p.1
United States Office of Research and EPA/600/8-90/066F
Environmental Protection Development October 1994
Agency Washington DC 20460
v>EPA Methods for Derivation of
Inhalation Reference
Concentrations and
Application of Inhalation
Dosimetry
-------
p. 2
EPA/600/8-90/066F
October 1994
METHODS FOR DERIVATION OF
INHALATION REFERENCE CONCENTRATIONS
AND APPLICATION OF INHALATION DOSIMETRY
Environmental Criteria and Assessment Office
Office of Health and Environmental Assessment
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
Printed on Recycled Paper
-------
p. 3
DISCLAIMER
This document has been reviewed in accordance with U.S. Environmental Protection
Agency policy and approved for publication. Mention of trade names or commercial
products does not constitute endorsement or recommendation for use.
11
-------
p. 4
TABLE OF CONTENTS
LIST OF TABLES vi
LIST OF FIGURES viii
AUTHORS, CONTRIBUTORS, AND REVIEWERS xiii
LIST OF ACRONYMS AND ABBREVIATIONS xix
GLOSSARY xxvii
1. INTRODUCTION AND OVERVIEW 1-1
1.1 INHALATION REFERENCE CONCENTRATION:
DEVELOPMENT, DEFINITION, AND DERIVATION 1-1
1.2 GENERAL PRINCIPLES OF DOSE-RESPONSE
ASSESSMENT FOR NONCANCER TOXICITY 1-8
1.3 GUIDELINES ON SPECIFIC ENDPOINTS 1-14
1.4 USE OF THE INHALATION REFERENCE
CONCENTRATION IN THE NATIONAL ACADEMY
OF SCIENCES RISK ASSESSMENT AND RISK
MANAGEMENT PARADIGM 1-15
1.5 OCCUPATIONAL EXPOSURE LIMITS VERSUS
INHALATION REFERENCE CONCENTRATIONS 1-16
1.6 PRIMARY NATIONAL AMBIENT AIR QUALITY
STANDARDS VERSUS INHALATION REFERENCE
CONCENTRATIONS 1-17
1.7 STATE-OF-THE-ART APPLICATIONS TO THE
DEVELOPMENT OF THE INHALATION REFERENCE
CONCENTRATION METHODOLOGY 1-18
2. QUALITATIVE EVALUATION OF THE DATA BASE 2-1
2.1 GUIDELINES FOR SELECTIONS OF KEY STUDIES 2-2
2.1.1 Human Data 2-2
2.1.1.1 Molecular Epidemiology and Biologic
Markers 2-3
2.1.1.2 Epidemiologic Data 2-16
2.1.1.3 Nonepidemiologic Data 2-20
2.1.1.4 Intraspecies Variability and Identifying
Sensitive Subgroups 2-22
2.1.1.5 Summary 2-24
2.1.2 Laboratory Animal Data , 2-26
2.1.2.1 Study Design 2-26
2.1.2.2 Impact of Experimental Protocol 2-27
2.1.2.3 Appropriateness of Laboratory Animal
Species as a Model for Humans 2-35
2.1.2A Study Validity and Relevance to
Extrapolation 2-43
2.1.3 Summarizing the Evidence 2-44
iii
-------
p. 5
TABLE OF CONTENTS (cont'd)
Page
3. CONCEPTUAL BASIS FOR INHALATION DOSE-RESPONSE
ASSESSMENT METHODOLOGY 3-1
3.1 FACTORS CONTROLLING COMPARATIVE
INHALED DOSE 3-1
3.1.1 Respiratory Anatomy and Physiology 3-3
3.1.1.1 Respiratory Regions and Branching Patterns .. 3-3
3.1.1.2 Clearance Mechanisms 3-18
3.1.2 Physicochemical Characteristics of the Inhaled
Toxicant 3-29
3.1.2.1 Particles 3-30
3.1.2.2 Gases and Vapors 3-30
3.2 MODELING COMPARATIVE DOSIMETRY OF
INHALED TOXICANTS 3-34
3.2.1 Particle Disposition Model Based on
Available Data 3-35
3.2.2 Gas Categorization Scheme Directs Default
Gas Modeling 3-36
3.2.3 Summary Considerations for Judging Model
Structures 3-39
4. QUANTITATIVE PROCEDURES 4-1
4.1 MINIMUM DATA BASE CRITERIA 4-2
4.1.1 Evaluation of Comprehensiveness 4-2
4.1.2 Route-to-Route Extrapolation 4-5
4.1.3 Not-Verifiable Status 4-12
4.2 DESIGNATION OF EFFECT LEVELS 4-12
4.3 CALCULATION OF HUMAN EQUIVALENT
CONCENTRATIONS 4-17
4.3.1 Conversion to Standard Units 4-19
4.3.2 Temporal Relationships of Toxicity and
Duration Adjustment 4-20
4.3.3 Use of Pharmacokinetic and Pharmacodynamic Data . . . 4-22
4.3.4 Default Dosimetric Adjustment and Physiological
Parameters 4-25
4.3.5 Dosimetric Adjustments for Particle Exposures 4-29
4.3.5.1 Respiratory Effects 4-37
4.3.5.2 Remote (Extrarespiratory) Effects 4-38
4.3.5.3 Additional Issues for Particle Dosimetry .... 4-39
4.3.6 Dosimetric Adjustments for Gas Exposures 4-44
4.3.6.1 Respiratory Effects 4-46
4.3.6.2 Remote (Extrarespiratory) Effects 4-57
4.3.6.3 Additional Assumptions and Default Values . . 4-63
4.3.7 Derivation and Dosimetric Adjustment Using
Human Studies 4-64
iv
-------
p. 6
TABLE OF CONTENTS (cont'd)
Page
4.3.7.1 Selecting the Threshold Estimate 4-65
4.3.7.2 Defining the Exposure Level 4-65
4.3.7.3 Dosimetric Adjustment for Human Data .... 4-66
4.3.7.4 Uncertainty Factors for Human Data 4-67
4.3.8 Data Array Evaluation and Choice of Principal
Study/Studies 4-68
4.3.9 Operational Derivation of the Inhalation Reference
Concentration 4-70
4.3.9.1 Application of Uncertainty Factors 4-73
4.3.9.2 Assignment of Confidence Levels 4-80
5. REFERENCES 5-1
APPENDIX A—Alternative Approaches to the Estimation of
No-Observed-Adverse-Effect Levels A-l
APPENDIX B—Criteria for Assessing the Quality of Individual
Epidemiological Studies B-l
APPENDIX C—Criteria for Causal Significance C-l
APPENDIX D—Adverse Human Respiratory Health Effects D-l
APPENDIX E—Guidance on Pulmonary Function Testing E-l
APPENDIX F—Criteria for Assessing the Quality of Individual
Laboratory Animal Toxicity Studies F-l
APPENDIX G—The Particle Deposition Dosimetry Model G-l
APPENDIX H—Particle Sizing Conventions H-l
APPENDIX I—Derivation of an Approach To Determine Human Equivalent
Concentrations for Effects of Exposures to Gases in
Categories 1 and 2 1-1
APPENDIX J—Derivation of an Approach To Determine Human Equivalent
Concentrations for Extrarespiratory Effects of Category 3 Gas
Exposures Based on a Physiologically Based Pharmacokinetic
Model Using Selected Parameter Values J-l
-------
p. 7
LIST OF TABLES
Number Page
2-1 Steps in the Development of a Biomarker 2-8
2-2 Qualitative Rating for Validity of Biologic Markers 2-10
2-3 Comparison of the Qualities of Field and Experimental Approaches
in the Study of Threshold Limit Value/Biologic Exposure
Indices Relationships 2-15
2-4 Prevalence of Subgroups Susceptible to Effects of Common
Pollutants 2-23
2-5 Agents Causing Wheezing and Bronchoconstriction 2-41
2-6 Approach for Summarizing the Evidence from Diverse Data 2-45
2-7 Human Data for Use in Health Risk Assessment 2-46
3-1 Respiratory Tract Regions 3-4
3-2 Comparative Lower Airway Anatomy as Revealed on Casts 3-8
3-3 Normal Surface Airway Epithelium: Cell Types 3-23
3-4 Some Specific Lung Cell Types and Their Functions 3-24
3-5 Main Species Differences in Epithelial Cells and Glands 3-25
3-6 Hierarchy of Model Structures for Dosimetry and Interspecies
Extrapolation 3-40
4-1 Minimum Data Base for Both High and Low Confidence in the
Inhalation Reference Concentration 4-3
4-2 Four Types of Effect Levels Considered in Deriving Inhalation
Reference Concentrations for Noncancer Toxicity 4-14
4-3 Effect Levels Considered in Deriving Inhalation Reference
Concentrations in Relationship to Empirical Severity
Rating Values 4-16
4-4 Default Surface Area Values for Respiratory Effects 4-26
4-5 Body Weight Default Values—Rats 4-28
VI
-------
p. 8
LIST OF TABLES (cont'd)
Number Page
4-6 Intercept and Coefficient Values Used in Algorithm To Calculate
Default Minute Volumes Based on Body Weight .............. 4-29
4-7 Comparison of the Highest Individual Species NOAELjE and
Its LOAEL[HEC] ................................. 4-71
4-8 Guidelines for the Use of Uncertainty Factors in Deriving Inhalation
Reference Concentration ............................ 4-76
4-9 The Use of Uncertainty Factors in Deriving an Inhalation
Reference Concentration ............................ 4-77
E-l Definition of Various Pulmonary Function Test Volumes and
Capacities ..................................... E-4
E-2 Change in Spirometric Indices Over Time .................. E-17
F-l General Clinical Biochemistry Examinations ................ F-5
F-2 Organs and Tissues Preserved for Histological Examination ....... F-6
G-l Regional Fractional Deposition ........................ G-4
G-2 Deposition Efficiency Equation Estimated Parameters ......... . . G-6
H-l Particle Diameter Definitions ......................... H-5
H-2 Lognormal Conversion Equations for Common Types of Diameters . . H-7
H-3 Percentage of Particles in the Reported Range Associated with the
Number of Standard Deviations Used To Calculate the Geometric
Standard Deviation ................................ H-10
H-4 General Particle Descriptions and Associated Sizes ............ H-12
1-1 Definition of Parameter Symbols Used in Appendix I ........... 1-9
J-l Definition of Symbols .............................. J-5
Vll
-------
p. 9
LIST OF FIGURES
Number Page
1-1 National Research Council (1983) framework for risk assessment
and risk management 1-2
1-2 Schematic characterization of comprehensive exposure-dose-response
continuum and the evolution of protective to predictive
dose-response estimates 1-11
2-1 Biological marker components in sequential progression between
exposure and disease 2-4
2-2 Schematic representation of possible relationships to research
using biologic markers 2-11
2-3 Schematic relationships between threshold limit values in air,
biologic exposure indices, and effects 2-13
3-1 Diagrammatic representation of three respiratory tract regions 3-5
3-2 Schematic representation of selected parameters influencing
regional deposition of particles in the respiratory tract 3-9
3-3 Regional deposition in humans of monodisperse particles by indicated
particle diameter for mouth breathing and nose breathing 3-10
3-4 Schematic representation of selected parameters influencing
regional deposition of gases in the respiratory tract 3-11
3-5 Net dose of ozone versus sequential segments along anatomical
model lower respiratory tract paths for human, rat, guinea pig,
and rabbit 3-12
3-6 Diagram of the nasal passages for the F344 rat modified from
Morgan et al. (1984) 3-15
3-7 Diagram of the nasal passages for the Rhesus monkey modified from
Monticello et al. (1989) 3-16
3-8 Inspiratory airflow patterns in upper respiratory tract of Fischer 344
rat and Rhesus monkey 3-17
3-9 Gas categorization scheme based on water solubility and reactivity
as major determinants of gas uptake 3-37
Vlll
-------
p. 10
LIST OF FIGURES (cont'd)
Number
4-1 Multiple route comparisons for mice and humans administered
chloroform at a dose of 100 milligrams per kilogram body weight ... 4-7
4-2 Differential effects of inhaled and ingested cadmium with increasing
inhaled and ingested doses 4-8
4-3 Decision tree for route-to-route extrapolation 4-10
4-4 Flowchart for calculation of human equivalent concentrations 4-18
4-5 Display Screen 1 of the computer program that calculates
regional deposited dose ratios 4-33
4-6 Display Screen 2 of the program that calculates regional
deposited dose ratios 4-33
4-7 Display Screen 3 of the computer program that calculates
regional deposited dose ratios 4-34
4-8 Display Screen 4 of the computer program that calculates
regional deposited dose ratios 4-34
4-9 Time course of periodicity for Fischer 344 rat exposed 6 hours per
day, 5 days per week to theoretical gas with partition coefficients
as shown 4-61
4-10 Relationship of blood:gas and fatrblood partition coefficients to the
attainment of periodic blood concentrations in the Fischer 344 rat ... 4-62
4-11 Example data array and inhalation reference concentration
derivation 4-72
A-l Graphical illustration of proposed low-dose risk estimation for the
proportion of abnormal responses in developmental toxicity A-7
A-2 Schematic of computing a posterior distribution from a likelihood
function and a prior distribution A-12
A-3 Incidence of nasal turbinate lesions in B6C3F1 female mice exposed
to n-hexane for 13 weeks A-13
A-4 Posterior distribution for the /z-hexane concentration associated with
the specified health effect in Figure A-3 A-13
IX
-------
p. 11
LIST OF FIGURES (cont'd)
Number Page
A-5 Posterior distribution for the concentration of «-hexane associated
with the specified health effects from the combined evidence of
Sanagi et al. (1980) and Dunnick et al. (1989) A-14
A-6 Posterior distributions for the manganese concentration associated
with specified health effects from each of three studies:
Roels et al. (1987a), Iregren (1990), and Chandra et al. (1981) A-15
A-7 Posterior distribution for the manganese concentration associated
with specified health effect using either exposure or estimated
exposure distribution A-16
A-8 Posterior distribution for the concentration of manganese associated
with specified health effect from the combined evidence of Iregren
(1990) and Roels et al. (1987a) with fixed exposure A-16
A-9 Posterior distribution for the concentration of manganese associated
with specified health effect from the combined evidence of Iregren
(1990) and Roels et al. (1987a,b) with exposure distribution A-17
A-10 Categorical data from published results on methyl isocyanante for
exposures of less than 8 hours in duration and shown as NOAEL,
AEL, or lethality A-20
A-ll Categorical data from published results as in Figure A-10, excluding
lethality data A-22
A-12 Categorical regression analysis for data on carboxyhemoglobin
in humans A-22
A-13 Categorical regression analysis of tetrachloroethylene: acute effects . . A-23
E-l Lung volumes and capacities E-3
E-2 Volume-time plot (spirogram) of the forced vital capacity maneuver . . E-6
G-l Comparison of regional deposition efficiencies and fractions for
the mouse G-10
G-2 Range of particles for lognormal distributions with same mass median
aerodynamic diameter but differing geometric standard deviations ... G-l2
-------
p. 12
LIST OF FIGURES (cont'd)
Number Page
G-3 Regional deposited dose ratios for rat: human as a function of mass
median aerodynamic diameter for monodisperse and polydisperse
particle size distributions G-14
H-l An example of the log-normal distribution function of an aerosol .... H-4
H-2 Plot of same aerosol as in Figure H-l on log-probability paper H-6
H-3 Measurement ranges of aerosol monitoring instruments H-8
H-4 Various airborne materials and their size ranges H-ll
1-1 Gas categorization scheme based on water solubility and reactivity
as major determinants of gas uptake 1-5
1-2 Schematic of two-phase mass transport resistance model 1-13
1-3 Schematic of modeling approach to estimate regional respiratory
tract dose of gases 1-18
1-4 Bohr model of ventilation and uptake 1-29
1-5 Schematic of physiologically based pharmacokinetic modeling
approach to estimate respiratory tract dose of gases in Category 2 ... 1-37
1-6 Schematic of surface-liquid/tissue phase concentration during
exhalation 1-40
1-7 Schematic of change in surface-liquid/tissue phase concentration
with distance and time 1-41
1-8 Schematic of change in mass during breathing cycle 1-42
J-l Schematic of the physiologically based pharmacokinetic model
assumed to describe the uptake and distribution of inhaled
compounds J-4
J-2 Plot of NOAEL*jHECj versus NOAEL*[Aj for the rat for four possible
methods of determining NOAELrHECj estimates as defined in
the text J-15
J-3 Plot of NOAEL*rHECj versus NOAEL*^ for the mouse for four
possible methods of determining NOAEL[HECn estimates as defined
in the text J-16
XI
-------
p. 13
AUTHORS, CONTRIBUTORS, AND REVIEWERS
This document was prepared under the direction of the Environmental Criteria and
Assessment Office (ECAO) in Research Triangle Park, NC, of the Office of Health and
Environmental Assessment (OHEA) within the U.S. Environmental Protection Agency's
(EPA's) Office of Research and Development (ORD).
The principal authors are:
Annie M. Jarabek
Environmental Criteria and Assessment Office
Office of Health and Environmental Assessment
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
Linda Hanna, Ph.D.
Sciences International, Inc.
King Street Station
1800 Diagonal Road, Suite 500
Alexandria, VA 22314
Margaret Menache
Center for Extrapolation Modeling
Duke University Medical Center
Durham, NC 27710
John Overton, Jr., Ph.D.
Health Effects Research Laboratory
Office of Health Research
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
The contributing authors to the current document, listed in alphabetical order, are:
Michael Dourson, Ph.D.
Environmental Criteria and Assessment Office
Office of Health and Environmental Assessment
U.S. Environmental Protection Agency
Cincinnati, OH 45268
Linda Erdreich, Ph.D.*
Environmental Research Information, Inc.
New York, NY 10018-3011
Judith A. Graham, Ph.D.
Environmental Criteria and Assessment Office
Office of Health and Environmental Assessment
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
Elaine C. Grose, Ph.D.t
537 Venard Rd.
Clark Summit, PA 18411
Mary Jane Selgrade, Ph.D.
Health Effects Research Laboratory
Office of Health Research
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
*Formerly with ECAO—Cincinnati.
formerly with the Health Effects Research Laboratory—RTF.
Xlll
-------
p. 14
The authors are also grateful to several other individuals who contributed to the
development of earlier versions of the methodology. Their thoughtful discussions at that time
were useful in delineating important issues. These individuals, listed in alphabetical order,
are:
Karen Blackburn, Ph.D.*
Proctor and Gamble
Cincinnati, OH
Christopher DeRosa, Ph.D.*
Agency for Toxic Substances and Disease
Registry
Atlanta, GA
Mark Greenberg, M.S.
Environmental Criteria and Assessment Office
Office of Health and Environmental Assessment
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
Richard Hertzberg, Ph.D.
Environmental Criteria and Assessment Office
Office of Health and Environmental Assessment
U.S. Environmental Protection Agency
Cincinnati, OH 45268
Bruce Peirano, Ph.D.
Environmental Criteria and Assessment
Office
Office of Health and Environmental
Assessment
U.S. Environmental Protection Agency
Cincinnati, OH 45268
William Pepelko, Ph.D.
Human Health Assessment Group
Office of Health and Environmental
Assessment
U.S. Environmental Protection Agency
Washington, DC 20460
Greg Theiss, Ph.D.
Office of Program Planning and Evaluation
Office of Research and Development
U.S. Environmental Protection Agency
Washington, DC 20450
*Formerly with ECAO—Cincinnati.
XIV
-------
p. 15
Lastly, the authors thank Bette Zwayer and Carol Haynes (ECAO-Cincinnati) for
diligently and graciously preparing the original drafts of the manuscript and Judith Olsen for
excellent editorial support. The authors further thank Marianne Barrier, John Barton, Ivra
Bunn, Lynette Cradle, Patricia Felix, Jorja Followill, Miriam Gattis, Lome Godley, Sheila
Lassiter, and Wendy Lloyd of ManTech Environmental Technology, Inc. (Research Triangle
Park, NC), for preparing several drafts for reviews (experts workshop, EPA's Science
Advisory Board [SAB], and external peer) as well as the current, final document. The
authors also are grateful to Donald Joyner of the Chemical Industry Institute of Toxicology
for graphics support. Lastly, the authors express a special tribute to the late Director of
ECAO—Cincinnati, Dr. Jerry F. Stara—his vision initiated the effort.
xv
-------
p. 16
The following individuals participated as peer reviewers at a publicly held expert review
at the EPA Environmental Research Center in Research Triangle Park, NC, on October 5 and
6, 1987, and provided valuable comments and written contributions on both the workshop
draft and a subsequently revised draft:
Charles Hobbs, D.V.M.
Inhalation Toxicology Research Institute
Lovelace Biomedical and Environmental
Research Institute, Inc.
P.O. Box 5890
Albuquerque, NM 87185
Michael D. Lebowitz, Ph.D.
University of Arizona College of Medicine
Respiratory Sciences Center
1501 North Campbell Avenue
Tucson, AZ 85724
Daniel B. Menzel, Ph.D.*
Laboratory of Environmental Pharmacology
and Toxicology
P.O. Box 3813
Duke University Medical Center
Durham, NC 27710
Richard Schlesinger, Ph.D.
Laboratory for Pulmonary Biology and
Toxicology
Institute of Environmental Medicine
New York University Medical Center
Long Meadow Road
Tuxedo, NY 10987
Vera Fisgrova-Bergerova Thomas, Ph.D.
Department of Anesthesiology
University of Miami School of Medicine
P.O. Box 016370
Miami, FL 33101
Theodore Torkelson, Ph.D.
Toxicology Consultant
315 Birch Street
Roscommon, MI 48653
Curtis Travis, Ph.D.
Center for Risk Management
P.O. Box 2008
Building 4500S
Oak Ridge National Laboratory
Oak Ridge, TN 37831-6109
*Current address:
Department of Community and Environmental
Medicine
University of California—Irvine
Irvine, CA 92717.
XVI
-------
p. 17
On October 26, 1990, an external review draft (1990) of this document was reviewed
by EPA's Science Advisory Board (SAB). The SAB reviewers were:
ACTING CHAIRMAN
Ronald Wyzga, Ph.D.
Electric Power Research Institute
3412 Hillview Avenue
P.O. Box 10412
Palo Alto, CA 94303
MEMBERS AND CONSULTANTS
Mel Andersen, Ph.D.*
Chemical industry Institute of Toxicology
P.O. Box 12137
6 Davis Drive
Research Triangle Park, NC 27709
David Gaylor, Ph.D.
Biometry Division
National Center for Toxicological Research
Jefferson, AR 72079
Marshall Johnson, Ph.D.
Department of Anatomy and Developmental
Biology
Jefferson Medical College
1020 Locust Street
Philadelphia, PA 19107
Fred Miller, Ph.D.1"
Center for Extrapolation Modeling
Duke University Medical Center
Durham, NC 27710
Richard Monson, Ph.D.
Department of Epidemiology
Harvard School of Public Health
677 Huntingdon Ave.
Boston, MA
D. Warner North, Ph.D.
Decision Focus Inc.
650 Castro Street, Suite 300
Mountain View, CA 94041
Guenter Oberdorster, Ph.D.
Department of Environmental Medicine
University of Rochester School of Medicine
Rochester, NY 14642
Martha Radike, Ph.D.
Department of Environmental Health
University of Cincinnati Medical Center
3223 Eden Avenue
Cincinnati, OH 45267
Bernard Weiss, Ph.D.
Department of Environmental Medicine
University of Rochester School of Medicine
Rochester, NY 14642
*Current address:
ICF Kaiser
1 Copley Park, Suite 102
Morrisville, NC 27650
f Current address:
Department of Inhalation Toxicology and
Biomathematical Modeling
Chemical Industry Institute of Toxicology
P.O. Box 12137
6 Davis Drive
Research Triangle Park, NC 27709.
XV11
-------
p. 18
After the 1990 SAB review, the document was revised and an additional peer review
was conducted in August* and September^ 1993 to evaluate the key revisions made in
response to SAB comments. The following experts participated:
Alan Dahl, Ph.D.*1"
Inhalation Toxicology Research Institute
Lovelace Biomedical and Environmental
Research Institute, Inc.
P.O. Box 5890
Albuquerque, NM 87185-5890
Vera Fiserova-Bergerova Thomas, Ph.D.*
Department of Anesthesiology
University of Miami School of Medicine
P.O. Box 016370
Miami, FL 33101
Clay B. Frederick, Ph.D.*1"
Toxicology Department
Rohm and Haas Company
727 Norristown Road
Spring House, PA 19477
Michael D. Lebowitz, Ph.D.*
University of Arizona College of Medicine
Respiratory Sciences Center
1501 North Campbell Avenue
Tucson, AZ 85724
John Morris, Ph.D.*1"
Department of Pharmacology and Toxicology
University of Connecticut
Storrs, CT 06269-2092
Curtis C. Travis, Ph.D.*
Center for Risk Management
P.O. Box 2008 Building 4500S
Oak Ridge National Laboratory
Oak Ridge, TN 37831-6109
James S. Ultman, Ph.D.*1"
Department of Chemical Engineering
Penn State University
106 Fenske Laboratory
University Park, PA 16802
Ron K. Wolff, Ph.D.*
Lily Research Laboratories
P.O. Box 708
Greenfield, IN 46140
xvni
-------
p. 19
LIST OF ACRONYMS AND ABBREVIATIONS
a Airway perimeter
ADI Acceptable daily intake
BEIs Biologic exposure indices
bw Body weight
C0 Initial concentration
Cajv Pulmonary region gas concentration
Ca(x) Gas concentration as a function of x
Cb Blood concentration
Cb/g Gas concentration in equilibrium with blood concentration
Cb/r Concentration of gas in its chemically transformed (reacted) state
Cf Concentration in the fat compartment
C Gas phase concentration in airway lumen
C i Gas-phase concentration at the interface of the gas phase with the surface
liquid/tissue phase
Cj Inhaled concentration
Cj Surface-liquid/tissue phase concentration
CLG Concentration in the lung compartment
Cy Surface-liquid/tissue concentration in equilibrium with the gas phase
Cfo Surface-liquid/tissue concentration at the interface of the gas phase and the
surface-liquid/tissue phase
Cs Imposed concentration
CT/A Concentration of reacted and unreacted gas in arterial blood
CT/V Concentration of reacted and unreacted gas in venous blood
xix
-------
p. 20
LIST OF ACRONYMS AND ABBREVIATIONS (cont'd)
Cz Concentration in the surface-liquid/tissue phase
CA Arterial (unoxygenated) blood concentration (mg/cm3)
CLfat Clearance from the fat compartment
CLLIV Clearance from the liver compartment
CLSYs Clearance from the systemic compartment
CNS Central nervous system
CV Concentration in venous (oxygenated) blood entering gas-exchange (PU)
region
CX(EXH)ET Concentration exiting from extrathoracic region on exhalation
CX(EXH)pu Concentration exiting from pulmonary region on exhalation
CX(EXH)TB Concentration exiting from tracheobronchial region on exhalation
CX(INH)ET Concentration exiting from extrathoracic region on inhalation
CX(INH)TB Concentration exiting from tracheobronchial region on inhalation
D Deposited fraction of mass
£>j Liquid diffusivity
dae Aerodynamic equivalent diameter
dar Aerodynamic resistance diameter
DAF Dosimetric adjustment factor
DNA Deoxyribonucleic acid
dp Particle diameter
dx Differential of axial distance into airway
dy Differential of axial distance into capillary segment
dz Differential of distance into the surface-liquid/tissue phase
xx
-------
p. 21
LIST OF ACRONYMS AND ABBREVIATIONS (cont'd)
ELG Elimination rate in the lung compartment
EMAX Maximum extraction efficiency
Ej Liver extraction efficiency
ER Extrarespiratory (systemic) or remote to respiratory tract
ERV Expiratory reserve volume
ET Extrathoracic respiratory tract region
f Respiratory frequency
F Flux fraction (unitless)
Fr Fractional deposition
PEL Frank-effect level
FEVj Forced expiratory volume at one second
fp Fractional penetration
fpET Fractional penetration through the extrathoracic region
fppu Fractional penetration through the pulmonary region
fpTB Fractional penetration through the tracheobronchial region
FRC Functional residual capacity
FVC Forced vital capacity
GI Gastrointestinal
Hb/g Blood:gas (air) partition coefficient
HEFF Effective partition coefficient
Ht/b Tissue: blood partition coefficient
Ht/ Surface-liquid/tissue:gas (air) partition coefficient
xxi
-------
p. 22
Ha
HEC
1C
iv
K,
K
K
SET
iTB
LG
KM
L
LEL
LOAEL
LOEL
Md
aPU
aTB
LIST OF ACRONYMS AND ABBREVIATIONS (cont'd)
Hatta number
Human equivalent concentration
Inspiratory capacity
Intravenous
Transport coefficient in the gas phase
Overall mass transport coefficient
Overall mass transport coefficient of the extrathoracic region
Overall mass transport coefficient of the pulmonary region
Overall mass transport coefficient of the tracheobronchial region
Transport coefficient in the surface-liquid/tissue phase
Elimination rate from lung compartment
Alveolar membrane diffusion coefficient
Reaction rate constant in the blood or tissue
Michaelis constant
Airway length
Lowest-effect level
Lowest-observed-adverse-effect level
Lowest-observed-effect level
Desorbed mass
Desorbed mass from extrathoracic region to blood
Desorbed mass from pulmonary region to blood
Desorbed mass from tracheobronchial region to blood
xxn
-------
p. 23
M
LET
Mpu
MTB
MF
MMAD
N
Ng
NOAEL
NOEL
OEL
PEL
PU
Qalv
Qb
Or
RGDr
RDDr
RDDR,.
RGDRET
RGDR,.
LIST OF ACRONYMS AND ABBREVIATIONS (cont'd)
Mass flux from extrathoric region to blood
Mass flux from pulmonary region to blood
Mass flux from tracheobronchial region to blood
Modifying factor
Mass median aerodynamic diameter
Overall transport or flux
Flux through the air phase
Flux through the surface liquid-tissue phase
No-observed-adverse-effect level
No-observed-effect level
Occupational exposure level
Permissible exposure level
Pulmonary respiratory tract region
Alveolar ventilation rate
Blood flow rate
Cardiac output
Regional gas dose to respiratory tract region (r)
Regional deposited dose of particles to respiratory tract region (r)
Regional deposited dose ratio of particles for respiratory tract region (r)
Regional gas dose ratio for the extrathoracic region
Regional gas dose ratio for the pulmonary region
Regional gas dose ratio for respiratory tract region (r)
XXlll
-------
p. 24
LIST OF ACRONYMS AND ABBREVIATIONS (cont'd)
RGDRTB Regional gas dose ratio for the tracheobronchial region
RfC Chronic inhalation reference concentration
RNA
RV
SP
SA
SA
ET
SA
TB
SA
PU
TB
TLC
TLV
TWA
UF
URT
V
Vu
'LG
Ribonucleic acid
Residual volume
Blood perfusion surface area
Surface area of unspecified respiratory region
Surface area of the extrathoracic region
Surface area of the tracheobronchial region
Surface area of the pulmonary region
Geometric standard deviation
Time
Time (duration) of exhalation
Tracheobronchial respiratory tract region
Total lung capacity
Threshold limit value
Time-weighted average
Uncertainty factor
Upper respiratory tract
Volumetric flow rate
Capillary blood volume
Minute volume (VT X f)
Lung compartment volume
xxiv
-------
p. 25
LIST OF ACRONYMS AND ABBREVIATIONS (cont'd)
VT Tidal volume
VMAX Maximum velocity of saturable (Michaelis-Menton) metabolism path
x Distance into the airway
Ay Thickness of the surface liquid-tissue layer
z Distance into the surface-liquid/tissue phase
Az Surface-liquid/tissue phase thickness
xxv
-------
p. 26
GLOSSARY
Activity Median Diameter (AMD)
Refers to the median of the distribution of radioactivity, lexicological, or biological
activity with respect to particle size.
Acute Exposure
A one-time or short-term exposure with a duration of less than or equal to 24 h.
Aerodynamic Diameter
Term used to describe particles with common inertia! properties to avoid the complications
associated with the effects of particle size, shape, and physical density.
Aerodynamic Equivalent Diameter (dae)
"Aerodynamic diameter" generally used. The diameter of a unit density sphere
(p = 1 g/cm3) having the same settling velocity (due to gravity) as the particle of interest
ofwhatever shape and density. Refer to Raabe (1976) and Appendix H for discussion.
Aerodynamic (Viscous) Resistance Diameter (d^)
The "Lovelace" definition for aerodynamic diameter. Characteristic expression based on
terms describing a particle in the Stokes' regime. Refer to Raabe (1976) for equation.
Aerosol
All-inclusive term. A suspension of liquid or solid particles in air.
ATPS
Ambient temperature and pressure, saturated (a condition under which a gas volume is
measured).
DTPS
Body temperature and pressure, saturated (a condition under which a gas volume is
measured).
Critical Effect
The first adverse effect, or its known precursor, that occurs as the dose rate increases.
Designation is based on evaluation of overall data base.
Chronic Exposure
Multiple exposures occurring over an extended period of time, or a significant fraction of
the animal's or the individual's lifetime.
Dosimetric Adjustment Factor (DAF)
A multiplicative factor used to adjust observed experimental or epidemiological data to
human equivalent concentration for assumed ambient scenario. See regional gas dose ratio
(RGDR) and regional deposited dose ratio (RDDR).
xxvn
-------
p. 27
Diffusion Diameter
Diameter of a sphere having the same diffusion mobility as the particle in question.
dp < 0.5 /xm.
Expiratory Reserve Volume (ERV)
The maximum volume exhaled from FRC (FRC - RV).
f Respiratory frequency (breaths/min).
Fr
Fraction of inspired particles deposited in respiratory tract region (r).
Functional Residual Capacity (FRC)
The lung volume at the end of tidal expiration (TLC - 1C).
Forced Expiratory Volume (FEVj) at One Second
The volume of air that can be forcibly exhaled during the first second of expiration
following a maximal inspiration.
Forced Vital Capacity (FVC)
The maximal volume of air that can be exhaled as forcibly and rapidly as possible after a
maximal inspiration.
Generation
Refers to the branching pattern of the airways. Each division into a major daughter (larger
in diameter) and minor daughter airway is termed a generation. Numbering begins with
the trachea.
Inhalation Reference Concentration (RfC)
An estimate (with uncertainty spanning perhaps an order of magnitude) of a continuous
inhalation exposure to the human population (including sensitive subgroups) that is likely
to be without an appreciable risk of deleterious noncancer health effects during a lifetime.
The inhalation reference concentration is for continuous inhalation exposures and is
appropriately expressed in units of mg/m3.
Inspiratory Capacity (1C)
The maximum inhaled from FRC (TLC - FRC).
Henry's Law Constant
The law can be expressed in several equivalent forms, a convenient form being:
C = HCj where C and C1 are the gas-(g) and liquid-(l) phase concentrations. The
constant (H) is the ratio at equilibrium of the gas phase concentration to the liquid-phase
concentration of the gas (i.e., moles per liter in air/moles per liter in solution).
Lowest-Effect Level (LEL)
Same as Lowest-Observed-Adverse-Effect Level.
XXVlll
-------
p. 28
Lowest-Observed-Adverse-Effect Level (LOAEL)
The lowest exposure level at which there are statistically and biologically significant
increases in frequency or severity of adverse effects between the exposed population and
its appropriate control group.
Mass Median Aerodynamic Diameter (MMAD)
Mass median of the distribution of mass with respect to aerodynamic diameter. Graphs for
these distributions are constructed by plotting frequency against aerodynamic diameters.
Minute Volume
The volume of air exhaled per minute body temperature and pressure, saturated (BTPS).
Modifying Factor (MF)
An uncertainty factor that is greater than zero and less than or equal to 10; its magnitude
reflects professional judgment regarding scientific uncertainties of the data base or study
design not explicitly treated by the uncertainty factors (e.g., the number of animals tested).
The default value for the MF is 1.
No-Observed-Adverse-Effect Level (NOAEL)
An exposure level at which there are no statistically and biologically significant increases
in the frequency or severity of adverse effects between the exposed population and its
appropriate control. Some effects may be produced at this level, but they are not
considered as adverse, nor immediate precursors to specific adverse effects. In an
experiment with several NOAELs, the assessment focus is primarily on the highest one for
a given critical effect, leading to the common usage of the term NOAEL as the highest
exposure without adverse effect.
Portal-of-Entry Effect
A local effect produced at the tissue or organ of first contact between the biological system
and the toxicant.
Regional Deposited Dose (RDDr)
The deposited dose (mg/cm2 of respiratory tract region surface area per minute) of
particles calculated for the respiratory tract region of interest (r) as related to the observed
toxicity (e.g., calculated for the tracheobronchial region for an adverse effect in the
conducting airways).
Regional Gas Dose (RGDr)
The gas dose (mg/cm2 of respiratory tract surface area per minute) calculated for the
respiratory tract region of interest (r) as related to the observed toxicity (e.g., calculated
for the tracheobronchial region for an adverse effect in the conducting airways).
Regional Deposited Dose Ratio (RDDR^
The ratio of the deposited dose in a respiratory tract region (r) for the laboratory animal
species of interest (RDDA) to that of humans (RDDH). This ratio is used to adjust the
observed particulate exposure effect level for interspecies dosimetric differences.
XXIX
-------
p. 29
Regional Gas Dose Ratio (RGDRj)
The ratio of the deposited gas dose in a respiratory tract region (r) for the laboratory
animal species of interest to that of humans. This ratio is used to adjust the observed gas
exposure level for interspecies dosimetric differences.
Reserve Volume
Volume of air remaining in the lungs after a maximal expiration.
Residual Volume (RV)
The lung volume after maximal expiration (TLC — VC).
Respiratory Bronchiole
Noncartilagenous airway with lumen open along one side to alveoli; when walls are
completely alveolarized it is usually referred to as an alveolar duct. Essentially absent in
rats.
Stokes' Law
The total drag force or resistance of the medium due to fluid motion relative to the particle
is the sum of form and friction drag. When particle motion is described by this equation,
it is said to be in the Stokes regime.
Subchronic Exposure
Multiple or continuous exposures occurring for approximately 10% of an experimental
species lifetime, usually over 3 mo.
Terminal Bronchiole
Noncartilagenous airway that conducts airstream to respiratory bronchiole.
Threshold
The dose or exposure below which a significant adverse effect is not expected.
Carcinogenicity is thought to be a nonthreshold endpoint, thus, no exposure can be
presumed to be without some risk of adverse effect. Noncancer toxic health effects are
presumed to have threshold endpoints, thus, some exposures are presumed to be without
risk of adverse effects.
Tidal Volume
Volume of air inhaled/exhaled during normal breathing.
Total Lung Capacity (TLC)
The lung volume at maximal inspiration.
xxx
-------
p. 30
Uncertainty Factor (UF)
One of several, generally 3- to 10-fold factors, used in operationally deriving the
inhalation reference concentration (RfC) from experimental data. UFs are intended to
account for (1) the variation in sensitivity among the members of the human population,
(2) the uncertainty in extrapolating laboratory animal data to humans, (3) the uncertainty in
extrapolating from data obtained in a study that is of less-than-lifetime exposure, (4) the
uncertainty in using LOAEL data rather than NOAEL data, and (5) the inability of any
single study to adequately address all possible adverse outcomes in humans. The RfC
methods use 3 for the UF for interspecies extrapolation due to the incorporation of default
dosimetric adjustments.
Vital Capacity (VC)
The maximum volume that can be exhaled in a single breath (TLC - RC).
xxxi
-------
p. 31
1. INTRODUCTION AND OVERVIEW
This document describes the U.S. Environmental Protection Agency (EPA)
methodology for estimation of inhalation reference concentrations (RfCs) (earlier terminology
was "inhalation reference dose" or "RfD^1) as benchmark estimates of the quantitative dose-
response assessment of chronic noncancer toxicity for individual inhaled chemicals.
Noncancer toxicity refers to adverse health effects other than cancer and gene mutations.
This overview chapter discusses general principles of dose-response assessment for noncancer
toxicity, the development of the RfC methodology, and its role within the context of the risk
assessment process. Subsequent chapters of the document discuss criteria and information to
be considered in selecting key studies for RfC derivation, provide an overview of the
respiratory system and its intra- and interspecies variables, and discuss areas of uncertainty
and data gaps in relation to the proposed methodology.
1.1 INHALATION REFERENCE CONCENTRATION:
DEVELOPMENT, DEFINITION, AND DERIVATION
The EPA has a history of advocating the evaluation of scientific data and calculation of
Acceptable Daily Intake (ADI) values for noncarcinogens as benchmark values for deriving
regulatory levels to protect exposed populations from adverse effects. For example, the
Office of Pesticide Programs has long used the concept of ADI for tolerance estimates of
pesticides in foodstuffs, the Office of Health and Environmental Assessment (OHEA) has
used ADI values for characterizing levels of pollutants in ambient waters (Federal Register,
1980), and the National Research Council (1977, 1980) has recommended the ADI approach
to characterize levels of pollutants in drinking water with respect to human health.
In 1983, the National Academy of Sciences (NAS) published a report entitled "Risk
Assessment in the Federal Government: Managing the Process" (National Research Council,
1983). The NAS had been charged with evaluating the process of risk assessment as
performed at the federal level in order to determine the "mechanisms to ensure that
government regulation rests on the best available scientific knowledge and to preserve the
1-1
-------
p. 32
integrity of scientific data and judgements" so that controversial decisions regulating chronic
health hazards could be avoided. The NAS recommended that the scientific aspects of risk
assessment should be explicitly separated from the policy aspects of risk management. Risk
assessment, as shown in Figure 1-1, was defined as the characterization of the potential
adverse human health effects of exposures to environmental hazards and consists of the
following four steps: (1) hazard identification: the determination of whether a chemical is or
is not causally linked to a particular health effect; (2) dose-response assessment: the
estimation of the relation between the magnitude of exposure and the occurrence of the health
effects in question; (3) exposure assessment: the determination of the extent of human
exposure; and (4) risk characterization: the description of the nature and often the magnitude
of human risk, including attendant uncertainty.
Research Risk Assessment
Laboratory and field
observations of
adverse health and
ecosystem effects and
exposures to particular
agents
Information on
extrapolation methods
for high to low dose,
for animal to human,
and between sensitive
biota
Field measurements,
estimated exposures,
and characterization of
populations
-
HAZARD
IDENTIFICATION
(Does the agent cause
the adverse effect?)
i
DOSE-RESPONSE
ASSESSMENT
(What is the
relationship between
dose and incidence in
humans or biota?)
EXPOSURE
ASSESSMENT
(What exposures are
currently experienced
or anticipated under
different conditions?)
\
/
RISK
CHARACTERIZATION
(What is the estimated
incidence of the
\ adverse effect in a
/ given population or
ecosystem?)
\
Risk Management
Development of
regulatory options.
1
Evaluation of public
health, economic,
social, and political
consequences of
regulatory options.
/
^^
Agency decisions
and actions.
/
Figure 1-1. National Research Council (1983) framework for risk assessment and risk
management. Key elements of each process are shown.
1-2
-------
p. 33
Following the NAS report, the EPA developed a methodology for evaluating available
data pertaining to xenobiotics for purposes of developing oral reference doses (RfDs) (Barnes
and Dourson, 1988). Although similar to ADIs in intent, RfDs were based upon a more
rigorously defined methodology that adhered to the principles proposed by the NAS and
included guidance on the consistent application of uncertainty factors for prescribed areas of
extrapolation required in the operational derivation. The RfD methodology represents a
quantitative approach to assess toxicity data in order to derive a dose-response estimate.
According to the NAS paradigm, the final step of the risk assessment process, risk
characterization, would involve the comparison of the RfD as a dose-response estimate with
an exposure estimate.
The RfC methodology to estimate benchmark values for noncancer toxicity of inhaled
chemicals significantly departed from the RfD approach. The same general principles were
used, but the RfC methodology was expanded to account for the dynamics of the respiratory
system as the portal of entry. The major difference between the two approaches, therefore, is
that the RfC methodology includes dosimetric adjustments to account for the species-specific
relationships of exposure concentrations to deposited/delivered doses. The physicochemical
characteristics of the inhaled agent are considered as key determinants to its interaction with
the respiratory tract and ultimate disposition. Particles and gases are treated separately, and
the type of toxicity observed (respiratory tract or toxicity remote to the portal-of-entry)
influences the dosimetric adjustment applied.
An inhalation reference concentration (RfC) is defined as an estimate (with
uncertainty spanning perhaps an order of magnitude) of a continuous inhalation
exposure to the human population (including sensitive subgroups) that is likely to be
without appreciable risk of deleterious noncancer health effects during a lifetime.
The derivation of any dose-response1 estimate, such as the RfC, to predict the potential
for noncancer toxicity of a chemical requires evaluation of the data array, defined as the
toxicity profile of adverse effects observed at the different levels tested among the available
Although the strict definitions of "dose", "response", and "effect" are recognized and discussed explicitly in
Section 1.2., the conventions of the NAS paradigm will be used in this document, with the RfC being synonymous
with a "dose-response" assessment. Therefore, in the broader sense, the term "dose" may encompass
administered dose (i.e., exposure concentration), delivered dose, or target tissue dose. Likewise, "response" in
the general sense, is an indication of an adverse influence regardless of whether the data were measured as
quantal, count, continuous, or ordered categorical.
1-3
-------
p. 34
data. A challenging aspect of this evaluation is that across the available data, often different
effects are measured in the same tissue; different endpoints are investigated in some studies;
different species are used in various studies; and each investigation may or may not be
performed at exposure concentrations that coincide with others. The effects measured may or
may not represent different and/or unequivocal degrees of severity or adversity within disease
continuums. The dose-response estimate must represent a synthesis of this entire array of
data. Therefore, the evaluation of this data array and choice of data on which to base the
operational derivation of a dose-response estimate are critical and require somewhat
sophisticated lexicological judgment.
In the simplest terms,2 the RfC derivation begins with the identification of a
no-observed-adverse-effect level (NOAEL) and a lowest-observed-adverse-effect level
(LOAEL), which are determined for the specified adverse effect from the exposure levels of a
given individual study on the various species tested. The NOAEL is the highest level tested
at which the specified adverse effect is not produced and is therefore, by definition, a
subthreshold level (Klaassen, 1986). This NOAEL/LOAEL approach, is also a function of
the exposure levels used in the experimental design or is the function of designating a
specified health effect measure (e.g., 10% incidence of a lesion) in the case of some
alternative modeling approaches, and thus, does not necessarily reflect the "true" biological
threshold.
The RfC methodology requires conversion by dosimetric adjustment of the NOAELs
and LOAELs observed in laboratory animal experiments or in human epidemiological or
occupational studies to human equivalent concentrations (HECs) for ambient exposure
conditions. These conditions are currently assumed to be 24 h/day for a lifetime of 70 years.
The dosimetric conversion to an HEC is necessary before the different adverse effects in the
data array can be evaluated and compared.
Definition of an HEC may be viewed as a naive presumption. However, because the
methodology acknowledges that accurate dose-response relationships depend on the degree to
which state-of-the-art research has achieved understanding and characterization of the
2
As discussed in Appendix A, there are alternative approaches under development aimed at deriving estimates of
exposures that are analogous in intent to the establishment of a NOAEL. The NOAEL/LOAEL approach outlined
here is not intended to discourage alternative or more sophisticated dose-response procedures when sufficient data
are available, but rather to present key issues involved in any approach for the assessment of noncancer toxicity.
1-4
-------
p. 35
exposure-dose-response continuum and will therefore be revised accordingly, it must be
recognized that the definition of HEC is iterative and dynamic as well. That is, the HEC is a
concentration back-extrapolated from an appropriate surrogate internal dose to the extent that
this has been defined.
Although it is preferable to use human studies as the basis for the dose-response
derivation, adequate human data are not always available, often forcing reliance on laboratory
animal data. Presented with data from several animal studies, the risk assessor first seeks to
identify the animal model that is most relevant to humans, based on comparability of
biological effects using the most defensible biological rationale; for instance, by using
comparative metabolic, pharmacokinetic, and pharmacodynamic data. In the absence of a
clearly most relevant species, however, the most sensitive species is used as a matter of
science policy at the EPA. For RfCs, the most sensitive species is designated as the species
that shows the critical adverse effect at an exposure level that, when dosimetrically adjusted,
results in the lowest HEC.
The critical toxic effect used in the dose-response assessment is generally characterized
by the lowest NOAEL.^™ that is also representative of the threshold region (the region
where toxicity is apparent from the available data) for the data array. The objective is to
select a prominent toxic effect that is pertinent to the chemical's key mechanism of action.
This approach is based, in part, on the assumption that if the critical toxic effect is prevented,
then all toxic effects are prevented (see Section 1.2, general principles of dose-response
assessment for noncancer toxicity). The determination of the critical toxic effect from all
effects in the data array requires toxicologic judgment because a chemical may elicit more
than one toxic effect (endpoint) in tests of the same or different exposure duration, even in
one test species. Further, as discussed in Appendix A, the NOAEL and LOAEL obtained
from studies depend on the number of animals or subjects examined and on the spacing of the
exposure levels. The NOAEL|-HECj from an individual study (or studies) that is also
representative of the threshold region for the overall data array is the key datum synthesized
from an evaluation of the dose-response data. Determination of this critical effect represents
the first scientific evaluation required by the RfC dose-response assessment.
The RfC is an estimate that is derived from the NOAELrHECj for the critical effect by
consistent application of uncertainty factors (UFs). The UFs are applied to account for
1-5
-------
p. 36
recognized uncertainties in the extrapolations from the experimental data conditions to an
estimate appropriate to the assumed human scenario. Determination of which UFs to apply
and the magnitude of each represents the second scientific evaluation required by an RfC
dose-response assessment. The standard UFs applied are those for the following
extrapolations (as required): (1) effects in average healthy humans to sensitive humans,
(2) laboratory animal data to humans, (3) studies of subchronic to chronic duration,
(4) a LOAELpjgq to a NOAEL^HECj, and (5) an incomplete to complete data base. The
UFs are generally an order of magnitude, although incorporation of dosimetry adjustments or
other mechanistic data has routinely resulted in the use of reduced UFs for RfCs. The typical
reduced UF is three or one-half Iog10 (i.e., 10'5). The composite UF applied to an RfC will
vary in magnitude depending on the number of extrapolations required. An RfC will not be
derived when use of the data involve greater than four areas of extrapolation. The composite
UF when four factors are used is generally reduced from 10,000 to 3,000 in recognition of
the lack of independence of these factors. An additional modifying factor (MF) may also be
applied when scientific uncertainties in the study chosen for operational derivation are not
explicitly addressed by the standard UFs. For example, an MF might be applied to account
for a statistically minimal or inadequate sample size or for poor exposure characterization.
Thus, notationally, the RfC is defined as
RfC = NOAEL*[HEC] / (UF X MF), (1-1)
where:
NOAEL*rHECj = The NOAEL or analogous effect level obtained with an alternate
approach as described in Appendix A, dosimetrically adjusted to a
human equivalent concentration (HEC);
UF = Uncertainty factor(s) applied to account for the extrapolations required from the
characteristics of the experimental regimen; and
MF = Modifying factor to account for scientific uncertainties in the study chosen as the
basis for the operational derivation.
1-6
-------
p. 37
Confidence levels of high, medium, or low are assigned to the study used in the
operational derivation, to the overall data base, and to the RfC itself. Confidence ascribed to
the RfC estimate is a function of both the confidence in the quality of the study and
confidence in the completeness of the supporting data base together, with the data base
confidence taking precedence over that assigned to the study. High confidence in the RfC is
an indication that the data base included investigation of a comprehensive array of noncancer
toxicity endpoints established from studies of chronic duration in various mammalian species
and that the study (or studies) established an unequivocal NOAEL. Therefore, a high
confidence RfC is not likely to change substantially as more data become available, with the
exception of additional mechanistic data or sophisticated tests that may change the perspective
of the evaluation. Low confidence in an RfC is usually applied to a derivation that is based
on several extrapolations and indicates an estimate that may be especially vulnerable to
change if additional data become available. For some chemicals, the data base is so weak
that the derivation of a low confidence RfC is not possible (see Section 4.1 for minimum, data
base criteria). In such cases, the data base supporting an RfC for a chemical is designated as
"not-verifiable". Upon the availability of new data, this not-verifiable status would be
reevaluated.
It must be emphasized that the RfC as a quantitative dose-response estimate is not
numeric alone. As risk assessments have become a more prevalent basis for decision-making,
their scientific quality and clarity of presentation have gained unprecedented importance
(American Industrial Health Council, 1989). Due to the complexity of many risk
assessments, desirable attributes include the explicit treatment of all relevant information and
the expression of uncertainty in each element (i.e., hazard identification, dose-response
assessment, exposure assessment, risk characterization). Any dose-response assessment, such
as the RfC, has inherent uncertainty and imprecision because the process requires some
subjective scientific judgment, use of default assumptions, and data extrapolations.
A complete dose-response evaluation should include communication of the rationale for data
selection, the strengths and weaknesses of the data base, key assumptions, and resultant
uncertainties (Habicht, 1992; American Industrial Health Council, 1989, 1992;
U.S. Environmental Protection Agency, 1984a). The rationale for the choice of the data
from which the RfC is derived, a discussion of data gaps, and the resultant confidence in the
1-7
-------
p. 38
RfC are all outlined in the summary of the RfC entered on the EPA's Integrated Risk
Information System (IRIS). A discussion and rationale for the UFs used in the RfC
derivation are also provided. This information is an important part of the RfC and must be
considered when evaluating the RfC as s dose^response estimate, in addition to assumptions
and resultant uncertainties inherent in an exposure assessment, when attempting to integrate
the assessments into a risk characterization.
In summary, the RfC methods presented herein were developed based on the NAS 1983
framework and are in keeping with the recent NAS report on science and judgement in risk
assessment (National Research Council, 1994). Default options for derivation of NOAELs
and LOAELs and for dosimetric adjustments of particle or gas exposures are presented.
Principles for modifying and departing from these default options are also provided. The
methods represent the currently available science. Uncertainty factors are utilized that allow
for RfC derivation in the absence of some data, but the UF and confidence statements
explicitly call out prescribed areas of extrapolation hi order to communicate data gaps. For
example, a UF is used to account for intraindividual variability, an area identified by the
NAS as one requiring additional data to more accurately characterize susceptibility of
subpopulations.
1.2 GENERAL PRINCIPLES OF DOSE-RESPONSE ASSESSMENT
FOR NONCANCER TOXICITY
Noncancer toxicity refers to adverse health effects or toxic endpoints, other than cancer
and gene mutations, that are due to the effects of environmental agents on the structure or
function of various organ systems. These effects include those on the tissue where the
chemical enters the body, such as the respiratory tract for inhaled agents, and also effects that
follow absorption and distribution of the toxicant to a site remote to its entry point. Most
chemicals that produce noncancer toxicity do not cause a similar degree of toxicity in all
organs, but usually demonstrate major toxicity to one or two organs. These are referred to as
the target organs of toxicity for that chemical.
Empirical observation generally reveals that as the dose of a toxicant is increased, the
toxic response also increases. "Response", in the context of the RfC methodology discussion
1-8
-------
p. 39
may be the degree or severity of an effect in an individual or the fraction of a population
responding. A distinction is sometimes made between response and effect as different
measurements. Effects are graded and measured; whereas responses are quantal and counted
(O'Flaherty, 1981). The distinction is necessary in order to determine an appropriate
mathematical or statistical model for analysis. For dichotomous responses, model estimates
describe probabilities of events in individuals. These probabilities can also be thought of as
the fraction of a population that will show the response. For continuous effects, models
estimate expected changes in individuals. These expected changes can be expressed as shifts
in population means. For practical and sound conceptual reasons, responses and effects can
be considered to be identical (Klaassen, 1986). That is, in a qualitative sense when'trying to
ascertain if a toxic agent exerts an adverse influence, the distinction is unimportant. It is
recognized that the distinction must be carefully applied when employing mathematical
models to calculate estimates.
The importance of understanding the relationship between concentration (applied dose)
and response has been established in the theory and practice of toxicology and pharmacology.
Dose-response behavior is exemplified by the following types of data: (1) quantal responses
(dichotomous), in which the number of responding individuals in a population increases as a
function of dose (e.g., number of animals with a specified effect at each exposure
concentration); (2) count responses, in which the number of measured events increases as
dose is increased (e.g., number of lesion foci in tissue); (3) dose-graded responses (ordered
categorical), in which the severity of the toxic response within an individual or system
increases with dose (e.g., pathology graded from mild to severe); and (4) continuous
responses, in which changes in a biological parameter (e.g., organ weight, nerve conduction
velocity) vary with dose.
Classic toxicology texts and the NAS framework for risk assessment refer to dose-
response assessment as the process of estimating an expected response at various exposure
levels (i.e., the response at various applied dose levels or exposure concentrations). Because
tissue dose of the putative toxic moiety for a given response is not always proportional to the
applied dose of a compound, emphasis has recently been placed on the need to clearly
distinguish between exposure concentration and dose to critical target tissues. The term
"exposure-dose-response assessment" has been recommended as more accurate and
1-9
-------
p. 40
comprehensive (Andersen et al., 1992). This expression refers not only to the determination
of the quantitative relationship between exposure concentrations and target tissue dose, but
also to the relationship between tissue dose and the observed/expected responses in laboratory
animals and humans.
As shown in Figure 1-2, the process of determining the exposure-dose-response
continuum is achieved by linking the mechanisms or critical biological factors that regulate
the occurrence of a particular process and the nature of the interrelationships among these
factors (Andersen et al., 1992). Although the mechanisms of interaction at the molecular
level are very different from the mechanisms involved at the population level, in each case
they refer to biological determinants that control the responses at the respective level of
organization. This figure illustrates that the exposure-dose-response continuum evolves from
protective to predictive as more information becomes available on mechanisms and toxic
events. Dose-response assessment estimates based on characterization at the first "black box"
level necessarily incorporate large uncertainty factors to ensure that the estimates are
protective in the presence of data gaps, which are often substantial. With each progressive
level, incorporation and integration of mechanistic determinants allow elucidation of the
exposure-dose-response continuum and thus, a more accurate characterization of the
pathogenesis process. Although utilization of these data reduces uncertainty in the
dose-response assessment (thus allowing it to be more predictive in nature), in reality, there
will always be some degree of uncertainty.
As this comprehensive continuum is characterized, mechanistic determinants of chemical
disposition, toxicant-target interactions, and tissue responses are integrated into an overall
model of pathogenesis. The three proposed stages in the continuum between exposure and
response are similar to the previously described division of "pharmacokinetics" versus
"pharmacodynamics". Pharmacokinetics was defined to encompass processes relating
exposure to consequent tissue doses, whereas pharmacodynamics encompassed processes that
determined response to the tissue dose. This comparison to the two traditional areas of
investigation is offered only as a context for the new terminology because any divisions are
artificial and a reflection of the degree of understanding of events in the pathogenesis process.
1-10
-------
p. 41
Protective
Predictive
Chemical
Exposure
Concentration
Exposure
"Dose"
Default
Exposure >•
w Mechanisms j
Disposition Models
Exposure
Toxlcological
Response
Response
Response
Response
Disposition Models Toxicant-Target Models
Disposition Models Toxicant-Target Models Tissue Response Models
Qualitative
Quantitative
Figure 1-2. Schematic characterization of comprehensive exposure-dose-response
continuum and the evolution of protective to predictive dose-response
estimates.
Adapted from Conolly (1990) and Andersen et al. (1992).
Disposition includes deposition, absorption, distribution, metabolism, and elimination of
chemicals. Mathematical models of the mechanistic determinants of the disposition of a
parent compound and/or its metabolites, such as physiologically based pharmacokinetic
(PBPK) or dosimetry models, have been useful in describing the relationships between
exposure concentration and target tissue dose (Overton, 1984; Andersen et al., 1987a).
These disposition models can be linked to other models that address the mechanistic
determinants of the toxicant-target tissue interaction and tissue response, respectively. These
latter models refine the designation of response. The tissue dose is linked to determinants of
target-tissue interaction, (e.g., critical mechanistic events such as cytotoxicity and rebound
1-11
-------
p. 42
cellular proliferation), which, in turn, may then be related via other mechanisms to the
ultimate production of lesions or functional changes that are typically defined as the disease
(pathogenesis) outcome. To the extent that these events are explanatory of the disease
outcome, they can be used to quantitate important nonproportionalities or as replacement
indices of the response function. It is important to emphasize that the integration of the
mechanistic determinants may not necessarily be achieved by linking respective models in a
series (i.e., the output of one model becomes input to the next) but may require simultaneous
solution (e.g., the mechanistic determinants of disposition are dynamically related "moment-
by-moment" to mechanisms of toxicant-target interaction). Eventually, causality of the
critical mechanistic toxic effect can be correlated to the internal toxic moiety as the dose
surrogate, rather than relating the exposure concentration to the "black box" of the organism
within a population. It should also be recognized that the history of toxicology shows that
the discovery of a mechanism of toxicity is often accompanied by the identification of a new
or more refined uncertainty. In spite of such knowledge dynamics, expanding the envelope
of "knowns" clearly improves quantitative dose-response assessment, while creating more
challenges to continue to define unknowns.
Predictive dose-response estimates are desired in order to increase the accuracy of the
estimates and eliminate attendant uncertainties. An advantage to the iterative process of
characterizing the exposure-dose-response continuum is that the models used to describe the
pathogenesis process are dynamic and can be updated by additional data and/or changes in
understanding of the process. As will be seen in later chapters, dosimetry and PBPK models
not only are considered the optimal approach for extrapolation of dose across species, but
also have provided insight on important mechanistic determinants that have been utilized in
the default dosimetry adjustments applied to RfC derivation.
Since the dosimetric adjustments incorporate mechanistic determinants of disposition,
they can be applied, after consideration of underlying assumptions described herein, to
adjustment of other inhalation exposures (e.g., acute exposures) or toxicity (e.g., cancer).
The framework evaluating alternative model structures would also be applicable.
Although RfCs are expressed as exposure concentrations so that units are comparable to
those of exposure assessment estimates, it must be emphasized that the RfC exposure
concentrations are back-extrapolated and based on target tissue dose and/or critical
1-12
-------
p. 43
mechanistic effects, to the extent possible. As more data become available and understanding
of the pathogenesis process changes, changes in the dose-response estimate are anticipated.
Generally, based on understanding homeostatic and adaptive mechanisms, most dose-
response assessment procedures operationally approach noncancer health effects as though
there is an identifiable threshold (both for the individual and for the population) below which
effects are not observable. However, it is recognized that there are inherent difficulties in the
identification of population thresholds (Gaylor, 1985). For example, although each National
Ambient Air Quality Standard (NAAQS) is based on noncancer toxicity, not one is based on
a threshold. This is likely the result of the extensive nature of the data base and the
investigation of the effects in identified sensitive subpopulations that support each of the
NAAQS. That is, the operational identification of a threshold is a function of the available
data and current understanding of the exposure-dose-response continuum, which may be
revised as more information such as data from studies encompassing additional endpoints or
more sensitive indicators of toxicity, such as mechanistic determinants, are developed and
evaluated.
For an individual, the threshold concept presumes that a range of exposures from zero
to some finite value can be tolerated by the organism without adverse effects. As an
example, there could be a large number of cells that perform the same or similar function
whose population must be significantly depleted before an adverse effect is seen. This
threshold will vary from one individual to another, so that there will be a distribution of
thresholds in the population. Because sensitive subpopulations (i.e., those individuals with
low thresholds) are frequently of concern in setting exposure standards, risk-assessment
efforts are aimed at estimating levels at which these sensitive individuals would not be
expected to respond.
The identification of a threshold currently distinguishes approaches for noncancer
toxicity assessment from those for carcinogenic endpoints, which dose-response assessment
procedures typically approach as resulting from nonthreshold processes. However, it should
be noted that as the exposure-dose-response continuum described above is characterized better
for both certain carcinogens and noncarcinogens, knowledge of the mechanistic determinants
may blur this distinction between approaches for noncancer toxicity and carcinogenicity.
As mentioned above, consideration of dosimetry determinants are applicable regardless of
1-13
-------
p. 44
toxicity endpoint. The EPA guidelines for cancer assessment are undergoing revision, and an
issue under review is how to incorporate mechanistic data (Federal Register, 1988a).
1.3 GUIDELINES ON SPECIFIC ENDPOINTS
As mentioned, one of the major challenges to performing dose-response assessment for
noncancer endpoints is that it requires the evaluation of effects measured in a number of
different tissues. Often different endpoints are investigated in different studies, in different
species, and at various concentrations. The effects measured may represent different degrees
of severity or adversity within disease continuums. Individual studies must be evaluated for
their usefulness for quantitative assessment, which will be discussed in Chapter 2. The
available information then must be synthesized into an assessment of the dose-response for
noncancer toxicity based on the entire array of data. The overall data array analysis and
integration of data are a critical aspect of the RfC methodology and are discussed in
Chapter 4 (Section 4.3.7).
In order to promote technical quality and consistency in risk assessment, guidelines have
been developed on how to evaluate toxicity data for cancer and a number of different
noncancer endpoints, how to evaluate mixtures (U.S. Environmental Protection Agency,
1987), and how to perform an exposure assessment (Federal Register, 1992a). Guidelines
have also been promulgated for the evaluation of developmental toxicity (Federal Register,
1991) and proposed for the evaluation of female and male reproductive toxicity (Federal
Register, 1988b,c). Guidelines under development for other noncancer endpoints include
those for neurotoxicity, immunotoxicity, and respiratory tract effects.
The historical and conceptual development of the guidelines and their role in the EPA
have been discussed elsewhere (U.S. Environmental Protection Agency, 1987; Jarabek and
Farland, 1990). Within the context of the RfC methodology, these guidelines present key
considerations and approaches to the evaluation of data within an individual endpoint to arrive
at a dose-response estimate. Therefore, the RfC methodology will look to the guidelines on
individual endpoints for ways to consider the data, organize the data, and conduct a dose-
response assessment. The RfC methodology then provides guidance on how to approach the
1-14
-------
p. 45
synthesis of the resultant dose-response estimate with estimates for other noncancer endpoints
to arrive at an overall dose-response estimate for the data array.
1.4 USE OF THE INHALATION REFERENCE CONCENTRATION IN
THE NATIONAL ACADEMY OF SCIENCES RISK ASSESSMENT
AND RISK MANAGEMENT PARADIGM
As discussed earlier, the 1983 NAS report on risk assessment in the federal government
recommended that the scientific aspects of risk assessment should be explicitly separated from
the policy aspects of risk management. The RfC approach described here represents one
component of the risk assessment process, the dose-response component, and as such must be
compared against an exposure estimate in order to characterize risk. The attendant
uncertainties and default assumptions of the RfC estimate should be evaluated in context with
those of the exposure estimate (e.g., averaging time of the measured exposure, exposure
pattern, particle size) to ascertain whether the two are appropriate to integrate. The explicit
treatment of all such relevant information and resultant uncertainties is a requisite for any
final risk characterization. One of the uncertainties that needs to be considered when
comparing an RfC to an exposure estimate is the "order-of-magnitude" imprecision of the
RfC itself, as stated in the definition of the RfC. From a purely mathematical viewpoint, this
refers to a Iog10 around the RfC (i.e., 3-fold above and below). However, such uncertainty
is not purely mathematical, but rather is an expression of the difficulty in translating a data
base (which is often very limited) into a single number that is thought to represent a relatively
safe exposure. This discussion is not intended to be a complete presentation on the use of
RfCs. Rather, it expresses a few of the issues that require consideration and illustrates that
simplistic comparisons of one dose-response value to one exposure value may be inadequate
to precisely represent risk characterization.
The EPA recognizes that regional, state, and local health protection departments need
uniform and scientifically sound procedures for the calculation of benchmark inhalation dose-
response estimates. The proliferation of diverse risk assessment values for inhalation
exposure and the resulting confusion this has caused attests to the importance of a consistent
approach. It is the intention of the EPA that the RfC approach described will be useful to
many in performing dose-response assessments as one piece of the risk assessment process.
1-15
-------
p. 46
1.5 OCCUPATIONAL EXPOSURE LIMITS VERSUS INHALATION
REFERENCE CONCENTRATIONS
Occupational exposure limit (OEL) is a generic term used to denote a variety of
standards that usually reflect a documented body of toxicological, epidemiological, and
clinical information pertaining to human exposure to airborne contaminants. Due to their
derivation methods, attendant assumptions, and intended application, they represent risk
management values, and this distinction with the RfC as a dose-response estimate must be
emphasized.
Occupational exposure limits have often been chosen by organizations for their risk
management programs because they are available for nearly 700 pollutants. The OELs
include the Occupational Safety and Health Administration Permissible Exposure Limits
(PELs) or full text standards, the National Institute of Occupational Safety and Health
Recommended Standards, and the American Conference of Governmental Industrial
Hygienists (ACGIH) threshold limit values (TLVs). The OELs differ among themselves in
regard to the philosophy of the sponsoring organization, legal mandate, objectives,
assumptions, and evaluation of scientific data. They share the common elements of the
evaluation of effects due to inhalation exposure and the goal of protection of human health.
The OELs are generally time-weighted average concentrations of airborne substances to
which a healthy worker can be exposed during defined work periods and under specific work
conditions throughout a working lifetime, without material impairment of health.
An important underlying assumption of most OELs is a workplace setting in which industrial
hygienists are able to control the environments. Therefore, the OEL can represent, in part, a
risk management decision that considers nonhealth issues such as the technological feasibility
of control measures and analytical detection limits. Some OELs, such as the ACGIH TLV,
also reflect the cost of controlling exposure levels. The appropriateness of some of these
assumptions and extenuating considerations to the application of deriving ambient air levels
for pollution control have been discussed elsewhere (Jarabek and Segal, 1994).
A number of these same assumptions and considerations preclude the use of OELs
directly for the derivation of RfCs. The OELs often are not based on chronic effects and
may differ from RfCs in severity of effect. The OELs further assume intermittent exposure
periods of the workplace, whereas RfCs are set to protect against continuous exposure. The
1-16
-------
p. 47
OELs may not incorporate the most current lexicological information because lexicological
review is not on a regular basis. Also, the unavailability of unpublished corporate
documentation precludes scientific scrutiny of the primary basis for a number of TLVs
(Castleman and Ziem, 1988). The evaluation of toxicity data by agencies deriving OELs may
differ from that of EPA with respect to weight-of-evidence classification, application of UFs,
and other issues. Finally, the use of OELs is established to protect the average healthy
worker (ages 18 to 65 years) against the adverse effects of inhaled pollutants to which they
are exposed only a fraction of a day (i.e., during a typical 8-h work shift). Inhalation
reference concentrations, however, are relevant to those of any age and health status and are
aimed at protecting the most sensitive members of the population, assuming long-term
continuous exposures. Therefore, the EPA does not endorse the use of OELs in deriving
RfCs. The OEL data base should be evaluated along with all other data according to the
methodology for RfC derivation. The biological endpoint, quality and nature of the
underlying data sets, the exposure scenarios, and applicability to highly sensitive
subpopulations are among those factors that must be considered for relevance to
nonoccupational exposures.
An issue paper on OEL values, developed by the Inhalation Technical Panel of EPA's
Risk Assessment Forum, discusses the history, use, and limitations of OELs as surrogates for
ambient exposure RfC values (U.S. Environmental Protection Agency, 1990).
1.6 PRIMARY NATIONAL AMBIENT AIR QUALITY STANDARDS
VERSUS INHALATION REFERENCE CONCENTRATIONS
The Clean Air Act requires that NAAQS be set for any ubiquitous air pollutant that, if
present in the air, may reasonably be anticipated to endanger the public health or welfare and
whose presence in the air results from numerous or diverse mobile or stationary sources.
These so-designated pollutants are called criteria pollutants. Primary standards are designed
to protect public health, and secondary standards are designed to protect public welfare (Code
of Federal Regulations, 1991). The primary NAAQS are solely health-based and designed to
protect the most sensitive group of individuals (but not necessarily the most sensitive
members of that group) against adverse health effects. Therefore, by definition, the primary
1-17
-------
p. 48
NAAQS define allowable pollutant concentrations that can be present in the atmosphere
without causing adverse health effects and represent a complete health risk characterization
according to the NAS risk assessment and risk management paradigm.
This RfC methodology will not be applied to the criteria air pollutants (carbon
monoxide, lead, ozone, nitrogen dioxide, paniculate matter, and sulfur dioxide) due to
legislative requirements in the Clean Air Act and major differences in the health data bases of
these pollutants. Development of NAAQS for the criteria pollutants is governed by
Sections 108 and 109 of the Clean Air Act. The health assessment is described more fully
elsewhere (Padgett and Richmond, 1983) and essentially is a scientific process that undergoes
extensive review by the public and the Clean Air Scientific Advisory Committee of EPA's
Science Advisory Board. The determination of adversity and identification of a NAAQS with
an adequate margin of safety is a decision reserved to the EPA Administrator by the Clean
Air Act. This is profoundly different from an RfC in which the determination of adversity
and uncertainty factors are part of the scientific assessment itself. Furthermore, the criteria
air pollutants have extensive health data bases that enable avoiding many of the simplifying
assumptions and default procedures of the RfC methodology. For additional details, refer to
the Code of Federal Regulations (199la), criteria documents for these chemicals (U.S.
Environmental Protection Agency, 1982a,b,c; 1984b,c; 1986a,b,c,d; 1991; 1992; 1993a,b),
and an overview article describing the NAAQS development process (Padgett and Richmond,
1983).
1.7 STATE-OF-THE-ART APPLICATIONS TO THE DEVELOPMENT
OF THE INHALATION REFERENCE CONCENTRATION
METHODOLOGY
All elements of risk assessment (i.e., hazard identification, dose-response assessment,
exposure assessment, risk characterization) involve some degree of reliance upon assumptions
or extrapolations that substitute for unavailable quantitative information and, by that, impart
varying degrees of uncertainty. Risk assessments ultimately serve as the basis for personal or
governmental risk management decisions on safeguarding health and have consequential
economic impacts. As the state-of-the-art of health risk science progresses, the accuracy of
risk assessments will be improved, insofar as these advancements are incorporated into risk
1-18
-------
p. 49
assessment procedures. This makes it imperative that, as scientific advancements in related
disciplines such as biologically motivated extrapolation modeling are made, they are
appropriately incorporated into the elements of the risk assessment process. The RfC
methodology, as a set of procedures to estimate a dose-response assessment, has inherent
uncertainty and imprecision because the process requires some subjective scientific judgment,
use of default assumptions, and data extrapolations. Therefore, OHEA, Office of Research
and Development, has committed to a regular reevaluation of the scientific advancements in
the field and will continue to make recommendations for significant improvements in the
methodology. Modifications are anticipated on approximately a 2-year basis or as
appropriate. If research advancements having a striking impact on the methodology were to
occur earlier or slightly later, the timing of the process may be altered.
In summary, one objective of the RfC methodology is that it always be scientifically
based, and thus, the methodology should be considered dynamic. Pertinent issues and their
solutions will be incorporated as identified and reviewed for applicability on a continuing
basis. These actions will make the methodology sufficiently reliable to serve as one of the
key bases for decisions on protecting the public health.
1-19
-------
p. 50
2. QUALITATIVE EVALUATION OF THE DATA BASE
This chapter outlines considerations for the collection and qualitative evaluation of
diverse data into a cohesive toxicity profile that then can be evaluated by means of the
quantitative procedures for dose-response analysis provided in Chapter 4. The conceptual
basis for the dosimetry adjustments applied to inhaled agents and other considerations specific
to this administration route are addressed in Chapter 3.
The aim of the inhalation reference concentration (RfC) methodology is to establish a
relationship between a particular agent in the air and a specific health effect (or effects).
To define such a relationship, evidence must be collected from diverse sources and
synthesized into an overall judgment of health hazard (Hackney and Linn, 1979). One of the
major challenges to performing dose-response assessment for noncancer endpoints is that it
requires the evaluation of effects measured in a number of different tissues. Often different
endpoints are investigated in different studies, in different species, and at various
concentrations. The effects measured may represent different degrees of severity (adversity)
within disease continuums. Qualitative evaluation of the data base, also known as the hazard
identification component of risk assessment, involves integrating a diverse array of data into a
cohesive, biologically plausible toxicity "picture" or weight-of-the-evidence relationship to
establish that the agent causes an effect (or effects) and is of potential human hazard.
Questions addressed by this process include whether the agent associated with an effect is
responsible for the effect, if the effect is biologically significant, and what the potential
public health implications might be. Answering such questions requires ascertaining the
validity and meaning of the toxicity data, determining whether the experimental results as a
whole suggest or show causality between the agent and the effect, and evaluating whether or
not the causal relationship is applicable under other sets of circumstances (e.g., in
extrapolating from test animals to humans). This entails consideration of all relevant human
and laboratory animal data of various study types, studies with differing results (e.g., positive
and negative), pharmacokinetic disposition data (deposition, absorption, distribution,
metabolism, elimination) mechanistic information, and structure-activity relationships. This
process integrates information needed for the dose-response assessment, which is discussed in
2-1
-------
p. 51
Chapter 4. Thus, qualitative evaluation of a diverse data base necessitates a systematic
approach for obtaining agreement on the validity, selection, and interpretation of studies to be
used in the quantitative methodological procedures of the dose-response assessment.
2.1 GUIDELINES FOR SELECTIONS OF KEY STUDIES
Key studies are those that contribute most significantly to the weight of evidence as to
whether or not a particular chemical is potentially hazardous in humans (Barnes and Dourson,
1988). These studies are of two types: (1) epidemiologic, clinical, or case reports on
humans and (2) experimental studies on animals. Each has unique considerations that will be
addressed separately here. However, whenever the data base permits, the most robust
qualitative evaluation typically involves an integrated interpretation of human and animal
data, taking advantage of the unique strengths of each. Once the key studies demonstrating
the critical toxic effect have been identified, the selection of effect level and the RfC
derivation arises from an objective scientific evaluation of the data array available on the
chemical as described in Chapter 4.
2.1.1 Human Data
Utilization of human data avoids the necessity of extrapolating from laboratory animals
to humans, thereby decreasing uncertainty in the risk assessment. Human data have often
been useful in developing oral reference doses (RfDs) (Barnes and Dourson, 1988). There
are significantly more human data on inhalation than on ingestion exposures, however, so that
criteria for evaluating studies and their results need to be stated explicitly, particularly if they
are to be used in a quantitative fashion. Since 1977, when the Clean Air Act identified goals
related to air quality and health, the task of clarifying how population studies can be used for
determining scientifically reasonable standards and how to define an adverse respiratory
health effect has been rigorously debated (Lebowitz, 1983; American Thoracic Society, 1985;
National Research Council, 1985). Many of the results from these efforts can be applied as
guidance for the RfC methodology.
Three types of human studies are most often utilized to obtain data pertinent to
understanding the risk of chemicals to humans in order to protect public health:
2-2
-------
p. 52
(1) epidemiologic studies, (2) qlinical studies or controlled exposure experiments, and
(3) case reports. In addition, recent advances in molecular epidemiology and physiologically
based pharmacokinetic (PBPK) simulation modeling provide other types of data useful to
evaluating and synthesizing data from these three types of human studies along with
laboratory animal data. When using these studies for risk assessment, several factors are
important in evaluating their quality and in determining the level of certainty associated with
their use. The factors that are most relevant to developing chronic RfCs from human data
relate to biomarkers and epidemiologic studies, which are discussed more fully below.
Clinical studies are typically of acute or short durations and therefore, as such, are less useful
as the basis of an RfC, but can be useful in the development of dosimetric data relevant to
biomarkers.
2.1.1.1 Molecular Epidemiology and Biologic Markers
In the early 1980s, the concept of "molecular epidemiology" was developed to describe
an evolving approach to research that attempts to synthesize advanced laboratory methods
with analytical epidemiology (Perera and Weinstein, 1982). Although originally defined for
cancer, molecular epidemiology can encompass any disease outcome and can provide
important insights and understanding of a wide variety of critical issues in current risk
assessment (Hattis, 1986). The approach is based on the combination of two biologic tenets:
(1) early biologic effects from a toxic exposure are far more prevalent in the population at
risk than the late events of direct (historical) interest such as disease, and may sometimes be
more specific to the exposure than the outcome itself; and (2) given technological advances,
most xenobiotics can either be directly quantified in the body or indirectly measured by
identification of some predictable, dose-related biologic response (Cullen, 1989). Thus, once
(prevalent, early) "markers" of effect and (accurate) "markers" of dose can be developed in
the laboratory, human epidemiology could, with appropriate research, proceed without its
prior methodologic constraints; relative risks are high because the events studied are either
very common among the exposed (i.e., sensitive markers) or very rare among the unexposed
(i.e., specific markers); exposures can be precisely classified by direct measurement and the
lapsed time between first human exposure and an opportunity for study is foreshortened
because endpoints are, by definition, "early" (Cullen, 1989).
2-3
-------
p. 53
Biologic markers are not new. Markers such as blood lead, mercury levels in hair, and
urinary metabolites or liver function assays after solvent exposure have long been used in
health research and practice to indicate exposures to or to predict effects of these compounds.
As defined by the National Research Council (NRC) Board on Environmental Studies and
Toxicology, a "biologic marker" is any cellular or molecular indicator of toxic exposure,
adverse health effects, or susceptibility (National Research Council, 1987). The markers may
represent signals—generally biochemical, molecular, genetic, immunologic, or physiologic—
in a continuum of events between a causal exposure and resultant disease as shown in
Figure 2-1.
Exposure
Effect
\ \
\ \
Susceptibility
Figure 2-1. Biological marker components in sequential progression between exposure
and disease.
Source: Schulte (1989).
2-4
-------
p. 54
The distinguishing aspect of this paradigm vis-a"-vis the previous use of biological
markers is that current technological advances and developments in basic sciences allow for
detection of smaller signals at diverse points in the continuum. Thus, the historical analytic
epidemiology approach for estimating risks by relating exposure to clinical disease (morbidity
and mortality) may be supplemented by a fuller method, one that identifies intervening
relationships more precisely or with greater detail than in the past. As a result, health events
are less likely to be viewed as dichotomous phenomena (presence or absence of disease) but
rather as a series of changes in a continuum from homeostatic adaptation, through
dysfunction, to disease and death (Schulte, 1987, 1989; National Research Council, 1991b).
Significant side benefits of this research modality include: (1) an improvement in the
accuracy of exposure variables; (2) a contribution to the understanding of underlying
pathogenic mechanisms inherent in the study of events at the molecular, cellular, or tissue
levels; (3) the potential for more accurate and etiologic classifications of environmental
diseases; and (4) the possibility that recognition of early effects could prompt strategies for
secondary prevention or early disease modification (Hulka and Wilcosky, 1988). Quantitative
consideration of the events in the exposure-dose-disease continuum has implications for dose-
response assessment and could provide insight on how to extrapolate from high to low
exposure levels, the reliability of extrapolation from laboratory species to humans, the
relevance of certain physiologic events to disease outcome, and an index of human
interindividual variation.
The progression from exposure to disease as shown in Figure 2-1 has been characterized
by a number of authors and scientific committees on the use of biomarkers (Perera, 1987;
Schulte, 1989; National Research Council, 1987, 1991a,b). It should be pointed out that
components in the progression shown in Figure 2-1 are not necessarily discrete or the only
events in the continuum. There may be a series of other components (steps or stages)
between or in parallel with these that have yet to be discovered (Schulte, 1989). The
similarity of this paradigm to that presented in Figure 1-2, as proposed by laboratory
toxicologists, is striking and emphasizes the interdisciplinary and collaborative nature that will
be required of future research on disease etiology and of associating causality to events along
the continuum for use in dose-response assessment. Due to the anticipated impact that
biological markers will have on future epidemiologic research and the potential for use of
2-5
-------
p. 55
such data in health risk assessment, this section will discuss the evolving concepts and
definitions of biological markers and provide a framework for their validation and use in
dose-response assessment. Methodologic issues and their effect on research design will be
discussed in subsequent sections on the use of epidemiologic and nonepidemiologic data.
It should be noted that many of these considerations are the same for any bioassay, as the
level of sensitivity of the measured effect moves from the macro (e.g., histopathology) to
molecular (e.g., receptor binding) level.
Concepts and Definitions
Because it is important that risk assessors understand the purpose of a given marker,
that is, the reason the marker is being considered and what aspect of the exposure-dose-
disease ("response") association it is supposed to indicate, markers are often classified into
three broad categories: markers of exposure, disease, or susceptibility. It must be
emphasized that this classification depends on the state of knowledge concerning the
mechanistic relationship between the marker and the conditions of exposure, disease, or
susceptibility that the markers represent. Thus, allocation of markers to one or more of three
categories is subjective and could change (National Research Council, 199Ib).
External exposure is defined as the sum amount of the xenobiotic material presented to
an organism, whereas internal dose is the amount actually absorbed into the organism.
An effect is defined as: (1) an actual health impairment or (by general consensus) recognized
disease, (2) an early precursor of a disease process that indicates a potential for impairment
of health, or (3) an event peripheral to any disease process but correlated with it and
therefore predictive of development of impaired health. An intrinsic genetic or other
characteristic or a preexisting disease that results in an increase in the internal dose, the
biologically effective dose, or the target tissue response can be markers of increased
susceptibility (National Research Council, 1987).
As shown in Figure 2-1, along the progression from exposure in the environment to the
development of clinical disease, four generic component classes of biologic markers can be
delineated: (1) indices of the internal dose, (2) indices of the biologically effective dose,
(3) early biologic effects, and (4) altered structure and function. Clinical disease can also be
represented by biologic markers for the current disease as well as by markers for prognostic
2-6
-------
p. 56
significance. Internal dose is the amount of xenobiotic substance found in a biologic
medium; the biologically effective dose is the amount of xenobiotic material that interacts
with critical subcellular, cellular, and tissue targets or with an established surrogate target
tissue. A marker of early biologic effect represents an event that is correlated with, and
possibly predictive of, health impairment. Altered structure and function are precursor
biologic changes more closely related to the development of disease. Markers of clinical
disease and of prognostic significance show the presence and predict the future of developed
disease, respectively. Markers of susceptibility are indicators of increased (or decreased) risk
for any component in the continuum. Even before exposure occurs, there may be biological
differences between humans that cause some individuals to be more susceptible to
environmentally induced disease (National Research Council, 1987,199la,b).
A marker may be: (1) an actual measure of an event, such as blood lead to indicate
exposure; (2) a surrogate for an event, such as creatinine clearance for renal function;
(3) a correlate of an event, such as DNA adducts to reflect organ-specific exposure; or
(4) a risk predictor, such as human lymphocyte antigen (HLA) B27 for ankylosing spondylitis
(Schulte, 1989). Therefore, biological markers are tools that can be used to provide greater
resolution of aspects of exposure-disease associations, that is, to clarify the relationship, if
any, between exposure to a xenobiotic compound and health impairment.
Framework for Validation and Use
Although the development and use of biologic markers is increasing at a rapid rate, the
validity and meaning of many of the markers need to be established before they can be used
as analogous to "exposure" or "disease" in classical epidemiologic research and prior to their
use in quantitative dose-response assessment. The key to relating variables in the exposure-
dose-disease continuum and to validation is agreement on what constitutes a "critical effect".
A critical effect is the biologic marker deemed most representative of a particular component
in the continuum and ultimately most pathognomonic (Schulte, 1989). There is a need to
have general agreement on which of these are critical (i.e., indicating some aspect of a
disease response) and which are merely adaptive. This usually requires a series of
independent studies, primarily toxicologic, and then clinical and epidemiologic, as delineated
in Table 2-1. Knowledge of these steps can be useful in evaluating data that may characterize
2-7
-------
p. 57
TABLE 2-1. STEPS IN THE DEVELOPMENT OF A BIOMARKER
Step
Action Required
Relative
Importance3
1. Chemical
Selection
2. Conceptualization
3. Confirmation of
Concept
4. Develop Method
of Measurement
5. Biomarker Practical
for Field?
6. Establish Dose-
Response
Relationship
7. Identify Variables
Affecting Relation-
ship with Dose
8. Measures Toxic
Effect?
9. Validation of
Applicability to
Humans
10. Conduct
Demonstration
Study
Prioritize based on occurrence, significant C
human exposure, potential for adverse
human health effects.
Identify logical consequence of chemical C
exposure that might serve as a useful
measure of exposure.
Experimentally confirm the validity of the C
basic concept.
Identify method for reliably detecting C
changes in biomarker at doses at or below
those producing toxic effects.
Develop feasible field methodology and L
develop sufficient sensitivity of biomarker to
monitor existing exposures.
Characterize pharmacokinetics and C,L
metabolism of chemical. (Consistent
relationship to systemic dose is critical;
knowledge of effective dose is limiting.)
Establish specificity of response and identify C,L
lifestyle, genetic, disease state, therapeutic,
or occupational variables that modify the
response.
Identify advantages of this biomarker among N
other biomarkers of equal efficacy as
measures of exposure.
Conduct pilot study in small groups of C
humans with defined exposure gradients to
the chemical of interest.
Determine whether variation in response in C
larger population can be accounted for by
known variables.
"C = Critical to the application of the biomarker; L = Limiting to the application of the biomarker (i.e., places
limits on interpretation of results for secondary purposes) (e.g., risk assessment); N = Nice to have, but not
essential to the application of the biomarker.
Source: Adapted from Bull (1989).
2-8
-------
p. 58
biomarkers as surrogates for dose or disease to determine dose-response relationships.
As more causal component associations are identified, it becomes necessary to elucidate
quantitative relationships of the kinetics, natural history, and rates of transition along the
continuum. The hypothesis of the role that the marker has in the disease development should
sustain throughout these refinements. Subsequently, it is necessary to relate critical effects to
dose estimates, to determine what factors affect dose, and to define a no-observed-adverse-
effect level (NOAEL).
Reliability and Validation
Because biological markers are measurements, they have inherent signals (true effects)
and noise (random errors). Measurement errors need to be acknowledged and controlled
since failure to do so may lead to a decreased sensitivity due to the lack of reliability in the
measurements, which may lead to systematic biases or correlations toward underestimation, a
need for increased sample size, and bias selection in case-control studies. It is recommended
that a pilot reliability study be performed as standard practice.
The validity of a biologic marker can be viewed in terms of "measurement validity" as
used in epidemiology (Schulte, 1989; National Research Council, 1991b). Three aspects of
validity have been defined: (1) construct validity (i.e., the ability to correspond to theoretical
constructs under study [e.g., if some event such as kidney function changes with age, then a
marker with construct validity should also change]), (2) content validity (i.e., the domain of
the phenomenon under study is incorporated [e.g., a DNA adduct for aromatic amines will
represent exposure from various routes and from occupational and lifestyle exposures]), and
(3) criterion validity (i.e., the extent to which the marker correlates with an external measure
of the phenomenon under study). There are two types of criterion validity: concurrent
validity and predictive validity. Concurrent validity is when the marker and the criterion
refer to the same point in time (e.g., exhaled breath measures could be validated against
ambient air measures of occupational exposure to a chemical). Predictive validity indicates
the ability of a marker to predict a criterion (e.g., detection of a marker can be validated
against the appearance of an effect).
It is necessary to have precise, accurate, sensitive, specific, and reliable assays for each
component estimate and an understanding of the factors that influence them (Schulte, 1987;
2-9
-------
p. 59
Griffith et al., 1988). A validated relationship between the various components along the
exposure-dose-disease continuum (Figure 2-1) would include knowledge established at four
levels (Gann, 1986): (1) the association between a marker and a preceding exposure or
subsequent effect; (2) the location, shape, and slope of the exposure marker, or of the
marker-effect relationship; (3) the threshold of "no observed adverse effect"; and (4) the
positive predictive value of the marker for exposure or for disease. The validity may be
assessed in terms of sensitivity, specificity, disease frequency, and predictive value. The
relationship between these parameters and ways to calculate them are provided in detail
elsewhere (Schulte, 1989; Khoury et al., 1985; Griffith et al., 1988). A qualitative rating
scale for the validity of biologic markers is provided in Table 2-2.
TABLE 2-2. QUALITATIVE RATING FOR VALIDITY OF BIOLOGIC MARKERS
(1) "Totally experimental", with complete uncertainty about health or exposure
significance of results.
(2) Experimental, but theoretical reasons exist to suggest that the marker will correlate
with exposure or disease.
(3) Correlates well with exposure or disease, but significance of the data is still
uncertain.
(4) Probably correlates well with exposure or disease, but truly conclusive data are not
available.
(5) Extensively studied and has been validated as a useful tool for monitoring exposure
or disease, but gives an unexpected positive response in 10% of people screened.
(6) Extensively studied and has been validated as a useful tool for monitoring exposure
or disease, but gives an unexpected negative response in 10% of people screened
who have a history of chronic abnormal exposure.
(7) Extensively studied and has been validated as a useful tool for monitoring exposure
or disease, with no or very rare false positives and negatives.
(8) Validated and is completely predictive of exposure or disease.
Source: Schulte (1989).
Conceptually, the goal of validation is to explore and establish links between markers
along the exposure-dose-disease continuum. The conventional approach to validation is to
2-10
-------
p. 60
relate a critical effect to exposure or dose, or to toxic effects. It has also been suggested that
validation of biologic markers include testing the association for one component of the
continuum and any other critical component elsewhere in the continuum (Schulte, 1989), as
shown in Figure 2-2. This approach is consistent with the iterative process of research and
the steps in development of biologic markers, as discussed. The risk assessor should consider
the degree to which these criteria have been addressed for a biomarker when considering its
application to dose response assessment. Hattis (1991) offers guidance on and provides
examples of how to incorporate biomarkers and pharmacokinetic analysis into risk
assessment.
E = Exposure
ID = Internal dose
BED = Biologically effective dose
EBE = Early biological effect
ASF - Altered structure/function
CD = Clinical disease
PS = Prognostic significance
Figure 2-2. Schematic representation of possible relationships (1 to 21 pairs) to research
using biologic markers.
Source: Schulte (1989).
2-11
-------
p. 61
Analytic Issues
The conventional techniques for assessing exposure-disease associations, for screening
for disease in populations, and for handling multiple variables can be practiced for any two or
more components in the continuum. The major assumption that permits this approach is that
there is an association between the component markers. Figure 2-2 shows the 21 possible
pairwise relationships that may be evaluated along the continuum between exposure and
disease. The ability to characterize these relationships is dependent on the degree of
mechanistic knowledge, whereas the importance of each of these will vary depending on the
priorities and objectives of the investigators and/or the application to dose-response
assessment.
Essentially, at issue is whether the marker is truly an intervening variable or a
confounding factor. Any marker that represents a step in the causal progression between
exposure and disease is not a confounding factor but, in fact, is an intervening variable.
When there is uncertainty about the mechanism, handling a potential confounding factor as
both confounding and not confounding in different analyses is justified. Seasoned judgment
of the best available information in the face of lack of mechanistic data will be required.
Relationships between components in the continuum can be modeled by two approaches:
empirical and process modeling. The empirical approach can be used when there are no
explicit hypotheses about components. The approach is to use statistical techniques to find
the combination of descriptors that "best" explain the observed effects (e.g., gauging the
relative appropriateness of different dose surrogates determined principally by the nature of
the pathogenesis process) (Schulte, 1989). For use in dose-response assessment, it is also
necessary to determine the extent that a marker reflects recent or past exposures, peak as
opposed to integrated exposures, and cumulative rather than noncumulative biologic effects
(Checkoway and Rice, 1992). The process modeling approach uses quantitative toxicologic
models to estimate concentrations in biological compartments and temporal patterns of
occurrence. It requires explicit hypotheses. Process modeling should be the goal as more is
learned about the continuum.
2-12
-------
p. 62
Biologic Exposure Indices
Perhaps the one area where use of biologic markers has achieved the most success as
applied to dose estimation is in setting biologic exposure indices (BEI) based on occupational
epidemiology and experimental studies. Figure 2-3 shows the relationship between air
monitoring and biologic monitoring as practiced for risk management of occupational
exposures. Air monitoring and its related threshold limit value (TLV), usually expressed as a
time-weighted average (TWA), is a measure of external dose, whereas biological monitoring
and the associated BEI relates to indirect monitoring of the internal dose (Droz, 1985). Air
monitoring as often conducted, however, does not reflect unexpected exposure resulting from
peculiarities of certain jobs or from poor working practices (Fiserova-Bergerova, 1990), so
that surveillance of workers by monitoring BEIs is recommended (American Conference of
Governmental Industrial Hygienists, 1986).
TLV BEI
External ^ Internal
Dose Dose
Effects
Figure 2-3. Schematic relationships between threshold limit values in air (TLV),
biologic exposure indices (BEI), and effects.
Source: Droz (1985).
2-13
-------
p. 63
In order to develop and set a BEI, the relationship between internal dose (i.e., the BEI)
and health effects should be established. However, most of the available toxicologic data
relate exposure dose directly to health effects. In order to make use of these data, approaches
to development of the BEIs recommended by the ACGIH have considered that the BEIs are
bioequivalent to the TLV (Droz, 1985). A similar type of reasoning can be used to establish
NOAELs or lowest-observed-adverse-effect levels (LOAELs) associated with occupational
epidemiology exposures. Exposure estimates such as a TWA (or other exposure measure
[e.g., duration or cumulative exposure]) are a measure of the composition of the external
environment surrounding a worker. The BEI is a measure of an internal dose farther along
the exposure-dose-disease continuum, and as such can better reflect individual exposure
variability and response. Therefore, appropriate BEI levels can serve as dose surrogates,
associated with an observed effect in a population (e.g., lower confidence limit on mean
metabolite in blood) then extrapolated back to exposure estimates in order to calculate a
human equivalent concentration (HEC).
The correlation between the degree of exposure and biological levels is influenced by
variability in the exposure concentration (temporal repetition, intraday concentration
variation, and interday concentration variation) and individual variability (workload, body
build, and metabolism). The relationships between exposure levels and BEIs can be
established using three main approaches: (1) epidemiologic field studies on groups of
workers or populations exposed to the chemical in question; (2) experimental or clinical
studies on volunteers exposed in controlled chambers; and (3) PBPK simulation studies, using
different kinds of mathematical models to allow the simulation of various exposure situations
and individual characteristics (Droz, 1985; Fiserova-Bergerova, 1990). These three
approaches are complementary and each has its own advantages and disadvantages, as
qualitatively summarized in Table 2-3. The ranking of these factors depend heavily on
experimental design and could be quite different for a particular chemical or set of studies.
The BEI documentation for individual chemicals should be consulted for considerations
pertaining to these modifying factors and their influence on interpretation of results
(American Conference of Governmental Industrial Hygienists, 1986).
2-14
-------
p. 64
TABLE 2-3. COMPARISON OF THE QUALITIES OF FIELD AND
EXPERIMENTAL APPROACHES IN THE STUDY OF THRESHOLD
LIMIT VALUE/BIOLOGIC EXPOSURE INDICES RELATIONSfflPS
Approach
Factor
Field
Experimental
Exposure (dose) measurement
Physical workload characterization
Timing of biological sampling
Effects of exposure repetition
Environmental variability
Representativity of the subjects
+ + + = Good; + + = Medium; + = Poor.
Source: Droz (1985).
Application of Physiologically Based Pharmacofdnetic Models
Physiologically based pharmacokinetic models are simulation models described by
simultaneous differential equations, the number of which is dictated by the number of
compartments needed to describe the physiological and metabolic processes involved. In the
context of characterizing the exposure-dose-disease continuum, simulation models can be
considered as complementary, providing critical insight on key processes related to the fate of
chemicals in the body and for depicting the contribution of various exposure and biological
factors to the variability of response. That is, these models can provide the following
information on which biological monitoring (e.g., BEIs) is designed and data are interpreted:
(1) concentration-effect relationships, (2) time-effect relationships, (3) matching exposure in
the workplace with integrated exposure, (4) depicting effects of external and internal factors
that alter the relationship between intensity of exposure and biological concentration and body
burden of the biologic marker, (5) extrapolation and prediction of biological concentrations
resulting from exposure to new compounds or new exposure conditions, and (6) verification
of data (Leung, 1992; Fiserova-Bergerova, 1990; Leung and Paustenbach, 1988; Droz,
1985). Simulation models, because of their ability to match the extent of exposures
associated with the predetermined dose or biological markers of exposure, are a valuable tool
2-15
-------
p. 65
in extrapolation of reference values for workers with unusual workshifts (Andersen et al.,
1987b; Saltzman, 1988).
2.1.1.2 Epidemiologic Data
There are essentially three areas of concern in assessing the quality of an epidemiologic
study. These involve the design and methodological approaches used for: (1) exposure
measures, (2) effect measures, and (3) the control of covariables and confounding variables
(Lebowitz, 1983). The study population and study design must adequately address the health
effect in question in order to support a risk assessment (Lebowitz, 1983). In order to
accomplish this goal, the exposure measures must be appropriate and of sufficient quality; the
statistical analysis methods must be suitable to the study design and goals; the health effect
measures must be reliable and valid; and the covariables and confounding variables need to
be controlled or eliminated. Additional guidance on evaluation of the quality of individual
epidemiologic studies is provided in Appendix B. Criteria for causal significance are
provided in Appendix C.
Assessment of Exposure Measures
The problem of the accuracy and relevance of exposure measurements is not unique to
epidemiologic investigations, but it can be exacerbated due to the long-term nature of these
studies. For example, the nature of aerometric data may change over time because of
different air sampling techniques. Exposures also change over time because of different
industrial hygiene practices and because individuals change jobs and residences. Accurate
documentation of air toxicant levels, therefore, is critical in determining the usefulness of an
investigation as well as documentation that the analysis of the air toxicant is appropriate and
of sufficient sensitivity. It also is advisable to have the concentrations of other pollutants
reported and considered in the statistical analyses to help rule out confounding or interactive
effects. The number, location, and timing of monitors should be suitable to allow an
appropriate determination of exposure of the subjects to the pollutant being studied and to the
pollutants that could confound the results. When appropriate, the exposure measure or
estimate should take into account indoor/outdoor exposures and activity and subject location
data. Unfortunately, exposure measures often are the weakest component of an
2-16
-------
p. 66
epidemiologic study. Minimally, the exposure measure or estimate needs to be representative
of the actual exposure.
Assessment of exposure measures should attempt to establish whether the following
wide range of aspects (National Research Council, 199la) were addressed:
• Contaminant and potential biological response
• Specification and selection of the target population
• Spatial and temporal variability of concentration distribution patterns
• Frequency and intensity of exposure
• Selection of the sampling period in appropriate relationship to the time scale of
biological effect (e.g., peak exposure versus TWA; short-term versus lifetime)
• Precision and accuracy requirements.
Exposure measures employed can either be direct (e.g., personal monitoring and in some
cases biological markers) or indirect (e.g., environmental monitoring such as area samples,
models that predict spatial and temporal concentration distributions of air contaminants in
microenvironments, questionnaires, and questionnaires or diaries). Each type has distinct
advantages and disadvantages, and depending on the nature of the agent in question, may
address the above aspects to greater or lesser degrees.
Assessment of Effect Measures
Effect measures refer to the methods used to define disease indices. For epidemiologic
studies, these include incidence, standardized mortality ratios, and relative risk ratios.
Criteria for assessment require the proper selection and characterization of both the
exposed and control groups. For example, criteria for inclusion in the control category of a
case-control study must ensure that this group has no exposure to the agent of concern. For
studies without internal control groups, reference populations are needed, particularly when
evaluating spirometric data (Ferris, 1978; American Thoracic Society, 1979; Crapo et al.,
1981; Knudson et al., 1976). Each population used to predict "normal" pulmonary function
tests has its own characteristics, which should be considered when used for comparisons.
2-17
-------
p. 67
Other considerations include the adequacy of study duration and quality of the follow-up.
A disease with a long latency before clinical presentation requires a longer study duration
than one with an acute onset. Valid ascertainment (such as verification according to the
International Classification of Diseases IX) of the causes of morbidity and death also is
necessary.
Evaluation of epidemiologic studies may require interpretation of a variety of subjective
health effects data. Questionnaire responses may be biased by the way questions are worded,
the training of an interviewer, or the setting. However, a study based on a high-quality
questionnaire can provide useful results. For example, a committee of the American
Thoracic Society (ATS) charged with defining an adverse respiratory health effect, has come
to a consensus that "in general, increased prevalence of chronic respiratory symptoms as
determined from questionnaire surveys should be considered to be an adverse health effect"
(American Thoracic Society, 1985). Questionnaires should be validated as part of the
investigation protocol, unless a standard questionnaire that has previously been validated is
used (Medical Research Council, 1960; Ferris, 1978; National Institute for Occupational
Safety and Health, 1986).
It is very important to consider differences between statistical significance and medical
or biological significance. Both the variability of an outcome measure and the magnitude of
an exposure's effect determine the level of statistical significance. For example, data from a
large study population analyzed with sophisticated techniques may yield statistically
significant effects of small magnitude that cannot readily be interpreted biologically.
Conversely, apparently large changes of clinical importance may not be statistically
significant if the study population is too small. In addition, some studies present false
negative or no-effect results due to the lack of power. Judgments concerning medical or
biological significance should be based on the magnitude and class of a particular effect. For
example, cough or phlegm production can be considered less important than effects resulting
in hospital admissions, but daily productive cough can be more important than infrequent
cough. Underlying assumptions and nuances of the statistical procedures applied to the data
also need to be considered. This will probably best be accomplished on a case-by-case basis.
Because the RfC considers both portal-of-entry and remote (systemic) effects, it would
be helpful to define an "adverse respiratory health effect." An ATS committee published
2-18
-------
p. 68
guidelines that defined such an effect as medically significant physiologic or pathologic
changes generally evidenced by one or more of the following (American Thoracic Society,
1985):
• Interference with the normal activity of the affected person or persons
• Episodic respiratory illness
• Incapacitating illness
• Permanent respiratory injury or
• Progressive respiratory dysfunction
Appendix D provides detailed descriptions of adverse respiratory effects in humans.
Assessing the Control of Confounding and Covariables
Epidemiologic investigations attempt to relate an exposure to a given health effect, but
this includes accounting for the "background" health effect (pathologic condition) that exists
in individuals due to predisposing factors and preexisting health conditions, or from other
variables, such as occupational exposures.
Various host factors contribute as risk factors for disease and can influence the health
indices assessed. For example, asthmatics may be particularly susceptible to effects from
exposure to irritant gases. Epidemiologic evaluation of these factors often not only accounts
for such interactions but also can help to characterize susceptible or sensitive groups.
Covariables can be as important as the major aerometric variables themselves in affecting
human health. Other exposures, such as concomitant occupational exposures and smoking, in
particular, can affect the disease outcome. Meteorologic variables such as air velocity,
temperature, and humidity also are very important factors when considering respiratory health
effects. These Covariables should be controlled by both the study design and analysis, as
appropriate.
The final step in the inferential process from an epidemiologic investigation is the
extension of the study results to persons, populations, or settings not specifically included in
the experimental design, that is, to demonstrate consistency of results within replicates in
2-19
-------
p. 69
different scenarios. The confidence with which this is done for positive results is usually
based implicitly on how successful the investigators have been in identifying and handling the
potential risk factors and covariables that produce or influence the pollution-effect association
they have observed. Uncertainties also arise because the general population includes some
people, such as children, who may be more susceptible than people in the epidemiologic
study. Factors such as the "healthy worker" effect and the bias of a predominantly male
worker sample must be considered when using occupational studies (National Research
Council, 1985). Intraindividual variability concerns are addressed in Section 2.1.1.4.
2.1.1.3 Nonepidemiologic Data
Human data also include clinical studies and case reports. The case reports may
provide support for the weight-of-the-evidence decision, but are often of limited utility in
establishing a quantitative relationship between environmental exposures and anticipated
effects (Barnes and Dourson, 1988). Controlled human clinical studies, properly conducted,
can be of great value to dose-response assessment. Although such studies for ethical reasons
are typically for acute durations and therefore, by definition, do not meet the criteria for
development of a chronic RfC estimate, they can be valuable in improving understanding of
the nature of the effect in humans. Some of the discussion found in Section 2.1.2.2, Impact
of Experimental Protocol (for laboratory animal studies), is also appropriate to consider.
Clinical Studies
Clinical studies may contain exposure-response information that can be used in
estimating effects. Most clinical studies combine the strong point of animal toxicology,
rigorous control of the experimental exposure and subject, with the strong point of
epidemiology, the unquestioned relevance to human health. In addition, clinical studies can
be independently confirmed somewhat more easily (requiring a reasonably short time and
resource commitment) than epidemiologic studies. There are limitations, however, that
include short exposure duration and "noninvasive" techniques that might not ascertain the full
array of effects. The test atmospheres are usually within the range expected to produce only
mild and temporary health effects. Certainly, clinical studies should be recognized and given
credence to the extent that they are scientifically rigorous, relevant to human health concerns,
2-20
-------
p. 70
and have been independently replicated. They may be particularly useful for acute or less-
than-lifetime dose-response assessment. The prediction of long-term effects from short-term
observations remains questionable, but confidence in clinical findings can be bolstered by
supporting evidence from epidemiology and laboratory animal toxicology, and vice versa.
Although clinical exposures and respiratory measurements (at least the noninvasive ones
for functional mechanics) are typically done on nonsedated humans, the breathing pattern
remains an important consideration. Experimental protocol often dictates the breathing
pattern (i.e., nonspontaneous breathing) where a subject patterns his or her breathing to a
metronome or is instructed to take a deep breath on every fifth inhalation. Because the
efficiency of time-dependent deposition mechanisms is greater during inspiration than
expiration, an ideal "academic" breathing pattern would keep the inspiration time/expiration
time ratio (tj/tg) constant (Heyder et al., 1975). Relevance of such an academic pattern to
risk assessment, however, remains equivocal and most investigations do not attempt to
maintain a constant ratio. Documentation of breathing patterns should be included in the
experimental protocol and considered in the extrapolation of dose.
The exposure mode is also important to consider. Because the nasal passages are more
efficient at removing particles (particularly for large particles) than the oral cavity, increased
lung deposition of larger particles could occur through mouth breathing. This would affect
both the amount and the size distribution of an inhaled aerosol in the lower respiratory tract.
Even the specific configuration of the mouthpieces used in inhalation exposures delivered
orally can affect the extent of deposition (Schlesinger, 1985). Miller et al. (1988) showed
that regional respiratory tract deposition of insoluble particles in humans is a complex
function of breathing route, ventilatory level, and the paniculate physicochemical and
aerodynamic properties. Some gases (especially highly water soluable and reactive ones) are
extensively removed in the nasal passages, making exposure mode important for gases as
well. Whether the subjects were free-breathing or whether they breathed through a
mouthpiece or used a facemask affects gas deposition as well and should be considered.
Case Reports
Individual case reports of adverse effects due to a specific agent also can provide some
help in evaluating the potential risk from exposure to a toxic air pollutant. These reports are
2-21
-------
p. 71
especially valuable qualitatively for indicating that the quantitative effect observed in animals
occurs in exposed humans. These reports must be examined carefully and used with
discretion because they represent a very small sample and are usually related to heavy
exposures (Goldstein, 1983). Nevertheless, these observations should not be overlooked,
especially when a large number of case histories exist with the same endpoint.
2.1.1.4 Intraspecies Variability and Identifying Sensitive Subgroups
In order to control factors other than the chemical being tested, laboratory animals
(e.g., rodents) used in toxicity studies are often bred for homogeneity. In contrast, the
human population is heterogeneous. The broad genetic variation of the human population in
processes related to chemical disposition and tissue response causes individual differences in
sensitivity to toxic chemicals. A susceptible individual is one who will experience an adverse
health effect to a pollutant significantly earlier in the course of exposure or at lower doses
than the average individual, because of host factors that predispose the individual to the
harmful effects. Sensitive individuals may be those whose genetic makeup puts them at the
extreme end of a continuous distribution of a biological function, such as the amount of
enzyme production, or those who possess a unique genetic difference, such as an altered
enzyme, that makes them markedly different from the general population.
In addition to genetic factors, personal characteristics such as age, sex, health status,
nutrition or personal habits make some people more susceptible (Calabrese, 1978). The
activity pattern of people is a major host factor influencing the dose-response by its effect on
delivered dose. Generally, exercise increases the delivered dose and alters the regional
deposition of the dose.
Environmental risk assessment also should consider host factors that both increase
susceptibility and that occur relatively frequently in the population. Erdreich and Sonich-
Mullin (1984) estimated the prevalence of population subgroups who are potentially
hypersusceptible to some common pollutants. Table 2-4 shows five subgroups of individuals
who, based on empirical observations or compromised physiological functions, are assumed
susceptible to the listed chemicals. Theoretically, elderly individuals could be more
susceptible to some chemicals and children to others. Unfortunately, very little is known
about this important area. Likewise, very little is known about gender differences.
2-22
-------
p. 72
TABLE 2-4. PREVALENCE OF SUBGROUPS SUSCEPTIBLE TO EFFECTS
OF COMMON POLLUTANTS
Susceptibility
Subgroup
Population
Prevalence
Chemicals*'*
Reference
Embryo, fetus,
neonate
Young children
Chronic obstructive
pulmonary disease
Circulatory
conditions
Liver disease
Pregnant
women:
21/l,000b
Ages 1-4:
70/1,000b
Carcinogens,
solvents, CO,
mercury, lead,
PCBs, pesticides
Hepatotoxins, PCBs,
metals, NO2
Chronic bronchitis: O3, Cd, paniculate
13,494,000 (5.4%)c matter, SO2, NO2
Asthma: 12,375,000
(4.9 %)c
Emphysema:
1,915,000 (0.8%)c
Rice (1981), Kurzel and
Cetrulo (1981), Saxena
et al. (1981), U.S.
Environmental Protection
Agency (1986a, 1991)
Calabrese (1981), Friberg
et al. (1979), U.S.
Environmental Protection
Agency (1993a)
Holland et al. (1979),
Redmond (1981), U.S.
Environmental Protection
Agency (1982b; 1993a,b)
Ischemic heart
disease: 8,155,000
(3.2%)c
Liver abnormalities:
20/1,000d
Chlorinated solvents, McCauley and Bull
fluorocarbons, CO (1980), Aviado (1978),
U.S. Environmental
Protection Agency (1991)
Carbon tetrachloride, Calabrese (1978)
PCBs, insecticides,
carcinogens
'Abbreviations:
CO = Carbon monoxide;
PCBs = Polychlorinated biphenyls;
O3 = Ozone;
Cd = Cadmium;
SO2 = Sulfur dioxide;
NO2 = Nitrogen dioxide.
"Representative samples of chemicals to which these individuals may be susceptible. Some evidence from
laboratory animal studies only.
""Estimates of Erdreich and Sonich-Mullin (1984) from 1970 census statistics data.
"Population base 251,448,000; estimate from U.S. Department of Health and Human Services (1992).
dEstimate of Erdreich and Sonich-Mullin (1984) from Health Interview Survey (National Center for Health
Statistics, 1975).
Source: Adapted from Erdreich and Sonich-Mullin (1984).
As a result of epidemiologic investigations, it is well recognized that a population of
adult workers experiences less morbidity and mortality than the general population (Fox and
Collier, 1976; Wen et al., 1983; Monson, 1986). However, sufficient qualitative and
2-23
-------
p. 73
quantitative information on interindividual variability and susceptibility for specific chemicals
rarely exists.
If the RfC is based on data derived from subgroups of the general population, such as
workers who are generally a selected group of healthy adults, the calculation procedures must
include an appropriate uncertainty factor (UF) to account for the anticipated broader
variability in the general population. Worker populations are nonrepresentative in terms of
sex, age distribution, and general health status. Susceptible subpopulations may not be
represented because they may not seek or sustain employment, particularly in situations such
as those represented in workplace exposure studies. Occasionally, data are available on more
sensitive subgroups such as children or asthmatics. In these cases, dose-response assessments
can be made for the general population with greater confidence. In the absence of data on
the more susceptible individuals in the population or lack of identification of such individuals,
UFs are used to protect unidentified individuals at greater risk.
There are two steps necessary to obtain information addressing the problem of sensitive
individuals: (1) examine chemical-specific data for empirical evidence of sensitivity and
hypersusceptibility, and (2) ascertain whether the mechanism of toxicity for a given chemical
suggests that any population group would be more sensitive.
In addition to this chemical-specific evaluation, guidance should be developed
concerning the prevalence of sensitive subgroups and the range of sensitivities in the general
population exposed to inhaled toxicants. Some research has assessed the magnitude of
interindividual variability in pharmacokinetic parameters related to the delivery of the
biologically effective dose, in order to develop guidance for appropriate UFs. Differences
among normal healthy adults may be as much as 10-fold (Hattis et al., 1987). Therefore, the
potential that exists for broad differences when children, the elderly, the ill, and those
previously exposed are included.
2.1.1.5 Summary
Based on the foregoing discussion, guidelines for the qualitative assessment of human
data are as follows:
2-24
-------
p. 74
Evaluation of the Epidemiologic Data Base
• Examine epidemiologic and clinical data for dose-response information in potential or
previously identified sensitive groups (e.g., studies in asthmatics and children).
• Examine laboratory animal data for models that may help identify potential sensitive
individuals.
• Evaluate epidemiologic studies to ascertain genetic and personal factors that increase
the risk of adverse response. Evaluate implications of these risk factors for
identifying sensitive groups.
• Examine data for reports of ranges of responses or response variables, and for
information on individual responses. This is particularly important in evaluating
human data for assessing the range of variability in response because epidemiologic
studies may find a LOAEL with no NOAEL.
• Evaluate available biological monitoring data and clinical and experimental data for
indications of characteristics of increased susceptibility. For example, irritants may
induce responses earlier in individuals with asthma.
• Evaluate data on mechanisms of toxicity, pharmacokinetics, and critical target organs
to identify characteristics that may imply broad interindividual variability or
susceptible individuals. For example, the elderly may be more sensitive to certain
chemicals in relation to age-related changes in oxidative metabolism potential.
Evaluation of Individual Studies
• Assess the makeup of the study population and control groups to identify the presence
or absence of sensitive individuals. Data on healthy workers, for example, are not
representative of the general population and will require reduction of NOAELS or
LOAELs by UFs.
• Consider the activity pattern of the subjects. Whether the subjects received exposure
while at rest or at level(s) of exercise that influenced the inhaled dose as well as the
pattern of deposition.
• In longitudinal (cohort) studies, evaluate information in relation to the natural history
of the disease (e.g., the progression of lesions). For example, normal changes over
time, such as increased forced expiratory volume at 1 s (FEV^) as children get older,
and decline of FEVj with aging in older adults, should not be adversely affected.
Cross-sectional studies may suggest such associations but will not support causality as
strongly as will cohort studies.
• For parameters that have known variability with age, such as FEVj, evaluate results
within age groups and ascertain whether appropriate reference populations were used.
2-25
-------
p. 75
2.1.2 Laboratory Animal Data
When the data base lacks adequate information on effects in humans, as is frequently
the case, the key studies are drawn from experiments conducted on nonhuman mammals.
Animals most often used include the rat, mouse, guinea pig, hamster, rabbit, monkey, and
dog. Such animal studies have often been conducted with controlled exposure conditions on
relatively homogenous populations, but nevertheless, present the risk assessor with concerns
about evaluating dose and exposure regimen. Unlike the human, inbred laboratory animals
have homogeneous constitutions. Genetic background differences and numerous inbred, have
homogeneous constitutions. Genetic background differences and numerous other interspecies
differences are confounding factors during key study selection.
Evaluation of the quality of individual animal toxicity studies requires consideration of
factors associated with the study's hypothesis, design, execution, analysis, and interpretation.
Guidelines for assessing individual animal studies are provided in Appendix F and are
adopted from a number of recommendations (National Research Council, 1984; Society of
Toxicology, 1982; James, 1985; Muller et al., 1984; Lu, 1985a). Refer to this appendix for
a more detailed description of those issues.
2.1.2.1 Study Design
An ideal study addresses a clearly defined hypothesis, follows a carefully prescribed
protocol, is conducted in adherence to good laboratory practice, and includes appropriate and
sufficient subsequent analysis to support its conclusions. The EPA Good Laboratory Practice
Standards (Code of Federal Regulations, 1991b,c) are designed to ensure the quality and
integrity of data used in hazard evaluation. These regulations contain detailed guidance on
provisions for personnel, facilities for animal care, animal supply, handling of test and
control substances, equipment, operation of testing facilities, characterization of test and
control chemicals, protocol and conduct of a laboratory study, report records, record storage,
and record retrieval. Studies that do not precisely follow these guidelines may still be judged
adequate if, in the context of overall results, the deviations are not important. The type of
deviation (from the guidelines) and its magnitude, as well as the potential for its interaction
among all the variables, must be assessed (National Research Council, 1984). For example,
a study may still be judged adequate, despite an insufficient number of test animals specified
2-26
-------
p. 76
by the appropriate reference protocol guidelines, if the results are so definitive that the
addition of more test animals would almost certainly not have affected the conclusion.
A dose-response assessment that is based on a study with deficiencies may include a
modifying factor to account for the added uncertainty (see Section 4.3.8.2).
The use of statistics in design and interpretation of studies is an area in animal toxicity
testing that is often neglected or applied inappropriately (Muller et al., 1984). Consideration
of statistical applications restricted to confirmatory analysis (i.e., outcome is dependent on the
mathematically randomized test condition and is independent of other observations) versus
exploratory analysis (i.e, many tests on a variable) should be emphasized.
2.1.2.2 Impact of Experimental Protocol
The techniques and measurements used in inhalation toxicology investigations may
affect the exposure conditions or the interpretation of toxic effects, thereby altering the results
used for risk assessment. Areas that introduce uncertainty into interspecies extrapolations of
inhaled dose include measurement techniques, the definitions and underlying assumptions
used in the procedures, and the exposure technology. Careful consideration should be given
to each when estimating the effective inhaled dose. This discussion is also appropriate to
consider when evaluating clinical human studies.
Equipment Specifications
The equipment used will impart restrictions on any interpretation (i.e., limitations of
sensitivity for exposure analysis or to monitor an effect) of investigative results and therefore
should be considered when evaluating test results.
Generation and Characterization of Exposures
Just as the working definitions and underlying assumptions alter the interpretation of
measurement techniques, the operative exposure level (e.g., for use in risk assessment,
prediction models, etc.) of a test agent is a function of how its particulate mass and
composition (mean particle diameter and distribution) and gas concentration are expressed.
Other specific characteristics (e.g., adequate test substance mixing in chamber,
hygroscopicity, charge density) should be accounted for as part of this description. The
2-27
-------
p. 77
soundness and interpretation of the animal data are dependent on the methods employed to
generate and analyze the test atmosphere data because the methods influence deposition
calculations.
The two most common ways in which particle size is expressed are the count median
diameter (CMD) and mass median diameter (MMD). The toxicity of a material is most
consistently related to its mass distribution. Measurement of mass has the further advantage
of a minor quantitative error at the small end of the size spectrum. To assess risk, however,
the activity diameter may be a more appropriate expression of particle size as discussed in
Appendix H. Methods of particle measurement include settling, filtration, wet and dry
impingement, multiple impaction, electrical precipitation, thermal precipitation,
centrifugation, and observation of optical effects. Each of these has its own principle of
operation and limits of sensitivity that, in turn, affect the expression or characterization of the
test aerosol. Fiber exposures are further complicated by the need to describe the aspect
criteria and distributions. As discussed in the section on anatomy and physiology, certain
mechanisms contribute to the deposition fraction in each respiratory region. Failure to
account for characteristics such as hygroscopicity or charge density when generating an
aerosol could change its deposition in certain regions. This variability in the aerosol
characterization would be expressed as uncertainty in the dose-response assessment.
Gaseous contaminant atmospheres are usually somewhat easier to characterize.
Delivered concentrations must be consistent across exposure location and duration and may be
less than the generated concentration. If the gas is extremely reactive, loss due to reactions
with the walls of the transport system (e.g., tubing) and chamber will occur. Losses due to
decomposition or alteration of the test substance during some generation procedures also may
be a factor. Gas flow rate (delivery) must be known, steady, and calibrated for the given gas
because it is density-dependent. Analysis of the air is limited by the detection device
specifications. If online analysis is not feasible, consideration should be given to the
frequency of samples taken. The period between samples for intermittent analysis should be
less than one-tenth of the total exposure time for any given day (McKenna, 1982).
For all generation and characterization of pollutants, periodic calibration of all
measurement systems is a critical quality control/quality assurance step. This also needs to
be considered when evaluating the study, as discussed in Appendix F.
2-28
-------
p. 78
Generation of the compound under study and subsequent exposure also will affect the
derived inhaled dose. Exact determination of the dose achieved in inhalation studies is a
complex process. Proper generation, appropriate characterization, and accurate delivery of
the test atmosphere are integral to this determination. Varieties and limitations of the
available technology must be considered when evaluating the selection of methods and
interpreting experimental results. The reader is referred to review articles for details on
inhalation exposure systems (Cheng and Moss, 1988; Barrow, 1988; Moss and Cheng, 1988;
Gardner and Kennedy, 1993).
Exposure Regimen
Extrapolation from one exposure regimen to another has uncertainties, most of which
are not quantified. For most chemicals, the quantitative relationship between the toxic effect
and concentration or duration of exposure is not studied. Some studies have indicated that
the relationship is dependent on many factors, including (1) the number of exposure hours per
day; (2) the exposure scenario, that is, continuous versus interrupted (e.g., 1 week of
exposure, 1 week of air, 1 week of exposure, etc.), versus intermittent (X hours per day,
Y days per week) regimens; (3) the time of endpoint assessment (e.g., acute versus
subchronic versus chronic studies or studies with recovery time before observation); (4) the
endpoint(s); and (5) the mechanisms of toxicity. Examples of particles and gases follow that
illustrate some of the complexities involved in extrapolating across exposure scenarios.
The actual amount of particles or gas found in the respiratory tract at any time is
determined by the relative rates of deposition and clearance. The efficiencies of the
deposition mechanisms are different in each respiratory tract region. The defense
mechanisms and clearance rates for each of these regions also are different. Therefore, it is
expected that the kinetics of the toxic effect of an exposure will be influenced by the duration
of exposure. There is experimental evidence for such a differential dependence of effect on
exposure duration. For example, Albert et al. (1971) showed that low single doses or early
effects of repeated exposure to cigarette smoke were associated with acceleration of clearance
rates in the tracheobronchial trees of both donkeys and humans. Heavier doses and long-term
repeated exposures were associated with sporadic clearance, stasis intervals, and some
retrograde movement. Unfortunately, there has not been a systematic comparison and
2-29
-------
p. 79
quantification of differential clearance rates across species. This will be necessary before the
effects of duration can be assessed in the same models or default values can be developed.
Ozone can be used as an illustration for gases because it has a large health effects data
base. Kenoyer et al. (1981) showed that rats exposed to O3 for 4 h showed delays in the
early clearance and an acceleration in the late clearance rate of tracer particles. These
investigators postulated that the delays in early clearance could be caused by effects that
decrease mucous transport (e.g., decreased ciliary beat rate or change in mucous properties),
whereas acceleration of the late clearance rate was most likely due to an increase in numbers
or activities of alveolar macrophages. Rats exposed intermittently (7 to 8 h/day to O3 for
approximately 1 week) had similar changes in lung antioxidant enzymes to animals exposed
continuously (24 h/day), even though the dose, expressed as the product of concentration (C)
and time (T) of exposure, was different (Mustafa and Lee, 1976). Monkeys exposed to
O3 for 18 mo continuously or for 18 mo bimonthly (equivalent to 9 mo of exposure) had
some similar alterations in lung morphology; additional alterations were observed in the
intermittent exposure group although they received a lower C x T (Tyler et al., 1985).
Using morphometric measurements of the proximal alveolar region of lungs of rats receiving
prolonged low-level exposures of O3, Huang et al. (1988) have shown that the increase in the
relative volume of Type I epithelial cells was related to the C x T, whereas other
morphometric indices were more dependent on concentration than on time.
For nitrogen dioxide (NO^, the data base is equally complex on the exposure scenario
issue. Using the mouse infectivity model (an index of antibacterial lung defenses),
concentration was found to be more important than duration of exposure in causing the effect
(Gardner et al., 1979). When a typical urban pattern of NO2 was used (i.e., a baseline of
continuous exposure to a low level of NO2 on which were superimposed two 1-h peaks of
NO2 each weekday), the study indicated that on a C XT basis, this regimen was not more
toxic than a continuous exposure to the baseline level after a short period of exposure
(Graham et al., 1987). After a chronic exposure, the spikes to the baseline increased the
effects relative to the baseline exposure only (Miller et al., 1987a).
The topic of extrapolating across different exposure scenarios is beyond the scope of
this document. However, the few examples provided illustrate the complexity of the issue
with respect to concentration and duration. Other factors that also influence interspecies
2-30
-------
p. 80
extrapolation (e.g., temperature, humidity, particle size, and distribution are discussed) in
Chapter 3. Risk assessors will have to consider the effects of exposure on a case-by-case
basis and utilize default assumptions until the needed research data are available.
Exposure Modes
The various exposure techniques can be divided according to the extent to which the test
species are exposed. The techniques range from whole-body exposure at the one extreme to
exposures limited only to the lower respiratory tract (Lippmann, 1980). These techniques
include whole-body, head-only, nose-only, nasal, oral, and trachea! cannula exposures, and
trachea! instillations. Practical considerations such as economic feasibility, special
precautions for safe and efficient generation, amount of material, test compound stability,
exposure duration, and the measurements desired dictate the selection of an exposure
technique for a given study design. For example, whole-body exposure of laboratory animals
in cages is the most common method to conduct chronic inhalation exposures for more than
1 to 2 h/day, whereas nose-only exposures are most often used for short durations particle
exposures.
Wolff et al. (1982) studied the deposition and retention of 0.1 ^m radiolabeled gallium
oxide (67Ga2O3) aggregate aerosols in Fischer 344 rats following whole-body and nose-only
exposures of 3 days duration. In this investigation, lung deposition for whole-body exposures
was similar to that for nose-only exposures (~ 15% of the inhaled particles). Due to
preening, passage of material into the GI tract, however, was 1.6-fold greater for whole-body
exposures than with nose-only exposures. This could be important in cases where there is
either a specific GI response (i.e., stomach lesions) or substantial GI absorption that may
result in a systemic effect.
Rotation of animals in whole-body chambers is recommended and should be included in
the experimental design (Griffis et al., 1981) to minimize dosimetric differences that would
result if the aerosol was not uniformly distributed in the chamber. The effects of factors such
as heat and/or other stress upon animals in confinement tubes used for nose- or head-only
exposures need to be considered, particularly because these factors may be species-dependent.
For example, rats in confinement tubes for short exposures have been shown to have
respiratory values and body temperatures that remain constant, although Syrian golden
2-31
-------
p. 81
hamsters exhibit increasing ventilation and temperature (Raabe et al., 1973). Adaptation to
exposure or measurements may be a function of behavior, such as ability to be trained
(Mauderly and Kritchevsky, 1979), but in general, animals in confinement tubes or animals
forced to breathe through mouthpieces will experience abnormal stress (Raabe et al., 1973).
Nose-only restraint was shown to induce indications of material toxicity but did not appear to
affect normal embryo/fetal morphologic development in mice exposed on gestational days
6 through 15 for 6 h per day (Tyl et al., 1994). The potential for stress should be accounted
for in the experimental protocol. The tubes can be modified into plethysmographs to monitor
respiratory function changes indicative of stress, or cooled to a constant temperature to
prevent it. If such modifications are not made, the risk assessor must be aware of potential
influences on results.
Anesthesia
Anesthesia greatly influences the respiration characteristics of the test animal. This is a
consideration when evaluating pulmonary function parameters for adverse effects. Prolonged
anesthesia can compromise the respiratory system, altering normal function and response.
Anesthesia also can alter the metabolism of the study compound. Anesthesia has been
reported to interfere with autonomic control, produce atelectasis, decrease lung compliance,
block reflex responses, and introduce an undesirable risk to animals committed to long-term
toxicology studies (Dorato et al., 1983). These alterations in ventilation and breathing
mechanics produced by anesthesia could have severe effects on the results of respiratory
function measurements. This possibility provided the impetus for the development of
procedures for measuring respiration in unsedated laboratory animals (Amdur and Mead,
1958; Mauderly et al., 1979). Data now are available on respiratory characteristics in
sedated and unsedated animals; consideration of anesthesia should be included in data analysis
to ensure appropriate comparisons.
Breathing Pattern
Consideration should be given to the possible alteration of the breathing pattern due to
the exposure concentration, which, in turn, would alter the delivered dose. Exposure of
certain agents, such as irritants, may lead to concentration-dependent changes in pulmonary
2-32
-------
p. 82
mechanics measurements (Costa and Tepper, 1988; Alarie, 1981). Correct quantification of
inhaled dose therefore may require measurement of breathing pattern (respiratory frequency
and VT) during the course of the exposure. Differences in delivered "dose" correlated with
the species-dependent differences in ventilation have been reported for formaldehyde toxicity
(Chang etal., 1983).
Measurement Techniques
Because measurements of ventilation and breathing mechanics often are used to evaluate
respiratory functional alterations or to estimate inhaled/retained dose, performance parameters
of such measurements are critical to their interpretation. The patterns of respiration
(breathing route, depth, and rate) affect the air flow characteristics, which, in turn, influence
the relationship between competing particle deposition mechanisms and the relative
contribution of gas transport processes. The penetration depth of the exposure air is
determined by the tidal volume (VT), the airway caliber, and the ratio of functional residual
capacity to total lung capacity (FRC/TLC). As the FRC/TLC increases, deposition would be
expected to increase (Schlesinger, 1985). For example, rapid shallow breathing often is
associated with increased deposition of larger particles in the upper respiratory tract, as
compared to slow, deep breathing. Therefore, performance parameters include both the
factors that influence the test species (including human) respiration characteristics and the
performance limitations of the techniques.
Pharmacologic Effects of Agents
The test agents may affect lung ventilation and function. Administration of a chemical
with narcotic properties will lower physical activity, whereas an irritant might increase
movement. The test agent could also alter clearance mechanisms. All of these states would
affect deposition, uptake, and retention of the dose. In addition, the agent could disrupt the
immune system and render the animal more susceptible to disease during long-term testing,
thereby altering the study results.
There are several examples of irritating or potentially anesthetic chemicals that can
depress ventilation. Chang et al. (1983) reported a 40% decrease in minute volume in mice
exposed to 15 ppm formaldehyde. This inhibition was maintained during the entire course of
2-33
-------
p. 83
the daily exposure period. Ventilation was decreased to as little as 1/15 of resting values
during exposure of mice to 10 ppm ozone (O3), and to as little as 1/3 of resting values during
exposure of mice to acrylate esters (Bruce et al., 1979).
Particle overloading in the lungs of laboratory animals is a recognized outcome of
excessive particle exposures, especially during chronic inhalation studies. The phenomenon
has been associated both with protracted retention time of particles in the lung and with
changes that can confound lexicological interpretations (Morrow, 1992). Concurrent and
persistent features of the progressive prolongation of pulmonary retention include histological
evidence of aggregated alveolar macrophages (AM) engorged with phagocytized dust
particles, chronic inflammatory response, increased uptake of particles in the intersitial
spaces, and an increased alveolar cell hyperplasia. Subsequent development of alveolitis,
granulomas, and fibrosis are related to the duration and severity of the overload condition.
Morrow (1988) has developed the hypothesis that excessive levels of dust (particles) in the
lungs lead to excessive engulfment of particles by AMs and after a certain degree of loading
occurred, the AMs become progressively immobilized and aggregated. The activated AM
can also release mediators that can affect the integrity of the epithelial barrier, inhibit
antiproteases, or cause influx of inflammatory cells. The relative or complete loss of AM
mobility increases the likelihood of direct particle-epithelial cell interactions and interstitial
localization of dust particles. The impact of this phenomenon is likely modulated by the
particle surface properties, the amount of dust phagocytized, the intrinsic cytotoxicity of the
particles and the persistence of the particle laden cells in the lung milieu.
It has been concluded that particle overloading seriously confounds lexicological
interpretations in the F344 rat (Morrow, 1992) and has important implication for most
species, including humans. At this juncture, differentiating overload effects from those
induced by the intrinsic toxicity of the inhaled material relies to a major extent on the
characterizing the toxic potency of the particles. If the possibility for a particle overload
phenomenon exists, caution is warranted in the use of first-order kinetics to describe
clearance kinetics. Models that incorporate realistic functional and cytological bases and
appropriate kinetic descriptions such as that of Yu and Yoon (1990) to describe diesel particle
clearance, are necessary to describe both reasonable and excessive particle dust burden
retention.
2-34
-------
p. 84
Definitions/Underlying Assumptions
Additional variability and uncertainty in evaluating available inhalation studies occur
because investigators have used different definitions of various respiratory regions and have
employed different methods to estimate total or regional deposition. For example, total
deposition often is estimated by calculating the difference between the amount of compound
in the inhaled air and that in the exhaled air. By making assumptions about mixing and dead
space, estimates of regional deposition may be obtained using measurements of the compound
concentration in different volume fractions of the expired air. As another example, the
definition of upper respiratory tract in various studies has included any or all of the following
anatomic regions: nasopharynx, oropharynx, larynx, or upper trachea. In other studies,
deposition values based on chemical or radiologic assays of tissues after exposure assume no
particle translocation before or during dissection. Some investigators include measurement of
material in the gastrointestinal (GI) tract in their reported value for upper respiratory tract
deposition, while others ignore this translocation. The underlying assumptions and working
definitions for different experimental conditions can contribute a large degree of variability in
reported results. Conversion to some common basis will be necessary in order to calculate
and accurately compare inhaled doses.
2.1.2.3 Appropriateness of Laboratory Animal Species as a Model for Humans
For inhalation studies in particular, there is a dichotomy in terms of the types of
endpoints monitored in human versus laboratory animal studies. Human data concerning the
consequences of inhalation exposure generally consist of information on subjective symptoms
along with clinical data concerning pulmonary function. The relationship between the clinical
picture and lung pathology is poorly defined. However, standard animal toxicological
protocols generally incorporate respiratory tissue evaluation as part of the routine necropsy,
but do not evaluate pulmonary function. Of course, once the lung has been identified as a
target tissue, more detailed studies of it as a target organ may be conducted. When these
more detailed data are available, two additional questions are raised: (1) What is the
significance of alterations in test species' pulmonary performance in terms of potential human
effects? and (2) If tests showing differences in pulmonary biochemistry are available, what is
the utility of the biochemical changes as predictors of disease? Correlations between
2-35
-------
p. 85
functional decrements and immunologic, biochemical, and pathologic changes need to be
quantitated. Work in progress on animal models (see Section 3.1.2.1), biological exposure
indices (Lowry, 1986), and in vitro alterations of lung biochemistry as predictive of lung
disease (Last, 1983) are contributing to this end.
Each inhalation study should be evaluated for possible indications that the respiratory
system is the critical target organ. Human studies that provide only cursory evaluation of
respiratory endpoints make careful evaluation of animal data essential. Human data should be
evaluated with special emphasis on the significance of respiratory system endpoints and
adequacy of their characterization. Extrapolation from oral to inhalation exposures may be
utilized only after careful consideration of factors presented in Section 4.1.2.
For compounds that appear to produce their critical effect within the respiratory system
itself, decisions concerning adversity need to be made on a case-by-case basis. Appendix D
provides specific information concerning evaluation of the severity of respiratory tract
endpoints in humans, while Appendix E provides a summary of issues and references for
pulmonary function evaluation.
Emphysema provides an example of some of the complexities involved in this issue.
Appropriate animal model selection may be contingent upon pathological identification of
early changes consistent with the human syndrome; for example, a clear choice of the most
appropriate laboratory animal species has not been established for emphysema (Snider et al.,
1986). The most recent definition of emphysema by the National Heart Lung and Blood
Institute, Division of Lung Diseases Workgroup (Snider et al., 1985), differentiates between
emphysema in human lungs and animal models of emphysema. When reports of emphysema
following exposures of animals are to be extrapolated to potential hazards for humans, the
definition of human emphysema, rather than that for laboratory animal models of
emphysema, must be used. Thus, the current definitions of emphysema in human lungs and
in laboratory animal models are critical to this review (U.S. Environmental Protection
Agency, 1993a).
The report from the National Institutes of Health (Snider et al., 1985) first defines
respiratory airspace enlargement. "Respiratory airspace enlargement is defined as an increase
in airspace size as compared with the airspace size of normal lungs. The term applies to all
varieties of airspace enlargement distal to the terminal bronchioles, whether occurring with or
2-36
-------
p. 86
without fibrosis or destruction." Emphysema is one of several forms of airspace
enlargement. In human lungs, "Emphysema is defined as a condition of the lung
characterized by abnormal, permanent enlargement of airspaces distal to the terminal
bronchiole, accompanied by destruction of their walls, and without obvious fibrosis."
Destruction is further defined: "Destruction in emphysema is further defined as
nonuniformity in the pattern of respiratory airspace enlargement so that the orderly
appearance of the acinus and its components is disturbed and may be lost." The report also
indicates that "Destruction...may be recognized by subgross examination of an inflation-fixed
lung slice..." Emphysema in laboratory animal models was defined differently. The stated
reason for this difference in the definitions of emphysema in humans and in laboratory animal
models was "In order to foster the development of new knowledge, animal models of
emphysema are defined as nonrestrictively as possible: An animal model of emphysema is
defined as an abnormal state of the lungs in which there is enlargement of the airspaces distal
to the terminal bronchiole. Airspace enlargement should be determined qualitatively in
appropriate specimens and quantitatively by stereologic methods." Thus, in laboratory animal
models of emphysema, airspace wall destruction need not be present. "Appropriate
specimens presumably refers to lungs fixed in the inflated state and is similar to the 1962
American Thoracic Society Committee's requirement for tissue fixation. This document
states "It is still not clear whether the airspace enlargement of age is due to age alone or to
the combination of age and environmental history, but the occurrence of these changes in
nearly all subjects suggests that the changes are normal" (Meneely et al., 1962). Control
animals of the same age as the experimental animals appear necessary to avoid potential
confusion due to age. This National Institutes of Health committee also noted that, to date,
animal models of emphysema fall into two general classes. "The first class centers on testing
the pathogenicity of agents suspected of being relevant to the genesis of emphysema; models
produced by NO2, cadmium, and tobacco smoke are examples of this type. The second class
of models is analytical, for testing specific hypotheses of the pathogenesis of emphysema."
Thus, in reviewing reports of emphysema following experimental exposure to a
toxicant, important considerations include (1) whether the tissue was fixed in an inflated state;
(2) whether airspaces distal to the terminal bronchiole were enlarged beyond normal and
whether that enlargement was determined quantitatively by stereologic methods (control
2-37
-------
p. 87
animals of identical age as exposed animals should be used for sterologic studies to exclude
the possibility that airspace enlargement was due to age); and (3) whether or not airspace wall
destruction, as defined by the NHLBI workgroup (Snider et al., 1985), was present. The
presence of airspace wall destruction, as defined by the NHLBI workgroup, is critical.
In published reports of emphysema following exposure to a toxicant evidence of airspace wall
destruction can only be obtained by careful review of the authors' description of the lesions
or by examining the micrographs the author selected for publication. Thus, although a
particular animal species may share a number of similarities with humans in respiratory tract
physiology, it may be dissimilar in crucial parameters and, therefore, be a less than adequate
source as a model.
Sensory Irritation
One endpoint that is specific to inhalation is sensory irritation. Sensory irritants are
defined as chemicals that stimulate trigeminal nerve endings in the cornea and nasal mucosa
and that evoke a stinging or burning sensation. This perception can be accompanied by
irritation of the throat and coughing from stimulation of laryngeal nerve endings. Sensory
irritants induce, among other effects, a postinspiratory apnea in experimental animals,
resulting in a decrease in breathing rate. A test for sensory irritation in laboratory animals
was developed, based on the premise that if sensory irritation can be prevented then systemic
effects will be prevented as well (Alarie, 1984). The test is based on the decrease in
respiratory frequency occurring in numerous laboratory animals (cats, dogs, mice, rats,
rabbits, and guinea pigs) when exposed to chemical irritants. The decrease in respiratory rate
was found to be concentration-related. The RD50 is the concentration that induces a 50%
decrease in respiratory rate and it has been proposed as the basis of comparison for the
irritating potencies of chemicals (Kane et al., 1979; Alarie, 1984). The test has become a
standard method adopted by the American Society for Testing and Materials.
It should be emphasized that the mechanism of sensory irritation is a different
mechanism than that by which stimuli (physical, toxicologic, or pharmacologic) cause
obstruction in the lower respiratory tract regions (tracheobronchial and pulmonary). In fact,
the epidemiology of bronchial or airway responsiveness and the mechanisms underlying the
physiologic phenomenon of airway hyperresponsiveness still are not completely understood.
2-38
-------
p. 88
Multiple mechanisms have been suggested and one or another may predominate in any given
individual. Possible mechanisms include: alterations in airway geometry, disordered
autonomic regulation of smooth muscle tone, structural alterations in airway smooth muscle,
increased accessibility of stimuli to the muscle, and the release of locally acting mediators of
inflammation. Atopy is a multifactorial trait, both genetically and environmentally
determined, and is only one mechanism by which levels of airway responsiveness can be
increased.
The relationship of sensory irritation to airway irritation is unknown. It is known that
irritation and toxicity can interfere with trigeminal nerve stimulation. An evaluation of the
sensory irritation test for the assessment of occupational health risk found that quantitative
evaluation with respect to human data was not possible due to a number of factors, including
interlaboratory differences in ability to perform the test and intra- and interspecies
inconsistencies in response (Bos et al., 1992), although correlation of RD50 values with TLV
values has been demonstrated (Schaper, 1993). Histopathology has also been reported after
short-term exposure to the RD50 concentration for some irritants (Buckley et al., 1984). For
these reasons, the suitability of the sensory irritation test results is limited to serving as an
indication of the potential for respiratory tract irritation. Dose-response assessment of the
sensory irritation test is not recommended especially for quantitative evaluation of chronic
effects.
Asphyxiation
Another effect specific to the inhalation route is asphyxiation. This effect is thought to
be brought about by reversible, "physical" interactions of gas molecules with biomolecules
(e.g., "displacement" of oxygen by carbon dioxide) (Tichy, 1983). The vapor pressure of a
liquid or solid at ambient temperatures determines the maximum exposure concentration
(MEC) for its vapor. The MEC in parts per million may be calculated from the vapor
pressure (VP) at 25 °C according to
VP9Sor(mmHg) , „ 1X
MEC (ppm) = —2S C _ X 106. (2-1)
^ 760mmHg
2-39
-------
p. 89
Knowing the VP of a liquid or solid is important for estimating its capacity to produce
reversible effects. A compound with a VP of less than 0.76 mm Hg at room temperature
will attain an air concentration of less than 1,000 ppm at the saturated vapor concentration.
This concentration is below the limits for which narcotic or anesthetic effects are generally
observed (Tichy, 1983). Therefore, if a material has a VP of less than 0.76 mm Hg, its
potential to produce such effects can reasonably be ruled out (Dahl, 1990).
Allergic Sensitization
Although most pollutants would be expected to elicit a dose-response upon exposure,
some pollutants cause tolerance/adaptation and some act by allergic or asthmatic mechanisms.
Allergic sensitizers may be considered a subgroup of the agents that produce their critical
effect in the respiratory system. Sensitization is typically caused by high initial doses.
Subsequently, any challenge level of exposure (including low concentrations) may be
sufficient to induce the asthmatic syndrome in sensitized individuals. There is evidence that
IgE antibody levels and inflammatory pulmonary reactions play a role in such syndromes.
Toluene diisocyanate is a well-known example of a sensitizing agent that affects
immunological and pharmacological mechanisms and induces asthma.
The potential for chemicals to induce an airway immune response is related to their
ability to interact with human airway proteins resulting in haptenization or the formation of
new antigenic determinants. Hence, if the structure of the compound suggests that it is
reactive or if it is related to one of the chemicals known to elicit hypersensitivity in humans
(Table 2-5), it is suspect as a potential sensitizing agent. Classes of compounds that have
been most extensively studied for the effects are the anhydrides, isocyanates, and some of the
metal salts.
Several methodologies are now available that test chemicals for their sensitizing
potential. Three of the major approaches include: (1) the Karol method (Karol et al., 1985;
Karol, 1994), (2) the Sarlo method (Sarlo et al., 1992), and (3) the Dearman/Kimber method
(Dearman et al., 1992). None of the methods have been well validated for a range of
chemicals and all have drawbacks. The reader is referred to the summary of workshop
entitled "The Status of Test Methods for Assessing Potential of Chemicals to Induce
Respiratory Allergenic Reactions" (Selgrade et al., 1994) and to Briatico-Vangosa et al.
2-40
-------
p. 90
TABLE 2-5. AGENTS CAUSING WHEEZING AND BRONCHOCONSTRICTION
Large molecular weight compounds
Animals proteins
Laboratory animals
Domestic animals
Birds
Sea squirts
Prawns
Grain weevils
Mites
Arthropods
Enzymes (animal)
Subtilisin
Trypsin, pancreatin
Plant proteins
Cereal grains
Legumes (coffee, soy, castor bean)
Pollen
Seeds (cotton, flax, linseed)
Enzymes (plant)
Papain, bromelain, pectinase, diastase
Vegetable gums
Karaya, tragacanth, acacia (arabic),
quillaja
Fungi
Mold
Inorganic and organic compounds of small
molecular weight
Abietic acid
Anhydrides
Phthalic, trimellitic, hexahydrophthalic,
tetrachlorophthalic, himic
Cyanuric chloride
Platinum salts
Dyes
Azo, anthraquinone, remazol black B dye
Diisocyanates
Toluene diisocyanate
Diphenylmethane diisocyanate
Hexamethylene diisocyanate
Antibiotics
Metallic salts
Nickel
Chromium
Aluminum
Fluxes
Colophony
Aminoethylethanolamine
Miscellaneous
Formaldehyde
Piperazine
Plicatic acid
Pyrethrins
Extract of henna
Adapted from: Moller et al. (1986); Selgrade et al. (1994); Briatico-Yangosa et al. (1994).
2-41
-------
p. 91
(1994) for additional information and guidance on hazard identification and assessment of
respiratory allergic reactions.
Summary
Identification of the most appropriate laboratory animal species is the end result of an
interpretative process that examines all facets of a data base from study design to data
relevance to the extrapolation methodology.
The most sensitive species is selected from evaluation of key studies. Although this
approach (i.e., NOAEL identification) may have the advantage of affording a greater degree
of protection, the species most sensitive to an agent may not be as lexicologically relevant as
other species for extrapolation to humans because of a variety of interspecies variables.
Selection of an appropriate animal model and key study depends on the depth of
understanding of the human disease syndrome, adverse effect, or indicator of toxicity selected
as the criterion for evaluation. For agents whose lexicological outcome is dependent on the
degree to which it is metabolized, the most appropriate animal species is contingent upon
proper evaluation of the numerous interspecies differences with respect to metabolism (see
also Section 3.2). The studies of Plopper et al. (1983) suggest that animal species differ
widely in metabolizing potential of the respiratory tract. Hamsters and rabbits have much
greater metabolizing potentials than do monkeys and rats. Interspecies differences in the
metabolic pathway, as shown for xylene (National Toxicology Program, 1986), may serve as
a basis for selecting one study for RfC derivation and rejecting another. Species-dependent
variables in mucous production and secretion are factors in selecting an appropriate animal
model (see also Chapter 3) for irritants.
The subject of appropriate animal models has been reviewed (Hakkinen and Witschi,
1985) and various mammalian species (rat, hamster, and rabbit) were identified as appropriate
species for extrapolation from several perspectives. Other reviews that discuss the current
limitations and need for the development of animal models as surrogates for humans include
those of Reid (1980), Slauson and Hahn (1980), and Calabrese (1983).
2-42
-------
p. 92
2.1.2.4 Study Validity and Relevance to Extrapolation
The validity of the study and its relevance to human extrapolation is another major area
to consider when assessing individual animal studies. It involves the evaluation of a number
of factors, including all elements of exposure definition (concentration, duration, frequency,
administration route, and physicochemical characterization of the chemical used), reliability
of and limits to the procedures used for both exposure and effects measurements, relevance of
the exposure level tested to the anticipated human exposure level, nature of the effect
(consistency with the area of toxicology assessed and the suspected mechanism of action), and
the similarities and differences between the test species and humans (e.g., in absorption and
metabolism).
Animal studies are conducted using a variety of exposure scenarios in which the
concentration, frequency, and duration of exposure may vary considerably. Studies may use
different durations (acute, subchronic, and chronic) as well as schedules (single, intermittent,
and continuous). All of these studies contribute to the hazard identification of the risk
assessment. Special consideration should be addressed to those studies of appropriate
duration for the reference level to be determined (i.e., chronic investigations for the RfC).
These exposure concerns (concentration 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 rationale
(e.g., comparable metabolism and pharmacokinetic profiles). In the absence of such a
model, the most sensitive species (i.e., the species showing a toxic effect at the lowest
administered dose) is adopted for use as a matter of science policy at the EPA (Barnes and
Dourson, 1988). This selection process is more difficult if the laboratory animal data are for
various exposure routes, especially if the routes are different from that in the human situation
of concern.
Because the data base may be deficient for the route of exposure of interest, it is the
EPA's view that the toxicity potential manifested by one route can be indicative of potential
toxicity via any other exposure route unless convincing contrary evidence exists (Barnes and
Dourson, 1988). Quantitative extrapolation, however, requires consideration of the
differences in the dosimetry for the chemical resulting from the different exposure routes.
Detailed consideration is given to route-to-route extrapolation in Section 4.1.2.
2-43
-------
p. 93
2.1.3 Summarizing the Evidence
The culmination of the hazard identification phase of any risk assessment involves
integrating a diverse data collection into a cohesive, biologically plausible toxicity "picture";
that is, to develop the weight of evidence that the chemical poses a hazard to humans. The
salient points from each of the laboratory animal and human studies in the entire data base
should be summarized as should the analysis devoted to examining the variation or
consistency among factors (usually related to the mechanism of action), in order to establish
the likely outcome for exposure to this chemical. From this analysis, an appropriate animal
model or additional factors pertinent to human extrapolation may be identified.
The utility of a given study is often related to the nature and quality of the other
available data. For example, clinical pharmacokinetic studies may validate that the target
organ or disease in laboratory animals is likely to be the same effect observed in the exposed
human population. However, if a cohort study describing the nature of the dose-response
relationship were available, the clinical description would rarely give additional information.
An apparent conflict may arise in the analysis when an association is observed in toxicologic
but not epidemiologic data, or vice versa. The analysis then should focus on reasons for the
apparent difference in order to resolve the discrepancy. For example, the epidemiologic data
may have contained other exposures not accounted for, or the laboratory animal species tested
may have been inappropriate for the mechanism of action. A framework for approaching
data summary is provided in Table 2-6. Table 2-7 provides the specific uses of various types
of human data in such an approach. These guidelines have evolved from criteria used to
establish causal significance, such as those developed by the American Thoracic Society
(1985) to assess the causal significance of an air toxicant and a health effect. The criteria for
establishing causal significance can be found in Appendix C. In general, the following
factors enhance the weight of evidence on a chemical:
• Clear evidence of a dose-response relationship;
• Similar effects across sex, strain, species, exposure routes, or in multiple
experiments;
• Biologically plausible relationship between metabolism data, the postulated
mechanism of action, and the effect of concern;
2-44
-------
p. 94
TABLE 2-6. APPROACH FOR SUMMARIZING THE EVIDENCE
FROM DIVERSE DATA
CONCEPT 1: STRENGTH OF THE ASSOCIATION
The stronger the association, the greater the confidence that the agent causes the effect.
• Presence of low LCj,,, low NOAEL, high potency index
• Dose-response gradient evident
• High incidence rate, large excess risk
• High level of statistical significance in relevant studies
CONCEPT 2: CONSISTENCY
The association is observed in various circumstances.
• Observed in a number of experimental species
• Various routes
• Different dose regimens
• Descriptive epidemiologic data
• Analytical epidemiologic studies
CONCEPT 3: BIOLOGICAL PLAUSIBILITY
The association is plausible in terms of other scientific information related to the putative causal mechanism.
• A gradient of responses observed
• Short-term or in vitro tests
• Pharmacokinetics
• Molecular action and pathology
• Structure-activity relationship
• Preclinical indicators
• Biological monitoring of exposure
Source: Erdreich (1988).
• Similar toxicity exhibited by structurally related compounds;
• Some correlation between the observed chemical toxicity and human evidence.
The greater the weight of evidence, the greater the confidence in the conclusion derived.
Developing improved weight-of-evidence schemes for various noncancer health effect
2-45
-------
p. 95
TABLE 2-7. HUMAN DATA FOR USE IN HEALTH RISK ASSESSMENT
Study (Alternative Terms)
Comment on Potential Use
Cohort (longitudinal, prospective,
incidence)
Case-control (retrospective, dose or
case-referent)
Cross-sectional (prevalence)1"
Geographic correlation11
Clinical trials
EPIDEMIOLOGIC DATA
Rates as percent response useful in risk assessment. Measure of
excess risk can be obtained. If dose or exposure data are
available, dose-response curves can be constructed. Studies with
ordinal exposure data support strength of evidence and hazard
identification.
No direct measure of disease rates. If exposure data are available,
a NOAEL may be identified.* Studies with ordinal or nominal
exposure data may support strength of evidence and hazard
identification.
Similar to case-control for short-term effects. Prevalence data less
reliable for effects from chronic exposures.
An inexpensive screening procedure. Crude indicator of potential
hazard. Rates are usually only indirectly related to exposure.
Generates hypotheses for analytical studies.
Generally not applicable to environmental issues, because
exposures are treatments or preventive measures. Intervention
trials in which an exposure is removed or changed (e.g.,
medication, smoking, diet) are useful in strength of the evidence
for evaluating causality.
Experimental studies
"Exposed-control" comparisons
(noncohort; see text for discussion)
Case series0
Case reports
NONEPIDEMIOLOGIC DATA
The only human data with controlled exposure levels. Usually
interval level exposure data but low dose, limited exposure time.
Use for hazard identification and dose-response assessment.
Rates may be biased because of self-selection or incomplete
ascertainment of exposed population. Cannot be used to support
absence of hazard. Clinical descriptions useful for hazard
identification.
Can be used to demonstrate hazard if syndrome is unusual.
Usually high level, short-term exposure. May yield data point for
adverse-effect levels. Cannot be used to show absence of hazard.
Suggests nature of acute endpoints in humans. Cannot be used to
support absence of hazard.
"Exposure history is difficult to reconstruct, particularly outside of the occupational setting.
""May be available pertinent to air pollution exposure.
"Several cases seen by or reported by a single investigator. Cases may be attributed to unique exposure
incident, but total exposed population is not defined.
Source: Adapted from Erdreich and Burnett (1985).
2-46
-------
p. 96
categories has been the focus of efforts by the Agency to improve health risk assessment
methodologies (Perlin and McCormack, 1988).
Another difficulty encountered in this summarizing process is that certain studies may
produce apparently positive or negative results, yet may be flawed. The flaws may have
arisen from inappropriate design or execution in performance (e.g., lack of statistical power
or adjustment of dosage during the course of the study to avoid undesirable toxic effects).
The treatment of flawed results is critical; although there is something to be learned from
every study, the extent that a study should be used is dependent on the nature of the flaw
(Society of Toxicology, 1982). A flawed negative study could only provide a false sense of
security, whereas a flawed positive study may contribute to some limited understanding.
Although there is no substitute for good science, grey areas such as this are ultimately a
matter of scientific judgment. The risk assessor will have to decide what is and is not useful
within the framework outlined earlier.
Studies meeting the criteria detailed in Sections 2.1.1 and 2.1.2 (epidemiologic,
nonepidemiologic data), and experimental studies on laboratory animals that fit into this
weight-of-evidence framework are used in the quantitative dose-response assessment discussed
in Chapter 4.
2-47
-------
p. 97
3. CONCEPTUAL BASIS FOR INHALATION
DOSE-RESPONSE ASSESSMENT METHODOLOGY
As discussed in Chapter 1, comprehensive characterization of the exposure-dose-
response continuum is the fundamental objective of any dose-response assessment. Species
differences in anatomical and physiological characteristics, the wide range of physicochemical
properties associated with inhaled chemicals, the diversity of cell types that may be affected,
and a myriad of mechanistic and metabolic differences combine to make the characterization
particularly complex for the respiratory tract as the portal of entry. This chapter attempts to
discuss these factors within the exposure-dose-response context in order to present unifying
concepts. These concepts are used to construct a framework by which to evaluate the
different available dosimetry models; appreciate why they are constructed differently; and
determine how the default approaches presented in Chapter 4 are derived.
3.1 FACTORS CONTROLLING COMPARATIVE INHALED DOSE
The various species used in inhalation toxicology studies do not receive identical doses
in comparable respiratory tract regions when exposed to the same external particle or gas
concentration (Brain and Mensah, 1983). The biologic endpoint or health effect, therefore,
may be more directly related to the quantitative pattern of mass deposited within the
respiratory tract than to the external exposure concentration. Regional deposition pattern
determines not only the initial lung tissue doses but also the specific pathways and rates by
which the inhaled agents are cleared and redistributed (Schlesinger, 1985).
This section discusses the issues associated with the two major factors controlling the
deposition pattern: (1) respiratory anatomy and physiology (Section 3.1.1) and (2) the
physicochemical characteristics of the inhaled toxicant (Section 3.1.2).
The factors that control inhaled dose are discussed relative to the significant mechanisms
by which particles and gases may initially be deposited or taken up in the respiratory tract.
Note that, in this document, disposition is defined as encompassing the processes of
deposition, absorption, distribution, metabolism, and elimination. Initial deposition is used in
3-1
-------
p. 98
reference to gases as well as particles because contact with the respiratory tract surface
precedes absorption. For particles, deposition mechanisms include inertia! impaction,
sedimentation (gravitational), diffusion, interception, and electrostatic precipitation, whereas
mechanisms important for gases include convection, diffusion, chemical reaction (including
metabolism), dissolution, and perfusion. Detailed consideration of these mechanisms is
beyond the scope of this discussion. The reader is referred elsewhere for more extensive
discussions of particle deposition (U.S. Environmental Protection Agency, 1982b, 1986c;
Hatch and Gross, 1964; Raabe, 1979; Hinds, 1982; Lippmann and Schlesinger, 1984) and
gas absorption (U.S. Environmental Protection Agency, 1986, 1993b; Fiserova-Bergerova,
1983; Overton, 1984; Overton and Miller, 1988).
It must be emphasized that dissection of the factors that control inhaled dose into
discrete topic discussions is deceptive and masks the dynamic nature of the intact respiratory
system. For example, although deposition in a particular respiratory region will be discussed
separately from the clearance mechanisms for that region, retention (the actual amount of
inhaled agent found in the lungs at any time) is determined by the relative rates of deposition
and clearance. Retention and the toxicologic properties of the inhaled agent are related to the
magnitude of the pharmacologic, physiologic, or pathologic response. Therefore, although
the deposition, clearance mechanisms, and physiochemical properties of the agent are
described in distinct sections, assessment of the overall toxicity requires integration of the
various factors.
As discussed in Chapter 1, comprehensive description of the exposure-dose-response
continuum requires integration of quantitative knowledge of appropriate mechanistic
determinants of chemical disposition, toxicant-target interactions, and tissue responses into an
overall model of pathogenesis. Improvements in this process will be accomplished in the area
of extrapolation modeling (Miller et al., 1983a; Fiserova-Bergerova, 1983). This involves
determining the dose delivered to the target organ of various species and the sensitivity of the
target organ to that dose. Once such dosimetry has been established and species sensitivity
accounted for, the effective pollutant concentration in laboratory animals can be quantitatively
related to concentration responses in humans. Extrapolation models should incorporate
parameters such as species-specific anatomical and ventilatory differences, metabolic
3-2
-------
p. 99
processes, and the physicochemical properties of the pollutant and should be physiologically
based upon the factors that govern transport and removal of the pollutant.
This chapter provides background information on the major determinants controlling
comparative inhaled dose that should be considered when evaluating the results of
lexicological and human studies for selection of the key studies for the determination of an
inhalation reference concentration (RfC). This background information also provides the
theoretical considerations that are addressed (to varying degrees) by different dosimetry
models, such as those described in Appendices G, I, and J that serve as the basis for the
dosimetric adjustments used in Chapter 4 to extrapolate from experimental conditions to
human equivalent concentrations. A framework by which to evaluate the degree to which
different dosimetry models address these considerations is provided as a summary in
Section 3.2.3.
3.1.1 Respiratory Anatomy and Physiology
The respiratory systems of humans and various experimental animals differ in anatomy
and physiology in many quantitative and qualitative ways. These variations affect air flow
patterns in the respiratory tract, and in turn, the deposition of an inhaled agent, as well as the
retention of that agent in the system. The variations in anatomy and physiology will be
discussed according to respiratory regions and branching patterns, clearance mechanisms, and
cell types. Clearance mechanisms as used here include processes such as the mucociliary
escalator, solubilization in various compartments, uptake, and metabolism.
3.1.1.1 Respiratory Regions and Branching Patterns
The respiratory tract in both humans and experimental animals can be divided into three
regions on the basis of structure, size, and function: the extrathoracic region (ET) that
extends from just posterior to the external nares to just anterior to the trachea, the
tracheobronchial region (TB) defined as the trachea to the terminal bronchioles where
proximal mucociliary transport begins, and the pulmonary region (PU) including the terminal
bronchioles and alveolar sacs. The thoracic (TH) region is defined as the tracheobronchial
and pulmonary regions combined. The anatomic structures included in each of these
respiratory tract regions are listed in Table 3-1, and Figure 3-1 provides a diagrammatic
3-3
-------
p. 100
TABLE 3-1. RESPIRATORY TRACT REGIONS
Region
Extrathoracic (ET)
Anatomic Structure
Nose
Mouth
Nasopharynx
Other Terminology
Head airways region
Nasopharynx (NP)
Upper respiratory tract (URT)
Tracheobronchial (TB)
Pulmonary (PU)
Oropharynx
Laryngopharynx
Larynx
Trachea
Bronchi
Bronchioles (to terminal
bronchioles)
Respiratory bronchioles
Alveolar ducts
Alveolar sacs
Alveoli
Gas exchange region
Alveolar region
Adapted from: Phalen et al. (1988).
representation. The retained dose of an inhaled agent in each of these regions is governed by
the exposure concentration, by the individual species anatomy (e.g., airway size and
branching pattern) and physiology (e.g., breathing rate and clearance mechanisms), and by
the physicochemical properties (e.g., particle size, solubility, reactivity) of the chemical as
discussed in Section 3.1.2.
In general, laboratory animals have much more convoluted nasal turbinate systems than
do humans, and the length of the nasopharynx in relation to the entire length of the nasal
passage also differs between species. This greater complexity of the nasal passages, coupled
with the obligate nasal breathing of rodents, is generally thought to result in greater
deposition in the upper respiratory tract (or ET region) of rodents than in humans breathing
orally or even nasally (Dahl et al., 199la), although limited data are available. The extent of
upper respiratory tract removal affects the amount of particles or gas available to the distal
respiratory tract.
Airway size (length and diameter) and branching pattern affect the aerodynamics of the
respiratory system in the following ways:
3-4
-------
p.1
Extrathoracic
Region
Tracheobronchlal
Region
Inspirable Mass
Fraction
(Enters via nose
or mouth)
Thoracic Mass
Fraction
(Penetration past
terminal larynx)
Respirable Mass
Fraction
(Penetration past
terminal bronchioles)
Pulmonary
Region
Figure 3-1. Diagrammatic representation of three respiratory tract regions.
• The airway diameter affects the aerodynamics of the air flow and the distance from
the agent molecule or particle to the airway surface.
• The cross-sectional area of the airway determines the airflow velocity for a given
volumetric flow.
• Airway length, airway diameter, and branching pattern variations affect the mixing
between tidal and reserve air.
Differences in airway sizes and branching among species therefore may result in significantly
different patterns of transport and deposition for both particles and gases. Alveolar size also
differs between species, which may affect deposition efficiency due to variations on the
distance between the airborne particle or molecule and alveolar walls (Dahl et al., 1991a).
3-5
-------
p. 2
Effect on Particle Deposition Mechanisms
Air flow in the extrathoracic region is characterized by high velocity and abrupt
directional changes. Therefore, the predominant deposition mechanism in the ET region is
inertia! impaction. In this process, changes in the inhaled airstream direction or magnitude of
air velocity streamlines or eddy components are not followed by airborne particles because of
their inertia. Large particles (> 5 jiim in humans) are more efficiently removed from the
airstream in this region.
Impaction remains a significant deposition mechanism for particles larger than 2.5 /*m
aerodynamic equivalent diameter (dae) in the larger airways of the TB region in humans and
competes with sedimentation, with each mechanism being influenced by mean flow rate and
residence time, respectively. As the airways successively bifurcate, the total cross-sectional
area increases. This increases airway volume in the region, and the air velocity is decreased.
With decreases in velocity and more gradual changes in air flow direction as the branching
continues, there is more time for gravitational forces (sedimentation) to deposit the particle.
Sedimentation occurs because of the influence of the earth's gravity on airborne particles.
Deposition by this mechanism can occur in all airways except those very few that are
vertical. For particles ~4 jum dae, a transition zone between the two mechanisms, from
impaction to predominantly sedimentation, has been observed (U.S. Environmental Protection
Agency, 1982b). This transition zone shifts toward smaller particles for nose breathing.
Differences in airway size and branching pattern are a major source of interspecies
variability in inhaled dose for the TB region. Larger airway diameter results in greater
turbulence for the same relative flow velocity (e.g., between a particle and air). Therefore,
flow may be turbulent in the large airways of humans, whereas for an identical flow velocity,
it would be laminar in the smaller experimental animal. Relative to humans, experimental
animals also tend to have tracheas that are much longer in relation to their diameter. This
could result in increased relative deposition in humans because of the increased likelihood of
laryngeal jet flow extending into the bronchi. Human airways are characterized by a more
symmetrical dichotomous branching than that found in most laboratory mammals, which have
highly asymmetrical airway branching (monopodial). The more symmetrical dichotomous
pattern in humans is susceptible to deposition at the carina because of its exposure to high air
3-6
-------
p. 3
flow velocities toward the center of the air flow profile. These comparative airway anatomy
differences are summarized in Table 3-2.
Sedimentation becomes insignificant relative to diffusion as the particles become
smaller. Deposition by diffusion results from the random (Brownian) motion of very small
particles caused by the collision of gas molecules in air. The terminal settling velocity of a
particle approaches 0.001 cm/s for a unit density sphere with a physical diameter of 0.5 ^m,
so that gravitational forces become negligible at smaller diameters. The main deposition
mechanism is diffusion for a particle whose physical (geometric) size is <0.5 /urn.
Impaction and sedimentation are the main deposition mechanisms for a particle whose size is
greater than 0.5 pm. Hence, dae = 0.5 /*m is convenient for use as the boundary between
the diffusion and aerodynamic regimes. Although this convention may lead to confusion in
A
the case of very dense particles, most environmental aerosols have densities below 3 g/cm
(U.S. Environmental Protection Agency, 1982b). Diffusional deposition is important in the
small airways and in the PU region where distances between the particles and airway
epithelium are small. Diffusion has also been shown to be an important deposition
mechanism in the ET region for small particles (Cheng et al., 1988, 1990).
These mechanisms for particle deposition in the respiratory tract are schematically
represented in Figure 3-2. Experimental deposition data and extrapolated estimates on
humans that illustrate these same concepts are shown by the curves for PU (alveolar) and TB
deposition in Figure 3-3. Deposition fraction is shown plotted against particle diameter. It is
important to note that over half of the total mass of a typical ambient mass distribution would
be deposited in the ET region during normal nasal breathing, with most of this being coarse
particles (U.S. Environmental Protection Agency, 1986c). With mouth-only breathing, the
regional deposition pattern changes dramatically compared to nasal breathing, with ET
deposition being reduced and both TB and PU deposition enhanced. Oronasal breathing
(partly via the mouth and partly nasally), however, typically occurs in healthy adults while
undergoing moderate to heavy exercise. Therefore, the appropriate activity pattern of
subjects for risk assessment estimation remains an important issue. Miller et al. (1988)
examined ET and thoracic deposition as a function of particle size for ventilation rates
ranging from normal respiration to heavy exercise. A family of estimated deposition curves
were generated as a function of breathing pattern. Anatomical and functional differences
3-7
-------
p. 4
•5 s « °
3 -8 -a -2
5 | ||
85 S
^\ 43 *•
2 .S
H S
O
3 Cfl
-I
^.3
S
00
3
CJ
c-i
to
S
8-
nd
1
-
Ifi
, 1
1 2
s s
J5 43
< <
8 8
cN cs
\o *°
p: »Q
*"t ON
1-a
t2 &
>1 1
• vj
§ o
• ^
w>
1 1 1
II GO
0 B 0 "«"
'5 £ 'S s
S.^ « E
3 -0 .E U
W *fl
I.I 1
g.sl a
5 -s 5
"s>
'S.
? |jf 1
dO- (2
CO
!
1
"B "3
•o >
n o
T3 TJ
^ C
g OS
•2 -C
§ <&
_c
it
*n
•*
&
1
g
•o
«
o
-1
o
«
CO
2
"B
cO
fr
1
^
s
o
CO
ri
!
«
!•"
2
u
^
0
3
cn
1
«
"f
>
CO
o
I
1
CO
•§
CO
o
to
^
r°
1
1
=?
1
CO
s
o
•£
a
1
&
•E
0
3
5
<
oo
o
CO
JO
»n
cs
&
g.,
4J
JO
O
^-
•al
1 1
1/3 1
-------
p. 5
Directional
Change
Very
Abrupt
Air
Velocity
•i
Impaction
Jl
XT'
Sedimentation
Less
Abrupt
Impaction
Mild
Diffusion
Electrostatic
Precipitation
Figure 3-2. Schematic representation of selected parameters influencing regional
deposition of particles in the respiratory tract.
Source: Adapted from Casarett (1975); Raabe (1979); Lippmann and Schlesinger (1984).
between adults and children are likely to yield complex interactions with the major
mechanisms affecting respiratory tract deposition, again with implications for risk assessment.
Age-dependent dosimetric adjustments may be possible, pending data availability for children.
3-9
-------
p. 6
1.0
0.9
0.8
0.7
| 0.6
2 0.5
•I 0.4
0> 0.3
Q
0.2
0.1
0
0,
\ I I I I I I I I T
Range of Pulmonary Deposition,
Mouth Breathing
Estimate of Pulmonary Deposition, Nose Breathing
Range of Tracheobronchial Deposition,
Mouth Breathing
Extrapolation of Above to Point (•) Predicted
by Miller etal. (1979)
O* Emmett et al. (1982); 337cm »«-\ 6-s Breathing Cycle
D • Heyder et al. (1986); 750 cma s-1.4-s Breathing Cycle
-A A Heyder et al. (1986); 250 cm'*-', 4-s Breathing Cycle/
O+ Svartengren (1986)
Open Symbols: Tracheobronchial Deposition
- Solid Symbols: Alveolar Deposition
\
\
\
0\
o \
o »v
O\
\ _
\ ~~
o I-
*A*
Ti 11 m
0.2
0.3 0.4 0.5
1.0
2.0 3.0 4.0 5.0
10 12141620
Physical Diameter i
Aerodynamic Diameter (^m)
Figure 3-3. Regional deposition in humans of monodisperse particles by indicated
particle diameter for mouth breathing (pulmonary and tracheobronchial)
and nose breathing (pulmonary). Deposition is expressed as fraction of
particles entering the mouth or nose. The PU band indicates the range of
results found by different investigators using different subjects and flow
parameters for PU deposition following mouth breathing. The TB band
indicates intersubject variability in deposition over the size range measured
by Chan and Lippmann (1980). The extrapolation of the upper bound of
the TB curve in the larger particle size range also is shown and appears to
be substantiated by data listed in the legend.
Source: U.S. Environmental Protection Agency (1986c).
Effect on Gas Deposition and Uptake
The major processes affecting gas transport involve convection, diffusion, absorption,
dissolution, and chemical reactions. These mechanisms are schematically represented in
Figure 3-4. Predictions of lower respiratory tract distribution of ozone from a detailed
dosimetry model that accounts for many of these processes is shown in Figure 3-5.
3-10
-------
p. 7
Inspiration
V
Velocity
Alveolar
Sacs
Figure 3-4. Schematic representation of selected parameters influencing regional
deposition of gases in the respiratory tract.
Source: Overton (1984).
Beginning at the trachea, the model predicts the net ozone dose (flux to air-liquid interface)
slowly decreases distally in the tracheobronchial region and rapidly decreases in the
pulmonary region (U.S. Environmental Protection Agency, 19935).
3-11
-------
p. 8
do*.
E
I
VT (ml) f (bpm)
Human—.— 8000 150
Rat---- 1.98 66.0
Guinea Pig * 2.63 60.9
Rabbit «..._.. 13.20 38.8
(No absorption in the URT)
Zone 023 4 5 67 8 Rabbit
Order 0234 5678 91011 12 13 14 Guinea Pig
Generation 0246 8 10 12 14 151617 19 21 23 Rat
Generation 0 2 46 8 10 12 14 161718 20 22 23 Human
Figure 3-5. Net dose of ozone versus sequential segments along anatomical model lower
respiratory tract paths for human, rat, guinea pig, and rabbit. In general,
each segment represents a group of airways or ducts, with common features
as defined by the designers of the anatomical model (human and rat:
generation; guinea pig: order; rabbit: zone). For a given species the
plotted dots represent a predicted dose that corresponds to a given segment.
The dots have been joined by lines for ease of interpreting the plots; these
lines do not represent predicted values except where they intercept the dots.
TB = tracheobronchial region. PU = pulmonary region.
Source: Overton and Miller (1988).
The bulk movement of inspired gas in the respiratory tract is induced by a pressure
gradient and is termed convection (U.S. Environmental Protection Agency, 1982b).
Convection can be broken down into components of advection (horizontal movement of a
mass of air relative to the airway wall) and eddy dispersion (air mixing by turbulence so that
individual fluid elements transport the gas and generate flux). Molecular diffusion is
superimposed at all times on convection (bulk flow) due to local concentration gradients.
Absorption removes gases from the lumen and affects concentration gradients.
3-12
-------
p. 9
The average concentration of a gas in a tube (i.e., an "idealized" airway) can be
described by one-dimensional convection and dispersion. A pulse of substance moves down a
tube with an average air velocity equal to the medium's (air's) average velocity, and its
spread in the axial direction is governed by an effective dispersion coefficient that can be
described by Pick's law of diffusion (Overton, 1984). This effective dispersion coefficient is
larger than the molecular diffusion coefficient except in the PU region. As illustrated in
Figure 3-4, perpendicular transport in this region can carry a gas molecule into the alveoli,
but because of the alveolar walls, there is minimal net axial transport with respect to that in
the central channel. The average axial transport is slowed because only a fraction of the
molecules in the cross-sectional average can move axially, generally resulting in a dispersion
process with a dispersion coefficient less than the molecular diffusion coefficient, although it
is possible for longitudinal mixing to be enhanced by the presence of alveolar septa leading to
dispersion coefficients that are actually greater than the molecular diffusivity (Federspiel and
Fredberg, 1989). The dispersion coefficient is a function of the molecular diffusion
coefficient, the total air volume, and the generation's alveolar airspace volume (Overton,
1984). The dispersion coefficient is also influenced by the absorption process (Dayan and
Levenspiel, 1969).
Molecules are transferred from the flowing gas into the liquid layer lining the airway
wall by molecular diffusion. A simple description for this process postulates a thin, stagnant
air layer based on the assumption that the air velocity becomes very small as the air-liquid
interface is approached. Transfer through this layer depends on the gas-phase diffusion
coefficient, layer thickness, and the gas concentrations at the boundaries of the layer. If the
molecules are absorbed, then the concentration of the gas in the diffusion layer is decreased
at the liquid boundary. As the ability of the liquid to remove the gas increases, the relative
concentration at the gas-liquid boundary decreases, and the mass transfer from the gas phase
to the liquid phase increases. For poorly soluble, hydrophobic, and nonreactive gases, little
gas is removed by the airways. The transport into and chemistry of the adjacent surface
liquid and tissue layers will be described in Section 3.1.2.2, which describes the
physicochemical characteristics of gases and vapors. These next layers can serve as a "sink"
to help "drive" the delivery of gas across this layer. Capillary blood flow (i.e., perfusion) is
important to the gas uptake in that it removes the gas or its chemical reaction products on the
3-13
-------
p. 10
other side of these liquid and tissue layers. Therefore, addressing species differences in
alveolar ventilation, regional perfusion rates, and cardiac output is critical to estimating initial
absorbed dose. The importance of regional differences (e.g., the distance from the air to the
capillaries in the tracheobronchial region is 7 to 20 times that in the pulmonary region
[Overton and Miller, 1988]) and interspecies differences in the anatomic relationship of the
airspace to capillary blood should be considered. Transfer also is enhanced by a reduction in
diffusion layer thickness that is dependent on the nearby rate of airflow; the higher the flow
velocity, the thinner the layer, again emphasizing the significance of airway morphology.
Although the preceding figures have only illustrated these concepts for the lower
respiratory tract, the influence of anatomy on comparative deposited dose is also important in
the ET region. Species differences in gross anatomy, nasal airway epithelia (e.g., cell types
and location) and the distribution and composition of mucous secretory products have been
noted (Harkema, 1991; Guilmette, 1989). The geometry of the upper respiratory tract
exhibits major interspecies differences (Gross and Morgan, 1992). Figures 3-6 and 3-7 show
diagrams of the ET region that illustrate the differences between Rhesus monkeys and F344
rats. Cross-sections for the four levels shown on the transverse section are at comparable
locations in the monkey and rat. Figure 3-8 shows the influence these differences have on
airflow patterns in the region. In both species shown in Figure 3-8, studies have
demonstrated complex inspiratory flow streams, exhibiting regions of simple laminar,
complex secondary (vortices, eddies, swirling), and turbulent flows (Morgan et al., 1991).
Differences in nasal air flow patterns between these two species and humans (Hahn et al.,
1993) is an important consideration for extrapolation of dose associated with nasal toxicity.
Good correlation has been shown between routes of flow, regional secondary flows,
turbulence, and impaction of airstreams on the airway wall, with the reported distribution of
formaldehyde-induced nasal lesions in these species, illustrating the influence of the nasal
anatomy on gas deposition for this reactive and soluble gas (Morgan et al., 1991; Kimbell
etal., 1993).
In order to model the effects that the intricate morphological structure of the respiratory
tract have on the nature of gas mixing and flows, representations of the mechanical mixing
imparted by tube bifurcations, turbulence, and secondary flows due to molecular diffusion
must be formulated. Location, diameter, and length of airways are considered to be the
3-14
-------
p. 11
L8W8I2
Levels
Level 4
Figure 3-6. Diagram of the nasal passages for the F344 rat modified from Morgan et al.
(1984). Cross-sections are shown at the four levels indicated and
correspond to comparable locations for the rhesus monkey illustrated in
Figure 3-7. Note the greater complexity of the posterior (ethmoid) region of
the rat nose compared to that of the monkey. Much of this region is
covered by olfactory epithelium, reflecting the macrosmatic nature of
rodents.
relevant measurements for gas transport (Overton, 1984). Because of the morphology of the
respiratory tract and air flow patterns, the relative contribution of these gas transport
processes is a function of location in the respiratory tract and point in the breathing cycle
(i.e., depth and rate) (U.S. Environmental Protection Agency, 1982b; Overton, 1984). The
interspecies differences in the nature and structure of the respiratory tract, as summarized in
Table 3-2, critically influence the differences in transport and deposition of gases across
3-15
-------
p. 12
Level 1
Level 2
Levels
Level 4
Figure 3-7. Diagram of the nasal passages for the Rhesus monkey modified from
Montketto et al. (1989). Broken lines on the transverse section indicate
the junction of squamous with the trarnitional/respiratory (anterior line)
epithelia and the respiratory with the olfactory epithelium (dorsal line).
Cross sections are shown at the four levels indicated and correspond to
comparable locations for rat illustrated in Figure 3-6.
species. The airways also show a considerable degree of within species size variability and
this is most likely the primary factor responsible for the deposition variability seen within
single species (Schlesinger, 1985). Sex also influences airway anatomy. Additionally, age
has dramatic influences on respiratory dynamics.
The differences in respiratory tract anatomy summarized in this Section 3.1.1 are the
structural basis for the species differences in gas and particle deposition. In addition to the
structure of the respiratory tract, the regional thickness and composition of the airway
epithelium (a function of cell types and distributions) is an important factor in gas absorption
3-16
-------
p. 13
Male F344 Rat
Male Rhesus Monkey
Superior Dorsal
Stream
Medial Anterior
Jet
Ventral Anterior
Vortex
Middle Lateral
Vortex
Superior Ventral Lateral
Stream
Inferior Ventral Lateral
Stream
' Ventral Anterior
Vortex
Figure 3-8. Inspiratory airflow patterns in upper respiratory tract of F344 rat and
Rhesus monkey. A = major medial streams; B = major lateral streams.
Black and white arrows depict high and low velocity airstreams,
respectively.
Source: Morgan et al. (1991).
and contributes to the solubility and extent of reaction of the gas. Other anatomic and
physiologic factors that influence gas uptake include (1) ventilation, which affects the tidal
volume and ventilation to perfusion ratios; (2) body build, which affects the volume of
distribution (including cardiac output and tissue volume); and (3) metabolic capacities. These
are all factors to evaluate when estimating inhaled dose, interpreting injury response, and
extrapolating effects between species.
3-17
-------
p. 14
3.1.1.2 Clearance Mechanisms
Deposited material is removed from the respiratory tract by clearance mechanisms,
which vary depending on the site of deposition and the properties of the inhaled toxicant.
The speed and efficiency by which the inhaled toxicants are cleared can be critical
determinants of their toxic potential. Rapid removal lessens the time available to cause
critical damage to the respiratory tract tissue and to permit systemic absorption of agents that
have target organs other than the respiratory tract (Menzel and Amdur, 1986). The clearance
mechanisms involved include (1) exhalation of volatiles; (2) mucociliary transport;
(3) macrophage phagocytosis; (4) chemical reactions; (5) metabolism by various cell types;
and (6) dissolution and absorption into the blood, lymphatic, or lung fluids.
Inhalation represents a route of exposure in which a variety of interrelated factors
influence not only the nature of the effects (respiratory versus systemic) but also the manner
by which they occur. The influence of target cell populations in the respiratory tract on the
nature of the response is a factor unique to the inhalation route of exposure. Unlike the liver,
a first-pass organ in oral exposures that has a more homogenous population of limited types
of cells, the respiratory tract has more than 40 cell types (Sorokin, 1970). Xenobiotics,
which exert their action by direct effects of the parent compound or by metabolites, can
manifest profound differences in the nature and degree of response, depending on the route of
exposure and subsequent availability to interact with various cell populations.
The likelihood of adverse effects in the respiratory tract can be affected by
(1) production, distribution, and reactivity of metabolites by and among specific cell types;
(2) the degree to which detoxication systems are overwhelmed (e.g., glutathione depletion);
(3) efficiency and sensitivity of repair processes (e.g., type II cell proliferation);
(4) efficiency of clearance processes; (5) airway mechanics; and (6) mechanism of action
(e.g., pharmacologic or immunologic) (Bond, 1989; Boyd, 1980; Calabrese, 1983; Gram
et al., 1986; Trush et al., 1982; Nadel et al., 1986; Marin, 1986).
Exhalation of volatile agents (including from administration routes other than inhalation)
is an important excretory pathway that is dependent on tissue levels and exposure regimen.
For inhalation exposures, the exposure duration influences the amount of chemical entering
the systemic circulation, the amount metabolized, and the concentration of the chemical in
tissues. Using a simulation model, Fiserova-Bergerova et al. (1984) demonstrated that for
3-18
-------
p. 15
chemicals that are not metabolized, tissue concentrations of "poorly soluble" (Hoil/ < 10)
chemicals change very minimally after 2 h of exposure. The pulmonary uptake rate
approaches zero at the end of a 2-h exposure and apparent equilibrium is established. "Easily
soluble" chemicals (10 < Hoil/gas < 10,000) require more than 1 day of exposure to reach
apparent equilibrium and "highly soluble" chemicals (Hoil/gas > 10,000) require more than
1 year of exposure. If the chemical is metabolized, pulmonary uptake and the amount
metabolized increase with exposure duration, but the effect of metabolism may be more
complex if exposure concentrations are so high that metabolic pathways approach saturation
kinetics and cause metabolism to deviate from first order kinetics.
Conversely, pulmonary clearance decreases with increasing biosolubility (refers to
solubility of gases and vapors in biologic materials) and thereby affects the accumulation of
chemicals during intermittent exposure regimens. Simulation of an 8 h/day, 5 days/week
schedule for a 3-week exposure duration to a 70 kg man showed that poorly soluble
chemicals (as defined previously) have no tendency to accumulate in the body, although
easily and highly soluble chemicals do have a tendency to accumulate because the
intermissions between exposures are not long enough to allow the chemical to be removed
from adipose tissue (Fiserova-Bergerova et al., 1984). Excursions in exposure concentrations
had a great effect on tissue concentrations of poorly soluble chemicals, but had little effect on
tissue concentrations of highly soluble chemicals. Concentrations in well-perfused tissues
were more affected by excursions in exposure concentrations than concentrations in muscle or
adipose tissues.
The results of these simulation efforts emphasize the uncertainty that the dual function
(i.e., uptake and exhalation) of the respiratory system adds to any attempt to estimate either
respiratory tract or extrarespiratory (remote) "dose" of volatile agents. These simulations
also emphasize the need for careful consideration of the uptake, metabolism, and excretion
parameters for these agents when attempting the exposure duration and concentration
conversions discussed in Chapter 4, and when ruling out the possibility of a respiratory tract
endpoint when using oral data as part of the data base.
There are numerous defense systems that protect the respiratory tract. While some
defense systems are truly protective, it must be kept in mind that many "activate" inhaled
agents and may be responsible for adverse effects. Defense systems can be physical in nature
3-19
-------
p. 16
(e.g., filtration of particles by nasal hair), mechanical (e.g., expiration), enzymatic, or
cellular (e.g., phagocytosis).
Nasal hair can be envisioned as a first line of defense since it can help prevent contact
of toxicants (e.g., particles) with underlying epithelia. However, trapping of agents in the
diffusion layer underlying cilia in the nose can serve as a source of irritation and more
serious adverse effects. Some agents (e.g., formaldehyde, acrolein) have been shown to
cause severe lesions in nasal epithelial cells (Morgan et al., 1986). The mouth also can be
envisioned as another first-line defense system. Mouth-breathing in humans can result in
solubilization of vapors in saliva and deposition of particles. Swallowing can reduce
pulmonary exposure but increase presentation of the agent systemically via gastrointestinal
tract absorption. Once an agent penetrates to the tracheobronchial region, agent deposition
and/or solubilization occurs in the mucous blanket covering the surface epithelium.
Deposited particles can be cleared from the respiratory tract completely or they may be
translocated to other sites within this system. Clearance mechanisms are regionally distinct,
in terms of both routes and kinetics (Dahl et al., 1991a). Particles deposited on the anterior
nares are cleared by mechanical processes such as nose wiping, blowing (humans), or
sneezing (animals/humans). Particles in this area can have long biological half-lives. Those
deposited in the nasopharynx or oropharynx, however, are swallowed within minutes and
passed through the esophagus down to the gastrointestinal tract.
Particles deposited in the TB region are transported out of the respiratory tract by the
mucociliary system, an interaction between the mucous secretions and the cilia that provide
the mechanisms of movement. Such transport occurs along the area from the larynx to the
terminal bronchioles. Insoluble particles are transported up to the esophagus where they are
swallowed. Soluble particles may dissolve in the mucus. Generally, the biological half-lives
of insoluble particles deposited in the TB region are on the order of hours.
Clearance of particles from the PU region of the lung generally takes the longest,
usually a rapid phase of hours, and slower phases with biological half-lives of days, months,
or years, depending on particle size and solubility. A major clearance process for "insoluble"
particles is phagocytosis by alveolar macrophages. These cells then may be removed from
the PU region after reaching the distal terminus of the mucociliary transport system or by
migrating through the interstitium to the lymphatic system. Highly soluble particles will
3-20
-------
p. 17
dissolve in alveolar lining fluid and enter the blood or lymph directly (Johanson and Gould,
1977;Dahletal., 1991a).
It is likely that dissolution rates and rates by which dissolved substances are transferred
into blood are related mostly to the physicochemical properties of the material being cleared
and are essentially independent of species. On the other hand, different rates of mucociliary
transport in the conducting airways or of macrophage-mediated clearance from the PU region
may result in species-dependent rate constants for these pathways (Dahl et al., 1991a). For
example, clearance of insoluble particles from the PU region of mice and rats is much faster
than that in dogs and humans, which have similar clearance rates of inhaled particles (Snipes,
1989a,b).
As discussed in Chapter 2, an overload phenomenon can occur with excessive particle
exposures that can alter the clearance kinetics of lung dust burdens and confound the
interpretation of lexicological effects (Morrow, 1992).
Conceptually, uptake of a gas requires that it move from the airway lumen through the
surface-liquid lining layer, the tissue layer, and the capillary endothelium, to reach the blood.
This passage is influenced by the physiochemical properties of the gas as well as the
biochemistry and thickness of the layers between the lumen and blood. For reactive gases,
the sequence in which anatomic sites are affected appears to be more dependent on
concentration than on exposure duration. However, at a given local anatomic site and at a
specific concentration, the stages in the pathogenesis of the lesion relate to the duration of
exposure (U.S. Environmental Protection Agency, 1986d, 1993b). The rate of mucous
transport also affects the gas transport mechanisms in the diffusion layer at the gas/liquid
interface along the airways. The rate varies with the depth of the airways (greater velocities
in the proximal airways) and across species. For example, a very highly reactive gas may
not reach the blood if it reacts biochemically with mucus and the mucus layer has sufficient
volume (thickness) to serve as a sink. This same gas may not react with the saturated lipid of
surfactant; and if deposited significantly in the PU region, could reach alveolar tissue. The
thickness and efficiency of the epithelial barrier also influences absorption. Both of these
main factors (liquid lining and epithelial barrier) are present in all species but have
species-specific differences, only a few of which have been quantified. Mucus is a complex
secretion with contributions from various epithelial cells. The numbers and distribution of
3-21
-------
p. 18
these cells may affect the composition and properties of the mucus, which in turn interacts
with the physicochemical properties of the agent. The species differences in the thickness of
the alveolar epithelial cells could account for variations observed in the diffusion of gases into
the bloodstream (Crapo et al., 1983). The lung also is a very efficient excretory organ for
volatile organic chemicals after the exposure ceases or is lowered. The efficacy of PU
excretion correlates directly with the saturated vapor pressure of the chemical and indirectly
to water solubility.
Cell Types
A variety of other cellular defense mechanisms can be marshaled, which can diminish
or sometimes exacerbate toxic insult. The numerous cell types found in different species
contribute to the varying clearance patterns from the respiratory regions and differences in the
nature of the response. Table 3-3 presents the distributions of various cell types across
species commonly used in inhalation toxicologic investigations. Different mammalian species
have different amounts and isozyme distribution of cytochrome P-450 in their Clara cells,
which could account for differences in metabolism of some agents. Recent investigations
have also shown species differences in cellular organization at the terminal respiratory
bronchioles/alveolar duct junctions and in the ultrastructure of the same cell type across
species (St. George et al., 1988). The possible functions of these cell types are provided in
Table 3-4, and the differences seen in the cell types across species are summarized in
Table 3-5. Such species differences are important to consider when determining if the
laboratory animal is an appropriate model for the chemical's mechanism of action. For
example, the rat may be an inappropriate species for the evaluation of hyper sensitivity
because of its lack of mast cells.
Alveolar macrophages are the predominant cell type responsible for clearance of
particles from the PU region. Particles are phagocytized and transported within macrophages
to the mucociliary escalator. This alveolar macrophage clearance of the PU region is
considerably slower (weeks to years) than clearance in the TB region. Gases and soluble
particles that escape phagocytosis by alveolar macrophages can be dissolved in the lining
fluid. This dissolution would be governed by physicochemical characteristics such as
3-22
-------
p. 19
K
A
1
o
-
E
I
1
a
PQ
+0
+ 60 I I + + + .1 +
+ +
rt
§
'
U
w
0> »g O
o •—' «—i 2
^ | "S f
l-s i §
^o 3S S
' +
- 4-
+ +
+
•a -
Cq (^
f
o «
ll
8
i
!
8
o
o
1
8
R o
I
U w
I...
t C ^^
*^ ^^ T—^
1 o 5
1 frJI
3 ' 3 .^
I-S|l
a-a o §
u 0 fc r «
•a u n-, oo 43 .« .-
"8
.3
1
1 if« ?
ll'
T5 O
o5 ^*
s< ?•>
en
oo
o\
•a
13
I
S3
.&
3-23
-------
p. 20
TABLE 3-4. SOME SPECIFIC LUNG CELL TYPES AND THEIR FUNCTIONS
Cell Types
Location and Function
Epithelium
Clara cells
Ciliated cells
Type II alveolar
Type I alveolar
Mucous
Serous
Brush cells
Globule
leukocyte
Endocrine
Submucosal
Goblet (mucus)
cells
Serous cells
Endocrine cells
Lymphocytes
Myoepithelial
Bronchoalveolar
mast cells
Macrophage
Endothelial cells
Fibroblasts
(interstitial)
Nonciliated cells of the tracheobronchial region; high xenobiotic
metabolic activity; secretory; function not well-defined; may serve as
precursor of goblet and ciliated cells
Most common epithelial cells in airways; may secrete mucous-like
substances; controls perciliary fluid
Generally covers < 5 % of alveolar surface; secrete surfactant; replace
injured Type I cells; high xenobiotic metabolic activity
Large and covers considerable surface area per cell; covers >95% of
alveolar surface; forms the alveolar epithelium and facilitates gas
exchange; low metabolic activity; incapable of self-reproduction
Mucus-secreting
Mucus-secreting; perciliary fluid; stem cell
Chemoreceptor cells; preciliated
Immunoglobulin transportation; releases inflammatory mediators
Secreto- and vaso-regulatory
Epithelial linings; common in trachea and bronchioles; contribute
to mucus production
Mucus-secreting; perciliary fluid; stem cell/proliferative
Secretes amines and neuropeptides
Immunoresponsive
Expulsion of mucus
Migratory cells located throughout respiratory tract; release mediators
of bronchoconstriction when antigens bind to IgE antibodies on surface
Phagocytic; secrete mediators of inflammatory reactions; modulate
lymphocytes and otherwise participate in immune response
Approximately 40% of lung parenchyma cells; metabolize blood-borne
substances; proliferative
Predominant in alveolar wall and constitutes the basement membrane;
become activated during disease states and produce elastin and
collagen; proliferation leads to fibrosis, modulation of growth,
bronchial tone, and mucosal secretion
Source: Jeffery (1983), Bowden (1983), Marin (1986), Nadel et al. (1986), Plopper et al. (1983), Burri (1985),
Brain (1986).
3-24
-------
p. 21
TABLE 3-5. MAIN SPECIES DIFFERENCES IN EPITHELIAL CELLS
AND GLANDS
Epithelial Morphology
Thickness and pseudostratification
Thickness and structure of "basement membrane"
Mucus-secreting cells
Number
Histochemistry
Predominant ultrastructure type
Clara cells
Morphology (smooth endoplasmic reticulum)
Distribution
Endocrine cell frequency
Ciliated cells
Extent of coverage
Structure of rootlet
Lamellar bodies
Glycogen stores
Presence of brush cell
Basal cells
Number
Shape
Tonofilaments
Presence of Globule Leukocytes
Innervation
Extent
Distribution
Type
Gland Morphology
Amount
Distribution
Mam histochemical cell type
Presence of collecting duct
Innervation
Source: Jeffery (1983).
3-25
-------
p. 22
reactivity, water solubility, lipophilicity, and ability to serve as substrate for activation and/or
detoxification enzymes.
Certain cell types can be stimulated to release mediators, such as mast cell release of
histamine. Histamine can cause bronchoconstriction, which can be protective, by limiting the
amount of pollutant inhaled, or can be toxic, by limiting oxygen uptake. Synthesis or
metabolism of prostaglandins (leukotrienes) also can affect airway and vascular caliber. The
chemotactic factors released can recruit phagocytic cells involved in clearance. It should be
recognized that the respiratory tract contains a variety of different cell types that possess
different metabolizing potential and are distributed in a manner that varies among species.
Lists of common cell types and their functions are provided in Tables 3-3 and 3-4.
Macrophages, for example, constitute a cellular protection system and not only protect inner
surfaces of the respiratory tract from damage caused by particles and microorganisms, but
also have the potential to cause damage themselves because the proteases and mediators that
are useful in destroying microbes or physical agents can also destroy healthy tissue (Rossi,
1986) (Brain, 1986). Although recruitment of macrophages to the lung is related to the
toxicant dose, the adaptive increase in macrophages can be exceeded (Bowden, 1986). This
threshold may vary among species. The alteration of macrophage functioning has the
potential to shift the balance between protective and adverse effects.
Epithelial secretions in response to injury may recruit scavenger cells such as
polymorphonuclear leukocytes, which can biotransform inhaled agents. More recent data on
cellular morphometrics and interspecies differences in cell populations (Mercer and Crapo,
1987; St. George et al., 1988) will aid in dosimetry adjustments for clearance, metabolism,
and uptake. As an example, modeling for the metabolic capacity of the human lung instead
of considering it only as a physical barrier can result in disparate estimates of extrapulmonary
dose (see Section 3.2). Estimates from models that account for respiratory tract metabolism
may better fit experimental data on systemic dose surrogates for some chemicals.
Concurrent with the action of inhaled agents upon critical cell types in the respiratory
tract, a portion of the dose in the PU region is likely to be transported across the alveolar
epithelium and enter systemic circulation. Changes in permeability can result from the action
of some of the mediators and proteases mentioned. The greater the amount reaching the
systemic circulation, the greater the likelihood for adverse effects in other systems (e.g.,
3-26
-------
p. 23
liver, kidney, central nervous system). The rapidity and extent to which systemic absorption
occurs and the time-to-steady-state blood levels are influenced by (1) ventilation rates and
airway mechanics, (2) blood transit time in capillary beds (i.e., perfusion limited),
(3) metabolic conversion in the respiratory tract and other organs, (4) alveolar surface area,
(5) thickness of the air-blood barrier, and (6) the blood:air and blood:tissue partition
coefficients. Many of these factors vary among species and, therefore, should be considered
in key study identification.
After the inhaled agent enters systemic circulation, the liver may produce additional
metabolites that, if the half-life is sufficiently long, may re-enter the lungs and exacerbate the
portal-of-entry effects or produce additional adverse effects (Boyd and Statham, 1983; Yost
et al., 1989). Some other agents, that do not require bioactivation, have been shown to
damage the lung when applied systemically (Kehrer and Kacew, 1985).
Metabolism
The effect of respiratory tract metabolism on the toxicity of inhaled materials is thought
to be important for many chemicals because (1) high concentrations of xenobiotic
metabolizing enzymes occur in the nose and substantial concentrations occur in the lower
respiratory tract; (2) the respiratory tract tissues are the first exposed to inhaled chemicals and
are exposed to the highest concentrations (barring tissue-specific uptake); (3) the products of
respiratory metabolism may have different toxicities from those of hepatic metabolism; and
(4) tissues at risk to toxic metabolites formed in the respiratory tract are different from those
formed in the liver (Dahl et al., 1988). The metabolic capacity of the lower respiratory tract
has been recognized for many years and nasal metabolism has recently been shown to be
significant for some compounds (Dahl et al., 1988). Accordingly, it is useful to consider that
inhaled chemicals may be extensively metabolized in the nose or in the lower respiratory tract
and both the metabolites and the parent compound may be cleared via the blood or by
mucociliary clearance.
Metabolism of potentially toxic inhaled compounds is achieved by a variety of enzyme
reactions involving oxidation, reduction, hydrolysis, and conjugation. The enzymes may
work individually, concurrently, or consecutively to detoxicate or, in some cases, activate
inhaled foreign compounds (Ohmiya and Mehendale, 1984; Minchin and Boyd, 1983; Dahl
3-27
-------
p. 24
et al., 1987). These enzymes may vary in activity across species and organs (Ohmiya and
Mehendale, 1984; Ziegler, 1980; Tynes and Hodgson, 1985; Plopper et al., 1983; Litterst
et al., 1975). Depending on the chemical being metabolized, each of these enzymes may
play a role in either an activation or detoxication pathway. The balance between activation
and detoxification governs the rate of delivery of bioactive metabolite to the macromolecular
target site (Dahl et al., 199la).
The oxidation and reduction reactions are catalyzed primarily by the cytochrome P-450
and flavin-containing monooxygenases (FAD-MO). The cytochrome P-450 isoenzymes are
ubiquitous hemoproteins located in the endoplasmic reticulum of a variety of cells and are
responsible for the oxidation of foreign compounds. Isoenzyme specificity, inducibility,
catalytic activity, and localization in the rabbit and rat lung (Domin and Philpot, 1986;
Vanderslice et al., 1987) have been elucidated. Until recently, it was thought that the
cytochrome P-450 isoenzymes were the only primary monooxygenases in the lung.
However, recent studies have shown that the FAD-MO play an important role in detoxication
of foreign compounds. FAD-MO have also been demonstrated to exist in various isoenzymic
forms, with substrate specificity and mechanisms different from those of cytochrome P-450
(Ziegler, 1988).
The Clara cells lining the respiratory and terminal bronchioles are thought to be the
primary site of cytochrome P-450 because of the presence of endoplasmic reticulum.
However, the ultrastructure of the Clara cell varies across species (Plopper et al., 1980).
In the ox, cat, and dog, more than 60% of the cytoplasmic volume is glycogen with a
relatively small proportion of the cell volume containing endoplasmic reticulum or
mitochondria. Therefore, species differences in Clara cell ultrastructure can be reflected in
significant differences in xenobiotic metabolism potential (Plopper et al., 1983; St. George
et al., 1988). Differences in localization of cytochrome P-450 activity have been suggested
as a likely basis for some differences in respiratory tract toxicity (O'Brien et al., 1985).
Epoxide hydrolases and carboxy esterases are hydrolytic enzymes found in both the
nasal cavity and lower respiratory tract tissues. The epoxide hydrolases further metabolize
potentially toxic oxidation products after initial cytochrome P-450-dependent metabolism of
aromatic compounds or alkenes. The carboxy esterases hydrolyze carboxylic esters to the
respective alcohols and carboxylic acids. At least two types of aldehyde dehydrogenases have
3-28
-------
p. 25
been detected in the nasal cavity and may be important in modifying the toxicity of volatile
aldehydes such as formaldehyde and acetaldehyde (Casanova-Schmitz et al., 1984).
Aldehyde dehydrogenase also occurs in the lower respiratory tract, particularly in the Clara
cells of the distal bronchioles.
Individually or in concert with the cytochrome P-450 isoenzymes, conjugation reactions
are catalyzed by the glutathione-5-transferases that transform potentially toxic parent
compounds or activated metabolites into nontoxic water soluble compounds. The glutathione-
5-transferases may catalyze conjugation reactions with toxic metabolites formed by the
cytochrome P-450, rendering them harmless and easier to excrete from the body. However,
GSH conjugation with certain substrates (e.g., 1,2-dibromoethane and several other related
haloalkenes) has been shown to provide reactive species capable of producing nephrotoxicity
(Monks and Lau, 1989). The cofactor required for these reactions is glutathione (GSH).
The GSH is synthesized in the lung, as well as in other major organs, and also is reduced
from the oxidized state (GSSG) to the reduced state (GSH) by GSH reductase. Under
extreme conditions of GSH depletion in the lung, it has been hypothesized that
extrapulmonary GSH is mobilized and transported to the lung from the liver (Berggren et al.,
1984). The GSH has been identified in isolated Type II epithelial cells, Clara cells, and
ciliated cells of the lung, but it is not known if it is present in all pulmonary cells. The GSH
also is the cofactor utilized by the enzyme GSH peroxidase. The GSH peroxidase catalyzes
the metabolism of hydrogen peroxide and organic peroxides formed by the ozonization of
unsaturated fatty acids. Other key antioxidant components in the lung include ascorbic acid,
or-tocopherol, superoxide dismutase, and catalase (Massaro et al., 1988).
3.1.2 Physicochemical Characteristics of the Inhaled Toxicant
The physicochemical characteristics of the inhaled agent will influence the deposition
and retention within the respiratory tract, translocation within the respiratory system,
distribution to other tissues, and ultimately, the toxic effect. Therefore, it is important to
consider characteristics of the inhaled agent as well when attempting to evaluate and
extrapolate the effects of a particular exposure.
3-29
-------
p. 26
3.1.2.1 Particles
For a given particle exposure, the two most important parameters determining
deposition are the mean diameter and the distribution of the particle diameters. The size,
density, and shape of the particles influence their aerodynamic behavior and, therefore, their
deposition (Raabe, 1979; U.S. Environmental Protection Agency, 1982b, 1986c). The
definition of diameter for a spherical particle is unambiguous, but for irregularly shaped
particles, a variety of definitions exist. Nonspherical particle size often is described by its
aerodynamic properties. Fibrous material may be described by actual length, actual diameter,
coil length, coil diameter, aspect ratio, or coil-to-aspect ratio.
Information about particle size distribution aids in the evaluation of the effective inhaled
dose (Hofmann, 1982). Recommendations defining the particle size ranges for inspirability to
the various regions have been published by an ad hoc working group of the International
Standards Organization (1981). Particle diameter and size distribution should be provided to
the risk assessor to completely characterize the aerosol in order to estimate respiratory tract
deposition with any confidence and to evaluate relevance to toxicologic potential. Appendix
H provides definitions of particle size diameters and distributions. Appendix G presents a
dosimetry model that accounts for interspecies differences in regional respiratory tract
deposition and illustrates the influence of particle size and distribution on deposition.
3.1.2.2 Gases and Vapors
The deposition site and rate of uptake of a volatile chemical are determined by its
reactivity and solubility characteristics. Therefore, the pharmacokinetics of gases and vapors
are governed by
• Rate of transfer from the environment to the tissue,
• Capacity of the body to retain the material, and
• Elimination of the parent compound and metabolites by chemical reaction,
metabolism, exhalation, or excretion.
As mentioned in Section 3.1.1.1, the transport processes in the liquid and tissue layers
adjacent to the airway lumen influence the relationship of the gas with the air-liquid
3-30
-------
p. 27
boundary. Physicochemical characteristics of the gas that contribute to the relative
importance of these processes include its chemical reactivity and solubility.
The chemical reactions of the gas with both the liquid and tissue layers may be
important. For example, reactions with the liquid layer could result in an increased flux
from the airway but reduce (relative to no reactions) the delivery of the gas to the tissue.
If the gas is the only toxic molecule, then this reaction would protect the tissue. Conversely,
if the reaction products are toxic, then reactions with the tissue layer would increase the
delivery of toxic molecules to the tissue (Overton, 1984). Chemical reactivity with the
biological constituents of the tissue is similarly important to the gas's toxic potential to the
respiratory tract tissue-and to the amount of gas and reaction products that enter the blood for
potential extrarespiratory toxicity. Theoretically, knowledge of all the chemical species
involved and the reaction rates of the reactants and products is necessary to characterize a
system for dosimetry. Sometimes the complexities may be reduced into relative
classifications (e.g., slow, fast, instantaneous) using approximation techniques for time and
spatial dependence (Overton and Miller, 1988).
Gases that are not soluble or reactive are relatively inert to the airways and penetrate to
the alveoli. Examples are nitrogen and volatile hydrophobic chemicals. The major factor
driving the uptake of these gases is the removal of the gas from alveolar air by capillary
blood. The concentration in alveolar air and capillary blood is generally considered to reach
equilibrium. Therefore, uptake of alveolar gases depends on blood:air partitioning,
ventilation/perfusion ratio, and air and blood concentrations.
For gases that are soluble, uptake is linearly related to solubility (Overton and Miller,
1988). There are many different expressions for the solubility of gases, differing in terms of
units as well as in terms of what chemical form of the gaseous species in the liquid phase is
related to the gas-phase quantities. As long as the concentration of dissolved gas is small,
and the pressure and temperature are not close to the critical temperature and pressure, then
Henry's Law is obeyed (Overton and Miller, 1988). It should be noted that the Henry's Law
constant is independent of chemical reactions so that it refers to the parent molecular form of
the gas in water and air, and not the total quantity absorbed in water to air quantities.
Considering the importance of chemical reactions as described above, solubilities as indicated
by Henry's Law constants may not be appropriate to fully describe uptake. Further,
3-31
-------
p. 28
extrapolation of Henry's Law constants from water data to biological fluids and tissues is not
always appropriate, particularly for organic compounds.
Because uptake and disposition of inhaled vapors and gases are driven by the
equilibration of their partial pressures in tissues with their partial pressures in ambient air,
solubility may be aptly described by Ostwald solubility coefficients at body temperature.
Ostwald solubility coefficients and partition coefficients (concentration ratios of the volatile
chemical in two phases with equilibrated partial pressures) have the same values (Fiserova-
Bergerova et al., 1984). Partition coefficients are essentially a measure of the affinity of a
chemical for one medium compared to another at equilibrium. The blood:air (or blood:gas)
partition coefficient is a critical determinant in the uptake and achieved blood concentration of
volatile organic chemicals (Dahl et al., 199la). Absorption generalizations based on
molecular weight are not recommended. As an example, the difference in solubility between
methanol and ethane, which have similar molecular weights, is a result of the presence of the
hydroxyl group on methanol. Interspecies comparisons necessitate consideration of the
effects of the differences in anatomy and physiology described previously, but it can
generally be stated that the less water soluble and less reactive the gas, the more similar the
deposition will be between humans and laboratory animals. The tissue:gas partition
coefficient of a chemical has been shown to correlate with its fat:gas and blood:gas partition
coefficients so that linear correlation equations may provide a useful means of estimating
tissue:gas and blood:gas partition coefficients (Fiserova-Bergerova and Diaz, 1986).
Similarly, the fat:air partition coefficient can serve as an index of whether high
concentrations of the chemical will occur in the fat. The fat compartment plays an important
role in accumulating and storing lipophilic chemicals both during and after exposure. The
chemical stored in fat becomes available for redistribution by the systemic circulation after
the end of exposure when the arterial blood concentration decreases relative to the fat. This
"postexposure" phenomenon due to fat solubility can be an important factor influencing the
amount of chemical metabolized, because that chemical that leaches from the fat compartment
after exposure is available for metabolism, which can continue for a signifiant period of time
after removal from the exposure atmosphere. Therefore, interspecies differences in body fat
can induce interspecies differences in uptake, distribution, accumulation, and toxicity of
lipophilic chemicals.
3-32
-------
p. 29
Metabolism of the parent compound can modulate uptake of inhaled gases from the
respiratory tract and is also probably the most important determinant of tissue dosimetry when
metabolites are the toxic moiety. The cells and tissues at risk from toxic metabolites depend
not only on the source of the metabolites but also on their kinetic properties. The toxic
effects of metabolites that react at fast rates are confined to the activating enzyme or cell.
If metabolite reaction rates are moderate, effects will largely be restricted to the activating
tissue and to nearby tissues. Slow-reacting metabolites may themselves be potential substrates
for further metabolism.
The effect of concentration and exposure time on the above parameters of reactivity and
metabolism should be addressed. Uncatalyzed reactions follow pseudo-first-order kinetics if
the gas is inhaled at "low" concentrations (Overton and Miller, 1988). "High" vapor
concentrations can qualitatively change the chemical fate and toxicity. Depletion of
biological coreactants, or just an increase in the concentration of the chemical to the point at
which reactions can no longer be treated as pseudo-first-order, may qualitatively change the
fate and potentially the toxicity of an inhaled gas. For chemicals metabolized according to
Michealis-Menten kinetics, metabolism may be saturated at high concentrations and become
described by zero-order kinetics. Further, saturation of metabolic pathways can alter the
metabolites formed and the resultant toxicity of the metabolized compound.
Such effects of inhaled vapor concentration on metabolism are not limited to systemic
enzymes, but also occur in localized areas within the respiratory tract. In general, the
concentrations of inhalants in the respiratory tract mucus will be higher than anywhere else in
the body, barring selective tissue uptake. Therefore, the xenobiotic metabolizing enzymes of
the respiratory tract will reach maximum reaction velocities at inhaled concentrations far
lower than those needed to bring extrarespiratory (systemic) enzymes to maximum velocities.
Therefore, it is likely (except at extremely low inhaled gas concentrations) that local
metabolizing areas within the respiratory tract, particularly the nasal tissues, will not follow
linear enzyme kinetics (Dahl, 1990).
The physicochemical gas characteristics of reactivity and solubility will interact with
physiologic parameters such as pulmonary ventilation, cardiac output (perfusion), metabolic
pathways, tissue volumes, and excretory capacities. The relative contribution or interaction of
these is, in turn, affected by the exposure conditions (concentration and duration), so that as
3-33
-------
p. 30
emphasized previously, integration of these various factors is necessary to estimate the
deposited (on airway surfaces) and absorbed doses in order to assess toxicity.
3.2 MODELING COMPARATIVE DOSIMETRY OF INHALED
TOXICANTS
The preceding discussion provides an overview of the various factors that affect the
disposition (deposition, uptake, distribution, metabolism, and elimination) of inhaled
toxicants. Major determinants include (1) the respiratory tract anatomy and physiology and
(2) the physicochemical characteristics of the inhaled toxicant. The relative contribution of
each of these factors is a dynamic relationship. Further, the relative contribution of these
determinants is also influenced by exposure conditions such as concentration and duration.
As discussed in Chapter 1, a comprehensive description of the exposure-dose-response
continuum is desired for accurate extrapolation from experimental conditions and dose-
response assessment. Therefore, an extrapolation model should incorporate all of the various
deterministic factors described in the previous section into a computational structure.
Clearly, many advances in the understanding and quantification of the mechanistic
determinants of chemical disposition, toxicant-target interactions, and tissue responses are
required before an overall model of pathogenesis can be developed for a specific chemical.
Such data do exist to varying degrees, however, and may be incorporated into less
comprehensive models that nevertheless are useful in describing delivered doses or in some
cases, target tissue interactions.
Because much information on the mechanistic determinants of target tissue dose,
toxicant-target interactions, and tissue responses is likely lacking for any given chemical to
which this RfC methodology will be applied, the default dosimetry adjustments are derived
from models that incorporate only the major determinants of chemical disposition. The
defaults are determined categorically for particles versus gases, and within gases, for those
more reactive (defined as including local metabolism) and soluble than nonreactive and
insoluble. It is recognized, however, that these are default dosimetry models, so that use of
models that incorporate a more comprehensive description of the exposure-dose-response
continuum may take precedence when such a model is judged to provide a more accurate
3-34
-------
p. 31
description. The next sections describe the rationale for the default models and dosimetry
adjustments provided in detail in Chapter 4 and the Appendices G, I, and J. Examples of
more robust models are provided to illustrate considerations of the appropriateness of the
default versus alternative model structures. The summary for this section then provides
considerations for judgement of the relative value of different modeling structures. This
judgment may be based on whether the structure of the alternative model is superior to that of
the default, (e.g., incorporates additional known mechanistic determinants) or if it empirically
results in a better correlation between "dose" and "effect".
3.2.1 Particle Deposition Model Based on Available Data
The preceding discussion in this chapter described the various mechanisms and
anatomical dependencies of deposition in the respiratory tract. A theoretical model to
describe deposition would require detailed information on all of these parameters (e.g., exact
airflow patterns, complete measurements of the branching structure of the respiratory tract,
pulmonary region mechanics) across the various species used in toxicity studies.
As described in Appendix G, an empirical model was instead developed due to the limited
availability of these types of data. An empirical model is a system of equations fit to
experimental data. Measurement techniques for deposition are such that deposition can be
defined only for the major respiratory tract regions (i.e., ET, TB and PU) and not for
localized areas such as the respiratory versus olfactory epithelium. The choice of the
experimental data and description of the model are provided in Appendix G.
The default model used in the RfC methodology estimates regional deposition. "Dose"
may be accurately described by deposition alone if the particles exert their primary action on
the surface contacted (Dahl et al., 1991a), but since the RfC is defined as a dose-response
estimate for chronic exposures, a more appropriate dose metric for particle exposures may be
to take into account clearance of the deposited dose and thereby calculate the retained dose
and the dose rate to extrarespiratory tissues. Incorporation of clearance kinetics into the
dosimetric adjustments awaits development of data enabling comparable modeling of
clearance across species. Often the physicochemical properties or mechanisms of action of
the inhaled toxicant (particle or gas) can be used to gauge the relative importance of the
various factors controlling inhaled dose. For example, the model of Yu and Yoon (1990) for
3-35
-------
p. 32
diesel exhaust incorporates clearance components such as transport of deposited particles to
the lymphatic system. A model that described the retained dose for diesel particles was
necessary because the toxicity is related to particle overload.
3.2.2 Gas Categorization Scheme Directs Default Gas Modeling
Numerous model structures have been used to describe toxicant uptake in the respiratory
tract. The type of model often reflects the physicochemical characteristics of the gases to
which they are applied. For example, the model of Miller et al. (1985) for the respiratory
tract uptake of ozone (highly reactive and moderately water soluble) is a detailed, distributed
parameter model. Key elements incorporated into this convective-diffusion-chemical reaction
model include (1) anatomic dimensions of the airspace and tissue thickness (2) dispersion in
the airspace, (3) reactivity in the liquid lining (mucus or surfactant) covering the cells of the
lower respiratory tract, and (4) lateral mass transport resistance from the airspace to the blood
(Overton et al., 1987). Models for highly reactive and highly soluble gases (e.g.,
formaldehyde, hydrogen fluoride) have emphasized the requirement to account for scrubbing
of the gas from the airstream by the upper respiratory tract (Aharonson et al., 1974; Morgan
and Frank, 1977; Morris and Smith, 1982; Hanna et al., 1989; Cassanova et al., 1991).
Such models are not applicable to a nonreactive gas such as styrene, however.
The chemical-specific or class-specific nature of these models has been dictated by the
physicochemical characteristics of the subject gases, and therefore, any single model is not
applicable to the broad range of gases that the RfC methodology must address. Dahl (1990)
categorized gases as stable, reactive, or metabolizable based on their thermodynamic and
kinetic properties. Various concepts of "dose" can be related to these properties and the
mechanism of action (e.g., macromolecular bound fraction as dose for reactive gases versus
inhaled dose for stable asphyxiants). A gas categorization scheme was constructed based on
physicochemical characteristics as determinants of gas uptake as shown in Figure 3-9.
A similar scheme has been developed by the International Commission on Radiological
Protection (1993). The definition of reactivity includes both the propensity for dissociation as
well as the ability to serve as a substrate for metabolism in the respiratory tract. The scheme
does not apply to inert gases that exert their effect by reversible "physical" interactions of
3-36
-------
p. 33
Reactivity.
Gas Category Scheme
Category 1: Do not penetrate to blood
(e.g., highly water soluble/
rapidly reactive)
Category 2: Water soluble/Blood
accumulation
Category 3: Water insoluble/
Perfuslon limited
Location
• Extrathoracic absorption
ffl Entire tract absorption
D Predominantly pulmonary
absorption
Figure 3-9. Gas categorization scheme based on water solubility and reactivity as major
determinants of gas uptake. Reactivity is defined to include both the
propensity for dissociation as well as the ability to serve as a substrate for
metabolism in the respiratory tract. Definitive characteristic of each
category and anticipated location (region) for respiratory tract uptake are
shown.
gas molecules with biomolecules (e.g., "displacement" of oxygen by carbon dioxide).
Consideration of this mechanism was discussed in Section 2.1.2.3.
The dominant determinants are used to construct a conceptual framework that directs
development of the default dosimetry model structures discussed in Appendices I and J.
These model structures are reduced further by simplifying assumptions to forms requiring a
minimal number of parameters in order to derive the default adjustments used in Chapter 4
for each category that are commensurate with the amount of data typically available for RfC
chemicals.
3-37
-------
p. 34
The two categories of gases with the greatest potential for respiratory effects are
(1) gases that are highly water soluble and/or rapidly irreversibly reactive and (2) water
soluble gases which may also be rapidly reversibly reactive or moderately to slowly
irreversibly metabolized in respiratory tract tissue. The objective of the default modeling
approach is to describe the effective dose to the three major regions of the respiratory tract
(ET, TB, PU) by addressing the absorption or "scrubbing" of a relatively water soluble
and/or reactive gas from the inspired air stream as it travels from the ET to PU region. That
is, the dose to the peripheral regions (TB and PU) is affected by the dose to the region
immediately proximal. The appropriateness of assessing proximal to distal dose
representative of the scrubbing (uptake) pattern is supported by the proximal to distal
progression pattern of respiratory tract toxicity with increasing concentration that is observed
with many chemicals (Jarabek, 1994). At low concentrations of highly water soluble and/or
irreversibly reactive gases, observed effects are largely isolated to the ET region. At higher
concentrations, more severe effects occur in the ET region and toxicity is also observed to
progress to the peripheral regions. The severity of toxicity also progresses distally with
increased exposure concentrations. As for the default particle deposition model described in
Appendix G, the default gas models do not describe respiratory tract uptake in detail to the
level of local airflow distribution (e.g., respiratory versus olfactory epithelium), but they do
adequately describe the scrubbing of the chemical from the inhaled airstream on a regional
scale.
The defining characteristic for Category 1 gases is that they are so highly water soluble
and/or rapidly irreversibly reactive in the surface-liquid/tissue of the respiratory tract that
they do not develop a significant backpressure (i.e., reversal in the concentration gradient at
the gas-liquid interface) from the surface-liquid tissue phase during exhalation. Category 1
gases are also distinguished by the property that the gas does not significantly accumulate in
the blood which would reduce the concentration driving force and, hence, reduce the
absorption rate. The default model structure is based on these characteristics. Examples of
gases classified as Category 1 are hydrogen fluoride, chlorine, formaldehyde, and the volatile
organic acids and esters.
Gases in Category 2 are moderately water soluble and rapidly reversibly reactive or
moderately to slowly irreversibly metabolized in respiratory tract tissue. Ozone, sulfur
3-38
-------
p. 35
dioxide, xylene, propanol, and isoamyl alcohol are examples of Category 2 gases. The
boundaries between categories are not definitive. Some compounds may appear to be defined
by either Category 1 or Category 2 because water solubility and reactivity are a continum.
Thus, although sulfur dioxide is reversibly reactive, which would categorize it as a
Category 2 gas, it is also highly soluble such as to be a Category 1 gas. Similarly, ozone is
highly reactive yet only moderately water soluble. More explicit delineation of categories
will be made upon review of the empirical data and the predictability of the model gases that
may appear to fit more than one category.
Because they are not as reactive in the respiratory tract tissue as Category 1 gases, gases
in Category 2 have the potential for significant accumulation in the blood and thus have a
higher potential for both respiratory and remote toxicity. Thus, the model structure used to
describe uptake of these gases is a hybrid of that for Category 1 and Category 3. The PBPK
model component of the structure is necessary to evaluate the steady-state blood concentration
which allows calculation of both absorption flux on inhalation and the desorption flux during
exhalation. The derivation of the model structures and their reduction to forms with a
minimal number of parameters are described in detail in Appendix I.
Gases in Category 3 are relatively water insoluble and unreactive in the ET and TB
surface liquid and tissue and thus result in relatively small dose to these regions. The uptake
of Category 3 gases is predominantly in the pulmonary region and is perfusion limited.
Styrene is an example of a Category 3 gas. The site of toxicity of these gases is generally at
sites remote to the respiratory tract and a compaitmental approach can be used to describe
distribution to various systemic tissues. Thus, the default model for Category 3 gases is
similar in structure to the PBPK model used by Ramsey and Andersen (1984) to describe
styrene distribution. The model structure and the derivation of the default dosimetric
adjustment based on this model are described in detail in Appendix J.
3.2.3 Summary Considerations for Judging Model Structures
Although a comprehensive description of the exposure-dose-response continuum is
desired for accurate extrapolation from experimental conditions and dose-response assessment,
often the data base is inadequate. The preceding chapter discussion illustrates that data on the
mechanistic determinants of chemical disposition, toxicant-target interactions, and tissue
3-39
-------
p. 36
responses vary in degree of availability for chemicals and species. Depending on the relative
importance of these various determinants, models with less detail may be used as a default to
adequately describe differences in dosimetry for the purposes of interspecies extrapolation
often required for the chemicals at which the RfC methodology is directed. The default
dosimetry models incorporated in the methodology represent structures that are commensurate
with the available data for both chemical-specific (e.g., reactivity and solubility) and
species-specific (e.g., respiratory tract airway dimensions, surface areas, ventilation rates,
deposition data, distribution of cell types, metabolic capacities) parameters.
An understanding of the basis for model structures also allow development of a
framework for the evaluation of whether an alternative model structure is considered optimal
relative to the default. For example, an alternate model structure might be considered more
optimal than the default for extrapolation when default assumptions or parameters are
replaced by more detailed, biologically-motivated descriptions or actual data, respectively.
For example, a model could be considered more optimal if it incorporates more chemical or
species-specific information or if it accounts for more mechanistic determinants. These
considerations are summarized in Table 3-6.
TABLE 3-6. HIERARCHY OF MODEL STRUCTURES FOR DOSJDVIETRY AND
INTERSPECIES EXTRAPOLATION
Optimal model structure
Structure describes all significant mechanistic determinants of chemical disposition,
toxicant-target interaction, and tissue response
Uses chemical-specific and species-specific parameters
Dose metric described at level of detail commensurate to toxicity data
Default model structure
Limited or default description of mechanistic determinants of chemical disposition,
toxicant-target interaction, and tissue response
Uses categorical or default values for chemical and species parameters
Dose metric at generic level of detail
3-40
-------
p. 37
The sensitivity of the model to these differences in structure may be gauged by their
relative importance in describing the response function for a given chemical. A model which
incorporates many parameters may not be any better at describing ("fitting") limited response
data than a simpler model. In these instances, the principle of parsimony might dictate the
use of the simpler model.
Woodruff et al. (1992) used Monte Carlo analyses to assess the impact that structure
and parameterization of PBPK models has on output predictions and variability.
Nonphysiologically based (NPB) models of three or two compartments were compared with
PBPK models that either used five compartments (PBPK5) to describe the body (well-
perfused, poorly-perfused, fat, bone marrow, and liver tissue compartments) or that "lumped"
the body into three (fat, bone marrow, and central) compartments (PBPK3). Comparisons
were run for different data sets from inhalation to benzene. The two main influences on
variability of model output predictions were (1) the quantity and type of data used to calibrate
the model and (2) the number of parameters in the model. While some differences existed
between the models' average predictions when calibrated to the same experimental data, these
differences were smaller than the differences between the predictions made by the same
model fitted to different data sets. An excessive number of parameters was shown to lead to
overparameterization and cause large variability in the output. The similarities in the average
predictions of the NPB and PBPK models supported the use of NPB models in some cases.
The NPB models have fewer parameters and are potentially easier to fit. The PBPK models
did show greater reliability for extrapolation but NPB models provided reliable results with
less effort needed in fitting data when the objective was to interpolate from the current data.
Issues addressed in the review by Woodruff et al. (1992) and others (Hattis et al., 1989;
Farrar et al., 1989; Portier and Kaplan, 1989; Bois et al., 1990) regarding evaluation of the
uncertainty in input parameters and the variability of predictions due to alternate structures
and data sets, should be considered when evaluating different available model structures for
replacing the default adjustments.
3-41
-------
p. 38
4. QUANTITATIVE PROCEDURES
This chapter presents the quantitative procedures for dose-response1 assessment for
noncancer toxicity. Once key studies have been identified in the available data base for a
chemical and evaluated for adequacy and limitations in terms of experimental design and
analysis, dose-response assessment for noncancer toxicity involves the designation of the
critical effects for each individual study, dosimetric adjustment of the associated exposure
concentrations to human equivalent concentrations (HECs), and an analysis of the overall data
array of these effects for that which is most representative of the threshold region. It is this
no-observed-adverse-effect level (NOAEL) for the critical effect, together with uncertainty
factors (UFs), that is used for the derivation of the inhalation reference concentration (RfC).
An RfC has a numerical value, and hence, a quantitative nature. As discussed throughout
this document, numerous theories, assumptions, and empirical data provide the quantitative
framework for these RfC calculations. To account for inherent uncertainties in the chemical-
specific data base and essential qualitative judgments, levels of confidence are assigned,
enhancing the interpretation of a numerical RfC.
This chapter begins in Section 4.1 with a discussion of the minimum data base criteria
to develop an RfC and of how to evaluate the available data to determine that a sufficient
number of appropriate endpoints were addressed to ascertain the potential for noncancer
toxicity. Guidance on how to designate effect levels (i.e., assign exposure levels as NOAELs
or lowest-observed-adverse-effect levels [LOAELs]) is provided in Section 4.2. The
remainder of the chapter is dedicated to the dosimetric adjustments used to extrapolate the
experimental data to an HEC, according to whether the observed toxicity is in the respiratory
tract or is extrarespiratory (sites remote to the portal of entry) and according to the type of
inhaled toxicant. Conversion of experimental units to the units of the RfC (mg/m3) and
'Although the strict definitions of "dose", "response", and "effect" are recognized and discussed explicitly in
Section 1.2., the conventions of the NAS paradigm will be used in this document, with the RfC being
synonymous with a "dose-response" assessment. Thus, in the broader sense, the term "dose" may encompass
administered dose (i.e., exposure concentration), delivered dose, or target tissue dose. Likewise, "response" in
the qualitative sense, is an indication of an adverse influence regardless of whether the data were measured as
quantal, count, continuous, or ordered categorical. The reader is referred to Section 1.2 for a detailed
discussion of the general principles of dose-response and assessment for noncancer toxicity.
4-1
-------
p. 39
adjustment for discontinuous experimental exposure duration are described in Sections 4.3.2
and 4.3.3, respectively. The dosimetric adjustments for particles are discussed in Section
4.3.5 and for gases in Section 4.3.6. Duration and dosimetric adjustment of human data is
discussed in Section 4.3.7. Once the effect levels are converted to HECs, the choice of the
critical effect and principal study is made according to the guidance in Section 4.3.8. The
operational derivation of an RfC is provided and the choice of UFs and assignment of
confidence levels discussed in Sections 4.3.8.1 and 4.3.8.2, respectively.
4.1 MINIMUM DATA BASE CRITERIA
Noncancer toxicity is defined as health effects other than cancer and gene mutations that
are due to the effects of environmental agents on the structure or function of various organ
systems. Therefore, by definition, a data base for derivation of a dose-response estimate for
noncancer toxicity should ensure that both appropriate and adequate numbers of endpoints
have been evaluated.
As shown in Table 4-1, the minimum laboratory animal toxicologic data base
requirement for derivation of an RfC with low confidence is a well-conducted subchronic
inhalation bioassay that evaluated a comprehensive array of endpoints, including an adequate
evaluation of portal-of-entry (respiratory tract) effects, and established an unequivocal
NOAEL and LOAEL. For a higher confidence RfC, chronic inhalation bioassay data, two-
generation reproductive studies, and developmental studies in two different mammalian
species are usually required. Considerations related to evaluating the comprehensiveness of
the available data according to these criteria follow in Section 4.1.1. Oral data may be used,
according to the criteria and guidance provided in Section 4.1.2, when inhalation data are not
available. Typically, the level of confidence in a given RfC will be higher if it is derived
from human data and supported by laboratory animal data. A more detailed discussion of
how to assign confidence levels is provided in Section 4.3.9.2.
4.1.1 Evaluation of Comprehensiveness
Data bases vary considerably in their completeness. The rationale supporting the
minimum data base requirements for derivation of an RfC, as outlined above, is that
4-2
-------
p. 40
TABLE 4-1. MINIMUM DATA BASE FOR BOTH HIGH AND LOW
CONFIDENCE IN THE INHALATION REFERENCE CONCENTRATION (RfC)
Mammalian Data Basea
Confidence
Comments
1. A. Two inhalation bioassays in different
species
B. One two-generation reproductive study
C. Two developmental toxicity studies in
different species
2. 1A and IB, as above
3. Two of three studies, as above in 1A and
IB; one or two developmental toxicity
studies
4. Two of three studies, as above in 1A and
IB
5. One of three studies, as above in 1A and
IB; one or two developmental toxicity
studies
6. One inhalation bioassayc
High
Minimum data base for
high confidence
Medium to high
Medium to high
Medium
Medium to low
Low
Minimum data base for
estimation of an RfC
"Composed of studies published in refereed journals, reports that adhered to good laboratory practice and have
undergone final QA/QC, or studies rated by the Office of Pesticide Programs as "core-minimum". It is
understood that adequate toxicity data in humans can form the basis of an RfC and yield high
confidence in the RfC without this data base. Pharmacokinetic data that indicate insignificant distribution
occurs remote to the respiratory tract may decrease requirements for reproductive and developmental data.
bChronic data.
°Chronic data preferred but subchronic acceptable.
well-defined and well-conducted subchronic toxicity studies are generally considered to be
reliable predictors of many forms of chronic toxicity, with the notable exceptions of
carcinogenic, teratogenic, and reproductive effects. Testing is required in two different
species, in the absence of a relevant laboratory animal model, in order to address potential
species sensitivity. The additional specific requirement for adequate respiratory tract
evaluation arises from the increased potential for the portal-of-entry tissue to interact
intimately with chemicals. The observation that approximately 70% of the RfCs derived to
date (October, 1994) have been based on respiratory tract endpoints is consistent with this
increased potential.
4-3
-------
p. 41
It should be recognized, however, that for some substances, results of other studies may
suggest the possibility of effects not detected in the subchronic studies. Current toxicity
testing strategies are hierarchical sequences of tests designed to develop a profile of a
chemical's toxicity (Environ Corporation, 1985). Initial testing tiers consist of relatively
rapid, inexpensive tests designed to identify acute toxicity. This information is not directly
useful in predicting chronic adverse effects in humans, but can be used to guide decisions as
to type and extent of other testing required, such as subchronic, chronic, or reproductive
bioassays. The toxicity "profiles" or information required as a minimum data base also are
somewhat structured according to this hierarchy. The magnitude of data insufficiency varies
on a case-by-case basis and should be defined by the nature of the plausible or possible
pathogenesis processes (i.e., defined according to the possible mechanism[s] of action for the
observed effect[s]). For example, the U.S. Food and Drug Administration (1982) suggests
that if a chemical tested in a subchronic study is found to cause focal hyperplasia, metaplasia,
proliferative lesions or necrosis, then a carcinogenicity study in two rodent species is
warranted. Likewise, if reproductive effects are found, then teratology testing also should be
conducted. If acute or subchronic data demonstrate reproductive organ toxicity or neurotoxic
effects, standard 2-generation reproductive assays, developmental testing, and a neurotoxicity
battery may be required for appropriate characterization. Pharmacokinetic data that indicate
insignificant distribution to sites remote from the respiratory tract at exposure concentrations
under consideration for derivation of an RfC can mitigate the requirements for reproductive
and developmental data, except when these endpoints are suggested as potential targets by
other inhalation data. Route-to-route extrapolation of oral data, according to the criteria
provided in the next section, may provide a qualitative gauge by which to judge the relative
sensitivity of these endpoints to those under consideration for the respiratory tract or other
target tissues.
The quantitative relationships between these various endpoints and how to evaluate the
entire data array for determination of the principal studies on which to base the derivation of
the RfC are discussed in Section 4.3.8.
4-4
-------
p. 42
4.1.2 Route-to-Route Extrapolation
When the data base for a given chemical is not adequate via inhalation, route-to-route
extrapolation is often practiced by some risk assessors using empirically derived factors that
are not necessarily applicable to the case at hand. For most route-to-route extrapolations, the
lack of data, lack of ability to interpret data, and underutilization of existing data due to
insufficient models and statistics reduce or eliminate the validity of these extrapolations
(Gerrity and Henry, 1990).
Data from other routes of exposure may be useful to derivation of an RfC only when
respiratory tract effects and/or "first-pass" effects (a pharmacologic phenomenon) can be
ruled out. First-pass effects refer to the metabolism that can take place in the portal-of-entry
tissue, prior to entry into the systemic circulation. For example, after oral administration,
many chemicals are delivered to the liver via the portal vein from the gastrointestinal (GI)
tract before they enter into the systemic circulation.
The respiratory tract can also exhibit a first-pass effect after inhalation due to its various
cell types and metabolic enzyme systems. This first-pass action can alter the disposition of
the parent and metabolites, thereby modulating the dose to remote or systemic target tissues
in a route-dependent fashion. Therefore, unless this first-pass effect and dosimetry are
adequately understood, there can be substantial error introduced in route-to-route
extrapolation that does not account for these parameters. In the absence of data to determine
dosimetry via inhalation, when a chemical is thought to be susceptible to first-pass effects
(e.g., metabolized), or where a potential for portal-of-entry effects is indicated but not well
characterized (e.g., respiratory toxicity after acute exposures or skin irritation after dermal
administration), then route-to-route extrapolation for derivation of an RfC is not appropriate.
For a more detailed discussion of important parameters to consider, refer to Gerrity and
Henry (1990), the National Research Council (1986, 1987), and Pepelko and Withey (1985).
Oral toxicity data are the most common data available as alternatives to inhalation data.
Oral data should not be used for route-to-route extrapolation in the following instances:
(1) when groups of chemicals are expected to have different toxicity by the two
routes; for example, metals, irritants, and sensitizers;
(2) when a first-pass effect by the respiratory tract is expected;
4-5
-------
p. 43
(3) when a first-pass effect by the liver is expected;
(4) when a respiratory tract effect is established, but dosimetry comparison
cannot be clearly established between the two routes;
(5) when the respiratory tract was not adequately studied in the oral studies;
and
(6) when short-term inhalation studies, dermal irritation, in vitro studies, or
characteristics of the chemical indicate potential for portal-of-entry effects at the
respiratory tract, but studies themselves are not adequate for an RfC
development.
Dose-response data from other routes of exposure, such as intravenous, intraperitoneal,
subcutaneous, dermal, and intramuscular routes also may be available. Intravenous data may
provide reliable information for certain chemicals (e.g., metabolizable or stable but not
rapidly reactive) on blood levels but such information would have to be supplemented by
knowledge of the quantitative relationship between inhalation exposure concentration and
blood levels in order to be useful. The other routes generally have a much more limited
usefulness in route-to-route extrapolation because the pharmacokinetics are, in general, poorly
characterized.
The ability to perform quantitative route-to-route extrapolation is critically dependent
upon the amount and type of data available. Regardless of the toxic endpoint being
considered, a minimum of information is required to construct plausible dosimetry for the
routes of interest. This information includes both the nature of the toxic effect and a
description of the relationship between exposure and the toxic effect. Illustration for this
rationale is provided by Figures 4-1 and 4-2.
Figure 4-1 shows physiologically based pharmacokinetic (PBPK) model simulations of
the concentrations required to result in a comparable "administered dose" (mg/kg/BW) after
gavage with different vehicles (oil or water), oral exposure via drinking water, or inhalation
for different durations (6 or 24 h). For gavage and drinking water studies, dose was defined
as the total amount of chloroform entering the gastrointestinal tract. Administered dose for
inhalation studies was defined as the product of inhaled air concentration (mg/L) and the
alveolar ventilation rate (L/h). Absorption efficiency was assumed to be 100%. For a
4-6
-------
p. 44
1,000
750 -
500-
250-
<
cc
\
-250
200
160
(a)
Gavage (water, 100 mg/kg); PTDEAD - 14.3%
Gavage (oil; 100 mg/kg); PTDEAD - 5.6%
Inhalation (48.3 ppm for 6h); PTDEAD - 0.3%
/ Drinking water (337 mg/L);
PTDEAD-0.1%
I
Inhalation (12.1 ppm for 24h); PTDEAD - 0.08%
10 15
(b)
20
25
^^»
0> 120 -|
80
40
Gavage (oil; 100 mg/kg); PTDEAD - 0.05%
Gavage (water, 100 mg/kg);
PTDEAD = 0.03%
inhalation (697 ppm for 6h);
PTDEAD - 0.05%
Inklr
Drinking water (3507 mg/L);
PTDEAD = 0.11%
_
Inhalation (172 ppm for 24 h);
PTDEAD = 0.06%
10 15 20
Time (h)
25
Figure 4-1. Multiple route comparisons for (a) mice and (b) humans administered
chloroform at a dose of 100 mg/kg body weight. Actual concentrations of
chloroform in air or in drinking water used to deliver a total body burden
comparable to a gavage dose of 100 kg/mg and the percentage of liver cells
killed (PTDEAD) as a result of the exposures are labeled for each
simulation. Model simulations are of the rates of metabolism in the liver
(RAML, mg/L of liver/h) for 24 h.
Source: Corley and Reitz (1990).
4-7
-------
p. 45
Conc./Dose
Inhalation
Ingestion
Low
t
High
Slight pulmonary irritation
Slight renal tubular damage
Chronic bronchitis
(COPD)
Progressing renal damage
Changes in Ca and Vit D metabolism
Lung cancer
Anaemia
Uremia
Osteomalacia and osteoporosis
Slight renal tubular damage
Decrease in intestinal Ca
absorption
Progressing renal damage
Changes in Ca and Vit D
metabolism
Intestinal mucosa damage
Anaemia
Uremia
Osteomalacia and osteoporosis
Figure 4-2. Differential effects of inhaled and ingested cadmium with increasing inhaled
and ingested doses.
Source: Oberdorster (1990).
detailed discussion of the PBPK model structure and parameter values, refer to Corley and
Reitz (1990). The figure is used here to highlight the differences in administered dose by
various routes required to achieve the same internal dose in a target tissue, in this case, the
liver. The model predicts the percentage of hepatocytes killed in the liver due to the
metabolism of parent compound. Note the different profiles of metabolism via the different
routes in Figure 4-1. The degree of cytotoxicity predicted by the model simulations was in
the order of gavage (water) > gavage (corn oil) » inhalation (6 h) > drinking water >
inhalation (24 h). This figure also illustrates the interspecies differences in the processes
involved. For example, to produce a total body burden of chemical comparable to that
4-8
-------
p. 46
achieved by an exposure dose of 100 mg/kg/day of chloroform, the concentration of the 24-h
inhalation exposure was increased from 12 ppm in the mouse to 172 ppm for the humans.
Figure 4-2 shows the qualitative relationships of differential effects after either
inhalation or ingestion of cadmium (Cd) (Oberdorster, 1990). It is apparent that the exposure
route influences the target organ effects. Respiratory effects have only been observed after
inhalation exposures and GI tract effects only after oral exposure. Portal-of-entry effects are
obviously of importance. In contrast, remote effects such as those on the kidney, bone, and
the hematopoietic system are observed after exposure by either route. However, the portal of
entry also modulates the dose rate to the remote tissues. Using a simplified steady-state
PBPK model of a few basic transfer kinetics (e.g., 90% of inhaled soluble Cd is absorbed;
5% of ingested) for soluble Cd compounds (e.g., cadmium oxide, cadmium chloride),
Oberdorster (1990) estimated that 1 /xg/m3 of inhaled (24-h) Cd was equivalent to 21.5 jwg
(daily ingested) and 1,000 /tg (daily ingested) for renal and respiratory effects, respectively.
The great disparity in potency by different routes for Cd emphasizes the point that dosimetry
should be established for each relevant route when either contact site or remote toxicity is a
concern.
Therefore, the actual impact of exposure by different routes can only be estimated by
taking account of factors that influence absorption at the portal of entry, such as
(1) physicochemical characteristics of the chemical (e.g., dissociation state, molecular weight,
partition coefficient, reactivity, solubility), (2) exposure factors (e.g., concentration, duration,
regimen), and (3) physiologic parameters (e.g., barrier capacity as related to variability in
species, blood flow, cell types and morphology, metabolism, pH, specialized absorption sites,
storage in cells), and those parameters that influence dose remote to the portal of entry,
including metabolism, clearance, tissue binding, tissue blood flows, tissue:blood partition
coefficients, and tissue volumes.
Evaluation of the adequacy of the available data to address the factors outlined above is
the basis for the decision tree shown in Figure 4-3 (Gerrity and Henry, 1990). As discussed
above, route-to-route extrapolation for quantitative dose-response assessment should only be
considered when concern for contact-site (portal-of-entry) toxicity has been ruled out
(Option 2 of Figure 4-3 is sufficient only for hazard identification). Although the fact that
4-9
-------
p. 47
Chemical Identity
Do adequate toxicology data
exist for at least one route?
Can SAP or
analogy be used?
No
No
Collect data
(Option 1)
Yes
Use data for
analogous
compound
(Option 2)
Yes
Is toxicity
remote from
contact site
likely?
No
(Contact site
concern)
Collect data
on all relevant
routes
(Option 3)
Yes
Candidate for
route-to-route
extrapolation
(Option 4)
Figure 4-3. Decision tree for route-to-route extrapolation (see text below for a
discussion of the options listed). SAR = structure activity relationship.
Source: Gerrity and Henry (1990).
the effect of a chemical is observed in the portal of entry does not necessarily preclude route-
to-route extrapolation, the requirements for quantitative data via each route (Option 3 of
Figure 4-3) in order to perform such an extrapolation usually obviate the reason (i.e., lack of
data) for which it was being considered in the first place.
If respiratory tract toxicity can be ruled out and remote site toxicity is of interest, then
route-to-route extrapolation becomes a possibility (Option 4 of Figure 4-3). Methods for
route-to-route extrapolation range in accuracy and therefore, inherent uncertainty. The
simplest approach is to use default absorption values for each exposure route appropriate to
the chemical class in question. Such values have only been developed for a limited class of
volatile organic chemicals. Because this approach entails an increased uncertainty compared
to those that use pharmacokinetic data and PBPK modeling, use of default absorption values
is considered inadequate for quantitative dose-response assessment.
4-10
-------
p. 48
Direct measurement of absorption efficiency for the routes of interest is an improvement
on the use of default values, but the approach still ignores many of the potentially important
factors mentioned above, invoking additional uncertainty that would have to be accounted for
when calculating a dose-response estimate. Measurement of bioavailability by the use of a
validated internal marker provides greater certainty. Comparative excretion data when the
associated metabolic pathways are equivalent by each route and regimen of interest or
comparative systemic toxicity data when such data indicate equivalent effects by each route
and regimen of interest may also provide useful information. However, the associated
uncertainty would have to be accounted for in the estimate derived using an extrapolation
based on such data.
The preferred method for performing route-to-route extrapolation involves the
development of a PBPK model that describes the disposition (deposition, absorption,
distribution, metabolism, and elimination) of the chemical for the routes of interest (Gerrity
and Henry, 1990). Such models account for fundamental physiologic and biochemical
parameters and processes such as blood flows, ventilatory parameters, metabolic capacities,
and renal clearance, tailored by the physicochemical (e.g., blood:air and tissue:blood
partitions) and biochemical properties (e.g., binding, depletion of co-factors) of the chemical
in question.
The use of a PBPK model is predicated on the assumption that an effective (target-
tissue) dose achieved by one route in a particular species is expected to be equally effective
when achieved by another exposure route or in some other species. A key determination is
the choice of the dose surrogate for the toxic effect. The more accurately the exposure-dose-
response continuum is characterized, and therefore the correlation of the chosen dose
surrogate with toxic effect, the more accurate this approach will be with respect to use in
quantitative extrapolation. For example, a measure of target-tissue dose for a chemical with
pharmacologic activity could be the tissue concentration divided by some measure of the
receptor-binding constant for that chemical. The behavior of a substance administered by a
different exposure route can be determined by adding equations that describe the nature of the
new input function. Similarly, because known physiologic parameters are used, different
species (e.g., humans versus laboratory animal species) can be modeled by replacing the
appropriate constants. It should be emphasized that PBPK models must be used in
4-11
-------
p. 49
conjunction with toxicity and mechanistic studies in order to relate the effective dose
associated with an effect for the test species and conditions to other scenarios. The use of an
existing model structure, essentially a template, can greatly reduce the effort required for
model development of analogous chemicals.
4.1.3 Not-Verifiable Status
When the available data do not meet the minimum data base requirements as discussed
above or when the existing data can not be synthesized into a compelling toxicity profile
without great uncertainty (see Section 4.3.8), the data base on a given chemical is designated
as "not-verifiable" and no RfC estimate is calculated. This status would require reanalysis
when new data become available.
4.2 DESIGNATION OF EFFECT LEVELS
The designation of effect levels, or the association of adversity2 with exposure
concentrations, is one of the most difficult procedures of any dose-response analysis for
noncancer toxicity. The critical effect for an individual study is often described as either the
adverse effect that first appears in the dose scale as dose is increased, or as a known
precursor to the first adverse effect. The premise of this designation is the underlying
threshold phenomenon and it assumes that if this critical effect is prevented then all observed
adverse effects at subsequent concentrations are prevented. In the simplest terms, a NOAEL
and a LOAEL are determined for the specified adverse effect from the exposure levels of a
given individual study for each of the various species tested. The NOAEL is the highest
level tested at which the specified adverse effect (i.e., a biologically and statistically defined
adverse effect) is not produced and is, thus, by definition, a subthreshold level (Klaassen,
1986). The NOAEL/LOAEL is a function of the exposure levels used in the experimental
2Here adverse effects are considered to be functional impairments or pathological lesions that may affect the
performance of the whole organism or that reduce an organism's ability to cope with an additional challenge
(Federal Register, 1980). One of the major problems encountered with this concept is the reporting of
"observed effect levels" as contrasted to "observed adverse effect levels." The terms "adverse" and "not
adverse" are at tunes satisfactorily defined, but because more subtle responses continue to be identified due
to increasingly sophisticated testing protocols, scientific judgment is needed regarding the exact definition of
adversity.
4-12
-------
p. 50
design, or is a function of designating a specified health effect measure (e.g., 10% incidence
of a lesion) as the outcome of interest in the case of some alternative approaches presented in
Appendix A3, and therefore, does not necessarily reflect the "true" biological threshold.
Table 4-2 presents the four types of effect levels that may be applicable when evaluating
an individual study. Historically, the distinction between adverse effects and nonadverse
effects has been and remains problematic. For example, although disease is a dynamic
process (injury, adaptation, or healing), a pathologist records a morphologic change at a
single point in time and these "freeze-frame" data are used to determine the probable cause
and pathogenesis (past) and probable progression or outcome (future). Designation of an
effect level (i.e., the designation of adversity) requires interpretation of the data based on an
ability to deduce the preceding events that have led to the observed change and to predict the
outcome or progression. The relationship between structural alterations to altered function is
not always simple, however.
Determining whether altered morphology is an adaptive response or truly an expression
of toxicity (functional impairment) can be extremely difficult and even controversial (Burger
et ah, 1989; Ruben and Rousseaux, 1991). In some cases, structural alteration can occur,
but normal function can continue in target tissues with functional reserve such as the lung,
liver, and kidney. Not all tissues demonstrate this high reserve. The central nervous system
can compensate to only a limited degree and where the damage occurs is vitally important for
the function of the system. Therefore, "focal" damage may be adverse in some but not all
target tissues. Also, the lack of observed functional change may be due to failure to detect
subtle or unknown functional changes rather than to their absence.
A similar morphologic alteration may have both functional and physiologic significance,
but often it is difficult to differentiate toxicity from physiologic response by morphologic
means alone. Not all functional abnormalities manifest themselves morphologically.
Temporal-spatial patterns are particularly challenging when evaluating toxicologic pathology.
Problems concerning time include reversibility, adaptation versus toxicity, progression versus
3There are alternative approaches under development (presented and discussed in Appendix A) aimed at deriving
estimates of exposures that are analogous in intent to the establishment of a NOAEL. The NOAEL/LOAEL
approach outlined is not intended to discourage alternative or more sophisticated dose-response procedures when
sufficient data are available, but rather to present key issues necessarily involved (e.g., dosimetric adjustment
and data array analysis) in any approach for the assessment of noncancer toxicity.
4-13
-------
p. 51
TABLE 4-2. FOUR TYPES OF EFFECT LEVELS3 (RANKED IN ORDER OF
INCREASING SEVERITY OF TOXIC EFFECT) CONSIDERED
IN DERIVING INHALATION REFERENCE CONCENTRATIONS
FOR NONCANCER TOXICITY
NOEL: No-Observed-Effect Level. That exposure level at which there are no statistically
and biologically significant increases in frequency or severity of effects between
the exposed population and its appropriate control.
NOAEL: No-Observed-Adverse-Effect Level. That exposure level at which there are no
statistically and biologically significant increases in frequency or severity of
adverse effectsb between the exposed population and its appropriate control.
Effects are produced at this level, but they are not considered to be adverse.
LOAEL: Lowest-Observed-Adverse-Effect Level. The lowest exposure level in a study or
group of studies that produces statistically and biologically significant increases in
frequency or severity of adverse effects between the exposed population and its
appropriate control.
PEL: Frank Effect Level0. That exposure level that produces frankly apparent and
unmistakable adverse effects, such as irreversible functional impairment or
mortality, at a statistically and biologically significant increase in frequency or
severity between an exposed population and its appropriate control.
"Note that these levels represent points on a continuum and are not discrete.
bAdverse effects are defined as any effects resulting in functional impairment and/or pathological lesions that
may affect the performance of the whole organism, or that reduce an organism's ability to cope with an
additional challenge.
Trank effects are defined as overt or gross adverse effects (e.g., severe convulsions, lethality, etc.).
regression, and peracute lethal toxicity. Problems concerning space are limited to missing the
lesion completely or missing a relevant area because of sampling method. For example,
histologic examination of the nasal cavity should select four tissue sections, not one, to
achieve a thorough examination (Young, 1981). Further, due to the proximal to distal
inspiratory airstream, some examination of the upper respiratory tract is indicated when
respiratory toxicity from an inhaled irritant is evident in the lower respiratory tract.
Due to the structural-functional and temporal-spatial problems discussed above, an
approach that integrates pathological studies (ultrastructural, histochemical, cellular, and
molecular) with functional methods is recommended (Ruben and Rousseaux, 1991). Morgan
(1991) has provided guidance on the identification and interpretation of URT lesions in
toxicologic studies. A systematic but flexible approach to evaluation of lesions in the URT is
4-14
-------
p. 52
recommended, one that considers selection of section level in context with the
physicochemical characteristics of the inhaled gas (e.g., water solubility and reactivity), the
role of factors that may account for lesion distribution (e.g., dosimetry and tissue
susceptibility), and development of a pathogenesis profile or a chronological order of events
(e.g., degenerative, adaptive, and adaptive/regenerative changes versus time). The nasal
diagrams proposed by Mery et al. (in press) offer an approach to recording data and mapping
lesions that aids this type of interpretation strategy. This approach is also likely the best to
compile the data and precludes the restraint to interpretation and mathematical modeling
presented by data scored categorically for severity (e.g., + = mild, ++ = moderate; and
+ + + = severe) and/or without sufficient section detail with respect to lesion location
(Jarabek, 1994).
In the early stages of respiratory disease, there is considerable uncertainty concerning
how to differentiate between acute reversible effects, which are the immediate consequence of
an exposure episode, and potential progression to chronic, nonreversible respiratory
pathology. The boundary between adaptive and toxic responses also remains controversial for
some respiratory tract lesions (Burger et al., 1989). These are important issues both in terms
of evaluation of respiratory tract effects per se, as well as for decisions concerning the critical
effect in inhalation studies. Inhalation-specific issues such as evaluation of pulmonary
function, sensory irritation, and allergic sensitization data are discussed in Section 2.2.
Designation of effect levels usually contains an element of scientific judgment in
addition to objective criteria. Considerable experience and precedent for such decisions have
accrued over the last several years in the process of developing oral reference doses, RfCs,
and other health-related benchmark estimates. Table 4-3 presents guidance as to how general
effects would usually be designated as different (adverse) effect levels. In general, effects
that may be considered marginal are designated as adverse only to the extent that they are
consistent with other structural and functional data suggesting the same toxicity. For
example, altered liver enzymes (statistically out of normal range) would only be considered
adverse in context with altered structure (pathology) and liver weight changes.
4-15
-------
p. 53
TABLE 4-3. EFFECT LEVELS CONSIDERED IN
DERIVING INHALATION REFERENCE CONCENTRATIONS
IN RELATIONSHIP TO EMPIRICAL SEVERITY RATING VALUES
(Ranks are from lowest to highest severity.)*
Effect or No-Effect Level
Rank
General Effect
NOEL
NOAEL
NOAEL
NOAEL
NOAEL/LOAEL
LOAEL
(LO)AELb
(LO)AEL/FEL
0
1
3
4
5
No observed effects.
Enzyme induction or other biochemical
change, consistent with possible mechanism
of action, with no pathologic changes and
no change in organ weights.
Enzyme induction and subcellular
proliferation or other changes in organelles,
consistent with possible mechanism of
action, but no other apparent effects.
Hyperplasia, hypertrophy, or atrophy, but
no change in organ weights.
Hyperplasia, hypertrophy, or atrophy, with
changes in organ weights.
Reversible cellular changes including
cloudy swelling, hydropic change, or fatty
changes.
Degenerative or necrotic tissue changes
with no apparent decrement in organ
function.
Reversible slight changes in organ function.
PEL
PEL
PEL
8
9
10
Pathological changes with definite organ
dysfunction that are unlikely to be fully
reversible.
Pronounced pathologic changes with severe
organ dysfunction with long-term sequelae.
Death or pronounced life shortening.
'Adapted from DeRosa et al. (1985) and Hartung (1986).
"The parentheses around the "LO" in the acronym "LOAEL" refer to the fact that any study may have a series
of doses that evoke toxic effects of rank 5 through 7. All such doses are referred to as adverse effect levels
(AELS). The lowest AEL is the (LO)AEL.
4-16
-------
p. 54
4.3 CALCULATION OF HUMAN EQUIVALENT CONCENTRATIONS
A key element of extrapolation of laboratory animal inhalation data to humans is
estimation of the "dose" (i.e., agent mass deposited per unit surface area or tissue volume)
delivered to specific target sites in the respiratory tract or made available to uptake and
metabolic processes for systemic distribution considered with mechanistic determinants of
toxicant-target interactions and tissue responses (Martonen and Miller, 1986; Andersen et al.,
1991). To this end, PBPK and other mathematical dosimetry models have evolved into
particularly useful tools for predicting disposition differences for risk assessment (Miller
et al., 1987b). Their use is predicated on the assumption that an effective (target-tissue) dose
in a particular species is expected to be equally toxic when achieved in some other species.
However, it is likely that species differences in sensitivity occur due to such species-specific
factors as host defense, repair processes, and genetics, so that the use of a 10-fold UF to
account for intraspecies variability, despite application of dosimetric adjustments, requires
additional research.
This section outlines the methods for calculating HEC estimates by using adjustment
factors that have resulted from similar modeling efforts of species-specific dosimetry
differences. The factors are used to adjust the observed exposure effect levels (i.e.,
NOAELs, LOAELs, etc.) in laboratory animals to estimate a concentration that would be an
equivalent exposure to humans (i.e., NOAELrHECjS, LOAELrHEC,s, etc). These HECs then
are the basis for comparison and choice of the critical effect and study.
As discussed in Section 3.2, the equations presented in this chapter are default
adjustments based on dosimetry models that incorporate only the major determinants of
particle or gas disposition. The use of models that may incorporate a more comprehensive
description of the exposure-dose-response continuum is considered the optimal approach in
each case. It should also be noted that because PBPK models allow for explicit handling of
intermittent exposure regimens (e.g., model can simulate 6 h/day, 5 days/7 days exposure
and predict resultant internal dose), the duration adjustment discussed in Section 4.3.2 is
obviated by the use of these models.
Figure 4-4 is a flowchart for the default calculation of HECs and provides an outline for
the contents of this section. Conversion of units from ppm to mg/m3 is required before
dosimetric adjustments can be applied. This calculation is discussed in Section 4.3.1. The
4-17
-------
p. 55
Convert
ppm to mg/m3
(Section 4.3.1)
Respiratory
Remote
(Extrarespiratory)
Adjust for
Exposure Regimen
(Section 4.3.2)
1. Evaluate
Generation
System
2. Characterize
by MMAD, ag, or
Default Values
1. Evaluate
Generation
System
2. Characterize by
Concentration,
Temperature,
Pressure, or
Default Values
Remote
(Extrarespiratory)
Identify
the Target
Effect(s)
Remote
(Extrarespiratory)
Remote
(Extrarespiratory)
/section 4.3.6.
Figure 4-4. Flowchart for calculation of human equivalent concentrations.
4-18
-------
p. 56
next step in calculating a HEC is to convert the exposure regimen of the experiment in
question to that of the human exposure scenario; that is, a continuous (24-h/day) lifetime
(70-year) exposure. The third step of the approach is to apply the dosimetric adjustments
appropriate for the type of toxicant to be assessed (particle or gas, and if a gas, what
category) and the effect to be assessed (respiratory tract or extrarespiratory toxicity) resulting
from an inhalation exposure. The default dosimetric adjustments to derive HECs for
respiratory tract effects and extrarespiratory effects of particles are provided in Section 4.3.5.
For gases, the determination of the appropriate gas category according to the scheme
provided in Section 3.2.2 is required to determine which dosimetric adjustment to apply to
calculate an HEC. Because the boundaries between the categories are not definitive (see
discussion in Section 3.2.2 and Appendix I), but instead were made to allow derivation of
default model structures, identification of the target effect(s) is used to further define the gas
category. Thus, remote (extrarespiratory) effects of Category 1 gases and respiratory effects
of Category 3 gases are treated according to the default dosimetric adjustments for each of
these respective effects of Category 2 gases (Section 4.3.5 and 4.3.6). The default
dosimetric adjustments to derive HEC values for respiratory effects of Category 1 gases are
provided in Section 4.3.5. The default dosimetric adjustment to derive HEC values for
extrarespiratory effects of Category 3 gases is provided in Section 4.3.6.
Although the presentation in this section divides the dosimetry calculations into those
applied to extrapolate respiratory tract effects versus extrarespiratory effects, it should be
recognized that there is no strict compartmentalization of effects for a chemical. A given
inhaled chemical could cause both respiratory tract and extrarespiratory effects. Therefore,
the decision on which of the equations to use in this chapter is governed by the endpoint of
interest in concert with the properties of the chemical to be assessed.
4.3.1 Conversion to Standard Units
In the rare event that investigations using paniculate exposures would report the
concentration in ppm, a mass-density relationship should be used to convert the exposure
o
concentration to mg/m . Inhalation toxicity studies on gases typically employ exposure levels
expressed as mg/m3 or ppm. Exposure levels for gases, including the NOAEL selected for
RfC derivation, should be expressed in standard units of mg/m3. For exposure levels
4-19
-------
p. 57
«5
expressed as ppm, the Ideal Gas Law should be used to derive the corresponding mg/m
level:
x _ _xx
m3 22.4L g-mole T 760mmHg m3 g
where:
ppm = concentration expressed on a volumetric basis L ,
106L
MW = molecular weight in grams,
22.4 L = the volume occupied by 1 g-mol of any compound in the gaseous state at
0 °C and 760 mm Hg,
T = actual temperature in degrees Kelvin, and
P = actual pressure in mm Hg.
At 25 °C and 760 mm Hg, 1 g-mole of a perfect gas or vapor occupies 24.45 L.
Therefore, under these conditions, the conversion becomes
mg/m3 =
4.3.2 Temporal Relationships of Toxicity and Duration Adjustment
Many inhalation toxicity studies using laboratory animals use discontinuous exposure
regimens. Often exposures are for 6 to 8 h/day and 5 days/week. Inhalation reference
concentrations are constructed to reflect a benchmark level for continuous exposure.
By extension, the RfC also is assumed to be protective for discontinuous exposures at the
same air concentration. A normalization to some given exposure (e.g., 24 h/day for a
lifetime of 70 years) is needed to adjust for the wide variety of experimental exposures to
permit comparisons between studies. As discussed earlier, the RfC proposed herein is to
reflect lifetime continuous exposure, making this scenario the objective of normalization.
Attention should be paid to the degree this scenario deviates from the experimental, and to
4-20
-------
p. 58
the physicochemical (solubility and reactivity) parameters of the inhaled agent and species-
dependent factors (e.g., distribution volumes and metabolic pathways) that might temper this
conversion.
To calculate duration-adjusted exposure levels in mg/m3 for experimental animals, the
equation is
NOAEL*[ADJ] (mg/m3) = E (mg/m3) X D (h/24 h) X W (days/7 days), (4-2)
where:
NOAEL*rADJi = the NOAEL or analogous effect level obtained with an alternate
approach as described in Appendix A, adjusted for duration of
experimental regimen;
E = experimental exposure level;
D = number of hours exposed/24 h; and
W = number of days of exposure/7 days.
NOTE: 1. This same duration adjustment is applied to LOAELs.
2. This duration adjustment is not applied when PBPK models are used (see
Section 4.3.3).
3. Duration adjustment for human data is discussed in Section 4.3.6.
The rationale for this linear prorate adjusment is that the resultant human exposure
concentration should be the concentration (C) x time (T) equivalent of the experimental
animal exposure level. This adjustment is weakly founded because steady-state conditions
may not be reached in laboratory animals for some chemicals and intermittent regimens and
because the influence of dose-rate is different for different toxicity mechanisms (e.g., an
effect mediated by peak blood concentration versus integrated tissue dose). Thus, depending
on the mechanism of action, such duration adjustment may be inappropriate. Toxic effects of
gases such as irritation, narcosis or asphyxia may be much more dependent on concentration
than duration. An attempt should always be made to take into account the mechanisms of
toxic action as related to the temporal parameters of duration and frequency, although
C x T is rarely investigated after subchronic or chronic durations. Unless more information
is available on a case-by-case basis, this default is used.
4-21
-------
p. 59
As the effect in question increases in its severity, the validity of this equation becomes
more tenuous. The toxicity of an exposure is dependent upon the character of the
"concentration-time" (C x T) curve, which may be described by a hyperbola whose arms
converge asymptotically toward the axes of the coordinates (Bliss, 1940). Bliss and James
(1966) have shown that such curves can be extrapolated with minimal error when the time
points in the experiment are located on the segment of the curve asymptotically approaching
the axes of the coordinates (i.e., high concentration acute exposures or low concentration
chronic exposures). The exposure duration should ideally embrace the time span in which the
rate of onset of specific toxic effects sharply change, reflecting the degree of arc in the curve
of the (C X T) relationship.
Fiserova-Bergerova et al. (1980), using a compartmentalized model based on first-order
kinetics, demonstrated that duration of exposure to a gas can have profound effects on the
fractions of uptake exhaled or metabolized. Concentrations in tissues reflected the
concentration fluctuations in exposure, but the fluctuation in tissues was greater during
exposure to low solubility gases than to lipid soluble vapors (blood:air partition coefficients
of 0.5 and 10.0, respectively), due to the faster equilibration of partial pressures of the low
solubility gases. Fluctuations between tissue and exposure concentrations were diminished if
the substances were metabolized. Because a toxic effect is usually related to tissue
concentration, consideration should be given to these duration and solubility effects.
Extrapolation on the basis of C x T should be attempted only if a steady-state was attained.
Likewise, linear extrapolation from one concentration exposure to another is scientifically
supportable only if all processes involved in the uptake and elimination of the inhaled agent
are first order. Differences are caused primarily by concentration-dependent metabolic
clearance.
4.3.3 Use of Pharmacokinetic and Pharmacodynamic Data
Pharmacokinetic and pharmacodynamic data (described in Section 1.2) can be used in a
range of applications, from providing adjustments to external exposures based on correlations
of exposure to effect, through gathering insight on various important mechanistic insights and
calculation of kinetic parameters, to developing a comprehensive exposure-dose-response
description that incorporates major determinants of toxicant disposition, toxicant-target
4-22
-------
p. 60
interaction and tissue response. These data can also be used to ascertain what laboratory
animal species is most appropriate, based on similarity of major mechanistic determinants, for
extrapolation to humans.
Empirical equations such as correlation equations (e.g., that relate the extent of external
exposure with the amount of internal biologic markers) can be used to describe kinetic
processes by a simple mathematical expression. These data, as described in Section 4.1.2,
may be useful as a qualitative index of uptake for a given route, but they provide no insight
into the other parameters controlling disposition of a toxicant (distribution, metabolism,
excretion) over time and therefore their use is rather restricted.
Experimental data that track the concentrations of various kinetic parameters during and
following exposures can be used to determine various measures of the intensity of tissue
exposure. The parameters that are proportional to the relevant measure of tissue exposure are
referred to as tissue dose metrics (Andersen, 1987). These metrics include estimates of time
integrals of tissue exposure to a parent toxicant or its metabolite(s) (e.g., area under the
blood [AUBC] or tissue curve [AUTC]), concentrations of these materials in tissues, or
receptor occupancy caused by the presence in tissues. This information provides little insight
into the mechanistic determinants or the biological effect of the parent or its metabolite(s).
The choice of which metric to use as an appropriate measure should be based on some
knowledge of the mechanism by which the toxic effects are induced.
This mechanistic knowledge does not necessarily have to be exhaustive, but can rather
be related to certain general aspects of the nature and causes of a particular toxic interaction.
For example, is the effect related to chemical reactivity or to occupancy of cellular receptor
molecules? Is the effect associated with the parent or with a metabolite? If it is a metabolite,
does the metabolite have a sufficiently long half-time in the body to circulate freely
throughout the body or is it so reactive that it likely produces its damage locally? Are the
effects themselves reversible cytotoxic phenomena or irreversible changes? Is there sufficient
time for the target tissue to recover from the damage within the exposure frequency interval?
If the critical damaging toxicant-target interaction is caused by direct chemical reaction
in which the toxicant reacts with and consumes cellular constituents, the degree of damage
should be related to the time integral of tissue exposure to the reactive chemical (e.g.,
AUTC). This definition would likely need to incorporate quantitative information on the
4-23
-------
p. 61
synthesis and normal catabolism of the macromolecules involved to describe chronic
exposures accurately. If the toxicant interacts with tissue by noncovalent binding to cellular
receptor molecules, the response of the cell is dependent on the occupancy of the receptor
and occupancy is determined by the binding constant for the chemical and the free
concentration of the toxicant in the cell.
The use of categorization schemes based on the physicochemical properties or
mechanisms of action of the inhaled toxicant have been proposed and different concepts of
"dose" related to these (National Research Council, 1986; Andersen, 1987; O'Flaherty, 1989;
Dahl, 1990). Considerations such as these are described in Section 3.2. and went into the
development of the default dosimetry adjustments provided in the following Sections 4.3.5
through 4.3.7. Details on the development of the dosimetry models are provided in
Appendices G, I and J. The default adjustments are determined categorically for particles
versus gases, and within gases, for those more reactive and soluble than nonreactive and
insoluble. Reactivity is defined to include both the propensity for dissociation as well as the
ability to serve as a substrate for metabolism in the respiratory tract. Because these are
default dosimetry adjustments, the use of models that may incorporate a more comprehensive
description of the exposure-dose-response continuum is considered the optimal approach for
extrapolation to HECs when such a model is judged to provide a more accurate description.
This judgment may be based on whether the structure of the alternative model is superior to
that of the default, (e.g., incorporates known mechanistic determinants) or if it empirically
results in a better correlation between "dose" and "effect". The reader is referred to
Section 3.2 for a discussion of modeling comparative dosimetry and to Section 3.2.3 for
summary considerations regarding judging model structures.
Use of more comprehensive models obviate the need for the duration adjustment
described above in Section 4.3.2 because such models employ parameters that describe time-
dependent determinants of toxicant disposition such as metabolic clearance, distribution
volumes and elimination constants. These models can therefore be used to simulate both the
experimental exposure regimen as well as the exposure scenario for the human. PBPK and
linear pharmacokinetic models have both been used to evaluate and to adjust for different
work place exposure durations (Droz, 1985; Andersen et al., 1987b; Saltzman, 1988). For
example, in order to extrapolate laboratory animal data using a PBPK model, the laboratory
4-24
-------
p. 62
animal regimen (e.g., 6 h/day, 5 days/week) is simulated and the resultant appropriate dose
metric (e.g., AUTC) calculated. This is done assuming steady-state conditions for chronic
studies if it is likely that these conditions were met for 90% of the time (see Section 4.3.5),
or the entire exposure can be simulated with the model. The model is then used with the
human parameters to ascertain the exposure concentration that results in an equivalent dose
metric under the human exposure scenario (e.g., 24 h/day). This exposure concentration
back-extrapolated from the equivalent dose metric is the HEC.
4.3.4 Default Dosimetric Adjustment and Physiological Parameters
As described in Sections 3.2 and 4.3.3., the dosimetric adjustment factors described in
in the following sections are default approaches to be used when more sophisticated or
chemical-specific models are not available. The HEC is calculated with the default
dosimetric adjustment factor as:
NOAEI/pjBcj (mg/m3) = NOAEL*[ADJ] (mg/m3) X DAFr (4-3)
where:
NOAEL*[HECj is the NOAEL or analogous effect level obtained with an alternate
approach as described in Appendix A, dosimetrically adjusted to an HEC,
NOAElA^j] is defined in Equation 4-2, and
DAFr is a dosimetric adjustment factor for respiratory tract region, r (ET, TB, PU, TH,
or TOTAL), either the regional deposited dose ratio (RDDRj) for particles or the regional gas
dose ratio (RGDRr) for gases.
The DAF represents a multiplicative factor used to adjust an observed exposure
concentration in a particular laboratory species to an exposure concentration for humans that
would be associated with the same delivered dose. The calculation of the RDDRr for
particles and the RGDRr for gases is described in section 4.3.5 and 4.3.6, respectively.
Depending on whether the observed toxicity is in the respiratory tract or at remote
(extrarespiratory) sites, the DAFr is used in conjunction with default normalizing factors for
the physiological parameter of interest. Because insoluble particles deposit and clear along
the surface of the respiratory tract, dose per unit surface area is a commonly used
4-25
-------
p. 63
normalizing factor for respiratory effects due to paniculate deposition; body weight is often
used to normalize dose to remote target tissues. In some cases, it may be appropriate to
normalize by regional volumes or target organ weights. For gases, use of mass flux (mass
per surface area-time) is considered a reasonably accurate predictor of the peak localized
concentration driving the absorption gradient for respiratory tract effects. For example, if the
observed toxicity is in the TB region, the dose deposited in that region for each species is
normalized to the TB surface area for each species.
Default values of surface area (SA) for the various respiratory tract regions of five
commonly tested animal species are provided in Table 4-4. Selection of the values was based
on a meeting of experts in laboratory animal and human morphometric measurements
convened in August 1991 (Jarabek, 1991). At that time, a thorough review of the literature
had been conducted and the group was presented with summary tables of surface area
measurements; animal information (as available) including strain, body weight, sex and age;
tissue preparation, and morphometric measurement technique. Based on discussion among
the expert group members, values were identified as most representative of a species and
designated as the default. These values do not always correspond exactly to the published
value that is cited in Table 4-4, most generally due to rounding.
TABLE 4-4. DEFAULT SURFACE AREA VALUES FOR
RESPIRATORY EFFECTS3
Human
Mouse
Hamster
Rat
Guinea Pig
Rabbit
ET
(cm2)
200.0
3.0
14.0b
15.0C
30.0
30.0
Source
Guilmette et al. (1989)
Gross et al. (1982)
Gross et al. (1982)
Schreider and Hutchens
(1980)
Kliment (1973)
TB
(cm2)
3,200.0
3.5
20.0
22.5
200.0
300.0
Source
Mercer et al. (1994a)
Mercer et al. (1994a)
Yu and Xu (1987)
Mercer et al. (1994a)
Schreider and Hutchens
(1980)
Kliment (1973)
PU
(m2)
54.0
0.05
0.3
0.34
0.9
5.9
Source
Mercer et al. (1994b)
Geelhaar and Weibel
Mercer et al. (1994b)
Lechner (1978)
Mercer et al. (1994b)
(1971),
Tenney and Remmers (1963)
Gehret al. (1981)
aET = Extrathoracic.
TB = Tracheobronchial.
PU = Pulmonary.
No measurements of hamster ET surface area were found in the literature. This value is estimated based on similarity of the other
regional surface areas to the rat.
cAdditional unpublished measurements of the surface area beyond the ethmoid turbinates are included.
4-26
-------
p. 64
Body weight is the recommended normalizing factor for remote (extrarespiratory)
effects. The default body weight values for the same five animal species are provided in
Table 4-5. The body weight for the human is the weight used by the International
Commission on Radiological Protection (Snyder et al., 1975) for the Reference Man. The
values in Table 4-5 are taken from U.S. Environmental Protection Agency (1988a) and
provide recommended values for body weights when evaluating subchronic or chronic studies
in a variety of strains for each species. Often information on the strain used in a particular
study can be obtained from the principal investigator in the rare event that it is not provided
in the journal articles. If different strains are used than those in Table 4-5 and the body
weight is reported, choose the strain with the most comparable body weight. Documents on
recommended values for use in risk assessment (U.S. Environmental Protection Agency,
1988a) and for use in physiologically based models (U.S. Environmental Protection Agency,
1988b) are useful sources of default values for parameters such as ventilation rates and body
weights for use in these equations when these values are not supplied in individual
investigations. Available allometric equations (Adolph, 1949; Weibel, 1972; U.S.
Environmental Protection Agency, 1988a), relating body size to the parameters of interest
such as ventilatory rates and lung surface areas also may be appropriate. It must be
emphasized that the use of default or derived values must be consistent with the dosimetric
modeling parameters and approaches used in adjusting concentrations to human equivalent
values, such as the parameters used to calculate the RDDR,. and RGDR,..
The default ventilation values for minute volume [VE = tidal volume (VT) x breathing
frequency (f)] are calculated using the allometric scaling equations provided in U.S.
Environmental Protection Agency (1988a). The general form for the equation is:
log (VE)=b0+b1 log(BW) (4-4)
where log refers to the natural logarithm, VE is in L/min and body weight (BW) is in kg.
The species specific parameters (b0 and bj) are listed in Table 4-6. At the present time, the
default body weight for the human is defined to be 70 kg, and the VE is defined to be
(13.8 L/min) 20 m3/day.
4-27
-------
p. 65
TABLE 4-5. BODY WEIGHT (kg) DEFAULT VALUES—RATS
Strain
Fisher 344
Fisher 344
Sprague-Dawley
Sprague-Dawley
Long-Evans
Long-Evans
Osborne-Mendel
Osborne-Mendel
Wistar
Wistar
Strain
B6C3F1
B6C3F1
BAF1
BAF1
Sex
F
M
F
M
F
M
F
M
F
M
BODY WEIGHT
Sex
F
M
F
M
Subchronic
0.124
0.180
0.204
0.267
0.179
0.248
0.201
0.263
0.156
0.217
(kg) DEFAULT VALUES-MICE
Subchronic
0.0246
0.0316
0.0204
0.0223
Chronic
0.229
0.380
0.338
0.523
0.344
0.472
0.389
0.514
0.297
0.462
Chronic
0.0353
0.0373
0.0222
0.0261
BODY WEIGHT (kg) DEFAULT VALUES-HAMSTER
Strain
Syrian
Syrian
Chinese and Djungarian
Chinese and Djungarian
Sex
F
M
F
M
BODY WEIGHT (kg)
Strain
Not specified
Not specified
Sex
F
M
Subchronic
0.095
0.097
0.025
0.03
DEFAULT VALUES-GUINEA PIGS
Subchronic
0.39
0.48
Chronic
0.145
0.134
0.038
0.041
Chronic
0.86
0.89
BODY WEIGHT (kg) DEFAULT VALUES-RABBITS
Strain
New Zealand
New Zealand
Sex
F
M
Subchronic
3.10
2.86
Chronic
3.93
3.76
Source: U.S. Environmental Protection Agency (1988a).
4-28
-------
p. 66
TABLE 4-6. INTERCEPT (b0) AND COEFFICIENT (bj) VALUES USED
IN ALGORITHM (Equation 4-4) TO CALCULATE DEFAULT MINUTE VOLUMES
BASED ON BODY WEIGHT
b0 t>!
Rat
Mouse
Hamster
Guinea pig
Rabbit
-0.578
0.326
-1.054
-1.191
-0.783
0.821
1.050
0.902
0.516
0.831
Source: U.S. Environmental Protection Agency (1988a).
4.3.5 Dosimetric Adjustments for Particle Exposures
Inhalation toxicologists have advanced their ability to measure the deposition of particles
in the various regions of the respiratory tract across species. Initially the data were primarily
total deposition values for polydisperse and sometimes unstable aerosols, but data now exist
for insoluble monodisperse aerosols of different sizes under different breathing conditions
(U.S. Environmental Protection Agency, 1982b). Data are available for many experimental
species of interest on the regional deposition of aerodynamic particle size ranges and on the
necessary physiologic parameters (e.g., ventilation parameters and regional surface areas)
incorporated in dose adjustments (Overton et al., 1987; Miller et al., 1987b; Miller et al.,
1988; Raabe et al., 1988; Patra et al., 1986; Patra, 1986). Deposition data are usually
reported as the deposition fraction for each respiratory tract region of the species of interest.
Deposition fraction is the ratio of the number or mass of particles deposited in the respiratory
tract to the number or mass of particles inhaled. Deposition data also may be expressed as
efficiencies, that is the amount deposited in a particular region normalized for the amount
entering that region.
Knowledge also has been gained in the technology and methods for generating and
characterizing aerosols. State-of-the-art inhalation toxicology studies characterize the
paniculate exposure by the particle diameter (e.g., aerodynamic equivalent diameter [dae],
aerodynamic resistance diameter [dar], mass median aerodynamic diameter [MMAD]), and
the geometric standard deviation (a0). Appendix H provides information on converting
4-29
-------
p. 67
reported particle units to those used in the calculation of the dosimetric adjustment factor and
guidance on default values.
These advances in quantitation of species-specific regional respiratory tract deposition
and characterization of physiologic parameters have been used in the development of an
empirical model that accounts for dosimetry differences using deposition data typical for
aerodynamic particles. This application is an adaptation (Miller et al., 1983b; Graham et al.,
1985) and an extension (Miller et al., 1988; Jarabek et al., 1989, 1990) of previous work.
A series of empirical equations were fit to experimental measurements of regional particle
deposition in various laboratory species and humans as described in Appendix G. These
equations are used to estimate fractional deposition and, in conjunction with normalizing
factors such as body weight or surface area, are used to adjust for dosimetric differences
between species in the calculation of an HEC. The approach is limited at this time to
relatively insoluble and nonhygroscopic particles.
The derivation of the NOAELrHECj for insoluble, approximately spherical particles is
described as
NOAEL*[HEC] (mg/m3) = NOAEL*[ADJ] (mg/m3) X RDDRf, (4-5)
where:
NOAEL*[HECj = the NOAEL or analogous effect level obtained with an alternate
approach as described in Appendix A, dosimetrically adjusted to an
HEC;
NOAEL*rADJi = is defined in Equation 4-2; and
RDDRr = a multiplicative factor used to adjust an observed inhalation
paniculate exposure concentration of an animal (A) to the predicted
inhalation particulate exposure concentration for a human (H) that
would be associated with the same dose delivered to the r region or
target tissue:
(RDD/Normalizing Factor)»
RDDRr = -.
r (RDDr/NormalizingFactor)H
The r regions and potential target tissues identified by this calculation are the three
respiratory tract regions (extrathoracic [ET], tracheobronchial [TB], or pulmonary [PU]).
4-30
-------
p. 68
Definitions of the three regions are provided in Chapter 3 of this document. The RDDRr can
also be calculated for the thoracic (TH) region (TB plus PU regions) or the total (TOT)
respiratory tract (all three respiratory tract regions). Total deposition (deposition summed for
all three regions) is assumed to be available for transport to other organs and is used to
calculate the RDDR for extrarespiratory (ER) effects.
It is frequently desirable to use a normalizing factor when comparing doses across
species. Because insoluble particles deposit and clear along the surface of the respiratory
tract, dose per unit surface area is a commonly used normalizing factor for particulate
deposition in the respiratory tract. In some cases, it might be desirable to normalize by
regional volumes, organ weight, or body weight. It might also be appropriate to examine the
dose ratio with no normalizing factor. The appropriate normalizing factor to use may also be
judged according to the guidance provided in Section 4.3.3 on the use of pharmacokinetic and
pharmacodynamic data, with heed to the cautionary notes provided in the following sections.
Regional deposited dose (RDDr) is estimated as
RDDr = 10"6 X Cj X VE X Fr, (4-6)
where:
RDDr = dose deposited in region r, mg/min,
CA = concentration, mg/m3,
VE = minute volume, mL/min,
Fr = fractional deposition in region r.
The RDDRr may be expressed as a series of four ratios:
RDDR = X x x r (4.7)
r (10"6 X q)H (Normalizing Factor)A (v£)H (Fr)H'
For the purposes of calculating the RDDRr, the exposure concentration for the
laboratory animal (A) and human (H) are assumed to be the same because it is assumed that
the observed effect in the laboratory animal is relevant to human health risk. Therefore, the
RDDRr provides a factor to adjust for the difference in dose delivered to the target tissue
4-31
-------
p. 69
under the same exposure scenario. The first term in Equation 4-7, therefore, equals one and
will not be discussed further.
The second term in Equation 4-7 is the ratio of the normalizing factors for the human
and laboratory animal of interest. For effects in any or all of the three regions of the
respiratory tract, surface area (see Table 4-4) is the recommended normalizing factor.
To evaluate extrarespiratory effects, body weight (see Table 4-5) is the recommended
normalizing factor. The third term of Equation 4-7 is the ratio of minute volumes (see
Equation 4-4).
The final term in the RDDRr equation is the ratio of regional fractional deposition in
laboratory animals and humans. By means of nonlinear regression, empirical equations have
been fit using experimentally measured regional deposition in both laboratory animals and
humans. Details on the estimation procedures are provided in Appendix G. These equations
provide predictions for approximately spherical, nonhygroscopic, insoluble particles in the
aerodynamic size range (particle diameter > 0.5 pm). Deposition fractions should not be
calculated using these equations if the particles deviate enough from spherical that they are
not reasonably described by an aerodynamic diameter (e.g., fibers) or if the particles are
smaller than 0.5 /xm (see Appendix H for a discussion of particle-related issues). Predicted
deposition of hygroscopic particles may be approximated by these equations using the
equilibrium particle size, if known. Other techniques to estimate fractional deposition are
required for particles falling outside the assumptions of this empirical model.
The RDDRr is most easily calculated using the software available as a supplement to
this document. For near monodisperse particles (a < 1.3), deposition fractions may be
calculated as described in Appendix G and the RDDRr calculated by hand. For polydisperse
particles (a > 1.3), however, deposition fractions are calculated by integrating the product
o
of the monodisperse deposition probabilities and the log-normal distribution. This calculation
must be done by computer. The software to perform the RDDRr calculation is written in
C and will run on any DOS-based personal computer. A math coprocessor chip is not
required. Figures 4-5 to 4-8 illustrate the four screens of the program. The first three
figures show how the program display screens will look during data entry, while Figure 4-8
reproduces the RDDRs that would be calculated using these input data.
4-32
-------
p. 70
This program calculates the regional deposited dose ratio
for rodent/human for a particle characterized by a given mass median
aerodynamic diameter (MMAD) and geometric variance (sigma g).
v.2.2
Enter the MMAD in microns (.5 • 30): 2.3
Enter the sigma g(>=1.0): 1.8
Computationally Sigma g < 1.3 will be considered monodisperse
Press Esc key to quit, Enter to
continue, any other key to repeat.
Figure 4-5. Display Screen 1 of the computer program that calculates regional deposited
dose ratios.
Hunan information:
default Body ueight = 76.68 (Kg)
default Minute Vlolune CUE) = 13.86 (Liters)
default ET surface area = 266.66 (cnz)
default IB surface area = 3266.66 (cm1)
default PU surface area = 54.66 (n*)
Press Esc key to quit. Enter to continue, any other key to repeat.
Figure 4-6. Display Screen 2 of the program that calculates regional deposited dose
ratios.
4-33
-------
p. 71
Select one of the following:
1. House
2. hanster
3. rat
4. guinea pig
5. rabbit
Select one (1-5): 3
Body weight from table 4-4 = 188.88 (g)
default Minute Volume CUE) = 137.38 (nl)
default ET surface area = 15.88 (en*)
default IB surface area = 22.58 (erf)
default PU surface area = 8.34 (n2)
neu value: 12.8B(cnz)
Press Esc key to quit, Enter to continue, any other key to repeat.
Figure 4-7. Display Screen 3 of the computer program that calculates regional
deposited dose ratios.
Regional deposited dose ratios
(MAD = 2.30
Signa g = 1.86
Body
SPECIES ueight(g) UE(nI)
rat 180 137.3
human 7OOOO 13800 .0
RATIO 0.003 6.010
RDDR
rat
human
RATIO
RDDR
Enter: saue screen
Extrathorac ic
SA(cn^Z) dep
IZ.eoO 0.473
200.000 0.396
0.060 1.196
0.198
Thoracic
SACn~2> dep
0.34Z 0.147
54.320 0.1Z5
0.006 1.174
0.713
» neu session.
Tracheobronch ia 1
SA(cn^Z) dep
ZZ.SOO 0.058
3200.000 6.095
0.007 0.610
0.863
Total RT
SA(n~Z> dep
0.343 0.6ZO
54.340 0.7Z1
0.006 0.860
1.353
Esc: save screen »
Pu 1 nonary
SA(nAZ) dep
0.340 0.089
54.000 O.Z31
0.006 0.387
0.611
Extraresp i ratory
BU(g) dep
180 0.6ZO
70000 0.7Z1
0.003 0.860
3.3Z6
quit. U. Z.3
Figure 4-8. Display Screen 4 of the computer program that calculates regional
deposited dose ratios.
4-34
-------
p. 72
To begin the program, type RDDR in upper or lower case letters. The program will
then prompt you to enter the MMAD and a for which you want to calculate an RDDRr
(Figure 4-5). Although most studies do report particle sizes as aerodynamic diameter, some
studies do not. Using incorrect units in the program will result in incorrect estimates of
the deposition fraction. Particle size definitions, a discussion on conversion among units,
and guidance on default values when there is inadequate information in a study to determine
the MMAD and a0 are provided in Appendix H.
&
The second screen (Figure 4-6) will print the default values for the minute volume, the
three respiratory tract surface areas, and the body weight for the human. As each one is
listed, the user has the option of changing the default value for the calculations. Although
the software is written so that default values may be changed, it should be noted that body
weight, surface areas, and minute volumes are all inter-related and should be changed so that
all values are consistent with each other. It is also not recommended to make changes
without being able to provide detailed documentation to support alternative values.
The third screen (Figure 4-7) provides a list of 5 animal species from which one must
be selected. The body weight (selected from Table 4-5) must be entered. The program then
calculates and lists the default minute volume and the default surface areas. Similar to the
humans, any of these values may be changed if desired. The same cautions and caveats for
changing human default values apply to the laboratory animals.
In the fourth screen (Figure 4-8), the input parameters are listed; the ratios described in
Equation 4-7 are printed; and the calculated RDDRs are listed for the three respiratory tract
regions, the thoracic region, the total respiratory tract and for extrarespiratory effects. This
screen may be output to an ASCII file and printed using DOS commands. The "PRINT
SCREEN" key will work for a stand-alone PC with its own local printer. Use of the
"PRINT SCREEN" key with networks depends on how files are treated in the buffer.
The program may be run sequentially and calculations made by hand to determine an
RDDRr based on a human activity pattern. First, the minute volumes to be included in the
activity pattern and the fractional time spent at each minute volume must be determined.
Then, the program must be run for each minute volume (keeping all other data the same—
4-35
-------
p. 73
MMAD, o , and surface area for the human, and species, minute volume, and surface area
o
for the animal). Then,
RDDRr = . . . (4-8)
[ACT1 tmXVF XFr +tmxVF xFr + ...+trn1xVF XFr
W ^HIl] fH[l] 1ZJ fiH[2] fH[2] lnJ ^HM rH[n]
where t^ is the fractional time spent breathing minute volume [i],
+t[2] +-+t[n]
All of the needed values can be read from Screen 4. The calculated value, a, should have the
same input values (i.e., surface areas, all animal input information) on each Screen 4
generated for the activity pattern, but the human values for deposition (Fr . ) and minute
H[i]
volume (VE . ) will be different.
H[i]
Although the default normalizing factor used in the program for the respiratory tract
RDDRs is surface area and the default normalizing factor for the extrarespiratory RDDR is
body weight, there are situations in which an alternative normalizing factor might be
appropriate. In this case, the deposition ratio and the VE ratio from the 4th screen of the
computer program (Figure 4-8) may be multiplied by hand calculations of the normalizing
factor in humans divided by the normalizing factor in animals to determine the RDDRr.
Alternatively, when the default surface area values are listed in Screens 2 and 2
(Figures 4-6 and 4-7), they may be changed to the values of the new normalizing factor.
A caveat when "tricking" the program this way is to pay attention to the units of the
n
normalizing factor. In the program, ET and TB surface areas are entered in cm while the
PU surface area is entered in m2. The program converts units internally when calculating the
TH and total RDDRs. Unless the units of the proposed normalizing factor bear the same
relationship to one another as the surface areas, the program calculated TH and total RDDRs
with the alternative normalizing factor will be incorrect. At the present time, because
VE calculations depend on entering correct body weight data, if the program is "tricked" by
4-36
-------
p. 74
entering information other than body weight to estimate an extrarespiratory RDDR, then the
correct VE must be used to replace the program calculated "default" VE.
The next two sections provide a summary of the default values used for respiratory tract
effects and extrarespiratory tract effects. As discussed above, details on estimation of the
deposition fractions are described in Appendix G.
4.3.5.1 Respiratory Effects
The general dosimetric approach for insoluble particles outlined above provides the basis
for estimating HECs. When the toxic effect of interest is in the respiratory tract, the
equivalent dose across species is assumed to be the particle mass (mg) per minute per unit
O O
surface area (cm or m ) of the respiratory tract region of concern.
When the toxic effect of interest is in the respiratory tract, the normalizing factor
described in Equation 4-7 should be replaced specifically by the surface area (SA) of the
respiratory tract region of interest.
(SAr)H (V^A (Fr)A
RDDR, = - — X _ — X ^A X r A (4-11)
(SAr)A (VE)R (Fr)H
The default surface area values are provided in Table 4-4.
It is preferable, when possible, to estimate the RDDRr for one of the three defined
respiratory tract regions (ET, TB, or PU). Sometimes the nature of the effect or the detail of
reporting precludes distinguishing between a TB and a PU effect so that an RDDRr for the
TH region would be preferred, or it might be possible only to identify the region of interest
as the entire respiratory tract. Either some aggregation must be used in calculating the
RDDR,., or the RDDRj for the region that results in the most conservative HEC could be
selected. There are several techniques to aggregate the deposition information for calculation
of TH or total respiratory tract RDDRs. The resulting RDDR can vary substantially, and in
some cases the determination of which species is more sensitive (human or laboratory animal)
may change. This is due to differences in fractional deposition (reflecting the complexities
of the mechanisms governing deposition) in the different regions (see Chapter 3) and to the
differences in regional surface areas, which may span several orders of magnitude. The
4-37
-------
p. 75
formula used to calculate the total respiratory tract RDDR (RDDRTOT) is given below.
Calculation of the thoracic RDDR (RDDRjjj) differs only in the exclusion of terms related to
the ET region.
First, for each species, regional fractional deposition (Fr) per unit surface area (SAr) is
calculated and weighted by the percent of the respiratory tract (TH region) accounted for by
that region.
FET SAET
rr'TD
Then, simplifying this expression and summing over the three (or two in the case of the
calculation for the TH region) regions gives
FTOT FET + FTB + Fpu (4-13)
O "* O Ap-p + O A«-pT3 + O ApT j
yielding
RDDRTOT - x X X
"
.
(10"6 x q)H (SATOT)A (VE)H (FTOT)H
4.3.5.2 Remote (Extrarespiratory) Effects
The respiratory tract might not be the target organ for an inhaled compound. The dose
actually delivered to other regions of the body will be affected by metabolism, clearance, and
distribution patterns. Particles depositing in the respiratory tract will clear rapidly (ET can be
within seconds of inhalation) or slowly (PU clearance may take weeks or months) to the GI
tract or be absorbed into the interstitium, lymphatics, or into the blood from the respiratory
tract. Once deposited, however, very few particles will clear by exhalation (sneezing or
coughing). Therefore, it is not unreasonable to estimate extrarespiratory deposition by total
4-38
-------
p. 76
deposition in the respiratory tract when information on dose delivered to nonrespiratory tract
organs is unavailable.
The current default normalizing factor for extrarespiratory effects is body weight.
In the case of extrarespiratory effects of particles, the equivalent dose across species is
assumed to be the mass of particles (mg) deposited per unit body weight (kg). Until
clearance and distribution parameters can be incorporated, it is assumed that 100% of the
deposited dose to the entire respiratory system is available for uptake to the systemic
circulation. Although this assumption will most likely result in an overestimate of the dose
delivered to the extrarespiratory target tissue, it is not possible to predict a priori the impact
on the dose ratio and resulting HEC (e.g., if the overestimate is of similar magnitude in both
the laboratory species and human, the HEC will be relatively unaffected). Use of deposited
dose is more accurate than using exposure concentration, however. Therefore, Equation 4-7
may be rewritten as:
RDDRER . ' x x x (4-15,
(10-6 x CJ)H BWA (VE)H
-------
p. 77
Hygroscopicity, Solubility, and Nonspherical Particles
The empirical equations used to estimate the predicted regional deposition fractions are
derived from exposures using monodisperse, approximately spherical, nonsoluble, and
nonhygroscopic particles. The cases outside the defined conditions for the equations include
polydisperse particle size distributions, nonspherical particles, and soluble and/or hygroscopic
particles. Also, Gerde et al. (1991) have shown that highly lipophilic chemicals and
chemicals either absorbed or precipitated onto particles behave fundamentally differently and
may require other modeling approaches.
As described in Appendix G, deposition fractions may be estimated for polydisperse
spherical particles by integrating the monodisperse deposition fraction over the size
distribution of polydisperse particle. The calculations made for this document assume a
lognormal particle size distribution (Raabe, 1971). When particle size distribution for an
exposure is reported as MMAD and a , it may be assumed that the particle size distribution
is described by the lognormal distribution (because MMAD and a are the first two moments
for a lognormal distribution). If exact size distribution information is given or the particles
are described as coming from a different, well-parameterized distribution, then an exact
calculation must be performed.
Nonspherical particles may be described in terms of their equivalent aerodynamic
diameter and, if this information is provided, deposition fractions may be calculated as
described in this chapter and Appendix G. Deposition fractions may not be calculated using
these equations if an aerodynamic diameter is not provided.
Many particles are hygroscopic and/or soluble. Hygroscopic particles may change size,
shape, and density as they traverse the warm, humid airways of the respiratory tract. Soluble
particles might or might not undergo hygroscopic changes as they travel along the airways.
Solubility will change the physicochemical interactions of the particle with the surface upon
which it deposits. Hygroscopicity is a phenomenon related to deposition whereas solubility is
related to clearance. This discussion, therefore, will focus on hygroscopicity and its potential
effects on predicted fractional deposition.
The ROD of a hygroscopic aerosol will often be different from that of nonhygroscopic
particles, although both had similar size distributions upon inhalation (Martonen et al., 1985).
The factors influencing changes in inhaled hygroscopic particle characteristics are being
4-40
-------
p. 78
studied experimentally and through development and analysis with complex theoretical models
(Martonen and Patel, 1981; Martonen, 1982; Perron and Hornik, 1984; Martonen et al.,
1985; Eisner et al., 1990), but application in risk assessment awaits definition of the primary
factors influencing hygroscopic growth on species- and agent-specific bases. The factors
include initial particle geometry and density, material hygroscopic growth characteristics,
respiratory parameters, and temperature and relative humidity profiles. Observations on the
data from modeling efforts to date indicate that hygroscopic particles in the diffusion-
dominated regime have reduced TH deposition relative to nonhygroscopic particles of
identical preinspired size, whereas those hygroscopic particles affected by inertial and
gravitational forces have an increase in TH deposition relative to nonhygroscopic particles
(Martonen et al., 1985). These observations may be explained by changes in the particle size
after inspiration. Accordingly, the calculated deposition efficiency for nonhygroscopic
particles would underestimate the TH deposited dose for the larger (affected by inertial and
gravitational forces) hygroscopic particles, and overestimate the deposited dose for the
smaller diffusion-dependent hygroscopic particles. The TH deposited dose of inhaled
nonhygroscopic particles, however, is always less than the initial total dose (exposure dose).
Also, the relative changes in deposition will be in a similar direction in experimental animal
species and humans. Dosimetric adjustment by the default insoluble (nonhygroscopic)
empirical deposition equations is recommended as a conservative default for the hydroscopic
particles, pending modification by the elucidation of the hygroscopic models.
Ventilation
It is recognized that this approach is based on deposition efficiency data obtained or
derived under a particular set of ventilatory parameters (i.e., the experimental parameters for
the laboratory animals and human subjects), coupled with default ventilation parameters (VE).
The assumption in this application is that it is valid to linearly extrapolate from these
experimental values to the default sets of ventilation parameters. The validity of this
assumption is being investigated. The effect of activity pattern on ventilation and the
allometric relationships between lung weight, lung surface area, and body weight have been
investigated (Adolph, 1949; Weibel, 1972; U.S. Environmental Protection Agency, 1988a;
4-41
-------
p. 79
1993b; Federal Register, 1992b). A discussion of the impact that breathing pattern has on
the human deposition estimates can be found elsewhere (Miller et al., 1988).
Differences Between Experimental and Ambient Exposures
The human ambient exposure scenario, when known, may be characterized by a
different MMAD and a than that used to derive the health risk assessment. Comparisons
c?
between ratios calculated with a MMAD and a the same as the animal exposure and
o
calculated with the human estimate using the anticipated ambient MMAD and a may provide
some insight on the uncertainty of this extrapolation.
Clearance and Retention
In addition to inspired air concentration, VE, surface area, and deposition efficiency,
the effective dose of inhaled particulate matter will vary with bioavailability. The fraction of
particulate matter dissolved and assumed to be bioavailable can be expected to increase with
greater particle solubility, as well as with longer residence time in the lungs. Until clearance
and distribution parameters can be systematically incorporated, 100% of the deposited dose to
the entire respiratory tract is assumed to be available for uptake to the systemic circulation.
As discussed, this assumption will most likely result in an overestimate of the dose delivered
to the extrarespiratory target tissue, although it is not possible to predict a priori the impact
on the dose ratio and resultant HEC. Use of deposited dose is more accurate than using
exposure concentration, however. Models have recently been used to simulate clearance and
estimate retention in various species (Snipes, 1989a,b; Yu and Yoon, 1990). The EPA has
recognized the importance of incorporating clearance components to its dosimetric
adjustments in order to calculate regional retained dose ratios (RRDRs) for estimates of long-
term lung burdens, but such models for classes of particles and different species used in
testing are not fully developed. In those cases where clearance and distribution have been
experimentally determined and a validated model exists, the more comprehensive model
should be used. For example, the model of Yu and Yoon (1990) was used to calculate the
HEC for diesel engine emissions (IRIS, 1992).
4-42
-------
p. 80
Population Variability
The calculation of an RDDRj. currently uses point estimates for all the terms in
Equation 4-7 and its variants; that is, a default VE for each species, a default regional surface
area, and an estimate of fractional deposition. These single values are assumed to be
representative of the average value of that term for a member of the laboratory animal species
or human population. In fact, as discussed in Chapter 3, there are many sources of
intraspecies variability that contribute to the range of responses observed to a given external
exposure to an inhaled toxicant. Host factors affect both the delivered dose of the toxicant to
the target tissue and the sensitivity of that tissue to interaction with the toxicant. The
procedures described in the preceding sections of this chapter on particle dosimetry provide
some limited capabilities to examine the effects of population variability on the RDDRr by
simply changing the default VE and surface areas in an iterative fashion. As indicated in
these sections and in Appendix G, however, because of the correlations between VE, surface
area, and body weight, such changes should be made with extreme caution. Although the
point estimates of the parameters used to predict deposition efficiencies (details in
Appendix G) are used to calculate fractional regional deposition, the empirical model also
provides estimates of variability that can be used to generate confidence intervals reflective of
population variability. Using iterative computational procedures, it is possible to generate
envelopes of regional fractional deposition that can be used with distributions of VE, surface
areas, and body weights to provide ranges of RDDR,.s. Actual implementation of this
procedure is not straightforward due to the complex nature of the correlation structures.
In future versions of the dosimetric model used to calculate RDDRj., it should be possible to
estimate a distribution for the RDDRr reflective of population variability in both laboratory
animals and humans.
Susceptible Subpopulations
The data used to estimate regional fractional deposition are based on experimental
measurements made in healthy laboratory animals and humans breathing under normal or
approximately normal conditions. It is recognized that deposition patterns might vary in
potentially susceptible subpopulations such as children, the elderly, or people with respiratory
diseases (see Chapter 2). Limited data are available at present for fitting deposition
4-43
-------
p. 81
efficiency equations for any of these subpopulations. If it is assumed that the same efficiency
relationships may be used, then the model may be used to examine predicted RDDRs
(in healthy children, for example) by scaling surface area and ventilation for size,. This
approach is consistent with deterministic models of deposition in which airway geometry and
ventilation are scaled to children's dimensions, but the mechanisms of deposition are
unchanged. Although the approaches are consistent, the predicted deposition patterns might
vary with measured data.
4.3.6 Dosimetric Adjustments for Gas Exposures
The approach described in Section 4.3.5 for the insoluble particles illustrates the
feasibility of interspecies dosimetry calculations to extrapolate the lexicological results of
inhaled toxicants to human exposure conditions for dose-response evaluation. Dosimetry data
facilitate evaluation of concentration-response data with respect to dose-response relationships.
As described in Section 3.2, predictive physiologically-based modeling for relatively insoluble
and reactive gases has been demonstrated (Overton and Miller, 1988). Predictive
physiologically based modeling has also been demonstrated for gases and vapors of organic
solvents that may be metabolically activated (Fiserova-Bergerova, 1983; Andersen et al.,
1987a; Overton, 1989), and for reactive and soluble gases (Aharonson et al., 1974; Morgan
and Frank, 1977; Hanna et al., 1989; Casanova et al., 1991; Morris and Blanchard, 1992).
As discussed in Section 3.2.2, the chemical-specific or class-specific nature of these models
has been dictated by the physicochemical characteristics of the subject gases and no single
model structure is applicable to the broad range of gases that the RfC methodology must
address. A gas categorization scheme was thus developed as a way to create separation
between types of gases so that model structures for each type could be developed. The
scheme developed in Section 3.2.2 should be used to categorize the type of gas for dosimetric
adjustment. The derivation of the model structure and its reduction to a form with a minimal
number of parameters as the basis of the default dosimetry adjustments for gases in
Category 1 and 2 are presented in Appendix I. The model structure and basis for the default
adjustment for gases in Category 3 are presented in Appendix J. The reader is referred to
these sections for proper understanding of the framework of default dosimetric equations
presented herein.
4-44
-------
p. 82
Consideration also should be given to the discussion by the National Research Council
(1986) and Dahl (1990) on interspecies extrapolation based on mechanism of action. Three
classes of mechanism were distinguished based on whether the parent compound, stable
metabolite, or reactive metabolite produces the toxic effect; measures of dose for each of
these classes were suggested. These factors are often species-specific and dose-dependent, as
well as being chemical-specific and, therefore, require a substantial data base (beyond that
which exists in most circumstances) in order to model comparative species dosimetry of gases
based on mechanism of action. O'Flaherty (1989) presented a framework within which to
consider measures of delivered dose and discusses procedures for interspecies conversion of
kinetically equivalent doses. Identification of the limiting anatomic and physiologic
parameters, physicochemical characteristics, and exposure concentration and duration
conditions will facilitate the application of these factors to improve the interspecies default
dose adjustments. This understanding can also be used to gauge the appropriation of the
default adjustments on a case-by-case basis.
Basically, the RGDRr is used as the DAFr in Equation 4-3 to dosimetrically adjust the
experimental NOAEL to an HEC as
NOAEL*[HEC] (mg/m3) = NOAEL*[ADJ] (mg/m3) X RGDRr, (4-16)
where:
NOAEL*[HECj = the NOAEL or analogous effect level obtained with an alternative
approach as described in Appendix A, dosimetrically adjusted to an
HEC;
NOAEL*rADJi = is defined in Equation 4-2; and
RGDRr = (RGD)A/(RGD)H, the ratio of regional gas dose in laboratory animal
species to that of humans for region (r) of interest for the toxic
effect.
The default equations to derive the RGDR,. for the different gas categories according to
toxicity in the respiratory tract versus remote sites follow in Section 4.3.6.1 and 4.3.6.2,
respectively. Because the boundaries between the categories are not definitive (see discussion
in Section 3.2.2 and Appendix I), but instead were made to allow derivation of default model
4-45
-------
p. 83
structures, identification of the target effect(s) is used to further define the gas category.
Thus, remote (extrarespiratory) effects of Category 1 gases and respiratory effects of
Category 3 gases are treated according to the default dosimetric adjustments for each of these
respective effects of Category 2 gases (Section 4.3.5 and 4.3.6). The default dosimetric
adjustments to derive HEC values for respiratory effects of Category 1 gases are provided in
Section 4.3.5. The default dosimetric adjustment to derive HEC values for extrarespiratory
effects of Category 3 gases is provided in Section 4.3.6. Note that the gas categorization
scheme dose not apply to inert gases that exert their effects by reversible "physical"
interactions of gas molecules with biomolecules (e.g., "displacement" of oxygen by carbon
dioxide or narcosis by some parent compounds). Consideration of the inert gases is discussed
in Section 2.1.2.3.
4.3.6.1 Respiratory Effects
The two categories of gases with the greatest potential for respiratory tract effects are
gases in Category 1 and 2. Category 1 gases are defined as gases that are highly water-
soluble and/or rapidly irreversibly reactive in the respiratory tract. Reactivity is defined to
include both the propensity for dissociation as well as the ability to serve as substrate for
metabolism in the respiratory tract. Gases in Category 2 are defined as gases that are
moderately water-soluble that may be rapidly reversibly reactive or moderately to slowly
irreversibly reactive in respiratory tract tissue. Examples of gases in Category 1 are
hydrogen fluoride, chlorine, formaldehyde, and the organic acids and esters. Examples of
gases in Category 2 are ozone, sulfur dioxide, xylene, propanol, and isoamyl alcohol.
Respiratory Effects—Category 1 Gases
Category 1 gases are distinguished by the property that the gas does not significantly
accumulate in the blood which would reduce the concentration driving force into the
respiratory tract tissue and hence reduce the absorption rate. This characteristic allowed the
default approach to be developed based on the integration of attributes of two empirical
models as discussed in Appendix I. The approach takes into account the loss of chemical in
the airstream to the upper respiratory tract as it progresses to the lower respiratory tract and
4-46
-------
p. 84
separate equations are provided to calculate dose in each region. The rationale and full
derivation of the equations is provided in Appendix I.
Extrathoracic Effects. For Category 1 gases that have an effect in the upper respiratory
tract, the following equation is used to calculate the ET regional gas dose ratio (RGDRET).
-K SA
SET
VE
RGDRET - (DOS6ET)A - SAETA , (4-17)
ET ™_ x -\CTSACT
)H
_ _
(DoseET)H -KSA
SAET
where:
VE = minute volume (mL/min = cm3/min),
s\
SET = surface area of the extrathoracic region (cm ), and
K0 = overall mass transport coefficient in the extrathoracic region (cm/min).
SET
A, H = subscripts denoting laboratory animal and human, respectively.
When the overall mass transport coefficient in the ET region (K0 ) is not known or can not
°ET
be reasonably approximated with experimental data for either species, the following equation
is used to calculate the default RGDRET (see Section 1.2.4.1):
(DoseET)H V
)H
o
ET
Tracheobronchial Effects. For Category 1 gases that affect the lower respiratory tract,
the scrubbing in the upper airways of the chemical is taken into account, and the
concentration of the air exiting the ET region is used in the derivation of dose to the
TB region.
4-47
-------
p. 85
The following equation is used to calculate the TB regional gas dose ratio (RGDRTB):
RGDRTB =
(DoseTB)A _
(DoseTB)H
VE
SATB
'VE '
SATB
A (^PET) ^
(fpET) H
H
(1 -(
(1 -,
-K SA-.
KTB ^B
* VE )A
-K*rB SATB
- VE )H
, (4-19)
where:
KSTB
frET
s\
= surface area of the tracheobronchial region (cnr),
= overall mass transport coefficient in the tracheobronchial region (cm/min),
and
= the fraction of inhaled chemical concentration penetrating the ET region and
thereby available for uptake in the TB region, calculated as
K SA™
SET ET
f?ET = e
(4-20)
If the penetration fraction is unknown due to the lack of data on K , it is reasonable to
&TB
assume that K is large, which is consistent with the definition of Category 1 gases, such that
o
the exponential term of Equation 4-19 reduces to zero. The same result may be achieved by
determining the conditions in which the third ratio of the right hand side of Equation 4-19
reduces to 1. These conditions will be a function of the default values for respiratory tract
surface area and minute volume as well as the absolute value of the overall mass transport
coefficient. Using the definition of fpET results in the following dose ratio:
4-48
-------
p. 86
RGDRTB =
(RGDTB)A _
(RGDTB)H
VE
'VE"
SATB
-K CT
A (e " VE)A
-nr CT
SET y
H (C " }H
(4-21)
When the overall mass transport coefficient in the extrathoracic region (K ) is not known
or can not be reasonably approximated with experimental data for either species, K is
*ET
further assumed to be one, and Equation 4-21 reduces further such that only minute volume
and surface areas are needed to evaluate the dose ratio:
RGDRTB =
(RGDTB)A _
(RGDTB)H
VE
SATB
' V
SATB
A
H
SACT
e \
SACT
e ^
A
H
(4-22)
If K is available for each species, Equation 4-21 would be the preferred default equation.
Pulmonary Effects. The gas concentration that reaches the PU region was affected by
the amount of uptake in both the ET and TB regions so that the derivation for the PU gas
dose ratio (RGDRpu) incorporates the penetration fraction both for the ET and TB regions,
respectively. The following equation is used to calculate RGDRpu:
KgPUSAPU
RGDR =
r U
(DosePU)
H
Qalv
SA
PU
^H
:, (4-23)
where:
Qalv = alveolar ventilation rate (mL/min = cm /min),
4-49
-------
p. 87
SApu
KgPU
fPTB =
surface area of the pulmonary region (cm2),
overall mass transport coefficient in the pulmonary region (cm/min),
the fraction of inhaled chemical concentration penetrating the extrathoracic
region and thereby available for uptake in the tracheobronchial region,
calculated as in Equation 4-20, and
the fraction of inhaled chemical concentration penetrating the
tracheobronchial region and thereby available for uptake in the pulmonary
region, calculated as
ex.
era
TB
ex
= e
ET
(4-24)
where:
CXET =
CXTB =
the concentration exiting the extrathoracic region, and
the concentration exiting the tracheobronchial region.
At large K values, Equation 4-23 reduces to
RGDRpu =
(RGDpu)A
(RGDpu)H
Qalv
SApu
Qalv "
SApu
A (fpTB)A MA
(fPTBJH MH
H
(4-25)
If the penetration fractions to each of the preceding regions are unknown due to lack of
data on K and K , the approach to deriving a default equation for the PU region is
SET
described below.
Using the definition of fpET and fpTB results in the following PU region gas dose ratio:
4-50
-------
p. 88
RGDRpu =
(e
SArB
-K.
(e
)A
-K,
SAF
i
SAC
E \ /„ E \
)H (e )H
(4-26)
which can be rearranged to
RGDpu)A
(RGDpu)H
Qalv
SApu
Qalv '
SApu
A
H
-SA,
e E
'SAj
e VE
B
F
(KgTB)A
A
(KITB)H
[
-SAET
e E
-SACT
e E
(\ET>*
A
(KgCT)H
H
(4-27)
If (KgET) A and (KgTB) A are assumed to be equal to (KgET)H and (KgTB)H, respectively,
then Equation 4-27 can be further simplified to
RGDRpu =
(RGDPU)A
(RGDPU)H
[ Qalv 1
SAPU
[ Qalv 1
SApu
A
H
f _ SA-1
e E
' _ SA,.;
e E
A
H
«™
f _ SAKr]
e E
' _ SACT
e E
A
H
K
gET
(4-28)
If it is further assumed that the value of K is equal to 1 for each region, the resulting default
o
equation (Equation 4-28) reduces to an equation requiring only surface area and VE
parameters. It should be noted that as comparative transport studies become available,
Equation 4-27 would be preferable because it includes the differences in mass transport in
each region for each species.
4-51
-------
p. 89
Respiratory Effects—Category 2 Gases
Category 2 or "transitional" gases have the potential for significant accumulation in the
blood and thus have the potential for both respiratory and remote (extrarespiratory) toxicity.
The accumulation in the blood will reduce the concentration driving force during inspiration
and thereby reduce the absorption rate or dose upon inhalation. They also have the potential
for significant desorption during exhalation. The model structure used as the basis for the
default dosimetric adjustment for Category 1 gases was insufficient for addressing this
property and a hybrid structure between that for Category 1 and Category 3 gases was
constructed. The rationale and full derivation of the equations is provided in Appendix I.
The default dosimetric adjustments for respiratory tract effects of Category 2 gases is
presented below and those for dosimetric adjustment of remote toxicity are provided in
Section 4.3.6.2.
Extrathoracic Effects. For Category 2 gases, the ET regional gas dose ratio (RGDRET)
is given by
RGDRET - - _ _ (4.29)
-------
p. 90
H
H
(,30)
Assuming the same inspired concentration, simplifies the RGDRET to
.
RGDRET =
H
(4-3D
If the overall mass transport coefficients (K ) are assumed equal as in the case of Category
1 gases, the RGDRET is reduced to the ratio of the blood term (1 - Cb/&/Cj).
Two cases were developed for the derivation of the blood term (see Appendix I). The
first case assumes systemic elimination is much greater than respiratory tract metabolism such
that
RGDRFT =
(RGDET) K (0.25QTHb/g)
-
(RGDET)
H
(0.25QTHb/g)H
(4-32)
and the second case where the respiratory tract metabolism is of equal significance to
systemic elimination such that
RGDRET =
(RGDET) K (0.5QTHb/
(RGDET)H
(4-33)
(0.5QTHb/g)H
4-53
-------
p. 91
where EMAX, the maximum extraction efficiency, is equal to 0.25 Qj- and Hb/ is the
blood:gas (air) partition coefficient of the chemical. Because the constants are equal in the
numerator and denominator, Equations 4-32 and 4-33 reduce to the same equation. Thus, the
default regional gas dose ratio for ET effects of Category 2 gases is:
(RGDET) K (QTHb/g)
RGDRFT = = - — . (4-34)
(RGDET)H
Equation 4-34 can be further reduced by the assumption that the overall mass transport
coefficients (K ) are equal when these values are not available. The value of 1.0 is used
for the ratio of (Hb/ )A / (Hb/ )H if (Hb/ )A > (Hb/ )H or if these partition coefficient values
are unknown. Gargas et al. (1989) and Jepson et al. (1994) provide discussion of techniques
to derive partition coefficients and report values for volatile and nonvolatile chemicals,
respectively.
Tracheobronchial Effects. The TB regional gas dose ratio (RGDRTB) for Category 2
gases is given by
Cb/a
/"I _ ^
( " "
^ /\ / P\ I . / 1 17 \
wJ/k'T'TJ /p C. » \_^ ft — f* n. »
1 r> A \e /A ! A v * c H
RGDRTB A A A '
" (Cj _) ^-^~ d - -^-) VB-^-
ISATB H (e E )H ci H (1 - e E )H
(4-35)
As in the ET region, K for Category 2 gases is by definition less than 1 and a power
series expansion of the exponential term for the TB region similarly reduces the last term on
the right hand side to the animal-to-human ratio of K (SATB/VE). The exponential term
for the ET term in Equation 4-35 is reduced by assuming K is the same for each species
as was assumed for Category 1 gases. At values of K less than or equal 0.5, the
4-54
-------
p. 92
ET exponential term approaches one. Thus, assuming the same inspired concentrations,
Equation 4-35 becomes
RGDRTB = = - . (4-36)
(RGDTB)H K^ _ C^
H
As above for the ET region, the case in which systemic elimination predominates is given by:
(RGDTB) K (0.25QTHb/a)
RGDRTB = ^ = —^ A , (4-37)
™ (RGDTB)H KgTBR (0.25QTHb/a)H
and the case in which respiratory tract metabolism and systemic elimination are of equal
significance is given by:
(RGDTB) K (0.5 QTHb/a)
RGDR™ • - • •
where EMAX is equal to 0.25 Qp. Because the constants are equal in the numerator and
denominator, Equations 4-37 and 4-38 reduce to the same equation. Thus, the default
regional gas dose ratio for TB effects of Category 2 gases is
(RGDTB) K (QTHb/a)A
RGDRTR = - = - . (4-39)
TB (RGDTB)H KgTBH (QTHb/a)H
Equation 4-39 can be further reduced by the assumption that the overall mass transport
coefficients (K ) are equal when these values are not available. The value of 1.0 is used
BXB
for the ratio of (Hb/ )A / (Hb/ )H if (Hb/ )A > (Hb/ )H or if these partition coefficient values
are unknown. Gargas et al. (1989) and Jepson et al. (1994) provide discussion of techniques
4-55
-------
p. 93
to derive partition coefficients and report values for volatile and nonvolatile chemicals,
respectively.
Pulmonary Effects. The PU regional gas dose ratio (RGDRpu) for Category 2 gases is
given by
(RGDpu) ^SApu' (e "" VE } (e '" VE ) (1
RGDRpu A A A A
H (e E )H (e E )H i H
(4-40)
The default ratio is obtained by assuming the mass transport coefficients for the ET and the
TB region are the same in each species. At values of K < 0.5, as per the definition for
BXB
Category 2 gases, the exponential term for both the ET and TB term in Equation 4-40
reduces to 1.0. Thus, assuming the same inspired concentrations, Equation 4-40 becomes
(RGDpu) SApl q
= - - - - . (4-41)
SAPU » H
The RGDRpu must be evaluated for both cases described above for the ET and TB regions.
In the case where systemic elimination determines the blood term, the ratio is given by
-
(RGDpu) SApl (0.25 QTHb/ )
RGDRPIT= -= 1^ . (4-42)
(RGDpu) Qa] (0.25QTHb/g)
SApu
H
4-56
-------
p. 94
In the case where respiratory tract metabolism and systemic elimination are equally
important, the ratio is given by
--
(RGDpu) VSAPUA (0.5QTHb/)
RGDRPII = - - = - - - _ . (4-43)
(RGDpu)H ^ Qaly ^ (0.5QTHb/g)H
SAPU
H
Because the constants are equal in the numerator and denominator, Equations 4-42 and 4-43
the same equation. Thus, the default regional gas dose ratio for PU effects of Category 2
gases is:
(RGDpu) SAPU A
= _ - _ — . (4-44)
H
The value of 1.0 is used for the ratio of (Hb/g)A / (Hb/g)H if (Hb/g)A > (Hb/g)H or if these
partition coefficient values are unknown. Gargas et al. (1989) and Jepson et al. (1994)
provide discussion of techniques to derive partition coefficients and report values for volatile
and nonvolatile chemicals, respectively.
4.3.6.2 Remote (Extrarespiratory) Effects
As discussed above in Section 4.3.6.2, Category 2 gases have physicochemical
characteristics that result in the potential for significant accumulation of the gas in the blood.
Thus, these gases also have the potential to cause remote (extrarespiratory) toxicity at target
tissues other than the respiratory tract. Gases or vapors in Category 3 are relatively water
insoluble and unreactive in the ET and TB regions. Thus, the relatively limited dose to these
respiratory tract regions does not appear to result in any significant toxicity, although some
respiratory tract toxicity may be related to recirculation. The uptake of these gases is
predominantly in the pulmonary region and is perfusion limited. The site of toxicity is
4-57
-------
p. 95
generally remote to the principal site of absorption in the PU region. An example of a
Category 3 gases is styrene.
Remote (Extrarespiratory) Effects—Category 2 Gases
In the event that remote toxicity is associated with a gas in Category 2, the dose to the
respiratory tract, and therefore to the blood, is necessary to establish the dose ratio.
However, in this case, the surface area of the respiratory tract is irrelevant, only the
absorption rate in mass/time (RGDRT) is important such that the dose ratio becomes
(RGDRT)A _
-------
p. 96
Because the constants are equal in the numerator and denominator, Equations 4-46 and 4-47
reduce to the same equation. Thus, the default regional gas dose ratio for remote
(extrarespiratory) effects of Category 2 gases is:
(RGDRT) (VE) (QTHb/0)
RGDR = _ _* = _* b/*A . (4-48)
H (QTHb/g)H
The value of 1.0 is used for the ratio of (Hb/g)A / (Hb/g)H if (Hb/g)A > (Hb/g)H or if these
partition coefficient values are unknown. Gargas et al. (1989) and Jepson et al. (1994)
provide discussion of techniques to derive partition coefficients and report values for volatile
and nonvolatile chemicals, respectively.
Remote (Extrarespiratory) Effects-Category 3 Gases
For gases in Category 3 that exhibit their toxic effects outside of the respiratory tract,
an approach for the scenario when the concentrations of the gas in the animal is periodic
(or could be expected to be) with respect to time is recommended. Derivation of the
procedure and Equation 4-48 for estimating NOAELrHECjS for extrarespiratory effects of
these gases is based on a PBPK model described in Appendix J. The procedure will give
equivalent or more conservative values for the NOAEL,HECjS than those obtained by using
the PBPK model, and can be used with compounds for which modeling would be applicable,
but for which some or all values of the important parameters (Hb/g, VMAX, KM, etc.) are
not available. The approach assumes that physiologic and kinetic processes can be described
by a PBPK model, allometric scaling of physiologic and kinetic parameters may be used, and
that all concentrations of the inhaled compound in the experimental animal are periodic with
respect to time. Based on the PBPK ventilation-perfusion model concept (e.g., Ramsey and
Andersen, 1984), algebraic equations that relate organ and tissue compartment concentrations
to exposure concentrations under equilibrium conditions were derived for humans; for
laboratory animals, equations were derived that relate time average concentrations. Because
toxic effects observed in chronic bioassays are the basis for the determination of NOAELs
from which RfC values for human exposures are derived, the procedure assumes that chronic
laboratory animal exposure scenarios are equivalent to human lifetime exposures. The
4-59
-------
p. 97
procedure also assumes that the toxic effects observed are related to the arterial blood
concentration (concentration leaving lung compartment in the model) of the inhaled
compound and that NOAELrHECjS should be such that the human time-integrated arterial
blood concentration is less than or equal to that of the exposed laboratory animal. This latter
assumption is equivalent to assuming that the laboratory animal time-averaged arterial blood
concentration is equal to the human equilibrium arterial blood concentration. Note that the
time average concentrations are the area under the curve over a period divided by the length
(time) of a period (e.g., average concentration over 1 week). A mathematical derivation was
used to obtain the proposed method of simple alebraic equations to compute NOAELrHEC^s.
A more detailed description of the development of the procedure is given in Appendix J.
Another assumption is that the concentrations of the inhaled compound within the
animal achieved periodicity with respect to time (i.e., periodic steady state—the concentration
versus time profile is the same for every week). An illustration of periodicity is provided in
Figure 4-9. Periodicity of the arterial concentration of the agent was not achieved until the
sixth week for the plotted theoretical exposure simulation. Practically, the conditions of
periodicity should be met during "most" of the exposure duration. For example, if this
condition is met for 90% of the time (e.g., periodic during the last 90 weeks of a 100 week
experiment), then estimates of average concentrations will be in error by less than 10%.
The following equation is used to calculate an HEC for extrarespiratory effects of gases
in Category 3:
NOAEL*[HEC] (mg/m3) = NOAEL*[ADJ] (mg/m3) x -^A (4-48)
( b/g^H
where:
NOAEL*rHECn - the NOAEL or analogous effect level obtained with an alternative
approach as described in Appendix A, dosimetrically adjusted to an
HEC;
NOAEL*rADJn = is defined in Equation 4-2; and
4-60
-------
p. 98
O)
Blood:Alr Partition Coefficient = 1,000
Fat:Blood Partition Coefficient = 100
Time(hx1(T)
Figure 4-9. Time course of periodicity for F344 rat exposed 6 h/day, 5 days/week to
theoretical gas with partition coefficients as shown (Jarabek et al., 1990).
(Hb/ )A/(Hb/0)H = the ratio of the blood:gas (air) partition coefficient of the chemical
for the laboratory animal species to the human value. The value of
1.0 is used for the ratio if (Hb/ )A > (Hb/ )H.
In the case where Hb/g values are unknown, the default value of (Hb/ )A/(Hb/0)H = 1 is
recommended. An analysis of the available data on rats for blood:air partition coefficients
shows that the (Hb/ )A is greater than (Hb/ )H in most cases. Gargas et al. (1989) and Jepson
et al. (1994) provide discussion of techniques to derive partition coefficients and report values
for volatile and nonvolative chemicals, respectively.
Figure 4-10 provides guidance on the relationship of the blood:air and fat:blood
partition coefficients with respect to achieving periodicity of an inhaled agent in the arterial
blood of a 380-g F344 rat. (It should be noted that often tissue:air partition coefficients are
4-61
-------
p. 99
C
0)
'o
it
-------
p. 100
10 ppm. Physiologic parameters, such as ventilation rate, were scaled as described in
Appendix J. No metabolic parameters were incorporated in the model for the simulations,
because the arterial blood concentration takes longer to reach periodicity without metabolism.
Therefore, this figure represents the most conservative values for the partition coefficients for
that exposure regimen. The blood:air and fat:blood partition coefficients were chosen based
on sensitivity analyses that indicated these two parameters were important to describing the
time course of the concentration of an agent in the arterial blood, and upon data availability.
The importance of the relationship between the partition coefficients and the attainment
of periodicity is particularly significant when extrapolating from studies of different
durations. For example, for an agent with a blood:air partition coefficient of 1,000 and
a fat:blood partition coefficient of 100, it would be inappropriate to extrapolate from a
subchronic exposure regimen because the criterion of attaining periodicity for 90% of the
exposure duration is not met. Periodicity is attained with these same parameters when the
study is carried out for a longer duration, however, so that the approach based on the ratio of
animahhuman partition coefficients can be used on a chronic study without violation of
critical assumptions.
Similar matrices to Figure 4-10 can be developed for the relationship between partition
coefficients and the attainment of periodicity of the agent in the arterial blood of each
experimental species of interest. Use of physiologic parameters for other species and
different exposure regimens at various concentrations will influence this relationship and
should be considered when determining the extrapolation approach to use for derivation of an
HEC.
Since the requirement for achieving periodicity over 90% of the exposure duration is
based on the objective of limiting error in the estimate to less than 10%, a modifying factor
to account for a greater amount of error should be applied (see Section 4.3.8.1) when the
nature of the inhaled agent (e.g., high fat:blood partition coefficient) suggests this condition
was not met.
4.3.6.3 Additional Assumptions and Default Values
As with aerosols, after evaluation of the adequacy of the generation system, the initial
step in the calculation of HECs is characterization of the exposure.
4-63
-------
p.1
Gas exposures are characterized by concentration (mg/m3), temperature, and pressure.
If the concentration is expressed in ppm, the actual temperature and pressure should be used
•3
to convert the units to mg/m . When the actual temperature and pressure values are not
provided in a study, it should be suspect for deficient reporting of important experimental
detail. Some studies, however, express values already corrected for these parameters, usually
corrected to 25 °C and 760 mm Hg. These values are the recommended default values for
temperature and pressure, respectively.
Other assumptions and default values for gas and vapor extrapolations are provided in
Appendix J.
4.3.7 Derivation and Dosimetric Adjustment Using Human Studies
Whenever possible, a human study is preferred as the critical study for derivation of an
RfC. This avoids the problems of extrapolating from laboratory animals to humans, but has
its own limitations. When using epidemiologic data to assess risk in the context of a method
designed for data on experimental animals, the dependence of epidemiologic studies on
existing exposure conditions and the necessity of using noninvasive diagnostic methods
present two complicating factors. One is that existing exposure levels may not include a
NOAEL. Toxicologic studies are generally designed to identify the NOAEL. For ethical
reasons, many clinical studies in humans often focus on exposure scenarios that are associated
with minimal effects and short exposure durations, although they also may identify a NOEL.
In contrast, epidemiologic studies cannot be designed as rigorously because exposure levels
are dependent on existing exposures. Furthermore, often exposures in epidemiological
studies are poorly characterized. In both controlled human and animal studies, the effect
level estimates are biased by the dose or exposure level selected or available for study. These
effect level estimates are subject to random error, the magnitude of which depends on various
design aspects, such as the size of the study population or test groups, and the underlying
variability of the test animals or study subjects.
The second factor to consider for epidemiological studies is that a broad spectrum of
potential adverse effects cannot be evaluated; therefore, it is difficult to determine the critical
effect. Prospective epidemiologic studies that investigate an array of likely biological
markers or preclinical endpoints are better sources of NOAELs/LOAELs to estimate the
4-64
-------
p. 2
threshold region. Clinical studies may be based on low exposure levels selected by the
investigator and investigate sensitive endpoints, but these studies are generally of short
duration and unless mechanisms of action are unequivocally established, are probably more
useful for estimating short-term effects or to identify potential target tissues for consideration
when evaluating chronic data. The following discussion describes approaches to address the
use of human data for RfC derivation.
4.3.7.1 Selecting the Threshold Estimate
In some epidemiologic studies, only severe effects such as mortality are examined so
that the concept of a NOAEL in inappropriate for RfC derivation. A study in which sensitive
endpoints are evaluated may identify a LOAEL but not a NOAEL. If the effect is sensitive
(i.e., it occurs early in the natural history of the disease), a LOAEL may be judged suitable
for use in calculating an RfC in lieu of a NOAEL, because the uncertainty of extrapolating
human data for a well-defined critical effect from a LOAEL to a NOAEL is judged to be less
than the uncertainty involved in extrapolating from animal data to humans. The
circumstances governing this selection include deficiency in toxicologic and physiologic data
bases, small sample size in the experimental studies, or physiologic or pharmacokinetic data
suggesting that animal data are unlikely to be good predictors for humans.
4.3.7.2 Defining the Exposure Level
Epidemiologists cannot control the exposure levels for a study in a systematic fashion,
but instead attempt to estimate or measure the levels to which the study population is
exposed, insofar as is possible for that study. In actual exposure situations, the levels vary in
time and location. Epidemiologic studies can utilize a variety of parameters to characterize
exposure, although in retrospective studies the available data are usually quite limited.
The ideal exposure measure for humans who move about in their environment is
individual data, such as might be obtained with the use of a personal monitor. However, in
addition to the expense and practical difficulties, this technology is available for measuring
only a few chemicals. Individual exposure can be constructed by mapping the individual's
time in various exposure zones, rooms, or areas. If information on levels in the environment
4-65
-------
p. 3
is not available, duration of employment in a particular job category often is used as a
surrogate for exposure.
Parameters commonly used to measure environmental levels are cumulative exposure,
peak exposure level, time-weighted average, and ratio of average to peak exposure.
Currently, it is unclear which of these is best related to disease. For example, cumulative
exposure is more appropriate as the half-life of a substance is increased. Therefore, to derive
RfCs that identify levels of environmental exposures free of adverse effects, cumulative
exposure or time-weighted averages are appropriate for substances with long half-lives. The
circumstances must be evaluated on a case-by-case basis and different exposure parameters
may be used if the rationale is presented. For conversion of units, the approach is the same
as that for laboratory animal data (Equations 4-la and 4-lb). Considerations for route-to-
route extrapolation would be the same as for laboratory animal data; however, it is highly
unlikely that human ingestion data would be available in a form useful for quantitative
derivation of an RfC.
4.3.7.3 Dosimetric Adjustment for Human Data
When human data are available and adequate to derive an RfC, adjustments are usually
required to account for differences in exposure scenarios (e.g., extrapolation from an 8 h/day
occupational exposure to a continuous chronic exposure). The optimal approach is again to
use a biologically motivated mathematical or PBPK model. An occupational exposure can be
extrapolated in the same fashion as described in Section 4.3.3 to extrapolate intermittent
exposure regimens from experimental laboratory animals, using particle deposition or PBPK
models with human exertion (work) ventilation rates and exposure durations appropriate to
the occupational setting.
In the event that a PBPK model or required physicochemical and physiological
parameters are not available, the default approach for human exposure scenarios is to adjust
by the default occupational ventilation rate and for the intermittent work week schedule:
NOAEL*IHECJ = NOAEL (mg/m3) x (VEho/VEh) x 5 days / 7 days (4-49)
4-66
-------
p. 4
where:
NOAEL*rHECn = the NOAEL or analogous effect level obtained with an alternative
approach as described in Appendix A, dosimetrically adjusted to an
ambient human equivalent concentration;
NOAEL = occupational exposure level (time-weighted average);
o
VEho = human occupational default minute volume (10 m /8 h); and
VEh = human ambient default minute volume (20 m3/24 h).
4.3.7.4 Uncertainty Factors for Human Data
Areas of extrapolation and the UFs applied to account for them are essentially the same
as those for extrapolating laboratory animal data described in Section 4.3.8. The use of
human data, in most cases, will obviate only the use of the UF for interspecies extrapolation.
The best data to use for calculating an RfC would be a population study of humans that
includes sensitive individuals exposed for lifetime or chronic duration, and that evaluates the
critical endpoint or an appropriate early marker for the disease. A NOAEL derived from a
well done epidemiologic study of this description may require no UF. A similar study in
humans that contains only a LOAEL would require the use of a factor of up to 10-fold to
reduce the exposure to the range of a NOAEL. Chronic studies on populations that do not
include sensitive individuals may require a 10-fold UF. For example, studies of workers are
considered to contain only relatively healthy adults. A NOAEL from a study that entails
subchronic exposure would require a reduction by a 10-fold UF. However, the amount of
exposure in a human study that constitutes subchronic is not defined, and could depend on the
nature of the effect and the likelihood of increased severity or greater percent response with
duration. In the absence of data on the relationship of animal to human lifespan for
predicting health effects, a linear correlation of percent lifespan is sometimes assumed. For
example, because a study in animals that is 10% of lifespan is considered subchronic, then
7 years or one-tenth of the assumed human lifetime (70 years) is used as interim guidance for
the superfund program to determine the working cut-off for deriving a subchronic human
study (Means, 1989). Information on the natural history and progression for the disease
should be considered and explained; information on follow-up after exposure, often available
in epidemiologic studies, is important.
4-67
-------
p. 5
In some cases, short-term studies of effects in humans can give important information
on irritation, sensory effects, or sensitivity and reversibility, yet give no information on the
effect of chronic exposure.
4.3.8 Data Array Evaluation and Choice of Principal Study/Studies
Inhalation reference concentrations are typically calculated using a single exposure level
and UFs that account for specific deficiencies in the toxicity data base. Both the exposure
level and the UFs are selected and evaluated in the context of all available chemical-specific
literature. After all toxicological, epidemiologic, and supporting data have been reviewed
and evaluated, a principal study (or studies) is selected that reflects'optimal data on the
critical effect. Dose-response data points for all reported effects are examined as a
component of this review. Issues of particular significance in this endeavor include
• A delineation of all toxic effects and associated exposure levels (see
Section 4.2).
• Dosimetric adjustment to HEC (see Section 4.3).
• Determination, to the extent possible, of effect-specific experimental threshold
regions (i.e., the NOAELrHECi-LOAELrHECi interface or bracket).
• Determination of the critical effect. Of the multiple toxic endpoints potentially
observed, the critical effect selected is defined as the one associated with the
lowest NOAELrHECi-LOAELrHEC] interface or bracket.
• Special consideration of species, portal-of-entry effects, and/or route-specific
differences in pharmacokinetic parameters and the slope of the dose-response
curve.
If multiple NOAELrHECiS for the same critical effect are available in one animal
species, the highest NOAELrHECj for that individual species is compared to NOAELjHECjS
for that effect from other species. If multiple NOAEL^HECjS for the critical effect are
available in different species, the lowest of these NOAELrHECjS, or the NOAEL^HECj for the
most sensitive species, generally is selected as the exposure level that most closely defines the
threshold for adverse effects of the dose-response curve. When disparity in dose-response
patterns is apparent between species, studies need to be evaluated to ascertain, if possible,
4-68
-------
p. 6
whether the differences are due to (1) differences in the monitored endpoints or procedures
across studies, (2) species differences in dose-respone curves, or (3) choice of dose-spacing
(if alternative approaches such as the benchmark or Bayesian approaches described in
Appendix A are not used). If species differences are apparent, the question arises as to which
species is the most appropriate model for humans. Differences in dose-effect curves could be
due to inherent differences in target receptor sensitivity (pharmacodynamics) or to differences
in concentration of the compound or metabolite reaching the receptor (pharmacokinetics).
This distinction is important when trying to identify the most appropriate species for
modeling the human response. Current controversy with respect to the URT in the area of
data array analysis involves the relevance of nasal lesions in laboratory rodents versus humans
or other primates (DeSesso, 1993) and whether nasal lesions in rodents are somehow sentinel
for effects in the lower respiratory tract of primates (Jarabek, 1994). It is consistent with
EPA policy to use data on the most sensitive animal species as a surrogate to humans unless
data exist to the contrary. In the RfC methodology, this evaluation is based on
NOAEL[HEC]s.
Often an appropriate NOAEL,HECi will not be available. In that event, other estimates
of effect-specific thresholds may be used. Based on the dose-effect classification system
presented in Tables 4-2 and 4-3, the following guidelines may be employed (adapted from
Federal Register, 1980):
• An FELrHECi from a study with no other dose-response levels (a free-standing
FELiHECi) is unsuitable for the derivation of an RfC.
• A NOELrHECi from a study with no other dose-response levels is unsuitable for
the derivation of an RfC. If multiple NOEL,HECjS are available without
additional data, NOAEL^HEC,s, or LOAEL^HECjS, the highest NOELiHEC,
should be used.
• A LOAELmECj from a study with no other dose-response levels (a free-standing
LOAELrHECi) is unsuitable for the derivation of an RfC.
• A NOAELrHECj or LOAEL,HEC, supported by other data may be suitable for
RfC derivation. In the case of a LOAEL,HEC,, an additional UF is applied for
extrapolation to NOAEL.
Note: Caution must be exercised not to substitute a FEL^HECjfor a LOAELrHEC,.
4-69
-------
p. 7
If, for reasonably closely spaced doses, only a NOEL[HECj and a LOAELjHECj
of equal quality are available, then the appropriate uncertainty factor is applied
to the NOEL[HEC].
In the course of many risk assessment discussions during the last several years, the EPA
has decided on the following conditions when choosing the appropriate animal effect or
no-effect level as a basis of an RfC. If an appropriate human study with a well-defined
NOAEL[HECj is available as to a chemical's critical effect, it is used in preference to
laboratory animal toxicity data in estimating RfCs. When such human data are not available,
the following sequence is used to choose the appropriate study, species and NOAELrHECn as
a basis of RfC estimation.
The EPA chooses the most appropriate NOAELrHECi of the critical effect from
a well-conducted study on a species that is known to resemble the human in
response to this particular chemical (e.g., by comparative pharmacokinetics).
When the above condition is not met, the EPA generally chooses the most
sensitive study, species, and NOAELrHECj, as judged by an interspecies
comparison of the NOAELrHECj and LOAELrHECi. Table 4-7 outlines
examples of this condition.
4.3.9 Operational Derivation of the Inhalation Reference Concentration
Choice of the effect and its associated concentration that serves as the basis for
derivation of the RfC requires the evaluation of the entire data array of NOAELjHECjS and
LOAEL[HECjS. An example data array is shown in Figure 4-11. The critical toxic effect to
be used in the dose-response assessment is generally the one characterized by the lowest
NOAELrHECj that is representative of the threshold region for the data array. For example,
note in Figure 4-11 that as concentration increases above the NOAEL, the incidence or
severity of the observed toxicity is also increasing. The objective when analyzing such a data
array is to select a prominent toxic effect that is pertinent to the chemical's mechanism of
action and which is at or just below the threshold for the relatively more serious effects.
This approach is based, in part, on the assumption that if the critical toxic effect is prevented,
then all toxic effects are prevented. The determination of the critical toxic effect from all
effects in the data array requires toxicologic judgment because a chemical may elicit more
4-70
-------
p. 8
TABLE 4-7. COMPARISON OF THE HIGHEST INDIVIDUAL SPECIES
AND ITS LOAEL[HEC]a
Effect Level
(mg/m3)
Species
Dog Rat Mouse
Comments
(Given the Same Critical Effect)
Example 1:
LOAEL(HEC)
NOAEL[HEC1
Example 2:
NOAEL[HEC1
100 120
50 60 80
120 100 90
90 75
The proper choice is generally the highest dog
NOAEL1HEC] of 50 mg/m3, since the potential
experimental threshold in dogs (i.e., the potential
LOAEL(HEC]) may be below the highest NOAELim:cjS in
both rats and mice.
Example 3:
NOAEL1I1EC]
Example 4:
The proper choice is generally the mouse LOAEL|H1ZC] of
90 mg/m3, since the potential experimental threshold in
mice may be lower than the highest NOAEL[IIECJs for
both dogs and rats. Judgment is needed in this example
to ensure that the adverse effects seen in all three species
are truly minimal. For example, if any of the
LOAEL1H1£C)s in the species represented an increase in a
severe effect, no firm basis for the development of an
RfC exists. This is based on the general observation that
overt toxicity data are far removed quantitatively from
chronic LOAEL1H1:C)s and NOAEL[m;c]s, and thus, the
data base has failed to establish the likely experimental
threshold for the most sensitive endpoint.
75 80 90 The proper choice is generally the dog LOAEL1Mlic) of
75 mg/m3, since by definition this represents the most
sensitive species (see, however, the caution in
Example 2).
NOAEL111HC]
100 90 120
The proper choice is generally the highest rat
NOAEL[m;c) of 90 mg/m3, since no assurance exists that
the experimental threshold in rats is not below the
highest NOAEL1HEC|s of both dogs and mice. This
situation is unusual and should be judged carefully; since
a LOAEL11ILC, has not been determined, the RfC may be
unduly conservative. Strict interpretation of this example
might lead to strikingly lower RfCs if other species are
tested at much lower doses. Such RfCs may not be
appropriate.
*NOAEL|H1:C] or LOAEL^^q refers to NOAEL or LOAEL concentrations adjusted for dosimetric differences
between laboratory animals and humans to human equivalent concentrations (HECs).
4-71
-------
p. 9
Indicator Enzyme
O Slight Body Weight Decrease
• Liver Fat Infiltration
A Respiratory Eptihelium Hyperplasia
Convulsions
PEL
Concentration
Figure 4-11. Example data array and inhalation reference concentration (RfC)
derivation.
than one toxic effect (endpoint) in tests of the same or different exposure duration, even in
one test species. Further, as discussed in Appendix A, the NOAEL[HECj and LOAELrHECj
obtained from studies depend on the number of laboratory animals or human subjects
examined and on the spacing of the exposure levels. The NOAELrHECj (or LOAELrHECj as
discussed above) from an individual study (or constellation of studies), that is also
representative of the threshold region for the overall data array is the key datum synthesized
from an evaluation of the data array. The study from which this NOAELrHECn
(or LOAELrHECj as discussed above) is estimated is known as the principal study.
Determination of the critical effect for the entire data array and identification of the principal
study represents the first scientific evaluation of the dose-response analysis per se. The
second is the selection of uncertainty factors and operational derivation of the estimate.
4-72
-------
p. 10
4.3.9.1 Application of Uncertainty Factors4
The RfC is a benchmark estimate that is derived from the NOAELrHECi for the critical
effect by consistent application of UFs. The UFs are applied to account for recognized
uncertainties in the extrapolations from the experimental data conditions to an estimate
appropriate to the assumed human scenario. Determination of which UFs to apply and the
magnitude of each represents the second scientific evaluation required for an RfC dose-
response assessment. The standard UFs applied are those for the following extrapolations (as
required): (1) data on effects of average healthy humans to sensitive humans; (2) laboratory
animal data to humans; (3) studies of subchronic to chronic duration; (4) a LOAEL[HECj to a
NOAELjHECj; and (5) from an incomplete to complete data base. The UFs are generally an
order of magnitude, although incorporation of dosimetry adjustments or other mechanistic
data has routinely resulted in the use of reduced UFs for RfCs. The composite UF applied to
an RfC will vary in magnitude depending on the number of extrapolations required. An RfC
will not be derived when use of the data involve greater than four areas of extrapolation,
however. The composite UF when four factors are used generally is reduced from 10,000 to
3,000 in recognition of the lack of independence of these factors. This coalescing of several
areas of uncertainty is based on the knowledge that each individual factor is generally
conservative from the standpoint of the behavior of the average chemical (Dourson and Stara,
1983), and that the multiplication of four or five values of 10 is likely to yield unrealistically
conservative RfCs.
An additional modifying factor (MF) may also be applied when scientific uncertainties
in the study chosen for derivation are not explicitly addressed by the standard UFs. For
example, an MF might be applied to account for a statistically minimal sample size or for
poor exposure characterization.
''Other authors have discussed these areas of uncertainty or UFs in general. The reader is referred to Zielhuis
and Van der Kreek (1979) for a discussion of these factors in setting health-based permissible levels for
occupational exposure, and Dourson and Stara (1983) for a summary of these factors regarding oral exposures.
Other publications include Gaylor (1983), who discusses the use of safety factors for controlling risk; Crump
(1984), who discusses problems with the current methods that includes UFs; Krewski et al. (1984), who
contrast safety factors and mathematical models as methods for determining "safe" levels of exposure; Calabrese
(1985), who discusses UFs and interindividual variation; and Lu (1983, 1985b), who discusses safety factors
from the perspective of the World Health Organization. Lewis et al. (1990) have proposed an operational
alternative approach. Renwick (1991) has outlined a flexible scheme based on the nature of toxicity, knowledge
of metabolism, and information of human heterogeneity.
4-73
-------
p. 11
Thus, notationally, the RfC is defined as:
RfC = NOAEL*[HEC] / (UF X MF), (4-50)
where:
NOAEL*[HECj = The NOAEL or analogous effect level obtained with an alternate
approach as described in Appendix A, dosimetrically adjusted to an
HEC;
UF = Uncertainty factor(s) applied to account for the extrapolations required
from the characteristics of the experimental regimen; and
MF = Modifying factor to account for scientific uncertainties in the study
chosen as the basis for the operational derivation.
It must be emphasized that the RfC as a quantitative dose-response estimate is not
numeric alone. As risk assessments have become a more prevalent basis for decision-making,
their scientific quality and clarity have gained unprecedented importance (American Industrial
Health Council, 1989; National Research Council, 1994). Due to the complexity of many
risk assessments, desirable attributes include the explicit treatment of all relevant information
and the expression of uncertainty in each element (i.e., hazard identification, dose-response
assessment, exposure assessment, and risk characterization). Any dose-response assessment,
such as the RfC, has inherent uncertainty and imprecision because the process requires some
subjective scientific judgment, use of default assumptions, and data extrapolations.
A complete dose-response evaluation should include communication of the rationale for
data selection, the strengths and weaknesses of the data base, key assumptions, and resultant
uncertainties (Habicht, 1992; American Conference of Governmental Industrial Hygienists,
1986). The rationale for the choice of the data from which the RfC is derived, a discussion
of data gaps, and the resultant confidence in the RfC are all outlined on the summary of the
RfC entered on the EPA's Integrated Risk Information System (IRIS). A discussion and
rationale for the uncertainty factors used in the RfC derivation are also provided. This
information is an important part of the RfC and must be considered when evaluating the RfC
as a dose-response estimate, in addition to assumptions and resultant uncertainties inherent in
an exposure assessment, when attempting to integrate the assessments into a risk
4-74
-------
p. 12
characterization. Additional guidance on the assignment of confidence levels is provided in
Section 4.3.9.2.
Uncertainty factors are associated with various specific recognized uncertainties in
extrapolating from the type of study serving as the basis for the RfC to the scenario of
interest for the risk assessment as outlined in Table 4-8. The processes thought to be
encompassed by each factor are provided in Table 4-9.
An additional MF may be used to account for uncertainties in the study chosen for
derivation. For example, a MF may be applied to account for a study of statistically minimal
sample size or with poor exposure characterization. The effect of small sample size has long
been recognized in toxicology (Bliss, 1938), and recent research has focused on adjusting for
this by taking the power of individual studies into account (Brown and Erdreich, 1989).
Considerations of the sensitivity of the NOAEL/LOAEL approach to sample size and dose
spacing has led to the development of the alternative approaches to derivation discussed in
Appendix A.
In general, the choice of UFs applied reflects the uncertainty associated with estimation
of an RfC from different human or laboratory animal toxicity data bases. When sufficient
human data are available on a chemical's critical effect and pharmacokinetics, the UFs may
be smaller than those described in Table 4-8, or unnecessary. For example, if sufficient data
from chronic duration exposure studies are available on the threshold region of a chemical's
critical toxic effect in a known sensitive human population, then the UF used to estimate the
RfC may be 1. That is, these data are judged to be sufficiently predictive of a human
population subthreshold dose, so that additional UFs are not needed. Likewise, in cases
where data do not completely obviate the uncertainty for which a given UF is applied, or
appears to be intermediate in fulfilling that requirement, an intermediate UF is suggested to
estimate the RfC (Federal Register, 1980). Composite factors are sometimes applied to
account for partial uncertainty under more than one UF. For example, a 10-fold factor may
be applied to account for partial uncertainty due to botli the use of less than chronic data and
a LOAEL, if data supported that the effect was of minimal severity and the lesion did not
progress significantly with duration. When a single subchronic study that does not define a
NOAEL is the only available information, the EPA recognizes that all five areas of
4-75
-------
p. 13
* THE USE OF UNCERTAINTY FACTORS IN
f REFERENCE CONCENTRATION (RfC)
£2
r^\ H^
S <
g j
c
00 -.
4 &
2 w
H Q
J
C2
<
H
£ C
Use a 10-fold factor when extrapolating from valid experimental results from studi
using prolonged exposure to average healthy humans. This factor is intended to
account for the variation in sensitivity among the members of the human populatio
c
re
3
f-i
c
re
r
D
O
C
D
•o
^H
re
T3
C
re
£ 3C
S «
Use an additional threefold factor when extrapolating from valid results of long-tei
studies on experimental animals when results of studies of human exposure are not
available or are inadequate. This factor is intended to account for the uncertainty i
c
c
3
JS
O
re
1
<
>,
extrapolating animal data to the case of average healthy humans. Use of a 3 is
recommended with default dosimetric adjustments. More rigorous adjustments ma'
allow additional reduction. Conversely, judgment that the default may not be
appropriate could result in an application of a 10-fold factor.
Use up to an additional 10-fold factor when extrapolating from less than chronic
results on experimental animals or humans when there are no useful long-term
o
1
o
c
0
'c
O
V-
"1
f
to
CO
human data. This factor is intended to account for the uncertainty in extrapolating
from less than chronic NOAELs to chronic NOAELs.
Use up to an additional 10-fold factor when deriving an RfC from a LOAEL^q,
instead of a NOAEI^HEq. This factor is intended to account for the uncertainty in
extrapolating from LOAEL^qS to NOAEL^qS.
FT
a
_j~
w
<
O
z
Q
E
_r
w
<
0
j
Use up to a 10-fold factor when extrapolating from valid results in experimental
animals when the data are "incomplete". This factor is intended to account for the
S
re
X)
re
re
-a
3)
13
c.
c
o
O
O
"o
G*
o
0
J3
a
u
•*-» V <4-l
« § &
& $ X
•1$^
fj. S §
1-al
- .52 -5
.29*-
III
o 12 ®
a 8 -
s 't; o
« o .^3
w >-, 'C
2 « -1
11 !
Ill
™ O w
0 0 «
~ w ^
•^ Sg .9
B ^j «
« o i
| D a
'S gj •£
« 60 '
u 2 2
"M w S
.9 ,- 8
§8^
•S:c |
^^
111
1^
derivation is the most sensitive or appropriate.
£
5
^,
c
o
re
U,
Cl
c
-, SO
— re m;
^ £ «
E re >*•
I 3 -s
&£ >
O 'C
— c« m
•O "^3
5 '> „
~ 7; C
^ ° C
C c o>
"O C ^
•2 3 §
t .> '-i
D S3 5
0 "° g
- 0 §
<~ * c
C u, u
« p u
CO "^ 0
C [7 V-
-------
p. 14
TABLE 4-9. THE USE OF UNCERTAINTY FACTORS IN DERIVING AN
INHALATION REFERENCE CONCENTRATION
Standard Uncertainty Factors (UFs)
Processes Considered in UF Purview
H = Human to sensitive human
Extrapolation of valid experimental results from studies
using prolonged exposure to average healthy humans.
Intended to account for the variation in sensitivity
among the members of the human population.
A = Animal to human
Extrapolation from valid results of long-term studies
on laboratory animals when results of studies of
human exposure are not available or are inadequate.
Intended to account for the uncertainty in extrapolating
laboratory animal data to the case of average healthy
humans.
S = Subchronic to chronic
Extrapolation from less than chronic exposure results
on laboratory animals or humans when there are no
useful long-term human data.
Intended to account for the uncertainty in extrapolating
from less than chronic NOAELs to chronic NOAELs.
L = LOAEL to NOAEL
Derivation from a LOAEL instead of a NOAEL.
Intended to account for the uncertainty in extrapolating
from LOAELs to NOAELs.
D = Incomplete to complete data
Extrapolation from valid results in laboratory animals
when the data are "incomplete".
Intended to account for the inability of any single
laboratory animal study to adequately address all
possible adverse outcomes in humans.
Pharmacokinetics/Pharmacodynamics
Sensitivity
Differences in mass (children, obese)
Concomitant exposures
Activity pattern
Does not account for idiosyncracies
Pharmacokinetics/Pharmacodynamics
Relevance of laboratory animal model
Species sensitivity
Accumulation/Cumulative damage
Pharmacokinetics/Pharmacodynamics
Severity of effect
Recovery
Duration of study
Consistency of effect with duration
Severity
Pharmacokinetics/Pharmacodynamics
Slope of dose-response curve
Trend, consistency of effect
Relationship of endpoints
Functional vs. histopathological evidence
Exposure uncertainties
Quality of critical study
Data gaps
Power of critical study/supporting studies
Exposure uncertainties
uncertainty are present, and an RfC will not be derived. An RfC will also not be derived in
the absence of data on the potential respiratory tract toxicity.
It should be noted that the basis for the UFs is empirical and has been derived from oral
data (Dourson and Stara, 1983). In most cases, support of each UF was based on analysis of
the ratios of effect levels. For example, analysis of the ratios of NOAELs from 90-day
4-77
-------
p. 15
studies compared to NOAELs from chronic studies was used to support a 10-fold factor to
account for subchronic to chronic extrapolation. Because the different types of toxicity
(portal-of-entry versus remote) may have different determinants underlying ethe exposure-
dose-response continuum for which the default dosimetry adjustments only partially account,
the appropriate magnitude for these UFs when using inhalation data is a topic of ongoing
research at the EPA. Estimation procedures that are not sensitive to the spacing of exposure
concentrations such as the benchmark and Bayesian approaches discussed in Appendix A are
being explored instead of the previously used ratios approach for this research.
A UF is generally used to calculate RfCs with appropriate chronic human data, and is
intended to account for intraspecies human variability to the adverse effects of a chemical
(i.e., H in Tables 4-8 and 4-9). Empiric support for a 10-fold value for this UF is based on
analyses of single-dose oral data (Weil, 1972; Dourson and Stara, 1983). Hattis et al. (1987)
also suggest that a value of 10 is generally appropriate for this UF based on an analysis of
human variability for key pharmacokinetic parameters.
For derivation of the RfC, the UF applied for interspecies extrapolation (i.e, A in
Tables 4-8 and 4-9) is 3 due to the incorporation of dosimetric adjustments. If more rigorous
adjustments can be made, an additional reduction of the UF would be warranted. The
threefold factor represents the reduction of the usual 10-fold factor by half (i.e., 10'5) since
the default dosimetry accounts for variability in disposition (pharmacokinetics). The residual
uncertainty is envisioned to address species differences in pharmacodynamics. A similar
scheme was proposed by Renwick (1991), although the dosimetry adjustments in the RfC
methods explicitly address disposition of particles and gases via inhalation. The empiric basis
of this UF was originally based on oral data (Dourson and Stara, 1983). An analysis by
Jarabek and Hasselblad (1991) showed that the deviation across species and chemicals for
HEC estimates was reduced approximately 2-fold versus that using previous (Federal
Register, 1980) derivation methods. The dosimetric adjustments have also been shown to be
consistently less than those calculated with previous methods so that a reduction in the
UF was further supported (Jarabek et al, 1989; Overtoil and Jarabek, 1989a,b).
An RfC based on a NOAEL|HEC, with satisfactory subchronic laboratory animal data
would require a factor to address the uncertainty in extrapolating data from subchronic to
chronic exposures (i.e., S in Table 4-8). Empirical evidence supporting the proposition that
4-78
-------
p. 16
subchronic toxicity data can be used with a 10-fold UF is again based on analyses of oral
toxicity data (Dourson and Stara, 1983; Weil and McCollister, 1963; Weil et al., 1969).
McNamara (1976) also demonstrated that a 10-fold factor applied to a subchronic NOEL
would predict a chronic NOEL for 95% of the 122 compounds for which both chronic and
subchronic data for the oral route of exposure were available. To the degree that route-
specific and duration-specific data are not available, increased reliance on additional
extrapolation assumptions and a larger UF is necessary. The lack of data with appropriate
duration becomes of greater concern when either the chemical itself or its damage has the
potential to accumulate. Conversely, if the effect is more dependent on concentration than
duration, and progression of the lesion (either in incidence or severity) is not evident, a
reduced UF may be considered.
Generally, a UF is applied to estimate RfCs using LOAELs if NOAELs are unavailable
(i.e., L in Tables 4-8 and 4-9). This UF is employed to define an exposure level below the
LOAEL expected to be in the range of a NOAEL. The empiric support for this UF was
based on frequency analyses of LOAEL to NOAEL ratios for oral toxicity data after either
subchronic or chronic exposures (Dourson and Stara, 1983; Weil and McCollister, 1963).
In practice, this UF has varied and its value is chosen based on the severity of the adverse
effect of the LOAEL. For example, if the LOAEL represents liver cell necrosis, a higher
value is suggested for this UF than would be suggested if the LOAEL were based on fatty
infiltration because the hypothesized NOAEL should be closer to the less severe LOAEL
(Dourson and Stara, 1983).
Under some circumstances, a UF is applied when the data base is deficient in
comprehensiveness; for example, if it lacks a two-generation reproductive study (i.e., D in
Tables 4-8 and 4-9). The rationale for the minimum data base criteria provided in
Section 4-1 can provide guidance on when a UF for lack of comprehensiveness is warranted.
Dourson et al. (1992) have shown this to be an appropriate factor for oral data. The
requirement for data in a second species is also supported by analyses that have shown lack of
concordance for target tissues across species (Appelman and Feron, 1986; Heywood, 1981,
1983). The U.S. Food and Drug Administration has addressed the data base deficiencies
issue with the use of a twofold safety factor. Therefore, in situations where a subchronic
animal bioassay was available, but information in a second experimental species was lacking,
4-79
-------
p. 17
a 2,000-fold safety factor (i.e., 2D x 10H x 10A x 10S) was used to estimate an acceptable
daily intake (Shibko, 1981). The influence that the requirement for portal of entry data and
dosimetric adjustments used in the RfC methods may have on this UF has not yet been
quantified.
There are certain circumstances specific to inhalation that may require changes in UFs.
For example, the UF used when extrapolating from a subchronic to a chronic study is
assumed to be adequate for oral studies in the great majority of cases. A UF of extrapolation
of subchronic to chronic exposures for inhalation studies also should be adequate with certain
exceptions. Possible exceptions include the following:
Exposure to chemicals that are considered likely to induce hypersensitivity (see
Section 2.1.2.3),
Exposure to chemicals that are considered likely to induce very slowly
developing ("smoldering") effects (e.g., beryllium), and
Exposure to inhaled relatively insoluble particulate matter where the clearance
rate may slow or stop when a threshold for clearance is reached. Therefore,
after long-term exposure, lung loads can reach much higher levels than could
reasonably be expected from lower level, chronic exposure conditions.
The appropriate UF for these situations should be decided on a case-by-case basis until
more definitive guidelines are available.
4.3.9.2 Assignment of Confidence Levels
The selection of a NOAELrHECj or other appropriate measure of threshold response
involves a process that incorporates scientific subjective judgment and statistical measures of
significance. The qualitative and quantitative nature of this process results in an RfC
associated with varying degrees of confidence that can be described as high, medium, and
low.
A confidence level of high, medium, or low is assigned to the study used in the
operational derivation, the overall data base, and to the RfC itself. Confidence ascribed to
the RfC estimate is a function of both the confidence in the quality of the study and
confidence in the completeness of the supporting data base together, with the data base
4-80
-------
p. 18
confidence taking precedence over that assigned to the study. High confidence in the RfC is
an indication that the data base included investigation of a comprehensive array of noncancer
toxicity endpoints, established from studies of chronic duration in various mammalian
species, and that the study (or studies) established an unequivocal NOAEL[HEq. Therefore,
a high confidence RfC is not likely to change as more data become available, with the
exception of additional mechanistic data or sophisticated tests that may change the perspective
of the evaluation. Low confidence in an RfC is usually applied to a derivation that is based
on several extrapolations and indicates an estimate that may be especially vulnerable to
change if additional data become available. If the individual study is of excellent quality, it
most likely will receive a high confidence rating, although it may be subchronic in duration.
Duration of the chosen study, as well as supporting studies and the spectrum of investigated
endpoints (e.g., reproductive effects), are considered in the rating of confidence in the data
base. Low confidence in the data base might be given to an excellent chosen subchronic
study with few supporting studies and few endpoints examined. The confidence in the RfC
then would reflect these two ratings by a rating of medium to low, indicating uncertainty
(lack of confidence) and suggesting that further investigations may refine this number.
The level of confidence in a particular threshold value will be higher if it is derived
from human data and supported by laboratory animal data. The parameters and factors
involved in the evaluation of human data are described in Section 3.1.1. The degree of
confidence in a particular laboratory animal study involves a number of parameters. These
parameters include, but are not limited to, the following.
• Adequacy of study design
- Is the route of exposure relevant to humans?
- Were an appropriate number of animals and of both sexes used for determination of
statistical significance?
- Was the duration of exposure sufficient to allow results to be extrapolated to humans
under different exposure conditions?
- Were appropriate statistical techniques applied?
- Were the analytical techniques sufficient to adequately measure the level of the test
substance in the exposure protocol, including biological media?
4-81
-------
p. 19
- Is the animal species and strain appropriate as a surrogate for humans?
- Are the techniques for measurement of the biological endpoints scientifically sound
and of sufficient sensitivity?
- To what degree may the biological endpoints be extrapolated (qualitatively or
quantitatively) to humans?
Demonstration of dose-response relationships
- Were sufficient exposure levels used to demonstrate the highest NOAEL for the
endpoint of concern?
- Is the shape of the dose-response curve consistent with the known pharmacokinetics
of the test substance?
- Has the dose-response curve been replicated by or is it consistent with data from
other laboratories and other laboratory animal species?
Species differences
- Are the metabolism and pharmacokinetics in the animal species similar to those for
humans?
- Is the species response consistent with that in other species?
- Is the species from which the threshold value was derived the most sensitive species?
Other factors
- The number of biological endpoints evaluated and associated with dose-response
relationships,
- Sufficient description of exposure protocol, statistical tests, and results to make an
evaluation, and
- Condition of animals used in the study.
4-82
-------
p. 20
5. REFERENCES
Adolph, E. F. (1949) Quantitative relations in the physiological constitutions of mammals. Science (Washington,
DC) 109: 579-585.
Aharonson, E. F.; Menkes, H.; Gurtner, G.; Swift, D. L.; Proctor, D. F. (1974) Effect of respiratory airflow
rate on removal of soluble vapors by the nose. J. Appl. Physiol. 37: 654-657.
Alarie, Y. (1981) Dose-response analysis in animal studies: prediction of human responses. Environ. Health
Perspect. 42: 9-13.
Alarie, Y. (1984) Establishing threshold limit values for airborne sensory irritants from an animal model and the
mechanisms of action of sensory irritants. In: Esmen, N. A.; Mehlman, M. A., eds. Occupational and
industrial hygiene: concepts and methods. Princeton, NJ: Princeton Scientific Publishers, Inc.;
pp. 153-164. (Advances in modern environmental toxicology: v. 8).
Albert, R. E.; Lippmann, M.; Peterson, H. T., Jr. (1971) The effects of cigarette smoking on the kinetics of
bronchial clearance in humans and donkeys. In: Walton, W. H,, ed. Inhaled particles III,
v. 1: proceedings of an international symposium organized by the British Occupational Hygiene Society;
September 1970; London, United Kingdom. Surrey, United Kingdom: Unwin Brothers Limited;
pp. 165-182.
Allen, B. C.; Fisher, J. W. (1993) Pharmacokinetic modeling of trichloroethylene and trichloroacetic acid in
humans. Risk Anal. 13: 71-86.
Amdur, M. O.; Mead, J. (1958) Mechanics of respiration in unanesthetized guinea pigs. Am. J. Physiol.
192: 364-368.
American Conference of Governmental Industrial Hygienists. (1978) Air sampling instruments: for evaluation of
atmospheric contaminants, 5th ed. Cincinnati, OH: American Conference of Governmental Industrial
Hygienists.
American Conference of Governmental Industrial Hygienists. (1986) Documentation of the threshold limit values
and biological exposure indices. 5th ed. Cincinnati, OH: American Conference of Governmental
Industrial Hygienists, Inc.; pp. 117, 313.
American Industrial Health Council. (1989) Presentation of risk assessments of carcinogens: report of an ad hoc
study group on risk assessment presentation. Washington, DC: American Industrial Health Council.
American Industrial Health Council. (1992) Improving risk characterization: summary of a workshop; September
1991; Washington, DC. Washington, DC: American Industrial Health Council.
American Thoracic Society. (1979) ATS statement—Snowbird workshop on standardization of spirometry.
Am. Rev. Respir. Dis. 119: 831-838.
American Thoracic Society. (1982) Evaluation of impariment/disability secondary to respiratory disease.
Am. Rev. Respir. Dis. 126: 945-951.
American Thoracic Society. (1985) Guidelines as to what constitutes an adverse respiratory health effect, with
special reference to epidemiologic studies of air pollution. Am. Rev. Respir. Dis. 131: 666-668.
5-1
-------
p. 21
American Thoracic Society. (1986) Evaluation of impairment/disability secondary to respiratory disorders.
Am. Rev. Respir. Dis. 133: 1205-1209.
American Thoracic Society. (1987a) Standardization of spirometry—1987 update. Am. Rev. Respir. Dis.
136: 1285-1298.
American Thoracic Society. (1987b) Single breath carbon monoxide diffusing capacity (transfer factor):
recommendations for a standard technique. Am. Rev. Respir. Dis. 136: 1299-1307.
American Thoracic Society. (1991) Lung function testing: selection of reference values and interpretive strategies.
Am. Rev. Respir. Dis. 144: 1202-1218.
American Thoracic Society. (1993) Guidelines for the evaluation of impairment/disability in patients with asthma.
Am. Rev. Respir. Dis. 147: 1056-1061.
Andersen, M. E. (1981) A physiologically based toxicokinetic description of the metabolism of inhaled gases and
vapors: analysis at steady state. Toxicol. Appl. Pharmacol. 60: 509-526.
Andersen, M. E. (1987) Tissue dosimetry in risk assessment, or what's the problem here anyway?
In: Pharmacokinetics in risk assessment: drinking water and health, v. 8. Washington, DC: National
Academy Press; pp. 8-23.
Andersen, M. E.; Clewell, H. L, III; Gargas, M. L.; Smith, F. A.; Reitz, R. H. (1987a) Physiologically based
pharmacokinetics and the risk assessment process for methylene chloride. Toxicol. Appl. Pharmacol.
87: 185-205.
Andersen, M. E.; MacNaughton, M. G.; Clewell, H. J., Ill; Paustenbach, D. J. (1987b) Adjusting exposure
limits for long and short exposure periods using a physiological pharmacokinetic model. Am. Ind. Hyg.
Assoc. J. 48: 335-343.
Andersen, M. E.; Clewell, H. M., Ill; Gargas, M. L.; MacNaughton, M. G.; Reitz, R. H.; Nolan, R. J.;
McKenna, M. J. (1991) Physiologically based pharmacokinetic modeling with dichloromethane, its
metabolite, carbon monoxide, and blood carboxyhemoglobin in rats and humans. Toxicol. Appl.
Pharmacol. 108: 14-27.
Andersen, M. E.; Krishnan, K.;*Conolly, R. B.; McClellan, R. O. (1992) Mechanistic toxicology research and
biologically-based modeling: partners for improving quantitative risk assessments. CUT Activities
12 (1): 1-7.
Andre, S.; Metvier, H.; Lantenois, G.; Masse, R. (1987) Solubility of particles using cultured alveolar
macrophages. Comparison with in vitro acellular model and data of in vivo solubility. Presented at: The
design and interpretation of inhalation studies; March; Hannover, Federal Republic of Germany.
Appelman, L. M.; Feron, V. J. (1986) Significance of the dog as 'second animal species' in toxicity testing for
establishing the lowest 'no-toxic-effect level.' J. Appl. Toxicol. 6: 271-279.
AviadOj D. M. (1978) Effects of fluorocarbons, chlorinated solvents, and inosine on the cardiopulmonary system.
Environ. Health Perspect. 26: 207-215.
Barnes, D. G.; Dourson, M. (1988) Reference dose (RfD): description and use in health risk assessments. Regul.
Toxicol. Pharmacol. 8: 471-486.
5-2
-------
p. 22
Baron, J.; Burke, J. P.; Guengerich, F. P.; Jakoby, W. B.; Voight, J. M. (1988) Sites for xenobiotic activation
and detoxication within the respiratory tract: implications for chemically induced toxicity. Toxicol. Appl.
Pharmacol. 93: 493-505.
Barrow, C. S. (1989) Generation and characterization of gases and vapors. In: McClellan, R. O.; Henderson,
R. F., eds. Concepts in inhalation toxicology. New York, NY: Hemisphere Publishing Corp.; pp. 63-84.
Becklake, M. R.; Rodarte, J. R.; Kalica, A. R. (1988) NHLBI workshop summary: scientific issues in the
assessment of respiratory impairment. Am. Rev. Respir. Dis. 137: 1505-1510.
Berggren, M.; Dawson, J.; Moldeus, P. (1984) Glutathione biosynthesis in the isolated perfused rat lung:
utilization of extracellular glutathione. FEES Lett. 176: 189-192.
Bigwood, E. J. (1973) The acceptable daily intake of food additives. Crit. Rev. Toxicol. 2: 41-93.
Bird, R. B.; Stewart, W. E.; Lightfoot, E. N. (1960) Transport phenomena. New York, NY: John Wiley and
Sons.
Bliss, C. I. (1938) The determination of the dosage-mortality curve from small numbers. Q. J. Pharm.
Pharmacol. 11: 192-216.
Bliss, C. I. (1940) The relation between exposure time, concentration and toxicity in experiments on insecticides.
Ann. Entomol. Soc. Am. 33: 721-766.
Bliss, C. I.; James, A. T. (1966) Fitting the rectangular hyperbola. Biometrics 22: 573-602.
Bogdanffy, M. S.; Taylor, M. L. (1993) Kinetics of nasal carboxylesterase-mediated metabolism of vinyl acetate.
Drug Metab. Dispos. 21: 1107-1111.
Bogdanffy, M. S.; Randall, H. W.; Morgan, K. T. (1986) Histochemical localization of aldehyde dehydrogenase
in the respiratory tract of the Fischer-344 rat. Toxicol. Appl. Pharmacol. 82: 560-567.
Bogdanffy, M. S.; Randall, H. W.; Morgan, K. T. (1987) Biochemical quantitation and histochemical
localization of carboxylesterase in the nasal passages of the Fischer-344 rat and B6C3F1 mouse. Toxicol.
Appl. Pharmacol. 88: 183-194.
Bogdanffy, M. S.; Kee, C. R.; Hinchman, C. A.; Trela, B. A. (1991) Metabolism of dibasic esters by rat nasal
mucosal carboxylesterase. Drug Metab. Dispos. 19: 124-129.
Bohning, D. E.; Atkins, H. L.; Cohn, S. H. (1982) Long-term particle clearance in man: normal and impaired.
In: Walton, W. H., ed. Inhaled particles V: proceedings of an international symposium; September 1980;
Cardiff, Wales. Ann. Occup. Hyg. 26: 259-271.
Bois, F. Y.; Zeise, L.; Tozer, T. N. (1990) Precision and sensitivity of pharmacokinetic models for cancer risk
assessment: tetrachloroethylene in mice, rats, and humans. Toxicol. Appl. Pharmacol. 102: 300-315.
Bond, J. A (1989) Factors modifying the disposition of inhaled organic compounds. In: McClellan, R. O.;
Henderson, R. F,, eds. Concepts in inhalation toxicology. New York, NY: Hemisphere Publishing;
pp. 249-298.
Bos, P. M. J.; Zwart, A.; Reuzel, P. G. J.; Bragt, P. C. (1992) Evaluation of the sensory irritation test for the
assessment of occupational health risk. Crit. Rev. Toxicol. 21: 423-450.
Bowden, D. H. (1983) Cell turnover in the lung. Am. Rev. Respir. Dis. 128: S46-S48.
5-3
-------
p. 23
Bowden, D. H. (1986) Macrophages, dust, and pulmonary diseases. Exp. Lung Res. 12: 89-107.
Boyd, M. R. (1980) Biochemical mechanisms in chemical-induced lung injury: roles of metabolic activation.
Crit. Rev. Toxicol. 7: 103-176.
Boyd, M. R.; Statham, C. N. (1983) The effect of hepatic metabolism on the production and toxicity of reactive
metabolites and extrahepatic organs. Drug Metab. Rev. 14: 35-47.
Boxenbaum, H. (1982) Interspecies scaling, allometry, physiological time, and the ground plan of
pharmacokinetics. J. Pharmacokinet. Biopharm. 10: 201-227.
Brady, M. E. (1976) Gastrointestinal drug absorption in rats expsoed to ^Co 7-radiation [dissertation]. Pullman,
WA: Washington State University. Available from: University Microfilms, Ann Arbor, MI; 76-27,715.
Brain, J. D. (1986) Toxicological aspects of alterations of pulmonary macrophage function. Ann. Rev.
Pharmacol. Toxicol. 26: 547-565..
Brain, J. D.; Mensah, G. A. (1983) Comparative toxicology of the respiratory tract. Am. Rev. Respir. Dis.
128: S87-S90.
Breysse, P. N.; Swift, D. L. (1990) Inhalability of large particles into the human nasal passage: in vivo studies
in still air. Aerosol Sci. Technol. 13: 459-464.
Briatico-Vangosa, G.; Braun, C. L. J.; Cookman, G.; Hofmann, T.; Kimber, I.; Loveless, S. E.; Morrow, T.;
Pauluhn, J.; Sorensen, T.; Niessen, H. J. (1994) Respiratory allergy: hazard identification and risk
assessment. Fundam. Appl. Toxicol. 23: 145-158.
Brown, K. G.; Erdreich, L. S. (1989) Statistical uncertainty in the no-observed-adverse-effect level. Fundam.
Appl. Toxicol. 13: 235-244.
Bruce, M. C.; Bruce, E. N.; Leith, D. E.; Murphy, S. D. (1979) Diethyl maleate and/or ozone (10 ppm) reduce
ventilation by 60-80% in awake mice. Physiologist 22: 16.
Buckley, L. A.; Jiang, X. Z.; James, R. A.; Morgan, K. T.; Barrow, C. S. (1984) Respiratory tract lesions
induced by sensory irritants at the RD50 concentration. Toxicol. Appl. Pharmacol. 74: 417-429.
Bull, R. J. (1989) Decision model for the development of biomarkers of exposure. Las Vegas, NV: U.S.
Environmental Protection Agency, Environmental Monitoring Systems Laboratory; EPA report no.
EPA/600/X-89/163.
Burger, G. T.; Renne, R. A.; Sagartz, J. W.; Ayres, P. H.; Coggins, C. R. E.; Mosberg, A. T.; Hayes, A. W.
(1989) Histologic changes in the respiratory tract induced by inhalation of xenobiotics: physiologic
adaptation or toxicity? Toxicol. Appl. Pharmacol. 101: 521-542.
Burri, P. H. (1985) Morphology and respiratory function of the alveolar unit. In: Immunopathology and
immunopharmacology of the lung: advanced course at the Allessandro Volta Center; June 1984; Villa
Olmo, Como, Italy. Int. Arch. Allergy Appl. Immunol. 76(suppl. 1): 2-12.
Calabrese, E. J. (1978) Pollutants and high-risk groups: the biological basis of increased human susceptibility to
environmental and occupational pollutants. New York, NY: John Wiley and Sons.
Calabrese, E. J. (1981) Nutrition and environmental health: the influence of nutritional status on pollutant
toxicity and carcinogenicity. New York, NY: John Wiley and Sons.
5-4
-------
p. 24
Calabrese, E. J. (1983) Principles of animal extrapolation. New York, NY: John Wiley & Sons, Inc.
Calabrese, E. J. (1985) Uncertainty factors and interindividual variation. Regul. Toxicol. Pharmacol. 5: 190-196.
Casanova-Schmitz, M.; Starr, T. B.; Heck, H. d'A. (1984) Differentiation between metabolic incorporation and
covalent binding in the labeling of macro-molecules in the rat nasal mucosa and bone marrow by inhaled
14C- and 3H-formaldehyde. Toxicol. Appl. Pharmacol. 76: 26-44.
Casanova, M.; Morgan, K. T.; Steinhagen, W. H.; Everitt, J. I.; Popp, J. A.; Heck, H. d'A. (1991) Covalent
binding of inhaled formaldehyde to DNA in the respiratory tract of rhesus monkeys: pharmacokinetics,
rat-to-monkey interspecies scaling, and extrapolation to man. Fundam. Appl. Toxicol. 17: 409-428.
Casarett, L. J. (1975) Toxicology of the respiratory system. In: Casarett, L. J.; Doull, J., eds. Toxicology: the
basic science of poisons. New York, NY: MacMillan Publishing Co., Inc.; pp. 201-224.
Castleman, B. I.; Ziem, G. E. (1988) Corporate influence on threshold limit values. Am. J. Ind. Med.
13: 531-559.
Chan, T. L.; Lippmann, M. (1980) Experimental measurements and empirical modelling of the regional
deposition of inhaled particles in humans. Am. Ind. Hyg. Assoc. J. 41: 399-409.
Chan, T. L.; Lee, P. S.; Hering, W. E. (1984) Pulmonary retention of inhaled diesel particles after prolonged
exposures to diesel exhaust. Fundam. Appl. Toxicol. 4: 624-631.
Chan-Yeung, M. (1987) Evaluation of impairment/disability in patients with occupational asthma. Am. Rev.
Respir. Dis. 135: 950-951.
Chan-Yeung, M.; Lam, S. (1986) Occupational asthma. Am. Rev. Respir. Dis. 133: 686-703.
Chandra, S. V.; Shukla, G. S.; Srivastava, R. S.; Singh, H.; Gupta, V. P. (1981) An exploratory study of
manganese exposure to welders. Clin. Toxicol. 18: 407-416.
Chang, J. C. F.; Gross, E. A.; Swenberg, J. A.; Barrow, C. S. (1983) Nasal cavity deposition, histopathology,
and cell proliferation after single or repeated formaldehyde exposures in B6C3F1 mice and F-344 rats.
Toxicol. Appl. Pharmacol. 68: 161-176.
Checkoway, H.; Rice, C. H. (1992) Time-weighted averages, peaks, and other indices of exposure in
occupational epidemiology. Am. J. Ind. Med. 21: 25-33.
Chemical Manufacturers Association's Epidemiology Task Group. (1991) Guidelines for good epidemiology
practices for occupational and environmental epidemiological research. Washington, DC: Chemical
Manufacturers Association.
Cheng, Y.-S.; Moss, O. R. (1989) Inhalation exposure systems. In: McClellan, R. O.; Henderson, R. F., eds.
Concepts in inhalation toxicology. New York, NY: Hemisphere Publishing Corp.; pp. 19-62.
Cheng, Y. S.; Yamada, Y.; Yeh, H. C.; Swift, D. L. (1988) Diffusional deposition of ultrafine aerosol in a
human nasal cast. J. Aerosol. Sci. 19: 741-751.
Cheng, Y. S.; Hansen, G. K.; Su, Y. F.; Yeh, H. C.; Morgan, K. T. (1990) Deposition of ultrafine aerosols in
rat nasal molds. Toxicol. Appl. Pharmacol. 106: 222-233.
Clausen, J. L., ed. (1982) Pulmonary function testing guidelines and controversies: equipment, methods, and
normal values. Orlando, FL: Grune and Stratton, Inc.
5-5
-------
p. 25
Clegg, D. J. (1979) Toxicological basis of the ADI—present and future considerations. In: Frehse, H.;
Geissbuhler, H., eds. Pesticide residues: a contribution to their interpretation, relevance and legislation.
New York, NY: Pergamon Press; pp. 74-77.
Cleveland, W. S. (1985) The elements of graphing data. Pacific Grove, CA: Wadsworth and Brooks.
Code of Federal Regulations. (1991a) National primary and secondary ambient air quality standards. C. F. R.
40: §50.
Code of Federal Regulations. (1991b) Good laboratory practice standards [FIFRA]. C. F. R. 40: §160.
Code of Federal Regulations. (1991d) Health effects testing guidelines. C. F. R. 40: §798.
Code of Federal Regulations. (1991c) Good laboratory practice standards [TSCA]. C. F. R. 40: §792.
Conolly, R. B. (1990) Biologically-based models for toxic effects: tools for hypothesis testing and improving
health risk assessments. CUT Activities 10: 1-8.
Corley, R. A.; Reitz, R. H. (1990) Dose-route extrapolations in quantitative toxicology: physiologically based
pharmacokinetics and pharmacodynamics of chloroform. In: Gerrity, T. R.; Henry, C. J., eds. Principles
of route-to-route extrapolation for risk assessment, proceedings of the workshops; March and July; Hilton
Head, SC ans Durham, NC. New York, NY: Elsevier Science Publishing Co., Inc.; pp. 195-216.
Costa, D. L.; Tepper, J. S. (1988) Approaches to lung function assessment in small mammals. In: Gardner, D.;
Crapo, J.; Massaro, E., eds. Target organ toxicology series: toxicology of the lung. New York, NY:
Raven Press, Ltd.; pp. 147-174.
Costa, D. L.; Tepper, J. S.; Raub, J. A. (1992) Interpretations and limitations of pulmonary function testing in
small laboratory animals. In: Parent, R. A., ed. Comparative biology of the normal lung: v. 1, treatise
on pulmonary toxicology. Boca Raton, FL: CRC Press; pp. 367-399.
Crapo, J. D. (1987) Personal communication to A. Jarabek, Environmental Criteria and Assessment Office,
Cincinnati, OH, concerning interspecies cell population data and the confirmation of rudimentary
respiratory bronchioles by 3-D reconstruction in the rat acinus. Durham, NC: Duke University.
Crapo, R. O.; Morris, A. H.; Gardner, R. M. (1981) Reference spirometric values using techniques and
equipment that meet ATS recommendations. Am. Rev. Respir. Dis. 123: 659-664.
Crapo, J. D.; Young, S. L.; Fram, E. K.; Pinkerton, K. E.; Barry, B. E.; Crapo, R. O. (1983) Morphometric
characteristics of cells in the alveolar region of mammalian lungs. Am. Rev. Respir. Dis. 128: S42-S46.
Cropp, G. J. A.; Bernstein, I. L.; Boushey, H. A., Jr.; Hyde, R. W.; Rosenthal, R. R.; Spector, S. L.;
Townley, R. G. (1980) Guidelines for bronchial inhalation challenges with pharmacologic and antigenic
agents. ATS News (Spring): 11-19.
Crump, K. S. (1984) A new method for determining allowable daily intakes. Fundam. Appl. Toxicol.
4: 854-871.
Crump, K. S. (1986) [Letter to the editor]. Fundam. Appl. Toxicol. 6: 183-184.
Cullen, M. R. (1989) The role of clinical investigations in biological markers research. Environ. Res. 50: 1-10.
Dahl, A. R. (1990) Dose concepts for inhaled vapors and gases. Toxicol. Appl. Pharmacol. 103: 185-197.
5-6
-------
p. 26
Dahl, A. R.; Miller, S. C.; Petridou-Fischer, J. (1987) Carboxylesterases in the respiratory tracts of rabbits, rats
and Syrian hamsters. Toxicol. Lett. 36: 129-136.
Dahl, A. R.; Bond, J. A.; Petridou-Fischer, J.; Sabourin, P. J.; Whaley, S. J. (1988) Effects of the respiratory
tract on inhaled materials. Toxicol. Appl. Pharmacol. 93: 484-492.
Dahl, A. R.; Schlesinger, R. B.; Heck, H. D' A.; Medinsky, M. A.; Lucier, G. W. (1991a) Comparative
dosimetry of inhaled materials: differences among animal species and extrapolation to man. Fundam.
Appl. Toxicol. 16: 1-13.
Dahl, A. R.; Snipes, M. B.; Gerde, P. (1991b) Sites for uptake of inhaled vapors in beagle dogs. Toxicol. Appl.
Pharmacol. 109: 263-275.
Davis, J. M.; Svendsgaard, D. J. (1990) U-shaped dose-response curves: their occurrence and implications for
risk assessment. J. Toxicol. Environ. Health 30: 71-83.
Dayan, J.; Levenspiel, O. (1969) Dispersion in smooth pipes with adsorbing walls. Ind. Eng. Chem. Fundam.
8: 840-842.
De Rosa, C. T.; Stara, J. F.; Durkin, P. R. (1985) Ranking chemicals based on chronic toxicity data. Toxicol.
Ind. Health 1: 177-191.
Dearman, R. J.; Spence, L. M.; Kimber, I. (1992) Characterization of murine immune responses to allergenic
diisocyanates. Toxicol. Appl. Pharmacol. 112: 190-197.
Dedrick, R. L. (1973) Animal scale-up. J. Pharmacokinet. Biopharm. 1: 435-461.
Dedrick, R. L.; Bischoff, K. B. (1980) Species similarities in pharmacokinetics. Fed. Proc. 39: 54-59.
DeSesso, J. M. (1993) The relevance to humans of animal models for inhalation studies of cancer in the nose and
upper airways. Qual. Assur. (San Diego) 2: 213-231.
Domin, B. A.; Philpot, R. M. (1986) The effect of substrate on the expression of activity catalyzed by
cytochrome P-450: metabolism mediated by rabbit isozyme 6 in pulmonary microsomal and reconstituted
monooxygenase systems. Arch. Biochem. Biophys. 246: 128-142.
Dorato, M. A.; Carlson, K. H.; Copple, D. L. (1983) Pulmonary mechanics in conscious Fischer 344 rats:
multiple evaluations using nonsurgical techniques. Toxicol. Appl. Pharmacol. 68: 344-353.
Doull, J.; Klaassen, C. D.; Amdur, M. O., eds. (1980) Casarett and Doull's toxicology: the basic science of
poisons. 2nd ed. New York, NY: MacMillan Publishing Co., Inc.
Dourson, M. L. (1986) New approaches in the derivation of acceptable daily intake (ADI). Comments Toxicol.
1: 35-48.
Dourson, M. L.; Stara, J. F. (1983) Regulatory history and experimental support of uncertainty (safety) factors.
Regul. Toxicol. Pharmacol. 3: 224-238.
Dourson, M. L.; Hertzberg, R. C.; Hartung, R.; Blackburn, K. (1985) Novel methods for the estimation of
acceptable daily intake. Toxicol. Ind. Health 1: 23-33.
Dourson, M. L.; Hertzberg, R. C.; Stara, J. F. (1986) [Letter to the editor]. Fundam. Appl. Toxicol.
6: 182-183.
5-7
-------
p. 27
Dourson, M. L.; Swartout, J. C.; Stara, J. F. (1987) Excursions above the acceptable daily intake. Toxicologist
7: 184.
Dourson, M. L.; Knauf, L. A.; Swartout, J. C. (1992) On reference dose (RfD) and its underlying toxicity data
base. Toxicol. Ind. Health 8: 171-189.
Droz, P. O. (1985) The use of simulation models for setting BEIs for organic solvents. In: International
symposium on occupational exposure limits. Cincinnati, OH: American Conference of Governmental
Industrial Hygienists; pp. 339-350. (Annals of the American Conference of Governmental Industrial
Hygienists: v. 12).
Dunnick, J. K.; Graham, D. G.; Yang, R. S. H.; Haber, S. B.; Brown, H. R. (1989) Thirteen-week toxicity
study of n-hexane in B6C3F, mice after inhalation exposure. Toxicology 57: 163-172.
Eddy, D. M.; Hasselblad, V.; Shachter, R. (1992) Meta-analysis by the confidence profile method: the statistical
synthesis of evidence. Boston, MA: Academic Press, Inc.
Eisner, A. D.; Graham, R. C.; Martonen, T. B. (1990) Coupled mass and energy transport phenomena in
aerosol/vapor-laden gasses—I. Theory of the hygroscopic aerosol effects on temperature and relative
humidity patterns of inspired air. J. Aerosol. Sci. 21: 833-848.
Emmett, P. C.; Aitken, R. J.; Hannan, W. J. (1982) Measurements of the total and regional deposition of
inhaled particles in the human respiratory tract. J. Aerosol Sci. 13: 549-560.
Environ Corporation. (1985) Background document on the development and use of reference doses. Part I: data
needs and apportionment. Part II: considerations related to the development of protocols for toxicity
studies [draft]. Washington, DC: U.S. Environmental Protection Agency, Office of Solid Waste; EPA
report no. EPA/530/SW-86/048. Availble from: NTIS, Springfield, VA; PB87-107173/XAB.
Epler, G. R.; Saber, F. A.; Gaensler, E. A. (1980) Determination of severe impairment (disability) in interstitial
lung disease. Am. Rev. Respir. Dis. 121: 647-659.
Erdreich, L. S. (1988) Combining animal and human data: resolving conflicts, summarizing the evidence.
In: Gordis, L., ed. Epidemiology and health risk assessment. New York, NY: Oxford University Press;
pp. 197-207.
Erdreich, L. S.; Burnett, C. (1985) Improving the use of epidemiologic data in health risk assessment. Toxicol.
Ind. Health 1: 65-81.
Erdreich, L. S.; Sonich Mullin, C. (1984) Hypersusceptible subgroups of the population in multichemical risk
assessment. In: Stara, J. F.; Erdreich, L. S., eds. Approaches to risk assessment for multiple chemical
exposures. Washington, DC: U.S. Environmental Protection Agency, Office of Health and Environmental
Assessment, Environmental Criteria and Assessment Office; EPA report no. EPA-600/9-84-008;
pp. 189-218. Available from: NTIS, Springfield, VA; PB84-182369.
Fabro, S.; Shull, G.; Brown, N. A. (1982) The relative teratogenic index and teratogenic potency: proposed
components of developmental toxicity risk assessment. Teratog. Carcinog. Mutagen. 2: 61-76.
Farrar, D.; Allen, B.; Crump, K.; Shipp, A. (1989) Evaluation of uncertainty in input parameters to
pharmacokinetic models and the resulting uncertainty in output. Toxicol. Lett. 49: 371-385.
Federal Register. (1980) Guidelines and methodology used in the preparation of health effects assessment chapters
of the consent decree water criteria documents. F. R. (November 28) 45: 49347-49357.
5-8
-------
p. 28
Federal Register. (1988a) Intent to review guidelines for carcinogen risk assessment. F. R. (August 26)
53: 32656-32658.
Federal Register. (1988b) Proposed guidelines for assessing female reproductive risk. F. R. (June 30)
53: 24834-24847.
Federal Register. (1988c) Proposed guidelines for assessing male reproductive risk. F. R. (June 30)
53: 24850-24869.
Federal Register. (1991) Guidelines for developmental toxicity risk assessment. F. R. (December 5)
56: 63798-63826.
Federal Register. (1992a) Guidelines for exposure assessment. F. R. (May 29) 57: 22887-22938.
Federal Register. (1992b) Draft report: a cross-species scaling factor for carcinogen risk assessment based on
equivalence of mg/kg3/4/day. F. R. (June 5) 57: 24152-24173.
Federspiel, W. J.; Fredberg, J. J. (1988) Axial dispersion in respiratory bronchioles and alveolar ducts. J. Appl.
Physiol. 64: 2614-2621.
Ferris, B. G. (1978) Epidemiology standardization project: II. recommended respiratory disease questionnaires
for use with adults and children in epidemiological research. Am. Rev. Respir. Dis. 118(suppl.): 7-53.
Perron, G. A.; Hornik, S. (1984) Influence of different humidity profiles on the deposition probability of soluble
particles in the human lung. J. Aerosol. Sci. 15: 209-211.
Finney, D. J. (1978) Statistical method in biological assay. London, United Kingdom: Griffin.
Fiserova-Bergerova, V., ed. (1983) Modeling of inhalation exposure to vapors: uptake, distribution, and
elimination: volumes I and II. Boca Raton, FL: CRC Press, Inc.
Fiserova-Bergerova, V. (1990) Application of toxicokinetic models to establish biological exposure indicators.
Ann. Occup. Hyg. 34: 639-651.
Fiserova-Bergerova, V.; Diaz, M. L. (1986) Determination and prediction of tissue-gas partition coefficients. Int.
Arch. Occup. Environ. Health 58: 75-87.
Fiserova-Bergerova, V.; Vlach, J.; Cassady, J. C. (1980) Predictable "individual differences" in uptake and
excretion of gases and lipid soluble vapours simulation study. Br. J. Ind. Med. 37: 42-49.
Fiserova-Bergerova, V.; Tichy, M.; Di Carlo, F. J. (1984) Effects of biosolubility on pulmonary uptake and
disposition of gases and vapors of lipophilic chemicals. Drug Metab. Rev. 15: 1033-1070.
Fisher, J. W.; Allen, B. C. (1993) Evaluating the risk of liver cancer in humans exposed to trichloroethylene
using physiological models. Risk Anal. 13: 87-95.
Folinsbee, L. J. (1988) Human clinical inhalation exposures experimental design, methodology, and physiological
responses. In: Gardner, D. E.; Crapo, J. D.; Massaro, E. J., eds. Toxicology of the lung. New York,
NY: Raven Press; pp. 175-199.
Fox, A. J.; Collier, P. F. (1976) Low mortality rates in industrial cohort studies due to selection for work and
survival in the industry. Br. J. Prev. Soc. Med. 30: 225-230.
5-9
-------
p. 29
Friberg, L.; Nordberg, G. F.; Vouk, V. B., eds. (1979) Handbook on the toxicology of metals. Amsterdam, The
Netherlands: Elsevier/North-Holland Biomedical Press.
Gann, P. (1986) Use and misuse of existing data bases in environmental epidemiology: the case of air pollution.
In: Kopfler, F. C.; Craun, G. F., eds. Environmental epidemiology. Chelsea, MI: Lewis Publishers, Inc.
Gardner, D. E.; Kennedy, G. L., Jr. (1993) Methodologies and technology for animal inhalation toxicology
studies. In: Contributions to the scientific literature from the Haskell Laboratory for Toxicology and
Industrial Medicine. Newark, DE: E. I. duPont de Nemours and Co.; pp. 295-324.
Gardner, D. E.; Miller, F. J.; Blommer, E. J.; Coffin, D. L. (1979) Influence of exposure mode on the toxicity
of NO2. Environ. Health Perspect. 30: 23-29.
Gardner, R. M.; Clausen, J. L.; Epler, G.; Hankinson, J. L.; Permutt, S.; Plummer, A. L. (1986a) Pulmonary
function laboratory personnel qualifications. Am. Rev. Respir. Dis. 134: 623-624.
Gardner, R. M.; Clausen, J. L.; Crapo, R. O.; Epler, G. R.; Hankinson, J. L.; Johnson, R. L.; Plummer, A. L.
(1986b) Quality assurance in pulmonary function laboratories. Am. Rev. Respir. Dis. 134: 625-627.
Gardner, R. M.; Clausen, J. L.; Cotton, D. J.; Crapo, R. O.; Epler, G. R.; Hankinson, J. L.; Johnson, R. L.
(1986c) Computer guidelines for pulmonary laboratories. Am. Rev. Respir. Dis. 134: 628-629.
Gargas, M. L.; Burgess, R. J.; Voisard, D. E.; Cason, G. H.; Andersen, M. E. (1989) Partition coefficients of
low-molecular-weight volatile chemicals in various liquids and tissues. Toxicol. Appl. Pharmacol.
98: 87-99.
Gaylor, D. W. (1983) The use of safety factors for controlling risk. J. Toxicol. Environ. Health 11: 329-336.
Gaylor, D. W. (1985) The question of the existence of thresholds: extrapolation from high to low dose.
In: Flamm, W. G.; Lorentzen, R. J., eds. Mechanisms in toxicology of chemical carcinogens and
mutagens. Princeton, NJ: Princeton Scientific Publishing Co., Inc.; pp. 249-260. (Advances in modern
environmental toxicology: v. 12).
Gaylor, D. W.; Kodell, R. L. (1980) Linear interpolation algorithm for low dose risk assessment of toxic
substances. J. Environ. Pathol. Toxicol. 4: 305-312.
Geelhaar, A.; Weibel, E. R. (1971) Morphometric estimation of pulmonary diffusion capacity: III. the effect of
increased oxygen consumption in Japanese Waltzing mice. Respir. Physiol. 11: 354-366.
Gehr, P.; Mwangi, D. K.; Ammann, A.; Maloiy, G. M. O.; Taylor, C. R.; Weibel, E. R. (1981) Design of the
mammalian respiratory system. V. Scaling morphometric pulmonary diffusing capacity to body mass:
wild and domestic animals. Respir. Physiol. 44: 61-86.
Gerde, P.; Dahl, A. R. (1991) A model for the uptake of inhaled vapors in the nose of the dog during cyclic
breathing. Toxicol. Appl. Pharmacol. 109: 276-288.
Gerde, P.; Medinsky, M. A.; Bond, J. A. (1991) Particle-associated polycyclic aromatic hydrocarbons—
a reappraisal of their possible role in pulmonary carcinogenesis. Toxicol. Appl. Pharmacol. 108: 1-13.
Gerrity, T. R.; Henry, C. J., eds. (1990) Summary report of the workshops on principles of route-to-route
extrapolation for risk assessment. In: Principles of route-to-route extrapolation for risk assessment,
proceedings of the workshops; March and July; Hilton Head, SC and Durham, NC. New York, NY:
Elsevier Science Publishing Co., Inc.; pp. 1-12.
5-10
-------
p. 30
Goldstein, B. D. (1983) Toxic substances in the atmospheric environment: a critical review. J. Air Pollut.
Control Assoc. 33: 454-467.
Graham, J. A.; Miller, F. J.; Davies, D. W.; Hiteshew, M. E.; Walsh, L. C., III. (1985) Inhalation studies of
Mt. St. Helens volcanic ash in animals: I. introduction and exposure system. Environ. Res. 37: 61-71.
Graham, J. A.; Gardner, D. E.; Blommer, E. J.; House, D. E.; Menache, M. G.; Miller, F. J. (1987) Influence
of exposure patterns of nitrogen dioxide and modifications by ozone on susceptibility to bacterial
infectious disease in mice. J. Toxicol. Environ. Health 21: 113-125.
Gram, T. E.; Okine, L. K.; Gram, R. A. (1986) The metabolism of xenobiotics by certain extrahepatic organs
and its relation to toxicity. Ann. Rev. Pharmacol. Toxicol. 26: 259-291.
Griffis, L. C.; Wolff, R. K.; Beethe, R. L.; Hobbs, C. H.; McClellan, R. O. (1981) Evaluation of a multitiered
inhalation exposure chamber. Fundam. Appl. Toxicol. 1: 8-12.
Griffis, L. C.; Wolff, R. K.; Henderson, R. F.; Griffith, W. C.; Mokler, B. V.; McClellan, R. O. (1983)
Clearance of diesel soot particles from rat lung after a subchronic diesel exhaust exposure. Fundam.
Appl. Toxicol. 3: 99-103.
Griffith, J.; Duncan, R. C.; Hulka, B. S. (1989) Biochemical and biological markers: implications for
epidemiologic studies. Arch. Environ. Health 44: 375-381.
Gross, E. A.; Morgan, K. T. (1992) Architecture of nasal passages and larynx. In: Parent, R. A., ed.
Comparative biology of the normal lung: v. I, treatise on pulmonary toxicology. Boca Raton, FL: CRC
Press, Inc.; pp. 7-25.
Gross, E. A.; Swenberg, J. A.; Fields, S.; Popp, J. A. (1982) Comparative morphometry of the nasal cavity in
rats and mice. J. Anat. 135: 83-88.
Guilmette, R. A.; Wicks, J. D.; Wolff, R. K. (1989) Morphometry of human nasal airways in vivo using
magnetic resonance imaging. J. Aerosol. Med. 2: 365-377.
Guth, D. J.; Jarabek, A. M.; Wymer, L.; Hertzberg, R. C. (1991) Evaluation of risk assessment methods for
short-term inhalation exposure. Presented at: 84th annual meeting and exhibition; June. Pittsburgh, PA:
Air and Waste Management Association; paper no. 91-173.2.
Guth, D. J.; Hertzberg, R. C.; Jarabek, A. M. (1993) Exposure-response analysis: modeling severity against
concentration and duration. In: Beck, B. D.; Connolly, R. B.; Dourson, M. L.; Guth, D.; Hattis, D.;
Kimmel, C.; Lewis, S. C. Improvements in quantitative noncancer risk assessment: symposium overview,
Society of Toxicology meeting; February 1992; Seattle, WA. Fundam. Appl. Toxicol. 20: 9-12.
Habicht, F. H., n. (1992) Guidance on risk characterization for risk managers and risk assesors [memorandum to
EPA assistant administrators and regional administrators]. Washington, DC: U.S. Environmental
Protection Agency, Office of the Administrator; February 26.
Hackney, J. D.; Linn, W. S. (1979) Koch's postulates updated: a potentially useful application to laboratory
research and policy analysis in environmental toxicology. Am. Rev. Respir. Dis. 119: 849-852.
Hackney, J. D.; Linn, W. S. (1983) Controlled clinical studies of air pollutant exposure: evaluating scientific
information in relation to air quality standards. Environ. Health Perspect. 52: 187-191.
Hahn, I.; Scherer, P. W.; Mozell, M. M. (1993) Velocity profiles measured for airflow through a large-scale
model of the human nasal cavity. J. Appl. Physiol. 75: 2273-2287.
5-11
-------
p. 31
Hakkinen, P. J.; Witschi, H. P. (1985) Animal models. In: Witschi, H. P.; Brain, J. D., eds. Toxicology of
inhaled materials: general principles of inhalation toxicology. New York, NY: Springer-Verlag;
pp. 95-114. (Handbook of experimental pharmacology: v. 75).
Hanna, L. M.; Scherer, P. W. (1986) Measurement of local mass transfer coefficients in a cast model of the
human upper respiratory tract. J. Biomech. Eng. 108: 12-18.
Hanna, L. M.; Frank, R.; Scherer, P. W. (1989) Absorption of soluble gases and vapors in the respiratory
system. In: Chang, H. K.; Paiva, M., eds. Respiratory physiology: an analytical approach. New York,
NY: Marcel Dekker, Inc.; pp. 277-316. (Lenfant, C., ed. Lung biology in health and disease: v. 40).
Harkema, J. R. (1991) Comparative aspects of nasal airway anatomy: relevance to inhalation toxicology. Toxicol.
Pathol. 19: 321-336.
Hartung, R. (1986) Ranking the severity of toxic effects. In: Hemphill, D. D., ed. Trace substances in
environmental health—XX: [proceedings of University of Missouri's 20th annual conference]; June;
Columbia, MO. Columbia, MO: University of Missouri-Columbia; pp. 204-211.
Hasselblad, V.; Jarabek, A. M. (1994) Dose-response analysis of toxic chemicals. In: Berry, D. A.; Stangl,
D. K., eds. Bayesian biostatistics. New York, NY: Marcel Dekker: in press.
Hatch, T.; Choate, S. P. (1929) Statistical description of the size properties of non-uniform particulate
substances. J. Franklin Inst. 207: 369-387.
Hatch, T. F.; Gross, P. (1964) Pulmonary deposition and retention of inhaled aerosols. New York, NY:
Academic Press, Inc.
Hattis, D. B. (1986) The promise of molecular epidemiology for quantitative risk assessment. Risk Anal.
6: 181-193.
Hattis, D. (1991) Use of biological markers and pharmacokinetics in human health risk assessment. Environ.
Health Perspect. 90: 229-238.
Hattis, D.; Erdreich, L.; Ballew, M. (1987) Human variability in susceptibility to toxic chemicals—a preliminary
analysis of pharmacokinetic data from normal volunteers. Risk Anal. 7: 415-426.
Hattis, D.; White, P.; Marmorstein, L.; Koch, P. (1990) Uncertainties in pharmacokinetic modeling for
perchloroethylene. I. Comparison of model structure, parameters, and predictions for low-dose
metabolism rates for models derived by different authors. Risk Anal. 10: 449-458.
Hertzberg, R. C. (1989) Fitting a model to categorical response data with application to species extrapolation of
toxicity. In: Proceedings of the 26th Hanford life sciences symposium,"Modeling for scaling to man";
October 1987. Richland, WA: Battelle Pacific Northwest Laboratories. Health Phys. 57 (suppl.):
405-409.
Heyder, J.; Rudolf, G. (1977) Deposition of aerosol particles in the human nose. In: Walton, W. H., ed. Inhaled
particles IV, part 1: proceedings of an international symposium organized by the British Occupational
Hygiene Society; September 1975; Edinburgh, United Kingdom. Pergamon Press; pp. 107-126.
Heyder, J.; Armbruster, L.; Gebhart, J.; Grein, E.; Stahlhofen, W. (1975) Total deposition of aerosol particles
in the human respiratory tract for nose and mouth breathing. J. Aerosol Sci. 6: 311-328.
Heyder, J.; Gebhart, J.; Rudolf, G.; Schiller, C. F.; Stahlhofen, W. (1986) Deposition of particles in the human
respiratory tract in the size range 0.005-15 fan. J. Aerosol Sci. 17: 811-825.
5-12
-------
p. 32
Heywood, R. (1981) Target organ toxicity. Toxicol. Lett. 8: 349-358.
Heywood, R. (1983) Target organ toxicity E. Toxicol. Lett. 18: 83-88.
Hinds, W. C. (1982) Aerosol technology. New York, NY: John Wiley and Sons.
Hoftnann, W. (1982) The effect of polydispersivness of natural radioactive aerosols on tracheobronchial
deposition. Radiat. Prot. Dosim. 3: 97-101.
Hofmann, W.; Koblinger, L. (1989) The effect of polydisperity of radioactive aerosols on the activity
distribution in the human lung. J. Aerosol Sci. 20: 1313-1316.
Holland, W. W.; Bennett, A. E.; Cameron, I. R.; Florey, C. du V.; Leeder, S. R.; Schilling, R. S. F.; Swan,
A. V.; Waller, R. E. (1979) Health effects of particulate pollution: reappraising the evidence.
Am. J. Epidemiol. 110: 525-659.
Huang, Y.; Chang, L.; Miller, F.; Graham, J.; Ospital, J.; Crapo, J. (1988) Lung injury caused by ambient
levels of oxidant air pollutants: extrapolation from animals to man. J. Aerosol Med. 1: 180-183.
Hulka, B. S.; Wilcosky, T. (1988) Biological markers in epidemiologic research. Arch. Environ. Health
43: 83-89.
IRIS, Integrated Risk Information System [database]. (1990) [Printout of reference concentration for chronic
inhalation exposure (RfC) for n-hexane as verified 4/19/90]. Cincinnati, OH: U.S. Environmental
Protection Agency, Office of Health and Environmental Assessment, Environmental Criteria and
Assessment Office. Available online from: TOXNET, National Library of Medicine, Rockville, MD.
IRIS, Integrated Risk Information System [database]. (1992) [Printout of reference concentration for chronic
inhalation exposure (RfC) for diesel engine emissions as verified 6/25/92]. Cincinnati, OH: U.S.
Environmental Protection Agency, Office of Health and Environmental Assessment, Environmental
Criteria and Assessment Office. Available online from: TOXNET, National Library of Medicine,
Rockville, MD.
Interagency Regulatory Liaison Group. (1981) Guidelines for documentation of epidemiological studies.
Am. J. Epidemiol. 114: 609-613.
International Commission on Radiological Protection. (1993) Gases and vapours. In: Human respiratory tract
model for radiological protection: a report of Committee 2 of the International Commission on
Radiological Protection. Sutton, Surrey, United Kingdom: International Commission on Radiological
Protection; pp. 57-66; April 15.
International Standards Organization. (1981) Report of ad hoc working group to technical committee 146-air
quality: recommendations on size definitions for particle sampling. Am. Ind. Hyg. Assoc. J.
42(5): A64-A68.
Iregren, A. (1990) Psychological test performance in foundry workers exposed to low levels of manganese.
Neurotoxicol. Teratol. 12: 673-675.
James, R. C. (1985) Risk assessment. In: Williams, P. L.; Burson, J. L., eds. Industrial toxicology: safety and
health applications in the workplace. New York, NY: Van Nostrand Reinhold Company; pp. 369-398.
5-13
-------
p. 33
Jarabek, A. M. (1991) Meeting to identify ventilation rate and respiratory surface area default values for use in
the inhalation RfC methodology [memorandum to attendees for August 14, 1991 meeting]. Research
Triangle Park, NC: U.S. Environmental Protection Agency, Environmental Criteria and Assessment
Office; August 7.
Jarabek, A. M. (1994) Inhalation RfC methodology: dosimetric adjustments and dose-response estimation of
noncancer toxicity in the upper respiratory tract. Inhal. Toxicol. 6(suppl.): 301-325.
Jarabek, A. M.; Farland, W. H. (1990) The U.S. Environmental Protection Agency's risk assessment guidelines.
Toxicol. Ind. Health 6: 199-216.
Jarabek, A. M.; Hasselblad, V. (1991) Inhalation reference concentration methodology: impact of dosimetric
adjustments and future directions using the confidence profile method. Presented at: 84th annual meeting
and exhibition of the Air and Waste Management Association; June; Vancouver, BC, Canada. Pittsburgh,
PA: Air and Waste Management Association; paper no. 91-173.3.
Jarabek, A. M.; Segal, S. A. (1994) Noncancer toxicity of inhaled air toxics: available approaches for risk
assessment and risk management. In: Patrick, D. R., ed. Toxic air pollution handbook. New York, NY:
Van Nostrand Reinhold; pp. 100-130.
Jarabek, A. M.; Me"nache, M. G.; Overton, J. H., Jr.; Dourson, M. L.; Miller, F. J. (1989) Inhalation reference
dose (RfD/): an application of interspecies dosimetry modeling for risk assessment of insoluble particles.
In: Proceedings of the 26th Hanford life sciences symposium, "modeling for scaling to man;" October
1987; Richland, WA. Health Phys. 57(supp. 1): 177-183.
Jarabek, A. M.; Menache, M. G.; Overton, J. H., Jr.; Dourson, M. L.; Miller, F. J. (1990) The U.S.
Environmental Protection Agency's inhalation RFD methodology: risk assessment for air toxics. Toxicol.
Ind. Health 6: 279-301.
Jeffery, P. K. (1983) Morphologic features of airway surface epithelial cells and glands. Am. Rev. Respir. Dis.
128: S14-S20.
Jepson, G. W.; Hoover, D. K.; Black, R. K.; McCafferty, J. D.; Mahle, D. A.; Gearhart, J. M. (1994)
A partition coefficient determination method for nonvolatile chemicals in biological tissues. Fundam.
Appl. Toxicol. 22: 519-524.
*
Johanson, W. G., Jr.; Gould, K. G., Jr. (1977) Lung defense mechanisms. Basics Respir. Dis. 6(2): 1-6.
Kane, L. E.; Barrow, C. S.; Alarie, Y. (1979) A short-term test to predict acceptable levels of exposure to
airborne sensory irritants. Am. Ind. Hyg. Assoc. J. 40: 207-229.
Karol, M. H. (1994) Assays to evaluate pulmonary hypersensitivity. In: Burleson, G. R.; Dean, J. H.; Munson,
A. E., eds. Modern methods in immunotoxicology. New York, NY: Wiley-Liss.
Karol, M. H.; Stadler, J.; Magreni, C. M. (1985) Immunotoxicologic evaluation of the respiratory system:
animal models for immediate- and delayed-onset pulmonary hypersensitivity. Fundam. Appl. Toxicol.
5: 459-472.
Kehrer, J. P.; Kacew, S. (1985) Systemically applied chemicals that damage lung tissue. Toxicology
35: 251-293.
Kenoyer, J. L.; Phalen, R. F.; Davis, J. R. (1981) Particle clearance from the respiratory tract as a test of
toxicity: effect of ozone on short and long term clearance. Exp. Lung Res. 2: 111-120.
5-14
-------
p. 34
Khoury, M. J.; Newill, C. A.; Chase, G. A. (1985) Epidemiologic evaluation of screening for risk factors:
application to genetic screening. Am. J. Public Health 75: 1204-1208.
Kimbell, J. S.; Gross, E. A.; Joyner, D. R.; Godo, M. N.; Morgan, K. T. (1993) Application of computational
fluid dynamics to regional dosimetry of inhaled chemicals in the upper respiratory tract of the rat.
Toxicol. Appl. Pharmacol. 121: 253-263.
Kimmel, C. A.; Gaylor, D. W. (1988) Issues in qualitative and quantitative risk analysis for developmental
toxicology. Risk Anal. 8: 15-20.
Klaassen, C. D. (1986) Principles of toxicology. In: Klaassen, C. D.; Amdur, M. O.; Doull, J., eds. Casarett
and Doull's toxicology: the basic science of poisons. 3rd ed. New York, NY: MacMillan Publishing Co.,
Inc.; pp. 11-32.
Kleinman, M. T. (1984) Sulfur dioxide and exercise: relationships between response and absorption in upper
airways. J. Air Pollut. Control Assoc. 34: 32-37.
Kliment, V. (1973) Similarity and dimensional analysis, evaluation of aerosol deposition in the lungs of
laboratory animals and man. Folia Morphol. (Prague) 21: 59-64.
Knudson, R. J.; Burrows, B.; Lebowitz, M. D. (1976) The maximal expiratory flow-volume curve: its use in the
detection of ventilatory abnormalities in a population study. Am. Rev. Respir. Dis. 114: 871-879.
Kokoski, C. J. (1976) [Written testimony of Charles J. Kokoski, docket no. 76N-0070]. Washington, DC: U.S.
Department of Health, Education and Welfare, Food and Drug Administration.
Krewski, D.; Brown, C.; Murdoch, D. (1984) Determining "safe" levels of exposure: safety factors or
mathematical models? Fundam. Appl. Toxicol. 4: S383-S394.
Kurzel, R. B.; Cetrulo, C. L. (1981) The effect of environmental pollutants on human reproduction, including
birth defects. Environ. Sci. Technol. 15: 626-640.
Kuykendall, J. R.; Taylor, M. L.; Bogdanffy, M. S. (1993) Cytotoxicity and DNA-protein crosslink formation
in rat nasal tissues exposed to vinyl acetate are carboxylesterase-mediated. Toxicol. Appl. Pharmacol.
123: 283-292.
Last, J. A. (1983) Biochemical alterations of lung structure as predictors of chronic lung disease. Environ. Health
Perspect. 52: 159-163.
Lebowitz, M. D. (1983) Utilization of data from human population studies for setting air quality standards:
evaluation of important issues. Environ. Health Perspect. 52: 193-205.
Lechner, A. J. (1978) The scaling of maximal oxygen consumption and pulmonary dimensions in small
mammals. Respir. Physiol. 34: 29-44.
Leung, H.-W. (1992) Use of physiologically based pharmacokinetic models to establish biological exposure
indexes. Am. Ind. Hyg. Assoc. J. 53: 369-374.
Leung, H.-W.; Paustenbach, D. J. (1988) Application of pharmacokinetics to derive biological exposure indexes
from threshold limit values. Am. Ind. Hyg. Assoc. J. 49: 445-450.
Lewis, S. C.; Lynch, J. R.; Nikiforov, A. I. (1990) A new approach to deriving community exposure guidelines
from "no-observed-adverse-effect levels." Reg. Toxicol. Pharmacol. 11: 314-330.
5-15
-------
p. 35
Lippmann, M. (1970) Deposition and clearance of inhaled particles in the human nose. Ann. Otol. Rhinol.
Laryngol. 79: 519-528.
Lippmann, M. (1977) Regional deposition of particles in the human respiratory tract. In: Lee, D. H. K.; Falk,
H. L.; Murphy, S. D.; Geiger, S. R., eds. Handbook of physiology, section 9: reactions to physical
agents. Bethesda, MD: American Physiological Society; pp. 213-232.
Lippmann, M. (1980) Aerosol exposure methods. In: Willeke, K., ed. Generation of aerosols and facilities for
exposure experiments. Ann Arbor, MI: Ann Arbor Science Publishers, Inc.; pp. 443-458.
Lippmann, M.; Albert, R. E. (1969) The effect of particle size on the regional deposition of inhaled aerosols in
the human respiratory tract. Am. Ind. Hyg. Assoc. J. 00: 257-275.
Lippmann, M.; Schlesinger, R. B. (1984) Interspecies comparisons of particle deposition and mucociliary
clearance in tracheobronchial airways. J. Toxicol. Environ. Health 13: 441-469.
Litterst, C. L.; Mimnaugh, E. G.; Reagan, R. L.; Gram, T. E. (1975) Comparison of in vitro drug metabolism
by lung, liver, and kidney of several common laboratory species. Drug Metab. Dispos. 3: 259-265.
Lou, S.-R. (1993) Modeling of gas absorption: upper airway scrubbing [submitted dissertation]. Baltimore, MD:
The Johns Hopkins University.
Lowry, L. K. (1986) Biological exposure index as a complement to the TLV. J. Occup. Med. 28: 578-582.
Lu, F. C. (1983) Toxicological evaluations of carcinogens and noncarcinogens: pros and cons of different
approaches. Regul. Toxicol. Phannacol. 3: 121-132.
Lu, F. C. (1985a) Basic toxicology: fundamentals, target organs, and risk assessment. New York, NY:
Hemisphere Publishing Corporation.
Lu, F. C. (1985b) Safety assessments of chemicals with thresholded effects. Regul. Toxicol. Pharmacol.
5: 460-464.
Lyman, W. J.; Reehl, W. F.; Rosenblatt, D. H. (1990) Handbook of chemical property estimation methods:
environmental behavior of organic compounds. Washington, DC: American Chemical Society.
Marin, M. G. (1986) Pharmacology of airway secretion. Pharmacol. Rev. 38: 273-289.
Marple, V. A.; Rubow, K. L. (1980) Aerosol generation concepts and parameters. In: Willeke, K.,
ed. Generation of aerosols and facilities for exposure experiments. Ann Arbor, MI: Ann Arbor Science
Publishers, Inc.; pp. 3-29.
Martonen, T. B. (1982) Analytical model of hygroscopic particle behavior in human airways. Bull. Math. Biol.
44: 425-442.
Martonen, T. B.; Miller, F. J. (1986) Dosimetry and species sensitivity: key factors in hazard evaluation using
animal exposure data. J. Aerosol Sci. 17: 316-319.
Martonen, T. B.; Patel, M. (1981) Modeling the dose distribution of H2SO4 aerosols in the human
tracheobronchial tree. Am. Ind. Hyg. Assoc. J. 42: 453-460.
Martonen, T. B.; Barnett, A. E.; Miller, F. J. (1985) Ambient sulfate aerosol deposition in man: modeling the
influence of hygroscopicity. Environ. Health Perspect. 63: 11-24.
5-16
-------
p. 36
Massaro, E. J.; Grose, E. C.; Hatch, G. E.; Slade, R. (1988) Antioxidant defense mechanisms of the lung.
In: Gardner, D. E.; Crapo, J. D.; Massaro, E. J., eds. Toxicology of the lung. New York, NY: Raven
Press; pp. 201-218. (Target organ toxicology series).
Mauderly, J. L. (1989) Effect of inhaled toxicants on pulmonary function. In: McClellan, R. O.; Henderson,
R. F., eds. Concepts in inhalation toxicology. New York, NY: Hemisphere Publishing; pp. 347-401.
Mauderly, J. L.; Kritchevsky, J. (1979) Respiration of unsedated Fischer 344 rats and the effect of confinement
in exposure tubes. In: Henderson, R. F.; Diel, J. H.; Martinez, B. S., eds. Inhalation Toxicology
Research Institute annual report 1978-1979. Albuquerque, NM: U.S. Department of Energy, Lovelace
Biomedical and Environmental Research Institute; report no. LMF-69; pp. 475-478. Available from:
NTIS, Springfield, VA; LMF-69.
Mauderly, J. L.; Tesarek, J. E.; Sifford, L. J.; Sifford, L. J. (1979) Respiratory measurements of unsedated
small laboratory mammals using nonrebreathing valves. Lab. Anim. Sci. 29: 323-329.
McCauley, P. T.; Bull, R. J. (1980) Experimental approaches to evaluating the role of environmental factors in
the development of cardiovascular disease. J. Environ. Pathol. Toxicol. 4: 27-50.
McJilton, C.; Thielke, J.; Frank, R. (1972) Ozone uptake model for the respiratory system. Presented at:
American Industrical Hygiene Association conference; May; San Francisco, CA. Seattle, WA: University
of Washington, Departments of Civil Engineering and Environmental Health.
McKay, R. T. (1986) Bronchoprovocation challenge testing in occupational airways disorders. Semin. Respir.
Med. 7: 297-306.
McKay, R. T.; Lockey, J. E. (1991) Pulmonary function testing: guidelines for medical surveillance and
epidemiological studies. Occup. Med. 6: 43-57.
McKenna, M. J. (1982) Production and characterization of exposure atmospheres: vapors and gases. Presented at:
Third AIHA/Lovelace inhalation toxicology workshop; September; Albuquerque, NM.
McNamara, B. P. (1976) Concepts in health evaluation of commercial and industrial chemicals. In: Mehlman,
M. A.; Shapiro, R. E.; Blumenthal, H., eds. New concepts in safety evaluation. Washington, DC:
Hemisphere Publishing Corporation; pp. 61-140. (Advances in modern toxicology: volume 1, part 1).
Means, B. (1989) Risk assessment guidance for Superfund. Volume 1. Human health evaluation manual. Part A.
Washington, DC: U.S. Environmental Protection Agency, Office of Solid Waste and Emergency
Response; EPA report no. EPA/540/1-89/002. Available from: NTIS, Springfield, VA;
PB90-155581/XAB.
Medical Research Council, Committee on the Aetiology of Chronic Bronchitis. (1960) Standardized questionaries
on respiratory symptoms. Br. Med. J. 2: 1665.
Me"nache, M. G.; Raabe, O. G.; Miller, F. J. (submitted) An empirical dosimetry model of aerodynamic particle
deposition in the rat respiratory tract. Inhalation Toxicol.: submitted.
Me'nache, M. G.; Miller, F. J.; Raabe, O. G. (1995) Particle inhalability curves for humans and small laboratory
animals. Ann. Occup. Hyg. 39: 317-328.
Meneely, G. R.; Renzetti, A. D., Jr.; Steele, J. D.; Wyatt, J. P.; Harris, H. W. (1962) Chronic bronchitis,
asthma, and pulmonary emphysema: a statement by the Committee on Diagnostic Standards for
Nontuberculous Respiratory Diseases. Am. Rev. Respir. Dis. 85: 762-768.
5-17
-------
p. 37
Menzel, D. B.; Amdur, M. O. (1986) Toxic responses of the respiratory system. In: Klaassen, C. D.; Amdur,
M. O.; Doull, J., eds. Casarett and Doull's toxicology: the basic science of poisons. 3rd ed. New York,
NY: Macmillan Publishing Company; pp. 330-358, 428, 430.
Mercer, R. R.; Crapo, J. D. (1987) Three-dimensional reconstruction of the rat acinus. J. Appl. Physiol.
63: 785-794.
Mercer, T. T.; Tillery, M. I.; Chow, H. Y. (1968) Operation characteristics of some compressed air nebulizers.
Am. Ind. Hyg. Assoc. 29: 66.
Mercer, R. R.; Russell, M. L.; Roggli, V. L.; Crapo, J. D. (1994a) Cell number and distribution in human and
rat airways. Am. J. Respir. Cell Mol. Biol. 10: 613-624.
Mercer, R. R.; Russell, M. L.; Crapo, J. D. (1994b) Alveolar septal structure in different species. J. Appl.
Physiol. 77(3): in press.
Mery, S.; Gross, E. A.; Joyner, D. R.; Godo, M.; Morgan, K. T. (1994) Nasal diagrams: a tool for recording
the distribution of nasal lesions in rats and mice. Toxicol. Pathol. 22(4): in press.
Miller, F. J.; Gardner, D. E.; Graham, J. A.; Lee, R. E., Jr.; Wilson, W. E.; Bachmann, J. D. (1979) Size
considerations for establishing a standard for inhalable particles. J. Air Pollut. Control Assoc.
29: 610-615.
Miller, F. J.; Graham, J. A.; Gardner, D. E. (1983a) The changing role of animal toxicology in support of
regulatory decisions. Environ. Health Perspect. 52: 169-176.
Miller, F. J.; Graham, J. A.; Overton, J. H., Jr. (1983b) General considerations for developing pulmonary
extrapolation models. In: Proceedings of the conference on environmental toxicology (13th); November
1982; Dayton, OH. Wright- Patterson AFB, OH: Air Force Aerospace Medical Research Laboratory;
report no. AFAMRL-TR-82-101; pp. 178-208. Available from: NTIS, Springfield, VA; AD-A134 150.
Miller, F. J.; Overton, J. H., Jr.; Jaskot, R. H.; Menzel, D. B. (1985) A model of the regional uptake of
gaseous pollutants in the lung: I. the sensitivity of the uptake of ozone in the human lung to lower
respiratory tract secretions and exercise. Toxicol. Appl. Phannacol. 79: 11-27.
Miller, F. J.; Graham, J. A.; Raub, J. A.; Illing, J. W.; Menache, M. G.; House, D. E.; Gardner, D. E.
(1987a) Evaluating the toxicity of urban patterns of oxidant gases. II. Effects in mice from chronic
exposure to nitrogen dioxide. J. Toxicol. Environ. Health 21: 99-112.
Miller, F. J.; Overton, J. H., Jr.; Smolko, E. D.; Graham, R. C.; Menzel, D. B. (1987b) Hazard assessment
using an integrated physiologically based dosimetry modeling approach: ozone. In: Pharmacokinetics in
risk assessment, drinking water and health, v. 8. Washington, DC: National Academy Press;
pp. 353-368.
Miller, F. J.; Martonen, T. B.; Menache, M. G.; Graham, R. C.; Spektor, D. M.; Lippmann, M. (1988)
Influence of breathing mode and activity level on the regional deposition of inhaled particles and
implications for regulatory standards. In: Dodgson, J.; McCallum, R. I.; Bailey, M. R.; Fisher, D. R.,
eds. Inhaled particles VI: proceedings of an international symposium and workshop on lung dosimetry;
September 1985; Cambridge, United Kingdom. Ann. Occup. Hyg. 32 (suppl. 1): 3-10.
Minchin, R. F.; Boyd, M. R. (1983) Localization of metabolic activation and deactivation systems in the lung:
significance to the pulmonary toxicity of xenobiotics. Annu. Rev. Phannacol. Toxicol. 23: 217-238.
5-18
-------
p. 38
Moller, D. R.; Baughman, R.; Murlas, C.; Brooks, S. M. (1986) New directions in occupational asthma caused
by small molecular weight compounds. Semin. Respir. Med. 7: 225-239.
Monks, T. J.; Lau, S. S. (1989) Sulphur conjugate-mediated toxicity. Rev. Biochem. Toxicol. 10: 41-90.
Monson, R. R. (1986) Observations on the healthy worker effect. J. Occup. Med. 28: 425-433.
Monticello, T. M.; Morgan, K. T.; Everitt, J. L; Popp, J. A. (1989) Effects of formaldehyde gas on the
respiratory tract of rhesus monkeys: pathology and cell proliferation. Am. J. Pathol. 134: 515-527.
Morgan, K. T. (1991) Approaches to the identification and recording of nasal lesions in toxicology studies.
Toxicol. Pathol. 19: 337-351.
Morgan, M. S.; Frank, R. (1977) Uptake of pollutant gases by the respiratory system. In: Brain, J. D.; Proctor,
D. F.; Reid, L. M., eds. Respiratory defense mechanisms: part I. New York, NY: Marcel Dekker, Inc.;
pp. 157-189.
Morgan, K. T.; Jiang, X.-Z.; Patterson, D. L.; Gross, E. A. (1984) The nasal mucociliary apparatus: correlation
of structure and function in the rat. Am. Rev. Respir. Dis. 130: 275-281.
Morgan, K. T.; Jiang, X.-Z.; Starr, T. B.; Kerns, W. D. (1986) More precise localization of nasal tumors
associated with chronic exposure of F-344 rats to formaldehyde gas. Toxicol. Appl. Pharmacol.
82: 264-271.
Morgan, K. T.; Kimbell, J. S.; Monticello, T. M.; Patra, A. L.; Fleishman, A. (1991) Studies of inspiratory
airflow patterns in the nasal passages of the F344 rat and rhesus monkey using nasal molds: relevance to
formaldehyde toxicity. Toxicol. Appl. Pharmacol. 110: 223-240.
Morris, J. B. (1989) Deposition and absorption of inhaled vapors in the nasal cavity. In: Feron; Bosland, eds.
Nasal carcinogeneis in rodents: relevence to human health risk; pp. 113-118.
Morris, J. B. (1990) First-pass metabolism of inspired ethyl acetate in the upper respiratory tracts of the F344 rat
and Syrian hamster. Toxicol. Appl. Pharmacol. 102: 331-345.
Morris, J. B.; Blanchard, K. T. (1992) Upper respiratory tract deposition of inspired acetaldehyde. Toxicol.
Appl. Pharmacol. 114: 140-146.
Morris, J. B.; Cavanagh, D. G. (1986) Deposition of ethanol and acetone vapors in the upper respiratory tract of
the rat. Fundam. Appl. Toxicol. 6: 78-88.
Morris, J. B.; Cavanagh, D. G. (1987) Metabolism and deposition of propanol and acetone vapors in the upper
respiratory tract of the hamster. Fundam. Appl. Toxicol. 9: 34-40.
Morris, J. B.; Smith, F. A. (1982) Regional deposition and absorption of inhaled hydrogen fluoride in the rat.
Toxicol. Appl. Pharmacol. 62: 81-89.
Morris, J. B.; Clay, R. J.; Cavanagh, D. G. (1986) Species differences in upper respiratory tract deposition of
acetone and ethanol vapors. Fundam. Appl. Toxicol. 7: 671-680.
Morris, J. B.; Clay, R. J.; Trela, B. A.; Bogdanffy, M. S. (1991) Deposition of dibasic esters in the upper
respiratory tract of the male and female Sprague-Dawley rat. Toxicol. Appl. Pharmacol. 108: 538-546.
Morrow, P. E. (1988) Possible mechanisms to explain dust overloading of the lungs. Fundam. Appl. Toxicol.
10: 369-384.
5-19
-------
p. 39
Morrow, P. E. (1992) Dust overloading of the lungs: update and appraisal. Toxicol. Appl. Pharmacol.
113: 1-12.
Moss, O. R.; Cheng, Y.-S. (1989) Generation and characterization of test atmospheres: particles. In: McClellan,
R. O.; Henderson, R. F., eds. Concepts in inhalation toxicology; New York, NY: Hemisphere
Publishing Corp.; pp. 85-120.
Muller, K. E.; Barton, C. N.; Benignus, V. A. (1984) Recommendations for appropriate statistical practice in
toxicologic experiments. Neurotoxicology 5: 113-125.
Mustafa, M. G.; Lee, S. D. (1976) Pulmonary biochemical alterations resulting from ozone exposure. Ann.
Occup. Hyg. 19: 17-26.
Nadel, J. A.; Widdicombe, J. H.; Peatfield, A. C. (1986) Regulation of airway secretions, in transport and water
movement. In: Handbook of physiology; pp. 414-445.
National Center for Health Statistics. (1970) Natality statistics analysis, United States, 1965-1967. Washington,
DC: Public Health Service, National Center for Health Statistics; report no. PHS No. 1000. (Vital and
health statistics series 21, no. 19).
National Center for Health Statistics. (1975) Selected vital and health statistics in poverty and nonpoverty areas of
19 large cities, United States, 1969-1971. Washington, DC: Public Health Service, National Center for
Health Statistics. (Vital and health statistics series 21, no. 26).
National Institute for Occupational Safety and Health. (1986) NIOSH cross-sectional and reproductive medical
industry-wide studies questionnaire. Cincinnati, OH: National Institute for Occupational Safety and
Health; report no. OMB 0920-0037.
National Research Council. (1977) Drinking water and health. Washington, DC: National Academy of Sciences;
pp. 801-804.
National Research Council. (1980) Drinking water and health, v. 2. Washington, DC: National Academy Press.
National Research Council. (1983) Risk assessment in the federal government: managing the process.
Washington, DC: National Academy Press.
National Research Council. (1984) Toxicity testing: strategies to determine needs and priorities. Washington,
DC: National Academy Press.
National Research Council. (1985) Epidemiology and air pollution. Washington, DC: National Academy Press.
Available from: NTIS, Springfield, VA; PB86-137163.
National Research Council. (1986) Dose-route extrapolations: using inhalation toxicity data to set drinking water
limits. In: Drinking water and health: v. 6. Washington, DC: National Academy Press; pp. 168-204.
National Research Council. (1987) Pharmacokinetics in risk assessment: drinking water and health, v. 8.
Washington, DC: National Academy Press.
National Research Council. (199 la) Human exposure assessment for airborne pollutants: advances and
opportunities. Washington, DC: National Academy Press.
National Research Council. (1991b) Environmental epidemiology; v. 1: public health and hazardous wastes.
Washington, DC: National Academy Press.
5-20
-------
p. 40
National Research Council. (1994) Science and judgment in risk assessment. Washington, DC: National Academy
Press.
National Toxicology Program. (1986) Toxicology and carcinogenesis studies of xylenes (mixed) (60% m-xylene,
14% p-xylene, 9% o-xylene, and 17% ethylbenzene) (CAS no. 1330-20-7) in F344/N rats and B6C3F1
mice (gavage studies). Research Triangle Park, NC: U.S. Department of Health and Human Services,
National Institutes of Environmental Health and Safety; report no. NIH /PUB-87-2583. (National
Toxicology Program technical report series no. NTP-TR-327). Available from: NTIS, Springfield, VA;
PB87-1896847 AS.
Nuckols, M. L. (1981) Heat and water vapor transfer in the human respiratory system at hyperbaric conditions.
Naval Coastal Systems Center technical report; report no: TR364-81.
O'Brien, K. A. F.; Smith, L. L.; Cohen, G. M. (1985) Differences in naphthalene-induced toxicity in the mouse
and rat. Chem.-Biol. Interact. 55: 109-122.
O'Flaherty, E. J. (1981) Dose, response relationships. In: Toxicants and drugs: kinetics and dynamics. New
York, NY: John Wiley & Sons; pp. 354-389.
O'Flaherty, E. J. (1989) Interspecies conversion of kinetically equivalent doses. Risk Anal. 9: 587-598.
OberdSrster, G. (1990) Equivalent oral and inhalation exposure to cadmium compounds: risk estimation based on
route-to-route extrapolation. In: Gerrity, T. R.; Henry, C. J., eds. Principles of route-to-route
extrapolation for risk assessment, proceedings of the workshops; March and July; Hilton Head, SC and
Durham, NC. New York, NY: Elsevier Science Publishing Co., Inc.; pp. 217-235.
Ohmiya, Y.; Mehendale, H. M. (1984) Species differences in pulmonary N-oxidation of chlorpromazine and
imipramine. Pharmacology 28: 289-295.
Orr, C., Jr.; Keng, E. Y. H. (1976) Sampling and particle-size measurement. In: Dennis, R., ed. Handbook on
aerosols. Oak Ridge, TN: U.S. Energy Research and Development Administration, Technical Information
Center; pp. 93-117. Available from: NTIS, Springfield, VA; TID-26608.
Overton, J. H., Jr. (1984) Physicochemical processes and the formulation of dosimetry models. In: Miller, F. J.;
Menzel, D. B., eds. Fundamentals of extrapolation modeling of inhaled toxicants: ozone and nitrogen
dioxide. Washington, DC: Hemisphere Publishing Corporation; pp. 93-114.
Overton, J. H., Jr. (1989) A respiratory tract dosimetry model for air toxics. Presented at: U.S.-Dutch expert
workshop on air toxics; May 1988; Amersfoort, The Netherlands. Toxicol. Ind. Health 6: 171-180.
Overton, J. H.; Graham, R. C. (1994) Modeling the uptake of gases by the dog nasal-pharyngeal region: effects
of morphometric and physicochemical factors. Inhal. Toxicol. 6(suppl.): 113-124.
Overton, J. H.; Jarabek, A. M. (1989a) Estimating equivalent human concentrations of no observed adverse
effect levels: a comparison of several methods. In: Proceedings of a symposium—assessment of inhalation
hazards: integration and extrapolation using diverse data; February; Hannover, Federal Republic of
Germany. Exp. Pathol. 37: 89-94.
Overton, J. H.; Jarabek, A. M. (1989b) Estimating human equivalent no observed adverse effects levels for
VOCs based on minimal knowledge of physiological parameters. Presented at: 82nd annual meeting of
the Air and Waste Management Association; June; Anaheim, CA. Pittsburgh, PA: Air and Waste
Management Association; paper no. 89-91.8.
5-21
-------
p. 41
Overton, J. H.; Miller, F. J. (1988) Absorption of inhaled reactive gases. In: Gardner, D. E.; Crapo, J. D.;
Massaro, E. J., eds. Toxicology of the lung. New York, NY: Raven Press; pp. 477-507. (Target organ
toxicology series).
Overton, J. H.; Graham, R. C.; Miller, F. J. (1987) A model of the regional uptake of gaseous pollutants in the
lung: II. the sensitivity of ozone uptake in laboratory animal lungs to anatomical and ventilatory
parameters. Toxicol. Appl. Pharmacol. 88: 418-432.
Padgett, J.; Richmond, H. (1983) The process of establishing and revising national ambient air quality standards.
J. Air Pollut. Control Assoc. 33: 13-16.
Paiva, M. (1973) Gas transport in the human lung. J. Appl. Physiol. 35: 401-410.
Pang, K. S.; Rowland, M. (1977) Hepatic clearance of drugs. I. Theoretical considerations of a "well-stirred"
model and a "parallel tube" model. Influence of hepatic blood flow, plasma and blood cell binding, and
the hepatocellular enzymatic activity on hepatic drug clearance. J. Phannacokinet. Biopharm. 5: 625-653.
Patra, A. L. (1986) Comparative anatomy of mammalian respiratory tracts: the nasopharyngeal region and the
tracheobronchial region. J. Toxicol. Environ. Health 17: 163-174.
Patra, A. L.; Gooya, A.; Menache, M. G. (1986) A morphometric comparison of the nasopharyngeal airway of
laboratory animals and humans. Anat. Rec. 215: 42-50.
Pepelko, W. E. (1987) Feasibility of dose adjustment based on differences in long-term clearance rates of inhaled
particulate matter in humans and laboratory animals. Regul. Toxicol. Pharmacol. 7: 236-252.
Pepelko, W. E.; Withey, J. R. (1985) Methods for route-to-route extrapolation of dose. Toxicol. Ind. Health
1: 153-171.
Perera, F. (1987) The potential usefulness of biological markers in risk assessment. Environ. Health Perspect.
76: 141-145.
Perera, F. P.; Weinstein, I. B. (1982) Molecular epidemiology and carcinogen-DNA adduct detection: new
approaches to studies of human cancer causation. J. Chron. Dis. 35: 581-600.
Perlin, S. A.; McCormack, C. (1988) Using weight-of-evidence classification schemes in the assessment of
non-cancer health risks. In: HWHM '88, proceedings of the 5th national conference on hazardous wastes
and hazardous materials; April; Las Vegas, NV. Springfield, MD: Hazardous Materials Control Research
Institute; pp. 482-486.
Perry, R. H.; Chilton, C. H. (1973) Chemical engineers' handbook. 5th ed. New York, NY: McGraw-Hill Book
Company.
Phalen, R. F.; Oldham, M. J. (1983) Tracheobronchial airway structure as revealed by casting techniques.
Am. Rev. Respir. Dis. 128: S1-S4.
Phalen, R. F.; Stuart, B. O.; Lioy, P. J. (1988) Rationale for and implications of particle size-selective
sampling. In: Advances in air sampling: [papers from the ACGIH symposium]; February 1987; Pacific
Grove, CA. Chelsea, MI: Lewis Publishers, Inc.; p. 6. (Industrial hygiene science series).
Pickrel, C. A.; Samuhel, M. E.; Chesson, J. (1986) Quality assurance in epidemiologic studies. Washington,
DC: Battelle, Columbus Laboratories; contract no. 68-02-4246.
5-22
-------
p. 42
Plopper, C. G.; Mariassy, A. T.; Hill, L. H. (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.
Plopper, C. G.; Mariassy, A. T.; Wilson, D. W.; Alley, J. L.; Nishio, S. J.; Nettesheim, P. (1983) Comparison
of nonciliated tracheal epithelial cells in six mammalian species: ultrastructure and population densities.
Exp. Lung Res. 5: 281-294.
Portier, C. J.; Kaplan, N. L. (1989) Variability of safe dose estimates when using complicated models of the
carcinogenic process. Fundam. Appl. Toxicol. 13: 533-544.
Raabe, O. G. (1971) Particle size analysis utilizing grouped data and the log-normal distribution. J. Aerosol Sci.
2: 289-303.
Raabe, O. G. (1976) Aerosol aerodynamic size conventions for inertia! sampler calibration. J. Air Pollut. Control
Assoc. 26: 856-860.
Raabe, O. G. (1979) Deposition and clearance of inhaled aerosols. Washington, DC: U.S. Department of
Energy, Laboratory for Energy-Related Health Research; report no. UCD-472-503. Available from:
NTIS, Springfield, VA; UCD-472-503.
Raabe, O. G.; Bennick, J. E.; Light, M. E.; Hobbs, C. H.; Thomas, R. L.; Tillery, M. I. (1973) An improved
apparatus for acute inhalation exposure of rodents to radioactive aerosols. Toxicol. Appl. Pharmacol.
26: 264-273.
Raabe, O. G.; Yeh, H.-C.; Newton, G. J.; Phalen, R. F.; Velasquez, D. J. (1977) Deposition of inhaled
monodisperse aerosols in small rodents. In: Walton, W. H.; McGovern, B., eds. Inhaled particles IV,
part 1: proceedings of an international symposium; September 1975; Edinburgh, Scotland. Oxford,
United Kingdom: Pergamon Press, Ltd.; pp. 3-21.
Raabe, O. G.; Al-Bayati, M. A.; Teague, S. V.; Rasolt, A. (1988) Regional deposition of inhaled monodisperse,
coarse, and fine aerosol particles in small laboratory animals. In: Dodgson, J.; McCallum, R. L; Bailey,
M. R.; Fischer, D. R., eds. Inhaled particles VI: proceedings of an international symposium and
workshop on lung dosimetry; September 1985; Cambridge, United Kingdom. Ann. Occup. Hyg.
32 (suppl. 1): 53-63.
Ramsey, J. C.; Andersen, M. E. (1984) A physiologically based description of the inhalation pharmacokinetics of
styrene in rats and humans. Toxicol. Appl. Pharmacol. 73: 159-175.
Redmond, C. K. (1981) Sensitive population subsets in relation to effects of low doses. Environ. Health
Perspect. 42: 137-140.
Reeves, A. L.; Deitch, D.; Vorwald, A. J. (1967) Beryllium carcinogenesis: I. inhalation exposure of rats to
beryllium sulfate aerosol. Cancer Res. 27: 439-445.
Reid, L. M. (1980) Needs for animal models of human diseases of the respiratory system. Am. J. Pathol.
101: S89-S101.
Reitz, R. H.; McDougal, J. N.; Himmelstein, M. W.; Nolan, R. J.; Schumann, A. M. (1988) Physiologically
based pharmacokinetic modeling with methylchloroform: implications for interspecies, high dose/low
dose, and dose route extrapolations. Toxicol. Appl. Pharmacol. 95: 185-199.
Renwick, A. G. (1991) Safety factors and establishment of acceptable daily intakes. Food Addit. Contain.
8: 135-149.
5-23
-------
p. 43
Rhoads, K.; Sanders, C. L. (1985) Lung clearance, translocation, and acute toxicity of arsenic, beryllium,
cadmium, cobalt, lead, selenium, vanadium, ytterbium oxides following deposition in the rat lung.
Environ. Res 36: 359-378.
Rice, J. M. (1981) Prenatal susceptibility to carcinogenesis by xenobiotic substances including vinyl chloride.
Environ. Health Perspect. 41: 179-188.
Roels, H.; Lauwerys, R.; Buchet, J.-P.; Genet, P.; Sarhan, M. J.; Hanotiau, I.; de Fays, M.; Bernard, A.;
Stanescu, D. (1987a) Epidemiological survey among workers exposed to manganese: effects on lung,
central nervous system, and some biological indices. Am. J. Ind. Med. 11: 307-327.
Roels, H.; Lauwerys, R.; Genet, P.; Sarhan, M. J.; de Fays, M.; Hanotiau, I.; Buchet, J.-P. (1987b)
Relationship between external and internal parameters of exposure to manganese in workers from a
manganese oxide and salt producing plant. Am. J. Ind. Med. 11: 297-305.
Rossi, G. A. (1986) Bronchoalveolar lavage in the investigation of disorders of the lower respiratory tract. Eur.
J. Respir. Dis. 69: 293-315.
Rowland, M. (1985) Physiologic pharmacokinetic models and interanimal species scaling. Pharmacol. Ther.
29: 49-68.
Ruben, Z.; Rousseaux, C. G. (1991) The limitation of toxicologic pathology. In: Haschek, W. M.; Rousseaux,
C. G. eds. Handbook of toxicologic pathology. New York, NY: Academic Press; pp. 131-142.
Rudolf, G.: Gebhart, J.; Heyder, J.; Scheuch, G.; Stahlhofen, W. (1988) Mass deposition from inspired
polydisperse aerosols. Ann. Occup. Hyg. 32: 919-938.
Ruppel, G. (1979) Manual of pulmonary function testing, 2nd ed. St. Louis, MO: The C. V. Mosby Company.
Saltzman, B. E. (1988) Linear pharmacokinetic models for evaluating unusual work schedules, exposure limits
and body burdens of pollutants. Am. Ind. Hyg. Assoc. J. 49: 213-225.
Sanagi, S.; Seki, Y.; Sugimoto, K.; Hirata, M. (1980) Peripheral nervous system functions of workers exposed
to n-hexane at a low level. Int. Arch. Occup. Environ. Health 47: 69-79.
Sarlo, K.; Clark, E. D. (1992) A tier approach for evaluating the respiratory allergenicity of low molecular
weight chemicals. Fundam. Appl. Toxicol. 18: 107-114.
Saxena, M. C.; Siddiqui, M. K. J.; Bhargava, A. K.; Krishna Murti, C. R.; Kutty, D. (1981) Placental transfer
of pesticides in humans. Arch. Toxicol. 48: 127-134.
Schaper, M. (1993) Development of a database for sensory irritants and its use in establishing occupational
limits. Am. Ind. Hyg. Assoc. J. 54: 488-544.
Schlesinger, R. B. (1985) Comparative deposition of inhaled aerosols in experimental animals and humans:
a review. J. Toxicol. Environ. Health 15: 197-214.
Schreider, J.; Hutchens, J. (1980) Morphology of the guinea pig respiatory tract. Anat. Rec. 196: 313-321.
Schreider, J. P.; Raabe, O. G. (1981) Anatomy of the nasal-pharyngeal airway of experimental animals. Anat.
Rec. 200: 195-205.
Schulte, P. A. (1987) Methodologic issues in the use of biologic markers in epidemiologic research.
Am. J. Epidemiol. 126: 1006-1016.
5-24
-------
p. 44
Schulte, P. A. (1989) A conceptual framework for the validation and use of biologic markers. Environ. Res.
48: 129-144.
Seinfeld, J. H. (1986) Atmospheric chemistry and physics of air pollution. New York, NY: John Wiley and
Sons.
Selgrade, M. K.; Zeiss, C. R.; Karol, M. H.; Sarlo, K.; Kimber, I.; Tepper, J. S.; Henry, M. C. (1994)
Workshop on status of test methods for assessing potential of chemicals to induce respiratory allergic
reactions. Inhal. Toxicol. 6: 303-319.
Shibko, S. (1981) [Personal communication to M. Dourson documenting safety factors]. December 24.
Shoaf, C. R. (1993) EPA's inhalation testing guidelines. Presented at: International symposium on respiratory
toxicology and risk assessment; October 1992; Hanover, Germany. In press.
Slauson, D. O.; Hahn, F. F. (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: S103-S129.
Snider, G. L.; Kleinerman, J.; Thurlbeck, W. M.; Bengali, Z. H. (1985) The definition of emphysema: report of
a National Heart, Lung, and Blood Institute, Division of Lung Diseases workshop. Am. Rev. Respir.
Dis. 132: 182-185.
Snider, G. L.; Lucey, E. C.; Stone, P. J. (1986) Animal models of emphysema. Am. Rev. Respir. Dis.
133: 149-169.
Snipes, M. B. (1989a) Long-term retention and clearance of particles inhaled by mammalian species. CRC Crit.
Rev. Toxicol. 20: 175-211.
Snipes, M. B. (1989b) Species comparisons for pulmonary retention of inhaled particles. In: McClellan, R. O.;
Henderson, R. F., eds. Concepts in inhalation toxicology. New York, NY: Hemisphere Publishing;
pp. 193-227.
Snipes, M. B.; Boecker, B. B.; McClellan, R. O. (1983) Retention of monodisperse or polydisperse
aluminosilicate particles inhaled by dogs, rats, and mice. Toxicol. Appl. Pharmacol. 69: 345-362.
Snyder, W. S.; Cook, M. J.; Nasset, E. S.; Karhausen, L. R.; Howells, G. P.; Tipton, I. H. (1975) Report of
the task group on reference man. New York, NY: Pergamon Press. (International Commission on
Radiological Protection no. 23).
Society of Toxicology, Task Force of Past Presidents. (1982) Animal data in hazard evaluation: paths and
pitfalls. Fundam. Appl. Toxicol. 2: 101-107.
Sorokin, S. P. (1970) Properties of alveolar cells and tissues that strengthen alveolar defenses. Arch. Intern.
Med. 126: 450-463.
St. George, J. A.; Harkema, J. R.; Hyde, D. M.; Plopper, C. G. (1988) Cell populations and structure-function
relationships of cells in the airways. In: Gardner, D. E.; Crapo, J. D.; Massaro, E. J., eds. Target organ
toxicology series: toxicology of the lung. New York, NY: Raven Press, Ltd.; pp. 71-102.
Stahlhofen, W.; Gebhart, J.; Heyder, J. (1980) Experimental determination of the regional deposition of aerosol
particles in the human respiratory tract. Am. Ind. Hyg. Assoc. J. 41: 385-398a.
5-25
-------
p. 45
Stahlhofen, W.; Gebhart, J.; Heyder, J.; Scheuch, G. (1983) New regional deposition data of the human
respiratory tract. J. Aerosol Sci. 14: 186-188.
Stahlhofen, W.; Rudolf, G.; James, A. C. (1989) Intercomparison of experimental regional aerosol deposition
data. J. Aerosol Med. 2: 285-308.
Stara, J. F.; Erdreich, L. S., eds. (1984a) Approaches to risk assessment for multiple chemical exposures.
Cincinnati, OH: U.S. Environmental Protection Agency, Office of Health and Environmental
Assessment, Environmental Criteria and Assessment Office; EPA report no. EPA-600/9-84-008.
Available from: NTIS, Springfield, VA; PB84-182369.
Stara, J. F.; Erdreich, L. S., eds. (1984b) Selected approaches to risk assessment for multiple chemical
exposures: progress report on guideline development at ECAO-Cincinnati. Cincinnati, OH: U.S.
Environmental Protection Agency, Office of Health and Environmental Assessment, Environmental
Criteria and Assessment Office; EPA report no. EPA-600/9-84-014a. Available from: NTIS, Springfield,
VA; PB84-226992.
Stara, J. F.; Dourson, M. L.; DeRosa, C. T. (1981) Water quality criteria: methodology and applications.
In: Hoch, R. J., ed. Conference proceedings: environmental risk assessment, how new regulations will
affect the utility industry; December 1980; New Orleans, LA. Palo Alto, CA: Electric Power Research
Institute; pp. 3-1 to 3-18; report no. EPRI/EA-2064.
Stara, J. F.; Hertzberg, R. C.; Bruins, R. J. F.; Dourson, M. L.; Durkin, P. R.; Erdreich, L. S.; Pepelko,
W. E. (1985) Approaches to risk assessment of chemical mixtures. In: Homburger, F.; Marquis, J. K.,
eds. Chemical safety regulation and compliance: proceedings of a course; October 1983; Cambridge,
MA. Basel, Switzerland: Karger; pp. 71-83.
Stara, J. F.; Bruins, R. J. F.; Dourson, M. L.; Erdreich, L. S.; Hertzberg, R. C.; Durkin, P. R.; Pepelko,
W. E. (1987) Risk assessment is a developing science: approaches to improve evaluation of single
chemicals and chemical mixtures. In: Vouk, V. B.; Butler, G. C.; Upton, A. C.; Parke, D. V.; Asher,
S. C., eds. Methods for assessing the effects of mixtures of chemicals. New York, NY: John Wiley &
Sons; pp. 719-743. [Scientific Committee on Problems of the Environment (SCOPE) no. 30; International
Program on Chemical Safety (IPCS) joint symposia no. 6].
Stott, W. T.; McKenna, M. J. (1984) The comparative absorption and excretion of chemical vapors by the upper,
lower, and intact respiratory tract of rats. Fundam. Appl. Toxicol. 4: 594-602.
Svartengen, M. (1986) Lung deposition and clearance of particles in healthy persons and patients with
bronchiectasis. Stockholm, Sweden.
Task Group on Lung Dynamics. (1966) Deposition and retention models for internal dosimetry of the human
respiratory tract. Health Phys. 12: 173-207.
Tichy, M. (1983) Prediction of adverse activities from physical and chemical properties of vapors and gases
(QSAR analysis). In: Fiserova-Bergerova, V., ed. Modeling of inhalation exposure to vapors; v. 2,
uptake, distribution and elimination. Boca Raton, FL: CRC Press; pp. 3-35.
Tenney, S. M.; Remmers, J. E. (1963) Comparative quantitative morphology of the mammalian lung: diffusing
area. Nature (London) 197: 54-56.
Travis, C. C.; White, R. K. (1988) Interspecific scaling of toxicity data. Risk Anal. 8: 119-125.
Travis, C. C.; White, R. K.; Ward, R. C. (1990) Interspecies extrapolation of pharmacokinetics. J. Theor. Biol.
142: 285-304.
5-26
-------
p. 46
Trush, M. A.; Mimnaugh, E. G.; Gram, T. E. (1982) Activation of pharmacologic agents to radical
intermediates: implications for the role of free radicals in drug action and toxicity. Biochem. Pharmacol.
31: 3335-3346.
Tyl, R. W.; Ballantyne, B.; Fisher, L. C.; Fait, D. L.; Savine, T. A.; Pritts, I. M.; Dodd, D. E. (1994)
Evaluation of exposure to water aerosol or air by nose-only or whole-body inhalation procedures for
CD-I mice in developmental toxicity studies. Fundam. Appl. Toxicol. 23: 251-260.
Tyler, W. S.; Tyler, N. K.; Barstow, T.; Magliano, D.; Hinds, D.; Young, M. (1985) Effects in young
monkeys of intermittent episodes of exposure to low levels of ozone. Am. Rev. Respir. Dis.
131(suppl.): A169.
Tynes, R. E.; Hodgson, E. (1985) Catalytic activity and substrate specificity of the flavin-containing
monooxygenase in microsomal systems: characterization of the hepatic, pulmonary and renal enzymes of
the mouse, rabbit, and rat. Arch. Biochem. Biophys. 240: 77-93.
U.S. Department of Health and Human Services. (1994) Vital and health statistics: current estimates from the
National Health Interview Survey, 1992. Hyattsville, MD: Public Health Service, National Center for
Health Statistics; DHHS publication no. (PHS) 94-1517. (Series 10: data from the National Health
Survey no. 189).
U.S. Environmental Protection Agency. (1982a) Air quality criteria for oxides of nitrogen. Research Triangle
Park, NC: Office of Health and Environmental Assessment, Environmental Criteria and Assessment
Office; EPA report no. EPA-600/8-82-026. Available from: NTIS, Springfield, VA; PB83-131011.
U.S. Environmental Protection Agency. (1982b) Air quality criteria for particulate matter and sulfur oxides.
Research Triangle Park, NC: Office of Health and Environmental Assessment, Environmental Criteria
and Assessment Office; EPA report no. EPA-600/8-82-029aF-cF. 3v. Available from: NTIS, Springfield,
VA; PB84-156777.
U.S. Environmental Protection Agency. (1982c) Air quality criteria for particulate matter and sulfur oxides: v. 1,
addendum Research Triangle Park, NC: Environmental Criteria and Assessment Office; pp. A1-A15;
EPA report no. EPA-600/8-82-029aF. Available from: NTIS, Springfield, VA; PB84-156801/REB.
U.S. Environmental Protection Agency. (1984a) Risk assessment and management: framework for decision
making. Research Triangle Park, NC: Office of Health and Environmental Assessment, Environmental
Criteria and Assessment Office; EPA report no. EPA 600/9-85-002. Available from: NTIS, Springfield,
VA; PB85-170157/HSU.
U.S. Environmental Protection Agency. (1984b) Revised evaluation of health effects associated with carbon
monoxide exposure: an addendum to the 1979 EPA air quality criteria document for carbon monoxide.
Research Triangle Park, NC: Office of Health and Environmental Assessment, Environmental Criteria
and Assessment Office; EPA report no. EPA-600/8-83-033F. Available from: NTIS, Springfield, VA;
PB85-103471/HSU.
U.S. Environmental Protection Agency. (1984c) Review of the NAAQS for carbon monoxide: reassessment of
scientific and technical information. Research Triangle Park, NC: Office of Air Quality Planning and
Standards; EPA report no. EPA-450/5-84-004. Available from: NTIS, Springfield, VA; PB84-231315.
U.S. Environmental Protection Agency. (1985) Mutagenicity and carcinogenicity assessment of 1,3-butadiene:
final report. Washington, DC: Office of Health and Environmental Assessment, Carcinogen Assessment
Group; EPA report no. EPA-600/8-85-004F. Available from: NTIS, Springfield, VA; PB86-125507/AS.
5-27
-------
p. 47
U.S. Environmental Protection Agency. (1986a) Air quality criteria for lead. Research Triangle Park, NC: Office
of Health and Environmental Assessment, Environmental Criteria and Assessment Office; EPA report no.
EPA-600/8-83/028aF-dF. 4v. Available from: NTIS, Springfield, VA; PB87-142378.
U.S. Environmental Protection Agency. (1986b) Lead effects on cardiovascular function, early development, and
stature: an addendum to U.S. EPA air quality criteria for lead (1986). In: Air quality criteria for lead,
v. 1. Research Triangle Park, NC: Office of Health and Environmental Assessment, Environmental
Criteria and Assessment Office; pp. A1-A67; EPA report no. EPA-600/8-83-028aF. Available from:
NTIS, Springfield, VA; PB87-142378.
U.S. Environmental Protection Agency. (1986c) Second addendum to air quality criteria for particulate matter
and sulfur oxides (1982): assessment of newly available health effects information. Research Triangle
Park, NC: Office of Health and Environmental Assessment, Environmental Criteria and Assessment
Office; EPA report no. EPA-600/8-86-020F. Available from: NTIS, Springfield, VA; PB87-176574.
U.S. Environmental Protection Agency. (1986d) Air quality criteria for ozone and other photochemical oxidants.
Research Triangle Park, NC: Office of Health and Environmental Assessment, Environmental Criteria
and Assessment Office; EPA report nos. EPA-600/8-84-020aF-eF. 5v. Available from: NTIS,
Springfield, VA; PB87-142949.
U.S. Environmental Protection Agency. (1986e) Addendum to the health assessment document for
tetrachloroethylene (perchloroethylene): updated carcinogenicity assessment for tetrachloroethylene
(perchloroethylene, PERC, PCE) [external review draft]. Washington, DC: Office of Health and
Environmental Assessment, Carcinogen Assessment Group; EPA report no. EPA/600/8-82/005FA.
Available from: NTIS, Springfield, VA; PB86-174489.
U.S. Environmental Protection Agency. (1987) The risk assessment guidelines of 1986. Washington, DC: Office
of Health and Environmental Assessment; EPA report no. EPA/600/8-87/045. Available from: NTIS,
Springfield, VA; PB88-123997/XAB.
U.S. Environmental Protection Agency. (1988a) Recommendations for and documentation of biological values for
use in risk assessment. Cincinnati, OH: Office of Health and Environmental Assessment, Environmental
Criteria and Assessment Office; EPA report no. EPA-600/6-87-008. Available from: NTIS, Springfield,
VA; PB88-179874.
U.S. Environmental Protection Agency. (1988b) Reference physiological parameters in pharmacokinetic
modeling. Washington, DC: Office of Health and Environmental Assessment, Exposure Assessment
Group; EPA report no. EPA/600/6-88/004. Available from: NTIS, Springfield, VA; PB88-196019/AS.
U.S. Environmental Protection Agency. (1988c) Applications of an exact NOAEL procedure for dichotomous
data from animal experiments [final]. Cincinnati, OH: Office of Health and Environmental Assessment,
Environmental Criteria and Assessment Office.
U.S. Environmental Protection Agency. (1990) Occupational exposure limit data in relation to inhalation
reference doses. Washington, DC: Office of Health and Environmental Assessment; prepared for the Risk
Assessment Forum.
U.S. Environmental Protection Agency. (1991) Air quality criteria for carbon monoxide. Research Triangle Park,
NC: Office of Health and Environmental Assessment, Environmental Criteria and Assessment Office;
report no. EPA/600/8-90/045F. Available from: NTIS, Springfield, VA; PB93-167492.
5-28
-------
p. 48
U.S. Environmental Protection Agency. (1992) Summary of selected new information on effects of ozone on
health and vegetation: supplement to 1986 air quality criteria for ozone and other photochemical oxidants.
Research Triangle Park, NC: Office of Health and Environmental Assessment, Environmental Criteria
and Assessment Office; EPA report no. EPA/600/8-88/105F. Available from: NTIS, Springfield, VA;
PB92-235670.
U.S. Environmental Protection Agency. (1993a) Air quality criteria for oxides of nitrogen. Research Triangle
Park, NC: Office of Health and Environmental Assessment, Environmental Criteria and Assessment
Office; EPA report no. EPA/600/8-9!/049aF-cF. 3v.
U.S. Environmental Protection Agency. (1993b) Air quality criteria for ozone and related photochemical oxidants
[review draft]. Washington, DC: Office of Research and Development; EPA report nos.
EPA/600/AP-93/004a-c. 3v.
U.S. Food and Drug Administration. (1982) Toxicological principles for the safety assessment of direct food
additives and color additives used in food. Washington, DC: U.S. Food and Drug Administration,
Bureau of Foods.
Ultman, J. S. (1988) Transport and uptake of inhaled gases. In: Watson, A. Y.; Bates, R. R.; Kennedy, D., eds.
Air pollution, the automobile, and public health. Washington, DC: National Academy Press;
pp. 323-366.
Ultman, J. S. (1994) Dosimetry modeling: approaches and issues. Inhal. Toxicol. 6(suppl.): 59-71.
Vanderslice, R. R.; Domin, B. A.; Carver, G. T.; Philpot, R. M. (1987) Species-dependent expression and
induction of homologues of rabbit cytochrome P-450 isozyme 5 in liver and lung. Mol. Pharmacol.
31: 320-325.
Vettorazzi, G. (1977) Safety factors and their application in the lexicological evaluation. In: Hunter, W. J.;
Smeets, J. G. P. M., eds. The evaluation of toxicological data for the protection of public health:
proceedings of the international colloquium; December 1976; Luxembourg, Sweden. Oxford, United
Kingdom: Pergamon Press; pp. 207-223.
Vettorazzi, G. (1980) Handbook of international food regulatory toxicology: v. I, evaluations. New York, NY:
Spectrum Publications; pp. 66-68.
Ward, R. C.; Travis, C. C.; Hetrick, D. M.; Andersen, M. E.; Gargas, M. L. (1988) Pharmacokinetics of
tetrachloroethylene. Toxicol. Appl. Pharmacol. 93: 108-117.
Weibel, E. R. (1963) Morphometry of the human lung. New York, NY: Academic Press Inc.
Weibel, E. R. (1972) Morphometric estimation of pulmonary diffusion capacity: V. comparative morphometry of
alveolar lungs. Respir. Physiol. 14: 26-43.
Weill, C. S. (1972) Statistics vs safety factors and scientific judgement in the evaluation of safety for man.
Toxicol. Appl. Pharmacol. 21: 454-463.
Weil, C. S.; McCollister, D. D. (1963) Relationship between short- and long-term feeding studies in designing
an effective toxicity test. J. Agric. Food Chem. 11: 486-491.
Weil, C. S.; Woodside, M. D.; Bernard, J. R.; Carpenter, C. P. (1969) Relationship between single-peroral,
one-week, and ninety-day rat feeding studies. Toxicol. Appl. Pharmacol. 14: 426-431.
5-29
-------
p. 49
WeiB, M.; Sziegoleit, W.; Forster, W. (1977) Dependence of phannacokinetic parameters on the body weight.
Int. J. Clin. Pharmacol. Biopharm. 15: 572-575.
Wen, C. P.; Tsai, S. P.; Gibson, R. L. (1983) Anatomy of the healthy worker effect: a critical review.
J. Occup. Med. 25: 283-289.
Willeke, K., ed. (1980) Generation of aerosols and facilities for exposure experiments. Ann Arbor, MI: Ann
Arbor Science Publishers, Inc.
Wolff, R. K.; Griffis, L. C.; Hobbs, C. H.; McClellan, R. O. (1982) Deposition and retention of
0.1 /un ^GajOj aggregate aerosols in rats following whole body exposures. Fundam. Appl. Toxicol.
2: 195-200.
Woodruff, T. J.; Bois, F. Y.; Auslander, D.; Spear, R. C. (1992) Structure and parameterization of
phannacokinetic models: their impact on model predictions. Risk Anal. 12: 189-201.
Yeh, H. C.; Schum, G. M.; Duggan, M. T. (1979) Anatomic models of the tracheobronchial and pulmonary
regions of the rat. Anat. Rec. 195: 483-492.
Yost, G. S.; Buckpitt, A. R.; Roth, R. A.; McLemore, T. L. (1989) Mechanisms of lung injury by systemically
administered chemicals. Toxicol. Appl. Pharmacol. 101: 179-195.
Young, J. T. (1981) Histopathologic examination of the rat nasal cavity. Fundam. Appl. Toxicol. 1: 309-312.
Yu, C. P.; Xu, G. B. (1987) Predictive models for deposition of inhaled diesel exhaust particles in humans and
laboratory species. Cambridge, MA: Health Effects Institute; research report no. 10.
Yu, C. P.; Yoon, K. J. (1990) Retention modeling of diesel exhaust particles in rats and humans. Amherst, NY:
State University of New York at Buffalo (Health Effects Institute research report no. 40).
Yu, C. P.; Diu, C. K.; Soong, T. T. (1981) Statistical analysis of aerosol deposition in nose and mouth.
Am. Ind. Hyg. Assoc. J. 42: 726-733.
Ziegler, D. M. (1980) Microsomal flavin-containing monooxygenase: oxygenation of nucleophilic nitrogen and
sulfur compounds. In: Jakoby, W. B., ed. Enzymatic basis of detoxication: v. 1. New York, NY:
Academic Press, Inc.; pp 201-227.
Ziegler, D. M. (1988) Flavin-containing monooxygenases: catalytic mechanism and substrate specificities. Drug
Metab. Rev. 19: 1-32
Zielhuis, R. L.; van der Kreek, F. W. (1979) The use of a safety factor in setting health based permissible levels
for occupational exposure. Int. Arch. Occup. Environ. Health 42: 191-201.
5-30
-------
p. 50
APPENDIX A
ALTERNATIVE APPROACHES
TO THE ESTIMATION OF
NO-OBSERVED-ADVERSE-EFFECT LEVELS
The inhalation reference concentration (RfC) approach based on a lowest-observed-
adverse-effect level/no-observed-adverse-effect level (LOAEL/NOAEL) paradigm is
consistent with current methods for estimating human health risks from exposure to threshold-
acting toxicants in water or food, such as those established by the Food and Drug
Administration (Kokoski, 1976), the National Research Council (1977, 1980), the World
Health Organization, the Food and Agricultural Organization (Bigwood, 1973; Vettorazzi,
1977, 1980; Lu, 1983), and other approaches used by U.S. Environmental Protection Agency
(Federal Register, 1980; Stara et al., 1981; Barnes and Dourson, 1988). To date, these
methods have generally considered only chronic or lifetime exposure to individual chemicals
based on the assumption that "lifetime" data in laboratory animals are directly applicable to
lifetime human exposures. As our understanding of the exposure-dose-response continuum is
refined and the temporal aspects of the pathogenesis mechanisms elucidated (see Section
4.3.2), dose-response benchmark estimates for health risk characterization may be able to
address intermediate duration, periodic, and other exposure scenarios with greater accuracy.
These methods generally estimate a single, constant daily dose that is low enough to be
considered "safe" or "acceptable" (referred to as an acceptable daily intake [ADI]) or without
appreciable risk (RfC). A number of scientific problems with this approach have been long
recognized (Krewski et al., 1984; Crump, 1984; Brown and Erdreich, 1989). The first
problem is that this method does not readily account for the number of animals used to
determine the appropriate NOAEL. As described in Section 4.2 on designation of effect
levels, the NOAELs or LOAELs that serve as the critical data in the RfC approach can be
based on statistically significant or biologically significant increases in the frequency or
severity of adverse toxic effects. For example, NOAELs have been defined for quantal
endpoints that have nonzero background incidences by choosing an experimental exposure
A-l
-------
p. 51
level that does not contribute to a statistically significant increase in incidence of adverse
effects when compared to a control group. Some NOAELs have been defined for continuous
data by choosing an experimental exposure level that does not constitute a significantly
different mean value for a parameter, indicating an adverse effect when compared to a mean
value for a control group. Statistical significance, however, depends heavily on the design of
the experiment, including sample size, the number of concentrations used, the spacing of the
concentrations, and the arbitrary alpha level. Often the only information gained from the
experiment used as the critical study is the presence or absence of statistical significance for
an arbitrary alpha level at a small number of concentrations. Similarly, biological
significance is often attributed to a concentration with little consideration of the impact of
experimental design and no strict definition of the biological changes, suggesting that the
designation of NOAEL or LOAEL is to an extent subjective. For example, if a chemical has
a NOAEL based on 10 animals and another NOAEL with the same value but based on
100 animals, the risk assessor often will choose the NOAEL based on the larger study
because it yields greater confidence in the resulting RfC. However, comparison of statistical
power is not routinely done and the influence of sample size may not be taken into account
when comparing disparate NOAELs. It has also been argued that the use of this approach
encourages studies with smaller sample sizes, which reduce the power of the test. If these
NOAELs were for different chemicals, similar RfCs might be derived, even though one
would be associated with much less confidence.
The second problem with the current NOAEL/LOAEL approach is that the slope of the
dose-response curve of the critical toxic effect is generally ignored in the estimation of the
NOAEL. Many scientists have argued that this slope should in some way directly affect the
estimate, with steep curves presumably yielding lower values because thresholds or greater
toxicities are more quickly obtained with increasing concentration.
Furthermore, the current NOAEL/LOAEL approach to noncancer dose-response
assessment yields an RfC estimate that is presented as a single number. As such, it reflects
neither the statistical variability in the NOAEL resulting from study design factors nor the
inherent variability for which uncertainty factors are applied to extrapolate from the data base
to the RfC. The result of this variability is the unknown range of uncertainty in the estimate.
Exposure estimates to which the dose-response estimate must be compared are also associated
A-2
-------
p. 52
with a range of uncertainty and many exposure models now express explicitly this variability
as a distribution. Risk management decisions for regulation or enforcement need more
quantitative information on the inherent and recognized uncertainties in this assessment.
This appendix defines and illustrates alternative approaches to derivation of estimates
that could be used as analogues to the NOAEL. Many of these approaches offer solutions to
some of the criticisms of the NOAEL/LOAEL approach outlined above and these attributes
will be highlighted. Even so, no method is without inherent problems. Guidance is under
development that describes the application of "benchmark" concentration-response modeling
(Section A.2) to derive dose-response estimates such as the inhalation RfC. Recently, EPA
and the Risk Science Institute of the International Life Sciences Institute (ILSI) sponsored a
workshop entitled, "Workshop on Benchmark Dose Methodology". A summary paper from
the deliberations at these meetings discusses definitions and criteria for the use of a
benchmark approach to estimate a reference dose or reference concentration (Barnes et al.,
1994). The Risk Assessment Forum is also working on guidance that is anticipated to be
published as a "purple book". The reader is referred to these additional sources and is
encouraged to appreciate that development of guidance awaits consensus on issues raised both
herein and in these additional materials.
It is worthwhile to emphasize, as it will be noted in subsequent sections of this
appendix, that the lexicological decision as to what constitutes adversity (i.e., the decision
that a specified effect is adverse and what the associated severity is), particularly across
different endpoints, remains perhaps the most sensitive parameter in any of these procedures
regardless of the mathematical model applied. Using quanta! data, for example, it is a
decision based on toxicological judgment that determines whether 10% or 30% incidence of a
given lesion should be a concern. Similarly, toxicological or clinical insight may be required
to determine if a particular change in a continuous parameter (e.g., pulmonary function
decrement) is adverse relative to a normal population value or between a control and an
exposed cohort. To date, there has not been adequate appreciation by toxicologists and
biostatisticians alike of the interdependence of the decision to designate an effect as
biologically significant and the decision to estimate a response at a given level from the
mathematical model. Perhaps awareness of the interdependence is the single most important
factor that requires systematic development before any of these approaches can be
A-3
-------
p. 53
implemented consistently. The decision on the definition of adversity or biological
significance is termed designation of the "specified health effect" for purposes of discussion
in this appendix. The concept of a specified health effect is not new and is related to the
concept of "relative potency" (Jarabek and Hasselblad, 1991). Finney (1978) defines a
relative potency in his description of a direct assay. A direct assay is one in which "...doses
of the standard and test preparations sufficient to produce a specified response are directly
measured. The ratio between these doses estimates the potency of the test preparation relative
to the standard...." Note that the choice of a "specified response" is key to the definition.
Because all of the approaches presented herein have not yet been applied routinely to the
types of data generally encountered when evaluating the health effects information available
on the majority of inhaled chemicals, aspects that require further development and
consideration in order to use these alternatives will also be presented.
A.1 NO-STATISTICAL-SIGNIFICANCE OF TREND (NOSTASOT)
A statistically more accurate approach than the traditional NOAEL/LOAEL for
estimating a NOAEL when several exposure levels are available is the "no statistical
significance of trend" (NOSTASOT) approach proposed by Tukey et al. (1985). The
underlying principle is to sequentially test for a linear trend until significance is no longer
reached. As described by Tukey et al. (1985), the procedure is applied to all of the data first
and then entails sequentially deleting the highest exposure groups in succession downward
(i.e., "top-down" analysis). In this manner, the highest exposure level at which the response
is not significantly different from controls is determined to be the NOSTASOT, which could
therefore be considered a NOAEL.
A. 1.1 Approach Advantages
The advantage of the NOSTASOT approach is that it offers a simple yet fairly robust
method to determine a NOAEL by testing for a trend in all exposure levels (including
controls). As such, it utilizes more of the concentration-response information than individual
comparisons of exposed and control groups.
A-4
-------
p. 54
A.I.2 Application Issues and Development Needs
As proposed by Tukey et al. (1985), the NOSTASOT procedure tests for a statistical
significant trend in a series of exposure levels (including controls) until the highest level at
which the trend is nonsignificant is reached. This highest level is defined as the NOSTASOT
exposure and would be used as a surrogate to the NOAEL in derivation of an RfC. The last
concentration at which statistical significance was achieved would be a LOAEL. The method
was developed for application to data from experiments involving multiple groups of animals
of approximately the same size at different dose/exposure levels, including a zero-dose
control. In such cases, the NOSTASOT method may be preferred because it includes more
information and may have greater statistical power than multiple comparisons of different
experimental groups to a control group. However, despite its robustness, the NOSTASOT
approach remains sensitive to dose spacing. It is also sensitive to sample size when applied
to grouped data. Application to epidemiologic data with individual exposure data or to
continuous response measures is not straightforward.
An alternative to the NOSTASOT approach is to start at the lowest noncontrol exposure
or dose level and move upward (i.e., "bottom-up" analysis). The objective is to determine
the highest level of nonsignificance before a significant difference is detected. The highest
nonsignificant dose would be declared the NOAEL and the first statistically significant dose a
LOAEL in this analysis. The sensitivity to sample size and dose spacing of the NOSTASOT
approach is illustrated by the difference between a "top-down" versus "bottom-up" analysis.
In most animal experiments for which the procedure was developed, with groups of the same
size exposed to typically only at one, two, three, or four levels, the NOSTASOT would be
the same whether analyzed from the top down or the bottom up, assuming that the response is
monotonic. However, in data sets with a large number of exposure levels or with individual
exposure data, the top-down and bottom-up analyses may yield very different estimates (i.e.,
a LOAEL from the bottom-up analysis may be below a NOAEL from the top-down analysis).
This is conceivable with some nonlinear and/or nonmonotonic data sets (Davis and
Svendsgaard, 1990). It is therefore necessary to apply the method in a manner that
recognizes possible nonlinearities in the data (e.g., due to a sensitive subpopulation
responding at low concentrations). Such complications warrant consideration when applying
the NOSTASOT approach.
A-5
-------
p. 55
A.2 "BENCHMARK" CONCENTRATION-RESPONSE MODELING
Concentration-response modeling, recently referred to as the "benchmark dose"
approach, has been proposed as an improvement on the NOAEL/LOAEL approach (Crump,
1984). The "benchmark" approach, as defined in this discussion, is the use of a specific
mathematical model (e.g., Weibull, logistic, polynomial) to determine a concentration
(applied dose) and its lower confidence bound that is associated with a predefined effect
measure (e.g., 10% response of a dichotomous outcome) as the "benchmark". Application of
this approach (Kimmel and Gaylor, 1988) has been proposed for developmental endpoints,
which are generally dichotomous (quantal) in nature, but it has yet to be applied widely to
other noncancer outcomes.
Figure A-l illustrates the benchmark approach as applied to laboratory animal
developmental data. A mathematical model (e.g., Weibull, logistic, polynomial) is applied
(fitted) to the experimental effects data to estimate a maximum likelihood estimate (MLE) or
concentration-response function. The 95% confidence limit is calculated using information
on sample size and variance. It has been recommended that limits based upon the distribution
of the likelihood ratio statistic be used as the method of choice for this calculation
(Crump, 1984). The possible analogues to a NOAEL can then be estimated. For example,
the 10th percentile of an effect level could be designated as synonymous to "no adversity"
and the concentration corresponding to the MLE of that effect level used as the "effective
concentration" (EC10). The lower confidence bound on the EC10 could also be used and is
shown as "LEC10". A linear interpolation has also been proposed (Gaylor and Kodell, 1980;
Kimmel and Gaylor, 1988) that allows estimation of upper limits on risk for convex dose-
response curves. For example, as shown in Figure A-l, at a dose of LEC10 divided by an
uncertainty factor (UF), the "true" unknown risk in the low-dose regions is expected to be
less than that associated with the linear extrapolation if the "true" dose-response curves
upward.
A.2.1 Approach Advantages
Compared to the NOAEL/LOAEL approach, benchmark concentration-response
modeling has the advantages that it utilizes more information from the dose-response curve, is
less influenced by experimental design (e.g., exposure level spacing), and is sensitive to the
A-6
-------
p. 56
Excess
Proportion
of Abnormal
Responses
0.1
95% Lower Confidence
Umit on Estimated Risk.
0.1/UF
0
Concentration-
response Fitted
to Experimental
Data(-)
LEC10/UF
'10
Concentration (c)
Figure A-l. Graphical illustration of proposed low-dose risk estimation for the
proportion of abnormal responses in developmental toxicity.
Adapted from Kimmel and Gaylor (1988).
influence of sample size. It is important to note that this approach is sensitive to sample size
only when the "benchmark" is defined as the lower confidence bound. The MLE alone is not
influenced by sample size.
A.2.2 Application Issues and Development Needs
Application of this approach to the myriad of endpoints that can constitute noncancer
toxicity will require significantly greater effort directed at modeling continuous data.
A limitation may be finding data sets appropriate for modeling. Guidance must be developed
on choice of model structures and on goodness-of-fit criteria for models, especially whether
or not it is appropriate to superimpose model structures on data that only have one dose
group associated with a nonzero response (relative to control or background). Whether or not
there is a biological basis (e.g, for certain endpoints) for selecting certain model structures
also warrants investigation.
A-7
-------
p. 57
Use of the benchmark approach still requires dosimetric adjustment to a human
equivalent concentration (Section 4.3) and for application of UFs to account for
extrapolations (Section 4.3.8.1). Dosimetric adjustment to account for interspecies
differences should be applied before the data are modeled.
Application of UFs in a fashion analogous to that used with the NOAEL/LOAEL
paradigm have been proposed for use with the benchmark approach (Dourson et al., 1985,
1986). That is, a benchmark estimate for a more severe endpoint (e.g., liver necrosis),
essentially equivalent to a LOAEL, would warrant application of an additional UF, whereas
the endpoint judged as less severe (e.g., slight body weight decrease) would not. Application
of UF for intraspecies variability, subchronic duration, and data base may also be
appropriate.
Another approach for the application of UFs for dichotomous data has been proposed
using the linear interpolation from the LEC10 through the origin as shown in Figure A-l
(Kimmel and Gaylor, 1988). As shown on Figure A-l, if UF represents an uncertainty
factor, then the true unknown risk at an exposure concentration of LEC10/UF is expected to
be less than 0.1/UF. This procedure is conservative with respect to risk when the dose-
response is convex (curving upward). Therefore, an advantage is that an upper limit on the
risk is estimated. The size of the factor depends on the desired level of risk. For example, a
factor of 10 applied to the LEC10 would result in a risk less than 10~2. This approach
assumes that the incidence in humans on which the "acceptable risk" decision is based is
equivalent to the observed incidence of a given lesion in the experimental animals.
An equivalent procedure for continous data would necessitate an assumption that the mean
severity or magnitude of the observed effect in the exposed population relative to the control
(or relative to a normal reference) was equivalent in experimental animals and humans.
These UF approaches essentially result in subthreshold estimates, similar in intent to the
RfC, provided the LEC10 is considered to be analogous to a NOAEL and if the designation
of the specified health effect is unequivocal. However, the designation of a specified health
effect is a question of both biological and statistical significance. Various levels (e.g., EC01,
EC05, and EC10) have been proposed that could be considered as a NOAEL criterion
A-8
-------
p. 58
(Gaylor, 1983; Kimmel and Gaylor, 1988; Fabro et al., 1982).1 If one incidence level were
to be designated as the NOAEL criterion (e.g., 10%), a dose-response estimate could be
based on either 10% nasal hyperplasia or 10% proximal tubule necrosis, unless the severity
of the endpoint is taken into account. Intimate knowledge of the spectrum of severity within
a pathogenesis continuum for an individual endpoint may be required before criteria can be
established for designating specified health effects. Further, in order to compare across the
various endpoints associated with noncancer toxicity, it may be necessary to "normalize"
(e.g., designate the 50th percentile as the criterion for a minimally adverse effect and the
1st percentile for a severe effect), but this would require consensus on definitions of severity.
The interaction with model structure may also be influenced by these criteria. For example,
model "fit" and variability in the resultant estimate would be different for lesions designated
at the EC50 and the EC01 and determined at the associated lower confidence bound. The
relationship of these different estimates to applied UFs would also be different. The choice
of the mathematical model structure generally makes relatively little difference down to
approximately the 1% risk level. Estimation of excess risk above background in the region
below that level can become more dependent on the choice of model structure than on the
true dose-response curve. Although previous use of the "benchmark" approach avoided this
controversy because developmental endpoints do not distinguish degrees of severity to a large
extent, such issues are critical for development of this approach as an application to all the
other common noncancer endpoints.
Derivation of a dose-response estimate by the benchmark approach also does not
preclude evaluation of the data base for completeness. A comprehensive array of endpoints
must be evaluated to identify potential hazard for various target tissues regardless of the way
individual endpoints may be modeled. Once the individual specified health effects are
decided, determinations of the appropriate species and critical effect representative of the
threshold for the overall data array must be evaluated as described for the RfC methodology
in Section 4.3.7.
1It could also be argued that the exposure estimated to be the 5th percentile is really a lower confidence limit
of the exposure causing a specified effect. In that sense, any point below it is associated with no effect, and
therefore the 5th percentile (or any other lower tail percentile) could be considered as a NOAEL. Designation
criteria for the LOAEL, however, will be problematic as outlined above.
A-9
-------
p. 59
A.3 APPLICATION OF BAYESIAN STATISTICS
As discussed in Section 4.3.7, the analysis of noncancer toxicity data often involves
evaluation of data and a synthesis of information together in order to determine a
representative level for the threshold region of the data array. For example, sometimes a
NOAEL from one study may be used in conjunction with a NOAEL from another. Data
from a "free-standing" NOAEL are often used in a qualitative sense but cannot be used in a
dose-response model. The advantage to such a synthesis is the utilization of more
information rather than the reduction of data to a single study and its effect level, a practice
that is recognized as a significant limitation to the RfC and benchmark approaches described
above.
A Bayesian statistical approach has been proposed that both statistically incorporates the
attributes of the benchmark approach (incorporates influence of sample size and shape of the
dose-response curve) as well as offers the advantages of (1) visual display and description of
the uncertainty in the risk estimates, (2) allows for explicit synthesis of dose-response
estimates together when determined appropriate, and (3) allows for explicit incorporation of
uncertainty in the exposure characterization (Jarabek and Hasselblad, 1991; Hasselblad and
Jarabek, 1994).
The general approach proposed has been published under the title of the Confidence
Profile Method (Eddy et al., 1992). It combines the standard classical and Bayesian
statistical methods to produce likelihood functions and posterior distributions for parameters
of interest. Although the likelihood functions and posterior distributions have very different
interpretations, their shape is usually extremely similar. The likelihood function can be used
to compute confidence intervals. The posterior distribution is a continuous plot describing
belief about the location of the parameter of interest (i.e., for dose-response estimation
purposes, about the dose associated with a specified health effect). The basic formula of
Bayesian statistics is
p'(0) = L(0 | data) p(0), (A-l)
where:
A-10
-------
p. 60
6 = parameter of interest,
p(0) = prior distribution for 6,
L(01 data) = likelihood for 6 given new data, and
p'(0) = posterior distribution for 6. Because p'(0) will become the prior for the
next experiment, it is denoted by the same letter.
Consider the following simple example of a continuous effect measure. Assume that the
health effect, y, is related to the exposure, x, given by the model
y = /3x. (A-2)
Assume further that we wish to specify a particular health effect, y0, and then estimate the
exposure corresponding to this effect as
6 = y0/)8. (A-3)
Because 8 is not defined for ft < 0, it is reasonable to choose the prior for 6 as p(/J) = 1 for
j8 > 0 and p(/5) = 0 elsewhere. This corresponds to the belief that exposure to a toxic
chemical is not beneficial. The prior just described is the horizontal dashed line in
Figure A-2. Assume that an experiment to determine information about /J was conducted,
resulting in the likelihood, L(/5), shown as a dotted line in Figure A-2. Note that this
likelihood is positive for values of /? less than 0. The posterior distribution, p'(/?), is the
product of the likelihood function and the prior (properly normalized to be a probability
distribution) and is shown as a solid line in Figure A-2. Note that this distribution has the
same general shape as the likelihood function, except that it has no mass below zero. This
kind of distribution is often referred to as a truncated distribution. The posterior distribution
of 9 can be calculated from the posterior distribution of /}. It should be emphasized that the
mathematical modeling of these data was not different for these effect measures than that
which could be achieved using a benchmark approach, but the expression as a normalized
posterior distribution is the difference that provides for visual inspection and statistical
combination of data. The posterior distribution, p'(0), can be used as a prior if another
experiment is conducted giving additional information about /3, and the application of Bayes'
A-ll
-------
p. 61
Figure A-2. Schematic of computing a posterior distribution [p; (/J^C)] from a
likelihood function [L (/31 data)] and a prior distribution [p (ft)].
Source: Jarabek and Hasselblad (1991).
formula repeated. (Note: In the following applications, 8 [the parameter of interest] is
designated as XQ, the exposure concentration associated with a specified health effect.)
The proposed Bayesian approach is illustrated in Figures A-3 through A-5. Figure A-3
shows the dose-response from a logistic model superimposed on the dichotomous data for
nasal turbinate lesions in mice exposed to «-hexane (data of Dunnick et al., 1989). For
illustration purposes, an incidence of 10% (shown by the dashed line) is designated as the
specified health effect (Jarabek and Hasselblad, 1991). Figure A-4 shows the posterior
distribution for the n-hexane concentration associated with that 10% incidence. Figure A-5
shows the statistical synthesis together of posterior distributions of two different
concentrations associated with specified health effects (one respiratory, the other
neurotoxicity) of two studies (Dunnick et al., 1989; Sanagi et al., 1980.) The results of this
synthesis were in general agreement with the NOAEL used for the RfC derivation for this
chemical (IRIS, 1990) and with the benchmark approach for either of the two studies (data
not shown). Although experimental details are provided elsewhere (Jarabek and Hasselblad,
1991), the two data sets represent both continuous and dichotomous effect measures,
illustrating the ability of the Bayesian approach to address different outcomes.
A-12
-------
1.0.
0.5-
0.1-
p. 62
3,000
6,000
(ppm)
0,000
12,000
Figure A-3. Incidence of nasal turbinate lesions in B6C3F1 female mice exposed to
n-hexane for 13 weeks. Data of Dunnick et al. (1989).
Source: Jarabek and Hasselblad (1991).
1
p(Xo)
Dunnick etal. (1888)
100
200
n-Haxane (mfl/m»)
300
400
Figure A-4. Posterior distribution for the n-hexane concentration associated with the
specified health effect in Figure A-3. Data of Dunnick et al. (1989).
Source: Jarabek and Hasselblad (1991).
A-13
-------
p. 63
T
PW
ComMmd
Dunnick «t«l. (1889)
S*ngd «•!. (1980)
~ I
100 200
n-Hncaiw
i
300
400
Figure A-5. Posterior distribution for the concentration of n-hexane associated with the
specified health effects from the combined evidence of Sanagi et al. (1980)
and Dunnick et al. (1989).
Source: Jarabek and Hasselblad (1991).
A.3.1 Approach Advantages
Visual presentation of data is a powerful tool for analysis and communication
(Cleveland, 1985). Visual inspection of the posterior distribution concurs with the variability
of the data and provides much information about the usefulness of the health effects data for
dose-response evaluation. The shape of the posterior distribution for the data of Dunnick
et al. (1989), in contrast to that of Sanagi et al. (1980), easily highlights that these data were
generated from an investigation with an adequate number of animals and test concentrations
with a resultant tighter distribution and reduced variance. The skewed posterior distribution
for the data of Sanagi et al. (1980) results from its greater variance and small sample size.
The value of visual presentation is again illustrated in Figure A-6. This figure shows the
posterior distributions for the concentration of manganese (Mn) associated with specified
health effects (all approximately the same measure of neurotoxicity) from three different
studies. The visual presentation of the posterior distribution easily communicates that the
data of Chandra et al. (1981) were highly variable and in fact do not add much information
A-14
-------
p. 64
t
p(xo)
Roelsetal. (1987a)
(fixed exposure)
0.5
Mn (mg/m3)
Figure A-6. Posterior distributions for the manganese (Mn) concentration associated
with specified health effects from each of three studies: Roels et al.
(1987a), Iregren (1990), and Chandra et al. (1981).
to the synthesis. An appreciation of this variability would not have been imparted from the
numeric reporting of the estimate alone. Even if the percentile values were reported and
some sort of analysis on the spread is done (e.g., compare the ratio of the 95th to 50th
percentile for all studies), the communication of the reliability of these data is not as
straightforward as that of the visual display.
The Bayesian approach also allows for explicit incorporation of uncertainty in the
exposure estimates of the studies being evaluated. The influence that direct application of
uncertainty in the exposure estimate can have on the resultant dose-response estimate is
illustrated in Figures A-7 through A-9. These figures illustrate the influence of variability in
exposure characterization for the health effect data used to determine the dose-response.
Because the posterior distribution now expresses the dose-response estimate as a distribution
instead of a point estimate, this approach allows the dose-response distribution to be
combined statistically with an exposure distribution for risk characterization. Therefore, such
A-15
-------
p. 65
Figure A-7. Posterior distribution for the manganese (Mn) concentration associated
with specified health effect using either exposure or estimated exposure
distribution. Data Roels et al. (1987a,b).
t
p(xo)
.Iregren (1990)
Combined
Roels etal.(1987a)
/ (fixed exposure)
1
0.5
Mn (mg/m3)
Figure A-8. Posterior distribution for the concentration of manganese (Mn) associated
with specified health effect from the combined evidence of Iregren (1990)
and Roels et al. (1987a) with fixed exposure.
A-16
-------
p. 66
T
P(X«)
lregren(1990)
Combined
/ Reels etal.(1987a,b)
/ (exposure distribution)
0 0.5 1
Mn (mg/m*)
Figure A-9. Posterior distribution for the concentration of manganese (Mn) associated
with specified health effect from the combined evidence of Iregren (1990)
and Roels et al. (1987a,b) with exposure distribution.
presentation will allow explicit incorporation of uncertainty in the dose-response and exposure
estimate to be carried through to the risk characterization step and could provide more
information on which to base management decisions.
The Bayesian approach offers the advantages of the benchmark approach in that it takes
into account the influence of sample size and shape of the dose-response curve. However, it
is the only currently viable approach that offers the ability to statistically combine evidence
from different investigations. Such synthesis is routinely done with data without explicit
statistical handling of experimental design.
The Bayesian approach offers the advantages of the benchmark approach in that it takes
into account the influence of sample size and shape of the dose-response curve. However, it
is the only approach that offers the ability to statistically combine evidence from different
investigations. Such synthesis is routinely done with data without explicit statistical handling
of experimental design.
A-17
-------
p. 67
A.3.2 Application Issues and Development Needs
As mentioned, the Bayesian approach is essentially the same method as the benchmark
approach up until the expression of the posterior distribution. Therefore, most of the issues
under Section A. 1.2 apply to the development of this approach as well.
In addition, application of the statistical synthesis capability of the approach will require
guidance development as well. Figure A-3 presents the statistical combination of data with
different endpoints: neurotoxicity (Sanagi et al., 1980) and respiratory tract effects (Dunnick
et al., 1989). The resultant posterior distribution for the combined evidence of different
endpoints was not drastically different relative to the individual distributions from which it
was derived. This may be due to the fact that both studies investigated very sensitive
endpoints (i.e., near the threshold or subthreshold region). Perhaps when data are not
comparable with respect to assayed endpoints, but the data represent very sensitive endpoints,
then the combination of these data provide a more likely estimate of the concentration of
concern. The data combined for Mn on the other hand, were all for the same specified
health effect (neurotoxicity). The exclusion of the Chandra et al. (1981) data was on the
basis of statistical considerations. Future development of this approach will have to develop
guidance on limitations for data combination both for statistical and biologically motivated
concerns.
A.4 CATEGORICAL REGRESSION: USE OF DOSE-GRADED DATA
Not all data are expressed as quantal or continuous data that are readily amenable to
available standard dose-response models. Results are often reported as "categorical" (i.e.,
descriptive or severity-graded results [e.g., a particular dose group exhibited "mild"
toxicity]). As mentioned in the advantages for the Bayesian approach, other studies that are
not explicitly designed to examine dose-response relationships, such as single-dose studies or
mechanistic studies, may nonetheless provide useful data that should be incorporated into the
data array analysis.
An analysis method that allows the combination of quantal data with categorical data
and models the relationship between the severity of the effect against the exposure
concentration and duration has been proposed for chronic oral toxicological data
A-18
-------
p. 68
(Hertzberg, 1989). Guth et al. (1991, 1993) have extended this work to inhalation exposures
and have proposed a regression analysis method that provides for incorporation of both
quantal and dose-graded data and for data across different durations. The method has been
proposed in order to utilize as much of the available data as possible for the evaluation of
short-term inhalation exposures defined as less than or equal to 24 h in duration.
A categorization scheme is used for the quantitative exposure-severity analysis, with
severity category as the dependent variable and with concentration and exposure duration as
independent variables. The severity scheme consists of three categories representing
NOAELs, adverse effect levels (AELs), and lethality. More complicated severity-ranking
schemes can be applied but become contentious due to the difficulty in equating severity of
effect measures across target organs, endpoints, and species (Guth et al., 1991; 1993).
The form of the model for regression analysis is
LN(p/l - p) = Aj + B1LN(Concentration) + B2(Duration), (A-4)
where p is the probability that, at a given concentration and duration of exposure, severity
will be less than or equal to the severity category with rank = i, and A and B are estimated
model parameters. The model is solved for P = 1 - p, or the probability that, at a given
concentration and duration, the severity will be greater than the severity category with
rank = i. The regression analysis assumes constant slope parameters, hence the values of
Bj and B2 are constant across severity categories. The order or rank of the categories is
used, rather than the numerical values.
The model output is readily interpreted in the context of risk assessment. Figure A-10
illustrates the method applied to categorical data for exposures of less than 8 h in duration
and shown as NOAELs, AELs, or lethality. Although longer exposure regimens are
appropriate as an alternate method to derive a NOAEL for the RfC, this example based on
acute data is offered. The maximum likelihood model fit is shown by the line representing
the model prediction of p = 0.1 that severity is greater than the NOAEL category (i.e., that
the predicted effect would be in the "adverse" range or higher) at the corresponding exposure
concentration and duration.
A-19
-------
p. 69
3 4
Exposure Duration (h)
Figure A-10. Categorical data from published results on methyl isocyanante for
exposures of less than 8 h in duration and shown as NOAEL (circles),
AEL (triangles), or lethality (squares). The maximum likelihood model
fit is shown by the line representing the model prediction (p = 0.1) that
severity is greater than the NOAEL category at the corresponding
exposure concentration and duration.
Source: Guth et al. (1993).
A.4.1 Approach Advantages
Health risk assessments generally require evaluation of several types of toxicity data
derived from several different species, different doses, different exposure durations, varied
endpoints, and varied quality. This variety often makes the health risk assessment extremely
difficult. Therefore, it is valuable to have all such toxicity data displayed simultaneously and
this approach offers the advantage of a graphic presentation. Exposure-duration response
trends, if present, are clearly delineated. This insight may provide a possible strategy for
disaggregation of data according to a duration window and/or for a particular endpoint.
This categorical analysis approach also offers the advantages of allowing the use of data
that is not otherwise amenable to quantitative concentration-response analysis, such as
categorical data and data from single-dose studies, and of incorporating both concentration
and duration of exposure as explanatory variables. Various types of data (dichotomous,
A-20
-------
p. 70
continuous, categorical) can be considered simultaneously by converting each to a categorical
descriptor. The estimates from this approach applied to short-term data have been shown to
be in general agreement with estimates obtained from both the benchmark and
NOAEL/LOAEL approaches (Guth et al., 1991; Guth et al., 1993).
The approach also offers the advantage of providing estimates for a range of exposure
durations. Interpolation along this NOAEL boundary can be performed to estimate the
NOAEL for any desired partial-lifetime exposure, rather than a linear prorate of the point
estimate value at one given duration as is currently done with many approaches. It should be
noted again that although the data shown here are truncated to exposures of less than 24-h
duration, data can be incorporated for any duration and have been applied to the entire data
sets on chemicals, regardless of duration (Dourson et al., 1986).
A.4.2 Application Issues and Development
Development of this approach requires guidance on model application, particularly
minimum data base requirements. For example, if data are too sparse or when the effect
levels are far apart, often the model will not converge. Figure A-11 shows the model fit to
the same data as in Figure A-10 but with the exclusion of lethality data. The presence of the
lethality data influences how the model addresses the boundary line between "adverse" and
"no-adverse" levels. It is also a question as to whether lethality data are appropriate to use
for dose-response assessment that intends to be protective of public health. When the data
are on one type of specified health effect (e.g., 2% carboxyhemoglobin in blood) in a single
species (humans), the model shows remarkable agreement with estimates generated by a
PBPK model for the same specified effect (Figure A-12). When an array of different
endpoints are available from a number of different species, as shown in Figure A-13, then the
choice of an endpoint may not be as straightforward. Therefore, the biological rationale for
model application also needs to be refined, especially on whether to aggregate or disaggregate
data on individual endpoints. If disaggregation results in convergence failure, then it could
be argued that this approach using all the available data provides a conservative estimate of a
NOAEL boundary and may be more certain than one derived from a single study. One
approach to disaggregation of data may be based on respiratory versus extrarespiratory effects
(and perhaps segregation of extrarespiratory endpoints) because it is likely that different
A-21
-------
p. 71
Duration (h)
Figure Aril- letliality data.
asinFigu*6
excluding
Source:
Dichloromethane, Human
COHb
""
A-22
-------
p. 72
3 4
Duration (h)
Figure A-13. Categorical regression analysis of tetrachloroethylene: acute effects.
Individual regression lines are based on model fit for all observations of
specified effects. Each point is an independent exposure group defined as
a specific concentration, duration, species, strain, and sex in a study.
Source: Guth et al. (1991).
mechanistic processes are involved for each of those types. As with the other alternative
approaches discussed, application of interspecies dosimetry adjustments and UFs for data
extrapolation are also warranted.
Development of this model application should also address the appropriateness of
combining data of different durations. For the RfC, subchronic and chronic data are of
interest to estimate "lifetime" effects. Consideration of temporal aspects of toxicity (see
Section 4.3.2) is required. The linearity of responses with exposures is often assumed, but
rarely investigated over "lifetime" bioassays.
A-23
-------
p. 73
APPENDIX B
CRITERIA FOR ASSESSING
THE QUALITY OF INDIVIDUAL
EPIDEMIOLOGICAL STUDIES1
Human data obviate the need for interspecies extrapolation and thus represent valuable
information to dose-response assessment. Scientific controversy sometimes surrounds the
interpretation and significance of results when the nature of the study was not experimental.
Guidelines for good epidemiology practices, documentation guidance, and guidance on
preparation of quality assurance studies for epidemiologic studies have been developed that
provide a surrogate to good laboratory practice standards aimed at laboratory animal studies.
These guidelines address the process of conducting epidemiologic studies in order to ensure
the quality and integrity of the data and to provide adequate documentation of the research
methods.
The criteria for assessing the quality of individual epidemiologic studies provided herein
are adapted from these guidelines and a number of sources. These criteria are intended to
serve as guidance on the evaluation of the quality of the practice with which the study was
conducted. These criteria fundamentally represent good scientific practice and thereby impart
an index as to the level of uncertainty when utilizing a particular study for dose-response
assessment. It is recognized that in some cases, information is not available to ascertain
whether all the criteria have been met, in which case judgment is necessary. For example,
the typical peer-reviewed journal article lacks some of the information provided in a detailed
study report.
1. The relationships, roles, and responsibilities of the organizations and/or individuals
sponsoring or conducting the study should be defined in writing. Sponsorship and
funding sources should be acknowledged.
1 Adapted from: Interagency Regulatory Liason Group (1981), Lebowitz (1983), American Thoracic Society
(1985), Pickrel et al. (1986), and Chemical Manufacturers Association's Epidemiology Task Group (1991).
B-l
-------
p. 74
2. A critical review of the relevant literature to evaluate applicable findings should be
provided. The review should encompass laboratory animal and human experiments,
clinical studies, vital statistics, and previous epidemiologic investigations. The review
should be sufficient to identify potential confounders and effect modifiers.
3. The objectives, specific aims, and rationale of the study should be clearly stated.
4 The overall research design, strategy, and rationale for choosing the proposed study
design should be described in relation to the objectives. Limitations of the study design
should also be stated. Underlying assumptions and limitations of the design also should
be given.
5. Clear definitions of health outcomes, exposure, other measured risk factors, and
selection criteria should be provided, as appropriate, for the study population and
comparison group (nonexposed and/or referent), morbidity and mortality cases. The
study population and comparison group description should include the specific
population from which they were drawn and the method of selection. The rationale and
criteria for inclusion or exclusion of participants in the study should be given,
particularly for exposure classifications. The appropriateness and limitations of the
comparison group should be discussed. The extent to which the choice of subjects
depended on existing or specially developed record systems, and implications of this
upon the analysis, should be considered. The steps taken to ensure confidentiality of
the subjects should be accounted for.
6. Data sources for exposure, health status, and risk factors should be described
(e.g., questionnaires, biological measurements, exposure/work history record reviews,
or exposure/disease registries). The limitations of these sources should be described.
7. Methods of data collection should be described in detail, because these procedures will
influence the derived interpretation and inferences. This should include a description
of, or reference to, methods used to control, measure, or reduce various forms of error
B-2
-------
p. 75
(e.g., bias due to misclassification, interviewer, or confounding factors) and their
impact on the study. The validity (accuracy) and reliability (reproducibility) of the
methods used to determine exposure should be stated. Response rates, including reasons
for implications of differing rates, should be given. The direction and possible
magnitude of any bias introduced into the study as a result of these rates should be
described. The procedures used for following the study, methods to ensure
completeness, and length of follow-up for each group or subgroup must be included.
Other validity checks (e.g., avoiding bias by the independent ascertainment and
classification of study variables, such as blind reading of histologic slides or clerical
processing of data) also should be included.
8. Major demographic and anthropometric confounding factors should have been accounted
for, such as age, sex, ethnic group, socioeconomic status, smoking status, and
occupational exposure. The methods employed for these adjustments and their
limitations should be discussed. Temperature, season, and day of the week are
particularly important for acute studies of respiratory effects and also should be
accounted for.
9. The procedures and statistical methods used to describe and analyze the data, estimate
parameters, or test specific hypotheses should be presented. References and/or specific
formulae also should be given for the statistical tests and for any programming
procedures or packages that were applied. The underlying assumptions and potential
bias of the statistical methods should be stated. Explicit description of any method used
to account for confounding factors (e.g., adjustment or matching) should be described
explicitly. This includes methods to account for missing data, such as from
nonresponse, attrition, or loss-to-follow-up. When reporting hypothesis tests, the
measure of effect, statistical significance, power, and other criteria (e.g., one- versus
two-tailed test rationale) should be given. Procedures for obtaining point estimates and
their standard errors and/or confidence intervals should be given when using estimation.
B-3
-------
p. 76
10. Criteria for interpreting results should be discussed, including the influence of the
limitations of the design, data sources, and analytic methods. Criteria for assessing
biologic plausibility, internal and external consistency of the findings, and causal
inference (see Appendix C) should be stated.
Often the detailed laboratory reports and documentation of studies are evaluated along
with peer-reviewed papers when evaluating data for derivation of an RfC. Quality assurance
and guidelines have been developed to ensure that essentially the same requirements provided
herein are met and these can be used to assess the quality and data integrity of completed
studies (Pickrel et al., 1986; Chemical Manufacturers Association, 1991). Each study should
have a written protocol that was approved before the study began. Data are usually
considered draft unless the final report has been signed. The following are suggested items
for inclusion in a written protocol that should accompany any formal report (Chemical
Manufacturers Association's Epidemiology Task Group, 1991).
A. Descriptive title.
B. The names, titles, degrees, addresses, and affiliations of the study director, principal
investigator, and all co-investigators.
C. The name(s) and address(es) of the sponsor(s).
D. An abstract of the protocol.
E. The proposed study tasks and milestones, including study approval date (date protocol
signed by all signatories), study start date (first date the protocol is implemented),
periodic progress review dates, and completion date.
F. A statement of research objectives, specific aims, and rationale (See criteria number 3
above).
G. A critical review of the relevant literature to evaluate applicable findings (See criteria
number 2 above).
H. A description of the research methods, including:
1. The overall research design, strategy, and rationale for choosing the proposed
study design.
2. The data sources for exposure, health status, and risk factors.
B-4
-------
p. 77
3. Clear definitions of health outcomes, exposure, and other measured risk factors as
well as selection criteria, as appropriate, for exposed and nonexposed persons,
morbidity or mortality cases, and referent groups.
4. Projected study size and, if appropriate, statistical power.
5. The methods to be used in assembling the study data.
6. Procedures for handling the data in the analysis.
7. Methods for data analysis.
8. Major limitations of the study design, data sources, and analytic methods.
9. Criteria for interpreting the results.
I. A description of plans for protecting human subjects.
J. A description of, or reference to, quality assurance and quality control procedures for all
phases of the study. As appropriate, include certification and/or qualifications of any
supporting laboratory or research groups.
K. A description of plans for disseminating and communicating study results.
L. Resources required to conduct the study.
M. The bibliographic references.
N. Addenda, as appropriate, including correspondence, collaborative agreements,
institutional approval, and samples of the informed consent forms, questionnaires, and
representative samples of other documents to be used in the study.
O. A dated protocol review and approval sign-off sheet for the study director, principal
investigator, co-investigators, and all reviewers.
P. Dated amendments to the protocol.
B-5
-------
p. 78
APPENDIX C
CRITERIA FOR
CAUSAL SIGNIFICANCE
Statistical methods cannot establish proof of a causal relationship but can define an
association with a certain probability. 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 toxicant and a health effect, a number of criteria must be used, no one
of which is pathognomonic by itself. These criteria include the following:
Consistency (reproducibility) of the association. Causal inferences are
strengthened when a variety of investigators have reproduced the findings
under a variety of circumstances.
Strength of the association. The larger the calculated relative risk, the
greater the likelihood that the observed association is causal.
Specificity of the association. Causality is more likely if a particular
exposure is associated with only one illness and vice versa. This guideline
rarely applies to air pollution research, in which all the diseases of major
concern are multifactorial.
Temporal relationship of the association.
Coherence of the association. An epidemiologic inference of causality is
greatly strengthed when it conforms to knowledge concerning the biologic
behavior of a toxicant and its mechanism of action. This evidence may be
obtained from clinical research or toxicologic studies.
Dose-response relationship.
C-l
-------
p. 79
APPENDIX D
ADVERSE HUMAN RESPIRATORY
HEALTH EFFECTS
These criteria were developed to assist in the interpretations of the epidemiologic
literature on what constitutes an adverse respiratory health effect of air pollution. Adverse
human health effects caused by air pollution are listed in hierarchical order, with the most
severe at the top and the least severe at the bottom. The reader is referred to the American
Thoracic Society (1982, 1985, 1986, 1993) guidelines, Epler et al. (1980), and Chan-Yeung
(1987) for more detailed discussion as to what constitutes respiratory impairment in humans
and to Appendix E for a discussion of pulmonary function testing data.
1. Increased mortality. ("Increased", as used here and subsequently, means significantly
[p < 0.05] increased above that recorded in some standard, comparable population.
In selected situations, p < 0.1 may be appropriate.)
2. Increased incidence of cancer.
3. Increased frequency of symptomatic asthmatic attacks.
4. Increased incidence of lower respiratory tract infections.
5. Increased exacerbations of disease in humans with chronic cardiopulmonary or other
disease that could be reflected in a variety of ways, including the following:
• Less able to cope with daily activities (i.e., shortness of breath or increased
anginal episodes);
• Increased hospitalizations, both frequency and duration;
• Increased emergency ward or physician visits;
• Increased pulmonary medication; and
• Decreased pulmonary function.
D-l
-------
p. 80
6. Reduction in forced expiratory volume at one second (FEVj) or forced vital capacity
(FVC) or other tests of pulmonary function such as the following:
• Chronic reduction in FEVj or FVC associated with clinical symptoms.
• A significant increase in number of persons with FEVj below normal
limits; chronically reduced FEVj is a predictor of increased risk of
mortality. Transient or reversible reductions that are not associated with an
asthmatic attack appear to be less important. It should be emphasized that a
small but statistically significant reduction in a population mean FEVj or
FEV0 75 is probably medically significant to them, but when diluted with
the rest of the population, the change appears to be small.
• An increased rate of decline in pulmonary function (FEVj), relative to
predicted value in adults with increasing age or failure of children to
maintain their predicted FEVj growth-curve. Such data must be
standardized for sex, race, height, and other demographic and
anthropometric factors.
7. Increased prevalence of wheezing in the chest, apart from colds, or of wheezing most
days or nights. (The significance of wheezing with colds needs more study and
evaluation.)
8. Increased prevalence or incidence of chest tightness.
9. Increased prevalence or incidence of cough/phlegm production requiring medical
attention.
10. Increased incidence of acute upper respiratory tract infections that interfere with normal
activity.
11. Acute upper respiratory tract infections that do not interfere with normal activity.
12. Eye, nose, and throat irritation that may interfere with normal activity (e.g., driving a
car) if severe.
13. Detection of odors.
D-2
-------
p. 81
APPENDIX E
GUIDANCE ON PULMONARY
FUNCTION TESTING
The two primary functions of the lung, oxygenation of mixed venous blood and removal
of carbon dioxide from that same blood, depend on the integrity of the airways, the vascular
system, and the alveolar septa. Inhaled toxic chemicals can affect the integrity of all three of
these components. Ideally, tests would be designed to assess the integrity and functional
relationships of these structures separately. However, because this is often difficult, many
pulmonary function tests evaluate the status of these structural components in an indirect, and
often overlapping, way. The myriad of tests include those of pulmonary ventilation,
mechanics, distribution, diffusion, and ventilation/blood flows.
During the last three decades, lung function tests have evolved from tools for
physiologic study to clinical tools widely used in assessing respiratory status. It has become
common to evaluate the results of lung function tests in terms of whether or not they are
considered to be within a "normal" range (i.e., represent an "adverse" effect or not). These
interpretations are increasingly becoming the basis of dose-response assessments. All clinical
measurements, including pulmonary function tests (PFT) are subject to (1) technical variation
related to instrument, procedure, observer, subject, and their interactions; (2) biologic
variation; and (3) variation caused by dysfunction or disease, the focus of interest for dose-
response assessment. Therefore, interpretation of PFT requires establishing the variation of
interest (the signal) and its relation to the other sources of variation (the noise).
To maximize the clinical value of lung function testing, the American Thoracic Society
(ATS) has outlined the steps necessary to achieve standardization: (1) equipment
performance, validation, and quality control; (2) subject performance; (3) measurement
procedures to determine acceptability and reproducibility; and (4) reference values and
interpretation. These steps form the basis of the criteria outlined here and can loosely be
applied to the evaluation of tests on both human and laboratory animals, although in most
instances, subject performance is not voluntary in the laboratory animals. Adherence to the
E-l
-------
p. 82
available guidance and recommendations discussed herein should help to ensure that changes
in lung function over time and from certain exposures can be correctly interpreted and
analyzed without reservation regarding their accuracy and quality.
For more detailed discussion on selection of reference values, interpretative strategies
and standardization approaches for pulmonary function testing in humans, the reader is
referred to American Thoracic Society (1979; 1987a,b; 1991), Gardner et al. (1986a,b,c),
McKay (1986), McKay and Lockey (1991), Folinsbee (1988), Clausen (1982) and Ruppel
(1979). This appendix discusses considerations affecting the evaluation of PFT performed on
human subjects. For a more detailed discussion as to what constitutes respiratory impairment
in humans, the reader is referred to Appendix D and to the American Thoracic Society
(1982, 1986, 1993) guidelines, Epler et al. (1980), and Chan-Yeung (1987). Although some
of the general concepts are applicable to laboratory animals, some of the procedures and
definitions of PFT for laboratory animals are different and these are highlighted at the end of
each section. For more detailed discussion of the interpretations and limitations of pulmonary
function testing in laboratory animals and their correlates to human PFT, the reader is
referred to Costa et al. (1992); Costa and Tepper (1988); Mauderly (1989), and Costa
(1985).
E.I GENERAL DEFINITIONS
This section provides the definitions of (1) the tests commonly used to evaluate
pulmonary function and (2) the basic ventilatory defects.
E.I.I Common Pulmonary Function Tests in Humans
Figure E-l is a diagrammatic representation of the various lung volumes and capacities
based on a typical spirogram and Table E-l provides the description, determination technique,
and significance of each in the context of possible diagnostic use. There are some causes for
changes in these tests (e.g., limitation of the movement of the diaphragm by pregnancy,
thoracic surgery, or neuromuscular disease) that are not addressed by these comments.
It should be recognized that this table is very general and any decision on the significance of
E-2
-------
p. 83
Maximal Inspiratory
Level
Resting Expiratory Level
Maximal Expiratory Level
Figure £-1. Lung volumes and capacities. Diagrammatic representation of various lung
compartments, based on a typical spirogram. TLC, total lung capacity;
VC, vital capacity; RV, residual volume; FRC, functional residual capacity;
1C, inspiratory capacity; VT, tidal volume; IRV, inspiratory reserve
volume; ERV, expiratory reserve volume. Shaded areas indicate
relationships between the subdivisions and relative sizes as compared to the
TLC. The resting expriatory level should be noted, since it remains more
stable than other identifiable points during repeated spirograms, hence is
used as a starting point for FRC determinations, etc.
Source: Ruppel (1979).
abnormality observed in any given study depends heavily on the circumstances under which
the testing was performed.
Pulmonary mechanics tests include the forced vital capacity (FVC), the forced
expiratory volume (FEVT) and the forced expiratory flow at 25 to 75% exhaled FVC
(FEF25_75%). All values should be expressed using volumes corrected to body conditions
(BTPS): normal body temperature (37 °C), ambient pressure (mm Hg) saturated with water
vapor.
E-3
-------
p. 84
TABLE E-l. DEFINITION OF VARIOUS PULMONARY FUNCTION
TEST VOLUMES AND CAPACITIES
Volume or
Capacity Description Technique Significance
TLC
VC
RV
1C
FRC
Total Lung Capacity. The Usually calculated by TLC is decreased in edema,
amount of air contained in the combination of other specific atelectasis, pulmonary
lungs at the end of a maximal lung volumes (e.g., FRC + 1C, congestion, and restrictive
inspiration. VC + RV). diseases. The TLC may be
normal or increased in
bronchiolar obstruction with
hyperinflation and in
emphysema.
Vital Capacity. The largest
volume measured on complete
expiration after the deepest
inspiration without forced or
rapid effort.
Vital capacity is measured from
maximal inspiration to maximal
expiration ("I-E") or maximal
expiration to maximal
inspiration ("E-I")1.
A decrease in VC may be
caused by a loss of distensible
lung tissue (e.g., bronchiolar
obstruction or pulmonary
congestion).
Residual volume. The volume RV must be measured indirectly Increases in RV are
of air remaining in the lungs
at the end of a maximal
expiration.
as a subdivision of the FRC,
using N2-washout (open-circuit)
method or tracer gas dilution
(closed circuit).
Inspiratory capacity. The
largest volume of air that can
be inspired from the resting
expiratory level.
characteristic of emphysema and
chronic air trapping, as well as
chronic bronchial obstruction.
RV is typically decreased in
restrictive diseases, particularly
those associated with extensive
fibrosis, such as sarcoidosis,
asbestosis, and silicosis.
RV may also be decreased in
diseases that occlude many
alveoli (e.g., pneumonia).
1C is measured by inhaling Changes in the absolute volume
maximally from the resting of 1C usually parallel increases
expiratory level or estimated by: or decreases in the VC.
VC — ERV. Compensatory hyperventilation
normally "dips into" the
inspiratory capacity because
both the end-inspiratory and
end-expiratory levels are altered.
Functional residual capacity.
The volume of air remaining
in the lungs at the end-
expiratory level.
SeeRV.
An increased FRC represents
hyperinflation that may result
from emphysematous changes,
asthmatic or fibrotic bronchiolar
obstruction. FRC is typically
decreased in restrictive diseases,
particularly those associated
with extensive fibrosis, such as
sarcoidosis, asbestosis, and
silicosis. FRC may also be
decreased in diseases that
occlude many alveoli (e.g.,
pneumonia).
E-4
-------
p. 85
TABLE E-l (cont'd). DEFINITION OF VARIOUS PULMONARY FUNCTION
TEST VOLUMES AND CAPACITIES
Volume or
Capacity Description Technique Significance
IRV Inspiratory reserve volume. IRV is measured by inhaling Changes parallel to those in VC.
The largest volume of air that maximally from the tidal
can be inhaled from the tidal inspiratory volume.
inspiratory volume.
VT Tidal volume. The volume of VT is measured directly by VT is not an adequate indicator
air inspired or expired during simple spirometry. The volume of alveolar ventilation and
each respiratory cycle. change is measured from the should not be considered outside
excursions of normal breathing, the context of rate and minute
Because no two breaths are volume.
identical, the VT inhaled or
exhaled should be measured for
at least 1 minute and then
divided by the rate to determine
the average.
ERV Expiratory reserve volume. ERV is measured by exhaling Changes parallel to those in VC.
The largest volume of air that maximally from the resting
can be expired from the end- expiratory level or estimated by:
expiratory level. VC — 1C.
'The closed circuit technique enables evaluation of whether a maximum inspiration was achieved prior to
expiration for the "I-E" maneuver (McKay and Lockey, 1991).
The FVC is the volume (liters) of air that can be exhaled as forcefully and rapidly as
possible after a maximal inspiration. The test's validity depends heavily on patient effort and
cooperation (see footnote to Table E-l). The FEVT is the volume of air exhaled over a
specified time interval (liters per seconds) during the performance of a FVC. The time
interval (in seconds) is stated as a subscript to FEV. An interval in common use is the
FEVj, the volume expired at 1 s. The FEF25_75% is the mean forced expiratory flow during
the middle half of the FVC, formerly called the maximal midexpiratory flow (MMEF).
Figure E-2 shows a typical volume-time curve (spirogram) for the FVC maneuver and
various FEVT are indicated.
Bronchial responsiveness is an integrated physiologic mechanism involving airway
epithelium, nerves, mediators, and bronchial smooth muscle. Bronchoprovocation challenge
testing (BPCT) involves evaluating the changes in FVC, FEVT, and FEVj/FVC ratio after
exposure to either specific or nonspecific agents capable of producing bronchoconstriction.
E-5
-------
p. 86
£2-
1
FVC
FEVa
EV
0.05
FEVa
Time (s)
8
9
10
Figure E-2. Volume-time plot (spirogram) of the forced vital capacity (FVC) maneuver.
The subject exhaled as forcefully and rapidly as possible from maximal
inspiratory level. Forced expiratory volume (FEV) as various time intervals
are indicated.
Parasympathomimetic drugs, such as methacholine and carbachol, are used as nonspecific
agonists because they cause bronchoconstriction by stimulating acetylcholine receptors located
directly on airway smooth muscle. Histamine is another commonly used bronchoconstricting
agent. Although its mechanism of action is somewhat controversial, it probably acts
indirectly by stimulating cholinergic nerve endings as well as having a direct effect via
histamine receptors on airway smooth muscle. Specific agents include common antigens or
chemicals such as the isocyanates that may provoke immediate, delayed or dual pulmonary
responses that may not resolve spontaneously. Guidelines for standardization have been
developed for bronchial inhalation challenges with the nonspecific agonists such as
methacholine (Cropp et al., 1980) and adherence to these guidelines should be considered
when evaluating such data. Bronchoprovocation challenge testing (BCPT) with specific
agents requires more time, expense, and sophisticated equipment and remains more in the
realm of research than does nonspecific BCPT, but can also be an extremely useful diagnostic
E-6
-------
p. 87
aid when performed by a quality laboratory. Factors influencing agent-specific BPCT have
been discussed elsewhere (McKay, 1986).
Response to bronchodilating agents may also be measured. The within-individual
difference in response to different bronchodilators is variable. Because the correlation
between bronchoconstriction and bronchodilator response is imperfect, it is not possible to
infer with certainty the presence of one from the other. There is no clear consensus on what
constitutes reversibility in subjects with airflow obstruction (American Thoracic Society,
1991), however, 20% reversibility is generally believed to be consistent with asthma.
Carbon monoxide diffusing capacity (DLCO) measures all the factors that affect the
diffusion of a gas across the alveolo-capillary membrane. Traditional units are mL
CO/min/mm Hg at STPD (standard conditions: 0 °C, barometric pressure of 760 mm Hg,
0 mm Hg water pressure). Steady-state or rebreathing techniques are commonly used for
human testing. But the single-breath technique (DLCO^) is also commonly used. In general,
DLCO is decreased in alveolar fibrosis (e.g., as associated with asbestosis or berylliosis) or
interstitial edema. Carbon monoxide diffusing capacity is also decreased in emphysema
because of the decrease in alveolar surface area, loss of capillary bed, increased distance from
the terminal bronchiole to the alveolocapillary membrane, and the mismatching of ventilation
and blood flow. Guidance on standardization has been published elsewhere (American
Thoracic Society, 1987b).
The nitrogen washout test measures the concentration of nitrogen in alveolar gas at the
end of breathing 100% oxygen for a prescribed period of time (e.g., 7 min). The value is
recorded as a percentage of nitrogen. The test is used to determine lung volumes (e.g., the
FRC and RV). The FRC and RV are often increased in diseases in which there is an
increased airway resistance such as emphysema, chronic bronchitis, and asthma. The RV is
raised in these conditions chiefly because airway closure occurs at an abnormally high lung
volume. A reduced FRC and RV are often seen in conditions of reduced lung compliance,
for example, in diffuse interstitial fibrosis. In this case, the lung is "stiff" and tends to recoil
to a smaller RV.
E-7
-------
p. 88
£.1.1.1 Common Pulmonary Function Tests in Laboratory Animals
The conceptual framework for analysis of pulmonary function is quite similar for
laboratory animals with the following noteworthy differences:
1. The relevance of the various ATS guidelines is questionable since in many
instances they specifically define terms, procedures, and equipment only as they
relate to humans. The reviews of Costa (1985), Costa and Tepper (1988),
Mauderly (1989), and Costa et al. (1992) put most of the information presented
here for the human in the context of small laboratory animals with the various
caveats and limitations. The reader is referred to these reviews for more detail
than can be provided in these guidelines. For example, if the animal test is done
under anesthesia, its body temperature will fall. Unless this measure is
monitored and used in the computation of the BTPS adjusted measure (assumed
to be 37 °C), the data can differ between studies. Many investigators use actual
body temperature (approximately 35 °C) as the BTPS basis so that temperature
need not be monitored.
2. The tests described for humans generally require the use of nomogram or other
standardized tables based on sex, age, height, and weight of the test subject to
compare and determine "normalcy". Laboratory animal studies almost always
require the use of comparable control groups (based on the same
specio-promorphic considerations) against which determination of effect is
established. Anomaly or effect is based on statistical grounds for the group and
rarely for the individual (except as part of the overall interpretation).
3. Measures of maximal lung volume in laboratory animals are determined by
imposed pressures derived from allometric evaluations of cross-species data, not
effort. Since these animal measures are determined under anesthesia, volition is
eliminated and the static mechanics of the system can be established. The forced
vital capacity measure in the rodent is created differently from that of the human.
The interpretation is similar, but reductions in these species respective volumes
can differ because of pain in the human maneuver (e.g., after ozone exposure)
which the animal will not feel. Nuances can be important. Similarly, the
measure of FRC in laboratory animals and humans is different based ont he
mechanisms establishing this volume. In the human, the volume is based on
apposed recoil of the lung and chest wall. In animals with compliant chests this
is not the case. Rather, it is set by breathing mechanics and central expiratory
control (turned off during anesthesia). Hence, true comparison is difficult.
In fact, because the rodent lung has the ability to in part regenerate after acute
injury, the FRC response may be the reverse of that of the human (larger than
normal instead of smaller). The DLCO can respond in much the same manner.
4. Airway reactivity in the rodent is measured in many different ways. Almost
none of the these directly parallels the human but the overall interpretations are
the same. However, there can be toxicant differential effects in animals when the
agonist is delivered to the lung directly versus intravenously. It should be noted
E-8
-------
p. 89
that even within the human study community that the methods used for the testing
of airway reactivity differ significantly from laboratory to laboratory.
Standardization is more commonplace in true clinical test laboratories than in
empirical-clinical study laboratories.
E.I.2 Interpretation of Pulmonary Function Tests and Basic Ventilatory
Defects
The vital capacity (VC), FEV1? and FEVyvC ratio are the basic parameters used to
interpret spirometry. Although FVC is often used in place of VC, it is preferable to use the
largest VC, whether obtained on inspiration (IVC), slow expiration (EVC), or forced
expiration (FVC) for clinical testing. Limiting primary interpretation of spirometry to three
variables avoids the problem of simultaneously examining a multitude of measurements to see
if any abnormalities are present, a procedure that will lead to an inordinate number of
"abnormal" tests (American Thoracic Society, 1991). As discussed in other sections of this
appendix, the first step in interpreting lung function data should be the evaluation of the
quality of the testing. Further, tests interpreted without additional clinical information are
limited in their utility to be definitive. Consideration must also be given to (1) the level of
reporting and control of technical variation (Section E.2) and (2) the selection of reference
values and statistical techniques used to generate predictive values that may be used for
interpretation (Section E.4).
E.l.2.1 Definition of an Obstructive Defect
An obstructive ventilatory defect may be defined as a disproportionate reduction of
maximal airflow from the lung with respect to the maximal volume (VC) that can be
displaced from the lung. It indicates airflow limitation and implies airway narrowing during
expiration. The earliest change associated with flow limitation in small airways is thought to
be slowing in the terminal portion of the spirogram even when the initial phase is unaffected.
This slowing is reflected in a proportionally greater reduction in the instantaneous flow
measured after 75% of the FVC has been exhaled (FEF75%) or in the FEF25_75%, than in the
FEVj. Abnormalities in these midrange flow measurements during a forced exhalation are,
however, not specific for small airway disease and, though suggestive, should not be used to
diagnose small airway disease in individual patients. As airway disease becomes more
E-9
-------
p. 90
advanced and/or more proximal airways become involved, earlier time segments of the forced
expiratory maneuver such as the FEVj will become reduced out of proportion to the
reduction in the VC (American Thoracic Society, 1991).
The FEVj/VC is recommended as the primary test for distinguishing obstructive from
nonobstructive patterns. The FEF25_75% may be used to confirm the presence of airway
obstruction in the presence of a borderline FEVj/VC. The severity of airway obstruction
should be based on the FEVj rather than the FEV^VC (American Thoracic Society, 1991).
£.1.2.2 Definition of a Restrictive Defect
A restrictive veritilatory defect is characterized physiologically by a reduction in TLC.
The presence of a restrictive ventilatory defect is inferred when VC is reduced and the
FEVj/FVC is normal or increased. However, severe airflow limitation is another common
cause of a reduced VC, either because airflow is so slow the subject can not continue to
exhale long enough to complete emptying or because airways collapse. Also, a small VC
with a normal FEVj/VC will occasionally be observed in patients with a normal TLC. Thus,
if there is a contradiction between VC and TLC in defining restriction, the classification
should be based on the TLC (American Thoracic Society, 1991).
£.1.2.3 Interpretation of Laboratory Animal Tests
A notable difference for interpretation of laboratory animal pulmonary function testing
is that these studies typically require cohort control groups because of the many influences on
the animal that can not be standardized in textbook nomograms. The reader is referred to the
reviews of Costa (1985), Costa and Tepper (1988), Mauderly (1989), and Costa et al. (1992)
for important distinctions from human clinical interpretations.
E.2 TECHNICAL SOURCES OF VARIATION (INSTRUMENTATION)
Maximizing the usefulness of spirometry for clinical, diagnostic, or epidemiologic
purposes depends on a number of factors that begins with equipment selection. Because
spirometry involves effort-dependent maneuvers that require careful patient/subject
instruction, understanding, coordination, and cooperation, performance recommendations are
E-10
-------
p. 91
also an important component of ensuring accurate testing. This section discusses specific
guidance available on testing equipment, testing performance, quality control, and technician
training. This guidance is intended to serve as a framework by which to evaluate the level of
certainty in the use of reported spirometry data.
E.2.1 Equipment
Measurement of the deterioration of pulmonary function as an effect of exposure to a
toxic chemical may be erroneous if inaccurate spirometers (or other instrumentation) or less
sensitive if imprecise spirometers are used. Thus, equipment selection and maintenance is
pivotal to ensuring accurate test results. The accuracy of a spirometer systems depends on
the resolution (i.e., the minimal detectable volume or flow) and linearity of the entire
system—from volume or flow transducer to recorder, display, or processor. Studies should
state that the equipment was validated as meeting ATS recommendations. Mention should
also be made that equipment quality control procedures were routinely performed, including
preventive maintenance, calibration checks, verification and that a quality assurance program
was in place to ensure accurate spirometry and test results (American Thoracic Society, 1991;
Gardner et al., 1986a,b). Attention must be given to the spirometer temperature where the
tests are performed and values reported in BTPS. Quality control should at least include
strict adherence to ATS guidelines for equipment performance and calibration (American
Thoracic Society, 1991) and additional equipment recommendations have been made by
McKay and Lockey (1991).
Measurement procedures have been recommended to ensure that uniform methods are
used and that comparable results are obtained (American Thoracic Society, 1991). Medical
surveillance and epidemiological studies may require more stringent guidelines to ensure the
higher level of quality needed to detect changes from one year to another (McKay and
Lockey, 1991). Measurement procedures include how to perform specific maneuvers and
thus also define equipment requirements as well. For example, if a test procedure should be
carried out for at least a specified amount of time, the spirometer should at a minimum be
able to compile data for that duration. Other spirometry system recommendations related to
performance procedures include specifications on volume range and accuracy, flow range,
resistance and back pressure, time scale (paper speed), volume scale, flow:volume scale,
E-ll
-------
p. 92
display axes orientation, and the type of signal used to test the performance for a given
maneuver.
E.2.2 Procedure Performance and Measurement
Performance recommendations are an important component of testing because PFT
involves effort-dependent maneuvers that require careful patient/subject instruction,
understanding, coordination, and cooperation. The largest single source of within-subject
variability is improper performance of the maneuvers (American Thoracic Society, 1991).
The performance recommendations involve obtaining a sufficient number of maneuvers that
are of adequate quality and then determination as to whether these acceptable maneuvers are
reproducible. Once maneuvers have been performed, measurement procedures are included
to help ensure that uniform methods are used and that comparable results are obtained.
Interpretations of spirometry should include a statement about test quality before any other
interpretation is rendered.
Guidance on how to perform specific maneuvers (i.e., the VC, FVC, FEVT, and
FEF25_75%) include recommendations on satisfactory start of test criteria, end of test criteria,
subject instruction, minimum maneuver time, maximum number of maneuvers, acceptability
criteria, use of noseclips, sitting versus standing position, reproducibility criteria, test result
selection, and result reporting. If a study does not explicitly state in the methods section that
ATS-recommended procedures were performed, the description of the methods for the
maneuvers should be compared against the available recommendations (American Thoracic
Society, 1991; McKay and Lockey, 1991) to ascertain their credibility.
Proper training of persons administering PFT is the single most important component of
a respiratory surveillance program (McKay and Lockey, 1991). Spirometry is not a set of
simple procedures to be performed by untrained or minimally trained individuals. The
persons administering PFT must do so with skill and understanding. The technician must be
able to (1) adequately prepare the subject for testing; (2) identify any preexisting
contraindications or reasons to postpone testing; (3) properly instruct, demonstrate, and coach
the subject regarding proper technique; and (4) visually inspect each maneuver tracing for
validity. The technician must be able to correct and adjust technical problems that may occur
and be capable of responding to questions that may arise. The technician must also be
E-12
-------
p. 93
capable of accurately performing hand measurements and calculation and be able to confirm
the adequacy of software used for automated calculations. This person should also be able to
interpret tests and recognize the effect submaximal effort has on the interpretation process.
Studies should state that qualified personnel were used and that a program was in place to
evaluate the personnel periodically in order to ensure that accurate and reliable test results
were obtained. Testing of commercially available spirometers showed that a major source of
errors was in computer software. Due to the increased use of automated systems and
computers in pulmonary laboratories, the ATS published "Computer Guidelines for
Pulmonary Laboratories" (Gardner et al., 1986c).
E.2.3 Technical Sources of Variation in Laboratory Animal Testing
Throughout this section, application to animal testing requires special considerations.
The most notable are:
1. rapid responding plethysmographic and transducing equipment is required since
most of the measures are in the 1-15 mL volume range and the flows perhaps as
high as 150 mL/s with a response time of 40 ms or better, and
2. laboratory animals are typically anesthetized and orally or surgically (via the
larynx) tracheotomized so that the nose and mouth have no influence.
E.3 BIOLOGICAL SOURCES OF VARIATION
This section outlines sources of variation in PFT related to individual performance on
the tests or to host factors, including environmental factors, of the individual tested. The
factors are provided here for readers to be aware of as factors that should be controlled for
(when possible) in studies that use PFT to index respiratory dysfunction. Some of these
factors are explicitly incorporated in algorithms available to calculate normal values for
various maneuvers and others are not (see Section E.4). Recommendations to control for
some of these (e.g., recommended body position for most maneuvers) have been made to
ensure consistency (American Thoracic Society, 1991).
The main sources of intraindividual variation of PFT are (1) body position, (2) head
position, (3) effort dependency of maximal flows, and (4) circadian rhythms. The study
design should include procedures that ensure consistency relative to these four factors.
E-13
-------
p. 94
The most important host factors that are responsible for interindividual variation in PFT
are (1) sex, (2) size, and (3) age, which account for 30, 22, and 8%, respectively, of the
variation between adults. Growth affects the relationship between indices of body size and
spirometric measurements in children and adolescents. The relationship of ventilatory
function to height from childhood through late adolescence to adulthood is not linear.
Different prediction equation should be used for the sexes at all ages. Other sources of
interindividual variation include (1) race and (2) past and present health.
Exposure to tobacco smoke is by far the most important factor known to alter lung
function. A clear choice for the most appropriate method of adjusting spirometric indices for
the effect of smoking is not readily evident from published data in which any of the following
have been used: smoking status (current smoker or exsmoker), amount currently smoked,
duration of smoking, and pack-years. Neglecting the correlation of some of these factors,
(e.g., pack-years) with age can introduce errors in analyzing the effects of smoking.
Smoking should be handled as an independent variable as its distribution in the reference
population and its relation to other health indicators will affect any predictive regression
terms calculated. Other environmental factors that contribute to interindividual variation
include (1) geographic factors, (2) exposure to environmental and occupational pollution, and
(3) socioeconomic status.
The reader is referred to the reviews of Costa (1985), Costa and Tepper (1988),
Mauderly (1989), and Costa et al. (1992) for specific considerations in laboratory animals.
E.4 REFERENCE VALUES: SOURCES, SELECTION AND
STATISTICAL ISSUES1
Predicting the presence or absence of disease requires knowledge about the distribution
of dysfunction in various disease states and the prior probability of disease. Subjects with
similar characteristics for the major variables that affect lung function (sex, age, height, and
race) can be grouped together in a stratum or a cell. Comparing the performance of an
individual subject with the values generated from a reference population requires knowledge
about the data in the appropriate cell (i.e., the number in the cell, measures of central
'Text adapted from American Thoracic Society (1991). Reader is referred to these guidelines for additional detail.
E-14
-------
p. 95
tendency such as the mean value, estimates of dispersion such as variance or standard
deviation [SD], and information about the symmetry of the distribution). If the number of
subjects in each cell is sufficient, PFT can be described by providing descriptors of the
distribution such as mean and SD. Such tabulations are infrequently used for PFT because
there are too many possible cells to consider all possible combinations of age and height.
Regression equations are an economical and efficient alternative method to describe
expected values as a function of sex, height, and age. Regression techniques assume that
PFT varies in a symmetric fashion about the mean value in each cell and that the variance
about the mean is constant from one cell to another. The closer the distribution of PFT
values come to symmetry or, better still, to a Gaussian distribution within cells, the more it is
possible to take advantage of the equations. Distributions of FEVj and FVC in population
studies are usually found to be close to Gaussian in the middle age range, not at the
extremes. Ideally, publications describing reference populations should include, not only the
prediction equations, but also a means of defining their lower limits. In the absence of
explicit recommendations, a lower limit can be estimated from a regression model. For
spirometry, values below the fifth percentile are taken as below the expected range (below the
"lower limit of normal") and those above the fifth percentile are taken as within the expected
range. This implies a 5 % false positive misclassification, a rate generally considered
acceptable.
The most commonly reported measures of how well regression equations fit the data are
the square of the correlation coefficient (r2) and the standard error of the estimate (SEE).
The proportion of variation in the observed data explained by the independent variables is
measured by r2. The SEE is the average SD of the data around the regression line. Because
these two statistics reflect average characteristics of the regression, r2 and SEE may not
reflect the ability of the equation to describe the tails of the distribution or the limits of
"normal", and therefore are not sufficient criteria on which to choose the best equations to
evaluate a population.
Linear regression is the most common but not the only model used to describe PFT data
in adults. Such equations perform less well at the edges of the data distribution and in those
cells where there are few data. Estimates are likely to be misleading if they go beyond the
range of the independent variables used to create the equation.
E-15
-------
p. 96
Criteria for selecting reference values to be used fall into three categories:
(1) methodologic, (2) epidemiologic, and (3) statistical. Reference values should be based on
data obtained with the same instruments and methods comparable to those used for the
population for which the reference values are being selected. The population from which the
subjects are drawn should be similar with respect to age, height, sex, and ethnic composition
to the population to whom the prediction values are to be applied. Prediction equations
should use age, height, sex, and ethnic group as independent variables. For most uses, they
should be based on cross-sectional studies of lifetime nonsmokers. Both biologic plausibility
and simplicity in the model used to develop prediction equations are important issues in the
selection of reference values. Other statistical aspects have been described above. Selected
published reference equations for adult whites and blacks and scaling factors for blacks
currently in use have been published. Studies should use the published reference equations
that most closely describe the population being tested.
The practice in many clinical laboratories has been to classify values of FVC and FEVj
less than 80% of predicted as abnormal. This fixed value has no statistical basis in adults.
Cross-sectional data are subject to a bias called "cohort" effect. A person who is
40 years of age today is different from one who became 40 two decades ago because of a
variety of host and environmental factors. The age-related lung function deficit predicted
from cross-sectional data tends to be greater than that predicted from longitudinal PFT data in
adults and children. Prediction equations based on cross-sectional data are appropriate for
determining the prevalence of PFT impairment in defined populations. They are less well-
suited to determine age-related events including the incidence or progression of impairment.
Reliance can be placed on the FEVj and VC for examining changes over time as they are the
only spirometric variables that will consistently and correctly reflect the direction of the
change in overall PFT. Difficulty remains, however, in determining whether a change is
"real" or only a result of test variability. All PFT measurements tend to be more variable
when made weeks to months apart than when repeated at the same session or even daily.
It is more likely that a real change has occurred when there are a series of tests that show a
consistent trend. As shown in Table E-2, significant changes, whether statistical or biologic,
vary by parameter, time period, and the type of patient.
E-16
-------
p. 97
TABLE E-2. CHANGE IN SPIROMETRIC INDICES OVER TIME
Percent Changes Required To Be Significant
FVC FEVj FEF25.75%
Within a day
Normal subjects >5 >5 >13
Patients with COPD >11 >13 >23
Week to week
Normal subjects >11 >12 >21
Patients with COPD >20 >20 >30
Year to year
E.4.1. Reference Values for Laboratory Animal Testing
The discussion above generally applies to laboratory animal studies with the exception
noted above that the study design should include empirical control cohorts. Considerations
for establishing such controlled studies are presented in the reviews of Costa (1985), Costa
and Tepper (1988), Mauderly (1989), and Costa and Tepper (1992).
E.5 INTERPRETIVE STRATEGIES: CONCEPTUAL ISSUES
CONCERNING NORMALITY AND THE LIMITS OF NORMAL
FOR DESIGNATING ADVERSE-EFFECT LEVELS
To draw inferences about the presence of disease from one test, the prior probability
that the patient has the disease and the distributions of test values for subjects with and
without the disease in question should ideally be known. Although this ideal is rarely met,
understanding of the testing situation should be used to put an interpretation of PFT in proper
perspective. The "normal" range only gives information about the distribution of test results
in the healthy population from which they were derived. It says nothing about the true
positive rate, the false negative rate, or the predictive power of a positive test.
As discussed in the preceding sections, consideration must be made of the
appropriateness of the equipment, performance maneuvers, biologic variation and selection,
including statistical procedures, used to derive normal reference values. In summary, studies
E-17
-------
p. 98
should indicate in the methods section the source of reference values used for their reports.
Prediction equations for adults should include age, sex, and height as independent variables.
It is preferable to choose reference values for both sexes from the same population source.
Smoking status as an independent variable has been discussed in Section E.3. Altitude can be
important in the selection of reference values for flow rates and for DLCO. The equations
should come from studies that present lower limits of normal or present information from
which such lower limits can be calculated. In general, the prediction equations should not be
extrapolated for ages or heights beyond those covered by the data on which they are based.
The use of 80% of predicted for a lower limit of normal for adult PFT maneuvers is not
recommended. Because of unexplained differences between published reference values, no
one set of reference values is likely to be applicable to all studies performed. It is preferable
that studies performed on populations in North America use reference values based on North
American populations. European studies should use reference values based on European
populations.
If there are any reasons to suspect the quality of the test performance, specific
designation of adverse effect levels should be avoided. Dysfunction discovered under these
conditions should indicate only the need for more definitive testing. General definitions of
respiratory dysfunction are provided in Section E.I.2.2, but determination of the severity or
degree of dysfunction must be made in the context of the other considerations discussed
above, particularly the appropriateness of the reference values and statistical procedures used
to describe "normal". Finally, borderline "normal" values should be interpreted with caution.
Such interpretations should, when possible, use additional clinical information in the decisions
in order to designate an adverse-effect level or a no-adverse-effect level.
E.S.I Interpretive Strategies for Animal Testing
Again, the concepts outlined above generally apply to animal testing with a few notable
differences. Although spirometric measures in animals appear to be consistent over time, no
real investigation of this has been conducted. It should be pointed out, however, that most
rodents grow throughout life and their age dependent spirometry appears to improve (by
anthropomorphic standards) over the same period until just before death. This is quite unlike
the huan which begins to have less than optimal performance beyond young adulthood
E-18
-------
p. 99
(around 21 years of age). The DLCO in rodents also improves over most of life but begins to
diminish before the fall in spirometry. This is not the case in humans.
E-19
-------
p. 100
APPENDIX F
CRITERIA FOR ASSESSING THE
QUALITY OF INDIVIDUAL LABORATORY
ANIMAL TOXICITY STUDIES1
A minimally acceptable study should meet the following criteria, which fundamentally
represent good scientific practice.
1. All elements of exposure should be clearly defined.
• The exposure concentration, administration route, exposure schedule, and
exposure duration must be described. Consideration should also be given to
the concentration and time of exposure used versus the expected level of
human exposure.
• If animal body weights, ages, or sex are not provided, consideration should
be given to the uncertainty in appropriate default values.
• Exposure information should include physicochemical characteristics of the
substance used, such as purity, stability, pH, partition coefficient, particle
size and distribution, breathing zone concentration, and vehicle. These
properties can influence the local effects and the rate and extent of
absorption, which can subsequently modify the toxic manifestations.
Concentrations should be reported as means and variances.
• Exposure information should include a description of generation and
characterization technology used (e.g., chamber design, type, dimensions,
uniformity of distribution, source of air, generating system, air
conditioning, and exhaust treatment). The number of air changes, air flow
rate, oxygen content, temperature, and relative humidity are exposure
chamber characteristics that should be monitored and reported as means and
variances. The description of the characterization method(s) should also
include frequency of measurement, calibration of the measurement
instrument, frequency of the calibration, and other quality assurance
elements. Cage (or other animal holder) rotation schedule should be
described.
Adapted from Society of Toxicology (1982), Muller et al. (1984), National Research Council (1984), James
(1985), and Lu (1985a).
F-l
-------
p.1
• Animal care and holding procedures should be described.
2. Controls should be comparable with test animals in all respects except the treatment
variable ("negative").
• Concurrent controls must minimally include an "air-only" exposure group;
if a vehicle is used, it is desirable that there be a "vehicle-only" group.
• Historical control data can be useful in the evaluation of results, particularly
where the differences between control and treated animals are small and are
within anticipated incidences based on examination of historical control
data.
3. Endpoints should address the specific hypothesis being tested in the study, and the
observed effects should be sufficient in number or degree (severity) to establish a dose-
response relationship that can be used in estimating the hazard to the target species.
• The outcome of the reported experiment should be dependent on the test
conditions and not influenced by competing toxicities.
4. The test performed must be valid and relevant to human extrapolation. The validity of
using the test to mimic human responses must always be assessed. Issues to consider
include the following:
• Does the test measure an established endpoint of toxicity or does it measure
a marker purported to indicate an eventual change (i.e., severity of the
lesion)?
• Does the test indicate causality or merely suggest a chance correlation?
• Was an unproven or unvalidated procedure used?
• Is the test considered more or less reliable than other tests for that endpoint?
• Is the species a relevant or reliable human surrogate? If this test conflicts
with data in other species, can a reason for the discrepancy be discerned?
• How reliable is high exposure (animal) data for extrapolation to low
exposure (human scenario)?
F-2
-------
p. 2
5. Conclusions from the experiment should be justified by the data included in the report
and consistent with the current scientific understanding of the test, the endpoint of
toxicology being tested, and the suspected mechanism of toxic action.
6. Due consideration in both the design and the interpretation of studies must be given for
appropriate statistical analysis of the data.
• Statistical tests for significance should be performed only on those
experimental units that have been randomized (some exceptions include
weight-matching) among the dosed and concurrent control groups.
• Some frequent violations of statistical assumptions in toxicity testing
include:
Lack of independence of observations.
Assuming a higher level of measurement than available (e.g., interval
rather than ordinal).
Inappropriate type of distribution assumed.
Faulty specification of model (i.e., linear rather than nonlinear).
Heterogeneity of variance or covariance.
Large Type II error.
7. Subjective elements in scoring should be minimized. Quantitative grading of an effect
should be used whenever possible.
8. Peer review of scientific papers and of reports is extremely desirable and increases
confidence in the adequacy of the work.
9. When the data are not published in the peer-reviewed literature, evidence of adherence
to good laboratory practices is highly recommended, with rare exceptions.
10. Reported results have increased credibility if they are reproduced (confirmed) by other
researchers and supported by findings in other investigations.
F-3
-------
p. 3
11. Similarity of results to those of tests conducted on structurally related compounds should
be considered.
The reader is also referred to Part 798, the Health Effects Testing Guidelines, of the
U.S. Toxic Substances Control Act Test Guidelines delineated in 40 Code of Federal
Regulations (199Id). The chronic testing guidelines for all administration routes are provided
in Subpart D, Section 798.3260. Subpart C, Section 798.2450, and Subpart B, Section
798.1150, describe the guidelines for subchronic and acute inhalation testing, respectively.
Guidelines for inhalation developmental toxicity testing are discussed in Subpart E,
Section 798.4350.
These guidelines provide recommendations on laboratory animal selection (e.g., species,
number, sex, age, and condition); on number of test concentrations and the objectives of
each; on physical parameters of exposure that should be monitored and recorded and with
what frequency (e.g., chamber air changes, oxygen content, air flow rate, humidity, and
temperature); on what testing conditions should be reported and how (e.g., mean and
variance of both nominal and breathing zone exposure concentration, particle size, and
geometric standard deviation); and on what gross pathology and histopathology, clinical,
biochemical, hematological, ophthalmological, and urinary excretion tests to perform,
intervals at which to perform them, which exposure levels to process these data for, and how
to report their results.
The recommendations on what diagnostic tests to perform and how to report the data
are particularly useful for evaluating a given study. Although the mechanism of action
should dictate the repertoire of tests performed, Table F-l provides a general list of
recommended clinical biochemistry examinations; and Table F-2 provides a list of organs and
tissues recommended for histological examination. If specific mechanisms of action are
hypothesized, specific assays or functional tests for those would be added. It is also
important to establish that appropriate removal and tissue processing was performed.
Results should be reported in tabular form, showing the number of animals at test start,
number with lesions, the types of lesions, and the percentage of animals with each type.
Group animal data should be reported to show number of animals dying, number showing
signs of toxicity, and number exposed. Individual animal data should include time of death;
F-4
-------
p. 4
TABLE F-l. GENERAL CLINICAL BIOCHEMISTRY EXAMINATIONS
Calcium
Phosphorus
Chloride
Sodium
Potassium
Fasting glucose
Serum glutamic-pyruvic transaminase (serum alanine aminotransferase)
Serum glutamic-oxaloacetic transaminase (serum aspartate aminotransferase)
Ornithine decarboxylase
Gamma glutamyl transpeptidase
Urea nitrogen
Albumin
Blood creatinine
Creatinine phosphokinasea
Total cholesterol
Total bilirubin
Total serum protein
Lipidsb
Hormonesb
Acid/base balance^
Methemoglobinb
Cholinesterase activityb
"Suggested for chronic inhalation toxicity test.
bMay be required for a complete toxicological evaluation.
Source: Shoaf (1993).
time of observed toxicity; body weight; food consumption; and results of hematological tests,
clinical biochemistry tests, necropsy, histopathology, and statistical analyses.
F-5
-------
p. 5
TABLE F-2. ORGANS AND TISSUES PRESERVED FOR
HISTOLOGICAL EXAMINATION
All gross lesions
Nasopharygeal tissues
Lungs3
Trachea
Pituitary
Thyroid/parathyroid
Thymus
Brain and sectionsb
Heart
Sternum with bone marrow
Salivary glands
Liver
Spleen
Kidney
Adrenals
Pancreas
Gonads
Uterus
Accessory genital organs0'6
Aorta
Gall bladder
Esophagus
Stomach
Duodenum
Jejunum
Ileum
Cecum
Colon
Rectum
Urinary bladder
Representative lymph node
Peripheral nerve
Thigh muscle0
Mammary glandc
Eyesc
Skinc
Spinal cordC)d
Exorbital lachrymal glands0
"Removed intact, weighed, and treated with fixative (e.g., perfusion) to ensure maintenance of lung structure.
bMedulla/pons, cerebellar cortex, and cerebral cortex.
°If indicated by signs of toxicity or as a target organ.
dCervical, mid thoracic, and lumbar.
"Epididymis, prostate, seminal vesicles.
Source: Shoaf(1993).
F-6
-------
p. 6
APPENDIX G
THE PARTICLE DEPOSITION
DOSEMETRY MODEL
In this appendix, the revised empirical model used to estimate fractional regional
deposition efficiency for calculation of RDDR (Equation 4-5) to be used as a dosimetric
adjustment factor is described (Mdnache et al., submitted). This revised model represents
refinement of previously published models used to calculate the RDDR in the 1990 interim
RfC methods (Jarabek et al., 1989, 1990; Miller et al., 1988). For example, rather than
linear interpolation between the published (Raabe et al., 1988) means for deposition measured
at discrete particle diameters, as previously done for the laboratory animal deposition
modeling, equations have now been fit to the raw data as described herein.
The equations to perform calculations for monodisperse particles are provided; how the
calculated efficiencies may be transformed to fractional depositions is indicated; how to use
the model to predict deposition fractions for polydisperse particles is explained; and the
effects of the mass median aerodynamic diameter (MMAD) and the geometric standard
deviation (a ) on the regional deposited dose ratio (RDDRr) calculations are illustrated.
Because VE must be calculated from the default body weights (Table 4-5) using allometric
scaling (Equation 4-4) for use as input to the empirical model, the example of hand
calculation of monodisperse deposition includes a VE calculation.
Fractional deposition of particles in the airways of the respiratory tract may be
estimated using theoretical or empirical models or some combination of the two. Progress is
being made in answering the data needs of theoretical models (e.g., exact airflow patterns,
complete measurements of the branching structure of the respiratory tract, pulmonary region
mechanics), however, many uncertainties remain. Empirical models are systems of equations
that are fit to experimentally determined deposition in vivo. These models do not require the
detailed information needed for theoretical models, however, they can not provide estimates
of dose to localized specific sites (e.g., respiratory versus olfactory nasal epithelium terminal
bronchioles, carinal ridges). Measurement techniques are such that only general regions can
G-l
-------
p. 7
be defined (Stahlhofen et al., 1980; Lippmann and Albert, 1969; Raabe et al, 1977) which
limits the regions that can be defined for a dosimetry model. Despite this level of generality,
regional information is available now for humans and a number of commonly used laboratory
animals. Empirical models of regional fractional deposition have been presented for humans
(Yu et al., 1981; Miller et al., 1988; Stahlhofen et al., 1989). The empirical model
described in this appendix was fit for five species of commonly used laboratory rodents using
experimental data received from Dr. Otto Raabe (Raabe et al., 1988). At the same time,
Dr. Morton Lippmann (Lippmann and Albert, 1969; Lippmann, 1970, 1977; Chan and
Lippmann, 1980; Miller et al., 1988) and Dr. Wilhelm Stahlhofen and colleagues (Stahlhofen
et al., 1980, 1983, 1989; Heyder and Rudolf, 1977; Heyder et al., 1986) provided the
individual experimental measurements from their published studies. Using these data, the
human model published in Miller et al. (1988) was extended by refitting the extrathoracic
(ET) (oral breathing) and tracheobronchial (TB) deposition efficiencies with the original raw
data as well as by fitting a pulmonary (PU) deposition efficiency equation (Menache et al.,
submitted).
G.I EMPIRICAL MODEL FOR REGIONAL FRACTIONAL
DEPOSITION EFFICIENCIES
The equations describing fractional deposition were fit using data on particle deposition
in CFj mice, Syrian golden hamsters, Fischer 344 rats, Hartley guinea pigs, and New
Zealand rabbits. A description of the complete study including details of the exposure may
be found elsewhere (Raabe et al., 1988). Briefly, the animals were exposed to radiolabelled
ytterbium (169Yb) fused aluminosilicate spheres in a nose-only exposure apparatus. Twenty
unanesthetized rodents or eight rabbits were exposed simultaneously to particles of
aerodynamic diameters (dae) about 1, 3, 5, or 10 /xm. Half the animals were sacrificed
immediately post exposure; the remaining half were held 20 h post exposure. One-half of the
animals at each time point were male and the other half were female. The animals were
dissected into 15 tissue compartments, and radioactivity was counted in each compartment.
The compartments included the head, larynx, GI tract, trachea, and the five lung lobes. This
information was used directly in the calculation of the deposition fractions. Radioactivity was
G-2
-------
p. 8
also measured in other tissues including heart, liver, kidneys, and carcass; and additionally in
the urine and feces of a group of animals held 20 hours. In the animals sacrificed
immediately post exposure, these data were used to ensure that there was no contamination of
other tissue while the data from the animals held 20 hours were used in the calculation of a
fraction used to partition thoracic deposition between the TB and PU regions. This partition
is discussed below briefly and described in detail elsewhere (Raabe et al., 1977). Finally,
radioactivity was measured in the pelt, paws, tail, and headskin as a control on the exposure.
Although there are some other studies of particle deposition in laboratory animals (see
review by Schlesinger, 1985), no other data have the level of detail or the experimental
design (i.e., freely breathing, unanesthetized, nose-only exposure) required to provide
deposition equations representative of the animal exposures used in many inhalation
toxicology studies. However, many inhalation toxicology studies are not nose-only
exposures. While this is a necessary exposure condition to determine fractional particle
deposition, adjustments for particle inhalability and ingestion can be made to estimate
deposition fractions under whole-body exposure conditions.
The advantages of using the data of Raabe et al. (1988) to develop the deposition
equations include:
• the detailed measurements were made in all tissues in the animal, providing mass
balance information and indicating that there was no contamination of nonrespiratory
tract tissue with radioactivity immediately post exposure,
• the use of five species of laboratory animals under the same exposure conditions,
• the use of unanesthetized, freely breathing animals, and
• the use of an exposure protocol that makes it virtually impossible for the animals to
ingest any particles as a result of preening.
Regional fractional deposition, Ff, was calculated as activity counted in a region
normalized by total inhaled activity (Table G-l). The proportionality factor, fL, in Equations
G-2 and G-3 is used to partition thoracic deposition between the TB and PU regions. It was
calculated using the 0 and 20-h data and is described in detail by Raabe and co-workers
(1977).
G-3
-------
p. 9
TABLE G-l. REGIONAL FRACTIONAL DEPOSITION
P _ Activity Counted in a Region
r Total Inhaled Activity
[head + GI tract + larynx]n h
Extrathoracic (ET): FFT = 1——
hi Total Inhaled Activity
5
h + fL x JL lobei,o h (G-2)
Tracheobronchial (TB): FTR = ^
10 Total Inhaled Activity
5
Pulmonary (PU): FPU = l"1
(1 - f^ x £ lobei>0 h (G.3)
Total Inhaled Activity
These regional deposition fractions, Fr, however, are affected not only by the minute
volume (Vg), MMAD and ag, but also by deposition in regions through which the particles
have already passed. Deposition efficiency, r?r, on the other hand, is affected only by VE,
MMAD, and o . The differences between deposition fraction and efficiency, calculated as
&
described below, are described in more detail later in this appendix. In the aerodynamic
domain, that is for particles with diameters > 0.5 /-em (see Appendix H for further discussion
of particle dimension issues), efficiencies increase monotonically and are bounded below by 0
and above by 1. The logistic function (Equation G-4) has mathematical properties that are
consistent with the shape of the efficiency function (Miller et al., 1988):
where E(?7r) is the expected value of deposition efficiency (r?r) for region r, and x is
expressed as an impaction parameter, dae2Q, for extrathoracic deposition efficiency and as
aerodynamic particle size, dae, for TB and PU deposition efficiencies. The flow rate, Q, in
the impaction parameter may be approximated by VE/30. The parameters a and ft are
estimated using nonlinear regression techniques.
G-4
-------
p. 10
To fit this model, efficiencies must be derived from the deposition fractions that were
calculated as described in Table G-l. Efficiency may be defined as activity counted in a
region divided by activity entering that region. Then, considering the region as a sequence of
filters in steady state, efficiencies may be calculated as follows:
J?ET = FET (G-5)
h + fL x £ lobei>0 h (G 6)
7776 =
5
(1 - f^ x £ lobei>0 h
(GJ?)
Using these calculated regional efficiencies in the individual animals, the logistic
function (Equation G-4) was fit for the ET, TB, and PU regions for the five animal species
and humans. The parameter estimates from these fits are listed in Table G-2. Curves
produced by these equations have been compared where applicable to the data reported in
Schlesinger (1985), and the results are not inconsistent. As discussed by Schlesinger (1985),
there are many sources of variability that could explain differences in predicted deposition
using this model and the observed deposition data in the studies reported by Schlesinger
(1985).
G.2 TRANSFORMING FITTED EFFICIENCIES TO PREDICTED
REGIONAL FRACTIONAL DEPOSITION
The fitted equations are then used to generate predicted efficiencies 0)) as a function of
impaction in the ET region and of aerodynamic particle size in the TB and PU regions.
Finally, the predicted efficiencies are multiplied together and adjusted for inhalability, I, as
shown in Equations G-8 through G-10 to produce predicted deposition fractions (Fr) for
monodisperse and near monodisperse (a < 1.3) particles.
G-5
-------
p. 11
TABLE G-2. DEPOSITION EFFICIENCY EQUATION
ESTIMATED PARAMETERS
ET (Nasal)
Species
Human
Rat
Mouse
Hamster
Guinea Pig
Rabbit
a
7.129*
6.559
0.666
1.969
2.253
4.305
ft
-1.957a
-5.524
-2.171
-3.503
-1.282
-1.628
TB
a
3.298
1.873
1.632
1.870
2.522
2.819
ft
-4.588
-2.085
-2.928
-2.864
-0.865
-2.281
PU
a
0.523
2.240
1.122
1.147
0.754
2.575
ft
-1.389
-9.464
-3.196
-7.223
0.556
-1.988
•Source: Miller et al., 1988.
FET - I X fiET
= I X (1 - rjgj) X
Fpu = I X (1 - J)ET) X (1 -
X
(G-8)
(G-9)
(G-10)
Inhalability, I, is an adjustment for the particles in an ambient exposure concentration
that are not inhaled at all. For humans, an equation has been fit using the logistic function
(Mdnache et al., 1995). Using the experimental data of Breysse and Swift (1990):
1 = 1-
1
1 + e
10.32-7.17 log.0d
((Ml)
10 ae
The logistic function was also fit to the data of Raabe et al. (1988) for laboratory animals
(M6nacheetal., 1995):
1 = 1-
1
2.57-2.81 log.0d
(G-12)
1 + e'
G-6
-------
p. 12
For example, calculation of Fpu for a female Syrian golden hamster exposed to a nearly
monodisperse particle (a < 1.3) with an MMAD of 1.8 in a subchronic study would
proceed as follows.
1. Calculate the default VE. (If the study for which the RDDRPU is being calculated
has information on experimentally measured VE, that information may be
substituted for the default value; however, this could necessitate changes to surface
areas and body weight (if extrarespiratory tract effects are being examined).
If there is inadequate information to change all of these values, then the default
values should be used.)
a. The default body weight for a female Syrian hamster in a subchronic study from
Table 4-4 is 0.095 kg.
b. Calculation of VE expressed in natural logarithms using hamster coefficients
from Table 4-5:
log (VE) = -1.054 + 0.902 X log (0.095)
= -3.177
c. Convert from natural logs to arithmetic units
exp (-3.177) = 0.0417
d. Convert from L to mL by multiplying by 1,000
VE = 41.7
o •
2. Calculate the impaction parameter as MMAD x VE/30 for the ET region
= (1.8)2 x (41.7/30)
= 4.504
and take the Iog10
= 0.654
A
3. Calculate TJET using the parameters from Table G-2
+ exp(1.969 - 3.503 x 0.654))
= 0.580
G-7
-------
p. 13
4. Calculate Iog10 (MMAD) for the TB and PU regions
= Iog10d.8)
= 0.255
A
5. Calculate T?TB using the parameters from Table G-2
+ exp(1.870 - 2.864 X 0.255))
= 0.242
6. Calculate J?PU
+ exp(1.147 - 7.223 X 0.255))
= 0.667
7. Calculate the inhalability fraction, I
= 1 - (1/(1 + exp(2.57 - 2.81 x 0.255)))
= 0.865
8. Calculate FET (if desired)
= 0.865 x 0.580
= 0.502
9. Calculate FTB (if desired)
= 0.865 x (1 - 0.580) X 0.242
= 0.088
10. Calculate Fpu
= 0.865 x (1 - 0.580) X (1 - 0.242) x 0.667
= 0.184
These hand-calculated fractional depositions for monodisperse particles might differ
slightly from the fractions generated by the computer program due to rounding errors.
In particular, the parameter estimates in Table G-2 are only reported to three decimal places
but are used with nine decimal places in the program. Because all of these digits are not
significant, however, the deposition fractions should never be reported to more than two
digits.
The human deposition fractions may be calculated using the same strategy. The only
default VE, however, is 13.8 L/min. As described in step l.d, this value should be
G-8
-------
p. 14
converted to mL by multiplying by 1,000. The information provided in Table G-2 allows for
estimation of deposition in humans for nasal breathing only. When exercising (VE greater
than 35 L/min), a portion of the inhaled air will enter through the mouth. The ET deposition
efficiencies for oral breathing are different from those for nasal breathing and are not
recorded in Table G-2. They are, however, included in the computer program as well as
proportionality factors defining flow splits between the nose and mouth at higher VE. The
additional complexities engendered in the calculation of the ET deposition fraction when both
oral and nasal breathing are involved are such that those calculations should not be performed
by hand.
Figure G-l illustrates the relationship between the predicted efficiencies and predicted
depositions using this model for the mice. A qualitatively similar set of curves could be
produced for any of the other four species. The calculations were made according to the
ten steps listed above. The particles were assumed to be monodisperse and the default body
weight (BW) for the mice, taken from Table 4-4, was 0.0261 kg. This is the default BW of
male BAF1 mice for chronic exposure study durations. Regional deposition efficiencies and
fractions were calculated for particles with dae ranging from 0.5 to 10 /*m. These calculated
points were connected to produce the smooth curves shown in Figure G-l. The three panels
on the left of Figure G-l are plots of the predicted regional deposition efficiencies; the three
panels on the right show the predicted regional deposition fractions derived from the
estimated efficiencies and adjusted for inhalability. The vertical axis for the predicted
deposition efficiency panels range from 0 to 1. Although the deposition fraction is also
bounded by 0 and 1, the vertical axes in the figure are less than 1 in the TB and PU regions.
The top two panels of Figure G-l are the predicted deposition efficiency and fraction,
respectively, for the ET region. These two curves are plotted as a function of the impaction
parameter described for Equation G-4. The middle two and lower two panels show the
predicted deposition efficiencies and fractions for the TB and PU regions, respectively.
These four curves are plotted as a function of dae. When a particle is from a monodisperse
size distribution, the dae and the MMAD are the same. If, however, the particle is from a
polydisperse size distribution, the particle can not be described by a single dae; the average
value of the distribution, the MMAD, must be used. (See Appendix H for further discussion
of particle sizing, units, and averaging methods). In the aerodynamic particle size range, the
G-9
-------
p. 15
Predicted Regional Deposition Efficiency Predicted Regional Deposition Fraction
uo
i ' i
1 10
Impactlon, (urn)4 ml/sec
TTT]
100
0.0
0.1
1 10
Impaction, (urn)2 ml/sec
100
10
Aerodynamic diameter, urn
Aerodynamic diameter, urn
Aerodynamic diameter, urn
0.1 i
Aerodynamic diameter, |im
"
10
Figure G-l. Comparison of regional deposition efficiencies and fractions for the mouse.
A default body weight of 0.0261 kg (from Table 4-4) was used in these
calculations. The fractional deposition (solid line) and inhalability (dashed
line) are shown in the upper right panel.
G-10
-------
p. 16
deposition efficiency curves all increase monotonically as a function of the independent
variable (i.e., either the impaction parameter or d^ and have both lower and upper
asymptotes. The curves describing the deposition fractions, however, have different shapes
that are dependent on the respiratory tract region. Deposition fractions in all three regions
are nonmonotonic—initially increasing as a function of particle size but decreasing as particle
sizes become larger. This is because particles that have been deposited in proximal regions
are no longer available for deposition in distal regions. As an extreme example, if all
particles are deposited in the ET region, no particles are available for deposition in either the
TB or PU regions. In the ET region, the nonmonotonic shape for fractional deposition is due
to the fact that not all particles in an ambient concentration are inhalable.
G.3 POLYDISPERSE PARTICLES
As discussed in Appendix H, particles in an experimental or ambient exposure are
rarely all a single size but rather have some distribution in size around an average value.
As this distribution becomes greater, the particle is said to be polydisperse. Panel A of
Figure G-2 illustrates the range of particle sizes from a distribution that is approximately
monodisperse (ffg = 1.1) and particles that come from a lognormal highly polydisperse
distribution (a = 3.0), although both distributions have the same MMAD of 4.0 f*m. Also
drawn in Panel A of Figure G-2 is a vertical line through the MMAD that represents the
extreme case of o = 1.0, that is, an exact monodisperse particle distribution in which all
particles are a single size, which is also the MMAD.
The empirical model described in this appendix was developed from exposures using
essentially monodisperse particles (which are treated as though they are exactly
monodisperse). It is therefore possible to multiply the particle size distribution function
(which is customarily considered to be the lognormal distribution) by the predicted
depositions (calculated as described in Equations G-8 through G-10) and integrate over the
entire particle size range (0 to «>). Mathematically, this calculation is performed as described
by Equation G-3, and is illustrated for the mouse and human ET regions in panels B and C
respectively, of Figure G-2.
G-ll
-------
p. 17
w
OJ)
04
0.4
02
OuO
ai
6"
4"
3"
2'
1 "
^
Logn
ag-1.0 M
Og -1.1
__<^-3.0
_
Lognormal Distribution
MMAD-4.0 |im
10
Extrathoracic Deposition
Mouse
1
MMAD,
(B)
10
oa
(A)
i, urn
50
ao
Extrathoracic Deposition
Human •
(C)
Og-1.0
1
MMAD, i
10
Figure G-2. Range of particles for lognormal distributions with same MMAD but
differing geometric standard deviations (A). Effect of polydisperse
particles on predicted extrathoracic deposition fractions in mice (B) and
humans (C).
= J [FJ
X
x exp
-1/2-
(logdae - logMMAD)2
logff.
dd,
ae
(G-13)
A
where log refers to the natural logarithim, [Fr] is the predicted polydisperse fractional
deposition for a given MMAD, and [Fr]m is the predicted monodisperse fractional deposition
for particles of size dae. The limits of integration are defined from 0 to « but actually include
only four standard deviations (99.95% of the complete distribution). For each particle size in
A
the integration, [Fr]m is calculated as described in the ten steps listed in this appendix, then
G-12
-------
p. 18
multiplied by the probability of observing a particle of that size in a particle size distribution
with that MMAD and ag.
Panels B and C of Figure G-2 illustrate one of the principal effects of polydisperse
particle size distributions on predicted deposition fractions in the ET region, which is to
flatten the deposition curve as a function of MMAD. This same effect is observed also in the
TB and PU regions. (Note that the curves in panels B and C are expressed as a function of
MMAD. They were generated as a function of the impaction parameter but are expressed as
a function of MMAD for ease of comparison between species. A VE of 37.5 mL/min was
used for the mouse and of 13.8 L/min for the human.) Rudolf and colleagues (1988) have
also investigated the effect of polydisperse particle size distributions on predicted regional
uptake of aerosols in humans and present a more detailed discussion of these and related
issues.
G.4 REGIONAL DEPOSITED DOSE RATIO CALCULATIONS IN
RATS AND HUMANS: AN EXAMPLE
Three respiratory tract RDDRj. values were calculated as described by Equation 4-7
using (1) the default body weight for a female Fischer 344 rat in a subchronic study from
Table 4-5 to estimate VE for the rat and (2) the regional respiratory tract surface areas as the
normalizing factor for the rat and human from Table 4-4. In Figure G-3, the ET, TB, and
PU RDDR, as a function of particle size, for particles in the aerodynamic size range are
compared for monodisperse and a highly polydisperse particle size distribution (a = 3.0).
o
When the RDDRr is 1, both human and rodent would be predicted to have a comparable dose
rate per unit surface area of the inhaled particles. Ratios with a value of less than 1 indicate
that for an equivalent external exposure concentration, the dose rate per unit surface area in
the human will be greater than in the rodent; while RDDRr which are greater than 1 occur
when the rodent receives a larger surface area adjusted dose rate than the human. Figure G-3
indicates that in the ET region, the human will be expected to have a higher dose rate per
unit surface area than the rat over the aerodynamic particle size range for both monodisperse
and polydisperse particle size distributions. For the highly polydisperse particle size
distribution, the RDDR in the ET region is relatively constant as a function of aerodynamic
G-13
-------
p. 19
I
CC
g
DC
g
O
DC
CC
g
CC
0.30
0.25
0.20
0.15^
0.10 T
0.05
0.00
-3.0
14 n
12
10 ~\
8
6H
i ' I
4 6
MMAD,
8
10
,-1.0
4-
2-
o-
{
<*&\
X-\
) 2 4
i ' i ' i
6 8 10
MMAD, urn
0.00
MMAD, \im
Figure G-3. Regional deposited dose ratios (RDDRr) for ratzhuman as a function of
mass median aerodynamic diameter (MMAD) for monodisperse (ag = 1)
and polydisperse (ag = 3) particle size distributions.
G-14
-------
p. 20
particle size. This may be interpreted to mean that for a highly polydisperse size distribution,
the dose rate per unit surface area to the ET region of the human will be approximately 10 to
15 times that to the ET region of the rat regardless of the particle MMAD. In the TB region,
the RDDR declines over the aerodynamic particle size range for both mono- and polydisperse
particle size distributions. For particles smaller than about 2 /*m MMAD, the rat is predicted
to have a higher dose rate than the human; for larger particles, the relationship is reversed.
In the PU region, a relationship that is qualitatively similar in shape to the RDDRET is
observed; however, the range of the RDDRPU is much larger and there is a suggestion that
the dose rate in the rat is greater than that in the human for particles of about 2 /*m MMAD
since the RDDRpu > 1.0.
This example illustrates the complexities in the relationships between dose rate per unit
surface area in the three respiratory tract regions for rodents and humans. The regional
differences as well as the differences due to MMAD and a indicate the importance of
replacing administered dose with dosimetric information whenever possible in making risk
evaluations.
G-15
-------
p. 21
APPENDIX H
PARTICLE SIZING CONVENTIONS
Solid or liquid particles suspended in a gas are called an aerosol. In lexicological
experiments and epidemiological studies, aerosol particles from a given exposure are
commonly described by some measure of the average size of the particle and some measure
of variability in that average size. Although the average size is usually expressed as a
diameter, there are numerous methods for calculating diameter. In this appendix, some of
the more common sizing conventions for spherical (or nearly spherical) particles are briefly
discussed. Conversions from reported units to the units required to use the particle dosimetry
model described in this document are provided. More detailed discussions of particle
properties and behavior may be found elsewhere (Raabe, 1971, 1976; Hinds, 1982).
H.1 GENERAL DEFINITIONS
Particles in an exposure are rarely all a single size but rather have some distribution in
size around an average value. It is generally accepted (Raabe, 1971) that the lognormal
distribution provides a reasonable approximation for most observed spherical particle
distributions. For this reason, particle exposures are frequently characterized by median
diameter and the geometric standard deviation (ffg). Statistically speaking, data from a
lognormal distribution may be completely described by the median and geometric standard
deviation. As a approaches 1.0, the distribution of the particles approaches a single size and
the particles are said to be monodisperse. As a matter of practicality, a distribution is
considered to be near monodisperse when a is less than 1.3. As a approaches infinity, the
distribution contains particles of many sizes and is said to be polydisperse. By definition,
a cannot be less than 1.0.
In lexicological experiments, researchers might use (or try to use) monodisperse
spherical particles because deposition is a function of particle size. Real world exposures,
however, are rarely to monodisperse particles. Accordingly, laboratory animal experiments
H-l
-------
p. 22
designed to mimic some real exposure will use polydisperse particle distributions. Studies of
diesel exhaust and of Mt. St. Helens volcanic ash, for example, used highly polydisperse
particles. In addition to being polydisperse, such particles also have irregular shapes.
Although some irregular particles may be described by their aerodynamic diameter and so be
considered to behave like spherical particles, other particles have such extreme differences in
shape that they must be described by other parameters. Fibers are one extreme example of
nonspherical particle shapes. Deposition fractions for these particles may not be calculated
with the particle dosimetry model used in the methodology.
Particle diameters are most frequently reported as geometric diameter (d ) or
aerodynamic diameter (dae). Aerodynamic diameter is defined as-the diameter of a unit
density sphere having the same settling velocity as the geometric diameter of the particle in
question. Geometric diameters may be converted to aerodynamic diameters according to the
following equation:
dae = dg^C(dg)/C(dae) -
where p is the particle density in g/cm3 and C(d) is the Cunningham slip correction factor.
The particle dosimetry model requires that the particle size be expressed in aerodynamic
diameter so studies reporting particle sizes in these units will most likely not require any
further conversion. There are, however, two commonly used definitions for dae; the
methodology uses the definition of the Task Group on Lung Dynamics (1966). Because of
the complexities in calculating dae by the Task Group equation, other investigators have
developed an alternate specification for aerodynamic diameter (Mercer et al., 1968; Raabe,
1976). This dae is called an aerodynamic resistance diameter, dar, and may be converted to
dae according to the following equation:
dae = (dar2 + 0.0067)0'5 - 0.082 . (H-2)
H-2
-------
p. 23
Summary information from an exposure will most often be presented as count (CMD),
mass (MMD), surface (SMD) or activity (AMD), median (geometric) diameter. The
summary information might be reported in terms of median aerodynamic diameters instead
(CMAD, MMAD, SMAD, AMAD).
Because the particle distributions are assumed to be lognormal, estimation of count,
surface area, or mass distributions for any given sample of particles may be made once one
of those distributions has been measured. Figure H-l shows a typical log-normal distribution
and the various indicated diameters encountered in inhalation toxicology literature.
Table H-l provides the definitions of these diameters. A series of conversion equations
originally derived by Hatch and Choate (1929) may be used to convert the reported units to
MMAD, the units required by the particle dosimetry model. Figure H-2 shows the same
aerosol as in Figure H-l plotted on log-probability paper and illustrates the various size
parameters that can be computed using the Hatch-Choate equations. The relevant conversion
equations are summarized in Table H-2. It is a characteristic of any weighted distribution of
a lognormal distribution (such as the conversions described in Table H-2) that the geometric
standard deviation will be unchanged.
Conversion of activity median diameter (AMD) to MMAD depends on how the
radiolabeling of the particle was done. If the label was generated in a liquid, then the label is
distributed throughout the volume of the particle and the AMAD may be considered to be
equivalent to the MMAD. If, however, the radioactivity was attached to the surface of the
particle, then the activity median diameter may be considered to be proportional to the
surface area median diameter. More information on the labelling procedure is required to
provide an estimate of the proportionality factor. Further discussion of issues related to
activity diameters may be found elsewhere (Hofmann and Koblinger, 1989).
The concept of activity diameter is also useful when considering the physical
characteristics of particles that are responsible for the health effect or toxicity of concern.
The activity diameter of a particle may be the most appropriate expression of size for this
purpose. This expression takes into account the "activity" of the physical property of the
particle. For example, if the toxicant is distributed only on the surface, then the activity
median diameter is equal to the surface median diameter; and conversions to MMAD would
H-3
-------
p. 24
0.8 -
0.6 _
0>
S
£0.4
O
CO
0.2
Count mode (0.619 //m) d
Count median (1.0 /urn) dg
Count mean (1.272 /urn) d
Diameter of average area
(1.6140m)d.
Diameter of average mass
(2.056 urn) dm
Area median (2.614 /urn) da
Area mean (3.324 /urn) d a
—Mass median
(4.226 m d
Mass mean
(5.374 0m) dm
0
2 4
Particle Diameter (|im)
Figure H-l. An example of the log-normal distribution function of an aerosol.
Definitions of diameters commonly used are provided in Table H-l. Note
that if aerodynamic diameters were being measured, then count or other
frequency distribution would be measured against that (e.g., count median
aerodynamic diameter [CMAD]).
Source: Orr and Keng (1976).
H-4
-------
p. 25
TABLE H-l. PARTICLE DIAMETER DEFDVITIONS
Count mode diameter
Count median diameter
Count mean diameter
Diameter of "average mass"
The most frequently characterized particle diameter. In Figure H-l, the
frequency is normalized to frequency (or number) per micron. This type of
graph is also used in analyzing cascade impactor data.
This diameter is used to describe any log-normal distribution. It is the
diameter of a particle that is both larger and smaller than half the particles
sampled.
The average particle diameter. It is calculated by first multiplying each
diameter measured by the number of particles having that diameter.
Summing all of these products over the entire range of diameters measured
and dividing by the total number of particles sampled gives the count mean
diameter.
Another average particle diameter related to the total mass of particles
sampled. The mass of the particle of "average mass" multiplied by the
total number of particles sampled, equals the total mass of particles
sampled. The total mass of particles sampled is calculated by first
multiplying the single-particle mass calculated for each diameter measured
by the number of particles having that diameter and summing all of these
products over the entire range of diameters measured. The average mass of
each individual particle sampled is obtained by dividing this number by the
total number of particles. The diameter is calculated by assuming a sphere
and applying the density of the material to convert from mass to volume
and then to diameter.
Mass median diameter
Mass mean diameter
Diameter of the particle having a mass that is both larger and smaller than
the mass of half the particles sampled.
Average particle diameter, calculated by first multiplying each diameter
measured by the cumulative mass of all particles having that diameter.
Summing all of these products over the entire range of diameters measured
and dividing by the total mass of the particles sampled gives the mass mean
diameter.
Source: Moss and Cheng (1989).
be the same as described above for radiolabeled activity. If the toxicant is soluble, the
surface area of the particle will influence the rate of dissolution because solubilization occurs
at the surface. Such a situation needs to be understood more thoroughly, especially for
complex particles.
H-5
-------
p. 26
CO
I
98
90
70
50
CO
«j 30
10
0.1
1.0
Particle Diameter, D (/urn)
10
30
Figure H-2. Plot of same aerosol as in Figure H-l on log-probability paper. The curves
illustrate the various size parameters that can be computed using the
Hatch-Choate equations.
Source: Marple and Rubow (1980).
H.2 GENERATION SYSTEMS AND CHARACTERIZATION
INSTRUMENTS
Aerosols may be generated by condensation of vapors, by dispersion of dry particles or
liquids, or by dispersion of a suspension of solids in a liquid. Each of the currently available
systems and applied techniques used to generate test atmospheres for inhalation toxicology
testing have operating specifications that should be evaluated when attempting to determine
whether the operating conditions and size range stated was appropriate to the technique and to
ascertain the probable particle size range. The operating specifications (pressure, flow rate,
output concentrations, particle diameters, and standard deviations) of various generation
H-6
-------
p. 27
TABLE H-2. LOGNORMAL CONVERSION EQUATIONS FOR
COMMON TYPES OF DIAMETERS
Count to Mass
MMAD = CMAD exp (3 [log ag]2)
MMAD = />°-5 CMD exp (3 [log ag]2)
Activity to Mass
MMAD = AMAD if label may be assumed to be distributed throughout volume of particle
MMAD = pSMAD if label is attached to a proportion, p, of the surface of the particle
Count to Surface
SMAD = CMAD exp (2 [log ag]2)
SMAD = p°'5 CMD exp (2 [log gg]2)
Note: log = natural logarithm.
systems have been reviewed elsewhere and can serve as references (Moss and Cheng, 1989;
American Conference of Governmental Industrial Hygienists, 1978; Willeke, 1980).
Characterizing test atmospheres includes defining the aerosol concentration, chemical
composition, and particle size and shape. Aerosols for toxicology testing should be
characterized to quantitate toxicant concentration, stability, and particle size distribution
during exposure. Mass concentration of aerosols can be measured directly using filters,
impingers, and impactors. Other methods of determining mass concentration include beta-
attenuation, piezobalance, and photometers. The latter two instruments are real-time
continuous samplers that enable monitoring and early detection of problems related to
generation and delivery. Number concentrations are obtained by using nucleus counters,
optical counters, electrical counting, and microscopy.
The shape and size of particles are determined by collecting particles on filters, on grids
mounted on thermal or electrostatic precipitators or by microscopy. Dynamic size
measurements made using inertial or mobility aerosol instruments are frequently used to
determine the aerodynamic and mobility equivalent diameters. These diameters can be used
in the conversions scenarios provided in the next section.
Inertial types of instruments are used to measure dae larger than about 0.5 /*m. For
particles less than 0.5 /im in diameter, most inertial instruments are not effective in
H-7
-------
p. 28
separating and measuring particle size. In these cases, the mobility type of aerosol
instrument is used. The mobility equivalent diameter determines the collection efficiency in
many processes that are controlled by the diffusion deposition mechanism. Two types of
mobility instruments are the electrical aerosol analyzer (EAA) and the diffusion batteries.
No single instrument can measure aerosol size distributions of particles with diameters from
0.005 to 10 /*m. Sometimes different exposure levels are generated or characterized with
different instruments. Careful analysis of data is required, because the inertia! instruments
(e.g., cascade impactor) give mass distribution, and the EAA and screen diffusion battery
give number distribution. Figure H-3 shows the measurement ranges of aerosol monitoring
instruments; Knowledge of the measurement range of the instruments used to characterize
the test atmosphere can be useful in determining the probable particle size range used in a
given study.
Cascade Impactor
I 1
Low Pressure Cascade Impactor
i 1
EAA
Diffusion Battery or PFDB
i \
Particle Diameter
0.001 0.01 0.1 1 10 100 fo1111)
Impaction and Sedimentation
Diffusion
Figure H-3. Measurement ranges of aerosol monitoring instruments.
Source: Moss and Cheng (1989).
H-8
-------
p. 29
H.3 CONVERSION SCENARIOS
Particle information reported in a study will most likely fall into one of the seven
categories defined below. The remainder of this appendix describes these seven possibilities
and outlines appropriate strategies to convert reported particle information to MMAD,
required to run the particle dosimetry model.
1. MMAD and ag.
In this case the information in the study has been reported in the units required for
analysis, and no conversions or changes are required to run the model.
2. A median diameter and a .
Conversions from the most commonly used medians to MMAD are provided in
Table H-2. No conversions are required for ag.
3. A median diameter and a range of particle sizes.
The variance, a , may be estimated as
og = exp
= exp
log(median/lower bound)
n
(H-3)
log(upper bound/median)
n
where log is the natural logarithm, exp is the irrational number, e, raised to the power
in the brackets, and the range falls between the reported upper and lower bounds. The
range will include some percentage of the particles that is reflected in the parameter n,
the number of standard deviations used in calculating a (Table H-3). The median
diameter may then be converted to MMAD according to the equations in Table H-2, if
necessary.
H-9
-------
p. 30
TABLE H-3. PERCENTAGE OF PARTICLES IN THE REPORTED RANGE
ASSOCIATED WITH THE NUMBER OF STANDARD DEVIATIONS (n)
USED TO CALCULATE THE GEOMETRIC STANDARD DEVIATION
Percentage of Particles
in the Reported Range n
0.68 1
0.95 2
0.997 3
> 0.999 4
4. A median diameter only.
An estimate of a must be derived from outside sources. In order of preference, the
o
following sources are recommended.
a: Other studies by the same group using the same compound for which the median
and a are reported.
b: A comparison of the variances reported by other studies for the same compound
could be used to determine reasonable bounds on a . Using the largest and
smallest a determined by this method, the dosimetry model can be run to
determine the sensitivity of the dose ratio to a for this particular particle size
and rodent to human comparison.
c: The particle size range can be estimated from Figure H-4 and a calculated
according to Equation H-3 and Table H-3 (letting n = 4). If necessary, the
median can then be converted to MM AD according to Table H-2.
5. A range of particle sizes is the only information provided.
A median, in the same units as the reported particle size range, must be estimated from
outside sources. In order of preference, the following sources are recommended.
a: Other studies by the same group using the same compound for which the median
is reported.
b: A comparison of the medians reported by other studies for the same compound
could be used to determine reasonable bounds on the median. If necessary, the
dosimetry model could be run using the largest and smallest medians estimated by
H-10
-------
p. 31
u. n 9
^SUcU
b&^n
N» CO,
Dk><
Tobacco
Mosaic
Virus
rtGaaMotecu
Toba
N«cr<
Vhu
e*
Normal bnpu
ZlncO:
SS Virus &
i Protein
Carbon Bta
Aeiosob
Weak)
It
ddeRi
df.
Quiet Outdoor Alj^
leMlunolcal Dutt and Fumes
Fog
Smeller Dust & Fumes
Ammonium Chtoodi
, Fume* f
Akal Fumes
(!)•• w
Tobacco Smoke
OH Smoke
Magnesium Oxide Smote
Sliver Iodide
Rosin Smoke
S
, *"
R
Sulflde
uKurlc Add Mist
wcfldde Dust
Bacteria
JEnamels) Pigment* (F
Combustion Nudel
SeaSaK
Nudel
RE
Foundry Dust
our MID
Dust ^_
Mist
Ground Limestone
Ore, Pulps for Floatation^
Cement Dust
• ,
ats)
^pray Dried MPk ^
FERENCEJ
SIZES 1
I
Screen
( Pulverized Coal
Plant Spores
Pollens
Sneezes
Fly Ash
^ Sand Tailings
r
Human
•lair Oh
Mesh*0 325*
1
Washed
uneter
Rain!
Fount
Visible to Eye
0 1
Y
*«2
35
'r
Jrops >
lySand
0.0001 0.0005 0.001 0.005 0.01 0.05 0.1 0.5 1 6 10
Particle diameter (urn)
60 100
500 1000 6000 10,000
Figure H-4. Various airborne materials and their size ranges.
Source: Hatch and Gross (1964).
this method. Note that running the model with alternate estimates of the median
will require alternate estimates of a .
Using the estimated median and the range of particle sizes, a may be estimated as
o
described above in scenario 3. Finally, the median diameter may be converted to
MM AD according to the equations in Table H-2, if necessary.
6. Descriptive information on particle size is the only information provided.
Some of the more commonly used expressions of particle characteristics and the
generally accepted particle sizes associated with these characteristics are shown in
Table H-4. Further information on ranges of sizes for some common classifications of
H-ll
-------
p. 32
TABLE H-4. GENERAL PARTICLE DESCRIPTIONS AND ASSOCIATED SIZES
Particle Description Size (jitm)
Coarse >2.5
Fine <2.5
Fumes, Smoke < 1
Fog, Mist < 1 -* 20
particles may be found in Figure H-4. Using this information, the median may be
estimated as described in scenario 5, and a estimated as described in scenario 4.
7. No information on particle sizes provided.
Studies that do not provide this information should be suspect for deficient quality.
Some of the older toxicology literature may not provide this information, however, and
a default value may need to be invoked. The first approach in this situation is to
attempt an estimate of particle size and distribution based on the generation apparatus
used. Operating specifications of various generation systems are available from the
manufacturer or reviewed elsewhere (Moss and Cheng, 1989; American Conference of
Governmental Industrial Hygienists, 1978; Willeke, 1980). In conjunction with this
information, the knowledge that prior to the late 1970s, the generation technology was
not sufficiently sophisticated to deliver consistent exposures of particle sizes above 3 jum
(MMAD) can be used to construct a default approach. The recommended default
approach is to use the MMAD and o characteristic for the given generation system that
is <3 /nm and that yields the smallest (i.e., most conservative) RDDR values for the
respiratory tract region of interest. Knowledge of the measurement range of the
instrument used to characterize the test atmosphere can also be used to estimate a
particle size. Figure H-3 provides the measurement ranges of some aerosol monitoring
instruments.
H-12
-------
p. 33
The second approach is to use particle size information from other studies to estimate
the particle characteristics for the exposure in question. If no such information is
available, Figure H-4 provides the general size ranges for most common classifications
of particles. Using this information, the median may be estimated as described in
scenario 5, and o estimated as described in scenario 4.
H-13
-------
p. 34
APPENDIX I
DERIVATION OF AN APPROACH TO DETERMINE
HUMAN EQUIVALENT CONCENTRATIONS FOR
EFFECTS OF EXPOSURES TO GASES
IN CATEGORIES 1 AND 2
As discussed in Sections 3.2 and 4.3, the optimal approach to describe regional
respiratory tract dose for extrapolation across species is to use a comprehensive dosimetry
model. The limited number of sophisticated dosimetry models that currently exist are either
chemical-specific or class-specific and require an extensive number of model parameters.
As discussed in Section 3.2, the chemical-specific or class-specific nature of these models has
been dictated by the physicochemical properties of the subject gases and therefore any single
resultant model is not applicable to the broad range of physicochemical properties of gases
(or vapors—herein referred to as gases) that this methodology is aimed at addressing.
In addition, sufficient data from which to estimate model parameters for each gas are often
unavailable.
A conceptual framework was thus developed to structure mathematical models that
require limited gas-specific parameters and that may be further reduced by simplifying
assumptions to forms requiring minimal information. The models in reduced form are the
basis of the default adjustments used in Section 4.3 and are used to estimate the human
equivalent concentrations (HECs) from no-observed-adverse-effect levels (NOAELs) of gases
when the lack of data for the required parameters precludes more comprehensive modeling.
This appendix provides the conceptual framework and background details of the default
derivation for the adjustments used in Section 4.3.
Because adverse respiratory effects may be observed in the extrathoracic (ET),
tracheobronchial (TB), or pulmonary region (PU), the conceptual framework is constructed to
derive a regional description of dose, defined as the regional absorption rate. The regional
absorption rate is used as a surrogate of regional dose and is assumed to represent the
1-1
-------
p. 35
effective dose for evaluation of the dose-response relationship. The physicochemical
properties such as the water solubility and reactivity (e.g., ionic dissociation or tissue
metabolism) of the gas in the respiratory system are major determinants of the regional
absorption rate. For example, styrene is relatively insoluble in water and unreactive with the
respiratory tract surface liquid and tissue. This gas is therefore absorbed primarily in the
lung periphery, where it partitions across the blood-gas barrier. Formaldehyde, however, is
both water soluble and reactive such that most of the gas is absorbed in the ET region. The
concept of differentiating gases based on their stability, reactivity, or tissue metabolic activity
has been proposed by Dahl (1990) who presented a methodology to assist in categorizing
gases. As discussed in Section 3.2, delineation of the categories is accomplished by
identifying dominant mechanistic determinants of absorption that are based on the
physicochemical characteristics of the gases. The dominant mechanistic determinants are
used to construct a conceptual framework that directs development of the dosimetry model
structures.
The categorization scheme discussed in Section 3.2 and developed more fully herein is
used to establish a dosimetry model structure for three categories of gases from which default
equations are developed by imposing additional simplifying assumptions. Model structures
for two of the three categories are developed in this Appendix. The structure for Category 3
gases is developed in Appendix J. Gases in Category 3 are relatively water insoluble and
unreactive in the ET and TB surface liquid and tissue and thus exposures to these gases result
in a relatively small dose to these regions. The uptake of these gases is predominantly in the
PU region and is perfusion limited. The site of toxicity of these gases is generally remote to
the principal site of absorption in the respiratory tract. Thus, the relatively limited dose to
the ET.and TB regions does not appear to result in any significant ET or TB respiratory
toxicity. Toxicity may, however, be related to recirculation. An example of a Category 3
gas is styrene. Gases that fall in Category 3 are modeled using the structure and default
equations presented in Appendix J.
The methodology developed in this appendix addresses the absorption of gases that are
relatively water soluble and/or reactive in the respiratory tract (Categories 1 and 2 of the
scheme described in Section 3.2). The focus here is on the description of dose for
respiratory tract effects. This is not to suggest that the toxicity is limited to the respiratory
1-2
-------
p. 36
regions and in fact, for Category 2, the model structure may be used to define a dosimetric
for remote toxicity because this category of gases has physicochemical characteristics that
may result in some systemic circulation of toxicant. The assumption of this modeling
approach is that the description of an effective dose to each of these regions for evaluation of
respiratory effects must address the absorption or "scrubbing" of a relatively water soluble
and/or reactive gas from the inspired airstream as it travels from the ET region to the PU
region. That is, the dose to the distal regions (TB and PU) is affected by the dose to the
region immediately proximal. The appropriateness of assessing proximal to distal dose
representative of the scrubbing pattern is supported by the often observed proximal to distal
progression pattern of dose-response for respiratory tract toxicity with increasing
concentration. At low concentrations of relatively water soluble and/or reactive gases,
observed effects are isolated to the ET region. At higher concentrations, more severe effects
occur in the ET region and toxicity is also observed to progress to the distal regions. The
intensity or severity of the distal toxicity also progresses with increased exposure
concentrations.
In the following section, the conceptual framework that directs development of
dosimetry models is discussed. The framework is constructed according to the categorization
scheme of gases based on physicochemical characteristics. The physicochemical
characteristics are used to define dominant mechanistic determinants of absorption and
thereby determine the mathematical model structure to describe regional dosimetry. The
model structures developed in the framework rely on models that are currently in use; a
detailed review of potential structures is presented elsewhere (Ultman, 1988) and some are
incorporated here. Description and derivation of the model structure for each category of gas
follows with the exception of gases that are relatively insoluble in water (Category 3). The
uptake of Category 3 gases is predominantly perfusion-limited and the dosimetry approach for
these is discussed in Appendix J. Thus, the focus of this appendix is on those gases that are
relatively water soluble and/or reactive in the respiratory tract. It should be noted that the
definition of reactivity includes both the propensity for dissociation as well as the ability to
serve as substrate for metabolism in the respiratory tract. The default equations are derived
after the development of the modeling structure for gases in Categories 1 and 2. These
equations result from the application of further simplifying assumptions necessary to reduce
1-3
-------
p. 37
the required parameters to perform the dosimetry adjustment when minimal data are
available.
I.I Conceptual Framework
Extrapolation of the dose-response relationship from laboratory animals to humans is
performed based on the absorption in the three respiratory tract regions as defined in
Chapter 3: extrathoracic (ET), tracheobronchial (TB), and pulmonary (PU). Although toxic
effects may sometimes be observed in a more local area within those regions (e.g., the
olfactory epithelium of the ET region), the parameters required to further subdivide the
description of dose within these regions are not available currently. Several active areas of
investigation, such as the evaluation of regional mass transport within the nasal cavity to
create maps of localized flows in rats and monkeys (Kimbell et al., 1993), of regional mass
transport in the human (Lou, 1993), and of metabolic activity of localized tissues in rodents
(Bogdanffy et al., 1986, 1987, 1991; Bogdanffy and Taylor, 1993; Kuykendall et al., 1993),
are anticipated to provide the data required to estimate the necessary parameters on a species-
specific basis.
The conceptual framework used to direct development of model structures for estimation
of regional gas dose is based on the categorization scheme presented in Section 3.2.2. This
categorization scheme is based on the physicochemical characteristics of water solubility and
reactivity as shown in Figure 1-1. These characteristics are used to define dominant
mechanistic determinants of absorption and thereby direct development of model structures.
As will be described, the modeling structure favored for this development has been used
extensively to quantify gas exchange or absorption of pollutants. This structure is in no way
promoted exclusively as the only one available; it is however used here to develop the
approach for dosimetric adjustment. Its application to each category will be presented and
the default equations for use with limited parameters will be derived.
1.1.1 Category Scheme for Gases with Respiratory Effects
This appendix focuses on those gases that are relatively water soluble and/or reactive in
the respiratory tract (i.e., those gases in Categories 1 and 2, initially defined in Section 3.2).
1-4
-------
p. 38
•. i ••:.»!
Reactivity.
Gas Category Scheme
Category 1: Do not penetrate to blood
(e.g., highly water soluble/
rapidly reactive)
Category 2: Water soluble/Blood
accumulation
Category 3: Water insoluble/
Perfusion limited
Location
• Extrathoracic absorption
ffl Entire tract absorption
D Predominantly pulmonary
absorption
Figure 1-1. Gas categorization scheme based on water solubility and reactivity as major
determinants of gas uptake. Reactivity is defined to include both the
propensity for dissociation as well as the ability to serve as substrate for
metabolism in the respiratory tract. Definitive characteristic of each
category and anticipated location (region) for respiratory tract uptake are
shown.
Those gases which are relatively insoluble in water, principally absorbed in the PU region,
and distributed remote to the respiratory tract (Category 3) are addressed in Appendix J.
There are two points of departure between the treatments of Appendix I versus Appendix J:
(1) uptake of Category 1 and 2 gases (Appendix I) is limited by absorption in the gas or
surface-liquid/tissue phase, whereas uptake to the blood from the airspace for gases in
Category 3 (Appendix J) is described by the blood-to-air partition coefficient only; and
(2) Appendix I considers absorption in the regions proximal to the PU region.
1-5
-------
p. 39
Two categories of gases with potential respiratory effects at the uptake site have been
identified for simplifying the methods for dose determination. The categories separate the
gases on the basis of the physicochemical absorption parameters and the consequent dominant
determinants of absorption. The two categories of gases with potential respiratory effects are
(1) highly water soluble and/or rapidly irreversible reactive gases; and (2) water soluble gases
and gases that may also be rapidly reversibly reactive or moderately to slowly irreversibly
metabolized in respiratory tract tissue.
The gases in Category 1, highly water soluble and rapidly irreversibly reactive, are
distinguished by the lack of a blood-phase component to the transport resistance (i.e., almost
none of the gas reaches the bloodstream), which allows the overall transport to be described
by the transport resistance through air and liquid/tissue phases only (i.e., the two-phase
transport resistance model). Examples of gases in this category are hydrogen fluoride,
chlorine, formaldehyde, and the volatile organic acids and esters.
Gases in Category 2 are distinguished from those in Category 1 by the potential for
accumulation of a significant blood concentration that could reduce the concentration gradient
driving the absorption process and thereby reduce the regional absorption rate. In addition,
the accumulated blood or surface liquid/tissue concentration may impose a backpressure (i.e.,
a significant reverse in the concentration gradient) during exhalation which could result in
desorption. Category 2 gases may be further subdivided by distinguishing between those that
react reversibly with the surface liquid or underlying tissue from those that react irreversibly.
A gas that is moderately to slowly irreversibly metabolized in the respiratory tract will
effectively reduce tissue concentration and thereby increase the concentration gradient during
absorption and decrease it during desorption. In contrast, reversible reactions will not affect
the gradient dramatically. Consequently, in the case of irreversible reactions, the reaction
rate may need to be included in the model. In the case of Category 2 gases undergoing a
reversible reaction, the reaction may be incorporated into the model by the use of an
enhanced solubility term. Examples of Category 2 gases are ozone, sulfur dioxide, xylene,
propanol, and isoamyl alcohol.
General physicochemical properties of the gases have been used to delineate each of the
categories. The boundaries between categories are not definitive. Some compounds may
appear to be defined by either Category 1 or Category 2 because water solubility and
1-6
-------
p. 40
reactivity are a continuum. Thus, although sulfur dioxide is reversibly reactive, which would
categorize it as a Category 2 gas, it is also highly soluble such as to be a Category 1 gas.
Similarly, ozone is highly reactive yet only moderately water soluble. More explicit
delineation of the categories will be made upon review of the empirical data and the
predictability of the model structures for gases that may appear to fit more than one category.
The modeling approach for the determination of dose for each of these categories of gases is
discussed separately in the following sections, along with the determination of the default
methods if sufficient detail from which to determine dose is not available for a specific gas.
1.1.2 General Model Structure
Numerous model structures have been used to describe toxicant uptake in the respiratory
tract. The structures range from compartmental models, such as physiologically based
pharmacokinetic (PBPK) models in which spatial details are ignored, to distributed parameter
models, such as the finite difference models of McJilton et al. (1972) and Miller et al.
(1985). The finite difference models have been applied to specific gases, but a generalized
structure was developed by Hanna et al. (1989) for water soluble gases. Several reviews of
the various structures are available (Morgan and Frank, 1977; Ultman, 1988, 1994).
Methodologies to describe respiratory uptake of gases have been successfully applied by
using two types of empirical compartmental models. These models are distinguished by the
gases to which they have been applied. The ventilation-perfusion model first applied to the
exchange of carbon dioxide/oxygen (CO2/O2) in the lung periphery has been principally and
most successfully employed to describe the stable and less soluble gases. The modeling of
the respiratory tract using the ventilation-perfusion model has become a central component of
PBPK models as described in Appendix J (Ramsey and Andersen, 1984; Andersen et al.,
1987a; Overton, 1989; Andersen et al., 1991). In a ventilation-perfusion model (or Bohr
model), the mass of inhaled chemical reaching the lung periphery, or PU region, is calculated
as the product of the ambient concentration and the alveolar ventilation rate. The
ventilation-perfusion model would overpredict the gas concentration that reaches the alveoli if
the gas is absorbed or reacts with the ET and TB airway surface liquid and/or tissue.
The second type of model was developed to describe the fraction of an absorbing or
reacting gas that penetrates the ET region. This model, which will be referred to as the
1-7
-------
p. 41
penetration fraction model, was first used by Aharonson et al. (1974) to demonstrate
empirically the different upper airway absorption efficiencies for gases with differing
physicochemical properties. This modeling concept has since been utilized by Kleinman
(1984), Morgan and Frank (1977), Ultman (1988), Hanna et al. (1989), Gerde and Dahl
(1991), and Morris and Blanchard (1992). A principal focus of these modeling efforts has
been to predict the scrubbing efficiency of the ET airway based on the ventilation rate and the
physicochemical properties of the gas. However, the general applicability of the penetration
model has often been limited by the assumption that the gas blood concentration approaches
zero, thereby requiring complete systemic elimination. Retaining the blood concentration in
the model allows greater flexibility for inclusion of the reduction in the concentration
gradient, which would reduce the absorption rate if the gas were to accumulate in blood.
In this conceptual framework, the methodology to adjust regional respiratory dose from
laboratory animals to humans for evaluation of respiratory tract effects is achieved for the
relatively water soluble and/or reactive gases (Categories 1 and 2) by integrating the above
two types of empirical models. These models have been used extensively and are therefore
favored due to their wide use and potential for empirical measurement of model parameters.
The penetration fraction model provides estimation of the ET and TB doses. These are used
to adjust the mass of inhaled gas reaching the PU region in the ventilation-perfusion model.
Additional systemic compartments (e.g., liver and fat) may be required in the model to
describe gases that accumulate significantly in the blood. The addition results in a model
structure similar to PBPK models; however, it also incorporates the mass transport
description of the scrubbing of the gas in the ET and TB regions.
The approach herein to determine the regional dose within the respiratory tract is
developed by relying on the overall mass transport coefficient, Kg, to characterize the
transport of gases between the airphase, the intervening surface liquid and tissue, and the
blood. In the absence of empiric measurement, K may be estimated or scaled for a given
D
gas based on its physicochemical properties and reactivity within the respiratory tract. In the
following section, a derivation of K is provided and the influence of gas physicochemical
o
characteristics on K is discussed. The definitions of parameter symbols used in the
o
following sections are provided in Table 1-1.
1-8
-------
p. 42
TABLE 1-1. DEFINITION OF PARAMETER SYMBOLS USED IN APPENDIX I
a
C0
C*
c.OO
Q,
cf
c,
C
LG
CA
CLfat
CLLIV
CLSYs
CV
CX(EXH)CT
CX(EXH)PU
CX(EXH)TB
CX(INH)CT
CX(INH)TB
D
A
dx
dy
dz
Airway perimeter (cm2)
Initial concentration (mg/cm3)
Pulmonary region gas concentration (mg/cm3)
Gas concentration as a function of x (mg/cm3)
Blood concentration (mg/cm3)
Gas concentration in equilibrium with blood concentration (mg/cm3)
Concentration of gas in its chemically transformed (reacted) state (mg/cm3)
Concentration in the fat compartment (mg/cm3)
Gas phase concentration in airway lumen (mg/cm3)
Gas-phase concentration at the interface of the gas phase with the surface-liquid/tissue phase
(mg/cm3)
Inhaled concentration (mg/cm3)
Surface-liquid/tissue phase concentration (mg/cm3)
Concentration in the lung compartment (mg/cm3)
Surface-liquid/tissue concentration in equilibrium with the gas phase (mg/m3)
Surface-liquid/tissue concentration at the interface of the gas phase and the
surface-liquid/tissue phase (mg/cm3)
Imposed concentration (mg/cm3)
Concentration of reacted and unreacted gas in arterial blood (mg/cm3)
Concentration of reacted and unreacted gas in venous blood (mg/cm3)
Concentration in the surface-liquid/tissue phase (mg/cm3)
Arterial (unoxygenated) blood concentration (mg/cm3)
Clearance from the fat compartment (cm2/min)
Clearance from the liver compartment (cm2/min)
Clearance from the systemic compartment (cm2/min)
Concentration in venous (oxygenated) blood entering gas-exchange (PU) region (mg/cm3)
Concentration exiting from extrathoracic region on exhalation (mg/cm3)
Concentration exiting from pulmonary region upon exhalation (mg/cm3)
Concentration exiting from tracheobronchial region upon exhalation (mg/cm3)
Concentration exiting from extrathoracic region upon inhalation (mg/cm3)
Concentration exiting from tracheobronchial region upon inhalation (mg/cm3)
Deposited fraction of mass (unitless)
Liquid diffusivity (cm2/min)
Differential of axial distance into airway (cm)
Differential of axial distance into capillary segment (cm)
Differential of distance into the surface-liquid/tissue phase (cm)
Elimination rate in the lung compartment (cm2/min)
Maximum extraction efficiency (unitless)
1-9
-------
p. 43
TABLE 1-1 (cont'd). DEFINITION OF PARAMETER SYMBOLS
USED IN APPENDIX I
Ef Liver extraction efficiency (unitless)
erf Error function (unitless)
ET Extrathoracic respiratory region
F Flux fraction (unitless)
fp Fractional penetration (unitless)
fpgr Fractional penetration through the extrathoracic region (unitless)
fppu Fractional penetration through the pulmonary region (unitless)
fpTB Fractional penetration through the tracheobronchial region (unitless)
Ha Hatta number (unitless)
HWg Blood:gas (air) partition coefficient (unitless)
HEFP Effective partition coefficient (unitless)
H^ Tissue:blood partition coefficient (unitless)
H,/g Surface-liquid/tissue:gas (air) partition coefficient (unitless)
Kg Overall mass transport coefficient (cm/min)
Kg Overall mass transport coefficient of the extrathoracic region (cm/min)
K Overall mass transport coefficient of the pulmonary region (cm/min)
K Overall mass transport coefficient of the tracheobronchial region (cm/min)
^TB
kg Transport coefficient in the gas phase (cm/min)
k, Transport coefficient in the surface-liquid/tissue phase (cm/min)
kLG Elimination rate from lung compartment (min"1)
k,,, Alveolar membrane diffusion coefficient (cm/min)
kr Reaction rate constant in the blood or tissue (min'1)
KM Michaelis constant (mg/cm3)
L Airway length (cm)
Md Desorbed mass (mg)
MdBr Desorbed mass from extrathoracic region (mg)
MdPU Desorbed mass from pulmonary region (mg)
MJTB Desorbed mass from tracheobronchial region (mg)
MCT Mass flux from extrathoric region to blood (mg/cm2-min)
Mpu Mass flux from pulmonary region to blood (mg/cm2-min)
MTB Mass flux from tracheobronchial region to blood (mg/cm2-min)
N Overall transport or flux (mg/cm2-min)
NR Flux through the air phase (mg/cm2-min)
Nj Flux through the surface liquid-tissue phase (mg/cm2-min)
PU Pulmonary respiratory tract region
Qiv Alveolar ventilation rate (cm3/min)t
1-10
-------
p. 44
Or
ROD
RGDR
CT
RGDRTB
SA
SACT
SA™
SAPU
TB
V
Vb
VLG
VMAX
VE
x
Ay
z
Az
TABLE 1-1 (cont'd). DEFINITION OF PARAMETER SYMBOLS
USED IN APPENDIX I
Local blood flow rate (cm3/min)t
Cardiac output (cm3/min)t
Regional gas dose (mg/cm2-min)
Regional gas dose ratio for the extrathoracic region (unitless)
Regional gas dose ratio for the pulmonary region (unitless)
Regional gas dose ratio for the tracheobronchial region (unitless)
Surface area of unspecified respiratory region (cm2)
Surface area of the extrathoracic region (cm2)
Surface area of the tracheobronchial region (cm2)
Surface area of the pulmonary region (cm2)
Blood perfusion surface area (cm2)
Time (min)
Time (duration) of exhalation (min)
Tracheobronchial respiratory tract region
Volumetric flow rate (mg/min)
Capillary blood volume (cm3)
Lung compartment volume (cm3)
Maximum velocity of saturable (Michaelis-Menton) metabolism path (mg/cm2-min)
Minute volume (cm3/min)t
Distance into the airway (cm)
Thickness of the surface liquid-tissue layer (cm)
Distance into the surface-liquid/tissue phase (cm)
Surface-liquid/tissue phase thickness (cm)
f 1 mL = 1 cm3, so cmVmin = mL/min.
Also note that concentrations are expressed as mg/cm3 (1 mg/cm3 = 10"6 mg/m3).
1.1.3 Overall Mass Transport Coefficient
The concept of the overall mass transport coefficient is based on a concentration
gradient analysis similar to Pick's Law of Diffusion and is utilized to describe transport
through several different phases such as air and liquid. The structure of the two-phase mass
transport resistance model simplifies the description of mass transport to a minimal number of
parameters that may be scaled to gases differing in their physicochemical properties as
described later. The more definitive evaluation of transport is to describe absorption by a
Ml
-------
p. 45
simultaneous solution of the conservation of momentum and mass in the complex three-
dimensional airway and tissue structure, which has yet to be performed in the respiratory
tract. A finite difference solution of Pick's Law has been obtained in the TB and PU region
by assuming no gas-phase component to mass transport, which eliminates the solution to the
momentum equation (Miller et al., 1985). To include the gas-phase component, k , Hanna
et al. (1989) and Lou (1993) used empirically determined k values in conjunction with
conservation of mass in the liquid phase to solve for local absorption rates in a finite
difference model.
Finite difference solutions are numerically intensive, however, and must be solved for
each gas. Scaling of the transport coefficients based on the physicochemical properties of the
gas thereby allows scaling of the absorption rate without labor intensive calculations.
Furthermore, the transport coefficients may be determined empirically, reducing concern for
the appropriateness of the modeling assumptions. Two-phase mass transport resistance
models incorporating overall mass transport coefficients have been used in other applications,
such as the evaluation of atmospheric absorption of gases by aerosols (Seinfeld, 1986),
volatilization or absorption of gases by surface water bodies (Lyman et al., 1990), operation
of air strippers (Perry and Chilton, 1973), and absorption in the respiratory tract (Miller
et al., 1985; Hanna et al., 1989).
To simplify the respiratory tract into a two-phase resistance model for illustration of the
overall mass transport approach, it must be assumed that the blood concentration is constant.
For very reactive gases, such as ozone, it can further be assumed to be zero. Under these
conditions, the transport of the gas would occur primarily through the air phase and surface-
liquid/tissue phase. It is assumed that the surface-liquid and tissue phases are a single phase
because of the limited data to identify differing transport parameters for these two phases.
The two-phase transport is shown in Figure 1-2. The overall transport or flux, N, through
these two phases is expressed by
N = Kg(Cg - C1/g) (M)
where C is the bulk gas phase (or air phase) concentration, and Cj/g is the gas phase
concentration in equilibrium with the bulk surface-liquid/tissue phase concentration, Cj
1-12
-------
p. 46
Airway Lumen
Gas Phase
Surface-Liquid / Tissue Phase
Distance From
Interface
Surface
Blood
C=C=0
(k,, , A AZ)
Figure 1-2. Schematic of two-phase mass transport resistance model. The definitions
for the parameter symbols are provided hi Table 1-1.
1-13
-------
p. 47
(Perry and Chilton, 1973), such that q/ is .equal to the ratio of the surface-liquid/tissue
concentration, Cj, to the gas partition coefficient, Ht/ .
The overall mass transport coefficient may be determined from the transport coefficients
of each individual phase. It is obtained by considering the flux through each phase (Perry
and Chilton, 1973) such that
where N is the flux through the air phase, k is the transport coefficient in the gas phase,
and Cgi is the gas-phase concentration at the interface of the gas phase and the surface-liquid
tissue phase, and
- q) d-3)
where Nj is the flux through the surface liquid-tissue phase, kj is the transport coefficient in
the surface-liquid/tissue phase, and Cy is the surface-liquid/tissue concentration at the
interface of the gas phase and the surface-liquid/tissue phase.
Steady state (or quasi-steady state as occurs in the respiratory tract during inhalation or
exhalation) requires the following condition:
N = Ng = Nj, (1-4)
o
or
N = Kg(Cg - C1/g) = kg(Cg - Cgj) = IqCCy - C,). (1-5)
In the two-phase resistance approach defined by Equation 1-5 above, the overall mass
transport resistance is defined by the reciprocal of the mass transport coefficient, 1/K , and is
composed of the resistance to lateral movement of the absorbing gas through the air and
1-14
-------
p. 48
through the liquid and tissue as shown in Figure 1-2. The resistance in series can be derived
from Equation 1-5 as
(1-6)
Kg kg Ht/gkj
In the case where the surface liquid and tissue cannot be assumed to be a single compartment,
a separate partition coefficient and transport coefficient would need to be incorporated into
Equation 1-6.
The definition of the overall mass transport coefficient provided in Equation 1-6 may be
used to evaluate the conditions in which a single phase, either gas phase or surface-
liquid/tissue phase, determines the overall mass transport coefficient. To demonstrate
predominance of a single phase, it is further assumed that blood flow does not contribute to
the overall mass transport coefficient (i.e., that there is no accumulation in blood). In the
case of a reactive gas and/or a gas relatively soluble in both the tissue and blood, the
transport resistance through the gas phase, 1/k , may be greater than the transport resistance
in the other phases (i.e., k « Ht/gkj) such that
l/Kg - l/kg. (1-7)
The gas phase term, k , is dependent on flow rate, flow geometry, and the gas phase
o
diffusivity. In cross-species comparisons, the flow geometry differences of the species are
likely to predominately determine k . Additionally, recent data found that k differed
o o
significantly between living subject geometry and cadaver geometry so that it is reasonable to
expect geometry to affect interspecies differences (Lou, 1993).
Liquid phase controlled absorption (i.e., Ht/ kj « k ) is typically identified by a gas
of moderate to low water solubility and low reactivity. In the case of a nonreactive gas,
Ht/ kj may be approximated by the surface-liquid/tissue:gas partition coefficient, the liquid
diffusivity (Z>j), and the thickness of the liquid-tissue layer (Ay), such that
1-15
-------
p. 49
For reactive gases, k| would need to be evaluated to include the transformation rate (Bird
et al., 1960; Perry and Chilton, 1973; Ultman, 1988). However, as the reactivity increases,
it is less likely that the absorption rate will remain liquid phase controlled due to the
increasing influence of the gas phase. In the case of a liquid-controlled absorption process,
Hf/gkj may be substituted for kg in Equation 1-7.
As discussed, each of the transport coefficients is dependent on the transport properties
of the gas within the respective phase that alter the concentration gradient indicated in
Figure 1-1. Thus, in the case of the gas phase mass transport coefficient, k , the mass
o
transport is affected by the flow rate (ventilation rate), the gas phase diffusivity, and the local
(regional) airway geometry. The dependence of k on these parameters is discussed in
greater detail by Hanna et al. (1989) and Lou (1993). The surface-liquid/tissue phase
transport coefficient is determined by the phase thickness (Az), the liquid phase diffusivity,
and the reactivity (e.g., ionic dissociation and metabolism) in the surface-liquid/tissue. The
dependence of kj on these parameters is discussed in greater detail by Ultman (1988).
The penetration fraction model may be used to empirically determine the overall mass
transport coefficient (Section 1.2.1), provided the fractional penetration, fp, is measured.
Because fp is both gas and species specific, the K value will similarly be gas and species
specific. However, data for fp and K specific to a gas or gases may be used in a predictive
fashion by scaling to the physicochemical properties of solubility, diffusivity, and reactivity.
Using values of K obtained in a single species, K can be scaled within the species for
a different gas by decomposing K to the individual transport resistances (i.e., the gas phase
and surface-liquid/tissue phase mass transport coefficient). In humans, empirical measures of
k have been obtained in casts of the human nasal cavity (Nuckols, 1981; Hanna and Scherer,
o
1986; Lou, 1993) and can be used to decompose K . Although k is species specific, it may
be scaled to other gases by scaling to the gas diffusivity (Hanna et al., 1989). Therefore, for
a gas in which k « Ht/ kj, this scaling may be sufficient to predict fp and dose. Similar
scaling using the diffusivity and reactivity of a gas for which fp is unknown may be
performed for the surface-liquid/tissue phase transport coefficient, kj. For gases in which the
1-16
-------
p. 50
prediction of K , and therefore fp, depends on the surface-liquid/tissue phase, the solubility
and reactivity of the gas must also be used in the scaling (Equation 1-6).
A difficulty arises due to the lack of k values in airways of laboratory animals.
o
Decomposition of an empirically determined K to the individual components must therefore
be made based on a data base in which Ht/ kj may be determined. An approach is under
development to obtain data from several gases to decompose Kg into each component and
perform an evaluation of k for gases in each category to obtain a measure of consistency
&
within a species. This effort is underway and will be published as a technical support
document to this publication.
Absorption within the respiratory tract cannot always be assumed to be modeled by a
two-phase transport resistance model ignoring the blood concentration. In cases where the
absorption in the blood contributes to overall absorption, additional mass transport resistances
must be considered. Accumulation in the bloodstream may reduce the concentration driving
force (and thereby reduce the absorption rate) as well as contribute to the development of a
back pressure, which may result in desorption during exhalation due to the reversal in the
concentration gradient between the air and tissue. Gases that are likely to exhibit these
characteristics are Category 2 gases that are water soluble and rapidly reversibly reactive or
those that are moderately to slowly irreversibly metabolized in the respiratory tract,
previously referred to as "transition gases" (Dahl, 1990). The contribution of the blood to
both the overall mass transport resistance and to the potential for desorption during exhalation
was considered in the categorization of gases based on their physicochemical properties. The
categorization of gases with respiratory effects is used as the basis for defining the model
structure and, in particular, the overall mass transport coefficient.
1.2 MODEL FOR CATEGORY 1 GASES
Category 1 gases are defined as gases that are highly water soluble and/or that may
react rapidly and irreversibly in the surface liquid and tissue of the respiratory tract. Due to
these physicochemical characteristics, these gases are distinguished by the property that a
significant back pressure from the surface-liquid/tissue phase during exhalation does not
develop. A back pressure resulting from the reversal in the concentration gradient between
1-17
-------
p. 51
the gas and the surface-liquid/tissue phase may cause significant desorption during exhalation
which would require an adjustment to the dose as is considered in the model for Category 2
gases. Category 1 gases are further distinguished by the property that the gas does not
significantly accumulate in the blood, which would reduce the concentration driving force and
hence reduce the absorption rate. Examples of gases classified as Category 1 are hydrogen
fluoride, chlorine, formaldehyde, and the volatile organic acids and esters.
In the following, the two empirical models discussed above are synthesized to allow the
doses of gases of differing physicochemical properties to be scaled across species. The
penetration fraction model will be utilized to determine the fraction absorbed in the ET
region, the concentration entering and dose to the TB region, and the remaining concentration
entering the PU region. The ventilation-perfusion model is used to evaluate dose to the PU
region by substituting the concentration of the air exiting the TB region in place of the
ambient concentration. The overall schematic for the approach is shown in Figure 1-3. The
definitions for the parameter symbols are provided in Table 1-1.
Blood
Extrathoracic
Region
Tracheobronchial
Region
Pulmonary
Region
Figure 1-3. Schematic of modeling approach to estimate regional respiratory tract dose
of gases. The definitions for the parameter symbols are provided in
Table 1-1.
1-18
-------
p. 52
1.2.1 Extrathoracic Region: The Penetration Fraction Model
The penetration fraction model, designed to evaluate the upper airway scrubbing
efficiency, is based on a mass-balance approach. The change in mass traversing the gas
phase of the extrathoracic region is balanced by the mass absorbed at the gas-liquid interface
of the airway. This balance is written as
where V is the volumetric flow rate; dC /dx is the rate of change of the airstream
concentration (gas phase) as a function of distance into the airway, x; K is the overall
SET
mass transport coefficient between the airstream and the blood in the ET region; a is the local
airway perimeter; Cj is the inspired gas concentration; and Cb/g is the gas concentration that
would be in equilibrium with the blood concentration. Cb/ is equal to the ratio of the blood
concentration, Cb, to the blood:gas (air) partition coefficient, Hb/ .
To evaluate the change in concentration over the length of the ET region, Equation 1-9
is integrated resulting in the following relationship:
-K aL
SET
, s
(CX(INH)ET-Cb/g) <— — > (1-10)
where CX(INH)ET is the gas concentration exiting the ET region and L is the length of the
airway such that the product of a and L is the surface area of the ET region, SAET.
Equation I- 10 indicates that CX(INH)ET will equal Cj at an infinite volumetric flow rate.
In the case of Category 1 gases/vapors, CX(INH)ET and Cj are much greater than Cb/g,
so that Equation I- 10 can be further reduced to
CX(INH)ET y (Ml)
- - e
1-19
-------
p. 53
where fpET is the penetration fraction through the ET region and is given as the ratio of the
gas concentration exiting the region, CX(INH)ET, to the gas concentration entering the
airway, Cj. The relationship shown in Equation I-11 suggests that the product of the overall
mass transport coefficient and the surface area may be obtained by plotting fpET as a function
of volumetric flow rate. Indeed, many investigators have used this method to present
empirical results (Aharonson et al., 1974; Kleinman, 1984; Morris and Blanchard, 1992).
As an example, provided that SAET is known, Equation 1-11 may be used to evaluate K
in the form of
In fpET = ~ Kg SA
^ET
ET
V
where K SAET is the slope if the relationship between In fpET and 1/V is linear.
Equation 1-12 is similar to the relationship developed by Morris and Blanchard (1992).
Morris and Blanchard chose to fit D/fpET to 1/V, where D is the deposited fraction, l-fpET.
Using Equation 1-12 in conjunction with a power series expansion of the exponential term of
Equation I-11 results in
D KgCTSAET
(1-13)
It should be noted, however, that plotting either In fpET or D/fpET against I/ V may not
be linear. The nonlinearity was first reported by Aharonson et al. (1974). Both Ultman
(1988) and Hanna et al. (1989) attribute the nonlinearity to the contribution of the gas phase
mass transport coefficient, k , to the overall transport rate. Thus, K is a function of
V when affected by k , thereby producing the nonlinearity.
Equation 1-6 may be used to evaluate Ks if sufficient information is available to
*ET
calculate the individual mass transport coefficient for each phase. Empirical determinations
of K may also be obtained from Equation 1-12. Furthermore, in the case of gas phase
SET
controlled absorption (i.e., in the case of Category 1 gases) where K EX ~ kgET>
Equation 1-12 can be used to evaluate the gas phase mass transport coefficient (k ) for each
species.
1-20
-------
p. 54
To evaluate k for a single species, empirical measures of the fractional penetration of a
gas in which the gas phase contributes to (or controls) the overall mass transport resistance
must have been determined empirically. The fractional penetration obtained at several flow
rates may be used to evaluate k from the following relationship, which is obtained by
combining Equations 1-12 and 1-6 such that
F
Kg
The value of k and its functional dependence on V is determined by curve fitting the
empirically determined fpET and V, provided that Ht/ and kj are known. In the case of a
nonreactive gas, kj may be simply estimated. Methods are also available to estimate kj for
reactive gases (Ultman, 1988). It should be noted that kg will differ among gases and that
k is a function of ventilation rate. Therefore, k must be scaled by the gas phase diffusivity.
The ventilation dependence of kg allows the two terms, gas phase and surface-liquid/tissue
phase, to be separated. Thus, k may be evaluated from data of fpET of several gases to
obtain a reasonable estimate of k in a single species, particularly for the rat, for which
empirical data are most available. Values of k in other species may also be obtained from
o
published uptake data (Morris and Smith, 1982; Stott and McKenna, 1984; Morris et al.,
1986, 1991; Morris and Cavanagh, 1987; Morris, 1990; Dahl et al., 1991b; Bogdanffy
et al., 1991; Morris and Blanchard, 1992; Bogdanffy and Taylor, 1993; Kuykendall et al.,
1993).
By using the heat and mass transfer analogy, measures of change in inspired
temperature may be used to obtain an independent estimate of k (Hanna et al., 1989). The
o
mass transport coefficient, k , has also already been determined in human casts based on both
o
mass transport studies (Hanna et al., 1989; Lou, 1993) and heat transport studies (Nuckols,
1981) from which k is directly calculated.
If the absorption is gas phase controlled (i.e., absorption is completely determined by
transport through the gas phase), Equation 1-12 may be used to determine the gas phase mass
transport coefficient for a single species such that
1-21
-------
p. 55
8 SAET
where kg is substituted for Kg. Because k is a function of flow rate, a plot of In fpET
against I/ V will be nonlinear. The nonlinearity determined from empirical data of a gas
phase controlled absorption process can be used to evaluate the flow rate dependence of k .
The flow rate dependence is of the form Vn where n is typically between 0.5 to 0.8 (Hanna
et al., 1989). Note also that k can not be determined if there is no penetration (i.e.,
fpET = 0) because In fpET is undefined. The value of k at a specific flow rate and in a
single species should be relatively constant, changing only slightly as a function of the gas
diffusivity. Using values for k determined in humans and in a laboratory animal species
allows the scaling for dose in gas phase controlled absorption (i.e., where k « Ht/ kj).
In the case of surface-liquid/tissue phase controlled absorption, where K ~ Ht/ kj, the
value will be chemical-specific due to the dependence of Ht/Jcj on solubility and reactivity.
Under these circumstances, Ht/ kj would replace k in Equation 1-14. The regression of
In fpx to 1/V would be linear provided that the reactivity is not saturable. Michaelis-Menton
kinetics can be used to define kj and incorporate saturation kinetics which may introduce a
nonlinearity. However, saturation kinetics may be better described by the model for
Category 2 gases in which there may be a significant accumulation in blood thereby reducing
the absorption concentration gradient during inhalation, as well as a potential reversal of the
concentration gradient, which would result in desorption during exhalation.
The rate of mass absorbed at the gas-surface interface of the airway in a region is
simply the product of the absorbed fraction, 1-fpE-p, and the total mass inhaled during a
single breath, VCj. With this knowledge, a suitable metric of dose must now be chosen.
If dose were to be defined on a mass per volume basis, it would implicitly assume that the
outcome would be determined by concentration (i.e., mass/volume). This assumption would
therefore argue that the most appropriate definition of dose should be one defined on the basis
of surface area (i.e., the mass flux, or dose, defined as mass per surface area-time). The
mass flux implies a concentration gradient in the tissue such that the localized concentration
would be highest at the surface. The mass flux is thereby a more accurate predictor of the
peak localized concentration and will be used to define dose for this application. The
1-22
-------
p. 56
/•*
regional gas dose (RGD), defined as the mass absorbed per surface area per minute (mg/cm -
min), to the extrathoracic region (ET) is given by
C V
RGDET (mass /cm2 - min) = (1 - fpET) —!—.
SAET
From Equation 1-11, the regional gas dose to the ET may also be expressed as
CV v (1-17)
RGDET = —1—(1 - e E ).
SAET
The dose expressed in Equation 1-17 is applicable to any animal species provided the
appropriate parameters of that species are used in the assessment. For example, because the
purpose of this appendix is to address extrapolation of respiratory effects from experimental
animal species to humans, the minute volume (VE) is used as the default volumetric flow rate
in the remainder of the derivations because it approximates the flow rate at which the animal
was breathing during the experimental exposure. Further justification of the use of minute
volume is the relatively little desorption that occurs during exhalation, a requirement by
definition of Category 1 gases, suggesting that the dose should be averaged over the entire
cycle. The default values for surface area and minute volume for the various species are
provided in Chapter 4.
The regional gas dose ratio for the extrathoracic region (RGDRET) of differing species
used to calculate NOAEL(HEC) can also be developed using Equation 1-17. A comparison
between humans and an experimental test species would result in the following regional gas
dose ratio (RDGR):
-K SA_
*ET ^T
qvE —^~
renn ^> (SA~)A(I ~e )A
RGDRET - (RGDET)A = J^E , d-18)
ET (RGDET)H -KgErSACT
C:V —v
e - )H
1-23
-------
p. 57
where the subscript A and H refer to values for laboratory animals and humans respectively.
Because it is assumed that the laboratory animals and humans are exposed to the same
concentration for purposes of extrapolating the observed toxicity, Cj can be deleted.
Equation 1-18 represents the most general form of the ratio of ET regional dose between
laboratory test species and humans for Category 1 gases. This equation will therefore serve
as the basis for the default dosimetric adjustment. To evaluate the ratio, each term will need
to be determined for the species of interest.
By definition, gases in Category 1 would be associated with large K values due to high
k and low Ht/ kj terms (Ultman, 1988). Thus, in these cases the exponent is greater than or
equal to 1. Under these circumstances, the exponential term (the penetration fraction)
approaches zero (less than 5% error when K SAET/VET is 3) and the RGDRET is simply1
VE
(RGDPT)A SAFT
RGDRFT = - _ El^ . _ E_ . (1-19)
(RGDET)H V
E
RGDRET is determined by the ratio of ventilation rates and surface areas in each species.
Assuming that the penetration fraction (i.e., the exponential term) reduces to zero is
equivalent to assuming the gas is absorbed entirely in the ET region. Furthermore, based on
Equation 1-16, the absorption is assumed to be distributed equally. Studies currently in
progress are anticipated to provide more localized measures of dose to the nasal cavity
(Kimbell et al., 1993; Lou, 1993).
'Note that Equation 1-19 may also be derived from Equation 1-18 by determining the conditions whereby
SAET
-K
"ET y
E \
A
E )
SAPT
-K -
SET y
(1 -exp E)H
These conditions will be a function of the default values for respiratory tract surface area and minute volume as well
as the absolute value of the overall mass transport coefficient.
1-24
-------
p. 58
1.2.2 Tracheobronchial Region: The Fractional Penetration Model
The penetration fraction model, and the analysis discussed in the previous section on the
ET region, may also be used to describe absorption in the TB region. The major difference
is that the concentration at the inlet of the TB region will be dependent on absorption in the
ET region. Therefore, the penetration fraction through the TB region, fpjg, may be
described using Equation I-11 such that
_ CX(INH)TB _ — (1-20)
" CX(INH)ET
where CX(INH)TB is the concentration exiting the TB region; CX(INH)ET is the
concentration exiting the ET region and subsequently entering the TB region; SATB is the TB
surface area; VE is the species-specific minute volume used in place of the volumetric flow
rate, due to averaging between inhalation and exhalation; and K is the overall mass
&TB
transport coefficient in the TB region. In the TB region, Kg and VE can be defined for
each specific bronchial generation or as a value for the entire region as a whole using the
trachea to characterize both parameters (Nuckols, 1981). Because measures of K are only
available regionally, the value for K as used in the remaining discussion should be
assumed to refer to a regional determination.
The regional gas dose to the TB region, RGDTB, may be defined similar to the ET dose
(Equation 1-17) such that
-K SA__
en ^B
CX(INH)ET VE v (1-21)
RGD = ™ E (1 -e VE ).
'TB
SA.
TB
The dose ratio for the TB region (RGDRTB) between an experimental animal species and
humans is therefore
1-25
-------
p. 59
(RGDTB)A _
(RGDTB)H
CX(INH)ET VE
SATB
fcX(INH)ET VE]
SATB
*^B~. «A—n
8TB 1B
A (1 - e VE )A
-^ SATB
•
H (1 - e )H
(1-22)
where A and H refer to laboratory animals and humans, respectively. Assuming the same
inhaled concentration, the above dose ratio will be divided by Cj/Cj resulting in the
concentration ratio (CX^NH^j/Cj) in the numerator and denominator for both laboratory
animal and human. This concentration ratio is fpET. The resultant gas dose ratio to the TB
region is thereby
RGDRTB =
(RGDTB)A _
(RGDTB)H
VE
SATB
'VE'
SATB
~K«-B SATB
A (^PET) A C1 ~ e E )A
(rpg-j-) H "~^STB ^^TB
H ^ ~ e E )H
(1-23)
Thus, the results obtained from an evaluation of ET penetration are used to determine the
dose to the TB region.
Similar to the ET region, Equation 1-23 may be simplified when K is large (less than
5% error if Kg SATB/ Vc is greater than or equal to 3) such that Equation 1-23 becomes2
2Note that Equation 1-24 may also be derived from Equation 1-23 by determining the conditions whereby
-K
_
«TB y
(1 -exp E
-K
SA.B
•1.
«TB
-exp
)H
These conditions will be a function of the default values for respiratory tract surface area and minute volume as well
as the absolute value of the overall mass transport coefficient.
1-26
-------
p. 60
RGDRTB =
(RGDTB) A
(RGDTB) H
SA
TB
SA
TB
(1-24)
H
1.2.3 Pulmonary Region: The Bohr Model
Ultman (1988) proposed a gas absorption model for the PU region by coupling the Bohr
model to predict expired air concentration with a model descriptive of the progressive
increase in the capillary blood concentration of the PU circulation. The steady-state model
was developed by a mass balance approach in which the rate of uptake in a capillary segment,
dy, is balanced by the differential increase in the total blood concentration, which includes
both the reacted (transformed) and the unreacted form of the absorbing gas. The PU
absorption model is given by
(Calv - Cb/g)dy/L = QT (dCb + dCb/r),
(1-25)
where, K is the overall PU transport coefficient from the gas phase to the blood; SAPU is
the PU surface area; Calv is the PU region gas concentration; Cb/ is the blood gas tension in
equilibrium with the blood concentration, Cb; QT is the cardiac output; and Cb/r is the
concentration of the gas in its chemically transformed state.
In the PU region, it is assumed that the absorption of the gas is not limited by
absorption in the bloodstream. Therefore, perfusion-limited absorption processes are not
considered in this appendix. As discussed earlier, perfusion-limited processes are more
appropriate for PBPK models such as described in Appendix J.
Equation 1-25 can be integrated such that
KgSAPu (caiv ~ cb/g) - QT (CT/V ~ CT/A)
PU
d-26)
1-27
-------
p. 61
where CT/V and CT/A are the concentration of the reacted and unreacted form of the
absorbing gas in the venous (oxygenated) and arterial (unoxygenated—entering the PU region)
blood, respectively. Consistent with the assumption that the blood concentration approaches
zero (i.e., eliminating perfusion-limited absorption), CT/A is assumed to be much greater than
CT/V. Under these circumstances, the right-hand side of Equation 1-26 is simply
(-QTCT/A).
The overall mass transport coefficient in the PU region, Ke , has been determined for
&PU
carbon monoxide (CO) to be of the form (Ultman, 1988)
KgSAPU kmSAPU
pu
(1-27)
where 1^ is the alveolar membrane diffusion coefficient and l^SApy is the alveolar
membrane diffusing capacity, kj. is the reaction rate constant in blood, and Vb is the capillary
blood volume. In the case of CO, the diffusion resistance (l/l^SApu) is three times greater
than the blood reaction term and the mass transport in the PU region is therefore limited by
the diffusion resistance. The PU diffusion capacity of CO may thereby serve as a reasonable
estimate of the diffusion resistance of a nonreactive gas. In the case of a gas reactive in the
PU tissue, K may be approximated by:
(1-28)
where Ha is the Hatta number, a dimensionless parameter that depends on the ratio of the
reaction to diffusion times. When the reaction is zero, the Hatta number is one and increases
as the rate constant increases (Ultman, 1988). An increase in the Hatta number thus reduces
the diffusion resistance and may increase the absorption rate.
The absorption rate to the bloodstream is also balanced by the change in airstream
concentration. Using the Bohr model (Figure 1-4), the mass balance of Equation 1-26 is
further expressed by:
1-28
-------
p. 62
cx
TB
Q
alv
Falv
cv
QT
CA
Figure 1-4. Bohr model of ventilation and uptake. The definitions for the parameter
symbols are provided in Table 1-1.
(CX(EXH)TB - Calv) =
(Calv - Cb/g) = - QTCT/A, d-29)
where Q^ is the alveolar ventilation rate.
The first two terms in Equation 1-29 are used to develop the ratio of the expired
concentration to inspired concentration, which will be used as a penetration fraction of the
PU region, fppy* defined as the ratio of Calv to CX(INH)TB such that
'alv
Oalv
CX(INH)TB K SApu (1 - Cb/g / Calv) + QaK
(1-30)
However, for the case of diffusion-limited absorption, Cb/ is much less than C^ such that
the denominator on the right hand side of the Equation 1-30 is simply (K PUSAPU +
1-29
-------
p. 63
The regional gas dose (ROD) to the pulmonary region (PU) is
RGDpu =(1 - fppu) alv CX(INH)TB. (1-31)
Combining Equation 1-30 and 1-31 results in the following relationship:
= (1 - Qalv . ) - CX(INH)TB, (1-32)
+ Qalv SAPU
where the regional gas dose ratio to the pulmonary region (RGDRPU) between laboratory
animal and humans is given by
(1 - ... ,0"v . )A
Qalv (Qaiv/SApu)A (CX(INH)TB)A
(Qalv/SApu)H (CX(INH)TB)H
^
-
=PU
Qalv
Dividing both numerator and denominator by the inspired air concentration converts the last
term to the product of the penetration fractions of the preceding regions such that
RGDR PU (Qalv/SA)A
Qalv , (Qalv/SA)H '
=PU
SAPU + Qalv
where the ratios (fpTB) /(fpj-g) and (fpET) /(fpEr) must ^e determined from the
A rl A rd
penetration fraction model for the TB and ET regions, respectively. Equation 1-34 may be
further reduced to
1-30
-------
p. 64
pu
RGDRpu . A . .
pu (RGDPU)H KgtuSAPU ft
from which the limiting values for the dose ratio may be obtained. At large values of Kg ,
as would be the case for Category 1 gases, Equation 1-35 reduces to:
(RGDPU)
_ -
(RGDpu)H
°alv 1
)A (fP) (*P)
. (1-36)
1.2.4 DEFAULT APPROACH FOR CATEGORY 1 GASES
As mentioned earlier, more elaborate models such as those using a finite difference
solution to the convective-diffusive equation have been developed and applied to specific
gases for evaluation of local absorption rates (McJilton et al., 1972; Miller et al., 1985).
The method in this appendix presents a reasonable alternative based on fewer parameters and
one that is amenable to the types of uptake data routinely generated in some laboratories
(Morris and Smith, 1982; Stott and McKenna, 1984; Morris and Cavanagh, 1986, 1987;
Morris, 1990; Morris et al., 1986, 1991; Dahl et al., 1991b; Morris and Blanchard, 1992;
Bogdanffy et al., 1991; Bogdanffy and Taylor, 1993; Kuykendall et al., 1993). It is hoped
that this approach encourages development of these types of data for the various toxic air
pollutants that the inhalation reference concentration RfC methods are intended to address.
Because uptake data on which to base Kg values are not available for many toxic
chemicals, this section presents default approaches to those presented in the preceding
Section 1.2. The default approaches have been derived based on analyses of the limiting
conditions described in that section. It is assumed that the values for VE and the SA values
for the various respiratory tract regions will be constants within each species.
1-31
-------
p. 65
1.2.4.1 Default Approach for Extrathoracic Region
By definition, Category 1 gases are associated with large K values, which simplifies
the regional gas dose ratio in the extrathoracic region (RGDRET) to
VE
(RGDET)A
- «*
)H
The ratio is based on an averaged dose over the entire ET region because more localized
dosimetry is not yet possible across all species. This default is appropriate when K0
SET
(SAET/VE) is greater than 3 or when
-K E
EET y
(1 -exp ' )A
-1.
SET v
exp VE )H
The objective of the dosimetric adjustment is to address interspecies extrapolation of gas
doses associated with toxic respiratory effects. Because it has been established (Dahl, 1990;
ICRP, 1993) that the types of compounds that are likely to cause respiratory tract toxicity
have high reactivity (either ionic dissociation or metabolism) and solubility (i.e., have
relatively high K _), Equation 1-37 is thus chosen as the default approach for dosimetric
adjustment of gases with ET effects. The regional gas dose ratio (RGDRET) calculated in
Equation 1-37 would be used as the DAFr or the multiplier of the NOAEL*(ADJ) as
described in Chapter 4 (Equation 4-3).
1.2.4.2 Default Approach for Tracheobronchial Region
As discussed above, the basis of the methods for Category 1 gases was the penetration
fraction model to determine the fraction of inhaled dose penetrating the ET region and
1,32
-------
p. 66
thereby available for uptake in the TB region. Thus, the regional gas dose ratio for the
tracheobronchial region (RGDRjg) is calculated as
(RGDTB)A _
(RGDTB)H
VE
SATB
' V
SAra
~Kn-B SATB
A (fpET) A (1 - e VE
)A
(fpET) H -K^ SAj.B
TT \A ~~ &
)H
(1-38)
If the penetration fraction is unknown due to the lack of data on K , it is reasonable to
&XB
assume that K is large, which is consistent with the definition of Category 1 gases, such that
the exponential term of Equation 1-38 reduces to zero. The same result may be achieved by
determining the conditions in which the third ratio of the right hand side of Equation 1-38
reduces to 1. These conditions will be a function of the default values for respiratory tract
surface area and minute volume as well as the absolute value of the overall mass transport
coefficient. Using the definition of fpET results in the following dose ratio
RGDRTB =
(RGDTB)A _
(RGDTB)H
VE
SATB
f VE 1
SATB
SA_
-K CT
tCT
BET \r
A (e E)A
SA^ '
-K CT
SET \r
H <6 ''H
d-39)
which can be rearranged to
RGDRTB =
(RGDTB)A _
(RGDTB)H
VE
SATB
"VE"
SATB
A
H
SACT
e E
^
' _ SA J
e E
(^ET>*
A
(K^H
H
(1-40)
1-33
-------
p. 67
If (K ) A can be assumed to be equal to (K ) H, then Equation 1-40 can be further
simplified to
RGDRTB =
(RGDTB)A _
(RGDTB)H
[vEl
SATB
[VF1
SA^
A
H
' _ SACT
e VE
f SAEr]
F H
A
H
SET
(1-41)
If K is further assumed to be one, Equation 1-41 reduces further such that only minute
SET
volume and surface areas are needed to evaluate the dose ratio, such that:
RGDRTB =
(RGDTB)A _
(RGDTB)H
[VE |
SAjB
[vF 1
SATB
A
H
SAe,
[e V*j
f SAET'
e V*
A
H
(1-42)
If K is available for each species, Equation 1-39 would be the preferred default equation.
1.2.4.3 Default Approach for Pulmonary Region
As discussed in Section 1.2.3, the regional gas dose ratio for the PU region (RGDRPU)
is given by Equation 1-35:
KSA
PU
RGDRpu =
(RGDpu)A KgDTSAPU + Q^
gpu
(RGDpu)H
Qal
KgpuSAPU - Qalv SAPU
which at large Kg values reduces to
:, d-35)
1-34
-------
p. 68
RGDRpu =
(RGDpu)
Qalv
SA
pu
alv
SA
pu
H
If the penetration fractions to each of the preceding regions are unknown due to lack of
data on K and Kg , the approach to deriving a default equation for the PU region is
described below.
Using the definition of fpET and fpTB results in the following gas dose ratio for the PU
region:
RGDRpu =
Qal
(e
SAn
'T".
-K,
)H (e
)H
(1-44)
which can be rearranged to
RGDRpu =
RGDPU)A
(RGDpu)H
Qalv
SApu
Qalv "
SAPu
A
H
-SATB
? \
'SA,;
^VE.
A
H
-SACT
6 ^.
' -SACT
^ gET't
A
H
(1-45)
If (Kg ) A and (K TB) A are assumed to be equal to (K )H and (KgTB)H, respectively,
then Equation 1-45 can be further simplified to
1-35
-------
p. 69
(RGDPU)A
(RGDPU)H
Qalv
SApu
Qalv'
SAPU
A
H
H
H
gET
(1-46)
If it is further assumed that the value of Kg is equal to 1 for each region, the resulting default
equation reduces to an equation requiring only surface area and minute ventilation
parameters. It should be noted that as comparative transport studies become available,
Equation 1-45 would be preferable because it includes the differences in mass transport in
each region for each species.
1.3 Model for Category 2 Gases
The Category 2 or "transitional" gases are those that have physicochemical properties
that are likely to result in the gas significantly accumulating in blood. Accumulation in the
blood will reduce the concentration driving force during inspiration and thereby reduce the
absorption rate or dose upon inhalation. In addition, these gases are distinguished from
Category 1 gases in that there exists the potential for significant desorption during exhalation.
A back pressure (i.e., reversal of the concentration gradient at the air-liquid interface) may
occur during expiration when the exhaled air concentration is less than the concentration of
the surface liquid established during inspiration. Category 2 gases include those which are
moderately water soluble. These gases may also either react rapidly and reversibly with the
surface liquid or they may be moderately to slowly metabolized irreversibly in the respiratory
tract.
A PBPK modeling approach as shown schematically in Figure 1-5 is proposed to
describe the determinants of absorption for this category of gas. A similar model with a
more detailed description of blood flow has been proposed by Overton and Graham (1994).
The PBPK approach is used to evaluate the steady-state blood concentration that is necessary
to calculate both the absorption flux on inhalation and the desorption flux during exhalation.
1-36
-------
p. 70
CX(EXH)
ET
CX(EXH)JB
CX(EXH)pv
Vr
ET
'J CX(INH)
TB
Q
alv ^
/CX(INH) TB
M
ET
\/
M
TB
\/
Blood
Figure 1-5. Schematic of physiologically based pharmacokinetic modeling approach to
estimate respiratory tract dose of gases hi Category 2. The definitions for
the parameter symbols are provided in Table 1-1.
The derivation of the dose to the three respiratory tract regions will be developed in a
similar fashion as that for Category 1 gases (Section 1.2). Each region will be considered
individually. Following the general description of the modeling approach for each region, a
mass balance approach using a PBPK analysis will be developed to determine the blood
concentration. A summary of the results and equations will be provided at the end of this
section along with the default formulation.
1.3.1 Model for Category 2 Gases: Extrathoracic Region
As with the Category 1 gases, the change in concentration in the ET region
(Section 1.2.1) can be described by Equation 1-9. If it is assumed that sufficient time has
passed to allow a steady-state blood concentration to be developed, Equation 1-9 can be
integrated, resulting in Equation 1-10. In the case of Category 1, the blood concentration was
1-37
-------
p. 71
assumed to be much less than the airstream or interfacial concentrations. For Category 2,
however, the blood concentration must be retained. Thus, the fraction of gas that penetrates
to the TB region is given by rearranging Equation 1-10 such that:
-K
SACT
fpET = e
*ET
C:
-K
SAC
1 -e
SET
(1-47)
As defined in Equation 1-16, the dose on inhalation to the ET region, RGD(INH)ET, may be
obtained by substituting Equation 1-47 into Equation 1-16 and rearranging to obtain
RGD(INH)ET =
_ b/g
c.
-K
C-V
(1-48)
SA
ET
The form of the overall mass transport coefficient in Equation 1-48 differs from that to
describe Category 1 gases because a term to describe the disposition of the gas in blood is
required. Approaches to include this term are reviewed by Ultman (1988). In the case
where there is either no reaction or the reversible nature of the reaction can be handled by
adjusting Ht/ to be in equilibrium with the dissociated form of the gas, the mass transport
coefficient for Category 2 gases may be determined from:
J_
%T
Ht/gkl
Hb/gQb
(1-49)
where S is the blood perfusion surface area, Hb is the blood:air partition coefficient, and Qb
is the local blood flow rate. The mass transport coefficient for gases, which are moderately
to slowly metabolized in the tissue phase is given by
1
1
K
SET
d-50)
1-38
-------
p. 72
where F is the flux fraction reaching the blood, and VLQ is the volume of lung tissue. The
flux fraction, F, is less than one if the absorbing gas reacts with constituents in the surface
liquid and tissue phases.
Equation 1-48 addresses the dose upon inhalation only. To evaluate the total dose,
including events occurring during exhalation, the potential for desorption and the desorption
flux must be evaluated. Desorption will reduce the total dose over a respiratory cycle; the
dose associated with an observed effect is therefore less than that if only the dose on
inhalation was considered. In the following section, the desorption term is developed by first
considering the tissue depth in which desorption may influence tissue concentration and, by
analogy, tissue dose.
1.3.1.1 Theoretical Considerations for Modeling Desorption
Empirical data has indicated that desorption can be important to estimating the
respiratory tract dose (Gerde and Dahl, 1991; Dahl et al., 1991b). Unless the tissue
concentration is greater than the exhalation airstream concentration, there will be no
desorption during exhalation and, in fact, there is actually the potential for additional
absorption. To evaluate the potential desorption, it is assumed that the blood concentration
attains a relatively constant concentration independent of the respiratory flow cycle. Because
it is assumed that the potential desorption will not impact the blood concentration, desorption
will only impact the concentration profile in the tissue. The tissue concentration profile
during exhalation is a function of the duration of exhalation. If desorption occurs, the
surface-liquid/tissue concentration will decrease during exhalation, as will the concentration
gradient between the air (gas phase) and blood. An example of the change in the tissue
concentration profile that may occur during desorption is shown in Figure 1-6.
In Figure 1-6, the change in tissue concentration is shown to penetrate the entire
surface-liquid/tissue phase as a result of the change in airstream concentration between
inhalation and exhalation, C(INH) and C(EXH), respectively. The extent to which the tissue
concentration profile changes is as yet unknown and must be evaluated to formulate the
desorption term. To estimate the depth in the tissue that may be influenced by the change in
flow direction and hence airstream concentration, an analytic solution is employed to evaluate
the time course of the concentration profile in the tissue when the surface is exposed
1-39
-------
p. 73
Airway Lumen
Gas Phase Surface-Liquid / Tissue Phase
C(INH)
t<0
C(EXH)
t>0
Blood
Figure 1-6. Schematic of surface-liquid/tissue phase concentration during exhalation.
to a step change in concentration associated with the flow reversal. To avoid assumptions
about the tissue thickness, it is assumed that the tissue is infinitely thick.
In Figure 1-7, the initial conditions prior to imposing the step change is shown in which
the tissue concentration is C0 throughout. At time zero (t=0), the step change in the
airstream concentration is imposed and the change in tissue concentration with distance and
time is illustrated. Given the conditions described above and further assuming no reaction in
the surface liquid-tissue layer, the solution is given in the form
(Cs -
(Cs - C0)
= etf
(1-51)
where Cs is the imposed concentration; Cz is the concentration in the surface liquid/tissue,
which is a function of time (t) and distance into the layer (z); C0 is the initial concentration;
1-40
-------
p. 74
Airway Lumen
Gas Phase
Surface-Liquid / Tissue Phase
Figure 1-7. Schematic of change in surface-liquid/tissue phase concentration with
distance (z) and time.
and erf is the error function. The term on the left hand side of the equation is the
nondimensional concentration such that, when C2 is in equilibrium with the gas phase
concentration Cs, the nondimensional concentration is zero whereas when C2 is equal to C0,
the nondimensional concentration is one.
To determine the depth to which the change in the surface concentration impacts the
tissue concentration, the time for exhalation in humans is estimated to be approximately
3 s during rest. The time would decrease at increased ventilation rates. Using the above
equation, the distance in which the nondimensional concentration attains 0.5 (i.e., the
distance in which the local concentration is one-half the concentration difference) is
determined to be approximately 70 /xm. This distance represents significant penetration into
the surface-liquid/tissue phase.
1-41
-------
p. 75
1.3.1.2 Formulation of the Desorption Term
The above estimate indicates that an imposed step change in air (gas phase)
concentration as occurs during exhalation results in the tissue concentration at 70 /*m attaining
50% of the equilibrium value within 3 s. Because this distance is of the order of the distance
between the air-surface liquid/tissue interface and blood (Miller et al., 1985), it will be
assumed in the derivation of the desorption term that the entire depth of the surface liquid-
tissue phase between its interface with the gas phase and the blood may be impacted by
desorption. Thus, a conservative assumption would be to assume that the gradient of an
absorbing, nonreactive gas achieves a linear profile quickly during both inhalation and
exhalation and that the gas phase transport resistance does not affect desorption. Therefore,
the desorbed mass due to the flow reversal may be obtained by evaluating the change in mass
necessary to effectively reduce the tissue gradient from the inhalation gradient to the
exhalation gradient as indicated in Figure 1-8.
Airway Lumen
Gas Phase Surface-Liquid / Tissue Phase Blood
C(INH)
C(EXH)
C(EXH)
Continued
Absorption
'"f"
Desorption
Figure 1-8. Schematic of change hi mass during breathing cycle.
1-42
-------
p. 76
In Figure 1-8, two cases are noted with respect to the exhalation tissue concentration
profile. In the first case, the exhalation airstream concentration, C(EXH), is greater than the
concentration in equilibrium with the blood concentration. Consequently, the gradient is still
directed inward such that absorption would continue during exhalation. However, there
would be an initial loss of mass associated with the change in the concentration profile as
shown in Figure 1-6. The second case shown in Figure 1-8 is that in which the C(EXH) is
less than Cb, such that the gradient is directed outward. In this case, desorption occurs due
to the step change in concentration associated with the flow reversal as well as a reversal in
the concentration gradient between the air and blood. It is assumed, however, that the mass
transferred during exhalation is associated primarily with achieving the exhalation tissue
profile.
The desorbed mass (Md) due to the step change in airstream concentration is therefore
assumed to be determined by subtracting the average concentration represented by the tissue
gradients between inspiration and expiration, such that
M,
CX(INH)ET - CX(EXH)TB
AZETSAET,
d-52)
where AZET is the surface-liquid/tissue phase thickness of the ET region. Because the blood
concentration is assumed to be the same during inhalation and exhalation, it does not appear
in the above equation.
In the case of a reactive gas, Equation 1-52 will overestimate the desorbed mass because
the concentration gradient is likely to be curvilinear and a decrease in the concentration
profile would also be achieved by reaction and not through desorption. It is also possible that
the reaction could be sufficient to effectively result in further absorption during exhalation
due to the reduced tissue concentrations (similar to Case 2 described for Figure 1-8). As an
estimate of the desorbed mass of a reactive gas, an exponential decay term is added to
Equation 1-52 to account for the tissue reactivity, such that
1-43
-------
p. 77
M,
CX(INH)ET - CX(EXH)TB
A ^
a-53)
where tEXH is the time of exhalation.
The regional desorbed dose (mass/cm2 -time) during exhalation is therefore Md divided
by the product of the surface area and the exhalation time, such that
RGD(EXH)ET =
M,
SAET tEXH
(1-54)
The total dose in the ET region, accounting for both absorption during inhalation and
desorption during exhalation, is therefore:
RGD(TOTAL)ET = -1
SA
ET
-K
(1 -
) -
(1-55)
SAET tEXH
1.3.2 Model for Category 2 Gases: Tracheobronchial Region
The model developed for the analysis of total dose to the ET region for gases in
Category 2 is directly applicable to the determination of the total dose to the TB region, such
that
RGD(TOTAL)TB =
CX(INH)ETVE
SATB
1 _ Cb/8
CX(INH)ET
>
(1 -e
B y IVl J
SATB tEXH
d-56)
where the desorbed mass is similarly defined as above. Thus, for gases that do not react
irreversibly, Md is given by:
M,
CX(INH)TB - CX(EXH)pu
AZTBSATB,
(1-57)
1-44
-------
and for those gases that do react irreversibly Md is is given by
p. 78
M,
CX(INH)TB - CX(EXH)PU
,-MO
d-58)
where AZ-j-g is surface-liquid/tissue phase thickness of the TB region.
Substituting for CX(INH)ET using Equation 1-47 and the definition of fp ,
C/ JL
Equation 1-56 becomes
RGD(TOTAL)TB = -LJ e
SA
TB
1-
b/g
KgTBSATB
(1 -
a-59)
1.3.3 Model for Category 2 Gases: Pulmonary Region
The dose to the PU region for the Category 2 gases may be derived on the basis of
equations provided previously (Section 1.2.3). From Equation 1-30 the ratio of the expired
concentration to the inspired concentration of this region (i.e., the penetration fraction of the
PU region) is defined as
CX(EXH)pu
Qalv
CX(INH)TB
KgPUSAPU
1
Cb/g
CX(EXH)pu
+ Qalv
(1-60)
where CX(EXH)PU is the concentration exiting the PU region and therefore includes the
desorption term.
In the previous section describing PU dose (Section 1.2.3), KgPU is defined only for a
reactive gas (Equation 1-28). However, for gases that are nonreactive or reversibly reactive,
a more appropriate form would be (Ultman, 1988)
1-45
-------
p. 79
KgPU Ht/gklSAPU Hb/gQb
The PU dose as defined for Category 1 gases was based on the assumption that Cb/
was less than Caiv. This assumption is not applicable to the transitional gases of Category 2
because of the potential for elevated blood concentrations. Consequently, the total dose to
the PU region for these gases is defined as
RGD(TOTAL)pu =
Qal
[K SAPU
S ru
CX(EXH)pu
Qalv
SA
PU
CX(INH)TB
(1-62)
Equation 1-62, although the most general form of the PU dose, can also be formulated
more simply by assuming
calv ~ CX(EXH)pu - Cb/g
(1-63)
because Category 2 gases are moderately water soluble and likely to reach equilibrium
between alveolar air concentration and the blood. Under these conditions, the PU dose is
simply the difference between the inhaled concentration, CX(INH)TB, and the exhaled
concentration, CX(EXH)pu, such that
prnnYYTAT ^ CX(INH)TB - CX(EXH)pu
RGD(TOTAL)pu = _ Qalv,
SA
PU
(1-64)
which by substitution for CX(INH)TB and CX(EXH)PU rearranges to
1-46
-------
p. 80
RGD(TOTAL)pu = q^*L
5Apu
(1-65)
Equation 1-65 represents the most generalized equation resulting from the simplifying
assumption that the PU dose is proportional to the difference between the inhaled and exhaled
concentrations.
1.3.4 Modeling the Blood Compartment for Category 2 Gases
As defined, Category 2 gases will accumulate in the blood. Thus, an explicit derivation
to determine concentration of the gas in the blood is required to solve for the dose into each
region. In particular, the term that must be evaluated is (1- q,/ /q), which appears in each
of the equations necessary to solve the regional dose. This term includes the blood
concentration because Cb/ is the concentration in the gas phase which would be in
equilibrium with the blood (i.e., Cb/g = q/Hb/g).
The blood concentration is derived by a mass balance approach. It is assumed that the
systemic blood compartment is well mixed so that the change in concentration is due to the
input mass delivered through the respiratory tract, loss due to metabolism in the lung tissue,
redistribution of the gas in the systemic compartments (including the fat compartment) during
intermittent exposures, and loss due to systemic metabolism (modeled in the liver
compartment), such that
Vb = £(RGD(TOT)RTSA(TOT)RT) - C^CL + CLfat) - VLGELG
where Vb and VLG are the volumes of the blood and lung compartments, respectively, Cb is
the average blood concentration; 2(RGD(TOT)RT SA(TOT)RT) is the summed product of the
dose and surface area of each region in the respiratory tract; C^ is the arterial blood
concentration; CLsys and CLfat are the clearance from the systemic (i.e., assumed to be
dominated by the liver compartment) and the fat compartments, respectively; and ELG is the
elimination rate in the lung compartment due to metabolism. The total mass transport rate to
1-47
-------
p. 81
the respiratory tract (mass/time) is given in the above equation as the summed product of the
dose and surface area of each region in the respiratory tract. However, the total dose to the
respiratory tract may also be obtained by the difference in inhalation and exhalation
concentrations, such that
£ (RGD(TOT)RTSA(TOT)RT) = VE (q - CX(EXH)ET). (1-67)
The term that implicitly includes the blood concentration and is necessary to solve
regional dose is obtained from Equation 1-67. Ignoring further absorption or desorption that
may occur during expiration, CX(EXH)ET may be approximated by CX(EXH)pu, which is
equivalent to C^v, the alveolar concentration, which is in equilibrium with the blood
(Equation 1-63). Thus
C
VE (q - CX(EXH)ET) = VEq (1 - _±§). (1-68)
To determine the respiratory tract dose during the exposure, it will be assumed that the
system is in quasi-steady state such that the change in the average blood concentration
(dCb/dt) is zero (Equation 1-66). Under these conditions, the mass delivery rate to the
respiratory tract surface (defined in Equation 1-67) is equal to the loss due to clearance from
the liver and fat as well as metabolism in the respiratory tract tissue. Combining
Equations 1-66 through 1-68 under steady state conditions results in
VE
-------
p. 82
represents an additional loss from the arterial blood concentration. At the end of the
exposure, leaching from the fat compartment may be an additional input to the blood.
Because the initial dose of gases with respiratory toxicity accounts for the dose that may be
leached subsequently from the fat, no additional dose following the end of exposure needs to
be accounted for. Furthermore, the contribution of the fat compartment is reduced for
Category 2 gases because the gas will not partition significantly to the fat because of its lower
fat to blood partition coefficient. In addition, the concentrations in the systemic
compartments are in equilibrium with the blood during steady state. Thus, the uptake by the
fat will be assumed zero, consistent with the definition Category 2 gases because of their
partition coefficient. The assumption of a steady state is conservative because it will
underestimate the dose to the respiratory tract compartments that are the objective of the
derivation in this appendix. The assumptions of steady state and of equilibrium between
tissue and blood compartments results in the elimination of CLfat from Equation 1-69.
Rearranging Equation 1-69 to solve for systemic elimination results in
- V (Ci " Calv) - VLGfeLG (1-70)
VE — ' v '
where CLfat is zero as described above. However, the ratio of the exhaled concentration,
Calv , to Cart is approximated by Hb/ . Equation 1-70 may therefore be rewritten as
SVS E__ _
sys b VTT tr
WEFF nb/g
where HEFF is the steady state blood to inhaled gas concentration ratio observed in an
experimental situation (Andersen, 1981), which is referred to here as an effective partition
coefficient.
It is now necessary to more specifically incorporate the loss terms for systemic
clearance and respiratory tract metabolism. It will be assumed that systemic clearance is
1-49
-------
p. 83
predominately due to metabolism in the liver and is given by (Andersen, 1981; Pang and
Rowland, 1977)
(1-72)
where CLLiy is me clearance from the liver compartment; Qp is the cardiac output; and
Ej is the liver extraction efficiency. The elimination rate from the lung compartment, ELG,
is defined according to Michaelis-Menton kinetics:
CLGVMAX t ^
where VMAX is the maximum velocity of saturable (Michaelis-Menton) metabolism path;
where CLG is the lung tissue concentration; KM is the Michaelis constant; and kLG is the
elimination rate from the lung compartment.
Combining Equations 1-71 to 1-73 provides the loss terms in relation to HEFF:
= VB(—— - __L) - Vr-kir-^ (1-74)
.bvTT IT LO IAJ /-i
MEFF Mb/g uart
Assuming the respiratory tract tissue concentration, CLG, is in equilibrium with the blood, the
ratio CLG/Cart is equivalent to the tissuerblood partition coefficient, Ht/b. Solving for HEFF
yields
VLGkLGHt/b
Hb/g
Combining Equation 1-68 with the definition of Hb/ and HEFF results in
1-50
-------
p. 84
v r n ^ -vrr-r ^ - v
VECi (L~-F~) ~ VE(-Ci Calv>> ~ V
Upon substitution of 1-75 into 1-76, the term necessary to solve the dose ratio in
Equations 1-55, 1-59, and 1-65 is obtained:
VLGkLGHt/bHb/g
Ci QTErHb/g + VLGkLGHt/bHb/g + VE
where Ht/ is equal to the product of Ht/b and Hb/g. Equation 1-77 may be simplified by
considering the range of partition coefficients, extraction efficiency and the respiratory tract
tissue concentration.
At large values of Ht/ (and consequently Hb/ since Hb/t x Ht/ = Hb/ ), the term on
the right hand side of Equation 1-77 approximates one. Therefore, Cb/ « Cj which is the
definition of Category 1 gases (i.e., those gases that are highly soluble and/or rapidly
irreversibly reactive) for which the approach presented in Section 1.2.4 applies. This case is
consistent with a greater extraction efficiency of the respiratory tract relative to the systemic
clearance as well as absorption proximal to the PU region. Conversely, at low values of
Ht/0, absorption proximal to the PU region is negligible and the relative efficiency of
systemic clearance is greater than that of the respiratory tract extraction (as well as uptake).
The approach for category 3 gases presented in Appendix J applies in this case.
The remaining gases are those which are moderately water soluble (intermediate value
of Ht/g) and are therefore the Category 2 gases. For Category 2 gases, Equation 1-77 reduces
to
d -
since VE/QrH^g » Ej- + (VLGkLG/Qr) Ht/b.
1-51
-------
p. 85
Equation 1-78 can be further reduced since Qp approximates VE. The magnitude of the
blood concentration is determined by the relative significance of the metabolism which occurs
in the respiratory tract versus systemic elimination (as shown in the ratio
. If systemic elimination is much larger, Equation 1-78 reduces to
d-79)
At the maximum, it will be assumed that the respiratory tract elimination would be equal to
that of the systemic elimination under which circumstances
(1 - f™*) = 2ETHb/g . d-80)
M
It will be further assumed that the systemic elimination term is defined for maximum
elimination, i.e. assuming liver saturation kinetics, such that Ej- is defined by EMAX, the
maximum extraction efficiency. The maximum extraction efficiency is approximately
0.25 Op due to the flow limitation to the liver (Andersen, 1981). Thus, Category 2 gases can
be defined based on systemic elimination and the relative significance of respiratory tract
metabolism to systemic elimination.
1.3.5 Default Approach for Category 2 Gases
The default approach is developed by ignoring the desorption associated with exhalation.
This assumption may be valid in as much as the mass in the tissue that is desorbed during
exhalation is replaced on inhalation. Whether, in general, this assumption results in an
overestimate or underestimate of the dose is not clear because ignoring the desorbed mass
may not significantly impact the concentration driving force (i.e., the concentration of the gas
at the surface-liquid/tissue interface and the concentration in the blood may be proportionately
affected).
In comparing cyclic absorption-desorption and unidirectional absorption, the
concentration at the interface between the air and surface liquid is likely to be lower in the
1-52
-------
p. 86
case of desorption and the driving force would therefore be lower than in the case of
unidirectional absorption if the blood concentration were equal in both cases. The net effect
would suggest that ignoring desorption would overestimate the absorbed mass or dose.
However, by overestimating the absorbed mass in the case of unidirectional absorption, the
blood concentration will be elevated over the absorption-desorption case. The elevated blood
concentration will also reduce the concentration driving force. Therefore, although ignoring
desorption will increase the surface liquid concentrations, the blood concentration will
similarly be overestimated, so that the concentration driving force may not be dissimilar than
with desorption described.
The dose to each region, ignoring desorption, is therefore obtained by combining
Equation 1-79 or 1-80 (depending on the significance of respiratory tract metabolism) with
each of the individual dosimetry calculations in Equations 1-55, 1-59, and 1-65 for the ET,
TB, and PU regions, respectively.
1.3.5.1 Default Approach for Extrathoracic Region
From Equation 1-54, the regional gas dose ratio (ignoring desorption) for the ET region
(RGDRET) is given by
VE
r)
RGDRFT = —^ = 11^ 1^ r_: ^ . (1-81)
tii n?
-------
p. 87
RGDRET . . _ . (,82)
-------
p. 88
where Ej^x is equal to 0.25 Qp Because the constants are equal in the numerator and
denominator, Equations 1-84 and 1-85 reduce to the same equation:
) K (QTHb/g)
RGDR = - - = - - - _ , (1-86)
which can be further reduced if the overall mass transport coefficients (K ) are assumed to
be equal.
1.3.5.2 Default Approach for Tracheobronchial Region
From Equation 1-58, the regional gas dose ratio (ignoring desorption) for the
tracheobronchial region (RGDRTB) is given by
RGDRTB =
-K _ ,i _ ^D/ES -K
(& ^ "\ ^'i (\ —- & E
ve /A 1 A v1
«TB ^
( Jx.Cj-D'T'T} ) VT- TT- ET ^H /a
M //~i & \ ""•^•» — : (\ — ^
^ i"cA — ^ v ^ r ;
SATB H (e E )H in
-K SApB
*1B y
0 - e E )H
(1-87)
As in the ET region, K for Category 2 gases is by definition less than 1 and a power
&TB
series expansion of the exponential term for the TB region similarly reduces the last term to
the ratio of the K . The exponential term for the ET term in Equation 1-86 is reduced by
assuming K is the same for each species as was assumed for Category 1 gases. At values
of K less than or equal 0.5, the ET exponential term approaches one. Thus, assuming the
°ET
same inspired concentrations, Equation 1-86 becomes
(RGDTR)
RGDRTB = _ _ ^ = - _ . (1-88)
™ (RGDTB)H ^
H
1-55
-------
p. 89
As above, Equation 1-88 is further reduced by substituting Equation 1-79 for the case in
which systemic elimination predominates:
(RGDTR) K (0.25 QTHh/a)
RGDRTB = ^ = -^ - T b/a A . d-89)
(RGDTB)H
By substituting Equation 1-80 for the case in which respiratory tract metabolism and systemic
elimination are of equal significance, Equation 1-88 becomes:
(RGDTB) K g (0.5QTHb/a)
RGDRTR = = , (1-90)
TB
-------
p. 90
The default ratio is obtained by assuming the mass transport coefficients for the ET and the
TB region are the same in each species. The exponential term for both the ET and TB term
in Equation 1-90 thereby reduces to one. Thus, assuming the same inspired concentrations,
Equation 1-90 becomes
A
RGDR = _ _ = - * - * . (1-93)
pu H
The RGDRpu must be evaluated for each case described in section 1.3.4. In the case where
systemic elimination determines the blood term, the PU regional gas dose ratio is given by
(RGDPU)A ^SApu'A(0.25 QTHb/g)A
"(RGDpu)H~ Qd (0.25 QTHb/g)H '
\Vi i'
SAPUH
where EMAX is equal to 0.25 Qp.
In the case where respiratory tract metabolism and systemic elimination are equally
important, the PU regional gas dose ratio is given by
(RGDPU) PU A (°-5 0THb,g)
RGDRp,, » * = * — , 0-95)
(RODPU)H ^ (0.5QTHWg)H
SAPU H
where EMAX is equal to 0.25 Qp. Because the constants are equal in the numerator and
denominator, Equations 1-94 and 1-95 reduce to the same equation:
1-57
-------
p. 91
(RGDPU) u A (QTHb/g)
RGDRPIJ = 2 = A 1^ . (1-96)
(RGDPU)H ^ Qaly) (QTHb/g)H
SAPU H
1.3.6 Model for Category 2 Gases: Total Respiratory Tract
In the event that remote (extrarespiratory) toxicity is associated with a gas in
Category 2, the dose to the respiratory tract, and therefore to the blood, is necessary to
establish the dose ratio. However, in this case, the surface area of the respiratory tract is
irrelevant, only the overall absorption rate in mass/time (RGDRT) is important which is given
by
RGDRT = VE(q - CX(EXH)ET) = VEq(l - —) , (1-97)
such that the dose ratio (assuming the same inspiratory concentration) is
(1 - _^1?)
(RGDRT) (VE) q '
A = A A (1-98)
(RGDRT)H (VE)H _ C^ '
1 H
to be evaluated for each of the cases described in Section 1.3.4. In the case where systemic
elimination determines the blood term, the regional gas dose ratio for remote
(extrarespiratory) effects of Category 2 gases is given by
where EMAX is equal to 0.25 Qp-
In the case where respiratory tract metabolism and systemic elimination are equally
important, the regional gas dose ratio for remote (extrarespiratory) effects of Category 2
gases is given by
1-58
-------
p. 92
(RGDRT) (VE) (0.25 QTHb/g)
RGDRER = ~ = — , (1-99)
(RGDRT)H (VE)H (0.25 QTHb/g)H
(RGDRT) (VE) (0.5QTHb/)
RGDRER = ^ = _^ gA , (MOO)
(RGDRT) (v^ (0.5QTHb/g)
where EMAX is equal to 0.25 Qp. Because the constants are equal in the numerator and
denominator, Equations 1-99 and I- 100 reduce to the same equation:
(RGDRT) (VE) (QTHb/g)
_ - = - - - _ . (1-101)
(RGDRT)H (VE)H (QTHb/g)H
1-59
-------
p. 93
APPENDIX J. DERIVATION OF AN APPROACH TO
DETERMINE HUMAN EQUIVALENT
CONCENTRATIONS FOR EXTRARESPIRATORY
EFFECTS OF CATEGORY 3 GAS EXPOSURES
BASED ON A PHYSIOLOGICALLY
BASED PHARMACOKINETIC MODEL
USING SELECTED PARAMETER VALUES
This appendix describes in detail the derivation of the procedure used in Chapter 4 to
estimate no-observed-adverse-effect level human equivalent concentrations (NOAEL^HEqS)
for extrarespiratory effects of gases (or vapors) in Category 3. The derivation is
mathematical in nature in that the equations of state that describe the disposition of inhaled
compounds in a generalized physiologically based pharmacokinetic (PBPK) model are
manipulated so as to obtain a conservative estimate (with respect to the model assumptions) of
NOAEL[HEC]S as a function of the average animal exposure concentrations (NOAEL^jj).
A PBPK model is used because of the success of this type of model. For example, PBPK
models that describe the body as five compartments (gas exchange and the fat, poorly
perfused, richly perfused, and liver/metabolizing tissue groups) have been applied
successfully to estimating the internal concentrations of chemicals (e.g., styrene, methanol,
and ethylene dichloride) for the purpose of risk assessment. Although PBPK modeling is the
choice procedure in risk assessment for dose extrapolation, this approach is not possible
without the values of physiological and biochemical parameters used in the modeling process,
nor without a thorough understanding of the agent's mechanism of action. These data
generally are not available for most compounds.
The proposed method is based on a PBPK model in which all of any number of
compartments are in parallel and in which for any compartment there can be any number of
paths of removal by linear and saturable processes. Selected relevant parameter values are
replaced by qualitative assumptions about species similarity and the response of internal
concentrations to exposure scenarios. In order to obtain a NOAEL[HECj, the assumption is
made that the effective dose for dose-response purposes is the arterial blood concentration of
J-l
-------
p. 94
the gas or its concentration multiplied by time (C x T). (These assumptions are specified in
detail in the METHODS section.) This latter assumption is consistent with our current
understanding of systemic toxicity for a majority of chemicals, because the toxicity of most
environmental chemicals is more directly related to the concentration of the parent compound
at the target site over a period of time than to the exposure concentration over an equivalent
time period.
In addition to deriving conservative NOAELrHECj estimates based on arterial blood
concentrations, the method also predicts that the average blood concentration of an inhaled
compound in any human tissue compartment does not exceed the average blood concentration
in the corresponding animal compartment.
J.I METHODS
J.I.I Assumption Imposed by the Inhalation Reference
Concentration Methodology
Assumption I. Noncancer toxic effects observed in chronic animal bioassays are the
basis for the determination of NOAELs and the operational derivation of inhalation reference
concentrations (RfCs) for human exposures, as described in Chapter 4. The animal exposure
scenario is experiment-dependent and usually intermittent (e.g., 6 h/day, 5 days/week for
many weeks) and is assumed periodic. Human exposure concentration is continuous and
constant for 70 years. The "lifetime" chronic animal exposure scenario is equivalent to the
human chronic exposure scenario for the purpose of extrapolating the NOAEL.
J.1.2 Additional Assumptions for the Proposed Method
Assumption II. All the concentrations of the inhaled gas within the animal's body are
periodic with respect to time (i.e., periodic steady state—the concentration versus time profile
is the same for every week). Figure 4-9 illustrates the time course to achieve periodicity for
a chemical with blood:air and fat:blood partition coefficients of 1,000 and 100, respectively.
Periodicity is achieved for this chemical after approximately 5 weeks. As discussed in
Chapter 4 (Section 4.3.6.2), it is practical to require that these experimental periodic
conditions should be met during "most" of the experiment duration in order for this model
J-2
-------
p. 95
application to result in an accurate estimate for use in the dose-response analysis. For
example, if the condition is met for nine-tenths of the time (e.g., periodic during the last
90 weeks of a 100-week experiment), then estimates of average concentrations will be in
error by less than 10%. Thus, the requirement for application of this model is that
periodicity is achieved for 90% of the exposure period. If this is likely not to have occurred,
additional uncertainty in the extrapolation is imparted and should be addressed by an
uncertainty factor (Section 4.3.6.2). During most of the time humans are exposed, given
Assumption I of continuous exposure, their internal concentrations are constant and in
dynamic equilibrium with their exposure concentration.
Assumption III. A PBPK model describes the uptake and disposition of inhaled
compounds in animals and humans. The model is diagramed in Figure J-l, and the equations
of state are given by Equations J-l through J-6. Table J-l defines the variables and constants
in the equations.
dMp/dt = Qdv X (CE - Cp + QT X (CV - CA) - rp(CA) (M)
dM/dt = Qj x (CA - Cj) - r^Cj); j = 1,2,3,.. .n (J-2)
r(CA) - VKF
x CA + £ [VMAX^ X CA/(KMpi + CA)]
-------
p. 96
cv
CE
Qalv
QT
»
cP
I • CA
i
rp
c,
V
°P ,
Qalv
Or
CA
•
\
i
Respiratory Tract
^Compartment
With Metabolism
»
»
»
°1
1
Cn
1
4!
«
«
'On
\
1
•
1
•
,
I Any Number of
1 Metabolizing and
\ NonmetaboMzing
I Compartments
Figure J-l. Schematic of the physiologically based pharmacokinetic model assumed to
describe the uptake and distribution of inhaled compounds.
respect to each other; (2) in each extrarespiratory (systemic) compartment, the blood and
tissue concentrations are in equilibrium with respect to each other; (3) the metabolism and
other loss mechanisms are taken into account in the tissue of the respiratory tract
compartment and in the extrarespiratory (systemic) compartments; and (4) both first-order
and saturable loss rates are represented and are defined in terms of blood concentrations
regardless of whether or not they occur in tissue or blood.
Equations J-l, J-2, J-4, and J-5 are the dynamical equations of state or mass-balance
equations for the model. Equations J-3a and 3b define the possible loss rates in each
compartment in terms of linear rates (e.g., VKF:i x C) and rates of the Michaelis-Menton
type (e.g., VMAX j X CA/[KM • + CA]). In each compartment, the model allows for
J-4
-------
p. 97
TABLE J-l. DEFINITION OF SYMBOLS
General
V
n
M
X
b/g
H
P
L
h
Subscripts
i
P
j
A
H
HEC
Compartment volume
The number of extrarespiratory compartments
Mass of inhaled compound in gas-exchange compartment
Mass in compartment other than gas exchange
Multiplication symbol
Overbar indicates average
Blood to air partition coefficient
Period of periodic exposure concentration
Liters
Hours
i-th path of loss of primary compound
Gas-exchange compartment
j-th extrarespiratory compartment
Animal
Human
Human equivalent concentration
Flow Rates (L/h)
Alveolar ventilation
Cardiac output
Extrarespiratory (systemic) compartment perfusion rate
Concentrations (mg/L)
C
CE
CP
CA
CV
Biochemical
r
VMAX
KM
KF
VKF
In venous blood within and leaving extrarespiratory (systemic) compartment
Exposure
In air of pulmonary region
In arterial (unoxygenated) blood
In venous (oxygenated) blood entering gas-exchange region
Removal rate due to metabolism, reactions, excretion, etc. (mg/h); when
denoted as r(c) this indicates dependence on given concentration
Maximum velocity of saturable (Michaelis-Menton) metabolism path (mg/h)
Michaelis constant (mg/L)
First-order rate constant (h"1)
Equals to V x KF (L/h)
J-5
-------
p. 98
more than one path of elimination or metabolism or for no losses (i.e., set both of a
compartment's kinetic parameters, VKF and VMAX, to zero). Equation J-6 gives the
assumed relationship between the arterial blood concentration and the concentration in the air
of the pulmonary region.
According to Assumption I, the exposure concentration is periodic for laboratory
animals and constant for humans; in both cases, concentration of exposure (CE) can be
written as
CE = f(t) x HE, (J~7)
where:
CE = the average exposure concentration, and
f = a periodic function of time (t) such that
f(t + P) = f(t) (J-8a)
t+P
J f(t) x dt = 1; (J-8b>
and P is the period of the periodic exposure concentration.
Assumption IV. Because the lexicologically effective dose to a given target tissue
depends on the animal species and chemical compound, its specification is typically not
available so that definition of a surrogate dose must be somewhat arbitrary. However, the
toxic effects of some compounds are expected to be directly related to the inhaled parent
compound in the blood. Furthermore, the use of the average blood concentration is an
internal dose "closer" to the target than a dose based on exposure concentration. Basing the
effective dose extrapolation on another surrogate (e.g., metabolite) would require knowledge
of the mechanisms of action and additional information about human and animal physiological
parameters. Thus, for animal to human exposure extrapolation, the human equivalent
J-6
-------
p. 99
exposure concentration (CE[HECp is defined in terms of the average arterial blood
concentration of the inhaled parent compound by requiring that the human equilibrium
concentration of arterial blood be less than or equal to the time-averaged arterial blood
concentration of the animal; that is, CAH < €A~A. Note that the time average concentrations
are the area under the curve over a period divided by the length (time) of a period (e.g.,
average concentration over 1 week). The equality condition defines the upper limit on an
acceptable human arterial blood concentration; thus, for mathematical simplicity this
assumption is formulated as:
CAH = CA"A. (J-9)
Because of this requirement, CAH is a function of CEA, because CA~A depends on CEA.
Assumption V. Similarity of species is assumed in that KM and the ratios Q/ Qjv,
VKF/Qajy, and VMAX/Q^ are defined as species independent for each removal process (see
Table J-l for definitions). The invariance of the first ratio is based on the assumption that
the percent of blood flow to any compartment is independent of species and that cardiac
output (Qj. = sum of all Qj) scales, with respect to body weight, in the same way as the
ventilation rate (Qaiy); (i.e., the ratio of Qp to Q^ is species-independent). The metabolic
constants VMAX and VKF are assumed to scale in the same way as Q^. Justification for
this assumption about rates is based on the observation that for many species, rates scale in
the same way with respect to body weight (e.g., in proportion to basal metabolism, body
surface area, or body weight to some power) (Dedrick, 1973; Weifl, 1977; Dedrick and
Bischoff, 1980; Boxenbaum, 1982; Rowland, 1985; Travis and White, 1988; Travis et al.,
1990; Federal Register, 1992b). The invariance of the ratios VKF/Q^ and VMAX/<^lv
follows.
Most of the above assumptions are well supported by data on comparative anatomy and
physiology, as detailed in the cited and other allometry references (Federal Register, 1992b).
Collectively, they embody the concept of a basically similar mammalian physiological and
anatomical plan that varies primarily in scale from one species to another. The most
J-7
-------
p. 100
problematic issue is the scaling of rates of individual metabolic transformation reactions as
BW3/4. Not only are there few data on such scaling, but some individual metabolic enzyme
systems have been shown to vary across species (Federal Register, 1992b). However, several
points should be made. First, there are data that support the proposition of BW3/4 in specific
cases (Federal Register, 1992b). For example, these same scaling assumptions have been
used in successful PBPK modeling across species (Ramsey and Andersen, 1984; Andersen
et al., 1987a; Ward et al., 1988; Allen and Fisher, 1993; Fisher and Allen, 1993). Second,
overall metabolic rate (oxygen consumption, resting metabolic rate) clearly scales as BW3/4.
Indeed, this is the issue around which physiological allometry was developed. Scaling an
individual metabolic step in this way corresponds to keeping it in proportion to general
metabolism, which seems the best default (Federal Register, 1992b). Third, daily intake of
natural toxins (the usual targets of toxicant-metabolizing enzymes) depend on intake of air,
water, and food which all scale as BW3/4. That is, scaling detoxification processes in
proportion to their anticipated load also predicts BW3/4. Variation around scaling as BW3/4
does not invalidate the general scaling argument, nor does it provide evidence for any
different scaling factor. Rather, the variation simply illustrates that any single conception of
interspecies scaling can accommodate only the general trends, not the diversity of particular
instances (Federal Register, 1992b). Clearly, as proposed in Section 3.2.2, when data or
more sophisticated models are available for interspecies extrapolation, they should be used in
preference to the default method presented herein.
Subject to the Assumptions, Equations J-l to J-9 must be manipulated to determine
CEHEC as a function of the average animal exposure concentration, CEA. Because the
concentrations and masses of a parent compound within a compartment are assumed to be
periodic, the integral of the left-hand side (LHS) of Equations J-l and J-2 over a time length
of the period is zero; for example
t+p
(dM/dt') X dt' = M(t + P) - M(t) = 0. (M°)
J-8
-------
p. 101
Also note that for equilibrium or steady state, as in the human case, the LHS of each of these
equations (J-l and J-2) is zero by definition. Performing the period average of both sides of
Equations J-l to J-6, the following are obtained:
(Ml)
(J-12)
0 =Q
= £VKFpi
The steady-state equations for
LHS of these equations to zero
of equations of state for human
redefining the average
overbars).
The above equations are
(Qalv/Hb/g + 04
and Equation J-l2 is expressed
x (C5 - Cj) - ij; j = 1, 2, 3,...n
X CA~ +
i + CA]] (M3a)
j X [C/CKMjj + C:)]|; j = 1 to n (J-13b)
J J J J
QT x CV =
= H
b/g
(J-14)
(J-15)
(J-16)
1; umans are obtained from Equations J-l and J-2 by setting the
(the equilibrium or steady-state condition). The complete set
can be obtained from Equations J-ll through J-16 by
concentilations or terms as equilibrium values (i.e., remove the
mplified by combining Equations J-ll and J-16 to give
x CS = (Qalv x CE)
x CV) - fp ,
(M7)
as
j x CA" = Qj x Cj + TJ; j = 1 to n.
(J-18)
J-9
-------
p. 102
Both sides of Equations J-17 and J-18 are divided by alvQ and (X respectively, to give
uxCA~=CE+wxCV- rp/Qalv, and (J-19a)
CA~ = Cj + Fj/Qj; j = 1 to n, (M9b)
where:
w = Q/Q, and
1 + QT/Qalv).
According to Assumption V, w is species independent. The parameter u is species-dependent
(via Hb/_) and will be identified as such with subscripts A and H for laboratory animal and
human, respectively. For simplicity and unless otherwise noted, averaged concentrations
(indicated by overbar) will be those of animals and nonaveraged (no overbar) concentrations
will be those of humans.
Applied to humans, Equations J-19a and J-19b are written as
UH x CA = CE + w x CV - rpH(CA)/Qalv ,and
CA = Cj + FjHCCp/QjH; j = 1 to n.
For laboratory animals, Equations J-19a and J-19b are written as
uAxCA~=CE+wxCV- rpA/Qalv and (J-20c)
CA = Cj + FJA/QJA; j = 1 to n. (J-20d)
The loss terms in Equations J-3, r (CA) and the ij(Cj)'s, are concave functions with the
property that their second derivatives with respect to CA and Cj, respectively, are less than or
J-10
-------
p. 103
and also,
UA ~ UH - Hb/gA ~ Hb/gH •
Thus, Equation J-24 can be written as
(Hb/ -1 - Hb/ ~*) x CA~ > CE -CE +w x (CV -CV), or (J-26a)
*^A "H
CE > CE + w x (CV - CV) + (Hb/2 -1 - Hb/e -1) x CA~. (J-26b)
*?H *>A
Comparing Equations J-22b and J-23b, and using J-25b one sees that the blood
concentration of the inhaled compound in any human compartment is less than or equal to the
average blood concentration in the corresponding animal compartment; that is
<
(J-27)
Because of Assumption V, (QjA/Qr = QjH/Qr )> it follows from Equation J-14
applied to both humans and animals, and from Equation J-27, that
CV < CV. (J"28)
Thus, the term w x (CV - CV) > 0 can be dropped from Equation J-26b without
affecting the inequality.
CE > CE + (HL. -1 - HK/n -1) X UK (J-29)
Note that CE is the constant inhaled human concentration that would give rise to a human
constant blood level that is no greater than US. If we choose the actual human exposure
concentration to be less than or equal to this CE, as defined by CA = UK, then the actual
human arterial blood concentration will be less than or equal to
J-12
-------
p. 104
equal to zero. As a consequence, the average of each of these functions is less than or equal
to the function evaluated at the average concentration. Suppressing the subscripts, this
property is expressed as
f < r(C). 0-21)
Considering Equations J-21, J-20c, and J-20d, the following is noted:
uAxCA">€E+wxUV- rpACCSVQ,! , and (J-22a)
CX < q + rjA(9/QjA; j = 1 to n. * (J-22b)
Using Equation J-9, Assumption IV (in the presentation notation, CA = CA~ ),
Equations J-20a and J-205 for human are written in terms of the animal arterial blood
concentration by replacing CA with CA~ as follows:
UH x CA~ = CE + w x CV - rpH(CS)/Qalv (J-23a)
US = Cj + rjH(Cj)/QjH; j = 1 to n. (J-23b)
Subtract the LHS and the right hand side (RHS) of Equation J-23a from the LHS and
RHS of Equation J-22a, respectively, to obtain
(UA - UH) X CS > CE - CE + (w X CV - w X CV) - (rpA(€X)/Qalv - rpH(CA-)/Qalv)
pAalvA - pHH
(J-24)
Because of Assumption V, for any concentration value, C,
= rjH(C)/QjH;
J-ll
-------
p. 105
The following two cases are now considered with respect to the partition coefficient.
Case I: Hb/gA> Hb/gjj
The second term on the RHS of Equation J-29 is greater than or equal to zero; thus, the
term can be dropped from the RHS without affecting the inequality. Obviously, with respect
to model assumptions, a conservative human exposure concentration is CE. Therefore, in
terms of the variables in Chapter 4, an estimated conservative NOAEL*jHECj is given by
NOAEL*[HEC] = CE = NOAEL*[ADJ]
where:
the observed NOAEL or analogous effect level concentration
obtained with an alternate approach as described in Appendix A,
adjusted for exposure duration (Equation 4-2).
Case II: Hb/gA < Hb/gH
The second term on the RHS of Equation J-29 is negative in this instance. The inhaled
concentration must be greater than or equal to the exhaled concentration; this requires that
CE > C^ or CA~ < Hb/g x CE. In Equation J-29, CS can be replaced by the larger
value, Hb/ x CE, and still preserve the inequality, hence
CE > CE + (H^'1 - H^'1) x Hb/gA x CE, or (J-31a)
CE > CE X (Hb/gA/Hb/gH). (J-31b)
In this case, an estimated conservative NOAEL*^HECj is given by
J-13
-------
p. 106
NOAEL*[HEC] - (Hb/gA/Hb/gH) X CE - (Rb/gJ^b/g) X NOAEL[ADJ] , (1-32)
where:
= the observed NOAEL or analogous effect level concentration
obtained with an alternate approach as described in Appendix A,
adjusted for exposure duration (Equation 4-2).
J.2 AN EXAMPLE OF THE RELATIONSHIP BETWEEN THE
PROPOSED AND OTHER METHODS
A perspective on the proposed method can be attained by examination of Figures J-2
and J-3, plots of NOAEL*j-HEC] versus NOAEL*jA] for the rat and mouse, respectively.
These plots were created by choosing the equivalent exposure concentration that resulted in
the human arterial blood concentration being equal to the average arterial blood concentration
of the animal, using several methods, for the representative volatile organic compound
dicholoromethane (DCM).
In Figures J-2 and J-3, the "previous" method refers to the method of using the ratio of
the ventilation rate divided by body weight in the laboratory animal to the ventilation rate
divided by body weight in the human ratio for calculating NOAEL*rHECj estimates (Federal
Register, 1980), with the modification that alveolar ventilation rates are used (U.S.
Environmental Protection Agency, 1988a). The NOAEL*|-ADJj of the laboratory animal
(Equation 4-2) is multiplied by the ratio to calculate the NOAEL\HEC] estimate using this
method. "Optimal" method refers to the use of a specific PBPK model with an extensive set
of experimentally determined physiological parameters for the three species (Andersen et al.,
1987a). The same model and human parameters were used for the "similar" method, but the
animal parameters were determined by scaling from the human values, as defined in
Assumption V. The "proposed" results are based on the methods proposed in this document
and derived in this appendix.
In keeping with the results of the derivation that is the subject of this appendix, the
"proposed" NOAEL*rHECj estimates are less than the "similar" method estimates. With
J-14
-------
p. 107
I.OOOi
I
*|100:
LU
10
Dichloromethane
Rat
10
Optimal
Similar
Proposed
Previous
100 1,000
NOAEL* (mg/ma)
10,000
Figure J-2. Plot of NOAEL*^^^ versus NOAEL*[A] for the rat for four possible
methods (proposed, previous, similar, and optimal) of determining
NOAELpjEQ estimates as defined in the text. For any given observed
NOAEL 1*^1, the corresponding HEC estimate is found by going up to the
method(s) line and over to the y axis. The inhaled compound is
dichloromethane. NOTE: NOAEL*[A] = annual NOAEL*[ADJ].
Source: Overton and Jarabek (1989a,b).
respect to the relationship of the proposed predictions to the other methods of calculation, the
following observations are noted.
The "proposed" method lines are parallel to the "previous" lines and result in 3.4 and
6.9 times smaller, or more conservative, NOAEL*jHECj estimates than the "previous"
method for the rat and mouse, respectively. The "proposed" rat NOAEL*jHECj estimates
also fall below (i.e., are more conservative than) those of the "optimal" method by a range of
1.4 to 2.4. Except at high exposure concentrations (above approximately 1,600 mg/m3),
where the estimates are smaller by about 1.3, the "proposed" mouse NOAEL*j-HECj estimates
are up to 1.5 times greater than the "optimal" NOAEL*rHECi estimates. This supports
current evidence that the mouse is not "similar" to humans in some cases (Reitz et al., 1988).
However, for this species, the "proposed" method estimates more closely approximate the
J-15
-------
p. 108
1,000 H
O)
LU
§
100-
10 j
Dichloromethane
Mouse
10
100
1,000
10,000
NOAEL* (mg/m3)
Figure J-3. Plot of NOAEL^pjgc] versus NOAEL*[A] for the mouse for four possible
methods (proposed, previous, similar, and optimal) of determining
NOAElAjjEQ estimates as defined in the text. For any given observed
NOAEL*jA], the corresponding HEC estimate is found by going up to the
method(s) line and over to the y axis. The inhaled compound is
dichloromethane. NOTE: NOAEL*[A] = animal NOAEL*[ADJ].
Source: Overton and Jarabek (1989a,b).
"optimal" method estimates than do the "previous" estimates and the "proposed" method is
conservative (estimates all fall below) the "similar" method. It also should be noted that the
"optimal", "similar", and "proposed" methods result in smaller NOAElAHECi estimates for
the mouse relative to the rat for the same exposure concentration, whereas the previous
methodology results in the opposite relationship of estimates between the two species.
J.2.1 Discussion
Considering the "optimal" method estimates to represent the best possible dose
extrapolation based on internal blood concentrations, then the "proposed" method is more
realistic than the "previous" method. Because the bloodrair partition coefficients are more
J-16
-------
p. 109
readily available than are complete physiological parameter data, the proposed method
represents a simple default approach when extensive PBPK modeling is not feasible.
J.2.2 Research and Development
The approach presented in this appendix has resulted from modeling research focused on
determining the key parameters of gas uptake, distribution, and target tissue accumulation.
Future efforts will incorporate the anatomic and some aspects of the clearance data being
compiled for research to support the particle model described in Appendix G. Model
evaluation plans include comparing the efficiency of various dose surrogates and an approach
to address the apparent nonsimilarity of the mouse. Application of the model to address
mixtures of gases and of dose partitioning between gas and particles is also envisioned.
J-17
-------
|