Manufacturing of a Subsequent Autologous CAR-T Product after Prior CAR-T Is Safe and Feasible

Introduction: Patients with large B-cell lymphoma frequently fail to achieve a durable remission following CD19-directed chimeric antigen receptor T-cell therapy (CAR-T). Unfortunately, the outcomes for patients who progress after CAR-T are dismal with the median overall survival of approximately 6 months. Trials are now demonstrating the possibility of a subsequent autologous CAR-T for treating these patients. We demonstrated a CD22-directed CAR-T (CAR22) achieved a 68% and 53% overall (ORR) and complete response (CR) rate in 38 subjects who progressed after a CD19 CAR-T (CAR19) (NCT04088890). Of the 20 patients who achieved CR, 17 never relapsed after a median follow up of 22.8 months. The safety and feasibility of manufacturing a second (or more) CAR-T from patients has largely not been addressed and some concerns have been raised regarding the theoretical risk of malignant transformation due to the repeatedly introduction of new genomic material through serial CAR-T manufacturing. Here we address the impact of residual CAR19 on CAR22 production and patient outcomes.

Method: We evaluated the patients who progressed after CAR19 and underwent manufacturing for CAR22. Single-cell RNA-sequencing (scRNAseq; 10X) was used to quantify the frequency of axi-cel transcripts in CAR22 products. Patients' peripheral blood mononuclear cells (PBMCs) were collected on day 7, 14, 21, and 28 after CAR22 infusion. The FMC63 and CD22.BB.z sequence (with albumin as a control) were quantified using quantitative polymerase chain reaction (qPCR) to determine the presence of CAR19 and CAR22 within the PBMCs. The assay's limit of detection and quantification (LOQ) were calculated as 1.68 and 5.1 copies per 50ng of DNA reaction with 95% confidence, respectively.

Results: Among 40 patients underwent manufacturing for CAR22, 2 failed manufacturing due to insufficient T cells in the apheresis material. One of the two patients received prior alemtuzumab in combination with a CAR19 product, likely contributing to the failure. Among the 38 who were successfully manufactured, 37 patients had received prior CAR19, including axi-cel (n = 28), tisa-cel (n = 3), liso-cel (n = 3), autologous CAR19 in clinical trial (n = 1), bispecific CD19/CD22 CAR-T (n = 1), and bispecific CD19/CD20 CAR-T (n = 1). The median progression-free survival (PFS) from the previous CAR19 infusion was 3.3 months (range from 1.0 to 40.4 months), with a median age of 64.4 years old (ranging from 24.9 to 83.9 years old) at the time of relapse. The median time from CAR19 infusion to leukapheresis for CAR22 manufacturing was 6.8 months (ranging from 1.2 to 42.9 months). Thirteen CAR22 products were available for scRNAseq; all from patients who had received axi-cel previously. The median frequency of CAR19 positive cells was approximately 68 (ranging from 0 to 2146) per million cells per CAR22 product. Residual CAR19 cell proportions were not statistically significantly different between patients who underwent CAR22 manufacturing within or after 6 months of CAR19 infusion.

Thirty-one patients had available PBMCs post-CAR22 infusion for analysis. Twelve patients (39%) exhibited an increase in CAR19 qPCR transgene levels above the LOQ during CAR22 treatment. Most of these patients (n = 8) underwent leukapheresis for CAR22 manufacturing within 6 months after CAR19 infusion. The median CAR19 qPCR transgene levels increased by 1.2 log (ranging from 0.2 to 2.1 log; equivalent to 1.4 to 133.7 copies per 50ng of DNA), while CAR22 transgene levels increased by a median of 3.7 log (ranging from 0.7 to 3.9 log; equivalent to 4.7 to 9027.1 copies per 50ng DNA). In the comparison between patients who underwent leukapheresis within or beyond 6 months after CAR19 infusion, there were no statistically significant differences in the maximal CAR19 and CAR22 qPCR transgene levels. The ORR, cytokine release syndrome severity, immune effector cell-associated neurotoxicity syndrome severity and PFS after CAR22 were not statistically significantly different between patients with detectable CAR19 qPCR or not.

Conclusion: We found that CAR19 was detectable in the majority of CAR22 products, but the levels of CAR19 transcripts were rare. While CAR19 cells were detected post-CAR22 infusion, the levels were frequently just at the LOQ and did not appear to contribute to toxicity or clinical outcomes. Overall, we find that serial manufacturing of autologous CAR-T products is safe and feasible.