Clinical Efficacy of Anti-CD22 Chimeric Antigen Receptor T Cells for B-Cell Acute Lymphoblastic Leukemia Is Correlated with the Length of the Scfv Linker and Can be Predicted Using Xenograft Models

Introduction. Anti-CD19 chimeric antigen receptor T cells (CART19 or CTL019) have shown impressive clinical activity in B-cell acute lymphoblastic leukemia (B-ALL) and are poised to receive FDA approval. However, some patients relapse after losing CD19 expression. Since CD22 remains highly expressed in relapsed/refractory (r/r) B-ALL even in these patients, anti-CD22 CART (CART22) have been developed. The National Cancer Institute (NCI) reported 4/9 complete remission (CR) in patients receiving CART22, with 100% CR at the highest T cell dose (NCT02315612)(S hah NN, ASH 2016 #650).

Patients and Methods. We generated a second-generation CAR22 differing from that used by the NCI only by the use of a longer linker [4x(GGGGS); LL vs. 1x(GGGGS); SL] between the light and heavy chains of the scFv (Fig. 1 A). This construct was tested in two pilot clinical trials in adults (NCT02588456)and children with r/r-ALL (NCT02650414). CART22 cells were generated using lentiviral transduction as in our previous studies. The protocol-specified CART22 dose was 2x10⁶-1x10⁷ cells/kg for pediatric patients <50kg and 1-5x10⁸ for pediatric patients ≥50kg and adult patients,. infused after lymphodepleting chemotherapy. Patient characteristics are described in Table 1. For the adult trial, 5 patients were screened, 4 enrolled (1 patient withdrew consent) and 3 infused (1 manufacturing failure). For the pediatric trial, 9 patients were screened, 8 enrolled (1 screen failure) and 6 infused (two patients were not infused for disease progression).

For the preclinical studies, we generated CART22^LL and CART22^SL and tested them in vivo using xenograft models. NOD-SCID gamma chain deficient (NSG) mice were engrafted with either a luciferase+ standard B-ALL cell line (NALM6) or primary B-ALL cells obtained from a patient relapsing after CART19 (CHP110R). We also used 2-photon imaging to study the in vivo behavior and immune synapse formation and flow cytometry to asses T cell activation.

Results. CART22 cells were successfully manufactured for 10/12 patients. In the adult cohort 3/3 patients developed CRS (gr.1-3) and no neurotoxicity was observed; in the pediatric cohort out of 5 evaluable patients (1 discontinued for lineage switch to AML on pre-infusion marrow), 3/5 developed cytokine-release syndrome (CRS) (all grade 2) and 1 patient had encephalopathy (gr.1). CART22 cells expanded in the PB with median peak of 1977 (18-40314) copies/ug DNA at day 11-18. Interestingly, in an adult patient who had previously received CART19 a second CART19 re-expansion was observed following CART22 expansion (Fig 1 B). At day 28, in the adult cohort the patient who was infused in morphologic CR remained in CR, while the other 2 had no response (NR); in the pediatric cohort 2/5 patients were in CR, 1 in partial remission (PR) that then converted to CR with incomplete recovery at 2 months, and 2 NR. No CD22-negative leukemia progression was observed.

Since our results with a long linker appeared inferior compared to the previously reported CART22 trial (short linker), we performed a direct comparison of the 2 different CAR22 constructs. In xenograft models, CART22^SL significantly outperformed CART22^LL (Fi 1 C) with improved overall survival. Moreover, CART22^SL showed higher in vivo proliferation at day 17 (Fig 1 D). Mechanistically, intravital 2-photon imaging showed that CART22^SL established more protracted T cell:leukemia interactions than did CART22^LL, suggesting the establishment of productive synapses (Fig 1 E). Moreover, in vivo at 24 hrs higher T cell activation (CD69, PD-1) was observed in CART22^SL from the BM of NALM-6-bearing mice.

Conclusions. Here we report the results of two pilot clinical trials evaluating the safety and feasibility of CART22 therapy for r/r B-ALL. Although feasible and with manageable toxicity CART22^LL led to modest clinical responses. Preclinical evaluation allowed us to conclude that shortening the linker by 15 amino acids significantly increases the anti-leukemia activity of CART22, possibly by leading to more effective interactions between T cells and their targets. Finally, with the caveats of cross-trial comparison, our data suggest that xenograft models can predict the clinical efficacy of CART products and validate the use of in vivo models for lead candidate selection

View largeDownload slide

View largeDownload slide

Close modal