P4 HE Audio Recording to Detect Minimal Hepatic Encephalopathy (“HEAR-MHE”) Study: Speech Identifies Minimal Hepatic Encephalopathy

Background: There is demand for a simple diagnostic test to identify minimal hepatic encephalopathy (MHE). Current tools are burdensome, therefore rarely used. We aimed to evaluate if recorded speech can diagnose MHE. Methods: We enrolled 212 patients (169 cirrhosis and 43 controls with non-cirrhotic liver disease) from 2 centers. In the office, patients underwent psychometric HE score (PHES), animal naming test (ANT), and audio recording while reading a paragraph. Patients performed additional home audio recordings. Speech variables were automatically extracted (Figure 1). Patients with cirrhosis were in 3 non-overlapping categories: (1) prior overt HE (OHE), (2) MHE with no prior OHE (PHES ≤ -4), (3) no MHE. Linear and logistic regression predicted PHES and MHE respectively. Vuong test compared areas under the curve (AUC). Results: Cirrhosis: median 63 years (IQR 55, 68), 52% male, median MELD 9 (IQR 7, 12), 39% alcohol and 33% fatty liver. Controls: median 55 years (IQR 42, 62), 56% male, and 58% had fatty liver disease and 16% viral hepatitis. Audio recordings were median 36 seconds (IQR 33, 41). Speech rate was significantly slower in patients with MHE and history of OHE than patients without MHE or controls (Figure 1). Office speech associates with PHES: 82 speech variables associated with PHES in the overall cohort (P<0.05 with false discovery rate [FDR] adjustment). A model of 5 speech variables (2 speech tempo and 3 acoustic) was associated with PHES (r2=0.28, P<0.0001). Office speech associates with MHE: 71 speech variables differed between patients with and without MHE (P<0.05 with FDR adjustment). A model of 2 speech variables (1 speech tempo and 1 acoustic) was comparable to ANT to identify MHE (AUC 0.70 vs 0.66, P=0.19). Adding age and ANT but not MELD or years of education improved the model (AUC 0.75, P=0.01). Home speech associates with MHE: Patients performed 43 home audio recordings. The same speech model from office recordings that predicted MHE was able to predict MHE from home recordings (AUC 0.73). Conclusion: Recorded speech was associated with PHES and identified MHE. Speech is simple to record, provides immediate point-of-care data, and represents a promising biomarker in HE.Figure 1.: Speech variables were automatically extracted (A), and speech rate is slower in patients with MHE and history of OHE than patients without MHE or controls (B).