Cancer risk assessment in historical perspective

During the late 1970s, a foundational paradigm shift occurred within environmental regulation (EPA 1976). The US Environmental Protection Agency (EPA) abandoned the FDA-long employed Mantel-Bryan log-probit model in favor of Crump's Linearized Multistage (LMS) model, establishing the Linear Non-Threshold (LNT) framework as the scientific default for carcinogen risk assessment. This article evaluates the historical developments and mathematical strategies utilized to secure the acceptance and long-term regulatory utility of the LMS model. By tracing the parameter evolution from unmeasurable biological transitions () to maximum likelihood estimates () and analyzing the structural re-engineering of the polynomial degree (), we demonstrate that the model's low-dose linearity is an algebraic artifact of high-dose experiments rather than a reflection of low-dose biology. We highlight the operational essentiality of the "additivity to background" assumption, showing how the mathematical derivation of mediated this biological assumption into policy/practice. Furthermore, we document the erosion of support for the LMS model inside the toxicological community and EPA, culminating in the 2005 Cancer Guidelines and the adoption of the Point of Departure (PoD) framework. Finally, we examine how recent molecular reappraisals have challenged the core assumption of "additivity to background mechanism," leaving the low-dose derivative at zero ().