CD19-directed CAR T cell therapy is a revolutionary tool in the fight against leukemias, however, in patients with chronic lymphocytic leukemia only a small subset benefits. In order to better predict which patients are likely to respond to therapy, and to explore strategies that might expand the portion of responding patients, Fraietta et al. analyzed the T cell intrinsic properties of 41 patients with advanced, heavily pretreated, high-risk CLL who received at least one dose of CD19-directed CAR T cells.

Patients were categorized as complete responders (CR), partial responders who had highly active T cell product and responded to treatment, but relapsed with CLL that had transformed into aggressive B cell lymphoma (PR_TD), partial responders (PR), and non-responders (NR). Responding patients (CR/PR_TD) exhibited dramatic in vivo expansion of CAR T cells that proved to be essential for antitumor efficacy, and exhibited long-term persistence of functional CAR T cells. To determine whether the distinction between responses might have to do with the intrinsic potency of the CAR T cell product (as opposed to disease type or burden), Fraietta et al. delivered CAR T cells generated from patients to mice engrafted with leukemia cells. CAR T cells generated from responders decreased tumor burden and prolonged survival compared to those generated from partial or non-responders. This served as evidence that there was something about the patient-specific T cell product that could predict antitumor efficacy.

The next step for researchers was determining what those intrinsic factors might be, so they compared the transcriptomes of the CAR T cell infusion products and found that engineered T cells from responders were enriched for expression of genes involved in early memory differentiation. T cells from non-responders, on the other hand, had expression profiles enriched for regulators of late memory, effector T cell differentiation, apoptosis, and aerobic glycolysis that contributes to hypoxia and exhaustion. The researchers demonstrated that inhibition of glycolysis could promote the formation of memory CTL019 lymphocytes with enhanced proliferative capacity after serial restimulation. Upon stimulation, T cells from CR/PR_TD patients exhibited a more robust increase in T cell activation genes.

Next, the researchers analyzed cytokines and chemokines produced by CD19-directed CAR T cells and found that those from CR showed higher levels of STAT3 signaling mediators, including serum IL-6, suggesting that STAT3 activation in CAR T cells might be involved in the generation of the potent, less-differentiated T lymphocyte populations. Blockade of the IL-6/STAT3 pathway diminished cells’ proliferative capacity upon serial restimulation, indicating the functional importance of this pathway for expansion and long-term survival after infusion.

Investigating whether certain patient cell populations could predict response to CAR T cell therapy, Fraietta et al. found that the frequency of CD45RO^-CD27⁺CD8⁺ T cell subset (composed of a unique population of memory cells which are antigen-experienced, persist in a resting state, and possess properties of long-lived memory cells) was significantly higher at the time of leukapheresis in patients who were ultimately responders. This biomarker profile was able to stratify an independent cohort of 8 patients into CR and NR patients with a high degree of specificity. An elevated frequency of cells in this subset was further associated with sustained remission, and accordingly, they were found to divide extensively in culture while maintaining an extended undifferentiated phenotype. When this T cell subset was enriched and used to generate CAR T cells, they potently lysed targets, showed elevated granzyme B production, and expressed lower levels of Ki-67.

Testing whether exhaustion markers might also affect CAR T cell potency, the researchers found that while expanded CAR T cell products from CR patients had lower percentages of PD-1⁺CD8⁺ CAR T cells, the presence of exhaustion markers at the time of leukapheresis did not appear to impact future outcomes. Interestingly though, the presence of PD-1^- cells within the CD27⁺CD8⁺ T cell population within the expanded CAR product strongly segregated responders and nonresponders, and the infusion of a T cell product that contained high doses of PD-1^-CD27⁺CD8⁺ CAR T cells was highly significantly associated with the likelihood of response to therapy. This subset was further found to mediate tumor control in mice, and was found to have upregulated pSTAT3 in response to IL-6 stimulation. Serum IL-6 may also be a biomarker of potency and persistence of CAR T cell product.

The ability to predict whether a patient will respond to CAR T cell therapy could be used to spare patients from unnecessary costs, side effects, and wasted time. Furthermore, understanding what predicts a successful treatment could be used to develop strategies to select for or enhance certain cellular phenotypes and expand the pool of patients who will benefit from therapy. If the predictive biomarkers identified here hold up in larger cohorts, this tool could become critical in the clinic.

by Lauren Hitchings