Computed tomography radiomics to predict microsatellite instability status and immunotherapy response in gastric cancer
Abstract Objectives To develop and validate a CT radiomics model for predicting microsatellite instability (MSI) status in preoperative gastric cancer (GC) patients and to explore the underlying immune infiltration pattern of the radiomics model. Materials and methods This study used three retrospec...
Saved in:
| Main Authors: | , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
SpringerOpen
2025-08-01
|
| Series: | Insights into Imaging |
| Subjects: | |
| Online Access: | https://doi.org/10.1186/s13244-025-02050-1 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Abstract Objectives To develop and validate a CT radiomics model for predicting microsatellite instability (MSI) status in preoperative gastric cancer (GC) patients and to explore the underlying immune infiltration pattern of the radiomics model. Materials and methods This study used three retrospective datasets from Tongji Hospital (n = 304, training set), Xiangyang Central Hospital (n = 48, external testing set 1) and public datasets from The Cancer Imaging Archive (TCIA) (n = 43, external testing set 2). The preoperative contrast-enhanced CT images of GC were evaluated. Radiomics features were extracted and selected to construct the radiomics model in the training set, and further validated in the other two external testing sets. The outcome cohort, including 68 advanced unresectable GC patients receiving immunotherapy, was used to assess the predictive value of the radiomics model for treatment response and outcomes. We analyzed RNA-sequencing data from TCIA to investigate the underlying genomics characterization and immune infiltration spectrum of the radiomics model. Results Four radiomic features were ultimately selected to develop the radiomics model. The model demonstrated good predictive performance for MSI status, achieving AUCs of 0.952, 0.835, and 0.879 in the training set and the two external testing sets, respectively. Radiomics scores (Radscores) was an independent predictor for PFS in the outcome cohort (HR: 0.145; 95% CI: 0.032–0.657; p = 0.012). Radscores were positively correlated with CD8+ T cells (R = 0.74, p = 0.013) and negatively related to M2-type macrophages (R = −0.67, p = 0.028). Conclusion Our CT radiomics model could effectively predict MSI status and immunotherapy outcomes in GC patients therefore, may act as a potential noninvasive tool for personalized treatment decisions. Critical relevance statement Our study develops a noninvasive biomarker based on readily available imaging to identify gastric cancer patients who may benefit from immunotherapy. It also reveals biological meanings of the radiomics biomarker, promoting further research into interpretability and clinical application of radiomics. Key Points A CT-based radiomics model was constructed to noninvasively predict gastric cancer (GC) microsatellite instability status. This immune-related radiomics model can effectively predict immunotherapy outcomes in GC. This noninvasive method can serve as a supplement for treatment decisions. Graphical Abstract |
|---|---|
| ISSN: | 1869-4101 |