Interobserver variability in histopathologic diagnosis of melanocytic neoplasms

Background Precise histopathologic diagnosis of melanocytic neoplasms is required to guide appropriate clinical management; however, significant interobserver variability remains a challenge and may lead to suboptimal management. Objective To quantify interobserver variability in melanocytic lesion diagnosis and identify factors influencing discordance. Methods This retrospective cohort study examines 1491 melanocytic lesions with diagnoses rendered by an external institution that were subsequently reviewed in consultation at our institution. We used an ordinal classification system to examine the magnitude and directionality of diagnostic discordance and used machine learning to assess feature importance in influencing discordance. Results Overall diagnostic discordance was 33%. Indeterminate-risk lesions (melanocytic proliferation with uncertain malignant potential) with explicit treatment recommendations exhibited high (79.2%; 122/154) discordance, with most (72.1%; 82/122) cases downgraded to benign entities, which may be consequential for treatment planning. Academic practice setting and dermatopathology board certification were associated with diagnostic concordance, while intent-to-consult was associated with discordance. Limitations Single-institution, retrospective study design precludes correlation with outcomes and diagnostic accuracy assessment, and referral bias may limit generalizability. Conclusion Significant interobserver variability persists in melanocytic lesion diagnosis, underscoring the need for refined classification criteria and robust biomarkers to improve diagnostic accuracy and optimize patient management.