Dissecting metabolic dysfunction- and alcohol-associated liver disease (MetALD) using proteomic and metabolomic profiles

Background & Aim
Metabolic dysfunction associated and alcohol associated liver disease (MetALD) is a poorly understood condition that bridges cardiometabolic and alcohol-related pathological characteristics. We aim to distinguish MetALD patients who share similar molecular signatures with alcohol-related liver disease (ALD) and those share signatures with metabolic dysfunction-associated steatotic liver disease (MASLD), and assess their prognostic risk for complications and mortality.
Methods
Our analysis involved 443,453 European participants from UK Biobank, including 34,147 with MetALD, 11,220 with ALD, and 124,034 with MASLD. We employed Elastic Net Regression to classify ALD and MASLD involving 249 plasma metabolites and/or 2,941 plasma proteins with various sensitivity analyses. We then used the selected concise model in MetALD patients to identify alcohol-predominant group (classified to ALD) and cardiometabolic-predominant group (classified to MASLD). Finally, we explored their 15-year risk of major outcomes (i.e., heart failure, myocardial infarction, stroke, cirrhosis, hepatocellular carcinoma and mortality) using Cox regression.
Results
The metabolome alone discriminated ALD from MASLD with an Area under the Curve (AUC) of 0.86, while the proteome alone achieved an AUC of 0.96. Adding age, sex, BMI, liver enzymes, or metabolome information did not enhance the AUC of the proteome model. A ten-protein model differentiated ALD and MASLD with an AUC of 0.93. This model identified that alcohol-predominant MetALD patients had significantly higher risks of mortality, and cirrhosis, along with elevated fibrosis scores and higher fibrosis stages, compared to cardiometabolic-predominant patients.
Conclusions
This study emphasizes the importance of subtyping differentiation using proteome data for personalized treatment and improved prognostic outcomes in MetALD patients.