MRI Radiomics Predicts Early Biological Progression in a Preclinical Colon Cancer Model Following Mesenchymal Stem
Jun Sun1,2, Qingning Yang1, Wei Ma2,3
1Department of Medical Imaging, Chongqing Emergency Medical Center, Chongqing Key Laboratory of Emergency Medicine, Chongqing University Central Hospital, School of Medicine, Chongqing University, Chongqing, 400010, People's Republic of China.
Background:
Stem cell therapy is a potential approach for tumor treatment; however, its efficacy and safety remain incompletely controllable. Therefore, early prediction of tumor progression and therapeutic evaluation post-treatment to guide timely therapeutic adjustments is essential. Radiomics has recently been widely applied to assess tumor behavior and characteristics. This study aims to investigate the predictive value of MRI radiomics for therapeutic outcomes in the early stages of stem cell intervention in colon cancer.
Methods:
Subcutaneous tumor models were established by implanting CM38 colon cancer cells into the inguinal region of C57BL/6 mice. Bone marrow mesenchymal stem cells (MSCs) were intravenously injected via the tail vein within 24 hours. T1 weighted MRI scans were performed on day 7 to extract radiomic features, monitor tumor growth, and evaluate neovascularization (CD34) and proliferation (Ki67) via immunohistochemistry. An early-stage MRI radiomics model was constructed to predict colon cancer progression.
Results:
Tumors in the stem cell intervention group exhibited faster growth compared to the control group, though no significant volume difference was observed in the early phase. A total of 1162 radiomic features were extracted, with 17 strongly associated features (including texture features, first-order features texture and shape features) selected to build the prediction model. The final model demonstrated an AUC > 0.8 across both training and testing datasets.
Conclusion:
Intravenous MSC injection promotes the progression of subcutaneously implanted CM38 colon cancer. An MRI radiomics-based predictive model was successfully established, enabling early prediction of tumor progression post-intervention.
