Leukemia (2008) 22, & 2008 Macmillan Publishers Limited All rights reserved /08 $

Transcription

1 ORIGINAL ARTICLE (2008) 22, & 2008 Macmillan Publishers Limited All rights reserved /08 $ Quantitative MRD monitoring identifies distinct GVL response patterns after allogeneic stem cell transplantation for chronic lymphocytic leukemia: results from the GCLLSG CLL3X trial M Ritgen 1,SBöttcher 1, S Stilgenbauer 2, D Bunjes 2, J Schubert 3, S Cohen 4, A Humpe 1, M Hallek 5, M Kneba 1, N Schmitz 6, HDöhner 2 and P Dreger 7, for the German CLL Study Group 1 Department of Medicine II, University of Schleswig-Holstein, Kiel, Germany; 2 Department of Internal Medicine III, University of Ulm, Ulm, Germany; 3 Department of Medicine I, University of Saarland, Homburg, Germany; 4 Department of Hematology, Hospital Maisonneuve-Rosemont, Montreal, Quebec, Canada; 5 Department of Medicine I, University of Cologne, Cologne, Germany; 6 Department of Hematology and Stem Cell Transplantation, Asklepios Klinik St Georg, Hamburg, Germany and 7 Department of Medicine V, University of Heidelberg, Heidelberg, Germany The purpose of this study was to prospectively analyze minimal residual disease (MRD) kinetics after reduced-intensity allogeneic stem cell transplantation (allo-sct) in high-risk chronic lymphocytic leukemia (CLL). Subjects were the first 30 consecutive patients from a prospective clinical trial, and seven pilot patients treated identically. Using real-time quantitative- PCR (RQ-PCR) and/or flow-based MRD monitoring (sensitivity X10 4 ), five distinct patterns of MRD kinetics could be identified: patients who promptly achieved durable MRD negativity without direct evidence of graft-versus-leukemia (GVL) effects (Group 1) (n ¼ 4; no clinical relapse); patients with complete and sustained MRD response after GVL induced by immunosuppression tapering (Group 2) or donor lymphocyte infusions (Group 3) (n ¼ 18; one relapse); patients without MRD response due to lack of GVL (Group 4) (n ¼ 2; two relapses); patients with incomplete and transient MRD response to GVL (Group 5) (n ¼ 4; three relapses). In summary, this study provides a comprehensive map of possible MRD courses and their prognostic implications after T-replete allo- SCT in high-risk CLL, indicating that effective GVL activity is induced virtually in all patients who develop chronic GVHD. However, in a significant proportion of cases, this does not translate into sustained disease control due to development of secondary GVL resistance. (2008) 22, ; doi: /leu ; published online 17 April 2008 Keywords: CLL; allogeneic stem cell transplantation; MRD; clinical trial; GVL Allogeneic stem cell transplantation (allo-sct) can be an effective treatment for chronic lymphocytic leukemia (CLL). We and others have previously shown in patients with genetically defined poor-risk CLL that allo-sctfeven if performed after reduced-intensity conditioning (RIC)Fmay Correspondence: Professor P Dreger, Department of Medicine V, University of Heidelberg, Im Neuenheimer Feld 410, Heidelberg 69120, Germany. peter.dreger@med.uni-heidelberg.de Presented in part in abstract form at the 9th International Conference on Malignant Lymphoma, Lugano, Switzerland, June Authorship: MR, PD and HD designed the trial; PD, SS, DB, JS, SC and NS were involved in patient care, sample and clinical data acquisition; MR, PD and MH were responsible for data management; SS and HD performed genetic analyses; MR, SB and MK performed MRD analyses; AH performed chimerism analyses; PD performed statistical analyses; MR and PD wrote the paper; and all authors checked the final version of the manuscript. Received 4 October 2007; revised 25 February 2008; accepted 17 March 2008; published online 17 April 2008 result in complete resolution of minimal residual disease (MRD) as assessed by sensitive flow cytometry- or allele-specific PCR-based assays. 1,2 Since only small series of patients were analyzed so far, it is unknown, however, how frequently MRD negativity after RIC allo-sct is achieved, and whether it is durable or has prognostic impact on defined high-risk CLL settings. Moreover, although there is no doubt that graft-versusleukemia (GVL) activity is the key therapeutic principle responsible for disease control after RIC allo-sct in CLL, 3 6 time course and clinical triggers of GVL as well as resistance mechanisms to it are unclear. To address these questions, we prospectively analyzed MRD kinetics in consecutive patients having undergone T-replete RIC allo-sct for high-risk CLL within a prospective clinical trial using sensitive quantitative MRD monitoring according to the recently approved international standard. 7 The results show that durable MRD negativity can be achieved in more than half of the patients and occurs essentially in the context of chronic graft-versus-host disease (GVHD) induced by immunosuppression tapering or donor lymphocyte infusions (DLI). Although the patterns of MRD kinetics indicate significant GVL activity even in those patients who fail complete MRD clearance, it seems that resistance to GVL effects can be acquired over time. Furthermore, our data provide evidence that quantitative MRD monitoring might be very useful for guiding preemptive immunotherapy after allo-sct for CLL. Patients and methods Patients The patient population investigated for the purposes of this analysis comprised the first 30 consecutive patients transplanted on the T-replete arm of the German CLL Study Group (GCLLSG) CLL3X trial (Arm A). CLL3X was an open, non-randomized, multicenter phase-ii clinical study primarily aiming at studying the feasibility and safety of allo-sct following conditioning with fludarabine and cyclophosphamide in patients with high-risk CLL. Although the trial has completed accrual after enrollment of 113 patients in March 2007, final clinical results will not be available before early However, a key predefined secondary objective was the assessment of clinical significance of quantitative MRD monitoring in 30 consecutive patients. Patients eligible for CLL3X were those with high-risk CLL as defined by one of the following: (i) refractoriness or early relapse (within 12 months) after treatment with a purine

2 1378 analogue-containing regimen; (ii) relapse after autologous SCT; or (iii) rapidly progressive/symptomatic disease in the presence of an unfavorable genetic feature (11q deletion (11q-); 17p deletion (17p-); and/or unmutated status of the immunoglobulin VH gene region; and/or usage of the VH3-21 gene). Patients had to be between 18 and 65 years old, and had to have an HLA-matched (6/6) sibling or unrelated donor available. Exclusion criteria comprised serious localized or systemic infections, concomitant malignant disease, impaired organ function (creatinine clearance o60 ml/min; ASAT, ALAT, bilirubine 42 normal; impaired cardiac function by electrocardiogram and echocardiographic examination; and impaired pulmonary function tests) and psychiatric disease. Patients with Richter s transformation were also ineligible. Conditioning consisted of daily fludarabine (30 mg/m 2 ) and cyclophosphamide (500 mg/m 2 ) over 5 days. In case of unrelated donors, ATG (10 mg/kg/d; Fresenius, Bad Homburg, Germany) was added over 4 days. Conditioning was performed as per protocol in 27 cases. In two patients with documented fludarabine intolerance, cladribine 0.11 mg/kg/d over 5 days was used instead; and in one patient, busulfan 10 mg/kg was added to the conditioning regimen to overcome refractory disease. Peripheral blood stem cell or marrow grafts were transfused without manipulation on day 0. GVHD prophylaxis was performed with cyclosporin A (CSA, target level ng/ml). CSA was tapered between day þ 60 and þ 100 post transplant. In addition to CSA, short-course methotrexate (10 mg/m 2 on days þ 1, þ 3 and þ 6) or mycophenolat mofetil (2 15 mg/kg orally administered from day 0 to day þ 50) was given. DLI were administered not earlier than 4 weeks after complete withdrawal of CSA in case of incomplete chimerism and/or increasing MRD levels in the absence of GVHD. Regular follow-up included infection, GVHD and adverse event assessment; physical examination with assessment of nodal status; basic hematology and chemistry screen; and MRD assessment. BM biopsies and (optional) restaging of the nodal status by ultrasound or CT scanning had to be performed 3, 6 and 12 months post transplant and were to be repeated every 6 12 months thereafter. The protocol including the informed consent form was approved by all responsible institutional review boards (Primary IRB: Ethics Committee of the University of Kiel, approval # A123/00). Patients gave written informed consent using studyspecific forms. The trial has been registered at the US National Cancer Institute (Protocol Identity # EU-20554, NCT , see for protocol synopsis). For the analysis of individual MRD kinetics in response to immunomodulating maneuvers and of durability of MRD negativity, all seven patients who were treated identically in the CLL3X centers during a pilot phase before the formal implementation of CLL3X and had MRD monitoring available were studied. Nine patients (05P04, 41P01, 41P02, 41P03; through 05004, 41001) had been included in a preliminary analysis of MRD kinetics published previously. 1 Genetic analyses Immunoglobuline heavy chain (IgH) sequence analysis was performed as described previously. 8,9 Briefly, genomic DNA was isolated from blood or bone marrow samples, subjected to PCR amplification of VDJ rearrangements using appropriate (5 0 -V H ) FRI and (3 0 -J H ) FRIV primers, and sequenced with a genetic analyzer (models ABI 310 or ABI 377, Applied Biosystems, Foster City, CA, USA). All analyses were performed in 2 4 replicates using independent PCR products. Obtained sequences were compared to published VH, DH and JH germline sequences (DNAPLOT software and IMGT Database; to quantify germline homology of the individual clone-specific CDR3 region. Molecular cytogenetic analysis was performed at the time of study entry by interphase fluorescence in situ hybridizationas described previously. 10 All genomic analyses were performed in a central laboratory. Quantitative IgH RQ-PCR MRD quantification by IgH real-time quantitative-pcr (RQ- PCR) was performed as described previously. 11 In brief, allelespecific oligonucleotide primers matching the hypervariable N-D-N region of the individual leukemic clone were used with reverse consensus JH germline primers and a Taqman probe annealing to a downsteam family-specific JH region for RQ-PCR on an ABI PRISM 7700 thermal cycler (Applied Biosystems). Standards were generated with serially diluted pooled DNA from samples collected at initial diagnosis or relapse. Specificity was tested by amplifying polyclonal genomic DNA of pooled healthy donors in every PCR reaction. In case of non-specific amplification, cycle threshold (C T ) values of the last dilution step with a specific amplification product had to be at least one cycle lower than the lowest C T value found in polyclonal DNA. Before analysis of follow-up samples, each RQ-PCR assay was optimized for the highest sensitivity and specificity by testing different annealing temperatures. Differences in the amount and quality of DNA samples were normalized by using the albumin gene as internal reference. Calculation of MRD levels was based on comparative C T analysis between follow-up samples and standards in allele-specific oligonucleotide-pcr and albumin PCR. MRD values were specified in relation to target copy number in the diagnostic sample. Using this method in all cases, a sensitivity of at least 1E-04, mostly between 1E-05 and 4E-05 could be reached. MRD flow MRD quantification by high-resolution 4-colour flow cytometry (MRD flow) has been described previously. 7,11 In brief, two million leukocytes per tube were stained with fluorochromelabelled CD20/CD79b/CD19/CD5 and CD20/CD5/CD19/CD43 antibodies, respectively and acquired to an FACSCalibur flow cytometer (Becton Dickinson (BD), Heidelberg, Germany). Isotype controls (incubated with four isotype-matched irrelevant monoclonal antibodies conjugated with FITC, PE, PerCP-Cy5.5 and APC, all BD) were used to optimize light scatter parameters, threshold and amplification for each sample. Acquisition stopped, when all available leukocytes of the sample had been processed. CLL cells were defined as light scatter events with positivity for CD19, CD5, CD43 and dim expression of CD20 and CD79b. Clusters of 20 events or more were considered as evidence for MRD. MRD levels were calculated by dividing the number of CLL cells by the number of leukocytes. Each MRD value determined was the average of two tubes. The sensitivity of this assay is one cell in 10 4 leukocytes. MRD response evaluation criteria The study protocol defined RQ-PCR as the method for MRD monitoring. In March 2002, however, MRD flow was introduced with a sensitivity almost equivalent to RQ-PCR, but much

3 easier and broader applicability. 7,11,12 Thus, it was decided to quantify MRD by MRD flow in addition to RQ-PCR. An informative RQ-PCR marker could be identified in 18 of 25 patients with sufficient follow-up sampling available. In these 18 patients, parallel results from RQ-PCR and MRD flow monitoring were largely concordant in terms of MRD kinetics and time to reach MRD negativity. Therefore, MRD flow was accepted as adequate substitute in those cases where RQ-PCR was not possible to take the advantage of a larger data pool. As MRD flow was not available for samples obtained before March 2002, MRD flow could not generally be used instead of RQ-PCR. MRD levels are reported as a proportion of CLL cells of total leukocytes if not otherwise indicated. According to previously published observations, 7 we found a high correlation between MRD levels in peripheral blood and bone marrow samples collected at the same day (Supplementary Figure 1). However, to avoid bias, which nevertheless might arise from mixing blood and marrow results, the analysis of MRD kinetics for the purposes of this study was restricted to peripheral blood samples if not otherwise indicated. Statistical analysis The Mann Whitney test and Fisher s exact test were used to compare quantitative and categorical factors, respectively, between two groups of patients. Significance levels were set at Survival time data were measured from the time of SCT. Survival curves were estimated using Kaplan Meier method. Events relevant for the endpoint event-free survival (EFS) were clinical progression, disease recurrence or death from any cause, and the event relevant for the endpoint time to MRD negativity was the time of achievement of durable MRD negativity, measured from the time of transplant, with nonrelapse deaths being considered as competing events. Calculations were carried out using GraphPad Prism software (release 3.02; San Diego, CA, USA). Data were analyzed as of 31 August Results Patients Between June 2001 and March 2005, 30 patients were enrolled on arm A. Primary indication for inclusion was fludarabine resistance in 14 patients (47%). However, due to effective salvage with antibody-containing regimens, only seven patients (23%) had uncontrolled disease at the time of transplant, 19 (63%) were in partial remission, and four (13%) fulfilled the criteria of complete remission. All 29 patients tested had an unfavorable VH status, and that 11q- and/or 17p- was observed in half of the patients. Of note, as defined by the CLL3X eligibility criteria, comorbidity according to the hematopoietic cell transplantation comorbidity index was low in the vast majority of patients. 13 A summary of patient characteristics is given in Table 1 for a detailed description of individual risk profiles see Table 2. Lymphohematopoietic chimerism Durable complete (495%) hematopoietic donor chimerism was observed in all but two patients with a median time to complete chimerism of 94 days. One patient (UPN 41003) received a marrow graft from an unrelated donor and never exceeded 10% donor chimerism. This patient had autologous recovery but experienced early disease progression and died 18 months post Table 1 Pretransplant characteristics of study and pilot patients Study patients Pilot patients P No. (female/male) 30 (12/18) 7 (3/4) NS age (years) 54 (27 65) 51 (40 57) NS HCT-CI score 0 1/2/42 a 29/1/0 7/0/0 NS VH unfavorable b (%) 29/29 (100%) 4/6 (67%) 0.04 FISH karyotype unfavorable c 14/28 (50%) 1/6 (17%) NS FISH karyotype 17p13 4/28 (14%) 0/6 NS Maximum binet stage A/B/C 2/13/15 1/4/2 NS Time from diagnosis (months) 53 (6 270) 44 (8 91) NS #Previous regimens 4 (1 11) 3 (1 5) NS Previous auto-sct 16/30 (53%) 0/ Fludarabine resistant d 14/30 (47%) 3/7 (43%) NS Refractory disease at SCT 7/30 (23%) 0/7 NS Donor matched unrelated 17/30 (57%) 1/7 (14%) NS Abbreviations: FISH, fluorescence in situ hybridisation; HCT-CI, hematopoietic cell transplantation comorbidity index; NS, not significant; SCT, stem cell transplantation. a Hematopoietic cell transplantation comorbidity risk score. 13 b Unmutated (498% homology to germline) or VH3-21. c del 11q23 or del 17p13. d Non-response or relapse within 12 months to fludarabine monotherapy or fludarabine-cyclophosphamide/fludarabine rituximab combinations. SCT due to CLL-related causes. Early progression (3 months after SCT) was also observed in UPN in the presence of only 62% lymphoid donor chimerism. Despite two doses of DLI, CLL progression was irreversible, resulting in disease-related death at month þ 17. Toxicity and GVHD Toxicity of conditioning and post-transplant recovery phase was generally mild, and fatal aplasia-related complications did not occur. However, during longer follow-up, three fatal infections, all in the context of acute GVHD, were observed, resulting in a 3-year probability of treatment-related mortality (TRM) of 11% (95%CI 0, 23%; (Figure 1). Clinically relevant acute GVHD (grade II IV) developed in 12 patients (40%) but was severe (grade III IV) in only three (10%). The cumulative incidence of chronic GVHD at 1 year was 75% (extensive 59%) among 26 patients at risk with no difference between sibling and unrelated donor transplants. A detailed toxicity analysis will be presented with the results of the full study. Clinical outcome Clinical relapse or disease progression occurred in nine patients. Eight patients died, three from toxicity, and five from progressive disease, translating into a 3-year EFS and overall survival of 58% (39, 74%) and 71% (53, 98%) (Figure 1). Median follow-up of surviving patients was 42 (7 70) months post transplant. Detailed outcome data is presented in Table 3. Pilot patients Seven patients were transplanted in an identical manner during a pilot phase between May 1999 and December All engrafted, but one patient (UPN 41P03) remained a mixed chimera, despite repeated DLI, which resulted in permanent MRD eradication. Although grade II IV (III IV) acute GVHD developed in three patients and chronic GVHD in four, fatal complications did not occur. One patient relapsed after

4 1380 Table 2 Clinical course of patients before allogeneic PBSCT UPN Sex/age (years) VH status FISH caryotype Documented PA resistance Time from diagnosis (months) Maximum Binet stage Previous therapy Previous auto-sct Allo-SCT indication Status at allo-sct (involved sites) 05P01 F/51 Mutated no 77 C CHOP 8, Flud 5, DB 1, no Mobilization PR2 (BM) ClB 4, FC 1 05P02 M/43 NA normal yes 20 C CHOP 3, Flud 3, CY 1, no PA resistant, PR3 (BM) FC 1, CD20 1 mobilization 05P03 F/53 NA NA NA 44 B Clb, Clb, AML no AML PR2 (BM) 05P04 M/53 Unmutated del 11q22-q23, no 44 B CHOP 4; DB 1; FC 4; CY 1 no Progressive disease Untreated relapse (BM, LN) 41P01 F/49 Unmutated Normal yes 91 B PmM 3; FE 5; DB 2 no PA resistant PR2 (BM, LN) 41P02 M/57 Unmutated Normal yes 51 B Clb 7; Flud 3; DB 1 no PA resistant; Richter s PR2 (BM, S) transform. 41P03 M/40 Unmutated Normal no 8 B FC 3 no Progressive disease PR1 (n.a.) M/48 Unmutated del 11q22-q23 no 129 C Clb 20; Trofo 1; CHOP 1; DB 2; TBI-CY; FC 4 yes Progressive disease; PR3 (BM, LN. lymphocytosis) M/43 Unmutated, t(14q32) no 44 B CHOP 3; DB 1; TBI-CY; FC 1 yes Progressive disease; PR2 (BM, LN, S) F/57 Unmutated no 57 B Clb 12; Flud 5; FC 2; DB 1 no Progressive disease; PR2 (BM, S) mobilization M/63 Unmutated del 17p13 yes 43 B FC 2; CHOP 3; DB 1; Rituximab 4 no PA resistant; RD (BM, LN) mobilization M/58 Unmutated NA yes 38 C CHOP 6; F 3; CD20 1 no PA resistant RD (BM, LN, lymphocytosis) F/65 Unmutated +12, t(14q32) yes 43 B F 6; FCR 6 no PA resistant PR2 (S) M/54 n.a. del 17p13 no 6 A FC 2; DB 1 no Progressive disease PR1 (BM, LN) F/46 Unmutated Normal no 51 C CHOP 4; DB 2; TBI/CY; FCR 4 yes Progressive disease; CR M/47 Unmutated Normal no 137 C CHOP 4, DB 1; BEAM; FC 6 yes Progressive disease; PR3 (BM) M/53 Unmutated del 11q22-q23, no 24 C FC 4; DB 2; CD52 2mo; FCR 4 no Progressive disease CR F/26 Unmutated del 11q22-q23, no 6 B FC 2 no Progressive disease PR1 (BM) M/59 Unmutated no 18 B FC 3; DB 1 no Progressive disease PR1 (LN, S) F/50 Unmutated del 11q22-q23 no 44 B FC 1; FCR 2 no Progressive disease CR M/58 Mutated;, yes 75 C COP 6; DB 2; TBI/CY; FC 2; yes PA resistant; RD (LN, S) VH3-21 CD52 12mo; CD M/52 Unmutated del 11q22-q23, no 84 B CHOP 4; DB 2; TBI/CY; FC 4 yes Progressive disease; CR M/54 Unmutated, t(14q32) no 85 C FC 3; DB 2; TBI/CY; FC 2 yes Progressive disease; PR2 (LN, S) M/46 Unmutated del 11q22-q23, no 22 B FC 4 no Progressive disease PR1 (BM, LN, S) del 8q M/58 Unmutated del 11q22-q23, no 41 C FC 4 no Progressive disease PR1 (LN, S) F/55 Unmutated Normal yes 111 B CHOP 3; DB 2; TBI/CY; FC 2; FCR 1 yes PA resistant; auto-sct RD (BM, LN) F/60 Unmutated yes 270 B Clb 2; F 9; FC 2; FCR 4; no PA resistant PR4 (BM) FCR 3; CD M/57 Unmutated yes 48 C COP 10; FR; ESHAP 2 no PA resistant RD (BM, S, lymphocytosis) M/42 Unmutated del 11q22-q23, yes 53 C MCP 3; Flud 2; DB 1; TBI/CY; yes PA resistant; auto-sct PR2 (BM, LN) FC 2; RFC M/42 Unmutated del 11q22-q23, no 60 B CHOP 3; DB 2; TBI/CY; FC 3 yes Progressive disease; PR2 (BM, LN) M/59 Unmutated Normal yes 53 C CHOP 3; DB 1; TBI/CY; FC 3; FCR 2 yes PR2 (BM)

5 Table 2 (Continued ) Status at allo-sct (involved sites) Allo-SCT indication Previous therapy Previous auto-sct Maximum Binet stage Time from diagnosis (months) VH status FISH caryotype Documented PA resistance UPN Sex/age (years) PA resistant; auto-sct RD (BM, S) yes PA resistant; auto-sct yes 67 C VAD 2, Clb, CHOP 3, DB 2; TBI/CY; FCR F/54 Unmutated del 11q22-q23, RD (BM, LN) yes PA resistant; auto-sct yes 92 C CHOP 4; DB 2; TBI/CY; FC 1; CD20 4wk; CD52 4wk F/51 Unmutated del 17p13, del 11q22-q23 PR3 (BM, LN) M/53 Unmutated +12 yes 84 B CHOP 3; DB 2; TBI/CY; FC 2; FC 2 yes PA resistant; auto-sct PR5 (BM, LN, S) yes PA resistant; auto-sct F/57 Mutated; VH3-21 del 6q21 yes 163 C Clb 13; DB 2; TBI/CY; Flud 5; Cy 2; FC 2; CD20 6; R-Cy 2; Benda 2; Trofo; CD52 3mo F/64 Unmutated NA yes 115 C Clb 19; F 6; Benda 3; R-Benda 4 no PA resistant PR3 (BM, LN, S) PR2 (LN, S) no 41 B FC 2; DB 1; TBI/CY; CHOP 4 yes Progressive disease; M/43 Unmutated del 17p13, del 6q21 Abbreviations: BEAM, high-dose carmustin, etoposide, cytarabine, melphalan; Benda, Bendamustin; BM, bone marrow involvement; CD20, rituximab; CD52, alemtuzumab; CHOP, CY, doxorubicin, vincristin, prednisolone; Clb: chlorambucil; COP, CY, vincristin, prednisolone; CY, cyclophosphamide; DB (Dexa-BEAM):, dexamethason, carmustin, etoposide, cytarabine, melphalan; ESHAP, epirubicine, prednisolone, high-dose cytarabine, cisplatin; F: female; FC: fludarabine, CY; FCR, fludarabine, CY, rituximab; FE: fludarabine, epirubicine; Flud: fludarabine; FR, fludarabine, rituximab; LN, nodal involvement; MCP, mitoxantrone, chlorambucil, prednisolone; NA, not available; PA resistant, refractoriness or early relapse (within 12 months) after treatment with a purine analogue-containing regimen; PBSCT, peripheral blood stem cell transplantion; PR: partial remission; R-Benda, rituximab, Bendamustin; R-Cy, rituximab, cyclophosphamide; RD, refractory disease; R-DHAP, rituximab, dexamethasone, high-dose cytarabine, cisplatin; S, spleen involvement; TBI, total body irradiation (12Gy); Trofo: trofosfamide; VAD, vincristin, doxorubicin, dexamethasone. Percent TRM Percent event-free Percent survival % (0, 23%) Months from SCT 58% (39,74) Months from SCT 3-y OS 71% (53, 98) Months from SCT Figure 1 Treatment-related mortality (a), event-free survival (b) and overall survival (c) of the study group. months, and six live disease free (median 81) months post transplant. MRD kinetics Altogether, 725 samples (bone marrow 68; peripheral blood 657) were collected from 25 study patients and seven pilot patients. RQ-PCR was performed on 327, MRD flow on 303, and both PCR and flow on 95 of these. All except five study group patients (one lost to follow-up, two TRM, two without post-transplant MRD sample asservation) were available for longitudinal MRD assessment for at least 12 months after SCT with either RQ-PCR (n ¼ 23) or MRD flow (n ¼ 9). The median number of follow-up samples analyzed was 16 (3 29). Although MRD kinetics showed tremendous interindividual variations, patients could be grouped according to certain characteristic MRD response patterns (Table 4, Figure 2, Supplementary Figure 2). First, there were four patients who displayed MRD negativity already immediately after (on the first measurement between days þ 22 and þ 36 days) transplant and stayed so. These patients were characterized by a relatively low MRD load before transplantftwo of them actually being MRD negative at the time of SCT, two with MRD levels of and 1381

6 1382 Table 3 Individual outcome Patient Id. Donor Best chimerism (day reached) Acute GVHD chronic GVHD (month onset) DLI (month) T-cell dose ( 10e6/kg) Response at landmark 3 a (sites involved) Best response Time to best response (month) Outcome 05P01 MSD 100% (+180) II Lim. (+4) F F PR (BM) CR 3 alive, 87+ month (MRD-) 05P02 MSD 100% (+58) III Ext. (+4) F F CR CR 1 b alive, 90+ month (MRD-) 05P03 MSD 100% (+49) 0 Lim. (+5) F F CR CR 1 c alive, 61+ month (MRD-) 05P04 MUD 100% (+49) II Ext. (+4) F F PR (BM, LN) CR 6 alive, 64+ month (MRD-) 41P01 MSD 100% (+255) 0 No 6, 15, 41 5, 10, 40 CR CR 1 c alive, 61+ month (relapse +44; 2nd SCT +60) 41P02 MSD 100% (+251) 0 No 6 5 CR CR 1 c alive, 83+ month (MRD-) 41P03 MSD 66% (+670) I (DLI) No 3, 4, 14 5, 10, 10 CR CR 1 c alive, 79+ month (MRD-) MSD 100% (+150) 0 Ext. (+4) F F PR (BM) b CR 5 alive, 61+ month (MRD-) MSD 100% (+56) I Ext. (+4) F F PR (n.a.) CR 5 dead, month +44 (relapse +33) MSD 100% (+151) I Ext. (+4) F F PR (BM) b CR 5 alive, 70+ month (MRD-) MSD 100% (+97) 0 Ext. (+3) F F PD (LN) SD 5 dead, month +19 (progress +3; SD after chronic GVHD onset; progress +19) MUD 100% (+97) II NA F F n.a. n.a. n.a. dead, month +2 (TRM) MUD 100% (+27) II Lim. (+6) F F CR b CR 1 c alive, 34+ month (MRD-) MUD 100% (n.a,) IV Ext. (+10) F F PR (BM) CR 16 alive, 16+ month (CR, MRD NA) MUD 100% (+28) I Lim. (DLI) CR CR 0 alive, 45+ month (MRD- after DLI for relapse +8) MUD 100% (+99) II Ext. (+5) n.a. CR 6 alive, 50+ month (MRD- after DLI) MSD 100% (+106) 0 Lim. (+5) F F CR b CR 0 alive, 12+ month (MRD-) MSD 72% (+55) 0 no F F CR CR n.a. alive, 7+ month (progress +5). Lost to follow-up MSD 100% (+117) Lim. (+6) F F PR (S) CR 10 alive, 41+ month (MRD-) MUD 100% (+210) I no F F CR CR 0 alive, 60+ month (MRD-) MUD 8% (+19) 0 no F F PR (LN, S) c PR 2 dead, month +18 (progress +10) graft MSD 100% (+30) II no F F CR CR 0 alive, 60+ month (MRD-) MSD 100% (+69) II Ext. (+7) F F SD (LN, S) PR (S) 12 alive, 47+ month (nodal progress +23) MUD 100% (+150) IV F F PR (S) PR 3 dead, month +6 (TRM) MSD 100% (n.a.) II Ext. (+7) F F PR (S) CR 6 dead, month +28 (TRM) MUD 100% (n.a.) I Ext. (+4) F F CR CR 1 c alive, 34+ month (MRD-) MUD 100% (+90) 0 Ext. (+4) F F CR CR 1 b alive, 26+ month (MRD-) MSD 62% (+105) 0 no 3, 7 1, 5 PD (BM, LN, S) PD 3 dead, month +17 (progress +3) MUD 100% (+92) II Ext. (+5) F F PR (BM) b CR 5 dead, month +30 (progress +11) MUD 100% (+76) 0 Ext. (+3) F F PR (BM) b CR 4 alive, 49+ month (MRD-) MUD 100% (+44) 0 Ext. (+17) 15 5 PR (BM) b CR 8 alive, 42+ month (MRD-) MUD 100% (+83) 0 Ext. (+4) F F PR (n.a.) CR 4 alive, 42+ month (MRD-) MUD 100% (+54) II Ext. (+8) F F PR (LN) CR 3 alive, 46+ month (MRD-) MUD 100% (+33) I no 16, 17 1, 5 CR c CR 2 alive, 33+ month (CR, MRD-) MUD 100% (+94) 0 Lim. (+6) F F CR c CR 2 alive, 34+ month (CR, MRD+) MSD 100% (+110) 0 no F F PR (n.a.) CR 6 alive, 30+ month (MRD-) MSD 100% (+141) 0 no F F PR (LN, S) PR (LN, S) 1 dead, month +24 (progress +10) Abbreviations: Ext, extensive; Lim, limited; MRD, minimal residual disease; MSD, matched sibling donor; MUD, matched unrelated donor; NA, not available; SD, stable disease; TRM, treatment-related mortality. Outcome of patients with del 17p-is highlighted in bold. Response reflects results of physical examination, blood counts, BM biopsy, and imaging (ultrasound or CT) if not other indicated: a Last restaging before CSA taper. b By physical examination, blood counts and BM biopsy only. c By physical examination and blood counts only.

7 0.0013, respectively. Three of these patients acquired chronic GVHD later on. A second group comprised 12 patients who showed only a modest decrease of MRD levels after conditioning without significant further decrease until the start of CSA tapering (t1fpretransplant 0.19 (0 2); t2fpost conditioning 0.02 ( ); t3fpre-csa tapering 0.01 ( ); t2 vs t3 p 0.53), but converted to MRD negativity subsequent to immunosuppression reduction (Figure 2a). Except one, all of these patients developed chronic GVHD. MRD negativity was durable in 11 patients, whereas one patient had MRD recurrence at the time of extramedullary CLL relapse. Eight patients had insignificant MRD decrease even after cessation of immunosuppression (0.03( ) pretransplant vs ( ) after CSA withdrawal; p 0.097) in the absence of chronic GVHD. DLI were given to seven of these, resulting in achieving MRD negativity in six patients with chronic GVHD occurring in two of them (Group 3; Figure 2b, Supplementary Figure 2b). The seventh patient showed no DLI effect and relapsed clinically later on (Table 2). This patient formed Group 4 together with the eighth patient who could not be treated with DLI due to donor refusal and also experienced relapse (Figure 2b, Supplementary Figure 2c). Similar to the second group, a fifth group consisting of four patients had only little MRD decrease until the start of CSA withdrawal but a profound reduction thereafter which was associated with chronic GVHD (0.033 ( ) pre-csa tapering vs ( ) after CSA withdrawal; p 0.029). However, MRD negativity was never reached, and three of these patients relapsed clinically (Figure 2c, Supplementary Figure 2d). Finally, there were three patients who could not be grouped because samples were available only from days þ 166, þ 983 and þ 1389 onwards. All were durably MRD negative. Two of Table 4 GVL response pattern groups them had chronic GVHD. The remaining patient (UPN 41003) was the one with autologous recovery who was MRD positive until clinical relapse 15 months post transplant. MRD level MRD level 1E+1 1E+0 1E-1 1E-2 1E-3 1E-4 1E-5 1E-6 CR PR SD 3 PD 1 1 Landmark E+1 1E+0 1E-1 1E-2 1E-3 1E Group 1: Group 2: Group 3: Group 4: Group 5: Patients who displayed MRD negativity immediately (on the first measurement between days +22 and +36 days) post transplant and stayed so (n ¼ 4; no relapse) Patients who converted to MRD negativity upon immunosuppression reduction (n ¼ 12, 1 relapse). Patients who converted to MRD negativity upon DLI (n ¼ 6, no relapse) Patients who did not respond to DLI or for whom DLI were indicated but not available (n ¼ 2, 2 relapses) Patients who responded to GVL but failed to reach complete MRD clearance (n ¼ 4; 3 relapses) 1E-5 1E-6 CR PR SD PD Landmark E+1 1E+0 1E-1 Figure 2 MRD response patterns. (a) CSA tapering complete responders (Group 2); (b), DLI responders and non-responders due to the absence of GVL (Groups 3, 4); (c), secondary GVL resistance (Group 5). The horizontal line at 1E-5 symbolizes the threshold of MRD detectability; that is, MRD-negative values are plotted below this line. The numbers tabulated at the bottom indicate clinical response at landmark. MRD-positive CRs in grey. Landmark: (1) pretransplant; (2) d20 40 post transplant; (3) pre-csa tapering*; (4) best response post- CSA tapering; (5) pre DLI; (6) best response post DLI; (7) last follow-up. Each curve depicts MRD kinetics of one patient. Black curves, RQ- PCR data; grey curves MRD flow data. *Pre-chronic GVHD in UPN 041_005 (compare Supplementary Figure 2d). CSA, cyclosporin A; DLI, donor lymphocyte infusions; GVHD, graft-versus-host disease; GVL, graft-versus-leukemia; MRD, minimal residual disease; RQ-PCR, real-time quantitative-pcr. MRD level 1E-2 1E-3 1E-4 1E-5 1E-6 CR PR SD PD 1 3 Landmark

8 1384 MRD level 1E+0 1E-1 1E-2 1E-3 1E-4 1E-5 CSA Rituximab GVHD days pre/post allo-sct Figure 3 MRD kinetics in a patient acquiring GVL resistance (UPN ). Bold line, MRD by RQ-PCR; thin line MRD by MRD-flow. GVL, graft-versus-leukemia; MRD, thin minimal residual disease; RQ- PCR, real-time quantitative-pcr. Probability of MRD response In the study group, the median time to MRD negativity measured from the time of transplant was 11 months. Considering all 30 patients including the two early TRM cases and the three patients without MRD monitoring, the probability of being alive and MRD negative was 50% at 12 months andfdue to the effects of DLIF60% at 24 months, respectively. Of the four patients with del 17p-, two (UPN 005_004 and 144_010) never achieved complete remission (CR), remained MRD positive and succumbed to progressive CLL, whereas the remaining two (UPN 013_001 and 144_005) reached sustained CR. While UPN 013_001 had no MRD samples available, UPN 144_005 became durably MRD negative upon withdrawal of immunosuppression (Table 3, Supplementary Figure 2). Acquisition of GVL resistance Four patients (Group 5) showed profound but not complete MRD response in spite of active chronic GVHD. One of them (UPN ) was available for very close MRD monitoring by both RQ-PCR and MRD flow. MRD kinetics in this particular patient suggests that GVL resistance can be acquired under selection pressure of GVHD (Figure 3): After tapering of CSA, the MRD level started to decrease soon followed by clinical signs of chronic GVHD. Reinstitution of immunosuppressive therapy was accompanied by a slight increase of MRD. After control of GVHD, immunosuppression was again tapered and stopped at day þ 254 with subsequent further two log MRD reduction. However, recurrence of extensive chronic GVHD forced to resume CSA treatment at day þ 280. Although GVHD could not be completely controlled and CSA was stopped again later on, MRD levels now started to rise and could not be reverted until clinical relapse at day þ 330. Durability and prognostic impact of MRD response Except one, all 25 patients who had become MRD negative by month þ 12 or beyond remained so throughout the whole follow-up. In UPN 05002, RQ-PCR became positive at the time of extramedullary relapse (presenting as aggressive lymphoma of the stomach) at month þ 33. This was the only case of disease recurrence among those who had reached MRD negativity by month þ 12. Another patient (41007) succumbed to infectious complications of GVHD at month þ 28, translating into an EFS of patients MRD negative 12 months post transplant or beyond of 91% (78, 100%). In contrast, five out of six patients who failed to clear MRD due to secondary GVL resistance or lack of GVL relapsed clinically (Supplementary Figures 2c and d). Concordance of MRD and clinical remission status We did not observe a single case of MRD negativity in the presence of clinical evidence of disease. On the other hand, 10 patients were MRD positive while being in clinical CR (two of 12 in Group 2; five of six in Group 3; one of two in Group 4; two of four in Group 5). A detailed correlation of clinical remission status and MRD kinetics is presented in Figure 2. Of note, two of four patients who became secondary GVL resistant did so when they were in clinical (MRD positive) CR (Figure 2c). Discussion Although there have been some reports describing MRD kinetics after allo-sct in high-risk CLL, 1,2 essential questions remained unanswered. These include (i) the likelihood of achieving MRD negativity, (ii) whether it is durable or (iii) has prognostic impact on clinical outcome and (iv) most importantly, how MRD kinetics might reflect time course, clinical triggers and resistance mechanisms of GVL as the key therapeutic principle. Taking advantage of a large prospective clinical trial on RIC allo-sct in high-risk CLL, this study is the first to provide a detailed correlation of MRD kinetics with immunomodulating events, namely immunosuppression tapering, DLI and GVHD, on an informative set of patients. This allowed the identification of certain distinct patterns of MRD response, which can be basically described as those who achieve MRD negativity without direct evidence of GVL (Group 1), those who show a complete MRD response to GVL (Groups 2 and 3), those who respond to GVL but fail to reach complete MRD clearance (Group 5) and those in whom a GVL effect with subsequent MRD suppression cannot be induced (Group 4). For the purposes of grouping, GVL was assumed if stable or increasing MRD levels became undetectable upon immunomodulation, or if more than 1 log decrease occurred upon immunomodulation followed by de novo chronic GVHD. With regard to the first question, our data show that RIC allo- SCT can induce MRD eradication in more than 50% of the patients with well-characterized high-risk CLL in a prospective multicenter setting, translating into a 3-year EFS of study patients of 58%. For the first time, we could show this phenomenon in a fludarabine-refractory patient with del 17p- (UPN 144_005). As both MRD-negative survival and EFS are critically dependent on TRM, the low TRM observed here has to be taken into account when evaluating these favorable results. The reasons for the modest TRM despite multicenter design are not clear. However, it has to be recalled that the inclusion criteria of the CLL3X trial virtually precluded enrollment of patients with higher comorbidity score (hematopoietic cell transplantation comorbidity index). A low hematopoietic cell transplantation comorbidity index score has been shown to be a major determinant of favorable TRM and overall mortality, especially in patients with indolent lymphoid malignancies treated with RIC allo-sct. 13,14 Apart from the pioneer Dana Farber data obtained with TCD transplants, 15 MRD studies after allo-sct for CLL at a sensitivity of 1 in 10 4 or higher have been sparse. The Seattle group provided information on qualitative PCR monitoring in a subset

9 of patients reaching clinical CR. They observed ongoing MRD negativity paralleling complete clinical remissions in eight of nine patients at risk. The ninth patient became also PCR negative but reconverted to MRD positivity at the time of clinical relapse. 16 Based on 13 patients, the Barcelona group showed by quantitative assays that MRD negativity can be achieved after allo-sct with myeloablative conditioning, mostly with delayed kinetics. Moreover, in four patients, sustained disease control was obtained despite persisting low-level MRD positivity. 2 In contrast, in this RIC series, all patients who failed MRD clearance due to lack of GVL or secondary GVL resistance showed increasing MRD levels on longitudinal analysis, and five of six subsequently relapsed. An Italian consortium monitored MRD by PCR in 15 patients achieving CR after RIC allo-sct and reported absence of clinical relapses in six of six patients who had achieved MRD negativity at the 6-month landmark, whereas a continuing pattern of CLL recurrence was observed in those patients who displayed persistent PCR positivity or a mixed pattern. 17 In terms of prognostic impact of MRD response, MRD negativity at 12 months post transplant or beyond was a very strong predictor of durable relapse-free and MRD-negative survival in this study. One patient subsequently died of late GVHD-related complications, and one of extramedullary/extranodal relapse. Although the patients who acquired GVL resistance illustrate that absent or very low MRD levels not necessarily mean cure, it may be concluded that MRD negativity at the 12-month landmark heralds long-term disease control and excellent survival. Accordingly, absence of MRD at this landmark may well serve as an early indicator of long-term treatment success. This has implications not only for patient care but also for definition of appropriate endpoints for interventional trials in CLL. MRD kinetics and correlation to clinical immunomodulation observed in this study impressively confirms that GVL is the key contributor to MRD response. 3,4 In the setting chosen here, immunosuppression tapering was the most important element for GVL induction. As GVL was almost always observed in the context of chronic GVHD, our results strongly suggest alloreactivity rather than CLL-specific effects as principle mechanism responsible for GVL activity. This implies that it should be difficult to reliably segregate GVL from GVHD in CLL, or, in other words, a certain risk of chronic GVHD has to be taken into account for successful immunotherapy in this disease. Another novel and particularly interesting finding was the fact that all patients who had chronic GVHD had some evidence of GVL activity. Even those who finally failed to achieve complete MRD eradication showed, nevertheless, a significant and usually rather prompt MRD response (Group 5) upon GVHD onset, which was, however, only temporary. This suggests a kind of secondary GVL resistance, which might be acquired during the course of the disease as documented by patient who lost MRD response after repeated episodes of systemic immunosuppressive therapy. As fluorescence in situ hybridization karyotyping was not performed after relapse, it remains unclear whether secondary GVL resistance was due to clonal evolution. Other explanations might be survival of clonogenic cells at GVL-sanctuary sites as illustrated by the course of patient 05002, or the tumor stem cell hypothesis. 18 Finally, secondary GVL resistance could be due to the development of tolerance, but this explanation is questioned by the observation that all four patients displaying this pattern had ongoing extensive chronic GVHD. Thus, in addition to the lack of GVL induction, our study identifies secondary GVL resistance as another basic mechanism for treatment after allo-sct in CLL. On the other hand, cases suggestive of primary GVL resistance, that is, no MRD response after chronic GVHD onset, could not be observed in this series. The core feature of our approach was the use of quantitative MRD monitoring. The principal advantage of quantitative over qualitative MRD assessment is that it documents not only the presence or absence of disease but also tumor load kinetics. This is particularly important for an ongoing therapeutic process such as GVL after allo-sct. Even if residual clinical disease is present, quantitative MRD measurement may provide valuable information; for example, could the direction of MRD kinetics influence therapeutic decisions in patients with persisting lymph nodes. Moreover, quantitative MRD monitoring seems to be a valid instrument for sensitive guidance of immune intervention directed at disease eradication after allo-sct in CLL. This could include timely initiation of DLI as well as MRD-adapted modulation of immunosuppressive medication in a way that allows for limited or modest chronic GVHD in case of persistent MRD. Although not formally defined in the protocol, the latter policy was adopted by two centers representing more than 50% of the patients and might be one reason for the high incidence of chronic GVHD and subsequent MRD clearance observed here. As a consequence, only few patients remained eligible for preemptive DLI. Including the pilot phase, DLI was given to eight patients, in six cases merely upon persisting or increasing MRD levels in the absence of clinical disease ( preemptive DLI). Notably, a response was seen in five of the latter (83%), which compares favorably to success rates reported previously for clinical relapse-triggered ( therapeutic ) DLI. 16,19 21 Thus, similar to other entities, GVL efficacy in CLL may depend on the tumor load. In summary, this study provides a comprehensive map of possible MRD courses after allo-sct in high-risk CLL. As illustrated by MRD kinetics, effective GVL activity is induced in virtually all patients who develop chronic GVHD. However, in a significant proportion of cases, this does not translate into sustained disease control due to development of secondary GVL resistance. This data may help to understand the physiology of GVL activity and resistance to it, and gives a rationale basis for designing MRD-guided interventional studies on allo-sct in CLL. Acknowledgements We thank the German CLL Study Group CLL3X investigators for their participation in this study; a complete list of investigators appears in Appendix. This study was supported by grants from the Deutsche Jose- Carreras Leukämiestiftung e.v. Projects R02/18, R05/02 (MR, PD and MK). The authors declare that no potential conflict of interest exists. References 1 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: Moreno C, Villamor N, Esteve J, Colomer D, Bosch F, Campo E et al. Clinical significance of minimal residual disease, as assessed 1385

10 1386 by different techniques, after stem cell transplantation for chronic lymphocytic leukemia. Blood 2006; 107: Dreger P, Corradini P, Kimby E, Michallet M, Milligan D, Schetelig J et al. Indications for allogeneic stem cell transplantation in chronic lymphocytic leukemia: the EBMT transplant consensus. 2007; 21: Ben Bassat I, Raanani P, Gale RP. Graft-versus-leukemia in chronic lymphocytic leukemia. Bone Marrow Transplant 2007; 39: Khouri IF, Saliba RM, Admirand J, O Brien S, Lee MS, Korbling M et al. Graft-versus-leukaemia effect after non-myeloablative haematopoietic transplantation can overcome the unfavourable expression of ZAP-70 in refractory chronic lymphocytic leukaemia. Br J Haematol 2007; 137: Dreger P, Brand R, Milligan D, Corradini P, Finke J, Lambertenghi- Deliliers G et al. Reduced-intensity conditioning lowers treatmentrelated mortality of allogeneic stem cell transplantation for chronic lymphocytic leukemia: a population-matched analysis. 2005; 19: Rawstron AC, Villamor N, Ritgen M, Bottcher S, Ghia P, Zehnder JL et al. International standardized approach for flow cytometric residual disease monitoring in chronic lymphocytic leukaemia. 2007; 21: Kröber A, Seiler T, Benner A, Bullinger L, Brückle E, Lichter P et al. VH mutation status, CD38 expression level, genomic aberrations, and survival in chronic lymphocytic leukemia. Blood 2002; 100: Ritgen M, Lange A, Stilgenbauer S, Dohner H, Bretscher C, Bosse H et al. Unmutated immunoglobulin variable heavy chain gene status remains an adverse prognostic factor after autologous stem cell transplantation for chronic lymphocytic leukemia. Blood 2003; 101: Döhner H, Stilgenbauer S, Benner A, Leupolt E, Krober A, Bullinger L et al. Genomic aberrations and survival in chronic lymphocytic leukemia. N Engl J Med 2000; 343: 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: Rawstron AC, Kennedy B, Evans PA, Davies FE, Richards SJ, Haynes AP et al. Quantitation of minimal disease levels in chronic lymphocytic leukemia using a sensitive flow cytometric assay improves the prediction of outcome and can be used to optimize therapy. Blood 2001; 98: Sorror ML, Maris MB, Storb R, Baron F, Sandmaier BM, Maloney DG et al. Hematopoietic cell transplantation-specific comrobidity index (HCT-CI): a new tool for risk assessment before allogeneic HCT. Blood 2005; 106: Sorror ML, Storer BE, Maloney DG, Sandmaier BM, Martin PJ, Storb R. Outcomes after allogeneic hematopoietic cell transplantation with nonmyeloablative or myeloablative conditioning regimens for treatment of lymphoma and chronic lymphocytic leukemia. Blood 2008; 111: Gribben JG, Zahrieh D, Stephans K, Bartlett-Pandite L, Alyea EP, Fisher DC et al. Autologous and allogeneic stem cell transplantation for poor risk chronic lymphocytic leukemia. Blood 2005; 106: Sorror ML, Maris MB, Sandmaier BM, Storer B, Jain-Stuart M, Al-Ali H et al. Hematopoietic cell transplantation after nonmyeloablative conditioning for advanced chronic lymphocytic leukemia. J Clin Oncol 2005; 23: Corradini P, Dodero A, Farina L, Fanin R, Patriarca F, Miceli R et al. Allogeneic stem cell transplantation following reducedintensity conditioning can induce durable clinical and molecular remissions in relapsed lymphomas: pre-transplant disease status and histotype heavily influence outcome. 2007; 21: Huff CA, Matsui W, Smith BD, Jones RJ. The paradox of response and survival in cancer therapeutics. Blood 2006; 107: 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: Delgado J, Thomson K, Russell N, Ewing J, Stewart W, Cook G et al. Results of alemtuzumab-based reduced-intensity allogeneic transplantation for chronic lymphocytic leukemia: a British Society of Blood and Marrow Transplantation study. Blood 2006; 107: Brown JR, Kim HT, Li S, Stephans K, Fisher DC, Cutler C et al. Predictors of Improved Progression-Free Survival After Nonmyeloablative Allogeneic Stem Cell Transplantation for Advanced Chronic Lymphocytic. Biol Blood Marrow Transplant 2006; 12: Supplementary Information accompanies the paper on the website ( Appendix CLL3X Investigators Berlin, Charite Benjamin Franklin (Lutz Uharek); Essen, Universitätsklinikum (Dietrich Beelen); Göttingen, Universitätsklinikum (Bertram Glass); Hamburg, AK St Georg (Norbert Schmitz); Hannover, Medizinische Hochschule (Bernd Hertenstein, Michael Stadler, Matthias Eder); Heidelberg, Innere Medizin V (Peter Dreger, Manfred Hensel); Homburg, Universitätsklinikum (Jörg Schubert); Kiel, II. Medizinische Klinik (Martin Gramatzki); Köln, Universitätsklinikum (Michael Hallek); Marburg, Universitätsklinikum (Andreas Burchert); Montreal, Hopital Maisonneuve Rosemont (Sandra Cohen); Leipzig, Universitätsklinikum (Dietger Niederwieser); Regensburg, Universitätsklinikum (Ernst Holler); Ulm, III. Medizinische Klinik (Donald Bunjes, Stephan Stilgenbauer, Hartmut Döhner).