Abstract
BACKGROUND: Qualification of non-invasive tests (NITs) and blood biomarkers for fibrosis severity and risk assessment is an unmet need in clinical care for cirrhosis. The paucity of reproducibility data for imaging and blood biomarkers in cirrhosis hinders their application for longitudinal monitoring. The aim of this study was to evaluate the reproducibility of imaging and blood biomarkers in a prospective cohort of adults with cirrhosis. METHODS: We prospectively enrolled adults with cirrhosis in an observational, longitudinal cohort study. Participants completed a research visit with vibration-controlled transient elastography (VCTE) with liver stiffness measurement (LSM) and controlled attenuation parameter (CAP) and blood testing, including AST, ALT, platelets, alpha-fetoprotein (AFP), and the Fibrosis-4 Index (FIB-4). Biomarkers were repeated a median of 3 months apart without any interim intervention. The primary outcome was the reproducibility of each biomarker as assessed by the reproducibility coefficient (RDC%), within-participant coefficient of variation (wCV), intraclass correlation coefficient (ICC), and Bland–Altman analysis. RESULTS: In total, 40 participants (mean age 62, 60% female, mean BMI 28.6 kg/m(2)) had VCTE/CAP, 38 AST/ALT, and 26 had AFP at both time points. The RDC% for VCTE-LSM was 54.4, wCV 0.19, and ICC 0.91. VCTE CAP had an RDC% of 56, a wCV of 0.30, and ICC 0.47. AFP had an RDC% of 58, a wCV of 0.21, and an ICC of 0.95. RDC% for AST, ALT, and FIB-4 ranged from 70% to 79%. On Bland–Altman analyses, VCTE-LSM demonstrated the lowest relative bias (0.4%). CONCLUSIONS: Prospective assessment of repeat biomarkers in adults with cirrhosis found that VCTE-LSM and CAP provide consistent assessments across individuals but may yield substantial within-individual variation. Blood-based biomarkers had poor reproducibility within individuals but were consistent in detecting between-patient variability. These data can inform ultrasound-based non-invasive testing qualification by defining thresholds that may reflect measurement variability rather than true biologic change over time.