MRI-based radiomics for noninvasive prediction of GPC3 expression in hepatocellular carcinoma: a systematic review and meta-analysis.

BACKGROUND: Hepatocellular carcinoma (HCC) is a leading cause of cancer death, and glypican-3 (GPC3) is a key diagnostic and therapeutic target currently assessed invasively. However, the overall diagnostic accuracy, between-study variability, and potential clinical relevance of Magnetic resonance imaging (MRI)-based radiomics models for predicting GPC3 expression have not been systematically quantified. AIM: Here, we systematically review and meta-analyze MRI-based radiomics models for noninvasive prediction of GPC3 expression in HCC. METHODS: Studies developing or validating MRI-based radiomics models against histopathologic GPC3 expression were identified through database searches, and data on diagnostic accuracy, radiomics workflows, and study design were extracted for a bivariate random-effects meta-analysis and quality assessment using the adapted Quality Assessment of Diagnostic Accuracy Studies-2 (QUADAS-2) tool and the METhodological RadiomICs Score (METRICS) tool. Results: Of 63 records, seven retrospective studies comprising 1,209 patients met the inclusion criteria for the systematic review, of which five studies contributing eight model estimates were included in the quantitative synthesis. The pooled sensitivity was 0.81 (95% CI: 0.68-0.89), the pooled specificity was 0.87 (95% CI: 0.79-0.92), and the summary receiver operating characteristic curve yielded an area under the curve (AUC) of 0.91 (95% CI: 0.88-0.93). Sensitivity showed substantial heterogeneity (I² = 87.4%), whereas no heterogeneity was observed for specificity (I² = 0%). METRICS scores ranged from 56.1% to 92.2%. CONCLUSIONS: This systematic review and meta-analysis demonstrate that MRI-based radiomics provides a previously inaccessible, noninvasive window on GPC3 expression in HCC. The synthesis establishes a benchmark evidence base that opens the door to standardized radiomics pipelines, radiogenomic-guided trials, and imaging biomarkers to select patients for GPC3-targeted therapies.