Pharmacology‐guided rule‐based adaptive dose escalation in first‐in‐human studies

First-in-human (FIH) studies typically progress through cohorts of fixed, standard size throughout the escalation scheme. This work presents and tests a pharmacology-guided rule-based adaptive dose escalation design that aims at making 'best use' of participants in early clinical drug evaluation; it is paper-based, not requiring real-time access to computational methods. The design minimizes the number of participants exposed to dose levels with low likelihood of being therapeutically relevant. Using criteria based on dose-limiting adverse event rate and on target exposure or target pharmacodynamics, the design increases the sample size when approaching the dose range of potential clinical relevance. The adaptive escalation design was retrospectively tested on actual data from a sample of 40 recently executed FIH studies with novel small and large molecules, and it was evaluated by simulating trials with three compounds with different therapeutic windows, i.e. representing a promising, unacceptable and dubious profile. In retrospective evaluation of the adaptive escalation design, none of the cases overshot the actually reported top dose; one case resulted in a top dose that was within 20% under the estimated maximum tolerated dose in the original study. The median reduction of total number of participants per study was 38%. Trial simulations confirmed the retrospective evaluation, showing a similar performance of the adaptive escalation design compared to the conventional 6+2 design, at a reduced study size for compounds with a presumed acceptable therapeutic window. The adaptive escalation design was shown to make 'best use' of participants in FIH studies without compromising safety.