A MARKOV MODEL OF THE NATURAL-HISTORY OF PROSTATE-CANCER

The objective of this study was to lay a foundation for future cost-be nefit analyses evaluating the public health impact of treatment and sc reening protocols for prostate cancer. Specifically we wanted to defin e the relative impact on cancer-specific mortality rates of the indivi dual epidemiological components: pathological incidences by age groups , cancer progression rates, and the effect of competing causes of deat h, assuming expectant management (i.e. no definitive treatment). A bio logical model of prostate cancer incidence and progression was convert ed into a standard Markov tree where competing causes of death could o ccur. Weighted averages of progression rates were obtained from clinic al studies. Separate cohorts of 30 year old black and white men were f ollowed for 50 years. The model yielded cancer-specific mortality rate s, overall mortality rates, and pathologic prevalences for both white and black males, consistent with the literature. Sensitivity analyses showed that of all the parameters studied, the pathological incidence of cancer in men under 50 years of age had the greatest impact on the cancer-specific mortality rates. Also important was the annual probabi lity of progression of A1 lesions. However the other parameters includ ing pathological incidence in older males, and progression from locall y-extensive to metastatic lesions had much smaller effects. In summary , this model correlates the clinical literature with the epidemiology of prostate cancer and can be used for further decision analyses. We r ecommend that future research be done to more precisely quantify the p athological incidence of prostate cancer in men under 50-60 years of a ge. More certainty is also needed before generalizing the results of r elatively small Al series to millions of men, since Al progression rat es critically affect the eventual cancer-specific mortality. Enough un certainty remains at this point however, that we cannot advocate wides pread screening for prostate cancer until its merit be demonstrated ei ther by the definitive long term study, or by examination of costs and quality-of-life-adjusted benefits.