Background: Sentinel lymph node biopsy (SLNB) is the most sensitive procedure for assessing nodal status in patients with primary melanoma.
Objective: To evaluate the predictive ability of usual primary melanoma prognosis factors of detecting sentinel lymph node (SLN) metastasis in patients with melanoma.
Patients and methods: A cohort of 612 consecutive patients presenting with primary skin melanoma who underwent a SLNB was evaluated. Assessment of the determinants of SLN metastasis was based on general linear model analysis. The model performance was studied using the concordance statistic and the net reclassification index. The calibration was estimated using the Hosmer-Lemeshow test.
Results: The discrimination ability did not differ significantly between Breslow thickness (0.57), Clark index (0.61), ulceration (0.57) and histological subtype (0.55). Clark index, ulceration and Breslow thickness were all significant and independent determinants of SLN metastasis. The predictive ability of the final model was 0.657.
Conclusion: Breslow thickness, Clark index and ulceration are independent predictors of a SLN metastasis.