LI-RADS to categorize liver nodules in patients at risk of HCC: tool or a gadget in daily practice?

Purpose: To determine the effectiveness of liver reporting and data system (LI-RADS) to diagnose hepatocellular carcinoma (HCC) and to retrospectively evaluate its impact on the adopted therapeutic strategy.

Materials And Methods: Preoperative imaging of 40 of 350 patients (median age 66, 31 M/9 F) submitted to liver resection for suspected HCC, between January 2008 and August 2019, has been retrospectively analyzed by two radiologists with different expertise, according to CT/MRI LI-RADS v2018, both blinded to clinical and pathological results and untrained to using aforementioned scoring system.

Results: The perfect agreement between the readers was about 62.5% (25/40) (Cohen k: 0.41), better for LR-5 category (16/25) and higher in magnetic resonance imaging (MRI) investigations (68%; 13/19), which has been demonstrated the modality of choice for diagnosis of high probable and certain HCC, with arterial phase hyperenhancement as the most sensitive and accurate major feature. Compared to final histology, LR4 and LR5 scores assigned by senior radiologist reached sensitivity, specificity, positive and negative predictive values (PPV, PNV) and diagnostic accuracy of 90,9%, 29,0%, 93,8%, 62,5% and 87,5%, respectively, slightly higher than junior's ones. Misdiagnosis of HCC was done by both radiologists in the same two patients: 1 primary hepatic lymphoma (PHL) and 1 regenerative liver nodule (RLN). If LI-RADS would have been applied at the time of pre-surgical imaging, treatment planning would be modified in 10% of patients (4/40); the patient scheduled as LR-3 and finally resulted a focal nodular hyperplasia would have avoided liver resection.

Conclusions: Application of LI-RADS, especially on MRI, may provide a more accurate evaluation of suspected HCC. PHL and RLN are the Achille's heels according to our experience.