To the Editor

Allogeneic hematopoietic stem cell transplantation (allo-HCT) has remained the best option for patients with relapsed/refractory acute myeloid leukemia (AML) [1]. This approach relies on graft-versus-leukemia (GVL) effects for leukemia eradication. In patients receiving grafts from HLA-matched donors, numerous studies have demonstrated a tight association between occurrence of graft-versus-host disease (GVHD) and lower risk of relapse [2,3,4].

Post-transplant cyclophosphamide (PTCy)-based GVHD prophylaxis has revolutionized the field of human leukocyte antigen (HLA)-haploidentical hematopoietic cell transplantation (Haplo-HCT) [5, 6]. Consequently, Haplo-HCT is nowadays frequently used as treatment for relapsed/refractory AML patients [7]. A recent systems biology analysis in patients with PTCy-based GVHD prophylaxis demonstrated different signatures associated with GVHD and GvL effects [8]. In addition, another study observed different T-cell phenotypes associated with GVHD or GvL in PTCy-Allo-HCT recipients [9]. These observations prompted us to perform a large retrospective study in the EBMT registry aimed at assessing whether PTCy given in the Haplo-HCT setting might dissociate GVL effects from GVHD in patients with active AML at transplantation, a subgroup of patients who particularly rely on GVL effects for leukemic cell eradication. Population selection criteria included ≥ 18 years of age at transplantation, Haplo-HCT between 2010 and 2020 with PTCy, no prior allo-HCT, and primary refractory or relapsed AML (i.e. all patients had active disease at the time of transplant conditioning initiation).

The analyses were carried out in a total of 670 patients (Additional file 1: Table 1). The 180-day incidences of grade II-IV and grade III-IV acute GVHD were 30.8% (95% CI 27.4–34.3%) and 13.3% (95% CI 10.9–16%), respectively. These incidences were 21% and 8%, respectively, in BM recipients versus 35% (P = 0.001) and 16% (P = 0.008), respectively, in PBSC recipients. The 2-year cumulative incidences of chronic and extensive chronic GVHD were 26.8% (95% CI 23.4–30.3%) and 13% (95% CI 10.5–15.8%), respectively. There was no impact of stem cell source on chronic GVHD incidence. However, in vivo T-cell depletion was associated with a lower incidence of chronic GVHD (17% versus 28%, P = 0.027).

Table 1 Final model stratified on landmark at time intervals from day of allo-HCT to day + 360 by 30 days
Full size table

The impact of GVHD on transplantation outcomes was assessed using dynamic landmarking i.e. a method including a series of landmark analyzes at each time interval of 30 days from allo-HCT to day 365 (Table 1, see Additional file 1 for more details) [10].

There was no impact of acute nor of chronic GVHD on relapse incidence (Table 1 and Fig. 1). There were no associations between either grade II acute GVHD nor limited chronic GVHD on NRM, LFS nor OS in dynamic landmarking models (Table 1). However, grade III-IV acute GVHD was associated with higher NRM (HR = 3.09, 95% CI 1.87–5.12, P < 0.0001) and a statistical trend for lower LFS (HR = 1.36, 95% CI 0.99–1.86, P = 0.056) (Fig. 1). In contrast, grade I acute GVHD was associated with a trend for lower NRM (HR = 0.53, 95% CI 0.27–1.01, P = 0.055) and better LFS (HR = 0.71, 95% CI 0.51–0.99, P = 0.042) and OS (HR = 0.68, 95% CI 0.48–0.97, P = 0.032). We do not have a biological explanation for the lower NRM in patients with grade 1 acute GVHD. Future studies needed to evaluate whether this is due to a better immune reconstitution in patients with grade I acute GVHD. Finally, extensive chronic GVHD was associated with higher NRM (HR = 3.3, 95% CI 1.81–6.04, P < 0.0001) and lower LFS (HR = 1.97, 95% CI 1.35–2.89, P = 0.0004) and OS (HR = 1. 95, 95% CI 1.29–2.94, P = 0.001) (Fig. 1).

Fig. 1
figure 1

A–D Day-100 landmark analyses (n = 477) showing the impact of grade I, II and grade III–IV acute GVHD on: A relapse incidence (P = 0.39); B incidence of nonrelapse mortality (NRM, P = 0.001); C Leukemia-free survival (LFS, P = 0.005); D overall survival (OS, P = 0.002). E–H Day-365 landmark analyses (n = 234) showing the impact of limited and extensive chronic GVHD on: E relapse incidence (P = 0.8); F NRM (P = 0.021); G LFS (P = 0.11); H OS (P = 0.014)

Full size image

Our results differ from what has been observed by the Baltimore group in patients receiving Haplo-HCT with PTCy-based GVHD prophylaxis after nonmyeloablative conditioning as treatment of various hematological malignancies (n = 340) [11]. Indeed, in that study, grade II acute GVHD was associated with a lower risk of relapse. Our observations are, however, concordant with recent observations in another large population of patients treated with Haplo-HCT as treatment for AML in CR (n = 805) [10] and with data in humanized mouse models of GVHD in which it was demonstrated that PTCy attenuated GVHD without abrogating graft-versus-leukemia effects [12].

The absence of association between GVHD occurrence and the risk of relapse might suggest that in vivo T-cell depletion could be particularly suitable in the Haplo-HCT PTCy setting. However, we observed that ATG was associated with higher relapse incidence in multivariate analysis, without significantly affecting OS and LFS.

In conclusion, we demonstrated in a cohort of patients with active AML at transplantation treated with PTCy-based T-cell repleted Haplo-HCT that occurrence of GVHD did not decrease the risk of relapse suggesting a dissociation of GvL effects from GVHD in this transplantation setting.