AUTHORS
Laura Mpofu, Vaishali Basu2, Audrey H. Heathcote3, Susan Gladney4, Julie O'Neal5, Matthew J. Christopher, MD, PhD6
AFFILIATIONS
1 Washington University in St. Louis School of Medicine, St. Louis, Missouri, United States
2 Washington University in St. Louis School of Medicine, St. Louis, Missouri, United States
3 Washington University in St. Louis School of Medicine, St. Louis, Missouri, United States
4 Washington University in St. Louis School of Medicine, St. Louis, Missouri, United States
5 Washington University in St. Louis School of Medicine, St. Louis, Missouri, United States
6 Washington University in St. Louis School of Medicine, St. Louis, Missouri, United States
ABSTRACT
AML relapse after HCT is often treated with infusion of donor immune cells, known as donor lymphocyte infusion, or DLI, with the goal of re-establishing immune mediated graft-versus-leukemia effect. DLI can result in transient responses but are limited both by the low frequency of donor T cells that recognize AML alloantigens and by the propensity for off-target alloreactivity, which manifests as graft-versus host disease (GvHD). Bispecific antibodies (BsAbs) target cell-surface cancer antigens while activating T cells in a non-TCR dependent fashion through CD3e. Recently, there has been interest in engineering donor grafts enriched for minor T cell subsets that are less likely to cause GvHD, including invariant natural killer T-cells (iNKT) cells, gamma delta (GD) T cells, and CD45RA-depleted T cells (enriched for memory T cells). As each of these subtypes express CD3e, we hypothesized that BsAbs targeting AML cells could be given with DLI composed of iNKT, GD T or CD45RA-depleted T cells, which may improve efficacy without increasing the risk of GvHD. To begin to test this, we expanded GD T and iNKT cells (which normally comprise a small proportion of T cells) from healthy donor PBMC. CD45RA-negative cells were purified directly from healthy donor PBMC. In vitro, all three subsets robustly killed AML cell lines and primary AML cells when cultured with either AMG330 (CD33 x CD3 BsAb) or MGD006 (CD123 x CD3 BsAb), performing at least as well or better than conventional unfractionated T (Tcon) cells in all experiments.
A major limiting toxicity of BsAbs is cytokine release syndrome (CRS), mediated by the release of IL6 and other cytokines by recipient myeloid cells during T cell killing of target cells. We used an established in vitro CRS assay where IL6 secretion by primed monocytes is quantified during killing of AML target cells by iNKT, GD T, and CD45RA-depleted T cells in presence of BsAbs. Cultures containing either iNKT or GD T cells generated significantly less IL6 then cultures containing unfractionated conventional T cells (iNKT: 2546 ± 127 pg/ml vs Tcon 7766 ± 299 pg/ml, p=0.02 by paired T test; GD T: 703 ± 439 pg/ml vs Tcon 3380 ± 1747 pg/ml, p<0.01 by paired T test, n=3-5 biologic replicates). In contrast, cultures with CD45RA-depleted T cell effectors had a trend to increased IL6 production compared to matched Tcon (CD45RA-depleted, 3702 ± 1193 pg/ml vs Tcon 2390 ± 215 pg/ml, p=0.13 by paired T test, N=3 biologic replicates). Together, these results raise the possibility that DLI consisting of purified iNKT or GD T cells may cause less CRS if given concurrently with bispecific antibodies.
Given the relative ease with which GD T cells can be expanded from cryopreserved samples, we tested whether GD T cells could be directed to clear AML cells by a bispecific antibody in a xenograft model of AML. Immune deficient NSG mice were injected with luciferase-expressing THP1 cells. On day 14, mice were injected with Tcon cells or ex vivo- expanded GDT cells. Starting the next day mice were treated with MGD006 or PBS twice per week for two weeks (N=5-8 mice in each cohort). Every week bioluminescence, weight, CBC, and peripheral blood donor T cell engraftment was measured. Mice receiving conventional T cells with MGD006 had excellent clearance of their tumor as measured by bioluminescence, however this did not translate to improved survival as the mice had significant weight loss and severe anemia, consistent with xenogenic graft versus host disease. Mice receiving GD T cells plus MGD006 had improved tumor control over the first two weeks of the experiment compared to controls but subsequently progressed and died of leukemia with only a modest survival advantage compared to controls (median survival 75 days vs 60 days, MGD006 vs vehicle, p=0.045 by log-rank test). Analysis of T cell engraftment showed significant expansion of conventional T cells in bone marrow and spleen, however GD T cells—while present in peripheral blood for 2-3 weeks after transfer at similar levels to conventional T cells—were not detected at later time points. Together, these results suggest that a DLI composed of GD T cells may clear AML cells when given with BsAbs, but that this approach may be limited by poor persistence of GD T cells. Further experiments are underway to test whether cytokine treatment may improve in vivo persistence of GD T cells in this model.
DISCLOSURES
Matthew J. Christopher, MD, PhD: No relevant conflicts of interest to declare.