An Internally Validated Prognostic Risk-Score Model for Disease-Specific Survival in Clinical Stage I and II Merkel Cell Carcinoma

Background Merkel cell carcinoma (MCC) is a rare cutaneous malignancy for which factors predictive of disease-specific survival (DSS) are poorly defined. Methods Patients from six centers (2005-2020) with clinical stage I-II MCC who underwent sentinel lymph node (SLN) biopsy were included. Factors associated with DSS were identified using competing-risks regression analysis. Risk-score modeling was established using competing-risks regression on a training dataset and internally validated by point assignment to variables. Results Of 604 patients, 474 (78.5%) and 128 (21.2%) patients had clinical stage I and II disease, respectively, and 189 (31.3%) had SLN metastases. The 5-year DSS rate was 81.8% with a median follow-up of 31 months. Prognostic factors associated with worse DSS included increasing age (hazard ratio [HR] 1.03, p = 0.046), male sex (HR 3.21, p = 0.021), immune compromise (HR 2.46, p = 0.013), presence of microsatellites (HR 2.65, p = 0.041), and regional nodal involvement (1 node: HR 2.48, p = 0.039; >= 2 nodes: HR 2.95, p = 0.026). An internally validated, risk-score model incorporating all of these factors was developed with good performance (AUC 0.738). Patients with <= 4.00 and > 4.00 points had 5-year DSS rates of 89.4% and 67.2%, respectively. Five-year DSS for pathologic stage I/II patients with > 4.00 points (n = 49) was 79.8% and for pathologic stage III patients with <= 4.00 points (n = 62) was 90.3%. Conclusions A risk-score model, including patient and tumor factors, based on DSS improves prognostic assessment of patients with clinically localized MCC. This may inform surveillance strategies and patient selection for adjuvant therapy trials.

Automated summary

A risk-score model incorporating patient and tumor factors improves prognostic assessment of Merkel cell carcinoma patients. Factors such as age, sex, immune compromise, microsatellites, and nodal involvement are associated with disease-specific survival. The model can help inform surveillance strategies and patient selection for adjuvant therapy trials.