By continuing to browse this site you agree to us using cookies as described in About Cookies
Notice: Wiley Online Library will be unavailable on Saturday 7th Oct from 03.00 EDT / 08:00 BST / 12:30 IST / 15.00 SGT to 08.00 EDT / 13.00 BST / 17:30 IST / 20.00 SGT and Sunday 8th Oct from 03.00 EDT / 08:00 BST / 12:30 IST / 15.00 SGT to 06.00 EDT / 11.00 BST / 15:30 IST / 18.00 SGT for essential maintenance. Apologies for the inconvenience.
Olle Ringdén, MD, PhD, Center for Allogeneic Stem Cell Transplantation, Karolinska University Hospital Huddinge, F79, SE-141 86 Stockholm, Sweden. (fax: +46 8 746 6699; e-mail: email@example.com).
Abstract. Ringdén O, Labopin M, Beelen DW, Volin L, Ehninger G, Finke J, Greinix HT, Kyrcz-Krzemien S, Bunjes D, Brinch L, Niederwieser D, Arnold R, Mohty M, Rocha V, for the Acute Leukaemia Working Party of the European Group for Blood and Marrow Transplantation (EBMT) (Karolinska University Hospital Huddinge, Sweden; CEREST-TC EBMT, Paris, France; University of Duisburg-Essen, Germany; Helsinki University Central Hospital, Helsinki, Finland; University Hospital, Dresden, Germany; University Hospital, Freiburg, Germany; Medical University Vienna, Vienna, Austria; Medical University of Silesia, Katowice, Poland; University Hospital, Ulm, Germany; Rikshospitalet, Oslo, Norway; University Hospital, Leipzig, Germany; Charité University Hospital, Berlin, Germany; Université de Nantes, Nantes, France; and Hôpital Saint-Louis, Paris, France). Bone marrow or peripheral blood stem cell transplantation from unrelated donors in adult patients with acute myeloid leukaemia, an Acute Leukaemia Working Party analysis in 2262 patients. J Intern Med 2012; 272: 472–483.
Background. No survival benefit of using blood stem cells instead of bone marrow (BM) has been shown in matched unrelated donor (MUD) transplantation.
Design and methods. In a retrospective registry analysis, we compared the use of blood stem cells (n =1502) and BM (n =760) from unrelated donors in patients aged 18–60 years with acute myeloid leukaemia (AML) undergoing myeloablative conditioning between 1997 and 2008. The blood stem cell recipients were older (P <0.01), had more advanced disease (P <0.0001) and received less total body irradiation (P <0.0001) and more antithymocyte globulin (P =0.01).
Results. Recovery of neutrophils and platelets was faster with blood stem cells (P <0.0001). The incidence of acute graft-versus-host disease (GVHD) was similar, but there was more chronic GVHD in the blood stem cell group [hazard ratio (HR) = 1.29, P =0.02]. There were no significant differences in nonrelapse mortality (NRM), relapse incidence and leukaemia-free survival (LFS) between the two groups amongst patients with AML in remission. In patients with advanced leukaemia, NRM was lower (HR = 0.61, P =0.02) and LFS was prolonged (HR = 0.67, P =0.002) when blood stem cells were used. At 3 years, LFS for all patients, regardless of remission status, was 41% for both treatment groups. The outcome was not affected after multivariable analysis adjusted for confounders.
Conclusion. Blood stem cells compared with BM in MUD transplantation for patients with AML in remission resulted in the same rates of LFS. In patients with advanced leukaemia, the blood stem cell group had reduced NRM and improved LFS.
Using HLA-identical sibling transplants, several prospective randomized studies and retrospective studies have compared granulocyte colony-stimulating factor-mobilized peripheral blood stem cells (PBSCs) and bone marrow (BM) as the source of haematopoietic stem cells [1–10]. During the last decade, PBSCs have become the most common source of stem cells in allogeneic haematopoietic stem cell transplantation (HSCT) . In the report from the European Group for Blood and Marrow Transplantation (EBMT) in 2006, PBSCs were the source of stem cells in 71%, BM in 24% and cord blood in 5% of transplants . In 2009, the corresponding figures for the three stem cells sources were 71%, 22% and 7% .
Regarding the use of unrelated donors, a few studies with limited numbers of transplant patients have lent support to wider use of PBSCs rather than BM [13–17]. One complication associated with the use of PBSCs is the higher risk of chronic graft-versus-host disease (GVHD) than with BM using HLA-identical sibling donors or matched unrelated donors (MUDs) [8, 14, 18, 19]. Chronic GVHD is associated with a beneficial graft-versus-leukaemia effect, resulting in a reduced probability of relapse [20–22]. Despite the increased incidence of chronic GVHD using unrelated donors, the use of PBSCs has not resulted in a reduced probability of relapse compared with BM transplantation in patients with leukaemia in remission [14, 16, 17, 19]. To determine which source of stem cells is superior in patients with acute myeloid leukaemia (AML), we compared the outcomes of using PBSCs and BM in MUD recipients treated with myeloablative conditioning (MAC).
Design and methods
The EBMT is a group of 605 HSCT centres that report all consecutive transplantations to a common registry. Adult patients (18–60 years) with AML who received MAC and MUD transplants between 1997 and 2008 were included in this retrospective registry analysis; patients with secondary AML were excluded. Only the first HSCT was included, and transplants were restricted to those receiving HLA-A-, HLA-B- and DRB1-identical grafts. MUDs were identified by serological or genomic tissue typing for HLA class I and genomic typing for HLA class II. Detailed HLA typing was lacking for the majority of cases. In most European centres, a 6/6 antigen match is required for MUD transplantation in AML. When detailed HLA matching was available, it was similar in the PBSC and BM groups. In total, 760 patients who received BM and 1502 patients who received PBSCs fulfilled the inclusion criteria. This study was approved by the Institutional Review Board and the Ethics Committee of the Karolinska Institutet.
The characteristics of the patients who received BM or PBSCs are shown in Table 1. Median follow-up was longer in the BM group, because the PBSC transplants were carried out more recently. In the PBSC group, the patients were older and had more advanced disease at transplant, fewer patients received total body irradiation (TBI), and more received antithymocyte globulin (ATG) or antilymphocyte globulin (ALG). PBSC recipients who underwent transplantation during complete remission (CR) for the first time (CR1) or in the advanced phase of the disease were transplanted more recently than those who received BM. Other prognostic variables such as female donor to male recipient, cytomegalovirus (CMV) seropositivity in recipients and donors, French–American–British nomenclature (FAB) and cytogenetic abnormalities were not statistically significantly different between the two groups.
Table 1. Characteristics of adult patients with acute myeloid leukaemia who received PBSC or BM grafts from unrelated donors
Graft-versus-host disease prophylaxis consisted of cyclosporine or tacrolimus and methotrexate in 83% of the BM recipients and 69% of the PBSC recipients. The remaining patients received cyclosporine alone or with mycophenolate mofetil or steroids.
In EBMT in 2004, 34% of patients received reduced-intensity conditioning ; in 2009, 39% were conditioned with such regimens . To maximize the homogeneity of the patient population, we decided to include only patients who received MAC, defined as established conditioning regimens that result in pancytopenia if haematopoietic stem cells are not infused. The most common conditioning regimen was cyclophosphamide (120 mg kg−1) combined with fractionated TBI, which was given to 68% of the patients in the BM group and 57% in the PBSC group (P <0.0001). Busulfan (16 mg kg−1 orally or intravenous Busulvex) combined with cyclophosphamide (120 mg kg−1) was the second most commonly used regimen. Other regimens were only used in a small number of patients.
Study end-points included engraftment, GVHD, relapse, leukaemia-free survival (LFS) and survival. Engraftment was defined as a sustained recovery of absolute neutrophil count of more than 0.5 × 109 L−1. GVHD was scored from 0 to IV according to established criteria . Nonrelapse mortality (NRM) was defined as any death without previous relapse or progression of leukaemia. Relapse was defined as haematological relapse with at least 20% blast cells in the BM aspirate, or the presence of extramedullary leukaemic cells (i.e. extramedullary relapse). LFS was defined as survival without any evidence of relapse or progressive leukaemia (i.e. >5% blasts in the BM, with the percentage increasing and not controlled by therapy). Cytogenetic abnormalities were classified as: ‘good’, including t(8;21), t(15;17) and inv or del (16); or ‘poor’, including 11q23 abnormalities, complex karyotypes (at least three abnormalities) and abnormalities of chromosomes five and seven. All other abnormalities, including trisomies, were included in an ‘intermediate’ group .
Cumulative incidence curves were used in a competing risks setting, death being treated as a competing event to calculate the probabilities of haematopoietic recovery and chronic GVHD [25, 26]. Cumulative incidence curves were also used to calculate the probabilities of relapse and NRM as death and relapse were competing events. Comparisons between cumulative incidence curves were performed using the Gray test. Probabilities of survival and LFS were calculated using the Kaplan–Meier estimate, and the log-rank test was used for univariable comparisons. Multivariable analysis for acute GVHD was performed using logistic regression analysis.
Patient-, disease- and transplant-related variables of the two groups were compared using the chi-squared test for categorical variables and the Mann–Whitney test for continuous variables. Recipient age and sex, disease characteristics, FAB classification, cytogenetic abnormalities, white blood cell count at the time of diagnosis, donor age and sex and recipient and donor CMV status were considered as variables. Patients were divided into those in CR1, or second or third complete remission (CR2 and CR3, respectively) and those with more advanced disease (>5% blast cells in the BM). Transplant characteristics included year of transplant, source of stem cells, pretransplant regimen and GVHD prophylaxis.
Factors that differed between recipients of BM and PBSCs, with P-values <0.05, and factors known to influence outcome were included in the final models. For all prognostic analyses, continuous variables were categorized and the median was used as a cut-off point. Associations between patient/graft characteristics and outcomes were evaluated in multivariable analysis, using the Cox proportional hazards model. To test for a centre effect, we introduced a random effect or frailty for each centre into the model [27, 28]. All tests were two-sided. The type-I error rate was fixed at 0.05 for determination of factors associated with time to event outcomes. Statistical analyses were performed using spss software (SPSS Inc., Chicago, IL, USA) and splus software (Math-Soft Inc, Seattle, WA, USA).
Neutrophil and platelet recovery
The cumulative incidence of neutrophil recovery at day 60 was 94% (95% CI: 92–96%) in the BM group and 96% (95% CI: 95–97%) in the PBSC group (P <0.001). Six patients in the BM group lost their grafts, compared with three in the PBSC group. Median time to reach an absolute neutrophil count >0.5 × 109 L−1 was 20 days in the BM group and 16 days in the PBSC group (P <0.001; Fig. 1a). Cumulative incidence of platelet recovery to >20 × 109 L−1 at 1 year was significantly higher in the PBSC group (93%; 95% CI: 90–95%) than in the BM group (89%; 95% CI: 85–92%; P <0.0001; Fig. 1b). In multivariable analysis, after adjustment for differences between the two groups, we found that PBSC recipients experienced faster neutrophil recovery [hazard ratio (HR) = 1.89, 95% CI: 1.67–2.14; P <0.0001] and platelet recovery (HR = 2.30, 95% CI: 1.92–2.75; P <0.0001) than patients who received BM.
Acute GVHD of grades II–IV occurred in 35% of patients in the BM group, which was not statistically significantly different from 33% of those in the PBSC group (P =0.41). Acute GVHD of grades III–IV was present in 11% of patients who received BM and in 12.5% of those given PBSCs. In multivariable analysis, graft source (BM or PBSCs) was not associated with the presence of acute GVHD of grades II–IV. However, the use of T-cell depletion and ATG or ALG in the conditioning regimen significantly reduced GVHD. The cumulative incidence of chronic GVHD at 2 years was 36% in the BM group (95% CI: 32–40%) compared with 46% (95% CI: 43–50%) in the PBSC group (P =0.001; Fig. 2). However, the cumulative incidence of extensive chronic GVHD was 16% (95% CI: 13–19%) and 19% (95% CI: 17–22%), respectively, (P =0.21). Multivariable analysis confirmed that PBSC recipients experienced more chronic GVHD (HR = 1.29, 95% CI: 1.04–1.61; P =0.02).
Chronic GVHD, NRM, relapse incidence and LFS according to disease status at transplantation
Patients in CR1. Median age was lower in the BM group: 36 years (range 18–60) compared with 39 (range 18–60) in the PBSC group (P = 0.001). The median year of transplantation was earlier in the BM group: 2004 compared with 2006 in the PBSC group (P < 0.0001). There were no significant difference probabilities in chronic GVHD, NRM, incidence of relapse or LFS in univariable or multivariable analysis between patients undergoing transplant in CR1 in the two groups (Tables 2 and 3). Three years after transplantation, LFS was 53% (95% CI: 47–59%) in patients in the BM group and 48% (95% CI: 44–53%) in PBSC recipients (P = 0.32; Fig. 3a). Amongst patients in CR1 with cytogenetic abnormalities classified as intermediate, poor or missing (defined as a lack of data), there were no statistically significant differences in NRM, relapse incidence or LFS between the BM and PBSC groups.
Table 2. Chronic GVHD, NRM, relapse incidence and LFS in patients with adult AML in CR1, CR2/3 or advanced disease, who received BM or PBSC grafts from unrelated donors
Table 3. Multivariable analysis of adult patients with AML who received BM or PBSCs from unrelated donors: chronic GVHD, NRM, relapse incidence and LFS were analysed
Chronic GVHD, PBSC versus BM
NRM, PBSC versus BM
Age >39 years
Relapse, PBSC versus BM
Time from diagnosis to tx >175 days
LFS, PBSC versus BM
Age >39 years (median)
Chronic GVHD, PBSC versus BM
Year of transplant after 2004
NRM, PBSC versus BM
CsA + MTX versus other GVHD prevention
Relapse, PBSC versus BM
LFS, PBSC versus BM
AML advanced stage, (>5% blasts in the MB)
Chronic GVHD, PBSC versus BM
NRM; PBSC versus BM
Relapse, PBSC versus BM
Time from diagnosis to transplant >251 days
LFS, PBSC versus BM
Patients in CR2 or CR3. Amongst patients in CR2 or CR3, median age was significantly lower in the BM cohort: 36 years (range 18–57), compared with 41 years (range 18–60) in the PBSC group (P <0.0001). Median year of transplantation was earlier in the BM cohort: 2002 compared with 2005 in the PBSC group (P <0.0001). TBI was part of the conditioning regimen in 78% of patients in the BM group and 68% in the PBSC group (P =0.005). Use of ATG/ALG during conditioning was more common in the PBSC group (50%) than in the BM group (39%; P =0.02). There was more chronic GVHD in the PBSC group (50%; 95% CI: 43–57%), compared with the BM group (40%; 95% CI: 33–47%; P =0.05; Table 2). Amongst patients in CR2 and CR3 with good, intermediate or missing cytogenetic abnormalities, there were no differences in NRM, incidence of relapse or LFS between the BM and PBSC groups. However, in multivariable analysis, chronic GVHD as well as NRM, incidence of relapse and LFS were not significantly associated with the source of the graft (Tables 2 and 3).
Patients with advanced AML. In patients with advanced AML, median age was significantly higher in the PBSC group: 42 years (range 19–60) compared with 36 years (range 18–58) in the BM group (P <0.0001). Median year of transplant was 2005 and 2001 in the two groups, respectively, (P <0.0001). TBI was given to 67% of the patients in the BM group, compared with 56% in the PBSC group (P =0.01). There were no other statistically significant differences in patient characteristics between the two groups.
The probability of chronic GVHD was 25% (95% CI: 17–33%) in the BM group and 41% (95% CI: 34–47%) in the PBSC group; this difference was statistically significant in univariable analysis (P = 0.002; Table 2) but not in multivariable analysis (HR = 1.52; P = 0.09). The incidence of relapse was similar in the BM and PBSC cohorts in both univariable and multivariable analyses (Tables 2 and 3). The only significant factor in the multivariable analysis that was associated with a reduced risk of relapse was time from diagnosis to transplantation of >251 days (HR = 0.70, 95% CI: 0.53–0.94; P =0.02; Table 3).
In multivariable analysis, patients in the PBSC group had a lower NRM than those in the BM group (HR = 0.61, 95% CI: 0.40–0.91; P =0.02; Table 3). Thus, LFS was significantly increased at 3 years for patients with advanced AML amongst those receiving PBSCs (23%; 95% CI: 18–27%), compared with BM recipients (13%; 95% CI: 8–18%; P =0.001; Table 2). In the multivariable analysis, receiving PBSCs versus BM was the only significant factor for improved LFS (HR = 0.67, 95% CI: 0.52–0.86; P =0.002; Table 3). Amongst patients with intermediate cytogenetic abnormalities, median 3-year LFS was 10% (2–18%) in the BM group compared with 17% (8–26%) in the PBSC group (P =0.14). Amongst those with missing cytogenetic data, the corresponding figures were 16% (9–23%) and 25% (19–30%) in the two groups, respectively, (P =0.02).
Overall survival. In all patients in CR (CR1–3), survival was similar for recipients of BM or PBSCs (Fig. 4a,b). In patients with advanced AML, the PBSC group had a 3-year survival of 25 ± 3%, compared with 15 ± 3% in recipients of BM (P =0.004, Fig. 4c). With regard to all patients with AML, 3-year survival was 43% (95% CI: 40–48%) in those who received BM and 44% (95% CI: 41–47%) in PBSC recipients (Fig. 5a). The corresponding figures for LFS were 41% in both groups (Fig. 5b). Median follow-up for surviving patients was 36 months (range 1–140) in the BM group and 16 months (range 1–144) in the PBSC group.
Peripheral blood stem cell transplantation from HLA-identical sibling donors started in the early 1990s [29–31]. The advantages of using PBSCs were higher yields of CD34+ cells, CD3+ cells, CD19+ cells and CD56+ cells . Because of the high content of T cells, there was a reluctance to use PBSCs from unrelated donors. It was feared that there would be an increased risk of acute GVHD; however, initial studies did not show any increased risk of acute GVHD using PBSCs from unrelated donors, compared with BM [16, 17, 33].
An advantage regarding the outcome after HSCT of using PBSCs rather than BM from both related and unrelated donors was faster recovery of platelet and absolute neutrophil counts [1, 3–10, 13, 15–17, 29]. This very important advantage of engraftment capacity using PBSCs was confirmed in the present study. This may be even more important using RIC; however, only patients who received MAC were included in the present study. Although the results of all retrospective studies comparing PBSCs and BM suggest that use of PBSCs is preferred, this issue can only be properly addressed in prospective randomized studies. Recently, a prospective randomized study comparing transplantation with PBSCs and BM has been closed to recruitment, and publication of the results is awaited.
There is controversy as to whether or not PBSCs increase the risk of acute GVHD (relative to BM). In HLA-identical sibling transplants, PBSCs and BM have been reported to be associated with a similar probability of acute GVHD in most studies [1–5, 7–10, 29–31]; however, a few studies have suggested that there is an increased incidence of acute GVHD using PBSCs as compared to BM [6, 34]. Regarding unrelated donor transplants, most studies so far have shown a similar probability of acute GVHD using PBSCs and BM [13, 15–17]. This finding is supported by results of the present analysis. By contrast, a study by Eapen et al.  showed an increased incidence of acute and chronic GVHD using PBSCs compared with BM.
It is well established that there is an increased incidence of chronic GVHD with PBSCs from HLA-identical siblings and also in MUD transplants [8, 14, 18, 19, 35]. In patients undergoing HSCT for leukaemia in remission, this increased incidence of chronic GVHD did not lead to a superior graft-versus-leukaemia effect, resulting in a reduced probability of relapse in patients receiving PBSC transplants compared with BM grafts [1–5, 7–10, 13–17]. During the last decade, PBSCs have been increasingly used not only from HLA-identical siblings, but also from unrelated donors [8, 11, 14, 16, 17]. In children receiving HLA-identical sibling transplants for leukaemia, the increased risk of chronic GVHD from using PBSCs resulted in a reduced probability of survival compared with BM . Such an effect has not been seen from the limited experience of using MUDs in children with leukaemia . Despite the relatively few advantages of using PBSCs compared with BM and the small number of reported studies, especially of using unrelated donor transplants, PBSCs are currently the predominant source of haematopoietic stem cells [11, 12]. Other advantages of using PBSCs such as the fact that there is no need for surgery or anaesthesia for the donor and convenience for the transplant physician are probably the major reasons for this. In general, unrelated donors prefer donation of PBSCs rather than BM for safety reasons and because of the convenience of the procedure.
A retrospective study from the Centre for International Blood and Marrow Transplant Research showed a lower NRM and a higher LFS with PBSCs compared to BM in patients with advanced leukaemia . The results of a randomized study in patients with advanced malignancy also showed that survival was prolonged in PBSC recipients compared with those who received BM . No difference in survival was seen in patients with less advanced disease . The results of a meta-analysis showed that overall and disease-free survival using PBSCs compared with BM were only statistically significantly improved in patients with late-stage disease . An improved LFS was also seen using MUD transplants in patients with advanced AML in the present study (P =0.002; Fig. 3c, Table 3). The significant effect on LFS using PBSCs compared with BM for advanced disease is probably due to the reduced effect on NRM. However, a nonsignificant trend towards a lower relapse risk in the PBSC group may also have contributed to the improved LFS and survival compared with the BM group. It is surprising that we did not find any significant association between PBSCs and a higher incidence of chronic GVHD and a subsequent lower incidence of relapse. The benefit of PBSCs in patients with advanced AML appears to be related to lower NRM. That such an effect is only seen in patients with advanced leukaemia and not in those in remission is interesting, but the cause of this difference remains unclear. It is possible that faster engraftment induced by PBSCs relative to BM may lower the risk of life-threatening infections and toxicity. This may be critical in patients with advanced leukaemia, but of marginal value in low-risk patients in remission. In patients with AML in CR, the probability of relapse, NRM and LFS were similar amongst those receiving PBSCs or BM (Table 2). Three studies of patients with HLA-identical sibling transplants showed improved survival and/or LFS using PBSCs compared to BM in those with advanced cancer, but not early disease [1, 2, 4]. The results of the present study show that, using MUDs, there is also an improved outcome with PBSCs in patients with advanced AML. The advantages of using PBSCs compared with BM for advanced leukaemia should, therefore, be considered seriously.
The positive effects on NRM and LFS of using PBSCs rather than BM may be due to the higher cell dose with PBSCs given (P <0.001, Table 1). The doses of nucleated cells and CD34 cells are important for outcomes after HSCT. A high BM cell dose has been variously associated with improved survival and reduced graft rejection, risk of relapse, death from GVHD, invasive fungal infection and CMV disease [38–40]. A PBSC CD34+ cell content of >6 × 106 kg−1 was associated with a reduced risk of leukaemic relapse and improved LFS . Therefore, the outcome of both BM and PBSC grafts may be improved by increasing the cell dose.
For patients with AML in CR, there were no differences in NRM, incidence of relapse or LFS when using PBSCs compared with BM (Table 3, Fig. 3a,b). Although we did not observe an increase in the probability of chronic GVHD in any remission group, there was an overall increased incidence of chronic GVHD using MUD PBSCs (Fig. 2). Because of the increased risk of chronic GVHD, there is no major advantage for patients in remission of receiving PBSCs instead of BM. However, in a study of long-term outcome and late effects following HLA-identical sibling transplantation with either PBSCs or BM (random assignment), there was no difference in general health status or late events in the two groups , despite an increased proportion of chronic GVHD in the PBSC group.
The limitations of the study should be acknowledged. First, it was a retrospective multicentre registry study, with various protocols used in the different centres. Of note, patients receiving PBSCs may differ in many important aspects from those receiving BM. Although several important variables have been controlled for, there may be some differences that cannot be accounted for. Furthermore, the possibility that some results may be biased for differences between the two groups cannot be excluded. Therefore, the data should be interpreted with caution. A further limitation is the amount of missing data, for instance regarding cytogenetic abnormalities. The strengths of the study include the large number of participants and the restriction to a single disease with MAC as the only conditioning regimen.
In conclusion, using PBSCs rather than BM for MUD transplantation in adults with AML resulted in faster recovery of neutrophils and platelets, a similar incidence of acute GVHD, and an overall increased probability of developing chronic GVHD. In patients with AML in remission, comparable NRM, incidence of relapse and LFS were observed using PBSCs and BM. Therefore, either PBSCs or BM may be effective for these patients. In patients with advanced AML, NRM was reduced and LFS improved with PBSCs compared with BM. Thus, for AML patients in relapse, we recommend the use of PBSCs.
Conflict of interest statement
The authors have no potential conflicts of interest to declare.
We thank all the EBMT centres for submitting data to the Acute Leukaemia Working Party (ALWP; see list in Data S1). We also thank all the members of staff at the various departments for excellent patient care, all the volunteer donors for their generous gifts to the patients and the staff at ALWP for data collection. Finally, we thank Inger Holmström for typing this manuscript and Hugh Kidd at Good Written English for language revision. This study was supported by grants from the EBMT. Olle Ringdén was supported by grants from the Swedish Cancer Society, the Children’s Cancer Foundation, the Swedish Research Council, the Cancer Society in Stockholm and Karolinska Institutet.