Leukemia (2005) 19, & 2005 Nature Publishing Group All rights reserved /05 $

Transcription

1 (2005) 19, & 2005 Nature Publishing Group All rights reserved /05 $ Reduced-intensity conditioning lowers treatment-related mortality of allogeneic stem cell transplantation for chronic lymphocytic leukemia: a population-matched analysis P Dreger 1,2, R Brand 3, D Milligan 4, P Corradini 5, J Finke 6, G Lambertenghi Deliliers 7, R Martino 8, N Russell 9, A van Biezen 3, M Michallet 10 and D Niederwieser 11, on behalf of the Chronic Working Party of the EBMT 1 Department of Hematology, Allgemeines Krankenhaus St Georg, Hamburg, Germany; 2 Department of Medicine V, University of Heidelberg, Germany; 3 Department of Medical Statistics, University Medical Centre, Leiden, The Netherlands; 4 Department of Haematology, Heartlands Hospital, Birmingham, UK; 5 Bone Marrow Transplantation Unit, Istituto Nazionale dei Tumori, University of Milano, Italy; 6 Department of Haematology and Oncology, University of Freiburg, Germany; 7 BMT Center, Ospedale Maggiore, Milano, Italy; 8 Servicio de Hematologica Clinica, Hospital de la Santa Creu i Sant Pau, Barcelona, Spain; 9 Haematology Department, Nottingham City Hospital, Nottingham, UK; 10 Unite de Greffe de Cellules Souches Hématopoïétiques, Hôpital Edouard Herriot, Lyon, France; and 11 Department of Hematology and Medical Oncology, University of Leipzig, Germany To elucidate whether reduced-intensity conditioning (RIC) decreases treatment-related mortality (TRM) after allogeneic stem cell transplantation (allo-sct) for chronic lymphocytic leukemia (CLL), we retrospectively compared 73 RIC cases from a recent EBMT survey with 82 patients from the EBMT database who had undergone standard myeloablative conditioning (MC) for CLL during the same time period. The two populations were matched by adjusting the primary risk factor, the conditioning regimen, in a series of Cox models for age, sex, donor type, remission status at transplant and analyzed for its effect on TRM, relapse incidence, event-free (EFS) and overall survival (OS). After adjustment, a significant reduction of TRM became evident for the RIC population (hazard ratio (HR) 0.4 (95% confidence interval ); P ¼ 0.03). On the other hand, RIC was associated with an increased relapse incidence (HR 2.65 ( ); P ¼ 0.054). There was no significant difference between RIC and MC in terms of EFS (HR 0.69 ( ); P ¼ 0.22) and OS (HR 0.65 ( ); P ¼ 0.21). We conclude that RIC appears to favorably influence TRM after allo-sct for CLL. This observation, as well as possible detrimental effects of RIC on relapse risk, should be confirmed by prospective studies. (2005) 19, doi: /sj.leu Published online 14 April 2005 Keywords: CLL; reduced-intensity conditioning; allogeneic stem cell transplantation; treatment-related mortality Introduction The prognosis of B-cell chronic lymphocytic leukemia (CLL) is poor if an aggressive course or adverse biological risk factors are present, such as unfavorable cytogenetics, an unmutated status of the variable region of the immunoglobulin heavy-chain gene, or expression of the ZAP70 gene product. 1 4 The only treatment modality with documented potential for long-term disease control in CLL is allogeneic stem cell transplantation (allo- SCT). 5 9 In spite of this, the overall survival (OS) after allo-sct is reported to be only 45 60% at 3 to 4 years post-transplant due to the substantial toxicity associated with the transplant procedure. In large registry analyses, treatment-related mortality (TRM) of allografting for CLL using standard myeloablative regimens exceeds 30% (Esteve et al. Blood 2001; 98: 480a, abstract; Michallet et al. Blood 2001; 98: 859a, abstract). The causes of Correspondence: Dr P Dreger, Department Medicine V, University of Heidelberg, Im Neuenheimer Feld 410, Heidelberg 69120, Germany; Fax þ ; peter_dreger@med.uni-heidelberg.de Received 30 August 2004; accepted 28 February 2005; Published online 14 April 2005 these discouraging results are not completely clear, but patient age, selection of poor risk patients with extensive pretreatment and the CLL-associated incompetence of the immune system may all have contributed to the high TRM observed. In order to overcome this problem, strategies of nonmyeloablative, reduced-intensity conditioning (RIC) are increasingly being used in allo-sct for CLL, 9 11 based on observations indicating sensitivity of CLL to graft-versus-leukemia effects. A recent EBMT survey documented a 2-year TRM of only 18% (95% confidence interval (95%CI) 9 27) in 77 patients who had undergone allo-sct after various RIC regimens. 12 As results of controlled prospective trials on RIC allo-sct in CLL are still lacking, the aim of the present study was to prove the anticipated beneficial effect of RIC in CLL by comparison with standard myeloablative transplants. For this purpose, we performed a population-matched analysis using the patients from the EBMT survey and CLL cases from the EBMT database who had received a fully myeloablative allo-sct during the same time period. Methods Patient populations The RIC population consisted of all patients from the previously published EBMT survey, 12 except for four cases who had received BEAM (carmustine, etoposide, cytarabine, melphalan) for conditioning because this regimen is not uniformly considered as reduced intensive. 13 For the remaining 73 patients, complete information was available regarding patient and disease characteristics (age, sex, date of diagnosis, pretreatment, previous autotransplant, remission status at transplant, performance status at transplant, concomitant disease), graft/transplant characteristics (stem cell source, T-cell depletion, CD34 þ cell number transfused, donor, date of SCT, conditioning regimen, graft-versus-host disease (GVHD) prophylaxis, peritransplant toxicity) and outcome (post-transplant toxicity, extent and kinetics of acute GVHD and chronic GVHD, maximum donor chimerism and time to maximum chimerism, response to transplant, time to relapse or progression, disease status and survival at last follow-up, cause of death). Patients had been allografted between 1998 and Myeloablative controls were all patients from the EBMT database who fulfilled the following requirements: Transplant date between 1998 and 2001 and complete information available on age, sex, time from diagnosis to transplant,

2 1030 remission status at transplant, stem cell source, donor type and use of myeloablative conditioning (MC). Conditioning regimens were considered as myeloablative for the purposes of this study if they fulfilled at least one of the following criteria: Total body irradiation (TBI) 8 Gy or more single fraction; TBI 9 Gy or more double fraction; TBI more than 10 Gy hyperfractionated; busulfan 16 mg/kg; busulfan 12 mg/kg plus melphalan 140 mg/ m 2 ; and busulfan 10 mg/kg plus thiotepa 20 mg/kg. When starting this project, we initially performed an analysis using the patients from the EBMT survey 12 plus all patients from the EBMT database who had been allografted between 1998 and In this analysis, assignment of patients to RIC vs MC was carried out according to the suggestions of the individual centers on EBMT Med A forms. This comparison included altogether 448 patients and was submitted to the 2003 Annual Meeting of the American Society of Hematology (ASH) (Dreger et al. Blood 2003; 102: 197a, abstract). Further validation by considering extensive information on agents and doses actually used for conditioning, however, disclosed that the center-based assignment was debatable in a high proportion of cases. Therefore, we decided to retract the findings reported in the ASH abstract (which was already done during the presentation of the study at the ASH meeting), and to restrict the analysis for RIC to the survey cases and for MC to those for whom detailed conditioning information was available. After that, the models were extended with a number of risk factors, which were entered in a backward selection procedure. Here, the final model is more data driven since many more risk factors were tested and hence the associated P-values must be interpreted with more care and the interpretation of their significance is more hypothesis generating than hypothesis testing by nature. Finally, we verified the possible confounding effect of each of the remaining risk factors due to the fact that patients are necessarily removed from the model if at least one of the remaining risk factors has a missing value. This was carried out by dichotimizing in turn each risk factor into two categories: valid and missing and fitting the same model again (this time including all patients), adding the interaction of that risk factor with all other dichotomized ones and checking whether the missing and nonmissing subgroups generated clinically relevant differences in the estimates of the other factors. Kaplan Meier curves were generated with GraphPad Prism software (release 3.02; San Diego, CA, USA). Ethical committee approval Ethical committee approval for this study was obtained at each center as required. Statistical analysis In principle, the subpopulation of the EBMT database fulfilling the above-mentioned criteria would have been reduced further in order to obtain a population-based matched control group for the RIC group by considering the distribution of all covariates in both subgroups and restricting the (multivariate) distribution of the main risk factors in the control group to become identical to that of the RIC group. However, this turned out to be unnecessary since the two distributions overlapped sufficiently to let a multivariate model take care of the proper adjustment without unnecessary removal of patients (and hence information) from the total data set to be analyzed. Survival time data were calculated from the time of transplant using Kaplan Meier product-limit estimates. End points were defined as follows: OS was death from any cause. For calculation of event-free survival (EFS), clinical relapse, progression, second transplant or death from any cause were counted as events. Events relevant for the end point time to clinical progression or relapse were clinical progression or disease recurrence. Events determining TRM were all deaths before or without clinical progression or disease recurrence. Fisher s exact test was used to compare qualitative parameters, and Mann Whitney tests to compare quantitative parameters between subgroups of patients. Survival times were compared by univariate log-rank testing using SPSS for Windows software (release 10.0; SPSS Inc., Chicago, IL, USA). For multivariate analyses, the Cox proportional hazards model was used. The multivariate analyses were performed with two different intentions. First we used only a priori chosen variables, selected on proven predicted power in independent analyses. A backwards selection procedure was performed, starting with this limited number of chosen variables. This allows the usual and proper interpretation of the P-values since the models are not data driven. The number of variables chosen is in accordance with the total sample size to avoid overfitting the data. Results Patients In all, 73 RIC patients were defined by participation in the EBMT survey excluding those conditioned with BEAM. Myeloablative cases were identified from 226 transplants reported to the EBMT registry between 1998 and 2001, which were indicated as myeloablative by the reporting center. Complete information, which allowed assignment to the myeloablative cohort as defined in the methods section, was available in only 82 of these. Details of the conditioning regimens for both cohorts are given in Table 1. Patient characteristics and transplant characteristics are presented in Table 2. Both groups were similar in terms of sex, remission status at SCT and donor type. However, RIC patients were significantly older, had a longer time from diagnosis to transplant and were more often grafted with PBSC instead of marrow than MC patients. Outcome An unadjusted comparison of RIC and MC patients did not reveal significant differences in terms of OS, EFS and TRM with a median follow-up of 22 (2 60) months for the RIC group and 29 (3 66) months for the MC group. However, there were fewer relapses after MC. The 2-year Kaplan Meier estimates for RIC and MC patients, respectively, were 70 vs 70% for OS (P ¼ 0.7; log rank); 58 vs 62% for EFS (P ¼ 0.88); 19 vs 26% for TRM (P ¼ 0.21); and 28 vs 11% for relapse (P ¼ 0.008). For multivariate analysis, in a first step adjustment for age, sex, donor source and remission status prior to transplant only was performed by Cox modeling to avoid overfitting of the data. These factors were selected because they had been shown to be of significant predictive value in previous analyses (Robinson et al, 15 and unpublished data). After multivariate matching of the two populations for only these covariates, a significant

3 reduction of TRM became evident for the RIC population (hazard ratio (HR) 0.4 (95% confidence interval ); P ¼ 0.03). On the other hand, RIC remained associated with an increased relapse incidence (HR 2.65 ( ); P ¼ 0.054). Although only borderline significant, this effect was strong enough to offset a benefit of RIC in terms of EFS (HR 0.69 ( ); P ¼ 0.22) and OS (HR 0.65 ( ); P ¼ 0.21). In a second step, extended Cox modeling was performed by backward selection of the variables stem cell source, time from diagnosis to transplant, transplant year in addition to age, sex, donor source and remission status at SCT, taking also into account interaction terms between these covariates. This analysis confirmed the effect of conditioning as found with four-factor matching for TRM, but slightly weakened it for relapse. Independent adverse predictors were identified as follows: MC, higher age for TRM; alternative donor, less than Table 1 Conditioning regimens Conditioning n (%) Reduced intensity 73 TBI 2 Gy 7 (10%) TBI 2 Gy/fludarabine 10 (14%) Fludarabine/busulfan combinations 13 (18%) Fludarabine/melphalan combinations 17 (23%) Fludarabine/cyclophosphamide combinations 21 (29%) Other 5 (7%) Myeloablative 82 TBI/cyclophosphamide 69 (84%) TBI/melphalan 4 (5%) Busulfan/cyclophosphamide 4 (5%) Other 5 (6%) TBI ¼ total body irradiation. Table 2 Patient characteristics RIC MC P-value n Sex female 23/73 (32%) 17/82 (21%) 0.14 Age (years) 53 (30 66) 45 (19 65) o Alternative donor 15/73 (21%) 18/82 (22%) 0.85 Less than PR at transplant 28/73 (38%) 36/82 (44%) 0.52 Months from dx to transplant 49 (8 146) 35 (4 193) Year of transplant o /73 (21%) 45/82 (55%) o Stem cell source PB 67/73 (92%) 41/82 (50%) o RIC ¼ reduced-intensity conditioning; MC ¼ myeloablative conditioning. Table 3 Prognostic factors for outcome (Cox s multivariate; n ¼ 155) PR prior to SCT, and later transplant year for relapse; and higher age and less than PR prior to SCT for OS (Table 3). Apart from methodological problems of inclusion of timedependent covariates, chronic GVHD could not be considered in the Cox models due to missing information on this variable in a large proportion of patients of the MC group. Since GVHD as possible indicator of GVL, nevertheless, seems to be very crucial for interpreting the outcome of allo-sct in CLL, we separately analyzed a subset of 109 patients for whom data on chronic GVHD was available. Altogether, presence of limited or extensive chronic GVHD at any time was reported for 19 of 42 MC cases (45%) and 37 of 67 RIC cases (55%). Notably, only two of 22 relapses (9%) observed in these 109 patients occurred after the onset of chronic GVHD (one case each in the MC group and RIC group, respectively), whereas the remaining 20 relapses developed in the absence of chronic GVHD (Po0.0001; Fisher s exact test). Discussion The aim of this study was to investigate the influence of RIC on TRM after allografting for CLL. Thorough multivariate matching for prognostically relevant confounding variables, namely age, donor source and remission status prior to transplant, disclosed a significantly reduced TRM in the population conditioned with reduced-intensity regimens. This observation confirms previous impressions from uncontrolled studies on RIC in CLL 10,12 and is in line with findings obtained in other indolent lymphohematopoietic neoplasms. 14,15 Taking into account the limitations inherent to retrospective registry studies, however, these and all of the following interpretations must be accepted with great care. Despite the reduction of TRM, RIC was not associated with better OS and EFS due to an increased relapse incidence in this series. This unexpected finding might reflect the effect of lower cytoreductive efficacy of the conditioning regimen in the RIC population. However, recent studies on MRD kinetics post alloand auto-sct suggest that in CLL the contribution of the conditioning regimen even if myeloablative to MRD level reduction is generally inferior to that of GVL. 7,16,17 Moreover, MRD studies do not provide evidence that myeloablative regimens result in greater reduction of the CLL load than fludarabine-containing RIC protocols, although it has to be taken into account that MRD data for MC regimens are available only for auto-sct. 16 Another line of evidence for the importance of GVL for disease control after allo-sct in CLL comes from the strong inverse correlation between chronic GVHD and relapse observed with both types of conditioning studied here, but it has to be kept in mind that chronic GVHD could be considered only 1031 End point variable Relapse TRM Overall survival HR (95% CI) P-value HR P-value HR P-value RIC 2.46 ( ) ( ) ( ) 0.21 Age (years) a 1.38 ( ) ( ) ( ) 0.01 Donor not identical sibling 2.92 ( ) ( ) ( ) 0.18 Status at SCT opr 3.14 ( ) ( ) ( ) 0.03 Year of SCT b 1.71 ( ) 0.03 NR c NR Sex female 0.71 ( ) ( ) ( ) 0.7 Additional variables not remaining in the models: Time from diagnosis to SCT, stem cell source. a HR by percentile as linear effect (p45; 45p50; 50p55; 455). Reference is age p45 years. b HR by calendar year (1998; 1999; 2000; 2001). Reference is year ¼ c NR ¼ not remaining in the final model. Bold indicates variables with Po0.05.

4 1032 for two-thirds of the patient population and not as timedependent covariate due to missing data. Accepting GVL as possible mechanism for disease control, one could argue that the increased relapse incidence might be due to a lower risk of GVHD associated with RIC. However, in the present study, differences between MC and RIC in terms of chronic GVHD occurrence could not be observed. This leaves the possibility that MC better than RIC provides temporary disease control until GVL can become effective. Alternatively, since relapse and TRM are competing risks, a possible contributor to the increased relapse incidence could be that those poor-risk features, which favor TRM after the more toxic MC procedure, are overlapping with those that favor relapse after RIC. In other words, succumbing to TRM might prevent poor-risk patients from experiencing relapse after MC. Finally, the increased relapse incidence after RIC observed here might be explained by bias from confounding variables not covered by this kind of data collection. Some examples of these are performance status, genetic risk factors, pretreatment with purine analogs, antibodies, autografts, etc, none of which were considered in the present analysis due to lack of information for the MC population. Indeed, ineligibility for MC due to reduced performance status has been indicated as reason for preferring RIC over MC in initial studies on reduced-intensity allotransplants including patients considered in this analysis. 11,18,19 The hypothesis of patient selection bias is further supported by the intriguing observation that the relapse risk increases per calendar year of transplant, implying that with the availability of RIC, poorer-risk patients are increasingly considered for allogeneic SCT during the most recent years. A similar explanation may hold for the adverse effect of unrelated donor use on relapse seen here. Taken together, the current study does not provide definite evidence that the intensity of conditioning is of crucial importance for disease control after allo-sct in CLL. Nevertheless, the relapse problem must be a key issue of future prospective studies on RIC allografting in CLL. Similarly, such studies must allow evaluation of additional questions such as proper indication, timing and optimum transplantation strategy (ie agents employed for conditioning, T-cell depletion, donor lymphocyte infusion, etc). Many of these objectives are addressed by the current EBMT CLL3X trial. The protocol of this prospective phase-ii study on RIC allografting for CLL can be downloaded from the EBMT web site ( In conclusion, RIC appears to favorably influence TRM after allogeneic SCT for CLL. This observation as well as the possible detrimental effects of RIC on the relapse risk have to be confirmed by prospective studies. References 1 Döhner H, Stilgenbauer S, Benner A, Leupolt E, Kröber A, Bullinger L et al. Genomic aberrations and survival in chronic lymphocytic leukemia. N Engl J Med 2000; 343: Damle RN, Wasil T, Fais F, Ghiotto F, Valetto A, Allen SL et al. Ig V gene mutation status and CD38 expression as novel prognostic indicators in chronic lymphocytic leukemia. Blood 1999; 94: Hamblin TJ, Davis Z, Gardiner A, Oscier DG, Stevenson FK. Unmutated Ig V H genes are associated with a more aggressive form of chronic lymphocytic leukemia. Blood 1999; 94: Crespo M, Bosch F, Villamor N, Bellosillo B, Colomer D, Rozman M et al. ZAP-70 expression as a surrogate for immunoglobulinvariable-region mutations in chronic lymphocytic leukemia. N Engl J Med 2003; 348: Khouri IF, Przepiorka D, van BK, O Brien S, Palmer JL, Lerner S et al. Allogeneic blood or marrow transplantation for chronic lymphocytic leukaemia: timing of transplantation and potential effect of fludarabine on acute graft-versus-host disease. Br J Haematol 1997; 97: Pavletic ZS, Arrowsmith ER, Bierman PJ, Goodman SA, Vose JM, Tarantolo SR et al. Outcome of allogeneic stem cell transplantation for B cell chronic lymphocytic leukemia. Bone Marrow Transplant 2000; 25: Esteve J, Villamor N, Colomer D, Cervantes F, Campo E, Carreras E et al. Stem cell transplantation for chronic lymphocytic leukemia: different outcome after autologous and allogeneic transplantation and correlation with minimal residual disease status. 2001; 15: Doney KC, Chauncey T, Appelbaum FR. Allogeneic related donor hematopoietic stem cell transplantation for treatment of chronic lymphocytic leukemia. Bone Marrow Transplant 2002; 29: Dreger P, Montserrat E. Autologous and allogeneic stem cell transplantation for chronic lymphocytic leukemia. 2002; 16: Schetelig J, Thiede C, Bornhauser M, Schwerdtfeger R, Kiehl M, Beyer J et al. Evidence of a graft-versus-leukemia effect in chronic lymphocytic leukemia after reduced-intensity conditioning and allogeneic stem-cell transplantation: the Cooperative German Transplant Study Group. J Clin Oncol 2003; 21: Khouri IF, Lee MS, Saliba RM, Andersson B, Anderlini P, Couriel D et al. Nonablative allogeneic stem cell transplantation for chronic lymphocytic leukemia: impact of rituximab on immunomodulation and survival. Exp Hematol 2004; 32: Dreger P, Brand R, Hansz J, Milligan D, Corradini P, Finke J et al. Low treatment-related mortality but retained graft-versus-leukemia activity after allogeneic stem cell transplantation for chronic lymphocytic leukemia using reduced-intensity conditioning. 2003; 17: Faulkner RD, Craddock C, Byrne JL, Mahendra P, Haynes AP, Prentice HG et al. BEAM-alemtuzumab reduced-intensity allogeneic stem cell transplantation for lymphoproliferative diseases: GVHD, toxicity, and survival in 65 patients. Blood 2004; 103: Perez-Simon JA, Kottaridis PD, Martino R, Craddock C, Caballero D, Chopra R et al. Nonmyeloablative transplantation with or without alemtuzumab: comparison between 2 prospective studies in patients with lymphoproliferative disorders. Blood 2002; 100: Robinson SP, Goldstone AH, Mackinnon S, Carella A, Russell N, De Elvira CR et al. Chemoresistant or aggressive lymphoma predicts for a poor outcome following reduced-intensity allogeneic progenitor cell transplantation: an analysis from the Lymphoma Working Party of the European Group for Blood and Bone Marrow Transplantation. Blood 2002; 100: Ritgen M, Stilgenbauer S, von Neuhoff N, Humpe A, Brüggemann M, Pott C et al. Graft-versus-leukemia activity may overcome therapeutic resistance of chronic lymphocytic leukemia with unmutated immunoglobulin variable heavy chain gene status: implications of minimal residual disease measurement with quantitative PCR. Blood 2004; 104: Böttcher S, Ritgen M, Pott C, Brüggemann M, Raff T, Stilgenbauer S et al. Comparative analysis of minimal residual disease detection using four color flow cytometry, consensus IgH-PCR, and quantitative IgH PCR in CLL after allogeneic and autologous stem cell transplantation. 2004; 18: Diaconescu R, Flowers CR, Storer B, Sorror ML, Maris MB, Maloney DG et al. Morbidity and mortality with nonmyeloablative compared with myeloablative conditioning before hematopoietic cell transplantation from HLA-matched related donors. Blood 2004; 104: Niederwieser D, Maris M, Shizuru JA, Petersdorf E, Hegenbart U, Sandmaier BM et al. Low-dose total body irradiation (TBI) and fludarabine followed by hematopoietic cell transplantation (HCT) from HLA-matched or mismatched unrelated donors and postgrafting immunosuppression with cyclosporine and mycophenolate mofetil (MMF) can induce durable complete chimerism and sustained remissions in patients with hematological diseases. Blood 2003; 101:

5 Appendix A 1033 EBMT participating institutions Australia Austria Belgium Czech Republic Czech Republic Denmark Finland France Germany Italy The Netherlands Poland UK Sweden Spain Royal Perth Hospital, Perth (RP Herrmann) Vienna (H Greinix) Clinique Universitaire St Luc, Brussels (A Ferrant) Charles University Hospital, Pilsen (V Koza) Institute of Hematology and Blood Transfusion, Prague (A Vitek) Rigshospitalet, Copenhagen (L Vindelöv) University Central Hospital, Helsinki (T Ruutu) University Central Hospital, Turku (K Remes) Hôpital St Jaques, Nantes (N Milpied) Hôpital L Herriot, Lyon (M Michallet) Hôpital Necker, Paris (B Varet) Hôpital Saint Antoine, Paris (NC Gorin) Pitie-Salpetriere, Paris (L Sutton) Department of Haematology and Oncology, Freiburg (J Finke) BMT Unit, University Hospital, Hamburg (N Kröger) Medizinische Hochschule, Hannover (B Hertenstein) BMT Unit, Universität des Saarlandes, Homburg (L Trümper, D Beelen) BMT Unit, University Hospital, Kiel (P Dreger) Department of Hematology and Medical Oncology, Leipzig (D Niederwieser) Department of Hematology, University Hospital, Mainz (G Derigs) Abteilung Hämatologie und Onkologie, Universität Rostock (J Casper) Medical Department III, University Hospital, Ulm (S Stilgenbauer, D Bunjes) Istituto Nazionale dei Tumori, Milan (P Corradini) BMT Center, Ospedale Maggiore, Milan (G Lambertenghi Deliliers) Ospedale V Cervello, Palermo (I Majolino) Ospedale San Martino, Genova (A Bacigalupo) University La Sapienza, Rome (G Meloni) University Medical Centre, Leiden (R Willemze) University Medical Centre St Radboud, Nijmegen (T de Witte) Daniel den Hoed Cancer Centre, Rotterdam (JJ Cornelissen) University Medical Centre, Utrecht (L Verdonck) Department of Hematology, University of Medical Sciences, Poznan (J Hansz) Heartlands Hospital, Birmingham (D Milligan) Department of Haematology, Christie Hospital, Manchester (G Morgenstern) Department of Haematology, Bristol, (J Hows) Department of Haematology, University College (A Goldstone) Centre for Allogeneic SCT, University Hospital, Huddinge (J Aschan) Akademiska Sjukhuset, Uppsala (G Öberg) Hospital de la Santa Creu i Sant Pau, Barcelona (R Martino) Dept. Hematology, Hospital Clínic, Barcelona (J Esteve) Complejo Hospital, Salamanca (D Caballero) Hospital Universitario M de Valdecilla, Santander (A Iriondo)