RESUMO
PURPOSE: To describe ultrasound (US) quality for hepatocellular carcinoma (HCC) screening/surveillance using the US LI-RADS scoring system, and to assess predictive factors of worse US quality scores. METHODS: This retrospective study included adult patients (n = 470; M/F 264/206, median age 59y) at risk for HCC that underwent US for HCC screening/surveillance. US examinations were independently reviewed by 2 radiologists that assigned a visualization score (A: no/minimal, B: moderate, C: severe limitation) and US diagnostic category (US LI-RADS 1: negative, US LI-RADS 2: subthreshold, US LI-RADS 3: positive) to each study. A generalized linear mixed model was used to assess the predictive factors of worse visualization score using OR (odds ratio) statistics. Simple Kappa coefficient (K) assessed inter-reader agreement. RESULTS: For readers 1 and 2, 295/320 (62.8%/68.1%) cases were scored A, 153/134 (32.6%/28.5%) were scored B, and 22/16 (4.6%/3.4%) were scored C, respectively. There was moderate inter-reader agreement for US LI-RADS visualization score (K = 0.478) and 100% concordance for US diagnostic category (K = 1), with 30 (6.4%) cases scored as positive (US LI-RADS 3). Cirrhosis and obesity were significant independent predictors of worse visualization scores (B/C) (cirrhosis: OR 10.4 confidence intervals: [4.25-25.48], p < 0.001; obesity: OR 3.61 [2.11-6.20], p < 0.001). Of the 30 lesions scored as US LI-RADS 3, 9 were characterized as probable or definite HCC on confirmatory CT/MRI, yielding a PPV of 30% (9/30) and a false-positive rate of 70% (21/30). CONCLUSION: Moderate to severe limitations in quality of US performed for HCC screening/surveillance was observed in approximately one-third of patients. Patients with cirrhosis and/or elevated BMI have poorer quality US studies and may benefit from other screening modalities such as CT or MRI.