Related Experiment Video
Updated: Jun 6, 2025
Detection of a Circulating MicroRNA Custom Panel in Patients with Metastatic Colorectal Cancer
Published on: March 14, 2019
Baseline MRI-based radiomics improving the recurrence risk stratification in rectal cancer patients with negative
Shuai Fu1, Ting Xia1, Zhenhui Li2
1Department of Radiology, The Sixth Affiliated Hospital, Sun Yat-Sen University, Guangzhou 510655, China; Biomedical Innovation Center, The Sixth Affiliated Hospital, Sun Yat-sen University, Guangzhou 510655, China.
Objectives:
Surveillance of rectal cancer recurrence in patients with negative pretreatment carcinoembryonic antigen (CEA) is challenging. This study aimed to develop the magnetic resonance imaging (MRI)-based radiomics models to predict rectal cancer recurrence in CEA-negative patients.
Materials And Methods:
This retrospective multicenter study consecutively enrolled rectal cancer patients with negative pretreatment CEA diagnosed between November 2012 and November 2018 from three medical centers. Radiomics features were extracted from the volume of interest on T2-weighted and diffusion-weighted images. Inter-class correlation coefficient (ICC) analysis, univariate Cox and least absolute shrinkage and selection operator (LASSO) Cox regression were then used to select features and construct a radiomics signature (RS). Multivariable Cox regression analysis was applied to develop two prediction models for disease-free survival (DFS) by incorporating the RS and independent clinicopathological predictors. The Kaplan-Meier method stratified tumor recurrence risk, and model performance was assessed using the Harrell concordance index (C-index).
Results:
A total of 600 rectal cancer patients were assigned to the development cohort (n = 358), internal test cohort (n = 120) and external test cohort (n = 122). The combination of ICC, univariate and LASSO Cox regression resulted in the selection of 25 radiomics features that comprise the RS. The RS can significantly stratify high-risk and low-risk patients (development: HR = 6.63, internal test: HR = 4.76, external test: HR = 4.62, all p < 0.05) and achieved good predictive performance (C-index = 0.720-0.758). The All-Clin+Rad model constructed by integrating RS and clinicopathological predictors showed superior performance with a C-index of 0.775 (95 % confidence interval [CI], 0.723-0.827), 0.739 (95 %CI, 0.654-0.823), and 0.822 (95 %CI, 0.720-0.924) in the development, internal test, and external test cohorts, respectively.
Conclusions:
The radiomics signature notably enhanced the prediction of tumor recurrence in rectal cancer patients with negative pretreatment CEA, assisting clinicians in making personalized follow-up plans.
