Identification of Candidates for MASLD Treatment with Indeterminate Vibration-Controlled Transient Elastography.

Abstract

Background & aims: A noteworthy proportion of patients with metabolic dysfunction-associated steatotic liver disease (MASLD) have an indeterminate vibration-controlled transient elastography (VCTE). Among these patients, we aimed to identify candidates for MASLD treatment by diagnosing significant fibrosis.

Methods: Real-world prospective study including a large dataset of MASLD patients with paired VCTE and liver biopsy from 6 centers. A total of n=1196 patients were recruited and divided in training (3 centers, Spain), internal validation (2 centers, Spain), and external validation (1 center, United States) cohorts. In patients with indeterminate liver stiffness measurements (LSM:8-12 kPa), a diagnostic algorithm was developed to identify significant fibrosis, defined as histological stage ≥F2. Statistical analysis was performed using gaussian mixture model (GMM) and k-means unsupervised clusterization.

Results: From the eligible population, 33%, 29%, and 31% had indeterminate VCTE in the training, internal and external validation samples, respectively. Controlled attenuation parameter (CAP) allowed the differentiation of GMM clusters with a cut-off of 280 dB/m (AUC:0.89 [95%CI:0.86-0.97]). Within patients with <280 dB/m, a LSM between 8.0-9.0 kPa showed a 93% sensitivity and a 91% negative predictive value to exclude significant fibrosis. Among patients with ≥280 dB/m, a LSM between 10.3-12.0 kPa diagnosed significant fibrosis with a 91% specificity. Applying this algorithm to the validation cohorts, 36% of the indeterminate VCTE were re-allocated. The re-allocated high-risk group showed a prevalence of 86% significant fibrosis, opening the therapeutic window for MASLD patients.

Conclusion: To identify candidates for MASLD treatment among indeterminate VCTE, an algorithm-based on the sequential combination of LSM and CAP thresholds can optimize the diagnosis of moderate-to-advanced fibrosis.