Dose-finding design and benchmark for a right censored endpoint

Abstract

Dose-finding trials aim to determine a safe dose to be tested in larger trials for efficacy. In oncology, designs generally assume conventional monotonic increasing dose-toxicity relationships, mostly with binary outcomes (e.g., dose-limiting toxicity or not), measured in the first cycle of therapy or for a fixed number of cycles. However, with new anti-cancer agents such as molecularly targeted therapies and immunotherapies, late-onset toxicities have become more frequent. Designs with prolonged observation windows and censored endpoints analyzed using survival models, appear particularly suited to these settings. Moreover, in this setting, the observation of the lateonset toxicity endpoint could be precluded by trial discontinuation due to death, progression, patient withdrawal, or physician discretion, defining a competing event to toxicity. We propose extensions of the Continual Reassessment Method (CRM) dose-finding design using survival working models for right-censored endpoints and for handling treatment discontinuation in the toxicity observation window, namely the Survival-CRM (Surv-CRM) and the informative survival-CRM (iSurv-CRM). We also developed a benchmark approach for its evaluation. In a simulation study, we compared the performance of the Surv-CRM and iSurv-CRM, to those of the Time-toevent (TITE)-CRM and the nonparametric benchmark. The performance of the proposed methods was consistent with the complexity of scenarios as assessed by the nonparametric benchmark. Without treatment discontinuations, the Surv-CRM provides proportions of correct dose selection close to those of the TITE-CRM with fewer observed toxicities and patients assigned to overtoxic dose levels. In the presence of treatment discontinuation, the iSurv-CRM outperforms the TITE-CRM in identifying the correct dose level.

A Road Map for Designing Phase I Clinical Trials of Radiotherapy–Novel Agent Combinations

“…Overall, 29 of 37 papers described designs whereby MTD determination was based on toxicity only, with the majority (29/37) categorizing DLTs as a binary outcome over a fixed period. Some designs extended this to incorporate adaptations such as the inclusion of time to treatment discontinuation ( 23 ) or wait times between patients ( 24 ). Only two papers described designs incorporating the dose escalation of both radiotherapy dose/schedule and drug ( 25, 26 ), and three incorporated formal subgroup analysis ( 27–29 ).…”

Surv‐CRM‐12: A Bayesian phase I/II survival CRM for right‐censored toxicity endpoints with competing disease progression

“…To evaluate performances of our proposed design, we developed a nonparametric benchmark approach providing an upper bound estimate on selecting the OD under a given scenario, and at a fixed sample size 12,30,31 . We revised the benchmark approach proposed by Cheung 31 for complex designs with a bivariate dose‐finding objective, to account for right‐censored endpoints and competing risks events.…”

“…However, considering progression as an independent censoring event ignores this information related to efficacy, and may impact the operating characteristics of the design, with decreased performances in selecting the correct dose 11 . Accounting for any discontinuation event as informative censoring, using the cumulative incidence of DLT in a survival CRM setting improved the design operating characteristics compared to the binomial TITE‐CRM 12 . The information provided by efficacy‐related outcomes, such as disease progression, could be also used in the process of selecting a dose, resulting in phase I/II trial designs 13‐20 .…”

Seamless phase I/II design for novel anticancer agents with competing disease progression

“…However, we showed that the operating characteristics of the design remain impacted to some extent, with decreased performances in correctly selecting the MTD and limiting overdose selection, while progression is considered as an independent censoring event with no use made of this efficacy information 19 . We found that accounting for any discontinuation event as informative censoring, using the cumulative incidence of dose‐limiting toxicities (DLT) in a survival CRM setting improved the design operating characteristics compared with the binomial TITE‐CRM 20 …”