Recently, the study "Noninvasive proteomic biomarkers for alcohol-related liver disease" by the team of Matthias Mann was published in Nature Medicine, a leading international medical research journal. The study proposes a paired liver-plasma proteomics approach to infer molecular pathophysiology, and explores the diagnostic and prognostic capabilities of the plasma proteome in 596 individuals (137 healthy people and 459 patients with ALD).
Background Alcohol-related liver disease (ALD) is a leading cause of liver-related death worldwide. ALD progresses through a series of histological lesions, from alcohol-related fatty liver to steatohepatitis with no overt clinical symptoms, gradually advancing to progressive fibrosis and ultimately cirrhosis. If ALD is diagnosed at an early stage, disease progression can be slowed or prevented through intensified alcohol rehabilitation and treatment of metabolic comorbidities. However, the slowly progressive and asymptomatic nature of the disease makes early diagnosis challenging, and accurate diagnosis of liver disease requires biopsy. There is therefore an urgent need for minimally invasive diagnostic strategies to screen patients in high-risk populations. Recent advances in mass spectrometry-based proteomics have greatly expanded its range of applications in biomedical research, making it possible to identify and quantify hundreds to thousands of proteins specifically in biological or clinical samples. Study Design Cohort selection: 1. ALD patients: the GALA-ALD cohort (n = 459), with inclusion criteria of a history of alcohol abuse >1 year (>24 g d-1 for women, >36 g d-1 for men), age 18-75 years, and informed consent to undergo liver biopsy. 2. Healthy individuals: the GALA-HP cohort (n = 137), which recruited healthy controls partially matched for age (40-75 years), sex and BMI. 3. The ALD validation cohort (n = 63) consisted of independent participants from a population screening study that began after the GALA-ALD study ended, with inclusion criteria of a history of alcohol abuse > 5 years (≥24 g d–1 for women, ≥36 g d–1 for men) and age 30-75 years.
Proteomic testing workflow Analytical Approach Integrated analysis of the differential proteomes in liver and plasma from the ALD and healthy cohorts revealed plasma proteins associated with disease stage that were consistent with those in the liver. Finally, a machine learning model was built to identify the early stages of liver fibrosis, inflammatory activity and steatosis, and the diagnostic performance of the identified protein markers was evaluated in an independent cohort of 63 patients with ALD.
Figure 1 Approach to biomarker discovery in liver disease Results Impact of liver pathology on the liver and plasma proteomes A total of 5,515 proteins were identified in liver tissue using DIA technology, and differential analysis yielded 717 differential unique proteins. Fibrosis had the greatest impact on the liver proteome, followed by inflammation, and 84% of all proteins dysregulated in inflammation were also dysregulated in fibrosis (Figure 2b). Among the top 20 proteins most closely associated with fibrosis stage, there were notable liver-secreted proteins (Figure 2d). Approximately 50% of the top 20 inflammation-related proteins were cytoskeletal proteins, which may indicate structural changes in the cells of the inflamed liver.
Figure 2 Differential changes in the liver proteome caused by liver pathology Integrated analysis of the liver and plasma proteomes A total of 420 proteins were quantified in both liver and plasma (Figure 3a), and 46 proteins in total were shown by ANCOVA analysis of the liver and plasma proteomes to be co-dysregulated. These proteins are involved in immune and inflammatory responses, cell adhesion, ECM organization, protease inhibitors and intracellular enzymes (Figure 3g). Their co-dysregulation may result from immune cell infiltration, increased systemic inflammation, ECM remodeling and scar tissue formation, and tissue leakage. Among them, PIGR, LGALS3BP, TGFBI and C7 are all previously reported markers of liver disease with recognized significance.
Figure 3 Integrated analysis of the liver-plasma proteome Biomarker panels for early ALD pathology The study first compared 22 state-of-the-art machine learning classifiers and selected a logistic regression model as the final classifier. A panel of 9 protein markers was ultimately chosen to identify significant fibrosis, 6 proteins to identify mild inflammation, and 12 proteins to identify hepatic steatosis. Of the 22 proteins included in the three panels, 8 were co-regulated in both plasma and liver, namely C7, LGALS3BP, TGFBI, ALDOB, CLEC3B, FBLN1, ATRN and SERPINF1.
Validation of model performance in an independent cohort In the independent ALD validation cohort (n=63), the proteomic models for fibrosis stage 2 (F2) and mild inflammation stage 2 (I2) produced ROC-AUC values higher than any of the 10 comparator measures available in the cohort, with ROC-AUC values of 0.87 and 0.81 respectively. An independent validation cohort reproduced the performance of the diagnostic model, laying the foundation for routine mass spectrometry-based testing for liver disease.
Summary The research team explored whether liver-related disease could be detected using state-of-the-art mass spectrometry-based proteomic assays. The results show convincingly that proteomic testing can effectively explore and discover biomarker panels for three key pathological features of the liver, performing considerably better than the best currently available clinical tests collected for comparison. The findings also carry implications for the development of potential ALD treatments. For example, 20% of the upregulated tissue proteins are related to signal transduction pathways such as receptor tyrosine kinases, as well as ALD signaling pathways that are less well studied but are nonetheless druggable, including G protein-coupled receptors. In summary, this study confirms previous findings and provides potential protein targets with diagnostic, prognostic and therapeutic value for ALD. |
