간이식을 시행한 간세포암 환자에서 LI-RADS 기반 영상의학적 평가와 병리학적 평가에 따른 Milan 기준의 일치도 및 예후적 연관성에 관한 연구

Abstract

Background/Aims: Given the limited availability and feasibility for pathologic confirmation of every hepatic lesion, imaging diagnosis is primarily used to assess Milan criteria in candidates for liver transplantation (LT) with hepatocellular carcinoma (HCC). There is little data on the correlation between explant pathology and radiologic measurement based on LI-RADS in determining LT eligibility in HCC patients. This study aimed to investigate the radio-pathologic correlation of Milan criteria using LI-RADS-based diagnosis, and also to identify factors affecting discordance and its prognostic impact. Methods: This retrospective study included 267 patients who had any hepatic lesion identified on dynamic liver CT within 3 months prior to LT and/or in the explant livers at Asan Medical Center. Two radiologists reviewed CT examinations, evaluating nodules and the Milan criteria based on LI-RADS v2018. Analyses were performed on a per-lesion and per-patient basis, comparing radiologic lesions with their matched pathology. LR-5 or LR-TR-V nodules were regarded as HCC to determine LI-RADS Milan criteria (MC). Overall survival (OS) and recurrence-free survival (RFS) were measured according to LI-RADS MC and pathologic MC, applying a competing risk analysis to 259 patients, excluding cases of in-hospital mortality. Results: In per-lesion analysis, among 79 LR-5 lesions and 48 LR-3/LR-4 lesions, 72 lesions (91.1%) and 37 (77.1%) were identified as HCCs, respectively. The 189 LR-TR-V lesions were matched with 176 HCCs (93.1%) in pathology. According to per-patient analysis, an overall concordance rate of 87.3% was presented between LI-RADS MC and pathologic MC. These concordances were not affected by pre-LT chemoembolization and type of LT. The 5-year OS and RFS were significantly greater for patients meeting the MC, compared to the counterparts: 96% vs. 86% and 90% vs. 62% for pathologic MC; and 96% vs. 76% and 90% vs. 45% for LI-RADS MC, respectively (Ps<0.003). When cases meeting LI-RADS MC and pathologic MC were compared, there were no significant differences in OS and RFS. Multivariate Cox-proportional hazard models indicated that being outside LI-RADS MC independently predicted OS and RFS (hazard ratios, 6.51 and 6.34; 95% confidence intervals: 2.37- 17.86 and 3.38-11.88, respectively). The presence of nodules other than LR-5/LR-TR-V did not affect survivals. Conclusions: The LI-RADS-based radiology presented high concordance and comparable prognostic performance with explant pathology in determining the MC. LT eligibility could likely be judged by CT LI-RADS in patients with HCC.

Keywords: LI-RADS v2018, LI-RADS treatment response algorithm, liver transplantation, hepatocellular carcinoma