Among the problems is our relative inability to identify the cells in the several compartments. For many tissues, the stem cell populations are still unknown or structurally indistinguishable from related cell populations. It is probably not correct to assume that all cells that can divide or form adducts are necessarily at risk of transformation. We need biologic markers to identify the susceptible cell populations. The intermediate cell populations are also often difficult to identify. The many putative preneoplastic lesions associated with the carcinogenic process include few for which a causal association has been demonstrated. Malignant cells themselves might be difficult to identify until a tumor clone has grown enough for histopathologic diagnosis.
Measuring birth and death processes and transition rates requires identification of the cells in the several compartments. Birth processes are relatively easy to measure with existing methods, but methods for measuring programmed and unprogrammed cell death are still under development. In addition, the intermediate cell clones themselves are not always homogeneous, and cells can differ considerably from one another in biologic potential.
Those considerations and many others (including the doses to the target cells and interindividual differences in chemical metabolism) apply not only to laboratory animals, but also to humans, for whom similar information on cell kinetics is required. Humans pose the additional complication of greater heterogeneity (genetic and environmental) with individual variability in susceptibility to tumor formation at various body sites. To assess risk, one needs information not only about processes that take place in unexposed subjects, but also about the effect of various doses on the processes themselves. That is true, regardless of the dose-response modeling procedures used.
Criteria for Adoption
Before the two-stage model can be adopted for routine health risk assessments, chronic bioassay methods will have to be changed to generate the necessary data. It will be helpful, too, to evaluate the methods through a series of studies that use various agents in multiple animal strains or species. Prospective hypothesis-testing studies are preferable to retrospective model-fitting exercises.