New Model Predicts Long-term Response to CAR T-cell Therapy in B-cell Lymphoma


Investigators identified a set of microbiome-based biomarkers that may be used to optimize patient selection or increase personalization of treatment for patients with B-cell lymphoma receiving CAR T-cell therapy.

Image Credit: ©freshidea -

Image Credit: ©freshidea -

Investigators identified a set of microbiome-based biomarkers that may be used to optimize patient selection or increase personalization of treatment for patients with B-cell lymphoma receiving CAR T-cell therapy, according to findings from German and American investigators.

In findings published online in Nature Medicine in March 2023, investigators observed significant correlations between pre­–CAR T infusion Bifidobacterium longum and microbiome-encoded peptidoglycan biosynthesis, as wellCAR T treatment-associated 6-month survival or disease progression. Conserved microbiome features observed across clinical and geographical variations may enable physicians to make cross-cohort microbiome-based predictions for patients receiving CAR-T cell immunotherapy.

Investigators concluded that antibiotic exposure during the 3-week period prior to CAR-T cell infusion was associated with significantly decreased progression-free survival (PFS; HR, 2.04) in antibiotics-pretreated patients. This was true in both the US (HR, 2.02) and German (HR, 2.20) cohorts. In the US cohort, 41.5% of patients received antibiotics compared with 27.2% among the Germans.

Similarly, investigators found that antibiotic exposure was associated with a significantly higher cumulative incidence of progression (odds ratio, 1.81) and a significantly reduced overall survival (OS; HR, 2.39).

“Patients who had higher tumor burden, who had highly aggressive lymphoma, require more antibiotics and were admitted more into the hospitals because of their lymphoma,” study coauthor Neeraj Saini, MD, said in an interview with OncLive®. “Those antibiotics protected them from infection but damaged their microbiome, which we think leads to poor prognosis.”

Saini, an assistant professor in Department of Stem Cell Transplantation at The University of Texas MD Anderson Cancer Center, said that investigators have discovered that the gut microbiome plays an important role in multiple diseases as well as response to immune checkpoint therapies over the past decade.

Patients received a variety of antibiotics during the pre–CAR T-cell infusion period, with piperacillin–tazobactam, cefepime or meropenem the most common. Investigators determined that meropenem, cefepime, ceftazidime and piperacillin–tazobactam were “high-risk antibiotics” because they were associated with significantly reduced responses to CAR T-cell therapy. These agents are known to severely disrupt the gut microbiome by targeting many anaerobic commensals.

Data from previous studies have shown that high-risk antibiotic treatment administered within 3 weeks of CAR T-cell infusion is associated with increased disease progression and reduced OS for patients with B-cell lymphoma. Furthermore, evidence collected from human studies and preclinical experiments have shown that the gut microbiome and its effectors may be major biological variables effecting the efficacy of T cell-driven cancer immunotherapies and their toxicities. Gut microbes such as Akkermansia muciniphila, Faecalibacterium prausnitzii, and Bifidobacterium longum have been associated with improved responses to immune checkpoint blockade.

It is well established that despite high initial response rates, only 40% of patients achieve long-term remission following CAR T-cell therapy. Investigators have attributed loss of antitumor efficacy to CAR T-cell exhaustion in the tumor microenvironment or, to a lesser extent, immune evasion of malignant cells with loss of tumor surface antigen expression.

To understand the relationship between the microbiome and CAR T therapy effectiveness and adverse effects, investigators established an international consortium of German (n = 66) and US (n = 106) cohorts of patients with lymphoma who received CD19-targeted CAR T-cell immunotherapies. They then studied the role of the gut microbiome and its potential confounders in clinical outcomes of CAR T therapy.

The vast majority (94.2%) of the cohort had diffuse large B-cell lymphoma (DLBCL) or transformed follicular leukemia (TFL). Overall, 122 patients received axicabtagene ciloleucel (axi-cel; Yescarta), 49 received tisagenlecleucel (tisa-cel; Kymriah) and 1 received lisocabtagene maraleucel (liso-cel; Breyanzi). The median age of participants was 62.0 years (range, 24.0-88.0) and 117 patients were male. The number of median prior lines of systemic therapy was 3.

Ninety-seven (56.4%) patients had an ECOG score of 1, 52 (30.2%) had a score of 0, 18 (10.5%) had a score of 2, and 5 (2.9%) had a score of 3. Most patients (76.2%) had stage III/IV disease. More than 65% of patients received bridging therapy.

At day 90, 82 (47.7%) patients had complete response (CR) to CAR T-cell therapy. Seventeen (9.9%) had partial response (PR), 1 (0.6%) had stable disease (SD), and 53 (30.8%) had progressive disease (PD). Nineteen (11.1%) patients died.

At day 180, 72 (41.9%) patients had CR to CAR T-cell therapy. Eight (4.7%) had PR, and 56 (32.5%) had PD. Thirty-six (20.9%) patients died.

Of patients who experienced progression, 52.4% had early progression and 47.6% had late progression.

Patients who received high-risk antibiotics had significantly higher rates of disease progression (odds ratio, 2.60). Those treated with low-risk antibiotics, in contrast, had a progression incidence similar to that of patients who were not exposed to antibiotics prior to CAR T infusion. The exposed patients had significantly reduced PFS (HR, 3.05) and OS (HR, 2.99).

Investigators questioned whether exposure to antibiotics in the pre-infusion time window was associated with subsequent development of CAR T-cell-mediated toxicity because immune-related toxicities constitute important factors contributing to morbidity and mortality in CAR T-cell immunotherapy. They found that high-risk antibiotics were not associated with cytokine release syndrome (CRS), but there was a correlation between high-risk antibiotics and a significantly higher rate of immune effector cell–associated neurologic syndrome (ICANS) of any grade in the combined cohort (63.8%) compared with those not exposed to high-risk antibiotics (43.4%).

Microbome Predicts Response

Investigators next evaluated changes in the gut microbiome observed over the course of CAR T-cell therapy in the entire CAR T-treated cohort, the segregated high-risk antibiotics-treated, and the non-high-risk antibiotics-treated patients. They hoped to identify associations between microbiome features and major clinical and immunotherapy readouts, as well as links to high-risk antibiotics exposure.

They performed shotgun metagenome sequencing on 351 prospectively collected stool samples from 116 of 172 patients at all 5 CAR T-cell centers from days −62 to 208 relative to CAR T-cell infusion to investigate bacterial species composition and abundance of metabolic pathways encoded in the microbiome and its gene contents. They then analyzed the effects of various clinical covariates on the compositional variance of the pre-infusion gut microbiome in the combined patient cohort by testing the differences in beta diversity between groups based on Bray–Curtis distance. Finally, investigators further analyzed the associations between pre-infusion microbiome compositions and major clinical outcomes of CAR T-cell therapy by comparing the compositions of microbiomes between outcome groups.

When considering the differences in individual microbiome features between different outcomes, investigators found Bifidobacterium longum, Eubacterium eligens, and Parabacteroides merdae to be more abundant in patients alive at 6 months after infusion when including all samples. Bifidobacterium longum was also found to be more abundant after excluding high-risk antibiotic-treated patients.

The exclusion of patients exposed to high-risk antibiotics strengthened the associations between microbiome features and outcomes such as survival and progression. Multiple features, including Bifidobacterium longum and peptidoglycan biosynthesis, were strongly correlated with long-term survival or disease progression following CAR T-cell therapy independent of other demographic or clinical variables.

Studying the microbiomes of patients with non-antibiotic-disrupted baseline microbiomes, investigators developed a model that enabled prediction of long-term response to CAR T-cell therapy.

High levels of A. muciniphila (HR, 2.2) or R. lactaris (HR, 1.83) were associated with increased PFS (P = 0.12). In contrast, high Bacteroides stercoris counts were linked to decreased PFS.

Saini said that the gut microbiome was so damaged in patients who received high-risk antibiotics that it was difficult to identify good microbiome markers. They found that microbiome markers better correlated in patients who did not receive antibiotics.

“The field has to move away from patients who have been whose microbiome has been damaged from these antibiotics to predict accuracy and markers of outcomes,” he said. “Further, we observed what we usually see with other immunotherapies—bacteria such as Bacteroides longum and Akkermansia muciniphila, these Ruminococcus bacteria, were associated with good CAR T responses.”

Investigators noted that species predictive for immunotherapy response in their model were mostly anaerobes that are targeted by broad-spectrum antibiotics. They therefore assessed the difference in abundance of these species following response and antibiotic exposure stratification.

As expected, exposure to high-risk antibiotics was commonly associated with lower abundance of these species, similar to the non-response setting. Similarly, gut flora species predicting non-response following CAR T-cell treatment were also downregulated in high-risk antibiotics-exposed specimens.

Saini said the next step would be to establish causation between microbiome damage and the responsive outcome. That would require a microbiome manipulation study in patients who have been exposed to high risk antibiotics and correlate it with a retrospective cohort or as a retrospective study,” he added.

“We are planning for a fecal microbiota transplant study in patients who have been exposed to antibiotics,” he said. “We will give them fecal microbiom, from a normal healthy population prior to CAR T cell therapy, and then the hope that improving their microbiome would lead to better outcomes down the road. That will also establish the causation as well. Right now, what we see is just an association. We have not established causation showing that the microbiome is manipulating the responses.”


Stein-Thoeringer CK, Saini NY, Zamir E, et el. A non-antibiotic-disrupted gut microbiome is associated with clinical responses to CD19-CAR-T cell cancer immunotherapy. Nat Med. Published online March 13, 2023. doi:10.1038/s41591-023-02234-6

Related Videos
Jakub Svoboda, MD
Matthew Matasar, MD, chief, Division of Blood Disorders, Rutgers Cancer Institute; professor, medicine, Rutgers Robert Wood Johnson Medical School
Sattva S. Neelapu, MD
Sattva S. Neelapu, MD
Julie M. Vose, MD, MBA
Lakshmi Nayak, MD
Jakub Svoboda, MD
Sattva S. Neelapu, MD
Saurabh Dahiya, MD, FACP, associate professor, medicine (blood and marrow transplantation and cellular therapy), Stanford University School of Medicine, clinical director, Cancer Cell Therapy, Stanford BMT and Cell Therapy Division