Machine learning to identify potential biomarkers for sarcopenia in liver cirrhosis
Department of Gastroenterology, Shanxi Provincial People's Hospital, Taiyuan 030000, Shanxi Province, China
Department of Gastroenterology, Shanxi Provincial People's Hospital, Taiyuan 030000, Shanxi Province, China
Department of Gastroenterology, Shanxi Provincial People's Hospital, Taiyuan 030000, Shanxi Province, China
Department of Gastroenterology, Shanxi Provincial People's Hospital, Taiyuan 030000, Shanxi Province, China
Department of Gastroenterology, Shanxi Provincial People's Hospital, Taiyuan 030000, Shanxi Province, China
Department of Gastroenterology, Shanxi Provincial People's Hospital, Taiyuan 030000, Shanxi Province, China
Department of Gastroenterology, Shanxi Provincial People's Hospital, Taiyuan 030000, Shanxi Province, China. doclfx@126.com
Abstract
BACKGROUND
The prevalence of sarcopenia progressively increases with as liver function deteriorates. Muscle wasting has been shown to independently predict adverse outcomes in liver cirrhosis patients.
AIM
To screen effective biomarkers for sarcopenia in liver cirrhosis.
METHODS
Untargeted metabolomics were performed on serum from 62 liver cirrhosis patients, including 41 with sarcopenia and 21 without sarcopenia. Candidate metabolite biomarkers were screened based on three machine-learning algorithms. The diagnostic or predictive value of potential biomarkers was evaluated by drawing receiver operating characteristic curves.
RESULTS
A total of 60 differential metabolites between cirrhotic sarcopenia and the non-sarcopenia group were identified. Kyoto Encyclopedia of Genes and Genomes pathway enrichment analysis revealed differential metabolites primarily involved in glycerophospholipid metabolism, alpha-linolenic acid metabolism, retrograde endocannabinoid signaling, and choline metabolism in cancer. Finally, four potential biomarkers were screened through machine learning algorithms, namely N-Acetylcarnosine, 2-Stearylcitrate, CerP (d18:1/12:0), and 3-Methyl-alpha-ionylacetate. Among these, N-Acetylcarnosine can provide better diagnostic accuracy.
CONCLUSION
This study unveiled different plasma metabolic profiles of liver cirrhosis patients with and without sarcopenia. These valuable biomarkers have the potential to improve the prognosis of liver patients with cirrhosis by early detection or prediction of sarcopenia.
Untitled section
Keywords: Cirrhosis, Sarcopenia, Untargeted metabolomics, Machine learning, Biomarkers
Article notes
Untitled section
Received 2025 Jan 20; Revised 2025 Mar 30; Accepted 2025 Apr 27; Issue date 2025 Jun 27.
Core Tip: This study unveiled different plasma metabolic profiles of liver cirrhosis patients with and without sarcopenia, which may identify valuable biomarkers for the early detection and prognosis prediction of the disease.
INTRODUCTION
Sarcopenia is the progressive loss of skeletal muscle mass and function, leading to disability, decreased quality of life, and even death[1]. Sarcopenia occurs in approximately 37% of liver cirrhosis cases, and its prevalence progressively increases with the deterioration of liver function[2]. Muscle wasting is an indicator of malnutrition, and has been shown to independently predict adverse outcomes in liver cirrhosis patients[3]. Recent studies revealed that sarcopenia significantly affects the progression of liver disease, increasing the risk of ascites, upper gastrointestinal varices, infections, hepatic encephalopathy, and death[4]. Sarcopenia in liver cirrhosis underlies complicated and multifactorial mechanisms for pathogenesis, including hyperammonemia, reduced caloric intake, systemic inflammation, hormonal changes, and physical inactivity[5].
As an emerging method, metabolomics can detect and quantify low molecular weight molecules (metabolites) in biological fluids, which can then be used to identify biomarkers and reveal the molecular mechanism of complex diseases, monitoring diseases, and evaluating risks[6,7]. A prospective cohort study revealed that valine and acetate serum levels were reduced in cirrhotic patients with sarcopenia compared with cirrhotic patients without sarcopenia. Multiple regression analysis showed that serum valine was an independent predictor of sarcopenia in liver cirrhosis. However, there were no significant differences in other stool and urine metabolites[8]. Metabolomics studies have shown some promise concerning biomarkers in sarcopenia and liver disease. Han et al[9] have reported that LysoPC (17:0), L-2-amino-3-oxobutanoic acid and palmitic acid could act as potential diagnostic biomarkers to distinguish sarcopenia from suburb-dwelling older Chinese populations. In addition, Calzadilla et al[10] identified 3-ureidopropionate, cis-3, 3-methyleneheptanoylglycine, retinol, and valine as potential biomarkers for the presence or absence of cirrhosis in alcohol-associated liver disease and suggested that their levels related to disease severity. However, few studies have investigated effective biomarkers and their diagnostic value of cirrhosis with sarcopenia through metabolomics analysis.
In this study, we collected clinical liver cirrhosis with sarcopenia and non-sarcopenia samples and conducted untargeted metabolomics to screen and identify differential metabolites in the clinic. In addition, we combined three machine learning algorithms to further explore effective biomarkers. The findings will help us better understand the development of sarcopenia in liver cirrhosis patients, providing clues to novel diagnostic, preventive, and therapeutic strategies.
MATERIALS AND METHODS
Study cohort
All patients with cirrhosis admitted to the Department of Gastroenterology, Shanxi Provincial People's Hospital from June 2021 to June 2022 were assessed for eligibility. Inclusion criteria were as follows: (1) Age ≥ 18 years; and (2) Complete abdominal computed tomography (CT) scan. Patients with malignant tumors, neuromuscular diseases, long-term bed rest, and complications with other wasting diseases such as tuberculosis or hyperthyroidism were excluded. The muscle area at the level of L3 (cm2) was measured by abdominal CT and normalized by area (m2) to obtain the skeletal muscle index at the level of the third lumbar vertebra (L3-SMI), which has been used to assess muscle mass and in the diagnosis of sarcopenia[11]. The cut-off value of L3-SMI for the diagnosis of sarcopenia is inconsistent across various races and etiologies. Based on a previous study from an Asian tertiary care center, sarcopenia was defined as L3-SMI < 36.5 cm2/m2 for men and L3-SMI < 30.2 cm2/m2 for women[12]. The study protocol was approved by the Ethics Committee of Shanxi Provincial People's Hospital, and all patients signed a written informed consent before participation in the study.
Clinical features
Demographic and anthropometric data including gender, age, height, body weight, duration of cirrhosis, and Child-Pugh class of cirrhosis were obtained from all participants. Fasting blood samples were collected to measure alanine aminotransferase, aspartate aminotransferase, albumin, total protein, and total bilirubin. Each patient underwent an abdominal CT to confirm the presence or absence of sarcopenia.
Biomarker identification
To identify differential metabolites between the two groups, model analysis was performed using orthogonal partial least squares discriminant analysis (OPLS-DA). The identified metabolites were mapped to biochemical pathways by metabolic pathway and enrichment analysis based on the Kyoto Encyclopedia of Genes and Genomes (KEGG) database. All analyses were performed in both positive and negative modes.
Least Absolute Shrinkage and Selection Operator (LASSO) regression was performed using the "glmnet" R package, applying L1 regularization to shrink some coefficients to zero and enabling feature selection[13]. In LASSO, 10-fold cross-validation (nfolds = 10) were used and the regularization parameter alpha was set to 1. The Support Vector Machine-Recursive Feature Elimination (SVM-RFE) algorithm evaluated the importance of features through the SVM model and used the iterative process of RFE to gradually eliminate unimportant features[14]. A 10-fold cross-validation was applied to rank metabolite importance in SVM-RFE. Random forest (RF) is an ensemble learning method based on decision trees, which makes predictions by constructing multiple decision trees[15]. For RF, the ntree parameter was set to 1000, and the top 10 metabolites were selected. Combining three machine learning algorithms to further screen the key metabolites of sarcopenia in liver cirrhosis ensures the robustness and consistency of results across different algorithms. The intersection of the results of the above three algorithms was utilized to obtaining the key candidate metabolites (Supplementary Table 1).
Statistical analysis
Statistical Package for the Social Sciences version 26.0 was used for statistical analysis of the clinical data. Continuous variables with a normal distribution were presented as mean ± SD and compared using independent samples t-test. Variables that did not follow a normal distribution were represented as medians and compared using the Mann-Whitney U test. Categorical variables were presented as percentages and compared using the χ2 test. Multivariate statistical analyses of untargeted metabolomics were performed by social identity model of collective action software. Differential metabolites were selected based on variable importance in the projection (VIP) values obtained from the OPLS-DA model (VIP > 1.0) and the P values from the two-tailed student t-test (P < 0.05). Machine learning was performed in R software (version 4.2.1). Receiver operating characteristic (ROC) curves were generated in GraphPad Prism 8.0 software to assess the diagnostic value of significantly different metabolites in cirrhotic patients with sarcopenia. Spearman rank correlation method was used to analyze the correlation coefficient between differential metabolites and clinical indicators, and a correlation heat map was drawn. Statistical significance was set at P < 0.05.
RESULTS
Clinical features of participants
In an untargeted metabolomics study, a total of 62 liver cirrhosis patients were recruited, including 21 non-sarcopenia patients and 41 sarcopenia patients. The demographic and clinical characteristics of study participants are presented in Table 1. The median age of non-sarcopenia patients was 50 years, with 90.5% being male. Meanwhile, the median age of sarcopenia patients was 58 years, and 63.4% of them were female. There were significant differences in the sex and age between cases and controls (P < 0.001 for sex and P = 0.002 for age). Body mass index (BMI) in the sarcopenia group was significantly lower than that in non-sarcopenia group (P = 0.002). A total of 57.1% of non-sarcopenia patients were diagnosed as Child-Pugh A and 53.7% of patients in the sarcopenia group were diagnosed as Child-Pugh C (P = 0.005). Compared to non-sarcopenia patients, those with sarcopenia showed lower creatinine (Cr) (P = 0.013) and serum ferritin (P = 0.014).
| Cirrhosis with sarcopenia (n = 41) | Cirrhosis without sarcopenia (n = 21) | P value | |
| Gender | < 0.001 | ||
| Male | 15 (36.6) | 19 (90.5) | |
| Female | 26 (63.4) | 2 (9.5) | |
| Age (years) | 58 (53.5, 63.5) | 50 (48, 56) | 0.002 |
| Body mass index (kg/m2) | 20.98 (19.06, 24.13) | 24.22 (23.03, 25.95) | 0.002 |
| Child-Pugh | |||
| A | 9 (22) | 12 (57.1) | 0.005 |
| B | 10 (24.4) | 6 (28.6) | |
| C | 22 (53.7) | 3 (14.3) | |
| Duration of cirrhosis | 0.353 | ||
| < 5 years | 21 (51.2) | 12 (57.2) | |
| 5-10 years | 10 (24.4) | 7 (33.3) | |
| > 10 years | 10 (24.4) | 2 (9.5) | |
| Alanine aminotransferase (U/L) | 22.04 (16.48, 37.47) | 29.1 (16.59, 45.79) | 0.661 |
| Aspartate aminotransferase (U/L) | 36.69 (27.64, 44.94) | 37.94 (26.34, 44.92) | 0.953 |
| Albumin (g/L) | 33.51 (27.34, 35.88) | 34.71 (28.1, 38.07) | 0.557 |
| Total protein (g/L) | 66.64 (63.37, 71.65) | 66.36 (62.34, 69.01) | 0.475 |
| Total bilirubin (mmol/L) | 22.79 (16.5, 34.85) | 32.64 (19.26, 43.5) | 0.356 |
| Urea (mmol/L) | 5.14 ± 2.56 | 5.14 ± 2.11 | 0.996 |
| Creatinine (μmol/L) | 58.12 ± 13.99 | 68.67 ± 17.66 | 0.013 |
| Ammonia (μmol/L) | 33.92 ± 7.76 | 36.24 ± 9.03 | 0.298 |
| Serum ferritin (μg/L) | 38.27 ± 38.23 | 85.92 ± 77.67 | 0.014 |
| Prothrombin time (second) | 14.2 (13.3, 15.5) | 14.8 (13.75, 17.75) | 0.054 |
Comparison of the metabolic profiles of the non-sarcopenia and sarcopenia groups
The OPLS-DA model demonstrates a clear separation in metabolic profiles between the non-sarcopenia and sarcopenia groups. In positive ion mode, a total of 12 differential metabolites were identified between the non-sarcopenia and sarcopenia groups, 5 of which were up-regulated and 7 of which were down-regulated (VIP > 1 and P < 0.5, Figure 1A and B). Figure 1C and D exhibits the results of the negative ion mode analysis, with a total of 48 differential metabolites between the two groups; 30 were significantly up-regulated and 18 were significantly down-regulated. These differential metabolites could be classified into 10 categories, with the most represented being lipids and lipid-like molecules (73.333%), followed by propanoid organic acids and derivatives (5%) and fatty acids (5%, Figure 1E). The results of KEGG pathway enrichment analysis for the differential metabolites are illustrated in Figure 1F. The 60 differential metabolites were mainly enriched in glycerophospholipid metabolism, alpha-linolenic acid metabolism, retrograde endocannabinoid signaling, and choline metabolism in cancer.
Screening of key differential metabolites
We applied three different algorithms, LASSO, SVM-RFE, and RF, to further screen key candidates from the 60 differential metabolites. LASSO analysis identified 15 core metabolites at the most appropriate lambda (λ) = 0.04029433 (Figure 2A). SVM-RFE was used to evaluate the importance of metabolites, and the top 10 metabolites were selected (Figure 2B). We then input the above 60 differential metabolites into the RF classifier, which identified the top 10 genes on an importance scale (Figure 2C). Finally, we generated the intersection of these three algorithms to identify four shared metabolic biomarkers [N-Acetylcarnosine, 2-Stearylcitrate, CerP (d18:1/12:0), 3-Methyl-alpha-ionylacetate] (Figure 2D).
Diagnostic value of key metabolites in cirrhotic sarcopenia
We performed ROC curve analysis to determine the diagnostic value of serum metabolites for diseases. The area under the ROC curve (AUC) values of N-Acetylcarnosine, 2-Stearylcitrate, CerP (d18:1/12:0), and 3-Methyl-alpha-ionylacetate were 0.8153,0.7387, 0.7085 and 0.7468, respectively (Figure 3A-D). The cut-off value of the combination of the four metabolites was 0.6747, the sensitivity was 85.37%, the specificity was 95.24%, the maximum value of the Youden index was 0.8061, and the area under the ROC value was 0.9384, showing increased predictive capacity (Figure 3E).
Clinical correlations of selected metabolites
Figure 4 illustrates the results of Spearman correlation analysis between key metabolites and clinical indicators. When we used the absolute value of the correlation coefficient r > 0.4 and P < 0.01 as a screening threshold, gender, Child grade, age, BMI, and Cr were significantly correlated with N-Acetylcarnosine. Only BMI (r = 0.41, P = 0.001) and Cr (r = 0.57, P < 0.001) were positively correlated with N-Acetylcarnosine, suggesting the potential predictive value of N-Acetylcarnosine levels for sarcopenia.
DISCUSSION
As a common complication in patients with liver cirrhosis, sarcopenia is an independent predictor of multiple adverse clinical outcomes[16,17]. Identifying biomarkers of sarcopenia in cirrhotic patients is essential for early diagnosis and subsequent treatment decisions. In this study, we used untargeted metabolomics analysis and multiple machine-learning algorithms to identify promising metabolic markers in cirrhotic sarcopenia patients. Our study identified a total of 60 differential metabolites between the two groups of cirrhotic sarcopenia and non-sarcopenia, which were significantly enriched in glycerophospholipid metabolism pathways.
All machine learning techniques contain feature selection methods, and the combination of LASSO, SVM-RFE, and RF helped us identify the variables most relevant to the disease. Additionally, the SVM-REF and LASSO algorithms reduced the risk of overfitting through cross-validation. LASSO analysis identified 15 core metabolites at the most appropriate λ = 0.04029433. We used SVM-RFE and RF to evaluate the importance of metabolites, and we selected the top 10 metabolites. We then took the intersection of all three algorithms to obtain four shared metabolic biomarkers, namely N-Acetylcarnosine, 2-Stearylcitrate, CerP (d18:1/12:0), and 3-Methyl-alpha-ionylacetate. Among them, N-acetylcarnosine was most significantly correlated with clinical indicators of sarcopenia.
Glycerophospholipids are essential components of cell membranes and function as integral membrane proteins, transporters, receptors, and ion channels[18]. Excessive glycerophospholipids in hepatocytes may cause lipid toxicity, ultimately damaging hepatocytes and increasing the risk of liver fibrosis[19]. Didymin inhibits the glycerophospholipid metabolism pathway by reducing the synthesis of phosphatidylethanolamines and phosphatidylcholines, thereby alleviating hepatocyte injury and fibrosis[20]. Lin et al[21] previously found that inhibition of glycerophospholipid metabolism is a promising strategy to reverse the progression of fibrosis. Hinkley et al[22] identified a negative correlation between phospholipids and muscle volume, suggesting that age-related muscle atrophy may be related to increased phospholipid levels in skeletal muscle. Furthermore, most phospholipid levels are increased in the skeletal muscle of old rodents compared to young rodents[23].
Carnosine is an amino acid dipeptide made of histidine and alanine. Previous in vitro and in vivo studies have shown that carnosine has antioxidant effects, metal ion chelation, and pondus hydrogenii buffering capacity[24,25]. Carnosine may be effective against diseases caused by oxidative stress, such as aging, Alzheimer's disease, atherosclerosis, and diabetic complications[26]. N-acetylcarnosine is produced by acetylation of carnosine and has similar functions to carnosine. Administration of carnosine to cirrhotic rats can effectively reduce liver hydroxyproline content, ameliorate liver fibrosis, and significantly improve locomotor activity[27]. L-carnosine is an essential element in muscle buffering capacity and in preventing loss of aging-associated muscle mass. Alanine is a key component of L-carnosine, which can promote L-carnosine synthesis in skeletal muscle, thereby increasing muscle endurance and physical activity[28]. Our results showed that N-Acetylcarnosine has the best predictive performance of liver cirrhosis complicated with sarcopenia (AUC = 0.8153), which was positively correlated with BMI and Cr. A previous study demonstrated that BMI was the main influencing factor of cirrhosis combined with sarcopenia, and patients with low BMI have a higher risk of sarcopenia[29]. In liver cirrhosis patients, Cr levels are often reduced due to anorexia and muscle loss. The creatinine/cystatin ratio, also known as the sarcopenia index, is an effective predictor of malnutrition in patients with cirrhosis[30].
As an important substrate in cellular energy metabolism, citrate is involved in a variety of biological processes, including metabolism, inflammation, cancer, insulin secretion, histone acetylation, and non-alcoholic fatty liver disease[31]. Mitochondrial damage and fatty acid accumulation in hepatocytes in metabolic dysfunction-associated steatotic liver disease can lead to dysfunction of the citrate cycle[32]. In an independent cohort of subjects with biopsy-confirmed metabolic dysfunction-associated steatohepatitis, increased citrate was positively associated with the degree of liver fibrosis[33]. Holeček and Vodeničarovová[34] confirmed that the weights and protein contents of muscles in the carbon tetrachloride-induced liver cirrhosis rat model are low, making it a suitable model to study muscle wasting in human liver cirrhosis. Cirrhotic rats had decreased α-ketoglutarate levels in their muscle and tricarboxylic acid cycle disorders. However, there was no significant change in intermediate citrate concentration[34]. Our data demonstrated that 2-Stearylcitrate levels were lower in cirrhotic patients with sarcopenia than in non-sarcopenia patients, which has some predictive value for cirrhotic sarcopenia (AUC = 0.7387).
CerP (d18:1/12:0) is a phosphorylated form of ceramide, which is a useful biomarker for predicting sarcopenia in liver cirrhosis. As the central molecule of sphingolipid metabolism, ceramide has pro-apoptosis and anti-proliferative effects and participates in tissue homeostasis[35]. It has been reported that ceramide regulates collagen production by activating the transforming growth factor beta (TGF-β)/small mother against decapentaplegic (Smad) pathway, leading to liver fibrosis. Iwanaga et al[36] found that miglustat can prevent and reverse liver fibrosis by inhibiting the TGF-β/Smad pathway. High circulating sphingolipids levels, especially ceramides and modified ceramides, are common features of cancer cachexia that lead to tissue depletion and skeletal muscle atrophy by inhibiting anabolic signals[37]. Consistent with previous results, CerP (d18:1/12:0) levels were higher in cirrhotic patients with sarcopenia than in non-sarcopenia patients.
Sesquiterpenes are potential novel bioactive compounds for the prevention and treatment of rheumatoid arthritis. They are primarily derived from plant extracts and demonstrate superior activity and tolerability[38]. 3-Methyl-alpha-ionylacetate is a sesquiterpene, whose expression level was lower in cirrhotic sarcopenia compared to non-sarcopenia, which has a certain value for disease diagnosis (AUC = 0.7468).
Despite its contributions, our study has certain limitations. Firstly, the present study was limited by small sample sizes. In future studies, we will expand the sample sizes to improve statistical power and increase the generalizability across different patient populations. In addition, we will further analyze the influence of different subgroups such as gender, age, BMI, and Child-Pugh on the metabolomics of sarcopenia in liver cirrhosis. Secondly, the biomarkers identified in this study should be further subjected to model building to judge their diagnostic performance in a validation cohort. Thirdly, the combined application of LASSO, SVM-RFE, and RF in this study to screen biomarkers associated with cirrhotic sarcopenia reduced bias to the maximum extent. However, the biomarkers identified in this study have not been mechanistically investigated. Further experiments are needed to elucidate the exact mechanism and potential therapeutic interventions of biomarkers in cirrhotic sarcopenia.
CONCLUSION
Using an untargeted metabolomics approach, the present study identified differential metabolites in liver cirrhosis patients with or without sarcopenia. A panel of four candidate biomarkers demonstrated good predictive performance for liver cirrhosis patients with sarcopenia. These results provide a new theoretical basis for studying the pathogenesis of cirrhosis complicated with sarcopenia and may provide potential targets for treatment.
Footnotes
Footnote Group
Contributor Information
Qian-Yu Liang, Department of Gastroenterology, Shanxi Provincial People's Hospital, Taiyuan 030000, Shanxi Province, China.
Jun Wang, Department of Gastroenterology, Shanxi Provincial People's Hospital, Taiyuan 030000, Shanxi Province, China.
Yun-Feng Yang, Department of Gastroenterology, Shanxi Provincial People's Hospital, Taiyuan 030000, Shanxi Province, China.
Kai Zhao, Department of Gastroenterology, Shanxi Provincial People's Hospital, Taiyuan 030000, Shanxi Province, China.
Rui-Li Luo, Department of Gastroenterology, Shanxi Provincial People's Hospital, Taiyuan 030000, Shanxi Province, China.
Ye Tian, Department of Gastroenterology, Shanxi Provincial People's Hospital, Taiyuan 030000, Shanxi Province, China.
Feng-Xia Li, Department of Gastroenterology, Shanxi Provincial People's Hospital, Taiyuan 030000, Shanxi Province, China. doclfx@126.com.
Data sharing statement
The datasets used and/or analyzed during the current study are available from the corresponding author upon reasonable request.
References
Untitled section
References
- 1.Cruz-Jentoft AJ, Bahat G, Bauer J, Boirie Y, Bruyère O, Cederholm T, Cooper C, Landi F, Rolland Y, Sayer AA, Schneider SM, Sieber CC, Topinkova E, Vandewoude M, Visser M, Zamboni M Writing Group for the European Working Group on Sarcopenia in Older People 2 (EWGSOP2), and the Extended Group for EWGSOP2. Sarcopenia: revised European consensus on definition and diagnosis. Age Ageing. 2019;48:16–31. doi: 10.1093/ageing/afy169.
- 2.Di Cola S, Khan S, Lapenna L, Merli M. Emerging drugs for the treatment of sarcopenia in cirrhosis of the liver. Expert Opin Emerg Drugs. 2024;29:81–91. doi: 10.1080/14728214.2024.2332428.
- 3.Kim G, Kang SH, Kim MY, Baik SK. Prognostic value of sarcopenia in patients with liver cirrhosis: A systematic review and meta-analysis. PLoS One. 2017;12:e0186990. doi: 10.1371/journal.pone.0186990.
- 4.Maslennikov R, Alieva A, Poluektova E, Zharikov Y, Suslov A, Letyagina Y, Vasileva E, Levshina A, Kozlov E, Ivashkin V. Sarcopenia in cirrhosis: Prospects for therapy targeted to gut microbiota. World J Gastroenterol. 2023;29:4236–4251. doi: 10.3748/wjg.v29.i27.4236.
- 5.Ponziani FR, Gasbarrini A. Sarcopenia in Patients with Advanced Liver Disease. Curr Protein Pept Sci. 2018;19:681–691. doi: 10.2174/1389203718666170428121647.
- 6.Rangel-Huerta OD, Pastor-Villaescusa B, Gil A. Are we close to defining a metabolomic signature of human obesity? A systematic review of metabolomics studies. Metabolomics. 2019;15:93. doi: 10.1007/s11306-019-1553-y.
- 7.Szczerbinski L, Wojciechowska G, Olichwier A, Taylor MA, Puchta U, Konopka P, Paszko A, Citko A, Goscik J, Fiehn O, Fan S, Wasilewska A, Taranta-Janusz K, Kretowski A. Untargeted Metabolomics Analysis of the Serum Metabolic Signature of Childhood Obesity. Nutrients. 2022;14:214. doi: 10.3390/nu14010214.
- 8.Aliwa B, Horvath A, Traub J, Feldbacher N, Habisch H, Fauler G, Madl T, Stadlbauer V. Altered gut microbiome, bile acid composition and metabolome in sarcopenia in liver cirrhosis. J Cachexia Sarcopenia Muscle. 2023;14:2676–2691. doi: 10.1002/jcsm.13342.
- 9.Han P, Yuan C, Chen X, Hu Y, Hu X, Xu Z, Guo Q. Metabolic signatures and potential biomarkers of sarcopenia in suburb-dwelling older Chinese: based on untargeted GC-MS and LC-MS. Skelet Muscle. 2024;14:4. doi: 10.1186/s13395-024-00337-3.
- 10.Calzadilla N, Zilberstein N, Hanscom M, Al Rashdan HT, Chacra W, Gill RK, Alrefai WA. Serum metabolomic analysis in cirrhotic alcohol-associated liver disease patients identified differentially altered microbial metabolites and novel potential biomarkers for disease severity. Dig Liver Dis. 2024;56:923–931. doi: 10.1016/j.dld.2023.10.006.
- 11.Zhang Y, Wei L, Chang C, Duan F, Quan M, Yang S. Sarcopenia defined with L3-SMI is an independent predictor of survival in male patients with ARLD in mainland China. Front Nutr. 2023;10:1238433. doi: 10.3389/fnut.2023.1238433.
- 12.Benjamin J, Shasthry V, Kaal CR, Anand L, Bhardwaj A, Pandit V, Arora A, Rajesh S, Pamecha V, Jain V, Kumar G, Loria A, Puri P, Joshi YK, Sarin SK. Characterization of body composition and definition of sarcopenia in patients with alcoholic cirrhosis: A computed tomography based study. Liver Int. 2017;37:1668–1674. doi: 10.1111/liv.13509.
- 13.Tibshirani R. Regression Shrinkage and Selection via The Lasso: A Retrospective. J R Stat Soc B. 2011;73:273–282.
- 14.Sanz H, Valim C, Vegas E, Oller JM, Reverter F. SVM-RFE: selection and visualization of the most relevant features through non-linear kernels. BMC Bioinformatics. 2018;19:432. doi: 10.1186/s12859-018-2451-4.
- 15.Mantas CJ, Castellano JG, Moral-garcía S, Abellán J. A comparison of random forest based algorithms: random credal random forest versus oblique random forest. Soft Comput. 2019;23:10739–10754.
- 16.Zeng X, Shi ZW, Yu JJ, Wang LF, Luo YY, Jin SM, Zhang LY, Tan W, Shi PM, Yu H, Zhang CQ, Xie WF. Sarcopenia as a prognostic predictor of liver cirrhosis: a multicentre study in China. J Cachexia Sarcopenia Muscle. 2021;12:1948–1958. doi: 10.1002/jcsm.12797.
- 17.Jiang M, Hua X, Wu M, Wu J, Xu X, Li J, Meng Q. Longitudinal changes in sarcopenia was associated with survival among cirrhotic patients. Front Nutr. 2024;11:1375994. doi: 10.3389/fnut.2024.1375994.
- 18.Chang H, Meng HY, Liu SM, Wang Y, Yang XX, Lu F, Wang HY. Identification of key metabolic changes during liver fibrosis progression in rats using a urine and serum metabolomics approach. Sci Rep. 2017;7:11433. doi: 10.1038/s41598-017-11759-z.
- 19.Kartsoli S, Kostara CE, Tsimihodimos V, Bairaktari ET, Christodoulou DK. Lipidomics in non-alcoholic fatty liver disease. World J Hepatol. 2020;12:436–450. doi: 10.4254/wjh.v12.i8.436.
- 20.Li Y, Li C, Xiong Y, Fang B, Lin X, Huang Q. Didymin Ameliorates Liver Fibrosis by Alleviating Endoplasmic Reticulum Stress and Glycerophospholipid Metabolism: Based on Transcriptomics and Metabolomics. Drug Des Devel Ther. 2022;16:1713–1729. doi: 10.2147/DDDT.S351092.
- 21.Lin X, Wei Y, Li Y, Xiong Y, Fang B, Li C, Huang Q, Huang R, Wei J. Tormentic Acid Ameliorates Hepatic Fibrosis in vivo by Inhibiting Glycerophospholipids Metabolism and PI3K/Akt/mTOR and NF-κB Pathways: Based on Transcriptomics and Metabolomics. Front Pharmacol. 2022;13:801982. doi: 10.3389/fphar.2022.801982.
- 22.Hinkley JM, Cornnell HH, Standley RA, Chen EY, Narain NR, Greenwood BP, Bussberg V, Tolstikov VV, Kiebish MA, Yi F, Vega RB, Goodpaster BH, Coen PM. Older adults with sarcopenia have distinct skeletal muscle phosphodiester, phosphocreatine, and phospholipid profiles. Aging Cell. 2020;19:e13135. doi: 10.1111/acel.13135.
- 23.Uchitomi R, Hatazawa Y, Senoo N, Yoshioka K, Fujita M, Shimizu T, Miura S, Ono Y, Kamei Y. Metabolomic Analysis of Skeletal Muscle in Aged Mice. Sci Rep. 2019;9:10425. doi: 10.1038/s41598-019-46929-8.
- 24.Tanaka KI, Kawahara M. Carnosine and Lung Disease. Curr Med Chem. 2020;27:1714–1725. doi: 10.2174/0929867326666190712140545.
- 25.Min J, Senut MC, Rajanikant K, Greenberg E, Bandagi R, Zemke D, Mousa A, Kassab M, Farooq MU, Gupta R, Majid A. Differential neuroprotective effects of carnosine, anserine, and N-acetyl carnosine against permanent focal ischemia. J Neurosci Res. 2008;86:2984–2991. doi: 10.1002/jnr.21744.
- 26.Mehmetçik G, Ozdemirler G, Koçak-Toker N, Cevikbaş U, Uysal M. Role of carnosine in preventing thioacetamide-induced liver injury in the rat. Peptides. 2008;29:425–429. doi: 10.1016/j.peptides.2007.11.008.
- 27.Jamshidzadeh A, Heidari R, Latifpour Z, Ommati MM, Abdoli N, Mousavi S, Azarpira N, Zarei A, Zarei M, Asadi B, Abasvali M, Yeganeh Y, Jafari F, Saeedi A, Najibi A, Mardani E. Carnosine ameliorates liver fibrosis and hyperammonemia in cirrhotic rats. Clin Res Hepatol Gastroenterol. 2017;41:424–434. doi: 10.1016/j.clinre.2016.12.010.
- 28.Gasmi A, Mujawdiya PK, Lysiuk R, Shanaida M, Peana M, Piscopo S, Beley N, Dzyha S, Smetanina K, Shanaida V, Resimont S, Bjorklund G. The Possible Roles of β-alanine and L-carnosine in Anti-aging. Curr Med Chem. 2025;32:6–22. doi: 10.2174/0109298673263561231117054447.
- 29.Xiao HJ, Ye Q, Zhang M, Qi YM, Han T, Wang X. [Risk factors of cirrhosis combined with sarcopenia and their impact on clinical outcomes] Zhonghua Gan Zang Bing Za Zhi. 2020;28:53–57. doi: 10.3760/cma.j.issn.1007-3418.2020.01.013.
- 30.Wu YK, Li M, Zhang YC, Gao RZ, Su Y, Zhou Y, Zhao KL, Chen C, Wang WX. The sarcopenia index is an effective predictor for malnutrition in patients with liver cirrhosis. Nutr Diet. 2022;79:563–571. doi: 10.1111/1747-0080.12738.
- 31.Iacobazzi V, Infantino V. Citrate--new functions for an old metabolite. Biol Chem. 2014;395:387–399. doi: 10.1515/hsz-2013-0271.
- 32.Chella Krishnan K, Kurt Z, Barrere-Cain R, Sabir S, Das A, Floyd R, Vergnes L, Zhao Y, Che N, Charugundla S, Qi H, Zhou Z, Meng Y, Pan C, Seldin MM, Norheim F, Hui S, Reue K, Lusis AJ, Yang X. Integration of Multi-omics Data from Mouse Diversity Panel Highlights Mitochondrial Dysfunction in Non-alcoholic Fatty Liver Disease. Cell Syst. 2018;6:103–115.e7. doi: 10.1016/j.cels.2017.12.006.
- 33.Morrow MR, Batchuluun B, Wu J, Ahmadi E, Leroux JM, Mohammadi-Shemirani P, Desjardins EM, Wang Z, Tsakiridis EE, Lavoie DCT, Reihani A, Smith BK, Kwiecien JM, Lally JSV, Nero TL, Parker MW, Ask K, Scott JW, Jiang L, Paré G, Pinkosky SL, Steinberg GR. Inhibition of ATP-citrate lyase improves NASH, liver fibrosis, and dyslipidemia. Cell Metab. 2022;34:919–936.e8. doi: 10.1016/j.cmet.2022.05.004.
- 34.Holeček M, Vodeničarovová M. Muscle wasting and branched-chain amino acid, alpha-ketoglutarate, and ATP depletion in a rat model of liver cirrhosis. Int J Exp Pathol. 2018;99:274–281. doi: 10.1111/iep.12299.
- 35.Simon J, Ouro A, Ala-Ibanibo L, Presa N, Delgado TC, Martínez-Chantar ML. Sphingolipids in Non-Alcoholic Fatty Liver Disease and Hepatocellular Carcinoma: Ceramide Turnover. Int J Mol Sci. 2019;21:40. doi: 10.3390/ijms21010040.
- 36.Iwanaga T, Chiba T, Nakamura M, Kaneko T, Ao J, Qiang N, Ma Y, Zhang J, Kogure T, Yumita S, Ishino T, Ogawa K, Kan M, Nakagawa M, Fujiwara K, Fujita N, Sakuma T, Kanzaki H, Koroki K, Kusakabe Y, Inoue M, Kobayashi K, Kanogawa N, Kiyono S, Kondo T, Nakagawa R, Ogasawara S, Nakamoto S, Muroyama R, Kato J, Kanda T, Maruyama H, Mimura N, Honda T, Murayama T, Nakamura H, Kato N. Miglustat, a glucosylceramide synthase inhibitor, mitigates liver fibrosis through TGF-β/Smad pathway suppression in hepatic stellate cells. Biochem Biophys Res Commun. 2023;642:192–200. doi: 10.1016/j.bbrc.2022.12.025.
- 37.Morigny P, Zuber J, Haid M, Kaltenecker D, Riols F, Lima JDC, Simoes E, Otoch JP, Schmidt SF, Herzig S, Adamski J, Seelaender M, Berriel Diaz M, Rohm M. High levels of modified ceramides are a defining feature of murine and human cancer cachexia. J Cachexia Sarcopenia Muscle. 2020;11:1459–1475. doi: 10.1002/jcsm.12626.
- 38.Han J, Wu B, Wang D. The potential efficacy of sesquiterpenes and their derivatives in treating rheumatoid arthritis: A systematic review. Int Immunopharmacol. 2024;141:112946. doi: 10.1016/j.intimp.2024.112946.
Associated Data
Data Availability Statement
The datasets used and/or analyzed during the current study are available from the corresponding author upon reasonable request.