Surveillance of Sentinel Node-Positive Melanoma Patients with Reasons for Exclusion from MSLT-II: Multi-Institutional Propensity Score Matched Analysis

Background: In sentinel lymph node (SLN)-positive melanoma, two randomized trials demonstrated equivalent melanoma-specific survival with nodal surveillance vs completion lymph node dissection (CLND). Patients with microsatellites, extranodal extension (ENE) in the SLN, or >3 positive SLNs constitute a high-risk group largely excluded from the randomized trials, for whom appropriate management remains unknown. Study design: SLN-positive patients with any of the three high-risk features were identified from an international cohort. CLND patients were matched 1:1 with surveillance patients using propensity scores. Risk of any-site recurrence, SLN-basin–only recurrence, and melanoma-specific mortality were compared. Results: Among 1,154 SLN-positive patients, 166 had ENE, microsatellites, and/or >3 positive SLN. At 18.5 months median follow-up, 49% had recurrence (vs 26% in patients without high-risk features, p < 0.01). Among high-risk patients, 52 (31%) underwent CLND and 114 (69%) received surveillance. Fifty-one CLND patients were matched to 51 surveillance patients. The matched cohort was balanced on tumor, nodal, and adjuvant treatment factors. There were no significant differences in any-site recurrence (CLND 49%, surveillance 45%, p = 0.99), SLN-basin–only recurrence (CLND 6%, surveillance 14%, p = 0.20), or melanoma-specific mortality (CLND 14%, surveillance 12%, p = 0.86). Conclusions: SLN-positive patients with microsatellites, ENE, or >3 positive SLN constitute a high-risk group with a 2-fold greater recurrence risk. For those managed with nodal surveillance, SLN-basin recurrences were more frequent, but all-site recurrence and melanoma-specific mortality were comparable to patients treated with CLND. Most recurrences were outside the SLN-basin, supporting use of nodal surveillance for SLN-positive patients with microsatellites, ENE, and/or >3 positive SLN.