Restrictive Versus Standard Antibiotic Treatment During CAR-T Therapy

CAR-T-cell therapy is associated with substantial exposure to antibacterial agents because patients frequently develop fever, neutropenia, and suspected or confirmed infections during the peri-infusion period. Broad-spectrum antibacterial agents, particularly those with extended anaerobic activity, may cause marked disruption of the intestinal microbiota. Retrospective studies have associated peri-CAR-T antibiotic exposure and reduced gut microbial diversity with inferior treatment response, increased toxicity, and shorter survival. However, these associations may be affected by confounding related to infection severity, disease status, and other clinical factors. Prospective randomized evidence is therefore needed to determine whether a microbiome-sparing antibacterial strategy can preserve gut microbial diversity while remaining clinically feasible and safe.

In the restrictive-strategy group, intravenous ceftazidime will be used as initial empirical antibacterial treatment when antibacterial therapy is clinically indicated. Oral ciprofloxacin may be used as protocol-defined step-down treatment after adequate infection control and clinical stabilization. Ciprofloxacin may also be used as antibacterial prophylaxis when prophylaxis is clinically indicated and there is no suspected ongoing infection. Ciprofloxacin will not be used as initial empirical monotherapy for suspected infection.

A switch from intravenous ceftazidime to oral ciprofloxacin may be considered when the participant is clinically stable; fever and other clinical signs of infection have resolved or clearly improved; no microbiologically documented infection or suspected infectious focus requires alternative or additional coverage; any identified pathogen is susceptible to ciprofloxacin; oral absorption is considered reliable; and there is no known colonization or previous infection with a fluoroquinolone-resistant pathogen that would make ciprofloxacin inappropriate.

Participant safety takes priority over adherence to the restrictive strategy. The treating physician may modify or broaden antibacterial treatment at any time in response to clinical deterioration, persistent or recurrent fever, microbiological findings, suspected resistance, organ-specific infection, or a need for additional Gram-positive, anaerobic, or other antimicrobial coverage. All decisions to modify or escalate treatment and the reasons for doing so will be recorded. In the standard-of-care group, antibacterial prophylaxis and treatment will be selected according to local clinical practice without protocol-defined restrictions.

The biological effects of the two strategies will be characterized through longitudinal collection of stool and blood samples. Samples will be obtained at randomization, around the time of CAR-T-cell infusion, at day 28 after infusion, and at the end-of-treatment evaluation approximately 3-6 months after infusion. Stool samples will be used to assess microbial diversity, community composition, and the relative abundance of selected bacterial taxa. Blood and stool samples will also be used for exploratory metabolomic analyses, including microbiota-associated metabolites such as short-chain fatty acids, tryptophan-derived metabolites, secondary bile acids, and other immunomodulatory metabolites, depending on assay coverage. Blood proteomic profiles will be explored in relation to microbiota changes, antibacterial exposure, toxicity, and clinical outcomes.

Longitudinal microbiota analyses will include measures of within-sample diversity and between-sample community composition. Repeated measurements will be evaluated using appropriate mixed-effects models incorporating treatment strategy, sampling time, treatment-by-time interaction, and disease cohort. Community-level differences may be assessed using ordination methods and permutational multivariate analysis of variance. Differential abundance analyses will use appropriate multivariable microbiome methods with adjustment for multiple testing.

Analyses will primarily compare participants according to their randomized treatment strategy. Complementary analyses will characterize actual antibacterial exposure, including total antibacterial days of therapy, timing of exposure in relation to CAR-T-cell infusion, and duration of exposure to broad-spectrum agents with extended anaerobic activity. Exploratory exposure groups may include participants receiving no antibacterial treatment, participants receiving ceftazidime and/or ciprofloxacin only, and participants receiving other standard-of-care antibacterial regimens. Where appropriate, actual antibacterial exposure may also be analyzed as a time-dependent variable.

Potentially relevant concomitant treatments, including proton pump inhibitors, statins, corticosteroids, and tocilizumab, will be considered in adjusted or sensitivity analyses. Exploratory analyses will assess whether microbiota and metabolomic patterns differ between participants with large B-cell lymphoma and multiple myeloma. Data may also be compared with external microbiota datasets from healthy individuals and patients with newly diagnosed, treatment-naive large B-cell lymphoma.

Infectious events and antibacterial exposure will be evaluated during predefined periods before and after CAR-T-cell infusion. CAR-T-related toxicities will be assessed using established grading systems for cytokine release syndrome, immune effector cell-associated neurotoxicity syndrome, and immune effector cell-associated hematotoxicity. Participants will be followed for clinical outcomes for 24 months, after which the final study analysis is planned.