37P - Dose transition pathways for time-to-event continual reassessment method (TITE-CRM): Will imposing a waiting time result in better performance?

Background

Novel agents, such as immunotherapies, are likely to cause late-onset toxicities. The inclusion of these toxicities can lead to prolonged trials since traditional phase I designs require full follow-up information for each participant. TITE-CRM, an extension of the CRM design, permits continuous recruitment of participants by incorporating incomplete follow-up data to determine the recommended dose for the next participant. Thus, it can shorten trial duration, but it may also treat participants at suboptimal doses due to the uncertainty of the safety at the current dose. Imposing a waiting window to allow for more safety data to accumulate may assign the next participant to a higher and potentially more effective dose if no DLT is observed. Our aim is to create a practical tool to help trialists “look ahead” to see if imposing a waiting window would make a difference to the model’s dose recommendation.

Methods

We propose the dose transition pathways (DTP) for TITE-CRM design with possible waiting window (DTP-TITE-CRM). The DTP is a practical tool to visualize the dose recommendations for subsequent cohort of participants and was first proposed by Yap et al 2017. In DTP-TITE-CRM, the DTP projects all possible recommendations until complete follow-up is achieved. The user can also specify the maximum time for a participant to wait before being assigned to a dose.

Results

Simulations show that DTP-TITE-CRM has comparable accuracy as TITE-CRM in most scenarios in terms of identifying the MTD. In settings where the MTD is at the highest dose, or all doses are tolerable, DTP-TITE-CRM outperforms TITE-CRM. We illustrate the use of DTP-TITE-CRM in a published trial which used TITE-CRM, where imposing additional waiting time could have led to different dose decisions.

Conclusions

DTP-TITE-CRM looks ahead and enumerates dose recommendations in advance and gives participants and trialists the choice to wait or to treat immediately. This approach is particularly appealing when patient numbers are limited and when the tested agents are expected to have a tolerable toxicity profile, as it allows more participants to be treated at higher doses if safe.

Clinical trial identification

Editorial acknowledgement

Legal entity responsible for the study

C. Yap.

Funding

National Institute for Health Research (Biomedical Research Centre at The Royal Marsden NHS Foundation Trust and The Institute of Cancer Research, London).

Disclosure

All authors have declared no conflicts of interest.