RISK ASSESSMENT
Risky Business: It’s a Gas
Assessing Risks of Gasoline Exposure
BY FRANKLIN MIRER
Now that we can look at something other
than the price on a gas pump, maybe you
have noticed this label on the side:
“Warning, Prolonged High Level Exposure May Cause Serious Health Problems,
Long Term Exposure Has Caused Cancer
in Laboratory Animals.” The label appears at some stations and not at others.
Someone (ok, it was me) coined the
term “Houdini risk assessment” for analyses that make a chemical’s danger disappear or enable an industry to escape
significant control costs. The last Risky
Business, in the April 2009 Synergist,
showed the Houdini application to potency
(exposure-response assessment) estimates
for formaldehyde. This column addresses
the application of Houdini to the hazard
identification step for gasoline. In this
case, the eraser of risk is a hypothesis that
explains the presence of kidney tumors in
rats exposed to gasoline fumes as a function of alpha-2µ-globulin, a protein not
found in people. Therefore, according to
this argument, the male rat kidney tumors
are “not relevant” to people and should
not be used for carcinogenic hazard identification. In the rest of this column, I’ll
refer to this hypothesis as “alpha-2.”
Classifying Gasoline
Were gasoline—and related aliphatic petroleum hydrocarbons—considered a human
carcinogen, we’d certainly wonder whether
the ACGIH® 300 ppm TLV® was protective
enough, and our views of the extent and
impact of exposure to carcinogens would
change. The carcinogenicity of gasoline,
which is a mixture of mostly aliphatic
straight chain and cyclic hydrocarbons up
to C12, also would call into question petroleum solvents—naphtha, Stoddard, etc.—
that contain C10 hydrocarbons. Terpenes
such as limonene (
citro-solve, lemon oil) and
beta-myrcene (essential
oil in personal care products), not to mention the
ubiquitous turpentine, are
also C10 hydrocarbons.
(Fans of organic chemistry
should look up the structures of these natural
products.) And many
other solvents, like perchloroethylene and methyl
isobutyl ketone (MIBK),
cause the rat kidney tumors which are the subject of this argument.
The International Agency for Research
on Cancer (IARC) website lists gasoline as
“possibly carcinogenic to humans” (group
2B). This classification is based on a 1989
review citing limited data in animals and
inadequate data in people (but supported
by mechanistic data regarding liver cancer). The full text isn’t available yet—
IARC is in the process of making older
monographs accessible online—but a
summary is posted at http://mono
graphs.iarc.fr/ENG/Monographs/vol45/
volume45.pdf. The Agency for Toxic
Substances and Disease Registry
(ATSDR) website includes a profile for
gasoline and notes that it is one of the
most requested compounds. According
to the ATSDR, the cancer risk of gasoline
is “not determined.” Gasoline is not
listed on the U.S. Environmental Protection Agency’s Integrated Risk Information System (IRIS) site. OSHA and NIOSH
mention gasoline as a potential occupational carcinogen, but no PEL or REL
exists. OSHA PELs for naphtha and
Stoddard solvent are 500 ppm.
Relevant Studies
The story starts with the 1984 publica-
tion of a bioassay sponsored by the
American Petroleum Institute (API)
(MacFarland 1984). The abstract notes,
“The most interesting observations were
. . . renal carcinomas or sarcomas, in . . .
male rats at all dose levels, with some ev-
idence of a dose-response relationship.”
The full text of the paper is not available
on the Internet. The abstract omits men-
tion of the excessive mouse liver tumors.
Three years later, EPA published a risk
assessment titled “Evaluation of the Car-
cinogenicity of Unleaded Gasoline,” which
noted the increased liver cancer among
female mice exposed to gasoline in the API
study. The EPA document also described
increased tumors in male rats at 2,056 ppm
(and perhaps at 292 ppm if appropriate
statistics had been applied). The EPA calcu-
lated a cancer risk of 3. 5 x 10-3 per ppm
based on the rat tumors. The mouse tumors
likely would yield a slope factor one-tenth
that rate, which is still a significant risk at
10 ppm (compared to the 300 ppm TLV).
