KEY WORDS. Acute lymphoblastic leukemia Etoposide Cyclophosphamide Total body irradiation Conditioning regimen Sibling allografts

Transcription

1 Biology of Blood and Marrow Transplantation 12: (2006) 2006 American Society for Blood and Marrow Transplantation /06/ $32.00/0 doi: /j.bbmt A Comparison of Cyclophosphamide and Total Body Irradiation with Etoposide and Total Body Irradiation as Conditioning Regimens for Patients Undergoing Sibling Allografting for Acute Lymphoblastic Leukemia in First or Second Complete Remission David I. Marks, 1 Stephen J. Forman, 2 Karl G. Blume, 3 Waleska S. Pérez, 4 Daniel J. Weisdorf, 5 Armand Keating, 6 Robert Peter Gale, 7 Mitchell S. Cairo, 8 Edward A. Copelan, 9 John T. Horan, 10 Hillard M. Lazarus, 11 Mark R. Litzow, 12 Philip L. McCarthy, 13 Kirk R. Schultz, 14 David D. Smith, 2 Michael E. Trigg, 15 Mei-Jie Zhang, 4 Mary M. Horowitz 4 1 Adult BMT Unit, Bristol Children s Hospital, Bristol, United Kingdom; 2 City of Hope National Medical Center, Duarte, California; 3 Stanford University Medical Center, Stanford, California; 4 Center for International Blood and Marrow Transplant Research, Health Policy Institute, Medical College of Wisconsin, Milwaukee, Wisconsin; 5 University of Minnesota, Minneapolis, Minnesota; 6 University of Toronto, Toronto, Ontario, Canada; 7 Center for Advanced Studies of Leukemia, Los Angeles, California; 8 Columbia University, New York, New York; 9 The Ohio State University, Columbus, Ohio; 10 Emory University, Atlanta, Georgia; 11 University Hospitals of Cleveland, Cleveland, Ohio; 12 Mayo Clinic Rochester, Rochester, Minnesota; 13 Roswell Park Cancer Institute, Buffalo, New York; 14 British Columbia s Children s Hospital, Vancouver, British Columbia, Canada; 15 Alfred I. DuPont Hospital for Children, Wilmington, Delaware Correspondence and reprint requests: Mary M. Horowitz, MD, MS, Center for International Blood and Marrow Transplant Research, Medical College of Wisconsin, 8701 Watertown Plank Rd., P.O. Box 26509, Milwaukee, WI ( marymh@mcw.edu). Received August 3, 2005; accepted December 5, 2005 ABSTRACT We compared the outcomes of 298 patients with acute lymphoblastic leukemia in first or second complete remission (CR1 or CR2) receiving HLA-matched sibling allografts after cyclophosphamide and total body irradiation (Cy-TBI) conditioning with 204 patients receiving etoposide and TBI. Consequently, 4 groups were compared: Cy-TBI <13 Gy (n 217), Cy-TBI >13 Gy (n 81), etoposide-tbi <13 Gy (n 53), and etoposide-tbi >13 Gy (n 151). Analyses of relapse, leukemia-free survival (LFS), and survival were performed separately for CR1 and CR2 transplantations. Transplant-related mortality did not differ by conditioning regimen. In CR1, there were also no significant differences in relapse, LFS, or survival by conditioning regimen. In CR2, these outcomes differed among conditioning groups. In comparison with Cy-TBI <13 Gy, the risks of relapse, treatment failure (inverse of LFS), and mortality tended to be lower with etoposide (regardless of TBI dose) or with TBI doses >13 Gy. For both CR1 and CR2 transplantations, causes of death were similar among the groups; disease recurrence accounted for 47% of deaths. We conclude that for HLA-identical sibling allografts for acute lymphoblastic leukemia in CR2, there is an advantage in substituting etoposide for Cy or, when Cy is used, in increasing the TBI dose to >13 Gy American Society for Blood and Marrow Transplantation KEY WORDS Acute lymphoblastic leukemia Etoposide Cyclophosphamide Total body irradiation Conditioning regimen Sibling allografts 438

2 Comparison of Conditioning Regimens for ALL Sibling Allografts INTRODUCTION Allogeneic stem cell transplantation cures hematologic malignancies through 2 major mechanisms. The pretransplantation conditioning regimen kills leukemic cells directly, either eradicating the malignancy or reducing it to a minimal residual disease state. The second mechanism is a graft-versus-tumor effect. The relative importance of these 2 mechanisms differs from disease to disease: graft-versus-tumor effects are most important in chronic myelogenous leukemia and less important, although demonstrable, in acute myelogenous leukemia and acute lymphoblastic leukemia (ALL) [1]. For patients with ALL, the cytotoxic effect of pretransplantation chemoradiotherapy may be important in determining outcome. A graftversus-leukemia (GVL) effect may be effective only if there is a low level of residual disease, given the rapid kinetics of ALL cell growth. The results of conventional chemotherapy for adults with ALL are unsatisfactory, and there has been little progress in the last decade. Even though chemotherapy fails in most patients, the role of HLAmatched sibling allografting remains controversial. An International Bone Marrow Transplant Registry (IBMTR) study showed that HLA-identical sibling allografting produced a clear survival advantage over chemotherapy for children with ALL in second complete remission (CR) [2]. In general, only children with very-high-risk disease (Philadelphia positive or induction failure) receive allografts in first CR. For adults with ALL in second CR, survival after chemotherapy alone is so low that allografting seems clearly superior [3]. However, the results of allografting for ALL in second CR are far from acceptable, particularly for patients who relapse while receiving therapy. In young adults with ALL in first CR, there are conflicting data about the value of transplantation. One IBMTR study showed no advantage, but a more recent study showed that transplantation was superior in younger adults [4,5]. A recent randomized French study of 284 adults with ALL in first CR showed 45% 5-year disease-free survival (DFS) with allografting compared with 31% with chemotherapy or autografting (P.10) [6]. Subgroup analysis suggested a significant survival advantage with allogeneic transplantation in patients with high-risk disease features. The major cause of treatment failure after allogeneic transplantation for ALL is relapse. Very few patients survive this complication. Although there is evidence for a GVL effect in ALL, the poor results of donor leukocyte infusions for leukemic relapse after allografting suggest that this effect may be limited [1,7]. The optimal conditioning regimen before HLAmatching sibling allografts for patients with ALL is unknown. Total body irradiation (TBI) is frequently used because of the predictable toxicity of radiotherapy BB&MT and its ability to kill leukemic cells in sanctuary sites, such as the central nervous system and the testis. Cyclophosphamide (Cy; 120 mg/kg) and TBI (12 Gy in 6 fractions), although developed on empirical observations, is regarded as the standard regimen. It has been used for 30 years and has relatively predictable shortand long-term toxicity [8]. In one study, higher doses of TBI seemed to decrease leukemic relapse but, because of extramedullary toxicity, did not improve survival [9]. However, the optimal dose of TBI is not known. A conditioning regimen of etoposide (60 mg/kg) and TBI (13.2 Gy) was first evaluated by the Stanford and City of Hope (COH) groups and seemed to be highly active in patients with leukemia [10]. This group subsequently demonstrated considerable activity of this regimen in patients with advanced acute leukemia [11]. A German study using the same regimen reported 64% DFS at 18 months in 17 high-risk patients (10 were not in remission at transplantation) [12]. With these encouraging results, the regimen was then adopted for patients with less advanced disease. Using the etoposide-tbi regimen for conditioning, these groups reported 64% DFS at 3 years for 34 patients with ALL in first CR and a relapse rate of 12% [13]. These data were updated at the 1999 American Society of Hematology meeting by Fung et al.; among 102 patients who underwent transplantation in first CR, DFS was 63%, and only 8% died of relapse [14]. Although the median follow-up of the expanded series was only 2.1 years, it is worth noting that relapses were rarely seen more than 2 years after transplantation. Initial results were sufficiently good for the large international Medical Research Council UKALL12/Eastern Cooperative Oncology Group study to adopt this conditioning regimen as its standard in A 5-year DFS of 54% was seen in 170 patients with Philadelphia-negative ALL conditioned with etoposide-tbi [15]. However, the 100-day transplant-related mortality (TRM) in the multicenter, multinational setting was 21% [15]. The Stanford and COH groups found that the etoposide-tbi regimen caused considerable mucosal and dermatologic toxicity but that TRM was low [13]. The mucositis seen with etoposide-tbi may result in less prophylactic methotrexate being given (D. Marks, unpublished data). Although the best reported results of the etoposide- TBI conditioning seem superior to those achieved with standard Cy-TBI, there has never been a study directly comparing the 2 regimens. Here we report a registrybased comparison of the outcome of more than 500 adult and pediatric patients with ALL in first or second CR who received either Cy-TBI or etoposide-tbi before an HLA-matched sibling transplantation. Outcomes studied were TRM, relapse, leukemia-free survival (LFS), overall survival, and causes of death. 439

3 D. I. Marks et al. Table 1. Variables Tested in Cox Proportional Hazards Regression Models Main effect variable* Conditioning regimen before transplantation: Cy-TBI, fractionated TBI dose <13 Gy vs. etoposide-tbi, fractionated TBI dose <13 Gy vs. etoposide-tbi, fractionated TBI dose >13 Gy vs. Cy-TBI, fractionated TBI dose >13 Gy Patient-related variables Age at transplantation: <20 vs. >20 y Sex: female vs. male Karnofsky performance status at transplantation: >90% vs. <90% vs. unknown Disease-related variables at diagnosis Immunophenotype: T-cell vs. other WBC at diagnosis: < /L vs /L vs /L vs. > /L vs. missing Cytogenetics: no abnormalities vs. hypodiploid, t(9;22), t(4;11), t(8;14) vs. other abnormalities vs. unknown Extramedullary disease: no vs. yes Disease-related variables at transplantation Relapse on chemotherapy: no vs. yes vs. unknown Time from diagnosis to first complete remission: <4 wk vs. 4-8 wk vs. >8 wk vs. missing Disease status before transplantation: CR1 vs. CR2 Duration of CR1: <12 mo vs mo vs mo vs. >36 mo vs. missing Time from remission to transplantation: <3 mo vs. >3 mo vs missing Treatment-related variables Nucleated cell dose, median (range), 10 8 /kg: continuous Donor age: <20 y vs. >20 y Donor-recipient sex match: male-male vs. female-male vs. male-female vs. female-female Donor-recipient CMV status: / vs. / vs. / vs. / vs. missing Source of stem cells: BM vs. PBSC Year of transplantation: vs GVHD prophylaxis: CSP other vs. MTX CSP alone vs. MTX CSP other WBC indicates white blood cell count; CMV, cytomegalovirus; BM, bone marrow; CSP, cyclosporine; MTX, methotrexate. *Included in all models. Reference group for analysis. Data not available for City for Hope patients. PATIENTS AND METHODS Data Sources Data for this study were obtained from 2 sources: the Center for International Blood and Marrow Transplant Research (CIBMTR; n 427), which pools data from many centers, and the COH Cancer Center (n 75), which did not contribute data to the CIBMTR during the study period. The CIBMTR is a research affiliation of the IBMTR, the Autologous Blood and Marrow Transplant Registry, and the National Marrow Donor Program that comprises a voluntary working group of more than 450 transplantation centers worldwide that contribute detailed data on consecutive allogeneic and autologous hematopoietic stem cell transplantations to a statistical center at the Health Policy Institute of the Medical College of Wisconsin in Milwaukee or the National Marrow Donor Program Coordinating Center in Minneapolis, MN. Participating centers are required to report all transplantations consecutively; compliance is monitored by on-site audits. Patients are followed up longitudinally, with yearly follow-up. The CIBMTR collects data at 2 levels: registration and research. Registration data include disease type, age, sex, pretransplantation disease stage and chemotherapy responsiveness, date of diagnosis, graft type (bone marrow derived and/or blood-derived stem cells), high-dose conditioning regimen, posttransplantation disease progression and survival, development of a new malignancy, and cause of death. Requests for data on progression or death for registered patients are at 6-month intervals. All CIBMTR teams contribute registration data. Research data are collected on subsets of registered patients and include comprehensive pretransplantation and posttransplantation clinical information. Computerized checks for errors, physician reviews of submitted data, and on-site audits of participating centers ensure the quality of the data. The COH Cancer Center is a National Cancer Institute designated Comprehensive Cancer Center and one of the largest providers of bone marrow transplantation services worldwide. Data for this study were provided to the CIBMTR statistical center for all patients who met the eligibility criteria and underwent transplantation during the study period: data were prospectively collected and maintained by the COH in its computerized database system. Patients This study included patients receiving HLA-identical sibling bone marrow or peripheral blood stem cell (PBSC) transplants for ALL in first or second CR between 1989 and 1998 after pretransplantation conditioning with Cy-TBI or etoposide-tbi, reported on comprehensive data collection forms to the CIBMTR or performed at the COH transplant center. Demo- 440

4 BB&MT 441 Table 2. Characteristics of Patients Receiving HLA-Identical Sibling Transplants for ALL in First or Second Complete Remission by Pretransplantation Conditioning Regimen Variable Cy-TBI <13 Gy Cy-TBI >13 Gy Etoposide-TBI <13 Gy Etoposide-TBI >13 Gy patients Median (range) age (y) (1-56) (2-46) (2-41) (4-54) <.001 Age at transplantation, n (%) <.001 <10 y 60 (28) 29 (36) 24 (45) 12 (8) y 64 (29) 22 (27) 15 (28) 38 (25) y 47 (22) 18 (22) 10 (19) 45 (30) y 30 (14) 10 (12) 3 (6) 37 (24) >40 y 16 (7) 2 (3) 1 (2) 19 (13) Male sex, n (%) (67) (56) (74) (68).13 Karnofsky status before transplantation <90%, n (%) (17) (19) (27) (18).36 CRI status at transplantation (56) (47) (45) (72) <.001 Immunophenotype, n (%) <.001 T-cell 40 (19) 16 (20) 10 (19) 30 (20) Mature B-cell 20 (9) 5 (6) 0 8 (5) Null 13 (6) 7 (9) 4 (7) 5 (3) CALLa 111 (51) 40 (49) 30 (57) 51 (34) Other 33 (15) 13 (16) 9 (17) 57 (38) Median (range) WBC at diagnosis (n 10 9 /L) (<1-850) (<1-680) (1-488) (<1-507).07 WBC at diagnosis, n (%) < /L 125 (63) 38 (53) 22 (52) 76 (53) 25-< /L 46 (23) 17 (24) 13 (31) 38 (27) 100-< /L 13 (6) 9 (13) 2 (5) 13 (9) > /L 15 (8) 7 (10) 5 (12) 16 (11) Cytogenetics, n (%) No abnormalities 77 (36) 20 (25) 17 (32) 35 (23) Hypodiploid, t(9;22), t(4;11); t(8;14) 29 (13) 6 (7) 6 (11) 30 (20) Other abnormalities 45 (21) 18 (22) 12 (23) 36 (24) Unknown 66 (30) 37 (46) 18 (34) 50 (33) Extramedullary disease at diagnosis or relapse, n (%) (30) (37) (23) (37).16 Site of extramedullary disease at diagnosis, n (%) CNS (5) 77 3 (4) 51 2 (4) (6).89 Testes (1) Mediastinum (5) 78 6 (8) 51 3 (6) (10).33 Other (15) (14) 51 3 (6) (9).16 Site of extramedullary disease at relapse before treatment, n (%) CNS (6) 51 6 (12) 32 3 (9) (12).34 Other (1) Relapse on chemotherapy, n (%) (13) (16) 50 9 (18) (5).021 Time from Dx to CRI, n (%) <.001 <4 wk 59 (30) 27 (36) 10 (21) 64 (45) 4-8 wk 106 (53) 30 (41) 20 (42) 60 (43) >8 wk 35 (17) 17 (23) 18 (37) 17 (12) Duration of CRI, n (%) <12 mo 22 (27) 8 (21) 5 (21) 2 (13) mo 15 (18) 4 (11) 8 (33) 5 (31) P Value* Comparison of Conditioning Regimens for ALL Sibling Allografts

5 442 Table 2. Continued Variable Cy-TBI <13 Gy Cy-TBI > 13 Gy Etoposide-TBI <13 Gy Etoposide-TBI >13 Gy mo 21 (26) 14 (38) 5 (21) 5 (31) >36 mo 24 (29) 11 (30) 6 (25) 4 (25) Median (range) time from remission to transplantation (d) (6-793) (9-831) (17-325) (7-508).007 Time from remission to transplantation >3 mo, n (%) (49) (65) (55) (45).029 Median (range) nucleated cell dose (n 10 8 /kg) (.01-18) 63 3 (.02-20) 37 4 (1-41) 74 3 (.02-17).05 Median (range) donor age (y) (<1-52) (1-51) (1-41) (<1-56) <.001 Donor-recipient sex match, n (%) Male-male 85 (39) 24 (29) 20 (38) 57 (38) Female-male 60 (28) 21 (26) 19 (36) 35 (23) Male-female 40 (18) 20 (25) 6 (11) 35 (23) Female-female 32 (15) 16 (20) 8 (15) 24 (16) Donor-recipient CMV status, n (%) / 84 (41) 20 (25) 20 (40) 70 (50) / 29 (14) 8 (10) 8 (16) 22 (15) / 15 (8) 11 (14) 6 (12) 17 (12) / 75 (37) 41 (51) 16 (32) 32 (23) Graft type, n (%) BM 201 (93) 72 (89) 47 (89) 146 (97) PBSC 16 (7) 9 (11) 6 (11) 5 (3) Year of transplantation, n (%) (18) 6 (8) 3 (6) 23 (15) (13) 18 (22) 11 (21) 30 (20) (18) 13 (16) 10 (19) 35 (23) (26) 26 (32) 17 (32) 36 (24) (25) 18 (22) 12 (22) 27 (18) Median (range) TBI doses, (Gy) ( ) ( ) ( ) ( ) <.001 Median (range) drug doses, mg/kg ( ) ( ) ( ) ( ) NA GVHD prophylaxis, n (%) <.001 CsA alone 37 (17) 13 (16) 2 (3) 7 (5) CsA steroids 8 (4) 7 (9) 4 (8) 25 (16) CsA other 1(<1) 1 (1) 0 1 (1) MTX CsA alone 160 (74) 56 (69) 43 (81) 36 (24) MTX CsA steroids 8 (4) 3 (4) 4 (8) 76 (50) MTX CsA other 3 (1) 1 (1) 0 6 (4) Intrathecal therapy, n (%) (7) (13) 50 2 (4) (23) <.001 Median (range) follow-up survivors (mo) 67 (8-156) 59 (17-149) 52 (7-144) 65 (3-156) indicates evaluable; ALL, acute lymphoblastic leukemia; Cy, cyclophosphamide; TBI, total body irradiation; CR, complete remission; WBC, white blood cells; CNS, central nervous system; BM, bone marrow; PBSC, peripheral blood stem cells; GVHD, graft-versus-host disease; CSA, cyclosporine; MTX, methotrexate; Dx, diagnosis; CMV, cytomegalovirus; calla, common acute lymphoblastic leukemia antigen. *The 2 test was used for discrete covariates; the Kruskal-Wallis test was used for continuous covariates. Thirty-four (49%) of these patients had t(9;22). For patients in CR2. Data not available for City of Hope patients. Other GVHD prophylaxis treatments were antithymocyte globulin, azathioprine, and in vivo anti T-lymphocyte monoclonal antibody. None of these patients received immune therapy posttransplant. P Value* D. I. Marks et al.

6 Comparison of Conditioning Regimens for ALL Sibling Allografts Table 3. Univariate Analyses of Transplantation Outcomes among Patients Receiving HLA-Identical Sibling Transplants for ALL in First Complete Remission by Pretransplantation Conditioning Regimen Cy-TBI <13 Gy Cy-TBI >13 Gy Etoposide-TBI <13 Gy Etoposide-TBI >13 Gy Outcome Event Prob (95% CI) Prob (95% CI) Prob (95% CI) Prob (95% CI) Pulmonary toxicity, n (%) (28) 35 9 (26) 23 7 (30) (37) Grade II-IV acute GVHD at 100 d (21-38) (11-39) (14-50) (17-34) Chronic GVHD At 1 y 23 (16-32) 23 (10-40) 19 (4-42) 34 (24-45) At 3 y 28 (20-38) 27 (13-45) 19 (4-42) 42 (32-53) At 5 y 30 (21-39) 27 (13-45) 19 (4-42) 44 (33-55) 100-d mortality (4-14) (4-26) (7-39) (7-19) TRM At 1 y 9 (5-15) 13 (4-26) 23 (8-43) 9 (5-15) At 3 y 12 (7-18) 19 (8-33) 27 (10-48) 10 (5-17) At 5 y 13 (7-19) 19 (8-33) 27 (10-48) 12 (6-19) Relapse At 1 y 23 (15-31) 16 (6-30) 9 (8-25) 12 (7-19) At 3 y 32 (24-41) 24 (11-40) 13 (5-31) 23 (15-31) At 5 y 33 (24-42) 28 (14-45) 13 (2-31) 27 (19-36) LFS At 1 y 68 (59-76) 69 (54-83) 67 (46-85) 79 (70-86) At 3 y 56 (47-65) 56 (39-71) 57 (36-77) 67 (58-76) At 5 y 54 (45-63) 51 (35-68) 57 (36-77) 61 (51-70) Overall survival At 1 y 74 (65-81) 74 (59-86) 71 (51-87) 80 (71-87) At 3 y 59 (50-68) 63 (47-77) 61 (41-80) 67 (58-75) At 5 y 55 (46-64) 60 (44-75) 61 (41-80) 62 (52-71) Prob indicates probability; CI, confidence interval;, evaluable; LFS, leukemia-free survival; GVHD, graft-versus-host disease; TRM, transplant-related mortality; Cy, cyclophosphamide; TBI, total body irradiation; ALL, acute lymphoblastic leukemia. Probabilities of 100-day mortality, leukemia-free survival, and overall survival were calculated by using the Kaplan-Meier product-limit estimate. Acute GVHD, chronic GVHD, relapse, and TRM were calculated by using the cumulative incidence estimate. graphics and survival of the study population were similar to all ALL patients in CR1 or CR2 registered with the CIBMTR during the study period. Patients who received T cell depleted grafts, methotrexate without cyclosporin A as graft-versus-host disease (GVHD) prophylaxis, nonfractionated TBI, or TBI fractionated doses 12 or 14.5 Gy were excluded from the analysis. Cy doses ranged from 100 to 130 mg/kg, and etoposide doses ranged from 40 to 60 mg/kg. Five hundred two patients met the on-study criteria: 217 patients received Cy-TBI 13 Gy, 81 patients received Cy-TBI 13 Gy, 53 patients received etoposide-tbi 13 Gy, and 151 received etoposide-tbi 13 Gy. COH did not routinely report cases to the CIBMTR during the years 1989 to However, COH provided the necessary data to the CIBMTR for 75 patients who received etoposide-tbi 13 Gy as their conditioning regimen during this period. Eligible cases came from 111 reporting teams. The median follow-up of survivors was 67 months (range, months) for Cy-TBI 13 Gy patients, 59 months (range, months) for Cy-TBI 13 Gy patients, 52 months (range, months) for etoposide-tbi 13 Gy patients, and 65 months (range, months) for etoposide-tbi 13 Gy patients. BB&MT End Points Primary end points were TRM, clinical leukemia relapse (hematologic and extramedullary), LFS, and overall survival. TRM was defined as death during a continuous CR. Relapse was defined as clinical or hematologic leukemia recurrence. For analyses of LFS, failures were leukemia relapses or deaths from any cause; patients alive and in CR were censored at the time of last follow-up. For analyses of overall survival, failure was death from any cause; surviving patients were censored at the date of last contact. Statistical Analysis Patient-, disease-, and transplant-related variables for patients receiving Cy-TBI 13 Gy, Cy-TBI 13 Gy, etoposide-tbi 13 Gy, and etoposide-tbi 13 Gy for the conditioning regimen were compared by using the 2 statistic for categorical variables and the Kruskal-Wallis test for continuous variables. Univariate probabilities of LFS and survival were calculated by using the Kaplan-Meier estimator; the log-rank test was used for univariate comparisons. Probabilities of TRM and leukemia relapse were calculated by using cumulative incidence curves to accommodate competing risks [16]. Assessment of potentially con- 443

7 D. I. Marks et al. Table 4. Univariate Analyses of Transplantation Outcomes among Patients Receiving HLA-Identical Sibling Transplants for ALL in Second Complete Remission by Pretransplant Conditioning Regimen Cy-TBI <13 Gy Cy-TBI >13 Gy Etoposide-TBI <13 Gy Etoposide-TBI >13 Gy Outcome Event Prob (95% CI) Prob (95% CI) Prob (95% CI) Prob (95% CI) Pulmonary toxicity, n (%) (33) (37) 29 6 (21) (35) Grade II-IV acute GVHD at 100 d (24-83) (27-58) (19-54) (12-39) Chronic GVHD At 1 y 22 (14-32) 28 (14-44) 31 (14-51) 37 (21-54) At 3 y 25 (16-35) 32 (17-50) 37 (18-58) 37 (21-54) At 5 y 25 (16-35) 32 (17-50) 37 (18-58) 37 (21-54) 100-d mortality (11-26) (7-29) (2-24) (7-29) TRM At 1 y 18 (11-26) 20 (10-34) 10 (2-23) 9 (2-19) At 3 y 18 (11-26) 20 (10-34) 10 (2-23) 9 (2-19) At 5 y 20 (13-29) 20 (10-34) 10 (2-23) 12 (4-24) Relapse At 1 y 30 (21-40) 11 (4-22) 21 (8-38) 29 (16-44) At 3 y 42 (31-53) 14 (5-26) 24 (10-42) 36 (22-52) At 5 y 46 (34-57) 17 (7-30) 33 (16-54) 36 (22-52) LFS At 1 y 51 (41-61) 67 (53-80) 68 (50-84) 60 (46-74) At 3 y 40 (30-50) 65 (50-78) 64 (46-81) 54 (39-69) At 5 y 33 (24-43) 62 (46-76) 54 (35-73) 50 (35-65) Overall survival At 1 y 64 (55-74) 77 (63-88) 75 (58-89) 67 (53-80) At 3 y 48 (38-58) 72 (58-84) 71 (53-86) 53 (39-68) At 5 y 40 (30-51) 72 (58-84) 56 (36-75) 50 (35-65) Prob indicates probability; CI, confidence interval;, evaluable; LFS, leukemia-free survival; GVHD, graft-versus-host disease; TRM, transplant-related mortality; Cy, cyclophosphamide; TBI, total body irradiation; ALL, acute lymphoblastic leukemia. Probabilities of 100-day mortality, leukemia-free survival, and overall survival were calculated by using the Kaplan-Meier product-limit estimate. Acute GVHD, chronic GVHD, relapse, and TRM were calculated by using the cumulative incidence estimate. founding risk factors for outcomes of interest was performed by using Cox proportional hazards regression [17]. Variables considered are listed in Table 1. First, we compared the likelihood from a model stratified on conditioning regimen to a model with different risk coefficients for each conditioning regimen. The likelihood ratio test constructed from these models determined whether there was an interaction between the conditioning regimen and the factor being examined. When the likelihood ratio test was significant, an interaction term was added to the model. We found an interaction between conditioning regimen drug and TBI dose, so final models included main effect terms reflecting the conditioning regimen according to levels of TBI dose. The median TBI dose (13 Gy) was used as the cutoff point for high versus low dose. Additionally, the Martingale residuals plot was used to verify the validity of this cutoff point [17]. However, for relapse, LFS, and survival, this interaction was restricted to patients receiving Cy; consequently, patients receiving etoposide were considered as a single group. We also found an interaction between conditioning regimen and disease status before transplantation for all outcomes except TRM; consequently, analyses of relapse, LFS, and survival considered conditioning regimen effects separately for transplantations performed in first and second CR. After determining interaction terms, we next tested for proportional hazards for each factor in the Cox model by using time-dependent covariates. When this indicated differential effects over time (nonproportional hazards), models were constructed by breaking the posttransplantation course into 2 time periods. The maximized partial likelihood method was used to find the most appropriate breakpoint. After this modeling of time varying effects, the final multivariate model was built by using a forward stepwise model selection approach. Each model contained the main effect for the conditioning regimen (Cy-TBI 13 Gy versus Cy-TBI 13 Gy versus etoposide-tbi 13 Gy versus etoposide-tbi 13 Gy). Factors that were significant at a 5% level were kept in the final model. Because, in these models, TBI dose did not seem to affect relapse, LFS, or survival when combined with etoposide, additional analyses of these outcomes were also performed that considered 3 groups: Cy-TBI 13 Gy versus Cy-TBI 13 Gy versus etoposide-tbi. To test for possible center effects, a random effects score test developed by Andersen et al. [18] was used. The test is performed on the final model for the study and tests the hypothesis that there is no center effect against the hypothesis of a random center effect. We found no 444

8 Comparison of Conditioning Regimens for ALL Sibling Allografts Figure 1. Cumulative incidence of TRM after HLA-identical sibling transplantations for ALL in first (A) or second (B) complete remission, according to the pretransplantation conditioning regimen (pointwise P value at 5 years for CR1 patients: etoposide-tbi 13 Gy versus Cy-TBI 13 Gy, P.49; etoposide-tbi 13 Gy versus Cy-TBI 13 Gy, P.17; etoposide-tbi 13 Gy versus etoposide-tbi 13 Gy, P.14; Cy-TBI 13 Gy versus Cy-TBI 13 Gy, P.40; Cy-TBI 13 Gy versus etoposide-tbi 13 Gy, P.33; Cy-TBI 13 Gy versus etoposide-tbi 13 Gy, P.81; pointwise P value at 5 years for CR2 patients: etoposide-tbi 13 Gy versus Cy-TBI 13 Gy, P.21; etoposide-tbi 13 Gy versus Cy-TBI 13 Gy, P.14; etoposide-tbi 13 Gy versus etoposide-tbi 13 Gy, P.79; Cy-TBI 13 Gy versus Cy-TBI 13 Gy, P.98; Cy-TBI 13 Gy versus etoposide-tbi 13 Gy, P.30; Cy-TBI 13 Gy versus etoposide-tbi 13 Gy, P.22). Vp16 indicates etoposide. evidence of a correlation between center and any of the outcomes. All P values are 2 sided. RESULTS Patients BB&MT The patients in the 4 conditioning groups were compared with respect to demographic factors and other prognostic variables known to influence outcome (Table 2). The proportion of patients over the age of 30 years was higher in the etoposide-tbi 13 Gy group. Adverse cytogenetic abnormalities [hypodiploid, t(9;22), t(4;11), and t(8;14)] were also slightly more common in this group. Additionally, the etoposide-tbi 13 Gy group had a smaller proportion of transplantations for which both donor and recipient were cytomegalovirus seronegative. More than half of the etoposide-tbi 13 Gy group received corticosteroids in addition to standard cyclosporine and methotrexate GVHD prophylaxis, whereas this method of GVHD prophylaxis was uncommonly used in the other 3 conditioning groups. This made it difficult to separate the effects of differing GVHD prophylaxis from conditioning regimens on outcome. The etoposide-tbi 13 Gy group had a higher percentage of patients who took 8 weeks to achieve remission (37% versus 17% in the other 3 groups combined). 445

9 D. I. Marks et al. Figure 2. Cumulative incidence of relapse after HLA-identical sibling transplantations for ALL in first (A) or second (B) complete remission, according to the pretransplantation conditioning regimen (pointwise P value at 5 years for CR1 patients: etoposide-tbi versus Cy-TBI 13 Gy, P.23; etoposide-tbi versus Cy-TBI 13 Gy, P.78; Cy-TBI 13 Gy versus Cy-TBI 13 Gy, P.60; pointwise P value at 5 years for CR2 patients: etoposide-tbi versus Cy-TBI 13 Gy, P.22; etoposide-tbi versus Cy-TBI 13 Gy, P.033; Cy-TBI 13 Gy versus Cy-TBI 13 Gy, P.001). Vp16 indicates etoposide. Univariate Analyses The major transplantation outcomes in the 4 conditioning groups are compared in Tables 3 and 4. These data require cautious interpretation because of the differences in patient characteristics described previously. There were no major differences in pulmonary toxicity or 100-day mortality. Among patients who underwent transplantation in CR1, chronic GVHD was more common at 5 years in the etoposide-tbi 13 Gy group. This difference occurred despite no differences in the incidences of grade II to IV or III/IV acute GVHD (data not shown). TRM was also somewhat lower in the etoposide- TBI 13 Gy groups (Figure 1). The highest risks of relapse (Figure 2) and the lowest probabilities of LFS and overall survival were in the Cy-TBI 13 Gy groups. Multivariate Analyses Transplant-related mortality. There was not a statistically significant difference in TRM according to conditioning regimen (overall P.05), although, among patients receiving 13 Gy TBI, there was a trend for lower TRM with etoposide versus with Cy (Table 5). Other factors significantly associated with higher TRM in multivariate analysis were age 20 years and receiving a PBSC rather than a bone marrow graft. Relapse. Among patients who underwent transplantation in CR1, there was not a significant association between conditioning regimen and relapse risk (overall P.05), although there was a trend toward lower relapse in patients receiving etoposide compared with those receiving Cy-TBI 13 Gy (Table 6). Among patients who underwent transplantation in 446

10 Comparison of Conditioning Regimens for ALL Sibling Allografts Table 5. Multivariate Analysis Comparing Transplant-Related Mortality after Etoposide-TBI Versus Cy-TBI Conditioning Regimen for Acute Lymphoblastic Leukemia in First or Second Complete Remission Variable n Relative Risk (95% Confidence Interval) P Value Conditioning regimen (1) Cy-TBI, TBI dose <13 Gy * P overall.0725 (2) Cy-TBI, TBI dose >13 Gy ( ) P (3) Etoposide-TBI, TBI dose <13 Gy ( ) P (4) Etoposide-TBI, TBI dose >13 Gy ( ) P Other significant covariates Age (y) < * > ( ).0011 Graft type Bone marrow * Peripheral blood ( ).0332 Cy indicates cyclophosphamide; TBI, total body irradiation. *Reference group. Three degrees of freedom. Other pairwise comparisons: 4 versus 2: relative risk 0.42 ( ), P.012; 4 versus 3: relative risk 0.46 ( ), P.071. CR2, relapse risk differed by conditioning regimen. There was a modest (not statistically significant) decrease in relapse in the groups that received etoposide for conditioning. Among patients who received etoposide, there was no significant difference between those who received 13 and 13 Gy of TBI. Among patients who received Cy for conditioning, the risk of relapse was significantly lower among those who received 13 Gy of TBI (P.0016). Other factors associated with an increased risk of relapse in multivariate analysis were age 20 years and the presence of cytogenetic abnormalities. Leukemia-free survival. Among patients who underwent transplantation in CR1, there was little difference in the risk of treatment failure (inverse of LFS) by conditioning regimen (overall P value.05; Table 7; Figure 3). Among patients who underwent transplantation in CR2, the risk of treatment failure differed by conditioning regimen (overall P value.001). There was a lower treatment failure rate with Table 6. Multivariate Analysis Comparing Relapse after Etoposide-TBI Versus Cy-TBI Conditioning Regimen for Acute Lymphoblastic Leukemia in First or Second Complete Remission Variable n Relative Risk (95% Confidence Interval) P Value Conditioning regimen First CR status at transplantation (1) Cy-TBI, TBI dose <13 Gy * P overall.0632 (2) Cy-TBI, TBI dose >13 Gy ( ) P (3) Etoposide-TBI ( ) P Second CR status at transplantation (1) Cy-TBI, TBI dose <13 Gy * P overall.0045 (2) Cy-TBI, TBI dose >13 Gy ( ) P (3) Etoposide-TBI ( ) P Other significant covariates Age (y) < * > ( ).0131 Cytogenetics No abnormalities * P overall.0042 Hypodiploid, t(9;22), t(4;11); t(8;14) ( ).0044 Other abnormalities ( ).0076 Unknown ( ).5784 Cy indicates cyclophosphamide; TBI, total body irradiation; CR, complete remission. *Reference group. Three degrees of freedom. Other pairwise comparisons: first CR: 3 versus 2: relative risk (RR) 0.55 ( ), P.10; second CR: 3 versus 2: RR 2.28 ( ), P.04. Comparison of etoposide 13 Gy versus etoposide 13 Gy: first CR: RR 1.16 ( ), P.81; Second CR: RR 1.03 ( ), P.94. BB&MT 447

11 D. I. Marks et al. Table 7. Multivariate Analysis Comparing Leukemia-Free Survival after Etoposide-TBI Versus Cy-TBI Conditioning for Acute Lymphoblastic Leukemia in First or Second Complete Remission Variable n Relative Risk of Relapse or Death (95% Confidence Interval) P Value Conditioning regimen First CR status at transplantation (1) Cy-TBI, TBI dose <13 Gy * P overall.0758 (2) Cy-TBI, TBI dose >13 Gy ( ) P (3) Etoposide-TBI ( ) P Second CR status at transplantation (1) Cy-TBI, TBI dose <13 Gy * P overall.0011 (2) Cy-TBI, TBI dose >13 Gy ( ) P (3) Etoposide-TBI ( ) P.0065 Other significant covariates Age (y) < * > ( ).0003 Karnofsky status before transplantation >90% * P overall.0057 <90% ( ).0013 Unknown ( ).4375 Graft type Bone marrow * PBSC ( ).0163 Cy indicates cyclophosphamide; TBI, total body irradiation; CR, complete remission; PBSC, peripheral blood stem cells. *Reference group. Three degrees of freedom. Other pairwise comparisons: first CR: 3 versus 2: relative risk (RR) 0.56 ( ), P.03; second CR: 3 versus 2: RR 1.38 ( ), P.28. Comparison of etoposide 13 Gy versus etoposide 13 Gy: first CR: RR 0.64 ( ), P.23; second CR: RR 1.13 ( ), P.74. Two degrees of freedom. use of etoposide in comparison to Cy-TBI 13 Gy (P.007). Among patients receiving etoposide, there was no significant difference between those receiving 13 and 13 Gy of TBI. Among patients receiving Cy for conditioning, the treatment failure rate was significantly lower among those receiving 13 Gy of TBI (P.0013). Other factors associated with poorer LFS were age 20 years, a poor performance score before transplantation, and the use of PBSCs rather than bone marrow. Overall survival. Among patients who underwent transplantation in CR1, there was little difference in overall survival by conditioning regimen (overall P value.05; Table 8; Figure 4). Among those who underwent transplantation in CR2, survival differed by conditioning regimen (overall P value.001). There was higher survival in the group receiving etoposide compared with those receiving Cy-TBI 13 Gy (P.024). Among patients receiving etoposide, there was no significant difference between those receiving 13 and 13 Gy of TBI. Among those receiving Cy for conditioning, there was higher survival with 13 Gy of TBI (P.0005). Other factors associated with increased mortality in multivariate analysis were age 20 years, a poor performance score before transplantation, and the use of PBSCs rather than bone marrow. Other Factors Potentially Affecting Transplantation Outcomes The knowledge that most patients in the etoposide-tbi 13 Gy group underwent transplantation at the Stanford or COH institutions led us to perform an analysis of the center effect (see Patients and Methods ). There was a nonsignificant trend toward a center effect (P.07). We did not find any interaction between age at transplantation and conditioning regimen for any of the major outcomes. Patients younger and older than 20 years had similar results; there was insufficient power to detect differences among smaller age-defined groups. Chronic GVHD did not significantly affect relapse, LFS, or overall survival. Causes of Death The major single cause of death was recurrence of disease (Table 9). Overall, the causes of death were similar in the 4 groups except that recurrent disease was somewhat more common as a cause of death in the Cy-TBI 13 Gy group. GVHD, infection, interstitial pneumonia, and organ failure were the other major causes of treatment failure. 448

12 Comparison of Conditioning Regimens for ALL Sibling Allografts Figure 3. Adjusted probability (derived from multivariate regression models) of LFS after HLA-identical sibling transplantations for ALL in first (A) or second (B) complete remission, according to the pretransplantation conditioning regimen (pointwise P value at 5 years for CR1 patients: etoposide-tbi versus Cy-TBI 13 Gy, P.21; etoposide-tbi versus Cy-TBI 13 Gy, P.17; Cy-TBI 13 Gy versus Cy-TBI 13 Gy, P.59; pointwise P value at 5 years for CR2 patients: etoposide-tbi versus Cy-TBI 13 Gy, P.002; etoposide-tbi versus Cy-TBI 13 Gy, P.23; Cy-TBI 13 Gy versus Cy-TBI 13 Gy, P.001). Vp16 indicates etoposide. DISCUSSION There have been many attempts to intensify conditioning regimens to increase leukemia cell kill and improve outcomes after transplantation, but few have been conclusively shown to be successful [19]. Allogeneic stem cell transplantation for patients with leukemia, in part, exerts its curative effect via immune recognition of host leukemia cells by alloreactive donor T cells and other accessory cells. However, because the GVL effect in ALL seems to be less, the myeloablative conditioning regimen chosen is also relevant. In addition to providing immune suppression sufficient for engraftment, cytoreduction may be necessary to produce a minimal residual disease state so that the immunologic properties of the graft may be more effective. Overt ALL, or even ALL at the minimal residual disease level, is seldom eradicated by BB&MT allogeneic transplantation, probably because immune recognition is insufficiently rapid and powerful to overcome the rapid growth kinetics [20]. Chemoradiotherapeutic conditioning regimens may also influence GVHD (and, hence, GVL) by causing tissue damage and secondary cytokine release; these effects may differ between conditioning regimens. This article provides evidence that the conditioning regimen is of importance, particularly for patients who undergo transplantation after failure of initial chemotherapy. In 500 patients with ALL in first or second CR who received either Cy-TBI or etoposide- TBI conditioning, significant interactions between Cy, etoposide, and the TBI dose used were found. The pretransplantation conditioning regimen used most widely is Cy and 12 Gy of TBI. In fact, nearly three quarters of the patients in this registry-based 449

13 D. I. Marks et al. Table 8. Multivariate Analysis Comparing Survival after Etoposide-TBI Versus Cy-TBI Conditioning Regimen for Acute Lymphoblastic Leukemia in First or Second Complete Remission Variable n Relative Risk of Death (95% Confidence Interval) P Value Conditioning regimen First CR status at transplantation (1) Cy-TBI, TBI dose <13 Gy * P overall.2534 (2) Cy-TBI, TBI dose >13 Gy ( ) P (3) Etoposide-TBI ( ) P Second CR status at transplantation (1) Cy-TBI, TBI dose <13 Gy * P overall.0010 (2) Cy-TBI, TBI dose >13 Gy ( ) P (3) Etoposide-TBI ( ) P Other significant covariates Age (y) < * > ( ) <.0001 Karnofsky status before transplantation >90% * P overall.0019 <90% ( ).0005 Unknown ( ).9416 Graft type Bone marrow * PBSC ( ).0108 Cy indicates cyclophosphamide; TBI, total body irradiation; CR, complete remission; PBSC, peripheral blood stem cells. *Reference group. Three degrees of freedom. Other pairwise comparisons: first CR: 3 versus 2: relative risk (RR) 0.70 ( ), P.22; second CR: 3 versus 2: RR 1.86 ( ), P.07. Comparison of etoposide 13 Gy versus etoposide 13 Gy: first CR: RR 0.71 ( ), P.36; second CR: RR 1.37 ( ), P.41. Two degrees of freedom. study received 13 Gy of TBI. Although differences in outcome associated with the conditioning regimen were minor in patients who underwent transplantation in CR1, this standard Cy-TBI regimen was associated with significantly poorer outcomes (relapse, survival, and LFS) in patients who underwent transplantation in CR2. The reduction in survival (10%-32%) compared with the other 3 regimens was of a magnitude that would be clinically important and resulted mainly from an increased probability of relapse. It was true for children and adults and for patients with and without prior extramedullary disease. Excluding Philadelphia-positive cases also did not affect the analysis of outcomes. Although imatinib, which was not available during this study period, may produce more durable remissions in Philadelphia-positive ALL and affect the outcome of a subsequent transplantation, it is uncertain how it might modify the effect of a conditioning regimen. The findings concerning TBI doses are different from those in the randomized Seattle study in patients who had allografts for acute myelogenous leukemia. In that study, Gy was not superior to 12 Gy in terms of survival [9]. It should be noted that the dose ranges we examined were different, and, of course, the disease we examined was ALL. Additionally, more of the patients in our study were younger than 18 years; adverse effects of TBI may be less in this age group. A surprising finding of the study is that the TRM was lower in the group that received etoposide and TBI 13 Gy on univariate analysis, although on multivariate analysis this was only a trend (P.06). A possible explanation is that this is due to a center effect: ie, the TRM may be lower at the COH and Stanford centers. However, analysis did not find evidence for a statistically significant center effect. We are not able to explain this effect on TRM. This borderline effect on TRM may have affected overall survival, but it is also worth noting that this regimen resulted in significantly less relapse in CR1 patients. There is concern, particularly among pediatricians, about the long-term neuropsychological and growth/endocrine effects of using TBI. Two regimens recently introduced for pediatric patients with ALL seem to be associated with low relapse rates even in heavily pretreated patients with multiply relapsed disease. The first, pioneered by the Gluckman group in Paris, combines TBI with melphalan and cytarabine (the TAM regimen) [21]. However, despite low relapse rates, overall survival with this regimen does not seem to be better than with other preparative regimens. The second regimen combines TBI with high- 450

14 Comparison of Conditioning Regimens for ALL Sibling Allografts Figure 4. Adjusted probability (derived for multivariate regression models) of overall survival after HLA-identical sibling transplantations for ALL in first (A) or second (B) complete remission, according to the pretransplantation conditioning regimen (pointwise P value at 5 years for CR1 patients: etoposide-tbi versus Cy-TBI 13 Gy, P.18; etoposide-tbi versus Cy-TBI 13 Gy, P.47; Cy-TBI 13 Gy versus Cy-TBI 13 Gy, P.87; pointwise P value at 5 years for CR2 patients: etoposide-tbi versus Cy-TBI 13 Gy, P.029; etoposide-tbi versus Cy-TBI 13 Gy, P.012; Cy-TBI 13 Gy versus Cy-TBI 13 Gy, P.001). Vp16 indicates etoposide. BB&MT dose cytarabine (36 gm/m 2 ). Again, very low relapse rates are reported, but regimen-related toxicity limits long term survival [22]. Although TBI-containing regimens are the gold standard, many pediatric transplant centers now use a busulfan-cy regimen to avoid radiation-related adverse effects. Engraftment rates are similar, but relapse rates are higher, and TRM is no less than with other regimens. A recently concluded randomized study comparing a busulfan-cy regimen with a TBI-containing regimen showed inferior survival results with the non-tbi approach [23]. Another finding of this study was lower leukemiafree and overall survival in the small number of patients who received a PBSC graft. PBSC transplants were associated with higher TRM but a similar relapse risk. Reasons for these findings are not apparent, and they may be due to chance alone. Beneficial effects of using PBSC are more easily seen in patients with advanced leukemia, but it was surprising that this group of patients had a worse outcome in our series [24]. This study has limitations. TBI is given in many different ways with different fractionation, dose rates, use of compensators, and variable lung shielding. The CIBMTR did not have these specific data available for all patients. Thus, there may be considerable variation in the effective biological dose administered. The method of administering TBI may influence outcome. For example, both Stanford and COH administer 13.2 Gy in 11 fractions, shielding the lungs from half the total dose but administering electron beam radiation to the bony structures of the thorax [25]. This study also lacks data about the precise doses of GVHD prophylaxis received. It is possible that patients in the 451

15 D. I. Marks et al. Table 9. Causes of Death among Patients Receiving HLA-Identical Sibling Transplants for ALL in First or Second Complete Remission by Pretransplantation Conditioning Regimen Cause of Death Cy-TBI <13 Gy Cy-TBI >13 Gy Etoposide-TBI <13 Gy Etoposide-TBI >13 Gy patients Recurrence disease 63 (56) 9 (33) 10 (50) 23 (37) GVHD 11 (10) 3 (11) 1 (5) 5 (8) Graft failure (2) Infection 11 (10) 2 (7) 3 (15) 4 (6) Interstitial pneumonia 7 (6) 2 (7) 1 (5) 5 (8) ARDS 3 (2) 2 (7) 2 (10) 3 (5) Organ failure 4 (4) 2 (7) 2 (10) 7 (12) Hemorrhage 3 (3) 4 (14) 1 (5) 4 (6) Other/missing 10 (9) 4 (14) 0 10 (16) GVHD indicates graft-versus-host disease; ARDS, adult respiratory distress syndrome;, evaluable. high-dose TBI groups received less methotrexate (because of mucositis) and that this may be the mechanism of the protection against relapse. However, in this study, acute and chronic GVHD were not associated with a reduction in relapse probability. This study found less treatment failure and mortality with etoposide, regardless of TBI dose, compared with the standard Cy-TBI 13 Gy TBI regimen, accounted for by modest, not individually statistically significant, reductions in both TRM and relapse. Although only a randomized study would definitely resolve this issue, such a study would be difficult, given the relatively small numbers of patients who undergo transplantation for ALL annually and the need to standardize TBI techniques across many centers. After the encouraging results achieved by the Stanford and COH groups, the large UK Medical Research Council/US Eastern Cooperative Oncology Group study of adult ALL adopted etoposide-tbi as its standard conditioning regimen. There are other important differences between Cy and etoposide. The dose-limiting toxicity of Cy is cardiotoxicity [26]. For etoposide, the major limiting toxicities are stomatitis and hepatotoxicity [10]. However, this study did not find evidence of excess TRM after etoposide-tbi. The incidence and type of secondary malignancies to be expected after Cy-TBI have been described in a number of studies, whereas there are no large-scale data available for etoposide-tbi [27]. Secondary leukemias involving 11q23 chromosomal abnormalities may be a significant long-term complication of etoposide. In summary, this study suggests that the conditioning regimen of Cy and 13 Gy of TBI results in inferior survival compared with the other 3 conditioning regimens studied in patients receiving HLA-identical sibling allografts for ALL in CR2. We conclude that among patients receiving HLA-identical sibling allografts for ALL in CR2, there seems to be an advantage to substituting etoposide for Cy in the conditioning regimen or, when using Cy, to increasing the TBI dose to 13 Gy. However, the limitations of this study are such that we believe that a randomized trial is warranted to address the issue of which is the superior conditioning regimen. ACKNOWLEDGMENTS Supported by Public Health Service grant no. U24- CA76518 from the National Cancer Institute, the National Institute of Allergy and Infectious Diseases, and the National Heart, Lung and Blood Institute; Office of Naval Research; Health Resources Services Administration (Department of Health and Human Services); and grants from AABB, Aetna; AIG Medical Excess; American Red Cross; Amgen, Inc.; an anonymous donation to the Medical College of Wisconsin; AnorMED, Inc.; Berlex Laboratories, Inc.; Biogen IDEC, Inc.; Blue Cross and Blue Shield Association; BRT Laboratories, Inc.; Celgene Corp.; Cell Therapeutics, Inc.; CelMed Biosciences; Cubist Pharmaceuticals; Dynal Biotech, LLC; Edwards Lifesciences RMI; Endo Pharmaceuticals, Inc.; Enzon Pharmaceuticals, Inc.; ESP Pharma; Fujisawa Healthcare, Inc.; Gambro BCT, Inc.; Genzyme Corporation; GlaxoSmithKline, Inc.; Histogenetics, Inc.; Human Genome Sciences; ILEX Oncology, Inc.; Kirin Brewery Company; Ligand Pharmaceuticals, Inc.; Merck & Company; Millennium Pharmaceuticals; Miller Pharmacal Group; Milliman USA, Inc.; Miltenyi Biotec; National Center for Biotechnology Information; National Leukemia Research Association; National Marrow Donor Program; NeoRx Corporation; Novartis Pharmaceuticals, Inc.; Novo Nordisk Pharmaceuticals; Ortho Biotech, Inc.; Osiris Therapeutics, Inc.; Pall Medical; Pfizer, Inc.; Pharmion Corp.; QOL Medical; Roche Laboratories; StemCyte, Inc.; Stemco Biomedical; StemSoft Software, Inc.; SuperGen, Inc.; Sysmex; The Marrow Foundation; THERAKOS, a Johnson & Johnson Co.; University of Colorado Cord Blood Bank; Valeant Pharmaceuticals; ViaCell, Inc.; ViraCor Laboratories; WB Saunders Mosby Churchill; and Wellpoint Health Network. 452