This year at ASH our group has presented two abstracts on a topic that we think is very interesting from the point of view of the translational impact. This is the role of the gut microbiota in patients receiving CAR T-cell immunotherapy. So we have one abstract that was presented by Dr Andrea Facciabene as an oral presentation at ASH and is more focussed on the preclinical modelling of the role of the gut microbiota in CAR T-cell immunotherapy and a more clinical abstract that is a collaboration with the Memorial Sloan Kettering Cancer Center that was presented by Dr Melody Smith, again as an oral presentation.

The two abstracts focus on the same topic, in other words what is the role of bacteria that patients have in their gut and how those bacteria in the gut can affect the outcomes of CAR T-cell immunotherapy. Recent work in the last five years has shown that bacteria in the gut can actually affect the outcomes of another type of immunotherapy that is called checkpoint immunotherapy so we wanted to ask the same question in the setting of CAR T-cells.

Starting with the preclinical abstract, in that work we used a model of lymphoma in mice and we treated those mice with CAR T-cells and we modulated the composition of the bacteria in the gut using an antibiotic. As you know, antibiotics can affect the survival of bacteria, that’s why they are used for infections. So by giving an antibiotic that is called vancomycin that is specifically active on gram positive bacteria and is not absorbed, so if we give it orally to mice it is just going to influence the bacteria in the gut and it doesn’t have a systemic effect. What we noticed with the group of Dr Andrea Facciabene is that the mice that were receiving CAR T-cell immunotherapy plus this antibiotic called vancomycin were actually having a higher anti-tumour effect as compared to the mice that were receiving the immunotherapy alone, indicating the modulation of the gut microbiota might affect the outcome.

Interestingly, what we have seen is that it looks like the mechanism by which this antibiotic enhances the effect of CAR T-cell therapy is that basically it increases the number of antigens that the CAR T-cell can recognise in cancer cells. So not only through the CAR specific antigen that could be, for example, CD19, but also through the T-cell receptor that can recognise other antigens that are expressed in the cancer cells.

So this preclinical work obviously is interesting but of course we wanted to have a sense of what happened in actual patients. So with the group at Memorial Sloan Kettering we analysed two cohorts of patients, one cohort was a retrospective cohort of patients with acute lymphoblastic leukaemia and with non-Hodgkin’s lymphoma. These were all patients receiving CAR-19 immunotherapy, so either 4-1BB costimulated CAR-19 or CD28 costimulated CAR-19, in clinical trials and in the commercial setting.  In these patients we looked at the antibiotic use in the four weeks before CAR T-cell infusion and the question was can the use of these antibiotics in the four weeks before CAR infusion affect the outcomes of CAR T-cells.

It was very interesting because when we looked at the antibiotic use first of all we saw that over 70% of patients were receiving some sort of antibiotic. So this means that these patients really have a strong need for antibiotics before CAR-19 therapy. But most interesting is that exposure to any antibiotic, so if you look at patients receiving any antibiotic versus patients receiving no antibiotic, actually patients receiving antibiotics had a poorer outcome. So they had a poorer progression free survival, poorer overall survival.

In particular, when we look at the different types of antibiotics there was a group of antibiotics that we call PIM, this is piperacillin-tazobactam, imipenem and meropenem, these are broad spectrum antibiotics with a specific activity on anaerobes that in the past have been shown to be really affecting the gut microbiota. So those antibiotics that again we call PIM were very much associated with poorer outcomes, so low progression free survival, low overall survival in both acute lymphoblastic leukaemia and lymphoma.

Interestingly, in patients with non-Hodgkin lymphoma we also saw increased neurotoxicity so this one of the common side effects of CAR Ts.

This was the first cohort that, of course, is retrospective but was suggesting that dysfunctional microbiota can be associated with different outcomes. Of course, we also ran multivariate analyses to make sure that the groups receiving antibiotics and not receiving antibiotics were actually comparable.

The second cohort was a much smaller cohort – the first cohort was about 200 patients, the second cohort about 50 patients – is a prospective cohort where we collected the stools at baseline and we analysed the gut microbiota in there using 16S sequencing or shotgun metagenome. Interestingly, what we saw is that, first of all, patients receiving CAR T-cell therapy had different composition of the gut microbiota as compared to healthy individuals. This can be due to the disease, this can be due to the treatment, many reasons. But, most importantly, we found specific taxa of bacteria that were associated with different outcomes. For example, we saw that Ruminococcus was actually associated with a higher complete response in patients receiving CAR T-cell therapy. We also identified many other genera of bacteria that can potentially be associated with better outcome.

There were also other strains of bacteria that were linked with the presence of toxicity. So, again, this is just to underlie the fact that modulation of the gut microbiota that could be either voluntary if you use specific antibiotics or probiotics or prebiotics and so on, or just happens because of the treatments and so on can actually have an effect on the outcomes of CAR T-cell immunotherapy. So obviously these studies show for the first time that that is the case but we will need to have larger studies to identify for sure what are the specific strains, if any, that are associated with differential outcomes or if it’s just the diversity that is affecting these outcomes. Once we learn more about that we can really start crafting, potentially, treatment so either using antibiotics before CAR T-cell or during or modulating with faecal material transfer and so on.

So this is the work that we presented at ASH and, again, I wanted to stress this is a collaboration between my group, the big group at Penn that is headed by Carl June, the laboratory of Dr Facciabene at Penn and for the clinical part the collaboration with Dr Marcel van den Brink at Memorial Sloan Kettering and Dr Melody Smith who is now an assistant professor at Stanford.

What we need to do next is to expand our cohort of patients for whom we have the stool samples to analyse, and we are actually doing that in collaboration with a larger multicentre group, not only looking at the stool samples at baseline but we also want to see over time how the microbiota evolves during CAR T-cell therapy. Of course, once we learn more about the mechanism, so what are the taxa and genera that are involved, the diversity and so on, we can look more at the mechanism. Once we identify more data on the mechanism, also using preclinical models, we can design a potential treatment, a potential strategy to modulate the gut microbiota in CAR T-cell patients.