Kei Ohnuma, Division of Clinical Immunology, Advanced Clinical Research Centre, Institute of Medical Science, University of Tokyo, 4-6-1, Shirokanedai, Minato-ku, Tokyo 108-8639, Japan. E-mail: email@example.com
Factors influencing the outcome for 39 children with haematological malignancy who were subjected to a cord blood transplantation (CBT) from genotypically HLA-mismatched unrelated donors were analysed. This retrospective study included 21 children with acute lymphoblastic leukaemia, 15 with acute myelogenous leukaemia and one each with chronic myelogenous leukaemia, refractory anaemia with myelodysplastic syndrome (MDS) and juvenile myelomonocytic leukaemia (JMML). Those subjected to CBT during the first or second complete remission (CR) and MDS without blasts were assigned to the standard-risk (SR) group (n = 16). Patients in third or subsequent remission, relapse or partial remission with refractory leukaemia at the time of CBT were considered to be in advanced phase, and placed in the high-risk (HR) group (n = 11). JMML and the second CR after a relapse (n = 8), or bone marrow failure after a rejection (n = 3), following haematopoietic stem cell transplantation (HSCT) in the first CR were included in the high-risk group. Kaplan–Meier estimates for neutrophil and platelet recovery were 83·7 ± 12·2 at d 60 and 55·4 ± 16·6% at d 100 respectively. The incidence of grades II–VI acute graft-versus-host disease was 58·5 ± 16·8%. The Kaplan–Meier estimate for 3-year event-free survival (EFS) was 49·2 ± 16·6. From multivariate analysis, the most important factor influencing EFS was disease status at CBT: SR patients had a 3-year EFS of 75·0 ± 21·6%, compared with 29·6 ± 20·6% for those with HR disease (P = 0·013, RR 4·746, 95% CI 1·382–16·298). These data confirm that HLA-mismatched, unrelated CBT is a feasible procedure to cure a significant proportion of children with leukaemia, especially if conducted in a favourable phase of the disease.
Previously published reports, neither specifically analysing the outcome of children with leukaemia nor using molecular techniques for HLA-class I typing, did not provide clinical data that could establish conclusively the safety and efficacy of CBT from genotypically HLA-mismatched unrelated donors in children with leukaemia (Rubinstein et al, 1998; Locatelli et al, 1999).
This Kanagawa Cord Blood Bank (KCBB) report describes the results obtained from 39 paediatric patients with leukaemia and myelodysplastic syndrome (MDS) who received CBT from genotypically HLA-mismatched unrelated donors. In particular, the HLA disparity and disease-related factors that have affected event-free survival, the incidence of acute GVDH, engraftment and the relapse of HLA-mismatched CBT from unrelated donors have been specifically analysed.
Patients and methods
Patients Fifty-two donor–recipient pairs underwent unrelated CBT between February 1997 and September 2000. Eligibility and donor selection have been previously described (Nishihira et al, 1998). Among these 52 patients, 41 suffered from haematological malignancy, and 39 (22 men and 17 women, median age 3·1 years, range 0·5–28·0 years and median body weight 12·5 kg, range 6·5–66·6 kg) had HLA-mismatched CBT from unrelated donors. Other patient characteristics are shown in Table I.
Table I. Demographic characteristic of 39 CBT cases.
High-resolution molecular typing for class I (A and B) and class II (DRB1).
†NCC denotes nucleated cell numbers.
TBI, total body irradiation. ATG anti-thymocyte globulin. CSA, cyclosporine A. MTX, methotexate.
Primary data and annual follow-up reports were submitted to the data centre of the KCBB by the investigators. Transplantations were performed at 20 institutes throughout Japan (see Appendix). Twenty-one patients had acute lymphoblastic leukaemia (ALL), 15 had acute myeloid leukaemia (AML) and one each had chronic myelogenous leukaemia (CML), refractory anaemia with MDS (MDS-RA), and juvenile myelomonocytic leukaemia in accelerated phase (JMML-AP). The disease status of leukaemia patients undergoing unrelated CBT are detailed in Table I. For patients with leukaemia and MDS, disease status has been reported to be the main factor influencing transplant outcome (Balduzzi et al, 1995; Casper et al, 1995; Locatelli et al, 1999; Uderzo et al, 2000), so children receiving transplants during the first or second complete remission (CR) and MDS without blasts were assigned to the standard-risk (SR) group (n = 16). Those in their third or subsequent remission, relapse or partial remission with refractory leukaemia at the time of CBT were considered to be in advanced phase and placed in the high-risk (HR) group (n = 11). JMML-AP and second CR after relapse of HSCT in first CR were included in the HR group (n = 8). Three patients who had CBT during bone marrow failure after previous HSCT were included in the HR group. The median time interval between initial diagnosis and CBT was 18·5 months (range 4·5–54·0 months).
Karyotype analysis at diagnosis was available for 12 of 39 children. MLL gene rearrangement was reported in six cases of infant ALL (five in the first CR and one in relapse at CBT), and in three case of AML (one in the first CR and two in relapse at CBT). Philadelphia chromosome was described in one with ALL in the first CR. One patient with JMML had a normal karyotype. The karyotype analysis was not taken into account in the statistical analysis because only a limited number of karyotypes were available. Notably, 11 patients had previously received some kind of HSCT (five with autologous bone marrow or peripheral blood progenitor cells, four with unrelated bone marrow, one with unrelated CBT, and another with HLA-matched sibling bone marrow). Serological typing for HLA-A, B and DR was performed using a modified lymphocyte cytotoxicity test. With the use of polymerase chain reaction single-strand conformation polymorphism (PCR SSCP), HLA alleles were identified as class I (A, B) and class II (DRB1). Two technicians in a single laboratory (Kanagawa Prefecture Red Cross Blood Centre) examined the assays to confirm that the results were credible. Informed consent on HLA allele analysis was obtained from the guardians of patients and the mothers of donors. The HLA disparity from serological and molecular assays of HLA-A, B and DR is shown in Table I. In two patients, HLA genotype data were not obtained because of abnormal haematological conditions found at the examination given before CBT. The cytomegalovirus (CMV) serological state was examined in 30 patients before unrelated CBT: 24 patients were CMV seropositive.
Transplantation procedures The methods used for collecting, processing, cryopreserving and storing CB have been described previously (Fraser et al, 1998; Nishihira et al, 1998). All CB samples but one (AML) were thawed and infused without washing. The median number of nucleated donor cells is shown in Table I.
The conditioning regimen and acute GVHD prophylaxis varied according to the centre policy, type of disease, previous treatment and disease status at the time of unrelated CBT. The parents of the patient gave consent to CBT after being informed of the potential risks and benefits of the procedure. A summary of the preparative regimens, GVHD prophylaxis, and the use of anti-thymocyte globulin (ATG) is shown in Table I. All but four patients (JMML and AML) were given recombinant human granulocyte colony-stimulating factor (G-CSF) beginning in the first week post transplant; one AML patient was given recombinant macrophage colony-stimulating factor (M-CSF), and three had no haematopoietic growth factors. Supportive therapy varied among the transplant centres. Protocols for intensive myeloablative therapy and the use of unrelated-donor CB for transplantation were reviewed and approved by the institutional review boards at each transplant centre.
End points The status of 39 patients with haematological malignancies was evaluated based on the last follow-up report (September 2000). Chimaerism of the marrow cells was evaluated by fluorescence in situ hybridization for the X chromosome in sex-mismatched grafts, or HLA allele-specific hybridization in cases of sex-matched transplantations. Neutrophil engraftment was defined as the first of 3 consecutive days when the absolute neutrophil counts reached 0·5 × 109/l. The definition of platelet engraftment was the first of 7 consecutive days when the platelet counts remained above 2000 × 109/l without transfusion. Patients in whom no engraftment occurred were censored if the patient died before d 60 post transplant. Those receiving a second transplant for non-engraftment were censored at the time of the second transplant. Children were considered at risk of developing acute GVHD from d 1 post transplant. Patients with sustained engraftment of donor haemopoietic cells who survived for more than 100 d post transplant were evaluated for the development of chronic GVHD. Acute and chronic GVHD were evaluated according to standard criteria (Shulman et al, 1980; Przepiorka et al, 1995). Relapse was indicated by the morphology of leukaemic cells in the bone marrow, cerebrospinal fluid or peripheral blood, or by cytogenetic assay of a leukaemic clone. The time interval between CBT and relapse was considered with censoring at death in CR. Overall survival (OAS) was defined as the time interval between CBT and death due to any cause. Event-free survival (EFS) was defined as the time between CBT and a first event, which included graft rejection, relapse and death in CR.
Statistical analysis The results of this study were re-analysed on 30 September 2000. The median duration of follow-up was 23·9 months (range 1·0–37·5 months). No patients were lost to follow up on 30 September 2000, the day on which all centres verified their data on transplant characteristics and each patient's outcome. Categorical data in cross-tabulation tables were compared using Fisher's exact test. OAS, EFS, acute GVHD and relapse after CBT (starting point interval) were evaluated using the Kaplan–Meier product limit method and compared using a log-rank test (Kaplan & Meier, 1958). All variables found to have a P-value of less than 0·1 by the log-rank test were included as binary covariates in Cox's proportional hazards model. Relative risk (RR) for the association between covariates and events was estimated from Cox's model (Cox, 1972). All statistical analyses were carried out with statview version 5·0 software (SAS Institute, Cary, NC, USA).
There was evidence of myeloid engraftment in 32 of the 39 CBT. Graft rejection occurred in two patients, and four died from complications early after CBT without haematopoietic recovery. For one patient, it was too early to evaluate the haematopoiesis after CBT. In the remaining 32, the median time to neutrophil recovery was 27 d (range 13–49 d). The Kaplan–Meier estimate for neutrophil recovery was 83·7 ± 12·2% on d 60. Platelet engraftment was found in 20 of 30 recipients. For platelet recovery, the median time was 54 d (range 18–121 d); and the Kaplan–Meier estimate was 55·4 ± 16·6% on d 100. In univariate analysis (Table II), the factors favourably affecting neutrophil recovery were disease status at CBT (SR group), an HLA genotypic mismatch at no more than two loci, and donor cell doses of more than 3·7 × 107/kg (P = 0·028, 0·037 and 0·027 respectively). On time to platelet recovery, HLA match at less than two loci had an observable effect (P = 0·012). By Fisher's exact test, proportional deviations among HLA disparity, donor cell doses and disease status were not found to be significant (data not shown). From multivariate analysis using Cox's proportional hazard model, disease status at CBT had a favourable influence on the neutrophil recovery after CBT (P = 0·044, RR 2·219, 95% CI 1·022–4·817). Cox's model did not show any important factor affecting time to platelet engraftment.
Table II. Univariate analysis using log-rank test for engraftment, 3 years EFS and relapse rate.
3 year EFS
3 year relapse
% ± 2SE
% ± 2SE
% ± 2SE
% ± 2SE
igh-resolution molecular typing was used for class I (A and B) and class II (DRB1).
†Unit for NCC is 107/kg of recipient body weight.
‡One drug was either methotrexate, cyclosporine A or FK506.
Of the 32 patients who were available for evaluation and had myeloid engraftment, 26 developed acute GVHD: 14 (43·8%) were in grade I, eight (25·0%) in grade II, four (12·5%) in grade III, and one (3·3%) in grade IV. All those with grades I and II acute GVHD developed only skin erythema. Those with grade I acute GVHD recovered without treatment with additional immunosuppressive drugs, and those with grade II acute GVHD recovered with conventional steroids or methylprednisolone pulse therapy (MPT). Two patients developed grade III acute GVHD (erythema and diarrhoea), which remitted with MPT; and one who had HLA two allele mismatched-donor CB died as a result of grade VI acute GVHD. The probability of developing grade II–IV acute GVHD by 100 d after CBT was 59·7% for those given HLA one allele mismatched CB, and 65·3% for those given HLA two or more HLA allele mismatched CB (P = 0·903) (Fig 1). No factors (e.g. recipients' age, weight, disease status, HLA disparity, CMV serology, donor cell doses and type of GVHD prophylaxis) were found to influence the incidence of acute GVHD.
Chronic GVHD affected only 4 of 23 patients who survived more than 100 d post transplantation with evidence of engraftment.
Relapse and death
Nine patients (23·1%) relapsed after CBT: eight were in the HR group at CBT (two ALL and six AML) and one in the SR group (ALL). Sixteen patients (41·0%) died: 10 (25·6%) as a result of transplantation complications (infection, n = 3; interstitial pneumonitis, n = 4; GVHD, n = 1; multiorgan failure, n = 2), and six (15·4%) as a result of relapse after CBT (Table III).
At 3 years after CBT, the probability of relapse was 30·4%. Univariate analysis showed that disease status at CBT had a significant effect on the relapse rate (P = 0·014) (Table II and Fig 2). From the multivariate analysis using Cox's proportional hazard model, disease status at CBT had aggravated the effect of the 3-year relapse rate (P = 0·043, RR 8·592, 95% CI 1·066–69·227).
The 3-year probability of OAS and EFS was 49·2% and 49·8% respectively (Kaplan–Meier estimate, Fig 3). Details on the univariate analysis for EFS are shown in Table II. The factors influencing EFS were disease status at CBT (P = 0·014) (Fig 4 and Table II).
From the multivariate analysis using Cox's proportional hazard model, disease status at CBT had aggravated the effect for the 3-year EFS rate (P = 0·013, RR 4·746, 95% CI 1·382–16·294). Other factors (donor cell doses, and serological and genotypic HLA disparity) did not have a significant influence on EFS.
This study confirmed that CBT from HLA-mismatched unrelated donors is a feasible procedure, and that it is able to produce a significant effect on EFS for children with leukaemia, especially if CBT is given during a favourable phase of the disease. As a result of controllable acute GVHD that occurred early during the course of their disease, children receiving CBT from an HLA-mismatched unrelated donor had a remarkably strong EFS (75·0% at 3 years), and a lower relapse rate (7·2% at 3 years).
Even though a formal, well-matched comparison between the relapse rate of our patients and that observed after marrow transplants from unrelated donors was not performed, the outcomes of our low-risk children are similar to those reported in studies on unrelated HSCT (Casper et al, 1995; Davies et al, 1997; Rubinstein et al, 1998; Locatelli et al, 1999; Thomas, 1999; Gustafsson et al, 2000). In one report, the 2-year EFS of leukaemic children who had received unrelated CBT was 49% in the good-risk group and 8% in the poor-risk group (Locatelli et al, 1999). In a study by a Minneapolis group of unrelated BMT for leukaemic children, the 2-year EFS was 30% and 33% for patients with ALL and AML respectively (Davies et al, 1997). A recently published single-centre study, comparing matched sibling and unrelated-donor BMT in children, showed a remarkably strong probability of survival: 52% for unrelated BMT recipients with haematological malignancies and 77% for ALL patients with unrelated BMT (Hongeng et al, 1997). It remains to be shown using a case–control study whether CBT can produce similar results to that of BMT.
The number of disparities of HLA alleles between the donors and the recipients did not predict the occurrence of acute GVHD in our limited number of patients. The results showed that the incidence of grade II (or higher) acute GVHD was 40·6%, which is not a remarkably low incidence when compared with other studies on HSCT (Casper et al, 1995; Davies et al, 1997; Rubinstein et al, 1998; Sasazuki et al, 1998; Locatelli et al, 1999). However, of our patients, all but one, who had acute GVHD, achieved a good response using first-line therapy and survived. Studies from Seattle and the International Bone Marrow Transplant Registry have shown a decrease in the relapse rates for patients receiving allogeneic transplants compared with the rates in twin transplants, and for allogeneic recipients who developed acute or chronic GVHD compared with those without GVHD. This suggests that genetic disparity and graft-versus-host reaction are associated with an anti-leukaemic effect (Horowitz et al, 1990; Gale et al, 1994). In fact, the differences between the amino acid position within the HLA chains is important for inducing alloreactive cytotoxic T cells (Fleischhauer et al, 1990). Although the group of patients in our study was heterogeneous, the higher incidence but lower morbidity of acute GVHD may suggest the usefulness of HLA-mismatched, unrelated CBT relative to the GVL effect.
In contrast to other studies on the outcome of HSCT for children with leukaemia (Saarinen et al, 1996; Lawson et al, 2000), we did not analyse factors such as tumour burden at diagnosis, karyotype abnormalities and immunophenotype of leukaemic cells because of the small number of patients. This must await further study. Another limitation in this study was a short follow-up period, which was not long enough to clarify the definitive rate of relapse and chronic GVHD after CBT. This also requires further analysis.
In summary, this study supports the conclusion that CBT from an HLA-mismatched, unrelated donor (by molecular typing for both class I and II) is a feasible procedure, and is capable of curing some children with leukaemia who failed to respond to conventional chemotherapy, or who are at risk of a relapse. It is therefore evident from this analysis that if the aim of unrelated CBT is to achieve a definitive cure for leukaemia in children, then this source of HSCT should not be considered as a treatment of last resort for patients with far-advanced or end-stage disease. Rather it should be thought of as a well-established and accepted therapy for those who have achieved a second or third remission after a relapse, and for children who are in first remission with a high risk that the leukaemia will recur. In this regard, CBT offers the advantage of rapid availability, absence of donor risk, and possibly less HLA restriction and less mortality from acute GVHD. For a promising conclusion to unrelated CBT, only analysis of a larger sample, including a comparison with well-matched controls given BMT or PBSCT, is needed to determine the relative risk of a relapse of leukaemia and the long-term outcome of the recipients.
This study was supported in part by a Research Grant for Umbilical Cord Blood Transplantation from the Ministry of Health and Welfare and by a Grant for the Yokohama Foundation for Advancement of Medical Science.
Transplant centres that performed CBT as reported in this article and that produced follow-up reports: Yokohama City University School of Medicine, Department of Paediatrics; The National Cancer Centre, Division of Paediatrics; Kyushu University School of Medicine, Department of Paediatrics; National Kyushu Cancer Centre, Division of Paediatrics; Hirosaki University School of Medicine, Department of Paediatrics; Iwate Prefectural Hospital, Division of Paediatrics; Nihon University Itabashi Hospital, Department of Paediatrics; University of Tokyo School of Medicine, Department of Paediatrics; National Children's Hospital, Division of Haematology/Oncology; Chiba University School of Medicine, Department of Paediatrics; Fukui Medical School, Department of Paediatrics; Kyoto City General Hospital, Division of Paediatrics; Osaka University School of Medicine, Department of Paediatrics; Hiroshima Red Cross Hospital and Atomic-Bomb Survivors Hospital, Division of Paediatrics, Tokai University School of Medicine, Department of Paediatrics; Yamanashi Medical School, Department of Paediatrics; Saitama Children's Medical Centre, Division of Haematology/Oncology; Hamanomachi Hospital, Division of Paediatrics; Ibaragi Children's Hospital, and Kanagawa Children's Medical Centre, Departments of Oncology and Transfusion Medicine.