How best to allocate backfill patients to doses in an ethical way to maximise the use of their information in oncology phase I dose-finding clinical trials.

Background Historically phase I dose-finding oncology trials aimed to explore the toxicity profile of a new treatment.

Among several dose levels of the treatment explored, a maximum tolerated dose (MTD) was selected based on toxicity data for further investigation in later phase trials. Although cytotoxic agents have an increasing dose-efficacy curve, toxicities increase too.

However, for non-cytotoxic agents (e.g. molecular targeted agents and immune therapy agents) the dose-efficacy curve may reach a plateau, resulting in several dose levels having similar efficacy with possibly less toxicities.

Nowadays, it is of paramount importance to detect the recommended phase II dose (RP2D) using both efficacy and toxicity data, as the MTD may not be the most appropriate dose for patients.

Moreover, patient-reported outcomes (PROs); such as self-reported adverse events (AE) and health related quality-of-life, contain crucial information to complement the treatment's tolerability and capture patients' satisfaction. Unfortunately, PROs are rarely used to identify RP2D in oncology dose-finding trials.

An increasing number of trials have adopted new ways of defining the RP2D by collecting information from backfill patients (BP) dosed at levels lower than the one dose-finding patients receive.

Including BPs increases the sample size, allowing more information to be collected and assists in accurately identifying the RP2D.

Limited methodological research has been done around allocating BPs to doses incorporating their efficacy, and using PROs & BPs toxicities to identify RP2D.

Research question How can we improve the allocation scheme of BPs in an ethical way to maximise the use of their information into RP2D identification in phase I oncology dose-finding trials. Aims and objectives Investigate different approaches of allocating BPs to dose levels in an ethical way. Investigate how to integrate BPs' toxicities to identify RP2D.

Investigate how to incorporate PROs to identify RP2D alongside efficacy and toxicity data. Establish the impact of the efficacy endpoint format in RP2D identification.

Methods and timelines for delivery Work package 1 - Exploring the performance of allocating BPs in unequal groups or in a pair of doses or using weight to randomise them in previous doses to maximise their information into the trial. (Months 1-8) Work package 2 - Incorporate BPs' toxicities within the analysis to inform RP2D, using Continual Reassessment Method model-based & Bayesian Optimal Interval model-assisted adaptive designs. (Months 8-19) Work package 3 - Incorporate patients' self-reported side effects or European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire C30 alongside efficacy and toxicity data to identify RP2D in trials with BPs. (Months 20-30) Work package 4 - Determination of the impact of efficacy endpoint format (binary, e.g. objective response versus continuous biomarker) on identifying the RP2D from both dose-finding patients and BPs. (Months 31-36) Impact and dissemination This project will generate robust methodologies for statisticians and clinicians to design phase I dose-finding trials using BPs.

Research findings will be disseminated via conferences/workshops/webinars and publications. Lay members will help identify channels and simple ways to communicate findings to the public.