Classification according to sequentially combined non-invasive tests carries prognostic value in patients with nafld

Abstract

BACKGROUND: Combination of FIB4 and liver stiffness measured by vibration controlled transient elastography (LSM-VCTE) is supported by cross sectional studies of diagnostic accuracy. Here we aim to study the prognostic value of FIB-4 followed by LSM-VCTE. METHODS: This was an individual participant data Meta Analysis of people with NAFLD and baseline biopsy, LSM-VCTE, and FIB-4 measured within 6 months. The outcome was a composite endpoint of all-cause mortality, cirrhosis decompensation (ascites, variceal haemorrhage, hepatic encephalopathy), hepatocellular carcinoma, liver transplantation or progression to model of end stage liver disease score >= 15. Study-specific cumulative hazard functions and derived aggregated survival curves were built for: G1 (FIB-4 < 1.3), G2 (1.3 <= FIB-4 < 2.67; LSM-VCTE < 10 kPa), G3 (1.3 <= FIB-4 < 2.67; 10kPa <= LSM-VCTE < 20 kPa), G4 (1.3 <= FIB-4 < 2.67; LSM-VCTE >= 20kPa) and G5 (FIB-4 >= 2.67). G3 was further stratified according to fibrosis (F0-2, F3, F4), and LSM-VCTE one (15 kPa) or two cut-offs (13.5 kPa, 17 kPa). G5 was similarly subclassified according to fibrosis and LSM-VCTE. Unreliable LSM-VCTE (LSM >= 7.1 kPa and IQR/ median > 0.3) readings were excluded. RESULTS: Data from 2383 (51% males, median age 54 (IQR 19) years, 42% with type II diabetes mellitus) were available, after exclusion of unreliable LSM-VCTE. Median follow-up was 56 months and the primary outcome was reached in 5.3% (n = 127). There were significant differences in the prognostic information carried by all groups (study-stratified log-rank tests; p < 0.0001; Fig. 1a). Within G3, patients with cirrhosis (G3-F4) showed the poorest prognosis, and similar survival probabilities were seen between patients in G3-F0-2 and G3-F3 (p = 0.02; Fig. 1b). LSM-VCTE did not further separate G3 (Fig. 1c-d). Within G5, LSM-VCTE (p < 0.001, Fig. 1e), and histology (p = 0.004, Fig. 1f) provided further prognostic granularity. CONCLUSIONS: Sequential application of FIB-4 and LSM-VCTE classifies patients into 5 groups with differing risk of future adverse events, further supporting the continued use of this approach in clinical practice. In those with FIB4 >= 2.67, LSM-VCTE and biopsy can provide further prognostic granularity. Biopsy can also separate out risk subgroups in those with 1.3 <= FIB4 < 2.67. Our analysis is limited by relatively small numbers in groups G3 and G5. Repeated NIT measurements over time or other approaches may provide additional prognostic insight but these were beyond the remit of our study.