Combining Dermoscopy and Reflectance Confocal Microscopy to Improve External Ear Melanoma Diagnosis: A Multicenter Retrospective Study

BACKGROUND: External ear melanoma remains poorly characterized by non-invasive imaging. We characterized dermoscopic and reflectance confocal microscopy (RCM) features of auricular melanocytic lesions and evaluated whether combining both modalities improves diagnosis. METHODS: This multicenter retrospective study included 77 auricular lesions evaluated at three tertiary centers. Blinded readers assessed predefined dermoscopic and RCM criteria. Firth-penalized logistic regression was used to develop dermoscopy-only, RCM-only, and combined models. Model development and discrimination were internally validated with 2000 bootstrap replications. RESULTS: Thirty melanomas, one melanocytic tumor of uncertain malignant potential, one intraepidermal atypical melanocytic proliferation of uncertain significance, and 45 nevi were analyzed. Lentigo maligna/lentigo maligna melanoma-like imaging features were observed among melanomas of several histopathological subtypes. The combined model included blue-white veil (odds ratio [OR] 10.79, 95% confidence interval [CI]: 1.72-67.72), regression structures (OR 30.16, 95% CI: 3.53-257.97), dendritic pagetoid cells (OR 5.12, 95% CI: 1.25-21.08), and medusa head-like structures (OR 10.50, 95% CI: 2.11-52.18). It outperformed the dermoscopy-only model (apparent area under the curve [AUC] 0.931 vs. 0.852, DeLong p = 0.047, Holm-adjusted p = 0.047). Selection-aware optimism-corrected AUCs were 0.899 and 0.831, respectively. Excluding the two lesions of uncertain malignant potential produced comparable results (AUC 0.952 vs. 0.863, Holm-adjusted p = 0.040). CONCLUSIONS: Auricular melanocytic lesions warrant site-specific and sublocation-aware evaluation. Combining dermoscopy with RCM improved diagnostic discrimination over dermoscopy alone in this cohort, although prospective external validation is required.