Abstract
Purpose: The National Cancer Institute's Molecular Profiling-Based Assignment of Cancer Therapy (NCI-MPACT) randomized phase 2 clinical trial assessed the utility of applying tumor DNA sequencing to treatment selection. Here, we report the results of a companion preclinical study in patient-derived xenograft (PDX) models to evaluate how each tumor responded to each of the treatment regimens studied in the NCI-MPACT trial instead of simply to the specific regimen targeting the study-actionable mutation of interest (aMOI). Methods: Fifty-one PDX models (46 with and 5 without NCI-MPACT aMOIs) were tested against both the arm that would have been assigned in the NCI-MPACT trial as well as every other study regimen: (1) veliparib plus temozolomide or (2) adavosertib plus carboplatin (targeting the DNA repair pathway); (3) everolimus (targeting the PI3K pathway); and (4) trametinib (targeting the RAS/RAF/MEK pathway). Durability of response was measured by relative median time to tumor quadrupling event-free survival (EFSx4 ≥ 2), and duration of tumor regression. Results: Eleven of 50 models (22%) treated with veliparib plus temozolomide responded according to one or both metrics, as did 2/47 models (4.2%) treated with adavosertib plus carboplatin, and 2/46 models (4.3%) treated with trametinib; no models responded to erlotinib. Follow-up studies demonstrated that temozolomide drove the activity of the veliparib plus temozolomide combination and drug sensitivity to temozolomide correlated with MGMT deficiency. Conclusion: This prospective preclinical study confirmed the modest response rates in the NCI-MPACT clinical trial. Substantial responses to temozolomide suggest that this drug represents an effective treatment for patients with MGMT deficiency, regardless of cancer type.