Preoperative Ultrasound Assessment of Regional Lymph Nodes in Melanoma Patients Does not Provide Reliable Nodal Staging: Results From a Large Multicenter Trial.

OBJECTIVE: To assess whether preoperative ultrasound (US) assessment of regional lymph nodes in patients who present with primary cutaneous melanoma provides accurate staging. BACKGROUND: It has been suggested that preoperative US could avoid the need for sentinel node (SN) biopsy, but in most single-institution reports, the sensitivity of preoperative US has been low. METHODS: Preoperative US data and SNB results were analyzed for patients enrolled at 20 centers participating in the screening phase of the second Multicenter Selective Lymphadenectomy Trial. Excised SNs were histopathologically assessed and considered positive if any melanoma was seen. RESULTS: SNs were identified and removed from 2859 patients who had preoperative US evaluation. Among those patients, 548 had SN metastases. US was positive (abnormal) in 87 patients (3.0%). Among SN-positive patients, 39 (7.1%) had an abnormal US. When analyzed by lymph node basin, 3302 basins were evaluated, and 38 were true positive (1.2%). By basin, the sensitivity of US was 6.6% (95% confidence interval: 4.6-8.7) and the specificity 98.0% (95% CI: 97.5-98.5). Median cross-sectional area of all SN metastases was 0.13 mm2; in US true-positive nodes, it was 6.8 mm2. US sensitivity increased with increasing Breslow thickness of the primary melanoma (0% for ≤1 mm thickness, 11.9% for >4 mm thickness). US sensitivity was not significantly greater with higher trial center volume or with pre-US lymphoscintigraphy. CONCLUSION: In the MSLT-II screening phase population, SN tumor volume was usually too small to be reliably detected by US. For accurate nodal staging to guide the management of melanoma patients, US is not an effective substitute for SN biopsy.