Abstract
Outcomes for melanoma patients vary within cancer stage. Prognostic biomarkers are potential adjuncts to provide more precise prognostic information. Simple, low-cost biomarker assays, such as those based on immunohistochemistry, have strong translational potential. 5-hydroxymethylcytosine (5 hmC) shows prognostic potential in melanoma but prior studies were small. We, therefore, analysed 5 hmC in a retrospective cohort to provide external validation of its prognostic value. Two hundred primary melanomas were evaluated for 5 hmC expression using immunohistochemistry. The primary objective was to assess the effect on overall survival while controlling for important confounders. Univariable and multivariable analyses were performed. REMARK guidelines were followed. The 5 hmC immunohistochemistry scoring showed very strong inter-observer agreement (ICC 0.88) and expression was significantly related to age, site, Breslow thickness, ulceration, mitotic rate, and stage. Kaplan-Meier analysis showed 5 hmC was associated with metastasis-free, melanoma-specific, and overall survival, P<0.0001 for each. In univariable Cox proportional hazards models, 5 hmC hazard ratios were significant and remained so in a multivariable model. A two-step cox model was created using stage and 5 hmC, as stage is the gold standard for clinical practice. The addition of 5 hmC produced significant improvement in the model and 5 hmC and stage were independent significant predictors. This is the largest study of the prognostic value of 5 hmC immunohistochemistry in melanoma. The 5 hmC scoring was easily and reproducibly performed and it was an independent predictor of metastasis-free survival, melanoma-specific survival, and overall survival. This work supports further development of 5 hmC as a prognostic biomarker and suggests that it could add more precision to American Joint Committee on Cancer staging.