Antibiotic-induced loss of gut microbiome metabolic output correlates with clinical responses to CAR T-cell therapy

Abstract

Antibiotic (ABX)-induced microbiome dysbiosis is widespread in oncology, adversely affecting outcomes and side effects of various cancer treatments, including immune checkpoint inhibitors and chimeric antigen receptor T-cell (CAR-T) therapies. In this study, we observed that prior exposure to broad-spectrum ABXs with extended anaerobic coverage such as piperacillin-tazobactam and meropenem was associated with worse anti-CD19 CAR-T therapy survival outcomes in patients with large B-cell lymphoma (N = 422) than other ABX classes. In a discovery subset of these patients (n = 67), we found that the use of these ABXs was in turn associated with substantial dysbiosis of gut microbiome function, resulting in significant alterations of the gut and blood metabolome, including microbial effectors such as short-chain fatty acids (SCFAs) and other anionic metabolites, findings that were largely reproduced in an external validation cohort (n = 58). Broader evaluation of circulating microbial metabolites revealed reductions in indole and cresol derivatives, as well as trimethylamine N-oxide, in patients who received ABX treatment (discovery, n = 40; validation, n = 28). These findings were recapitulated in an immune-competent CAR-T mouse model, in which meropenem-induced dysbiosis led to a systemic dysmetabolome and decreased murine anti-CD19 CAR-T efficacy. Furthermore, we demonstrate that SCFAs can enhance the metabolic fitness of CAR-Ts, leading to improved tumor killing capacity. Together, these results suggest that broad-spectrum ABX deplete metabolically active commensals whose metabolites are essential for enhancing CAR-T efficacy, shedding light on the intricate relationship between ABX exposure, microbiome function and their impact on CAR-T efficacy. This highlights the potential for modulating the microbiome to augment CAR-T immunotherapy. This trial was registered at www.clinicaltrials.gov as #NCT06218602.

Graphical abstract

Figure 1.

Impact of ABX exposure before CAR-T therapy on survival outcomes in the US cohort. (A) PFS in the US cohort stratified by ABX exposure. (B) Serum levels of markers of tumor burden (LDH) and systemic inflammation (CRP and ferritin) in CAR-T recipients exposed to ABX compared to those who were not exposed. (C) Peripheral blood counts including WBC, ALC, hemoglobin, and platelets, stratified by ABX exposure. (D) Incidence of ABX exposure before CAR-T therapy stratified by year. (E) Volcano plot showing effect of individual ABX classes given in the 6 weeks before CAR-T infusion on progression incidence. (F) PFS stratified by exposure to different ABX classes (no ABX, CEF, and ANA-ABX). (G) Progression incidence stratified by ABX classes (no ABX, CEF, and ANA-ABX). (H) PFS stratified by exposure to different ABX classes (no ABX, CEF, and PIM∗). CEF includes beta-lactams and antipseudomonal cephalosporins; ANA-ABX includes ABX targeting anaerobic commensals. ALC, absolute lymphocyte count; CRP, C-reactive protein; WBC, white blood cell count.

Figure 2.

Impact of ABX exposure on the stool SCFAs metabolome in the German and US cohorts. (A) SCFAs and lactic acid levels stratified by ABX exposure in the US cohort. (B) Stool SCFAs and lactic acid levels stratified by ABX class exposure in the US cohort. (C) Stool SCFAs and lactic acid levels stratified by PIM-ABX vs no ABX exposure in the German cohort. (D) Kaplan-Meier curves demonstrating improved PFS in both the German and US cohorts stratified by median stool VA levels.

Figure 3.

Impact of ABX on the plasma metabolome in the US cohort. (A) Unsupervised hierarchical clustering heat map illustrating associations between 21 circulating microbial metabolites detected via mass spectrometry and ABX use in the US cohort. (B) Box and whisker plots depicting the relationship between plasma TMAO, IS, CS, and UDCA with ABX use. (C) Kaplan-Meier survival curves demonstrating the relationship between plasma levels of TMAO, IS, and CS with PFS in patients with R/R LBCL who subsequently received CAR-T therapy. Optimal cutpoints were derived from the Cox model. (D) Associations between plasma levels of TMAO, IS, CS, UDCA, and microbes detected in the stool from whole shotgun sequencing data. The heat map illustrates normalized abundance/counts of stool microbes from patients with R/R LBCL. Dot plots depicting relative abundances (autoscaled) of plasma TMAO, IS, CS, and UDCA for matching patients are provided at the top. The heat map on the right illustrates Spearman correlation coefficients between respective stool microbes and circulating microbial metabolites. ANAAbx, anaerobic ABX which includes ABX targeting anaerobic commensals; CRS, cytokine release syndrom; ICANS, immune-effector cell-associated neurologic syndromes; OSD, overall survival duration; PFD, progression-free survival duration; UDCA, ursodeoxycholate.

Figure 3.

Impact of ABX on the plasma metabolome in the US cohort. (A) Unsupervised hierarchical clustering heat map illustrating associations between 21 circulating microbial metabolites detected via mass spectrometry and ABX use in the US cohort. (B) Box and whisker plots depicting the relationship between plasma TMAO, IS, CS, and UDCA with ABX use. (C) Kaplan-Meier survival curves demonstrating the relationship between plasma levels of TMAO, IS, and CS with PFS in patients with R/R LBCL who subsequently received CAR-T therapy. Optimal cutpoints were derived from the Cox model. (D) Associations between plasma levels of TMAO, IS, CS, UDCA, and microbes detected in the stool from whole shotgun sequencing data. The heat map illustrates normalized abundance/counts of stool microbes from patients with R/R LBCL. Dot plots depicting relative abundances (autoscaled) of plasma TMAO, IS, CS, and UDCA for matching patients are provided at the top. The heat map on the right illustrates Spearman correlation coefficients between respective stool microbes and circulating microbial metabolites. ANAAbx, anaerobic ABX which includes ABX targeting anaerobic commensals; CRS, cytokine release syndrom; ICANS, immune-effector cell-associated neurologic syndromes; OSD, overall survival duration; PFD, progression-free survival duration; UDCA, ursodeoxycholate.

Figure 4.

Impact of ABX on the plasma metabolome in the German cohort. (A) Unsupervised hierarchical clustering heat map illustrating associations between 15 circulating microbial metabolites detected via mass spectrometry with ABX use in the German cohort. (B) Box and whisker plots depicting the relationship between plasma TMAO, IS, and CS with ABX use. (C) Kaplan-Meier survival curves demonstrating the relationship between plasma levels of TMAO, IS, and CS and PFS in patients with R/R LBCL who subsequently received CAR-T therapy. Cut points were the same as described in Figure 3. (D) Associations between plasma levels of TMAO, IS, and CS and microbes detected in the stool from whole shotgun sequencing data. The heat map illustrates normalized abundance/counts of stool microbes from patients with R/R LBCL. Dot plots depicting relative abundances (autoscaled) of plasma TMAO, IS, and CS for matching patients are provided at the top. The heat map on the right illustrates Spearman correlation coefficients between respective stool microbes and circulating microbial metabolites. CRS, cytokine release syndrom; ICANS, immune-effector cell-associated neurologic syndromes; OSD, overall survival duration; PFD, progression-free survival duration; TDCA, taurodeoxycholic acid.

Figure 4.

Impact of ABX on the plasma metabolome in the German cohort. (A) Unsupervised hierarchical clustering heat map illustrating associations between 15 circulating microbial metabolites detected via mass spectrometry with ABX use in the German cohort. (B) Box and whisker plots depicting the relationship between plasma TMAO, IS, and CS with ABX use. (C) Kaplan-Meier survival curves demonstrating the relationship between plasma levels of TMAO, IS, and CS and PFS in patients with R/R LBCL who subsequently received CAR-T therapy. Cut points were the same as described in Figure 3. (D) Associations between plasma levels of TMAO, IS, and CS and microbes detected in the stool from whole shotgun sequencing data. The heat map illustrates normalized abundance/counts of stool microbes from patients with R/R LBCL. Dot plots depicting relative abundances (autoscaled) of plasma TMAO, IS, and CS for matching patients are provided at the top. The heat map on the right illustrates Spearman correlation coefficients between respective stool microbes and circulating microbial metabolites. CRS, cytokine release syndrom; ICANS, immune-effector cell-associated neurologic syndromes; OSD, overall survival duration; PFD, progression-free survival duration; TDCA, taurodeoxycholic acid.

Figure 5.

Meropenem exposure results in loss of antitumor efficacy of murine CAR-Ts. (A) Experimental design of in vivo recapitulation of CAR-T efficacy failure upon ABX exposure. (B) Quantification of absolute bacterial density of fecal samples collected from meropenem-exposed and vehicle-exposed A20 tumor-bearing mice on day 0 or the day of CAR-T transfer. (C) Inverse Simpson diversity index of the gut microbiome in meropenem-exposed and vehicle-exposed A20 tumor-bearing mice on the day of CAR-T transfer. (D) Principal coordinate analysis of fecal samples by meropenem exposure using weighted UniFrac distances on day 0 or the day of CAR-T transfer. (E) Changes in fecal bacterial composition upon exposure to meropenem displayed as stacked bar graphs of bacterial genera on the day of CAR-T transfer compared with vehicle control. Each column represents 1 mouse. (F) Volcano plot of differentially abundant bacterial genera enriched in fecal samples from meropenem-exposed and non-exposed mice on day 0 or day of CAR-T transfer. (G) Tumor growth curve of murine lymphoma, A20, after exposure of mice to meropenem and CAR-T treatment compared with vehicle and/or non–CAR-T controls. (H) Kaplan-Meier curves of meropenem-exposed or vehicle-treated mice without infusion of CAR-Ts (left) and with infusion of CAR-Ts (right). (I) Box-and-whiskers plot depicting the autoscaled abundance of the uremic metabolites TMAO (leftmost), phenol sulfate (center-left), IS (center-right), and CS (rightmost) of tumor-bearing mice exposed to meropenem compared with vehicle control. Data are derived from combining 3 individual experiments with 2 to 5 mice per group. The box in each box-and-whiskers plot spans the interquartile range, with the lower and upper boundaries of the box marking the 25th and 75th percentiles, respectively. The median value is indicated by a line within the box. The whiskers extend to the minimum and maximum values observed in each group. Statistical significance of panel G was analyzed using the Mann-Whitney U test on day 21. Statistical significance of survival difference between vehicle and meropenem groups in panel H (left) and vehicle + CAR-T and meropenem + CAR-T groups in panel H (right) was analyzed using the log-rank test. Statistical significance in panels B, C, and I was analyzed using the Mann-Whitney U test. CTX, cyclophosphamide; PC, principal coordinate. Figure panel A created with BioRender.com.

Figure 6.

Impact of VA on the CAR-T metabolome in vitro. (A) PLSDA of metabolic profiles of CD4 (left) and CD8 (right) CAR-T cells treated with vehicle control or 1-mM VA. (B) Relative abundances (area units) of acetylcarnitine, carnitine, deoxycarnitine, propionyl-carnitine, C4-carnitine, and C6-carnitine in CAR-Ts after 1-week exposure to 1-mM VA or control media. (C) Heat map depicting relative abundances (z scaled) of metabolites involved in purine (top) or pyrimidine (bottom) metabolism in CAR-Ts after 1-week exposure to 1-mM VA or control media. Statistical significance was determined by 2-sided Welch t tests. PLSDA, partial least squares discriminant analysis. ADP, adenosine 5′-diphosphate; AMP, adenosine 5′-monophosphate; dCMP, deoxycytidine monophosphate; dGMP, deoxyguanosine monophosphate; dGTP, deoxyguanosine triphosphate; GDP, guanosine diphosphate; GMP, guanosine monophosphate; GTP, guanosine triphosphate; IDP, inosine diphosphate; IMP, inosine monophosphate; SEM, standard error of the mean; UMP, uridine monophosphate; UDP, uridine diphosphate; XMP, xanthosine monophosphate.