New approaches in quantitative mechanistic modeling of radiation-induced carcinogenesis.

摘要

Humans have always been exposed to natural sources of ionizing radiation, for example to radon gas rising from the ground, cosmic rays bombarding the Earth from space, and radioactive isotopes in the soil and inside the human body itself. In recent decades the average yearly doses of ionizing radiation to the US population have grown mainly due to increases in man-made exposure sources. Chief among these new sources of radiation are diagnostic medical procedures, e.g. computed tomography (CT), nuclear medicine, and fluoroscopy. Cancer radiotherapy also contributes, especially because the number of treated patients is increasing and their life expectancy after therapy is becoming longer. The nuclear power industry and space travel represent sources of occupational irradiation, and the potential of radiological terrorism is a grim prospect for malicious population-scale exposure.;A unified approach of integrating short- and long-term methods is needed because it would combine the advantages of both model classes: a detailed dose response would be generated by the short-term component of the formalism, even for a complicated irradiation scheme such as cancer radiotherapy, and modulation of this dose response over time after exposure until cancer diagnosis would be tracked by the long-term component. Developing and implementing such a combined short-long-term model of spontaneous and radiation-induced carcinogenesis is the purpose of this dissertation.;After the introduction (Chapter I), the short-term processes were addressed by formalisms in the class called iip or iir models: such models focus on the production of pre-malignant cells by radiation (initiation), killing of normal and pre-malignant cells by radiation (inactivation), and repopulation of surviving cells (proliferation or repopulation). A deterministic iip model was applied to radiogenic leukemias (Chapter II). An overview of iip modeling prospects for solid tumors was provided next (Chapter III), followed by use of a stochastic iip formalism on radiogenic second cancer risks in survivors of Hodgkin's disease (Chapter IV).;Next, plausible assumptions about long-term processes were grafted on to the short-term framework. The long-term equations use a classic two-stage approach, tracking the numbers of pre-malignant (stage 1) cells until the appearance of the first fully malignant (stage 2) cell, which can give rise to cancer. Novel features of the long-term formalism include: (1) emphasis on stem cell niches and clones rather than individual cells and (2) stem cell aging along with aging of the whole organism. The combined short-long-term formalism was applied to cancer risk data in atomic bomb survivors and radiotherapy patients, and to background cancer incidence (Chapters V and VI). Modified versions of the model were then used to analyze mouse carcinogenesis data (Chapter VII) and radon-induced lung cancer in rats and humans (Chapter VIII). Recent data on cancer risks from atomic bomb survivors were analyzed to clarify the mechanisms behind age at exposure dependencies of these risks (Chapter IX). Finally, future model applications, e.g. for second cancer risk minimization during radiotherapy treatment planning, were discussed (Chapter X).;The task of radiation risk estimation is a difficult one, particularly for complex exposure scenarios such as cancer radiotherapy. The fractionation schemes and dose distributions are changing rapidly, making extrapolation of risks from decades-old protocols to current practices suboptimal. Risks induced by newer protocols cannot be assessed directly because of the long latency period for many cancers (e.g. 10 years). This situation can be addressed by biologically-based mathematical models of spontaneous and radiation-induced carcinogenesis. Such models estimate cancer risk by implementing plausible assumptions about the mechanisms of carcinogenesis, e.g. the population dynamics of pre-malignant cells under background conditions and in irradiated individuals. Model parameter values can be obtained from the literature and/or by fitting the formalism to data on background cancer rates and radiation-induced risks. The calibrated model can then be used to predict cancer risks associated with modern or prospective irradiation protocols.;Taken together, these results suggest that the combined short-long-term modeling approach is a promising method for predicting radiogenic cancer risks from radiotherapy as well as from other exposure sources such as radon, and for interpreting the underlying biological mechanisms. (Abstract shortened by UMI.)