Patients with certain blood cancers may respond differently to autologous CAR T-cell therapy depending on the makeup of their gut microbiome and whether they’ve received antibiotics in the weeks leading up to treatment, recent research suggests.

In a study published Monday in Nature Medicine, researchers from Memorial Sloan Kettering and the University of Pennsylvania described how patients with non-Hodgkin lymphoma (NHL) and acute lymphoblastic leukemia (ALL) experienced reduced survival outcomes and increased adverse events following autologous CD19-directed CAR T-cell therapy when they received antibiotics — specifically, broad-spectrum antibiotics — in the four weeks before treatment. They also found that patients with a higher abundance of specific bacteria species — namely Ruminococcus, Bacteroides, and Faecalibacterium — in their microbiomes were more likely to respond to CAR T-cell treatment.

“There’s a plethora of studies that have demonstrated that the use of antibiotics before or even during checkpoint blockade, for instance, can have a negative impact on outcomes,” said Marcel van den Brink, head of the hematologic malignancies division at MSK and co-senior author on the Nature Medicine study. “So that was the backdrop to this.”

To explore the role of antibiotic use and outcomes on CAR T-cell therapy, the researchers analyzed retrospective data from patients treated with both commercially available and investigational CD19-directed CAR T-cell therapies at MSK and UPenn. Specifically, the cohort included 137 NHL patients and 91 ALL patients who received either Novartis’ Kymriah (tisagenlecleucel), Gilead’s Yescarta (axicabtagene ciloleucel), or one of two investigational therapies that were very close in nature to the commercial products, according to Melody Smith, the study’s first author who previously worked in van den Brink’s lab at MSK but is now a hematologic oncologist at Stanford University School of Medicine.

Although about 60 percent of all the patients had received at least one antibiotic leading up to the cell therapy infusion, the specific antibiotics used varied. Specifically, 47 percent of all the patients analyzed had been treated with the obligate anaerobe-targeting broad spectrum antibiotics piperacillin/tazobactam, imipenem/cilastatin, and meropenem (collectively dubbed “P-I-M”), mainly to control neutropenic fever. These patients, relative to patients not treated with P-I-M, experienced significantly shorter overall survival following CAR T-cell infusion. Patients who received P-I-M were also more likely to experience serious adverse events following CAR T-cell therapy, including cytokine release syndrome and neurotoxicity.

“The clinical relevance could be that we need to be careful with the antibiotic stewardship,” van den Brink said, highlighting that the negative impact on CAR T-cell therapy outcomes was not as pronounced among patients on the study treated with other antibiotic regimens such as cefepime.

Indeed, as the researchers described in the Nature Medicine paper, cefepime does not target anaerobes to the same degree as P-I-M does.

Going forward, “it will be important to look at how [P-I-M] alters T cells mechanistically,” Smith said, adding that a randomized study that prospectively treats patients with P-I-Ms versus other types of antibiotics and assesses outcomes on CAR T-cell therapy could be valuable for validation, too. Researchers in van den Brink’s lab have already conducted this kind of research validating the difference among patients who received bone marrow transplants and are looking at doing the same in the CAR T-cell therapy setting.

“[These findings] help us to understand how we can look at the different sensitivities of antibiotics and … if there might be a different antibiotic we can use as compared to another one in light of this data,” Smith continued, noting that many patients undergo intensive chemotherapy regimens before they are referred for CAR T-cell therapy, making infection and unavoidable antibiotic treatment a big consideration.

“We should think broadly, not just before CAR T-cell therapy, but before all treatments, about the use of antibiotics,” van den Brink echoed, pointing out that the research also showed that patients entered the CAR T-cell therapy setting with already damaged microbiomes, likely because of the many rounds of gut-altering treatment they previously underwent.

Sequencing, future directions

Separate from their findings on broad-spectrum antibiotics and CAR T-cell therapy outcomes, van den Brink, Smith, and colleagues also looked into the specific makeup of patients’ microbiomes prior to CAR T-cell therapy. For this part of the study, they prospectively collected stool samples from 48 patients at MSK and UPenn and used taxonomic profiling with 16S ribosomal RNA gene sequencing and metagenomic shotgun sequencing to characterize the microbiome composition before CAR T-cell therapy infusion.

Pairing the sequencing findings with patients’ clinical outcomes — which, in this case, they defined as response to therapy at day 100 following infusion — they found that baseline samples with higher relative abundance of microbial taxa in the class Clostridia were more likely to respond to treatment.

These bacteria — including Ruminococcus and Faecalibacterium — were obligate anaerobes, which makes sense given the findings that the broad-spectrum antibiotics that targeted obligate anaerobes to a greater degree were associated with worse survival outcomes.

Homing in on the specific species of bacteria associated with treatment response could ultimately lead to interventions and improved treatment selection strategies going forward. This is already an active area for drug developers trying to improve immune checkpoint inhibitor response rates. Indeed, several startups and academic spinouts are well on their way to developing microbiome signatures to include in oral capsules, which could be administered alongside checkpoint inhibitors to improve response. Others are developing sequencing tests for gut microbiome signatures as companion diagnostics to identify which patients are best suited for checkpoint inhibitors.

In the context of autologous CAR T-cell therapy, more research is needed to identify a specific signature for this type of intervention. According to Smith and van den Brink, though, the possibility is already top of mind.

“The way that we think of the microbiome is as another organ,” said van den Brink. “And just as we do organ function checks of the hearts, lungs, or kidney before we do any kind of therapy, we should also do it for the gut microbiome.”

In the case of a gut microbiome that appears damaged on one of these initial “function checks” — which would likely involve metagenomic shotgun sequencing — he said, “we hope that we will eventually get to therapies that can fix a damaged flora before a therapy starts.”

Van den Brink’s lab is already working with a company dubbed Seres Therapeutics to evaluate whether a “cocktail” of bacteria in the form of an oral capsule could benefit patients when administered before allogeneic bone marrow transplant. “Something similar could be done for CAR T-cell therapy,” he mused, pausing to add that, while the bacteria-cocktail-in-a-capsule approach is attracting a lot of attention, we shouldn’t rule out pre- and postbiotic interventions, including the role that diet can play in improving microbiome composition.

This article originally ran on Precision Oncology News click here for more info.