Gut Microbiome Composition As a Prognostic Marker in CD19-Targeting CAR T-Cell Therapy for Relapsed/Refractory Diffuse Large B-Cell Lymphoma

Introduction CD19-targeting chimeric Antigen Receptor (CAR) T-cell therapy represents a groundbreaking treatment for relapsed/refractory diffuse large B-cell lymphoma (R/R DLBCL). However, some patients experience disease progression despite this therapy, often due to CAR T-cell dysfunction or a hostile tumor microenvironment. This study aims to investigate resistance mechanisms to CAR T-cell therapies by analyzing the gut microbiome and exploring synergistic strategies to modulate the microbiome and overcome these resistance mechanisms. Methods We analyzed 47 patients with R/R DLBCL who were schedule to undergo anti-CD19 CAR T-cell therapy, including 29 patients receiving tisagenlecleucel (tisa-cel) and 18 receiving anbalcatagene autoleucel (anbal-cel). With patient consent, fecal and serum samples were collected at baseline before apheresis and one month after CAR T-cell infusion. Shotgun metagenomic sequencing was performed using the Ez-Mx platform. Treatment responses were assessed at one- and three-months post-infusion, with ongoing follow-up for progression-free survival (PFS). Results All patients had R/R DLBCL, with a median of two prior treatments (range: 1-4). Following CAR T-cell therapy, the complete response (CR) rate was 61%. The gut microbiome of these R/R DLBCL patients was compared with data from 150 treatment-naïve DLBCL patients and 47 age- and sex-matched healthy controls. R/R DLBCL patients exhibited a significantly different gut microbiome composition, characterized by a higher abundance of Enterobacteriaceae compared to both treatment-naïve DLBCL patients and healthy controls (P < 0.001). At baseline, significant differences in microbiome composition were observed between responders and non-responders (p=0.007, PERMANOVA), though no significant differences were noted post-treatment (p=0.4332, PERMANOVA). Taxonomic biomarker analysis using ANCOM-BC identified Bacteroides fragilis and Negativibacillus MSSCI00191739 as markers for responders, while Faecalibacterium prausnitzii, Faecalibacterium MSSCM00806154, and Faecalibacterium PRLE were identified as markers for non-responders. Bacteroides fragilis was significantly elevated in patients who achieved CR compared to those who did not (p=0.007, Mann-Whitney U test), whereas Faecalibacterium prausnitzii was more abundant in non-responders (p=0.03, Mann-Whitney U test). The presence of Bacteroides fragilis prior to CAR T-cell therapy was associated with significantly longer PFS (P = 0.022), while the absence of Faecalibacterium prausnitzii also correlated with extended PFS (p=0.008). Conclusions Our study suggests that the gut microbiome may serve as a prognostic marker for resistance to CAR T-cell therapy in R/R DLBCL patients. Pre-treatment samples revealed gut microbial dysbiosis in R/R DLBCL patients, characterized by high levels of Enterobacteriaceae compared to treatment-naïve DLBCL or healthy controls. Non-responders exhibited distinct pre-treatment microbiome compositions, with lower Bacteroides fragilis and higher Faecalibacterium prausnitzii associated with shorter PFS. These findings underscore the potential for microbiome modulation as a strategy to overcome CAR T-cell therapy resistance and improve patient outcomes. We are currently conducting functional studies to explore the interplay between the tumor microenvironment and microbiome.