Detection of actionable BRAF missense mutations by ctDNA-based genomic analysis in prostate cancer.

Abstract

306 Background: Activating BRAF fusion proteins are rare in prostate cancer (PCa) patients. Driver missense BRAF mutations have not been reported in detail in this population. Methods: We examined ctDNA-derived genomic profiles (Guardant 360) from 2,721 unique PCa patients, to identify BRAF genomic anomalies (SNVs, amplification). The ctDNA results were compared with PCa tissue-based genomics from the TCGA database (1,851 unique patients). Results: BRAF missense mutations were found in 76 ctDNA patients (2.8%) and were from all known mutation classes (I, II, III) as well as variants of unknown significance (VUSs). Only 4 patients had the V600E mutation. Multiple examples of known, autonomously active, non-canonical mutations were found (27), including K601E (12), G469A (5), D594G (2), and G466E (2). There were 45 VUSs. Mutations were primarily clonal but subclonal mutations were also found. In addition BRAF was commonly amplified, usually in the presence of multiple other amplified genes. BRAF missense mutations were more common with ctDNA than TCGA (2.8% vs 1.4%). Neither dataset identified frequent V600E mutations (ctDNA: 4/2,721; TCGA 1/1,851). However patients with the same non-canonical BRAF mutations were found in each dataset (K601E, G469A, G466E, D594G). Each dataset contained unique mutations found in only one patient. BRAF mutations potentially treatable with BRAF or MEK inhibitors (class I, II) were about half of all mutations (ctDNA 40.8%; TCGA 50%). We treated a PCa patient with a clonal BRAF(G469A) mutation with targeted therapy. The patient was resistant to multiple lines of hormonal and cytotoxic therapy. Trametinib produced a clinical and RECIST response. Conclusions: ctDNA-based genomic analysis identified multiple BRAF amplifications and missense SNVs in PCa patients. SNVs are largely non-canonical, but include known activating mutations that could act as drivers. The analysis also identified more BRAF missense mutations than did tissue genomic profiling, but the mutational landscape, overall frequency of mutations was similar with either method. ctDNA-based genomic profiling can identify actionable BRAF driver mutations that may respond to MEK and BRAF inhibitors.