A retrospective chart review was performed on children with acute leukemia who received an allogeneic HSCT at the Samsung Medical Center, Korea from January 2005 to December 2010. We excluded recipients who received two or more types of stem cell sources. Patients who had received previous allogeneic transplantation were excluded. UCBT was performed if no suitable HLA-MRD or URD were available. HLA compatibility was determined by high-resolution typing for HLA-A, -B, and -DRB1 as well as by serotyping for HLA-C for HSCT from URD. UCBT grafts were matched with at least 4-6 HLA-A and -B at antigen level and -DRB1 at the allele level to the recipient and in patients receiving 2 UCB units to each other. Double unit UCBT was performed if no suitable UCB unit with nucleated cell counts>3.0×10⁷/kg or CD34+ cell count>1.7×10⁵/kg was available. All patients underwent a myeloablative conditioning regimen. The GVHD prophylaxis was as follows: cyclosporine alone for HSCT from related BM or PBSC (MRD), cyclosporine and methotrexate for HSCT from unrelated BM or PBSC (URD), and cyclosporine and mycophenolate mofetil for HSCT from UCB. While antithymocyte globulin (ATG) was mostly used in the conditioning regimen for the UCBT, ATG was not administered for HSCT when other stem cell sources were used. All included recipients were divided into three groups according to the donor source as follows: MRD, URD, and UCB.

Neutrophil recovery was defined as the estimated time from transplantation to the first of three consecutive days when an absolute neutrophil count (ANC) of at least 0.5×10⁹ per liter was obtained. Platelet recovery was defined as the achievement of a platelet count of 20×10⁹ per liter without support of transfusions for seven continuous days. Graft failure was defined as the lack of donor-derived neutrophil recovery or the requirement for either a boost from the same donor or a second transplant for lack of count recovery. Advanced disease was defined as more than a complete remission (CR)1 for acute myeloid leukemia (AML), more than CR2 for acute lymphoblastic leukemia (ALL), and refractory disease. Diagnosis and grading of acute and chronic GVHD were based on standard clinical criteria and biopsy when available7,⁸⁾. Recipients who survived at least 100 days with sustained engraftment were evaluated for chronic GVHD. Relapse was defined as a recurrence of leukemia. Non-relapse mortality (NRM) was defined as death after transplant that was not preceded by recurrent or progressive malignancy. Event-free survival (EFS) was defined as survival in a state of continuous complete remission.

Patient characteristics and the outcomes of transplantation were compared across groups using the chi-square and Fisher exact tests. Continuous variables were compared using the Kruskal-Wallis test with the Tukey test using ranks. The five-year probabilities of overall survival (OS) and EFS were estimated with the Kaplan-Meier methodology and compared using the log-rank test for univariate comparisons. For analyzing EFS, leukemic relapse or death was considered an event. For analysis of OS or EFS, data from recipients without an event were censored at the last follow-up after HSCT. The cumulative incidence was estimated for NRM and relapse in order to take competing risks into account according to the method of Fine and Gray. Multivariate analysis using Cox proportional hazard regression model and logistic regression model of outcomes after transplantation was performed using variables with a P value of 0.1 or less according to groups. The variable for graft type was retained in all steps of model building. Results were expressed as hazard ratios that represented the relative risk (RR) of occurrence of the outcomes with URD or UCB as compared to MRD. Other variables that were considered in multivariate analysis for GVHD included age of transplant recipient, status of the disease at transplantation, the conditioning regimen (i.e., with irradiation vs. without it), the use of ATG, and the total dose of nucleated cells. In multivariate analysis for relapse, NRM, and EFS, other variables were age of transplant recipient, status of the disease at transplantation, the conditioning regimen (i.e., with irradiation vs. without it), acute GVHD, and extensive chronic GVHD. All statistical analyses were performed using SPSS ver. 17.0 (SPSS Inc., Chicago, IL, USA) and STATA release 11 (StataCorp LP., College Station, TX, USA).

The characteristics of the patients are shown in Table 1. All of the recipients included in the study were classified under three groups: MRD (n=33), URD (n=46), and UCB (n=41). The median age of all study patients was 8.3 years. The male to female ratio was 1.3:1. Three groups were similar with respect to sex, age at HSCT, and conditioning regimens. Among the patients, the proportions with AML, ALL, and acute biphenotypic leukemia were similar in all three groups (P=0.85). The three groups also had a similar disease status (P=0.31). Although ATG was not administered to recipients of MRD and URD, 78% of UCB recipients received ATG (P<0.01). All UCB grafts, with the exception of one, were HLA-mismatched, whereas 48% of URD transplants were HLA-matched (P<0.01). In 24 recipients of HLA-mismatched URD graft, four patients received 3-5 loci mismatched grafts, seven patients from 2 loci mismatched graft, and 13 patients from 1 locus mismatched grafts. Among 13 recipients of 1 locus mismatched graft, mismatched locus was HLA-A in 3 patients, HLA-B in 2 patients, HLA-C in 5 patients, and HLA-DRB1 in 3 patients. Among 41 recipients of UCB, 31 patients received double-unit UCBT. The median number of nucleated cells and CD34+ cells infused in the UCB group were lower than other groups.

Table 2 shows the outcomes after HSCT. All recipients in the MRD and URD groups showed successful engraftment, whereas graft failure occurred in 4 recipients in the UCB group (P<0.01). Neutrophil and platelet recoveries were significantly delayed after UCBT compared to HSCT from MRD or URD (both P<0.01).

In univariate analysis, the rate of grades II to IV acute GVHD was higher in recipients of URD and UCB transplants compared to MRD recipients (P=0.04). Using the MRD group as the reference category in a multiple regression analysis, the risk of grades II-IV acute GVHD was similar among recipients of transplants from URD and UCB and the RR was 2.4 and 2.8, respectively, however; which did not reach the statistical significance (P=0.1 and P=0.09, respectively) (Table 3). The rate of grades III to IV acute GVHD was 20% in the URD group. In the multiple regression analysis, there was a higher risk of grade III to IV acute GVHD (RR, 15.2; 95% confidence interval [CI], 1.2 to 186.2; P=0.03) in recipients of URD grafts (Table 3). Despite the high-degree of HLA mismatch between donor and recipient, the risk of acute GVHD grades III to IV after UCB transplant was not significantly higher than those after MRD transplant (Table 3). The rate of extensive chronic GVHD was 53.8% in the URD group, which was significantly higher than the MRD and UCB groups (Table 2). In the multiple regression analysis, the RR of extensive chronic GVHD was 6.9 (95% CI, 1.9 to 25.2) compared to the MRD group (P<0.01). The risk of extensive chronic GVHD was not higher in the UCB group compared to the MRD group (P=0.32).

When the risk of GVHD was compared in URD group according to the HLA disparity, the rate of grades III to IV acute GVHD was 9.5% and 29.5% in recipients of HLA-matched URD graft and HLA-mismatched URD graft, respectively (P=0.1). The rate of extensive chronic GVHD was 50.0% and 57.9% in recipients of HLA-matched URD graft and HLA-mismatched URD graft, respectively (P=0.6).

There was a trend toward a higher cumulative incidence of relapse in the MRD group, but this trend did not reach statistical significance (Fig. 1A). In a multivariate regression analysis, the only factor that had a significant impact on the risk of relapse was the advanced disease (RR, 12.7; 95% CI, 2.3 to 66.9; P<0.01). Moreover, the extensive chronic GVHD showed a protective effect on the relapse (RR, 0.1; 95% CI, 0.04 to 0.6; P<0.01).

The cumulative incidence of NRM was higher in transplants with UCB (Fig. 1B). In a multivariate analysis, extensive chronic GVHD increased the risk of NRM (RR, 9.0; 95% CI, 1.5 to 54.1; P=0.02). When evaluating the causes of death according to the donor source, disease recurrence was the most frequent cause of death in recipients of MRD (8/9, 86.3%) and URD (9/14, 64.3%). However, the most frequent cause of death in recipients of UCB was infection (7/15, 46.7%). The proportion of GVHD as the cause of death was 11.1% in the MRD group, 28.6% in the URD group, and 13.3% in the UCB group (P=0.46).

EFS and OS were similar among all groups (Fig. 2). The 5-year EFS was 48.7% for the MRD group, 48.3% for the URD group, and 45.9% for the UCB group, respectively (P=0.98). The factors associated with EFS advanced disease at transplantation (RR, 3.2; 95% CI, 1.3 to 7.8; P<0.01) at transplantation and total body irradiation (RR, 2.1; 95% CI, 1.0 to 4.3; P=0.04) based on multivariate analysis. The 5-year OS was 53.3% for the MRD group, 56.6% for the URD group, and 55.2% for the UCB group, respectively (P=0.80).