UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
                      WASHINGTON, O.C. 20460


                                       EPA-SAB-EHC-9Q-013
                                                     THE ADMINISTRATOR
April 23,  1990

Honorable William K. Reilly
Administrator
U.S. Environmental Protection Agency
401 M Street, s.W.
Washington, D.c. 20460

Subject: Science Advisory Board's review of the Office  of Research
and Development document Proposed Amendments_To The Guidelines  For
The. Health  Assessment Of Suspect  Developmental  Toxicants. 54  FR
9386-9403
Dear Mr. Reilly:

     On March  6,  1989,  EPA proposed amendments to the Guidelines
for  the Health  Assessment  of  Suspect  Developmental Toxicants.
These amendments expanded and clarified points made in the original
guidelines  and added new information based  on advances  in the
field.

     The Science  Advisory Board (SAB) was asked  to focus on the
major proposed amendments to the  Guidelines, and  to comment  on
other  aspects  of  the   Guidelines.    The  charge  to the  SAB's
Environmental Health Committee contained the  following elements:

   a. Assess the  technical changes to the Guidelines for  sound
      scientific  support.

   b. Review the  proposal to use broad weight-of-evidence
      categories.

   c. Review the applicability of the (RfDDT) concept of a  reference
      dose for developmental toxicity.

   d. Review  the proposed changes in  the  relationship between
      maternal and developmental toxicity.

-------
   e» Discuss alternative approaches to risk assessment  for
      developmental toxicity,

     The Environmental Health Committee met on  October 27,  1989,
in Bethesda, Maryland to  receive  briefings  from Agency officials
involved with the development of the proposed Guidelines revisions,
and to discuss in detail the issues noted above.

     With two exceptions,  the Committee considers the proposals to
be adequately founded in toxicological and teratological science,
and to reflect the current status in these fields.  There are minor
technical points which the Committee believes could be improved or
presented more  clearly.   Such  items  need attention, but do not
detract   from   the   overall   competency  of   the   proposals.
Consequently,  comments  on  these  points  have  been  supplied
separately to the Agency,  and are not addressed in the report,

     As noted,  there are two  areas in which the Committee suggests
the  need for significant  rethinking  and revision.    First,  the
weight-of-evidence classification scheme tends to be confusing vis-
a-vis current Agency and general  usage,  and still  reflects too
strongly its origins  in the  classification  of carcinogenicity as
an   unitary  endpoint,   rather  than   the  multiple   possible
developmental outcomes  of exposure to   developmental toxicants.
Functionally, it does not  provide  a more powerful conceptual basis
for risk assessment in the developmental area than now exists.  A
more  powerful  system  or scheme would  provide  a  biological,
functional basis for  assigning  priorities to  the questions  which
arise during an assessment by offering  a closer coupling between
dose and the nature of the expected outcome(s).

     The possibility  of a decision analysis-based  approach,  as
noted  in   the   report,   is  attractive,  and   is   offered  for
consideration with the understanding that considerable effort would
be required for implementation.  The Agency is advised to consider
it, along with  any  other methodology which could move towards a
more conceptually powerful, yet more economical, biologically-based
approach to developmental risk.

     The same rationale underlies the Committee's thoughts on the
subject of the RfDDT and alternative approaches to assessment.  The
traditional LQ&EL/NQ&EL process  ignores available data, is somewhat
insensitive to trends in the data, and ignores the uncertainty in
the  level   of  risk  at  the  NQAEL.    It tends to  reward  less
statistically precise  studies  by translating their  results into

-------
higher RfD levels and so provides a disincentive to better science.
The benchmark dose approach discussed  in  the  report avoids these
problems,  and  is the  subject  of a growing  body  of  literature
(including some  fine contributions by  EPA staff  scientists),   It
seems to be the next logical step  towards uniform risk assessment,
and the  Agency  is  urged to  begin moving  in this  direction  by
incorporating such an  approach in the Guidelines to be  used  in
conjunction with the current approaches.

     The  Science Advisory  Board is  pleased to  have  had  the
opportunity to review the proposed revisions to the Guidelines and
to offer  its  advice.   We would appreciate your  response  to  the
major points we have raised.
                                   Dr. Raymond Loehr, Chairman
                                   Science Advisory Board
                                   Dr. Arthur Upton,, Chairman
                                   Environmental Health Committee

-------
CD A       U.S. Environmental       Washington, DC
  ^
                Protection Agency        1PA-SAB-EHC-90-013
          Report of the Environmental
               Health Committee
       Review of Proposed Revisions to
     the Guidelines for Health Assessment
     of Suspect Developmental Toxicants
             (54 FR 9386-9403)
A SCIENCE ADVISORY BOARD REPORT                  April, 1990

-------
              U.  8.  ENVIRONMENTAL PROTECTION AGENCY

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

-------
                             ROSTER
CHAIRMAN
Dr. Arthur Upton
New York University Medical
  Center
Institute of Environmental
  Medicine

MEMSERS/CQNSUMMTTS
Dr. Gary Carlson
Department of Pharmacology
  and Toxicology
School of Pharmacy
Purdue University

Dr. David, G-aylor
National Center for
  lexicological Research
Department of Health and Human
  Services

Dr. Maureen Hatch
Department of Epidemiology
Columbia University
Dr. Marshall Johnson
Professor,    Department
Anatomy
Jefferson Medical College
of
Dr. Nancy Kim
Director, Division of
  Environmental Health
New York Department of Health

Dr. Martha Radike
Department  of Environmental
Health
Medical Center
University of Cincinnati

Dr. Stephen M. Rappaport
Department of Biomedical and
  Environmental Health Sciences
School of Public Health
University of California
Dr. Patricia Rodier
Department of Gbsteterics
  1 Gynecology
University of Rochester School
of Medicine

Dr. Bernard Schwetz
National    Institute    of
Environmental
  Health Sciences
Department of Health and Human
  Services

Dr. Bernard Weiss
Professor of Toxicology
Environmental Health  Sciences
Center
University of Rochester
  School of Medicine

Dr. Ronald Wysga
Electric    Power   Research
Institute

Executive Secretary
Mr* Samuel Rondberg
Environmental Protection Agency
401 M Street,  s.w., A-IOIF
Washington, D.c. 20460

Staff Secretary
Ms. Mary Winston
Environmental Protection Agency
401 M Street,  S.W., A-101F
Washington, D*C.  20460

Staff Director
Donald G.  Barnes
Environmental Protection Agency
401 M Street,  S.W., A-1Q1
Washington, B.C.  20460
                               11

-------
                     TABLE OF CONTENTS


1.0  Executive Summary	,,........   i
2+0  Introduction  ...... 	 ......   2
3.0  Detailed Charge ...,.,.. 	 .....   3
4.0  Detailed Findings 	 ..........   4
  4.1  Assessment of Technical Support of  Proposed
       Changes ......................   4
      4.1.1  Assumptions Regarding Adverse Effects in
             Animals . . 	 +,.......„..   4
      4.1.2  Manifestations of Developmental Toxicity  .  .   4
      4.1.3  Animal to Human Correlations  ........   5
      4,1.4  Use of the Most Sensitive Species	*   S
      4.1.5  Threshold Assumption	  .   5
  4.2. Weight of Evidence Categories » *  *	  ,   6
  4.3  Applicability of the RfDDT Concept   .	   8
  4.4  Maternal Toxicity ...,».......»..„.  11
5.0  Conclusions and Recoxtonendations ,	  11
                            1X1

-------
 1.0  Executive Summary   On March 6,  1989, EPA proposed amendments
to  the   Guidelines  for   the   Health  Assessment   of   Suspect
Developmental Toxicants (Proposed Amendments To The Guidelines For
The Health Assessment  Of Suspect Developmental Toxicants,  54 FR
9336-9403),   These  amendments expanded  and  clarified points made
in the  original guidelines  and added  new  information based on
advances in the field.

     The Science Advisory  Board was asked to focus  on  the major
proposed amendments to the Guidelines,  and  to comment on other
aspects of the  Guidelines.  More specifically,  the charge to the
Environmental Health Committee contained the following elements:
   a. Assess the technical changes to the Guidelines for sound
   scientific support.

   b. Review the proposal to use broad weight-of-evidence
   categories*

   c. Review the applicability of the (RfDDT) concept of a
   reference dose for developmental toxicity.

   d. Review the proposed changes in the relationship between
   maternal and developmental toxicity,

   e- Discuss alternative approaches to risk assessment for
   developmental toxicity.

     The Environmental Health Committee met on October 27, 1989,
in Bethesda, Maryland to  receive briefings  from Agency officials
involved with the development of the proposed Guidelines revisions,
and to discuss in detail the issues noted above.

     The Committee supports many of the proposed revisions to the
Guidelines? there are, however/ areas in which improvements could
be made.  Detailed comments on specific technical items have been
furnished to the Agency.   These items notwithstanding,  there was
a consensus that the proposed revisions were adequately founded in
developmental toxicology  and represented  a  step forward for EPA.
The  Agency  is advised to revisit  the  weight-of-evidence scheme
proposed, in order  to avoid  confusion with  more commonly applied

-------
uses  of  such  classifications,  and  to develop  a more  powerful
conceptual approach.  Further, the Agency should begin to move away
from  the  current use of  the No Observed  Adverse Effects/Lowest
Observed Adverse Effects Level  (NOAEL/LOAEL) basis for calculating
the Reference Dose, to a benchmark dose/confidence limit approach,
tied to empirical models of dose-response relationships.

2.0   Introduction   The  U.S.  EPA published  proposed and final
Guidelines  for the  Health  Assessment  of Suspect  Developmental
Toxicants in November,  1984  (49 FR 46324} and September,  1986 (51
FR 34028) , respectively.  These guidelines represented a consensus
of the scientific  community in developmental  toxicity on  how to
interpret data  in  this area.   However,  shortly  after the final
guidelines were published, it became apparent that a good deal of
new  information had become  available that  should  and  could be
incorporated into the Guidelines,  Therefore,  on March 6,  1989, EPA
proposed amendments to the Guidelines for the Health Assessment of
Suspect  Developmental  Toxicants  (Proposed  Amendments  To  The
Guidelines , For .The  Health  Assessment	0,£	.Suspect  Developmental
Toxicants, 54 FR 9386-9403).  These amendments expand and clarify
points made  in the  original guidelines  and add  new information
based on advances in the field.

     The Science Advisory Board was asked to  focus  on  the major
proposed amendments  to  the Guidelines,  in addition  to commenting
on  other aspects  of  the  Guidelines.    The  major  changes  are
summarized below:

     a*   The  original  risk assessment  guidance was  developed
     around several  basic  assumptions  that were  implicit but not
     stated in  the earlier document;  in  the  proposed amendments,
     these are clearly stated.

     b» Several consensus  workshops were held following completion
     of the 1986 guidelines and the conclusions of these workshops
     have been incorporated as revisions to the guidelines.  Areas
     affected  as  a result  of these   workshops  include  the
     relationship  of maternal  and  developmental toxicity,  the
     status of the Chernoff/Kavlock assay, and the development of
     an approach for a weight-of-evidence classification.

     c. A reference dose for developmental  toxicity (RfDpT)  is
     proposed which  is based on short-term exposure as is used in

-------
     assessing developmental toxicity potential.  This approach is
     distinguished from the RfD, which  usually applies to chronic
     or long-term exposures.

     d. The functional developmental toxicity section  has been
     expanded to reflect the Agency's recent testing guidelines
     for assessing potential developmental neurotoxicity.

     e. An expanded human studies section now gives more guidance
     on the use of human data in risk assessment*

     f. A number of other proposed minor changes are discussed in
     the  Supplementary Information  section  of  the Proposed
     Amendments,

3.0  Detailed Charge  More specifically, the Environmental Health
Committee was asked.to:

   a. Assess the technical  changes to  the Guidelines for sound
      scientific support.

   b, Review the proposal to use broad weight-of-evidence
      categories,

   c. Review the applicability of the  RfDDT for developing short-
      term reference dose estimates for developmental toxicity.

   d. Review proposed changes in the  relationship between maternal
      and developmental toxicity,

   e. Discuss alternative approaches to risK assessment for
      developmental toxicity. In particular, address alternatives
      to the National Academy of Science/National Research Council
      model,  as well  as  more  quantitative  approaches  to  risJc
      assessment than the  RfDOT approach  (It was anticipated that
      the points in  this element of  the charge,  because of their
      nature, could  best be addressed in  concert  with the other
      aspects of the charge listed above rather  than as a "stand
      alone" section,  consequently,  these issues are addressed as
      part of the discussion in section 4.3) .

     To carry out the  charge, the  Environmental  Health Committee
met on October 27, 1989, in Bethesda,  Maryland to  receive briefings

-------
from Agency officials  involved with the development of the proposed
Guidelines revisions,  and to discuss  in detail the issues noted in
the Charge.

4 • .Q_._Detailed Findings

4.1  Assessment .of Technical Support Of	Proposed Changes   It was
the consensus  of the Committee that  the changes  generally had
adequate rationale and support.  It was felt that the changes were
not only proper,  but  that they  moved the Guidelines  forward in
terms of current thinking.   Specific  comments were addressed to
five basic assumptions underlying the proposed revisions,

4.1.1   Assumptions Regarding- Adverse  Effects	in  Animals   The
revisions  state  that,   "An agent   that produces  an  adverse
developmental effect in experimental animal studies is assumed to
pose  a potential hazard  to  humans  following  exposure  during
development."    This  is  a  proper,  conservative  stance.  The
assumption  is   consistent    with   our   knowledge  of  biology,
toxicology,  and  clinical  experience  to  date.     If  there  are
exceptions,  they do  not provide  a  basis  for  ruling  out  this
position   based on  chemical structure of a specific  agent under
consideration,  or the type of response  observed in  a  specific
instance.

4.1.2  Manifestations  of Developmental Toxicity It is posited that
all   four   manifestations  of  developmental  toxicity   (death,
structural  abnormalities,  growth  alterations,   and  functional
deficits)  are of concern.  Although the relative importance of the
four  manifestations  is  not well established,  there  is general
agreement that the assumption is sound.

     More specifically,  the  importance of  structural variations is
not agreed upon by developmental toxicologists.  Some workers feel
that variations are  as important as malformations or fetal deaths?
others regard variation as  less predictive  of adverse effect in
humans than more serious manifestations of developmental toxicity.
In  addition,  we have less experience  in detecting  functional
alterations,  as  well  as less  experience  in looking  at  their
correlates in the human.  Further, the  importance of variations
is  confounded  by their  common  occurrence  in  the presence  of
maternal toxicity. Many  developmental  toxicologists consider that
a significant increase in structural variations noted only in the

-------
presence of significant maternal toxicity is a weaker signal than
the presence of major mal format ions or fetal deaths  in the absence
of maternal toxicity.   Thus the assumption stated in the proposed,
revisions  is  a  generally  accepted  one,  albeit  a  point  of
disagreement in  the field as to  the relative  importance  of the
various subsets of variation or  adverse effects noted within these
four major manifestations of developmental toxicity.

4.1.3  Animal	tp__Roman.. Correlations  The proposed revisions assume
that "..the types of developmental effects seen in animal studies
are  not  necessarily the  same  as those  that  may be produced in
humans."  This  is generally accepted by toxicologists, although the
supporting data  base is not  strong.   Further  examination of this
issue is needed,  including the generation of proper data to permit
a thorough analysis. -

4.1.4  use of  the  Most Sensitive  Species  The revised guidelines
call for using the most sensitive experimental species to estimate
human risk.   This  is  an  appropriate default position  when more
relevant data are not available, but the basic position should be
to use the most  relevant  data to estimate human  risk.   We would
like to see the  Guidelines pushing  the  field  in the direction of
trying to develop the  most relevant  data,  rather than routinely
developing data  in rats and rabbits without serious concern about
the relevancy to man or the use of other species.

4_._1_._5_	T_hreshoia__Assumption  The Guidelines assume a threshold in
the dose-response function for most developmental toxicants.  There
is general agreement that  this assumption is reasonable, but there
is a lack of consensus as to the importance of this assumption in
the risk assessment process.   For instances in which developmental
toxicity  is  already  observed  in  untreated  control  animals,
endogenous or  exogenous factors may be  sufficient to produce the
developmental toxicity.  The addition of substances which augment
these factors may produce additional developmental toxicity.  For
those  effects  which  are  caused  by  non-mutational events,  and
perhaps even  for those caused  by  mutations,  it  is important to
develop more comprehensive approaches to  risk assessment which use
the  total  data  available,  not  just  the LOAEL or  NOAEL (Lowest
Observed  Adverse Effects  Level or  No  observed  Adverse Effects
Level) .   This  concern  is relevant not only to  the Developmental
Guidelines,  but to most of EPA's endeavors in risk assessment, and

-------
has been a continuing  concern  of  the SAB1 -   It  is also addressed
below  (see  section   4.3).     Risk  assessment  procedures  for
developmental effects  should  be  based on  our  knowledge  of  the
biology of development.   The assessment of  carcinogenic  risk is
still  usually  done  with methods  based on  assumptions  for  the
mechanism  of action  that  have  been  in  use  for  decades;  in
developmental  toxicology,  where   we   know  very  little  about
mechanisms of action,  we  should be   careful not to tie ourselves
to mechanistic-based  risk assessment procedures  where potential
errors could be quite significant.

4.2. weight of Evidence Categories  The Committee believes that the
revised guidelines reflect a commendable attempt to structure and
conceptualize risk assessments  for developmental toxicants so that
they fit neatly into a regulatory framework.  They represent the
thoughtful application of toxicological  principles and insights to
a set of difficult, even volatile, issues.

     The Committee sees  two significant problems, however,  one
rooted in current usage within  the Agency, and the other rooted in
the  conceptual   approach employed   in  weight-of-evidence  (WOE)
schemes ger se.

     Assessing the completeness and quality  of  the data base for
a  specific  agent  is  an  important  part  of  the  risk  assessment
process.   The term "weight-of-evidence" as used  in  the proposed
Guidelines nay be  inappropriate,  since  many in  the field tend to
think of the  WOE as the total composite  of the information that is
available on which to make a judgement about risk, rather than to
assess the  quality  of  the  data base.   Within the  Guidelines
themselves, and  as evidenced  by the  comments  submitted  by  the
public, there is  confusion  about the term WOE  as  applied to the
data base,  compared to  the application  to a  specific chemical
agent.

     In terms of the underlying conceptual structure, this proposed
WOE scheme (like others), is encumbered by ambiguities.   Recall,
for example,  the repeated discussions within the Agency about the
proper labelling  of carcinogens,  and how to differentiate, say,
     1See the  recent SAB  Environmental  Health  Committee  letter
report on modifying and uncertainty factors  in  RfD calculations
(EPA-SAB-EHC-90-005).

-------
categories  B2  and C.   Those discussions, moreover,  are circum-
scribed  by  their  focus  on a  single endpoint—carcinogenicity—•
albeit an endpoint manifested in many ways.

     The  proposal  is  sensitive  to  the  web  of  complexities
surrounding  the WOE  classification  scheme,  and  takes  care  to
recognize its multiple facets.   Following a  fundamental tenet of
toxicology,   it  stresses the  need to  incorporate dose  into the
scheme.   The exposition stops short,  however, of offering guidance
on how to do so.  Further, it fails to provide guidance on how to
evaluate the contribution  of maternal  toxicity?  on how to define
the  dimensions  of  "adequate evidence?"  or  how  to  construct  a
coherent model  of  functional endpoints.   The exposition fails to
provide guidance on how to structure an inquiry about developmental
toxicity  that   does   not  extend  over   the  entire  realm  of
possibilities—with the concomitant potential to consmne enormous
resources-«in the application of the WOE classification scheme.

     Each of these questions suggests many others,*  we note them
here out of  the disquiet  that WOE designs  arouse.  This particular
WOE approach seems  tightly bound  to its  origins  in cancer.   We
urge EPA to  develop an approach suited to developmental toxicity
that can be implemented with available resources,

     Given the  above  comments,  we should at least  touch on some
other approaches or schemes.  If one considers the major difference
between  developmental  and  carcinogenic  risk  assessment,  the
significant factor is that in dealing with the developmental area,
many different  outcomes,  as opposed  to  one,  are possible.   For
example, the guidelines distinguish the four major manifestations
noted  above:    death,  malformations,  growth  alteration,  and
functional deficits.   Is  there  some way in which these are linked?
Does embryolethality imply the other possibilities at lower doses
or  at  other exposure  times?    Do malformations  imply  growth
alterations at lower doses?  Is  there an element of intransitivity
or a  directional bias in  these mutual relationships?   As noted
above, are they all considered  to be  of equal importance?  If not,
how should the relative importance of these endpoints be evaluated?

      An approach borrowed from decision analysis may clarify this
dilemma.    if  one  considers each of  the major  manifestations
described in the  guidelines  as a  primary category,  each primary
category would encompass a set of associated aspects.  Death, for

-------
example, might include embryolethality, stillbirths, reduced litter
size, early  neonatal mortality,  and  even premature  death after
maturity.  Malformations would include all of the indices described
in  the  guidelines.   Functional  deficits,  although  spanning an
almost  infinite  number  of measures,  could be assigned  to a much
smaller number of primary categories.   Even growth alterations may
be  expressed  in many ways?  for  example, prenatal  methylmercury
exposure revealed an influence on growth  in monkeys only when they
approached sexual maturity*

     One could then conceptualize the relationships  between the
major categories in  terms of    an  estimate of  the dose  ratio
required to  evoke  such an effect.      The  same  scheme  may be
extended to elements within each of four major outcomes, and to the
influence of maternal toxicity as well.   Is  it  reasonable to try
to  guess the  extent to which  some functional measure,  based on
preweaning  tests in  rats,  might portend the  value of  another
functional measure based on adult performance?

     One advantage of such an  approach might  be  that it helps us
to  "map" the location of important gaps in knowledge.  Such a map
might enable  us  to apply a probability  value to an un-evaluated
outcome, which despite its  grossness, might be better  than no
information at all.

     A basis for assigning priorities to questions will be required
if   regulations   and   resources  are   to   enjoy  even   modest
compatibility.   The volume of data  implied  as  necessary by the
guidelines is  so overwhelming  that it is likely  to arouse fierce
resistance or,  worse,   evasion.   The  decision analysis approach
involves new concepts, and would require further development.  Care
is  needed  so  that  this approach   also  does  not  require  an
unreasonable amount  of  data  and analysis.  However,  the  type of
scheme outlined above may offer a closer  coupling between dose and
the nature of the expected outcome(s), which, after  all,  is the
aim of risk assessment.

4_._3	applicability  of  the RfDDT_pj>_o_cjq3j:   The proposed Guidelines
substitute the use of an   RfDDT (Reference Dose  for Developmental
Toxicity),  based on  the use  of short-term or acute developmental
toxicity data, for the RfD, which is usually based on chronic data.
This substitution is appropriate, given the endpoints considered,
but the use of the RfD concept  has weaknesses that are recognized

                                8

-------
in  the  Guidelines?  specifically,  the value  of the  RfD  can  be
influenced by the characteristics of the experimental design used
to generate the toxicity data.  Hence, an  RfD based upon a study
with low statistical power  may  be (inappropriately) considerably
greater than one based upon a study with greater power.  For that
reason/   RfDs  based  upon  very   limited   studies   may  not  be
sufficiently protective. In addition, the current procedure makes
no explicit use of any trend in response to dose.

     Alternatives to this approach require some assumption about
an underlying dose-response model.  Not enough is known about the
biology  of  a  specific case  to  know  which model(s)  properly
represents that case.   This is  particularly problematic if there
is need to extrapolate response  to a  level of  risk much lower than
that  observed  in  the  experiments  used  to  generate the data.
Moreover,  it is often  supposed  that  developmental  toxicity  is
associated with a threshold  phenomenon, yet  there are no clear-cut
 models which can be used in a risk assessment.

     Both the choice of models and the existence of threshold doses
need to be addressed in the generation  of  developmental toxicity
guidelines.  The latter are discussed and used to justify the use
of the RfDDT.   Some  of its weaknesses  are acknowledged and there is
some  indication  that the weaknesses are  factored  into  the RfD
estimate  (although no  specifics  or examples  are  given).    It
generally is not recognized that  the incidence of  developmental
toxicity at the NOAEL, which is  the starting point for calculation
of the RfD, may be  as high as 6%2.   Hence, at the RfD, there exists
a non-zero baseline risk, which is independent of exposure*

     An alternative  which may overcome some  of the difficulties
with  the RfD  applications of  the above approaches  has  been
presented.  A recent paper  by Kiramel and Gaylor3 suggests the use
of a benchmark  dose associated with a response  level  of 10 percent.
The  benchmark dose,  defined by  the lower confidence level  for the
ED10  (the dose level at which  a  10  percent risk  is associated), is
estimated through use of an empirical model that makes use of all
dose-response data.   If we choose the lower  confidence  limit to the
     2Gaylor,  D.W,,  Environmental Health Perspectives 79:243-246,
1989

     3
      Risk Analysis,  8(1),  15-20,  1988

                                9

-------
ED10,  denoted  by  LED10,  then  studies with  higher  statistical
precision  generally will yield  larger values  for the  LED10 and
larger RfDs than studies with lower precision.  Hence, the greater
uncertainty  associated  with  limited  data   (and  thus  lower
statistical precision)  is factored into the benchmark dose.

     The model used to estimate the ED10 is  less critical  than when
extensive  extrapolation  is  required  because  the ten percent
response  is  likely to  be in  the range of  dose  levels used to
generate the toxicity data.  Thus, any empirical model which fits
the  data  well  is  likely to provide  a reasonable estimate of the
ED10;  in fact,   several   models could  be  applied  to  suggest how
robust the estimates of the SD10 actually are.  It is realized that
the nature of developmental toxicity data presents  specific model
problems; e.g., individual data derived from a given litter are not
independent.   Suitable  models are available  to  account  for this
lack of independence,  and they should be applied.   Since  the model
is  only used  to  extrapolate  (or interpolate)  to the  £D10,  no
assumptions about the existence (or non-existence)   of a  threshold
are needed.

     The choice of  uncertainty factors  applied  to  the LED10 could
be similar to those applied to a LOAEL.  If the data are adequate
to estimate the EDQ1,  which is closer to a threshold  dose if one
exists, the choice  of uncertainty factors applied to the I£D01 could
be similar to those applied to a NOAEL.  Another advantage of the
benchmark dose  over the current approach  is  that benchmark doses
likely  allow upper bound health risk  estimates.   As Gaylor4 has
argued, since dose-response relationships are often sub-linear in
the low dose range, decreasing the dose by an uncertainty  factor
will generally lead to a proportionately greater  reduction in risk.
Thus, if a maximum  risk  is indicated,  a dose level can be  estimated
through the benchmark/uncertainty  factor procedure such  that the
dose level is  a lower bound for dose  levels  associated  with the
risk.

     The  above approach is  not   entirely new  (in  1984,  Crump
suggested  replacing the  NOAEL with  a  benchmark  dose5),  but it
     4JOurnal  of  Toxicology and Environmental Health, 11, 329-336,
1983

     5Crump, K.S. Fundamental Appl.  Toxicology,  4:854-871,  1984.

                                10

-------
offers some advantages over the NQAEL/LQAEL approach.  We recommend
that it be considered for incorporation into the health assessment
guidelines for developmental toxicants,  to be used in conjunction
with other currently "standard" techniques.  This will facilitate
understanding of both approaches and  allow them  to be compared.
4.4  Maternal JCgxioity  One of the major points of uncertainty in
developmental toxicology today is the relationship between maternal
and developmental toxicity.  With the exception of some specific,
minor technical points  (which have been separately transmitted to
the Agency), the  Committee endorses the proposed  revisions,  and
considers them to be a good explication of the basic tenets of the
teratological literature.

     The proposed revisions do not  (and  probably  cannot)  "solve"
the question of the  relative weight or significance to be placed
on the manifestation of developmental toxicity in the presence of
observed maternal toxicity.   This  subject  elicited considerable
comment  from the  public,  and  was the source  of  considerable
discussion by the committee.  We suggest that the Agency retain the
current  statement   in  the  proposal,  i.e.,   "..when  adverse
developmental effects are  produced only at maternally toxic doses,
they are still considered to represent developmental toxicity and
should not be discounted as being  secondary to maternal toxicity,"
making only those modifications which do not weaken the thrust and
basic sense of the proposal.

5.Q  Conclusions and Recommendations  On the whole, the proposals
are adequately founded in  toxicological and teratological science,
and reflect the current status in these fields.  There are numerous
minor technical points, ranging from considerations of utilizing
human epidemiological data,  to the use of  in vitro testing (perhaps
the  technically  weakest  element  of  the  proposals)  which  the
Committee feels could  be  improved or whose  presentation could be
clarified.   Such  items need attention, but do not detract from the
overall competency  of the proposals.  Consequently,  comments on
these points have been supplied separately to the Agency,  and are
considered beyond the scope and purpose of this report.

     There are two areas in which the Committee suggests the need
for significant rethinking and  revision.  First,  the weight-of-
evidence classification scheme  tends  to be confusing vis-a-vis

                               11

-------
current Agency and general usage, and still reflects too strongly
its origins  in the  classification of unitary endpoint carcinogen
effects, rather than the manifold possible outcomes of exposure to
developmental toxicants.  Functionally,  it does not provide a more
powerful conceptual basis for risk assessment in the developmental
area than  now exists,   A more  powerful  system or  scheme would
provide a biological, functional  basis for assigning priorities to
the questions which  arise during an assessment by offering a closer
coupling between dose  and  the nature of the expected outcome(s),

     The  possibility  of a  decision  analysis-based approach  as
discussed above  is  attractive, and is  offered  for consideration
with the understanding that considerable effort would be required
for implementation.  The Agency  is advised  to  consider it, along
with  any  other  methodology which  could  move  towards  a  more
conceptually  powerful,  yet  more  economical,  biologically-based
approach to developmental risk.

     The same rationale underlies the Committee's thoughts on the
subject of the RfDDT and alternative approaches to assessment.  "The
traditional LQAEL/NQAEL process ignores available data, is somewhat
insensitive to trends  in the data, and ignores the uncertainty in
the  level  of risk  at  the  NQAEL.    It  tends  to  reward  less
statistically precise  studies by yielding higher RfD  levels and
provides a  disincentive to  better science.   The benchmark dose
approach discussed above avoids these  problems, and is the subject
of a growing body of literature (including some fine contributions
by EPA staff scientists).  It seems to be the next logical step in
assessing risk,  and  the Agency is urged to begin moving  in this
direction by  incorporating such  an approach  in the  Guidelines to
be used in conjunction with current methods.
                                12

-------