EUR 15606 EN
             EPA / 600/R-94 7042
       European Commision
 Science, Research, and Development
U.S. Environmental Protection Agency
 Health Effects Research Laboratory
                "Human genetic risks from exposure to
               chemicals, focusing on the feasibility of a
                      parallelogram approach"

                       Proceedings of the

             EC/US Workshop on Risk Assessment

                      October 11 -14,1993

                   Durham, North Carolina, U.S.A.
    Mouse Somatic Cells

    Measured Mutations
       and Adducts
     Mouse Germ Cells

    Measured Mutations
       and Adducts
     Human Somatic Cells

     Measured Mutations
         and Adducts
      Human Germ Cells

     Estimated Mutations
                           Comparisons

                              Estimates

-------
 DISCLAIMER
 EUR 15606 EN

 This report EUR 15606  EN of the  ENVIRONMENT Research Programme of the
 European  Commission  (Directorate-General  XII  for  Science,  Research,  and
 Development). The texts in this volume reflect the views of their authors and do not
 in any way engage the Commission.

 EPA/600/R-94/042

 The following is  a  record  of presentations  and discussions at the European
 Commission  (EC)/United States (U.S.) Workshop on Genetic Risk  Assessment:
 "Human Genetic Risks from Exposure to Chemicals, Focusing on the Feasibility of a
 Parallelogram Approach," which was held at the Omni Durham Hotel and Convention
 Center in Durham, North Carolina, on October 11-14,1993. Each speaker was given
the opportunity to edit the verbatim transcript of his/her presentation.  This material
was prepared by Research and Evaluation Associates, Inc., under contract to the U.S.
 Environmental Protection Agency. The views or opinions expressed in the attached
do not necessarily represent the views or policies of the U.S. Environmental Protection
Agency.

-------

-------
                           TABLE OF CONTENTS

SECTION                                                           PAGE

List of Figures	 iv

List of Tables	 ix

Overview	 1
       Introduction  	 3
             Workshop Organization	 4
       Synopses	10
             Plenary Sessions	10
             Compound Reviews	18
             Rapporteur's Reports	18
             Working Group Summary 	27
             Crosscutting Papers  	28
             Concluding Remarks	29

Plenary Sessions	33
       Introductory Remarks   	35
             Dr. Lawrence Reiter  	35
             Dr. Heinrich Ott   	37
       Parallelogram Concept (Dr. Alan Wright)  	41
             Key Scientific Issues in Assessing Genetic Risks	41
             Merits of Parallelogram Approach	55
             Limits of Parallelogram Approach	57
             Discussion	66
      ICPEMC Efforts (Dr. David Brusick)   	71
      Regulatory Perspective  (Dr. Kerry Dearfield)  	87
             What Are the Reasons for Mutagenictty Testing?	88
             How to Determine if There is a Genotoxidty Concern?	91
             How to Determine Real Risks to Humans?	97
             Discussion	109
      Research Background for EO (Dr. Lars Ehrenberg)  	113
             The Use of Hemoglobin Adducts	114
             Ethylene Oxide Data  	131
      EO Mutagenicity Risk Assessment (Dr. Vicki Dellarco)  	139
             Discussion	159
     A Reconsideration of EO Risk Assessment (Dr. Julian Preston)	161
             Translocation Mechanisms and the Stages of
              Spermatogenesis 	163
             Translocation Mechanisms and Oogenesis  	168
             Implications for Risk Assessment	170
             Discussion	177

-------
                      TABLE OF CONTENTS (continued)

 SECTION                                                           PAGE
 Data Summaries of Compounds
       Ethylene Oxide (Dr. AT. Natarajan) ............................ 183
             Discussion ......................................... 195
       Acrylamide (Dr. Kerry Dearfield)  .............................. 199
             Discussion ......................................... 221
       1,3-Butadiene (Dr. Ilse-Dore Adler)  ............................ 225
             Discussion ......................................... 242
       Cyclophosphamide (Dr. Diana Anderson) ....................... 247
             Adducts In Vitro ..................................... 251
             Adducts In Vivo ..................................... 253
             Adducts in Humans .................................. 253
             Somatic Mammalian Cell Genotoxicity In Vitro and
              In Vivo ........................................... 254
             Discussion ......................................... 270

 Rapporteur's Reports ............................................ 275
       Ethylene Oxide (Dr. Julian Preston)  ........................... 277
             Initial Report ....................................... 277
             Final Report ........... ............................. 287
             Discussion ......................................... 303
       Acrylamide (Dr. George Douglas)  ............................. 309
             Initial Report ....................................... 309
             Discussion ......................................... 310
             Final Report ................... ...................... 311
             Discussion ......................................... 318
       1,3-Butadiene (Dr. Marja Sorsa)  .............................. 321
             Initial Report ....................................... 321
             Final Report ........................................ 333
             Discussion ......................................... 339
       Cyclophosphamide (Dr. Paul Selby)  ........................... 341
             Initial Report ....................................... 341
             Final Report ........................................ 345

Crosscutting Remarks ........................................... 353
       Differential Repair of Chemically-Induced Damage:
        Somatic Vs. Germ Cells (Dr. Ekkehart Vogel)  .................... 357
            The Male Germ-cell Cycle  ................................
             DMA Damage and Repair in Drosophila ......................
            DMA Damage and Repair in the Mouse  ......................
            Comparison of  Mouse and Drosophila .......................
            Conclusions and Perspectives .............................

-------
                      TABLE OF CONTENTS (continued)

 SECTION                                                            PAGE

       Chemically-Induced Mutation: Comparative
        Outcome in Somatic Vs. Germ Cells (Dr. Susan Lewis)   	372
             Special Attributes of Germ Cells	
             Female Germ Cells  	
             Cell Environment	
             Sex Specificity of Mutagens	
             Induction of Mutagens Resulting in Inborn
               Errors of Metabolism	
       Chemically-Induced Mutation: Comparative
        Outcome in Somatic Vs. Germ Cells (Dr. James Allen) 	376
       Qualitative and Quantitative Results from
        Short-Term Tests (Dr. Michael Waters)  	383
       Chemicals for Future Study of Heritable
        Genetic Damage in Human Populations
        (Dr. Michael Shelby)	403
             Discussion	

 Wrap-Up and Concluding Remarks  	413
       Wrap-Up (Dr. David Brusick)	415
       Concluding Remarks (Dr. Canice Nolan)	425

 References  	427
       References	429

Appendix A - List of Participants  	A-1

Appendix B - Agenda  	B-1

-------
                              UST OF FIGURES

 Figure                                                                E§Qe

 1-1.    The Sobels' Parallelogram  	  5

 2-1.    A Current Perception of the Principal Stages
         & Systemic Determinants of Chemical Mutagenesis  	42
 2-2.    Induction of Mutation by Genotoxic Chemicals	44
 2-3.    Management of Genotoxic Risks	44
 2-4.    Quantitative Genotoxic Risk Assessment	47
 2-5.    Factors Contributing to Mutation Load  	47
 2-6.    Experimental Determination of Mutation  	49
 2-7.    Extrapolative Stages in Assessing Heritable Risks	49
 2-8.    Systemic Determinants of Mutation	51
 2-9	53
 2-10.  The Parallelogram  	56
 2-11.  Induced Mutations in V79 Chinese Hamster Cells and the
         Mouse Specific-locus Assay	56
 2-12.  In Vivo-In Vivo Extrapolative Model  	58
 2-13.  In Vitro - In Vivo Extrapolative Model	62
 2-14.  Cytogenetic Damage in Human and Mouse Peripheral
         Blood Lymphocytes Exposed to Radiation In Vitro and In Vivo  	64
 2-15.  Conclusions	65
 2-16.  Mutation Frequency or DMA Dose	69
 2-17.  Mutation Frequency	69
 2-18.  Operational Definition	72
 2-19	72
 2-20.   Risk Extrapolation Road-Map 	74
 2-21.   Primary Project Objectives  	75
 2-22.   Data Availability for Risk Extrapolation	76
 2-23.   Proposed Scaling Factors (REF) for Mouse to Human
         Risk Extrapolation	77
 2-24.   Calculations Used in Risk Assessment	79
 2-25.   Limitations of the  Methods  	80
 2-26.   Summary of the Risk Extrapolation Approach	82
 2-27.   Ethylene Oxide	,	83
 2-28.  Acrylamide	84
 2-29.  What Do These Numbers Mean?  	85
 2-30.  Purpose of Mutagenicity Testing From Subdivision F	89
 2-31.  Initial Battery	92
 2-32.  USEPA OPPTS Mutagenicity Test Scheme 	92
 2-33	95
 2-34.   Examples of Human Diseases and Conditions Caused by
        Mutations in Germ Cells	98
2-35.   Difficulty of Identifying Human Germ-Cell Mutagens  	99
2-36	101
                                     vi

-------
                         LIST OF FIGURES (continued)

 Figure                                                               Page

 2-37.   Quantitative Estimation of the Genetic Risk Associated with the
         Induction of Heritable Translocations at Low-Dose Exposure:
         Ethylene Oxide as an Example	101
 2-38.   Ethylene Oxide	102
 2-39.   Germ Cell Risk Extrapolation from Mice to Humans  	105
 2-40.   Ethylene Oxide (EO)	113
 2-41	115
 2-42	115
 2-43.   Relationships between Alkylation of Hb and DNA by
         Ethylene Oxide	117
 2-44	120
 2-45	122
 2-46	124
 2-47.   Hb Adduct Levels in EO-Treated Rats	124
 2-48.   Persistence of 7-(2-hydroxyethyl)guanine in Exposure of Rats (a) to
         300 PPM Ethylene Oxide and Ethylene Oxide	125
 2-49.   Concentration-Response Curves for Induction of Dominant Lethal
         Mutations in Male Germ Cells	127
 2-50.   Deviations from Linearity	128
 2-51.   Non-linearity; Saturation of Detoxification at High Doses	129
 2-52.   Comparison of the Dose Response of N-(2-hydroxyethyl)valine in
         Hemoglobin and 7-(2-hydroxyethyl) guanine in DNA of
         Rats and Mice  	130
 2-53	132
 2-54	132
 2-55	134
 2-56	135
 2-57	138
 2-58	142
 2-59	145
 2-60.   ETO Dose-Rate Studies  	148
 2-61	148
 2-62	150
 2-63	150
 2-64	152
 2-65.   Biological Consequences of Rearranged Chromosomes	153
 2-66.   Two Chromosome "Breaks" in Close Proximity	153
 2-67.   Weibull (Extreme Value Distribution)	154
 2-68.   Genetic Risk to Human by Extrapolations	162
 2-69.   Dose Response Curve for X-Ray-lnduced Reciprocal Translocations in
        Mouse Spermatogonial Stem Cells	162
 2-70.   Spermatogenesis	165
2-71.   Cell Cycle	165
2-72.   Oogenesis  	169

                                    vii

-------
                         LJST OF FIGURES (continued)
 2-73	171
 2-74.   Calculation of Total Unbalanced Genotypes that are Present in
         Live-Born Individuals	173
 2-75.   The Induction of Transiocation in Mouse Post-Miotic Germ
         Cells by EO 	176

 3-1.   Acrylamide, Glycidamide	202
 3-2	205
 3-3.   AA-Protein Adducts, GLY-DNA Adducts	207
 3-4	212
 3-5.   Gene Mutations, Clastogenicity	214
 3-6.   Doubling Dose Method	219
 3-7.   REFs for Acrylamide	220
 3-8	226
 3-9	226
 3-10.   Bone Marrow Micronudeus Test After 1,3-Butadiene Exposure for
         (6h/d on 2 d, Nose Only) 	228
 3-11.   Bone Marrow SCE Test After 1,3-Butadiene Exposure for
         (6h/d on 2 d, Nose Only) 	229
 3-12	231
 3-13	231
 3-14.   Detection of Germ Cell Specific Mutagens (1991)	232
 3-15.   Cooperations	232
 3-16.   Micronucleus-Test Results 18-24 h after 1,3-Butadiene Inhalation
         Exposure	234
 3-17.   Micronuclei in Early Spermatids of Mice	236
 3-18.   Dominant Lethal Test after Inhalation Exposure to 1,3-Butadiene	238
 3-19.   Cyclophosphamide	249
 3-20.  The Metabolites of Cyclophosphamide Metabolism	249
 3-21.  Cytotoxic Metabolites of Cyclophosphamide Metabolism  	250
 3-22.  The Structures of Guanine Adducts Formed by the Reaction with
         Phosphoramide Mustard	252
 3-23.  Cyclophosphamide: Effect of the Dose Fractionation on the
         Frequency of Dominant Lethals Induced in Mice	265

4-1.   Reciprocal Translocations Induced in Mouse
         Germ Cells by EO 	278
4-2.   Dominant visible Mutation Data for the Following
         Ethylene Oxide Exposure	282
4-3.   Calculation  of Dominant Mutation Rate in Humans
        Following  Ethylene Oxide Exposure	283
                                    vw

-------
                         LIST OF FIGURES (continued)
 4-4.    HPRT Mutants in Mouse Splenocytes, HPRT Mutants in Human
         Splenocytes	285
 4-5.    Dominant Visible Mutation Data for the Following
         Ethylene Oxide Exposure	288
 4-6.    Calculation of Dominant Mutation Rate in Humans
         Following Ethylene Oxide Exposure  	289
 4-7.    HPRT Mutants in Mouse Splenocytes, HPRT Mutants in Human
         Lymphocytes  	290
 4-8.    A Parallelogram for Predicting Mutation induction in
         Human Gene Cells Following Ethylene Oxide Exposure  	292
 4-9.    The "Perfect" Data Set for Genetic Risk Assessment by the
         Parallelogram Approach	302
 4-10.   Specific Differences in the Metabolism of Butadiene In Vivo  	322
 4-11.   Cartinogenicity of BD	324
 4-12.   Interspedes Differences 	325
 4-13.   HPRT Mouse Splenocytes  	326
 4-14.   Human Monitoring of BD  	327
 4-15.   HPRT Humans	328
 4-16.   Genetic Risk Estimates for cGy of Paternal Exposure to X or
         Gamma Radiation Made by the Direct Method  	352

 5-1	358
 5-2	360
 5-3.      ..'.	360
 5-4.    Sertoli Cell Barrier	377
 5-5	379
 5-6.      	384
 5-7.    OPP Mutagenicity Test Guideline   	384
 5-8.    OTS Mutagenicity Test Scheme 	386
 5-9.    How Was the Database Assembled?	387
 5-10.  Germ Cell Mutagen Tests	387
 5-11.  Short-term Test Results for Agents Positive in More than One
        Germ Cell Assay	388
 5-12.  Short-term Test Results for Agents Confirmed Positive in One
        Germ Cell Assay	390
 5-13.   Positive Germ Cell Vs. Somatic Cell Tests	391
 5-14.   Short-term Results for Agents Negative in More that One Germ
        Cell Assay  	392
 5-15.   Short-term Results for Agents Negative in One Germ
        Cell Assay  	394
5-16.   Negative Germ  Cell Vs. Somatic Cell Tests	394
5-17.   Performance of Somatic Vs. Germ Cell Tests for Agents
        Confirmed in Germ Cell Studies	396

-------
                       LIST OF FIGURES (continued)

Figure                                                             Page

5-18.  LEDs of Germ Cell Mutagens Tested in CBA Vs MVM	398
5-19.  LEDs of Germ Cell Mutagens Tested in CBA Vs MVM	398
5-20.  LEDs of Germ Cell Mutagens Tested in DLM Vs MHT	399
5-21.  LEDs of Germ Cell Mutagens Tested in DLM Vs MVM	399
5-22.  LEDs of Germ Cell Mutagens Tested in MHT Vs SLT  	401
5-23.  Agents Positive in Mammalian Germ Cell Tests	402

6-1.   Catch 22 Situation 	416
6-2.   Consequences	416
6-3.   Summary  	419
6-4.   Application of Parallelogram to Risk Assessment 	422

-------
                              LIST OF TABLES
 1-1.   Questions Posed to Working Groups and Answers	  6

 2-1.   Adduct Levels in Rats Exposed to EO	118
 2-2.   Human Data, Hb Adduct Levels from Ethylene Oxide  	119
 2-3.   Best Values for Parameters and for Increments of Adduct Level or
         Dose per Unit of Exposure Dose or Absorbed Dose	121
 2-4.   Annual Dose of EO in the 8.5 Million Population of Sweden  	136
 2-5.   Excess Risk per 10,000 Live Births	155
 2-6.   Death of Conceptuses and Defects among Living Fetuses 	157

 3-1.   Cytogenetic Studies and Point Mutation Studies in
         Cells of Humans Exposed to Ethylene Oxide 	187
 3-2.   Micronucleus Studies in Peripheral Blood Lymphocytes or
         Other Types of Cell*	188
 3-3.   Sister Chromatid Exchange (SCE) Studies in
         Peripheral Blood Lymphocytes	191
 3-4.   Acrylamide Exposure  	216
 3-5.   Genotoxicity of 1,3-Butadiene In Vitro	227
 3-6.   Testing of 1,3-Butadiene in the Spot Test with (TxHT)F1
         Mouse Embryos	235
 3-7	240
 3-8.   Species  Comparison of Dose of Expoxybutene in Blood,
         Calculated from  Hemoglobin Adduct Levels, and
         Predicted Dose in the Body of Epoxybutene Following
         Exposure to Butadiene	241
 3-9.   EC/US Workshop on Risk Assessment Parallelogram/
         Chemical Matrix - Somatic Genotoxic In Vitro
         Effects of Cydophosphamide on Mammals
         Including Humans  	255
 3-10.  EC/US Workshop on Risk Assessment Parallelogram/
        Chemical Matrix - Somatic Genotoxic In Vivo
        Effects  of Cydophosphamide on Mammals and
        Human Somatic Cells	256
 3-11.  Examples of Somatic In Vivo Effects of CP in Mammals
        Induding Man	258
 3-12.  Activity of Cydophosphamide Detected in Individual
        Assay Systems	259
3-13	264
3-14.   Doubling Dose of Cydophosphamide in mg/kg For Different
        Genetic Endpoints in Mice	266

4-1.    Proposed Scaling Factors (REF) for Mouse to
        Human Risk Extrapolation	312
                                    xl

-------
                         LIST OF TABLES (continued)

Table                                                                  Paoe

4-2.    Estimate of Comparative Doubling Dose of HPRT Mutations by
        Exposure to Butadiene	336

5-1.    Specific-locus Test Results for Chemicals that have given
        Conclusive  Results in Both Stem-Cell Spermtogonia and
        Post-Cell Stages'	362
5-2.    Genetic Activity Profiles of Three Categories of Alkylating
        Agents in Visible Specific-locus Assays of Drosophila
        (vermillion locus) and the Mouse (7 loci) in Comparison
        with Tumorigenic Potency in Rodents	365
5-3.    Factors to be Considered in Assessing Gene Mutations in
        Germ Cells	:	374
                                    XM

-------
 Overview

 This document contains the Proceedings of the European Commission/ United States
 (EC/US) Workshop on Genetic Risk Assessment held October 11 -14,1993 in Durham,
 North Carolina, U.SA  The title of the workshop was "Human Genetic Risks from
 Exposure to Chemicals, Focusing on the Feasibility of a Parallelogram Approach".
       The parallelogram approach to genetic risk assessment and the workshop
 itself were proposed by the late Emeritus Professor Frits Sobels of the University of
 Leiden, The Netherlands. Professor Sobels was deeply involved in the planning of the
 meeting until his death on the 6th  of July, 1993. It is appropriate, therefore, that the
 Workshop and these Proceedings are  dedicated to his memory.
                                                i
       The objective of the workshop was to  identify the  methodology, data
 requirements, and mechanistic research to understand the human impact of germ cell
 mutagens. The European Commission and the U.S. Environmental Protection Agency
as joint sponsors of the meeting anticipate that the discussions and conclusions from

-------
the Workshop (see Synopses and Table 1-1) will be of substantial value in planning



future research to address the information requirements of genetic risk assessment.

-------
 Introduction

 This workshop was the concept of Emeritus Professor Frits Sobels of the University
 of Leiden, The  Netherlands.  The underlying  idea  of the Sobels' parallelogram
 approach is that an estimate (corrected by DMA adduct dosimetry) of the genetic
 damage in human germ cells can be obtained by measuring a common endpoint in
 human and mouse somatic cells. The main objective of the workshop was to identify
 the methodology, data requirements, and mechanistic research needed to understand
 the human health impact of germ cell mutagens.
        Four chemicals were selected for review at the meeting: ethylene oxide, 1,3-
 butadiene, acrylamide, and cyclophosphamide. The first three are important industrial
 chemicals with substantial use worldwide and, therefore, considerable potential human
 exposure.    The  fourth,  cyclophosphamide,  is   a  commonly  used  cancer
 chemotherapeutic agent.
       This first  EC/US  workshop on risk assessment was highly focused on the
 feasibility of the parallelogram concept to estimate  potential germ cell effects in
 humans. It represented an evaluation of current knowledge and the identification of
 future research needs for a more precise assessment of human genetic risks from
 exposure to mutagenic chemicals.
       Following introductory remarks by Dr. Lawrence Reiter of the US Environmental
 Protection Agency (EPA), the workshop was dedicated by Dr.  Heinrich Ott of the
 European Commission (EC) to his and our friend, the late Professor  Frits Sobels.
Professor Sobels was the initiator of the workshop, and he was deeply involved in its
planning, until his death on the 6th of July 1993.

-------
        Some years ago, Sobels (1977,1982) began to formulate strategies applying
 DNA adduct dosimetry to improve the assessment of risks for genotoxic chemicals,
 particularly heritable risks.    The  underlying idea of  the  Sobels' parallelogram
 approach (Figure 1 -1) is that an estimate of the genetic damage in human germ cells
 can be obtained by measuring a common endpoint in human and mouse somatic
 cells, (such as gene mutation in lymphocytes), and in germ cells of mice, the desired
 target tissue inaccessible  in humans.

 Workshop Organization
 The workshop was organized into plenary and working group sessions.  The plenary
 sessions began with a general introduction to the parallelogram concept and genetic
 risk assessment, and continued with an examination of mutagenicfty data on ethylene
 oxide (EO) as an example. Papers were presented on the research background for
 EO, a draft risk assessment and a re-evaluation of the EO risk assessment.  Following
 overview presentations by each of the four chemical working group chairpersons, the
 working groups were convened to carefully  review the existing data for each
 compound.  Working group progress reports were presented in plenary by each of
 the four rapporteurs.  After two rounds of  review and  analysis, the rapporteurs
 concluded their reports  with answers to a series of questions (Table 1-1) that had
 been posed to each working group.  Finally, the workshop participants heard four
crosscutting papers representing current research on germ cell mutagens and future
research directions in genetic risk assessment.

-------
                  The Sobel's Parallelogram
Mouse Somatic Cells
 Mouse Germ Cells

 Measured Mutations
    and Adducts
Measured Mutations  ^
    and Adducts
Human Somatic Cells

Measured Mutations
    and Adducts
 Human Germ Cells

Estimated Mutations
                    Comparisons <	^

                       Estimates	>•
                          Figure 1-1

-------
Table 1-1.  Questions Posed to Working Groups and Answers.
1)     Which  are the most important studies  available to support  genetic risk
       assessment activities?
ACRYLAMIDE (AA)
EPA grout worker
study.
OSHA occupational
exposure study.
Ingestion through
drinking water.
BUTADIENE (BD)
Germ cell studies.
Human exposure data
CYLOPHOSPHAMIDE (CP)
Mouse heritable
translocatlon, specific locus
and
malformation data.
Human exposure data
ETHYLENE OXIDE (EO)
Mouse germ cell data
(Lewis, et al., 1986).
Somatic cell data
(Walker and Skopek,
1993).
Human data (Fates et
al., 1991).
1 a)    What endpoints (including other than genetic ones) are considered "adverse
       effects" from a risk assessment perspective?  Congenital malformation and
       infertility data, spontaneous abortions, tumor incidence.
2)     Identify inadequacies and gaps in the data base for genetic risk assessment.
ACRYLAMIDE (AA)
Human somatic cell
response needs more
definition with better
largei ussue oooiinoiiy
data
BUTADIENE (BD)
Need metabolism and
pharmacokinetics data
before proceeding with
genetic studies.
CYLOPHOSPHAMIDE (CP)
Need adduct data in mouse
and human somatic cells
and in mouse germ cells.
Human dose-response data
are inadequate.
ETHYLENE OXIDE (EO)
Need better human
exposure data,
additional somatic
mutational
endpoints.
Need doslmetry and
additional endpoints in
all germ cell stages.
                                     6

-------
Table 1-1.  Questions Posed to Working Groups and Answers (Cont).

3)      Is the necessary dosimetry data available?  If not, what else is needed?
ACRYLAMIDE (AA)
No. Need the same
biological endpoints and
dosimetry In humans
and In animals.
Need better exposure
assessment in humans.
BUTADIENE (BD)
No. Target dose and
pharmacokinetlcs
data needed.
CYLOPHOSPHAMIDE (CP)
No. Good target dose data
needed.
Need gene mutation data in
human cells and aneuploidy
data In mouse germ cells.
ETHYLENE OXIDE (EO)
No. See 2. (above).
4)      What are the major assumptions upon which the genetic risk assessment is
        based?
    ACRYLAMIDE (AA)
   BUTADIENE (BD)
CYLOPHOSPHAMIDE (CP)
ETHYLENE OXIDE (EO)
  Mouse equals man; all
  other assumptions are
  inherent In this one
  major assumption.
Rodent DMA response
Is equivalent to human
DMA response.

Equivalent background
and Induced mutation
rates In animal model
and humans.

Equivalent germ cell
specificity in mouse
and human at all
Assume linearity of dose
response; similar sensitivity
of mice and humans;
similar dose responses and
levels of cytotoxictty;
relative sensitivity somatic
cell types; relative sensitivity
of males vs. females; high
to low dose; similar dose
rate effects; background or
"spontaneous" mutations;
similar synerglstlc effects.
Similar sensitivity of
mice and humans,
somatic and germ cells.

No dose rate effects.

Equivalent sensitivity of
males and females.

Assume linear high to
low dose extrapolation.

Assume that HPRT
mutations assayed are
representative of the
genome as a whole.

-------
Table 1-1. Questions Posed to Working Groups and Answers (Coot).

5)    What are the major uncertainties in the risk assessment?
ACRYLAMIDE (AA)
Similar extrapolation
factors; target
dosimetry;
Induction kinetics for
different endpoints.
BUTADIENE (BD)
Extrapolations; for
doses, dose rates
and species.
CYLOPHOSPHAMIDE (CP)
All of the assumptions.
ETHYLENE OXIDE (EO)
All of the assumptions.
6)    What  research  is  required  to  address the  major  assumptions  and
      uncertainties?
ACRYLAMIDE (AA)
Demonstrate a
human germ cell
specific mutagen.
Elucidate risk
extrapolation factors.
Improve
pharrnacokinette
modelling for animal
models and for
humans.
BUTADIENE (BD)
Confirm occupational
exposure study (HPRT
variant frequency)
using another method;
Research needed
on human, mouse and
rat metabolism, DMA
relationships, and
mutation spectra
Induced by BD and
metabolites.
CYLOPHOSPHAMIDE (CP)
Adducts, human data;
female data required.
ETHYLENE OXIDE (EO)
In addition to 2. (above);
research Is needed to
validate assumptions,
need work on
retrospective
dosimetry (particularly
for reciprocal
translocatlons using
hybridization
techniques).
      Are you able to establish a priority for the needed research?  No.
                                    8

-------
 Table 1-1. Questions Posed to Working Groups and Answers (Con?).

 8)     Is  the  parallelogram  approach   applicable  to  the  compound  under
        investigation?  If not,  why  not?  What alternative approaches appear more
        suitable?  Ust pros and cons of each.
     ACRYLAMIDE (AA)
    BUTADIENE (BO)
  CYLOPHOSPHAMIDE (CP)
  ETHYLENE OXIDE (EO)
   Yes/no. Conceptually,
   It is a very useful tool to
   focus needs for
   extrapolation to
   humans; many
   components are
   already In use In other
   risk assessment
   methods, do not need
   all four comers to make
   progress.
  Not yet. There Is too
  much missing data
  and It Is not yet clear
  which Is the
  appropriate animal
  model, If It is not the
  mouse, then
  reconsideration is
  required on the
  appropriate endpoints
  to study.
 Yes, It is possible If some
 major assumptions are
 made out too many parts of
 the parallelogram are
 missing; there are not
 enough data
 Yes, It is possible to
 construct the
 parallelogram but the
 group was
 uncomfortable in doing
 so because of missing
 data
 9)      How would one address the problem of "no human germ cell mutagens*?
    ACRYLAMIDE (AA)
   BUTADIENE (BD)
CYLOPHOSPHAMIDE (CP)
 ETHYLENE OXIDE (EO)
  Perhaps an
  epidemtologlcal study
  showing a positive link
  between exposure to a
  mutagen and a birth
  defect
Need more btomarker
studies in an
epidemiologies)
approach in humans;
but for BD It Is too
early to attempt an
epidemtotogical
approach; the
problem is not'no
germ cell mutagens', ft
Is 'no germ cell
mutagens
demonstrated'.
Perhaps radiation and
cytostatlcs are human germ
cell mutagens; need to
show Inherited damage, but
It Is satisfactory to
extrapolate from animal
models with good
comparative mechanistic,
metabolism and
pharmacoklnetlcs data
The problem Is not 'no
germ cell mutagens', It
Is, 'no germ cell
mutagens
demonstrated'; the best
way forward is to look at
cancer Incidence in
progeny; there are
indicative
epidemtologlcal data
worth pursuing.
10)    What would you consider an "ideal data set"?
       assumptions validated.
                                   All four comers filled and
                                          9

-------
 Synopses

 The following sub-section provides a synopsis for each of major sections of the
 conference: Plenary sessions, Data Summaries of Compounds, Rapporteur's Reports,
 Crosscutting Remarks, and Concluding Remarks.

 Plenary Sessions

 Alan Wright introduced the parallelogram concept, prefacing his  remarks with an
 overview on genetic risk assessment:  "An appreciation  of the mechanisms of
 genotoxic action has permitted simplifications in correcting for the concerted action
 of factors (related to dose and effect) by allowing correction to be applied  in two
 stages. The first stage compensates for differences in determination of the  critical
 dose which, in turn, determines the rate of formation of the promutagenic lesions in
 DMA.  The second stage involves compensating for differences in the operation of
 factors to determine the progression of the key promutagenic lesions into mutations."
       In its original form, the concept known as the parallelogram approach was
 developed by  Sobeis (1977) to estimate mutations in rodent germ cells, by
 determining mutagenic potency  in  rodent somatic  cells in vitro and  applying
 measurements of the amounts of specific DMA adducts, formed in the in vitro test
system and in mouse germ cells in vivo.  [Note that Brewen and Preston (1974,1975)
presented a mouse-human, somatic-germ cell proportionality approach for radiation-
induced chromosome aberrations and were able to confirm the human germ cell data
using cells from irradiated human testes].
                                     10

-------
        Evidence to  support the contention that measurements of specific DMA
 adducts may provide a basis for extrapolating mutagenicrty data from one cell type
 to another, within an individual or species or even between species, was reported by
 van Zeeland  (1985).  Very  similar mutation  frequencies were observed when
 measurements of 06-ethylguanine in DNA were used as a basis for calculating the
 mutagenic potencies of ethyl nitrosourea in inducing HPRT mutations in V79 cells and
 specific locus mutations in mouse spermatogonia.
        In attempting to estimate mutation frequencies in human germ cells, according
 to the extended parallelogram model  (Sobels, 1982, 1989), a common genetic
 endpoint is measured  in the somatic cells of human and  an experimental animal.
 Then, the subsequent determination of the potency of the agent in the germ cells in
 the experimental model permits the ratio of somatic to germ cell mutation to be
 calculated in the experimental model. Provided this ratio holds for humans, one can
 then calculate the risk to man.
       The key assumption in the extended parallelogram model, of course, is that
 the ratio between somatic and germ cell mutation is the same  in the experimental
 model and in man.  As Wright pointed out, given the marked variations in gene
 expression between tissues and species, there is no theoretical justification for this
 assumption [Note, however, the previous work of Brewen and Preston (1974,1975)].
       Thus, the validation of the extended Sobels' parallelogram model depends on
 producing convincing evidence that the relative mutagenic effectiveness of a given
 genotoxic agent towards target somatic cells,  (for example, peripheral  blood
 lymphocytes,) and germ cells in the test species parallels that in humans over a broad
dose range.  This may  not be an easy task since, in practice, when environmental

                                     11

-------
 exposures are reduced to very low levels, the small incremental effect caused by a
 specific chemical may no longer be detected within the relatively high and variable
 background of mutation.
        Furthermore, according to Wright, one must consider that, for any given dose
 of a genotoxic chemical, the principal determinants of interspecies extrapolation
 constants are the rates and fidelity of DMA repair and replication.  Intuitively, one
 assumes proportionality or  parallelism between the repair of damage induced by
 different agents, in the same  organism.  However, high dose exposures typically
 employed in experimental models may lead to rapid saturation of repair systems and,
 therefore, to results which may not adequately reflect the role of repair in determining
 mutation, especially  at the low dose exposure levels operative in  the human
 environment.
       Wright pointed out that a number of approaches have been proposed and
 developed to compensate for species differences in DMA repair and replication; one
 notable approach is the radiation  equivalence concept which is also designed to
 assess cancer risks.  However,   all  of  these approaches are  based  on  the
 determination of relative mutagenic potencies, and they all necessitate proportionality
 or parallelism between the repair of damage induced by different agents, in the same
 test system  or, alternatively, the same agent in different tissues of an individual or
 strain. Also, accurate assessment of the dose  of the ultimate mutagen delivered to
the target DMA is fundamental to all of these approaches.
       Wright concluded, "I think in  view of the potential  limitations  of the
(parallelogram) approach, it would  be prudent to explore alternative strategies to
correct for species differences and (differences in) the progression phase of mutation."

                                      12

-------
        David Brusick summarized the recent efforts of the International Commission
 for Protection against Environmental Mutagens and Carcinogens (ICPEMC) in genetic
 risk assessment.  He noted that Frits Sobels was one of the founding fathers of
 ICPEMC and was the organization's first chairman.
        ICPEMC was asked by Health and Welfare Canada and by the EPA to examine
 whether risk data in mice can be extrapolated to human populations so as to describe
 the impact of exposure to germ cell mutagens in terms of the number of probable new
 mutations in the F, population. Brusick described the process used by ICPEMC (for
 EO and acrylamide) for determining from the animal data the appropriate scaling
 factors, the assembly of human  population exposure  data and the exposure
 assessment, the calculation of the risk using the direct or the doubling dose method,
 and finally, estimation of (based on the exposure) the induced genetic disease in the
 F,  population.   The assumptions employed by ICPEMC in the reports (ICPEMC
 1993a,b)  described by Brusick played  an important role in the discussions that
 followed during the meeting.
       Kerry DearfiekJ described the risk assessment guidelines of the EPA Office of
 Prevention,  Pesticides, and Toxic Substances (OPPTS) (USEPA I986a,b) and
 particularly its Mutagenicrty Testing Guidelines (Dearfield, 1991).  He discussed the
 assay methods used by the Agency to evaluate the capacity of the chemicals to alter
 genetic material and the way in  which these findings  are incorporated into an
 assessment of risk of heritable genetic alterations in humans. While not explicitly
employing the parallelogram method, he said, "We have utilized many of the concepts
of the parallelogram in our regulatory battery."
                                    13

-------
        Reflecting  on the  EPA's draft risk assessment for EO,  he observed that
 "OSHA's permissible exposure level (PEL) is one ppm. At one-half of that, which is
 a very real occupational exposure, there are going to be an additional three persons
 in  10,000  having. . .heritable  translocations.  How  do  you communicate that
 risk? . . .You can also say that 3 x 1CT4 may be included in the variability of the
 background.  The background is 19 x 10"*. So, how do you detect 3 over 19 in the
 human population?  More importantly, how does this number '3' get translated into
 a phenotypically-expressed adverse health consequence?" Finally, Dearfield made
 a strong point of the fact that EO shows  reproductive and developmental effects as
 well as  neurotoxicity, all of which may be related to the electrophilicity and/or the
 genetic activity of the chemical.
        Lars Ehrenberg presented a research perspective  on the EO exposure
 assessment and target tissue dosimetry (via hemoglobin  and  DMA adducts).  He
 pointed out that measuring the adduct level in hemoglobin, as  a surrogate dose
 monitor, has the advantage that the molecules are not repaired: They are more long-
 lived, and from the steady state level of adducts from long-term exposed persons, we
 can estimate the contribution for the adduct level per unit time, and then evaluate the
 dose as for DMA	There is a tendency to overestimate exposure doses with the
 effect that the calculated in vivo dose per unit of exposure dose  becomes too low.
 The tests where we get the higher figures are from those studies where we have the
 most careful measurements  of exposure dose."   In describing  dominant lethal
 mutations as a function of EO exposure from Waldy Generoso's work (Generoso, et
 al., 1990), Ehrenberg pointed to a strong increase in lethality at higher doses.  "What
could be the  causes of this?   There could be saturation of detoxification and
                                     14

-------
 saturation of DMA repair to make the curve turn upwards. There could be induction
 of detoxification that will turn it the other way. Induction of repair could affect the curve
 at very low doses	You see that the curve drops at very high doses because of
 the damage to sensitive cells."  Thus,  Ehrenberg provided a specific example of
 concepts discussed by previous speakers.
        Vicki Dellarco described the EPA Draft EO risk assessment (Rhomberg et at.,
 1990) based on heritable translocations in mice (Generoso et al.,  1990). The project
 was intended to develop a framework for quantitative genetic risk assessment, to
 highlight the  issues  and  uncertainties, and  to promote discussion  on data
 requirements and future research directions.
        In discussing the assumptions used by the EPA, she said, "Ethylene oxide
 dominant lethals (in mice) are induced in mid to late spermatids to early spermatozoa.
 The window (in the mouse) is 10 days. So,  one major assumption we are  making is
 that there is not a stem cell effect. That is a critical assumption, because if  there was
 a stem cell effect, even  after exposure had ceased, the individual would be at an
 increased risk for his entire reproductive life;  whereas with a post-stem cell effect,
 increased  risk is only a function of recent exposures.  Risk would recede with
 termination of exposure."
       She pointed out that EO concentrations  in tissue are  in balance  with a
 constant external air concentration. "Assuming proportionality at low to  moderate
 doses isn't such a bad assumption	Where our approach fails is looking at risk
 from episodic high exposures, given the known dose rate effects of EO	So, for
 the ppm-hour dose-metric, we have to assume that response is a function of exposure
times time.  We have to  determine the species equivalence of dose in mouse and

                                      15

-------
 man. The assumption is that the probability of inducing a heritable translocation is
 a function of adduct load irrespective of the species; also, the ratio of ppm-hour of
 external exposure to the concentration times time product, in the testis, is the same
 for mice and man."
        In discussing the dose-response modelling for EO, she said, "We are assuming
 (that the target is) DMA.  If it is protamine, then one may chose to select different
 mathematical models to fit the observed data on translocation frequency."  Another
 point emphasized by Dellarco was that it was necessary to assume that males and
 females are  at similar risk. She said, "This is a weak assumption given that the male
 is  unlikely to be a reliable predictor of female risk. More studies are needed in
 females to assess population risk.  Finally, the zygote should  be included in an
 assessment of genetic risk.*
        Julian Preston introduced a number of mechanistic considerations related to
 the EPA risk assessment on EO and to genetic risk assessment in general.  He
 pointed out that the probabilities in differentiating spermatogonial cells for chemical
 induction of a mutation or a translocation will be related to the total cell cycle and the
 proportion of that cell cycle that is in DMA synthesis phase. "Although you have a
 higher probability of producing a mutation or a translocation in a rapidly cycling cell,
 you have a higher probability  of  killing  that same cell, because  of a secondary
 event	In the case of the stem cell, the probability of getting a mutant cell either
 with a translocation or a mutation is rather high relative to cell killing.11
       Preston said that he believes that one should strongly consider the stem cells,
especially under chronic exposure conditions. He went on to say that for post-meiotic
germ cells,  the next S phase occurs following fertilization. Thus, there is a long period

                                      16

-------
 of time to accumulate damage (several days), and there is no repair taking place in
 these cells during this time.  The repair takes place after fertilization occurs. Preston
 said, "There is a race between repair of damage induced  in this cell stage and
 replication at the S phase. You  have to divide up the spermatogenic cell cycle into
 the different components; and then relate those to exposure, or to effective dose to
 those particular cell stages, based upon duration within the spermatogenic cell cycle."
 He also asserted that there are likely to be different probabilities of mutation across
 species, as related to the spermatogenic cell cycle.
        Discussing similar considerations for oogenesis in the female, he said: "For
 translocations and for mutations, there would be a rather low probability of induction
 for the female.  The higher probability, one would anticipate, would be for aneuploidy
 events."
        Regarding translocation carriers, Preston argued, "I consider only the ones
 that don't result in spontaneous abortions as a component of genetic risk."  Preston
 calculated that only 9% of the unbalanced genotypes  result in a viable, but
 unbalanced offspring. With regard to the modelling of the dose response curves,
 Preston said, "I would argue that at low exposures, it would really be most applicable,
for translocations, to use something which is very much a dose squared component
curve. That might not be the same as the rest of the curve, at high exposures; but
at low exposures, you need two lesions to produce a translocation, and you can well
argue that those are produced as independent events, because you need replication
errors on two chromosomes."
                                     17

-------
 Compound Reviews
 Nat Natarajan provided a general overview of the published data base on EO, as did
 Kerry Dearfiekf for Acrylamide (AA), Use-Dore Adter for 1,3-Butadiene (BD) and Diana
 Anderson for Cyclophosphamide (CP). Space does not permit reporting of these
 presentations.  However, it is anticipated that the Work Group reports on the risk
 evaluations for these compounds will be published in the near future in  Mutation
 Research Reviews in Genetic Toxicology. The overall conclusions of the Working
 Groups regarding risk evaluations are captured under the Rapporteur's Reports that
 follow below for each of the four compounds.
 Rapporteur's Reports
 Ethytene  Oxide (EO).  The EO Working Group consisted  of Nat Natarajan, as
 Chairman, Vicki Dellarco, Lars  Ehrenberg, Waldy Generoso, Susan Lewis, Julian
 Preston, and Ad Tates.
       Julian Preston, as rapporteur, discussed the  heritable translocation risk
 estimate for EO with regard to the applicability of the Generoso et al. (1990) data set
 and the necessary additional data for purposes of genetic risk estimation.
 The working group decided there were two factors that were of importance, that led
 to a specific conclusion on the utility of the heritable translocation assay.  One factor
 was Lars Ehrenberg's calculation of dose, and the second one was the independence
 of spontaneous and  EO-induced lesions.  Ehrenberg calculated the target dose to
 DNA, which flattened the curve at the low exposure levels and reduced the effect. For
reciprocal  translocations, as a function of dose to the DNA, this further reduced the
                                     18

-------
 risk estimate from the linear extrapolation on the ppm curve discussed by Vicki
 Dellarco.
       The group also discussed the second point in some detail, whether to use an
 additive assessment of slope. That is, do spontaneous lesions interact with the EO-
 induced ones, or are these independent events? They concluded that because of the
 very low probability of having a spontaneous event, they would consider those as
 additive and not interactive.  Again, these decisions reduced a linear response to a
 curve  that is convex  downward, and,  again, reduced the estimated effect at low
 exposure levels.
       The conclusion from these two points was that genetic risk from  reciprocal
 translocations at low exposures and for post-meiotic cells (for which the data set was
 collected), would be extremely low.
       There was  no new data  on gene mutations in germ cells, so  the group
 examined some data that was published  by Susan Lewis and her colleagues on
 electrophoretic variants and dominant visible mutations  in the mouse (Lewis, et al.,
 1986).  The exposures covered  the entire spermatogenic cycle and yielded four
 dominant mutations in 1891 offspring. Two were heritable translocations.
       Using an average exposure time for the four mutants (in fact, the first one
 showed up at around 60,000 ppm hours), 20 phenotypes (50 loci) scored and an
 assumed 2000 similar dominant loci in the human, the estimate was  160 dominant
visible  mutations in humans, assuming that  all sensitivities are equal between the
species. That would be 160 mutations times 10"5 per ppm hour, or one mutation per
1600 individuals  per year of exposure, based upon a one ppm exposure.
                                     19

-------
       Preston: "We took the one ppm exposure and converted that into dominant
 visible mutations	We progressed from that germ cell calculation, on the only data
 that we  could  do any manipulative calculation with, and moved into somatic cell
 studies,  to try and link mouse with man by another approach."
       The group assumed that from the somatic cell data, for HPRT mutations in
 mouse and man, they could use a sensitivity factor of one.  Again, they assumed
 2,000 dominant loci, and assumed that each was  giving rise to an independent
 phenotype. From the mouse data, the group had derived the value 4.0 x 10'10 per
 locus, per ppm-hour. The human value would be equivalent, for per locus frequency
 for the total number of dominant mutations.  One would expect 8.0 x 10*7 dominant
 mutations per ppm-hour.
       Preston concluded, That is not to be taken as an extremely accurate number,
 but I think it gives the types of data sets that are available, and what you can do with
 those. We felt that was a reasonable approach to take.  Lars Ehrenberg did a quick
 comparison with radiation numbers, and it would indicate that the ethylene oxide
 number here for dominant mutations might be of the order of ten to 100 times higher
 than a radiation equivalent mutation frequency, on a per rad basis. The approach that
 we took was based upon the data we had available. We had four dominant mutations
 in the mouse study that we utilized, and the HPRT mutation in somatic cells. We
 finished UP making a decision that the  parallelogram approach could be utilized on
 those data.  And, of course, we are certainly open to discussion on that approach."
       The group  also discussed the feasibility  and appropriateness of using  the
 relationship  between tumor incidence  as a measure of relative sensitivity on  the
assumption that the initiation of a tumor, at least, is modeling mutation induction in

                                     20

-------
 somatic cells. They specifically discussed the use of leukemias, on the assumption
 that the human myelogenous leukemias are the same tumor type as in a mouse.
 They  also  considered the  appropriateness of spontaneous  abortion for risk
 assessment; the group agreed that it is clearly an important component of genetic
 risk, but could not distinguish genetic, zygotic, or cell killing responses. Preston said:
 "We still thought, it is worth considering more.. .and  if it is not a genetic component,
 then you  still need the model. . .for the zygotic data, as a separate assessment of
 risk."

 Acrylamide (AA). The AA Working Group consisted of Kerry Dearfield, as Chairman,
 David Brusick, George Douglas, Udo Ehling, Martha Moore, and Gary Sega.
       George Douglas served as rapporteur for the AA group.  With regard to the
 parallelogram approach, the group concluded that there were no human data, other
 than one  hemoglobin adduct study, that would permit entry into the parallelogram.
 The parallelogram model was, therefore, used instead to delineate what additional
 tests were required to understand the risk involved from AA.  There was discussion
 (without a satisfactory conclusion) about the possible partitioning of effects involving
 metabolites;  for example,  there  was a  question  whether glycidimide was the
 metabolite causing gene mutation, versus AA causing heritable translocations.
       The ICPEMC Report, described earlier by David Brusick, served as a basis for
the working group's risk evaluation. In general, the same default risk extrapolation
factors  were  applied, although different  numbers were  used  in  some  cases,
particularly for the exposure assessment.  The Ehling and  Neuhauser-Klaus  paper
(1992) was used for the gene mutation risk assessment; therefore, the gene mutation
numbers were based  on  a single dose point (a serious deficiency in terms of the
                                     21

-------
 exercise).  For heritable translocations, Shelby,  et  al.  (1987) and Adler (1990)
 contributed the necessary data.    There were appropriate dosage data, for the
 heritable translocations  in the ICPEMC  reports.  Risk numbers were calculated
 separately for gene mutations and chromosomal aberrations.
        Using the ICPEMC approach,  all exposure numbers were  converted to
 milligrams/kilogram/day.  The group used a doubling dose approach, and elected to
 derive different doubling dose estimates for separate data points when the dose
 response was curvilinear, as in the Adler study (1990).  In  addition, they used a
 modification of the direct method, in that recessive mutation data for the mouse was
 used  to estimate dominant mutations in humans.  At  a 100 mg/kg  dose, the gene
 mutation doubling dose was 53.  The direct estimate did not resemble the doubling
 dose estimate.
        For the chromosomal aberrations, the direct and doubling dose estimates were
 much closer.  Three different sets of estimates were made.  Based on the Shelby, et
 ai. (1987) study, the doubling  dose was  0.1  mg/kg.   For the Adler (1990) data, a
 single dose estimate of 50 mg/kg yielded a doubling dose of 3.3 mg/kg. By modeling
 the dosage from the curve representing her three points, additional numbers were
 calculated. The net result was that 531 x 10"6 new genetic diseases were predicted,
 based on a one-day exposure-obviously a very high number.
       Douglas:  "I should point out that these numbers do not include  are any
factoring of tissue  accessibility, which is something that was a consideration in the
ethylene oxide analysis that has been published by EPA. Female exposure was not
considered at all. The numbers are based on daily exposure. They are not based on
time-weighted averaging, or reproductive life span of the people exposed	I would

                                     22

-------
 say that I am comfortable with the.. .default assumptions.  But I would be a lot more
 comfortable if we had  empirical  data to tell us, in the  case of AA, what these
 extrapolation factors are."
 Butadiene (BD).  The BD Working Group was chaired by Ilse-Dore Adler, and the
 members of th group were Judi Cochrane, Siv Osterman-Golkar, Thomas Skopek,
 Marja Sorsa and Ekkehart Vogel.
        Marja Sorsa, rapporteur for the group, reported that a special problem with BD
 is its metabolism and the specific metabolism in the animal germ cell stages  as
 compared to metabolism in humans. The mutagenic activity of diepoxybutane, in
 various  studies  in mice, seems to be from  10 to 100 times higher than for the
 monoepoxybutane.   And there  are additional putative  reactive intermediates:
 dihydroxyepoxybutane,  crotonaldehyde,  and hydroxymethytvinylketone.    Siv
 Osterman-Golkar informed the group that the presence of the metabolite M-1 [or 1,2-
 dihydroxy-4-(N-acetylcysteinyl)butane] in urine from humans exposed to BD suggests
 that hydroxymethyMnylketone is one of the reactive metabolites in humans. As for
 the parallelogram  approach at the  endpoint level, there were three  cytogenetic
 studies,  one published and two unpublished,  which seemed to confirm that, at the
 exposure level encountered in the manufacturing industry (generally below the three
 ppm level), there was no response.
       In mice, however, the lowest effective doses are surprising low, such that both
for micronuclei, as well as SCE, only a week's  exposure at 6.25 ppm was needed for
a positive response.  In rats, the only data that suggested a positive response was
for SCE, with a positive response at 500 ppm.  There was no response in rats in the
case of chromosome aberrations or micronuclei at doses as high as 1300 ppm five
                                    23

-------
  days per week. For germ cells, there were dominant lethal results in mice at two dose
  levels  (and two different laboratories), and a heritable translocation experiment is
  presently ongoing at the same exposure levels in Adler's laboratory.  Based on the
  available cvtoaenetic data,  it was not possible  to  continue the  parallelogram
  calculations for cytogenetic effects.
        The group confirmed that two studies on gene mutations in rodents in vivo are
  about to be published.  Fortunately, Cochrane and Skopek's data (1993a, 19935)
  were available to the group (the study is positive for HPRT mutations in mice).  In the
  Cochrane and Skopek data, the lowest dose was 625 ppm.
        In humans, there was only one study identified  by the group, from  BD
  manufacturing (Legator et al., 1993). The number of workers in this study was very
 small, but the exposure assessment provided a data base on ambient concentrations
 in the production facility that suggests that time-weighted-average values do  not
 increase above three ppm, and that the general ambient levels are between one and
 three ppm. Additionally, there was a measurement of the specific M-1 metabolite that
 allowed individual correlation of the amount of the specific BD metabolite in the urine
 samples of these workers, with the  HPRT variant frequencies in their lymphocytes.
 The correlation was excellent. There were control persons both from outside the
 facility, and persons working in the BD factory, but not in production. Furthermore,
 all of the individuals in the study were nonsmokers.
       Unfortunately, DMA adduct studies were not yet available for humans.  Siv
 Osterman-Golkar's preliminary results (Osterman-Golkar, 1993), showthat hemoglobin
adducts of the  monoexpoxide can be measured in humans. These  adducts were
measurable in workers from the Texas manufacturing plant mentioned above. The

                                     24

-------
 monoexpoxide adducts  have also  been determined in rodents.  For equivalent
 exposures, the adduct levels are three to five times higher in mice than in rats.
        Marja Sorsa indicated that much of the material available to the group is now
 in press in the Journal of Occupational Medicine. [This  is a meeting  report on the
 International Symposium on Health Hazards of Butadiene and Styrene, held in Espoo,
 Finland, 18-21 April 1993]. Sorsa reported that the meeting was concerned primarily
 with BD as a carcinogen: The lowest effective dose for cancer in mice is only six
 ppm.  Rats are far less sensitive, and the main carcinogenic response  is in the
 mammary gland. Human cohort studies have shown a slight effect on hemolymphatic
 cancers.   The most heavily discussed human  studies showed a large excess of
 leukemia.  (ARC, in 1992, classified BD as a probable human carcinogen in Group 2A.
 Sorsa:  There were several  presentations at the Espoo meeting concerning the
 metabolic rate of elimination as the key issue in the difference between mice  and rats.
 And I think you can safely state, that BD monoepoxide is  higher in mice than in rats.
 ...  Very low levels of the monoepoxide were reported to be effective  in the mouse
 bone marrow. The role of other metabolites is still uncertain."

 Cyctophosphamkte (CP). The CP Working Group consisted of Diana Anderson, Jack
 Bishop, Colin Garner, Patricia Ostrosky, and Paul Selby.
       Paul Selby served as the rapporteur of the group and began his presentation:
 "It seems  clear  that with cyclophosphamide,  metabolism needs to be better
 understood; there are  some  peculiar findings that have been found after certain
treatment regimes. Possibly some of that might result from differences in metabolism
over time and that needs to be pursued— The information on adduct formation for
this chemical is extremely complex; it is not clear what adducts should  be measured
                                     25

-------
 in man. Apparently, there are no data in man or at least nothing more than sketchy
 data. Adducts have been studied in mammals, but, curiously, not in reproductive
 cells	Clearly, pharmacokinetic information is vital to any risk estimate that might
 be made  and cell stage specificity has to be  taken  into consideration.  There are
 problems  with high dose/low dose extrapolation. In this situation, most of the people
 who are exposed, are exposed in many fractionated exposures so low level exposure
 would not be very common in the population."
        Approximately 200 studies have been carried out using CP on somatic cells,
 although most of these have focused on clastogenic effects.  There is relatively little
 information on induction of gene mutations in vivo in somatic cells; there is one study
 in humans and one in experimental animals that reported positive results.  Selby
 pointed out that more information was needed on point mutations in in vivo studies
 using methods that measure, for example, hemoglobin adducts, HPRT mutations, and
 glycophorin variants.
        Many studies have shown that dominant lethals and heritable translocations,
 as well as specific locus mutations, are induced by CP.  For heritable translocations
 and specific locus mutations, the effects are seen primarily in the second and third
 weeks following treatment. For specific locus mutations there is no suggestion of any
 effect  in spermatogonial stem cells.
       The specific locus test is one of the most important sources of information and
 represents a  precise method  of  measuring mutations in  all germ cells.  From
 molecular genetics, knowledge is rapidly expanding about events at the DMA level
 near the seven genes monitored in the test. But a major problem is how to relate
these  recessive mutations to clinical effects  in first generation  offspring?   With

                                      26

-------
 clastogenic effects, the heritable translocation test also has much historical data. But
 it is also difficult to relate translocations to damage in first generation offspring.
        Selby:   "A figure of 9%  (congenital malformations) was given the  1986
 UNSCEAR Report (UN, 1986)	But the data, many of which were from humans,
 are incredibly soft	One quite accurate figure is that approximately 8% of live born
 humans will have a serious abnormality by the time they are 25 years old; not all of
 those are genetic, but many of them do have genetic components."
        In the data of  Jenkinson  and Anderson (1990), with a treatment that was
 approximately 4-5 milligrams CP/kilogram/day, dominant lethality was found, and a
 very similar pattern was found for congenital malformations. According to Selby, "If
 you subtract control from experimental, then you get to a plateau where about 9% of
 the offspring are exhibiting quite striking effects.... One of the difficulties is that it
 is dealing with a rather small array of different abnormalities	Of the 9% induced,
 about 60% of them are runts, growth retardations	About a third of the animals
 with  these  abnormalities  were  demonstrated  to  have  gross  chromosomal
 abnormalities....  But there is very little understanding of whether this is really of
 clinical relevance in many of the cases."

 Working Group Summary

 In summary, and as reported in Table 1-1, each of the working groups was able to
 make significant progress in applying a parallelogram approach to the available
 experimental data for mutagenicity. Additional endpoints considered "adverse effects"
from a risk assessment perspective included infertility, spontaneous abortions, tumor
incidence,  and, especially., congenital malformations. The importance of addressing

                                     27

-------
 all  relevant risk factors was emphasized repeatedly.  The need for better human
 exposure and target dosimetry data, especially in mouse germ cells, was consistently
 mentioned. The crucial assumption for genetic risk assessment, that "mouse equals
 man," also remains the major uncertainty. The working groups were unanimous in
 concluding that  The problem is not  'no germ cell mutagens,' it is, 'no germ cell
 mutagens demonstrated'; the best way forward is to carefully examine genetic and all
 other related 'adverse effects' in progeny of humans (and in experimental animals)
 exposed to  mutagenic agents." Clearly, there are indicative epidemiological and
 experimental data worth pursuing.

 Crosscutting Papers
 Four crosscutting papers were presented on the final afternoon of the meeting. These
 papers were intended to provide perspective on the topic of genetic risk assessment
 in the context of the present research knowledge base and future research directions.
 Since it is anticipated that these papers will be submitted for publication separately,
 they will  not be discussed in detail here.
       The first paper was entitled "DMA Damage and Repair in Somatic Cells and in
 Germ Cells in vivo" by E.W. Vogel and AT. Natarajan. In this presentation, Ekkehart
 Vogel compared the activity of germ cell mutagens in Drosophila vs. the mouse vs.
 cancer in  rodents,  with regard  to cell (stage) specific  molecular adducts, and
 mechanisms of damage and repair. Following was a report on "Chemically-induced
 Mutation: Comparative Outcome in Somatic vs. Germ Cells" by J. Allen, S.  Lewis, M.
Moore, and U. Ehling. The paper was jointly presented by James Allen and Susan
Lewis and focused on the unique responses of germ cells to chemical exposures,

                                     28

-------
 particularly as  related to the synaptonemal complex and the events taking place
 during meiosis vs. mitosis in somatic cells.  The third presentation was by Michael
 Waters, on  The Performance of Short-Term Tests in the Detection of Germ Cell
 Mutagens: Qualitative and  Quantitative Results," by M.D. Waters, H.F. Stack, MA
 Jackson, B.A. Bridges, and  I-D. Adler. This paper analyzed the performance of short-
 term microbial, cultured mammalian cell and rodent  bone marrow tests in the
 discriminating confirmed germ cell mutagens  and nonmutagens.  Both qualitative
 responses and effective doses were analyzed. The final crosscutting presentation
 was given by Michael Shelby, based on a paper by M.D. Shelby and M.D. Waters on
 "Chemicals for Future Study of Heritable Genetic Damage in Human Populations."
 Shelby described what is known about chemicals that induce germ cell damage in
 animals, and discussed how this information may be beneficial in designing future
 genetic epidemiology studies in humans.

 Concluding Remarks

 David Brusick  introduced  the wrap-up  presentation  for  the workshop with the
 comment, "It appears to me, that we certainly do not have an absence of information
 to do risk assessment	I think, in fact, we know too much.. .and we end up with
 two problems: One is.. .paralysis by analysis-analyzing so much that you fail to act.
 . .  .  The second problem is that we have intimidated potential users of this
 information.. .with very fine  details which may or may not have any relevance to the
overall number	"
      Brusick continued, The regulator says, "We.. .do not have enough information
on  the dosimetry, dose-effect rates,  dose, route of exposure, pharmacokinetic

                                    29

-------
 information, et cetera'	The regulated community says, "We are not going to give
 you this information until you can show us how you are going to use it'....  As a
 consequence, data are generally developed out of our R&D programs. And most of
 those, as we know, are not [well] funded.  . . .  They do not  have. . .the data
 development power that industry has in developing products, if that information were
 made available	
        We have to. . .increase the visibility, and demonstrate the importance of
 genetic risk assessment.. .broaden it to include congenital anomalies in the zygotes,
 male transmitted congenital anomalies.. .cancer in the F1 and F2, and.. .fertility, and
 other associated reproductive and developmental  problems.
        The third (problem), is how to communicate the risk numbers that we develop.
 We need  to simplify the description without oversimplifying it. . .explain it  as a
 relatively simple process. We must express risk in terms of adverse outcomes, and
 not genetic endpoints	We need to reach a consensus on the basic rules for the
 process..  .on the assumptions that we are going to make..., then examine how to
 express it. Is it going to be a population risk? Are we going to talk about individual
 risks? .. .Are we going to include reproductive behavior? Are we going to talk about
 risk per mating opportunity, risk per conception, risk per Ft?  All of these things may
 become important....
       We have  risks for chromosome damage.  We have risks for gene mutation.
 Is it possible, once we have calculated these, to then come up with the single number
 or will we always be faced with  a  matrix of risk numbers?  . . .I think  we should
discuss cumulative exposures, versus  acute. . .for spermatogonial mutagens, you
accumulate risk on a daily basis	

                                     30

-------
        Finally, application of the parallelogram to risk assessment	It is a method
 that is intuitive.. .many of the extrapolation factors are built in.  We do not have to
 worry about extrapolating from the somatic cell to the germ cell.. .in the same animal.
 . . .  Those factors are taken into account, but the limitations  probably are more
 important than the advantages:  If.. .the shape of the [dose response] curve is linear
 that is  one thing; but if it is not linear, then you do not know that the relationship
 between somatic and germ cells in the animal is [valid] over a range of doses.. .and
 it is difficult to compare across compounds. We should standardize the somatic cell
 determinations in  humans, to one or two endpoints.  . .hemoglobin adducts,. . .
 dosimetry, or SCEs.. .then begin to look at one chemical versus another....  Now
 we do not know  the  relative sensitivities of these comparisons across  different
 compounds... ."
        Brusick concluded,  "From my perspective, it [the parallelogram] is certainly
 a model that we can use, but.. .1 think we have enough power in using the direct, or
 modified direct analysis.. .for risk assessment."
       Cantee Nolan closed the meeting with the following remarks: "Returning to the
 objectives  of the meeting, to identify  the methodology,  data  requirements, and
 mechanistic research needed to understand the health impact of germ cell mutagens,
 I think we  have gone  a long way towards doing that. When we chose  the four
 chemicals,  we knew that we would have problems, that there were data  missing.  I
think problems have been identified in applying the parallelograms to each  of those
compounds. ... We have  identified needs for data. I think that the parallelogram
does have advantages over other assessment methods and I think It should be given
a chance with a full set of data	

                                     31

-------
       We are quite pleased with the meeting. We think we have what we want, we
had some good overviews, and some crosscutting papers. I would hope, that this will
not be a one of a kind meeting, that we will have follow-up meetings to this one. I
would like to thank you all for participating, but I would like to thank particularly our
hosts.  .  .and  I would like to thank Frits,  because  Frits  really did  most of the
preparatory work on the European side for this.
                                    32

-------
Plenary Sessions
This section contains the introductory remarks and the following six plenary presentations of the

conference.
             Introductory Remarks
             Parallelogram Concept
             ICPEMC Efforts
             Regulatory Perspective
             Research Background
             EO Mutagenicity Risk Assessment
             A Reconsideration of EO Risk Assessment
                                       33

-------
34

-------
 Introductory Remarks

 The introductory remarks were given by Dr. Lawrence Reiter.  Dr. Reiter is from the
 Health Effects Research Laboratory at the US Environmental  Protection Agency in
 Research Triangle,  North Carolina, U.S.A.  The conference was dedicated to the
 memory of Emeritus Professor Frits Sobels by Dr. Heinrich Ott of the ENVIRONMENT
 Research Programme of DG XII of the European Commission.
       DR. REFTER: It is a pleasure for me to welcome you to the Research Triangle
 Park area and to this first EC/US Workshop on Risk Assessment. We are especially
 happy to welcome our European guests for what we hope will  become a continuing
 interchange through an ongoing scientific collaboration between the Health Effects
 Research Laboratory at EPA and  the Directorate  General for  Research and
 Development of the EC.
       As you know, the title of this workshop  is "Human Genetic Risks from
 Exposure to Chemicals,  Focusing on the Feasibility of a Parallelogram Approach."
 The title  and the workshop were the concept of Professor Frits Sobels who passed
 away on July 6,1963.  I am sure that all of you were greatly saddened at his passing.
 Those who knew him personally will be able to  recount better than I his many
 scientific and  administrative achievements.  Among  them were:   Professor and
 Chairman of the Department of Radiation Genetics and  Chemical Mutagenesis at
 Leiden University, and Member of the Royal Academy of Sciences of The Netherlands.
 His energy and enthusiasm for science and his strong personal friendships with many
of you provided much  of the motivation for what we are to accomplish here over the
next three days. His presence is greatly missed.
                                    35

-------
        As stated in your invitation letter, the main objective of this workshop is to
  identify the methodology, data requirements and mechanistic research needed to
  understand the health impact of germ cell mutagens. The rationale for choosing this
  topic is that a coordinated effort to quantify the genetic risks resulting from exposure
  to mutagenic  chemicals is greatly needed.  Chemicals  appropriate for such an
  approach are those for which there is considerable population exposure and which
  produce genetic effects in human somatic cells.
        Professor Sobels  believed that extrapolation from data on population exposure
  and genetic effect to what can be expected in germ cells of the exposed individuals
  should be possible by using good dosimetry and a parallelogram approach. The
  underlying idea of the parallelogram approach is that for most chemicals an estimate
 of the genetic damage in  human germ cells can be obtained by measuring a common
 endpoint in humans and mice, like genetic damage in lymphocytes, and a comparable
 genetic endpoint  in germ  cells  of mice, the desired target  issue inaccessible in
 humans.
       The chemicals selected for review at this meeting are:  Ethylene Oxide, 1,3-
 Butadiene, Acrylamide, and Cyclophosphamide. The first three of these are important
 industrial chemicals with substantial use worldwide, and therefore, considerable
 human  exposure.   The fourth, Cyclophosphamide, is one of the most widely used
 cancer chemotherapeutic agents.  We are very interested, therefore, in what you will
 decide  about  the  applicability of the  parallelogram  approach, or  alternative
 extrapolation methods to these agents.  But more importantly, we are interested in
your views about future research directions, especially as they will contribute to our
fundamental understanding of the impact of environmental  chemicals  on human
                                     36

-------
 genetic risk which, of course, includes a significant component of cancer risk and
 some component of developmental and reproductive risk.
       We realize, of course,  that this  is the first  EC/US conference on risk
 assessment and that it is highly focused on the parallelogram concept and germ cell
 effects in humans,  tt is anticipated that an important outcome from this workshop will
 be an evaluation of current knowledge and future research directions for the more
 precise assessment of human genetic risks.  We also want to emphasize, however,
 that our concerns for human health risk from exposure to environmental agents must
 be broader than this.
       Therefore, while we look to a successful outcome of this workshop, we also
 look forward to future workshops  encompassing cancer, developmental,  and
 reproductive risk assessment and other health effects as new knowledge and new
 methodology becomes available
       DR. WATERS:  Thank you, Larry.  I would like to call now on Heinz Ott from
 the EC for some introductory remarks.
       DR. OTT: Good morning, ladies and gentlemen. It is a great pleasure for me
 to welcome you on behalf of the European Commission and, in particular, on behalf
 of Professor Paolo Fasella (Director General for Science and  Research) to this first
 EC/US workshop on risk assessment which is dedicated to the late Frits Sobels.
       I do hope that this will  be the start for a  more intense and more formal
cooperation in this area which is to be based on an agreement between the EC and
the United States which  is, at the present, being negotiated. We hope that this will
facilitate organizing such things  in the future.  I also hope that we can organize the
next workshop in a not too remote future in Europe.

                                    37

-------
        If you say EC/US, I have to make a small parenthesis, because we have here
 friends from European countries which are, not members of the EC, in particular, from
 Sweden and Finland. They are, however, part of the European scientific community
 and they have been participating in the Community research programs since some
 time. We are now finalizing what is called the European Economic Area.  It means a
 partial  extension of the EC to the so-called EFTA States. These are the non-EC
 countries in Western Europe, in particular, the Nordic countries, Iceland, and Austria,
 and we look forward to that. Maybe we will have to rename the next workshop in that
 sense.
        As I mentioned at the beginning, we have decided to dedicate this workshop
 to the  late Frits Sobeis.  He was actually the man who proposed  to have this
 workshop, and he was deeply involved in Its preparation. We know that he was very
 enthusiastic about  It  in order to make a  point on the utility of the parallelogram
 approach for risk assessment which he considered, and I think he was right, his baby
 and his invention.
       I regret personally very much that he cannot be here.  I should add here a
 personal remark. When I came to  Brussels in 1972 to build up the first environment
 research program for the European Community,  we had also in mind to make a
 program specifically to develop tests for the genetic effects of chemicals.  I was
 advised to consult Frits Sobeis at  that time on what to do.  He was, from the very
 beginning, enthusiastic about the idea, and I may say that almost all what has been
done in this area of the program in  the last 20 years had been inspired and triggered
in one or another way by Frits Sobeis.
                                    38

-------
       The  program  developed from  a small program with a few  laboratories
developing genetic effects tests for regulatory purposes to a rather big program which
is encompassing now also population monitoring and other technologies which go
far beyond the testing of chemicals for regulatory purposes.
       I think I should underline here the merits of Frits Sobels for the European
research in this area and I think also for the research around the globe.
       I may kindly ask you perhaps to remain a minute in silence for the memory of
our friend, Frits.
       All that remains to me is to thank very much the EPA and Research Triangle
Park laboratories, in particular,  Dr. Larry Reiter and Dr. Mike Waters, for taking on the
trouble to organize this and for the hospitality. I wish the symposium to be a success.
   Thank you very much.
                                     39

-------
40

-------
 Parallelogram Concept

 The first plenary presentation was on the Parallelogram Concept and was given by
 Dr. Alan. S. Wright.  Dr. Wright is from the Tunstall Laboratory  of Shell Research
 Limited, Sittingbourne Research Centre, Kent, United Kingdom.

        DR. WRIGHT:  Ladies and gentlemen, I want to talk about the assessment of
 genetic risks, particularly heritable risks and the possible role of the parallelogram
 approach in making such assessments.
        First of all, I want to dwell briefly on why it is important to develop rational
 approaches to assess mutagenic risks. Then, I will touch on the key scientific issues
 that are fundamental to assessing genetic risks before going on to discuss the merits
 and limitations of the parallelogram approach as a model approach to assess genetic
 risks.
        I will leave any judgment about the performance of the parallelogram to this
 workshop.  This assessment is a major task of the workshop.

 Key Scientific Issues in Assessing Genetic Risks

 Concerns about mutagenic chemicals stem not only from what they do but also from
 how they do it.
       The majority of mutagenic chemicals operate by a genotoxic mechanism in
which they undergo direct covalent reactions with tissue DMA (Figure 2-1). However,
most of the compounds we are exposed to require metabolic activation into reactive
forms (electrophiles) to permit reaction with DMA.  The products  of these reactions,
e.g. DMA adducts, are generally promutagenic.
                                     41

-------
 A Current Perception of the Principal Stages & Systemic
 Determinants of Chemical Mutagenesis
                                                              S31997.7
         Precursor Mutagen	
               I Metabolic         Metabolic          *• Detoxification
              1 activation         detoxification	*- Products
         Ultimate  Mutagen 	
               I Penetration to
              1 molecular target
         Tissue DNA

DNA^repair  !  1Key (PrimarY & critical) interaction
         Chemically modified DNA (key lesions)
           I
Miscoding during cell
replication or error-prone repair
         Mutation in individual cells (heritable lesions)


                           Figure 2-1
                                42

-------
        Although the rates of formation of these key chemical lesions in DMA can
 deviate quite markedly from linearity in terms of exposure dose, there is no theoretical
 reason, nor is there any experimental evidence for a threshold in the induction of such
 primary DNA damage. Similarly, there is no evidence that the repair of this damage
 is so rapid and effective as to prevent the progression of even low levels of primary
 DNA damage into mutation. Consequently, there is no absolutely safe exposure level.
 Exposures to genotoxic chemicals pose an increased risk of mutation and cancer and
 congenital abnormalities.
        Furthermore, the mutagenic actions of genotoxic chemicals are additive,
 cumulative and, essentially, irreversible (Figure 2-2). For these reasons, even very low
 exposures are viewed with concern. Indeed, concerns about genotoxic chemicals are
 invariably focused on low-level risks posed by low exposures encountered in the
 general, domestic or occupational  environments.   High exposures to these agents
 cannot be countenanced except in chemotherapy. Therefore, our concerns should
 be focused on the development of methods to assess low-level risks posed by these
 agents. Quantitative risk assessment is essential for the proper management of these
 hazards.
       This overhead defines the four key elements necessary for the management
 of genotoxic risks (Figure 2-3).  The emphasis in this presentation is on the fourth
 element - the quantitation of risks.
       Certain exposures, for example, to indigenous genotoxic agents and some
 natural food components are unavoidable. The elimination or reduction of exposures
to potentially avoidable hazards,  for example, combustion products or certain base
chemicals, is often very difficult and costly.

                                     43

-------
  Induction of Mutation by
  Genotoxic Chemicals
                                                 B3I9B7.1
                   1.  No threshold
                   2.  Additive
                   3.  Cumulative
                   4.  Irreversible
                       Figure 2-2
Management of
Genotoxic Risks
                                                 931M7.2
      1.  Hazard detection
      2.  Hazard identification
      3.  Human exposure monitoring
      4.  Quantitative assessment of human risks
         (Human dose-response relationships)


                      Figure 2-3
                          44

-------
        Quantitative risk assessment is needed to prioritize these hazards and, most
 importantly, to determine safety margins.  Certainly, a failure to determine potency
 would lead to some uncertainty about the adequacy of safety margins and, possibly,
 to unnecessary measures to reduce  exposures.  Thus, despite their additive and
 cumulative effects, even genotoxic chemicals can pose a negligible risk to human
 health.
        The definition as to what constitutes a negligible or acceptable risk is, of
 course, a socio-political judgment; nevertheless, this judgment has to be realistic in
 the case of unavoidable hazards, (for example, indigenous hazards), and it has to be
 achievable in the case of avoidable hazards.
        I think the advent  of modem adduct technology which, for the first time
 permits the detection of hazards at exposure levels which may be consistent with
 negligible or acceptable risks, emphasizes the fact that quantitative risk assessment
 is absolutely necessary; that is, quantitation  is needed to discriminate  between
 acceptable and unacceptable risks.
       Thus, it is becoming clear that,  as we progress towards increasingly sensitive
 methods, quantitative  risk assessment becomes  an indispensable facet of  the
 procedures for the qualitative detection of hazards.  This is an inevitable outcome of
 improvements in sensitivity and resolving power.
       The  emergence of adduct  technology and its  applications in molecular
dosimetry to detect and monitor critical exposures provides the key  to quantitative
assessment of cancer  risks and heritable risks posed by exposures to genotoxic
chemicals.  Frits Sobels was well aware of this, and some years ago, he began to
                                     45

-------
 formulate strategies based on DNA dosimetry to improve the assessment of genotoxic
 risks, particularly heritable risks.
        As I see them, the aims of genotoxic risk assessment are to prioritize human
 genotoxic risk factors and to determine safety margins (Figure 2-4). The needs are
 for a strategy and methods to identify and quantify low-level risks.
        Direct epidemiological approaches to assess heritable  risks posed by low
 exposures to genotoxic chemicals are fraught with  problems.  Probably the most
 serious problem is the fact that the biological methods employed to determine
 heritable effects in human populations lack intrinsic resolving power. In particular, the
 limitations in resolving power make it increasingly difficult to attribute any observed
 effect to any specific genotoxic chemical as the exposure to the chemical is reduced
 to low levels.
        The next figure is a schematic representation of the factors contributing to the
 mutation  load (Figure 2-5).  The spontaneous component  is shown in inverted
 commas because there is perhaps some doubt about the fundamental basis of these
 so-called spontaneous mutations.
       The subdivisions in  the  increment  caused by  exposures to genotoxic
 chemicals signify increments due to  particular (specific) chemicals. When exposure
 is reduced to the very low  levels encountered in the  environment, the small
 incremental effect caused by a specific chemical may no longer be detectable within
the relatively high and variable background of mutation due to all other causes.
       Increasing the sensitivity of  the detection systems may not improve this
situation.  The overriding need is for  methods possessing the high resolving power
required to discriminate between effects induced by individual genotoxic chemicals.

                                     46

-------
Quantitative Genotoxic
Risk Assessment
        Aims
          1. Prioritize human risk factors
          2. Determine safety margins

        Needs
          1. Strategy and methods to determine
            low level risks
                                                 831997.3
                      Figure 2-4
 Factors Contributing to
 Mutation Load
                                                KU907.1S
                      Chemicals
                      (Indigenous and Exogenous)


                      Radiation



                      "Spontaneous"
                      Figure 2-5
                        47

-------
        Indeed, it is unlikely that even the most sensitive of the emerging mutation
  assays (for example, the assays based on mismatch technology which, incidently, can
  probably be applied to human sperm) will prove effective in this respect, at least at
  low exposure levels.
        The next figure illustrates the experimental determination of mutation (Figure
  2-6). The relatively high and variable background necessitates high doses of the test
  chemical. It may even be possible to observe compound-specific mutation spectra
  at such high doses.  However, it  is extremely unlikely that such spectra can be
  observed at realistic low doses.
        Currently, only adduct technology has the potential to actually discriminate
  between the effects, that is, induction of promutagenic lesions, of different genotoxic
 chemicals or classes of chemicals at very low exposures.
        The limitations in  current  epidemiological approaches to  assess human
 heritable risks necessitate alternative indirect approaches based on experimental
 models. In order to generate measurable effects, the experimental studies normally
 use higher doses or concentrations than those encountered in the environment.
       As a consequence, the estimation of the risk to humans requires two essential
 extrapolative stages (Figure 2-7). The first step involves extrapolation from the high
 dose range to the low dose range of interest. A linear projection is normally used and
 this may give rise to a gross overestimation or underestimation of the true risks,
 dependent upon the doses or concentrations required to saturate, induce, or inhibit
 metabolic activation or detoxification systems and DMA repair systems.
       The emergence of increasingly sensitive methods to determine mutagenesis
may reduce the scale of extrapolation by  permitting the use of lower doses, but the
                                     48

-------
  Experimental Determination
  of Mutation
                      Test Chemical
                      Chemicals
                      (Indigenous and Exogenous)


                      Radiation


                      "Spontaneous"
                      Figure 2-6
Extrapolative Stages in
Assessing Heritable Risks
       1. High-dose—How-dose extrapolations
          in experimental model


       2. Translation of low-dose extrapolation
          to humans
                                                 931997.4
                      Figure 2-7

                         49

-------
  practicability of such reductions is dependent upon the variability of background
  mutation in the model.
        The second extrapolative step involves translation of the experimental, (that is,
  the low-dose risk estimate), to humans.  Of course, this process is very complex and
  is fraught with uncertainty. The aim is to reduce extrapolative errors by correcting or
  compensating for differences between prospective risk models and humans in the
  operation of systemic factors that determine the quantitative relationships between
  exposure dose and the biological effect or the response.
        I have summarized these factors in the next figure (Figure 2-8). (This figure is
  a literal statement of the first figure).
        Clearly, the estimation of the individual contributions of each of these factors
  is a complex and daunting task which probably cannot be effected at the necessary
 level of sensitivity.  However, an appreciation of the mechanisms of genotoxic action
 has permitted simplifications allowing correction to be applied in two stages.  The first
 stage is designed to compensate for differences in determination of the critical dose
 which, in turn, determines the rate of formation of the promutagenic lesions in DMA.
        The second stage seeks to compensate for differences in the operation of
 factors determining the progression of the key (promutagenic) lesions into mutations.
        First consider the determination of DNA dose.  The key advance in this respect
 was the introduction of the target dose concept by Ehrenberg in the early 1970s. This
 new dose concept was developed to provide a measure of the doses of the ultimate
genotoxic agents reaching the  DNA of cells in tissues following  exposure to a
particular genotoxic chemical or a precursor genotoxic agent.
                                      50

-------
Systemic Determinants of Mutation
                                                  831987.5
1.  Factors determining the qualitative nature
   and magnitude of primary DNA damage:
    • Metabolic activation
    • Metabolic detoxification
    • Absorptive and transportation processes
    • Cellular and tissue compartment transfer processes
2.  Factors determining the progression of primary
   chemical damage into mutations:
    • Rates of repair of  promutagenic lesions.
    • Fidelity of repair of promutagenic lesions.
    • Cell replication rates
                       Figure 2-8
                          51

-------
        Target or DMA dose can be determined on the basis of the reaction between
 the genotoxic chemicals and DMA, but may also be based on the use of surrogate
 monitors. For example, the hemoglobin in circulating red cells. The latter approach
 is particularly valuable in monitoring human exposure and estimating  DMA doses in
 human tissue.
        Probably the  most effective of the hemoglobin methods is based on the
 determination of adducts formed by reaction with the amino groups of the n-terminal
 valine residues of the alpha and beta chains.  This approach has been developed by
 Margareta Tdrnquist in Stockholm.
       The important point about target dose is that the determinations are designed
 to automatically compensate for all individual and species differences in metabolism
 and related toxicokinetic factors that control the quantitative relationships between the
 exposure dose and the dose of the specific ultimate toxicant delivered to the target.
       The relationship between the exposure dose and target dose may be very
 complex.  For example, exposure to a specific genotoxic chemical or a precursor
 genotoxic agent, (like 1,3-butadiene) may lead to the generation of more than one
 ultimate reactive species.  Actually, the activation profile of butadiene is probably a
 lot more complicated than shown in this figure (Figure 2-9). However, the mono- and
 di-epoxides are probably the two most important genotoxic metabolites of butadiene.
 Of course,  the  problem is that these metabolites  display disparate mutagenic
 potencies. Complications of this sort have to be considered when addressing the
 problem of quantitative risk assessment or extrapolation.
       if we now consider attempts to correct for species differences in the systemic
determinants of the progression phase of mutation, the principal determinants are:

                                     52

-------
                            831W7.6
CH2=CH-CH = CH2
         [O]
        *  A
CH2=CH-CH-CH2
         [O]
            0
CH,
   -CH-CH-CH
      Figure 2-9
         53

-------
 firstly, the rates and fidelity of repair and, secondly, the rates of DNA replication
 (Figure 2-8).
        Clearly, any model approach employed to compensate for species differences
 in the operation of these factors  should avoid high  exposures  which may exert
 differential inhibitory effects on DNA repair and replication which, of course, could
 confound extrapolation to low doses and interspecies comparisons.  A number of
 approaches have  been  proposed  and  developed to compensate  for  species
 differences in DNA repair and replication;  one notable approach is the  radiation
 equivalence concept which is, in my opinion, the best validated and which is also
 designed to assess cancer risks.
       All of these approaches are based on the determination of relative mutagenic
 potencies, and they necessitate proportionality or parallelism between the repair of
 damage induced by different agents (for example, radiation and alkylating agents in
 the same test system or, alternatively,  the  same agent in different tissues of an
 individual or strain/species).  Accurate assessment of the dose of the  ultimate
 mutagen delivered to the target (DNA) Is fundamental to all of these approaches.
       Advances in molecular dosimetry in the 70s and early '80s led Frits Sobels to
 suggest  that the determination of specific DNA adducts such  as the  strongly
 promutagenic adducts formed by the reaction of alkylating agents with the  06 atom
of guanine residues would provide a basis for estimating heritable risks that were
otherwise difficult or impossible to determine directly.
                                     54

-------
 Merits of Parallelogram Approach

 In its original form, the  concept,  known  as the parallelogram  approach, was
 developed to estimate mutations in rodent germ cells by determining mutagenic
 potency in rodent somatic cells in vitro and applying determinations of the amounts
 of specific DNA adducts formed in the in vitro test system and in mouse germ cells
 in vivo to estimate mutations in the germ cells.
       The original parallelogram model is reproduced (Figure  2-10) exactly as in
 Frits' first publication on this topic.  In retrospect he would probably have signified the
 specific use of mouse somatic cells for the in vitro determinations rather than the
 more generalized approach signified by the term  •mammalian  cells". The critical
 assumptions implicit in this approach are: (1) the mutagenic effect of a genotoxic
 chemical in different cells or tissues of a given individual is a direct function of the
 amounts of specific DNA adducts, and (2) the proportionality between the amounts
 of DNA  adducts  and mutagenic action of a given genotoxic agent is constant for
 different cell types within a given individual  or strain.   Those are the fundamental
 presumptions of this model.
       Evidence  to support the contention that measurements of  specific  DNA
 adducts  may provide a basis for extrapolating mutagenictty data from one cell type
 to another within an individual or species or even between species has been reported
 by Bert van  Zeeland's group  (Figure 2-11).   They observed very similar mutation
frequencies when measurements of the levels of 06-ethylguanine in DNA were used
as  a basis for calculating the mutagenic potencies  of ENU  in inducing HPRT
mutations in VTO cells and specific locus mutations in mouse spermatogonia.  Clearly,
the correlation is excellent.
                                     55

-------
The Parallelogram
     Mammalian
     cells in vitro
      Mouse
    germ cells
   Measured specific
     DNA adducts
 Measured mutations
 Measured specific
   DNA adducts
Estimated mutations
                      Figure 2-10
 Induced mutations in V79 Chinese hamster cells (T)
 and the mouse specific-locus assay (n)
           250
         ^ 200
         b
           150
         0)

         CT
           100
         a
         "5
            50
                    93IW7.17

       (from van Zealand era/., 1985)
                  5    10   15   20
             Ethylations per 106 nucleotides
                      Figure 2-11
                         56

-------
        Part of the rationale for such an approach is that the direct determination of
 DNA adducts may, in fact, reflect DMA repair as well as adduct formation.  However,
 on  a cautionary note,  care  should be  taken, insofar  as is  possible  in such
 experiments, to avoid very high exposures which may lead to rapid saturation of repair
 systems and, therefore, to results which may not adequately reflect the important role
 of repair in determining mutation.  Low doses should therefore be used wherever
 possible.
        Turning to the problem of interspecies extrapolation, the parallelogram concept
 has been extended in an attempt to estimate mutation frequencies in human germ
 cells (Figure 2-12).  In principle, a common genetic endpoint is measured in somatic
 cells of humans and an experimental species in  vivo.  For example, one might
 measure mutation or genetic damage in peripheral  blood lymphocytes in a rodent
 species and in  humans  exposed to the same genotoxic  agent.   The subsequent
 determination of the potency of the agent in the germ cells in the experimental model
 would permit the ratio of  germinal to somatic damage  to be  calculated in the
 experimental species. Provided this ratio holds for human, this numerical value can
 be applied to calculate the risk to human germ cells.

 Limits of Parallelogram Approach

 You will note that some of the arrows and boxes employed in Figure 2-12 are broken.
 The broken lines signify a major problem inherent in this approach, i.e.  the difficulty
 in obtaining adequate human data.
      There are several assumptions in this type of model. The key assumption  is
that  the ratio of  germinal to somatic damage has the same numerical value in the

                                     57

-------
In vivo •*-*"//) vivo Extrapolative Model
       Mouse
                                                931997.10
       Human
   Germ cells in vivo
  Measured potency
 Somatic cells in vivo
  Measured potency
  Germ cells in vivo
  Estimated potency
 ________
\ Somatic cells in vivo\
  Measured potency  !
                      Figure 2-12
                         58

-------
 experimental model and in man. Given the marked variations in gene expression
 between tissues and species, there is obviously no theoretical justification for this
 supposition. Validation is necessary.
        Validation hinges on producing  convincing evidence  that  the  relative
 mutagenic effectiveness of a given genotoxic chemical towards a selected class of
 target somatic cells, (for example, peripheral blood lymphocytes), and germ cells in
 the test species, (for example, mouse), parallels that in humans over a broad dose
 range,  including very low doses.
        I have already touched on the emergence of a new generation of mutation
 assays based on mismatch technology. The new methods may provide a basis for
 direct validation of the model by permitting direct determinations of mutations in
 human sperm caused by high accidental or clinical exposures to genotoxic chemicals;
 but, as already noted, it is unlikely that even the most  sensitive of the emerging
 mutation assays will  actually permit the determination of very small increments in
 mutation caused by low environmental exposures. In my view, such direct validation
 of the parallelogram concept is not yet possible, and alternative indirect strategies
 need to be developed.
       One possible approach would  be to attempt  to demonstrate that the
 proportionality or extrapolation "constants* relating somatic and germ cell damage are
 independent of species by performing comparative assessments in  a number of
different species with a broad range of chemicals tested at a range of doses,
particularly low doses.
      As pointed out by ICPEMC Committee 4 (Lyon et al., 1983), the problem in
developing an effective parallelogram  approach to predict germ cell damage is in

                                     59

-------
 finding a suitable genetic endpoint that can be measured in the somatic cells of both
 experimental species and humans and which has extrapolative value for the prediction
 of germ cell damage.
        The determination of chromosome aberrations in peripheral blood lymphocytes
 provides an effective method for the qualitative prediction of genetic damage to germ
 cells, but the results of comparative studies reported by Van Buul suggests major
 differences in the ratio of somatic and germ cell damage between mice and Rhesus
 monkeys treated with X-rays. These results have cast some doubt on the value of
 chromosome aberrations for quantitative  applications in the  parallelogram mode.
 However, it should be stressed that the data generated in Van Buul's studies are
 probably too scanty to warrant firm conclusions, and I think it is quite clear that much
 more work is needed to test the general feasibility of the parallelogram approach. Of
 course, this is a major aim of the current workshop.
       The use  of radiation  in such validation studies  make it relatively easy to
 determine the critical dose and so facilitates valid interspecies comparisons. Accurate
 target dosimetry  with genotoxic chemicals is more difficult, but it is, nevertheless,
 entirely feasible.   Such target  dosimetry  provides a basis for making  valid
 comparisons of mutagenic potency between species and between tissues within an
 individual.
       However,  even when  target doses are identical, there is no  guarantee that
 major species differences will not occur in the ratio between somatic and germ cell
 mutation due, for example, to species differences in the ratio of the rates of DMA
repair in germ  and somatic cells.  The occurrence of such differences would, of
course, militate against a simple parallelogram approach  and would necessitate
                                     60

-------
 additional corrections such as those employed in the rad-equivalence model to be
 applied.
        As already noted, one of the problems in  developing and validating  a
 parallelogram approach to predict human germ cell mutations is in obtaining reliable
 mutagenicity data in human somatic cells exposed in vivo, particularly at low dose
 levels. Consequently, the ability to extrapolate in vitro  mutagenicity data to the same
 type of somatic cells in vivo is, therefore, potentially very valuable.
        If such in vitro/in vivo extrapolation proves to be valid then we can envisage
 an in vitro-in vivo parallelogram illustrated in the next figure (2-13). (Note the absence
 of broken lines).  Alternatively we may envisage a combination of the in vitro/in vivo
 (Figure 2-13) - in vivo/in vivo (Figure 2-12) extrapolative models in which the potencies
 in somatic cells in vivo are estimated on the basis  of the measured potencies in vitro.
        Information about the extrapolative value of in vitro data may be obtained by
 conducting appropriate in vitro-in vivo comparisons In the same type(s) of somatic
 cells in a range of  experimental  species.   Consistent  results  would  promote
 confidence in the extrapolative value of the corresponding human in vitro data to the
 human situation in vivo.
       This type of parallelogram approach  is,  in fact, employed as a standard
 procedure for validating the metabolizing systems of preparations of human tissues
 employed  to detect  and  identify  human genotoxic  carcinogens  in vitro.   In
 Pharmacokinetics, such dosimetry approaches are described as matrix approaches
and  have a very high  resolving power. As already mentioned, biological mutation
assays lack such intrinsic resolving power. Nevertheless, studies in humans exposed
                                      61

-------
In vitro •*-*- In vivo Extrapolative Model
        Mouse
                                                 931997.9
      Human
   Germ cells in vivo
  Measured potency
 Somatic cells in vitro
  Measured potency
 Germ cells in vivo
 Estimated potency
Somatic cells in vitro
 Measured potency
                      Figure 2-13
                        62

-------
 to gamma radiation have provided valuable quantitative data relating the induction of
 genetic damage in human somatic cells to radiation dose in vivo.
        These in vivo data, taken from one of Mike Waters' papers (Waters et al., 1986)
 (Figure 2-14), are compared with the results of laboratory studies to determine the
 cytogenetic  effects induced by  gamma radiation in human  peripheral blood
 lymphocytes exposed in vitro and in mouse lymphocytes exposed in vivo and in vitro.
 In the original paper, these data were used to construct a series of parallelograms
 corresponding to doses of 25,50,100, and 200 rads. However, a graphical construct
 is perhaps more revealing.  In this particular case, it is clear that the mouse in vitro
 data is highly predictive of the in vivo response. Application  of the mouse in vitro/in
 vivo proportionality constants to the human in vitro data is reasonably predictive of
 the damage in human cells in vivo, that  is, within a factor of 2.
        I have summarized the main conclusions in this final figure (Figure 2-15). The
 data,  reported  by  Mike Waters et al.  (1986),  using the radiation model  are
 encouraging, but fall short of the requirements for validating  models to estimate the
 risks of heritable  mutations in humans  posed by low exposures to genotoxic
 chemicals.  Comprehensive studies using a range of genotoxic agents and  a range
 of species are needed to promote confidence in the use of experimentally determined
 values of the relative risks of germinal and somatic cell mutations to predict the risk
to human germ cells.
       Direct validation is probably not possible, certainly at low exposures.  Indirect
validation  via human somatic cells in vivo is  difficult at low exposures where it is
difficult to obtain reliable human data. Indirect validation via human somatic cells in
                                     63

-------
§ «o -
.i
                                      Rads
          Cytogenetic damage in human and mouse peripheral blood lymphocytes
          exposed to -y-radiation in vitro and in vivo. (Data from Waters etal., 1986).
                               Figure 2-14
                                    64

-------
                           CONCLUSIONS
        1.     DIRECT VALIDATION NOT POSSIBLE AT LOW EXPOSURES.
        2.     INDIRECT VALIDATION VIA HUMAN SOMATIC CELLS IN VIVO-
              DIFFICULT AT LOW EXPOSURES.
        3.     INDIRECT VALIDATION VIA HUMAN SOMATIC CELLS IN VITRO
              POSSIBLE.
        4.     TARGET DOSIMETRY ESSENTIAL
        5.     EXPLORE ALTERNATIVE STRATEGIES TO CORRECT FOR
              SPECIES DIFFERENCES IN "PROGRESSION".
                               Figure 2-15
 vitro is obviously possible, but we still have the problem of high dose-low dose
 extrapolation. Perhaps some of the more sensitive new techniques may assist with
 this type of extrapolation.  However, the practicability of these new approaches in
 actually reducing the scale of the  extrapolation  is largely dependent upon the
 background variability in the model.
      Target dosimetry is an essential requirement of all of these approaches.
 Precise determination of the critical dose is needed, particularly, when extrapolating
from the in vitro to the in vivo situation. Determination of the critical dose of the
ultimate genotoxic agent is probably most readily and accurately determined, at least
in humans, by using a surrogate dose monitor such as the haemoglobin of circulating
erythrocytes.

                                   65

-------
        In view of the potential limitations of the parallelogram approach it would be
 prudent to explore alternative strategies to correct for species differences in the
 progression phase of mutation.
        Thank you very much.
 Discussion

        DR. ANDERSON: When you have a lot of data, your last point, I mean, if you
 come to that conclusion in general terms, for compounds where you have a lot of
 data,  surely we should  be  able to apply it.  Are  you making this  last point for
 everything?
        DR WRIGHT:  For everything?
        DR. ANDERSON: Yes, you are saying use alternative strategies.  Because of
 the problems with the approach, we should have alternative strategies lined up.
        DR. WRIGHT: Yes, I believe that, yes.
        DR. ANDERSON: But where you have got a lot of data for all chemicals and
 even human data, do you not think it is applicable then?  Are you making a broad
 statement at this point?
        DR WRIGHT: My statement is addressed to the broad issue of assessing low-
 level risks.  I do not believe we have any reliable human data relating genetic effects
 to low-level exposures to genotoxic agents.   The  point I  want to make is that, in
 human populations, it is  extremely difficult to attribute any observed biological effect
to exposure to a specific chemical when the exposure is very low and is in a mixed,
(for example, environmental), exposure situation.  It is relatively easy to attribute such
an effect to general environmental factors but it is much more difficult to identify the

                                     66

-------
 specific causative agent(s). Such identification can only be achieved using adduct
 technology.
       DR ANDERSON: Okay, thank you.
       DR. GARNER:  Alan, you mentioned something which is a little perhaps
 controversial.  You said that there was no such thing as a spontaneous mutation or
 there may not be such a thing as a spontaneous mutation.
       DR WRIGHT:  I said may not.
       Da GARNER:  Will you define what you implied there?
       Da WRIGHT:  I am suggesting that background exposures to radiation and
 genotoxic chemicals may contribute to the frequency of "spontaneous" mutation. We
 do not know to what extent these  background exposures may contribute to the
 "spontaneous" mutation rate. This remains to be resolved. We are probably exposed
 to low levels of thousands of genotoxic chemicals both indigenous and exogenous.
 The exposures may, in large measure, be responsible for the spontaneous mutation
           »                                        f
 rate.
       Da SORSA: Maybe at this point  it is a bit useless to throw these problems,
 because we will need to discuss these further.  I agree very much to all the hesitations
 which you showed in the conclusions and especially to the problems of the  species
 specific differences. Of course, concerning humans, we are a polymorphic species.
       Da WRIGHT: Yes.
       Da SORSA:  And we  never  know which  mammalian or  rodent  is
representative, we represent.
       One further problem which you did not specify and I think which we have to
discuss during these  days here is  the endpoint specificity, because there  are

                                   67

-------
 differences in endpoints when we measure the damage in somatic cells either in
 humans or in experimental species.
        DR WRIGHT: Yes, this is one of the reasons why I very much favor the
 development of the techniques based on mismatch technology. This new technology
 provides the basis of a  universal approach which may be applied to any tissue or
 sample of DMA. The approach does not rely on a phenotypic endpoint of any type.
        DR WATERS: Fred De Serres?
        DR. DE SERRES: You raised the issue in your talk about the shape of the
 dose-response curves and, you know, the assumption that most of them are  going
 to be linear.
        I think this is a point that needs to be discussed very thoroughly. Most dose
 response curves  are not linear.  You  can  get very different induction kinetics for
 different classes,  and, you  know,  different types of genetic  damage will have very
 different consequences in the human population.
       So, I just think this is a point that should not be glossed over but dealt with
 very thoroughly.
       DR. WRIGHT: Yes, I would accept that view and totally agree with you, Fred.
 I think the target dose approach helps to reduce some of these problems (Figure
 2-16).
       Certainly, if we examine data relating, let's say, the generation of DMA adducts
to exposure dose, these  relationships are often not linear (Figure 2-17).  However, if
we then examine the data relating DMA adducts to mutation, then these relationships
become much more linear.
                                    68

-------
                                             031987.11
 Mutation
 frequency
 or
 DNA dose
                            Exposure dose
Mutation
frequency
                  Rgure2-16
                                            931997.12
                               DNA dose
                  Figure 2-17

                     69

-------
        Da DE SERRES: Right, but, you know, the problem is trying to extrapolate
 to very low exposures.
        DR. WRIGHT:  One of the problems is that, it is extremely difficult to make
 accurate measurements in the very low dose range. I think we have to assume that
 the dose response curve, is linear in the very low dose range. It is not possible to
 measure biological effects at such vanishingly small exposures.
        Da DE SERRES: I think this is a really extremely difficult area and one we
 have to address very thoroughly.
        Da WRIGHT:  Yes, I would agree with that.
        Da DE SERRES: Because I think it is just as dangerous to extrapolate, you
 know, to what we expect at low levels of exposure and say that we are going to get
 an order of magnitude higher effect than you would actually get as  it is to
 underestimate. So, you have to have caution in both directions.
       Da WRIGHT: Yes. I did in fact state that such extrapolative problems could
 lead to gross overestimation or underestimation of the true risks.
       Da DE SERRES:  Right.  Well, it is particularly important where you are
 dealing with exposures in the workplace, and you are trying to determine the effects
 of that exposure on workers or on people who are exposed to  agents where the
 exposure cannot be avoided.
       Da WRIGHT: Yes. However, I believe that it is important that when we reach
a point where the exposures are too low or the populations are too small to permit
the detection or measurement of genetic effects, then we must employ conservative
models to estimate the risks.
      Da  DE SERRES:  Right, thank you.
                                    70

-------
 ICPEMC Efforts

 Dr.  David Brusick summarized the risk assessment efforts of the International
 Commission for protection against  Environmental Mutagens and  Carcinogens
 (ICPEMC).  Dr. Brusick is from the Hazleton Labs America, Inc., located in Vienna,
 Virginia, U.S.A.

        DR. BRUSICK: Frits Sobels was one of the founding parents of ICPEMC and
 was the first chairman of ICPEMC; so even though what I am going to talk about is
 not the parallelogram approach, I think it is still appropriate to review it.
        We  were contacted by Health Canada and also the EPA to look at the
 following questions: if you have risk data in mammals, in mice, can we extrapolate the
 risk  data from that model to human populations and, in some way,  describe the
 impact relative to the number of new mutations in the F1 so that this information can
 be useful to decision-makers?
       The major problem in this field has been that regulators or decision makers,
 whether they be in the regulatory agencies or in industry or whatever, are faced with
 a large amount of data, but are not given very good guidance as to how that data can
 be used in making decisions that affect risks to human populations.
       I was the project manager for ICPEMC.  George Douglas was the Health
 Canada project manager, and Kerry Dearfield was the EPA project manager. The
 members of the team included  Heinrich Mailing who was the chairman, Jack Favor
who is not here, Gary Sega who is here. David Layton did a lot of exposure analysis,
environmental exposure  information.   John Wassom was very instrumental in  his
                                    71

-------
group in providing data, and we also had input from Jim Burkhart at NIEHS and also

John Mulvihill, and Andrew Ceizel also was involved (ICPEMC, I993a,b).

      As I said, this is a very, very practical approach.  So, the operational definition

(Figure 2-18) of what we were trying to do in a genetic risk assessment was a

determination of the number of additional dominant genetic diseases, not mutations

but diseases, that would be added to the existing genetic burdens of the F1 progeny

derived from human individuals exposed to a mutagenic compound (Figure 2-19).
                   OPERATIONAL DEFINITION

 Genetic Risk Assessment: A Determination of the Number of
 Additional Dominant Genetic Diseases Added to the Existing
 Genetic Burden of the F1 Progeny, Derived from Individuals
 Exposed to a Mutagenic Compound.
                             Figure 2-18
 While it may be possible to wait until all of the needed information gaps
 are filled, there are disadvantages in waiting until all of the uncertainties
 have been eliminated.  The genetic disease burden is a serious health
 issue for most societies. In addition to the emotional burden, the medical
 costs associated with each induced phenotype result in a significant
 contribution to the total health care cost. Further, genetic toxicology has
 an obligation to explain the human hazard significance of positive results
 produced by to vitro tests and mammalian somatic cell assays.
                             Figure 2-19
                                72

-------
       To start with, we had to define pretty much a road map of where we were
 going, and that is described here. This is the risk extrapolation road map that we
 used, and it starts with basic information (Figure 2-20).
       First of all,  look at the  induced mutation frequency per unit dose  in
 experimental animals. Most of the data that we have comes from models.  It can be
 heritable translocations,  dominant-specific locus mutations, or recessive-specific
 locus mutations (Figure 2-21).
       We then have to factor in the kinds of exposures that humans are experiencing
 and, as  a result, should be able to get the per  locus induced mutation frequency
 following mutagenic exposure at the level that humans are exposed to (Figure 2-22).
       We would then  have some sort of an estimate of the expected  per locus
 mutation frequency which could then be factored by the number of disease-causing
 loci in the human genome. You need to know how many of the genes in the genome
 are responsible for genetic disease.
       You can then take the mutation frequency given per unit dose multiplied by
 that number of genes. You could then calculate the expected frequency of disease-
 causing mutations in an F1 population following a given exposure of the mutagen.
       Then, that would be taken times whatever F, population is derived from the
 exposed population.  You would then have an estimate  of the  number of new
 diseases, dominant diseases, in that F1 population.
       Now, the only other thing  that we wanted to add into this  is some way to
account for differences between the rodents and humans and also to account for the
differences  in the kinetics of induction,  (that is, dose rate differences, route  of
exposure differences).

                                     73

-------
                        RISK EXTRAPOLATION ROAD-MAP
(1)      Induced mutation frequency
        per target locus per unit dose,
        determined in animal
        experiments
Exposure dose In
humans
Expected per locus
induced mutation
frequency following
mutagen exposure
                                              (9
                               (m,)
(2)      Expected per locus mutation
        frequency due to mutagen
        exposure
Number of disease
causing loci In
human genome
Expected frequency of
disease-causing
mutations in F1
population following
mutagen exposure
                   (m,)
                      (m,)
(3)      Expected frequency of
        disease causing mutations in
        F1 population following
        mutagen exposure
F, population size
Expected number of new
mutations causing
disease in the F1
individuals resulting from
exposure of the parental
population
                   (m,)
                               ("J
            THE BASTC FORMULA WILL BE MODF1O) BY SCALING FACTORS
                              nm s REF x m1 x n0
            where REF Is a product of several uncertainty and scaling factors
                          added at each step of the equation
                                   Figure 2-20
                                       74

-------
               PRIMARY PROJECT OBJECTIVES

DEVELOP A RISK ASSESSMENT MODEL FOR USE AS A "STRAW MAN"

   •     THE MODEL SHOULD BE SIMILAR TO MODELS
         TRADITIONALLY USED FOR OTHER TYPES OF TOXIC
         RISK ASSESSMENT
              Animal mutation induction studies and human
              exposure information are the basic data
              requirements.

   •     DATA REQUIREMENTS SHOULD BE FLEXIBLE: THE
         MODEL COULD USE DOMINANT OR RECESSIVE/GENE
         OR CHROMOSOME MUTATION DATA
              The model should be able to use whatever data
              are available.

   •     THE MODEL SHOULD ALLOW FOR THE
         INCORPORATION OF EXTRAPOLATION SCALING
         FACTORS
              The model must be able to extrapolate
              responses from the mouse to humans through
              the use of scaling factors based on experimental
              evidence.

   •     RISK EXPRESSION SHOULD BE A QUANTITATIVE
         ESTIMATE OF THE NUMBER OF NEW DISEASES
         EXPRESSED IN THE F1 POPULATION
              Risk must be expressed as the number of new
              genetic diseases in the progeny produced from
              exposed populations.  This limits the risk to
              dominant disease phenotypes.
                       Figure 2-21
                           75

-------
           Data Availability for Risk Extrapolation
   Presumed Human
   Mutation Process
     Mouse
     Model
  Mouse to Human
   Extrapolation
   Uncertainties
      Exposure
   Compound
 Administration
  Exposure Route
    Dose Route
  Pharmacokinetics
Pharmacokinetics
 Metabolic Variability
   Sex Specificity
   DNA Reactivity
 DNA Reactivity
DNA Repair Capacity
  Germ Cell Stage
    Specificity
   Induced Germ
    Cell Muation
 Induced Germ
  Cell Muation
                        Estimate of Induced
                          Human Disease
                             Burden
  Locus Specificity
 Mutant Expression
Germ cell risk extrapolation from mouse to humans requires data from
several sources including (a)  human exposure, (b) mouse mutation
inductions and  (c) extrapolation scaling and uncertainty assumptions.
This figure attempts  to  demonstrate how  these  three types  of
information  are related  to each other  and  the  overall process.
Information  appearing in  the boxes in  columns  1  and 2 can be
experimentally obtained.  The remaining  information  is derived from
estimates.  Column 3  identifies the  factors considered essential in
making extrapolations from mouse to  human.
                            Figure 2-22
                               76

-------
        So, again, reaching into the toxicological sciences, there is a whole science

 of looking at scaling factors or safety factors or whatever you wish to call them, and

 coming up with some estimate of the risk.

        So, we added in what we call risk extrapolation factors; and risk extrapolation

 factors are just nothing more than a combination of all of these numbers that we

 might be able to estimate that have to then be factored into this equation to come up

 with the estimate of the number of new mutations causing dominant diseases in the

 F1 population (Figure 2-23).
       Proposed Scaling Factors (REF) for Mouse to Human Risk Extrapolation
   Parameters                  Experimentally Determined*           Default Factor
   Locus Specificity            Compound specific          2 to 5 (assumes human
                                                       sensitivity)
   DNA Repair Variability        Compound specific          0.1 (assumes mouse
                                                       sensitivity)
   Metabolic Variability          Compound specific          1 (70*)
   Dose Rate Variability          Compound specific          0.1
   Exposure Route             Compound specific          0.5
   Germ Cell Stage Specificity    Compound specific          1 (assumes equivalence)
   Dose Response Kinetics       Compound specific          1 (assumes linearity)
          Each factor that is experimentally determined for the compound under assessment will
          be used. If compound specific Information Is not known, the appropriate default factor
          will be used.
          If metabolism and detoxification are dependant upon glutathtone transferase
                                    Figure 2-23


       Once you have that number, it is then possible to look at some estimate of the

cost to treat or to manage the health care for each of those genetic diseases. So,

you will  have an incremental  cost that can be then compared to the economic


                                        77

-------
 advantages of the material, of the process, or whatever it might be, and do your risk
 and risk/benefit calculations.
        So, when we actually came up with the final list - locus specificity, DMA repair
 variability, metabolic variability, dose rate variability, exposure route, germ cell stage
 specificity, and dose response kinetics - again, very little in the way of dose response
 data) we  looked  for the chemical-specific information and some sort of a default
 factor.
        The calculations that we used essentially included the standard calculations
 using a direct method for mutation analysis or the doubling dose method (Figure
 2-24).  Possibly, the doubling dose data is one that we can apply a little bit better,
 because there is  at least some reasonable  amount of data on doubling dose per
 mutation in human populations to specific types of endpoints.
        Risk is calculated for dominant traits.   We are not calculating the risk for
 recessive  mutations in the Fr  There will be  recessive mutations  induced.  The
 problem is that estimating what the impact of recessive mutation will be is difficult.
        First of all, the contribution of recessive mutations is probably going to be
 relatively small (Figure 2-25).  It may accumulate in the population and, in 100 years
 or so, might be very important. We hope that in 100 years from now we can probably
 do a lot more with respect to controlling  and  diagnosing and treating  genetic
 diseases.  So, the F1 dominant diseases are  probably the most important to look at,
 at least initially.
       Some risks are not even going to be dealt with at all, for example, aneuploidy.
We don't have good data that we can put in and do a risk assessment for aneuploidy,
                                      78

-------
            CALCULATIONS USED IN RISK ASSESSMENT
 Direct Method
                  RISK
Where
                   MMo».xLHuni-nxDxN   I
                  ^••••^^^^^^^^^^^^••••BM
LHu

D
N
   m*n
                   '     induced per locus mutation rate per unit
                         dose exposure estimated in the mouse
                   number of loci in humans at which dominant
                   disease mutation may arise
                   -     exposure dose
                   -     number of offspring descendant from
                         exposed parents
Doubling Dose Method
Where
Spon

D
DD
     H(JRWnt
N
Risk Extrapolation
                         overall spontaneous mutation rate to
                         dominant disease alleles in humans
                         exposure dose
                         doubling dose estimated in the mouse (that
                         dose which induces a mutation rate equal
                         to the spontaneous mutation rate). The DD
                         is calculated as the mouse spontaneous
                         mutation rate per unit dose
                         number of offspring descendent from
                         exposed parents
        Number of New
        Diseases in the = REF x SponHinMm x D/DD x N
        Offspring
                              OR
        Number of New
        Diseases in the = REF x
        Offspring
                              xlOOOxDxN
                         Figure 2-24
                             79

-------
                        UMITAT1ONS OF THE METHODS

          EACH RISK MODEL HAS DATA REQUIREMENTS WHICH CANNOT BE FILLED WITH

          EXISTING INFORMATION SOURCES (I.e. Number of Loci Producing Dominant Disease
          Traits In Humans, Spontaneous Mutation Rates for Humans).

          ANIMAL DATA MAY ONLY BE AVAILABLE FOR ONE TYPE OF ALTERATION (I.e. Gene
          Mutation) AND NOT OTHERS. THUS, A COMPLETE RISK ASSESSMENT CANNOT BE
          PERFORMED.

          ANIMAL DATA MAY ONLY BE AVAILABLE FOR DOMINANT MUTATIONS ESTIMATES
          WILL REQUIRE FACTORING DOMINANT.-RECESSIVE INDUCTION RATIOS.

          RISK IS CALCULATED ONLY FOR DOMINANT TRAITS. RECESSIVE MUTATIONS WILL
          BE INDUCED BUT WILL HAVE MINIMAL IMPACT ON THE F1.

          SOME RISKS WILL NOT BE ADDRESS IN THE ASSESSMENTS (i.e. Data for
          Aneuploidy are Seldom Available).

          SOME CRITICAL DATA GAPS MUST BE FILLED WITH ASSUMPTIONS (l.e. Male and
          Female Risks are Equivalent; Mouse and Human Mutations nave Equivalent
          Penetrance; There Is no Age Influence)
                                  Figure 225


 and, obviously, we have got some critical data gaps in male to female risk.  We are

 looking at risk that is not sex specific, although most of the data is coming from

 males.  Age effects, penetrance of mutations and so on, again, are limitations that

 have to eventually be addressed, but I think we can go with what we have got.

       We were asked by Health Canada to look at acrylamide (ICPEMC, 19935), and

 by EPA to review ethylene oxide (ICPEMC, 1993a).  First of all, acquire the animal

 data. Determine from that animal data using information specifically directed to those

 chemicals, either ethylene oxide or acrylamide, come up with some  sort of scaling

factors that have  to be added  to the equation, then assemble human population

exposure data, get the exposure data assessment if it is available, calculate the risk
                                     80

-------
 using the direct or the doubling dose method,  and then estimate the exposure
 induced genetic burden in the Ft population.
        So, the next two figures essentially summarize the exposure data and the
 numbers that were generated for ethylene oxide and acrylamide using the approach
 we have (Figures 2-26 and 2-27).  Human exposure numbers that we used in the
 calculations for ethylene oxide were 36 ppm for a six-hour day, five days of the week.
 This is an occupational exposure, relatively high exposure.  Using the direct method
 and scaling factors that we used, single gene dominant mutations per 10,000 births
 calculated out at 12, and for chromosome diseases per 10,000 births, 123.
        Using the doubling dose method, though, it was considerably higher, and that
 was based upon an estimate of 1000 loci producing dominant F1 diseases.  Doubling
 dose was 26 single gene diseases per 10,000 and 207 chromosome diseases, and
 we had a scaling factor of 0.2 that was used.
        Acrylamide used fairly low exposure (Figure 2-28). These are the permissible
 exposure levels for occupational and environmental exposure to acrylamide: 0.07
 mg/kg  per day for occupational and 0.03 mg/kg per  day for regular general
 environmental exposure.
       Again, now looking at single gene diseases, F1 dominant per million births for
 occupational exposure was estimated at 24  and  environmental 10,  and for
 chromosome, 169 per million births and 73 for environmental exposures.
       The doubling dose method was considerably lower in this case, about 0.4 and
 0.2 for single  gene diseases and 1 and 0.5 for chromosome diseases. Again, based
 upon the fact that we didn't have a lot of information, we ended up with a scaling
factor of about 0.2, which ended up being a lot of default scaling factors.

                                     81

-------
        Summary of the Risk Extrapolation Approach
                         Produce or acquire the
                      Appropriate Animal Germ Cell
                             Mutation Data
                          Determine the REFs
                          for the Extrapolation
  Assemble human population
      and exposure data
              Assemble exposure
               assessment data,
                 if appropriate
                            Calculate Risk
          Option A
     Direct Method x REF
OR
        OptionB
Doublind Dose Method x REF
                       Estimate exposure-induced
                        genetic disease burden in
                             F, offspring
Use of Options A or B depends upon the availability of critical data
regarding the number of disease associated loci in humans (needed for
the Direct Method)  or  the  spontaneous background mutation  rates
(needed for the Doubling Dose Method).
                             Figure 2-26
                                 82

-------
 HUMAN EXPOSURE:



 RISK ESTIMATES:


         METHOD


 Direct


 Doubling Dose
ETHYLENE OXIDE


Occupational exposure at 36 ppm for an six
hour day, five days per week.
SGD/104 BIRTHS
       12
       26
CD/104 BIRTHS
     123
     207
REFS EMPLOYED:
        Parameter            REF

Locus Specificity               2
DMA Repair Variability           0.1
Metabolic Variability            1
Exposure Route                1
Germ Cell Stage Specificity      1
Dose Response Kinetics         1
                             1

                   REF =    0.2
                            Figure 2-27
                                83

-------
 HUMAN EXPOSURE:
   ACRYLAMIDE

Occupational exposure at 0.07 mg/kgper day.
days per week. Environmental exposures at 0.03
mg/kgper day.
 RISK ESTIMATES:
          METHOD
     SGD/106 BIRTHS
CD/106 BIRTHS
 Direct
       Occupational
       Environmental
                 24
                 10
          169
           73
 Doubling Dose

       Occupational
                 (0.4)
                 (0.2)
            1
           (0.5)
REFS EMPLOYED:
         Parameter

Locus Specificity
DMA Repair Variability
Metabolic Variability
Exposure Route
Germ Cell Stage Specificity
Dose Response Kinetics
                     REF =
               REF

                  2
                  0.1
                  1
                  1
                  1
                  1
                  1

                  0.2
                             Figure 2-28
                                84

-------
       So, the numbers that we have are essentially estimates of the frequency of
 dominant genetic disease burden in the offspring of persons exposed to those given
 doses of the environmental mutagens (Figure 2-29). These numbers can be used to
 assess the health burden following exposure with background, or they can be used
 to compare one chemical to another for comparative risk.
                     WHAT DO THESE NUMBERS MEAN?
   The risk numbers generated above are estimates of the frequency of dominant
   genetic disease burden in the offspring of persons exposed to a given dose of
   an environmental mutagen. These numbers can be used to assess the health
   burden following exposure with the background or they can be used to
   compare one chemical to another for comparative risk. Estimates of treatment
   cost per genetic disease are also available for use in economic considerations
   of risk:benefit decisions.
                                 Figure 2-29
       The only other elements that we addressed In the reports were, for Health
 Canada, we were asked to look at the possibility of taking somatic cell mutation data
 and using a way of coming up with some sort of risk reduction factors to determine
 the upper level permissible exposure to a mutagen that would protect the germ cells.
       Again, this was an  interesting problem  to tackle, but  we  did  produce
 essentially a proposal that would allow one to look at mutations in somatic cells and,
 using a series of safety factors based upon comparing germ cell and somatic cell
 data, estimate an upper exposure limit for protecting germ cells.
       EPA also asked this group to answer two other questions. One was, would
 it be possible, in the absence of any germ cell data, to utilize risk assessment derived
from somatic cell or, "cancer risk analysis" to act as a surrogate for genetic risk?  In
                                     85

-------
 other words, would we then have a conservative estimate of mutational risk that we
 could apply directly to germ cells? So, if you are protecting for cancer, would you,
 in a sense, also protect for the germ cell effects?
       The other question was, like cancer, is it possible to develop a level of induced
 mutation which would not be of a significant health concern and would not, therefore,
 require the generation of risk management and risk benefit type of processes being
 initiated?  In other words, is there an acceptable level of new mutation?
       DR. SELBY:  In fact, referring to McCusick's catalog, I  think most clinicians
 would agree that even the number in there would be a very great underestimate of
 dominance, because so little of human genetic disease Is really even understood at
 the locus level yet
       However, this is perhaps not really much of a problem, because the direct
 method, at least in the way it has been applied for radiation risk estimation which is,
 of course, much more developed than that for chemicals, does not use strictly the
 number of genes anyway.  It uses a multiplication factor that can be arrived at in
 various ways, one of which would depend on a guess as to the number of genes, but
 there are other ways.
       So, in this case, you just gather information on a body system and use a
 multiplication factor to go from that to total damage. So, it may not be that critical to
know what the number of genes is.
                                     86

-------
 REGULATORY PERSPECTIVE

 Dr. Kerry Dearfield discussed the regulatory perspective of genotoxic risk assessment.
 Dr. Dearfield is from the Health Effects Division, Office of Pesticide Programs at the
 EPA in Washington, D.C., U.S.A.

       DR.  DEARFIELD:  Thank you.  I am really glad this meeting is being held
 because I think it is a very good  extension of the meeting we had in Melbourne,
 Australia earlier  this year with the  International  Conference on  Environmental
 Mutagens.  When we were there, we highlighted the real need for genetic risk
 assessments and why they are important in assessing chemical exposures.
       I  think, in this workshop, we are going to have  a  very  good working
 assessment of the  different chemicals that have been identified.  We will try to
 determine what some of the real numbers might be from the exposures from these
 particular chemicals. Also, I think this will provide us with real examples in trying to
 determine what is a real genetic risk.
       I think, as responsible genetic lexicologists, we have become very good at
 identifying genotoxic chemicals.  Just look at the proliferation of genetic tests that we
 have and how we have spent an almost inordinate amount of time trying to develop
these tests.  Just how many different ways can you identify a genotoxic agent?
       We have also been trying to characterize some of the genetic activity of these
chemicals.   It is very important to consider such things as mechanisms and how
chemicals interact with the genome.  These aspects are  crucial when trying to
determine how the chemical causes a genetic effect.
                                    87

-------
        A major  point I would  like to make is once  we  have identified and
 characterized the genetic effect, how do we communicate risk to the people who may
 be exposed to these chemicals? This is probably the major point we have overlooked
 in these many years in trying to do genetic risk assessment.
        We say we can do genetic risk assessment. We can determine how many
 induced mutations from heritable translocation tests, for example; but how do we then
 translate that to a phenotypic expression of that particular event in a human that has
 been exposed to that chemical?
        We haven't really done that very well, and I believe that is one of the reasons
 why this workshop is very  important.  We can  start making examples with the
 chemicals selected here that have real human  exposure and then extrapolate them
 to what it might mean to adverse outcomes in humans exposed to these chemicals.
 What Are the Reasons for Mutagenicity Testing?
 So, why do we do mutagenicity testing? From the regulatory perspective which is
 what I have been asked to  present today, and being from the Office of Pesticide
 Programs of the U.S. Environmental Protection Agency, we have devised a scheme
 of how we would try to, first of all, identify genotoxic chemicals. Then, hopefully,
 characterize the activity as best we can and enter into a genetic risk assessment. Our
 risk managers then decide whether we should take some type of regulatory action on
these particular chemicals. And that is where we need help from you all, the experts
in genetic risk assessments.  Major questions are:  How do we set levels of concern?
How do we determine if there  is a real genetic risk to the exposed human population?
                                    88

-------
        These are questions for later discussion, so let's go back to the beginning

 now.  Why do we do this? (Figure 2-30)
                         PURPOSE OF MUTAGENICrTY
                        TESTING FROM SUBDIVISION F
          For each test substance, required tests are necessary to assess the
   potential of a test chemical to affect genetic material. Results from such assays
   may be used as part of a qualitative and/or quantitative risk assessment. The
   objective underlying the use of the results from mutagenicity assessment are:

          (1) To detect, with appropriate assay methods, capacity of a chemical to
          alter genetic material in cells;

          (2) To incorporate these findings in

                (A) The assessment of heritable genetic alterations of concern to
                humans;

                (B) The weight-of-evidence approach for  a carcinogenicity
                classification of a  chemical when results from carcinogenicity
                studies are being considered.  Furthermore, mutagenicity testing
                information may be helpful in the selection of an appropriate high
                to low  dose risk  extrapolation  model  if the chemical is  a
                demonstrable carcinogen.
                                 Figure 2-30


       From our introductory materials to our mutagenicity risk assessment guidelines

from the Office of Pesticide Programs, we use these assays to assess the potential

of a test chemical to affect genetic material. This helps us to identify whether the

chemical poses a genotoxic concern  or not.
                                     89

-------
        We do this with appropriate assay methods to check the capacity  of the
 chemical to alter genetic material. Once we have done that, we move forward and
 incorporate these findings  into an assessment of  heritable genetic alterations of
 concerns to humans.  This is our genetic risk assessment approach.  This is  a very
 simple statement, but there is a lot of time, thinking, and resources that go into this
 apparently simple consideration. But it is not simple, as we know.
        Part B (Figure 2-30) presents somewhat of the conundrum that we have gotten
 ourselves into over the years as genetic toxicologists.  For many, many years, we
 have  used genotoxic evidence as part of the weight of evidence  approach for
 carcinogenicity classification, in particular, how best to  do extrapolations from high
 to low doses, for example.  This is where we have gotten into the areas of genotoxic
 versus non-genotoxic carcinogenic chemicals. This Is a whole different issue that isn't
 part of this workshop.
       But what I want to mention is that we have gotten ourselves into a frame of
 mind that genotoxicity testing is sort of a subset of carcinogenicity testing. That is
 something we have to rethink; while it is still very important for carcinogenicity, it is
 also important for all the other toxicities.  I will have  a slide later to show you that it
 is also important for not only cancer but for other things such as developmental,
 neurological, and other effects.
       We have to think of genetic toxicology as a major field that can provide insight
 not only for cancer but for all sorts of other toxicities.  I think that is one major
 message we need to get out; and when we communicate genetic risk, it is not just
for cancer anymore.
                                      90

-------
        So, this Is the background of why we do mutagenictty testing in the Agency.
 Now, to get into the actual details, our Office has come up with an initial battery of
 tests that we like to use to identify whether there is a genotoxicity concern or not
 (Figure 2-31).

 How to Determine if there is a Genotoxicity Concern?
 We have a minimum three-test battery.  It can be a little more extensive  if you like.
 Generally, it is anchored by the Salmonella test. We have a second test of in vitro
 gene mutation assay.  We generally look to the mouse lymphoma assay for this, but
 there are other choices that can be used.  One can  use other cell lines such as the
 AS52 cell line.  Also, you can use the CHO cell line in concert with an in vitro
 cytogenetic assay.
        I am  detailing this battery because I am going to mention these  assays as
 examples in  a parallelogram I will show in a few minutes.  The third test we use, an
 in vivo  cytogenetics test, is satisfied by what most people have been likely using, a
 mouse micronucleus assay.
        This is an initial battery to give us an  idea of whether there  is genotoxic
 potential for chemicals that are being tested for our purposes of pesticide registration.
 Now, if these tests are all  negative, then, most likely, there will not be much of a
 mutagenicrty concern and there probably will  be no more testing required of that
 chemical.
       Now let me put this initial battery into the larger scheme (Figure 2-32). In the
top row is the initial battery we just discussed.  We have a tiered approach here and
here are the second and third tiers. Once you have identified a concern out of the
                                     91

-------
            *ln Vitro Gene Mutation Assay (choice):

            (A) Mouse lymphoma L5178Y cells, tk locus,
                small and large colonies
            (B) Chinese hamster ovary cells strain AS52
            (C) Chinese hamster ovary (CHO) or
                Chinese hamster lung fibroblasts (V79)
                cells + appropriate in vitro test for
                clastogenicity

            * In Vivo Cytogenetics (usually in rodent bone
                marrow)(choice):

            (A) Metaphase analysis (aberrations)
                          Figure 2-31
   Salmonella
+     In Vitro     +      In Vivo Bone
   Gene Mutation     Marrow Cytogenetics
                          •Aberrations
                               or
                         • Micronucleus
Interactions with Gonadal DNA
              *
        Specific Locus
           •Visible
              or
        • Biochemical
                        Dominant Lethal

                                *
                      Heritable Translocation
                          Figure 2-32

                              92

-------
 initial battery, then we begin going into the characterization of that particular genotoxic
 effect.  In the second tier, we are looking for whether it interacts with germ cells and
 has potential for heritable effects.
        This does not erase the idea that these data still can support other toxicity
 concerns such as cancer, reproductive effects, and other toxiclties; but we are only
 discussing heritable effects now. That is something one needs to remember when
 coming into the Agency. Once you begin moving up the mutagenicrty scheme for
 heritable risk, you are moving out of the cancer realm.
        As one moves into the second tier, we want to start asking does the chemical
 interact with the germ cells. This isn't asking if it is heritable or not yet, but is there
 an interaction with germ cell DMA.  The arrows on the slide  indicate likely paths of
 direction we may take.  Say you would get a positive micronucleus test. More likely
 than  not, we might ask you to follow up with a dominant lethal assay. Other assay
 types include aberration studies in spermatogonia or spermatocytes, or SCE or UDS
 in germ cells.
        If these tests are negative, we have to make a decision. We would most likely
 say based on the information we have, there is no evidence for interaction with the
 germ cells.  Therefore, our concern for heritable effects would drop at this point. We
 may have to make the decision even though there might be limited data. That is what
 is required of a regulatory agency. Others may want to follow up if they like, if they
 have other ideas.
       However, if these tests are positive, then we move to the third tier  where we
could be looking at either specific locus tests and/or the heritable translocation test
                                      93

-------
 (Figure 2-32, bottom). These are quantitative tests because now we are going to try
 to identify and characterize a possible heritable concern.
        If these tests are negative, then again the concern for heritable effects drops.
 But if they are positive, we take the data and model it into a genetic risk assessment;
 that is, we do a quantitative risk analysis.  This is what our guidelines say we will do.
       While I have put the specific locus and heritable translocation tests at the third
 tier, this does not mean these are the only two tests you can use. Obviously, there
 are other tests that are possible for quantitation; but  for now, the  field of genetic
 toxicology has demonstrated to us, the Agency, that these are the most appropriate
 tests to use for quantitative risk analysis right now. However, if a case can be made
 for dominant skeletal effects or some other test, that is a possibility to use in the third
 tier.  Our scheme does not absolutely rule it out.
       How does this relate to the parallelogram approach? I have constructed a
 couple of parallelograms here, not exactly the one that Dr. Sobels had put together,
 but things that we think about in terms of assessing the risks the  chemicals have
 (Figure 2-33).
       We have a lot of in vitro animal data. A lot of times, we have in vivo animal
 data as well. Sometimes we have in vitro human data such as aberrations in cultured
 lymphocytes. We don't have too many in vivo human data, but we are starting to get
that more and more now with such things as the HPRT assay in human lymphocytes.
       In terms of the regulatory area, this top parallelogram (Figure 2-33) gives us
a good qualitative feel that yes, there is a genotoxic concern and that we need to
continue our efforts to see if we can do a human germ cell risk assessment.
                                     94

-------
           r	T	j  Qualitative
Regulatory battery looks at this
                                        Genetic
                                         Events
                                       Exposure
 Regulatory battery looks at this
         Figure 2-33
             95

-------
        Now let's look at somatic versus germ cells (Figure 2-33, bottom).  This is
  conceptually part of the parallelogram that Dr. Sobels constructed.  Many times we
  get in vivo somatic animal data.   We also have in vivo germ cell data which is
  generally from the heritable translocation test or specific locus test.  Sometimes, we
  have in vivo somatic human data. Most always, we do not have in vivo germ cell data
  for humans, so we are trying to make extrapolations to this particular point.
        We talk about specific genetic events such as clastogenic mutagenic events
  versus point mutations.  Also, this type of parallelogram can be used to talk about
  exposures (Figure 2-33).  I will show you more of those in a little while or more
  probably this afternoon when I  talk about acrylamide.
        I am showing  this  figure (Figure 2-33)  because,  as far  as the regulatory
 aspects go,  I think that we use the parallelogram approach without actually really
 thinking about it, without  actually  consciously constructing the battery with the
 parallelogram. I think that, conceptually, we have utilized many of the concepts of the
 parallelogram in our regulatory  battery.
       You can see that the testing battery that I showed you before captures all of
 these elements found within the dashed lines.  It gives you a good idea if there might
 be a risk for humans. So, once you have gotten the in vivo somatic data as well as
 the in vivo germ cell data as  shown at the bottom of our battery, you start making
 some estimations about the in vivo  germ cell risk to humans. So, even though we
 didn't construct our battery  with the parallelogram specifically in mind,  I think,
 conceptually, those of us in the field have been using these concepts over and over
 and over again in trying to make extrapolations from mouse to man, from high dose
to low dose.
                                     96

-------
 How to Determine Real fl/sfcs to Humans?
 So, why do we do genetic toxicology research and want to regulate chemicals that
 present a genotoxic concern? This figure is from the UNEP/ICPEMC (1992) document
 Dave Brusick talked about (Figure 2-34). These are examples of human diseases,
 conditions caused by mutation in germ cells.
        As you can see from this whole list here of genetic diseases and the number
 of cases just found in the  USA alone, that the genetic burden causes a lot of
 problems. This list runs from dyslexia all the way down to various cancers, cancer in
 general and specifically retinoblastoma.  But you can see  it is just not cancer; for
 example, there are a lot of neurological effects such as the fragile-X syndrome and
 manic depression.
        It is very important that we do genetic risk assessment, because it feeds into
 all of the other toxicities, and I think this slide highlights that very well.  There are
 many genetic diseases that are not all related to cancer alone. That is why, in the
 regulatory program, we want to identify the mutagenic process as one of the major
 points we look at.
       However, having said all that, it has been very difficult to identify human germ
 cell mutagens. We are very good at detecting germ cell mutagens in animal systems,
 but we haven't really identified a bona fide human germ cell mutagen. I have to admit
that is one of the major questions our Agency managers ask us.
      We say we have concern about human mutagens.   When asked  for an
example, it is difficult to respond because we cant identify  a  chemical that has
caused  a particular mutation that has caused a particular disease or an adverse
outcome through the germ line, even though we believe it happens.
                                     97

-------
                    EXAMPLES OF HUMAN DISEASES AND
                    CONDITIONS CAUSED BY MUTATIONS
                               IN GERM CELLS
   Genetic Disease or Condition
   Dyslexia
   Hardening of arteries
   Cancer
   Manic depression
   Schizophrenia
   Juvenile diabetes
   Adult polycystic kidney disease
   Familial Alzheimer's disease
   Multiple sclerosis
   AAT deficiency
   Myotonic muscular dystrophy
   Fragile X chromosome syndrome
   Sickle-cell anaemia
   Duchenne's muscular dystrophy
   Cystic fibrosis
   Huntington's disease
   Hemophilia
   Phenylketonuria
   Retinoblastoma (childhood eye
   cancer)

   From: UNEP and ICPEMC (1992)
Estimated No. of Cases in USA
     15,000,000
      6,700,000
      5,000,000
      2,000,000
      1,500,000
      1,000,000
       500,000
       250,000
       250,000
       120,000
       100,000
       100,000
        65,000
        32,000
        30,000
        25,000
        20,000
        16,000
        10,000
                                Figure 2-34


       Here are a couple of reasons why it is difficult to identify them (Figure 2-35).

There is a rarity  of situations which permit  enough  detectable mutations, thus

reducing the effect of an epidemiology study.
                                    98

-------
       There is the rarity of the individual genetic diseases with marker genes to
 identify associated genetic disease. Also, there is difficulty in identifying and studying
 a suitable population exposed to substantial levels of mutagens. That has been a real
 problem, but I think with some of the examples we have at this workshop, we are
 beginning to identify some of those populations.
                       DIFFICULTY OF IDENTIFYING
                     HUMAN GERM-CELL MUTAGENS
   1)

   2)

   3)
Rarity of situations in which enough mutations are induced to be
detected in an epidemiological study.
Rarity of individual genetic diseases and of 'marker' genes identified as
associated with a genetic disease.
Difficulty in identifying and studying suitable populations exposed to
substantial levels of mutagens.
   Therefore, it has not been possible to link any human genetic disease with
   specific mutations induced by radiations or chemicals.
   From: UNEP and ICPEMC (1992)
                                 Figure 2-35

       For example, with acrylamide, we have some very highly exposed small
populations we can use in our exposure assessment.  So, I think we are beginning
to address point 3 (Figure 2-35).  The problem is that the individual genetic disease
                                     99

-------
 linked to a specific genetic marker that has not been found yet that has been induced
 by a particular chemical.
        Just to drive that point home a little bit, I would like to show this little paradigm
 (Figure 2-36) because this demonstrates why It is tough to sell our approach to risk
 managers. From animal studies, we know a particular chemical can cause a particular
 genetic alteration, a heritable translocation, for example, in the ethylene oxide case.
 We also know in the human situation that a particular adverse health effect has
 associated with it a specific genetic alteration, for example, the liver cancer induced
 by aflatoxin B1, which has a very specific point mutation in the ras locus in the liver.
 But we cannot make this final connection between a particular chemical which causes
 a particular genetic alteration that is associated with a particular adverse outcome.
 This is where we are missing the human mutagen. This is what we have to deal with.
 Intuitively,  we know ft occurs, but we haven't got the examples yet.
        Having said all that, what we want to do is start trying to provide examples to
 show  how we would do a genetic risk assessment.   So, a group  of us from the
 Agency got together and performed a genetic risk assessment with ethylene oxide as
 an example (Figure 2-37).
       Notice in the title  it says ethylene oxide as an example.  This is not the
 definitive risk assessment of ethylene oxide, but it was something we wanted to have
the genetic toxicology community start thinking about, start focusing on.  For
example, what is needed to do a genetic risk assessment? What are the appropriate
studies that need to be done to obtain appropriate information to  do genetic risk
assessment?  More importantly, finally, what do these numbers  mean?
                                     100

-------
    Chemical     —^    Genetic Alteration(s)
                                                      9
Specific Genetic  	^   Adverse  Health
     Alterations                        Effect(s)
                               Figure 2-36
                                 Environmtnlal and Molecular Mulageneits 16:104-125 (1990)
          Quantitative Estimation of the Genetic

           Risk Associated With the Induction of

                 Heritable Translocations at

            Low-Dose  Exposure: Ethylene  Oxide

                          as an Example

               Lorenz Rhomberg, Vicki L Dellarco, Cheryl Siegel-ScoH,
                  Kerry L Dearfield, and David Jacobson-Kram
          Office of Health and Environmental Assessment (L.K., V.L.D.. DJ.-K.), Office of
            Toxic Substances (CS.-S.), and Office of Pesticide Programs (K.L.D.), US
                  Environmental Protection Agency, Washington, D.C.

     This paper explores how quantitative risk as-  alkyloTmg ogenl, expected equivalency of doses
     sessmert methods' might be extended to analysis  across species,  germ-cell sensitivity, and «x-
     of risks to the human germ line. High inhalation  tropolation of dose-response relationship to low
     exposures to ethylene oxide ore reported to  exposure levels. Various dose-response models
     cause heritable translocations in male mice with  ore discussed in terms of their applicability to
     a steep and nonlinear dose-response curve. We  genetic end points and their ability to reflect the
     explore quantitative estimation of risk to humans  underlying basis of induced heritable Iranslcca-
     from law exposures based on these animal data,  tions.
     addressing questions of tissue dosimetry for this

            Kty words: genitic riik aueiiimnt, gum-cell mutogwiieily, genetic toxicology
                               Figure 2-37


                                  101

-------
       I like to put this example up (Figure 2-38). You all will probably be seeing this

 continually for the next day and a half, but this example crystallizes, I think, the whole

 thinking about how you would approach a genetic risk assessment.
                        ETHYLENE OXIDE

   Upper bound estimate of the risk from continuously exposing
   a male parent to 0.5 ppm EO for eight hours a day, five days
   a week for three weeks is:

                            2.8 x1(T*

   or, 2.8 offspring with  heritable translocations  per 10,000 live
   births from exposed  fathers.  This estimate represents the
   additional risk of heritable transiocations which are induced by
   exposure to  EO excluding spontaneous,  or  background
   genetic damage.

   Calculated from assessment is:
   "A quantitative estimate of the genetic risk associated with the
   induction of heritable translocations at  low-dose exposure:
   Ethylene  Oxide as an example".

   Published in Environmental & Molecular Mutagenesis 16: 104-
   125, 1990.
                            Figure 2-38


      Just assume everything is correct in our assumptions here. We know we have

made a lot of assumptions, like Dave Brusick said earlier.  We also made a lot of

naive simplifications. We said there were no dose rate effects for example, but we do
                               102

-------
 know there are dose rate effects with ethylene oxide; but just assume that everything
 here is right.
        If you have a male who is continuously exposed to 0.5 ppm for eight hours a
 day, five days a week for three weeks, you obtain a 2.8 x 1O"4 risk or 2.8 offspring with
 heritable translocations per 10,000 live births from exposed fathers.  I will just say 3.
 It is easier to conceptualize 3 births than 2.8. This represents an additional risk over
 background.
        So, now  we know a number that  is correlated with a particular  human
 exposure.   But really, what does this mean in terms of the person who has been
 exposed to 0.5 ppm and subsequent risk?
        Remember the OSHA's PEL, permissible exposure level, is 1 ppm.  So, you are
 telling exposed persons that at one-half of that which is a very real occupational
 exposure,  they are going to have heritable translocations in  an additional 3 births in
 10,000.  How do you communicate that risk?
       As you saw from David Brusick's slide earlier, you can translate this number
 into diseases. In our example, we considered what happens  at a higher exposure.
 When you  get exposure at about 100 ppm, you double the background rate.
       Now, the background rate was calculated to be 19 x 10**.  Is a doubling of the
 background where you really express concern and tell the exposed person there
 might be problems? These are questions that  are real tough, especially in many
 instances a regulatory mandate behind it says if you feel that is a  real level of
concern, then we have got to intervene and reduce the exposure to these people.
So, I think  this is a very good example in trying to highlight where is  the level of
concern that you want to intervene at.

                                    103

-------
        From another perspective, you can also say that 3 x 10"* is within the variability
 of the background. The background, as I mentioned, is 19 x 10"*. So, do you or how
 do you protect 3 over 19 in the human population? That is a very tough question to
 address.
        More importantly, how does this number 3 get translated into a phenotypically
 expressed adverse health consequence? I think that is a real difficult thing to do right
 now, and that is why we are trying to have ICPEMC give us some guidance, some
 process that we all can use and get an idea of what we want to do with the whole
 process.
        So, I think with this workshop we are going to build on this particular example
 with other examples and try to get everyone thinking about the important pieces that
 you need to do a genetic risk assessment and how you would extrapolate that to
 human beings.  In addition, there is a lot of research involved with this.
        DR. NOLAN:  Kerry, is that 19 in 10,000 live births?
        DR. DEARFIELD: Yes, 19x10"*. That came out of the Mary Lyon et al. paper
 from ICPEMC (Lyon et al., 1983). I had an earlier draft of ICPEMC's report and it had
 this figure here (Figure 2-39).  Dave Brusick redrew this graph, but it is  the same
 picture he showed earlier.
       Like I said, there is a lot of research that can be done. I think that is one of
the things this workshop will need to identify. What are some of the things we can
do?
       Well, I think these risk extrapolation factors (Figure 2-39) comprise one area
that we can address with research. We can use the mouse model and examine such
                                    104

-------
Human
REFs
Mouse Model
                      Route
                      Matrix
                      Dose/Rate
                      Bioavailability
                      Metabolic
                      Specificity
                      DNA Repair Capacity
                      Dose-response Kinetics
                      Target Size
                      Sex/Strain Specificity
                      Germ Cell Stage Specificity
                      DNA Lesion Homology
                      Locus Specificity    „ - -


                   .-
                      Disease Homology
                      Mutation Expression
                      Private Genes
                            Pharmacokinetics
          Germ Cell Risk Extrapolation From Mice To Humans

         The presumed process of genetic disease induction in
         humans modeled from experimental data derived from
                               Figure 2-39
                                  105

-------
 things as  pharmacokinetics.   We  need  to  determine  if the  same  set  of
 pharmacokinetics that you see in the human is modeled by the animal systems.
 Concerning DMA reactivity, is that similar in mouse and in man? We can go down
 that whole  list ICPEMC assembled here.  These are all the different things that are
 being looked at.  I now, for example, for ethylene oxide, the Ethylene Oxide Task
 Force from industry is very interested in providing us with some of this information.
 They are looking at some of the pharmacokinetics, at the metabolic specificity, and
 other aspects.
        So,  these are research  avenues that can be taken to help us refine what
 measured germ cell mutations mean for new diseases. Actually, this is another whole
 extrapolation in going from the induced amount of heritable translocation, for example,
 in the human  being to what that means in terms of new disease.
       These  are all the different types of information we need to be thinking about;
 and for me, from the regulatory  point of view, these are the real tough things to talk
 about when I  brief our risk managers.  It is very easy to say yes, it  is Salmonella
 positive.  We  have followed-up for example with a dominant lethal  test.   Now  a
 heritable translocation test has been done, and now we have this ethylene oxide risk
 assessment, for example.
       Let me point out the difficulty with communicating this example from a briefing
 we gave to our risk managers on the EO quantitative risk assessment.  Ethylene oxide
 is in what we call special review in the Office of Pesticide Programs. It is a fumigant
 for spices. It is actually not a real heavy-duty use in terms of pesticide use. However,
 we are interested, obviously, in hospital workers and others; but right now, because
the courts have mandated it, OSHA takes  care of hospital workers and not us.  So,

                                     106

-------
 we are only looking at spices, but we are trying to work this genetic risk assessment
 up anyway.
        We have an EO carcinogenicrty risk, and this is 2.6 x 10"2.  K is fairly high, for
 all of you who are familiar with carcinogenicrty risk assessments.  This might actually
 drive the whole risk assessment, and that is an issue that Dave Brusick mentioned at
 the very end of his talk.
        As an aside, in carcinogenicrty risk assessments, and additional question that
 is thought about is if you regulate on the basis of cancer, does it actually protect you
 against genetic risk?  That is a real question we need to ask, because there are a lot
 of resources to do both assessments.
        Right now, industry is approaching both assessments (that is, cancer and
 mutagenicity) and  they  are  part of our regulations.   However, routine  heritable
 translocation or  specific locus  testing  is not fully accepted yet.  There is real
 resistance to the top tier. Again, this is fall-out from the problem where we haven't
 really communicated that these are the tests that we need to  use to quantitate
 heritable risk, and it is not just for cancer.
        Right now, I cant tell you which way that debate is going, but if something
 else is going to be used as a surrogate for genetic risk assessment, then I don't know
 what genetic lexicologists are in business for. That sounds pretty dire, I admit, but
 I think we have some of the answers.
       Back to the EO example,  let me get to the numbers. This is our number, 2.8
x 10~* and these are our assumptions. We made a conservative assumption that each
heritable translocation will lead to an unacceptable health condition.  That is a very
conservative assumption!

                                    107

-------
       When you think about it, this is only heritable translocations. We didn't factor
 in specific locus mutations, for example,  or other  events we know ethylene oxide
 causes. So, this estimate may actually be an underestimation; but maybe because
 it is so conservative a number, it may actually be an overestimation.  We just dont
 know.  So, this is why we need efforts like this workshop to get a handle on this
 number.
       The major response from the risk managers was that 2.8 x 10"4 is pretty high
 risk.  But they are thinking in terms of 10* for cancer, but you are mixing apples and
 oranges when comparing to genetic risk.  But cancer risk is historically where these
 risk managers have  experience.   We haven't  really educated them on genetic
 risk—mainly because we haven't considered it much ourselves and do not have many
 examples.
       We have to figure out what the numbers we  generate really mean in terms of
 adverse outcomes.  This is a real problem when you try to communicate genetic risk.
       The last point I want to make with  the EO example is we have other effects
 that we are concerned with.   EO does have reproductive and developmental
 problems.  There are reports of neurotoxicity problems.  No risk assessment have
 been made for that, but these effects may  have impact on a total assessment.
       As a final note, I think we need to communicate out there that genetic risk
 assessment is not just for cancer anymore.  All these genetic effects by ethylene
 oxide, or whatever chemical, can cause a whole spectrum of effects, and we have got
to make sure that people out there understand that. That is why genetic toxicology
and associated research is so important.
                                    108

-------
       We cant allow genetic toxicrty to be subsumed into these other fields either,
 because I think, as a basic mechanism, we can contribute to whatever field versus
 just being part of that field. We can collaborate with these people examining adverse
 effects.
       This is the sort to perspective I have.  Right now, we can easily identify
 genotoxic concerns, but I think it has come to the point where we have to understand
 how to communicate genetic risk. We have gotten to the point in our testing where
 we  are generating numbers,  and we are asking to spend a lot of resources to
 generate those numbers. We have got to decide on what those numbers mean in
 terms of real risk out there for the people who may be exposed to those chemicals.
       Thank you.
 Discussion

       DR. NOLAN: Thank you.  I will take a couple of brief comments on that. Udo?
       OR. EHLJNG:  I agree, of course, with the tendency you were saying, but  I
 have one question and one remark.
       The question is I am very impressed with the testing scheme of EPA, but does
 the Agency enforce it? Who does the third tier test? I am  very interested in getting
 this answer.
       Okay, and the statement is about the animal experiments. We know from the
 Russian example that is used, that in the offspring, you  couldn't show a significant
difference between the exposed and the unexposed. No one in radiation protection
would say okay, there is no effect. We didnt see it in the man.  There, we accept the
animal data.

                                   109

-------
        The same is true in the Sellafield case. There, you have human data, but the
 final judgment is based on animal experiments, and the case is decided that there is
 no cause relationship between the leukemia and the radiation exposure, but again,
 the essential data comes from animal experiments.
        Why should it be in genetic toxicology so different to point out these cases?
        DR. DEARFIELD: I think that is a good point. It shouldn't be different. Let me
 answer your first question. Under FIFRA anyway and under TSCA, we do have the
 legal clout, if you will, to require those particular tests, all the way up to the top tier
 of the battery. If we have sufficient reason to move up to the next tier, we will require
 the companies to go ahead and have their chemical tested.
        Now, up to the second tier, there is no problem. Anybody can do those tests,
 and most contract labs have that capability, and there are a lot of labs in the industry
 that can do it in-house as well.
       The problem comes down to the specific locus and heritable translocation or
 any other of those top tier tests. I thing the heritable translocation can be done a little
 bit more easily than the specific locus.  We have almost come up to a specific locus
 problem where we are trying to contract through Oak Ridge. The problem we had
 was with GLPs.
       You have to remember when you do a regulatory test that things have to  be
 done with the GLPs, and that is a real different consideration than just doing a straight
test in  a research laboratory. There is a difficulty in that, and the best  answer  I can
give you right now is it a case-by-case basis.  I depends on the capabilities of the
laboratories involved and who else is willing to do it.
                                    110

-------
       I  know, in some cases, some of the industrial laboratories are contacting
 themselves out now, and I would love to be able to see if we could do the same type
 of thing  with some of the  national  laboratories, for example, with your  group in
 Munich,  because I think it would be a perfect place to try to do something or at Oak
 Ridge, because that is not a trivial test. It is a problem, but we can require them to
 do it if we believe it is necessary.
       DR OSTROSKY-WEGMAN: From your talk, the impression  is that we need
 more human studies, and the main problem remains that usually in the First World,
 we have people that are exposed to very low levels of dangerous substances. As an
 example, there  is  a  wonderful study by Marja  Sorsa  on  workers exposed to
 cyclophosphamide, and she couldn't find anything, because the levels were low and
 the protection was good.
       Now,  we do have people in the Third World exposed  to high doses  of
 substances, and I do think  we should promote a little bit more to do collaborative
 studies in humans exposed  to substances without protection.
       Da DEARFIELD: I think you are absolutely right. As a matter of fact, we have
 been talking with different people about trying to begin sampling of some of the highly
 exposed  populations  in Eastern Europe, for example,  and  also  in China.  For
 acrylamide example I have here, there is a highly exposed group of people in the
 United States.
       You are right, we need to get  some highly exposed human populations on, I
don't think, every chemical but at  lease enough chemicals to give  us  enough
confidence that we can extrapolate back and,  hopefully, validate the animal studies.
                                    111

-------
So that when we do animal studies in a prospective sense, we can make better
judgments on the human exposure without actually having to expose the humans.
                                 112

-------
Research Background for EO

Dr.  Lars  Ehrenberg provided  the research  background for the ethylene oxide
assessment.   Dr. Ehrenberg is from the Department of Radiology at Stockholm
University, Stockholm, Sweden.

      DR. EHRENBERG: Mr. Chairman, ladies and gentlemen, I will speak about the
data base of doses of ethylene oxide I was asked to give you for this meeting and
then say  something about the limitations and critiques of the use of Hemoglobin
adducts.  I will use the other adduct dose-rate effects which have been referred to by
several speakers here, and I will say something, in case of time, about the importance
of ethylene oxide, not only as an example but also, in general, about sources of this
compound (Figure 2-40).
                        Ethylene Oxide (EO)
                    In Vivo Doses:  Data Base
                    Hemoglobin Adducts
                    Limitations and Critique
                                              i
                    Dose-Rate Effects
                    EO: Sources, Importance

                              Figure 2-40
                                 113

-------
 The Use of Hemoglobin Adducts
 We use the measurements of adducts in order to estimate in vivo doses, but since our
 data available, (for instance, for epidemiologic studies and what we want to control
 by various risk limiting measures) is based on exposure we have to establish (this
 relationship between exposure and adduct level//n vivo dose, (Figure 2-40). Also, the
 risk produced and risk to exposure relationships need a good exposure assessment.
        In the case of a precursor of an ultimate mutagen such as ethylene oxide (A
 in Figure 2-41), we have also to know the rate the conversion to the ultimate mutagen,
 and we have the reaction of a specific mutagen with a specific site J in DNA for the
 formation of an adduct (Figure 2-42).
        There  is  a  relationship  between the cumulative  degree  of the specific
 alterations in DNA and the risk, and this cumulative alkylation is a function of the rate
 of formation of the adduct and the time integral of the concentration of the ultimate
 mutagen. This is what we call the dose.
       So, we express the dose  in concentration times time, for instance, in mg/kg
 body weight times hours. In short-term experiments where we can allow for the rate
 of repair, and may determine dose from an injected or ingested amount from the value
 of this rate constant (kg) and measuring the adduct level. With the hemoglobin as a
 surrogate dose monitor, it has the advantage that the molecules are not repaired.
They are more long-lived, and from the steady state level of adducts from long-term
exposed persons, we can estimate the contribution to the adduct level (a) per unit of
time and then evaluate the dose as for DNA.
       There are several experiments, though, that show the similarity  between the
levels, at low doses, of adducts to DNA and to hemoglobin.  This is from the  study
                                     114

-------
Exposure **->- Adducts •< - >• Risk
           In Vivo Doses
          RX — >- detoxified product
             Figure 2-41
     R,X -f DMA, —*> R, - DMA, + X

[B-DNAJcum
[DMA]

        j_  ^  [R, -DNAJ
        k,   *   g DNA
       D,=
          i -HbJ
          gHb
             Figure 2-42
               115

-------
 of Potter and colleagues (Figure 2-43).  It has to be pointed out that this is a short-
 term experiment where you can allow for repair.
        For other data bases available, I will not go into much detail.  I show here the
 data for adduct levels in rats (Table 2-1), with respect to adducts to N-terminal valine
 in hemoglobin. In some cases, we calculated that from measurement of some other
 adduct to hemoglobin and give data for cases where we have measurements of DMA
 adducts.
       These are now expressed per ppm hour, that is, per unit of exposure dose,
 and that gives a rather consistent figure of about 15 pmol per gram per ppm-hour in
 rats and a figure which is approximately the same or a little higher with respect to
 DMA adducts.  It should be about twice as high because of a difference  by a factor
 of 2 in the  rate  constants. So, there is still some uncertainty.
       Another important thing is that here, the lower figure (index c in Table 2-1) here
 is from the  experiment of Potter and colleagues, and this is from the studies of Walker
 and colleagues where very high levels were used; but you see that in the gonads, the
 dose is about half of that in the average of the body. There are similar data for the
 mouse.  I will come to that.
       When we look at the human data (Table 2-2), we have the great problem of
 the exposure assay. When we make the corresponding calculations of pmol per gram
 per ppm hour, we see a variation of the adduct level, from about 2  up to about 14.
 From various deliberations, we believe that the higher figures are the  correct ones.
 I cannot go into all  the arguments for that  here, but there  is  a tendency  to
overestimate exposure doses with the effect that the calculated in vivo dose per unit
of exposure dose becomes too low. I think that is enough for that. The tests where

                                    116

-------
     Relationships between Alkylation of Hb and DNA by Ethylene Oxide
<
z
Q
LU

X
                    I	I
I     I     I     I     I
                       n.moles N -HEH/g Globin
     Relationship between alkylation of DNA (n.mol N7-HEG/g DNA) and
     of globin (n.moles N'-HEH/g globin) in Fischer 344 rats exposed (6
                h, inhalation) to rC]EO (1,10 or 33 ppm)
                            Figure 2-43
                                117

-------
we get the higher figures are from those studies where we have the most careful

measurements of exposure dose.
Table 2-1.  Adduct Levels in Rats Exposed to EO.

                Strain, exposure condition    HOEtVal

  Inhalation
                                                       7-HOEtGua
  IP injection
                F344
-------
 Table 2-2. Human Data, Hb Adduct Levels from Ethytene OxkJe
                                          HOBVaJ
cxpoHure
situation (na of
exposed/no, of
controls)
Stsrftzation
(5/4)

(10/10)
(11/3)


Dally (7/9)

Occasional
(8/9)
Nurses (9/8)



(4/0)


Petroleum
industry

uonunntnon
TWASh/d
(pom)


6-30

0.02-0.3
<0.1-0.7


17

<17

0.15
0.05
0.1

1-ca 2





(nmovgj




5-20

0.05-0.4
low-2


13.2
±2.6
2.7
±1.9
0.1
0.15'
0.23

1.7-
ca 10




{pmovyj/ppmn




2

7
. 7 (3-15)
(20(pmol/g)/
(prnoi/kg)]
2



2
8a
6

5-ceL 14


12 (10-14)
P2(pmovg)/
(MmoVkg)]
i minimi ii




Calleman et al.
1987
Sarto et al 1991
Duus et al. 1989


Tates et al. 1991




Mayer et al. 1991
Hagmar et al.
1991

Osterman-Golkar
and Bergmark, 1
van Slttert al. to be
published

 ' Adjusted for difference in calibration method (Tdrnqvlst et al. 1992)
       In order to sum up this data base (Figure 2-44), we have to consider that the
 dose in blood of humans or animals is a constant times the alpha, the fraction of the
 inhaled amount  which stays in the  body, which is not exhaled or excreted; the
alveolar ventilation (V.J, the exposure dose; and in the denominator, the volume of
distribution (A'), a basic pharmacokinetic concept; and lambda, the rate of elimination
of the compound.
                                    119

-------
                   1  -h)  -  const           A'*-25 (L mol'1)
                                  Figure 2-44
        Now look at the various estimates we have of these constants, in Table 2-3,
 where we compare these parameter values for mouse, rat, and man.
        We see here that the target doses of the valine adduct level per ppm-hour, that
 is, per unit of exposure are approximately the same if we stick to these higher human
 values, as I have said before. We have also data here for hydroxyethylated guanine
 in testis.  This is the mouse value from Dr. Sega's experiments a few years ago, and
 we can see that there is generally a lower level  of adduct in the products.
       In this equality of values per ppm-hour, the equality in level of adducts per
 ppm hour is due to two effects which operate in opposite directions:  the longer
 lifespan of adducts with the increasing lifespan of erythrocytes, (that is, of hemoglobin
 when you go from mouse via rat to man), and  the decrease in alveolar  ventilation,
 (that it is the amount of air inhaled per minute per kilogram of body weight), and the
 decreasing rate of detoxification when you go through the species.
                                                  i
       The data give thus a rather consistent picture. I would say they are consistent
within about a factor of 2, and they might, therefore, be used for risk estimation.
       In the recent studies of Walker et al, they have criticized the use of hemoglobin
adducts for several  reasons.  Here is data for rats and mice (Figure 2-45), the rats
exposed to 300 ppm for four weeks.  This is the expected rate of disappearance in
                                    120

-------
the 60 days lifespan in the rat, and this is the found one (which decreases much

faster).
Table 2-3.    Best Values for Parameters and for Increments of Adduct Level or
  t.r (days)

      (L g'V1)
  V^ (L kg'V1)

  A' (L kg'1)
  [HOEtVal],
  increment from
  1 ppmh
  (pmol g"1 ppmh'1)

  [HOEtGua],
  increment from
  1 ppmh
  (pmol g"1 ppmh'1)

  Dose from 1 ppmh
       ppmh'1)
 Dose from
 1 /jmo1/kg
* ^»i h^^r^r*r*« **
Mouse
40
3.2-10-5
80-100%
60
-1
4.5
•^ww «ri VIMFW^^I Mr^^^fl •
Rat
60(55-65)
4.5 -10"6
-80%
36
<1 (0.5?)
3
^ «^«* ^
Man
120 (126)
5(4.5) -10-5
-80%
6-9
?'
1 (if A'-1)
12
20 (testis)
0.5 (blood)
0.2 (testis)

0.5 (liver)
0.2 (testis)
16
12-24
10 (testis)
0.35 (blood)
0.10 (testis)

0.30 (blood)
0.12 (testis)
12
.(?0.05-)0.3
(blood)
                                   121

-------
A 40°
£ 300
0
O)
ra 200
| 100
0
B 140
120
c
•§ 100
CD
E 80
[5 60
X
I 40
20
0
"2\^^-—
; V_^
_ —
0 24 6 6 10 t
-S^^.
1 V ^^D -
1^^ ^
— —
— —
— ' —
0 24 6 8 10 1
400
300
200
100
0
2
140
120
100
80
60
40
20
0
2
Time (Days postexposure)
Comparison of the projected loss of N-(2-hydroxyethyl)valine to the
observed loss of adduct after 4 weeks of exposure of rats (A) to 300
ppm ethylene oxide and mice (B) to 100 ppm ethylene oxide.
Projected ( °,a); observed (•,•). Data points are mean ± SE(n=5).
The derivation of the curves for the projected loss of
N-(2-hydroxyethyl)valine is discussed in Walker et al. (23) and
Figure 2*45





  122

-------
        However, these very high levels of adducts are something we have never
 encountered  before.  We see a toxicrty to the erythrocytes develop, and this is
 something one should be careful about when working with very high levels.
        Something similar was seen in our study (Figure 2-46),  a study carried out
 mainly by Dr. Osterman-Golkar, on the Bushy Run rats which had been exposed for
 two years to  10, 33 or 100 ppm of ethylene oxide,  30 hours per week.  There is a
 drop in the curve at the highest dose level, the very highest concentration, which
 could be due to a toxicrty of this kind. We have not seen that kind of effect in rats
 following acute exposure or short-term exposures. They are not expected to be there
 at low levels of exposure, that is those which are most interesting (Figure 2-47).
        One should say, however, that there are mammalian species where the kinetics
 of zero-order kinetics of hemoglobin turnover is not at hand, but where there is a first-
 order of decay after exposure.
       Another rather remarkable statement in one of the papers by Walker and
 colleagues is  that you cannot  use hemoglobin adducts to predict, at any time after
 an exposure, the DNA adduct level. Here is the expected level of hemoglobin adduct,
 and you will see that the DNA adducts in the different organs decay at different rates,
 because of repair (Figure 2-48).
       Who has asked  to be able to use hemoglobin adducts as something
 completely parallel  to the DNA adducts?  You can use each of these.  You can
 calculate the cumulative dose if you determine first the kinetics of the change of the
 quantities.
       I will jump over some of the figures to make it a little shorter. There are other
pieces of criticism which could be easily handled.

                                    123

-------
     100 •
:§>
          10    33                 100
                     ppm

              F344,6h/d, 5 d/wk, 2y
              Figure 2-46
  Hb Adduct Levels in EO-Treated Rats
           4 x 26 mg EO/kg
  40
  30
  20
  10
„  HOE + Val
    nmol/g
                         6    weeks
              Figure 2-47
                  124

-------
                 350

              <  300

              ^250
              o>

              §200
              HI
              I  150
              o
                 50
-^^  HG
       ^  ^_
                      0246
               35


            < 30
            •••^M
                        2468
                           Time
                  (Days postexposure)
Persistence of 7-(2-hydroxyethy!)guanine in exposure of rats (A) to
300 ppm ethylene oxide and ethylene oxide. Brain (A), lung (O),
leukocytes (*), sp liver (A), testis (X). Data points are means ±
SE(n)
                      Figure 2-48

                           125

-------
        I should like to say a few words on dose and effect. Here is what everybody
 has in our documents, I think, for dominant lethal mutations as a function of ethylene
 oxide exposure, discussed from Dr. Generoso's work, and we have a strong increase
 upwards (Figure 2-49).
        What could be the causes of this convexity? If we look at possible causes of
 non-linearity (Figure 2-50), we have that there could be saturation of the detoxification
 and saturation of DNA repair which we suppose make the curve turn upwards like
 that.  There could be induction of detoxification that will turn the other way around,
 induction of repair which could affect the curve at very low doses, thereby increasing
 the risk at very low doses or decreasing risk at very low doses (Figure 2-51). ft could
 be a true two-hit effect which is involved,  ft could also be a loss of damaged cells.
 I  may come to that later, but you see often that curves drop at very high doses
 because of the damage to sensitive cells.
        Here (for  EO), we  don't discuss the additional effects  of saturation of
 bioactivation and so  on.  These ones are valid for an ultimate mutagen, such as
 ethylene oxide, and they are all causative to the influence on mutation rate of dose
 rate.
       We see such  effects if we measure hemoglobin adducts  as a  function of
 injected amounts, for example, in a rat study in cooperation with Dr. Tates. At about
 25 mg/kg, there is a switch upwards (Figure 2-51).
       One can now look at the data from Walker et al, we have this bend upwards
 here (Figure 2-52). The line I have drawn here is to see what a linear curve would be.
 It varies at the highest dose in the mouse  exposed for four weeks; here it is higher
than expected.  For DNA adducts, the deviation is still higher.
                                     126

-------
       JO
       CO
       f
        CO



       I
               0     50   100   150  200   250   300

                    EO Concentration (ppm)


Concentration-response curves for induction of dominant lethal
mutations in male germ cells. A, matings with T-stock females;O,
matings with (SEC x 101)F, females.
                                      i

                       Figure 2-49
                            127

-------
       Deviations from Linearity
Potential Causes:                 F
  - Saturation of detoxification
  - Saturation of DMA repair
  - Induction of detoxification
  - Induction of repair
    Error-free
    Error-prone
    True two-hit effects
  - Loss of damaged cells
  (For pre-mutagens:
  - Saturation of bioactivation
  - Induction of bioactivation)
    All these effects are influenced
              by dose rate
                 Figure 2-50
                     128

-------
Hemogtobin adduct (HOEtVal) increment per
     injection, in experiment with rats
  Relation between Hb Adduct Level
       and Dose of EO in Rats
   100
JO»
   50
             26
 50 60
mg/kg
80   100
      Non-linearity; saturation of
      detoxification at high doses

               Rgure 2-51
                  129

-------
                                               - o
          0      20   40    60     80  100   120

              Exposure Concentration (ppm ETO)
Comparison of the dose response of N-(2-hydroxyethyl)valine in
hemoglobin and 7-{2-hydroxethyl) guanine in DNA of rats (A) and
mice (B) exposed to ethylene oxide for 4 weeks (6 hr/day, 5 days/
week). Globin (D), brain (A), lung (O), spleen (•). Data points are
means ± SE(n=3 to 5). Note the scale of the X-axis is different in A
and B and that the scale of the Y-axes in A are adjusted so that the
data points for N-(2-hydroxyethyl)valine and 7-(2-hydroxyethyl)
guanine overlap at 3 ppm ethylene oxide.

                          Figure 2-52
                               130

-------
        If you now calculate from the elimination kinetics a kind of joined effect of the
 two curves, using the data of the curve from Generoso's study (Figure 2-53).  Let me
 recalculate here the real dose we  have got at the DMA level where we also consider
 the prolongation of the akylated condition in the DNA as an increased DMA dose, that
 is, looking at the DNA dose over time as a measure of the degree of alkylation. So,
 the curve becomes much less steeply changing upwards which should be taken into
 consideration when you apply a formula for estimating the risk at very low doses from
 such a curve.
        When  I was here, I think, six years ago, there was a discussion of ENU as a
 mutagen (Figure 2-54). This is a most effective point mutagen in the  mouse with
 respect to induction of specific locus mutations.  In this rather complicated figure
 (Figure 2-54), I have drawn the curve for mutations per locus per rad of radiation as
 a function of dose rate in rad per  minute and then the curve for ENU from the data
 of Bill Russell and colleagues where I put equality between number of alkylations of
 guanine-O6 giving rise to the same  mutation frequency as one rad of ionizing radiation
 at high  dose rates.
       Here is the data of Russell  with 20 times 5 mg/kg injected, so what I did was
to make an extrapolation of this curve, calculating the linear component  of the
function for that curve.

Ethylene Oxide Data

      Here is now Liane Russell's ethylene oxide data; the question was whether or
not you  got an increase of specific locus  mutations following the treatment with
                                     131

-------
60


50


40


30


20


10


0
        -r*

                           w
                          t   O Dominant lethals in the study
                         r        of Generoso et al.

                               x With doses estimated from
                 500
                              1000
1500
                    Figure 2-53
 Mx107
 per locus
 perrad
or rad equ.
                     w               •/
                    X              T /
                          T   T^ JP5*— 20X6|M

                i	i	1^1     i	i	i
                                        lENU
                                        20 x 5 ng/kg
               -3   -2   -1E°0   -1-1   -1-2   -t3

                                        Log (rad/min)
                    Figure 2-54
                         132

-------
 ethylene oxide at very high doses.  The  dose rate in that study is here, and the
 question was are these levels (of EO and the expectation form ENU) significantly
 different. They are at about the 5 % significance level, that is, on the edge of being
 significantly different.
        Here (at lower dose rates) are effects which are of a different kind.  The
 bending upwards here is due to effects found for both radiation and ethylene oxide
 for inducibility of repair. So, at very low dose rates, we may have a higher mutagenic
 effectiveness.
        In the work on plants 25 years  ago, we saw  that when hydroxyethylating
 compounds  were  compared  to  methoxyethylating  compounds,  they  were
 approximately equal in mutagenicity (Figure 2-55). However, there was an enormous
 difference with  much  stronger frequency of chromosomal aberrations when you
 introduced hydroxyethylations.
        In efforts to understand this, there was a model developed (Figure 2-56) where
 you have hydroxyethylation of phosphate groups in DNA that leads to the formation
 of a cyclic phosphate compound with pentavalent phosphorus which then breaks to
 give rise to a chain break.
       Furthermore, this end of the strand carrying the adduct is alkylating.  ft is a
 phospho-triester, which is alkylating and could alkylate the other strand. So, you have
the possibility, perhaps, of a double strand break being formed.
       This is a mechanism which we are studying further. I will finish with  one table
(Table 2-4), of importance, I think, to epidemiological studies. I think of Dr.  Brusick's
presentation here and so on, and we have quite well limited ourselves to the industrial
exposure (to ethylene oxide), and that is the reason for the discussion.

                                     133

-------
M  CA
75   15  -
50   10  -
25    5  -
                                     Chrom.
                                     Aberrations (CA)
                  100       200        300

                       mM,24hat18°C
   A* HOCH2CH2

   AO MeOCH2CH2
(From Gichner et al. Heroditas 59,1968,253)
                    Figure 2-55
                        134

-------
dR-O-f-O-dR-

  ?  r
  CH2-CH2
                             9
-dR-O-P-O-dR- -> -dR-Of + HO-dR-
   CH-CH
CH-CH
(dR, deoxyribose; B1, B2, bases.)
        Model Alkylators:

         CH3SO2OCH2CH2OH

         CH3SO2OCH2CH2OCH3
         CH3SO2OCHCH2OH

                 CH3

         CH3SO2OCHCH2OCK
                 I     *    *
                 CH,
         H2NCON-CH2CH2OH

               NO

         HoNCON-CHoCHCH,
               NO   OH
              Figure 2-56
                 135

-------
 Table 2-4. Annual Dose of EO in the 8.5 Million Population of Sweden
   Source
   Chemical and
   pharmaceutical
   industry
   0.1-1 ppm EO

   Ethene in air,
   average 2 ppb,
   mainly from burning
   organic matter

   Endogenous
   production of
   ethene (partly from
   gut flora and unsat.
   dietary lipids)

   Cigarette smoking
   10cig. cf1;0.25mg
   ethene/cig
Individual dose
of EO (g a'1)
0.2-2
0.0003
0.007
0.03
                                           (N)
      Collective dose of EO

                  (kg a'1)
250
8.5-106
8.5-106
2-106
0.05-0.5
2.5
60
60
       I have here in the last column summarized for the Swedish population of 8.5

 million, the amounts, in kilograms, of ethylene oxide from various sources in the whole

 population annually, and at 1 ppm of ethylene oxide during working  hours.  In

 approximately 250 workers we have in the ethylene oxide industry in Sweden, there

 will be 0.5 kg. This is a level to the whole population.

       If you take ethane inhaled, 2% of the inhaled amount of which is converted to

ethylene oxide, that gives 2.5 kg. If we take the endogenous production of ethylene

oxide,  partly from the intestinal flora, and  partly from the diet, we think due to

unsaturated fatty acids, it is 60 kg per year. That is 100 times more than the industrial

contribution.
                                     136

-------
       So, this is something which we might find to give mutations collectively of
importance.  This corresponds to the contribution  from ethylene in the cigarette
smoke from smoking by about 2 million cigarette smokers which is about a quarter
of the population smoking cigarettes.
       Perhaps this ethylene oxide to the population would be one of the mutagens
operating in these curves which show the frequency of mutations of new dominant
mutations, as a function of father's age, from, Friedrich Vogel's work (Figure 2-57).
       Now, these are autosomal; none of them is x-chromosome inherited where the
maternal grandfathers play a role.
       We calculate that endogenously produced ethylene oxide could be the initiator
in 1 % of the cancers in knowingly unexposed persons, and the question is which are
the others.
       Thank you.
                                    137

-------
     4x
O—O Achondroplasia
• • • -• Achondroplasia
       (new series from USA)
 I—• Apert's Syndrome
     3x
co
    2x
     1x
                                    Population
                                    • •» •• ^ ^ ^ «D
                                     average
              24     29     34     39    44    47
   Father's age
                      Figure 2-57
                            138

-------
 EO Mutagenicity Risk Assessment

 Dr. Vicki Deliarco presented an Ethylene oxide mutagenicity risk assessment.  Dr.
 Dellarco is from the Human Health Assessment  Group, Office of Health and
 Environmental Assessment at the EPA in Washington, D.C., U.S.A.

       DR. DELLARCO:   The previous speakers have laid down  a very good
 background for me to present this ethylene oxide genetic risk assessment.
       As  Dr. Dearfield  mentioned, this is something we did as an example risk
 assessment a couple of years ago. This assessment was published in Environmental
 and Molecular Mutaoenesis in 1990 (Vol. 16, pages 104-125), and is in your handout.
 It was not a formal risk assessment per se, but we wanted to develop a framework of
 thinking  for quantitative genetic risk  assessment to highlight the issues and
 uncertainties (many of those were pointed out by Dave Brusick); furthermore, we
 wanted to promote  discussion on these  uncertainties and issues  and suggest
 approaches for genetic risk assessment and eventually identify data requirements and
 future research directions.
       Why did we choose ethylene  oxide?  It is not a bad model.  Also, it is a
 reactive alkylating mutagen, it is widely distributed to all tissues, its pharmacokinetics
 are relatively simple, there is human exposure, and more so, there is a large data
 base on  its genetic  toxicology.   Within that data base, we chose the heritable
translocation test in the  mouse, because ethylene oxide was evaluated at several
dose levels, and  it was given by a  relevant route of human exposure, namely
inhalation.
                                    139

-------
        Before I get to the assessment, I just want to point out that we were not the
 first to attempt to look at the genetic risks of ethylene oxide. Dr. Ehrenberg published
 a  risk assessment back in  1974 where he  used the radiation dose equivalent
 approach, taking mutation data from barley and molecular dosimetry in the mouse.
        Then, shortly before we published our assessment, Thomas Smith published
 a genetic risk assessment based on pharmacokinetic modeling using DMA alkylation
 in the mouse testis, Dr. Generoso's dominant lethal data, and Dr. Sega's DNA single
 strand break data to come up with a risk calculation.
        This morning, Dr. Wright pointed out some of the steps that we have to take
 in genetic risk assessment, and so what I am going to do is apply many of those
 steps to this example. The first thing you have to do is define response which I have
 already done; it is "heritable translocations" in mouse germ cells.
        Then you have to define the dose. This is critical not only to this approach but
 the parallelogram approach and involves deciding what the actual target is, selecting
 the dosimetrics, and trying to establish a relationship between external exposure to
 dose at the target (which Dr. Ehrenberg has gone over). It involves some concepts
 concerning the window of the response, where the adducts  accumulate, and dose
 rate issues.
       Once you  have established that, you have to determine equivalent doses in
 experimental animals and humans; in other words, what is the dose in mouse and
 human that will cause similar risks. Then you take the heritable translocation data
 and try to establish mathematical relationship; in  other words,  define the dose-
 response.  This involves considerations of mechanisms  of  action, how heritable
translocation are produced, and a concept called "additivity" which I will come to later.
                                    140

-------
        Then, once you do all of that, the final step in a genetic risk assessment is to
 determine what the potential human health impact is; in other words, what is adverse.
 Dr. Dearfield emphasized how important that is for us  as a regulatory agency.
        This is the translocation study in the mouse and the sequential method was
 used to identify translocation carriers. Male mice were exposed to ethylene oxide for
 six hours a day, five days a week for six weeks and then daily for ten more days.
        As you can see from this figure (Figure 2-58), the dose response is very steep,
 and it flattens out at lower exposures. There is some background, but it is very low.
 All doses tested were significant over the background.
        So, the question is how you take these data and estimate the risk. You must
 consider the various extrapolations that are going to be necessary  to do  this.
 Because we are using mouse data,  we  are  going to need to do a species
 extrapolation: the mice experiment was conducted at high exposures, so we are going
 to need to do a high to tow dose extrapolation: and the last calculation is a mutation
 to disease extrapolation.
        Now, the species extrapolation involves a couple of assumptions.  The first
 assumption we make up front is that the biological mechanisms that link levels of
 ethylene oxide exposure to risk of heritable translocations are common in mouse and
 man; in other words, the mechanisms that are causing, the induction of the  breaks
 that lead to the formation  of a translocation that would be  passed on to the F1
 offspring are similar in mice and man. So, we are going to have to make assumptions
 about DMA repair, cell cycle kinetics, and transmission to the offspring.
       This isn't so unreasonable to do despite repair differences among mammalian
species. If you  look at all the in vivo data that you have for inhaled ethylene oxide in

                                     141

-------
30 -r
25--
20--
15--
10--
 5--
       ppm   Incidence  % Translocations
  0
165
204
250
300
 1/2068
32/1143
52/1021
 88/812
109/427
 0.05
 2.79
 5.09
10.84
25.53
            50       100         165     204
                        EtO Exposure (ppm)
                                          250
                                            300
                         Figure 2-58
                             142

-------
 various species, you can see there is a commonality of genotoxic responses in test
 animals and humans.
        So, because you see this concordance of different responses, it gives you
 confidence that, in general, similar mechanisms are operating. The similar data that
 we have in humans, are micronuclei induction in exposed humans and in the mouse;
 we have sister chromosome exchanges being observed across several  species,
 including humans; and chromosome aberrations are also observed in sites in humans
 and the monkey. Although the rat cytogenetics are negative, in every system ethyiene
 oxide has been tested, it has been positive.  That is the exception.
        So, in essence, our species extrapolation  Is going  to become one  of an
 interspecies dosage calculation; in other words, we try to determine equivalent doses,
 but we have to define dose, and there are the questions that need to be addressed:
 What is the biological effective target?  Is it DNA?  Is it protamine?
       We went with the hypothesis that the target is DNA, however, one cannot rule
 out protamine as the target because of the work that has been done by Gary  Sega.
 Dr. Sega has shown that there is a concordance with protamine alkylation and
 dominant lethal mutations. Despite that, we think it is still reasonable to assume DNA
 is the target because ethyiene oxide produces mutations and cytogenetic damage in
 somatic cells where you dont have protamine, and furthermore, DNA adducts are
 detected in germ cells. Nevertheless, this is a major assumption that can affect the
 modeling, but we are going to go with DNA.
      The next question is what is the biologically effective dose? In other words,
how do adducts accumulate in germ cells?  When you talk about dose, you just can't
                                    143

-------
 talk about dose in nominal units like parts per million. You really have to epitomize
 dose and generalize it over time.
        So,  is there a window that is  critical for adduct accumulation? We used
 dominant lethal data to determine this. I We have to define the window over which
 adducts accumulate.
        This involves looking at different germ cell stages throughout spermatogenesis,
 and, as you know, there are  stem cells, spermatocytes, spermatids, and  mature
 sperm. Ethylene oxide produces a post-stem cell effect and, looking at the next figure
 (Figure 2-59), you will see the germ stage sensitivity pattern from the mouse dominant
 lethal test for ethylene oxide. This is just the germ cell stage pattern. At the  bottom
 is the post-mating schedule in days, and Y axis is dominant lethal frequency, and in
 the boxes at the top represent the germ cell types that are being sampled.
       As you can see from this study, ethylene oxide dominant lethals are induced
 in mid to late spermatids to early spermatozoa.  So, the mouse window is 10 days.
 So, we are going to assume that there is not a stem-cell effect, where adducts would
 accumulate over the reproductive life of the entire exposure history of the individual.
       We are fairly confident in this assumption. Most of the chemical mutagens that
 have been evaluated to date basically show post stem-cell effects, but we cant rule
 out stem cell effects unequivocally.
       There was a study that Dr. Waldy Generoso did where he exposed  mice to
ethylene oxide for two weeks and then eleven weeks, and he found a higher dominant
lethal response for the eleven weeks of exposure.  Now, that could  have been due
to a stem cell effect, or It could have been due to some pharmacokinetic factor.
                                    144

-------
     30
 Oi

 o-
 Ol
_  20
 ID
O
o
                                         I  * I  »*jl I
                   S          10         IS         20



                   Post-Exposure Mating Schedule (Days)
                        Figure 2-59
                           145

-------
        So, we are going to assume in the mouse that a ten-day window exists.  The
 next question is what is the delivered dose? We are using the dosimetric ppm»hour,
 and this involves looking at the relationship of external exposure to target dose.  If the
 delivered dose is  proportional  to  external exposure,  then we  can use  external
 exposure as a surrogate. So, we are looking for proportionality.
        We didnt do any sophisticated pharmacokinetics modeling or analysis.  We
 basically relied  on the work of  the  modeling  done by Dale Hattis  in 1987 which
 assumes linear pk.   Although Hattis'  model  doesn't  include  predictions of
 concentrations of ethylene oxide in  testis, the model results for other tissues might
 predict that.
        During inhalation  exposure, as has been discussed, ethylene oxide is readily
 absorbed and it reaches a fairly rapid steady state. The tissue concentration is in
 balance with the constant external air concentration.
        So, assuming proportionality at low to moderate exposures isn't such a bad
 assumption.  Also, Smith's model had low dose predictions, and concentration of
 ethylene oxide in the testis is proportional to external air concentration after a brief
 equilibration period. The assumption of proportionality is a very  critical one in risk
 assessment.  For the dosimetric ppmehour, we have to make the assumption that
 there is no dose rate effect.
       Where our approach fails is looking  at risk to high episodic exposures,
because there is some non-proportionality at very high concentrations of ethyiene
oxide.
                                     146

-------
       So, for the ppm»hour dosimetric, we have to assume that response is a
 function of exposure multiplied by time and this is not true if there is a dose rate
 effect.
       As Dr. Ehrenberg discussed, there are dose rate effects. This is shown in a
 study that Dr. Waldy Generoso did for dominant lethal mutations  (Figure 2-60).  Dr.
 Sega looked at DNA strand breaks and UDS and yes, he saw a dose rate effect, too;
 this effect was less as the total exposure was lowered. Gary went down to a 300 ppm
 hour exposure looking at sperm head DNA alkylation and saw a very slight dose rate
 effect.
       So, like radiation, it appears that as you reduce the total dose, the dose rate
 effect diminishes.  In rabbit somatic cells you dont see a dose rate effect.
       Now, what we did is we summarized all the  mouse dominant lethal  data that
 we had on ethylene oxide (Figure 2-61).  The X axis is the total hours of inhalation
 exposure, the Y axis is percent dominant lethals,  and the numbers here are the total
 ppm«hours.
        If there is no dose rate effect these lines would be horizontal.  The 7200 ppm
 total hours is the dose rate study that Waldy did for dominant iethals. You can see
 from these limited data that short intense exposures produce a dose rate effect lower
 than longer exposures totalling the same ppm»hour. The dose rate effect appears
 to diminish with lower total exposures and longer times.  So, we think it is reasonable
to assume no dose rate effects when you are looking  at low constant, continuous
exposure scenarios.
      Now that we have determined our dose and we are using  external exposure
as a surrogate for target dose, we have to determine the species equivalence of dose

                                    147

-------
ETO DOSE-RATE  STUDIES
CELL TYPE
HOUSE GERM CELLS:
HALE
POSTHEIOTIC


RABBIT SOMATIC
CELLS:
Peripheral
Lymphocytes
END POINT
DOMINANT
LETHALS
DMA Strand breaks
G Unscheduled ONA
Synthesis
SPERM HEAD
-ONA Alkylation
Sister Chromatic!
Exchange
TOTAL
EXPOSURE
7200 ppn/hr
1800 ppm/hr
300 ppn/hr
48. 000 ppn/hr
EXPOSURE
RATE
300ppn X 6hr X 4days
600ppn X 3hr X 4days
1200ppa X l.Shr X 4days
4SOppn X 4hr
900ppn X 2hr
ISOOppn X Ihr
75ppm X 4hr
ISOppn X 2hr
300ppn X l.hr
200ppn X 6hr X 40days
400ppn X 6hr X SOdays
ISOOppm X .5hr X 64days
DOSE-RATE
EFFECT
YES
(6X)
YES
(4X - 2X)
SLIGHT
(2X)
NO
      Figure 2-60
       4V         6U
       Hours of Inhulution
      Figure 2-61
         148

-------
 in mouse and  man.  The  assumption that you have to make here is  that the
 probability  of  inducing  a heritable translocation  is  a function of adduct load
 irrespective of the species;  also, one  must assume that ppm»hour of  external
 exposure to the concentration multiplied by time product in the testis is the same for
 mice and man.
       Therefore, external exposure is a surrogate for the biologically effective dose,
 and we believe that the Hattis and Smith models support the species equivalence of
 steady state blood concentration. Also in support of this is the hemoglobin alkylation
 data.  Here we have plotted all the human hemoglobin alkylation data that we had
 (Figure 2-62). Many of these studies came from Ehrenberg and Osterman-Golkar.
       You can see that even though this is a log-log plot, for ppm»hours, there
 appears to be proportionality among  different  species.  Even for comparable
 ppm«hours, the rate of hemoglobin alkylation among these different species is
 proportional to exposure level even at quite high air concentrations. So, we feel this
 is an indirect support for the proportionality assumptions that we are making.
       In the next figure (Figure 2-63), we took all the sister chromatic exchange data
 that we had at the time and we then plotted it in ppm«hours. We think that this also
 helps legitimize the species extrapolation that we  are making in that even though it
 is a log-log plot, we feel there is fairly  good agreement between ppm*hours and
 genotoxic damage. This is a very crude plot, because we are dealing with human
 data and with very gross estimations of exposure.
       So, therefore, we can go ahead and scale our dose across species just using
 external exposure as our surrogate. The 10-day window in the mouse  would equate
to about a 21-day window in humans. For humans, there is a more prolonged period

                                    149

-------
10,000

1.000
s*
0
X
^ 100
•S?
"o
3 10
1
9
UkJ
i

0
^_
O
—
A Mouse *
• Rat A
— O Rabbit
A Human AA
A * A
A
A
A **
• A
•—
A*
1 1 1 1 i 1
                              10        100        1,000

                           EtO Exposure (ppm-hours)
                                                            10.000
                                                                      100,000
                              Figure 2-62
    10
 a
^.  is
I
•o
a
•o
c
• Rat
O Rabbit
A Monkey
A Human
                             100        1.000       10,000

                              EtO Exposure (ppm-hours)
                                                           100.000
                                                                     1.000.000
                              Figure 2-63
                                  150

-------
 to accumulate adducts. So, when you calculate your ppm»hour exposure, you would
 just multiply It out over 21 days even if the humans were exposed for 2 years.
        Now I will go to the dose-response modeling.  We simply could take a ruler
 and just draw a straight line, but that is overly conservative. It ignores the information
 being provided by the higher dose points, and you would like to factor in biological
 considerations when you extrapolate your dose response curve to the low dose
 range. Also, it doesn't make any sense, because you would have a sharp break
 between here and the straight line and that would be hard to explain biologically.
        So, what we tried to do here (Figure 2-64), is to fit the data on a empirical
 basis using curve fitting models, and we chose the multi-stage model, which is a
 modified polynomial model that has a great deal of flexibility.
        So, we are just looking for the best fit to the curve, and that is represented as
 "M".  Again, that approach is somewhat unsatisfying, because we are not taking any
 biological considerations in. So, we looked for another model to use where we would
 find some biological reasoning to apply to the selection of the model.
       As shown in the next figure (Figure 2-65), simply to get heritable translocation,
 you have to have breaks  in two non-homologous chromosomes that are  in close
 proximity and we are selecting a model on the basis of that.
       We went to the Weibull model which has been used in cancer; and we think
 that the logic that is applied in cancer is somewhat similar to the logic that you would
 apply in this case  in that,  in cancer, you have many cells that are  at risk,  and you
 have to have certain rare events coinciding before that cell becomes transformed.  In
this case, the nucleus is divided up in little boxes, as shown in Figure 2-66,  and you
                                    151

-------
      0.25 -
      0.20 -
VI

O
O
•55
o

O>

O>
      0.15 -
                      5,000
10,000
15.000
                     ETO Exposure (ppm«hours)
                        Figure 2-64
                           152

-------
Biological Consequences of Rearranged Chromosomes
        Infertility/low fertility
        Spontaneous abortion/miscarriage/stillbirth
        Developmental effects (e.g., skeletal, neurological)
                      Figure 2-65
            MECHANISTIC CONSIDERATIONS
             TWO CHROMOSOME "BREAKS"
                 IN CLOSE PROXIMITY
                      Figure 2-66

                         153

-------
 have to have these two chromosome breaks occurring in close proximity; once that
 happens, you get a transiocation-bearing sperm.
        In the next figure (Figure 2-67), we used the Weibull model. A linear quadratic
 model does not fit the data. It has to be a model with a fourth power, and the Weibull
 is represented here by a "W. I am not going to go into these models too much. We
 probably could have chosen some other models for curve-fitting purposes, but I would
 like to make a point about additivity.
                          P(d) = 1 - e
                      Weibull (Extreme Value Distribution)
                                 Figure 2-67
       The issue came up about the background level of genetic damage. Does the
 so-called spontaneous noise level of genetic damage, participate with the ethylene
 oxide-induced damage to give rise to heritable translocation risk?  You have to have
 two chromosomes, and you have to have two breaks arising in dose proximity.  One
 is induced by ethylene oxide, but can  the  other be induced  by this so-called
 background level?
       If the case is the background damage does not contribute to the risk, in other
words, independent background, then you can see that the curve does move down
very steeply; in other words, the two chromosome breaks both have to be induced
by ethylene oxide.
                                   154

-------
        We tended to favor the assumption of additive background, in other words, the
 background damage contributes to the risk, and I think what supports this are the
 data that were presented on ethane; that there is this endogenous and exogenous
 exposure to ethane which produces ethylene oxide.  So, it is reasonable to assume
 that  you dont have to have both chromosomes broken by the ethylene oxide
 exposure.  However, this is an issue that needs to be discussed further by this
 workgroup.  Some individuals think the background is dependent.  If we were wrong
 about this assumption, you  could see that the independent background essentially
 just lays on  its axis, and K shows essentially no risk or very little risk at all at lower
 exposures.
        Table 2-5, to shows you,  again, the big  difference between an additive
 background versus an independent background assumption. You would predict that
 at 1200 ppm*hour total, about 7 translocation carries for 10,000 offspring versus 0.1,
 respectively. So, that assumption makes a big difference in the risk assessment.

 Table 2-6. Excess Risk per 10,000 LJve Births
EXTRAPOLATION
METHOD
Linear extrapolation
from lowest dose
-Multistage Moder
Weibull Model
Additive Background
Independent Background

OSppm
OOppnvftr*
2.1
0.79

0.21

-------
 For the purposes  of this assessment, we  just chose  a factor of 1,  that every
 translocation carrier would have some adverse effect.
        We know that is really conservative, and, in fact, that the factor is probably less
 than 1, but how off are we? I believe that Julian is going to get into some of these
 issues later on in his talk.
        Another assumption that we are making that I consider a major weakness in
 this assessment is that in the population both males and females are at risk and the
 data we are using comes from male mice only; and we know that the male is not a
 good surrogate for female risk, but we have to assume that the female is at a similar
 risk.   We  certainly  need  more studies on the female.   Because  we  know
 gametogenesis between males and females is very different,  metabolism of the
 chemical in the female  germ cells is different,  chromosome configuration  or
 condensation is different, and the biochemistry is different. That is a lot of differences.
 Also,  this assessment  is limited,  because we are  only looking at  heritable
 translocations, and we know that ethylene oxide causes other types of genetic effects.
 There have been some studies  looking at the effects of ethylene oxide on inducing
 specific locus mutations.
       The ultimate germ cell stage in the female is the zygote, and there have been
 some studies by Dr. Waldy Generoso looking at exposure of ethylene oxide to the
 zygote; and in  Table 2-6, you see that there is a high frequency of a congenital
 malformations that are induced in addition to mid to late gestational deaths.
       This fetus has some limb defects. This one is the normal.  This one is growth
retarded, and this one has hydropy.  In fact, there was quite a high incidence of
                                     156

-------
 hydrops after ethytene oxide exposure.   This may not be a transmissible effect;
 nevertheless, it is an effect on the offspring.

— - -»
TreejiimiK
EO
EMS
DES
MMS
ENU
TEM
X-ray**
Control
noauipUoii
bodteefK)
53
31
24
25
49
57
58-66
5
Mid-andtate-
nnarftrflnn
DHBtaQQli
deaths (K)
26
31
37
2.5
2
Z2
1.2-2.0
1.5
LVefatuaas
exarmnad
87
231
157
451
33
112
326
3,436
UveMuaee
WffJl
dofuct (K)
37
29
24
3.8
15
10
4.6
1.3
* fi hr rvMatmatlnn ovrar* Inr Y-rau whtoh tuaa 9 It hr
       b Data for 150 R and 200 R doses were pooled.
       Dr. Dearfield was pointing out that the cancer risk assessment for ethylene
 oxide seemed to be the driving factor in regulatory decision making. It only takes 900
 ppm for 1.5 hours, to get the zygote.effect. This could be much more sensitive than
 the cancer effect, particularly for short-term exposures.
       This is a big issue because there is the thinking that the cancer risk calculation
 is more protective given the non-threshold mathematical approach,  and because
 nobody has shown that there is a germ cell mutagen that doesn't cause cancer; but
we dont have a large data base on germ cell mutagens, so we don't know if that is
true, yet that is the thinking that we have to deal with.
                                     157

-------
        Ethylene oxide might not be the best case for germ cell risk, but what about
 acrylamide? Is the cancer risk more protective or the genetic risk? I dont know. We
 need to look at that. So, in some cases, the cancer risk may protect for the genetic
 risk, and in other cases, it may not at all;  but we won't know that until we start
 generating some genetic risk assessments and doing those comparisons.
        Okay, so let me just summarize by going back to the outline I had about the
 steps  that we took, because many of the assumptions that we made in this risk
 assessment are relevant to the parallelogram approach and I just want to go through
 this again.
        Nature of the target:  we are assuming Is DMA. If it is protamine, there may
 be a threshold response.  Maybe you have to accumulate a certain  adduct level
 before you get a break.  If so, that, you would take a different approach for modeling
 the data.  So, we could be wrong there.
       I don't think we are so wrong in assuming proportionality for low to moderate
 exposures; and true, we didn't do any pharmacokinetics modeling, but  that wouldn't
 improve upon this assessment.  The window of response; we are assuming a post-
 meiotic effect, and there may be a stem cell effect, and that could be argued. That
 would  have a tremendous  impact on the dosimetric that we used and the dose
 calculations that we did.
       We are assuming no dose rate effect, and that is okay, I think, as long as we
are looking at risk to low-moderate continuous exposures; but if you are looking at
high episodic  exposures,  that is  where this  assessment fails,  and  maybe
pharmacokinetics analysis can explain that non-proportionality and correct for it. This
                                    158

-------
 is also an issue in the parallelogram approach when you extrapolate across species
 and there are the acute exposure scenarios.
       The additivity assumption; there is disagreement, that the background is not
 additive, but independent resulting in a lower risk.
       What is adverse?  Well, we are using a factor of 1. We need to discuss that
 later in the workshop.
       So, maybe through the course of this meeting, we can identify the approaches
 that should be taken for genetic risk assessment and what sort of data we should
 generate.
 Discussion
       DR. NOLAN: Thank you very much. Any short comments?
       DR. DELLARCO: Well, I was going to say that since Julian is going to critique
 this assessment, some of the issues, we can open it for comment after that and just
 maybe a few questions if there are urgent questions now.
       DR. NOLAN: Good idea.  Just the one comment, then, from Lars Ehrenberg.
       DR EHRENBERG:  In this, the formations in the effects, was that following
 exposure of the teammates of the conception?
       DR DELLARCO: Right, it was 6 hours, 6 hours after mating.
       DR. EHRENBERG:  There were spontaneous abortions in females exposed
to ethylene oxide during pregnancy, but they were exposed also during the period of
conception. It is difficult to say what the consequences are of it.
       I think this story is important, because it was published ten years ago, and
nobody has tried to repeat it.

                                    159

-------
      DR DELLARCO: That is why I wanted to emphasize it in this talk, because I
think it is a critical effect of ethylene oxide.
      DR. NOLAN: One last question, then.
      DR ANDERSON: I think you can use the males in just the same way showing
the pups, because you get very nice data after low dose exposure, continuous
exposure, and I think that is very relevant to this sort of assessment, because it is a
disease state which may afflict males.
                                  160

-------
 A Reconsideration of EO Risk Assessment

 The final Plenary session was a reconsideration of EO Risk Assessment by Dr. Julian
 Preston.  Dr. Preston is from CIIT in Research Triangle Park, North Carolina, U.S.A.
        DR. PRESTON:  I would like to explain some points to ponder on the risk
 assessment model that you just heard.
        This is my version  of a parallelogram (Figure 2-68), and one I wanted to
 emphasize since the four corners for response have been talked about- these various
 associations with the dotted lines and the full lines. It should be noted that this figure
 is only for reciprocal translocations, and what I will mention in a little bit is a rather
 hefty distinction, between mutations and heritable translocation, when you are looking
 at risk assessment models.
        In two places, I have added underlying mechanism, and that is the underlying
 mechanism for the formation of translocations in somatic cells and in germ cells in
 humans and in rodent species which, I think, form a major component of the means
 of  extrapolation.  You can extrapolate by effect, and you can extrapolate by
 mechanism.  I am  going  to try and throw in a little bit of mechanism to add to the
 effect for our extrapolation considerations.
       We have heard  quite a lot about radiation responses and the parallelogram
 approach for radiation effects.
       This figure is to remind me of two or three items (Figure 2-69). This is a dose
response curve for reciprocal translocations induced in spermatogonial stem cells and
analyzed in primary spermatocytes.  These are some data from Grant Brewen and
myself some years  ago.  It  is to  remind me  that  we  have been  looking at
                                    161

-------
        GENETIC RISK TO HUMANS BY EXTRAPOLATIONS


             Example for chromosomal translocations
    Reciprocal translocations
    In rodent germ cells
              • -> Reciprocal translocations
                 in human germ cells
r
    Reciprocal translocations  <-
    in rodent lymphocytes
              -> Reciprocal translocations
                 in human lymphocytes
V
                         1
    underlying mechanism <-
              -> underlying mechanism
J
  Reciprocal Translocations Induced in Mouse Germ Cells by
  EO (Data from Generoso et al., 1990, Environmental Molecular
  Mutagenicity 16,126-131)
                         Figure 2-68
            r * LOB • io~4o » 5.00 - tcr'o2
         0.20
         0.10-
                  200
400     600     600    1000
     DOSE(R)
                                                    1200
    Dose Response Curve for X-Ray-lnduced Reciprocal
    Translocations in Mouse Spermatogonial Stem Cells (from
    Preston and Brewen, Mutat. Res., 19,215-223,1973)
                         Figure 2-69


                            162

-------
 parallelogram approaches for  a long  period of time;  but it is also important to
 remember that  here, as we have heard several times, the dose is  very easy to
 measure.
        We know something about dose, dose to the target tissue, if not necessarily
 to the target cells, and with radiation, the effect is produced in the cell of interest at
 the time of  the irradiation.  The translocation is  produced, in this  case,  in the
 spermatogonial stem cell at the time of exposure to radiation.
        That makes for a very different way in which you can consider the response
 curve, and here you can see a nice dose response formula which has a linear dose
 component and a dose- squared component.  You can define it, because you know
 something about dose and dose distribution.
        Thus it is (Figure 2-69) to remind me that radiation is simple, and chemicals
 are a lot more difficult to assess risks from exposure. The reason is that there is a
 very different mechanism of induction of the end-product. A translocation induced by
 radiation is induced in the cell type exposed and close to the time of exposure, and
 it  is a feature  of  errors  of repair.   Repair  errors will  be the  mode by  which
 translocations or mutations are produced by radiation.
       That is not the case with chemical exposures, as most of you are well aware,
 but it certainly affects the type of modeling that we do.

 Translocation Mechanisms and the Stages of Spermatogenesis

So, understanding  radiation effects is  relatively simple; chemicals  are a lot more
difficult.  Why do we have to consider mechanism, and why do we have to consider
all the different stages in Spermatogenesis?

                                     163

-------
        Figure 2-70 illustrates the steps of spermatogenesis. This is my linear version
 of spermatogenesis, but what I want to emphasize is not only the different stages in
 spermatogenesis,  but also where I have indicated what  I think are the critical
 components, that is, cell division and DMA synthesis.
        Let me return to the saga here  and explain that for radiation, the effect is
 produced in the target cell in any of the stages of spermatogenesis, but for chemical
 exposures, it is necessary to have a replication phase in order to produce your end
 product of interest, either a translocation or a mutation. Of importance here also is
 to note for replication errors that will be required to produce a translocation, you have
 to have two events on separate chromosomes. For a mutation, you need a singular
 event where an error of replication will produce your mutation.
        So, the sites of DMA synthesis are the critical components, and they will relate
 to the probability of getting your endpoint of Interest. For example, this is why I think
 any risk assessment model must  consider separate components of, in this case,
 spermatogenesis, because  they will all have very different probabilities for risk
 assessment.
       Here, for example, is the stem cell, and I would like to just deal with that briefly,
 because I really do think it is a critical component of the spermatogenic cycle when
 considering risk assessment You must understand that although a number of studies
 have been performed with the mouse to  look for effects in stem cells, they have
 largely been hefty doses and rather acute exposures, and I will just demonstrate why
 I think it is  important to consider long-term chronic exposures for the stem cell.
       These are just points to ponder.  Figure 2-71 shows the average somatic cell
or differentiating spermatogonial cell cycle  divided up into the G,, S, Gg, and mitotic

                                     164

-------
         SPERMATOGENESIS
               Stem Cell
             Spei
Differentiating
Spermatogonia


    •4f
SpermMocyles


    I
   irmatids


    I
 Spermatozoa  Q


 Fertilization

    I
   Zygote
                     Mitosis
                     (S^hase)
                                 Mitoses
                                (S-phases)


                                 Melosls
                             (pre-melollc S-phese)
                                 Mitosis
                                 (S-phase)
A Diagram of the Various Stages of Speratogenisis Showing
the Points at Which Cells Undergo an S Phase
                       Figure 2-70
                 CELL CYCLE
    Average Somatic Cell
Differentiating Spermatogonia
                          Spermatogonial Stem Cell
                                                     Mitosis
          24 Hours
                                   8 Days
 A Comparison of the Cell Cycles for a Somatic Cell and a
 Spermatogenial Stem Cell
                       Figure 2-71

                          165

-------
 phases.  The cell cycle duration is given as 24 hours, which is a very mean number
 for a somatic cell.  You could consider it less or greater than that, but let's use 24
 hours as an average.
        Now, if one is interested in alterations to the DNA,  either adducted DNA or
 strand breaks (in this case for ethyiene oxide), we will consider adducted DNA; this
 is the  critical component for producing a mutation or  a  translocation.  Damage
 induced  in the  S phase has a reasonably high probability of being involved in a
 replication error. The probabilities for cells in other stages of the cell cycle will be
 significantly lower because of repair that can take place prior to replication.
        So, the probabilities for a differentiating spermatogonial cell for producing a
 mutation or a translocation will be related to the total cell cycle and the proportion of
 that cell cycle which is a DNA synthesis phase. So, for differentiating spermatogonia,
 there will be some probability of having the opportunity to produce an error.
        How about the stem cell? Well, it has approximately an eight-day cell cycle
 in a mouse. I don't know the cell cycle time in the human.  I would presume it is
 rather long.  Note here  that the DNA synthesis phase represents a very  small
 component of that cell cycle.  So, a large proportion of the cells will be subject to
 repair prior to replication.
       Similarly, if you use a high exposure situation, you have a high probability of
 killing the cell that is in DNA synthesis because of associated lethal responses. So,
 although you have a higher probability of producing a mutation or a translocation, you
 have a higher probability of killing that same cell because of a secondary event.
       The probability of having coincident events in the stem cell is a lot smaller,
especially under conditions of chronic exposures. The  probability  of producing a

                                      166

-------
 mutant cell either with a translocation or a mutation is rather high relative to cell killing
 for the stem cell, under chronic exposures.
        I believe that you really should, especially under chronic exposure conditions,
 consider responses in the stem cell.
        I want to return to spermatogenesis for a second.  So, I  have done what I
 would  call the  abbreviated comparison between  a stem cell and a differentiating
 spermatogonia, but how about considerations of post-meiotic germ cell stages, the
 ones that were utilized in the risk assessment model that Vicki Dellarco just presented
 which, incidently, is a very good attempt to model  risk as a first shot.
        But here we have a completely separate consideration for  post-meiotic germ
 cells, and it  is a point that is often forgotten or not recognized, and that is that the
 next S phase for these cells is following fertilization. So, there is a fairly hefty period
 of time here for cells to accumulate damage for several days, and there is no repair
 taking place in these cells.  The repair takes place after fertilization. So, there is sort
 of a race between repair of damage induced in this cell stage and replication at the
 S phase.
       So, there is a completely separate set of probabilities of producing a mutation
 or a translocation for post-meiotic germ cell stages. You can't consider probabilities
 of effect similar in stem cells, spermatogonia, or post-meiotic cells.  You have to divide
 up, I think, the spermatogenic cell cycle into the different components and then relate
exposure or  effective dose to those particular cell stages based upon duration within
the spermatogenic cell cycle.
       For risk assessment, I think you have to consider a lot more than just the  post-
meiotic stage of the cell cycle,  and the probabilities will be quite  different; but also,

                                      167

-------
 there is a strong possibility that those probabilities will be different across species.
 We know very little about the human. We know something of the macro mechanism,
 the fact that DMA synthesis is required will cross species, but the probabilities that are
 found for the different stages or the different regions of the cell cycle could be quite
 different for the different species.
        So,  I  think it is very important to try and learn something  about those
 probabilities for the species that one is comparing.
 Translocation Mechanisms and the Stages Oogenesis

        I wanted to present a similar consideration for oogenesis to show that in the
 female, again, a quite different situation exists that would indicate, in fact, that for
 translocations and for mutations, there would be a rather low probability of induction.
 The higher  probability, one would anticipate, would be for aneuploidy events.
       The  reason for that is that here in Figure 2-72, I have identified again the
 periods of S phase; note that there is a very long period of time in here between this
 S phase (oogonia) and the next S phase which is after fertilization. So, there is a
 hefty period of time for repair to take place with a rather low probability of getting any
 adverse outcome in terms of a translocation or a mutation.  There is a very long
 period of time between the two S phases.
       Here (zygote) is the S phase that is important for mutation production by most
chemical agents other than those that are truly radiomimetic. Again, this contrast with
radiation will be quite obvious, because radiation does not require the S phase. The
event of interest is produced in the cell type of interest at the time of exposure. So,
                                     168

-------
                     OOGENESIS


                       Stem celt
                          i
                       Oogonia
     Mitoses
    (S phase)
   1st polar body
                       1° oocyte
                       (arrests)
                       matures
                          I
                       2° oocyte
                      Mature ovum
                      Fertilization
                        Zygote
     Meiosis I
(pre-meiotlc S-phase)
                                              Meiosis II
     Mitosis
     S-phase
A Diagram of the Various Stages of Oogenesis Showing the
Points at Which Cells Undergo an S Phase
                          Figure 2-72
                             169

-------
 I  think there is a very different set of considerations that one has to have for the
 different cell types based upon the mechanism of production.
        I had several other points that related to hemoglobin adducts and dose.  Dr.
 Ehrenberg raised these as they related to the importance of considering dose as
 measured by hemoglobin adducts based upon the lifetime of the erythrocyte.  I think
 this is a singularly important consideration.

 Implications for Risk Assessment
 The only other point I would like to make here that I think we should consider during
 this workshop, and Vicki Oellarco certainly alluded to it, and that is the question of
 what is genetic risk.  There are different interpretations of genetic risk.  I wanted to
 just present one on the assumption that tt is not the only interpretation of genetic risk.
       What we have to decide is what is the health impact of induced translocations.
 Here is just a simple reminder of what happens to chromosomes and translocations
 during meiosis.  On the left  of the figure is the normal situation.  There are two
 homologous chromosomes depicted, white and black.  These will pair up at meiosis
 and then segregate  in two different ways  (1,4:2.3 and 2,4:1,3).  There will be four
 different products and, of course, in this case, they will all be normal, normal haploid
 genotypes.
       In  the right  hand case (Figure 2-73) there  is a translocation  between
 chromosome  number 2  and  number  3, and  because of the  propensity for
 chromosomes to want to pair  up accurately, for homologous regions to pair, in
 meiosis, you get a configuration involving all four of the chromosomes; there are two
ways that these can segregate, giving four different products.
                                     170

-------
   normal
chromosomes
 pairing
at meiosis
           normal complement
 translocated
chromosomes
 pairing
at meiosis
                                        I
                                   2    4/13
                                     segregation


                                 IN
                                   •     o     •>•
                                    J
                                     J
                   translocation  1 mostly spontaneous
                      carrier      abortion
                   (semi-sterility)  2 sometimes severe
                                 congenital abnormalities
   Induction of a translocation in a stem cell spermatogonium with the
   pairing configuration in the primary spermatocyte and the most common
   segregation products normally and in the event of a translocation; A and
   B are translocation breakpoints.
                               Figure 2-73
                                  171

-------
        Without following this in any detail, all I need to point out is that the four
 possibilities are one normal, one translocation carrier, and two unbalanced.  The
 unbalanced products have duplications for the black chromosome and a deficiency
 for the white chromosome or, of course, the reciprocal of a duplication for the white
 chromosome and a deficiency for the black chromosome.
        So, these are unbalanced  genotypes, and in the vast majority of situations,
 these are lethal events.  So, what we have to consider here for genetic risk, and I
 consider genetic means transmission from generation to generation, we have to
 consider whether a translocation carrier represents a  genetic risk  and whether
 unbalanced genotypes represent a genetic risk.
        Does a translocation  carrier have, per se, a risk, such that it might be that it
 has some decrease in fitness? There are certainly some adverse outcomes being a
 translocation carrier, and certainly there is a component of the unbalanced genotypes
 which can be viable.
        Do we consider all of these  as adverse genetic outcomes?  I argue no.  I
 consider only the ones that don't  result in spontaneous abortions as being a
 component of genetic risk.
       If you take that proportion or that group of unbalanced genotypes and do a
 calculation, it comes out as shown in Figure 2-74.  I took this out of the UNSCEAR
 report (1986, p. 102), and it is not  very complicated and  I don't need to go through
 it.  tt is simply a way of representing  the proportion of conceptions that result in live
 birth multiplied by the unbalanced complement in those versus the total number of
those unbalanced complements in a population of the  live born  and the aborted
group, assuming that 85 % of conceptions result in live births.
                                    172

-------
       If you go through the calculation, you can calculate that unbalanced offspring

 arising from reciprocal translocations are about 9% of the total unbalanced genotypes.

 So, 9% of the unbalanced genotypes result in a viable but genetically unbalanced

 offspring.
                                Pi*,
               p1 is the proportion of conceptions that result in
               live birth (0.85).

               a1 is the frequency of unbalanced  complements
               in live-born arising from reciprocal translocation
               carriers (0.00012).

               p2 is the proportion of conceptions that result in
               abortions (0.15).

               ag is the frequency of unbalanced  complements
               in abortions arising from reciprocal translocation
               carriers (0.00681).

            Unbalanced offspring arising from RT = 9%
  Calculation of Total Unbalanced Genotypes that are Present
  in Live-Born Individuals
                             Figure 2-74
      So, we have to consider here, when we talk about genetic risk, do we wish to

consider 100 % of  translocation frequencies?  That means  that all reciprocal

translocations present a genetic risk. Or should we consider that it is a 9 % outcome
                                173

-------
 that is of significance in these cases?  I have argued that the latter is the correct
 approach.
        Also, I would like to point out that, again, there are two different components
 to translocation transmission.  You should realize that for stem cells, translocations
 produced in a pre-meiotic cell appear in the F, individual by transmission.  For the
 post-meiotic germ cells, that translocation is actually produced in the F, offspring, the
 DNA damage is transmitted.
        There are two very different considerations here. One, the translocation is
 transmitted from the parent to the offspring.  In the other case, it is produced in the
 offspring. I think these are important considerations which will really affect the genetic
 risk assessment. So, they are really additions to what you heard in the previous talks.
        We obviously need to consider what doses and what components there are
 to dose. The situation for chemical agents is not like radiation. We really don't know
 what the important ONA lesion is that leads to the replication errors that I am talking
 about here. We dont know what the real measure of dose is. We ought to consider
 that here.
       Also, we really need to  consider the  different responses of the cells in the
 different stages of the cell cycle, and we have to relate those to the DNA replicative
 phases and the probabilities of making errors on a damaged DNA template.
       So, I think there are several additional components to the EPA model that are
of importance, and there is one other that relates to the modeling that is  used.   I
would  argue that at  low exposures,  it would  really be  most applicable for
translocations to apply something which is very  much a dose squared component
curve.  This is in contrast to the various models fitted in the EPA risk assessment, one
                                     174

-------
 of which is shown in Figure 2-75, a strictly linear extrapolation from high exposure
 values.
       That does not imply the same mechanism as the rest of the curve at high
 exposures.  However,  at  low exposures, you need two  lesions to produce  a
 translocation, and you can well argue that those are produced as independent events,
 because you need replication errors on two chromosomes simultaneously.  You can
 assume those are independent events unless you wish to assume that one alteration
 on one DMA strand is sufficient to initiate a process of recombination which I would
 rule out at this point with the available data.
       So, I think at low exposures, (and as I say, this might well not be the case at
 the  high exposures which is the mode that we are trying to extrapolate from very
 often), just on mechanistic considerations, I think that a two-track fit, the radiation two-
 track type of curve, would be the appropriate one to consider.
       These are some of the points that bring in the macro mechanism into genetic
 risk  assessment and that is really a very simple look at mechanism.  Basically, It just
 involves DMA replication and its fidelity.  There are also micro mechanisms involved.
 Whether you have a GC to AT base pair change resulting in a mutation is what  I
 would call micro mechanism, which I don't think we are ready to or need to consider
 at this point.  However, I do think we need to consider macro mechanism in our
 deliberations of which type of parallelogram approach to use and how can you fill in
the  probabilities around that parallelogram.
                                    175

-------
       10 —
       II-
       M-
       u-
       10 —
       I —
                   too
                EtO Eipeture (ppm)
                             M<   «0   MO
The Induction of Translocation in Mouse Post-Miotic Germ
Cells by EO (from Generoso et al, 1990, Environ. Mol
Mutagen. 16,126-131)
                      Figure 2-75
                         176

-------
 Discussion

       DR. EHRENBERG: Just a very brief comment. If you have dependence of
 dose squared, the deliberative of that curve would be zero at very low doses, but if
 it is additive to a background of some kind, there will be a linear component.
       DR. PRESTON: A linear background or an interaction with the background?
       DR. EHRENBERG: Yes, if you have additivity of background.
       DR. PRESTON: I  am suggesting that this is probably a very low probability
 event, as  background, and the probability of interaction between  induced and
 background events is equally low or even lower.

       DR. BISHOP: Julian, just quickly. If a chemical is causing apurinic sites or is,
 say, interacting with some protein component causing stress and thereby causing
 breakage,  would that chemical have  a radiomimetic profile  or a non-radiomimetic
 profile or something intermediate?
       DR. PRESTON: An apurinic site would have a non-radiomimetic profile.  An
 apurinic site would behave as a potential site for a replication error.  It can result in
 the A rule, that is, what do you incorporate opposite a gap. You know, the tendency
 is to stick an A opposite a gap from replication of an apurinic site.
       DR. BISHOP: What did you say about the protein?
       DR. PRESTON: You could argue either way, and I think it would be whether
the torsion or  stress of a protein effect is  more likely when the DMA is unwinding
during  replication or when it is stationary or when it is transcribing.  Since I don't
know the answer, one could argue either way.
       DR.  BISHOP: Or it could be intermediate?
                                    177

-------
       DR PRESTON:  Well, if the stress is greatest when the DMA is unwinding
 during replication, it would be S phase dependent. If it is transcription which is driving
 the torsional stresses on DMA, it could be a G1 event.
       DR. NATARAJAN: In Drosophila, when you treat the flies with the alkylating
 agents and then you store the sperm, you get a storage effect.
       DR PRESTON:  Right.
       DR. NATARAJAN: This storage effect increases the frequency of chromosome
 aberrations with time.
       DR PRESTON:  Transiocations.
       DR. NATARAJAN: If you assume that model, you could expect without an S
 phase dependency, also S independent chromosome aberrations arising from the
 alkyiation at the activity sites following ethylene oxide treatment.
       DR. PRESTON:   I agree.   In fact, some years ago, I produced non-S
 dependent effects with chemicals by inhibiting DNA repair, which is essentially what
 you are doing in the Drosophila storage effect. Yes, I agree.  I think that both S-
 dependent and S-independent processes might be equally important in assessment.
       DR GENEROSO: And, Julian, in follow-up on that, the translocations we
 produced with ethylene  oxide or EMS which are very much the same, there is no
 evidence for heterozygosis with these translocations; to me, this means they are
 chromosome type aberrations which, means that these rearrangements must have
 been completed before the S phase after fertilization.
       DR. PRESTON: It might be very different in the case of the post-meiotic germ
cell stage.  I did point out it is a race between repair and replication.  If repair is rapid
                                    178

-------
in that stage, then it may be more a chance of producing a chromosome aberration
by a repair process at that stage.
      DR GENEROSO: That is what I think is possible.
      DR PRESTON:  Yes, but again, it is a mechanistic consideration, I think, that
we need to predict these probabilities.
                                  179

-------
180

-------
Data Summaries of Compounds
This section contains the following compound summaries:

     •    Ethylene Oxide
     •    Acrylamide
     •    1,3-Butadiene
     •    Cydophosphamide
                           181

-------
182

-------
 Ethylene Oxide

 Dr. AT. Natarajan presented the data summary for Ethylene Oxide.  Dr. Natarajan is
 from the Department of Radiation  Genetics  and Chemical  Mutagenesis,  State
 University of Leiden, The Netherlands.

       DR. NATARAJAN:  The parallelogram approach for estimation of genetic risk
 in man due to exposure to genotoxic agents utilizes the available data on tissue dose
 in somatic cells of experimental animals and man, induced chromosomal aberrations
 and point mutations in somatic cells of experimental animals and man, and induced
 chromosomal aberrations and mutations in germ cells of experimental animals.
       Thus, we have to  utilize the data from the animal systems, particularly the
 mouse, and all the available data in man except the germ cells in man to which we
 don't have any access at the moment.
       In the following, results will be discussed that have recently appeared in the
 literature and that have recently been generated in our own laboratory on genetic
 effects induced by ethylene oxide (EO) in rodents and in human populations. These
 data pertain to somatic cells because, according to Dr. Generoso (pers. comm. 1993),
 there are no recent new data available for germ cells.
       From the previous presentation by Dr. Ehrenberg, It has become apparent that
 DNA and hemoglobin adducts can be used to measure EO exposure in human sub-
jects and experimental mammals. Hemoglobin adducts can be used as a surrogate
for DNA adducts.  As was pointed out earlier, the problem with DNA adducts is -
especially in humans chronically exposed to low levels of EO - that the DNA adducts
measured may not represent the real tissue dose, because adducts can constantly
                                    183

-------
 disappear through chemical hydrolysis or they can be repaired by intracellular DMA-
 repair processes.  On the other hand,  hemoglobin adduct measurements  reflect
 exposure in the last three or four months  of the individual because the half-life of
 erythrocytes is about 120 days.
       EO adducts in DMA are uniformly distributed in the body except for the liver
 where there are higher levels of adduct formation, and in the testis where adduct
 levels tend to be lower (for example, due to the presence of a blood-testis barrier).
       As was shown earlier by Ehrenberg and Rhomberg et al., (1990), the frequency
 of EO adducts, especially the hemoglobin adducts per a given dose, appears to be
 very similar in mice, rats, rabbits, and humans. This makes a comparison between
 experimental animals and humans valid, at  least in the case of EO.
       When we consider chromosome aberrations in somatic cells, there is an old
 study by Ribeiro et al. (1987), which has been already referred to in an earlier review
 by Deliarco et al., in 1990. Ribeiro et al., found that inhalation of 200,400, or 600 ppm
 for six hours, and 200 or 400 ppm for 6 hr/day, 5 days/week for 2 weeks, resulted in
 dose-related increases in aberrations in bone-marrow cells  and spermatocytes of
 mice.
       Acute exposure was found to be more efficient in inducing aberrations than
 chronic exposure. More recently in our laboratory, Farooqi et al. (1993) carried  out an
in vivo experiment in which mice were treated i.p. with four different doses  in the
range of 0-4 mmol/kg (=  0-150 mg/kg of EO). For bone marrow cells, they found a
linear increase in the frequency of chromosomal aberrations per cell: aberrations/cell
= 0.08 (±0.06)  + 0.267 (± 0.03)  D (mmol/kg).
                                    184

-------
       Another endpoint which has been studied is the frequency of micronuclei in
 polychromatic erythrocytes bone marrow. In the early study by Appelgren et al.,
 (1978) mice were treated i.v. with two i.v. injections of 50-200 mg/kg, 24 hr apart. A
 dose-related response was obtained.  Rats were treated according to the same
 regimen and here also  there  was a dose-related increase of the frequency  of
 micronuclei. Positive, dose-related responses in mice were also reported by Conan
 et al., (1979) and Jenssen and Ramel (1980). In our own study with mice, (Farooqi et
 al.,  1993) the  frequency of micronuclei in polychromatic erythrocytes from bone
 marrow increased linearly with dose in the range of 30-150 mg/kg. There are no new
 data for induction of micronuclei in rats.
       If we look at the data on induction of sister chromatid exchanges, in the bone
 marrow cells of mice treated with 30-150 mg/kg (four doses), we obtain a very good
 linear dose response: SCEs/cell = 0.50 (± 0.07) + 0.94 (± 0.03) D (mmol/kg).  Also
 in rats, Kligerman et al., (1983) reported a concentration-dependent increase of SCEs
 in peripheral blood lymphocytes of rats exposed to EO by inhalation (0, 50,150, or
 450 ppm for 6 h/day for 1 or 3 days).
       After the publication of the review by Dellarco et al., 1990, several reports were
 published on  the  induction  of genetic damage in lymphocytes from workers
 occupationally  exposed to EO.  These papers have been reviewed for the present
 meeting by Tates (see separate document provided for the meeting; this document
 did not carry author's name; the name of the document was "Genetic Damage in
 Human Cells Exposed In Vivo to Ethylene Oxide").
       In some of these papers the induction  of chromosomal aberrations  was
studied (Hdgstedt et al., 1990; Mayer et al., 1991; Tates et al., 1991 and Lerda and

                                    185

-------
 Rizzi, 1992) (see Table 3-1). One of the problems with the studies by Hogstedt et al.,
 (1990) and Mayer et al., (1991) is that the fixation times for cultured cells were quite
 long (72 and 66 hours, respectively). Thus the lymphocytes would have gone through
 more than one cell division,  thereby  reducing the frequency  of aberrations in
 lymphocytes because of the dilution with second division cells. Thus, this could result
 in an underestimation of the aberration frequency and therefore to the absence of an
 effect in both studies. Another reason for the negative results is that the exposure
 levels were very low compared to the studies by Tates et al., (1991) and Lerda and
 Rizzi (1992).  In  our laboratory we looked at two EO-exposed groups (Tates et al.,
 1991). One with a low exposure, and another with a high exposure. We detected a
 dose-related increase in the frequency of aberrations in both groups of workers. In the
 study by Lerda and  Rizzi (1992), the frequency of aberrations was also significantly
 enhanced.
        Several groups of investigators studied induction of micronudei in lymphocytes
 from EO-exposed workers (see Table 3-2). Sarto et al., 1991 could not detect micro-
 nucleus induction in lymphocytes of sanitary workers exposed to low doses of EO.
 The same investigators studied micronucleus induction in nasal cells and buccal cells
 (Sarto et al., 1990,1991) but, again, there was no proof for induction of micronuclei
 in such  cells.  In  the studies of Mayer et al. (1991)  and Perrera et al., (1992), the
 exposure level was low,  as in the previous  study, and here again  no induction of
 micronuclei  could be established. A similar result was obtained in the  study by
 Schulte et al., (1992)  where exposure levels were also quite low. Thus it appears that
 at low levels of exposure there is no significant induction of micronuclei. In our own
study with hospital workers where the exposure level was low, we found a significant

                                     186

-------
      3-1-  Cytogenetle studies and Point Mutation Studies In Celts of
Chromosomal aberrations in peripheral blood lymphocytes
  Exposed to EthyUne Oxide
Population
Hospital
workers 8
Hospital
workers 2-6
Factory
workers 3-27
Hospital
workers 3
Factory
workers
tears Exposure information X Chromosomal Result
exposed (TUA values are In ppn) aberrations
Exp. Contr.
A: Contr. (23)c Smokers*
•: Exp. (34) TUA-8 h S 1 2.3 1.3 Ma
range: 0.008 - 2.4 Monsmokere
2.5 2.9 N
A: Contr. (8) . 48
•: Exp. (9); TUA-8 h « 20 4.6 2.0 Pb
with bursts of 22 - 72
C: Contr. (15)
D: Exp. (15); TUA-8 h - 17 - 33 6.9 1.9 P
with bursts of 14 - 400
A: Contr. (10) 48
•: Exp. (10); TUA-8 h - 60 - 69 8.7-10.9 1.2-7.9 P
A: Contr. (23) propylene-oxide 72
exposed workers
I: Exp. (6); sterilizers TUA-8 h • 2 5.8 4.7 N
C: Exp. (12); assemblers 4.8 4.7 M
Culture Reference
time (h)
ooMayer et el. (1991)
Tates et al. (1991)
Lenta and Rizzi (1992)
Hogstedt et al. (1990)
                    • * C
                                                               5.0
4.7
                                                                           187

-------
Table S-2.  Nicronuetw atudlee in peripheral Mood l>a*hocytee or other type, of coll*
Population
Hospital
worker* 5
(0.5 - 12)


Sanitary
workers 2
8.6
5.3




Hospital
workers 8


Hospital
workers




Year* Exposure Information
exposed (TUA values are In pp»)
A: Contr. (27)
B: Exp. (5); TUA-8 h - 0.025b.
C: Exp. (4); TUA-6.5 h * 0.38°
TUA-30 • « 4.4
B «• C
B:
C:
B + C:

B:
C:
B * C
A: Contr. (23)

B: Exp. (34); TUA-8 h • S 1
range: 0.008 - 2.4
A: US: Contr. (8)
B: US: 0-32 ppVh (32)
C: US: > 32 pp*/h(11)(TWA-8 h-0.16)
A: Hex: Contr. (1)
B: Hex: 0-32 ppw/h (9)
C: Hex: > 32 pp»/h(12)(TUA-8 h-0.54)



O.J7

0.60
0.63


13.34
13.00

15.60
NonsMok
12.62
480"
1060

3020
1810
NN /
Exp.
Masai cat la

0.44
Buccal eel la
i!oo
1.00
IvnohocYte*
12.65
11.00
11.00
S*oker»d
16.50
,eff
13.90
5108


2150

1000 Result Method Reference
Contr.
Old
(no Cyt-B Sarto et al. (1990)
or BrdU)

«c
OldSarto et al. (1991 )"
1.00 H (no Cyt-B
N or BrdU)
N
Cyt-B
11.00 H
N
N
BrdUNayer et al. (1991)
N Perera et al. (1992)

N
Old Schulte et al. (1992)
N (no Cyt-B
N or BrdU)

N
N
                                                                           188

-------
Table 3-2.  Nlcronucleus stadias In peripheral Mood lymphocytee or other types of cell*  (continued).
Hospital            A: Contr. (8)
workers 2-6         B: Exp. (9); TWK-8 h « 20"
                       bursts of 22 - 72
Factory             C: Contr. (15)
workers 3-27        0: Exp. (15); TtM-8 h « 17 - 33
                       bursts of U - 400
29
27
Cyt-B
                                              Tstes ct si.  (1991)
Factory A: Contr.: propylena-oxlds
workers
B
C
B
exposed workers (23)
: Exp.: stsrl Users (6); TtM-8 h » 2
: Exp.: assemblers (12)
* C:

6
5
5

.0
.4
.7

2
2
2

.6
.6
.6

P
P
P
Ndgstedt et al. (1990)
Old
(no Cyt-B)
(no BrdU)


So effect
"Statistically significant difference between control and exposed subjects
Sunber between brackets indicates muter of subjects per group
"current smokers; data on subdivision into current and former  smokers are not included in the table
'cells with high frequency of SCEs
 Cumulative exposure in ppVh during 4 months prior to blood sampling
"Frequencies  indicated in Table 3 of Schulte et al. (1992) seem  to be unreal 1st leal ty high (error?)
 Lower values were found when internal exposure was measured via Nb-adduct formation.
                                                                             189

-------
 induction of chromosomal aberrations but no evidence for induction of micronuclei
 (Tates et al., 1991). Thus for low exposures to EO, the micronucleus frequency does
 not seem  to be a sensitive Indicator for EO exposure. In more heavily exposed
 workers from a factory involved in the production of sterilized disposables, we found
 a statistically significant induction of both chromosomal aberrations and micronuclei
 (Tates et al., 1991). Finally, Hogstedt et al., (1990)  reported a significant increase in
 the frequency of micronuclei in factory workers exposed to rather low doses of EO.
 Thus, when the exposure is high,  it is possible to detect micronucleus induction.
 Chromosomal aberrations seem to be a more sensitive indicator of induced damage
 than  micronuclei under chronic low exposure conditions.
       Several recent reports describe results of investigations into induction of sister
 chromatid  exchanges in  lymphocytes from  EO exposed workers (see Table 3-3).
 Mayer et  al., (1991) looked  at  smoking  as a  confounding factor and  could
 demonstrate induction of SCEs in smokers but not in non-smokers. In our own study
 (Tates et al., 1991), we found that both low and high exposures induced statistically
 significant amounts of SCEs. Lerda and Rizzi (1992) also found a positive response.
 Finally, positive effects were also reported by Sarto et al., (1991) and Schulte et al.,
 (1991).  The latter two groups of investigators did not  find induction of SCE's in
workers that received relatively low doses of EO.
       From the foregoing, it appears that SCEs represent a very sensitive endpoint
following chemical exposure such as EO and, as expected, it comes out to be positive
in most of these  studies.
                                    190

-------
Table 3-3.   Sister chroMtid
(MX) studies In peripheral Mood lymphocytes
Population
Reference

Sanitary
workers 2
8.6

5.3
Hospital
workers
8

Hospital
workers 2-6

Factory
workers 3-27

Hospital
workers




Hospital
workers 3
Years Exposure information
exposed (TiM values are in pan)
A: Contr. (10)
•: Exp. (5); TlM-8 h > 0.025
C: Exp. (4); TtM-6.5 h - 0.38
TIM-30 m « 4.4
B * C:
Contr. (23)
Exp. (34); TMA-8 h « 5
range 0.008 - 24

A: Contr. (8)
•: Exp. (9); TUA-8 h * 20 - 25
with bursts of 22 - 72
C: Contr. (15)
0: Exp. (15); TUA-8 h « 17 - 33
with bursts of 14 - 400
A: US Contr. (8)
I: US 0-32 ppm/h (32)
C: US > 32 pp*/h<11) 32 ppn/h(12)(TUA-8 h-0.54)
A: Contr. (10)
B: Exp. (10); TWA-8 h « 60 - 69



10.0
12.4

11.2

13.0

10.2

6.0


12.2


5.9
6.9

6.5
6.5

13.3



10.8
10.8

10.8
Smokj
VT6
ponsmifri
9.8

5.0


6.3


4.6
4.6

5.6
5.6

6.1
SCE/cell
Exp. Contr.

H
P

N
t£i
P
ma.
N

P


P


P
P

H
N

P
Result HFCs* Result
Exp. Contr.
Sarto at al. (1991)




Smokers Mayer at al (1991)
6.9 4.0 N
Monsisokers
1.8 1.2 N
Tates et al. (1991)
12.8 4.8 P


79.4 4.2 P

Schulteet al. (1992)





Lerda and Rlizl


-------
        When one carefully analyzes data on sister chromatid exchanges in exposed
 workers, one can often recognize a subpopulation of lymphocytes with a relatively
 high frequency of SCEs (Mayer et al., 1991 and Tates et al., 1991) (see Table 3-3).
 Therefore, the number of high frequency cells, along with haemoglobin adducts, can
 be used as very sensitive indicators of exposure.
        In addition to cytogenetic studies on effects of EO exposure, there are several
 publications on induction of point mutations by EO. Very recently, Walker and Skopek
 (1993) reported a dose-dependent induction of HPRT mutations in splenocytes of
 preweanling mice (12 days old) exposed to fractionated doses of EO (cumulative
 doses of 200, 600, or 900 mg/kg). In this study an expression time of 8 weeks was
 used  and  the mutant frequency was increased  respectively  5-, 15-,  and 24-fold
 (control frequency: 3x10"6). The dose-effect curve was found to be linear. In another
 experiment with doses ranging from 150-600 mg/kg, an expression time of 20 weeks
 was used. In the latter case, the increase in mutant frequency was much lower (range;
 1,4-5 fold). Thus, the proper selection of the expression time for mutations is critically
 important for demonstrating a positive effect of exposure. The optimal expression time
 seems to be about  10 weeks. It is of interest to note that evidence for mutation
 induction in EO-exposed mice could only be obtained in preweanling mice and not
 in adult mice.
       In Leiden, we have extensively studied  induction of HPRT mutations in lym-
 phocytes from spleen and cervical lymph nodes from adult rats (Dam van, et al.,
 1992). Rats were given 1 x 80, 3 x 80, 4 x 60, or 4 x 26 mg/kg EO with one day
intervals between fractions. Splenocytes had higher spontaneous mutant frequencies
than lymphocytes from cervical lymph nodes. Despite the fact that several expression

                                    192

-------
 times were used, there was generally little or no evidence for mutation induction by
 EO in these  experiments. However, statistical analysis of our data by L Ehrenberg
 suggests that a 50 percent increase of the mutant frequency over the control value
 cannot be ruled out. These unpublished data are in agreement with those of Walker
 and Skopek (pers. comm.) who also could not detect mutation induction by  EO in
 adult mice. The difference in mutagenic effects of EO in preweaniing and adult mice
 may be explained in two ways that mutually support each other. Firstly, there  are
 more dividing lymphocytes in preweaniing mice than in adults so that the conditions
 for fixation of pre-mutational lesions during S-phase, and consequently induction of
 mutations, are better in preweaniing mice than in adults. Secondly, the detoxification
 capacity of preweaniing rodents is lower than in adults, so that more pre-mutational
 damage persists for mutation fixation in preweaniing mice.
       The above results with mice and rats do not exclude the possibility that  EO
 may still induce mutations in adult mice or rats. If, for example, EO is  administered
 chronically over longer periods of time one increases the probability that lymphocytes
 will  be exposed to EO  when undergoing DNA-synthesis so that pre-mutational
 damage can be converted more effectively into mutational damage. To test this possi-
 bility we are  presently performing mutation studies with rats that are  chronically
 exposed to EO in their drinking water. Preliminary results indicate mutation induction
 by EO in adult rats.
       It is known that hydroxy ethyl nttrosourea  (HOENU) induces similar adducts
as ethylene oxide (Segerback,  1990). The ratio of O6/N7 guanine adducts for the two
chemicals is very different (for EO O6 constitutes 0.5% of the alkylation of guanine-N-
7, whereas for HOENU O6-(2-hydroxyethyl) guanine was found to  be 63% of N-7-(2-

                                     193

-------
 hydroxyethyl) guanine). At present, we are conducting experiments with HOENU in
 adult rats, and found already that 75 mg/kg leads to a 30-fold increase in mutant
 frequency compared to the control value. So, in principle, one would expect that with
 EO we should get also induction of mutation at the HPRT. As mentioned above, we
 indeed found induction of  mutants in rats treated for one month with EO in the
 drinking water.
        We would now like to go back to our study where workers were occupationally
 exposed to EO (Tates et al., 1991). This study involved a multilaboratory investigation
 on  workers from the former German Democratic  Republic. Blood samples were
 analyzed for frequencies of chromosome aberrations, micronuclei and sister chromatid
 exchanges (Leiden, Bad Elster), HPRT mutations (Leiden) and hemoglobin adducts
 (Stockholm and London). In addition, we sequenced in Leiden a collection of mutants.
 Among the different endpoints studied, their sensitivities to detect an exposure was
 in decreasing order: Hb adducts, high frequency SCEs, chromosomal aberrations,
 SCEs,  micronuclei, and HPRT mutations. The  relative insensitivity  of the HPRT
 mutation assay to detect exposure is expected, because in the case of all the other
 endpoints, we are looking at the whole genome, whereas with HPRT, we are  looking
 at a DMA target of about 43 KB only.
       The spectrum of mutations induced by EO was not found to  be different from
 that in smoking and non-smoking controls. When we look, however, at the distribution
 of mutants over the HPRT coding region, a hot spot for EC-induced mutants was
found.
       Coming back to the parallelogram method for risk estimation, K is of interest
to note that mutants induced by EO and ENU in somatic cells in vitro predominantly

                                    194

-------
 involve changes in guanine, but in the case of in vwo-exposed germ cells and somatic
 cells, the mutational changes frequently involve AT base pairs. When comparisons are
 made of the spectrum of mutations in germ cells and somatic cells of mice, rats and
 monkeys exposed in vivo to ENU, one finds a striking similarity. These findings are
 very reassuring and indicate that we can extrapolate with confidence from in vivo data
 with mice, rats and monkeys to the human situation.
 Discussion
       DR. SORSA:   Thank you, Nat,  for this very comprehensive background
 information concerning genetic effects of EO. I would now like to invite a discussion
 on the various points raised in the presentation by Natarajan.
       DR. VOGEL Nat, in the last slide which you showed on the ENU spectrum,
 you had data on the monkey and rat?
       DR NATARAJAN: Mouse, monkey, and rat.
       DR VOGEL Did your data pertain to somatic cells or also to germ cells?
       DR NATARAJAN: Somatic cells.
       DR VOGEL  I am asking you, because you also showed Drosophila data
 which are quite different. The germ cell mutation data for the specific locus test of the
 mouse are for pre-meiotic germ cell stages whereas the germ line data for Drosophila
 pertain to post-meiotic germ cell stages. In Drosophila the mutations induced are pre-
 dominantly GC -> AT transitions whereas in the mouse AT -> GC changes prevail.
 However, Dr. Maria Serra (pers. comm.)  recently collected mutation data (vermillion
 gene) for pre-meiotic germ cell stages of Drosophila treated with ENU. In that case
the mutation spectrum corresponded very well with the mouse spectrum for mutations

                                   195

-------
 in pre-meiotic germ cell stages, also showing a preponderance of changes at AT base
 pairs.  In other words, in Drosophila the mutation spectra shifted from GC -> AT
 transitions to changes involving the AT base-pairs, when comparing postmeiotic and
 premeiotic germ cells. Thus, when you compare the same germ-cell stages in different
 species, you apparently find similar types of mutational changes.
        DR PRESTON: I just wanted you to clarify. When you were showing the early
 data on chromosome aberrations, you had linear dose response curves. Were those
 just deletions mostly?
        DR NATARAJAN:  Yes.
        DR PRESTON: Okay.   You mentioned that the curves for chromosomal
 exchanges and deletions were non-linear. What was the power of the dose in these
 curves?
        DR NATARAJAN:  The power of the dose {= n) in these curves could be 2,
 3 or perhaps 4.
        DR PRESTON: You said "n" is greater than 1.  I wondered what that meant
 or what the number exactly was. Just for clarification.
       DR GARNER: Nat, in connection with the ethylene oxide exposed individuals,
 you mentioned that you could detect hemoglobin adducts, but that DMA adducts
 weren't detectable. I just wondered what the limits of sensitivity of the two assays
 were, because I would expect, on the whole, DMA adducts to be a more sensitive
 measure than protein adducts.
       DR TATES: Very restricted efforts were made to measure DMA adducts. Rob
 Baan's group, in Rijswijk the Netherlands, made preliminary attempts to measure ring
open N7-ethylguanine adducts by means of a specific antibody developed in their

                                   196

-------
laboratory. Unfortunately the technique was not yet standardized at the time these
measurements were performed. There was insufficient biological material available to
repeat these measurements at a later date.
                                    197

-------
198

-------
 Acrylamide (AA)

 Dr. Kerry Dearfield presented the data summary for Acrylamide. Dr. Dearfield is from
 the Health Effects Division, Office of Pesticide Programs, EPA, in Washington, D.C.,
 U.S.A.
       DR. DEARFIELD:  Before I start, I would like to relate a short story on how I
 met Vicki Dellarco, who you heard earlier.  When I was in the former Office of Toxic
 Substances at the EPA,  acrylamide was a major compound that  I was asked to
 examine in terms of regulatory action. The Office was looking at grout workers who
 work with acrylamide as a sealant for sewers that have cracks in them. That was and
 still is a major action that the EPA is examining for acrylamide.
       We examined all the  data at that time and wrote up our data assessment
 which culminated in the review article that is in your package. At the time we were
 looking at the acrylamide data, we had some dominant lethal data from Waldy
 Generoso. We were trying to decide if we could do a genetic risk assessment for
 acrylamide with just dominant lethal data. This was about 1986,1987.  We had just
 gone through the material on the dominant lethal, and we were thinking maybe we
 could give a genetic risk assessment a try.
      Around this time, we heard that the EPA's Office of Research and Development
 (ORD) had been  discussing ideas for genetic risk assessment. So, on the advice of
 my senior science advisor, I went over to the ORD, knocked on Vicki Dellarco's door,
 and asked what have you done on genetic risk assessment.
      She said they tried something with ethylene oxide, but the effort essentially
died.  I said it is time to resuscitate it, because genetic risk assessment is the coming
                                    199

-------
 thing.  We were beginning to receive a lot of data that I talked about this morning.
 The regulatory schemes are pointing toward obtaining this type of data from the
 quantitative tests, so we  needed to know how to do a genetic risk  assessment,
 especially since the EPA is now requiring companies to perform these tests. In other
 words, if we couldn't do an assessment with the data, then why are  we requiring
 companies to spend hundreds of thousands  of dollars to do these tests?
        So, we went ahead and resuscitated the ethylene oxide assessment effort that
 culminated in the report that is in your  background materials.   I like  to think that
 acrylamide was responsible for  the birth of the ethylene oxide risk assessment!
        That was about 1988.  Since we put together that particular review with
 acrylamide, there has been an incredible amount of data that has been generated.
 We now have heritable translocation data, and we have specific locus data.
        We have looked at many different questions that were  highlighted in that
 particular review.  At that time,  we had thought that maybe acrylamide might be a
 special compound, that is, it might have been a specific germ cell mutagen versus a
 somatic cell mutagen. We have since found that not to be true, but there was some
 discussion about that at that time.
       There was also a discussion of whether it was the parent compound versus
 a possible metabolite that might need consideration in a genetic risk for acrylamide.
That was very unclear at the time, and studies showed that acrylamide appeared to
have a direct acting capability. We were also trying to figure out what is the primary
genetic activity of acrylamide.  We had some discussion about that, and I would like
to talk about it some today.
                                    200

-------
        So, there has been a tremendous amount of research on acrylamide.  Many
 people looked at pharmacokinetics and other aspects.  We have also done much
 exposure analysis  for acrylamide.  One of the risk assessments for acrylamide
 culminated in the contract that Dave Brusick talked about earlier with ICPEMC and
 Health Canada.
        We are going to work with that assessment in our break-out group later. So,
 let's meet the players (Figure 3-1 A).
        Acrylamide is a fairly simple compound.  It has an alpha, beta unsaturated
 bond and this imparts much of its activity. The main use for acrylamide is actually to
 make polyacrylamide, and there are many uses for this. Most obvious to us are the
 polyacrylamide gels you see many researchers use in laboratories, but there is a lot
 of industrial use for acrylamide. Probably one of the largest uses for acrylamide is as
 a flocculent in drinking water.
       So, we have a lot of exposure to polyacrylamide and some form of unreactive
 monomer in your drinking water.  This is the main reason why the EPA's Office of
 Drinking Water is very concerned with acrylamide, because millions of people drink
 water that has been treated with polyacrylamide.
       The main reason why I was interested in acrylamide was through the efforts
 of the EPA's Office of Toxic Substances, the name has recently been changed to the
 Office of Pollution Prevention and Toxics (OPPT). Acrylamide is used heavily by grout
workers. What they do is mix it like cement and use it as such.  The workers use big
bags of monomer, mix it like cement, and then go  down into sewers for repairs. They
trowel it into cracks, and when it polymerizes, it results in a seal to fix the crack in the
sewer line.
                                    201

-------
A.   ACRYLAMIDE
                O
     H,C=CH— C— NH
B.   GLYCIDAMIDE
          Figure 3-1
           202
2
O
/ \
H2C — CH -
O
II
-c-


-NH2

-------
       These workers are covered with acrylamide, so there is a large amount of
 exposure. This is the work practice that the OPPT is examining for regulatory action.
 I have some exposure numbers from this.
       Now, the  other player I want to discuss is glycidamide (Figure 3-1B). When
 you first  look at the  parent compound, acrylamide, you automatically assume an
 epoxide is going to form.  That is one of its apparent natural metabolites.  However,
 the data at the time about 1988 did not suggest much of a role for glycidamide.  It
 was  questionable whether that glycidamide  was a natural in vivo  metabolite of
 acrylamide.  We now know that, most likely, it is.
       There is a second reaction for acrylamide. It is called a Michael-type reaction
 in which  the beta-carbon  reacts with a nucleophile.  This is a kind of alkylating
 reaction.  This is a direct-acting type of mechanism, and  is generally the type of
 activity we have seen in many of our studies with EPA; for example, it was seen in the
 study with Martha Moore and Jim Rabinowitz.  This is the type  of reaction that this
 class of compounds, these Michael-type addition compounds, generally undergo.
       These reactions are important because we will discuss in a few minutes what
 is the primary target of acrylamide and glycidamide; so it is  important  to understand
 the chemistry here.
       First of all, I would like to give you an idea of some of the data.  I put together
 an extensive  table for this workshop effort. I think it is about six or seven pages of
just the studies that have been done since  1988. This table does not  include any of
the studies that were in the earlier review I  mentioned.
       It is just amazing how much has been done. For example, at the time of our
earlier review, we were hustling to get Mike Shelby et al.'s paper on the  heritable

                                     203

-------
 translocations as a footnote!  But since then,  look at all these studies just on
 dominant lethal assay, for example, that have been done (Figure 3-2).
        An important point is that acrylamide has been tested for dominant lethal
 effects via different routes of administration. You have i.p.  injection,  gavage, via
 drinking water, and dermal exposure.
        in the drinking water test, we have the lowest dose  found yet in which a
 positive response for a dominant lethal effect was  observed, 9.2 mg/kg/day. Most of
 the studies we have seen are generally up around 40 to about 100 mg/kg/day.  And
 I know even more data is being generated as we  meet this week.
        The point is that acrylamide can be incorporated via many different routes of
 administration and cause an effect.
        Also, once acrylamide gets into the system, it is distributed almost  everywhere.
 We also know that it is cleared very quickly as well. A major reaction is  the adducts
 it forms from Michael addition to a lot of protein in the blood, particularly, hemoglobin.
 We will talk about that in  a minute.
       The only two places it appears to accumulate after the first few hours is in the
 skin and in the testis. The testis accumulation was another piece of evidence from
 earlier that suggested acrylamide may be a unique germ cell mutagen.  It  may yet turn
 out to be more preferential to the germ  cells than to somatic cells.
       Another thing I want to point out, not from this slide, but when we were first
thinking about whether it  was a unique  germ cell mutagen or not, Ilse-Dore Adler et
al. came out with their  micronuclei  data to show  that in the bone marrow, it was a
somatic cell  mutagen.   The  data however showed, generally speaking, those
responses in the bone marrow were not that dramatic. There is a dose-response, but

                                     204

-------
Assay
Dominant lethal
assay
Heritable
translocations
Reciprocal
translocations

Test System
Male (C3H X 101)
F, mice
Male Fischer 344
rat
Male Pzh:SFlSS
mice
Male (102/E1 X
C3H/EDF, mice
Male CD-I mice
Male (C3H/R1 X
101/RDF, mice
Male (C3H X '
I01)F, mice
Male C3H/E1
mice

Concentration/
Dose1
5 X 40 mg/kg/day;
i.p. injection
5 X 30 mg/kg/day;
gavage
75-125 rag/kg;
i.p. injection
50-125 mg/kg:
i.p. injection
0.72-9.2 mg/kg/day;
drinking water for
20 weeks
5 X 25-125 mg/kg/
day; dermal
5 X 40-50 mg/kg/
day: i.p. injection
5 X SO mg/kg/day:
i.p. injection

HID or
LED2
40 mg/kg/
day
30 mg/kg/
day
125 mg/
kg
75 mg/kg
9.2 mg/
kg/day
25 mg/kg/
day
40 mg/kg/
day
50 mg/kg/
day

Assay Result
Positive
Positive
Positive
Positive
Positive
Positive
Positive
Positive

Reference
Shelby et al.,
1987
Working et al.,
1987
Dobrzyiiska et
al.. 1990
Ehiing &
Ncuhiuser-
Klaus. 1992
Fail et al.. 1992
Cutierrez-
Espelcta et al.,
1992
Shelby et al.,
1987
Adler. 1990

Figure 3-2
   205

-------
 the response was not real large compared to some of the responses you see with the
 germ cell observations (for example, dominant lethal effects).
        The other thing I wanted to add is that many of the other studies in germ cells
 for aberrations generally turn out to be negative.  But there are not very many studies
 in the germ cells for aberrations. There are also some for micronuclei.  These are just
 some things in the general data base I wanted to point out for our assessment later.
        One thing that was of interest to me is how can we sort through the pieces of
 information and get to an understanding of how acrylamide and glycidamide may be
 interacting?
        In a lot of discussions I had with Gary Sega in the workgroup and others, we
 came up with this working hypothesis, if you will (Figure 3-3). I just want to present
 this for the workgroup to consider. We will talk about this in the break-out sessions
 and whether this  has impact on our genetic risk assessment or not.
        First we have  acrylamide and we now know it does convert somewhat to
 glycidamide. The first thing we notice in working with acrylamide is that it binds quite
 avidly to protein.  It goes  after thiol-containing compounds very well.  This is a
 component of the pharmacokinetics one has to take into account.  Once acrylamide
 gets into the system, it is binds extensively to protein.
       Another consideration is that, early in  our first assessment, there were some
in vitro  studies looking at DMA adduct formation with  acrylamide.  However, the
problem with this observation was it took acrylamide something like 40 days to obtain
detectable binding, only then was found weak activity.
                                    206

-------
           converted
AA
weak binding  '   1 binding to
to DNA       /    I proteins
AA-DNA adducts


       AA-protein adducts


  o AA is SAL negative

  o long time to detect DNA
     binding after exposure

  o delay  in maximum UDS after
    'exposure, suggests conversion

  o similar sensitive germ cell cycle
     stages as seen for DLM & HTT
     (suggests protamine binding)

  o synaptonemal complex effects
     (proteinaceous structure)

  o spindle poison?

  o P450  inducer (affects proteins?)
                                            GLY
                  binding to /   x   some binding
                       DNA/    N to protein
                                   GLY-protein
                                   adducts

                     GLY-DNA adducts


                    o GLY is SAL positive

                    o no delay in UDS after exposure

                    o epoxide less discriminant .'. may
                      contribute to gene mutations in
                      spermatogonia

                    o predominant adduct  in DNA in
                      vivo rodents was GLY-DNA
                      adduct; major adduct, as
                      expected for epoxide, is at
                      N-7 of guanine

                    o DNA adduct level in  vivo is
                      similar in several tissues
                      .'. GLY evenly distributed
          Figure 3-3
             207

-------
        So we wondered, if acrylamide is an avid alkylator of DMA, why would it take
 so long in a pure DNA-acrylamide environment to cause adducts? This posed several
 questions, for example, is acrylamide itself really the DMA adduct moiety?
        When you look down these lists on this slide, you begin to sort out activities
 between acrylamide and glycidamide. Another thing that perplexed us for a long time
 was that acrylamide was Salmonella negative with and without activation. What we
 didnt know for awhile is that glycidamide is Salmonella positive.
        In the study that Gary Sega performed with unscheduled DMA synthesis (UDS),
 there was a delay in UDS response after exposure.  He suggested there was some
 conversion to some metabolite before eliciting a response.  Gary started to examine
 glycidamide. The work is unpublished, I think as of now.  But it was found that there
 is no delay in UDS after exposure to glycidamide. This suggests that it might be quite
 appropriate to consider a glycidamide-DNA adduct.
       The next point addresses the similar sensitive germ cell cycle status for the
 dominant lethal  and heritable translocation studies in post-spermatogonial cells.  I
 think in the slide that you saw from Vicki Deliarco earlier this morning, acrylamide had,
 apparently, from the dominant lethal studies, the same window of susceptibility in the
 mouse studies and  in the rat studies.
       Gary, at that time, suggested that the most sensitive time for dominant lethal
 effects was the same time as protamine appearance in germ cells.  Unlike ethylene
 oxide where you can get some detectable DMA adducts, again, it was very difficult to
find acrylamide-DNA adducts in the germ cells.  Whereas we made the assumptions
that the DMA adducts were probably the more important adduct for ethylene oxide,
because of the known affinity for acrylamide to proteins, this aspect of protamine may

                                    208

-------
 weigh a little bit more in the acrylamide-induced dominant lethal and the heritable
 translocation effects, according to the Sega hypothesis I am pretty sure you are all
 familiar with.
        Other pieces of evidence about the protein affinity of acrylamide came from
 some other work. We did some work with Jim Allen on the synaptonemal complex.
 We were looking for aberrations and other types of events in the synaptonemal
 complex.  We didn't find any aberrations,  but did find some interactions with the
 complex itself in the form of irregularities. Since this is a very proteinaceous structure,
 again, this suggests a binding mechanism to a protein. In the material that Ilse-Dore
 Adler was reporting  in some recent publications, there is some disturbances of the
 spindle apparatus, possibly even  leading to aneuploidy.  Again, the spindle is a
 proteinaceous structure.
       The last point here  is  from an article that suggests that acrylamide might
 induce P450.  It might have some type of interaction with the protein, but I am not
 dear on the mechanism.
       All this evidence suggests that this big arrow (Figure 3-3) represents that the
 major property for acrylamide  is more likely for acrylamide to form protein adducts
 versus DMA adducts. This does not rule out the possibility that DMA adducts may
 form from an acrylamide  parent.
       Now examining the glycidamide evidence (Figure 3-3), epoxides, by inference,
 are generally positive in Salmonella tests. This gives you an idea that there is some
 DNA reactivity here.  Again, we talked about the timing of the UDS response earlier.
We know that epoxides are less discriminate in what they tend to interact with, and
                                     209

-------
 we also know in general they cause gene mutations. This may contribute to the gene
 mutation effect in spermatogonia.
        What I am going to suggest as  a hypothesis is that the glycidamide-DNA
 adducts may be more responsible for the gene mutation aspect of this compound,
 whereas the acrylamide protein adducts such as the protamine binding may explain
 some of the clastogenic outcomes that we see in the germ cell tests.
        In an unpublished study performed by Dan Segerback et al., glycidamide DMA
 adducts were found in vivo. The predominant adduct in DMA was the glycidamide
 adduct, and the alkylation was  found  in N-7  of  guanine, which is expected for
 epoxide.  So, this is direct evidence that glycidamide adducts are forming in vivo.
        He also found  a similar DMA  adduct  in vivo in several  tissues.   So,
 glycidamide, although you would think an epoxide would  be rather reactive, does
 seem to distribute rather evenly. Acrylamide itself also distributes rather evenly. That
 is another consideration for when we do  a risk assessment for this compound. So,
 these are different pieces of information that give you an idea of the clastogenicity
 versus the gene mutation activities of acrylamide.
       I just might mention that there  is some suggestion that this may be  a
 mechanism for neurotoxicity. Acrylamide binding to protein  can occur for example to
 proteins in myelin sheaths.
       By the way, other toxicities besides neurotoxicity helped drive the regulatory
concern for this compound. Since the beginning of our efforts, the EPA has had two
cancer studies submitted. Those  two cancer studies probably reproduce each other
almost  as  well  as  any two  cancer  studies  we have  seen;  they  were done
                                    210

-------
 independently at different times.  Now we are getting all the genetic data as well as
 reproductive and developmental data.
        When assessing these various effects, we started to see effects at doses that
 were  below  doses you  would  see for overt neurotoxicrty.   Neurotoxicity  from
 acrylamide is a result of cumulative exposure.  You have to be exposed for several
 days, but some of these other effects are seen at doses lower than you would see in
 a cumulative approach and in lesser time frames.   So, these other toxicities are
 becoming as important as neurotoxicrty for regulatory concerns.
        DR. SHELBY: Can you say anything about the relative affinity for acrylamide,
 glycidamide, and ethylene oxide? We just got a lot of ethylene oxide protein binding
 data but none on acrylamide. I must ask is there any difference.  Is there reason to
 think there is a difference between ethylene oxide and glycidamide or acrylamide with
 regard to their protein binding characteristics?
        OR DEARFIELD:  No, I have not looked at that yet or thought about ft.  This
 hypothesis I'm presenting has only crystallized in the last month, so I haven't really
 done any comparisons to ethylene oxide.
        Since this is a meeting on the parallelogram approach, I wanted to show some
 parallelogram  ideas.  I  apologize  here, I put  arrows on  each  end  of every
 parallelogram.  I know this is not appropriate, but since I had to hand draw this, I just
thought I had better  put as many  arrows on as I could!
       I think this particular parallelogram on Hb adducts is very useful (Figure 3-4)
because ft also helped us  in  discerning whether glycidamide was actually formed in
the body.   I think this is one of the uses of a parallelogram, to give you qualitative
information.
                                     211

-------
Hb adducts in vitro
     animal
            Hb adducts in vivo
                 animal
Hb adducts in vitro
     human
            Hb adducts in vivo
                 human
Figure 3-4
  212

-------
        This Hb work was done by Bergmark et al. and I've discussed this with Carl
 Calleman. They were looking at hemoglobin adducts in in vitro systems to calibrate
 essentially  how  they would  detect  these particular hemoglobin  adducts from
 glycidamide.  They first calibrated their machines and for the levels they were trying
 to detect.  So, they were using an in vitro situation as their standard.
        They then looked in mouse and they found the same Hb adducts in the
 animal. So, they were using this experience to see if you can detect hemoglobin
 adducts in in vivo situations.
        These researchers obtained access to a cohort of workers in China who were
 in the manufacturing process of making acrylamide. They went in and tested their
 erythrocytes for hemoglobin adducts.  They found glycidamide-hemoglobin adducts
 in vivo.  That is evidence that there is metabolism from acrylamide to glycidamide
 which is detected in vivo.
        In 1988, we didnt have any of these kinds of information, so we really weren't
 sure about the role of glycidamide.  Now I think we are pretty sure that glycidamide
 will be a major player in the risk assessment for acrylamide.
       I want to show a couple more parallelograms to reinforce what I was saying
 about the gene mutation versus the clastogenidty activities (Figure 3-5).  For gene
 mutation, we have the somatic cell in vivo  animal tests from the  big table  we put
together, for example, the mouse spot test  and the transgenic mouse somatic cell
data.  From the germ cell in vivo animal data, we have the specific locus test. We can
try to extrapolate to human germ cell in vivo situations using the techniques that were
talked about this morning.
                                    213

-------
               Gene Mutations
Somatic cell in vivo       Germ cell in vivo
    animal          *—*      animal
(e.g. spot test, transgenic)     (e.g. specific locus)
                               I
Somatic cell in vivo       Germ cell in vivo
    human         *—»      human
                Clastogenicity
Somatic cell in vivo       Germ cell in vivo
    animal          «—»      animal
  (e.g. MVM, abs)            (e.g. DIM, HTT)
       1
Somatic cell in vivo       Germ cell in vivo
    human          «—»      human
                  Figure 3-5
                     214

-------
        For clastogenicity (Figure 3-5), the somatic cell data we have is micronucleus
 and aberration tests.  For me germ cell data, we have the dominant lethal and
 heritable translocation tests. Again, we will try to extrapolate to the human germ cell
 situation.
        This presents a question. We are probably going to derive two different types
 of genetic risk assessments, one for gene mutation based on one set of studies, and
 one for clastogenic events based on another set of studies, each with their own
 assumptions.  Is it possible to combine these into one assessment,  or  is it too
 difficult? These are things we will need to discuss.  What is the total  overall risk? I
 don't know if you can combine or not yet, but I suggest that someone is going to ask
 what is the overall risk to a human being? Because there are so many endpoints that
 this chemical or any of the other chemicals we are looking at can cause, we may have
 to address an overall risk at some time.
       To just present one of the assessments from one endpoint may be thought as
 an underestimation of the risk because there are so many other things that can be
 caused by that particular chemical.  So how are we going to integrate all these
 different risks from the various specific endpoints into one assessment, or should we?
       Now, to do a genetic risk assessment for humans, we need  to have some
 human exposure figures. Generally speaking, the EPA is interested in exposure from
 inhalation, ingestion, and skin absorption for acrylamide (Table 3-4).
       For inhalation exposure, examples include exposure from the manufacture of
the monomer, and from the sewer line maintenance workers that I mentioned earlier.
      The OSHA  permissible  exposure level, PEL, is 0.03 mg/m2  for an 8-hour
average. When the EPA became concerned about the grout workers, this was one

                                    215

-------
Table 3-4. Acrylamide Exposure

Exposure occurs via inhalation, ingestion, and skin absorbtion

Inhalation: e.g., manufacture, sewer line maintenance

      • OSHA permissible exposure limit (PEL) is 0.03 mg/m3
       (8 hour average)
      • EPA sampling of line maintenance workers found 8-h
       time-weighted average exposures of 0.0008 -
       0.12 mg/m3

Ingestion: e.g., in drinking water from floccujent-treated water,
eating sugar (polyacrylamides used  in refining process)

      •Assume 100% exposure from drinking water
      • FDA limit of residual monomer in food processing not
       exceed 0.05% (500 ppm) (e.a. sugar refining limit)
      • Estimated exposure level in flocculent-treated water is
       0.01  ug/l

Skin absorbtion: e.g., grout workers for sewer repair, research
laboratories using polyacrylamide gels, manufacture


      • Assume -25% of dermal exposure is absorbed via skin
      • EPA sampling of sewer repair workers found dermal
       contact values estimated up to 5.0 mg/h
                           216

-------
 of the few times EPA itself has actually gone out on their own, sent their own people,
 and did all the monitoring.  The Agency monitored a group of line workers, and did
 a sampling of these people for inhalation and for skin absorption exposures.  What
 they found with workers who have an 8-hour time weighted average, was exposure
 from 0.008 to 0.12 mg/m2. The top value is about four times the PEL for inhalation.
 These are some of the numbers we can use in our break-out sessions. It would be
 very interesting to see how our calculations for government-regulated levels compares
 with the actual exposures. For ingestion, as I mentioned, you can get exposure from
 drinking water treated with acrylamide. And here is something that I didn't realize until
 recently. You can also get exposure from sugar.  Polyacrylamides are used in the
 refining process.  So,  for ingestion, it is not just  a  cohort  of 50 or 60 people
 working on the sewers we are  looking at.  There are millions of people being
 exposed.
       For ingestion, we are going to assume 100 percent exposure for drinking
 water. From one of the drinking water documents that came from the EPA's Office
 of Drinking Water, estimated exposure level for flocculent treated water is 0.01 mcg/1.
 When you run these numbers, the assumptions for drinking water are like 2 liters a
 day of drinking water for, I think, a 70 or 75 kg person. So, we need to perform these
 conversions when we do our risk assessment via this particular route of exposure.
       From the dominant lethal studies we know acrylamide can be absorbed and
cause effects by all these routes of exposure. So, we are  more confident these
exposure routes are appropriate for human  genetic risk assessment
                                    217

-------
        Now, for the monomer in sugar, I just thought I would put this up (Table 3-4)
 because it is of interest.  The FDA limits the residual monomer in food processing
 which should not exceed 500 ppm.
        The last exposure shown here is skin absorption (Table 3-4).  This is the one
 exposure scenario we may have some problems with because when these sewer
 workers are working, they are supposed to be wearing protective equipment. But the
 EPA has found they do have detectable exposures via the dermal route.
        Another major skin route is in the research laboratories since a lot of people
 use polyacrylamide gels.  How many of you know people who have come back with
 tingling fingers and skin peeling off their hands?  Those are some of the signs  of
 acrylamide exposure.
        What  the EPA has assumed is that 25% of dermal exposure is absorbed via
 the skin. We need to take the assessment figures we get and divide by 4 as one of
 our  risk extrapolation factors.  The EPA sampling of sewer  repair workers found
 dermal contact values up to 5 mg/hour.
       You can see that these workers are  getting measurable exposures. That is
 one  of the main reasons why we think there are problems with a high exposure like
 this,  and the  EPA is considering a ban on this use for acrylamide.  The Agency is
 using section 6 of TSCA which permits banning of adverse practices. This is one of
 the few times this section is being invoked.
       This slide is just a reiteration of what Dave Brusick discussed earlier (Figure
3-6).  These are the formulas for the doubling dose and the direct method using the
assumptions that the ICPEMC workgroup have presented on both the ethylene oxide
                                    218

-------
                         DOUBLING DOSE METHOD

         DD=   	Spont MR	

                          Induced MR / Unit Exposure



  RISK = SponHumiin. x  D / DD x N x REF


         Assumptions: UNSCEAR (1986) estimate of spon mutation
                     rate-1.5 X10"3
                     Sankaranarayanan (1982) estimate of spon
                     chromosomal aberrations rate = 6.2 X 10"8


                            DIRECT METHOD


         RISK = MMou" x LHunwn x D x N x REF
        Assumptions: Dominant single gene diseases = 1,000
                     Dominant chromosomal diseases -1-
        For point mutations, Ehling and Neuhauser-Klaus specific locus test
        For heritable translocations, Shelby et al. reports and Adler reports
                                Figure 3-6


and acrylamide assessments they performed. Some variables for the methods are

shown in Figure 3-7.

       In particular for acrylamide, we have the Ehling and Neuhauser-Klaus specific

locus test data to use for point mutations, and for heritable translocations, we will

probably use the Shelby et al. reports and the Adler reports.
                                  219

-------
                            REPS FOR ACRYLAMIDE
                Parameter	REF
                Locus Specificity                        2
                DNA Repair Variability                    0.1
                Metabolic Variability                      1
                Dose Rate Variability                     1
                Exposure Route                         1
                Germ Cell Stage Specificity               1
                Dose Response Kinetics                  1
                                  REF = 0.2
                                  Figures-?

        Now, the problem with the heritable translocations data is that, generally, they
 are point estimates.  I know Shelby's material has two dose points, but they were so
 close together they are essentially one dose point.
        Last night, Ilse-Dore Adler told me she actually has dose response data now
 for heritable translocations.  So, we will obtain that data from her and also use that
 in our assessment.
        For the point estimates, we weren't able to use  any of the mathematical
 modelling to describe a curve.  Now, with the newer data, we might be able to do
 some of the modelling we did with ethylene oxide and, hopefully, use the data to help
 extrapolate to lower doses.
       I think that is essentially it.  I expect we will have a lively discussion and I look
forward to the break-out sessions.  Thank you.
                                     220

-------
 Discussion

       DR. SORSA: Thank you very much. This is especially important to get this
 new information right now, because IARC is evaluating acrylamide next February.
       DR DEARFIELD: That is for cancer. Right?
       DR. SORSA: Yes, that is for cancer, but they are considering, of course,
 other....
       DR. DEARFIELD: Cancer data is available.
       DR. SORSA: Yes, and I was wondering. You didn't cite any human biomarker
 data. You just commented on the exposure levels.  So, there is as yet no data?
       DR. DEARFIELD: There is some hemoglobin adduct work that was done on
 the Chinese workers, but there is not a whole lot more biomarker data available with
 acrylamide.
       DR. SORSA: Yes, and then Udo.
       DR. DEARFIELD: His lab might actually have more on that.
       DR. EHUNG: You  were asking the question  what should we do with the
 different endpoints.  I think the answer is quite clear from the last figure you showed.
       Keep the different endpoints separated, because risk assessment starts with
 an endpoint, and for translccation, you get a different answer than for specific locus
 mutation, and if you use a doubling dose or the direct method, that you have to keep
 separated.  But what UNSCEAR  has done for many years is summarize the data.
 Here was the different method, and then for translocation, for gene mutation, and so
on, and I think that is the proper way to approach it.
      DR. SORSA:  I think most of us, at least, seem to agree with this.
      DR DEARFIELD: I don't think I entirely agree.
                                   221

-------
       DRSORSA: Aha, you don't?
       DR. DEARFIELD: The problem is, again, I go to my risk managers, and they
 ask, which one is more important, or should we add this all together. They don't
 understand  there are  differences between genetic endpoints.  So I am not really
 arguing with you, but emphasizing how we communicate the assessments.
       DR.  EHUNG:  UNSCEAR went one step further not only to estimate the
 mutation frequency you can expect, but what does it mean in clinical terms. That, of
 course, I mean risk estimation is a very risky business, and to try to translate it in
 clinical terms is more difficult, but it can be done, and I think the more brain you put
 into that work, the better the outcome.
       DR. DEARFIELD: That is the same message from this morning.
       DR. GENEROSO: Let me just bring in a few more points.  I would like to first
 emphasize that acrylamide is the very first compound that we have shown to be
 mutagenic in germ  cells by way of  dermal exposure.  There was  a good dose
 response, and it is almost a linear dose response where the lowest dose is fairly low.
       If you used that dose and tried to detect carcinogenesis or neurological
 syndromes, I think you would find it very difficult to find any effect of the lowest dose
 that produced a positive response a  dominant lethal mutation  system by dermal
 exposure.
      Second, acrylamide also is effective when you expose the mother during the
zygote stage of her conceptus and produces malformations very much similar to that
of ethylene oxide and ethylmethane sulfonate.
                                   222

-------
       Third, you discussed  a great deal  the  compound glycidamide  and the
possibility that it is the active metabolite of acryiamide in germ cell mutagenesis. We
have new dominant lethal data in mice for all of the spermatogenic stages.
       DR DEARFELD: I have been trying to get it.
       DR. GENEROSO: And gtycidamide is an effective inducer of dominant lethal
mutations.  The pattern of germ cell stage sensitivity is exactly the same as that of
acryiamide. The difference in mg/kg between acryiamide and glycidamide is very little.
       Lastly, glycidamide also produces the same malformations that acryiamide
produces when the mothers were exposed during the zygote stage of the conceptus.
       So, the similarities between glycidamide endpoints effects and acryiamide is
incredibly the same.
                                   223

-------
224

-------
 1,3-Butadiene

 Dr. Ilse-Dore Adler presented the summary for 1,3-Butadiene.  Dr. Adler is from the
 Instttut fur Saugetiergenetik in Neuherberg (Munchen), Germany.

       DR. ADLER:  I am supposed to introduce this figure (Figure 3-8). There is a
 categorization for butadiene as a carcinogen in animal experiments. There is a TWA
 level of 22 mg/m3 which is equivalent to 10 ppm.  Butadiene is metabolized to the
 mono- and the diepoxide.
       Marja gave me this overhead (Figure 3-9) that shows the activation  to the
 epoxybutene, and then the deactivation and excretion in the urine as mercapturic acid
 and butenediol.
       There is also activation from the monoepoxide to the diepoxide and then from
 there  on deactivation.  There  are probably a lot of other  metabolites  that Siv
 Osterman-Golkar might be able to tell us more about.
       The world production of butadiene comes to about 4 million tons. Butadiene
 is mainly used in the production of synthetic elastomers and polymers.  There is a
 growing market in paper lamination and styrene-butadiene rubber production.  About
 65,000 U.S. workers are exposed.  That is an estimate made that Marja got from the
 literature. Urban air contamination can reach up to 15 |tg/m3. Automobile exhaust
 is also a source of environmental contamination  with butadiene, up to 44pg/m3, and
 a side-stream tobacco smoke component is about 24jjg/tigarette.
       Concerning the genotoxicity in vitro data (Table 3-5), again this is a figure that
 Marja gave me, and was more or less positive if S-9 mix was provided for metabolic
activation in Salmonella tests and in CHO cells for SCEs. In human lymphocytes for
                                    225

-------
                     1 ,3-butadiene
                   H2C-CH-CH-CH2
MAK Category IIIA2 (carcinogenic in animal experiments)
               TWA 22 mg/m3 (10 ppm)
          Metabolized to mono- or di-epoxides
                CH2-pH-CHOH-CH2OH
                   CH2^pH-CH-CH
                       Figure 3-8
         Activation.
 Butadiene
Epoxybutene
          Activation^
Dlepoxybutane
           Deactlvatlon.
                     Deacttvatlon and
                    excretion In urine
               OH
                      + mercapturic acids
            OH
             Butenediol
                       Figure 3-9
                         226

-------
 SCE induction, S-9 mix was not required. Yet in mouse lymphoma cells, for gene
 mutations the results were negative even with S-9 provided.
 Table
                  GENOTOXICfTY OF 1 ,3-BUTADIENE IN VITRO
TEST SYSTEM
Salmonella
CHO cells
Human lymphocytes
L5178Y mouse
lymphoma cells
ENDPOINT
reversion
SCEs
SCEs
gene
mutations
RESULTS
+ (with S9 mix)
+ (with S9 mix)
+ (without S9 mix)
or
- (with S9 mix)
        In the animal, there is a striking difference between species. These are two
 graphs from the paper of Cunningham et al, 1986 (Figure 3-10). You can see in the
 upper graph that the toxicity to the erythro-poietic system is similar in mice and rats.
 The dashed line is the rat The solid line is the mouse. Only at the highest dose is
 there a difference in cytotoxicity, which  is higher in the mouse than in the rat. if you
 look for micronucleated PCEs in the lower graph, again, the solid line is the mouse,
 and the dashed line is the rat; there is no effect in the rat, and there is a strong effect
 from 100 ppm upwards in the mouse.
       In the same paper, data were provided for SCE-induction in bone marrow, and
 you see on the upper graph (Figure 3-11) in the  mouse a linear increase with dose
 followed by a plateau at higher doses.  On the lower graph A, in the low dose range,
there seems to be a slight increase in SCEs also in the rat. If you go to higher doses,
there is virtually no increase, and in the repeat experiments, this small initial increase
here was not confirmed.
                                    227

-------
      Bone Marrow Micronucleus Test After
           1,3-Butadiene Exposure for
             (6h/d on 2 d, nose only)
             Cunningham et al., 1986
    £  1.0
    s
    111
    o
        0.5
                  10      100    1000   10,000
                     1,3-butadiene (ppm)

 	  of BD in bone marrow cells.  PCE/NCE ratios of
five B6C3F, mice, •, and five Sprague-Dawley rats, O, in
each dose group were measured 24 hours after treatment.
Error bars represent the standard error of the mean.
                        100     1000    10,000
                      1,3-butadiene (ppm)
Induction of micronucleated PCEs in bone marrow of
B6C3F, mice (•, •) and Sprague-Dawley rats (O, D) from
two independent experiments. Error bars represent the
standard error for the mean of five animals.
                      Figure 3-10
                        228

-------
              Bone Marrow SCE Test After
               1,3-Butadiene Exposure for
                 (6h/d on 2 d, nose only)
                 Cunningham et al., 1986
                 100         300         500
                       1,3-Butadiene (ppm)

  Induction of SCEs in bone marrow cells of B6C3F, mice. Error bars
  represent the standard error of the mean of five animals
                              300
500
                                              B
                                          f
                  2000        6000        10,000
                       1,3-Butadlene(ppm)

Induction of SCEs in bone marrow cells of Sprague-Dawley rats. Error
bars represent the standard error of the mean of five animals. (A) lower
exposures; (B) higher exposures.
                          Figure 3-11
                            229

-------
        Basically, in the mouse, there is a good micronucleus response with various
 protocols and various strains of animals as was compiled by Tom Skopek in his
 summary paper on somatic in vivo effects (Figure 3-12).  The fold increase is given
 on the Y-axis here, and accumulated ppm times days as accumulated doses is given
 on the X-axis. You can see that there was virtually a positive response for all sorts of
 strains of mice.
        A little bit more scanty was the chromosomal assay data (Figure 3-13), and the
 chromosomal aberrations don't respond as sensitively.  That is the general picture
 one gets from this chart as compared to micronucleus induction.
        This was about the state of the art when we started with our germ cell project
 which was funded by the European Commission (EC), Environmental Programme, in
 Brussels; Frits Sobels convinced us to take up butadiene as one of the chemicals to
 be studied in germ cells.
       The project was initially called "Detection of Germ Cell-Specific Mutagens", and
 then we felt that we were looking for a needle in a haystack. After long discussions,
 it was renamed to "Detection of Germ Cell Mutagens", leaving  out  the specific.
 because until today, we have not found a chemical that  only acts in germ cells and
 not in somatic cells.
       We coordinate that project, and there are, altogether, six laboratories involved
 in the studies (Figure 3-14). Additionally, there is a lot of collaboration going on with
other groups (Figure 3-15).  Cooperation with Johannes Riser at the Toxicology
Institute of the GSF  provided  inhalation  chambers  and the knowledge of how to
expose the animals.  Also within the GSF, we collaborated with Michael Nusse who
has performed the measurement of micronuclei in peripheral blood reticulocytes.

                                    230

-------
O
30


25


20




10


5
-B-B6C3F1.PCE.Ref. 1
•4- B6C3Fl.NCE.Ref. 1
 *  B6C3F1.PCE.Ref.3
-»-B6C3Fl.NCE.Ref.7
 e-  B6C3F1.PCE. Ref. 7
 •>  NMRI, PCE, Ref. 8
•*- 102XC3H.7. Ref.9
-*- 102XC3H.PCE.Ref.9
-+-B6C3Fl.BM,Ref.9
-»-B6C3Fl.PCE.Ref.9
    10     30     100   300    1,000  3,000  10,000  30.000  100,000

                       ppm x days
                                 Figure 3-12
    10
                                                           NIHSwisa,CA, Ref.2
                                                           B6C3F1.CA,Ref.2
                                                           BALBxDBA.LacZ.Ref.6

                                                           B6C3F1,CA,Ref.7

                                                           102 x C3H, HPRT, Ref. 9

                                                           B6C3F1.HPRT, Ref. 10
     20     SO   100  200    500   1,000 2,000    5,000 10,000
                           ppm x day*
                                 Figure 3-13
                                    231

-------
                      STEP-CT91-0144
        Detection of Germ Cell Specific Mutagens (1991)
      Detection of Germ Cell Mutagens (since Estoril, 1992)
 Coordinator:
 Project Leaders:
I.-D. Adler, Neuherberg, Germany
U.H. Ehling, GSG, Neuherberg, Germany
R. Benign!, ISS, Roma, Italy
D. Anderson, BIBRA, Carshalton, UK
A.D. Tates, MGC, Leiden, The Netherlands
F. Pacchierotti, ENEA, Roma, Italy
J. Laehdetie, University of Turku, Finland
                         Figure 3-14
Johannes Riser,
Neuherberg

Michael Nusse,
Neuherberg

Marja Sorsa,
Helinski
Hans-Gunter
Neumann, Wurzburg

Siv Osterman-
Golkar, Stockholm
Inhalation exposure of animals to 1,3-butadiene
and ethyleneoxyde

Micronucleus analysis in blood (flow cytometry) of
1,3-butadiene- and ethyleneoxyde-exposed mice

DMA adduct measurement in tissues (testes, heart,
lung) and micronucleus analysis in blood of
1,3-butadiene-exposed mice

Hemoglobin adduct measurement in blood of
1,3-butadiene-exposed mice
                      •
Hemoglobin adduct measurment in blood of
1,3-butadiene- and ethyleneoxyde-exposed
mice (rats)
                        Figure 3-15
                            232

-------
       Marja Sorsa  collaborated with us from  another EC project on butadiene
 biomonitoring.  She also exposed animals in the facility of Johannes Riser. Hans-
 Gunter Neuman and his coworker Olaf Albrecht from the Toxicology institute at the
 University of Wurzburg collaborated by performing hemoglobin-adduct measurements
 for both the experiment with Marja and the experiment with our group. And I think that
 Ad Tates sent specimens from mice exposed in the same setup at the  GSF to Lars
 Ehrenberg's group in Stockholm for a collaboration there with Siv Osterman-Golkar.
       Now I would like to show you  the micronucleus data that we got  after
 butadiene exposure for 6 hours per day on five consecutive days (Figure 3-16).  This
 graph contains the data from both experiments, the one carried out in Marja Sorsa's
 Institute but exposed at the GSF, and the one from our lab.  There seems to be a
 strain-related difference in the absolute yields  of micronudei, but the shape of the
 dose-response curve is pretty much the  same, namely a linear increase up to 500
 ppm followed by a  plateau up  to  1300 ppm.   Circles  designate bone  marrow
 experiments, squares designate  experiments with peripheral blood.   The open
 symbols represent bone marrow  and peripheral blood data from Helsinki.  Both of
 these curves for peripheral blood were obtained with the conventional technique of
 microscopic scoring, and we have started to do a flow analysis of peripheral blood
 micronudei (closed triangles). We have now obtained the data for the highest dose
 also, and they all fall in the same range of the dose-response curve. The micronucleus
responses in the same strain are pretty  much the same for bone marrow and for
peripheral blood.
       Some of the exposed animals were pregnant females, and Agelika Neuhause-
Klaus at the  Institute did a mouse spot test (Table 3-6). The animals were 500 and
                                    233

-------
Micronucleus-Test Results 18-24 h after 1,3-Butadiene-
Inhalation Exposure (6h per day on 5 consecutive days)
         200
400    600    800    1000   1200
     Exposure (ppm)
1400
Circles = bone marrow micronucleus test.
Squares = peripheral blood micronucleus test, manual scoring.
Triangles = peripheral blood micronucieus test, flow cytometry.
Bars represent 95% confidence limits.

Hollow symbols, dashed lines = M. Sorsa Helsinki
Filled symbols, solid lines = I.-D. Adler, Neuherberg
                     Figure 3-16
                        234

-------
 Tabte3-6. Testing of 1,3-butadiene in the spot test with OxHTJF, mouse embryos
 (500 ppm for 5 h/d on the 8th, 9th, 10th, 11 th and 12th day of gestation)
Treatment
Control'
Butadiene
Females
treated with
litter
37/34
19/17
Birth
No. of Ft
(8V.I.S.)
271 (8.0)
164(9.6)
Weaning
No. of F,
(av.l.s.)
257 (7.6)
156(9.2)
Offspring with
coat color spot
No. %
6 2.3
15b 9.6
 'Untreated  control in SPF (2/133) and controls kept under similar conditions as
 treated animals (4/124), P = 0.16 (2-tailed).
 bP + 0.0056a
 exposed on the 8th, 9th, 10th, 11th, and 12th day of gestation, and Angelika looked
 for genetic relevant spots in the fur of the progeny about three weeks after birth.  You
 see here a statistically significant increase of animals with .genetically relevant color
 spots in the fur.  The background rate was 2.3%, and the observed  rate in the
 experimental group after 500 ppm was increased by about four-fold to 9.6%.
       This result indicates that butadiene-inhalation exposure of females has a trans-
 placental effect, first of all, and,  secondly, it caused gene mutation.  Even though
 there are other possible genetic origins for these color spots,  they  are probably
 mostly gene  mutations, because the genetic markers (coat color genes) that are
 looked at in the mouse spot test are partly the same as in the specific locus test.
       Then,  Tates performed the spermatid micronucleus test, and he got animals
from us that were exposed to 500 and 1500 ppm (Figure 3-17).  He sampled the
sperm 2, 5,11,  and 15 days after exposure.  As you can see, there is an increase at
                                    235

-------
Micronuclei in Early Spermatids of Mice
               A.D. Tates

      M-SPTD/1000SPTD
                5      11
              Sampling Days
15
    -*- Control • 500 ppm CH 1300ppm
                Figure 3-17
                  236

-------
 1300 ppm 15 days after the end of exposure. With the higher dose of 1300 ppm a
 slight effect was also seen after 5 and 11 days. The samples taken after 15 days
 correspond to treated spermatocytes around the time of the last DNA synthesis
 period. Interestingly, though,  the dose-response seems to be inverted. The higher
 effect is seen at 500 ppm, and cytotoxicity at the higher dose might be the reasons
 for this observation.
       We also did a dominant lethal experiment,  and  another  dominant lethal
 experiment was done by Diana Anderson at BIBRA (Figure 3-18).  We exposed our
 animals to 1300 ppm of butadiene for 6 hours per day over 5 days, and then mated
 the animals at the end of that exposure time at weekly intervals over a period of 4
 weeks. As you can see from  the hatched bars, the increase of the dominant lethal
 effect was significant in the second week after treatment, and then it came down to
 control levels at the fourth week. Diana treated the animals with 1250 ppm, 6 hours
 per day, 5 days per week for 10 weeks, and then she mated the animals for one week
 after that. As you can see from the striped bar, she got about 29% dominant  lethals.
 I think that was the early calculation including late death. Later on, it came down to
 more like 28% early dead implants in the uteri of the females. And this is interesting,
 if you sum up the values obtained during the first three weeks in our experiment
 (hatched bars) after one week of exposure, you will come to about 23%, as compared
to the 21% in Diana's experiment with 10 weeks of exposure. So, I would conclude
here that what she observed after 10 weeks of treatment were actually only the effects
of the last 3 weeks of treatment that have accumulated.
       Now, both of these experiments were done at a very high dose, but they do
show  that dominant lethality was induced, and together with the spermatid data we

                                   237

-------
      Dominant lethal test after Inhalation exposure
      to 1,3-butadiene
CO

f
   30
   25
   20
CO

I 15
o
Q
xP
* 10
            * *
             1
                  1250 ppm, 6h/d, 5d/w, 10w
                  (BIBRA)
1300 ppm, 6h/d,5d(GSF)

 * *
                    Mating Week
                   Figure 3-18
                     238

-------
 demonstrated a mutagenic effect of butadiene-inhalation exposure in mouse germ
 cells.
        In her chronic experiment, Diana did  not only look  for the early  dead
 implantation rate, but she also looked at late deaths, and at abnormal fetuses (Table
 3-7). This middle column here does not represent the high dose level but the 12.5
 ppm dose given for 5 days per a week on 6 hours per day over 10 weeks. So, at this
 stage, remember the TWA levels with 10 ppm and the experimental exposure of 12.5
 ppm, which was very dose.  At this low dose, Diana observed a significant increase
 of late  deaths and also a significant increase of malformed fetuses.  Both of these
 effects  were also seen at the high dose. These values for the high dose are lower
 than for the lower dose,  but nevertheless they are significantly above the control.
 Again,  there seems to be  a reversed dose-response similar to the spermatid
 micronudeus data.  Probably, the early embryonic deaths prevailed at the high  dose
 and thereby reduced the malformed fetuses and late deaths.
        So, this is where we  stand at the moment (Table 3-8).  Genotoxicity of 1,3-
 butadiene inhalation exposure in vivo and DNA/hemoglobin adducts in the mouse are
 positive. Single strand breaks in DMA extracted from treated mice is positive. SCEs
 is positive.  Micronudeus induction is positive both in bone marrow and in spermatid.
 Gene mutations is positive in mouse spot tests. Chromosomal aberrations are positive
 both in bone marrow and in germ cells, in terms of dominant lethal induction, and we
 do get congenital malformations induced in male germ cells.
      Contrastingly, in the rat, adduct measurements did not show an effect. Single
strand breaks in DMA is the only positive observation.  For SCEs, there is a positive
and a negative response in the two experiments I showed you.

                                    239

-------
 Table 3-7.

Number of males
Number (and %) of males mated to at
least on* female
Number of females
Number (and%) of pregnant femeJee
Number of pregnant female* used In
dominant lethal assay"

T
O
T
A
L
S
M
E
A
N
S

NtlfflD^f Of IfflpMtflfll
Number of early death*
Number of late death*
Number of lei* death*
Including dead foetuses
Number of abnormal foetuee*

Implantation*
Early death* per Implantation
Late death per Implantation
foetuee* per Implantation
Abnormal foetuses per
ifHpwnuoof)

Control
25
23<92%)
SO
41 (82%)
23
278
13
0
2
0
12.00*1.276
0.050*0.0587
0
0.007*0.0222
0

i reainienT group
12.5 ppm
25
25(100%)
50
45(90%)
24
306
16
7
6
7*
12.75BS±Z507
O.OS3M±0.0561
0.023**±0.038
0.026*±0.0424
0.024**0.0620

1250 ppm
48d
43(90%)
96
74(77%)
36
406
87
6
7
3f
10.68**±3.103
0.204** ±0.1611
0.014*"*±0.032
4
0.016°' ±0.0339
0.011**±0.0439
         6 hr per day, 5 day* per week, 10 week*
         Each male housed win 2 females for up to one week
         Not more than one from each mala.  (The other female* were allowed to litter • data to be reported
         2/50 make died during the treatment (one at week 5, one at week 7)
         4 Exenoephalle* (3 In litter), 2 runt* (575% of mean body weight of remainder of litter), 1 foetus with blood
         In the amntotic sac but no obvtoue gross malformation (stgnlfloano* of difference not altered If tht* foetus

         1 HydreoepheJy, 2 runt* (1 runt was In a litter of 2)
         not significantly different from control
*, **, ••*, significantly different from control, pSO.05; psO.01, sO.001
Statistical analysis by analysis of variance and least significant different tost on transformed data (arc-sine)
ns
                                                 240

-------
 Table 3-8. Species comparison of dose of expoxybutene in blood, calculated from
 hemoglobin adduct levels, and predicted dose in the body of expoxybutene following
 exposure to butadiene
Sp«ciM





Mo.
B6C3F1
CD2F1
ftato
Wtotv
Spr.-Dmv.
HumM.

Adduct
L*v*l
(pmol/g
porppmh)



0.5
0.3

0.00
0.3/0/1**
0.004

DOM
(nMn pot
ppmh)*




17
10

2.3
7.5/2.5**
0.30.510.
15
rlvwHlVv
DOM





7.0
4.1

0.9
3/1**


Predicted
DOM for
*6h
Expouir*
to 100
ppmBD
10.5


2.2




DOM
(nMh p»r
ppmh)
n-^-ii.g-.J
rnKUCwO


17.5


3.6




FMattv*
DOM




4.8


1




 * calculated from adduct levels,
 ** determined at 10ppm and 100ppm, respectively
 "Osterman-Golkar et al., 1993
 BKohn and Melnick, 1993
       Micronuclei were not induced in the human population biomonitoring studies;
likewise, SCEs were  not found to be increased, neither  were micronuclei nor
chromosomal aberrations.
       So, there  is a  real difference between these three mammalian species.
Although I might get into trouble if I explain this table here in detail which was put
together by  Siv, I wanted you to just look at this last column.  This is a species
comparison of dose of the monoepoxide in blood calculated from hemoglobin adduct
levels. The  relative dose differs vastly between mice, rats,  and  humans. So, the
species difference in metabolism of butadiene is the take-home message. And we
                                    241

-------
 have to discuss in the individual butadiene working group the following questions:
 Do we really have a good basis, a firm basis, for extrapolating mouse data to humans,
 or is the distinct difference between the two rodent species a cause for assuming that
 the rat is a better model for man than the mouse?
 Discussion
        DR. OSTERMAN-GOLKAR: I compared dosage of the monoepoxide in these
 species based on adduct measurements  and based in pp-pk modeling, and these
 two methods compare well,  but the monoepoxide is probably not the  important
 epoxide. Probably, the diepoxide is much more important. So, you cant really base
 risk estimations on these comparisons.
       The adduct levels are very low of the monoepoxide, and the difference of a
 factor of 2 to 5 between mice and rats is much less than the difference In other tests
 for genotoxicity.
       DR. ADLJER: Does that adduct level reflect the amount of monoepoxide formed
 or present before ft is going to be metabolized into something else?
      DR. OSTERMAN-GOLKAR: This is the dose.  It is the concentration of the
 epoxide times the time the epoxide is there.
      DR. ADLER: But even if the diepoxide was the more genotoxic agent, it could
 only be formed by the monoepoxide.
      DR OSTERMAN-GOLKAR:  Yes.
      DR. ADLER: So, if there is more monoepoxide, there is a better chance that
the diepoxide is formed.
                                   242

-------
       DR. OSTERMAN-GOUCAR:  There is only one study where the DMA adducts
 have been measured. This is from the study by the Institute and co-workers, and they
 looked at different types of DMA  adduct, and here you have the adduct  of the
 diepoxide, and this is the adduct of the monoepoxide, and this is from the mouse.
       You  can see that these peaks are of the same  amount of time, and the
 diepoxide is so much more effective than the monoepoxide. I think this indicates that
 the diepoxide is the important metabolite. In the last one of these peaks, we are not
 seeing that.
       DR. SORSA: Well, butadiene is a very interesting compound, as you  will all
 leam to know, and I think one of the problems which we are facing now  is the
 differences which were seen already in the somatic genotoxicity endpoints.
       So, we have in the mouse in the liver the main metabolic pathway towards the
 diepoxide which is the more potent also for genetic toxicity endpoints. In the rat, the
 major metabolic pathways goes towards the  diol formation.  Obviously, the human
 pathways are more towards the diol formation; but I think,  here again, it is important
 to point out that the polymorphism in human metabolism as well seems to play  a role,
 but we just don't know enough about that yet.
       For instance, the cytochrome 2E1 polymorphic enzyme system is responsible
 for the butadiene metabolism. So we might find individuals among the human species
 which have a pathway, which is nearer to the mouse pathway.
       So, I think taking butadiene as an example for the parallelogram approach has
 many possibilities and it is  certainly an important compound, and highly genotoxic.
 It is a carcinogenic compound, and, therefore, it is of importance as well concerning
the genetic risk information.

                                   243

-------
       Certainly, I think during these days, we will pick up the genotoxlclty knowledge
 and which are the important steps and information which need to be made and
 reached during the next couple of months.
       DR. ANDERSON: Just to take up on that point of Marja's at the end there,
 where she was saying that it might be too early; but if you do recall, Frits' original
 parallelogram had mouse data in it. Now, if we didn't know about the rat, we would
 be doing a very nice risk assessment with the mouse.
       So, we have to bear in mind, I think, that rats could be equally important in this
 methodology that Frits has suggested, the parallelogram.
       DR SHELBY:  Agreed.  I don't think it is premature to work with butadiene.
 I think it presents a unique data set that we will have to deal with In the future, and
 we are just fortunate here to know the species differences.
       DR. EHRENBERG:  I wanted to say first that Ilse-Dore put my name in the table
 at the bottom. She talked about my group.  I am an old man. This was a long time
 ago, but I will make the proper references.
       Since I am an old man,  I would like to show just some data. The first here is
 on the mutagenicity per alkylation of DMA from ethylene oxide and epoxybutane in
 mutation in the 1000 loci. As you see, it is a factor of 100 in favor of the diepoxide.
 This is from my own publication in 1957.
       It  is correct that I  am involved in some work but more as  an advisor in
 independent study rather than the one you referred to on comparisons of the various
 metabolites of butadiene and efforts to measure mutation,  measure doses of  the
diepoxide.
                                    244

-------
       This shows here the relative mutagenidty by a factor of 100 between the
 butadiene monoepoxide and the diepoxide, in CHO cells.  If you use the CHO cells
 which are deficient in UV damage repair this increases to a factor of 10,000, about.
       So, as in bacteria, there is a relationship between repair of the damage caused
 in this type of cells and the UV damage which is one type of damage. In 3,4,6,7, you
 have also facts by about a factor of 100.
       I see that the paper referred to Skopek, or was it Moore?
       DR SHELBY: Cochrane and Skopek.
       OR  EHRENBERG:  Cochrane and   Skopek, yes, which are submitted to
 carcinogenesis which we have not seen so far with similar data for differences
 between the epoxides related. It is certainly true, as she has said here, that it is very
 difficult to make the risk estimation without proper knowledge of the doses in the
 target tissues of the diepoxide.
       DR SHELBY: Thank you, Lars. Are there other questions or points to make?
       Martha,  is there an obvious explanation for the L51 negative data that she
 referred to?
       DR MOORE: I don't know. There may be someone who can address this.
       DR SHELBY: I think Marja has something else to say.
       DR SORSA: Thank you.  One obvious explanation is that it is very difficult to
work with butadiene, because you have to have it in liquid form, and you can only
have it in liquid form at minus 4 degrees. So, I think this is one of the reasons for
discrepancies.
       Since we do have some time, maybe I will continue with a further discussion
about the importance of the monoepoxide as compared to the diepoxide.

                                   245

-------
       These are our lowest effective concentrations in humans in vitro, in human
 peripheral blood lymphocytes and in germ cells, so that you get the activation. The
 difference in focusing on the dose effect relationship of the diepoxide as compared
 to the monoepoxide in vitro in human lymphocytes is 20 times, in the germ cells 10
 times.
       So, I think even though  the diepoxide is more potent as a genotoxic agent
 and, obviously, as an alkylating agent, (and, actually, it  is used as a diagnostic tool
 to diagnose for lymphoma in patients), still, the monoepoxide is important, because
 it is certainly genotoxic.
       Da SHELBY:  Martha Moore?
       DR. MOORE:  Actually, Mike, in response to your question, can someone tell
 me whether the lab that did the mouse lymphoma data had good recovery of the
small colony mutants?  Do you know?  That may be the  answer to the question,
because it appears that butadiene is positive in other in  vitro assays, so that may be
the problem.
                                   246

-------
 Cyclophosphamide (CP)

 Dr.  Diana Anderson  presented  the  data summary for  Cyclophosphamide.   Dr.
 Anderson is from  BIBRA Toxicology  International, Department of Genetic and
 Reproductive Technology in Carshalton, Surrey, United Kingdom.

       Da ANDERSON:  Cyclophosphamide, N,N-bis(2-chloroethyl)tetrahydro-2H-
 1,3,2-oxaphosphorin-2-amine,2-oxide monohydrate, has a molecular weight of 279.1.
 It is a fine, white, almost odorless crystalline product.  It is soluble  at 20 degrees
 centigrade in 25 parts of water, 1 part ethanol in benzene, chloroform, dioxane, and
 glycols.   Aqueous solutions may be held for two hours at room temperature, and
 hydrolysis occurs at about 30 degrees centigrade with removal of chlorine atoms.
 During its production, it is  treated  with propenalamine  in the  presence of
 trimethylamine and dioxane, and  1000 kg are produced annually.
       It is used only as a drug and particularly widely so in the treatment of various
 cancers.  It is an  immuno-suppressive agent used  in a variety of  non-malignant
 diseases including rheumatoid arthritis, systemic lupus, erythematosus, scleroderma,
 glomerulenephritis, chronic intestinal pneumonia, psoriatic arthritis, multiple sclerosis,
 and chronic hepatitis; and it is often used as an immuno-suppressive agent following
 organ transplantation.
       It has various special physiological properties. It has an intravenous \JDX of
 160 mg/kg body weight for the rat, 400 for the guinea pig, 130 for rabbits, and 40 for
dogs, and it has an oral LD^, of 180 mg/kg body weight for rats.
       Leucopaenia is the predominant hematological effect in mice, rats, dogs, and
man.  It requires metabolic activation and it is rapidly absorbed and  metabolized in
                                    247

-------
 animals and man. It has a half-life in plasma of 6.5 hours. It is excreted in the urine,
 but not in air or in the faeces, and CP and its metabolites are found in bile, milk,
 sweat, saliva, and cerebral spinal fluid.
        Much  of the work I shall refer to from now on has been provided to me as
 mini-reviews on DMA adducts (Colin Garner and McConnell), mammalian cells in
 culture (Patricia Ostrosky), and germ cells (Jack Bishop).
        Cyclophosphamide is used extensively as a therapeutic agent, used in several
 regimens for the treatment of cancer.  It is a cyclic derivative of the alkylating agent
 nor-nitrogen mustard and consists of a phosphoramide ring linked to a bifunctional
 moiety containing two chloro-ethyl groups (Figure 3-19).
        Tumor cells have been shown to contain high activities of phosphoramidase
 and phosphatase enzymes.  It is  a pro-drug, and it becomes selectively activated
 within tumor cells by phosphoramidase cleavage of the ring, and this results in the
 release of the toxic bifunctional alkylating species at the target site.
        It needs to be metabolized before it becomes active and the primary activation
 occurs in the liver. Hepatic microsomal cytochrome P-450 monooxygenase enzymes
 metabolize   Cyclophosphamide,   producing   the   proximate   metabolite
 4-hydroxycyclophosphamide (Figure 3-20).  This  product equilibrates with its ring
 opened tautomer aldophosphamide which is further metabolized by aldehyde oxidase
 and aldehyde  dehydrogenase  to  the  metabolites aldophosphamide,  4-keto
 phosphamide  and  carboxyphosphamide.   Aldophosphosphamide may undergo
spontaneous  p   elimination yielding  the  cvtotoxic   products  acrolein,  and
phosphoramide  mustard (Figure  3-21 a  and 3-21 b).   Carboxyphosphamide is
enzymatically cleaved to nor-nitrogen mustard  (Figure 3-21 c). Chloroacetaldehyde
                                    248

-------
                              /CH2CH2CI
                          P-N
                    o/    ^CH2CH2CI
                  Cyclophosphamide
                      Figure 3-19
CyclophosphamM*      44
-------
a)              CH2=CH-CHO
                    Acrolein
b)         CI-CH2-CH2        /NH2
                         N-P=0
           CI-CH2-CH/          OH
              Phosphoramide Mustard

c)             CI-CH2-CH2
               CI-CH2-CH
                Nomitrogen Mustard

      Cytotoxic metabolites of cyclophosphamide metabolism
             Figure 3-21
                250

-------
 is  also  formed by  N-oxidation.   Phosphoramide  mustard is  the  main ultimate
 alkylating species generated by cyclophosphamide metabolism.

 Adducts In Vitro

       Cyclophosphamide in the absence of metabolic activation fails to bind to ONA.
 The reactions of phosphoramide mustard with guanosine and deoxyguanosine at 37
 degrees centigrade at pH 7.4 was investigated by Mehta et al. (1980, 1982). The
 products were separated by HPLC and identified by UV spectroscopy and field
 desorption and revealed as N7guanine derivatives. The main adducts were labile
 with a half-life of 2.3  hours and were readily transformed into imidazole ring-opened
 derivatives, suggesting that they would readily  undergo secondary reactions with
 biological systems.
       Vu et a/. (1981) found that phosphoramide mustard reacts with guanosine-5'-
 monophosphate  forming  three  types   of  adduct:  a  cross  linked  dimer
 N,N-bis(2-[5'phospho-7-guanosinyl)-ethyl]phosphorodiamidic acid, (Figure 3-22c) two
 monoalkylated  products,   N-[2-(N7-guanosine-5'-monophosphate)ethyl]
 phosphorodiamidic    acid   (Figure   3-22a)   and   2-chloro,2'-(N7-guanosine-5'
 -monophosphate) diethylamine (Figure 3-22b).
       More recent work (Macubbin ef a/., 1991) has revealed that phosphoramide
 mustard forms a phosphoester adduct with 2'-deoxyguanosine 3'-monophosphate and
 DMA. A modification of the ^P-postlabelling technique (Macubbin, 1991) was used
to detect adduct levels of one adduct for every 100,000 normal nucleotides. Early
work investigating the biological properties of cyclophosphamide within microsome
enzyme systems demonstrated the generation of genotoxic alkylating species (IARC,
                                    251

-------
                          CHaCH,CI
CH,CH2CI
                                       B
The structures of guanine adducts formed by the reaction with phosphoramide mustard
•RP = ribose phosphate

                              Figure 3-22
                                  252

-------
 1987).   Work  by Hemminki (1985) studying the binding of  (chloroethyl-3H)
 cyclophosphamide to DMA revealed 3 different N7 substituted guanine adducts.

 Adducts In Vivo
       DNA adducts have been formed following the treatment of mice and rats with
 3H labeled cyclophosphamide; the work has been carried out by Hemminki (1985)
 and Benson ef a/. (1988), respectively.  3H-cyclophosphamide to  mice produced
 maximal DNA binding between 2 and 7 hours after injection.
       As  in in vitro studies, in vivo  studies show that cyclophosphamide  was
 metabolized to phosphoramide mustard. This reacted with guanine, producing a DNA
 adduct   undergoing   spontaneous  dephosphoramidylation   forming   the
 hydroxynornitrogen mustard.  Adducts were present at the levels of 4 adducts x 107
 normal nucleotides for mice and 2 adducts x 107 for rats.
       Macubbin ef a/. (1991) detected a cyclophosphamide derived phosphodiester
 DNA adduct in mouse liver. DNA was at a level of 3 adducts/108 normal nucleotides.
Adducts In Humans

       As to adducts in humans, molecular dosimetry work by Dr. Gamer showed that
DNA adducts increase an individual's  risk  of developing cancer for a number of
different DNA adducts, but the use of molecular techniques for CP derived alterations
in humans have yet to be carried out successfully.
       The ^P-postlabelling method developed by Macubbin et al in 1991 represents
the best prospect for examining adducts. Monitoring of adduct formation in patients
                                    253

-------
 may help to optimize therapy and detect individual differences in metabolism and
 excretion.  Thus, it would be useful to get a such measurement in humans.

 Somatic Mammalian Cell Genotoxicity In Vitro and In Vivo
        Dr. Ostrosky has examined over 200 papers on the somatic genotoxic effects
 in vivo and in vitro of cydophosphamide in mammals, including humans.
        These papers have confirmed that cydophosphamide shows positive effects
 in the presence of metabolic activation in all endpoints evaluated in somatic cells.  It
 has been employed as a positive control in many studies, in fact, where activation of
 the compound is evaluated.
        Only  one negative result has been found in humans, and this was in  a
 monitoring study designed to evaluate the adequacy of protective measures among
 the workers; and this study was performed by Marja Sorsa and colleagues.
        Twenty-eight percent of the positive data  without exogenous metabolic
 activation have been  reported in vitro and some authors daim  their cells have a
 capacity to metabolize CP when there is enough time to induce it. So, positive effects
 can be obtained even in the absence of metabolic activation when there is a sufficient
 period in culture. However, metabolism may not be due to the usual route in the liver,
 but to some spontaneous hydrolysis products and in  the  presence of metabolic
 activation; the effects are several times more potent than in the absence of metabolic
 activation.
       Different sensitivities  to cydophosphamide have been reported in animal
spedes, strains, and individuals. It has been found that mice are more sensitive than
rats. These are more sensitive than hamsters, which are more sensitive than humans.
                                    254

-------
       Two HPRT gene mutation studies have been carried out with exposed human
 beings, and only one study in animals.  These are positive.
       In general, the gene mutations in vitro have been performed in V79 HPRT and
 L5178Y mouse lymphoma cells.  Nine positive results have been found with metabolic
 activation and one negative without.
       Dose-related effects for all the endpoints have been found with CP and there
 appears to be  a maximum dose and an optimum time for the detection of effects.
 This is due to toxicity, since very high doses of CP will affect cell division in vivo and
 in vitro, and genotoxic effects will not be detected if the doses are too high.  DNA
 repair and adaptive responses also influence the outcome of the studies.
       The tables summarize the results. While there are some weakly positive and
 negative results most of them are, in fact, positive (Table 3-9).
 Tabie 3-9.    ECAJS Workshop on Hsk Assessment F>aralletogranVChernicaI Matrix
             Somatic Genotoxic In Wfiro Effects of Cydophcspnamide on Mammals
             Including Humans
             No. Assays                         Results
              NM   77                        22    8      47
              M     59                        54    5      0
              M    with exogenous metabolic activation
             NM    without
              +    positive
             (+)    weakly positive
                    negative
       Also different endpoints, cells and tissues have  been examined  including:
body fluids, host mediated assays,  DNA damage, sister chromatid exchanges,

                                    255

-------
 micronuclei,  chromosome aberrations, lymphoma cells,  lymphocytes  and gene

 activation. The number of assays conducted and the results are tabulated (Table 3-

 10). Again, there are very few negative results; there are some equivocate, but in fact,

 most are positive. So, there is a great deal of data on cydophosphamide in mammals

 and human somatic cells.


 Tabte3-10.   EC/US Workshop on Risk Assessment Parallelogram/ Chemical Matrix
             Somatic Genotoxic A? vim Effects of Cvctophosphamide on Mammals
             and Human Somatic Cells
 Endpoint
No. Assays
F
H
MAMMALS
D
S
M
CB
CL
HUMAN
S
C
G
12
13

4
29
35
23
6

7
6
1
 F     Body fluids
 H     Host-mediated
 D     DNA damage
 S     Sister Chromatid exchange
 M     Micronuclei
 C     Chromosomal aberrations
CB     Chromosomal aberrations in bone marrow
CL     Chromosomal aberrations in lymphocyte
 G     Gene mutation
 +     positive
(+)     weakly positive
       negative
Results
10
9
2
28
31
22
6
5
5
1
1
4
2
1
4
1
0
1
1
0
1
0
0
0
0
0
0
1
0
0
                                  256

-------
       Table 3-11 examines the cell type, the  dose range, the results, and the
 numbers of studies in vivo, in mice, rats, hamsters, and humans. Nearly all results are
 positive, and there are a vast number of studies.
       There is also a large continuing table (Table 3-12) from a review paper that
 was produced by Heddle, (1982).  The first part shows £ coli and Salmonella
 bacterial data, the types of responses that were found, and whether activation or not
 was required. The result column shows the lowest positive result obtained  or the
 highest dose tested and  not  found positive.   Mostly,  results  represent positive
 responses.
       The other parts of Neddie's review table move from  the  prokaryotic to
 eukaryotic systems including  yeasts  and  plants,  Drosophila, mouse, human
 lymphocytes, rabbit bone marrow, mouse bone marrow, Chinese hamster and all the
 way through to humans. This paper was published in the early '60s, and there have
 been more data since, but even then, there were already many studies.  Again, the
 doses that were  used are listed, with comments about the shape of  the  dose
 response curve (for example,  describing whether dose responses  are simple or
 complex, and the  sampling times used and  so on).
       Work is described on the more complex  animal systems involving not only
 somatic cells but germ cells as  well, including the mouse spot test, mouse oocytes,
 and  aneuploidy systems.   More recently,  some work has been done  by  Dr. F.
 Pacchierotti in Italy with aneuploidy and spermatocyte aberrations, Chinese hamster
spermatogonial aberrations, mouse sperm treated at meiosis or later stages, and the
mouse translocation test. Heddle (1982) also describes dominant lethals in pre- and
                                    257

-------
post-meiotic males, various other stages in the males, and effects on females, as well

as mouse sperm abnormalities, mouse-specific locus mutations.
Table 3-11.   Examples of Somatic in vivo Effects of CP in Marnmals Including Man

ASSAY END-POINT CELL TYPE    DOSE RANGE   RESULTS  NO. STUDIES
                                 (mg/kg)
MICE M




C
S
RAT M

C
D
HAMSTER M

C
S
HUMANS M

C
S
G

* Environmental
M Mteronuclei
C Chromosomal
ERYT
EPYT
ERYT
ERYT
ERYT


ERYT
ERYT

LYS
ERYT
LYS
LYS
LYS

LYS
LYS
LYS
LYS

exposure

aberrations
(BM)
(COLON)
(PB)
(UV)

-------
Table 3-12.  Activity of Cydophosphamide Detected In Assay Systems"
        Test system
                            Activation
Result6
Comments
  g. coll. forward mutation

  g. coH, fm, ip HMA
  (mouse, SC)

  g. cofipolA*/
  P01A"
     Disc
     Liquid

  B. subtllte.
  RecvRec"
  S tvphlmurium
    bm

     IP HMA (1000
                                            1.4 rev/nmote

                                             870 mg/kg
                                             200 mg/kg
                                             200 mg/kg
                                               870 mg/kg
                                               870 mg/kg
                                             250jig/ml



                                            10,000 jig/ml
                                                              Base pair
                                                              substitutions
                                                              Under reexamlnatlon
               TA1535 base-pair
               substitution
               G46
               TA1950
               Dose to mouse;
               TA1535

               Strain a21
               Strain a13
                                                                Activation required
               Active In 03,05 but
               not 04
               Increasing activity as
               the compound aged
               In buffer at 30-C
    Gene conversion
    IP HMA (mouse,
    PO)
    Gene conversion
    testesHMA
    (mouse, SC)
    Gene conversion
    In vitro
    urine, rat, PO
    Gene conversion
    In vitro
    urine, human, IV
                                             500 mg/kg


                                             380 mg/kg




                                             340 mg/kg


                                             3.1 mg/kg
               Renal excretion
               blocked

-------
Table 3-12.  Activity of Cydophosphamide Detected in Individual Assay Systems'
             (continued)
         Test system
Activation6
    Result0
Comments
  V. faba. chromosomal
  chromosal aberrations

  D. mejanogaster,
  sin
  Mouse (ymphoma TK*
  Human lymphocytes,
  In vitro wtth rat
  plasma, IV
  Heta cells, c.
  aberr.
  Human lymphocyte,
  SCE
  Human lymphocyte,
  SCE
  Mouse spermato-
  gonla,SCE
                    34pg/ml
                                                 Adult feeding
  m/s/pg/ml/hr


500-1000 mg/kg




84 /ug/ml, 48 hr



  500pg/ml



3.7 at



 5x20 mg/kg
                    Renal excretion
                    blocked
                                                                    Postsper-
                                                                    matogonial
                                     Actual data not
                                     available

                                     Diluted 10-SOx
  Rabbit bone marrow, SCE
  Mouse bone marrow, aberr.
 Mouse bone marrow
 Mouse bone marrow,
 mteronudel
                  2.4/cdl at 20
                  mg/kg, 24 hr

                   5 mg/kg IP
                   5 mg/kg IP
                  200 mg/kg,
                     96 hr
                    Curve probably
                    bending up

                    Curve approxi-
                    mately linear
                                                                    Position at
                                                                    earlier times
                                       260

-------
lame a-i£. Mcnvny or Vsyoopnospnanwae uoieciea m aiamauai *vroay oymmiiB
(continued)
Twt system Activation6
Chinese hamster, bone
marrow aberrations +
Chines hamster bone
marrow
Human bone marrow
+
Human leukocytes, In vivo,
aberr.

Human leukocytes
+
Human leukocytes, SCE
+

+
Mouse spot test
Mouse oocytes, +
aneuptoioy at Meta II
Result6
20 mg/kg.
10x64 mg/kg
daily
40 mg/kg IV
24 hr
3-5 mg/day
4-6 weeks

500mo-44g
total dose
3xat100
mg/day x 13

30x at S mg/kg
74% at 145
mg/kg IP
Comments
Complex dose
response curve
72 hr after
last treatment,
positive results
atChr

-0.065 mg/kg/
day « -2 mg/kg
total 74 hr
culture
72 hr culture
No Increase prior to
13
days; decrease
to Dover 2
wvvKS

Spontaneous 14%;
12 hr post treatment
 Mouse spermato-
 gonla/spermatocytes
 aberrations
Chinese hamster
spermatogonia,
aberrations

Mouse spermato-
gonJa/spermatocytes
treated as
stem cells
 20% abnormal
ceHs at 97 mg/kg.
      oral
  2 x 96 mg/kg,
      oral
  0/400at200
    mg/kg IP
Metaphase: 1, 13
days post-
treatment; 3
transkxaUons
m 80 cells
treatments
                                        261

-------
 Table 3-12. Activity of cydophosphamide detected in Individual assay systems*
               (continued)
          Test system
Activation6
Result6
Comments
    UDS mouse sperm,
    treated at meiosis
    or later stages

    Mouse transtocatlon
    test
      Sperm

      Spermatlds

      Spermatocytes

      Spermatogonial
      stem cells

    Dominant lethals, mouse
      Postmelotlc
      males
      Various stages, males
   Various stages, males
      Premelotlc males
      Premetotic males
      Premelotlc females
      Premelotlc females
   Mouse, sperm
   abnormality
   Mouse, specific
   locus spermatogonla
                   3-6xat200
                    mg/kg IP
                   5/35 AT 350
                    mg/kg IP
                   34/84 at 350
                    mg/kg IP
                   Z/287at350
                    mg/kg IP
                  1/1350 at 350
                    mg/kg IP

                  40 mg/kg, IP

                 80 mg/kg, oral
                0.005% In drinking
                water for 4 weeks

                    0.01% In
                  drinking water
                   for 20 days

                  210 mg/kg, IP
                  210 mg/kg, IP
                  210 mg/kg, IP
                 200 mg/kg, oral

                  2x at 5 x 130
                  mg/kg daily IP

                3/(3642x7) at 350
                     mg/kg
'Abbreviations:  fm = assay for forward mutations; bm = assay for back mutations; co = assay for
crossing over; c = chromosome; aberr. = chromosomal aberrations; uds = unscheduled DMA synthesis;
hma « host-mediated assay; slri = sex-linked recessive tathals.

''Shows the use of In vitro activation by +.

cResult column shows the lowest positive result obtained or the  highest dose tested and not found
positive.  In some cases many other studies have been carried out at other doses In the same test
system.
                                          262

-------
       A review of cyclophosphamide data has not only been performed by Heddle
 but by Anderson and Sram (1982) (Table 3-13) in the same book which was edited
 by Drs. Fred de Serres and  Mike Shelby. All the dominant lethal data were examined
 and doses were stated at which effects were produced.  Dr. Sram has also done a
 great deal of work with dominant lethals, and he has also used fractionated doses
 (Figure 3-23).
       It is possible to obtain very good dose response curves with different doses
 of cyclophosphamide. One of the aims in our laboratory was to do so and various
 other  laboratories have also done so.  We have examined acute dose-response
 relationships and others have examined the effect of dose-fractionation  on  the
 frequency of dominant  lethals,  pregnancy index, and  mating weeks.   The most
 positive responses for cyclophosphamide are obtained in post-meiotic weeks 1 and
 2, and sometimes in 3. Sotomayer and Gumming (1975), however, found a random
 distribution of translocations in mouse germ cells.
       Thus just about every conceivable type of endpoint has been investigated for
 cyclophosphamide and with this vast array of data, it should be possible to carry out
 some risk assessments.
       Dr. Udo  Ehling has reviewed different germ cell stages and  examined the
 doubling dose  of cyclophosphamide  in mg/kg  bodyweight for different germ  cell
 endpoints in mice (specific locus mutations, heritable translocations,  and dominant
 lethals) and has determined different doubling doses for the post-spermatogonia and
the spermatogonia.   He has compared  them  with doubling doses for somatic
mutations and sperm abnormalities. The doubling dose was based on the regression
                                    263

-------
Table 3-13.
                        Pn>             Poat-
                     npiantation       implantation        Dominant
                      loe»eec           loMa«c          lethality0        Comment*
       150              -               +               +           Firat 10 day* after doalng

       150

       100              -                +                +           Pro-aatrua

       100              +               +                +

       50               +               +                +           Poatmeiotic

       100              +               +                +

       50               o               o                o           PoatfMiotic fertility reduced
                                                                      atalldoaaa

       100              o                o                o

    2.5 (x25)            ooo

     5 ^(25)             o                o                o

    10 (x25)             ooo

       3.5

       6

       9

       4.4               +                +                +

       60                +                +                +

      210               +                +                +

       «0                +                +                +           Poatmaiotic

      210               +                +                +           (Pramatotlcloaaaaat210
                                                                      mg/kg); 420 mg/kg group
                                                                      did notaurvlve

      420               ooo
                                             264

-------
A
0.4-
0.3-
0.2-
0.1 -
^.:Nx:rAv.-
* --^V_l^
246
B
0.8-
0.6-
•* 0.2-
\f
• M
*
•
\ .*

1 1 I 1 1
8 2468 week
.7. Cyclophosphamide: effect of the dose fractionation on the
frequency of dominant lethals induce mice. (A) Dominant lethals; (B)
pregnancy index. ( — ) 100 mg/kg b.w.; ( 	 ) 20 x 2.5 mg/kg
( — ) 20 x 5 mg/kg.
Figure 3-23
   265

-------
 coefficient; the other values are on-point estimates (Ehling, 1981) (Table 3-14).  Dr I.D.


 Adler has also examined spermatogenesis and Dr. A. Wyrobeck investigated sperm


 abnormalities, and he found increases. In contrast, we didn't find sperm abnormalities


 with cyclophosphamide, so we must have been using the wrong dose or showing


 strain differences.



 Table 3-14.  Doubling Dose of Cyclophosphamide in mg/kg for Different Genetic
             Endpointsin Mice

   GERM CELL        SPECIFIC       HERITABLE      DOMINANT     SOMATIC    SPERM
   STAGE            LUCUS         TRANS-     .    LETHALS     MUTATIO    ABNOR-
 	MUTATIONS	LOCATIONS	NS	MAUTY

   POST                              (a)                       NT        (b)
   SPERMATOQ          4              6            88+                   130
   ONIA

   SPERMA-            320            -             -          NT
   TOQONIA	

       + Doubling do** la baaad on tha ragraaalon eoaffioient; tha other valuea are on point eatimatea.
       (a) I.-D. Adler, Terat,Carc.Mutag. 1 (1980,7546.
       (b) A.J. Wyrobek end W.R. Bruc«7 Chemical Mutag. 5 (1978,257-258.


       Dr. Ehling showed doubling doses for specific locus mutations ranging from


 4 to 320 mg/kg.  These are the actual values he produced in his paper, but he has


 done a more recent paper in 1988 with Dr. Angelica  Neuhauser-Klaus (Ehling and


 Neuhauser-Klaus, 1988), and  he has got even more values now for specific locus


 mutations, but that is what he did at that time.


      Now I am moving on to risk estimates made by Meddle (1982). There follows


 an outline of his approach.


      The specific locus mutation rates observed  in the mouse after a dose of 350


 mg/kg body weight is 1.2 x 10** mutations per locus per gamete. Allowing 1 mg/kg


body weight  to be one unit of dose (p), this  rate can be expressed as 3.4  x 10'7


mutations per locus per gamete per mg/kg  body weight (or \i).  Since the average
                                     266

-------
 person is about 70 kg, one unit of dose equals 70 mg per person, that is, 70 mg
 equals one person per mg/kg body weight.  Using the annual production figure of
 about 1000 kg (109 mg), is equivalent to 109 mg divided by 70 mg per person/)*
 equals 1.4 x 107 person p in post-spermatogonia.
       If there is a linear response curve, the distribution of these units among people
 will  not influence the number of mutations produced, and  using the mouse rate
 produces 3.4 x 10~7 mutations gamete 11 times 1.4 x 107 person \i which equals 5
 mutations per locus per gamete times per person.  Allowing a reproductive rate of
 nominal gametes per person, there will be six nominal gametes per mutation per locus
 produced.
       Assuming that these mutations will arise in a population of about 500 millions,
 that is approximately that of Western Europe and North America, (appropriate for this
 meeting, because it is EC and U.S. EPA meeting), the spontaneous number  is 500 x
 10* persons times nominal gametes times 7.5 x 106 mutations per locus per gamete,
 and that is equivalent to 3.8 x 103 nominal mutilations per locus.
       Taking into account the annual  production of  cyclophosphamide this is
 equivalent to 0.13% of a spontaneous rate. If this rate of exposure continues for 30
 years, and this is an estimate of the average reproductive lifespan, (it is recognized
 that men generally reproduce for a longer time span than women), this would be 30
 times 0.13%, and that value  equals 4% of the spontaneous rate.
      This 4% might well be an upper limit because not all of the cyclophosphamide
 is used for human exposure and many of the exposed persons may be beyond the
reproductive age or may not have children because of their disease status or for other
reasons.  Also the population at risk  may be smaller now that manufacturers  are
                                    267

-------
 recommending its use only in life-threatening situations, and physicians are warned
 that a genetic risk may exist. This is happening of late, and the estimate depends on
 the similarity between mouse and  man in the induced and spontaneous mutation
 spectra.
       Thus the doubling dose as outlined by Ehling has been considered as well as
 the population  approach of Neddie. Now I wish to consider exposures of the male
 which produce abnormalities as varied as those produced by females exposed in
 utero. These are for example runts, which are 75% or less of the body weight of the
 normal size fetus, fetuses with gastroschisis where the intestine hangs outside the
 body, exencephaly where the brain comes outside of the head, and anasarca which
 is a total bloating of the fetus.  There are others also. All are thought to be truly
 genetic events, since they are transmitted through the exposed male.
       We examined the incidence of grossly abnormal fetuses that we found after
 treating males with chronic low doses of cyclophosphamide given at the therapeutic
 dose rate of 5 mg/kg for 30 weeks. Mating took place at weekly intervals for a long
 time over this period. There was a saline control and ally! alcohol treatment also.
 Ally! alcohol is involved in the metabolism of cyclophosphamide and was of interest
 to the U.K. Ministry of Agriculture, Fisheries, and Food who funded the project.
      We found that we got large  numbers, up to 10%, of the runted and grossly
 abnormal fetuses after cyclophosphamide treatment, but not with allyl alcohol and
 saline treatment, after about week 7; and this response plateauxed up to week 30.
      We also  examined alizarin stained fetuses for skeletal abnormalities with the
following gross abnormalities: anasarca,  exencephaly, hydrocephaly, other skull
                                    268

-------
 defects, spleen abnormalities, gastroschisis, abnormal placenta, and growth-retarded
 fetuses.
       There were large numbers of genetically transmitted congenital malformations.
 These can be closely correlated with genetic defects in humans. Man would not want
 his children to have such defects, and similar abnormalities can be identified in human
 offspring.  This makes the comparison between malformations in animals and man
 very meaningful and tangible.
       Thus, this may well be a very good model for using in genetic risk assessment,
 but it  is  also  a  good  model  for  chronic  low dose exposure.   We used
 cyclophosphamide and butadiene in this way.
       MuMhill did some work with exposed populations, that is, patients who had
 been on therapeutic doses of cyclophosphamide, and he found that when they mated
 after these doses, there were no abnormal  children.  However, he believed the
 numbers examined may not have been sufficiently large to detect if there truly was an
 effect, or it might simply be that the patients did not mate immediately after treatment.
 If they  had  waited, and it is extremely likely that they did,  by  that time the
 spermatogonia which were not affected would have formed sperm.
      So, for a chemical that just affects the post-meiotic stages, provided a sufficient
 length of time, elapses after treatment then there will not necessarily be a risk to the
 germ cells.
      However, CP is a mutagen which produces  point mutations, cytogenetic
damage, and DMA adducts in several organisms. In mice, the active form reaches the
gonads in both males and females, and it  also does so in rats. Thus similar effects
might be expected in exposed people.  It  is known that active metabolites are

-------
 excreted in  the  urine and chromosomal aberrations are present in peripheral
 lymphocytes in blood as well in exposed people.
       According to Heddle, CP is likely to produce less than 4% of the spontaneous
 mutation rate, but Dr. Ehling may  have some thoughts on that. There is a significant
 risk of a chromosomally abnormal child if conception occurs within a short time of the
 end of treatment, and that is a very important point to bear in mind.
       With regard to Professor Frits Sobels' parallelogram, there are positives for
 mouse somatic  cells, including mutations  and adducts.  There are also positive
 responses in mouse germ cells and positive human somatic cell responses, which are
 mutations; but there are no positives for human germ cell data.  It is  not known,
 however, whether MuMhill's data  represent a true response.
       There are thus many areas of the parallelogram almost complete, so we should
 be able to make some predictions, perhaps in our workshop deliberations this week.
 Discussion '

       DR. SHELBY:  Dr. Ehling, would you like to consider this presentation before
 we take questions?
       DR. EHUNG: First, you were asking about the 4% where I have some doubts,
 because it would end up as 1/1000th of the spontaneous mutation rate.  If you make
 such a calculation, then John Heddle had a 'hit'.
      The other point is, of course, the MuMhill study.  The numbers you cited were
from the Proceedings of the International EMS Meeting in Japan. It just indicates a
big difference between post-spermatogonia stages and  other germ cell stages.
                                    270

-------
 Mulvihill published  a large study from  offspring from  parents  treated  with
 cyclophosphamide without considering the date of conception.
       I think in doing such a thing, we can get a false sense of security, because, of
 course, you debated that cyclophosphamide may not be mutagenic in man, because
 you mentioned the human data from Mulvihill; but his study, I want to emphasize once
 more, was without consideration of the conception date and in fact cyclophosphamide
 may well be a human germ cell mutagen.
       To make the story a little bit more complicated, I would like to present a figure.
       Here you see radiation and the frequency of cyclophosphamide is the same
 as the control.
       You see here treatment due  to X-irradiation. That is with 3 rads and with 6
 rads. ft is a very conventional mutation rate. The same was observed in Oak Ridge
 and by Hemminki in Harvard.
       But if we pre-treat for 24 hours the animals with cyclophosphamide, then we
 double  the mutation rate in these animals. The types of mutations observed are
 exactly  the types of mutations that will be observed in the radiation experiment.
      Very likely, what cyclophosphamide does is to block the repair mechanism, so
 enabling a doubling of the mutation rate.  So, it is not that simple that we do not
 observe mutations in spermatogonia, but in such a situation, an actual mutation rate
 with  cyclophosphamide can be observed in spermatogonia.
      DR BISHOP:   Dr. Ehling, is that only in  spermatogonia, or  is that post
spermatogonia as well?
                                    271

-------
       DR. EHUNG: We were not so much concerned about post-spermatogonia in
 this experiment, because we wanted to see the effect in spermatogonia, so I cannot
 answer that question.
       DR. SHELBY:  Dr. Sorsa?
       DR. SORSA: The exposure assessment is very important, and I think that the
 idea of having 4% of the spontaneous  mutation rate related to cyclophosphamide is
 not a sound one, because not everybody is exposed to cyclophosphamide.
       We have some measurements, like many other people, on the potential work
 exposure of nurses when they are diluting  anti-cancer agents,  and, of course,
 cyclophosphamide still is in most countries, as the most often used anti-cancer agent.
 If you dont have any protection, it is about 0.5 mg/m2 which are the measurements
 of levels to which humans are exposed.
       However, there are certain  occupations involved in the manufacturing of the
 drug, cyclophosphamide, where the exposure levels might be as high as 800 mg. So,
 there is a huge difference in exposure  due to the manufacturing  process, up to 400
 mg. If no protection is used, this is a severe exposure.
       So, one might calculate that a nurse without protection working for 2 hours per
 day gets about 1 mg of cyclophosphamide 5 days  a week and 5 mg, 40 weeks per
 year, that is 0.2 mg. In  10 years' time  working in this same situation, this equals, 2
 mg.
      Of course, this amount is nothing by comparison with the amount which the
patients are receiving. The patients  are generally receiving about 4000 mg.
      We should not be talking about cancer rates; but in this case, it might be easy,
because we do have patient studies where cyclophosphamide therapy has increased

                                   272

-------
 the risk of secondary cancers, namely, secondary acute non-leukocytic leukemia, and
 the risk is about 1% in two of the published studies.
       If the dose has been about 4000 mg, we can extrapolate that in this situation
 of the nurses working, in 20 years' time, there would be 1 acute non-leukocytic
 leukemia in 100,000 exposures.  So, the risk is minimal.
       But if we would just use these groups, of course, the risk clearly becomes, very
 soon a much higher one, and I think that maybe this kind of calculation could be
 performed also for the general population risk.
       DR ANDERSON: I have a comment. This relates to the mechanisms involved.
 There would appear to be a lestes barrier" that protects the germ cells.  For this or
 some other reason, germ cells are relatively protected, but we have already seen an
 example of a chemical today such as butadiene that reaches the germ cells at a very
 low level.
       If a chemical affects spermatogonia, it could be a lasting problem and Dr. Vicky
 Dellarco was earlier highlighting this. If a chemical only affects post-meiotic cells, that
 is sperm and spermatids, it might be of less concern because the cells will cycle and
 there will be some non-affected stages. If man waits until the non-sensitive cells are
 cycled in order to mate, there may not be a problem at all.
       The area where I believe there is a real cause for concern, is from chronic low
 dose exposure in the workplace or from low level environmental exposure, because
 man is continually exposed and mating. Since such exposure occurs all the time, if
 a chemical only ever affects the post-meiotic cells, it will matter because sensitive
 stages are continually cycling and being affected. Thus mating will always take place
with damaged sensitive cells.

                                     273

-------
      Many of the arguments concerning  risk to man are based on acute high
exposures (for example, as occurred after the bomb at Hiroshima and Nagasaki).
Such  extrapolations do not take into  account the chronic low  dose exposures
encountered in real life situations. It is like comparing apples and oranges instead
of apples and apples and I think this is a real issue we ought to be concerned about.
                                  274

-------
Rapporteur's  Reports
This section contains the rapporteur's reports for the following compounds:

      •     Ethylene Oxide
      •     Acrylamide
      •     1,3-Butadiene
      •     Cyclophosphamide
                               275

-------
276

-------
 Ethylene Oxide (EO)

 Dr. Julian Preston was the rapporteur for the group discussing Ethylene Oxide.  Dr.
 Preston is from CUT is Research Triangle Park, North Carolina, U.S.A.

 Initial Report on Ethylene Oxide
        DR. PRESTON:  What I have are five figures,  that represent  some of our
 conversations and discussions this morning. The first  thing that we discussed was
 the heritable effects in germ cells and the heritable translocation risk estimate; so, we
 started with germ cells and what we wanted to discuss  was the applicability of using
 the particular data set used by EPA in their genetic risk assessment for ethylene oxide
 and seeing what additional data we would need to make such a risk estimate as that
 presented yesterday appropriate.  We decided there were two factors that  were of
 importance,  that led us  to  a specific conclusion  on the  utility of  the heritable
 translocation assay.  One was Lars Ehrenberg's brief calculation  of dose; we talked
 a lot yesterday about what one could use as dose along the abscissa. What he
 pointed out was that what was appropriate, was the dose to the DMA; that  is, what
 was the target dose. What I plotted here (Figure 4-1) was the reciprocal translocation
 frequency; first you will see data, I plotted  as ppm, from the original Waldy Generoso
 study and then an approximation of the recalculation of the doses to DNA, which will
 shift all the points over to the right What that will do is, in fact, flatten  out the curve
 at the tow exposure levels, so  that will reduce the effect, also, for reciprocal
translocations as a function of dose to the DNA; that will further reduce  the risk
estimate from the linear  extrapolation,  on the ppm curve, that was discussed
                                     277

-------
  0)

  o
  (0
  g
  o.
 '8
 DC
          x  ppm (Qeneroso data)
          o  Dose to DMA (estimate)
Reciprocal Translocations Induced in Mouse Germ Cells
      by EO (Data from Generoso et al., 1990,
  Environmental Molecular Mutagenicrty 16,126-131


                 Figure 4-1
                    278

-------
 yesterday. Also we discussed in some detail, at the beginning, as to whether to use
 an additive assessment of slope; that is, do the spontaneous lesions interact with the
 ethylene oxide induced ones or are those independent events.  We concluded that
 because of  the very low  probability of having a spontaneous  event, we  would
 consider those as additive and not interactive and you will appreciate that that will
 reduce a linear response, for either of these, to something that is convex downward.
 Thus, it will reduce again,  the effect at low exposure levels.  Our conclusion from
 these two points and others we discussed  both yesterday and this morning, but
 particularly based on these two, is: genetic risk from reciprocal translocations at low
 exposures and for post-meiotic cells, for which  this data set  was collected, will be
 extremely low. We indicated that further experiments really would not change that
 situation. We could not see that collecting more data was really warranted to improve
 the genetic risk estimate, for post-meiotic cells and for reciprocal translocations.
 There might  be  reasons for  considering other germ cell stages, for additional
 information.  This does refer to reciprocal translocations also.  I  will move on to
 mutations in a minute.
       DR GENEROSO: Do you mean independent, rather than additive?
       DR. PRESTON: I meant the lesions are additive, the responses are simply an
 additive response. The spontaneous will be present at all of the dose levels;  it will
 be an equal contribution. I do not consider the two responses as interacting.  I was
thinking in terms of mechanisms, and independent and additive are basically the
same.  To return to our discussion session. That was point number one. Then we
had a presentation from Waldy. His presentation related to the work that he has both
published and has in unpublished form, on effects on the zygote, where he has

                                    279

-------
 shown that there is a very sensitive stage, somewhere around from one hour to six
 hours after mating, that is sensitive to the production of congenital abnormalities with
 ethylene  oxide.   He  had  additional  information on  retinoic  acid  for other
 developmental stages, but what we have concentrated on here are the discussions
 on effects on the zygote. First of all, we want to decide what sort of effect this was.
 Well, we agreed it was not a genetic effect, according to what was being defined
 yesterday.  It is  obviously not a component of genetic risk because it is not
 transmitted from an individual to the offspring.  Of course it is transmissible from cell
 from cell, because you get an effect in the zygote and a response in the embryo; so
 it is clear you have transmissible effect, but it did not fall under the category of genetic
 risk.  Waldy discussed the possibility that it was a so-called epigenetic effect; that is
 a word that I feel covers a multitude of sins, if you do not know quite what it is.  We
 could not  decide what the mechanism was, which suggests some confusion in the
 end.  We do have a transmissible effect, and because it is such a clear response and
 there is certainly a sensitive period of response, I think we felt that there was need for
 separate risk assessment consideration here.  A completely separate one from the
 genetic risk, this is zygotic risk, maybe exhibits a broader period of action for other
 chemicals  that covers  a several day period.  We do think there is a good case for
 making a separate risk assessment. Waldy in particular noted that there was a higher
 sensitivity here, with the female pro-nucleus in the zygote, considerably higher than
the male, which was essentially insensitive.  So here is a female effect, that I think
bears further consideration; it is quite a clear response and you appreciate that in the
pronucleus stage the two nuclei, the original sperm and egg,  are separate until the
first cleavage division,  so you can in fact look at effects  independently on the two

                                     280

-------
 nuclei. There are a number of reasons why that might be the case and we had only
 touched upon those; but our conclusion here was that it was a very important area
 of research and that further study, in particular, mechanistic information was needed
 and indicated from these original, rather exciting studies that Waldy described,  hie
 did have a little  bit of new data, which I probably will be able to present tomorrow; I
 did not put it on an overhead today.  He attempted to produce a dose-response
 curve, but what Waldy pointed out was that he used the six hours after fertilization
 period for treatment, with a one and a half hour treatment regime; that certainly was
 not ideal. This was not the most sensitive stage, and so he got a no response at 600
 ppm, whereas that probably would not be true at a more sensitive stage, because the
 six hour treatment, (that is, the six hour treatment start time), with an hour and a half
 treatment, takes you into a rather insensitive period of seven hours after fertilization;
 so he would like to move the treatment time back towards fertilization. But I think
 mechanistic information is certainly needed here, in order to help to produce a risk
 estimate.
       Then we did a series of dubious calculations on a data set; I will present them
 as we did them. Now, we have moved to consider mutational data, having covered
 reciprocal translocations and then the zygotic effect and the effect on the conceptus
 back into germ cells. There were no new data on mutations in germ cells that Waldy
 could indicate, but we returned to some data that was published a few years ago by
 Susan Lewis and her colleagues, on the mouse. These were for electrophoric variants
 and also dominant visible mutations.  What Susan did in these experiments was to
 look for some 20 defined phenotypes (assumed to be 50 genetic loci), for dominant
visible mutations (Figure 4-2). For these 20 phenotypes, she found four  dominant

                                    281

-------
  Dominant visible mutation data for the mouse following ethytene oxide exposure
  (Source: Lewis et al., 1986, Environ. Mutagen. 8, 867-872)
  •      4 dominant visible mutations (including 2 heritable translocations)
         Number of offspring 1891 (assume 2000)
         Average exposure for mutation induction 100,000 ppm h
  •      Visible mutations for 20 phenotypes that are assumed to be
         representative of 50 genetic loci
  •      The overall mutation frequency calculates as
                            2 x 10"8 mutations/ppm h
                            i.e., 4 x 10~10 mutations/locus/ppm h
                                 Figure 4-2
 visible mutations in a study were there was a long, several week exposure.  In fact,
 the animals were not mated until seven weeks into the exposure regime, thus, the
 exposures covered the whole speratogenetic cycle.  Of these four dominants, two
 were heritable translocations and I think this is in approximately 2,000 offspring. I
 think it was between 1,800 and 1,900.
       DR LEWIS:  1,891.
       DR PRESTON:  That is 4 mutations in  1,891  offspring, and the average
 exposure, at the time of mating, for these four mutations, was 100,000 ppm hours; so
 these mutations showed up at different times after the start of exposure. We just
 simply took an average exposure time for the four mutations; in fact, the first one
 showed up around 60,000 ppm hours. So there were 20 phenotypes (50 loci) scored
 and we  did a calculation that assumed that there are 2,000 similar dominant loci in
the human, which might be an over estimate  for dominant visibles, but it  is a
reasonable estimate of dominant mutations overall, if you just do a simple calculation,
                                    282

-------
you would find you have 160 dominant visible mutations in humans in a study similar
to the mouse one, assuming all sensitivities are equal. That would be 160 mutations
x 10"5 per ppm hours. That would work out at one mutation per 1,600 individuals per
year of exposure, based upon a one ppm exposure (Figure 4-3).  Now, there is a lot
of flop in that calculation; I mean, do you use all four dominant visible mutations,
when two of them were heritable translocations?   We assumed that  amongst the
offspring in humans, there would be similar numbers of heritable translocations.  We
just wanted to see what the situation would be if you simply take mouse mutational
data and convert it into human data.  That is the sort of approach one might take,
without putting an enormous amount of reliance upon the end product; I think it is a
 Calculation of dominant mutation rate in humans following ethylene oxide
 exposure

 •      Mouse data (Figure 4-2) give 2 x 10* mutations/ppm h
 •      Assume 2000 dominant loci in humans
        Equivalent sensitivity
 •      Mutation frequency is
                           160 x KT5 mutations/ppm h
                           Le., 1 mutation/1600 persons/ppm h
                           (based upon a 1 ppm exposure)

                                Figure 4-3
                                   283

-------
 ball park type of number. We progressed from that germ cell calculation on the only
 data that we found for which we could do any manipulative calculation, and moved
 into somatic cell effects, to try and link sensitivity of mouse with man, by another
 approach.  Here, we just took a direct approach, mutation for mutation.  What we
 shown for a 20 week expression time and the reason that this was selected although
 there were two expression times, an eight week and a 20 week, was that this would
 be more appropriate for a human exposure, which is presumed to be continuous. So
 have tried to do now is see what the relationships in sensitivity between the mouse
 and  human  might be, for mice, when you look at  the same endpoints in both
 organisms.  Somatic cells, HPRT mutations are in splenocytes. I have presented just
 part of the data set from Skopek and Walker because it is one that you might be able
 to use to compare with the human. The data are presented in Figure 4-4. This is we
 took the longer  expression time.   That  is the  time you  sample the cells, the
 splenocytes for mutation analysis after exposure.  These exposures are in milligrams
 per kilogram and they were from more than one IP injection; for this particular set, I
 assume this is like 4 times 150 (600 mg/kg) and other triple combinations,  double
 combinations. They were for more than one day for the higher exposures.  Here the
 mutation frequency is times 1CT6 (Figure 4-4).  The control is 3.6, and up to 15.5 at
 600, it is approximately linear.  So we did a doubling dose  calculation.  This is
 approximately 200 milligrams per kilogram, which converts to, with Lars Erhenberg's
 help, to 4,400 ppm hours, which is about 110 ppm hour per week. That is a doubling
 dose, based upon the mutation frequency in the mouse. There are some human data
 available of the same sort, Ad Tales' study (Tales et al., unpublished). We considered
the highest exposure group; there were two exposure groups in this particular study.

                                    284

-------
        HPRT mutants in mouse spienocytes
        (Source:  Walker, V.E., and Skopek, T.R., 1993, Mutat Res. 288,161-162)
        20 week expression time
        •     Doubling dose - 200 mg/kg
        •     This converts to 4400 ppm h or 110 ppm h/week
        •     Using a doubling dose approach, the mutant frequency is
                           4.0 mutants/4400 ppm/h
                           i.e., 1x10"° mutations/locus/ppmh

        HPRT mutants in human lymphocytes
        (Source:  Tales et al., unpublished)
        •     Control -    8.6 x 10"6 mutations/locus
        •     Exposed  -    13.8 x 10"6 mutations/locus
                           (internal dose 5 ppm h)
        •     Using a doubling dose approach (doubting dose about 200 ppm
              h/week)
              Mutation frequency is
                          9.0 x 10* mutations/200 ppm h/week
                          i.e., 1.1 x 10"* mutations/locus/ppm h
                                 Figure 4-4
We took the control value for the humans, 8.6 x 10"8.  Now, these are mutations, the
mouse ones are mutants.  There is an important distinction with the spienocytes
because of donal expansion of induced mutants in progenitor cells. However, for the
mouse, there was a short exposure, where clonal expansion would be less likely to
be a problem, so we actually got mutants, but their frequency would  be very similar
                                   285

-------
 to mutations.  For the human, because of the long exposures, mutants or mutations
 require some separation - the value used was for mutations and is 13.8 x 106.  They
 were sequenced.  The internal dose is estimated from the hemoglobin adducts as
 5 ppm. If you simply do a doubling dose calculation where here the exposed group
 were approximately double the control group. I say approximately because we came
 up with a doubling dose of approximately 200 ppm hours per week, which is rather
 similar to that for the mouse data.  Now, again there is some flop in there, but I  think
 that is the sort of approach  that would be appropriate and we get reasonably similar
 sensitivities for these two data sets for mouse and human, both on HPRT mutations.
       One more quick point.  We looked at somatic cells, again a comparison
 between what we hope would be mouse and human for chromosome aberrations in
 peripheral  blood lymphocytes, because  there is quite a large human data base.
 When we considered the mouse data, there were none, and so we could not do a
 mouse to human comparison. There are SCE data, but there are no aberration  data.
 For the rat, it is rather interesting, and this presents quite a fascinating story, in that
 for the rat, it is negative for aberrations as shown by Andy Kligerman;  I have single
 data point which was also  negative for chromosome  aberrations.  Natarajan also
 reported that ethylene oxide was negative for mutations In the rat, from their study.
 So, we had this  interesting data set for the  rat, but unfortunately no mouse data to
 compare it with.  Clearly there are gaps in the available information; one of the gaps
 is data for the  mouse, and the other is trying to understand what is going on  in  the
 rat. There was some data for the monkey, which is positive, especially for dicentrics;
 but we felt that was a rather short extrapolation to make and also it would be rather
dead-ended, because the  only data you  will  get for the monkey  are for  the

                                    286

-------
 lymphocytes, since you probably will not get the germ cell data. There are a lot of
 gaps in the aberration data, which was a bit of a surprise, and we could not do any
 of our parallelogram comparisons.  We did have some somatic cell comparisons for
 mutations; we feel that the germ cell data are what is needed in the mouse for
 mutations at different stages in order for us to go around the parallelogram.  I think
 I have approximately represented our morning discussions; if I am not here tomorrow,
 it means I did not accurately represent the  group.
 Final Report on Ethylene Oxide
        DR.  PRESTON:  It is a significant pleasure to  be able to hand  all these
 overheads over to the distinguished Chairman, who will then make sense out of them.
 He has not seen them yet, because I wrote them last night, but I think they are
 representative of what we discussed.  But what I will talk about today is really an
 extension of our discussion yesterday.  I am not going to cover the same material,
 except where it is important, to show where we have gone in the last two of our
 discussions sessions.
       We have tried to address the questions, but hopefully not in a generic way, but
 in a rather Ethylene Oxide specific way. I should spread the responsibility around and
 let you know that Vicki Dellarco, Susan Lewis, Waldy Generoso, Lars Ehrenberg, and
Ad Tates,  Natarajan and myself were the Ethylene Oxide group.
       What I was trying to do yesterday,  on behalf of the group, was to use the
parallelogram approach for estimating potential genetic effects, having defined genetic
effects yesterday as being transmissible effects from germ cells to the offspring, and
not including the zygotic effects.  We decided to apply the parallelogram approach

                                    287

-------
for those pieces of data that were available. So question one:  which are the best
available data?  Clearly, we feel the ones we are using here are the best available
data, that is why we are using them. They are rather limited, but I think they do, at
least in our estimate, allow some initial attempt at predicting genetic risk from ethylene
oxide exposure.
       Now, yesterday, the germ cell data that we discussed was from Susan Lewis,
and we made a mini-adjustment to the data that I presented yesterday.  Following
discussions with  Susan,  although 20  different phenotypes  were assessed  for
predominant visibles,  she  feels that  the actual number of genetic loci that are
responsible for those 20 phenotypes, appreciating  that these are relatively gross
phenotypes, was more like 50.   So,  it is not a one-for-one relationship between
genotype and phenotype.  So, we recalculated using the 50 tod and that, instead of
one mutation per 700 persons per year to an exposure of one ppm we come up with
one per 1,600 (correction already made in Figures 4-5 and 4-7).
                 Dominant Visible Mutation Data lor the Mouse
                      Following Ethylene Oxide Exposure
           (Source: Lewis et al., 1986, Environ. Mutagen. 8,867-672)
        4 dominant visible mutations (including 2 heritable translocattons)
        Number of offspring 1891 (assume 2000)
        Average exposure for mutation induction 100,000 ppm h
                                                   *
        Visible mutations for 20 phenotypes that are assumed to be representative of
        50 genetic loci
        The overall mutation frequency calculates as
                            2x10"* mutations/ppm h
                            i.e., 4 x 10'10 mutations/tocus/ppm h
                                  Figure 4-5
                                     288

-------
        We have moved away from this particular approach for expressing the data in
 order to include it in our parallelogram. I just wanted to point out that we had made
 a slight reassessment. And because 50 loci came into the calculation, I used 2,000
 dominant loci for the humans so that my calculations were easier (Figure 4-6).
                Calculation of Dominant Mutation Rate in Humans
                       Following Ethylene Oxide Exposure
         Mouse data (Figure 2) give 2 x 10* mutations/ppm h
         Assume 2000 dominant loci in humans
         Equivalent sensitivity
         Mutation frequency is
                            160 x 10"5 mutations/ppm h
                            i.e., 1 mutation/1600 persons/ppm h
                            (based upon a 1 ppm exposure)
                                  Figure 4-6

       And again, just to remind you of the somatic cell data that we used; this is
 exactly the same set as yesterday (Figure 4-7).  The mouse data are from, Tom
 Skopek who has shown up this morning, so he gets to see, some of his own data on
 the screen. So, if he has any concerns about the way we used them, he can express
 them.
       This was, as I expressed yesterday, a 20-week expression time, for the set of
 mutations,  and we calculated yesterday a doubling dose.  Again, we have departed
from this approach to make comparison between our various  data sets a little more
                                    289

-------
         HPRT mutants in mouse sptenoctyes
         (Source: Walker, V.E., and Skopek, T.R., 1993, MutaL Res. 288,
         161-162)
         20 week expression time
         •     Doubling dose - 200 mg/kg
         •     This converts to 4400 ppm h or 110 ppm h/week
         •     Using a doubling dose approach, the mutant frequency is
                           4.0 mutants/4400 ppm/h
                           i.e., 1 x 10"8 mutations/locus/ppmh
  B.     HPRT mutants in human lymphocytes
         (Source: Tales et al., unpublished)
         •     Control-    8.6 x 10"6 mutations/locus
         •     Exposed -   13.8 x 10"6 mutations/locus
                           (internal dose 5 ppm h)
         •     Using a doubling dose approach (doubting dose about 200 ppm
               h/week)
               Mutation frequency is
                           9.0 x 10"8 mutations/200 ppm h/week
                           i.e., 1.1 x 10*8 mutations/locus/ppm h
                                 Figure 4-7

sensible for the parallelogram approach. The human data are the same as yesterday
from Ad Tales' data set
       It should also be noted, because that comes up later, that the human data
does have hemoglobin adduct information for an internal dose. This particular mouse
mutation data set does not; these are exposure data from IP injection, but we felt H
                                   290

-------
 that  was  not  unreasonable to equate  the IP injection  data to internal dose
 measurements.
       So, how did we go about doing some sort of parallelogram calculations? I am
 going to start in the top left-hand corner of my parallelogram (Figure 4-8), which is
 with the mouse germ  cell data.  All we have done here is simply to convert our data
 sets, using the various assumptions, and we have done this with all three data sets
 that you will see, into  mutants or mutations per locus per unit of dose. In this case,
 in the mouse, it is mutations per locus, per ppm hour (Figure 4-5). It was just so that
 we had some point of comparison. And it is the number of mutations, per locus, per
 ppm hour. As in the  previous day's presentation, we arrived at  a value of 2 x 10"8
 mutations  per ppm hour.  I can return to that calculation if necessary, but it is based
 on this set of assumptions.  The 4 mutations, in approximately 2,000 offspring, at the
 50 loci, were assumed to arise at an average exposure of 150,000 ppm hour. That
 is just an average, rounded off. This converts to 4 x 10*10 mutants  per locus, per ppm
 hour.  All right, that is  how we used the germ cell data for the mouse.
       DR. BISHOP:  I am sorry, Julian, I missed  that, what germ cell  stage was
 created for the mouse here?
       DR. PRESTON: Because of the nature of exposure, it covers the whole of the
 spermatogenic cycle.  The first matings were done at seven weeks, and I do not know
 the exact dates after that that those animals,  or the mutants arose, but, it was from
 exposure of the whole spermatogenic cycle. So, you cannot define which cell stage
 would be affected, or the germ cell's phase relevance.  That is important, but we felt
that this was a particularly useful data set, because it approximates a human exposure
scenario.
                                    291

-------
         A parallelogram for predicting mutation induction in
         human gene cells following ethylene oxide exposure
           Mouse                                Human
         Germ Cells                ^         Germ Cells
      Mutation Frequency                    Mutation Frequency
      Somatic Cells         ^      ,.          Somatic Cells
(splenocytes/lymphocytes)    ^      ^        (lymphocytes)
   Mutation Frequency                        Mutation Frequency
                           Figure 4-8
                              292

-------
       But remember, also, that there are a number of assumptions.  I am trying to
 minimize those, but I think these are going to have to be made, whatever data set is
 available. So, with the somatic cells for the mouse, and it is based on the doubling
 dose that we had calculated in the previous figure I showed (Figure 4-7A), there are
 about four mutants per 110 ppm hour per week.  This simply converts to 1 x 10*9
 mutants, per locus, per ppm hour, for these mouse somatic cells. These are mutants,
 rather than mutations but we felt that, again, this was a small compounding factor.
       And here (Figure 4-7B)  for the human somatic  cells, again, based on a
 doubling dose  calculation, we come up with an estimated mutation frequency of 1.1
 x 1O"8, mutations per locus,  per ppm hour. I put the 1.1 in just to make it marginally
 different from that for the mouse. It was rather remarkable; rather I think that it is a
 fluke, not necessarily a measure of the absolute sensitivity equivalence between the
 two.
       What we used these data for was not for  absolute numbers, but  rather to
 suggest that in order to extrapolate from mouse to human a number, a sensitivity
 factor  of one  would be  appropriate.  We have to make some  assumptions for
 converting from mouse germ cell to human germ cell. So, we assumed that from the
 somatic cell data, for the HPRT mutation, under these particular treatment regimes,
 and the various assumptions, that we could use a sensitivity factor of one. So, if you
 feed that into a human germ cell, the unknown in our parallelogram box, we assume
 that equivalence of sensitivity, as I just pointed out, with the mouse germ cells, based
 on the somatic cell mutations.
       For the human, we assume 2,000 dominant loci, and of those 2,000 we had
to assume something about the phenotype.  So, we assumed that each one was

                                    293

-------
 giving an independent phenotype.  So that  our  loci represent  2,000  different
 phenotypes. That might well be inaccurate, but that was the best that we could do.
 These loci, identified as dominants, are represented in McKusick, for example, as
 independent gene mutations.
       So from the mouse data that I presented in Figure 4-7, we had a mutation
 frequency of 4 x  10'10 per locus,  per ppm hour.   And so the human would  be
 equivalent, for per locus frequency, for the total number of dominant mutations. You
 have 8 x 10'7 dominant mutations per ppm hour.
       That is just, a "flavor type" of number; it is not to be taken as an extremely
 accurate number, but I think it gives the types of data sets that are available, and what
 you can do with those.  I think that we felt that that was a reasonable approach to
 take.
       Lars Ehrenberg did a quick comparison with radiation numbers, and it would
 indicate that the ethylene oxide number here for dominant mutations might be of the
 order of ten to a 100 times higher than a radiation equivalent mutation frequency on
 a per rad basis.
       As I say, the approach that we took, based upon the data we had available,
 realizing that we had four dominant mutations that we utilized in the mouse study, and
 the HPRT mutation in somatic cells, we  finished up with a  decision that the
 parallelogram approach could be utilized on those data.  And of course  we are
certainly open to discussion on the approach.
      We were trying to see whether there were other methods that would be better
used to make comparisons between somatic cell mutations in humans and mice, but
                                    294

-------
 we were not totally happy. Thus, we used the HPRT mutations but were not sure that
 this was truly representative of somatic mutations as a whole.
       We had a fair amount of discussion on the feasibility and appropriateness of
 using the relationship between tumor incidence as a measure of relative sensitivity on
 the assumption that the initiation of a tumor, at least, is modeling mutation induction
 in somatic cells. So, if you assume that the initiation event for tumors is a mutational
 event, can you use tumor incidence? Specifically, we discussed the use of leukemias
 on the assumption that that same tumor type, the leukemias that you would see in
 humans, say from an ethylene oxide exposure, really  is the same tumor type in a
 mouse. Just because it is called a leukemia does not mean they are exactly the same
 tumor type.  We made the assumption that they might  be, and we chose leukemias
 because the mechanism of formation of leukemias, in terms of number of steps, is a
 lot simpler than many of the solid tumors.
       We discussed this approach and we came up with one that is really a form of
 a direct method, using the rad equivalents as the dose term.  We discussed the
 appropriateness of doing this and, we think it might have some merits.  We did not
 feel, in our discussions, that we really wanted to replace our somatic cell mutations
 with the tumor incidence, at this point,  because I think, in general, as this is our
 consensus, we felt there were maybe more assumptions than we would have to make
 to the somatic cell mutations approach.
       However, we wanted to present this, as it was a fairly lengthy discussion, and
we think it is worth looking at further, because you will see, as an answer to a couple
of the questions that were posed, maybe tumor incidence, or tumor formation might
be an appropriate endpoint to address other issues.

                                   295

-------
       We did discuss the topic of the spontaneous abortions in hospital workers
 raised by Dr. Sorsa yesterday.  The only data that were available to us, in fact, were
 the 1982 publication of Hemmlnki, we did not have the other data that Marja Sorsa
 presented yesterday, or  rather mentioned yesterday.   So,  we  discussed  the
 appropriateness  of  that particular  data set  on spontaneous abortion  for  risk
 assessment. We all agreed that it is clearly an important component of genetic risk;
 but as to its utility in genetic risk assessment,  at this point, we could not establish
 what component of the endpoint was genetic, what component might be zygotic, a.
 la Generoso discussion yesterday, and what component is what I described as a toxic
 response, a cell killing response.
       So, we felt that there was need for further information to be able to incorporate
 such data into one component of our risk assessment.  And  if it is not a genetic
 component, then you still need the model, as we suggested, for the zygotic data, as
 a separate assessment of risk, and take  It out of the genetic risk department.  We
 would like to establish these factors a little more clearly and maybe utilize other data
 sets that might be available.
       Then we entered into a discussion around the questions that were posed by
 the organizing group. Without going through all of them, since we clearly addressed
 a number of them, by utilizing the parallelogram approach, by selecting a particular
 data set, by pointing out assumptions:  What additional data do we need? Well, we
want to remove the assumptions. What data do we need? We need more mutation
data.  What data do we need for somatic cells?  We need additional mutational
endpoints.  So, we obviously answered some just by the assumptions that we had to
make.
                                    296

-------
       However, we obtained some  additional  information  that  relates to the
 questions.  First, what was needed to enhance dosimetry data, and what were the
 major uncertainties?  And with ethylene oxide, there is dearly a need for better
 exposure data; there is not only a need for dose data, but exposure data needs to be
 much better defined. That is a very important component to the sort of extrapolations
 we were doing, because we were basing our frequencies of mutational endpoints on
 ppm hours.
       Also there was a need for establishing a relationship between DNA dose,
 which one could equate with adducts, maybe specific adducts, and a response in the
 target cells. We have very little information on the germ cell stages. This is not only
 unique to the dosimetry information  because we have a lot of gaps  in biological
 endpoints in the targets cells.  But we did identify those as areas that clearly would
 help in the genetic risk assessment process.
       A second question relates to the assumptions that are made for genetic risk
 assessment approaches.  Again, I used a number of those in our calculations, but
 these were sort of the highlight ones. A major one is the equivalence of somatic to
 germ cell for sensitivity.  I pointed out that we decided that the mouse and the human
 had equivalent sensitivity to HPRT mutation induction. In somatic ceils we simply took
 that as an equivalence in germ cells, while we do not really know whether that is true
 at all. We had to use some approach, and there are no data to impact negatively
 upon the utility of equal sensitivity.
      Are the mutations assayed representative of the genome? Well, the mutations
in the somatic cells were just HPRT mutations. We have a lot of information on HPRT,
                                    297

-------
 and it is appropriate for this type of study, but you need to know if it is representative
 of other regions of the genome.
        In the germ cells we considered dominant visible mutations. Clearly it would
 be important to have other types of mutations, not just that small set.  And maybe
 something that has a dearer definition of genotype to phenotype.  This is what we
 had available, and we are not sure that the 50 loci is a good firm figure.  I think Susan
 Lewis would say it has got quite a bit of slop in it. We would like to have something
 that was better defined as a relationship between number of genes and phenotypes.
        And then at low doses and low exposure rates, the cell stage of importance
 for a cumulative dose estimate will be the stem cell. And we think that it is important
 to look at stem cell effects under these other exposure conditions, particularly low
 dose and low exposure rates.
        Moving to the next question:  What research is required to address the major
 assumptions? Well, they almost fall out of our parallelogram approach, but there were
 some additional ones that relate to this data set and other assumptions that might be
 made.
       One issue that is important is to look at some retrospective dosimetry, and that
 is an area in which Natarajan is particularly interested. Because we are now in a
 position to  look  at  stable chromosomal  aberrations,  particularly  reciprocal
translocations,  using the in situ hybridization method, retrospectively.  These  are
stable aberrations which  are transmitted through multiple somatic cell generations.
So the opportunity to do retrospective dosimetry allows for the endpoint, one of the
endpoints in germ cells,  the translocation, to be related to the somatic  cell  event
rather directly.

                                     298

-------
       We felt that rather little attention is being paid to aneuploidy, which clearly is
 a genetic risk feature. We really did not have the data sets upon which to work.  Part
 of that is due to the fact that it needs to be determined if the mouse is a reasonable
 model for aneuploidy. It is pretty tough to demonstrate aneuploidy in the mouse, in
 the offspring, certainly from exposure of the germ cell.  There was some discussion
 of this, and whether other models were more appropriate.  In particular we  had
 discussions of: What is the mechanism of aneuploidy here, and in what areas did we
 think  that chemicals such as ethylene oxide would  impact  upon induction of
 aneuploidies? So, we reached a generally, what I would describe as a holding pattern
 on aneuploidy.  We think it is well  worth  considering additional studies from
 reconsidering the mouse as a model, or to utilize Natarajan's favorite, the Djungarian
 hamster.  There is rather  little data available on the offspring of the Djungarian
 hamster.
       Several mutations,  rather mutational assays, should  be used,  including,  I
 emphasize, relevant ones.  I do not mean to  imply that the type of information we
 used from HPRT was irrelevant information; it is Just that there is a question whether
 one should assess mutations which would have a specific human impact in our germ
 cell assessment.  One needs to look at additional mutational assays to see whether
 the sensitivity is equivalent between the human and mouse somatic cells as the HPRT
 approximately was.
       We need to look at more  mutational data in  germ  cells at all  stages of
spermatogenesis and oogenesis.  This impacts upon the last question, which is:
What is available for assessing genetic effects in female germ cells? We did not have
any data set to work with in the female germ cell.  We could make estimates  of a

                                     299

-------
 lower sensitivity than the male germ cell, and anywhere up to equal sensitivity.  For
 heritable translocations, as I said yesterday, the female germ cell is probably rather
 insensitive, certainly in the oocyte.
        What is needed to fill in these assumptions, or data gaps that we have? There
 is very little, little or nothing on chronic low-dose, chronic exposures and low-dose
 exposures, for somatic or germ cells.  And that is clearly needed.  That is  the
 exposure scenario we are interested in.  You really cannot model, with chemical
 agents, from acute exposures to chronic exposures.  It is a little easier with radiation,
 because of the definition of dose and the physical characteristics of radiation.
        Another area that  really impacts very heavily upon both extrapolation and
 genetic risk is repair in germ cells, and how it impacts upon the effective dose and
 the response. The repair  of the appropriate lesion in DMA, a DMA adduct or strand
 break, or whatever one is interested in, the repair of that will impact upon the effective
 dose, and the response, be it a chromosomal event or a mutational event.
       An oft-repeated question  is: There have been no human germ cell mutation
 inducers demonstrated:   how or what are you going to do about it? The question
 really is: How do you demonstrate an effect?  We had some discussion about  the
 difficulty of actually demonstrating an effect, since you are looking at a low frequency
 event amongst a fairly high frequency background. And we discussed the likelihood
that the response one would look at in an F,, is probably a cancer one, that we would
have the greatest chance of being identified, rather than  a birth defect.  We  can
expand on that if appropriate.
       There are  some indicative studies, particularly from epidemiology and these
need to be pursued.  And I think that I am not going to really enlarge upon those too

                                     300

-------
 much; I think Mike Shelby is discussing a little bit of that this afternoon. They are only
 indicative, but we think that those are well worth looking at a little more closely to see
 whether we really have an opportunity to demonstrate germ cell effects in humans,
 or identifying germ cell  mutagens.
       So, the final question on our list for consideration was:  What would be the
 ideal data set for assessing genetic risk? Although the ideal data set is not the one
 we utilized in our  calculations,  it is the ideal one for us because it was the one
 available.  It is not very  difficult to come up with the ideal data set, as we all agreed.
 The ideal  data set is one that has all four corners of the parallelogram in it (Figure
 4-9), and whether you have the  germ cell and the somatic cell data, either direct or
 by using epidemiology as a surrogate.  Obviously if you have direct data for all four
 points, it is easy to put arrows on both ends of the extrapolation arrow. So, I put
 arrows here on both ends, because I felt that if you had all four corners you could do
 that.
       So, we know what the ideal data set is, we know that if you remove all the
 assumptions, and collect all the data that was indicated, I suppose you finish up with
 this (Figure 4-9).
       That is a summary of our deliberations. We felt with ethylene oxide that we
 had at least a data set that we could take some liberties with for extrapolations, and
we do appreciate the fact that there were liberties taken.  But I think the end product
was  quite  enlightening  to us.   It certainly has perked  my interest.   That is  my
presentation on behalf of ethylene oxide.
                                     301

-------
 The "perfect" data set for genetic risk assessment
         by the parallelogram approach
Rodent
Human
                   Surrogate
                 epidemiology
                  Figure 4-9
                     302

-------
 Discussion

       DR. LEWIS:  I just thought I would mention among those data which we did
 not really consider from our experiment, that one of the loci that was affected  by
 ethylene oxide was one that does give rise to a human genetic in-born error. So, with
 certain systems you can actually measure induction of human related disorders with
 chemicals, and this was one of them.
       DR. PRESTON: I should have mentioned in that same sentence that Susan
 mentioned her own data, there are two other mutations that were observed. These
 were electrophoretic variants that we did not consider. If they were seen in the same
 number of offspring, and that,  maximally two, were observed, then the germ cell risk
 would be half that which we have presented for the dominant visibles. But we did not
 consider these because we thought to use four was a bit brave.   We felt that
 estimates of risk from two was perhaps a little foolhardy.
       DR. ANDERSON:  I was very interested in what you said about the cancer
 endpoints. Would you consider that for genotoxic carcinogens, that cancer could be
 considered as an endpoint in the germ cell risk assessment?
       DR. PRESTON: We were particularly considering the Nomura data on the
 process that he has presented for radiation and for chemicals, where exposure of the
 parent did  result in increased cancer instances in the offspring.  Yes, we felt that
 maybe, it would be simpler to detect Ft cancer outcomes, or F, sensitivity to cancer
 production than detecting an induced birth defect.
       DR. EHRENBERG: There  have been experiments, you know very well, all of
you know,  Muller's experiments,  there are  several Russian studies published and
ongoing, proving increased cancer incidences in the offspring, second, third, fourth
                                   303

-------
 generation, after treatment with radiation, mostly radiation, but occasionally mutagenic
 chemicals. Considering diagnostic techniques to detect some effect, leukemia and
 some other cancers for example, the technique is so uniform and well diagnosed. I
 think this is one of the endpoints one could really look for in humans.  This pertains,
 in a way, to the Sellafield discussion.
        Besides that, if you compare the risk estimates, the ICPEMC group made
 some six or seven years ago, with respect to magnitude of a genetic risk, it is a very
 large response, per unit dose, which you have found in the animal tests, looking for
 cancer related induction. It is much larger than the other genetic endpoints specific
 locus  mutations, etc., etc.
        DR. PRESTON: That might be on the edge of the view that we took, but I think
 it is representative of our discussion.
        DR. SELBY: I would certainly be of the view that cancer endpoints would not
 be the useful ones to look at. The major problem is that for many of them, heritability
 is very low. If you look, for example, at what is known about human genetic disease,
 there are very large numbers of hereditary disorders that have high heritability. And
 many of the cancers would have low heritability, which would make the studies more
 complicated. I would tend to think a much better approach would be to use a cohort
 study of an exposed and  unexposed  pattern, somewhat  like what was done in
 Hiroshima  and Nagasaki.  I do not think here is the proper place to go into long
 detailed discussion of the application of the high frequencies found by Nomura as to
 what this might imply as far as future research directions.
       If some of you would be interested in seeing Court transcripts, or the Judge's
summary they can be distributed.

                                     304

-------
       DR PRESTON: I think that is an important point of discussion. I would point
 out that the induction of cancer is not necessarily what was meant here.  It is the
 induction of genetic alterations that increase the propensity to get cancer.   So,
 Nomura is one end of that. Mutated genes, any repair genes mutated in any of those
 increase the propensity.  I think that might show up as a cancer at this point not a
 genetic birth defect.
       DR SKOPEK:  You mentioned that HPRT might not be a relevant biomarker,
 and I agree, but there is also another layer of complexity with our studies as well.
 And that is, as I mentioned in our working group, the actual number of mutations you
 get at the HPRT locus in the animal model is very strongly a function of the age of the
 animals when they are treated and the time of sampling. And just by changing those
 two variables you can adjust those numbers by an order of magnitude, easily.
       DR. PRESTON: We did select the 20-week expression time because we felt
 that was more representative of the human situation, and it might not change much
 beyond 20 weeks. And we took that rather than the eight-week exposure.
       DR. SKOPEK: But if you treated young animals the numbers would be higher.
       DR PRESTON: That is right, I agree.
       DR. GENEROSO: I do not mean to contradict the conclusions of our group,
 but I think it is important to do so because the issue is very important. And I think it
 is important to express my dissenting opinion in our discussion.
       First of all, what I said was, my gut feeling, that it is going to be extremely
difficult in any human population, considering our experience with Hiroshima and
Nagasaki, and with the myriad of studies that have been filed, the most recent one
                                    305

-------
 being Sellafield, to come up with cause and effect for increased mutations in human
 populations.
       The other point that I made, and of course Lars has one opinion on this, and
 I would like to hear other opinions, is that the cancer rate in the children of Hiroshima,
 Nagasaki exposed population seems to be an obvious thing to measure. What is the
 experience in this?  This seems to be a good cohort to start with the possibility for
 using cancer in our genetic epidemiology.
       DR. SELBY: The extensive studies in Hiroshima and Nagasaki looked at many
 endpoints, one of which was cancer in children, and specifically leukemia. And there
 was no indication at all that there was an increase.  In fact, if you looked at the area
 that is considered to have people that were exposed, and look at their children
 compared to those outside of that circle, the  point estimate is slightly lower for the
 exposed.
       DR. PRESTON: We could discuss that, cause and effect, and all of these
 comments were ones that did arise.  We did also point out that the exposure, of
 course, was an acute exposure.
       DR. GENEROSO:  Just one more point that I think is important to bring up.
 And I think, while I do not believe in our lifetime that we are going to be able to come
 up with a smoking gun in genetic toxicology, I think we  ought to really think and
 develop another strategy. We all know that genetic toxicology is important.  We all
 know that chemicals can induce mutations that have  some effect.  And I think we
 ought to develop another strategy, and I do not know what that should be. We need
to  answer, or to counter, the question of genetic risk in humans from chemical
                                    306

-------
exposures without resorting to discussions of the significance or importance of
genetic toxicology.
                                    307

-------
308

-------
 Acrylamide (AA)

 Dr. George Douglas was the rapporteur for the Acrylamide group. Dr. Douglas is from
 Health Canada, Environmental Health Center in Ontario, Canada.

 Initial Report on Acrylamide

       DR.  DOUGLAS:  The  acrylamide group had the advantage, I  suppose, of
 having a look at the ICPEMC Report on acrylamide that was prepared for Health
 Canada, and, in brief, we decided that there were appropriate studies that we could
 use in modeling separately risk for both gene mutations and heritable translocations.
 The Ehling and Neuhauser-Klaus paper was used for gene mutation risk  assessment.
 There were two groups producing heritable translocation data: Shelby et al. and llse-
 Dore Adler. The ICPEMC Report will be used as a basis for the risk assessment
 modeling. Although, in some cases we used  different values for exposure. We used
 both the doubling dose approach, and the  direct method.  I think one thing that
 became  evident from our discussions was that the doubling dose approach really
 assumes a  linear response,  because in the formula, there are two points, the
 spontaneous and a data point, and so it results in linearity.  In the case where there
 is obvious curvilinearity in the  dose response, which appears to be the case in Use-
                                                 t
 Dore's data, we are limited by using single data points from that curve. Therefore, we
 have elected to derive different doubling dose estimates for the different data points
 in the dose response.  In addition, we are using a modification of the direct method,
which we are going to call the modified direct method.  We not using  precisely the
direct method as applied in the UNSCEAR reports, where there are dominant mouse
                                    309

-------
 mutation data available in the calculation.  In our case, we only have recessive mouse
 data with which to estimate dominant human data, so there is a broad assumption
 that  we are making, and so it  is not, strictly speaking, the direct approach,  In
 addition, later, when Kerry Dearfield returns to the EPA office, he is going to use their
 computer  to  derive some curve fitting models  to apply to the  risk assessment
 calculations. We are going to use the default risk extrapolation factors that you heard
 about in Dave Brusick's talk. The exposure data are those that were outlined as well
 in Kerry's presentation, in particular the occupational exposure by inhalation in sewer
 grout workers, that the  EPA collected.  As well, we are going to be making risk
 calculations for ingestion and skin absorption. There was some talk of whether or not
 there was  a  partitioning of effects  involving different  metabolites,  and whether
 glycidamide was the metabolite causing gene mutation effect versus acrylamide
 causing heritable translocation effects. We have not really worked that one to any
 satisfactory conclusion. With regard to the parallelogram approach, there is really no
 human data available, other than the Chinese hemoglobin adduct study, that will
 permit  us to use the  parallelogram approach.  I think we will  be using the
 parallelogram model to delineate  what additional tests are required to understand the
 risk involved from acrylamide. We are going to do a lot of different calculations. Have
 I missed anything?

Discussion

       OR WATERS:  How are you planning to address the issue of the contribution
of glycidimide versus acrylamide  and the two types of endpoints?
       OR DOUGLAS: We do not know.
                                     310

-------
       DR. DEARFIELD:  I just have one question from what Julian presented. This
 is a comment that related to what Udo said yesterday, as well, talking about the
 number of loci that we need to consider in our calculations, whether we are going to
 be using the default value of 1,000 that the  ICPEMC paper is presenting; but I noticed
 that you all put down 1,800, so, we all  are going to be trying to do the same
 calculations. I think as a whole group, we need to figure out which type of default
 values that we are going to use, if we are going to do similar calculations.
       DR WATERS:  I think Dave pretty  much Indicated yesterday, that that 1000
 figure was just the beginning of their thinking and Udo's point yesterday was pretty
 clear.
       DR. EHUNG: Yes, but I mean for the calculations, for the exercise, 1,000 was
 a wonderful figure.
       DR. WATERS:  Okay, scratch that,  we will make it 1,000, but somewhere in
 there, put a footnote, I think.
 Final Report on Acrylamide

       DR. DOUGLAS: I must say it is always a pleasure to work with the group of
 people present here today, and the extended group that you represent. Our group,
the acrylamide group, consisted of Kerry Dearfield (Chairman), Martha Moore, Udo
Ehling, Dave Brusick, Fred de Serres, Gary Sega, and myself.  I think,  as you recall
from Dave Brusick's talk, and in my preliminary report of what we were doing in the
Acrylamide group, that we are going to rely on the risk paradigm developed in the
ICPEMC report;  nevertheless, we diverged from that considerably, in that we used
                                    311

-------
 other exposure estimates to factor into the equation.  This table summarizes the
 information.
 Table 4-1.    Proposed  Scaling  Factors  (REF)  for Mouse to  Human  Risk
              Extrapolation
Parernetefs
Locus Specificity
DNA Repair Variability
Metabolic Variability
Dose Rate Variability
Exposure Route
Germ Cell Stage Specificity
Dose Response Kinetics
Experirnentsiy DuUMiiikMKJ^
Compound Specific
Compound Specific
Compound Specific
Compound Specific
Compound Specific
Compound Specific
Compound Specific
Defauk Factor
2 to 5 (assumes human sensitivity
0.1 (assumes mouse sensitivity)
1(70*)
0.1
.05
1
1
        • Each factor that is experimentally determined for the compound under assessment will be used.
         If compound specific Information Is not known, the appropriate default factor will be used.
        * If metabolism and detoxification are dependent upon glutathtone transferase (Bell et al., 1992).
       We have exposure through three different routes: 1) ingestion through drinking
 water; 2) two sets of exposure data on inhalation (the EPA grout worker's study with
 low and high exposures, and also the OSHA PEL for aery lam ide). We have a dermal
 exposure estimate, as well, from the same grout workers study conducted by EPA.
       All the exposure numbers have been converted to milligrams per kilograms per
 day, for convenience of comparison. We calculated risk numbers for gene mutations
 and chromosomal aberrations separately. The gene mutation numbers were derived
from the study by Udo Ehling, and are based on a single dose point (which is one
thing that we regard as a serious deficiency in the kind of data that is available). We
have some germ cell data, which are hard to obtain because funding is definitely
                                     312

-------
 inadequate.  We really need to have appropriate dose-response data for  gene
 mutation, similar to  the heritable translocation data.
       There are two estimates for gene mutation: the doubling dose estimate, and
 the modified direct  estimate.  At the 100 mg\kg dose, the gene mutation doubling
 dose was 53.  This was used with the different routes of exposures to derive risk
 estimates.
       The direct estimate for gene mutation risk does not bear much resemblance
 to  the doubling dose method estimate.  There seems to be, in this case, a  wide
 discrepancy  in the estimates.  And, as you will see later,  for the  chromosomal
 numbers, the direct and doubling dose estimates are much closer. One reason for
 this difference  could be that the number of mutant human mutations  leading  to
 measurable dominant detrimental effects, which is a key variable in the direct method
 equation, may  be over estimated,  leading to an over estimate of the risk.  The
 magnitude of this variable, and the equivalent one for chromosomal disease, needs
 to be considered carefully.
       For chromosomal aberrations we made three different sets of estimates.  One
 is based on the Shelby et al, study which gave a doubling dose of 0.1.  As you can
 see, the risk number are amazingly close in comparison to the direct method estimate.
       We made two estimates from  Ilse-Dore's data.  We used  a single  point
 estimate of 50 mg/kg, which yielded a doubling dose of 3.3, and with the associated
 risk numbers. And then we also used a doubling dose number that Ilse-Dore  gave
 us that was based on curve modelling  of the three point dose response curve.  As
you would expect, the people at the highest exposure level have the highest risk. 531
new genetic diseases per million people exposed is a very high number.  If we are

                                    313

-------
 going to start using these numbers we are going to have to explain them, and be able
 to justify them. We have to be able to tell regulatory managers that the numbers are
 meaningful.  I, personally, am not totally comfortable yet with these large estimates
 of risk.
        Now, I should point out that not included in these numbers is any factoring of
 tissue accessibility, which is something that was a consideration in the ethylene oxide
 risk analysis that has been published by EPA.  As well, female exposure  was not
 considered at all. The numbers are based on daily exposure, they are not based on
 time-weighted averaging, or reproductive life span of the people exposed. However,
 I would say the default factors that we used are, as assumptions, entirely valid. They
 are based mostly on assumptions that are fairly well established in the toxicological
 community, at least the ones that relate to exposure route and dosage and things like
 that.   Obviously we  have some work to do in the genetic related areas here,  in
 particular.
       We made an attempt to answer the questions posed to all groups, and as all
 the other groups probably found, did not have enough time to go into detail for some
 questions.
       Which are  the most important studies  available to support genetic risk
 assessment activities? We found, and I alluded to this earlier in my presentation, that
tests which give quantitative data, and dose-responses are required. I think that we
all recognize this need, but I think that we also recognize that the costs of the germ
cell tests are increasingly expensive, and that the budgets available are diminishingly
small.  So, I think we need to ensure that funding continues, and increases,  because
if we are really going to do modelling  we need dose-response data. And in the

                                     314

-------
 acrylamide data set, we are really lacking data on human exposure to permit us
 empirically to determine what these extrapolation factors are.  This is really PB/PK
 modelling, which just means using toxicology data to fine tune risk extrapolation
 factors and minimize assumptions.
        And we really should not forget about in vitro data, because it really helps us
 model  methods in a cost effective way that really cannot be done otherwise.
 However, we have to  keep in mind the limitations  of  extrapolating to in vivo
 conditions.
        What endpoints,  including, other than genetic are considered adverse affects?
 Obviously gene mutation and chromosomal mutation, Including aneuploidy, are the
 key events. I think that other indicative endpoints in the reproductive effects field are
 also relevant; but if you are thinking of these kinds of effects In a genetic context, you
 have to, in your studies,  elaborate the exposure/conception history of the people you
 are studying. In other words, you have to show a transmitted effect that is genetically
 based.  If you are studying exposed males, you have to ensure that the male is
 exposed prior to conception, that the chemical to which he was exposed to was not
 transmitted in the semen, and that he does not cany it into the home and expose the
 spouse post-conception and create a teratogen effect. I have talked on this subject
 previously, and described the concept in a paper in which nobody was interested.
       DR. DEARFIELD:  What was the name of that journal we used?
       OR DOUGLAS:   It was in  Mutagenesis, 5:421-423, 1990, if anybody is
interested.
       DR. DEARFIELD:  Have you got any reprints, George?
           DOUGLAS:  I have lots!
                                    315

-------
        Data gaps?  We really think that in the parallelogram paradigm, the most
 under-represented part of the model, in terms of data, is the human somatic cell
 corner.  It is essential, in these kind of studies, to have the appropriate dosimetry,
 preferably target tissue dosimetry.  And  also, where possible, to relate this to germ
 cell responses. These need not necessarily be  heritable endpoints, but could be
 indications of biological response in germ cells.
        Is the necessary dosimetry available? In the case of acrylamide, no.  To fulfil
 the parallelogram model you really need to have the same biological endpoint
 dosimetry in both  animals  and  human  somatic cells, or  germ  cells in vivo.
 Unfortunately, this is something that in most cases is really unachievable.  And of
 course, the biggest deficiency in any risk assessment is human exposure.
        What are the major assumptions in genetic risk assessment?  Probably the
 only one that really matters is: mouse equals man.  If we do not have the answers,
 we make a lot of assumptions,  in  toxicology we  may think we know a lot,  but we
 know less related to the genetic endpoints.  This is where the upper line of the
 parallelogram can help us.
       What are the  major uncertainties  in quantitative, in risk assessment?  After a
 short discussion, we came up with the  following. We understand a bit about the
 process of mutation, and can make some fairly comfortable assumptions about risk
 extrapolation factors; but we know  very little about the complexity of the responses
of different loci, and how different individuals, human  and animals, respond to toxic
insults at the organ and cellular level. These are all critical if one is  really going to
make informed estimates of risk.
                                     316

-------
       What research is required? Research on finding a human mutagen is needed.
 Frankly, it is seemingly an impossible task,  and I think we need to look at this in new
 ways. I think Diana Anderson had the right idea in terms of rethinking our concept
 of what  is a suitable endpoint. Are there other things we can look at? You might
 even want to read my Mutagenesis paper mentioned earlier.  Probably the biggest
 problem we have is funding, and I think this is a perception problem among funding
 bodies.  There is some inertia there among those that provide funding, because they
 do not see ft as a problem.
       Now, I would say that in my Department the lack of a human mutagen is more
 of an annoyance than a complete roadblock.  It is not the problem.  For classification
 of Priority Chemicals, germ cell mutation is a parallel endpoint to carcinogenesis. We
 have a rating scheme similar to the  (ARC scheme for carcinogenesis, and one with
 virtually  parallel wording for germ cell effects. The  only difference is that if we have
 a Class I  or Class  II carcinogen (a  human carcinogen,  or  probable human
 carcinogen), the chemical will be declared "toxic" under the terms of the legislation.
 Whereas, if it is  a Class I or Class II germ cell mutagen, ft mav be declared toxic.
 There is  a distinction between "will" and "may" that indicates the impact of the lack of
 a human mutagen.
       What is the priority?  Well, to find a human mutagen  is a priority.  But we
 should not  stop at that. I think that we need to not let the lack of this human
 mutagen inhibit us from moving ahead.
       Is the parallelogram approach applicable to acrylamide? Well,  yes and no.
 Conceptually ft is a very useful tool to focus our efforts; ft has become so much a part
of the way we think that we use ft all the time, without calling ft a parallelogram, ft is

                                    317

-------
 an ideal to strive for, and it tells you where your data gaps are, but lack of all the data
 needed to fill 3 of the corners does not prevent you from estimating human genetic
 risk from animal data (that is, the fourth corner).
        How would you address the problem of human cell mutagens?  Just let me
 say that I think that one would be cynical to think that humans are that much different
 from animals in the way the germ cells are affected.
        However, I think that one could do studies short of human epidemiologies!
 studies to detect increases in genetic effects in offspring in order to sort that out.
 Human germ cell studies on sperm, such as the work of Andy Wyrobek will, perhaps,
 give us some  clue  to the effects that  are  transmissible in human germ cells.
 Nevertheless, I  think the assumption that mouse equals man is a  reasonably
 comfortable one for me.
 Discussion
       DR ADLER:  Since George has mentioned Andy Wyrobek's work, and also
 Renee Martin's is important in this respect, I thought that we all have to think about
 an extension of the information that we get from the parallelogram. And particularly
 in terms of genetic risk assessment, we need to get away from the somatic cells as
 we try to use the data as they are available.  We found all sorts of problems with
 relating somatic cell mutation data from rodents to humans, splenocytes, the other is
 using lymphocytes, and all sorts of problems. But if we use sperm samples, and
 particularly in the exposed populations, both of animals or humans, as long as the
 samples are taken within  the exposure period, we are likely to  be able to measure
adducts, mutations, aberrations,  aneuploidy; and when we have a common cell type

                                    318

-------
that we look at, we can relate sensitivity correlations. The parallelogram, then, would
boil down to the lower part of this chart where we compare sperm to sperm, and we
estimate, or we experimentally determine the effect in progeny in rodents, and we
estimate the human risk.
      So, that gets us a bit closer to where we actually need to be, including germ
cells.
                                    319

-------
320

-------
 1,3-Butadiene(BD)

 Dr. Marja Sorsa was the rapporteur for the 1,3-Butadiene group. Dr. Sorsa is from the
 Finnish Institute of Occupational Health in Helsinki, Finland.

 Initial Report on 1,3-Butadiene

       DR. SORSA: Well, our small group was eagerly discussing the problems, but
 I think we  very quickly ended up  in realizing that plainly we have to be discussing
 data gaps.  There  are special  problems with butadiene.  Most of the data is  just
 coming in and therefore after a couple of years, we would certainty know much more.
 A special  problem  with  butadiene, that already has been  discussed here, is its
 metabolism and the specific metabolic features in the animal species as  compared
 to humans (Figure 4-10).  The genetically important metabolites, if we begin from  1,3-
 butadiene  here,  are circled in the figure here. We have mainly been discussing the
 monoepoxybutene and diepoxybutane, the  activity of which in various test results
 seems to be from 10 to 100 times higher than that of monoepoxybutene. There are
 additional genetically active metabolites, as well, (for example, crotonaldehyde,  and
 also the 1,2-dihydroxy-3,4-epoxybutane). Dr. Siv Osterman-Golkar informed us  that
 there is an additional suggested metabolite,  obviously also active genetically, which
 must be  there, since   in humans  exposed  to 1,3-butadiene, this specific M-1
 metabolite, the 1,2-dihydroxy-4-(N-acetylcysteinyl)butane has been observed. So when
going to the parallelogram possibility, we realized, right from the beginning, that we
are dealing with a compound with a specific metabolism. The parallelogram approach
                                    321

-------
              Specific differences in the metabolism of butadiene in vivo
     CHj-CtWH-CH,
               J
      CH^CHCH-CH,
     jCOfjctV
         i     *
         LJ Jfcii ••• W'

J    ctviwawpt.oH   J<-
    OH
                     (»f
                                                 QBH
.    CHrCM-CH-CHflH


I    OH  OH

•• ^m MB ^m ^B •• •• *•> •*• I


      CHVcH-CKCH,  —r
                                         K-
                                                 or


                                                        R
CH^CH CHCH,
                                           •(•*
                                                   CtV"CH-CH Cl^ 6R
                          WV-CHCH-CH.

                               OHOH
                                p.,

                                      Bl
    Compounds endosed in boxes have been identified in vivo as
    metabolites of butadiene; those that have been identified only
    tentatively are enclosed with broken lines. 3SH, glutathione
    s-transferase; EH, epoxide hydrolase


    Metabolism of 1,3-Butadiene (Henderson et al., 1993)
                            Figure 4-10
                                322

-------
 can only be taken using similar endpoints. So we first went to see what is known
 about the cytogenetic endpoints, concerning rodents and concerning humans.  We
 realized that so far, and there is one published and two unpublished studies, which
 all seem to confirm that at the exposure levels of the manufacturing industry, which
 are generally below 3 ppm level, none of the cytogenetic endpoints gives response.
 So, this is difficult then to continue with the chromosome damage parallelogram. We
 went through the available cytogenetic data, which clearly suggests that in mice, the
 lowest effective doses are surprising low (Figure 4-11), so that both for micronucleus,
 as well as the sister chromatid exchange endpoints in vivo only 6.25 ppm for a week's
 exposure is needed to get a positive response. Micronucleus is obviously the most
 sensitive endpoint.  In rats in vivo, the only data which suggests a positive response
 so far, seems to be the sister chromatid exchange, which is giving an  effect at 500
 ppm level,  but going as high as 1300  ppm with chromosome aberrations  or
 micronudei; there is no positive response in rats.
       DR. EHUNQ:  How long  was the time?
       DR. SORSA: Five days, two weeks, six hours per day. Then to the germinal
 data, which actually were important in picking up this compound for the exercise for
 parallelogram. We have dominant lethal mutation data from mice, from two exposure
 levels and two different laboratories, so that this data is .obviously true. The heritable
 translocation experiment is ongoing,  at the same exposure  level  as Dr.  Adler's
 laboratory.
       DR. SORSA: I  think, at present, it is too early to really judge the significance
of the male mediated  congenital malformations; it is a warning sign and  I think this
experiment certainly is something which  must be repeated. From the cytogenetic

                                     323

-------
                     Mice:  LED 6.25 ppm

                                 - Lung most sensitive

                                 - Heart angiosarcomas



                     Rats:  LED 1000 ppm

                                 - Mammary gland tumors
                     Humans     *     Cohort studies showed a slight
                                       excess   of  lymphatic/
                                       hematopoietic cancers  or no
                                       excesses

                                 *     Nested    case-control   study
                                       showed   large   excess  of
                                       leukemia
                    (ARC:  1987:Group2B
                           1992:Group2A
                                Figure 4-11


endpoint, I think it is going to be, at present, extremely difficult to continue the

parallelogram. Point mutations in vivo, give a little bit better chance.  In rodents, we

have two studies, (Figure 4-12). Both are just going to be published, but fortunately

we had Dr. Cochrane's data available in the working group and then we have data

from Dr. Tates; both are positive in mice for HPRT mutations (Figure 4-13).  In Dr.

Skopek's and Cochrane's data, the lowest effective dose was also the lowest dose,

and actually could be even lower, but the points where the effect was measured, i.e.


                                    324

-------
625 ppm. There is also positive  mouse spot test data at 500 ppm and what is
important, as Dr. Adler was pointing out, was that in general, the experience tells us,
that there is a very high correlation with positive mouse spot test data, and specific
locus mutations. This would also urge the importance of specific locus tests being
carried out.
                   INTERSPECIES DIFFERENCES
 •      Metabolic rate of elimination of BD Is greater in mice than in rats and
        that of BME in rats than in mice.
        At steady state the concentration of BME is greater in mice than rats.
       Tissue differences exist:  a very high level of BME is detected in mouse
       bone marrow.
       Role of other metabolites?
       Role of human variability and polymorphisms?

       Major detoxification pathway in mice in glutathione conjugation.
       Major detoxification pathway in rats, and obviously in monkeys and
       humans, is epoxide hydroxylase hydrolysis.
                                Figure 4-12
                                   325

-------
HPRT Mouse Splenocytes
Cochrane & Skopek 1993
1U

8

Or i i 0 600 Dose (ppm) Figure 4-13 326


-------
      in the limited human studies (Figure 4-14), so far, the only possibility for us to
use the same genetic endpoint in mice and man seems to be the point mutations in
HPRT locus. This study by Legator, et al, from butadiene manufacturing is the only
study in humans; it is in press now. The problems with the study are that the number
of workers is very small and the additional problem is that the methodology used here
is the autoradiographic detection method (Figure 4-15). The good thing in this study
is, that the exposure assessment is quite well performed; the data base of ambient
concentrations in that facility suggest that the time weighted average values, do not
increase 3  ppm, so that the general ambient levels are between  1  and 3 ppm.
Additionally the biological  monitoring measurements of the specific M-1  metabolite
were performed on the same individuals. This allowed the individual correlation to be
made, on the amount of the specific butadiene metabolite in the urine samples of the
workers and the HPRT mutation frequency measured as variant frequencies. The
correlation is extremely high.
                  HUMAN BIOMONITORING OF BD
       HPRT mutations increased in correlation to urine metabolite
       concentration.
       Chromosomal aberrations, sister chromatid exchanges or micronuclei
       are not associated with exposure at <3 ppm level.
       HB-valine adducts of BME detected at level 1-10 pmol/g from BD
       exposures 1-3 ppm; no adducts below 1 ppm.
                               Figure 4-14
                                   327

-------
                            HPRT Humans
                          Legatoretal.,1993
         10
         6

,.1  I
pi  f<
!§>"§  i
W (0
O-
	  - 2
          130
       1300

  Metabolite (ng/ml)
Low number of persons
Autoradiographic method
has to be repeated!
13,000
                       Figure 4-15
                          328

-------
        DR. SORSA: These are control persons from outside the facility and the other
 symbols are the persons working in the butadiene factory in the production. All of the
 subjects were non-smokers; this is also an additional strength of this study. The third
 possibility for the parallelogram, which we discussed in detail, is the possibility to use
 adducts; however, the DNA adduct studies are not yet available from humans. There
 is a monoexpoxide adduct which has been Identified by post-labelling in experimental
 animals in mice; diepoxide adduct standards exists, and we have good hopes to be
 ready with the results in a short while.   The possibilities  which  are available
 concerning adducts of 1,2-epoxybutene with the N-terminal valine in hemoglobin, were
 already discussed by Dr. Osterman-Golkar, and so the results suggest that in humans,
 the adducts can finally be measured. Adduct levels in the range of 1-3 pmol/g globin
 were detected among the  same persons in the Texas manufacturing plant as the
 HPRT mutations. No adducts were detected if the exposure levels are below 1 ppm.
 Adducts have been detected also in mice and they are three to five times higher in
 mice than in rats.   I think I will stop now and  go at  this point, to  any further
 discussions.
       DR BISHOP: Questions?  Lars?
       DR. EHRENBERG: Can you show the first or second overhead?  One of your
 overheads. What I wanted to discuss is these minus signs; they are discussed here
 in terms of yes or no.  They are all figures with a statistical uncertainty and a non-
 significant effect or absence of a significant effect could  always give you an upper
 level of a possible effect at a certain probability level and then it is a question whether
this, under current exposure conditions leads to an acceptable risk or not. The risk
is below an acceptability level; this means that throughout, I think, and one should

                                    329

-------
 also draw the conclusions from the absence of significant effects. They are quite
 useful.
       DR. SORSA:  But it does not help in parallelogram calculations, if we do not
 have the levels which would give us the equivalent response in humans and animals.
       DR. EHRENBERG:  You might, under certain conditions, use the upper 95%
 confidence level of a possible effect for a parallelogram comparison and in that way
 arrive at the estimate for humans, which you may ask, is it acceptable. This concerns
 all this problem.
       DR. EHUNG:  Of course you can go one step further and say there is no
 negative, there is only a certain probability to exclude, and mostly with a very high
 percentage in effect.
       DR SORSA:  Yes, I think this should be cited as a non-positive effect.
       DR. BISHOP: Diana, you do not want to comment on the male mediated?
       Da ANDERSON:  We discussed  the mechanistic thinking  behind the male
 mediated; we had quite a discussion in our group this morning, but it is  possibly a
 dominant effect, which is present in all dominant lethal studies, if we had looked in the
 past and we  have not and it occurs in a window.  It does not occur at very high
 doses, because the level of lethality is so great that you cannot pick up the response.
 If you come down to a low enough dose and you dissect the animals just before they
 come to term, you will pick up these congenital malformations; and to me, these are
 effects which  the human population would recognize, because you would not want
any of your children to have these abnormalities.   They persist, children are born
sometimes with exencelphy or definitely with spina bffida; some children are bom as
dwarfs and you can pick this up. So those are real effects that persist into adulthood

                                    330

-------
 or certainly dwarfism persists into adulthood and spina bffida children exist for some
 time, so those are real genetic events transmitted through the male.  They are not the
 same as Waldy is talking about, although I accept all that Waldy said about his being
 special epigenetic, but also indicative, maybe not genetic in the way we are thinking
 about it, maybe genetic in the terms of switching on, switching off genes from cell to
 cell; but this is transmission through the male germ cell and I just do not think we are
 thinking enough about those sorts of issues.  Look at how the cancer people and
 other areas of toxicology make their risk assessments; I know it is not our business
 at the moment, but I think we ought to think about it and this is a real mechanism
 which we can use and the public would identify with.  They identify with tumors, so
 why cannot they identify with congenital malformations? They would, and in terms
 of genetic effect it comes through the sperm and therefore it is a true inherited event.
       OR. DOUGLAS:  I wrote a discussion paper on this topic, mutagenesis to stir
 up the pot a bit, that this should be  considered in a human  population study.
       DR. ANDERSON: I just do not think we are taking it seriously enough; it has
 been there for years  in our data and we just have not explored it; it was just by
 chance that we got interested in it.  But It is a narrow window, that is all, because
 otherwise you get death in dominant lethality, or you do not get death, you get these
 malformations.
       DR. DOUGLAS: I think that from a population epidemiology perspective, that
this endpoint is something that has been overlooked.  I think that in some so-called
reproductive studies, where males have been able to partition exposure of males and
look at reproductive outcomes, that they sort of got at this idea, but they did not get
there with that intent and that end in mind; it was just part of a study that they broke

                                    331

-------
 up. I wrote a discussion paper in mutagenesis just to stir up the pot a little bit and
 so far, it has been over a year, I have not gotten any reprint requests; so obviously
 this is not foremost in peoples' minds and it did not stir up anybody's pot.  But I think
 that there is a lot there that we are not looking at that may be helpful if we are ever
 to get data on populations, because I think, based on the numbers, the spontaneous
 levels, that the studies are probably a little easier to do anyway.
       DR. ANDERSON: May I make just one more point?
       DR BISHOP: Sure.
       DR ANDERSON: We are losing credibility as a population, in terms of doing
 risk assessment, I think, because the public cannot identify with what is a heritable
 translocation; I am not decrying anything  that has gone before, but they do not
 actually understand what it means, they only understand what it means if they see
 things.  If they see things like tumors, they know ft can effect them; if they see things
 like congenital malformations, they know it can effect them.  You talk about a
 translocation or a specific locus, recessive mutations can take many years to come
 about; I know if you got a coat color change, but we do not have those sort of things
 in the human population. That is not meant to be funny or flippant, it is just what the
 public can identify with.  Now, I think we really ought to think about this as something
 that we can handle.  And if you see other areas of toxicology and what Kerry was
 saying yesterday, our manager wants to know what this means,  but predictions are
 made all the time using numbers.  And we are so pure, such purists  as geneticists;
we do not have to be purists, we ought to be able to tell the public, "you ought to be
concerned about this", but yet nobody, as yet, listens to us properly.  People listen
                                    332

-------
 in the field of radiation, because there have been methodologies worked out for years,
 but not for chemicals and I think it is our responsibility to do something about it now.

 Final Report on 1,3-Butadiene

       DR. SORSA:  Well, our working group was very small. It was chaired by Dr.
 Ilse-Dore Adler, and the participants in the group were Dr. Ekkehart Vogel, Dr. Siv
 Osterman-Golkar, then Drs. Cochrane and Skopek, who today appeared, and myself.
 But the smallness of the group, I think, was reflected by the fact that there is not very
 much  material available to do a proper genetic risk estimation on 1,3-butadiene.
 Since the previous Rapporteurs made some announcements, advertisements, maybe
 I should  also do the same.   Much of the material which is  available now was
 presented in a meeting on butadiene and styrene on April 1993 and in a proceedings
 publication will appear in the beginning of November 1993 with the (ARC Scientific
 Publications number 127. Also there is a report of the happenings and the highlights
 of the meeting in the November issue of the Journal of Occupational Medicine.
       Very briefly, I will give background material from that meeting.  The major
 reason why the meeting was convened on Butadiene was that it is carcinogenic, and
 I will very quickly go through that evidence.
       Mouse is a very sensitive species, the lowest effective dose for cancer being
only about 6 ppm.  Lung  is the most sensitive target organ and there are some
extremely unusual cancers, also, for example, the hemangisarcomas of the heart.
Rats are far less sensitive, and that we already discussed yesterday. The main target
organ for cancer is the mammary gland.
                                    333

-------
        Human cohort studies have either shown or not shown a slight excess in the
 lymphatic and hematopoietic cancers.  The heavily discussed studies are the new
 nested case-control studies, which seem to show a very large excess of leukemia.
 And as you know, the updating of Butadiene by (ARC took place in 1992 when it was
 upgraded into category Group 2A, a probable human carcinogen.
        Also concerning Butadiene, as was discussed in the meeting, the species
 differences were much discussed. There were several presentations on that, and the
 evidence seems to  be quite convincing concerning the metabolic rate of elimination
 being the key issue and the biggest difference between  mice and rats. And I think
 we can safely state that the  butadiene monoepoxide level is higher in mice than in
 rats.  Also, what is quite interesting, and personally I fee! should be a reason for future
 studies, is that there seem to be metabolic differences between different tissues. Very
 high levels of butadiene monoepoxide have been reported, for example, in the mouse
 bone marrow.
       The role of the other metabolites is still quite uncertain, and we discussed that
 yesterday.   Human variability and the  polymorphisms involved  are also certainly
                                                              ^
 something  about which we know extremely little. And then, there is variability in the
 detoxification pathway,  as we  already  discussed.  It  seems to be  so that the
 resemblance of humans in this respect is more related to rats than mice.
       Concerning the human biomonitoring data, and this is what we are going to
 have as the basis in  our extrapolation, there is only one study which shows that there
 is an HPRT mutation increase among workers exposed to Butadiene, as I told you
yesterday.  None of the cytogenetic studies shows any response.
                                    334

-------
       There  are  at  least two  research  groups who have  been studying the
 hemoglobin adduct levels in relation to butadiene exposure:  the group of Dr. Siv
 Osterman-Golkar and the group of Dr. Hans-Gunther Neumann. The detection limit
 of adducts seems to be at exposure levels exceeding 1 ppm, while the detection of
 adducts at lower than 1 ppm levels has not been reported.
       Now we move to the very brave and very inaccurate estimates of our working
 group, which are based on the HPRT mutations in somatic cells, which is the only
 endpoint available both in humans as well as in mice.  The data (Table 4-2) derives
 from a human study by Dr. Legator et al. (1993) and the mouse study by Dr. Skopek
 and Dr. Cochrane in an article submitted for publication. Both of latter authors are
 here so we have the possibility to discuss their data with the authors themselves.
       In the butadiene worker study, as I showed to you yesterday, the number of
 exposed workers was very small, only  13; and this consisted of workers in the low
 exposure category and workers in the higher exposure category, with mean exposure
 time of 7.8 years. And as has been reported in the publication (Legator et al. 1993),
 the overall average exposure level has been in the range of 1 ppm. We can calculate,
 if the workers work about 1,600 hours per year, that the ppm hours for the exposed
 worker category was 12,480 ppm hours.
       The HPRT mutation frequency levels, as I mentioned yesterday, were detected
 by the autoradiographic method.  Mutation frequencies in the exposed group were
 2.5 x 10"6, and for the controls very low, only 1.03 x 10"*. And we can then calculate
that the doubling dose of HPRT mutations would be about 10,000 ppm hours in the
lymphocytes of  these workers.  The mouse studies have been performed on
splenocytes of very young animals; altogether the number of animals was  8. The

                                    335

-------
 exposure was 625 ppm, for 2 weeks time, 6 hrs per day, total of 10 days. Total
 exposure in ppm hours was 37,500. The HPRT mutation frequency in the controls was
 1.3 x 10*, and  in the exposed group it was roughly five times higher, so that the
 doubling dose of butadiene, if we expect a linear dose response is about 15,000 ppm
 hours.  Surprisingly similar to the value in humans.

 Table 4-2,    Estimate of Comparative Doubling Dose of HPRT Mutations by
••"f— "••
Reference study
Study group
Size
Exposure data
Estimate of
ppm hrs
HPRT mutation frequency
in controls
In exposed
Estimate of
doubling dose
MICE
Cochrane & Skopek 1993,
In press
B6C3F1
8
625 ppm
6 hrs/d, 5 d/w,
2 weeks
60x625 = 37500
1.3 x 10*
6.0 x 10*
15 000 ppm hrs
HUMANS
Legator et al. 1993, In press
Butadiene production workers
13
TWA1 ppm
mean exposure 7.8 yrs
1600x7.8x1 =12480
1.03 x Iff6
2.5 X10"6
10000 ppm hrs
       But since we had the problems in the metabolic differences between mice and
humans and in the type of the tissue examined, our estimations in general are based
on butadiene and not the metabolites; this estimation cannot be considered relevant,
only a brave attempt.
                                   336

-------
       Among the other interesting results on human biomonltoring, which in general
 is among the future research priorities, the hemoglobin adduct monitoring studies
 performed by Siv Osterman-Golkar have produced results which show that there is
 some dose related effect on the adducts level. These adducts measured are now the
 monoepoxide adducts but we are still uncertain how well these are representing the
 genetically significant adducts. Future research certainly should try to focus on the
 other metabolites as well.
       But anyway I think this is reassuring, and it has even been repeated in another
 study population  exposed to roughly similar levels of butadiene but in  another
 country, in  Portugal. Also these so far unpublished results show that there  is an
 exposure related effect in the workers, as compared to the control levels.
       Our working group actually was discussing these human biomonitoring results
 and the urgent research needed for better understanding of genetic risks with
 Butadiene.  The confirmation of the worker study concerning the HPRT mutation
 frequency needs to be performed, preferably with another methodology. Confirmation
 is also needed on the  important findings from Diana Anderson's group which were
 discussed yesterday. The male mediated malformations at the low dose of 12.5 ppm
 were observed and urgently need to be confirmed.
       The metabolism studies, especially towards better understanding of human
 metabolism and in relation to the genetic polymorphism of metabolic enzyme systems
 in humans, are among  the research priorities.
       DMA adduct studies are urgently needed, and especially, I think, this would be
 important for the tissue specific adduct determinations.  One example of a need is in
the testicular tissue for the genetic risks estimation. In our group, Ekkehart Vogel was

                                    337

-------
 suggesting mutation sequence analysis for butadiene studies,  and I think this is
 certainly a way which the research should be focused on. Mutation spectrum analysis
 could also give us a better understanding of the  Importance  of the different
 metabolites of butadiene,  and this should be performed  for germinal mutations.
 Heritable translocation data is especially needed if we want to have some confirmation
 in our parallelogram approaches.
        Ladies and gentlemen, I am not going to go  repeat all the discussions which
 we already have had with the other two groups concerning the questions which we
 were given.  I will maybe just take some highlights, because to  a  large extent our
 discussions have already been presented by the other Rapporteurs.
       The first question is the important  studies needed  to support genetic risk
 assessment activities.  Of course, we cannot deny that we do need the germinal cell
 studies, and I think especially the heritable translocation data. It was pointed out that
 concerning human risk, the balanced transiocations are important because also the
 carriers show impairment and reduced fertility.
       There is now test data available on some 30 chemicals concerning the genetic
 heritable effects. When we discussed which might be the important priority chemicals
 for risk assessment efforts, benzene came up because of its importance as a human
 exposing agent, but  also because of its obvious in vivo genetic toxicrty activity.
       The Questions 6, 7  and  8 were  the  ones which were  actually  already
 answered.  I think just one additional point concerning the future prospectives in
 addressing the problem of "no human germ cell mutagens", which theoretically, does
 not exist. An "ideal data set" is not available, since we are not living in a dream world.
The way to go would be to try to do the more biomarker studies, because this would

                                    338

-------
 enable, a better focusing of epidemiologic studies. This would give a better possibility
 to identify the internal, biologically relevant exposures in individuals. Exposure
 assessment is very often a flaw in many epidemiologic studies. Proper biomarker
 studies would enable the possibility for earlier prevention.  We do not need dead
 bodies for biomarker studies, so this would enable better focusing  of epidemiologic
 research, and in that sense it would be also cheaper.
 Discussion

       DR. DOUGLAS: I just want to make the observation that I think that when we
 express risk numbers we should do it with a common unit of exposure.  I do not
 necessarily know what it is, but in some cases we are talking about ppm hours, other
 cases we are talking milligrams per kilograms per day, and so on.  I think we really
 need to work on that, in particular for the numbers that come out in this report; if they
 are going to  come out as numbers, they ought to be  based on a common
 denominator.
       DR. SORSA: So, is there a consensus?
       DR. ANDERSON: For drugs you cannot do quite the same  thing.
       Da DOUGLAS: Of course.
       DR. SORSA: Well, ppms are, of course, for gases are the most easily available
 numbers to relate to real life situations.
       DR. ANDERSON: Just to take an example, ppm hours is very relevant in an
exposure  for  a work force; but if you have got  a drug, let  us  just take
cyclophosphamide in milligrams per kilogram per animal data studies, then when you
get up to the human studies they have this other sort of exposure  where they have

                                    339

-------
millimeters squared, which is related to the skin surface area. So, if you are trying to
equate the animal data, you have to then convert it to a unit for the clinical situation.
So, it is not as easy as all that.
       OR DOUGLAS: But there are units that are used in particular application. But
if you are going to put all the numbers together, you should as well make them as
much in common as possible.
       DR. ADUER: You just have to produce conversion term to table.
       DR.SORSA:  I think so.
       DR. DOUGLAS: We do that routine with gases.
       DR. WATERS: I think we will make an attempt to do that, although I do not
know that this is a major point.
                                   340

-------
 Cyclophosphamide (CP)

 Dr. Paul B. Selby presented the report on Cyclophosphamide. Dr. Selby is from Oak
 Ridge National Laboratory, Biology Division, in Oak Ridge, Tennessee, U.S.A.

 Initial Report on Cyclophosphamide

        DR. SELBY:   We have  not  yet  gotten very far in  trying  to  work out
 parallelograms for Cyclophosphamide. How far we will get in this task is questionable,
 but we will make an honest attempt to develop parallelograms. The group working
 on Cyclophosphamide consists of Dr. Diana Anderson as chairperson, and of Drs.
 Jack Bishop, R. Colin Gamer, Pat Ostrosky-Wegman, and myself. Thus far we have
 spent most of our time discussing reports prepared by these members and identifying
 gaps in knowledge. We have not yet prioritized the gaps in knowledge, but my list of
 those identified will give you some feel for how things  are moving.
       It  seems clear  that  metabolism  needs to  be  better understood  for
 Cyclophosphamide.   Peculiar findings  have  been found after  certain  treatment
 regimens. Some of these peculiar findings might result from differences in metabolism
 over time,  and this possibility needs  to be pursued.  The Information on adduct
 formation for this chemical is extremely complex. It is not clear what adducts should
 be measured in humans.  Apparently there are no clearly defined data available on
 adducts in humans. Adducts have been studied in mammals, but curiously they have
 not been studied in reproductive cells. One peculiar thing about Cyclophosphamide
is that while there are voluminous amounts  of data in the literature, rather little of it is
recent.  It might be worthwhile to see if results from alkaline-elution or unscheduled
                                    341

-------
 DMA synthesis studies might serve as a surrogate for data on adduct formation.
 There is some question how useful information on adducts will be. Probably more
 than 200 studies have been carried out on somatic cells, but most of them have
 focused on clastogenic effects.  Some studies show effects of metabolic activation,
 but many more do not.   Incidentally, in this regard, there was an  error in  the
 viewgraph that was shown yesterday related to this.  The lines "with metabolic
 activation" and "without metabolic activation" got switched, which caused confusion
 about the importance of metabolic activation.
        There is relatively little information available for cyclophosphamide on induction
 of gene  mutations  in somatic  cells in  vivo;  one study  in humans  and one in
 experimental mammals gave positive results.  There is also very little information
 available on the persistence of gene mutations after treatment ends. More information
 is needed on induction of point mutations in In vivo studies, using hemoglobin, HPRT,
 and giycophorin.
       Jack Bishop described the extensive information available on germ-cell effects;
 many studies have shown very clearly that there is induction of dominant lethality.
 There  are good data available  for  induction   of  heritable  translocations  and
 specific-locus mutations.   Heritable translocations were induced primarily in  the
 second and  third weeks following treatment, and  specific-locus mutations were
 induced in the first three weeks following treatment.  There is no suggestion of any
 induction of specific-locus mutations in a large study on exposed spermatogonial
 stem cells.
       In regard to what was said in the prior presentation, there is a Jenkinson and
Anderson paper that gives results  on induction of  congenital  malformations in

                                     342

-------
 progeny of exposed rats.  Even though it is difficult to know how to apply those
 results, that paper is important because it strongly suggests that there are mutations
 induced that have effects in heterozygotes.  UNSCEAR (United Nations Scientific
 Committee on the Effects of Atomic Radiation) has considered whether congenital
 malformation data can be applied in risk estimation, perhaps in a way similar to the
 direct method.   For various reasons, it has thus far never applied those data in risk
 estimation.  We will give some more thought to the possibility of  doing so for
 cyclophosphamide since there are no other data available that can be  used to make
 a direct estimate of genetic risk. Perhaps, if we use sufficient caveats,  we can apply
 those data in quantitative risk estimation.
        Clearly,  for a  chemical  such  as  cyclophosphamide,  pharmacokinetic
 information is essential to the process of making any genetic risk estimate. Stage
 specificity must also be taken into consideration. There are special problems for this
 chemical in making extrapolations from high dose to low dose.  Most people exposed
 to cyclophosphamide  are  exposed in many fractionated  exposures during  their
 therapeutic treatment  regimen.   As a result, real low-level exposure would be
 extremely uncommon in the human population.
       Our work group had a fairly long discussion of what could be done in applying
 available  data  in  a  case  like this,  in  which  the  heritable-translocation  and
 specific-locus data show clearly that mutations are induced. I mentioned briefly some
 findings from my laboratory that I will develop more fully in my presentation tomorrow.
 Some of you know that we are conducting a series of long-term experiments using
the Assessment of Dominant Damage (ADD) method, in which the two most important
endpoints evaluated in first-generation  progeny are skeletal malformations and

                                    343

-------
 cataracts.  Our results show that while ethylnitrosourea (ENU)  is very effective in
 inducing dominant mutations with phenotypic effects, chlorambucil is surprisingly
 ineffective. Both treatments induce very high frequencies of specific-locus mutations,
 with the frequency being about twice as high for ENU.  Our samples are too small to
 exclude the possibility that chlorambucil may be inducing some dominant mutations.
 However, if it does, there is still a much smaller effect than would be expected based
 on the findings for specific-locus mutations.   These results are mentioned here
 because they may have relevance to cyclophosphamide.  Cyclophosphamide and
 chorambucil both induce dominant lethals, heritable translocations, and specific-locus
 mutations at about the same times following treatment in the male.  Unfortunately,
 there is much too little known presently to judge whether these similarities might also
 suggest that the mutations induced by cyclophosphamide are also quite unlikely to
 cause serious dominant effects.
       There was discussion as to what we might hope to accomplish in estimating
 genetic risk for cyclophosphamide. There was speculation that populations may be
 so infrequently exposed to low concentrations of cyclophosphamide that there may
 be little  need for risk assessment for the population at large.   If the  exposed
 population is primarily the group of people being given chemotherapy, the question
 is whether there is sufficient reliability  in any genetic risk estimates that  might be
 made for them to be applied in any useful way.  Something that must always be kept
in mind when estimating genetic risk is that the human population already has a large
genetic burden. One quite accurate figure, discussed in detail in the 1993 UNSCEAR
report, is that approximately 8% of liveborn humans will manifest a serious genetic or
partially genetic abnormality by the time they are 25 years old.  In the presence of

                                    344

-------
 such a large current burden, even if genetic risk estimates suggested that a patient
 might have a 1% chance of having a child  with an induced effect, it must be
 questioned how such an estimate would be applied. I doubt that it would be justified
 to suggest that such a patient not have children as a result of his treatment because
 his total risk is now 9% instead of 8%? Furthermore, any estimate of 1% induced risk
 might be a considerable overestimate.
        I have briefly described some of the findings and uncertainties that we must
 deal with when attempting to estimate genetic risk for cyclophosphamide. This is as
 far as we have gotten.  I  hope that we can accomplish much more by tomorrow.
 Thank you.
 Final Report on Cyclophosphamide
        DR. SELBY:  Thank you, Ekkehart.  The cyclophosphamide group consisted
 of Diana Anderson, Jack Bishop, R. Colin Gamer, Pat Ostrosky-Wegman, and myself.
 Much of what we did was to identify gaps in knowledge. I mentioned almost all of the
 major gaps in knowledge yesterday. I may touch on a few of them this morning, but
 I will not take time to review those gaps in knowledge today. We have not yet made
 enough progress to make a parallelogram calculation, but we intend to make one
 when working up a review.
       Some of the questions that we discussed were as follows: What data are
 available?   Are they acceptable data?  How  can they be applied in quantitative
 genetic risk estimation?  Ideally there would be data for relevant doses, at least for the
 germ-cell data.   Ideally we could apply  linearity of dose.  There is probably no
alternative but to assume that humans and rodents  are approximately equal  in

                                    345

-------
 sensitivity to mutational damage, or at least that any differences between them can
 be corrected for by using known differences found for somatic data. It would be nice
 to have the same cells on both comers of the parallelogram, but oftentimes the
 desirable data are lacking.
       With cyclophosphamide we have a somewhat peculiar situation regarding the
 doses applied in some  experiments. The doses were split into many  fractions
 administered on many successive days. For this reason, at first sight, the treatments
 may seem to be chronic exposures somewhat similar to the situation in which Dr. W.L
 Russell found that when ENU was administered in many small fractions, there was a
 reduction in the mutation frequency.  However, with cyclophosphamide, it must be
 kept in mind that just because  a dose is fractionated into smaller amounts does not
 mean that one is dealing with  a low-level exposure. It  turns out that the individual
 doses used in the fractionated cyclophosphamide experiments are rather similar to
 the individual doses used in fractionated therapeutic exposures for people.  As a
 result, the findings apply to high-dose therapeutic treatments instead of to long-term
 chronic exposures.
       We discussed the questionnaire supplied by the organizers, and I will not take
time to recite our responses now.  They are very similar to those given by other
working groups. Although we have not yet been able to make any quantitative risk
estimates for  cyclophosphamide, when we do so, we shall  attempt to  use a
parallelogram approach.  However, in some cases, in order to make a risk estimate
it may prove necessary just  to look at  the  germ-cell data and  make certain
assumptions about relative risk.
                                    346

-------
        Most of the rest of my comments will relate to the corner of the parallelogram
 that deals with germ-cell effects measured in mammals. Needless to say, one man
 who has contributed much to knowledge about induction of mutations in germ cells
 of mammals,  by radiation and chemicals, is W. L Russell.   Bill Russell  and Frits
 Sobels were good friends with a common Interest in genetic risk estimation for several
 decades. I no longer see Bill very often, but I happened to see him the day before
 I came to this meeting.  He asked me to mention one thing for him.  He told me that
 he had always hoped to attend a meeting on the parallelogram approach with Frits.
 He would have liked to have had a chance to tease his friend about the parallelogram
 approach by  suggesting that it is really the paralogism approach.  Paralogism is
 defined in the Shorter Oxford English Dictionary as  (1) a piece of false reasoning, or
 (2) a fallacy, especially one of which the reasoner is himself unconscious.  For various
 reasons, some of which have been discussed at this meeting, Bill has always been
 skeptical about the validity of applying  the parallelogram approach; however, I am
 sure that he appreciated the importance of what Frits was trying to accomplish by
 promoting the parallelogram approach.  Many of the uncertainties that must be dealt
 with when trying to apply the parallelogram approach must be dealt with in  any
 attempt to extrapolate from data collected in controlled experiments on laboratory
 animals to humans.
       Without question, one of the important sources of data that we have for
 cyclophosphamide is specific-locus data. The specific-locus method offers a precise
 means of counting mutations induced in mouse germ cells. Another important feature
of those data is that knowledge is rapidly expanding on the molecular genetics (that
is, what is actually happening at  the DMA level) near these seven genes. There is

                                     347

-------
 also a vast amount of historical data available for comparison.   But there are also
 some problems with  specific-locus data.  A  major difficulty  concerns how to
 extrapolate from recessive mutations found at only seven genes to estimations of
 clinical effects  in first-generation  offspring.  Another  important difficulty is that
 mutations are only scored at seven genes per animal, and it is now becoming very
 expensive to raise large numbers of mice for observation.
        If one is dealing with clastogenic effects, the most sophisticated method for
 counting  transmitted  mutations induced in  mouse  germ  cells is the heritable
 translocation test. This method also has the advantage of having much historical data
 available for comparison.  However, as mentioned by Dr. Julian Preston yesterday,
 there  is the difficulty of how to relate induced translocations to damage found in
 first-generation offspring.  One figure that has been used to make such an estimate,
 in the  1986 UNSCEAR Report, was that 9% of the unbalanced segregants of heritable
 translocations will survive  birth and have serious handicaps.  Dr. Udo Ehling  and  I
 took part in the discussions that led to the use of that figure of 9%.  The data on
 which  that figure was based were extremely soft; many of them were observations
 made  studying human pregnancies, and  we could easily have used those data to
 estimate that a much  smaller percentage of  unbalanced segregants would  have
 serious health problems, perhaps only 2%. Not only is this a very uncertain figure,
 but it is unknown whether the same percentage applies to different mutagens and to
different germ-cell stages.  It would be useful to obtain experimental data to explore
these questions.  At the present time, the best that can be done is to use the data
obtained for heritable translocations following treatment with cyclophosphamide and
make various assumptions recognizing these difficulties.
                                    348

-------
       The most precise data that we have for cyclophosphamide are (1) the data on
 induction of heritable translocations, for which there is induction predominantly in the
 second and third weeks after a single exposure, (2) specific-locus data, and (3) data
 on induction of dominant lethals. A recent paper from Dr. Udo Ehling's group shows
 that, following single exposures, there is induction of specific-locus mutations for
 matings made in the first three weeks after treatment. We shall use these data when
 we attempt to make a risk estimate for cyclophosphamide.
       Once you get away from mutations that can be precisely counted, such as
 heritable translocations and specific-locus mutations, and try to move in the direction
 of risk estimation, it becomes necessary to deal with the difficult question of what
 mutations really mean in terms of human health. How many induced mutations have
 clinical relevance?  The types of data that relate more directly to health effects deal
 with  mutations that cannot be counted as precisely.   I will briefly discuss one
 interesting set of data of this type.  These are the data of Jenkinson and Anderson
 referred  to yesterday. They exposed rats to cyclophosphamide in a fractionated
 treatment that was spread out over many weeks. Individual exposures were to 3.5 or
 5.1 mg/kg per day (at times it was 3.5 and at other times it was 5.1). They found that
 the percentage of dominant lethality increased until it reached a plateau of about 70%
 after about 7 weeks. The explanation for this plateau is unknown.  It could be that
 an equilibrium is reached and that they are mainly seeing the additive effects from the
first few weeks of treatment.  Alternatively, perhaps the long continued fractionation
leads to  a change in  metabolism of cyclophosphamide that thereby produces the
response that was observed. Of course, this finding is for dominant lethality, which
                                    349

-------
 does not relate to health effects seen in humans that survive birth.  Risk estimates, at
 least for radiation, usually only apply to abnormalities found in liveborn humans.
        Interestingly, following the same exposure regimen, Jenkinson and Anderson
 found a very similar  pattern of response for fetal abnormalities.  Such abnormal
 fetuses are often referred to as congenital malformations in the literature, and they are
 abnormalities found during uterine examinations made shortly before the time of birth.
 For this endpoint,  the  plateau was  reached by the seventh week  of continuous
 treatment, at which time about 10% of the fetuses were abnormal. An estimate of the
 induced frequency of abnormal fetuses can be made by subtracting the control
 frequencies from the experimental ones. The data seem to show convincingly that
 about 9% of the offspring had induced abnormalities. This study seems to be free of
 the difficulties present in some other studies of congenital malformations.
       The problem with such data, however, comes in deciding how to apply them
 in making a quantitative risk estimate. One of the difficulties  is that the experiment
 deals with only a rather small array of different abnormalities. Another difficulty is that
 about 60% of the fetuses said to have congenital malformations had only growth
 retardation. Some investigators call the small fetuses "dwarfs", but they are clearly not
 dwarfs in the clinical sense. In the Jenkinson and Anderson experiment, about a third
 of the animals with congenital malformations were shown to have gross chromosomal
 abnormalities.  Some of the fetuses with malformations may thus be   aneuploid
 segregants for translocations.  It is hard to know what clinical significance, if any, to
 attach to  many  of the malformations.  Some of  the  smaller fetuses might just be
toward the lower end  of the normal distribution of size and might develop normally
after birth.
                                     350

-------
       The Jenkinson-Anderson experiment deals with rats observed just before the
 time of birth, at which time they would probably be roughly equivalent in stage of
 development to a human at about four months of pregnancy. Already at this time, 12
 of the 41 mice with congenital malformations were dead. Thus, many of the fetuses
 said to have congenital malformations clearly do not relate to birth defects seen in
 liveborn humans.  Probably the abnormal fetuses relate partly to anomalies with no
 clinical relevance and  partly  to  late miscarriages, still  births,  and some gross
 congenital abnormalities present at the time of birth. We will attempt to make a risk
 estimate using  these data, but in doing so it will be necessary to make assumptions
 to cover the above uncertainties.
       For radiation risk, by the direct method of genetic risk estimation, for one cGy
 (that is, 1  rad) of paternal exposure, an estimate of risk that I consider reasonable is
 shown in Figure 4-16. You will find this estimate to be similar to the risk estimate for
 paternal exposure that will be in the 1993 UNSCEAR Report, which will be published
 soon.  Incidentally, that report provides a very useful discussion of the uncertainties
 in genetic risk estimation, and it describes the types of information needed to improve
 risk estimates for hereditary effects of radiation.  That report also provides a valuable
 discussion of risk estimation from a clinical perspective. The main consultant in the
 development of that report was Dr. Judith  Hall, who is the  President-elect of  the
 American Society of Human Genetics.
       To better understand how risk estimates can be made for chemicals, it is
 useful to consider how risk estimates are made for  radiation, for which quantitative
risk estimates have been made by committees for  several decades. In the direct
method, as applied in Figure 4-16, the range of from 10 to 20 per million liveborn is

                                    351

-------
based on damage detected in first-generation offspring of exposed males. The 10 is
based on cataracts and the  20 on skeletal malformations. The estimate of 20 is
unchanged since it was first  made in 1977 by UNSCEAR. The second  part of the
estimate, of from 5 to 10 per million liveborn, is based on early deaths, that is, on
deaths that occur before mice grow up so that they can be examined for other effects.
That estimate is explained in the 1986 UNSCEAR Report and in the upcoming 1993
report.  Part of the induced damage estimated in both rows of the figure is expected
to result from induced translocations.
      For radiation we have much more data on damage in first-generation offspring
to use in making a rough estimate of genetic risk.  For chemicals, comparable data
is scarce or nonexistent.
      Genetic risk estimates for 1 cGy of paternal exposure
       to X or gamma radiation made by the direct method
                                     Expected number of seriously affected
                                     children  in  the  first  generation per
                                     million live births.
 Estimate based on phenotypic changes                 10-20
 seen in mice living 3 weeks or longer
 Estimate based on frequency of mice                  5-10
 dying  between birth and 3 weeks of
 age
Total                                              15-30

                               Figure 4-16

                                  352

-------
Crosscutting Papers
      This section contains the following crosscutting topics:

      •     Differential Repair of Chemically-induced Damage:
              Somatic vs. Germ Cells
      •     Chemically-induced Mutation:
              Comparative Outcome in Somatic vs. Germ Cells
      •     Qualitative and Quantitative Results from Short-term Tests
      •     Chemicals for Future Study of Heritable Genetic Damage in Human
            Populations
                                 353

-------
354

-------
 Differential  Repair of Chemically-induced Damage:  Somatic vs.
 Germ Cells
 The first crosscutting paper was delivered by Or. Ekkehart W. Vogel. Dr. Vogel is from
 the Department of Radiation Genetics and Chemical Mutagenesis, State University of
 Leiden, The Netherlands.

       DR VOGEL At this point, Dr. Natarajan and I had been asked to deal with
 differential repair of chemically induced damage in somatic versus germ cells. The first
 thing we did was to delete the term "differential", because germ cell mutagenesis itself
 is so decidedly complex that it seemed impossible to contrast on one side genetic
 damage in somatic cells with  mutagenic effects seen in germ cells on the other. The
 next question, then, was what might be relevant for the purpose of this meeting. We
 first had in mind to just compare in mammals mutagenesis in somatic tissue versus
 germ cells. At the same time we felt inclusion of a non-mammalian species such as
 Drosophila would not be appropriate, because  a general view  seems to be that
 Drosophila is not a model system for the mouse. However, when  checking the
 literature we became surprised and fascinated by a number of similarities seen in both
 species in the responses to mutagens. Thus in the following, experimental data with
 respect to some major principles of  germ-cell mutagenesis in  male mice  vs.
 Drosophila males is discussed against the mechanistic mode of action of genotoxins
assayed in germ cells. The chemicals with data for germ cells are predominantly
alkylating agents.
                                   355

-------
 The Mate Germ-cell Cycle

 The meiotic cell cycle is associated with changes from diploid to haploid. Of most
 interest for our understanding of genetic damage formation in germ cells is a period
 of repair incompetence which lasts 3-4 days in Drosophila and 14 days in the mouse
 (from differentiation of  late  spermatids  onwards). The efficient  functioning of the
 maternal repair system, which operates  in the egg after fertilization, is therefore of
 utmost importance for removal of DMA damage carried by male germ cells. The total
 duration of spermatogenesis  is  about  35 days  in  the  mouse and 11  days in
 Drosophila. Another  characteristic feature of germ-line mutagenesis is germinal
 selection prior to  meiosis (Muller, 1954). Thus,  the interaction of various complex
 biological  processes, (that is, the efficiency of repair  enzymes, the capacity of the
 germ cells for metabolism, and germinal selection during and prior to meiosis), largely
 determine  the mutational spectrum finally observed in  each individual cell stage.
       For a comparison of mutation spectra in mouse versus Drosophila, there is
 data from specific-locus experiments for both species. In addition, for Drosophila there
 is information on multi-locus recessive lethal assays (Lee et al., 1983); that is, there
 is information from the  X-linked recessive test,  where activity is measured in 700
 genes, and from the  lind chromosome  recessive  lethal assay, a system covering
 about 1400 to 1500 genes located on the second chromosome.

DNA Damage and Repair in Drosophila

For processing of chemically induced DNA lesions in Drosophila, two loci, namely the
mei-9 (X-chromosomal) and mus(2)201 (iind chromosome), are of particular interest.
Both mutants have  been shown to block nucleotide excision repair (NER) and to have
                                    356

-------
 no UDS (Boyd et al., 1976; Lee and Kelley, 1986). Mutants of these two loci have
 extensively been used to study the consequences of a deficiency in NER for (Figure
 5-1) alkylation-induced mutagenesis. In the example shown in Figure 5-1, wild-type
 (exr+, Base) males treated with 0.15 mM MMS (methyl methanesulphonate) for 24
 hours were mated to either exr  (mei-9L1) or exr+ (Berlin-K) females,  and X-linked
 recessive lethal (RL) induction was measured at 10 successive broods, corresponding
 to the entire spermatogenesis cycle. The MMS dose was such that it would induce
 approximately 0.4 to 1% RL (« 2- to 5-times the spontaneous rate) in the cross with
 wild-type genotypes. The  time-related  potentiating effect seen for the postmeiotic
 stages, and the subsequent decline mutational yield in meiotic and pre-meiotic stages,
 supports the notion that DMA repair functions lose their efficiency as spermatogenesis
 progresses.  With MMS, up to 95% of  the mutations seen in the exr  group do not
 occur when the maternal excision repair is functioning (Figure 5-1). That this decline
 in mutation induction by MMS in the premeiotic stages is due to repair but not a result
 of germinal selection is indicated by the low yield in premeiotic stages of both X-
 chromosomal and of  ll-chromosome  RL The RL^,:  RL^  ratio (RL,p and RL^
 denotes recessive lethals  in spermatozoa and spermatogonia, respectively) is 23:1
 for the X and 19:1 for the second chromosome.
       The next point I want to discuss is the response of different germ-cell stages
 of  Orosophila   to   END  (N-ethyl-N-nitrosourea),   MMS,  and   EMS  (ethyl
 methanesulphonate) because for these three alkylating  agents there is also
 information  for the  mouse  specific-locus test.  Forward mutation  induction in
 Drosophila by  ENU, MMS,  EMS was  measured in  both  the lind and  the X
chromosome. The iind chromosome (40% of the genome) has about twice the length
                                    357

-------
   0)

   'w
   CO
   V
   o
   4)
   oc
            1      3      5      7      9      11     13     15


                                     Days

             Haploid   —>  —>      Diploid  germ-cells
% Recessive lethal mutations (RL) induced in Base males mated to either
(exr-) or to wild-type (exr+) females. Brood pattern analysis of 10 successive broods
(total 14 days). The enhanced induction of RL seen for the exr- groups indicates the
amount of DNA damage (N-alkylation) normally  repaired after fertilization by the
maternal repair system (From EW. Vogel, unpublished).
                             Figure 5-1
                                358

-------
 of the  X  chromosome  (20%).  Three  germ-cell  stages  were  compared: the
 postmeiotic spermatozoa and spermatids versus the premeiotic spermatogonia.
 Important for this analysis is that premeiotic male cells have two copies of the second
 but only one copy of the  X chromosome. With ENU (and  also with DEN,  N-
 nitrosodiethylamine) high incidences of mutations are found for the X and the second
 chromosome in both post- and premeiotic germ-cells, providing no indication for a
 significant germinal selection (Figure 5-2). For EMS the ratio of lind- to X-linked lethals
 was 1.8 for spermatozoa and 14:1 for spermatogonia, indicating a selection against
 mutations located on the X-chromosome prior to meiosis. It was already mentioned
 that with MMS a decrease in RL is seen for both the second and the X chromosome.
 Thus, for three  of the five alkylating  agents  depicted in Figure 5-2, a substantial
 portion of lesions induced on the X-chromosome do not pass through the filter of
 meiosis.
       Another analysis made was to compare the ratios of lind-chromosome RL in
 spermatozoa versus those in spermatogonia stages (Figure 5-3). In this case germinal
 selection should not have a big impact. While there were no significant changes in the
 RL.p:RL^ ratios for ENU and DEN, the low yields of RL in spermatogonia for MMS
 and (less so) for EMS indicate efficient error-free repair of alkylation-induced damage
 in premeiotic cells.  Surprisingly, the cross-linking agent DEB was very active in  all
 stages (Shukla and Auerbach, 1980).
       In summarizing the data for Drosophila,  the  following  can be concluded:
 basically, efficient paternal and maternal repair systems are available for DMA damage
 removal; there is, however, accumulation of premutagenic damage from N-alkylation
during a period of  repair  incompetence (3 days). For  the  purpose of a general

                                    359

-------
                     II:X  Spermitoioi
Spcrmtto(oni>
                20
                      Germinal selection:  MMS,  EMS.  DEB
                      MMS    EMS    ENU    DEN    DEB
 Ratios of second chromosome recessive lethals (RL,,) to X-linked recessive lethals
 (Ri-x)  for spermatozoa and spermatogonia. For spermatozoa the ratios are as
 expected (• 2:1)  because the  lind  chromosome  is twice the  length of the  X
 chromosome. The increase in  the RL^R!^ ratios seen for spermatogonia indicates
 germinal   selection   for  MMS  (methyl  methanesulphonate),   EMS  (ethyl
 methanesulphonate) and DEB  (1 ,2;3,4-diepoxybutane), but not for ENU (/v-ethylW-
 nitrosourea) and DEN (A/-nitrosodiethylamine). Data for MMS, EMS and ENU from
 Vogel and ZiJIstra, 1967; data for DEN and DEB from Shukla and Auerbach, 1980.
                             Figure 5-2
                                     Repair in Sg:
                                     MMS and EMS
                                        ENU??
                       MMS    EMS    ENU    DEN    DEB
Comparison of Induced lind-chromosome (autosomal) recessive lethal mutations In
spermatozoa (and some spermatids) versus Induction In spermatogonia. The high
RL.piRL^ ratio obtained for MMS and (less so) for EMS Indicates efficient DMA repair
in spermatogonia. RL.p:RL^ ratios are not significantly different from 1 for ENU and
DEB. A decrease is found ror DEN because mature spermatozoa  cannot activate
promutagens. For abbreviation and references see Figure 5-2.
                             Figure 5-3

                                 360

-------
 comparison, three distinct categories of alkyiating agents can be identified. Category
 1  chemicals, prototypes are MMS, DMS (dimethyl sulfate), EMS, and EO (ethylene
 oxide), are very efficient for ring-nitrogen alkylation. These agents show major activity
 in postmeiotic cells. Both DNA repair and germinal selection seem to be responsible
 for the low mutation yield from premeiotic male germ cell stages. To category 2
 belong agents with preferential ability for alkylation at oxygen atoms in DNA (Singer
 and Grunberger, 1983). Thus, ENU, DEN, and MNU (A/-methyl-A/-nitrosourea), are all
 highly genotoxic at all germ cell stages; a major role of germinal selection is  not
 evident. Category 3 (DEB, 1,2;3,4-diepoxybutane; BCNU, bis[chloroethyl]nitrosourea;
 CHL,  chlorambucil; HMPA, hexamethylphosphoramide)  represents  cross-linking
 agents which, in addition to their property to form mono-adducts, can crosslink DNA.
 Although germinal selection was demonstrated for this category (for example, for DEB,
 see Figure 5-2), they are highly genotoxic to all germ-cell stages. They are expected
 to mainly produce large deletions in postmeiotic cells, as was recently demonstrated
 for HMPA (Aguirrezabalaga et al., 1993; 1994). The data further suggest for premeiotic
 cells less efficient repair of cross-links in comparison to N-alkyl mono-adducts.

DNA Damage and Repair in the Mouse

The response of different male germ-cell stages in the mouse to the induction of
visible specific-locus mutations (data from the laboratories  at Neuherberg and Oak
Ridge) show striking  similarities to Drosophila. (Table 5-1). ENU (category 2) is
predominantly mutagenic in spermatogonial stages whereas MMS, EMS and DES
(diethyl sulfate)  (category 1), which  induce  relatively  high levels of  N-alkylations
                                     361

-------
Chemical

Acrylamfde monomer*'"
Category ]
1,2-Dibramoethane*
Ethylene oxide?
Methyl methanesulfonate*'d
Ethyl methanesulfgnate*fd
Di ethyl sulfate*'*
MMS+EMS+OES
Cateaory 2
g-Nethyl -N.-nf trosourea*'9
Procarbazine 'a
M- Ethyl -Jl-nl trosourea*^'"
MMHPRC+EMJ
Cateaory 3 .
TriethylenemelMlne*' '
Ch I orambuc 1 1
Nelphalan ' .
Nechlorethamlne1
Nitomycin C*'°
PlatinoP
Cyclophosphamide''*
Trophosphamide

IEN+CHL+NLP+CPP

AA
DBE
EO

PRC

TEN
CHL
NLP

NNC
CPP
TPP

Abbrev. Chemical
properties

s • 0.96
NNS s * 0.84; Nt
ENS s • 0.67; Et
DES s • 0.64; Et

MM s « 0.42; Nt
UkeNNU
EMI s • 0.26; Et

cross- 1 inking
cross- I Inking
cross- 1 inking
NEC cross- I Inking
cross- I Inking
PLA cross- I Inking
cross- 1 inking
cross- I Inking

Post-stem cell,
Resp. LL
* 3
I
15
:
5
* 1
* 11
* 13
*
?
*
* 3
28 9
OL UC
2
10 4 3
0 1 15
6 20
2 1 1
23 0
42 3
4
2 1
3
2

2
1
Resp. LL
*
-

:
n
+ i
+ 0
+ 1
•
* 0
2 21
Stem- cell snermatogonla
OL UC



1 3
2 44
8 246
WS 131
4 1
2 2
3 3

12 1
7



0
9
122







' L.B. Russel et al., 1990; b L.B.  Russell and Rlnchlk, 1993; c U.L. Russell,  1984; also negative with mouse dominant lethal assay and the electrophoretJc specific-
locus test  (Barnett et al.. 1992);  d Ehling and Favor.  1984; Ehling and NeuMuser-Klaus.  1984,  1989b.  1990; * Ehling  and  Neuhauser-Klaus. 1988«; r Ehling and
Neuhltuser-Klaus,  1988b;   • U.i.  Russel  and Nunsicker, 1983;                 ,
H W.L. Russell et al.. 1979; rCattanech, 1966; * L.B. Russell et al.. 1989;  l Ehling and Neuhauser-Klaus, 1989s; " Ehling and Neuh«user-Klaus,  1994; n Ehling and
NeuhCuser-Klaus,  1992; ° L.B. Russel et  al.,  1992a,  1992b;  p L.B. Russel et al., 1984 (positive for mouse dominant lethals and heritable trans locations (Generoso
et al., 1980;  ICPEMC,  1983);  q Favor  et  al.,  1994.

Abbreviations: Et,  ethylating; LL,  large (multi-locus) deletions; Nt, methyl sting; OL other lesions (Intra-locus); UC, unclassified; although the germ-cell stage
is a major determinant of the nature  of  mutations recovered, the type(s) of  initial DMA  lesions induced also have an impact.
                                                                               362

-------
 (predominantly 7-alkylguanine), are much more mutagenic in post-meiotic stages. This
 may be largely  due to N-alkylation products leading to  DMA breaks  and the
 accumulation of apurinic sites in these stages which have lost the capacity of repair.
 This hypothesis assumes that mutations are fixed following fertilization. In view of their
 preponderance for ring nitrogen alkylation, there should be a smaller chance for MMS-
 type mutagens to display a genotoxic effect in the early stages. This has indeed been
 found, as is evident from the negative test results in stem cells for MMS, EMS and
 OES (Table 5-1).  An intermediate position is taken by MNU and PRC (procarbazine)
 which produce specific-locus mutations in both spermatogonia and post-gonia cells.
 Thus a positive correlation seems to exist between increasing efficacy for alkylation
 at oxygens in DNA (category 2), in  particular at oxygens of pyrimidine bases, and
 mutation induction in premeiotic cells. The specific-locus test results summarized in
 Table 5-1 also contain 8 chemicals known to form cross-links with DNA. Interestingly,
 4 of these 8 mutagens displayed genotoxic effects in premeiotic male germ cells of
 the mouse.  Clearly, the general pattern emerging for the 3 categories in the mouse
 is very similar to what is seen in Drosophila.
       Retrospective analysis by Dr. Lee Russell of mutations recovered in chemical
 mutagenesis experiments revealed very many more  large deletions (LL)  than  other
 lesions  (OL; unknown  but  most likely all  intragenic  alterations  such as base-
 substitutions and intralocus  deletions) in postspermatogonial stages compared U>
 spermatogonia (Russell et al., 1990; Russell and Rinchik, 1993). If, however, the three
 agents MMS, EMS, and DES (all negative in premeiotic cells) are again regarded as
a group (group  1) distinct from MNU,  PRC,  and  ENU (group  2), and all the
monofunctional agents to be different from the sub-set of cross-linking agents (group

                                     363

-------
 3), It appears that the ratio of LL versus OL is not the same for the three groups. The
 LLOL ratio is as high as 2.5:1 (15 LL, 6 OL) for group 1, approximately 1:1 (5 LL, 6
 OL) for group 2, and again very high, 3:1 (27 LL, 9 OL) for group 3.  This suggests
 that although the germ-cell stage is a major determinant of the nature of mutations
 recovered, the types(s) of initial DNA lesions induced also have an impact.
 Comparison of Mouse and Drosophila
 This study suggests that a classification of alkylating agents into sub-groups based
 on their mode of DNA modification, rather than a consideration of individual agents,
 may be worthwhile. In fact, when data analysis is based on a consideration of initial
 interaction patterns with the DNA, three distinct categories of alkylating mutagens can
 be distinguished for both Drosophila and the mouse.
       The prototype of the first category is  MMS, an agent predominantly reacting
 with  ring  nitrogens in  DNA.  Broadly  speaking,  efficient,  error-free  repair  in
 spermatogonia appears to be the  protective mechanisms responsible for the low
 mutation yield in Drosophila and the absence of significant numbers of mutations in
 germ cells of the mouse. By contrast, there are no problems with most of  category
 1 mutagens to be detected in postmeiotic cells of Drosophila.  For three of theses
 agents (MMS, EMS,  and DES) molecular mutation spectra have become  available
 (Table 5-1).  MMS and EMS are quite efficient deletion makers in  Drosophila. In
postmeiotic germ-cells of the mouse, a high LLOL ratio of 2.5:1 is found for MMS,
EMS  and  DES   (Table  5-2).  Strikingly,  for  MMS,  DMS,  EO,  and  DBE  (1,2-
dibromoethane)  high TDg, values have been  estimated for rodents (Barbin and
                                    364

-------
Table 5-2. Genetic activity profiles of three categories of alleviating agent.
comparison with tumor 1 genie potency in rodents.

postmetotle cells
US XIGR

XLL
CEA weakly * for RL
EO"
VkBhS*
DBE
MMS
DES
ENS
* responses In RL test
positive for RL
None 78
45 23
93 7
89 6

22
32
no test
15
Mouse
XBS+IGR
inactive 7
inactive ?
inactive*

29
117]
mutants not
In visible specific-locus assays of Orosophila C vermilion locus) and the mouse (7 loci) In
Mouse
DOBtmeiotlc cells Stem-eel I en
XLL



71
83
specified
BS+I6R 	 LL 	
inactive 7
inactive 7
inactive
inactive
inactive
inactive
Inactive
prmatooonia

high 7
13700
10100
1650
1240
[100]
000]
Comments/predictions
mg/kg bw DD (mg/kg)
efficient repair expected
like CEA
like CEA

DO - 2 (PSG)
DO • 17 (PSG)
DO « 9 (PSG)
Category 2
NNU
PRC
DMN
NNNG
DEN
MN
ENU
52 32
positive in RL test
56 20
30 40
89 11

100 0
16

24
30
no test

0
P3]
[60]
no test
no test
inactive

[67]
[67]
[40]




(331
[75] [25]
96 4
no test
no test
inactive

97 3
28
262
139
27
56

11
DO - 110 (SG)



active In sperms t Ids for

DO - 4 (SG)
Category 3
DEB
CHL
MLP
NEC
positive for RL
positive for RL
positive for RL
positive for RL




cJsDDP positive for RL
MMC
TEN
positive for RL
positive for RL


positive 7
21
13
positive
Inactive
Inactive
[67]

79
87



[331
positive ? tow?
100 0
[75] [25]
inactive
inactive
100 0
[80] [20]

92
67
17
20
0.68
T

DD - 0.7 (PSG)
DO « 0.6 (PSG). 3.9 (SG)
DD « 0.1 (PSG)

DD « 1 (SG)
TD-n's: 5, trsnlmon; 110,
TEPA
HEMEL 0 33
HMPA 6 39
CPP
positive for RL
67
55

no test
7
[40]


[60]
no test
Inactive
Inactive
no test
27
2080


DD - 6 (PSG)
  positive for mouse dominant lethaIs and heritable translocatlons (ICPEMC, 1983);

Chemicals: CEA, 2-chloroethylamine; CPP, cyclophosphamlde; cisOOP. cisplatln; DBE,  1,2-dibromoethane; DEB, 1,2:3.4-diepoxybutane; DEN, K-nitrosodlethylamine; DES, diethyl
sulfate; DMN,
JJ-nitrosodimethylamine;  ECH, epichlorohydrin; EO*,  ethylena oxide; ENS,  ethyl methanesulphonate;  ENU,  Jl-ethyl-JJ-nitrosourea;  HEMEL,  hexamethylmelamine;  HMPA, hexamethyl-
phosphoramide; NEC, mechlorethamine; MLP, melphalan; MMC, mitomycin C; MMS, methyl methanesulphonste; MNNG, N.-methyl-NJ.-n1trojJ-nitrosoguen1dlne; NNU. M-methyl-N/nitrosourea;
PRC, procarbazine;  TEN,  triethylenemelemine.

Abbreviations: BS, base substitutions; DD, Doubling Dose in mg/kg (from Favor et al., 1994); IGR, intralocus rearrangements; LL, large (multi-locus) deletions; MN, micronuclei;
PSG, postspermatogonia; RL, X-linked recess ice lethal mutations; SG. apermatogonia; TD5Q, median life-time dose (mg/kg body weight) producing in SOX of animals (rodents) tumors
(Bar-bin and Bartsch,  1989; Vogel  et al., 1990).

Nutation spectra at the vermilion aene. DBE: Ballering et al., 1994; HNS: Nivard, 1991; Nlvard et al., 1992; MNU, DMN: Nlvsrd,  1991; EMS:  Pastlnk et al., 1991; MLD for EMS,
estimate based on sterile and male-lethal vermilion mutants; ENU: Pastlnk et al.. 1989;  DEN: L.M.  Sierra (personal communication); HEMP A, HEMEL: Aguirrezabalaga et al., 1993;
1994.

Nouse-SLT: For references see Table I.
                                                                                    365

-------
 Bartsch, 1989; Table 5-2); these values support the hypothesis of a generally low
 genotoxic efficiency of N-alkyl DNA-mono-adducts.
       To the second category belong MNU, PRO, DMN (A/-nitrosodimethylamine),
 DEN,  MNNG (/V-methylW-nrtro-N-nitrosoguanidine), and  ENU  (Table 5-2). These
 agents in common have a considerable ability for modification  at oxygens in DNA
 (Singer and Grunberger, 1983). Category 2 mutagens are monofunctional agents of
 low nucleophilic selectivity, which are expected to show activity in a wide range of
 both post-  and  premeiotic  germ-cell stages.  Base-pair  substitutions were the
 predominant type of mutations found in postmeiotic germ-cells of Drosophila with
 ENU and DEN (note  the higher proportion  of deletions by MNU, DMN, and MNNG).
 In agreement with a high mutation induction by these agents in Drosophila, a relatively
 low LL:OL ratio of 1:1 was observed for the mouse specific-locus test for MNU, PRO,
 and ENU (Table 5-2). A number of these agents have been demonstrated  to be
 among the most potent carcinogens in rodents, as is evident from their generally
 lower TDgo estimates compared to class 1 chemicals (Barbin and Bartsch, 1989; Vogel
 etal., 1990).
       Agents capable of cross-linking DNA form category 3. The most unexpected
 finding of this study  was the realization that the cross-linking agent DEB (1,2:3,4-
 diepoxybutane), with regard to the production of autosomal mutations, is equally
 active in both post- and premeiotic Drosophila germs cells (Shukla and Auerbach,
 1980).  In  accordance with  that observation, four  cross-linking  agents  (TEM,
triethylenemelamine;  CHL, chlorambucil; MLP, melphalan;  and MMC, mitomycin C)
display activity in stem-cell spermatogonia of the mouse (Table 5-2). The negative
data for c/sDDP (cisplatin), CPP (cyclophosphamide), and MMC is not necessarily

                                    366

-------
 against the argument of cross-linking agent being a potential risk factor for the entire
 spermatogenic cycle. Classification of 11 mutations induced by either chlorambucil
 or melphalan in spermatogonia revealed that 10 (91 %) of the 11 were either intragenic
 (IG) or other lesions (OQ; no large deletions, LL). By contrast, 24 (83%) of the 29
 mutations induced in postspermatogonial stages were of the LL category (Russell et
 al., 1990; Russell and Rinchik, 1993). The predominant occurrence of the LL category
 in postmeiotic germ-cells seems to be a general characteristic of the activity of cross-
 linking agents: 75% of all visible specific-locus  mutations were of this type (Table
 5-2). However, since the OL category should  include a considerable fraction of
 intragenic deletions, the total yield of rearrangements may reach more than 90%, as
 has recently been shown for mutations induced by HMPA at the  vermilion locus of
 Drosophila (Aguirrezabalaga et al., 1993; 1994). Thus cross-linking agents are highly
 effective in inducing deletion mutations in postmeiotic stages of both Drosophila and
 the  mouse. Regarding their tumorigenic potency in  rodents, (direct-acting) cross-
 linking agents  have  generally  lower TDg,  values than  their monofunctional
 counterparts. It therefore seems to us that the considerable genotoxic potential of
 cross-linking agents for  both tumor induction and incidence of heritable genetic
 damage is often  under-estimated, because the emphasis is usually more on their
 cytotoxicity - the desired effect for the use as cytostatic agents of chemicals such as
 cisplatin, melphalan, and chlorambucil.
       With all the caution  indicated  when trying to summarize  the  very complex
pattern of alkylation-induced mutagenesis in the two species, the following trends in
responses of germ cells occur:
                                     367

-------
 Category 1:   MMS, EMS, DES, and EO.
                     Active in postmeiotic germ-cells In both mouse and Drosophila
                     Efficient repair of ring-nitrogen alkylation in spermatogonia
                     Better deletion makers than category 2 chemicals.
 Category 2:   MNU, PRC, DMN, MNNG, DEN, and ENU
                     Activity expected for a wide range of germ-cell stages
                     O-alkylation adducts (at pyrimidines) seem responsible for
                     mutation induction  in  premeiotic  cells  because  changes
                     involving AT basepairs are the major event seen with ENU in
                     both Drosophila and the mouse.
 Category 3:   DEB, CHL, MLP, MEC, JEM, HEMEL, and HMPA
                     Activity expected for a wide range of germ-cell stages
                     Poor repair of cross-links in spermatogonia ?
                     Most efficient deletion makers in postmeiotic cells.
 Conclusions and Perspectives


 The most unexpected but encouraging outcome of the study is the identification of

 shared genotoxic activity profiles for three vastly different  biological indicators of

 genotoxictty: mutations in Drosophila vs.  mutations the mouse vs. tumors (TO^

 estimates) in rodents. It should be noticed that the distinction into two categories of

 monofunctional agents has a purely operational purpose; it facilitates the recognition

 of some common genetic action principles in the two germ-line systems.

       Based on the above 3-category classification scheme the following tentative

 conclusions can be drawn. Monofunctional agents belonging to category 1 display

 genotoxic effects in male germ-cell stages that have passed the meiotic division. We

 believe the reason for the rather impressive resistance of premeiotic stages and the

generally high TDg, estimates observed for this class in rodents is efficient error-free

repair of N-alkylation damage at low exposure levels. When accepting the hypothesis

of efficient error-free repair  of N-alkylation damage in vivo, then the  increased

                                    368

-------
 carcinogenic potential in rodents, seen when passing from category 1 to 2, along with
 their ability for genetic damage in premeiotic stages, must be due to the enhanced
 ability of category 2 type of chemicals for alkylations at oxygens in DNA; it is this
 property which distinguishes the two groups from each other.
        Both in terms of hereditable genetic damage and the initiation of tumors (low
 TDggS) cross-linking agents form a considerable genotoxic hazard. Doubling doses
 have been determined for 4 cross-linking agents not requiring metabolic conversion
 and in all four cases they were lower than those for MMS, DES, and EMS (Table 5-2).
 Four members of this class display activity in stem-cell spermatogonia indicating that
 this type of agent is active in a wide range of germ-cell stages. The negative test
 results obtained for MEC  (mechiorethamine),  c/sDDP, HMPA, CPP, and TPP
 (trophosphamide) in stem-cell spermatogonia is not in disagreement with such a
 conclusion  because the cause  could be rather trivial (for example, metabolic
 clearance). In contrast to what was observed in unicellular systems  (Yamada et al.,
 1992)  and in mammalian cells in culture (Nivard, 1991; Spelt et al., 1992), both cross-
 linking agents and MMS-type mutagens (high s value) predominantly produce deletion
 mutations in postmeiotic male germ-cell stages. This is the uniform picture found for
 both Drosophila and the mouse. It is concluded that in vitro systems, in contrast to
 Drosophila germ cells,  fail to predict this very intriguing  feature of mouse germ-line
 mutagenesis.  In addition to  their potential for  induction of deletions and  other
 rearrangements, cross-linking agents are among the most efficient inducers of mitotic
 recombination in Drosophila (Vogel  and  Nivard, 1993). Thus there are several
 mechanisms  by which cross-linking  agents  may cause loss of  heterozygocity,
generating genetic damage and/or cancers.

                                    369

-------
       We feel much more effort should be spent to learn about the time course of
 repair in germ-cells in vivo in relation to the types and amount of DMA lesions. This
 study suggests that DMA-crosslinks are not well repaired by premeiotic cells. This
 problem could be examined by studying pairs of structurally-related agents in repair-
 proficient and -deficient premeiotic cells of Drosophila, and in repair-competent mice.
 For  example, it would  be interesting  to compare the monofunctional  agents 2-
 chlorethylamine, ethylene oxide, and ethyl methanesulphonate with their  respective
 crosslinking counterparts nitrogen  mustard,  1,2;3,4-diepoxybutane and chloroethyl
 methanesulphonate. The prediction would be a considerably higher genotoxic activity
 of the latter sub-group compared to the agents only producing mono-alkyi adducts,
for which efficient DNA repair is anticipated.
                                    370

-------
 Chemically-induced Mutation:  Comparative Outcome in
 Somatic vs. Germ Cells
 Dr. Susan Lewis presented "Chemically-induced Mutation: Comparative Outcome in
 Somatic vs. Germ Cells."  Dr. Lewis is from Research Triangle Institute in Research
 Triangle Park, North Carolina, U.S.A.

       DR. LEWIS: What Jim and I are trying to do is point out some of the attributes
 special to germ cells that make human germinal risk calculations difficult. Taking into
 account the basic biology of both the male and female germ cell, perhaps versus the
 same somatic cells that are commonly used.  A lot of the relevant germ cell biology
 was covered in Banbury Report #34 edited by Jim Allen, Lee  Russell, Mary Lyon,
 Montrose Moses, and Bryn Bridges. This Banbury volume has a lot to do with the
 correlation of mutagenic responses and the biology of developing germ cells at all
 stages.

 Special Attributes of Germ Cells

 I want to emphasize the primary differences between germ cells  and somatic cells of
 all kinds. Germ cells and somatic cells both go through mitosis cell cycles at various
 stages, but meiosis is unique and special to germinal tissue.  I do not mean to make
 the somatic cells appear less complex as opposed to more complicated germ cells.
The variety of differentiation that various somatic cells do go through is considerable.
 For example, certain cell types can become polyploid.
      I want to remind you of a concept derived from work in carcinogenesis. If you
examine the NTP database on carcinogens, you find that different somatic tissues
                                   371

-------
 respond differentially to different carcinogens.   Thus, I  am not sure you can
 extrapolate mutation data from various somatic cells to possible consequences in
 germ cells.  At the very least, I think you have to be careful.  But, just remember,
 effects may vary according to mutagen, and exact correlation  may be very difficult.
 So, you cannot assume exact correlation between the somatic cells that are usually
 used for in vivo mutagenesis studies and germ cells.

 Female Germ Cells
 We will start with a consideration of the female germ cell. As a matter of fact, I would
 like to remind you that there is such a thing as a female germ  cell, and it is very
 different in its biology and response to mutagens from that in the male. The germ cell
 stages progress  in a very different manner in males and females.  The female germ
 cell is arrested in meiotic prophase.  In contrast, germ cells in  the male go through
 a continual cycle of spermatogonial renewal through its entire mature life span.
 Cell Environment

 Germ cells at different stages are in different environments that could significantly
 affect the access of mutagens. Udo Ehling has often reminded us more than once
 that the mature sperm, which is a mutagen-susceptible gtage of spermatogenesis, is
 no longer in the testes at all, but is in the epididymis. Also, the spermatogonia are
also outside what is now called the Sertoli cell barrier.  Both meiotic cells and
spermatids are within this barrier but in different positions relative to the barrier itself.
                                     372

-------
Sex Specificity of Mutagens

There are so called sex-specific mutagens in germ cells.  There are a number of
recent studies that show that there are some chemicals that are mutagenic in late
oocytes but not in any male germ cell stage.  Of course, there are a number of
mutagens that are effective in males but not females.  So, extrapolating from the male
to the female, which is often done, may not be valid. For purely logistical reasons,
it is much more difficult to do mutagenesis experiments on females than males.

Induction of Mutations Resulting in Inborn Errors of Metabolism

Finally, in regard to the contribution of chemical exposure of germ cells to the human
disease problem, there have been  several inborn errors in metabolism parallel to
those in humans or induced by mutagens in mice. The phenylalanine hydroxylase
seems to be an extremely susceptible loci to mutagens.
       In summary, the following table  (Table 5-3) presents some of the aspects of
mammalian germ cell, developmental biology, and genetics that should be taken into
account when assessing germinal risk.
                                    373

-------
Table 5-3. Factors to be Considered in Assessing Gene Mutations in Germ Cells

       •     There are sex-specific mutagens in germ cells. These act either in the
             mate or female but not in both.

       •     Germ cells  at different stages of gamelogenests are in different
                      xits that could affect access of mutagens to the largest cells.
             Stage-specific susceptibilities  are  important  both  for mutation
             frequency and for spectra of induced mutations.

             There are stage-specific differences in sensitivity to induced mutations
             in females as well as in males.  The conditions of exposure differ
             according to developmental stage. As a general rule, large lesions are
             induced in postgonial stages and smaller ones in spermatogonia.

             Effects of differential gene expression

             Expression of a gene may alter repair and thus induction of mutation.
             Coordinated  changes  in  gene  expression  occur  during  the
             development and moderation of germ cells of both sexes.

             Condition of the chromatin in germ cells

             The mature sperm looks like an impossible target for mutagens, but
             mature sperm respond to a significant number of mutagens.

             Lack of cell cycle in germinal tissue during stages of gametogenesis
             sensitive to mutation

             Spermatids through mature sperm in males and oocytes are arrested
             in meiotic prophase during the entire reproductive lifetime of females.

             Repair-deficient, germ-cell stages in males

             Fixation of mutations induced in late male germ cells occur in the egg
             after fertilization.  The  genotype of the  mother affects  frequency of
             mutations induced in the male.
                                   374

-------
 Chemically-induced Mutation:  Comparative Outcome in
 Somatic vs. Germ Cells
 Dr. James W. Allen gave the Second Part of the talk on "Chemically-induced Mutation:
 Comparative Outcome in Somatic vs.  Germ Cells."  Dr. Allen is from the Genetic
 Toxicology Division of the EPA in Research Triangle Park, North Carolina, U.S.A.

       DR.  ALLEN:     There are a  number  of  factors modulating  germ line
 mutagenesis.  I have organized them as: 1) those which affect the distribution  of
 mutagenic substances to the target molecules-such as physiological barriers in the
 testes, 2) germ cell properties which mediate the induction of damage-such as cell
 cycling, gene expression, various chromatin states, and 3) germ cell properties which
 may mediate the fate and potential recovery of damage-such as repair and selection.
       As many of these factors are operative at meiotic prophase, I would like  to
 focus on this particular stage.  It encompasses many of the elements that are
 considered to be protective of the germ line, as well as those which may involve
 unique targets for damage to the germ line. Some factors relate to accessibility of the
 mutagen to the meiotic prophase cells, and others to the crucial genetic processes
 that may be affected, or to the various outcomes. The Sertoli cell barrier is generally
 appreciated as being protective of germ line cells.  It has been suggested several
 times in this conference that,  perhaps one of the reasons why spermatogonial cells
 are showing relatively lower responses to mutagens, as  compared with some other
 cell types in spermatogenesis, is that  they might be protected by the Sertoli cell
 barrier. Actually, this should not be the case. Lonnie Russell has pointed out that a
chemical in the vascular supply  should have access to the spermatogonial and
                                    375

-------
 prelepotene cells, which  are in the basal compartment (Figure 5-4). It is not until
 leptotene-zygotene stage cells are moved through the tight junctions of the Sertoli
 cells to reach the adluminal compartment, where pachytene and later stages develop,
 that these cells become protected by the barrier. Sperm leave this relatively sheltered
 environment upon passing to the epididymis.
       There is not very much known about how protective this barrier is against
 different mutagens.  In general, it is believed to be effective against chemicals with
 high molecular weight, low fat solubility and/or high particle size.  However, a number
 of potential limitations of the barrier, even for these types of chemicals, have been
 pointed out (by Lonnie Russell and Irving Fritz). Germ cells may take up the chemical
 on one side of the barrier (in the basal compartment) , and then cross the barrier
 where that chemical may have a delayed effect on meiotic chromosomal processes.
 Or, damage may be carried across the barrier; that is, spermatogonial or preleptotene
 stages may suffer the chemical exposure effects and then carry aberrations across
 the  barrier to further  interfere with meiotic chromosomes,  for example, their
 segregation at cell division. Furthermore, some chemicals may directly damage the
 Sertoli cells to gain access to the germ cells, and perhaps also to secondarily inhibit
 nutrient supply to the germ cells.  Several agents have been found to specifically
 damage Sertoli ceils.
       In  some  collaborative studies, Ilse-Dore Adler and I studied the effects of
 bleomycin, which has a high molecular weight (>1400), on different cell types in the
testis.  We determined that this chemical  is very damaging to basal compartment
cells, as evidenced by subsequent detection of synaptonemal complex aberrations
in meiotic prophase.  Even after bleomycin exposure of zygotene and pachytene

                                     376

-------
              SERTOLI CELL BARRIER
                 Adlumlnal Compartment
                                     pachytene
                                         Sertoll Cell
                                    preleptotene
                                      spermatogonial
                   Basal Compartment
mutagen in vascular supply
                            (Based on LD. Russell, 1977; 1990)
                       Figure 5-4
                          377

-------
 stages of prophase, extensive induction of synaptic errors and breakage was
 apparent in synaptonemal complexes. This agent clearly caused damage evident in
 adluminal compartment cells.
        It was expected that breakage events in synaptonemal complexes would also
 be detectable as breakage events when those cells progressed to meiotic metaphase
 stages. However, in  both Ilse-Dore's lab and ours, only control levels of structural
 damage were observed.  Similar patterns of higher levels of damage observed in
 prophase (SC aberrations),  as compared to metaphase (chromosome aberrations)
 have been found for cyclophosphamide, acrylamide, and other chemicals tested.
 Apparently, much  of the damage that is evident In  synaptonemal complexes  at
 prophase disappears by the time those cells reach metaphase. The extent to which
 repair and/or selection are operative in  late  prophase to reduce the amount  of
 damage being transmitted further is unknown.  Some references made during this
 conference to germ cells appearing less sensitive; to "induction* of damage may, in
 fact, reflect instances in which the cells were very sensitive, but once damaged, they
 did not survive long enough  to be recovered at meiotic metaphase.  Imperfect
 synapsis is believed to be implicated in cell loss, and in aneuploidy of surviving cells.
 Electron microscopy studies of synaptonemal complex formation can be used to track
 homologous  chromosome  pairing through early prophase stages to complete
 synapsis  in pachytene stage  of mid-prophase.   The process of  crossing-over
 accomplishes chiasma formation, which holds the homologues together and helps to
 regulate their separation in cell division. Figure 5-5 shows synaptonemal complexes
at pachytene from a mouse treated with cyclophosphamide.  A normal-appearing
autosomal synaptonemal complex, with lateral elements of two homologues held in

                                    378

-------
Synaptonemal complexes at pachytene from
  a mouse treated with cyclophosphamide
                Figure 5-5
                  379

-------
 continuous register, is indicated by the asterisk (*). All of the remaining synaptonemal
 complexes in the cell reveal various types of structural and/or synaptic errors. An
 important question is-which types of damage will cause the cell to die, and which
 types may be compatible with cell survival but may  undermine recombination and
 disjunction?
        It is generally believed (with evidence from mutant mouse strains and humans)
 that improper synapsis and recombination of the XY  chromosomes can lead to cell
 death.  In this cell, the XY pair is sufficiently abnormal to  be unrecognizable.  In
 surviving cells, failed recombination and chiasma formation in any of the bivalents may
 lead to homologue nondisjunction.  This concept is supported by the recent work of
 Warren  and  Hassold  and  others studying individuals with trisomy conditions,
 specifically Down syndrome and Klinefelter syndrome. They  have reported evidence
 of  failed parental  meiotic recombination in the  extra chromosome of aneuploid
 offspring.
       I think it is important to understand the risks associated with mutagen effects
 on  processes of synapsis and recombination.   However,  it is also important  to
 consider this in light of new views about meiosis.  We have all been taught that
 homologous  chromosomes synapse so that they can cross-over.  Recently, there
 have been some very interesting findings in yeast (as reviewed by Hawley and Arbel)
 that indicate this perceived order of events in meiosis may, in fact, not be correct.  It
 was found that double-strand  breakage serving to  initiate recombination actually
 precedes synapsis.  The alignment of chromosomes in synapsis  is believed to
function in the maturation of recombination intermediates necessary to complete the
exchange and form chiasmata.  There is increasing evidence to suggest that proper

                                    380

-------
 disjunction  depends upon crossing-over in connection with synapsis and  SC
 formation. Consequently, mutagen-induced interference with any of the steps in this
 chromosome orientation and exchange process may pose an aneuploidy risk.  It is
 very apparent that synapsis is  severely perturbed  following mutagen exposure.
 However, it is not clear which types of abnormalities are significant for undermining
 recombination.  It could be that they all are-we would like to know more about the
 importance of such damage.
        In the Armenian hamster XY bivalent, homology and synapsis is limited to the
 short arms.  We know from earlier bromodeoxyuridine  labelling-differential staining
 studies that most cross-overs occur in a proximal segment of the synapsing region.
 This provides an animal model for analyzing synaptonemal complex anomalies that
 may impact, or otherwise involve, crossing-over.  We have found that there are, in
 fact, some curious types of abnormalities that occur in that region.  For example,
 breakage across both lateral elements to result in complete scission of the SC has,
 thus far in our analyses of XY chromosomes, only been observed in the cross-over
 region. Our interpretation is that  both homologues may be broken at the same site
 as a result of disruption to cross-over exchange-leaving both chromosomes involved
 in the exchange with double-strand breakage. If this type of abnormality predisposes
 aneuploidy, it represents a germ-line specific mechanism; there is no regular
 counterpart process in somatic cells subject to such interference. The same may be
 said of induced  alterations to synapsis, for example,  "point synapsis",  in these
 chromosomes which can result from radiation or chemical exposure. It could cause
 recombination to fail, and  signify  a germ-line specific mechanism  predisposing
aneuploidy.

                                    381

-------
       The unique biology of germ cells affords many possibilities for germ cell-
specific pathways of mutagenesis.  The consequences may be equally specific for
the germ line;  some examples  are infertility, germ cell tumors including those
characterized by aneuploidy, and abnormal sexual differentiation due to anomalous
meiotic exchange. These fundamental points are not always taken into account in
efforts to extrapolate from mutagenic mechanisms and outcomes associated with
somatic cells to those associated with germ cells.
                                   382

-------
 Qualitative And Quantitative Results From Short-term Tests

 Dr. Michael D. Waters presented "Qualitative and Quantitative Results From Short-term
 Tests."  Dr. Waters is from the  Health Effects Research Laboratory at the EPA in
 Research Triangle Park, North Carolina, U.S.A.

        DR WATERS:  The main source of the information that I will discuss is the
 EPA/IARC Genetic  Activity Profile  (GAP) Database, which  appears in the  (ARC
 Monographs. The figure is an example of a genetic activity profile  (Figure 5-6).
        We are looking at lowest effective doses for positive test results along the top,
 and the highest ineffective doses for negative test results along the bottom (positive
 up, negative down). That is the way the profiles are plotted.  In this particular plot,
 the tests are organized in a phylogenetic sequence; we also do an endpoint plot.
        Kerry Dearfield  alluded to the EPA Office of Pesticide Programs Mutagenicity
 Testing Guidelines in his presentation. What we were interested in doing was to look
 at the performance of the short term tests, in particular, those being required by the
 EPA under these Guidelines.
                 \
       We also have been considering international testing requirements and together
 with  Bryn  Bridges  recently  published a paper that  is  an assessment of the
 performance of short-term tests relating to the UK guidelines.
       This is the scheme that is being used in the U.S. for pesticide registration
 (Figure 5-7).  The initial three-tier test is required for all pesticide chemicals.  It is
 actually  four tests,  if  you consider the L5178Y gene mutation  assay as being
 performed with an assessment of small and large colonies,  or in conjunction with an
in vitro cytogenetics assay.

                                     383

-------
    flCRYLRMIDE
                             79-06-1
                                                   8-JUN-93

8-
6-
4-
2-
-2-
-4-

















/I











01
IK


jj-






	 1"""""


AA









.








S
1
C





t
1










1
t
H








I
C
L









I 1
K SH [ED
1 II IB UK
v IN i s n




















~ SB I
mm r
csi r

PROKM VOTES
LOVER CUKM.
PLANTS I1HSCCT
rftHMHlS 1M VITRO
HUMMS If/Hj l*m<
IN VITRO j 1
"~
-

!
||
ji

-
-


-
-
c
i
tLS IN VIVO MUr«WS |
IN VIVO 1
                      Figure 5-6
    OPP MUTAGENICITY TEST GUIDELINE
Salmonella   +     In Vitro
                Gene Mutation
+      In Vivo Bone
   Marrow Cytogenetics
      • Aberrations
            or
      • Micronuclei
        In Vitro Gene Mutation (choice):
        (a) Mouse lymphoma L5178Y cells, TK locus,
          small and large colonies
        (b) Chinese hamster ovary cells strain AS52
        (c) Chinese hamster ovary or lung fibroblasts
          with the appropriate in vitro test for clastogenicity
                       Figure 5-7

                          384

-------
       By way of example, to follow up a positive response for chromosomal effects,
 the dominant lethal test or the heritable translocation test would be performed.  Or for
 positive gene mutation, one would undertake studies to look at the interaction of the
 chemical with gonadal DMA, to be followed by a specific locus test.
       This is the battery (Figure 5-8), that is required by the EPA Office of Toxic
 Substances to be used by industry in the evaluation of Industrial chemicals according
 to specific test rules.
       Our germ cell database was assembled primarily from the EPA/IARC GAP
 database; this was supplemented with data from the EPA Gene-Tox database. Gene-
 Tox, was established to review test systems, whereas GAP was set up to facilitate
 evaluation of chemicals.  Sometimes either of the two of these databases  may be
 somewhat out of date. So, by combining the two, we hoped that we would get most
 of the available data on germ cell mutagens.
       We started out with 64 chemicals  (Figure 5-9) and reduced the number to 30
 confirmed positives.  We subsequently reviewed the literature for putative germ cell
 nonmutagens, in order to create a negative data  set. There were 26 of those. This
 overhead just describes how that process occurred. With the germ cell data in hand,
 we looked at the corresponding mammalian somatic cell and in vitro data and then
 we updated all of it.
      These are the test codes for the germ cell assays (Figure 5-10), including the
 specific locus tests, dominant lethal tests, and chromosomal aberration assays, in
various germ cell stages.
      This is the confirmed germ cell positive database (Figure 5-11).  This is just to
give you a sense of  how much  information is in the database. Here is bacterial

                                    385

-------
      OTS MUTAGENICITY TEST SCHEME
Salmonella   +    In Vitro
             +     In Vivo Bone
Gene Mutation     Marrow Cytogenetics
                  •  Aberrations
                       or
                  •  Micronuclei
                  Dominant Lethal

                   (DLM, DLR)
Interaction with Gonadal DNA

 (CCC, CGC, CGG, COE)
Specific Locus (SLP, SLO)     Heritable Translocation (MHT)

  • Visible

  • Biochemical



                  Figure 5-8
                     386

-------
   HOW WAS THE DATABASE ASSEMBLED?

• Data were reviewed for 64 chemicals with evidence of germ
 cell mutagenicity; 40 from the EPA/IARC GAP database and
 24 from the GENE-TOX database.

 The germ cell data were updated and the list of chemicals
 was reduced to 30 confirmed positives.

• Data for the GENE-TOX chemicals (10) were adapted to the
 EPA/IARC GAP format and the datasets were combined.

1 Chemicals without data for specific somatic cell tests were
 updated from the recent literature.

1 The updated dataset was used for the analysis presented here.

1 An anatogus procedure was followed for preparing the dataset
 for the 26 germ cell nonmutagens.
                    Figure 5-9
          GERM CELL MUTAGEN TESTS
TEST CODE  DEFINITION
   Mouse Specific Locus Tests
   SLP    Post-Spermatogonial
   SLO    Other Stages

   MHT    Mouse Heritable Translocation Test

   Dominant Lethal Tests
   DIM    Mice
   DLR    Rats

   Chromosomal Aberration Tests
   CCC    Spermatocytes Treated In Vivo, Spermatocytes Observed
   CGC    Spermatogonia Treated In Vivo, Spermatocytes Observed
   CGG    Spermatogonia Treated In Vivo, Spermatogonia Observed
   COE    Oocvtes or Embryos Treated In Vivo
                    Figure 5-10

                       387

-------
    Short-term test results for agents positive in more than one germ cell assay
    CTCLOPHOSPHAUIOE
           MYIBMN
       ETHVIENE OXIDE
         WTOUTCMC
              ENU
              UNU
         ACmiAMIDE
  TWETHY1ENEMQAM1NE
      HYCAKTHONEMS
DIBROMOCHLOBOPROPANE
          THOTB>A
        TMABOUONE
       6OPROPYLMS
  TWMETHYL PHOSPHATE
             TEPA
    6-M6BCAPTOPUWNE
          OSPLATW
         BlEOMYCm
          ETHANOL
      OIETHYLSUIFATE
   GERM CELL
   MUTATION
SS CCCC OOM
LL CQQO LUH
PO CCOE MHT
•O  O*  ••
                   O«
                   • O »O •
                   •O
O«
••  O*
           •
                  eo  o  •
                             •
oo
      •    •
     o  •
     o  •
OO •
      o* ••
                                    BACTERIAL
                                    MUTATION
                                  SSSSSSSSS
                                  AAA A AAAA A
                                  234S7BCOS
                                      •ooo»»
                   •o o»
                                      •ooo»
                       oo*
                   ooooo
                                     oooooe
                   •  o***
                   oooo«
                   •ooo««
                   ooo*»*
                       ooo
                   oooooo
                                       •oo*«
                                   MAMMALIAN CELLS
                               MUTATION        ABS.
                             GOGOOOGQCG
                             CCS 9M5S 1 1 H
                             LOHOLT1AHT
                              •o
CCCCCCCCCC
Mill) HHH 1
CMRSTAFLTH
                                                o
                                                 o  o
                 BONE
                MARROW
MMMMMCC
VVVVV BB
MRCAH AH
                                                          o«
                                                          oc
                                         • Positive
                                         C Conflicting
                                    O Negative
                                    F • Positive in females
                                    Figure 5-11
                                       388

-------
 mutation, here are the mammalian cell assays, both mutation and chromosomal
 aberration, and here are the bone marrow assays, the mouse micronucleus test, and
 chromosome aberrations assay. The positive responses are in black, open circles are
 negative, conflicting calls are the half-shaded figures. "P is a positive response in a
 female dominant lethal assay, which is in the DIM column.
        This is the positive data set (Figure 5-11), where you have a confirmed positive
 response in more than one germ cell assay. The next illustration (Figure 5-12) shows
 a confirmed positive response, but only in one germ cell assay, and again you see
 the corresponding bacterial, mammalian cell and bone marrow data.
        To segregate this information for  your  evaluation (Figure 5-13), we have
 ordered the chemicals most-to-least tested.  Please look at the consistency of the
 black dots.   It is  immediately apparent from this kind of assessment that the short
 term in vitro tests do well in picking up the confirmed germ cell mutagens. I will show
 you the actual numbers in a later slide. One exception, ethanol, is not picked up.
 But, for the  most part,  the short-term tests are quite sensitive.  Another  notable
 exception, hexametapol, is not detected in the in vitro tests but is picked up in the
 bone marrow assay.  It may be hard to appreciate,  but the bone marrow assay is
 doing quite well here, and as we look into the other negative data, you will see that
 it is doing even better.
       Now examine the negative data set (Figure 5-14). It became critically important
 if one were going to evaluate the performance of these  short-term tests to build a
 negative germ cell data set.  As far as I know, this has not been done previously. It
 is a challenging task, because, one  has to review a lot of data and confirm that the
data is negative.  We used the Gene-Tox criteria to call these results negative. There

                                     389

-------
Short-term test results for agents confirmed positive in one germ cell assay


BENZD(A)PYRENE
ADMAMYCM
NrmOGEN MUSTARD
METHOmEXATE
HEXAMETAPOL
METEPA
METHYL CHLORIDE
GERM CELL
MUTATION
SS CCCC DOM
LL COQO LLK
PO CCQE Ufl T
O •
• F
•
•
•
9
•
BACTERIAL
MUTATION
8S888SSSS
AAAAAAAAA
234 S 7»S OS
••••••
oo»*>»«>
•ooe*
ooooo
oooooo
o o»
• •
MAMMALIAN CELLS
MUTATION ABS.
aooooaaooQ cccccc cccc
CCB9MSS 1 1 H 1 1 1 1 1 1 HHH 1
LOHOLT1AHT CMRSTAFLTH
••• •••• • •
•• • • • • •
• • •••
• • • ••
OC 0 0 •

•
BONE
MARROW
MMMMMCC
VVVVVBB
MRCAHAH
«
• • •
•
•• ••
•• •
0

                                     • Positive
                                     C Conflicting
O Negative
¥ -Positive in females
                                Rgure 5-12
                                    390

-------
     POSITIVE GERM CELL VS. SOMATIC CELL TESTS
   • Positive
   O Negative
   O Conflicting
     CYCLOPHOS.
        MYLERAN
     ETHYLENEOX
           MMS
           EMS
     MYTOMYCINC
           EMU
           MNU
     ACHYLAMIDE
           TEM
    PROCARBAZINE
 OIETHYL SULPHATE
        7HIOTEPA
     TRIA3QUONE
            IMS
            TMP
           TEPA
           HMS
           DBCP
6-MERCAPTOPURINE
       CISPLAT1N
      BLEOMYC1N
        ETHANOL
           B(A)P
      AORIAMYQN
      N. MUSTARD
            MTX
    HEXAMETAPOL
         METEPA
       METHYUCL
GERM
CELL
H  NH
  BONE
MARROW.
MM CA.
IARC
EVAL
                       Positive In specific-locus tests and MHT
                        Positive in MHT or in nonherltable tests
                       Single germ cell test confirmed positive
                                        2A
                                        2A
                                        2A
                                         3
                                        2B

                                         3
          * 1: Carcinogenic to humans; 2A: Probably carcinogenic to humans; 2B: Possibly
            carcinogenic to humans; 3: Not classifiable as to its carcinogenicity In humans
                         Figure 5-13
                            391

-------
Short-term test results for agents negative in more than one germ ceJI assay
      MCHLORVOS
            PCS
ETHYLENEMBROMIOE
    ACRYLONtTTVLE
  CYCtOHEXYLAMINE
   MTROFURANTON
 2.4-OlNrmOTOLUENE
    FORMALDEHYDE
CALOUM CYCLAMATE
        CAFFEINE
       PARATMON
           2.4.S.T
        CADMIUM
       ARSENIC +3
        BENOMYL
  PROPYLENEOXnE
    VINYL CHLORIDE
 METHYL PARATHON
GERM CELL
MUTATION
SS CCCC DDM
LL COQO LLH
PO CCQE MHT
OOO O
o o
oo
o o
oc o
o o
oo
o o
o ooo
ooo
O 0
oo
o oo
ooooo
o ooo
o o
o o
oo
oo
o o
BACTERIAL
MUTATION
SSSSSSSSS
AAA AAAAAA
234 5 7 »» OS
•OOO»»
OOOOO
• •oo*»*
o» •000*0
ooooo
ooo***
ooe**
• *ooo**e
o oo
oooooo
ooooo
ooooo
ooooo
• *ooo*o
oo oo
o ooooo
o ooooo
•0 O*
• *ooo*e
•ooo*
MAMMALIAN CELLS
• MUTATION ABS.
QOQQGQQQQG CCCCCCCCCC
CO »M55 MH 1 1 1 1 1 1 HHH 1
LOHOLT1AHT CMRSTAFLTH
O • • • O
O •
• • • •
• *e** • o o
o • e
• o o
o • o
• • ••
o o •
ooo 00*0 •••

• o •
•
• • o •
• • • o e
o o • • •• •
• • •
• • •
•••
• •
BONE
MARROW
MMMMMCC
VVVVVBB
MRCAHAH
0 0
O O
0 O
O O
O
o
o
o» o
o e
ooe e
e
• o
o •
e
• e
• e
• e
e
• ••
•• •
                                           • Positive
                                           e Conflicting
O Negative
                                      Figure 5-14
                                          392

-------
 are a number of compounds that are surprisingly negative, but this is the negative
 data set. Again, we are looking at the black dots. That is, what the short term tests
 are doing wrong? We are also interested, especially, in this bone marrow column.
        Here is the residual negative data set (Figure 5-15). These are the chemicals
 that only have a single confirmed negative response.  In terms of the strength of the
 evidence, you see that the negative data set is much weaker. We do not have any
 heritable assays represented, and we have much less data for the non-heritable
 assays, but all of these test results are in agreement. We have a consistently negative
 data set.
        Now we  have reduced the negative data set (Figure 5-16), hi order to give you
 the ability to look quickly at the germ cell column. This is the standard by which we
 are measuring  the  other short-term  tests.   And now we want to examine the
 performance characteristics of the various short term tests. So, (in the darker shaded
 area)  we have bacterial gene mutation,  mammalian  cell gene mutation  and
 chromosome  aberration  and then bone marrow micronucleus, and  chromosomal
 aberrations.   Finally,  in  the  last column  there is  the  IARC  evaluation of the
 carcinogenicity  of these  compounds.  Many  of them are quite clearly animal
 carcinogens.
       The horizontal line that is drawn across the middle, of this figure is to show you
where the bone  marrow assays agree with the negative germ cell response.  Then,
in the bottom half, you have either a bone marrow call that is either in question, or a
clearly positive call.  If you look at the bone marrow data, I think the results are a bit
more encouraging, than in the case of the in vitro data, where you have quite a few
black dots,  at the top and at the bottom of the slide.

                                     393

-------
 Short-term test results for agents confirmed negative in one germ cell assay
        MEUMN
t.l-OIMETHYUmMAaNE
   EPtCHLOWXYDWN
      CHLOKMNE
         KXPET
 HEXACHLOHOBENZENE
GERM CELL
MUTATION
SS CCCC DDM
LL CGOOLLH
PO CCOE MR T
o
o
o
o
o
o
BACTERIAL
MUTATION
SSSSSSSSS
AAAAAAAAA
234 57«»OS
oooooo
o 000*00
•••••
ooooo
••oo »o
oo oo
MAMMALIAN CELLS
MUTATION ABS.
QQQQOQOQaQ CCCCCCCCCC
CC89MS5 1 IH 1 1 1 1 1 1 HHN 1
LOHOLT1AMT CMRSTAFLTH
•
•O
•• ••• • •
o» • oo
• • ••
o
BONE
MARROW
MMMMMCC
V VVVV BB
MRCAH AH
O
O
o e



                                 • Positive
                                 e Conflicting
O Negative
F - Positive in females
                             Figure 5-15
       NEGATIVE GERM CELL VS. SOMATIC CELL TESTS
                 PCB
             CAFFEINE
             CADMIUM
             DIELDRIN
     CYCLOHEXYLAMINE
    2.4-DINrTROTOLUENE
       NrTROFURANTOIN
  1.1-DIMETHYLHYDRAZ1NE
        FORMALDEHYDE
   ETHYLENE DIBROMIDE
        ACRYLONrmiLE
          DICHLORVOS
            PARATHION
    CALCIUM CYCLAMATE
     EPICHLOROHYDRIN
            ISONIAZIDE
               ^4.5-T
           ARSENIC +3
         THEOBROMINE
             BENOMYL
      PROPYLENE OXIDE
        VINYL CHLORIDE
     METHYL PARATHION
  HEXACHLOROBENZENE
          CHLORDANE
              FOLPET
                             Figure 5-16

                                 394

-------
       And now, we come to the performance characteristics of the short-term tests
 (Figure 5-17). Since we have 30 agents positive in germ cell assays and 26 agents
 negative, we are in a good position to compare these performance characteristics.
 I am going to point out the numbers for sensitivity and specificity. These numbers
 are of primary  concern,  but since prevalence was not bad, we also  calculated
 predictivity and  accuracy.
        Sensitivity and specificity are as follows: 1) bacterial  mutation, 79% for
 sensitivity,  specificity  50%, 2) sensitivity for  mammalian cell gene mutation, 96%,
 specificity 21 % and 3) chromosomal aberration, sensitivity 92%, specificity 11 %. It has
 been a concern, on the part of many people, that in vitro chromosomal aberration
 assays are giving too many false positive responses, and this evidence indicates that
 their specificity is quite low.
       If, on the other hand, you examine the in vim bone marrow assays, you have
 high sensitivity   (96%), whether  you  use micronucleus  test,  the  chromosome
 aberrations assay or either of the two.  Specificity is 63% for the micronucleus test,
 64% for chromosomal aberrations, and if you use either, it  is 55%. The number of
 chemicals tested is small, but the indication is that clearly the bone marrow assays,
 either the micronucleus or the chromosomal aberration assays, are performing much
 better, in terms of specificity than are the in vitro systems.
       Since we had the lowest effective dose (LED)  information, we wanted to
examine how the tests under investigation were performing, relative to one another,
in a quantitative  sense. We decided to regress comparable bone marrow tests, and
then to regress comparable germ cell tests to see whether  the two data sets would
                                    395

-------
Performance of Somatic vs. Germ Cell Tests
for Agents Confirmed in Germ Cell Studies

           IN VITRO TEST SYSTEMS
GERM "*"
cat
MUTATION _
PreetlctMty
GERM +
ffl I
MUTATION _
BACTERIAL
MUTATION
+ " -
23
13
6
13
64% 68%
MCRONUd£US
+
22
7
1
12
Pndletmty 78% 92%
Sens*
78%
Specil.
50%
Accur.
65%
INVIV
Sensit
96%
Specif.
63%
Accur.
81%
MAMMAUAN
MUTATION CHROM.ABS.
+ - H- -
22
15
1
4
59% 80%
0 BONE MARROV
CHROM. ABS.
+ -
22
5
1
9
8f% 90%
Sensit.
96%
Specil.
21%
Accur.
62%
(V TESTS
Sensit.
96%
Specil.
64%
Accur
84%
22
16
2
2
58% 50%
EITHER
+ -
27
a
1
11
75% 92%
Sens/I.
92%
Specif.
11%
57%
Sensit.
96%
Specil.
55%
Accur.
79%
         Figure 5-17
             396

-------
 be in quantitative agreement in terms of LEDs. This (Figure 5-18) was the regression
 of the mouse micronucleus test LEDs versus the mouse bone marrow cytogenetics
 assay LEDs.  The data base has been culled for only mouse data, and only IP
 injections. So we are looking at a common route, ail in the mouse and the correlation
 coefficient is 0.77.  There are only 11  chemicals in common between the two tests.
 We do not have a large data set here, but for other chemicals not tested in germ cell
 assays,  we still have  a good quantitative correlation between the bone marrow
 chromosome aberration and  micronucleus tests.
       These are compounds for which we do not have germ cell data (Figure 5-19)
 and the correlation coefficient is still good, so  I think we  have some degree of
 confidence that these two types of bone marrow assays are giving us the same sort
 of quantitative LED data.
       Here we are looking  at the dominant lethal test and the mouse heritable
 translocation test (Figure 5-20) - two tests that you would expect to correlate quite
 well in terms of LED, and, in fact, they do. The correlation coefficient is 0.92, and the
 slope is 0.98.  So, I think this gives one confidence that the dominant lethal test and
 mouse heritable translocation test are yielding similar data, in terms of LED.
       We look now at the dominant lethal test in the mouse, versus the micronucleus
 test in the mouse, performing the same regression (Figure 5-21). This solid line is the
 actual  regression line,  the dotted line is what you would expect for a one-to-one
 relationship. So, you see that the mouse micronucleus test is responding at a slightly
 lower dose, on the average (as indicated by the Y intercept), than the dominant lethal
test. For a screening assay (such as the micronucleus test), this is quite desirable.
There are 18 chemicals represented in this pair-wise regression.

                                    397

-------
LEDs of Germ Cell Mutagens Tested in CBA vs. MVM
                 MVM Dose (log(mg/kg))
Correl. Coeff. = 0.77

     Slope = 0.88


  Y Intercept = 0.11

         N = 11
                    (D
                    (2)
TEM
                            A Triaziquc ne
  A
PRO
     MMC
             AA
               EM:
                MMS
                A EMU
                           (1)    0     1     23
                               CBA Dose (log(mg/kg))
                      Figure 5-18
         LEDs of Agents Tested in CBA vs. MVM
             (Not Tested in Germ Cell Assays)
                  MVM Dose (log(mg/kg))
4
3
Correl. Coeff. = 0.75
2
Slope = 0.89
1
Y Intercept = 0.05
0
N = 19
(1)
(2)
P




>
XI


i

X
t



XJ




Y
14


A
-s
' A



A ./
A S
7*.
k




!) (1) 0 1 2 3 4
CBA Dose (log(mg/kg))
Figure 5-1 9
398

-------
LEDs of Germ Cell Mutagens Tested in DLM vs. MHT
              MHT Dose (log(mg/kg))
                  4
Correl. Coeff. = 0.92
      Slope = 0.98
  Y Intercept = 0.05
             15










/






TEPA

/ATE
/A Tnajiq







z
L/*ThioTE
»
one



CPA
IMS
INU* Y
/A
/MMS
W
'A



IMP
PRO ./
y
^/ENU
EMS
AEO





^/










                          DLM Dose (log(mg/kg))
                   Figure 5-20
LEDs of Germ Cell Mutagens Tested in DLM vs. MVM
                MVM Dose (log(mg/kg))
Correl. Coeff. - 0.72


     Slope = 0.98


 Y Intercept = -0.32


         N = 18
                   -2^
ThioTEP.  A X
                            A Tnazit jone
        MylemA.*
     / MMC
              EMS ^
                
-------
       What we did next (Figure 5-22) was to select the most sensitive stage of the
 specific locus test and regress those results against the results of the mouse heritable
 translocation test, and this is the line that you see.  These two tests are quantitatively
 quite comparable in terms of LED even though they represent different endpoints.
       In summary, if one wants high sensitivity, short-term in vitro tests provide that
 sensitivity at low cost. Obviously, you can test a lot of chemicals with these assays.
 But if you want to discriminate potential germ cell mutagens, you need to move
 quickly into the bone marrow tests. They seem to be  quite  useful in a qualitative
 sense, and, from the present data, also in a quantitative sense in predicting what the
 response might be in the germ cells. And then the germ cell assays themselves seem
 to be  correlating quite well, in terms of lowest effective dose. This argues for the
 value of the dominant lethal test as a relatively inexpensive  confirmatory germ cell
 assay.
       What kinds of agents would we want to examine in  the future.  I think we
would want to add more data particularly for those chemicals that were positive in a
single  germ cell study (Figure 5-23). Mike Shelby is going to talk about some of the
agents on this list. He will go into some of the details about why we might want to
look further at some of these agents.
                                     400

-------
LEDs of Germ Cell Mutagens Tested in MHT vs. SLT
                 SLT Dose (log(mg/kg))
^
3
Correl. Coeff. = 0.92
Slope = 0.88 2
Y Intercept - 0.27 1
N = 9
0
(D
(1




/ATEM


X
xL


MMS EMS
*
X



CPA ./
A >/
kX PRO
«NU ENU


/





) 0 1 234
MHT Dose (log(mg/kg))
                    Figure 5-22
                       401

-------
AGENTS POSITIVE IN MAMMALIAN GERM CELL TESTS
            Heritable (H)   NonHeritable (NH)
• Positive
O Negative
e> Conflicting
F Female Pos.


CYCLOPHOS.
MYLERAN
ETHYLENEOX.
MMS
EMS
MYTOMYCINC
ENU
MNU
ACRYLAMIDE
feu
ICM
PROCARBAZINE
HHYL SULPHATE
THIOTEPA
TRIAZIQUONE
IMS
TMP
TEPA
HMS
DBCP
=RCAPTOPURINE
aSPLATIN
BLEOMYON
ETHANOL

B(A)P
ADRIAMYCIN
N. MUSTARD
MTX
HEXAMETAPOL
METEPA
METHYL CL
S S M
L L H
POT
C C C C
C G G O
C C G E
Positive in specific-locus tests and
• *
• O
0 •
• O
• o
0 •
• •
•
• •
: • •
• O • •
0 •

• O •

• 0 • •
O •

• 0 •
D D
L L
M R
MHT
• •
• •
• •
•
•
•
•
•
• •
:
Positive In MHT or in nonheritable tests
e
e
•
•
0
e o
O 0
o o o



• • •
o •
o •


o •
•
•
• e •
• • •
o •
•
•
•
•
•
F •
0 •
•
F
F
• •
Single germ cell test confirmed positive
0






• F
                 Figure 5-23
                   402

-------
 Chemicals  for Future  Study  off Heritable   Genetic  Damage  In
 Human Populations

 Dr.  Michael Shelby presented "Chemicals for Future Study of Heritable Genetic
 Damage  in Human  Populations*.  Dr. Shelby is  from  the National Institute  of
 Environmental Health Sciences in Research Triangle Park, North Carolina, U.SA

       DR. SHELBY: One of the concerns that has been expressed throughout this
 meeting is the fact that there are no known human germ cell mutagens. In fact, we
 do have evidence of human germ cell  mutagens.  One of them is ionizing radiation.
 Brewen, Preston, and Gengozian published a paper in 1975 on cytogenetic analysis
 for translocations in irradiated testes of humans.  Rene Martin's work using human
 sperm and  hamster egg fusion experiments has also shown  that structural and
 numerical chromosome aberrations result from testicular radiation.
       More recent studies, again using  Rene Martin's basic technique in cancer
 chemotherapy patients, have clearly  shown chromosomal aberrations in human
 sperm.   So, we do have evidence  of  human germ cell mutagenicity, not the
 transmission of  genetic  damage expressed as genetic disease in subsequent
 generations, but  there can be little doubt that such effects will result.
      The purposes of animal germ  cell mutagenicity studies include the simple
 identification of germ cell mutagens. Secondly, we study the mechanisms by which
these mutations are induced, and finally, we generate data for risk estimation. Many
people lose track of these different types of studies and they try to do risk estimation
based on experiments that are conducted to identify germ cell mutagens. I think that
is a problem many of you have run into here today.
                                   403

-------
       To do germ cell mutation risk assessments, there are several factors that have
 been brought up that are typically not a part of germ cell  mutagen identification
 experiments. One is the gender that is affected; we know now that there are specific
 mutagens for each gender. The cell stages affected, the dose-response, dose rate
 effects, and the types of mutations induced whether you want to think of these as
 deletions versus base pair substitutions or dominant lethal mutations versus specific
 locus mutations, are all critical when one is trying to estimate genetic risk.  As was
 discussed some yesterday, I think it is important that we look at reproductive outcome
 whenever possible.   Our standard mutational endpoints are very informative but
 reproductive outcomes provide an additional, medically relevant, endpointthat we can
 review.
       What chemicals should we study in humans in our efforts to toy to identify
 germ cell mutagens?  I  have come up with three exposure situations that seem
 reasonable to me.
       First are the  chemotherapy agents in which you have  a  population of
 individuals who  have been exposed to high,  acute, defined exposures to known
 mammalian germ cell mutagens.  A study is now underway to investigate the health
 outcomes of some 25,000 people who have survived childhood cancer, who were
 diagnosed with cancer before the age of 21 and have survived more than five years.
The primary objective of this study is to look at the health outcomes in these 25,000
individuals.  But, there is a big opportunity to  do studies on the  offspring of these
people where we have both the parents and the offspring available for evaluation.
      The second situation  involves  high  occupational or accidental exposures
because they provide unique  opportunities to study human populations. We hope

                                    404

-------
 those do not occur; but if they do, we should take advantage of them.  Finally, we
 should study large populations exposed to low doses of potent germ cell mutagens,
 a situation which, theoretically has the greatest potential for affecting the human
 population.  Fortunately, as far as we know, there are no potent germ cell mutagens
 to which large segments of the population are exposed.
        These are the chemicals that I would suggest are the primary ones we should
 study in human populations. They are all cancer chemotherapy agents. Chlorambucil
 is positive in all the endpoints we have looked at. Post-meiotic stages in male mice
 are affected.  It is biologically active in humans. In addition, these are all human
 carcinogens. That sometimes detracts people's interest in germ cell mutagens but
 from our point of view,  it provides evidence of their biological activities in humans.
 Because they are human carcinogens, they must be metaboiically activated or direct
 acting in humans.  Chlorambucil is used in breast and ovarian cancer chemotherapy,
 and we think it would be very important to study it in females.
        The second chemical, melphan, is positive again in dominant lethal, heritable
 translocations, specific  locus, and affects all germ cell stages in the male and is
 positive cytogenetically in humans.   All of these compounds down to tresulphan,
 which is potent in in vitro and in vivo mutagen tests and an alkylating agent, is very
 similar to myleran and busulphan. We have been studying these chemicals in our
 program, mainly through the work of Waldy Generoso and Lee Russell, for the reason
 that we think if we have a sound mouse germ cell mutagenesis database, we will be
 prepared when the time comes for the studies to be done in humans. We will know
whether the female or male is affected, and what germ cell stages are most sensitive.
These are factors that are sometimes ignored by epidemiologists.  If you look at the

                                     405

-------
 papers in the literature, they often give no consideration to what germ cell stage is
 affected and how long a period has elapsed between exposure and conception. But
 in this case I hope that we are at least partially prepared to work in support of the
 human epidemiology studies.
       So with regard to the title of the talk here today, that is the one slide I have,
 chemicals for future studies in  humans.   There is another interesting group of
 compounds, none of which are chemotherapy agents except for azathioprine. Again,
 it is a human carcinogen.  It is a rodent clastogen and there is a single report of
 mouse dominant lethal positive. All these results were thanks to Mike Waters and the
 massive database he has on chemical mutagens.  Benzene, to which there is the
 most widespread exposure to a known potent clastogen and human carcinogen, is
 virtually lacking in animal germ cell mutagenidty data. Benzene is, I think, well worth
 studying and it simply has not been done.  N-butyl glycidyl ether and diazepam, are
 potential germ cell mutagens.  There are  reports of  bone marrow micronucleus
 positive results in animals with diazepam so here may be one case of widespread
 exposure in the population of what might be a potent germ cell mutagen; we just do
 not have the data to say whether it is or not.
       Ziram is in the same situation.  There is also a smattering of information in the
 literature that ethanol may affect genetic integrity of germ cells and there is certainly
widespread chronic exposure to this chemical.  So, there is a another partial list of
compounds on which I think we need to extend the animal data base before we
consider these for studies in humans.  If not before, at least concurrently with the
human studies that might be conducted.
                                    406

-------
       There are few special topics that I had to throw in here that are not part of my
 title. DBCP remains enigmatic in the fact that it sterilizes humans, it induces dominant
 lethals in male rats, and it seems to have no effect on the male mice. DBCP is one
 issue that should be sorted out in our literature I believe.
       We have Dr.  Generoso's female specific germ cell mutagens with regard to
 dominant  lethal  induction.  Additional  work certainly  needs  to  be  done there.
 Aneuploidy work is going ahead very well I think. A big study is being conducted in
 Europe.  There are  such  chemicals  as  chloral hydrate which are used to sedate
 children, and  trichlorfon which I  have another slide on  here, that was reported to
 induce a high frequency of abnormal birth outcomes in a small village in Hungary.
 I do not know if you all have seen that paper or not; I have a couple of copies with
 me, but it is some of Andrew Czeizel's work in Hungary and some very impressive
 data.
       Another topic involves the triplet repeat mutations that have been shown now
 to be associated with at least six human genetic diseases.  We do not know if these
 are chemically inducible because, as far as I know, we do not have a mouse or an
 in vitro model  to study it and a mouse model would be very convenient to have to
 study this type of genetic alteration. Finally what role should transgenic animals play?
 There  has been a world of hype about what the transgenics could do for us and  I
 think it is a great deal, but their role in studying germ cell mutations has not yet been
 defined and I think this is desperately needed at this time.
       I will show you one set of Rene Martin's data because she refers to it in her
paper  as important  to those who  are  considering  use  of the  parallelogram in
assessing genetic damage. This is simply the percent of aberrant cells. It is from 13

                                    407

-------
 patients on radiation therapy where she looked at chromosome aberrations in these
 patients' lymphocytes and in their sperm using fusion of sperm and hamster eggs.
 These are the months after radiation therapy running from one month to five years.
 These men were azoospermic for about a year after they were exposed to radiation
 therapy. So, there was no sperm to do cytogenetic analysis on, but if you look at the
 24 month to five year samples, you see that the data are not real tight, but they seem
 relatively steady in the sperm through that four year period while in the lymphocytes,
 the frequency goes down with each year of sampling. If you try to use this kind of
 data in a parallelogram, just an example, at this point it looks like you have a one to
 one ratio of aberrations in lymphocytes and sperm. If you look out here you have five
 times as many sperm with aberrations as you do in the lymphocytes. It is the kind
 of  data that simply raises a red flag that you have  got to think about using the
 parallelogram to understand what is happening to the lesions you are looking at.
       This is the Czeizel study, the paper on  environmental trichlorfon and the
 cluster of congenital abnormalities.  In this small village over a two year period, there
 were 15 births.  Eleven of them or 73% had congenital abnormalities.  There were
 three sets of twins in these  15.  The number of Down Syndromes is very high. The
 number of births was lower than  the expected number, so there may also have been
 sterility or abortions.
       Anyway, it is interesting because this effect was attributed to consumption by
the mothers of trichlorfon-contaminated fish, trichlorfon they used to eliminate
parasites from the fish on the fish farms. They sometimes used ten times more of this
chemical than recommended. It killed the fish, but they were picked up and eaten
                                    408

-------
 because food was short.  We have this chemical on study in the continuous breeding
 protocol for reproductive effects and for the induction of dominant lethals.
 Discussion

        DR. EHRENBERG:  I have one  big question to each of the Mike's...any
 qualification on that. What are the criteria used for positive and negative results?
        DR. WATERS: As I mentioned we followed the Gene-Tox criteria so that we
 went through each study to match the Gene-Tox criteria for a positive or negative
 response and I cannot quote you all those criteria right now but they are as published
 in Mutation Research.  It is an obvious question. It is the sort of thing that we are
 very  concerned  about,  but we felt that  in order to get  some  handle on the
 performance of the short-term test we needed to create that data set. I do not really
 have any apologies for it.  I accept your criticism and we are aware of it.
       DR EHRENBERG: It is a question I think of historical data that we have. The
 question  to you  Mike, is why not include nitric oxides and nitrogen dioxide as
 chemicals in the  test.  With the nitric oxide you have both the  endogenous and
 exogenous exposure.   The  endogenous  exposure  varies very  much between
 conditions of the human - ten fold over production in certain allergies.
       DR SHELBY: Very good idea. We will include it. I would also like to answer
 part of the question you asked Mike Waters.  Ilse-Dore Adler is very aware of this
 question and so much concerned about  it that she recently organized a meeting in
 Neuherburg to address this issue of the analysis of a few of these tests.  Their
 recommendation was that the power be clearly stated in manuscripts so that when
you see a negative result you know the level of effect ruled out by the study.

                                    409

-------
        DR. ADLER: That is future music.  I mean the literature results, in particular
 the negative ones.  I still need to look at them and It is fine if you find a key to the
 positive effect criteria for that is very simple and straightforward; and I have done or
 I have reviewed all the positive data and confirmed that they were really truly positive
 but with the negatives there is a lot of work ahead of me.
        DR WATERS: I was going to say that Ilse-Dore quite kindly has gone through
 all of the  positive agents for us, and she is hopefully going to agree to look at the
 negative data set also. But that has not been done yet so we are a little premature
 in presenting this.  We thought it would give you some idea of how the information
 seems to be sorting out so far.  I think at least the bottom lines to me are  very
 interesting with respect to the performance of the in vitro tests versus the bone
 marrow assays.
        Da NATARAJAN:   Mike, you have a table from Renee Martin, the sperm
 chromosomes and the lymphocytes.
       DR. SHELBY: Right.
       DR. NATARAJAN:  I think this is what one would expect, I mean nobody has
 adequately compared the numbers if you have been in the field of radiation biology.
       DR BISHOP: And you are a cvtogenetidst.
       DR NATARAJAN: Yes, because the lymphocyte population is renewed, the
 half life of lymphocytes in the human body is about 200 days or 220 days; so that is
what you expect, that the population depleting and you are comparing the sperm then
they are coming from the stem cells, and it is good through meiosis and the selection,
and then what do you get. So, you get this big difference in what is expected.  This
can be combatted, too.
                                    410

-------
       DR SHELBY: Right, if you are not a cytogeneticist and you do not know
 about cell cycles then you are a risk estimator.
       DR SHELBY: But that does not keep them from doing it.
       DR ANDERSON: Do we now all accept from what Mike has just put up that
 there are in fact human germ cell mutagens, so we could all stand up with our hands
 on our hearts and now say we believe in x-rays, germ cell mutagens, human germ cell
 mutagen, or a human mutagen, and also that chemotherapeutic agents are; so if we
 are asked, do we know any human germ cell mutagens, or human mutagens we will
 say yes. Is that right? Does everybody accept that?
       DR ADLER:  If we start to do more sperm studies we might come up with
 better examples even than that.
       DR. ANDERSON: Do we accept that x-rays are then, human mutagens.
       DR. BISHOP: I think the discrepancy, Diana is that mutation is a heritable
 change and these chromosome aberrations that are being induced; these studies are
 not looking at a heritable change, they are not looking at factual transmission so I am
 not sure everybody would automatically accept that those are human mutagens
 because we are not showing a heritable change.
       DR. ANDERSON:  But sperm changes can be transmitted, sperm morphology
 changes can be transmitted, some of them.
       DR PRESTON:  I would very quickly say we saw reciprocal translocations in
spermatocyte in humans; you find exactly the same alteration in the mouse and in the
mouse you can show that it is transmitted to the F, so I think that is probably fairly
solid.
      DR ANDERSON:  Do we accept it?

                                  411

-------
      DR. ADLER: Any more questions, comments? If not, thank you all again and
I think it is has come up to right about that time, who is going to do the wrapping up?
                                 412

-------
Wrap-up And Concluding Remarks
                     413

-------
414

-------
 Wrap-up

 Dr. Brusick  gave the wrap-up presentation representing his views on the scientific
 sessions.  Dr. Brusick is from the Hazleton Labs America, Inc., located in Vienna,
 Virginia, United States.

       DR  BRUSICK:  It appears to me, after spending two or three days talking
 about this,  that we certainly do not have an absence of  information to do risk
 assessment. If anything I would say that we have too much information available to
 us.  If you go to other areas of toxicology,  the cancer risk assessment people are
 very happy  if they have one good study to do their risk assessment on, and they do
 their risk assessment and then live and die by that risk assessment, until someone
 else comes along and argues with them that there is another study.
       We have compounds here where we say, "well we have got one specific locus
 test, we have two heritable translocation tests, and maybe three  or four dominant
 lethal studies. We need some more data to do  a risk assessment."  And I do not
 think that that is really true.
       I think in fact we know too much about what we are doing,  and as a result of
 knowing too much and too many of the subtleties about the process of mutation, we
 end up with two problems.
       One  is, I think a function of paralysis by analysis (Figure 6-1).  What that
 means is that you  are analyzing so much that you fail to act; you, in a sense, paralyze
yourself because you spend so much time down  among the trees, that you lose the
forest, and I  think that we have done that to some extent.
                                    415

-------
                     CATCH 22 SITUATION
  REGULATOR:  WE CANT PREDICT HOW TO REGULATE ON THE BASIS OF
  GENETIC RISK UNTIL WE HAVE  ENOUGH DATA (EXPOSURE, ROUTE OF
  ADMINISTRATION, PK, DOSE RATE) TO KNOW EXACTLY HOW TO APPLY THE
  RISK ASSESSMENT.
  REGULATED:  WE ARE NOT OBLIGATED  TO SUPPLY ANIMAL GERM CELL
  DATA UNTIL THE AGENCY CAN TELL US HOW IT WILL BE INTERPRETED AND
  USED.
                            Figure 6-1

      The second problem (Figure 6-2) that I think has happened as a result of
knowing too much, is that we have intimidated potential users of this information.  It
is so complicated and so intimidating, with the information that we have, and
acrylamide was a good example.  Every time we begin a discussion, we would talk
about beginning to do an estimate of risk for a certain type of mutation, and someone
would throw out something more that we know about the window of opportunity for
                       CONSEQUENCES
      NEW DATA MUST BE DEVELOPED FROM R&D PROGRAMS WHICH
               ARE POORLY FUNDED OR HAVE MISSIONS
                 INCOMPATIBLE WITH THE AGENCIES

                            Figure 6-2
                              416

-------
 mutation or heritable translocation, because ft is this kind of mechanism; these are
 very, very fine details which may or may not have any relevance to the overall number.
 The variance on these numbers is so big that some of these little fine things just really
 do not contribute that much, but they intimidate anyone trying to understand what is
 going on and I think to some extent they intimidate us to the point that we are afraid
 to sort of lay this out. Because if I do not make it complicated, then I am going to be
 criticized by my peers for trying to simplify this process too much. So in a sense, we
 are sort of losing because we know too much, I believe.
       So it is not surprising that when we go through all this, and you give it to the
 regulatory agencies, that one of the things that happens is that they run right back
 and adopt the Ames test because it is simple, you understand it and boy,  I could take
 the Ames test and I can make all the  conclusions.  I can extrapolate it all over the
 world because of the perception of the simplicity and the way it is explained.
       So I think that there is really the core of the problem that I see.  Taking into
 account that we do not want to oversimplify or throw out numbers at people are going
 to grab, and  I really realize that.  But I think that what  we have is that genetic
 toxicology to some extent is a technology in search of a problem.
       We do not have the corpse and  that has been alluded to, many different times.
 We have a lot of smoking guns, I think.  We have got animal data, which is a smoking
 gun.  We have got all the information that Mike talked about, which in a sense is
 indirect information; but we still do not have the corpse, and we may never have the
 corpse. So I do not think that we ought to try to hang this field particularly on finding
the corpse.
                                    417

-------
        Another problem we have is the Catch-22 situation and this is between the
 regulators and the regulated. The regulator says we cannot predict how to regulate
 your compound because we really do not have enough information on the dosimetry,
 dose-effect rates, dose, route of exposure, PK information, et cetera, please give us
 that information so we will  understand how to use this technology. The regulated
 community says we are not going to give you this information until you can show us
 how you are going to use it.
        So we have got a dilemma in that nobody has access to information, to
 develop a way to use the information, and therefore we cannot actually solve the
 problem of the regulated, by telling him what we are going to do, because we will
 never get the information.  As a consequence, data are generally developed out of
 our R&D program.  And most of those, as we know, are not funded to this same
 extent.   They do not have the buying power, the data development  power, that
 industry has in developing products, if that information were made available. In many
 cases, without apologizing,  these R&D programs are not directed necessarily at the
 missions of the agencies like EPA, or some of the others. They want to go and study
 some other information, where in fact EPA needs some very fundamental information,
 which is not particularly a scientific interest.
       So  I would suggest that we might want to try  another  meeting, and in a
 second meeting talk about some issues that relate to  how we  communicate this,
 rather than how we do it. Because I do not think there are really any problems on
 how to do it.
       It became reasonably clear that the four work groups (Figure 6-3), even though
they approached each chemical a little bit differently, had no problem coming up with

                                    418

-------
                         SUMMARY
  1.    GENERAL CONSENSUS IS THAT THE TIME IS UPON US TO INCREASE
       THE VISIBILTTYAND IMPORTANCE OF GENETIC RISK ASSESSMENT

            -CREDIBILITY OF SCIENCE IS AT RISK
            -IMPACT OF THE RISK MAY BE UNDERESTIMATED
            -SURVIVAL OF FIELD IS TIED TO INTEREST OF GOVERNMENT
            AND PRIVATE INDUSTRY

  2.    EXAMINE HOW WE VIEW GENETIC RISK

            -CONGENITAL ANOMAUES FROM TREATMENT OF ZYGOTES
            -MALE TRANSMITTED CONGENITAL ANOMALIES
            -INCREASED CANCER INCIDENCE IN F1 AND F2
            -FERTILITY AND ASSOCIATED ENDPOINTS

  3.    EXAMINE HOW WE COMMUNICATE GENETIC RISK

            -SIMPLIFY THE DESCRIPTION WITHOUT OVER-SIMPLIFICATION
            -EXPRESS RISK IN TERMS OF ADVERSE OUTCOMES RATHER
            THAN  INCREASES IN SPECIFIC ENDPOINTS
            (TRANSLOCATIONS,
             DELETION ETC.)
            -ATTEMPT TO REACH CONSENSUS ON BASIC RULES FOR THE

             PROCESS

  4.     EXAMINE HOW WE EXPRESS RISK

            -POPULATION OR INDIVIDUAL
            -INCLUDE REPRODUCTIVE BEHAVIOR: RISK PER
             MATING/CONCEPTIONS/F1
            -COMBINE INDIVIDUAL RISKS OR NOT
            -INCLUDE CUMULATIVE EXPOSURES OR ONLY ACUTE
                           Figure 6-3


some numbers and we can argue for days and days about whether those numbers

are good numbers or bad and you know, you saw George, he had a page full of
                             419

-------
 numbers that were risk numbers for the F, population. So generating the numbers is
 not a problem, and I think here is where we have to look.
        We have to then look at trying to find out how we can increase the visibility,
 and demonstrate the importance of genetic risk assessment. We have got credibility
 at risk, we have got an impact issue; survival in the field, I think is tied to how well we
 can convince people in the government, and private industry, that this is an important
 issue.  We have got to examine how we view genetic risk. Are we going to make ft
 very narrow, or are we going to broaden it to include congenital anomalies in the
 zygotes, male transmitted congenital anomalies, or are we even going to look at the
 issue of cancer in the Fv F2, as an indicator and even broaden it maybe into fertility,
 and other associated reproductive developmental problems. That is another issue
 that we should talk about.
       The third, is how are we going to communicate the risk numbers that we
 develop. Number one, we have got to simplify the description without oversimplifying
 it, and that is a very difficult thing to do, but it  is absolutely essential.  If we are not
 going to take an approach to begin to simplify the process,  and explain it  as a
 relatively simple process, we will  never sell it, we will never, ever sell it.
       We have got to be able to express risk in terms of adverse outcomes, and not
 genetic endpoints, and I think that was made pretty clear by most of the speakers and
 we have got to reach a consensus on basic rules for the process. I think just some
 very simple basic rules like what kind of data you need, what are certain assumptions
that we  are going to make. Without these kinds of basic rules, we will not fit in again,
good or bad, to the mainstream of  toxicology.  And the mainstream of toxicology
when you look at  other risk analyses, analysis processes, they have basic rules that

                                    420

-------
 they follow. Some people diverge from those rules and are quickly beaten around a
 little bit, and are brought back into tow.
       Then examine how we are going to express it; is It going to be a population
 risk? Are we  going to talk about individual risks? That is very important.  Are we
 going to include reproductive behavior, we are going to talk about risk per mating
 opportunity, risk per conception, risk per F.,? All  of these things  may become
 important.  If you want to counsel individuals who are concerned about an exposure
 on the job, they might want to know risks such as what is the risk per conception, or
 what is the risk every time I  have sex; and we can get numbers like that, they are not
 totally out of reason, and can we combine individual risks.
       We have risks for chromosome damage, we have risks for gene mutation.  Is
 tt possible, once we have calculated these, to then come up with the single number
 or we will always be faced with this matrix of risk numbers?
        And finally, I think we should discuss cumulative exposures, versus acute, and
 that came up over and over again; that in  a sense what we do is take the animal data
 and we calculate a risk for acute exposure then, in a sense, in many cases  with
 occupational exposures, especially for spermatic gonial mutagens, you accumulate
 that risk on a daily basis to that spermatogonia population. And therefore, what is the
 dose? Is it a cumulative dose, over 20 years or is it a dose for a week, or whatever
 it may be.
      So my suggestion te to think about that for future meetings, before we  then
and go back in and do more number crunching, and then finally application of the
parallelogram to risk assessment (Figure  6-4).
                                    421

-------
             APPLICATION OF PARALLELOGRAM TO
                       RISK ASSESSMENT

  ADVANTAGES
        •    METHOD IS INTUITIVE AND SIMPLE TO EXPLAIN
        •    SOME EXTRAPOLATION FACTORS ARE INTRINSIC
  LIMITATIONS
        •    WITHOUT ADEQUATE DOSE:EFFECT DATA,
             PARALLELOGRAM WILL HAVE TO BE CONSTRUCTED
             FOR EACH POTENTIAL HUMAN EXPOSURE
        •    MAY BE DIFFICULT TO COMPARE ACROSS
             COMPOUNDS UNLESS DATA OF THE THREE CORNERS
             ARE SIMILAR (I.E. MAKE ONE OR TWO HUMAN SOMATIC
             ENDPOINTS STANDARD)
        •    HUMAN SOMATIC CELL DATA MAY BE DIFFICULT TO
             OBTAIN
                             Figure 6-4
      The advantage of it is, I think, that it is a method that is pretty intuitive. If you
look at it, it makes sense, it is fairly simple, and it has a strong appeal from that
standpoint It also, because you are only looking at two endpoints in the same
animal, and then in human, some of the extrapolation factors are automatically built
in. We do not have to worry about extrapolating necessarily from the somatic cell to
the germ cell, because that is built into the same animal.  So some of those factors
are sort of taken into account, but the limitations probably are more important than
the advantages.
                               422

-------
       If we do not have dose effect data, every time you  want to look at the
 parallelogram for a different dose exposure, you  have to go back and redo the
 parallelogram again, because you do not know the shape of the curve so if it is linear
 that is one thing; but if it is not linear, then you do not know that the relationship
 between somatic and germ cells in the animal is going to hold up over a range of
 doses.  It is going to be for that particular exposure, and then the same relationship
 will hold only for that particular exposure in the human, and it is difficult to compare
 across compounds.
       I think it would be nice if we are going to use this as an approach to try and
 standardize, maybe the  somatic cell determinations  in humans, to one or two
 endpoints. Maybe hemoglobin abducts would be a good one, dosimetry, or SCEs or
 something of that sort, so you can then begin to look at one chemical versus another.
       Now we are only able to look essentially at a compound in isolation and not
 necessarily  compare it because we do not know the relative  sensitivities of these
 comparisons across different compounds.
       So if you are looking at SCEs with one compound, as a somatic thing and you
 make those chromosome  aberrations  or abducts in another,  there are some
 differences and that is difficult to get human somatic cell data.
      So I think from my perspective it is certainly a model that we can  use,  but  I
think that at this point we do not have to go to models like parallelograms, if we do
not want to;  I think we have got enough power in using the direct, or modified direct
or whatever analysis we would like to use for doing a risk assessment to go right
ahead and do that. That is it.
                                    423

-------
424

-------
 Concluding Remarks

 Dr. Canice Nolan gave the concluding remarks.  Dr. Nolan is from the European
 Commission located in Brussels, Belgium.

       DR NOLAN:  It falls on me to give the wrap up of the wrap up. Going back
 to the objectives of the meeting - to identify the methodology, data requirements, and
 mechanistic research needed to understand the health impact of germ cell mutagens,
 I think we have gone a long way towards doing just that. When we chose the four
 chemicals we  knew already that we would have problems with one or two, because
 there were data missing.  Problems were identified in applying the parallelograms to
 each of these four compounds.  However, I do not think you should throw out the
 method because of problems with the data. We have identified needs for data and
 shown that the parallelogram does have advantages over other methods.  I am not
 pro- or anti-parallelogram or indeed any other assessment methods but I think that
 it should be given a chance, with a full set of data and I look forward to the day that
 we can do so. That is about as much as I am going to say on that.
       We are quite pleased with the meeting and we have what we wanted. We also
 had some good overview papers and cross-cutting papers.  I have not been able to
 sit in on all the working group meetings and I await getting the full transcripts from
 those. I would hope, and I hope I speak for Dr. Ott as well, that this will not be a one-
 off meeting and that we will have follow-up meetings.
       I would like to thank you all for participating  but  I would  like to thank
particularly our hosts here at Research Triangle Park, especially Mike Waters, who I
know has done a lot of work putting this meeting together.  I would also like to thank
                                    425

-------
Frits because he really did a lot of the preparatory work on the European side for this.
 It is a real pity that he could not be here with us.
       DR. WATERS: I would also like to thank my counterpart, Canice, because I
failed  to do that In the beginning, which I had intended to and he certainly has done
his share to make this meeting happen.  Hopefully it will not be the last, and we will
look forward to the next one.
       DR. NOLAN: Thanks Mike and thank you all.
                                   426

-------
References
                       427

-------
428

-------
 References
 Adler, I-D. (1990) Clastogenic effects of acrylamide in different germ-cell stages of
       male mice.  In: J. Allen, B. Bridges, M. Lyon, M. Moses & L Russell (Eds.),
       Biology of mammalian germ cell mutaqenesis.  Banburv Report. Vol. 34 (pp
       115-131). Cold Spring Harbor:  Cold Spring Harbor Laboratory Press.

 Aguirrezabalaga, L, M.J.M. Nivard, & E.W. Vogel (1993).  Mutation Res.. 293. p.286.
       (Abstract).

 Aguirrezabalaga, I., M.J.M. Nivard, MA  Comendador, & E.W.  Vogel,  (1994) in
       preparation.

 Appelgren, L.E., G. Eneroth, C. Grant, LE. Lanstr&m & K. Tenghagen (1978). Testing
       of ethylene oxide for mutagenicity using the micronucleus test in mice and
       rats. Ada Pharmacol. Toxicol.. 43, 69-71.

 Ballering, LAP., M.J.M. Nivard, & E.W. Vogel, (1994).  Paper submitted.

 Barbin, A., &  H.  Bartsch (1989).   Nucleophilic selectivity  as a determinant of
       carcinogenic potency (TDgo) in rodents:  a comparison of mono- and  bi-
       functional alkylating agents and vinyl chloride metabolites. Mutation Res.. 215.
       95-106

 Bamett, L.B., D.P. Lovell, C.F. Felton, B.J. Gibson, R.R. Cobb, D.S. Sharps, M.D.
       Shelby,  & S.E. Lewis (1992). Ethylene dibromide: negative results with the
       mouse dominant lethal assay and  the electrophoretic specific-locus test.
       Mutation Res.. 282. 127-133.

 Benson,   A.J.,   C.N.  Martin,  &  R.C.   Gamer,  (1988).   N-(2-hydroxyethyl)-N-[-
       (7-guaninyl)ethyl]amine, the putative major DNA adduct of cyclophosphamide
       in vitro and in vivo in the rat. Biochem. Pharmacol.. 37, 2979-2985.

 Boyd,  J.B., M.D.  Golino,  T.D.  Nguyen,  & M.M.  Green  (1976).   Isolation and
       characterization of X-linked mutants of Drosophila melanogaster which are
       sensitive to mutagens.  Genetics. 84. 485-506.

Brewen, J.G., & R.J. Preston (1974).  Cytogenetic effects of environmental mutagens
       in mammalian cells and the extrapolation to man. Mutation Res.. 26,297-305.

Brewen, J.G.,  R.J.  Preston & N. Gengozian (1975).   Analysis of X-ray induced
       chromosomal translocations  in human and marmoset spermatogonial stem
       cells. Nature. 253. 468-470.
                                     429

-------
 Buul, P.P. W. van (1976). A comparative study of the frequencies of radiation-Induced
       chromosome aberrations in somatic and germ cells of the Rhesus monkey.
       Mutation Res.. 36, 223-236.

 Buul,  P.P.  W.  van  (1977).    Dose-response  relationship  for  radiation-induced
       translocations in somatic and germ ceils of mice. Mutation Res.. 45, 61-68.

 Calleman,  C.J.,  L Ehrenberg, B.  Jansson, S. Osterman-Golkar,  D. Segerback, K.
       Svensson, & C.A. Wachtmeister, (1978). Monitoring and risk assessment by
       means of alkyl groups in haemoglobin in persons occupationally exposed to
       ethylene oxide. Environ. Path. ToxicoL g, 427-42.

 Cattanach, B.M. (1966). Chemically induced mutations in mice.  Mutation Res.. 2,
       346-353.

 Cochrane, J.E., & T.R. Skopek (1993a) (in press). Mutagenicity of butadiene and its
       epoxide  metabolites:  I.  Mutagenic potential  of 1,2-epoxybutene, 1,2,3,4-
       diepoxybutane, and 3,4-epoxy-1,2-butanediol in cultured human lymphoblasts.
       Carcinoaenesis.

 Cochrane, J.E., & T.R. Skopek (1993b) (in press). Mutagenicity of butadiene and its
       epoxide metabolites: II. Mutational spectra of butadiene, 1,2-epoxybutene, and
       diepoxybutane at the hprt locus in splenic T-cells from exposed B6C3F1 mice.
       Carcinoaenesis.

 Conan, R.A., G.T. Waggy, M.H.  Spiegel &  R.L Berglund (1979).  Study  of  the
       mutagenic action  of ethylene oxide,  ethylene glycol, and  2-chloroethanol
       residues  in plastic material sterilized by ethylene oxide. Ann. Falsif.  Expert
       Chim.. 72.141-151.

 Dam F.J. van, AT. Natarajan & A.D. Tates (1992). Use of a T-lymphocyte clonal assay
       for determining HPRT  mutant frequencies in individual rats.  Mutation Res..
       271. 231-242.

 Dean, B.J., D. Anderson, & R.J. Sram, (1982). The mutagenioity of selected chemicals
       in the dominant lethal assay.  In:  F.J. de Serres & M.D. Shelby  (Eds.),
       ComparativeChemical Mutaoenesis (DP. 487-5381. New York: Plenum Press.

 Dearfield, K.L  (1991).  Pesticide Assessment Guidelines, Subdivision F. Hazard
       Evaluation:    Human  and Domestic  Animals, Series  84,  Mutagenicity,
       Addendum  9,  (EPA Publication  No.  540/09-91-122).  U.S. Environmental
       Protection Agency, Office of Pesticide Programs, Washington, DC (NTIS Order
       No.  PB9M 58394)

Dellarco, V.L, W.M. Generoso, GA Sega, J.R. Fowle III & D. Jacobson-Kram (1990).
       Review of the mutagenicity of ethylene oxide.  Environ. Molec. Mutaaen.. J6,
       85-103.
                                    430

-------
 Ehling, U.H. (1981). Risk estimations based on Germ-cells Mutation in Mice.  In:
       T. Sugimura, S. Kondo, & H. Tokebe (Eds.), Environmental Mutaoens and
       Carcinogens.   Proceedings of  the 3rd  International  Conference  on
       Environmental Mutagens (pp. 709-719). Tokyo: University of Tokyo Press.

 Ehling, U.H., & J. Favor (1984).  In:   E.H. Chu & W.M. Generoso (Eds.), Mutation.
       Cancer and Malformation, (pp. 389-428).  New York:  Plenum Press.

 Ehling, U.H., & A. Neuhauser-Klaus  (1984).   Dose-effect relationships of germ-cell
       mutations in mice.  In: Y. Tazima, S.  Kondo, & Y. Koroda, (Eds.), Problems
       of Threshold In Chemical Mutaoenesis. The Environmental Mutagen Society
       of Japan (pp. 15-25). Tokyo: Kokusai-Bunken Printing, Co., Ltd.

 Ehling, U.H.  & A.  Neuhauser-Klaus, (1988).   Induction  of  specific-locus and
       dominant-lethal   mutations   by   cyclophosphamide  and   combined
       cyclophosphamide-radiation treatment in male mice.  Mutation  Res.. 199.
       21-30.

 Ehling, U.H.,  & A. Neuhauser-Klaus  (1988b).   Induction  of  specific-locus and
       dominant-lethal mutations in  male mice by diethyl sulfate (DES).  Mutation
       Res.. 199.191-198.

 Ehling, U.H., & A. Neuhauser-Klaus (1989a). Induction of specific-locus and dominant
       lethal mutations in male mice  by chlormethine. Mutation Res.. 227. 81-89.

 Ehling, U.H., & A. Neuhauser-Klaus (1989b).  Induction of specific-locus mutations in
       male mice by ethyl methanesulfonate (EMS).  Mutation Res.. 227. 91-95.

 Ehling, U.H., & A. Neuhauser-Klaus (1990). Induction of specific-locus and dominant
       lethal   mutations  in  male mice  in  the low  dose  range  by  methyl
       methanesulfonate (MMS).  Mutation Res.. 230. 61-70.

 Ehiing, U.H., & A. Neuhauser-Klaus (1992). Reevaluation of the induction of specific-
       locus mutations in spermatogonia of the mouse by acrylamide. Mutation Res..
       283. 185-191.

Ehling, U.H., & A. Neuhauser-Klaus (1994) (in press).  Mutation Res.

Ehrenberg, L (1979).  Risk assessment of ethylene oxide and other compounds. In:
       V.K. McElheny & S. Abrahamson (Eds.), Assessing chemical mutaoens: the
       risk to humans. Banburv Report 1 (pp. 157-90).  Cold Spring Harbor, New
       York: Cold Spring Harbor Laboratory Press.
                                    431

-------
 Ehrenberg, L (1980).  Purposes and methods of comparing effects of radiation and
       chemicals.  In:  Radiological equivalents of chemical pollutants (pp. 23-36).
       Vienna:  International Atomic Energy Agency.

 Ehrenberg, L, K.D. Hiesche, S. Osterman-Golkar, & I. Wennberg, (1974). Evaluation
       of genetic risks of alkylating agents:  tissue doses in the mouse from air
       contaminated with ethylene oxide. Mutation Res.. 24, 83-103.

 Farooqi, Z., M. Toornqvist, L Ehrenberg & AT. Natarajan (1993). Qenotoxic effects
       of ethylene oxide and propylene oxide in mouse bone marrow cells.  Mutation
       Res.. 288. 223-228.

 Favor, J., E.W. Vogel, A.A. van Zeeland, J.G. Filser, M.L Mendelsohn, & U.H. Ehling
       (1994).  In:    DJ.   Brusick (Ed.), Methods for  Genetic  Risk  Assessment
       ICPEMC/UNEP Working Group III: Dose and effect assessment. Boca Roton:
       Lewis Publishers.

 Fomenko, V.N.  & E. Ye Strekalova (1973). The mutagenic effect of some  industrial
       toxins as a function of industrial toxins as a function of concentration and
       exposure time. Tokslkol. Nov. Prom Khim Veshchestv. 13, 51-57.

 Generoso, W.M., K.T. Cain, C.V. Cornett, N.LA. Cacheiro & LA.  Hughes (1990).
       Concentration-response   curves  for   ethylene  oxide-induced   heritable
       translocations and dominant lethal mutations. Environ. Molec. Mutaaen.. 16,
       126-131.

 Generoso, W.M., K.T. Cain, M. Krishna, C.W. Sheu, & R.M. Gryder (1980).  Heritable
       translocation and dominant-lethal mutation induction with ethylene oxide in
       mice. Mutation Res.. 73.133-142.

 Meddle, JA (1982).  Comparative mutagenidty of myleran and cydophosphamide.
       In FJ. de Serres & M.D. Shelby (Eds.), Comparative Chemical Mutaaenesis.
       New York: Plenum Press.

 Hemminki, K. (1985). Binding of metabolites of cydophosphamide to DMA in a rat
       liver microsomal system and in vivo in mice. Cancer Res.. 45, 4237-4243.

 Henderson R.F., W.E. Bechtold, P.J. Sabourin, K.R. Maples & A.R. Dahl (1993).
       Spedes differences in the metabolism of 1,3-butadiene in vivo. In M. Sorsa, K.
       Pettonen, H. Vainio & K. Hemminki (Eds.), Butadiene and Stvrene: Assessment
       of Health Hazards. (ARC Scientific Publications No. 127, Lyon: IARC.

Hdgstedt, B., E. Bergmark,  M. T6mqvist & S. Osterman-Golkar (1990). Chromosomal
       aberrations  and micronudei in lymphocytes in relation to alkylation of
       hemoglobin in workers exposed to ethylene oxide and propylene oxide.
       Hereditas. 113.133-138.
                                    432

-------
 ICPEMC  (1983). Screening strategy  for chemicals that are potential germ-cell
       mutagens in mammals.  International Commission for Protection  Against
       Environmental Mutagens and Carcinogens. Mutation Res.. 114.117-177.

 ICPEMC (1993a). Genetic risk extrapolation from animal data to human disease. Final
       Report  prepared for the  U.S.  Environmental  Protection Agency  by the
       International Commission for Protection Against Environmental Mutagens and
       Carcinogens, under Interagency Agreement No. 1824-D094-A1 with Oak Ridge
       National Laboratory, September 1993.

 ICPEMC  (1993b). Taskgroup on Risk Extrapolation, Final Report.  Use of In  Vivo
       Genetic Toxicology Data to Construct Human Risk Assessments. Submitted
       to Ministry of Health and Welfare Canada by International Commission for
       Protection Against  Environmental Mutagens and Carcinogens.  Contract No.
       3138, Collator No.  151103, Project No. K332202, October,  1993.

 Jenkinson,  P.C., & D. Anderson (1990).   Malformed  foetuses  and karyotype
       abnormalities in the offspring of cyclophosphamide and allyl-alcohol treated
       male mice. Mutation Res.. 229.173-184.

 Jenssen,  D. & C. Ramel  (1980).  The micronucleus test as  part of  a short-term
       mutagenicity test program for the prediction of carcinogenicity  evaluated by
       143 agents tested.  Mutation Res.. 75,191-202.

 KJigerman, A.D., G.L Erexson, M.E. Phelps & J.L Wilmer (1983). Sister-chromatid
       exchange induction in peripheral  blood lymphocytes of rats  exposed to
       ethylene oxide by inhalation.  Mutation Res.. 120. 37-44.

 Lee, W.R. (1976).  Molecular dosimetry of chemical mutagens.  Determination of
       molecular dose to the germ line. Mutation Res.. 38, 311-316.

 Lee, W.R., & M.R.  Kelley (1986).  In Risk  Assessment  in Relation to Environm.
       Mutaoens and Carcinogens (pp. 99-102). New York: Alan R. Uss, Inc.

 Lee, W.R., S. Abrahamson, R. Valencia, E.S. von Halle, F.E. Wurgler  & S. Zimmering
       (1983).  The sex-linked recessive lethal test for mutagenesis in Drosophila
       melanogaster.  Mutation Res.. 123.183-279.

 Legator,  M.S.,  W.W. Au,  M. Ammenheuser,  & J.B. Ward Jr.  (1993).  Genetic
       biomonitoring of butadiene exposed populations. In: M. Sorsa, K. Peltonen,
       H. Vainio & K. Hemminki (Eds.), Butadiene and Stvrene: Assessment of Health
       Hazards. IARC Scientific Publications No. 127, Lyon: (ARC.

Lerda, D. & R. Rizzi (1992).  Cytogenetic study of persons occupationally exposed to
       ethylene oxide. Mutation Res.. 281. 31-37.
                                    433

-------
 Lewis, S., L Bernet, C. Felton, F. Johnson, L Stow, N. Cacheiro & M. Shelby (1986).
       Dominant visible and electrophoretically expressed mutations induced in male
       mice exposed to ethylene oxide by inhalation. Environ. Molec. Mutaaen.. 8,
       867-872.

 Lu, A-L, & I-C Hsu, (1992). Detection of single DMA base mutations with mismatch
       repair enzymes.  Genomlcs. 14, 249-255.

 Lyon, M.F., I-D. Adler, B.A. Bridges, L Ehrenberg, L  Golberg, D.J. Kilian, S. Kondo,
       E. Moustacchi, A. Patrament, K. Sankaranarayanan, F. Sobels, R.J. Sram, G.
       Streisinger, & K. Sundaram, (1983). Estimation of genetic risks and increased
       incidence of genetic disease due to environmental mutagens.   (ICPEMC
       Committee 4 Report).  Mutation Res.. 115. 255-291.

 Maccubin, A.E., L Caballes, J.M. Riordan, D.H. Huang, & H.L Gurtoo, (1991).  A
       cydophosphamide-DNA phospoester adduct formed in  vitro and in  vivo.
       Cancer Res.. §1, 886-892.

 Mayer, J., D. Warburton, A.M. Jeffrey, R. Pero, S. Walles, L Andrews, M. Toor, L
       Latriano, L Wazneh, D. Tang, W.-Y. Tsai, M. Kuroda & F. Perera (1991).
       Biologic markers in ethylene oxide-exposed workers and controls. Mutation
       Res.. 248.163-176.

 Mehta, J.R.,  M. Przbylski, & D.B.  Ludlum,  (1980).   Alkylation of guanosine and
       deoxyguanosine by phosphoramide mustard. Cancer Res.. 40, 4183-4186.

 Mehta, J.R. & D.B. Ludlum, (1982).   Trapping of DMA reactive metabolites  of
       therapeutic  or  carcinogenic   agents   by   carbon-14-labelled   synthetic
       potynucleotides.  Cancer Res.. 42, 2996-2999.

 Mohn, G.R., & AA van Zeeland, (1985). Quantitative comparative mutagenesis in
       bacteria, mammalian cells and animal-mediated assays: A convenient way of
       estimating genotoxic activity in vivo?  Mutation Res.. 150.158-175.

 Muller, H.J. (1954).  In: A. Hollaender (Ed.), Radiation Biology. Vol. 1, McGraw Hill,
       New York, 475-626.

 Newbold,  R.F., W. Warren, A.S.C. Medcalf, & JA Amos, (1980).  Mutagenidty of
      carcinogenic  methylating  agents  is  associated  with  a  specific  DMA
      modification. Nature. 283.596-598.

Nivard, M.J.M. (1991). Thesis, University of Leiden.

Nivard, M.J.M., A. Pastink, & E.W. Vogel (1992). Molecular analysis of mutations
      induced in  the  vermilion  gene  of  Drosphila melanogaster by  methyl
      methanesulfonate. Genetics. 131. 673-682.
                                    434

-------
 Norppa, H. & M. Sorsa (1993).  Genetic toxicity of 1,3-butadiene and styrene. In:
       M. Sorsa, K. Peltonen, H. Vainio & K. Hemminki (Eds.), Butadiene and Stvrene:
       Assessment of Health Hazards. IARC Scientific Publications No. 127, Lyon:
       (ARC.

 Osterman-Golkar,  S.M., J.A. Bond, J.B. Ward Jr. & M.S. Legator (1993).  Use of
       hemoglobin adducts for biomonitoring exposure to 1,3-butadiene. In M. Sorsa,
       K.  Peltonen, H. Vainio & K. Hemminki  (Eds.),  Butadiene and  Stvrene:
       Assessment of Health Hazards. (ARC Scientific Publications No.  127,, Lyon:
       (ARC.

 Osterman-Golkar,  S., L Ehrenberg, D. Segerback, & I. Hallstrom, (1976).  Evaluation
       of  genetic risks of alkylating agents  is associated with  a specific DMA
       modification. Nature. 283. 596-598.

 Pastink, A., C.  Vreeken, M.J. Nivard, LL Searles, & E.W. Vogel (1989). Sequence
       analysis of N-ethyl-N-nitrosourea-induced vermillion mutations in Drosophila
       melanogaster. Genetics. 123.123-129.

 Pastink, A.,  E. Heemskerk,  M.J. Nivard,  C.J. van Vliet,  & E.W. Vogel  (1991).
       Mutational  specificity of ethyl methanesulfonate in excision-repair-proficient
       and -deficient strains of Drosphila melanogaster.  Mol. Gen. Genet.. 229.213-
       218.

 Peltonen K., P. KoMsto, I. Neagu, R. Kostiainen, I. Wlpelainen , & M. Sorsa (1993).
       Estimating internal dose of 1,3-butadiene: preliminary data on use of modified
       purine bases as markers of exposure.  In M. Sorsa, K. Peltonen, H. Vainio &
       K. Hemminki (Eds.), Butadiene and Stvrene: Assessment of Health Hazards.
       IARC Scientific Publications No. 127, Lyon: IARC.

 Perera, F., D. Brenner, A. Jeffrey, J. Mayer, D. Tang, D. Warburton, T-L Young, L
       Wazneh, L  Latriano, G. Motykiewicz, E. Grzybowska, M. Chorazy, K. Hemminki
       & R. Santella (1992).  DNA adducts and  related biomarkers in  populations
       exposed to environmental carcinogens. Env. Health Perspec.. 98,133-137.

 Rhomberg, L,  V.L Dellarco,  C. Siegel-Scott, K.L Dearfield & D.  Jacobson-Kram,
       (1990).   Quantitative  estimation of the  genetic risk associated with  the
       induction of heritable translocations at low-dose exposure: ethylene oxide as
       an example. Environ. Molec. Mutaaen.. 16,104-125.

Ribeiro, L.R., M.N. Rabello-Gay, D.M.F. Salvador! & C.A.B. Pereira (1987). Cytogenetic
       effects of inhaled ethylene oxide in somatic  and germ cells of mice. Arch.
       Toxicol.. 59, 332-335.

Russell, L.B., R.B. Gumming, & P.R. Hunsicker (1984). Specific-locus mutation rates
       in the mouse following inhalation of ethylene oxide, and application of the
       results to estimation of human genetic  risk. Mutation Res.. 129. 381-388.
                                     435

-------
 Russell, LB., P.R. Hunsicker, N.LA. Cacheiro, J.W. Bangham, W.L Russell, & M.D.
        Shelby (1989).  Chlorambucil effectively induces deletion mutations in mouse
        germ cells.  Proc. Natl. Acad. ScL (USA), fi§, 3704-3708.

 Russell, LB., W.L Russell, E.M. Rinchik, & P.R. Hunsicker (1990).  In: J.W. Allen, B.A.
        Bridges, P.M. Lyon, M.J. Moses, & LB.  Russell (Eds.), Pactore affecting the
        nature of Induced mutations. Banburv Report 34. (pp. 271-289). Cold Spring
        Harbor:  Cold Spring Harbor Laboratory Press.

 Russell, LB., P.R. Hunsicker, & M.D. Shelby (1992a). Melphalan, a second chemical
        for which specific-locus mutation induction in the mouse is maximum in early
        spermatids. Mutation Res.. 282.151-158.

 Russell, LB., P.R.  Hunsicker, N.LA. Cacheiro, & E.M. Rinchik (1992b).  Genetic,
        cytogenetic, and molecular analyses of mutations induced by melphalan
        demonstrate high frequencies of heritable deletions and other rearrangements
        from exposure of postspermatogonial stages of the mouse. Proc. Natl. Acad.
        ScL (USA), 89, 6182-6186.

 Russell, LB., & E.M. Rinchik (1993).  Structural differences between specific-locus
        mutations induced by different exposure regimes in mouse spermatogonial
        stem cells.  Mutation Res.. 288.187-195.

 Russell, W.L, E.M. Kelly, P.R. Hunsicker, J.W. Bangham, S.C. Maddux, & E.L Phipps
        (1979).   Specific-locus test shows ethylmitrosourea to be the most potent
        mutagen in the mouse. Proc. Natl. Acad. ScL (U.SA), Z§, 5918-5819

 Russell, W.L, & P.R. Hunsicker (1983). Extreme sensitivity of one particular germ-cell
       stage  in  male  mice to induction  of specific-locus mutations  by
       methylnitrosourea. Environ. Mol. Mutaoen.. §, 498.

 Russell, W.L (1984).   In Y. Tazima, S.  Kondo, & S.  Koroda (Eds.),  Problems of
       threshold in chemical  mutaoenesis. The Environmental Mutaoen  Society of
       Japan, (pp. 153-160).  Tokyo:  Kokusai-Bunken Printing Co., Ltd.

 Sarto,  F., R. Tomanin, L  Giacomelli, G. lannini & A.R.  Cupiraggi  (1990).  The
       micronucleus assay in human exfoliated cells of the nose and  mouth:
       application to occupational exposures to chromic acid and ethylene oxide.
       Mutation Res.. 244. 345-351.

Sarto,  P., MA  T6mqvist, R. Tomanin, G.B. Bartolucd, S.  Osterman-Golkar & L
       Ehrenberg (1991). Studies of biological and chemical monitoring of tow-level
       exposure to ethylene oxide. Scand. J. Work Environ. Health. 17, 60-64.

Schulte, PA, M. Boeniger, J.T. Walker, S.E. Schober, MA Pereira, O.K. Gulati, J.P.
       Wojdechowsky, A. Garza, R. Proelich, G. Strauss, W.E. Halperin,  R. Herrick,
       & J. Griffith (1992). Biologic markers in hospital workers exposed to tow levels
       of ethylene oxide. Mutation Res.. 278. 237-251.

                                    436

-------
 Segerback, D.,  (1990).  Reaction products in hemoglobin and DMA after in vitro
       treatment  with  ethylene   oxide   and  N-(2-hydroxyethyl)-N-nitrosourea.
       Carcinoaenesis. 1J., 307-312.

 Shelby,  M., K. CaJn, C. Cornett & W. Generoso (1987).  Acrylamide: induction of
       heritable translocations in male mice. Environ. Molec. Mutaoen.. 9, 363-368.

 Shukia,  P.T., &  C. Auerbach (1980). Genetic tests tor the detection of  chemically
       induced small deletions in drosophila chromosomes. Mutation Res.. 72, 231-
       243.

 Singer, B., & D. Grunberger (1983). Molecular Biology of Mutaaens and Carcinogens.
       New York: Plenum Press.

 Sobels,  F.H. (1977). Some problems associated with the testing for environmental
       mutagens and a perspective for studies in comparative mutagenesis. Mutation
       Res.. 46, 245-260.

 Sobels,  F.H. (1982).  The parallelogram: An indirect approach for the assessment of
       genetic risks from chemical mutagens.  In:  K.C. Bora, G.R., Douglas, & E.R.
       Nestman, (Eds.).  Progress In Mutation Res.. 3, 323-327.

 Sobels,  F.H. (1989). Models and assumptions underlying genetic risk assessment.
       Mutation Res.. 212. 77-89.

 Sotomayor,  R.  &  R.  B.  Gumming,  (1975).    Induction  of translocation  by
       cydophosphamide in different germ cell stages of male mice: Cytological
       characterization and transmission.  Mutation Res.. 22, 375-388.

 Speit, G., W. Menz, C. Roscheisen, & B. Kdberle (1992).  Cytogenetic and molecular
       characterization of the mutagenicity of chlorabucil in V79 cells. Mutation Res..
       283.75-81.

 Sram, R.J. (1 976). Relationship between acute and chronic exposures in mutagenicity
       studies in mice.  Mutation Res.. 41, 25-42.

 Streisinger, G. (1983).  Extrapolations from species to species and from various cell
       types in assessing risks from chemical mutagens.  Mutation Res.. 114. 93-105.

 Strekalova, E.Ye (1971). Mutagenic action of ethylene oxide on mammals. Toksikol.
       Nov. Prom. Khim Veshchestv.. 1£, 72-78.
Tales, A.D., T. Grummt, M. Tomqvist, P.B. Farmer, F.J. van Dam, H. van Mossel, H.M.
       Schoemaker, S. Osterman-Golkar, C. Uebel, Y.S. Tang, A.H. Zwinderman, AT.
       Natarajan  & L Ehrenberg (1991).  Biological and chemical monitoring of
       occupational exposure to ethylene oxide. Mutation Res.. 250. 483-497.
                                    437

-------
 Thilly, W.G. (1991). Mutational spectrometry: opportunity and limitations in human
        risk assessment. In R.C. Gamer, P.B. Farmer, G.T. Steel, & A.S. Wright, (Eds.),
        Human carcinogen exposure:  Biomonitorina and risk assessment (pp. 127-
        133). New York: Oxford University Press.

 United Nations. (1986). Genetic and somatic effects of ionizing radiation.  United
        Nations Scientific Committee on the Effects of Atomic Radiation, 1986 Report
        to the General Assembly, with annexes.  (United Nations sales publication
        E.86.IX.9).  New York: United Nations.

 United Nations.  (1990).  Assessing the risk of genetic damage.  United  Nations
        Environment Programme, Nairobi.

 UNEP and ICPEMC. (1992). Assessing the risk of genetic damage. In:  D.J. Brusick,
        H.N.B. Gopalan, E. Heseltine,  J.W.  Huismans, &  P.H.M. Lohman  (Eds.),
        Gaborone, London, Sydney, Auckland:  Hodder & Stroughton.

 USEPA (1986a).  Guidelines for carcinogen risk assessment. Fed. Reg.. 51.. 33992-
        34003 (NTIS Order No. PB91 -158394).

 USEPA (1 986b). Guidelines for mutagenicity risk assessments. Fed. Rea.. §1. 34006-
        34012 (NTIS Order No. PB91 -158394).

 Vogel,  E.W., & JA Zjlstra (1987).  Mechanistic and methodological aspects of
        chemically-induced  somatic  mutation  and  recombination  in  Drosophila
       melanogastor.  Mutation Res.. 182. 243-264.

 Vogel, E.W., A. Barbin, M.J.M. Nivard, & H. Bartsch (1990). Nudeophilic selectivity of
       alkylating agents and their hypermutability in Drosophila  as predictors of
       carcinogenic potency in rodents.  Carcinoaenesis. H, 2211-2217.
Vogel, E.W., & M.J.M. Nivard (1993). Performance of 181 chemical in a Drosophila
       assay predominantly monitoring interchromosomal mttotic recombination.
       Mutaaenesis. 8, 57-81 .

Vu, V.T. & C.C. Fenselau, (1981). Identification of three alkylated nucleotide adducts
       from the reaction  of guanosine  S'-monophosphate  with  phosporamide
       mustard.  J. Am. Chem. Soc.. 103. 7362-7364.

Walker, V.E., T.R.  Fennell, P.B.  Upton,  J.P. MacNeel &  JA Swenberg  (1993).
       Molecular dosimetry of DMA and  hemoglobin adducts in mice and rats
       exposed to ethylene oxide.  Env. Health Persoec.. 99, 11-17.

Walker, V.E. & T.R. Skopek (1 993). A mouse model for the study of in vivo mutational
       spectra: sequence specificity of ethylene oxide at the HPRT locus. Mutation
       Res.. 288. 151-162.
                                    438

-------
Waters, M.D., J.W. Allen, LD. Claxton,  N.E. Garrett, S.L Huang, M.M. Moore, Y.
       Sharief, & G.H. Strauss, (1984).  Evaluation of genotoxic effects in human
       populations.  In F J. de Serres & R.W. Pero (Eds.), Individual susceptibility to
       genotoxic agents In the human population, (pp. 53-87).  New York:  Plenum
       Press.

Waters, M., J. Allen, C. Doerr, G. Erexson, V. Hasselblad, S. Huang, A. Kligerman, M.
       Moore, S. Nesnow, A. Stead, G. Strauss, J. Tabor, B. Westbrook-Collins, & J.
       Wilmer, (1986).  A preliminary investigation of the parallelogram concept in
       genetic monitoring and risk estimation.  In M. Sorsa, & H. Norppa (Eds.),
       Monitoring of Occupational Genotoxicants. (pp. 203-215). New York: Alan R.
       Uss, Inc.

Wright, A.S. (1991). Emerging strategies for the determination of human carcinogens:
       detection, identification, exposure monitoring  and risk evaluation.   In R.C.
       Garner, P.B. Farmer, G.T. Steel,  & A.S. Wright, (Eds.),  Human carcinogen
       exposure: Biomonitorina and risk assessment, (pp. 3-23). New York: Oxford
       University Press.

Wright, A.S., & W.P. Watson, (1989).  Prospective assessment of human carcinogens:
       The detection of genotoxic action in human skin. In: T.J. Slaga, A.J.P. Wein-
       Szanto, R.K. Boutwell, D.E. Stevenson, H.L Spitzer, & B. D'Motto, (Eds.), Skin
       Carcinoaenesis:  Mechanisms and human relevance (pp. 313-329). New York:
       Alan R. Uss, Inc.

Yamada,  M.,  T.  Sofuni,  & T. Nohmi  (1992).    Preferential  induction of AT-TA
       transversion,  but not deletions,  by chlorambucil at the hisG428 site of
       Salmonella typhimurium TA102. Mutation Res.. 283. 29-33.

Zeeland, AA van, G.R. Mohn, A. Neuhauser-KJaus, & U. Ehling, (1985). Quantitative
       comparison of genetic effects of ethyiating agents on the basis of DMA adduct
       formation. Use of O6-ethylguanine as a molecular dosimeter for extrapolation
       from cells  in culture to the mouse. Environ. Health Persoec.. 62,163-169.
                                     439

-------
Appendix A
Final List of Participants
                        A-1

-------
A-2

-------
                   EC/US WORKSHOP ON RISK ASSESSMENT

                     Omni Durham Hotel and Convention Center
                   201 Foster Street, Durham, North Caroina, USA
                               October 11-14,1993

                          FINAL LIST OF PARTICIPANTS
 Dr. Ilse-Dore Adter
 GSF - Forschungzentrum fur Umwelt
  und Qesundheit GmbH
 Institut fur Saugetiergenetik
 Ingotetradter Landstrasse 1
 D - W 8042 NEUHERBERG (MOnchen)
 GERMANY

 Dr. James W. Allen
 Mutagenesis and Cellular
  Toxicology Branch
 Genetic Toxicology Division (MD-68)
 U.S. Environmental Protection Agency
 Research Triangle Park, NC 27711
 USA

 Dr. Diana Anderson
 BIBRA Toxicology International
 Dept of Genetic & Reproductive
  Toxicology
 Woodmansteme Road
 UK - CARSHALTON, SURREY SMS 4DS
 UNITED KINGDOM

 Dr. Jack Bishop
 A2-10
 National Institute of Environmental
 Health Sciences
 P.O. Box 12233
 Research Triangle Park, NC 27709
 USA

Dr. David Brusick
Hazteton Labs America, Inc.
9200 Leesburg Pike
Vienna, VA 22180
USA
Dr. Larry Claxton*
Genetic Toxicology Division
MD-68
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
USA

Ms. Judi Cochrane
Pathology Department
355 Rosenau Hall 201H
University of North Carolina
Chapel Hill, NC 27599
USA

Dr. Kerry DearfiekJ
Health Effects Division
Office of Pesticide Programs (H7509C)
U.S. Environmental Protection Agency
401 M St, SW
Washington, DC 20460
USA

Dr. Vicki Dellarco
Human Health Assessment Group
Office of Health and Environmental
 Assessment (RD-689)
U.S. Environmental Protection Agency
401 M St, SW
Washington, DC 20460
USA

Dr. Frederick J. de Serres*
Research Triangle Institute
P.O. Box 12194
Research Triangle Park, NC 27709
USA
                                      A-3

-------
 Dr. George R. Douglas
 Health Canada
 Environmental Health Center
 Tunney's Pasture
 Ottawa, Ontario K1A OL2
 CANADA

 Dr. Udo Ehling
 Instrtut fur Genetik Gesellschaft
  fOr Strahten- und Umweltforschung
 Neuherberg/
 Ingoistradter Landstrasse 1
 D - 8042 OBERSCHLEOSSHEIM
 GERMANY

 Dr. Lars Ehrenberg
 Department of Radiobtotogy
 University of Stockholm
 S -10691 STOCKHOLM
 SWEDEN
 Dr. R. Colin Gamer
 The Jack Birch Unit
  of Environmental Carcinogenesis
 University of York
 Heslington Road
 UK-YORKY015DD
 UNITED KINGDOM

 Dr. Walderico Generoso
 Oak Ridge National Laboratory
 Biology Division
 P.O. Box Y
 Oak Ridge, TN 37831-6050
 USA

 Dr. Susan E. Lewis
 Research Triangle Institute
 P.O. Box 12194
 Research Triangle Park, NC 27709
 USA

 Dr. Barry Margolin*
 Dept. of Biostatistics
CB# 7400 Rosenau Hall
University of North Carolina
Chapel Hill, NC 27599
USA
Dr. Martha Moore
Mutagenesis and Cellular
 Toxicology Branch
Genetic Toxicology Division (MD-68)
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
USA

Dr. A.T. 'Nat* Natarajan
Dept. of Radiation Genetics
 & Chemical Mutagenesis
State University of Leiden
Wassenaarseweg 72
NL - 2300 RA LEIDEN
THE NETHERLANDS

Dr. Canice Nolan
Commission of the European
 Communities
DG XII Science, Research
 and Development
200 Rue de la  Loi
B -1049 BRUSSELS
BELGIUM

Dr. Siv Osterman-Golkar
Department of Radiobiology
University of Stockholm
S -10691  STOCKHOLM
SWEDEN

Dr. Patricia Ostrosky-Wegman
Institute de Investigaciones Biomedicas
UNAM
P.O. Box 105-285, PoJanco
Mexico D.F. CP 11581
MEXICO

Dr. Heinrich Ott
Commission of the European
 Communities
DG XII Science, Research
 and Development
200 Rue de la Loi
B -1049 BRUSSELS
BELGIUM
                                     A-4

-------
 Dr. Julian Preston
 GIFT
 P.O.  Box 12137
 6 Davis Dr.
 Research Triangle Park, NC 27709
 USA

 Dr. Lawrence W. Reiter
 Health Effects Research Laboratory
 MD-51
 U.S.  Environmental Protection Agency
 Research Triangle Park, NC 27711
 USA

 Dr. Gary Sega
 Oak  Ridge National Laboratory
 Analytical Chemistry Division
 4500 South Building
 P.O.  Box 2008
 Oak  Ridge, TN 37831-6120
 USA

 Dr. Paul B. Selby
 Oak  Ridge National Laboratory
 Biology Division
 Bldg. 9210, MS-8077
 Oak Ridge, TN  37831
 USA

 Dr. Michael Shelby
 E4-03
 National Institute of Environmental
 Health Sciences
 P.O. Box 12233
 Research Triangle Park, NC 27709
 USA

 Dr. Thomas R. Skopek
 (represented by Ms. Judi Cochrane)
 Pathology Department
 355 Rosenau Hall 201H
 University of North Carolina
 Chapel Hill, NC  27599
 USA

 Dr. Marja Sorsa
 Finnish Institute of Occupational
 Health
Topeliuksenkatu 41 a A
F1 - 00250  HELSINKI
FINLAND
 Dr. A.D. mMT Tales
 Department of Radiation Genetics
  and Chemical Mutagenesis
 State University of Leiden
 Wassenaarseweg 72 / P.O. Box 9503
 NL - 2300 RA LEIDEN
 THE NETHERLANDS

 Dr. Ekkehatt W. Vogel
 Dept. of Radiation  Genetics &
  Chemical Mutagenesis
 State University of  Leiden
 Wassenaarseweg 72
 NL - 2333 AL LEIDEN
 THE NETHERLANDS

 Dr. Michael D. Waters
 Health Effects Research Laboratory
 MD-51A
 U.S. Environmental Protection Agency
 Research Triangle  Park, NC 27711
 USA

 Dr. Alan S. Wright
Tunstall Laboratory
Shell Research Limited
Sittingboume Research Centre
 UK - SITTINGBOURNE,  KENT ME9 SAG
UNITED KINGDOM
*lnvted Observer
                                      A-5

-------
A-e

-------
Appendix B
Agenda
                    B-1

-------
B-2

-------
                 EC/US WORKSHOP ON RISK ASSESSMENT
                 Omni Durham Hotel and Convention Center
               201 Foster Street, Durham, North Carolina, USA
                           October 11-14,1993
Monday. October 11
      8:00-10:00 p.m.    Fteceptk>n at the Shef aton LMiversity Center Hotel in the
                         Executive dub, 2800 MkJoMon Avenue, Durham
                         (Compliments of Hazleton Labs America, he. -
                         Dr. D. Brustek)
Tuesday. October 12
Morning Session (8:45 am. - 12:30 p.m.) Room 107 - M. Waters and C. Nolan,
Session Co-Chairs
      8:45 - 9:00        Introductory Remarks              H. Ott, L Reiter
      9:00 - 9:25        Parallelogram Concept             A. Wright
      9:25 - 9:30        Discussion
      9:30 - 9:55        ICPEMC Efforts                   D. Brusick
      9:55 -10:00       Discussion
     10:00 -10:30       Break, Compliments of the Omni Durham Hotel and
                       Convention Center
     10:30 -10:55       Regulatory Perspective             K. Dearfield
     10:55-11:00       Discussion
     11:00 -11:25       Research Background for EO        L Ehrenberg
                      Assessment
     11:25-11:30       Discussion
     11:30 -11:55      EO Mutagenkaty Risk              V. Dellarco
                      Assessment
     11:55 -12:00      Discussion
                                  B-3

-------
      12:00 -12:25      Critique of EO                     J. Preston
                        Assessment
      12:25 -12:30      Discussion
      12:30-  2:00       Lunch
 Afternoon Session (2:00 - 5:30 p.m.) Room 107 - M. Shelby and M. Sorsa, Session
 Co-Chairs
                        Data Summaries of Chemicals (Effort to apply
                        parallelogram and/or alternative methods)
       2:00 - 2:30       Ethytene oxide                     A. Natarajan
       2:30 - 2:45       Discussion
       2:45 - 3:15       Acrylamide                        K. Dearfield
       3:15- 3:30       Discussion
       3:30- 4:00       Break
       4:00- 4:30       1,3-Butadiene                      I-D. Adler
       4:30- 4:45       Discussion
       4:45- 5:15       Cydopnospnamide                 D.Anderson
       5:15- 5:30       Discussion

Evening (6:00 - 9:00 p.m.)
      6:00 -  9:00       Social and Dinner, Parizade Restaurant, 2200 W. Main St.,
                        Durham
                        (Meet in hotel lobby at 6:00 p.m. for transportation to the
                        restaurant)
Wednesday. October 13
Morning Session: Chemical Work Groups (9:00 a.m. -12:00 p.m.)
     Data Presentations (parallelogram component contributions-see matrix)
     Parallelogram Calculations (chairpersons)
                                    B-4

-------
      Identification of Data Gaps (chairpersons)
      Alternative Approaches (rapporteurs)
      10:00 -10:30
Break, Compliments of the Omni Durham Hotel and
Convention Center


Etnylene oxide
Chair: A Natarajan
Rapp.: J. Preston
Aaylamide
Chain K. Dearfield
Rapp.: G. Douglas
1.3-Butad»ne
Chain I-D. Adter
Rapp.: M. Sorsa

cyctopnospnamKie
Chain D. Anderson
Rapp.: P. Selby
Mouse Germ
Gets
W. Generoso
U. Ehling
I-D. Adler
J. Bishop
Adducts
L Ehrenberg
G. Sega
S. O.-Golkar
C. Gamer
Mouse and
Human
Somatic
Cefe
A. Tates
M. Moore
T. Skopek
P. Ostrosky
Room No.
and
Weekday
Boardrm. I
(Wedmiur)
Boardrm. II
(Wed/Thur)
Rm. 108
(Wed/Thur)
Rm.106
(Wed) and
Rm. 105
(Thur)
Afternoon Session (12:30 - 5:00 p.m.) - J. Bishop and U. Ehling, Session Co-Chairs
     12:30-  2:00       Lunch
      2:00- 3:00       Status  Reports to  Plenary-Rapporteurs (see  above)
                       Room 107
      3:00- 3:30       Break
      3:30- 5:00       Chemical Work Groups-Meet lor further discussion, report
                       preparation
                                   B-5

-------
 Evening
      6:00 -  9:30        Excursion and Dinner, Angus Barn, Ltd., Highway 70 East
                        & Airport Boulevard, Raleigh
                        (Meet in hotel lobby at 5:30 p.m. for transportation to the
                        restaurant)
 Thursday. October 14

 Morning Session:  Chemical Work Groups - Prepare Written Reports (9:00 am. -12:30
 p.m.) - J. Allen and A. Wright, Session Co-Chairs

      9:00 -10:00       Finalize Reports

      10:00 -10:30       Break, Compliments of the Omni Durham Hotel and
                        Convention Center

      10:30 -12:30       Rapporteurs' Presentations of Work Group Reports to
                        Pierian/- Room 107

                        EO                              J. Preston

                        Acrylamide                       G. Douglas

                        1,3-Butadiene                    M. Sorsa

                        Cydophosphamide                P. Selby

      12:30-  1:30       Lunch
Afternoon Session (1:30 - 4:30 p.m.) Room 107 - M. Moore and I-D. Adter, Session
Co-Chairs

                       Cross-cutting Papers Focusing on Several General Topics
                       Pertinent to Somatic and Germ CeN Mutation

      1:30- 2:00       Differential  Repair of               E. Vogel and
                       Chemically-irKluced Damage:       A. Natarajan
                       Somatic vs. Germ Cells

      2:00 - 2:30       Chemically-induced Mutation:       J. Allen, S. Lewis
                       Comparative Outcome in Somatic   M. Moore and
                       vs. Germ Cells                    U. Ehling
                                   B-6

-------
2:30- 3:00        Qualitative and Quantitative         M. Waters and
                  Results from Short-Term Tests       I-D. Adier

3:00- 3:30        Break

3:30- 4:00        Chemicals for Future Study of       M.Shelby and
                  Heritable Genetic Damage          M. Waters
                  in Human Populations

4:00 - 4:30        Wrap-up                          D. Brusick
                                                   C. Nolan
                              B-7

-------
B-8

-------