A nomogram construction and multicenter validation for predicting overall survival after fruquintinib application in patients with metastatic colorectal cancer: a multicenter retrospective study
Related Experiment Video
Updated: Jun 11, 2025
Competing-Risk Nomogram for Predicting Cancer-Specific Survival in Multiple Primary Colorectal Cancer Patients after Surgery
Published on: September 27, 2024
Xiao-Xuan Wang1, Yu-Wen Zhou1, Bo Wang1
1Colorectal Cancer Center, West China Hospital, Sichuan University, Chengdu, Sichuan, China.
Background:
Fruquintinib is a third-line and subsequent targeted therapy for patients with metastatic colorectal cancer (mCRC). Identifying survival predictors after fruquintinib is crucial for optimizing the clinical use of this medication.
Objectives:
We aimed to identify factors influencing the prognosis of patients with mCRC treated with fruquintinib and to leverage these insights to develop a nomogram model for estimating survival rates in this patient population.
Design:
Multicenter retrospective observational study.
Methods:
We collected patient data from January 2019 to October 2023, with one healthcare institution's data serving as the training cohort and the other three hospitals' data serving as the multicenter validation cohort. The nomogram for overall survival was calculated from Cox regression models, and variable selection was screened using the univariate Cox regression analysis with additional variables based on clinical experience. Model performance was measured by the concordance index (C-index), calibration curves, decision curve analyses (DCA), and utility (patient stratification into low-risk vs high-risk groups).
Results:
Data were ultimately collected on 240 patients, with 144 patients included in the training cohort and 96 included in the multicenter validation cohort. Predictors included in the nomogram were CA199, body mass index, T stage, the primary site of the tumor, and other metastatic and pathological differentiation. The C-index of the nomogram in the training set and multicenter validation was 0.714 and 0.729, respectively. The models were fully calibrated and their predictions aligned closely with the observed data. DCA curves indicated the promising clinical benefits of the predictive model. Finally, the reliability of the model was also verified through the risk classification using the nomogram.
Conclusions:
We constructed a nomogram for mCRC treated with fruquintinib based on six variables that may be used to assist in personalizing the use of the drug.
