Longitudinal BRAF ctDNA monitoring in melanoma reflects disease status and predicts relapse

BACKGROUND: Early detection of melanoma relapse after curative-intent surgery remains challenging, and routine follow-up relies on imaging and imperfect serum biomarkers. OBJECTIVE: To assess whether longitudinal BRAF-mutant circulating tumour DNA (ctDNA) reflects disease status and whether time-updated ctDNA detection and concentration are associated with relapse hazard during post-resection surveillance. METHODS: We conducted a prospective real-world cohort study including 127 AJCC stage II-IV melanoma patients with tumour-confirmed BRAF V600-mutant melanoma. We analysed 318 plasma samples. BRAF-mutant ctDNA was quantified by droplet digital PCR using the Bio-Rad BRAF V600 Screening Kit. Disease status was classified as no evidence of disease (NED), locoregional disease, or metastatic disease. Serial ctDNA detection was analysed using logistic generalised estimating equations. In the post-resection surveillance cohort, relapse-free interval was analysed from the first post-resection surveillance sample using time-dependent Cox models with ctDNA updated serially. RESULTS: ctDNA detection was higher in locoregional disease (OR 5.90, 95% CI 2.34-14.88; P < 0.001) and metastatic disease (OR 7.21, 95% CI 2.86-18.21; P < 0.001) than in NED. During post-resection surveillance, detectable ctDNA (HR 17.17, 95% CI 6.12-48.14; P < 0.001) and higher ctDNA concentration (HR 1.53 per doubling, 95% CI 1.36-1.71; P < 0.001) were associated with increased relapse hazard. CONCLUSION: Longitudinal BRAF p.V600E/K/R ctDNA monitoring reflects disease status and identifies surveillance intervals with increased relapse risk, despite limited sensitivity. These findings support further refinement of ctDNA-based approaches to complement standard melanoma follow-up.