BackgroundIn light of the evolving landscape of adjuvant therapy in melanoma and the recently confirmed absent survival benefit of completion lymph node dissection (CLND), it becomes important to explore possible consequences of omitting CLND, and whether it is possible to adequately stratify positive sentinel node (SN) patients solely based on information retrieved from the melanoma up to the sentinel lymph node biopsy (SLNB). MethodsA retrospective cohort from nine European Organization for Research and Treatment of Cancer Melanoma Group centres was used. Patients were staged based on SLNB and CLND result according to the American Joint Committee on Cancer (AJCC) criteria and stratified by ulceration and SN tumour burden. These were incorporated in Cox regression models. Predictive ability was assessed using Harrell's concordance index (c-index) and the Akaike information criterion (AIC). ResultsIn total, 1015 patients were eligible. CLND led to upstaging in N-category in 19% and in AJCC stage in 5–6%. The model incorporating only ulceration and SN tumour burden performed equally well as the model incorporating substages after CLND. The model incorporating substages based on SLNB had the lowest predictive ability. Stratifying by ulceration and SN tumour burden resulted in four positive SN groups from which low-, intermediate- and high-risk prognostic classes could be derived. ConclusionsAdequate stratification of positive SN patients was possible based on ulceration and SN tumour burden category. The identification of low-, intermediate- and high-risk patients could guide adjuvant therapy in clinical practice. Omitting CLND seems to have little consequences.