Risk factors and survival analysis of human leukocyte antigen loss in relapsed acute myeloid leukaemia/myelodysplastic syndrome patients after allogeneic haematopoietic stem cell transplantation

To investigate the clinical characteristics and risk factors of specific human leukocyte antigen loss (HLA loss) in relapsed acute myeloid leukaemia (AML)/myelodysplastic syndrome (MDS) patients after allogeneic haematopoietic stem cell transplantation (allo-HSCT), and compare the responses of patients with HLA loss relapse with those without HLA loss (non-HLA loss) to different treatment regimens. Clinical data of traceable patients with AML/MDS after myeloablative allo-HSCT in our centre between January 2010 and June 2021, who experienced disease relapse after the transplantation, were collected. The patients were divided into the HLA loss relapse group and the non-HLA loss relapsed group based on HLA loss gene test findings by next-generation sequencing. The patients' median overall survival (OS) after the relapse were compared, and univariate and multivariate analyses were performed using the Kaplan-Meier survival curve and Cox proportional hazard model to explore the responses to different treatments after relapse. A total of 2359 patients were selected. Retrospective HLA gene loss gene detection was performed for the deoxyribonucleic acid in 179 relapsed patients, including 47 patients in the HLA loss group (27.2%), 126 patients in the non-HLA loss group (72.8%) and 6 patients were excluded due to a lack of confirmed results. There was no significant statistical difference in the baseline characteristics of patients between the two groups, but as to transplantation-related characteristics, the donor-recipient relationship and HLA mismatched loci were statistically different between the two groups (both p < 0.001). Multivariate Cox analysis showed that more HLA mismatched loci ≥3 (HR = 3.66; 95% CI: 1.61-8.31; p = 0.002), time (≤6 months) from HSCT to relapse (HR = 7.92; 95% CI: 3.35-18.74; p < 0.001) and donor chimerism (CD3) in bone marrow at relapse (HR = 1.02; 95% CI: 1.00-1.03; p = 0.036) were independent factors affecting HLA loss relapse. The ratio of negative conversion of FLT3-ITD or CEBPA mutation was significantly lower in patients with post-transplantation HLA loss relapse than in the non-HLA loss group (0.0% vs. 45.5%, p = 0.003; 0.0% vs. 80.0%, p = 0.035), with none of the patients with FLT3-ITD or CEBPA mutation turned negative in the HLA loss group. The number of gene mutations turned negative when relapse in the non-HLA loss group was remarkably higher than that in the HLA loss group (p = 0.001). Using donor lymphocyte infusion (DLI) could not prolong OS for the HLA loss group (p = 0.42). Nevertheless, second transplantation had a significant positive impact on OS in the HLA loss group (p = 0.017), although only five patients in the HLA loss group underwent second transplantation. However, patients in the non-HLA loss group using DLI had a relatively longer OS time than those without DLI (p = 0.017). Second transplantation could also prolong OS in the non-HLA loss group, but the effect was not as significant as in the HLA loss group (p = 0.053). In summary, HLA loss detection is essential for patients with recurrence after transplantation, especially for those with more HLA mismatched loci and non-sibling donor. Furthermore, the detection of HLA loss has a guiding role in choosing subsequent therapy when relapsed, as secondary transplantation is more suitable than DLI for those with HLA loss.