Related Experiment Video
Updated: Jul 18, 2026
Establishing a Competing Risk Regression Nomogram Model for Survival Data
Published on: October 23, 2020
Development and validation of a nomogram for predicting cardiovascular mortality risk for diffuse large B-cell
Kai Mu1,2, Jing Zhang2, Yan Gu2
1Pediatric Heart Center, Children's Hospital of Fudan University, Shanghai, China.
Objective:
This study aimed to construct and validate a nomogram for predicting cardiovascular mortality (CVM) for child, adolescent, and adult patients with diffuse large B-cell lymphoma (DLBCL).
Materials And Methods:
Patients with only one primary tumor of DLBCL first diagnosed between 2000 and 2019 in the SEER database were extracted. We used the cumulative incidence function (CIF) to evaluate the cumulative rate of CVM. The outcome of interest was CVM, which was analyzed using a competing risk model, accounting for death due to other causes. The total database was randomly divided into a training cohort and an internal validation cohort at a ratio of 7:3. Adjustments were for demographics, tumor characteristics, and treatment modalities. Nomograms were constructed according to these risk factors to predict CVM risk at 5, 10, and 15 years. Validation included receiver operating characteristic (ROC) curves, time-dependent ROC, C-index, calibration curves, and decision curve analysis.
Results:
One hundred four thousand six hundred six patients following initial diagnosis of DLBCL were included (58.3% male, median age 64 years, range 0-80, White 83.98%). Among them, 5.02% died of CVM, with a median follow-up time of 61 (31-98) months. Nomograms based on the seven risk factors (age at diagnosis, gender, race, tumor grade, Ann Arbor stage, radiation, chemotherapy) with hazard ratios ranging from 0.19-1.17 showed excellent discrimination, and calibration plots demonstrated satisfactory prediction. The 5-, 10-, and 15-year AUC and C-index of CVM in the training set were 0.716 (0.714-0.718), 0.713 (0.711-0.715), 0.706 (0.704-0.708), 0.731, 0.727, and 0.719; the corresponding figures for the validation set were 0.705 (0.688-0.722), 0.704 (0.689-0.718), 0.707 (0.693-0.722), 0.698, 0.698, and 0.699. Decision curve analysis revealed a clinically beneficial net benefit.
Conclusions:
We first built the nomogram model for DLBCL patients with satisfactory prediction and excellent discrimination, which might play an essential role in helping physicians enact better treatment strategies at the time of initial diagnosis.
