RESUMEN
BACKGROUND: Two RCTs found no survival benefit for completion lymphadenectomy after positive sentinel lymph node biopsy compared with observation with ultrasound in patients with melanoma. Recurrence patterns and regional control are not well described for patients undergoing observation alone. METHODS: All patients with a positive sentinel node biopsy who did not have immediate completion lymphadenectomy were identified from a single-institution database (1995-2018). First recurrences were classified as node only, local and in-transit (LCIT) only, LCIT and nodal, or systemic. Regional control and factors associated with recurrence survival were analysed. RESULTS: Median follow-up was 33 months. Of 370 patients, 158 (42·7 per cent) had a recurrence. The sites of first recurrence were node only (13·2 per cent), LCIT only (11·9 per cent), LCIT and nodal (3·5 per cent), and systemic (13·8 per cent). The 3-year postrecurrence melanoma-specific survival rate was 73 (95 per cent c.i. 54 to 86) per cent for patients with node-only first recurrence, and 51 (31 to 68) per cent for those with initial systemic recurrence. In multivariable analysis, ulceration in the primary lesion (hazard ratio (HR) 2·53, 95 per cent c.i. 1·27 to 5·04), disease-free interval 12 months or less (HR 2·38, 1·28 to 4·35), and systemic (HR 2·57, 1·16 to 5·65) or LCIT and nodal (HR 2·94, 1·11 to 7·79) first recurrence were associated significantly with decreased postrecurrence survival. Maintenance of regional control required therapeutic lymphadenectomy in 13·0 per cent of patients during follow-up. CONCLUSION: Observation after a positive sentinel lymph node biopsy is associated with good regional control, permits assessment of the time to and pattern of recurrence, and spares lymphadenectomy-related morbidity in patients with melanoma.