Dose-response relationships in chemical carcinogenesis: superposition of different mechanisms of action, resulting in linear-nonlinear curves, practical thresholds, J-shapes.

The shape of a carcinogen dose-cancer incidence curve is discussed as the result of a superposition of dose-response relationships for various effects of the carcinogen on the process of carcinogenesis. Effects include direct DNA damage, e.g., by covalent binding, indirect DNA damage, e.g., by increased formation of reactive oxygen species or interaction with DNA replication or chromosome integrity. The 'fixation' of a DNA adduct as a heritable mutation depends on its pro-mutagenic potency and on the rates of DNA repair and DNA replication. Endogenous and unavoidable DNA damage is responsible for a background rate of the process of mutagenesis and carcinogenesis and forms the basis of spontaneous cancer incidence. For DNA-reactive carcinogens, linearity of the dose response at the low-dose end is expected. With increasing dose, saturation of DNA repair can introduce a sublinearity (example: dimethylnitrosamine). Stimulation of cell division as a result of high-dose toxicity and regenerative proliferation also results in a sublinear deviation from low-dose linearity. If the DNA-damaging potency of the carcinogen is low in comparison with the high-dose effects, the linear part of the low dose-cancer incidence curve might be hidden within the background variability. Under such conditions, 'practical thresholds' could be discussed (formaldehyde). If a carcinogen increases the rate of cell division or the level of oxidative stress at high dose but has an antimitogenic or antioxidative effect at low dose, a J-shaped (or: U-shaped) curve with a decrease of the spontaneous tumor incidence at low dose could result (caffeic acid; TCDD). This phenomenon has been observed even under conditions of a genotoxic contribution (ionizing radiation; diesel exhaust particles). For a mechanism-based assessment of a low-dose cancer risk, information on the various modes of action and modulations should be available over the full dose range, and models should be refined to incorporate the respective information.