The aim of this study was to validate the Liver Imaging Reporting and Data System (LI-RADS) v2014 category 4 (LR-4) and 5 (LR-5) criteria on gadoxetate disodium–enhanced magnetic resonance imaging (MRI) in patients with chronic liver disease.
Materials and Methods
Between January and December 2012, 300 patients with chronic liver disease who had hepatic nodules 3.0 cm or smaller at surveillance ultrasonography and gadoxetate disodium–enhanced MRI were included. LI-RADS category was retrospectively assigned to each nodule on MRI. Final diagnosis was assessed using pathologic diagnosis only (operation or core-needle biopsy) or pathologic and clinical diagnosis (marginal recurrence after treatment or a change in lesion size on follow-up imaging). To validate the LR-4 and LR-5 criteria, the sensitivity, positive predictive value (PPV), and false referral rate for diagnosing hepatocellular carcinoma were examined.
Based on major imaging features only, 137 nodules were initially assigned as LR-3, but 133 (97.1%) were upgraded into LR-4 by the presence of ancillary features. Excluding the remaining 4 LR-3 and 3 LR-M nodules, we analyzed 379 nodules in 294 patients, consisting of 211 LR-4 and 168 LR-5 nodules. Using pathologic diagnosis only, the sensitivity and PPV with 95% confidence intervals (CIs) for LR-5 were higher than those for LR-4 (57.3% [50.6–63.7] vs 42.7% [36.3–49.4]; 94.6% [89.0–97.5] vs 82.0% [73.7–88.1]), with a lower false referral rate (5.4% [2.5–11.0] vs 18.0% [11.9–26.3]). Using pathologic and clinical diagnosis, PPV and 95% CI for LR-5 were higher than that for LR-4 (95.2% [90.7–97.7] vs 79.1% [73.1–84.1]), whereas sensitivity and 95% CI for LR-5 was similar to that for LR-4 (48.9% [43.6–54.3] vs 51.1% [45.7–56.4]).
In patients with chronic liver disease, LR-5 criteria on gadoxetate disodium–enhanced MRI had excellent PPV for diagnosing hepatocellular carcinoma, whereas LR-4 criteria showed good PPV, but are only of limited use.