Co-clinical assessment identifies patterns of BRAF inhibitor resistance in melanoma

Multiple mechanisms have been described that confer BRAF inhibitor resistance to melanomas, yet the basis of this resistance remains undefined in a sizable portion of patient samples. Here, we characterized samples from a set of patients with melanoma that included individuals at baseline diagnosis, on BRAF inhibitor treatment, and with resistant tumors at both the protein and RNA levels. Using RNA and DNA sequencing, we identified known resistance-conferring mutations in 50% (6 of 12) of the resistant samples. In parallel, targeted proteomic analysis by protein array categorized the resistant samples into 3 stable groups, 2 of which were characterized by reactivation of MAPK signaling to different levels and 1 that was MAPK independent. The molecular relevance of these classifications identified in patients was supported by both mutation data and the similarity of resistance patterns that emerged during a co-clinical trial in a genetically engineered mouse (GEM) model of melanoma that recapitulates the development of BRAF inhibitor resistance. Additionally, we defined candidate biomarkers in pre- and early-treatment patient samples that have potential for predicting clinical responses. On the basis of these observations, we suggest that BRAF inhibitor-resistant melanomas can be actionably classified using protein expression patterns, even without identification of the underlying genetic alteration.