Abstract

Background: While obesity is a risk factor for NAFLD, patients across the BMI spectrum are affected by the disease, creating a need for reliable non-invasive tests (NITs) with performances that are not affected by BMI. While most of standard NITs are designed to detect advanced fibrosis, NIS2+™, an optimization of the blood-based NIS4® technology, is specifically designed to detect at-risk NASH (NAS≥4; F≥2). We aimed to isolate the effect of BMI on NITs and assess their clinical reliability across the BMI spectrum.

Methods: Among all non-cirrhotic NASH patients enrolled in the RESOLVE-IT Phase 3 trial (NCT02704403), those with data for NIS2+^TM, APRI, NFS, FIB-4, ELF^TM and FibroScan (FS) were selected (n=898). This cohort was split in 4 BMI-based subgroups: non-obese, Class 1, 2 and 3 obesity. To isolate the effect of BMI from confounding factors, we matched the 4 groups for the histology and other comorbidities using a propensity score matching algorithm, resulting in 4 groups of n=113 patients. One-way ANOVA tests were used to evaluate the BMI impact on NITs and biomarkers distribution. Impact on clinical performances (sensitivity, specificity) was also analyzed using fixed cutoffs.

Results: NFS was impacted by BMI (p<0.0001), with scores increasing along with BMI. The significant decrease in albumin concentration with BMI (p<0.0001) and the presence of BMI in the NFS equation explain the NFS results. FS distribution was significantly impacted by BMI (p<0.0001), displaying increased mean scores in Class 3 obesity compared to other groups (14.3 kPa vs 10.1-11.0kPa). The BMI impact on NFS and FS distributions resulted in a decrease in specificity with increasing BMI when ruling-out (NFS: 76% to 20%; FS: 49% to 33%) and ruling-in (NFS:100% to 83%; FS 76% to 48%) F≥3. While NFS sensitivity progressively increased with BMI when ruling-out (NFS: 52% to 90%) and ruling-in (NFS:2% to 33%) F≥3, FS achieved the highest sensitivity in class 3 obese patients compared to other groups (rule-out: 94% vs 76-88%; rule-in: 82% vs 60-68%). NIS2+^TM, APRI, ELF^TM and FIB-4 were not significantly impacted by BMI, resulting in stable clinical performance.

Conclusion: NIS2+^TM, FIB-4, APRI and ELF^TM were not significantly impacted by BMI. NFS was, however, significantly impacted by BMI, notably due to BMI-associated differences in albumin levels. Liver stiffness by FS was also significantly impacted in Class 3 obesity. This suggests a need for BMI-adapted cutoffs for these particular NITs.