Pharmacogenetics: detecting sensitive populations

Model-based translation of DNA damage signaling dynamics across cell types

Interindividual variability in DNA damage response (DDR) dynamics may evoke differences in susceptibility to cancer. However, pathway dynamics are often studied in cell lines as alternative to primary cells, disregarding variability. To compare DDR dynamics in the cell line HepG2 with primary human hepatocytes (PHHs), we developed a HepG2-based computational model that describes the dynamics of DDR regulator p53 and targets MDM2, p21 and BTG2. We used this model to generate simulations of virtual PHHs and compared the results to those for PHH donor samples. Correlations between baseline p53 and p21 or BTG2 mRNA expression in the absence and presence of DNA damage for HepG2-derived virtual samples matched the moderately positive correlations observed for 50 PHH donor samples, but not the negative correlations between p53 and its inhibitor MDM2. Model parameter manipulation that affected p53 or MDM2 dynamics was not sufficient to accurately explain the negative correlation between these genes. Thus, extrapolation from HepG2 to PHH can be done for some DDR elements, yet our analysis also reveals a knowledge gap within p53 pathway regulation, which makes such extrapolation inaccurate for the regulator MDM2. This illustrates the relevance of studying pathway dynamics in addition to gene expression comparisons to allow reliable translation of cellular responses from cell lines to primary cells. Overall, with our approach we show that dynamical modeling can be used to improve our understanding of the sources of interindividual variability of pathway dynamics.

Susceptibility to develop cancer varies among people, partially due to differences in genetic background. Ideally, healthy human-derived cells are used to investigate intracellular signaling pathways and their interindividual variability contributing to cancer susceptibility. Because cells from healthy human tissue are difficult to obtain and culture for periods longer than a few days, cell lines are often used as substitute. However, it is unclear to what extent signaling dynamics in cell lines represent dynamics in healthy human tissue. We asked whether we could reproduce interindividual variability in DNA damage response gene expression in a set of 50 human liver cell donors. Therefore, we built a mathematical model that simulates temporal expression dynamics of the DNA damage response in the HepG2 liver cell line upon chemical activation and used the simulations to create virtual donors. Our virtual donors displayed similar relations between genes as the samples from human donors, provided that we adjusted the strength of specific molecular interactions. Thus, our approach can be used to examine the applicability of widely used cell systems to healthy human tissue in terms of their dynamic responses.

[Childhood leukemia: a genetic disease!]

Cancer affects 1 in every 500 children before the age of 14. Little is known about the etiology of this heterogeneous group of diseases despite the fact that they constitute the major cause of death by disease among this population. Because of its relatively higher prevalence, most of the work done in pediatric oncogenetics has been focused on leukemias, particularly acute lymphoblastic leukemia (ALL). Although it is now well accepted that genetic variations play a significant role in determining individual's cancer susceptibility, few studies have explored genetic susceptibility to childhood leukemia with respect to polymorphisms. The main biological mechanisms contributing to cancer susceptibility can be grouped into broad categories : (1) cellular growth and differentiation, (2) DNA replication and repair, (3) xenobiotic metabolism, (4) apoptosis, (5) oxidative stress response and (6) cell cycle. To evaluate whether candidate genes in these pathways are involved in childhood leukemogenesis, we conducted association studies. We showed that leukemogenesis in children may be associated with genetic variants and that the combination of genotypes seems to be more predictive of risk than either of them independently. These results indicate that the genetic investigation of several enzymes (or metabolic pathways) is needed to explain the physiopathology of childhood leukemia because of the complexity of the environment and that of the inter-individual variability in cancer susceptibility.

Oestrogen receptor beta over expression in males with non-small cell lung cancer is associated with better survival

BACKGROUND

Adenocarcinoma of the lung is more frequent in females than in males and the association with smoking is less pronounced than for the other histological subtypes of lung cancer. Oestrogen induction of cell proliferation has been found in breast adenocarcinomas, and since oestrogen receptors (ER) have been demonstrated in lung tumours, a similar role of oestrogens in the development of lung cancer has been suggested. We examined the expression of ERalpha, ERbeta and progesterone in a well defined cohort of patients with NSCLC with more than 15 years of follow up, and related the results to gender and survival.

METHODS

Paraffin embedded, histological material was collected from 104 patients (71 men and 33 women), operated in the period 1989-1992 for NSCLC (56 squamous cell carcinomas, 40 adenocarcinomas and 8 large cell carcinomas). ERalpha, ERbeta and progesterone were immunohistochemically analysed. Staining frequency and intensity was scored semi-quantitatively. A tumour was defined as positive when more than 10% of the tumour cells were positive with at least a weak nuclear staining. Kaplan-Meier survival curves were generated to evaluate the significance of ERalpha, ERbeta and progesterone expression for the prognosis.

RESULTS

ERbeta positivity was demonstrated in 69% (72 of 104) of the tumours. There was no statistically significant correlation between ERbeta positivity and age, gender, stage, or histology. After adjusting for gender, age, stage at diagnosis and histology there was no difference in survival between subjects with ERbeta positive and ERbeta negative tumours. When stratifying by gender women with ERbeta-negative tumours had a non-significant (P=0.26) decrease in mortality compared with women with ERbeta positive tumours. In contrast, men with ERbeta positive tumours had a significantly reduced mortality (P=0.035) compared to men with ERbeta negative tumours. Using multivariate regression analysis the interaction between gender and positive ERbeta staining was the only significant prognostic factor. There was no correlation between the ERalpha immunohistochemical staining and any of the clinical variables, including survival. None of the 104 patients had tumours positive for progesterone.

CONCLUSION

The presence of ERbeta in a tumour seems to be a positive prognostic factor for men with non-small cell lung cancer. The finding confirms another recent study and suggests that the relation between oestrogens and lung cancer be investigated further.

Challenges identifying genetic determinants of pediatric cancers--the childhood leukemia experience

Pediatric cancers affect approximately 1 in every 500 children before the age of 15. Little is known about the etiology of this heterogeneous group of diseases despite the fact they constitute the major cause of death by disease among this population. Because of its relatively high prevalence, most of the work done in pediatric oncogenetics has been focused on leukemias, particularly acute lymphoblastic leukemia (ALL). Although it is now well accepted that genetic variation plays a significant role in determining individual's cancer susceptibility, few studies have explored genetic susceptibility to childhood leukemia with respect to common polymorphisms. The biochemical and genetic mechanisms contributing to cancer susceptibility are numerous and can be grouped into broad categories: (1) cellular growth and differentiation, (2) DNA replication and repair, (3) metabolism of carcinogens (4) apoptosis, (5) oxidative stress response and (6) cell cycle. To evaluate whether candidate genes in these pathways are involved in childhood leukemogenesis, we conducted case-control studies. We showed that leukemogenesis in children may be associated with DNA variants in some of these genes and that the combination of genotypes seems to be more predictive of risk than either of them independently. We also observed that, at least at some loci, the parental genetics might be important in predicting the risk of cancer in this pediatric model of a complex disease. Taken together, these results indicate that the investigation of a single enzyme and/or a single genotype might not be sufficient to explain the etiology of childhood leukemia because of the complexity of the environment and that of the inter-individual variability in cancer susceptibility.

Pharmacogenetic screening and therapeutic drugs

BACKGROUND

Pharmacogenetics is the science of the influence of heredity on pharmacological response.

ISSUES

The cost of severe adverse drug reactions in individuals has been estimated in the US alone to be in excess of US$4 billion. It has been argued that in a significant proportion of cases, the efficacy and toxicity profiles of drug therapy would be substantially improved in individuals if characteristics due to genetic variation were taken into account. Methods are now available, which make screening for susceptibility feasible.

CONCLUSIONS

There are several therapeutic areas in which screening may give rise to significant improvements in outcome with cost-benefits to both the individual and the community. However, there is currently a lack of data on which cost-benefit analysis can be based. The challenge is to provide this information for new drugs, and for drugs with established therapeutic roles.

Usefulness of genetic susceptibility and biomarkers for evaluation of environmental health risk

Recent attention is focused on understanding the genetic basis for individual susceptibility to the development of chronic disease. An emphasis is concentrated on establishing an association between inheritance of polymorphic chemical metabolizing genes and development of environmental cancer (e.g., lung cancer among cigarette smokers). The early reports of such associations have been very encouraging. However, some reported positive associations were not substantiated in subsequent studies using larger sample sizes and different ethnic populations. In this review, some confounding factors that contribute to the discrepancies are presented (e.g., ethnic-dependent distribution of variant gene alleles, differential expression of metabolizing genes, and inadequate study design). It is possible that the precision of the association can be improved if the mentioned investigations are complemented with concurrent studies of biological activities/effects. The usefulness of integrating metabolic susceptibility with biomarker measurement for understanding the development of lung cancers is presented. The importance of using adequate sample size and experimental design is emphasized. Development of a reliable approach for prediction of environmental disease not only will provide fundamental information regarding the genetic basis of human disease but will be useful for reducing disease burden in the population and for advancing patient care. Environ. Mol. Mutagen. 37:215-225, 2001. &copy 2001 Wiley-Liss, Inc.

Variant metabolizing gene alleles determine the genotoxicity of benzo[a]pyrene

Understanding the mechanisms involved with genetic susceptibility to environmental disease is of major interest to the scientific community. We have conducted an in vitro study to elucidate the involvement of polymorphic metabolizing genes on the genotoxicity of benzo[a]pyrene (BP). Blood samples from 38 donors were treated with BP and the induction of sister chromatid exchanges (SCE) and chromosome aberrations (CA) were evaluated. The latter is based on the tandem-probe fluorescence in situ hybridization (FISH) assay. The data indicate that the induction of genotoxicity was clearly determined by the inherited variant genotypes for glutathione-S-transferase (GSTM1) and microsomal epoxide hydrolase (EH). In a comparison of the two biomarkers, the CA biomarker shows a more definite association with the genotypes than does SCE. For example, the presence of the GSTM1 null genotype (GSTM1 0/0) is responsible for the highest level and significant induction of CA, irrespective of the presence of other genotypes in the different donors. This effect is further enhanced significantly by the presence of the excessive activation EH gene allele (EH4*) and decreased by the reduced activation EH gene allele (EH3*). Overall, the modulation of genotoxicity by the susceptibility genotypes provides support of their potential involvement in environmental cancer. Furthermore, the data indicate that the variant enzymes function independently by contributing their metabolic capability toward the expression of biologic activities. Therefore, studies like this one can be used to resolve the complexity of genetic susceptibility to environmental disease in human.

Genetic susceptibility to childhood acute lymphoblastic leukemia

Acute lymphoblastic leukemia (ALL) is the most common pediatric cancer. The origin of this disease can be explained by a combination of genetic susceptibility factors and environmental exposures. For the purpose of our study it can be considered as a complex disease, caused by the "carcinogenic" effect of the environment modified by a series of genes. In population, these genes tend to occur in allelic forms representing functional polymorphisms thus explaining inter-individual variability in cancer susceptibility. The latter can be evaluated more realistically in childhood ALL than in sporadic cancers of the adult because of its relatively short latency period. We asked therefore, the question about the role of genes controlling the efficiency of xenobiotics metabolism in childhood leukemogenesis. Xenobiotics (drugs and carcinogens) are excreted from the body after metabolic conversion by enzymes mediating oxidation activation (Phase I) and conjugation detoxificaton (Phase II). Functional variants of these enzymes, resulting from known DNA polymorphisms in the corresponding genes, were shown to influence the risk to a variety of solid tumours in adults. A case-control study on ALL patients and healthy controls in a French-Canadian population was carried out by examining the loci of Phase I, CYP1A1 and CYP2D6, as well as Phase II enzymes, GSTM1, GSTT1, NAT1 and NAT2. The NAT2 slow-acetylator, CYP1A1*2A and GSTM1 null genotypes were shown to be significant risk determinants of ALL (OR=1.6, 1.8 and 1.8, respectively), whereas, polymorphisms in CYP2D6 and GSTT1 genes did not seem to play an important role in the aetiology of ALL. Interestingly, the risk associated with NAT2 slow-acetylators was most apparent among males homozygous for NAT1*4 (OR=3.3) whereas girls carrying the CYP1A1*4 allele were significantly underrepresented in the patient group (OR=0.2). These findings point to a gender-specific effect of DNA variants which, at least in part, may explain why ALL is more prevalent among boys. To assess gene-gene interactions, NAT2 slow-acetylators were considered together with GSTM1 null genotypes and CYP1A1*2A alleles. The combined presence of two risk-elevating genotypes appeared to confer an increased risk of ALL among the carriers (OR=2.6). This risk was increased further (OR=3.3) when all three genotypes occurred in the same individuals indicating that the combination of susceptibility variants is more predictive of risk then either of them independently. The association of leukemogenesis in children with metabolising gene variants suggests causal relation to environmental exposures.

Analysis of PBPK models for risk characterization

Adoption of a Bayesian framework for risk characterization permits the seamless integration of different kinds of information available in order to choose and parameterize risk models. It also becomes easy to disentangle uncertainty from variability, through hierarchical statistical modeling. Appropriate numerical techniques can be found, for example, in the recently developed arsenal of Markov chain, Monte Carlo simulations. The developments in this area can actually be viewed as extensions of the traditional or standard Monte Carlo methods for uncertainty analysis. Following a brief review of the techniques, examples of Bayesian analyses of physiologically-based pharmacokinetic models are presented for tetrachloroethylene and dichloromethane. The discussion touches on some open problems and perspectives for the proposed methods.

Cytogenetic effects from exposure to mixed pesticides and the influence from genetic susceptibility

Exposure to pesticides remains a major environmental health problem. Health risk from such exposure needs to be more precisely understood. We conducted three different cytogenetic assays to elucidate the biological effects of exposure to mixed pesticides in 20 Costa Rica farmers (all nonsmokers) compared with 20 matched controls. The farmers were also exposed to dibromochloropropane during the early employment years, and most of them experienced sterility/fertility problems. Our data show that the farmers had consistently higher frequencies of chromosome aberrations, as determined by the standard chromosome aberration assay, and significantly abnormal DNA repair responses (p < 0.05), as determined by the challenge assay, but no statistically significant differences in the tandem-probe fluorescence in situ hybridization (FISH) assay (p > 0.05). Genotype analysis indicates that farmers with certain "unfavorable" versions of polymorphic metabolizing genes (cytochrome P4502E1, the glutathione S-transferases mu and theta, and the paraoxonase genes) had significantly more biological effects, as determined by all three cytogenetic assays, than both the farmers with the "favorable" alleles and the matched controls. A unique observation is that, in individuals who had inherited any of the mentioned "unfavorable" alleles, farmers were consistently underrepresented. In conclusion, the Costa Rican farmers were exposed to genotoxic agents, most likely pesticides, which expressed the induction of biological and adverse health effects. The farmers who had inherited "unfavorable" metabolizing alleles were more susceptible to genotoxic effects than those with "favorable" alleles. Our genotype data suggest that the well-recognized "healthy worker effect" may be influenced by unrecognized occupational selection pressure against genetically susceptible individuals.

Genetic polymorphism and their contribution to cancer susceptibility

Since the majority of chemical carcinogens are not capable of causing hazardous effects per se, the metabolism of these compounds is a crucial part of the initial host response to the environmental exposure. Disturbances in the balance between activation and detoxification may thus explain the individual variations in responses to exposures to carcinogens. The amount of the ultimate carcinogen produced depends on the action of competing activation and detoxification pathways involving cytochrome P450 and glutathione-S-transferases enzymes.

Immunochemical characterization of hepatic cytochrome P450 isozymes in the channel catfish: assessment of sexual, developmental and treatment-related effects

The profiles of immunoreactive proteins recognized by antibodies raised against purified trout P-450 isoforms (CYP1A1, CYP2M1 and CYP2K1) were examined in channel catfish liver by Western blot analysis. Gender differences in basal expression of these isoforms, as well as responses to known inducers of mammalian isoforms (ethanol, beta-naphthoflavone and clofibric acid) and early life stage (3 and 6 months) profiles are described. Two similar protein bands were detected by Western blotting in mature untreated catfish with CYP2K1 and CYP2M1 antibodies. A third band is detected by anti-2K1 in fish treated with beta-naphthoflavone; this band was verified as CYP1A, with about twice the level of expression in males versus females. No difference between sexes was seen in the expression of the 51-kDa CYP2-reactive bands; however, a significant difference (female > male) was seen in the lower molecular weight CYP2 band (47-kDa). Ethanol treatment caused a dose-dependent decrease in the 47-kDa CYP2-reactive isoforms but no change in the 51-kDa band. Clofibric acid treatment caused an increase in both the 51-kDa CYP2 protein as well as in liver somatic index. Age-dependent changes in isoform expression were also detected in CYP2-reactive forms, with a novel protein (53-kDa) detected in 3-month-old fish. The results from this study provide insight into the regulation of constitutive catfish CYP isoforms and prepares a foundation for further examination of the biotransformation capabilities of an important aquatic species.

Cancer risk assessment of extremely low frequency electric and magnetic fields: a critical review of methodology

This review provides a discussion of cancer risk assessment methodology pertinent to developing a strategy for extremely low frequency electric and magnetic fields (EMF). Approaches taken for chemical agents or ionizing radiation in six key topic areas are briefly reviewed, and then those areas are examined from the perspective of EMF, identifying issues to be addressed in developing a risk assessment strategy. The following recommendations are offered: 1) risk assessment should be viewed as an iterative process that informs an overall judgment as to health risk and consists of a complex of related activities incorporating both positive and negative data, tumor and nontumor end points, and human and nonhuman sources of information; 2) a hazard identification resulting in a conclusion of weak or null effects, such as may be associated with EMF, will need to assign significant weight to animal cancer bioassays conducted under defined exposure conditions as well as to human epidemiologic studies; 3) a default factor to account for possible age differences in sensitivity to carcinogenesis should be included in an EMF risk assessment; 4) lack of evidence of dose response and the apparent lack of DNA reactivity of EMF suggest that a safety (or uncertainty) factor or margin of exposure type of risk characterization may be most appropriate; and 5) an EMF risk assessment should permit at least tentative conclusions to be reached as to the limits of carcinogenic risk from exposure to EMF, and should also define an efficient research agenda aimed at clarifying uncertainties appropriate to a more complete assessment.

Carcinomas of the renal pelvis associated with smoking and phenacetin abuse: p53 mutations and polymorphism of carcinogen-metabolising enzymes

Phenacetin abuse and smoking are established risk factors for transitional cell carcinomas of the urinary tract. In the present study, we analysed exposure and the clinical course of patients who underwent nephrectomy for transitional cell carcinoma of the renal pelvis. PCR-SSCP of archival, paraffin-embedded histological sections followed by direct DNA sequencing revealed that 29 of 89 (33%) renal pelvic carcinomas contained a p53 mutation. Double mutations were found in 4 tumours and triple mutations in 1 tumour. The incidence of p53 mutations was significantly higher in tumours with grades 3 and 4 than in those with grades 1 and 2 and higher in invasive than in non-invasive tumours. Furthermore, patients with carcinomas carrying a p53 mutation showed poorer survival than those without mutation. The type of p53 mutation in renal pelvic carcinomas was similar to that reported for bladder cancer, G:C-->A:T transition mutations being most frequent (45%, 33% of these at CpG sites), followed by G:C-->T:A and G:C-->C:G transversions. The incidence and type of p53 mutation did not differ significantly in patients with a history of phenacetin abuse, smoking or neither of these habits. This was also true for G:C-->T:A transversions (17.5% of mutations), which are considered typical of smoking-induced carcinomas at other sites, e.g., lung, oral cavity and oesophagus. Our results indicate that the frequency and pattern of p53 mutations are similar in transitional cell carcinomas of the bladder and the renal pelvis and do not reflect exposure to phenacetin and/or smoking. The frequency of genetic polymorphism in genes coding for carcinogen-metabolising enzymes (CYP1A1, NAT1, GSTT1 and GSTM1) was also independent of exposure. Although the sample size of our study does not allow definite conclusions, these data are compatible with chronic tissue damage as a causative factor in the evolution of urothelial carcinomas rather than pointing to a direct mutagenic effect of phenacetin and tobacco-specific carcinogens.

Pharmacogenetics and cancer chemotherapy

Cancer chemotherapy is limited by significant inter-individual variations in responses and toxicities. Such variations are often due to genetic alterations in drug metabolising enzymes (pharmacokinetic polymorphisms) or receptor expression (pharmacodynamic polymorphisms). Pharmacogenetic screening prior to anticancer drug administration may lead to identification of specific populations predisposed to drug toxicity or poor drug responses. The role of polymorphisms in specific enzymes, such as thiopurine S-methyltransferases (TPMT), dihydropyrimidine dehydrogenase (DPD), aldehyde dehydrogenases (ALDH), glutathione S-transferases (GST), uridine diphosphate glucuronosyl-transferases (UGTs) and cytochrome P450 (CYP 450) enzymes in cancer therapy are discussed in this review.

Factors contributing to discrepancies in population monitoring studies

A review of the scientific literature on population monitoring studies (on non-accidentally exposed populations) frequently show that many of these studies using similarly exposed populations and the same laboratory techniques do not produce consistent results. To illustrate the problem, a brief review of studies using well validated techniques (chromosome aberrations and hprt gene mutation) to elucidate genotoxic effects of cigarette smoking is presented. Although many factors can contribute to the generation of discrepant results, two obvious factors are small sample sizes and inadequate experimental data. In addition, a new factor on genetic susceptibility should be considered in population studies whenever appropriate. The new factor is based on recent data showing the influence of polymorphic metabolizing genes on response to environmental mutagens towards biological effects and disease outcome. The common ones include the cytochrome P450 and the glutathione S-transferase genes. The inclusion of susceptibility factors in population monitoring may revolutionize the approach for health risk assessment and for environmental regulations.

Variation in DNA repair is a factor in cancer susceptibility: a paradigm for the promises and perils of individual and population risk estimation?

The repair of DNA damage protects the genome of the cell from the insults of cancer causing agents. This was originally demonstrated in individuals with the rare genetic disease, xeroderma pigmentosum, the prototype of cancer genes, and subsequently in the relationship of mismatch repair to colon cancer. Recent studies suggest that individuals with less dramatic reductions in the capacity to repair DNA damage are observed at polymorphic frequency and these individuals have an increased susceptibility to several types of cancer. Screening of individuals for DNA sequence variation in the exons of 9 DNA repair genes has resulted in identification of 15 different polymorphic amino acid substitution variants. Although the studies to relate these variants to reduced DNA repair capacity and cancer status have not been completed, the available information is sufficient to suggest that DNA repair genes should be incorporated into molecular epidemiology and cancer susceptibility studies. The availability of molecular epidemiology data presents exciting opportunities for refinement of risk estimation models and identification of individuals at increased risk of disease, with resultant opportunities for effective surveillance and early intervention and treatment. The opportunities to acquire susceptibility data are associated with possible perils for establishment of regulations for permissible exposures to carcinogenic agents and also stigmatization of 'at risk' individuals that may result in decreased access to employment opportunities and health care.

Ethnic differences in cancer incidence: a marker for inherited susceptibility?

Cancer incidence varies markedly by ethnicity and geographic location. Ethnic variation in cancer occurrence has traditionally been ascribed to differences in social, cultural, economic, and physical environments. However, this interpretation of the epidemiologic evidence may need to be revised as a result of new biological evidence and theories of carcinogenesis. Carcinogenesis is now recognized to be a multistep process during which mutations or heritable changes in expression occur in genes involved in cellular growth control and genome stability. Inherited cancer susceptibility may be a stronger determinant of ethnic differences in cancer incidence than is currently appreciated. To examine the potential role of inherited susceptibility, the theoretical contribution of inherited susceptibility to ethnic differences in rates in considered using a simple probability model. Germline mutations in tumor suppressor genes BRCA1 and p53 are used to illustrate the magnitude of the ethnic differences for breast cancer that might arise from differences in inherited susceptibility. Our simple model suggests that ethnic differences in cancer occurrence can result from differences in genetic susceptibility. However, the magnitude of ethnic relative risk is likely to more strongly reflect differences in the distribution of susceptibility genotypes between groups than the magnitude of the disease risk associated with the genotypes. For many scenarios, the ethnic relative risk arising from differences in susceptibility may be bounded by the ratio of the proportion of susceptible individuals in each group.

DNA adducts: biological markers of exposure and potential applications to risk assessment

DNA adducts have been investigated extensively during the past decade. This research has been advanced, in part, by the development of ultrasensitive analytical methods, such as 32P-postlabeling and mass spectrometry, that enable detection of DNA adducts at concentrations as low as one adduct per 10(9) to 10(10) normal nucleotides. Studies of mutations in activated oncogenes such as ras, inactivated tumor suppressor genes such as p53, and surrogate genes such as hprt provide linkage between DNA adducts and carcinogenesis. The measurement of DNA adducts, or molecular dosimetry, has important applications for cancer risk assessment. Cancer risk assessment currently involves estimating the probable effects of carcinogens in humans based on results of animal bioassays. Estimates of risk are then derived from mathematical models that fit data of tumor incidence at the high animal exposures and extrapolate to probable human exposures that may be orders of magnitude lower. Molecular dosimetry could extend the observable range of mechanistic data several orders of magnitude lower than can be achieved in carcinogenesis bioassays. This measurement also compensates automatically for individual and species differences in toxicokinetic factors, as well as any nonlinearities that affect the quantitative relationships between exposure and molecular dose. As a result, molecular dosimetry can provide a basis for conducting high- to low-dose, route-to-route, and interspecies extrapolations. The incorporation of such data into risk assessment promises to reduce uncertainties and produce more accurate estimates of risk compared to current methods.