Gut microbiota and metabolism associated with CAR-T therapy response in multiple myeloma

Gut microbiota and metabolism associated with CAR-T therapy response in multiple myeloma

A recent work published by Mireia Uribe-Herranz¹⁺², Aina Oliver-Caldés¹⁺³, Neus Martínez-Micaelo⁴, Marta Español-Rego², Maria Val-Casals¹⁺⁵, Roberto Martínez-Soler¹, Elisa Rubio-Garcia⁶⁺⁷⁺⁸⁺⁹, Valeria Brunello¹, Erik Z. Mihelic¹, Nela Klein-González¹⁺²⁺⁵, Daniel Benítez-Ribas², Núria Amigó⁴⁺¹⁰⁺¹¹, Andrea Vergara⁶⁺⁷⁺⁹⁺¹², Valentin Ortiz-Maldonado³, Luis Gerardo Rodríguez-Lobato¹⁺³, Julio Delgado³, Iñaki Ortiz de Landazuri¹⁺², Verónica González-Calle¹³, Valentín Cabañas¹⁴, Beatriz Martin-Antonio¹⁵, Lorena Pérez-Amill¹⁺⁵, Juan Luis Reguera-Ortega¹⁶, Paula Rodríguez-Otero¹⁷, Bruno Paiva¹⁷, Joaquín Martínez-López¹⁸, Maria-Victoria Mateos¹³, Mariona Pascal¹⁺², Álvaro Urbano-Ispizua¹⁺³, Europa Azucena González-Navarro², Carlos Fernández de Larrea¹⁺³, and Manel Juan¹⁺²⁺¹⁹⁺²⁰. has demonstrated that the gut microbiota and metabolite profiles influence the clinical response of patients with multiple myeloma treated with BCMA-directed CAR-T therapy (ARI0002h). Published in Blood Cancer Discovery (2025), the study integrates for the first time metagenomic, metabolomic, and immunophenotyping data to understand how the intestinal microbial environment affects the persistence and functionality of CAR-T cells.

Figure 1. Pre CAR-T-cell infusion gut microbiota taxonomic profiles by hospital of origin from the first cohort (n=28). (A) PCA of gut microbiota composition, with each dot representing a patient and colored by hospital. (B) Arrows represent the direction and magnitude of each variable’s influence on the separation of patients, with cos2 values indicating the significance of each variable’s contribution to the main component. (C) Reciprocal Simpson diversity index of the gut microbiome in patients from different clinical trial sites. Each dot represents an individual patient (D) Stacked bar plot showing the mean relative abundance of bacterial families in the gut microbiome of myeloma patients across different clinical trial sites. PCA: Principal component analysis

Study design and methodology

The study analyzed fecal and serum samples from 51 patients enrolled in the CARTBCMA-HCB-01 clinical trial, collected at different time points: before apheresis, before lymphodepletion, and after CAR-T infusion. The microbial composition was characterized by 16S rRNA sequencing, while the metabolomic profile was evaluated by nuclear magnetic resonance (NMR) and gas chromatography–mass spectrometry (GC–MS), allowing quantification of 35 fecal metabolites and serum short-chain fatty acids (SCFAs). In parallel, CAR-T cell persistence and clinical response were monitored at days 28, 100, and 180.

Main findings

The authors found that higher stool succinate levels before infusion correlated with a greater proportion of CD4+ central memory T cells and enhanced CAR-T cell persistence after treatment. In cell culture, succinate supplementation during ex vivo expansion increased the respiratory capacity and metabolic profile linked to longer CAR-T cell lifespan, results that were also confirmed in murine models fed with fructooligosaccharide (FOS)-enriched diets.
Additionally, integrative analyses revealed that certain bacterial families —such as Acidaminococcaceae, Barnesiellaceae, and Akkermansiaceae— were associated with complete responses, while Monoglobaceae and Erysipelotrichaceae were linked to suboptimal or absent responses.

Figure 2. Characterization of taxa and metabolites associated with CD8+ T cells in the apheresis product from the first cohort: (A) Tc1 cells, (B) Tc17 cells, (C) memory subsets, and in the final product (D) memory subsets. Significant associations are indicated as blue (negative) and red (positive). Correlations were then analyzed using Spearman’s rank correlation. Significance cut off p<0.05. Variables related to both microbiome and metabolome in feces and serum metabolome profiles (analyzed by NMR and GC-MS respectively) were included, and the results are graphically represented using volcano plots for features that reached statistical significance.

Multimodal predictive models

By combining microbial, metabolomic, and immunological data, the researchers developed multimodal models for predicting clinical response, integrating information from bacterial composition, metabolic profiles, and cellular markers collected throughout the treatment.
Using machine learning and multivariate analysis techniques, they evaluated the relative contribution of each biological variable to patient outcomes, building models capable of predicting complete response with high accuracy (AUC ≈ 0.9) at both 100 and 180 days after CAR-T cell infusion.

These predictive models not only make it possible to identify patients more likely to benefit from the therapy, but also provide a valuable tool to personalize treatment strategies and clinical monitoring, incorporating the influence of the microbiome and host metabolism into the overall efficacy of immunotherapy.

Implications

Taken together, these results reinforce the idea that the gut microbiota is a key modulator of cellular immunotherapy and highlight the potential of a multi-omics approach to understand the complex interactions between metabolism, the microbiome, and antitumor immunity.
Looking ahead, the challenge will be to translate this knowledge into clinical practice by integrating microbial and metabolomic characterization into patient selection and monitoring protocols for CAR-T therapies.
Moreover, the possibility of modulating the microbiome through nutritional, prebiotic, or probiotic strategies opens a new field of research aimed at enhancing the efficacy and durability of CAR-T therapies in hematologic cancers, moving toward truly personalized immunotherapy based on the patient’s own biology.

Figure 3. Comparison of gut microbiome profiles in MM patients from the first cohort at baseline (apheresis day) and 12 months post–CAR-T-cell infusion. (A) PCA of bacterial community composition at baseline (apheresis day) vs 12 months post CAR-T cell infusion microbiome in MM patients. (B) VIP scores from PLS-DA identifying the top 10 discriminating parameters from 16sRNA metagenomics. Scores plot of PLS-DA based on metagenomics of samples 12 months post-infusion (purple, n = 4) and at baseline (blue, n = 4) from paired patients. Colored dots illustrate individual samples. Axes are labeled with the first and second components and their respective percentages of explained variance. PCA: Principal component analysis, VIP: Variable importance in projection, PLS-DA: partial least square-discriminant analysis.

Read the paper here. 

Author Affiliations

  1. August Pi i Sunyer Biomedical Research Institute (IDIBAPS), Barcelona, Spain

  2. Department of Immunology, Centre de Diagnòstic Biomèdic (CDB), Hospital Clínic de Barcelona, Barcelona, Spain

  3. Department of Hematology, Hospital Clínic de Barcelona, University of Barcelona, Barcelona, Spain

  4. Biosfer Teslab, Reus, Spain

  5. Gyala Therapeutics S.L., Barcelona, Spain

  6. Department of Clinical Microbiology, CDB, Hospital Clínic de Barcelona, Barcelona, Spain

  7. Barcelona Institute for Global Health (ISGlobal), Barcelona, Spain

  8. Molecular Core Facility, CDB, Hospital Clínic de Barcelona, Barcelona, Spain

  9. Department of Clinical Foundations, Faculty of Medicine and Health Sciences, University of Barcelona, Barcelona, Spain

  10. Department of Basic Medical Sciences, Universitat Rovira i Virgili (URV), Institut d’Investigació Sanitària Pere Virgili (IISPV), Reus, Spain

  11. Biomedical Research Networking Center in Diabetes and Associated Metabolic Disorders (CIBERDEM), Instituto de Salud Carlos III (ISCIII), Madrid, Spain

  12. Biomedical Research Networking Center in Infectious Diseases (CIBERINFEC), Madrid, Spain

  13. University Hospital of Salamanca, Biomedical Research Institute of Salamanca (IBSAL), Cancer Research Center (IBMCC–USAL, CSIC), Salamanca, Spain

  14. Virgen de la Arrixaca University Hospital, IMIB-Arrixaca, University of Murcia, Murcia, Spain

  15. Department of Advanced Therapy Medicinal Product Development, Instituto de Salud Carlos III, Madrid, Spain

  16. Virgen del Rocío University Hospital, Institute of Biomedicine of Seville (IBIS/CSIC/CIBERONC), University of Seville, Seville, Spain

  17. Clinica Universidad de Navarra, Center for Applied Medical Research (CIMA), IDISNA, CIBERONC, Pamplona, Spain

  18. 12 de Octubre University Hospital, Complutense University of Madrid, i+12, CNIO, Madrid, Spain

  19. Clinical Immunology Unit, Hospital Sant Joan de Déu–Hospital Clínic de Barcelona, Barcelona, Spain

  20. University of Barcelona, Barcelona, Spain