Abstract 1976 Multiple single center and registry reports have documented the critical impact of donor-recipient HLA match on engraftment, transplant-related mortality (TRM) and survival after umbilical cord blood (UCB) transplantation. However, nearly all reports have only considered HLA A and B at antigen level and HLA DRB1 at allele level typing without consideration of HLA C or DQ. Therefore, we retrospectively performed allele level HLA typing for HLA-A, B, C, DRB1, DQB1 for UCB donor-recipient pairs in order to assess the importance of high resolution HLA typing on transplant outcomes. After 2002, most patients received a dUCB transplant in order to achieve the desired cell doses of ≥3, ≥4 and ≥5 × 10e7 NC/kg for grafts that were HLA 6/6, 5/6 and 4/6 matched by original typing resolution, respectively. Therefore, the analysis was limited to 275 recipients of dUCBT for hematological malignancy and whom DNA from both units was available. The effect of HLA match was based on the HLA type of the predominant long term engrafting unit. The median recipient age and weight was 44 years (range, 0.6–69) and 76.9 kg (range, 7.1–148), respectively. Conditioning was myeloablative (40%) consisting of cyclophosphamide (CY) 120 mg/kg, fludarabine (FLU) 75 mg/m2 and total body irradiation (TBI) 1320 cGy, or non-myeloablative (60%) consisting of CY 50 mg/kg, FLU 200 mg/m2, TBI 200 cGy with 95% receiving cyclosporine A (CsA) and mycophenolate mofetil (MMF) immunosuppression. Patients had acute leukemia (62%), standard risk disease (62%), cytomegalovirus seropositive (59%), and received at least one UCB unit that was sex mismatched to the recipient (78%). Results reported are based on the long-term predominant UCB unit. Notably, survival was not adversely affected by HLA mismatch. The probability of survival at 5 years was 46% (95%CI, 33–58%), 47% (95%CI, 38–54%) and 29% (95%CI, 13–47%) in patients engrafting with a 3–5/10, 6–8/10 and 9–10/10 HLA-matched UCB grafts, respectively (p=.47). In multivariable analysis after adjusting for disease risk, CMV serostatus, and KPS, there was similar risk of overall mortality for all groups regardless of HLA matching level. All other transplant outcomes including the incidence of acute and chronic GVHD were similar for all HLA-matching groups (data not shown). In the subset with acute leukemia (n=174), however, greater HLA mismatch was associated with a significantly lower risk of relapse without a deleterious effect on risk of TRM, resulting in a benefit in LFS (inverse of treatment failure) as shown below. Together these data indicate that UCB units with greater HLA mismatch may confer greater GVL effect without greater TRM compared to HLA better-matched UCB grafts. These results suggest importance of evaluating allele level HLA typing in the setting of dUCB transplantation. If confirmed, these results could have major implications not only on graft selection (ie avoidance of HLA matched units), but also the target size of the international UCB banking inventory. Disclosures: No relevant conflicts of interest to declare.

Abstract Abstract 225 Double umbilical cord blood (dUCB) transplantation (dUCBT) is a strategy to overcome dose limitation in adult recipients. It is established that after dUCBT, one unit will predominate by day +100 after transplant in >95%. While in some studies order of infusion has been associated with unit predominance, this has not been reproduced in an analysis at our center. However, significant differences in UCB infusion between these two analysis were present. In particular, at our center, unit order of infusion is random and the second infusion is within minutes of the first, while this prior study separated infusion time by 6 hours. Between 2001 and 2009, 262 patients with hematologic malignancies underwent a dUCBT and engrafted. Of these, 233 were >18 years of age with 39% conditioned with cyclophosphamide (CY) 60 mg/kg, fludarabine (FLU) 75 mg/m2 and total body irradiation (TBI) 1320–1375 cGy and 61% with CY 50 mg/kg, FLU 200 mg/m2 and TBI 200 cGy with 1/3 also receiving antithymocyte globulin (ATG); 100% received cyclosporine and mycophenolate mofetil posttransplant immunosuppression. Median recipient weight was 78 kg and median follow-up was 2.7 years (0.5-7.2). The following factors were considered in the logistic regression model: total nucleated cell (TNC), CD34+ and CD3+ cell and colony forming units-granulocyte macrophage (CFU-GM) doses, HLA match, sex and ABO-match, CXCR4 expression on CD34+ cells, order of infusion and cell viability. Cell viability, infused TNC, CD34+ and CD3+ cell and CFU-GM doses were remarkably similar between the predominating and non-predominating unit. By day 21, the predominating unit (i.e., representing >70% of hematopoiesis) was achieved in 73 of 90 (81%) patients after MA conditioning and in 88 of 145 (61%) patients after reduced dose conditioning (p<0.01). Subsequently, predominant unit chimerism in the bone marrow between MA and NMA was similar by day 100 (95% vs. 97%, p=0.35), day 180 (97% vs. 100%, p=0.3), day 365 (97% vs. 98%, p=0.84) and day 730 (94% vs. 93%, p=0.81). Notably, CD3+ cell dose and HLA were strongly associated with unit predominance. In the MA setting, CD3+ cell dose was the most significant factor that predicted unit predominance [OR 4.4 (95% CI, 1.8–10.6, p<0.01)]; while CD3+ cell dose [OR 2.1 (95%CI, 1.0–4.2, p=0.05)] and HLA-match [OR 3.4 (95%CI 1.0–11.4, p=0.05)] were independent predictors in the reduced intensity setting. In summary, immunological graft-graft interactions are likely responsible for unit predominance. While the combined CD34 dose and CFU-GM dose from the two UCB units are critical for rate of neutrophil recovery (data previously reported), CD3 dose and HLA match after reduced intensity conditioning are important in determining which unit will ultimately predominate. These findings have potential implications in the algorithm of graft selection. Disclosures: No relevant conflicts of interest to declare.

Abstract Abstract 3731 Both T cell and natural killer (NK) cell reconstitution have been shown to affect clinical outcomes after hematopoietic stem cell transplantation (HSCT). Killer immunoglobulin-like receptor (KIR) interactions between alloreactive NK cells and their targets can prevent relapse, but may be dysregulated, especially after T cell replete HSCT. T cell recovery is also affected by the stem cell source and T cell content of the graft. To better understand the effects of various NK and T cell subsets we evaluated lymphocyte recovery in 304 adult patients who received either UCB (n=116), Sib (n=84) or Auto (n=94) HSCT for hematologic malignancies between 2003 and 2010 at the University of Minnesota. Peripheral blood mononuclear cells obtained at 3 months after HSCT were stained with CD56, CD3, CD4, CD8, and a cocktail of anti-NK cell KIR antibodies to determine the relative percentage of lymphocyte subsets by flow cytometry. The absolute lymphocyte count (ALC) was measured and used to calculate the absolute (Abs) number of T and NK cells and their subsets. ALC recovery at 3 months was similar among groups (UCB: 901.9 ± 74.5, Sib 890.2 ± 73.0 and Auto 1076.7± 69.4 cells/ul). Abs NK cells were highest in the UCB cohort (375.4 ± 24.9) vs. Sib (183.8 ± 15.4; p<0.0001) or Auto (160.7 ± 11.0; p<0.0001), as were the CD56bright and KIR+ subsets (data not shown). In contrast, Abs T cell recovery was lowest in the UCB group (300.8 ± 39.6) vs. Sib (578.5 ± 57.9; p<0.0001) or Auto (737.3 ± 60.4; p<0.0001). Accordingly, the lowest Abs CD4 count was in the UCB group (158.8 ± 14.7) vs. Sib (272.5 ± 23.5; p<0.0001) or Auto (223.6 ± 20.2; p=0.01), with a similar pattern observed for Abs CD8 counts. We then examined the effect of lymphocyte recovery on clinical outcomes. Multivariate models were constructed for each transplant group with relevant covariates (risk status, conditioning, sex, age, number of UCB units, CMV status, HLA matching (4/6, 5/6, or 6/6), and ABO matching). The most significant effect of lymphocyte recovery on outcomes was observed specifically in the UCB group, where higher ALC was associated with improved OS with a hazard ratio (HR) of 0.86 (95% CI 0.78–0.95) for each unit increase in ALC of 100 cells/ul (p <0.01). A similar trend was observed in Sib recipients but not in the Auto group. Specifically, increases in Abs T cells (HR 0.75 [95% CI 0.58–0.98]; p=0.034), Abs CD4 count (HR 0.63 [95% CI 0.42–0.95]; p=0.03), Abs CD8 count (HR 0.31 [95% CI 0.13–0.73]; p=0.01) and to a lesser extent Abs NK cells (HR 0.85 [95% CI 0.71–1.02]; p=0.085) were associated with improved OS. In the Sib cohort, higher Abs CD4 count was associated with improved OS (HR 0.43 [95% CI 0.20–0.92]; p=0.03) and decreased relapse (HR 0.37 [95% CI 0.37–1.00]; p=0.02), with no other factor having a significant impact. In the Auto group, only Abs NK (HR 0.40 [95% CI 0.16–0.99]; p=0.05) and to a lesser extent Abs KIR+ NK cells (HR 0.17 [95% CI 0.02–1.36]; p=0.09) were associated with improved OS but no other outcomes. The effect of Abs CD4 count on OS in all groups is shown in Figure 1 with survival stratified by quartiles. Figure 1: Figure 1:. In summary, rapid recovery of T cells predicts significantly better survival in patients undergoing UCB and Sib HSCT, while the NK cell effects are less pronounced. In contrast, NK cell effects predominate after Auto HCT. This suggests that more rapid T cell recovery is critical for survival and that defects in NK cell education after allogeneic HSCT may affect their function such that just increasing numbers may not be sufficient for clinical benefit. Appropriate modifications to immune suppression or the use of agents that promote T cell (IL-7) and/or NK cell (IL-15) function and survival may positively influence survival outcomes. Disclosures: No relevant conflicts of interest to declare.

Chronic myelogenous leukemia (CML) was primarily treated with HSCT until imatinib mesilate was shown to be effective and safe for patients with early chronic phase CML. However, patients who fail imatinib therapy due to disease progression or drug intolerance still require HSCT. Umbilical cord blood (UCB) has been an increasingly used source of hematopoietic stem cells for transplantation (HSCT) of patients with hematologic malignancies who lack a suitable sibling donor. We report here on 20 adult patients who underwent UCB transplantation (UCBT) for Ph+ CML at the University of Minnesota between 1998 and 2005. Patient received myeloablative (MA, n=12) or nonmyeloablative (NMA, n=8) conditioning, The median age was 46 y (r: 18-58), 12 (60%) were male, and median weight was 78 kg (r: 57-103), and 13 (65%) were CMV positive. The MA conditioning was Bu/Cy (n=2) Cy/TBI ± ATG (n=4), or Cy/Fludarabine(Flu)/TBI (n=6). The NMA conditioning was Bu/Flu/TBI (n=2) or Cy/Flu/TBI± ATG (n=6). Posttransplantation immunosuppression was CsA alone (n=1), CsA/methylprednisolone(MP) (n=5), or CsA/MMF (n=14). Eleven patients (55%) receive a double UCB graft. The highest HLA disparity of UCB units was 4/6 (n=13), 5/6 (n=5), and 6/6 (n=2). Six patients (30%) were in first chronic phase (CP1) and 14 (70%) were in accelerated phase (AP) CML. The median TNC dose infused was 2.9 × 107/kg (r:1.2-5.3) and median CD34 dose infused was 4.8 × 105/kg (r: 0.7-12.7). The median time from diagnosis to transplant was 24.5 months (r: 6.7-118.8), and the median follow-up of surviving patients was 2.9 yrs (r: 0.7-.7.0). There were no failures of neutrophil engraftment. In the MA setting median time to neutrophil engraftment was 21d (r:13-33), grade II-IV acute GVHD was 58% (95%CI, 28-88%), 1-yr transplant related mortality (TRM) 41% (95%CI, 13-69), 2-yr relapse rate 10% (95%CI, 0-26), and overall survival 58% (95%CI, 30-86). In the NMA setting median time to neutrophil engraftment was 13d (r:5-32), grade II-IV acute GVHD was 63% (95%CI, 28-98%), 1-yr transplant related mortality (TRM) 38% (95%CI, 6-70), 2-yr relapse rate 13% (95%CI, 0-34), and overall survival 50% (95%CI, 15-85). There was no statistically significant difference between MA and NMA conditioning regimens on all outcomes. In this report we show that UCB appears to be a safe and effective HSC for transplantation of patients with CML.

The clinical importance of HLA class II gene disparity in unrelated stem cell transplantation is not entirely known. The impact was evaluated of matching donors and recipients for HLA-DR, HLA-DQ, and HLA-DP genes on clinical outcome after stem cell transplantation for chronic myeloid leukemia (CML) performed between 1988 and 1997. HLA-DRB1, -DQA1, -DQB1, -DPA1, and -DPB1 alleles were identified in 831 transplant pairs using a combination of sequence-specific oligonucleotide probes, sequence-specific priming, and sequencing methods. Among the 831 pairs, 696 (84%) were HLA-A and -B serologically matched; of these, 565 (81%) were also matched for HLA-DRB1. HLA-DRB1 matching correlated with significantly improved survival (relative risk [RR], 1.29 [95% confidence interval (CI), 1.02-1.64; P = .04]) independently of HLA-DQA1 or HLA-DQB1 (RR, 1.01 [95% CI, 0.81-1.26; P = .94]) and HLA-DPA1 or HLA-DPB1 (RR, 1.11 [95% CI, 0.84-1.48;P = .46]). Single-locus HLA-DQ or HLA-DP disparity was not associated with significantly poorer survival. For patients who underwent transplantation in the first chronic phase (CP) from HLA-A, B matched donors, the presence of DRB1 allele mismatching was independently associated with increased incidence of grades III-IV acute graft-versus-host disease (GVHD). No significant associations of class II allele mismatching with risk for delayed engraftment or chronic GVHD disease were detected. This study clearly demonstrates the importance of precise matching of HLA-DRB1 alleles for successful transplantation. Furthermore, a good-risk population of patients whose transplantations were performed in the first CP of disease from HLA-A, B, DRB1 matched unrelated donors can be shown to have superior survival.

Glucocorticoids remain the standard approach to initial systemic management of acute graft-versus-host disease (aGVHD). For patients refractory to steroids, antithymocyte globulin (ATG) is frequently used as salvage therapy. We decided to test whether the combination of corticosteroids and equine ATG would improve the outcome of initial management of aGVHD, especially in high-risk patients such as recipients of unrelated donor (URD) transplants. One hundred patients with grade II. to TV aGVHD having undergone a related or URD marrow transplant were enrolled in the study. Of the patients, 46 were randomly assigned to therapy with prednisone (60 mg/m(2) per day x 7 days) and 50 received ATG/prednisone (15 mg/kg ATG bid plus 20 mg/m(2) prednisone bid x 5 days, each followed by an 8-week prednisone taper). An intent-to-treat analysis of the overall response at day 42 revealed equivalent complete plus partial response rates of 76% in both the prednisone and ATG/prednisone therapy groups (P > .80). In univariate analysis, patient age, donor type, site of involvement, or aGVHD stage did not influence overall response to therapy (all P > .2). When treatment arms were studied separately, no single clinical feature predicted outcome in either group. Complications were more frequent in the ATG/prednisone arm; patients experienced more infections with cytomegalovirus (44% versus 22%; P = .02) and more frequent pneumonitis, both infectious and noninfectious (50% versus 24%; P < .01). Epstein-Barr virus lymphoproliferative disease was uncommon (4 cases) and comparable in both arms (P = .35). There was no significant difference in survival at day 100, 6 months, and 2 years between the 2 treatment arms. The more intensive immunosuppressive combination of ATG/prednisone failed to improve control of aGVHD and may have affected survival by causing more infectious complications. Combination therapy with ATG should thus be reserved as second-line therapy in the management of aGVHD.

Twenty-six cases of B cell lymphoproliferative disorder (BLPD) were identified among 2395 patients following hematopoietic stem cell transplants (HSCT) for which an overall incidence of BLPD was 1.2%. The true incidence was probably higher, since 9/26 of the diagnoses were made at autopsy. No BLPD was observed following autologous HSCT, so risk factor analyses were confined to the 1542 allogeneic HSCT. Factors assessed were HLA-mismatching (⩾1 antigen), T cell depletion (TCD), presence of acute GvHD (grades II–IV), donor type (related vs unrelated), age of recipient and donor, and underlying disease. Factors found to be statistically significant included patients transplanted for immune deficiency and CML, donor age ⩾18 years, TCD, and HLA-mismatching, with recipients of combined TCD and HLA-mismatched grafts having the highest incidence. Factors found to be statistically significant in a multiple regression analysis were TCD, donor age and immune deficiency, although 7/8 of the patients with immunodeficiencies and BLPD received a TCD graft from a haploidentical parent. The overall mortality was 92% (24/26). One patient had a spontaneous remission, but subsequently died >1 year later of chronic GVHD. Thirteen patients received therapy for BLPD. Three patients received lymphocyte infusions without response. The only patients with responses and long-term survival received alpha interferon αIFN). Of seven patients treated with αIFN there were four responses (one partial and three complete). These data demonstrate thatαIFN can be an effective agent against BLPD following HSCT, if a timely diagnosis is made.

Cancer PracticeVolume 6, Issue 4 p. 202-205 Weight Loss, Skin Rash, and Cough Following Bone MarrowTransplantation for Chronic Myelogenous Leukemia Robert P. Witherspoon MD, Robert P. Witherspoon MD Medical Director Ambulatory Clinics, Clinical Research Division, Fred Hutchinson Cancer Research Center,Associate Professor of Medicine, University of Washington, Seattle, Washington,Search for more papers by this authorPhilip McGlave MD, Philip McGlave MD Professor of Medicine Director Division of Hematology, Oncology, and Transplantation University of MinnesotaMinneapolis, Minnesota,Search for more papers by this authorJudy Beach Campbell RN, Judy Beach Campbell RN Long-Term Follow-Up Nurse Fred Hutchinson Cancer Research Center, Seattle, Washington,Search for more papers by this authorTrish Sigley RD, LD, Trish Sigley RD, LD Clinical Dietitian Specialist Johns Hopkins Oncology Center, Baltimore, Maryland,Search for more papers by this authorKenneth V. I. Rolston MD, Kenneth V. I. Rolston MD Professor of Medicine Section of Infectious Diseases, The University of Texas, M. D. Anderson Cancer Center, Houston, TexasSearch for more papers by this author Robert P. Witherspoon MD, Robert P. Witherspoon MD Medical Director Ambulatory Clinics, Clinical Research Division, Fred Hutchinson Cancer Research Center,Associate Professor of Medicine, University of Washington, Seattle, Washington,Search for more papers by this authorPhilip McGlave MD, Philip McGlave MD Professor of Medicine Director Division of Hematology, Oncology, and Transplantation University of MinnesotaMinneapolis, Minnesota,Search for more papers by this authorJudy Beach Campbell RN, Judy Beach Campbell RN Long-Term Follow-Up Nurse Fred Hutchinson Cancer Research Center, Seattle, Washington,Search for more papers by this authorTrish Sigley RD, LD, Trish Sigley RD, LD Clinical Dietitian Specialist Johns Hopkins Oncology Center, Baltimore, Maryland,Search for more papers by this authorKenneth V. I. Rolston MD, Kenneth V. I. Rolston MD Professor of Medicine Section of Infectious Diseases, The University of Texas, M. D. Anderson Cancer Center, Houston, TexasSearch for more papers by this author First published: 25 December 2001 https://doi.org/10.1046/j.1523-5394.1998.006004202.xRead the full textAbout ToolsRequest permissionExport citationAdd to favoritesTrack citation ShareShare Give accessShare full text accessShare full-text accessPlease review our Terms and Conditions of Use and check box below to share full-text version of article.I have read and accept the Wiley Online Library Terms and Conditions of UseShareable LinkUse the link below to share a full-text version of this article with your friends and colleagues. Learn more.Copy URL No abstract is available for this article. Volume6, Issue4July 1998Pages 202-205 RelatedInformation

Autologous transplantation can induce extended remission in some patients with advanced breast cancer and lymphoma yet nearly 80% and 50%, respectively, will ultimately relapse. In vitro studies suggest that activated natural killer cells (NK) mediate lytic activity against breast cancer and lymphoma cell lines. Therefore, immunotherapy with interleukin-2 (IL-2, Amgen) to activate NK may improve long-term disease-free survival when administered in a post-transplant minimal residual disease setting. To determine the feasibility of administering IL-2 and activation of NK post-transplant, twelve patients (6 breast cancer, 6 lymphoma) were enrolled on a phase I dose escalation study after autologous transplantation (median day + 94, range 50-166). IL-2 was self administered at 0.25 x 10(6) (n = 6) or 0.5 x 10(6) (n = 6) U/m2/day subcutaneously for 84 consecutive days. The best tolerated dose was 0.25 x 10(6) U/m2/day (75% of planned doses given vs. 48% at the higher dose). Dose limiting toxicity occurred in 6 patients (n = 2 at 0.25 x 10(6) U/m2/day, n = 4 at 0.5 x 10(6) U/m2/day) consisting of decreased performance status (n = 2), thrombocytopenia (n = 3, 1 at the lower dose), and mild neutropenia (n = 1 at the lower dose). However, all symptoms resolved within a week following discontinuation of IL-2 and no patient required hospitalization. Circulating soluble IL-2 receptor levels were significantly increased in all patients receiving IL-2. Patients receiving at least 28 days of IL-2 exhibited a greater than 10-fold increment in circulating CD56+bright/CD3- NK. Furthermore, lytic function was increased against NK resistant targets, MCF-7 (breast cancer), and Raji (lymphoma). In vivo IL-2 primed NK cells obtained by lymphapheresis were activated in large-scale ex vivo incubation in high dose IL-2 (1,000 U/mL) at high cell density (10 x 10(6)/mL), in gas permeable bags, and using serum-free media. NK lytic function against MCF-7 and Raji targets was further enhanced. We conclude that low dose subcutaneous IL-2 based immunotherapy is feasible, relatively safe, can be administered in an outpatient setting and hypothesize that additional ex vivo incubation in IL-2 may be used to generate NK cells with potent antitumor effects in vivo.

Autologous transplantation for non-Hodgkins lymphoma and Hodgkin's disease is widely used as standard therapy for those with high-risk or relapsed tumor. Peripheral blood stem cell (PBSC) collections have nearly completely replaced bone marrow stem cell (BMSC) harvests because of the perceived advantages of more rapid engraftment, less tumor contamination in the inoculum, and better survival after therapy. The advantage of PBSC, however, may derive from the hematopoietic stimulating cytokines used for PBSC mobilization. Therefore, we tested a randomized comparison of GM-CSF vs. G-CSF used to prime either BMSC or PBSC before collection for use in autologous transplantation. Sixty-two patients receiving transplants (31 PBSC; 31 BMSC) for non-Hodgkin's lymphoma (n = 51) or Hodgkin's disease (n = 11) were treated. All patients received 6 days of randomly assigned cytokine. Those with cellular marrow in morphologic remission underwent BMSC harvest, while those with hypocellular marrow or microscopic marrow tumor involvement had PBSC collected. Neutrophil recovery was similarly rapid in all groups (median 14 days; range 10-23 days), though two patients had delayed neutrophil recovery using GM-CSF primed PBSC (p = 0.01). Red cell and platelet recovery were significantly quicker after BMSC mobilized with GM-CSF or PBSC mobilized with G-CSF. This speedier hematologic recovery resulted in earlier hospital discharge as well. However, in multivariate analysis, neither the stem cell source nor randomly assigned G-CSF vs. GM-CSF was independently associated with earlier multilineage hematologic recovery or shorter hospital stay. Relapse-free survival was not independently affected by either the assigned stem cell source or the randomly assigned priming cytokine, though malignant relapse was more frequent in those assigned to PBSC (RR of relapse 3.15, p = 0.03). These data document that BMSC, when collected following cytokine priming, can yield a similarly rapid hematologic recovery and short hospital stay compared with cytokine-primed PBSC. Using primed BMSC, no difference in malignant relapse or relapse-free survival was observed. These findings suggest that despite widespread use of PBSC for transplantation, BMSC, when collected following hematopoietically stimulating cytokines, may remain a satisfactory source of stem cells for autologous transplantation. G-CSF and GM-CSF are both effective in priming autologous PBSC or BMSC for collection.

We analyzed the incidence of posttransplant chronic myelogenous leukemia (CML) relapse in 283 consecutive related-donor (n = 177) and unrelated-donor (n = 106) allogeneic transplant recipients. Twenty-two of 165 related-donor recipients with stable or advanced disease at the time of transplant had hematologic relapse of CML following transplant (5-year Kaplan-Meier estimate of relapse, 20%; 95% confidence interval [CI], 11 to 30%). One of 12 patients transplanted in second stable phase following blast crisis also relapsed. Fifteen related-donor transplant recipients relapsed within 5 years of transplant; however, seven relapsed between 5 and 9 years after transplant. Factors independently associated with an increased risk of posttransplant relapse for related-donor recipients included prolonged interval between diagnosis and transplant (relative risk, [RR], 3.81; P = .009) and bone marrow basophilia (RR, 5.62; P = .01). Related-donor recipients with posttransplant chronic graft-versus-host disease (CGVHD) had a decreased risk of relapse (RR, 0.24; P = .005). Only two of 106 unrelated-donor transplant recipients relapsed following transplant (5-year Kaplan-Meier estimate of relapse, 3%; 95% CI, 0% to 7%). When both related- and unrelated-donor recipients were considered, the use of an unrelated donor was independently associated with a decreased risk of relapse (RR, 0.24; P = .07). Twelve of 16 relapsing patients who received further therapy (nine of 13 who underwent second transplant and three of three who received donor leukocyte infusions) remain alive. This analysis shows that relapse, sometimes occurring long after transplant, is an important adverse outcome in allogeneic transplantation for CML. Early transplant, posttransplant CGVHD, and use of an unrelated donor are associated with a reduced incidence of relapse, perhaps due to allogeneic disparities enhancing the graft- versus-leukemia effect.