many uncertainties
associated with them.  Total tr1chlorinated metabolites in urine have usually
been determined colorimetrically and therefore are difficult to evaluate
quantitatively.  By this method, production of chlorinated metabolites is
equated with the amount metabolized.  This approach does not account for the
possibility that some metabolites may be produced that are not chlorinated,
such as C02, the thioether, and oxalic acid.
     Studies qf  PCE metabolism 1n humans have consistently identified  TCA  as
the principal metabolite.   TCA production has been measured to estimate  the
extent of metabolism of PCE and  also  to characterize  the kinetics of urinary
elimination.  Volunteers exposed  to 87 ppmv of  PCE  for  3 h excreted about
0.40 mg/h of TCA (Ogata et  ajk,  1971).  Monster  et  a]..  (1979) analyzed  the TCA
content of  blood and urine  from volunteers exposed  to 72 or 144 ppmv of  PCE for
4 h.   The mean production of  TCA (over a  70-h period) was  6.0 and  11.0 mg.
Blood  levels of  TCA  increased over  the course of the  experiment and  continued
to rise  until  about  20  h after the  end of exposure.   TCA was  eliminated  from
blood  by first-order processes;  the half-life of elimination  was  65  to 90  h.
Urinary  elimination  of  TCA  followed the  disappearance of TCA  from blood.
 Fernandez et a].. (1976)  also measured excretion of  TCA  from individuals  exposec
 to 150 ppmv for  8  h.  Over  a 72-h collection  period,  the  average  amount of TCA
 produced was 25  mg.   This  1s equivalent  to about 0.34 mg  of TCA per hour.
 Ikeda (1977) and Ikeda  and  Imamura (1973) qualitativelymeasured  the half-life
 of urinary trichloro compounds and calculated thejnean  half-life  to be 144 h
 (range of 123 to 190 h).   The lengthy half-life may be due to the continued
 formation of metabolites,from PCE mobilized from tissues where it has
 accumulated.
                                        30

-------
                  e
                  41
                  u
                  41
                  H-
                  41,
                  at
                  VI
                  Ol
«
k
41
C

41
4)
O

a
44
41
a
a

VI
I/I
         1
          41
          U
               «  e
               44  3

               i:-
               4i  -a

               44  §,
               c  e
               41  14
               U  A
               U  U
               4i  e
               a.  3
                  u
                   VI
                   41
               44   a

               ,23
                   VI

                   41
               £
 I/I

2
               a.  «
               x
               41
     £  i
     3  ••-

     S  «

     II
     44
     14 -a
     W   41
     44   VI
     e   a
     4i  -a
     u

     S   £
     u
                  ex
3
                            £
                             in
                  m

                  >*3
    2
 41  3
r-  VI
 01  a
 e  a.

£  S
                             Ol

s
^ •»
-
•s
td

s
41




CO






CM
PO







^

CO










(SJ
^^









_g

^D



^^
Ol
JC
^^

1

CM
2 3









a

44 , .
•2
45

1
rtl • ' ^ft ^ft
2 2 2
5 - -
1 ^ '. *
m rtj (jj
4ta*
C CO O) O)
rtj CD - O) O)
k ty\ 4u - 4V
u. — a. a.



§. CO 0
CO ^^






CM CM • O
""""""







CM \O

r. vo *T



F'






CM . CM CM
r» . p»'-. r»






41
« 3
«— VI
? S.
CM — X





^
01
-* Ol

j*> §. s" "oS

0» CM
• o ci a
co vo — • in



^^
u
41

- s
O>
— e
.x a
e —

•a —
"- o

1
-
•si

m

2

z



r>






' in .








•T

CM








.S

CM



Iff
x •

^y

Jt
X
•a

in





O)


i1

a
o
o













2
o

-



















^
41

•*•
•3

44
4l
g'

VI

X

44
.C
44

41
VI
a


•o
41
w
41

a

2

, *a

^J
e
41
U
U
41
0.

41


e
4)
VI
41

O.
41
U

I/I
41
3
"?
. 41
VI
41
fi-
(0
Ol
S
CM
a
*£
Ol
41
X
41
Ol
U
41
C
e
•a"
"^
"a
CO
0
o
41
44
U
01
!c
41
U
.a

- *
Ol
• C
'i
3
VI
VI
TS


.
i1

c

41
I/I
a
•o

41
g*
44

VI
41
U
3 .
VI
e
a.
X
41

e
a

^
2
.S "5

- k •*«
a e
U. 19
* i
a.






















•
e
o
44
41
U
X
41

14*
a

VI
41
3
O



f*m
(Q



U
^B

^

41
O
U
41
U

41
44

^M
a

flj

i

44
e
41
VI
41
U
a.
41


VI
41
3
H
41
VI
41
J=
U
                                                                  31

-------






























gj
u

n

e
41
C
«
*><

At
41
2
0
u
2

4)
«J

V4-
o

C
O

£
O
0.
VI
•o
I
r»
.5
PMB
o
(U
•
^3
f*i
4)
i

41
U
C
01
u
01
**•
oc


VI
T3 OI
oi e
At 3
s -
"* 2
•5 '>
*« tj
01
g 1
U -C
u u
 a
N O J3
•"- AI
5 2
i»
1 *
VI
e x AI
41 U u-
u m •—
U CO
4i ••- -a
a. i. *
=> |

qj
i. -a
1 3 0
At I/I —
M a v.
O a. 0)
a. x a.


4i e
U 0
3 —
VI At
i.2
X 3
ui -a


§
*£*
2%,
At M
e a
4i -a
I»


O
*«
i
et


f 1 •'•*?•
• I
"5 • «

Ite ^M ^^

O H VO O
4» O\
"c c c
1
1 * 1 1 1 ->
3 ^^ U ^™ 3 Gft
CO - «« > OB -






S ^ 23
Z ^* n» Z








0
^ S i 2
V





in m o in
CM CM •

i W >«,
•a At ^ -o

M V0 CM in
j
^^
a> jf 9 Jt
I |y s* >
o a CM • a
CM i— •-• •"• CM



e e
o o
Al At
1— F-
1 ' 1 11
>>
Al
U
01
i

lA.

O
•Q
e
tj
e
oi m
JS CO

CO —






2









$





VO

CM





A

0


-5

^o


Jtf
>»
•o

m


at
JC
>»







2
e
o
CO
<*V in
•— CO
•~

*fll •

«l *
G ^(
e
u —
W X





vo in
CM r«
co in








^" CM
^ CM
— CM





<*» rr

O »T
*•• ^




' Si J=

CM CO


J<

^y
01
U
,41 3 JC
^» 0 "X

C CV ^


-------




















































































3
a
e
•^
e
a



•
^D
d,
41

^
^
^















_


•Q
4)

1
"5

p^
41
^J
C
41
U
W
41
a.





•a *«
41 •<•>
N 0
•» I—
f^
e
.a
nj
1

e >i
41 U
U  V»
vi e
o a.
a. x
41


£
3
VI
a
X
UJ

§


4
i.
•

U
4t

*
e
13
•o
e


•
•o
-s
WB

IT)
5
o

o
(4.
a

.2
nj
u

o»
!c

19
41
' U
A
13
g>

i
VI
v>


•o
41


3 ~
U


U

VI
M
X

Ol
-N
?
e


VI
a
•o
5
4^

•
^^
VI
"v
u.
3
I/I

X
41

e
.2 •
^rf ^*
•i— S
ns —
^ e

«^
^ I
u.
91
10
O
CM



















































.
^^
e
o

VI
41

.«•

a
^

^^
n

«
c
.^

3
a
u
41

a
u
41
U

e
41
VI
£

S"
u
VI
41
a

^

41
i/l
41

ja















41
e
41
"5,
^j
4rf
4)
a
a
2
u

u
41

•-
^3
41
. U
•^
X
4>


a
e
a

4J
Q.
41
U
X
41
41

^
4^

X

•
4)
U
a
VI
a
Q.
X
41


a

vi
4)
a
a
. w
^•B
/Q

§
*•
t.
41

a
u
41


e
41
VI
£
a.
41
V
VI
41
3

^

4)
VI
41
I—

U






















































































41

jg
ifl
131
(3
«

^J
a
c
•'i

Q


33

-------





















,
VI

e
ei
•a
o
k
e
41
e

*>•

*j
o
JZ
u
1.
*l
<*»
o

VI
1
VI
o
u

u

'en
a
o
u
1
m
£
r»
A
ti
1


41
U
2
41
41



«> e
k o
3 •-
VI *J
O 19
a. k
X 3
Ul Tl
I
3 *
2 S
«j -a
41 k
U 0
a


«

3
a
oc






*^
VI **
.2 *
a. s




3

>











U
41
I
^ft ^ft 1J1
^* ^^ ^^
o o^ o
*"" *~
•a 3 3
tJ "S3 "Si

en en en
en en en
41 4) 4)
a. a- a.

41 , 41
u k
41 3 41 3
•—VI f— VI
en o en o
CQ. j= e e. .e
— X •«• X

•x S.
€ , I =
en • o \e
8 S. in
^W «
k O
a o o -a
o o co
in « *.(»*.
I , 1
4J e *»


— fl i— <•
>a -c m .c
k c k e
a N* o —



.. ._•


en en en
o o o
*> in *• in -w in
* CM It CM 19 CM





S
0.
f * * *
vo r» r>. -~




19 19

^ ^ *
f- f- Q
S § S
5 '5 3
C C r-
•5 i2 9^
1 "« !s
1*
a a i CM
•2 ^ "si
S S a. *T

Cft ^^
^ft ^ft

•d -a
44 ^M

en en
en en
d! 2

41
k
41 3
r— VI
en a
CO. J=
•^ X
VI 41 V0

en
1 1

S S
lO vO
S
JJ
iq

^**' iQ
<9 -C
C e
0 «






en en
o o
•M m *• in
fl CM « CM
a* •— ex —





"L "L
CM O
e d






*? C
e e
3 3
VI VI
s ' e
a e
w w
1 I
cr'£ cr'i
J» 8 -"8
s 3 * 3
s " s *
03'— CO v
34
ON
2

j
'•a •

s
41
a.

41
k
41 3
1— VI
en o
e a.
— X
> «4
k *
<9 e
||

41
<«
U.
O
•a
S
e
.0 03
3 91
CO —
X
vfl
• •
jf
X

•a
in
1
e '
a
o
CM
O

a
CM




^»
19
k
0






4>
VI
' i




13
£

SP

CO



Vto
O

fM -O
41 41
1 = 3
ill
• -

2 >» * 2
i k •«
(O 41 V
x e • CM
3— C
> s 5--.S
.
2
j
^
03
•a
e
19
k
41
VI





^J

CM
"c
e
i-s S
a. ••- c
*> 41
o — u
ace
o ••• e
.— ~- u
S
'£
19

/t)
J=
e'


^ .
en
CM

e

U CM
ae ~-



en

1
^
i

p*
V





u
'o
3rm +* -^
8 e
'x ••-
3 21 *
•i .- e
> > U

1
J
'o
•o
e
i>
k
41
VI
'• iZ1'





^J

CM

i
a
o
a
o
=
'3
«

4
&
e



en
a
CM

O
« 0
oe CM






»

U1
a






S S
1 1--
.^
3 *"
cr c
41 »
VI
er e
41 O
^ U


-------

u
e
41
b
Vto

2 S
3 —
Is
x a
ui -a
e -
'a
* vt
i> a
^j ^9
e
41 k
u a
§


4)

at

VI *»
.2 i,
u ••-
,*i

S
i






4)
*>
i
I.
2
C ^
^" ^n *
•3
i -s ^

a 73 §
33 1
«l 41 U


A
• ^™
^9 O \O

* .

| 1 i

2 1 2
S S §
'-S '^ *^
* * «
•s -<= -2
£ 5 '- -5


0 0 S
*• vrt *• ui a
<« PH <« CM O
OC — 06 — £

o> e> S
||f
* 2 "
a? f> >—




! ! 1
J J J
•S1 -S" -o*
s s i


























e
a

e
'i

41
1
Y
JZ
41
5
VI
/H

35

-------
    •Measurements of TCA production suggest that human metabolism of PCS is
limited.   For example, the values published by Monster et a_[. (1979) (6.0 and
11.0 mg TCA after a 4-h exposure to 72 or 144 ppmv PCE) represent only about
1  to 2% of the estimated absorbed dose.  Ogata et a].. (1971) reported that TCA
excretion was equivalent to 1.8% of the retained PCE, and that total excretion,
of organic chloride accounted for only 2.8X of the retained dose.  These data
are in agreement with those of Fernandez et a_L (1976), who estimated that
13SO mg of PCE would be absorbed after" exposure to 150 ppmv PCE for 8 h.  The
25 mg of TCA produced corresponds to metabolism of 1.85% of the absorbed dose.
Ikeda.et aj.. (1972) and Ohtsuki et a]..- (1983) have estimated that only about
2% of an 8-h exposure to 50 ppmv PCE would be metabolized and that 381 would
be eliminated through the lungs unchanged .by the end of the exposure period,
the remaining 60% of the inhaled dose  was hypothesized to be stored in the
body and available for subsequent metabolism and/or pulmonary elimination.
Ohtsuki et a]..-(1983) found that human urinary-metabolite production-did no
appear to be linearly related to exposure concentration.  A graph of total
trichloro compounds (from urine) against PCE concentrations in air  showed  tr :
metabolite production appeared to be dose-dependent, leveling off a't
approximately 400 ppmv PCE -(8-h exposure), suggesting metabolic  saturation;
however, no  statistical test of departure from linearity was performed  in  this
study, and a questionable nonlinear model was assumed.
      In contrast to the findings just  reviewed, Bolanowska and Golacka  (1972)
proposed that at steady state approximately  62% of respired PCE  is  metabolized
(based on their measurements of respiratory  PCE retention discussed above).
This  conclusion  is contradicted by other studies of  PCE  uptake and  metabolism
in humans (Ogata et aj...  1971;  Ikeda  et aj...  1972; Fernandez  et  aj..,  1976;
Ikeda, 1977; Monster  et al..,  1979; Ohtsuki  et  aj.., 1983).
      A long  period of time  is  necessary for  pulmonary  elimination of
unmetabolized  PCE.  Stewart and  co-workers  (Stewart  et  aj..  1970;  Hake  and
Stewart,  1977)  have analyzed the  pulmonary  excretion of PCE  following
experimental human  exposures and  found that  it is  biphasic.   Initially,
elimination  is  rapid  but  the second  phase  is prolonged,  with a half-life of
approximately 65 h.   Monster et aj..  (1979)  determined that  human pulmonary
elimination of PCE has three different phases, with  half-lives of 12 to 16 h,
30 to 40 h,  and 55 to 50  h, respectively.   Fernandez et aj..  (1976)  note that
humans exposed to 100 ppmv  for 8 h will  require about 2 wk to eliminate PCE.
                                        36

-------
                           4.  TOXIC EFFECTS IN ANIMALS

      Estimates of human-health risks resulting from exposure to a toxic
 substance are frequently based on an assessment of animal dose-response data
 because specific human data are often Inadequate for this purpose.   In this
 section, we review studies of the toxldty of PCE to animals,  Including data
 from bioassays conducted to evaluate the cardnogenlcity of PCE.   Bloassay
 results are also used as the basis of the quantitative assessment of.
 carcinogenic potency In Section 7.  The toxldty of PCE has also  been  reviewed
 by IARC (1979),  Re1chert (1983),  HHO (1984),  and the  U.S.  EPA  (1980, 1982
 1984b,  1985a,  1985b).
      We begin  our discussion of PCE toxldty  with analyses  of  toxic effects  to
 major body organs and systems.  For these effects/Appendix A  presents  a
 review and summary of dose-response Information  for different  routes and
 periods of exposure.   The Information  1n Appendix A would be relevant  to the
 development of safety limits for  PCE concentration In  drinking water to
 protect against  acute,  subchronlc,  and  certain (specifically, noncardnogenlc
 and nonmutagenlc) chronic toxlcologlcal  end points. In  the  absence of  adequate
 human-toxldty data.   We then  examine studies dealing with  the teratogenidty
 of PCE.   The section  concludes with  a review of  the mutagenlc potential of PCE
 and Its  metabolites and  a summary of the  results  of animal  cardnogenicity
 bioassays.

 TOXIC EFFECTS  ON  ORGANS  AND  SYSTEMS

 Liver

     Cornish et a].. (1973) administered 0.3 to 2.0 mL/kg (0.33  to 4.95  mg/kg)
 of  PCE IntraperUoneally  (IP)  to  rats.  Liver damage was measured by an
 increase In serum glutamlc oxalacetlc transamlnase levels (SCOT),  and  was
 observed at all doses (Cornish et a].., 1973).   Ogata et ah  (1968) observed  a
 decrease 1n the adenoslne tHphosphate (ATP)  content of liver as  well as an
 Increase 1n the content of Uplds  and trlglycerldes after mice  were  exposed  to
800 ppmv for 3  he.  Elevation of serum glutamlc pyruvate transamlnase  (SGPT)
levels In mice  was elicited by exposure  to 3700  ppmv for 9 to 12  h,  as  well  as
by IP administration  of 3900 mg/kg (GehHng,  1968). Klaassen and  P1aa.<1966>

                                      37

-------
also measured increased levels  of SGPT. in  mice that received single
intraperitoneal  doses of 2.9 ml/kg.   These animals  had  enlarged  hepatocytes
and slight liver necrosis.   A single intraperitoneal  dose of 1.23  ml/kg
elevated SGPT levels in dogs.
     Cornish and Adefuin (1966) studied the effects of  PCE administered  in
combination with .ethanol.  A single dose of ethanol was administered  by
stomach tube (5 g/kg).  Rats were then exposed to 4000, 5000,  or 10,000  ppmv
PCE for 6, 4, or 2 hours, respectively.  None of these  treatments  had a
statistically significant effect on levels of SCOT, SGPT, or SICO  (serum
isocitric dehydrogenase).
     Carpenter (1937) studied the subchronic and chronic inhalation  toxicity
of PCE in rats.  Although no effects were observed in animals treated with
70 ppmv (8 h/d, 5 d/wk for 7 months), rats that received 150 exposures of
230 ppmv had less glycogen storage than unexposed animals.  Exposure to
470 ppmv (150 d) caused liver congestion and swelling.   Rowe et a!-  (1952)
exposed guinea pigs to  100 ppmv 7 h/d over a period of 17 to 185 d.   No
effects were apparent in animals that received 13 exposures in 17 d.  However,
when the number of  exposures was increased to 132 over 185 days, females had a
significant  increase  in  liver weight  (p - 0.01) and animals of both sexes
exhibited  Hpid  accumulation in the  liver.
     Schumann  et a]..  (1980)  dosed mice and rats orally with 100, 200, 500,  or
 1000 mg/kg  of  PCE daily for  11 d.   In mice, all dose levels produced
hepatocellular  swelling, a  significant increase in absolute liver weight
 (p<0.05),  and  a significant  decrease  in hepatic DNA content (p<0.05).  All  of
 these  changes  are indicative of hypertrophy  (the enlargement of an organ due
 to an  increase  in size  of  its  constituent  cells).  Mice  that received 100 mg/kg
 displayed an increase in hepatic  DNA synthesis.  In contrast to the  pathologies
 that  developed in mice, only the  highest  dose,  1000 mg/kg,  induced hepatic
 toxicity in rats.   These animals  had a statistically significant  increase in
 relative liver weight (p<0.05).   A change in the  staining affinity of
 hepatocytes (for hematoxylin and  eosin) was  also observed.   The significance
 of the latter observation  is not  known.
      Buben and 0'Flaherty  (1985)  treated  mice by  gavage with  dosages ranging
 from 20 to 2000 mg/kg, 5 d/wk for 6 wk.   Liver weights and liver  triglycerides
 were significantly greater than controls  at doses  of 100 mg/kg  and  above
 (p<0.001).  A Juje-dependent increase in  liver degeneration and karyorrhexis

                                        38

-------
 was  also  observed at  TOO mg/kg  and  greater.  Activity  of  glucose-6-phosphatase
 (G6P)  was Inhibited,  and there  was  a  significant  Increase 1n  SGPT  at  500,
 1000,  1500,  and 2000  mg/kg  (p<0.001).  Hepatic DNA  content  was  measured  in
 mice treated with 200 pr 1000 mg/kg;  animals that received  1000 mg/kg  had
 significantly  lower levels  of DNA (p<0.01).
      In an NTP-sponsored study  (NTP,  1986) of the effects of  PCE on mice and
 rats,  animals  were exposed  to PCE by  Inhalation 6 h/d, 5  d/wk for  103  wk.  Mice
 were exposed to 100 or 200  ppmv and rats to 200 or  400 ppmv.  Male and female
 mice of both exposure groups developed liver degeneration and necrosis.
 Development  of these  pathologies appeared to be dose related.   The incidence
 of liver  degeneration in male mice was as follows:  controls,  2/49; low dose,
 8/49;  and high dose,  14/50.  The observed incidence 1n female mice 1n  the
 control group  was 1/49; in  the  low dose group, 2/50; and  in the high dose
 group, 13/50.  The number of male mice exhibiting necrosis  in the treatment
 groups increased with increasing exposure concentrations  (I.e., controls,
 1/49;  low dose, 6/49; and high dose,  15/50).  For female mice the incidence of
 necrosis was:  controls, 3/48; low dose, 5/50; and high dose,  9/50.  Treated
 male mice (but not females) had a greater incidence of hepatic nuclear
 inclusion than controls (i .e., for controls, 2/49;  low dose,  5/49; and high
 dose,  9/50).   The statistical significance of these data was  not evaluated.
 Under  the conditions of this study,  rats did not develop hepatic lesions in
 response to  exposure  to PCE.

 Kidney                                            • ,
                 «.
     Klaassen and Plaa (1967) administered PCE to dogs intraperitoneally and
 measured excretion of phenolsulforiephthalein (PSP),  a substance used  to test
 for  renal  function.   Control dogs excreted 561 of the PSP within 30 minutes;
 excretion of less than 39X was .considered to be an Indicator of kidney
 dysfunction.   Kidney function was significantly affected after a single IP dose
 of 1.4 ml/kg (statistical  significance was not given).   Plaa and Larson (1965)
 reported that all  mice given a single IP dose of 2.5 ml/kg exhibited  swelling
of the proximal convoluted  tubule,  and one animal  (of six  treated)  developed
necrosis of the proximal  convoluted  tubule.   Mice that received  a  single  dose
of 2.5 or 5.0 ml/kg  excreted protein in their urine.  The  statistical
significance  of these  responses  was  not reported.  Carpenter (1937)  found that

                                       39

-------
rats given 230 ppmv (8 h/d, 5 d/wk)  for 21  d developed swelling and congestion
of the kidneys.  This response was exacerbated when the concentration was
increased.to 470 ppmv.  Subchronic exposure of mice aand guinea pigs to
400 ppmv, 7 h/d for 169 times in 236 d caused swelling of the tubular
epithelium along with an increase in kidney weight (Rowe et §1-, 1952).
     The NCI (1977) cancer bioassay of PCE documented a high incidence of toxic
nephropathy in both species of rodents and in all  dose groups (toxic
nephropathy was defined as degenerative changes in the proximal convoluted
tubule, fatty degeneration, and necrosis.'of the tubular epithelium).  PCE was
administered by gavage.  In this study, mice received time-weighted-average
(TWA) daily doses of 386 to 1972 mg/kg; toxic nephropathy was observed in 82
to 100% of the animals.  Rats received TWA daily doses of 471 to 949 mg/kg; 58
to 941 of these animals developed toxic nephropathy (see'Table A-7 for
specific data).
     A bioassay sponsored  by  the NTP  (NTP, 1986) documented kidney casts,
nephrosis, and tubular  cell karyomegaly in mice (animals received  100 or
200  ppmv of PCE 6  h/d,  5 d/wk for 103 weeks).  Casts occurred more .frequently
in  treated male mice  than  in  controls (incidence in controls,  3/49;  low dose,
9/49;  and  high dose,  15/50).  The trend in  female mice was  not  clearly dose
related  (incidence in controls,  4/48;  low dose, 4/49; and  high  dose,  15/50).
Nephrosis  developed at a greater incidence  in  treated female  mice  (control,
5/48;  low  dose,  14/49;  high  dose, 25/50).   For male mice,  the incidence  of
nephrosis  in  the  controls  was 22/49;  low dose, 24/49; and  high  dose,  28/50.
Karyomegaly of tubular cells  was treatment-related.   The Incidence of this
pathology in  male mice was control,  4/49;  low dose,  17/49;  high dose,  46/50.
 In female mice,  the incidence of nephrosis  was 0/48,  16/49, and 38/50 in the
 controls,  low-dose, and high-dose groups,  respectively.
      The same study (NTP,  1986) reported  a dose-related increase of renal
 tubular cell  karyomegaly in rats of both  sexes.   Low-dose  animals received
 200 ppmv of PCE;  high-dose animals  received 400  ppmv (the  exposure regime was
 the same as listed above for mice).  The  incidence of karyomegaly in male rats
 for the corresponding ccntrol group was 1/49; low dose, 37/49; and high dose,
 47/50.  In female rats, the  incidence'was 0/50,  8/49, and  20/50, respectively.
 Male rats exhibited a dose-related increase in renal tubular cell hyperplasia
 (controls, 0/49;  low dose, 3/49; and high dose,  5/50).  Only one high-dose
 female rat had renal tubular cell hyperplasia.

                                        40

-------
  Pancreas
      Hamada and Peterson (1977) studied the effects of PCE on  the electrolyte
 and protein concentration in bile duct-pancreatic fluid (BOPF).  The actual
 source of this fluid (bile duct and/or pancreas) is not known.  Rats were
 given a single IP dose of PCE (1.3 ml/kg in corn oil).  Animals were then
 fasted for 24 h, at which time BOPF was collected and analyzed.
      PCE caused a significant increase in BOPF flow, a decrease in
 concentration of protein in the BOPF, and an increase in the concentration of
 chloride and potassium (p«3.05 for all parameters).   The mechanism of
 enhanced BDPF is not known.   Although Hamada and Peterson (1977) discussed
 possible mechanisms that may be analogous  to secretion or cholinergic
 stimulation, they concluded that PCE (and  other chlorinated aliphatic
 hydrocarbons)  altered BDPF  by an unknown  mechanism,  and  that the toxicological
 significance of the reported effects is not known.

 Lungs.  Skin,  and  Eyes

     PCE  is  an  eye  and  skin  irritant.   Application of  PCE  to  the eye  of
 rabbits  caused  abrasion of the  epithelium and  conjunctivitis.   PCE was also
 extremely  irritating  when applied  topically to  the skin of rabbits (Duprat   '
 e£  a_L,  1976).  However, Jakobsen  et  aj.. (1982)  saw  no visible  sign of skin
 irritation when guinea  pigs were exposed to  liquid PCE.
     The NCI (1977) reported a high  incidence of pneumonia  in animals used in
 the bioassay of PCE.  Sixty-two to 791 of treated rats, and 29  to 661 of
 treated mice developed  pneumonia.  However, 95 to 1001 of control rats, and 28
 to  3SX of control mice  also developed  pneumonia.  Because of the relatively
 high incidence of pneumonia observed in control animals, PCE probably did not
 contribute directly to  tHe infection.  In a separate study, chronic inhalation
 of PCE caused a dose-related incidence of passive congestion of the lungs in
 mice (NTP, 1986).

 Reproductive System

     The only indication that PCE has any  effect on  the reproductive  system
comes  from the  work of Rowe  et a!.  (1952).   Seven male  guinea pigs  were

                                       41

-------
 exposed  to  1600  ppmv,  7  h/d  for 8  exposures  within 10 d.   Microscopic
 examination of tissues revealed slight degenerative changes in the germinal
 epithelium  of the  testes.  The Implication of this observation was not
 discussed and subsequent studies have not confirmed-the finding.   Therefore,
 1t is difficult  to evaluate  the significance.of this report.

 Cardiovascular System                                             .      '      -

      PCE has been  associated with sudden death from cardiac failure (Rowe
 et aJL.,  1952; Reinhardt  et a].., 1973).  It has been suggested that PCE (as
 well as  a number of other solvents) may sensitize the heart to the effects of
. endogenous 1y produced epinephrine (Price and Dripps, 1970; Reinhardt et a]..,
 1973).  If sensltization occurs, epinephrine-induced stimulation can lead to
 tachycardia and cardiac failure.  This response is apparently precipitated by
 physical exertion and exposure  to high concentrations.
      Kobayashi et aj.. (1982)  Investigated the action of intravenously
 administered  PCE on cardiac rhythm.   Rabbits were  anesthetized with urethane,
 while cats and dogs were anesthetized with  pentobarbital.  A  mean dose of
 10  mg/kg of  PCE administered  with 0.7 yg/kg  of epinephrine produced
 tachycardia  In  rabbits  (although  the  most sensitive  animals were  effected  by
 5 mg/kg of PCE).  Tachycardia also occurred in dogs  given a mean  dose of
 13  mg/kg PCE with 4.2 yg/kg of epinephrine.  while doses of 30 to  40 mg/kg
 decreased  left  intraventricular pressure.   Cats  exhibited ventricular
 arrhythmias  after 24  mg/kg  of PCE was administered in  conjunction with  13  to
  14  jig/kg of epinephrine.
       Rowe  et a!.  (1952) speculated that death in some  rats exposed to
  concentrations  of 3000  ppmv PCE or more In air was caused by cardiac  failure;
  however, it is  possible that cardiac failure occurred as  a result of  extreme
  CNS depression.  Reinhardt  et aj.. (1973) studied the cardlotoxicity of PCE by
  exposing unanesthetized dogs to 5000 or 10,000 ppmv.  Arrhythmias and cardiac
  failure were not observed,  and there was no evidence of'sensltization.•
       The dose levels used in the aforementioned studies  are  not representative
  of typical human exposure levels.  Furthermore, the work of  Kobayaahi et  a]..
   (1982)  utilized anesthetized animals, administered PCE Intravenously, and used
                                         42

-------
                                                               e ,
relatively large amounts of epinephrine, none of which readily facilitates
extrapolation of results to humans.  The cardiotoxic potential of PCE requires
additional experimental  work before any conclusions can be drawn.

Central Nervous System

     Acute exposure to PCE typically induces CNS depression.  Initial
depression can progress  to loss of consciousness, anesthesia, and respiratory
failure with prolonged or massive exposure.  Single oral  doses of 1623 mg/kg
produced reversible CNS  effects in cats (Maplestone and Chopra, 1933) while a
single dose of 6492 mg/kg caused lethal CNS depression (Lamson et a_L, 1929).
Death from CNS depression resulted from single doses of 4700 mg/kg (rat) and
6492 mg/kg (dog) (Smyth  et aj.., 1969; Lamson et a].., 1929).  Over a 4-h period,
2300 ppmv caused an impairment of muscular coordination in female rats, which
contributed to a loss of 80 percent of avoidance and escape responses.  Animals
apparently developed some tolerance to PCE, because this effect did not persist
when dosing was continued over a two-week period (Goldberg et aK, 1964).
     Rats exposed to 6000 ppmv lost consciousness within a few minutes;
decreasing the concentration to 3000 ppmv required several hours to elicit the
same effect (Rowe et aj.., 1952).  The NTP (1986) study found that exposure of
mice to 2917 ppmv for 4 h was lethal to all animals.  Rats appear to be less
sensitive to PCE, because a 4-h exposure to 5163 ppmv was required to produce
100 percent mortality (NTP, 1986).
     Carpenter (1937) studied the effects on rats of chronic exposure to 70,
230, 470, or 7000 ppmv of PCE.  Although various pathological changes were
                                                                            »
observed at 230 ppmv and above, no CNS effects were reported.  Savolainen
et a_K (1977) observed a slight decrease in brain RNA content and an increase
in nonspecific cholinesterase in rats exposed to 200 ppmv (6 h/d for 4 d).  A
one-month study conducted by Honma and co-workers (1980) documented a dose-
dependent decrease in dopamine content of the striatum in rats exposed to 200,
400, or 800 ppmv 12 h/d.  The decrease was not statistically significant.
Norepinephrine content of the hypothalamus and serotonin levels of the cortex
and hippocampus increased after exposure to PCE.(all three concentrations).
None of the increases were statistically significant.  A significant decrease
in acetylcholine (ACh)  levels in the striatum was measured after exposure to
800 ppmv  (p<0.05).  Drowsiness and other symptoms indicative of CNS depression

                                       43

-------
were reported by Rowe et a].. (1952) after rats were exposed to 1600 ppmv 7 h,
5 d/wk over a 25-d period.  At 2500 ppmv, 13 exposures within 18 d caused the
death of most rats and guinea pigs from CNS depression.  Rabbits that received
the same treatment displayed signs of CNS depression but did not lose
consciousness.                                           '          ,

TERATOGENICITY

     The teratog'enlc activity of PCE has been studied in rats (Nelson et aJL,
1980; Schwetz et a_K, 1974; Schwetz et a_L, 1975) and mice (Schwetz et a]..,
1975).  Maternal exposure levels ranged from 100 ppmv to 1800 ppmv.  Although
some minor effects were seen, in the progeny, PCE is not considered to be a
teratogen.  A summary of these studies is provided in Table 4-1.
     Investigations of the teratogenic and/or developmental effects of PCE have
most commonly shown evidence of maternal toxldty, rather than adverse effects
on the progeny.  Toxicity was evident at 300 ppmv (Schwetz et a]..  1974; Schwetz
et ah, 1975) and 900 ppmv (Nelson et aj.. 1980), while maternal death occurred
at 1800 ppmv (Nelson et aj.., 1980).  Dams exposed to 300 ppmv 7 h/d on days 6
to 15 of gestation had reduced body weights (rats) or an increase  in liver
weight (mice).  The pups of these mice had lower body weights, and there was a
slight increase 1n the number of runts.  Some fetuses had subcutaneous edema
or  delayed ossification of the skull and sternebrae, as well as splits in the
sternebrae.  These pathologies probably reflect developmental delays, and as
such, are considered to be reversible.  Developmental delays are believed to
result from maternal toxlcity rather than from any direct teratogenic activity
of PCE (Schwetz et §1., 1975).  The mechanism of maternal toxicity is unknown,
but is thought to Involve CNS depression (weight loss) and cytotoxicity
(change in liver morphology).  Because fetal health is often a reflection of
the health of the mother, maternal toxlcity is a significant concern.  The
loss of maternal weight, most probably due to decreased  feed consumption from
sufaclinical.effects (ataxla and anesthesia), can have a  great impact on the
growth and maturation of the fetus.  Maternal malnutrition can cause
developmental retardation of the fetus (Doull et aj.., 1980).  The
hepatotoxicity observed 1n mice could also have a profound effect  on  the
                                       44

-------































VI
"fl
g

'e
^

o

41
e
41
^S
ol
O
e

4=
U
JM
U
41
*J
i^
9
0)
9

2
fl
u •
41 '
1—

*_,
^.

41
ii—
"fl



 41
fl V)
U O

e
41 1-
U 0
e
o



VI
41

U
41
m , m
o r*» f** o • o
en ••• •" ^ ° • ^
• => . • - • •
"fli "^ _ -^i -^l
^ji ^ji
*J( 44 41 ^4 *J(
4l| Qjl jjj
N N
c •* -u c e
O 41 41 9 9
£ X J JJ VI
41 U U 41 41
Z .W CO Z Z
-^ -» c ' -e t- " c

"fl e >* 15 vi . . — u • —
C O U1 •»•> U 4! C VI . 4> •>-> O 41 *J

41 -^93*^-^ fl Cvl flO)> flOlv)
«< U^>ja«l 41 41 94 4) .— T3 fl 41 .— -o m
fl 3 C •<- U VI JJ 41 U 41 41 J= U 41 41 *J
S -a T *J •— .— u fl u U2JS4) w2j=
41-^^jC C*>>4li-»W 4fVIXl 4IVI-W
u u ai4ii»>, —^w-a^-c -a-o— -a -o — . c
9 «J .— vl — O) U 41 3 •— C C 41 C C 4i
• • .C ,— fl •• •— 41 ^ J< fl •• fl •— E . . fl .— u
i— >. Ol VI 41 *. >i 41 "O fl VI • C >» SO >> Svl
fl -J — U O ^iZi— 4lS *i C •- VI *J C — fl
*> .^ 4i .. u .— . . 41 i*. fl -a — o -a •— o ~9
4i u j >, e a\ u u >, -a 9 u 41 u — e u— e
.1^ .«• » •_ _ .— 4) «> ^ ..- «J .. — •— *J •• O
x >> .— x > •— • c 4i vi x a. >< xo.>>
•a . O -Q U —J VI 9 — • U «l -9 e 3 9K«'4I O S ••" 41
41 -ui O •- C C **•«.• j: •— U O *J 3 •— U jJ 3 •— u
> J3XIQ9 X O» ** 41 41 VI U C VI U C
U >» ^- O u •— i— 41 9 .^ fl *J C — C •— fl — - C •— fl
4) *J fl C -^ — *J fl > «J 11 U VI fl fl O X C fl 9 X E
vi— c fl vi. a. e— X— •* c u 9 C c u o C
0 — 4) S fl C 9 4) * « X — — U 4)13 <*-4rf 41-0 *.
X J-> ^ CD VI *< ^ *J -o vl i— J3 «J4IO>^VI -J 41 Q. U
99 fl C 41 •— 41 fl 41 '41 O VI i 3 fl 41 3 41 41 fl 41 3 41
Z •«-> Z •— U. VI 1. X I- U. ^ 9 VI VI Z'^a.&w Z'^&O.
^
O ' O
CMC me me me CM
9 •— 9 ^9 —9
9 — — — — 9
«>«! O *J ' O *> 9 *J *J
fl ' JJ fl *J fl *J fl

^ VI VO VI \O VI 1^ VI •—
41 41 41 41
VI O) VIS) VIS) VIS) VI
X >, >. >. >.
fl Vk. fl •«- fl >*- fl V>- fl
o o Q o a 9 09 a -


-Q -^ ^ • « -a ~o
»x -V >» «s. ^' .
fl .c j: fl fl

(X [*% f^ f^ f^.



--- — -- - .__-—-— — "--
.1 11 1 1
a. a. a. a. a.

o o o o o
O O O O 0
^» r^ PO ' ^^ ^^



II 1 1
41 41 — • 41 4)
O)">i O) ">« ,VI4I Oi"^ Sl>>
fl 41 fl 4) VI i-> fl 41 fl 41
U — U -- 41— VI U>- 1- —
o.x a.x vix^a o,x Q. i
.tolt/Yfl
fl i^ O fl ^^ O ' • 9 ^^ 2 fl *"^ O fl *<^ O
ae ce X ae at
45

-------
              4>
              U
              C
              41
              4)
             oc
              u
              41
 
           e
           o
 e
 at   u
 u   o
 e
 o
o
              .<*
                 <«  >» -e  u
                 c   c  a.  «s  c  c
                >M  •••  V> tj  1Q  •••
                     s
                          ui
                           01  O
 w   ••


58
                                                                                     46

-------
 growth of the fetus.  Maternal toxicity is also suspected of causing a small
 but significant (p<0.05) increase in the number of resorptions in treated rats
 (300 ppmv, 7h/d) (Schwetz et aK, 1975).
      Nelson et aJL (1980) performed a series of behavioral and biochemical
 tests on the offspring of exposed rats.  There were no adverse effects to
 mothers or their pups following exposure of the mothers to 100 ppmv of PCE
 (7 h/d) on days 14 to 20 of gestation.   Exposure at 900 ppmv for 7 h/d during
 days 7 to 13 of gestation produced significant differences in neuromuscular
 coordination (p<0.02) and wire mesh ascent e(p<0.05).   Exposure at 900 ppmv for
 7 h/d during days 14 to 20 of gestation caused diminished performance in  the
 wire-mesh ascent test, but increased the performance  in the neuromuscular
 coordination test.  A neurochemical  analysis of whole brain (minus cerebellum)
 was performed on newborn and 21-day-old pups.   Twenty-one-day-old pups  from
 dams exposed during either period had a significant decrease in acetylcholine
 (p<0.05).  A significant decrease in dopamine  (p<0.05)  was measured in  21-day-
 old pups from dams exposed during days  7 to 13 of  gestation.
      Elovaara et a]..  (1979)  injected 5  to 100  nmol  of PCE into the air  space
 of chicken eggs and studied  the gross effects  on the  embryo.   Malformations
 observed were exteriorization of viscera,  as well  as  skeletal  and eye
 abnormalities.   These deformities occurred in  six  embryos (out of 61  examined).
      Tests conducted  in rodents have not clearly demonstrated  that PCE  is a
 teratogen.   However,  there is some evidence that inhalation exposure  of
 pregnant rodents to PCE can  induce developmental delays  and altered
 performance in  behavioral  tests of the  offspring.
                               f.
 MUTAGENIC EFFECTS                                                   •       •

      Short-term assays have  been  conducted  to  evaluate  the ability of PCE to
 permanently alter genetic  information.   Most of the tests  of genetic  activity
 have been microbial assays that measured forward or reverse mutations.
 Chromosomal  effects have been studied In cultured mammalian eel Is.
      The ability of short-term  assays' to.detect mutagens  is compromised by  lack
 of knowledge  of  the mechanisms  involved, different sensitivity  and predictive
 ability  of each  test,  and  by variations  in  protocols used  by separate labs.
 Despite  these problems,  short-term assays  provide supportive evidence in  the
 evaluation  of a  compound's carcinogenic  potential.

^                                      47

-------
H1crob1al  Assays

     Greim et a_K (1975) evaluated tne mutagenic activity of PCE (purity
>99.91) in Escherichia coli  K12 with and without metabolic activation (S-9).
The results were negative at four loci tested.  Three of these loci are back
mutation systems (gal*, arg*. and nad*), while one measures a forward mutation
that produces resistance to 5-methyl-OL-tryptophan.  Only one concentration was
used (0.9 mM), and detailed data were not reported.  This study was the only
one cited by Fishbein (1976) in his review of the mutagenicity of halogenated
aliphatics.
     Cerna and Kypenova (1977) reported in an abstract that PCE of unspecified
purity produced base-pair substitutions and frameshift mutations in Salmonella
typhimurium without S-9 (Ames test).  Concentrations of 0.01, 0.1, and
1.0 mg/mL of PCE produced a dose-dependent increase in the number of
revertants.  This response was significant only in TA100 (51 level of
significance), a strain sensitive to base-pair substitutions.  In a host
mediated assay that used ICR mice and S. tvphimurium strains TA1950, 1951, and
1952, PCE reportedly caused a significant increase in the number of revertants
(level of significance was not given).  The doses used were lifted as LD5Q
and l/2LDgo, but exact quantities were not specified.  Because no information
was provided on  the purity of PCE used, revertant counts, or controls, the
significance of  these data cannot be evaluated.
     The results of Bartsch et aj..  (1979) conflict with those of Cerna and
Kypenova (1977).  PCE  (99.71 pure)  was studied in the Ames test with strain
TA100.  Concentrations up to 4 x  10"3 M were  not mutagenic in the presence
of an S-9  liver  fraction  from mice  pretreated with phenobarbital.  Toxicity
                                                  -4
was observed at  concentrations greater than 5 x 10   M.
     Kringstad  et ii-  (1981) evaluated the mutagenic activity of PCE in the
Ames test.   PCE  (99.01 pure) was  tested at a  single concentration
(0.1 mg/plate)  1n S. tvphlmurium.  strain TA 1535.  No source of exogenous
metabolic  activation (S-9) was used.  A slight  increase  in the  number of
revertants  was  observed  (31/piate after PCE compared to  19/plate  in  the ether
controls);  however,  the  response  was  considered negative.
     The NTP (1986)  reported  the  results of a series of  Ames  tests on  PCE
conducted  by the Environmental Mutagen  Test Development  Program.   Salmonella
tvphimurium (strains TA  98,  100,  1535,  and  1537),was incubated  with  technical-

                                        48       ,

-------
 grade PCE  in  covered  test  tubes  for  20  min.   The  test was  conducted  both  with
 and without S-9  (S-9  fractions were  prepared  from the livers  of male Sprague-
 Dawley  rats and  Syrian hamsters  pretreated with Aroclor 1254).   The  greatest
 number  of  revertants  was observed  in  strain TA100 (all  doses,  both with and
 without S-9); high doses (333 yg/plate)  were  toxic  to TA1535  and 1537 in  the
 absence of S-9.  However,  PCE was  judged  not  to be  mutagenic  in any  of the
 strains, regardless of the concentration  tested.
     Callen et a_K (1980)  used Saccharomvces  cerevisiae. strain D7,  to study
 the frequency of gene conversion (trp5  and ilvl loci)  and mitotic recombination
 (ade2 locus).  PCE (purity not given) was added directly to a cell suspension
 (log phase) and  incubated  in a closed vial for one  hour.  Samples were
 centrifuged,  resuspended in buffer, and plated on a medium with glucose."
 Survival decreased as dose increased  from 0 to 4.9, 6.6, or 8.2 mM.   At 8.2 mM
 survival was greatly  reduced and assessment of* genetic  activity was  precluded.
 Exposure to 6.6  mM elicited substantial increas.es in mltotic  recombination
 (52.6 recombinants per 10  survivors  vs 3.3 recombinants per  104 survivors
 in controls).  The number of gene  conversions (trpS) also increased  at 6.6 mM
 (8.3 convertants per  10  survivors vs 1.4 x 105 convertants per 105  survivors
 in controls); the number of revertants at the ilvl  locus was not measured at
 this exposure concentration.  Fabre (1978) has proposed that mitotic
 recombination and gene conversion  may.be induced by the same mechanism.  If so,
 the response may have been inaccurately evaluated, as ade2 recombinants were
 estimated  from plates that had been previously used to determine the  number of
 trp5 convertants.  Consideration of this possibility as well  as  a lack, of
 statistical analysis of results limit the strength of evidence  presented by
Call en  et §1.  (1980).  However, mitotic recombination and gene  conversion are
 indicative of interaction of a substance with DMA.  Since the increase in
 frequency of mitotic recombination was pronounced, this response may warrant
additional  study.                                                      .    .
     Bronzetti et a].. (1983) also  studied the effect of PCE on  the trpS,  ade2,
and ilvl loci  in S.  cerevisiae.  Cells in the stationary phase were exposed for
two hours to 5,  10,  20,  60, or 85 mM PCE.  All results were negative.  The PCE
used was 99.51 pure,  whereas the purity of that used by Callen |t a_L (1980)
was not reported.  When  comparing  the results  of these two studies,
consideration must be given to the possibility that cells in  the stationary
phase may be re'sistant to the mutagenic or toxic action of xenobiotics.

                                       49

-------
     In addition, Bronzetti  et a_[. (1983) conducted an intrasanguineous host-
mediated assay that tested the ability of PCE to induce genetic effects in S.
cerevisiae.  Yeast were exposed to PCE and its metabolites in the liver, lungs,
and kidneys of mice.  PCE did not induce point mutations, mitotic
recombination, or mitotic gene conversion.

Drosophila                         ,

     The NTP (1986)" reported the results of an assay that tested the ability
of PCE to induce sex-linked recessive lethal mutations in Drosophila.  Males
were exposed by injection or by feeding and were mated to a series of untreated
females.  Neither route of administration produced a statistically significant
increase in sex-linked recessive letnal mutations.
Mouse Cells
                            *
     PCE was not mutagenic in L5178Y/TK*'" mouse lymphoma cells, with or
without metabolic activation.  Cells were treated for four hours with 6.25,
12.5, 25.0, 50.0, and  100.0 nL/mL of PCE.  Following incubation for 48 h, celi,
were plated in medium  supplemented with trifluorothymidine for selection of
cells mutant at  the  thyraidlne kinase (TK) locus.  No statistically significant
increase  in mutation frequency was observed at any dose  level (NTP, 1986).
     Somatic mutations are thought to occur when a substance or one of its
metabolites interacts  with DNA.  Alkylation of DNA is therefore one possible
indicator of genotoxlcity.   Schumann et aj.. (1980) measured binding to hepatic
macromolecules of mice after administration of radiolabeled PCE.  No
radioactivity was detected bound to  hepatic DNA.  The specific activity of the
tetrachloroC14C]ethylene used in this  study was  too  low  to permit detection  ;
of low  levels of DNA binding.   Therefore, these  results  do not rule out the
possibility of  DNA  alkylation following  exposure to  PCE.

Genetoxic Activity  of  Metabolites

      Tetrachloroethylene oxide (PCE  oxide)  is believed  to be  the first
 intermediate  formed by mlcrosomal  oxidation  of PCE.   A  concentration-dependent
 mutagenic response  was produced by PCE oxide (0.5,  1.3,  2.5,* 5.0,  and 25.0 mM)

                                        50

-------
in S. typhimurium TA1535 (without metabolic, activation) but PCE oxide was not
mutagenic to £. coli WP2 uvrA.  The mutagenicity of this epoxide was also
evaluated,in a DNA-repair-deficient strain of E. coli (E. coli pol A 1").
In the latter assay, genotoxicity is measured by comparing .differential growth
inhibition of the DNA-polymerase-deficient strain, pol A V",  with a polymerase-
proficient strain, pol A 1*.  PCE oxide gave positive results in this test;
growth of the pol A 1" strain was inhibited at all dose levels used (Kline
et aj... 1982).
     Trichloroacetic acid (TCA) is the principal metabolite of PCE excreted in
the urine of rodents and humans; trichloroethanol has also been identified as
a metabolite of PCE.  In an Ames test conducted with metaboTic activation, TCA
CO.45 mg/plate) and trichloroethanol (7.5 mg/plate) were not  mutagenic to
strains TA^98 and TA100.  There is some indication, however, that
trichloroethanol can induce sister-chromatid exchange in cultured human
lymphocytes (Gue_ta_L, 1981).

CHROMOSOMAL DAMAGE
  •   Cerna and Kypenova (1977) did not observe any cytogenetic effects in mouse
bone-marrow cell's following single or repeated IP Injections (daily injections
for five days).  Cells were recovered for analysis 6, 24, or 48 hours'after the
last injection of PCE.  No details were provided regarding doses used or the
specific end points that were studied (the study was published only as an
abstract).  The NTP (1986) published the results of assays for PCE induced
chromosomal aberrations and sister-chromatid exchanges in Chinese Hamster Ovary
(CHO) cells, both with and without S-9.  The S-9 fractions were obtained from
livers of male rats pretreated with Aroclor 1254.  Data were reported in tables
only, with no supportive discussion (such as the number of cells scored for
each dose level).  However, at least three dose levels were used (with and
without S-9), as were both positive and negative controls.  PCE had little, if
any, cytogenetic effect in.either assay.

CARCINOGENICITY IN ANIMALS
                         *
     Two lifetime bioassays have been completed on PCE (NCI, 1977; NTP, 1986).
Additionally, there are three other studies that have addressed the question

                                       51

-------
of PCE card nogeni city (Rampy et a_L, 1978; Thelss et a_L, 1977; Van Ouuren
it aK, 1979).  In Table 4-2 we have summarized results of the bioassay,
studies that resulted in significant increases in malignant neoplasms among
the exposed animals.
     The National Cancer Institute (NCI, 1977) conducted a study in which
86C3F1 mice and Osborne Mendel rats were administered PCE in corn oil by
gavage, 5 d/wk for 78 wk.  Animals were then observed for 32 wk (rats) or 12 wk
(mice).  Mice were 25 days old at initial  treatment;  rats were 35 days of age.
The time-weighted average daily doses ofjPCE were 536 and 1072 mg/kg for male
mice, 386 and 722 mg/kg for female mice, 471 and 941  mg/kg for male rats, and
474 and 949 mg/kg for female rats.
     PCE causeQ a statistically significant increase (p<0.001) in the incidence
of hepatocellular carcinoma in mice of both sexes and both dosage sroups
(Table 4.2).  The time to first tumor development was considerably shorter in
treated mice than in controls.  Hepatocellular carcinomas were first detected
at weeks 91 and 90 in untreated and vehicle controls.  However, in male mice,
hepatocellular carcinomas were detected after 27 weeks (low dose) and 40 weeks
(high dose).  The first hepatocellular carcinomas were observed in female mice
at week 41 (low dose) an'd week 50 (high dose).
     Exposed mice also exhibited a high incidence of toxic nephropathy, a
conditibn that was not observed in controls.- Median survival times of mice
were inversely related to dose.  Control males had median survival times of
             »                                                •
more than 90 weeks; this decreased to 78 weeks in low-dose males and 43 weeks
in high-dose males.  The median survival time of control females was also
greater than 90 weeks.  Median survival times of low- and high-dose females
were 62 and 50 weeks, respectively (NCI, 1977).
     Early mortality occurred in all groups of rats dosed with PCE.  Half of
the high-dose males had died by week 44; half of the high-dose females died by-
week 66.  The median survival time of control animals was 88 to 102 weeks,
depending on sex.  The National Cancer Institute (NCI, 1977) determined that
there was a statistically significant association (p<0.001) between increased
dosage of PCE and increased mortality.  The early mortality obser /ed in rats
and its statistical association with dose of PCE indicate that the doses given
to rats in this bioassay were inappropriately high*.  Because the optimum dosagi
was not used, and because significant early mortality occurred, these  results
preclude any conclusions regarding the carcinogenicity of PCE  in rats.

                                       52

-------








V)
CU
•5
^j
^**
c/l

ITS
(/I
vt
tj
o
3
1
'c
s
£
ItJ
•o
CU
u
CU
t3
U
e
•^
i
w
•M
0

1_
E
1
1
"

CU
3
.•ts
H-

cu
"<*»

1
CL
CU
u
e
•ft CU
CU •Q
e *u
O c
cu

u
§
h- eu
st
P""

0
+* CU
Concentra
or dos



X
CU
OO

CU ~-
•#• ^0
U i-
CU -4-1
CL 



>» .
a
oo \
J3 J3 . .Q J3
^•* P—1* «^ V"^
< o o  vo CM
CO CO OO ^^
o 10 i— o en I**



^
^C Lw


0
VO
CO

u
s
o*

*
'X .
JS JD JS JS

< O O  O (_)
%%% ^== ^^^^$=



•
ill ill llllll
CL CL CL CL
CLO O CLO O . CLO O CLO O
0 — CM O°-CM O 2 CM 0 — CM




^» ^_ .^
^" ^b UB

U.
g

CD

U
•^
OO
at . ,
^~
a."
z
'  53

-------



































•a
3
*#"•

C
8



CM
^*
41
JS
IQ



41
3
fd
>
I
a.
4)
O
e
v» O>
««—
^3
vi •«-
e u
o e
VI
41
V-
u
cu
>»

o
4-> 4>
(0 VI
J- 0
c
41 U
U O
O
0



X
4)
oo


vi e
**** rtJ
u u
O. VI







~ "O

4-1
00
u u u u
i^ ^* f^ CO
SO CM LO
5 O O
< d d < o o
z z
II .11

0.0. 0.0.


000 OO O
LO to in in tr> 10
co r^^ ^* oo o ^ft
CM n n •^ ro CM




zzz zzz


^ — > >
• E s e
> o. o. > o. a.
S 0. 0. E 0. 0.
0. 0.
O» C3 ^*? CC ^3 ^3
^j ^^ ^^ ^3
O CM ^ O CM ^






Z *J-


s
^
fl
u.

V)
4-1



VO
03
CT»


a.
H-
Z










*




•a
c
c
3
u
5

o.
5
1
(J
 VI
S —
^^ ^9
4J <
U

f 41
U. _J
J" _•
49 49
41 41
^™ ^*»

4^ £
f~ ^"*
H^ *^
j3 J3
^9 ^9
"o "o
u u
a. a.
.a u



























•
c
*
u
' «^
<*-
gj
en
•*"
>•
• "*
41 U
!a 4J
4 (/)
pM» H^
2 2
<3 V)

^J <^B!
O ' O
e e
i N
< 00
z z


.54

-------
      Questions  have  been  raised  about  the  purity of  PCE  used  in  the  NCI mouse
 and  rat  bioassays.   The PCE  was  produced by  Aldrich  Chemical  Co.  and had a
 purity of  99%.  However,  epichlorohydrin (ECH)  was apparently used as  a
 stabilizer.   It has  been  suggested  that the  presence of  this  contaminant may
 have directly contributed  to tumor  induction.   ECH is a  direct-acting
 alkylating agent  and-is mutagenic (Kucerova  et  a_L,  1977;  Bridges, 1978).  Van
 Duuren et  aT. (1974) demonstrated that ECH was  carcinogenic in mice  when
 injected subcutaneously.   A  subsequent study by Laskin et  aj_.  (1980)  showed
 that ECH induced  neoplastic  lesions of the nasal  cavity  of rats.  Most of these
 tumors were  carcinomas of  "the  squamous epithelium.   Interestingly, 30-d
 exposures  to 100  ppmv produced a much  greater incidence  of cancer than lifetime
 exposures  of 30 ppmv (exposures were 6 h/d,  5 d/wk).  A  study by  Konishi et
 a_l.  (1980) and  Kawabata (1981) also showed that ECH  fed  discontinuously to
 rats  in drinking  water at  a  concentration of 1500 ppm (and at a  lifetime TWA
 dose  of approximately 40.2 mg/kg-d) induced  a significantly increased
 incidence  of papillomas and  squamous cell carcinomas  of  the forestomach above
 that  of control animals.
      The exact  quantity of ECH present in the PCE used in  the'NCI study is not
 known, but it has been estimated that  high-dose male  mice  received 0.42 mg/kg-d
 (U.S. EPA, 1985a).   This represents a  small  fraction  of  the dose that elicited
 squamous cell carcinomas in  rats.  Furthermore, ECH appears to initiate tumors
 by a  localized  tumorigemic reaction at sites where it is in direct contact with
 tissue, such as nasal or forestomach squamous-cell epithelium  (U.S.  EPA,
 1984c).  No  animal in the  NCI bioassay developed  tumors  at these sites.  ECH
 is,among the weakest of the more than  50 suspect  carcinogens  evaluated by the
 U.S.  EPA Carcinogen Assessment Group,  having an estimated  upper-bound
 carcinogenic potency, or effect per unit dose at  low  doses; to humans of
 9.9 x 10~  (mg/kg-d)~ , based on data  indicating  increased nasal  cavity tumor
 incidence in rats exposed  to ECH via drinking water "(U.S.  EPA, 1984c).  Using
 the methodology of U.S. EPA (1984c), the equivalent potency to mice would be
 9.9 x 10~3 x (fm/fh), where fm and fh  are the fractions of body weight
 consumed as water per day, equal  to 0.17 and 0.029 for mice and humans
 respectively.  The potency for ECH to mice is therefore estimated to be
0.058 (mg/kg-d)"  .  Using this potency estimate, the highest dosed animals
(high-dose male  mice) in the NCI  (1977) bioassays*would be expected to incur
                                       55

-------
an increased cancer risk of (0.42)  x (0.058)  - 0.024,  or  less  than  2.5%.
Therefore, it is unlikely that ECH  contributed significantly to the observed
increased tumor incidence in PCE-exposed mice in  the  NCI  (1977) bioassay.
     Rampy et al_. (1978) exposed male and female  Sprague-Dawley rats to PCE by
inhalation (300 or 600 ppmv) 6 h/d, 5 d/wk for 12 months.   Animals  were
subsequently observed for 18 months.  High-dose males  had  slightly  greater
mortality than did controls, but neither sex  exhibited an  increased incidence
of tumors, regardless of dose.  Interpretation of this study is limited by the
duration of the exposure and by the fact that it  was  reported  only  as an
abstract.
     Theiss et a].. .(1977) studied the ability of  PCE  to induce lung adenomas
in A/St male mice.  Animals six to eight weeks old were given  80, 200, or
400 mg/kg of PCE in tricapryliri (intraperitoneally) three times a week.  Each
group received 14, 24, or 48 injections.  Animals were sacrificed 24 weeks
after the first injection and were examined historically for the presence of
pulmonary tumors.  Treated animals did not exhibit a significant increase in
the average number of lung tumors,when compared to controls.  The relevance and
validity of these test  results are of questionable significance, though, as
this test has  not produced positive  results with several known animal
carcinogens.
     The  ability of  PCE to  initiate  and/or promote skin tumors in ICR/Ha Swiss
mice was  investigated by Van  Duuren  et a].. (1979).  In one group, 163 mg of PCE
was applied once tq  surface skin.   Fourteen  days after this, phorbol myristate
acetate,  a  promoter, was applied to  the  same  area  three times  a week for 428
to 576  d.   A  second  group,received  54 mg of  PCE  by topical application  three
times a week for.440 to 594 d.   PCE did  not  show any  initiating activity.  It
also gave negative results  in the  portion "of the experiment that tested  its
ability to act as  a  complete  carcinogen.  It is  difficult  to  interpret  these
data in relation to  the carcinogenic action  of PCE .because the significance
and sensitivity of skin application tests are not  thoroughly  understood.
     The most definitive study of  the carcinogenic potential  of PCE was
conducted by Battelle  Pacific Northwest Laboratories  for  the  National
Toxicology Program (NTP, 1986).   B6C3F1  mice and F344/N  rats  were  exposed to
99.9%  pure PCE by inhalatfon, 6' h/d, 5 d/wk  for  103  wk.   Mice were exposed to
concentrations of 0, 100,  or 200 ppmv;  rats  were exposed to concentrations of
0, 200, or 400 ppmv.                    .

                                        56   .

-------
      Treated male rats had lower survival rates than control animals (controls,
 23/50; 200 ppmf, 20/50; 400 ppmv, 12/50).  Survival rates among female rats
 showed little variation (controls, 23/50; 200 ppmv, 21/50; 400 ppmv, 24/50).
 Both exposure concentrations produced significant increases in mononuclear cell
 leukemia in female rats (incidence in controls, 18/50;  in rats receiving
 200 ppmv, 30/50; and in rats receiving 400 ppmv, 29/50).   Life Table analysis
 showed the significance of these increases to be p = 0.023 (200 ppmv) and
 p = 0.053 (400 ppmv).   Treated male rats also developed mononuclear eel 1
 leukemia in greater numbers than controls (controls, 28/50;  200 ppmv, 37/50;
 400 ppmv, 37/50).   Levels  of significance (evaluated by Life Table  analysis)
 are p = 0.046 (200 ppmv)  and p » 0.004 (400 ppmv)  (Table  4-2).
      Renal  tubular-cell  adenomas are rare neoplasms with  a historical
 occurrence  at Battelle Laboratories  of less than one percent (Appendix  F,  NTP,
 1986).   Renal  tubular-cell  adenocarcinomas are even less  common,  and have not
 been documented  in any NTP studies  (NTP,  1986).   Male rats (at  the  200  and
 400 ppmv  exposure  levels)  exhibited  an increased incidence of both  of these
 neoplasms (see Table 4-3).   Although the  increases  were not  statistically
 significant,  they  appeared  to  be dose-related.   Tubular-cell  hyperplasia  was
 observed  in  eight  treated males  but  only  in one  treated female  rat.   Tubular-
 cell  karyomegaly developed  in  a  majority  of male rats but  was  less  common in
 females.
      Brain glioma  is a rare  tumor of neuroglial  cells (the cells  that compose
 the supporting structure of  nervous  tissue).   Brain  gliomas  were  observed in
 one male  control rat and in  four male  rats  that  were  exposed  to 400  ppmv PCE
 (NTP, 1986).   This  increase was  not  statistically significant.  However,
 because the historical incidence of  these  tumors is  quite  low (0.2%  at Battelle
 Laboratories), the  increased incidence in  treated animals  is noteworthy.
      In the NTP  study (NTP, 1986), the survival  of  low-dose male mice (after
 week  74)  and of  high-dose male mice  (after week  75) was significantly lower
 than  controls 
-------
Table 4-3.  Incidence of renal  tubular cell  adenomas  and  adenocarcinomas  in
male rats exposed to PCE by inhalation.4
Neoplasm
Tubular cell adenoma
Tubular cell adenocarcinoma

Control
1/49
0/49
Treatment
200 ppmv
3/49
0/49

400 ppmv
2/50
2/50
Tubular cell adenoma or
  adenocarcinoma
1/49
3/49
(p = 0.259b)
4/50
(p . 0.070b)
  a Data are from NTP, 1986.
  b P-values are based on Life Table Tests (Appendix E,  NTP,  1986).
         • *               ,                          '                • '
     Hepatocellular adenomas  occurred in both sexes of mice,  and at  both
concentrations of PCE (Table  4-2).  The incidence of adenomas was not
statistically significant.  However, the combined incidence of hepatocellular
adenomas and hepatocellular carcinomas was significant.   In males, the combined
incidence was: controls, 16/49; low dose, 31/49; (p « 0.002); and high dose,
40/50 (p<0.001).  In females, the incidence of hepatocellular adenomas and
carcinomas was: controls, 4/48; low dose, 17/50 (p = 0.001);  and high dose,
38/50 (p<0.001).

Summary of Evidence of Carcinoqenicitv in Animals

     The NCI  (1977) bioassay of PCE found that administration of PCE by gavage
was associated with a statistically significant increased incidence (p<0.001)
of hepatocellular carcinoma.   This increase was documented in low- and high-
dose B6C3F1 mice of both sexes.  A decrease in the time to first tumor
development was also observed in treated mice of both sexes and both dose
groups.  Early mortality in rats prevented an analysis of PCE's carcinogenic
potential  in  this species.  The NCI (1977) concluded  that under the conditions
of this study, PCE was a  liver carcinogen to B6C3F1 mice of both  sexel.
     In 1979,  IARC reviewed the NCI study on PCE as well as the animal
carcinogenicity studies of Rampy  et a]..  (1978)  and -Theiss et a]..  (1977).   Only
two short-term assays were evaluated  (Cerna and Kypenova, 1977; Greim et  aj...
1975).  IARC  (1979) determined that there was "limited evidence"  that PCE is
carcinogenic  in mice.
                                        58

-------
     IARC re-evaluated the evidence of carcinogenocity of PCE to animals in a
1982 publication (IARC, 1982).  Although this review was extended to include
an epidemiologic study as well as recent information on PCE's activity in
short-term tests, no additional animal data were available.   As in 1979, IARC
concluded that there was only "limited evidence" that PCE is carcinogenic to
animals (IARC, 1982).
     The final report of the NTP inhalation bioassay on PCE  was released in
1986 (NTP, 1986).  The NTP determined that under the conditions of this study,
there was "clear evidence of carcinogenicty" of PCE for male F344/N rats, "some
evidence of carcinogenicity" of PCE for female F344/N rats,  and "clear evidence
of carcinogenicity" of PCE for both sexes of B6C3F1 mice.  In rats, these
conclusions were based on an increased incidence of mononuclear cell  leukemia
in males and females.  Male rats also developed renal  tubular cell neoplasms
(a rare type of tumor).  The evaluation of carcinogenicity in mice was based
on an increased incidence of hepatocellular adenoma and hepatocellular
carcinoma in males, and an increased .incidence of hepatocellular carcinoma in
females. '
                                      59

-------
                           5.   TOXIC EFFECTS  IN  HUMANS

     To assess correctly the health risks from a chemical, consideration of
human toxidty data is essential.  Unfortunately, information on human toxicity
for many substances is limited or is anecdotal in nature.  For PCE, however,
there have been some controlled inhalation exposures.for the purpose of
defining occupational limits.   In addition, epidemiological  studies have be.en
conducted to explore the relationship between exposures to PCE vapors and .
potential health effects.  We begin this section with  a brief overview of the
health effects of PCE exposures.  That discussion is followed by a review of
different epidemiological studies dealing with PCE.   Finally, we examine human
data on the toxic effects of PCE on specific organs  and systems.

GENERAL TOXICITY                           ,
     Acute exposure to PCE can produce skin irritation and burns, as well as
irritation of the eyes and respiratory tract.  Central nervous system (CNS)
depression is the most immediate effect of exposure, but high concentrations
can also cause loss of consciousness and respiratory failure.  Liver and kidney
toxicity can result from single exposures (Stewart et aj... 1961a; Stewart,
1969; Hake and Stewart, 1977), but the concentration and duration are typically
greater than those that cause transient CNS effects.  Chronic occupational
exposure to PCE has caused headache, dizziness, hangover, intoxication,
diminished cognitive abilities, and a decreased performance in the Romberg and
Flanagan Coordination Tests (Stewart "et ah, 1961b; Stewart et aJL,  1970;
Stewart et a].., 1974; Stewart et a].., 1976).  Extended exposure to PCE (2-1/2
months to several years) has also caused changes in kidney and liver function,
cirrhosis, and toxic hepatitis (Coler and Rossmiller, 1953; Meckler and
Phelps, 1966; Hake and Stewart, 1977).
     In experimental studies human volunteers have been exposed to PCE by
inhalation at various concentrations and for various durations.  Because
subjects were allowed to leave exposure chambers .when they felt discomfort,
observed adverse effects have been restricted to the respiratory tract and CNS
(Carpenter, 1937; Rowe et al_., 1952; Stewart et aj.., 1961b; Stewart et a_L,
1970; Stewart et al_., 1974; Hake and Stewart, 1977).  Accidental exposure  to
PCE has occurred primarily as a result of its use as an industrial solvent.

                                       60

-------
 Although specific exposure  levels have not always been determined,
 concentrations have been high enough to cause liver and kidney dysfunction
 (Coler and Rossmiller, 1953; Hake and Stewart, 1977; Koppel e_t aj.., 1985).

 EPIDEMIOLOGIC EVIDENCE FOR CARCINOGENICITY IN HUMANS

     Epidemiologic studies of PCE exposure have been reviewed by Reichert
 (1983) and by the U.S. EPA  (EPA, 1985a).  Blair et aj. (1979) analyzed the
 death certificates of 330 union laundry and dry-cleaning workers (out of a
 cohort of 10,000).  Of 330 decedents, 279 had worked solely in dry-cleaning
 establishments (while union members).  The solvent(s) used by the dry cleaners
 were not identified.  Length of union membership ranged from one to 25 years,
 with a mean of 13 years.  The number of expected deaths from cancer was 67.9
 (based on proportionate mortality of the U.S. population) while 87 deaths from
 cancer were observed.  The authors also compared the number of years of union
 membership with cause of death.  With the exception of nonwhite males, length
 of union membership was nearly identical for cancer and noncancer deaths.
 Increased mortality (p<0.05) from cancers of the respiratory tract, cervix,
 and skin, was documented.  When all malignancies were evaluated together, the
 number of observed deaths was also significantly greater than expected
 (p<0.05).  The authors noted that the excess of cervical  cancer may be related
 to the typically low wages and socioeconomic class of this occupational
 group.  Although an excess of liver cancer and leukemia was also observed,
 these increases were not statistically significant.
     The increases in cancer deaths among the study group probably contributed
 to a lower than expected relative frequency of deaths from other causes.   It
 is noteworthy that death from circulatory disease was significantly lower than
 expected (p<0.005).   The factors contributing to this phenomenon are not  known.
Causes of death were not determined separately for laundry and dry-cleaning
workers in this study.  The actual  solvent(s) used were not identified,  and
smoking history was  not documented.   The lack of control  for smoking was  a
significant deficiency of the study because lung cancer was a major
contributor to the total  number of cancers.   Although this study identifies a
potential  occupational hazard,, data provided are not adequate to evaluate the
carcinogenic  potential of PCE.
                                       61

-------
     Katz and Jowett (1981) analyzed the mortality patterns  of 671  white  female
laundry and dry-cleaning workers.   Data were obtained from the death
certificates of. individuals who died in the period 1966 to 1977.  Occupational
code's listed on the certificates did not distinguish between the  two  types  of
work.  Data on the duration of employment were not available,  nor were  the
investigators able to determine to which solvent(s) the individuals were
exposed.  Smoking history was not known.  Cause-specific proportionate
mortality"ratios were calculated for 25 causes of death.  A  significant
increase in risk of death from cancer of the kidneys (p<0.05)  and genitals
(p<0.01) was documented.  An excess risk from skin and bladder cancer was also
found; however, neither increase was statistically significant.   Individuals
in the group under study had a greater risk of death from cancer  of the cervix,
ischemic (obstructive) heart disease, and diabetes mellitus.  However,  when,
the effect(s) of low-wage occupations were accounted for, only the  risk for
diabetes mellitus remained statistically, significant (p<0.05).
     Other studies of laundry and dry-cleaning workers have also  reported an
increased risk of death from cervical cancer (Blair et a_L,  1979; Kaplan,
1980); however, these investigators have not compared mortality data  by
low-wage occupation.  Although not definitive, the findings  of Katz and Jowett
(1981) suggest that factor(s) other than (or in addition to) solvent  exposure
are important contributors to cervical cancer.
     Kaplan (1980) completed a retrospective mortality study of 1597  dry-
cleaning workers exposed to PCE for at least one year (prior to 1960).   By  the
end of the study period, 1028 of the cohqrt were alive, 285 had died, and the
status of the remaining 254 was not known.  Although a considerable effort  was
made to determine the history of solvent exposure, the solvent history of
approximately half of the dry-cleaning establishments was unknown.   Of those
shops with known solvent history, none had used trichloroethylene;  individuals
who had worked in shops that had used carbon tetrachloride were eliminated  from
the study.  However, prior to 1960  (the period of  interest in this  study),  the
majority of dry cleaners used petroleum solvents (NIOSH, 1980).  In keeping
with this information, Kaplan decided that employment in a shop with unknown
solvent history probably involved exposure to petroleum  solvents.  Similarities
in the physical properties and physiological effects of  gasoline (which  has
been associated with kidney cancer  in rats)  (Kitchen, 1983), and petroleum
solvents suggest that use of these  solvents  may contribute to an increased

                                        62

-------
risk of cancer.  The inability of Kaplan (1980) to quantify solvent exposure
adds further uncertainty to the interpretation of this study.  The mean
exposure concentration of individuals to PCE was calculated to be 22 ppmv for
                                                      ^
dry-cleaning machine operators and 3.3 ppmv for all  other jobs.  These values
are based on a NIOSH survey (NIOSH, 1980) cited by Kaplan (1980).
     A Standardized Mortality Ratio (SMR)* was used to compare the number of
observed deaths to the number of expected deaths, by cause.  Kaplan (1980)
found an elevated SMR (182) for malignant neoplasms  of the colon (11 observed
deaths, 6.77 to 6.98 expected deaths).  In discussing this observation, Kaplan
(1980) points out that those individuals of high socioeconomic status are at
greater risk for cancer of the colon than individuals of low socioeconomic
status.  Because dry-cleaning workers generally receive low wages, the study
cohort may have overrepresented individuals of low socioeconomic status, and
therefore, included a disproportionate number of individuals who are at low
risk for colon cancer.  If this risk trend is valid,  the elevated SMR reported
for colon cancer may actually be an underestimate.
     In addition to colon cancer,  SMR's for malignant neoplasms of the rectum
(158), pancreas (152), respiratory system (140), urinary organs (198), an9
"other and unspecified sites (major)" (156) were observed.  Although Kaplan
did not evaluate SMR's for statistical significance,  a review of this study by
the EPA (U.S. EPA, 1985a) included an evaluation of  significance of these
malignant neoplasms.   The SMR's for cancer of the rectum,  pancreas,
respiratory system, urinary organs, and "other and  unspecified sites (major)"
were not significant at the p<0.05 level.   However,  cancers of the respiratory
system, urinary organs, and "other and unspecified  sites"  were of borderline
significance (0.10
                                       63

-------
     Although the relatively small  cohort 1n this study limits conclusions
about the carcinogenic potential  of PCE,  the study results suggest a
relationship between colon cancer and solvent exposure.  Because the NTP study
of PCE (NTP, 1986) found hyperplastlc and neoplastlc changes In the kidneys of
treated rats, the Increase In the SMR from malignant neoplasms of the urinary
organs raises the possibility that occupational  exposure to solvents may
Increase the risk of cancer In these organs.  Since petroleum products have
also been linked to kidney cancer in rats (Kitchen, 1983), and because it is
probable that some of the cohort were exposed to petroleum solvents (as well
as to PCE), the possible contribution of PCE to an increased risk of urinary
organ cancer cannot be ascertained.  An additional problem in this study was
the inability of investigators to collect data on (and thus control for)
smoking history.  Since smoking is associated with an elevated risk of many
types of cancers (including lung and kidney), i.ts contribution to the elevated
SMR's reported by Kaplan needs to be evaluated (U.S. DHEW, 1979).
     Duh and Asal (1984) studied the cause(s) of mortality among 440  laundry
and dry-cleaning workers from Oklahoma who  died during 1975 to 1981.  This
study had the  same  problem  as the studies of Blair et aj[. (1979) and  Katz and
Jowett  (1981)—smoking  histories were not available and separation of the two
groups  by occupation  was not possible.  Therefore, duration or characterization
of individual  exposure  was  not reported.  However, Duh and Asal  noted that  the
two groups  of  workers probably experienced  substantially  different solvent
exposure.   (NIOSH  (1980)  reported  that, although  75% of dry-cleaning
establishments in  the U.S.  use PCE,  Oklahoma may  be unique  in  that petroleum
solvents account for  more  than 50% of  total  solvents used.)   A Standardized
Mortality Odds Ratio  (SMOR)*  revealed  elevated SMOR's  for all  digestive
diseases (1.5),  cirrhosis  of the liver (1.3), and homicide  (3.8).   SMOR's less
 than 1.0 were  reported  for diabetes mellitus  (0.7),  ischemic  (obstructive)
 heart disease  (0.8),  emphysema  (0.8),  and  suicide (0.2).  A SMOR less than 1.0
 suggests that  laundry and dry-cleaning workers may be  at  low risk for these
 diseases.
   *SMOR was defined by Duh and Asal (1984) as a method that compares the
 number of deaths by specific cause to the number of deaths due to other causes
 in the exposed population (the odds) to the expected odds derived from a
, comparison population.
                                        64

-------
      Analysis  of deaths  due to cancer showed an increase in the SMOR for
 cancers of the respiratory system (1.8),  lung (1.7),  and kidney (3.8).   Deaths
 from breast cancer were  considerably less  than expected (SMOR = 0.1).
      Brown and Kaplan  (1987)  conducted a  retrospective, cohort-mortality study
 of workers employed in the dry-cleaning industry to evaluate the carcinogenic
 potential  from occupational  exposure to PCE.   The study cohort consisted of
 1,690 members  of four  labor unions  (located  in Oakland, Detroit,  Chicago,  and
 New York City).   Individuals  selected for  the study had been employed for  at
 least one  year prior to  1960  in  dry-cleaning  shops  using PCE as  the  primary
 solvent.   Complete solvent-use histories were not known for about half of  the
 shops included in  the  study.   Because petroleum solvents were widely used  by
 dry cleaners prior to  1960, most  of  the cohort had  known or potential exposure
 to solvents other  than PCE (primarily,  various  types of Stoddard  solvents).
 The investigators  also identified a  subcohort of  615 workers who  had been
 employed only  in establishments where PCE  was  the primary solvent.
      PCE exposure  in shops  included  in  the study  was evaluated  independently
 (Ludwig et al- .1983).  The  geometric  mean  of  time-weighted-average exposures
 was  22  ppmv for machine operators, and approximately 3  ppmv  for other workers.
      Brown and Kaplan  (1987) calculated person-years-at-risk (PYAR)  for  each
 worker.  The PYAR  were then .combined  into  five-year calendar periods and
 five-year  age  groups by the life-table-analysis-system  (Waxweiler et al..,
 1983).  PYAR values were also evaluated by length of employment and  by time
 lapsed  since first  employment in a shop that  used PCE.   The  expected number of
 deaths  was calculated  by multiplying  PYAR  (by age and calendar period*) by the
 U.S.  mortality rates.  Risk of death  due to a specific  cause was calculated by
 means of a SMR.
      Among the,(main) cohort, the number of observed deaths  from all  causes
 (considered together) was  less than expected  (493 observed,  575.5 expected;
 SMR = 86).  No deaths occurred from liver cancer, although 3.5 were expected.
 There were also fewer deaths due to diseases of the'circulatory system and
 nervous system (SMR -i 70  and 73,  respectively).
     However,  observed  deaths from all types of neoplasms were higher than
 expected (142 observed, 122.9 expected; SMR = 116).  Elevated SMR's from
malignant neoplasms of  the intestine (136)  and pancreas (172) were documented.
Malignant neoplasms of  the urinary tract caused a significant excess  of  deaths
 (12 observed,  4.7 expected; SMR = 255).  Of these urinary tract cancers,

                                       65

-------
 kidney cancer caused four deaths  (2.0 expected; SMR - 200), while eight deaths
 from  bladder cancer were observed  (2.7 expected; SMR = 296). Mortality from
 calculi of the urinary system  (a  nonmalignant disease) was also greater than
 expected  (2.0 observed deaths, 0.3 expected; SMR = 667).  Although Brown and
 Kaplan note that there may be  an  association between calculi and malignant
 disease of the urinary tract,  the  association is speculative.  An elevated SMR
 for cancer of the  cervix (196) and a decreased SMR due to cancer of  the breast
 (87)  were attributed to factors associated with the low  socioeconomic status
 of the cohort (Hoover et al_.,  V975).  The subcohort (workers employed only in
 shops where PCE was the primary solvent) had only one death from urinary tract
 cancer (1.3 deaths expected).  All deaths from urinary calculi  (2) occurred  in
 this  group.                                                         •
      In  summary, a statistically  significant excess of deaths  from urinary
 tract cancer was observed  in  those workers potentially exposed to both PCE and
 petroleum solvents.  Individuals  employed in shops where PCE was the primary
 solvent  did  not  have an  increased risk of mortality from kidney or bladder
 cancer.   Although  these  findings  do  not  rule out PCE as  the causative agent  of
 urinary  tract  cancer,  the  data suggest that other  factors or agents.may have
 contributed  to  the development of neoplastic disease.
      The possible  relationship between exposure  to petroleum  solvents and
 kidney cancer  has  already  been noted (kitchen,  1983).  An excess  risk of
 bladder  cancer has been  associated with  cigarette  smoking (Matanowski and
 Elliot,  1981).   Brown  and  Kaplan  (1987)  were  not able  to document  smoking
 history; however,  they calculated the possible  effects of smoking  on the  risk
 for bladder cancer (based  on Axelson, 1978).   They concluded  that  the
 three-fold increase in bladder cancer among  the cohort  could  not  be  attributed
 to smoking.
      The epidemiologic studies that are  currently available add limited
 information to our understanding of the  health  hazards  associated with
 occupational  exposure to PCE.  Although  there is some indication that use of
 dry-cleaning solvents poses a health risk,  the contribution of individual
 solvents to the overall  problem is far from clear.  Until studies are
 completed that include a thorough analysis and quantification of PCE exposures,
. the  usefulness of epidemiological studies in the assessment of the human
 nealth risks of PCE will be limited.    :
                                        66

-------
 TOXICITY  TO MAJOR ORGANS AND.SYSTEMS

      Information on  human  toxicity  following  exposure  to  PCE  has  been obtained
 from  case reports of accidental  exposures,  as  well  as  from  a  limited number of
 experimental  studies.  Human-health effects from  short- or  long-term exposures
 are similar to  those observed  in animals.   PCE initially  affects  the.CNS, and
 larger doses  cause various degrees of  hepatic  and renal damage.   Table 5-1
 summarizes human-health effects  resulting from experimental inhalation
 exposures to  PCE.
 Liver
     The earliest reports of PCE-induced  liver damage are associated with its
use as an anti-helminth  in the 1920's and  1930's.  Hundreds of thousands of
individuals were treated with PCE for hookworm infestations.  PCE was typically
given as a single oral dose of 0.12 ml/kg  (maximum of 5 ml) (Reichert, 1983).
PCE reportedly produced  hepatic necrosis on some occasions.  Damage was
transient, and recovery  took place within  1 to 2 wk (Hall and Schillinger,
1925; Lambert, 1933).  Meckler and Phelps  (1966) documented a case in which an
individual developed hepatitis after a massive inhalation exposure to PCE (of
undetermined concentration).  The individual's liver remained enlarged six
months after the exposure.  Interestingly, in a separate case of oral
overexposure reported by Koppel et a_L (1985), .an individual did not have any
measurable liver or kidney damage after ingestion of 8 to 10 ml of relatively
pure PCE.  Clinical measurements of organ function that were within normal
limits included SCOT, SGPT, alkaline phosphatase, red and white Wood-cell
counts, and serum creatinine.  However, the individual was hospitalized for
treatment within 1  h of ingestion, which probably averted organ damage (Koppel
et a].., 1985).
     Stewart and co-workers have investigated several  cases of acute
overexposure to PCE that caused liver damage.   The earliest report documented
the clinical  effects of occupational  exposure to the vapor of a petroleum-based
solvent mixture that contained approximately 50% PCE.   The individual  was
exposed to this mixture for about 3.5 h,  which caused  a loss of consciousness.
Simulation of exposure conditions gave,an estimated  concentration of 250 ppmv,
with levels  that reached 1000 ppmv for the last 30 min.   Nine days after the

                                       67

-------
x>





















LU
Du

O

CU
5-
3
to
O
Q.
cu

c
o
4-»
rd
x:
c
•r-

Id
c
p-
•r*
t-
CL
X
CU
et
s
u


co
U
eu
<*-
cu
x:
4-*
rd
cu
i
c
3
«••
•Jv.
*

1
LO
cu
X)
rd






eu
u
c
cu
cu
eu
cu














4-"
u
cu
<*-
<*-
LU







cu
3 O
CO *f~
O i-
cx cu
X CL
LU









CU
cu E
CO «r-
acn
CD
V-
,





O
4-1

U *>
1 « c3
C CL
CU CU
O *-^
s
•* •—

en en


* *
• •
'rdl "id

•Ml -M
CU1 CU
•M | •
V- U
^3 ^3
3 3
cu cu
4-> 4->
CO CO




C
0
•r—

rd rd
LU T—
LU S-

«*%
o -t ' •—

J
^^" r
1*** • C
3 *"*
—• r»~
O CO
^y\ ^^ o~t o^ ^^ crt
r— - *•— r—

* : » ' **
* • •
15| ^| Id
«LJ| ^^1 luf
cu| cu| cu
1 . 4J 4J
V- J- S-
CM •— r
LO
en
»— •
i— »•
id
r— 1 4->
rd| CU
-Ml 4-> -i
eu| u
— CM •—
UO
» cr* *
• r^ . •
rd Idl
-* r« 4»i|
a> id eu|
P -Ml ' 4->
u cu| s-
3 3 X O) 32CU2
cu eu cu
1 « • 4^ 4~*
CO CO CO

en
T3 J-
c cu
, id xi eu c
E > O
co » O id «i—
co CU OS 3 4->
co co • - i , id
c o i= -a rd 4->
•r- C O CU -Mrd «r-
Q. .r- .1- r— (J V_
CO -4-" <4- CU LU S-
> • -+-> s- t_ rd
en cu eu e M- •— o ' eu >> •—
C x: CU rd U co -M rd
rd [ *
«— f—
U •»— -t-> E co rd «i— CL
rd co id c_ CU >
o en -o e o O -M i-.r- 4->
O i — eu U x: x) cu cu o
Z CO


^
'i
x: **
0
LO -M




cu
j
—
c/;
O
s- a
o x
C|_ CU
OC4-»-l->rd4-» i-Hid CO


.^
•o
x:
o
x: LO -M
^^ ^^ ^^
*



eu cu
v. v_
3 3
CO CO
3 eu cu 3 cu
O -M 4
M O -M
Oi CO CO OS . CO

• M
c <=
O rd
.r- en
. 4-> < — rd
rd id C
4-> -M rd -
•r- C •—
S- O LL.
c- S-
•i— <4- C
-M en
r^ W
3 CU
c ox>
O ••- E
co T- **_ o
3 -M <*_ Q£
M C id T-
10 . •!— «r- T3 • —
1> co c_ rd
• • CU co c_
•i- en c c co o
cu M—
J_ o "- CU C
>, O co eu -r- c:
cu cu x> -M • • xu en
c u E id c cu c .
4->O O O -MOT3"-
C .r- U OS • .r- «i- Id C
O) 4-> co CO -Q V--M eu-r-
•r- CO CU . '
CO CU CO C_ O
c en 3 cu -M
id c c 3 co
s- O ••- o cu
i— u co _i -M :

,
*
— CO V* CO X. rd
id CU -i- CO 4-> -4->X)
E s- en x: c -M
u O eu c cn.r- co
O a >> O •*- re) cu
Z CO LU O — 1 E -M

x: c
| CM 'i
o r-
O • -M CO
-o co •—
x:

x: LO




cu
s_
3
V)
O O O
CL i- a. a. >-
X O X X O
eu M— eu eu <4_
^/ QJ co \y cu cu -^£
2 r— r— • 3 *"" l~m f
•aaecc -o3cc TSS
•r- O
O S -M
4->
LO O
xi "3- . co




CU CO . CU
c_ c_ t_
3 33
CO CO CO
O O O
CL CL Q.
XXX
cu cu cu
cu cu cu
^yt c?) en
c c c
Loiotvotr) LO'LOCOCO in 10 - O) •
-------

























^_^
T3
O)
3
CI
•r™
4-*
C
o
o
x_s


•
^-»
1
LO
Ol
1—


O)
u
(U
cu
"oi
Q£






o
O)
(4-
LU


Ol
i- -0
3 O
in .i—
0 i-
Q. O)
X Q.




m

C Q.
ai a.

Q.
Q.
O
oo
CM













congestion; transient
nausea





























•
m

u
CU
c
CU
Q.
O
Ol
c
>f~
*~~
01 O)
3 <4-
c
oo JZ
Ol
O 01
4-" • -
rd C C
4-> O O
S- 4-> -M
I- Ol rd
•r- CU •—
Ol CU
0)' C
>> O <*-
UJ u o



£
0
m
*""" '


CU
3
0
Q.
X
cu
Ol
c
CO












^
CL
O.
LO
CM
LO
—
r
rd

"cu

CU
O


* «h
C
O
4->
S_
• r-*
CU
00
O
c:
T3
C
rd
UJ

,-

C
'i
o
^""


CU
3
01
O
CL
X
CU
CU
Ol
CO













ex
CL
o
0
VO


c






O  O
o
E 01
01
•o o
O) r—
01 CU
= O
•I— 00
E
•a c"
o
• -.r- C
oo 4-> O
01 rd •!—
CU C 4->
C T- -r-
"- T3 J3
N i- -f-
N 0 J=
•r- O C
•a u .1—
« «

"



























m
"" -
**
CU
C
cu
ex
S-



• 


OJ
3
Ol
X
OJ
at
Ol
c
CO














Ch
CM
LO
2
*
"id

4->
O)

O>
• O

>,
s_
O
rd
s_
Q.
Ol
CU
i_
O)
O.
3
C
o>
UJ

c
'i
CM
' 0
—


«
• 3
01
O
0.
X
CU
O)
Ol
c
CO














o
0


rn
^™
jj"
 4->
C rd <—
O c CU 4->
•.— 4-> U
S- (U
I- CU -r-3
••- 1- J3
3 3
4-> 01 Ol
U O
S- X r—
•+-> LU rd



C
E
LO
^
-

O)
3
01
O
Q.
X
cu
01
Ol
<=
•f—
CO







"






o
o
0
CM

















.'





£
s_
Ol
o
"id
.C
Q.
CU
U
C
0)
o

4-)
O
CU
»— •
ID

UJ
UJ

















.
O
c
rd ~^y
•i- S-
X O
fO ' O
rd
i °
4->
-t-> CZ
C QJ
cu E
E 0)
O) 1-
S- 3
3 01
01 re
rd OJ
OJ E
E

• • i •
4-> 01
01 O)
OJ 4->
4->
C
Ol rd
i- Ol
O) rd
E ri
5 L7
rd 43 U
69

-------
Incident, urinary urobilinogen and serum bilirubin levels were elevated,  an
indication of liver impairment.  On the 18th day after exposure,  a slight
elevation of SGPT was measured (Stewart,et a_L,  1961 a).   In a separate incident
reported by Stewart (1969), a worker was overcome by PCE vapors (of unknown
concentration).   The exposure lasted about ten  minutes and produced hepatic
dysfunction.  Clinical measurements of liver function were normal  shortly after
exposure.  A slight increase in SCOT levels was measured on the third and
fourth days following exposure, and urinary urobilinogen levels were elevated
on the ninth day.  Elevated serum enzymes were  also measured by Hake and
Stewart (1977) after a massive overexposure to  PCE (estimates of the exposure
concentration were not made).  'The individual was found lying in a pool of PCE,
and some was probably absorbed dermally as well  as through the lung.  Recovery
was complete within 21 days.
     Eight of nine firemen exposed to PCE vapors for three minutes (unknown
concentration) had elevated SCOT levels.  Hepatomegaly and splenomegaly were
also found in one individual.  Normal function was regained within 22 to
63 days  (Saland, 1967).                 ,
     The  effects of chronic occupational exposure to PCE vapor were studied in
seven individuals exposed  for  2 toa6.y  (Coler and Rossmiller,  1953).  Short-
term tests indicated  that  exposure levels ranged  from 232  to 385 ppmv.  The
authors  assumed  that  these measurements were representative of levels that
workers ^were  exposed  to 8  h/d, 5 d/wk.  Of  the seven individuals,  three had
abnormal  liver-function tests, and one  was  diagnosed as  having cirrhosis.

Kidneys

      In  animals, kidney damage generally occurs  at  exposure  levels greater.
than  those  that  cause liver  toxicity.   This trend seems  to hold true  for
humans  as well.  Several  cases of  overexposure  great  enough  to cause  loss  of
consciousness have  not produced  measurable  kidney damage (Stewart  et  aj..,
1961a;  Stewart,  1969; Patel  et a]..,  1973;  Patel  et  a]..,  1977).   In one case  of
overexposure  (where an individual  was  unconscious in  a  pool  of PCE for 12  h),
kidney damage was  measured by proteinuria and  hematuria.  These  effects  lasted
for 20 d and  for 8 d, respectively (Hake and Stewart,  1977)..
                                        70

-------
 Lungs
   •  Pulmonary edema has been  documented  in only one  instance.   Exposure to
PCE was probably massive (>1500 ppmv)  because  the  individual was  comatose and
required mechanical ventilation upon admission  to  the hospital.   Recovery was
complete within 4 d (Patel e_t  al_.,  1977).

Skin and Eyes

     Dermal contact with PCE causes localized  irritation; prolonged exposure
can cause erythema, first- and second-degree burns, and blistering (.Gold,
1969; Stewart et a]_., 1961b; Hake and  Stewart,  1977).  PCE vapor  is also
irritating to the eyes (Carpenter,  1937;  Rowe  et al_.,  1952).  Exposure of human
volunteers to 106 ppmv produced transient eye  irritation; this became more
pronounced when the concentration was  Increased to 216 ppmv (Rowe et aJL,
1952).  Stewart et a].. (1970)  has also reported eye irritation when humans
were exposed to 100 ppmv for a period  of  7 h.

Connective Tissue

     Sparrow (1977) described  a patient who had a connective tissue disorder
with similarities to a syndrome observed  in vinyl chloride workers.  The
individual in question was exposed to  PCE vapor during his work at a dry
cleaners.  No measurements of  concentration were reported, but at least once a
week (over a four-year period), exposure was high enough to cause dizziness and
sleepiness.  The individual displayed  pathological changes in the skin of the
hands, acrocyanosis, and polymyopathy.  Abnormalities in the immune system,
and in hepatic function (mild hepatitis) were also documenfed.
     The patient.,may have been abnormally sensitive to PCE, perhaps related to
an existing abnormality in the immune  system.  Indications that the individual
did have an abnormal immune system are suggested by intermittent alopecia
areata (since childhood), vitiligo (an apparently autoimmune condition
characterized by destruction "of melanocytes), and an absence of immunoglobulin
A (IgA).  This app'ears to be a unique'case report, whose information is
                                       71

-------
confounded by the abnormal medical  history of the individual.   Although PCE
exposure may have contributed to the disease, it is not clear  if PCE was the
sole causative agent.

Central Nervous System

    'Acute exposure to PCE generally causes temporary CNS effects such as
dizziness, headache, and confusion.  However, massive single exposures can
cause loss of consciousness (Patel  et a_L, 1973; Patel et a].., 1977; Hake and
Stewart, 1977), and have been fatal in at least two instances  (Lukaszewski,
1979; Levine et al.., 1981).  Protracted exposure produces symptoms similar to
those observed after short-term exposure, although the effects apparently
persist, even after exposure is terminated.
     Rowe et a].. (1952) did not observe any CNS effects in humans exposed to
106 ppmv (single exposure of unspecified duration).  Minor CNS effects were
produced by 216 ppmv (45 min to 2 h), and a 10-min exposure to 600 ppmv
significantly affected motor coordination.  Eleven volunteers  exposed for a
single 7-h peri'od to 100 ppmv experienced headache, dizziness, and somnolence
(Stewart et al.., 1970).  Three of the eleven had abnormal scores on the Romberg
test, which measures ataxia.  Tests of coordination, visual inspection, visual
acuity, and depth perception were normal.  Carpenter (1937) noted only minor
and transient CNS effects in individuals exposed to 500 ppmv for 2 h.  When
the same subjects were exposed to 911 ppmv, they complained of lassitude,
exhilaration, and inebriation.
     In a study conducted by Stewart et al.. (1974), individuals were exposed
to 20  to 150 ppmv of PCE, 7.5 h/d over a 5-wk period.  No alterations in  the
EEG were noted  at 20 ppmv, but some aberrant EEC tracings were seen after
100.ppmv.  A decrease  in  Flanagan Coordination Test scores was observed
following exposure  to  150 ppmv.  In a subsequent experiment (Stewart et  al..,
1977),  individuals  were repeatedly exposed to 0, 25, or  100 ppmv of PCE,
5.5 h/d over 11 wk.  CNS  effects were observed  by  measuring the  subjects'
response on the Romberg,  Michigan Hand-Eye Coordination,  and  Flanagan
Coordination Tests,  and on the EEG.  In  contrast to the  1970  study, exposure
to 100 ppmv did not produce  abnormal scores  on  the Romberg Test.   PCE  at
100 ppmv did cause  a significant decrease  in Flanagan  Coordination  Test scores.
                                       •72

-------
      Coler and RossmiHer (1953) documented subjective compla.ints of malai.se,
 dizziness, headache, light-headedness, and intoxication in individuals
 regularly exposed to concentrations of 232 to 385 ppmv PCE.  Gold (1969)
 described a case history of an individual  exposed to PCE vapors, six to seven
 days a week for three years.   The individual  was hospitalized after exhibiting
 confusion, disorientation,  agitation,  and  an  inability to concentrate.  A
 neurological  examination revealed a normal EEC.   However, his performance on
 psychological  tests that required concentration  was "poor", and he showed
 "marked confusion."  These  problems persisted over a 12-month follow-up period,
 although there was  no further exposure to  PCE (the individual was lost to
 follow up at  this time). Gold (1969)  concluded  that there was suggestive
 evidence of both cerebral and cortical  damage, and basal  ganglia involvement.
 However, it was not possible  to obtain conclusive evidence of any neurological
 damage.  Gregersen  et al_. (1984) examined  65  Danish workers exposed to various
 organic solvents for neurotoxic effects.   Although only 15% of the cohort were
 exposed to PCE (approximately 100 ppmv), the  results are  intriguing.
 Individuals were examined for intelligence, given a neuropsychological  exam,
 and  were subjected  to a  number of neujropsychological  tests.   The authors
 concluded that solvent exposure was  correlated with acute neurotoxic  symptoms,
 as well  as longer-lasting symptoms  of - intellectual  impairment.   A relationship
 between exposure and signs of peripheral neuropathy was also  observed.
     McMullen  (1976)  reported on a  case in which  an individual  was  exposed to
 PCE  vapor (500 ppmv)  for an undetermined period of time.   CNS depression  was
 observed;  the  effects were described as resembling alcohol  intoxication.  No
 clinical  measurements of sensory or organ  function were made,  but  the
 individual  apparently recovered  within six hours.

 Reproductive System

     No  studies  have addressed the question of whether exposure  to PCE affects
 the  human  reproductive system.   At present, there  is no evidence of human
 reproductive toxicity from PCE exposure.         .
                  £
Cardiovascular System
     Some organic solvents have been associated with cardiac arrest due to
ventricular fibrillation.  It is thought that these compounds sensitize the
                                      73

-------
heart td epinephrine-induced arrhythmias.   There is  suggestive evidence that
PCE has this effect in animals (see Section 4),  and  there is one report on
PCE-induced cardiac arrhythmias in humans.   Abedin et al_. (1980) observed that
occupational exposure to PCE probably caused dizziness and premature
ventricular contractions in one case.  Although  PCE may not have been the only
factor in this response, removal from exposure to PCE alleviated the symptoms.

TERATOGENIC EFFECTS
                                                         \
     There is no published information on the teratogenicity of PCE in humans.

MUTAGENIC EFFECTS

     Ikeda  et al_.  (1980) examined  lymphocytes from individuals  exposed to PCE
for  3  months t
-------
for evaluating evidence on the carcinogenic activity of a substance.   The
International  Agency for Research on Cancer (IARC,  1982) separates  strength  of
evidence of carcinogenic activity into four groups:  sufficient evidence,
limited evidence, inadequate evidence, and no evidence of carcinogenicity.
Inclusion in any of these categories is based on data from short-term assays,
as well as animal and human studies (if available).
     The U.S.  Environmental Protection Agency uses  the same groupings, but
places a substance in one or the other category solely on the basis of animal-
bioassay data (U.S. EPA, 1984a).  To assess overall  evidence of carcinogenic
potential to humans, six additional categories are  used: Group A -  Human
Carcinogen, Group B - Probable Human Carcinogen (further separated  into Bl  and
B2), Group C - Possible Human Carcinogen, Group 0 - Not Classified  (due to
inadequate animal evidence), and Group E - No Evidence of Carcinogenicity for
Humans (U.S. EPA, 1984a).
     In the absence of sound epidemiological data,  the greatest weight of
evidence in a carcinogen assessment is typically given to the results of
lifetime animal  bioassays.  The criteria employed in analysis of bioassay data
include an increase in the incidence of tumors in treated animals over those
noted  in controls, a decrease in latency (time to tumor development),
development of rare tumors, and an  increase in the number of tumors in
individual animals.
     In  1982, the  IARC evaluated available information on PCE and determined
that there was "inadequate" evidence  to conclude that PCE is carcinogenic to
humans  (IARC, 1982).  This assessment  was  based primarily on the epidemiologic
study  of Blair et  aj.. (1979).   (An  analysis of animal data gave "limited
evidence" of  carcinogenicity; evidence for any activity of PCE in  short-term
tests  was "inadequate" to  judge  its carcinogenic potential)  (IARC, 1982).   The
U.S. EPA Health  Assessment Document for Tetrachloroethylene  (Perchloroethylene)
(EPA,  1985a)  has also analyzed  the  evidence of carcinogenicity of  PCE.  Thts
evaluation  included an extensive review of short-term test results, data from
animal  tests, and  several  epidemiological  studies.   EPA concluded  that the
evidence  for  the carcinogenicity of PCE  in animals  is "limited," and  that the
epidemiological  data were  inconclusive.   PCE  was placed in Group C, a possible
human  carcinogen (EPA,  1985a);
                                        75

-------
     It should be noted however,  that since these two analyses were published,
other epidemiologic and animal  study results have become available.  It is
possible that a subsequent evaluation by either the U.S. EPA or IARC would
result in a different conclusion.
                                       76

-------
           6.  HUMAN EXPOSURES TO PCE  FROM CONTAMINATED NATER SUPPLIES

      In this section,  we describe  the  procedures we use  to assess human
 exposures attributable to contaminated water supplies.   Efforts to assess
 human exposure to contaminated  drinking water have revealed that significant
 exposures to volatile  organic compounds (VOC's)  can occur from  pathways other
 than the direct  ingestion of water.  Several  researchers  have investigated  the
 relative importance of a variety of  VOC exposure routes  in the  home from use
 of contaminated  water  supplies  (Cothern et a_L,  1986; Shehata,  1985;  Andelman,
 1985).   In addition, there have  been studies  of  the contribution to indoor
 exposures of waterborne radon-222, another highly volatile substance  (Hess
 et al_., 1982;  Prichard and Gesell, 1981).  These studies  indicate that  exposure
 to volatile  chemicals  from routes other than  direct ingestion of fluids may be
 as large as  or larger  than exposure  from ingestion alone.   These other  routes
 include inhalation from indoor  air of  contaminants mobilized by showers, baths,
 toilets, dishwashers,  washing machines,  and  cooking;  ingestion  of contaminants
 in food; and dermal  absorption of contaminants while  washing, bathing,  and
 showering.  We divide  exposures  attributable  to  contaminated ground water into
 ingestion, inhalation,  and dermal absorption  pathways.  Our discussion  of human
 exposure to  PCE  is divided into  three  subsections.   In the first subsection,
 we provide a background on the general  approach  we use for assessing  human
 exposure to  VOC's.  This approach addresses  relative  contributions  from
.ingestion, inhalation,  and dermal absorption.  Our focus  in this  subsection  is
 on the  type  of information that  is needed for risk assessments.   The  second
 subsection covers  the  method used to estimate *ingestion,  inhalation,  and
 dermal-absorption  dose  factors.  These  factors convert water concentrations  in
 mg/L into human-population exposures in  mg/kg-d.   The third subsection  presents
 our calculations of the magnitude and distribution  of human exposures
 attributable to  PCE contamination in California  groundwater supplies.
             +
 BACKGROUND ON  HUMAN-EXPOSURE ESTIMATES
      The  purpose of the human-exposure estimate is to provide a distribution
 of  population dose of a chemicalfrom the various phases of the environment.
 Three primary pathways must be addressed — inhalation, ingestion, and dermal
                                      77

-------

-------
                          B.4
EXPOSURE ASSESSMENT
               AtmDUTe 2  The specific populations and subpopulations that are the
                              subjects of the assessment are clearly identified,,and the
                              reasons for their selection and any exclusions are given.
                   SOURCE  Case Study D. Formaldehyde (Pages' 6-2 to 6-4).
                        Note  References to previous studies are given as background and as a basis
                              for selecting two.population groups for study.

-------
         Assessment of Health  Risks
to Garment Workers and Certain Home Residents
        from Exposure to Formaldehyde
                  April 1987
  Office of Pesticides and Toxic Substances
     U.S.  Environmental Protection Agency

-------
 derivatives are irreversibly Corned and usually contain only
 residual levels of unreacted HCHO.  Under extreme conditions,
 such as very high temperatures or highly acidic conditions,
 of  the derivatives may degrade and release HCHO.
      HCHO's major nonconsumptive uses are (1)  disinfectant,  (2)
 preservative,  (3)  deodorant,  and (4)  .textile' and oaoer uses.
      The major" pseudo-consumptive uses are (I)  urea-HCHO resins
 which  are  used: in  fiberboard,  particleboard,  plywood,  laminates,
           %                    •
 urea-HCHO  foams,  molding  compounds,  and paper,  textiles, and
 protective  coatings;  (2)  urea-HCHO concentrates  which  are  used to
 produce  time-release  fertilizers,  and (3)  hexanethylenetetramine
                                                     i
 which  is used  as a  special  anhydrous  form  of  HCHO to cure  resins
 and  to  treat textiles  and-rubber.
     The major  consumptive  uses  are  (1)  melanine-HCHO  resins
which  are used  for  molding  compounds,  fiberboard,  particleboard,
plywood, laminates, paper and  textiles,  (2) phenol-HCHO resins
which  are used  in  fiberboard,  particleboard, olywood moldina
compounds, and  insulation;  (3) nentaerythritol which is  used to
produce alkyd  resins.  (4) 1,4-butanediol which  is  used  to  produce
tetrahydrofuran, (5) acetal resins which are used  in the
manufacture of  engineering  plastics,  and  (6) trimethyloloropane
which is used  in the production of urethanes.
6.2.   Batiaates of Current Human  Exposure
     To obtain  estimates of human  exposure to HCHO, the  Agency
commissioned a contractor study  (Versar,  1982).  This  studv
integrated the  existing monitoring data, engineering or  -nodelir.^
                                                   •»,
                               6-2

-------
estimates, use data, population estimates, and assessment of the


likelihood of exposure from HCHO-related activities into an


exposure assessment detailing those activities having a high HCHO


exposure potential.  EPA updated some portions of this assessment


to reflect new data received in response to the FEDERAL REGISTER


notice of November 18, 1983 and other data gathered by EPA.  The


combined data were used as the basis for the May 1935 draft risk


assessment.

                                                      #
     Subsequent to the draft risk assessment, the Agency


commissioned additional contractor studies to assess garment


worker  (PEI, 1985) and residential (Versar, 1986a,b,c) exposure


to HCHO in more depth.  The, exposure estimate's from these reports


were used as the primary basis  for this risk assessment.  The


conclusions of these contractor reports are summarized in this .


document; more detailed information  regarding exposure can be


obtained by referring to the contractor reports.


6.3.    Populations  at Risk


     The  two populations at risk examined here are certain home


residents and garment workers.


6.3.1.    Home Residents


     Based on a projection  of  manufactured housing starts by


Schweer (1987), it  is estimated that 7,800,000 persons may occupy


new manufactured homes during  the next  ten years.  This  figure


assumes 295,000 starts per  year and  2.64  persons per home.


     Similarly, an estimated 214,000 new  conventional  homes


containing significant quantities of pressed  wood  products as
              «"•     .

construction materials will be started  each  year  for  the next  ten
                                6-3

-------
 years with an occupancy rate of 2.95 persons  for  a  total of
 6,310,000 persons.
 6-3.2.   Garment Workers
      The number of potentially exposed garment workers is
 estimated to be 777,000 (Versar, 1982) out of 1,100,000 workers
 employed in the U.S. apparel industry (Ward, 1984).  This figure
 may drop in the future due to-increased foreign competition and
 the introduction of labor saving equipment.
 6.3.3.    Summary
                                            «
      Table 6-1  presents population estimates ,for the two  housing
 segments.   Assuming that the number of potentially exposed
 garment  workers  remains steady  at  777,000,  then a  total of  almost
 15,000,000  persons  over the  next ten years  may have  the potential
 to  be exposed to elevated  levels of HCHO.
                            Table 6-1.
                       POPULATIONS AT RISK
   Category
Manufactured homes
Conventional homes
* Schweer (1987)
per yr
779,000
631,000
Population
Estimates	
        10 yraT
      7,790-, 000
      6,310,000
                               6-4

-------

-------
                          B.4

EXPOSURE ASSESSMENT
               Attribute 3
Available data are considered and critically evaluated, and the
degree of confidence in the data expressed. (Reasons for any
data exclusion are presented.)
                  SOURCE  Case Study I. Para-dichlorobenzene (Pages B-3. to B-5).


                       Note  Limitations in drawing conclusions from the data in a large multi-
                             chemical monitoring study are pointed out.

-------
                ASSESSMENT OP HUMAN CANCER RISKS
                    FROM PARA-DICHLOROBENZENE
                        FEBRUARY 13- 1987
Prepared by Barbara  Mandula,  Project  Manager   Karl Baetcke
Diane Beal, C.J.  Nelson,  Vanessa  Rodriguez, Gary Thorn
Risk Analysis  Branch  Office  of Toxic Substances  EPA.

-------
                               a - 3
                l)
Measurements
                     a)   TEAM Study
                          Extensive monitoring of PDCB exposure
 was carried out as part of.EPA's Total Exposure Assessment
 Methodology (TEAM) study (Pellizzari jt .al_. ,  1985).   The  TEAM
 study used personal air monitoring devices to measure the
 concentration of 20 organic compounds in daytime and  overnight
 air samples among 600 persons selected in four geographical
 regions.   The samples were  taken in two consecutive  12-hour
 intervals.   In addition,  exhaled breath samples were  taken at the
 end of the 24 hours.
                          Although meta- and  para-dichlorobenzene
 were  measured together,  the values until now  have been assumed  to
 represent  PDCB,  since there were no known common sources  of  the
 meta-  compound.   However, a recent  experiment found  that a
 liquid  deodorizer contained approximately equal amounts of meta
 and para isomer  (Wallace, 1986c).   Therefore,  the TEAM
 measurements  may  in some  cases  reflect meta-  plus para- isomers
 of dichlorobenzene,  and therefore  represent an upper  limit on
possible PDCB levels.   It is  likely that most .of the:  measured m-
 »p-DCB  in the homes  sampled represented PDCB  since production of
the meta- compound is  approximately an order  of magnitude lower
than that of  the  para- compound  (U.S.  EPA,  1985).
                          Tables  B-2 and B-3 show the  TEAM results
 for daytime and nighttime personal  air samples.   Daytime  and
overnight exposures  are similar.   It is unclear to what extent
these findings reflect differences  in  exposure,at home and away.

-------
 0)
 M

 0)
£>
 O
 u
 O
 o
•r*
Q
 I
 a

IU
 o
 4)
*-4
 a
 s
 10
V)
 a

 o
 (0
 — 4

O
CN

3

1
0)
C
3

in
r-t





CN

z

o

in
i
CN
CN
O


O
CN
CN



0
CN
i-4

O
CN
"*

vO
•
en
in



00
03
ON
— i

_Q
0)
EC4
C
"
ON






en'

§

0 0
en CN . fn
CN f» T '
1 1 1
ON eN .en
Tf ~4 O
0 0 O


< < O
Z Z O
— 1


000
CN in en
in r» r»


o o o
en ON vO
-4 CN CN

vO ON
, J3
0 (0  0)
cu u <
Q O J

O -i
in in
CN ^4
i i
in in
0 0
o o
o
•
O ^0
f^ ^
^4


^5 , ^^
^^ ^\
CN CN


t en
03 in
0 O
.'
CN
»* I~"
i-4



03
ON
i—4
^*
0) 03 .
C ON
3 -l
1 0)
>i C
(0 3

^4 CO
in o
i i
o \o
ir> ^

en
0
U
*0
S-l
CN -P
C
< o
3 o
                                                                                                        in
                                                                                                        00
                                                                                                        ON
0

XI
(0
                                                                                                  fl

                                                                                                  rtj
                                                                                                  o
                                                                                                  z
        0)
        cu

-------
•o
3
•U
CO
s
3 s
_ C
5 5
u "•
U-l
*"£
^ <5) of
$ -. in
•H CIN
c *i
.5 c
« 8
c !*
s *-
i £
0)

O
u
o
-•§ *
S £
? 1
• a s
0
« Z
0> •|~l
-U *•
1 i S
to ' * ^
•^
^ <
-^
I «
2 o|
f fil
e
(0
•
ro
i
03
J O
51-1
0
.a
U} CO
^ c
'-I CN -H <1) -H CN
 o
2 ^ i-j 
-------
        Ficjure
       POMHXnOM tXCttOHM CONCJWTRATION SHOWN
             iifjDi   t»sat  iu»  u»
1000


1.000
  i  m,p-DICHLOROBENZENE
100
                                  NWMT
           <.ICINO
           > MMONM. AUI j
            •IO.MM    !
          -. ourooe* urn
            N-.M1
                                         i.ooi
                                          100
                                               I
                                               N
                                  .NIGHT
                   /
              ~0*    90%    «%   W*
              12JOO    **OOD   115JOO 127.000
          KJWUkTION MUOW* CONCtNTHATKJN SHOWN

                              nonj of per»n»l exposure*.
       collected in the vic.mtv of tne participant*
        Samples  taken  Fall,   1981

        Source:   Wallace et,al.,  1985

-------
                              B - 4


although-overnight samples would be expected to reflect primarily

indoor horn* air.  In some cases, daytime measurements may also

reflect home air to a great extent, depending on an individual's

daily routine.  -The populations sample'd were chosen statistically

to reflect a much larger population.  Thus, for example, the

sample included children (older than 7 years old), and one-half

of the persons sampled were not in the workforce.

                         The distribution of concentrations

provides some indication of range of exposures.  Thus, the median

value is-close to the outdoor measurements, which are usually

below 2 ug/m3, whereas the 75th and 95th percentile values are
                                            £               •   '
higher.  Approximately a 10,000-fold range  was  found  for measured

values of this compound.  Figure B-l shows  the  distribution-in

New Jersey.

                         For the samples  in New Jersey,  if. we

take 50 ug/m  as an arithmetic mean concentration of  daily

personal exposure, and assume an -inhalation rate of  1 m^/hour,  we

get a daily intake of about  1.2 mg. The nighttime median  is about

3.5 ug/m3 and the 75th percentile  about  13  ug/m3.  Mean 24-hour

values in the other sampled  areas  are  6  to  18  ug/m3,  giving a  24-

hour daily  intake of 0.14  to 0.42  mg/day.   The 12-hour  75th

percentile  concentration values  excluding New Jersey range from-

2.5 to 9.4  ug/m3, and the  median ranges  from 0.5 to  2.9 ug/m3.

                         The TEAM measurements presumably include

some persons  not  using  any PDCB products at home, as-well as  some

using  PDCB-products  in  several  locations.   The 10,000-fold range

in  nighttime  personal exposures is consistent with this

-------
                               3 ,-  5





explanation  (Wallace,  1985).   There  is  essentially no information



on whether the persons with high exposure  were  using PDCB



products.  Follow-up studies are planned in  the near future to



determine the sources of  PDCB  in homes  and their overall



contribution to PDCB loading.  Such  follow-up studies might



provide an explanation for the higher  levels  found in New Jersey



as compared with the other sampled areas.



                    b)    Home  with Known Source



     In a recent experiment related  to  the TEAM study, Wallace



(1986c) reported measurements  of PDCB  in a home with a known



source.  Two consecutive  8-hour measurements  established a



baseline indoor PDCB concentration of  about  1 ug/m3.  A room air



freshener containing PDCB was  then opened  and PDCB levels



measured in the home, and detected by  personal  monitors, averaged



300 ug/m3 over the next six 8-hour measurement  periods.  Breath



concentrations of the residents increased  approximately linearly



over the same two-day period to 50 ug/m3.



                    c)    Concentration  of  PDCB  in Public Restrooms



                          In another  short-term  experiment,



Midwest Research Institute (MRI 1986)  determined PDCB



concentrations and related variables in two  public bathrooms.



Figures B—2 and B-3 provide information about the bathrooms.



Bathroom 1 had one urinal block in the open  air on the  floor,   .



whereas bathroom 2 had two blocks, one  in  a  urinal and one  in(  a



toilet bowl.  These experiments were carried out  in  large well-



ventilated bathrooms under conditions  that do not necessarily



reflect the conditions that would  be encountered  in  a home.   In

-------
                          B.4

EXPOSURE ASSESSMENT
               Attribute 4
If models are used, their bases are described, along with their
validation status.
                   SOURCE  Case Study I. Para-dichlorobenzene (Pages B-8 to B-ll).
                       Note  The limitations of the model are discussed. None of the case studies
                              treated model validation status extensively.

-------
                 ASSESSMENT OF HUMAN CANCER RISKS

                     FROM PARA-DICHLOROBENZENE
                         FEBRUARY 13-  1987
Prepared by Barbara  Mandula,  Project Manager  Karl Baetcke
Diane Beal, C.J.  Nelson,  Vanessa  Rodriguez,  Gary Thorn
Risk Analysis  Branch-  Office  of Toxic Substances- EPA.

-------
                              B - 8

                         For Table B-4, worst case and be'st
estimate values are given for adults, and  for children
approximately one year old.  The calculations are based on the
modeled .concentration in a given room, an  estimate of time spent
per day in that room (1-14 hours) and the  indicated inhalation
rate.  In Table B-5 best estimates for exposure are obtained by
considering the concentration throughout a house, the residence
time in the house (1-14 hours/day) and the indicated inhalation
rate.
                         Even using the house concentrations, the
values obtained are in the highest range of those found in the
TEAM study.  Therefore, it seem likely that these models do not
adequately reflect the actual exposures that people experience
from PDCB.
                    b)   Simplified Model   .
                         Wallace (1986a) has suggested a
simplified equation that may be applied to PDCB sources in a
home.  The equation
          cin =S/V(a+k)
where C^n is the indoor concentration  in  ug/m3
     S is th« source generation fate in ug/hour
     V ia the volume of the home in m3
     a is the air exchange rate/hour
     and k is a decay constant due to  absorption,  chemical
reaction etc in h~^ , between 0 and 1.

-------
                              B - 9

                         This is the basic equation that can be-
used to estimate PDCB concentrations in homes with sources.  The
source generation rate may be estimated from knowledge of the
Loss rate as provided by the manufacturer.  The volume of the
home and the air exchange rate may be estimated or measured.
However, little is known about the magnitude of the decay term k,
representing the loss of the vapor through adsorption or other
chemical processes.  Presumably some absorption on fibers occurs,
since clothes that have been recently stored near moth crystals
retain odors.  Absorption on surfaces such as paint or wood is
also possible.
                         A second major problem in employing this
equation is our lack of knowledge of air  flow in homes.  The
above equation assumes perfect mixing within the home, whereas in.
fact the mixing is affected by many factors including the type of
ventilation system (central forced air provides better mixing
                            «L
than other systems),  use of fans, presence of closed doors
between interior rooms, etc.  A "mixing factor" jn can be defined
that acts as a multiplier to the air exchange rate.  The value of
m may range between 0.1 and 0.5 in practical situations.
Alternatively, we may simply use somewhat smaller values of the
air exchange rate in the above equation to,take into account
imperfect mixing.  Similar concerns apply to any model.
                         Without better knowledge of absorption
processes and mixing factors, we cannot hope to use models  to
provide accurate predictions of observed  concentrations.
However, by assuming realistic values for the various model

-------
                              B  -  10

parameters./ we may establish a range of concentrations that can
then be compared to measured values.
                         We can now estimate concentrations for
various scenarios, using suitable values from Table B-7.
                    Worst Case
                    A small tight apartment with strong sources
of PDCB and no materials to absorb the PDCB vapors.  V = 100;
k = 0.0; a = 0.1; S = 100 g/month.  We convert S to ug/h by
dividing by about 720 h/month.
                    C =   140,000 ug/h     14,000 ug/m3
                        100 m3 x 0.1 h'1

                    Typical Case
                    An average-size home without special
insulation.  V = 400; k = 0.5; a = 0.5; S = 10 g/month.-
                          14,000 ug/h
                    400 m3 X (0.5 •)- 0.5) h"1
= 35 ug/m
This calculation yields 35 ug/m3, or about 840 ug/day.  This value
is in the range of the observed arithmetic mean  values  of  from  5
to 78 ug/ra3 found by the TEAM study.  Furthermore,  this value  is
1/3 to 1/12 of that estimated by the house method  in  Tables  3-5
and B-6.
                    The simplified  equation  was  also  applied to
the MRI bathroom in which a value for S  of 102 g/month had  been
measured.  The concentration was calculated  as 400 ug/m3;  the
measured value was 600 ug/m3.   It appears that the simplified

-------
         B-7.   Parameters  for Simplified Indoor Air Model
Parameter
Volume (m3)
Air exchange rate (h"1)
Source generation
rate (g/ month)
Decay rate (h"1)
Symbol
V
a
S
k
Low
100a
0.1
1
0.0
Medium
,400b
0.5
10
0.5
High
700
1
100
1

.0

.0
a Small apartment (about 400 sq. ft.)



b Average new home (about 1700 sq. ft.)





From Wallace, 1986c.

-------
                              B - 11


 model  yields  concentrations at  least  in  some  cases,  that  agree

 with measurements.

                3)    Number of Households  Using  PDCB  Products

                     The  number  of households  using PDCB as  space

 deodorants, toilet  deodorants,  or as  moth repellent  is not  -
                           «*
 known.    We can get some estimate of  the  numbers  of  households

 using  these products by  assuming 4 million kg/year total  used  for

 each of  the three uses,  80% of  use by consumers,  and reasonable
                                                  ,/•
 estimates  for  numbers of grams/month/household.   Also, to the

 extent that the exposures in the TEAM study indicate household

 use, we  can presume that approximately one-half the  households

 may  be exposed  to close  to ambient levels.

                     Simmons (1982) reports that about 25  %  of

 households, or  about 20  million households, use "in-bowl  toilet

 fresheners."  Two-thirds of households use air  fresheners and

 room deodorizers; approximately 25%,  or 20 million, households,

 use a solid product.  It is unclear to what extent these  products

 represent  PDCB  use, but  it seems likely that  much of the  use of

 these solid products reflects PDCB use.

                     Using the data on total consumption,  if we

assume 10  g per  month/household for toilet deodorant, we  get.25

million households  using the product; assuming  16 g  per month  for

space deodorant  gives 16' million households using space

deodorant, and  37 g per  month (a one-pound package per year) of

moth repellent  gives 7 million  households.  If  the

use/month/household is overestimated, then more households  are

using the  products;  if  the use/month/household, is

-------

-------
EXPOSURE ASSESSMENT
              Attribute 5
Potential sources, pathways, and routes of human exposure
are identified and quantified; the reasons why any are not,
included in the assessment are presented.
                 SOURCE  Case Study K. Tetrachlproethylene (Pages 4-8; 73-93).

-------
                                UCRL-15831
   Health Risk Assessment
of Tetrachloroethylene (PCE)
in California Drinking Water
 K. T. Bogen, L. C. Hall, T. E. McKone,
    D. W. Layton, and S. E. Patton
   Environmental Sciences Division
Lawrence Livermore National Laboratory
       University of California
           RO. Box 5507
        Livermore, CA 94550
           April 10, 1987

            Prepared for
  California Public Health Foundation
            P.O. Box 520
         Berkeley, CA 94701

-------
                       structyr? °f  tetrachloroethylene. alternative names
                numbers, empirical  formula, and molecular weight.
 Chemical structure:
Cl - C - C - Cl
     Cl   Cl
 Empirical formula:  C2C14             Molecular weight:  165.85
 Chemical Abstracts Service registry number:  127-18-1
 NIOSH Registry of Toxic Effects of Chemical Substances number:  KX3850000
 Alternative names:  PCE, Perc,  tetrachloroethene,  perchloroethylene,  ethylene
                     tetrachloride,  carbon dichloride
 Common trade names:  Antisol,  Dee Solv,  Per Sec,  and Texranec.
 ENVIRONMENTAL TRANSPORT AND  TRANSFORMATION

      Tetrachloroethylene tends  to  partition  primarily  to  the  atmosphere.   It
 has  been  estimated  that 85 to 90%  of  the  PCE produced  is  eventually released
 to the  atmosphere (U.S.  EPA, 1985a; HHO,  1984).  The key  properties of PCE
 that affect  its movement in  the environment  are its high  vapor pressure and
 low  solubility in water.
      GEOTOX  (McKone and  Layton, 1986) was used to estimate the equilibrium
 distribution  of PCE in  air,  soil,  and water.  GEOTOX is %a multimedia
 compartment model that  simulates the environmental transport and transformation
 of a chemical, based on  its  physical and chemical characteristics and the
 properties of the landscape  into which it is released.  A simulation of the
 environmental partitioning of PCE was run using California landscape data,
 properties of PCE (see Table 2-2), and PCE source-emission data from the
 California Air Resources Board (CARB).  The PCE source term was represented by
 an annual  release of 1.83 x 107 kg/y over an area of 411,000 km2 (CARB,
 1984); of this source,  10% is assumed released into soil,  1% to surface  water,
and  the remainder directly to the atmosphere.  The simulated equilibrium
distribution of PCE  is  shown  in Fig. 2-1.   Most of the PCE released to the
environment is found in the atmosphere.   However,  the equilibrium
distributions, 86% in  the air and  11% in  surface water, reflect both  the

-------
     Atmospheric gas
                                 86%
                                 Atmospheric particles

                                        4 x 10~8%
             Bio miss
                    0.02%
      Upper soil
             0.4%
     4
                       t
      Lower soil
               0.5%
                        •
|	
      Ground water
     6
                             1%
Surface water

         11%
Sediments
                                                                1%
                                                      8
Figure 2-1.   Environmental distribution of PCE under  steady-state conditions.
Partitioning  between compartments  Is  predicted by the computer model GEOTOX
(McKone and-Layton,  1986).
                                        .5

-------
  Table  2-2.  Chemical, physical,  and organoleptic  properties of
  tetrachloroethylene.
 Property
   Units
Value
                                                           Reference
 Boiling point at 760 mm Hg     8C
 Freezing/melting point         °C
 Density at 20°C                g/cm3
 Vapor pressure at 20°C         mm Hg
 Henry's law constant at 20°C   atm-m3/mo1
 Conversion factor              mg/m3-ppmv
 Diffusion constants
   at 1  atm, 20°C
     Air                        m2/s
     Water                      m2/s
             121
             -22.4
               1.65
              15.8
               0.0227
               6.89
 Solubility in water at 258C    mg/L
           7.4 x  10-6
           7.6 x  10-10
            150
 Log octanol/water
   partition  coefficient
 Odor threshold  in  water
Unitless
mg/L       3.0 x 1Q-1
 3.14
 2.46
              Hawley (1981)
              Hawley (1981)
              Hawley (1981)
              Sittig (1985)
              Mackay and  Shiu  (1981)
              Verschueren* (1983)

              Lyman  et al.  (1982) *
             Mackinson et al.
             (1981)
Leo (1983)
Callahan et al.. (1979)
Zoeteman et a_K (1974)
 relative magnitude of the source  (89%  to air and  1% to water) and the
 effective  residence times.  The loss rate of PCE  in air is an order of
 magnitude  greater than that in surface water.  This accounts for the apparent
 "enrichment" of PCE in surface water.
Air
     PCE in the atmosphere is subject to relatively rapid chemical or
photochemical degradation.  In the troposphere, it photodegrades, ultimately
leading to the formation of hydrochloric acid, trichloroacetic acid, and carbon
dioxide in the presence of atmospheric water (U.S. EPA, 1985a).  PCE can also
be removed by scavenging mechanisms,  primarily through hydroxyl radicals
(Dimltriades et a!., 1983).   Estimates of Its atmospheric residence time are
on the order of one year or  less (see U.S.  EPA, 1985a).

-------
     Singh et al-  <1981>  compiled monitoring data for the concentrations  of
several  volatile organlcs 1n ambient air and found that for the western  half
of the U.S. the average PCE concentration was 4.3 jig/m3 and the overall  range
was 0.23 to 51.6 ng/m3.  The U.S. EPA (1985a) reported ambient air PCE^
concentrations 1n California (1972-1980) ranging from 0,2 to 19.0 jig/m .   The
California Air Resources Board (Nystrom, 1986), based on preliminary data,
found average ambient air PCE concentrations for several California locations:
Los Angeles 1180 * 900 parts-per-trillion by volume (pptv) (8.1 ± 1.2 Hg/m );
San lose 490 ± 330 pptv (3.4 ± 2.3>g/m3); Long Beach 1030 * 560 pptv
(7.1 ±3.9 vg/m3); Stockton 450 ± 170 pptv (3.1 ± 1.2 yg/m ), and Si mi
Valley 330'* 250 pptv  (2.3 ± 1.7 iig/m3).  These data indicate that PCE
concentrations  in the  ambient air of urban areas are higher than those in
rural areas  (or less densely populated.areas).
 Hater

      In surface waters,  PCE  rapidly  volatilizes  into  the  atmosphere.  Wind
 speed,  agitation of the  water,  and water  and  air temperatures  affect
 evaporation rates.   Photodegradation,  in  contrast,  is a slow decay  process  and
 does not appear to  be an important transformation mechanism in water.   The
 half-life of PCE in shallow  water due  to  volatilization has been  estimated  at
 24 to 28 rain in laboratory experiments (Dilling  et al-,  1975).  Zoeteman
 et al.  (1980) measured PCE persistence in surface waters  of the Netherlands
 from 3 to 30 days (half-life),  while in lakes and ground  waters,  the  half-life
 was estimated to be 10-fold  higher.
      In ground water, PCE is relatively persistent, with  degradation  occurring
 through hydrolysis and biotransformation.  It is denser than  water as an
 undissolved liquid, consequently it tends to sink in-ground water.   Vogel and
 McCarty (1985) have shown that PCE biotransforms to  trichloroethylene (TCE),
 dichloroethylene, and vinyl  chloride via reductive dehalogenation under
 anaerobic  conditions.  They further suggest  that the potential exists for the
 complete mineralization of PCE to carbon dioxide in  aquifer systems.   The
 half-life  of PCE due  to aqueous  hydrolysis in natural waters  can be on the
 order  of months  (Oilling  et alv 1975) to several years  (Pearson and
 McConnell,  1975).

-------
      The U.S.  Environmental  Protection Agency (1985a) reported a mean PCE
 concentration  of 1  yg/L from 1102 surface water measurements in 45 states
 (from August 1975 to September 1984).   An important source-of data on the
 concentrations of PCE in drinking water supplies is a survey of large water
 utilities  in California (i .e.,  utilities  with more  than  200  service
 connections) that was conducted by the California Department of Health Services
 (1986).   From  January 1984 through^ December  1985, the Dwells  in 819 water
 systems  were sampled for contamination by organic chemicals.   The  water systems
 considered  included  a total  of 5650 wells, 2947 of  which  were sampled.  The
 wells  sampled  were  selected  based on the  likelihood of contamination.   PCE  was
 found  in  199 wells  in concentrations up to 166  jjig/L,  with  a-median
 concentration  of  1.9 jig/L.   Generally,  the highest  fraction  of contaminated"
 wells  and the  wells  with  the highest concentrations  were  found in  the  heavily
 urbanized areas of the  state..   Contamination  was state-wide.   Los  Angeles
 County registered the greatest  number  of  contaminated wells  (i.e.,  140).

 Soil
     There is limited information .on the behavior of PCE in soil.  The solvent
can be adsorbed to soil or leached through soil when dissolved in water c-r as
a separate organic phase (as in large spills).  PCE associated with soil air
or soil water is more mobile than the absorbed portion (Schwarzenbach and
Westall, 1981).          *
   * The adsorption of PCE to soils appears to be correlated to its octanol/
water partition coefficient, the organic carbon content of the soi.l,  and the
concentration of PCE in the liquid phase.  PCE appears to leach rapidly
through soils of low (<0.1%) organic carbon content (U.S. EPA, 1985a;
Schwarzenbach and Westall,  1981).
     Several  studies have documented the mobility of tetrachloroethylene in
soil/groundwater systems (Piet et  §_[.,  1981;  Schneider et a].., 1981;
Schwarzenbach and Hestall,  1981).   Wilson et  aK  (1981) showed that most of
the chemical  was lost from the soil  via leaching  or volatilization to the
atmosphere.   Persistence in soil  ranges from  months to years.

-------
     Coler and Rossmlller (1953) documented subjective complaints of malaise,
dizziness, headache, light-headedness, and intoxication in individuals
regularly exposed to concentrations of 232 to 385 ppmv PCE.   Gold (1969)
described a case history of an individual  exposed to PCE vapors, six to seven
days a week for three years.  The individual  was hospitalized after exhibiting
confusion, disorientation, agitation, and  an  inability to concentrate.  A
neurological examination revealed a normal EEC.   However, his performance on
psychological tests that required concentration  was "poor",  and he showed
"marked confusion."  These problems persisted over a 12-month follow-up period,
although there was no further exposure to  PCE Cthe individual was lost to
follow up at this time).  Gold (1969) concluded  that there was suggestive
evidence of both cerebral and cortical damage, and basal ganglia involvement.
However, it was not possible to obtain conclusive evidence of any neurological
damage.  Gregersen et- al_. (1984) examined  65  Danish workers  exposed to various
organic solvents for neurotoxic effects.  Although only 15% of the cohort were
exposed to PCE (approximately 100 ppmv), the  results are intriguing.
Individuals were examined for intelligence, given a neuropsychological exam,
and were subjected to a number of neuropsychological tests.   The authors
concluded that solvent exposure was correlated with acute neurotoxic symptoms,
       ft.
as well as longer-lasting symptoms of intellectual impairment.  A relationship
between exposure and signs of peripheral neuropathy was also observed.
     McHullen (1976) reported on a case in which an individual was exposed to
PCE vapor (500 ppmv) for an undetermined period  of time.  CMS depression was*
observed; the effects were described as resembling alcohol intoxication.  No
clinical measurements of sensory or organ  function were made, but the
individual apparently recovered within six hours.

Reproductive System

     No studies have addressed the question of whether exposure to PCE affects
the human reproductive system.  At present, there is no evidence of human
reproductive toxicity from PCE exposure.

Cardiovascular System                                .
     Some organic solvents have been associated with cardiac arrest due to
ventricular fibrillation.  It is thought that these compounds sensitize the
                                      73

-------
heart to epinephrine-induced arrhythmias.  There is suggestive evidence that
PCE has this effect in animals (see Section 4), and there is one report on
PCE-induced cardiac arrhythmias in humans,  Abedin et aK (1980) observed that
occupational exposure to PCE probably caused dizziness and premature
ventricular contractions in one case.  Although PCE may not have been the only
factor in this response, removal from exposure to PCE alleviated the symptoms.

TERATOGENIC EFFECTS

     There is no published information on the teratogenicity of PCE in humans.

MUTAGENIC EFFECTS

     Ikeda et a_L (1980) examined lymphocytes from individuals exposed to PCE
for 3 months to 18 y.   Chromosomal  aberrations, sister-chromatid exchanges,  and
alterations of the mitotic index were the cytogenetic  effects studied.   The
exposure lev.el  in one group of workers was 92 ppmv (geometric mean), while a
second group was exposed to 10 to 40 ppmv (the authors did not give TWA
exposure concentrations).   Although a control  group was  included,  the criteria
usedJ:o select and/or match controls was not reported.  Exposed individuals  did
not have a significantly greater frequency of chromosomal  aberrations or
sister-chromatid exchanges, nor were there any substantial  differences  in  the
mitotic index.
     Trichloroethanol  is a metabolite of PCE isolated  from the urine of humans.
                                                                    #
Gu et §1.  (1981) reported  a slight  increase in the number of sister-chromatid
exchanges  per cell  in  human lymphocytes  exposed to 178 mg/L  of
trichloroethanol.  Neither of these studies is adequate  to assess  the
mutagenic  potential  of PCE and its  metabolites in  humans.

SUMMARY OF EVIDENCE OF HUMAN CARCINOGENICITY

     Evaluation of the carcinogenic potential  of a chemical  is  based on the
results of short-term  assays of mutagenesis,  pharmacological  data  (e.g.,
distribution and metabolism),  lifetime animal  bioassays,  and  epidemiological
evidence.   Several  agencies and groups have developed  systems  of classification
                                       74

-------
for'evaluating evidence on the carcinogenic activity of a  substance.   The
International  Agency for Research on Cancer (IARC,  1982) separates  strength  of
evidence of carcinogenic activity into four groups:  sufficient evidence,
limited evidence, inadequate evidence, and no evidence of  carcinogenicity.
Inclusion in any of these categories is based on data from short-term assays,
as well as animal and human studies (if available).
     The U.S.  Environmental Protection Agency uses  the same groupings, but
places a substance in one or the other category solely on  the basis of animal-
bioassay data (U.S. EPA, 1984a).  To assess overall  evidence of carcinogenic
potential to humans, six additional categories are  used: Group A -  Human
Carcinogen, Group B - Probable Human Carcinogen (further separated  into Bl  and
B2), Group C - Possible Human Carcinogen, Group D - Not Classified  (due to
inadequate animal evidence), and Group E - No Evidence of Carcinogenicity for
Humans (U.S. EPA, 1984a).
     In the absence of sound epidemiological data,  the greatest weight of
evidence in a carcinogen assessment is typically given to the results of
lifetime animal  bioassays.  The criteria employed in analysis of bioassay data
include an increase in the incidence of tumors in treated animals over those
noted  in controls, a decrease in latency (time to tumor development),
development of rare tumors, and an  increase  in the number of tumors in
individual animals.
     In  1982, the IARC evaluated available  information on PCE and determined
that there was "inadequate" evidence  to conclude that PCE is carcinogenic to
humans (IARC, 1982).  This assessment  was  based primarily on the epidemiologic
            <*
study  of Blair et aj_.  (1979).   (An  analysis  of animal data gave "limited
evidence" of  carcinogenicity; evi'dence for any activity of PCE in  short-term
tests  was  "inadequate" to  judge  its carcinogenic potential)  (IARC, 1982).   The
U.S. EPA Health  Assessment Document for Tetrachloroethylene  (Perchloroethylene)
(EPA,  1985a)  has also  analyzed  the  evidence of carcinogenicity of  PCE.  this
evaluation  included an  extensive review of short-term test results, data from
animal tests, and  several  epidemiological  studies.   EPA concluded  that the
evidence  for  the carcinogenicity of PCE  in animals  is  "limited," and  that the
epidemiological  data were  inconclusive.  PCE was placed in Group C, a possible
human  carcinogen (EPA,  1985a).
                                        75

-------
     It should be noted however,  that since these two analyses were published,
other epidemiologic and animal  study results have become available.  It is
possible that a subsequent evaluation by either the U.S. EPA or IARC would
result in a different conclusion.
                                        76

-------
          6.  HUMAN EXPOSURES TO PCE FROM CONTAMINATED WATER SUPPLIES

     In this section,  we describe  the procedures we use to assess  human
exposures attributable to contaminated  water supplies.   Efforts to assess
human exposure to contaminated drinking water have revealed that significant
exposures to volatile  organic compounds (VOC's)  can occur from pathways other
than the direct ingestion of water.   Several researchers  have investigated the
relative importance of a variety of VOC exposure routes in the home from use
of contaminated water  supplies (Cothern et a_L,  1986;  Shehata,  1985;  Andelman,
1985).  In addition, there have been studies of the contribution to indoor
exposures of waterborne radon-222,  another highly volatile substance  (Hess
et a].., 1982; Prichard and Gesell,  198T). .These studies  indicate  that exposure
to volatile chemicals  from routes  other than direct ingestion of fluids may be
as large as or larger  than exposure from ingestion alone.  These other routes
include inhalation from indoor air of contaminants mobilized by showers, baths,
toilets, dishwashers,  washing machines, and cooking; ingestion of contaminants
in food; and dermal absorption of contaminants while washing, bathing, and
showering.  We divide  exposures attributable to contaminated ground water into
ingestion, inhalation, and dermal  absorption pathways.   Our discussion of human
exposure to PCE is divided into three subsections.  In the first subsection,
we provide a background on the general  approach we use for assessing  human
exposure to VOC's.  This approach addresses relative contributions from
ingestion, inhalation, and dermal  absorption.  Our focus  in this subsection is
on the type of information that is needed for risk assessments.  The  second
subsection covers the method used to estimate ingestion,  inhalation,  and
dermal-absorption dose factors.  These factors convert .water concentrations in
mg/L into human-population exposures in mg/kg-d.  The third subsection presents
our calculations of the magnitude and distribution of human exposures
attributable to PCE contamination in California groundwater supplies.
BACKGROUND ON HUMAN-EXPOSURE ESTIMATES  .

     The purpose of J:he human-exposure estimate is to provide a distribution
of population dose of a chemical from the various phases of the environment.
Three primary pathways must be addressed — inhalation, ingestion, and dermal

                                      77            ".

-------
absorption.  The exposure estimates form the basis for determining the absorbed
doses, which are expressed as the amount of chemical (in mg) absorbed or
metabolized per unit body weight (in kg).
*                  -                                           *      •*
Exposure. Dose, and Risk                       '

     Exposure refers to human contact with a chemical or physical agent.
Exposure can be expressed in terms of a concentration, such as the airborne
level of PCE in mg/m3, or in terms of the quantity that comes in contact with
the human system through lung, gut wall, or skin, expressed in mg/d or mg/kg-d.
An individual breathing 20 m3/d of air containing 1 mg/m3 is exposed to
20 mg/d.  Dose, or dose rate, expresses the amount of chemical actually
absorbed into the body where it can be subsequently metabolized and/or
transported to other tissues.  Our risk estimates are based on the equivalent
lifetime dose rate expressed in mg/kg-d absorbed througlf the lung, skin, or
gut.
     For a risk-based approach, in which there is uncertainty about the level
of exposure and the dose-response function, it is appropriate to employ a
stochastic approach for estimating the incidence of health effects within a
population.  As recommended by the United Nations study on population
exposures to ionizing radiation (UNSCEAR, 1977; UNSCEAR,  1982), a general
approach for estimating the incidence of health effects within an exposed
population is obtained from an expression of the form:
    J* dD / dR(D)n(D) p[R(D)]
    o    o
(6-1)
where
     H       = expected number of health effects within the population;
     D       * absorbed dose, mg/kg-d;
     R(D)    - risk factor that expresses the lifetime probability of health
               effects at a dose level  in the range D to D + dD,  kg-d/mg;
     n(D)    = number of people receiving a dose level in the range D to
               D + dD; and
     pCR(D)] m probability density function that expresses, the probability
               that the dose/response function at dose level D has a value
               between R(D) and R(D)  +  dR(D).
                                      78

-------
If exposure levels are low,  such that the risk factor R(D)  can be approximated
by a linear function that is independent of dose rate D,  then R(D) ~ q1,
where q^ is the cancer potency in kg-d/mg.   (The derivation of q] is taken
up in Section 7.)  Nith a linear dose-response function,  Eq.  6-1  becomes
H - q S dD n(D)
      o
(6-2)
Our goal in this section is to estimate the distribution of n(D)  for water-
borne PCE exposures within the population of California.  For the three .
pathways we consider—ingestion, inhalation, and dermal  absorption—the total
dose rate is given by the expression
                                                                          (6-3)
where
     D    - total lifetime average absorbed dose rate, mg/kg-d;
     i    m index referring to pathway (l=ingestion, 2=inhalation,  3-dermal
            absorption); and
     a,   - absorption factor for the ith pathway, unitless; and
     E.   - daily average lifetime exposure by the ith pathway,  mg/kg-d.

Anatomical and Dietary Parameters for Humans
     Our human-exposure estimates are structured to provide input to a risk
assessment.  To achieve this we must determine the lifetime average exposure
within the human population in terms of the daily intake per unit body weight,
mg/kg-d.  In this subsection we review the information required for these
estimates and provide tables of representative values for infants, children,
and adult males and females.
     Table 6-1 lists values of human-body mass and surface area as a function
of age.  The lower portion of this table lists the values used in this study
to characterize each of the major age categories — adult, child, and infant.
The values listed include arithmetic means and standard deviations.  The
surface area as a function of body weight is calculated from a formula taken
from ICRP (1975):
                                      79

-------
 Table  6-1.  Human body weight and surface area  by  age  and  sex  (from  ICRP,
 I .7 / D / •                           '-                           " .
Age
(y)
Newborn
1
2
4
8
12
16

20

40

Infant
Newborn to 2 y
Child
2 to 16 y
Adult
16 to 70 y
Sex
male/female
male/female
male/female
male/female
male/female
male/female
male
female
male
female
male
f emal e
male/female
male/female
male
female
Arithmetic mean + one standard
SA . 4W + 7
iM W + 90
in which
SA - surface
W - body we


2
area, m , and
ight, kg.
Mass (kg)*
3.5 ±
10 ±
12 ±
18 ±
26 ±
41 ±
62 +
55 ±
70 +
58 ±
75 ±
62 ±
8.5 ±
32 ±
73 +
60 ±
deviation




0.6
2
2
2
5
8
8
8
10
9
10
10
3'
16
10
9
•




Surface areaa
(m2)
0.22 ±
0.47 ±
0.54 ±
0.73 ±
0.96 ±
1.3 +
1.7 ±
1.6 +
1.8 ±
1.6 ±
1.8 +
1.7 +
' 0.42 + 0
1 .1 ± 0
1.8" + 0
1.6+0





0.02
0.07
0.06
0.06
0.1 .
0.2
0.1
0,1
0.1
0.1
0.1
0.2
.1
.4
.1
.1

(6-4)



The standard deviation in surface area is calculated as the product of the

derivative of surface area with respect to body weight (W) and the standard
deviation in body weight:
                                      80

-------
 7SA
1  ^
90
 4H + 7
(H + 90)'
'w
                                                                          (6-5)
     In Table 6-2 we present values of the hourly breathing rate by age and
activity level.  Also provided are the daily average breathing rates based on
the time spent at' rest or awake and the daily average breathing rate per unit
body weight.  These values represent the volume of air that enters and leaves
the lungs within a one-hour period.  Based on the information in this table,
we use a single breathing-rate-per-unit-body-weight for male and female adults
of 14 L/kg-h for the daily average air intake.. However, to estimate time-
varying exposures we assume that the infant, child, and adult, respectively,
breathe 29, 24, and 17 L/kg-h during waking hours and 11, 9, and 6.1 L/kg-h
while resting.
     Table 6-3 lists reference values for intake of fluids by infants,
children, and male and female adults.  We have listed the information by fluid
source.  Tap water refers to direct consumption of tap water.  Other sources
of fluids refer to all intakes of fluids, exclusive of milk and direct tap-
water consumption.  It should be noted that intake of fluid in beverages such
as coffee, tea, and soft drinks may consist of indirect tap-water consumption.
Also listed in Table 6-3 is the fluid intake per unit body weight.  For infants
and children, we find this ratio to be relatively high, 0.11 and 10.044 L/kg-d,
respectively.  For adults, the ratio of intake varies from a typical value of
about 0.026 L/kg-d to values as high as 0.05 L/kg-d at high environmental
temperatures or for moderately active adults (ICRP, 1975).  Appendix B provides
reference values for water consumption in an arid environment.  The values of
water consumption discussed in this section are intended for use in assessing
chronic exposures.  Values discussed in Appendix B apply to maximum acute
exposure.
     Table 6-4 is a summary of the range of values that can be used to estimate
age-dependent intake of various food types.  For each category in which a  range
of values is listed, the lower value corresponds to a median or "typical"  value
associated with the reference adult or child.  The upper value corresponds to
a high annual average rate of intake.  The high annual  average is the value
recommended by the U.S.  Nuclear Regulatory Commission (1977) to determine
exposure of the maximally exposed individual from routine radioactive releases.
                                      81

-------
  ICRP?  ?975).ReferenCe  breath1n9  rates  f°r  infants,  children, and adults  (from
 Activity
  Infant        Child        Adult         Adult
,  /i1 #,,       (1° y>       female        male
L/h  (h/d)    L/h  (h/d)    L/h  (h/d)   L/h  (h/d)
 Working, light activity,
 or recreation
 Resting
 Daily average

 Daily average breathing
 rate per unit body weight
250   (10)    780   (16)    1100  (16)   1200  (1
160
 19
                                                     L/h
620           850
     L/kq - h
 19
                             14
                                   6)
93    (14)    288   (8)     360   (8)    450
                                                                             (8)
                                         950
                                                                        13
 Table  6-3.   Fluid  intakes for  infants, children, and adults (from ICRP, 1975).
Milk
Tap water
Othera
Total fluids
Fluid intake per unit body weight

  a Includes tea,  coffee,  soft drinks,  beer,  and other beverages.
Infant
(1 v)
Child
(10 v)
Adult
female
Adult
male
L/d
0.9
0.9

t .0.11
0.45
0.2
0.75
1.4

0.044
0.2
0. 1
1.1
1.4
L/kq-d
0.023
.0.3
0:15
1.5
2.0

0.027 .
                                     82

-------
Table 6-4. Food intake for infants,  children,  and adults.

Food type

Fruits, vegetables, and grains
Milk and dairy products
Milk fat
Meat and poultry
Freshwater fish
Saltwater fish and other seafood
Infant
(1 V)


-.-
0.90
0.036
—
-r
-'-
Child
(10 V)

kg/d
0.60 to 1.8
0.51 to 1.1
0.021 to 0.044
0.13 to 0.18
0.0077 to 0.044
0.0031 to 0.054
Adult
(male and female)

—
0.52 to 1.6
0.30 to 0.85
0.012 to 0.034
0.21 to 0.30
0.014 to 0.057
0.0074 to 0.071
   a The  range of values expressed here.reflects the range of values
 oublished  in ICRP-23 (ICRP, 1975) and in Reg Guide 1.109 (U.S. NRC, 1977).
 The lower  value expresses the "typical" value; the upper value represents that
 suggested  to calculate maximum annual average exposures.
 WATER-BASED PATHWAYS AND DOSE  FACTORS

      In this section,  we develop  the unit-pathway-dose factors for contaminated
 tap-water supplies.  We consider  three  pathways—water ingestion, inhalation,
 and dermal  absorption.  For a  given pathway,' i  (i.e., 1 =  ingestion,
 2 - inhalation,  and  3  - skin absorption),  the unit  pathway dose  factor, F.,
 translates the water-supply concentration,  Cw,  in mg/L into  an equivalent
 lifetime absorbed dose rate, D.,  in mg/kg-d,
 Ji
Ficw
                                                                           (6-6)
 Because we are interested in the equivalent lifetime dose within  a population
 composed of three age categories (infant, child,  and adult),  we calculate the
 overall dose factor, F., as the weighted sum of the pathway-dose  factors,
 fi (age group), for each of the three age categories.
                                     54
 F  » y  -f,(Infant) +    f^child) +    f,(adult)
                                                                     (6-7)
                                       83

-------
 In  this  expression,  the. factors  2/70,  14/70,  and  54/70 reflect  the  fraction  of
 time  the population  cohort  spent in  each  of  the age  categories.   He also
 assume that  the  population  is*stationary.

 Water Ingestion
      For  the water-ingestion pathway  (i =  1),  the  unit-pathway-dose  factor,  f.,
 for each  age category  is obtained by  dividing  daily water  intake  by  body weight
 and multiplying by an  absorption factor, a1.   The  ratio of fluid  intake to
.body  weight for each age group comes  from  Table 6-3.  We also use data compiled
 by the ICRP (1975) on  fluid intake by adults at high environmental temperatures
 and during moderate activity.  The ICRP reports that at high environmental
 temperatures (to 32°C) adults consume 2.8  to 3.4 L/d of fluids and that
 moderately active adults can consume  3.7 L/d.  Using an average adult weight
 of 66.5 kg, from Table 6-1, we calculate that  this corresponds to a  fluid
 intake as high as 0.056 L/kg-d.  We make the conservative  assumption that all
 fluids consumed by the members of a household  with contaminated water are at
 the same  concentration level.  The absorption  of PCE across the gut  wall is
 assumed to be 100% (see Section 3).   We further assume that a reasonably
 conservative estimate of the pathway  dose  factor can be bracketed using the
 lifetime  average fluid intake per unit body weight with 2-L/d adult  fluid
 intake as the reference value and the fluid intake by moderately active adults
 as the upper bound.  Thus; F, is bracketed by
     y§ x 0.11 + 75 x 0.044 + |$ x 0.025

     0.031 mg/kg-d per mg/L, and
                                        ff

     7§ x 0.11 + j^ X 0.044 + |4 x 0.056
                                        *>
     0.055 mg/kg-d per mg/L.
(6-8)
(6-9)
     The reference value, 0.031 (mg/kg-d)/(mg/L) is similar to the value
obtained under the assumption that intake of drinking water over a lifetime-
approximates 2 L/70 kg or 0.028 (mg/kg-d)/(mg/L).  Cothern et al_. (1984) report
that a weighted average derived from water-consumption curves gives a lifetime
                                      84

-------
dose factor on the order of 0.034 (mg/kg-d)/(mg/L).  Our upper-bound limit
corresponds to a lifetime average daily fluid intake of 3.8 L/70 kg.  To date,
population studies on the variability of water consumption have not been
conducted in the United States (Cothern et a].., 1984).  However, the Canadian
Environmental. Health Directorate (1981) has studied the variability of water
consumption in Canada and found 13-16% of adults may consume more than 2 L/d
of fluid.  It is not clear how these results apply to drinking-water
consumption in California, but the numbers do suggest it is usefuj to consider
the upper-bound dose factor in ex'posure estimates.

Inhalation Exposure

     Several researchers have addressed the relative contribution of the
respiratory pathway to overall human exposures from VOC's in tap water.  All
have found that volatile compounds in water supplies can result in inhalation
exposures that are comparable to the ingestion pathway.  Prichard and Gessell
(1981) found that public water supplies provide a major pathway for indoor
exposures to radon and measured the amount of radon transferred to air from
water during various household activities.  Cothern et aj.. (1984) found tlrat
the respiratory uptake of VOC's from household air attributable to tap wafer
is approximately equal to oral uptake from fluids. • An.de 1 man (1985) developed
a model shower that he used to study human exposures while showering or
bathing.  He found that for a volatile pollutant, overall indoor inhalation
exposure may be as much as six times higher than direct ingestion exposures.
Foster and Chrostowski (1986) have developed an integrated household exposure
model for assessing human uptake of VOC's from tap water.  Using this model,
they have estimated that the ratio of hjjman uptake by inhalation to uptake by
ingestion is 1.5 for TCE.
     McKone (1987) has developed a model that describes the daily
concentration profile of VOC's within the various components of the indoor air
volume of a dwelling.  For the mode.l, the indoor air volume is divided into
three compartments—the shower/bath stall, the bathroom, and the household
volume.  We use this model to calculate two bounding pathway dose factors that
correspond to the use of contaminated water in the indoor environment.  One
corresponds to the average daily lifetime dose absorbed by an individual

                                      85

-------
 living  in a  "typical" California  household.  The other  represents an  upper
 bound estimate of exposure and dose  in which the model  parameters are set at
 values  that  provide the upper limit  on exposure and dose estimates.   For the
 typical home, we assume that the  household has four occupants and uses 900 L/d
 of water contaminated with 1 mg/L of PCE and that PCE has water-to-air transfer
 properties similar to radon-222.  The water-to-air transfer .efficiency for PCE
 is calculated as the product of the efficiency measured for radon and the ratio
 of the mass-transfer coefficient for radon to the mass-transfer coefficient for
 PCE.  The derivation of this estimation technique is found in McKone (1987)".
     The time-dependent concentration profile of PCE in shower stall,  bathroom;
and household air and  the resulting effective lifetime doses were estimated
using two sets of assumptions.   These assumptions were intended to define
typical  values and a likely upper limit on dose.   The two sets of assumptions,
are listed  below.
     Assumptions  used  for typical  doses:
         Occupants  spend  100% of their time  in  the  house from 11:00 pm  to
         7:00 am.
     •    Bathroom is  used for showers/baths  from 7:00 am to  8:00  am.
         Each adult and  child spends 20  min  in  the  bathroom  during  the period
         from 7:00 am to  9:00 am.                           ,
         Each adult and child spends an•  additional'20 min in  the  bathroom
         during any 22-h  period  (excluding the hours  7:00 am  to 9:00  am).
     •    Each adult spends 10 min  in the  shower or  bath.
         Adults spend 25% of the  time  from 7:00 am  to 11:00 pm in the house.
         Children spend an average of 20 min/wk in  showers or baths.
         Children spend 60% of the time between 7:00 am and 11:00 pm in the
         house.
         Infants spend 100% of their time in the house and 2% of that time in
         a bathroom.
         Fifty percent of the PCE inhaled is available for pulmonary uptake.

    Assumptions used for upper-bound doses:
    •    Each adult and child spends 40 min in the bathroom between 7:00 am
         and 9:00 am.
         The bathroom  is  used for showers/baths  from 7:00 am to 8:30 am.

                                     86

-------
     •    Each adult spends  20  min  in  the  shower or  bath.    .      ,
     •    All  age groups  spend  100% of their  time  in the  house.
     •    Children spend  an  average of 40  min/wk in  showers  or  baths.
     •    Absorption of PCE  in  the  lung for all age  categories  is  100% of  that
          inhaled.

     Using this model,  we estimate  that the pathway  dose  factor for  inhalation
is in the range
F2 « 0.041 mg/kg-d per mg/L for typical  households,  and

F0 - 0.16 mg/kg-d per mg/L for an upper-bound estimate
(6-10)
(6-11)
(but note the modification of Eq. 6-10 required for the application appearing
at the end of Section 7).  These numbers are based on the assumption that an
adult showers every day and that children bathe every second day.   Table 6-4
summarizes the relative contribution to the pathway dose factor, F2> from
each age category and household compartment.  For adults or children who take
baths instead of showers these numbers are likely to be reduced somewhat.  He
have not examined the extent of reduction that taking baths in place of
showers would give.  Table 6-4 reveals that exposures to adults in the shower
and bathroom are the major contributors to indoor inhalation exposures
attributable to contaminated water.  A local sensitivity analysis of the
different parameters involved in the calculation of the pathway dose factor
for the inhalation route of exposure has been prepared in McKone (1987).
According to that analysis, the pathway dose factor was most sensitive to 1%
increases in the following parameters:  the absorption fraction in the lung,
the transfer efficiency of a VOC in shower water to air, the water use of
individual's in showers, and the ratio of breathing rate to body weight (adult).
     Although the assumption that adults spend 25% of the time  from 7:00 am  to
11:00 pm in the house may seem low, we believe this is a plausible value for a
typical adult who spends 10 h per day in work and travel, leaving  6 h of
leisure time of which we assume roughly 2/3 is actually spent  in the home.
Furthermore, we assume that roughly 3/4 of all adults work outside the home
                                       87

-------
and that those who do not can be accounted for by the upper bound estimate.
Finally, we have found that the pathway .dose factor is very insensitive to
this occupancy factor.
     When calculating "typical doses," we assume that only 50% of the PCE
inhaled is available for pulmonary uptake in order to account for the
observation that.the alveolar volume of the lung — which can be considered
the actual interface between the human system and inhaled air — is
approximately one-half o*f the resting total lung capacity (ICRP, 1975).  This
value of 50% is actually somewhat conservative in that it does not take into
account the effect of dead space and lung expansion during respiration.  For
example, assuming an average tidal  volume of approximately 1.0 L, dead space
of 0.16 L, alveolar volume of 3.0 L, and resting total lung capacity of 5.6 L
(ICRP, 1975),  the actual  percent of inhaled air ventilated through alveolar
space (assuming instantaneous.equilibrium between alveolar and non-alveolar
air, excluding dead -space) would be approximately 35% to 40% (or from 4.9 to
5.6 L/min assuming a 20 m /d respiration rate).
  •   Table 6-5 compares the ratio of inhalation intake to ingestion intake
(based on 2 L/d per 70 kg) projected using the McKone model  and compares this
to the value of this ratio that has been estimated by other researchers.

Dermal Absorption            *

     We reviewed the literature on absorption rates of volatile solvents having
direct contact with the skin to estimate the likely value of dermal  absorption
from normal daily use of contaminated water.  Over the last 20 y, several
investigators  have examined the transport of dissolved chemicals across the
skin (Stewart  and Dodd, 1964; Riihimaki  and Pfaffli, 1978; Brown et a].., 1984;
Cothern et al... 1984; Scheupleih and Blank, 1971; Bronaugh,  1985; and Foster
and Chrostowski, 1986).  Although a complex process, dermal  uptake of
compounds occurs mainly through passive diffusion through the stratum corneum.
     We assume that dermal absorption occurs during bathing and showering.  To
determine the  pathway dose factor for dermal absorption, we had to make a
number of simplifying assumptions.   These are

          Resistance to diffusive flux through layers other than the stratum
          corneum is negligible.

-------
Table 6-4.  Percent contribution to the lifetime average dose factor

(inhalation) from specific age and household exposures.
Exposure

% Contribution
Typical household pathway dose factor
Adult, in
fn
in
Child, in
in
in
Infant, in
in

Upper-bound
Adult, in
in
in
Child, in
in
in
Infant, in
in

shower
bathroom
remainder of house
shower
bathroom
remainder of house
bathroom
remainder of house

pathway dose 'factor
shower
bathroom
remainder of house
shower
bathroom
remainder of house
bathroom
remainder of house

51
20
7.9
5.2
7.2 .
• 7.0
0.17
1.2
100

' 42
25
14
4.2
9.0
4.9
0.16
0.56
100
Table 6-5.  Estimates of the ratip of indoor inhalation dose to ingestion dose
for a unit concentration of 1  mg/L.


Inhalation Uptake (mg/kg-d)
Ingestion Uptake* (mg/kg-d)
1.4 to 5.0
1
6
1.5

Compound
PCE
VOC
"Volatile pollutant"
TCE

Reference
LLNL Model (McKone, 1987)
Cothern et aK (1984)
Andelman (1985)
Foster and Chrostowski
(1986)
    Assuming 2 L/d for a 70-kg individual.

                                      89

-------
      •    Steady-state diffusive flux is proportional to the concentration
           difference between the skin surface and internal  body water.
      •    An adult spends from 10 to 20 min in a bath or shower each day.
           During,bathing, roughly 80% of the skin is in contact with water,
           and during showers,  roughly 407/of the skin is in contact with water.
           Children and infants  spend approximately 1  h/wk in bathing or
           swimming (U.S.  NRC,  1977).

      The  absorbed  dose from  dermal  absorption  is  given  by the expression
          fs SA
                                                                          (6-12)
where
     T

     fs
     SA
           = steady-state flux  across  the stratum corneum,  mg/cm?-h;
           = duration  in  the  shower or bath,  h;
           - fraction  of  the  s'kin  surface in  contact  with water,  unitless;  and
           • surface area of  the skin,  cm2.
Chemical transport across the skin is assumed to follow Pick's law, so that
the flux J$ across skin tissue is given by
                                                                         (6-13)
where
     AC.
               permeability constant across the stratum corneum, L/cm2-h; and
               concentration difference of the solute across the tissue, mg/L.
Brown et a].. (1984) have determined, from an analysis of chemical transfer
through the skin layer, that Kp is on the order of 0.001 L/cm2-h for VOC's.
They used this value to characterize Kp for six measured. skin-absorption rates
on four different chemicals.  In these experiments, K  ranged from 0.0006 to
0.001 L/cm -h.   For dilute solutions, AC$ is approximate.^ equal to the
chemical concentration at the skin surface.   However, the concentration at the
skin surface is not necessarily the same as  the concentration in the water
supply.   For showers,  we assume that C
                                         Cw, the PCE concentration in tap
                                     90

-------
water.  But for bathing,  in which water stands for a period of time,  we use
C  » C   where C  is the  average water concentration over the period  of the
bath.  Assuming an exponential  loss of one-half of the dissolved PCE  over a
period of 10 min, we find Cw = 0.72 enduring a 10-min bath and
"w
    0.54 C  during a 20-min bath (Foster and Chrostowski, 1986).
          W
     We substitute Eqs. 6-12 and 6-13 into Eq.  6-7 to obtain a lifetime
equivalent dose factor for dermal  absorption, .
70 (^ x'fs x m x Vinfant
                                         (
-------
Higher estimate (assuming 20-min bath for adults):
F-. = 0.001 —5—    fi <°-17   x °-80 x 490 '- * 0.72 CJ
 J         cm -h   70       d              Kg          w
        (0.17   x 0.80 x 340   - x 0.72 GW> +
.g (0.33   x  0.80  x  260   - x  0.54

            per
   = 0.037

(6-16)
Summary of Pathway Dose Factors
     Table 6-6 provides a summary of the three pathway dose factors.   We list
our best estimate and the "high-average" or upper bound values.   The  pathway
dose factors are used to convert the water-supply concentration  of PCE,  Cw,
(mg/L) Into an equivalent lifetime average dose rate in mg/kg-d.   Thus,  the
pathway dose factor serves to account for the exposure conditions within a
population and the absorbed dose corresponding to an exposure for each pathway.
     Table 6-6'also lists the lifetime average fluid intake of contaminated
water that would give the equivalent dose to a 70-kg adult.  These values are
listed for comparison.  Finally, Table 6-6 lists the percent of  the lifetime
equivalent dose attributable to contaminated water that is contributed by each
pathway.     '

HUMAN EXPOSURES TO PCE-CONTAMINATED GROUND WATER
     In this .subsection, we estimate the integrated population dose
attributable to PCE contamination in California groundwater supplies.   To
determine the population doses, we focus on the distribution of contamination
in household water supplies.  The pathway dose factors derived above are used
to translate household water-supply concentrations into equivalent lifetime
doses in mg/kg-d.  We use AB1803 survey data (CDHS, 1986) to estimate the
distribution of household water-supply concentrations of PCE.  Following this,
we use the distribution of household-supply concentrations and dose factors to
calculate the distribution of doses within the California population.
                                      92      *

-------
Table 6-6.   Summary of the pathway dose factors  for PCE.
Pathway
  Fluid      Indoor      Dermal
ingestion  inhalation  absorption
                                                                      Total
Variable
                                  FT
Best estimates:
(mg/kg-d)/(mg/L)
                                  0.031
Equivalent lifetime daily fluid
intake by 70-kg adult, L          2.2
Percent of total                 31
Upper bounds;
(mg/kg-d)/(mg/L)                  0.055
Equivalent lifetime daily fluid
             0.040

             2.8
            41

             0.15
 0.028

 2.0
28

 0.037
  0.099

  6.9

100

  0.24
intake by 70-kg adult, L
Percent of total
3.8
23
10.5
62
2.6 17 .
15 ' 100
     At present, the only data available on the distribution of PCE in
household supplies are the AB1803 data (CDHS, 1986).   It should be recognized
that this data base has a number of limitations with  regard to estimating PCE
concentrations in water supplies.  Among the major limitations are   • '

     •    Incomplete sampling data (even though all large systems were
          sampled, only a limited number of wells in  each system were sampled)
     •    Limited data on how well-water concentrations change with time.
     •    No information on the percent of system water supplied-by an
          individual well.
     •    No information on the relation between concentration in individual
          wells and the concentration in water-supply lines that enter
          individual households.

In spite of these limitations', we estimate the distribution of human doses in
California using the AB1803 data and two sets of assumptions.  The first set
of assumptions is
                                      93

-------
           Surface  water  and  negative  samples  are  assumed  to contain  no PCE.

r     *     A11  We11s  within a given  water  system are  assumed to  flow  at the
           same rate.

           The  average concentration of PCE  in unsampled wells is  the  same as
           the  average concentration of PCE  in sampled wells.

Under this  set of  assumptions the daily dose  rate to people in a  given water
system  is given by the expression
  i =Ci fwi
where
                                                       (6-17)
      'wi
       » F2' F3
daily average lifetime dose rate to individuals obtaining
water from water system i, mg/kg-d;
average concentration of PCE in all wells of system i  that
were tested for PCE and are still in use, mg/L;
fraction of the water in system i that is taken from
groundwater supplies (obtained from Appendix C), unitless;
and
pathway dose factors derived above.
The second set of assumptions is the same as the first, with the exception that
          Unsampled wells are assumed to contain no PCE.
Under the second set of assumptions the daily dose rate to people in a given
water system i takes the form
         * -


                   +F2.+F3)
                                                       (6-18)
where the parameters are the same as above and the additional  parameters N.
and N. are the number of wells tested and the total  number of wells in
system i, respectively.
                                      9.4

-------

-------
                         B.4

EXPOSURE ASSESSMENT
               Attribute O   Central estimates and upper and lower bounds on exposures
                              or, ifpossibk, the full population distribution of exposures are
                              described; any preferred estimates are noted, together with
                              supporting documentation.
                   SOURCE  Case Study K. Tetrachioroethylene (Pages 95-96).
                       Note  Support for the upper bound and "best" estimates can be inferred from
                             the step-by-step derivation shown in Exposure Assessment Attribute 5.

-------
                                 UCRL-15831
   Health Risk Assessment
of Tetrachloroethylene (PCE)
in California Drinking Water
 K. T. Bogen, L. C. Hall, T. E. McKone,
     D. W. Lay ton, and S. E. Fatten
   Environmental Sciences Division
Lawrence Livermore National Laboratory
       University of California
            P.O. Box 5507
        Livermore, CA 94550
           April 10,1987
                              *•

            Prepared for
  California Public Health Foundation
            P.O. Box 520
         Berkeley, CA 94701

-------
      Table 6-7 shows the distribution of PCE exposures among the California
 population as calculated from Eqs.  6-17 and 6-18.   The table was constructed
 using AB1803 data to obtain the average concentration in  tested wells  and the
 data compiled in Appendix C on the  number of people associated  with  each  water
 system.
      We  also used Eqs.  6-17 and 6-18  to calculate  integrated population dose
 rates attributable to sampled water systems.   The  integrated population dose
 rate (IPO) in person-mg/kg-d is calculated by summing the  product  of the  daily
 dose rate  and the population over all  systems having  positive PCE
 concentrations
 IPD
                                                                         (6-19)
 where
     IPD =   integrated population dose rate to PCE from large water systems
             in California, person-mg/kg-d;
     Pi =•    population associated with water system i;
     D. =    lifetime daily dose rate to PCE in system i, calculated using
             Eq. 6-17 or 6-18, mg/kg-d; and
     n  -    number of large systems with positive PCE measurements.

     Table 6-8 provides a summary of the integrated population dose as it is
calculated, based on the two sets of dose factors  and two sets of assumptions
regarding the interpretation of sample data.   It should be  noted that these
population doses do not include contributions from surface  water and private
wells.
                                     95

-------
Table 6-7.  Distribution of population doses to PCE from ground*water in
California.
System average dose, mg/kg-d
Population receiving the corresponding dosea
Using best-estimate       Using upper-bound
   dose factors             dose factors
Assuming only positive test wells
are contaminated
10"7 < D < 3 x 10~6
3 x 10~6 < D < 10~5
10~5 i D < 3 x 10~5
3 x 10~5 1 D < 10"4
•1(T4 < D < 3 x 10~4
3 x l(f4 < D < 10~3
10~3 < D < 10"1


10"7 £ D < 3 X 10~6
3 x 1(T6 < D < 10~5
10~5 < D < 10~4
10~4 < D < 3 x 10"4
3 x TO"4 < D < 10~3
10" 3 < D < 10"1
1,000,000
4,400,000
870,000
470,000
60,000
800
, —
Extrapolating average
tested wells
800,000
760,000
5,039,000
. 140,000
77,000
800
710,000
510,000
4, 300 ,-000
950,000
240,000
59,000
800
concentrations in
to' all wells
•
960,000
1,920,000
3,800,000
71 ,000
29,800
   a The sum of each  column  is  the  number  of  people who obtain water  from a
 system having  one  or more contaminated  wells.

 Table 6-8  Summary of integrated population  dose  rates in  person-mg/kg-d
 calculated from Eq.  6-19.

                Integrated population dose (IPD)  in person-mg/kg-d
                                               Calculated
                                             using Eq.  6-17
                                                 for 0
                                   Calculated
                                 using Eq.  6-18
                                     for 0
Based on best-estimate dose factors
Based on upper-bound dose factors
250
610
92
220
                                       96

-------
                          B.4

EXPOSURE ASSESSMENT
               Attribute 7   Uncertainties in the estimates are described, and the relative
                               importance of key assumptions and data is highlighted.
                   SOURCE  Case Study D. Formaldehyde (Pages 6-1 to 6-47).


                        Note  None of the case studies presented a cumulative highlighting of key
                              assumptions or a particularly quantitative treatment of uncertainty.
                              Excerpts from the formaldehyde report illustrate a qualitative descrip-
                              tion of factors that would result in variations from the upper bound
                              estimates.

-------
         Assessment of Health Risks

to Garment Workers and Certain Home Residents

        from Exposure to Formaldehyde
                  April 1987
  Office of Pesticides and Toxic  Substances
     U.S.  Environmental Protection Agency

-------
                      6-   EXPOSURE ASSESSMENT

 6.1.   Introduction


      The sources of HCHO can. be grouped  into two major


 categories:  commercial production and indirect production.  'The


 chemical is not 'imported in. any appreciable quantities.


      Commercially,  HCHO is  produced from the catalytic oxidation

 of methanol, using  either silver oxide or a nixed-metal oxide as


 the catalyst.   Processes accounting for the indirect production

 of HCHO include the photochemical  oxidation of  airborne

 hydrocarbons released  from  incomolete  combustion processes,  the


 oroduction  of  HCHO  during  incomplete combustion of  hydrocarbons


 in fossil fuels  and refuse, and  certain natural processes.


     .The  1984  commercial production of HCHO amounted to about 6

 billion pounds.  The major derivatives are  urea-HCHO resins,

 phenol-HCHO resins,  acetal  resins,  and butanediol.   The urea- a-.d

 phenol-HCHO resins  account  for about 53 percent of  HCHO


 production.  Adhesives  and plastics.are the major end  uses.

     The "consumption"  of HCHO can  be  broken down into  three

 major categories:   nonconsumptive  uses,  pseudo-consumptive  uses,

 and consumptive uses.   In nonconsumptive  uses,  the  chemical


 identity of  the HCHO does not change. , In pseudo-consumptive
                                            s&
 uses, the chemical  identity of HCHO does  change, .but  it  is  not


 irreversibly altered.   Under aporopriate  conditions, some or  all


of the original HCHO may be regenerated.  Consumptive uses, on

 the other hand, are those uses in which HCHO serves  as  a -


 feedstock for the preoaration of other  chemicals.   The
                               6-1

-------
derivatives are irreversibly formed and usually contain only
residual levels of unreacted HCHO.  Under extreme conditions,
such as very high temperatures or highly acidic conditions, some
of the derivatives may degrade and release HCHO.
     HCHO's major nonconsumptive uses are (1) disinfectant, (2)
preservative, (3) deodorant, and (4) textile and pacer uses.
     'The major pseudo-consumptive uses are (1) urea-HCHO resins
which are used in flberboard, particleboard, plywood, laminates,
urea-HCHO foams, molding compounds, and paper, textiles, and
protective coatings; (2) urea-HCHO concentrates which are used to
produce time-release fertilizers, and  (3) hexamethylenetetramine
which is used as a snecial anhydrous form of HCHO to cure resins
and to treat textiles/and  rubber.
     The major consumptive uses are  (1) melamine-HCHO resins
which are used for molding, compounds,  fiberboard, particle'ooard,
plywood, laminates, paper'.and textiles,  (2) phenol-HCHO  resins
which are used in fiberboard, particleboard, plywood moldina
compounds, and insulation;  (3) nentae.rythri tol which is  used  to
produce alky.d resins,  (4)  1,4-butanediol  which  is used  to  produce
tetrahydrofuran,  (5) acetal  resins  which  are used in the
manufacture of engineering plastics, and  (6)  trimethyloloropane
which is used in  the production  of  urethanes.
6.2.   Estimates  of  Current Human Exposure
     To obtain estimates  of  human exposure  to  HCHO,  the Agencv
commissioned a contractor  study  (Versar,  1982).   This  studv
integrated  the  existing monitoring  data,  engineering or
                                6-2

-------
 estimates,  use data,  population estimates,  and assessment of 'the

 likelihood  of exposure  from HCHO-related activities into an

 exposure  assessment detailing those .activities having a high HCHO

 exposure  potential.„. EPA updated  some portions of this assessment

 to  reflect  new data received in response to the FEDERAL'REGISTER

 notice of November  18,  1933 and other data  gathered by EPA.   The

 combined  data were  used  as. the basis for the May 1985 draft risk

 asses sment.        ,

     Subsequent to  the draft risk assessment,  the Agency

 commissioned  additional  contractor studies  to  assess garment
                                      f
 worker (PEI,  1985-)  and residential (Versar,  1986a,b,c)  exposure

 to HCHO in  more depth.   The exposure estimates from these reports

 were used as  the primary basis for this  risk assessment.   The

 conclusions of  these  contractor reports  are  summarized  in this.

 document; more  detailed  information regarding  exposure  can be

 obtained  by referring to the contractor  reports.

 6^3.   Populations  at Risk

     The  two  populations at risk.-examined here are  certain home

 residents and garment workers.

 6.3.1.    Home  Residents

     Based on a projection  of- manufactured housing  starts by

 Schweer" (1987),  it  is estimated that•7,800,000 persons  may occupy

 new manufactured homes during  the  next ten years.   This  figure

 assumes 295,000  starts per  year and 2.64 persons  per home.

     Similarly,  ah  estimated 214,000 new conventional homes

containing significant- quantities  of pressed wood products as

construction materials will  be started each  year  for the  next  ten
                               6-3

-------
years with an occupancy  rate  of  2.95  persons  for a total of

6,310/000 persons.

6.3.2.   Garment Workers

     The number of potentially exposed  garment workers is

estimated to be 777,000  (Versar,  1982)  out .of 1,100,000 workers

employed in the U.S.  apparel  industry (Ward,  1984)'.  -This figure

may drop in the future due to increased  foreign  competition and

the introduction of labor saving  equipment.

6.3.3.    Summary
                        *
     Table 6-1 presents population, estimates  for the  two housing

segments.  Assuming'that the number of  potentially exposed

garment workers remains  steady at 777,000, then  a total of  almost

15,000,000 persons over the next  ten  years may-have the potential

to be exposed to elevated levels  of HCHO.
   Category
Manufactured homes

Conventional homes

* Schweer  (1987)
                            Table 6-1.
                        POPULATIONS 'AT RISK
                                        Population
                                        Estimates
per yr

779,000

631,000
  10 yrs

7,790-, 000


6,310,000

-------
 6-4.   Souroaa of HCHO in Population Categories of  Concern
      The principal sources of HCHO in the two population
 categories of concern are HCHO-based resins, principally urea-
 HCHO (UF)  resins.  In homes, these resins are used  -to bond .the
                              *   ...
 wood plys  used, to make plywood and to bind the wood particle and
 fibers  used to make particleboard and medium density
 fiberbpard.  For garments,  HCHO-based resins are used to impart
 permanent  press finishes  to the. garments.
 6.4.1.   Homes Containing Pressed-Wood  Products
 6.4.1.1.    Pressed-woo.d  product  descriptions
      Pressed-wood products  are used  in  flooring, interior walls
 and  doors,, cabinetry,  and furniture. ' The three  principal types
 of products containing uT-resin  are  particleboard,  medium-density
 fiberboard  (MDF),  and  hardwood plywood.
      Particleboard is  composition  board  comprised  of 6 to 10
 percent resin  (by  weight),  and small  wood particles.  UF resin  is
 used  in the  majority of particleboard (about  90  percent  of. total
 production  capacity).  The  1983  production of particleboard  was
over  3 billion  square  feet,  of which  70  percent  was  used  in
furniture,   fixtures, cabinets, and similar products.   The
remaining 30 percent was  used  for  construction,  including deckinc
in manufactured home manufacture and  flooring underlayment  in
conventional housing.
     Recent data  indicate that particleboard  is  used in  home
construction at a rate of 0.16 square  feet  (ft2)  (~ 0.5  ^n2)
cubic foot (ft3) o.f indoor air volume  in mobile  homes.   The
                                                             ner
                               6-5

-------
loading rate (ft2/ft3) in conventional homes is lower on average;
approximately 0.05 ft2/ft3 ("ST 0.17 m2/m3) (see Table 6-2) ,
However, loading rates in conventional homes may vary
considerably from homes that contain-only particlebpard as a
cabinet material to homes whose floors are constructed with
particleboard underlayment.
     MDF is also a composition board.  It is comprised of wood
fibers and 8 to 14 percent UF resin  solids by  weiqht.
Approximately 95 percent of MDF production  (over 600 million
square feet in  1983)  was used to manufacture furniture, doors,
fixtures/ and cabinetry.   No data  are  available on  the precise
extent of MDF's use in either mobile or  conventional homes.
     Unlike the two composition boards discussed above,  hardwood
plywood is a laminated product; the  resin  is used as a qlue  to
hold thin layers of wood and veneers together. Of  the nearly  4.3
billion sguare  feet consumed in  1983,  55 percent was used  for
indoor paneling, 30 percent  for  furniture and  cabinets,  and  15
percent for doors and laminated  flooring.
                                6-6

-------
  Table 6-2 .  U» of Pressed-wood Products in How Construction
 Cate<»rv
 Mew taws
 Percent  units  containing.
   Hardwood  plywaod.paneling
   Particleboard undarlaynwnt

 Average  loading rates,6  (m^/m3)
   Hardwood  plywood paneling
   Particleboard underlayment
   Particleboard shelving*
   Particleboard kitchen  cabinets
   Total  particleboard
    Honas (Canada) d
Ptrctnt units containing
  Particltooard

Av«raot loading rat»s (m2/™3)
  Total p«rticj«ooard

Eiistinq Hemrs (U.S.)*

Ptrcont units containing
  Hardbood plywood paneling
  Parti cleooard

Avcraot loading rat« (m2/™3)
  Hardtaood plywood paneling
  Particlaboard
                                       7.6
                                      30.5
0.05*
0.118
0.010
0.039
0.167
                                      100
                                      35.5
                                      90.3
                                      0.098
                                      O.OSS
                                              Type o'f taw
                                                TH
          9.3
          9.2
                                               0.039
                                               0.092
                                                .016
                                                .052
0.
0.
                                               0.160
         100
                                      0.145    0.100
                    f

                    ff
           8.5
           1.7
0.049
0.033
0.020
0.059
0.112
          100
         most.
         most
                   1.0
                                                                  0.5
                                                                  100
                   0.079    0.479
                           most
                           most
                            1.0
                            0.5
  Data rtf1»ct only inttrior us«s of UF pr«s»d wood products.
  Loading rates ar* for tnou taws containing tfws* products.

*Sourc«:  HPA (1984) and HPHA (1984) for conventional  hows  - Based on
 interpretation of tr» results of a survey of 900 taw builders  (103
 responses) regarding the extent of use of particleooard  and hardwood
 plywood paneling in new hones containing these products  (NAH8 1984).
bSource:
 hows.
          «eyer and Hermanns (19eAa),  NAH8 (1984).  wi  (1984) >or mobile
(Footnotes continued on nest page)
                                  6-7

-------
               Table  6-2.   Footnotes (continued)
    of produce surface  anea/m3 of  indoor air volume.

<%ource:   InterArt  (1983V -  based  on  in-hcme surveys at 9 SFO, 1 TH, 1  rf
 and 1 W.   Total  loading includas underlaynent, shalving and cabinets.
  SFO loading* ranged fron 0.029 to 0.491 n^/m3.

'Sourca:   Schutte  (1981)  - Sased on in-ncme surveys at 31 SFO.  Average
 loadings based on  hows  containing those products.

f    SFO » Single  fanly  c^elling
     TH • TcMnhouse
     HP » ftjltifamily dwelling
     m * Mobile hone
                                    6-8

-------
 6.4.1.2.    HCHO release from oressed-wood products   '      -._  -
      Each  of  the pressed-wood products described above contain L'F
 resins  which  release  HCHO over'time.   The release is attributable
 to  two  basic  sources  (Podall, 1984):
     1.  - Free  (unreacted)  HCHO present as a  result of incomplete
         crosslinking  during resin  cure/
     2.   Decomposition  of  unstable  UF  resin  or  resin-wood
         chemical  species  as a result  of their  intrinsic  '
         instability and/or  due  to  hydrolysis.
 Free  HCHO,  which  is present  in cured  resin at  low levels  (<{
 percent).is the most significant  source  of HCHO release  from
 pressed-wood products  in  the initial period  after they are
 manufactured  (Podall,  1984).   The specific time period in  which
 free  HCHO dominates releases is  not known.
      The second source, decomposition  and hydrolysis,  pertains  to
 the large proportion of HCHO-bearing species like methylene
 ureas, urea methylene ethers,  and cellulose-crosslinked species
 that  may release HCHO for  a  much  longer  period  of time (Podall,
 1984).  These species differ  in  their  susceptibility  to
 hydrolytic  attack and decomposition., and  their  relative rates  and
durations of release can only  be hypothesized at  this  time.  «...
      Release of HCHO from  UF-resin  containing pressed-wood
products is complex, with  numerous  interrelated  aspects.  Th.e
pressed-wood product manufacturing  process, and  other  factors,
affect the   amount of each  HCHO-releasing  species  present in the
 finished product.  The  resin  formulation  has a  direct  effect on
                               6-9

-------
release; resins with a low HCHO:urea ratio have, when cured, a ,
lower level of free HCHO but may be less stable and more
susceptible to hydrolysis (Myers, 1984).  Other additives  to the
resin, such as acid catalysts, chanqe the resin-chemistry  and
influence the -release profiles.  The conditions under which  the
resin is cured affect bond strength, determining to some extent.
the stability of the resin components.  The character of the wood
itself also affects HCHO release; the more acidic the wood,  the
greater the tendency for acid -hydrolysis and  HCHO release
(Podall, 1984).
      Under normal use conditions, the release of HCHO decreases
with  time, as discussed previously.  Emission reductions linked
to product aging relate to a  decrease over  time  in  both the  HCHO
present in the board as a residual  from nanufacturinq  and  the
latent HCHO present  in  the board in hydrolyticall.y  labile  resin
and wood components.  The emission  rate decay curve for a  board
is apparently  exponential with  time; the  resid.ual  HCHO is  emitted
at relatively  high rates  followed by a  slow, release of  latent
HCHO.   Althouah  the  short-term  emission rate behavior  of boards
                                                                  *
has been reported  in  numerous studies,  little quantitative
information  is available  on' the long-term emission rates,
particularly  for*newer  products made with low HCHO-urea ratio
resins  or  treated  with .scavengers.
6.4.1.3.   Other Sources  of  HCHO
      Indoor  HCHO concentrations may be attributable to sources
other than pressed-wood products containing  UF resin.   The other
sources can  be characterized as follows:
                                6-10

-------
    o    Urea-HCHO  foam  insulation  (UFFI)  (existing homes  only)

    o.    Products with phenol  HCHO  resins  (PF)
         --    softwood plywood
               hardboard
         --  •  wafer-board  •
               oriented strand  board
               fibrous glass  insulation
               fibrous glass  ceiling tiles

    o    Consumer products  that  may contain  HCHO  resins
         .—  . .upholstry  fabric
               drapery fabric   '
         --    other textiles

    o    Combustion products '
               unvented kerosene .a"hd gas  appliances
               smoke from tobacco  products
             •  combustion of  wood  or coal in 'fireplace's

    o    Outdoor air                     '           .        •   .  .
               ventilation system  air exchange

     Compared  to pressed-wood  products,  with^the  exception  of

UFFI, the other sources  are  usually minor  contributors to  HCHO

concentrations in conventional and  manufactured homes.

     The Consumer Product Safety. Commission  (G-PSC)  in-1982

prohibited the installation  of UFFI in residential  buildings and

schools.  Although It was later overturned by a Federal  court,

the CPSC ban on UFFI caused  the virtual  elimination of the  UFFI

industry (Formaldehyde Institute, 1984).   There is  considerable

debate among the regulatory  agencies'and the UFFI industry  as to

the extent of  long-term  HCHO emissions from UFFI presently  in

place (Hawthorne et al.,   1983).   UFFI is not discussed in detail

in this section? refer to Versar  (1986c) for further  information

and references.

     Though no residential sources  of HCHO have been  as'  well-

studied as urea-HCHO foam insulation and pressed-wood products
                                v
made from UF resins, there are fairly complete data on the  '.
                               6-11

-------
importance of pressed-wood products with PF resins, on fabrics



treated with UP resins for permanent press, on fueled appliances,



and on cigarette smoke as sources of residential  levels.



     Common applications of PF resin pressed-wood  products



include roof and wall sheathing, subflooring,'and  siding.  Small



amounts are used for' shelving, cabinets, indoor paneling, and

                                                              «

fixtures (APA, 1934).  Phenol-HCHO resins  are inherently nore



stable than are UF resins, and pressed-wood  products  nad-e of  ??



resin emit HCHO at much  lower  rites  than do  products  made with O'F .

       »

resins.  The small amount of HCHO that  is  emitted  from the panel



products is the result of residual HCHO that remains  in  the  resin




(APA, 1984).
                w


     There are several published studies on  HCHO  emissions  from



PF pressed-wood panel products.  Myers  and Nagaoka (1981)  found



that HCHO levels in chamber tests  rarely exceeded 0.1 ppm  in the



presence of PF particleboard at  25°C.   Matthews  et al.  (1983,



reports X-XV) tested  PF  hardboard  and  softwood  plywood and



obtained similar results.  Myers  (1983) measured  higher  levels



 (0.3 ppm) initially  in  tests  of  waferboard and  particleboard made



with PF resins, but  levels declined  rapidly.  The American



 Plywood Association  (APA,. 1984)  has  submitted data (reviewed by



 Versar, 1986c)  indicating that PF-resin pressed wood products




 emit  little HCHO.



     Other  generic product lines- containing PF that are used in



 construction  applications are fibrous glass, insulation and
                                6-12

-------
 ceiling tiles.  In 1983, as a result of a study on HCHO  release



 from consumer products (Pickrell et al., 1982), CPSC decided to



 further evaluate HCHO emissions from fibrous glass insulation and



 ceilinq tiles.  These products,  when comoared with other products



 tested, were among the hiqhest group of emitters tested  by  *



 Pickrell'ej: a'l.  (1982).   Concern about these test results arose



 because of 'the high loading rates of these'products in homes.   *



 Under normal use conditions (in  attics),  insulation would be



 subjected  to temperatures much higher than normal room



 temperatures,  thereby increasing ootential HCHO emissions.



      Further evaluation  by Matthews et al.  (1983)  and Matthews



 and  Westley (1983)  (under contract  to CPSC)  indicated that  a



 predicted  increase  of no  more  than  0.022  ppm in indoor HCHO level



 would  result from.use of  new ceiling tiles'or new  insulation.   As



 the-  products aqe,  the HCHO emission rates  and resulting indoor



 concentrations would  be expected  to decline  sianificantly.



      Available data on treated fabrics  (Pickrell  et al.,  1982,



 1984)  indicate that,  with emission  rates only as high  as  115  '



 ug/m2/hr,  these  can be relatively  important  sources in  homes only



 with  large  surface areas  of  furnishings  like  draperies  (at  least



when  new).   The  data  on combustion  appliances show  that HCHO



 release  is  a function  of  whether the  applia.nce  is  tuned and



 functioning  properly.  Gas  stoves may  emit  less . than  2  t.o'nearlv



 30 mg HCH'O  per hour, of use; ;gas heaters can  emit less  than  5 to



over 60 mg/hr, depending  on"the efficiency of burning;  and  new  '



kerosene heaters emit  up  to  of 3.9  mg/hr of  HCHO.  (Traynor et al.,
                               6-13

-------
1982; Girman et al., 1983; Fortmann et al., 1984; Traynor et al. ,
1983; Caceres et al, 1983).
     The emissions data on sidestream cigarette  stnoke range from
20 ug per cigarette .(Bardana, 1984) to nearly 1.5 rug/cigarette
(reported by Matthews et al., 1984).  Several studies, however,
concur on an emission rate of 1.0 to 1.2 mg/c'igarette.  The
importance of this  source  is obviously related to' use patterns.
Studies where numerous persons chain-smoked in a poorly
ventilated room  (Timm and  Smith, 1979) did indeed show that HGHO
levels were elevated after a short period of time, but other
studies  (Traynor  and Nitschke,  1984) in  the homes of  smokers
indicated that,  at  a smoking rate of 10  cigarettes per day,. HCHO
levels were  not  elevated  over controls with similar  loading rates
of other sources.
6.4.2.   Garment Manufacture
      The principal  source of HCHO  in  the garment manufacturing
workplace,  is  the release of HCHO  from  fabric  treated with resins
that impart durable or  permanent press  properties.   The  textiles
normally treated are blends  of'cotton,  acetate,  and  rayon. These
fabrics  account for 60-80 percent  of  the textile produced
annually.
      The re«in of choice is dimethyldihydroxyethylene urea
 (DMDHEU) and its alkylated derivatives.   It is estimated that
approximately 90 percent of the durable press resin market is
accounted  for by DMDHEU.  Other resins used are urea-HCHO,
melamine HCHO, and carbamate resins,  plus a HCHO/sulfur dioxide
vapor phase process.

                              ..  6-14       '   •   ' '     •       •

-------
      HCHO is  released  from treated  fabric  in three phases.   In



 Phase I,  any  HCHO loosely  held  by Van  der  Waal  forces  is  released



 as  the fabric is  dried.  Release of HCHO by  this- mechanism  is



 usually complete  by the  time  garment workers receive  the



 fabric.   Surface  desorption occurs  during  Phase  II.   This
            S?


 represents the  release of  HCHO  which is  not  covalently  bound to



 the  fabric, and can last up to  240  hours.  The material is



 normally  stored during this phase,'and increased  ventilation can



 increase  the  rate  at which HCHO is  desorbed.  Phase III,  in -which



 hemiacetal hydrolysis is the  mechanism of  release, -is thought to



 be the phase  of HCHO release  which  results in worker  exposure at



 the manufacturing  site.  Release of HCHO by  the hydrolysis



 mechanism  is.  independent of air changes, but dependent  on



 humidity and  temperature (Ward, 1984)



 6.5.   HCHO Levels in Homes and Garment  Manufacturing Sites



 6.5.1.   HCHO Levels in  Homes     '                   *



      Table 6-3 briefly summarizes the residential  HCHO monitoring



 studies reviewed by Versar (19S6a,   c).    However, because of the



 changing nature -of pressed-wood products with UF  resins and the



 constant- evolution and improvement  in monitoring techniques, this



 residential monitoring data base is not  the  most  appropriate for



describing current HCHO exposure in homes.    Many data sets are



based on investigation of homes from which complaints of HCHO



symptoms have been filed; these data sets may not be



 representative of average exposure because of bias toward high



concentrations.  Homes studied before 1980 were built with
                               6-15

-------


1
1







t
• «•• q
|
•«•
a
1
1
u»
"5
MM*
«J
1
O
y.
I
»5

ri
1
o
O
2
6-



j*
C
3




1

Ol
i
% i
- i
l"
S o
It
ut
at
1

*s
u
J


1
O
1
Study/sapling date(s) Wu
/reference)





^i
e
i%
|


\fi
d
i
d



d "
&

i
i






*
OOUVENIIOUAL HONES
Fleming 1 Associates

^
^
"S
i
j
h
e
•f5
8 M
U -
Primarily
measureme

o
A
«i
d
V-



i


«n
^-






s-
New York Study
(Iraynor t Nitschke 1964)
Univ. Washington (1962-1963)
(Breysse 1964)

• —
-* S
|j
UJ ^
«> j*
e &
•^ i
^» Q
? 2
>. —
2-2
2 7
01 S
Includes
weather \i
the U.S.
Z 2
CM O
d d
2 2
u*i un
l»



!


i
i





is
SS
LBL (1919-present)
(Ginun et al. 1963)
•^5 — /i

^ VI
-"&i.-
•« S j?
-* u S
S c
i -^
s. • •<
5^2
_« -a 'S

- " w
e vi s
1 5 Si
O 2£ vi "O
W VI >»
- ? 1
_aj 5 'a
3 • VI ^
— U 01 1
u * e ^ 9
. «« 4| .S3 » <
S^j -fl * T
is a- !•
a -c s 3 s >,
S *l s & • "§
S* J 22 3
2 «< 3 — O)

-------










I
•~
'I

r*l
1
d
j"»4
_t$
'e-










wt
e



1

£
J
s i
- 8
It!
C Ql
^u ^

S
"a.
5
M
tte
O
|
ut
1
O


Study/sampling date(s)
(reference)
5
j
3
1
' - i
^
2.S
'Z 8
. a. w
i
>•> .••>
** ® 1
O 'fl K




1
1




t
1





*


I
if
if'
1*.
* s
' IT "3 *
^ M
2 *•"
.If | • J "•
§^± -- ™
^ o 2
2. §• i w
«• V i '* •
- s -5 ^
< ^  • U
* ^ si
Is S 5
ill i i!
«> •< — > "S. a
a a -S .£ | v
i I «» f we
si 1 i a
<^< f 9 5 5 -
>,! v v .|
i^I I II
I x S§
d d d d
22 22
0 0 2 0
d d d d



S -. | i
«



rf» . T
|?|

in
2
2
i



S
d




o
2




§


Indiana Board of Health
study (19)9-1983)
(Konopin&ki 1983)

-------
1





^,
i
|
<•*>
1
O
O
"fl
i-1














Connenls
S '
£
£•
ee


M
ll
1 S,
i!


Ul
01
1
*s
1
*
u
i


-j
01
3
1 s| |
if'l- f ' 1
5 <=• S 0 2 . -a •$
f -S 5 o ' 3 T! J ' ' ?
i!1: * . li « •
l|?i 2; |2 "ri la-
1 1 1 11 ! H
Ml I'll* !•!
"8 I 3 S . - "3 22 Is
fill * il-'i P
s^ias 5 3s £ * s
^ 2 ° S
O ° pj
2 2 o 2
••*
S 5 . - 5
, ? ' ° °
ut
s
o
2
8 -5 1 a 5
0 0 | 0 0 —



1 I ! i ! '.I.


f m 4  —
1 * S 3 '*"'—. . i "*
e— s i * s •* 2 * ^
— -J S •— * .£ * S* 8 e
!l § i I] li ll ll
xl ull jg ^m# g WC V* « ^ jl
t-a | i li II IS il
333 3 a- a- .•*•" a •
Assuming 601 of total aldehydes is
formaldehyde, (ton-complaint hones.
2
2
•*»
o
0


„
d



S


N


2
2
,j'
^
«
u» S3
i s|
_ i flj gf
5 §2
i 1-
31 of 822 Measurements >I.O ppn. Co^>!
hones.
3-oonth old home built specifically f"'
o ^
T 0
o e
^tm «•
o ^
v . o,



*
! • ®



«M £
• •

8


1
i
i
^
s <-»
.11
3 S« $
. J c

Men complaint, occupied and nouotcupiei
Concentration by home age evaluated.
*9
2 'S


O *-*


f
2|



!
•

CSI




~
S S
^ 2 .""
i 1 7;
1 '! t

^^•H

-------













I
"Z
%
w
rl
I
a .
s-





1

wt
J




s

gj
s

i i
I1

S o
l!
2
1.
•3
i
wt
1
1

Study/sailing date(s)
(reference)
i s
If
|? 1
S * sf
ii £•
u £ v
"So a
11 *
8 § |

a'.
4
d d

i i
i i


- a

lennes&ee (1982-1983)
(Hodges 1984)
f ?.
1 H
a. |l
Ol M. Al
— . U C
4 * 41
2 - 1
U s| s
«* 2 c
- . 1 5 i
.£ | |§2
II ill
£ 3
— d
2 2
o 3
d d




£3 ^B ;
d d

it
i i


i s

«^
a ?->
i ^
1 ^ . O* *••
^
1
s
i"
i
* *
f
oa
.2
vt
ut
5
d

2
^f
d




^r
d

i
i


«n
I
SAI California survey (
(SAI 1984)
ta» •
S !
?
1
i
o
• il •*'
s .£
| 1
v< e
^ .
d
o
S
o
o
9




I
d

i


3.
ii
California State survey
(Sexton et al. I985a.



















i
2
^rf
^ .5
w ••*
41 ^
1 8
1 *
- Insufficient data in
NO = Not Detectable, or

-------
with products made of high HCHO:urea ratio resins that are no
longer on the market; they cannot be considered as baseline
exposures for that reason.  The most appropriate data for
describing current exposures in mobile and conventional homes
are, therefore/ those generated by random sampling of
noncomplaint homes after 1980, preferably after 1982 (when
manufacturers began using resins with mole ratios of 1.5 (F:U)'or
less)..  These restrictions on t-he "appropriate" data base still
leave a considerable volume of monitoring data on levels in
homes.  Table 6-4 summarizes the noncomplaint  (random) data on'
HCHO levels in conventional and mobile homes.
6.5.2.   Manufactured Homes
     HUD has recently promulgated changes in i.ts Manufactured
Home Construction and Safety Standards (24 CFR 3280).  The
changes, published in the FEDERAL REGISTER of  August 9, 1984 (47
FR 31996), set product emission standards for  particleboard (0.3
ppm) and plywood (0.2 ppm).  HUD believes that if the. produce
standards are met and no other major emitters of HCHO are present
(e.g., medium density fiberboard), ambient levels will not exceed
0.4 ppm (EPA estimate of 0.15 ppm as a 10 year average) under
certain t«p«rature, humidity, and ventilation rate conditions.
Th© HUD regulations, however, were designed to reduce acute
reactions to HCHO and are not based oh HCHO's  potential
carcinogenicity in humans.
                               6-20

-------
Table 6-4.   Sunnary of Residential Monitoring Data from Randomly-Sampled Haws
Nutter
of homes
Conventional
30
40
17
29
3.1
6
120
29
103
78
51
Mobile
2
259
137
121
3
663
Mean (pom)
-- ' •
0.040
' —
0.05
0.060
0.063
0.084
0.09
0.05
0.027
0.07
0.038

0.21.
0.62
0.38
0.18
0.114
0.091
Range (pom)

0.007 - 0.151
0.25
0.013-0.34
0.046 - 0.153
—
0.03 - 0.07
—
<0.008 - 0.29
0.013 - 0.085

0.07 - 0.46
0.02 - 2.9
0.02 - 2.26
0.04 - 0.35
0.068 - 0.144
<0.01 - 0.48
Investigator, date
(date of monitor tinig)

Traynor 1984
Sirmtn 1983 (1979-83)
noschandreas 1978 (1978)
Hawthorne 1984 (1982)
Schutte 1982 (1980)
SAI 1984 (1984)
Koncpinski 1984 (1979-1983)
Godish 1983
Conn 1981
Stock and ntndez 1985 (1980)
(includes apartments and
condominiums)
Sexton et al . 198Sb (1984)

aoschandreas 1978 (1978)
Singh 1962 (1980-1981)
Anderson 1983 (I960)
University of Texas 1983 (1982)
SAI 1984 (1984)
Sexton et al. 19656 (1984)
                                    6-21

-------
     EPA estimates a ten-year average ambient HCHO level of. 0.10
Dpm for new manufactured homes.  EPA has used this estimate and
the estimated 10-year average for new homes that  just meets the
HUD target level of 0.4 com  (0.15 ppm) in the quantitative cancer
risk assessment.  Another study has reported average levels of
0.54 pom for manufactured homes less than three years old and
0.19 ppm for homes older than three years (State  of Wisconsin,
1983).  The Exposure Panel of the Workshop (1984) reported
studies that showed average  ambient levels of 0.38,ppm  for
manufactured homes not subject to complaints about HCHO odor -by
residents, and averages of 0.38 ppm to 0.90 ppm for complaint
homes.  Thus, an unrealistic worst case exposure  estimate was not
used to estimate human risk.  Also, only 10 years of exposure
were assumed for manufactured homes.  Specific exposure data
follow.
     The average HCHO level  in mobile homes appears to  have
declined in recent years due to the use of lower-emittina stressed
wood products in mobile home construction and to  the natural
decay of HCHO emissions from products in existing mobile hones.
Average levels in the existing stock  of mobile homes are now
around 0.2 to 0.5 ppm/ with  mean levels in individual- homes
(including-complaint" homes)  ranging  from less  than  0.;  to over
1.0 ppm.                               .
     This apparent decline  is  shown graphically  in  Fiqure 6-1.
The Conyers  (1984) study of  complaint mobile homes,  initiated  in
1980, showed mean HCHO levels  of 0.85 ppm  in new  homes.  An
                               6-22

-------
a ' '
a
2
a . ,
« 2
1 • '» I i i
ft 5 § 5
i - : * i i
a * * -JS s- . >
S * ~ • S • a
< * 3} 2 • j 2
H Si S v 3 ^ 2
« < s < - 8 S
I i \ 1 : 1 I
• • a Z a a _
1 3 i 5 I 5 2
a • '
i !•«•*+$
* .

•
*













;


' •
• 1 1 1 i 1
9 0 6 00 <





•3> ^
T-- •



" *

^ • ^>
^ ^

* • *

4 •


* 4 *

• ^b
^F

• **

• +«

• «^

• *«
• •'

1 i 1 r ^
• rt w ^
i oo «





-I
-m
^ U
^5
kw
- | ^
u.
2
-| I
u.
O
^ 9
" >
O
9 ^*
"™ 2 2
II
mm 3 y S
~ ^ V3
• • i
2 r
K < -,
-S 2 3
is
« 22
•• S > —
•• u
a
-a I
•z
* —
_. 2 .u
2 >
-u
^
-3 ' -
•* i

-------
exponential function describing the relationship between HCHO
level and home age (r2=0.35) for the combined Singh (1982) and
Anderson (1993) data (i.e. the Clayton/Wisconsin data set) (1200
data points) predicts an average level of 0.5 pom in new 1970 to.
1980 vintage mobile homes (noncomplaint).   Results of studies
begun in more recent years (University of Texas, 1984; MHI, 1984;
Sexton et al., 1985; Groah et al. ,  1985) indicate that initial
HCHO levels in new homes on average fall within the range of 0.2
to 0.3 ppm.
     Using the exponential, function describing  the Clayton/
Wisconsin data to estimate decay of HCHO emissions over time, 10
year average concentrations can be estimated.   For initial
concentrations in new homes of 0.5 ppm  (i.e., Clayton/Wisconsin
data set), 0.4 ppm (i.e./ the HUD target level), and 0.25 ppm
(i.e., midpoint of range of recent study of new home levels),, the
10-year average concentration estimates are 0.19 ppm, 0.15 ppm,
and 0.10 ppm, respectively.
     The fraction of homes with elevated levels of HCHO also
appears to have declined in recent years.  Figure 6-2 shows that
the majority of homes less than 215 days old in the Clayton/
Wisconsin data set had HCHO concentrations above 0.4 ppm.  More
recent studies indicate that this fraction is decreasing.  The
California survey of 663 mobile homes (Sexton et al., 1985)
reported levels exceeding 0.4 ppm only  in two and three-year old
homes.  The Texas study (University of  Texas, 1984) reported that

-------

i i i
Q 9 f*
- • ' • '
r A -v

'•»!••»•
z z z
5W W W
•:•.•. . -
8 <* • •
§
w




«
'
•
• ' .
•
• •
I
5

•
"* • . <

1
4
«
•
• • . - 4
.
. !
• "
•
• ' • +
'
9«
'
• 4
• 4
• 4
4


. s
s

*•*
-2 .
•«' <
••5
S «•
"I |
• 1
IS
" 2
-1

"™ «
"!
*N
8
                 > J"
                 s I
                1^1
                 s «
                 SC v
                 6-25

-------
the highest mean in any-=group of homes was 0.35 ppm (ten hones in
one county less than one year old); it is likely that one or more
of these had levels above 0.4 pom, but not approaching 1.0 ppm.
     Levels measured at any one temperature and humidity can,
however, be misleading.  Table 6-5 which  illustrates the effect
of temperature and humidity changes on a  0.4 ppm reading at 25'C
and 50 percent relative humidity (the HUD target) shows'  that
under more extreme conditions (30°C/70 percent RH) , the .predicted
level could rise to 0.92 ppm.  Because changes in temperature  and
humidity occur over the course of a day and with seasonal weather
f luctuationsr homes without constant climate control would
therefore be affected.
     These data illustrate clearly tha.t HCHO levels  in homes  are
the functions of multiple variables; neither age nor temperature
and humidity, nor  any  other variables can account  for all
variations in residential levels  (Versar, 1986b).
     As  the foregoing  illustrates, HCHO levels  in  new
manufactured homes were  tending  toward 0.4 ppm and  in some 'cases
above,  until about 1979.  After  that date, mean  HCHO levels  in
new manufactured homes began  to  fall or level  off  slightly  below
0.4 ppm.   Even so, peak levels  above  0.4  .ppm  can be expected  at
times du« to adverse temperature and humidity conditions.
                                                            The
 frequency for such peaks is not known with confidence, but based
 on  the  data  available (see. Tables 6-6 and 6-7,  and Figure. 6-1)
 they could be expected to occur in a substantial fraction of new
 manufactured homes.
                                6-26

-------
Table 6-5.  Potential  Effects of Temperature and Relative
               CMn9»*.an Formaldehyde Air Concentration* (pan)*
Relative hutidity
Temperature 301
S9*f (IS^C) . 0.08
68*f <20«C) 0.15
77»f (25»C) 0.24
9S*F (30*C) 0 . 40
401
0.11
0.19
0.32
0.53
SOI
0.14 .
0.24
0.40
O.M
601
0.17
0.29
0.48
0.79
701
0.19
0.33
0.54
0.92
  Calcul*t*d (rting Kjuationsin  Myers,  l984uAicft MT* 
-------
                           Table 6-6.
        FREQUENCY OF OBSERVATIONS FOUND  IN CONCENTRATION
         INTERVALS BY CLAYTON'ENVIRONMENTAL  CONSULTANTS
Concentration
Interval (ppm)
0.0 -
. 11 -
.21 -
.31 -
.41 -
.51 -
.61 -
.71 -
.81 -
.91 -
1.1 -
2.1 -
Number
.10
.20
.30
..40
.50
.60
. 70
.80
.90
1.00
2.00
3.00
of homes
1 Percent of Sampled Homesa
<0.5 yrs > 0.5-1 yr All Hones
3.6
7.9
6.5
7. 2
5.8
6.5
5.8
5.8
6.5
12.2
24.5
7.9
139
8.0
4.0
36.0
16.0
0.0
12.0
16.0
4.0
0.0
4.0
0.0
0.0
25
8". 1
19.7/
14. 3
9. 3
5.0
4.6
4.6
3.9
3.9
7.7
14.7
4.2
259
a 259 "noncomplaint" mobile homes up to eight years old were
  sampled in 1980-1981.  Three measurements were typically- taken
  in each single-wide home and-four measurements were taken in
  each double-wide home.  The data in the Table reflect the
  average concentration measured in each home.

Source:  Versar  (1986a) statistical analysis of data  supplied by
         Singh et al.  (1982).

-------
                             Table 6-7.

          FREQUENCY OF OBSERVATIONS FOUND IN CONCENTRATION

             INTERVALS BY WISCONSIN DIVISION OF HEALTH
Concentration
Interval (ppm)
0.0 -
.11 -
.21 -
.31 -
.41 -
.51' -
.61 -
.71 -
.81 -
.91 .-
1.1 -
2.1 -
Number
.10
.20
.30
• 40
.50
.60
• 70
.80
.90
1.00
2. .00
3.00
of observations
• Per;
_<_0. 5 yrs
*
2. .63'
29.0
0.0
10. 5
10.5
13.2
10.5
7-9
2.6
2.6
10.5
0.0
38
:ent of Observations'* ' '
>0.5-1 yr All -iorr.es
3.8
13.6
21.1
14.6
11.3
12.2
8.9
5.6
3.3
0.0
5.2
0.5
213
14.1
20.4
18.4
14.0
9.2
8.0
5.2
3.6
2.2
0.7
3.8
0-. 3
976
                            h°mes Up t0 nine vears old were
                         Each home was sampled at least six

      rel13'  ^ data in the table reflect
      results of 976 measurements.


Source:   Versar (1986a) statistical analysis of data supplied by
         Wisconsin Division of Health (1984).
                              6-29

-------
6.5.3.   Conventional Homes

     The average HCHO levels reported in several monitoring

studies of conventional homes range from less than 0.03 to 0.09

ppm (see Table 6-4).  Newer homes and energy efficient homes with

low air exchange rates tend to have hi-gher HCHO Levels (often

exceeding 0.1. ppm) than older homes (Versar, 1986c) .   Results of
                               *
recent studies indicate that initial HCHO levels in new

conventional homes generally fall within the range of 0.05 tp 0.2

ppm; few .neasurements exceeded 0.3 pom (Stock and Mendez, 1985;

Hawthorne et al., 1984; SAI, 1984; Wagner, 1982).  Computer

modeling to estimate initial HCHO levels in conventional homes

built 'using significant amounts of pressed wood  (i.e., either

underlayment, paneling or both) yields values ranging from 0.1 to

0.2 ppm  (Versar,  1986  ).  Using the exponential decay function

described in Section 6.5.2, the 10 year average concentration  for

a'home with an initial concentration of 0.15 ppm  (i.e.,

approximate midpoint of range of new home levels)  is estimated to

be 0.07 ppm.  Summaries of  some of the major HCHO monitoring

studies are presented below.                                 .

     The Lawrence Berkeley  Laboratory  (LBL) has  summarized HCHO

concentrations in 40 residential  indoor environments since  1979

 (Girman  et al.,  1983).  They have  found that HCHO  concentrations

in homes designed to be energy-efficient  are somewhat higher than

concentrations  in conventional homes.  The  maximum reported  value

is 0.214 ppm  in  an  energy-efficient home  in  Mission  Viejo,

California.   Data are  not  sufficient  to allow  calculation of mean


levels.                •     ' '

-------
      As part of the development of an  indoor air pollution model
 'based on outdoor pollution and air 'exchange rates, Moschandreas
 et al. (1978) studied the patterns'of  indoor aldehyde'levels
 monitored in 17-houses in the U.S.  These data can be useful i£
 'we assume HCHO constitutes 60 percent of total aldehydes, based
 on L8L data (Girman et al., 1983).  The 17 houses had an average
 aldehyde concentration of 0.09 ppm.  Applying the 60 percent
 factor,  the average HCHO concentration for the houses would be
 0.05 ppm.   The highest mean for any one home was 0.26 ppm; the
 range for  that home was 0.2 to 0.45 ppm.  Another home with a
 mean of  0.20 ppm reported a range of 0.07 to 0.5 pom.  For -no
.other conventional  home.did levels exceed 0.4 ppm,
      A University  of Towa Study . (Schutte et al.., 1981),  oerforted
 for  the  Formaldehyde Institute,  monitored 31 conventional,
                                    81
 detached  homes not  containing  urea-HCHO* foam insulation  (UFFI)
 for  HCHO concentrations  in the indoor air.   Samples  were
 evaluated, in relation to outdoor  HCHO concentrations, age of the
 home,  and  other  environmental  factors monitored at each  of the
 sampled  homes.   The average indoor concentration found in the
 homes  was  0.063  ppm (standard  deviation = 0.064)  with a  ranrje of
 0.013  to 0.34  ppm.   In  only 5  o^f  the 31 homes were average
 concentrations higher  than or  equal to 0.1  ppn.
      The  1981  Canadian  study  (UFFI/ICC,  1981)  also studied nor.-  '
 UFFI  homes.  Table  6-8  summarizes  these data,  showing that levels
 in none of  the 378  homes  exceeded  0.2 ppm.
                               6-31

-------
   Table 6-8.  Comparison of Non-UFTI Canadian Homes
                 by Average HCHO Concentration
Avtrag*
foraaloirtyo*
concentration (ppi)
<.01
.01-025
.025-040
.040-055
.055- 070
.070-. 085
.085-. 10
.1-15
.15-. 20
Totals
NuA*r of
taws
48 .
Ill
97
67
30
15
—
9
* 1
378
*~,u.
12.7
29.4
25.7
17.7 .
7.9
4.0
. — .
2.4
0.3
100.1
Cunulativ*
p«re*flUgt .
12.7
42.1
67.8
85. 5
93.4
97.4
—
99.8
100.1

Sour«:  UFFI/IOC (1981).
                           6-32

-------
      A report,by Virgil J. Konopinski (1983) of the Indiana.State
 Board of Health summarizes the 'results -of a series'of
 investigations conducted from, 1979 through 1983 to determine HCHO
 levels in  conventional  homes in Indiana.   The mean' HCHO level in
 the 120 homes without UFFI was 0.09 ppm  .(0.05 for homes-with
 UFFI).  That  mean could be skewed  by the  maximum concentration of-
 1.35 ppm reported in -one home.  Neither  the age of the homes nor
 the age of  the UFFI installations  was reported.
      From  April to mid-December 1982, Oak Ridge National
 Laboratory  (ORNL)  with  the U.S.  Consumer  Product Safety
 Commission  (CPSC)  studied indoor air quality in 40 east Tenessee
 homes.   The objective of the study was to increase the data base
 of  HCHO monitoring in a variety of American homes and further.
 examine the effect of housing types,, inhabitant lifestyles, and
 environmental  factors on indoor pollutant levels.
      Homes  to  be' sampled were selected based on a. stratification
,                .                                                  &
 to  ensure  representative home age, insulation types,  and  heatina
 sources.  All  were voluntarily enrolled.   Twice a month,  four
 samplers at each location monitored HCHO  levels in three  rooms
 and  outside the  house.   Samplers were exposed to the  air  for
 24-hour periods.  'No  modifications to the residents'  life styles
 were  requested  during these  measurements.
      Table  6-9  summarizes these  data by  home age and  season
 (indicative of  temperature-and humidity).  HCHO measurements  in
the  40-home east Tennessee study  led to  the following major
conclusions:
                               6-33

-------
Table  6-9.
ORNl/CPSC «ean Formaldehyde Concentrations  (pom)
as a Function of Age and Season (Outdoor Ae«ns Ar«
Less Than 25 ppo Detection Limit)
Age of nous*
«n
0-S years
5-15 years
Older
0-S y»*rs

=
5-15 ywrs


olo»r


all


S«a«en
all
all
all
all
spring
suMwr
fall
spring
sunwr
fall
spring
SUMT
fall
spring
SUMW-
fall
X*
0.012
O.OS4
0.042
0.032
0.087
0.111
0.047
0.043
0.049
0.034
6.036
0.029
0.026
0.062
0.083
0.040
s*
0.077
0.091
0.042
0.042
0.093
0.102
,0.055
0.040
0.048
0.03S
0.051
0.037
0.023
0.076
0.091
0.047
• n
5903 40
3210 18
1211 11
1482 11
1210
1069
931
626
326
259
757
341
384
2593
1736
1574
Hot*:   x  « BMII concentrations.
       s  » standard deviation.
       •  • nuttar of

       Include* hows with and without UFfl.
         Hawthorne et al. (1984).
                              6-34

-------
     (1)   Th« average  HCHO levels exceeded 100 ppb .(0.1 ppm)  in 25
          percent  of the  homes.                      .
                                   *
     ,(.2)   HCHO levels  were found to be positively related to
          temperature  in  homes.-  Houses with UFFI were freauently
          found to exhibit a temperature-dependent' relationship
          with measured HCHO levels.

     (3)   HCHO levels  generally  decreased with increasing age of
          the .house.   This is  consistent .with decreased emission
          from materials  due to  aging.

     (4)   H.GHO levels  were found to fluctuate significantly both
          during the day  and seasonally.

      Studies by Breysse  (1984)  evaluated conventional/ non^UFFI

 homes.   The  University of Washington studied 59 such homes;

 private  laboratories,  in  the state studied an additional 25.   The

 freauency distribution for measured  levels are presented in  Table

"6-10.  A total  of 6 of. the 189  samples.(3.1 percent)  were over

 0.5  ppm  and  56 samples (26.5  percent) were over 0.1 ppm.

      Traynor and  Nitschke (1984) monitored indoor air pollutants

 in« 30 homes  with  and  without  suspected combustion (and other)

 sources,   the average HCHO level observed in all the test homes

 was  40'ppb;  a high value of 151 ppb  was found in one of the

 tested residences categorized as containing new furnishings  and

 new  paneling as a suspected pollution source.
                               6-35

-------
         Table 6-10.   Frtqmncy Distribution of FonMldthy 1-0
> 0.5 -0.99
> 0.1 • 0.49
< 0.1
TOTAL OBSERVATIONS
2
2
41
«S
113
0
2
9
65 ,
76
1.0
2.1
2*.S
70.4

Sourca:   Irtyss*  (1964)
                                     6-36

-------
     The results can be summarized as follows:

    o    The 4 homes with no  identified source had a range of
         means of 0.007. to 0.034 ppm.

    o    The 3 homes with new furnishings had a range 'of means of
         0.015 to 0.061 ppm.

    o    The 4 homes with cigarette smokers had a range of means
         of 0.032 to 0.060 ppm.

    o    The 18 homes with gas, coal, and wood fueled
         appliances/heaters had a range of means of 0.012 to
         0.056 ppm.

    o.   The 12 homes.with a combination of sources reported a
         range of means from 0.013 to 0.064.
                                -                *
Variations in home levels could not be attributed to combustion

sources.

     Stock and Mendez (1985) measured HCHO concentrations inside

78 homes in the Houston, Texas area during the summer of 1980.

No mobile homes, UFFI homes, or complaint homes were samoled.

Indoor concentrations ranged from less than 0.008 ppm to 0.29 ppm

with an average value of 0.07 ppm for detectable concentrations

(Number of samples, N»75).  Three energy efficient condominiums

had, as a housing category, the highest mean level (0.18 ppm).

Condominiums (Mali), apartments (N=*19), and energy-efficient

houses (N*7) represented the mid-range with mean levels of 0.09,

0.08, and 0.07 ppm/ respectively; the mean of 38 conventional

houses was 0.04 ppm.

     Wagner (1982) measured HCHO levels in 12 California homes

that fall into a prescribed "worst-case" category~of buildinq and

occupancy characteristics' (i.e.-, low infiltration and ventilation

rates, new construction, presence of gas stoves).  Weekly averaqe
                               6-37

-------
concentrations ranged from 0.078 to 0.163 ppra with a mean of
0.106 ppm.
     Sexton et al.  (1985) measured HCHO- levels in 51 home
dwellings.  Weekly average concentrations ranged from 0.013 to
0.085 ppm with a geometric mean of 0.035 pom and an arithmetic •
mean of 0.038 ppm.   Seventy-six percent of the homes were more
than 10 years ,old and only two were less than six -years old.
     A downward tend in HCHO levels in conventional homes is seen
in Figure 6-3.  The relative proportion of low HCHO lev-els in
homes that have been monitored has increased over the past six
years, and the proportion of high  levels have decreased.  These
data are limited and caution in interpretation is recommended
(Versar, 1986a).
6.5.4.   Garment Worker Exposure
     HCHO levels in apparel manufacturing facilities were
generally below 3 ppm prior to 1980 (see Table 6-11).  OSHA had
established a 3 ppm TWA (time-weighed  average) in 1967.  However,
OSHA is presently considering establishing a new level  (see 50 FR
50412; December 10, 1985).  The ACGIH  (American Conference of
Government Industrial Hygienists)  recommended level, is  1 ppm
TWA.  In r«c«nt years, HCHO levels observed were generally below
1 ppm (see Table 6-12).  The data  in Tables 6-11 and 6-12 must be
viewed with caution because in 1983, the National Institute for
Occupational Safety and Health  (NIOSH) discovered that  the
commercially prepared inpregnated  charcoal tubes, which  had been
used in previous personal monitoring studies  were unstable.
                               5-38

-------
Z

§
i
*M
a     TI
1     •!
                              <
                              a
                       <      §
                      :      1
                                                                                                                           I
                                                                                                                           s
                                                                                                                           <
                                                                                                                           2
                                                          5-39

-------
    Table 6-11.   PRE-1980 MONITORING DATA FOR GARMENT MANUFACTURING  AND
                                CLOSELY RELATED INDUSTRIES
                        •i

                        *
                        et»1««
                         0.1 - 1.4 (1171)*

                         1-11
                                                 0.3 • 2.7 (TWA-4f*«. IHt)


                                                 •0.1 - 1.4 (T\a, 117*)'

                                                 o.is - O.M (TV*, mo*

                                                 O.I • 2.7 (TWU*rt». 1H4)

                                                 o.ooc-4.»*
(i.t. - irfti »»•*». eiatn-
Iftf •*•*!•*» Utr«l. «te
t?

Mi



•s

•s
                                                        . Wl)
                                                 2.600 •
                                                  1179)
                                                 2.2M
                                                  1*79)
0.030

O.I • 3.3 (Ct1Hfli.«r««. 19t<)

0.13 • O.«i (l»i»)4

0.04 . 0.73 (TWA, U71J*
                            4)

                            •S


                            32


                            10

                             1



                            «c
 1

Hi

•S

22
                                                                                        >S
           1


           1

           I

           2

         i.:


          Ml


           t

          Mi
        CM 125.
            >  t CM 125.
    * I CM 31».
    eu
           J54.
             «r
             i*
                                                                                          15
 ••t
                                              5-40

-------
o —


<«

   2as
   3
LiJ


UJ

ae
_
li
• T
: i
M

«
M
l\

* • *

1 •
*



|









1

s V
•I *

s



«*
»
1
X
3
|

•
o
a








"s
1

1*
w X
P
2 ;
S





i
*? %
MH IJi
s s • • — v
r 3 s s rss
s s s s ?••
« • «• 4 • •
a, - • • «--
»w •' •« <«• • ••
*a
ft ^
«•
*
V • » *
* , * *
J If
* *i
1 J3*
s 5i s s -S3 5
I * .
6 "3
: 212

I £* |8.lt- *3
f 51 I|:| ss-
41 ^E ^ ^ ^ M C * w
^S • 7 ^ • * ^S • J»
Jill iili li
s s





i
1 11
« ' » *
« • *
nil
f_ T-n-i» *
** s;5 «
^J^v ^ ^«i
rjjda
• — o' • o
_,
^
J
*

j
1?
^ a
si


J9
^rf
if
•tf •
wJ

9


2» * * «M«to
S3 3 -
! i
III if
••
• • te •
55- ?5
- ii1. 4i
^ •**
ss: «?
• M • ^^ MA
J82S ?2
— o o' o° * o'
J{
•i

s •
t 1
! r
b b
^ » ^9
V . W
s'i si


« _
fcrf ^rf
I": i«.
a= ii
•il 1!
W X W «






s •
s - L I-
r 1 |s --is
! if |i iiis
1 Ij 41 ill5.

£ S1™* "t?'85SS
• • • ®M * * * 
-------
cs
1-1


VO


2 ••( •« »« ••
i S s 1 i
                                        i
                                        *
                                        i
                                        •
                                           t


                                           r


                                           1

                                           •

                                           1
                                           i
                                           >

                                           5
                                            I

                                            i'

                                            S  -

                                            11
                                            a  c
                                            J  r
           i
          t  £ s

          I  5 1

          7  i -

          5  r s
          •:'?
                                            . i- i
                                            t -s r-
                                        |  as ts X2-

                                        ;  I! ^! *i
                                                   iUr
" *
il^-w,
:53
                            5-42

-------
 Thus,  the monitoring  data above may be suspect since the Loss of
                                                          «
 HCHO from the  tubes was  not  consistent. .  Consequently,  the HCHO

 levels  recorded  most  likely  represent lower levels than actual

 conditions.  The NIOSH method  at that time  was also used by OS HA.
          ?
     NIOSH subsequently  developed a stable  medium for collecting

 the  HCHO  and did two  in-depth  industrial  hygiene  studies.  _The

 surveys were done at  two large manufacturing  sites producing

 men's dress shirts.   HCHO exposure  levels were determined  for 54

 of 72 job titles in two  different plants.   The number of

 individuals within each  job  title whose exposure  levels were

 sampled was based on  the. total number of  employees in that

 category  and reflect  a 95jpercent confidence  level that the

 highest and lowest exposed individuals were iacluded in the

 sampling.  A summary  of  the  data  are  presented in Tables 6-1.3 and

 6-14.  These tables show that  all levels  of exposure were  less

 than 0.51 ppm TWA.  Also, as Table  6-13 illustrates,  the combined

 range of  data was  very narrow  (0.01-0.39  ppm)  for 5  of  the 6

departments in the two plants.   The range of  mean concentrations

of all departments (0.13-0.20  ppm)  is  very  narrow and compares

well  within the  overall  combined  mean  exposure level  of 0.17 or.n,

which was used for the quantitative ca.ncer  risk assessment.   In

addition,  the average exposure  levels used  in  EPA's  section 4(f)

determination (SPA, 1984), 0.23  ppm  (area)  and 0.64  ppm

 (personal) (Versar, 1982), were  also  used for  this -cancer  risk

assessment.      ,                               '      .  -
                               6-43

-------
 cc

 (9
 (X

 I
 ra
 *j
 m
 Q

 60

 •V4
 U w
 e c


 51
I!
Z .a
vo
       -


                                                                                 UJ
                                                                            •  •
                                                                            '•o   £
                                     
-------
             o

             fi£
                                   o
                                   CM
                         O   O
                                  O  O
      Q.
      Q_
      CO
                         en
                                   tn  in
      —o
vo
I

•w
us
3S
r*r


0
o

1
pH
• o
9
0

4
o

1 .
1-H
o
0
V
^
0

1
O4
o
o

f^
o

1
1— t
o
O
V
K\
p

I
f— t
O
o

-------
     All of the determinations made in the NIOSH studies were at
one point in time and may not reflect the variation of exposure
over a longer period.  -Factors that could affect variation in
HCHO levels in these plants include variation in ambient
temperature, humidity/ type of fabric or resin system/ and volume
of stored materials or completed work.
     The exposure range across departments/ within plants/ as
well as between plants/ appears to be narrow.  Both these plants
were large manufacturing sites, producing similar products.  Both
plants had central ventilation/cooling systems.  This type of
plant may potentially represent only  10 percent of the total
number of manufacturing sites (though up to 25-30 percent of the
workforce may work in such plants) (Ward/ 1984).
6.6.   Summary
     The data presented above indicate that HCHO levels  in new
manufactured homes are generally below 0.5 ppm/ with  10-year
averages for new HUD Standard homes of 0.15 ppm or less.
                        3*
However/ some fraction of new homes experience ^peak levels that
could exceed 1.0 ppm for periods of time.  It would be expected
that as temperature/humidity exceed 75°F/50% RH, HCHO levels
would rise as Table 6-3 illustrates.  Thus, depending on heating
and cooling preferences/ HCHO levels  in new homes may
substantially exceed the reported mean for new homes.
     The situation is similar for conventional homes/ although
reported mean levels are lower/ 0.03  to 0.09 ppm.  However/
because conventional housing is much  more heterogeneous, peak
                               6-46

-------
levels  in  some  new  homes  may  substantially exceed  reported
means.  Although temperature  and  humidity  conditions  play  a  large
role, construction  techniques which  tend to limit  air exchanges,
such as in energy efficient homes, and  building  product mixes  are
also of major importance.  The ten-y.ear average  HCHO
concentration for a new home  built with significant amounts of
pressed wood is estimated to  be 0.07 ppm.               -  '    •
     Reported HCHO levels during  garment manufacture  are below
1.0 ppm and in some plants below  0.5 ppm,  and the  NIOSH data
indicate rather tight ranges  (none exceeding 0.51  ppm).  However,
much of the reported monitoring data must  be approached with
caution due to the technical  fault discussed earlier.  Building
design, ventilation, and temperature/humidity changes may  be
responsible for daily or seasonal variations.
                              6-47 .

-------

-------
                          B.4

EXPOSURE ASSESSMENT
               Attribute 8
Research or data necessary to improve the exposure assessment
are described.                         •
                   SOURCE  Case Study B. TCDD (Page xxii).  -
                       Note  Several of the case studies noted lack of data on aspects of exposure but
                              contained little discussion of approaches. Excerpts from the case study
                              on TCDD illustrate a limited treatment of the prospect of then on-
                              going human studies.

-------
ncasl
special repoi
NATIONAL COUNCIL OF THE PAPER INDUSTRY FOR AIR AND STREAM IMPROVEMENT. INC, 260 MADISON AVENUE. NEW YORK. MY.
               EXECUTIVE SUMMARY




      DIOXIN:  A CRITICAL REVIEW OF ITS DISTRIBUTION,



       MECHANISM OF ACTION, IMPACTS ON HUMAN HEALTH,



       AND THE SETTING OF ACCEPTABLE EXPOSURE LIMITS
             SPECIAL REPORT NO, 87-07
                  MAY 1987

-------
                            -  xxii -
generally accepted as "safe," although it is likely that higher
levels are also safe)  That is -gratifying because it supports
the idea that levels of 1 ppb or more are not"widespread in the
environment.  -Instead-, they are limited to where .they are
expected—production and disposal sites.   The EPA survey also
revealed that some fish taken downstream from some bleached
pulp mills contained measurable levels of dioxin,,  Those
results identified a possible yet unconfirmed source.

J.    Better Information About Exposure

     Dioxin* is stored for long periods of time in the human
body.  It is stored in adipose (fat) tissue, and the time
necessary to eliminate 1/2 the body burden of dioxin may be 5
years or longer.  Therefore, if a person were heavily exposed
10 or even 20 years ago, he or she might still have elevated
levels of dioxin in his or her fat.  Scientists have standard
formulae to calculate what the exposure levels must have been
years ago to produce a given level in fat today if the
half-life for elimination is known.

     As was mentioned in connection with the proposed Agent
Orange study, the ability to measure dioxin in blood promises
that much more information about exposures will be      ;
forthcoming.   Until that method was developed,  the necessity of
surgical removal of adipose tissue for dioxin measurements
limited the number of ..samples.  Based on measurements from a
handful of studies involving a few score people, the "average"
concentration of dioxin in adipose tissue in North Americans is
estimated at between 2 (perhaps less) and about 6 parts per
trillion (ppt).  Those estimates may be too high.  It may be
that the persons who have been sampled are not representative
of the United States population.  If they are among the
more-exposed people, then the average body burden will be
overestimated.  Although perhaps on the high side, those
averages can be taken as a measure of "background" exposures to
dioxin.

     As more measurements of dioxin in humans are accumulated,
the estimates of,background exposures may become more precise.
However, it is also necessary to have better information about
the half-life of dioxin in humans to refine the calculations.
The estimate of 5 years, based on three measurements in a
single man, may well change.

-------

-------
                              B.5
RISK CHARACTERIZATION
                   Attribute 1
The major components of risk (hazard identification, dose-
response, and exposure assessment) are presented in summary
statements, along with quantitative estimates of risk, to give a
combined and integrated view of the evidence >.
                        SOURCE  Case Study D. Formaldehyde (Pages 1-23 to 1-31)!
                            Note  See Dose-Response Attribute 1 in this Appendix for summary state-
                                   ments on epidemiological data and animal data. Also, see Dose-
                                   Response Attribute 2 for quantitative estimates of risk with a treat-
                                   ment of uncertainty associated with dose-response evaluation.


                                   Another section of the report deals with exposure. Uncertainty in
                                   exposure estimates is not integrated into the risk characterization.
                        SOURCE  Case Study B. TCDD (Pages xxiv to xxvii).
                            Note  These excerpts are from the executive summary. Both Attributes 1
                                   and 2 for risk characterization are illustrated by this text.
                        SOURCE  Case Study H. Methylene Chloride (Pages 108-112)
                            Note  This report did not include an exposure assessment. Thus, the integra-
                                   tion primarily focuses on hazard and unit risk without consideration
                                   of overall risks from specific exposure scenarios. The excerpts also
                                   .illustrate Risk Characterization Attribute 2.

-------
ncasl
special report
NATIONAL COUNCIL OF THE PAPER INDUSTRY FOR AIR AND STREAM IMPROVEMENT, INC, 260 MAOtSON AVENUE. NEW YORK, N.Y. 1001
               EXECUTIVE SUMMARY
      DIOXIN:  A CRITICAL REVIEW OF ITS DISTRIBUTION,



       MECHANISM OF ACTION, IMPACTS ON HUMAN HEALTH,
       AND THE. SETTING OF ACCEPTABLE EXPOSURE LIMITS
             SPECIAL REPORT NO. 87-07
                  MAY 1987

-------
                            -xxiv -
decide which model is a better predictor.  Although there is no
convincing evidence that dioxin has caused human cancer, the
available data do not permit choosing a model.  The failure to
find human cancer could mean that humans are less sensitive to
the carcinogenic effects of dioxin than animals.  Or it could
be that humans have not been exposed to doses comparable to
those that cause cancer in laboratory animals.  Or it could be'
that dioxin-caused diseases are sufficiently uncommon in number
and common in kind that they have gone undetected.  For
example, there is no evidence that dioxin causes lung cancer,
but if it were to, a small increase could go undetected against
the large incidence caused by smoking.

     In the absence of human data to validate a model, ENVIRON
Corporation chooses a threshold model because of the scientific
evidence that dioxin does not act through a genotoxic mechanism
and because what is known about its mechanism of action
strongly suggests a threshold dose must be exceeded and
sustained 'before the events leading to cancer are set in
motion.   It also recognizes that some scientists and the
regulatory agencies of the United States while accepting that
dioxin is a promoter,  prefer to use a no threshold model to
estimate cancer risks.  Finally, we. emphasize that we do not
suggest that a threshold model be applied to all carcinogens
that do not have genotoxic activity.   Each chemical must be
evaluated individually.   However, we suggest use of the
threshold model in the case of dioxin because of the
substantial additional evidence that exists concerning its
mechanism of action as a promoter.

L.   Examples of Risk Calculations

     The discovery of dioxin-contamination of fish downstream
of some bleached pulp mills focused attention on wastes from
those facilities.   Waste sludge from pulp production has been
disposed by land-spreading in Maine.   The material is used for
soil enrichment and can be either applied to the land surface
or plowed in.

     1.    Risks from Consumption of Dioxin Containing Milk

          Some sludges ^ave detectable dioxin levels,  and the
     amounts of dioxin that would be applied to the land in any
     sample can be accurately estimated.   Using sludge
     containing, fifty ppt of dioxin at an application rate of
     10  tons/acre, sludge can be spread on land for 45 years.

-------
                        -  xxv -
At the end of that time, metals in the sludge that will be
transferred to the soil will be near acceptable limits in
many circumstances, and application must end.

     Computer programs were designed to calculate the
concentration of dioxin in the soil for each of the 45
years of disposal plus each of the next 25 years.  The
average concentration over that 70 year span is projected
to be 6.4 ppt.

     People can be exposed to the land-disposed dioxin
through ingestion of dust or contaminated food, through
inhalation of contaminated dust, and through absorption
across the skin.  According to Envirologic Data (a Maine
consulting firm), the most exposed people would be farmers
and their families who obtain all their food from land
spread with pulp mill sludge.  The exposure estimates are
deliberately biased toward the high side; it is unlikely
that a farm family would eat food only from'its own land
and that sludge would be spread across all their land.
Consumers would be exposed to significantly less dioxin.
They obtain their food from a variety-of sources, they are
more distant from the sites of land-spreading, and their
risks would be less.

     Dioxin, if taken up at all by plants, is taken up
very poorly, and little enters the food chain that way.
The most important food chain source is from cattle
ingesting dirt, which they do when grazing.   Dioxin
absorbed from the ingested dirt will be stored in the
animals' fat tissues and excreted in milk.

     The highest exposures were estimated for farmers who
drank milk and ate beef only from cattle grazed on
sludge-augmented lands, who ate corn from the same land,
and who were exposed to the same soils.  The largest
single source is from beef consumption.
                                »
     Use of EPA's estimate of dioxin's carcinogenic
potency produces the largest estimated risks.  It predicts
about 2 extra cancers per million farmers from drinking
TCDD-contaminated milk at the maximum contamination level
every day for a full 70-year lifetime.   Use of FDA's or
CDC's estimate of potency produces calculated risks of
less than one in a million.

-------
                     - xxvi -
     A lifetime risk of (10~6), is a de facto acceptable
dose; concern about carcinogenicity drops off remarkably
when the lifetime risk is less.  Only use of EPA's
estimate of carcinogenic potency with exposure estimates
from land spreading generates lifetime risks greater than
10~6.  Although EPA's estimate is most generally used in
this country, its scientific and theoretical underpinning
are no better than FDA's or CDC's, and probably poorer
than for the ADI approach.   This example illustrates that
risk characterization very much depends on the models used
for estimating risks.
      ^
2.  "  Risk from Fish          .

     Fish are a special problem in considerations of
dioxin.  Although the chemical is, for practical purposes,
insoluble in water, it binds strongly to organic materials
in soils and sediments.   Dioxin-contaminated sediments in
bodies of water can be ingested by fish, and the chemical
is stored in fish fatty tissue.  FDA,-making assumptions
that people eat a variety of fish from different fisheries
and using its estimate of carcinogenic potency,  advised
that fish from fisheries where concentrations were less
than 25 ppt in edible portions could be consumed without
limit.  EPA, making assumptions that people eat  only
contaminated fish and using its estimate of carcinogenic
potency, estimates that fish containing 0.069 ppt (about
1/350 FDA's acceptable limit) pose a 10~6 lifetime risk
of cancer.   The best analytical methods can barely detect
0.069 ppt,  and routine methods cannot detect that low
level in fish.  One conventional response to a
non-detectable level is to  assume that the chemical is
present at 1/2 the detection level.   Therefore any method
that could not detect 0.14  ppt (essentially 2 x  0.069) can
lead to the assumption that dioxin is present at a level
that poses a 10~6 risk.   These very different
conclusions illustrate the  importance that assumptions
play in risk assessments even when similar models (in
these cases, no-threshold extrapolation) are used.
                             >
3.    Risk Calculations and  Models

     No-threshold models are designed to minimize any
probability of understating risk.   Therefore they
incorporate assumptions that cause them to err on the
conservative side (that is  to predict higher risk than are
actually expected).  For instance, EPA's estimates of
carcinogenic potency, including the estimate for dioxin,
are upper bounds on risk.   Use of the upper bound makes it
unlikely that the actual risk is higher than estimated,
and EPA states that "true value" for potency "may be much

-------
                       -xxvii -
lower, with a lower bound approaching zero."  In addition
to the assumptions common to all no-threshold models,
EPA's model incorporates other features that cause it to
predict a higher risk than does FDA's or CDC's model.  The
importance attached to EPA's model because it is widely
used can be balanced against the fact that the other
models are equally valid.

     The threshold model predicts much lower risk.   It is
well supported by biological research and tests, and it
has been adopted by foreign countries.  Given current
understanding of the mechanism of dioxin toxicity,  a
threshold model appears to be sufficiently protective of
human health.

-------
                   United States
                   Environmental Protection
                   Agency
Office of Health and
Environmental Assessment
Washington DC 20460
EPA/600/8-87/030A
July 1987
Extarnal Review Draft
                   Research and Development
EPA
                    Update to the Health Assessment Document

                    and Addendum for Dichloromethane  (Msthylene

                    Chloride):  Pharmacokinetics, Mechanism of

                    Action, and Epidemiology
                         Review

                         Draft

                         (Do Not
                         Cite or Quote)
                         (Permission to
                         reproduce  granted
                         by  EPA)
                                         Notice

                   This document is a preliminary draft. It has not been formally
                   released by EPA and should not at this stage be construed to
                   represent Agency policy. It is being circulated for comment on its
                   technical accuracy and policy implications.

-------
defined, and such risk estimates do hot,  as yet,  take into
account changes in the pharmacokinetic model required by the
adjustment of a particular kinetic parameter.
     A comparison of risk estimates made directly from the human
data provided by the Kodak epidemiology study to risk estimates
derived from .the results of the pharmacokinetic model used by
Andersen and Reitz does not show the animal-based risk estimates
to be overestimates, using upper-bound risk estimates from
respiratory cancer deaths or using either maximum likelihood
estimates or upper-rbound estimates from the pancreatic cancer
deaths  in the Kodak study.             •
     In summary, EPA concludes that the animal evidence of
carcinogenicity conforms to the definition  for "sufficient" in
the EPA Guidelines for Carcinogen Risk Assessment.  The
epidemiology studies, while showing no evidence  of either liver
or lung cancer attributable to DCM, are not sufficient to rule
out a risk  to humans; the data on deaths  from pancreatic cancer
give some weight to the possibility that  DCM may cause cancer  in
humans  at sites other than those  found in animal species.
Overall, the epidemiologic data conforms  to the  definition  in  the
Guidelines  for "inadequate"  insofar as the pancreatic cancer
deaths  cannot  be used  to  establish a  connection  between  exposure
to DCM and  human carcinogenicity,  yet neither  can the possibility
                                                 Thus, DCM meets .
                                                human carcinogen.
      The available body of evidence on the carcinogenic mechanism
of a such a connection be entirely discounted.
the Guidelines criteria for Group B2, probable
                                108

-------
 of action of DCM and on'species differences in utilization of the
 carcinogenic metabolic pathway are not sufficient to support an
 estimate of zero cancer risk to humans.   An evaluation of the
 weight of evidence does lead to the conclusion,  however,  that
 risks should be estimated on the basis of internal dose of the
 GST metabolite (s)..  A comparison of the results  of the available
 studies indicates that the GST pathway is the most likely source
 of the excess tumorigenesis observed in the NTP  mouse bioassay.
      Additional research  on pharmacokinetic model parameters and
 on the carcinogenic mechanism of action underway -by CEFIC are
 expected to lead to refinement of the risk estimates presented in
 Chapter 8.   it should be  noted that data from the experiments may
 lead to human risk estimates below those estimated from the
 currently available data.   To the degree that the estimates  of
 relative metabolism by the GST pathway change as  a result of-
 these  data,  the  extrapolated risks  will  change.   (Method  2, which
 does not use metabolism to extrapolate across  species, would  be
 expected to  give  virtually the  same  risk estimates  as before,
 irrespective of  interspecies  differences  in metabolism that may
be discovered.)
     Using the pharraacokinetic model with its original kinetic
parameters to estimate the  internal dose of the GST metabolite*,
then following Method 1, and correcting  internal dose for
interspecies differences in sensitivity by using the surface area
correction factor, leads to a unit risk estimate for continuous
inhalation exposure to 1 ug/m3 of 4.7 x 10~7.

                               109

-------
     It would be unwise to read too much importance or
significance into changes in the unit risk of a few fold when
pharmacokinetic data are employed by either Method .1 or Method 2.
The previous chapters have outlined uncertainties in the
structure and parameter values of the model formulated by
Andersen et al.  (1986, 1987).  Although it is difficult to define
these uncertainties in quantitative terms  (such _as confidence
limits), it is clear that model projections of internal doses
could vary, perhaps by up to several fold, without contradicting
currently available model validation data.  Moreover, there are
large uncertainties as to the biological effects of those
internal doses that overshadow any  error in their  estimation.
Species differences in responsiveness—and within-species
differences in susceptibility of various tissues—are-unclear.
Perhaps the largest uncertainty lies in the question of the
relative carcinogenicity of  high and low doses, owing to the  lack
of knowledge  about the mechanism of DCM's  carcinogenic action.
It is  somewhat  ironic that the  area of risk extrapolation  that
has the least uncertainty as far as pharmacokinetics  is
concerned—relative  internal doses at  high and low exposures—
also has the  greatest uncertainty  in terms of the  degree of
carcinogenic  response that those  internal  doses can be  expected
to engender.   (Such uncertainty continues  to  be accommodated by
the use of an upper-bound,  linearized  multistage model  for low-
dose extrapolation,  which recognizes  that the true dose-response
 curve may fall off more rapidly at low doses.)

                                110

-------
      In  view of  the  uncertainties  involved,  the  changes  in  DCM's


 carcinogenic potency that  result from  different  uses  of  the


 available pharmacokinetic  information  are not, in practical '


 terms, very  distinct.   Discussion  of the issues  has been


 worthwhile because of their theoretical importance rather than


 their practical  significance  in the present  case.  For other


 compounds  (or for DCM itself, upon the introduction of new  data),


 the distinction  among extrapolation methods  may  have  much greater


 practical consequences.                               -


      Rather  than focusing  on  exactly how much the risk


 extrapolation has been  changed by  the  use of pharmacokinetic


 information,  it  is instructive to  examine how"little  it  has been


 changed.  Perhaps the most important result  of the foregoing


 analysis is  that, in the case of DCM,  pharmacokinetic
                    »

 considerations have  not revealed a great error inherent  in  using


 applied dose as  a surrogate for internal or  delivered dose.


 According to current understanding as  expressed  in the


 pharmacokinetic  model used by Andersen and Reitz, there  is  little


 difference between mice and humans in  the proportion  of  a given


 applied dose that is metabolized.  There are differences in this


 proportion from  high to low doses,  but they  are .riot- especially


 large.  Having uncovered these pharmacokinetic factors,  it  is


well to incorporate  our best understanding of them into the risk


 extrapolation process, despite remaining questions as to their


 exact magnitudes.  The uncertainty in the resulting potency


estimates is reduced (compared to the extrapolation based on




                               111

-------
r
         applied  dose)  because  the potential  for the  influence  of
        ' pharmacbkinetic factors has been markedly  circumscribed.
                                        112.

-------
                            B.5

RISK CHARACTERIZATION
                 Attribute 2  The report ckarly identifies key assumptions, their rationale,
                                and the extent of scientific consensus; the uncertainties thus
                                accepted; and the effect ofreasonabk alternative assumptions
                                on conclusions and estimates.
                     SOURCE  Case Study J. Red Dye No. 3 (Pages 87-97).
                          Note  Each critical assumption is discussed.
                     SOURCE  Case Study B. TCDD (Pages xxiv to xxvii).


                          Note  See Risk Characterization Attribute 1 in this Appendix.
                     SOURCE  Case Study H. Methylene Chloride (Pages 108-112).
                         Note  See Risk Characterization Attribute 1 in this Appendix.

-------
           A REPORT BY THE FD&C RED NO. 3  PEER  REVIEW  PANEL
AN INQUIRY INTO THE MECHANISM OF CARCINOGENIC ACTION OF FD&C RED NO.  3
               AND ITS SIGNIFICANCE FOR RISK ASSESSMENT
                                     Prepared  by:

                                     Dr.  Ronald  W.  Hart,  NCTR/FDA (Chairman)
                                     Dr.  Thomas  Burka,  NIEHS'/NIH
                                     Dr.  Stan  C.  Freni,  CEH/CDC
                                     Dr.  Robert  Furrow,  CVM/FDA
                                     Dr.  David W.  Gaylor,  NCTR/FDA .
                                     Dr.  Theodore  Meinhardt,  NIOSH/CDC
                                     Dr.  Bernard Sass,  NCI/NIH
                                     Dr.  Elizabeth K. Weisburger,  NCI/NIH
                                      Executive  Secretaries -
                                       Dr.  Paul  Lepore,  ORA/FDA
                                       Dr.  Angelo Turturro, NCTR/FDA
                                 July,  1987

-------
                                                                  July, 1987
                      CHAPTER 8 - RISK CHARACTERIZATION
A.  Introduction
     Uncertainties  are the  major  source of  problems in  conducting  a risk
assessment.    There are  uncertainties about  the quality  of  measurements,
such as  the  degree of accuracy or  sampling  error.   This  can  be dealt with
by  applying  a  variety  of statistical  tools.   Another  kind  of uncertainty
presents  itself when  interpreting  study results,  when it is  unknown what
mechanisms  have led  to  the  study  results.    For instance, how should the
effect  of gender and  species on the. tumorigenic effect  of  R-3  be  inter-
preted,  given  that it is  unknown  what  caused  this  dependency?  In some
situations,  available additional information may clear  up part  or  all  of
the uncertainties..   In other cases,  deductions  or  circumstantial evidence
from analogous studies may  be  of  help in interpreting study  results.   To
what extent  this may  occur  is  largely  determined  by the willingness and
ability  to  bridge  gaps  in  knowledge with  assumptions,  and  to  weigh the
arguments of .their  validity.
                                                     1
     Many  assumptions  have,  knowingly  or  unwittingly^ been  used  in the
preceding chapters.  There  are a number of  situations,  however,  in which
the Panel felt that there was insufficient scientific  information for for-
mulating  assumptions  with  the  confidence  that  the  balance  of  arguments
favoring  an  assumption  would  outweigh  those of  rejecting  it.    From  a
scientific viewpoint,  forcing a decision in such situation would render  it
more or less arbitrary.   It  is the  Panel's opinion  that  deciding in  a con-
troversial issue  with lack of  supporting evidence for  either  of the sides
may  be  based  on  health policy  or   regulatory  considerations,  which  is
clearly within  the  domain of  risk management.                           ,
     To allow the  reader to understand  the  extent  and  the  impact  of as-
sumptions used  by the  Panel  in the preceding  chapters,  and in harmony with
recommendations made in  a number  of  documents  (10,15),  assumptions,  con-
sidered  to  be  of crucial importance in the  decision process,  are  listed
below.     Where appropriate,  options   have  been  presented  to .assist  risk
managers in making  decisions  in issues open  to alternative interpretations.
Some of these  options may lead  to an  alternative  dose level,  and thus to  an
alternative risk  level.   Chapters  6  and 7 provide  the- tools  to assess the
risk associated with different dose  levels.    In other cases,  the  options
                                      87

-------
                                                                  July,  1987

are limited  to validity or invalidity of an assumption,  and  thus to either
acceptance or ^non-acceptance  that  there  is  some  risk to  humans.

B. Characterization of Assumptions
 1.  It is assumed that  a  tumorigenic response in any animal  species is  an
     indication of possible carcinogenicity in humans.
     Comment;   Most human carcinogens  have  shown  a  positive response  in
     animals.   Conversely, most animal  tumorigens  have  not  been shown  to
     have an effect in humans.   The latter  can be explained in a number  of
     ways:   there is  a lack  of,  or  a  deficiency  in,  human studies;  the
     tumorigenic  potency of the chemical  is  too  low to be  observed in  a
     human study of manageable size;  or  the chemical may  not be tumorigenic
     to humans.   The validity  or   invalidity  of  the  assumption  cannot  be
     judged on the basis of current  scientific evidence.

 2.  It can be concluded that  the  tumorigenic response in R-3 treated  rats
     has  been  mediated  through  or  induced by TSH.   Following  the  former
     assumption, it %is  then assumed that R-3  has  a  tumorigenic response  in
     humans  regardless   of  whether or not  TSH has  a pivotal  role  in  the
     development of tumors.
     Comment;  Implicit  in the above  statement  is the assumption that  the
     target organ in humans is necessarily  the same  as that in rats.  There
     is evidence  that  R-3  acts  on  the hypothalamus-pituitary-thyroid  axis
     of the  rat,  in which TSH  and iodine have a  central role.   Given  the
 f               *                                 L
     lack of evidence  for  an alternative  pathway,  there would  be  no  evi-
     dence that,  in humans, organs other than the thyroid may be affected..
    .. The  evidence that  TSH is  crucial to the response  in rats is not  con-
     clusive, and  neither  is the  evidence  that  excess  TSH causes  thyroid
     tumors in humans.   It is thus  possible that  thyroid  tumors may  develop
     in humans  even  if  TSH were not  essential  in the tumorigenesis.   Al-
     though  the  Panel feels  that  there is some  evidence that  R-3  is  not
     tumorigenic to humans, the  weight of evidence  is insufficient  to  rule
     out  tumorigenicity.   A decision  has therefore- to be made using  other
     considerations.

 3.  It is  assumed  that  the  tumorigenic  response  observed  at  high  dose
                                     88

-------
                                                                 July, 1987

    levels in  rats  can be extrapolated, by means  of  a mathematical model,
    to the dose levels  of human  dietary intake which are several orders of
          /•                       •                        ,
    magnitude  lower.
    Comment;   Many  mathematical  models  can  be fitted equally well  to  the
    various observed  data  points.   A number of  studies  have been,  and  are
    in the process  of  being,  done  to evaluate the use of  different  models
    (157-161).   An alternative  to  biologically-based mathematical  models
    is the use of a NOEL  with safety factors  (see assumption  6), with  the
    understanding that  the  NOEL  in  this case  does not  imply that  there is
    a threshold for effect.   It  is  presently  not  clear which approach is
    more accurate  in  defining actual risk  in  carcinogenesis, and the  de-
    cision to  use  either  approach,  or  some  other,  is a matter  of  science
    policy.

4.  It is  uncertain  whether  dose  levels  associated  with  decreased  body
    weight  and  ultramicroscopical   lesions exceeded  the  MTD.   Whether
    exceeding  the MTD in this way  would invalidate  the study  results  is
    also controversial.
    Comment:  .A decrease in body weight (in week 82,  in male rats 8.3%  and
    23.2% in female rats)  was significant for the 4%  dose level only.   In
    another semichronic rat  study (114), smaller decreases  in body  weight
    were observed  for  1%  and 2% as  well,  although  they  were  not  always
    statistically significant.   One method  to  address  this is  to disregard
    the  response  at the 4%  dose level.   One  consequence  of  disregarding
    this response  is  to remove  any  evidence  that  R-3 induced carcinomas.
    The  excess of  ultramicroscopical  lesions  was  observed  at all dose
    levels, but  the observations were  not quantitative.    Although  these
    lesions appear  R-3 related,  there   is no  information   suggesting that
    they are indicative of  an abnormal  metabolic mechanism that would  not
    have occurred at lower dose  levels.  The Panel feels that,  the issue of
    weight loss indicating  that  the MTD is  exceeded requires a risk man-
   . agement decision.   Risk  estimates   have  been  provided both with  and
    without the 4% group.

5.  In choosing the NOEL  it  is assumed  that the observation of  no  effects
    in a study of limited size can be applied  to the  entire US population.
                                    89

-------
                                                             July, 1987

Comment:  The  invalidity of  this  assumption is clear.   One  method for
addressing  this  would be to apply  a  safety factor to  account  for the
difference  in population  size.    There are  several  options open for
selecting  a NOEL,  each associated with different additional  assump-
tions or problems.  Although frequently used to indicate that there is
no  risk below a  certain threshold dose,  actually the NOEL does not
imply such  threshold.   The NOEL  (actually  the NOAEL  or  "no observed
adverse  effect level")  simply  means  that  under  the  study  conditions
"(size, duration of  exposuref and  observation, exposure level, and route
of exposure)  no adverse effect,  in this case tumors,  have been observ-
ed.   It  does  not rule out  that a  larger  study  size  or a  longer
observation period could have resulted in an observable  excess tumor
incidence.  Defining  the target  effect for a NOEL is  a difficult and
controversial issue,  because  of  the  ambiquity in  the  term "adverse
effect".  The following options  are offered:
a.  One  option is to assume that TSH is  the  mediator  of the  tumori-
    genic  effect  of  R-3 in rats  and  humans,  and thus  an increase in
    the TSH-level  in  humans can  be used as a parameter of, effect.
    Comment:  The  NOEL may be set at 20  or 60  mg  R-3/day. However, in
    part because  of the problems associated with  interpretating Study
    Wl,  the Panel cannot conclude with  certainty  on  what is the high-
    est  observed  NOEL in humans.   It  has  been discussed  in chapter 6
    and  in  assumption  #2 in this chapter  that there is evidence that
    TSH may not  be the mediator  of thyroid  carcinoma,  and,  therefore,
    tumorigenesis  in  humans.   This evidence  is  not  sufficiently con-
    clusive to reject using the  assumption of TSH mediation.  Thus, as
    stated  before, this  decision requires  science policy.   If  it is
    decided that  TSH  is NOT a mediator in  tumorigenesis  in humans, a
    NOEL  for  TSH  would  have no meaning.
b.  If  it  is  decided  that  TSH  is not a suitable  parameter  indicating
    the  tumorigenic potency of R-3 in  humans,  one may opt to choose a
    NOEL for  excess tumor incidence.
    Comment:  The Panel  has developed  two-possibilities for illustra-
    tion  of the  technique to use if this  option  is chosen,  reflecting
    the  science policy  nature  of this  issue.
       Possibility 1. Use  Study  410-002 in rats  (72),  which shows an
                                 90

-------
                                                         July, 1987

 increase  in tumor  incidence at  the  lowest dose  tested  in males,
 0.1%  (or  44 mg/kg-d).   If, traditionally,  one  divides this value
 by  safety factors of 10 for each of  the following: the difference
 in  species;  for the intraspecies  variation; and for the use of an
 effect  level;  then  the    ADI  becomes  44  mg/kg-d  /  1000  or  44
 ug/kg-d.                                              '
    Possiblity  2.   Since there  has been no long-term human study of
 R-3 induced tumors  to  derive  a NOEL, use  either the other multi-
 generational study in rats, (Study Id in  Chapter 4) which did not
 show  an excess of tumors  at the highest  dietary dose of 4% R-3,  or
 the animal  study  with the largest number of animals per dose group
 (Study  2a in  Chapter 4)  involving mice with no excess  tumors  at <
 the highest  dietary dose  of   3% R-3 in  a one-generation study.
 Using  a  NOEL  from  a rodent study requires  that it is established
 in  a  study  of  sufficiently   long  duration   and  size  (current
 standard  is  about 50 animals per dose group).
  Study Id above  was  too  short  (14 months) and  comprised  only 25
 animals  per dose  group.    On  the other hand,  this study involved
 uhree generations compared  to  the two-generation IRDC studies.  If
.one applies a  total safety factor of  100  to account  for:  a) the
 too small study size; b) the  too short duration of the study;  c)
 intraspecies variation  among   humans;  and  d)   the  possibility  of
 humans being more  sensitive  to the tumorigenic  effect  of  R-3 (if
 the action  is  not  involving TSH); then the ADI derived from this
 rat study would  be  0.04% of the  diet,  equal- to approximately 1.6
 mg/kg-d  (using food  intake and  body  weight observed  in  the IRDC
 study), or  the equivalent,  for a human,  of 100 mg/d.
  For  study  2a, in  mice,  one can  use a  total  safety factor of 100
 to  compensate  for:  a)  the study was  not multigenerational;  b)
 intraspecies variations among humans; and c) possible greater sen-
 sitivity  of humans.   The ADI would  then  be 1/100  of  3%  or (at 5
 g/d food intake  and 30 g  body weight)  50  mg/kg-d, equivalent in
 the human  to  3 g/d.   As   stated  before,  the  magnitude  of safety
 factors   and  their  application  is   considered   to  belong  in the
 domain of risk management.   It  needs mentioning that the value of
 10  for a safety  factor is  based  on  tradition.  , There would be no
                             91

-------
                                                             July,  1987.

    reason  why larger  or smaller  values may  not better  approximate
    reality.  There is  some  evidence that in each of  the  above cases,
    a value  of 10 for a  safety  factor over or under  protects  against
    the effect of the uncertainty of interest.
       The use of possibility 1  assumes that one will  not accept  bas-
    ing a NOEL on wholly inadequate studies.  Using possibility 2 can
    be done  if one feels  that Study  410-002  is  as inaccurate  as the
    negative studies.
c.  If one would opt for R-3  not being a  human  tumorigen,  a NOEL could
    then be  chosen  for any adverse, effect.   In that  case,  the rabbit
    study mentioned in Chapter 4 under C2  is the  study with the lowest
    exposure levels.   It appears that 12.5  mg/kg-d of  R-3 (equivalent
    to 760  mg/d  in humans) was  the NOEL  for  fetal toxicity.   Again,
    there is the question of  the appropriate safety factors to  use.

When reporting on R-3 studies,  it   is  assumed  that the "compound ef-
fects" are, in fact, attributable to the  halogenated  fluorescein,  R-3.
Comment;   There is sufficient evidence  that  excess  iodide  can cause
thyroid tumors  in the rat, while  its effects on TSH,  T^,  and  1^  both
in rats and  humans  have long been  known.   In ' this light,  most of the
findings of the R-3  toxicity studies  can better be explained by al-
ready known properties  of. iodide   (derived  from  deiodination  or  from
free iodide  present as  an impurity) rather than by assuming that the
halogenated  fluoroscein  is the responsible factor.  The suppression of
T^-T3 conversion  appears to  be R-3  specific, but  this  feature  has not
been adequately  studied for  iodide.  Although there  is evidence  that
iodide  rather  than R-3  may  be  the  oncogen,  there   is  insufficient
evidence  to preclude  R-3 as  the   toxic  factor.   A  decision  in  this
issue hence  requires  risk management  considerations.   It  is worthy to
mention that,  in  view  of option  c  for assumption 5, a NOEL for iodide
pertaining  to non-cancer  effects   was  established  at  0.27  mg  iodide/
kg-d from a  study involving  humans  exposed to iodide  in drinking water
(64).   This is confirmed by  a study  showing that humans  exposed  to 1
ppm iodide  in  drinking water (about 0.3 mg/kg-d)  for  over three years
did not appear to have  an increase  of T4  (64).
                                 92

-------
  7.
 8.
                                                              July, 1987

 It is assumed  that  the  dose scale for rats and  humans  is  the same,  if
 the dose is expressed in mg/kg body weight.
 Comment;  Hoel  et^ al. (157) have  suggested use of mg/m2  body  surface
 area as  an  alternative scale  on the  basis  of  the  acute  toxicity  of
 anticancer drugs in humans.  The use  of this metameter would increase
 the extrapolated (from  rats) cancer risk estimates  in  Chapter 7 by  a
 factor  of 7.   However, that  this may not  be  appropriate for  general
 use is  shown  by Crump , _ejt  al.   (158),  who  concluded  that  the NTP
 bioassays of  over  200  chemicals  gave the closest  correlation  among
 animal  species when  the mg/kg  body weight  metameter was  used.  This
 was  also  supported  by  Crouch  and  Wilson  (153),  although  there are
 qualifications  in  using  this  study.    The   Panel,  considering the
 available information, has opted  for using this metameter.

 Any  risk assessment has  to  rely  on the  accuracy of the data.   It is
 known that the interpretation of  thyroid tumor histology is difficult,
 leading  to different outcomes.   The thyroid lesions  of  the IRDC  study
 were not  an  exception.  The Panel  felt unable,  within the constraints
 of  its   charge,  to  attempt  another  round of  diagnostic  readings  by
 pathologists  other  than  the four  listed in  Tables  13-15.   Although
 there  were  no  sufficient  arguments  to  favor  one  reading   over the
 others,  risk estimates in Chapter  7  have been based  on  the  CCMA con-
 sultant's  data for  females  and .the" FDA data  for  males because  they
were the  most  complete of the,  readings.   An estimate of  the slope of
 the various  sets  of  readings would yield a range of  estimate of which
 the extremes have a  ratio  of  approximately 2.

The use  of the  linearized multistage  model  to extrapolate  risk  from
high to  low  dose does  not  imply that the  Panel assumes  that  this  par-
ticular model  is  more  appropriate than others.   As stated earlier  in
assumption #3,  various models may equally well fit the observed  data
points.    The choice of  the  multistage model,  and its  impact  on the
risk estimate, has been explained extensively in  Chapter 7. '
11.  There are  several options  for  selecting the  pathology data  for  risk
     estimation:
 9.
                                     93

-------
                                                                 July, 1987

     a.   Use  the sum of  follicular adenomas and carcinomas  in most sensi-
         tive strain/species (rats) with all doses for each sex.
     b.   Use  the sum of  follicular adenomas and carcinomas  in most, sensi-
         tive strain/species  (rat) for  each sex,  excluding  the  4%  group
         because it may have'exceeded the MTD.
     c.   Use the  combined'  incidence  of  follicular adenomas  in  male  and
         female   rats.    The  rationale  would  be   that  the  risk  is  then
         relevant  to   the  combined  population  rather  than  to  each  sex
                                                                 ' i
         individually.
     d.   Use  the incidence of all  tumors combined,  regardless  of  the histo-
         logic  type or  gender,   i.e.,  add  C-cell  and  follicular  tumor
         together.
     Comment;  Current practice does  not support option d.  The  scientific
     support   for  the  first  three options  seems   to  be  of  equal weight.
     Which of the options a-c should  prevail can,  therefore, not  be decided
     on the  basis  of  scientific arguments  alone.   Chapter 7  provides risk
     estimates for the options a and  b.   Option c was not included because
     the doses on a mg/kg basis were  different  for the two sexes.
C. Analysis
     Compared to the R-3 component  of  a risk assessment of the external us.e
of six  dyes (4),  this Report  seems much less conclusive  in  its  assessment
of the  risk to humans  from R-3.  However, it should  be appreciated that in
the  former document,  the  assumptions that  were  quantified  were,  essent-
ially,  ones involved  in  exposure  assessment  of  externally applied  R-3.
Estimation  of  exposure  is  Very much less  dependent  on assumptions,  as
available  information usually allow  making some  rough  estimates  in the
absence of more accurate  data.  However,  careful reading of the previous
document will  show that a  number of  qualitative assumptions  remained.  In
addition,  the previous document did not address  the carcinogenicity of R-3.
It was  assumed  to be  carcinogenic  based  on a science  policy decision  that
the  Panel  felt  was made  by FDA,  that a  positive effect in any adequate
chronic study indicated  that the compound was a  carcinogen.    This assump-
tion was not made in- this document.   Finally,  much  less  information was
then available  to  the Review Panel than is now.
                                      94

-------
                                                                    July,  1987

       The assumptions and the options outlined and discussed  in  this  chapter
  may seem .to be confusingly complex to the reader not very  familiar with the
  subject.  However,  it should 'be recognized that, different from the  case  of
  R-3 for cosmetic use (4),  the  Panel has now ventured  into the uncertain
  path  of separating  science  from "science policy" and  policy.   Its  charter
  offered the Panel ^sufficient  room  to conduct a  risk  assessment based  only
  on  scientific considerations.  The Panel  has  made  considerable  and  consis-
  tent  efforts  to, explicitly  indicate  where  it  feels  that decisions  in
  certain issues cannot be( made  on the basis of  scientific arguments alone.
  In  doing so,  the Panel  realizes that this  report  lacks  the simplicity  of
  the conclusions  of  the  first report on R-4 (4),  but  at the same  time it can
  be  claimed  that  this report  attempts a comprehensive  risk assessment  of a
  chemical not  complicated by  policy  considerations.   Responding to its char-
  ter,  the Panel has  also outlined in  the Addendum a number of studies   that
 may be  considered  to provide  information to fill  the gaps  in knowledge,
 thus reducing  the necessity  of  assumptions.
      The above assumptions and  options can easily be distinguished in three
 categories pertaining to  issues dealing with:   1)  the question whether or
 not R-3  is  toxic to  humans,  2)  the- magnitude  of the risk or the exposure,
 and 3)  how  to express risk.   With  respect to the first  category, the  op-
 tions  offered  to the risk manager, and  their  .impact on risk, is clear  and
 simple,  since  the choice is  between  yes or no.   If it  is  decided  that  a,
 given  assumption  is  invalid, e-.g.,  that  animal  tumors  can be extrapolated
 to humans (assumption 1), then the outcome of  the risk  assessment is  simply
 that there is  no tumorigenic risk to  humans.   The second category provides
 exposure options which  would alter  the  risk estimate  in either  direction,
 but  not  to  the extent that there is  no  risk.  If a  change in the exposure
 estimate is  made  due  to a decision  in one  of  the controversial  issues,  the
 new  exposure  estimate can  be  translated in  a new risk estimate by  direct
 proportionality.   The last  category  of  options  may  be the  most difficult
 one  offered:   whether  to  manage  the risk  through the  NOEL approach  or
 through  mathematical model  risk estimates.
     The Panel does  not express  its  preference  for  either,  but it  would
 like to  repeat that*a distinction should be maintained between the toxicity
•of iodine and  that of R-3  proper.  Integrating  all  available information,
 it  appears  that  the  preponderance  of information is  compatible with   the
                                      9.5

-------
                                                                 July,  1987

statement  that  the toxicity  of commercially  available R-3  is  largely  or
entirely  due  to  the  iodine  component, present  as  an impurity  and  from
deiodination of  R-3.    The important  feature not  yet  explained, by  iodide
alone is  effect  on the  TS - TA conversion,  which seems  an effect  of  R-3
proper.   With  the currently  available information, ascribing all  observed
toxic effects to  R-3  proper thus  seems  largely conjectural.  In  contrast,
                              / f4
reduced  body  weight  despite   creased  food  intake,  effect on TSH-T3-T4,
thyroid  tumors  and hyperplasia, and  fetotoxic effects  are all well-known
toxic properties of iodine.  Although  iodine  is a  recognized animal carcin-
ogen, a  Recommended Dietary Allowance  of  0.15  mg/d has been  issued  (161),
equal  to 50% of  the  NOEL,  and present  in 2  grams  of iodized table  salt
(which is about  the daily per  capita used of  salt in the USA),  while  the
range  of iodine  intake is  0.240  - 0.740  mg/d (64).   Another example of
iodine as  a food  additive  is  the use  o'f iodine  containing  dough conditioner
in  bread.   A dose of 1.4  mg R-3 equals an intake  of  approximately 0.01 mg
iodine,  as compared to  0.076 mg in 1  gram of  table salt.

D.  Conclusions

1.   R-3  is a  rat oncogen with equivocal  evidence of  carcinogenicity and
     with some  evidence  for causing benign thyroid tumors.

2.   The  tumorigenic effect of  R-3  on  rats is more likely to  be the result
     of an indirect (secondary) mechanism.

3.   It  is  likely, not certain,   that  the  tumorigenic  effect  of  R-3  is
     attributable to its iodine component present in part  as an  impurity.

4,   There is  insufficient evidence  to support the assumption  that R-3  is
     tumorigenic to humans, but this possibility cannot  be ruled out.

 5.   The average per capita exposure  to R-3 through food  and  internal  drugs
     is estimated at 1.41  mg/d.                      ,   »

 6.   If it is assumed that R-3 poses  a tumorigenic risk to humans, the risk
     from  ingesting R-3  containing food  and drugs  is  small,  that  is, the
                                       96

-------
 7.
                                                              July,  1987

 number  of  people with  R-3 .induced  tumors  would  be  too small  to be
 observed  by epidemiologic or other human studies.           •

 Several  uncertainties in  the risk assessment  of R-3 cannot  be solved
 on the basis of  scientific arguments  alone.   These  situations have been
 outlined,  and  options  have  been  presented  for  risk management  deci-
 sions,  m addition,  Chapter  9  lists  a  large  number of studies that may
be conducted to  -provide additional information to  clear a  part of the
uncertainties.
8-  Two options have: been offered  as  a  basis  for setting any human exposure
    level:  the  use  of  biologically-based mathematical  models or  the  NOEL
    approach.
                                    97

-------

-------
RISK CHARACTERIZATION
                 Attribute 3  The report outlines specific ongoing or potential research
                               projects that would probably clarify significantly the extent of
                               uncertainty in the risk estimation.
                    SOURCE  Case Study J. Red Dye No. 3 (Pages 98-101).


                        Note  See Hazard Identification Attribute 4 in this Append!

-------

-------
                             B.5

RISK CHARACTERIZATION
                  Attribute «l   jftg report provides a sense of perspective about the risk
                                 through the use of appropriate analogy.
                          Note  None of the case studies provided an example illustrating the use of
                                appropriate analogy.
                                                    *U.S. GOVERNMENT PRINTING OFFICE:  1990-718-159/20166

-------

-------

-------
                                                                          ""** 03

                                                                          ""* W
                                                                          75'
                                                                          = • CD

                                                                          I $
                                                                          CD
                                                                          tn
                                                                          CD
                                                                          CO
                                                                          O
                                                                          o
<
f
VD
o
u>
         r  fa?
        S *C   3" i-i
3-00
III
      2 o
      f *
        If
        (S cn
        ,-i to

        35

        5 =
        5 m
        (D O
        — m
        ?P
             5.° 3
             §s!
         C 03

         3 •<

         S3-
         ~ Q)
         3" 3
               38"
                                                                                        CD Sli
                                                                                          CD
                                                                                          O
                                                                                          5"
                                                                                          D'
                                                                                        O.3

                                                                                        0°

                                                                                        =' 3
                                                                                        3 0)
                                                                                        Q) rt-

                                                                                        — o
                                                                                        O5
                                                                                        00

-------