Background
Sentinel lymph node biopsy (SLNB) is widely used in melanoma. Identifying nodal involvement preoperatively using high-resolution ultrasound may offer less invasive staging. This study assessed feasibility and staging results of clinically-targeted ultrasound (before lymphoscintigraphy) compared to SLNB.
Methods
From 2005–2009, 325 patients with melanoma underwent ultrasound before SLNB. We reviewed demographics and histopathologic characteristics, then compared ultrasound and SLNB results. Sensitivity, specificity, positive and negative predictive value (PPV, NPV) were determined.
Results
325 patients, 58% male; 42% female, median age 58 years (range 18–86) were included. A total of 471 basins were examined with ultrasound. Only 6 patients (1.8%) avoided SLNB by undergoing ultrasound-guided fine needle aspiration of involved nodes, then therapeutic lymphadenectomy. 65 patients (20.4%) had 69 SLNB positive nodal basins; 17 nodal basins from 15 patients with positive ultrasounds were considered true-positive. 45 SLNB positive basins had negative ultrasounds (false-negative). 7 node-positive basins did not undergo ultrasound because of unpredicted drainage. 253 patients with negative SLNBs had negative ultrasounds in 240 nodal basins (true-negative) but false-positive ultrasounds occurred in 40 basins. Overall, sensitivity of ultrasound was 33.8%, specificity 85.7%, PPV 36.5% and NPV 84.2%. Sensitivity and specificity improved somewhat with increasing Breslow depth. Sensitivity was highest for the neck, but specificity was highest for the groin.
Conclusions
Routine preoperative ultrasound in clinically node-negative melanoma is impractical due to low sensitivity. Selected patients with thick or ulcerated lesions may benefit. Because of variable lymphatic drainage patterns, preoperative ultrasound without lymphoscintigraphic localization will provide incomplete evaluation in many cases.