Diagnostic Performance of Magnetic Resonance Elastography Versus Vibration‐Controlled Transient Elastography for Liver Fibrosis Staging in Steatotic Liver Disease

ABSTRACT Gravitational magnetic resonance elastography (GT‐MRE) enables comprehensive biomechanical liver characterization and may overcome limitations of conventional elastography techniques. This study evaluated the diagnostic performance of two‐ and three‐dimensional (2D and 3D) GT‐MRE compared with 2D acoustic MRE (AC‐MRE) and vibration‐controlled transient elastography (VCTE) for liver fibrosis staging. In this prospective study, patients with suspected or known steatotic liver disease underwent GT‐MRE (2D and 3D, gradient‐echo sequences) and AC‐MRE (2D, spin‐echo echo‐planar imaging) at 1.5 T, plus VCTE with a median interval to MRE of 37 days (IQR, 20–124 days; range, 0–372 days). Statistical analyses included generalized additive models, ordinal logistic regression, and decision tree‐based classification. Diagnostic performance was assessed using cumulative receiver operating characteristic analysis and corresponding areas under the curve (AUCs). Ninety‐one participants (median age, 55 years [IQR, 45–64]; 41 women) were included, of whom 35 (38%) had histopathological correlation. In the biopsy‐confirmed subgroup, 3D GT‐MRE‐derived viscoelastic parameters showed promising exploratory performance for early fibrosis classification. Wave attenuation achieved AUCs of 0.91 for F0 versus ≥ F1 and 0.99 for ≤ F1 versus ≥ F2, while phase angle yielded an AUC of 0.88 for F0 versus ≥ F1, with numerically higher AUCs than conventional stiffness‐based metrics. For advanced fibrosis thresholds, all modalities performed well (AUC ≥ 0.84). Obuchowski measures for 3D‐derived parameters were 0.95–0.96. Internally cross‐validated multiparametric modeling achieved an accuracy of up to 92%. VCTE measurements plateaued beyond F3, whereas AC‐ and GT‐MRE metrics continued to increase with fibrosis severity. 3D GT‐MRE‐derived viscoelastic parameters showed promising performance for liver fibrosis classification and provided biomechanical information complementary to conventional stiffness measurements. Given the limited biopsy‐confirmed sample and marked class imbalance at F0, these findings require validation in larger independent cohorts before clinical implementation.