Post-transplant cyclophosphamide after matched sibling, unrelated and haploidentical donor transplants in patients with acute myeloid leukemia: a comparative study of the ALWP EBMT

Background The use of post-transplant cyclophosphamide (PTCy) is highly effective in preventing graft-versus-host disease (GVHD) in the haploidentical (Haplo) transplant setting and is being increasingly used in matched sibling (MSD) and matched unrelated (MUD) transplants. There is no information on the impact of donor types using homogeneous prophylaxis with PTCy. Methods We retrospectively compared outcomes of adult patients with acute myeloid leukemia (AML) in first complete remission (CR1) who received a first allogeneic stem cell transplantation (SCT) with PTCy as GVHD prophylaxis from MSD (n = 215), MUD (n = 235), and Haplo (n = 789) donors registered in the EBMT database between 2010 and 2017. Results The median follow-up was 2 years. Haplo-SCT carried a significantly increased risk of acute grade II–IV GVHD (HR 1.6; 95% CI 1.1–2.4) and NRM (HR 2.6; 95% CI 1.5–4.5) but a lower risk of relapse (HR 0.7; 95% CI 0.5–0.9) that translated to no differences in LFS (HR 1.1; 95% CI 0.8–1.4) or GVHD/relapse-free survival (HR 1; 95% CI 0.8–1.3). Interestingly, the use of peripheral blood was associated with an increased risk of acute (HR 1.9; 95% CI 1.4–2.6) and chronic GVHD (HR 1.7; 95% CI 1.2–2.4) but a lower risk of relapse (HR 0.7; 95% CI 0.5–0.9). Conclusions The use of PTCy in patients with AML in CR1 receiving SCT from MSD, MUD, and Haplo is safe and effective. Haplo-SCT had increased risk of acute GVHD and NRM and lower relapse incidence but no significant difference in survival.

Since the first reports of Haplo-SCT using PTCy, there has been significant interest in comparing this platform with those using other donor types such as cord blood from unrelated donors [4][5][6]8] or bone marrow (BM) or peripheral blood (PB) from MSD, MUD, and MMUD transplants [9][10][11][12][13][14][15]. However, an important limitation of these studies is that each type of transplant received different GVHD prophylaxis. Therefore, comparisons were made of different transplant platforms instead of different donor types. Two prospective studies compared T cell replete Haplo and MSD transplants in acute myeloid leukemia (AML) [16] and acute lymphoblastic leukemia [17] using non-PTCy GVHD prophylaxis with similar results but increased risk of GVHD after Haplo. In addition, safety and feasibility of PTCy-based GVHD prophylaxis in transplants using different donor types was further evaluated in small single-center prospective non-randomized studies showing comparable outcomes in Haplo-SCT compared to 9/10 MUD [18], as well as in MSD, MUD, and MMUD transplants [19].
The aim of this study was to investigate the impact of donor type in the outcome of patients with AML undergoing unmanipulated allogeneic SCT (allo-SCT) using PTCy as GvHD prophylaxis. We analyzed patients who received allo-SCT from MSD, MUD, and Haplo for acute myeloid leukemia (AML) in first complete remission (CR1) and reported to the European Society for Blood and Marrow Transplantation (EBMT) registry from 2010 to 2017.

Study design and data source
This is a retrospective registry-based analysis on behalf of the Acute Leukaemia Working Party (ALWP) of the EBMT. The EBMT is a voluntary working group of more than 600 transplantation centers that are required to report all consecutive stem cell transplantations and follow-up once a year. Audits are routinely performed to determine the accuracy of the data. All transplantation centers are required to obtain written informed consent before data registration with the EBMT in accordance with the 1975 Helsinki Declaration.

Patient eligibility
All adults (age ≥ 18 years) with AML in CR1 at transplantation, reported to Promise-EBMT, who underwent first allogeneic SCT, from an unmanipulated graft, using PT-Cy from Haplo, MUD, or MSD donors between 2010 and 2017, were analyzed. Haplo was defined as recipient-donor number of human leukocyte antigen (HLA) mismatches ≥ 2.

Endpoints and definitions
The primary endpoint was to compare leukemia-free survival (LFS) after MSD, MUD, and Haplo donor transplants. Secondary endpoints were neutrophil engraftment, acute GVHD (aGVHD) and chronic GVHD (cGVHD), relapse incidence, nonrelapse mortality (NRM), GVHD-free and relapse-free survival (GRFS), and overall survival (OS) within the same subgroups and to perform analysis of risk factors for each outcome.
Neutrophil recovery was defined as the first day of an absolute neutrophil count of 0.5 × 10 9 /L lasting for 3 or more consecutive days. aGVHD and cGVHD were defined and graded according to standard criteria [20,21]. Relapse was defined as disease recurrence and appearance of blasts in the peripheral blood or BM (> 5%) after CR. LFS was calculated until the date of first relapse, death from any cause, or the last follow-up for patients in CR. NRM was defined as death from any cause other than relapse. The composite endpoint GRFS was defined as survival without the following events: stage III-IV aGVHD, severe cGVHD, disease relapse, or death from any cause after SCT [22]. Myeloablative conditioning (MAC) was defined as a regimen containing either total body irradiation with a dose greater than 6 Gray, a total dose of oral busulfan greater than 8 mg/kg, or a total dose of intravenous busulfan > 6.4 mg/kg or melphalan at doses > 140 mg/m 2 . In addition, regimens containing 2 alkylating agents were considered as MAC. All other regimens were defined as reduced intensity (RIC).

Statistical analysis
Patient characteristics according to donor type were compared using chi-squared tests for categorical and Kruskal-Wallis tests for continuous variables. GRFS, LFS, and OS were estimated using the Kaplan-Meier method. Cumulative incidence functions were used to estimate neutrophil engraftment, aGVHD, cGVHD, relapse incidence, and NRM. Competing risks were death for relapse incidence and neutrophil engraftment, relapse for NRM, and relapse or death for aGVHD and cGVHD. Univariate analyses were done using the logrank test for LFS, GRFS, and OS and Gray's test for cumulative incidence. Multivariate analyses were performed using the Cox proportional hazard model. Donor type, gender, age at transplantation, performance status, cytogenetic risk group according to the Medical Research Council [23], type of AML (primary vs secondary) stem cell source, transplantation year, cytomegalovirus serostatus, and conditioning regimen were included in the final model. To take into account the center effect, we introduced a random effect (also named frailty effect) for each center into the model. The significance level was fixed at .05, and p values were 2-sided. Statistical analyses were performed using R software version 3.2.3 (R Development Core Team, Vienna, Austria) software packages.
Haplo patients were older (p < 0.001) and had higher proportion of secondary AML (p < 0.001), while differences in gender, performance status, and cytogenetic risk category were not statistically significant. Regarding transplant characteristics, Haplo-SCT recipients received more frequently MAC (p = 0.006) and BM as stem cell source (p < 0.001), while the proportion of in vivo T cell depletion was higher in MUD transplants (p < 0.001). Although the vast majority (93%) of Haplo patients received GvHD prophylaxis with PTCy combined with 2 other immunosuppressive (IS) drugs, only 47% and 26% of MUD and MSD patients received such combination, respectively. In contrast, a higher proportion of MSD and MUD patients received 1 or no additional IS drugs than Haplo patients (p < 0.001).

Discussion
The use of PTCy for GvHD prophylaxis in patients with AML in CR1 receiving SCT from MSD, MUD, and Haplo is safe and effective, resulting in low rates of GVHD, especially chronic, in all transplant settings. Using this approach, our results demonstrate that patients undergoing Haplo-SCT had higher rates of aGVHD and NRM, but lower relapse incidence. As seen in other transplant scenarios, PB was also associated with more GVHD and less relapse. Due to the retrospective nature of a registry-based study, some potential bias cannot be completely ruled out. In order to minimize one of the most important, such as the disease status at transplantation, all patients included in the analysis had AML in CR1. Although all patients received PTCy prophylaxis as the main inclusion criteria for the study, a variety of conditioning regimens were used and there were obvious differences in additional GvHD prevention strategies, such as in vivo TCD or the addition of other IS drugs, depending of the type of donor. In fact, Haplo patients received more frequently MAC, BM, and PTCy with 2 IS drugs, while a higher proportion of MUD patients received in vivo TCD. Although some of these variables could be adjusted in multivariable analysis, TCD and combination of IS drugs for GvHD prophylaxis were strongly associated with type of donor and their effect could not be evaluated. Despite all these pitfalls, we aimed to compare MSD, MUD, and Haplo using a homogeneous GvHD prophylaxis with PTCy in a large series of patients, which allowed us to segregate the effect of donor from the effect of GvHD prophylaxis.
PTCy was highly effective in preventing acute and chronic GvHD in MSD, MUD, and Haplo-SCT and  (14) Interstitial pneumonitis 3 (4) 4 (6) 6 (2) Sinusoidal obstruction syndrome 0 (0) 1 (2) 7 (3) seems to compare favorably with standard GvHD prophylaxis with calcineurin inhibitor and methotrexate in MSD and MUD transplants [24]. In fact, the incidence of GvHD after PTCy seems similar to that reported with anti-T cell globulin in both scenarios [25], with the potential advantage of avoiding complications associated with prolonged TCD. An interesting observation was the higher incidence of acute grades II-IV and a trend towards a higher severe acute GvHD of Haplo-SCT compared to MSD transplants. It should be noted that most previous studies had not been able to demonstrate this effect [9,11,14,26], probably due to the use of different GvHD prophylaxis for each procedure. In fact, Haplo was associated with increased risk of GVHD compared to MSD in two prospective studies using similar non-PTCy transplant platforms [16,17]. Particularly relevant was the low overall chronic and chronic extensive GvHD observed in our study in all cohorts, as it has been previously reported with PTCy [1,4,7,24,27]. We should highlight that we did not observe increased risk of chronic GvHD in Haplo in multivariable analysis. We could speculate that PTCy abrogates the detrimental effect of HLA disparity for this particular outcome, but we should also consider that most patients in the Haplo cohort received PTCy with a double combination of IS drugs, while patients in the MSD and MUD cohorts received less intensive GvHD prophylaxis. In fact, a recent study of the ALWP-EBMT has recently reported that the addition of IS drugs to PTCy enhances its effect and reduces the risk of severe chronic GvHD, reducing mortality and improving survival [6]. The most important observation from our study is that, using PTCy as GVHD prophylaxis, NRM was higher in the Haplo setting compared with the MSD and MUD cohorts. Previous studies comparing Haplo (with PTCy) with MSD and MUD transplants (with standard GvHD prophylaxis) have reported discrepant results in terms of NRM. While some studies reported a higher rate of NRM in Haplo [11,15], some others reported similar [12,14] or even improved outcomes [10]. The reasons of these discrepancies remain unexplained and are probably multifactorial, but differences in transplant platforms could explain, at least in part, some of these results. Under similar GvHD prophylaxis, a greater HLA disparity in the Haplo compared with the MSD and MUD settings could explain a higher NRM. This finding suggests that HLA-matched donors, when available, should remain as the first choice. Although the negative impact of Haplo in NRM was partially counterbalanced with a decreased incidence of relapse that translated in similar LFS, other strategies aiming at reducing relapse such as maintenance or MRD-guided therapy with a growing targeted therapy strategies could be investigated.
The fact that Haplo was associated with lower risk of relapse deserves special attention. It is possible that this was a spurious finding since more patients in the Haplo cohort died from NRM and were therefore no longer at risk of relapse. However, Haplo procedure could have offered enhanced anti-leukemic efficacy, intriguingly in a way that was independent of chronic GVHD. Superior graft-versusleukemia effect of Haplo compared to MSD transplants for high-risk AML has already been observed in previous comparative studies. Two prospective trials with biological randomization from China showed decreased posttransplant minimal residual disease (MRD) positivity [28] or relapse [29] in patients undergoing Haplo, particularly relevant for those with detectable pretransplant MRD. In addition, a recent retrospective study of EBMT also showed decreased relapse incidence in patients with highrisk cytogenetics undergoing Haplo [15].
The immunological pressure of Haplo grafts has been illustrated with the observation that loss of the mismatched HLA haplotype is a frequent mechanism of escape associated with relapse [30]. The biological explanation is unknown but NK-mediated alloreactivity has been previously proposed to induce enhanced efficacy and GVHD protection in the context of T celldepleted Haplo-SCT [31]. The hypothesis of an increased anti-leukemic efficacy independent of GvHD of Haplo-SCT compared to matched donors in the context of PTCy should be further explored from a clinical and biological point of view.
Despite the risks and benefits of BM over PB have been widely investigated, the effect of the stem cell source on transplant outcomes deserves special consideration. In MUD transplants, BM reduced the risk of chronic GVHD in a randomized study [32] and improved long-term GRFS and overall survival in a large retrospective registry study [33]. In Haplo-SCT, the use of PB resulted in an increased risk of acute GvHD, uncertain impact of chronic GvHD, and decreased risk of relapse in patients with acute leukemia, but not with lymphoma [34,35]. In the present study, we confirm that PB was associated with increased risk of acute and chronic GvHD, less relapses, but no final influence on NRM and survival.
Although most retrospective studies comparing MAC with RIC in patients with AML have suggested similar survival, since the latter has been associated with increased relapse but reduced NRM [36,37], we observed a significant reduction of relapse with MAC that translated into improved survival when compared with RIC. Unfortunately, the only randomized study, designed to address this issue in patients with AML, was closed to patient accrual early due to excess of relapse and reduced survival in the RIC cohort [38]. Since the efficacy of RIC in SCT mainly relies on graft-versus-leukemia effect, it may be particularly relevant to increase conditioning intensity in transplant platforms with effective GvHD control such as with the use of PTCy.

Conclusion
In patients with AML undergoing allo-SCT, PTCy for GvHD prophylaxis showed promising outcomes. Future studies comparing PTCy standard regimens are warranted to establish the standard of care. In this specific scenario, Haplo-SCT had increased risk of acute GVHD and NRM that was counterbalanced by a lower relapse incidence that translated into no significant difference in LFS and OS. Haplo-SCT offers a good alternative to matched donor transplants.