Factors predictive of tumor-positive nonsentinel lymph nodes after tumor-positive sentinel lymph node dissection for melanoma

Purpose Approximately 20% of sentinel node (SN) positive melanoma patients have additional non-SN (NSN) metastasis. The rationale for this study was to identify the factors associated with additional nodal disease, as a method to determine which patients may most benefit from completion lymph node dissection (CLND). Patients and Methods During 1990 to 2002, 1,599 patients have undergone SN biopsy at our institute. 19.5% underwent CLND for tumor-positive SN. One hundred ninety-one of these patients had clinicopathologic information available for review. Univariate analyses used chi(2) test, Wilcoxson rank sum test, and chi(2) test for trend. Multivariate analyses used logistic regression and Wald test. Results Forty-six (24%) patients had tumor-positive NSN. Univariate analyses showed that primary thickness (Breslow and Clark), primary site, SN tumor size, and number of tumor-positive SNs were significantly associated with tumor-positive NSN. Multivariate analysis (167 patients), confirmed that Breslow and SN tumor size were independently predictive. Sex, histology, ulceration, mitotic index, and SN basin location were not predictive. Risk stratification by the number of prognostic factors present (Breslow greater than or equal to 3 mm and SN tumor size greater than or equal to 2 mm) showed that probability of finding tumor-positive NSN was 12.3% in the low-risk group (0 factors), 30.9% in the intermediate-risk group (1 factor), and 41.9% in the high-risk group (2 factors). Conclusion Thicker primary and larger SN tumor size are factors that correlate best with tumor-positive NSN. Although none of these factors are absolutely predictive of residual nodal disease, these factors must be strongly considered if the SN contains metastasis, as they provide enhanced risk assessment for NSN tumor-positivity. (C) 2004 by American Society of Clinical Oncology.