Inter-observer agreement using the LI-RADS version 2018 CT treatment response algorithm in patients with hepatocellular carcinoma treated with conventional transarterial chemoembolization

Aim To determine inter-reader agreement in categorization of imaging features using the Liver Imaging Reporting and Data System (LI-RADS) treatment response (LR-TR) algorithm in patients with hepatocellular carcinoma (HCC) treated with conventional transarterial chemoembolization (cTACE). Methods Two radiologists used the LR-TR algorithm to assess 112 computed tomography (CT) examinations of 102 patients treated with cTACE. The inter-observer agreement in categorization of LR-TR features was assessed using kappa (kappa) statistics. Results There was substantial inter-observer agreement between the two reviewers using the LR-TR algorithm (kappa = 0.70; 95% CI 0.58-0.81). The two reviewers categorized tumors as non-viable in 37 (33.0%) and 39 (34.8%) of 112 examinations, viable in 58 (51.8%) and 62 (55.4%) examinations, and equivocal in 18 (16.1%) and 11 (9.8%) examinations, respectively. There was almost perfect inter-observer agreement for the LR-TR non-viable category (kappa = 0.80; 95% CI 0.68-0.92), substantial agreement for the viable category (kappa = 0.78 95% CI 0.67-0.90), and fair agreement for the equivocal category (kappa = 0.25; 95% CI 0.02-0.49). Conclusion The LR-TR algorithm conveys high degrees of inter-observer agreement for the assessment of CT imaging features in the viable and non-viable categories. Further refinement of indeterminate features may be necessary to improve the correct categorization of equivocal lesions.

Automated summary

This study aimed to determine the inter-reader agreement in categorization of imaging features using the LI-RADS LR-TR algorithm in HCC patients treated with cTACE. The results indicated substantial inter-observer agreement in categorizing tumors as viable, non-viable, and equivocal, with further refinement needed for equivocal lesions.