Adults with acute myeloid leukemia who reach complete remission but still carry measurable residual disease face one of the most difficult decisions in modern hematology: which donor should provide the transplant that offers the best chance of staying in remission. A large retrospective analysis from the European Society for Blood and Marrow Transplantation, drawing on registry data from 269 centers, now suggests that haploidentical transplantation with post-transplant cyclophosphamide may be associated with a lower risk of relapse than transplants from matched sibling or matched unrelated donors in this high-risk setting, even though overall survival did not differ significantly among the three donor types.
The study, published in Experimental Hematology & Oncology, analyzed 3385 adult patients with acute myeloid leukemia who were in first or second complete remission when they received an allogeneic hematopoietic cell transplant between 2015 and 2022. Of these, 609 received a haploidentical transplant, 1057 received a transplant from a matched sibling donor, and 1719 received a transplant from a matched unrelated donor. The median follow-up for the entire cohort was 2.1 years, with an interquartile range of 2.1 to 2.3 years.
Measurable residual disease, detected either by multiparameter flow cytometry or by molecular biology methods, is the commonly used approach to predict relapse after transplant in acute myeloid leukemia. Several studies have reported that the risk of relapse after transplant significantly increases in patients who are MRD positive before the procedure. Disease relapse remains the major challenge for these patients, and the choice of donor is a major issue when planning an allogeneic transplant. While a matched sibling donor remains the priority in most guidelines, haploidentical transplantation has surged globally over the recent decade, and an increasing number of studies have suggested that haploidentical transplants might exert superior graft-versus-leukemia effects than HLA-matched donor transplants, especially in high-risk disease.
The baseline characteristics of the three cohorts were not identical, and the authors emphasized that these differences may represent potential confounders, which were adjusted for in multivariable analyses. Patients in the haploidentical cohort more frequently had intermediate or high-risk disease according to the European LeukemiaNet 2022 risk stratification based on cytogenetic abnormalities, at 88.7 percent, compared with 83.2 percent in the matched sibling cohort and 85.9 percent in the matched unrelated cohort. Patients in the haploidentical cohort were also more frequently transplanted in second complete remission, at 22.2 percent, versus 15.2 percent for matched sibling and 18.8 percent for matched unrelated recipients. Myeloablative conditioning was used less frequently in the haploidentical cohort, at 43.8 percent, than in the matched sibling cohort at 55.6 percent or the matched unrelated cohort at 50.1 percent.
A crucial difference in GVHD prophylaxis shaped the comparison. In this study, haploidentical transplants were restricted to recipients receiving post-transplant cyclophosphamide-based GVHD prophylaxis, while post-transplant cyclophosphamide was used in only 11.4 percent of the matched sibling cohort and 8.6 percent of the matched unrelated cohort as the backbone of GVHD prophylaxis. The authors explicitly flagged this as a major limitation: post-transplant cyclophosphamide was highly collinear with donor type, and the data cannot fully disentangle whether the observed advantage associated with haploidentical transplantation reflects the donor source itself, the use of post-transplant cyclophosphamide, or a combination of both.
On multivariable analysis using a Cox cause-specific model with competing events, both matched sibling and matched unrelated transplants were associated with significantly higher relapse incidence compared with haploidentical transplantation as the reference. The hazard ratio for relapse incidence was 1.55 for matched sibling donors, with a 95 percent confidence interval of 1.26 to 1.91 and a p value below 0.01, and 1.26 for matched unrelated donors, with a 95 percent confidence interval of 1.03 to 1.54 and a p value of 0.03. Despite a lower non-relapse mortality with matched sibling donors, with a hazard ratio of 0.63 and a 95 percent confidence interval of 0.46 to 0.84, the higher relapse incidence led to inferior leukemia-free survival compared with haploidentical transplantation, with a hazard ratio of 1.19 and a 95 percent confidence interval of 1.01 to 1.41.
The graft-versus-host disease picture was more nuanced. Both matched sibling and matched unrelated transplants were associated with significantly lower incidence of grade II to IV acute GVHD compared with haploidentical transplantation, with hazard ratios of 0.68 and 0.80 respectively. Although overall chronic GVHD incidences were not different between matched sibling and haploidentical transplants, the matched sibling group was associated with a significantly higher incidence of extensive chronic GVHD, with a hazard ratio of 1.77 and a 95 percent confidence interval of 1.28 to 2.46. Collectively, matched sibling transplantation was associated with significantly inferior GVHD-free, relapse-free survival, with a hazard ratio of 1.22 and a 95 percent confidence interval of 1.05 to 1.41. No significant difference was observed between matched unrelated and haploidentical transplantation regarding leukemia-free survival and GVHD-free, relapse-free survival, and no significant difference was observed among the three donor types on overall survival.
Several other factors influenced outcomes across the cohort. Transplantation in second complete remission, compared with first complete remission, was associated with a significantly higher relapse incidence, with a hazard ratio of 1.5, and with lower leukemia-free survival, overall survival and GVHD-free, relapse-free survival, with hazard ratios of 1.21, 1.33 and 1.26 respectively. Myeloablative conditioning was significantly associated with a lower relapse incidence and better leukemia-free survival, overall survival and GVHD-free, relapse-free survival compared with reduced-intensity conditioning, with hazard ratios of 0.75, 0.78, 0.77 and 0.83 respectively. Older patient age, measured as the effect for each 10-year increase, was associated with higher non-relapse mortality, with a hazard ratio of 1.23, and worse overall survival, with a hazard ratio of 1.10. A total of 1069 patients died during the study period, and relapse or progression of the underlying disease was the most common cause of death in all three cohorts, at 71.4 percent in the matched sibling group, 59 percent in the matched unrelated group and 55.7 percent in the haploidentical group. GVHD was the second most frequent cause of death, occurring more often in the matched unrelated cohort, at 16.6 percent, than in the haploidentical cohort at 12.9 percent or the matched sibling cohort at 10.5 percent.
The authors concluded that these findings suggest haploidentical transplantation with post-transplant cyclophosphamide may offer superior relapse prevention compared with matched sibling or matched unrelated transplantation for MRD-positive acute myeloid leukemia patients in remission, and may serve as a good alternative for these patients lacking a matched sibling donor. However, they urged caution in interpreting the apparent relapse reduction. Haploidentical transplantation was associated with a lower risk of relapse compared with matched sibling transplantation, but also with higher non-relapse mortality and similar overall survival. The apparent reduction in relapse after haploidentical transplantation may at least partly reflect the competing effect of increased non-relapse mortality, meaning that some patients who might have relapsed died of other causes first. The retrospective design is an inherent limitation, and the lack of detailed information on MRD detection methods in the EBMT registry further complicates the interpretation of the results, although even minute levels of MRD positivity have been shown to carry clear prognostic significance.
Because a well-designed randomized study comparing haploidentical, matched sibling and matched unrelated transplantation is not yet available, the authors argued that their registry-based results are of major clinical importance. They called for dedicated studies directly comparing the three donor types with all groups using post-transplant cyclophosphamide-based GVHD prophylaxis, which would allow the field to determine whether the relapse advantage seen here belongs to the haploidentical donor itself or to the prophylaxis platform that almost always accompanies it.
Subject of Research: Donor type comparison for allogeneic transplantation in adults with AML in remission with measurable residual disease
Article Title: Comparison of haploidentical, matched sibling and matched unrelated donor allo-HCT in adult patients with AML in complete remission with measurable residual disease: a study from the Global committee and the ALWP of the EBMT
Article References: Ye, Y., Galimard, J.-E., Labopin, M., Ayuk, F., Yakoub-Agha, I., Stelljes, M., Salmenniemi, U., Schroeder, T., Bug, G., Gallego-Hernanz, M.-P., Forcade, E., Passweg, J., Itala-Remes, M., Piemontese, S., Sanz, J., Bazarbachi, A., Nagler, A., Brissot, E., Li, L., … Gorin, N. C. (2026). Comparison of haploidentical, matched sibling and matched unrelated donor allo-HCT in adult patients with AML in complete remission with measurable residual disease: a study from the Global committee and the ALWP of the EBMT. Experimental Hematology & Oncology, 15(1), Article 96. https://doi.org/10.1186/s40164-026-00827-8
Image Credits: AI Generated
DOI: 10.1186/s40164-026-00827-8
Keywords: acute myeloid leukemia, allogeneic hematopoietic cell transplantation, haploidentical transplant, matched sibling donor, matched unrelated donor, measurable residual disease, post-transplant cyclophosphamide, relapse incidence, graft-versus-host disease, EBMT, registry study, leukemia-free survival
News Source: Nathaniel Bowman. (October 29, 2026). Haploidentical transplants show lower relapse risk for MRD-positive AML patients in EBMT registry analysis. Scienmag.
