ABSTRACT
BACKGROUND: Non-sentinel node (NSN) positivity impacts the prognosis of melanoma patients; however, the benefits of completion lymph node dissection in patients with positive sentinel nodes (SNs) are limited. OBJECTIVE: We aimed to present a predictive nomogram for NSN positivity in melanoma patients with a positive SN biopsy. METHODS: This retrospective analysis from patients who underwent SN biopsy in a Brazilian institution from 2000 to 2015 was used for the construction and internal validation of the nomogram. This nomogram was then externally validated in a cohort of Dutch patients. RESULTS: The Brazilian cohort comprised 1213 patients, with a mean follow-up of 5.11 years. Breslow thickness (odds ratio [OR] 1.170, 95% confidence interval [CI] 1.043-1.314]; p = 0.008), number of positive SNs (OR 1.092, 95% CI 1.034-1.153; p = 0.001), and largest diameter of the metastatic deposit (OR 3.217, 95% CI 1.551-6.674; p = 0.002) were statistically significant for NSN positivity. Internal validation was performed using a bootstrapping technique. A good performance was observed (Brier score 0.097) and an excellent power of discrimination was achieved (area under the curve [AUC] 0.822). The nomogram was then applied to the Dutch cohort, and its overall performance (Brier score 0.085), calibration (Hosmer-Lemeshow goodness-of-fit test; p = 0.198), and discriminatory power (AUC 0.752, 95% CI 0.615-0.890) were all adequate. CONCLUSIONS: We presented a nomogram for assessing NSN probability that should not only be used for surgical considerations but also for risk stratification and clinical decisions. Internal validation has shown that this is an adequate model, while external validation increases the model's reliability and suggests that it can be globally incorporated.