Sentinel lymph node melanoma metastases: Assessment of tumor burden for clinical prediction of outcome in the first Multicenter Selective Lymphadenectomy Trial (MSLT-I)

Purpose: As clinical management decisions in patients with Stage III melanoma have become more complex, precise pathologic characterization of sentinel lymph node (SLN) metastases has become critical to guide management. The extent of SLN involvement correlates with risk of adverse outcomes, but reported methods of disease quantification vary. We examined SLN metastases from patients participating in an international clinical trial and compared several methods of tumor burden quantification. Methods: SLNs from 146 node-positive patients in the first Multicenter Selective Lymphadenectomy Trial (MSLT-I) were centrally-reviewed and characterized by number of tumor-positive nodes, percent nodal area tumor replacement, maximum dimension of largest metastasis, tumor penetrative depth, number of tumor foci, metastasis microanatomic location, and extracapsular extension. These data were analyzed for correlation with non-SLN metastasis and melanoma-specific survival (MSS). Results: The median number of tumor-involved SLNs was 1. The median maximum metastasis dimension was 1.11 mm. Median SLN area involvement was 1.5%. Tumor burden measures were highly correlated with each other. Factors associated with non-SLN metastasis by univariable analysis were primary tumor ulceration and extent of metastases. Tumor thickness, ulceration, non-SLN metastasis and multiple measures of SLN tumor burden were significantly related to MSS on univariable analysis. After multivariable adjustment, number of involved SLNs (p = 0.05) and percent nodal area tumor replacement (p = 0.02) were independent predictors of MSS. Conclusion: Central review of MSLT-I pathology indicates that primary tumor and SLN tumor characteristics predict non-SLN metastasis and MSS. Percent nodal involvement was more powerfully prognostic than the more commonly used maximum dimension of largest metastasis.