Society: AASLD

Introduction
The rising prevalence of non-alcoholic fatty liver disease (NAFLD) will result in increased referrals to gastroenterology and hepatology units. Non-invasive tests (NITs) predict advanced liver fibrosis in NAFLD, and NIT thresholds help triage referrals for patients who require assessment in secondary care. However, these cut-offs may not appropriately predict long term risk of liver-related morbidity, and many patients continue having secondary care follow-up with low-risk disease. This study prospectively assessed the utility of NITs in predicting hepatic morbidity and mortality.

Methods
The NAFLD-fibrosis score (NFS) and Fibrosis-4 Index (FIB-4) were calculated at baseline in patients with NAFLD recruited between 2002 to 2021. The primary study endpoints were incident liver-related events (composite of hepatic decompensation, hepatocellular carcinoma, liver transplantation, and liver-related death) and all-cause mortality. Univariable and multivariable Cox’s regression models assessed associations of baseline factors with study endpoints. Multivariable models tested NIT thresholds (Low risk, NFS -1.455, FIB-4 1.3; Medium risk, NFS 0.20, FIB-4 2.0; High risk, NFS 0.676, FIB-4 2.67) in predicting liver-related events (LRE) and mortality.

Results
In total, 487 patients were included (187 females; median age 49 years (IQR 38-58); median body mass index 32.3 kg/m²(IQR 29.0-37.0); median follow up 44 months (IQR 17-88)). LRE was observed in 18 patients at a median follow-up of 43 months (IQR 16-87). Baseline NFS (adjusted Hazard ratio (aHR)=2.17, 95% CI 1.51-3.13) and FIB-4 (aHR=1.88, 95% CI 1.50-2.35) scores were independently associated with LRE. Death was observed in 27 patients at a median follow-up of 69 months (IQR 36-112). Baseline NFS (aHR=2.08, 95% CI 1.50-2.88) and FIB-4 (aHR=1.40, 95% CI 1.14-1.72) scores were independently associated with all-cause mortality. Low risk NIT thresholds (NFS -1.455, FIB-4 1.3) that are commonly used to triage patients from primary to secondary care did not predict either study endpoint (p>0.05) (Table 1). Medium NIT thresholds predicted LRE (aHR_NFS=10.70, 95% CI 2.68-42.74, Negative predictive value (NPV)_NFS=0.99; aHR_FIB-4=10.05, 95% CI 2.74-36.87, NPV_FIB-4=0.99) and all-cause mortality (aHR_NFS=6.71, 95% CI 2.28-19.71, NPV_NFS=0.98; aHR_FIB-4=5.51, 95% CI 1.91-15.88, NPV_FIB-4=0.98) (Table 1, Figure 1). Liver-related events were observed in 1 patient (<2%) with indeterminate NITs between low and medium risk thresholds (NFS -1.455 – 0.20; FIB-4 1.3 – 2.0).

Conclusions
NITs accurately predict liver-related events and all-cause mortality in patients with NAFLD. Higher NIT thresholds better predict liver-specific prognosis than those used in current referral pathways. Higher NIT thresholds predict liver-related events for up to 3 years and can guide resource utilization once patients enter secondary care.

<b>Table 1.</b> Associations of non-invasive tests of liver fibrosis (FIB-4, fibrosis-4 index; NFS, non-alcoholic fatty liver disease fibrosis score) at different thresholds in predicting the risk of liver-related events and death in patients with non-alcoholic fatty liver disease using multivariable Cox regression models.

Table 1. Associations of non-invasive tests of liver fibrosis (FIB-4, fibrosis-4 index; NFS, non-alcoholic fatty liver disease fibrosis score) at different thresholds in predicting the risk of liver-related events and death in patients with non-alcoholic fatty liver disease using multivariable Cox regression models.

<b>Figure 1.</b> Kaplan Meier curves showing survival free of liver-related events and death in patients with non-alcoholic fatty liver disease (NAFLD) when stratified by baseline non-invasive tests of liver fibrosis including the NAFLD fibrosis score (NFS, A and C, respectively) and fibrosis-4 index (FIB-4, B and D, respectively).

Figure 1. Kaplan Meier curves showing survival free of liver-related events and death in patients with non-alcoholic fatty liver disease (NAFLD) when stratified by baseline non-invasive tests of liver fibrosis including the NAFLD fibrosis score (NFS, A and C, respectively) and fibrosis-4 index (FIB-4, B and D, respectively).

Background and Aims: A multidisciplinary task force commissioned by the American Gastroenterological Association (AGA) recently developed a clinical care pathway with guidance on screening high-risk populations including those with type 2 diabetes (T2DM). We aimed to validate the pathway in a prospectively recruited cohort of patients with T2DM and evaluate the diagnostic performance of new American Association for the Study of Liver Disease (AASLD) practice guidelines with enhanced liver fibrosis (ELF) testing.

Method: This prospective study enrolled adults age ≥ 50 years with T2DM recruited from primary care or endocrinology clinics. Participants underwent a standardized clinical research visit with magnetic resonance imaging proton-density-fat-fraction (MRI-PDFF), magnetic resonance elastography (MRE), vibration-controlled transient elastography (VCTE) and controlled attenuation parameter (CAP) and ELF testing. The primary outcome was the diagnostic performance of the AGA clinical pathway and AASLD pathway with ELF for advanced fibrosis using MRE ≥ 3.63 kPa as the reference.

Results: 417 patients (36% men) with T2DM with FIB-4 and MRE data were included. The mean (±SD) age and BMI were 65 (±8) years and 30 (±5) kg/m2, respectively. The prevalence of NAFLD (MRI-PDFF ≥ 5% after exclusion of other liver diseases) was 64% and 12% had advanced fibrosis (MRE ≥ 3.63 kPa). FIB-4 values of < 1.3, 1.3-2.67 and ≥ 2.67 were present in 208 (50%), 183 (44%), and 26 (6%) of patients respectively. VCTE values < 8 kPa, 8-12 kPa and ≥ 12 kPa were present in 309 (77%), 58 (14%), and 36 (9%) of patients respectively. ELF scores of < 9.8, 9.8-11.3 and > 11.3 were present in 236 (58%), 146 (36%) and 22 (6%) patients respectively. Applying the AGA pathway (N=403), 202 patients were low risk by FIB-4 and 128 additional patients had VCTE < 8 kPa and were classified as low risk. The false negative rate was 3.3% and 18% would qualify for specialty referral (Figure 1A). Applying the expanded pathway with ELF (N=404), 200 patients were low risk by FIB-4 and 85 additional patients had low ELF and were low risk. The false negative rate was 4.8% and 29% would qualify specialty referral (Figure 1B).

Conclusion: Validation of the AGA clinical pathway in a well-phenotyped, prospectively recruited cohort with T2DM revealed a low false negative rate, 3.3% and identified 82% of patients as low-risk. An alternative pathway with FIB-4 + ELF, which does not require VCTE prior to specialty referral, had a low false negative rate and similar performance and may serve as a viable alternative for primary care and endocrinology clinics without access to VCTE.

<b>Figure: </b>Validation of the AGA Clinical Pathway for NAFLD (A) and an expanded pathway using ELF (B) with MRE as the reference standard in patients with type 2 Diabetes Mellitus.

Figure: Validation of the AGA Clinical Pathway for NAFLD (A) and an expanded pathway using ELF (B) with MRE as the reference standard in patients with type 2 Diabetes Mellitus.

Veeral Ajmera

University of California San Diego

Appearances