Augmenting Efficacy of T-Cell Bispecific Antibodies in AML through a Tumor Stroma-Targeted 4-1BB Agonist

Bispecific antibodies represent a promising treatment option for acute myeloid leukemia (AML). We have recently described a novel T-cell bispecific antibody (TCB) targeting the intracellular tumor antigen Wilms tumor 1 (WT1) in the context of HLA-A*02 (Augsberger et al. Blood 2021). Based on these findings a multicenter first-in-human clinical trial was initiated in relapse/refractory AML (NCT04580121). Possible immune escape mechanisms against T-cell based immunotherapy are provided by the tumor microenvironment (TME) of the bone marrow by co-inhibition of T cells or stromal cells shielding leukemic cells from immune effector cells. To overcome the immunosuppressive effect of the TME and to enhance T-cell responses, we evaluated the combination of the WT1-TCB with an antibody fusion protein that targets a stromal antigen (Fibroblast-activation protein; FAP) and provides a positive costimulatory signal (4-1BBL) to T cells. FAP is upregulated on cancer-associated fibroblasts after remodulation of the bone marrow niche by leukemic cells, and the FAP specificity of the molecule therefore provides T-cell co-stimulation tightly restricted to the tumor niche.

Efficacy of the combination (WT1-TCB + FAP-4-1BBL antibody fusion protein) was evaluated in co-culture assays over 4 days with primary HLA-A*02 ⁺ AML cells, healthy donor (HD) T cells and three NIH-3T3 fibroblast cell lines. NIH-3T3 cell lines were genetically modified to express low and high levels of FAP, respectively. Wild-type NIH-3T3 cells were included as control. Additionally, a control (Ctrl)-TCB and a Ctrl-4-1BBL antibody fusion protein recognizing a non-tumor target derived from the human germline repertoire were included. Enhancement of T-cell mediated cytotoxicity by the FAP-4-1BBL antibody fusion protein was evaluated by (1) specific lysis of primary AML cells, (2) upregulation of the T-cell activation markers CD25 and 4-1BB, (3) T-cell expansion calculated as fold change compared to day 0, and (4) Granzyme B-expression which was evaluated by intracellular staining.

After 4 days of co-culture, with an E:T ratio of 1:2, we observed a mean specific lysis of 55.1±8.2% (±SEM; n=4) of primary AML cells mediated by HD T cells and WT1-TCB. Notably, this was reduced to 19.4±5.9% (±SEM; n=4) in the presence of NIH-3T3 cells. However, AML cell lysis was restored by the addition of the FAP-4-1BBL antibody fusion protein in the presence of high FAP expressing NIH-3T3 cells (mean specific lysis: 62.8±7.3%; ±SEM; n=4). Concomitantly, the FAP-4-1BBL antibody fusion protein led to increased expression of the activation molecules CD25 (MFI ratio: 22.1±5.3 vs. 10.4±1.3; ±SEM; n=4) and 4-1BB (MFI ratio: 10.4±6.0 vs. 2.1±0.3; ±SEM; n=4) on CD3 ⁺ T cells. Furthermore, lysis was accompanied by increased frequencies of granzyme B expressing T cells (45.0±2.5% vs. 16.1±5.3%; n=3). Importantly, the FAP-4-1BBL antibody fusion protein led to improved T-cell proliferation, especially of CD8 ⁺ T cells (fold change on day 4 vs day 0: 5.7±2.2 vs. 1.0±0.3; ±SEM; n=4). Overall similar observations were made in the presence of low FAP expressing NIH-3T3 cells.

Taken together, we have established an in vitro model system mimicking the immunoprotective bone marrow TME using NIH-3T3 cells resulting in impaired AML cell lysis. Providing additional T-cell co-stimulation by a tumor-stroma targeted 4-1BB agonist, however, restored WT1-TCB-mediated cytotoxicity of primary AML cells in the presence of FAP expressing cell lines. Importantly, the combination overcame the immunosuppressive effect of the NIH-3T3 cells on T cells as further demonstrated by improved T-cell activation and expansion. The tumor-stroma targeted 4-1BB agonist therefore represents a promising combinatorial approach to enhance T-cell activity at the local tumor site and warrants further investigations in an in vivo model system.