Treatment profile and prognostic modeling in early-stage Merkel cell carcinoma: a retrospective cohort study

Merkel cell carcinoma (MCC) is a rare but highly aggressive neuroendocrine skin cancer. The prognostic associations of different treatment approaches remain incompletely defined. This study aimed to characterize treatment patterns, evaluate survival outcomes, and develop prognostic models for early-stage MCC. We analyzed SEER data from 2004 to 2021 for patients aged > = 20 years with node-negative, non-metastatic primary cutaneous MCC. Associations with overall survival (OS) and cancer-specific survival (CSS) were evaluated using Cox regression and inverse probability of treatment weighting (IPTW). A Fine-Gray competing-risk analysis was additionally performed for MCC-specific mortality in relation to adjuvant radiation therapy. Prognostic models were developed using Cox regression and XGBoost and evaluated using Harrell's C-index and time-dependent AUCs. Among 5742 patients, surgery was the predominant treatment modality. In direct comparisons among surgical approaches, gross excision showed only a modest and model-dependent association with OS and no consistent association with CSS relative to wide excision, whereas Mohs micrographic surgery (MMS) was associated with higher hazards of all-cause and MCC-specific mortality. Adjuvant radiation therapy was associated with improved OS but not CSS. In the competing-risk analysis, the adjusted subdistribution hazard of MCC-specific death was modestly higher among patients receiving adjuvant radiation (sHR 1.17, 95% CI 1.02-1.34; P = 0.026). In the testing cohort, the Cox and XGBoost models achieved OS C-indices of 0.66 and 0.67, respectively. Treatment modality was associated with survival outcomes in early-stage MCC, although the observational treatment comparisons should be interpreted cautiously because of potential residual confounding. Cox and XGBoost models showed comparable prognostic discrimination in the testing cohort and may support further development of risk-stratification approaches.