Factors affecting lymphocyte subset reconstitution after either related or unrelated cord blood transplantation in children – a Eurocord analysis


Dr Tim Niehues, Zentrum für Kinderheilkunde, Klinik für Pädiatrische Hämatologie und Onkologie, Heinrich Heine Universität (HHU), Moorenstr. 5, 40001 Düsseldorf, Germany. E-mail: niehues@uni-duesseldorf.de


Immune recovery after cord blood transplantation (CBT) is of concern owing to the low number of lymphocytes transferred with the graft and their immaturity. Risk factors influencing lymphocyte subset reconstitution related to disease, patient, donor and transplant were studied in 63 children (< 16 years), given either related (n = 14) or unrelated (n = 49) CBT for malignant (n = 33) or non-malignant diseases (n = 30). Only children with sustained myeloid engraftment were analysed. Absolute numbers of T (CD3+, CD4+, CD8+), B and natural killer (NK) cells were reported 2–3, 6, 9, 12 and 12–24 months after CBT. Median patient age was 4·0 years (0–15) and median follow-up was 23 months (1·7–61·0). Twenty-six patients received human leucocyte antigen (HLA)-matched CBT and 37 received HLA-mismatched CBT. The median number of nucleated cells (NCs) collected/recipient weight was 6·1 × 107/kg. In this selected population, the estimate 2 year survival was 85%. Lymphocyte reconstitution (defined as the median time to reach the normal value of age-matched healthy children) was 3, 6 and 8 months for NK, B and CD8+ cells, while it was 11·7 months for both CD3+ and CD4+ lymphocytes. In the multivariate analysis, factors favouring T-cell recovery were: related donor (P = 0·002); higher NCs/kg (P = 0·005) and recipient cytomegalovirus (CMV)-positive serology (P = 0·04). Presence of acute graft-versus-host disease (GVHD) delayed T-cell recovery (P = 0·04). To summarize, in children with sustained myeloid engraftment the concern that lymphocyte recovery after CBT could be delayed does not appear to be substantiated by our results.

Cord blood (CB) can be used as a stem cell source for transplanting patients with haematological malignant disease, bone marrow failure syndromes, haemoglobinopathies, immunodeficiencies and inborn errors of metabolism (Kurtzberg et al, 1996; Wagner et al, 1996; Gluckman et al, 1997; Rubinstein et al, 1998; Locatelli et al, 1999). The immaturity and the unique immunological properties of CB cells have been associated with a relatively low incidence of graft-versus-host disease (GVHD) after cord blood transplantation (CBT) (Madrigal et al, 1997; Rocha et al, 2000). In view of the reduced incidence and severity of GVHD after CBT, placental blood can be used as an alternative stem cell source in the human leucocyte antigen (HLA)-identical sibling setting and, more importantly, when there is no allogeneic bone marrow donor with an acceptable degree of compatibility (Rocha et al, 2000, 2001). Other important sources for stem cell transplantation in paediatric practice are family mismatched or unrelated bone marrow donors, as well as peripheral stem-cell donors (Sanders, 1997).

Immune recovery and T-cell reconstitution in particular are of central importance for the occurrence of relapse, GVHD and infectious complications after stem cell transplantation. The current model of T-cell reconstitution after conventional bone marrow transplant (BMT) relies on two pathways (Mackall et al, 1995; Mackall & Gress, 1997; Parkman & Weinberg, 1997): a thymus-dependent pathway which resembles the ontogeny, and a thymus-independent generation of T cells which involves expansion of mature donor T cells transferred with the graft (peripheral expansion). With respect to the thymus-independent pathway, the quality and quantity of donor lymphocytes transplanted with bone marrow (BM) and CB clearly differ. CB lymphocytes have repeatedly been characterized as immature and therefore a less efficient expansion of donor CB T cells in the thymus-independent pathway may be expected. Apart from qualitative differences, there is also a quantitative difference, as CBT recipients receive approximately 1 log less T cells than patients given BMT.

We addressed the following questions to centres performing CBT: is there long-term reconstitution of normal lymphocyte numbers? What time period is needed for lymphocyte recovery after CBT? Which factors may influence lymphocyte reconstitution? In order to answer those questions, we analysed immune reconstitution in 63 children with sustained myeloid engraftment reported to Eurocord with a median follow-up time of almost 2 years. This large series allowed us to perform a multivariate analysis in order to identify independent risk factors predicting lymphocyte recovery after CBT.

Patients and methods

Patient and transplant characteristics Data on lymphocyte reconstitution from 71 patients given either related or unrelated CBT were reported to Eurocord. Inclusion criteria were: age < 16 years and sustained donor myeloid engraftment after d +60. Exclusion criteria were age > 16 years, graft failure, autologous reconstitution after CBT or donor lymphocyte infusion after related CBT. Administration of cytokines in vivo or ex vivo was not considered as an exclusion criterion. Eight patients were excluded on the basis of age > 16 years (n = 2), incomplete and non-valid data (n = 1), or non-engraftment of donor haematopoiesis (n = 5). Sixty-three children receiving CBT between January 1995 to January 2000 in 13 centres and with a median follow-up of 23 months (range: 1·7–61·0) were analysed.

Patient and transplant characteristics are summarized in Table I. Cytomegalovirus (CMV) serological status was studied before transplantation in 61 out of 63 children. CB units were provided by cord blood banks in New York (n = 21), Düsseldorf, Germany (n = 12), Milan, Italy (n = 10), Barcelona, Spain (n = 3), Paris, France (n = 2), Louvain, Belgium (n = 2), and local banks, mainly for related transplants (n = 13). CB units were collected, processed and thawed following previously published procedures (Rubinstein et al, 1995; Kögler et al, 1996). The method of counting the number of granulocyte–macrophage colony-forming units (CFU-GM) and CD34+ cells collected and infused varied considerably between centres and was therefore not included in the analysis. HLA typing of recipients and CB units was performed by standard National Institutes of Health (NIH) microlymphocytotoxicity and low-resolution generic oligotyping for DRB1 antigens. GVHD prophylaxis and conditioning regimens varied according to centre policy, previous treatments, original disorder and disease status at time of transplantation (Table I).

Table I.   Patient and transplant characteristics.
  1. †Diagnoses in detail; malignancies: acute lymphoblastic leukaemia, 19; acute myeloid leukaemia, 3; solid tumours, 2 (non-Hodgkin's lymphoma, neuroblastoma); chronic myeloid leukaemia, 2; myelodysplastic syndrome, 4; haemophagocytic lymphohistiocytosis, 3; bone marrow failure syndromes, 6; primary immune deficiencies, 14 (TB, 1; TB+, 5; Omenn syndrome, 1; Zap deficiency, 2; leucocyte adhesion deficiency, 1; Wiskott–Aldrich syndrome, 2; X-linked lymphproliferative,1; other severe combined immunodeficiency, 1); metabolic diseases, 4 (osteopetrosis, 2; Hurler,1; adrenoleucodystrophy, 1); haemoglobinopathies, 6.

  2.    ‡Good prognosis is considered 1st and 2nd complete remission (CR) for acute leukaemia (and 1st chronic phase for CML), poor prognosis all other status and diagnoses.

  3.    *CMV serology was only available in 61 patients.

  4.    §CSA, cyclosporine A; MTX, methotrexate; FK506, tacrolimus.

 Median age, years (range):4·0 (0–15) 
 Median weight, kg (range):16·9 (4–66) 
 Positive CMV serology3152·5*
  Good prognosis1648·5
  Poor prognosis1751·5
 Bone marrow failure syndromes69·5
 Immunodeficiencies and metabolic diseases1828·6
 ABO matched812·7
 HLA matched2641·3
 HLA (low resolution)
  1 difference2844·4
  2 differences914·3
 Median number (range) of nucleated cells infused (n = 61):4·7 × 107/kg (0·9–19) 
 Median number (range) of nucleated cells collected (n = 55):6·1 × 107/kg (1·9–25) 
GVHD prophylaxis§
 CSA alone1625·4
 CSA + Prednisone2641·3
 CSA + MTX1117·5
 MTX + FK5061015·9
Preparative regimen:
 Total body irradiation-containing regimen2031·7
 Busulphan + Endoxan1015·9
 Prophylactic growth factors (starting between d 0–7)4074·1

Determination of lymphocyte subsets and kinetics of lymphocyte reconstitution Lymphocyte subpopulations in peripheral blood after CBT were analysed using flow cytometry with monoclonal antibodies for classic T- (anti-CD3, CD4 and CD8), B- (anti-CD19) and natural killer (NK)-cell antigens (anti-CD56, CD16). Absolute numbers were calculated using the percentages of CD3, CD4, CD8, CD19 and CD56/CD16 staining, and the lymphocyte count of the same sample. The number of patients analysed at each time point varied. The latest evaluation of lymphocyte number available at 2–3 (n = 38), 6 (n = 41), 9 (n = 20), 12 (n = 42) and 12–24 months (n = 20) was used for analysis. The mean number of lymphocyte subset counts per patient was 3·35 (25th percentile: 2; 75th percentile: 5).

Definitions and statistical analysis Standard definitions were used to describe overall survival, relapse, event-free survival (EFS), acute GVHD (aGVHD) and chronic GVHD (cGVHD), as published in previous Eurocord analyses (Gluckman et al, 1997; Locatelli et al, 1999). Grades > II were counted as acute GVHD. White blood cell (WBC) engraftment was defined as the first of three consecutive days when the absolute neutrophil count was > 0·5 × 109/l with evidence of donor haematopoiesis, and platelet engraftment as the time to reach a sustained platelet count > 20 × 109/l in the absence of platelet transfusions for seven consecutive days.

As normal values in childhood vary considerably with age, absolute numbers of CD3+, CD4+, CD8+, B and NK cells were related to age-specific normal values, taken from a study on paediatric reference values for blood lymphocytes (Comans-Bitter et al, 1997). The endpoint for the determination of lymphocyte reconstitution was the median time needed to reach the age-related normal median value. Kaplan–Meier estimates were employed for calculating reconstitution of NK, B and T cells, as well as for CD4+ and CD8+ lymphocyte subpopulations.

All survival function estimates and time to immune reconstitution cumulative incidence function estimates were based on Kaplan–Meier estimates (Kaplan & Meier, 1958). Survival estimates were compared according to levels of qualitative variables with the log-rank test, whereas associations between quantitative variables or dichotomized quantitative variables and outcomes were assessed using the Cox proportional hazard model and the Wald test (Cox, 1972). Cox proportional hazard models were also used to assess the relationship between covariates and time to lymphocyte reconstitution. Occurrences of acute and chronic GVHD were regarded as time-dependent covariates. Multiple Cox proportional hazard models were carried out to identify factors independently associated with immune recovery. All variables achieving statistical significance at a 10% level in the univariate analyses were simultaneously considered in the multivariable models: age as a dichotomized variable (> 5 years), CMV status, sex, disease (malignancy, bone marrow failure syndrome, immunodeficiency and metabolic disease, haemoglobinopathy), related/unrelated donor, HLA match, ABO match, cell dose and aGVHD as a time-dependent covariate. A variable selection procedure based on Hurvich and Tsai criterion was implemented to remove non-significant variables from the multiple models (Hurvich & Tsai, 1989). The validity of the proportional hazards assumption was tested using the Grambsch and Therneau goodness-of-fit test (Grambsch & Therneau, 1994). All tests were two-sided, with a significance level of 5%. All analyses were performed using the SPlus 2000 (MathSoft, Seattle, WA, USA) software package.


Outcomes, haematopoietic engraftment and GVHD

In this selected population, the estimated 2 year survival was 80% for malignant diseases and 93% for non-malignant diseases. Deaths were owing to relapse or disease progression (n = 5) or transplant-related causes (n = 3; one interstitial pneumonitis, one bacterial infection and one fungal infection). The median time to reach neutrophil and platelet engraftment was 21 d (range 8–50) and 47 d (range: 39–67; n = 58) respectively. Neutrophil recovery is delayed after HLA-identical CBT or unrelated CBT compared with HLA-identical or unrelated bone marrow transplants (Rocha et al, 2000, 2001). At 100 d after CBT, aGVHD had occurred in 30·3% of patients (Grade II: n = 13, III: n = 4, IV: n = 2), and at 2 years, cGVHD had occurred in 10·7% of patients enrolled in this study, which are somewhat higher percentages than in T-depleted unrelated BMT, but lower percentages than non-manipulated unrelated BMT (Rocha et al, 2001).

Lymphocytes subset recoveries

B- and NK-cell reconstitution kinetics B and NK cells recovered quickly in the majority of individuals (Fig 1A and B). Interestingly, the number of NK cells did not change considerably in the post-transplant period. Kaplan–Meier estimates documented that the median time for B- and NK-cell reconstitution was 5·9 months (CI 95%: 5·8; 6·5) and 3·0 months (CI 95%: 2·3; 3·3) respectively.

Figure 1.

 Kaplan–Meier estimates: time to T-, B- and NK-cell reconstitution (A), and CD4+ and CD8+ cell reconstitution (B).

T-cell reconstitution kinetics The increase of CD3+, CD4+ and CD8+ cells after CBT was slow (Fig 1A and B). While CD4+ cells showed a recovery very similar to total T cells (CD3+), the reconstitution of CD8+ cells was somewhat faster. Kaplan–Meier estimates showed that the median time for both CD3+ and CD4+ cell reconstitution was 11·7 months (CI 95%: 9·2; 12·1), while the median time for CD8+ cell reconstitution was 7·9 months (CI 95%: 6·0; 12·0).

Factors affecting lymphocytes subset recoveries

The results of multivariate analysis regarding the influence of patient-, disease-, donor-and transplant-related variables on time to lymphocyte subset reconstitution are summarized in Table II.

Table II.   Multivariate analysis.
Lymphocyte subset recoveryParameterHR95% CIP valueFavourable
CD3 cellsCMV status recipient2·20[1·08; 4·46]0·03CMV positive
Donor related/unrelated0·24[0·095; 0·60]0·002Related
Cell dose1·12[1·03; 1·21]0·005High number of NCs/kg collected
aGVHD0·44[0·20; 0·96]0·04No aGVHD
CD4 cellsaGVHD0·41[0·20; 0·86]0·02No aGVHD
CD8 cellsCMV status recipient1·96[1·00; 3·84]0·048CMV positive
Donor related/unrelated0·22[0·10; 0·51]0·0004Related
Cell dose1·10[1·03; 1·17]0·006High number of NCs/kg collected
NK cellsAge (years)1·83[0·99; 3·38]0·054Age > 5
CMV status recipient3·40[1·21; 9·55]0·02CMV positive

B- and NK-cell reconstitution NK-cell reconstitution was observed to be faster in children with haemoglobinopathy [Hazard ratio (HR): 3·40; 95% CI: 1·21, 9·55; P = 0·02] and in children older than 5 years (HR: 1·83; 95% CI: 0·99, 3·38; P = 0·054). None of the variables showed a significant influence on B-cell recovery.

T-cell reconstitution: patient-related factors Interestingly, age did not affect T-cell recovery, either as a continuous or as a dichotomized variable. A positive CMV serology favoured T-cell recovery in general and the recovery of CD8+ cells (CD3+ cell recovery: HR: 2·20; 95% CI: 1·08–4·46; P = 0·03 and CD8+ cell recovery: HR: 1·96; 95% CI: 1·00, 3·84; P = 0·048).

Disease-related factors T-cell recovery was not influenced by diagnosis.

Donor-related factors In multivariate analysis, unrelated CBT recipients had a worse T-cell reconstitution (CD3+ recovery: HR: 0·24; 95%CI: 0·095–0·60; P = 0·002). While CD8+ cell reconstitution was delayed in CBT performed using an unrelated donor (HR: 0·22; 95% CI: 0·10–0·51; P = 0·0004), the recovery of CD4+ cells appeared not to be affected either by unrelated donor or by donor/recipient HLA disparity.

Transplant-related factors While conditioning regimens and GVHD prophylaxis used in this study did not affect T-cell recovery, occurrence of aGVHD clearly had an adverse effect on CD3+ cell recovery in multivariate analysis as a time-dependent covariate (P = 0·04; HR: 0·44; 95% CI: 0·20, 0·96). Interestingly, aGVHD affected CD4+ cell recovery (HR: 0·41; 95% CI: 0·20, 0·86; P = 0·02) rather than CD8+ cell reconstitution. The number of NCs collected in the graft (as a continuous variable) correlated significantly with better CD3+ cell (HR: 1·12; 95% CI: 1·03–1·21; P = 0·005) and CD8+ cell recovery (P = 0·006; HR: 1·10; 95% CI: 1·03, 1·17). However, this positive correlation was not found for the number of cells infused.


In the present analysis, we investigated which factors influence lymphocyte reconstitution after CBT. When interpreting our data, some important caveats have to be kept in mind. Patients were selected on the basis of myeloid engraftment, which excluded patients with non-engraftment or autologous reconstitution and patients who died within the first 2–3 months after CBT. Although we report a series with a high number of patients and a long observation period of lymphocyte recovery, it needs to be noted that the number of patients analysed at each time point after CBT varied. The minimum number of patients analysed at one time point was 20 patients (at 12–24 months). Owing to the nature of this analysis, using registry data, it also has to be taken into account that conditioning regimens, supportive care, GVHD prevention, as well as treatment and patient selection, differed between centres. Similarly, the determination of lymphocyte subset counts may have varied. However, we do not believe that this resulted in significant variations, as in each institution standardized flow cytometry analyses were performed by laboratories specialized in immunophenotyping blood samples of children before and after stem cell transplantation. Lastly, it has to be pointed out that lymphocyte subset reconstitution in this analysis represents numerical reconstitution as there were no data on lymphocyte function and receptor repertoire available.

Among the patient characteristics influencing lymphocyte recovery, the analysis reveals interesting results: an influence of age on T-cell reconstitution has been postulated because of the age-related decrease in thymic function (Mackall et al, 1995; Mackall & Gress, 1997; Parkman & Weinberg, 1997). However, as in T cell-depleted (TCD) BMT (Kook et al, 1996), this was not observed in the present study. An influence of age may become apparent in a group with more adolescents and adults. A positive CMV status of the recipient favoured T cell and, in particular, CD8+ cell recovery. This has to be interpreted with caution: negative IgG titres of a recipient may not reflect CMV status precisely (e.g. in the immunosuppressed), as some seronegative children may indeed be positive using PCR. Still, a correlation of T cell and, in particular, CD8+ cell recovery with CMV seropositivity has been previously observed in children given TCD BMT and in adult studies (Kook et al, 1996; Behringer et al, 1999). In both studies as well as in our analysis, the reason for a favourable influence of CMV on T-cell recovery is not clear. It would be interesting to evaluate whether rapid T-cell recovery may be induced by CMV itself by showing that lymphocyte reconstitution is donor-derived and that a large fraction of T cells after CBT responds to CMV antigen/peptide. Unfortunately, in our study chimaerism data were not available for the majority of the patients.

We observed that, among donor characteristics, unrelated donor clearly had an adverse effect on T-cell reconstitution, mostly on the CD8+ subset. While it may have been expected that delayed recovery of the CD8+ subset is associated with HLA disparity, HLA matching did not appear as an independent variable for T- and CD8+ cell reconstitution. The different use of GVHD prophylaxis in related and unrelated donors may have affected lymphocyte recovery. However, among the transplantation characteristics tested in this analysis, neither different GVHD prophylaxis nor conditioning regimens were found to significantly influence lymphocyte recovery. This result may be influenced by different transplantation protocols between centres as well as different underlying diseases. Concerning graft characteristics, there was a positive correlation of T-cell recovery with the number of graft NCs/kg collected, but no correlation with the number of NCs infused. Previous Eurocord Registry studies have shown that the number of cells infused is the most important factor for haematopoietic recovery (Gluckman et al, 1997; Locatelli et al, 1999; Rocha et al, 2001). After thawing, there is a median loss of 20% of the cells collected (unpublished observations). In our study, the lack of correlation between cells infused and haematopoietic recovery may be related to the small number of patients and to the selection criteria. A positive correlation of T-cell recovery with NCs in the graft would suggest that (similar to BMT), despite their presumed immaturity, CB T cells and particularly CD8+ cells can be peripherally expanded in vivo. To determine the role of peripheral expansion after CBT precisely, routine T-cell counts in CB grafts before and after thawing would have been necessary, but there were no data available.

In previous analyses, occurrence of aGVHD has been shown to adversely influence immune reconstitution in BMT (Lum, 1987). This was confirmed for CBT in our analysis as acute GVHD was associated with impaired T-cell recovery, particularly with CD4+ cell reconstitution. It has been hypothesized that damage to the thymic epithelium by acute GVHD may affect CD4+ cell differentiation specifically (Seddik et al, 1984).

The mechanisms for reconstitution of NK and B cells are poorly characterized. NK-cell engraftment in CBT reported by others (Abu-Ghosh et al, 1999; Knutsen & Wall, 1999; Gireaud et al 2000; Thomson et al, 2000) and in our cohort shows evidence of a fast engraftment at 3 months after CBT, similar to that of BMT recipients (Lum, 1987). There was fast B-cell recovery, as described by others (Locatelli et al, 1996; Abu-Ghosh et al, 1999; Knutsen & Wall, 1999; Thomson et al, 2000). B-cell recovery may be associated with a high number of B-cell precursors present in CB (Arakawa-Hoyt et al, 1999). B-cell as well as T-cell recovery is probably functional, as normal levels of IgM and IgA and a normal response of lymphocytes to T- and B-cell mitogens have been demonstrated after CBT (Locatelli et al, 1996; Abu-Ghosh et al, 1999; Knutsen & Wall, 1999; Thomson et al, 2000).

Most data on immune reconstitution after stem cell transplantation in children are on TCD bone marrow grafts (Cowan et al, 1987; Daley et al, 1987; Foot et al, 1993; Kook et al, 1996; Small et al, 1999). The detailed analysis of Kook et al (1996) on immune reconstitution after TCD BMT in 102 children reported normalization (defined as > 5th percentile) of CD3+ cells within 18–36 months. At the latest time point (18–24 months) in the analysis of 42 children by Small et al (1999), there were also several children with T-cell counts well below 1 × 109/l (percentiles not given). In our analysis of lymphocyte recovery after CBT, we found that, within 18–24 months, all patients were above the 5th percentile, which is similar, if not faster, than TCD BMT in the large studies (Kook et al, 1996; Small et al, 1999). The kinetics of lymphocyte recovery after CBT has systematically been reported in smaller cohorts and mostly in CBT with unrelated donors (Abu-Ghosh et al, 1999; Knutsen & Wall, 1999; Gireaud et al, 2000). Most recently, Thomson et al (2000) studied 27 patients, 25 of whom had leukaemia, who received CBT: fast NK (2 months) and B (6 months) and slower CD4+ (12 months) and CD8+ cell (6 months) recovery were observed. In contrast, a relatively fast T-cell reconstitution (> 5th percentile) within 6 months was observed in three children transplanted with CB of related donors for acute lymphoblastic leukaemia (ALL) (Locatelli et al, 1996).

What are the differences in lymphocyte recovery between CBT and BMT? The most recent case–control study by Moretta et al (2001) shows that immune recovery after either related or unrelated CBT gives similar results to matched related or unrelated BMT. Much faster recovery of CD8+ than CD4+ cells with a characteristic inversion of the CD4/CD8 ratio has been repeatedly observed after BMT (Lum, 1987). This has been attributed to peripheral expansion of mature CD8+ cells transplanted with the graft. Even in TCD allogeneic BMT, reconstitution of CD8+ cells is faster than that of CD4+ cells (Kook et al, 1996; Small et al, 1999). In CBT, however, our data and that of others (Abu-Ghosh et al, 1999; Gireaud et al, 2000; Thomson et al, 2000) do not show the characteristic inversion of the CD4:CD8 ratio seen after BMT. At present, it is unclear to what extent a reduced CD8+ cell number after CBT may affect anti-infective and anti-tumour immunity.

In conclusion, a CMV-seropositive recipient, a related donor with a high number of graft NCs collected and lack of acute GVHD after CBT are the most important predictors for fast T-cell recovery after CBT in children. The concern that lymphocyte recovery after CBT could be delayed appears not to be substantiated.


This work was supported by the Elterninitiative der Kinderkrebsklinik e. V. Düsseldorf, an EEC grant for Eurocord BIOMED II contract QLRT-1999–00380, and by grants from IRCCS Policlinico San Matteo to Franco Locatelli.

We would like to thank the participants R. Pasquini, Curitiba, Brazil, B. Brichard, Brussels, Belgium, S. Fisher, RN, Los Angeles, USA, A. Nagler, Jerusalem, Israel, A. Busca, Torino, Italy, C. Urban, Graz, Austria, W. Nürnberger, Düsseldorf, Germany, J. Stary, Prague, Czech Republic, O. Feyen for help in the data analysis, S. Özbek for help in the preparation of the manuscript, doctors F. Garnier and I. Ionescu, and data managers in centres for collecting data.