Noninvasive assessment of liver disease severity in patients with nonalcoholic fatty liver disease (NAFLD) and type 2 diabetes

Abstract

Background: We evaluated the diagnostic accuracy of simple non-invasive tests(NITs) in NAFLD patients with type 2 diabetes(T2D). Methods: This was an individual patient data meta-analysis of 1780 patients with biopsy-proven NAFLD and T2D. The index tests of interest were FIB-4, NAFLD Fibrosis Score(NFS), APRI, liver stiffness measurement(LSM) by vibration-controlled transient elastography(VCTE) and AGILE 3+. The target conditions were advanced fibrosis, nonalcoholic steatohepatitis(NASH) and fibrotic NASH(NASH plus F2-F4 fibrosis). The diagnostic performance of NITs individually or in sequential combination was assessed by area under receiver operating characteristic curve(AUROC) and by decision curve analysis(DCA). Comparison with 2278 NAFLD patients without T2D was also made. Results: In NAFLD with T2D LSM and AGILE 3+ outperformed both NFS and FIB-4 for advanced fibrosis(AUROC:LSM 0.82,AGILE 3+ 0.82,NFS 0.72,FIB-4 0.75,APRI 0.68;p < 0.001 of LSM-based versus simple serum tests), with an uncertainty area of 12%-20%.The combination of serum-based with LSM-based tests for advanced fibrosis led to a reduction of 40% to 60% in necessary LSM tests. DCA showed that all scores had modest net benefit for ruling-out advanced fibrosis at the risk threshold of 5%-10% of missing advanced fibrosis. LSM and AGILE 3+ outperformed both NFS and FIB-4 for fibrotic NASH(AUROC LSM 0.79,AGILE 3+ 0.77,NFS 0.71,FIB-4 0.71;p < 0.001 of LSM-based versus simple serum tests). All noninvasive scores were sub-optimal for diagnosing NASH. Conclusions: LSM and AGILE 3+ individually or in low availability setting in sequential combination after FIB-4 or NFS have a similar good diagnostic accuracy for advanced fibrosis and an acceptable diagnostic accuracy for fibrotic NASH in NAFLD patients with T2D.