Azacitidine (AZA) and donor lymphocyte infusion (DLI) is a viable salvage approach for chimerism-only relapse following allogeneic blood or marrow transplantation (BMT) Free

Background: Post-BMT relapse of myeloid malignancies is fraught with limited options for salvage. AZA-DLI is commonly used, but the outcomes, especially in haploidentical donor (haplo) BMTs, are not well characterized. Here, we present data from our uniformly managed institutional cohort to identify optimal timing and population for AZA-DLI as a salvage strategy for post-BMT relapse.

Methods: We conducted a retrospective chart review of all patients who underwent AZA-DLI for relapse after BMT at our center between January 2016 and November 2024. Patients received treatment for mixed chimerism only (≤95% donor chimerism) or for clinical relapse [increased blasts for acute myeloid leukemia (AML) or abnormal blood counts for myelodysplastic syndrome (MDS) and myeloproliferative neoplasm (MPN), including MDS/MPN overlap neoplasms]. Full donor chimerism (FDC) was defined as >95% donor chimerism in bone marrow or peripheral blood. Clinical improvement (CI) was evaluated in patients with clinical relapse, and defined as decrease in marrow and blood blasts in AML, or an improvement of blood counts in patients with MDS and MPN. Stable Disease (SD) was defined as a stable percentage of blasts without worsening chimerism. Overall response rate (ORR) included patients achieving FDC or CI. Standard grading was used for acute and chronic graft versus host disease (a/cGVHD). Standard statistical methods were used for comparison of responses, and overall survival (OS) was calculated from the date of established post-BMT relapse.

Results: We identified 50 patients who met study criteria: 22 (44%) AML, 15 (30%) MDS, and 13 (26%) MPN. All except one received reduced-intensity conditioning, and all received post-transplant cyclophosphamide as GVHD prophylaxis backbone. Thirty-nine (78%) were above 50 years of age at BMT, 37 (74%) underwent BMT using haplo donors, and 33 (66%) received a peripheral blood graft. Time to relapse from BMT was ≤6 months in 11 (22%) patients, and>6 months in 39 (78%). Mixed chimerism relapse and clinical relapse were noted in 14 (28%) and 36 (72%) patients, respectively. The median (range) days from BMT to the first DLI was 525 days (125 - 6123). Patients received 1-3 DLIs with a median of 4 AZA cycles (range 1-70). The first DLI dose was 1 x 10⁶cells/kg for haplo and mismatched unrelated donors, 1 x 10⁷-10⁸cells/kg for matched siblings, and at least 1x10⁸cells/kg for matched unrelated donors. Doses were escalated for subsequent DLIs if given. At 6 months from DLI, FDC was achieved in 64.3% of mixed chimerism patients versus 19.4% of clinical relapse patients (p = 0.0053). CI was noted in 4 out of 36 patients (11.1%) with clinical relapse. ORR was 9 (64.3%) in the mixed chimerism group and 11 (30.6%) in the clinical relapse group (p = 0.0519). In an additional 2 patients, stable disease was observed. OS at 1 year and 2 years was 85.71% (95%CI 54-96) and 60% (95%CI 29-81) in the mixed chimerism group and 72.2% (95%CI 55-84) and 34% (95%CI 19-50) in the clinical relapse group (p = 0.0383), respectively. Six patients in the mixed chimerism and one patient in the clinical relapse group remain in remission at the time of this analysis.

There was no significant difference in the ORR between AML versus MDS or MPN patients (50% vs. 32.1%, p = 0.2514), with an OS at 1 year and 2 years in AML of 77% (95%CI 54-90) and 36% (95%CI 17-56) and in MDS or MPN, 75% (95%CI 54-87) and 45% (95%CI 25-63), respectively (p = 0.7885). In the analysis by time to relapse post-BMT by ≤6 months versus >6 months, there was no significant difference in ORR (54.5% vs. 35.9%, p = 0.3109) or OS [1 year OS: 73% (95%CI 37-90) versus 77% (95%CI 60-87); 2 year OS: 42% (95%CI 13-68) vs 41% (95%CI 25-56 p = 0.7619)]. In 11 patients, AZA-DLI was used >3 years from their BMT, of whom 4 (36.4%) achieved FDC and 1 (9%) noted clinical improvement.

When AZA-DLI was used for mixed chimerism, rates of grades II-IV aGVHD were 35.7%, grades III-IV aGVHD 21.4%, and cGVHD requiring systemic immunosuppression 21.4%. For clinical relapse, these rates were 16.7% (p = 0.2521), 5.6% (p = 0.1262), and 5.6% (p = 0.1262), respectively.

Conclusion: Our data support AZA-DLI as a feasible strategy for managing post-BMT relapse even in a predominant haplo-BMT population, or when used many years after BMT. Pre-emptive intervention with DLI in a setting of mixed chimerism is associated with superior outcomes, while monitoring for GVHD is warranted.