£fA-7f-t- -/?- 93 -
         Guidelines for
Developing Community
   Emergency Exposure
   Levels for Hazardous
           Substances
        COMMITTEE ON TOXICOLOGY
       NATIONAL RESEARCH COUNCIL

-------

-------
    Guidelines for Developing
Community Emergency Exposure
 Levels for Hazardous Substances
    Subcommittee on Guidelines for Developing
  Community Emergency Exposure Levels (CEELs)
           for Hazardous Substances

           Committee on Toxicology
   Board on Environmental Studies and Toxicology
         Commission on Life Sciences
           National Research Council
        NATIONAL ACADEMY PRESS
           Washington, D.C.  1993

-------
NATIONAL ACADEMY PRESS 2101 Constitution Aye., N.W., Washington, B.C. 20418

NOTICE- The project that is the subject of this report was approved by the Governing Board of the
National Research Council, whose members are drawn from the councils of the National Academy
of Sciences, the National Academy of Engineering and the Institute of Medicine. The members of
the committee responsible for the report were chosen for their special competenc.es and with regard
for appropriate balance.                                                             .
  This  report has been reviewed  by a group other than the authors accordmg o  procedures
approved fay a Report Review Committee consisting of members of the National Academy of
Sciences, the National Academy of Engineering, and the Institute of Medicine.
  The  National Academy  of Sciences  is a private,  non-profit, self-perpetuatmg society of
distinguished scholars engaged in scientific and engineering research, dedicated to the fortherance
of science and technology and to their use for the general welfare. Upon the authonty of the charter
granted to it by the Congress in 1863, the Academy has a mandate that requires ,t to^advise the
federal government on scientific and technical matters. Dr. Frank Press is president of the National

^The'NatioMl Academy of Engineering was established in 1964, underthe charter of the National
Academy of Sciences, as a parallel organization of outstanding engineers.  It is autonomous in its
administration and in the selection of its member.!, sharing with the National Academy of Sciences
the responsibility for advising the federal government.  The National Academy of Engmeenng also
sponsors engineering programs aimed at meeting rational needs, encourages education and research
and recognfees the superior achievements of engineers.  Dr.  Robert M. White  ,s president of the
National Academy of Engineering.
   The Institute of Medicine was established in 1970 by the National Academy of Sciences to secure
the services of eminent members of appropriate professions  in the examination of pohcy matters
pertaining to  the  health of the public.  The  Institute  acts under the responsibility given to the
National Academy of  Sciences by its  congressional  charter to be an adv.ser to  the federal
government and, upon its own initiative, to identify issues of medical care, research, and educaUon.
Dr. Kenneth I. Shine is president of the Institute of Medicine.
   The National Research Council was organized by the National Academy of Sciences in 1916 to
associate the broad community of science and technology with  the Academy's purposes of furthering
knowledge and advising the federal government.  Functioning in accordance with general policies
determined by the Academy, the Council has become  the principal operating agency of both the
National Academy of Sciences and the National Academy of Engineering in providing services to
the government,  the public, and  the scientific and engineering communities.  The Council is
administered  jointly  by both Academies and the Institute  of Medicine.  Dr. Frank Press and
Dr. Robert M. White are  chairman and vice chairman, respectively, of the  National Research

  'rhe'project was supported by the U.S. Environmental Protection Agency under contract No. CR-
815682-01 -0 and the Agency for Toxic Substanc es and  Disease Registry.
   Additional copies of this report are available from the Board on Environmental Studies and
Toxicology, 2101 Constitution Avenue, N.W., Washington, D.C.  20418
 Copyright 1993 by the National Academy of Sciences.

 Printed in the United States of America

-------
   Subcommittee on Guidelines for
   Developing Community Emergency
   Exposure Levels  (CEELs) for
   Hazardous Substances
JOHN DOULL (Chair), University of Kansas Medical Center, Kansas
  City, Kan.
R. HAYS BELL, Eastman Kodak Company, Rochester, N.Y.
CHARLES E. FEIGLEY, University of South Carolina, Columbia,
  S.C.
DONALD E. GARDNER, ManTech Environmental Technology, Inc.,
  Research Triangle Park, N.C.
IAN GREAVES, University of Minnesota, Minneapolis, Minn.
ROGENE F. HENDERSON, Lovelace Biomedical and Environmental
  Research Institute, Albuquerque, N.M.
DAVID P. KELLY, E.I. DuPont de Nemours & Co., Newark, Del.
RALPH L. KODELL, Food and Drug Administration, National Center
  for Toxicological Research, Jefferson, Ark.
TRENT R. LEWIS, Consultant, Cincinnati, Ohio
BERNARD M. WAGNER, New York University Medical Center,
  Orangeburg, N.Y.
BAILUS WALKER, JR., University of Oklahoma, Oklahoma City,
  Okla.

Staff

RICHARD D. THOMAS, Program Director
KULBIR S. BAKSffl, Project Director
RUTH E. CROSSGROVE, Editor
BEULAH S. BRESLER, Senior Editorial Assistant (until June 1992)
CATHERINE M. KUBEK, Senior Program Assistant
RUTH P. DANOFF, Project Assistant
                          in

-------
                                              Committee on Toxicology
                                          ROGENE HENDERSON (Chair), Lovelace Biomedical and
                                            Environmental Research Institute, Albuquerque, N.M.
                                          BERNARD WAGNER (Vice-Chair), Wagner Associates, Inc.,
                                            Millburn, N.J.
                                          R. HAYS BELL, Eastman Kodak Company, Rochester, N.Y.
                                          DEAN E. CARTER, University of Arizona, Tucson, Ariz.
                                          CHARLES E. FEIGLEY, University of South Carolina, Columbia,
                                            S.C.
                                          DONALD E. GARDNER, ManTech Environmental Technology, Inc.,
                                            Research Triangle Park, N.C.
                                          MARY ESTHER GAULDEN, University of Texas Southwestern
                                            Medical School, Dallas, Tex.
                                          WALDERICO GENEROSO, Oak Ridge National Laboratory, Oak
                                            Ridge, Tenn.
                                          IAN GREAVES, University of Minnesota, Minneapolis, Minn.
                                          RONALD A. HTTES, Indiana University, Bloomington, Ind.
                                          CAROLE A. KIMMEL, Environmental Protection Agency,
                                            Washington, D.C.
                                          RALPH L. KODELL, Food and Drug Administration, National Center
                                            for Toxicological Research, Jefferson, Ark.
                                          LOREN D. KOLLER, Oregon State University, Corvallis, Oreg.
                                          ERNEST EUGENE MCCONNELL, Raleigh, N.C.
                                          MICHELE A. MEDINSKY, Chemical industry Institute of Toxicology,
                                            Research Triangle Park, N.C.
                                          ROBERT SNYDER, Rutgers University, Piscataway, NJ.
                                          BAILUS WALKER, JR., University of Oklahoma, Oklahoma City,
                                            Okla.
                                          HANSFETER R. WTTSCfll, University of California, Davis, Calif.
                                          GERALD N. WOGAN, Massachusetts Institute of Technology,
                                            Cambridge, Mass.
                                          GAROLD S. YOST, University of Utah, Salt Lake City, Utah
                                                                      IV
f, ! :.-:,

-------
Committee on Toxicology

RICHARD D. THOMAS, Program Director
KULBIR S. BAKSffl, Senior Program Officer
MARVIN A. SCHNEIDERMAN, Senior Staff Scientist
RUTH E. CROSSGROVE, Editor
BEULAH S. BRESLER, Senior Editorial Assistant (until June 1992)
CATHERINE M. KUBEK, Senior Program Assistant
RUTH P. DANOFF, Project Assistant

-------
    Board on Environmental Studies
    and Toxicology
PAUL G. RISSER (Chair), University of Miami, Oxford Ohio
FREDERICK R. ANDERSON, Cadwalader, Wickersham & Taft,

JO^C^ABLAR, in, McGill University School of Medicine,
  Montreal, Quebec, Canada
GARRY D  BREWER, University of Michigan, Ann Arbor, Micti.
JOHN CAIRNS, JR., Virginia Polytechnic Institute and State
  University, Blacksburg, Va.
EDWIN H. CLARK, Department of Natural Resources and
  Environmental Control, State of Delaware, Dover, Del.
JOHN L. EMMERSON, Lilly Research Laboratories, Greenfield, Ind.
ROBERT C. FORNEY, Unionville, Pa.
ALFRED G. KNUDSON, Fox Chase Cancer Center, Philadelphia, Pa.
KAI LEE, Williams College, Williamstown, Mass
GENE E. LIKENS, The New York Botanical Garden, Millbrook, N.Y.
JANE LUBCHENCO, Oregon State University, Corvalhs, Oreg.
DONALD MATTISON, University  of Pittsburgh, Pittsburgh, Pa.
HAROLD A. MOONEY. Stanford University, Stanford, Calif.
GORDON ORIANS, University of  Washington, Seattle, Wash.
FRANK L. PARKER,  Vanderbilt University, Nashville, Term., and
   Clemson University, Anderson, S.C.
GEOFFREY PLACE, Hilton Head, S.C.
MARGARET M. SEMINARIO, AFL/CIO, Washington, D.C.
I  GLENN SIPES, University of Arizona, Tucson, Ariz.
BAILUS. WALKER, JR., University of Oklahoma, Oklahoma City,
   Okla
WALTER J. WEBER, JR., University of Michigan, Ann Arbor, Mich.

-------
Board on Environmental Studies and Toxicology Staff

JAMES J. REISA, Director
DAVID J. POLICANSKY, Associate Director and Program Director
  for Natural Resources and Applied Ecology
RICHARD D. THOMAS, Associate Director and Program Director for
  Human Toxicology and Risk Assessment
LEE R. PAULSON, Program Director for Information Systems and
  Statistics
RAYMOND A. WASSEL, Program Director for Environmental
  Sciences and Engineering
                             Vtl

-------
    Commission on Life Sciences
THOMAS D. POLLARD, Johns Hopkins Medical School, Baltimore,
  Md.
BRUCE M. ALBERTS, University of California, San Francisco, Calif.
BRUCE N. AMES, University of California, Berkeley, Calif.
J. MICHAEL BISHOP, Hooper Research Foundation, University of
  California Medical Center, San Francisco, Calif.
DAVID BOTSTEIN, Stanford University School of Medicine,
  Stanford, Calif.
MICHAEL T. CLEGG, University of California, Riverside, Calif.
GLENN A. CROSBY, Washington State University, Pullman, Wash.
LEROY E. HOOD, University of Washington, Seattle, Wash.
MARIAN E. KOSHLAND, University of California, Berkeley, Calif.
RICHARD E. LENSKI, University of Oxford, Oxford, United
  Kingdom
STEVEN P. PAKES, Southwestern Medical Center, Dallas, Tex.
EMIL A. PFTTZER, Hoffmann-La Roche Inc., Nutley, N.J.
MALCOLM C. PIKE, University of Southern California School of
  Medicine, Los Angeles, Calif.
PAUL G. RISSER, Miami University, Oxford, Ohio
JOHNATHAN M. SAMET, University of New Mexico School of
  Medicine, Albuquerque, N.M.
HAROLD M. SCHMECK, JR., Armonk, N.Y.
CARLA J. SHATZ, University of California, Berkeley, Calif.
SUSAN S. TAYLOR, University of California at San Diego, La Jolla,
  Calif.
P. ROY VAGELOS, Merck and Company, Inc., Whitehouse Station,
  N.J.
TORSTEN N. WBESEL, Rockefeller University, New York, N.Y.

Staff

ALVIN G. LAZEN, Acting Executive Director
                            viu

-------
     Other Recent Reports  of
     The Board on Environmental
     Studies and Toxicology
Issues in Risk Assessment (1993)
Setting Priorities for Land Conservation (1993)
Protecting Visibility in National Parks and Wilderness Areas (1993)
Biologic Markers in Immunotoxicology (1992)
Dolphins and the Tuna Industry (1992)
Environmental Neurotoxicology (1992)
Hazardous Materials on the Public Lands (1992)
Science and the National Parks (1992)
Animals as Sentinels of Environmental Health Hazards (1991)
Assessment  of the U.S. Outer Continental Shelf Environmental Studies
  Program, Volumes I-IV (1991-1993)
Human Exposure Assessment for Airborne Pollutants (1991)
Monitoring Human Tissues for Toxic Substances (1991)
Rethinking the Ozone Problem in Urban and Regional Air Pollution
  (1991)
Decline of the Sea Turtles (1990)
Tracking Toxic Substances at Industrial Facilities (1990)
Biologic Markers in Pulmonary Toxicology (1989)
Biologic Markers in Reproductive Toxicology (1989)
           Copies of these reports may be ordered from
                  the National Academy Press
                       (800) 624-6242
                             IX

-------

-------
             Acknowledgments
  The Subcommittee  on  Guidelines  for  Developing  Community
Emergency Exposure  Levels  (CEELs) for  Hazardous  Substances
gratefully acknowledges the valuable assistance provided by Mr. Ralph
C. Reynolds and Mr. David P. Richardson of Eastman Kodak Company
for preparing sections of this report.
  The subcommittee also acknowledges the following scientists for the
valuable contributions to the work of the subcommittee by making oral
presentations and/or providing written material:

  Dr. Dean Branson, Dow Chemical Company
  Dr. Edward Calabrese, University of Massachusetts
  Dr. Roger Garrett, U.S. Environmental Protection Agency
  Dr. Daniel Guth, U.S. Environmental Protection Agency
  Dr. Annie Jarabek, U.S. Environmental Protection Agency
  Dr. Gary Kimmel, U.S. Environmental Protection Agency
  Dr. Robert MacPhail, U.S. Environmental Protection Agency
  Dr. Alan Susten, Agency for Toxic Substances and Disease
    Registry
  Dr. Paul Tobin, U.S. Environmental Protection Agency
                             XI

-------

-------
       Preface
Massive amounts of extremely hazardous substances (EHSs) can be re-
leased accidentally as a result of chemical spills, industrial explosions,
fires, or accidents involving railroad cars and trucks transporting EHSs.
The people in communities surrounding industrial facilities where EHSs
are manufactured, used, or stored, and in communities along the nation's
railways and highways are potentially at risk of being exposed to air-
borne EHSs during accidental releases. The U.S. Environmental Protec-
tion Agency (EPA) has identified 366 EHSs based on acute lethality data
in rodents.
   Various governmental agencies and private organizations have recom-
mended inhalation exposure limits for airborne substances for workplace
exposures (8 hr/day, 5 days/week, 50 weeks/year for up to 50 years of
worklife) and for ambient exposures to low levels of common air pollut-
ants (24 hr/day for lifetime of a person). However, inhalation exposure
limits for  emergency exposures to EHSs (which might occur once in a
lifetime of a person) have not been recommended for most EHSs. Emer-
gency exposure limits for EHSs would be useful to local emergency
planning  committees, as well as state and federal  agencies involved in
emergency planning and public health.
   At the request of EPA and the Agency for Toxic Substances and Dis-
ease Registry (ATSDR), the National Research Council's Committee on
Toxicology (COT) convened the Subcommittee on Guidelines for Devel-
oping Community Emergency Exposure Levels for Hazardous  Substan-
ces, which prepared this report.  The members of the subcommittee had
expertise in toxicology, epidemiology, medicine, pharmacology, chemis-
try, industrial hygiene, mathematical modeling, biostatistics,  and risk
assessment.
   This report addresses risk assessment of short-term exposures to high
levels of chemical toxicants.  It describes the criteria and methods for
developing community emergency  exposure levels for EHSs for use by
                               xin

-------
xiv    PREFACE

many organizations, including EPA, the ATSDR, state and local govern-
ments, and chemical industries.                           .   .
   The subcommittee gratefully acknowledges the following scientists for
making presentations on issues relevant to setting emergency exposure
limits for hazardous substances: Dr.  George Alexeef of California De-
partment of Health Services; Dr. Desai Branson of Dow Chemical Com-
pany Dr Edward Calabrese of the University of Massachusetts; Dr.
Alan'susten of ATSDR; and Drs. Roger Garrett, Daniel Gulh, Annie
Jarabek,  Gary Kimmel, Robert MacPhail, and Paul Tobm of EPA.
   This report could not have been produced without the untiring efforts
of the National Research Council staff: Richard D. Thomas, program
director; Ruth E. Crossgrove, editor;  Beulah Bresler, senior editorial
assistant; Catherine Kubik, senior program assistant; and Ruth Danoff,
project assistant.                                          '.      '
   Finally  the subcommittee gratefully acknowledges the persistence,
patience, and expertise of Kulbir S. Bakshi, project director of the sub-
committee, for preparing the white paper that served as the building
block for the report and for bringing this report to its final form.

               John Doull, Chair
               Subcommittee on Guidelines for Developing
               Community Emergency Exposure Levels (CEELs)
               for Hazardous Substances

               Rogene F. Henderson, Chair
               Committee on Toxicology

-------
       Contents
     EXECUTIVE SUMMARY                              1

 1    INTRODUCTION                                     9
     Acceptable Inhalation Exposure Limits Recommended
        by Agencies and Organizations                        11
     Limitations of the Recommended Exposure
       Limits for Use as CEELs                             14

2    CRITERIA FOR ESTABLISHING CEELS               19
     Populations of Concern                                 19
     Exposure Periods and Patterns                           20
     Nature and Severity of Toxic Effects                      20
     Proportion of the Exposed Population Showing
       Each Category of Toxic Effect                         21

3    SOURCES OF DATA FOR DEVELOPING CEELS       23
     Human Data                                          24
     Animal Toxicity Studies                                29
     Short-Term In Vitro Tests                              30
     Chemical and Physical Characteristics of Toxicants           33

4    TYPES OF DATA USED IN DEVELOPING CEELS       35
     Pharmacokinetics and  Metabolism                        35
     Toxicity End Points: Humans and Animals                 44
     Other Organ Effects                                   54

5    RISK ASSESSMENT                                 57
     Overview                                            57
     Noncarcinogenic Effects                                61
     Carcinogenic Effects                                   67
                            xv

-------
jevi     CONTENTS
                     '
6    DEVELOPMENT OF CEELS
     Evaluation of Data
     Presentation and Treatment of Data
     Translation of Animal Data to Humans
     Threshold Toxicity
     Nonthreshold Effects
     Calculation of CEELs and Rationale
     Presentation of CEELs
75
76
84
86
88
93
95
96
REFERENCES
                                                        99

-------
       Executive Summary
In the Bhopal disaster of 1984, 2,000 residents living near a chemical.
plant were killed and approximately 20,000 more suffered irreversible
damage to their eyes and lungs following an accidental release of methyl
isocyanate.  The toll was particularly high because the community had.
little idea what chemicals were being used at the plant, how dangerous
they might be, and what steps to take in case of emergency. This trag-
edy served to focus international attention on the need for governments
to identify hazardous substances and to assist local communities in plan-
ning how to deal with emergency exposures.
   In the United States, the Superfund Amendment and Reauthorization
Act (SARA) of 1986 required the U.S. Environmental Protection Agency
(EPA) to identify extremely hazardous substances (EHSs) and, in cooper-
ation with the Federal Emergency Management Agency, to assist local
emergency planning committees (LEPCs)  in developing emergency re-
sponse plans for sites where EHSs are produced, stored, transported, or
used. SARA also required the Agency for Toxic Substances and Disease
Registry (ATSDR), part of the Public Health Service, to determine
whether chemical substances that have been identified at hazardous  waste
sites or in the environment present a public health concern.
   As a first step in  assisting LEPCs, EPA has identified 366 EHSs,
based largely on their "immediately dangerous to life and health" (IDLH)
values in experimental animals.  However, few of these substances have
adequate toxicity information from human studies.  And while several
public and private groups have established  exposure limits  for  some
substances and some exposures (e.g., workplace or ambient air quality),
these limits are not easily or directly translated into the kind  of  limits
required for emergency exposures, which  typically involve exposure at
high levels but of short duration, usually less than 1 hour, and only once
hi a lifetime.

-------
2     CEELS FOR HAZARDOUS SUBSTANCES
                       '!•          I            ,      ,     :
   Consequently, EPA and ATSDR requested that the Committee on
Toxicology (COT) of the National Research Council  (NRC) develop
guidelines that could be used to develop community emergency exposure
levels (CEELs) for EHSs, in order to assist LEPCs in developing plans
for responding to sudden exposures to toxic substances. COT established
the Subcommittee on Guidelines for Developing Community Emergency
Exposure Levels for Hazardous Substances to respond to this request.
   The subcommittee's report is intended to be useful to EPA, ATSDR,
state and local governments, LEPCs, and private industry.  It provides
step-by-step guidance for EPA, ATSDR, and state and local governments
in setting CEELs for EHSs: what data are needed, what data are avail-
able, how to evaluate them, and how to present the results. This should
also assist LEPCs in developing emergency response plans to deal with
accidental releases—establishing the quantities of EHSs that can be safely
stored, estimating evacuation zones, and taking remedial action to protect
the health of their citizens.
         SETTING EMERGENCY EXPOSURE LIMITS

   CEELs are ceiling exposure values for the public (i.e., a ceiling is a
 concentration of a substance that should never be exceeded), and they are
 applicable to emergency exposures of foreseeable magnitude and duration
 (usually 1 to 8 hours, although few exposures will exceed 1 hour). The
 subcommittee determined that three CEELs are needed for each EHS,
 thus providing a series of exposure values that provides more information
 than a single value (see Table 1).

   • CEEL-1 refers to the concentration of an airborne substance (such
 as a gas, vapor, or aerosol) below which direct toxic effects are unlikely
 to lead to discomfort in the exposed population, but above which discom-
 fort becomes increasing common—for example, eye and nose irritation
 or headaches (the description of CEEL values for a chemical must speci-
 fy the symptoms to be expected).
   • CEEL-2 refers to the concentration below which direct toxic effects
 are unlikely to lead to disability,  but above which disability becomes
 increasingly common—for example, severe eye or respiratory irritation,
 disorientation, and organ damage.

-------
                                     EXECUTIVE SUMMARY    3

   • CEEL-3 refers to the concentration below which  death  or  life--
 threatening effects are unlikely, but above which death or life-threatening
 effects  become  increasingly likely—for  example,  pulmonary  edema,
 cardiac failure, or cancer.

   These levels take into consideration the wide range of human  respon-
 ses to toxic substances.  CEELs must be set low enough to protect most
 of the population that might be exposed, including those with increased
 susceptibilities such  as  children,  pregnant women, and persons with
 specific illnesses such as asthma.  However, CEELs must also  be high
 enough to minimize false alarms and over-response.  As a result, they
 cannot be expected to protect all individuals in a community, particularly
 those with unusually high or idiosyncratic susceptibilities.
   Most existing exposure limits apply to specific populations, low con-
 centrations, and  long durations.  Hence, workplace standards apply to
 healthy workers  exposed for 8 hours per day, 5 days per week, for up
 to 50 years of working life; ambient air quality standards apply to the
 entire  community, including susceptible  individuals, exposed  for  24
 hours per day for their entire lives. Because of differences in duration, •
 concentration, and population,  CEELs cannot be derived from these
 standards by simple extrapolation. Two organizations have issued stan-
 dards that might be more useful in developing CEELs:  the NRC's short-
 term public emergency  guidance  levels (SPEGLs)  and the American
 Industrial Hygiene Association's 1-hour emergency response planning
 guidelines (ERPGs).  However, SPEGLs and ERPGs have been devel-
 oped for a  limited number of hazardous substances—4 and 43,  respec-
 tively—while CEELs must be developed for 366 EHSs.
             CRITERIA FOR DEVELOPING CEELS

  In the absence of existing standards, what criteria and methods are to
be used in developing CEELs for hazardous substances? Such standards
must reflect the relationships among four factors:

  •  Population potentially exposed. In the case of CEELs, this means
the general population, including susceptible individuals but not neces-
sarily those who are seriously debilitated or hypersusceptible.

-------
4     CEELS FOR HAZARDOUS SUBSTANCES

   •  Degree and pattern of exposure! Duration and concentration should
reflect realistic scenarios for accidents  involving the release of EHSs,
including inventories, location, dispersion, and weather.
   •  Nature, severity, and reversibility of the anticipated effects.  Acute
health effects include mortality and adverse clinical signs and symptoms
such as pulmonary effects, effects on eyes and skin, and neurobehavioral
effects. Long-term effects mat also need to be considered include repro-
ductive and developmental toxicity, mutagenicity, and carcinogenicity.
   •  Proportion of those exposed who  are subject to toxic effects.  At
each threshold concentration (CEEL-1,  CEEL-2, CEEL-3), a small per-
centage of the population might exhibit  effects. Because this proportion
cannot be predicted precisely, it is necessary to build adequate safety
factors into the CEELs.


             DATA NEEDS ANT' DATA SOURCES

   Because there are so few detailed, descriptions of general community
exposures, those  developing CEELs will  need  to conduct a thorough
literature search for other lexicological  data that can be used to estimate
CEELs.   Human data should be emphasized; they  are most likely to
come from case studies of exposed individuals or epidemiological studies
of acutely exposed populations, usually from workplace accidents. These
studies are retrospective, however, so they seldom provide reliable meas-
urements of actual exposure.  Controlled-exposure studies using human
volunteers can  define thresholds for odor and various toxic effects, but
relatively few such studies have been conducted. The resulting lack of
human data presents a serious problem in developing emergency ex-
posure guidelines.
   In the absence of human data, risk assessors must depend on data from
animal studies  and other laboratory results.  These experimental data
have the advantage that the dose is well  known and the conditions can be
closely controlled and monitored.  However,  animals do not always
respond to a toxin exactly as humans would, and some species are known
to be highly susceptible to specific substances and effects.  Nevertheless,
well-designed animal studies can provide a great deal of information
from which CEELs can be developed,  including

-------
                                      EXECUTIVE SUMMARY     S

   •  Organs or systems likely to be affected.
   •  Specific nature of the toxic effects.
   •  Dose-response relations.
   •  Possible recovery or reversibility.
   •  Cumulative effects (e.g., neurotoxicity and cancer).
   •  Effects of interaction with  other pollutants.

   Other useful information can  come from so-called  "short-term"  or in
 vitro tests. These techniques are particularly useful for evaluating poten-
 tial carcinogens; more than 100  short-term tests have  been developed to
 detect chemicals that  might be carcinogenic in  animals  and humans.
 Additional information can come from the molecular structure of an EHS
 (which can determine its chemical  activity) and from its pharmacokinet-
 ics—how it is absorbed, distributed, metabolized, and eventually excreted
 by the body.
                       RISK ASSESSMENT

   Estimation of CEELs requires anticipation of a wide range of adverse
effects, from headaches and eye irritation to pulmonary edema and death.
However, the techniques of quantitative risk assessment are rarely used
for effects other than cancer—for example, estimating the exposure level
corresponding to a lifetime cancer risk of one in 1 million.  This expo-
sure also corresponds to a CEEL-3, the concentration  at  which life-
threatening effects become more likely. Clearly, other approaches will
be needed to estimate CEELs for noncarcinogens, materials that have
toxic effects other than causing cancer.
   The use of animal data to establish the toxicity of a chemical is called
hazard evaluation. This process involves four steps:

   • Identify the  adverse effects  produced by  acute exposure to  the
agent.
   • Establish dose-response relationships for each of the effects,  ideally
using inhalation exposure route at different rates and durations.
   • Determine whether this information is relevant for extrapolation to
humans under actual exposure conditions.

-------
6     CEELS FOR HAZARDOUS SUBSTANCES

  •  Determine the no-observed-adverse-effect level (NOAEL) and then
divide the NOAEL by appropriate uncertainty factors to arrive at an
acceptable limit for human exposure to the material.
                    •.!•:,      I       '      '          • ,  ' " •''' •
It is vital to select uncertainty factors that reflect the quality and rele-
vance of the data, differences between test species and humans,  and
variation within the human population. Typically, in the past the permis-
sible human exposure has been reduced by a factor of 10 for each ad-
ditional source of variation or uncertainty.
          :  •     i      i          i                        .     .    .

                  DEVELOPMEJ'ilT OF CEELS

   The CEEL team that reviews the available data will be required to
evaluate the quality, adequacy, and relevance of the studies from which
the data were derived, as well as the consistency and plausibility of the
entire data base. Human data should be emphasized, and the most useful
data are (1) the quantitative relationship between dose and response and
(2) the identification of sensitive members of the population. Studies that
show toxic effects of short-term exposure in humans or animals might be
more relevant than long-term studies, but both immediate and delayed
effects of exposure need to be considered.
   Animal studies should be evaluated on the basis of methodology, as
well as the appropriateness of the species for extrapolation to humans.
(Rats,  mice, hamsters, and rabbits have been identified  as species ap-
propriate for extrapolation.) In the absence of a most-relevant species,
data from the most sensitive species should be used.  The most useful
studies  (human or animal) employ a number of dose levels, determine a
level at which there is no effect, and characterize the toxic responses at
several higher levels.
   To expedite  assimilation by the LEPCs, data derived from literature
review  should be summarized hi a  consistent format. Documentation
should include  the major assumptions, uncertainties, and rationales for
handling of data.  The subcommittee has provided checklists  for the
presentation of data from human studies and animal experiments, as well
as guidelines for determining uncertainty factors and procedures for
calculating the CEELs themselves (see Figure 1).  Proposed CEELs will
be reviewed by a committee of experts, and where there  are significant

-------
                                   EXECUTIVE SUMMARY     7

data gaps the proposed CEELs should be considered preliminary.  All
CEELs should be reviewed and updated every 5 to 10 years, or sooner
if additional data become available.

-------

-------
 J.     Introduction
The possibility of sudden contamination of air at sites where hazardous
substances are produced, stored, or used has created the need for guid-
ance regarding potential exposure of people to hazardous chemicals under
emergency conditions.  The possibility of acute exposure is considered
when granting an operating permit for a new industrial facility or when
evaluating an existing facility's potential for a catastrophic release.  One
of the factors involved in emergency planning for averting or managing
toxic gas releases is consideration of "acceptable" exposure levels for a
"once-in-a-lifetime exposure."  As  described  below,  several sources
provide acceptable exposure levels, none of which is ideal.
   In 1984, after the Bhopal, India, incident in which methyl isocyanate
was released, killing 2,000 people and causing irreversible health effects
(damage to eyes and respiratory system) in up to 20,000 people, the U.S.
Environmental Protection Agency (EPA) identified 366 extremely hazar-
dous substances  (EHSs) on the basis  of acute lethality data  from the
Registry of Toxic Effects of Chemical Substances (RTECSs).  Title III
of the Superfund Amendments and Reauthorization Act of 1986 (SARA)
(also known as the Emergency Planning and Community-Right-to-Know
Act)  requires that EPA  in  collaboration with the Federal Emergency
Management Agency (FEMA)  provide information to local emergency
planning committees to help them develop and implement comprehensive
emergency-response plans for sites where EHSs are produced, stored, or
used.
  In response to SARA, EPA decided that it needed to develop emergen-
cy exposure levels for all 366 EHSs for the use of local planning com-
mittees.   The Agency  for Toxic  Substances  and  Disease  Registry
(ATSDR), which is responsible for conducting health-hazard evaluations
at hazardous waste sites,  also is interested in emergency exposure levels
for EHSs  for hazardous-site personnel and the general public living
around those sites. Because the National Research Council's Committee

-------
10     CEELS FOR HAZARDOUS SUBSTANCES
on Toxicology (COT) has recommended emergency exposure levels for
40 years to the Department of Defense (DOD) and the National Aero-
nautics and Space Administration (NASA), EPA and ATSDR requested
that COT review various guidelines used to define exposure limits, iden-
tify their limitations for use in community exposures to EHSs, and re-
commend its own guidelines for developing emergency exposure levels
for EHSs.  In response, COT established the Subcommittee on Guide-
lines for Developing Community Emergency Exposure Levels (CEELs)
for Hazardous Substances.
   CEELs are ceiling exposure values for the public and are applicable
to emergency exposures of foreseeable magnitude and duration (usually
1-8 hr)  However, few exposures; will exceed 1 hr in duration.  The
subcommittee agreed to set three CEEL values for each EHS. The series
Of three exposure values-CEEL-1, CEEL-2, and  CEEL-3-provide
much more information than a single value because the series indicates
the slope of the dose-response curve. As explained below, CEEL-2 and
CEEL-3 will not protect against health-related effects for all exposed
persons.                                                     ,
   CEEL-1 refers to the concentration of an airborne substance (sucn as
a gas  vapor, or aerosol)  for an exposure lasting 1-8 hr below which
direct toxic effects are unlikely to  lead to  discomfort in the exposed
population  (including  susceptible but excluding hypersusceptible in-
dividuals) and above which discomfort becomes increasingly common.
The symptoms for each category will depend on the chemical. For ex-
ample, discomfort due to  an  irritant gas  might cause eye and nose  ir-
ritation and an organic vapor might cause headaches.  The description of
CEEL values for each chemical must specify the symptoms expected for
each CEEL category.
   CEEL-2 refers to the concentration of an airborne substance (such as
 a gas vapor, or aerosol) for an exposure lasting 1-8 hr below which
 escape is not impaired and direct toxic effects are unlikely to lead to
 disability in the exposed population (including susceptible but excluding
hypersusceptible individuals) and above which disability becomes increas-
 ingly common.
   CEEL-3 refers to the concentration of an airborne substance (such as
 a gas, vapor, or aerosbl) for an exposure lasting 1-8 hr below which
 death or life-threatening effects  are unlikely in the exposed  population
 (including susceptible but excluding hypersusceptible individuals) and

-------
                                          INTRODUCTION     11

 above which death or life-threatening effects become increasingly com-
 mon.
   In an emergency situation, personnel responsible for protecting the
 public health need to know the dose-response characteristics of the hazar-
 dous substances that have been released. For example, emergency per-
 sonnel need to know if the concentration of a compound that causes eye
 irritation is  close to the concentration that causes death or if the lethal
 concentration is several thousandfold higher than the irritant level.
   The characteristics of the three CEEL categories are shown in Table
 1.
                 ACCEPTABLE INHALATION
           EXPOSURE LIMITS RECOMMENDED BY
              AGENCIES AND ORGANIZATIONS

   Many organizations worldwide are engaged in developing and issuing
inhalation exposure guidelines or promulgating inhalation health stan-
dards.   Some are private organizations; others are governmental agen-
cies. Although the approaches for determining inhalation exposure limits
have generic similarities, the philosophies, organizational structures, and
chartered mandates of each group active in such endeavors have finite
differences.   However,  all the inhalation  exposure  limits have been
deemed adequate within the framework of the various organizations to
protect the health of an exposed individual throughout his or her lifetime
under defined exposure conditions.
  The  greatest number of inhalation exposure limits and their issuing
organizations relate to occupational exposures—i.e., exposure for 40 hr
per week over a working lifetime. Among the private organizations in
the United States active in  recommending permissible occupational ex-
posure  guidelines are the Threshold Limit Values (TLV) Committee of
the American Conference  of Governmental  Industrial Hygienists
(ACGIH), the American Industrial Hygiene Association  (AIHA), the
American Society for Testing Materials (ASTM), and the American
National Standards Institute (ANSI).  Of these, the TLV Committee of
ACGIH has had the greatest national and international impact.  Recom-
mendations by these private organizations are not legally binding but are
jften voluntarily utilized by industry. Industrial corporations and trade

-------
12      CEELS FOR HAZARDOUS SUBSTANCES
                         !           •


TABLE 1  Characteristics of CEELs 1-3	
Category
                          Characteristics
Death or life-threatening   Death or life-threatening effects immediately or
effects                    soon after exposure
                   CEEL-3

 Disability                 External assistance is needed because
                          •  persons; are disabled by exposure and cannot
                             take actions necessary to protect themselves or
                             escape;
                         j •  exposed persons acquire an illness or condition
                             with permanent or long-lasting residual ef-
                             fects, including effects on the outcome of an
                             existing or subsequent pregnancy.

                   CEEL-2

 Discomfort               Exposed persons might request assistance, but
                          their condition,  though unpleasant and possibly
                          amenable to symptomatic relief, does not
                           • impair escape;
                        i   • produce disablement;
                           • result fin permanent or long-lasting effects.

                    CEEL-1
                   •'."'!    '       !            •             '  i  "'
 Detectability              Exposed persons might complain, inquire, or
                           express  aoxiety, but exposure, if perceived at all,
                           will be perceived only by smell, taste,  or sight or
                           by sensations (mild sensory irritation) that do not
                           persist after exposure ceases.  There are no direct
                           effects of exposure on health.
                        .:'.':      I            i               i  .
 Source:  Adapted from ECETOC,  1991.

-------
                                          INTRODUCTION    13

 associations also develop occupational exposure limits to apply to their
 particular facilities and customers.  These limits are particularly impor-
 tant when an industry is the sole manufacturer or user of the substance,
 the  number of potentially  exposed workers is small and limited to a
 single company, and other organizations are committed to developing
 limits for the vast number of other chemicals used in commerce.
   In the United States, the authority to promulgate federal health stan-
 dards is delegated by Congress to the executive branch of the  govern-
 ment (Administrative Conference of the United States, 1983).  The pro-
 cedure of administrative rule-making encompasses the processes of notice
 and  comment.  The public is given notice of an agency's proposed rule
 that  would affect the conduct of activities in the workplace or have an
 impact on the ambient environment.   The public, which includes indus-
 try,  labor,  academia, and government, participates in the rule-making
 process by presenting written or oral comments and scientific data on the
 various issues addressed in the proposed rule. Subsequently, the agency
 publishes a final rule that is the product of the original proposed rule and
 the comments and data received from the public. The final rule has legal
 status and is binding on the  regulated  industry or party. Violations may
 be punishable by fines or imprisonment or both.
  Legal health standards are the responsibility of a number of federal
 agencies. The Occupational Safety and Health Administration (OSHA),
 of the U.S. Department of Labor (DOL) is responsible for promulgating
 and enforcing health standards in the majority of work environments (the
 Occupational Safety and Health Act of 1970, U.S. Public Law 91-596).
The Mine Safety and Health Administration (MSHA) of DOL is respon-
sible  for promulgating and  enforcing metal and nonmetal mining air-
quality standards (the Federal  Mine Safety and Health Amendment Act
of 1977, U.S. Public Law 95-164).  The National Institute for Occupa-
tional Safety and Health (NIOSH) of the U.S. Department of Health and
Human  Services (DHHS) is responsible for developing health-related
criteria and making recommendations to OSHA and MSHA for the prom-
ulgation of legal standards. NIOSH also provides written comments and
oral testimony during the OSHA or MSHA process.   A fourth  federal
agency,  EPA, is charged with the establishment of community exposure
limits for air and water pollutants (the Air Quality Act of 1967 as amend-
ed by the Clean Air Act Amendments of 1970, U.S. Public Law 91-604,
and the Safe Drinking Water Act of 1974, U.S. Public Law 93-523). In
sases  in which the  substance to be regulated is a known or suspected

-------
14     CEELS FOR HAZARDOUS SUBSTANCES

human carcinogen (animal carcinogen), other federal agencies, such as
the National Cancer Institute and the National Toxicology Program of
DHHS, may be active in the federal rule-making process.
   On the international level, exposure limits have been established by a
number of countries or multinational groups.  Occupational exposure
limits have been developed by Germany, Russia, The Netherlands the
Nordic countries, the International Labor Organization  and the World
Health Organization. Other countries, such as the United Kingdom and
Australia, which previously used ACGIH's TLV  list as a  mandatory
standard, have recently introduced guidelines of their own for establish-
ing occupational exposure standards.  The newly founded  European
Economic Community is also active in establishing initiatives regarding
toxic chemicals and protecting potentially exposed individuals   The
International Agency for  Research  on Cancer  (IARQ  of  the World
Health Organization through its monographs on carcinogenic risks to nu-
mans provides valuable assistance to nations and international groups
active in establishing permissible exposure guidelines or standards.


     LIMITATIONS OF THE RECOMMENDED EXPOSURE
                  LIMITS FOR USE AS CEELS

   There are several key issues that influence how previously established
 exposure limits are used for the development of CEELs.  The issues are
 (1) the duration of exposure and the concentration of the toxic agents,
 determinants of the total dose, or the threshold of response in the ex-
 posed individual; (2) the margin of safety or uncertainty applied to the
 levels that induce the principal health risks; and (3) whether the limit is
 intended to protect  all exposed persons,  including the young,  aged,
 chronically ill, and those with genetic or acquired hypersusceptibihties
 to chemical insult.
   With the exception of  carcinogens, inhalation  exposure limits are
 based on the premise that the  amount of damage to a person's health is
 dependent on the total dose of the toxic agent and that, with a continuous
 decrease of dose, a quantity will be reached at which damage, or devia-
 tion from the norm, will not occur or will escape detection by the most
 sophisticated procedures.   It is imperative to recognize that this no-ob-
 served-adverse-effect level (NOAEL) will vary among individuals be-
 cause of the variability in health status and genetic composition of hu-

-------
                                           INTRODUCTION     15

 mans.  The variation in human susceptibility to a toxic substance, cou-
 pled with the imprecision of laboratory,  clinical, and epidemiological
 studies and the difficulties of extrapolating animal data to humans, neces-
 sitate the application of "safety" or "uncertainty" factors to the NOAEL
 hi establishing all exposure guidelines and standards.
   Occupational and community air exposure limits differ regarding the
 fundamental bases inherent to their establishment; likewise CEELs will
 differ from other exposure limits for the same reasons.  Occupational
 exposure limits  are established to protect against adverse health effects
 in nearly all workers who are repeatedly exposed over a 40-hr workweek
 for a working lifetime.  The population at risk in occupational settings
 is composed of adults whose health is usually better than that of the
 general population.
   Evidence for the healthier state of worker populations has been found
 repeatedly in epidemiological studies in which morbidity  and mortality
 rates are lower  in working populations than in the general population
 except when diseases or deaths are caused by specific toxic agents pres-
 ent at the worksite. This phenomenon  of apparent selection of persons
 at lower risk at the time of employment for  many causes  of disease or
death has been termed "the healthy worker effect" by epidemiologists and
 is a form of selection bias that inadvertently  occurs when the mortality
or morbidity experience of workers is compared with that of the general
population.  Conversely, community air exposure limits must  be estab-
lished for all elements of the general population over a chronological
lifetime and must reflect exposures that can be tolerated by the general
populace 24 hr per day for a lifetime. In contrast, CEELs are designed
to protect community populations, including potentially sensitive popula-
tions, that might be exposed to a  chemical  on a rare occasion, most
likely not more than once in a lifetime.  Thus, even though  differing
safety or uncertainty factors must be used to  derive CEELs that protect
the young, the old, the ill, and the idiosyncratic, they might not protect
all hypersensitive persons all the time.
  Because occupational  exposure limits and  community  air exposure
limits have differing exposure durations and populations at risk for ex-
posure to hazardous substances, CEELs cannot be derived  by simple
jxtrapolation from other  existing exposure limits. Discussion of CEELs
rbr each chemical substance should  represent committee judgments,
Allowing evaluation of known mechanisms of injury, experimental ani-
nal and  human  clinical and  epidemiological data, and conditions in

-------
16     CEELS FOR HAZARDOUS SUBSTANCES

which emergency exposures might occur.  Effects must be reversible and
not incapacitate an exposed individual either mentally or physically; he
or she must be able to recover or return to normal without significant
medical treatment.
   Several types of exposure guidelines have been developed for use in
emergency situations involving a single high-level exposure to substances
that might produce adverse health effects.
   COT has recommended emergency exposure guidance levels (EEGLs)
for exposures of 1-24 hr for about 70 substances. An EEGL is defined
as a concentration of an airborne substance (such as a gas  vapor  or
aerosol) that may be judged by DOD to be acceptable for the perfor-
mance of specific tasks during rare emergency conditions lasting 1-24 hr.
Such an exposure could result from a fire, a spill, a line break, or any
event mat is unanticipated  but nevertheless has a rate of occurrence that
is predictable.  Exposure at an EEGL might produce reversible effects
that do not impair judgment and  do not interfere with proper responses
to the emergency. It must be noted, however, that EEGLs are not hy-
gienic or safe.  The EEGL is a suggested peak level of exposure and is
not to be regarded as a standard in any form or use.  EEGLs were devel-
oped for military use and are intended for healthy young military person-
nel  Therefore, EEGLs are not applicable to general populations consis-
ting of the elderly, the very young, pregnant women, and ill persons
   COT also recommended short-term public emergency guidance levels
 (SPEGLs) for  civilians (comprising healthy  adults, infants, children,
 elderly, and ill persons) residing in and near  military installations  A
 SPEGL is defined as a concentration of an airborne substance (such as
 a gas, vapor, or aerosol)  for unpredicted single short-term emergency
 exposure of the general public.  In contrast to  the EEGL, the SPEGL
 takes into account the wide range of susceptibility of the general public
 SPEGLs include sensitive subgroups, such as  children, the elderly, and
 Ul persons.  Effects of exposure on the fetus and on the reproductive
 capacity of both men and  women also would be considered. However,
 SPEGLs were recommended for only four substances by NRC.  Further-
 more  EEGLs  and SPEGLs do not exist for many commercially  impor-
 tant chemicals.  Thus, most of the NRC values are not directly applicable
 to the general public.                                      .
    In the  1970s, NIOSH developed a series of exposure values-immedi-
 ately  dangerous to life and health (IDLHs).   The IDLH values were
 developed for  respirator-selection purposes, not for setting permissible

-------
                                           INTRODUCTION     17

 exposure limits.  The EDLH value represents a maximum concentration
 from which, in the event of respirator failure, a person could escape
 within  30 min without  experiencing any escape-impairing effects  or
 irreversible health effects.
   EDLH values are based primarily on secondary-literature review. The
 basis for each EDLH value has not been published formally, nor were the
 values subject to peer review. No formal mechanism or procedure exists
 to update the values.
   Other potentially useful  inhalation  exposure limits are short-term
 exposure limits (TLV-STEL) and ceiling limits recommended by ACGIH
 for workplace  exposures.  Although developed for daily occupational
 exposures, the values have been designed for healthy workers and are not
 relevant for the general public.
   Other organizations, such as the World Bank and the European Eco-
 nomic Community, provide guidance in terms of acutely hazardous quan-
 tities rather than concentrations.   Their recommendations are based on
 LC50 values. LC50 values are less appropriate for emergency planning
 since lethality is the only health effect they consider.  Human-experience
 data, escape-impairing effects, and serious but nonlethal effects, such  as
 lung or liver injury,  are not considered.
   Emergency response planning guidelines (ERPGs) have been devel-
 oped under the  auspices of AIHA.  The AIHA committee decided to set
 three planning guidelines: ERPG-3, ERPG-2, and ERPG-1.

   •  ERPG-3 is the maximum airborne concentration below  which it is
 believed that nearly all individuals could be exposed for up to 1 hr with-
 out experiencing or developing life-threatening health effects.

  The ERPG-3 level is a worst-case planning level above which there  is
 the possibility that some members of the community will develop life-
 threatening health effects.

  •  ERPG-2 is the maximum airborne concentration below which it  is
believed that nearly all individuals could be exposed for up to 1 hr with-
out experiencing or developing irreversible or other serious health effects
or symptoms that would impair an individual's ability to take protective
action.

  Above ERPG-2, for some members of the community, there might be

-------
18     CEELS FOR HAZARDOUS SUBSTANCES

significant adverse health effects or symptoms that could impair an in-
dividual's ability to take protective action.  Those symptoms  include
severe eye or respiratory irritation oir central nervous system effects.

  • ERPG-1 is the maximum airborne concentration below which it is
believed that nearly all individuals could be exposed for up to 1 hr with-
out experiencing any symptoms  other than mild transient adverse health
effects or without perceiving a clearly defined objectionable odor.

  The ERPG-1 identifies a level thatt does  not pose a health risk to the
community but might be noticeable because  of slight odor or mild ir-
ritation.  In the event of a small nonthreatening release, the community
may be notified that odor or slight irritation might be noticeable but that
concentrations are below those that could cause adverse health effects.
  Although ERPGs have been widely used by states, local communities,
departments of transportation and energy, and industry, there are ERPGs
for only 43 substances.  Information from these ERPG reports might be
useful for developing CEELs on those 43 substances, but CEELs are
needed for about 350 chemicals.
  This report describes the criteria and methods for developing CEELs
for hazardous substances, which will  be useful to EPA, ATSDR, in-
dustry, and state and local governments in responding to chemical emer-
gencies.

-------
 2    Criteria for Establishing CEELs
 Any numerical index of toxic effect or any standard that is set to protect
 against toxic effects must reflect the relationships among the following
 factors:

   1.  The population potentially exposed.
   2.  The degree and pattern of exposure.
   3.  The nature, severity, and reversibility of the anticipated toxic
 effects.
   4.  The proportion of those exposed who are subject to toxic effects.

   Thus,  for example, LC50 values refer to a specific animal species (1),
 an exposure concentration for a specified time (2), a specific toxic effect
 (death) (3), and a specific proportion of animals  affected (50%) (4).
 TLV-STEL values refer to workers (1), represent the exposure  for 15
 min by inhalation (2), refer to a level not expected to have significant
 adverse health effects (3), and refer to nearly all workers (4).
   Those four factors are necessary to establish a definition for CEELs.
                 POPULATIONS OF CONCERN

   The population concerned  in  any  industrial accident  involving the
release of hazardous substances is the work force and others within the
premises and the general public who live or work in the  vicinity. The
general public might include people who are more susceptible to chemi-
cal exposure than the average person (e.g., the elderly, the young, preg-
nant women, and those with acute or chronic  illnesses). CEELs should
take into account such susceptible individuals, but not necessarily more
seriously debilitated persons or those with enhanced susceptibilities (e.g.,
those with pneumonia or myocardial infarction) current or historical.

                               19

-------
20     CEELS FOR HAZARDOUS SUBSTANCES

            EXPOSURE PERIODS AND PATTERNS
                                                          I Js
   After an accidental release of a quantity of a chemical, the concentra-
tion at any point where the chemical disperses will be influenced by
meteorological conditions and topography. As the gas cloud disperses,
the concentration at a particular point generally will rise to a maximum,
which will be sustained for  a period and  men fall toward zero.  It was
considered that the most practical descriptors of exposure would be the
exposure duration and the concentration that was not exceeded during the
exposure.  CEELs would be most useful if the exposure period or peri-
ods to which they refer were of a duration pertinent to realistic scenarios
for accidents involving the release of hazardous substances. The descrip-
tion of the scenarios should incorporate the evolution of release over time
as determined by product inventory, isolation, and  other mitigating de-
vices and dispersion factors as determined by meteorological and topo-
graphical conditions. In some cases, the exposure period will be fairly
clear and similar for a wide range of concentrations—for example, when
release  is uniform and  ceases abruptly and  wind velocity  is high.   In
other cases, exposure duration will vary considerably according  to the
concentration  of interest; in those cases, an iterative process must be
used to  establish the appropriate duration for each CEEL beginning with
an arbitrary period.
   CEELs should be established for each toxic effect for a number ot
exposure periods, up to 1-8 hr, when the data allow it, so that values
most appropriate for predicted exposure periods  can be chosen.  How-
ever, few exposures will exceed 1 hr in duration.
        NATURE AND SEVERITY OF TOXIC EFFECTS

   Toxic effects are of many types!  and each can occur with varying
degrees of severity. The important effects following accidental industrial
releases of toxic materials are those that occur in humans following an
acute exposure to an airborne subsitance.   Systemic effects  will result
chiefly from absorption of materials through the lungs; local effects will
involve the  eyes, respiratory tract, and skin.
   The possibilities are limitless for describing the grades of severity of
the immediate toxic effects of acute exposures.  It is essential in setting

-------
                      CRITERIA FOR ESTABLISHING CEELS     21

 CEELs that each indicates a type of toxic effect that clearly requires a
 specific emergency response.  The subcommittee recognized the need for
 three indexes (i.e., CEELs 1-3) to represent the transitions between four
 graded groupings (see Table 1).
   Death or life-threatening effects are the most severe effects for which
 an exposure index can be provided; they are easily defined and are used.
 by society to judge the severity of accidents.  Two other grades of ef-
 fect—disability and discomfort—though less well defined, place distinct
 demands on emergency and health-care services.  The term disability is
 used here to indicate the situation where persons will require assistance
 or where the effects of exposure will be more severe or prolonged with-
 out assistance.   Persons suffering discomfort, though distressed and
 possibly requesting assistance, will not be dependent on assistance for
 minimizing the severity or duration of the effect of exposure. Exposure
 insufficient to cause discomfort or adverse health effects might be per-
 ceived nevertheless by means of smell, taste, or sensations (mild sensory
 irritation) that are not uncomfortable.  The awareness of exposure might
 lead to anxiety and complaints and constitutes what is termed here detec-
 tability.   Except for death, these graded categories of effect are not
 sharply demarcated, but each merges into adjacent categories.
       PROPORTION OF THE EXPOSED POPULATION
       SHOWING EACH CATEGORY OF TOXIC EFFECT

   CEELs should  indicate  exposures that would be thresholds  for the
occurrence of (1) death or  life-threatening effects, (2) disability, or (3)
discomfort in the population.  At such a threshold concentration, a small
proportion of the population might exhibit effects.
   Precision in defining "a small proportion" is impossible and unneces-
sary for the following reasons:

   • The categories of effect are not precisely demarcated.
   • The data available for setting  CEELs  are imprecise and derived
from various sources,  including experiments and clinical observations
usually not designed for setting CEELs.
   • The incidence of effects at CEELs will depend on the proportion of
susceptible people  in the population, which is variable and uncertain.

-------
22     CEELS FOR HAZARDOUS SUBSTANCES

  •  The response will depend on the actual pattern of exposure rather
than the predicted exposure category!                 '..'''.
  •  Models used to predict the pattern of exposure provide information
that is inherently imprecise.

-------
        Sources  of Data for
        Developing CEELs
 Ideally, the data from which CEELs are developed would be well-docu-
 mented descriptions of the clinical effects seen in a representative sample
 of the general population exposed to accurately measured concentrations
 of the substance under consideration for relevant exposure periods.  In
 practice, such data never exist.  Reliance has to be placed on toxicologi-
 cal data from animal studies  and human data, usually from workplace
 accidents.  Thus, it is essential that a thorough literature search be con-
 ducted for data relating to accidental and experimental human exposures
 and to lexicological studies in animals.  Although the most immediately
 or directly applicable data would come  from reports of acute inhalation
 studies  in animals and humans, the search should not be confined to
 those studies but should be widened to include all routes of exposure and
 all exposure periods. When the reports  of inhalation studies are scant or
 inadequate, the  subsidiary information might be helpful  and, in all cases,
 would provide confirmatory information on the toxicity  of the substance.
 In some circumstances, for example,  when the  data  available on  the
 chemical itself are inadequate, data relating to a close chemical analogue
 might contribute significantly  to the information to be considered.
   All data sources should be consulted, including the published scientific
 literature and any unpublished information  from industry and  public
bodies.  Every attempt should be made to obtain the full text of original
articles  to extract the details of the  experimental conditions and results.
Secondary sources should be used only when original articles cannot be
obtained.  Unpublished reports and individual expertise may be used to
supplement published reports  provided they are scientifically credible.
   Various types of evidence should be assessed in establishing CEELs.
Information should  be collected from (1)  studies and case reports  of
human exposures, (2) animal toxicity studies, (3) in vitro toxicity studies,
and (4) chemical and physical  characterizations of the potential toxicant.
                               23

-------
24     CEELS FOR HAZARDOUS SUBSTANCES

                         HUMAN DATA

  Human data relating to the acute effects of hazardous chemicals have
been collected in relation to unintentional and intentional exposures to
toxic agents.
  Unintentional man-made disasters that are life-threatening often are
sudden and devastating (e.g., Bhopal and Chernobyl) but also might be
chronic and insidious (e.g., Minamata). In both circumstances, however,
the health consequences are most often a combination of acute and long-
term effects.
  Emergency exposures to chemicals usually arise from various sources:

  • Accidental  releases at industrial sites.
  • Spills during transportation.
  • Explosions, fires, or other  mishaps.
  • Escape from toxic waste sites.

  The timing and location of toxic releases are difficult to predict and
have often been unanticipated.  Thus, the at-risk populations and the
possible health effects among them have seldom been determined before
a toxic release.  As  a result, health studies among populations exposed
to toxic releases almost always have been retrospective in nature.
  Two main types of health data have been collected following acute
exposures to toxic agents:  (1) case studies of individuals affected to
various degrees  by acute exposures; and (2) epidemiological studies of
acutely exposed  populations.
  Epidemiological and  clinical  data on  toxic  health effects also come
from workers exposed to toxic agents during the course of their work.
Those data include some of the most detailed studies of the relationship
of human exposure and response to toxic agents. Health  data on com-
mon industrial chemicals, such  as sulfur dioxide, ammonia, chlorine,
hydrogen sulfide, and phosgene, agents that have killed  workers, are
readily available. Exposures to workers, particularly following industrial
spills or explosions, generally are higher than  those encountered by the
public during acute emergency releases.  Workers usually  are at greater
risk of life-threatening exposures because they are closer  to the source
of toxic release.  When large quantities of highly toxic agents are stored

-------
             SOURCES OF DATA FOR DEVELOPING CEELS     25

 close to populations (as occurred in Bhopal), however, there is the poten-
 tial for severe, life-threatening toxic exposures to large numbers of the
 general public.   Under  such circumstances, an environmental release
 leads to health  effects among the population similar to those seen in
 industrial workers.
   Another important source of human data comes from volunteers who
 •are exposed intentionally to  toxic agents under controlled clinical  con-
 ditions, again at levels less than those usually found during acute emer-
 gencies.
                           Case Studies

   Important clinical information has been obtained from survivors of
 acute massive exposures to toxic agents.  By their nature, case studies
 describe individuals who develop clinical problems  caused by such ex-
 posures.  When affected subjects are followed over time, information is
 obtained about the natural history of health effects caused by particular
 agents, as well as the  clinical  patterns  of recovery or progression to
 chronic disease.
   The  initial description  of  reactive airways dysfunction syndrome
 (RADS) exemplifies how a series of individual case histories can demon-
 strate an important effect following a single acute toxic exposure (Brooks
 et al., 1985).  Some individuals, even though they have neither asthma
 nor atopy beforehand, might inhale a high concentration of a respiratory
 irritant and quickly develop highly reactive airways  with a clinical pic-
 ture indistinguishable from asthma (Brooks et al., 1985; Boulet, 1988).
 Asthmatic symptoms typically appear within 24 hr of the acute exposure
 and persist for many months.  Further case reports have confirmed that
 inhalation of high concentrations of many  different irritant gases, fumes,
 vapors, and aerosols might lead to RADS.
   Recent  findings among 56 workers exposed to an acute  release of
 acetic acid showed that the prevalence of RADS was 21.4% among those
most highly exposed and that the risk of RADS in the highest exposure
group was 10  times greater than in less exposed  individuals (Kern,
 1991). Increasing recognition of RADS and the extent to which it might
occur following a single acute exposure to a respiratory irritant make it

-------
26     CEELS FOR HAZARDOUS SUBSTANCES

an important health effect to be considered by the risk assessor.  Criteria
for the diagnosis of RADS have been discussed (Brooks et al.,  1985;
Kern, 1991).

                ''                I                      .' '  '
                      Epidemiological Studies

   Virtually all attempts to study the effects of acute releases of toxic
agents on exposed populations have been retrospective.  Longitudinal
data on exposed groups (in contrast to case studies of affected individuals
over time) are relatively uncommon and often subject to methodological
problems.                                             ,  .       ,
   A major problem is defining the exposed or at-nsk populations and a
similar unexposed population that can serve as a valid comparison group.
In the immediate  response to  an acute emergency,  records are seldom
obtained of everyone who might have been exposed. To reconstruct that
information after the fact is difficult, and incomplete definition of the
exposed group is a source of potential bias.
   Reliable exposure measurements are seldom obtained at the time ot a
toxic release, so most exposures a*e estimated retrospectively. For ex-
ample, when a toxic agent is released into the atmosphere from a source,
exposure can be estimated by mathematical modeling  based on  the
amount and rate of the release, the prevailing atmospheric conditions
and knowledge of where the subject was located relative to the point of
release.   That approach contains  several  assumptions (uniformity ot
release-exposure and exposure-dose relationships based solely on large-
scale physical factors) that cannot account for differences in the immedi-
ate environment of individuals, differences that might affect  exposures
greatly or might affect the substantial  individual variation in exposure-
 dose relationships.                                  .
    Misclassification of an individual's exposure is  inevitable whenever
 exposure is estimated.  Such misclassification might occur  randomly or
 nonrahdomly and can influence the data adversely.  Random misclas-
 sification of exposure causes bias toward the null hypothesis:  noisy data
 lead to a decrease in statistical power for any exposure-related effect and
 therefore  the possible dismissal of an  exposure effect because of a lack
 of statistical significance, when a real effect actually might be present.
 If exposure misclassification is nonrandom, it becomes a source of bias

-------
             SOURCES OF DATA FOR DEVELOPING CEELS     27

 that can cause a spurious association or lack of association between the
 exposure and health effects.
   A different approach to exposure assessment has been to measure the
 concentration of toxic agents and their metabolites in body fluids and
 tissues.  That method of determining exposure is best suited for agents
 that are absorbed systemically and are excreted slowly. The association
 between such exposure measurements and related health effects will be
 closest and most meaningful when the fluid or  tissue sample is taken
 from the target organ or a related tissue. The most frequently sampled
 fluids and tissue are blood, urine, and fat cells.  Exhaled  air also can
 provide useful information.   Blood and urine are used widely for such
 biological  monitoring,  and the relevance  of exposure measurements
 derived from each fluid depends greatly on the pharmacokinetics of the
 agent and the timing of the sample in relation to exposure.  Samples of
 fat tissue have been used particularly to determine exposure to organo-
 chlorine compounds.  Many organochlorines accumulate and persist for
 years in fatty tissue. The persistence of those compounds has been use-
 ful, for example, in determining past exposures to phenoxyacetic acid
 derivatives and polychlorinated biphenyls. Recent exposures to volatile
 organic compounds can be detected by the presence of the compound in
 exhaled breath.   Workers are exposed frequently to toxic agents in the
 workplace. Epidemiological studies of working populations have provid-
 ed evidence of the nature of adverse health effects caused by many toxic
 agents and their exposure-response relationships.  It is important to note,
 however, that workers usually are exposed to low to moderate levels for
 8 hr per day  and that standards for workplace exposures are often aver-
 aged over the work period.
  Extrapolation of exposure limits developed for workers might not be
 relevant to the large fraction of nonworking people in the general popula-
 tion.   Currently, workers usually are healthier than  nonworkers.  By
 their nature,  working groups exclude the very young and the elderly,
 who are regarded as more susceptible to many of the adverse effects of
toxic hazards.  Workers are thus a biased sample of the general popula-
tion in terms of their health  status and their  susceptibility to hazardous
agents.
  In addition, workers who experience adverse health effects from toxic
agents in the workplace often migrate to jobs where they are not exposed
to such hazardous agents; that self-selection process is a  form of the

-------
28     CEELS FOR HAZARDOUS SUBSTANCES

healthy-worker effect and has been noted in many working populations.
The healthy-worker effect causes bias in epidemiological studies and is
seen most commonly in studies of cross-sectional design; it is less likely
in longitudinal cohort studies in which self-selection effects can be moni-
tored by the investigator.


                     Controlled Exposures of
                         Human Subjects
                        ' I   	„ .'     !• ,   •       ,;              II , 	"

  Important and unique data have been obtained from studies of volun-
teers subjected to controlled exposures to toxic agents at low concentra-
tions, that approach complements case studies and epidemiological data
by  defining exposure-response relationships among healthy subjects and
susceptible subgroups of the general population (such as asthmatics).
Such data also are useful in validating animal studies, conducted at high-
er concentrations, and extrapolating to human effects.
   Low-level controlled exposures are designed to investigate single
agents or simple mixtures. They have been used  to define odor thresh-
olds  exposures at which respiratory, skin, and eye irritation occur, and
neurobehavioral effects. Although such studies are not usually designed
to  assess nonspecific systemic  effects, exposure of volunteers at low
levels also might help identify exposures that are associated with mild
subjective symptoms such as nausea, headache, dizziness, or fever.
   Controlled exposures to airborne agents have been either brief individ-
ual exposures via a nasal mask or mouthpiece or exposures to one or
more people in an environmental chamber.   The latter procedure has
greater flexibility in terms of the exposure route, the number of subjects
exposed, their activities, the duration of exposure, and  the ease of con-
comitant measurements of physiological function and other indexes of
health effects.
    Much of the data available from controlled clinical exposures to air-
borne hazards relate to the criteria for air pollutants defined  by EPA
 (such as ozone, nitrogen dioxide, sulfur dioxide, and  respirable parti-
 cles); less information is available on common industrial pollutants (such
 as  ammonia and chlorine).
    For many agents, the lack of human data on acute health effects at low
 exposure levels presents a serious problem for the risk assessor who is

-------
             SOURCES OF DATA FOR  DEVELOPING CEELS     29

 required to evaluate the human toxicity of a given chemical and to devel-
 op emergency exposure guidelines.  In  the absence of relevant human
 data, the risk assessor must resort to less direct information, including
 acute toxicity studies in animals.
   When the Bhopal disaster occurred in  1984, limited human data exist-
 ed on the health effects of methyl isocyanate, and very limited data exist-
 ed that applied directly to that chemical's exposure-response relationship
 in humans.  Thus, at the  time of the release there was little quantitative
 information on the toxicity of methyl isocyanate in people.
                  ANIMAL TOXICITY STUDIES

   The data necessary to evaluate the relationship between exposure to a
 pollutant and its effects on a population are frequently not available from
 human experience.  For many  air pollutants,  studies in animals have
 provided the only useful data. Ideally, the data should be derived from
 at least two species and by the inhalation route.
   Inhalation experiments with animals provide a basis  for estimating
 possible effects in humans and the concentrations at which the effects
 occur.  They are useful in the identification of adaptations that might
 occur following repeated exposure. They permit the testing of hypothe-
 ses about the mechanisms of the toxic actions of pollutants.  They offer
 a good opportunity to explore interactions between pollutants and other
 factors that might affect toxicity.
   Data from skin absorption,  ingestion, and parenteral studies also are
 potentially useful. The usefulness of animal data depends in part on the
 species used.  Relevance to humans might be limited in the absence of
 information on target organs and pharmacokinetics  in animals and hu-
 mans.
   Well-conducted animal studies will report the following:

   •  The most sensitive target organs or body  systems affected  by ex-
posure to the contaminant in question.
   •  The nature of the effect on the target organs.
   •  Data to establish dose-response relationships for the target organs—
from no observed effect  to severe effects.   (The distinction between
exposure and dose needs to be made.)

-------
30     CEELS FOR HAZARDOUS SUBSTANCES

  •  The rate of recovery from reversible effects, if any.
  •  The nature and severity of injury for effects that are not reversible.
  •  Cumulative effects, if any (e.g., neurotbxicity and cancer)^
  •  Pharmacokinetic data for comparison with data obtained from hu-
U13HS.                                                  • i_   u    •
  •  The effects  of interaction, if my, of the toxicant with other air
pollutants and the minimum concentrations at which the interaction ap-
pears to  occur.
  •  Techniques used to assure quality and avoid bias.


                SHORT-TERM IN VITRO TESTS

  The major effort to date in the development of short-term tests has
been in genetic toxicity testing for predicting carcinogenicity after obser-
vation of genotoxicity. More than 100 short-term in vitro tests have been
developed or proposed to detect rodent carcinogens and, by implication,
chemicals that might be carcinogenic in humans.  Test combinations (or
batteries) and sequential testing schemes have been proposed to improve
the predictability.  In  general, the batteries and combinations have not
proved to be better predictors than the best of the  single tests.  Too
often  batteries of tests have increased sensitivity (finding more positives)
at the cost of decreased specificity (labeling materials positive that were
really negative).                                       .      .
   Short-term tests  have been developed for other lexicological end
points  such as reproductive and developmental toxicity; however, the
level o'f investigation for the majority of those end points has not been as
 extensive  as that for  carcinogenicity.  Therefore, short-term  tests for
 carcinogenicity can be used as the paradigm for short-term tests in gener-
 al  and the mistakes made and lessons  learned in this field can be used
 to' guide development of comparable tests  for other pathological end

 P°Snort-tenn tests often are defined with respect to the long-term  effect
 that they are to identify or measure.  There are three reasons to perform
 short-term tests:

   1     For screening purposes  to  replace  costly  long-term testing or
 provide information that could be used to help design subsequent tests.

-------
             SOURCES OF DATA FOR DEVELOPING CEELS     31

    2.    As an adjunct to other tests to aid in making regulatory or in-
 dustrial decisions or to aid in elucidating the mechanisms of toxicity.
    3.    To replace existing tests with easier tests.

    Short-term tests need to be validated.  As in any predictive test, short-
 term tests will give false positives  and false negatives.  The tests are
 useful in elucidating the mechanism of action.
                     In Vitro Reproductive and
                       Developmental Effects

   A number of in vitro approaches have been used for evaluating repro-
 ductive and developmental toxic effects. Whole-embryo cultures, in vitro
 fertilization procedures, organ-tissue cultures,  and cell-culture systems
 have been  developed; some of those systems  have been tested  for
 replicability and reliability with both positive and negative test agents.
   In vitro test systems can provide significant information on target-site
 interaction and mechanisms of action, but none of the test systems alone
 is considered adequate for risk-assessment purposes or useful in setting
 CEELs, because most in vitro systems do not account adequately  for
 effects that might occur in the whole animal.  For example, the maternal
 influence on dose to the developing organism via piacental transfer dur-
 ing pregnancy or via milk transfer during the preweaning period cannot
 be modeled in in vitro systems.  In the case of in vitro fertilization  as-
 says, several steps in the interaction between the sperm and the female
 reproductive tract are bypassed.  Thus, those test systems may be used
 in conjunction with  whole-animal tests of developmental toxicity, fer-
 tility, or reproductive organ function, but they may not be used as the
 only indicator of an adverse effect.
                   In Vitro Genotoxicity Assays

   Many of the in vitro carcinogenicity tests are in fact tests for detecting
agents causing damage to DNA.  The existence of an extremely high
positive correlation for inducing the two end points (tumors and  muta-
tions) for a compound indicates a relationship that is  more than coin-

-------
32     CEELS FOR HAZARDOUS SUBSTANCES

cidental, and much of current cancer research is devoted to elucidating
the mechanisms related to the steps of tumor initiation. One feature of
chemicals that appears to be significant for inducing both mutations and
tumors is the ability to form reactive electrophilic molecular species that
will interact with nucleic acids. Most direct-acting carcinogens are high-
ly reactive electrophilic agents per se and most procarcinogens typically
form electrophilic metabolites.
   A  significant body of other indirect evidence also associates one or
more critical steps in cancer induction with change in cellular genetics.
For example, humans with hereditary defects in their ability to repair
DNA damage have a statistically higher susceptibility to cancer induc-
tion  The active forms of most initiating chemical carcinogens are also
the active mutagens. Most tumors appear to be of single-cell origin, and
that origin would be a requirement if mutation were an etiological com-
ponent in the process of initiation.
   In vitro mutagenicity tests include those using microorganisms, such
as the Ames Salmonella strains, as indicators of genetic activity and those
using mammalian tissue extracts as  sources of metabolic  activation to
identify enzyme-formed mutagens. Yeasts can detect the kinds of genetic
damage  that are identifiable only in  diploid eukaryotic cells. Other in
vitro test systems include those using mammalian cells (e.g., Chinese
hamster ovary  cells) for detecting  point  mutations  and chromosomal
aberrations and damage.
   Mutagenicity assays  are designed to identify compounds that are con-
verted to  reactive electrophilic compounds (or are converted to these
compounds) hypothesized to be involved in tumor initiation.  According
to this postulate, only carcinogens that have a reactive electrophilic state
are expected to be detected by mutagenicity assays.  Chemicals that
promote initiated  cells or induce tumors by indirect means, such as hor-
mones, physical carcinogens, heavy  metals, and certain other chemicals
are unlikely to induce genetic alterations.  One important function or
short-term in vitro genotoxicity tests is to distinguish between direct-
 acting carcinogens (chemicals that produce neoplastic transformation) and
 indirect-acting carcinogens (chemicals that promote tumor development).
 It might be possible to develop acceptable exposure levels for indirect
 carcinogens. The results of carcinogenicity tests appear to support that
 model because some apparent indirect carcinogens (e.g., saccharin, some
 chlorinated  hydrocarbons,  and nitrilotriacetic acid)  require high-dose
 levels to produce effects, whereas most direct carcinogens are active at

-------
             SOURCES OF DATA FOR DEVELOPING CEELS     33

 low doses. Thus, negligible risk can be expected from very low environ-
 mental levels  of many indirect carcinogens but not necessarily  from
 direct carcinogens.
                  Structure-Activity Relationships

   Structure-activity relationships (SARs) are commonly used in phar-
 macology and toxicology to predict chemical structures that have the
 ability to cause certain responses.  Because carcinogenesis is a multistage
 process, SARs tend to be more complicated than SARs associated with
 simpler processes—for example, the inhibition of specific enzymes, such
 as acetylcholinesterase, by certain pesticides.  Certain chemical  struc-
 tures, however, are often but not always associated with carcinogenic
 activity.  Those chemicals include such compound classes as polycyclic
 aromatic  hydrocarbons,  aromatic  amines,  Af-nitroso compounds, and
 biological alkylating and arylating agents.
   SARs are of great value in the pharmaceutical industry in deriving new
 products  effective against very specific targets.   A carcinogen might
 induce cancer by different mechanisms in different tissues  of the same
 species.  SARs should be used with care and are most useful if the can-
 didate chemical contains  one or more of the key groups associated with
 the induction of cancer.  In such cases, the known carcinogenicity or
 noncarcinogenicity of structurally related compounds might be sufficient
 to indicate whether the chemical is likely or unlikely to be a carcinogen.
 For further information on this topic, the reader is referred  to the work
 of Klopman (1985) and Klopman and Rosenkranz (1991).
   Besides predicting carcinogenicity, SAR is useful  in predicting other
 toxic  effects, such as reproductive and developmental toxicity.   When
 there  are no data for the compound under consideration, its  CEELs can
 be estimated from toxicity of surrogate chemicals that are closely related
 to it in its chemical class.
                 CHEMICAL AND PHYSICAL
             CHARACTERISTICS OF TOXICANTS

  The chemical and physical characteristics of a chemical provide valu-
able information on both the dosimetry of the compound within the body

-------
34     CEELS FOR HAZARDOUS SUBSTANCES
                         1   •       !                      '
and on the likely toxic effects.  A highly lipophilic substance can be
expected to accumulate in fatty depots in the body and might be retained
and available for metabolism to toxic metabolites for some time after an
exposure. Water-soluble substances can be expected to clear the body
more rapidly except for metal ions that bind to specific proteins.  Much
is known about the substrate specificity of xenobiotic metabolizing en-
zymes and metabolic pathways and potential metabolites for many chem-
icals can be predicted by the structure of the chemical.  For aerosols, the
physical size, shape, and density of the aerosols can be used to determine
the probable sites of deposition in the respiratory tract and the degree to
which the aerosols are respirable. The water solubility of inhaled gases
is known to strongly influence where the gases are deposited and retained
in the respiratory tract. Thus, the chemical and physical characteristics
of a substance can be used to predict the  potential toxicity of that sub-
stance and to set CEELs, especially for substances for which few, if any,
toxicity studies have been conducted.

-------
        Types of Data  Used in
 4     Developing CEELs
 The preceding chapter summarized the sources of data for use in devel-
 oping CEELs. In this section, the two types of data obtained from those
 sources and their importance in developing CEELs are described.  The
 types are (1) dosimetry and (2) toxicity end points.
          PHARMACOKINETICS AND METABOLISM

   The evaluation of the health effects of any chemical in a given envi-
ronment is greatly improved if there is an understanding of its physiolog-
ical disposition in the body (pharmacokinetics and metabolism). Metabo-
lism of a chemical might lead to detoxification or might result in meta-
bolic activation leading to toxic effects.  Because metabolic events usual-
ly are enzyme-mediated, they usually are driven by the concentration of
substrate available for the reaction. The concentration of substrate in the
target site is a result of the level of exposure and the pharmacokinetics
of the chemical.  Modern pharmacokinetic studies are aimed increasingly
at developing  dispositional models, which compare input of chemical
with outflow for the whole system as well as for individual organs.
   The toxic metabolites of each chemical and factors that control the rate
at which they form are important for developing CEELs. It is important,
however, to distinguish between the production of toxic metabolites in
organs where toxicity is observed and metabolism in the liver where the
bulk of most chemicals are metabolized. The relationship between me-
tabolism in liver and toxic effects  in either liver or  other organs is an
important factor because metabolism of chemicals  in a  liver already
damaged by other  agents or conditions (diet, microbial  disease, toxic
chemicals, etc.) is  likely  to differ  from metabolism  in a normal liver.
Furthermore, if toxic metabolites are exported from the liver to other
organs,  damaged livers may alter the production of toxic responses.

                               55

-------
36     CEELS FOR HAZARDOUS SUBSTANCES

  Toxic metabolites have been observed to be highly reactive chemically
and have been characterized as  "biologically reactive intermediates."
The reactive metabolites can covalently bind to nucleic acids, proteins,
or lipids, which  in turn can alter  DNA replication or the process of
transcription.  Attempts to repair damage to  DNA might involve
misrepair, leading to erroneous DNA replication or function, inhibition
of protein synthesis if RNA is the target, or inhibition of an enzyme or
other activity if proteins are the target. In addition to the formation of
reactive intermediates that are metabolites of the chemicals, metabolic
activity can give rise to species of active oxygen, which  can damage
nucleic acids or proteins or yield lipid peroxidation. The results might
range from target-organ toxicity to  carcinogenesis.
                        Biological Markers

   Recently, the utility of biological markers as indicators of exposure to
polluted air has been investigated (NRC, 1989).  Biological markers of
various types within an exposed individual can indicate the degree of
exposure to a pollutant, the initial structural, functional, or biochemical
changes induced  by exposure, and, eventually, the changes associated
with adverse health effects.  Thus,  biological markers are indicators of
change within an organism that link an exposure to polluted air to subse-
quent development of an adverse health effect. It is convenient to divide
biological markers into three classes: (1) biological markers of exposure,
(2) biological markers (early predictors) of effects of exposure, and (3)
biological markers of susceptibility to effects of exposure.
   Biological markers of exposure can be thought of as "footprints" that
the chemical leaves behind upon interacting with the body.  Such mark-
ers contain the chemical itself 01 a metabolic fragment of the chemical
and thus are chemical specific. Examples of such markers are the chemi-
cal adducts formed with macromolecules such  as  nucleic acids or pro-
terns, particularly blood proteins.  Another example is the presence of
volatile chemicals in exhaled air. That type of marker has been used by
Wallace (1987) to assess exposure to benzene during filling of gasoline
tanks in passenger cars, exposure to tetrachloroethylene in dry cleaning
shops,  exposure to chloroform from contaminated hot shower water in
homes, and exposure to volatile aromatic compounds in tobacco smoke.

-------
            TYPES OF DATA  USED IN DEVELOPING CEELS     37

 Physiologically based pharmacokinetic models (Ramsey and Andersen,
 1984) have been used to relate biological markers of exposure to prior
 exposure conditions.
   Biological markers of the effects of inhalation exposure are the total
 list of clinical signs of chemically induced disease states. The biological
 markers of greatest interest are those that are early predictors of late--
 occurring effects.  Such markers would be invaluable hi assessing what
 levels of pollutants can be tolerated without causing irreversible deleteri-
 ous health effects. For  example,  cell proliferation can result in clonal
 expansion of initiated cells (Swenberg, 1989).  Thus, persistent cell
 proliferation could be a biological marker predictive of an increased
 incidence of late-developing neoplastic lesions. Few markers, however,
 have been validated as predictive of late-stage diseases.
   The third type of biological marker, markers of susceptibility to the
 effects of exposure to airborne chemical pollutants, is a potentially im-
 portant, useful tool. For example, polymorphisms related to acetylation
 and DNA repair can be related to susceptibility to chemically induced
 tumors.  One must be cautious, however, in the use of such indicators
 of susceptibility, as they pose numerous moral and ethical problems.
   The appearance of metabolites in excreta have been useful in the past
 and will be even more useful in the future as biological markers of expo-
 sure, effect, and susceptibility.  When the metabolites are specific to the
 chemical, they will be most useful. When similar compounds related to
 normal diet are found, the  interpretation of the results  might be more
 complex. Other types of biological markers that are undergoing evalua-
 tion include chromosome damage and covalent binding of reactive metab-
 olites to DNA or proteins in circulating blood cells.
                       Route Considerations

Inhalation Route

   The primary function of the respiratory tract, to provide the organism
with oxygen and remove carbon dioxide, presents a particular dilemma
with respect to exposure to airborne toxicants because, unlike ingestion
or dermal exposure, we cannot choose not to breath.  Extreme measures
are often required to remove a toxic pollutant from the breathing zone of

-------
38     CEELS FOR HAZARDOUS SUBSTANCES

the individual.  Understanding the key features of the respiratory tract as
a portal of entry will  permit better predictability of exposure scenarios
resulting in inhalation hazards.   The key features are the respiratory
tract's unique anatomy, the mechanisms by which inhaled particles and
gases deposit in the respiratory tract/and the mechanisms for removal or
clearance of deposited materials.  More detailed information can be
found in reviews by McClellan and Henderson (1989), Gardner et al. (in
press), Bond (1990), Snipes (1990), and Schlesinger (1990).
   The respiratory tract is grossly divided into three compartments:  the
nasopharyngeal region,  consisting of the nasal  cavity,  pharynx, and
larynx; the tracheobronchial region, consisting of the trachea and bron-
chial tree down to and including the terminal bronchioles; and the pulmo-
nary region, composed of respiratory bronchioles, alveolar ducts, alveoli,
and  alveolar sacs.  The last compartment is  the functional area of the
lungs, the site for  gas  exchange.
   Because inhaled pollutants move through the respiratory tract in an
airstream,  the  principles of fluid  dynamics govern their transport and
eventual interaction with the respiratory-tract  surface,  a phenomenon
termed deposition. One of the most important factors in determining if
and  where a pollutant will deposit in the respiratory tract is the aerody-
namic diameter of the pollutant particle.  The aerodynamic diameter of
a particle is defined as  the diameter of  a unit-density sphere with the
same settling velocity as the particle in question.  Thus, a small, very
dense particle will act like a larger, less dense particle  on the basis of
where it deposits in the respiratory tract.
   Fluid-dynamic mechanisms that govern the deposition of uncharged
aerosol particles include sedimentation, inertia! impaction, interception,
and  diffusion.  Diffusion, or movement due to Brownian motion, is most
important for particles with physical diameters of < 0.2 /mi. Sedimenta-
tion arid impaction are the most important mechanisms for deposition of
particles > 0.5 /mi diameter. One generalization that can be made is that
particles  >5 /mi  will deposit primarily in the nasopharyngeal region
because of inertial impaction; particles 1-5 /mi will deposit  in the tra-
cheobronchial  and pulmonary regions because of sedimentation;  and
smaller particles will deposit in the pulmonary region because of diffu-
sion.  The smallest particles (<0.01 /mi),  such as radon progeny, will
deposit in the nasopharyngeal region because of diffusion.
   The site for deposition of fibers in the respiratory tract is also depen-

-------
             TYPES OF DATA USED IN DEVELOPING CEELS     39

 dent on aerodynamic diameter.  The aerodynamic diameter of fibers such
 as asbestos depends primarily on the diameter of the fiber and is indepen-
 dent to a large extent of the length of the fiber.  Inhaled fibers align
 parallel to the airways and act as spheres of approximately equivalent
 diameter. An important mechanism for deposition of fibers  is intercep-
 tion or contact of the fiber with the respiratory-tract walls at the carina.
    The structure of the respiratory tract affects particle deposition.  For
 example,  airway diameter determines the distance that a particle must
 traverse before contacting a surface, and the cross-sectional  area deter-
 mines the velocity at which the airstream travels through the particular
 airway segment.   Variations in pathway lengths in  the bronchial tree
 influence regional  deposition of pollutants.  Pathways with the shortest
 lengths generally will have the highest concentrations  of deposited parti-
 cles in the alveolar region at the end of the pathway. Finally, the pattern
 of breathing and the mode of breathing, such as nasal, oral, or a combi-
 nation, will alter the deposition patterns of inhaled pollutants.
   Critical determinants in the deposition and uptake of inhaled gases and
 vapors include the solubility of the gas in body fluids and its reactivity
 with tissue components or fluid lining the respiratory tract.   Gases or
 vapors that are very  water soluble (e.g., methanol, acetone, hydrogen
 chloride, and ammonia) deposit predominantly in the nasopharyngeal and
 tracheobronchial regions (Buckley et al.,  1984; Gerde and Dahl, 1991;
 Johanson,  1991). For those pollutants, respiratory airflow patterns play
 a major role in determining sites of deposition on the airway lining.  Not
 all water-soluble gases are toxic to respiratory-tract tissues.  However,
 if the gases react with respiratory-tract tissues, deposition patterns influ-
 ence the distribution and severity of toxic responses and might account
 for interindividual differences in susceptibility to associated respiratory-
 tract disease or injury. Formaldehyde is an example of a reactive, highly
 water-soluble vapor.  Studies in rats have shown that the nasal mucosa
 is  the site for deposition of inhaled formaldehyde.  Toxicity studies in
 rats have shown that the anterior nasal cavity is the site for development
 of tumors (Heck et al., 1990).
   Reactive gases with low water solubility, such as ozone, will penetrate
 farther into the respiratory tract.   Exposure to  low  concentrations of
ozone causes injury predominantly in the small airways and the proximal
region, which is the alveolar region closest to the terminal bronchioles.
The proximal alveolar portions of the airways have a small total volume

-------
40     CEELS FOR HAZARDOUS SUBSTANCES

and are protected by a thick coat of mucus.  As the inhaled ozone moves
into the large volume comprising the distal alveoh, the concentration is
diluted. Tlie dilution results in a localized injury to a specific portion of
the respiratory tract (Crapo et al., 1990).                   r^wi;-
   Gases and vapors that are not water soluble but are more lipophihc
(such as the volatile organic compounds, benzene,  styrene  methylene
chloride, and trichloroethylene) are not completely removed by interac-
tion with the surfaces of the upper respiratory tract.  As a result, those
gases can continue down  into the pulmonary region where they might
diffuse across the alveolar epithelium into the blood (Gerde and Dahl,
1991)  Once in the systemic circulation, their disposition will be similar
to the disposition after other routes of administration. The theoretical
equilibrium that can be attained between the volatile pollutant m  fee
blood and that in the alveolar air is generally termed the blood-air parti-
tion coefficient. The extent of pulmonary exhalation of volatile organic
compounds is determined by the blood-air partition coefficient (Pb).  The
proportion of circulating  volatile compound that will be exhaled  in  a
single pass through the lungs when inhaled  air contains none of the
Seal is approximately 1/[1 +  PJ  (NRC. 1986a).   That situation
often occurs if the volatile pollutant is administered by  another route
such as oral, dermal, or injection. Many of the highly toxic, unregulated
pollutants identified by the 1990 Clean Air Act Amendment,  such as
benzene and  1,3-butadiene, fall into the category of volatile organic
     opa       or gases that are inhaled will be deposited on the
 surfaces of the respiratory tract or absorbed into the blood.  A large
 fraction of inhaled pollutants are exhaled.  Fractional deposition of parti-
 cles is dependent upon particle sire.   For large (>5/™) or s™
 «0 01 urn) particles, deposition efficiencies approach 80-90%.  Mini-
 mum deposition occurs for particles in the 0.3- to 0.5-Mm range.  Parti-
 cles in that range are minimally influenced by impaction, sedimentation,
 and diffusion.  As a result, they undergo minimal deposition after inhala-
 tion and most are carried out of the respiratory tract in the exhaled air.
 Average deposition efficiencies for those particles range from approxi-
 mately 10% to 30%.                                       A.^,11.
    Likewise, not all inhaled gases and vapors are deposited on the walls
 of the respiratory tract or absorbed into the blood.  For very reactive
 volatile chemicals, deposition efficiencies are high, and because these

-------
             TYPES OF DATA USED IN DEVELOPING CEELS     41

 reactive gases are not likely to re-entrain into the airstream, uptake by
 the respiratory-tract surface can approach 100%.  For nonreactive va-
 pors, the rate of uptake depends on the Pb. As the Pb increases, the rate
 of uptake increases. The rate of uptake of the chemical will also vary as
 equilibrium  with  blood is achieved.  For vapors with a low Pb,  total
 uptake might amount to only a few percent of the amount inhaled.  The
 majority of the inhaled vapor will be exhaled.  Even for vapors with a
 high Pb, the upper limit for uptake appears to be about 60%.  That is due
 to the cyclic nature of respiration. On inspiration, vapors with a high Pb
 are  temporarily  absorbed  by  the respiratory-tract surfaces.    On
 exhalation, a portion of those vapors will re-entrain into the airstream
 and be exhaled. The remainder will be absorbed into the blood (Gargas
 et al., hi press).
   For pollutants  that have deposited on the respiratory-tract lining,
 mechanisms  for clearance or removal include mechanical clearance via
 the mucociliary escalator and transfer to the gastrointestinal tract, phago-
 cytosis  by  pulmonary macrophages  followed  by migration  of  the
 macrophages to the lymph nodes or to the upper airways, and dissolution
 of the particle in the fluid lining the respiratory tract and diffusion into
the blood.  Mucociliary clearance is a rapid process beginning in a mat-
ter of minutes after the particle has deposited on the ciliated cells lining
large portions of the upper respiratory tract.  Transfer to the gastrointes-
tinal tract, by swallowing of mucus containing particles,  means that all
inhalation exposures to particulate material can result in a significant oral
exposure component (ICRP Task Group on Lung Dynamics,  1966).
  The surface of the bronchial tree is nonhomogeneous. There are areas
that are not covered by ciliated cells,  usually at the bifurcations. Mucus
flow is interrupted over those regions and clearance is decreased. Addi-
tionally, the bifurcations may be sites for increased deposition of parti-
cles due to inertia! impaction as the airstream changes direction. Com-
bined slower clearance and increased deposition indicates that the bifur-
:ations may receive an increased dose of pollutants compared with other
ureas of the bronchial tree.
  Particles also can be cleared from  the respiratory tract by  dissolution
uto the blood.  In  general, particles that are water soluble will dissolve
within minutes or days.  Chemicals that are not water soluble but have
i moderate degree of lipophilicity (log octanol-water partition coefficient
<  106), such as benzo[a]pyrene? nitropyrene, and anthracene, will also

-------
42      CEELS FOR HAZARDOUS SUBSTANCES

clear rapidly into blood (Bond et al., 1985).  Those that are not water
soluble or are highly lipophilic (octanol-water partition coefficient >
106) will not dissolve and will be retained in the lungs for long periods.
Dioxin and other highly chlorinated hydrocarbons are examples of highly
lipophilic pollutants (Henderson et al.,  1988).  Inorganic particles pro-
duced  during high-temperature combustion processes,  such as metal
oxides: oxides of transuranic elements, including plutonium oxides; and
carbon particles, including diesel soot, are examples of water-insoluble
particles.  Those kinds of particles might be retained in the lungs for
years  They will be removed slowly through phagocytosis by pulmonary
macrophages and through dissolution.  In addition to the solubility char-
acteristics of the particle, dissolution rate is also a function of the surface
area of the particle. Large specific surface area (square centimeter per
gram) facilitates dissolution.
   The presence of xenobiotic metabolizing enzymes in respiratory-tract
tissues raises the possibility of both a first-pass metabolism of inhaled
pollutants and a bioactivation of these materials to compounds that can
be directly toxic to  respiratory-tract tissue (Bond, 1990).  All segments
of the respiratory tract, from the nasal cavity  to the periphery of the
pulmonary compartment,  contain enzymes that metabolize  xenobiotic
compounds. However, the distribution of those enzymes throughout the
respiratory tract is  not uniform.   Additionally, within a region of the
respiratory tract, enzymatic activity varies with different cell types.  A
number of studies have demonstrated that the nonciliated Clara cells in
broncniolar tissue contain the highest concentrations of xenobiotic metab-
olizing enzymes. Those cells are also a target for many  toxicants, in-
 cluding carbon tetrachloride, chlorinated ethylenes, and naphthalenes.
 Skin—Percutaneous Absorption

    The skin, as the outermost organ of the body («1.7 m2), provides an
 important protective shield for underlying organs.  Xenobiotic substances
 that make contact with the skin must circumvent this barrier tissue to
 gain entrance into the body. The protective function of skin is imperfect,
 quite variable,  and influenced by a host of factors (Marzulli, 1962).
    Percutaneous absorption (skin penetration) provides a more indirect
 route to the bloodstream than inhalation; nevertheless, both routes share
 certain similarities.  In both cases, the structure, function, and metabo-

-------
            TYPES OF DATA USED IN DEVELOPING CEELS     43

lism at the point of entry modulate the dose of the xenobiotic and its
metabolites that reach the circulatory system before distribution within
the body. Both local and systemic effects are possible, therefore, when
xenobiotic substances make contact with the skin (EPA, 1992).
  When a xenobiotic substance is inhaled, some is retained in various
parts of the respiratory tract and some is exhaled.  Similarly, losses oc-
cur when xenobiotics make contact with skin. The first loss is the evap-
orative loss that occurs at the skin surface.  Additional losses occur if the
substance binds to the outermost layers of dead tissue (stratum corneum).
Metabolism in the underlying living epidermis might bring about chemi-
cal changes in the parent molecule (Bronaughs and Collier, 1991).
  The amount of xenobiotic that reaches the systemic circulation is
determined by certain chemical  and physical factors (concentration,
molecular size, dissociation state, lipid and aqueous solubilities, and
reactivity) and exposure factors (skin contact area, skin site, duration of
xmtact, occlusion, and decontamination efforts). Intact skin is important
in functioning as a barrier to foreign substances in contact with the skin
Flynn, 1990).
  In percutaneous  absorption, the xenobiotic compound bypasses the
iver, so that first-pass liver metabolism is avoided; nevertheless, some
netabolism, albeit with less activity than the liver, may take place in the
;kin itself because it contains metabolic enzymes.
  Metabolism may occur during percutaneous absorption, preceding the
;enobiotic's entry into the bloodstream.  At present, it is not known to
vhat extent metabolic activity in animal skin duplicates that taking place
n human skin. Some exploratory information suggests that human skin
s not as metabolically active as animal skin (Kao and Carver, 1991).
  Skin penetration is quantitatively estimated by one of two techniques.
"he first technique involves measuring the steady-state rate  at which the
«netrant traverses the skin's barrier tissues in an in vitro diffusion-cell
ystem.  The  second technique involves measuring the fraction of the
pplied amount that penetrates skin in vivo (usually with radiolabeled
laterials) (EPA, 1992). With the first technique, results are reported in
;rms of amount penetrating per unit of skin per unit of time.  In the
econd method, results are reported as percent of applied amount that is
bsorbed.
 To normalize effects of concentration, when using the  steady-state
pproach, results can be reported in terms of a permeability constant.
hus, with the simple form of Pick's law,  r = pC, where r = rate in

-------
44     CEELS FOR HAZARDOUS SUBSTANCES
    '      '     "	'    '  -i   •'	• .   	".	• ..•;.,•.    .    ,  	" •„.). '  ....<•   .  •
micrograms  per square centimeter per hour, p  = permeability constant
in microcentimeters per hour, and C = concentration in grams per milh-
liter (Marzulli et al., 1969).
      TOXICITYEND POINTS:  HUMANS AND ANIMALS

                            Mortality

   The most commonly reported end  point for acute animal  toxicity
testing is lethality. The information is expressed as either median lethal
concentration (LC50), when the exposure is by inhalation, or as median
lethal dose (LDSO), when the test substance is administered by some other
route   The data should be expressed  in statistical terms necessary to
describe the lethal response of a substance in a particular population
under a discrete set of experimental conditions.  The exact numerical
value of the median lethal dose should not be used as an absolute num-
ber because many extraneous  factors, such as animal species, strain,
sex', age,  and duration of exposure, can influence or alter the precision
of the data  Lethality data also might not reflect accurately the tull
spectrum of toxicity associated with such exposures, because some chem-
icals (e.g., some  carcinogens and teratogens) might have a low acute
toxicity but produce serious long-term health effects.
   In the absence of median lethality data, alternative data might be
useful in determining acute toxicity, such as identifying the  lowest dose
or concentration (LDLO or LCLO) mat results in the test animal s death
following  exposure.  Such information should be expected to  be quite
variable, but it can be used to compare toxicity of various chemical types
when LDSO data are not available.                         _
   Lethality is  not the only significant end point obtainable from acute
toxicity testing. Much information  can be derived from the slope of the
 dose-response  curve, time of death, and signs of toxicity, all of which
 are critical responses that are useful in characterizing acute toxicity.
                    Clinical Signs and Symptoms
    Physical examination, paying attention to signs and  symptoms, can
 contribute to the biomedical and behavioral assessment of exposure. Eye

-------
            TYPES OF DATA USED IN DEVELOPING CEELS     45

irritation and tearing, for example, are among the most sensitive indica-
tors of excessive exposure to oxidants.  Complaints by exposed individu-
als provide an important criterion on acceptability of pollutants.  Al-
though they might not be related to any direct health effects, complaints
need to be considered as an indication of unacceptable exposure.
                   Effects on Respiratory Tract

  The majority of exposures in a community emergency situation most
likely will be through the inhalation route, which can result in immediate
danger to life and health.  In addition to causing pulmonary effects, the
inhaled toxicants can be rapidly transported from the initial target organ
and cause potentially serious lexicological effects on  other biological
systems.
  The earliest indicator of pulmonary toxicity to high concentrations of
airborne contaminants might be respiratory irritation. Chemicals that are
sensory irritants usually act relatively rapidly, and the irritation tends to
:ease quickly when the chemical exposure is removed.   Such substances
iave a characteristic odor and a tendency to evoke a burning sensation
jf the nasal turbinates, to cause tearing, coughing, and headaches, and
:o induce bronchoconstriction of the airway passages. The irritant effects
:end to be more concentration-dependent than time-dependent and, as
such, are generally noncumulative in nature.  Such responses can provide
i certain warning of potential danger, but they  cannot be considered a
•eliable indicator of pending hazards.
  Inhalation experiments with animals can provide a useful  basis for
>redicting possible immediate and  delayed health effects in humans and
:an indicate the concentration at which such effects might occur.  Vari-
>us pulmonary-function parameters have been measured in animals and
lumans. Many of the techniques are noninvasive and are useful for com-
laring effects in  animals with similar measured responses in humans.
[Tie data can be useful in detecting the presence of impairment, the type
•f impairment, and its severity. The most  common measurements in-
lude lung volumes (tidal volume, vital capacity, residual volume, and
Dtal lung  capacity), mechanical  behavior of the lung (airway resistance
nd compliance), gas-exchange proficiency (diffusion capacity and distri-
>ution), and  ventilatory exchange  (minute ventilation and respiratory
ate).

-------
46     CEELS FOR HAZARDOUS SUBSTANCES

  Whenever possible, multiple end points should be considered in evalu-
ating -the toxicity of hazardous materials because the lung can respond to
airborne pollutants in different ways.  Multiple end points can be used
to assess toxic injury in an emergency situation and are valuable in estab-
lishing safe exposure guidelines.  Table 2 lists toxic effects that result
from short-term inhalation exposures.


                           Neurotoxicity

  When administered in large doses, many chemical substances produce
marked changes in human behavior and neurological function, such as
depression of the central nervous system, that disappear rapidly and with
no apparent sequelae  after exposure has ceased. The same compounds
might produce other potentially irreversible neurotoxic effects in persons
chronically exposed to lower doses for long periods.  The mechanisms
underlying those two (or more) neurotoxic effects might be entirely
different.   Although some compounds, such as n-hexane and toluene,
cause various  types of neurodegenerative diseases after prolonged expo-
sure, others, such as methyl ethyl ketone and acetone, do not appear to
caiise such effects.                                            .
   Evidence that  environmental  agents produce adverse effects in tne
human nervous system can  be provided through clinical evaluations of
exposed humans and studies in experimental animals. Definitive demon-
stration that a substance is neurotoxic comes from complementary inves-
tigations showing that the suspect agent produces the same type of disor-
der  in humans and in one or more appropriate test species. In the ab-
sence of data  on  humans, convincing demonstration that a  substance is
neurotoxic in  an appropriate animal species can be taken as evidence that
the agent is probably neurotoxic in humans.  For example, certain well-
studied classes of chemicals, such  as  the  organophosphates,  reliably
produce human-like disorders in fowl.
   Chronic neurotoxicity might  develop many  years  after  exposure to
certain chemical substances  (e.g., phenothiazines).  Negative results ob-
tained from studies conducted over shorter periods should be interpreted
carefully, because lifetime studies to evaluate agents  for neurotoxic prop-
erties rarely have been undertaken.
   Short-term in vitro methods for assessing neurotoxic activity are being

-------
         TYPES OF DATA USED IN DEVELOPING CEELS
47
TABLE 2 Data for Identifying Effects Associated with Short-Term
  Inhalation Exposures3
Immediate
Effect Effect
Respiratory Effects:
Pulmonary edema +
Pulmonary +
hemorrhage
Spasm-suffocation +
Chronic obstructive —
lung disease
Restrictive lung —
disease
Emphysema —
Cancer —
Hypersensitivity +
Hyperreactivity —
Predisposition to +
infection
Nauseating odor +
Eye Effects:
Corneal clouding +
Burning +
Lachrymators •*•
Availability of
Delayed Procedures and
Effect Protocols

•*• Yes
+ Yes
Yes
+ Yes
+ Yes
+ Yes
+ Yes
+ Yes
+ Yes
+ Yes
- Yes

+ Yes
+ Yes
Yes
Quality
of Data

•H- +
+ + +
•H- +
++
+
+
—
±
±
+ +
±

•f -H-
+•(-+
+ H- +
aData from EPA, 1990a.

-------
48     CEELS FOR HAZARDOUS SUBSTANCES

developed. Tissue-culture systems,  especially those that exploit a com-
plex cellular structure and function comparable to that foundm parts of
the human nervous system,  ultimately might provide surrogates for in
vivo animal bioassays of potential neurotoxicants.  Organotypic tissue
cultures composed of structurally and functionally coupled spinal cord,
dorsal root ganglia, peripheral nerve, and muscle develop specific types
of pathological change  (e.g.,  sensory neuronopathy, axonopathy,  and
myelinopathy) that also are seen in humans and animals challenged with
the same substances.  Tissue-culture systems also have been used to
study alterations of neurotoxic response to one agent («-hexane) by con-
current exposure to a second agent (e.g., methyl ethyl ketone, ethanol,
and toluene).                                                    .
   In some cases, structure-activity considerations  might be usetut in.
assessing whether a substance poses a hazard to the  human nervous
system.  Examples of compounds with structural similarities that consti-
tute neurotoxicants include  some organophosphorus esters (phosphates
and phosphonates), pyrethrins, and 1,4-dicarbonyl aliphatic hydrocar-
bons.


             Reproductive  and Developmental Toxicity

   Single or short-term exposures to chemical agents have been shown to
result in reproductive and developmental toxicity in a number of cases
(Schardein, 1985).  A variety of outcomes are possible,  including alter-
ations in sexual behavior, onset of puberty, fertility, gestation, parturi-
tion, lactation, pregnancy outcomes; (including prenatal and early postna-
tal death,  structural  abnormalities, growth retardation, and postnatal
functional deficits), premature reproductive senescence, and modifica-
tions in other functions that are dependent on the integrity of the repro-
ductive system (Kimmel, 1991).
   A number of stages in the reproductive cycle and during development
are known to be sensitive to exposure to toxic agents,  and the types of
effects  seen may depend on the timing of exposure  (Wilson, 1973;
Dixon, 1986).  Available data are often from studies employing expo-
sures to pregnant animals during  major organogenesis, and some extend
the study into the preweaning period to examine the products of concep-

-------
            TYPES OF DATA USED IN DEVELOPING CEELS     49

tion and the growth and development of the liveborn offspring.  Addi-
tional information on reproductive performance in exposed adults and on
pregnancy outcomes is sometimes available from two-generation studies.
Data from those studies usually form the primary data base for evaluating
reproductive and developmental hazards (Manson and Kang,  1989;
Zenick and Clegg, 1989).  When human data are available, they should
be considered with the animal studies to make judgments about hazard
and risk (EPA,  1991).
  Inhalation exposure, but more often oral exposure, is used in conduct-
ing reproductive  and developmental toxicity studies in  experimental
animals. Data from oral exposure studies should be considered in deter-
mining CEELs  when evaluating the potential  for an  agent to produce
reproductive  or  developmental  effects  from  inhalation exposure.
Pharmacokinetic data from appropriate studies also should be considered
for route-to-route extrapolation of reproductive and developmental toxici-
ty data.
  Acute or short-term exposure levels that produce adult systemic toxici-
ty are likely in some, but not all, cases to cause reproductive or develop-
mental effects. With some agents, reproductive or developmental chang-
es may  be produced at doses that  do not cause  other forms of overt
toxicity, and these agents are of greatest concern in assessing risk for
humans  (Zenick and Clegg, 1989; EPA, 1991).
  The adjustment of doses  to account for duration of exposure in repro-
ductive and developmental toxicology has not been explored  in great
ietail. EPA's "Guidelines for Developmental Toxicity Risk Assessment"
[1991) gives the most recent guidance on that issue.  As indicated in the
EPA guidelines, a single exposure is assumed to be sufficient to produce
idverse  reproductive and developmental effects.  Therefore, exposure
jstimates and doses usually are based on a daily dose that is not adjusted
ibr duration or pattern of exposure unless pharmacokinetic data are avail-
ible that indicate bioaccumulation with repeated exposure. In developing
DEELs,  data from studies  that are  most similar to the  scenario  (i.e.,
ihort-term exposures from one to several hours per day) of exposure by
nhalation are most appropriate for use.  When such data are not avail-
ible, data  from other types of studies should be considered,  and the
guidance discussed  above should be employed, incorporating scientific
udgment as  necessary.   Further  development of short-term exposure

-------
50     CEELS FOR HAZARDOUS SUBSTANCES

guidelines is under discussion by several groups, including scientists at
EPA, and work is in progress to develop more appropriate methods tor
risk assessment in this area.
                         Immunotoxicity

   Immunotoxicology is a scientific discipline that explores the effectsof
physical, biological, and chemical agents on the immune system.  The
immune system is complex and highly sophisticated and is self-regulated
as well as influenced by other systems, namely the endocrine and central
nervous systems.   Immunotoxicology has  experienced  considerable
growth since its inception nearly two decades ago. Today, several exper-
imental  procedures  are available that  have been validated  to assess
xenobiotic-induced immune dysfunction in animals.  Fewer assays are
available to evaluate immune dysfunction in humans.  Nevertheless, pro-
cedures are available and others are being  developed to assess immune
function in animals and humans. Care must be taken in using immune
data for risk analysis to ensure that the data are a correlate of immune
function (i.e., that it represents some type of response that  correlates
with innate or naturally acquired resistance to disease, neoplasia, autoim-
munity, and hypersensitivity).
   Procedures that can be used in animals and humans to assess  immune
function are measurement  of humoral (B-cell) antibody responses to
 challengeby T-dependent antigens; delayed-type hypersensitivity respons-
 es1  cytotoxicity of T lymphocytes, macrophages, natural killer cells  and
 lymphokine-activated cells; and cytokine production and activity. Tests
 developed specifically  for humans include assessing antibody levels to
 ubiquitous antigens (blood groups); responses to protein and polysaccha-
 ride antigens (such as diphtheria, tetanus, poliomyelitis, pneumococcus,
 and meningococcus); primary (keyhole limpet hemocyanin) and second-
 ary (Candida, diphtheria, and tetanus) delayed-type hypersensitive reac-
 tions; use of multiple-antigen skin-test kits; natural-killer-cell  cytotoxici-
 ty  and serum levels of cytokines (such as interleukins 1  and  2).  Al-
 though tests for  humoral, cellular, and nonspecific immunity have great
 value in studies of profound hereditary immunodeficiency states,  they
 have not been specifically evaluated for their ability to detect more mod-
 est immunodeficiencies  that  might occur  in individuals exposed  to
 immunotoxic  agents.  Therefore, although these procedures have been

-------
            TYPES OF DATA USED IN DEVELOPING CEELS    51

 validated in animal models, their use as biomarkers in humans must be
 validated prospectively in populations exposed to putative immunotoxi-
 cants and in control groups to establish their ability to predict develop-
 ment of disease associated with xenobiotic-induced immunodeficiency.
   Several other immune procedures might be applied eventually in as-
 sessing xenobiotic-induced immune dysfunction in humans.  However,
 those procedures will require considerable testing before they can be
 accepted as positive correlates of immune function for risk-assessment
 purposes.
                          Carcinogenicity

   Current theories based on experimental research in animals indicate
that at least three distinct stages exist in the development of carcinogene-
sis:  initiation, promotion, and progression.  Each of the stages appears
to be influenced by such factors as age, heredity, diet, metabolic activity,
and previous and current exposures to xenobiotic compounds, which can
interact and potentiate, antagonize, or synergistically influence the devel-
opment of tumors.  Mouse skin serves as  a good model for chemical
carcinogenesis; the model's additional use is for studying the mechanisms
of action of initiators and promoters.  Rat and mouse liver models also
have led to useful discoveries about stages of tumor development, and
promotion has been demonstrated in other tissues, such as the lung and
breast.  Some chemicals, such as asbestos, estrogens, and some constitu-
ents of cigarette smoke,  appear to be mainly tissue-specific promoters,
whereas others, such as 2-acetylaminofluorene,  can  act as promoters in
the mouse bladder but as initiators  (and promoters) in the liver  of the
same animal.  The effect of a promoter can vary depending on the se-
quence in which it is administered with an initiator.  Therefore, it might
be desirable to test potential promoters and inhibitors of promotion in
vivo in several species and strains, to examine several organs for the
response,  and to test different sequences of administration.
  Recent  investigations into the role of oncogenes seem to support the
multistage theory of carcinogenesis. Of the oncogenes thus far discov-
ered, most have been either of two types, immortalizing genes or trans-
forming genes, which may be regarded as the biological counterparts of
initiation and progression in tumor induction.  A third type of oncogene
recently identified might cause a cancer cell to metastasize.  Ideally,

-------
                                                                    , '"i!
                                                                    i
52     CEELS FOR HAZARDOUS SUBSTANCES

then, the risk modeler should consider at least three stages of tumor
induction to be consistent with human and experimental evidence. Risk
models also must take into account the diversity of promoting agents and
their multiplicity of action at the cellular level.
  The most widely used tests for carcinogen evaluation and regulation
are long-term animal bioassays, which are considered predictors of re-
sponse in humans.  However, animal bioassays not only are time con-
suming but also are difficult to conduct, are extremely expensive, do not
necessarily predict the target organ in humans, and cannot duplicate the
wide range of susceptibilities, simultaneous exposures, and genetic back-
grounds of the human  experience.
   Short-term in vitro assays of cultured cells, requiring only a fraction
of the time, can test for toxic end points manifested during the initiation
phase of carcinogenesis. Other tests, such as cell transformation assays,
examine toxic end points for later events in carcinogenesis.  Short-term
tests provide information on the mechanism of carcinogenicity and can
contribute to the weight of evidence.
   There is a need to develop short-term tests for promotion. A battery
of tests is necessary for screening chemicals for genetic toxicity. False-
negative and false-positive results might occur because the  tests do not
reflect the complexity  of interactions in the  whole animal, and although
knowledge of the chemical's principal mechanism of action  is crucial to
the  selection of short-term  tests with meaningful predictive value, that
knowledge is rarely available.  Despite the fact that  in vitro tests are
technically difficult to design and interpret, they have considerable poten-
tial  and should be developed further.
                            Mutagenicity

   Genetic toxicology is" concerned with health effects resulting from the
 mutagenic action of physical and chemical agents on DNA. Mutagenesis
 involves gene (molecular) and chromosomal (gross structural and numeri-
 cal aberrations) changes in germ and somatic cells.  Mutations  in both
 types of cells do not result in immediate adverse health effects but might
 result in late effects.   Germ-cell mutations affect the subsequent and
 future generations of the exposed person, whereas somatic-cell mutations
 can affect the individual in whom they are produced, primarily by initiat-
 ing carcinogenesis and, in the embryo-fetus, teratogenesis.   Low-dose

-------
             TYPES OF DA TA USED IN DEVELOPING CEELS     53

 exposure to some chemicals is capable of inducing mutations, but acute
 high-dose exposure that permits survival  requires special attention be-
 cause not only is the probability of mutation induction higher but also
 more mutations can be induced by high-dose exposure than by low-dose
 exposure. The increase in mutations gains significance in view of recent
 evidence suggesting that additional gene mutation is involved in the later
 stages of carcinogenesis (Stowers et al., 1987; Kennedy, 1991).
   Germ-cell mutations, either gene or chromosomal, usually are deleteri-
 ous to  some degree in the individual who inherits them,  the mode of
 expression—recessive,  dominant, or polygenic—varying with the gene
 affected.  The involvement of somatic-cell mutation in carcinogenesis is
 now widely accepted.  Carcinogenesis is a complex multistage phenome-
 non, three stages of which have been identified:  initiation, promotion,
 and progression.  The best understood is  initiation (i.e., a change in a
 cell that leads eventually, through  a chain of subsequent events, to  a
 clinical tumor).  The finding that most human tumors have arisen  from
 a single  cell (Nowell, 1991) and  the  discovery that single  mutated
 growth-regulating genes or growth-suppressing genes (oncogenes) can
 cause cancer (cf. Bishop, 1987) leave little doubt that mutation is the
 main initiating event.   Progression is probably controlled by different
 genes that permit transformed cells to  invade and metastasLze  (Hill,
 1990).
   Over the last two decades, many data have been accumulated on the
 mutagenic effects of many chemicals, primarily  with non-human cell
 systems. In recent years, data from human cells have been  emphasized.
 Research  on in vivo gene mutation has centered on red blood  cells
 (RBCs) and T  lymphocytes (cf. Albertini et al., 1990).  In RBCs, the
 end  points  are mainly mutations  in the hemoglobin and  the  GPA
 (glycophorin-A) genes, and in T lymphocytes, the end points are mainly
 mutations in the  HGPRT (hypoxanthine-guanine  phosphoribosyltrans-
 ferase) and the HLA (human leukocyte antigens) genes.  Chromosomal
 changes have been and are being extensively examined in lymphocytes
 hi vivo and in vitro including structural and numerical aberrations, sister
 chromatid exchanges,  micronuclei,  unscheduled DNA synthesis, and
single-strand breaks in  DNA.
   The rapid growth of molecular biology has resulted in a new approach
to in vivo genetic toxicity  in humans, namely, molecular epidemiology
(Weinstein and Perera,  1982; Perera et al., 1985,  1991; Perera, 1990;
Marx, 1991). This interdisciplinary field ties the highly sensitive labora-

-------
III I  !	 I  t!»
                                                    54     CEELS FOR HAZARDOUS SUBSTANCES

                                                    tory techniques for detecting initial chemical interactions with cell targets
                                                    to the analytical methods of epidemiology, one goal of which is early
                                                    detection of individuals at high risk for cancer, thereby permitting inter-
                                                    vention or prevention of the disease.
                                                       Molecular epidemiology focuses on detecting molecular markers (pri-
                                                    marily carcinogen-DNA or carcinogen-protein  adducts) of damage in
                                                    populations of people who are exposed to chemicals known to be or
                                                    suspected of being carcinogenic.  Perera (1990)  provided a summary of
                                                    the laboratory techniques.  One of the advantages of molecular epidemi-
                                                    ology is that adducts are formed during or very quickly after exposure
                                                    but their early formation can be  a disadvantage when late sampling of
                                                    subjects is necessary because many adducts  are relatively short-lived
                                                    Among other advantages are the ability  to  detect minute  amounts of
                                                    molecular markers (one adduct in 109-1010 nucleotides in DNA); protein
                                                    adducts in body fluids (blood, urine, breast milk, and semen), exhaled
                                                    air, and adipose tissue;  and  new proteins, such as those produced by
                                                    activated oncogenes.
                                                       With one exception (cf. Perera et al., 1991), results do not yet allow
                                                    quantitative risk assessment, but they do show that molecular epidemiolo-
                                                    gy is promising (e.g., Hemminki et al., 1990; Mayer et al., 1991) and
                                                    probably  will be one of the prominent methods for identifying and as-
                                                    sessing cancer risks in humans.  As Perera (1990) has pointed out  mo-
                                                    lecular epidemiology is not a panacea but  supplements existing methods
                                                    for environmental monitoring,  experimental testing, and epidemiology.
                                                        '      '   •       ".!.•„'• .'••'•' • t ,,i. •      •             . ...... , I
                                                        • •     ••"                   '   '      •   '        '     '
                                                                        OTHER ORGAN EFECTS

                                                                               Ocular Effects
  Acute eye irritation (including inflammation or opacity of the cornea
and irritation of the bulbar and palpebral conjunctiva and the iris) is a
prominent feature of ocular effects that must be considered in relation to
emergency exposures to hazardous chemicals.  The cornea and ins are
important in vision and thus those effects warrant special attention.
  A measurement of the pH and buffering capacity of an acid or alkali
provides a  first look at the potential' of the xenobiotic to irritate the eye
and produce possible serious deep injury. The intimacy and duration of
contact between the offending substance and the eye as well as the pene-

-------
             TYPES OF DATA USED IN DEVELOPING CEELS    55

 tration capacity and chemical reactivity of the xenobiotic with tissue
 components of the eye are important predictors of ocular damage.  Use-
 fulness and availability of water as an eye decontaminant also might be
 critical determinants of ocular damage.
    The most capable test of disclosing ocular irritation potential is the
 Draize rabbit eye irritation test; however,  if serious ocular damage is
 expected, it can be avoided by first applying the test material to skin for
 a short time.  As the skin and eye are anatomically similar, skin, which
 is tougher and more easily repaired, can serve as a surrogate for the eye.
    Organic  solvents, alkylating agents, surfactants, and detergents  are
 known to produce  serious deep ocular damage or delayed injury on
 contact with the eye; less toxic chemicals might produce only stinging,
 lacrimation, and discomfort (Grant,  1974).
                          Dermal Effects

   As the outermost organ of the body, skin might be the first tissue to
 come in contact with xenobiotic chemicals during emergency exposures.
 Systemic or local effects are possible  following  skin contact with
 xenobiotics.  Organophosphate pesticides and other toxic chemicals might
 penetrate skin without apparent gross skin damage, producing systemic
 effects.  Some skin penetrants might be more toxic by the percutaneous
 route than by the oral route if metabolic activity is high by the oral
 route. Herbicides, fungicides,  and rodenticides are typical farm chemi-
 cals that might produce systemic toxicity when they contact skin and are
 absorbed into the bloodstream following percutaneous penetration.
   Skin irritation, skin sensitization, and phototoxicity are local effects
 that might occur when xenobiotic substances come in contact with the
 skin during emergency exposures. Skin irritation is a local inflammatory
 response, skin sensitization is  an immune response involving skin and the
 immune  system, and phototoxicity is  an exaggerated sunburn-like skin
 response produced by chemicals that are activated by light.
                         Hepatic Effects

  The liver is the single most important organ concerned with protecting
the body against invasion by potentially toxic xenobiotic chemicals and

-------
!"'ii!!i!i: I, •'": I 	 !	.I'lE1 •' \\S' ih iiilllJ^l' "'I!1'?"' ''	'il "'•!"" "!'ii,	ill  ' lii'll1 'It'1
                                                          T'li1' !N.' '"• '"S,"!1 !' ,,i.I'.
                                                      56     CEELS FOR HAZARDOUS SUBSTANCES
                                                                                      1 " i1 •  '"       ' ' '         ' '  ' !j 	''
                                                      therefore is among the organs most vulnerable to chemical insult.  The
                                                      liver avidly accumulates many chemicals, often after oral administration,
                                                      leading to the so-called first-pass effect.  However,  accumulation of
                                                      chemicals by the liver and toxic responses of the liver also  can occur
                                                      after exposures via inhalation or other routes.
                                                        Some chemicals, such as phosphorus, can be directly toxic to the
                                                      liver.  Chemicals passing through the hepatic circulation can be acted
                                                      upon to yield detoxification products  or can be activated metabohcally to
                                                      become more toxic chemical entities.   Metabolic activation of many
                                                      chemicals leads  to  liver damage.  Mechanisms by which liver injury
                                                      might occur include the accumulation of lipids, lipid peroxidation, cova-
                                                      lent binding of reactive metabolites to critical cellular  macromolecules,
                                                      depletion of antioxidants such as glutathione leading to oxidative stress,
                                                      interactive lexicological effects originating in Kupffer  cells that lead to
                                                      damage in hepatocytes, disruption of calcium  compartmentation, and
                                                      other aberrations that can lead to cell death.  The effects  of chemicals
                                                      include local or generalized lesions  leading to cell death and necrosis,
                                                      inflammation, fatty infiltration, cholestasis, cirrhosis, or carcinogenesis.
                                                         Underlying these injuries are factors such as excessive alcohol intake,
                                                      diabetes mellitus, or starvation that induce increases in the levels of cyto-
                                                      chrome P-450HE1,an enzyme involved in the metabolic activation of
                                                      many halogenated hydrocarbons and other chemicals to lexicologically
                                                      active or carcinogenic species.  Some halogenated hydrocarbons, such as
                                                      halothane, can cause immunological responses in liver following repeated
                                                      exposure. Nutrilional deficiencies rnighl exacerbate potenlial liver injury
                                                      caused by chemicals such as elhanol.
                                                         Liver damage can be delected by the appearance of jaundice, measure-
                                                      ment of hepatic enzyme levels in circulating blood, finding of radioiso-
                                                      lope accumulation  in the liver, use of various imaging  techniques, or,
                                                      more invasively, evaluation of liver samples obtained  by biopsy.

-------
        Risk Assessment
                           OVERVIEW

 Recommendations for permissible exposure limits to chemicals usually
 have been determined by the nature of the adverse effect, the exposure
 scenario, and the  anticipated susceptibility of the population for which
 the permissible exposure limit is being set.  The quality of such recom-
 mendations for materials that have been in use for a long time is highest
 when the recommendations are based on epidemiology or clinical toxicol-
 ogy studies carried out under conditions that are identical to those for
 which the recommendation  is intended.  Even in that ideal situation,
 however, the recommendations are for the average or typical population
 and  might not be predictive  for a specific individual within the popula-
 tion. When the information  on humans is not ideal for setting permissi-
 ble exposures for a chemical or exposure situation, the first approach is
 to extrapolate from the best available human data, recognizing that differ-
 ences in dose, exposure route, rate, or duration; population differences
 in sex,  age, occupation, or dietary habits; and other differences might
 bias or invalidate the permissible limits set. If human data on the desired
 chemical is lacking or inadequate, several other approaches suggest them-
 selves.  For new materials, laboratory data usually from animal studies
 combined with data on surrogate chemicals might be helpful. The chem-
 ical and biological properties  of the target chemical should closely resem-
 ble those of the surrogate chemical, for which human data are adequate.
 The  use of animal data to predict the effects of human  exposure to a
 chemical is most useful when the test species and humans respond in the
 same manner to the chemical.  Unfortunately, data on corresponding
 responses are rarely available.  Studies can be carried out in animals that
 are neither ethical nor practical in humans.  Animal  studies can be de-
 signed, however, to provide information that can contribute to the appro-
priate setting of guidelines for human exposure to chemicals.

                               57

-------
5S     CEELS FOR HAZARDOUS SUBSTANCES

  The use of animal data to establish the toxicity of a chemical is called
hazard evaluation. Full hazard evaluation involves developing informa-
tion concerning the chemical, the exposure conditions, and the adverse
effects in the test species. The first step in the hazard evaluation process
is to identify the adverse effects produced by (acute or chronic) exposure
to the agent. The second step is to establish dose-response relationships
for each  of the adverse effects. Ideally, the studies would provide infor-
mation on the effects of administration by various routes at different rates
and durations  of exposure arid information on other test species.  That
information and data on the chemical  and physical-chemical properties
of the  agent, kinetic data in various species, genotoxicity studies, devel-
opmental toxicity and reproductive studies,  and  other types of target-
organ  and mechanistic  studies constitute the toxicity data base for the
chemical.  Of course, if the material  turns out to be so toxic that the
manufacturer does not wish to put it on the market or if the likely per-
missible exposure is too low; much of the additional work need not be
done  The third step is to determine, whether the information in the data
base is appropriate and relevant for extrapolating to humans under actual
exposure conditions.  In situations where the information is adequate and
relevant and where there is a threshold or no-effect level for a specific
adverse  effect on which to base the extrapolation, the final  step in the
process is to divide the experimental threshold or no-observed-adverse-
effect  level (NOAEL) by appropriate  uncertainty factors to develop an
upper  limit for exposure to the material. A modification of that approach
has been proposed by Crump (1984a), in which the fitted dose-response
curve  is used to calculate benchmark doses or concentrations for low-
incidence responses (1%, 0.1%, etc.), and then appropriate uncertainty
factors are used to develop the reference dose or concentration (RrD or
RfC)  The benchmark-dose approach  is more responsive to the number
of animals used at each dosage level  and avoids  the use of an arbitrary
NOA^L value but is basically an extension of the traditional RfD-RfC
approach.  One of the advantages of the RfD-RfC approach is that it is
simple and easy to understand.  However, in practice it involves a great
deal of  judgment in selecting test data and uncertainty factors.   The
 selection of an uncertainty factor should reflect not only the confidence
of the ©valuator in the quality and relevance of the data for the prediction
but also the differences  in susceptibility between  the test and target spe-
 cies and the differences  among individual members of the exposed popu-

-------
                                         RISK ASSESSMENT     59

 lation.  The goal of the benchmark approach is to predict a reference
 dose for exposure to an agent below which a significant increase in toxic
 effects is not expected above background levels.  This approach for
 protecting the health of specific groups, such as  workers,  is the most
 well-established and most widely used approach against adverse effects
 from  exposure to chemicals, but it is not applicable for  materials that
 produce nonthreshold toxicity, such as cancer.
   Although it has been argued that there should be a threshold for every
 adverse effect, the threshold might  be virtually impossible to distinguish
 from the zero- or no-exposure level.  In addition, there are theoretical
 reasons why some toxicities are no-threshold phenomena.  Therefore, in
 those  situations where a threshold  or no-effect level cannot be demon-
 strated or theoretically should not exist, an alternative approach is need-
 ed.  For example, a so-called quantitative risk-assessment (QRA) ap-
 proach is currently used for carcinogens, which assumes zero response
 only at zero dose.
   Several models have been proposed for use in extrapolating low-dose
 (lower than the lowest experimental dose) responses to carcinogenic
 materials within the test species.  The earliest models were  so-called
 susceptibility models—the probit  and logit models.  Those models were
 rapidly replaced by the multistage model of Armitage and Doll (1961),
 which was consistent with the age-related incidence patterns of human
 cancer.  More recently, other biologically driven models  have been
 proposed and  are being studied extensively, including the Moolgavkar
 two-stage model (Moolgavkar and Venzon, 1979) and several pharmaco-
 dynamic models.  At low doses, the Armitage-Doll model predicts a
 linear dose-response curve. Regulatory agencies are using the linearized
 multistage model most frequently.
   For between-species extrapolations, there are concerns about the ap-
propriateness of extending the results from one species to another.  Cur-
rently, extension of results from one species to  another hinges on wheth-
er milligrams per kilogram of body weight, milligrams per square meter
of surface area, or some other allometric approach is used.  The surface-
area approach  (milligrams per square meter) leads to the most risk-aver-
sive "acceptable" doses and is  favored by EPA.  In addition,  findings
from pharmacokinetic studies might assist in extending the results from
experimental species to humans. There are arguments that some carcino-
gens operate through indirect or secondary mechanisms, which are relat-

-------
                                                         -,!! " 'I!!' •• < ?
                                                          I ('I:'  If
60     CEELS FOR HAZARDOUS SUBSTANCES

ed at times to excess cell proliferation (thus increasing the probability of
mutations leading to cancer) or perhaps to hormone-like activity.  Under
those circumstances, it is argued that secondary carcinogens would exhib-
it thresaold-like behavior and for them the uncertainty factor and refer-
ence dose approach would be appropriate.
   Exposure limits frequently are established for different purposes, some
of which  are not related directly to health.  At times, exposure limits
have been set on best-available technology (BAT), analytical chemistry
detection  limits, and the flammability and  explosive properties of the
chemical.  There are exposure limits established for  different popula-
tions, such as workers (ACGIH's TLVs,  OSHA's  PELs,a NIOSH^s
RELs,b AIHA's WEELs,c and the International Labor Organization s
EOLs^),  and for military and  space applications (COT's EEGLs e
SPEGLs f and SMACs&).  The purpose of most exposure limits is self-
explanatory.   For those limits that are not health-based, it would be
helpful if the basis for establishing the values were provided. Although
the general procedure used to establish most of the values is documented,
it  is not always  clear from the documentation on individual chemicals
how the decision was  reached and whether the recommendation was
based on  human or animal data or on threshold or QRA analysis. Ex-
posed persons need to be better informed about the level of protection
that is provided  by the  exposure limits, although current  right-to-know
laws have improved communication.  Those using exposure limits also
should be concerned about the level of protection provided by the expo-
sure limit for different segments of the population, such as  pregnant or
older workers. Setting a range of limits might provide a good indication
of the protective value  by reflecting the biological variation associated
with responses to exposures.  Using a  range of values  rather than a
single number could also be advantageous in dealing with different expo-
    permissible exposure limits.
    bRecommended exposure limits.
    ''Workplace emergency exposure limits.
    dEmergency occupational exposure limits.
    Emergency exposure guidance levels.
    fShort-term public emergency guidance levels.
    ^Spacecraft maximum allowable concentrations.

-------
                                         RISK ASSESSMENT     61

 sure situations.  The range concept needs to be  evaluated by all  the
 groups involved in establishing exposure limits.
    Setting exposure limits for workers,  military personnel, and other
 population groups exposed to chemicals is the major practical application
 of the risk-assessment process. Groups such as the  ACGIH's TLV com-
 mittee and the NRC's COT have been engaged in this effort for over 40
 years; the U.S. Food and Drug Administration has been using essentially
 the same process to regulate drugs, cosmetics, and  food additives for an
 even longer period. The basic elements of the process are case-by-case
 evaluation, priority of human over  animal data, use of the threshold
 approach to establish safety, and reliance on good science and judgment
 rather than rigid protocols to achieve the optimal balance between bene-
 fits and risks.
                 NONCARCINOGENIC EFFECTS

             Traditional Threshold-Based Approaches

   Risk assessment for health effects other than cancer traditionally has
 been based on the concept that an adverse health effect will not occur
 below a certain level of exposure, even if exposure continues  over  a
 lifetime. The existence of a so-called threshold dose is supported by the
 fact that the toxicity of many agents is manifest only after the depletion
 of a known physiological reserve.  In addition, the biological repair-
 capacity of many organisms can accommodate a certain  degree of dam-
 age  by reversing toxic processes  (Klaassen,   1986;  Aldridge,  1986).
 Above the  threshold dose, however, the homeostatic physiological pro-
 cesses that  allow compensatory mechanisms to maintain normal biologi-
 cal function might be overwhelmed, leading to organ dysfunction. Thus,
 the objective of traditional lexicological risk assessment is to establish a
 threshold dose below which adverse health effects  are not expected to
 occur or are extremely unlikely to occur.
  The concept of a no-effect level was  introduced by  Lehman  and
Fitzhugh (1954),  who proposed that an acceptable daily intake (ADI)
could be calculated for contaminants in human food. That concept was
endorsed by the Joint FAO-WHO (Food and Agricultural Organization
and World  Health Organization) Expert Committee on Food Additives

-------
62     CEELS FOR HAZARDOUS SUBSTANCES

in 1961 and subsequently adopted by the Joint FAO-WHO Meeting of
Experts on Pesticide Residues in 1962 (cf. McColl, 1990).  Formally,
the ADl was defined by

                        ADI = NOEL/SF,

where NOEL  stands for the noK)bserved-effect level in toxicological
studies (the highest exposure level at which there are no statistical y _ or
biologically significant increases in frequency or seventy of effec ^s be-
tween the exposed population and its appropnate conttol) and SF repre-
sents a safety factor to allow for variations in sensitivity to fce tes  agent
in humans as  compared with experimental animals and for variations
within tie human population. Those two sources of vana ^f*™
been accommodated through the use of a 10  x 10 = 100-fold SF as
reviewed  by the NRC's Food Protection Committee (NRC, 1970).
   In 1977  the  NRC's Safe Drinking Water Committee reviewed the
methods that had evolved for establishing ADIs and  made several signifi-
cant recomiileiidations. First; the committee proposed that the NOEL be
expressed in milligrams per kilogram of body weight rather than ^milli-
grams per kilogram of diet to  adjust for dietary consumption ipatt m.
Second, the committee suggested reducing the trad.tional 100-fold SF to
only 10-fold in the presence of adequate dose-response data derived from
human studies. Third, the committee proposed augmenting the tradition-
 al  100-fold SF  to 1,000-fold in the absence of adequate toxicity data
        uthe use of SFs is now accepted practice in toxicological risk
 assessment for health effects other than cancer, the NOEL-SF approach
 is subject to certain limitations (Munro and Krewski, 1981) Because toe
 ADI is only an estimate of the assumed population threshold dose or the
 true no-effect  level,  it  does not  provide absolute assurance of safety
 (Crump,  1984a).  Smaller experiments tend to yield larger NOELs  and
 hence larger ADIs, than do  larger, more sensitive experiments (Mantel
 and Schneiderman, 1975). SFs of 10-fold that are used to account for
 both inter- and intraspecies variation in sensitivity might not aftord ade-
 quate protection in all cases. Therefore, the ADI should not be viewed
 as possessing a high degree of mathematical precision but rather as a
 guide to  exposure limits that are  not expected to present serious health
 risks to humans.
                                                                  .-:	11

-------
                                         RISK ASSESSMENT    63

    EPA has recommended using the term uncertainty factor (UF) rather
 than safety factor in recognition of the uncertainty associated with the
 ADI, and the ADI has been relabeled as a reference dose (EPA, 1988;
 Barnes and Dourson, 1988). EPA also introduced an additional modify-
 ing factor  (MF) to account for specific scientific uncertainties in  the
 experimental data used to establish the RfD.  With the NOAEL defined
 as the highest experimental dose at which there are no statistically signifi-
 cant adverse effects  in the test animals  beyond that exhibited by  the
 control group, the RfD is defined by the equation

                    RfD  = NOAEL/(UF x MF).

 In the present  context, an adverse effect  is defined  as any  effect that
 contributes to the functional impairment of an organism or that reduces
 the ability of the organism to respond to additional challenges (Dourson,
 1986).  When the data do not demonstrate a NOAEL, a LOAEL (lowest-
 observed-adverse-effect level) may be used. A LOAEL is defined as the
 lowest experimental dose at which a statistically significant adverse effect
 occurs.
   Five factors may contribute to the composite UF x MF. The factors
 are (1) the need to accommodate human-response variability, including
 sensitive subgroups; (2) the need to extrapolate from animal data to
 humans when human exposure response data are unavailable or inade-
 quate; (3) the need to extrapolate from subchronic to chronic exposure
 data when the latter are unavailable; (4) the need to account for using a
 LOAEL when a NOAEL is unavailable; and (5) the need to extrapolate
 from a data base that is inadequate or incomplete. The first four factors
 are incorporated into the UF, whereas the fifth, which addresses the
 quality of the data base, is represented by the MF.  Factors between 1-
 and 10-fold commonly are used to account for each of those sources  of
 uncertainty.
   EPA has adapted the oral RfD method to estimate inhalation reference
 concentrations (RfCs) to be consistent in setting exposure levels for
health effects other than cancer (EPA, 1990b).  The inhalation RfC
method departs from the oral RfD paradigm by using dosimetric adjust-
ments to scale the exposure concentration for animals to a human equiva-
lent concentration.
   For health effects other than cancer, the RfD and RfC methods would

-------
	ll1
64     CEELS FOR HAZARDOUS SUBSTANCES

be useful in setting CEELs.  In that approach, UFs between 1 "and 10
might be used for each source of uncertainty listed above, depending on
the nature and severity of the adverse effects.  For setting CEELs; the
duration extrapolation, if any, would be in the opposite direction of that
indicated in factor 3 above (i.e., it would be  from long-term to acute
exposure).   An exposure-extrapolation approach such as Haber s law
might be accompanied by the highest UF, 10, if there were considerable
uncertainty as to the validity of Haber's law.
    '       '''"    '	
                                                                 Mathematical-Model-Based Methods

                                                     Recently,  efforts have been devoted "to mathematical  and statistical
                                                   modeling of adverse health effects other than cancer, the objective being
                                                   to exploit the shape of the dose-response curve and to improve the preci-
                                                   sion of the estimates of acceptable exposure levels of a chemical.
                                                   Benchmark Dose

                                                     The concept of a benchmark dose (BD) has been proposed "as a re-
                                                   placement for the NOAEL (Crump, 1984a; see also Kimmel and Gaylor,
                                                   1988)  The BD is defined as a statistical lower confidence limit on the
                                                   dose that produces a predetermined relatively small increase in response
                                                   rate  (risk) of 1-10%.  It is calculated with the aid of a matoematical
                                                   dose-response model that is fitted to the experimental data.  The BD is
                                                   promoted as representing a toxicologically relevant quantity because it is
                                                   in the experimental dose range, although the BD does not usually coin-
                                                   cide with an actual experimental dose level. The BD makes appropriate
                                                   use of the sample size, as  reflected in the magnitude of the confidence
                                                   limit  It exploits the shape (steepness) of the dose-response curve in the
                                                   experimental range but does not depend strongly on the particular mathe-
                                                   matical model used, because the model is not followed below the  1%
                                                   response level.
                                                      Crump (1984a) recommended redefinition of the ADI as

                                                       "",':    	,     "'.    ADI = BD/SF.  '; '""",; .  	"'   '  '"

                                                   Gaylor (1983) proposed that the BD could be used to determine the  size

-------
                                        RISK ASSESSMENT     65

 of the UF needed to achieve a desired low level of risk. He argued that
 the relationship ADI = BD/SF can be viewed as equivalent to linear
 low-dose extrapolation. EPA's recent publication "Guidelines for Devel-
 opmental Toxicity Risk Assessment" (EPA, 1991) recommends that a BD
 be calculated to supplement the NOAEL or LOAEL used in determining
 an RfD for developmental toxicity  (RfDDT).   The BD  is considered
 useful  as a comparison and  as a possible  option to the LOAEL or
 NOAEL.
   Gaylor (1992) examined 120 bioassay experiments on developmental
 toxicity.  He showed that the estimated risk of embryo lethality or mal-
 formations at the NOAEL ranged from 0 to  4.5% and exceeded 1% in
 one-fourth of the cases.  Gaylor (1992) suggested that the BD often
 would be closer to a true no-effect level than the NOAEL.
   Various dose-response models (e.g, probit and logit) can be used to
 facilitate calculation of the BD for most binary quantal-response data.
 Provided the model  gives an adequate description of the data, it need not
 have strong biological motivation because it is used almost entirely in the
 observed  experimental range.  Thus, popular mathematical models for
 carcinogenesis (e.g., one-hit and multistage)  frequently suffice.  There
 have been, however, special efforts directed  at dose-response modeling
 for developmental effects (Rai and Van Ryzin,  1985; Faustman et al.,
 1988; Chen and Kodell, 1989; Ryan, 1989). Perhaps more important
 than the model  itself is the overdispersion (greater than binomial varia-
 tion) exhibited by the data due to litter effects (Kupper et al., 1986).
   Modeling on a probability scale (to characterize risk) is not well devel-
 oped for continuous quantitative-response data, such as changes in organ
 and body  weights, clinical chemistry end points, hematology measure-
 ments, and neurotoxic effects.  Crump (1984a) suggested modeling such
 effects on their untransformed scale without a specific risk connotation.
 Application of that method to behavioral-response data was demonstrated
 by Glowa (1991). A fundamental approach to dose-response modeling
 of quantitative health effects other than cancer on a probability scale was
 illustrated by the NRC's Safe Drinking Water Committee (NRC, 1980)
 and further developed by Gaylor and Slikker (1990).  The process  in-
volves four steps. First, a  dose-response relationship for the expected
value of a given quantitative end point must  be postulated.  Second, a
statistical distribution of individual measurements about the dose-response
curve (e.g.,  normal) must be assumed.  Third, an abnormal or adverse
range of the distribution for the given end point must be defined (e.g, a

-------
.!" Mi:
66     CEELS FOR
                                                                                      SUBSTANCES
                        ,  , „,                         .....
low percentage point, say 0. i% , of the distribution in control subjects).
Finally the previous three steps are combined to calculate the probability
(risk) of an adverse effect as a function of dose.  Methods for calculating
statistical confidence limits for this methodology to define BD appropri-
ately remain to be worked out. As with the quantal-response models, the
dose-response model for quantitative responses is intended to improve the
estimation  of a BD  at the lower end of the data range and is not for
extrapolation below that point.
                                                     New Approaches
                                                          	r t   ,„   ,„      i    ri||  „   f    i I' •  • " •	1111 i1   i' •'•  » i    i
                                                        Recent research within EPA has focused on improving risk-assessment
                                                     methodology for noncarcinogenic effects. Jarabek and Hasselblad (1991)
                                                     have proposed a confidence profile method (CPM) to improve on the
                                                     RfC-RfD approach. With the CPM, a posterior probability distribution
                                                     is estimated for a concentration of chemical that either (1) produces a
                                                     defined response level of a specified quanta! effect, or (2) produces a
                                                     defined level of change in a specified quantitative effect.  Thus, the CPM
                                                     enlarges on the BD concept in providing more information on the uncer-
                                                     tainty in the data; it emphasizes  total  distribution as opposed to single
                                                     percentage points (or level of confidence), whereas  the BD is associated
                                                     with a  particular  level of confidence  (e.g., 95%).   Jarabek  and
                                                     Hasselblad  (1991)  illustrated that the  CPM could  be used to combine
                                                     effects from different experiments, even if those effects were defined
                                                     according to the two approaches  expressed in points 1  and 2  above.
                                                        Another  approach that has been developed by EPA and continues to
                                                     be refilled is categorical regression analysis (Hertzberg and Miller, 1985;
                                                     Hertzberg,  19^89; Guth et al.,  1991).  With that approach, health effects
                                                     other than cancer are assigned to  severity categories based on evaluation
                                                     of the reported information and consideration of biological and statistical
                                                     significance. The aggregate group of animals at any particular dose level
                                                     and exposure duration is classified as giving evidence of a specific sever-
                                                     ity of response.  The actual response level, if available, is used only to
                                                     make the severity determination;  it is not used in the regression analysis.
                                                       The logistic regression model is applied with the  severity code as  the
                                                     dependent variable and  the exposure concentration and duration as  the
                                                      independent variables.  Advantages of the  method are that it allows
                                                      incorporation of both quanta! and quantitative data and it enables  the

-------
                                         RISK ASSESSMENT     67

 simultaneous analysis of many studies.  Also, it provides a concentration-
 by-duration profile for any desired probability level either as a point
 estimate (Guth et al., 1991) or as a confidence limit (Clement Interna-
 tional Corp., 1991).  That capability is especially useful for acute-expo-
 sure durations, since extrapolation according to an approach such as
 Haber's law is not required.  Finally, if sufficient human data are avail-
 able to include in the regression analysis, then interspecies uncertainty
 is reduced.   One criticism of the approach has been that it sacrifices
 information on good studies by reducing the data to severity categories.
 Another criticism is that the severity rankings themselves are subjective.
 Because the method  does not use individual animal data, it cannot be
 used to analyze single experiments.  Although disaggregation of the data.
 from multiple experiments according to different target sites and mecha-
 nisms of action might be desirable, it can result in apparently poorer fits
 and possibly even failure to converge (Guth et al., 1991).  However, a
 recent refinement in the method (Clement International  Corp.,  1991) to
 allow the use of individual animal data holds promise for alleviating the
 problem.
   As. indicated above, the RfD-RfC approach  should constitute  a first
 approach in setting CEELs for noncarcinogenic effects. However, the
 more refined mathematical-model-based methods should be given consid-
 eration as secondary approaches. In particular,  BDs should be calculated
 whenever possible for comparison with NOAELs and, optionally, ,used
 as substitutes for  calculating RfDs and RfCs. Finally, the mathematical
 approaches  should  be reviewed by a  committee of experts  (such as
 COT), which would take them into consideration  in  deriving CEEL
 values.
                   CARCINOGENIC EFFECTS

                      Genotoxic Carcinogens

Extrapolation from High to Low Doses

  For carcinogenic effects that are known or assumed to be due to direct
genetic  events (e.g., mutations), thresholds are assumed not to exist.
Thus, beginning with the pioneering work of Mantel and Bryan (1961),
attempts have been made  to estimate carcinogenic risks on a precise

-------
                                     68    CEELS FOR HAZARDOUS SUBSTANCES

                                     quantitative basis all the way down to zero dose rather than to seek a
                                     positive experimental dose level associated with presumed zero risk, such
                                     as a NOAEL. Because clinical or epidemiqlogical data rarely have been
                                     sufficient to assess the carcinogenic risk to  humans  at low levels of
                                     chemical exposure, cancer risk assessment largely has involved fitting
                                     mathematical  models to animal bioassay data and extrapolating those
                                     models downward to predict risks at doses well below the experimental
liin
   The probit-log-dose model used by Mantel and Bryan (1961) was a
carryover from  the  tolerance distribution models  used for analyzing
dose-response mortality data from bioassays. After initial acceptance,
the probit-log-dose model was replaced by mathematical models derived
on the basis of presumed mechanisms of carcinogenesis.  The mechanis-
tic model most  frequently used as the underlying basis for low-dose
extrapolation is the multistage model of Armitage and Doll (1961).  A
historical perspective on the evolution of the multistage model has been
provided by Whittemdre and Keller (1978).  According to the multistage
theory,  a malignant  cancer cell develops from  a single stem cell as a
result of a number of biological events (e.g., mutations) that must occur
in a specific order.  The model  predicts that the  age-specific cancer-
incidence rate should increase  in proportion to age raised  to a power
related  to the number of  stages in the model,  and it provides a good
description of many forms of human cancer, if two to six stages are
assumed.
   When the transition between stages in the multistage model are ex-
pressed  as linear functions  of dose,  the  model  is characterized by
low^dose linearity, a property generally assumed for chemical carcino-
gens mat interact directly with genetic material. Low-dose linearity is
supported by theoretical considerations in carcinogenesis (Krewski et al.,
1989) and by the linearity of DNA binding observed at very low doses
with a number of chemical carcinogens (Lutz, 1990). Although low-dose
linearity might not be applicable in cases in which carcinogenesis occurs
by certain nongenotoxic mechanisms (e.g., as a result of tissue toxicity),
its assumption is widely made in regulatory applications of low-dose risk
assessment in the absence of clear information  to the contrary (OSTP,
 1985).  Indeed,  if the mechanism of the carcinogen whose dose response
Is being modeled is  assumed  to be the same as  the background mecha-
nism of tumor  induction  (i.e., the background is  dose-additive), then

-------
                                        RISK ASSESSMENT    69

 even for nonlinear dose-response models that do  not exhibit low-dose
 linearity from mechanistic considerations, the dose response will  ap-
 proach linearity at low doses (Crump, et al., 1976;  Hoel, 1980; Krewski
 and Van Ryzin, 1981).
   It has become common policy among regulatory agencies in the United
 States to extrapolate downward in some linear fashion from an upper
 confidence limit on excess risk (FDA, 1985; EPA, 1986).  That proce-
 dure generally is followed by COT for setting emergency and short-term
 emergency guidance levels (EEGLs and SPEGLs) for chemicals of inter-
 est to the U.S. Department of Defense (NRC,  1986b). Perhaps the best
 known procedure  is what EPA calls  the linearized multistage (LMS)
 model (EPA,  1986).  With this procedure,  an upper confidence limit,
 #!*, is obtained on the coefficient of the linear term that appears in the
 exponent of a generalized version  of the multistage model (Crump,
 1984b).  The value 
-------
I	
                                                      70     CEELS FOR HAZA&DOUS WJ3STANCES
                                                                                             :      .„            'I
                                                      and PO represents a specified "acceptable" excess risk (e.g., 10"4 to KT6),
                                                      then the corresponding dose, d^, is found by solving
                                                      Thus, PO is an upper bound on excess risk corresponding to dose d0, and
                                                      dQ represents a lower bound on dose for risk level P0.
                                                        Some chemical carcinogens are known to require metabolic activation
                                                      to exert their effects.  If metabolic activation could be characterized
                                                      adequately in terms of a suitable pfaarmacokinetic model, then the dose
                                                      delivered to the target should be used in place of the administered dose
                                                      for puiposes of dose-response modeling and low-dose extrapolation (Hoel
                                                      et al., 1983; Starr, 1990). In general, the use of delivered rather than
                                                      administered doses might be expected to lead to  more accurate predic-
                                                      tions of carcinogenic risk (NRC, 1987).  Greater accuracy  might not
                                                      occur, however, if complicated pharmacokinetic models with many pa-
                                                      rameters,  each subject to uncertainty, are used for tissue  dosimetry
                                                      (Portier and Kaplan,  1989).
                                                      Extrapolation from Long-Term
                                                      To Shprt-Term Exposures
                                                        "  •-''; .' :, : •:'   :".	»,:'. v ",.fc,: .  .  ' !	!	^	I,,",,;	;,  >;,(:• j ^ £:  '  ••
                                                        Most bioassay data suitable for low-dose extrapolation reflect continu-
                                                      ous exposure to a carcinogen, necessitating the translation of risks calcu-
                                                      lated for continuous exposures to risks associated with short-term expo-
                                                      sures.  Crump  and Howe (1984)  developed methods for applying the
                                                      multistage model to carcinogenic  risk assessment when exposure to a
                                                      carcinogen is of short duration. Their results have been adapted by COT
                                                      for setting EEGLs and SPEGLs (NRC, 1986b). Kodell et al. (1987) and
                                                      Murdoch and Krewski (1988) have studied the multistage model with
                                                      respect to accommodating age at the time of exposure. The results indi-
                                                      cate that early exposures will be of greater concern than late exposures
                                                      when an early stage  iii the carcinogenic process is dose-dependent.
                                                      Conversely, late exposures have higher risk than early exposures when
                                                      the late stage is dose-dependent.  In the context of the classical multistage
                                                      model, the potential increased excess risk from exposure for a fraction,
                                                      /, of a lifetime at a given dose rate, d, will never be more than k times
 11

-------
                                          RISK ASSESSMENT     71

 the excess risk from full lifetime exposure at dose rate^f, where k is the
 number of stages in the model (Kodell et al., 1987).
    The determination of CEELs for genotoxic carcinogens generally will
 require extrapolating risks estimated for long-term exposure conditions
 to short-term situations.  The methods proposed by Kodell et al. (1987)
 on the basis of the multistage model  can be  used  to set those CEELs.
 Suppose that data from a long-term animal bioassay or human epidemiol-
 ogy study have been used to calculate an average daily lifetime exposure
 level,  d,  of a particular  chemical, corresponding to  an  "acceptable"
 lifetime excess carcinogenic risk (e.g.,  lO"6).  Assume that d has been
 estimated  using the linearized multistage model along with any available
 pharmacokinetic information on target-tissue dose level (and interspecies
 conversion,  if necessary).
    Suppose that the carcinogen of interest affects only a single stage of
 a multistage process and that there are k stages in total.  Let / (i — 1, 2,
 ..., K) denote the index of the stage that is dose-related.  From the results
 of Kodell  et al. (1987), it can be shown that an individual exposed to
 constant daily dose D from t0 to ra  will have the same excess risk as an
 individual exposed at constant daily dose d from  birth to age t, where
and t denotes the time when excess risk is evaluated (usually the average
human lifetime).
   In the case of near-instantaneous exposure (e.g., single-day duration
or less), the total dose D at time tQ that will give equivalent risk to that
of exposure at constant daily dose d from birth to age t is
As an example of how large an adjustment might be made, consider a
three-stage process with only the first stage dose-related, and assume that

-------
                                                         r i :,:4'
                                                         -I	i
72     CEELS FOR HAZARDOUS SUBSTANCES
a 70-year lifetime exposure at dose d is collapsed to an instantaneous
(1-hr) exposure D at age 20.  Then, D = 400,624 d. For an 8-hr expo-
sure, D = 50,078 d.                                ,_.   *    -r-
  The latter expression and others like it derived on the basis of specific
multistage models can be useful for setting 1-hr and 8-hr CEELs, but
generally information will be insufficient to postulate the total number of
stages in the cancer  process and the stages that are dose-related.  In such
situations, the limiting case of the time-weighted-average dose can  be
used whereby instantaneous dose D at time f0 having equivalent lifetime
excels carcinogenic risk as daily dose d up to time t is approximated  by
     .   •  •  •  •  ......   -..  .    ,.   -                       • *,   .
As shown by Kodell etal.  (1987), the actual risk at dose D will not
exceed k times the presumed risk. In the above example, the approxima-
tion gives D =  613,200 d for a 1-hr CEEL and D = 76,650 d for an
8-hr CEEL.                                            .   .        '
   If prudence dictates conservatism, then the above approximation to L>
may be reduced by a factor  of 2 to 6 (the number of assumed stages in
the multistage model), according to the result of Kodell et al. (1987). In
determining cancer-based CEELs, extrapolation from a daily lifetime
exposure  level to near-instantaneous exposure levels  should follow this
approach  on the basis of the multistage theory of carcinogenesis, unless
there is evidence to the contrary.
                     Nongenotoxic Carcinogens

 Extrapolation from High
 To Low Doses
     • i  ;   •     ,",/:„ ii    ,	:t  f   •       '"'.   •('','   ;:	rj;»;. >  1;:
   Dissatisfaction with the linearized' multistage procedure for (perceived)
 nongenotoxic chemical carcinogens has prompted the search for an alter-
 native  method for establishing exposure levels.   Much  attention has
 focused on a stochastic model that includes only two genetic events but
 allows for tissue growth and for proliferation of cells that have experi-
 enced  the first of  these  events  (Moolgavkar  and  Venzon,  1979;
 Moolgavkar and Knudsori, 1981; Greenfield et al.,  1984).  This model

-------
                                         RISK ASSESSMENT     73

 has various descriptions:  a two-stage clonal expansion model, a stochas-
 tic birth-death-mutation model, and an initiation-promotion-progression
 model.  Initiation refers to the occurrence of the first genetic event that
 predisposes a cell to malignancy, and progression to the occurrence of
 the second genetic event that transforms the initiated cell to a malignant
 one. Promotion generally refers to the expansion of the pool of initiated
 cells.  Carcinogenic agents can be classified as initiators, promoters, or
 progressors according to the components of the model that the agents are
 presumed to affect (Moolgavkar, 1986; EPA, 1987). Without the pro-
 motional component, this model reduces to the ordinary  Armitage-Doll
 two-stage model (Armitage and Doll, 1961).  Recently, Bogen (1989) has
 investigated the two-stage clonal expansion model in the  context of a
 general class of cell-kinetic multistage models.
    Application  of the two-stage clonal expansion model  in risk assess-
 ment for a subclass of nongenotoxic carcinogens (i.e, putative tumor
 promoting agents) has been investigated on a limited basis (Thorslund
 and Charnley, 1988). When the background tumor incidence arises from
 a  mechanism that is independent of the promoting agent,  then  the
 dose-response model can be nonlinear at low doses and can have a zero
 slope at zero dose; but if additive background is assumed, models  for
 both pure initiators and pure promoters show low-dose linearity (Portier,
 1987). The ability to distinguish between initiation and promotion effects
 by using only standard bioassay data is poor, even for independent back-
 ground mechanisms; with zero background risk, the ability to distinguish
 improves (Portier, 1987).  Thus, the use of the two-stage clonal expan-
 sion model requires  a good understanding of the biological processes.
 Before the model can be recommended for routine application, its statisti-
 cal properties require further study, particularly with respect to predic-
 tions of risk at low doses.
   For nongenotoxic carcinogens that act to induce cancer by secondary
 mechanisms other than tumor promotion (e.g., hormonal  mechanisms),
no general class  of mathematical models has been proposed. It has been
suggested that thresholds do exist for certain  secondary carcinogens.  In
fact, even for certain tumor promoters (i.e., those that cause cell prolif-
eration as a result of toxic tissue injury), there might be a threshold dose
below which cellular proliferation, and hence cancer, will not be induced
(EPA,  1987).  For nongenotoxic carcinogens,  a procedure other than
linear low-dose  extrapolation can  be  used  to establish cancer-based

-------
        	  •  ' ISIill1,)'"!1!!!"'
                                                                      .',	!,. ' !J'l'l|:li, '; :flv	:,ll|lli.	>!,"',,"Hi 	,11,	i;!»||' i|
it'
IK
74     CEELS FOR HAZARDOUS'SUBSTANCES

CEELs, if known mechanisms of action so indicate.  In some cases, risk-
assessment techniques  that are  used  to  set  acceptable  levels  for
honcafcinogens may be applied (e.g., BD/SF), provided knowledge of
the mechanism of action is sufficient to provide a high degree of confi-
dence with respect to the existence of a threshold. For a putative tumor
promoter (with or without presumed threshold), the pure-promoter model
may be used from the class of initiation-promotion-progression models;
however, the importance"of knowing the underlying biological processes
when using the pure-promoter  model cannot be overemphasized.
  Like the multistage model, the two-stage model has been studied with
respect to accommodating age at  time of exposure (Chen et al., 1988;
Murdoch and Krewski, 1988).  Recall that for a fc-stage multistage mod-
el, the potential increased excess risk from exposure for a fraction,/, of
a lifetime at a given dose rate, d, will never be more than k times the
excess risk from full lifetime exposure at dose  rate fd.  With  the
two-stage clonal expansion model, however, the increased risk can be
substantial when the agent of interest greatly increases the proliferation
rate of the initiated  cell population,  thus, for a tumor promoter whose
mechanism of action is adequately described by the  two-stage clonal
expansion model, care must be taken  riot to underestimate risk when
extrapolating from long-term to short-term exposures.
   When nongenotoxic carcinogens  act through secondary mechanisms
having presumed thresholds, acceptable exposure levels are calculated for
long-term exposures, but those levels might not be useful for estimating
acceptable exposure levels for short-term exposures.

-------
 U    Development of CEELi
 CEELs were defined  in Chapter  1 as a series of concentrations that
 indicate certain toxic effects that might be observed in some members of
 the general population exposed to an agent over a specified time.  The
 exposure is assumed to result from the accidental release of the agent.
 CEELs, which are defined in Table 1, delineate concentrations above
 which exposure may lead to death or life-threatening effects (CEEL-3),
 disability requiring rescue or treatment  (CEEL-2), or discomfort that
 does not cause incapacity to take appropriate actions (CEEL-1). Below
 CEEL-1, persons may detect the presence of the agent, which may lead
 to complaints, inquiries, or anxiety, but for the most part they will not
 suffer any direct adverse effects on their health.  CEELs for 1-8 hr are
 likely to apply to  most emergency situations, but for  most situations,
 only the 1-hr CEEL will be needed because few exposures will exceed
 1 hr.
   CEELs are designed to protect almost all people in the general popula-
 tion from the toxic effects of an agent; thus, extrapolation factors (i.e.,
 safety or uncertainty factors) will be incorporated into setting CEELs.
 Although CEELs are designed to protect "sensitive"  individuals, some
 hypersusceptible individuals might not be protected,  even at  CEEL-1.
 Because emergency exposures might occur once in a person's lifetime,
the acute toxic effects of the agent are of primary importance.
  The process of setting CEELs involves issues of data collection, evalu-
ation of the adequacy of data bases, presentation and treatment of data,
use of other standards and guidelines in setting CEELs, and calculation
of CEELs on the basis of key studies chosen during the process.
  The sources and  types of data useful for setting CEELs are discussed
in Chapters 3 and 4.  It is important to emphasize that those who set
CEELs should use primary sources of data; secondary sources should be
used only if an original report cannot be obtained.  Documentation used
by other organizations to set standards and guidelines for the agent (e.g.,

                               75

-------
IMS*
ii	I,!),
                                                     76
       CEELS FOR HAZARDOUS SUBSTANCES
                                                                                                              Kit	En
TLVs  PELS, IDLHs, EEGLs, SPEGLs, and STELs) also should be
available for use.  Although those values might not be extrapolated di-
rectly to set CEELs,  the documentation provides a useful  checklist ot
studies to ensure that all pertinent data have been identified.
   Before starting the process for setting CEELs for a particular com-
pound, the practicality of setting levels and the need for them should be
considered.  Some very toxic materials do not warrant CEELs because
they do not pose the types  of hazards that are addressed  by CEELs.
Those materials might never be used or stored in significant quantities,
or they might not be  liable to sudden, massive airborne generation be-
cause of their physical properties.  Alkali hydroxides, for example, are
highly corrosive in aqueous solution, but the possibility that they or their
solutions could become airborne  in large amounts should be considered
carefully before  setting CEELs for them. Similar considerations might
be appropriate for organic liquids with very high boiling points or low
vapor pressures.
                                                                         EVALUATION OF DATA
  t'.lllll	 	ii,
                                                                             Qualitative Review

                                                        The aim of the CEEL methodology is to establish relationships be-
                                                     tween airborne concentrations of an agent and specific human health
                                                     effects   Data will be evaluated from many sources and disciplines of
                                                     varying adequacy.  The CEEL team that evaluates data should include
                                                     reports from scientific specialties encompassing the range of disciplines.
                                                     The reviewers will be required to judge the adequacy of the experimental
                                                     procedures with respect to the following:

                                                        •  Description of the methods.
                                                        •  Statistical analysis .
                                                        •  Atmospheric analysis:  estimated, nominal, or actual.
                                                        •  identification of NOELs, NOAELs, LOELs, or threshold levels.
                                                        •  Dose-response characterization with quantitative grading of respons-
                                                           •M- ' .......  . ......... ......... - •' ....... - - ' . .....  + ......... I ......... ' ...............         ..... ............ I ' ...........................
                                                     es.
                                                        •  Clearly defined elements of exposure.
                                                        •  Controls comparable with test subjects in all respects except treat-
                                                     ment Variable.
      II I
                                                                                                              !, 'Jliilill1, i

-------
                                 DEVELOPMENT OF CEELS     77

    • Relevant toxicity end points measured.
    • Conclusions justified by data in report and consistent with current
 scientific understanding.
    • Study conducted according to sound scientific practices.

    The key studies for setting CEELs will be those studies with quantita-
 tive data that can be extrapolated to provide numerical CEELs that are
 protective of human health.  Inherent  in this process is evaluating the
 weight of evidence of an agent's hazard to humans.  CEELs cannot be
 set without sufficient data from in vivo studies that provide quantitative
 dose-response information.  Thus, most in vitro studies and nonmam-
 malian in  vivo screening studies are of limited use in setting CEELs
 except for hazard identification.  The two types of studies that will pro-
 vide useful information for setting CEELs are epidemiological, clinical,
 or case studies in humans and experimental studies in mammalian  ani-
 mals.  Information on the structure-activity relationship of closely related
 chemicals  can be used to support CEELs;  however, that information
 alone is seldom sufficient for setting CEELs.
   It is obvious that human data should be emphasized in the setting of
 CEELs. Studies that show toxic effects of acute or short-term exposure
 in humans  or animals might be more relevant than long-term studies  that
 show only subtle effects.  Animal data derived from  well-conducted
 studies might be preferred over human studies that are not appropriately
 conducted.  If no acute or short-term data are available, then data from
 subchronic or chronic exposures must be used,  if available.  Although
 short-term  exposures are more likely to  give pertinent information, both
 immediate  and delayed effects of exposure should be considered. Except
 for reproductive or developmental  effects, however, delayed effects of
 short-term  exposures rarely are determined in animal studies or available
 from human experience.
   Studies  of human experience  and studies in  animals have specific
 qualities that should be considered in assessing  their relevance to  the
 setting of CEELs.
                         Human Studies

   The use of data from human experience removes one of the major
uncertainties involved in setting CEELs—i.e., the need for cross-species

-------
M'f li id ........... •" ,- }'!,.! ; ..... '!-
                               Siiii'ist;11;;1 ........ :;•
                           j.	: i    si
                                                  78     CEELS FOR HAZARDOUS SUBSTANCES

                                                  extrapolation of data.  Nevertheless, adequate data from short-term expo-
                                                  sures of humans rarely are available.  Although acute effects in humans
                                                  are sometimes observed'as a result of accidental exposure, quantitative
                                                  exposure data are seldom available:  The most useful data in  setting
                                                  CEELs are (1) the quantitative relationships between dose and response
                                                  for certain cohorts, and (2) the identification of sensitive members of the
                                                  population.  The specific requirements germane to studies of human ex-
                                                  perience have been reviewed by EPA (1990b).
                                                     Human data also might include clinical data and case reports. Clinical
                                                  studies might contain exposure-response information that can be used in
                                                  estimating health-related effects.  Individual case reports of adverse ef-
                                                 ;^~ duetto, a specific agent can  provide some help  in evaluating the
                                                  potential risk from exposure.
Si1 K'1'11,1   , ii
                                                      Criteria for Identifying
                                                      Sensitive Individuals
                                                                                       it  	I!!'*
                                                         As CEELs are designed to protect most people in the general popula-
                                                      tion,  information from human studies, including  epidemiological and
                                                      clinical studies, can help identify susceptible people so that appropriate
                                                      extrapolation factors can be incorporated into the calculation of CEELs.
                                                      The criteria recommended by EPA should be considered when examining
                                                      studies to identify sensitive or  liypersusceptible individuals  (EPA,
                                                      1990b).
                                                      Establishing Causality

                                                         Statistical methods used in epidemiological  studies cannot establish
                                                      proof of a causal relationship but can define an association with a certain
                                                      probability.  Determination of the causal significance of an association
                                                      is a matter of judgment that goes beyond any statement of statistical
                                                      probability. To assess the  causal significance of an air contaminant with
                                                      respect to a health effect, a number of criteria,  none of which is patho-
                                                      gnomonic, must be used.  The criteria have  been provided by EPA
                                                      (EPA, 1990b).

-------
                                 DEVELOPMENT OF CEELS     79

                          Animal Studies

   When, as is frequently the case, the data base lacks appropriate and
 adequate information on the effects of an agent in humans, CEELs must
 be based on experiments conducted in animals.  Even in cases in which
 apparently adequate human data are  available, the animal data base is
 useful as supportive evidence to confirm, if possible,  that the  critical
 toxicities have been identified for each CEEL level and that the numbers
 generated from the human data are not widely disparate compared with
 the animal data.  The experimental animals most often used for toxicity
 studies include rats, mice, guinea pigs, hamsters, dogs, and non-human
 primates. Although studies in animals are usually conducted under con-
 trolled exposure conditions on relatively homogeneous populations, they
 present the risk assessor with uncertainties when extrapolating dose and
 exposure regimens to the human  situation.  Some of the factors that
 influence the interpretation of animal studies  and the extrapolation of
 quantitative data to humans are discussed here.
Quality of Animal Studies

   Evaluation of the quality of toxicity studies requires consideration of
factors associated with each study's hypothesis, design, execution, analy-
sis, interpretation, and reporting. Guidelines for assessing animal studies
have been provided (Society of Toxicology, 1982; Muller et al., 1984;
NRC, 1984; James, 1985; Lu, 1985).
   An ideal animal study addresses a clearly defined hypothesis, follows
a carefully prescribed protocol, adheres to good laboratory practices, and
includes appropriate analysis to support its conclusions.  EPA's Good
Laboratory Practice Standards (EPA, 1983a,b) are designed to ensure the
integrity of data used in hazard evaluation. Studies that do not precisely
follow those guidelines might still be judged adequate if it is determined
that the  deviations are not important.  The types of deviation, their mag-
nitude,  and their potential interaction with other variables must be as-
sessed (NRC,  1984).  For example, a study might be judged adequate
despite an insufficient number of test animals as specified by the appro-
priate reference protocol guidelines if the results are so definitive that the

-------
80     CEELS FOR HAZARDOUS SUBSTANCES

addition of more test animals almost certainly would not affect the con-
                                                       	 (	'I	.ili'l!'	„ » • ' n.v.Vt
elusion.
  U&iUlJt.                                         .     T  J U  TTT) A
  Ideally, an acceptable study should meet the criteria outlined by EPA,
which fundamentally represent good scientific practice (EPA,
Appropriateness of Species
As a Model for Humans

   Selection of an animal model and key study for each CEEL level
depends on the depth of understanding of the human disease syndrome,
adverse effect, or indicator of toxicity selected as the criterion for evalua-
tion  Even though a particular animal species shares a number of simi-
larities  with humans in respiratory-tract physiology, dissimilarity in
crucial factors make it inadequate as a model. This subject area has been
reviewed (Hakkinen and Witschi, 1985) and various mammalian species
(rat hamster, and rabbit) have been identified as appropriate species tor
extrapolation from several perspectives.  Other reviews that discuss the
current limitations and need for the development of animal models as
surrogates for humans inclule thBse of Reid (1980), Slauson and Hahn
 1980), ^ Calabrbse  (1983)_                        .
   For agents with toxicological outcomes that depend on degree to which
they are metabolized, selection of the most appropriate animal species is
contingent on proper evaluation of the numerous interspecies differences
with respect to metabolism.  The studies of Plopper et al. (1980) suggest
that animal species  differ widely in the metabolizing potential  of the
respiratory tract. The metabolizing potential is much greater in hamsters
and rabbits than in monkeys and rats.  Interspecies differences in meta-
bolic pathway might serve as  a  basis for selecting one study for  CEEL
derivation and rejecting another.                                  .
   Appropriate animal  model selection might be contingent on pathologi-
cal Identification of early changes in a species that are consistent with the
human syndrome; for  example,  a clear choice of an appropriate  animal
species has not been established for emphysema (Snider et al.,  1986).
People with deficient antitrypsin blood concentrations because of a genet-
ic defect are characterized as  a high-risk subgroup for  emphysema.
Hamsters  which have low antiprotease concentrations, might represent
a good choice, but primates, which have antitrypsin profiles comparable
to humans, might offer a better model.
                                                        '..' "I!*

-------
                                 DEVELOPMENT OF CEELS      81

    Species-dependent variables in mucous production and secretion also
 are factors to be considered in selecting an appropriate animal model for
 inhalation assays.
 Establishing Relevance
 Of Animal Studies

    Identification of the most appropriate study (key study) for each toxici-
 ty end point that will be used to set a CEEL is the end result of an inter-
 pretive process that examines all aspects of each study in the data base
 from study design, performance, and interpretation to relevance to the
 human situation.  Although selecting the most sensitive species might
 have the advantage of affording the greatest degree of protection, it could
 lead to undue alarm  and  unnecessary evacuation of a population.  The
 species most sensitive to an agent might not be as toxicologically relevant
 as other  species for extrapolation  to humans;  therefore,  all available
 information on the agent  must be considered in choosing the key study
 for each toxic end point.
   Determination of validity and relevance involves evaluating a number
 of factors, including all elements of exposure definition (dose, duration,
 administration route, and  physical and chemical characterization of the
 chemical), relevance of the dose level tested to the anticipated human
 exposure level, nature of the effect, and similarities and differences
 between the test species and humans (e.g., deposition, absorption, reten-
 tion, clearance, and metabolism).
   Animal studies are conducted with a variety of exposure scenarios, in
 which the magnitude, frequency, and duration of exposure might vary
 considerably.  Studies might use different durations (acute, subchronic,
 and chronic)  and schedules (single,  intermittent, and  continuous) of
 dosing. All the studies contribute to the hazard identification portion of
 the risk assessment.  Special consideration should  be  given to those
 studies in which exposure duration is appropriate for the CEEL to be
 determined.
   Exposure concerns (dose and duration) are compounded when the risk
 assessor is presented with  data from several animal studies.  An attempt
to identify the animal model  most relevant to humans  should be made on
the most defensible biological rational (e.g., morphology,  comparable
metabolism, and pharmacokinetic profiles).  In the  absence of  such a

-------
                         ""il!"1!" "    ,:!!;!
ill11  I»"! '" *   •  »'"
                                                   82     CEELS FOR HAZARDOUS SUBSTANCES
                                                   model the most sensitive species (i.e., the species showing a toxic effect
                                                   at the lowest administered dose) should be used.  The selection process
                                                   is more difficult when the animal data are for various exposure routes,
                                                   esfieciaily when the routes are different from that in the human situation
                                                   bfconcern. For example, if data are insufficient for the inhalation route,
                                                   CEEL reviewers might have to use toxicity information obtained from a
                                                   less relevant  exposure  (unless convincing contrary evidence exists).
                                                   However, consideration  must be given to the differences in pharmacoki-
                                                   netics resulting from  different  exposure  routes  of the chemical.
                                                   Bioavailability of the chemical administered is another important factor
                                                   for  consideration  in the evaluation of dose.  Extrapolation from the
                                                   noninhalation route might be inappropriate when acute high-level expo-
                                                   sure primarily affects the pulmonary system.
                                                                                      '" di*:"^',;<•;>	if-.V ,':- '•' .••• .•••.:•{"	•'. ,'jy
                                                                         Appropriateness of Data
                                                        • "      '"  ' \;''* ;i:;'"'i": !fpV' Extrapolation

                                                      Several factors will contribute to the evaluation of the complete data
                                                    base as supporting the hazard assessment and the choice of the key stud-
                                                    ies for quantitative extrapolation.  Those factors include the consistency
                                                    of findings among studies in humans or animals and the biological plausi-
                                                    bility of the associations.
                                                    Consistency of Findings Among
                                                    Studies in Animals and Humans

                                                      The strength of an association between exposure and response is in-
                                                    creased if the specific response to an exposure is seen as follows:

                                                      • In a single species having different dosing regimens.
                                                      • In a number of tests in a single species.
                                                      • In a single species having different routes of exposure.
                                                      • In different species of animals.
                                                      • In studies of animals and humans.
                                                      • In different epidemiblogical studies of humans.
                                                                             ."111"..
                                                                                                               	I!,!;,  •!„

-------
                                DEVELOPMENT OF CEELS     83

   In general, the strength of the association increases as one goes down
 the list.
 Biological Plausibility

   Biological plausibility is derived from an examination of the available
 data. The following increase confidence in the association between the
 data and the effect:

   • Dose-response gradient is observed in epidemiological or animal
 tests.
   • Short-term in vitro or in vivo tests indicate plausible mechanism of
 action.
   • Biotransformation and pharmacokinetic studies indicate constancy
 of mechanism across species or show that mechanisms are of types gen-
 erally found among all relevant species.
   • Pharmacodynamic studies or organ pathology indicates constancy
 of target organs and toxic mechanisms across species.
   • Structure-activity relationships suggest constancy of action for simi-
 lar structures.
           Consideration of Dose-Response Relationship

   The primary purpose of the evaluation of experimental and epidemio-
logical evidence is to provide quantitative data that can be extrapolated
to humans exposed to the agent under the hypothetical conditions of an
accidental release  of the agent.   The data needed to define the three
CEEL levels ideally would show a gradient of effects  with increasing
dose. The primary objective in the case of threshold toxicity is to identi-
fy NOELs, NOAELs,  LOELs, and frank-effect levels (FELs) for the
most sensitive end  points and to set CEELs on the basis of those values.
The most useful studies are experimental animal or human studies that
have employed a number of dose levels, have  determined a  level at
which there is no effect, and have characterized the toxic responses at
several doses above NOEL. For epidemiological studies, the best studies

-------
                              I1IH1II' •
                              en <\
                                                                                   './'iiiiiii'ioi;"1".;!!!,;.]
pi 4r' / :  ' ll
I", +"iil" If"1  1' I  III1,.
84     CEELS FOR HAZARDOUS SUBSTANCES

have sufficient exposure monitoring data and a large enough cohort to
grade responses over a range of exposure levels. It should be noted,
however, that the procedure for setting CEELs  for threshold toxicity
might involve mathematical modeling rather than use of NOAELs or
LOAELs (see discussion of benchmark dose below).
  Also of importance in  setting CEELs is the question of whether the
severity of a toxic response is related to the cumulative dose—i.e., the
response is  a function of concentration and time—or to the absolute
exposure concentration—i.e., the response is independent of time.   One
relationship  that describes the time dependence of severity of response
is Haber's law, which states simply that the severity  of effect varies
directly as the product of concentration times exposure time.  That is, of
course, most often an oversimplification of the true relationship, but if
some information can be obtained on the relationship between severity
of response and concentration and time, that information should be incor-
porated into the evaluation of the data.
 MI'1,1.
i. ffVS"! 'i 1
   : . -. '. ' ;, i- i  .•,' •  ' • r - .:'' » PkESENf ATl<5N "AND '
   ;•  .• ,;  "       '   TREATMENT OF DATA          ............ ' _' ........
    ;. • '''.         |          1 1                 '' ii iV . ,  ' ,;' "'"I* i ij'iwi1'- K','!i i
   Data derived from literature review should be summarized to expedite
assimilation of the material by those involved in setting CEELs.  The
following types of data should be presented:
                                                                   • t". '
                                                           i •  ii    i ii.   i    ni  illl ........ • '  ,,,,,, ,j ,     i      ^    ^  , ,    i ii
                                                       • Human experience:  epidemiology and case histories.
                                                       • Animal toxicology .
                                                       • Toxicologicai properties of structurally related compounds.
                                                       • Current exposure standards and guidelines.
                                                       Checklists for 'the presentation of data from human epidemiological
                                                     and case-history studies and  from animal experiments are presented
                                                     below.
                                                                         Human Experience Data
                                                              ,i' '' I,*!     M   '!'  .' • ..  fiam  ,•   ii .if ', -  •«,   ,!• i     	   • ,111 , n 	 '

                                                       Data should include information on groups who might be more suscep-

-------
                                 DEVELOPMENT OF CEELS     85

 tible to chemical exposure—e.g., the elderly, the young, the pregnant,
 and those with minor illnesses compatible with participation in normal
 daily activities.  In addition, subjective response information should be
 included, such as vapor concentrations that caused eye or respiratory-
 tract discomfort or produced objectionable odors. The following infor-
 mation should be presented:

   •  Odor data (include odor threshold and odor description).
   • Toxicity data (include data on humans who were exposed to known
 concentrations).
   • Clinical-, workplace-, and accidental-exposure experience.
   • Epidemiological data.
                           Animal Data

   Each animal study should be summarized and critically reviewed with
respect to adequacy and relevance.   Duration, route, species, sex, and
effects at each dose should be given, and NOELs, NOAELs, and LOELs
should be reported for important effects. Negative and positive findings
should be included. Lengthy detail is not required for studies judged to
be of minimal relevance for establishing CEELs. The following types
of animal studies should be reviewed:

   • Acute toxicity (include studies of less than 5-day exposures),  e.g.,
oral LD50s and eye  and skin irritation (provide exposure duration for all
LC50s; state whether nominal or actual concentration; include pathology
if available;  give lowest lethal concentrations and highest nonlethal con-
centrations if available).
   • Subacute toxicity (include 5- to  15-day exposure studies).
   * Subchronic toxicity (include 15-day to 6-month exposure studies).
   • Chronic toxicity and  carcinogenicity (include studies of exposures
of greater than 6 months).  Provide regulatory classifications based on
chronic effects (e.g., EPA and OSHA) and other agency classifications
(e.g., IARC  and ACGIH).  Mathematical models adopted by NRC might
be useful in  assessing the risk of developing cancer from a single expo-
sure (see Chapter 5).
   •  Reproductive and developmental toxicity (include  developmental

-------
86     CEELS FOR HAZARDOUS SUBSTANCES

toxicity and reproductive studies).  Provide regulatory and other classifi-
cations based on those effects (e.g., EPA).
  •  Genotoxicity (include short-term In vivo or in vitro studies).
  •  Metabolism and pharmacokinetics.
               Other Air Standards and Guidelines

   As previously noted, the use of standards and guidelines recommended
by other governmental and private-sector organizations to set CEELs is
not recomniended. Nevertheless, the values published by these organiza-
tions along with the rationale for the limit set should be summarized and
presented to provide some indication of how those organizations consid-
ered the data and to assure that the CEELs set are not widely divergent
from the existing values when proper adjustment is made for differing
contexts: Some of the standards arid guidelines are presented in Table
3.
                  TRANSLATION OF ANIMAL
                       DATA TO HUMANS

   Interpretation of data derived from animal experiments and their ex-
trapolation to humans require experienced scientific judgment in a variety
of disciplines.   The evaluation should consider the conditions under
which the data were obtained  and,  in particular, their relevance to the
conditions of human exposure.  In addition, the evaluation should consid-
er the similarity of the test species to humans with respect to target or-
gans, metabolism, and disposition of the agent.
   Development of CEELs requires that  animal data be extrapolated
quantitatively to humans. As noted earlier, the lexicological, biological,
and pharmacokinetic characteristics of the species most representative of
humans should be used to set CEELs if adequate data on that species are
available. If data on that species are not available or if the most appro-
priate species cannot be identified from the data, it is prudent to use data
from the most sensitive species to set the appropriate limits.

-------
87





;Ub ana uuiuelmes
1
§
"3
'•5
M
^
"•i
J
o
2
0
-t
1
I
H
1
8,§
fie!



Definition
|
'I
E?
O


J
>

II
W 0
*• — — Ui W
"C u •§ •§ fi •*« "**
5^5 ^o"30^ c^o
«— ' »H oo oo^Hoocn<— ip±!'-S
l?ifill«ll'-?l>lrll|fi&.sl
§-S £ 2>3,u-vE6g-lrgrt.t;t-i§3'S w a — 5
§|||f!S~ir£!!l|^.|sS^:llt§
S*S i.^o-2
-»-§ £-£3'2te'2c8<2e8 1 1 S § « S S S«
|i!INIfIt|f|li|p!H
H4>HmJ«SOO
MeuW JWggHSWW
BJ,CO&H H PS; :> ti eo ow u

2^1 IIIIIHI
I II 1 1 1 1 1 ll 1

-------
:' ,,tt
                                                                   in';
ss     CEELS FOR mz^ous SUBSTANCES

                   THRESHOLD TOXICITY
                        ii   i   i   1 1   i
                1       '  'I  V i     II     i
                       Use of Uncertainty
                     And Modifying Factors
                           ill II     II        'i ...... !/." •.   nL ini'i  f, ,); 1,1 ill 'ill,!' '' i",!!!';/1
  When appropriate for threshold toxicity (usually noncarcinogens and
nongenotoxic carcinogens),  uncertainty factors should be  applied to
NOAELs (or NOELs, LOELs, LOAELs, FELs) or to benchmark doses
if they are available.  In the application of uncertainty factors, a factor
must be incorporated that approximates the likely range of susceptibilities
among humans.  People at increased risk include those at either extreme
of age those with poor nutritional status, those with preexisting diseases
such as certain heart diseases, that are fairly widespread in the general
population, those with enlancid hereditary susceptibility, or those who
are Overexposed because of unusual physical exertion.  Nevertheless, the
intent of CEELs is not necessarily to provide absolute assurance that
everyone at risk will be protected under all  circumstances, and thus the
uncertainty factors should be chosen  with the understanding that a few
hypersusceptible persons might not be protected. In the case of CEEL-Z
uncertainty factors must be balanced against the inherent risk associated
with" actions, such as "evacuation, that might be taken as a result of appli-
cation of (JEELs; Large uncertainty factors, which might be appropriate
with chronic exposure limits, such as PELs, might be associated with
increased risk to the community in the application of CEEL-2.
   In addition, the CEEL is derived from the NOAEL through the appli-
cation of uncertainty factors that reflect various specific recognized un-
certainties in extrapolating from the animal  or human studies. An addi-
tional modifying factor also is applied that reflects professional judgment
of m"e entire data base available on the specific agent.
   The CEEL can be derived from the NOAEL (or other effect level) as
follows :   ........... „ ...... ' ...... \ ..........................

 .    :': . 'i' ; • ' ' ..-•'   ." ;; CEEL" ="NOAEL/(UF x  MF),  ' ' "_'   _  ^ '' ^ ' '"
                                    •         ,             j •; ...........
 where UF is the product of uncertainty factors described below, and MF
 is the modifying factor.                              ,
    Generally, a  10-fold UF will be applied when extrapolating from valid
 experimental' results of studies involving appropriate exposures to aver-

-------
                                 DEVELOPMENT OF CEELS     89

 age healthy humans or to experimental animals.  That factor is intended
 to account for the variation in sensitivity among the human population.
 If data are available on a sensitive subpopulation, the factor might be as
 low as 1.  .
    An additional 10-fold UF generally is applied when extrapolating from
 valid results of studies on experimental animals because results of studies
 of human  exposure are not available or are inadequate.  That  factor is
 intended to account for the uncertainty in extrapolating animal data to
 humans.
    An additional 10-fold UF may be introduced when deriving a CEEL-1
 from a LOAEL instead of a  NOAEL or a CEEL-2 or CEEL-3  from
 appropriate LOAELs or FELs. That factor is intended to account for the
 uncertainty in extrapolating from a LOAEL to a NOAEL and so forth.
 An intermediate factor may be applied when the data are very strong and
 relevant animal-to-human extrapolation factors have been derived, when
 limited data are available for a sensitive human subpopulation, and when
 dosimetric adjustments have been made on the basis of species consider-
 ations. The intermediate factor is usually 3—i.e., the geometric  mean of
 1  and  10, rounded to one significant figure.
   Professional judgment must be used to determine another UF,  the MF,
 which is more  than 0 and less than 10.  The magnitude of the MF de-
 pends on the professional assessment of the scientific uncertainties of the
 study and data  base that are not  explicitly treated by the UF (e.g., the
 completeness of the overall data base and the number of species tested).
   In selecting UFs for deriving CEELs, it is important to recognize that
 the intent is to avoid unnecessary  conservatism that might result in expo-
 sure levels with little or no biological plausibility.  An example of that
 would be the application of several levels of uncertainty to a concentra-
 tion of an agent that produced  irritation in an appropriate and adequate
 animal inhalation study.   Although it is desirable to consider the uncer-
 tainty around varying sensitivities and extrapolations, it is not practical
 to establish an  inflexible system  of UFs that simply become multipli-
 cative in their application. Therefore, UFs must be determined  case by
 case.  Of course, the determinations would be associated closely with the
quantity and quality of the data and the end points in question.
   It has long been recognized,  however, that there are two major prob-
lems with the use of NOAELs  (or LOAELs, etc.) in deriving reference
concentrations (RfCs) (and derivative values such as CEELs). One prob-

-------
                                                                                 "I"	I"	"
             'I11. '•• -    ;  "ii   .'[.ni i'. .• -;,    '  '      •   •:" i ?•;•?.   • j           i  ii   i       I                       lii
                                                                                                         i  i

                                                90    CEELS FOR HAZARDOUS SUBSTANCES

                                                iem is that the use of NOAEL's does not readily account for the number
                                                of ammate used to  determine' the NOAEL.  It might be useftil if the
                                                number of animus used to determine the NOAEL would affect the value
                                                of the resulting RfC in some way other than through the level of confi-
                                                dence (as reflected in the MF).  A second major problem with the use of
                                                NOAELs is that the slope of the dose-response curve for the critical toxic
                                                effect is generally ignored in the estimation of the RfC.  Many scientists
                                                have argued that the slope should directly affect the resulting RfC; steep
                                                '"effect' slopes would presumably yield higher values.   Thus, in some
                                                cas&s; the use of some other procedure for quantitative risk assessment
                                                for noncancer end points might have greater validity.
                                                   The current principal alternative to  NOAELs is the benchmark dose
                                                (BD) procedure,  the BD has  certain advantages over the NOAEL ap-
                                                proach   The BD makes appropriate use of the sample size, as reflected
                                                in me magnitude of the confidence limit. It exploits the shape (steepness)
                                                of the dose-response curve in the experimental range but does not depend
                                                strongly on the particular mathematical model used, because the model
                                                 is not followed below me 1% response level.  EPA's  "Guidelines for
                                                Developmental Toxicity Risk Assessment" (EPA, 1991) recommends that
                                                 a BD be calculated  to supplement the NOAEL or LOAEL used in deter-
                                                 mining a reference  dose for developmental toxicity.
                                                   EPA considers the BD to be useful for comparison with RfC results
                                                 or as a possible alternative to NdAELor LOAEL.  The use of the BD
                                                 in the risk assessment of noncarcinogenic toxic effects should be encour-
                                          1 '        ", „ ''^	'" '      ''  	'	'''"   ''   ''!  I	  	     	1|""	
III   I  II ill I    ,  5" ",    ••   ' ,'ili!   •:•!	

   I   II     .  j'l	I   ., ,    ',!!  ; SB
   One unresolved issue with the BD procedure, however, is the size of
the UF that should be applied to any particular BD to account for varia-
tion in sensitivity among members of a population. For example, would
applying a 10-fold UF to the BD01 be equivalent to applying a 10-fold
UF to a well-defined NOAEL, or would it be more or less protective?
That question will be answered only after more experience is gained with
the BP procedure.


                   Cross-Species Extrapolation:
1  :;'•,';.;••  ,'^::r:"' 	   Specific Dose Adjustment
                    •,'.:"       ' I         "   '           i
   The extrapolation of quantitative  animal data to humans, whether for

-------
                                 DEVELOPMENT OF CEELS     91

 threshold toxicity or for assumed nonthreshold toxicity (e.g., genotoxic
 carcinogens) requires cross-species extrapolation.  Traditionally, cross-
 species extrapolation has been based on body weight (e.g., milligram per
 kilogram of body weight in animals is equivalent to milligram per  kilo-
 gram of body weight in humans). The U.S. Food and Drug Administra-
 tion has used this formula until the present time.  Many physiological
 factors (e.g., consumption of water, food, or oxygen), however, vary as
 a fractional power of the body weight.  Dourson and Stara (1983) in a
 review of the data pointed out that the toxicity of many compounds ap-
 pears to be more closely related to body-surface area than to body weight
 (Pinkel, 1958; Freireich et al., 1966).  Because body-surface area varies
 approximately in proportion to the 2/3  power of body weight (as it would
 for spherical bodies  of equal densities),  that value has often been used.
 Extrapolation on the basis of concentration of an agent in a medium
 (e.g., in diet or air) generally yields a factor that falls between one calcu-
 lated on the basis of body weight and  one calculated on the basis of the
 2/3 rule.  After examining the data of Freireich et al. (1966), Travis and
 White (1988) suggested that body weight to the 3/4 power might be more
 appropriate for interspecies dose conversion, but other values between
 0.67 and 0.75 have  also been suggested. Recently, three federal agen-
 cies proposed the use of body  weight  to the 3/4 power as  a default
 interspecies dose scaling factor for carcinogens (EPA, 1992).  In some
 instances, physiologically based pharmacokinetic models can be used for
 cross-species extrapolation. Although  such models or the data needed to
 create the models usually are available only for well-studied genotoxic
 and nongenotoxic carcinogens, use of the models should be considered
 whenever they are available for either threshold or nonthreshold toxicity.
 If cross-species extrapolation is  made on the basis of pharmacokinetic
 data in humans and the experimental animal, the cross-species extrapola-
 tion factor (maximum value =10)  may be adjusted appropriately.
            Cross-Species Extrapolation:  Inhaled Dose
             In Animals to Inhaled Dose in Humans

  The effect that an agent has  on an animal or human might be a func-
tion of concentration alone or  a function of concentration and time (or
something in between).  If the effect is directly on the surfaces of the

-------

                                                    92     CEELS FOR HAZARDOUS SUBSTANCES

                                                    lung then the effect might well depend on concentration, although the
                                                    effect also will  depend on the breaming rate and the geometry of the
                                                    airways and lungs of each species.  Ways to approach these situations
                                                    have been proposed by EPA. For systemic toxicants, on the other hand,
                                                    the important consideration is the dose of the active compound (possibly
                                                    a metabolite) to the site of action.  It is assumed that the action of the
                                                    compduhd is a function of the specific dose in the body.
                                                       If data are sufficient, the percentage uptake of a chemical through the
                                                    lungs for the animal species and humans can be factored into the extrapo-
                                                    lation, but, in general,  100% absorption through the  lungs is assumed.
                                                    In the absence of sufficient data, the amounts absorbed are assumed  to
                                                    be determined by the respiratory rates of the subjects and the time  of
                                                    exposure.  The rates are available from various sources, although actual
                                                    data on the animals in question should be used if available.  For exam-
                                                    ple,  the accepted |^re!a^fng rate for a resting human is usually 7-10
                                                    L/min,  whereas the value for a 0.113-kg rat is about 73  mL/mm.  The
                                                    conversion of data in rats, expressed as concentration in  milligrams per
                                                    cubic meter, to humans can be calculated on the basis of the total volume
                                                    breathed by the two species over the relevant time as follows:
                                                     •. 1J•'•':! iv!! '. i ,  '     • i  .' 	:',n~ >;,.,	';'   -it:,' I"1""	   "  	• ' •   ,  • . •,  	i	i	 •	.? -
                                                     !  .•"•/! . i' •• ••;   ;       , .;. i "i-ji" •;;!,?,,;.	    LI,: [_"_  : ,  .w i 'f . •:    -,n  ,  - .. ItW"1 ail,; ,,-
                                                    111    ,	:":         •.':'J."1;?>ii xbAxBWH •'
                                                                                         BWA
                                                                                                              '1! il'i'i
••	i! 3
                                                    where C is exposure concentration in milligrams per cubic meter, R is
                                                    respiratory rate in cubic meters per unit of time, BW is body weight, H
                                                    is human, and A is animal.  Note that the interspecies conversion is a
                                                    one-to-one unit weight conversion.
                                                               Cross-Species and Dose-Route Extrapolation:
                                                             Oral Dose in Animals to Inhaled Dose in Humans

                                                       In some instances, the only data available for cross-species extrapola-
                                                    tion is that obtained from oral doses in animals.  In that case, the extrap-
                                                    olation should be tjie same as that for inhalation extrapolation, except
                                                    that conversion from the air concentration for the animal will not have
                                                    to be made.  As before, if toxicokinetic  information on uptake of the
                                                                                     •  ''i " •      '	!	'	
	ii), i  : j: ,! >

-------
                                DEVELOPMENT OF CEELS     93

 compound of interest (or a similar compound) is available for humans,
 the value should be corrected for the absorption ratio.
                   NONTHRESHCMLD EFFECTS

   Decisions on the carcinogenicity of chemicals in humans should be
 based on considerations of all relevant data, whether they are indicative
 of a positive or negative response, and should embody sound biological
 and statistical principles.  Because animal  carcinogens are not the same
 with respect to potency, target organs, mechanism, and so forth, and thus
 are not equally relevant to humans, hazard  evaluation should be made on
 a weight-of-evidence basis that includes  consideration of  all  relevant
 factors, giving appropriate weight to each factor on a case-by-case basis.
 The weight-of-evidence  evaluation of carcinogenic hazard to humans
 provides the basis for carcinogen classification.  In  addition,  it is the
 initial step in determining the adequacy and appropriateness of the data
 for quantitative cancer risk assessment.
   Data addressing the strength of the conclusions to be drawn from
 animal bioassays and data addressing the appropriateness of the animal
 model as a predictor for humans should include information on compara-
 tive metabolic paths, pharmacokinetics, routes of exposure, mechanisms
 of action, and organ or species differences  in response, as well as infor-
 mation on humans.  If pharmacokinetic models for calculating delivered
 dose  and  cross-species  extrapolation  have  been  developed,  the
 pharmacokinetic information should be incorporated into the  quantitative
 risk estimates.
   When an agent under evaluation is determined to present a potential
 carcinogenic  risk to humans, a separate quantitative risk assessment is
undertaken in recognition of the fact that.even limited  exposure to such
agents theoretically can increase the risk of cancer.  Calculating CEELs
for chemical  carcinogens is a complex process.  Many carcinogenicity
studies involve high-dose, long-duration exposures to compensate for the
small number of animals used  in the studies.  Data on short-term or
single exposures that assess  carcinogenicity in animals or humans are
virtually nonexistent for agents  of interest  to organizations that will set
CEELs (NRC, 1986b).
   In the absence of human data, usually it is assumed that carcinogenic

-------
                                                                              • si' , A:    • r!'. i  •
 (if	• • i
                                                                 V. '  ill1.!1'!; ;'f(V '«,, '"  "i" Sr"!	
                                                    94     CEELS FOR HAZARDOUS SUBSTANCES

                                                    risk derived from animal data is directly and quantitatively applicable to
                                                    humans.  Extrapolation from high-dose animal exposures to low-dose
                                                    human exposures is often required.  For known genotoxic carcinogens
                                                    and for carcinogens with mechanisms of action that are not well under-
                                                    stood^ setting CEELs should involve linear low-dose extrapolation from
                                                    an upper confidence limit on  excess risk.   Implementation might be
                                                    accomplished with a computer program such as GLOBAL82 (Howe and
                                                    Crump, 1982), whether one employs the basic procedure  of Crump
                                                    (1984b) or Gaylor and Kodell (1980).   In  the equation below, if ft
                                                    denotes a computed upper confidence limit on the excess-risk low-dose
                                                    slope for a specified level of confidence (e.g., 95%) and P0 represents
                                                    a specified "acceptable" excess risk (e.g., 10"4 to lO"6), then the corre-
                                                    sponding dose, d0,  is found by solving
                                                    Thus, PQ is an upper bound on excess risk corresponding to dose d0, and
                                                    d0 represents a lower bound on dose for risk level P0. See Chapter 5 for
                                                    greater details.
                                                       Most bioassay data suitable for low-dose extrapol ation reflect continu-
                                                    ous exposure to a carcinogen, necessitating the translation of risks calcu-
                                                    lated for continuous exposures to risks associated with short-term expo-
                                                    sures.  Crump arid Howe (1984) developed a method for applying the
                                                    multistage "model to carcinogenic risk assessment when exposure to a
                                                    carcinogen is of short duration. Their results have been adapted by COT
                                                    for setting  EEGLs and SPEGLs (NRC,  1986b).  The determination of
                                                    CEELs for carcinogens generally will require extrapolating risks estimat-
                                                    ed for long-term exposure conditions to short-term exposure situations.
                                                       In the case of near-instantaneous exposure (e.g., single-day duration
                                                    or less), total dose D "at "time "fp ''that will give equivalent risk to that of
                                                    exposure at constant daily dose d from birth to  age t is
                                                         D  =
                                                                                     dt*

                                                       Expressions derived on the basis of specific multistage models can be
t/iiiin	i	, i , ,i .v mil:,,,  
-------
                                DEVELOPMENT OF CEELS     95

 useful for setting 1-hr and 8-hr CEELs, but generally information will
 be insufficient to postulate the total number of stages in the cancer pro-
 cess and the stages that are dose-related. In such situations, the limiting
 case of the  time-weighted-average dose can be used, whereby instanta-
 neous dose  D at time f0 having equivalent lifetime excess carcinogenic
 risk as daily dose d up to time t is approximated by


                             D = dt.

 As shown by Kodell et al. (1987),  the actual risk at  dose D will not
 exceed k times the presumed risk, where k is the number of stages hi the
 multistage model.
   If prudence dictates conservatism, then the above approximation to D
 may be reduced by a factor of 2  to 6 (the number  of assumed stages in
 the multistage model), according to Kodell et al.  (1987). In determining
 cancer-based CEELs, extrapolation from a daily lifetime exposure level
 to near-instantaneous exposure levels should follow that approach based
 on the multistage theory of carcinogenesis, unless there is evidence to the
 contrary.  See Chapter 5 for greater detail on risk-assessment techniques
 for carcinogens and noncarcinogens.
                  CALCULATION OF CEELS
                       AND RATIONALE

   CEELs are calculated on the basis of the principles embodied in the
previous discussions. The steps are outlined here. It is stressed, howev-
er, that sound scientific principles should always guide the calculation of
CEELs.  Thus, if some other method for identifying levels of concern,
for example, the benchmark dose, can be substituted for NOELs and
LOAELs with scientific probity, then the alternative method should be
used.  In addition,  agents demonstrating nonthreshold toxicity should
undergo evaluation using both threshold and nonthreshold methods; the
value most protective of human health for any  particular CEEL level
should be chosen as the CEEL.
   The following list briefly outlines one rational approach to the calcula-
tion of CEELs; the procedure is diagrammed in Figure 1.

-------
96    CEELS FOR HAZARDOUS SUBSTANCES

ASh |
• 4-
Select
Physiological
Reference
Values
4-
Select Risk
Level
	 4r
Calculate
Acceptable
Total Dose


.: "" Select I
Averaging 1
Period 1

|
Review Overall
Data Base 1

4
Select Key I
Studies 1


^^V No Identify
reshold\^ 	 ^ NOAEL or
ects ^ LOAEL
^r |
Determine
Absorption
Contribution
™"""^""""~™^
Select
Physiological
Reference
Values

Select
Extrapolation
Procedure
T
	 	 Select 1
Uncertainty 1
1 Factors |
Calculate
Recommended
CEELs Based
on Threshold

Select CEELs Most L
Protective of
Human Healthr
4- •

Present All Assumptions I
and Uncertainties |
i
r
Compare CEELs to Other 1
Exposure Levels |
"" !;' . . !,!: '"'
r
Present CEELs 1
          FIGURE 1
process.

-------
                                DEVELOPMENT OF CEELS     97

   1.   Select  exposure averaging time, e.g.,  1 hr.   CEELs may be
 established for more than one period, e.g., 1 hr and 8 hr.
   2.   Review overall data base.
   3.   Select the key studies.   Identify the most relevant or sensitive
 species and target organ for each limiting toxicity and each time period.
   4.   Select most appropriate no-effect or threshold levels for each
 limiting toxicity.
   5.   Determine contribution of absorption data to dose calculation.
   6.   Select physiological and anthropometric reference values, e.g.,
 body weight, surface area, minute volume, and volume of inspired air,
   7.   Select extrapolation procedures.
   8.   For agents that are considered to produce threshold effects, select
 an appropriate uncertainty factor of 1 or greater for end points affording
 NOELs, NOAELs, LOELs, or  thresholds.
   9.   Calculate the  recommended CEELs on the basis of threshold
 toxicities.
   10.  For agents that are considered to produce nonthreshold effects
 (e.g., genotoxic carcinogens), select an appropriate model, appropriate
 physiological reference values,  and an appropriate risk  level ranging
 from 10"4 to 10"6 depending on the population potentially exposed.  Cal-
 culate the acceptable dose levels (1 and 8 hr, etc.). Note that for materi-
 als that might cause threshold and nonthreshold effects, both approaches
 should be used to determine the CEELs.
   11.   Select CEELs  most protective of human health  based on the
 overall data base.
   12.   Present and discuss all  assumptions and uncertainties used to
 derive CEELs.
   13.   Compare CEELs with other recommended exposure levels for the
 same potentially exposed individuals  in the general population or for
 different groups, such as workers.
   14.  Present CEELs, provide documentation,  and  briefly discuss
rationale.
                  PRESENTATION OF CEELS

  The procedure used to derive CEELs should be fully documented. As
well as  presenting all the calculations, the rationale for selecting the

-------
                                                           it. f
98     CEEIS FORtfAZARtiOUS SUBSTANCES

relevant studies, end points, and no-effect or threshold levels should be
discussed.
  Furthermore, because risk assessments are based on the use ot many
assumptions to bridge gaps in understanding and data, the major assump-
tions, uncertainties, and their rationale are important and should be pro-
vided to the risk manager. It is important for agents with nonthreshold
effects that the range of values from different models or approaches be
presented along with an analysis of key assumptions  and discussion of
key critical judgmental factors-e.g., the key assumptions that might bear
on the likelihood that the true risk is below the upper bound.
  After CEELs have been calculated according to the procedures de-
scribed in this document; they must be submitted for  review by a com-
mittee of experts (e.g., COT or a comparable group).  If necessary, pro-
posed CEELs will be modified in response to recommendations of the re-
view committee before'being; finalized. Recommended CEELs for which
significant data gaps exist should Be considered preliminary and should
be reviewed and updated as new data become available.  All chemical
substances studied thus far have toxicity data gaps; therefore, all recom-
mended CEELs should be considered for possible updating every 5-10
years, or even earlier if new data on the compounds under investigation
become available.
  When toxicity data for a chemical under consideration are judged to
be inadequate to recommend CEELs with an acceptable level of confi-
dence, relevant testing should be recommended.

-------
        References
 Administrative Conference of the United States: Office of the Chairman.
   1983.  A Guide to Federal Rulemaking.  Washington, D.C.:  U.S.
   Government Printing Office.
 Albertini, R.J.,  J.A. Nicklas,  J.P. O'Neill, and S.H. Robison.  1990.
   In vivo somatic mutations in humans:  Measurement and analysis.
   Annu. Rev. Genet. 24:305-326.
 Aldridge, W.N. 1986. The biological basis and measurement of thresh-
   olds.  Annu. Rev. Pharmacol. Toxicol. 26:39-58.
 Annitage, P., and R. Doll. 1961.  Stochastic models for carcinogenesis.
   Pp. 19-38 in Proceedings  of the Fourth Berkeley Symposium on
   Mathematical Statistics and Probability, Vol.  4,  J. Neyman,  ed.
   Berkeley, Calif.: University of California Press.
 Barnes, D.G., and M.L. Dourson. 1988.  Reference dose (RfD):  De-
   scription and use  in health risk  assessments.   Regul. Toxicol.
   Pharmacol. 8:471-486.
 Bishop, J.M.    1987.   The molecular genetics  of cancer.   Science
   235:305-311.
 Bogen, K.T.  1989.  Cell proliferation kinetics and multistage cancer
   risk models.  J.  Natl. Cancer. Inst. 81:267-277.
 Bond, J.A.  1990.  Contributions of the respiratory tract to the metabolic
   fate of inhaled  organic chemicals.   Pp. 131-154 in Principles of
   Route-to-Route Extrapolation for Risk Assessment, T.R. Gerrity  and
   C.J. Henry, eds. New York:  Elsevier.
Bond, J.A., S.M. Baker, and W.E. Bechtold.  1985. Correlation of the
   octanol:water  partition  coefficient  with  clearance  halftimes   of
   intratracheally instilled  aromatic hydrocarbons in rats.  Toxicology
   36:285-295.
Boulet, L.P. 1988. Increases in airway responsiveness following acute
   exposure to  respiratory  irritants:  Reactive airway dysfunction syn-
   drome or occupational asthma?  Chest 94:476-481.

                               99

-------
I  I
                                                                                                      111 < II
                                                   100     CEELS FOR HAZARDOUS SUBSTANCES
                                                                   \\. ",'  ' •'  "!i'i!|li:"°'J: •• .IS'"	i'"1 ifi •»	i1' ••'n"|i'i ,	it1 i"' /'i!.', "i.', " '   it   i!1 iiU'11'. I i'llir in'"'':'!1 ' *
                                                   Bronaughs, R.L., and S.W. Collier.  1991.  In vivo percutaneous ab-
                                                     sorption.  Pp. 61-73 in Dermatotoxicology, 4th Ed., F.N.  Marzulh
                                                     arid H.I. Maibach, eds.  New York:  Hemisphere.
                                                   Brooks  sK;''M.AVWeiss,'and"i:L: Bernstein.  1985. Reactive airway
                                                     dysfunction syndrome (RADS): Persistent asthma syndrome after high
                                                     level  irritant exposures': Chest 88:376-384.
                                                   Buckley, L.A.,  X.Z. Jiang, R.A. James, K.T. Morgan, and C.S. Bar-
                                                     row.  1984.  Respiratory tract lesions induced by sensory irritants at
                                                     the RD50 concentration. Toxicol. Appl. Pharmacol. 74:417-429.
                                                   Calabrese, E.J.  1983. Principles of Animal Extrapolation. New York:
                                                     John  Wiley & Sons'.'  603 pp.                                 .
                                                   Chen, JJ., and R.L. Kodell.  1989.  Quantitative risk assessment for
                                                     teratological effects.  J. Am. Stat. Assoc. 84:966-971.
                                                   Chen, J.J., R:L. Kodell, and D.W. Gaylor.  1988.  Using the biological
                                                     twb-slage model to assess risk from short-term exposures. Risk Anal.
                                                  :  ' "8:223-230.
                                                   Clement International Corp.  1991. Health Effects and Dose-Response
                                                     Assessment for Hydrogen Chloride Following Short-Term Exposure.
                                                     Interim Final Report:  Prepared for the Office of Air Quality Planning
                                                     and Standards, Pollution  Assessment Branch, U.S. Environmental
                                                     Protection Agency, Research Triangle Park, N.C.
                                                   Crapo,  J.D., L.-Y.  Chang, F.J. Miller, and R.B. Mercer.  1990. As-
                                                     pects of respiratory tract structure arid function important for dosime-
                                                     try modeling: Interspecies comparisons.  Pp. 15-32 in Principles of
                                                     Route-to-Route Extrapolation for Risk Assessment, T.R. Gerrity and
                                                     C.J.  Henry, eds. New York:  Elsevier.
                                                   Crump, K.S.  1984aT" A new method for  determining allowable daily
                                                     intakes. Fundam. Appl. Toxicol. 4:854-871.
                                                   Crump, K.S. l984b^ An improved procedure for low-dose carcinogenic
                                                     risk' assessment  from animal data.  J.  Environ.  Pathol.  Toxicol.
                                                     Oncol. 5:339-348.                                     '
                                                   Crump, K.S., and  R.B.  Howe.  1984.  The multistage model with a
                                                     time-dependent dose pattern: Applications to carcinogenic risk assess-
                                                     ment.  Risk Anal. 4:163-176.
                                                   Crump, K.S., D.G. Hoel, C.H. Langley, and R. Peto.  1976. Funda-
                                                     mental carcinogenic processes and their implications for low dose risk
                                                      assessment.  Cancer Res. 36:2973-2979.
                           'I!  ' '

-------
                                          REFERENCES     101

Dixon, R.L.  1986. Toxic responses of the reproductive system. Pp.
   432-477 in Casarett and Doull's Toxicology:  The Basic Science of
   Poisons, 3rd Ed., C.D.  Klaassan, M.O.  Amdur, and J. Doull, eds.
   New York: Macmillan.
Dourson, M.L.   1986. New approaches in the derivation of acceptable
   daily intake (ADI).  Comments Toxicol. 1:35-48.
Dourson, M.L., and J.F. Stara.  1983.  Regulatory history and experi-
   mental  support  of uncertainty (safety)  factors.   Regul.  Toxicol.
   Pharmacol. 3:224-238.
ECETOC  (European  Chemical  Industry,  Ecology  and Toxicology
   Centre). 1991.  Emergency Exposure Indices for Industrial Chemi-
   cals. Technical Report No. 43. Brussels, Belgium:  European Chem-
   ical Industry, Ecology and Toxicology Centre.
EPA (U.S. Environmental Protection Agency).  1983a.  Toxic substances
   control:  Good laboratory practice standards—Final rule.  40 CFR
   Part 792.  Fed. Regist. 48:53922-53944.
EPA (U.S. Environmental Protection Agency).  1983b.  Pesticide pro-
   grams:  Good laboratory practice standards—Final rule. 40 CFR Part
   160. Fed.  Regist. 48:53946-53969.
EPA (U.S. Environmental Protection Agency).  1986.  Guidelines for
   carcinogen risk assessment.  Fed. Regist. 51:33992-34003.
EPA (U.S. Environmental  Protection Agency).   1987.  Report of the
   EPA Workshop on the Development of Risk Assessment Methodolo-
   gies for Tumor Promoters. EPA 600/9-87/013. Office of Health and
   Environmental Assessment.  Washington, D.C.: U.S. Environmental
   Protection Agency.
EPA (U.S. Environmental Protection Agency). 1988. Reference Dose
   (RfD):  Description and Use in Health Risk Assessment. Integrated
   Risk Information System (IRIS). Online.  Office of Health and Envi-
   ronmental Assessment, Environmental Criteria and Assessment Office.
   Cincinnati, Ohio: U.S. Environmental Protection Agency.
EPA (U.S. Environmental  Protection Agency).   1990a.  Summary of
   EPA Workshop on Effects Resulting from Short-Term Exposures to
   Toxic Substances, March 12-13. Clement Associates and ICF Incor-
   porated, Fairfax, Va.
EPA (U.S. Environmental Protection Agency).  1990b.  Interim Methods
   for Development of Inhalation Reference Concentrations. EPA 600/8-

-------
 102     CEELS FOR W&AKDOUS SUBSTANCES

    90/066A. Environmental Criteria and Assessment Office.  Research
    Triangle Park, N.C.:  U.S. Environmental Protection Agency.
 EPA (U.S. Environmental Protection  Agency).   1991.  Guidelines for
    developmental   toxicity   risk   assessment.      Fed.   Regist.
                                                         I11:'1!1
                                                          i	if
                   .
 EPA (U.S. Environmental Protection Agency).  1992.  A cross-species
    scaling factor for carcinogenic risk assessment based on  equivalence
    of mg/kg3/4/day:  Draft report.  Fed. Regist. 57:24152-24173.
 Faustnian, E.M.,  D.G. Wellington, W.P.  Smith, and C.A. Kimmel.
    1988.   Characterization of a developmental toxicity  dose-response
    model. Environ. Health Perspect! 79:229-241.
 FDA (U.S. Food  and Drug Administration^  1985.   Sponsored com-
    pounds in food-producing animals:  Proposed rule and notice.  Fed.
    Regist. 50:45530-45556.	
 Flynn, G.  1990.   Physical chemical  determinants of  skin absorption.
    Pp. 93-127 in  Principles of Route-to-Route Extrapolation for Risk
    Assessment, T.R. Gerrity and C.J. Henry, eds.  New York: Elsevier.
 Freireich, E.J., E.A. Gehan, D.P. Rail, L.H. Schmidt, and H.E. Skip-
    per.  1966.  Quantitative comparison of anticancer agents in mouse,
    rat,  hamster, dog, monkey and man.   Cancer  Chemother.  Rep.
    50:219-244.   ,	'	."".'"	,	
 Gardner, D.E., Crapo, J.D., and E.J. Massaro, eds. In press. Toxicol-
    ogy of the Lung.  New York:  Raven.
 Gargas, M.L., M.A. Medinsky,  and  ME. Andersen.  In press. Ad-
    vances in physiological modeling approaches for understanding the
    disposition of inhaled vapors.   Pp. 461-483 in Toxicology of the
    Lung, D.E. Gardner, J.D. Crapo, and E.J. Massaro, eds.  New York:
......   Raven.   , n	, ^,,	\	(  ,,: ,	,  (	,„  	^ 	 ^	  [	
 Gaylor, D.W.  1983? The use of safety factors lor controlling risk.  J.
;.'.,.  TpxicQl. Environ..Health.11:329-336.  i		|_.	
 Gaylor" D.W   1992.  Incidence of  developmental defects "at the "no
    observed adverse effect level (NOAEL). Regul. Toxicol. Pharmacol.
	,	I:	    p	 ,.,-,	, i ,,	,	 v,..i.	I ..ni" i .. .'in	 -	'	 .  i	 '.I	 I-	
.,'  15:151-160.	,	 ,	  ,	r		   	 ,h.	  	t	
 Gaylor,  D.W., and R.L. Kodell.  1980.  Linear interpolation algorithm
    for low dose risk assessment of toxic substances. J. Environ. Pathol.
  , 'Toxipoi. 4(5&6):305-312.	'"  	'"	""""	   	'  ' """"	' '	
 Gaylor, D.W., and W.L. Slikker.  1990. Risk assessment  for neurotoxic
    effects. Neurotoxicology 11:211-218.

-------
                                           REFERENCES    103

 Gerde, P., and A.R. Dahl.  1991.  A model for the uptake of inhaled
   vapors in the nose of the dog during cyclic breathing. Toxicol. Appl.
   Pharmacol. 109:276-288.
 Glowa,  J.R.   1991.   Dose-effect approaches to risk assessment..
   Neurosci. Biobehav. Rev. 15:153-158.
 Grant, W.M.  1974. Toxicology of the Eye, 2nd Ed.  Springfield, 111:
   Charles C Thomas.
 Greenfield, R.E., L.B. Ellwein, and S.M.  Cohen.   1984.   A general
   probabilistic model of carcinogenesis:  Analysis of experimental uri-
   nary bladder cancer. Carcinogenesis 5:437-445.
 Guth,  D.J., A.M. Jarabek,  L.  Wymer,  and R.C.  Hertzberg.   1991.
   Evaluation of Risk Assessment  Methods for Short-term  Inhalation
   Exposure.  Paper presented at the 84th Annual Meeting of the Air and
   Waste Management Association, Paper No. 91.173.2, June 16-21,
   1991, Vancouver, B.C.
 Hakkinen, P.J., and H.P. Witschi.  1985.  Animal models. Pp. 95-114
   in Toxicology of Inhaled Materials:  General Principles of Inhalation
   Toxicology, Handbook of Experimental Pharmacology, Vol. 75, H.P.
   Witschi and J.D. Brain,  eds.  New York:  Springer-Verlag.
 Heck,  H.d'A., M. Casanova, and  T.B. Starr.   1990.  Formaldehyde
   toxicity: New understanding.  CRC Crit. Rev. Toxicol. 20:397-426.
 Hemminki, K., E. Grzybowska, M. Chorazy, K. Twardowska-Saucha,
   K.L. Patma, K. Randerath, D.H. Phillips, A. Hewer,  and R.M.
   Santella. 1990. DNA adducts in humans environmentally exposed to
   aromatic compounds  in an industrial area of Poland.  Carcinogenesis
   11:1229-1231.
 Henderson, R.F., W.E. Bechtold, M.A.  Medinsky, J.P.  Fischer, and
   T.T. Lee.  1988.  The  effects of molecular weight/lipophilicity on
   clearance  of organic compounds from  lungs.    Toxicol.  Appl.
   Pharmacol. 95:515-521.
Hertzberg, R.C.   1989.  Fitting a model to categorical response data
   with application to species extrapolation of toxicity.  Health Phys.
   57(Suppl. 1):405-409.
Hertzberg, R.C., and M. Miller.  1985.  A statistical model for species
   extrapolation using categorical response data.  Toxicol. Ind. Health
   1:43-57.
Hill, R.P. 1990.  Tumor progression: Potential role of unstable genom-
   ic changes. Cancer Metastasis Rev. 9:137-147.

-------
i",:: iiiiiui   < 
-------
                                           REFERENCES     105

 Kimmel, C.A., and D.W. Gaylor. 1988. Issues in qualitative and quan-
   titative risk analysis for developmental toxicology.   Risk  Anal.
   8:15-20.
 Klaassen, C.D.  1986.  Principles of toxicology.  Pp. 11-32 in Casarett
   and Doull's Toxicology:  The Basic Science of Poisons, 3rd Ed.,
   C.D.  Klaassen, M.O.  Amdur,  and J. Doull, eds.   New  York:
   Macmillan.
 Klopman, G.  1985.  Predicting toxicity through a computer automated
   structure evaluation program.  Environ. Health Perspect. 61:269-274.
 Klopman, G., and H. Rosenkranz.  1991.  Structure-activity relations:
   Maximizing the usefulness of mutagenicity and carcinogenicity data-
   bases.  Environ. Health Perspect. 96:67-75.
 Kodell, R.L., and C.N. Park.   1991.  Linear Extrapolation in Cancer
   Risk Assessment.  Proceedings of the First Workshop of the Cancer
   Dose-Response Working Group.  Washington, D.C.: ILSI Risk Sci-
   ence Institute.
 Kodell, R.L., D.W. Gaylor, and J.J. Chen.  1987. Using average life-
   time dose rate for intermittent exposures to carcinogens. Risk Anal.
   7:339-345.
 Krewski, D.R., and J.A. Van Ryzin.  1981.  Dose response models for
   quanta! response toxicity data.  Pp. 201-231 in Statistics and Related
   Topics, M. Csorgo, D.W. Dawson, J.N.K. Rao, and A.K. Sal eh, eds.
   New York: North Holland.
 Krewski, D.R., T. Thorslund, and J. Withey. 1989.  Carcinogenic risk
   assessment of complex mixtures.  Toxicol. Ind. Health 5:851-867.
 Kupper, L.L., C. Portier, M.D. Hogan,  and E. Yamamoto. 1986. The
   impact  of  litter effects on dose-response modeling in teratology.
   Biometrics 42:85-98.
Lehman, A.J., and O.G. Fitzhugh. 1954.  100-fold  margin of safety.
   Assoc. Food Drug Off. U.S. Q. Bull. 18:33-35.
Lu, F.C.  1985.  Basic Toxicology:  Fundamentals, Target Organs, and
   Risk Assessment. New York:  Hemisphere. 276 pp.
Lutz, W.K.  1990. Dose-response relationship and low-dose extrapola-
   tion in chemical carcinogenesis. Carcinogenesis 11:1243-1247.
Manson, J.M., and Y.J. Kang.  1989.   Test methods for assessing fe-
   male reproductive  and  developmental toxicology.  Pp.  311-359 in
   Principles and Methods  of Toxicology, 2nd Ed.,  A.W. Hayes, ed.
   New York:  Raven.

-------
 106     CEELS FOR HAZARDOUS SUBSTANCES
  :• ."                      i        i      i   | in      i      i i
 Mantel, N., and W.R.  Bryan.  1961.  "Safety" testing of carcinogenic
   agents.  J. Natl. Cancer Inst. 27:455-470.
 Mantel, N., and M.A.  Schneiderman.  1975.  Estimating safe levels, a
   hazardous undertaking.  Cancer Res. 35:1379-1386.
 Marx, J.  1991.  Zeroing in on individual  cancer risk.   Science
   253:612-616.
 Marzulli, F.N. 1962. Barriers to skin penetration. J. Invest. Dermatol.
   39:387-393.
 Marzulli, F.N., D.W.C. Brown, and H.I. Maibach.  1969.  Techniques
   for studying skin penetration.  Toxicol. Appl. Pharmacol.  15(Suppl.
   3):76-83.
 Mayer,  JV, D.  Warburton, A.M. Jeffrey, R.  Pero, S.  Walles,  L.
   Andrews, M. Toor,  L. Latriano, L. Wazneh, D. Tang, W.-Y. Tsai,
   M. Kuroda, and F. Perera.  1991. Biologic markers in ethylene
   oxide-exposed workers and controls. Mutat. Res. 248:163-176.
 McClellan, R.O:, and RTF. Henderson, eds. 1989. Concepts in Inhala-
   tion Toxicology. New York:  Hemisphere.
 McColl, R.S.  1990.  Biological Safety Factors in Toxicological Risk
   Assessment. SSC H49-49/1990E. Ottawa, Ont.:  Supply and Servic-
   es Canada.  Available from NTIS,  Springfield,  Va., Doc.  No.
   MIC90-06409.
 Moolgavkar, S.H.  1986.   Carcinogenesis  modeling:  From  molecular
   biology to epidemiology.  Annu. Rev. Public Health 7:151-169.
 Moolgavkar, S.H., and A.G. Knudson, Jr.  1981.  Mutation and cancer:
   A model  for  human  carcinogenesis.    J.  Natl.  Cancer  Inst.
   66:1037-1051	
 Moolgavkar,  S.H., and D.J. Venzon.  1979.  Two-event models for
   carcinogenesis:   Incidence  curves for childhood and adult tumors.
^v Math. Biosci. 47:55-77.'
 Muller, K.E., C.N. Barton, and V.A^  Benignus.  1984.  Reconimenda-
   tions  for appropriate statistical practice  in toxicologic experiments.
   Neiirotoxicology 5:113-125.
 M^nro, I.C., and D.R. Krewski.  1981. Risk assessment and regulatory
   decision making.  Food Cosmet. Toxicol. 19:549-560.
 Murdoch, D J., and D. Rrewski.  1988. Carcinogenic risk assessment
   with time^ependent exposure patterns.   Risk Anal. 8:521-530.
 Nowell, P.C.  1991.  Commentary:  How many human cancer genes?
   J. Natl. Cancer Inst. 83:1061-1064.

-------
                                           REFERENCES    107

 NRC (National Research Council). 1970. Evaluating the Safety of Food
   Chemicals.  Washington, D.C.: National Academy of Sciences.
 NRC (National Research Council).  1977.  Drinking Water and Health.
   Washington, D.C.:  National Academy Press.
 NRC (National Research Council).  1980.  Drinking Water and Health,
   Vol. 3.  Washington, D.C.:  National Academy Press.
 NRC (National Research Council).  1984.  Toxicity Testing: Strategies
   to  Determine  Needs and Priorities.  Washington, D.C.:  National
   Academy Press.
 NRC (National Research Council). 1986a. Drinking Water and Health,
   Vol. 6.  Washington, D.C.:  National Academy Press.
 NRC (National Research Council).  1986b.  Criteria and Methods for
   Preparing Emergency Exposure Guidance Level (EEGL), Short-Term
   Public Emergency Guidance Level (SPEGL), and Continuous Expo-
   sure Guidance Level (CEGL) Documents.  Washington, D.C.:  Na-
   tional Academy Press.
 NRC  (National  Research  Council).    1987.   The application  of
   pharmacokinetic data in carcinogenic risk assessment. Pp. 441-468 in
   Pharmacokinetics  in Risk Assessment: Drinking Water and Health,
   Vol. 8.  Washington, D.C.:  National Academy Press.
 NRC (National Research Council).  1989.  Biologic Markers in Pulmo-
   nary Toxicology.  Washington, D.C.: National Academy Press.
 OSTP (Office  of Science and Technology Policy).  1985.  Chemical
   carcinogens: A review of the  science and  its associated principles.
   Fed. Regist. 50:10371-10442.
 Perera, P.P.  1990.  Molecular epidemiology:   A new tool in assessing
   risks of environmental carcinogens.  CA Cancer J. Clin. 40:277-288.
 Perera, P.P.,  R.M. Santella, and  M.C.  Poirier.   1985.  Potential
   methods to  monitor human populations exposed to carcinogens:
   Carcinogen-DNA  binding as an  example.    Pp.  211-223 in Risk
   Quantitation and Regulatory Policy, Banbury Report 19, D.G. Hoel,
   R.A. Merrill,  and F.P.  Perera, eds.  Cold Spring Harbor, N.Y.:
   Cold Spring Harbor Laboratory Press.
Perera, F., J. Mayer, R.M. Santella, D. Brenner, A. Jeffrey,  L.
   Latriano, S. Smith, D.  Warburton, T.L.  Young, W.Y. Tsai,  K.
   Hemmiaki, and P. Brandt-Rauf.   1991.  Biologic markers in risk
   assessment for environmental carcinogens. Environ. Health Perspect.
   90:247-254.

-------
'it,'1:	mi:.- i"'<;i, IK: .i;
                                                     108     CEELS FOR HAZARDOUS SUBSTANCES

                                                     Pinkel, D.  1958. The use of body surface area as a criterion of drug
                                                       dosage hi cancer chemotherapy. Cancer Res. 18:853-856.
                                                     Plopper, C.G., A.T. Mariassy,  and L.H. Hill.  1980.  Ultrastructure of
                                                       the nonciliated bronchiolar epithelial (Clara) cell of mammalian lung:
                                                       I. A comparison of rabbit, guinea pig, rat, hamster, and mouse. Exp.
                                                       Lung Res.  1:139-154;
                                                     Portier, CJ.  1987.  Statistical properties of a two-stage model of carci-
                                                       nogeriesis.  Environ. Health Perspect. 76:125-131.
                                                     Pprtier, CJ., ami N.L. Kaplan.   1989.  Variability of safe dose esti-
                                                       niates when using complicated models of the carcinogenic process. A
                                                       case  study:    Methylene   chloride.    Fundam.   Appl.  Toxicol.
                                                      «.-;i3:5^3-544.	""	' '	'	"	""	'	    "  "	"	"'
                                                     Rai, K., and J. Van Ryzin. 1985.  A dose-response model for teratolog-
                                                       ical experiments involving quanta! responses. Biometrics 41:1-9.
                                                     Ramsey^  J'.(I', anSKO; Andersen.   1984.  A physiologically based
                                                       description of the inhalation pharmacokinetics of styrene in rats and
                                                       humans. Toxicol. Appl. Pharmacol. 73:159-175.
 'fi{'.f j V. .;"'.*;' )  ''::v    	,    :'',.;:	•  ..   '   Reidr^M.	J'i98^	NeSds for animal models of hui
   _  _                                   of human diseases of the
  Respiratory system.  Am. J. Pathol.  101(3 Suppl.):S89-S101.
Ryan, L.M.  1989; Dose response models for developmental toxicity.
  Pages 154-159 in the Proceedings of the American Statistical Associa-
  tion, Biopharmaceutical Section. Alexandria, Va.: American Statisti-
  cal Association.
Schardein, LLi  1985.  Chemically-Induced Birth Defects. New York:
  Marcel Dekker,  Inc.
Schlesinger, R.B;  1990* The interaction of inhaled toxicants with respi-
  ratory tractclearance  mechanisms.   CRC  Crit.  Rev.  Toxicol.
 "20:257-286.  	
Slauson, D-t)., and F.F. Hahn.  1980.  Criteria for development of
  animal models of diseases of the respiratory system: The comparative
  approach in respiratory disease model development.  Am. J. Pathol.
  101(3 Suppl.):S103-S122.
Snider, G.L., B.C. Lucey, and P.J. Stone.   1986.  Animal  models of
  emphysema.  Am. Rev. Respir. Dis.  133:149-169.
Snipes,  M.D.   1990   Long-term retention and clearance of particles
  inhaled by mammalian species. CRC Crit. Rev. Toxicol. 20:175-211.
Society of Toxicology, Task Force of Past Presidents.  1982.  Animal

-------
                                           REFERENCES     109

   data in  hazard  evaluation:   Paths  and pitfalls.   Fundam.  Appl.
   Toxicol. 2:101-107.
Starr, T.B. 1990. Quantitative cancer risk estimation for formaldehyde.
   Risk Anal. 10:85-91.
Stowers, S.J., R.R. Maronpot, S.H. Reynolds, and M.W.  Anderson.
   1987. The role of oncogenes in chemical carcinogenesis.  Environ.
   Health Perspect. 75:81-86.
Swenberg,  J.A.  1989.  Carcinogenic responses of the nasal passages.
   Pp.  299-311  in Concepts in Inhalation Toxicology,  R.O.  McClellan
   and  R.F. Henderson, eds.  New York: Hemisphere.
Thorslund, T.,  and G. Charnley.   1988.  Quantitative dose-response
   models for tumor promoting agents. Pp. 245-255 in Carcinogen Risk
   Assessment:  New Directions in the Qualitative and  Quantitative As-
   pects, Banbury Report 31, R.W. Hart and F.D. Hoerger, eds. Cold
   Spring Harbor, N.Y.: Cold Spring Harbor Laboratory Press.
Travis, C.C., and R.K. White. 1988.  Interspecific scaling of toxicity
   data. Risk Anal. 8:119-125.
Wallace, L.A.  1987.  The Total  Exposure Assessment Methodology
   (TEAM) Study: Summary and Analysis, Vol. I. EPA 600/6-87/002a.
   Washington, D.C.:  U.S. Environmental Protection Agency.
Weinstein, I.E., and P.P. Perera.  1982. Molecular cancer epidemiolo-
   gy.  Pp.  3-15 in Indicators of Genotoxic  Exposure,  Banbury Report
   13, B.A. Bridges, B.E. Butterworth, and I.B. Weinstein, eds.  Cold
   Spring Harbor, N.Y.: Cold Spring Harbor Laboratory Press.
Whittemore, A., and J. Keller.  1978.  Quantitative theories of carcino-
   genesis.  SIAM Rev. 20:1-30.
Wilson, J.G.  1973.  Environment and Birth  Defects.  New York: Aca-
   demic.
Zenick H.,  and  E.D.  Clegg.  1989.  Assessment of male reproductive
   toxicity:  A risk assessment approach.  Pp. 275-309 in Principles and
   Methods of Toxicology, 2nd Ed., A.W. Hayes, ed.   New York:
   Raven.

-------

-------
NATIONAL ACADEMY PRESS
The National Academy Press was created by the
National Academy of Sciences to publish the reports
issued by the Academy and by the National Academy
of Engineering, the Institute of Medicine, and the
National Research Council, all operating under the
charter granted to the National Academy of Sciences
by the Congress of the United States.

-------
|!   '   ';            !!•«!,

-------