Comparison of two transient elastography (TE) systems for measuring liver stiffness in clinical practice.

Inter-instrument variations of liver stiffness measure (LSM) by transient elastography might be problematic in clinical practice. LSMs provided by 2 different systems were compared in outpatients with chronic liver disease (CLD). In 777 consecutive asymptomatic outpatients admitted at the Hepatology Unit of Pisa University Hospital the agreement of LSMs measured by FibroScan (Echosens, France) and FT9000 (Hisky Medical, China) (LSM/LSM) was tested using Pearson correlation and Bland-Altman analyses (BAA). Delta FT9000-FibroScan LSM (LSM-D) variations were analysed according to clinic-pathologic characteristics. ALT/AST/GGT and platelets-counts/portal-vein-caliber/spleen-bipolar-diameter were used as proxies of necro-inflammatory and portal hypertension, respectively. LSM and LSM were highly correlated (r = 0.781, p < 0.001) overall, but agreement varied according to BMI (normal-weight r = 0.841, over-weight r = 0.747, obese r = 0.647, p < 0.001). LSM showed a + 0.52 kPa bias (p = 0.01) with - 11/12 kPa as 95% Limit of agreement, with higher values measured by LSM below 10 kPa and lower values above 20 kPa. In bivariate analysis LSM-D correlated with LSM and AST (β = 0.002, p = 0.001 with significant interaction, p = 0.018) and was associated with spleen-bipolar-diameter (β = 0.033, p = 0.027) and platelets-count (β=-0.008, p = 0.018) with significant interaction with LSM (p < 0.001). LSM and LSM are strongly correlated overall with optimal agreement around 10 kPa but limited at ends of the dynamic range: LSM provides higher values below 10 kPa whereas LSM higher values over 20 kPa with different measures distribution. LSM discrepancies associate with anthropometric characteristics, necro-inflammatory activity and portal hypertension proxies. These findings suggest the need of systems specific cut-offs in clinical practice.