Circulating tumor DNA as a predictor for response to treatment in BRAF V600E mutant malignant melanoma.

Abstract

9564 Background: Available biomarkers LDH and S100B possess limited sensitivity and specificity to predict outcome in melanoma. In this pilot study we evaluated the use of circulating tumor (ct)DNA harboring BRAF and NRAS mutations as a predictive biomarker for treatment response and progression-free survival (PFS) in patients with locally advanced or metastatic melanoma. Methods: We analyzed 89 retrospective plasma samples from 32 unselected pts, and 158 samples from 12 pts included in a prospective trial (DRKS00009507). We included stage III disease with planned resection or stage IV disease before initiation or change of medical treatment. Blood samples were taken at baseline at d +8, d +28, and thereafter at 3 months intervals for up to two years. We developed a hydrolysis probe based, Locked Nucleic Acid assay to detect BRAF V600E and wild type ctDNA by droplet digital PCR. Results were correlated with LDH, S100B and PFS. Results: Sensitivity of BRAF V600E specific assay was 0.01% with a limit of Blank of 0.28 copies/well. Of 31 stage IV pts with retrospective samples, 23 were positive for BRAF V600E ctDNA at least once (74%). Positive pts had a mean of 9 (range: 1-17) and 483 (range: 0.1-16,388) BRAF V600E copies/mL for stage III and stage IV respectively. The presence of ctDNA at baseline predicted poor PFS (hazard ratio [HR] 1.487, 95% CI 0.34-6.45). A negative slope in BRAF V600E ctDNA was a favorable prognostic factor for PFS (hazard ratio [HR] 0.230, 95% CI 0.04-1.20) with a median PFS of 3.42 vs. 2.56 months (Range 1.87-8.9 vs. 0.89-5.02). Residual ctDNA at the first time point after initiation or change of treatment was related to a shorter PFS (hazard ratio [HR] 2.02, 95% CI 0.39-10.53). Based on 144 measurement pairs, BRAF ctDNA strongly correlated with S100 (r = 0.73) and LDH (r = 0.52). Conclusions: Residual ctDNA early after change or institution of treatment predicted tumor progression at first clinical response assessment. A positive to negative conversion or a decrease indicated a more favorable course. These data support the use of ctDNA as an early predictive marker for treatment response. We will examine whether two or more detected mutations indicate clonal heterogeneity and confer adverse prognosis. Clinical trial information: DRKS00009507.