Anti-CD19 chimeric antigen receptor (CAR) T cells have powerful activity against B-cell lymphoma, but improvement is clearly needed. Toxicity, including cytokine-release syndrome (CRS) and neurologic toxicity, occurs after anti-CD19 CAR T cell infusions. Most CAR T-cell toxicity is caused, either directly or indirectly, by cytokines or other proteins that are secreted from CAR T cells.

The structure of a CAR is an extracellular antigen-recognition domain connected by hinge and transmembrane (TM) domains to intracellular T-cell signaling moieties. In vitro, T cells expressing CARs with hinge and TM domains from the CD8-alpha molecule released significantly lower levels of cytokines compared with T cells expressing CARs with hinge and TM domains from CD28; however, T cells expressing CARs with hinge and TM domains from CD8-alpha retained sufficient functional capability to eradicate tumors from mice (Alabanza et al. Molecular Therapy. 2017. 25(11) 2452).

To reduce cytokine production with a goal of reducing clinical toxicity, we incorporated CD8-alpha hinge and TM domains into an anti-CD19 CAR. The CAR also had a human antigen-recognition domain, a CD28 costimulatory domain, and a CD3-zeta domain. This CAR was designated Hu19-CD828Z and was encoded by a lentiviral vector. Hu19-CD828Z was different from the FMC63-28Z CAR that we used in prior studies. FMC63-28Z had hinge and TM domains from CD28 along with a CD28 costimulatory domain, a CD3-zeta domain, and murine-derived antigen-recognition domains.

Twenty patients with B-cell lymphoma were treated on a phase I dose-escalation clinical trial of Hu19-CD828Z T cells (Table). Patients received low-dose cyclophosphamide and fludarabine daily for 3 days on days -5 to -3. Two days later, on day 0, CAR T cells were infused. The overall response rate (ORR) after 1st treatments with Hu19-CD828Z T cells was 70%, and the complete response (CR) rate 55%; the 6-month event-free survival was 55%. The anti-lymphoma activity of Hu19-CD828Z T cells in the current trial was comparable to the anti-lymphoma activity of FMC63-28Z T cells in a similar prior trial that also enrolled patients with advanced B-cell lymphoma. In the prior trial, we observed a 73% ORR, a 55% CR rate, and a 6-month event-free survival of 64% in 22 patients treated with FMC63-28Z T cells (Kochenderfer et al. Journ. Clin. Oncology. 2017 35(16) 1803).

In our previous clinical trial of FMC63-28Z T cells, the rate of Grade 3 or 4 neurologic toxicity among 22 patients treated was 55%. Strikingly, in our trial of Hu19-CD828Z T cells, the rate of Grade 3 or 4 neurologic toxicity was only 5% (1/20 patients). In addition, the rate of Grade 2 or greater neurologic toxicity with FMC63-28Z T cells was 77.3% while the rate of Grade 2 or greater neurologic toxicity with Hu19-CD828Z T cells was 15%. To explore the mechanism for the difference in neurologic toxicity in patients receiving FMC63-28Z T cells versus Hu19-CD828Z T cells, we assessed serum levels of 41 proteins in patients treated with these CAR T-cells. This comparison is valid because the same Luminex methodology was used for the serum protein analysis for both trials, and controls of known amounts of each protein were assayed to ensure that protein levels were comparable on the different trials. Lower levels of several serum proteins that might be important in CAR toxicity were found in patients treated with Hu19-CD828Z T cells versus patients treated with FMC63-28Z T cells: Granzyme A (P<0.001), Granzyme B (P<0.001), interferon gamma (P=0.011), interleukin (IL)-15 (P=0.007), IL-2 (P=0.0034), and macrophage inflammatory protein-1A (P<0.001). Median peak patient blood CAR+ cell levels were 44 cells/µL for Hu19-CD828Z and 46.5 cells/µL for FMC63-28Z (P=not significant). We hypothesize that lower levels of potentially neurotoxic proteins in patients receiving Hu19-CD828Z T cells versus FMC63-28Z T cells led to a lower frequency of neurologic toxicity in patients receiving Hu19-CD828Z T cells. The lower levels of immunologically active proteins found in the serum of patients receiving Hu19-CD828Z T cells compared with patients receiving FMC63-28Z T cells is consistent with our in vitro experiments showing lower cytokine production by T cells expressing CARs with CD8 hinge and TM domains versus CD28 hinge and TM domains. Altering CAR hinge and TM domains can affect CAR T-cell function and is a promising approach to improve the efficacy to toxicity ratio of CAR T-cells.

Disclosures

Rossi:KITE: Employment. Shen:Kite, a Gilead Company: Employment. Xue:Kite, a Gilead Company: Employment. Bot:KITE: Employment. Rosenberg:Kite, a Gilead Company: Research Funding. Kochenderfer:Kite a Gilead Company: Patents & Royalties: CAR technology, Research Funding; Celgene: Research Funding.

Author notes

*

Asterisk with author names denotes non-ASH members.

This icon denotes a clinically relevant abstract

Sign in via your Institution