Multi-parametric numerical simulation of age-specific cancer rates in human populations

Abstract

The CancerFit computer program allows cancer researchers to analyze epidemiologic data describing the age-specific risk of cancer in terms of hypotheses about historical environmental risks, the heritability of cancer, the role of gender and the processes embedded in cancer formation. The program was based on the theories of Professor W. Thilly (MIT), Professor S. Morgenthaler (ETH) and their students. Written as a Fortran program by Prof. Morgenthaler it was transported into Java by David Hensle (MIT) who introduced a number of characteristics that enabled MIT students to perform basic parametric analyses for thesis and coursework in cancer epidemiology. In this thesis, the CancerFit application has been extended to include new functionality that allows computation and subsequent analysis of the ratio of two age-specific cancer incidence or mortality datasets. Originally this ratio was proposed to compare the lifetime risks of children of parents with a specific form of cancer and of parents with at least one child with the same cancer; it does this task as intended. However, its use has permitted me to discover a previously unrecognized excess of colon cancer deaths in women relative to men in the 30-64 year age interval. As this is the same age interval for early breast cancer and ovarian cancer onset in women, this finding points to a more general cancer risk in pre-menopausal women than has been previously recognized. Furthermore, the CancerFit program has been improved by permitting the cancer researcher to include historical age-specific survival rates, overall mortality rates, and reporting error rates when these are available. A user can now input data for each of these rates, which the program uses to adjust the mortality data

(cont.) to better approximate the age-specific rate of cancer appearance for the cohort studied. These improvements and new clinical data have permitted a clearer understanding of the age-specific risks and in the case of colorectal cancer appear to permit calculation of the critical parameters in this form of human cancer.