Competing risk nomogram and risk classification system for evaluating overall and cancer-specific survival in neuroendocrine carcinoma of the cervix: a population-based retrospective study.

Abstract

Objective: Neuroendocrine carcinoma of the cervix (NECC) is a rare malignancy with poor clinical prognosis due to limited therapeutic options. This study aimed to establish a risk-stratification score and nomogram models to predict prognosis in NECC patients.

Methods: Data on individuals diagnosed with NECC between 2000 and 2019 were retrieved from the Surveillance Epidemiology and End Results (SEER) database and then randomly classified into training and validation cohorts (7:3). Univariate and multivariate Cox regression analyses evaluated independent indicators of prognosis. Least absolute shrinkage and selection operator (LASSO) regression analysis further assisted in confirming candidate variables. Based on these factors, cancer-specific survival (CSS) and overall survival (OS) nomograms that predict survival over 1, 3, and 5 years were constructed. The receiver operating characteristic (ROC) curve, the concordance index (C-index), and the calibration curve estimated the precision and discriminability of the competing risk nomogram for both cohorts. Finally, we assessed the clinical value of the nomograms using decision curve analysis (DCA).

Results: Data from 2348 patients were obtained from the SEER database. Age, tumor stage, T stage, N stage, chemotherapy, radiotherapy, and surgery predicted OS. Additionally, histological type was another standalone indicator of CSS prognosis. For predicting CSS, the C-index was 0.751 (95% CI 0.731 ~ 0.770) and 0.740 (95% CI 0.710 ~ 0.770) for the training and validation cohorts, respectively. Furthermore, the C-index in OS prediction was 0.757 (95% CI 0.738 ~ 0.776) and 0.747 (95% CI 0.718 ~ 0.776) for both cohorts. The proposed model had an excellent discriminative ability. Good accuracy and discriminability were also demonstrated using the AUC and calibration curves. Additionally, DCA demonstrated the high clinical potential of the nomograms for CSS and OS prediction. We constructed a corresponding risk classification system using nomogram scores. For the whole cohort, the median CSS times for the low-, moderate-, and high-risk groups were 59.3, 19.5, and 7.4 months, respectively.

Conclusion: New competing risk nomograms and a risk classification system were successfully developed to predict the 1-, 3-, and 5-year CSS and OS of NECC patients. The models are internally accurate and reliable and may guide clinicians toward better clinical decisions and the development of personalized treatment plans.