Global and mitosis-specific interobserver variation in mitotic count scoring and implications for malignant melanoma staging.

Abstract

Staging is the gold standard for predicting malignant melanoma outcome but changes in its criteria over time indicate ongoing evolution. One notable recent change from the 8th edition of the American Joint Committee on Cancer (AJCC) staging manual was removal of mitotic count. We explore the extent to which this feature is limited by interobserver error in order to find ways to improve its fitness for use should it be revisited in future staging versions. In a cohort of 476 patients with melanoma ≤1.0mm, a mitotic count of 0 versus 1 was significant for metastasis-free survival, but not melanoma-specific or overall survival. In 10 melanomas that were 0.9-1.0mm thick, the mitotic count intraclass correlation coefficient for histopathologists was 0.58 (moderate agreement). Uniquely, we also assessed agreement for specific putative mitotic figures, identifying precise reasons why specific mitotic figures qualified for scoring or elimination. A kappa score was 0.54 (moderate agreement). We also gathered data on other staging features. Breslow thickness had an intraclass correlation coefficient of 0.41 (moderate agreement) and there was a systematic difference between histopathologists among cases (P=0.04). Every case had a range that crossed the AJCC8 0.8-mm pT1a/pT1b staging boundary. Ulceration was only identified in two of the 10 cases. For ulceration, kappa agreement score was 0.31 (fair). This study supports the removal of mitotic count from staging, but shows that its scoring is substantially affected by interobserver variation, suggesting that more prescriptive guidelines might have a beneficial impact on its prognostic value.