A comparison between allogeneic stem cell transplantation from unmanipulated haploidentical and unrelated donors in acute leukemia

Background In the absence of a HLA-matched related or matched unrelated donor, allogeneic stem cell transplantation (allo-SCT) from mismatched unrelated donors or haploidentical donors are potential alternatives for patients with acute leukemia with an indication to allo-SCT. The objective of this study was to compare the outcome of allo-SCT from T cell-replete haploidentical (Haplo) versus matched (MUD 10/10) or mismatched unrelated donor at a single HLA-locus (MMUD 9/10) for patients with acute leukemia in remission. Methods Two hundred sixty-five adult patients with de novo acute leukemia in first or second remission that received a Haplo-SCT between January 2007 and December 2013 were compared with 2490 patients receiving a MUD 10/10 and 813 receiving a MMUD 9/10. Propensity score weighted analysis was conducted in order to control for disease risk imbalances between the groups. Results The weighted 3-year non-relapse mortality and relapse incidence were 29 and 30% for Haplo, 21 and 29% for MUD 10/10, and 29 and 25% for MMUD 9/10, respectively. The weighted 3-year leukemia-free survival (LFS) and overall survival (OS) were 41 and 46% for Haplo, 50 and 56% for MUD 10/10, and 46 and 48% for MMUD 9/10, respectively. Using weighted Cox model, both LFS and OS were significantly higher in transplants from MUD 10/10 compared from those in Haplo but not different between transplants from MMUD 9/10 and Haplo. The type of donor was not significantly associated with neither acute nor chronic graft-versus-host disease. Conclusions Patients with acute leukemia in remission have better outcomes if transplanted from a MUD 10/10. We did not find any significant difference in outcome between transplants from MMUD 9/10 and Haplo, suggesting that both can be equally used in the absence of a 10/10 MUD. Key point 1 Better outcomes using fully (10/10) matched unrelated donor for allo-SCT in acute leukemia in remission. Key point 2 Similar outcomes after allo-SCT from unmanipulated haploidentical graft or mismatched (9/10) unrelated donor in acute leukemia in remission. Electronic supplementary material The online version of this article (doi:10.1186/s13045-017-0394-2) contains supplementary material, which is available to authorized users.


Background
Allogeneic stem cell transplantation (allo-SCT) represents the only possible cure for most adult acute leukemias (AL). HLA-matched related (MRD) or unrelated donors (MUD) are usually considered the preferable donors, but they are not available for all the patients with an indication for allo-SCT. In the absence of a HLA-matched donor, allo-SCT from mismatched unrelated donors (MMUD), cord blood units (CB), or haploidentical (Haplo) donors are potential alternatives.
A Haplo donor is available for virtually all AL patients, enabling minimal delay and access to repeated stem cell (SC) donations or donor lymphocyte infusions, if needed. These are the main reasons for the increasing numbers of Haplo-SCT for AL in recent years. Unmanipulated (non ex vivo T-depleted) grafts from Haplo donors, in comparison to T-depleted ones, result in lower incidence of serious infections due to faster immune reconstitution and stronger graft versus leukemia effect [1,2]. In addition, the introduction of more effective regimens for graft-versus-host disease (GvHD) prophylaxis for Treplete Haplo-SCT contributed to reduce GvHD incidence and to increase the use of unmaniplated grafts for the Haplo setting [3][4][5][6][7][8][9][10].
Recently, several reports have shown comparable allo-SCT outcomes between Haplo and historical MRD, MUD, and MMUD series [11][12][13][14][15]. However, these are mostly but one [14] single center studies with limited number of patients [11,12,15] in various disease categories and status. For these reasons, we decided to perform a large, registry-based study, using the European Society of Bone Marrow Transplantation (EBMT)-Acute Leukemia Working Party (ALWP) registry, comparing T-replete Haplo-SCT to transplants from MUD and MMUD for AL patients in first or second remission.

Methods
In order to be included in the study, the patients had to fulfill all the following criteria: age ≥18 years; de novo AL; disease status at transplant: complete remission 1 (CR1) or 2 (CR2); family donor with host/donor number of HLA mismatches ≥2 (Haplo), or MUD 10/10 or MMUD 9/10 (patients and donors should have HLA A, B, C, and DRB1 and DQB1 allelic typing performed); peripheral blood (PB) or bone marrow (BM) or both as source of SC; no ex vivo T cell depletion; and first allo-SCT (previous autologous SCT was allowed). All patients underwent transplantation between January 2007 and December 2013. We were able to verify the inclusion criteria for 265 Haplo-SCT, 2490 MUD 10/10-SCT, and 813 MMUD 9/10-SCT. This was a retrospective multicenter analysis. Data were provided and approved for this study by the ALWP of the EBMT group registry.
The EBMT is a non-profit, scientific society representing more than 600 transplant centers mainly in Europe. The EBMT promotes all activity aiming to improve stem cell transplantation or cellular therapy, which includes registering all the activity relating to stem cell transplants. Data are entered, managed, and maintained in a central database with internet access; each EBMT center is represented in this database. There are no restrictions on centers for reporting data, except for those required by the law on patient consent, data confidentiality, and accuracy. Quality control measures included several independent systems: confirmation of validity of the entered data by the reporting team, selective comparison of the survey data with minimum essential data A (MED-A) data sets in the EBMT registry database, cross-checking with the National Registries, and regular in-house and external data audits. Since 1990, patients have provided informed consent authorizing the use of their personal information for research purposes.

Definitions and statistical analysis
The primary endpoints were leukemia-free survival (LFS) and overall survival (OS). The secondary endpoints were engraftment, acute and chronic GVHD (aGVHD and cGVHD), relapse incidence (RI), nonrelapse mortality (NRM), and graft-versus-host relapsefree survival (GRFS) [16]. LFS was defined as time to death or relapse, whichever came first. OS was defined as time to death from all causes. NRM was defined as death without evidence of relapse. Engraftment was defined as the first of three consecutive days with an absolute neutrophil count > 0.5 × 10 9 /L. Acute GVHD was graded according to the modified Seattle-Glucksberg criteria [17] and cGVHD according to the revised Seattle criteria [18].
We used propensity scores (PS) weighting to control for pre-treatment imbalances on observed variables. The following factors were included in the PS model: patient age, time from diagnosis to transplantation, year of transplant, diagnosis (AML versus ALL), status at transplant (CR1 versus CR2), cytogenetics group, donor/ patient CMV serology, conditioning (RIC versus MAC), and sex matching (female donor to male recipient versus other). The estimation of propensity score was performed using generalized boosted models [19].
As the study question was whether Haplo could replace 10/10 or 9/10 MUD, we weighted the groups receiving either MUD 10/10-and MMUD 9/10-HSCT to match the characteristics of patients receiving Haplo-SCT, by estimating the average treatment effect among the treated (ATT); Haplo-HSCT being the treated group. The ATT weights equal one for Haplo-HSCT, and it equals the ratio of the propensity score to one minus the propensity score in the two UD-HSCT groups. Therefore, UD patients that were significantly different from average haplo-grafted patients had a low contribution in the comparisons. We checked the balance between the groups looking to ATT weighted means. Then, we used pairwise ATTs to fit the weighted Kaplan-Meier and Cox models separately for Haplo-versus MUD 10/10-HSCT and Haplo-versus MMUD 9/10-HSCT.
The type I error rate was fixed at 0.05 for determination of factors associated with time to event. Analyses were performed using the R statistical software version 3.2.3 (R Development Core Team, Vienna, Austria); propensity score analysis was performed using the mnps function of the Twang package and weighted analyses using the survey package [20].

Patients' and donors' characteristics
Patient's and host/donors' characteristics are showed in Table 1. Regimens, stem cell source, and GvHD prophylaxis are described in Table 2. Anti-thymocyte globulin

Discussion
In the absence of a MRD for allo-SCT in acute leukemia, the ideal donor still remains to be determined. The current report represents a large registry study comparing the outcome between transplants from Haplo and MUD 10/10 or MMUD 9/10 for adult patients with de novo AL in remission.
As a recent ALWP-EBMT survey on unmanipulated haploidentical transplantation in AL showed a lower incidence of aGvHD in patients receiving PT-Cy but no significant statistical differences in either LFS or OS [21], in our study we included all the non T-depleted Haplo-SCT registered in the EBMT database.
Previous reports described worse outcomes after MMUD in comparison to MUD [22][23][24][25][26], and this finding was in accordance to our preliminary data and to a recent ALWP publication [27]. We did not find any differences in terms of LSF and OS in the 813 MMUD-SCT according to HLA-DQ mismatch in univariate analysis (p = 0.35). Therefore, we decided to compare separately MUD 10/10 and MMUD 9/10 transplants with Haplo. The 30-day CI of engraftment proved to be lower in Haplo than in MUD 10/10 but not different between Haplo and MMUD 9/10. In addition to the type of donor, the number and type of chemotherapy cycles pretransplant and/or early post-transplant infections, in particular viral, could have had an impact on differences in engraftment. A lower incidence of engraftment in Haplo compared in UD was previously reported in both reduced intensity [14] and myeloablative conditioning transplants [15].
Notably, we did not find any difference in the incidence of either aGvHD > II or cGvHD according to donor type in line with similar comparisons in literature. This was true also for grades III-IV aGvHD and extensive cGvHD. In the recent report comparing Haplo using PT-Cy to UD from the Center for International Blood and Marrow Transplant Research (CIBMTR), the authors describe a lower incidence of acute and cGvHD in Haplo on univariate analysis [14], but most of the Haplo were performed using BM as stem cell source, and most of the UD using PB without receiving any in vivo T cell depletion. Of note, they did not find any difference in the 3-year cGvHD incidence according to donor type when analyzing only patients receiving BM. The role of stem cell source as risk factor for cGvHD after Haplo-SCT remains to be determined. While some reports do not show any difference in terms of cGvHD between PB and BM in haploidentical transplantations using PT-Cy [28,29], a recent comparison from CIBMTR  concludes for a high incidence of cGvHD using PB instead of BM [30]. Notable, in the context of UD, cGvHD has been shown to be higher using PBSC [31] especially if not using in vivo T cell depletion [32]. The 3-year NRM was higher using Haplo (29%) compared with MUD 10/10 (21%) but identical compared with MMUD (29%), in contrast with other reports on similar comparisons where the authors did not find any difference [11][12][13][14][15]. This higher NRM in Haplo compared in MUD 10/10 could be in part due to the use of in vivo T cell depletion other than PT-Cy in the 45% of our Haplo [33]. In addition to the type of in vivo T cell depletion, also, differences in conditioning regimens could explain our finding [14]. Patients' comorbidities, mostly unknown in restrospective study, could also have influenced NRM. The 3-year CI of relapse was not different between the three groups.
The lower NRM in MUD 10/10 resulted in both a higher 3-year LFS and OS compared in Haplo but no difference in GRFS. No differences in either OS, LFS, or GRFS were observed between Haplo and MMUD 9/10. Analyzing separately patients with AML and ALL, we found no differences of GRFS in Haplo compared in MUD 10/10 and compared in MMUD 9/10, higher LFS and OS in MUD 10/10 compared in Haplo but no differences between Haplo and MMUD 9/10 in terms of LFS and OS (Additional file 1).
Based on these results, we can assert that patients with acute leukemia in remission showed better outcomes if transplanted from a MUD 10/10. We did not find any significant difference in outcome between MMUD 9/10 and Haplo, suggesting that both can be equally used in the absence of a MUD 10/10 and that other factors, such as urgency of transplant and center expertise should dictate the choice between these two alternative donor sources.
Ongoing and future prospective clinical studies, including transplants from CBU, will ultimately be required to determine the best alternative donor for adult AL patients who lack an HLA-matched sibling one.