Identification of Functional Determinants of Response and Resistance to CD19 Chimeric Antigen Receptor (CAR) T-Cell Therapy of Chronic Lymphocytic Leukemia

The adoptive transfer of autologous T cells genetically modified to express a chimeric antigen receptor (CAR) targeting CD19 has shown remarkable activity for patients with B cell malignancies. Using a second- generation CAR signaling through 4-1BB/CD3ζ, (CTL019), over 90% of patients with relapsed or refractory (R/R) acute lymphoblastic leukemia (ALL) will achieve complete remission (Maude, S.L. et. al. New England Journal of Medicine, 2014). Similar therapy can induce long-term remissions for R/R chronic lymphocytic leukemia (CLL), but only in a subset of patients (Porter, D.L. et. al. Science Transl. Medicine, 2015). The ability to identify individuals most likely to respond will have a dramatic impact on patient selection and provide critical information that may allow immune modification to enhance responses using this approach. We here present updated analyses and functional validation of our previously reported biomarker profiles of CTL019 cells in CLL (Fraietta, J.A. et. al., Blood, 2016). Retrospective samples of clinically manufactured CAR T cells from forty-one patients with advanced, heavily pre-treated and high-risk CLL were evaluated for functional determinants of potency. Transcriptomic analysis of CTL019 cells from subjects who had complete responses (CR) to this therapy revealed enrichment in early memory T cells and IL-6/STAT3 gene signatures, while T cells from non-responding (NR) patients were enriched in genes belonging to known pathways of terminal differentiation, exhaustion, apoptosis and glycolysis. Accordingly, CAR-stimulated T cells from NR relative to CR patents demonstrated an increased uptake of a glucose analog; inhibition of glycolysis using 2-deoxy-D-glucose resulted in the generation of early memory CAR T cells in culture. Furthermore, CTL019 cells that were highly functional in CLL patients had increased levels of phosphorylated STAT3 (pSTAT3) upon activation compared to CAR T cells from NR patients with poor anti-tumor activity. These data were concordant with the highly significant direct correlation between serum IL-6 and the degree of CTL019 proliferative capacity in patients. Increasing pSTAT3 activity by addition of exogenous IL-6 to culture media elevated the absolute numbers of CAR T cells following repeated stimulation with CD19-expressing tumor targets, while pSTAT3 inhibition diminished CTL019 cell growth.

Single-cell proteomic analysis of CAR-stimulated CTL019 cells from CLL patients revealed marked heterogeneity in cytokine secretions and enhanced polyfunctional strength in CAR T cells from responding over non-responding patients. There were multiple contributors to CTL019 cell polyfunctional strength, including granzyme B, MIP-1α, IFNg, TNFα, IL-5, IL-8, MIP-1β and perforin. We next performed deep immunophenotyping on the above pre-infusion CAR T cells and, through unbiased computational analysis of the flow cytometry data, defined a single subset (CD8+CD27-PD1-) that segregated CRs from NRs. Importantly, we noted a highly significant association between the likelihood of having a response to CTL019 therapy and the infusion of CTL019 products containing a high dose of CD8+PD1-CD27+ CAR T cells. To confirm that this cell population drives the response to CTL019 therapy, we depleted it from CR patient CTL019 cells and observed loss of tumor control and decreased survival in leukemia-bearing NOD scid gamma (NSG) mice compared to animals treated with subject-matched unmanipulated cellular products. These findings, coupled with our previous results, reveal a functionally relevant T lymphocyte population that can serve as a novel predictive biomarker and a framework for future mechanistic dissection to enhance the efficacy of CAR T cell therapies.