Phase 1 Study of Trispecific Anti-CD20, Anti-CD19, Anti-CD22 (CAR20.19.22) T Cells for Relapsed, Refractory B-Cell Malignancies

Introduction:

Antigen downregulation is a well defined mechanism of resistance to single targeted CAR T cell therapies. To improve outcomes, new CAR constructs targeting >1 B cell receptor have been developed to enhance B-cell targeting and reduce the risk of relapse. Building on our prior work with 41BB, CD3z LV20.19 (bispecific) CAR T cells at the Medical College of Wisconsin (Shah NN. Nat Med 2020) which had high overall response rate (ORR) we designed a new Phase 1 clinical trial using a single trispecific lentiviral CAR construct encoding a tandem CAR with the same anti-CD20/anti-CD19 targeting domains but with ICOS/CD3z signaling, and a mono CAR with anti-CD22 targeting domain and CD3z signaling domain (CAR20.19.22). Preclinical data (Schneider et al. Sci Transl Med 2021) demonstrated ability of trispecific duoCAR T cells to target each antigen effectively in xenografted mouse models suggesting this approach would be ideal to handle heterogenous disease with variable B-cell expression.

Methods:

We conducted a Phase 1 single center, prospective trial (NCT05094206) evaluating CAR20.19.22 trispecific T cells as part of a dose-escalation study with a starting dose of 1x10^6 cells/kg utilizing an iBOIN Bayesian interval design. Patients were stratified into two cohorts: Indolent B-cell Lymphomas including Follicular Lymphoma, CLL/SLL and Marginal Zone Lymphoma, and Aggressive B-cell Lymphomas which include DLBCL and associated subtypes, Burkitt Lymphoma and Mantle Cell Lymphoma. Manufacturing was performed via CliniMACS Prodigy ® with an 8-12 day flexible process to limit the processing time. Standard fludarabine/cyclophosphamide lymphodepletion was administered prior to infusion.

Results:

In total, 4 patients (pts) received trispecific CAR20.19.22 T cells; 2 pts at a dose of 1x10^6 cells/kg and 2 pts at a dose of 2.5x10^6 cells/kg. All products were manufactured in a point of care fashion with target dose achieved in 8 days. Two pts received a fresh CAR infusion and 2 pts received a cryopreserved product. The median age was 66 years, and 2 pts had a diagnosis of DLBCL and 2 pts had MCL. All pts were CAR naïve except 1 MCL patient who had received LV20.19 bispecific CAR T cells as part of a prior clinical trial. The median number of lines of therapy was 5 (3-8).

The ORR response at Day 28 was 0%. Three patients had disease progression at Day 28 evaluation and one patient had stable disease with rapid progression shortly after. In terms of safety, two patients had Grade 1 CRS and received tocilizumab for management. No patients developed ICANS. Correlative studies demonstrated no meaningful post-infusion CAR expansion in any of the patients. Two patients went on to receive commercial CD19 CAR T cell therapy following this trial without meaningful response. All patients eventually passed as a result of disease progression. Due to lack of efficacy this study was closed to future enrollment.

From a product characteristic standpoint, CAR20.19.22 T cells had similar cytotoxicity, CD4:CD8 ratios, CD3 T-cell expansion during manufacture, and transduction efficiency as patients treated with LV20.19 CAR T cells on a prior clinical trial. Isoplexis single-cell cytokine secretion studies also demonstrated high polyfunctionality and polyfunctional stimulation index of CAR20.19.22 T cells when exposed to target K562-CD19 stimulators. Further in-vivo xenograft studies are being conducted to understand the mechanism of failure of this CAR product.

Conclusions:

In this first-in-human study using a novel CAR construct with ICOS signaling and trispecific targeting of CD19, CD20 and CD22, there was no CAR T expansion in patients or clinical efficacy despite administering therapeutic dosages of CAR T cells seen with other constructs. While preclinical xenograft studies for this vector demonstrated potent in-vivo efficacy, these data demonstrate the importance of human investigation and how manipulations in CAR design can impact construct efficacy. To our knowledge this is the first ICOS CAR administered to humans, and these early findings suggest careful consideration is needed for future CARs dependent on ICOS signaling rather than traditional 41BB or CD28 co-stimulation.