Modeling intercellular interactions during carcinogenesis.

Abstract

By modulating the microenvironment of malignant or premalignant cells, inhibitory or stimulatory signals from nearby cells can play a key role in carcinogenesis. However, current commonly used quantitative models for induction of cancers by ionizing radiation focus on single cells and their progeny. Intercellular interactions are neglected or assumed to be confined to unidirectional radiation bystander effect signals from cells of the same tissue type. We here formulate a parsimoniously parameterized two-stage logistic (TSL) carcinogenesis model that incorporates some effects of intercellular interactions during the growth of premalignant cells. We show that for baseline tumor rates, involving no radiation apart from background radiation, this TSL model gives acceptable fits to a number of data sets. Specifically, it gives the same baseline hazard function, using the same number of adjustable parameters, as does the commonly used two-stage clonal expansion (TSCE) model, so it is automatically applicable to the many data sets on baseline cancer that have been analyzed using the TSCE model. For perturbations of baseline rates due to radiation, the models differ. We argue from epidemiological and laboratory evidence, especially results for the atomic bomb survivors, that for radiation carcinogenesis the TSL model gives results at least as realistic as the TSCE or similar models, despite involving fewer adjustable parameters in many cases.