Predictors for establishing recommended phase 2 doses: analysis of 320 dose-seeking oncology phase 1 trials

Abstract

For decades, determination of the recommended Phase 2 dose (RP2D) was based on the toxicity (especially the maximum tolerated dose or MTD) experienced by patients enrolled in dose-escalating Phase 1 trials investigating anti-cancer agents. Recent studies suggest that this toxicity-based strategy is not suitable for modern anti-cancer agents. We conducted a retrospective study to identify the risk factor(s) for failing to determine the RP2D according to the MTD. We analyzed 320 recently published (1997-2008) Phase 1 trials using the maximum tolerated dose (MTD) to define the P2RD. We analyzed the current definitions of RP2D and then identified the risk factors for not establishing the RP2D using the odds ratio and 95% confidence intervals. Interactions between these risk factors were explored using the logistic regression model and CHAID algorithm. 18% of contemporary dose-seeking Phase 1 trials did not identify a RP2D. The logistic regression analysis showed that the risk factors for not identifying the RP2D were: investigation of molecular targeted therapies (RR = 3.0, p = 0.0017), lack of justification of the starting dose (RR = 5.9, p = 0.0121) and lack of definition of the MTD (RR = 8.4, p = 0.0006). The CHAID algorithm confirmed the importance of the methodological parameters. This study confirms the difficulty of determining the RP2D of molecular targeted therapy using conventional methods. However, it underlines the major importance of two methodological points: definition of the MTD and justification of the starting dose.