Related Experiment Video
Updated: May 1, 2026
Detection of a Circulating MicroRNA Custom Panel in Patients with Metastatic Colorectal Cancer
Published on: March 14, 2019
Radiomics for predicting microsatellite instability-high status in colorectal cancer: a systematic review and
Qitao Gou1, Xin He2, Xiang Zhang3
1Department of Oncology, Laboratory of Immunity, Inflammation & Cancer (Chongqing Municipal Health commission), The First Affiliated Hospital of Chongqing Medical University, Chongqing, China. 357937698@qq.com.
Objective:
Microsatellite instability (MSI) has emerged as a key predictive biomarker for chemotherapy and immunotherapy response, and as a prognostic indicator in colorectal cancer (CRC). The current clinical standard for MSI detection relies on polymerase chain reaction (PCR) or immunohistochemical analysis of tumor biopsy specimens. CT, PET-CT, and MRI-based radiomics models present a promising non-invasive alternative for this purpose.
Materials And Methods:
To identify studies assessing the diagnostic efficacy of CT, MRI and PET-CT-based radiomics in detecting MSI status in CRC, a systematic search was performed across PubMed, Embase, the Cochrane Library, and Web of Science. The pooled area under the curve (AUC), sensitivity, and specificity were estimated using a random-effects model in Meta-DiSc 1.4, RevMan 5.4, and Stata 15. Data were visualized through forest plots and a summary receiver operating characteristic (SROC) curve. A comprehensive heterogeneity assessment was conducted via I² statistics, sensitivity analyses, threshold effect evaluation, subgroup analyses, and meta-regression.
Results:
This meta-analysis included 34 studies with a total of 7,959 patients. The overall model demonstrated a pooled area under the curve (AUC) of 0.90 (95% CI: 0.87-0.93). The pooled sensitivity and specificity were 0.85 (95% CI: 0.79-0.89) and 0.82 (95% CI: 0.78-0.86), respectively, both marked by substantial heterogeneity (I² = 87% and 92%, respectively; P < 0.01).
Conclusion:
Radiomics holds significant promise as a non-invasive tool for MSI status prediction in CRC. In particular, machine learning and deep learning offer enhanced potential for model performance. These results pave the way for future research to develop and validate more accurate predictive models, thereby improving diagnostic precision, therapeutic decision-making, and prognosis in colorectal cancer.
