89Zr-immunoPET-guided selection of a CD33xIL15 fusion protein optimized for antitumor immune cell activation and in vivo tumour retention in acute myeloid leukaemia.

Purpose: Immune cells are capable of eliminating leukemic cells, as evidenced by outcomes in hematopoietic cell transplantation (HCT). However, patients who fail induction therapy will not benefit from HCT due to their minimal residual disease (MRD) status. Thus, we aimed to develop an immunomodulatory agent to reduce MRD by activating immune effector cells in the presence of leukaemia cells via a novel fusion protein that chimerises two clinically tolerated biologics: a CD33 antibody and the IL15Ra/IL15 complex (CD33xIL15).

Methods: We generated a set of CD33xIL15 fusion protein constructs with varying configurations and identified those with the best in vitro AML-binding, T cell activation, and NK cell potentiation. Using ⁸⁹Zr-immunoPET imaging we then evaluated the biodistribution and in vivo tumour retention of the most favourable CD33xIL15 constructs in an AML xenograft model. Ex vivo biodistribution studies were used to confirm the pharmacokinetics of the constructs.

Results: Two of the generated fusion proteins, CD33xIL15 (N72D) and CD33xIL15wt, demonstrated optimal in vitro behaviour and were further evaluated in vivo. These studies revealed that the CD33xIL15wt candidate was capable of being retained in the tumour for as long as its parental CD33 antibody, Lintuzumab (13.9 ± 3.1%ID/g vs 18.6 ± 1.1%ID/g at 120 h).

Conclusion: This work demonstrates that CD33xIL15 fusion proteins are capable of targeting leukemic cells and stimulating local T cells in vitro and of concentrating in the tumour in AML xenografts. It also highlights the importance of ⁸⁹Zr-immunoPET to guide the development and selection of tumour-targeted antibody-cytokine fusion proteins.