Obesity Modifies the Performance of Fibrosis Biomarkers in Nonalcoholic Fatty Liver Disease

Abstract

Context: Guidelines recommend blood-based fibrosis biomarkers to identify advanced nonalcoholic fatty liver disease (NAFLD), which is particularly prevalent in patients with obesity.

Objective: To study whether the degree of obesity affects the performance of liver fibrosis biomarkers in NAFLD.

Design: Cross-sectional cohort study comparing simple fibrosis scores [Fibrosis-4 Index (FIB-4); NAFLD Fibrosis Score (NFS); aspartate aminotransferase to platelet ratio index; BARD (body mass index, aspartate-to-alanine aminotransferase ratio, diabetes); Hepamet Fibrosis Score (HFS)] and newer scores incorporating neo-epitope biomarkers PRO-C3 (ADAPT, FIBC3) or cytokeratin 18 (MACK-3).

Setting: Tertiary referral center.

Patients: We recruited overweight/obese patients from endocrinology (n = 307) and hepatology (n = 71) clinics undergoing a liver biopsy [median body mass index (BMI) 40.3 (interquartile range 36.0-44.7) kg/m2]. Additionally, we studied 859 less obese patients with biopsy-proven NAFLD to derive BMI-adjusted cutoffs for NFS.

Main outcome measures: Biomarker area under the receiver operating characteristic (AUROC), sensitivity, specificity, and predictive values to identify histological stage ≥F3 fibrosis or nonalcoholic steatohepatitis with ≥F2 fibrosis [fibrotic nonalcoholic steatohepatitis (NASH)].

Results: The scores with an AUROC ≥0.85 to identify ≥F3 fibrosis were ADAPT, FIB-4, FIBC3, and HFS. For fibrotic NASH, the best predictors were MACK-3 and ADAPT. The specificities of NFS, BARD, and FIBC3 deteriorated as a function of BMI. We derived and validated new cutoffs for NFS to rule in/out ≥F3 fibrosis in groups with BMIs <30.0, 30.0 to 39.9, and ≥40.0 kg/m2. This optimized its performance at all levels of BMI. Sequentially combining FIB-4 with ADAPT or FIBC3 increased specificity to diagnose ≥F3 fibrosis.

Conclusions: In obese patients, the best-performing fibrosis biomarkers are ADAPT and the inexpensive FIB-4, which are unaffected by BMI. The widely used NFS loses specificity in obese individuals, which may be corrected with BMI-adjusted cutoffs.

Keywords: biomarkers; cirrhosis; fibrosis; nonalcoholic fatty liver disease; nonalcoholic steatohepatitis; obesity.

Figure 1.

Performance of fibrosis biomarkers in the overweight/obese cohort (n = 378). (A) Areas under the receiver operating characteristic (AUROCs) for the composite scores to identify advanced fibrosis (F3-F4), significant fibrosis (F2-F4), or fibrotic nonalcoholic steatohepatitis (NASH + NAFLD Activity Score ≥ 4 + ≥F2). Whiskers denote 95% CI. The Delong’s test was used. *P < 0.05; **P < 0.01. (B) AUROCs for the composite scores to identify advanced fibrosis, based on groups divided by body mass index (BMI) quartiles (Q1-Q4). The DeLong’s test was used. (C-F) Sensitivities and specificities for the (C) lower and (D) upper cutoffs of the NAFLD Fibrosis Score (−1.455 and 0.676), and for the (E) lower and (F) upper cutoffs of the Fibrosis-4 Index (1.30 and 2.67) for advanced fibrosis, in groups divided based on BMI quartiles (Q1-Q4). Black circles and solid lines denote specificity, and white circles and dashed lines denote sensitivity. Regression lines were fitted using a quadratic model for visualization purposes.

Figure 2.

Body mass index (BMI)-adjusted cutoffs improve the performance of the NAFLD Fibrosis Score (NFS) for advanced fibrosis in all patients (n = 1232). Sensitivities and specificities for the (A) lower and (B) upper cutoffs of NFS using the standard cutoffs of −1.455 and 0.676 for advanced (F3-F4) fibrosis and for the BMI-adjusted (C) lower and (D) upper cutoffs in groups divided based on the degree of obesity. Black circles and solid lines denote specificity, and white circles and dashed lines denote sensitivity. Regression lines were fitted using a quadratic model for visualization purposes. (E) Schematic illustration of the use of BMI-adjusted NFS cutoffs to either rule in or rule out advanced fibrosis. Patients who have NFS between the upper and lower cutoffs are classified as indeterminate.

Figure 3.

Use of body mass index (BMI)-adjusted cutoffs for the NAFLD Fibrosis Score (NFS) significantly improves diagnostic performance in the overweight/obese cohort (n = 373). Flowcharts illustrating use of either the (A) standard or (B) BMI-adjusted cutoffs of NFS to identify advanced fibrosis (F3-F4). In white rectangles are shown the allocation of patients with different stages of fibrosis (F0-F4) into low (rule out ≥F3), high (rule in ≥F3), and indeterminate risk categories. Percentage values separated by slashes indicate the proportion of patients in the risk category as well as the proportion out of all patients having the same fibrosis stage. ^†−2.022 (BMI < 30.0 kg/m²), −1.083 (BMI 30.0-39.9 kg/m²), 0.544 (BMI ≥ 40 kg/m²). ^‡0.326 (BMI < 30.0 kg/m²), 1.076 (BMI 30.0-39.9 kg/m²), 2.054 (BMI ≥ 40 kg/m²). Abbreviations: LR+, positive likelihood ratio; LR-, negative likelihood ratio; NPV, negative predictive value; PPV, positive predictive value.

Figure 4.

Sequential use of the Fibrosis-4 Index (FIB-4) followed by another biomarker increases the diagnostic yield for advanced fibrosis. (A) Proposed algorithm to test patients with FIB-4 in the indeterminate range. (B) Accuracy of using FIB-4 alone compared to sequential use with either the NAFLD Fibrosis Score (NFS), the Hepamet Fibrosis Score (HFS), FIBC3, or ADAPT to identify advanced fibrosis in the overweight/obese cohort (n = 378). Cutoffs used for NFS were BMI-adjusted as follows: −2.022 (BMI < 30.0 kg/m²), −1.083 (BMI 30.0-39.9 kg/m²), and 0.544 (BMI ≥ 40 kg/m²); cutoffs used for the other scores: HFS, 0.12; FIBC3, 0.4; and ADAPT, 6.3287. Light gray bars show true positives and dark gray bars false positives. The Chi-squared test was used. *P < 0.05. Abbreviation: NS, not significant.