Improving basal cell carcinoma subtype diagnosis with dermoscopy and reflectance confocal microscopy: a multicentre prospective study

Background: Basal cell carcinoma (BCC) is the most prevalent form of skin cancer. Accurate histological subtyping is essential for appropriate therapeutic planning, particularly for infiltrative variants associated with higher recurrence rates. Noninvasive imaging modalities such as dermoscopy and reflectance confocal microscopy (RCM) may enhance the diagnostic precision of BCC subtyping prior to treatment. Objectives: To validate dermoscopic criteria and identify RCM features associated with distinct BCC subtypes in a larger cohort and to improve the accuracy of noninvasive BCC subtyping and support precision treatment planning. Methods: This was a prospective multicentre study conducted between August 2017 and June 2019 across three centres in Northern Italy. In total, 281 histologically confirmed BCCs from 256 patients were analysed. Each lesion underwent standardized clinical, dermoscopic and handheld RCM evaluation prior to biopsy. Subtype-specific features were recorded and correlated with histopathology. Interobserver agreement was assessed using Cohen’s kappa, and diagnostic performance was evaluated via sensitivity, specificity and the area under the receiver operating characteristic curve (AUC). The study was registered at Clinicaltrials.gov (NCT04789421). Results: Among the 281 BCCs, 15% were superficial, 38% were nodular and 47% were infiltrative. Dermoscopically, superficial BCCs exhibited brown globules and shiny white–red structureless areas; nodular BCCs presented blue structures, arborizing vessels and ulceration; infiltrative BCCs showed white porcelain areas and lacked pigmentation. RCM identified streaming and cords in superficial BCCs (P = 0.003; threefold risk), large tumour islands in nodular BCCs, and dark silhouettes in infiltrative BCCs (P = 0.005; two–threefold risk). Diagnostic accuracy improved with combined dermoscopy and RCM. Sensitivity for superficial BCC increased from 74.4% [95% confidence interval (CI) 60.2–85.8] to 81.4% (95% CI 68.1–91.0), specificity from 97.9% (95% CI 95.5–99.5) to 98.3% (95% CI 96.1–99.5), and AUC reached 0.899. Conclusions: The integration of dermoscopic and RCM findings significantly enhances the noninvasive preoperative classification of BCC subtypes. This approach improves diagnostic accuracy and may inform more precise, subtype-tailored management, reducing treatment failures and supporting personalized dermatological care.