Reliability of Hepatic Steatosis Quantification by Controlled Attenuation Parameter in Steatotic Liver Disease

Assessment of hepatic steatosis is key for diagnosis and monitoring of steatotic liver disease. Controlled attenuation parameter is widely used, but concerns regarding its measurement variability exist. We evaluated the repeatability and reproducibility of controlled attenuation parameter and FibroScan-aspartate aminotransferase score across different clinical steatotic liver disease settings and disease stages.

We assessed same-day repeatability in subjects with suspected steatotic liver disease in secondary care (cohort I, n = 40) and tertiary care (cohort II, n = 58). Different-day reproducibility within 28 days was assessed in 4 cohorts: suspected steatotic liver disease (cohort III, n = 37), suspected fibrosis (cohort IV, n = 163), compensated advanced steatotic liver disease (cohort V, n = 32), and clinically significant portal hypertension (cohort VI, n = 28). Agreement was evaluated using the intraclass correlation coefficient, Bland-Altman analysis, and minimal detectable change. Categorical agreement was assessed by observed agreement and Cohen's Kappa.

Same-day controlled attenuation parameter agreement was moderate to good (intraclass correlation coefficient, 0.64-0.80), but variability was substantial, with a minimal detectable change of 57 to 69 dB/m, corresponding to a relative minimal detectable change of 18% to 25%. Agreement for dichotomized controlled attenuation parameter cutoffs was moderate to strong (Cohen's Kappa = 0.33-0.65). Across different days, controlled attenuation parameter reproducibility was lower and worsened with disease severity. Intraclass correlation coefficients ranged from 0.44 to 0.67 in cohorts with suspected steatotic liver disease/fibrosis, decreased to 0.52 in compensated advanced steatotic liver disease, and was poor in clinically significant portal hypertension. Correspondingly, minimal detectable change increased from 80 to 94 dB/m (26%-34%) in less advanced disease to 128 to 174 dB/m (49%-72%) in advanced disease. FibroScan-aspartate aminotransferase showed higher reproducibility in cohorts III to V (intraclass correlation coefficient, 0.73-0.82; minimal detectable change, 0.19-0.33) but also declined in cohort VI (intraclass correlation coefficient, 0.54).

The minimal detectable change of controlled attenuation parameter is >57 to 69 dB/m (>18%-25%) and substantially greater in a longitudinal setting and advanced disease.