A Phase I Trial of Activated Donor-Derived Marrow Infiltrating Lymphocytes to Treat Relapsed Hematologic Malignancies after Allogeneic Transplantation with Post-Transplantation Cyclophosphamide As Graft-Versus-Host Disease Prophylaxis

Background: Donor lymphocyte infusion (DLI) or a second allogeneic transplant (HCT) are the only curative treatments for hematologic malignancies relapsing after initial HCT. Both approaches lack significant efficacy and may cause considerable toxicity. Post-transplant cyclophosphamide (PTCy) used as graft-versus-host disease (GVHD) prophylaxis eliminates alloreactive T cells and enriches for tumor-specific T cells in the bone marrow. We hypothesized that expanded, donor-derived marrow infiltrating lymphocytes (alloMILs) collected post-HCT could restore antitumor immunity in relapsed recipients without the additional GVHD and toxicity seen with DLI or second HCT.

Methods: We conducted a phase I dose-escalation trial (NCT02342613) of expanded alloMILs to treat bone marrow infiltrating hematologic malignancies relapsing after HCT with PTCy. Key inclusion criteria were age ≥ 18 years; relapsed hematologic malignancy ≥ 6 months after HLA matched or mismatched HCT with PTCy; donor CD3⁺ chimerism ≥ 30% in the peripheral blood; and lack of immunosuppressant use for 2 weeks prior to alloMILs harvest. Enrolled patients underwent salvage chemotherapy prior to or after alloMILs harvest to control the disease during the expansion process. Harvested bone marrow mononuclear cells were co-incubated with anti-CD3/CD28 beads and IL-2 for 7 to 14 days until the target CD3⁺ dose was reached. Given the differential risk of GVHD with matched versus mismatched DLI, dosing progressed from 1 x 10⁷ to 5 x10⁷ CD3⁺ cells/kg for matched HCT recipients and 1 x 10⁶ to 5 x10⁷ CD3⁺ cells/kg for mismatched HCT recipients. The primary endpoint was the feasibility of the product, defined as: successful harvest; expansion; infusion; and absence of grade III-IV acute GVHD for 90 days after infusion. Evaluation for HLA loss was performed using next-generation sequencing (NGS) of HLA loci. Single cell RNA sequencing (scRNAseq) was performed using 10x Genomics technology, Cell Ranger (v6.0.2), and Seurat (v4.3.0).

Results: Thirty-two patients enrolled and underwent bone marrow harvest. Two patients were inaspirable, and five patients were not expanded due to low cell counts or high tumor volume. Twenty-five patients underwent alloMILs expansion, of which five failed to expand. Twenty patients received an infusion and six received a second infusion. Of the twenty patients receiving an infusion, eleven had multiple myeloma (MM) or plasma cell leukemia; five had acute myeloid leukemia (AML); one had chronic myeloid leukemia in blast phase; and three had chronic lymphocytic leukemia. Harvest, expansion, and infusion were well tolerated with minimal adverse events. No patient developed acute or chronic GVHD after infusion. Dosing advanced to the highest level of 5 x10⁷ CD3⁺ cells/kg in both the mismatched and matched HCT cohorts. At day 60, the overall response rate (composite of complete (CR) and partial responses) was 40 percent (8/20) in the entire cohort; 33.3 percent (2/6) in the myeloid cohort (Figure 1A); 30 percent (3/10 evaluable patients) in the MM cohort; and 66.7 percent (2/3) in the lymphoid cohort. One MM patient achieved sustained > 2 years long CR and one AML patient is in ongoing CR, both of whom received alloMILs with no or minimal disease followed by maintenance therapy. HLA loss analysis was performed on pre-expansion samples from three AML patients receiving a mismatched HCT and alloMILs. Both non-responders in this cohort had loss of the host mismatched HLA haplotype. ScRNAseq on pre-expansion samples from ten AML patients, eleven MM patients, and four AML patients in remission after HCT (control) yielded 197 522 total cells and 32 587 CD8⁺ T cells. CD8⁺ T cells expressing a senescent-like gene signature previously associated with dysfunctional tumor killing in AML patients (Mazziotta et al ASH 2022) were associated with non-response primarily in AML patients (observed log2 fold change -2.48 compared with responders; FDR = 0.0017) (Figure 1B).

Conclusions: AlloMILs demonstrated feasibility, a favorable safety profile, and early evidence of clinical activity in patients with measurable residual disease. The maximum dose of 5 x10⁷ CD3⁺ cells/kg was reached, and no patient developed acute or chronic GVHD. HLA loss and CD8⁺ T cell senescence may be associated with non-response in AML patients. A clinical trial utilizing alloMILs as maintenance therapy for patients with measurable residual disease is in development.