Risk stratification prior to neck dissection for stage III or higher head and neck malignant melanoma: A retrospective UK study of 83 neck dissections.

Abstract

Head and neck (H&N) melanomas account for approximately one-fifth of cutaneous melanomas and carry higher mortality owing to complex lymphatic drainage and increased invasiveness. However, predictors of nodal metastasis specific to this region remain poorly defined. To identify clinicopathological predictors of nodal metastasis in patients undergoing neck dissection for H&N melanoma. A retrospective cohort study of 83 patients undergoing neck dissection at Whiston Hospital (2017-2025), classified as ND-positive (n=45) or ND-negative (n=38). Univariable and multivariable logistic regression with Firth's penalised MLE, Kaplan-Meier survival analysis, and Cox proportional hazards regression were performed. Larger tumour size (median 16.0 vs 10.0 mm, p=0.011), greater Breslow thickness (3.9 vs 2.3 mm, p=0.017) and lymphovascular invasion (26.3 vs 5.4%; OR 6.25, p=0.014) were significantly associated with ND-positive outcome. Cumulative epithelioid morphology across all preoperative biopsies was the sole independent predictor of nodal metastasis (aOR 25.67, 95% CI 3.55-185.60, p=0.001). Among sentinel lymph node biopsy (SLNB)-positive patients, larger metastatic deposit size (3.0 vs 0.6 mm, p=0.014) and subcapsular deposit diameter (3.0 vs 0.5 mm, p=0.015) predicted ND-positive outcome. Eight patients developed nodal metastasis despite negative SLNB. Cumulative epithelioid morphology assessment is the strongest independent predictor of nodal metastasis in H&N melanoma. The high SLNB false-negative rate in this region supports risk-adapted surveillance imaging regardless of SLNB status, with early therapeutic neck dissection considered in selected high-risk cases.