EPA450/4-84-990
PROTECTION
AGENCY
ESTIMATION OF THE PUBLIC HEALTH DALLAS, TEXAS
RISK FROM EXPOSURE TO GASOLINE VAPOR VIA THE LIBRARY
GASOLINE MARKETING SYSTEM
A Staff Paper
Submitted for Review
to the Science Advisory Board
by the
Office of Health and Environmental Assessment
Office of Air Quality Planning and Standards
Environmental Protection Agency
June 1984
-------
TABLE OF CONTENTS
PAGE
1-1
2-1
2-1
2-1
2-3
2-3
2-4
2-7
2-7
2-10
3-1
3-1
3-1
3-2
3-3
3-3
3-4
4-1
4-1
4-2
4-2
4-4
4-4
4-4
4-4
4-5
4-7
5-1
5-1
5-1
5-2
5-16
5-18
5-25
SECTION
TITLE OF SECTION
1.0 INTRODUCTION
2.0 BACKGROUND
2.1
2.1.1
2.1.2
2.1.3
2.1.3.1
2.1.3.2
2.1.3.3
2.2
3.0 DECISIONS
3.1
3.2
3.2.1
3.2.2
3.2.2.1
3.2.2.2
4.0 ANALYSIS
4.1
4.1.1
4.1.1.1
4.1.1.2
4.1.1.3
4.1.1.4
4.1.2
4.1.2.1
4.1.2.2
The Gasoline Marketing Industry
Industry Structure
Emission Sources
Options for Emission Control
Stage I Controls
Stage II Controls
Onboard Controls
Regulatory History
FACING EPA
Health Basis for Regulation
Control Strategy Decisions
Actions to Reduce Ozone Precursors
Actions to Reduce Risks from Gasoline Vapor Exposure
Regulation of Automobile Refueling
Regulation of Bulk Terminals, Bulk Plants, and
Stations
OF THE GASOLINE MARKETING INDUSTRY
Exposure/Risk Analysis
Selection of Suspect Agents for Evaluation
Benzene
Unleaded Gasoline
Ethyl ene Dichloride (EDO
Ethyl en e Di bromide (EDB)
Service
Assessment of Exposure and Estimated Cancer Incidence
Location and Distribution of Plants
Self-Service Exposure
5.0 EVALUATION OF THE CARCINOGENICITY OF UNLEADED GASOLINE
5.1
5.2
5.2.1 '
5.2.2
5.2.3
5.2.4
Introduction
Animal Studies
Lifetime Inhalation Bioassay in Rats and Mice
(American Petroleum Institute 1983)
90-Day Inhalation Exposure Study with Gasoline
Vapor in Rats and Monkeys (MacFarland 1983)
Renal Toxicity of Gasoline and Related Petroleum
Naphtha in Male Rats (Haider et al . 1983)
Renal Effects of Decalin in General Laboratory
Mammalian Species (Alden et al . 1983)
ii
-------
TABLE OF CONTENTS
SECTION TITLE OF SECTION PAGE
5.2.5 Toxicity of Synthetic Fuels and Mixed 5-26
Distillates in Laboratory Animals
(MacNaughton and Uddin 1983)
5.2.5.1 Studies with RJ-5 Synthetic Fuel 5-26
5.2.5.2 Studies with JP-10 Synthetic Fuel 5-28
5.2.5.3 Studies with JP-4 Mixed Distillate 5-29
5.2.5.4 Studies with JP-5 5-30
5.2.6 Influence of Benzene on the Renal Carcinogenic 5-31
Effects of Unleaded Gasoline Vapor in Male Rats
5.2.7 Conclusions of the UAREP Report (1983) on 5-33
Toxicological Interpretation of
Hydrocarbon-induced Kidney Lesions
5.2.7.1 Assessment of the API Chronic Inhalation Study 5-33
with Unleaded Gasoline Vapor in Rats and Mice
5.2.7.2 Interpretation of the Toxicological Carcinogenic 5-35
Findings in the Carcinogenicity Study with
Unleaded Gasoline by UAREP
5.2.7.3 Review of Human Kidney Lesions 5-36
5.2.7.4 Species and Sex Comparison of the Kidney 5-38
5.2.7.5 Rodent Kidneys and Other Hydrocarbons 5-39
5.2.7.6 Old-Rat Nephropathy 5-40
5.2.7.7 Comparative Nephrotoxicity and Nephrocarcinogenicity 5-41
5.2.7.8 Significance to Humans of the Chronic Inhalation 5-42
Study with Unleaded Gasoline Vapor in Rats and Mice
5.2.8 Summary of Animal Studies 5-43
5.3 Epedemiologic Studies of Petroleum Workers 5-44
5.3.1 Thomas et al. (1980) 5-44
5.3.2 Thomas et al. (1982) 5-47
5.3.3 Rushton and Alderson (1982) 5-48
5.3.4 Summary of Epidemiologic Studies 5-50
5.4 Quantitative Risk Estimation 5-51
5.4.1 Procedures for the Determination of Unit Risk 5-54
5.4.1.1 Low-Dose Extrapolation Model 5-54
5.4.1.2 Selection of Data 5-56
5.4.1.3 Calculation of Human Equivalent Dosages 5-57
5.4.1.3.1 Oral 5-58
5.4.1.3.2 Inhalation 5-60
5.4.1.3.2.1 Case 1 5-60
5.4.1.3.2.2 Case 2 5-61
5.4.1.4 Calculation of the Unit Risk from Animal 5-62
Studi es
5.4.1.4.1 Adjustments for Less Than Lifespan 5-63
Duration of Experiment
5.4.2 Lifetime Risk Estimates 5-63
5.4.2.1 Data Available for Risk Estimation 5-63
5.4.2.2 Choice of Low-Dose Extrapolation Models 5-64
5.4.2.3 Calculation of Unit Risk (Risk at 1 ppm) 5-66
5.4.2.4 Comparison of Risk Estimates by Different 5-67
Low-Dose Extrapolation Models
m
-------
SECTION
TABLE OF CONTENTS
TITLE OF SECTION
PAGE
5.4.2.5 Uncertainties of Quantitative Risk
Assessment
5.4.2.5.1 Uncertainties Associated with Potency
Estimates
5.4.2.5.2 Uncertainties Associated with the Use of
Potency Estimates to Predict Individual
Risks in Real-Life Exposure Patterns
5.4.2.6 Cancer Risk Attributable to Benzene
Content in Gasoline Vapor
5.4.3 Summary of Quantitative Risk Estimation
5.5 Summary and Conclusions
5.5.1 Summary
5.5.1.1 Qualitative
5.5.1.1.1 Animal Studies
5.5.1.1.2 Epidemiologic Studies
5.5.1.2 Quantitative
5.5.2 Conclusions
6.0 ISSUES TO BE ADDRESSED BY THE SCIENCE ADVISORY BOARD
6.1
6.2
APPENDICES
Quality of Evidence
Quantitative Risk Assessment
Appendix A, API Inhalation Study
Appendix B, Comparison Among Different Extrapolation Models
Appendix C, Carcinogenic Potency of Benzene
Appendix D, References
5-71
5-71
5-72
5-73
5-77
5-77
5-77
5-77
5-77
5-79
5-80
5-81
6-1
6-2
A-l
B-l
C-l
D-l
-------
LIST OF TABLES
TABLE TABLE TITLE PAGE
2-1 Estimated Gasoline Vapor Emissions in 1982 2-9
4-1 Unit Risk Factor Summary 4-3
4-2 Lifetime Exposure Estimates for Emission Sources 4-6
Considered in Risk Analysis
5-1 Physicochemical Characteristics of the Test 5-3
Materi al
5-2 Formulation of Unleaded Gasoline 5-4
5-3 Inhalation Exposure Concentrations for a 5-5
Carcinogenicity Study on Unleaded Gasoline
Vapor in Fischer 344 Rats and B6C3F1 Mice
5-4 Body Weight Trends in a Carcinogenicity Study of 5-7
Unleaded Gasoline Vapor in Fischer 344 Rats
5-5 Body Weight Trends in a Carcinogenicity Study of 5-8
Unleaded Gasoline Vapor in B6C3F1 Mice
5-6 Effect of Chronic Exposure of Unleaded Gasoline 5-9
Vapor on Kidney Weights and Kidney/Body Weight
Ratios in Male Fischer 344 Rats
5-7 Kidney Tumor Incidence in Male Fischer 344 Rats 5-12
from Chronic Exposure to Unleaded Gasoline Vapor
5-8 Individual Data on Mineralization of the Renal 5-13
Pelvis and Kidney Tumors in Male and Female
Fischer 344 Rats Exposed to Unleaded Gasoline
Vapor
5-9 Hepatocellular Tumor Incidence in Female B6C3F1 5-15
Mice from Chronic Exposure to Unleaded Gasoline
Vapor
5-10 Design of the 90-Day Inhalation Exposure Study 5-17
5-11 Summary of the Composition and Boiling Ranges of 5-19
the Test Materials
5-12 Nephrotoxic Effects in Rats Following a 21-Day 5-20
Inhalation Exposure to Light Straight-Run
Naphtha
5-13 Nephrotoxic Effects in Rats Following a 21-Day 5-2D
Inhalation Exposure to Light Catalytic Cracked
Naphtha
-------
LIST OF TABLES
TABLE TABLE TITLE PAGE
5-14 Nephrotoxic Effects in Rats Followina a 21-Day 5-21
Inhalation Exposure to Light Catalytic Reformed
Naphtha
5-15 Nephrotoxic Effects in Rats Following a 21-Day 5-21
Inhalation Exposure to Heavy Catalytic Reformed
Naphtha
5-16 Nephrotoxic Effects in Rats Following a 21-Day 5-22
Inhalation Exposure to Full-Range Alkylate
Naphtha
5-17 Nephrotoxic Effects in Male Rats Following a 5-22
21-Day Inhalation Exposure to Full-Range
Naphtha
5-18 Nephrotoxic Effects in Rats Following a 21-Day 5-23
Inhalation Exposure to Polymerization Naphtha
5-19 Nephrotoxic Effects in Rats Following a 21-Day 5-23
Inhalation Exposure to an Unleaded Gasoline
Blend
5-20 Nephrotoxic Effects in Rats Following a 90-Day 5-24
Inhalation Exposure to an Unleaded Gasoline
Blend
5-21 Biological Testing of Decalin, a Prototype 5-25
Volatile Hydrocarbon
5-22 Description of Fuel Inhalation Exposures 5-27
5-23 Incidence Rates of Total Kidney Tumors in Male 5-65
Fischer 344 Rats Exposed to Unleaded Gasoline
Vapor
5-24 Incidence Rates of Hepatocellular Tumors in 5-65
Female Mice (B6C3F1) Exposed to Gasoline Vapor
5-25 Estimates of Carcinogenic Potency Due to Exposure 5-67
to 1 ppm of Unleaded Gasoline Vapor
5-26 95% Upper Bound (and Maximum Likelihood) 5-68
Estimation of Lifetime Risk at Various
Dose Levels, Using Three Different Low-Dose
Extrapolation Models
vi
-------
LIST OF TABLES
TABLE TABLE TITLE PAGE
5-27 Estimates of Carcinogenic Potency of Benzene 5-74
(Risk at 1 ppm)
1 Specifications of Unleaded Motor Gasoline A-27
2 Design of Study A-28
3 Chromatograph of Operating Conditions A-29
4 Termination Times for Animal Groups A-30
5 Tissues Selected for Weighing A-31
6 Tissues Prepared for Microscopic Examination A-32
7 Primary Renal Neoplasms in Rats A-33
1A Formulation of Unleaded Gasoline A-38
2A Specifications of Light Catalytic Cracked Naptha A-39
3A Specifications of Heavy Catalytic Cracked Naptha A-40
4A Specifications of Light Catalytic Reformed Naptha A-41
5A Specifications of Light Alkylate Naptha A-42
6A Detailed Composition of Gasoline A-43
7A Identification of Major Contributors A-44
B-l Maximum Likelihood Estimates of the Parameters B-2
for the Three Extrapolation Models Based on
Three Data Sets in API Unleaded Gasoline Study
C-l Incidence of Zymbal Gland Carcinomas in Female C-l
Sprague-Dawley Rats Administered Benzene by
Gavage
C-2 Incidence of Hematopoietic Neoplasma in C57BL C-3
Male Mice Exposed by Inhalation
C-3 Incidence of Zymbal Gland Carcinomas in Male C-3
Rats (F344) Administered Benzene by Gavage
C-4 Incidence of Zymbal Gland Carcinomas in Female C-4
Rats (F344) Administered Benzene by Gavage
vii
-------
LIST OF FIGURES
FIGURE FIGURE TITLE PAGE
2-1 Gasoline Marketing in the U.S. 2-2
2-2 Bulk Plant Vapor Balance System 2-5
2-3 Service Station Vapor Balance System 2-6
2-4 Onboard Controls 2-8
1 Schematic Diagram of Vapor Generation and A-34
Exposure System
2 Histologic Appearance of a Renal Casrcinoma A-35
Composed of Epithelial cells arranged in a
Tubulo-acinar Pattern.
vin
-------
1.0 INTRODUCTION
The Environmental Protection Agency is conducting a regulatory analysis
of the gasoline marketing system to determine whether the emissions of
gasoline vapor or specific constituents of gasoline vapor warrant control
to protect public health. EPA's analysis also evaluates available options for
the control of emissions from the gasoline marketing system, and the impacts
of such controls. The background for the gasoline marketing industry and
the regulatory history of emission standards that currently control gasoline
vapor emissions are discussed in Section 2 of this paper. The decisions
facing the Agency and the analytical approach used in this analysis are
described in Sections 3 and 4, respectively.
To aid in determining the extent to which regulation may be warranted,
EPA has estimated the potential public health risks posed by gasoline
vapors emitted from bulk storage facilities, delivery trucks, service
station storage tanks, and from motor vehicle tanks during refueling.
The risk assessment focuses on the potential cancer risks associated with
gasoline vapor and three gasoline cemponents: benzene, ethylene dibromide
(EDB), and ethylene dichloride (EDO, substances for which EPA's Carcinogen
Assessment Group (CAG) has derived carcinogenic risk factors. The assessment
estimates cancer risks for exposed populations residing near gasoline
storage facilities and service stations, and as a result of exposure during
self-service refueling of motor vehicles. The results of this analysis are
expected to be released in July 1984.
Based on the preliminary results of this analysis, the cancer risks
estimated from use of the gasoline vapor risk factor dominate the risks
attributable to the three constitutent substances (benzene, EDB, and EDO.
*
The gasoline vapor risk factor is derived from a chronic inhalation study
1-1
-------
in two rodent species, sponsored by the American Petroleum Institute (API).
This study was forwarded to EPA as a "draft final report" in March 1982 and
as a final report in early 1984. At the request of EPA's Office of Mobile
Sources, CAG evaluated the results of the API study and developed a human
carcinogenic risk factor. The review by CAG of the health literature
pertinent to the issue of the carcinogenicity of gasoline vapor and the
derivation of unit risk factors by CAG are presented in Section 5.
The final report of the inhalation study performed by International
Research and Development Corporation for API consists of six volumes.
One copy of the complete six volume set has been forwarded to the chairman
of the Science Advisory Board. In addition, a paper summarizing the results
of the API inhalation study are included within this paper (Appendix A).
Because of the importance of the API study in EPA's gasoline marketing
analysis, EPA requests that the Science Advisory board provide peer review
of this unpublished work as well as the CAG's evaluation of the implications
of this research for human exposure to gasoline vapor. In addition, EPA
requests that the Science Advisory Board address the issues listed in
Section 6 of this paper.
1-2
-------
2.0 BACKGROUND
2.1 The Gasoline Marketing Industry
In 1982, about 103 billion gallons of motor vehicle gasoline were
distributed in the U.S. An extensive network of storage, transportation,
and dispensing facilities were used by refiners, marketers, distributors
and dealers to deliver an average 280 million gallons of gasoline each day
to ultimate consumers. Gasoline produced at domestic refineries or that is
imported (about 3%) is distributed by ship, barge, or pipeline to the
gasoline distribution system of bulk terminals, bulk plants, tanker trucks,
and service stations. The diagram in Figure 2-1 shows the path of motor
vehicle gasoline distribution through these facilities.
2.1.1 Industry Structure
Bulk terminals typically receive most of the motor vehicle gasoline
delivered by ships, barges, or pipelines. About 1500 bulk terminals in
current use store gasoline in large above-ground storage tanks. Separate
tanks are used for each type of petroleum product distributed (e.g., three
grades of gasoline). Typical bulk terminals have 4 to 5 tanks and larger
terminals have more tanks that may be spaced over several acres. From
these bulk terminals, gasoline is loaded into large tank trucks that make
deliveries to local distributors operating bulk plants or directly to
service stations. Gasoline is delivered to a national network of 15,000
bulk plants operating smaller above-ground storage tanks. The gasoline is
distributed from these plants by smaller trucks to businesses, institutions,
or dealers operating retail outlets.
2-1
-------
FI6URE 2-1 GASOLINE MARKETING IN THE U.S.
Imported
Gasoline
Barge
Pipeline
Tanker
Imported
or
Domestic
Crude
Wholesale
Distribution
Level
Conroercial,
Rural
Consumer
Storage
Transport
2-2
-------
About 210,000 retail outlets such as service stations or convenience
stores are currently dispensing gasoline to the public. A roughly equal
number of outlets exist for dispensing gasoline in a wide range of governmental
and private business uses. Government owned outlets fuel Federal, State
and local passenger vehicles, trucks, buses, and military vehicles. Business
outlets include auto rental, utility, taxi, delivery, and commercial trucking
operations.
2.1.2 Emission Sources
Emission of gasoline vapor occurs at each distribution facility in the
gasoline marketing chain. From the time gasoline is received by a bulk
terminal (e.g., from a pipeline) until delivered to the ultimate consumer
through a network of bulk storage tanks, gasoline delivery trucks, and
service stations, gasoline vapor is expelled to the atmosphere each time a
transfer occurs. Displacement of saturated gasoline vapor from tanks during
these gasoline liquid transfers is the primary source of gasoline vapor
emissions. However, the evaporation of gasoline through tank pressure
equalization vents is also a significant contributor to gasoline vapor
emissions from very large storage tanks at bulk terminals.
2.1.3 Options for Emission Control
Control methods for containment of gasoline vapors are available and
are currently being used at many facilities. Equipment designed to contain
and recover gasoline vapors as they are expelled from storage tanks or tank
trucks being filled are designated by EPA as "Stage I" control systems.
Stage I systems provide for the recovery of gasoline vapors from the vessel
being filled into the vessel from which the liquid gasoline is being
discharged (i.e., from the service station underground storage tanks to the
gasoline tank delivery truck, from the truck to the bulk plant, and via
2-3
-------
tank truck, to the bulk terminal where they are liquified by refrigeration
and compression equipment and returned to storage). Stage II control
systems complete the chain of custody of gasoline vapor emissions by
recovering gasoline vapor displaced from an automobile's fuel tank during
filling and returning these vapors to the service stations' underground
storage tanks. In addition to Stage I and Stage II controls, gasoline
vapor emissions from very large storage tanks at bulk terminals can be
controlled by use of a floating roof tank system. With this tank design,
the tank roof is not fixed to the walls at the top of the vessel, but is
supported by pontoons floating on the surface of the gasoline pool. Unlike
a fixed-roof tank where a large amount of gasoline vapor may exist above
the level of liquid gasoline in the tank and may be expelled as the tank is
filled, the floating roof moves with the liquid level and does not create a
large vapor space as the tank is emptied.
2.1.3.1 Stage I Controls
Stage I control systems at bulk terminals and bulk plants collect and
recover gasoline vapors from empty, returning tank trucks as they are
refilled with gasoline from the storage tanks (Figure 2-2). These systems
are now in place at about 2/3 of the bulk terminals and at about half of
the bulk plants in operation in the U.S. Floating roof storage tanks
are also generally in use at those bulk terminals where Stage I controls
have been applied.
Stage I controls are also in use at roughly half of U.S. service
stations (Figure 2-3). Stage I controls at service stations contain the
gasoline vapors within the station's underground storage tanks for transfer
to empty gasoline tank trucks returning to the bulk terminal or bulk plant.
2-4
-------
1/1
>>
OO
V
(J
co
S-
o
Q.
i.
(Q
Sfll
,• ® ^
1 II
CO
CM
I
CM
OJ
i-
3
cn
o
oe
o
Q.
2-5
-------
tn
o
CO
CO
UJ
CO
_l
O
ce
o
u
UJ
00
co
Z
_J
z
UJ
cr
ill
UJ
o.
Q.
U.
O
UJ
o
cr
UJ
2
CD
3
V)
-------
2.1.3.2 Stage II Controls
In addition to underground storage tanks, the other source of gasoline
vapor emissions at service stations (420,000 in use nationally) results
from the uncontrolled venting of displaced vapors from the motor vehicles
fuel tank during filling. With the prevalent use of self-service pumps,
relatively high exposures of a large segment of the population to gasoline
vapors results from motor vehicle refueling operations. About 70 percent
of all gasoline used by the public is dispensed by self-service pumps.
Stage II controls which recover the vapors from vehicle tanks to service
station storage tanks (Figure 2-3) are currently used at about 38,000 service
stations nationally (less than 10 percent) located primarily in California
and in the District of Columbia.
2.1.3.3 Onboard Controls
A feasible alternative for recovery of Stage II emissions displaced
from motor vehicle fuel tanks during refueling consists of the use of
vapor control system designed into new model automobile and light duty trucks.
The onboard system includes a sealed fill pipe and carbon canister that
adsorbs displaced vapors during filling and purges them to the carburetor for
combustion during operation (Figure 2-4). Use of onboard carbon canisters
for control of evaporative emissions of gasoline vapors has been required
on all new automobile production since 1971. Enlargement of this canister
would be necessary to accomodate an increased recovery of gasoline vapor
during motor vehicle refueling.
A summary of the sources and amounts of annual gasoline vapor emissions
are shown in Table 2-1. These emission estimates reflect the influence of
Stage I and Stage II controls that are currently being applied by EPA and
»
State regulations.
2-7
-------
FIGURE 2-4 ONBOARD CONTROLS
CARBURETOR
GASOLINE
GASOLINE TANK NOZZLE
GASOLINE
VAPORS
PURGE CARBON
CONTROL CANISTER SEPARATOR
FILL PIPE MODIFICATIONS
TRAP DOOR
SPOUT
LEAD
2-8
-------
Table 2-1. ESTIMATED GASOLINE VAPOR EMISSIONS IN 1982
Facility Annual Emissions^3)
(Mg/year)
Bulk Terminals
- Storage tanks 52,000
- Truck loading 140,000
Bulk Plants 208,000
Service Stations
- Storage tanks 222,000
- Motor vehicle refueling 407,000
Total National 1,029,000
Emissions
Emission estimates assume controls required by current
EPA and State standards are in place.
2-9
-------
2.2 Regulatory History
The gasoline marketing industry has come under regulatory scrutiny by
EPA in several contexts. Volatile organic compound (VOC) emissions,
including gasoline vapors, participate in atmospheric photochemical reactions
that produce ozone and other constituents of "smog." Emissions of gasoline
vapor are also of concern to EPA due to the potential health risks of
exposure to certain gasoline constituents (benzene, ethylene dibromide and
ethylene dichloride) and due to exposure to gasoline vapor itself. EPA
listed benzene as a hazardous air pollutant in June 1977 and uses of EDB as
a citrus and grain fumigant have been curtailed. In addition, a number of
regulatory actions have been taken by EPA and the States to control gasoline
vapors for the purpose of reducing atmospheric ozone.
Because National Ambient Air Quality Standards (NAAQS) for ozone have not
been attained in all air quality control regions of the U.S., States are
developing additional regulations to control volatile organic compounds to
attain these standards. Control of gasoline vapors has been incorporated
in some State plans since 1974 in ozone non-attainment areas.
In addition to State plans for attainment of the ozone NAAQS, EPA has
issued Federal New Source Performance Standards (NSPS) that require controls
for new, modified, or reconstructed storage tanks. EPA issued these standards
in June of 1973 for bulk gasoline storage tanks with a capacity over 40,000
gallons. Bulk terminal or bulk plant storage tanks affected by these
standards are required to have floating roofs or a vapor recovery system.
EPA has acted to require additional control of gasoline vapor emissions
by bulk terminals. NSPS were proposed in December of 1980 and were promulgated
in August 1983. This action required Stage I controls for all new, modified
or reconstructed storage tanks nationally regardless of ozone NAAQS attainment
2-10
-------
status. The standards also require control of tank truck gasoline vapor
displacement emissions when filled from loading racks servicing these
storage tanks.
Although the gasoline marketing industry has frequently been considered
by EPA as a candidate for regulation on the basis of its photochemical
reactivity and ozone formation contribution, and on the basis of its benzene
content, the Agency has not addressed the need for the regulation of this
industry on the basis of the carcinogenic potential of gasoline vapor itself.
The questions raised by the recent API study concerning the carcinogenic!ty
of unleaded gasoline vapor and the known health effects of gasoline constituents,
including benzene, have stimulated the public debate on this issue and led
advocacy groups to press EPA for a regulatory determination.
On July 14, 1983, The Environmental Defense Fund (EOF) and the National
Resources Defense Council (NRDC) filed a citizen suit to compel EPA to
either take final action on benzene emission source categories or find that
benzene clearly is not a hazardous pollutant pursuant to section 112 of the
Clean Air Act. NRDC v. EPA (D.D.C). The plaintiffs requested that the
court require EPA to promulgate standards on the categories for which
proposals had been made, propose an emission standard for coke oven by-product
plants, and propose standards for benzene emissions from the gasoline
marketing system and "chemical manufacturing plants". On January 27, 1984,
the Court ordered EPA to publish in the Federal Register, by May 23, 1984,
its final action on the source categories for which proposals had been made
(maleic anhydride and EB/S process vents, benzene storage vessels, and
benzene fugitive sources), and to propose action for coke oven by-product
recovery plants. The court did not issue an order concerning chemical
manufacturing or gasoline marketing. In response to this suit, EPA announced
2-11
-------
on May 23, 1984 final standards for benzene fugitive emissions, proposed
standards for coke oven by-product recovery plants, and also acted to
withdraw previously proposed standards for benzene storage vessels, and
maleic anhydride and ethylbenzene/styrene process vents.
A regulatory analysis of the gasoline marketing industry cateogry is
now being completed and a notice of availability of the background document
will soon be published in the Federal Register. No decision has been made
at this time to regulate the gasoline marketing industry under section 112
(hazardous air pollutants) of the Clean Air Act.
In addition to risk to human health resulting from the benzene content
of gasoline, additional animal studies provided to EPA by the American
Petroleum Institute in 1982 seem to indicate that constituents other than
benzene alone in gasoline vapor may have health related impacts. Thus there
are three environmental impacts that may contibute to a decision to further
regulate gasoline marketing sources: gasoline vapor contributions to photo-
chemical smog (ozone), benzene related health risks, and additional gasoline
constituent health risks. All of these potential environmental impacts and
other factors relating to the implementation and costs of controls must be
evaluated by EPA to reach a decision on an appropriate regulatory strategy.
2-12
-------
3.0 DECISIONS FACING EPA
3.1 Health Basis for Regulation
The impact on public health will be a key factor in EPA's decision
on what further controls (if any) may be warranted for the gasoline marketing
industry. Before a decision can be reached with respect to additional
control of gasoline marketing emissions, the health basis for any action
selected must be determined.
Gasoline vapor emissions are precursors to ozone formation and on this
basis alone EPA could consider action to control gasoline marketing operations.
However, actions under these authorities may affect only ozone non-attainment
areas and exposure to gasoline marketing emissions would not be reduced for a
significant portion of the U. S. population. Thus, controls implemented by
EPA to control ozone formation in the atmosphere may not be an adequate
strategy if a significant health risk from exposure to gasoline marketing
emissions is found to exist.
EPA has prepared an analysis of the health risks of exposure to
gasoline vapor from the gasoline marketing system, the regulatory control
alternatives, and associated costs. The document containing this analysis
is expected to be released in July 1984. A description of the risk analysis
methodology is presented in Section 4 of this paper.
3.2 Control Strategy Decisions
In considering the impacts of emissions from the gasoline marketing
system and reaching a decision on the additional controls (if any) that may
be warranted, EPA has authority under several sections of the Clean Air Act
(CAA) by which gasoline vapor emissions could potentially be further regu-
lated. Each authority was granted to address specific risks or categories
of sources. Pollutants for which National Ambient Air Quality Standards
3-1
-------
(NAAQS) have been established (criteria pollutants) may be regulated
through several mechanisms. Authority under CAA section 110 provides for
attainment of NAAQS through a program of State standards development for
new and existing stationary sources. CAA section 111 provides authority
for establishment of Federal standards for new stationary sources, and
Title II of the CAA provides authority for EPA to require installation of
controls by manufacturers on new motor vehicles. For pollutants to which
no NAAQS applies and for which EPA determines a health risk of mortality or
serious illness would result (a hazardous pollutant) CAA section 112 provides
for Federal regulation of new and existing sources of emissions.
Thus, to a degree, the magnitude of the public health risk posed by
gasoline marketing emissions may influence the selection of control strategies
for gasoline marketing emission sources.
3.2.1 Actions to Reduce Ozone Precursors
With NAAQS attainment programs, control is applied only in specific
areas of the U.S. to provide an effective air quality improvement strategy
at minimum cost. Previous actions by EPA to control gasoline marketing
emissions from existing sources have focused on specific regions of the
U.S. where attainment of the ozone NAAQS has been and continues to be a
problem. To assist States in attainment of the ozone NAAQS, EPA has
prepared and published Control Technique Guideline (CTG) documents for
every sector of the gasoline marketing system with the exception of
automobile refueling. EPA has also taken action to control new bulk
terminals under CAA section 111 as part of an overall strategy to reduce
VOC emissions. Thus, many existing bulk terminals, bulk plants, and service
station storage tanks primarily in ozone non-attainment areas and all
storage tanks and loading racks for newly constructed bulk terminals in all
3-2
-------
areas of the U.S. are currently required to install Stage I gasoline vapor
recovery systems. However, gasoline vapor recovery during automobile
refueling remains largely uncontrolled. Stage II recovery systems have
been applied only in California and the District of Columbia.
Should the potential risk to human health resulting from exposure to
gasoline vapor or its constituents be determined to be a significant hazard,
a more widespread systan of controls for the gasoline marketing industry
would be a logical consideration. Since many of the current regulations in
effect are based on a strategy for ozone attainment, many areas of the
U.S. remain uncontrolled with respect to existing gasoline marketing sources
because they either do not have an ozone attainment problem or because
control of VOC emissions from stationary sources other than gasoline marketing
have been adequate to attain the ozone NAAQS.
3.2.2 Actions to Reduce Risks from Gasoline Vapor Exposure
A hazardous pollutant designation for gasoline vapor or a decision
to pursue regulation of the gasoline marketing system on the basis of a
gasoline constituent such as benzene would impose a requirement upon EPA to
reduce exposure for a broader spectrum of the U.S. population than would
be provided by a program to control ozone. The outcome of listing under
section 112 would be regulations for those source categories found to pose
significant health risks,
3.2.2.1 Regulation of Automobile Refueling
EPA could decide that a national program of gasoline marketing vapor
emission reductions should include the automobile refueling operation since
these emissions are currently uncontrolled in all but two areas of the
U.S., and a large segment of the population is exposed to relatively high
concentrations of vapor during self-service gasoline pump use. Two methods
3-3
-------
(Stage II and Onboard) are available for control of gasoline vapor emissions
from fuel tanks. Either of these methods would greatly reduce gasoline
vapor emissions during refueling. However, these approaches would produce
different costs and emission results over time.
Stage II control systems could be retrofitted at the nations's service
stations within a relatively short time span (e.g., in 3 years). This
action would provide more immediate reductions in gasoline vapor emissions.
However, the Stage II nozzles and hoses for gasoline pumps would receive
constant use and would be subject to being torn, crushed, or otherwise
degraded. The experience in California has been that a continuing program
of inspection is needed to maintain the effectiveness of Stage II control
systems. As a result, the annual average control efficiency of Stage II is
estimated to be less than with vehicles using on-board controls.
On-board control systems could be designed into new automobiles and
other light-duty vehicles. Even though the controls would be installed
only on new vehicles and would not provide as much control as Stage II
during the first several years of implementation, the efficiency of the
controls is likely to be better over time than Stage II. In comparision to
Stage II, on-board controls would provide about one-half as much emission
reduction after the fourth model year, about the same amount by the ninth
model year, and superior emission reductions in subsequent years.
3.2.2.2 Regulation of Bulk Terminals, Bulk Plants, and
Service Stations
If gasoline vapors or its constituents are determined to impose a
significant health risk on the public, a national program for implementation
of Stage I vapor recovery systems at bulk terminals, bulk plants, and
service stations would also be evaluated. Regulatory requirement for-
3-4
-------
Stage I control systems now in effect in ozone non-attainment areas could be
extended by section 112 standards to all gasoline marketing facilities in
the U.S. Currently, about one-third of bulk terminals and about one-half
of bulk plants and service stations remain uncontrolled. These uncontrolled
facilities could attain a 90 percent or greater reduction in their losses
of gasoline vapor emissions by implementing Stage I controls.
3-5
-------
4.0 ANALYSIS OF THE GASOLINE MARKETING INDUSTRY
To provide a basis for any regulatory action that may be warranted
and to evaluate potential control strategies, EPA has performed an analysis
of the gasoline marketing industry. The analysis includes an estimate of
the national incidence of cancer expected to result from the public's
exposure to gasoline vapors from all segments of the gasoline marketing
industry. These national incidence estimates were developed for the current
level of gasoline vapor emission control applied by the industry and at
alternative levels of control that could be applied. The analysis evaluates
the emission control efficiency, costs, and cost-effectiveness of alternative
gasoline vapor emission control strategies.
4.1 Exposure/Risk Analysis
Several animal studies of cancer risk resulting from exposure to
gasoline vapor and three of its constituents, benzene, ethylene dibromide
(EDB), and ethylene dichloride (EDO have been reported. Human epidemiological
studies of benzene exposure are also available. From these studies, EPA
has derived unit risk factors for gasoline vapor and those constituents
currently identified as potential cancer risks (i.e., benzene, EDB, and
EDO. Together with other information on the exposed population and exposure
concentration levels for each emission source, EPA has estimated national
cancer incidence resulting from gasoline vapor emissions from the gasoline
marketing system. Incidence estimates were developed assuming no change in
the current level of emission control in all segments of the gasoline
marketing distribution system, and as a result of application of additional
controls on specific segments of the industry. An explanation of the
approach taken to derive these estimates are contained in the following
sections. The results of the risk and incidence analyses are describjed
in a document scheduled for release by EPA in July 1984.
4-1
-------
4.1.1 Selection of Suspect Agents for Evaluation
Gasoline is a complex mixture of over 200 hydrocarbons with the
paraffinic and aromatic compounds constituting the largest fraction.
Aromatics including benzene and toluene are about 20 to 35 percent of the
total gasoline mixture by volume. The majority of these aromatics are alkyl-
benzene compounds; pure benzene accounts for 0.2 to 4.0 percent of the total
gasoline mixture based on 1977 analyses of gasoline produced by several
refiners. The average benzene concentration was found to be 1.3 percent.
In addition to benzene, leaded gasoline contains ethylene dibromide
(EDB) and ethylene dichloride (EDO which are used as lead scavengers. To
improve octane ratings, gasoline contains a large number of hydrocarbons that
have been cracked, reformed, or otherwise chemically altered. At present,
quantitative cancer risk factors are available for only three consist!'tuents
of gasoline vapor (i.e., benzene, EDB, and EDO. A more recent study, conducted
for the American Petroleum Institute, suggests that gasoline vapor may be a
potential human carcinogen. The maximum lifetime risks associated with
gasoline vapor exposure, based on a preliminary analysis, were much greater
than those attributable to the three constituents for which unit risk
factors were available. For this reason, EPA selected gasoline vapor in
addition to benzene, EDB, and EDC for the risk assessment analysis. The
basis for these selections is described for each substance in the following
sections and a summation of the unit risks values are shown in Table 4-1.
4.1.1.1 Benzene. An association between benzene exposure and
leukemia has been documented in several human studies of occupationally
exposed populations. Benzene has also been found to be carcinogenic in
both rats and mice by gavage and inhalation routes of exposure. The benzene
unit risk factor (the risk of cancer resulting from a 70 year lifetime of
4-2
-------
TABLE 4-1. UNIT RISK FACTOR SUMMARY
Unit Risk
(probability of
Pollutant cancer given lifetime
exposure to 1 ppm)
Health Effects
Summary
Comments
Gasoline Vapor
Plausible Upper Limit:b
Rat studies 3.5 x 10~3
Mice studies 2.1 x 10'3
Maximum Likelihood Estimates:
Rat studies 2.0 x 10~3
Mice studies 1.4 x 10-3
Benzene0
Ethylene
Oibromide
Ethylene
Oichloride
2.2 x 10-2
4.2 x 10-1
2.8 x 10-2
Kidney tumors in rats,
liver tumors in mice.
Human evidence of
leukemoginicity
Zymbal gland tumor
in rats; lymphoid
and other cancers
in mice.
Evidence of carci-
nogenicity in animals
by inhalation and
gavage. Rats: nasal
tumors; Mice: liver
tumors.
Evidence of carcino-
genicity in animals.
Rats: Circulatory
system, forestomach,
and glands; Mice:
liver, lung, glands,
and uterus.
Gasoline test samples
in the animal studies
were completely volat-
ilized, therefore may
not be completely repre-
sentative of ambient
gasoline vapor exposures.
EPA: listed as a hazardous
air pollutant, emission
standards proposed.
IARC&: sufficient evidence
to support a causal associ-
ation between exposure
and cancer.
EPA: suspect human
carcinogen; recent
restrictions on
pesticidal uses.
EPA: suspect human
carcinogen. Draft
health assessment
document released
for review March 1984.
a Unit risk factor is in terms of the probability of a cancer incidence (occurrence)
in a single individual for a 70-year lifetime of exposure to 1 ppm of pollutant.
b 95% confidence interval.
c Derived from human epidemological data; not the same factor shown in Table 5-27 based
on animal data.
4-3
-------
exposure to a unit concentration) was derived from the average of three
occupational studies, assuming a linear dose-response function. A unit
factor risk derived from the animal data is very close to the value derived
from the human studies thereby indicating a similar dose-response relationship.
4.1.1.2 Unleaded Gasoline. The evidence of carcinogenicity
comes primarily from the American Petroleum Institute chronic inhalation
study of unleaded gasoline vapor in rats and mice. The unit risk estimates
for each species based on a linear non-threshold dose extrapolation were
derived from this study. Although API studied unleaded gasoline, other
gasoline grades (e.g., leaded gasoline) are expected to have as much
carcinogenic potency. A detailed discussion of the study results and
the risk factor derivation is presented in Section 5 of this paper.
4.1.1.3 Ethylene Pi chloride (EDC). No human evidence of
carcinogenicity is available. The animal evidence consists of positive
responses at several sites in male rats and mice via gavage. The unit risk
for EDC inhalation was estimated by two separate methods: (1) a direct
estimation based on the EDC gavage study, assuming that the absorption rate
by inhalation is one-third of that by the oral route; and (2) an indirect
estimation from the EDB inhalation study. The potencies calculated from
both approaches are similar.
4.1.1.4 Ethylene Dibromide (EDB). No human evidence of
carcinogenicity is available. The animal evidence consists of positive
reponses in mice, in both inhalation and gavage bioassays, as well as nasal
cavity tumors in rats following inhalation exposure. The unit risk was
obtained from the rat inhalation experiment using the linear dose-response
extrapolation procedure.
4-4
-------
4.1.2 Assessment of Exposure and Estimated Cancer Incjdenee
This section briefly outlines the methodology and assumptions used to
estimate the concentrations of benzene, EDB, EDC, and gasoline vapors from
each source category to which the nation's population as a whole (and to
which selected individuals subject to high exposures) would be expected to
be exposed as well as their associated health risks. Estimates of gasoline
vapor exposure were made for each of the source categories of emissions in
the gasoline marketing industry (see Table 4-2). These estimates were
developed for uncontrolled and controlled emission rates. A national mix
of uncontrolled and controlled rates were used to determine current exposure
levels according to the number of emission sources that had controls in
operation in 1982. Estimates of incidence due to EDB, EDC, and gasoline
vapor were projected for the years 1986 through 2020 in proportion to the
total or leaded (for EDB and EDC) gasoline throughput for the source category.
4.1.2.1 Location and Distribution of Plants
Since there are about 1,500 bulk terminals, 15,000 bulk plants, and
420,000 service stations in the United States handling gasoline, limited
resources would not allow modeling each plant individually, even if data
were available regarding exact location and throughput. Model plants
(four for bulk terminals, four for bulk plants, and five for service stations)
for a range of representative gasoline throughputs were used to estimate
exposures nationwide.
In order to calculate exposure to emissions in specific locations (and
the resultant risk) from bulk terminals and plants, assumptions were made
concerning their geographical distribution. The fundamental assumption was
that facilities were located in proportion to the gasoline throughput for
an area. For example, the largest model plants would be located in large
4-5
-------
TABLE 4-2. LIFETIME EXPOSURE ESTIMATES FOR EMISSION SOURCES
CONSIDERED IN RISK ANALYSIS (a)
SOURCE CATEGORY
Bulk Terminals
- loading racks (b)
- storage tanks
- vapor processors
Bulk Plants
- loading racks (b)
- storage tanks
Service Stations (c)
- underground storage tanks
- automobile refueling (d)
Self-Service (fi)
- automobile refueling
UNCONTROLLED CONTROLLED
(ppm) (ppm)
1.41 0.18
0.073 0.015
0.026 0.003
0.029 0.003 (f)
(a) The HEM model was used to estimate community exposure for bulk terminals,
bulk plants, and service stations for the highest exposed population. Actual
measurements of exposure by service station attendants were used for
self-service estimates. The self-service estimates are based on the average
of all the attendant exposure measurements; not the maximum exposures.
(b) The exposure estimates are based on emissions from displacement of gasoline
vapors from the tank trucks. Loading racks are used to fill the tank trucks.
lc) Exposure is estimated for communities nearby service stations.
(d) Exposure estimates include vehicle fuel tank emissions displaced from fill pipe
and emissions from spillage in the vicinity of the gasoline pumps.
(g) Exposure estimates for individuals engaged in self-service from emissions
displaced from the vehicle's fuel tank through the fill pipe.
*
(f) Exposure varies from 0.001 to 0.004 ppm depending on the control system applied,
4-6
-------
urban areas where throughput (and population density) were highest. Further,
each model plant type in each source category (bulk terminals and bulk
plants) was distributed over a range of ten urban area sizes. The largest
terminals, for instance, were assumed to be located in cities ranging in
size from New York City to Des Moines, Iowa; the smallest terminals were
assumed to be located in cities ranging in size from Spokane to Effingham,
Illinois. Estimates were also made of the extent of existing control at
these terminals. Most of those in the large cities (likely to be ozone
nonattainment areas) are currently controlled with proportionately fewer
facilities controlled in the less densely populated areas.
In a similar fashion, model service stations were allocated to 35
localities (multi-county metropolitan areas or single counties), grouped
by seven population size ranges. The model plants were selected to be
representative of the total national service station distribution. The
localities and seven population size ranges were selected to be representa-
tive of the total national population distribution.
Amient concentrations, exposure, and incidence for bulk terminals,
bulk plants, and service stations were calculated using the SHEAR version
of the EPA Human Exposure Model (HEM). The HEM is a model capable of
estimating ambient concentrations and population exposure due to emissions
from sources located at any specific point in the continguous United States.
4.1.2.2 Self-Service Exposure. As with calculation of
incidence due to community exposure, calculation of incidence due to self-
service exposure involves estimates for the unit risk factor, the concen-
trations to which people are exposed, the length of time they are exposed
to the concentrations, and the number of people exposed. The same unit risk
factors were used for self service exposure as for commmmunity exposure.
The concentrations to which people are exposed were estimated based on a
4-7
-------
study conducted by API in which benzene and gasoline vapor concentrations
in the region of the faces of persons filling tanks were measured.
Concentrations for the other pollutants (i.e., EDB and EDO were
calculated using the ratio of the emissions of those pollutants to benzene
emissions. The length of exposure during filling was calculated using a
pumping rate of 8 gallons per minute, or 1.25 minutes per 10 gallons. It
was assumed that 70 percent of gasoline consumption is purchased through
self-service. Calculation of self-service user exposure assumes that
someone is exposed to the concentrations measured for benzene and gasoline
vapors (and prorated for the other pollutants) for 1.25 minutes for each 10
gallons purchased. Since the linear dose-response model is the basis for
the unit risk factor, any exposure (no matter how small) is assumed to
result in some risk of cancer. The risks across the exposed population is
summed to determine the total cancer incidence expected. For self service,
wherein some person is always pumping fuel, the total annual incidence is
directly proportional to annual self service gasoline throughput. Thus,
knowing the throughput, pumping rate, and pollutant concentration, total
annual incidence was calculated.
"Lifetime risk due to high exposure" was calculated using the same
assumptions for exposure as were used for the incidence calculations. These
include the assumptions for the concentration in the person's face during
tank filling (based on API measurements) and the length of time for a tank
filling (1.25 minutes per 10 gallons) However, total gasoline consumption
is not a relevant variable for this calculation. Rather it is important to
know how much of their lifetime individuals experiencing high exposures may
spend filling their tank at self-service stations. The EPA predicted that
*•
people with high exposures (e.g., outside salesmen) could purchase 40 gallons
4-8
-------
of gasoline per week from such stations (i.e., 5 minutes of exposure per
week while the vehicle is refueled) for 50 years of their life.
The estimates of risk, in terms of individual lifetime risk from high
exposure and aggregate incidence, are applicable to the public in the
vicintiy of gasoline marketing sources and those persons that refuel their
vehicles at self-service pumps. This analysis did not examine the risk
to workers from occupational exposure (e.g., terminal workers and service
station attendants). The lifetime risk from high exposure for these
workers is probably substantially higher than for the general public.
In addition, the estimates of aggregate incidence would be higher if such
worker populations were included in the analysis. Of course, any controls
to reduce gasoline marketing emissions would reduce exposure for workers
as well as for the general public.
4-9
-------
5.0 EVALUATION OF THE CARCINOGENICITY OF UNLEADED GASOLINE
5.1 Introduction
The International Research and Development Corporation (IRDC) has
recently conducted an inhalation study of unleaded gasoline vapor in Fischer
344 rats and B6C3F1 mice. The study was conducted at the request of the
American Petroleum Institute (API) to determine the carcinogenicity of
inhaled gasoline vapor. Preliminary results of this study were forwarded
to EPA on March 3, 1982 and a draft report (Appendix A) was forwarded later
on February 24, 1983.
Results of the study have been accepted for publication in the Journal
of the American College of Toxicology. This Section contains a review by
EPA's Carcinogen Assessment Group (CAG) of the draft IRDC report and other
studies related to the health impact of gasoline vapor or its constituents.
5.2 Animal Studies
An evaluation of the likelihood that unleaded gasoline is a human
carcinogen and a basis for estimating its possible public health impact,
including a potency evaluation in relation to other carcinogens, is presented
in this section. The evaluation of carcinogenicity depends heavily on animal
bioassays and epidemiologic evidence. However, other factors, including
mutagenicity, metabolism (particularly in relation to interaction with DNA),
and pharmacokinetic behavior have an important bearing on both the qualitative
and the quantitative assessment of carcinogenicity. This chapter presents an
evaluation of the animal bioassays and relevant toxicity studies, the human
epidemiologic evidence, the quantitative aspects of assessment, and finally,
a summary and conclusions dealing with all of the relevant aspects of the
carcinogenicity of unleaded gasoline.
5-1
-------
5.2.1 Lifetime Inhalation Bioassay in Rats and Mice (International
Research and Development Corporation 1983)'
The following is a review of a final report of an unpublished study on
the carcinogenicity of unleaded gasoline vapor in Fischer 344 rats and B6C3F1
mice. The study was completed by the International Research and Development
Corporation (IRDC) for the American Petroleum Institute (API) in 1983. A
report of this study has been accepted for publication (MacFarland et al. 1984).
The physicochemical properties and the formulation of unleaded gasoline
test sample, as described by the sponsor, are presented in Tables 5-1 and 5-2.
The unleaded gasoline used in the API inhalation study was blended specifically
for the experiment. The test gasoline contained no EDB or EDC (as does leaded
gasoline). The benzene content of the test gasoline was 2.0 percent. In
comparison to commercial unleaded gasoline the test gasoline contained a higher
proportion of benzene (average percentage in commercial gasoline was 1.3 in
1977) and of heavy catalytic cracked naptha (HCCN). Six fractions of the
gasoline blend (HCCN) has been evaluated seperately in animal inhalation
tests. Five of the gasoline fractions induced renal lesions. The HCCN
fraction was the only fraction for which no renal effects were noted (see
section 5.2.3).
Exposures in the API animal inhalation study of the total gasoline vapor
were conducted in 16-m3 glass and stainless steel chambers. Humidity and
temperature within the chambers were approximately 55% and 25°C, respectively.
Gasoline vapor was generated by metering liquid through a heated vaporization
column; the vapor was carried by dry nitrogen to the inlet port of the chamber,
where the vapor was diluted with filtered air at a flow rate of 910 to 1,900
L/min to achieve the desired atmospheric concentrations.
Exposure concentrations are given in Table 5-3. Actual concentrations,
measured by gas chromatography, and nominal concentrations approximated
desired concentrations rather closely.
5-2
-------
The same protocol was used for the study in rats and mice. Rats and mice
were about 6 weeks old when the study began. Initial body weights were: male
rats, 95 to 129 g; female rats, 79 to 105 g; male mice, 14 to 26 g; female mice,
12 to 20 g. Animals were randomly assigned to exposure groups according to
body weight. Three treatment groups, each composed of 100 males and 100 females,
were exposed to measured levels of 67, 292, or 2,056 ppm of gasoline vapor.
An untreated group of 100 males and 100 females was exposed to filtered chamber
air only. Animals were exposed 6 hours/day, 5 days/week until final sacrifice
at 107 weeks (male rats and male mice), 109 weeks (female rats), and 113 weeks
TABLE 5-1. PHYSICOCHEMICAL CHARACTERISTICS* OF THE TEST MATERIAL
(IRDC 1982)
Research octane no.
Motor octane no.
(R+M)/2
Reid vapor pressure, Ibs.
Distillation, ASTM 0-86
IBP, °F
10% evap., °F
50% evap., °F
90% evap., °F
End point, °F
API gravity
Gum, ASTM D381, mg/gal
Sulfur, ppm
Phosphorus, g/gal
Lead, g/gal
Stability, hours
HC analysis, ASTM D1319
Aromatics, vol . %
Olefins, vol . %
Saturates, vol . %
Benzene content, vol . %
92.0
84.1
88.1
9.5
93
116
216
340
428
60.6
1
97
<0.005
<0.05
24+
26.1
8.4
66.5
2.0
aAll of the above information was supplied by the sponsor, the American
Petroleum Institute.
5-3
-------
TABLE 5-2. FORMULATION OF UNLEADED GASOLINE
Generic stream9 CAS number Volume %
Light catalytic cracked naphtha
Heavy catalytic cracked naphtha
Light catalytic reformed naphtha
Light alkylate naphtha
Benzene added to bring to 2%
Butane added to increase Reid vap
64741-55-5
64741-54-5
64741-63-5
64751-66-8
or pressure
7.6
44.5
21.3
22.0
0.8
3.8
Plus: Antioxidant 5 lbs/1,000 bbl
Metal Deactivator 5 lbs/1,000 bbl
aToxic Substance Control Act (TSCA) PL 94-469: Candidate List of Chemical
Substances, Addendum 1, Generic Terms Covering Petroleum Refinery Processed
Streams, January 1978.
(female mice). Ten males and 10 females per group were sacrificed at 3, 6, 12,
and 18 months.
Animals were observed -daily, and body weights were recorded monthly for
the first 17 months and biweekly thereafter. Hematology was evaluated for
seven males and seven females per group at 18 and 24 months. Serum from seven
males and seven females per group was biochemically analyzed at 3, 6, 12, 18,
and 24 months. Ten animals from each dose/sex group were killed after 3, 6, 12,
and 18 months of exposure to provide for periodic histopathologic evaluation.
Survivors, interim sacrificed animals, and decedents were necropsied,
and tissues, organs, and tumors were examined microscopically. Major organs
were weighed.
5-4
-------
TABLE 5-3. INHALATION EXPOSURE CONCENTRATIONS FOR A CARCINOGEN1CITY
STUDY ON UNLEADED GASOLINE VAPOR IN FISCHER 344 RATS
AND B6C3F1 MICE
(IRDC 1982)
Exposure
group
Low
Mid
High
Desired
concentration
(ppm)
50
275
1,500
Nomi nal
concentration3
(ppm)
129
596
2,963
Actual
concentration3
(ppm)
49.7
273
1,501
3The actual concentration data have not been corrected for the "nitrogen
effect" on instrument calibration. Furthermore, an error in chamber airflow
rate calibrations was reported which increased the actual airflow rate to
approximately twice the assumed flow rate. If the corrections discussed in
the study report are applied, the most probable nominal and actual
concentrations were as follows:
Nominal Actual
Exposure concentration concentration
group (ppm) (ppm)
Low 72 67
Mid 310 292
High 1,713 2,056
5-5
-------
Exposure to unleaded gasoline vapor did not affect survival. All groups
of rats and female mice had greater than 50% survival for the entire study, and
survival for all groups of male mice was greater than 50% for at least 95 weeks.
Body weight trends are given in Tables 5-4 and 5-5. Modest reduction of
weight gain was found in male and female rats and male mice in the high-dose
groups. No effect of gasoline vapor on weight gain in female mice was observed.
Organ weights (absolute and organ/body) did not appear to be affected by
treatment with gasoline vapor, with the exception of significant (P < 0.05)
increases in kidney weights and kidney/body weight ratios in high-dose male
rats, as shown in Table 5-6.
At the 3-month interim sacrifice, dose-related nonneoplastic histopathologic
changes were observed in the male rats. These consisted of cortical multifocal
renal tubular basophilia, protein casts, and chronic interstitial inflammation.
The basophilia was present in epithelial cells of renal tubules. The protein-
aceous tubular casts occurred within dilated renal tubules and were commonly
located at the corticomedullary junction. The incidence was 70 and 100% in
mid- and high-dose males, respectively. Chronic interstitial inflammatory foci
with a predominantly lymphoid cell type were observed at 20 and 70% incidence
in mid- and high-dose males; respectively. In addition, renal congestion and
very small foci of renal cortical mineralization were noted in several rats.
In animals dying in the 3- to 6-month interval or sacrificed at 6 months,
the nonneoplastic renal changes in male rats described above were again evident.
The incidence of tubular basophilia was 0, 40, 100, and 100% in control, low-,
mid-, and high-dose male rats, respectively. Proteinaceous casts were observed
in 27% of the control male rats, 80% of the mid-dose male rats, and 100%
of the high-dose male rats. The incidence of chronic interstitial inflammation
was 18, 20, 100, and 100% in control, low-, mid-, and high-dose male rats,
respectively. Mineralization in a radial pattern within the renal pelvis, with
5-6
-------
TABLE 5-4. BODY WEIGHT TRENDS IN A CARCINOGENICITY STUDY OF
UNLEADED GASOLINE VAPOR IN FISCHER 344 RATS
(adapted from IRDC 1982)
Study week
Males
0
13
26
52
78
106
Females
0
13
26
52
78
108
Control
112 + 8
306 ~ 18
348 +" 19
409 + 27
401 + 31
416 +_ 29
93 + 6
173 ~ 11
209 + 12
250 ~ 18
264 T 19
288 ~ 35
Mean body
67 ppm
113 + 8
316 + 15b
361 T 19b
412 ~ 27
406 + 41
403 + 44
93 + 6
186 + lib
210 T 11
256 + 163
274 + 19a
282 + 31
weight + S.D. (grams)
292 ppm
113 + 9
312 + 16a
350 + 20
398 + 24a
393 +~ 20
388 _+ 33
92 + 6
177 + 12a
201 + 12b
249 ~ 18
263 + 21
289 +" 48
2,056 ppm
112 + 8
290 ~ 18b
340 + 16b
376 T 20b
376 + 25b
364 +_ 32
92 + 6
173 ~ 9
192 + IQb
225 + 13b
246 + 16b
255 T 27
aStatistically different from control group at P _< 0.05.
bstatistically different from control group at P < 0.01.
5-7
-------
TABLE 5-5. BODY WEIGHT TRENDS IN A CARCINOGENICITY STUDY OF
UNLEADED GASOLINE VAPOR IN B6C3F1 MICE
(adapted from IRDC 1982)
Study week
control
Mean body weight + S.D. (grams)
67 ppm 292 ppm - 2,056 ppm
Males
0
13
26
52
78
102
Females
0
13
26
52
78
112
22 + 2
30 + 2
33+2
38 + 4
38 + 4
39 7 4
18 + 2
25 + 1
28 + 1
31 + 3
35 + 3
34+3
21 + 2
29 + 2
32 + 29
36 + 3b
37 + 4
37 + 5
18 + 2
25 + 1
28 + 2
32 + 4
35+5
35 + 4
22 + 2
31 + 2b
32 + 2a
35 + 3b
37 + 39
38 7 3
18 + 2
26 + 19
28 + 1
30 7 2
34 + 3b
34 + 3
22 + 2
31 + 2
34 + 2a
35 T 3b
35 +" 3b
35 + 3
18 + 2
26 T lb
29 + 2
30 T 29
32 + 3*>
32 + 3
^Statistically different from control group at P <^ 0.05.
bStatistically different from control group at P <_ 0.01.
5-8
-------
CO
I
13
LU
LU
Z
Q
»-4
Z
O
ce
o
S£
=- <
LU
J CO
co o:
Q CO
LU >-i
3
LU LU
Z 3*
LU Z — •
O i— ' CM
CO
LU CO Ol
3 «•«
CO H- O
o «e o
0. Q£ Q£
LU h— —
X
o to
Z LU
0 3!
ae.
x >-
o o
o
LU CO
o --
1— UJ
o z
LU a
u. — i
Lu 14.
LU
o
- ^c
VO
1
LO
LU
_J
CO
h-
0
•r»
"8
S-
01
CO ^" •"•* ^" ^* *^
CNJ ^* ^H co (*} r** •— < ^ co o
C3O OO OO OO O^H
•*•! •*•! *!•*-( •*•!*! •*•! •*•! -H+l
^3 ^^ *f ^O m CO -3 f^ fO t-rt
f**. (*i 00 CO ^^ f^ '30 OO ^H f**.
csi cfi cu 00 (*) co CM r*» PO co
^ ^ ~ ~ ^ ^ ~ ^ _ ^
OO OO OO OO ^PO
zic^ ^— c^i ^ici ^2, *~ '
u
cviun ^H<*> rom ^m c\j<^.
oo oo oo oa oca
*^ OO ^f ^O C3 CT^ f"*^ f** ^^ ^^
^•f-< voo oocu r«*ro cOt^
CSJCO CMCO <^)00 CMP*. CSJr*-
«-^ —• •. .*-• — fc *— t ^-« -— • — ^ .— - ^^
OO OO OO OO COCO
*-* •—, ^— ^* -^^ ^^ Cl ^. ^_- ^M.
^J O ^O ^" CT*O O^OO i*O>
c*^CO CxJ'sO t-4^- ,— t ^H roo
oo oo oo o-^ o^4
+•1 -»•! •*-! -H -«-l -t-l -t-l +1 -H -t-l
ro GO P1^ P*» **^ ^i <^ ^J C^ ro
• * •• •* *• ••
CMP** CMf*. ror*. CMr*. CM^*
^3
oo ao oo oo vo<^
d *^ dc^ c^ci ^i^l
CM O CMU** Cftvrt CT>O*l COlD
CO CO ^1 *^ *^ tf** ^^ ^f ^O Cft
• • •• • t •• * •
oo oo oo oo oo
•i-l-H -t-l-H + 1 -H +1 -t-l *l +1
cnoi CMr~ r^cM 3* od
CO ^^ ^O CO CD CO r*» O\ CO ^^
CMCO CMP- COCO OJ^O CM>>)>> >1 >>
O O O O O
aTa? a? t? S?4? ara? ai'ai'
cc cc cc cc c c
•a-o -a-a 13-0 -a -a -o-o
x:^ m * n n -*i *i iis<:
. .
LO -H
O O
V | V |
a. a.
^** *"^
Q. a.
1^ O^
el
•u <->
c c
0 0
u u
0 0
*•> *J
^ ^3
I- t-
cx. a.
§§
0 -J
v ai
o u
c c
4) CU
i- u,
a> ai
«- <<-
*- ih-
• -^ t—
•a •« -a
* e c
3 «O <*-!*-
41 ••- t-
c c
VI Ol O>
^— *^ >r«
« I/I VI
e
^ •— t— '
• 13 •— I—
• "*-
« e
-------
material located within tubules or the collecting ducts of the renal pelvis,
was observed in 20% of the high-dose males.
At the 12-month interim sacrifice, the occurrence of proteinaceous casts
in the kidneys of male rats was nearly equal in all groups: 20, 30, 30, and
30% in control, low-, mid-, and high-dose male rats, respectively.
Mineralization in the renal pelvis occurred in 20% of the mid-dose male rats
and in 80% of the high-dose male rats. Progressive glomerulonephrosis was
diagnosed in one high-dose male rat. Another new finding was karyomegaly
(very large nuclei within renal tubular epithelial cells) in male rats.
The complexity of nonneoplastic morphologic alterations observed in the
kidneys of all rats, especially males, increased after 18 months of exposure.
Progressive glomerulonephrosis occurred with higher incidence than previously.
The lesion was characterized by atrophied or sclerosed glomeruli, dilated renal
tubules containing proteinaceous casts, tubular damage with regeneration or
scarring, and the presence of foci of chronic inflammatory cells. The incidence
of glomerulonephrosis in male rats was 20% in controls, 30% in the mid-dose
group, and 20% in the high-dose group; the incidence in female rats was slightly
lower. Proteinaceous casts in the kidneys of male rats were noted in 50, 50, 40,
and 60% of control, low-, mid-, and high-dose male rats, respectively.
Mineralization in the renal pelvis was seen in 20% of the mid-dose and 80% of
the high-dose male rats. Renal congestion was commonly seen, and karyomegaly
was again noted in male rats. A benign renal cortical adenoma was diagnosed in
a high-dose male rat. Mononuclear cell leukemia was diagnosed in the kidney of a
female rat that died during the 12- to 18-month interval.
At the final sacrifice, nearly all male rats exhibited progressive
glomerulonephrosis. The incidence rates were 100, 95, 97, and 100% in control,
low-, mid-, and high-dose male rats, respectively. A slightly lower rate
5-10
-------
of occurrence was seen in female rats. Mineralization in the renal pelvis
occurred in 0, 5, 63, and 91% of the control, low-, mid-, and high-dose males, re-
spectively. Karyomegaly was observed occasionally in the male rats. One mid-
dose male rat had renal tubular epithelial hyperplasia at termination. The
lesion was characterized by the presence of a large dilated tubule containing
a cystic lumen lined by epithelial cells. Renal cysts, epithelial cell
pigmentation, hydronephrosis, chronic interstitial inflammation, congestion,
cortical and pelvic mineralization in female rats, and necrosis were among the
nonneoplastic lesions observed in the 18-month to terminal sacrifice period.
Pathologic examination of the rats revealed a small incidence of renal
tumors in each treated group of male rats (Table 5-7). The first of these
tumors was detected at the 18-month interim kill. Renal carcinomas were found in
each treated group of male rats, with those in high-dose males being significantly
(P < 0.05) increased compared to controls (Table 5-7). A statistical test for
linear trend was significant at the 0.05 level. Renal carcinomas generally
consisted of epithelial cells in a tubular or acinar pattern in the cortex, and
renal adenomas mainly included small masses of epithelial cells forming tubular
or papillary structures in the cortex. Renal sarcomas consisted primarily of
spindle cells in a more pelvic location. The following percentages of final
sacrificed male rats had mineralization of the renal pelvis: control, 0%;
low-dose, 5%; mid-dose, 63%; high-dose, 91%. Mineralization of the renal pelvis
was not found in each kidney with a tumor (Table 5-8); hence, mineralization of
the renal pelvis does not appear to have been a requirement in the etiology of
kidney tumor formation in rats exposed to unleaded gasoline vapor.
Spontaneous kidney tumor formation is rare in male Fischer 344 rats; for
example, Goodman et al. (1979) reported a historical control incidence of one
kidney adenoma (0.05%), two kidney adenocarcinomas (0.11%), and three* benign
5-11
-------
UJ
ai
3
CO
o
Q.
X
UJ
o
1—4
•^
o
ai
:E
o
ai
u.
CO
i—
«t
as
«•
-
UJ
Z
a
1—4
^£
(^«.
1
UO
UJ
_1
CO
<
H-
***
S
n
i.
o o
+j s
3
OI 4->
£
•5 4->
1— VI
U
t-
s_
o
Q.
ia
>
OI
c
*o
v>
ia
OI
i.
a.
CL
i
OI
OI
t_
3
V)
o
a.
X
UJ
c
o
(J
o
OI
^
>L
<3>
w
-j
t—
J3
«r
in
o
o
^^
f-4
o
o
r— (
x^
CM
§
**
0
IO
«—4
X*.
O
X
-x. 01
§«->
i.
o
c u
OI
t3 10
>o S
>— c
10 Ol
C TJ
ai ie
a:
CM
Cfl
IO
<*•
•O
*k
o
o
o
*-4
*x»
<*3
o
o
p«l
-X,
CM
u
§
*•*
•*•*.
,— 1
o
§
p-4
o
TJ
•—. 10 o>
•0 S 4->
§5 «
c •—
C »- 4J
f- U C
U V. 41
t. 10 1.
IO (J Ol
4
^x, ^x.
o a
Ol
g
u
i_
a
VI
13 *«
c »->
41 O
OS t—
VI t- ,—
O O iO
•0 a.e
1 Ol
f t. Ol
Ol I/I
•*- >— o
f 10 -o
C 1
in •>- c
»- 01
•a i-
C 4-> JS
10 C
Ol t£
0) 3
VI CT-0
O * C
•O I/I "3
1 J2
X 3 01
o vi 1/1
r- O
Ol T3
•— CM J= 1
Ol *> X
> C O
01 .- C ,—
ui « -a
O VI Ol C
• 0 4-> -^
a c 10
01 o vi
ai ia •>- ie
.c •- -o e
*J T3 C O
•— c
4-> 41 t-
IO i- I/I U
Ol >0 S.
«-> x «
c « u
IQ VI VI
0 10 C .—
•^ s o 10
*- O •!- C
»- C *J Ol
C •- 0 L.
Ol U Ol
f- fc. VI «4-
VI IO O
u >>
Ol C
10 •— c o
t m -a •—
C i- +J
T3 Ol j* iO
C L. >
Ol Ol L.
• t. -j= Ol
•O -U 4-> *> VI
01 L. ja
C S- o *- O
T-
Oi f- s vi
•5 >, c
t- t. U »— T-
Ol O 41 iO
J2 1- 4-> C -O
§c 10 oi
VI f- 4->
C T- S- r-
-x VI —. 41 3
t- >,CM S. VI
§»- SO «-> 4»
10 ej> (. S-
5C — . 3
« — u. -
•>.
c. ~- o -o
*J - o ae vi *j
3 f *- • *-> v>
*J 10
!_ VI Ol 1- Ol
oi •— j= jr
fU 4-> 4) +J
4-> C "5 C
Z co i-i S o
10 .a o
•o oi o
c > e «. c
10 i- ai o i- 4->
> 13 >»- ia vi
" *- € <"
CM 3 j= oi o! t-
*+ i/l O > l-
S-r- 4->
> oi u
~O 4-> L. i— IO
§3 J3 X
vi ia ui
ai t. 4->
X '*- o vi
4-» O 4-> VI
IO ^— "O T- &.
^ j=
41 3 X VI
U Ol r— O C • f-
•i— 4- O . — •»- CL U.
<«- 4) X >— 3
i- X Ol iO C O >,
S- X t- J3
-------
suggest that a maximum tolerated dose was approached; however, mineralization
in the kidney indicates that exposure to unleaded gasoline vapor produced
toxicity in this organ in each treated group of male rats. Applying the
International Agency for Research on Cancer (IARC) classification approach for
carcinogens, the Carcinogen Assessment Group concludes that these studies
furnish sufficient evidence for the carcinogenicity of unleaded gasoline vapor
in animals under the conditions of the bioassay.
TABLE 5-9. HEPATOCELLULAR TUMOR INCIDENCE IN FEMALE B6C3F1 MICE
FROM CHRONIC EXPOSURE TO UNLEADED GASOLINE VAPOR
(IRDC 1982)
Hepatocellular
tumor type
0 (control)
Exposure group (ppm gasoline vapor)
67 292 2,056
Adenoma
Carcinoma
Adenoma and carcinoma
combined
1/lOOa
7/100
8/100
4/100
6/100
10/100
3/100
9/100
12/100
7/100
20/lOOb.c
27/10Qb,C
aNumber with tumor/number examined.
bStatistically significant (P < 0.01) increase compared to control group.
cThe 100 animals in each denominator in this table includes 40 animals sacrificed
at 3, 6, 12, and 18 months and decedents and survivors in the remaining 60 ani-
mals which were allowed to survive for the duration of the study. If the 40
interim sacrificed animals are excluded from each denominator to allow replace-
ment of the 100 total animals with the 60 animals allowed to survive for the
duration of the study, the statistically significant differences shown in this
table remain significant at P < 0.01.
5-15
-------
5.2.2 90-Day Inhalation Exposure Study With Gasoline Vapor In
Rats and Monkeys (MacFarland 1983)
A 90-day inhalation exposure study of the toxicity of unleaded gasoline
vapor in Sprague-Dawley rats and squirrel monkeys was performed as a pre-
chronic test in preparation for the carcinogenicity study with unleaded
gasoline in rats and mice. In the 90-day study, rats and monkeys were exposed
6 hours/day, 5 days/week for 13 weeks to vapors of an unleaded EPA reference
gasoline and a leaded commercial gasoline, as shown in Table 5-10. The hydro-
carbon composition of the two gasolines was similar, but the unleaded gasoline
contained 5 mg/gallon of lead and the leaded gasoline contained 1.94 g/gallon of
lead. The animals were examined for mortality, body weight, food consumption,
toxic signs, hematological changes, urinary changes, tissue lead levels, and
pathology. Pulmonary function tests and cortical flash-evoked response tests
were also done on the monkeys.
Some female monkeys in Groups III and V showed emesis. Body weights in
male rats in Groups II and IV were significantly greater at termination.
Female rats in Group III had increased reticulocyte counts, and some rats in
Group V had increases in hematocrit and mean corpuscular volume, and decreases
in white cell count and mean corpuscular hemoglobin concentration.
Male monkeys in Groups III and V had an increased minute volume. Female
monkeys in Group III had a reduced respiratory rate, and female monkeys in
Group V had a decreased tidal volume at termination.
Liver weights were increased in male rats in Groups II and IV and
decreased in Group V female rats. Kidney weights were increased in Group IV
female rats and Group V male monkeys. Thyroid weights were" increased in male
monkeys in Groups II and III. Heart/body weights were decreased in male rats
5-16
-------
in Groups IV and V, and brain weights were decreased in male rats in Groups
II and III. Group V female rats had decreases in liver/body and adrenal/body
weights.
The initial pathological examinations showed no treatment-related effects.
Histopathologic reexamination of tissue sections showed subtle but discernible
changes in kidneys of Group III male rats. These changes were described as
an increase in the incidence and severity of regenerative epithelium, and
proteinaceous material in dilated tubules was found.
TABLE 5-10. DESIGN OF THE 90-DAY INHALATION EXPOSURE STUDY
Number3 and Species of Animals
Concentration
group
I. Control
II. Unleaded gasoline
III. Unleaded gasoline
IV. Leaded gasoline
V. Leaded gasoline
Rats
40
40
40
40
40
Monkeys
8
8
8
8
8
Dose (ppm)
0
384
1,552
103
374
aEqually divided as to sex.
5-17
-------
5.2.3 Renal Toxicity of Gasoline and Related Petroleum Naphtha In
Male Rats (Haider et al. 1983)
The renal effects of subchronic inhalation exposure of male and female
Sprague-Dawley rats to vapors of unleaded gasoline and related petroleum
naphthas described in Table 5-11 have been reported by C. A. Haider, T. M. Warne,
and N. S. Hartoum at the Workshop on the Kidney Effects of Hydrocarbons, held
in Boston, MA, on July 18-20, 1983. The results of this study are presented
in Tables 5-12 through 5-20. This study is especially pertinent to the
unleaded gasoline carcinogenicity study in that it gives an indication as to
which fractions in unleaded gasoline can produce kidney toxicity.
Exposure of male and female Sprague-Dawley rats to unleaded gasoline for
21 days induced mild tubular degenerative and regenerative changes with
increases in hyalin droplets in the renal cortex in males. Corticomedul lary
tubular dilatation and necrosis were found in one high-dose male rat.
A 90-day exposure to unleaded gasoline resulted in a treatment-related
incidence of tubular dilatation and necrosis at the corticomedullary junction
in male rats, along with a dose-related severity. The persistence of these
lesions during a 4-week recovery period suggests an irreversible effect.
Similar 21-day exposures of rats to full-range alkylate naphtha,
polymerization naphtha, light catalytic reformed naptha, and light straight-
run naphtha induced renal lesions in males similar to those obtained with
unleaded gasoline treatment. Milder renal lesions were found in males exposed
to light catalytic cracked naphtha. No renal effects were noted with exposure
to heavy catalytic reformed naphtha.
The results of these studies suggest that paraffin and alkene fractions
are effective as renal toxicants and that aromatics are relatively non-toxic.
The unleaded gasoline blend included some of the naphtha materials tested in
5-18
-------
this study, and although the unleaded gasoline composition is proprietary,
it was mentioned that it contained 22% full-range alkylate naphtha, a fraction
which could be a significant factor in renal toxicity induced by unleaded
gasoline exposure.
TABLE 5-11. SUMMARY OF THE COMPOSITION AND BOILING RANGES
OF THE TEST MATERIALS
Composition (%)
Material
Light straight-run naphtha
Light catalytic-cracked naphtha
Light catalytic-reformed naphtha
Heavy catalytic-reformed naphtha
Full -range alkylate naphtha
Polymerization naphtha
Unleaded gasoline blend
Paraffins3
96
39
67
7
98
8
45b
Olefins
0
32
2
0
2
92
12b
Aroma tics
4
29
31
93
0
<1
43b
Boiling range (°F)
IQ% bp
71
174
137
290
124
205
112
90% bp
222
346
230
364
315
353
326
alncludes cyclo-, normal, and branched.
bEstimated.
5-19
-------
TABLE 5-12. NEPHROTOXIC EFFECTS IN RATS FOLLOWING A 21-DAY INHALATION
EXPOSURE TO LIGHT STRAIGHT-RUN NAPHTHA
Group and Incidence^
concentrations3 ftF
Environmental control 0/10 0/10
Sham control 0/10 0/10
1.50 mg/L (395 ppm) 0/10 0/10
5.13 mg/L (1349 ppm) 0/10 0/10
14.56 mg/L (3829 ppm) 3/10 0/10
Analytical time-weighted average in mg/L (ppm).
^Incidence of tubular dilation and necrosis at corticomedullary junction,
TABLE 5-13. NEPHROTOXIC EFFECTS IN RATS FOLLOWING A 21-DAY INHALATION
EXPOSURE TO LIGHT CATALYTIC CRACKED NAPHTHA
Group and
concentrations3 Effects
Sham control
0.20 mg/L (43 ppm)
Evidence of early
2.04 mg/L (434 ppm) degenerative changes in
kidneys of treated male rats.
13.06 mg/L (2,777 ppm)
Analytical time weighted average in mg/L (ppm).
5-20
-------
TABLE 5-14. NEPHROTOXIC EFFECTS IN RATS FOLLOWING A 21-DAY INHALATION
EXPOSURE TO LIGHT CATALYTIC REFORMED NAPHTHA
Group and Incidenceb
concentrations3
M F
Environmental control 0/10 0/10
Sham control 0/10 0/10
2.00 mg/L (544 ppm) 0/10 0/10
5.85 mg/L (1,591 ppm) 1/10 0/10
20.30 mg/L (5,522 ppm) 3/10 0/10
aAnalytical time weighted average in mg/L (ppm).
blncidence of tubular dilation and necrosis at corticomedullary junction.
TABLE 5-15. NEPHROTOXIC EFFECTS IN RATS FOLLOWING A 21-DAY INHALATION
EXPOSURE TO HEAVY CATALYTIC REFORMED NAPHTHA
Group and Incidence**
concentrations
Environmental control
Sham control
1.03 mg/L (215 ppm)
2.81 mg/L (587 ppm)
10.20 mg/L (2132 ppm)
M
0/10
NEC
NE
0/10
0/10
F
0/10
NE
NE
0/10
0/10
Analytical time weighted average in mg/L (ppm).
blncidence of tubular dilation and necrosis at cortico-medullary junction.
CNE = Not examined. Pathology was not done due to lack of adverse effects
at higher concentrations.
5-21
-------
TABLE 5-16. NEPHROTOXIC EFFECTS IN RATS FOLLOWING A 21-DAY INHALATION
EXPOSURE TO FULL-RANGE ALKYLATE NAPHTHA
Group and Incidence^3
concentrations^
M F
Environmental control 0/10 0/10
Sham control 0/10 0/10
1.54 mg/L (345 ppm) 10/10 0/10
4.92 mg/L (1,104 ppm) 10/10 0/10
15.31 mg/L (3,434 ppm) 10/10 0/10
Analytical time weighted average in mg/L (ppm).
^Incidence of tubular dilation and necrosis at cortico-medullary junction
TABLE 5-17. NEPHROTOXIC EFFECTS IN MALE RATS FOLLOWING A REPEAT 21-DAY
INHALATION EXPOSURE TO FULL-RANGE ALKYLATE NAPHTHA
Group and Incidence^ in males
concentrations3
Sham control 0/40
0.015 mg/L (3 ppm) 0/20
0.152 mg/L (34 ppm) 4/10
1.538 mg/L (345 ppm) 11/20
Analytical time-weighted average in mg/L (ppm).
Incidence of tubular dilation and necrosis at corticomedullary junction.
5-22
-------
TABLE 5-18. NEPHROTOXIC EFFECTS IN RATS FOLLOWING A 21-DAY INHALATION
EXPOSURE TO POLYMERIZATION NAPHTHA
Group and
concentrations3
Incidence*5
M
Environmental control
Sham control
1.04 mg/L (215 ppm)
3.05 mg/L (632 ppm)
9.89 mg/L (2,049 ppm)
0/10
0/10
0/10
2/10
4/10
0/10
0/10
0/10
0/10
0/10
^Analytical time-weighted average in mg/L (ppm).
bIncidence of tubular dilation and necrosis at corticomedullary junction,
TABLE 5-19. NEPHROTOXIC EFFECTS IN RATS FOLLOWING A 21-DAY
INHALATION EXPOSURE TO AN UNLEADED GASOLINE BLEND
Group and
concentrations3
Incidence'3
M
Environmental control
Sham control
0.11 mg/L (29 ppm)
1.58 mg/L (416 ppm)
12.61 mg/L (3,316 ppm)
0/10
0/10
0/10
0/10
1/10
0/10
0/10
0/10
0/10
0/10
Analytical time-weighted average in mg/L (ppm).
blncidence of tubular dilation and necrosis at corticomedullary junction.
5-23
-------
TABLE 5-20. NEPHROTOXIC EFFECTS IN RATS FOLLOWING A 90-DAY
INHALATION EXPOSURE TO AN UNLEADED GASOLINE BLEND
Incidence15
Group and
concentration^
Environmental control
Sham control
0.15 mg/L (40 ppm)
1.44 mg/L (379 ppm)
14.70 mg/L (3,866 ppm)
Terminal
M
0/10
0/10
1/10
7/10
5/10
sacrifice
F
0/10
0/10
0/10
0/10
0/10
Four-week
M
0/10
0/10
1/10
5/10
4/10
recovery
F
0/10
0/10
0/10
0/10
0/10
Analytical time-weighted average in mg/L (ppm).
blncidence of tubular dilation and necrosis at cortico-medullary junction.
5-24
-------
5.2.4 Renal Effects of Decat in in Several Laboratory Mammalian
Species (ATden et al. 1983)
A comparison of the renal effects in various laboratory mammalian species
exposed to decalin, a prototype volatile hydrocarbon, was discussed in "The
Pathogenesis of the Nephrotoxicity of Volatile Hydrocarbons in the Male Rat,"
by Carl L. Alden, R. L. Kanerva, G. Ridder, and L. C. Stone, at the Workshop
on the Kidney Effects of Hydrocarbons, held in Boston, MA, on July 18-20,
1983. One of the observations presented was that a 91-day inhalation exposure
to 5 ppm and 50 ppm decalin induced renal toxicity in male Fischer 344 rats
and not in females, male or female mice, male or female dogs, and male or
female guinea pigs (Table 5-21). The observed renal effects in male rats
included hyalin droplets in the cytoplasm of proximal convoluted tubular
epithelial cells, granular casts at the junction of the inner and outer band
of the outer zone of the medulla, and augmented chronic glomerulonephropathy.
These droplets consist of an alpha2U globulin, a protein synthesized in the male
rat liver under the control of testosterone and endogenous corticosterone.
They occur spontaneously in sexually mature male rats but not in castrated
males, in female rats, or in humans.
TABLE 5-21. BIOLOGICAL TESTING OF DECALIN, A PROTOTYPE VOLATILE HYDROCARBON
Species tested Renal injury Reference
Rat (male/female) +/- AFAMRL-TR-79-121
(Wright-Patterson AFB)
Mice (female)
Dog (male/female) -/-
Guinea pig (male/female) -/- AFMRL-TR-78-55
(Wright-Patterson AFB)
»
Mice (male) - Dr. Logan Stone
(personal communication)
5-? s
-------
5.2.5 Toxicity of Synthetic Fuels and Mixed Distillates In Laboratory
Animals (MacNaughton and Uddin 19337
Toxicity studies on mixed distillates and synthetic fuels in experimental
animals have been or are being done by the United States Air Force, and
preliminary results of these studies were reported by M.G. MacNaughton and
D.E. Uddin at the Workshop on the Kidney Effects of Hydrocarbons, held in
Boston, MA, on July 18-20, 1983. The studies are summarized below. The
design of the experiments, in which the agents were given by inhalation, is
shown in Table 5-22. Beagle dogs, Fischer 344 rats, Golden Syrian hamsters,
and C57BL/6 mice were used.
5.2.5.1 Studies with RJ-5 Synthetic Fuel
RJ-5 fuel consists of hydrogenated dimers of norbornadiene with a vapor
pressure of 1.3 kPa at 103°C.
Results of the studies with a one-year exposure to 30 mg/m3 and 150 mg/m3
were as follows:
1. Decreased body weight gain in rats and dogs, with possible appetite suppression.
2. Acute lung inflammation and some bronchopneumonia in rats and dogs
sacrificed immediately after 6 months of treatment.
3. After a 1-year holding period, there was a 25% incidence of alveolargenic
carcinomas in CF-1 mice (the strain shown in the workshop proceedings),
a strain predisposed to this tumor type.
The results of the studies in which dogs, mice, hamsters, and rats were exposed
to 30 or 150 mg/m3 for 1 year followed by a 1-year holding period were as follows:
1. Decreased body weight gain in exposed male rats and male hamsters throughout
the study. Increased body weight gain in exposed rats during treatment
was reversed during the post-treatment period.
5-26
-------
TABLE 5-22. DESCRIPTION OF FUEL INHALATION EXPOSURES
Fuel
Exposure
(months)
Concentration
(mg/m3)
Sped esa
End date
Synthetic
JP-10
RJ-5
RJ-5
Mixed Distillate
12, intb
6, contc
12, int
560
155
30, 150
D,R,M/F,H
D,R,M/F,M
D,R,M/F,H
*D (dogs); R (rats); M (monkeys); M/F (mice, female); H (hamsters)
^Intermittent (6 hours/day, excluding weekends and holidays).
cContinuous.
dShale.
Completed
Completed
Compl eted
JP-4
JP-4
JP-4
JP-5
JP-5 (S)d
JP-7
JP-8
JP-TS
DFM
DFM(S)
8,
3,
12,
3,
3,
12,
3,
12,
3,
3,
int
cont
int
cont
cont
int
cont
int
cont
cont
2,500,
500,
500,
150
250,
150,
500,
200,
50,
50,
5
1
1
1
1
,000
,000
,000
750
750
750
,000
,000
300
300
D,
D,
R,
D.
D,
R.
R,
R.
R.
R,
,R,M,M/F
,R,M/F
,M/F
,R,M/F
,R,M/F
,M/F
,M/F
,M/F
,M/F
,M/F
Compl
Dec.
Jul.
Compl
Compl
Dec.
Jul.
Dec.
Compl
Compl
eted
1983
1984
eted
eted
1985
1985
1985
eted
eted
5-27
-------
2. Decreased (P < 0.05) kidney/body weights in exposed female rats.
3. Four (7%) renal cell adenomas and five (8%) renal cell carcinomas in high-
dose male rats; one (2%) renal cell carcinoma in a low-dose male rat. No
renal cell carcinomas were seen in controls.
4. Other kidney lesions in exposed male rats were:
Incidence
Lesion
Renal medullary
mineralization
Moderate pelvic
urothelial hyperplasia
Hyalin droplets
Cortical cysts
5.2.5.2 Studies
150 mg/m3
57/62 (92%)
58%
18%
24%
with JP-10 Synthetic
30 mg/m3
2/59 (3%)
7%
19%
2%
Fuel
Control
0%
2%
2%
0%
JP-10 fuel is a bicyclic, bridged compound; exotetrahydrodi(cyclopentadiene),
with a vapor pressure of 1.87 kPa.
Results of the studies with a 1-year exposure at 562 mg/m3 followed by a
1-year recovery period were:
1. Slight weight loss in exposed rats and hamsters.
2. Hepatocellular vacuolization in 50% of the control and 75% of the exposed
female mice.
3. Nine renal cell carcinomas in treated male rats compared to one in contols;
poorly differentiated malignant neoplasms in one control and one treated
male rat.
4. Other renal effects in male rats were:
5-28
-------
Lesion
Incidence
Treated
Control
Augmented renal
tubule degeneration
compatible with old-
rat nephropathy
Medullary mineral deposits
(mineralized cell debris)
Papillary hyperplasia of
renal pelvic epithelium
43/49 (87%)
100%
26/49 (53%)
32/49 (65%)
2/49 (4%)
5. No toxic lesions in female mice.
6. Adrenal cortical adenomas and carcinomas were found in 27% of the control
and 28% of the treated male hamsters; however, adrenal zona glomerulosa
adenomas and adrenal zona glomerulosa hyperplasia were found in 14% and
72% of treated male hamsters, respectively, and 5% and 45% of control male
hamsters, respectively.
5.2.5.3 Studies with JP-4 Mixed Distillate
JP-4 mixed distillate has characteristics similar to gasoline and has a
vapor pressure of 13 kPa. JP-4 represents 85% of the turbine fuel used by
the Department of Defense.
Results of studies with an 8-month intermittent exposure to 2,500 and
5,000 mg/m3 (containing 80 mg/m^ benzene) were:
1. Increased organ and organ/body weights for kidney, liver, spleen, and
lung in exposed male rats.
2. A 27% incidence of bronchitis in exposed rats.
3. A transient increase in red blood cell fragility in female dogs.
5-29
-------
Results of studies with a 90-day continuous exposure to 500 and 1,000
mg/m3 with a 19-month holding period were:
1. Increases in serum globulin and total protein and BUN in low-dose and high-
dose dogs.
2. Decreased body weight gain in exposed male and female rats during treatment.
3. Centrilobular hepatocellular fatty change in 88% of the low-dose and 89% of
the high-dose female mice.
4. Kidneys of all exposed male rats contained hyalin droplets in the proximal
tubular epithelium, and focal dilatation of renal tubules near-the
corticomedullary junction with plugging by cellular debris was found in 96%
of the low-dose and 100% of the high-dose male rats.
Results of studies with a 1-year exposure to 500 and 1000 mg/m3 were:
1. Decreases in body weight and in kidney and liver weights in high- and low-
dose male rats.
2. Decreases in spleen and kidney weights in low dose female rats.
3. Histopathological examination of tissues is ongoing.
5.2.5.4 Studies with JP-5
JP-5 mixed distillate is the other major turbine engine fuel besides
JP-4. Results of studies with a 90-day continuous exposure of dogs, rats,
and mice to 150 and 750 mg/m3 and a 19 month post-exposure period were:
1. Decreased body weight in exposed male rats.
2. Increased BUN and serum creatinine in male and female high-dose rats.
3. Mild, diffuse fatty change with small vacuoles in hepatocytes of 3% of
the control, 73% of the low-dose, and 24% of the high-dose mice.
"Foamy" hepatocellular cytoplasmic vacuoles were found in 18% of the control,
15% of the low-dose, and 44% of the high-dose female mice.
5-30
-------
4. Male rats sacrificed at the end of the 90-day exposure period had dilated
renal tubules filled with granular necrotic debris at the cortico-
medullary junction.
5. By 19 months post-treatment, old-rat nephropathy was evident in 96%,
96%, and 84% of the high dose, low dose, and control males, respectively.
Old-rat nephropathy was more severe in treated males. Renal medullary
tubular mineralization was found in 82% of the high-dose, 59% of the low-
dose, and none of the control male rats. A dose-related focal hyperplasia
of the renal pelvis was reported.
In summary, chronic exposure to RJ-5 and JP-10 synthetic fuels induces
a common pattern of nephrotoxicity leading to renal carcinomas. Similar
studies with JP-4 and JP-5 show the same preneoplastic lesions, but histo-
pathological analysis is currently incomplete and no information is available
about neoplastic response.
5.2.6 Influence of Benzene on the Renal Carcinogenic Effects of
Unleaded Gasoline Vapor in Male Ra"t?
According to an October 1983 draft report of a bioassay of benzene by
the National Toxicology Program (NTP 1983) in which Fischer 344 rats and
B6C3F1 mice were tested, several organ sites in mice and rats had benzene
induced carcinomas. These included Zymbal gland and oral cavity carcinomas
in male and female rats; skin carcinomas in male rats; Zymbal gland, prepugial
gland and lung carcinomas in male mice and mammary, ling, and hepatocellular
carcinomas in female mice.
Data from the NTP report are discussed in section 5.4.2.6 and Appendix B,
and from the Maltoni et al. experiment are discussed here. In these experiments,
male and female Sprague-Dawley rats were either dosed with 50 or 250 mg/kg of
benzene by gavage 5 days/week for 52 weeks, exposed by inhalation 4 to 7
5-31
-------
hours/day to 200 ppm benzene for 15 weeks or to 200-300 ppm benzene for 104
weeks, or dosed with 500 mg/kg of benzene by gavage 4-5 days/week for 104
weeks. Animals in these studies were allowed to survive until spontaneous
death before pathological examination. Treated animals were compared with
controls.
In the study in which rats were dosed with 50 and 250 mg/kg of benzene,
there was a dose-related increase in mortality. Dose-related increases in
Zymbal gland carcinomas, "haemolymphoreticular" neoplasms, and mammary
carcinomas were reported. Two carcinomas of the oral cavity, one subcuta-
neous angiosarcoma, and one hepatoma were also found in treated animals.
No carcinogenic effect from 15 weeks exposure of 200 ppm benzene was
observed. Exposure to 200-300 ppm benzene for 104 weeks increased mortality,
and Zymbal gland carcinomas and hepatomas were attributed to treatment in
this study.
Gavage treatment with 500 mg/kg of benzene decreased body weight and
induced hematological effects. Zymbal gland carcinomas and carcinomas of
the oral cavity were concluded to be treatment-related.
None of the tumor types attributed to benzene treatment in the studies
by Maltoni et al. (1982) and NTP (1983) were found as treatment related
effects in the carcinogenicity study with unleaded gasoline vapor (104-week
exposure) in male Fischer 344 rats. Conversely, the kidney tumors in male
Fischer 344 rats in the study with unleaded gasoline vapor were not a
treatment-related effect in the studies of Maltoni et al. (1982) in
Sprague-Dawley rats. Nonneoplastic renal lesions of similar morphology have
been found in male Sprague-Dawley rats and male Fischer 344 rats exposed to
unleaded gasoline blend as well as other hydrocarbons in toxicity studies
5-32
-------
with inhalation exposure. Renal toxicity was not indicated in the Maltoni et
al. (1982) report. Taking all of this evidence together, it would appear
that the kidney has not yet been shown to be a target organ for benzene
carcinogen!city in male rats. Furthermore, the benzene level compared to other
ingredients in unleaded gasoline, which was completely volatilized in the
carcinogenicity study, was relatively low at 2%, and comparison with similar
toxic and carcinogenic effects induced in the kidneys of male rats by other
hydrocarbons, including mixed distillates, synthetic fuels, and other hydro-
carbons without benzene, indicates that the hydrocarbon nature of unleaded
gasoline was pivotal in the induction of renal carcinomas in male Fischer 344
rats. Snyder et al. (1980) found an increased number of hematopoietic neoplasms
in male C57BL mice exposed to benzene, but there was no indication of renal
neoplasia from exposure to benzene in this study.
5.2.7 Conclusions of the UAREP Report (1983) on Toxicological
Interpretation of Hydrocarbon-Induced Kidney Lesions
An analysis of the toxicology and carcinogenicity of unleaded gasoline
and other hydrocarbons, issued by the Universities Associated for Research
and Education in Pathology, Inc. (UAREP) was published in December, 1983.
This section summarizes the review and interpretation of the data presented
in that document.
5.2.7.1 Asessment of the API Chronic Inhalation Study with
Unleaded Gasoline Vapor in Rats and Mice
1. There were significant increases in renal adenoma and carcinoma inci-
dence in male Fischer 344 rats and in hepatocellular adenoma and carci-
noma incidence in female B6C3F1 mice exposed to unleaded gasoline vapor.
The bioassay was well designed and conducted, and there were several
independent examinations of the kidney slides.
5-33
-------
2. Male Rat Kidney Lesions in Exposed Groups
a. Three-month findings: Focal degeneration in the proximal tubules,
hyalin droplets in the proximal tubules, granular casts at the
junction between the inner and outer stripes of the outer medulla,
some evidence of regeneration. The lesions seemed dose-related,
with the greatest prominence in the high-dose group.
b. Findings by 12 months: Karyomegaly, probably in the ?3 segment
of the proximal tubules. Old-rat nephropathy, in treated as
well as control rats, shown as atrophy of the PI segment of the
proximal tubule with basement membrane thickening, mesangial
thickening in the glomeruli, interstitial fibrosis with chronic
inflammation, and periodic acid-Schiff (PAS)-positive tubular
colloid casts in the distal nephron. Calcium hydroxyapatite
deposition in the papilla was evident in the exposed groups.
c. Findings after 12 months: Areas of hyperplasia. Progression of
the severity of old rat nephropathy, which was greater in the
treated groups than in controls, as well as preneoplastic and
neoplastic lesions.
d. Findings at 18 months: Renal adenoma in one rat.
e. Findings at 24 months (final sacrifice): Renal adenomas and
carcinomas in treated rats.
3. Liver Pathology in Female Mice
No treatment-related lesions were found between 3 and 18 months.
It could not be determined whether preneoplastic lesions preceded the
liver tumors. Acute effects, such as fatty metamorphosis from
exposure of mice to other hydrocarbons, were neither reported nor
looked for in the carcinogenic!"ty study with unleaded gasoline.
5-34
-------
5.2.7.2 Interpretation of the Toxicological Carcinogenic
findings In the Carcinogem'city Study with Unleaded
Gasoline by UAREP
1. Male Rat Kidney - Nonneoplastic Lesions
Signs of acute and chronic renal toxicity were evident. There
was some necrosis, but more often there was cell degeneration and/or
blebbing with release of cell debris forming casts between the pars
recta (?3 segment) and the thin limb. The kidney lesions in exposed
male rats were unique in that they were unlike those induced by known
nephrotoxins such as mercuric chloride, halogenated hydrocarbons, or
nitrilotriacetic acid.
The mechanism of gasoline-induced nephrotoxicity is obscure. There
was no uniform necrosis in the P$ segment, and many lesions were found
in the PI and ?2 segments. It is currently not possible to accurately
characterize the nature and location of the toxic lesions in male rat
kidneys in this study.
The toxic kidney lesions were clearly distinguishable from old-rat
nephropathy, which in the latter involved the whole kidney and showed
atrophy of the PI segment of the proximal tubule and glomerular
sclerosis. However, exposure to unleaded gasoline vapor augmented
the severity of old-rat nephropathy.
The most striking chronic nonneoplastic lesion in exposed rats was
severe mineralization of the tubules in the papilla. The deposits were
characterized as calcium hydroxyapatite. The etiology behind the mine-
ralization is uncertain, but it was hypothesized that chronic damage in
the proximal tubule and higher segments in the nephron leads to
phospholipid vesicle-induced calcification. The observed calcification
5-35
-------
pattern is unique to hydrocarbon exposures, and chronic exposure of
male rats to JP-10 synthetic missile fuel has also induced minerali-
zation in the kidney.
2. Male Rat Kidney - Preneoplastic and Neoplastic Lesions
The neoplastic process resulting from treatment of male rats
with unleaded gasoline resembles that induced by several other renal
carcinogens: karyomegaly, probably in the P$ segment, followed by
hyperplasia, followed by adenomas that were often cystic, and carci-
nomas. These preneoplastic and neoplastic lesions were found in
association with the increased severity of old-rat nephropathy, but
it was not possible to characterize their precise location, nature,
or progression. It was not possible to establish the influence of
old-rat nephropathy on these lesions, and the mechanism for the
induction of these preneoplastic and neoplastic lesions is unknown.
3. Liver Lesions in Female Mice
Acute toxicity or preneoplastic lesions were not found in the livers
of female mice, according to the carcinogenicity study report; however,
hematoxylin and eos-in staining is not sensitive for the detection of
preneoplastic lesions. Acute toxicity studies with other hydrocarbons
in mice have revealed fatty metamorphosis in the liver. The neoplasms
in livers of female mice exposed to unleaded gasoline vapor resembled
mouse liver neoplasms found in other studies, but such neoplasms are
often phenotypically similar regardless of etiology.
5.2.7.3 Review of Human Kidney Lesions
1. Acute - There are no thorough studies on human kidney lesions from
acute exposure to hydrocarbons. Case reports indicating structural
5-36
-------
and functional changes in the kidney show the main lesions as variants
of an immume complex type of glomerular nephritis, a type of lesion
that has not been found in rodents. The mechanism of acute toxicity in
the case reports is difficult to pinpoint because of confounding factors.
2. Chronic - There is no persuasive evidence that human exposure to gasoline
is associated with renal cancer. There is no evidence for the calcification
of papilla or calculi in the bladder or kidney from human exposure to
gasoline. Human renal adenocarcinomas are morphologically similar to
those found in male rats chronically exposed to unleaded gasoline
vapor, as well as to well-characterized models of rodent renal neoplasia
induced by chemical carcinogens. However, renal adenocarcinomas
develop in human kidneys that are normal except for a putative increase
in hyperplasia and adenomas, whereas adenocarcinomas in the kidneys
of rats chronically exposed to unleaded gasoline vapor occurred with
a background of chronic, and often severe, renal disease. An apparent
increase in the incidence of adenomas and carcinomas in the kidneys
of dialysis patients is the only possible equivalent to adenocarcinoma
induction in rats. This possible similarity between humans and rats
needs to be further investigated, but it is consistent with the view
that any type of chronic renal injury, e.g., old-rat nephropathy, can
possibly act as a promoter and/or cocarcinogen in the induction of
renal neoplasia.
Human renal cancer can occur along with chronic interstitial
nephritis, e.g., chronic analgesic nephropathy. This type of cancer
arises in the renal pelvic epithelium to yield transitional cell
carcinomas totally different in location and structure from the
lesions seen in the chronic rat study with unleaded gasoline.
5-37
-------
5.2.7.4 Species and Sex Comparison of the Kidney
Renal morphology has been most thoroughly studied in the rat, rabbit, and
dog. There has been no detailed ultrastructural study in the mouse kidney. A
detailed ultrastructural study has been done with renal biopsies from 10 human
males who were screened for renal dysfunction.
The human kidney is multilobular, without the distinct zonation caused
by the alignment of nephrons in the unilobar rodent kidney. The rodent kidney
has a long loop of Henle and long papillae to allow extensive concentration of
urine. The human kidney has nothing like the outer stripe of the outer medulla
in the rodent kidney, which contains the pars recta (?3 segment) of the proximal
tubule and the ascending limb of the loop of Henle. The human kidney has an
ultrastructurally simple proximal tubule in contrast to the kidney of the rat,
mouse, and rabbit. No morphological differences between the kidneys of male
and female humans have been described.
The size and number of lysosomes in the PI, ?2, and ?3 segments in the
male rat kidney are larger than in the female rat kidney. This may be correlated
with the unique production of alpha-2-microglobulin and resorption of this
protein in the proximal tubule in the male rat kidney. Endoplasmic reticulum
and microbodies are more prominent in female than in male rat kidneys, which
may indicate a difference between them in metabolic capacity. Castration and
hypophysectomy of male rats decreases the differences in the proximal tubule,
particularly lysosomes, between male and female rat kidneys.
Hyalin droplets in the PI and P£ segments have been found in kidneys
of male rats exposed to hydrocarbons other than unleaded gasoline vapor,
e.g., decalin. These droplets consist mainly of protein, including alpha
5-38
-------
2-microglobulin, within the phagolysosomal system of the male rat kidney.
In the male rat kidney it is presumed that, with no evidence of acute
glomerular lesions, these droplets represent accumulations of endogenous
proteins that are produced in the male rat liver and resorbed by the kidney,
to be phagocytized by lysosomes. Protein accumulation by the kidney could be
due to increased synthesis and uptake and/or decreased degradation. The
mechanism of hydrocarbon nephrotoxicity is presently unclear. Exposure
to decalin produces hyalin droplets in male but not in female rats, and
these droplets have been observed to disappear quickly after cessation of
treatment with decalin. However, it is not known whether chronic lysosomal
overload can produce cell injury in rat kidney proximal tubules.
Less is known about renal mixed function oxidase (MFO) than about liver
MFO. There are marked species, strain, and sex differences in the metabolic
capability of the rodent kidney. There are potentially significant quantita-
tive and, to a lesser degree, qualitative differences in renal MFO components
and activity among species. In all species studied thus far, MFO activity
has been localized in the proximal tubule and usually in the ?2 and/or P£
segments. Little data exist on MFO in the human kidney. Metabolites from
other organs can possibly go to the kidney to produce toxicity in vivo.
Specific studies on renal MFO and unleaded gasoline toxicity are lacking, but
there is some evidence that renal MFO may play a role in the renal toxicity
of other hydrocarbons.
5.2.7.5 Rodent Kidneys and Other Hydrocarbons
The only hydrocarbon fuel other than unleaded gasoline that has been
tested in a chronic rodent bioassay is the synthetic missile fuel JP-10.
Exposure to JP-10 was found to induce renal carcinomas in male Fischer 344 rats.
5-39
-------
Most solvents and hydrocarbons have induced similar acutely toxic
4
lesions in the rodent kidney; however, no specific mechanism of action
has been determined, nor has an ultrastructural analysis been done.
Paraffin and isoparaffin fractions have been found to be more acutely
toxic in rodent kidneys than are other fractions of petroleum products.
The mutagenicity of unleaded gasoline has been reported as negative.
However, in the in vitro assays, only S-9 fractions from rat liver were
used, which may not relate to other organs: also, there may have been a
problem with the volatility and solubility of the gasoline in these assays.
None of the tested hydrocarbons has produced a uniform necrosis in
the pars recta epithelium, as is commonly seen in the rodent kidney with
other renal toxins, such as mercuric chloride.
Acute lesions in the rodent kidney from hydrocarbon exposure have
been characterized as hyalin droplet accumulations, focal areas of
degeneration and necrosis, epithelial regeneration, and granular casts in
the corticomedullary junction.
5.2.7.6 Old-Rat Nephropathy
Old-rat nephropathy is characterized by interstitial fibrosis, thickening
of tubular basement membranes, interstitial chronic inflammation, vascular
thickening in interlobular and afferent arterioles, glomerular hyalinization,
and tubular atrophy, especially in the PI segment of the proximal tubule.
In the unleaded gasoline carcinogenicity bioassay, increased numbers of
mitoses, hyperplasia, karyomegaly, and other preneoplastic lesions were not
seen in control rats.
Old-rat nephropathy may start at an early age. Old females show a
lesser degree of nephropathy than males. The severity of old-rat nephropathy
»
varies among strains.
5-40
-------
Old-rat nephropathy was morphologically different from preneoplastic
lesions in the API carcinogenic!ty study of unleaded gasoline vapor. However,
a possible etiologic relationship between old-rat nephropathy and toxic lesions
from exposure to unleaded gasoline vapor cannot be ruled out.
In humans, chronic renal disease has been associated with renal adeno-
carcinoma only in kidney dialysis patients. Lesions in the kidneys of these
patients are morphologically similar to those seen in male rats in the chronic
unleaded gasoline vapor study, as well as in other rodent studies with hydro-
carbons. No control human group has been studied along with the dialysis
patients.
Old-rat nephropathy resembles focal and segmental glomerulosclerosis
in human disease and, to a lesser extent, arteriolar and arterial nephro-
sclerosis in aging humans. Several patients with renal carcinoma following
dialysis had renal failure secondary to nephrosclerosis, and one of these
patients had multiple calculi in the kidney.
Studies with the liver indicate a greater ability of younger than of
older rats to metabolize carcinogens; however, a similar comparison with
the kidney still needs to be explored.
5.2.7.7 Comparative Nephrotoxicity and Nephrocarcinogenicity
Species, strain, and sex differences in response to nephrotoxins are
clearly evident. Many chemical classes of nephrotoxins induce similar morphological
effects, and most nephrotoxic chlorinated hydrocarbons affect primarily the
?3Vsegment of the proximal tubule, which is apparantly lacking in humans.
However, few agents have well-character!" zed mechanisms of acute and chronic
renal toxicity. Some chlorinated hydrocarbons, e.g., chloroform, have chronic,
carcinogenic effects that do not always correspond to acute effects in terms
of target organ response.
5-41
-------
Chloroform induces a selective and uniform degeneration and necrosis
of proximal tubule epithelial cells, with the effect being greatest in areas
with the greatest MFO activity, as supported by studies in rats and dogs.
Similar patterns of proximal tubule degeneration from exposure to chloroform
have been observed in these two species.
There is evidence that similarity of morphological endpoints does not
necessarily indicate similarity between mechanisms. For example, nitriloacetic
acid may act as a promoter in the induction of renal adenocarcinomas in rats,
but the renal adenocarcinomas induced by this agent are morphologically similar
to those observed in rats in the chronic unleaded gasoline vapor study. Acute,
but not chronic, renal lesions from mercuric chloride treatment are unlike
those induced by exposure to hydrocarbons. Mercuric chloride, as well as
chlorinated hydrocarbons, initially induces selected necrosis in the renal ?3
segment, and with higher doses also induces progressive necrosis in the PI and
?2 segments.
5.2.7.8 Significance to Humans of the Chronic Inhalation
Study with Unleaded Gasoline Vapor in Rats and Mice
1. The relationship between old-rat nephropathy and renal neoplasia in
the chronic unleaded gasoline study is presently uncertain. Although
renal neoplasia in male rats exposed to unleaded gasoline does not
appear to stem from basophilic cells in the old-age renal lesions, a
role of the old-age lesions in the etiology of renal neoplasia pre-
sently cannot be ruled out. In humans, chronic renal disease has been
associated with an increased incidence of renal cancer.
2. No statistically significant association between renal epithelial neo-
plasia and environmental agents, except for cigarette smoke, has been
found in humans. Tumors of the renal pelvis in humans have been
*
associated with exposure to environmental agents.
5-42
-------
3. Anatomical and physiological differences between rat and human kidneys
may contribute to differences in renal responses to environmental agents,
including unleaded gasoline. This issue needs further study.
5.2.8 Summary of Animal Studies
A lifetime inhalation bioassay of unleaded gasoline in Fischer 344 rats and
B6C3F1 mice induced a statistically significant incidence (6/100) of renal
carcinomas in the kidney cortex of male rats and a larger, also statistically
significant incidence (20/100) of hepatocellular carcinomas in female mice.
Female rats and male mice had no significant treatment-related induction of
tumors at any organ site. The incidence of renal tumors was statistically
significant at the highest dose tested (2,056 ppm) but not at the two lower
doses (292 ppm and 67 ppm). In mice the incidence of liver carcinomas
alone and adenoma and carcinoma combined was also statistically significant in
the highest but not the two lower dose groups. Moderate decrements in body-
weight gain in the high-dose groups indicate that the maximum tolerated dose
was reached. Glomerulonephrosis occurred in nearly all male rats, and
mineralization of the pelvis was correlated with dose. However, there was no
correlation between animals with tumors and those with mineralization.
The acute and subchrooic renal toxicity of decalin, a volatile hydrocarbon
of the same general type as those contained in gasoline, is confined to male
rats and does not occur in female rats or in mice, dogs, or guinea pigs. In
a series of 21-day inhalation exposures of male rats to a variety of chemical
fractions of gasoline, renal toxicity was correlated with the paraffin components
and not with the aromatic compounds in the mixture. The same pattern of
renal toxicity as well as a positive renal tumor response occurs in response
to chronic inhalation of two synthetic fuels (RJ-5 and JP-10). Chronic
inhaltion studies with the jet fuels used by the Air Force and Navy (JP-4 and
5-43
-------
JP-5) have shown the same nephrotoxic lesions, but no statements can be made
about the carcinoma response until histopathological analysis has been completed.
The renal toxicity pattern observed with exposure to hydrocarbon mixtures,
involving protein accumulation in renal tubules, is clearly different than
the kidney lesions occurring spontaneously in old rats, and occurs in males
of both Fischer 344 and Spraque-Dawley strains, but not in females of these
strains or in mice or monkeys.
Mutagenesis tests of unleaded gasoline have been carried out in Salmonella,
yeast, mouse lymphoma in vivo cytogenetics, and mouse dominant lethal systems.
Various gasoline feedstocks have been tested in mouse lymphoma and i^n vivo
cytogenetics assays. The results of most of these assays have not met the
criteria for positive responses. A detailed examination of their adequacy is
in process.
5.3 Epidemiologic Studies of Petroleum Workers
Animal studies involving mice and rats have indicated that unleaded gas-
oline exposure may increase the risk of cancer, especially kidney and liver
cancers, in humans. The purpose of this section is to review the epidemio-
logic literature in order to determine whether there is any epidemiologic
evidence suggested by the animal findings. Three epidemiologic studies have
been reviewed: two published [Thomas et al. (1980, 1982)] and one unpublished
[Rushton and Alderson (1982)].
5.3.1 Thomas et al. (1980)
Death records of individuals, who at the time of death were active members
of the Oil, Chemical, and Atomic Workers International Union (OCAW), were
reviewed for specific causes of death by the Environmental Epidemiology Branch
of the National Cancer Institute (Thomas et al. 1980). The study group
5-44
-------
consisted of 3,105 males whose deaths were reported to the OCAW Internatioal
Headquarters by Union locals in Texas between 1947 and 1977 and for whom
death certificates were obtained. Death certificates were not available for
10% of the reported deaths. Approximately 40% of the decedents were less
than 50 years of age at death. Also, 40% of the decedents were union members
for less than 10 years.
The individual plants in which members had worked were classified into
one of five major categories according to major processes. The most interes-
ting and the one for which approximately 70% of the deaths were classified
was the petroleum refinery and petrochemical plant category. In the discus-
sion presented here, the results will be restricted to this category.
Proportionate mortality ratios (PMRs), adjusted for age and calendar
time using the United States general population, were computed and tested for
statistical significance.
The PMR for all cancer deaths (1.26) was significantly elevated for
whites (P < 0.01) but not for blacks (P > 0.05). Also, the relative frequency
for arteriosclerotic heart disease deaths was elevated significantly for both
racial groups (P < 0.01). However, the relative frequencies for respiratory
and digestive disease deaths were quite a bit lower than expected for whites
and blacks. Both races had significantly elevated (P < 0.01) PMRs for
non-motor vehicle accidents, whereas the PMR for motor vehicle accidents was
significantly greater for whites only.
With regard to the relative frequencies of cause-specific cancer deaths,
greater than expected frequencies (P < 0.05) were observed for cancers of
the digestive organs and peritoneum, respiratory system, and skin for whites.
The PMRs for cancer of the stomach (2.69) and kidney (2.14) were signi-
ficantly elevated (P < 0.05) only for white males who joined the union 20 or
5-45
-------
more years prior to death. Black males whose lengths of union membership were
either less than 10 years or 10 or more years experienced significantly
greater than expected frequencies (PMRs of 2.42 and 2.80, respectively) of
stomach cancer deaths.
These results provide very weak evidence for the carcinogem'city of
gasoline vapors because of the following study limitations. The underlying
cause of death was unable to be determined for 10% or approximately 350 of
the total number of deaths. This underreporting of the causes of death could
heavily influence the cause-specific mortality frequencies, especially if the
underreporting were occurring for a small number of causes of death.
A serious problem is inherent in the usefulness of PMRs. If the study
group has a lower mortality rate than the comparison group for all causes of
death, PRMs represent inflated estimates of cause-specific risks. Furthermore,
excesses for one or more causes will automatically force others to be in
deficit.
This study has the obvious limitation of resricting its investigation
to active members of the union, thereby excluding union members who retired
or left the union for other, reasons as well as excluding non-union members.
The results of this study, therefore, may overrepresent diseases with very
low survival rates and underrepresent diseases which tend to occur in
retirees.
No exposure information regarding mesured levels of gasoline vapor is
given. Furthermore, the question has to be raised as to whether this study
actually investigates the risk of leaded gasoline exposure or health in
contrast to unleaded gasoline exposure. This study examined the mortality
experience of active OCAW members between 1947 and 1977. The advent .of
unleaded gasoline use did not take place until the late 1970s.
5-46
-------
Along with the problem of determining the exact agent to which the study
population was exposed, there is a question with regard to the adequacy of the
latency period. A surrogate measure of latency is length of union membership.
Approximately 40% and 25% of the decedents were OCAW members for less than 10
years and more than 20 years, respectively. Thus, for a significantly large
proportion of the study population, the period between first exposure and
death was less than the generally accepted average latency period of 10 to
30 years for environmentally induced cancers.
5.3.2 Thomas et a!. (1982)
For three of the refineries included in their earlier study, Thomas et
al. (1982) examined the cause-specific mortality experience of an expanded
group of union members. The number of male deaths from the original study
of these three refineries was 1,161. This number was expanded to include
1,194 retiree deaths and 154 additional active union member deaths. In the
earlier study, reported deaths were for the period between 1947 and 1977.
The period of observation for the present study was extended through 1979.
Thus, 2,509 active and retired members of the OCAW were available for analysis
in the present study. As with the earlier study, death certificates were
unable to be located for 8% of the reported deaths. However, no information
was given regarding length of union membership.
The PMR for stomach cancer was significantly elevated (P < 0.05) for
whites (1.41) and nonwhites (1.96). The relative frequencies of deaths
attributable to cancer of the pancreas (1.42), prostate (1.46), brain (2.28),
and hematopoietic and lymphatic system (1.72), [including leukemia (1.89) for
whites only] were significantly greater than expected (P < 0.05). Although
the PMR (1.51) for kidney cancer was elevated for white males, it was not
*
statistically significant (P > 0.05).
5-47
-------
There were differences in mortality patterns for white males among active
and retired union members. Significant relative excesses {P < 0.05) of stomach,
pancreatic, and brain cancer deaths were seen among active members.
However, among retired members, the PMRs were significantly elevated for
prostrate cancer, Hodgkin's disease, multiple myeloma, and leukemia deaths.
The limitations of this study are similar to those already noted in
relation to the earlier study. There are serious concerns regarding the loss
of individuals due to the unavailability of death certificates, lack of expo-
sure information, and the inherent validity of PMRs.
5.3.3 Rushton and Alderson (1982)
Rushton and Alderson (1982), in an unpublished report, presented the results
of a retrospective cohort mortality study of workers at distribution centers
from three oil companies in Great Britain. This study was funded by 23 oil
companies in Great Britain, and was coordinated by the Institute of Petroleum.
The study population consisted of men employed for at least one year between
January 1, 1950 and December 31, 1975. The comparison population used was
the entire male population of England and Wales.
A total of 762 distribution centers contributed 23,358 men to the study
population. Ninety-nine percent of the population was followed sucessfully
to determine their vital status as of December 31, 1975. The study population
accounted for 397,568.60 person-years, with an average follow-up period of
17.1 years.
The number of deaths of the study population was decidedly lower than
that of the comparison population both from all causes (3,925 observed, 4,632
expected) and from all neoplasms (1,002 observed, 1,157 expected).
These deficits may in part reflect the "healthy worker" effect. However,
*
consideration must also be given for the criteria of inclusion of participants
5-48
-------
into the study. Although a feasibility study suggested that a minimum of 10
years of employment should be required for admittance into the study population,
it was decided to reduce this requirement to one year in order to increase the
number of people in the study. Although this action did, in fact, increase the
study population by one and one-half, it undoubtedly contributed to an over-
estimation of the expected numbers of deaths, causing the observed number
of deaths to be in deficit.
With regard to cancer of the kidneys and suprarenals, there were slightly
more deaths than expected (23 observed, 19.05 expected). However, among
drivers the 12 deaths observed were significantly larger (P < 0.05) than the
7.03 deaths to be expected. It also should be noted that all but two of the
drivers had started work before 1940 and had over 20 years of service.
It is indeed unfortunate, given the size and scope of this epidemio-
logic survey, that this report prepared by Rushton and Alderson is best
characterized as superficial, anecdotal, and generally incomplete. For
example, years of employment are parenthetically discussed in the authors'
explanation for the selection criteria of workers into the study. No accom-
panying tables are presented. It can only be inferred that 36% of the study
population had under 10 years of employment. Virtually no other information
regarding years of employment and, hence, latency period can be gleaned from
this document, with the exception of an occasional reference in the discussion
of the mortality from a few diseases.
Pertinent information regarding measured levels of gasoline vapor at the
762 distribution centers is missing. Also missing is exposure data for the
various occupational categories at these centers. Specifically, to what
extent and amount were the drivers subject to gasoline vapors? Did the drivers
of gasoline tank trucks assist in the unloading of gasoline into tanks?'
5-49
-------
Furthermore, what constitutes a distribution center? The background material
is at best incomplete in this report.
5.3.4 Summary of Epidemiologic Studies
Three epidemiologic studies of workers exposed to petroleum products
including gasoline and cancer have been reviewed: two published [Thomas et
al. (1980, 1982)] and one unpublished [Rushton and Alderson (1982)].
Thomas et al . (1980, 1982) reported on two studies in which the death
records of men who worked in oil refineries and petrochemical plants were
reviewed for specific causes of death. The 1980 paper examined the records
of workers who at the time of death were active members of the Oil Chemical
and Atomic Workers International Union (OCAW). The 1982 paper was expanded
to include retired members of the OCAW at the time of death. Proportionate
Mortality Ratios (PMRs) adjusted for age and calendar time using the United
States general population were computed.
The 1980 study showed that the PMRs for kidney and stomach cancers were
significantly increased (P < 0.05) for white males who joined the union 20 or
more years prior to death. However, black males whose lengths of union mem-
bership were less than 10 years, as well as those whose union membership were
10 or more years, experienced significantly greater than expected frequencies
of stomach cancer deaths. As indicated in the 1982 paper, there were dif-
ferences in mortality patterns for white males among both active and retired
union members. Significant relative excesses (P < 0.05) of stomach, pancreatic,
and brain cancer deaths were seen among active members. However, among
retired members, the PMRs were significantly elevated for prostate cancer,
Hodgkin's disease, multiple myeloma, and leukemia deaths.
Rushton and Alderson (1982) presented the results of a retrospective
»
cohort mortality study of male workers at distribution centers from three
5-50
-------
oil companies in Great Britain. The notable result was that among drivers
there were 12 kidney cancer deaths which were significantly in excess
(P < 0.05) compared to the 7.03 expected number of deaths.
All three studies suffer from insufficient documentation of exposure and
employment histories and questionable applicability for determining the carcino-
genicity of unleaded gasoline. The studies by Thomas et al. (1980, 1982)
present inadequate definitions of the study populations and methodologies.
Moreover, the limitations inherent in proportionate mortality ratios (PMRs)
are in themselves sufficient to cast doubt on the results of these studies.
PMRs reflect inflated estimates of mortality if the study group has a lower
mortality rate than the comparison group for all causes of death. Also,
excesses for one or more causes may automatically lead to a deficit in others.
Because of its incomplete nature, the study by Rushton and Alderson (1982) is
judged to be inadequate.
5.4 Quantitative Risk Estimation
This quantitative section deals with the estimation of cancer risk due
to exposure to unleaded gasoline vapor. The unit risk is defined here as the
lifetime incremental cancer risk from exposure to 1 ppm of gasoline vapor in
air. Uncertainties about the risk estimate and the possible role of benzene
content in gasoline vapor are also addressed in this section.
The risk estimate for gasoline vapor represents an extrapolation below
the dose range of experimental data. There is currently no solid scientific
basis for any mathematical extrapolation model that relates exposure to cancer
risk at the extremely low concentrations, including the unit concentration
given above, that must be dealt with in evaluating environmental hazards. For
practical reasons the correspondingly low levels of risk cannot be measured
5-51
-------
directly either by animal experiments or by epidemiologic studies. Low-dose
extrapolation must, therefore, be based on current understanding of the mechanisms
of carcinogenesis. At the present time the dominant view of the carcinogenic
process involves the concept that most cancer-causing agents also cause irrever-
sible damage to DMA. This position is based in part on the fact that a very
large proportion of agents that cause cancer are also mutagenic. There is
reason to expect that the quanta! response that is characteristic of mutagenesis
is associated with a linear non-threshold dose-response relationship. Indeed,
there is substantial evidence from mutagenicity studies with both ionizing
radiation and a wide variety of chemicals that this type of dose-response model
is the appropriate one to use. This is particularly true at the lower end of
the dose-response curve; at high doses, there can be an upward curvature,
probably reflecting the effects of multistage processes on the mutagenic response.
The linear non-threshold dose-response relationship is also consistent with the
relatively few epidemiologic studies of cancer responses to specific agents
that contain enough information to make the evaluation possible (e.g., radiation-
induced leukemia, breast and thyroid cancer, skin cancer induced by arsenic in
drinking water, liver cancer induced by aflatoxins in the diet). Some supporting
evidence also exists from animal experiments (e.g., the initiation stage of the
two-stage carcinogenesis model in rat liver and mouse skin).
Because its scientific basis, although limited, is the best of any of the
current mathematical extrapolation models, the non-threshold model which is
linear at low doses, has been adopted by CAG as the primary basis for risk
extrapolation to low levels of the dose-response relationship. The risk estimates
made with such a model should be regarded as conservative, representing the
most plausible upper limit for the risk (i.e., the true risk is not likely to
*
be higher than the estimate, but it could be lower).
5-52
-------
For several reasons, the unit risk estimate based on animal bioassays is
only an approximate indication of the absolute risk in populations exposed to
known carcinogen concentrations. First, there are important species differences
in uptake, metabolism, and organ distribution of carcinogens, as well as species
differences in target site susceptibility, immunological responses, hormone
function, dietary factors, and disease. Second, the concept of equivalent
doses for humans compared to animals on a mg/surface area basis is virtually
without experimental verification as regards carcinogenic response. Finally,
human populations are variable with respect to genetic constitution and diet,
living environment, activity patterns, and other cultural factors.
The unit risk estimate can give a rough indication of the relative potency
of a given agent as compared with other carcinogens. Such estimates are, of
course, more reliable when the comparisons are based on studies in which the
test species, strain, sex, and routes of exposure are similar.
The quantitative aspect of carcinogen risk assessment is addressed here
because of its possible value in the regulatory decision-making process, e.g.,
in setting regulatory priorities, evaluating the adequacy of technology-based
controls, etc. However, the imprecision of presently available technology for
estimating cancer risks to humans at low levels of exposure should be recognized.
At best, the linear extrapolation model used here provides a rough but plausible
estimate of the upper limit of risk from exposure to a unit concentration of
gasoline vapor (i. e., with this model it is not likely that the true risk
would be much more than the estimated risk, but it could be considerably lower).
The risk estimates in this paper relate only to exposure to gasoline
vapor. Risks related to the entire range of compounds that may be present in
air are not estimated here.
5-53
-------
5.4.1 Procedures for the Determination of Unit Risk
5.4.1.1 Low-Dose Extrapolation Model
The mathematical formulation chosen to describe the linear nonthreshold
dose-response relationship at low doses is the linearized multistage model.
This model employs enough arbitrary constants to be able to fit almost any
monotonically increasing dose-response data, and it incorporates a procedure
for estimating the largest possible linear slope (in the 95% confidence limit
sense) at low extrapolated doses that is consistent with the data at all dose
levels of the experiment.
Let P(d) represent the lifetime risk (probability) of cancer at dose d.
The multistage model has the form:
P(d) = 1 - exp L-(q0 + q^ + q-^2 + ...+ qkdk)]
where
qi >_ 0, i = 0, 1, 2, .... k
Equivalently,
Pt(d) = 1 - exp [-(q^ + q2d2 + ... + qkdk)]
where
P (d) = P(d) - P(0)
t 1 - P(0)
is the extra risk over background rate in the animal control group at dose d.
The point estimate of the coefficients qi, i = 0, 1, 2, ..., k, and
consequently, the extra risk function, Pt(d), at any given dose d, is
calculated by maximizing the likelihood function of the data.
5-54
-------
The maximum likelihood estimate and the 95% upper confidence limit of the
extra risk, Pt(d), are calculated by using the computer program, GLOBAL79,
developed by Crump and Watson (1979). At low doses, upper 95% confidence
limits on the extra risk and lower 95% confidence limits on the dose producing
a given risk are determined from a 95% upper confidence limit, q^, on parameter
qj_. Whenever qi > 0, at low doses the extra risk PtU) has approximately the
form Pt(d) = q£ x d. Therefore, q^ x d is a 95% upper confidence limit on the
extra risk and R/q^ is a 95% lower confidence limit on the dose, producing an
extra risk of R. Let LQ be the maximum value of the log-likelihood function.
The upper-limit q* is calculated by increasing q^ to a value q* such that when
the log-likelihood is remaximized subject to this fixed value q* for the linear
coefficient, the resulting maximum value of the log-likelihood LI satisfies the
equation:
2 (Lo - LI) = 2.70554
where 2.70554 is the cumulative 90% point of the chi-square distribution with
one degree of freedom, which corresponds to a 95% upper-limit (one-sided). This
approach of computing the upper confidence limit for the extra risk ?t(d) is an
improvement on the Crump et al. (1977) model. The upper confidence limit for
the extra risk calculated at low doses is always linear. This is conceptually
consistent with the linear nonthreshold concept discussed earlier. The slope,
q*, is taken as an upper-bound of the potency of the chemical in inducing
cancer at low doses.
In fitting the dose-response model, the number of terms in the polynomial
is chosen equal to (h-1), where h is the number of dose groups in the experiment,
including the control group.
5-55
-------
Whenever the multistage model does not fit the data sufficiently well, data
at the highest dose is deleted and the model is refit to the rest of the data.
This is continued until an acceptable fit to the data is obtained. To determine
whether or not a fit is acceptable, the chi -square statistic
2 m
is calculated where N^ is the number of animals in the ith dose group, X^ is
the number of animals in the i™ dose group with a tumor response, P.,- is the
probability of a response in the itn dose group estimated by fitting the
multistage model to the data, and h is the number of remaining groups. The
fit is determined to be unacceptable whenever X2 is larger than the cumulative
99% point of the chi-square distribution with f degrees of freedom, where f
equals the number of dose groups minus the number of non-zero multistage co-
efficients.
5.4.1.2 Selection of Data
For some chemicals, several studies in different animal species, strains,
and sexes, each run at several doses and different routes of exposure, are
available. A choice must be made as to which of the data sets from several
studies to use in the model. It may also be appropriate to correct for metabolism
differences between species and for absorption factors via different routes of
administration. The procedures used in evaluating these data are consistent
with the approach of making a maximum-likely risk estimate. They are as follows:
1. The tumor incidence data are separated according to organ sites or
tumor types. The set of data (i.e., dose and tumor incidence) used in the
model is the set where the incidence is statistically significantly higher
5-56
-------
than the control for at least one test dose level and/or where the tumor
incidence rate shows a statistically significant trend with respect to dose
level. The data set that gives the highest estimate of the lifetime car-
cinogenic risk, q*, is selected in most cases. However, efforts are made to
exclude data sets that produce spuriously high risk estimates because of a
small number of animals. That is, if two sets of data show a similar dose-
response relationship, and one has a very small sample size, the set of data
having the larger sample size is selected for calculating the carcinogenic
potency.
2. If there are two or more data sets of comparable size that are
identical with respect to species, strain, sex, and tumor sites, the geometric
mean of q*, estimated from each of these data sets, is used for risk assessment.
The geometric mean of numbers AI, A2, ..., Am is defined as
(A x A x ... x
3. If two or more significant tumor sites are observed in the same study,
and if the data are available, the number of animals with at least one of the
specific tumor sites under consideration is used as incidence data in the model
5.4.1.3 Calculation of Human Equivalent Dosages
Following the suggestion of Mantel and Schneiderman (1975), it is assumed
that mg/surface area/day is an equivalent dose between species. Since, to a
close approximation, the surface area is proportional to the two-thirds power
of the weight, as would be the case for a perfect sphere, the exposure in
mg/day per two-thirds power of the weight is also considered to be equivalent
exposure. In an animal experiment, this equivalent dose is computed in the
following manner:
5-57
-------
Let
Le = duration of experiment
le = duration of exposure
m = average dose per day in mg during administration of the agent (i.e.,
during le), and
W = average weight of the experimental animal
Then, the lifetime exposure is:
2/3
Le x W
5.4.1.3.1 Oral.
Often exposures are not given in units of mg/day, and it becomes necessary
to convert the given exposures into mg/day. Similarly, in drinking water
studies, exposure is expressed as ppm in the water. For example, in most
feeding studies exposure is given in terms of ppm in the diet. In these cases,
the exposure in mg/day is:
m = ppm x F x r
where ppm is parts per million of the carcinogenic agent in the diet or water,
F is the weight of the food or water consumed per day in kg, and r is the
absorption fraction. In the absence of any data to the contrary, r is assumed
to be equal to one. For a uniform diet, the weight of the food consumed is
proportional to the calories required, which in turn is proportional to the
surface area, or two- thirds power of the weight. Water demands are also
assumed to be proportional to the surface area, so that
m « ppm x W x r
5-58
-------
or
rW
As a result, ppm in the diet or water is often assumed to be an equivalent
exposure between species. However, this is not justified for the present
study, since the ratio of calories to food weight is very different in the
diet of man as compared to laboratory animals, primarily due to differences
in the moisture content of the foods eaten. For the same reason, the amount
of drinking water required by each species also differs. It is therefore
necessary to use an empirically-derived factor, f = F/W, which is the
fraction of an organism's body weight that is consumed per day as food,
expressed as follows:
Fraction of body
weight consumed as
Species W ffood ^water
Man 70 0.028 0.029
Rats 0.35 0.05 0.078
Mice 0.03 0.13 0.17
Thus, when the exposure is given as a certain dietary or water concentration in
ppm, the exposure in mg/W2/3 is
m = ppm x F = ppm x f x W = ppm x f x W1/3
rWZ/3 ~
When exposure is given in terms of mg/kg/day = m/Wr = s, the conversion is
simply
= s x W1/3.
5-59
-------
5.4.1.3.2 Inhalation.
When exposure is via inhalation, the calculation of dose can be considered
for two cases where 1) the carcinogenic agent is either a completely water-soluble
gas or an aerosol and is absorbed proportionally to the amount of air breathed
in, and 2) where the carcinogen is a poorly water-soluble gas which reaches an
equilibrium between the air breathed and the body compartments. After equilibrium
is reached, the rate of absorption of these agents is expected to be proportional
to the metabolic rate, which in turn is proportional to the rate of oxygen
consumption, which in turn is a function of surface area.
5.4.1.3.2.1 Case 1
Agents that are in the form of particulate matter or virtually completely
absorbed gases, such as sulfur dioxide, can reasonably be expected to be absorbed
proportionally to the breathing rate. In this case the exposure in mg/day may
be expressed as:
m = I x v x r
where I = inhalation rate per day in m3, v = mg/m3 of the agent in air, and
r = the absorption fraction.
The inhalation rates, I, for various species can be calculated from the
observations of the Federation of American Societies for Experimental Biology
{FASEB 1974) that 25 g mice breathe 34.5 liters/day and 113 g rats breathe 105
liters/day. For mice and rats of other weights, W (in kilograms), the surface
area proportionality can be used to find breathing rates in m3/day as follows:
For mice, I = 0.0345 (W/0.025)2/3 m3/day
For rats, I = 0.105 (W/0.113)2/3 m3/day
For humans, the value of 30 m3/day* is adopted as a standard breathing rate
(International Commission on Radiological Protection 1977). The equivalent
5-60
-------
exposure in mg/W2/3 for these agents can be derived from the air intake data
in a way analogous to the food intake data. The empirical factors for the air
intake per kg per day, i = I/W, based upon the previously stated relationships,
are tabulated as follows:
Species W i = I/W
Man
Rats
Mice
70
0.35
0.03
0.29
0.64
1.3
Therefore, for particulates or completely absorbed gases, the equivalent
exposure in mg/W2/3 is
d = ro = Ivr = iWvr = -jwl/3vr
^273
In the absence of experimental information or a sound theoretical argument
to the contrary, the fraction absorbed, r, is assumed to be the same for all
species.
5.4.1.3.2.2 Case 2
The dose in mg/day of partially soluble vapors is proportional to the 02
consumption, which in turn is proportional to W2/3 and is also proportional to
the solubility of the gas in body fluids, which can be expressed as an absorption
coefficient, r, for the gas. Therefore, expressing the 02 consumption as 02 =
k W2/3, where k is a constant independent of species, it follows that:
m = k W2/3 x v x r
or
d = rc = kvr
w2/3
Trom "Recommendation of the International Commission on Radiological
Protection," page 9. The average breathing rate is 10? cm3 per 8-hoor workday
and 2 x 107 cm3 in 24 hours.
5-61
-------
As with Case 1, in the absence of experimental information or a sound theoretical
argument to the contrary, the absorption fraction, r, is assumed to be the same
for all species. Therefore, for these substances a certain concentration in
ppm or ug/m3 in experimental animals is equivalent to the same concentration
in humans. This is supported by the observation that the minimum alveolar
concentration necessary to produce a given "stage" of anesthesia is similar
in man and animals (Dripps et al. 1977). When the animals are exposed via
the oral route and human exposure is via inhalation or vice versa, the
assumption is made, unless there is pharmacokinetic evidence to the contrary,
that absorption is equal by either exposure route.
5.4.1.4 Calculation of the Unit Risk from Animal Studies
The risk associated with d mg/kg2/3/day is obtained from GLOBAL79 and, for
most cases of interest to risk assessment, can be adequately approximated by
P(d) = 1 - exp (-q*d). A "unit risk" in units X is simply the risk corresponding
to an exposure of X = 1. This value is estimated simply by finding the number
of mg/kg2/3/day that corresponds to one unit of X, and substituting this value
into the above relationship. Thus, for example, if X is in units of ug/m3 in
the air, then for case 1, d = 0.29 x 701-/3 x 10~3 mg/kg2/3/day, and for case 2,
d = 1, when ug/m3 is the unit used to compute parameters in animal experiments.
If exposures are given in terms of ppm in air, the following calculation
may be used:
1 ppm = 1.2 x molecular weight (gas) mg/m3
molecular weight (air)
Note that an equivalent method of calculating unit risk would be to use mg/kg
for the animal exposures, and then to increase the jth polynomial coefficient
by an amount:
5-62
-------
(Wh/Vj/3 J = i, 2, ..., k,
and to use mg/kg equivalents for the unit risk values.
5.4.1.4.1 Adjustments for Less Than Lifespan Duration
of Experiment
If the duration of experiment Le is less than the natural lifespan of the
test animal L, the slope qf, or more generally the exponent g(d), is increased
by multiplying a factor (l_/l_e)3. We assume that if the average dose d is
continued, the age specific rate of cancer will continue to increase as a constant
function of the background rate. The age-specific rates for humans increase at
least by the third power of the age and often by a considerably higher power,
as demonstrated by Doll (1971). Thus, it is expected that the cumulative tumor
rate would increase by at least the third power of age. Using this fact, it is
assumed that the slope qf, or more generally the exponent g(d), would also
increase by at least the third power of age. As a result, if the slope q^
[or g(d)] is calculated at age Le, it is expected that if the experiment had
been continued for the full lifespan L at the given average exposure, the slope
qf [or g(d)] would have been increased by at least (L/Le)3.
This adjustment is conceptually consistent with the proportional hazard
model proposed by Cox (1972) and the time-to-tumor model considered by Daffer et
al. (1980), where the probabiity of cancer by age t and at dose d is given by
P(d,t) = 1 - exp[-f(t) x g(d)].
5.4.2 Lifetime Risk Estimates
5.4.2.1 Data Available for Risk Estimation
The chronic inhalation study of unleaded gasoline vapor conducted by the
International Research and Development Corporation (IRDC 1983) and sponsored by
the American Petroleum Institute (API) is the only study that can be used to
derive the carcinogenic potency of unleaded gasoline vapor. Tables 5-23 and
5-63
-------
5-24 present dose-response data used in these calculations. The data in Table
5-23 were taken from Tables 21 and 22 of Volume 5 of the API report. The data
in Table 5-24 were taken from Tables 23 and 24 of Volume 6 of the API report.
All of the tumors reported in Tables 5-23 and 5-24 were observed after 18
months of study. One kidney tumor that was observed in the 40 animals sacrificed
before 18 months in the highest dose group is not included in Table 5-23.
5.4.2.2 Choices of Low-Dose Extrapolation Models
In addition to the multistage model currently used by the CAG for low-dose
extrapolation, estimates of risk from exposure to gasoline vapor were also
determined using two other models (the probit and the Wei bull models). These
models cover almost the entire spectrum of risk estimates that could be generated
from existing mathematical extrapolation models. These models are generally
statistical in character, and are not derived from biological arguments, except
for the multistage model, which has been used to support the somatic mutation
hypothesis of carcinogenesis (Armitage and Doll 1954, Whittemore 1978, Whittemore
and Keller 1978.) The main difference among these models is the rate at which
the response function, P(d), approaches zero or P(0) as dose, d, decreases. For
instance, the probit model would usually predict a smaller risk at low doses than
the multistage model because of the difference of the decreasing rate in the
low-dose region. However, it should be noted that one could always artificially
give the multistage model the same (or even greater) rate of decrease as the
probit model by making some dose transformation and/or by assuming that some of
the parameters in the multistage model are zero. This, of course, is not
reasonable without knowing, a priori, what the carcinogenic process for the
agent is. Although the multistage model appears to be the most reasonable or
at least the most general model to use, the maximum likelihood estimate generated
*
from this model does not help to determine the shape of the dose-response curve
5-64
-------
TABLE 5-23. INCIDENCE RATES OF TOTAL KIDNEY TUMORS IN MALE FISCHER 344 RATS
EXPOSED TO UNLEADED GASOLINE VAPOR
(International Research and Development Corporation 1983)
Experimental dose (ppm)
0
67
292
2056&
Standardized
lifetime dose (ppm)a
0
11.96
52.14
367.14
Incidence
0/49
2/59
5/56
5/45
rate
(3.4%)
(8.9%)
(11.1%)
aThe dose in ppm is assumed to be equivalent between humans and animals. Since
the doses were given only 6 hours/day and 5 days/week, the lifetime dose is
calculated by multiplying the factor (5 x 7) x (6/24) to each of the experimental
doses.
bThe data from this group is not used in calculation.
TABLE 5-24. INCIDENCE RATES OF HEPATOCELLULAR TUMORS IN FEMALE MICE (B6C3FL)
EXPOSED TO UNLEADED GASOLINE VAPOR
(International Research and Development Corporation 1983)
Experimental
dose (ppm)
0
67
292
2056
Standardized
lifetime dose (ppm)
11.96
52.14
367.14
Carci noma/adenoma
incidence rate
8/57 (14.0%)
10/52 (19.2%)
13/57 (22.8%)
28/56 (50.0%)
Carcinoma
incidence rate
7/57 (12.3%)
6/52 (11.5%)
9/57 (15.8%)
20/56 (35.7%)
5-65
-------
beyond experimental exposure levels. Furthermore, maximum likelihood estimates
at low doses extrapolated beyond experimental doses could be unstable depending
on the amount of the lowest experimental dose; the upper-bound estimates from
the multistage model at low doses are relatively more stable than maximum
likelihood estimates. The upper-bound estimate can be taken as a plausible
estimate at low doses if the true dose-response curve is actually linear. The
upper-bound estimate means that the risks are not likely to be higher, but
could be lower if the compound has a concave upward dose-response curve or a
threshold at low doses. Because the estimated risk is a probability conditional
to the assumption that an animal carcinogen is also a human carcinogen, the
actual risk could range from a value near zero to an upper-bound estimate.
5.4.2.3 Calculation of Unit Risk (Risk at 1 ppm)
In the calculation of unit risk, ppm in air is assumed to be equivalent
between animals and humans. The data from the highest dose group in Table 5-23
is excluded in the calculation because the model dose not fit well if these data
are included (See 5.4.1.1 above). Furthermore, the data seem to indicate the
toxic effect in the highest dose group because only two-thirds of the animals
survived beyond 18 months. Using the tumor incidence data and the corresponding
lifetime dose presented in Tables 5-23 and 5-24, the cancer risks at 1 ppm are
calculated using the multistage model. The results are presented in Table
5-25. Both the 95% upperbound estimate and the maximum likelihood estimate
are given. Because the maximum likelihood estimate of the linear component in
the multistage model is not zero, the upper-bound estimate is only about two
times the corresponding point estimate. The cancer risk estimates in Table
5-25 can be used to represent the carcinogenic potency of unleaded gasoline
vapor. The kidney data in rats and the combined hepatocellular adenoma/carcinoma
5-66
-------
data in mice are closely similar, spanning a range from 2.1 x 10*3 to 3.5 x 10-3,
This range represents one measure of uncertainty in the upper limit of potency,
the lower limit being zero potency.
TABLE 5-25. ESTIMATES OF CARCINOGENIC POTENCY DUE TO EXPOSURE TO
1 PPM OF UNLEADED GASOLINE VAPOR
q* 95% upper-bound Maximum
Data base 1 estimate likelihood estimate
(1) Kidney tumor in
male rats 3.5 x 10'3 2.0 x 10-3
(2) Hepatocelluar carcinoma/
adenoma in female mice 2.1 x 10-3 1.4 x 10-3
Hepatocellular
carcinoma in
female mice 1.4 x 10-3 8.5 x 10-4
Geometric mean of (1) and (2) 2.7 x 1Q-3 1.7 x 10-3
5.4.2.4 Comparison of Risk Estimates by Different Low-dose
Extrapolation Models
For comparison, the probit and the Weibull models are also used to calculate
cancer risks at various dose levels. The calculated results are presented in
Table 5-26. The maximum likelihood estimates of the parameters in each model
are presented in Appendix B. The results shown in Table 5-26 indicate that all
three of the models predict comparable risks (within an order of magnitude) at
5-67
-------
1/5
O
O
0? 01
C r—
58
PO
CM 1
1 O
O-l
-4 X
X CO
O -t
PO •
CM —~
CM 1
1 O
x en
CM O
• CM
PO 1
1 O
o ••«
-4 X
X O
i-» in
in •
• PO
PO —
_
PO 1
1 0
o •*
x en
CO *
r*. •
• t-*
in
1 O
O -t
X CM
CMC.
CO •
• *-4
CM —
*r i
1 O
o <-*
-4 X
IO CM
• ^
r-.
in i
1 0
O -H
^4 X
X CM
co en
* ^f
-4 —
^j
A
2
0.
PO
CM 1
1 O
O —l
-4 X
xo
oo *
pv •
• iO
CM *-•
PO"
O -4
— I X
XO
O O
o •
• ^
PO"
PO 1
1 0
O i-l
-4 X
x in
p*. •
OO —
«*•
1 O
o ->
^^ X
^H ^f
0 •
• 00
^
1 O
O -4
X PO
in co
• CM
CM —
PO 1
1 O
0 -•
r-4 X
s^
• .-4
^fc 1
1 O
O — I
"x m
CM en
• in
10 —
^
jg
"5
<*> 1
o2
•-« X
• .— t
^j I
o2
— < X
xo
r-- CM
O •
•r—
4- i
1 0
o --«
—• X
x «>•
• >-t
CM —
in"
1 0
o **
US
<~- CM
0 •
in
in i
1 0
O -H
X "»•
* •«•
• ,^
CM •—
in i
• o
0 ^
•^ X
x o
S£
IO 1
2=3
X V
*t *t
• «»4*
CM —
4)
en
*
u>
^^
i
u
IO
^_
^ ^^
^ 2
O) O
u c
o •—
4-> O
IO L.
|S1
1-4 X
X «*
CM CM
PO *
ff* I
o2
-^ X
X CM
• ••*
in i
I O
O fH
-4 X
X *
en ^^
• PO
^»
IO I
' O
0 -H
xo
r» in
• in
en"
I 0
O —•
X r-«
CM en
• in
cS"
CO I
I O
0 -4
-4 X
X ^^
CO •
• IO
CM"
-* i
I O
2 ~*
x in
• CM
en >—
*»
"Z
o
a.
0)
-"5
U S
3 l«-
CM 'l
o2
•^4 X
•~l •
CM t
«r in
• CM
PO I
I O
O --I
"•« X
X —I
PO pg
in •
_
PO I
o2
^•4 X
—4 —4
so .
^tf
I 0
o -*
X OO
• va^
en —
in
1 O
0 —
••* X
PO QO
IO •
in
«»• i
o2
*^ X
x «»•
en en
• *-».
> V
•— o
§1
PO 1
o2
ooS
OO *
f*\ i^
I O
O -l
-4 X
x en
r* en
• CM
PO ~—
in"
^ i
1 O
o ~>
~+ X
X -H
P-4 •
• in
in
1 O
O -4
x m
-H P«.
CO •
• CM
PO - -
^o
in i
1 O
O -"
X r-
• in
co —
m i
1 O
0 -•
»-4 X
PO m
PO •
• PM
IO 1
I 0
O -<
x in
10 r~
CM •
en —
^2.
3
"v
5-68
-------
1 ppm, but the difference becomes greater as the dose becomes smaller. For
instance, on the basis of hepatoceulular tumors, the miltistage model predicts
a much higher risk than that predicted by the probit model at a dose level of
0.001 ppm. This observation is not surprising, since the tangent (slope) of
the probit curve approaches zero as dose approaches zero, while the slope of
the multistage curve is linear at low doses. The risks predicted by the Wei bull
model on the basis of kidney tumors and hepatocellular carcinoma/adenoma are
higher on the entire exposure range (0.001 ppm to 1.0 ppm) than the multistage
model because the Weibul model shows a sub-linear dose-response relationship
which is not considered biologically plausible (see section 5.4). For this reason,
low-dose linearity has intuitive appeal. For example, the incidence of hepatocellular
tumors at the lowest experimental dose (11.96 ppm) is 10/52, and the incidence
in controls (0 ppm) is 8/57. In the absence of knowledge as to the shape of
the dose-response relationship below the lowest experimental dose level, the
only reasonable method of estimating cancer potency without having the possibility
of seriously underestimating the true risk is to use linear extrapolation.
That is, the slope (potency) is calculated by:
(10/52 - 8/57)/11.96 = 4.3 x 10-3/ppm
This crude estimate is about threefold greater than the maximum likelihood
estimate (1.44 x 10-3/ppm) calculated from the multistage model which utilizes
all the data points, including the lowest data point used in the above calculation.
If one assumes that the dose-response curve is concave upward at low doses,
the risk calculated by the low-dose linear model can be considered an upper-
bound estimate of the true risk; it is the only plausible estimate that does
not have the potential for underestimating the true risk on the basis of the
5-69
-------
data given. Any more precise estimate would require either further assumptions
about the shape of the dose-response curve or biological knowledge of the
mechanism of carcinogenic action.
In comparing the models in Table 5-26, note that the multistage model
has higher maximum likelihood estimates than the probit model for all doses
lower than 0.01 ppm. This result is attributed to the low dose linearity
characteristic of the multistage model's prediction of the dose-response
relationship. In fact, the multistage model produces a linear relationship
over the entire range of exposure estimates (0.001 ppm to 1.0 ppm) that would
be produced by emissions of controlled and uncontrolled gasoline vapors.
Based on its low dose linearity characteristic, the multistage model is selected
as the model EPA should rely upon to estimate risk of exposure to gasoline
vapor because it provides conservative estimates at low doses (i. e., below
0.01 ppm) and an adequately conservative approximation of risk at higher doses.
At higher dose levels in the range of ambient exposure levels that would result
from continued release of uncontrolled gasoline vapor emissions (i.e., between
0.01 ppm and 1.0 ppm), the multistage and probit models produce approximately
the same maximum likelihood risk estimates (within the error band of these
models estimates). Although the multistage model maximum likelihood risk
estimates are about one-half of the probit model estimates at dose levels of
0.05 ppm to 1.0 ppm, this result is not considered to be a significant factor
given the inherent uncertainties in developing risk estimates. In fact, the
maximum likelihood risk estimates of each model being within a factor of 2 of
one another is considered good agreement.
5-70
-------
5.4.2.5 Uncertainties of Quantitative Risk Assessment
5.4.2.5.1 Uncertainties Associated with Potency Estimates
It is well known that different models, all of which might fit well with a
given set of data over the experimental dose range, might nevertheless predict
drastically different responses at low doses. Gasoline vapor data are no
exception. As shown in Table 5-26, the multistage model predicts much higher
risk than the probit model at 0.001 ppm, on the basis of liver tumor incidence.
The risk estimate at low doses for unleaded gasoline vapor is calculated by
using the linearized multistage model, which is linear at low doses. The
potency estimate derived from such a model has been considered an upper-bound
estimate on the assumption that the shape of the dose-response curve is upwardly
concave at low dose levels. The carcinogenic potency, qf, as derived from the
multistage model, represents the 95% upper-bound confidence estimate, reflecting
only the statistical variability of the response data.
The low-dose risk estimate derived from animal data must further be extra-
polated to humans. There are many factors that must be considered in extrapo-
lating risk from animals to humans. Included among these factors are differ-
ences between humans and animals with respect to life span, body size, genetic
variability, and pharmacokinetic effects such as metabolism and excretion
patterns. In assessing the risks of gasoline vapor, it was assumed that ppm
in air will induce the same tumor response in humans as in animals. It is
questionable, however, whether this simple assumption is capable of accounting
for all the differences between humans and the animals that were used in the
gasoline experiment of the IRDC (1983).
An important but often neglected factor in risk assessment is the weight
of the evidence that gasoline vapor is carcinogenic to humans. The risk
5-71
-------
estimate derived from animal data gives only a conditional probability of cancer
on the assumption that the agent is carcinogenic to humans.
5.4.2.5.2 Uncertainties Associated with the Use of Potency
"Estimates to Predict Individual Risks in Real-life
Exposure Patterns'
The carcinogenic potency estimate for unleaded gasoline can be used to
predict the human cancer risk from continuous gasoline exposure, subject to the
uncertainties previously discussed. The actual human exposure to gasoline
vapor, however, is likely to be only a few minutes per week. The questions
then arise as to whether this exposure can be averaged over the entire week
in order to arrive at a continuous exposure estimate, and whether overestimation
or underestimation of the risk of intermittent doses would result from such
averaging. The available data for analogous situations indicate that either of
the two possibilities may be true. In studying factors modulating the
carcinogenici ty of benzidine, Vesselinovitch et al. (1975) demonstrated that
twice-weekly administration of benzidine by stomach intubation was less effective
in inducing liver and harden an gland tumors but more effective in inducing
lung adenomas than the continuous (daily) feeding of equivalent doses.
Another example which may or may not be relevant to the case of gasoline
vapor exposure are the studies of low-linear energy transfer (LET) radiation.
After reviewing all the data on radiation-induced genetic and tumorigenic
effects in plants, "simple" biological systems, animals, and humans, the National
Council on Radiation Protection and Measurements (NCRP 1980) concluded that,
for a given total dose, the high-dose rate exposure is more effective than the
low-dose rate exposure in producing the response and that the difference in
response between the two exposure patterns diminishes as the total dose decreases.
5-72
-------
The applicability of this observation to the human exposure to gasoline vapor
is not known. If one assumes that the gasoline vapor has the same dose-rate
effect as the low-LET radiation exposure, then the use of averaging dose would
not overestimate the risk and would give a close approximation to the true
risk when the exposure level is small. In this discussion, it is assumed that
the dose-response relationship obtained previously predicts accurately the true
risk when the dose is continuous.
In general, three possible situations can occur in estimating cancer risk
due to gasoline vapor exposure in the real-life situation when the averaging
dose is used:
1. The real-life (intermittent) exposure pattern and the continuous
(averaging dose) exposure patterns are equally effective. In this case the
risk estimate is unbiased.
2. The real-life exposure pattern is more effective than the continuous
exposure pattern. In this case, the risk is underestimated when the dose is
averaged.
3. The real-life exposure pattern is less effective than the continuous
exposure pattern. In this case, the risk is overestimated when the dose is
averaged.
Not enough is known about the mechanism of action to state which possibility
is the most likely or to know the magnitude of either the overestimation or
the underestimation.
5.4.2.6 Cancer Risk Attributable to Benzene Content in Gasoline
Vapor
To estimate the cancer risk which could be quantitatively attributable
to the benzene content in gasoline vapor, the following assumptions are made:
5-73
-------
1. Tumor response due to benzene content in gasoline vapor is additive.
That is, benzene does not act synergistically or antagonistically with other
chemical compounds in the gasoline vapor complex mixture.
2. Tumor responses due to benzene exposure need not be site-specific
among different species or strains. This assumption is made for the purpose of
quantitative analysis, but may not be valid from a biological point of view.
3. Part per million (ppm) in air is assumed to be equally effective in
inducing tumors among different species.
4. The absorption rate for rats is similar irrespective of the route
of exposure (gavage or inhalation).
Table 5-27 summarizes the cancer risk of benzene at 1 ppm. Both 95% upper-
bound and maximum likelihood (point) estimates are presented. Details on the
data and calculations are presented in Appendix C. It should be noted that the
potencies presented in Table 5-27 are to be used solely for determining the frac-
tion of tumor response in the gasoline vapor study that is attributable to
benzene content, and should not be construed as the CAG's estimates of the
carcinogenic potency of benzene in humans. The fraction of the unleaded gaso-
line tumor response attributable to benzene content can be expressed as:
AR = potency (benzene) x 0.02/potency (gasoline)
where 0.02 is the reported benzene content. Since both gasoline and benzene
potency estimates calculated on the basis of different data sets are comparable,
it is appropriate to use the geometric means presented respectively in Table 5-25
and Table 5-27 to calculate AR.
5-74
-------
TABLE 5-27. ESTIMATES OF THE CARCINOGENIC POTENCY
OF BENZENE (RISK AT 1 PPM)
Data base
Female ratsa
Male ratsb
Female rats*5
Male micec
Geometric mean
Upper-bound
estimate
1.3 x 10-2
7.9 x 10-3
1.3 x 10-2
1.4 x 10-2
1.2 x 10-2
Maximum likelihood
estimate
8.0 x 10-3
5.2 x 10-3
8.7 x 10-3
6.9 x 10-3
7.1 x 10-3
aZymbal gland carcinoma (gavage); Mai torn' et al. (1982).
bZymbal gland carcinoma (gavage); NTP (1983).
cHematopoietic neoplasms (inhalation); Snyder et al. (1980).
5-75
-------
When upper bound estimates are used:
AR = 1.2 x 10-2 x 0.02/2.7 x 10-3 = Q.Q9
When maximum likelihood estimates are used:
AR = 7.1 x 10-3 x 0.02/1.7 x 10-3 = Q.08
These calculations indicate that from the quantitative viewpoint alone,
the benzene content accounts for less than 10% of the tumor responses observed
in the IRDC (1983) unleaded gasoline study.
Another way to determine the quantitative benzene contribution to the
tumor response is to calculate the expected increase of tumor-bearing animals
and compare it with the corresponding observed response (after adjusting for
the background rate) at each of the two lowest experimental doses in which
toxic effects were not observed. These calculations (not shown here) also
indicate that about 10% of responses to gasoline could be due to benzene content.
One of the main uncertainties associated with the conclusion made above is the
assumption that the risk due to benzene is additive to that of gasoline vapor.
There is no evidence to support or deny this assumption. However, it can be
shown that, under the multistage theory of carcinogenic!"ty, if two carcinogens
act on different stages of carcinogenesis, a multiplicative effect will result.
There is abundant evidence that a carcinogen or a non-carcinogen could modify
(enhance or inhibit) the carcinogenic action of another compound. Since gasoline
vapor contains more than one chemical compound, such interactive effects are
likely. Further research is needed to identify which compound (e.g., benzene)
or fraction of compounds is responsible for the carcinogenic effect.
5-76
-------
5.4.3 Summary of Quantitative Risk Estimation
Data from the API study on kidney tumors in male rats and liver adenomas
and carcinomas in female mice were used to derive an estimate of the incremental
upper-limit unit risk due to continuous human exposure to 1 ppm of unleaded
gasoline. Since the animals breathed the complete mixture under laboratory
conditions, whereas humans are expected to breathe only the more volatile
components of the mixture, the estimates are uncertain. The estimates from
the mouse and rat data are similar: 2.1 x 10-3 (ppm)-l from mouse data and
3.5 x 10-3 (ppm)"1 from rat data.
The presence of 2% benzene in the unleaded gasoline mixture could
theoretically contribute to the response, although the mouse liver and rat
kidney have not been target organs in animal experiments with benzene. Based
on those experiments, it is estimated that the contribution of benzene to the
response observed in the API unleaded gasoline studies could be on the order
of 10%. However, there is no qualitative evidence that benzene actually
is contributing to the response.
5.5 Summary and Conclusions
5.5.1 Summary
5.5.1.1 Qualitative
5.5.1.1.1 Animal Studies
A lifetime inhalation bioassay of unleaded gasoline in Fischer 344 rats
and B6C3F1 mice has induced a statistically significant incidence (6/100) of
renal carcinomas in the kidney cortex of male rats and a larger, also statistically
significant, incidence (20/100) of hepatocellular carcinomas in female mice.
Female rats and male mice had no significant treatment-related induction of
tumors at any organ site. The incidence of renal tumors was statistically
significant at the highest dose tested (2,056 ppm) but not at the two.lower
5-77
-------
doses (292 ppm and 67 ppm). In mice, the incidence of liver carcinomas alone
and adenoma and carcinoma combined, was also statistically significant in the
highest but not the two lower dose groups. Moderate decrements in the body
weight gain in the high-dose groups indicate that the maximum tolerated dose
was reached. Glomerulonephrosis occurred in nearly all of the male rats, and
mineralization of the pelvis was correlated with dose. However, there was no
correlation between animals with tumors and those with mineralization.
The acute and subchronic renal toxicity of decalin, a volatile hydrocarbon
of the same general type as those in gasoline, is confined to male rats and
does not occur in female rats or in mice, dogs or guinea pigs. In a series of
21-day inhalation exposures of male rats to a variety of chemical fractions of
gasoline, renal toxicity was correlated with the paraffin components and not
with the aromatic compounds in the mixture. The same pattern of renal toxicity,
as well as a positive renal tumor response, occurred in response to chronic
inhalation of two synthetic fuels (RJ-5 and JP-10). Chronic inhalation studies
with the jet fuels used by the Air Force and Navy (JP-4 and JP-5) have shown
the same nephrotoxic lesions, but no statements can be made about the carcinoma
response until histopathological analyses are completed. The renal toxicity
pattern observed with exposure to hydrocarbon mixtures involving protein accumu-
lation in renal tubules, is clearly different than the kidney lesions occurring
spontaneously in old rats, and occurs in males of both Fischer 344 and Sprague-Dawley
strains but not in females of these strains or in mice or monkeys. Mutagenesis
tests of unleaded gasoline have been carried out in Salmonella, yeast, mouse
lymphoma in vivo cytogenetics and mouse dominant lethal systems. Various gasoline
feedstocks have been tested in mouse lymphoma and in vivo cytogenetics assays.
The results of most of these assays have not met the criteria for positive
responses. A detailed examination of their adequacy is in process. -
5-78
-------
5.5.1.1.2 Epidemlologic Studies
Three epidemiologic studies of workers exposed to petroleum products
including gasoline have been reviewed: two published [Thomas et al., (1980,
1982)1 and one unpublished [Rushton and Alderson (1982)].
Thomas et al. (1980, 1982) reported on two studies in which the death
records of male individuals who worked in oil refineries and petrochemical
plants were reviewed for specific causes of death. The 1980 paper examined the
records of workers who at the time of death were active members of the Oil
Chemical and Atomic Workers International Union (OCAW). The 1982 paper was
expanded to include retired members of the OCAW at the time of death.
Proportionate Mortality Ratios (PMR) adjusted for age and calendar time using
the United States general population were computed. The 1980 study showed that
the PMRs for kidney and stomach cancers were significantly increased (P < 0.05)
for white males who joined the union 20 or more years prior to death. However,
black males whose lengths of union-membership were less than 10 years, as well
as those whose union membership we>e 10 or more years, experienced significantly
greater than expected frequencies of stomach cancer deaths. As indicated in
the 1982 paper, there were differences in mortality patterns for white males
among both active and retired union members. Significant relative excesses
(P < 0.05) of stomach, pancreatic, and brain cancer deaths were seen among
active members. However, among retired members, the PMRs were significantly
elevated for prostate cancer, Hodgkin's disease, multiple myeloma, and leukemia
deaths.
Rushton and Alderson (1982) presented the results of a retrospective cohort
mortality study of male workers at distribution centers from three oil companies
in Great Britain. The notable result was that there were 12 kidney cancer
5-79
-------
deaths among drivers, which represented a significant excess (P < 0.05) in
comparison with the 7.03 expected number of deaths.
The studies by Thomas et al. (1980, 1982) present problems in their
definitions of the study population and in their methodology. The limitations
inherent in proportionate mortality ratios (PMRs) are sufficient to cast doubt
on the results of these studies. PMRs reflect inflated estimates of mortality
if the study group has a lower mortality rate than the comparison group for
all causes of death. Also, excesses for one or more causes may automatically
lead to a deficit in others.
Because of the incomplete nature of the study by Rushton and Alderson
(1982), it is judged to be inadequate. All three studies suffer from insufficient
documentation of exposure and employment histories and questionable applicability
for assessing the carcinogenicity of unleaded gasoline.
5.5.1.2 Quantitative
Data from the API study on kidney tumors in male rats and liver adenomas
and carcinomas in female mice were used to derive an estimate of the incremental
upper-limit unit risk due to continuous human exposure to 1 ppm of unleaded
gasoline. Since the animals breathed the complete mixture under laboratory
conditions, whereas humans are expected to breathe only the more volatile
components of the mixture, the estimates are uncertain. The estimate from the
mouse and rat data are similar: 2.1 x 10-3 (ppm)-l from mouse data and 3.5 x 10-3
(ppm)'1 in rat data.
The presence of 2% benzene in the unleaded gasoline mixture could
theoretically contribute to the response, although the mouse liver and rat
kidney have not been the target organs in animal experiments with benzene.
Based on those experiments, it is estimated that the contribution of benzene to
5-80
-------
the response observed in the API unleaded gasoline studies could be on the
order of 10%. However, there is no qualitative evidence that benzene actually
is contributing to the response.
5.5.2 Conclusions
The occurrence of a small but definite kidney tumor response in male
rats and a significant hepatocellular response in female mice furnish sufficient
evidence, using the criteria of the International Agency for Research on Cancer
(IARC), for the carcinogen!city of unleaded gasoline in animals. The similar
pattern of response in rats to the synthetic fuels RP-5 and JP-10, and the renal
toxicity observed in chronic bioassays with JP-4 and JP-5, support the findings
with unleaded gasoline, indicating that some agent or combination of agents
common to these mixtures is responsible for the observed effects.
The scattered reports of kidney cancer in workers exposed to gasoline-
related compounds hint that some effect may be occurring in humans, but the
evidence is judged to be too poor to justify anything but a classification of
inadequate under the IARC criteria for epidemic!ogic evidence. Therefore,
unleaded gasoline should be placed in IARC category 2B, meaning that unleaded
gasoline is a probable human carcinogen.
5-81
-------
6.0 ISSUES TO BE ADDRESSED BY THE SCIENCE ADVISORY BOARD
Before EPA can evaluate the risk gasoline vapor emissions may pose
to human health, EPA needs the Science Advisory Board's advice on the
soundness of the scientific studies performed by the American Petroleum
Institute and other relevant studies discussed in section 5.0. Advice
is also needed on methdology for deriving unit risk factors from these
studies. It would be especially helpful if the Science Advisory Board
would also address the following questions:
6.1 Quality of Evidence
1. Do you see any defect in the design or conduct of the animal
studies that would cause you to seriously question the results?
2. Do you agree with our conclusion that the test gasoline is
an animal carcinogen?
3. Does the available evidence permit any conclusion on the
likelihood that gasoline vapor is carcinogenic in humans?
a. does the available human evidence support the API animal
results?
b. is the difference in the composition of the test gasoline
compared to ordinary gasoline a serious drawback to the
relevancy of the animal studies for estimating human risks?
c. does the fact that the studies used completely volatilized
gasoline rather than the mixture of higher volatiles
characteristic of partial evaporation represent a serious
drawback to the interpretation of the results for human
exposure?
6-1
-------
4. Does the SAB agree with the explanation given for the absence
of benzene-like carcinogenic response in the API studies? Were any factors
overlooked? Is the assumption that the risks from benzene and gasoline vapor
are additive valid for the purposes of analysis?
6.2 Quantitative Risk Assessment
1. Given the need for quantitative estimates of gasoline vapor
health risks, does EPA's methodology for the derivation of the unit risk
factors constitute a reasonable approach?
a. is there reason to believe that the use of a linear model for
dose/response extrapolation is inappropriate in this case?
b. is it reasonable to combine malignant and benign tumors in
extrapolating cancer risks from the animal data?
2. Are the rat and mouse strains used in the API studies equally
applicable as the basis for risk extrapolation to humans?
3. Does the lifetime exposure regimen characteristic of the
animal bioassays seriously compromise the use of these studies in estimating
risks for human populations intermittently exposed to gasoline vapor (e.g.
self-service refueling)?
4. Are the uncertainties in the gasoline vapor unit risk factors
adequately described?
6-2
-------
APPENDIX A
SUMMARY OF
API INHALATION STUDY
-------
A CHRONIC INHALATION STUDY
WITH UNLEADED GASOLINE VAPOR
H.N. MacFarland,1 C.E. Ulrich,2 C.E. Holdsworth,3
D.N. Kitchen,4 W.H. Halliwell,5 and S.C. Blum6
Gulf Life Sciences Center, 260 Kappa Drive, Pittsburgh, PA 15238,
to whom reprint requests should be addressed.
2
International Research and Development Corporation, Mattawan, MI
American Petroleum Institute, Washington, D.C.
4
Biolabs, Inc., Las Cruces, NM
5 Westpath, Inc., Ft. Collins, CO
Exxon Research and Engineering Co., Linden, NJ
Accepted for Publication
by the Journal of American Toxicology
March 1984
A-l
-------
ABSTRACT
A chronic inhalation study of unleaded gasoline vapor was
conducted in mice and rats. The gasoline employed was typical of
gasoline used in the U.S. and contained 2 percent benzene. Groups of
both sexes of B6C3Fj mice and Fischer 344 rats were exposed to three
concentrations of vapor, 67, 292, and 2056 ppm. Exposures were for 6
hours per day, 5 days per week, for periods ranging from 103 to 113
weeks. Interim sacrifices were conducted at 3, 6, 12, and 18 months.
Laboratory studies, including hematological and biochemical deter-
minations, were performed on rats at the interim sacrifices and at
termination. Histopathological studies were conducted on both
species at every interval.
No consistent compound-related changes were seen in pharma-
cotoxic signs, mortality, hematological or biochemical indices in
either species. Significant depression of body weight gain was seen
in both sexes of rats and male mice exposed to the highest level of
gasoline vapor. On gross necropsy, a compound-related increase in
liver nodules and masses was seen in female mice exposed to the high
level.
The most interesting observations were made on histopath-
ological examination of the rats' tissues and, of these, pathological
changes in the kidneys were the most striking. Renal carcinomas or
sarcomas, in the cortex or near the renal poles, were seen in the
male rats at all dose levels, with some evidence of a dose-response
relationship. One female rat in the intermediate dose group exhib-
ited a renal sarcoma. Two mice had renal tumors, considered to be
spontaneous neoplasms. Mention is made of new studies that have
been prompted by the present findings.
A-2
-------
INTRODUCTION
Although gasoline, a fuel for the internal combustion en-
gine, has been manufactured and used for several decades, no chronic
investigation of its toxicological properties has been undertaken.
To rectify this gap in our knowledge, the American Petroleum
Institute began in the early 1970s to sponsor a program of longer
term studies. A 90-day inhalation investigation with leaded and
unleaded gasoline in rats and monkeys was completed in 1976 and later
a paper was written for publication (Kuna and Ulrich, 1983). During
the long hiatus between the original 90-day study report and the
later paper of Kuna and Ulrich, a careful re-evaluation of the
study's kidney tissues was undertaken for toxic signs consistent with
those being observed for other hydrocarbon solvents. Upon reexamina-
tion by pathologists familiar with nephrotoxic lesions, subtle regen-
erative changes were discovered in the renal tubules. These minimal
changes were seen only in male rats.
Shortly after the completion of the 90-day study in 1976,
but before re-evaluation of the kidney slides from that.study, the
present chronic study was begun in rats and mice . The study
protocol was adapted from that recommended by the National Cancer
Institute (NCI)(1976). Unleaded gasoline was utilized in an
inhalation investigation in which exposures were continued for 24 to
26 months.
Nephrotoxic lesions were seen in the chronic study. An
unexpected finding was primary renal neoplasms in male rats near or
at termination of the study. Both nephrotoxic and nephrocarcjnogenic
findings in male rats have stimulated further exploratory programs
now in progress.
A-3
-------
MATERIALS AND METHODS
Gasoline Sample
The unleaded automotive motor fuel (gasoline) used in the
study was prepared to conform with the specifictions of unleaded gas-
oline in use in the United States in 1976, as determined by a road
octane survey (DuPont Road Octane Survey, Summer 1976). At the time
the gasoline was blended for the study, benzene concentrations in
U.S. gasolines averaged about 1 percent with a maximum approaching 2
percent; therefore, benzene content of the gasoline was adjusted to
the upper limit of U.S. gasolines. The specifications are shown in
Table 1, but more detailed information on chemical composition is
provided in Appendix 1.
Table 1
Animals
Fischer 344 albino rats and B6C3F^ mice, each species
equally divided as to sex, were utilized. After a 2-week quarantine
period just prior to initiation of exposures, weight ranges were as
fol1ows:
Rats, male 95-129 g.
Rats, female 79-105 g.
Mice, male 14—26 g.
Mice, female 12—20 g.
At this time, both mice and rats were approximately 6 weeks
of age. They were provided with Purina* Laboratory Chow* #5001-up to
A /I
-------
week 38; thereafter Purina* Laboratory Chow* #5002 was used. Tap
water and chow were available ad libitum except during the actual
exposures.
Design
From larger groups of a given species and sex, only animals
which appeared healthy were selected. They were further restricted
as to weight range, using only those rats, both sexes, and female
mice whose weights were within +1.5 standard deviations of the group
mean; +1.6 standard deviations was permitted for the male mice. The
animals were assigned at random, with 100 animals of each species and
sex, i.e., a total of 400 per chamber in each group, in the design
shown in Table 2.
Table 2
Interim sacrifices of 10 randomly selected animals of each
species and sex were performed at 3, 6, 12, and 18 months.
Chamber Operations
Exposures were conducted in 16 nr stainless steel and glass
exposure chambers (Fig. 1), designed by Leong (Leong, 1976; Drew,
1978). the supply air was filtered and controlled for temperature
and humidity, and flow rates between 900 and 1900 liters per minute,
depending on the desired chamber concentrations, were established by
the main exhaust pump. Temperature and humidity were measured each
day at the start of exposure and at 1, 3, and 5 hours. Gasoline was
delivered from a liquid metering pump to a heated counteccurrent
A-5
-------
vaporization column and completely volatilized. Dry nitrogen at 5-6
liters per minute was used to carry the vapor into the main inlet
pipe of the chamber. The exposure pattern was 6 hours per day, 5
days per week, for periods which ranged from 103 to 113 weeks. The
target concentrations of gasoline were 50, 275, and 1500 ppm.
Cheaical Analysis
Nominal concentrations were determined daily and
calculations of concentration in ppm were made by using weight loss
data and assuming an average molecular weight of 108 for the
gasoline.
Analytical concentrations were determined by drawing samples
from the chambers into a gas chromatograph equipped with a flame
ionization detector. The operating conditions for the chromatograph
are shown in Table 3.
Table 3
These conditions resulted in the appearance of a single peak for the
complex hydrocarbon mixture, thereby facilitating expression of
results as total hydrocarbon concentration.
Standard curves for calibration were prepared by injecting a
known volume of liquid gasoline into a 25-liter Saran bag filled
with nitrogen. It was found, after the experiment had been in pro-
gress for 24 weeks, that the gas chromatograph responded differently
to gasoline standards prepared in nitrogen as compared to chamber
samples of gasoline vapor in air. The magnitude of the correction
A-6
-------
factor to be applied for each of the three concentrations under
investigation was determined; it varied nonlinearly depending on the
absolute concentration. This showed that the target concentrations
of 50, 275, and 1500 ppm had been, in fact, 67, 292, and 2056 ppm,
with standard deviations of +3.1; 11.0, and 110.4, respectively, and
the study was continued at these concentrations.
Biological Estimations
Animals were observed twice daily for signs of toxicity,
behavioral changes, general appearance, and deaths. Each animal was
individually examined for clinical signs and palpable tissue masses
once a month. Individual body weights were recorded monthly for the
first 17 months and biweekly thereafter.
Serum biochemical determinations were performed on seven
male and seven female rats randomly selected from each group at the
interim sacrifices (3, 6, 12, and 18 months) and at termination. The
rats were fasted overnight, blood withdrawn from the orbital sinus,
and the following enzyme activities determined as recommended by NCI
(1976): alkaline phosphatase, glutamic oxalacetic transaminase,
glutamic pyruvic transaminase, ornithine carbamyl transferase, and
isocitrate dehydrogenase.
Hematologic evaluations were conducted at the 18-month
interim and terminal sacrifice on the same rats used for biochemical
A-7
-------
determinations at those time points. The fallowing variables were
measured: hemoglobin, hematocrit, erythrocyte count, total and
differential leucocyte count, platelet count, reticulocyte count,
mean corpuscular volume, mean corpuscular hemoglobin, and mean
corpuscular hemoglobin concentration.
Gross and microscopic examinations of tissues were performed
on animals dying during study, those obtained at the interim sacri-
fice periods, and those sacrificed at termination. A 40 percent
survivability criterion was used to terminate each group; this
resulted in the termination times shown in Table 4.
Table 4
At the 3, 6, and 12 month interim sacrifies, ten rats and
ten mice of each sex were asphyxiated with carbon dioxide, and a
complete necropsy was performed. At the 18-month interim sacrifice
and at termination, animals were sacrificed by sodium pentobarbital
anesthesia and exsanguinated. The trachea and lungs were removed at
maximum inspiration and examined while inflated and deflated. The
contents of the abdominal, thoracic, and cranial cavities were
examined in situ and after dissection.
After trimming of fat and connective tissue, the tissues
listed in Table 5 were weighed.
Table 5
The tissues listed in Table 6 were fixed in phosphate-buff-
ered neutral formalin; hematoxylin and eosin stained paraffin sec-
tions were prepared for microscopic examination.
A-8
-------
Table 6
Statistical Procedures
Body weight, hematologic, and serum biochemical data were
tested for homogeneity of variance (Steel and Torrie, 1960), followed
by a parametric analysis of variance. When a significant F-ratio was
obtained, individual group comparisons were performed, utilizing
student's t-test when variances were heterogeneous and Dunnett's test
(1964) when homogeneous.
In some cases where the number of animals was small and the
variances heterogeneous, the nonparametric multiple-group test of
Kruskal-Wallis was applied and where appropriate, individual group
comparisons were made with the Mann-Whitney U test (Siege!, 1956).
Data from male rats were.analyzed for mortality, all renal
tumors, malignant tumors, and renal adenomas, carcinomas and undif-
ferentiated tumors combined, using procedures outlined in Thomas et
al. (1977). Life table curves were computed and tested for homo-
geneity by both approximate and exact methods. A pair-wise compar-
ison of groups was made. In addition, each datum set was examined
for linear trend in the proportions, using both unadjusted and time-
adjusted tests. The exact test for trend and approximate test for
homogeneity and departure from trend were performed. Differences in
pairs of proportions were examined.
A-9
-------
RESULTS
Chamber Conditions
As indicated above, the actual concentrations of gasoline
vapor in the chambers (67, 292, 2056 ppm) were higher than the orig-
inally planned target concentrations (50, 275, 1500 ppm), but when
the calibration discrepancy was recognized, it was decided to
continue the animal exposures at these higher concentrations through-
out the study.
The temperatures and humidities in the four chambers +_S.D.
ranged from 24 +_1.4 to 26 +1.3C, and 52 +9.5 to 56 +7.2,
respectively.
General Aniaal Observations
Some minor signs were noted intermittently in the study,
including ocular discharge and apparent irritation in all four groups
of rats. In mice, a significant number of animals developed alope-
cia, ranging in size from a small restricted area to a generalized
hair loss over as much as two-thirds of the animals' bodies. The
alopecia was seen in all groups, including controls, with approx-
imately equal incidence.
No significant differences in spontaneous death rate were
seen in female rats and mice. Male control rats, Group I, exhibited
a significantly higher death rate after week 80 than any of the
exposed groups. The male rats in Groups II, 67 ppm, had a partic-
A-10
-------
ularly low spontaneous death rate. The following significant dif-
ferences were noted in male mice: Groups II and III, 67 and 292 ppm,
had a higher death rate than controls, but the Group IV male mice,
2056 ppm, exhibited a lower death rate when compared to controls.
Some statistically significant depressions in body weight
were encountered. Male rats in Group IV had significantly lower body
weights than controls from Week 13 to termination. Female rats in
Group IV showed a similar depression which was significant from Week
26 to the end of the study. Male mice in Group IV exhibited a lower
body weight than controls; the differences were significant from Week
66 to termination. In addition changes were noted in relative {in
relation to body weight) and absolute organ weights in rats. The
kidney weights of male rats of Group IV were elevated, both abso-
lutely and relatively, from the 3-month interim sacrifice through to
termination. At termination, the relative kidney weights of Group
III male rats and Group IV female rats were also elevated. There was
a dose-related relative increase in the testes and ovaries of Groups
III and IV rats, and a slight depression in absolute heart weights
was noted in Group IV males and females.
In mice, statistically significant alterations in organ
weights were noted sporadically throughout the study, but none of
these changes showed consistent trends, and thus they were not
considered to be exposure related. Neither kidney nor liver weights
were remarkable.
-------
Clinical Observations
The usual slight variability in the various hematological
indices was noted during the course of the study, but not considered
to be related to the gasoline exposure.
The evaluated biochemical variables were unremarkable
throughout the study. The serum ornithine carbamyl transferase
values were judged unreliable because of methodological problems and
were discounted.
Pathological Findings - Mice
The microscopic examination of tissues from the mice showed
a large variety of neoplastic and nonneoplastic changes throughout
the study which were not dose-related and were seen in both control
and treated groups. In the 18-month to final sacrifice period and at
final sacrifice, the female mice of Group IV exhibited an increased
incidence of hepatocellular tumors. The incidence for all groups
during the 18-month to final sacrifice time period was 45, 36, 45,
and 44 percent in male mice, and 14, 19, 21, and 48 percent in the
female mice, Groups I-IV, respectively.
There was some indication of a trend in the female mice in
Groups I, II, and III; however, the high incidence, 48 percent, in
the Group IV females was considered to be related to the exposure to
gasoline.
The tumors were of two types. Hepatocellular adenomas were
usually small and less than 1 cm in diameter. They were generally
spherical, did not contain distinct sinusoids or portal areas, and
were composed of hepatocytes that were usually larger than those of
-------
the surrounding parenchyma. The juncture of the tumor with the
surrounding parenchyma was distinct, and there was usually evidence
of compression of the surrounding hepatocytes. The hepatocellular
carcinomas were characterized by great variability of cell size, some
containing large nuclei. The border of the tumor with the surround-
ing hepatocytes was indistinct with evidence of invasion of the
surrounding parenchyma. The pattern of growth varied and included
trabecular and solid patterns with areas of necrosis or hemorrhage.
Several of the hepatocellular carcinomas in mice metas-
tasized to the lungs. In the final sacrifice, tumors in 7 percent
of the male mice in Group III and 2 percent in Group IV metastasized
to the lungs. No hepatocellular carcinomas in the final sacrifice
female mice metastasized to the lungs. In the moribund male mice and
those that died on test, tumors in 20 percent in Group I metastasized
to the lungs. In the moribund female mice and those that died on
test, tumors in 6 percent in Group I, 10 percent in Group III, and
7 percent in Group IV metastasized to the lungs.
Two female mice in Group IV exhibited renal tumors. One
mouse, killed at final sacrifice, had a papillary cystic adenoma of
the cortex. This adenoma consisted of a cystic space into which
projected small papillae composed of cells morphologicaly similar to
renal tubular epithelium. There was no evidence of peripheral
invasion; it had distinct and discrete morphologic limits. The other
mouse, which died during the 18-month to final sacrifice period,
A-13
-------
exhibited bilateral renal tubular adenocarcinomas. These tumors
replaced large portions of each kidney and contained large coalescing
areas of necrosis and hemorrhage.
Pathological Findings - Rats
At the 3-month interim sacrifice, dose-related histopathol-
ogical changes were observed in the male rats. These consisted of
cortical multifocal renal tubular basophilia, protein casts, and
chronic interstitial inflammation. The basophilia was characterized
by the presence of renal tubules containing basophilic epithelial
cells. The proteinaceous tubular casts occurred within dilated renal
tubules and were commonly located at the corticomedullary junction.
The incidence was 70 and 100 percent in Groups III and IV, respec-
tively. Chronic interstitial inflammatory foci with a predominantly
lymphoid cell type were observed at 20 and 70 percent incidence in
Groups III and IV, respectively. In addition, renal congestion and
very small foci of renal cortical mineralization were noted in several
rats.
In animals dying in the 3- to 6-month interval or sacrificed
at 6 months, the renal changes in male rats described above were again
evident. The incidence of tubular basophilia was 0, 40, 100, and 100
percent in Groups I to IV, respectively. Proteinaceous casts were
observed in 27 percent of the rats of Group I, 80 percent in Group
III, and 100 percent in Group IV. The incidence of chronic intersti-
tial inflammation was 18, 20, 100, and 100 percent in Groups-I to IV,
respectively. Mineralization in a radial pattern within the renal
A-14
-------
pelvis, with material located within tubules or the collecting ducts
of the renal pelves, was observed in 20 percent of the males in Group
IV.
At the 12-month interim sacrifice, the occurrence of protein-
o^v.ous casts in the kidneys of male rats was nearly equal in all
groups, 20, 30, 30, and 30 percent in Groups I to IV, respectively.
Mineralization in the renal pelvis occurred in 20 percent of the male
rats of Group III and in 80 percent in Group IV. Progressive glomer-
ulonephrosis was diagnosed in one male rat from Group IV. Another new
finding was karyomegaly, very large nuclei within renal tubular
epithelial cells in male rats.
The complexity of morphologic alterations observed in the
kidneys of all rats, especially males, increased after 18 months of
exposure. Progressive glomerulonephrosis occurred in higher inci-
dence than previously. The lesion was characterized by atrophied or
sclerosed glomeruli, dilated renal tubules containing proteinaceous
casts, tubular damage with regeneration or scarring, and the presence
of foci of chronic inflammatory cells. The incidence of glomerulo-
nephrosis in male rats was 20 percent in Group I, 30 percent in Group
III, and 20 percent in Group IV; the incidence in female rats was
lightly lower. Proteinaceous casts in kidneys of male rats were noted
in 50, 50, 40, and 60 percent in Groups I to IV, respectively.
Mineralization in the renal pelvis was seen in 20 percent of Group III
and 80 percent of Group IV male rats. Renal congestion was.commonly
seen and karyomegaly was again noted in male rats. A benign renal
A-15
-------
cortical adenoma was diagnosed in a Group IV male rat. Mononuclear
cell leukemia was diagnosed in the kidney of a female rat that died
during the 12- to 18-month interval.
At the final sacrifice, nearly all male rats exhibited
progressive glomerulonephrosis. The incidence rates were 100, 95, 97,
and 100 percent in Groups I to IV, respectively. A slightly lower
rate of occurrence was seen in female rats. Mineralization in the
renal pelvis occurred in 5, 63, and 91 percent of the males in Groups
II, III, and IV, respectively. Karyomegaly was observed occasionally
in the male rats. One male rat in Group III had renal tubular epithe-
lial hyperplasia at termination. The lesion was characterized by the
presence of a large dilated tubule containing a cystic lumen lined by
epithelial cells. Renal cysts, epithelial cell pigmentation, hydro-
nephrosis, chronic interstitial inflammation, congestion, cortical and
pelvic mineralization in female rats, and necrosis were among the
nonneoplastic lesions observed in the 18-month to terminal sacrifice
period.
Of great interest were primary renal neoplasms diagnosed at
termination or in those rats which died after 18 months. The total
number of these primary renal tumors was 14, with zero, one, six, and
seven in Groups I to IV, respectively, as shown in Table 7.
Table 7
All but one of these primary renal neoplasms occurredJn male
rats making the occurrence in males three adenomas, nine carcinomas,
A-1S
-------
and one sarcoma. The neoplasm in the female was a renal sarcoma or a
mixed malignant tumor.
The renal adenomas were characterized by the presence of
cuboidal to columnar epithelial cells, generarfy located in the
cortex, which formed tubular or papillary structures. The masses were
small, circumscribed, and the mitotic index was low.
The renal carcinomas varied in cellular morphology but gen-
erally contained epithelial cells arranged in a tubular or acinar
pattern. Cellular pleomorphism, cellular anaplasia, central hemor-
rhage and/or necrosis was common. The mitotic index varied but was
generally moderate to high. The histologic appearance varied greatly
within some individual neoplasms and contained well-formed to ill-
defined tubules. Other areas contained cells arranged in solid sheets
with little structural arrangement and a scanty connective tissue
stroma. Figure 2 is a photomicrograph of a typical renal carcinoma
obtained from a Group IV male rat at termination.
Histologically, the renal sarcomas displayed a variety of
cell types. The predominant type was a spindle cell, commonly seen
invading the edge of the lesion and infiltrating between normal renal
tubules. Other areas contained more solid sheets of spindle cells
arranged in a whorl-like pattern. Some areas within the neoplasms
were very anaplastic and pleomorphic in nature.
The renal adenomas and carcinomas were generally located" in
the cortex, but several were located near the renal poles. The
A-17
-------
sarcomas had a central or pelvic anatomic location.
After 12 months, both sexes of rats exhibited a mild, multi-
focal, pulmonary inflammatory response characterized by an accumula-
tion of alveolar macrophages in the alveolar spaces of the lungs. At
termination, the incidence of these aggregates of macrophages was 19,
5, 43, and 38 percent in males, and 40, 46, 34, and 67 percent in
females, in Groups I to IV, respectively.
DISCUSSION
Rats exhibited ocular discharge and appeared to be suscept-
ible to the irritant effects of the airborne gasoline vapor. Death
rates in male rats exhibited some differences among groups throughout
the study, but none of these were considered to be related to the
exposure. The depression in body weights seen in both sexes of rats
exposed to the high concentration, Group IV, is regarded as a toxic
stress effect of the gasoline exposure. Increases in kidney weights,
both absolute and relative, were noted particularly in the male rats
in the intermediate and high dose groups. There was also a slight
increase in the relative weights of gonads in these groups. These
changes in gonad weight may be, in part, a reflection of decreased
body weights. The hematological and biochemical findings in rats were
unremarkable.
The nephrotoxic changes seen at the 3-month and 6-month
interim sacrifices are in accord with the observations of several
investigators. Carpenter et al. (1975 a, b; 1977) reported renal
tubular regenerative changes and dilated tubules containing
A-18
-------
eosinophilic debris at the corticomedullary junction in male
Harlan-Wistar rats exposed to the vapors of Stoddard solvent, 60
solvent, and High Naphthenic solvent, all derived from petroleum.
These studies were performed under contract for the American Petroleum
Institute, as were the 90-day inhalation studies in Sprague-Dawley
rats and squirrel monkeys with leaded and unleaded gasolines reported
to API in 1976 and subsequently written for publication by Kuna and
Ulrich (1983). An initial reading of the slides of the kidney
sections from this latter investigation revealed no remarkable
observations but, after a careful reexamination some years later,
subtle changes were detected in the male rats exposed to a high
concentration (approx. 1500 ppm) of unleaded gasoline vapor. These
consisted of an increase in the incidence and severity of regenerative
epithelial changes, and dilated tubules containing proteinaceous
material were observed. Other investigators have also noted similar
alterations following -administration of certain petroleum solvents.
Other characteristics of the early nephropathy in the present study
included interstitial inflammatory focal reactions and a progressive
cortical mineralization. At the 12-month point, there was a decrease
and equalization in the incidence of proteinaceous casts, increase in
mineralization, and occurrence of karyomegaly in the renal tubular
epithelial cells of male rats.
The further progression of the early nephropathy becomes
increasingly obscured by the advent of "old rat nephropathy,* a pro-
gressive glomerulonephrosis. This condition was first diagnosed in
A-19
-------
one male rat In the high concentration group at the 12-month interim
sacrifice. By 18 months, 20 to 30 percent of the male rats were
affected and the incidence in the females was only slightly lower.
However, the mineralization in the renal pelves and karyomegaly in
male rats, seen prior to the onset of old rat nephropathy, were still
readily distinguishable at 18 months. At termination, essentially all
male rats and nearly all female rats exhibited old rat nephropathy.
The incidence of pelvic mineralization was increased and karyomegaly
was observed occasionally in the male rats.
It should be noted that, in the second year, two disease
processes seemed to be occurring in parallel, the old rat nephropathy
and a number of preneoplastic changes that appeared not to be concom-
itants of old rat nephropathy. These changes included karyomegaly,
hyperplasia, and an early benign neoplasm.
The surprising finding at termination was the primary renal
neoplasms, 13 of which were diagnosed in the male rats with evidence
of a dose relationship, and one sarcoma seen in a female rat in the
intermediate dose group. The spontaneous incidence of primary renal
tumors in the Fischer 344 rat 1s extremely low in both sexes (Coleman
et al. 1977; Goodman et al. 1979). It must, therefore, be concluded
that the dose-related incidence of such tumors in male rats in the
present study is to be ascribed to the exposure to wholly vaporized
gasoline.
The nonneoplastic pulmonary inflammatory response, seen after
12 months and at a slightly higher incidence in female rats, may be
A-20
-------
related to the slight irritant effect of the gasoline vapor. It is
interesting to note that no evidence of the progressive focal inter-
stitial fibrosis reported by Lykke and Stewart (1978) was found in the
present study. These authors exposed rats to 100 ppm of the vapors of
a leaded gasoline for periods ranging from 6 to 12 weeks.
In mice, alopecia was a frequent occurrence during the expo-
sure phase, but was seen in all groups, including controls; thus, it
does not appear to be related to the gasoline exposure. No remarkable
changes in death rate or organ weights were seen in the mice.
The pathological finding of interest in the mice was an
increased incidence of hepatocellular tumors, noted in the females in
the period from the 18-month sacrifice to termination. These tumors
are commonly seen in mice and have a significant spontaneous incidence
which is higher in males (Tarone et al., 1981). Whether the exposure
promoted the appearance of additional tumors or even initiated them
cannot be determined from the present study. In some cases, metas-
tasis to the lungs and kidneys were noted.
The most important findings in this chronic study are the
early and progressive renal tubular disease seen in male rats in the
first year, the advent and enhanced development of old rat
nephropathy in the second year with a parallel appearance of certain
preneoplastic changes, and the final appearance of primary renal
neoplasms in the male rats. The hypothesis has been advanced that
there may be causal connections between the early nephropath/ies and
the late appearance of renal neoplasms, with the preneoplastic changes
A-21
-------
In the second year as a possible link. New studies are planned to
explore this question.
In analyzing the results of this study, attention has been
directed to the gasoline, which is a complex mixture of several
hundred hydrocarbons (See Appendix 1). There are five main classes:
n-alkanes, isoalkanes, cycloalkanes, alkenes, and aromatics. Some
evidence is beginning to accrue which suggests that the renotoxic
effect of whole gasoline may be largely due to the presence of one or
two of the main types of hydrocarbons. In particular, the isoalkanes
are suspect (Cockrell et al. 1983; Pitts et al. 1983). Studies are in
progress to examine the relative activity of the five hydrocarbon
classes and individual molecular species.
Finally, the relevance of the results of this study to man is
under active investigation. Collectively, epidemiological studies of
populations that are exposed to gasoline in occupational situations
has not revealed any statistically significant increase in renal
carcinoma although slight increases have been detected in some studies
(Hanis et al. 1979; Hanis et al. 1982). It should be noted that, in
real-life situations where gasoline vapors are released, the vapors
tend to be richer in the low-boiling constituents. Analyses of such
atmospheres reveal total hydrocarbon concentrations generally less
than 60 ppm for approximately two minutes (0.28 ppm based on an 8-hour
Time Weighted Average).
A-22
-------
ACKNOWLEDGEMENTS
A number of investigators have contributed to various aspects
.v
of this study: its design, performance and evaluation. We wish
especially to thank Drs. B.K.J. Leong, W.R. Richter, J.F. Hardesty,
and Mr. N.K. Snyder for their assistance. Drs. R.N. Roth and C.A.
Lapin, along with Drs. S.C. Lewis and O.K. Baldwin, Ms. B.K. Hoover,
and Messrs. R.M. Siconolfi and R.C. Anderson, were particularly active
in the quality assurance review and evaluation of the detailed final
report of the study.
A-?.?
-------
REFERENCES
Carpenter, C.P., Kinkead, E.R., Geary, D.L., Jr., Sullivan, L.J., and
King, J.M. (1975 a). Petroleum Hydrocarbon Toxicity Studies.
III. Animal and Human Response to Vapors of Stoddard
Solvent. Tox. Appl. Pharmacol. 32_, 282-297.
Carpenter, C.P., Kinkead, E.R., Geary, D.L., Jr., Sullivan, L.J., and
King, J.M. (l!>75 b). Petroleum Hydrocarbon Toxicity Studies.
VI. Animal and Human Response to Vapors of "60 Solvent."
Tox. Appl. Pharmacol. 34. 374-394.
Carpenter, C.P., Geary, D.L., Jr., Myers, R.C., Nachreiner, D.J.,
Sullivan, L.J., and King, J.M. (1977). Petroleum Hydrocarbon
Toxicity Studies. XV. Animal Response to Vapors of "High
Naphthenic Solvent." Tox. Appl. Pharmacol. • 41_, 251-260.
Cockrell, B.Y., Iverson, W.O., and Phillips, R.D. (1983). Anatomical
Kidney Changes in Rats Following Inhalation Exposure to
C10"^ll Jsoparaffinic Solvent. The Toxicologist 3_, 25.
Coleman, G.L., Barthold, S.W., Osbaldiston, G.W., Foster, S.J., and
Jonas, A.M. (1977). Pathological Changes During Aging in
Barrier-Reared Fischer 344 Male Rats. J. Gerontol. 32.
258-278.
Drew, R.T., editor (1978). Proceedings, Workshop on Inhalation
Chamber Technology. Brookhaven National Laboratory, U.S.
Department of Energy, U.S. Department of Commerce,
Springfield, Virginia. PP. 14, 15.
A-24
-------
Dunnett, C.W. (1964). New Tables for Multiple Comparisons with a
Control. Biometrics 2£, 482-491.
DuPont Road Octane Survey, Summer 1976.
Goodman, D.6., Ward, J.M., Squire, R.A., Chu, K.C., and Linhart, M.S.
(1979). Neoplastic and Nonneoplastic Lesions in Aging F 344
Rats. Tox. Appl. Pharmacol. 48, 237-248.
Ham's, N.M., Stavraky, K.M., and Fowler, J.L. (1979). Cancer Mortality
in Oil Refinery Workers. J. Occup. Med. 21., 167-174.
Hani's, N.M., Holmes, T.M., Shellenberger, L.G., and Jones, K.E.
(1982). Epidemiological Study of Refinery and Chemical Plant
Workers. J. Occup. Med. 24, 203-212.
Kuna, R.A. and Ulrich, C.E. (1983). Subchronic Inhalation Toxicity of
Two Motor Fuels. J. Amer. Coll. Toxicol. (in press).
Leong, B.K.J. (1976). Proceedings, 7th Annual Conference on
Environmental Toxicology. National Technical Information
Service, Springfield, Virginia, pp. 141-149.
Lykke, A.W.J. and Stewart, B.W. (1978). Fibrosing Alveolitis
(Pulmonary Interstitial Fibrosis) Evoked by Experimental
Inhalation of Gasoline Vapors. Experientia J54_, 498.
National Cancer Institute (NCI) (1976). Guidelines for Carcinogen
Bioassay in Small Rodents. DHEW Pub. f(NIH) 76-801.
Pitts, L.L., Bruner, R.H., D'Addario, A.P., and Uddin, D.E. (1983).
Induction of Renal Lesions Following Oral Dosing with
Hydrocarbon Fuels. The Toxcologist 3^» 70.
Siegel, S. Nonparametric Statistics for the Behavioral Sciences.
McGraw-Hill Book Co., Inc., New York (1956).
A-25
-------
Steel, R.G.D. and Torrie, J.H. Principles and Procedures of
Statistics. McGraw-Hill Book Co., Inc., New York (1960).
Tarone, R.E., Chu, K.C., and Ward, J.H. (1981). Variability in the
Rates of Some Common Naturally Occurring Tumors in Fischer
344 Rats and (C57BL/6N X C3H/HeN)Fj (BSCSFj) Mice. J. Nat.
Cancer Inst. 6£, 1175-1181.
Thomas, D.G., Breslow, N., and Gart, J.J. (1977). Trend and
Homogeneity Analyses of Proportions and Life Table Data.
Computer Biomed. Res. _lp_, 373-381.
A-26
-------
Table 1
SPECIFICATIONS OF UNLEADED MOTOR GASOLINE
Research Octane No.
Motor Octane No.
(R+M)/2
Reid Vapor Pressure, Ibs.
Distillation, ASTM D-86
IBP
5
10
20
30
40
50
60
70
80
90
95
EP
Recovery
10% Evap., °F
50% Evap., °F
90% Evap., °F
API Gravity
Gum, ASTM D381, mg/gal
Sulfur, ppm
Phosphorus, g/gal
Lead, g/gal
Stability, hrs
HC Analysis, ASTM D1319
Aromatics
Olefins
Saturate
Benzene Content
Sample
Used in
Study
92.0
84.1
88.1
9-5
93
105
116
138
164
190
216
238
256
294
340
388
428
97%
112
211
331
60.6
1
97
<0.005
<0.05
24+
26.1 Vol. %
8.4 Vol. %
65.5 Vol. %
2.0%
Unleaded
Commercial
Average*
92.1
83.6
87.9
9.9
92
124
220
332
412
59.3
1
27%
7%
66%
1.0%'
* DuPont Road Octane Survey, Summer 1976
**" Average benzene content typical of U.S. gasolines,
A-27
-------
Table 2
DESIGN OF STUDY
Target
Group Designation Concentration
1 Chamber Control 0 ppm
II Low Concentration 50 ppm
111 Intermediate Concentration 275 ppm
IV High Concentration 1500 ppm
A-28
-------
Table 3
CHROMATOGRAPH OPERATING CONDITIONS
Gas Chromatograph:
Detector:
Co 1 umn:
Sample Loop Size:
Co1umn Temp.:
Detector Temp.:
Injector Temp.:
Air Flowrate:
N2 Flowrate:
H2 Flowrate:
Range:
Attenuation:
Chart Speed:
Varian 2400
Flame lonization
5' x 1/8 inch O.D.
Stainless Steel
1.5* OV-101 on 100/120
Mesh Chromosorb GHP
5 cc
200° C
270° C
250° C
300 ml/min
60 ml/min
30 ml/min
1C'"
1024 for 1500 ppm
8 for 275 ppm
64 for 50 ppm
2.5 cm/min for 1500 ppm
0.25 cm/min for 275
2.5 cm/min for 50 ppm
A-29
-------
Table V
TERMINATION TIMES FOR ANIMAL GROUPS
Group
No.
1
II
III
IV
103 weeks
10? weeks
Species/ No.
Sex at
Rat-M
Rat-F
Mouse-M
Mouse-F
Rat-M
Rat-F
Mouse-M
Mouse-F
Rat-M
Rat-F
Mouse-M
Mouse-F
Rat-M
Rat-F
Mouse-M
Mouse-F
• 23.9 months
« 2k. 7 months
of Animals
Initiation
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
109 weeks
1 1 3 weeks
Duration of
Exposure
(Weeks)
107
109
107
113
107
109
103
113
107
109
103
113
107
109
107
113
- 25.2 months
» 26. 1 months
A-30
-------
Table 5
TISSUES SELECTED FOR WEIGHING
brain thyroid/parathyroid complex
**
heart kidneys
*
liver pituitary
**
test is lung with trachea
* *
ovaries adrenals
Tissues weighed after fixation
Tissues weighed in toto, prior to dissection
A-31
-------
Table 6
TISSUES PREPARED FOR MICROSCOPIC EXAMINATION
Both Species*
gross lesions and tissue masses
(and regional lymph. noJes, if
possible)
blood smear (as required by the
pathologist)
mandibular lymph node
salivary gland
sternebrae, femur, or vertebrae
including marrow
thyroids
parathyroids
jejunum
colon
liver
gallbladder (mice)
prostate
testes
ovaries
lungs and
larynx
mainstream bronchi
nasal cavity
heart
esophagus
stomach
uterus
brain (three sections, including
frontal cortex and basal ganglia,
parietal cortex and thalamus,
and cerebellum and pons)
thymus
trachea
pancreas
spleen
kidneys
adrenals
urinary bladder
pituitary
spinal cord
eyes
Rats Only
optic nerve
Harder!an gland
Zymba1 g1 and
oral mucous membrane
duodenum
ileum
cecum
mammary gland
mesenteric lymph node
skeletal muscle
sciatic nerve
skin
epididymides
seminal vesicles
cervix
Fallopian tubes
head
*As recommended by NCI. 1976.
A-32
-------
Table 7
PRIMARY RENAL NEOPLASMS IN RATS
Test Group
I
II
ill
IV
Neoplasm
None
Carcinoma
Adenoma
Ca re i noma
Sarcoma
Carcinoma
*
Adenoma
Number of
Males
0
1
2
2
I
6
1
Neoplasms
Fema 1 es
0
0
0
0
1
0
0
Occurred in male rat at 18 months.
A-33
-------
o
o.
to
(A
>•
m
~ «
o u
3
O «»
~ O
JJ Q.
ra x
e uj
u
CO
0)
3
X
IU
A-34
-------
Figure 2
Histologic appearance of a renal carcinoma composed of epithelial
»
cells arranged in a tubulo-acSnar pattern. Note cellular
pleomorphism and anaplasia.
A-35
-------
APPENDIX 1 - COMPOSITION Ot- GASOLINE
The specifications used to define petroleum products such
as gasoline are directed towards performance characteristics, usually
stated in terms of physical properties; little attempt is made to
determine detailed chemical composition, as can be seen in the data
of Table 1 in the text. The gasoline used in the present study was
formulated by blending four refinery streams, as shown in Table 1A.
Table 1A
The antioxidant consisted of 76 percent 2,6-di-tertiary
butylphenol, with the remainder about equal parts of 2-tertiary
butylphenol and 2,4,6-tri-tertiary butylphenol. The metal de-
activator was a 50 percent solution of N, N'-disalicy?idene-1,2-
diaminopropane in commercial xylene. The concentration of 5 Ibs/
1000 bbl corresponds to approximately 20 ppm w/w or 1*» ppm w/v.
Like gasoline, the four refinery streams in Table 1A are
specified largely by physical parameters, with only minimal chemical
compositional information, as shown in Tables 2A, 3A, ^A, and 5A.
Tables 2A, 3A, *»A, 5A
The most detailed compositional information available on
the unleaded gasoline employed in this study, based on gas chroma-
tographic and mass spectrometric analyses, covers 151 compounds out
of over 5*»2 that are possible. These data are provided in Table 6A.
Table 6A
A-36
-------
The specific individual compounds identified as major
contributors in Table 6A are listed in Table 7A.
Table 7A
It should be noted that about 75 percent of the gasoime is
comprised of 42 of the compounds determined. In Table 1A, benzene
adjustment to approach 2 percent Is indicated, based on an infrared
analytical method. However, when the more precise gas chromatographic-
mass spectrometric analytical procedure was used to obtain the results
shown in Table 6A, the benzene content was estimated to be 1.69
percent. More recent re-analyses of the gasoline by an improved
method indicates that the actual benzene content was 1.80 to 1.96
percent, a satisfactory approximation to 2 percent.
We thank Richard W. King of Sun Tech, Inc., for providing
the detailed information on the chemical composition of the gasoline.
A-37
-------
Table 1A
Formulation of Unleaded Gasoline
Generic Stream
Light Catalytic Cracked Naphtha
Heavy Catalytic Cracked Naphtha
Light Catalytic Reformed Naphtha
Light Alky late Naphtha
Benzene added to bring to 2%
Butane added to increase Reid
CAS Number
6W-55-5
6W-5*-4
6A741-63-5
6k 751 -66-8
Vapor Pressure
Volume %
7.6
Mi. 5
21.3
22.0
0.8
3.8
plus:
Antloxidant 5 lbs/1000 bbl
Metal Deactivator 5 lbs/1000 bbl
1 Toxic Substances Control Act (TSCA) PL 3k-k6$: Candidate List of
Chemical Substances, Addendum I, Generic Terms Covering Petroleum
Refinery Processed Streams, January 1978.
A-38
-------
Table 2A
Specifications of Light Catalytic Cracked Naphtha
A complex combination of hydrocarbons produced by the
distillation of products from a catalytic cracking process. It
consists of hydrocarbons having carbon numbers predominately in the
range of C4 through C11 and boiling in the range of approximately -20
degrees C to 190 degrees C (-4 to 374 degrees F). It contains a
relatively large proportion of unsaturated hydrocarbons.
Range of
Tests Company Data*
Gravity, degrees API 50-75
Sulfur, weight % 0.02-0.3
Nitrogen, ppm 10-50
Reid Vapor Pressure, psia 2-12
Distillation (ASTH D-86), °F
IBP 80-125
10* 103-160
50% 152-265
30% 235-408
35% 240-430
EP 295-460
Paraffins, % 21-44
Olefins, % 15-68.5
Napthenes, % 10-16
Aromatics, % 6-28
Saturates, %
Jk
Based on data submitted by 11 companies.
A-39
-------
Table 3A
Spec If1 cations of Heavy Catalytic Cracked Naphtha
A complex combination of hydrocarbons produced by a distillation
of products from a catalytic cracking process. It consists of hydro-
carbons having carbon numbers predominantly in the range of C* through
C|2 and boiling In the range of approximately 65°C to 230° C (148°F
to 446°F). It contains a relatively large proportion of unsaturated
hydrocarbons.
Range of
Tests Company Data*
Gravity, °API 36 - 4?.1
Sulfur, wt. % 0.08- 0.3
Nitrogen, ppm 21 -110
Reid Vapor Pressure, psia 0.3 - 4.1
Distillation, °F
(ASTM D-86 Equiv.)
IBP 118 -275
}0% 245 -33,3
50* 324 -372
30% 388 -412
95% 412 -422
EP 420 -450
PONA, % by MS
Paraffins 22.8 - 32.7
Olefins 9-8 - 20.8
Naphthenes 10.6
Aromatics 45.0 - 56.6
Saturates 40.0
Aniline Pt., °F 64.0
MON (Clear) 77-6 - 81.3
RON (Clear) 85.0 - 90.8
Based on data submitted by 6 companies.
A-40
-------
Table 4A
Specifications of Light Catalytic Reformed Naphtha
A complex combination of hydrocarbons produced from the
distillation of products from a catalytic reforming process.
It consists of hydrocarbons having carbon number predominantly In
the range of C§ through Cjj and boiling in the range of approximately
35 degrees C to 190 degrees C (95 to 374 degrees F). It contains a
relatively large proportion of aromatic and branched chain hydrocarbons.
This stream may contain 10 vol. % or more benzene.
Range of
Tests Company Data"
Gravity, degrees API 40 - 59
Sulfur, weight %
Nitrogen, ppm
Reid Vapor Pressure, psia 3-7- 11
Distillation (ASTM D-86), °F
IBP 74 -149
10% 136 -225
50% 186 -299
90% 229 -360
95% 292 -381
EP 356 -J
Paraffins, % 28-55
Olefins, % 0 - 2.A
Napthenes, % 0.5- 4.4
Aromatics, % 30.9~ 69-9
Saturates, %
Benzene, vol. % 0.6-3.97
Based on data submitted by 9 companies.
A-41
-------
Table 5A
Specifications of Light Alkylate Naphtha
A complex combination of hydrocarbons produced by distillation
of the reaction products of isobutane with monoolefinic hydrocarbons
usually ranging in carbon numbers from Ci through Cj. It consists
of predominantly branched chain saturated hydrocarbons having carbon
numbers predominantly in the range of Cj through C\Q and boiling in
the range of approximately 90°C to 160°C (194°F to 320°F).
Tests
Gravity, °API
Sulfur, Wt. %
Nitrogen, ppm
Flash Pt., °F
Aniline Pt., °F
RVP, IDS.
Distillation, °F
(ASTM D-86)
IBP
10*
50*
90*
95*
EP
P, %
0, %
N, %
A, %
Saturates, *
RON (clear)
HON (clear)
Range of
Company Data
70. A -
0.002-
1.1
122
166
4.2 -
104
154
208
235
0.01
6.9
-120
-175
-230
-300
258 -335
99+
0.01 -
1.0
0.0 -
98.5
93.8 -
90.5
0.5
1.0
95.2
92.5
Based on data submitted by 3 companies.
A-42
-------
TabIt 6A
Detailed Composition of 6«toline
Compound Class
Alkanes
Isoalkanes
Total Isoalkanes
Cycloatkanes
Total Cycloalkancs
Alkenes
Total alkenes
•enzene
Alky benzenes
Total Al kyl benzenes
Indans/Tetratlns
Naphthalenes
Total Aromatic*
Alkanes
Isoalkanes
Cycloalkanes
Alkenes
Aromatic*
TOTAL
Carbon
Mo. Range
Cj thru C|Q
C4
C5
C6
C7
CB
°9
C|o thru C13
C* thru C)3
C5
- C6
C7
CB
°9
C!0 thru C13
Cj thru Cj3
C2
C3
C4
C5
c|
Cy thru Cu
C2 thru Cjj
C6
c?
CB
°9
CID
cj?
CI2
C* thru C|2
Cj thru 013
C,Q thru Cu
C^ thru Ci3
£3 thru C)0
Cit thru C)3
05 thru €13
Cj thru CJ2
C^ thru 013
No. of Isomers
Possible
8
1
2
4
8
17
3*
>75
>1*1
1
2
7
23
76
>76
»85
1
1
A
6
17
>128
><57
1
1
*
8
22
>22
»22
»36
targe
15
>51
8
>l«tl
>I85
»57
>51
>$42
Analyzed For
8
1
2
4
8
1*
22
"
51
1
2
7
16
23
•
*9
1
1
k
6
17
™
29
1
1
4
8
-
-
™
14
-
-
14
Summary
8
51
*9
29
14
151
A-43
Volume
« In Fuel
11.40
1.14
10.26
8.99
4.77
16.73
2.01
2.65
*6.55
0.15
1.05
1.03
0.7
-------
Table 7A
Identification of Major Contributors
Alkanes (3)
n-butane
n-pentane
n-hexane
Isoalkanes (17)
isobutane 2,2,4-trimethylpentane
Isopentane 2,3,'ftrimethylpentane
2-methylpentanc 2,3,3-trlmethylpentane
3-methylpentane 2,2,3-trlmethylpentane
2,3~dimethyl butane 2-methyloctane
2-methylhexane 3-methyloctane
3-methylhexane 4-methyloctane
2,3-dimethylpentane 2,2,5-trimethylpentane
2,A-dimethyl pentane
Cycloalkanes (5)
methylcyclohexane cyclopentane
l,cis, 3~dtmethylcyclopentane methylcyclopentane
1,trans, 3-dimethylcyclopentane
Alkenes (8)
propylene trans pentene-2
trans butene-2 els pentene-2
cis butene-2 2-methylpentene-1
pentene-1 2-methylpentene-2
Aromatics (9)
benzene p-xylene
toluene 1-methyl, 3-cthylbenzene
ethylbenzene 1-methyl, k- ethyl benzene
o-xylene 1,2,4-trimethylbenzene
m-xylene
A-44
-------
APPENDIX B
COMPARISON AMONG DIFFERENT EXTRAPOLATION MODELS
Three models used for low-dose extrapolation, assuming the independent
background, are:
Multistage: P(d) = 1 - exp [-(q^ + ... + qkdk)]
where q-j are non-negative parameters;
A + B ln(d)
Probit: P(d) = ; f(x) dx
where f(.) is the standard normal probability density function; and
Wei bull: P(d) = 1 - exp [-
where b and k are non-negative parameters.
The maximum likelihood estimates (MLE) of the parameters in the multistage
model is calculated by means of the program GLOBAL82, which was developed by
Howe and Crump (1982). The MLE estimates of the parameters in the probit and
Weibull models are calculated by means of the program RISK81, which was
developed by Kovar and Krewski (1981). Table B-l presents the MLE of parameters
in each of the four models.
B-l
-------
TABLE B-l. MAXIMUM LIKELIHOOD ESTIMATES OF THE PARAMETERS FOR
THE THREE EXTRAPOLATION MODELS BASED ON THREE DATA SETS
IN API UNLEADED GASOLINE STUDY
(International Research and Development Corporation 1983)
Data base
Kidney tumor
in male rats
Hepatocellular
carci noma/adenoma
in female mice
Mul ti stage
model
qx = 2.01 x ID'3
qp = 0
qj_ = 1.44 x 10'3
q2 = q3 = 0
Probit
model
A =
B =
A =
B =
-2.64
0.33
-3.29
0.52
b
k
b
k
Wei bull
model
= 6.42 x
= 0.68
= 5.15 x
= 0.78
io-3
10~3
Hepatocellular qx = 8.53 x 10'4 A = -3.98 b = 6.95 x 10~4
carcinoma in q2 = 3.83 x 10'8 B = 0.57 k = 1.04
female mice
B-2
-------
APPENDIX C
CARCINOGENIC POTENCY OF BENZENE
The carcinogenic potencies calculated in this appendix are to be used
only for determining the contribution of benzene content to the tumor response
observed in the unleaded gasoline vapor bioassay. In this endeavor, it is not
necessary to consider the species conversion factor where the data from the
gavage study are used to estimate cancer risk via the inhalation route.
Benzene has been shown to produce Zymbal gland carcinoma in rats (Maltoni
et al. 1982 and NTP 1983) and in male mice (NTP 1984) and hematopoietic neoplasms
in male mice (Snyder et al. 1980). These data are used to calculate the
carcinogenic potency for benzene. Preliminary calculation of the benzene
potency on the basis of the male mice data (not presented here) shows comparable
results. A complete document on benzene risk assessment, including human and
animal data, is currently being prepared by the CAG. In the present appendix,
only information relevant to the objectives stated above is presented. The
presentation has been kept as simple as possible because the time allocated to
this document is very limited. Details will be provided in the CAG risk assessment
document on benzene which is now in preparation.
In estimating the relative potency of benzene, the CAG has utilized the
following equivalency ratio:
1 ppm of benzene = 3,250 ug/m3
The volumetric breathing rate for a rat weighing 300 grams is 0.2 m^/day
(see section 5.4.1.3.2.1). For purposes of convenience, the body weight for
C-l
-------
all of the rats used in the Maltoni et al. (1982) and the NTP (1984) studies is
assumed to be 300 grams.
Thus, 1 ppm of benzene in air is calculated as being equivalent to
3,250 ug/m3 x 0.2 m3/day x 10-3 mg/Ug/0.30 kg = 2.17 mg/kg/day
Tables C-l to C-4 present the tumor incidence data that have been used
by the CAG to calculate the carcinogenic potency of benzene. The maximum
likelihood estimate of the parameters in the multistage model, and the 95%
upper-bound estimate, q*, of the linear component, are also presented at the
bottom of each table.
TABLE C-l. INCIDENCE OF ZYMBAL GLAND CARCINOMAS IN FEMALE
SPRAGUE-DAWLEY RATS ADMINISTERED BENZENE BY GAVAGE
(Maltoni et al. 1982)
Experimental dose
(mg/kg/day) Response
0 0/30
50 2/30
250 " 8/32
Remarks:
1. The number of animals surviving at the 26th week are used for the denomi-
nators.
2. Animals were treated by gavage 4 to 5 times a week for 52 weeks. The
lifetime dose is calculated by d x (4.5/7) x (52/104) = 0.32 d, where d
is the experimental dose.
3. The 95% upper-bound estimate of the linear component in the multistage model
is qj = 5.97 x 10~3/mg/kg/day or, equivalently, q| = 1.29 x 10'2/ppm
using the fact that 1 ppm of benzene in air is equivalent to a dose of
2.13 mg/kg/day. The maximum likelihood estimates of the parameters in the
multistage model are qi = 8.03 x 10-3/ppm, q2 = 0.
C-2
-------
TABLE C-2. INCIDENCE OF HEMATOPOIETIC NEOPLASMA IN
C57BL MALE MICE EXPOSED BY INHALATION
(Snyder et al. 1980)
Dose (ppm) Response
0 2/40
300 8/40
Remarks:
1. Mice were exposed by inhalation to 300 ppm of benzene 6 hours/day, 5 days/
week for 488 days, at which time all of the benzene-treated animals died.
2. Lifetime dose is calculated as d = 300 x (6/24) x (5/7) = 53.57 ppm.
3. Because the lifespans of the treated animals were shorter than those of
the controls, the risk calculated from the data is further adjusted by
multiplying by a factor of (630/488)3, where 630 days are assumed to be
the lifespan of the control mice. The carcinogenic potency of benzene is
calculated to be q| = 1.4 x 10'Vppm. The maximum likelihood estimate is
q = 6.9 x 10"3/ppm.
TABLE C-3. INCIDENCE OF ZYMBAL GLAND CARCINOMAS IN MALE RATS
(F344) ADMINISTERED BENZENE BY 6AVAGE
(NTP 1984)
Dose (mg/kg/day) Response
0 2/48
50 6/50
100 10/50
200 17/50
Remarks:
1. Benzene was administered by gavage 5 days/week for 103 weeks.
»
2. The lifetime dose is calculated by d x (5/7), where d is the experimental
dose.
C-3
-------
3. The carcinogenic potency of benzene is estimated to be:
q* = 3.64 x 10-3/mg/kg/day
or, equivalently,
q* = 7.90 x 10-3/ppm
The maximum likelihood estimates of parameters using the multistage model are:
ql = 5.16 x l(T3/ppm, q2 = 5.56 x l(T6/(ppm)2, q3 = 0
TABLE C-4. INCIDENCE OF ZYMBAL GLAND CARCINOMAS IN FEMALE RATS (F344)
ADMINISTERED BENZENE BY GAVAGE
(NTP 1984)
Dose (mg/kg/day) Response
0 0/50
25 5/50
50 5/50
100 14/49
Remarks:
1. The study design is the same as that described in Table B-3.
2. The carcinogenic potency of benzene is calculated as:
q* = 5.96 x 10-3/mg/kg/day
or, equivalently,
q* = 1.29 x 10-2/ppm
The maximum likelihood estimates of the parameters in the multistage model
are:
qi = 8.68 x 10-3/ppm, q2 = 0 and q3 = 5.56 x lO-7/(ppm)3.
C-4
-------
APPENDIX 0
REFERENCES
Alden, C.L., R.L. Kanerva, G. Ridder, and L.C. Stone. 1983. The patho-
genesis of nephrotoxicity of volatile hydrocarbons in the male rat.
G.P. Hemstreet III, J.J. Thorpe, and M.L. Kane, eds. Proceedings of
the Workshop on the Kidney Effects of Hydrocarbons, Boston, MA, July
18-20. American Petroleum Institute, pp. 154-170.
Armitage, P., and R. Doll. 1954. The age distribution of cancer and a multi-
stage theory of carcinogenesis. Br. J. Cancer 8:1-12.
Cox, C.B. 1972. Regression model of life tables. J. Roy Stat Soc. B
34:187-220.
Crump, K.S., and W.W. Watson. 1979. GLOBAL79: A Fortran program to extrapo-
late dichotomous animal carcinogenicity data to low doses. Natl. Inst.
Environ. Health Science Contract No. l-ES-2123.
Dripps, R.D., J.E. Eckenhoff, and L.D. Vandem. 1977. Introduction to
anesthesia: The principles of safe practice. 5th ed. Philadelphia, PA:
W.B. Saunders Company, pp. 121-123.
Federation of American Societies for Experimental Biology (FASEB). 1974. 2nd
ed. Vol. III. Philip L. Altman and Dorothy S. Dittmen, eds. Bethesda,
MD. Library of Congress No. 72-87738.
Freireich, E., E. Gehan, D. Rail, L. Schmidt, and E. Skipper. 1966. Quantita-
tive comparison of toxicity of anticancer agents in mouse, rat, hamster, dog,
monkey, and man. Cancer Chemotherapy Reports 50(4):219-244.
Goodman, D.G., J.M. Ward, R.A. Squire, K. C. Chu, and M.S. Linhart. 1979.
Neoplastic and nonneoplastic lesions in aging F344 rats. Toxicol. Appl.
Pharmacol. 48:237-248.
Haider, C.A., T.M. Warne, and N.S. Hatoum. 1983. Renal toxicity of gasoline and
related petroleum naphtha in male rats. G.P. Hemstreet III, J.J. Thorpe,
and M.L. Kane, eds. Proceedings of the Workshop on the Kidney Effects of
Hydrocarbons, Boston, MA, July 18-20. American Petroleum Institute, pp. 107-
129.
Howe, R.B., and K.S. Crump. 1982. GLOBAL82: A computer program to extrapolate
quantal animal toxicity data to low doses. OSHA Contract No. 41 USC252C3.
International Commission on Radiological Protection (ICRP). 1977. Recommenda-
tion of the ICRP. Pub. No. 26, adopted Jan. 17, 1977. Oxford, England:
Pergamon Press.
International Research and Development Corporation (IRDC), Mattawan, MI. 1982.
Motor fuel chronic inhalation study. 6 vols. Sponsored by the American
Petroleum Institute. Interim report.
D-l
-------
International Research and Development Corporation (IRDC), Mattawan, MI. 1983.
Motor fuel chronic inhalation study. 6 vols. Sponsored by the American
Petroleum Institute. Sept. 15, 1983.
Kitchen, D.N. 1983. Neoplasic renal effects of unleaded gasoline in Fischer
344 rats. G.P. Hemstreet III, J.J. Thorpe, and M.L. Kane, eds. Proceedings
of the Workshop on the Kidney Effects of Hydrocarbons, Boston, MA, July
18-20. American Petroleum Institute, pp. 92-102.
Kovar, J., and D. Krewski. 1981. RISK81: A computer program for low-dose
extrapolation of quantal response toxicity data. Department of Health
and Welfare, Canada.
MacFarland, H.N. 1983. Xenobiotic-induced kidney lesions: Hydrocarbons the
PS-6 90-day and 2-year gasoline studies. G.P. Hemstreet III, J.J. Thorpe,
and M.L. Kane, eds. Proceedings of the Workshop on the Kidney Effects
of Hydrocarbons, Boston, MA, July 18-20. American Petroleum Institute,
pp. 68-74.
MacFarland, H.N., C.E. Ulrich, C.E. Holdsworth, D.N. Kitchen, W.H. Halliwell, and
S.C. Blum. 1984. A chronic inhalation study with unleaded gasoline vapor.
Accepted by J. Amer. College of Toxicology.
MacNaughton, M.G., and D.E. Uddin. 1983. Toxicology of mixed distillates and
high-energy synthetic fuels. G.P. Hemstreet III, J.J. Thorpe, and M.L.
Kane, eds. Proceedings of the Workshop on the Kidney Effects of Hydro-
carbons, Boston, MA, July 18-20. American Petroleum Institute, pp. 171-185.
Maltoni, C., B. Conti, G. Cotti, A. Mandrioli, C. Scarnato, and L. Valgimigli.
1982. Benzene: An experimental multipotential carcinogen. Institute
of Oncology, Bologna and Bentivoglio, Italy.
Mantel, N., and M.A. Schneiderman. 1975. Estimating "safe" levels, a hazar-
dous undertaking. Cancer Res. 35:1379-1386.
National Council on Radiation Protection and Measurements (NCRP). 1980.
Influence of dose and its distribution in time on dose-response
relationships for low-LET radiation. NCRP Report No. 64. Washington,
D.C.
National Toxicology Program. 1983 (Oct. 19). NTP Technical report on the
toxicology and carcinogenesis studies of benzene (CASNO 71-43-2). Board
draft.
Rushton, L., and M.R. Alderson. 1982. An epidemiology survey of oil
distribution centers in Great Britain. Unpublished.
Snyder, C., B. Goldstein, A. Sellakumar, I. Bromberg, S. Lasking, and R. Albert.
1980. The inhalation toxicology of benzene: Incidence of hematopoietic
neoplasms and hematotoxicity in AKR/J and C57BL/6J mice. Toxicol. Appl.
Pharmacol. 54:323-331.
D-2
-------
Thomas, T.L., P. Decoufle, and R. Moure-Eraso. 1980. Mortality among workers
employed in petroleum refining and petrochemical plants. J. Occup.
Med. 22:97-103.
Thomas, T.L., R.J. Waxweiler, R. Moure-Eraso, S. Itaya, and J.F. Fraumeni.
1982. Mortality patterns among workers in three Texas oil refineries.
J. Occup. Med. 24:135-141.
Universities Associated for Research and Education in Pathology, Inc. (UAREP),
Bethesda, MD. 1983 (Dec.). Hydrocarbon toxicity: Acute, subchronic and
chronic effects in relation to unleaded gasoline exposure of rodents, with
comments on the significance to human health. Prepared for the American
Vesselinovitch, S.D., K.V.W. Rao, and N. Mihailovich. 1975 (Oct.). Factors
modulating benzidine carcinogenicity bioassay. Cancer Res. 35:2814-2819.
Whittemore, A. 1978. Quantitative theories of carcinogenesis. In:
Advances in cancer research, Vol. 27. New York: Academic Press, pp. 55-58,
Whittemore, A., and B. Keller. 1978. Quantitative theory of carcinogenesis.
Society for Industrial and Applied Mathematics Review 20:1-30.
D-3
------- |