Characterization of the gut‐liver‐muscle axis in cirrhotic patients with sarcopenia

Sarcopenia is frequent in cirrhosis and is associated with unfavourable outcomes. The role of the gut‐liver‐muscle axis in this setting has been poorly investigated. The aim of this study was to identify gut microbiota, metabolic and inflammatory signatures associated with sarcopenia in cirrhotic patients.


| INTRODUC TI ON
Muscle wasting and physical function impairment affect 30%-70% of patients with cirrhosis. 1 Several factors are invoked in the pathogenesis of muscle atrophy in this patient population. Reduced use of carbohydrates as a source of energy at the expense of proteins, hormonal changes, malnutrition, gastrointestinal dysfunction, increased resting energy expenditure and low physical activity are responsible of the loss of muscle mass in cirrhotic patients. 2 Sarcopenia has been associated with poor quality of life and has been shown to independently predict negative outcomes, including reduced survival, in patients with cirrhosis. 3 Interestingly, while liver transplantation can cure cirrhosis, its effects on sarcopenia are variable with either improvement, stability or worsening being reported in literature. 3,4 Therefore, a better understanding of the pathophysiology of sarcopenia in this setting is needed to plan targeted therapeutic interventions. 5 In older adults, sarcopenia has been associated with a state of chronic, low-grade inflammation, termed inflamm-ageing. This condition shares the same pathophysiological traits of metainflammation, which is typical of metabolic disorders and is associated with changes in the gut microbiota. 6 Cirrhosis is a paradigm of chronic inflammation driven by gut dysbiosis and increased intestinal permeability. However, whether alterations in gut microbiota are associated with sarcopenia in cirrhotic patients has yet to be established.
To fill this gap in knowledge, the present study aimed at characterizing the gut-liver-muscle axis and identifying gut microbial, metabolic and inflammatory signatures of sarcopenia in patients with cirrhosis.

| Patients selection and study procedures
All patients with cirrhosis consecutively admitted from 1 January to 31 June 2019 at the Hepatology Clinic of the Fondazione Policlinico controls. A multiomic analysis, including gut microbiota composition and metabolomics, serum myokines and systemic and intestinal inflammatory mediators, was performed.

Results:
The gut microbiota of sarcopenic cirrhotic patients was poor in bacteria associated with physical function (Methanobrevibacter, Prevotella and Akkermansia), and was enriched in Eggerthella, a gut microbial marker of frailty. The abundance of potentially pathogenic bacteria, such as Klebsiella, was also increased, to the detriment of autochthonous ones. Sarcopenia was associated with elevated serum levels of pro-inflammatory mediators and of fibroblast growth factor 21 (FGF21) in cirrhotic patients. Gut microbiota metabolic pathways involved in amino acid, protein and branched-chain amino acid metabolism were up-regulated, in addition to ethanol, trimethylamine and dimethylamine production. Correlation networks and clusters of variables associated with sarcopenia were identified, including one centred on Klebsiella/ethanol/FGF21/Eggerthella/Prevotella.

Conclusions:
Alterations in the gut-liver-muscle axis are associated with sarcopenia in patients with cirrhosis. Detrimental but also compensatory functions are involved in this complex network.

K E Y W O R D S
cirrhosis, ethanol, gut microbiota, metabolomics, sarcopenia

Lay summary
• The role of the gut-liver-muscle axis in the development of sarcopenia in patients with cirrhosis is largely unknown.
• The gut microbiota contributes to sarcopenia through metabolic and pro-inflammatory networks, but, at the same time, it may exert compensatory functions to limit muscle wasting.
• Our findings can help to identify patients with cirrhosis who may benefit from personalized treatments to halt the progression of muscle wasting.
Universitario Agostino Gemelli IRCCS in Rome were assessed for eligibility. Inclusion criteria were as follows: age ≥ 18 years, absence of systemic or intestinal pathologies associated with gut microbiota alterations (eg celiac disease, inflammatory bowel diseases, diabetes mellitus etc) and complete abstinence from alcohol consumption for at least one year. Patients with previous or active tumours, chronic neurodegenerative or muscle diseases, use of probiotics, prebiotics or antibiotics during the previous 3 months, and those on vegetarian or vegan diet were also excluded. Actively exercising subjects were also excluded. A group of subjects without cirrhosis comparable for age and sex distribution and meeting the same eligibility criteria were enrolled as controls. The only comorbidities allowed for the control group were hypertension, prior stroke (more than 1 year before the enrolment) or non-obstructive peripheral vascular disease and mild chronic obstructive pulmonary disease. Among controls, a small group of sarcopenic subjects was also included to compare sarcopenic cirrhotic patients with sarcopenic controls.
The two participant groups received a nutritional evaluation, including a 7-day food frequency questionnaire, standard anthropometry and body composition analysis by whole-body dual X-ray absorptiometry (DXA). The diagnosis of sarcopenia was based on the presence of low muscle mass and strength, according to the criteria released by the Foundation for the National Institutes of Health (FNIH) Sarcopenia Project, 7 as follows: (a) appendicular lean mass (ALM) to body mass index (BMI) ratio (ALM BMI ) < 0.789 and < 0.512 in men and women; or (b) crude ALM < 19.75 kg in men and < 15.02 kg in women when the ALM BMI criterion was not met; and (c) handgrip strength < 26 kg for men or < 16 kg for women.
Faecal and blood samples were also collected for the analysis of the gut microbiota composition and metabolomics profile, as well as for the quantification of circulating cytokines/chemokines, markers of intestinal inflammation (calprotectin) and permeability (zonulin-1 [ZO1], lipopolysaccharide [LPS]). In particular, the panel of inflammatory cytokines measured in the present study was designed based on their involvement in pathways and processes relevant to cirrhosis and sarcopenia pathophysiology. A detailed description of laboratory techniques and procedures is provided in the Supplementary

Methods.
The study was approved by the Ethics Committee of the Fondazione Policlinico Universitario Agostino Gemelli IRCCS (protocol ID 741). All procedures were conducted in accordance with the principles laid down in the Declaration of Helsinki. A written informed consent was obtained from all participants prior to enrolment.

| Statistical analysis
Descriptive statistics were run on all data. Differences in demographic, anthropometric, clinical, functional characteristics, inflammatory and metabolic markers, and gut permeability/ inflammation markers between cases and controls were assessed The gut microbiota alpha diversity in experimental groups was evaluated by the Chao1 index, which was calculated on raw data. Afterwards, data were processed to remove taxa not seen more than three times in at least 20% of samples and normalized by applying regularized logarithm transformation (rlog). 8 Principal coordinates analysis (PCoA) was performed on weighted UniFrac distance, using permutational multivariate analysis of variance on the distance matrix to unveil differences between cirrhotic patients and controls, as well as among subgroups of sarcopenic participants.
The analysis of gut microbial differential abundance at the phylum, family and genus level was carried out using a negative binomial distribution on raw counts normalized by 'size factors', taking into account sequencing depth between samples. 8 This method was chosen based on its good performance in experiments involving relatively small samples. 9 Differences in bacterial abundance were expressed as log2 fold change (log2FC); a log2FC higher or lower than ± 1.5 with a P-value < 0.05 adjusted for multiple comparisons (false discovery rate [FDR] control according to Benjamini-Hochberg method) was considered statistically significant.
Gut microbiota metabolomics analysis was conducted by and METLIN databases. The library specific to Homo sapiens was chosen as reference. Global test algorithm was used for pathway analysis, while out-degree centrality was used for topological analysis. All p-values were adjusted for multiple testing using the FDR control according to the Benjamini-Hochberg method.
Finally, to obtain a multiomic picture of sarcopenia in cirrhosis, a correlation network based on Spearman's coefficients was built up, including gut bacteria, metabolites, inflammatory parameters and gut barrier markers found to be differentially represented in sarcopenic cirrhotic patients. The Girvan-Newman algorithm 10 was used to further detect cohesive subgroups of variables in the network, in order to unveil additional information on variables interconnection.
All of the analyses were performed using the R statistics program (version 3.6.2), Metaboanalyst (version 4.0) and Cytoscape (version 3.7.2).  Table 1. No differences were observed among groups for BMI or waist circumference. The weekly consumption of red meat, non-red meat, milk, fish, eggs, cured meats, cereals and bakery products, legumes, vegetables and fruits were similar between groups. Controls reported a higher intake of dairy products (cirrhotic patients vs controls P =.04; sarcopenic cirrhotic patients vs sarcopenic or nonsarcopenic controls P <.05).
We also investigated whether the gut microbiota was different between sarcopenic cirrhotic patients and sarcopenic controls.

| Intestinal permeability and systemic inflammation
We explored the integrity of the intestinal barrier using indirect markers of permeability/inflammation. As expected, serum levels of ZO1 (P =.0003) and LPS (P <.0001), and faecal concentration of calprotectin (P =.04) were higher in cirrhotic patients than in controls. However, no difference in any of these markers was observed between sarcopenic and nonsarcopenic cirrhotic patients (Table 2).

| Characterization of the gut microbiota metabolomics of cirrhotic patients vs controls
To explore the relationship between the gut microbiota metabolomic profile and sarcopenia, we conducted untargeted faecal metabolomics analyses.
These metabolites were functionally involved in several metabolic pathways, including aminoacyl-tRNA biosynthesis, valine, leucine and isoleucine biosynthesis and degradation, pantothenate and CoA biosynthesis, pentose and glucuronate interconversions, and cysteine and methionine metabolism ( Figure 3C and Table S3b).
Among all these pathways, cysteine and methionine metabolism showed the highest capacity in discriminating between cirrhotic patients and controls (P =.0002, impact 0.105) ( Figure 3D, Table   S3c).

| Characterization of the gut microbiota metabolomics of sarcopenic vs nonsarcopenic cirrhotic patients and controls
The gut microbiota metabolomic profile of cirrhotic patients with and without sarcopenia was clearly different at both univariate analysis (Table S4a) Figure S2A and B). The metabolic pathways differentiating sarcopenic cirrhotic patients from sarcopenic controls were glycolysis/gluconeogenesis and butanoate metabolism (P =.03, impact 0.133) (Table S5 and Figure S2C and D).
*Overall p-value; statistically significant comparisons (P <.05) between the subgroup a, b, c and d are also specified.

| Integrated approach to sarcopenia in cirrhotic patients
We finally combined data of differentially represented gut microbial components, intestinal barrier and inflammatory parameters, and gut microbiota metabolomics to explore their interconnection and networks correlated with sarcopenia in the group of cirrhotic patients.
Overall, 269 statistically significant correlations (Spearman's r < −0.250 or > 0.250 and adjusted P-value < 0.05) were observed (Table S6). The Girvan-Newman algorithm also allowed to identify three clusters of aggregation, the main nodes of which are displayed

| D ISCUSS I ON
The gut microbiota is one of the main actors in the development of metabolic disorders. With this study, we have shown that gut dysbiosis is associated with sarcopenia in patients with cirrhosis. better physical functioning. [13][14][15] The decrease in Prevotella is a renowned marker of frailty, [16][17][18] while its increase is associated with physical functioning and lean mass in older adults 19 and in young professional athletes. 20 Conversely, the genus Eggerthella, which was increased in sarcopenic cirrhotic patients, is overabundant in physical frailty, a condition that often overlaps with sarcopenia. 21,22 Conversely, only differences related to advanced liver disease, such as a lower abundance of Ruminococcaceae/Ruminococcus,

Veillonellaceae and Rikenellaceae, and an increase in Klebsiella and
Streptococcus, were observed by comparing sarcopenic cirrhotic patients with sarcopenic controls. 23,24 Therefore, gut microbiota modifications associated with sarcopenia do not overlap with alterations distinctively caused by liver disease, thus possibly producing additional, unfavourable, effects.
Interestingly, no difference was determined in the degree of alteration of intestinal barrier integrity or local inflammation between cirrhotic patients with and without sarcopenia. This observation suggests that the gut microbiota composition may be a more relevant factor for the development of inflammatory and metabolic changes associated with sarcopenia in cirrhosis than gut barrier impairment.
Sarcopenic cirrhotic patients also showed a pro-inflammatory cytokine profile which may further contribute to muscle atrophy.
Indeed, sarcopenia is associated with persistent, low-grade inflammation, 25 which is shared also by cirrhosis, ageing, cancer and other diseases. In our sample, the inflammatory effects of dysbiosis associated with liver disease, and the lack of specific microbial components with anti-inflammatory properties such as Akkermansia, 26 may favour the over-production of several cytokines/chemokines expressed in cirrhotic patients, most of which have already been associated with sarcopenia. 27,28 Furthermore, we found higher serum concentrations of FGF21 in sarcopenic cirrhotic patients. This finding might reflect a stressinduced increased secretion of FGF21 by the liver and/or the muscle in the setting of perturbed inter-organ crosstalk. 29 Notably, FGF21 has been shown to induce muscle atrophy via inhibition of protein synthesis and stimulation of autophagy. 29,30 Another interesting finding of our study was the strict clustering of the gut microbiota and inflammatory features associated Whether TMAO can also contribute to the pathogenesis of sarcopenia in cirrhotic patients through the promotion of inflammation warrants investigation.
Pathways including metabolites involved in amino acid and protein metabolism, in particular glutamine, glutamate, alanine, aspartate, arginine, aminoacyl-tRNA biosynthesis and branchedchain amino acids (BCAAs) were also up-regulated in cirrhotic patients with sarcopenia. Liver impairment causes a shift from carbohydrates to amino acids as a source of energy, which contributes to muscle wasting. 39 In healthy people, half of the ammonia produced from this process is metabolized by the liver in the urea cycle, while the remaining ammonia is detoxified by the skeletal muscle. 40 Conversely, in patients with cirrhosis glutamine synthesis from glutamate in the muscle becomes the most important detoxification route. 40 This process involves BCAAs, such as valine, leucine and isoleucine, which are also relevant to muscle homeostasis. 41 While humans have no biosynthetic pathways for BCAAs, the gut microbiota can synthesize them. Hence, our results confirm the enrichment of the gut microbiota in bacteria involved in BCAA metabolism in cirrhotic patients, which was especially evident in those with sarcopenia. 42 Furthermore, our data indicate that, in patients with mild liver dysfunction, this 'gut microbial buffer' may support the gut-liver-muscle axis, as none of our patients presented high ammonia serum levels or clinical signs of hepatic encephalopathy.
Finally, we found that the gut microbiota of sarcopenic cirrhotic patients may contribute to the generation of intermediates of glycolysis/gluconeogenesis and pentose and glucuronate interconversions pathways, such as xylose and arabinose, as well as to the production of metabolites with antioxidant properties (ie arginine, a precursor of nitric oxide, from glutamate and selenocompounds from methionine). These metabolites may further support muscle homeostasis and counteract sarcopenia in cirrhotic patients. [43][44][45][46][47] The present study has several strengths. First, we obtained a landscape of sarcopenia in patients with cirrhosis. We focused our analysis on patients with mild liver dysfunction, to capture the early metabolic and compositional changes of the gut microbiota that could contribute to the development of sarcopenia, avoiding the influence of other factors (eg hyperammonemia, malnutrition, inactivity) usually associated with a more advanced liver impairment. This information may be crucial in clinical practice, as it can help identify patients who may benefit from personalized treatments (eg gut microbiota modulation, anti-inflammatory treatments, nutritional support) to halt the progression of muscle wasting. The prevalence of cirrhosis in older adults has increased, and this is also confirmed by the median age of our patient population. 48 For this reason, we matched patients and controls by Klebsiella, installs a metabolic and pro-inflammatory network involving ethanol, TMA, myokines such as FGF21, cytokines and chemokines, which might ultimately contribute to muscle wasting. At the same time, the peculiar gut microbial composition in cirrhotic patients may act as a metabolic buffer, to limit muscle decay. Whether the dysfunction of the gut-liver-muscle axis is amenable for corrective interventions needs to be established to improve patients' outcomes.

| DATA AVAIL AB ITIT Y S TATEMENT
Data are available upon reasonable request and with permission of all the Authors.

ACK N OWLED G M ENT
None.