Since the first successful umbilical cord blood transplantation (UCBT) for a patient with Fanconi anemia , umbilical cord blood (UCB) has been considered to be a reasonable alternative to bone marrow (BM) or G-CSF-mobilized peripheral blood (PB) as a source of hematopoietic stem cells, especially in children [2–6]. In Korea, the history of UCBT dates back to 1996, when it was used to treat a child with severe aplastic anemia (SAA). Although the outcome was unsuccessful, a child with acute lymphoblastic leukemia (ALL) was treated successfully with UCBT 2 years later. However, UCBT was not a realistic alternative to transplantation with BM or PB in Korea until January 2003, when it was started to be covered by the National Health Insurance for children under the age of 16 years. Since then, the number of pediatric cases of UCBT in Korea has increased dramatically. In March 2004, the first successful double-unit UCBT in Korea was accomplished for a 14-year-old girl with acute myelogenous leukemia (AML), which paved the way for its application in older children and adults. In January 2009, the Korean National Health Insurance policy started to cover all UCB recipients including adults without an appropriate familial donor.
While it is estimated that more than 20,000 cases of UCBT have been performed worldwide to date , the current status of UCBT in Korea has not yet been reported. To address this, a retrospective multicenter study of pediatric UCBT in Korea was performed. Several prognostic factors that affect the outcomes of these patients were also identified.
The clinical outcomes of 236 cases of pediatric UCBT performed at 13 transplant centers in Korea between 1996 and 2006 were reviewed retrospectively. Details about UCBT from a given institution were registered to the Korean Cord Blood Transplantation Registry, and data were verified by comparing reports to primary data sources. Written informed consents were obtained prior to transplantation and patients and/or guardians were also informed that data could be reported back to the registry. The study was approved by the Institutional Review Board of Samsung Medical Center. The HLA disparity between the recipient and the graft was determined at the 6-loci antigen level only (i.e., HLA-A, -B, and -DR) because most UCB banks in Korea do not provide high-resolution HLA typing. In terms of the intensity of conditioning regimens, reduced-intensity regimens were generally defined as reported previously [8, 9]. Briefly, the upper limits of oral busulfan, intravenous busulfan, melphalan, and total body irradiation (TBI) were 8 mg/kg, 6.4 mg/kg, 140 mg/m2, and 6 Gy, respectively, for consideration as reduced-intensity conditioning regimens. The transplant outcomes in terms of various UCBT-related parameters were analyzed. Events were defined as engraftment failure, relapse or progression of the primary disease, and deaths due to any cause whichever occurred first. For statistical analyses, the Kaplan-Meier method was used to estimate the probabilities of the survival which were compared using the log rank test. The cumulative incidences of engraftment and transplant-related mortality (TRM) were also calculated. The paired t-test was used to identify the factors that determine the “winning” unit after double-unit UCBT. The chi-square test or the Fisher's exact test was used to compare the frequencies, and the Mann-Whitney U-test was employed to compare the continuous variables of two groups. To identify the prognostic factors affecting engraftment, graft-versus-host-disease (GVHD), TRM, relapse and survival, the following variables were used in univariate analyses: age, body weight, the number of transplanted UCB units, cell dose, conditioning intensity, TBI, T-cell depleting agents (in vivo T-cell depletion), salvage or primary transplantation, methotrexate (MTX) in GVHD prophylaxis, early complete chimerism (at 1 month following UCBT), blood stream infection within 30 days after UCBT, cytomegalovirus (CMV) infection, CMV disease, and severity of acute or chronic GVHD. Variables with P < 0.1 in univariate analysis of engraftment, GVHD, TRM and relapse were entered in the logistic regression model for multivariate analysis. For multivariate analysis of survival, the Cox proportional hazard model was used. SPSS version 12.0 was used for all statistical analyses, and statistical significance was accepted when the P-value was less than 0.05. Given the small numbers of sibling (n = 1), autologous (n = 6), and unknown (n = 3) donor transplants, only the results of unrelated donor transplants (n = 226; 95.8%) were included for the statistical analyses.
The characteristics of unrelated UCBT cases are summarized in Table I. The median age and weight at the time of UCBT were 7.0 years (range, 0.5–18.8) and 22.7 kg (range, 7.0–73.8), respectively. The underlying primary diseases varied, but acute leukemia (n = 167) comprised almost three-quarters of all the cases. The remaining patients had inborn errors, (n = 21), myelodysplastic syndrome (MDS) or juvenile myelomonocytic leukemia (JMML) (n = 15), SAA (n = 12), chronic myelogenous leukemia (n = 5), solid tumors (n = 3), and others (n = 3). The CMV seropositive rate of recipients was 91.7%. The majority of patients (71.7%) received one unit of UCB; the remaining 28.3% received two units. UCBT was performed as salvage transplantation in 22 patients whose previous autologous (n = 9) or allogeneic (n = 13) transplantation had been unsuccessful. The preparative regimens varied depending on the patients' underlying diseases and general condition, but most patients (90.7%) received a conventional myeloablative preparative regimen. In addition, 78.8% received antithymocyte globulin (ATG) or similar agents for in vivo T-cell depletion. The mainstay of GVHD prophylaxis was cyclosporine (CsA) plus corticosteroid, or CsA plus mycophenolate mofetil (MMF).
Table I. Characteristics of the Unrelated Umbilical Cord Blood Transplantation Cases
The degree of HLA disparity between the recipients and the grafts is indicated in Table II. Of the single-unit UCBT recipients, only 8.6% matched the donors fully; the majority (73.5%) was mismatched in 1 of the 6 loci. Of the double-unit UCBT recipients, more than a half of those (53.1%) did not match each of their two donors by 1 of the 6 loci (i.e., there were 1/6 and 1/6 mismatches).
Table II. HLA Disparity in A, B, and DR Loci in Unrelated UCBT
Among the parameters such as cell dose, TBI, in vivo T-cell depletion, and MTX for GVHD prophylaxis that could potentially affect engraftment, nucleated cell dose ≥1.5 × 107/kg (P = 0.016), CD34+ cell dose ≥1.0 × 105/kg (P = 0.021), and TBI-containing regimen (P = 0.011) favored successful engraftment in univariate analysis, whereas in vivo T-cell depletion (P = 0.175) and use of MTX (P = 0.731) had no significant impact on engraftment. TBI was the only significant factor which favored engraftment in multivariate analysis (P = 0.009, Odds ratio (OR) = 3.50 [95% C.I. 1.36–9.00]). Table III shows the details of cell doses and engraftments after the single- and double-unit UCBT. Overall, the median nucleated cell and CD34+ cell doses were 4.84 × 107/kg (range, 0.06–22.11) and 2.00 × 105/kg (range, 0.32–84.00), respectively. It took a median of 18 days to attain neutrophil engraftment (>0.5 × 109/L) and 45 days to achieve platelet recovery (>20 × 109/L). The cumulative incidence of neutrophil recovery (>0.5 × 109/L) by day 60 was 90.7%, while that of platelet recovery (>20 × 109/L) by day 100 was 74.7%. When the single- and the double-unit UCBT groups were compared in terms of cell doses and engraftments, no significant differences were observed. However, the double-unit recipients were significantly heavier (18.8 vs. 35.8 kg, P < 0.0005) and older (5.6 vs. 10.3 years, P < 0.0005) than the single-unit recipients. Of 40 patients who achieved complete donor chimerism (CC) at 1 month following double-unit UCBT, 13 patients (32.5%) showed balanced co-engraftment of the two donors, while the other 27 revealed only one donor chimerism. The impact of cell dose parameters on determining the dominant unit was evaluated for those who achieved one donor chimerism and whose paired cell dose information was available. The engrafted units had higher numbers of nucleated cells (P = 0.003) and colony-forming units-granulocyte/macrophage (CFU-GM; P = 0.008) than the non-engrafted units, whereas the numbers of CD34+ cells and CD3+ cells had no impact on determining the dominant unit (P = 0.382 and P = 0.161, respectively) (Fig. 1).
Table III. Cell Doses Used and Engraftment Characteristics in Unrelated UCBT
Comparison of the single-UCBT and double-UCBT groups.
NC × 107/kg, median (range)
CD34+ cells × 105/kg, median (range)
Days to neutrophil >0.5 × 109/L, median (range)
Days to platelet >20 × 109/L, median (range)
Cumulative incidence of neutrophil recovery by day 60 (%)
Cumulative incidence of platelet recovery by day 100 (%)
Age in years, median (range)
Body weight in kg, median (range)
Grade 2–4 and 3–4 acute GVHD developed in 41.1 and 15.0% of all patients, respectively. The single- and the double-unit UCBT groups exhibited similar incidences of acute GVHD: 41.3 and 40.6% developed Grade 2–4 GVHD (P = 0.92), and 14.7 and 15.6% developed Grade 3 or 4 GVHD, respectively (P = 0.86). Factors which contributed to the development of Grade 2–4 GVHD in univariate analysis were nucleated cell dose ≥2.5 × 107/kg (P = 0.020), CD34+ cell dose ≥1.0 × 105/kg (P = 0.045), TBI-based conditioning regimens (P = 0.019), the absence of in vivo T-cell depletion (P = 0.001), the presence of CMV infection (P < 0.0005), and the presence of CMV disease (P < 0.0005). T-cell dose or GVHD prophylaxis regimens did not affect the development of GVHD. In multivariate analysis, in vivo T-cell depletion (P = 0.003, OR = 0.22 [95% C.I. 0.08–0.59]) and CMV infection (P < 0.0005, OR = 4.66 [95% C.I. 2.16–10.07]) were the factors that significantly affected the development of Grade 2–4 GVHD. Of the 147 evaluable patients, 53 (36.1%) developed chronic GVHD that was either limited (n = 31) or extensive (n = 22). The single- and double-UCBT groups were similar in terms of the incidences of limited (19.8% vs. 23.9%, P = 0.57) and extensive (14.9% vs. 15.2%, P = 0.95) chronic GVHD. In multivariate analysis, acute GVHD ≥ Grade 2 (P = 0.006, OR = 6.76 [95% C.I. 1.73–26.40]) and the presence of CMV disease (P = 0.027, OR = 8.35 [95% C.I. 1.27–54.93]) significantly affected the development of chronic GVHD.
CMV infection, as indicated by the presence of CMV antigenemia or PCR positivity, was noted in 98 of the 215 evaluable patients (45.6%), and 45 of the 215 patients (20.9%) developed CMV disease in the lungs (n = 20), gut (n = 10), retinas (n = 6), liver (n = 2), multiorgan (n = 5), or other sites (n = 2). All but one of those who developed CMV infection and all of those who developed CMV disease had been seropositive for CMV before transplantation.
Survival and prognostic factors
The 5-year overall survival and event-free survival were 47.5 and 36.9%, respectively, with a median follow-up of 34 months (range, 1–99) (Fig. 2a). Figure 2b illustrates the overall survival of the patients after they were divided according to their primary diseases. In the single-unit UCBT group, the survival rate was slightly better with 0 or 1/6 antigen mismatches than with 2/6 antigen mismatches, although this did not achieve statistical significance (49.1% vs. 36.9%; P = 0.19). The single- and double-unit UCBT groups were similar in terms of long-term overall survival (48.1% vs. 46.9%, respectively; P = 0.96).
Table IV shows the results of univariate and multivariate analyses of overall survival. Univariate analysis revealed that factors that adversely affected the survival rates of all patients were the use of TBI-based conditioning (P = 0.006), the absence of in vivo T-cell depletion (P = 0.004), salvage transplantation (P = 0.013), the failure to achieve early (1 month post-transplant) CC (P = 0.023), and the presence of CMV disease (P = 0.004). Multivariate analysis revealed that the TBI-based conditioning regimen (P = 0.007, Hazard ratio (HR) = 1.98 [95% C.I. 1.21–3.40]), salvage transplantation (P = 0.001, HR = 2.64 [95% C.I. 1.48–4.72]), the lack of CC at 1 month (P < 0.0005, HR = 2.55 [95% C.I. 1.59–4.08]), and the presence of CMV disease (P = 0.001, HR = 2.28 [95% C.I. 1.40–3.72]) were significant risk factors for survival. Analysis of the 167 patients with acute leukemia by univariate analysis revealed that an advanced disease status (>CR2; P < 0.0005), the use of the TBI-based conditioning regimen (P = 0.001), the absence of in vivo T-cell depletion (P = 0.002), the failure to achieve early CC (P = 0.015), and the presence of CMV disease (P = 0.005) were the factors that adversely affected their survival. Multivariate analysis showed that the survival of the acute leukemia patients was adversely affected by an advanced disease status (P = 0.003, HR = 3.31 [95% C.I. 1.50–7.34]) and the presence of CMV disease (P = 0.047, HR = 2.11 [95% C.I. 1.01–4.42]). Age, gender, CMV serological status, type of GVHD prophylaxis, cell dose, and the presence of acute and/or chronic GVHD did not significantly affect the survival rates (data not shown). An advanced disease status (>CR2; P = 0.001), myeloablative dose of TBI (≥10 Gy; P = 0.007), the failure to achieve early CC (P = 0.035), and the presence of CMV infection (P = 0.038) were the high risk features for developing leukemic relapse, but only an advanced disease status was independently significant in multivariate analysis (P = 0.037, OR = 4.10 [95% C.I. 1.09–15.45]). The cumulative incidence of TRM at 3 months, 1 year, and 2 years after transplantation was 19.0, 33.9, and 35.6%, respectively. Multivariate analysis revealed that salvage transplantation (P = 0.008, OR = 7.34 [95% C.I. 1.70–31.81]) and acute GVHD Grade 3–4 (P = 0.039, OR = 4.34 [95% C.I. 1.08–17.43]) were the independent risk factors for developing TRM.
Table IV. Univariate and Multivariate Analyses of Overall Survival in Unrelated UCBT
5-year survival rate ± S.E.
Hazard ratio (95% C.I.)
Abbreviations: CC, complete chimerism; CMV, cytomegalovirus; CR, complete remission; TBI, total body irradiation; UCBT, umbilical cord blood transplantation.
All patients (n = 226)
0.538 ± 0.045
0.347 ± 0.062
In vivo T-cell depletion
0.324 ± 0.077
0.505 ± 0.041
0.498 ± 0.040
0.273 ± 0.095
Lack of CC at 1 month
0.592 ± 0.050
0.388 ± 0.067
0.508 ± 0.043
0.318 ± 0.074
Acute leukemia (n = 167)
0.575 ± 0.054
0.134 ± 0.088
0.565 ± 0.056
0.319 ± 0.067
In vivo T-cell depletion
0.333 ± 0.122
0.486 ± 0.048
Lack of CC at 1 month
0.604 ± 0.060
0.371 ± 0.077
0.512 ± 0.051
0.269 ± 0.092
It has been shown that for both children and adults, the leukemia-free survival rate after mismatched UCBT is similar to that of matched unrelated BMT [3, 5, 10, 11]. However, a relatively high incidence of engraftment problems associated with various infectious complications often leads to high early TRM after UCBT. In the present study, the cumulative incidence of engraftment of our cohort was 90.7%. While this is similar to those reported recently in other studies [5, 12, 13], it is apparently lower than that following BM or G-CSF-mobilized PB transplants.
The pioneering work of Barker et al.  suggested that the use of two partially HLA-matched UCB units can elevate the engraftment rate. This double-unit approach may overcome a major problem associated with single-unit UCBT, especially when the number of cells that can be transplanted is limited. Consequently, double-unit UCBT is believed to increase the chance of engraftment in older children and adults. While there are some concerns that the use of two different UCB units may increase the incidence of GVHD, this does not seem to be a problem for patients with acute leukemia as this may also enhance the graft-versus-leukemia effect [15, 16]. However, in the present study, there were no detectable differences between the single- and the double-unit UCBT groups in terms of engraftment parameters, GVHD, and overall survival. This may be explained by the fact that the cell doses relative to the recipient body weights of these two groups were almost exactly the same. These results suggest that the double-unit approach is both safe and suitable for those patients who are likely to need more than one unit. Further supporting the feasibility of the double-unit approach is the work of Yoo et al. . who reported that some UCB units lacked clonogenic capacity after thawing and that the recipients who received such functionally exhausted units eventually experienced primary engraftment failure. Thus, the double-unit approach may actually be safer than the single-unit approach since it not only ensures that the cell dose is adequate, it also reduces the chance of graft problems caused by the transplantation of a functionally impaired unit. These potential advantages explain why double-unit UCBT is being performed with increasing frequency in Korea. Indeed, it is sometimes employed on a routine basis, even if one unit would be able to deliver an acceptable cell number. However, given that the routine use of double-unit UCBT strategy could potentially elevate the frequency of harmful events such as severe GVHD in some patients, a clinical trial testing this approach is warranted.
Another way to augment the cell numbers in a single unit is to expand the UCB cells ex vivo before transplantation. However, a number of clinical studies have suggested that ex vivo-expanded UCB cells may not provide long-term benefits because it is the more committed cells that tend to expand, which reduces the long-term repopulating potential of the UCB cell population [18–20]. Further studies are needed to overcome this problem.
CMV disease was associated significantly with a poorer outcome in the present study. Notably, we found that the incidence of CMV disease (20.9%) was much higher among our study cohort as compared with that reported in other UCBT studies which ranged 0–10.8% [6, 21–23]. This extremely high incidence of CMV disease may be due to several reasons. First, the recipients in the present study had a very high prevalence of CMV seropositivity (91.7%), which is generally the case for developing countries like Korea. Second, a preemptive strategy was exclusively adopted instead of a prophylactic one in using ganciclovir for CMV infection. Third, in the majority of cases (83.2%), ATG or other anti-T cell agents were included in the preparative regimens; this depletion of T-cells in vivo may have attenuated the host defense mechanisms against CMV. While the latter possibility appears to be discounted by the fact that univariate analysis showed in vivo T-cell depletion was a good prognostic factor, the multivariate analysis showed that in vivo T-cell depletion was not a significant favorable prognostic factor (Table IV). In contrast, the multivariate analysis revealed that the absence of CMV disease was a favorable prognostic factor. Notably, while large studies of UCBT have routinely employed pre-transplant ATG treatment [10, 12, 24], a recent study  showed that a good engraftment rate (94.8%) could be achieved without ATG, even in adult recipients. These observations together suggest that particularly for countries like Korea, where the CMV seroprevalence is very high, the incidence of post-UCBT CMV disease could be reduced by developing novel conditioning regimens that do not contain ATG or similar agents, and that work in conjunction with more effective anti-CMV strategies.
A recent study by van Heeckeren et al.  showed that UCB graft mismatching at DRB1 may be an important risk factor of acute GVHD and survival. However, as mentioned earlier, most UCB banks in Korea cannot afford high-resolution molecular typing of HLA-DRB1, which means it was not possible to accurately determine the influence of HLA-disparity on the outcome of UCBT in this study. Given that a good HLA match between graft and recipient significantly improves the outcome of UCBT, it is strongly recommended that all UCB banks in Korea perform high-resolution molecular typing of HLA-DRB1, as this will enable the physicians to select the best-matched UCB units. The outcome of UCBT in Korea would also be improved by the establishment of a government-controlled management system that supervises the UCB banks in Korea and ensures high quality product delivery; such a system is not currently in place in Korea.
In conclusion, double-unit UCBT seems to yield good outcomes for older or heavier children without increasing the risk of GVHD. However, this protocol did not appear to improve survival relative to single-unit UCBT. The incidence of CMV disease in the present study cohort was very high, which indicates the importance of thorough surveillance and proper management of CMV infection in Korea, which is a country where CMV seropositivity exceeds 90%. Finally, since an advanced disease status was found to be the most important risk factor for poor survival in acute leukemia, it is strongly recommended that a timely UCBT is performed when an appropriate BM or PB donor is not available.