Strengthening the Foundation of Risk Assessment
The growing controversy over the risk assessment policies of U.S. regulatory
agencies is emotional and divisive. At the root of the problem is the huge
magnitude of uncertainty that often accompanies risk estimates for a given
exposure level of an environmental agent. This uncertainty is created by
weaknesses and/or inadequacies in the available scientific data and the
difficulties in translating biological information into mathematics. Reliance
on default assumptions that are open to legitimate criticisms fuels the
controversy. Regulatory agencies are often the victims in risk assessment
wars because they must make regulatory decisions whether or not appropriate
scientific information is available and ensure that these decisions protect
the public from adverse health effects.
NIEHS has a long record of accomplishment, through both extramural and
intramural programs, on the mechanisms whereby chemicals cause disease and
on the relationship between chemical exposure and adverse health outcomes
in humans. The National Toxicology Program, the nation's most comprehensive
toxicity testing program, is centered at NIEHS and has contributed substantially
to the hazard identification component of risk assessment. The major goals
of NIEHS in risk assessment are 1) to strengthen the scientific foundation
on which risk assessments are based by increased understanding of mechanisms
and dose-response relationships, the identification of sensitive subpopulations,
and evaluation of the relevance of animal models for estimating human risks;
2) to develop novel and more reliable approaches to estimate human risks;
3) to collaborate with regulatory agencies on conducting risk assessments;
and 4) to communicate findings to the public in an understandable and objective
way.
Laboratory of Biochemical Risk Analysis
Although many components of NIEHS conduct research that is directly relevant
to various aspects of risk assessment, the Laboratory of Biochemical Risk
Analysis (LBRA) has served as the focus for the development and application
of laboratory approaches relevant to risk assessment. George Lucier has
been chief of LBRA since its inception in 1984. Many notable contributions
have been made by LBRA scientists, but the most visible is research on dioxins,
genetic susceptibility, and biomarkers. The work on dioxin has addressed
dose-response relationships and comparison of human and rodent responses.
The dose-response studies have demonstrated that the response to different
levels of exposure to dioxin cannot be predicted solely on the basis that
the response is receptor mediated. It is generally accepted that most, if
not all, of dioxin's effects require interaction with a cellular protein
called the Ah receptor. The Ah receptor appears to function like receptors
for steroid hormones. In a series of papers by Lucier and co-workers Angelika
Tritscher, Charles Sewall, Jack VandenHeuvel, and George Clark, evidence
was presented that the amount of dioxin required to activate cellular enzymes
and affect growth factors is linearly related to concentrations of dioxins
in certain body tissues. It is also clear that ovarian hormones, probably
estrogens, are necessary for dioxin-mediated liver cancer; the model dioxin,
2,3,7,8-tetrachloro-p-dioxin (TCDD), promotes liver tumors in intact
rats but not in rats from which the ovaries have been removed.
George Clark--researching
sensitivity to dioxins. NIEHS |
In contrast to data on the activation of cellular enzymes, TCDD's effects
on proliferation of liver cells and growth of existing cancer cells are
not strictly proportional to the dose of TCDD. Much of the dose- response
work has been conducted on liver, but the NIEHS work has recently demonstrated
that the mechanism responsible for TCDD-mediated lung cancer is different
from that for liver cancer; ovarian hormones are necessary for liver cancer
but protect against lung cancer. This finding is especially relevant in
light of several studies which demonstrated that dioxin exposure in the
workplace is associated with increased risk of respiratory tumors.
LBRA scientists are now attempting to characterize the factors that control
dose- response relationships for different effects mediated by the Ah receptor.
LBRA molecular dosimetry studies are now using sensitive methods such as
reverse-transcriptase polymerase chain reaction to detect dioxin-mediated
changes in gene expression for exposure levels encountered by the overall
population.
One of the important questions concerning risk assessment for dioxins
is the sensitivity of humans to these compounds. This issue is being addressed
by a large multilaboratory collaborative study that includes George Clark
and Angelika Tritscher of LBRA, Neil Caporaso of the National Cancer Institute,
Larry Needham of CDC; Maria Teresa Landi and Pier Bertazzi of the University
of Milan, Paolo Mocarelli and Paolo Brambilla of Desio Hospital, Detlev
Jung of the University of Mainz, Lutz Edler of the University of Heidelburg,
George Lambert of Loyola University, Oliver Hankinson of the University
of California-Los Angeles, and Linda Birnbaum and Dagmar Lang of EPA. One
goal of the study is to identify human genetic markers for susceptibility
to the toxic and carcinogenic effects of dioxins. By characterizing the
receptor and the changes in gene expression, markers may be identified that
correlate with adverse human health effects.
For this study, two cohorts of people exposed to high concentrations
of dioxins have been assembled. One of the cohorts is from Seveso, Italy
and was exposed to high levels of dioxins after a chemical plant explosion
in 1976. The second cohort includes workers exposed occupationally at a
chemical plant synthesizing 2,4,5-trichlorophenol and other chemicals contaminated
with dioxins. Within these cohorts certain individuals developed chloracne,
a skin lesion that is a response to dioxin exposure, whereas others exposed
to similar concentrations did not develop chloracne. Analyzing the genetic
and biochemical differences in these people and correlating the results
to other health effects should provide insight into the sensitivity of humans
to dioxins. LBRA scientists are working closely with NIEHS's new Laboratory
of Quantitative and Computational Biology to develop dose-response models
for the cancer and noncancer effects of dioxins, described later in this
section.
Douglas A. Bell is carrying out studies on human genetic susceptibility
to cancer-causing agents. Inherited variability in the ability to detoxify
carcinogens has been associated with increased cancer susceptibility. Presumably,
individuals who carry high-risk genotypes suffer more genetic damage as
a result of chemical exposure, and this damage translates into greater risk
of developing cancer.
NIEHS investigators have developed sensitive assays to detect susceptibility
to carcinogens in cigarette smoke, foods, industrial by-products, and environmental
pollution. Based on tests of more than 1000 individuals for these susceptibility
genes, the frequency of the at-risk genotypes vary significantly among Asians,
Caucasians, and African-Americans. Such variation suggests that some of
the differences in cancer incidence among ethnic groups may be due to genetic
differences as well as exposure differences.
In collaboration with Jack Taylor of the NIEHS Epidemiology Branch, LBRA
is testing the effect of certain cancer susceptibility genes in studies
of bladder cancer, lung cancer, and liver cancer. Individuals who carry
the at-risk genotype for glutathione transferase µ, an enzyme that
detoxifies constituents of cigarette smoke, suffer a 70% increased risk
of bladder cancer. In ongoing studies with researchers at the National Cancer
Institute, Columbia University, University of North Carolina, and University
of Keele, England, NIEHS is exploring how genetic variability in the metabolism
of carcinogens affects risk for cancer of the bladder, lung, liver, stomach,
colon, head, and neck.
Joyce Goldstein--isolating
P450 genes. NIEHS |
The cytochrome P450 enzymes catalyze the oxidation of drugs, carcinogens,
and other xenobiotics. Joyce Goldstein's group in LBRA is looking for genetic
defects in these enzymes that affect the ability of humans to metabolize
chemical agents. Population studies have shown that some people are poor
metabolizers of the drug S-mephenytoin, and defective metabolism
is inherited. Metabolism of other drugs, including barbiturates, the antimalarials,
and the antiulcer drug omeprazole, may be mediated by the same enzyme. Goldstein's
group has isolated two new genes in the P450 subfamily. These studies use
conventional cloning techniques and polymerase chain reaction to identify
genetic differences in genes from poor or extensive metabolizers of the
drugs. The ability of proteins coded by these genes to metabolize drugs
and chemicals is being studied by Burhan Ghanayem of LBRA.
Laboratory of Quantitative and Computational Biology
High-speed computing, the improved ability to collect a broad range of
data at the biochemical and molecular levels, and recent advances in mathematics
and statistics have significantly enhanced the utility of mathematical modeling
in describing and studying environmental risks. The emergence of these new
technologies requires a multidisciplinary approach in the quantitative sciences
and creates the need for research teams in quantitative and computational
areas. To address this need, NIEHS is creating a new laboratory, the Laboratory
of Quantitative and Computational Biology (LQCB). The primary responsibility
of LQCB is to investigate the application of mathematics, statistics, computational
chemistry, electrical engineering, and computer science to the understanding
of human health risks from exposure to environmental agents. LQCB will initially
combine scientists from three research groups at NIEHS: a group developing
new methodology and applying existing methodology to the application of
quantum and statistical mechanics in environmental health, a group focusing
on research in mathematical modeling aimed at developing new methodologies
for risk estimation, and the NIEHS Scientific Computing Laboratory, whose
primary purpose is computational support and direction for intramural research
at NIEHS. In addition to these scientists, the LQCB plans to add expertise
in a variety of related fields, including artificial intelligence and virtual
reality. LQCB will be able to model biological mechanisms at all levels
of complexity from molecular to demographic. The collaboration of LQCB and
other NIEHS branches will lead to more efficient use of NIEHS resources
through improved experimental design and the formal use of data from multiple
sources. Current research and short-term research plans for the LQCB can
be divided into seven broad areas: carcinogenic modeling, molecular modeling,
biochemical and pharmacological modeling, modeling noncarcinogenic endpoints,
computer science, artificial intelligence, and risk communication. Christopher
Portier will be acting chief of this new laboratory.
Challenging a Dioxin Hypothesis
Dioxin is believed by many to be one of the most potent carcinogens in
the environment. Emerging information at the molecular level concerning
the mechanism of dioxin's toxicity has reignited controversy regarding safe
exposure levels for dioxin and prompted the reassessment by the EPA. As
part of these efforts, scientists at NIEHS developed mechanistic models
to challenge the dioxin threshold hypothesis.
At the heart of the dioxin controversy is a proposed hypothesis that
toxic effects of dioxin are receptor mediated, and at sufficiently low exposure
to dioxin (i.e., a threshold), too few receptors would be occupied to produce
a significant biological consequence. Researchers Christopher Portier and
Michael Kohn of the Laboratory of Quantitative and Computational Biology,
in collaboration with the Laboratory of Biochemical Risk Analysis and the
Biostatistics Department of the German Cancer Research Center, used data
on the effects of dioxin, including tissue concentrations, changes in expression
of liver proteins, modification of plasma membrane epidermal growth factor,
interactions with estrogens, cellular proliferation, and carcinogenesis,
to create a comprehensive mechanistically based model of dioxin's effects.
A clonal two-path/two-stage model of carcinogenesis
is being developed by members of the LQCB.
Portier and Kohn found that although the classical receptor-mediated
models theoretically allowed for both nonlinear behavior that mimics a threshold
and linearity at low doses, these models failed to predict a nonlinear relationship
at low doses. The models predict that binding of dioxin to the Ah receptor
follows linear kinetics at low doses, and induction of the Ah receptor by
the dioxin-Ah receptor complex does not alter this curvature. Binding of
dioxin to other liver proteins does not seem to significantly affect the
dose-response curve for expression of any of the proteins modeled. Not only
were the NIEHS models unable to detect any nonlinearity in cell kinetics,
they also indicated that dioxin potentially produces premalignant lesions
in the liver.
Michael Kohn--modeling
effects of dioxin. NIEHS |
Because changes in gene expression do not necessarily predict toxicity,
current studies are attempting to develop dose-response models to determine
if the toxic effects of dioxin exhibit linear or nonlinear behavior. For
example, Portier and Kohn have undertaken a theoretical analysis of the
impact of receptor-based models on the shape and magnitude of tumor incidence
rates.
EPA Reevaluation of Dioxin's Risks
In 1991, then EPA administrator William Reilly initiated a reevaluation
of dioxin's risks. George Lucier and Christopher Portier have been involved
in this reevaluation in a number of ways. Lucier and Michael Gallo (EOHSI)
co-chair the committee that prepared the dose-response models chapter, the
cornerstone of the reevaluation. Portier played a key role in the development
of biologically based dose-response models for dioxin's effects. Lucier
also prepared the chapter on animal carcinogenicity. The dose-response models
and animal carcinogenicity chapters received favorable reviews from the
EPA Peer Review process in September 1992. The Scientific Advisory Board
will review the background papers and the risk characterization in late
1993.
Species Differences in Butadiene Carcinogenesis
1,3-Butadiene, a gaseous hydrocarbon used in the production of synthetic
rubber and other resins, is a carcinogen in rodents and is associated with
leukemia and lymphoma in humans. Mice develop tumors at lower exposures
to butadiene than rats. Recent studies have shown that mice have a higher
capacity to oxidize butadiene to 1,2-epoxy-3-butene, a mutagenic and carcinogenic
compound, than either rats or humans. Some investigators have concluded
that species differences in tumor development are due to differences in
metabolic activation of butadiene and detoxification of expoxide intermediates.
To validate this conclusion, two NIEHS scientists, Michael Kohn and Ronald
Melnick, constructed physiologically based pharmacokinetic models of the
distribution and clearance of inhaled butadiene in mice, rats, and humans.
In contrast to the conclusions of earlier investigators, the models predict
that species differences in the uptake of butadiene and the blood concentration
of epoxybutene are much more sensitive to the physiological parameters (e.g.,
ventilation rate and cardiac output) than to the biochemical parameters.
In addition, the model predicts that, because of these physiological differences,
butadiene accumulates in the fat of humans, but not mice, on repeated exposure.
According to the model, butadiene released from fat during the periods between
exposures continues to be converted into epoxybutene, adding to the carcinogenic
risk.
Ronald Melnick--modeling
butadiene. NIEHS |
In a recent editorial in Science, it was stated that after exposure
to 10 ppm butadiene in the ambient air, blood epoxybutene levels are 590
times higher in mice than in monkeys. Yet, in Kohn and Melnick's models,
mice produce only 5.5 times as much epoxybutene as humans at exposures that
result in equivalent amounts of butadiene absorbed into the body. Risk assessments
of inhaled carcinogens are normally performed on the basis of internal dose
rather than on the basis of atmospheric concentration. Inhalation studies
can lead to different implications relevant to human risk depending on the
manner in which the results are reported.
Computed epoxybutene concentrations, by themselves, were found not to
correlate with tumor incidence in mice and rats. Rats exposed to 1000 ppm
butadiene generate about twice the concentration of epoxybutene in lung
as mice exposed to 60 ppm. Yet mice develop lung tumors under those conditions
and rats do not. Kohn and Melnick conclude that other biochemical processes
(e.g., formation of DNA adducts, efficiency of DNA repair) not included
in their models are more important determinants of the differential response
of the two species than the concentration of the putative carcinogen.
Risk Assessment Seminar Series
As an adjunct to its initiatives in risk assessment, NIEHS is hosting
a seminar series featuring prominent scientists in the risk assessment field.
The first seminar was June 8, with Christopher Portier, chief of the NIEHS
Laboratory of Quantitative and Computational Biology. Other speakers in
the series will include David P. Rall, internationally recognized environmental
health researcher and retired director of NIEHS; Joe Rodricks of Environ,
a Washington, DC firm; John Graham of the Harvard Institute of Risk Assessment;
William Farland and John Vandenberg of U.S. EPA; Henry Falk of CDC; Ellen
Silbergeld of the University of Maryland at Baltimore and the Environmental
Defense Fund; Leslie Staynor of NIOSH; Gil Omenn of the University of Washington;
and Roger McClellan of the Chemical Industry Institute of Toxicology.
The series is designed to allow professionals in risk assessment to discuss
critical issues. For information on the series contact George Lucier, (919)
541-3802.
Last Update: August 21, 1998