Individualized Prediction and Risk Factors of Recurrence in Chinese Patients with Sebaceous Carcinoma: A Multicenter Study of 418 Patients

Background: Recurrent eyelid sebaceous carcinoma (SC) after primary resection still remains a great challenge for ophthalmologists. The predictors of recurrence are multifactorial, the identification of core risk factors to construct an individualized prediction model is warranted. The purpose of this study is to develop and validate a nomogram for individualized recurrence prediction of eyelid SC and to determine the independent risk factors. Methods: A multicenter cohort study. The study included 418 consecutive patients with eyelid SC, and this sample was divided into training (n=293) and validation cohorts (n=125). Least absolute shrinkage and selection operator (LASSO) regression was applied to select the features for the nomogram. The model was evaluated using the receiver operating characteristic (ROC)-derived area under the curve (AUC), calibration plot, and decision-curve analyses (DCAs), and it was compared with the TNM staging system. These results were externally validated with bootstrap resampling in an independent cohort. Multivariate Cox regression was used to explore the independent predictors of recurrence. Findings: This nomogram displayed satisfactory discriminative ability and good calibration for both the training (C-index: 0.83; AUC: 0.84) and validation (C-index: 0.80; AUC: 0.82) cohorts. The discriminative ability compared significantly favorable than TNM staging (C-index: training cohort: 0.67, validation cohort: 0.71; AUC: training cohort: 0.67, validation cohort: 0.71; all p