Precision in Liver Diagnosis: Varied Accuracy Across Subgroups and the Need for Variable Thresholds in Diagnosis of MASLD

Abstract

Background and Aims: The performance of non-invasive liver tests (NITs) is known to vary across settings and subgroups. We systematically evaluated whether the performance of three NITs in detecting advanced fibrosis in patients with metabolic dysfunction-associated steatotic liver disease (MASLD) varies with age, sex, body mass index (BMI), type 2 diabetes mellitus (T2DM) status or liver enzymes. Methods: Data from 586 adult LITMUS Metacohort participants with histologically characterised MASLD were included. The diagnostic performance of the Fibrosis-4 Index (FIB-4), enhanced liver fibrosis (ELF) and vibration-controlled transient elastography liver stiffness measurement (VCTE LSM) was evaluated. Performance was expressed as the area under the receiver operating characteristics curve (AUC). Thresholds for detecting advanced fibrosis (≥F3) were calculated for each NIT for fixed (high) sensitivity, specificity and predictive values. Results: Differences in AUC between all subgroups were small and statistically not significant, indicating comparable performance in detecting ≥F3, irrespective of these clinical factors. However, different thresholds were needed to achieve the same level of accuracy with each test. For example, for a fixed sensitivity and specificity, the thresholds for all three NITs were higher in patients with T2DM. Effects for sex, age and liver enzymes were less pronounced. Conclusions: Performance of the selected NITs in detecting advanced liver fibrosis does not vary substantially with clinical characteristics. However, different thresholds have to be selected to achieve the same sensitivity, specificity and predictive values in the respective subgroups. Large prospective studies are called for to study NIT accuracy considering multiple patient characteristics.