Long-Term Remission of CLL Sustained By Pauciclonal Anti-CD19 Chimeric Antigen Receptor T (CTL019) Cell Clones

We recently demonstrated that sustained remission in 41 CLL patients treated with the CD19-specific, 4-1BB/CD3zeta-signaling chimeric antigen receptor (CAR19) T-cells correlated strongly with the expansion and persistence of the engineered T cells and that important pathways such as T cell exhaustion, glycolysis and T cell differentiation segregated responders from non-responders (Fraietta et al., 2018, Nature Medicine). We here report two advanced, chemotherapy-resistant CLL patients with the longest (7 years) follow-up on any trial of CART19 cells. Both patients had received five therapies before being treated at the University of Pennsylvania with autologous, murine CTL019 (tisagenlecleucel) cells for their CLL in 2010, receiving 1.1e9 and 1.4e7 CAR19+ T cells, respectively. Both patients have persistence of CAR-engineered T cells and both patients are still in remission as determined by flow cytometry and deep sequencing of IgH rearrangements for 5.5-7 years. Thus, the infused CAR-T cells have maintained these patients in deep molecular remission of their disease for the longest period of time that has been reported to date.

To understand the fate of the infused CAR-T cells we determined the phenotype, function, and clonal nature of the persisting CTL019 cells. Flow cytometric CART19 cell analyses demonstrated that early during the anti-leukemia response, activated, HLA-DR-expressing CD8+ CAR-T cells rapidly expanded, followed by similarly activated CD4+ CAR-T cells. With tumor clearance the CAR-T cell population contracted, but an activated CD4+ CAR-T cell population was maintained and was still detectable at the last follow-up of 7 years. The CD8+ CAR-T cell pool remained present at low frequencies. Both populations had acquired and maintained an effector memory phenotype, a phenotype most consistent with active disease control. Furthermore, the analysis of the classical immune checkpoint inhibitory markers PD1, TIM3, LAG3, and CTLA4 showed that only PD1 was expressed from the earliest to the latest time point on >80% of all CAR-T cells, whereas LAG3 and TIM3 were expressed only early on but lost after tumor clearance. These data suggest that the initial tumor clearance was mediated by CD8+ CAR-T cells, but sustained by a CD4+ CAR-T cell population that still actively engages with target cells. To understand the clonal nature of these long-term persisting CAR-T cells we used two complementary methods: a) CAR T cells were sorted from post-infusion aliquots during the first two years for T cell receptor-beta deep-sequencing (TCR-seq); b) the CAR integration sites in the genome were sequenced in the infusion product and in circulating CAR-T cells. TCR-seq analysis of early post-infusion time points demonstrated that the circulating CAR-T cell populations consisted of hundreds to thousands of distinct clones which in patient 1 and 2 displayed clonal focusing by 21 and 1 month post-infusion, respectively, with some clones making up as much as 12% (patient 1) and 48% (patient 2) of the CAR-T cell repertoire. The analysis of clonotype sharing at the various time points via Morisita's overlap index analysis similarly showed repertoire stabilization late (21 months; patient 1) and early (1 month; patient 2) after infusion. Lastly, fate mapping of the infused CART19 cells via CAR integration site analysis in the infusion product until the latest time point indicated that the infusion products for both patients had a very diverse, non-clonal make-up, containing over 8,000 and 3,700 integration sites in patients 1 and 2, respectively. The higher degree of clonality in patient 2 but not 1 CAR-T cells as seen by TCR-seq was confirmed by integration site analysis, as was the sharing of CAR-T cell clones over time. Importantly, whereas the CAR integration site repertoire in patient 1 was diverse in the first two years, it stabilized and trended towards oligoclonality 21 months after infusion. Lastly, CAR integration site analysis revealed a high degree of clonal persistence, suggesting that tumor control and B cell aplasia were maintained by few, highly functional CD4+ CAR-T cell clones.

In summary, we demonstrate that in both patients with the longest persistence of CAR-T cells reported thus far, early and late phases of the anti-CLL response are dominated by highly activated CD8+ and CD4+ CAR-T cells, respectively, largely comprised of a small number of persisting CD4+ CAR-T cell clones.