Genetic biomarker-driven trials are powerful ways to test targeted drugs, but are often complicated by the rarity of the biomarker-positive population. “Umbrella” trials circumvent this issue by testing multiple hypotheses to maximize accrual. However, allocation strategy (ie which drug to administer first, in case multiple actionable mutations coexist) greatly affects sample size and should be carefully planned based on relative mutation frequencies. Inadequate planning results in lack of statistical power and inconclusive trials. The bigger the trial, the higher the chance of conflicting drug allocation. Biomarker-based trial design may be facilitated by leveraging data from public sequencing projects, but no specific computational tool is available.
We developed the package Precision Trial Designer (PTD) in the R programming language. We used TCGA data to show its potential in a simulated 10-arm imaginary trial on multiple cancers, based on genetic alterations suggested by the 2017 Molecular Analyses for Personalised Therapy (MAP) conference. We validated PTD predictions versus real data from the SHIVA trial (Lancet Oncol 2015). Finally, we deployed PTD as an open access, web-based app using the Shiny-R Studio platform.
PTD estimates parameters useful for trial design, most importantly 1) the fraction of patients with ≥ 1 actionable alteration 2) the frequency matrix of mutation co-occurrence 3) the number of patients needed to molecularly screen (NNMS) for a given design (time-to-event or proportion-based) 4) the allocation rule that maximizes patient accrual (systematically assigning patients to the drug associated with the rarer mutation). In the MAP imaginary trial, PTD “optimal” design reduces NNMS by up to 71.8% (3.5x) vs non-optimal designs. In SHIVA, the fraction of patients with actionable alteration was correctly predicted (33.51% vs 32.92%, within 95% confidence interval), as well as allocation to specific treatment groups and statistical power.
PTD correctly estimates crucial parameters to overcome issues in designing precision oncology trials. Its availability as an open-access web app creates a useful resource for the community of clinical researchers in the field of targeted therapy.
European Institute of Oncology
AIRC-15988, ANR-10-EQPX-03, Italian Ministry of Health RF-2013-02357231
All authors have declared no conflicts of interest.