Reply to A. Stenzinger et al.

Abstract

Basket trials investigate the effects of targeted therapies against specific molecular alterations across multiple histologic subtypes, including tumors arising in parallel from different primary organs. The rationale is that the presence of a specific molecular aberration would predict the response to a targeted therapy regardless of the histologic subtype. Patient enrollment in basket trials is based on the molecular profiles of their tumors. The CUSTOM (Molecular Profiling and Targeted Therapies in Advanced Thoracic Malignancies) trial used a basket trial design to simultaneously evaluate targeted therapies in patients with advanced non–small-cell lung cancer, small-cell lung cancer, and thymic malignancies; patients were grouped by molecular alterations along with tumor histology. Stenzinger et al raise concerns about the viability of basket trial designs in their current form: directing targeted treatments of patients with disparate histologies on the basis of single genetic alterations to address the mutational complexity of cancer. We fully agree. As Stenzinger et al point out, the biologic relevance of molecular drivers has been reported to vary between tumors arising from different primary sites; for example, the effect of BRAF V600E is different in nevi, malignant melanoma, and colorectal cancer. Further extending these observations are reports that show the effect of coexisting mutations in determining response to the same targeted therapy, even among primary cancers that harbor identical genetic mutations. In KRAS-mutant non–small-cell lung cancer, for example, therapeutic response to a combination of the MEK inhibitor selumetinib and docetaxel has been reported to vary, depending on concurrent alterations in tumor suppressor genes, including P53 and LKB1; in KRAS-/P53-mutant mice, selumetinib plus docetaxel led to responses in 61% of mice whereas only 33% of mice with KRAS-/ LKB1-mutant cancers achieved a partial response with the combination. Indeed, most basket trial designs ignore the biologic context not only with respect to coexisting mutations but also with respect to other interactions that may have an impact on the clinical response, including the proteome, epigenome, and the immune milieu. Basket trials are a relatively new and evolving area in oncology trial design. Treatment arms of basket trials will show clinical efficacy only when the tumor depends on the target pathway and the targeted therapy reliably inhibits the target. Basket trial designs are therefore best used to identify early signals of clinical activity of targeted therapies with a small number of patients and to validate a clinical target. This is particularly useful when the cancer type or the mutation is rare. So where do we go from here with basket trials? Future basket trials, as indicated by Stenzinger et al, should use more comprehensive characterization of molecular alterations to overcome limitations of existing histology-based and single genetic alteration–based approaches. The CUSTOM trial is a necessary first step toward identifying better clinical trial designs to investigate in parallel the clinical efficacy of multiple targeted therapies in multiple cancers.