Gut Microbiome and CAR T-Cell Therapy: Impact on Response, Toxicity, and Outcomes

Pharmacy Times interviews Robert R. Jeng, MD, about the emerging role of the gut microbiome in shaping outcomes of chimeric antigen receptor (CAR) T-cell therapy. He discusses current evidence linking microbial composition and diversity to treatment response, as well as how the microbiome may influence CAR T-cell expansion, persistence, and exhaustion. The conversation also explores challenges posed by lymphodepleting chemotherapy and antibiotic use, which can disrupt the microbiome prior to infusion. Jeng highlights the potential of microbiome-based interventions, including fecal microbiota transplant, while addressing safety and clinical trial considerations. He also examines whether microbiome modulation could impact CAR T–related toxicities such as cytokine release syndrome and neurotoxicity.

Pharmacy Times: What does the current evidence show about the link between gut microbiome composition and CAR T-cell therapy outcomes, and which specific microbes or diversity measures seem most predictive of response?

Robert R. Jeng, MD: The evidence, I would say, is not bad. There have been multiple studies that were done by…different groups that have had similar findings, and the studies are pretty large in size,…numbering on the order of 500-plus patients total across the different studies and spanning multiple institutions in the US and Europe,…so I think it’s pretty robust. [One important] finding is that antibiotic exposure, particularly…the stronger antibiotics that are more active against anaerobic bacteria,…prior to CAR T-cell therapy, is associated with a less than ideal response. So these patients that needed an infection treatment with antibiotics…tend to not do as well with the CAR T-cell therapy, and that seems to be holding true for a variety of different cancers. Lymphoma and all leukemia have been studied the best, but newer studies in the myeloma population have shown pretty similar findings. I would say that that antibiotic association is the one that is most robust…most reproducible in terms of…particular bacteria, particular microbes, or like a microbial signature that really predicts…for a response that’s been a little bit trickier to reproduce.

And that’s also been the case in other arenas, like in the cancer immunotherapy context…where people were looking for microbiome signatures that predicted response to, say,…PD-1 blockade, or something like that. It was also trickier to find that it’s…I think probably some of that is that the names of the bacteria aren’t really the underlying cause. It’s not the crux of what's going on…. We’re not smart enough yet to know exactly what the bacteria are doing. If we knew that, then we would have…a bacterial product, or something like a metabolite, or some other bacterial part, like a cell wall part, or something that really nicely explained which patients have the correct microbiome profile and which ones didn’t. So that’s my…guess in terms of…why we don’t have a great microbial signature that’s reproducible.

Pharmacy Times: How might the microbiome influence CAR T-cell expansion, persistence, and exhaustion in vivo, and what immunological pathways are most relevant here?

Jeng: I think we and other groups are actively trying to do this. What hasn’t…been done really well yet is a good analysis of the CAR T cells themselves in the patients. And you know these things that you’re mentioning,…like persistence and exhaustion, those have been identified by the CAR T-cell field as strong predictors of who does well with CAR T cells and who doesn’t. But they’re not, for…whatever reason,…routinely collected in every single patient that undergoes CAR T-cell therapy. It’s still kind of like a research question, and so I guess…the papers that are out there currently are primarily…first-wave papers, where people collected stool samples and antibiotic exposure data for the first time in every patient that was undergoing a CAR T-cell product or procedure at the institution. And so we were just…collecting stool samples and seeing how patients did, and lo and behold, we saw this interesting association. And now we have to do…the second-wave studies, where we can collect all the research-type questions, like the samples, where we do quantify CAR T cells, and how well they are expanding. What’s their phenotype like? Are they happy, or are they exhausted? That sort of thing.

Pharmacy Times: Prior to CAR T infusion, many patients receive lymphodepleting chemotherapy and broad-spectrum antibiotics. How does this standard conditioning regimen complicate our ability to harness the microbiome therapeutically?

Jeng: The patients that come to CAR T-cell therapy [have] undergone other types of therapies prior, right? So the [patients with] leukemia have had induction chemotherapy, [those with] lymphoma have had first-line chemotherapy, and even the [patients with] myeloma, [while] they may not have had…the toughest chemotherapy in terms of hurting the bone marrow,…they’ve usually undergone multiple rounds of treatment too. And there actually is…a sense amongst oncologists that CAR T-cell therapy [has] been held in reserve,…kind of like to salvage or to treat relapsed patients, but it might work better in the upfront setting.

So more and more, it seems like every few years [there is] a new study showing that if you do CAR T cells earlier, after patients are diagnosed, the outcomes are better. And it also means that the patients are coming into the CAR T-cell procedure with a better and better microbiome…over time, as you know, as these things have evolved, so it kind of makes it harder, in a weird sense, to intervene on the microbiome. I’ll give you an example. I mentioned earlier that the patients who’ve had antibiotics are probably the most likely to not do as well with the CAR 'T-cell therapy. So we’ve opened studies at clinical trials of trying to restore the microbiome in these patients, and we use that as an eligibility. So we say, “Oh, if you’re a patient,…your doctor wants to give you CAR T-cell therapy, [if] but you’ve been treated with these strong antibiotics prior, then you’re eligible to be on our study.”

We’re trying to enrich for the patients that are most likely to benefit, and those studies have actually been kind of hard to do because…[fewer] patients are getting treated with these really strong antibiotics. I think part of it is that the message has gotten out, and so the oncologists are trying to be mindful of the microbiome and not injure it unnecessarily. And the other thing that’s happened is that patients are being referred for the CAR T-cell procedure earlier in the course of their cancer treatment, so they’re coming in without necessarily [having had many rounds] of infections and antibiotic treatments. Which is a good thing, of course, but it means it’s a little bit harder to study. It’s a little harder to prove the point that fixing the microbiome can really mediate a benefit.

Pharmacy Times: If you were designing a clinical trial to evaluate a microbiome-based intervention, such as fecal microbiota transplant or a defined bacterial consortium, in CAR T-cell recipients, what would your primary end point be, and how would you address safety concerns?

Jeng: The primary end point that we’d be most interested in would be response rates. The microbiome association with toxicities like CRS [cytokine release syndrome] or the neurotoxicities, [those] data are not the strongest. There have been some papers that show an association, but the response association has really been the strongest, I would say. So probably what’s happening is some of these toxicities are…maybe microbiome modulated, but it’s not the primary driver, whereas...the microbiome really does seem to be really important for…an optimal response to CAR T-cell therapy. So response, I think, would be the primary outcome that we’d be interested in.

But then, that said, usually when we do these types of interventional studies, they’re small phase 1 studies, like single-center phase 1 studies, and they’re not powered; they’re not statistically powered for an outcome like response. And so what the biostatisticians tell us to do is that we should power it for some sort of surrogate that we are statistically powered for. So one thing that we tend to reach for [is] some sort of…microbiome profiling readout. So if you look at some of these early phase 2 studies where they’re intervening on the microbiome, of course they really care about response, as they want to, and they want to see if there’s a hint of an improvement in response. But the formal primary objective is usually something like microbiome diversity.

Pharmacy Times: Cytokine release syndrome and immune effector cell–associated neurotoxicity are major toxicities of CAR T-cell therapy. Do you see any potential for microbiome modulation to influence the severity of these toxicities, and if so, how?

Jeng: I’m not going to say that there aren’t studies that have found associations between the microbiome and these CAR T-cell toxicities, but the strength of those associations seems to be not as strong, and the response rate seems to be stronger.

Pharmacy Times: Beyond bacteria, what role might the virome, mycobiome, or microbiome-derived metabolites, such as short-chain fatty acids, play in modulating CAR T-cell efficacy, and how far away are we from translating these insights into routine clinical practice?

Jeng: I would say that there’s been a lot of active work, primarily in preclinical models, like in mouse models, to try to identify mechanisms behind how the microbiome could be modulating CAR T-cell therapies. And there are several hits. A lot of them have focused on bacterial products, like metabolites. So there [are] some studies looking at different short-chain fatty acids like valerate, for example, or nucleotides, like in a scene. And then there are other bacterial products too. So whenever this type of finding comes up, the question comes up, can you just give the bacterial product as a drug? Is it possible? Maybe you don’t need to restore the microbiome; you just have to restore what the microbiome is generating that leads to the nice CAR T-cell response rates. And I think it’s a really interesting question.

That said, I’m not aware of any drug companies, biotech companies, that have gone in that direction yet. Some of these products [are] naturally occurring metabolites, right? And they’re not necessarily really well thought out in terms of the delivery, the optimal delivery method, the pharmacokinetics. So short-chain fatty acids, for example…have a really short half-life in the serum. They’re cleared by the kidneys very rapidly. So it’d be really challenging to come up with a way to administer a short-chain fatty acid; you'd have to basically just do a continuous drip or something like that in the micro. When it’s produced by the microbiome, the bacteria are producing these metabolites…continuously around the clock. And so it’s kind of interesting to think about how you can replicate what the bacteria are doing pharmacokinetically. I think that’s a really interesting area. It’d be sort of ripe for, you know, a smart chemist to develop some sort of analogue…that has different [or] better pharmacokinetic properties, like a longer half-life and that sort of thing, but [that] would still hit the same target. I think a short-chain fatty acid analogue would be very interesting.

It’s still early days. I think if you need antibiotics and your doctors want to give you antibiotics, you shouldn’t fight back. I think antibiotics are important, and it’s not the end of the world if you get treated with antibiotics. There are centers that have clinical trials that are studying strategies to improve the microbiome before the CAR T-cell therapy, and in general,…healthy eating, eating a lot of variety of foods, especially a variety of different fruits and vegetables, can have a beneficial effect in terms of restoring your microbiome. There’s a lot of growing evidence for that.