Fig. 4: IMPACT2 study algorithm.

IMPACT2. Study design Part A, May 2014-March 2017 A Patients with metastatic cancer undergo a tumor biopsy and genomic profiling. B If patients have targetable molecular aberrations and FDA-approved drugs within labeled indication are available, the patients will receive FDA-approved targeted therapy. C If patients have targetable molecular aberrations and there are no FDA-approved drugs within labeled indication, the patients are considered for commercially available targeted therapy or clinical trial. Patients are presented at tumor board. D Patients are offered randomization to one of two arms: MTT or non-MTT E Criteria (biomarker present, available clinical trial, eligibility criteria met) are met. F Analysis. IMPACT2. Study design Part B, March 2019-present. A Patients with metastatic cancer undergo a tumor biopsy and genomic profiling. B If patients have targetable molecular aberrations and FDA-approved drugs within labeled indication are available, the patients will receive FDA-approved targeted therapy. C If patients have targetable molecular aberrations and there are no FDA-approved drugs within labeled indication, the patients are considered for commercially available targeted therapy or clinical trial. Patients are presented at tumor board. D Patients are offered randomization between two arms: MTT or non-MTT E Criteria (biomarker present, available clinical trial, eligibility criteria met) are met. F In March 2019, we amended the trial to include a “patient-preference” cohort for each arm. Patients who decline randomization are offered their choice of arm. G The primary analysis will use both randomized and patient-preference cohorts based on a Bayesian hierarchical model that “borrows” from the patient-preference cohorts to the extent to which its PFS agrees with that in the randomization cohort.