Induction of Antigen-Specific T-Cell Responses through Dendritic Cell Vaccination in AML: Results of a Phase I/II Trial and Ex Vivo Enhancement By Checkpoint Blockade

Postremission therapy is critical for successful elimination of minimal residual disease (MRD) in acute myeloid leukemia (AML). Innovative treatment options are needed for patients that are not eligible for allogeneic stem cell transplantation. As the intrinsic immune response against leukemia-associated antigens (LAAs) is generally low, the clinical application of checkpoint inhibitors as monotherapy is less promising in AML compared to other hemato-oncological diseases. Therapeutic vaccination with autologous dendritic cells (DCs) loaded with LAAs is a promising treatment strategy to induce anti-leukemic immune responses.

We have conducted a phase I/II proof-of-concept study using monocyte-derived next-generation DCs as postremission therapy of AML patients with a non-favorable risk profile in CR/CR_i after intensive induction therapy (NCT01734304). These DCs are generated using a GMP-compliant 3-day protocol including a TLR7/8 agonist, loaded with RNA encoding the LAAs WT1 and PRAME as well as CMVpp65 as adjuvant/surrogate antigen, and are applied intradermally up to 10 times within 26 weeks. The primary endpoint of the trial is feasibility and safety of the vaccination. Secondary endpoints are immunological responses and disease control.

After the safety and toxicity profile was evaluated within phase I (n=6), the patient cohort was expanded to a total of 13 patients. DCs of sufficient number and quality could be generated from leukapheresis in 11/12 cases. DCs exhibited an immune-stimulatory profile based on high costimulatory molecule expression, IL-12p70 secretion, migration towards a chemokine gradient and processing and presentation of antigen. In 9/9 patients that are currently evaluable, we observed delayed-type hypersensitivity (DTH) responses at the vaccination site, but no grade III/IV toxicities. TCR repertoire analysis by next-generation sequencing revealed an enrichment of particular clonotypes at DTH sites. In the peripheral blood, we detected vaccination-specific T cell responses by multimer staining and by ELISPOT analysis: 7/7 patients showed responses to CMVpp65, including both boosting of pre-existing T cells in CMV⁺ patients and induction of a primary T cell response in CMV^- patients. 2/7 patients exhibited responses to PRAME and WT each. 7/10 vaccinated patients are still alive, and 5/10 are in CR, with an observation period of up to 840 days.

In vitro, DC-activated T cells showed an upregulation of PD-1 and LAG-3, while the DCs expressed the respective ligands PD-L1 and HLA-DR. Therefore, we studied the capacity of checkpoint blocking antibodies to further enhance T-cell activation by DCs. We found that blockade of PD-1 and particularly of LAG-3 was highly effective in enhancing both IFN-γ secretion and proliferation of T cells. Both pathways seem to target different T-cell subsets, as PD-1 blockade resulted in increased IFN-γ secretion by T_N- and T_EM-subsets, while blockade of LAG-3 significantly affected T_N- and T_CM-subsets.

We conclude that vaccination with next-generation LAA-expressing DCs in AML is feasible, safe, and induces anti-leukemic immune responses in vivo. Our in vitro data supports the hypothesis that T-cell activation by means of the vaccine could be further enhanced by blocking PD-1 and/or LAG-3.