Reliability criteria for controlled attenuation parameter (CAP) for assessing hepatic steatosis using transient elastography

Background:

Due to partially conflicting results of previous studies, there is an ongoing debate on reliability criteria for controlled attenuation parameter (CAP; an imaging biomarker for of hepatic steatosis). Thus, we assessed the diagnostic performance of CAP according to different reliability criteria.

Methods:

Patients undergoing measurement of CAP and liver biopsy (± 180 days) at the Medical University of Vienna were included. Steatosis (S0-S3) was assessed according to Brunt/Kleiner Score.

Results:

Among 408 patients (age: 49 years, 53% male) with predominantly non-alcoholic fatty liver disease (NAFLD, n = 180, 44.1%) and viral hepatitis (n = 62, 15.2%), the distribution of histological steatosis and fibrosis stages were: 148 (36.3%) S0, 118 (28.9%) S1, and 78 (19.1%) S2 and 64 (15.7%) S3; 149 (46.3%) F0/F1, 59 (14.5%) F2, and 153 (37.5%) F3/F4.

In the overall cohort, the area under the receiver operating characteristic curve (AUROC) of CAP was 0.825 (95% confidence interval [CI]: 0.783 – 0.866) for diagnosing any steatosis (≥S1), 0.797 (95%CI: 0.753 – 0.841) for ≥S2 and 0.780 (95%CI: 0.728 – 0.832) for ≥S3. Applying CAP corrections as suggested by Karlas et al (J Hepatol 2017;65(6): 1022 – 1030) did not improve diagnostic performance (AUROC). Diagnostic performance for any steatosis decreased in patients with F3/F4 (AUROC: 0.788 vs. 0.851 in F0/F1/F2) and in patients with liver stiffness ≥20kPa (0.757 vs. 0.845 in < 20kPa). Importantly, AUROC of CAP for ≥S1 was substantially better in patients with more homogenous individual CAP values: 0.856 (0.807 – 0.905) in CAP-IQR/median< 20% (n = 278) vs. 0.665 (0.558 – 0.772) in CAP-IQR/median≥20% (n = 130).

Noteworthy, 104 patients (25.5%) had significant disagreement (≥2 stages of steatosis) between CAP and histology. In multivariate analysis, disagreement was independently associated with higher CAP (P< 0.001) and CAP-IQR/median (P= 0.040) but neither with histological fibrosis nor with markers for necroinflammatory activity (AST/ALT) or cholestasis parameters (ALP/GGT).

Next to histological steatosis (unstandardized B = 19.949, P< 0.001), CAP values were independently linked to BMI (B = 3.047, P< 0.001), NAFLD etiology (B = 36.210, P< 0.001) and diabetes (B = 16.887, P= 0.0116).

Conclusion:

About 25% of patients showed a considerable disagreement between steatosis staging by CAP and histology. However, a CAP-IQR/median< 20% can be used to identify more reliable CAP measurements.