Comparable outcomes after busulfan- or treosulfan-based conditioning for allo-HSCT in children with ALL: results of FORUM

Radiochemotherapy with total body irradiation (TBI) and etoposide (VP16) has been established as “gold standard” conditioning regimen before allogeneic hematopoietic stem cell transplantation (HSCT) for pediatric patients with acute lymphoblastic leukemia (ALL) aged >2 years.^1,2 Its superiority over chemotherapy-based conditioning has recently been demonstrated in the prospective, international, randomized phase-3 FORUM (For Omitting Radiation Under Majority age) study.¹ The FORUM study enrolled 417 patients aged ≤18 years at diagnosis (4-21 years at HSCT) in complete remission (CR), who underwent HSCT for ALL from a compatible donor; either an HLA-matched sibling donor (MSD), or an HLA-matched unrelated donor (MD). The use of either of the 2 protocol-prespecified chemotherapy conditioning regimens resulted in a significantly inferior event-free survival (EFS), overall survival (OS), nonrelapse mortality (NRM), and cumulative incidence of relapse (CIR) as compared with TBI/VP16.¹ 

Despite its proven superiority, TBI conditioning cannot be implemented in all cases because of either unavailability or individual contraindications, for example, very young patient age, previous irradiation, or comorbidities. Myeloablative chemotherapy conditioning regimens based on fludarabine (FLU), thiotepa (THIO), and either busulfan (BU) or treosulfan (TREO) had shown high efficacy in previous pediatric HSCT studies^3,4 and NRM rates for either regimen were acceptable in the FORUM study.¹ The optimal chemotherapy-based conditioning for patients ineligible for TBI is, however, yet unknown. In light of these considerations, this study compares the outcomes of patients who received a BU-based regimen with those who were given a TREO-based conditioning in FORUM centers in both randomizing and nonrandomizing countries from 2013 until 2018.

The FORUM study is a prospective, global, multicenter, open-label, phase 3 clinical trial (EudraCT 2012-0032-22; ClinicalTrials.gov identifier: NCT01949129). A detailed trial description has been published previously.¹ This analysis focuses on the comparison of the 2 myeloablative chemotherapy conditioning protocols. In accordance with the FORUM study’s inclusion criteria, patients had high-risk/relapsed ALL and were aged ≤18 years at diagnosis, (aged 4-21 years at HSCT). The recommended stem cell source was either bone marrow (BM) or cord blood in case of a HLA-MSD; or BM, cord blood, or peripheral blood (PB) stem cells in case of a HLA-MD (matching in ≥9/10 HLA loci). This study describes outcomes of patients treated in countries, in which randomization was not possible for legal or technical reasons, and in countries, in which randomization to receive either of the 2 chemotherapy conditioning regimens occurred before the implementation of the stopping rule (supplemental Table 1). The decision to use either a BU-based or TREO-based regimen was not randomized but was made at the national level of each participating country to accommodate country-specific drug availability and practices, before the study’s initiation. The study was performed in compliance with ethical guidelines outlined in the Declaration of Helsinki and applicable national laws and was approved by the relevant institutional review boards and ethics committees in each center. Before participating, written consent was obtained from all patients and/or their parents/legal guardians.

Before HSCT, patients received intravenous FLU (30 mg/m² per day for 5 days; total 150 mg/m²) and THIO (5 mg/kg twice daily for 1 day; total 10 mg/kg) together with either BU or TREO according to a predetermined choice taken at the national level. BU was administered over 4 consecutive days according to the summary of product characteristic (supplemental Table 8), and TREO (14 g/m² per day for 3 days; total 42mg/m²) was administered once daily for 3 consecutive days. BU pharmacokinetics were not monitored routinely. Graft-versus-host disease (GVHD) prophylaxis was standardized and administered according to donor type: patients undergoing transplantation from an MSD were given cyclosporine alone (and additional short-term methotrexate in case of a PB stem cell graft), whereas recipients of MD grafts received a combination of cyclosporine, short-term methotrexate, and antithymocyte globulins (Grafalon or Thymoglobuline).

CR was defined as <5% blasts in the BM, coupled with the absence of extramedullary disease. Minimal residual disease (MRD) positivity in the BM was defined as ≥0.01% in flow cytometry and ≥ 1 x 10⁻⁴ in polymerase chain reaction (PCR) assays. Relapse was defined by the presence of ≥5% blasts and/or evidence of extramedullary disease. Adverse events (AEs) were graded by the National Cancer Institute common terminology criteria for adverse events (CTCAE version 3.0). Acute GVHD (aGVHD) was assessed according to the Glucksberg criteria,⁵ whereas chronic GVHD (cGVHD) was classified as either limited or extensive.⁶ Veno-occlusive disease (VOD) was graded according to the 2018 European Group for Blood and Marrow Transplantation criteria.⁷ 

OS was calculated from the date of transplantation with death from any cause considered as an event. EFS was calculated from date of transplantation to disease progression or relapse, secondary malignancy, or death from any cause, whichever occurred first. When calculating CIR, death without relapse/progression and secondary malignancy were considered competing events. Relapse/progression and secondary malignancy were competing events for calculating the cumulative incidence of NRM; and death without malignancy was a competing event for calculating the cumulative incidence of secondary malignancy. Relapse, death, and secondary malignancy were competing events for aGVHD and cGVHD. GVHD-free and relapse-free survival (GRFS) was calculated from date of transplantation to the first event: grade 3 or 4 aGVHD were considered events at day of onset, furthermore, extensive cGVHD, relapse/progression, and death from any cause and secondary malignancies were considered an event for GRFS. For all time-to-event end points, patients lost to follow-up without an event were censored at last follow-up. The rate of nonhematological AEs at day +100 (according to common terminology criteria for adverse events) and aGVHD at day 100 are given, as well as cumulative incidence of later onset of aGVHD (aGVHD any time) considering relapse and death as competing events.

The comparison as presented had been prespecified in the FORUM study protocol and covers all standard end points used to evaluate the main randomization question. The significance level was set at 0.05 for all statistical analyses. OS, EFS, and GRFS were estimated using the Kaplan-Meier method and the differences among groups were compared using the log-rank test. 3-year estimates and 95% confidence intervals using the log-log transformation are reported.

Baseline characteristics evaluated included sex, age at HSCT and diagnosis, immune phenotype and genetic aberration of ALL, MRD before HSCT (MRD threshold of 1 × 10⁻⁴ was selected to stratify patients according to their MRD level: patients with undetectable MRD and those who were MRD positive but at <1 × 10⁻⁴ in quantitative PCR or <0.01% in flow cytometry, were defined as MRD- negative/low, whereas patients with positive MRD at ≥1 × 10⁻⁴, in quantitative PCR or ≥0.01% in flow cytometry were defined as MRD-positive/high), donor type and HLA-matching with the recipient (9/10 or 10/10), stem cell source, remission status, and site and time of first relapse in patients in second CR (CR2).

The proportion of patients with grade 3/4 aGVHD and other AEs of grade 3/4 at day +100 as well as the distribution of baseline characteristics were compared using the χ² test. CIR, NRM, and secondary malignancy were estimated taking into account competing events and compared using the Gray test.

For multivariable analysis, Cox regression was used to explore the impact of baseline characteristics and conditioning regimen on OS and EFS. Factors evaluated were conditioning regimen, remission status, source of stem cells, age at HSCT, donor type, MRD, immune phenotype, and whether the patient’s respective country participated in the randomization. Propensity score matching was used to further reduce confounding biases of these factors when comparing both chemotherapy conditioning regimens. Forest plots show the estimated hazard ratios and their 95% CIs of the Cox regression performed univariably in subgroups. The results within subgroups and a test for the interaction between conditioning regimen and risk-factors from a Cox-model were calculated. Median follow-up was estimated using the inverse Kaplan-Meier method.

Between April 2013 and December 2018, 308 patients from 23 countries (Table 1; supplemental Table 1) underwent HSCT according to the FORUM protocol and received either BU-based (n = 180) or TREO-based (n = 128) chemotherapy conditioning. The study included 193 children from randomizing countries and additional 115 patients from nonrandomizing countries. Patient characteristics were balanced between the 2 treatment groups (Table 1). Overall, 194 patients were male (63%) and 114 (37%) female. Most children (n = 159; 52%) were aged 4 to 10 years. Children aged 10 to 14 years accounted for 24% (n = 73), and patients aged >14 years represented 25% (n = 76) of the cohort. There was a trend toward more adolescent patients (aged >14 years) in the TREO group (30% vs 21% in the BU group). Median age at HSCT was 9.9 years (range, 4-19.5; Table 1). In total, 221 patients (72%) had B-cell precursor ALL, 25% of patients had T-cell ALL, and 3% had other phenotypes. Genetic analysis showed BCR::ABL in 29 patients (10%), ETV6::RUNX1 in 32 patients (11%), and KMT2A rearrangements in 6 patients (2%). Overall, 157 patients (51%) underwent HSCT in first CR (CR1), 131 patients (43%) in CR2, and 20 in CR3 or beyond. Among patients evaluated for MRD (n = 222), 76% were negative/low and 24% were positive/high. Notably, a higher percentage of patients in the BU group received HSCT with MRD of <1 x 10⁻⁴ (81%) compared with the TREO cohort (66%; P = .012; Table 1).

In total, 87 patients (28%) received transplantation from an MSD and 221 (72%) from an MD. The TREO group had a higher percentage of MD donors (78%) than the BU group (67%; P = .036). Most patients received stem cells from BM (77%), followed by PB (20%), and cord blood (2%). BM was more frequently used in the TREO group (84%) compared with the BU group (73%; P = .051).

Patient outcomes were analyzed based on data cutoff in February 2023, allowing for a median follow-up of 4.2 years (range, 0.3-9.1). The 3-year OS probability was similar between the FLU/THIO/BU (0.71 [0.64-0.77]) and FLU/THIO/TREO (0.72 [0.63-0.79]) groups (P = .977; Table 2; Figures 1A and 2) as was the probability of 3-year EFS (0.60 [0.53-0.67] for BU and 0.55 [0.46-0.63] for TREO; P = .626; Figures 1B and 2). No significant differences in the 3-year CIR were observed between the BU (0.31 [0.25-0.38]) and TREO cohort (0.36 [0.27-0.44]; P = .779; Figure 1C). The 3-year NRM was similar for the BU (0.08 [0.05-0.13]) and the TREO (0.09 [0.05-0.15]) groups (P = .831; Figure 1D), with 16 patients of the BU group and 12 patients of the TREO group, respectively, dying in CR (Table 2; supplemental Table 2).

Graft failure occurred in 7 patients of the BU group and 3 patients of the TREO group. Overall, patients who had received BU conditioning showed a faster leukocyte engraftment, reaching a white blood cell count (> 1 x 10⁹/L for 3 consecutive days) at a median of 18 days (TREO median of 21 days; P < .001). Neutrophil engraftment (> 5 × 10⁹/L for 3 consecutive days) was detected at a median of 19 days in the BU group and 22 days in the TREO group (P < .001). Engraftment of lymphocytes and platelets also occurred earlier in the BU cohort (supplemental Table 3).

The number of patients experiencing severe aGVHD was low, and there was no significant difference between the treatment groups (Table 2). Of the patients evaluable on day +100, 16 of 178 (5%) in the BU group and 13 of 124 (4%) in the TREO group developed grade 3-4 aGVHD (P = .664; Table 2). The BU group had a slightly higher incidence of cGVHD (0.20 [0.15-0.26]) than the TREO group (0.13 [0.08-0.19]; P = .077; Figure 1E), especially for patients in ≥CR2 (supplemental Figure 2E). However, propensity score Cox-regression analysis found no influence of conditioning on incidence of cGVHD (Table 3). GRFS at 3 years was similar between both cohorts (BU, 0.48 [0.41-0.55]; TREO, 0.45 [0.37-0.54]; P = .89; Figure 1F).

Both regimens were overall well tolerated with an acceptable safety profile. In the first 100 days after transplant, the most frequent nonhematological grade 3-4 AEs observed were infection, stomatitis, nausea and vomiting, elevated liver enzymes, diarrhea, and hypoxia (supplemental Table 4). Overall, 10 patients who had received TREO-based conditioning developed posttransplant lymphoproliferative disorder (PTLD), whereas PTLD did occur in 2 BU-treated patients (any grade). In the BU cohort, 15 patients developed VOD of any severity, with 4 patients (2%) developing (very) severe VOD, whereas 9 patients in the TREO cohort developed any grade of VOD, with 2 severe cases (2%). In either group, VOD was listed as cause of NRM in 1 case (supplemental Tables 2 and 4). Up until 3 years after HSCT, only 1 secondary malignancy (sarcoma) was observed (in the TREO cohort).

Among 157 patients who received transplantation in CR1, there were no differences in 3-year OS and 3-year EFS. The probability of 3-year OS for the BU group was 0.78 [0.68-0.86], whereas it was 0.75 [0.63-0.84] for the TREO group (P = .675; supplemental Figure 1A). The 3-year EFS probability for patients who received transplantation in CR1 was comparable at 0.74 [0.63-0.82] and 0.66 [0.54-0.76] for BU and TREO, respectively (P = .415; supplemental Figure 1B). Evaluating the CIR in CR1 patients, the 3-year risk was 0.18 [0.10-0.27] for the BU arm compared with 0.27 [0.17-0.38] for the TREO arm (P = .255; supplemental Figure 1C). The 3-year NRM of the CR1 study population was 0.08 [0.04-0.16] for the BU and 0.07 [0.03-0.15] for the TREO arm (P = .747; supplemental Figure 1D). The 3-year cumulative incidence of any cGVHD (limited + extensive) for patients who received transplantation in CR1 was not significantly different between the treatment arms (supplemental Figure 1E). The probabilities of GRFS at 3 years in CR1 patients were 0.59 (BU, [0.47-0.68]) and 0.52 (TREO [0.40-0.63]; = .486) (supplemental Figure 1F).

For 151 patients who received transplantation in ≥CR2, 3-year OS and EFS were 0.65 [0.54-0.73] and 0.48 [0.38-0.58] for the BU arm, compared with 0.67 [0.53-0.78] and 0.40 ([0.27-0.53] for the TREO arm, respectively (P = .927 and P = .433; supplemental Figure 2A-B). The 3-year CIR was high in both groups, with 0.43 [0.33-0.53] and 0.47 [0.33-0.60] in the BU and TREO cohorts, respectively (supplemental Figure 2C). Comparing the 3-year probability of NRM in chemotherapy-conditioned patients in ≥CR2, there were no statistically significant differences (supplemental Figure 2D). In ≥CR2 patients, the 3-year cumulative incidence of cGVHD was 0.24 [0.16-0.23] for the BU group (n = 23) compared with slightly less (0.13 [0.06-0.23]) in the TREO group (n = 7; P = .078; supplemental Figure 2E). GRFS at 3 years in both populations was comparable (supplemental Figure 2F). Timing of CR2 relapse had a significant impact on outcome, with early CR2 relapses (<18 months after diagnosis) presenting with a better OS and EFS than late CR2 relapses (18-30 and >30 months; both P < .001; supplemental Table 6). Overall, remission status at transplant significantly affected outcome, with patients in ≥CR2 showing worse OS, EFS, and CIR (Table 3; supplemental Table 6), whereas the type of conditioning regimen used did not affect the outcome of the 2 cohorts (supplemental Table 6).

There was no difference in 3-year OS for the 2 subgroups of MRD-negative/low vs MRD-positive/high patients. MRD-negative/low patients showed a 3-year OS of 0.73 [0.63-0.80] for BU- and 0.83 [0.69-0.91] for TREO-treated patients, and a 3-year OS for MRD-positive/high patients of 0.69 [0.47-0.83] for BU- and 0.63 [0.42-0.78] for TREO-conditioned patients (P = .289; supplemental Table 6). However, we observed differences in 3-year EFS and CIR between the 2 conditioning groups among MRD-positive/high and MRD-negative/low patients: among MRD-positive/high patients, the BU-conditioned patients showed a higher 3-year EFS (BU, 0.65 [0.43-0.80]; TREO, 0.39 [0.21-0.56]; P = .049), with a lower incidence of relapse (BU, 0.27 [0.12-0.46]; TREO, 0.54 [0.33-0.71]; P = .016). Meanwhile, among MRD-negative/low patients, BU-conditioned patients showed a lower 3-year EFS and higher 3-year incidence of relapse than TREO-treated patients (supplemental Table 6).

Propensity score Cox-regression analysis confirmed that, overall, conditioning regimen had no impact on OS, EFS, and CIR (Table 3). When analyzed separately for the individual diseases, BU and TREO conditioning were associated with equal outcomes (EFS, OS, and CIR) in patients with B-cell precursor ALL and T-cell ALL alike (supplemental Table 7). However, when considering the subgroup of MRD-positive/high patients, patients who had received BU conditioning benefited from a higher EFS and lower CIR (Table 3; supplemental Table 6). Overall, patients in CR1 were shown to have a higher OS and EFS with a lower CIR than patients in ≥CR2 (Table 3).

The FORUM study represents, to our knowledge, the largest international, multicenter, randomized trial comparing TBI plus etoposide vs chemotherapy conditioning regimens (consisting of FLU, THIO, and either intravenous BU or TREO) in pediatric patients with high-risk ALL, to date.¹ Despite challenges faced when initiating the trial in >20 countries and 5 continents, the FORUM trial was a remarkable success thanks to the collaboration among numerous international study groups. Patients underwent allogeneic HSCT according to standardized transplant indications, donor selection criteria, GVHD prophylaxis, and conditioning regimen. The study protocol provided investigators with guidance for monitoring MRD levels and intervention strategies to reduce MRD before treatment and after transplant. Strong adherence to randomization, rigorous data collection, and a careful and timely interim analysis detected the inferiority of chemotherapy conditioning with a significantly higher relapse rate and poorer OS, EFS, and GRFS.¹ These clear effects led to an early termination of randomization.¹ 

TBI/VP16 conditioning is now the data-driven state-of-the-art conditioning in allogeneic HSCT in pediatric patients,² but certain instances remain in which TBI conditioning cannot be performed. TBI is not available in some transplant centers or regions. Furthermore, individual contraindications like infancy, prior radiation treatment, or comorbidities may require chemotherapy conditioning. In both, retrospective and prospective trials, it was shown, that chemotherapy conditioning can represent a suitable alternative to TBI, at least for younger patients and in case of favorable prognostic characteristics before HSCT (CR1 or age of <4 years).^2,8-11 In a large retrospective analysis of >3000 pediatric patients with ALL, no significant difference in OS was observed between patients who received transplantation in CR1 using TBI vs chemotherapy conditioning, whereas in CR2, the TBI arm had significantly better OS, EFS, and less relapse and TRM.¹⁰ 

Here, we demonstrate comparable outcomes for a homogenous cohort of pediatric patients with ALL who received BU-based regimens vs those who were given TREO-based conditioning in FORUM centers in both randomizing and nonrandomizing countries.

Children who received transplantation in CR1 had a remarkable 3-year OS of 0.78 [0.68-0.86] for the BU arm and 0.75 ([0.63-0.84] for the TREO arm, which are, to our knowledge, the highest OS-rates in ALL after chemotherapy-based conditioning regimen published so far. The same applies for EFS, with a 3-year EFS of 0.74 [0.63-0.82] for the BU and 0.66 [0.54-0.76] for the TREO cohort. Compared with the outcomes of TBI-conditioned patients from the FORUM study (3-year OS, 0.9 [0.84-0.93)]; 3-year EFS, 0.81 [0.74-0.85])¹² the established superiority of TBI remains indisputable. However, these results show that CR1 patients with contraindications to, or without access to TBI can benefit from chemotherapy-based regimens. On the contrary, the outcome of chemotherapy-conditioned patients who received transplantation in CR2 remains unsatisfactory, with a 3-year EFS of 0.48 for BU [0.38-0.58] and 0.40 [0.27-0.53] for TREO, respectively, in line with the FORUM results¹ and a previous European Group for Blood and Marrow Transplantation pediatric diseases working-party study by Willasch et al.¹⁰ This is a clear message from our study to use a TBI conditioning regimen for patients in CR2, whenever possible.

It has been previously demonstrated that high MRD levels before transplant increase the risk of posttransplant relapse.^13-15 In our study, TREO-conditioned patients presented with a slightly higher rate of pretransplant MRD than BU-treated children (Table 1). Among all MRD-positive/high patients, TREO conditioning was associated with lower EFS and higher CIR. Because there was no difference in OS for MRD-positive/high patients, we speculate that TREO-treated patients could be rescued with subsequent salvage therapies (eg, chimeric antigen receptor T cells) in case of relapse. It is difficult to draw final conclusions considering the low numbers of MRD-positive/high patients analyzed (n = 54) and further studies are warranted. Given the clear advantage of TBI conditioning for MRD-positive/high patients, as demonstrated in the FORUM study, TBI should remain the first-line approach. However, if TBI is not feasible, our results suggest that BU conditioning may be preferable to TREO-based regimens for MRD-positive/high patients. It is important to note, however, that confounding factors, such as differences in graft sources (eg, more PB and cord blood in the BU group) and a trend toward higher GVHD rates, could have influenced these results, making it difficult to isolate the specific effects of the conditioning.

Remarkably, we recorded a similar EFS in children who received transplantation with either chemotherapy conditioning in randomizing and nonrandomizing countries. The observed trend toward better OS in chemotherapy-conditioned children from randomizing countries (Table 3), potentially linked to better access to effective and new salvage therapies in those countries, raises awareness of treatment disparities.¹⁶ This highlights the need for equitable access to advanced therapies globally.

Despite their proven inferiority to TBI/VP16, it has to be noted that both reported chemotherapy-based conditioning regimens appear feasible, safe, and relatively nontoxic. The trend toward increased incidence of GVHD in the BU group could be because of its impact on the thymus, which may disrupt T-cell neogenesis.^17-19 Variations in center size and heterogeneity of supportive care notwithstanding, NRM incidence in chemotherapy-conditioned patients in the FORUM trial was lower than in other reports.^20-23 Centers reported 6 cases of (very) severe VOD (4 in the BU arm, 2 in the TREO arm) and only a single secondary malignancy in the TREO arm. The apparent differences in PTLD incidence are most likely not attributable to the conditioning regimens but linked to varying types and modes of application of serotherapy. Future analyses will help elucidate the role of serotherapy and its impact on posttransplant infectious complications, immune-reconstitution, and GVHD.

Although relapses after chemotherapy conditioning remain an issue both in our and other previous studies,^10,11 emerging cellular and immunotherapies may represent successful salvage therapies, potentially leading to deeper remissions and higher survival.^23-28 

The use of either of the 2 protocol-prespecified chemotherapy conditioning regimens resulted in significantly worse outcomes compared with TBI in the FORUM trial.^1,2 Impairment of growth, endocrine and cognitive function, cataracts, and secondary malignancies are more frequent after TBI compared with irradiation-free conditioning regimens.^29-31 However, these late effects can occur after myeloablative conditioning as well.^10,32,33 More efforts are needed to study how we can maintain the survival benefits of TBI conditioning while mitigating the life-long adverse effects. The ideal age threshold to perform TBI conditioning has not been conclusively demonstrated.^2,33,34 However, based on the demonstrated superiority of TBI/VP16,² its recommendation has now been extended to include children aged 2 to 4 years to enhance outcomes in this group. This extension will be incorporated in the upcoming FORUM-2 trial, which is currently in preparation.

In the analysis presented here, we observed comparable long-term outcomes after BU/THIO/FLU or TREO/THIO/FLU conditioning in children with ALL undergoing allogeneic HSCT enrolled in the FORUM trial. In conclusion, either of the 2 regimens can be effectively and safely used in patients aged >4 years with contraindications to radiotherapy or treated in centers/countries unable to deliver TBI. These data collected in various countries can provide practical guidance for clinicians dealing with cases in which recommended TBI conditioning is not an option.

This study received funding from the St. Anna Children’s Cancer Research Institute, Vienna, Austria. The funders were not involved in the study design, collection, analysis or interpretation of data, the writing of this article, or the decision to submit it for publication. The authors thank all participating national coordinators and investigators from other FORUM trial countries: Sandra Formisano (Argentina); Nina Minakovskaya and Dzmitry Prudnikau (Belarus); Dobrin Konstantinov (Bulgaria); Tony Truong (Canada); Julia Palma (Chile); Ioannidou Dikaia Eleni (Greece); Hany Ariffin (Malaysia); Alberto Olaya Vargas (Mexico); Arjan Lankester (The Netherlands); Jacek Wachowiak (Poland); Smaranda Teodora Arghirescu, Anca Colita, and Cristian Jinca (Romania); Mohammed Essa (Saudi Arabia); and Simona Avčin (Slovenia). The authors thank the trial sponsor St. Anna Children’s Cancer Research Institute; the sponsor’s trial manager Tijana Frank; and all physicians, nurses, trial coordinators, and data managers involved. And lastly, the authors are grateful to all patients, their families, and caregivers involved in the FORUM trial.

Contribution: K.Kalwak, L.M.Moser, C.Peters, M.Ansari. and F.Locatelli designed the study and were primarily responsible for writing the manuscript; U.Pötschger provided all biostatistical support and data analysis. All authors contributed patients, collected and checked data, and edited and read and approved the manuscript.

Conflict-of-interest disclosure: K. Kalwak discloses speakers bureau fees from Novartis, medac, and Pierre Fabre; and reports travel grants from Pierre Fabre. P.B. reports travel grants from medac, Novartis, Vertex; and speakers bureau fees from medac, Novartis, Amgen. K. Kleinschmidt serves on advisory boards for Jazz and Vertex Pharmaceuticals; reports writing support grant from Jazz; and reports travel grants from Sobi. J.-H.D. reports consultancy with, and honoraria from, Orchard, Pierre Fabre Medicaments, Novartis, Pfizer, Vertex, and SymbioPharm; reports travel grants from Pierre Fabre Medicaments; and is a current equity holder in the private company Teva. A.B. reports speakers bureau fees from Novartis, Amgen, medac, and Neovii; and reports meeting support from medac and Neovii. H.P. reports travel grants from Jazz Pharmaceuticals and Neovii. T.G. serves on the advisory board for Novartis, Jazz, and Merck Sharp & Dohme; and reports travel grants from Neovii. J.B. reports advisory board participation for Janssen, Pfizer, and Novartis; is a study steering committee member with Novartis; and discloses speakers bureau fees from Novartis. C.P. reports study support from Sanofi; reports travel grants, study support and speakers bureau fees from Neovii. F.L. reports speakers bureau fees from Amgen, Bristol Myers Squibb, Miltenyi Biotec, Neovii, Novartis, Sobi, and Vertex; consultancy for Sanofi, Vertex, and Amgen; and reports honoraria from, and advisory board participation for, Amgen. The remaining authors declare no competing financial interests.

Correspondence: Laura M. Moser, Goethe University Frankfurt, University Hospital, Department of Pediatrics, Division for Stem Cell Transplantation and Immunology, Theodor-Stern-Kai 7, 60590 Frankfurt am Main, Germany; email: moser@med.uni-frankfurt.de.