ARRS 2022 Abstracts


E1303. Inter-Exam Variability of Ultrasound LI-RADS Visualization Score Over Time: A Three-Center Study
  1. Thodsawit Tiyarattanachai; Department of Radiology, Stanford University School of Medicine; Faculty of Medicine, Chulalongkorn University
  2. David Fetzer; Department of Radiology, University of Texas Southwestern Medical Center
  3. Aya Kamaya; Department of Radiology, Stanford University School of Medicine
In 2017, the Ultrasound Liver Imaging Reporting and Data System (LI-RADS) was launched to standardize management and reporting of ultrasound (US) exams in patients at risk for hepatocellular carcinoma (HCC). Visualization Score (VIS) was established to help standardize a radiologist’s assessment of relative exam quality, and potentially convey the expected level of sensitivity of each US exam. It is unclear, however, whether a single suboptimal VIS warrants complete change in surveillance strategy given the variability in patient conditions and extrinsic factors that may affect VIS. Thus, we sought to determine the relative stability of VIS on subsequent exams in patients referred for surveillance US.

Materials and Methods:
A consecutive 15,727 US HCC surveillance exams in 8,632 at-risk patients were retrieved from 3 centers in the United States: Stanford University Medical Center, University of Texas Southwestern Medical Center, and Parkland Health and Hospital System. VIS were reported in 14,674 exams from 8,241 patients. 3,169 patients had at least 2 exams yielding a total of 9,602 US exams comprise our final analysis group. Of these, 8,030 (83.6%) exams were VIS A (no or minimal limitations), 1,378 (14.4%) were VIS B (moderate limitations), and 194 (2.0%) were VIS C (severe limitations). We examined the probabilities that VIS would remain stable or change, given VIS in the preceding exam(s).

The median time intervals after exams with VIS A, B and C were 192 (IQR 75), 199 (IQR 118) and 194 (IQR 101) days, respectively (p = 0.11). Mixed-effects ordinal logistic model showed that suboptimal VIS were significantly associated with older age (OR 1.02, 95% CI 1.01 – 1.03) and male sex (OR 1.37, 95% CI 1.11 – 1.69). VIS were also significantly different among patients from different centers (p < 0.001). VIS A tended to remain stable with probability of 0.88 (CI 0.87-0.89), 0.90 (CI 0.89-0.92) and 0.93 (CI 0.92-0.95) after 1, 2 and 3 consecutive VIS A’s. For VIS B, the probability of remaining stable was more variable, increasing from 0.45 (CI 0.41-0.49) after 1 exam, to 0.50 (CI 0.41-0.59) after 2 exams, and 0.52 (CI 0.31-0.74) after 3 consecutive VIS B’s. The probabilities of transition from VIS B to A were 0.50 (CI 0.46-0.54) and 0.43 (CI 0.34-0.52); B to C were 0.05 (CI 0.03-0.06) and 0.07 (CI 0.02-0.11) after 1 and 2 consecutive VIS B’s, respectively. For VIS C, the probability of remaining VIS C after 1 exam was 0.41 (CI 0.32-0.51), increasing to 0.56 (CI 0.33-0.79) after 2 exams, and 0.80 (CI 0.45-1.00) after 3 consecutive VIS C’s. The probabilities of transition from VIS C to A were 0.21 (CI 0.13-0.30) and 0.22 (CI 0.03-0.41); C to B were 0.37 (CI 0.27-0.47) and 0.22 (CI 0.03-0.41) after 1 and 2 consecutive VIS C’s.

A single suboptimal VIS should not be a reason to move from US to alternative imaging modalities as over half of subsequent exams are scored VIS A or B. However, for patient in whom there are multiple exams with suboptimal VIS, the clinical care team may consider alternative imaging modalities to increase sensitivity for focal liver observations.