5570
905R85102
                          Preliminary scientific report of the
                           L'PA risk assessment guidelines for
                             carcinogenicity, mutagenicity,
                         chemical mixtures,  developmental effects
                                 and exposure assessment
                       Oh5'-.- -.-

-------
Agency

-------
            UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
                          WASHINGTON. D.C. 20460
     19, 1985

Hon. Lee M. Thomas
Administrator
U. S. Environmental Protection
  Agency                    "                                   THE AOMINISTPATOR
401 M Street S. W.
Washington, D.C.  20460

Dear Mr. Thomas:

     The Science Advisory Board has completed its preliminary scientific
review, as requested, of the EPA Risk Assessment Guidelines for Carcino-
genicity, Mutagenicity, Chemical Mixtures, Developmental Effects and
Exposure Assessment.  To carry out its review the Board's Executive
Committee formed five Risk Assessment Guidelines' Review Groups consisting
of scientists with expertise in the subject matter of each guideline.
These groups met in public session on March 4 for a preliminary planning
meeting and on April 22-23 to discuss and evaluate a number of technical
issues with EPA staff as well as members of the public.  The Executive
Conmittee prescheduled its meeting for April 25-26 to expedite its evalu-
ation of the Review Groups' draft reports and to enable you to receive
the final SAB reports in a timely fashion.  Enclosed are. preliminary
reports on each of the five guidelines.  As agreed to by EPA staff, the
chairs of each SAB Review Group will have the opportunity to see revisions
to the guidelines to determine which of the SAB suggested changes were
incorporated.  If needed, the Ghairs may distribute the revised guidelines
to all members of the Review Group.  Following this second review of the
revised guidelines, the chairmen will expeditiously prepare final reports.

     The Risk Assessment Guidelines' Review Groups addressed two major
issues in their respective reviews.  These included the adequacy of the
scientific rationale that identifies the need for guidelines and articu-
lates the conceptual framework behind their development, and the adequacy
of the Agency staff's interpretation of a number of technical issues
associated with each guideline.  The major SAB conclusions and recommend-
ations for each guideline can be summarized as follows:

     • The Carcinogenicity Guidelines Review Group advises that the
       proposed, updated guidelines for cancer are needed.  Since the
       publication of the 1976 Interin Guidelines for Carcinogenicity,.
       considerable change has occurred in both the underlying scientific
       data base and in the development of risk assessment methodologies.
       The proposed guidelines are reasonably complete in their current
       form.  Two additional issues that need to be addressed include:
       1) transplacental, multi-generational or total lifetime animal
       bioassays, and 2) sensitivity analysis of the cuantitative esti-
       mate of potency.  Regarding the Agency's interpretation of the
       scientific issues, EPA's proposed guidelines are generally
       consistent with the recently published Office of Science and
       Technology Policy principles for chemical carcinogens, and they
       address the important issues raised in the public ccmment period.

-------
              /            - 2 -

Contingent on the adoption of minor revisions outlined  in the
technical report prepared by the Review Group,  the proposed guide-
lines are scientifically adequate  for EPA staff to develop risk
assessments  in support of Agency decision making.

The Mutagenicity Guidelines Review Group unanimously concurs  that
the proposed guidelines are based  on an adequate scientific
rationale as determined by the existing state of knowledge in this
field.  The guidelines represent a successful effort in articulating
the background information necessary to develop a risk  estimation
approach for judging the potential human mutagenicity of a chemical.
Such an appoach can be .used to develop qualitative weight-of-evidence
assessments and, in-special cases, provides guidance to develop
quantitative risk assessments.  The Review Group recommends that
an eight level rank ordering of studies be used in the  weight-of-
evidence determination of qualitative risk.  The Group  also provides
the Agency with research recommendations that,  if pursued, would
supply EPA with needed and more precise tools for risk  assessment.

The Chemical Mixtures Review Group connends the EPA  work group for
developing a very good draft of rational guidelines  for health
risk assessment in this very complicated area of environmental
health.  These guidelines were published later  than  the others and
EPA staff did not have time to incorporate changes based on public
comments.  Agency staff agreed to  submit a revised draft for  the
panel's review prior to finalizing the guidelines.   The principal
scientific recommendations which the Review Group brings to your
attention include:

     1. A need for revision of a summary table  which describes the
preferred approach to risk assessment for chemical mixtures.   The
table needs to be clarified and expanded to provide  more options.
In particular, there may be conditions in which the  poor quality
of data and absence of relevant information could lead  to a "non-
definable risk" outcome.

     2. Developing and incorporating a system to express the  level
of confidence associated with various steps in  the risk assessment
process.  This might include a weight-of-evidence approach analagous
to the classification system for carcinogens developed  by the  Inter-
national Agency for Research on Cancer and would require development
of a taxonomy of chemical mixtures.  Sensitivity analysis  is also
recommended to assess the impact of various uncertainties  on  risk
estimates.

     3. Because the scientific background of health  effects of
chemical mixtures is very broad and diverse, and at  the  same time
full of critical gaps, there is a need for a separate document
which in effect would be a technical support document for  the guide-
lines.  The study of interactions of chemicals  in biological systems
is a comparatively immature science.   Progress  in improving risk
assessment will be unusually dependent upon progress  in  the science
of interactions.

-------
                                  -  3 -
     • The Developmental Effects Review Group concludes  that,  in  general,
       the proposed guidelines are scientifically adequate and  ccmmensurate
       with the present state of the science.  The field of developmental
       toxicology is, however, particularly weak with respect  to  quantitative
       assessment.  The proposed guidelines could be improved  in  the  section
       addressing the relationships of maternal toxicity in comparison
       to toxicity of the fetus.  The basis for quantitative assessment also
       needs elaboration.

     • The Exposure Assessment Review Group finds that the proposed guide-
       lines provide a very good overview of the general principles to be
       followed in conducting exposure assessments and of the use of models
       to estimate exposures in the absence of reliable  exposure  data based
       on measurements.  However, in their present state, they  fail to
       sufficiently emphasize that exposure assessments  based on  reliable
       measurements of pollutant concentrations in relevant environmental
       media should take precedence over exposure estimates based on
       unvalidated models.  In addition, they fail to provide guidance on
       the general principles to be followed in measuring pollutant
       concentrations in environmental media.  EPA should prepare a
       supplementary guideline addressing this latter issue.

     The Risk Assessment Guidelines' Review Groups also  addressed and made
recommendations on several generic issues relating to the Risk Assessment
Guidelines.  These include:

     • The proposed guidelines provide a conceptual framework that is
       generally applicable to radioactive-substances as well as  to
       chemicals.   However, it is recognized that differences exist in
       the physical and biological properties of these classes of
       pollutants as well as differences in their human health and environ-
       mental fate and effects.  The Agency should consider whether to
       develop separate guidelines for radiation that recognize these
       similarities and differences and serve as the basis for radiation
       risk assessments.

     • The development of Risk Assessment Guidelines suggests a number of
       areas for improving the scientific understanding of pollutant
       behavior and effects.   The Review Groups have recommended a number
       of research needs to the Agency's attention.   The Office of Research
       and Development should make use of these reconmendations and use
       the guidelines'  development process to enhance its research planning
       process.

     • The Science Advisory Board concurs with the Agency's recommendation
       to periodically update the guidelines as additional scientific
       knowledge may warrant.

-------
                                  - 4 -
     The Science Advisory Board recognizes the significance of  the effort
to develop Risk Assessment Guidelines and applauds the Agency for taking
this initiative,  fthen finalized, they should provide the basis  for greater
consistency in the development and utilization of risk assessments by all
the program offices and should assist the Agency's senior managers in ccm-
municating complex scientific issues to the public.  The Board appreciates
the opportunity to provide its advice on the guidelines and will provide
any additional review that is requested.  An Agency response to  the Board's
advice is requested.


                                    Sincerely,
                                    Norton Nelson, Chairman
                                    Science Advisory Board

-------
         REPORT 3Y THE SAB CARCINOGENICITY GUIDELINES  REVIEW GROUP
  INTRODUCTION

       On April 22-23, 1985, the SAB Risk Assessment Guidelines'  Review
  Group for Carcinogenicity met in public session  to review a revised draft
  of the Agency's proposed Guidelines for Carcinogen Risk Assessment.  EPA
  proposed the Guidelines in November, 1984.  Public ccmments were received,
  and a revised draft that incorporated the Agency's response to  these
  comments was rnade available to the Review Group.  The  ccmments  below
  refer to this "revised text and not to the November,  1984  draft  document.

       The Review Group advises that new guidelines for  assessing cancer
  risks are needed.  Since the publication of the  Interim Guidelines
  in 1976,! both the underlying scientific information and  risk assessment
  methods have changed considerably although basic principles remain  the same.

       The draft Guidelines are reasonably complete in their conceptual
  framework and are sound in their overall interpretation of the  scientific
  issues.  They are also generally consistent with the principles for
  chemical carcinogens issued by the Office of Science and  Technology
  Policy (OSTP),2 and address the important issues raised by the  public
  ccmments.  Contingent on the revisions outlined  below, the draft Guidelines
  should provide reasonable guidance for EPA personnel to develop scientifi-
  cally adequate based risk assessments in support of  Agency decision making.

       The Review Group identifies three kinds of  reconnendations including:
  (1) issues nissing fron the draft guidelines that should  be added;  (2)
  general advice that can be implemented by the Agency without further
  review; and (3) specific reconmendations for wording changes,   '-tost of
  the Review Group's ccmments are in the latter form and represent a
  carefully developed consensus of the reviewers.  This specificity indicates
  the extent to which only minor changes are advised.  The  Review Group
  requests that EPA respond to this report and, if necessary,  to  particularly
  indicate what advice was not used together with  a rationale for why it
  was not accepted.


                                  U.S. Environmental Protection Agency
                                  Region V, Li!"nry
                                  ?^n 'v-v!!) <>~rsfv'n Street
                                  £. •>. J U v. S ^' ^ - ' ' l f -' - ' ' '
 	        r>,;^n '!!'- '-.  :."<"' '-04
1 U.S. Environmental Protection Agency, Health Risk and Economic  Impact
  Assessments of Suspected Carcinogens: Interim Procedures and Guidelines
  Federal Register 41; (May 25, 1976), 21402-21405.

2 Office of Science and Technology Policy, Chemical Carcinogens - A Review
  of the Science and its Associated Principles - February, 1985   Federal
  Register 50: (March 14, 1985), 10371-10442.

-------
UtS. Environmental Protection Agency

-------
I.   PROPOSALS FOR ADDITIONS TO THE GUIDELINES

     The Review Group concludes that  two additional  issues  need  to  be
addressed in the guidelines before they are finalized.  "These  include:
(1) transplacental, multi-generational or  total  lifetime  animal  bioassays,
and (2) sensitivity analysis of the quantitative estimate of potency.
EPA staff should develop the wording  of the section  on  transplacental,
multi-generational or total lifetime  bioassays.  The Review Group has
provided specific language to be added for sensitivity  analysis  of  quan-
titative estimates.  (See "Recommendations for Specific Wording  Changes"
p. 29, below.)

II.  GENERAL COMMENTS

     The Review Group advises that the Agency add a  paragraph  to the
introduction that describes the general differences  between the  current
version of the guidelines and the Interim  Guidelines of 1976.  The  numbering
of the major sections also should be  revised, with the  dose-response
section as part II, the exposure section as part III, and the  risk  char-
acterization section as part TV.  The distinction between qualitative
and quantitative assessment is implicit thoughout the document,  but the
introduction should explicitly define these terms.   The references  in the
proposed guidelines to the Office of Science and Technology Policy  principles
need to define the relationship between the two documents more explicitly,
that is, the OSTP document contains a general discussion  of the  scientific
basis for risk assessment whereas the EPA  document sets cut practical
guidelines for carrying cut risk assessment.

     The panel members proceeded through the revised draft  guidelines on an
issue by issue basis, first discussing an  issue  in detail and  then  testing
the revised draft against the discussion.  The panel believes  that, in
general, the guidelines adequately define  the assumptions made in evaluating
the scientific issues that are addressed.  However,  the Agency needs to
distinguish recommendations based on scientific evidence  from  those based
on science- policy considerations.   In essence, EPA's recommendations
represent choices between competing scientific hypotheses for which the
available data do not always permit a clear decision.   The  draft guidelines
explain this distinction well in discussing the use  of  the  "linearized
multi-stage model" to develop plausible upper bound  estimates of potency.
The Review Group suggests a similiar approach for questions such as
extrapolation between routes of administration or extrapolation  between
species (body weight versus surface area).   The latter  topic should refer
to more recent information and make clear  that consideration of  more
sensitive human subpopulations forms the rationale for  selecting the most
sensitive animal species for quantification purposes.   The  recommendations
for extrapolation between routes of administration need to make  the procedure
more explicit.

     The Review Group concurs with the Agency's stated  positions on cancer
promoters.   The term "promoter" is operationally defined, and current
bioassay procedures do not permit the designation of a  chemical  as
exclusively a promoter or an initiator.

-------
                                  - 3 -


     The Agency's senior policy officials should have sufficient flexibility
to define various options in deciding whether to regulate specific.compounds.
Although the guidelines represent a logical input to establishing a policy
response, they should not delimit the series of factors, both scientific and
non-scientific, considered in regulatory decision making.  Control of exposure,
discussed in.the guidelines, seems to the Review Group to constitute only one o
a number of available options.  In cases where data were inadequate to make a
regulatory decision, the Agency could recommend further testing or exposure
evaluation on a priority basis.  (See page 6 of the revised draft.)

      In developing the section on the elements of hazard identification
the Review Group suggests that the guidelines draw the risk assessor's
attention to a series of issues that will influence the interpretation of
cancer risk.  These include:

   (1) Strength of the evidence
   (2) Promotion or co-carcinogenesis
   (3) Maximum tolerated dose
   (4) Mouse liver tumors
   (5) Benign tumors
   (6) Statistical considerations
   (7) Historical controls
   (8) Negative data
   (9) Positive data
  (10) Interract ions
  (11) Variability in human response

      Regarding the issue of less than lifetime exposure, the Review Group
reconmends that the Agency improve on the methods used in the guidelines
to develop ranges of estimates,  '/hen using the multi-stage model, reference-
to the method of Crump and Howe will be useful.  More generally, the
method of Gaylor and Kodell is suggested.  The document should describe
alternative methods to deal with less than lifetime exposure.

      In the description of the weight-of-the-evidence for epidemiological
data, biological plausibility and temporal sequence should be mentioned as
factors to be considered in the evaluation.  Throughout the document the
term "weight-of-the-evidence" should be used consistently, and reference
to apparent synonyms that may confuse, such as "degree-of-evidence" or
simply "evidence," should be deleted.   The Review Group concurs with the
overall weight-of-the evidence categories developed by the Agency, but we
advise that the Agency should:

         (1) not use letters tc represent the categories unless a descrip-
             tive paragraph on the substance is always used in conjunction;
         (2) add a table to the appendix of the guidelines showing the
             suggested combined weight-of-the-evidence category for each
             possible permutation of categories of animal and human
             weights-of-evidence; and
         (3) clarify that regulatory decisions should not be based
             on the classification system alone but will include
             consideration of factors such as potential human exposure.

     Finally, the insertions added to the revised test, including the
Review Group's recommendations, make the text awkward in places, and a final
editorial revision for readability is reccmmended.

-------
                                   -4-
III. RECOMMENDATIONS FOR SPECIFIC WORDING CHANGES

[Insert on page 1, after line 4 as a new paragraph]

     0 A major theme of the approach being advocated for risk assessment
       is the use of a weight-of-the-evidence nethodology.  This is
       because inadequacies in our knowledge of carcinogenesis preclude
       advocating a step by step prescriptive "cook-book" approach
       applicable to all suspect carcinogens.  Of necessity, the use of
       a weight-of-the-evidence approach places particular emphasis on
       the exercise of scientific judgment in reviewing and selecting
       information to be used in either the qualitative or quantitative
       elements of the carcinogen risk assessment process.  In selecting
       information for inclusion in the process, priority will be given to
       information validated through replication and peer review.  Special
       weight will be given to human data when they are available and
       adequate for purposes of risk assessment.  Generally, however,
       such data are not adequate and it is necessary to rely on experi-
       mental data.  Mien data are selected for use in a risk assessment
       frcm among multiple data sources, the basis for selection or
       exclusion of the data will be explicitly stated.

[Page 8, Section II.B.2., insert into line 2]

       ...  "that argue for or against the prediction ...,

[Page 8, Section II.B.3., last paragraph, line 4, change to]

       ...  molecular interactions,  biologically effective dose, and
       transport in, fate in, storage in and excretion frcm the body, as
       well as ...  metabolism should be discussed and critically evaluated.

[Insert on, page 9 before "6. Long  Term Animal Studies" (add the following)]

     0 The  implications of information presented in this section regarding
       shape of the dose-response relationship,  biologically effective dose,
       saturation effects, immune system effects, short-term test results,
       species differences,  etc., should be summarized so as to facilitate
       the  interpretation of data fron long-term animal studies.

[Page 9, Section II.B.4., first paragraph,  line 3,  add]
                                                                   i
     0 These- should include  interactions with other chemicals or agents
       and  with lifestyle factors,  v*here applicable.   Storage in the body
       and  possible recycling should also- be considered.

[Page 9, Section II.B.6., and forward throughout the  text,  where appropriate,
begin the  subsections of the Guidelines by stating]

     0 "This section of an assessment should include	-.	"

-------
                                  - 3 -
[Page 9,  Paragraph 3, Line 3. add:]

       ... However, lack of positive results in short tern tests for genetic
       toxicity do not, by themselves, provide a oasis for discounting
       positive results in animal bioassays.

[Page 10,  Paragraph 1, line 9, delete]

       ... "of the same "Torpholcgic type."

[Page 10,  Paragraph 1, line 11, add]

       ... malignant tumors, in most cases the evidence will be considered...

[Page 10,  last paragrapn, 1st sentence, at the end of the sentence add
 references to]

     0 (NTP,  1984) (IA?C preamble).

[Page 11,  paragraph 1, line 14, change to]

       ... high exposures alter tumor responses by mechanisms that may be ...

[Page 11,  paragrapn 1, line 15, changes to]

       ... Icwer exposures should be ...

[Page 11,  paragraph 2, line 4, change to]

       ... and implanta-jcn site sarcomas.

[Page 12,  paragraph 2, line 1, change to]

     ° To  evaluate carcinogenic!ty,  the primary comparison is tumor
       response in cosed anirals as  compared with that in contemporary
       matched control aninals.  Historic  control data are often valuaole,
       however,  and ...

[Page 12,  2nd paragrapn, end of last sentence,  add reference to]

     0 '(NTP,  1984)

[Page 12,  paragraph 2, line 3, change to]

       ...  group may oeccne questionable

[Page 12,  paragraph 2, line 10 change to]

      ...  an  unusual i:ac:
-------
                                  - 6 -
[Page 12, last paragraph, line 5, add]

     ° (Haseman, 1984)

[Page 13, replacing lines 3 & 4]

       ... classification could be changed on a case-fcy-case basis to
       "limited," if warranted by the available information on each specfic
        substance.  Factors to be taken into account in this determination
        could include; the occurrence of tuners ...

[Page 13, paragraph 1, line 8 delete]

       "showing no evidence of metastases or invasion:"

[Page 13, paragraph 1, line 9, replace with]

       ... tumors:

       [Delete]

 ___    "the occurrence of excess tumors only as a single sex.

[Page 13, paragraph 2, add at the beginning]

     0 All bioassay data are to be considered in the evaluation of carcino-
       genicity.

[Page 13, paragraph 2, line 3, delete]

      "... that are essentially identical in all other respects to a positive
       study ..."

[Page 14, Section II,  8.7., paragraph 2, line 6, change to]

       ... causes of morbidity and death, and the power to detect an effect.
       The power of studies should be included in the "assessment.

[Page 15, 2nd paragraph,  line 5.  change to]

       ..."available evidence including exposure data."

[Page 16, Add]

       ... Category F  - r.o data

[Page 17, Add at bottom]

       ...  (5)  no data

-------
                                  - 7 -



[Page 18,  Section II. D.,  paragraph 2, line 9, restore]

        Groups D and E generally would not have ...

[Page 20,  paragraph 1, line 6, change to]

      ...  uncertainty, may be useful, for example, in setting regulatory
      priorities and for evaluating ...

[Page 21,  Section III. A.I., paragraph 1,  line 3 change to]

      ...  to the scientists doing the ...

[Page 21,  Section III. A.1., paragraph 2,  line 4, add at the end]

      However, both estimates should be presented.

[Page 21,  paragraph 2, line 15,  change to]

       ... human sensitivity may be more or less than the most ...

[Page 23,  paragrapn 2, line 16:  Add]

      ...  "mechanisms of the carcincgenesis process 'are largely unknown and
           data are generally limited	

[Page 23,  paragraph 2, line 19,  change to]

      ...  the same carcinogenic  processes  ...

[Page 24,  paragraph 1, line 9, change to]

       In  the absence of information to the contrary  ... employed,  where
      appropriate, the results  of..."

[Page 24,  paragraph 2, line 1, replace first two sentences (The remainder
of  the  2nd paragraph, Beginning "In certain cases..."  becomes a new
paragraph)]

     0 It  should be emphasized that the  linearized multi-stage model leads
      to  a plausible upper limit to the risk  that is consistent with
      widely accepted nsohanisms of carcinogenesis.  However, such an
      estimate does not necessarily give a realistic prediction of the
      risk.   The true value of  the risk is uncertain,  and for many
      substances the lower bound estimate  of  risk is zero.  The Agency's
      procedures lead to  a range of risk,  defined  by the upper limit

-------
                                  - 8 -


       estimate frcm the linearized multi-stage model and a lower limit,
       which should be explicitly stated.  An established procedure that
       is applicable to a variety of substances does not yet exist for
       making most likely or best estimates of risk within the range of
       uncertainty defined by the upper and lower limit estimates.  How-
       ever, on a case-by-case basis where the data and procedures are
       available, the Agency will strive to provide most likely or best
       estimates of risk for use in risk management.  This will be most
       feasible when human data are available and when exposures are in
       the dose range of the data.

[Page 26, paragraph 1: Add]

       ... Regardless of the classification of a carcinogen, EPA should
       evaluate the available data on human and environmental exposure.

[Page 26, line 3 frcm bottom, change to]

       ... from all relevant sources of exposure including multiple
       avenues of intake frcm the same source.

[Page 26, last sentence]

       ....and modeling.  The data and methods by which doses to the target
organ are calculated should be described and critically evaluated.

[Page 27, paragraph 1, line 6, change to]

       .;. recommended as an appropriate ...

[Page 27, paragraph 2, line 6, add to the end of paragraph]

       Subpopulations with heightened susceptibility (either because of
       exposure or predisposition) should, when possible, be identified.

[Page 29, change the current 3 to 4, and add a new 3 called Host Factors.
Genetic, sex, health or age related factors known or believed to influence
susceptibility should also be taken into account.]

[Page 29, Section III.C.3., after line 8, add]

     0 In presenting quantitative estimates of risk and where sufficient
       data are available, it will be appropriate to provide the results
       of sensitivity analyses of the various selected models used toi
       calculate risk.  Such an analysis requires an explicit statement
       of the assumptions and selected parameters used in the models and
       will aid in identifying the influence of changes in the assumptions
       and parameters.  This approach clarifies those assumptions and
       parameters that are most significant in influencing the final
       risk estimate values.

-------
                                  - 9 -



[Page 32:  footnote:  delete]

      ...  life-threatening  benign tumors.

'Page 35,  line 16, change to]

       ...  used for  agent(s) with equivocal or inadequate human and
       animal  evidence  of carcinogenicity.  Inadequate animal evidence
     •  rcay result fron  animal  tests that are  technically flawed or when
       there is a single  negative animal bioassay.

[Page 35,  line 18, change to]

     Group E - No Positive  Evidence of ...

     Group F - No data  available

-------
          REPORT OF THE SAB MUTAGENICITY GUIDELINES REVIEW GROUP
         Review Group found EPA's proposed Guidelines for Mutagenicity Risk
Assessment to be well researched and commends the staff who prepared them
for toing a thorough job in presenting the scientific issues associated
witr. assessing the risk of environmental mutagens, discussing the approaches
for qualitative and quantitative risk estimation and in addressing the
r.ajcr areas of scientific concern and uncertainty.

     Although it was not a major area of disagreement, the Review Group
has prepared, for Agency consideration, a rank ordering of studies to be
used in the veight-of-evidence determination of qualitative risk.  The
proposed ranking, in numerical order, deals with test attributes that EPA
sr.culd use to define the levels of evidence relating to human mutagenicity
induced by a chemcial(s).  These eight ranking levels include:

     1.  Positive-data derived from human germ-cell mutagenicity studies,
         xhen available, will constitute the highest level of evidence
         for hunan mutagenicity.

     2.  Valid positive results from studies on heritable mutational
         events (of any kind) in mammalian germ cells.

     3.  Valid positive results from mammalian germ-cell chromosome
         aberration studies that do not include an intergeneration test.

     4.  Sufficient evidence for a chemical's interaction with mammalian
         germ cells, together with valid positive mutagenicity test results
         from two assay systems, at least one of which is mammalian (in
         vitro or in vivo).  The positive results may both be for gene
         mutations or ooth for chromosome aberrations:  if one is for
         gene mutaticrs and the other for chromosome aberrations, both
         must be fron mammalian systems.

     5.  Suggestive evidence for a chemical's interaction with mammalian
         germ cells, together with valid positive mutagenicity evidence
         from two assay systems as described under 4 above.   Alternatively,
         positive nutagenicity evidence of less strength than defined ,
         under 4 above, -^hen combined with sufficient evidence for a
         chemical's interaction with mammalian germ cells.

     6.  Positive "njtagenicity test results of less strength than defined
         under 4 above, combined with sugestive evidence for a chemical's
         interaction with mammalian germ cells.

     7.  Although definitive proof of nonmutagenicity is not possible,  a
         chemical could be classified operationally as  not a human germ-cell
         mutagen, if it gives valid negative .test results for all endpoints
         of concern that are enumerated in the guidelines.

     3.  Inadequate evidence bearing on either mutagenicity or chemical
         interaction with nammalian germ cells.

-------
     In addition to this ranking system for evaluating evidence for
mutugenicity, the Review Group made several reconrnendations relating to
other aspects of the guidelines.  These include:

     •   EPA should include additional references within the guidelines
pertinent to established and accepted genetic-risk estimation procedures
that specifically deal with the characterization of the human disease
burden introduced by an increase in mutation.

     •   The Agency should modify the section on testing systems to more
specifically detail existing mammalian.germ cell tests and provide a
brief discussion of their strengths and weaknesses and the limited avail-
ability of data for certain tests.

     •   The quantitative risk section should place greater emphasis on
data derived from exposure, exposure rate and or other regimens that most
closely simulate the human experience.

     The Review Group recognizes a number of research areas that, if
pursued diligently, could provide the Agency with needed tools to assess
human genetic risk more adequately.  Although not exhaustive, the following
issues deserve attention:

     1.  More experimental work on mammalian oocyte response to mutagens,
         in a variety of experimental mammalian species.

     2.  Enhancing the attempts to correlate mammalian germ and somatic
         cell mutagenic data with equivalent human somatic cell data
         in an effort to predict induced human germ risk.  These studies
         should utilize some well-defined mutagens.

     3.  Further studies employing dominant mutant phenotypes (e.g.,
         skeletal and cataract) that could be used in both sexes and
         between different mammalian species.  Such studies would improve
         the confidence in interspecies extrapolation to man.  Moreover,
         additional work needs to be done to characterize the mutation
         component of these qenetic endpoints.

     4.  Continued development of aneuploidy technology in germ cells of
         both sexes and particularly in mammals with an aim toward
         improving quantitative risk estimation.

     5.  Studies that will better define the baseline incidence of hunan
         genetic disease and the mutational components associated with the
         various subcategories of disease determined by various genetic
         mechanisms.

-------
                                  - 3 -
     6.  More experimental studies on varying exposure regimes and rates
         of exposure are needed in eukargate test systems in order to
         provide needed models for extrapolation to low doses.

     The Review Group concludes that the scientific adequacy of the proposed
Guidelines for Mutagenicity Risk Assessment accurately reflect the state
of the art of this field.  We have presented the Agency staff with a
series of minor changes in the document that we believe will improve the
clarity and general intent of the guidelines.  Since these are noted in
the transcript of the meeting, we do not intend to specify a line-by-line
editing but note that the staff is in general agreement with these changes.

-------
          REPORT OF THE CHEMICAL MIXTURES GUIDELINES REVIEW GROUP

 Introduction

      The problem of assessing health risks of exposure to chemical mixtures
 is a very important issue for the Agency.  In terms of public concern, it
 is one of the highest priorities in public health, highlighted by the
 reactions of communities impacted by hazardous waste dump sites.

      Most environmental exposures are to mixtures rather than to single
 chemicals.   In a strict sense, it is correct to think of all exposures
 as exposures to complex mixtures of chemical and physical factors, variable
 with time and interacting in complex ways.  Focusing regulatory attention
 on single chemicals results from many practical considerations but tends
 to oversimplify the problem of assessing health risks.

      This is the Agency's first attempt to develop guidelines for risk
 assessment of chemical mixtures.  It is a good first effort, but should
 be considered only as a step in the evolution of guidelines which should be
 revised and improved as more scientific information on interactions becomes
 available and as risk assessors evaluate their experience periodically.

      The proposed guidelines submitted to the Review Group were released
 for public comment on January 9, 1985.  An internal Agency working group
 has reviewed the public connents submitted by March 11, 1985, but it has not
 had time to incorporate changes in a revised set of guidelines.  EPA staff
 will consider the comments of the present Review Group and of the public
 for preparation of a new draft.  The Review Group expects that the revised
 document will be submitted for its review.

 Conceptual Framework of the Guidelines

      The Review Group considers the conceptual framework of the current
 guidelines to be scientifically sound.  It notes that the conceptual
 issues on which these guidelines are based are very different fron chose
 of other risk assessment guidelines, for example, those which address
 health effects such as cancer.

      The Review Group also recognizes that there is a relatively neager
 data base on the health effects of chemical mixtures and little experience
 in preparing these kinds of risk assessments.  Scientific evaluation of
 potentially adverse health effects of chemical mixtures is an area of
 toxicology that has not undergone extensive research and testing either
 in humans or in laboratory animals.  Furthermore, such research and
 testing is complicated by the number and variety of different cherucal
 mixtures encountered, either nan-made or those that occur naturally,  in
 addition, when interactions of more than two chemicals are studied
 systematically, the complexity of experimental design is increased and
.the size of the project and the costs can increase enormously.

      Although not a matter of conceptual framework, the guidelines would
 be better understood and more easily used by risk assessors and the public
 if they were substantially rewritten.  Information on potential uses of
 the guidelines would be helpful to individuals outside the Agency.  Numerous
 reviewers have noted the absence of any definition of chemical mixtures.
 A few examples are given, but no limitation is placed .on what might
 constitute a complex mixture.   There is also no recognition of the temporal
 component of exposures to mixtures.  In several parts of the document
 more care is needed in precise definition of terminology.

-------
 Recommendation

      The current proposed guidelines should be revised addressing conments
 received at and since the March 4, 1985 preliminary meeting of the Review
 Group as well as those received fron the April 22-23, 1985 meeting.  This
 redraft should be reviewed by mail by the Review Group at which time the
 need for further review will be considered, including the possibility of
 resubmission for public comments.

 Scientific Issues

      1.  Issues relating to the guidelines' summary in Table 1 (page 1171)

      The proposed guidelines recognize that the amount and types of data
 available on a chemical mixture may vary considerably.  The recommendation
 is to anphasize "flexibility, judgment and clear articulation of the
 assumptions and limitation in any risk assessment."  EPA staff have
 incorporated a table summarizing the proposed approach to chemical mixtures.

      The Review Group found this tabular summary helpful but has several
 suggestions intended to increase its utility.  These include:

      o  The concepts in Table 1 should be presented in flow chart form
         such that the decisions to be made at each branch point are
         illustrated, the consequences of that decision made explicit
         and, if possible, the data needs articulated.  The table format
         does not allow for recycle loops or multiple options if appro-
         priate.  This revision in format could also reflect the differences
         in the quality of available information where one set of decisions
         are clearly data-based and another are highly inferential.

      o  In the text or in an appendix, EPA should provide illustrative
         examples of how the guidelines would be used.  These examples would
         be drawn from actual cases performed by the .Agency or would be
         simulations designed to elucidate the procedure.   It has been
         noted elsewhere in this report that the guidelines contain too
         few exanples.  This table (or chart), which for some readers
         will be the only thing they read, would benefit frcm such an
         illustration.

      o  The Review Group found that not all important outcomes were
         accounted for in the table.  It recommends that the flow chart
         or comparable device be expanded to include additional options.
         For example, where more than one outcome is of importance to the
         risk manager, the interaction of such outcomes should be indicated.
         This interaction could involve carcinogenic and systemic toxicity
         endpoints or mutagenic and developmental effects.

      In the evaluation of "sufficient similarity," there is the need for
'the guidelines to recognize that this concept could be defined in terms
 of composition (e.g. an homologous series of hydrocarbons) or in terms of
 end-effects (e.g. a series of Central Nervous System depressants such as
 alcohols,  ketones and simple esters).

      The importance of exposure data in identifying and quantifying risk
 is not integrated into the proposed scheme.  Such data are needed for the
 proper use of Step 3 which deals with indices of acceptable exposure.

-------
      Most importantly, there should be a branch in the decision logic
 which leads to a "non-definable risk" outcome.  The Review Group believes
 that conditions can exist in which there are no toxicity data available
 on the chemical mixtures; there is no "sufficiently similar" mixture
 which has been studied; the number of unidentified components is large
 and/or the amount of data on the few identifiable components is small.
 In such instances, hazard identification, the acknowledged first step in
 risk assessment, is not possible.  Consequently, the decision process snculd
 move to a "non-definable risk" outcome.  The alternative is an assessment
 with such wide uncertainty bands as to be meaningless or misleading.

      2.  Greater emphasis on levels of confidence in risk assessments

      The scientific basis for estimating the risks of chemical mixtures is
 highly variable depending upon the type cf mixture, the complexity of the
 mixture, and data availability.  A risk assessment based upon bicassay
 data for the mixture in question, for example, probably has a mucn higher
 degree of confidence associated with it than dees a risk assessment based
 upon the toxicity of some of the components of a poorly-characterized
 mixture.  There is a need to express the degree of confidence to be
 associated with various risk assessments of chemical mixtures.  Accordingly,
 the Review Group recommends that the guidelines be revised to develop and
 incorporate a system to express the level of confidence associated witr. a
 risk assessment.  Such a systan might, for example, be analogous to tr.e
 IARC classification system for carcinogens, which summarizes the weight
 of evidence for carcinogenicity of a chenical.  The development cf sucn a
 system will probably require a taxonomy of chemical mixtures.  Sucn a
 systan should take into account the nature cf risk associated witn the
 components of the mixture (all carcinogens, all toxicants, roo present),
 the complexity of the mixture (well characterized, composed of few or
 many chemicals, poorly characterized)-, and data availability >available
 for the same or similar mixtures, available for some or all components, and
 knowledge of potential interactions).

      The Review Group also notes that there are more assumptions and
 uncertainties associated with the risk assessment cf chemical mixtures
 than of single chemicals.  The mixtures may be poorly characterised; in
 some cases, only some of the components may re identified cr the mixture
 may be defined to be similar to another one.  little information is
 probably known about potential antagonistic or synergistic reactions, ?.nd
 often this information may be qualitative in nature.  Theories and motels
 developed for single chemicals may not be appropriately transferred to
 chemical mixtures; for example, the extrapolation of high cose response data
 on a mixture having carcinogenic properties may not be the same =s for single
.chemicals; this is also true for interspecies extrapolations, arc any
 interactive mechanisms which may be present may not 're constant across all
 dose levels.  The Review Group urges that tr.es-? 'uncertainties be clearly
•articulated and that sensitivity analyses re 'undertaken,  where appropriate,
 to reflect the likely impact of some of tr.is 'uncertainty  on the risk
 estimates.  The sensitivity analyses snculd reflect differences in ways
 to address the risk of a chemical mixture 'e.g. ,  -vhether it is rased en
 the toxicity of a "similar mixture" or on knowledge cf the toxicity cf
 constituent components),  differences in tr.e application of the acditivity
 assumption in calculating the hazard index I different groupings cf
 mechanisms), and differences due to interaction assumptions that may b-e -.

-------
                                  - 4 -
based upon qualitative or suggestive data.  Variations inherent in single
chemical risk assessments should also be covered by the sensitivity
analyses; these variations include the use of alternative data sets, extrap-
olation from different species, use of alternative pathology results, use
of confidence intervals and alternative exposure levels.

     3.  Need for a technical support document

     The technical and scientific background from which these guidelines
must draw their validity is so broad and varied that it can't reasonably
be summarized within the framework of a brief set of guidelines.  Numerous
reviewers, including the Review Group, believe that the guidelines would
greatly benefit by the preparation of a technical support document.  Such
a document, as a minimum, would describe the biologic principles of
toxicologic interactions associated with multiple chemical exposures;
would identify alternative approaches, the assumptions behind these
approaches and the impacts of violating the assumptions; and would use
illustrative examples liberally throughout the document.  The support
document would also supply the technical references which are somewhat
deficient in the guidelines.  The Review Group suggests the following as
sons of the topics that might be included in such a technical document
(it is assumed that the EPA working group would use this as a guide1", and
would undoubtedly add other topics of its own):

     o  Summaries of toxicity studies of chemical mixtures, including
        their relevance to the theoretical basis of interactions.

     o  Definition and basis of dose additivity models, with illustrative
        examples.

     o  Similar delineation of response additivity, with examples.

     o  Definition and examples of risk additivity.

     o  How to quantify uncertainties.

     o  How to deal with mixtures that contain both carcinogenic and
        systemic toxicants that are non-carcinogenic.

     o  Role, importance, need for mechanistic data on toxicologic
        interactions.

     o  Effect of temporal sequence on additivity models.

     o  Recommendations for methods of testing mixtures.

     o  How to test mathematical models for plausibility.

     o  How to deal with the Maximum Tolerated Dose (MTD) concept in
        carcinogen bioassay as it deals with mixtures since no one
        conponent may be present at KTD.

-------
                                  - 5 -
     o  Is there a theoretical or practical limit to the number of agents
        in a dose additive model?

     o  Role of matrix sensitivity—toxicity of an agent may be a function
        of other agents being present.

     o  Criteria and methods for evaluation of data quality.

     o  Relationship between mechanisms of interactions and mechanisms of
        toxicity.

     o  Research needs

Other Issues

     1.  Identification of research needs

     Although the toxicity of chemical mixtures has been ..recognized for
decades as an important area for toxicolcgic research, there has been
little emphasis on this area until recently.  Consequently, the existence of
large gaps in available information is not surprising.  The Review Group
believes that an important mechanism for identifying critical research
needs is inherent in the process of carrying out risk assessments.  The
Office of Research and Development should ensure that there is an effective
linkage with the research planning process to take full advantage of—
these opportunities.

     The preparation of the technical support document will also identify
nany research needs, which likewise should be made available and utilized in
zhe research planning process.

     2.  Use of case-by-case approach.

     As indicated in the guidelines and elsewhere in this report, the
oasis of scientific information is very meager for quantitative risk
assessment for exposure to multi-conponent mixtures, with "the exception
of a few complex single source emissions or products (such as coke-oven
exhaust, gasoline, PCBs).  Consequently it is appropriate, as the guidelines
suggest, to be very flexible in approaches to risk assessment.  The Review
Group suggests that risk assessments of chemical mixtures be done on a
case-by-case basis, until a substantial background of experience is built
up in this area.

     3.  "Enshrining the uncertainty"                                ,

     The guidelines themselves, and to a greater extent this Review Group
report, have emphasized the need to specify .the level of confidence (or
degrees of uncertainty) in various steps of the risk assessment.  The
Review Group believes these estimates should not only be a component of
the risk assessment but should be carried forward through all steps in
the risk management process and be made a part of releases of information
to the public.  The Review Group indentifies the objective of this process
to be "enshrining the uncertainty."

-------
                                  - t> -
Overall scientific adequacy of the Risk Assessment Guidelines for Chemical
Mixtures.

     The Review Grcup believes it would be inappropriate to state a
conclusion on scientific adequacy at this time, inasmuch as the draft
guidelines are in the process of revision.  The document was on a different
schedule by several '*eeks, and the closure of public comments did not
occur until March 11, 1935.  The EPA working group is already preparing a
revised draft, based on the public conments, Agency comments, and the SAB
meetings cf March 4 and April 22-23.  EPA staff anticipate completing the
revision within approximately 30 days and has asked the Review Group to
consider the revised draft.  This will be done by mail and by a conference
call; another meeting will be held if deemed necessary after the conference
call.

     The Review Grcup believes that present operational guidelines for
chemical mixtures should continue to be used as interim guidelines.

     The technical support document should be submitted to the usual
peer review process for technical documents and should be submitted to the
SAB for reviev.  The SAB should evaluate the document in terms of its suit-
aoility as support for the chemical mixtures guidelines, and should
recommend whether the draft support document should be made available for
cuolic comment.

-------
      REPORT OF THE DEVELOPMENTAL EFFECTS GUIDELINES REVIEW GPOUP


 Scientific Adequacy of the Guidelines

      The Review Group concludes that, in general, the proposed guidelines
 are scientifically adequate and commensurate with the present state of
 the science.  However, the field of developmental toxicology is particularly
 weak with respect to quantitative assessments.  The guidelines could be
 improved by revisions in the section relating to the relationships of the
 maternal toxicity in comparison to toxicity for the fetus.  The bases for
 quantitative assessment also require elaboration.  The Review Group
 requests responses from the Agency on the specific points  raised and the
 opportunity to review a revised draft at which tine we will render a
 final opinion.

 General Conclusions

      The proposed Guidelines for the Health Assessment of  Suspect Develop-
 mental Toxicants are generally well written and appropriately express
 the kinds of actions which are possible, given the present state of the
 art.  Developmental toxicity studies have largely addressed qualitative
 responses, and the guidelines reflect this situation.  Most of the testing
 methodologies employed at present are not as readily adaptable to quanti-
 tative assessments, especially with respect to chronic exposures.

 Specific Issues

      1.  Integration With Other Guidelines

      The Review Group understands that the guidelines for  reproductive .
 effects that are not maternally mediated are still under development.  It
 is also clear that many instances will occur in which it will be difficult
 to differentiate between paternally mediated effects, effects originating
 during fertilization, maternally mediated reproductive effects and direct
 effects on the fetus.  As EPA develops these additional guidelines for risk
 assessment it should integrate them with the existing guidelines as they
 evolve.

      2.  Definitions

      The Review Group briefly discussed definitions used in the proposed
 guidelines and concurred that the concept of "functional developmental
 toxicology" was useful.  However, it also noted that the well-established
 concept of "teratogenicity" appeared to be subsumed under other definitions
; and concepts in a number of instances.   EPA needs to make distinctions,
. in the definitions and use of these concepts.

       3.  Qualitative Assessments

      The possible number of adverse effects in reproductive toxicology,   •
 in contrast to the possible number of outcomes in carcinogenicity testing,
 is very large.  Consequently, the probability of missing an adverse effect,
 because it was not anticipated in the experimental design, is a much more
 likely event.  This fact introduces complexities into the assessment of
 developmental toxicity that EPA should keep in mind.

-------
                                  -  2 -

     If the available animal data indicate  that a substance disrupts
development only at maternally toxic doses, and if exposure potential  may
also include the maternal toxic dose range, then the  substance merits
careful scrutiny and is a possible candidate  for additional study.

     In evaluating dose-related effects observed in maternal and  developing
organisms  (at any dose used), it is  important to apply  judgment as  to  the
comparative nature and severity of these effects.  Moreover, effects in
the developing organism in the presence of maternal effects should  not
necessarily be considered secondary  effects or of lesser significance  in
evaluating developmental toxicity.

     The proposed guidelines anphasize that the responses  in the  most
appropriate and/or sensitive species should be used in  risk assessment.
The Review Group suggests that the most appropriate species be determined
on the basis of metabolic and pharmacokinetic characteristics with  direct
bearing on the toxic mechanism in the experimental organism in comparison
to humans.

     EPA should evaluate functional  tests in  assessing  developmental toxicity
on the basis of sensitivity of such  tests to  subtle toxic  effects,  the
recognition that functional tests may assist  in interpreting the  biological
significance of other effects of exposures, and the fact that data  obtained
fron human populations have sometimes indicated functional  deficits.   For
example, behavioral testing in animals may also be useful  in assessing
questions of toxicity to the nervous system.

     The Review Group expresses concern about the use of screening  tests  for
developmental toxicity.  It endorses the use  of single  high dose  screening
studies for the purpose of prioritization (as developed by Chwernov),
with reservations.  Panel members share the concern of  several previous
conmenters that materials found to be positive in this  screening  test
might be permanently labelled as "teratogenic" by the public at large, a
label which we consider to be possibly premature in some instances  since
the test is likely to produce a relatively high number  of  false positives.
Although we believe that it is desirable that  screening tests  produce
very few false negative responses, and thus are bound to produce  false
positive responses, the tendency to place immediate labels  on  the positive
responders is unfortunate and unjustified in  some instances.

     The Review Group also considered the possible utility  of  structure-
activity relationships in screening  new chemicals for possible developmental
effects.  Structure-activity relationships, although potentially  useful,
are not sufficiently advanced at this time to  be generally  useful
even for preliminary assessments, with the possible exception  of  preliminary
assessments of hormone analogs.

     4.  Quantitative Assessments

     On theoretical grounds,  in most cases a  threshold should  exist  for
developmental effects in terms of the dose required to elicit  toxicity.
However, there is not a sufficient basis to rule out the possibility that
non-threshold models may he more appropriate  for some endpoints of certain
toxicants.

-------
     It is important to recognize that the No-Observed-Effect-Level (NOEL),
No-Observed-Adverse-Effect-Level (NOAEL) and Lowest-Observed-Effect-Level
(LOEL) approaches attempt to approximate an apparent threshold region
that incorporate judgment as part of the quantitative approach.  However,
the presence or absence of a threshold is, at present, almost impossible
to ascertain, and that an exact threshold cannot be readily determined by
mathematical inference.  The mere existence of an NOEL neither proves r.or
disproves the existence of a real threshold.  An experimentally determined
NOEL is not necessarily at or below a threshold dose and, in that context,
it is important to examine the quantitative relationships of NOEL dose
rates to dose rates that result in effects.  The examination of the
effects needs to take the type of severity of the effect into account
including the dose rates at which they are elicited.  The application of
a safety factor or an uncertainty factor to an NOEL may still result in a
residual risk.

     In some instances it may be necessary to conduct a reanalysis of
experimental data to determine whether the conclusions drawn by the
authors of a study were valid, especially when the original types of
statistical analyses seem to have been inappropriate, or when the reported
statistical outcomes do not seem to be supported by the data.

     The Review Group considered whether EPA should introduce a weighting
factor into the range of adverse outcomes that could result fron exposure
to a developmental toxin.  The consensus opinion was that the state of
the science was not sufficiently developed to enable a determination of
the relative importance of structural abnormalities, altered growth,
functional deficiencies, and intrauterine death; the considerable
uncertainties in extrapolating from one such effect to another, as well
as frcm species to species for a given effect, was also noted.  For hunan
data, such determinations extend beyond the area of risk assessment into
the risk management area.  The question of malformations versus anatomic
variations was also discussed.  The Review Group believes that, althougn
the significance of anatomic variations is at present not well understood,
a dose related increase in their incidence should be taken into account
in the risk assessment process.

-------
        REPORT OF THE SAB EXPOSURE ASSESSMENT GUIDELINES REVIEW GROUP
       The proposed Guidelines  for Exposure Assessment provide a  very good
  statement on  the general principles of exposure assessment and  logical
  procedures to follow in the absence of reliable measurement data on
  exposure.  Seme specific conments which relate to the discussion on
  uncertainty are offered in a  later section of this review and should be
  considered in the revisions and corrections that were given to  EPA  staff
  during the course of the review.

       Despite  the generally satisfactory conditions of the material
  included within the proposed guidelines, the Review Group concludes that
  it  is unsatisfactory to entitle them as the EPA's "Guidelines for Exposure
  Assessment" because they give virtually no guidance on the uses of  environ-
  mental measurements for exposure assessment.  Some of the EPA program
  offices make  extensive usage of measurement data in their exposure  assess-
  ments, and quite properly so.  Agency-wide guidelines should certainly
  acknowledge such usage and encourage more of it.  The Review Group  was
  unanimous in  concluding that direct measurements of concentrations  of
  pollutants in air, water, food, and soil, made with suitable sampling
  protocols and quality assurance practices, usually provide better indices
  of human exposure than theoretical models.  A secondary benefit in  the
  use of this approach to exposure assessment is that accumulated data can
  provide a basis for the validation of exposure models.

       The proposed guidelines appear to have been written fron the per-
  spective that exposure assessment should depend primarily on nodels,
  supplemented  when possible by measurements of the characteristics and
  strengths of  pollutant sources such as stacks, discharge pipes, fugitive
  industrial emissions, and waste dumps.  They give virtually no recognition
  to more directly relevant environmental media at the receptors.  Specific-
  ally, the guidelines state (page 46307) that "The analysis of monitoring
  data should be considered a complement to environmental pathway and fate
  analysis..."  They devote space to a general discussion of environmental
  fate, transport and transformation, and identification of principal path-
  ways of exposure.  However, not until the last sentence on Monitoring
  (page 46308)  is an acknowledgement made that "Reliable, analytically
  determined values should be given precedence over estimated values  whenever
  significant discrepancies are found" (between model and measurement data).
  The Review Group believes that measurements provide the best basis  for
  exposure assessment, and that modeling should be used only when there are
  too limited numbers and/or inaccuracies in available measurements.

       The Review Group unanimously concludes that the guidelines will only
  be partially complete until they are expanded to include a whole new
^  section on the general principles of the measurement of pollutant concen-
  trations in the various environmental media.  Furthermore, the preface to
. the existing proposed guidelines should indicate that environmental
-  measurements, vftenever feasible should have•precedence over estimates
  based on theoretical or empirical models.   Where a limited amount of good
  environmental measurement data are available and need to be expanded with
  modelled exposures, the measurements provide a means of guiding the
  selections of model input parameters which can give a greater degree of
  confidence in the model predictions.  The guidelines should also address
  the limitations of environmental concentration data for exposure assessment.

-------
                                  - 2 -


A section should be added which delineates errors encountered such  as  trie
representativeness of the samples, sampling error, laboratory analysis
errors, and data manipulation errors.  It should also discuss the differences
between precision and accuracy.

     The supporting documentation for the proposed guidelines now includes
a background paper using TCDD as a model for exposure assessment.   This
provides a good example of how one can proceed to nake an exposure  assess-
ment in the absence of relevant data.  The documentation for the final
guidelines would benefit frcm the inclusion of two other examples as well.
One should be for a pollutant for which a relative wealth of environmental
exposure data are available, such as carbon monoxide or lead.  Another
should be for pollutant for which a limited amount of data are available,
e.g., benzol or arsenic.

     The proposed guidelines have not specified the form of the output to be
generated from the exposure assessment.  We reccrmend that a specific fom
of the output be proposed, such as an estimated exposure distribution,
specific of each target population, supplemented by appropriate indications
of the uncertainty (e.g., confidence limits), for the estrnated distribution
function.

     When containing estimated exposures across various scenarios, the
association or correlation among the various components needs to be addressed.

     When dealing with synergistic responses in the combined risk assessment
context, the association or correlation anong the agents needs to be addressed.

     In summary, the Review Group finds that the proposed guidelines provide
a very good overview of general principles to be followed in measuring
pollutant concentrations in environmental nedia.  However, in their present
state, they fail to sufficiently emphasize that exposure assessments rased
on reliable measurements of pollutant concentrations in environmental nedia
should take precedence over exposure estimates based on unvalidated rccels.
In addition, they fail to provide guidance on the general principles to be
followed in measuring pollutant concentrations in environmental media.  These
critical deficiencies need to be properly addressed before they can be
recommended for approval as Agency-wide Guidelines for Exposure Assessment.
At the same time, Agency staff should proceed to finalize those sections cf
the guidelines that state general principles of exposure assessment and
procedures to follow in the absence of measurement data, following  further
consultation with the Review Group.

     In discussing the report of the Review Group the SAB Executive Cxnittee,
by majority vote, further recommended that:  1)  the preface of the Guidelines
be revised to snphasize that direct measurements, whenever feasible, -ave
precedence over estimates based on unvalidated nodeIs as the basis  fcr the
Agency's exposure assessments; 2) the revised guidelines, which should include
the new preface and the corrections resulting from the Review Group and puolic
comments, be retitled to indicate that it represents general guidance and
procedures to be followed when direct measurements are not feasible, and to
indicate that they are not overall guidelines for exposure assessment; and
3)  that the Agency commit itself to the preparation, on a priority  basis,
to the preparation of a guidelines document on the performance of exposure
assessments based in whole, or in part, on direct measurements of pollutant
concentrations in environmental media.

-------
                                  -3-
Specific Garments on Uncertainty

     1. We commend Dr. fchitmore's work on the uncertainty analysis which
        is utilized in the proposed guidelines.  The analysis is useful
        where environmental measurements are not available or feasible,
        and exposures and exposure distributions need to be assessed
        using a theoretical model.

     2. An important component of uncertainty that is not sufficiently
        addressed is uncertainty about the form of the model, i.e.,
        systematic error or bias arising from the use-of a model.  Mien
        environmental measurements are available, the validation of the
        model can be used to determine the presence and magnitude of
        systematic errors.  When exposure measurements are not available,
        some subjective judgment will be needed.

     3. Another important component of uncertainty is the random error
        not predicted by the model, assuming that the model predicts the
        expected exposure under the assumed input variables, or some
        other summaries of the conditional exposure distribution (condi-
        tioned on the assumed input variables), leaving random components
        unexplained.  It is important that this component of variation be
        included in the assessment of the distribution of exposures
        across the members in the target population.

     4. It is important to impose probabilitic interpretations on the
        uncertainty assessment.  One specific instance where this is lack-
        ing is the consideration of the range of an input variable.  The
        range appears to be interpreted in an absolute sense:  all members
        of the target population are expected to fall inside the range.
        Such an absolute range is either impossible to assess, or too wide
        to serve any useful purpose.  It will be more useful to impose a
        probabilistic interpretation on the ranges.  (For .example, ranges
        in which 90% of the cases are expected to fall.)

     5. The enhanced sensitivity analysis is not clearly presented,  it
        should be emphasized that the true distribution of the input
        variable is likely to be less variable  than the uniform distri-
        bution assumed in this analysis,  and the exposure distribution
        imputed frcm this analysis will show more variability than the
        true exposure distribution.   It should, therefore, be viewed as a
        conservative estimate (overestimate)  of the prevelance in the
        upper range of exposures.

     6. The sensitivity analyses should be used as screening tools to
        identify the input variables which require further assessment.

     7. Since it is likely that certain subjective assessments will be
        used, either in the form of expert judgements or range-finding
        calculations, a review of the techniques that can be used to
        reach the best subjective assessments from individual experts
        also needs to be addressed,   toen limited environmental  measure-
        ment data are available, the issue on combining subjective assess-
        ments and empirical measurements needs  also to be addressed.

-------
                  U. S. ENVIRONMENTAL PROTECTION AGENCY
                          Science Advisory Board
                 Risk Assessment Guidelines' Review Groups
Dr. Morton Nelson   (Chairman)
Professor of Environmental Medicine
Institute of Environmental Medicine
New York University Medical Center
500 First Avenue
New York, New York  10016
Dr. Terry F. Yosie
Director, Science Advisory Board
Q. S. Environmental Protection Agency
401 M Street, S. W.
Washington, D. C. 20460
Carcinogenicity Guidelines Review Group

Dr. Richard A. Griesemer  (Panel Chair)
DirectorF Biology Division
Box Y
Oak Ridge National Laboratory
Oak Ridge, Tennessee  37830

Dr. Roger O. McClellan
Director of Inhalation Toxicology
Research Institute
Lovelace Bicmedical and Environmental
Research Institute
P. 0. Box 5890
Albuquerque, New Mexico  87185

Dr. Frederica Perera
School of Public Health
Division of Environmental Sciences
(B-109)
Columbia University
60 Haven Avenue
Mew York, New York  10032

Dr. Norton Nelson
Professor of Envirormental Medicine
Institute of Environmental Medicine
New York University Medical Center
500 First Avenue
New York, New York  10016
Dr. Ronald Hart
Director
National Center for lexicological
  Research
Jefferson, Arkansas  72079

Dr. Warner North
Principal
Decision Focus, Incorporated
Los Altos Office Center
49fl4 El Camino Peal
Suite 200
Los Alto, California  94022

Dr. Henry Pitot
Director
McArdle Laboratory
450 Nbrtn Randall Avenue
Madison, Wisconsin  53706

-------
                                  - 2 -
Mutagenicity Guidelines Review Group

Dr. Seymour Abrahamson
Professor of Zoology and Genetics
Department of Zoology
University of Wisconsin
Madison, Wisconsin  53706
Dr. Robert Neal (Panel Chair)
President, Chemical Industry
   Institute of Toxicology
P. 0. Box 12137
Research Triangle Park, NC  27709
Dr. Michael Shelby
National Institute of Envirormental
   Health Sciences
Post Office Box 12233
Research Triangle Park, N. C.  27709
Dr. Mary Esther Gaulden
Department of Radiology
University of Texas Health
Science Center—Dallas
5323 Harry Hines Blvd.
Dallas, Texas  75235

Dr. Liane Russell
Section Head
Biology Division
Oak Ridge National Laboratory
Post Office Box Y
Oak Ridge, Tenn.  37831

-------
                                  - 3 -
Chemical Mixtures Guidelines Review Group

Dr. John Doull                                Dr.
Professor of Pharmacology and
  Toxicology
Department of Pharmacology
College of Health Sciences
  and Hospital
The University of Kansas
39th and Rainbow Boulevard
Kansas City, Kansas  66103

Dr. Robert Scala -
Exxon Corporation - REHD
Post Office Box 235
Metier's Road
East Millstone, New Jersey  08873
Dr. Ronald wyzga
Program Manner
Electric Power Research
  Institute
3412 Hillview Avenue
Post Office Box 10411
Palo Alto, California  94303
                                                  Ciarles Reinhardt
                                              Director, Haskell Laboratory
                                                for Toxicology and
                                                Industrial Medicine
                                              E I du Pont de Nenours and Company
                                              Elkton Road
                                              Newark, Delaware  19711
                                              Dr. James Whittenberger  (Panel Chair)
                                              Director, Southern Occupational
                                                Health Center
                                              19722 MacArthur Blvd.
                                              University of California
                                              Irvine, CA  92717

-------
                                  - 4 -
Developmental Effects Guidelines Review Group

Dr. Larry Feohter
Kresge Hearing Research Institute
University of Michigan Medical School
1301 East Ann Street
Ann Arbor, Michigan  48109

Dr. Rolf Hartung (Panel Chair)
School of Public Health
University of Michigan
Ann Arbor, Michigan  48109
Dr. Marshall Johnson
Department of Anatomy
Thomas Jefferson University
1020 Locust Street
Philadelphia, PA  19107

Dr. Ellen Silbergeld
Chief Toxics Scientist
Environmental Defense Fund
1525 18th Street, N. :•/.
Washington, D. C.  20036
Dr. David Gaylcr
Division of Bicnetry
National Center for
  Toxicological Research
Jefferson, Arkansas  72079

Dr. Ronald D. Hood
Professor, Developmental Biology
Section, Department of Biology
University of Alabama and
  Principal Associate, R. D. Hood
  and Associates
Consulting Tbxicologist
Post Office Box 1927
University, Alabama  35486

Dr. Susan Sieber
National Cancer Institute
9000 Rockville Pike
Building 31  Roan 111.03
Bethesda, Maryland  20 20 5

-------
U.S  Environmental  Protection Agency
Region V, Library
230  South Dearborn Street
Chicago,  Illinois   60604

-------
                                  - 5 -
Exposure Guidelines Review Group

Dr. Richard Cuddihy
Senior Scientist
Inhalation Toxicology
  Research Institute
Lovelace Bionedical and
  Environmental Researach Institute
Post Office Box 5890
Albuquerque, New Mexico  87185

Dr. Thomas Kittleman
Loubiers Building
13W10
Dupont Company
Wilmington, Delaware  19898
Dr. Jerry Wesolowski
California Department of Health
2151 Berkeley Way
Berkeley, California  94704
Dr. Naihua Duan
University of Wisconsin
Mathematics Research Center
610 Walnut Street
Madison, Wisconsin  53705
Dr. Morton Lippmann (Panel Chair)
Institute of Environmental
  Medicine
Lanza Laboratory
Long Meadow Road
New York University
Tuxedo, New York 10987

-------
U.S.  Envron.v <•'1 :~.s-. .?.-;on Agency
R2g;on  V,  Lib":,  y
230 SojLi'i Dc-ribc-rn °';?'=t
C'-.^.-ro, Illinois   6U'.".04

-------