Colon cancer cachexia remodels gut microbiota and metabolite profiles in a murine model.

Introduction
Cachexia is a multifactorial syndrome characterized by involuntary weight loss, muscle atrophy, and adipose tissue depletion, often accompanied by anorexia and systemic inflammation (1). This syndrome occurs in up to 80% of patients with advanced cancer and is directly responsible for approximately 20% of cancer-related deaths (2). Patients with gastrointestinal malignancies, lung cancer, and hematological neoplasms are particularly susceptible, and its presence is strongly associated with poor prognosis and reduced tolerance to anticancer therapies. Despite its high clinical relevance, therapeutic strategies remain limited, as nutritional supplementation alone is generally ineffective in reversing cachexia progression (3,4).
Accumulating evidence suggests that the gut microbiota exerts profound effects on host physiology by modulating nutrient metabolism, immune responses, and systemic energy homeostasis (5,6). Dysbiosis of gut microbial communities has been linked to metabolic disorders, inflammation, and impaired intestinal barrier function, which are key features of cachexia (7,8). Furthermore, microbial metabolites, including short-chain fatty acids, bile acids, and amino acid derivatives, act as essential mediators of host-microbiota interactions and influence skeletal muscle maintenance and metabolic regulation (7,9). A recent study in a murine model demonstrated that cancer cachexia altered gut microbiota composition, characterized by increased Bifidobacterium and Romboutsia and decreased Streptococcus, along with reduced acetate and butyrate levels (10). Notably, a specific member of the Ruminococcaceae family was identified as a key contributor to the observed butyrate reduction. Further analysis revealed that a two-fold acceleration in intestinal transit serves as a central modulator, reshaping microbial community structure and function, collectively promoting fecal loss of proteins and amino acids (11). 16S ribosomal RNA (rRNA) analysis demonstrated an overrepresentation of Escherichia-Shigella and Hungatella in the cachexia patients with advanced non-small cell lung cancer, whereas the non-cachexia patients exhibited a higher abundance of Anaerostipes, Blautia, and Eubacterium (12). Although studies using 16S rRNA sequencing have reported shifts in microbial composition in cachexia, the functional consequences of these changes and their impact on metabolite production remain poorly understood.
To address this knowledge gap, we employed an integrative approach combining metagenomic sequencing and non-targeted metabolomics to characterize the gut microbiota and metabolic alterations in a murine colon cancer cachexia model. By comparing cachexia-inducing colon carcinoma 26 (C26) with non-cachexia-inducing C26 [colon carcinoma tumor 26 cells (CT26)] tumors, alongside healthy controls, we aimed to distinguish cachexia-specific effects from those related to tumor burden alone. This study identifies significant taxonomic and functional alterations in the gut microbiota and corresponding metabolic disturbances, providing new insights into the pathophysiology of cancer cachexia and suggesting potential microbial and metabolic targets for therapeutic intervention. We present this article in accordance with the ARRIVE reporting checklist (available at https://jgo.amegroups.com/article/view/10.21037/jgo-2025-720/rc).

Methods

Cell culture
Cachexia-inducing colon carcinoma 26 (C26, from Cell Line Service, Eppelheim, Germany) and non-cachexia-inducing C26 colon carcinoma cells (CT26, from Shanghai Cell Bank, Chinese Academy of Sciences, Shanghai, China) were maintained in 1640-RPMI medium supplemented with 10% fetal bovine serum (Gibco, Grand Island, NY, USA), 100 µg/mL streptomycin and 100 IU/mL penicillin (Beyotime Biotechnology, China) at 37 ℃ with 5% CO2.

Colon cancer cachexia and non-cachexia colon cancer model
Animal experiments were performed under a project license (No. XM992022-0023) granted by the Wenzhou Medical University Ethics Committee, in compliance with Chinese national guidelines for the care and use of animals. Male BALB/c mice (5 weeks old, Hangzhou Medical College, Hangzhou, China) were kept in specific pathogen-free conditions and housed in individually ventilated cages with a 12 h light/dark cycle and fed an irradiated chow diet. After one week of acclimatization, mice were randomly assigned to control group (CTRL group), C26 group, CT26 group based on their body weight and were subcutaneously injected in the upper flank with a saline solution, C26 (5×105), or CT26 cells (2×106 cells in 0.1 mL saline). Food intake, body weight and tumor size were recorded. The size of the tumor was measured using a caliper, and the tumor volume was calculated using the formula: 0.5 × (length × width2). The mice’s body weight and food intake were measured every day. Seventeen days after cancer cells injection, mice were fasted for 6 h, and tissue samples were harvested following anesthesia. Tissues were weighed and frozen in liquid nitrogen. All samples were stored at −80 ℃ until further analyses.

Metagenomic sequencing of gut microbiota
Genomic DNA was extracted from the cecal content of mice using the E. Z. N. A. ® Soil DNA Kit from Omega Bio-tech, a US-based company. Following extraction, the concentration and purity of DNA were assessed, and DNA integrity was analyzed through 1% agarose gel electrophoresis. DNA extract was fragmented to an average size of about 350 bp using Covaris M220 (Gene Company Limited, Shanghai, China) for paired-end library construction. Paired-end library was constructed using NEXTFLEX Rapid DNA-Seq (Bioo Scientific, Austin, TX, USA). Paired-end sequencing was performed on Illumina NovaSeq™ X Plus (Illumina Inc., San Diego, CA, USA) at Majorbio Bio-Pharm Technology Co., Ltd. (Shanghai, China) using NovaSeq X Series 25B Reagent Kit according to the manufacturer’s instructions (www.illumina.com).

Non-targeted metabolome analysis
Liquid chromatography-mass spectrometry (LC-MS)/mass spectrometry (MS) analysis was conducted on the cecal content of mice utilizing a SCIEX UPLC-Triple TOF 5600 system equipped with an ACQUITY HSS T3 column (100 mm × 2.1 mm i.d., 1.8 µm; Waters, USA) at Majorbio Bio-Pharm Technology Co. Ltd. (Shanghai, China). The pretreatment of LC/MS raw data was performed by Progenesis QI (Waters Corporation, Milford, USA) software. The MS and MS/MS mass spectrometry information was integrated with the metabolic public databases HMDB (http://www.hmdb.ca/) and Metlin (https://metlin.scripps.edu/), the self-compiled Majorbio Database (MJDB) of Majorbio Biotechnology Co., Ltd. (Shanghai, China).

Statistical analysis
The experimental data were plotted using GraphPad Prism 9.0 software and expressed as mean ± standard deviation (SD). The statistical significance of differences among groups was performed using one-way analysis of variance (ANOVA), followed by Tukey multiple comparisons test. P<0.05 was defined as statistically significant. The analysis of Metagenomic sequencing and non-targeted metabolome was processed by the Majorbio Cloud Platform (https://cloud.majorbio.com/) (Shanghai Majorbio Bio-pharm Technology, Shanghai, China).

Results

Establishment of the cancer cachexia model and non-cachexia cancer model
We established cancer cachexia and non-cachexia cancer models in mice by transplanting cachexia-inducing colon carcinoma 26 cells and non-cachexia-inducing C26 colon carcinoma cells, respectively, following the sequence shown in Figure 1A. To confirm the success in establishing the cancer cachexia mouse model, the physiological characteristics, such as food intake, body weight, tumor volume and several other variables, are shown in Figure 1B-1H. Cancer cachexia mice (C26 group) were compared with those of healthy control mice (CTRL group) and non-cachexia cancer mice (CT26 group). As shown in Figure 1C, the food intake of the C26 group is lower than that of the CTRL group and the CT26 group. During the experimental period, the body weight of the CTRL group and the CT26 group gradually increased while the body weight of the C26 group did not increase and even decreased strongly from Day 11 after tumor implantation (Figure 1E). Furthermore, the C26 group exhibited a significant decrease in tumor-free body (Figure 1F), skeletal muscle (gastrocnemius and quadriceps femoris) weight (Figure 1B,1G) and fat (inguinal white adipose tissue and epididymal white adipose tissue) weight (Figure 1B,1H). The tumor growth curve of the C26 group and the CT26 group is shown in Figure 1D. The significant decrease in body weight and the considerable loss in weight of muscle tissue and fat tissue suggested the successful establishment of the cancer cachexia model in the present study.

Gut microbiota dysbiosis in colon cancer cachexia mice
Multiple studies have established correlations between gut microbiota dysbiosis and cancer cachexia, predominantly using 16S rRNA gene sequencing. We employed metagenomic sequencing to investigate associations between cancer cachexia and the structure of gut microbiota. In comparison to the CTRL group, CT26 group and C26 group, the Shannon index, Ace index, Sobs index, and Chao index in the C26 group were significantly decreased (P<0.01) (Figure 2A-2D), and the Simpson index was increased (P<0.01) (Figure 2E), indicating that the species diversity and richness in the C26 group were significantly lower than those in the CTRL group and the CT26 group. The principal coordinate analysis (PCoA) revealed significantly different community distributions between the C26 group and the other two groups (R=0.4898, P=0.001). The CTRL and CT26 group exhibited close clustering, indicating similar microbial community structure (Figure 2F).
To study the variation of the cancer cachexia mice microbiome, we conducted microbiological composition and difference analysis on the sequencing data of samples in the CTRL, CT26, and C26 groups at the levels of kingdom, phylum, genus, and species. The detected taxa compositions included six microbial kingdoms, including bacteria, Heunggongvirae, viruses, Loebvirae, archaea, and Eukaryota (Figure S1A-S1C). Over 90% of the sequences in all samples were aligned to bacteria, while Heunggongvirae, viruses, Loebvirae, archaea, and Eukaryota accounted for only a small portion. It was notable that eukaryota was found in the C26 group, while it was not found in the CT26 or the CTRL groups; loebvirae was found in the CT26 and CTRL groups, but not in the C26 group. Figure S1D shows the distribution of bacteria, Heunggongvirae, viruses, Loebvirae, archaea, and Eukaryota in all the samples. All kingdoms were classified into 187 phyla and 2,734 genera. At the phylum level, the C26 group was dominated by three phyla: Bacteroidetes (50%), Bacillota (20%) and Pseudomonadota (20%); the CT26 group was dominated by Bacillota (55%) and Bacteroidetes (35%); the CTRL group was also dominated by Bacillota (59%) and Bacteroidetes (31%) (Figure S2A). To identify the microbial taxa that exhibited the greatest dissimilarities in the C26 model, we conducted comparative two-group analyses and employed linear discriminant analysis effect size (LEfSe) analyses [the linear discriminant analysis (LDA) score was set at 4.0]. The differential phyla and LEfSe analysis at the phylum level are shown in Figure S2B-S2F. In our study, unclassified species accounted for a high proportion. The top 10 species at the species level are shown in Figure S3.
The genus level was mainly analyzed and the major genera in the CTRL, CT26 and C26 group is showed in Figure 3A. Compared with the CTRL group, the relative abundance of Bacteroides, Phocaeicola, Escherichia, Helicobacter, and Enterobacter was significantly increased in the C26 group (P=0.009), while Alistipes, Acetobacter, Eubacterium, Roseburia, and Clostridium were significantly decreased in the C26 group (P<0.001) (Figure 3B). Notably, Bacteroides, Phocaeicola, Escherichia, Enterobacter, Helicobacter, and Proteus were the prominent genera in the C26 group (Figure 3C). When comparing the C26 group to the CT26 group, Bacteroides, Phocaeicola, Escherichia, Helicobacter, and Enterobacter were again significantly increased (P<0.001), whereas Alistipes, Acetobacter, Eubacterium, Hungatella, and Roseburia were significantly decreased (P<0.001) (Figure 3D). The same set of genera—Bacteroides, Phocaeicola, Escherichia, Enterobacter, Helicobacter, and Proteus were identified as predominant in the C26 group in this comparison (Figure 3E). In the comparison between the CT26 and CTRL groups, only Acutalibacter was significantly increased in the CT26 group (P=0.045), while Eubacterium and Clostridium were significantly decreased (P=0.01) (Figure 3F).
To sum up, our study suggested that changes in Bacteroides, Phocaeicola, Escherichia, Enterobacter, Helicobacter, and Proteus may have potential for the development of cachexia predictive biomarkers.

Functional and metabolic alterations in colon cancer cachexia mice
The Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis revealed that metabolism, environmental information processing, and genetic information processing occupied the top 3 highest functions in each group (Figure 4A). The functional structure of the microbiota in each group was evaluated at Level 3 by PCoA (Figure 4B), the overall functional structure of C26 group was clearly separated from that of CTRL group and CT26 group, and the functional structure of CTRL group and CT26 group tended to be the same (R=0.3708, P=0.001). In the Level 3 functional comparison, the top 10 enriched functions between the C26 group and CTRL group (Figure 4C) included metabolic pathways, microbial metabolism in diverse environments, biosynthesis of cofactors, biosynthesis of amino acids, ABC transporters, biosynthesis of nucleotide sugars, purine metabolism, quorum sensing, ribosome, and nucleotide metabolism, and all of which were significantly different from each other (P<0.05). When comparing the C26 and CT26 groups, the top 10 functions: metabolic pathways, biosynthesis of cofactors, biosynthesis of amino acids, biosynthesis of nucleotide sugars, purine metabolism, quorum sensing, nucleotide metabolism, pyrimidine metabolism, starch and sucrose metabolism, and homologous recombination showed significant differences (P<0.05) (Figure 4D). In addition, the comparison between the CT26 and CTRL group revealed significant differences in some pathways, such as metabolic pathways, biosynthesis of secondary metabolites, ribosome, cysteine and methionine metabolism, methane metabolism, cell cycle-caulobacter, fructose and mannose metabolism, porphyrin metabolism, base excision repair, and oxidative phosphorylation (P<0.05) (Figure 4E). When the LDA score was set at 3.0, three functions were impacted in C26 group in contrast to CTRL group, including metabolic pathways, lysosomes, and microbial metabolism in diverse environments (Figure 4F). Compared with the CT26 group, the C26 group exhibited alterations most notably in metabolic pathways (Figure 4G). When comparing CT26 group with CTRL group, there were no functional alterations between the two groups. The above analysis results suggested that the enriched functions in the C26 group were mainly metabolic pathways.
To further explore the metabolic changes in the cancer cachexia mice, non-targeted LC-MS/MS metabolomics was performed on mouse cecal content from each group. The overall metabolic differences among the CTRL group, CT26 group, and C26 group were assessed by PCA analysis. In the positive-ion (POS), the metabolic profiles of the C26 group were clearly distinguished from those of the CTRL group and CT26 group, whereas the metabolic profiles of CTRL group and CT26 group showed a high degree of similarity (Figure 5A). A similar trend was observed in the negative-ion (NEG), where the C26 group exhibited a distinct metabolic pattern compared to the other two groups, which remained largely indistinguishable from each other (Figure 5B). These results indicated that endogenous metabolites were altered in the C26 mice.
The common and unique metabolites among the three groups were analyzed. The results showed that there were 1,422 common metabolites among the three groups (Figure S4A). An Orthogonal Projections to Latent Structures-Discriminant Analysis (OPLS-DA) model was constructed. The OPLS-DA score plot and permutation testing (Figure S4B-S4G) showed significant differences between the C26 group and the CTRL group [R2X(cum) =0.666, R2Y(cum) =0.998, Q2(cum) =0.969], between the C26 group and the CT26 group [R2X(cum) =0.641, R2Y(cum) =0.996, Q2(cum) =0.958], and between the CT26 group and the CTRL group [R2X(cum) =0.5, R2 Y(cum) =0.998, Q2(cum) =0.669]. This indicated that the OPLS-DA model was indeed reliable and predictive. Then, the hierarchical clustering method was used to identify the differential metabolites with variable importance in projection (VIP) scores >1 and P<0.05. A total of 636 different abundance metabolites were identified between the C26 group and the CTRL group. Compared with the CTRL group, 216 metabolites were upregulated and 420 were downregulated in the C26 group (Figure S5A,S5B). Between the C26 group and the CT26 group, 652 different abundance metabolites were identified. Compared with the CT26 group, 222 metabolites were upregulated and 430 were downregulated in the C26 group (Figure S5A,S5C). Between the CT26 group and the CTRL group, 155 different abundance metabolites were identified. Compared with the CTRL group, 65 metabolites were upregulated and 90 metabolites were downregulated (Figure S5A,S5D) [screening criteria: t-test P<0.05, VIP >1, fold change (FC) >1].
Figure 5 shows the diversity of the heat map and the difference bar chart (VIP >1, P <0.05). Compared with the CTRL group (Figure 5C), the metabolites in the C26 group, such as didanosine, citpressine ii, spirotaccagenin, etc., were significantly increased (P<0.001), while Sm(D19:1/Txb2), Dg(20:0/Pge2/0:0), dodecyl gallate, etc., were significantly decreased (P<0.001). The KEGG pathway enrichment analysis (Figure 5D) revealed that the C26 group showed significant enrichment in ascorbate and aldarate metabolism, nucleotide metabolism, and steroid hormone biosynthesis (P<0.05). Compared with the CT26 group (Figure 5E), the levels of various metabolites in the C26 group showed significant changes. Glu His Arg, hydratopyrrhoxanthinol, spirotaccagenin, etc. were significantly increased (P<0.001), while Sm(D19:1/Txb2), Dg(20:0/Pge2/0:0), piceatannol, etc., were significantly decreased (P<0.001). The KEGG pathway enrichment analysis (Figure 5F) revealed that the C26 group showed significant enrichment in ascorbate and aldarate metabolism, nucleotide metabolism, and hypoxia-inducible factor 1 (HIF-1) signaling pathways (P<0.05). For the HIF-1 signaling pathway, this enrichment was driven by metabolites detected in the cecal contents that are annotated within this pathway, including ascorbic acid, pyruvic acid, and Dg[18:1(11Z)/14:1(9Z)/0:0]. It should be noted that this enrichment reflects metabolite-driven pathway annotation based on KEGG analysis and does not represent direct measurement of HIF-1α activity or secretion. Compared with the CTRL group (Figure 5G), the levels of Ps[Lte4/22:6(4Z,7Z,10Z,13Z,16Z,19Z)], didanosine, sorbitan palmitate, etc., in the CT26 group were significantly increased (P<0.05), while 1-nitro-7-glutathionyl-8-hydroxy-7,8-dihydronaphthalene, 10,20-dihydroxyeicosanoic acid, and phenylalanylphenylalanine were significantly decreased (P<0.05). The KEGG pathway enrichment analysis (Figure 5H) showed that the pathways related to insulin resistance, regulation of adipocyte lipolysis, fat digestion and absorption, etc., in the CT26 group were significantly enriched (P<0.05). In addition, the concentrations of gut microbial metabolites, such as ursodeoxycholic acid (UDCA) and hyodeoxycholic acid (HDCA), were significantly decreased in the C26 group (P<0.01). Bacterial amino acid metabolites (bAAms), such as kynurenic acid, were significantly decreased in the C26 group (P<0.01). Branched chain amino acids (BCAAs), such as valine, leucine, and isoleucine, were also significantly decreased in the C26 group (P<0.01). Other amino acids, such as glutamine, histidine, aspartic acid, and threonine were significantly decreased in the C26 group (P<0.01); arginine, glutamic acid, tyrosine, and methionine were significantly increased in the C26 group (P<0.01).
To sum up, the results of non-targeted metabolome analysis revealed distinct metabolic profiles between the three groups. The changed metabolites in the C26 group were mainly enriched in ascorbate and aldehyde metabolism, nucleotide metabolism, steroid hormone biosynthesis, and HIF-1 signaling pathways. Notably, gut microbiota-derived metabolites, including secondary bile acids, bAAms, and BCAAs, were significantly decreased in the C26 group.

Integration of microbiota and metabolite profiles
Using procrustes analysis to conduct an in-depth analysis of the correlation between the intestinal microbiome and the overall metabolites, it was revealed whether the changes in metabolites and microorganisms in each sample were consistent. The analysis results (Figure 6A) showed that the intestinal microbiome and metabolite profiles had a strong synergy (M2 =0.31, P<0.001). To investigate the association between the gut microbiota and its metabolic profile in C26 mice, Spearman correlation analyses of differential metabolites obtained by metabolomics and differential species obtained by metagenomic sequencing analyses were performed to explore how the microbial community affects metabolic functions.
The results showed significant differences at the genus level between the C26 group and the CTRL group (the top 30 genera and the top 30 metabolites at the classification level). The results indicated that Phocaeicola, Escherichia, Enterobacter, and Proteus were positively correlated with 3-ketosphinganine, lysope(15:0/0:0), 15-deoxy-delta-12,14-Pgj2-D4, 3-alpha-hydroxy-6-ketocholanic acid, 3-oxo-4,6-choladienoic acid and negatively correlated with butaprost, 1-(4-O-beta-D-glucopyranosyl-3-methoxyphenyl)-3,5-dihydroxydecane, UDCA, oxypurinol, 13(S)-hpode, 2’,3-dihydroxy-4,4’,6’-trimethoxychalcone, etc. Helicobacter was positively correlated with 3-alpha-hydroxy-6-ketocholanic acid, 15-deoxy-delta-12,14-Pgj2-D4 and negatively correlated with isoferulic acid, cellobiose, 5-hydroxyindole-3-acetic acid, UDCA, etc. Bacteroides was positively correlated with lysope(15:0/0:0), 3-ketosphinganine, 15-deoxy-delta-12,14-Pgj2-D4, Ile Thr Tyr Asp and negatively correlated with oxypurinol (Figure 6B). Between the C26 group and the CT26 group, Helicobacter, Phocaeicola, Escherichia, Enterobacter, and Proteus were positively correlated with 3-oxo-4,6-choladienoic acid, Ile Thr Tyr Asp, 3-ketosphinganine, 15-deoxy-delta-12,14-Pgj2-D4 and negatively correlated with cellobiose, isoferulic acid, UDCA, Lys-Lys-Oh, octadecanedioic acid, etc. Bacteroides was positively correlated with lysope(15:0/0:0), 15-deoxy-delta-12,14-Pgj2-D4, 3-oxo-4,6-choladienoic acid, methyl linoleate and negatively correlated with 7-methyl-cholic acid (Figure 6C). The clustering effect in the heatmap was poor between the CT26 group and the CTRL group (Figure 6D).
Above all, correlation analysis at the genus level revealed significant microbial-metabolite associations. Phocaeicola, Escherichia, and other genera showed distinct positive and negative correlations with specific metabolites, such as bile acids and sphingolipids in the C26 group, compared to CT26 and CTRL groups. In contrast, clustering between the CT26 and CTRL groups was poor, indicating less distinct associations.

Discussion
In this study, we demonstrated that cancer cachexia profoundly altered the structure and function of the gut microbiota, accompanied by distinct shifts in cecal content metabolites. Using a C26 cachexia model and a CT26 non-cachexia model, we showed that cachexia was associated with decreased microbial diversity, enrichment of Bacteroides, Phocaeicola, Escherichia, Enterobacter, Helicobacter, and Proteus, and depletion of butyrate- and bile acid-producing taxa such as Alistipes, Eubacterium, Roseburia, Clostridium, and Hungatella. Functionally, these microbial changes were associated with enrichment of metabolic pathways, while metabolomics revealed depletion of BCAAs, secondary bile acids (UDCA, HDCA), and bAAms, together with increased arginine and other amino acids.
These findings are consistent with previous reports describing gut microbiota dysbiosis in cancer cachexia, including reduced Lactobacillus and Clostridiales and increased Enterobacteriaceae (13-15). Importantly, our integrative analysis further identified significant correlations between microbial taxa and metabolites, suggesting a potential association of gut dysbiosis in metabolic reprogramming during cachexia. For example, the depletion of UDCA strongly correlated with reduced abundances of Alistipes, Eubacterium, and Roseburia, genera known to modulate bile acid metabolism. Secondary bile acids have been implicated in maintaining intestinal and metabolic homeostasis (16,17), and their loss may contribute to systemic inflammation and muscle wasting. Similarly, the reduction of BCAAs, essential for protein synthesis and skeletal muscle maintenance (18), highlights the metabolic consequences of altered microbial function.
Notably, our study provides clarifying evidence by separating cachexia-specific changes from those associated with tumor presence alone. The CT26 non-cachexia tumor group displayed microbial and metabolic profiles similar to healthy controls, confirming that the observed alterations are not merely consequences of tumor growth but are specific to cachexia. This strengthens the evidence that host-microbiota-metabolite interactions play a central role in driving the syndrome.
The current study also highlights potential therapeutic opportunities. Strategies aimed at restoring beneficial microbial taxa or supplementing key metabolites may help mitigate cachexia progression. Probiotic or synbiotic interventions have shown promise in experimental models (14), and targeting bile acid metabolism has been proposed as a strategy to improve metabolic outcomes in cachexia (16,19,20). Our data further suggest that BCAAs supplementation, in combination with microbiota-directed therapies, could be explored as an adjuvant treatment approach.
Several limitations should be acknowledged. First, this study was restricted to murine models, and validation in clinical samples will be essential to confirm translational relevance. The murine models employed here serve as a powerful discovery tool but represent a simplified system. Consequently, future research must prioritize the analysis of patient-derived samples to verify whether the key microbial and metabolic disturbances we observed are conserved and clinically actionable in human cancer cachexia. Second, although strong correlations between specific bacteria and metabolites were identified, causal relationships remain to be established. It should be noted that this inference is based on targeted correlation analyses following an assessment via procrustes analysis. This method is designed to assess global congruence at the dataset level and can not imply causality or identify specific drivers of the observed concordance between the microbiome and metabolome datasets. Therefore, future studies incorporating germ-free or microbiota-transplantation approaches will be required to determine whether specific microbial taxa directly drive metabolic dysregulation in cachexia.

Conclusions
This study demonstrated that cancer cachexia induced specific and profound alterations in the gut microbiota and metabolome, characterized by reduced microbial diversity, depletion of beneficial butyrate- and bile acid-producing bacteria, and distinct metabolic disturbances including loss of secondary bile acids and BCAAs. Through integrated multi-omics analysis, we established robust microbiota-metabolite correlations and provided evidence that these changes are specific to cachexia rather than tumor presence alone. These findings suggested the key role of host-microbiota-metabolite interactions in cachexia progression and offered new avenues for therapeutic intervention, such as microbial restoration and metabolite supplementation. Further validation in clinical cohorts and mechanistic studies is warranted to translate these insights into effective treatments for cancer cachexia.

Supplementary
The article’s supplementary files as