Nomograms for predicting cancer-specific and overall survival in patients with invasive extramammary Paget's disease

Aim: To develop nomograms for predicting cancer-specific survival (CSS) and overall survival (OS) in patients with invasive extramammary Paget's disease (iEMPD). Patients & methods: Retrospective data of 1955 patients with iEMPD were collected from the Surveillance, Epidemiology, and End Results database. Nomograms for predicting CSS and OS were established using competing risk regression and Cox regression, respectively, and were internally validated. Results: Five (age, surgery, tumor location, stage and concurrent malignancy) and eight (gender, age, race, marital status, surgery, tumor location, stage and lymph node metastasis) clinicopathological factors were utilized to construct nomograms for predicting CSS and OS, respectively. The concordance indices of the nomograms for predicting CSS and OS were 0.78 and 0.73, respectively. The validation of the nomograms showed good calibration and discrimination. The decision curve analyses confirmed the clinical utility of these nomograms. Conclusion: The nomograms can be a reliable tool for treatment design and prognostic evaluation of iEMPD. Lay abstract Invasive extramammary Paget's disease (iEMPD) is a rare type of cutaneous malignancy with a heterogeneous prognosis. The prognostic factors remain poorly described, resulting in unclear risk stratification of the patients with iEMPD. The purpose of this study is to identify the prognostic factors associated with cancer-specific and overall survival rates in iEMPD and to develop accurate risk stratification models to guide the design of individualized treatment regimens. Clinicopathological data of 1955 patients pathologically diagnosed with iEMPD were retrospectively collected from the Surveillance, Epidemiology, and End Results database, and were utilized for analysis and construction of models for predicting the long-term survival in patients with iEMPD. Eventually, five (age, surgery, tumor location, stage and concurrent malignancy) and eight (gender, age, race, marital status, surgery, tumor location, stage and lymph node metastasis) factors were chosen to develop models for predicting cancer-specific and overall survival, respectively. The prediction accuracy and clinical utility of the established models were confirmed in subsequent evaluation. Because iEMPD is an extremely rare disease that a lot of clinical practitioners might not be familiar with, the availability of these quantifiable predictive models will provide convenience in daily practice.