Dermoscopy and reflectance confocal microscopy of solitary flat pink lesions: A new combined score to diagnose amelanotic melanoma

Abstract Background Differential diagnosis of amelanotic/hypomelanotic melanoma among solitary flat pink lesions is challenging, due to limited clinical and dermoscopic clues. Dermoscopy and reflectance confocal microscopy assessments improve diagnostic accuracy, but their combined capacity among solitary flat pink lesions is yet to be defined. Objectives To determine (i) whether diagnostic accuracy is improved with combined dermoscopy and reflectance confocal microscopy, (ii) a model to estimate probability of flat amelanotic/hypomelanotic melanoma among solitary flat pink lesions. Methods A retrospective single‐centre study of solitary flat pink lesions, excised for suspected malignancy between 2011 and 2022 was performed. Images were independently evaluated by two dermatologists, blinded to histopathological diagnosis. Diagnostic performance was evaluated on the receiver operating characteristic curve and the area under the curve. Predictive features were identified by univariate and multivariate logistic regression analyses. A final predictive nomogram of independent risk factors was calculated by backward likelihood ratio. Hypothesis being tested was formulated before data collection. Results A total of 184 patients (87 females, 47.3%) were included; mean age was 57.6 years (19–95). Combined dermoscopy and reflectance confocal microscopy was more sensitive (83%, CI 69.2–92.4 and 91.5%, CI 79.6–97.6) than dermoscopy alone (76.6%, CI 62.0–87.7 and 85.1%, CI 71.7–93.8). Predictive features defined the new model, including linear irregular vessels (4.26‐folds, CI 1.5–12.1), peripheral pigment network (6.07‐folds, CI 1.83–20.15), remnants of pigmentation (4.3‐folds, CI 1.27–14.55) at dermoscopy and atypical honeycomb (9.98‐folds, CI 1.91–51.96), disarranged epidermal pattern (15.22‐folds, CI 2.18–106.23), dendritic pagetoid cells in the epidermis (3.77‐folds, CI 1.25–11.26), hypopigmented pagetoid cells (27.05‐folds, CI 1.57–465.5), and dense and sparse nests (3.68‐folds, CI 1.24–10.96) in reflectance confocal microscopy. Diagnostic accuracy of the model was high (AUC 0.91). Conclusions Adjunctive reflectance confocal microscopy increases diagnostic sensitivity of flat amelanotic/hypomelanotic melanoma differential diagnosis. The proposed model requires validation.