Prognostic significance of ctDNA detection at the time of relapse during melanoma surveillance.

Abstract

e21543 Background: Circulating tumor DNA (ctDNA) is an emerging biomarker of progression and relapse in multiple cancers. Our earlier single-center data demonstrated that site of melanoma relapse impacted the sensitivity of ctDNA detection. In this multi-institutional study, we further examined the prognostic implications between ctDNA detection and melanoma relapse site. Methods: A retrospective cohort analysis was performed across three centers using a personalized, tumor-informed ctDNA assay (Natera) on prospectively collected plasma from patients with melanoma from December 2021 to January 2025 with longitudinal follow up. Inclusion criteria were patients with no evidence of disease (NED) following definitive surgery or clinical remission with systemic therapy; and undetectable ctDNA. Sensitivity of ctDNA detection at biopsy- and/or radiographically confirmed relapse was evaluated, along with overall survival (OS) from date of relapse to date of death or last follow-up. Analyses were stratified based on anatomic sites of relapse. Logistic regression was used to evaluate predictors of ctDNA detection. Log-rank p-values and Cox proportional hazards were calculated for OS measurements. Results: Of 284 patients on longitudinal MRD surveillance, 78 (27.5%) had confirmed relapse. Of these, 75 (96.2%) had NED following surgery and 3 (3.8%) had clinical remission with systemic therapy, with 49 (62.8%) resected stage II/III, 26 (33.3%) resected stage IV, and 3 (3.8%) unresectable stage III/IV in remission. Primary sites included 61 (78.2%) cutaneous, 12 (15.4%) mucosal, and 5 (6.4%) unknown. ctDNA was detected at relapse in 40/78 patients (51.3% sensitivity). Similar to prior data, sensitivity varied by relapse site with 84.6% (22/26) for lymph node (LN) relapse and 28.6% (6/21) for skin/soft tissue (SST) relapse. LN metastases (OR = 10.4, 95% CI 3.1 – 34.8; p < 0.001) and multiple (2+) metastatic sites (OR = 4.43, 95% CI: 1.13 – 17.4; p = 0.037) were associated with detectable ctDNA at relapse, while SST metastases (OR 0.27 (95% CI: 0.09 – 0.80; p = 0.021) were associated with an undetectable ctDNA. Median follow-up time from relapse was 13.4 months. Survival analysis from time of relapse showed decreased OS in those with undetectable versus detectable ctDNA (log-rank p = 0.041; HR 0.321, 95% CI: 0.102 – 1.011), including in the SST relapse subgroup (log-rank p = 0.0016; HR not estimable). Conclusions: Detectable ctDNA at the time of melanoma relapse appears to be associated with inferior OS. ctDNA sensitivity varied by site of relapse, with improved survival observed for those with undetectable ctDNA at relapse. These findings support the prognostic relevance of ctDNA while highlighting important site-specific limitations, warranting validation in larger prospective cohorts.