Melanoma Recurrence Prediction Using a Tissue-Free Epigenomic Molecular Residual Disease Assay: a Multicenter Prospective Observational Study (COSMOS-MEL01)

Despite recent adoption of perioperative therapy for high-risk melanoma, its overall survival benefit remains unclear, highlighting the need for biomarkers to identify patients likely to benefit from treatment. Detection of circulating tumor DNA (ctDNA) after definitive treatment predicts recurrence, and emerging molecular residual disease (MRD) assays may influence management. We validated a tissue-free MRD assay in 48 patients with clinical stage II-III melanoma treated with surgery (52% non-acral cutaneous, 31% acral, and 17% mucosal). The assay detects MRD through bioinformatic analysis of methylation signals across >20,000 epigenomic regions. Plasma samples (n = 281; 280 successfully reported) were prospectively collected before surgery, at Day 28, and every 3-6 months thereafter. Associations between MRD detection, recurrence, and overall survival were analyzed. At a median follow-up of 34.6 months with 27 recurrence events, postoperative surveillance sensitivity for distant recurrence was 81% with 100% post-treatment specificity. Median lead time between MRD detection and radiographic recurrence was 70.5 days. Surveillance MRD detection was associated with increased recurrence risk (HR 26.55, 95% CI 2.86-246.58, P = 0.004). Similarly, MRD detection 28 days postoperatively independently predicted recurrence (HR 11.70, 95% CI 3.09-43.76, P <0.001) and death (HR 31.14, 95% CI 5.37-221.1, P <0.001). Tissue-free epigenomic MRD detection was sensitive, specific, and significantly associated with recurrence and melanoma-related death. These findings support use of this assay in melanoma as a rapid, noninvasive approach for earlier identification of patients at high risk for recurrence, enabling refined treatment and surveillance strategies.