Page last modified November 6th 2005

Published Papers
Wilson Home page
Public Hearings
Science and Law
Arsenic Project
Harvard Physics
Radiation and Risk
Cyclotron History
This page
Particle  Physics
Human Rights


The group has recently concentrated on Cancer at Old Age

(A)    It has been the conventional wisdom for a longtime that if a person does not die of something else (car accident, heart disease etc.) he/she will die of cancer.  This view is being challenged by the fact that the age-specific incidence of cancer does not seem to rise indefinitely but flattens off about age 80 and even falls above that age.    This was pointed out by Dr Francesco Pompei who successfully defended his thesis on this subject in April 2002.   A paper Age Distribution of Cancer:  The Incidence Turnover at Old Age  was published in  Human and Ecological Risk Assessment in November  2001.     Better data are presented by  Ellen Lee, Frank Pompei and Richard Wilson, (2004) in a comment on a paper by Campisi, (  "Cancer Turnover at Old Age"    Nature Reviews (1994) published paper 868)) .  These were expanded in a  poster at 57th annual scientific meeting of the Gerontological Society of America and in  a paper submitted to the Journal of Gerontology.  Some implications were presented in paper number 872. by  Pompei F., and Wilson R., "A Quantitative Model of Cellular Senescence: Influence on Cancer and Longevity" Toxicology and Industrial Health  18:365-376 (2002) and in a more recent paper by Pompei  and Wilson  "A Quantitative Model of Cellular Senescence: Influence on Cancer and Longevity" in  Toxicology and Industrial Health  18:365-376 (2002).   We are studying the same data for a number of observational periods to address the isue of cohot bias (which we beleive to be small)  and the study of cancer-caused mortality in the same dta base. 

(B)  We are studying the age dependence of tumors in laboratory animals in the ED01 data.  Pompei,  Polkanov  and Wilson showed that there is a fall off in cancer incidence after 800 days of life in mice in the ED01 data.  Most animal bioassays have a terminal sacrifice at 700 days, so this is not always apparent.   We hope to extend this work.  We repeatedly urge chemical industry and the National Toxicology Program to carry out their bioassays till the end of life so that this phenomenon may be studied further.

(C)   The conventional theories of cancer incidence, the multistage theory examined by Doll and Armitage and the clonal expansion theory of Moolgavkar and Knudsen cannot explian this fall off at old age.   It except by assumption that the number of completely non susceptible persons varies with cancer site in such a way that the turn over happens at the same age for each cancer.  This means 99% completely non susceptible people for a rare cancer and 20% foa common cancer.   Some people had speculated that the problem was in the mathematical approximations used.  Ritter, Burmistrov,  Pompei,  and Wilson R., in   The Multistage Model of Cancer Development: Some Implications.   Toxicology and Industrial Health (2002) examined the mathematically exact formulation and showed that it does not alter the above conclusion.  

Earlier work on Animal/Man comparisons in Chemical Carcinogenesis
    This group pioneered the use of NCI/NTP CBDS database for addressing many problems and questions about carcinogenesis. This data base is of the tumors and other lesions in over 300,000 rodents exposed to a variety (over 400) of chemicals.  The highlights of previous work by this group using the CBDS database include (numbers refer to the number in the list of published papers by Richard Wilson and links when a copy is available).    There have been several recent discussions suggesting that regulation based upon  these animal bioassays causse unnecessary panic in the American populace.   Richard Wilson disagrees and in a letter to the Washington Times claims that the problem is the attempt to regulate miniscule risks.   

(1)  Crouch and Wilson (1979) "Interspecies Comparison of Carcinogenic Potency," J. Tox. Environ. Health. 5:1095 (Wilson published papers number 211) and subsequent papers by Crouch (Crouch, 1983a, 1983b) on interspecies comparison of carcinogenic potency)  compare potency in sveral species -- principally comparing potency in rat and mouse.    It is shown that when animals are fed the same amount, daily,  as a fraction of body weight, the lifetime incidence of cancer is comparable.

(2) Zeise et al.  confirmed and elucidate an unexpected correlation between acute toxicity and carcinogenic potency;  (Zeise, L. PhD Thesis, Harvard University. (1984);  Zeise, L., E.A.C. Crouch, and R. Wilson (1984). "Use of Acute Toxicity to Estimate Carcinogenic Risk." Risk Analysis 4:pp 187-199;   "Reply to Comments: On the Relationship of Toxicity and Carcinogenicity," Risk Analysis 5: 265-270.   Zeise, L., Crouch, E. and Wilson, R (1985) Uncertainty in Risk Assessment, Banbury Report 19:133;  Zeise, L., E.A.C. Crouch and R. Wilson (1986). "A Possible Relationship Between Toxicity and Carcinogenicity," J. Am. Coll. Toxic. 5:137.

(3)  Initially Zeise, and later Metzger et al. demonstrated that the correlation of item 2 is both weaker, and has a larger coefficient, for chemicals that are not in the NCI/NTP database which were presumably tested earlier; (Metzger et al., 1989).(Wilson published papers number 381)

(4) Goodman and Wilson showed that the correlation between carcinogenicity and acute toxicity, although present for both mutagens and non-mutagens, is slightly stronger for non-mutagens. Goodman, G. and R. Wilson. (1991). "Quantitative Prediction of Human Cancer Risk from Rodent Carcinogenic Potencies: A Closer Look at the Epidemiological Evidence for some Chemicals Not Definitively Carcinogenic in Humans," Regul. Toxicol. Pharmacol. 14: 118.  Goodman G. and R.Wilson (1992), "Comparison of the Dependence of TD50 on Maximum Tolerated Dose for Mutagens and Nonmutagens," Risk Analysis 12:525-533..

(5) Shlyakhter, Goodman and Wilson (1992) performed a statistical (Monte Carlo) study of the relationship between acute toxicity and carcinogenic potency to address the extent to which they are real and the extent to which they might be statistical artifacts  (Wilson published papers number 461).

(6) Byrd et al, 1988, 1990 studied the concordance between different organ sites for different species (Byrd et al., 1988, 1990) (Wilson published papers numbers 437,500).

(7) Byrd et al. (1988,1990) demonstrated that rare tumors in one species are better predictors of tumors in another species than are common tumors; this was the first definitive quantitative justification for a procedure adopted by both IARC and FDA (Wilson published papers numbers 437,500).

(8) Byrd et al. (1990) demonstrated that, contrary to our initial presumption, liver tumors in B6C3F1 mice are as good a predictor of tumors in rats as other common tumors.  Moreover, benign liver tumors (adenomas) are as predictive as malignant tumors (carcinomas)  (Wilson published papers number 500).

(9) We prepared a simplified version of the CBDS data base, together with listing of all tumors, available on a floppy disk, and made this available to all who requested a copy.

(10) Gray et al., 1995, showed that concordance studies across sites and species improves when common regulatory criteria (as used by FDA for example) are used.   This is a quantitative justification for these procedures.  (Wilson published papers number 620 ).

(11) Linkov et al (1998a), using Monte-Carlo modeling, studied false-positive detection rates in long-term rodent bioassays, Linkov et al., 1998a) (Wilson published papers number 650)

(12) An initial study of anticarcinogenic effects in the CBDS database (Linkov et al., 1998a, 1998b) (Wilson published papers 642 ,650 and erratum )

<>(13) In related work not using the CBDS data base tumor site concordance between humans and laboratory animals was explored directly (Gray and Evans, 1992). With humans and mice there was no evidence of tumor site concordance. A similar comparison of humans with rats results in a statistically significant level of discordance, that is a response in one tissue in one species is actually statistically associated with the lack of that response the other species. In this data set there is no statistically significant association (at the p = 0.05 level) of tumor site between rats and mice for those human carcinogens tested in both species, although evidence of discordance is present with 0.10 < p < 0.05.   We are studying these site specific correlations in detail with a view to comparing the predictions of an Absolute Risk (AR) and a Relative Risk (RR) model..

(14)  We have been exploring an anticorrelation between the incidence of liver tumors in control (non dosed) rodents and lymphomas in control animals.   This seems to be most pronouned for histiocytic lymphoma and adenomas.  "Correlations Among Tumor Types in Mouse Cancer Bioassays: Liver Adenomas, Liver Carcinomas,  Leukemias and Lymphomas" , I. Linkov, M. Polkanov, A. Shagiahmetov, R.Wilson and G.M. Gray, Toxicology and Industrial Health 16, 16 (2000)  

(15)   Liver tumors in rodents have been a puzzle for a long  time.  Some people combine them before analysis but there is strong evidence of a biological difference.  Some of these biological diferences show up in the CBDS and TDMS data bases and are the basis for the paper:: "LiverAdenomas and Carcinomas: Correlations and Relationship to Body Weight in Long-Term Rodent Cancer Bioassays" ,

(16)  We are examining in a little detail the effects of tumor groupings  (classifications) on the fraction of chemicals being studied that are assigned  to be carcinogens or anticarcinogens,   A preliminary report on  this was presented as a poster at the Society Society of Toxicology (SOT)  meeting in March 1999 and a more complete paper is published in Regulatory Toxicology and Pharmacology, 36:139-148, (2002)   as published;   as submitted with extra figures.

(17) In addition members of the group have been authors of a number of other papers on specific risks of chemicals, or asbestos (Wilson published  papers numbers 207, 209, 212, 218, 240, 258, 304, 309, 317, 327, 332, 343,  346, 347, 378, 391, 408, 414, 440, 451, 455, 499, 495, 499, 701.).  See also papers by Alexander Shlyakhter.

(18)    We have always been supporters of the argument that a linear dose response relationship between any pollutant and an effect (benign or adverse) should be considered as a default (to be used unless good data say otherwise).  This follows from the general multistagemodel of Armitage and Doll, but is more general than that:  the papers in 1975 of Crump, Guess, Langley and Peto show that it should apply to cancer incidence if and when these cancers are indististinguishable from cancers caused by natural (background) processes.  Crawford and Wilson (paper number 568 ) show that same argument applies to many non-cancer medical outcomes also - including respiratory problems caused by air pollution.  Attached are slides from a recent lecture on the problem. For example for radiation RW has argued that there exist no definitive data that force a departure from this view.  He has also presented a discussion of how one might regulate Environmental Hazards.

(19)     We have been studying the effects of arsenic on people and have a website on the subject.  In addition we have presented to the US EPA an argument, similar to that in (18) above, that a default dose-response relationship should be linear.

(20)  We have been studying the extent to which Absolute Risk or Relative Risk best describes an extrapolation from tumor at a given sitein one rodent species to the same site in another species.  At a specific site the difference between the absolute risk and the relative risk predictions is smaller than the variation between chemicals but average overall, the data seem to prefer the Relative Risk approach.  "Absolute Risk or Relative Risk? A Study of Intraspecies and Interspecies Extrapolation of Chemical-Induced Cancer Risk" .  J. Kuo, I. Linkov, L.Rhomberg, M. Polkanov, G. Gray, and R. Wilson.   Risk Analysis (about 2002)

(21)    Richard Wilson has been involved with scientists at other institutions in considering the effects of various fibers on health (papers, 455,561,647,701,719,810).   A conference on the effects of some minerals mined in Minnesota is was held in April 2003.   The group also assessed the asbestos exposures and risk of someone near the world trade center when it collapsed.     Down load an MSWORD file  Nolan R., Ross, M., Nord, G., Axten C., Osleeb J., and Wilson R.   Presented to Conference in Asbest, Urals, Russia

(22)   Notwithstanding the arguments in 21 above,  we  discussed in an unpublished lecture that in the same animal one can often have both anticarcinogenic at one tumor site and carcinogentic effects at another tumor site.   The total cancer rate has a dip.


The following persons have worked with the group in the above work at various times:   D.M. Byrd, E.A.C.Crouch, J.S Evans, M. Fiering, G. Goodman, G. M.Gray, E. Hakonoglu, Jeanne Kuo,  E.E. Lee, P. Li, I. Linkov, R.Macdonald, B. Metzger, M. Polkanov, F. Pompei, A. Shaghiametov, A.I. Shlyakhter, I. Shlyakhter, R.Wilson, and L. Zeise. In addition and most importantly, the work would not have been possible without the dedicated work of the staffs of the Surveillance, Epidemiology and End Results (SEER) program and the National Toxicology Program (NTP).

Possible future work that we would like to see

Further evaluate anticarcinogenic responses in the NTP rodent bioassay database and to maintain an ability to perform calculations quickly. The database will be refined and adjusted the results for identified confounding effects (weight and survival depression and false-positive detection).

Evaluate the role of various attributes of tumor responses in prediction of carcinogenicity and anticarcinogencity across species. These attributes include tumor site, background tumor rate, and known mechanism of action data.

Develop and validate methods for quantitative integration of both increases and decreases in tumor rates to assess a chemical's overall carcinogenic potential.

Analyse the carcinogenic and anticarcinogenic responses due to chemical administration for different levels of statistical significance.

Evaluate the contributions of the random effects as well as weight and survival decreases by Monte-Carlo modeling as well as by correcting tumor rates based on empirical data for dietary restricted animals.

Build upon insights developed in our studies of cross-species concordance of carcinogencitiy (Gray, et al., 1995) and anticarcinogenicity (Linkovet al., 1998a).

Explore further the observation that the sites of increased and decreased tumors are almost always different between mice and rats.

Evaluate whether response predicts across species better for similar sites or sites with similar background rates.    We have found that the background tumor rate has a strong influence on the likelihood of both carcinogenic and anticarcinogenic responses in bioassays.  This will also have important implications for dose-response evaluation.

Correlate the observations with simple mathematical models of carcinogenesis.

Sudt the age distribution of caners in laboratory animals at old age to see whether the turn over at old age found in people applies to rodents also.

Critical examination of  the correspondence of organ toxicity and tumor formation in the long-term NTP bioassay. The specific hypotheses that will be investigated are:
that subsets of the data, identified through biological or chemical properties of the chemicals, are more correlated than the total data set. i.e. there is a higher correspondence between target organ toxicity and tumorigenicity. Some of the specific subsets to be investigated include in vitro mutagenesis or compounds with chemical structure alerts as described, for example inaseries of papers by Ashby and Tennant (1988,1992), as well as compounds which act at particular sites, the mouse liver for example.

The role of bioassays in understanding risks of chemicals

 Although animal bioassays have been used for half a century to discover which chemicals cause cancer, their use in quantitative assessment of risk is only about 20 years old. The U.S. Food and Drug Administration and the Environmental Protection Agency (EPA) have made the most extensive use of quantitative risk assessments and the assumptions they make dominate the field.   However the assumptions are rrely clearly stated.

The four most crucial assumptions are:

(1) that a substance that is carcinogenic in animals is carcinogenic with a similar potency (measured in appropriate units) in humans;

(2) that there is a linear (proportional) relation between dose and carcinogenic response (probability of developing cancer); and

(3) the slope of the dose response relationship at low doses can be derived from data at high doses

(4) it is reasonable to treat all carcinogens, regardless of proposed mechanism of action, in the same way.

Studies on animals are used to detect potential human carcinogens before harm to people is obvious. The usefulness of animal tests is usually justified by reference to the fact the all compounds identified as human carcinogens through epidemiology have been demonstrated to be animal carcinogens (Wilbourn et al., 1986; Tomatis et al., 1989; Huff, Haseman, and Rall, 1991). However, the converse of this argument, that all animal carcinogens are or can be expected to be human carcinogens, has been challenged. Some challenges can be addressed by a careful consideration of potency and exposure. Ennever et al.(1987) listed several animal carcinogens which had not been demonstrated to cause cancer in humans in spite of careful and responsible epidemiological studies. But Goodman and Wilson (1991) showed that the carcinogenic potency in humans predicted on the basis of chemical bioassays and the "usual" interspecies relationship is not large enough to expect tumors to appear in the small, moderately exposed group of humans in these studies.

Each of the major regulatory assumptions listed above has come to be challenged by the scientific community. In many ways, the objections hinge on assumption 4, that all carcinogens are the same. Empirical studies have demonstrated that there are instances when assumptions 1 and 2 are indeed reasonable (i.e. Crouch and Wilson, 1979), and times when they are not (Cohen and Ellwein, 1990; Swenberg, 1991). In particular assumption 1 is wrong for arsenic and in our view an adherence to this assumption, without realizing that it is an assumption is a major reason for the tragedy in Bangladesh.  In our work we study ways in which differences between carcinogens can be demonstrated empirically hoping to help bridge the gap between scientists and regulatory scientific policy.

We note in particular that there are many natural chemicals, many of them in our foods, that we eat regularly.  This is illustrated by the attached Holiday Menu.

Interpecies Comparisons of carcinogencity

Historically it seemed reasonable that cancer might appear in man in the same organ as it appears in animals (rodents). In 1973 Tomatis noticed that there is a greater correlation between animal carcinogenicity and human carcinogenicity if the sites were allowed to be different for the two species, than if they were forced to be the same. This was further confirmed by this group by Crouch and Wilson (1979) and by Crouch (1983a,1983b) who studied the comparison of carcinogenic potency between rats and mice. In 1983 (Anderson et al .) the Carcinogenic Assessment Group (CAG) of the EPA, made this assumption explicit for regulatory purposes. No explicit biological rationale was ever put forth to support this assumption. CAG now assumes that if a chemical produces a statistically significant excess of tumors in a group of animals, usually rats or mice, at any one site, then it will be considered to pose a probable risk of cancer to humans at some unspecified site. There are several problems with this interpretation. Since carcinogenic potency and acute toxicity have been found to be correlated, and there is an inter-species correlation of acute toxicities, there is a strong empirical connection between toxicity and carcinogenicity (Bernstein et al.1985, Crouch et al. 1987).This could be explained simply if acute toxicity is the cause of tumors with most chemicals, (Gold et al.1992) although other explanations are possible.There are problems with this simple explanation. In particular if the Gold explanation applies to most chemicals it is hard to understand why there appears to be little concordance across tumor sites.

EPA did not formally consider it to be relevant to the quantitative risk assessment whether the site in which tumors are found in rats differs from the site in which they are found in mice, and makes no implication that tumors will be found at the same site in humans. However, such considerations appear in discussing the general toxicological profile of the substance, for instance the appearance of tumors in the same site in both sexes or species tested is taken as stronger evidence for potential human carcinogenicity. In this sense, all carcinogens are considered the same. However, other uses of bioassay data by EPA, PBPK models for example, have the concordance of sites as an implicit assumption. Although EPA developed a fairly well defined method for risk assessment (Anderson et al., 1983; EPA, 1986), the logical and scientific underpinning for this method remains weak. The EPA theory is "unified" in the sense that the risk is calculated for most chemicals in the same way, regardless of any mechanism for carcinogenesis that may have been proposed. The EPA assumes a linear dose-response at low doses, but allows data to suggest differences from linearity at high doses. Although often called a "linearized multistage" model, it is more accurately called a "truncated polynomial" procedure, (Zeise, Wilson and Crouch 1987) since it is a mathematical model whose relationship to the biological multistage model is not unique. This method generates a "plausible upper bound" estimate of carcinogenic potency that has been widely criticized in the scientific community.  It has more recently been replaced by a default approach that gives a similar low dose risk - to draw a straight line from the lowest response at a lowdose (LD10) to the origin.

Tumor sites are said to be concordant across species if the tumors appear at the same anatomical site in different species when they are exposed to the same carcinogenic agent. Although usually ignored in risk assessment, concordance plays a very large implicit role in the process. In the hazard identification portion of risk assessment the underlying assumption of perfect concordance plays a role in both epidemiology and the interpretation of the relevance of animal bioassay results to humans. Tumor site concordance often guides the design of epidemiology studies. The results of an animal bioassays may lead to case/control epidemiologic studies examining the risk of cancer in humans in the same site in which it was found in animals. If carcinogenic agents frequently affect humans and animals at the same site this strategy can help to focus the efforts of epidemiologists. On the other hand, if tumor site concordance is not to be expected, this approach may hinder attempts to identify human carcinogens by focusing the attention of epidemiologists on single tumor sites. In addition to its influence on epidemiology, in hazard identification tumor site concordance is frequently invoked by those who wish to ignore or downgrade the relevance to humans of tumors in rodent tissues with no human equivalent (e.g. zymbal gland) or when concordance is lacking (Ashby et al., 1990).

In dose-response evaluation tumor site concordance is important both as the basic assumption of dose-response modelling and in the determination of "dose." The standard models of dose response used in risk assessment,such as the "linearized multistage (LMS) model" (more accurately called atruncated polynomial model) that used to be favored by the U.S. Environmental Protection Agency (EPA), actually calculate a site specific cancer potency from animal data which theoretically is only applicable to the same site in humans. The use of this model to estimate risk to humans at any site is out of necessity rather than theory. Tumor site concordance is also an important but usually unacknowledged assumption in the use of physiologically-based pharmacokinetic (PBPK) models in risk assessment. In order to improve the cross-species and high dose to low dose evaluation of risk it is often advocated that PBPK models be used to calculate the dose to the target organ rather than the dose administered to animals or humans for purposes of dose-response evaluation (Krewski, Murdoch, and Withey, 1987). If, however, target organs are not concordant across species, then target organ dose may be an inappropriate dose metric. Examining the issue of tumor site concordance, and perhaps identifying cases when it is an appropriate assumption, and when it is not, should increase our knowledge in these issues of cross-species extrapolation. In addition, study of concordance should generate biologically based hypotheses about factors known to influence tumor site concordance - such as similarities or differencesin pharmacokinetics, metabolism, or gene expression, in different species.

Most of the examination of cross-species extrapolation of carcinogenic response has focused on the fact of carcinogenicity itself, rather thanthe site of response. In addition to the purported 100% correlation between humans and animals it has been demonstrated that rats and mice have a fairly high, about 75% concordance of carcinogenic response when challenged with the same chemical (Haseman and Huff, 1987; Haseman et al., 1987, Byrd et al. 1988, Huff and Haseman, 1991). Others have shown, however, that crossspecies prediction is somewhat better for mutagens that nonmutagens and for chemicals with high acute toxicity (Gold et al., 1989) again indicating that scientific research can demonstrate qualitative and quantitative differences between chemical carcinogens.

Organ Toxicity and Tumor Formation
Do cancers form at the same site as toxic lesions?

    Twenty years ago, it seemed natural to search for evidence of carcinogenicity after chronic administration of a substance at the same site where the substance had been shown to cause toxicity. Many historical examples exist. Benzene at high doses produces pancytopenia (kills blood cells), and it also produces leukemia. Exposure to asbestos results in asbestosis (of the lung) and later was shown to cause lung cancer. Vinyl chloride is toxic to both human and rat liver, and also causes angiosarcoma in both humans and rats. Aflatoxin B1 is both toxic to the liver and is an established risk factor for liver cancer. In two of these cases (benzene and asbestos) it is still an open question whether it is the substance that causes the cancer directly or whether the cancer is caused by the toxicity.

    In spite of the regulatory assumption that all carcinogens are alike,  scientists have endeavored to find distinctions between different types of carcinogens. Following the seminal work of Meselson and Russell (1977), that suggested that animal carcinogenicity is quantitatively related to mutagenicity Parodi et al.(1982) began to study possible correlations between mutagenicand carcinogenic potencies. Parodi et al. discovered, to most scientists' surprise, that carcinogenic potency is more strongly correlated with acute toxicity, than with mutagenicity. This was found independently by our group, using a larger database. (Zeise (1984), Zeise et al.(1985, 1986a,1986b). Associated with this correlation, and a possible cause of it, is the fact that many substances only produce statistically significant tumor responses at the highest dose tested. This Maximum Tolerated Dose (MTD) is related to the acute toxic effects of the substance. This led to concern that this correlation might be merely a statistical artifact of the bioassay experimentaldesign (Bernstein et al., 1985; Crouch, Wilson, and Zeise, 1987; Reith and Starr,1989). An extensive Monte Carlo simulation of the bioassays (Shlyakhter et al., 1990, 1991) showed that while the correlation could in principlebe a statistical artifact of the bioassay itself, it would be so only ifmanysubstances cause tumors in 100% of the animals in a bioassay. The actualvaluesof the various parameters preclude such an explanation. Only 2% orless ofchemicals produce tumors in 100% of the animals in long-term bioassays.

    Possible biochemical reasons for the existence of this correlation (in which it could be described as a biological artifact of the experimental design of the NTP bioassays) have recently been discussed (Ames et al. , 1990; Cohen and Ellwein, 1990). In particular, substances at high doses, approaching the maximum tolerated dose, kill cells. Ames et al. propose that this results in cell proliferation, and subsequent tumor development by a mitotic effect. If this is the case, such substances would be expected to show a very non-linear dose response relationship at these doses. However, at lower doses, a linear dose response might still be appropriate, especially for initiators.

    More recently several examples have emerged of situations in which carcinogenesis indeed appears to be secondary to direct organ toxicity, and there is a clear biological explanation. For example, sodium saccharin causes microcrystals to form in the bladder at high doses, and it is probable that these crystals cause the bladder cancers that are observed, presumably by an irritant effect (Cohen and Ellwein, 1990). As calls to deemphasize the results of animal bioassays have intensified (Ames and Gold, 1990b), so has the need to empirically investigate this relationship in as many ways as possible.

    One of the most contentious issues in carcinogen risk assessment is the extrapolation of risk from high doses in animal studies to the low doses encountered by humans. This makes factors influencing the dose-response all important.In a review, Zeise et al.,(1987) found that the direct evidence for or against a linear, or near linear, dose response relationship, is weak. One well studied carcinogen (2-AAF) seems to show a threshold response for bladder tumors, but the bioassay results are consistent with a linear response for liver tumors. This suggests that some carcinogens at some sites might exhibit a linear dose-response and others might exhibit a non-linear dose response relationship characteristic of acute toxicity. If the high dose carcinogenicity is secondary to an acute toxic effect, the dose-response relationship will be that of the acute toxic effect, which is often expected to be non-linear, rather than the linear relationship usually assumed for carcinogens. Moolgavkar and Luebeck, (1990) are pursuing methods of explicitly incorporating the effects of high dose toxicity on cell proliferation into general dose-response models for risk assessment although empirical support is lacking (Hoel et al ., 1988). If these two types could be distinguished, that is chemicals likely to have nonlinear dose response relationships and those which might, this characteristic could be included in the risk assessment and different substances could be treated differently. If this could be achieved, it would be most useful for carcinogen risk assessment. We believe that one of the factors most likely to lead to strong nonlinearities in the dose-response relationshipis toxicity to the target organ.

    Investigation of the relationship between organ toxicity and tumor formation in animal bioassays is therefore a primary goal which we wish to address in due course.


    Long-term rodent bioassays are conducted with the intention of identifying chemicals with the potential to increase cancer rates. It has long beenknown, but often dismissed, that statistically significant dose-related decreases for certain tumor sites and types are also found in many long-term rodent bioassays. A well known example is the work of Kociba et al, 1978 on dioxin. Anticarcinogenicity has recently come to be taken more seriously.It has been demonstrated that anticarcinogenic effects in animal bioassays apparently have a biological basis (Linkov et al., 1998a, 1998b). In addition,a just published human study (Clark et al., 1996) has demonstrated convincing anticarcinogenicity for selenium administered at low doses to humans. The fact that rodent bioassays also shows anticarcinogenicity for selenium indicatesthe potential generality of protective effects.

    In spite of the findings of anticarcinogenicity, it has been regulatory practice to consider only tumor increases when classifying and regulating a substance as a carcinogen. Factors contributing to the view that anticarcinogenic effects are not important included the inability to account for biases due to experimental conditions, inadequate attention to random responses, and the presence of significant weight loss and other effects of exposing animals at the maximum tolerated dose. Although Haseman and Johnson (1996) suggested that these effects could generate spurious "anticarcinogenic" effects inthebioassays, our preliminary work (Linkov et al., 1998a, 1998b) has shownthatthey can not completely explain the correlation.

    Linkov et al maintain that anticarcinogenic tumor decreases are as common in the bioassays as increases (Davis and Monro, 1995) and are not a single isolated phenomenon.  This contention continues even after some corrigenda and addenda are made to the analysis.    Wilson discussed in an unpublished lecture that in the same animal one can often have both anticarcinogenic at one tumor site and carcinogentic effects at another tumor site.   No one has discussed the implication of this for regulation.

The CBDS and other DATABASES

    The Carcinogenesis Bioassay Data System (CBDS) database of the National Toxicology Program (NTP) is the primary tool for these studies. Not only are sites of all tumors listed, but also toxic lesions are listed. The weights of the animals are listed at the end of life and of the groups of animals at other times.

    However the group hope to expand the work to include the later TBDS data base of NTP as well as data bases from chemical and pharmaceutical industry. The Gold et al. data base may also be useful although this data base does NOT include data on animal weights that is likely to be important for future understanding..

Regulatory Implications

    The results so far indicate that any attempt to a priori predict the target organ for human carcinogenicity from experience of an individual chemical would be fraught with uncertainty. When the goal of testing a chemical for carcinogenicity in animals is to identify potential human carcinogens it appears that the rodent tests can qualitatively confirm whether a compound has carcinogenic potential (Tomatis et al., 1989) but rodent tests in general appear to have little ability to predict the target tissue for human carcinogenicity.

    The regulatory expression of these observation is confused but the scientific community already has discussed an important implication, namely that all carcinogens are not the same (e.g. Butterworth, 1989; Weisburger, 1990). Therefore, a unified regulatory method to identify carcinogens and estimate carcinogenic potency may no longer be appropriate. In principle this is already recognized in the EPA carcinogen policy; the three key assumptions discussed above are stated to be only default assumptions to be replaced when better data or more generally accepted theories are available (EPA,1986). Then risk assessors can adopt a case-by-case approach that utilizes more biological information in the assessment of potency (Thorslund et al. , 1987). No general rule for when it is appropriate to replace default assumptions has yet been found. Some have attempted to argue that nonmutagens are one case, having a nonlinear dose response relationship rather than the default linear relationship. But while it seems to be now accepted that there are examples of this, such as saccharin (Cohen and Ellwein, 1990) and the case of dioxin is being seriously discussed this has not been generally accepted for other nongenotoxic compounds.

    On the other hand we have been emphasizing the generality of linear dose-response relations (Crawford and Wilson 1996, Heitzman and Wilson 1997). They apply whenever the mechanism of the medical ailment being discussed is similar for the carcinogen and for whatever causes the background.


Ames, B.N. and L.S. Gold (1990a). "Too Many Rodent Carcinogens: Mitogenesis Increases Mutagenesis." Science 249:970-971.

Ames, B.N., and L.S. Gold (1990b). "Chemical Carcinogenesis: Too ManyRodent Carcinogens," Proceedings of the National Academy of Sciences, USA 87:7772-7776.

Anderson E. and the Carcinogenic Assessment Group of the Environmental Protection Agency (1983). "Quantitative Approaches in Use to Assess Cancer Risk," Risk Analysis 3: 277-295.

Ashby, J. and R.W. Tennant (1988). "Chemical Structure, Salmonella Mutagenicity and Extent of Carcinogenicity as Indicators of Genotoxic Carcinogenesis Among 222 Chemicals Tested in Rodents by the U.S. NCI/NTP," Mutation Research 204:107-115.

Ashby, J., N.G. Doerrer, F.G. Flamm, J.E. Harris, D.H. Hughes, F.R. Johannsen, S.C. Lewis, N.D. Krivanek, J.F. McCarthy, R.J. Moolenar, G.K. Raabe, R.C. Reynolds, J.M. Smith, J.T. Stevens, M.J. Teta, and J.D. Wilson (1990). "A Scheme for Classifying Carcinogens," Regulatory Toxicology andPharmacology 12:270-295.

Ashby, J., and R.W. Tennant (1991). "Definitive Relationships Among Chemical Structure, Carcinogenicity, and Mutagenicity for 301 Chemicals Tested bytheU.S. NTP," Mutation Research 257:229-306

Bernstein, L., L.S. Gold, B.N. Ames, M.C. Pike and D.G. Hoel (1985). "Some Tautologous Aspects of the Comparison of Carcinogenic Potency in Rats and Mice," Fund. App. Toxicol. 5:79.

Butterworth, B.E. (1989). "Nongenotoxic Carcinogens in the Regulatory Environment," Regulatory Toxicology and Pharmacology 9:pp 244-256

Byrd, D.M. III, E.A.C. Crouch and R. Wilson (1991). "Organ Specificity of Rodent Tumor Induction: A Test of Statistical and Graphical Methods by Examination of Tumor Induction from Ingestions of Selected Substances," in Risk Analysis (C. Zervos ed.) Plenum Press, New York, NY.

Byrd, D. (1990). "New Statistical Approaches to the Quantitative Interpretation of Toxicology Data". J. Amer.Coll.Toxic. 7:559-563

Byrd, D., E.A.C. Crouch, and R. Wilson (1988). "Concordance Between Tumors at Different Sites in Different Species," Meeting of Society for Risk Analysis, Washington, DC, November.

Byrd, D., E.A.C. Crouch, and R. Wilson (1990) "Do Mouse Liver Tumors Predict Rat Tumors: A Study of Concordance Between Tumors Induced in Different Sites in Rats and Mice." Mouse Liver Carcinogenesis: Mechanism and Species Comparisons. pp 19-41, Alan R. Liss, Inc.

California (1985) "Guidelines for chemical carcinogen risk assessments and their scientific rationale" Department of Health Services, State of California, November

Charles, M.W., M.P. Little (1990). "Theoretical and epidemiological evidence for time variations in radiation-induced cancer risk and implications for population risk." Nuclear Electric, June

Clark, L.C., Combs, G.F., et al. (1996). Effects of Selenium supplementation for cancer prrevention in patients with carcinoma of the skin. JAMA 276:1957-1963

Cohen, S.M. and L.B. Ellwein (1990). "Cell proliferation in carcinogenesis." Science 249: pp 1007-1009

Cohen, S.M. and L.B. Ellwein (1991). "Genetic Errors, Cell Proliferation, and Carcinogenesis," Cancer Research 51:6493-6505.

Crouch, E.A.C. and Richard Wilson (1979), "Interspecies Comparison ofCarcinogenic Potency," J. Tox. Environ. Health. 5:1095

Cohen, S.M. and L.B. Ellwein (1993). "Biological Theory of Carcinogenesis: Implications for Risk Assessment," Environmental Health Perspectives (in press).

Crouch, E.A.C. (1983a). "Uncertainties in Interspecies Extrapolation of Carcinogenicity. Environ. Health Perspect. 50:321-327.

Crouch, E.A.C (1983b). "Carcinogenic risk assessment - the consequences of believing models." In: Organ and Species Specificity in Chemical Carcinogenesis, Langebach, Newnow and Rice, eds., pp. 653-665; Plenum Press, New York.

Crouch, E.A.C. and R. Wilson (1981). "Regulation of Carcinogens," Risk Analysis, 1:47.

Crouch, E.A.C., R. Wilson, and L. Zeise (1983). "The Risk of DrinkingWater", Water Resources Research 19:1359-1375.

Crouch, E.A.C., R. Wilson, and L. Zeise (1985). "Uncertainty in Risk Assessment," Banbury Report 19:33.

Crouch, E.A.C., R. Wilson and L. Zeise (1987). "Tautology or Not Tautology," J. Tox. and Env. Health. 20:1.

Davies, T.S., and Monro, A. (1994). The rodent carcinogenicity bioassay produces a similar frequency of tumor increases and decreases: Implications for risk assessment. Reg. Toxicol. Pharmacol. 20, 281-301.

Ennever F.K., T.J.Noonan and H.S. Rosenkranz (1987) "The Predictivityof Animal Bioassays and Short Term Genotoxicity Tests for Carcinogenicityand Noncarcinogenicity to Humans" Mutagenesis 2:73-78

EPA (1986), U.S. Environmental Protection Agency, "Guidelines for Carcinogen Risk Assessment," Federal Register 51:3992-347003

Gold, L.S., L. Bernstein, R. Magaw, and T.H. Slone (1989). "Interspecies Extrapolation in Carcinogenesis: Prediction Between Rats and Mice," Environmental Health Perspectives 81:211-219

Gold L.S., N.B.Manley and B.N.Ames (1992), "Extrapolation of Carcinogenicity Between Species: Qualitative and Quantitative Factors" Risk Analysis 12:579-588.

Goodman, G., and R.Wilson (1989) "Interspecies Comparison of Carcinogenic Potency: The Use of Data Bases" AST/SRA Conference on Interspecies comparisons, Baltimore, Md, October

Goodman, G., A.I. Shlyakhter, and R. Wilson (1991). "The RelationshipBetween Carcinogenic Potency and Maximum Tolerated Dose is Similar for Mutagensand Non-mutagens," in Chemically induced Cell Proliferation: Implicationsfor Risk Assessment. T.Slaga and B.Butterworth (ed.): pp 501-516.

Goodman, G., and R.Wilson (1991) "Predicting the Carcinogenicity of Chemicals in Humans from Rodent Bioassay Data", Environ. Health Perspect. 94:195.

Goodman, G. and R. Wilson. (1991). "Quantitative Prediction of Human Cancer Risk from Rodent Carcinogenic Potencies: A Closer Look at the Epidemiological Evidence for some Chemicals Not Definitively Carcinogenic in Humans," Regul. Toxicol. Pharmacol. 14:118.

Goodman G. and R.Wilson (1992), "Comparison of the Dependence of TD50on Maximum Tolerated Dose for Mutagens and Nonmutagens," Risk Analysis 12:525-533.

Gray, G.M., and J.S. Evans (1992). "Human/Rodent Tumor Site Concordance for Human Carcinogens: Implications for Risk Assessment," Harvard Center for Risk Analysis Working Paper - Available Upon Request

Gray, G.M., Li, Ping, Shlyakhter, I., and Wilson, R. (1995). An empirical examination of factors influencing prediction of carcinogenic hazard across species. Reg. Toxicol. Pharmacol. 22:283-291.  

Haseman, J.K, and J.E. Huff (1987). "Species Correlation in Long-Term Carcinogenicity Studies," Cancer Letters 37:125-132.

Haseman, J.K., J.E. Huff, E. Zeiger, and E.E. McConnel, (1987). "Comparative Results of 327 Chemical Carcinogenicity Studies," Environmental Health Perspectives 74:229-235.

Haseman, J.K., and Johnson, F.M. (1996). Analysis of national toxicology program rodent bioassay data for anticarcinogenic effects. Mutation Res. 350:131-141

Hoel, D.G., J.K. Haseman, M.D. Hogan, J.Huff, E.E. McConnell (1988). "The impact of toxicity on carcinogenicity studies: implications for risk assessment." Carcinogenesis 9:pp 2045-2052.

Huff, J., J. Haseman, and D. Rall (1991). "Scientific Concepts, Value, and Significance of Chemical Carcinogenesis Studies," Annual Review of Pharmacology and Toxicology 31:pp 621-652

Huff, J. and J. Haseman (1991). "Long-Term Chemical Carcinogenesis Experiments for Identifying Potential Human Cancer Hazards: Collective Database of the National Cancer Institute and National Toxicology Program (1976-1991)," Environmental Health Perspectives 96:23-31.

IARC (1996). Chemoprevention in Cancer Control (M.Hakama, V. Beral, E. Buiatti, J. Faivare and D.M. Parkin, eds.) IARC Scientific Publication 136, IARC, France.

Karp, J.E., Chiarodo, A., Brawley, O. and Kelloff, G. (1996). Prostate Cancer Prevention: Investigational approaches and opportunities. Cancer Research 56 5547-5556.

Kociba, R.J., Keyes, D.G., Beyer, J.E., Carreon, R.M., Wade, D.A., Dittenberger, D.A., Kalnins, R.P., Frauson, L.E., Park, C.N., Bernard, S.D., Hummel, R.A., and Humiston, C.G. (1978) Results of a Two-Year Chronic Toxicity and Oncogenicity Study of 2,3,7,8-Tetrachlorodibenzo-p-dioxin in Rats. Toxicol. Appl. Pharmacol. 46:279-303.

Krewski, D., D.J. Murdoch, and J.R. Withey (1987). "The Application of Pharmacokinetic Data in Carcinogenic Risk Assessment," in Risk Assessment: Drinking Water and Health, National Academy Press, Washington, D.C.

Linkov, I., Wilson, R., and Gray, G. (1998) "Anticarcinogenic Responses in Rodent Cancer Bioassays are Not Explained by Random Effects" Fundam. Appl. Toxicol.  ( paper number 650 and erratum )

Linkov, I., Wilson, R., and Gray, G. (1998) "Anticarcinogenic Effectsin Rodent Cancer Bioassays are Not Explained by Depression of Animal Weigh tor Survival" Fundam. Appl. Toxicol. ( paper number  642 and erratum )

Meselson, M. and K. Russell. (1977), "Comparisons of Carcinogenic and Mutagenic Potency." in Origins of Human Cancer ( H.H. Hiatt, J.D.Watson andJ.A. Winstein, eds.) , pp. 1473-1581; Cold Spring Harbor Laboratory,NY.

Metzger B, E.A.C. Crouch, and R. Wilson (1989). "On the Relationship Between Carcinogenicity And Acute Toxicity," Risk Analysis 9:169-177.

Moolgavkar, S. and G. Luebeck (1990). "Two Event Model for Carcinogenesis: Biological, Mathematical, and Statistical Considerations," Risk Analysis 10:323-341.

Morse, M.A. and Stoner, G.D. (1993). Cancer chemoprevention: principles and prospects. Carcinogenesis 14:1737-1746.

Parodi, S., M. Taningher, P. Baero and L. Santi (1982). "QuantitativeCorrelations amongst Alkaline DNA Fragmentation, DNA Covalent Binding, Mutagenicityinthe Ames Test and Carcinogenicity for 21 compounds," Mutation Res. 93:pp1-24.

Reith, J.P, and T.B. Starr (1989). Chronic Bioassays: Relevance to Quantitative Risk Assessment of Carcinogens. Regul. Toxicol. Pharmacol.10:160-173.

Shlyakhter A., Goodman G., and Wilson R. (1992) "Monte Carlo Simulation of Rodent Carcinogenicity Bioassays," Risk Analysis 12:73-82.

Shlyakhter A.I. and Wilson R. (1993) "Estimating Uncertainty and Combining Data Sets in Observational Studies: Lessons from Trends in Physical Constants," to be submitted to Biometrics.

Swenberg, J.A. (1991). "Risk Assessment of Chemicals Causing alpha-2 microglobulin Nephropathy," Regulatory Toxicology and Pharmacology 13:1

Thorslund, T. W., C.C. Brown, and G. Charnley (1987). "Biologically Motivated Cancer Risk Models. Risk Analysis 7:109-119.

Tomatis L., C. Partensky and R. Montesano (1973). "The Predictive Value Of Mouse Liver Tumor In Carcinogenicity Testing -- A Literature Survey," Int. J. Cancer: 12:1-20.

Tomatis, L., A. Aitio, J. Wilbourn, and L. Shuker (1989). "Human Carcinogens So Far Identified," Japanese Journal of Cancer Research 80:795-807.

Wattenberg, L.M. (1996). Prevention of cancer: The role of chemoprevention for the general population and for individuals at high risk due to genetic defects or other factors. In: Proceedings of the American Association for Cancer Research 37:17-618.

Weisburger, E.K. (1990). "Mechanistic Considerations in Chemical Carcinogenesis," Regulatory Toxicology and Pharmacology 12::41-52.

Wilbourn, J., L. Haroun, E. Heseltine, J. Kaldor, C. Partensky, and H. Vainio (1986). "Response of Experimental Animals to Human Carcinogens: An Analysis Based Upon the IARC Monographs Programme," Carcinogenesis 7:1853-1863

Zeise, L. PhD Thesis, Harvard University. (1984)

Zeise, L., E.A.C. Crouch, and R. Wilson (1984). "Use of Acute Toxicity to Estimate Carcinogenic Risk." Risk Analysis 4:187-199.

Zeise, L., E.A.C. Crouch, and R. Wilson (1984). "Reply to Comments: On the Relationship of Toxicity and Carcinogenicity," Risk Analysis 5:265-270.

Zeise, L., Crouch, E. and Wilson, R (1985) Uncertainty in Risk Assessment, Banbury Report 19:133.

Zeise, L., E.A.C. Crouch and R. Wilson (1986). "A Possible Relationship Between Toxicity and Carcinogenicity," J. Am. Coll. Toxic. 5:137.

Zeise, L. Wilson. R, and Crouch, E.A.C. (1987). "Dose Response Relationship for Carcinogens: A Review" Environmental Health Perspectives 73:259-308

Zeise, L. and Crouch E.A.C. (1984) unpublished report

Places forinformation about chemicals and environmental issues
Gold et al potency list
Related work by Richard Wilson on arsenic
Related work by Richard Wilson on radiation
top of page
Richard Wilson home page