A comparative analysis of the EPA's 2005 cancer risk assessment restructuring vs. the historical Crump linearized...

In 2005, the US EPA finalized its Guidelines for Carcinogen Risk Assessment, transitioning from the 1986 Linearized Multistage (LMS) framework to the Benchmark Dose (). This shift leveraged software to fit empirical curves to biological data, with a focus on chronic animal bioassay cancer studies. This paper conducts a meta-analysis evaluating the "before and after" risk metrics for 28 carcinogens during this transition. These findings demonstrate that for linear extrapolations, the exposure limits generated by the consistently land within a single log-scale factor () of LMS estimates for 89.3% of chemicals (25 of 28). Furthermore, 64.3% of the cohort (18 of 28) exhibits near numerical parity (), with several agents showing equivalence. We show that cases of divergence () are driven by changes in biological inputs (e.g., human vs. animal datasets, multi-organ tumor aggregation) rather than low-dose curve-fitting. While this restructuring offers theoretically improved model data fits for sublinear and threshold carcinogens, the findings indicate that the BMD approach, like the LMS model, is still primarily limited by the use of the few-dose/high-dose NTP protocol. The present findings are in close agreement with previous studies comparing LMS and BMD risk estimates for a wide range of EPA carcinogens.