Comment on Wages et al, Coherence principles in interval‐based dose finding. Pharmaceutical Statistics 2019, doi:10.1002/pst.1974

作者：D. C. Norris · 发表于：Pharmaceutical statistics · 年份：2020 · DOI：10.1002/pst.2016 · 研究领域：Medicine

To “stud[y] the principle of coherence in interval-based dose-finding methods,” Wages et al carry out simulations which by their very construction negate the supposed importance of the peculiar coherence notion of Cheung which motivated them. Yet at the same time, these simulations contain the seeds for a newer precautionary coherence concept that addresses modern imperatives of precision medicine and patient-centeredness in dose finding. To construct their single-trial simulation, Wages et al introduce a “latent toxicity tolerance” ui U[0, 1] that characterizes individual patients. The authors, however, immediately discount ui as mere technical artifice: “Of course, in a real trial, it is impossible to observe a patient's latent tolerance, but it is a useful tool in simulation.” Yet the very real I-PREDICT trial, where patients “received escalating doses of drugs to tolerance, while being monitored closely by their treating physicians,” belies this ready denial of the possibility of clinical titration. I contend that ui is best understood as the cumulative distribution function of a titration-based concept: the individualized maximum tolerated dose (MTDi). 5 If Wages et al would allow that their ui admits this realistic interpretation, what becomes of the coherence principle they sought to study? Cheung's problem, “where the objective is to estimate a targeted quantile of the unknown dose-toxicity curve,” now translates as seeking an arbitrarily selected quantile of the popu...