https://orcid.org/0000-0002-7776-3263
Visual Abstract
Antibiotics disrupt the delicate balance of bacteria, fungi, and viruses in the human microbiome. Growing evidence indicates a significant relationship between the intestinal microbiome and cellular therapy, which aligns with the established influence of the microbiome on immune responses. When examining the link between cellular therapy and the microbiome, it is essential to understand how disruptions in the microbiome, especially those caused by antibiotics, affect these therapies. Here, we discuss the impact of antibiotics on the intestinal microbiome, cellular therapy outcomes, and associated toxicities, particularly in the context of hematopoietic cell transplantation and chimeric antigen receptor T-cell therapy. Furthermore, we examine the mechanisms through which antibiotics affect cellular therapy, the future implications of this knowledge, and the areas that warrant further investigation.
Introduction
Cellular therapy continues to gain an increasingly important role in medicine.1 Aside from blood transfusion, the oldest and most common cellular therapy is hematopoietic cell transplant (HCT), in which hematopoietic progenitor cells from the patient (autologous HCT) or donor (allogeneic HCT [allo-HCT]) are used in treating malignant and nonmalignant life-threatening conditions.2 Immune effector cell therapies (IECTs), a newer class of cellular therapy that includes chimeric antigen receptor (CAR) T-cell therapy (CAR-T), tumor-infiltrating lymphocytes (TILs), and T-cell receptor (TCR) therapy, trigger therapeutic immunologic responses.1,3 IECTs have revolutionized the treatment of multiple diseases and continue to be an area of development.
As cellular therapy expands, so does the need to optimize its safety and efficacy. An emerging focus is on exploring the microbiome’s impact on cellular therapy. The microbiome includes diverse microorganisms colonizing the body’s mucosal and skin environments; the metabolome is the metabolite profile produced by these microorganisms. The wide roles of the microbiome are still being defined, but specific changes have been linked to a range of illnesses.4 The microbiome is intimately involved in the immune system.5 Thus, it would follow that the microbiome plays a prominent role in cellular therapy in which multilineage hematopoietic and immune reconstitution or therapeutic immunologic cascade are direct treatment impacts.1
Characteristics of a healthy microbiome include high alpha diversity and resident anaerobic commensal organisms such as Bacteroides, Prevotella, and Alistipes spp.6,7 The composition of the microbiome is affected by environmental,4 dietary,4,8,9 lifestyle,4 and medication4,6 exposures. Antibiotics are one of the most well-studied factors leading to dysbiosis; antibiotics cause decreased alpha diversity, increased relative abundance of facultative anaerobes, and impaired metabolic function.4,6 Such perturbations can lead to antibiotic resistance, opportunistic infections, increased propensity to certain disease states, and impaired efficacy of immunotherapies.4
Patients receiving cellular therapy experience frequent antibiotic exposure. It is important to understand the impact of antibiotics and resultant dysbiosis on the efficacy and toxicity of cellular therapy. Here, we examine the current evidence regarding how antibiotic use affects the microbiome and cellular therapy outcomes. We also analyze mechanistic data and explore future implications of these findings.
HCT
Impact of antibiotics on the microbiome and HCT outcomes: preclinical data
In mouse HCT models, evidence suggests that recipient, not donor, antibiotic exposure plays a crucial role in determining the outcomes of transplant10,11 (Table 1). Peri-transplant administration of gut-decontaminating antibiotics has been shown to impair engraftment. This effect has been noted with antibiotics that are systemically well absorbed, such as ampicillin and enrofloxacin, and with antibiotics that are poorly absorbed, including oral vancomycin, amikacin, and neomycin.12,13 Notably, impaired hematopoiesis has been observed in mice treated with antibiotics in the absence of allo-HCT.10 Mechanistically, the decrease in hematopoiesis was associated with reduced energy harvest by the gut microbiota from the diet, with sucrose supplementation significantly improving hematopoietic recovery in the presence of antibiotic exposure.12
Preclinical evidence: role of antibiotics in modulating microbiota and outcomes of HCT
| Antibiotics | Model | Method | Effects on microbiome | Outcomes | Ref |
|---|---|---|---|---|---|
| Syngeneic | |||||
| Vancomycin, 0.5 g/L Amikacin, 0.5 g/L | B6 → B6 | Fed water ad libitum from d –5 to the end of the experiment | ↓ Colonic bacterial abundance | ↓ Posttransplant hematopoiesis | 12 |
| Ampicillin, 0.5 g/L Enrofloxacin, 0.25 g/L | B6 → B6 | Fed water ad libitum from d –5 to the end of the experiment | ↓ Colonic bacterial abundance | ↓ Posttransplant hematopoiesis | 12 |
| Vancomycin, 0.5 g/L Neomycin, 1 g/L | B6 T cells → NK-depleted RAG1–/– B6 | Fed water ad libitum from d –7 to d 28 of transplant | ↓ Microbial diversity ↓ Firmicutes ↑ Verrucomicrobia ↑ Proteobacteria ↓ Bacteroidetes ↓ Spirochaetes | ↓ BM cell number ↓ BM myeloid and NK cells ↓ Colonic inflammation | 13 |
| Bacitracin, 5 g/L Neomycin, 2 g/L Natamycin, 1.2 mg/L Meropenem, 1 g/L Vancomycin, 1 g/L | B6 → B6 | Fed water ad libitum from d 28 to d 42 of transplant | ↓ Firmicutes ↓↓ Bacteroidetes ↑ Firmicutes to Bacteroidetes ratio ↑ Proteobacteria | ↑ BM-derived monocyte infiltration into the brain | 14 |
| Allo-HCT | |||||
| Ampicillin, 5 mg Metronidazole, 4 mg Clindamycin, 5 mg Vancomycin, 5 mg | MHC-mismatched B10.BR → B6 | Gavaged daily for 6 d before transplant | ↓ Obligate anaerobes | ↓ Survival | 15 |
| Ampicillin, 250 mg/kg Metronidazole, 250 mg/kg Kanamycin, 250 mg/kg Vancomycin, 125 mg/kg | MHC-mismatched BALB/c → B6 | Gavaged daily from d –1 to d 21 of transplant | ↓ Bacteroidetes ↑ Proteobacteria | ↓ Survival | 16 |
| Piperacillin-tazobactam, 100 mg/kg | MHC-matched minor antigen–mismatched B6 → 129S1 | Subcutaneous injected daily from d 10 to d 24 of transplant | Near-complete gut-decontamination | ↓ Survival∗ | 17 |
| Ampicillin, 200 μg Gentamicin, 200 μg Metronidazole, 200 μg Kanamycin, 200 μg Vancomycin, 100 μg | MHC-mismatched BALB/c → B6 | Gavaged daily from d 0 to d 5 of transplant | ↓ Butyrate-producing Clostridiales species ↑ Non-butyrogenic genus Lactobacillus, phylum Proteobacteria | ↓ Survival | 18 |
| Ampicillin, 0.5 g/L Enrofloxacin, 0.25 g/L | MHC-matched minor antigen–mismatched 129 → B6 | Fed water ad libitum from d –5 to the end of the experiment | ↓ Colonic bacterial abundance | ↓ Post-transplant hematopoiesis | 12 |
| Vancomycin, 0.5 g/L Neomycin, 1 g/L | MHC-mismatched BALB/c T cells → NK-depleted RAG1–/– B6 | Fed water ad libitum from d –7 to d 28 of transplant | ↓ Microbial diversity ↓ Firmicutes ↑ Verrucomicrobia ↑ Proteobacteria ↓ Bacteroidetes | ↓ BM and spleen conventional T cells | 13 |
| Imipenem/cilastatin, 100 mg/kg | MHC-matched minor antigen–mismatched B6 → 129S1 | Subcutaneous injected daily from d 10 to d 24 of transplant | ↓ Clostridia∗, ↑ Verrucomicrobiales∗, ↑ Akkermansia∗ | ↓ Survival∗, ↑ GI GVHD∗, ↑ Donor CD4+ T cells, IL-23, and colonic granulocyte number∗ | 17 |
| Imipenem/cilastatin, 50 mg/kg | MHC-mismatched B6 → BALB/c | Gavaged daily from d –3 to d –1 of transplant | ↑ Bacteroidaceae and Bacteroidales_S24-7_group†, ↓ Porphyromonadaceae† | ↓ Survival ↑ GVHD† | 19 |
| Ampicillin, 1 g/L | MHC-mismatched B10.BR → B6 | Fed water ad libitum from d –21 to d –14 of transplant | ↓ Lactobacillus ↑ Blautia | ↓ Survival ↑ GI GVHD | 20 |
| Meropenem, 0.625 g/L | MHC-mismatched B6 → B6D2F1 | Fed water ad libitum from d 3 to d 15 of transplant | ↓ Bacterial density ↓ α-diversity ↓ Clostridia ↑ Bacteroides, Enterococcus, Erysipelatoclostridium, Bifidobacterium, and Akkermansia | ↓ Survival ↑ GI GVHD | 21 |
| Gentamicin, 0.5 g/L | MHC-matched minor antigen–mismatched B10.D2/nSnSlc → BALB/cCrSlc | Orally administered from d –14 to d 28 of transplant | Not mentioned | ↓ cGVHD | 22 |
| Polymyxin B, 100 mg/kg | MHC-mismatched B6 → B6D2F1 | Gavaged daily from d –4 to d 28 of transplant | ↓ E coli | ↑ Survival ↓ GVHD ↓ Donor T-cell expansion in the small intestine | 23 |
| Azithromycin, 100 mg/kg | MHC-mismatched B6 → BALB/c | Gavaged daily from d –2 to d 2 of transplant | Not mentioned | ↑ Survival ↓ GVHD | 24 |
| Antibiotics | Model | Method | Effects on microbiome | Outcomes | Ref |
|---|---|---|---|---|---|
| Syngeneic | |||||
| Vancomycin, 0.5 g/L Amikacin, 0.5 g/L | B6 → B6 | Fed water ad libitum from d –5 to the end of the experiment | ↓ Colonic bacterial abundance | ↓ Posttransplant hematopoiesis | 12 |
| Ampicillin, 0.5 g/L Enrofloxacin, 0.25 g/L | B6 → B6 | Fed water ad libitum from d –5 to the end of the experiment | ↓ Colonic bacterial abundance | ↓ Posttransplant hematopoiesis | 12 |
| Vancomycin, 0.5 g/L Neomycin, 1 g/L | B6 T cells → NK-depleted RAG1–/– B6 | Fed water ad libitum from d –7 to d 28 of transplant | ↓ Microbial diversity ↓ Firmicutes ↑ Verrucomicrobia ↑ Proteobacteria ↓ Bacteroidetes ↓ Spirochaetes | ↓ BM cell number ↓ BM myeloid and NK cells ↓ Colonic inflammation | 13 |
| Bacitracin, 5 g/L Neomycin, 2 g/L Natamycin, 1.2 mg/L Meropenem, 1 g/L Vancomycin, 1 g/L | B6 → B6 | Fed water ad libitum from d 28 to d 42 of transplant | ↓ Firmicutes ↓↓ Bacteroidetes ↑ Firmicutes to Bacteroidetes ratio ↑ Proteobacteria | ↑ BM-derived monocyte infiltration into the brain | 14 |
| Allo-HCT | |||||
| Ampicillin, 5 mg Metronidazole, 4 mg Clindamycin, 5 mg Vancomycin, 5 mg | MHC-mismatched B10.BR → B6 | Gavaged daily for 6 d before transplant | ↓ Obligate anaerobes | ↓ Survival | 15 |
| Ampicillin, 250 mg/kg Metronidazole, 250 mg/kg Kanamycin, 250 mg/kg Vancomycin, 125 mg/kg | MHC-mismatched BALB/c → B6 | Gavaged daily from d –1 to d 21 of transplant | ↓ Bacteroidetes ↑ Proteobacteria | ↓ Survival | 16 |
| Piperacillin-tazobactam, 100 mg/kg | MHC-matched minor antigen–mismatched B6 → 129S1 | Subcutaneous injected daily from d 10 to d 24 of transplant | Near-complete gut-decontamination | ↓ Survival∗ | 17 |
| Ampicillin, 200 μg Gentamicin, 200 μg Metronidazole, 200 μg Kanamycin, 200 μg Vancomycin, 100 μg | MHC-mismatched BALB/c → B6 | Gavaged daily from d 0 to d 5 of transplant | ↓ Butyrate-producing Clostridiales species ↑ Non-butyrogenic genus Lactobacillus, phylum Proteobacteria | ↓ Survival | 18 |
| Ampicillin, 0.5 g/L Enrofloxacin, 0.25 g/L | MHC-matched minor antigen–mismatched 129 → B6 | Fed water ad libitum from d –5 to the end of the experiment | ↓ Colonic bacterial abundance | ↓ Post-transplant hematopoiesis | 12 |
| Vancomycin, 0.5 g/L Neomycin, 1 g/L | MHC-mismatched BALB/c T cells → NK-depleted RAG1–/– B6 | Fed water ad libitum from d –7 to d 28 of transplant | ↓ Microbial diversity ↓ Firmicutes ↑ Verrucomicrobia ↑ Proteobacteria ↓ Bacteroidetes | ↓ BM and spleen conventional T cells | 13 |
| Imipenem/cilastatin, 100 mg/kg | MHC-matched minor antigen–mismatched B6 → 129S1 | Subcutaneous injected daily from d 10 to d 24 of transplant | ↓ Clostridia∗, ↑ Verrucomicrobiales∗, ↑ Akkermansia∗ | ↓ Survival∗, ↑ GI GVHD∗, ↑ Donor CD4+ T cells, IL-23, and colonic granulocyte number∗ | 17 |
| Imipenem/cilastatin, 50 mg/kg | MHC-mismatched B6 → BALB/c | Gavaged daily from d –3 to d –1 of transplant | ↑ Bacteroidaceae and Bacteroidales_S24-7_group†, ↓ Porphyromonadaceae† | ↓ Survival ↑ GVHD† | 19 |
| Ampicillin, 1 g/L | MHC-mismatched B10.BR → B6 | Fed water ad libitum from d –21 to d –14 of transplant | ↓ Lactobacillus ↑ Blautia | ↓ Survival ↑ GI GVHD | 20 |
| Meropenem, 0.625 g/L | MHC-mismatched B6 → B6D2F1 | Fed water ad libitum from d 3 to d 15 of transplant | ↓ Bacterial density ↓ α-diversity ↓ Clostridia ↑ Bacteroides, Enterococcus, Erysipelatoclostridium, Bifidobacterium, and Akkermansia | ↓ Survival ↑ GI GVHD | 21 |
| Gentamicin, 0.5 g/L | MHC-matched minor antigen–mismatched B10.D2/nSnSlc → BALB/cCrSlc | Orally administered from d –14 to d 28 of transplant | Not mentioned | ↓ cGVHD | 22 |
| Polymyxin B, 100 mg/kg | MHC-mismatched B6 → B6D2F1 | Gavaged daily from d –4 to d 28 of transplant | ↓ E coli | ↑ Survival ↓ GVHD ↓ Donor T-cell expansion in the small intestine | 23 |
| Azithromycin, 100 mg/kg | MHC-mismatched B6 → BALB/c | Gavaged daily from d –2 to d 2 of transplant | Not mentioned | ↑ Survival ↓ GVHD | 24 |
↑ represents increase; ↓, decrease; and →, transplant to.
BM, bone marrow; cGVHD, chronic GVHD; IL-23, interleukin-23; NK, natural killer; Ref, reference.
Compared with aztreonam-treated mice.
Compared with imipenem/cilastatin- and galactooligosaccharide-treated mice.
Administering nonabsorbable gut-decontaminating antibiotics has been shown to disrupt the blood-brain barrier (BBB) during syngeneic HCT, increasing brain-infiltrating bone marrow–derived cells. This disruption was linked to the depletion of gut microbiota. Fecal microbiota transplantation (FMT) from mice without antibiotic exposure significantly restores the integrity of the BBB.14
In allo-HCT models, administering gut-decontaminating antibiotics either before or after transplantation significantly accelerates recipient mortality.15-18 In major histocompatibility complex (MHC)–mismatched models, anaerobe-targeting antibiotics, including ampicillin, metronidazole, vancomycin, clindamycin, and kanamycin peri-transplantation, significantly decreased survival.15,16 Additionally, in an MHC-matched minor antigen–mismatched model, subcutaneous administration of piperacillin/tazobactam from days 10 to 24 after allo-HCT was associated with reduced survival.17 In line with clinical evidence, nonrelapse mortality of the allo-HCT mice was characterized by reduced hematopoietic reconstitution and increased graft-versus-host disease (GVHD).12,13,17,19-21 Notably, the administration of imipenem/cilastatin, ampicillin, and meropenem has been linked to an elevated risk of gastrointestinal (GI) GVHD.17,19-21
Clostridiales may play an important role in regulating the outcomes of allo-HCT. Clostridiales produce butyrate, a short-chain fatty acid, which plays a role in regulating intestinal T cells, improving the integrity of the intestinal lining, decreasing apoptosis, and potentially mitigating GVHD.15,25,26 In an MHC-mismatched model, recipient mice were given a gut-decontaminating antibiotic cocktail daily via gavage for 6 days starting from the day of transplantation. This cocktail of ampicillin, gentamicin, metronidazole, kanamycin, and vancomycin reduced the abundance of butyrate-producing Clostridiales species and increased mortality.18 Class Clostridia, which belongs to Clostridiales, also decreased in imipenem/cilastatin- and meropenem-treated mice and was associated with reduced survival and increased GVHD. Interestingly, although no difference was found in mice treated with imipenem/cilastatin, meropenem significantly reduced the fecal concentration of butyrate.17,21 Treating recipient mice with a 17-strain Clostridial cocktail after antibiotic treatment effectively mitigated the increased mortality associated with antibiotics, highlighting the crucial role of Clostridiales.15
Although many antibiotics are linked to worse outcomes, some have been associated with reduced GVHD and improved survival. For example, peri-transplant gentamicin exposure ameliorates chronic GVHD, as evidenced by enhanced systemic and ocular clinical phenotypes.22 These beneficial effects may be attributed to targeting bacteria known to contribute to GVHD or affecting certain cell types involved in GVHD development. For instance, polymyxin B has effectively eliminated intestinal Escherichia coli, a bacterium associated with GVHD severity after HCT, resulting in milder GVHD and prolonged survival.23 Azithromycin acts as an NF-κB inhibitor in murine dendritic cells (DCs), inhibiting their maturation and functional activity. This inhibition can disrupt the interaction between DCs and T cells, potentially leading to decreased acute GVHD (aGVHD).24
Antibiotics, the microbiome, and HCT: clinical background
HCT patients have baseline dysbiosis with lower alpha diversity and distinct enterotypes compared with healthy controls.27-29 Further dysbiosis is created via conditioning-induced mucosal epithelial damage,30 dietary alterations,31 antibiotic exposure,6,32-36 and a new immune system.37 Microbiome changes include decreased diversity and new domination by single predominant taxa.27-29,38-40 Dysbiosis during HCT is associated with negative outcomes. Low intestinal alpha diversity has been associated with increased GI symptoms,38 neutropenic fever,38 disease progression or death,27,28,41 transplant-related mortality, aGVHD, and GVHD-related mortality.27,39,42 Initial preclinical research suggested that antibiotic-induced “gut decontamination” may help lessen aGVHD, but this was not seen in the clinical setting.43
Despite advances in HCT outcomes, both relapse and nonrelapse mortality, primarily driven by GVHD and infection, remain substantial.44,45 HCT creates immunosuppression, necessitating frequent antibiotic exposure.33,45-47 Given the evidence showing a close relationship between the microbiome and HCT outcomes, understanding the role of antibiotics in this connection is crucial.
HCT: antibiotics increase dysbiosis and correlate with poor outcomes
The post-HCT microbiome has been characterized by lower diversity and predominance of Enterococcus, Streptococcus, and Proteobacteria.40,43,48 Changes are exacerbated by peri-transplant antibiotic exposure.35,39 A study involving 8767 samples from 1362 patients across 4 centers found that exposure to piperacillin/tazobactam and meropenem was linked to reduced gut microbial diversity; similar results have been demonstrated in other studies.27,49,50 β-lactams and metronidazole have been associated with decreased alpha diversity in HCT patients.40,41
A study of 577 samples from 233 patients undergoing autologous HCT or allo-HCT found correlations between the duration of peri-transplant antibiotic exposure and the richness and diversity of antibiotic-resistance genes.34 Effect sizes were largest for ciprofloxacin on species measures and meropenem on the diversity of antibiotic-resistant genes. Piperacillin/tazobactam and IV vancomycin were most associated with vancomycin-resistant Enterococcus (VRE).36 Domination of the intestinal microbiome by VRE is associated with VRE bacteremia.51 Sulfamethoxazole/trimethoprim and IV vancomycin were linked to specific antibiotic-resistance genes, whereas piperacillin/tazobactam, meropenem, and ciprofloxacin had broader effects.34
Fluoroquinolones are frequently used for neutropenic fever prophylaxis; the impact of this class on the microbiome appears variable.33,52-54 Studies within the HCT population suggest fluoroquinolone exposure is not associated with increased enterococcal domination, reduced diversity, or GVHD.40,41,55 Fluoroquinolone exposure is associated with decreased risk of gram-negative rod bacteremia and decreased Proteobacteria domination.40
HCT: antibiotics correlate with increased incidence of intestinal GVHD
Preclinical data suggested that resident intestinal bacteria played a key role in the pathogenesis of GI aGVHD and led to the hypothesis that gut decontamination would reduce this risk. This strategy did not demonstrate clinical benefit.43 As methods of studying microbiome have evolved from culture-based assays to sequencing, the relationship between antibiotics, gut microbiome, and GI GVHD has proven to be complex. Dysbiosis associated with GVHD is characterized by loss of diversity, predominance of Enterobacteriales and Lactobacillales, loss of Clostridia, and subsequently decreased butyrate production.19,23,56-58
Investigators examined the effects of anaerobic-targeting antibiotics on the microbiome, GI GVHD, and response in 1214 pediatric and adult patients who experienced peri-HCT neutropenic fever. Exposure to anaerobic antibiotics was associated with increased risk of GI aGVHD (hazard ratio [HR], 1.26; 95% confidence interval [CI], 1.03-1.54; P = .02) and aGVHD-related mortality (HR, 1.63; 95% CI, 1.08-2.46; P = .02) compared with nonanaerobic spectrum antibiotics. There was no significant difference in chronic GI GVHD between the 2 cohorts.59 A cohort of 857 adults demonstrated similar results, with exposure to imipenem/cilastatin and piperacillin/tazobactam significantly associated with increased GVHD-related mortality at 5 years compared with untreated patients. This association was not seen in patients treated with aztreonam and cefepime.17 In an adult cohort of 399 patients, cumulative antibiotic exposure (HR, 2.46; 95% CI, 1.59-3.81; P < .001) and sequential treatment with penicillin derivatives and carbapenems (HR, 6.22; 95% CI, 1.27-30.31; P, not reported [NR]) were associated with GI GVHD at day 100.60 Another study of 211 adult patients identified that broad-spectrum antibiotic use during the neutropenic period is associated with GI GVHD (HR, 3.25; 95% CI, 1.13-9.34; P= .029).58
The microbiome was assessed in 36 pediatric patients, 26 of whom received cefepime and 10 piperacillin/tazobactam in a study by Tanaka et al. Alpha diversity significantly declined from a Shannon index of 2.0 (interquartile range, 1.6-2.5) to 0.9 (interquartile range, 0.3-1.2) in the piperacillin/tazobactam but not in the cefepime cohort. Postantibiotic composition differed significantly from preantibiotic composition in both groups. Exposure to anaerobic antibiotics resulted in a decline in the relative abundance of Bifidobacteriales and Clostridiales, an increase in the relative abundance of Lactobacillales, and a decrease in the abundance of genes for butyrate biosynthesis.59 In an adult cohort, piperacillin/tazobactam therapy was associated with larger decreases in Bacteroidetes and Lactobacillus than those receiving cefepime or aztreonam, although the decreases in Clostridia and Actinobacteria were not significantly different.17 Decreased Clostridia has been associated with GI GVHD, whereas Blautia and Akkermansia muciniphila are associated with reduced risk of GI GVHD and associated death.40,56,61,62
In a prospective cohort study involving 2023 HCT patients, machine learning (ML) models for proportional hazard regression and marginal structural analysis were developed to explore the relationship between antibiotic exposures and the risk of grade 2 to 4 aGVHD. There were many significant associations between antibiotic exposure across various classes and peri-HCT time points with an increased risk of grade 2 to 4 aGVHD. Interestingly, in the marginal structural model, exposure to penicillin with a β-lactamase inhibitor during week 1 was associated with an increased risk of aGVHD (HR, 6.55; 95% CI, 2.35-18.20; P, NR), whereas exposure to these antibiotics before HCT was associated with decreased risk (HR, 0.59; 95% CI, 0.37-0.94; P, NR).63
HCT future directions
Evidence highlighting the influence of antibiotics on the microbiome and HCT outcomes carries significant implications because it underscores the necessity of antibiotic stewardship during the peri-HCT period. Areas of interest include optimizing the duration of exposure and the choice of agent to balance the high risk of infection with the negative impact of antibiotics on the microbiome and outcomes. Absolute neutrophil count-driven antimicrobial prophylaxis, in which fluoroquinolone prophylaxis is delayed until the onset of severe neutropenia as opposed to conventional initiation on day –1 from transplant, is a reasonable strategy in reducing total antibiotic exposure without compromising infection control.64 The choice of agent in the treatment of neutropenic fever is being investigated in an ongoing phase 2 trial comparing the impact of piperacillin/tazobactam vs cefepime on changes in Clostridiales abundance.65 Results could help inform antibiotic choice to balance the treatment of neutropenic fever while minimizing harmful dysbiosis.
Targeting antibiotic-related dysbiosis to improve HCT outcomes is an additional area of interest.66,67 Increased GVHD has been associated with the predominance of Enterococcus, which is commonly seen in the post-HCT microbiome. Enterococcus growth depends on lactose, and preclinical models have shown that avoidance of dietary lactose decreases the severity of GVHD in mice. Consequently, a lactose-free diet or lactase supplementation may reduce Enterococcus dominance and its negative impacts.68 Probiotics, which are live microorganisms found in foods or supplements such as Lactobacillus spp. and Bifidobacterium spp., may help alleviate dysbiosis caused by antibiotic use.66 One small study demonstrated potential benefits of treatment with a prebiotic and probiotic mixture in reducing aGVHD.69-71 There is an ongoing phase 1 study of CBM588, a probiotic Clostridium butyricum strain, in patients undergoing allo-HCT to investigate the intervention’s impact on diarrhea, infection, GVHD, and response.72 Although some studies have demonstrated the safety of this approach, cases of probiotic-related infections, including bacteremia, do raise concern.73
FMT may mitigate the negative impacts of antibiotics on the microbiome of patients undergoing HCT. FMT has been used experimentally in the HCT setting to treat recurrent Clostridioides difficile infection or steroid-resistant acute GI GVHD and as a preventative strategy to decolonize resistant flora.67 There are 4 actively recruiting studies investigating the role of FMT in the prevention74,75 and treatment76,77 of GVHD. There is 1 planned study that will investigate the impact of FMT 2 weeks before allo-HCT on microbiome diversity throughout the transplant course, rates of infections and GVHD, and survival outcomes.78
Gut decontamination has fallen out of favor, but careful selection of peri-transplant antibiotics may positively modulate the microbiome. For example, given the decreased risk of aGVHD associated with exposure to penicillin with β-lactamase inhibitors in the week before HCT, this may be a valid strategy for decreasing the risk of aGVHD.63 Rifaximin has also gained attention as a microbiome modulator due to evidence suggesting a reduction in microbial diversity loss and a decrease in enterococcal load.79 Ongoing studies are investigating the impact of rifaximin on microbial diversity, antibiotic resistance, infections, GVHD, and mortality outcomes.80,81
IECTs
IECTs initiate an immune response to manage disease.1 The infusion of TILs, helper T cells, regulatory T cells, and genetically modified T cells, including CAR and TCR constructs, represent exciting strategies in the treatment of many disease states.1 US Food and Drug Administration approval has been granted for several CAR-T products in hematologic malignancies, 1 TCR product for sarcoma, and 1 TIL product for melanoma.82 Data on the relationship among antibiotics, microbiome, and IECT effectiveness and toxicity are limited, aside from CAR T cells. However, given the known impacts of antibiotics on T cells, this will be an important area of investigation.
Impact of antibiotics on T cells and IECTs: preclinical evidence
Antibiotics downregulate proinflammatory cytokines and alter the development and function of T cells. Antibiotics, including tetracycline, fluoroquinolone, penicillin, azithromycin, and metronidazole, exhibit suppressive effects on CD4+ helper T cells; this occurs via various mechanisms, including direct proliferation inhibition, antiapoptotic protein downregulation, mitochondrial function impairment, and effector cytokine reduction.83 The impact of antibiotics on CD8+ T cells appears more variable, with some preclinical studies showing increased effector functions in response to antibiotics, whereas others show decreased proliferation and activity.83
Preclinical studies have suggested some positive effects of antibiotic therapy on CAR-T outcomes (Table 2). For example, in a systemic A20 lymphoma model with anti-CD19 CAR T cells, gut-decontaminating antibiotics enhanced CAR T-cell persistence in long-term survivors, but survival was unchanged.84 In solid A20 lymphoma and B16-CD19 melanoma models treated with anti-CD19 CAR T cells, oral vancomycin treatment increased tumor-infiltrating DCs and improved CAR T-cell antitumor activity, leading to smaller tumors and increased T-cell infiltration. This effect depended on interleukin-12.85,89
Preclinical evidence: role of antibiotics in modulating microbiota and outcomes of IECTs
| Antibiotics | Model | Method | Effects on microbiome | Outcomes | Ref |
|---|---|---|---|---|---|
| Ciprofloxacin, 0.15 g/L Gentamicin, 0.2 g/L Bacitracin, 1 g/L Streptomycin, 2 g/L | CD19-CAR T cell → BALB/c with CD19+ lymphoma | Fed water ad libitum from 29 d before IECT throughout the experiment | Not mentioned | ↓ Spleen B-cell recovery ↑ Spleen CAR T-cell persistence | 84 |
| Vancomycin, 0.5 g/L | CD19-CAR T cell → BALB/c or B6 with CD19+ tumor cells | Fed water ad libitum for the duration of the experiment | ↓ α-diversity ↓ Ruminococcaceae ↓ Lachnospiraceae | ↓ Tumor size ↑ Spleen and tumor-infiltrating T cells | 85 |
| Vancomycin, 1 g/L Metronidazole, 1 g/L Cefoxitin, 1 g/L Gentamicin, 1 g/L | Marilyn T cells (Y chromosome–targeting TCR T) → male B6 | Fed water ad libitum for 2 wk before IECT | Gut decontamination | ↓ Marilyn T-cell infiltration in the gut ↓ GVHD | 86 |
| Vancomycin, 1 g/L Metronidazole, 1 g/L Cefoxitin, 1 g/L Gentamicin, 1 g/L | B6 + Marilyn T cells → female B6D2F1 | Fed water ad libitum from 14 d before IECT throughout the experiment | ↓ Commensal microbiota | ↓ Differentially expanded T-cell clones | 87 |
| Ciprofloxacin, 50 mg/kg per day | pmel-1 T cells (melanoma-targeting TCR T) → B6 with melanoma | Fed water ad libitum for 2 wk from 2 d before IECT | ↓ Microbial translocation in the mesenteric lymph nodes | ↑ Tumor size | 88 |
| Ciprofloxacin, 0.15 g/L Gentamicin, 0.2 g/L Bacitracin, 1 g/L Streptomycin, 2 g/L | HA-targeting CD4+ T cell → BALB/c with CT26HA tumors | Fed water ad libitum from 29 d before IECT throughout the experiment | Not mentioned | ↑ Tumor size | 84 |
| Polymyxin B, 1 mg/kg per day | pmel-1 T cells (melanoma-targeting TCR T) → B6 with melanoma | Fed water ad libitum for the duration of the experiment | Block the biological effect of gram-negative LPS | ↑ Tumor size | 88 |
| Antibiotics | Model | Method | Effects on microbiome | Outcomes | Ref |
|---|---|---|---|---|---|
| Ciprofloxacin, 0.15 g/L Gentamicin, 0.2 g/L Bacitracin, 1 g/L Streptomycin, 2 g/L | CD19-CAR T cell → BALB/c with CD19+ lymphoma | Fed water ad libitum from 29 d before IECT throughout the experiment | Not mentioned | ↓ Spleen B-cell recovery ↑ Spleen CAR T-cell persistence | 84 |
| Vancomycin, 0.5 g/L | CD19-CAR T cell → BALB/c or B6 with CD19+ tumor cells | Fed water ad libitum for the duration of the experiment | ↓ α-diversity ↓ Ruminococcaceae ↓ Lachnospiraceae | ↓ Tumor size ↑ Spleen and tumor-infiltrating T cells | 85 |
| Vancomycin, 1 g/L Metronidazole, 1 g/L Cefoxitin, 1 g/L Gentamicin, 1 g/L | Marilyn T cells (Y chromosome–targeting TCR T) → male B6 | Fed water ad libitum for 2 wk before IECT | Gut decontamination | ↓ Marilyn T-cell infiltration in the gut ↓ GVHD | 86 |
| Vancomycin, 1 g/L Metronidazole, 1 g/L Cefoxitin, 1 g/L Gentamicin, 1 g/L | B6 + Marilyn T cells → female B6D2F1 | Fed water ad libitum from 14 d before IECT throughout the experiment | ↓ Commensal microbiota | ↓ Differentially expanded T-cell clones | 87 |
| Ciprofloxacin, 50 mg/kg per day | pmel-1 T cells (melanoma-targeting TCR T) → B6 with melanoma | Fed water ad libitum for 2 wk from 2 d before IECT | ↓ Microbial translocation in the mesenteric lymph nodes | ↑ Tumor size | 88 |
| Ciprofloxacin, 0.15 g/L Gentamicin, 0.2 g/L Bacitracin, 1 g/L Streptomycin, 2 g/L | HA-targeting CD4+ T cell → BALB/c with CT26HA tumors | Fed water ad libitum from 29 d before IECT throughout the experiment | Not mentioned | ↑ Tumor size | 84 |
| Polymyxin B, 1 mg/kg per day | pmel-1 T cells (melanoma-targeting TCR T) → B6 with melanoma | Fed water ad libitum for the duration of the experiment | Block the biological effect of gram-negative LPS | ↑ Tumor size | 88 |
↑ represents increase and ↓ decrease.
LPS, lipopolysaccharides; Ref, reference.
In preclinical models, exposure to gut-decontaminating antibiotics has been linked to decreased MHC class II expression on intestinal epithelial cells and T-cell infiltration in the gut.86 This resulted in better GVHD mitigation. The beneficial effect of lessening GVHD may also be attributed to limited T-cell clonotype expansion.87
However, antibiotic exposure seems to negatively affect the antitumor effect of TCRs. Exposure to ciprofloxacin, with or without gentamicin, bacitracin, and streptomycin, was associated with larger tumor sizes in melanoma and colorectal carcinoma models.84,88 This could be due to antibiotic-induced decreases in lipopolysaccharides. Notably, the administration of polymyxin B, a cyclic cationic polypeptide antibiotic that selectively inhibits the biological effects of lipopolysaccharides, resulted in increased tumor sizes in the melanoma model.88
Impact of antibiotics on CAR-T clinical outcomes
Antibiotic exposure before CD19 CAR-T has been associated with poor outcomes.90,91 Smith et al conducted a review of 228 patients receiving CD19 CAR-T for non-Hodgkin lymphoma (NHL) or acute lymphoblastic leukemia at 2 US institutions. Antibiotics were used within 4 weeks before CAR-T in 60% of the cohort and were associated with worse overall survival (OS; HR, 1.71; 95% CI, 1.12-2.59; P = .011). The impact was more pronounced when considering exposure to piperacillin/tazobactam, imipenem/cilastatin, and meropenem (P-I-M), which were selected for their ability to target anaerobes and their use in neutropenic fever (OS; HR, 2.58; 95% CI, 1.68-3.98; P ≤ .001). This result remained significant in multivariate analysis. P-I-M exposure was associated with decreased progression-free survival (PFS) and OS in both NHL (PFS [HR, 1.83; 95% CI, 1.03-3.27; P = .038]; OS [HR, 3.37; 95% CI, 1.77-6.44; P ≤ .001]) and acute lymphoblastic leukemia (PFS [HR, 1.96; 95% CI, 1.15-3.35; P = .012]; OS [HR, 2.12; 95% CI, 1.2-3.76; P = .008]) cohorts. P-I-M exposure before CAR-T was also associated with increased immune cell–associated neurotoxicity syndrome (ICANS; P= .023) but not cytokine release syndrome (CRS; P = .058).90
Stein-Thoeringer et al reviewed 172 patients receiving CD19 CAR-T for NHL at 5 institutions across Germany and the United States with similar results of decreased PFS (HR, 2.04; 95% CI, 1.38-3.00; P= .0009) and OS (HR, 2.39; 95% CI, 1.46-3.91; P= .0027). They identified meropenem, cefepime, ceftazidime, and piperacillin/tazobactam as “high-risk” antibiotics, given the strength of association with progression. Patients who received these antibiotics had higher incidence of progression (odds radio, 2.60; 95% CI, 1.15-5.90; P ≤ .0001) and a decrease in PFS and OS. Patients receiving “low-risk” (ie, non “high-risk”) antibiotics experienced a similar rate of progression as those with no antibiotic exposure. In multivariate analysis, “high-risk” antibiotic exposure remained significantly associated with worse PFS but not OS. Exposure to “high-risk” antibiotics was significantly associated with ICANS but not CRS. Peripheral CD4 T cells, effector/effector-memory CD4 T cells, and CD8 T cells were significantly lower in those exposed to “high-risk” antibiotics, which may suggest impaired quality of the T-cell product. ML models were created from this data set to predict outcomes from baseline microbiome features. Notably, predictions from models were poor, when the model was trained on cohorts including those exposed to “high-risk” antibiotics. The predictability of the probabilistic and deterministic models was markedly improved when the “high-risk” antibiotic-exposed patients were excluded.91 These data highlight the substantial and unpredictable influence of “high-risk” antibiotic exposure on CAR-T results, making it a factor to consider for future initiatives aiming to use the microbiome as a predictive biomarker.
Both studies noted a significant association between antibiotic exposure and the occurrence of ICANS but not CRS, which is an area worthy of future investigation. Preclinical data suggest antibiotic-mediated depletion of gut microbiota disrupts the BBB and facilitates central nervous system infiltration of immune cells, which theoretically could increase the risk of ICANS.14 The impact of antibiotics on cytokine levels appears variable. One study noted significantly increased levels of proinflammatory cytokines associated with antibiotic courses over 14 days,92 whereas another noted an association between some antibiotics and downregulated cytokine levels.83 The impact of antibiotics on cytokine profiles and, in turn, CRS occurrence and severity likely depends on the antibiotic type and duration and should be further defined.
Impact of antibiotics on the microbiome of patients receiving CAR-T
Similar to HCT, patients undergoing CAR-T show baseline dysbiosis, characterized by reduced alpha diversity and altered compositions.90 Alpha diversity declined through the treatment course, with significant decreases associated with exposure to “high-risk” antibiotics.91 One study found that higher alpha diversity before CAR-T was positively associated with a 100-day complete response (CR). In contrast, another study found no significant association between baseline alpha diversity and day +180 response. Both noted no significant association between baseline alpha diversity and CRS or ICANS.90,91
Gut microbiome compositions varied depending on antibiotic exposure (Table 3). Those exposed to “high-risk” antibiotics were more likely to have an abundance of Prevotella, Veillonella, or Enterococcus spp., with domination of lactose and galactose degradation and peptidoglycan biosynthesis pathways. Roseburia, Bifidobacterium, Lactobacillus, and Eubacterium spp. were predominant in the nonexposed or “low-risk” antibiotic-exposed cohorts.91 Smith et al identified a higher relative abundance of obligate anaerobes, specifically Akkermansia, Bacteroides, Ruminococcus, and Faecalibacterium, among responders. The latter 2 genera were associated with a lack of CRS or ICANS.90 Stein-Thoeringer et al noted that species, pathway, and gene variability were weakly associated with survival and toxicity outcomes among the entire cohort; yet, when excluding those exposed to “high-risk” antibiotics, the impact of variance was stronger. Bifidobacterium longum, Eubacterium eligens, and Parabacteroides merdae were more abundant in responders, although there was no significant species association with CR.91 Metagenomic analysis results were distinct between the 2 studies. In one, the peptidoglycan biosynthesis pathway showed enrichment among responders, whereas in the other, this pathway was more prevalent in patients with early disease progression or mortality.90,91 The nonoxidative branch of the pentose phosphate pathway, which produces molecules involved in tryptophan metabolism, was enriched among those who experienced toxicity.90
Clinical data: microbiome characteristics and IECT outcomes
| Disease | CAR product | Level | Bacteria | Sample size | Country | Outcome or association | Ref |
|---|---|---|---|---|---|---|---|
| ALL NHL | 1928z Axi-cel Tisa-cel CTL019 Brexu-cel | Phylum Class Order Family Genera Species | Bacteroidetes Clostridia Bacteroidia Bacteroidiales Rumminococcaceae Bacteroidaceae Bacteroides Ruminococcus Faecalibacterium Akkermansia Faecalibacterium prausnitzii Ruminococcus bromii | 48 | United States | D 100 CR | 90 |
| Class Genera Species | Clostridia Ruminococcus Faecalibacterium Faecalibacterium prausnitzii | No toxicity | |||||
| NHL | Axi-cel Tisa-cel Liso-cel | Genera | Roseburia Bifidobacterium Lactobacillus Eubacterium | 116 | United States Germany | Non- or “low-risk” antibiotic exposure∗ | 91 |
| Species | Bifidobacterium longum Eubacterium eligens Parabacteroides merdae | Alive at 6 mo | |||||
| Species | Bacteroides eggerthii Ruminococcus lactaris Eubacterium spp. CAG 180 Akkermansia muciniphila Erysipelatoclostridium ramosum | CR based on deterministic ML model | |||||
| Species | Lachnospira pectinoschiza Akkermansia muciniphila | Higher CD3 and CD4 T-cell levels | |||||
| MM | BCMA CAR-T | Genera | Bifidobacterium Prevotella Sutterella Oscillospira Paraprevotella Collinsella Faecalibacterium Roseburia Ruminococcus | 43 | China | CR | 93 |
| NHL | Not specified | Genera | Faecalibacterium Bifidobacterium Ruminococcus | 12 | |||
| ALL NHL | 1928z Axi-cel Tisa-cel CTL019 Brexu-cel | Order Family | Veillonellales Veillonellaceae | 48 | United States | D 100 non-CR | 90 |
| NHL | Axi-cel Tisa-cel Liso-cel | Species | Prevotella Veillonella Enterococcus | 116 | United States Germany | “High-risk” antibiotic exposure∗ | 91 |
| Species | Bacteroides stercoris | Nonresponse based on deterministic ML model | |||||
| Species | Bacteroides uniformis Bacteroides ovatus Blautia spp. Faecalibacterium prausnitzii | Lower CD3 and CD8 T-cell levels | |||||
| ALL | Not specified | Genera | Bifidobacterium Roseburia Collinsella | 23 | China | Nonresponse | 93 |
| MM | BCMA CAR-T | Genera | Bifidobacterium Leuconostoc | 43 | Severe CRS |
| Disease | CAR product | Level | Bacteria | Sample size | Country | Outcome or association | Ref |
|---|---|---|---|---|---|---|---|
| ALL NHL | 1928z Axi-cel Tisa-cel CTL019 Brexu-cel | Phylum Class Order Family Genera Species | Bacteroidetes Clostridia Bacteroidia Bacteroidiales Rumminococcaceae Bacteroidaceae Bacteroides Ruminococcus Faecalibacterium Akkermansia Faecalibacterium prausnitzii Ruminococcus bromii | 48 | United States | D 100 CR | 90 |
| Class Genera Species | Clostridia Ruminococcus Faecalibacterium Faecalibacterium prausnitzii | No toxicity | |||||
| NHL | Axi-cel Tisa-cel Liso-cel | Genera | Roseburia Bifidobacterium Lactobacillus Eubacterium | 116 | United States Germany | Non- or “low-risk” antibiotic exposure∗ | 91 |
| Species | Bifidobacterium longum Eubacterium eligens Parabacteroides merdae | Alive at 6 mo | |||||
| Species | Bacteroides eggerthii Ruminococcus lactaris Eubacterium spp. CAG 180 Akkermansia muciniphila Erysipelatoclostridium ramosum | CR based on deterministic ML model | |||||
| Species | Lachnospira pectinoschiza Akkermansia muciniphila | Higher CD3 and CD4 T-cell levels | |||||
| MM | BCMA CAR-T | Genera | Bifidobacterium Prevotella Sutterella Oscillospira Paraprevotella Collinsella Faecalibacterium Roseburia Ruminococcus | 43 | China | CR | 93 |
| NHL | Not specified | Genera | Faecalibacterium Bifidobacterium Ruminococcus | 12 | |||
| ALL NHL | 1928z Axi-cel Tisa-cel CTL019 Brexu-cel | Order Family | Veillonellales Veillonellaceae | 48 | United States | D 100 non-CR | 90 |
| NHL | Axi-cel Tisa-cel Liso-cel | Species | Prevotella Veillonella Enterococcus | 116 | United States Germany | “High-risk” antibiotic exposure∗ | 91 |
| Species | Bacteroides stercoris | Nonresponse based on deterministic ML model | |||||
| Species | Bacteroides uniformis Bacteroides ovatus Blautia spp. Faecalibacterium prausnitzii | Lower CD3 and CD8 T-cell levels | |||||
| ALL | Not specified | Genera | Bifidobacterium Roseburia Collinsella | 23 | China | Nonresponse | 93 |
| MM | BCMA CAR-T | Genera | Bifidobacterium Leuconostoc | 43 | Severe CRS |
ALL, acute lymphoblastic leukemia; axi-cel; axicabtagene ciloleucel; brexu-cel, brexucabtagene autoleucel; liso-cel, lisocabtagene maraleucel; MM, multiple myeloma; tisa-cel, tisagenlecleucel.
“High-risk” antibiotics defined by study as meropenem, cefepime, ceftazidime, and piperacillin-tazobactam. “Low-risk” antibiotic exposure defined by study as exposure to any antibiotic not in the “high-risk” group.91
Hu et al studied longitudinally collected stool samples during B-cell maturation antigen CAR-T.93 Diversity decreased after CAR-T, and differences in operational taxonomic unit level bacterial abundance were more pronounced when comparing samples during/after CRS with those before CAR-T. Increases in Enterococcus, Lactobacillus, and Actinomyces and decreases in Bifidobacterium and Lachnospira were noted over time. When comparing those with CR vs partial response (PR), the cohorts had similar baseline alpha diversity, but those in the PR group experienced a more significant decline after CAR-T. There was a significantly higher operational taxonomic unit enrichment at peak CRS in the CR vs PR groups. Prevotella, Collinsella, Bifidobacterium, and Sutterella were significantly enriched in the CR vs PR cohort at baseline and after CAR-T infusion by multiple means of analysis. Bifidobacterium was significantly more enriched in those experiencing severe CRS compared with mild CRS.93
IECT future directions
Data on the effects of antibiotics and the microbiome in IECTs are limited. As the use of IECTs increases, it will be essential to assess these interactions. Mechanistic studies are warranted to understand why the changes in diversity and microbiome composition are associated with specific outcomes.
The microbiome is of interest as both a source of predictive biomarkers and as a modifiable therapeutic target in IECTs. Among a cohort of patients receiving CAR-T not exposed to “high-risk” antibiotics, a logistic regression-based ML model was created and validated across international cohorts with Bacteroides, Ruminococcus, Eubacterium, and Akkermansia spp. identified as predictors of CAR-T response. Notably, these models had no predictive value when including patients exposed to “high-risk” antibiotics.91 To accurately train predictive models, gathering more data on how antibiotics affect microbiome and the resulting outcomes in this population is essential.
Therapeutic alteration of the microbiome to improve IECT responses and tolerability is an area of interest for future investigation. Much of this is driven by data from the immunotherapy and HCT setting, in which alterations of gut microbiome via diet, medications, prebiotics and probiotics, and FMT have shown promise.6 Notably, a clinical trial will investigate the impact of β-hydroxybutyrate supplementation on the microbiome of patients with lymphoma receiving CD19 CAR T cells,94,95 because β-hydroxybutyrate has shown some antitumor properties in preclinical studies.96,97
Considering the link between antibiotic use and poorer CAR-T results, antibiotic stewardship is a potential clinical intervention. Antibiotic prophylaxis practices differ during CAR-T. Cytopenias, arising from lymphodepleting chemotherapy, and, in some cases, baseline myelosuppression from heavily pretreated patients, as well as immunosuppressants used for the treatment of CRS and ICANS, predispose patients to infectious complications.98,99 Improved risk stratification could help inform prophylaxis in IECT patients. For example, one study found that patients with CAR-HEMATOTOX scores of ≥2 experienced a significantly reduced likelihood of severe bacterial infections from fluoroquinolone prophylaxis.100 Perhaps not every IECT patient requires prophylactic antibiotics, although further validation of risk stratification is needed. Additionally, antibiotic choice in the treatment of neutropenic fever varies.99 The specific antibiotics that most significantly affect the microbiome are not clearly defined. For example, among the studies investigating the impact of antibiotics on the microbiome and CD19 CAR-T outcomes, one classified cefepime as a high-risk antibiotic, whereas others did not.90,91,101 Nonetheless, further studies are needed to inform antibiotic choice and duration to balance the risk of dysbiosis with the need for infection prophylaxis and treatment in the IECT setting.
Authorship
Contribution: A.D. and J.X. contributed to the writing of the manuscript; J.X. and M.S. contributed significant edits to the manuscript; M.S. was responsible for conception of the manuscript; and all authors agreed to the contents of the manuscript before submission.
Conflict-of-interest disclosure: M.S. reports consultancy for CVS Caremark; advisory for A28 Therapeutics; and patents pertaining to the intestinal microbiome and chimeric antigen receptor T-cell therapy (US 17/229184; US 63/303461). The remaining authors declare no competing financial interests.
Correspondence: Melody Smith, Stanford University, CCSR Building, Room 2255, 269 W Campus Dr, Stanford, CA 94305-5170; email: melodysm@stanford.edu.
