Abbreviated-protocol screening MRI vs. complete-protocol diagnostic MRI for detection of hepatocellular carcinoma in patients with cirrhosis: An equivalence study using LI-RADS v2018

Background The high operational cost of MRI limits its utility for hepatocellular carcinoma (HCC) screening. Abbreviated-protocol dynamic contrast-enhanced MRI (aMRI) may help lower cost while maintaining the high accuracy of complete-protocol diagnostic MRI (cMRI). Purpose To compare aMRI to cMRI for HCC detection in cirrhosis patients. Study Type Cross-sectional study. Study Population Cirrhosis patients undergoing MRI for suspected HCC. Field Strength/Sequence 1.5T and 3T; aMRI (coronal T-2-weighted, axial dynamic contrast-enhanced T-1-weighted fat-suppressed sequences); cMRI (aMRI sequences and unenhanced axial T-2-, T-1-, and diffusion-weighted sequences). Assessment From each cMRI, an abbreviated exam was created by extracting only the aMRI sequences. Five radiologists independently reviewed aMRI and cMRI and assigned per-patient screening results by the presence/absence of any actionable observation per Liver Imaging and Reporting Data System v2018 (LI-RADS 4, 5, M, or TIV categories). Per-patient HCC status was determined by the composite reference standard of histopathology, follow-up imaging, consensus expert panel imaging review, and clinical follow-up. Statistical Tests Interreader agreement between aMRI and cMRI was compared with that of cMRI and tested for interchangeability against a tolerance margin of 0.05. Per-patient screening sensitivity, specificity, and accuracy were compared between aMRI and cMRI and tested for equivalence against a tolerance margin of 0.05. Results In 93 cirrhosis patients, five radiologists recorded on average 121 liver observations. Interreader screening agreement probability (and 95% confidence interval confidence interval [CI]) was 0.914 [0.900, 0.926] between aMRI and cMRI, and 0.927 [0.908, 0.942] for cMRI; their difference was within the 0.05 margin for interchangeability. In 86 patients in whom a final HCC status could be determined, the detection sensitivity and specificity of aMRI was 0.921 [0.864, 0.956] and 0.886 [0.844, 0.918], within the 5% equivalence margin to cMRI, 0.936 [0.881, 0.965] and 0.883 [0.840, 0.915], respectively. Data Conclusion Abbreviated-protocol screening MRI is interchangeable with, and equivalent to, complete-protocol diagnostic MRI for per-patient HCC detection in cirrhosis. Technical Efficacy: Stage 6 J. Magn. Reson. Imaging 2020;51:415-425.