A nomogram for predicting overall survival in patients with Appendiceal mucinous carcinoma: A population-based analysis

Abstract

Abstract Background Appendiceal mucinous adenocarcinoma (AMA) is a rare tumor and prognostic prediction models have rarely been reported. The aim of our study was to establish and evaluate a nomogram to predict the overall survival of AMA patients. Methods We selected patients diagnosed with AMA from 2000 to 2020 from the Surveillance, Epidemiology, and End Results (SEER) database. They were randomized in a 7:3 ratio to be further divided into a training cohort and a validation cohort. Univariate and multivariate COX regression analyses were used to select prognostic independent risk factors and further to select variables for nomogram. The validity of the nomogram was assessed using the consistency index (C index), area under the curve of time-dependent ROC curves (time-dependent AUC), and calibration curves. The net benefit of nomograms with different threshold probabilities was quantified using Decision Curve Analysis (DCA) and compared to the net benefit of AJCC standard-based tumor staging. The Net Reclassification Index (NRI) and Integrated Discriminatory Improvement (IDI) were also employed to compare the clinical use of the nomogram with AJCC standard-based tumor staging. Comparison of risk stratification for nomogram and AJCC standard-based tumor staging. Results A total of 2489 patients were enrolled according to the admission and exclusion criteria and divided into training cohort (n = 1739) and validation cohort (n = 750) in a 7:3 ratio. Six variables were selected to establish the nomogram of AMA based on univariate and multivariate regression analyses. C-index (0.724 for the training cohort, and 0.693 for the validation cohort) and time-dependent AUC (> 0.7) indicated that the nomogram had a satisfactory discriminatory ability. In both the training and validation cohorts, the calibration curves demonstrated a decent agreement between the predictions of the nomogram and the actual observations. the NRI values (training cohort: 0.308 for 1 year, 0,300 for 3 years, 0.187 for 5 years OS prediction; validation cohort:0.178 for 1 year, 0.172 for 3 years and 0.148 for 5 years OS prediction) and IDIs (training cohort: 0.052 for 1 year, 0.076 for 3 years, 0.069 for 5 years OS prediction; validation cohort: 0.035for 1 year, 0.053 for 3 years, 0.058 for 5 years OS prediction) indicated that the constructed nomogram significantly outperforms tumor staging based on the AJCC standard alone (P < 0.05). Furthermore, DCA showed that the nomogram has clinical utility to identify patients at higher risk than AJCC standard-based tumor staging. Conclusion We established and validated a prognostic nomogram to help clinicians evaluate the prognosis of patients with AMA. In the future, we hope that more external validation will be added to increase its efficacy.