Biomedical subjects
Publications and source records attributed to S H Moolgavkar.
A two-mutation model for carcinogenesis is reviewed. General principles in fitting the model to epidemiologic and experimental data are discussed, and some examples are given. A general solution to the model with time-dependent parameters is developed, and its use is illustrated by application to data from an experiment in which rats exposed to radon developed lung tumors.
The recessive oncogenesis model, according to which inactivation of both alleles of specific genes leads to cancer, has received much recent attention. A mathematical formulation of a two-mutation model for carcinogenesis, which includes the recessive oncogenesis model as a special case, was fitted to data from a large experimental study in which rats exposed to radon daughters developed malignant lung tumors. The model described the data well. The results indicate that fractionation of exposure increased the lifetime probability of tumor. Examination of the parameters of the model suggests that the effect of fractionation can be explained by the relative effects of radon daughters on the mutation rates and on the kinetics of growth of initiated cells. The first mutation rate is very strongly dependent upon the rate of exposure to radon daughters, the second mutation rate much less so, suggesting that the nature of the two mutational events is different. The model makes predictions which are testable in future experiments.
Attention has focused recently on the recessive oncogenesis model, according to which inactivation of both alleles of specific genes leads to cancer. A mathematical formulation of this model was fitted to the lung cancer incidence data from a cohort study among British doctors. The model described the data well. One implication is that age influences lung cancer risk among smokers independently of duration of smoking. A study of dose-response within the framework of the model shows that the data are consistent with various interpretations regarding the relative importance of daily level of smoking and duration of smoking in determining lung cancer risk.
Explore the source record for details and available documents.
Mathematical expressions are derived, under different dosing patterns, for the number and size of premalignant clones within the framework of a two-mutation model for carcinogenesis, which has previously been shown to be consistent with a large body of epidemiologic and experimental data.
Risk of breast cancer increases with age at first birth, and is lower in women who bear their first children while young than in nulliparous women. While previous studies have investigated risk of breast cancer in birth cohorts by examining partial aspects of cohort childbearing, the present ecological study assesses total cohort childbearing risk in Connecticut women born between 1855 and 1945. In each cohort, the proportion of women nulliparous and first bearing children in their twenties, thirties, and forties are weighted by relative risks associated with these events as ascertained in previous studies. Summed cohort childbearing risks are compared to the incidence of breast cancer in women 40 years of age and older in the same cohorts. Changes in decade of first birth and nulliparity do not explain the changes in breast cancer incidence observed: while cohort childbearing risk has declined over the period examined, breast cancer incidence has increased in the same cohorts. Alternative explanations for cohort increases in breast cancer incidence are reviewed.
A model for carcinogenesis is presented that incorporates two features: (1) transition of target stem cells into cancer cells via an intermediate stage in two irreversible steps, and (2) growth and differentiation of normal target and intermediate cells. Implications of the model for initiation-promotion and environmental carcinogenesis are considered.
We present a mathematical treatment of a two-mutation model for carcinogenesis with time-dependent parameters. This model has previously been shown to be consistent with epidemiologic and experimental data. An approximate hazard function used in previous papers is critically evaluated.
This paper uses a detailed example to illustrate how to detect individual observations that disproportionately influence the results of hypothesis testing with relative risk regression models.
Three parametric families of relative risk functions for the analysis of case-control data are discussed. A desirable feature for any general relative risk function is that inference based on it be independent of the coding of a binary covariate. Only one of the three families considered has this property. Additionally, when the relative risk is not multiplicative, methods of inference based on the asymptotic covariance matrix are likely to be seriously misleading unless the sample size is very large, as has been noted previously in other papers. This is illustrated by means of examples. Likelihood-based procedures should routinely be employed when nonmultiplicative relative risk functions are used for analysis of case-control data.
This paper describes recent developments in failure time regression, and the application of such methods to epidemiologic data. Relative risk and odds ratio estimation procedures are described under cohort and case-control sampling, and under some less standard designs for sampling within a cohort. Some special issues that arise in the analysis of vital data are also mentioned. While methods for descriptive analyses are emphasized some attention is also given to mechanistic models for carcinogenesis and their uses. Topics meriting further methodologic development are delineated throughout.
Explore the source record for details and available documents.
Initiation, promotion and the role of hormones in carcinogenesis are viewed within the framework of a two-stage model for malignant transformation that has been shown by Moolgavkar and Knudson to be consistent with the main body of epidemiological and experimental data on carcinogenesis. It is argued that hormones influence carcinogenesis principally by clonal expansion or contraction of initiated or first-stage cells. The role of hormones in breast cancer and endometrial cancer is briefly discussed.
Explore the source record for details and available documents.
Vital statistics data on mortality from malignant melanoma from five countries are simultaneously fit by the method of cohort analysis. After the effects of geographic and temporal trends are removed, the proportion of deaths occurring in the younger age groups is found to be higher in females than in males. Relative increases in mortality rates by birth cohort are approximately equal in different regions and appear to be slowing down in recent years.
Lung cancer mortality in England and Wales among males in the age group 25 to 84 over the years 1941 to 1975 was analyzed. Using cigarette consumption information from the Tobacco Research Council and a statistical model, relative and attributable risks were estimated. The lung cancer deaths could be partitioned into deaths among smokers and deaths among nonsmokers; 88% of total male deaths during the study period could be attributed to smoking. Whereas the mortality rate among smokers increased over the entire span of the analysis, the rate among nonsmokers rose gradually until 1956-1960 and then declined until 1971-1975. The beginning of the decline was coincident with the implementation of the Clean Air Act of 1956. In order to address possible systematic biases in the smoking data, Monte Carlo simulations were performed.
Age-specific incidence of melanoma of the trunk and of the limbs increases slowly with age, while incidence of melanoma of the face increases rapidly. This observation has led to speculation that cumulative ultraviolet exposure determines risk for the face, while acute exposure affects the trunk and limbs. However, there have been strong secular trends in melanoma of the trunk and limbs, but not in melanoma of the face. Analysis of incidence data from Connecticut and Denmark shows that steadily increasing rates in successive birth cohorts account entirely for these secular trends and for the differences seen among the cross-sectional age curves of the various sites. Thus, incidence data do not support the hypothesis that melanoma of the face and melanoma of the trunk and limbs involve distinct pathogenic mechanisms.
The relative merits of the likelihood ratio statistic, the Wald statistic, and the score statistic are examined by an empirical evaluation based on matched case-control data. A mixture model for the relative-odds function is used. The likelihood ratio statistic is relatively constant for reasonable values of the mixture parameter, but the Wald statistic is unstable. The score statistic is shown to be independent of the mixture parameter. An exact expression is derived for the change in the score statistic upon deletion of risk sets, and an approximation is numerically evaluated.