United States       Science Advisory         EPA-SAB-EHC-95-014
    Environmental        Board (1400)            May 1995
   Protection Agency	

    AN SAB REPORT:GUIDELINES
&EPA
    FOR REPRODUCTIVE

    TOXICITY RISK ASSESSMENT
    REVIEW OF THE OFFICE OF
    RESEARCH AND DEVELOPMENT'S
   GUIDELINES FOR REPRODUCTIVE
   TOXICITY RISK ASSESSMENT BY THE
   ENVIRONMENTAL HEALTH
   COMMITTEE

-------
                                 May 2, 1995

EPA-SAB-EHC-95-014

Honorable Carol M. Browner
Administrator
U.S. Environmental Protection Agency
401 M Street, S.W.
Washington, D.C.  20460

      Subject:     Science Advisory Board's review of the Draft Guidelines for
                  Reproductive Toxicity Risk Assessment (EPA/600/AP-94/001,
                  February, 1994)

Dear Ms. Browner:

      The Guidelines for Reproductive Toxicity were originally proposed in 1988 as
separate Guidelines for Assessing Male Reproductive Risk and Guidelines for Assess-
ing Female Reproductive Risk. Following public comment, the EPA's Science Advisory
Board (SAB) reviewed the proposed guidelines and in its report (EPA-SAB-EHC-89-
005) recommended several changes, including the combining of the two guidelines into
a single guideline for reproductive toxicity risk assessment. The current draft effects
this combination; in addition, the female component was expanded substantially while
retaining the original basic concepts.

      At the request of the Office of Research and Development (ORD), the SAB's
Environmental Health Committee met on July 19, 1994 to review the subject draft
Guidelines. The review addressed a series of issues developed (through discussions
between the Committee Chair, SAB Staff, and ORD Staff) constituting the formal
Charge for the activity.  This Charge included one major over-arching question: In
general, does the document reflect current scientific knowledge relevant to reproduc-
tive toxicity risk assessment?  Addressing this issue, the Committee found the overall
scientific foundations of the draft Guidelines' positions to be generally sound. This
finding notwithstanding, the Committee provided suggestions for improvement and
some specific criticisms in its report.

      The remainder of this letter summarizes each element of the Charge (the full text
of which will be found in section 2.2 of the enclosed report) and the Committee's
findings and recommendations.

-------
a)    Combining hazard identification and dose response evaluation

            The Committee does not support the combination of the hazard
            identification and dose-response evaluation, preferring the four-
            step risk assessment paradigm proposed by the 1983 National
            Research Council committee on risk assessment. In addition, the
            Committee has suggested revisions for Table 5 of the draft Guide-
            lines (the table provides a scheme for judging the available evi-
            dence on the reproductive toxicity of a particular agent).  These
            revisions include redefinition of some  of the categories of evidence
            to bring them into agreement with other portions of the draft docu-
            ment.

b)    Gender-neutral default assumption

            The Committee agrees that it is reasonable to assume that an
            agent which acts as a reproductive toxicant in one sex may also
            adversely affect reproductive function in the other sex; additional
            discussion to support this default assumption is suggested

c)    Default assumption of a threshold for non-genotoxic agents

            The Committee believes that the threshold assumption should be
            invoked only after an evaluation of the likely biological mechanism
            and mechanistic information indicates that linear responses would
            not be expected.

d)    Endocrine disrupters

            The Committee recommends that more discussion on this issue be
            incorporated in the Guidelines, noting the evidence that
            xenobiotics with estrogenic activity, as well as antiestrogens,
            androgenic, and antiandrogenic compounds, can adversely affect
            reproduction and development.  The Committee agreed that
            exposure to such chemicals is a potentially serious public health
            hazard.
e)    Need for multiple negative reproductive toxicity studies to adjudge a
      toxicant as "unlikely to pose a hazard."

            The Committee felt that the Guidelines accurately reflected the

-------
                  underlying science - it is not possible to state with confidence that
                  an agent is unlikely to constitute a hazard until one has ruled out
                  the possibility that a lack of response is due to an idiosyncratic
                  insensitivity of the species tested, or that all aspects of
                  reproduction have been comprehensively and sensitively
                  assessed.  The Committee also felt that the burden of proof should
                  lie with showing a lack of hazard, consistent with public health
                  protection.

      f)     Susceptible populations

                  Given the increasing evidence that individuals and populations
                  vary in sensitivity to  environmental toxicants, the Committee
                  recommends that EPA substantially expand the coverage of this
                  topic in the draft document.  The Committee also recommends that
                  the Guidelines require that relevant information on differential risks
                  to subsets of the population be incorporated into risk assessments
                  for environmental toxicants when possible.

      g)    Complex mixtures and exposures

                  The Committee accepts the basic substance of the draft document
                  via-a vis the discussion of these  subjects, but, suggests the
                  addition of discussion of several  exposure assessment issues to
                  strengthen  the Guidelines. The Agency should develop an overall
                  strategy to evaluate  exposures to mixtures, exposures to multiple
                  single agents, and exposures to  the same agent via multiple
                  pathways.

      We appreciate the opportunity to review this document, and  look forward to your
response to the issues we have raised.
                              Dr. Genevieve Matanoski, Chair
                              Science Advisory Board

-------
                            Dr. Frederica Perera, Chair
                            Environmental Health Committee
ENCLOSURE

-------
                                Distribution List
Administrator
Deputy Administrator
Assistant Administrators
Deputy Assistant Administrator for Pesticides and Toxic Substances
Deputy Assistant Administrator for Research and Development
Deputy Assistant Administrator for Water
EPA Regional Administrators
EPA Laboratory Directors
EPA Headquarters Library
EPA Regional Libraries
EPA Laboratory Libraries
Staff Director, Scientific Advisory Panel
Library of Congress
National Technical Information Service

-------
                                   NOTICE

This report has been written as a part of the activities of the Science Advisory Board, a
public advisory group providing extramural scientific information and advice to the
Administrator and other officials of the Environmental Protection Agency. The Board is
structured to provide balanced, expert assessment of scientific matters relating to
problems facing the Agency.  This report has not been reviewed for approval by the
Agency and, therefore, the contents of this report do not necessarily represent the views
and policies of the Environmental Protection Agency, nor of other agencies in the
Executive Branch of the Federal government, nor does mention of trade names or
commercial products constitute a recommendation for use.

-------
                          SCIENCE ADVISORY BOARD
                    ENVIRONMENTAL HEALTH COMMITTEE

                                 APRILS, 1994
CHAIR
Dr. Frederica Perera, School of Public Health, Columbia University, New York City, NY

MEMBERS
Dr. William B. Bunn, Mobile Administrative Services, Company, Inc, Princeton, NJ

Dr. Kenny S. Crump, Clement International Corp., Ruston, LA

Dr. Rogene F. Henderson, Chemistry & Biochemical Toxicology Group, Inhalation
Toxicology Research Institute, Albuquerque, NM

Dr. Richard Jackson1, National Center for Environmental Health, Center for Disease
Control, Atlanta, GA

Dr. Donald R. Mattison, Graduate School of Public Health, University of Pittsburgh,
Pittsburgh, PA

Dr. Richard Monson, Harvard School of Public Health, Boston, MA

Dr. Martha J. Radike, Dept. of Environmental Health, University of Cincinnati, Cincinnati,
OH

Dr. Ellen K. Silbergeld2, University of Maryland,  Dept. of Epidemiology, Baltimore, MD

CONSULTANTS
Dr. George P. Daston, Miami Valley Laboratories, The Proctor and Gamble Co., Ross,
OH

Dr. Elaine Faustman, Dept. of Environmental Health, University of Washington, Seattle,
WA

Dr. David P. Rail, Washington, D.C.
      1 At the time of the meeting, with the California State Department of Public Health

      2 Member, Science Advisory Board Executive Committee

-------
Dr. Knut Ringen, Center for Worker's Rights. Washington, DC

Dr. John T. Wilson, Shreveport, LA

Dr. Lauren Zeise, Reproductive and Cancer Hazard Assessment Section, California
Environmental Protection Agency, Berkeley, CA

FEDERAL EXPERT
Dr. George W. Lucier, National Institute of Environmental Health Sciences, Research
Triangle Park, NC

DESIGNATED FEDERAL OFFICIAL
Mr. Samuel Rondberg, Environmental Health Committee, Science Advisory Board
(HOOF), U.S. Environmental Protection Agency. Washington, D.C. 20460

STAFF SECRETARY
Ms. Mary L. Winston, Environmental Protection Agency, Science Advisory Board
(HOOF), Washington, D.C. 20460

-------
                                 ABSTRACT
      The Draft Guidelines for Reproductive Toxicity were reviewed by the Science
Advisory Board's Environmental Health Committee on July 19, 1994. The Committee
found the overall scientific foundations of the draft Guidelines' positions generally sound,
but provided suggestions for improvement and some specific criticisms. The following
text summarizes the Committee's findings:

a) The Committee does not support the combination of the hazard identification and
dose-response evaluation, preferring the four-step risk assessment paradigm presented
by the 1983 National Research Council committee (NRC, 1983) on risk assessment.

b) The Committee agrees that it is reasonable to assume that, in the absence of
contraindicating information, an agent which acts as a reproductive toxicant in one sex
may also adversely affect reproductive function in the other sex.  However, discussion to
support this default assumption is incomplete and should be developed more fully.

c) The Committee believes that a threshold should be assumed as a default only after
an evaluation of other possibilities. Use of the threshold assumption should occur only
after an evaluation of the likely biological mechanism and mechanistic information
indicates that linear responses would not be expected.

d) The Committee recommends that more discussion on the issue of endocrine
disrupters be incorporated in the Guidelines.  The Committee agreed that exposure to
such chemicals is a potentially serious public health hazard

e) The Committee felt that the Guidelines accurately reflected the underlying science on
the need for multiple negative reproductive toxicity studies to adjudge a toxicant as
"unlikely to pose a hazard."

f)  Given the increasing evidence that individuals and populations vary  in sensitivity to
environmental toxicants, the Committee recommends that EPA substantially expand the
coverage of susceptible sub-populations in the draft document.

h) The Committee accepts the basic substance of the draft document via-a vis the
discussion of complex  mixtures and exposures, but, suggests the addition of discussion
of several exposure assessment issues to strengthen the Guidelines.

KEYWORDS: Reproductive toxicity; environmental toxicants; guidelines; risk
assessment; endocrine disrupters; exposure thresholds
                                       IV

-------

-------
                          TABLE OF CONTENTS
1.  EXECUTIVE SUMMARY	  1

2.  INTRODUCTION	  4
      2.1  Background	  4
      2.2  Charge 	  4

3.  DETAILED FINDINGS	  7
      3.1  Combining of Hazard ID/Dose Response Evaluation	  7
            3.1.1 Reliance upon scientific judgement	  10
            3.1.2 Discussion of statistical evaluation	  10
            3.1.3 Suggested revisions to Table 5	  11
      3.2  Gender-neutral default assumption  	  11
      3.3  Threshold default assumption	  13
      3.4  Endocrine disrupters	  14
      3.5  Need for multiple negative studies	  15
      3.6  Susceptible populations	  16
      3.7  Risks from complex mixtures and exposures 	  16
            3.7.1 Specific guidance on exposure assessments for reproductive
                 toxicants	  17
      3.8 Scientific underpinnings of the guidelines	  18

4.  CONCLUSIONS 	  22
                                     VI

-------
                         1. EXECUTIVE SUMMARY

      The Guidelines for Reproductive Toxicity were originally proposed in the Federal
Register in 1988 as separate Guidelines for Assessing Male Reproductive Risk and
Guidelines for Assessing Female Reproductive Risk.  Following public comment, the
EPA's Science Advisory Board reviewed the proposed guidelines and recommended
several changes, including the combining of the two guidelines into a single guideline for
reproductive toxicity risk assessment (SAB, 1989).  A draft document effecting this
combination; and expanding the female component was reviewed by the Science
Advisory Board's Environmental Health Committee on July 19, 1994.

      The Committee found the overall scientific foundations of the  draft Guidelines'
positions generally sound, but provided suggestions for improvement and some specific
criticisms.  The following discussion summarizes the specific elements of the Charge for
this review (see section 2.2 for the full Charge) and the Committee's findings.

      a)  Combining hazard identification and dose response evaluation

      The Committee does not support the combination of the hazard identification and
      dose-response evaluation, preferring the four-step risk assessment paradigm
      presented by the 1983 National  Research Council committee  (NRC, 1983) on risk
      assessment. The Committee bases this position on three considerations:

            1)    The NRC paradigm is not restricted to non-threshold responses but
                  is equally applicable to both threshold and non-threshold responses.

            2)    Consistency in risk assessment and communication of ideas will be
                  fostered by continuing adherence to the paradigm unless there are
                  compelling reasons for a departure.

            3)    Preserving a distinction between the two steps may enhance our
                  knowledge of reproductive toxicity.

      In addition, the Committee has suggested revisions for Table  5 of the draft
      Guidelines (the table provides a scheme for judging the available evidence on the
      reproductive toxicity of a particular agent).  These revisions include a change in
      title to make it more descriptive,  and redefinition of some of the categories of
      evidence to bring it into agreement with other portions of the draft document.

-------
b)    Gender-neutral default assumption

      The Committee agrees that it is reasonable to assume that, in the absence
      of contraindicating information, an agent which acts as a reproductive
      toxicant in one sex may also adversely affect reproductive function in the
      other sex. However, discussion to support this default assumption in the
      main section of the document is incomplete and should be developed more
      fully. Also, a more detailed presentation on contraindicating information
      which would obviate the need for using the this default assumption is
      needed.

c)    Default assumption of a threshold for non-genotoxic agents

      The Committee believes that a threshold should be assumed as a default
      only after an evaluation of other possibilities.  Although some of the many
      mechanisms by which toxicants can exert  their effects may indicate
      threshold behavior,  others do not. For example, recent studies in several
      laboratories have demonstrated that the shape of the dose response curve
      cannot be predicted solely on the knowledge  that a response is
      receptor-mediated.  Since many chemicals exert their reproductive and
      developmental effects by mimicking or blocking hormone action, selection
      of a threshold default assumption without  other mechanistic or biological
      information could be inappropriate. Consequently, use of the threshold
      assumption should occur only after an evaluation of the likely biological
      mechanism and mechanistic information indicates that linear responses
      would not be expected.

d)    Endocrine disrupters

      The Committee recommends that more discussion on this issue be
      incorporated in the Guidelines, noting the  evidence that xenobiotics with
      estrogenic activity, as well as antiestrogens, androgenic, and anti-
      androgenic compounds, can adversely affect reproduction and
      development.  The Committee agreed that exposure to  such chemicals is a
      potentially serious public health hazard. The  Committee also recommends
      that:

            1)    The revised guideline document include a list of estrogen-
                  sensitive reproductive endpoints (as identified by EPA staff at

-------
                  the review meeting).
            2)    The Agency consider the use of risk assessment procedures
                  that are mechanism-specific (when data permit) for assessing
                  agents that have been identified as  acting via a hormone
                  receptor-mediated mechanism.

            3)    That measures of decreased sperm concentration/count be
                  considered as a basis for regulatory action.

e)    Need for multiple negative reproductive toxicity studies to adjudge a
      toxicant as "unlikely to pose a hazard."

      The Committee felt that the Guidelines accurately reflected the underlying
      science.  It is not possible to state with confidence that an agent is unlikely
      to constitute a  hazard until one has ruled out the  possibility that a lack of
      response is due to an idiosyncratic insensitivity of the species tested, or to
      the failure to assess comprehensively, using sensitive methods, all aspects
      of reproduction.  The Committee also felt that the burden of proof should lie
      with showing a lack of hazard, consistent with public health protection.
      Consequently,  the Committee recommends that Table 5 of the Guidelines
      explicitly state  that data from a second species is necessary to classify an
      agent as being unlikely to pose a hazard.

f)     Susceptible populations

      Given the  increasing evidence that individuals and populations vary in
      sensitivity  to environmental toxicants, the Committee recommends that
      EPA substantially expand the coverage of this topic in the draft document.
      The Committee also recommends that the Guidelines require that relevant
      information on  differential risks to subsets of the population be
      incorporated into risk assessments for environmental toxicants when
      possible.

h)    Complex  mixtures and exposures

      The Committee accepts the basic substance of the draft document via-a vis
      the discussion  of these subjects, but, suggests the addition of discussion of

-------
several exposure assessment issues to strengthen the Guidelines. The
Agency should develop an overall strategy to evaluate exposures to
mixtures, exposures to multiple single agents, and exposures to the same
agent via multiple pathways. In addition, exposures to multiple chemicals
with a common mechanism of action should be discussed.

-------
                             2. INTRODUCTION

2.1  Background

      The Guidelines for Reproductive Toxicity were originally proposed in the Federal
Register in 1988 as separate Guidelines for Assessing male Reproductive  Risk and
Guidelines for Assessing Female Reproductive Risk. Following public comment, the
EPA's Science Advisory Board reviewed the proposed guidelines and recommended
several changes, including the combining of the two guidelines into a single guideline for
reproductive toxicity risk assessment (SAB, 1989). The current draft effects this
combination; in addition, the female component was expanded substantially while
retaining the original basic concepts.

      Given the amount of reworking and updating required in these guidelines, and the
time elapsed since the original proposals, EPA decided to solicit review and comments
from interested parties before finalizing these guidelines for publication.  In addition to
the  peer review, a Federal Register notice was published on March 4,  1994 announcing
the  availability  of these guidelines for general comment.

      In general, the reaction to these proposed Guidelines has been favorable from
both the peer reviewers and the public commentors. A few substantive issues were
raised in the peer reviews and  public comments,  most of which are included in the issues
listed below.

2.2  Charge

      Following discussions between the Chair,  SAB Staff, and EPA Staff, the
Environmental  Health Committee identified the following issues on which tofocus its
review:

      a)    Is it appropriate to combine the hazard identification and dose-response
            evaluation to reflect more accurately the process used for non-cancer
            health effects? (Although this is a change from the original NAS paradigm,
            it is an approach  that EPA has been working with for some time.  This
            organization was used in the Guidelines for Developmental Toxicity Risk
            Assessment (1991)).

      b)    A default assumption that supports a "gender-neutral" approach to risk

                                       5

-------
      assessment for reproductive toxicity (see Section I). (This assumption has
      been included to deal with the frequent situation in which sufficient data are
      available on only one sex that demonstrate reproductive toxicity; in this
      case, it is assumed that the agent may also adversely affect reproductive
      function in the other sex unless sufficient mechanistic evidence is available
      to negate the assumption.)

c)     The default assumption of a threshold for non-genotoxic agents of
      reproductive toxicity (see  Section I). (The argument against this
      assumption is that a background level of impairment already exists, and
      that toxic effects can add to that impairment at any level of exposure,
      especially for agents that also are endogenous or that act to mimic or
      compete with endogenous agents.)

c)     Adequate coverage of issues involving endocrine disrupters and
      development (see Section III).

            With the recent prominence of these issues, is there adequate
            emphasis and guidance on hormonally mediated developmental
            effects and on use of data showing hormonal activity? Also, on a
            related point, is it appropriate to base regulatory action on
            decrements in sperm measures?

d)     The requirement for more than one negative reproduction studies to judge
      that an agent is "unlikely to pose a hazard" for reproductive toxicity.

            Is this requirement excessive, given the cost of conducting multi-
            generation reproduction studies? Currently, in the RfD/RfC process,
            no additional uncertainty factor is applied  if only one acceptable
            multi-generation reproduction test is available showing  no effect.

            Are the criteria for evaluating the adequacy (design, power, etc.) of
            studies sufficiently discussed?  As a scientific and/or policy matter,
            should a single valid positive study suffice to judge that an agent is
            "likely to pose a hazard," whereas a negative finding should require
            confirmation?

e)     Is there adequate treatment of susceptible populations and individuals?

-------
f)     Are the risks from complex mixtures and multiple exposures (including
      agents that may act by a similar mechanism) adequately considered?
g)    In general, does the document reflect current scientific knowledge relevant
      to reproductive toxicity risk assessment?

-------
                                    3.  DETAILED FINDINGS

         3.1 Combining of Hazard ID/Dose Response Evaluation

                The draft Guidelines for Reproductive Toxicity Risk Assessment modify the risk
         assessment paradigm3 proposed by the 1983 National Research Council (NRC)
         Committee on this topic (NRC, 1983) by combining the steps of hazard identification and
         dose-response assessment. The outcome of this combined step is described in the
         Guidelines document (page 6) as a "characterization of the health-related data as
         sufficient or insufficient to proceed with a quantitative risk assessment."   The reason
         given for this approach is that whereas the NRC paradigm was developed for assessing
         carcinogens, EPA believes the paradigm is not applicable for assessing a threshold
         response, which the Agency is assuming generally holds true for reproductive effects.
         EPA postulates that reproductive effects are not expected to occur below some threshold
         level of exposure.  Consequently, in the policy as drafted by EPA, determination of
         whether or not an agent poses a reproductive hazard (i.e., the hazard assessment)
         depends upon the threshold value and the level and pattern  of human exposure.
         Specifically, the document (page 6) states: "A hazard is defined in terms of the range of
         effective doses, routes of exposure, timing and duration of exposure and other relevant
         factors."  The threshold value would be estimated in the dose-response assessment
         step, and exposure would be evaluated in the exposure assessment step.

                The Committee does not support the combination  of the hazard identification and
         dose-response evaluation steps. The Committee wishes to make the following points
         and suggestions with respect to this issue:

                a)     The NRC paradigm is not restricted to non-threshold responses but is
                      equally applicable to both threshold and non-threshold responses.
                      Moreover, as  discussed in section 3.3 below, the Committee questioned
                      whether a threshold should be assumed for all reproductive toxicants.

                b)     Preserving a distinction between the two steps may enhance our
                      knowledge of reproductive toxicity by allowing the identification of potential
   3  The NRC paradigm consists of four steps: a) Hazard identification: The determination of whether a particular chemical is or is
not causally linked to particular health effects; b) Dose-response assessment: The determination of the relation between the
magnitude of exposure and the probability of occurrence of the health effects in question; c) Exposure assessment:  The
determination of the extent of human exposure before or after application of regulatory controls; and d) Risk characterization: The
description of the nature and often the magnitude of human risk, including attendant uncertainty.

                                                 8

-------
            hazards in the absence of sufficient data to conduct a dose-response
            analysis. Exposures that fall into this category should become high
            priority subjects for further testing and data collection, to support dose-
            response analysis.

      c)     The Committee feels that, since the NRC paradigm is widely applied and
            accepted, consistency in  risk assessment and communication of ideas will
            be fostered by continuing adherence to the four step risk assessment
            paradigm unless there are compelling reasons for a departure.  This
            paradigm has been widely used by agencies and groups involved in risk
            assessment since it was promulgated over ten years ago. The value and
            relevance of the paradigm was recently reaffirmed by the review of risk
            assessment practices conducted by the NRC  in response to directives
            incorporated in the 1990  Clean Air Act Amendments (NRC, 1994;  EPA
            1994).

      As noted, the NRC risk assessment paradigm has been widely used and has
become a standard for conduct of risk  assessments. Its use has helped to foster
understanding and communication in risk assessment.   The present draft of the
reproductive Guidelines could be made more understandable by adopting functional
separation of hazard identification and  dose-response assessment. The Committee
noted that there are several places in the guidelines where such functional separation
would help to avoid potential confusion. For example, since the title to Table 5
(CATEGORIZATION OF THE HEALTH-RELATED DATA BASE HAZARD
IDENTIFICATION/DOSE RESPONSE  EVALUATION) involves hazard
identification/dose-response evaluation, a reader familiar with the NRC paradigm would
assume that this table involves elements of both hazard identification and dose-
response assessment.  However, the  categorization does not actually involve the dose-
response evaluation step. The categorization defined by Table 5 (additional comments
on Table  5 follow below) is in reality a  part of hazard identification and should be used
to assess whether the dose-response step should be undertaken. If the guidelines
were slightly reorganized to reflect this point of view then the discussion at the bottom
of page 87 would follow much more logically and naturally.   What is now being referred
to as "completing a hazard identification/dose-response evaluation" becomes simply the
dose-response step.

      The hazard identification step, as defined  in the NRC "Red Book" is basically an
evaluation of the available data to determine if the data support continuing with the
dose response evaluation and subsequent steps in the risk assessment process.  For

-------
         potential reproductive toxicants, hazard assessment involves an evaluation of the
         evidence that an agent is a reproductive toxicant (at any dose in any species), and
         whether the data are sufficient to support a dose-response assessment. Deciding to
         proceed with a dose-response assessment should not be interpreted as implying
         necessarily that humans are at risk from any particular exposure level.

                In the hazard evaluation step, the evaluation of data should focus upon the
         consistency of available information as to the nature of any reproductive effect observed,
         and its relevance to human health. This type of evaluation is essentially qualitative in
         nature,  and may include evaluation of mechanism  of action, the animal model(s) used,
         type of effect observed, the pattern of dose-response relationships, and overall
         consistency of available data. This step is distinct from dose-response assessment, in
         which data (when available) are utilized to evaluate in quantitative terms the relationship
         between dose or exposure and severity or probability of effect.

               The Committee notes that there are some instances in which the data available
         are sufficient to identify a hazard, but insufficient to provide a credible basis for a dose-
         response analysis. That is, information may be of a qualitative, but not quantitative in
         nature.  Qualitative information may be very important: as in, for instance, the Minimata
         episode, where quantitation of the risk of methyl mercury for intrauterine development
         was undertaken long after the identification of the epidemic, and the exact dose-
         response for human methyl mercury toxicity remains in some dispute (e.g., at the recent
         international symposium on mercury,  Little Rock, November, 1994).4  There may well be
         instances where qualitative information may be appropriately used to support a
         precautionary approach, even in the absence of quantitative data sufficient to support a
         full dose-response analysis. On the other hand, if hazard identification is based on a
         reproductive response in animals at a dose that is sufficiently high to disrupt the animal's
         general physical well-being, a precautionary approach may not be warranted. If we
         insist that the two stages should be combined, then we run the risk of throwing out
         information that identifies a hazard but does not support a dose response evaluation.

                The Committee notes that only minimal  revision of the document would be
         required to separate hazard identification from dose response assessment,  Sections
         III.A, III.B, III.C, III.E, and III.G discuss hazard identification, whereas Sections III.F and
         III.H discuss dose-response assessment.
   4 The Committee notes that the Minimata incident involved developmental toxicity, not reproductive effects, but believes that the
lessons learned are relevant to the point concerning identification of hazard.

                                                 10

-------
      Pharmacokinetic considerations (III.D) are relevant to all steps of risk
assessment,  are most closely associated with dose assessment and should be included
in depth where data are available (e.g., on changes in compartments during pregnancy).
They could be addressed in a separate section or included with hazard assessment or
dose-response evaluation as appropriate. In cases where pharmacokinetic data are
included in the hazard assessment, it is important to determine that the differences in
pharmacokinetics are qualitatively absolute,  rather than quantitatively relative (e.g.,
absence of a critical metabolic pathway, rather than the differences in metabolic rate).

  3.1.1 Reliance upon scientific judgement

      Specific guidance is not provided for many of the critical decisions in risk
assessment, and the document stresses use of "scientific judgement" in these decisions.
Although scientific judgement is a critical element in all risk assessment, it is important
for the Guidelines to define those elements of judgement sufficiently to enable all users
of risk assessments to understand and evaluate the process and its relationship to
explicitly stated scientific principles.  This tendency is particularly apparent in the critical
decision regarding determination of the No Observed Adverse Effects Level (NOAEL).
The Agency should review each place in the Guidelines where scientific judgement is
called for to determine whether more explicit guidance, with robust underlying scientific
support, can be provided.

  3.1.2 Discussion of statistical evaluation

      Whereas detailed discussion is provided for various endpoints and experimental
protocols, very little guidance is presented regarding the appropriate statistical tests for
assessing reproductive toxicity. Statistical evaluation is critically important in
establishing a NOAEL. The problems encountered in determining NOAELs (with
statistical robustness and reliability) should be discussed, as should the need to provide
information about the uncertainty surrounding any NOAEL or Lowest Observed Adverse
Effects Level (LOAEL) (Gaylor, 1989).  In particular, the discussion on evaluation of
dose-response trends (page 23) could benefit from discussing specifically the use of
statistical tests of trend, including the No Statistical Significance of Trend  (NOSTASOT)
procedure (Tukey et al.,  1985), used previously by EPA in specific reference
dose/reference concentration (RfD/RfC) analyses for determining a NOAEL. Such
procedures will generally have greater statistical power than pair-wise comparisons.

      Also, for balance,  the discussion (page 23) of non-significant trends or
associations that may be biologically real should  discuss also  the likelihood that such

                                       11

-------
         trends can also occur by chance.  It could also emphasize that the likelihood that such a
         trend is a result of exposure is enhanced if similar trends are observed in a number of
         endpoints. Other statistical approaches should also be discussed.
                The Agency appropriately emphasizes evaluation of statistical power in the
         evaluation of negative studies (page 67).  However, formal statistical power calculations
         do not involve the actual data in a study and are difficult also to interpret. A posteriori
         power considerations can be better addressed through the use of confidence intervals
         for the effect rather than through formal power calculations. The following language is
         suggested as a replacement for the paragraph on page 67:

                It is important to carefully evaluate results of a negative study, including the power
                of the study, and to compare the degree of concordance or discordance between
                that study and other studies (including careful analysis of comparability in the
                details, such as strain or species used, timing, reproductive status, similarity of
                adverse endpoints, etc).  A power calculation does not reflect the observed
                outcomes of a study. Therefore, instead of making a formal power calculation, it
                may be more important to evaluate the ranges of outcomes consistent with
                various studies by calculating statistical confidence limits for the effects found in
                different studies.  Studies with lower power will tend to provide wider confidence
                intervals.  If the confidence intervals from a negative study and a positive study
                overlap, then there may be no conflict between the results of the two studies.

           3.1.3 Suggested revisions to Table 5

                Table 5 (page 89) of the draft Guidelines document provides a scheme for judging
         the available evidence on the reproductive toxicity of a particular agent.  Two broad
         categories ("Sufficient" and "Insufficient") are defined within the Table and data from all
         available studies are evaluated and used to judge whether available evidence allows a
         hazard assessment for reproductive toxicity.  As mentioned earlier, the contents of Table
         5 are not in full agreement with the text of the Guidelines document.  As a result, the
         Committee has developed a suggested revision for this Table, including a change in title
         to make it more descriptive. The proposed revision follows on the next page.5

         3.2  Gender-neutral default assumption
    The proposed revisions make the Table fairly terse; consequently the Agency may wish to add an explanatory statement or Concordance to the Guidelines . The
Concordance can be longer and can be changed periodically as new scientific information becomes available or as regulatory guidance needs change

                                                  12

-------
      It is reasonable to assume, as stated in the draft document's Overview (p. 11),
that "In the absence of information to the contrary, ... a chemical that acts as a
reproductive toxicant in one sex may also adversely affect reproductive function in the
other sex." The
                                       13

-------
                                 TABLE 5 (REVISED)
      CATEGORIZATION OF HEALTH-RELATED HAZARD IDENTIFICATION EVIDENCE
                             FOR REPRODUCTIVE TOXICITY

                         SUFFICIENT EVIDENCE FOR TOXICITY

      The Sufficient Evidence for Toxicity category includes data that collectively provide enough
information to judge that a reproductive hazard exists.  This category includes human and animal
evidence.

Sufficient Human Evidence: This category includes data from epidemiologic studies that provide
adequate evidence to judge that a causal relationship exists between exposure to an agent or
mixture and human reproductive toxicity. A case series in conjunction with strong supporting
evidence may also be used.  In addition, mechanistic information based on animal studies that is
directly to humans may be used.

Sufficient Experimental Animal Evidence/Limited Human Evidence:  This category  includes
sufficient data from experimental animal studies and/or limited data from human studies to judge
that the potential for reproductive toxicity to humans exists.  The minimum information necessary
is data from one study that demonstrate an adverse reproductive effect  in one test species.
Alternatively, minimal information for this classification  may consist  of data on humans for which a
causal interpretation  may be credible, but for which chance, bias, or confounding cannot be ruled
out with reasonable confidence.

                        INSUFFICIENT EVIDENCE FOR TOXICITY

      The Insufficient Evidence for Toxicity category includes evidence for which  there are
inadequate data in animals or in humans upon which to base a judgment of causality.

                      EVIDENCE SUGGESTING LACK OF TOXICITY

      Because lack of toxicity can never be demonstrated with certainty, exposures in this
category must always be judged as tentative.  More than one negative study is required to support
the designation of lack of toxicity

Suggestive Human Data:  Data are  available from  multiple epidemiological studies for which there
is quantitative information on exposure, credible evidence for lack of bias and confounding,
sufficient information on most reproductive outcomes, and adequate sample size to lead to a
narrow confidence interval around a rate ratio of 1.0 for all important reproductive outcomes.

Suggestive Animal Data: Data are available from at least two experimental animal studies for
which there is quantitative information on exposure, credible evidence for lack of error, sufficient
information on most reproductive outcomes, and adequate sample size to lead to a narrow
confidence interval around a rate ratio of 1.0 for all important reproductive outcomes.
                                          14

-------
level of discussion of this and the other default assumptions in the Overview section is
appropriate; however, discussion to support this assumption in the main section of the
document (e.g., in the Dose Response/Hazard Identification section (III)) is incomplete
and should be developed more fully.  In addition, clearer guidance on how default
assumptions should be addressed  in a risk characterization (Section V) is  needed.
      A fuller discussion of "information to the contrary" which would obviate the need
for making this default assumption is also needed. In this regard, the EPA may wish to
consider the two examples about which the Committee expressed concern at the review
meeting:

      a)     Gender differences in target organs or in tissue and temporal metabolic
            profiles related to potential adverse endpoints
      b)     Gender differences in susceptibility to xenobiotic-modified expression of
            isozymes (e.g., P450 and the resultant impact on the steady state
            metabolism of steroids important for gender organization of tissue and
            maintenance of reproductive status).

3.3 Threshold default assumption

      The Committee believes that use of an assumed threshold should occur only as a
default after an evaluation of other possibilities.  Reproductive and developmental
toxicants can exert their effects by a wide variety of mechanisms. While some of these
mechanisms may indicate threshold behavior; i.e., cytolethality-induced effects, other
mechanisms do not imply a threshold.  For example,  recent studies in several
laboratories have demonstrated that the shape of the dose response curve cannot be
predicted solely on the knowledge that a response is receptor-mediated (Lucier et a\.,
1993; Sewall and Lucier, 1994).  Since many chemicals  exert their reproductive and
developmental effects by mimicking or blocking hormone action (e.g. dioxin,  and
environmental  estrogens) selection of  a threshold default assumption without other
elements of mechanistic or biological information could be inappropriate.

      In addition, the level of the threshold or No Observed Effects Level (NOEL) is
dependent on the experimental design. Because the number of animals used in
experimental studies is small (typically 10-50 per group) simply increasing the sample
size will increase the chance of finding statistical significance between  groups. For
example, if a specific dose is given to a group of ten animals, three may have an  effect
compared to one out of ten in the controls. This would likely not be statistically
significant, but if 100 animals were used and 30% of the animals exhibited an effect, the
result would be significant.  In this case the NOEL would become a Lowest Observed
Effects Level (LOEL). The point is that the safety factor approach  is inherently flawed,
not that more animals should necessarily be used in experimental studies. The
                                       15

-------
"Benchmark Dose" method has been recommended to the Agency by the SAB (SAB
1990; 1993) as a possible alternative to the NOAEL/LOAEL approach .

      Lastly, a recent study that analyzed dose response patterns for genotoxic and
non-genotoxic carcinogens studied by the National Toxicology Program (315 chemicals)
has implications for the threshold assumption for reproductive and developmental
toxicants (Hoel and Portier, 1994). This analysis reported that non-genotoxic
carcinogens were somewhat more likely to exhibit linear behavior under the test
conditions than the genotoxic carcinogens. This finding was surprising and it implies
that since some non-genotoxic carcinogens can act by similar mechanisms as
reproductive and developmental toxicants (receptors,  signal transduction  pathways,
enhanced mitogenesis, enzyme activation or inhibition), use of threshold  assumption
may not be appropriate.

      There was considerable discussion of this issue by the Committee, and there was
agreement that this issue was complex and our understanding of it still evolving;
consequently methods used now may be inappropriate later. The Committee agreed
that:

      a)    Although many reproductive and developmental toxicants will exhibit
            threshold behavior, some reproductive and developmental toxicants may
            not. Examples might include certain xenobiotic hormone-blocking or
            hormone-mimicking agents, or other chemicals that are adding to an
            existing molecular  lesion (e.g., oxidative DMA damage) caused by
            background exposure to endogenous or exogenous compounds.

      b)    Use of the threshold assumption should occur only after an  evaluation of
            the likely biological mechanism and available data provides evidence that
            linear responses would not be expected. In other words, a threshold for a
            reproductive or developmental toxicant should not be automatically
            assumed.

3.4 Endocrine disrupters

      The Committee recommends that more discussion on this issue be incorporated
in the Guidelines. It noted the evidence indicating that interference with endocrine
targets, particularly  by xenobiotics with estrogenic activity, as well as antiestrogens,
androgenic, and antiandrogenic compounds, can adversely affect reproduction and
development.  Examples  include the synthetic hormone diethylstilbestrol which has been
documented to be a human developmental toxicant and the insecticide chlordecone
which has been identified as a human reproductive toxicant (Williams and Uphouse,
1991). A number of other hormonal agents have been shown to be reproductive  and/or
developmental toxicants in animal species and are potential human hazards.  The
                                      16

-------
Committee agreed that this is one of the known mechanisms of reproductive and
developmental toxicity, and
 that exposure to chemicals with hormonal activity is a potentially important public health
hazard.

      EPA Staff provided a summary of estrogen-sensitive reproductive endpoints that
are measured in a number of the reproductive toxicity screening assays described in the
draft Guidelines. The Committee recommended that the revised guideline document
include a list of these endpoints. EPA Staff also indicated that the Agency intends to
insert information regarding the effects of  estrogens on the development of the male
reproductive system into the section of the Guidelines on male endpoints, and on in vitro
methods for detecting hormonal activity (The draft Guidelines already contain
information on developmental effects in females - pp. 55, 57). The Committee supports
these changes.

      In addition to the above, the Committee had two recommendations:
      a)    Where adequate data are available, the Agency should consider the use of
            risk assessment procedures that are mechanism-specific for assessing
            agents that have been identified as acting via a hormone
            receptor-mediated mechanism.

      b)    The Committee supports consideration of the use of measures of
            decreased sperm concentration/count as a basis for regulatory action. This
            position is based on the observation that the distribution of human sperm
            counts appears to include the minimum value necessary to ensure fertility;
            for individuals with sperm counts near this minimum, any measurable
            decrement in sperm count would adversely affect their fertility.  In addition,
            this endpoint is one of the few biomarkers of effect available for
            reproductive toxicity; indications of change in this monitored endpoint
            should be closely evaluated.

3.5 Need for multiple negative studies

      Some public comments submitted on the draft Guidelines questioned the Agency
position that more than one negative study was required to constitute evidence
suggesting a lack of hazard, and suggested that it is possible to categorize a substance
as likely to pose a reproductive hazard based on a single positive study.  These
com mentors felt that the expense of a  comprehensive reproductive toxicity assessment
is sufficiently high that it is unlikely that most companies would be willing to  sponsor
more than one such study on a compound that has not produced reproductive effects in
a single well-conducted study.
                                       17

-------
      Although the Committee was sensitive to this position, it was felt that the
Guidelines, as stated, more accurately reflect the underlying science.  It is not possible
to state with confidence that an agent is likely to be without hazard until one has ruled
out the possibility that a lack of response is due to an idiosyncratic insensitivity of the
species tested, or to the failure to assess comprehensively, using sensitive methods, all
aspects of reproduction.  The Committee noted that in developmental toxicology, a lack
of effect in a single species is also considered to be insufficient to ensure lack of human
hazard; therefore, it would  be difficult to classify with confidence an agent as being
unlikely to have reproductive effect based on a single study. The Committee also felt
that the burden of proof should lie with showing a lack of hazard, consistent with public
health protection It was recommended by the Committee that Table 5
(CATEGORIZATION OF THE HEALTH-RELATED DATA BASE HAZARD
IDENTIFICATION/ EVALUATION) of the  Guidelines explicitly state that data from a
second species are necessary to classify an agent as being unlikely to pose a hazard.

      It was noted that data from a second -species generated as part of other study
designs, such as sub-chronic studies, can increase the confidence in negative results;
however, the magnitude of this confidence will depend on the endpoints measured.  It
was also  noted that the Agency does not consider a lack of multiple studies to be a data
gap of sufficient importance to warrant the imposition of an additional uncertainty factor.
The Committee agrees with these practices.

3.6 Susceptible populations

      The Committee recommends that EPA substantially expand the relevant discus-
sion in  the draft guidelines  document to summarize the available data on individual and
population sensitivity. There is increasing evidence that individuals and  populations
vary with  respect to risk from environmental toxicants (NRC, 1993; 1994; Perera etal.,
1991).  Inter-individual variability can result from variability in exposure (e.g., differences
in patterns, timing, and intensity  of exposure), as well as from host susceptibility (e.g.,
genetic, acquired, and developmental factors that may lead to heightened biologic
response to exposure).  Examples of the  latter include polymorphisms  in  genes
controlling xenobiotic metabolism and DNA repair, preexisting impairment or disease,
nutritional deficits and stage of development. Normal physiological variation can also
affect dose.  Examples include mouthbreathing affecting lung deposition, as well as
pregnancy, lactation, and the menopause being major events in women's life histories
that can affect uptake, retention, and deposition of toxic chemicals stored in fat and bone
(Mattison, Blann, and Malek, 1991).
      The Committee also recommends that the Guidelines explicitly require that, when
available,  relevant information on differential risks to subsets of the population be
                                       18

-------
incorporated into risk assessments for environmental toxicants.  Both exposure-related
and biologic factors should be considered. One approach would be to present the range
or distribution of risks across the population including children, women and minorities
where data indicate differential exposure/susceptibility.

3.7 Risks from complex mixtures and exposures

      The Committee accepts the basic substance of the draft document via-a-vis the
discussion of these subjects. There are, however, several areas where we believe
additional discussion of exposure assessment issues would strengthen the Guidelines.
In general, the Committee recommends that the Agency develop an overall strategy to
evaluate exposures to mixtures, exposures to multiple single agents, and exposures to
the same agent via multiple pathways for reproductive toxicity endpoints. In addition,
exposures to multiple chemicals with a common mechanism of action should be
discussed.  Human populations are generally exposed to complex mixtures through the
uncontrolled general environment. Animal models should be developed to evaluate
similar complex exposures under a controlled laboratory environment.  Some more
specific  comments follow.

  3.7.1  Specific guidance on exposure assessments for reproductive toxicants

      The purpose of the draft document's Section IV (Exposure Assessment) is to
provide  clear guidance on estimating human exposure for reproductive toxicants.  The
discussion should be more prescriptive and provide more specific guidance for exposure
assessments on issues such as patterns of exposure and reversibility of effect.  A partial
listing of endpoints where pattern of exposure is important to evaluate, and an indication
of the significant parameters to be considered in the evaluation  (e.g., age of individuals;
differential exposures; peak versus average exposure) would be particularly useful.
Guidance should also  be given on other exposure assessment issues that arise when
doing a  risk assessment for a suspect reproductive toxicant. For example, the document
should address ways of dealing with the frequent case where the exposure data are
insufficient to assess adequately risks for specific reproductive endpoints (e.g.,  adding
uncertainty/adjustment factors).  It should also discuss aspects of the exposure
assessment requiring emphasis in the risk characterization (such as inability to
characterize exposures of populations that may be particularly susceptible to the
reproductive endpoint  of concern).

      Much of the section is devoted to a discussion of exposure issues in interpreting
reproductive toxicity studies (e.g., top paragraph p. 97; second & third paragraph p.  98;
p. 100) and in addressing endpoints covered in the developmental toxicity guidelines
(e.g., p.  101). We suggest that these items should be moved and incorporated as
appropriate into earlier sections (e.g., Section III).
                                       19

-------
      Questions arose during the Committee discussion of "margin of exposure."
Margin of exposure (MOE) is given as "the ratio of the NOAEL from the more appropriate
or sensitive species to the estimated human exposure level from all potential sources."
Given that the NOAEL has not been adjusted to account for interspecies differences in
pharmacokinetics or study sample size (e.g., number of animals in the selected NOAEL
dose group) and that susceptibilities within the human population can be quite variable
(and generally cannot be assessed), the MOE could be misleading in certain
circumstances.  As a simple example, consider a chronic exposure where the mouse
NOAEL (in daily mg/kg) is 10 times the human dose;  i.e., the MOE is 10.  If the agent is
active in parent form and metabolically eliminated, however, the effective dose to the
human may be roughly equivalent to the mouse NOAEL and, in terms of pharmacokinetic
or "effective" dose, the MOE would be roughly one.  If there is significant variability in
susceptibility or in patterns of exposure  among people (and pattern is important), the
"effective" MOE for some individuals could potentially be significantly less than one. We
suggest that the MOE discussion should be modified to address these concerns.

 3.8 Scientific underpinnings of the guidelines

      The Committee especially endorses the separation of reproductive risk from
developmental risks.  There is ample scientific foundation to suggest that, for
physiological, pharmacological, toxicological as well as chronological reasons, the
methodological approaches for testing chemicals in animals and extrapolating those data
to humans should be managed in a substantially different way for reproductive and
developmental endpoints. However, it is also clear that many of the endpoints overlap
(i.e., could be considered manifestations of either reproductive, developmental or
mutagenic toxicity).  It would strengthen this document to indicate how overlapping
endpoints  should be managed in characterizing risk,  if they are considered
manifestations of reproductive or developmental toxicity, and which risk assessment
approaches might be considered appropriate depending upon the interpretation  of the
mechanism of toxicity producing the endpoint. In addition,  some reproductive endpoints
may overlap with endpoints evaluated for mutagenicity, and the guidelines should
suggest alternative ways of interpreting and utilizing those endpoints depending on
evidence of the structure and mechanism of action of the chemical.

      With respect to the specific default assumptions, the EHC believes that it is
reasonable to assume that an adverse reproductive effect observed in experimental
animals represents presumptive evidence of a similar potential for humans. In addition,
we support (as being reasonable positions) the default assumptions of conservation of
the site and mechanism of action of xenobiotics across species and of a threshold for the
dose response relationship (in many, but not all instances, as discussed above). Also, in
the absence of more detailed data on site and mechanism of action, the Committee
accepts the use of the most sensitive species as the most appropriate for estimating
human risk. The Committee noted, however, the case of human fertility where certain
                                       20

-------
parameters of the biological process in certain human populations have little reserve
capacity. For example, it has been proposed that any decrease in human sperm count is
associated with a considerable decrease in male fertility; this stands in contrast to
observations in laboratory rodents (Meistrich, 1984 et seq.).

      With the caveats expressed above, we agree that, in the absence of detailed data
to the contrary, it is reasonable to assume that a chemical acting in one sex may also
adversely effect reproductive function in the other sex. However, we believe that testing
in both sexes should be encouraged.

      The above notwithstanding, some comments on the use of defaults perse, are
also in order. The Committee appreciates the many problems faced by the Agency in
assessing reproductive risk due to the limits in the state of scientific knowledge.
Because of these limits, the proposed Guidelines  must rely extensively on the use  of
defaults as

discussed above. We have two general suggestions concerning defaults, which are
intended to improve both the current document and future revisions.

      a)    The current revision should provide  additional detail on the appropriate use
            of default assumptions.  The Committee encourages EPA to provide
            general  criteria for deciding when to deviate from defaults, as well as
            examples of the proper use of defaults.

      b)    The Agency should develop a specific research agenda to reduce in the
            future the reliance on default assumptions in assessing reproductive  risks.

      We also recommend that more details  on the mechanics of risk assessment for
reproductive endpoints be included. We would encourage the agency to use this
document as an opportunity to lay out some reference dose, and benchmark dose
approaches for reproductive endpoints.  Given the attention that the agency has directed
to this issue, this would be the appropriate place to summarize these methods for
consideration and discussion.

      The Committee appreciates the considerable effort that has gone  into combining
both male and  female reproductive endpoints and believes that it  has added
substantially to the  quality and utility of the document. Although this does not mean that
reproductive health effects in males or females (in the absence of a partner or child) are
not important, one of the important components of the merging of the male and female
reproductive risk assessment guidelines is that the unit of the couple becomes an
important focus.  This is also another reason why we  suggest that the risk assessment
strategies or methods need to be laid out in more  detail.  We are not aware of any other
endpoint which has a couple-based biological basis; this makes risk assessment for
                                      21

-------
reproductive endpoints unique.

      Despite the initial statement that reproduction is going to be considered
separately from development, the document confuses them throughout the text and in
the references. We would encourage the authors to go back through the document and
make certain that the discussion and references are relevant to reproductive endpoints.
For example, many of the references refer not to reproductive, but to developmental
studies; however the text in the draft suggests they refer to reproductive endpoints.

      The document should incorporate discussion of "windows of vulnerability" both
during spermatogenesis and follicular development, ovulation, and corpus luteum
formation. Different biological processes can result in differential susceptibility to
environmental  toxicants.

      One of the notable gaps in the document is a lack of information on the validation
of reproductive toxicity testing paradigms. We suggest that greater attention should be
given by the authors to testing methods or approaches that have been validated and
also

to better identifying where the lack of validation hinders or impairs the reproductive risk
assessment process.

      In discussion of additional test protocols (section III A7), the authors spend some
time talking about dominant lethal assays but do not include discussion of other
dominant assays. This is unfortunate because these other tests may be equally
informative.

      In the section describing sexual behavior (pg. 33), the authors assert that "Most
human information comes from clinical reports in which the detection of exposure-effect
associations are unlikely."  The Committee disagrees with this, given that there is a large
literature on effects on sexual behavior of a range of drugs. This literature is important
information, and the authors are missing an opportunity to draw attention both to that
literature as well as  inferences from it for sexual behavior risk assessment. In addition,
that literature allows the exploration of similarity of effect across species.

      In the section that discusses the use of sperm evaluations (III B3d), the authors
comment on a series of longitudinal study designs which have improved sensitivity.  One
of the complications of those longitudinal study designs, however, is that there is a
substantial correlation structure within the data and none of the longitudinal study
designs have used statistical techniques to account for the correlation structure in the
data.  Some attention  to that limitation in statistical techniques in this section would be
appropriate.
                                       22

-------
      The issue of correlation of the data is also important in the sections in which the
authors of the report describe approaches for collecting and using data from human
populations. Some attention needs to be given to recommendations for handling the
correlation structure in human data. This complication again points out the rationale for
specific guidance for risk assessment approaches for reproductive endpoints which
cannot be handled well in a generic risk assessment guideline document.

      On Page 72, the authors also comment that "obtaining specimens with a high
level of participation in the work force has been difficult." This points to the need to
develop more sensitive approaches to educating potential study populations, and the
Guidelines should offer suggestions about tactics that improve the cooperation of study
populations.

      Lastly, we wish to point out some small but significant issues of terminology and
the use of references.  In the document, the authors describe couple-mediated
approaches. Given that most of this material is directed toward the analysis of data
derived from animals, it may be appropriate to consider changing "couple" to "breeding
pair" when discussing animal data, but retaining "couple" when discussing human data
or effects in humans. Another point of terminology is the inappropriate or confusing use
of rates and indices. The authors need to make sure that they are using those terms
correctly and consistently throughout the document.  This is also relevant to the use of
the terms "fertility" and "fecundity." The authors would be well advised to adopt the
demographer's definition of those terms, with fecundity being the ability of a male or a
female to reproduce and fertility being the actual production of live offspring.  Those
definitions should be used consistently throughout the document.

      Throughout the document, there are places where references were omitted or are
out of date.  In the discussion on Page 31 of length of gestation, the authors suggest that
as the length of gestation increases, birth weight may be higher. This is an example of a
statement which needs to be referenced.  Lengthened gestation can result in either
lower or higher birth weights than term deliveries, and decreased birth weight puts the
infant at greater risk of adverse outcome during or following parturition (Kassis et al.,
1991; McLean etal., 1991; Goldenberg, etal., 1989; and Eden etal., 1987).

      Finally, as a general comment, we were disappointed that no academic societies
were asked to comment on the guidelines - given the number of societies that  exist in
reproductive biology and risk assessment.  It would have been appropriate to give
academic and clinical societies an opportunity to review and comment on the guidelines.
                                       23

-------
                             4. CONCLUSIONS

      The Committee found the overall scientific foundations of the draft document's
positions generally sound. Suggestions for improving the documents underpinnings
include incorporating discussions of possible "windows of vulnerability" during the
reproductive cycle or during gestation, focusing on the use of couples-based outcomes
as a measure of reproductive toxicity, and providing better descriptions of categorical
studies supporting the various default assumptions.

      The following text reiterates each specific element of the Charge for this review
(see section 2.2) and summarizes the Committee's findings.

      a) Combining hazard identification and dose response evaluation

      The Committee does not support the combination of the hazard identification and
      dose-response evaluation. In addition to the major points made below, it is
      inconsistent with the four-step risk  assessment paradigm  presented by the 1983
      National Research Council committee (NRC, 1983) on risk assessment. Since it
      was established over ten years ago, this paradigm has been widely used by
      agencies and groups involved in risk assessment.  The value and relevance of the
      paradigm was recently reaffirmed by the review of risk assessment practices
      conducted by the NRC in response to the Clean Air Act Amendments (NRC, 1994;
      EPA, 1994).

      The Committee wishes to make the following points with respect to this issue:

            1)   The NRC paradigm is not restricted to non-threshold responses but
                 is equally applicable to both threshold and non-threshold responses.

            2)   Consistency in  risk assessment and communication of ideas will be
                 fostered by continuing adherence to the paradigm unless there are
                 compelling reasons for a departure.

            3)   Preserving a distinction between the two steps may enhance our
                 knowledge of reproductive toxicity.

      b)    Gender-neutral default assumption
                                      24

-------
      The Committee agrees that it is reasonable to assume that, in the absence
      of contraindicating information, an agent which acts as a reproductive
      toxicant in one sex may also adversely affect reproductive function in the
      other sex. However, discussion to support this default assumption in the
      main section of the document (e.g., in the Dose Response/Hazard
      Identification section (III) is incomplete and should be developed more fully.

      Also, a more detailed presentation on contraindicating information which
      would obviate the need for using the this default assumption is also needed.
      In this regard, the EPA may wish to consider the two examples about which
      the Committee expressed  concern at the review meeting:

            1)    Gender differences in target organs or in tissue and temporal
                  metabolic profiles related to potential adverse endpoints

            2)    Gender differences in susceptibility to xenobiotic-modified
                  expression of isozymes

c)    Default assumption of a threshold for non-genotoxic agents

      The Committee believes that use of an assumed threshold should occur
      only as a default after an evaluation of other possibilities. Reproductive and
      developmental toxicants can exert their effects by a wide variety of
      mechanisms. Although some of these mechanisms may indicate threshold
      behavior, others do not.  Since many chemicals exert their reproductive and
      developmental effects by mimicking or blocking hormone action (e.g. dioxin,
      and environmental estrogens) selection of a threshold default assumption
      without other pieces of mechanistic or biological information could be
      inappropriate.  Consequently, use of the threshold assumption should occur
      only after an evaluation of the likely biological mechanism and mechanistic
      information indicates that linear responses would not be expected.  In other
      words, a threshold for a reproductive or developmental toxicant should not
      be automatically assumed.

d)    Endocrine disrupters

      The Committee recommends that more discussion on this issue be included
      in the Guidelines. It noted the evidence indicating that interference with
                                 25

-------
      endocrine targets, particularly by xenobiotics with estrogenic activity, as well
      as antiestrogens, androgenic, and antiandrogenic compounds, can
      adversely affect reproduction and development. The Committee agreed
      that exposure to such chemicals is a potentially serious public health
      hazard.

      The Committee also recommends that the revised guideline document
      include a list of estrogen-sensitive reproductive endpoints identified by
      EPA staff at the review meeting.  EPA Staff also indicated that the Agency
      will add information on the effects of estrogens on the development of the
      male reproductive system into the section of the Guidelines on male
      endpoints, an improvement the Committee supports.

      The Committee also recommends that the Agency consider the use of risk
      assessment procedures that are mechanism-specific (when data permit) for
      assessing agents that have been identified as acting via a hormone
      receptor-mediated mechanism; and that measures of decreased sperm
      concentration/count as be considered as a basis for regulatory action.

e)    Need for multiple negative reproductive toxicity  studies to adjudge a
      toxicant  as "unlikely to pose a hazard."

      Although  the Committee considered carefully some  of the public comments
      on this issue which disagreed with the position taken,  it was felt that the
      Guidelines more accurately reflect the underlying science. It is not
      possible to state with confidence that an agent is likely to be without hazard
      until one  has ruled out the possibility that a lack of response is due to an
      idiosyncratic insensitivity of the species tested, or that all aspects of
      reproduction have been comprehensively and sensitively assessed.  The
      Committee also felt that the burden  of proof should  lie with showing a lack
      of hazard, consistent with public health protection.  Consequently, the
      Committee recommends that Table  5 (CATEGORIZATION OF THE
      HEALTH-RELATED  DATA BASE HAZARD EVIDENCE IDENTIFICATION
      FOR REPRODUCTIVE TOXICITY)  of the Guidelines explicitly state that
      data from a second species was necessary to classify an agent as being
      unlikely to pose a hazard.

f)     Susceptible populations


                                26

-------
            There is increasing evidence that individuals and populations vary with
            respect to risk from environmental toxicants, and the Committee
            recommends that EPA substantially expand the relevant discussion in the
            draft Guidelines document to summarize the available data on individual
            and population sensitivity.  The Committee also recommends that the
            Guidelines require that relevant information on differential risks to subsets

            of the population be incorporated into risk assessments for environmental
            toxicants when possible.

      h)    Complex mixtures and exposures

            The Committee accepts the basic substance of the draft document via-a vis
            the discussion of these subjects.  However, additional discussion of several
            exposure assessment issues would strengthen the Guidelines.  The
            Committee recommends that the Agency develop an overall strategy to
            evaluate exposures to mixtures, exposures to multiple single agents, and
            exposures to the same agent via multiple pathways. In addition, exposures
            to multiple chemicals with a common mechanism of action should be
            discussed. Finally, the body of this report (section 3.7.1) provides  some
            specific suggestions in improving exposure assessment for reproductive
            toxicants.
                               REFERENCES

Eden, R., Seifert, L, Winegar, A., and W. Spellacy. 1987.  Perinatal characteristics of
      uncomplicated postdate pregnancies. Obs. and Gyn. 69(3 Pt 1):296-299.

EPA.  1994.  Response from the Science Policy Council of the US Environmental
      Protection Agency to "Science and Judgment in Risk Assessment,"  A Report by
      the National Research Council (NRC). Memorandum from Robert M. Sussman,
      Chair SPC, May 31, 1994.

Gaylor, D. 1989. Quantitative risk analysis for quantal reproductive and developmental
      effects. Environ. Health 79:243-246.

Goldenberg,  R.,  Davis., R., Cutter, G., Brumfield, C., and J. Foster.  Prematurity,
      postdates, and growth retardation: the influence of use of ultrasonography on
      reported gestational age. Am. J.  of Obs. and Gyn. 160(2):462-470.


                                      R-1

-------
Hoel, D.G., and C. Portier.  1994.  Nonlinearity of dose-response functions for
      carcinogenicity. Environ. Health Perspect. 102(Suppl 1): 109-133

Kassis, A., Mazor, M., Leiberman, J., and V. Insler.  1991.  Management of post-date
      pregnancy: a case control study. Israel Journal of Med. Science 27(2):82-8Q.

Lucier, G. W., Portier, C. J., and M. A.  Gallo. 1993.  Receptor mechanisms and
      dose-response models for the effects of dioxins.  Environ. Health Perspect.
      101(l):36-44

Mattison, D., Blann, E., and A. Malek.  1991.  Physiological alterations during pregnancy:
      impact on toxicokinetics. Fund. andAppl. Toxicol. 16:215-218

McLean, F., Boyd, M., Usher, R., and M. Kramer. 1991. Postterm infants: too big or too
      small? Am. J. ofObs. and Gyn. 164(2):619-624

Meistrich, M. L, 1989a.  Calculation of the Incidence of Infertility in Human Populations
      from Sperm Measures Using the Two-Distribution Model.  Sperm Measures and
      Reproductive Success: Institute for Health Policy Analysis, Forum on science,
      Health and Environmental Risk Assessment, pp.  275-290.

Meistrich, M. L., 1989b.  Interspecies Comparison and Quantitative Extrapolation of
      Toxicity to the Human male Reproductive System.  In: Toxicology of the Male and
      Female Reproductive Systems,  Ed. P. K  Working, Hemisphere Pub. Co., pp.
      303-321.
Meistrich, M. L and C. C. Brown. 1993. Estimation of the Increased Risk of Human
      Infertility from Alterations In Semen Characteristics.  Fertility and Sterility, Vol 40,
      No. 2 (August).

Meistrich, M. L. and M. E. A. B. Van Beek.  1990. Radiation Sensitivity of the Human
      Testis.  Advances in Radiation Biology, Vol, 14, pp. 227-268

Meistrich, M. L., Finch, M. and C. C. Lu.  1984.  Strain Differences in the Response of
      Mouse Testicular Stem Cells to Fractionated Radiation.  Radiation Research 97,
      pp. 478-487.
                                      R-2

-------
Perera, F., et al. 1991. Markers in Risk Assessment for Environmental Carcinogens.
      EMP, pp. 247-254.

NRC. 1994. Science and judgement in risk assessment, National Research Council
      Committee on Risk Assessment of Hazardous Air Pollutants, Board on
      Environmental Studies and Toxicology, National Academy Press, Washington DC.

NRC. 1993. Pesticides in the diets of infants and children.  National Research Council
      Committee on Pesticides in the  Diets of Infants and Children, Board of Agriculture
      and Board on Environmental Studies and Toxicology, Committee on Life
      Sciences, National Academy Press, Washington DC.

NRC.  1983.  Risk Assessment in the Federal Government: Managing the Process.
      National Research Council, National Academy Press, Washington, DC.

SAB. 1993. Environmental Health Committee (EHC), U.S. EPA Science Advisory
      Board.   Review of the Office of Solid Waste and Emergency Responses Draft
      Risk Assessment Guidance for Superfund Human Health Evaluation Manual.
      EPA-SAB-EHC-93-007.

SAB. 1990. Environmental Health Committee (EHC), U.S. EPA Science Advisory
      Board.  Use of Uncertainty and Modifying Factors in Establishing Reference Dose
      Levels.  EPA-SAB-EHC-90-005

SAB. 1989. Environmental Health Committee, (EHC) U.S. EPA Science Advisory
      Board.  Male and Female Reproductive Guidelines.  EPA-SAB-EHC-89-005.

Sewall, C. H.,  and G. W. Lucier.  1994. Dose-response and risk assessment
      considerations for receptor-mediated effects: case study with a TCDD hepatic
      tumor promotion model. In: Receptor-Mediated Biological Processes: Implications
      for Evaluating Carcinogenesis.  Wiley-Liss, pp. 155-171.

Tukey J., Ciminera J., and B. Heyse.  1985.  Testing the statistical certainty of a
      response to increasing doses of a drug. Biometrics 41:295-301.
Williams, J., and L. Uphouse.  1991. Vaginal cyclicity, sexual receptivity, and eating


                                     R-3

-------
behavior of the female rat following treatment with chlordecone.  Rerpro. Tox.
5:65-71
                                 R-4

-------