Background

The immunometabolic interface between gut microbiota and host immunity has emerged as a critical regulator of systemic malignancies. In diffuse large B-cell lymphoma (DLBCL), this crosstalk remains poorly defined, limiting early-response biomarker development. Using metagenomic, metabolomic, and immunomic analyses, we systematically profiled the microbiota–metabolism–cytokine axis in DLBCL patients and its evolution during immunochemotherapy, aiming to define microbiota-derived immunometabolic signatures associated with disease progression and therapy outcomes.Methods We prospectively enrolled 40 newly diagnosed, treatment-naïve DLBCL patients and 32 healthy controls, collecting paired fecal and serum samples before and after 4 cycles of R-CHOP or R-miniCHOP chemotherapy. Metagenomic sequencing (WGS), untargeted serum metabolomics (LC-MS/MS), and multiplex cytokine profiling were performed. Canonical correspondence analysis (CCA), random forest modeling, and Kaplan-Meier survival analysis were used to integrate cross-domain data and evaluate predictive performance.Results

  • Microbial Dysbiosis and Diversity Loss in DLBCL: DLBCL patients exhibited significantly reduced α-diversity (Shannon/Simpson indices, P<0.05) and altered β-diversity (PCoA; P<0.01) versus healthy controls. Progressive depletion of butyrate-producing genera (e.g., Lachnospiraceae, Roseburia, Faecalibacterium) correlated with disease stage. Fungal overgrowth (Candida, Tremellaceae) and expansion of Enterococcus defined late-stage microbial networks with enhanced cross-kingdom pathogenicity (R > 0.99). LEfSe identified 92 differentially abundant species (FDR<0.05), highlighting SCFA depletion and fungal dominance in advanced disease.

  • Metabolomic Perturbations Reflect Microbial Disruption: Significant alterations in lipid, amino acid, and organic acid metabolism were detected. Proinflammatory and immunosuppressive metabolites, including kynurenic acid, 2-arachidonoylglycerol (2-AG), and N,N-dimethyl-L-arginine, were enriched in stage III–IV. SCFA-linked metabolites (e.g., propionate, Cys–Cys) were depleted and positively associated with microbial diversity (r>0.4, FDR<0.01). Plant-derived compounds were less stage-specific.

  • Multi-Omics Integration Reveals a Core Immunometabolic Network: CCA explained 61.2% of the variance in microbiota-metabolite-cytokine relationships (P<0.01). Healthy controls clustered with SCFA-producing bacteria, IL-12p70, IL-23p19, and 2-AG, while DLBCL patients exhibited a shift toward TNF-α, IL-10, and neuroinflammatory metabolites (e.g., mannitol-1-phosphate). Butyrivibrio and Clostridium negatively correlated with MCP-1 and G-CSF (r=−0.37 to −0.44), suggesting suppression of myeloid and Th1 pathways.

  • Chemotherapy Amplifies Dysbiosis and Metabolic Imbalance: After 4 cycles of immunochemotherapy, beneficial genera (Bifidobacterium, Faecalibacterium) declined, while opportunists (Shigella, Escherichia) increased. Metabolomics revealed elevated phospholipid remodeling (e.g., PC/PE 38:4), oxidative stress markers (malonic acid ↑, allantoin ↓), and depletion of L-histidine and acetylcarnitine, contributing to immune dysregulation.

  • Microbiota-Metabolite Signatures Predict Treatment Response: Among 34 evaluable patients, 25 (62.5%) achieved CR. Linear discriminant analysis (FDR<0.05) identified 23 bacterial species discriminating CR from non-CR (NCR). CR was associated with Roseburia, Eubacterium, and Lachnospiraceae, linked to solasodine, 2-AG, and decahydrogambogic acid. NCR patients had increased Escherichia and Kluyvera, associated with inflammasome activity.

  • Risk Models Predict PFS: A microbial-metabolite classifier (13 species + 9 metabolites) achieved AUCs of 0.982 (training) and 0.961 (validation, n=106). Kaplan-Meier analysis showed significant PFS stratification by microbial (Log-rank P=0.02) and metabolite (Log-rank P=0.02) risk scores.

Conclusions This study delineates a dynamic immunometabolic network in DLBCL, shaped by disease stage and chemotherapy. Depletion of SCFA-producers, fungal overgrowth, and disrupted endocannabinoid metabolism define resistant states. Early recovery of these profiles aligns with remission and improved survival. Multi-omics signatures hold promise for microbiota-informed diagnosis, real-time monitoring, and targeted interventions in precision DLBCL care.

This content is only available as a PDF.
2025
Sign in via your Institution