Clinico-pathologic features of primary melanoma and sentinel lymph node predictive for non-sentinel lymph node involvement and overall survival in melanoma patients: a single centre observational cohort study.

OBJECTIVE Completion Lymph Node Dissection (CLND) is the current standard of practice for patients with a positive Sentinel Lymph Node Biopsy (SLNB). Significant morbidity is associated to CLND, so we tried to evaluate which prognostic variables could predict NSLN invasion in SLN-positive patients and their impact on the overall survival (OS). METHODS A retrospective chart review of 603 patients that had undergone SLNB for melanoma between 2000 and 2009 at our department was done. 100 SLN were positive at the histopathological analysis of SLN. Demographic variables, primary melanoma, SLN pathologic features and results of CLND were analysed. Multivariate logistic regression and OS analyses were carried out to test the prognostic relevance of clinico-pathologic variables on CLND results and disease course. RESULTS Breslow thickness, ulceration and micro/macrometastatic pattern of SLN invasion carried a significantly independent higher likelihood of NSLN involvement; Starz classification did not maintain a statistical significance in multivariate analysis. Only one patient (4.3%) without adverse prognostic factors showed NSLN involvement, which was found in 33.3% of patients with one and 55.9% with two or more adverse parameters (p = 0.0001). OS analyses confirmed the prognostic significance of these factors. CONCLUSION Waiting for the results of Multicenter Selective Lymphadenectomy Trial II, our study suggests a clinically useful and easily applicable means of identifying patients with an unfavourable disease course. The presence of one or more adverse factors identifies patients in whom CLND is mandatory to include thereafter in a more strict follow-up program. Moreover, the finding of no adverse prognostic indicators associated to the presence of significant co-morbidities and/or elderly age, could be useful in identifying patients not to treat by CLND.