Improving outcomes of cord blood transplantation: HLA matching, cell dose and other graft- and transplantation-related factors


Vanderson Rocha, MD, PhD, Eurocord-Netcord-EBMT office, Institut Universitaire, Hôpital Saint Louis, Université de Paris 7, 1 Av Claude Vellefaux, 75010 Paris, France. E-mail:


The use of unrelated umbilical cord blood (UCB) as an alternative source of haematopoietic stem cells transplantation (HSCT) has been widely used for patients lacking a human leucocyte antigen (HLA) matched donor. One of the disadvantages of using UCB is the limited number of haematopoietic stem cells and, consequently, delayed engraftment and increased risk of early mortality. Many approaches have been investigated in the attempt to improve engraftment and survival. Among those, studies analysing prognostic factors related to patients, disease, donor and transplantation have been performed. Variable factors have been identified, such as factors related to donor choice (HLA, cell dose and others) and transplantation (conditioning and graft-versus-host disease prophylaxis regimens). This review will focus on the interactions between HLA, cell dose and other modifiable factors related to the UCB unit selection and transplantation that may improve outcomes after UCB transplantation.

Umbilical cord blood transplantation (UCBT) has extended the availability of allogeneic haematopoietic stem cell transplantation (HSCT) to patients who would otherwise not be eligible for this curative approach. Progress in the field of UCBT parallels the huge interest in establishing and developing cord blood banks (CBB) worldwide. Today, more than 450 000 cord blood (CB) grafts are available in more than 50 CBB and we estimate that more than 20 000 unrelated UCBT have been performed worldwide.

In comparison to other sources of allogeneic HSCT, umbilical cord blood (UCB) offers substantial logistic and clinical advantages (Rocha & Locatelli, 2008) including: (i) significantly faster availability of banked cryopreserved UCB units, with patients receiving UCBT a median of 25–36 d earlier than those receiving an unrelated bone marrow graft; (ii) extension of the donor pool due to tolerance of 1–2 human leucocyte antigen (HLA) mismatches out of six; (iii) lower incidence and severity of acute graft-versus-host disease (GVHD); (iv) lower risk of transmitting infections by latent viruses, such as cytomegalovirus (CMV) and Epstein–Barr virus (EBV); (v) lack of risk to the donor; (vi) higher frequency of rare haplotypes compared to bone marrow registries, because it is easier to target ethnic minorities. However, the main problem of using UCB for transplantation is the low number of haematopoietic progenitor cells and HSCs in UCB compared with bone marrow or mobilized peripheral blood stem cells (PBSC), which translates into increased risk of graft failure, delayed haematopoietic engraftment (Rocha et al, 2000, 2001; Rocha et al, 2004a; Laughlin et al, 2004; Eapen et al, 2007) and delayed immune reconstitution (Niehues et al, 2001; Komanduri et al, 2007).

In order to improve engraftment, decrease the period of aplasia and improve survival after UCBT, many approaches have been investigated, such as the use of double unit UCBT (Barker et al, 2005; Brunstein et al, 2007), intrabone injection of CB cells (Frassoni et al, 2008), use of reduced intensity conditioning (RIC) regimen (Barker et al, 2003; Brunstein et al, 2007; Ballen et al, 2007; Rocha et al, 2009); transplantation of in vitro amplified CB cells (de Lima et al, 2008) and co-infusion with a haploidentical T cell depleted graft (Fernández et al, 2003; Bautista et al, 2009), or mesenchymal stem cells (Macmillan et al, 2009). With regard to these approaches, many prognostic studies for outcomes after UCBT have been performed, analysing factors related to patients, disease, donor and transplantation (Gluckman et al, 1997; Rubinstein et al, 1998; Wagner et al, 2002; Laughlin et al, 2001; Sanz et al, 2001; Gluckman et al 2004; Arcese et al 2006; Kurtzberg et al 2008). Variable factors have been identified, such as HLA, cell dose and others related to the choice of graft or factors related to the conditioning regimen (Nabhan et al, 2008) or GVHD prophylaxis (Locatelli et al, 2003).

This review will focus on the interactions between HLA, cell dose and other variable factors related to the donor choice that can guide clinicians in order to improve engraftment and increase early mortality after UCBT.

Impact of HLA and cell dose in myeloablative single unit UCBT from unrelated donors in children and adults: trying to establish an algorithm

Retrospective and prospective studies have shown that unrelated donor UCBT: (i) reconstitutes haematopoiesis and achieves sustained engraftment, but with delayed myeloid recovery; (ii) is associated with a low incidence of graft-versus-host disease (GVHD); (iii) does not result in a higher relapse risk and has similar survival rates to other sources of allogeneic HSCT. Almost all series concerning UCBT in children and adults from unrelated donors have demonstrated the profound impact of cell dose, measured as pre-freeze or infused total nucleated cells (TNC), colony-forming cells, and CD34+ cells, on engraftment, transplant-related events and survival (Rocha et al, 2004b). HLA matching was also recognized in the earlier series as an important factor for engraftment; however its impact on long term survival, mainly in patients with malignant diseases is still controversial.

In 1997, the Eurocord group first described the association of TNC dose and HLA match with neutrophil and platelet recovery and survival after UCBT, in 143 patients, mostly children, given a related and unrelated UCBT (Gluckman et al, 1997). The best cut-off value associated with a higher probability of neutrophil and platelet recovery and improved survival rate was the median infused TNC dose (3·7 × 107/kg). In this series, only 20 adults were transplanted with a CB unit, and 1-year survival was only 16%. A closer HLA match (defined as matched or mismatched based on HLA-A and -B low resolution and HLA-DRB1 high resolution typing) was also associated with better engraftment and survival, but due to the small number of patients, the number of HLA disparities associated with outcomes was not studied (Gluckman et al, 1997). These results have since been confirmed in a series of 562 children and adult unrelated CBT recipients (Rubinstein et al, 1998). The authors found that higher cell dose, and number of HLA disparities (6/6; 5/6 or 4/6, considering same above HLA definition) were independent factors associated with better engraftment and decreased transplantation-related mortality. The impact of TNC dose in the graft (pre-freeze) was analysed using four categories of cell dose and an arbitrary cut-off point chosen. It was shown that 1-year mortality was 70% for patients transplanted with a UCB graft containing a pre-freeze cell dose <2·4 × 107/kg. Also, 1-year mortality was increased 2·0 times after UCBT for patients given a two or 3/6 HLA disparate graft. In this cohort only 18% were adults who experienced worse outcomes after UCBT.

With these two important studies, it was clear that HLA matching and cell dose were crucial factors for improving outcomes after UCBT and, in all probability, the number of TNC collected or infused should not be <2·5 or 2·0 × 107/kg (considering a TNC loss of around 20%). Also the number of HLA disparities should not be exceed three out of six (following the above definition). Results of patients given a two of six HLA disparities showed 1-year mortality of 60% that it was still acceptable in this high risk group of patients. Other factors that were associated with outcomes in both studies (Gluckman et al, 1997; Rubinstein et al, 1998) were related to patient factors (CMV serology), disease (acute leukaemia and status at transplant) and location of transplant centre.

Between 1994 and 2004, progress in the field of CBT was largely restricted to children, mainly because of the impact of cell dose on engraftment. Almost all studies in children showed that a higher cell dose, established by the median nucleated cell dose cut-off value (as high as 5·2 × 107/kg), was associated with better outcomes (Michel et al, 2003). Also, the Minnesota group reported that both CD34+ cell dose (more than 1·7 × 105/kg, percentile cut-off value) and number of HLA disparities (6/6 and 5/6 vs. 4/6 or more) were associated with better engraftment and decreased transplant-related mortality (TRM) in 102 UCBT recipients (mostly children) (Wagner et al, 2002). Based on these results in children, we can also suggest that the TNC dose should be as high as possible, and should be <2·5 × 107/kg at collection. HLA matching is also important in children, and one should avoid 3/6 HLA disparities. Results of outcomes after UCBT in children with haematological malignancies have improved over the recent years as a result of higher numbers of TNC in CB grafts, lower number of HLA disparities, transplantation in early or intermediate phase of the diseases and, of course, other factors, such as supportive care (better infections surveillance and treatment) and centre experience (Fig 1A, B).

Figure 1.

 (A) Survival after single unrelated UCBT for children with haematological malignancies by transplant period. OS, overall survival. (B) Median of collected and infused total nucleated cell dose (107/kg), number of HLA disparities and status of the disease in children with haematological malignancies transplanted with a single unrelated UCBT according to time of transplant. TNC, total nucleated cells; Adv. dis., advanced disease.

The improvement in paediatric UCBT results is clearer in adults, where the cell dose effect is more evident. Outcomes after single unit UCBT in adults have also improved in recent years (Fig 2A, B) due to the same reasons described above. Some studies have reported the outcomes and risk factors of single unit unrelated UCBT in adults (Laughlin et al, 2001; Sanz et al, 2001; Ooi et al, 2003, 2004; Takahashi et al, 2004; Rocha et al, 2004a; Laughlin et al, 2004; Arcese et al, 2006). As expected from retrospective and multicentre studies, the series were heterogeneous in terms of recipients and disease-related characteristics, such as type and status of the disease at transplant. The myeloid engraftment rate at 60 d ranged from 72 to 100% and the probability of platelet engraftment at 180 d was 65–90%. Median time to achieve a neutrophil count above 0·5 × 109/l varied from 22 to 32 d.

Figure 2.

 (A) Survival after single unrelated UCBT for adults with haematological malignancies by transplant period. OS, overall survival. (B) Median of collected and infused total nucleated cell dose (107/kg), number of HLA disparities and status of the disease in adults with haematological malignancies transplanted with a single unrelated UCBT according to time of transplant. TNC, total nucleated cells; Adv. dis., advanced disease.

The impact of cell dose was also found to be associated with engraftment or survival in different cohorts of UCBT in adults after myeloablative conditioning regimen. However, unexpectedly, the number of HLA disparities was not associated with any outcome. This observation was probably related to the small number of patients reported. Laughlin et al (2001) described the association of TNC and CD34+ cell dose with engraftment and survival in 68 UCBT in adults. The median number of TNC collected and infused and CD34+ cells infused were 2·1, 1·6 × 107/kg and 1·2 × 105/kg, respectively. Delayed engraftment was observed in 23 patients transplanted with a graft containing <1·9 × 107/kg compared to those (n = 21) receiving more than 1·9 but <2·4 × 107/kg or those given more than 2·4 × 107/kg (n = 24). Interestingly, a higher number of CD34 cells infused (median; >1·2 × 105/kg) was associated with improved event-free survival (EFS) rate; however it was not associated with TNC dose infused. There was a trend towards better survival in patients given a more HLA compatible graft (Laughlin et al, 2001). Recently, another Eurocord study also showed the association of CD34+ cell dose with engraftment after single and double UCBT in a group of 104 patients with lymphoid malignancies (Rodrigues et al, 2009). This study included single and double UCBT, RIC and myeloablative conditioning regimens. The cumulative incidence (CI) of neutrophil engraftment was 85% by day 60, with higher engraftment rate in recipients of higher (>1·0 × 105/kg) CD34+ per kg cell dose (P = 0·0001) but it was not associated with decreased non-relapse mortality (NRM) or overall survival. The number of HLA disparities was not associated with any outcomes.

We have published risk factors for outcomes after UCBT based on 171 adults with haematological malignancies given a single CB unit after myeloablative conditioning between 1998 and 2003 (Arcese et al, 2006). Ninety-one patients (53%) were transplanted with advanced disease and an autologous transplant had failed in 32 (19%). Most patients (87%) received an HLA-mismatched CB unit with 1–2 HLA disparities. The median number of TNC and CD34+ cells infused was 2·1 × 107/kg and 1 × 105/kg, respectively. The CI of neutrophils recovery at day 60 was 72 ± 3%, with a median of 28 d (11–57). A higher TNC cell dose (>2·0 × 107/kg) and use of haematopoietic growth factors were independently associated with faster neutrophils recovery (Arcese et al, 2006). However, TNC dose and number of HLA disparities were not associated with TRM, survival or disease-free survival (DFS).

We have observed in recent retrospective studies that the impact of TNC dose on outcome after UCBT in adults was not as ‘strong’ as it was before 2000, when the median TNC dose infused was 1·7 × 107/kg compared to 2·4 and 2·7 × 107/kg in recent years (Arcese et al, 2006; Eurocord, unpublished observations) (Fig 2B).

Impact of HLA and cell dose in RIC regimen prior to single or double UCBT from unrelated donors in adults

The number of adults patients transplanted with UCBT has increased following the use of RIC regimen and double CB transplants. Very few data has addressed the issue of cell dose and number of HLA disparities in this setting. The Minnesota group has evaluated the efficacy of UCB in the setting of a non-myeloablative regimen consisting of fludarabine (200 mg/m2), cyclophosphamide (50 mg/kg), and a single fraction (200 cGy) of total body irradiation (TBI) with ciclosporin (CsA) and mycophenolate mofetil (MMF) for post-transplantation immunoprophylaxis (Brunstein et al, 2007). The target cell dose for the UCB graft was 3·0 × 107 nucleated cells/kg, resulting in the selection of a second partially HLA-matched UCB unit in 85%. One hundred and ten patients with haematological diseases were enrolled. Neutrophil recovery was achieved in 92% at a median of 12 d. Neither TNC, CD34+ and CD3+ cell doses, HLA matching, nucleated cell viability, ABO typing, gender match, or order of unit infusion was predictive of engraftment or which unit eventually dominated. TRM was 26% at 3 years. Survival and EFS at 3 years was 45% and 38%, respectively. Favourable risk factors for survival were absence of high-risk clinical features (Karnofsky score 50–60, serious organ dysfunction, recent fungal infection, P < 0·01) and absence of severe GVHD (P = 0·04), and favourable risk factors for EFS were absence of high-risk clinical features (P < 0·01) and use of two UCB units (P = 0·07). However, cell dose and HLA were not associated with final outcomes in this cohort of patients (Brunstein et al, 2007).

More recently, Société Française de Greffe de Moelle-Thérapie Cellulaire (SFGM-TC) – Eurocord collaboration reported results of 155 consecutive UCBT performed using a RIC in 21 French centres, between 2003 and December 2007 with a median follow-up of 18 months (2–56) (Rio et al, 2009). The median age was 47 years (18–69). Sixty-nine patients had acute myeloid leukaemia (AML) and 22 had acute lymphoblastic leukaemia (ALL; 59%), 33 patients had other lymphoid malignancies and there were 18 myelodysplastic syndrome, eight myeloma and five chronic myeloid leukaemia. At time of transplantation, 20% of the patients had active disease. Conditioning regimen was fludarabine (150–200 mg/m2), cyclophosphamide (50 mg/kg) and TBI (2 Gy). Two CB units were infused in 59 (38%) patients. In case of double UCBT, HLA and ABO classification of the unit with the highest degree of disparity was taken into consideration. Therefore, HLA match was 6/6 in five patients, 5/6 in 28, 4/6 in 93, ≤3/6 in eight; 42 patients had a minor and 65 a major ABO incompatibility. The number of TNC and CD34+ cell infused was 3·1 × 107/kg and 1·2 × 105/kg respectively; it was 2·8 × 107/kg and 1·4 × 105/kg, in single UCBT and 3·6 × 107/kg and 1·6 × 105/kg in double UCBT, respectively. CsA and MMF were used for GVHD prophylaxis. The CI of neutrophil engraftment at day +60 was 80 ± 3% with median time to achieve a neutrophil count >0·5 × 109/l of 20 d; autologous recovery was seen in 14% of the patients. In multivariate analysis, factors independently associated with better neutrophil recovery were CD34 cell dose (>1·2 × 105/kg) [Hazard ratio (HR) 1·51, P = 0·04], HLA compatibility (0–1 vs. 2–3) (HR 1·5, P = 0·05) and previous autograft (HR 1·8, P < 0·01). The NRM CI was 18 ± 3% at 18 months. ABO major incompatibility (HR 2·22, P = 0·02) and recipients’ positive CMV serology (HR 3·15, P = 0·03) increased the risk of NRM. The probability of overall survival (OS) and DFS at 18 months was estimated at 62 ± 5% and 51 ± 4%, respectively. In multivariate analysis, two factors were associated with improved DFS rates: disease in remission at UCBT (HR 2·08, P < 0·01) and HLA compatibility (HR 2·19, P = 0·02). DFS was 70% for patients given a six or 5/6 UCBT compared to 42% for ≤4/6 HLA compatibility.

In summary, larger series of patients are needed to draw definitive conclusions on the impact of cells dose and HLA matching in the UCBT-RIC setting, but both factors, as in the myeloablative conditioning regimen setting, are also associated with outcomes after RIC-UCBT and should be targeted when choosing a CB unit.

Selection of CB units based on the interactions between cell dose, HLA disparities and diagnosis

Given that cell dose and HLA disparities are important and independent factors associated with outcomes, it has been suggested that both factors interact mutually on engraftment and on other outcomes. Thus, a higher cell dose in the graft could partially overcome the negative impact of HLA for each level of HLA disparity, but this hypothesis has not been yet been fully demonstrated.

The Eurocord group has tried to analyse the interaction of cell dose and HLA in 550 UCBT for adults and children with malignant disorders (Gluckman et al, 2004). This study found that the 60-d CI of neutrophil engraftment for all patients was 74%, whereas the incidences for those with no HLA disparity (6/6) versus three or more of six were 83% and 53%, respectively. The number of HLA disparities was correlated with neutrophil recovery, with a log-linear relationship between HLA disparity and risk of graft failure, suggesting inferior engraftment with increased disparity. The CI of neutrophil and platelet recovery was also associated with number of NC before freezing and use of granulocyte colony-stimulating factor. A specific cell dose cut-off for the different HLA mismatched groups could not be determined.

Eapen et al (2007) also tried to analyse the interaction between cell dose and HLA disparity, comparing outcomes of 503 UCBTs and 282 unrelated bone marrow transplants (UBMTs) in children with acute leukaemia. The cut-off point for collected cell dose was defined as 3·0 × 107/kg for survival in children given a 5/6 HLA disparity graft, but a cell dose cut-off point associated with survival of children given a 6/6 or 4/6 grafts were not found. The probability of neutrophil recovery by day 42 and platelet recovery by 6 months were similar after UBMT or matched UCBT (6/6). Higher cell doses (>3·0 × 107/kg) resulted in a higher probability of engraftment in 5/6 UCBT but had no effect in 4/6 UCBT, probably suggesting that cell dose may not be able to overcome the adverse impact of HLA mismatching in the setting of 4/6 UCBT. However, larger series of patients should be analysed to enable definitive conclusions to be drawn.

In both previous analyses, the interaction between cell dose and HLA disparity were studied in patients with malignant disorders. However, other factors, such as diagnosis, have an important role in rate of engraftment and other outcomes. This is due to the fact that most patients have a complete marrow and have not received chemotherapy or immunosuppression before conditioning, or, in the cases of aplastic anaemia, they have often received previous multiple transfusions or had a severe infection at the time of transplantation, thus increasing the risk of non-engraftment. Recently, in light of the observation that requirements regarding cell dose and HLA matching may differ in malignant and non-malignant diseases (Rocha & Locatelli, 2008), we attempted to construct an algorithm to guide clinicians in choosing the ‘best’ CB unit, taking into account the impact of diagnosis, cell dose and HLA incompatibilities, in patients receiving a single UCBT. If the required cell dose is not achieved with a single unit, a double UCBT should be investigated. With this objective, two different cohorts of patients who had received a single UCBT between 1994 and 2005 were analysed: 925 patients had a malignant disease and 279 had a non-malignant disease (Eurocord, unpublished observations). Donor–recipient histocompatibility was determined by serology or antigen typing (low resolution) for HLA-A and HLA-B and by allele typing for HLA-DRB1.

Patients with malignant diseases

In patients with malignant diseases, the median age was 11 years (range 1 month–56 years). Diagnoses included AML in 24·6%, ALL in 44·3%, chronic leukaemia in 9·1% and myelodysplastic syndrome in 10% of cases. Only 9% were classified as having class I antigens and class II alleles identical to their donor; 42% had one HLA difference, 40% had two HLA differences and the remainder had three or four HLA differences. For patients who had one HLA mismatch, 67% had a class I difference and 33% had a class II difference. For patients with two HLA differences, 38% had two class I differences, 7% had two class II differences and 55% had one class I and one class II difference. The median number of TNC infused was 3·1 × 107/kg (range 2–5 × 107 TNC/kg). At day 100, the CI of neutrophil recovery (the first day the neutrophil count reached >0·5 × 109/l) was 77·4% and the CI of platelet recovery (the first day platelet count reached >20 × 109/l) was 54·7%. This related to the number of cells infused (P < 0·0001). HLA was a second factor that affected neutrophil engraftment, with a difference between 0–1 (81%), 2 (75%) and 3–4 (63%) HLA incompatibilities (P = 0·037). The role of HLA mismatching was partially abrogated by an increase in cell dose, except in the group of patients who received a 3–4 HLA-mismatched transplant.

Cell dose was the most important factor influencing outcome in the malignant disease group; a minimum of 3 × 107 TNC/kg at collection and 2 × 107 TNC/kg at infusion needed to be targeted. The number of HLA mismatches increased the risk of delayed engraftment and led to a higher incidence of TRM and chronic GVHD; however, it decreased the risk of relapse, resulting overall in a lack of influence of HLA mismatching on OS and DFS. Type of HLA mismatch did not influence outcome, but matching for HLA-DRB1 appeared better for patients receiving a graft that had two HLA incompatibilities. As stated earlier, increasing the cell dose abrogated the effect of HLA mismatching, but not for grafts with three or four HLA incompatibilities.

Patients with non-malignant diseases

The median age of patients with a non-malignant disease was 3 years (range 3 months–10 years). The diagnosis was bone marrow failure syndrome in 40%, primary immunodeficiency in 36% and a hereditary metabolic disorder in 24% of patients.

Only 18% were classified as identical for HLA class I antigens and class II allelic typing, whereas 43% had one HLA difference, 35% had two HLA differences and the remainder had three HLA differences. For patients with one HLA mismatch, 69% had a class I difference and 31% a class II difference. For patients with two HLA differences, 43% had two class I differences, 3% had two class II differences, and 54% one class I and one class II difference. The median number of TNCs infused was 6·4 × 107/kg (range 0·8–66 × 107/kg).

At day 100, the CI of neutrophil and platelet recovery was 69·3% and 50%, respectively. Both outcomes were associated with the median number of cells infused (P < 0·000055). HLA was also an important factor associated with neutrophil recovery, with a statistical difference between 0–1 and ≥2 HLA mismatches (P = 0·046). The role of HLA mismatching was abrogated by increasing cell dose, except in the group of patients who received a 3–4 HLA-mismatched transplant. There was no correlation between neutrophil recovery and class of HLA mismatch (class I or class II, or HLA-A, -B or -DRB1).

The CI of acute GVHD grade II–IV was 31·8% and grade III–IV 18%, and was only associated with the number of HLA incompatibilities (P = 0·0029). The CI of chronic GVHD was 24% and was also associated with the number of HLA incompatibilities (P = 0·01). The CI of OS at 100 months was 49%, and was influenced by cell dose and by the number of HLA mismatches. The group of patients who received an UCBT with <3·5 × 107 TNC/kg at infusion and a 2–3 HLA-mismatched transplant, had <10% survival. Increasing cell dose partially abrogated the effect of HLA mismatches; there was no statistical difference between the groups who received >3·5 × 107 TNC/kg with a 0, 1-, 2- or 3-HLA-mismatched UCBT.

Thus, patients with a non-malignant disease should receive a higher cell dose to obtain engraftment than patients with a malignant disease; this should not be below 4·9 × 107 TNC/kg at collection and 3·5 × 107 TNC/kg at infusion. In non-malignant disorders, HLA mismatching played a major role in engraftment, GVHD, TRM and survival that was partially abrogated by increasing cell dose. A CB graft containing two or more HLA disparities with a cell dose inferior to 3·5 × 107 TNC/kg should be avoided. Experience of double UCBT in non-malignant disorders is still too limited to allow routine recommendation of this type of transplant (Ruggeri et al, 2008).

In summary, we were not able to determine, for each HLA disparity, the best cut-off point for cell dose that is associated with engraftment or survival in patients with malignant or non-malignant disorders. However, we found that the minimum cell dose for all patients (children and adults) with malignant disorders with should be higher than 2·5 × 107/kg at collection or 2·0 × 107/kg after thawing. As stated above, Eapen et al (2007) found that, in children with acute leukaemias given a 5/6 HLA CB graft, the best cell dose cut-off point at freezing is 3·0 × 107/kg (or 2·5 × 107/kg at infusion). For non malignant disorders, HLA matching is crucial, and for patients who need a 4/6 HLA graft a CB unit containing more than 4·0–5·0 × 107/kg at collection or 3·5 × 107/kg at infusion should be targeted.

It has been suggested that, for each HLA disparity, the TNC dose should be increased by an increment of 1·5 × 107/kg, however this recommendation has not yet been validated by any study and the use of this algorithm would mean that many patients would not find a suitable single CB unit and will need a double UCBT. Double UCBT has been investigated; despite the fact that this has been associated with lower relapse rate in patients with lymphoid malignancies (Rodrigues et al, 2009), the incidence of acute and probably chronic GVHD is higher (Macmillan et al, 2009). Few cases of double UCBT have been reported in patients with non-malignant disorders (Ruggeri et al, 2008).

Other considerations when selecting a CB unit

Although the availability of HLA-matched CB units with an adequate cell dose is important, there are a number of other factors that need to be considered when selecting a unit for transplantation.

Cell content marker

The first consideration is related to the best cell content marker before freezing or after thawing. Most of the studies have analysed the impact of TNC dose at freezing or after thawing. When selecting a unit, one may be aware that after thawing, there is a median loss of 25% nucleated cells, mostly granulocytes. Among the nucleated cell subpopulations, it seems that the number of erythroblasts should also be considered (Stevens et al, 2002). The loss of CD34+ cells after thawing is less important, therefore when the number of CD34+ cells at freezing is available, this marker should be considered first; it is probably a better cell content marker than the TNC as it reflects better the HSC content.

Another aspect, not yet studied, is the association of viable TNC or viable CD34 infused with the outcome, as the methodology of counting viable cells can vary among transplant centres. In double UCBT, viable CD34+ cell dose has been associated with the engrafted unit (Barker et al, 2009).

The number of TNC or CD34+ cells does not reflect the functional aspects of haematopoietic stem cell content. Granulocyte-macrophage colony-forming units (CFU-GM) in the CB graft represent clonogenic potential but few studies have clearly documented its association with engraftment or survival (Migliaccio et al, 2000; Yoo et al, 2007; Iori et al, 2004). Moreover, one study showed the impact of colony-forming cells pre-freezing on engraftment (Migliaccio et al, 2000); the other two studies showed the impact of post-thawed CFU-GM on engraftment and survival (Yoo et al, 2007; Iori et al, 2004). Therefore due to logistic, technical and economic aspects, the use of this cell marker is difficult to apply routinely as a surrogate marker for CB unit choice. The number of CFU-GM present in the CB unit bag after thawing or in a small vial frozen (same conditions of the CB bag), should be determined in the transplant centre and should help the clinicians make a decision regarding a second transplant for patients with CB graft failure.

HLA allelic typing

Studies in UBMT have identified the importance of allelic disparity of HLA class I (-A, - B and -C) and class II (mainly DRB1) in determining transplant outcomes. Although HLA matching at antigen level (low or intermediate resolution) for HLA-A and -B and allele level matching for HLA-DRB1 continues to be the current standard for CB unit selection, some retrospective analyses have evaluated the impact of undetected allelic disparities in a subset of UCBT recipients. The data from Kögler et al (2005) regarding 122 donor–recipients pairs showed important changes in HLA-A, -B and -DRB1 matching status between the initial assignments (antigen level for HLA-A and-B) and allele level for HLA-A and -B. Over half of the pairs (nine out of 16) previously matched at 6/6 were matched at the allele level. When additional HLA loci, such as HLA-C and HLA-DQB1, were determined, only 14% of patients were matched for 9–10/10 alleles, 63% were matched for 6–8/10 and the remaining 23% were more extensively mismatched. It was not possible to find any association between allele typing and outcomes, because the number of patients were small in the different categories of mismatches and the patient cohort was heterogeneous (Kögler et al, 2005).

More recently, results of the Cord Blood Transplantation Study (COBLT) have been published (Kurtzberg et al 2008). HLA matching was assigned in 179 patients–donors pairs using high resolution typing. When typed for all six alleles (HLA-A, -B and -DRB1), approximately one third of the pairs were found to be more disparate. However, matching by high resolution typing was associated with decreased incidence of acute GVHD, but not with neutrophil or platelet engraftment. There was a trend towards better survival for patients given a 6/6 match by high resolution typing and worse survival rates for those with 3/6 but the numbers were too small to reach statistical significance (Kurtzberg et al 2008).

In fact, to determine the real value of allele typing in UCBT, thousands of patients–donors pairs will be needed to reach statistical significance. We encourage all CBB and transplant centres to perform this typing or to freeze DNA samples in order to further evaluate the impact of HLA allele typing in outcomes after UCBT.

Influence of HLA-C

Recently, Eurocord, in collaboration with Centre of International Blood and Marrow Transplant Registry, have analysed the impact of matching for HLA-C in UCBT, knowing that the current selection of CB units is based on antigen-level HLA typing at HLA-A and -B and allele-level at DRB1. We analysed haematopoietic recovery, acute GVHD and mortality in 619 UCBT recipients who received a single CB unit, a myeloablative preparative regimen and a calcineurin inhibitor for GVHD prophylaxis (Eapen et al, 2008). Eighty-three per cent of patients received UCBT for leukaemia or lymphoma and 17% for immunological, metabolic or histiocytic diseases. Seventy per cent of patients were ≤16 years of age at transplantation. HLA typing (using molecular methods) was performed for 96% (n = 593) of donor–recipient pairs; the method of typing was not available for the remaining 4% (n = 26). For all analyses, donor–recipient HLA matching was evaluated at the antigen-level (first two of four digits) for HLA-A, -B, -C and allele-level (four digits) for -DRB1. The median infused cell dose was 4·0 × 107/kg and median follow-up, 2 years. We first examined the effect of donor–recipient HLA matching and cell dose on haematopoietic recovery and mortality considering the current standard for selection of CB units. Fifteen per cent (n = 94) were matched at HLA A, B and DRB1, with 43% (n = 265) mismatched at one locus and 42% (n = 260) mismatched at two loci. As reported previously, compared to matched UCBT, neutrophil recovery at day-42 was lower after UCBT mismatched at one locus [relative risk (RR) 0·48, P = 0·042] and two loci (RR 0·38, P = 0·007). After adjusting for infused cell dose, year of transplant and disease status, platelet recovery and 1-year mortality rates were not different after matched and mismatched UCBT. We then examined whether the addition of another mismatch at the C locus impacted outcomes and again there was no statistically significant effect of HLA-C on outcomes. Briefly, this data suggest that HLA-C does not affect haematopoietic recovery, acute GVHD and 1-year OS after UCBT. However, definitive conclusions can only be achieved in a larger series. In the mean time, CB unit selection need not consider matching at the C-locus.

Killer-immunoglobulin receptor (KIR) ligand matching

Donor KIR (-ligand) incompatibility in the GVHD direction is associated with decreased relapse incidence (RI) and improved leukaemia-free survival (LFS) after haplo-identical and HLA-mismatched unrelated HSCT. Recently, we assessed outcomes of 218 patients with AML (n = 94) or ALL (n = 124) in complete remission (CR) who had received a single unit unrelated UCBT from a KIR-ligand compatible or incompatible donor (Willemze et al, 2009). Grafts were HLA-A, -B, -DRB1-matched (n = 21) or -mismatched (n = 197). Patients and donors were categorized to their degree of KIR-ligand compatibility by determining whether or not they expressed HLA-C group 1 or 2, HLA-Bw4 or HLA-A3/-A11. Sixty-nine patient–donor pairs were KIR-ligand incompatible and 149 were compatible in the GVHD direction. KIR-ligand incompatible UCBT showed decreased RI (HR 0·53, P = 0·05), improved LFS (HR 2·05, P = 0·0016) and overall survival (HR 2·0, P = 0·004). Those results were more evident for AML transplant recipients in whom the 2-year RI and LFS with or without KIR-ligand incompatibility were 5% vs. 36% (P = 0·005), and 73% vs. 38% (P = 0·012), respectively.

However, the Minnesota group has, more recently, published different results (Brunstein et al, 2009). Of note, the authors examined the clinical impact of KIR-L mismatch in a heterogeneous cohort of 257 patients with various malignant disorders, transplanted in remission or advance phase of the disease, given RIC (n = 102) or myeloablative conditioning regimen (n = 155) and single (n = 91) or double (n = 166) unit UCB grafts. Analyses of double unit grafts considered the KIR-L match status of the dominant engrafting unit. After myeloablative conditioning, KIR-L mismatch had no effect on grade III–IV acute GVHD, TRM, relapse and survival. In contrast, following RIC, KIR-L mismatch between the engrafted unit and the recipient resulted in significantly higher rates of grade III–IV acute GVHD [42% (CI, 27–59) vs. 13% (CI, 5–21), P < 0·01] and TRM [27% (CI, 12–42%) vs. 12% (CI, 5–19%), P = 0·03] with inferior survival [32% (CI, 15–59%) vs. 52% (CI, 47–67%), P = 0·03]. Multivariate analysis identified KIR-L mismatch as the only predictive factor associated with the development of grade III–IV acute GVHD [RR 1·8, CI (1·1–2·9); P = 0·02] and demonstrated a significant association between KIR-L mismatch and increased risk of death (RR 1·8, 95%CI, 1·0–3·1, P = 0·05) (Brunstein et al, 2009).

Therefore other series of patients analysing the impact of KIR-L matching are needed before this factor can be included in the algorithm of CB choice.

Unrelated donor CBB: economic and quality aspects

Economic aspects, distance of CBB and, more importantly, quality of the CB units are considerations when selecting a CB unit. The number of unrelated CBB, and in consequence, the number of available CB units for unrelated use is increasing worldwide. We estimate that more than 450 000 CB units are available for transplantation in more than 50 CBB in many countries ( The price of a CB unit varies between 15 000 and 22 000 Euros. Currently, there is an increasing number of internationally-exchanged CB units. For example, in France, 63% of UCBT were performed using a CB unit from abroad between 1994 and 2005. Therefore, many national regulatory agencies and transplant centres are aware of the need for international standards for CB collection, processing, testing, banking, selection and release. In 2006, Netcord-FACT (Foundation for the Accreditation of Cellular Therapy) published the third edition of the International standards for CB ( Founded in 1998, Netcord is the international CB banking arm of Eurocord, with a mission to promote high quality CB banking and clinical use of CB for allogeneic HSCT. Through its on-line virtual office (, CB units from member banks are made available for unrelated donor transplantation. Approximately 20 CBB, mostly European, are Netcord members and account for almost 50% of worldwide available units. To be an active Netcord member, a Netcord-FACT accreditation is required among other criteria. Some Netcord CBB have already been accredited and others are in the process of gaining accreditation. The major objective of these standards is to promote quality medical and laboratory practices throughout all phases of CB banking with the goal of achieving consistent production of high quality placental and CB units for transplantation. These standards cover: (i) collection of CB cells, regardless of the methodology or site of collection; (ii) screening, testing and eligibility determination of the maternal and infant donor according to local law; (iii) all phases of processing and storage, including quarantine, testing and characterization of the unit; (iv) making the CB unit available for transplantation, either directly or through listing it with a search registry; (v) the search process for selection of specific CB units; (vi) all transport or shipment of CB units, whether fresh or cryopreserved. To be compliant with Standards, CBB must use validated methods, supplies, reagents and equipment. They must maintain a comprehensive, properly documented Quality Management Programme and track the clinical outcomes of patients who receive CB units from that bank. The accreditation process includes submission of written documentation and on-site inspection of collection, processing and storage facilities. Netcord-FACT accredited CBB are re-inspected routinely every 3 years.

All the practical aspects of CB banking, such as need to obtain the mother’s informed consent, collection techniques, labelling and identification, infectious disease and genetic disease testing, HLA typing, methodology of cell processing, cryopreservation, transportation and release have been extensively published (Davey et al, 2004; Rubinstein, 2006). All of these issues are detailed in the last version of the Netcord-FACT Standards (

Very few clinical studies have been performed that correlate laboratory aspects of CB banking (such as methods of volume reduction, transportation presence of dimethyl sulphoxide) and outcomes of transplanted units (Nagamura-Inoue et al, 2003). Also the effect of ‘cord blood bank’ on outcomes after UCBT has not been studied.

Other considerations

Some other aspects related to the CB unit selection and transplantation can be associated with outcomes, however the data published is still preliminary and need confirmation in larger series of patients. ABO major incompatibility has been described to be associated with decreased survival and DFS rates in UCBT for adults with haematological malignancies (Arcese et al, 2006). ABO major incompatibility was also found to be associated with higher TRM after RIC-UCBT in adults with haematological malignancies (Rio et al, 2009). Therefore when many cord blood units are available, the use of a unit that is ABO compatible or with minor incompatibilities should be taken into consideration.

Patient’s pre-transplant anti-HLA antibodies also seem to have an impact on neutrophil and platelet engraftment after UCBT. Since most UCBT are HLA-mismatched, the presence of anti-HLA antibodies in the patient against the HLA of the CB should be identified. In a recent Japanese study, 648 recipients of a single unit and myeloablative UCBT were analysed retrospectively and 155 were antibody-positive. Of the positive group, 117 recipient antibodies had no specificity against the CB HLA antigens (positive group); however in 38 patients the anti-HLA antibodies were against the CB HLA (cross positive group). The CI of neutrophil recovery was 79% for the negative group, 69% for the positive group and 50% for the positive cross group. In a multivariate analysis, the cross positive group had lower neutrophil and platelet recovery compared to the negative group. Therefore, patients’ pre-transplantation anti-HLA antibodies should be examined and considered when selecting CB units (Takanashi et al, 2009). Another concern raised during selection of the CB unit is the influence of storage duration of the unit and its influence of UCBT outcomes. Some CBB in USA and in Europe were established in the early 1990s, therefore the ‘age’ of the CB units is around 10–15 years. No data has yet been published showing an impact of storage period and outcomes. In the Eurocord registry, the association of CB storage duration with engraftment and survival was assessed in 1351 UCBT recipients given a single unit following a myeloablative conditioning regimen. Storage duration of the CB unit (median 2·3 years, range 0·3–14) was not associated with neutrophil recovery or survival (Eurocord, unpublished preliminary observation).

Other factors related to the technique of transplantation, such as route of CB infusion, conditioning regimen and GVHD prophylaxis, may also to be associated with more rapid time to engraftment. In a preliminary matched-pair analysis comparing patients transplanted with CB injected intravenously (IVCB) versus CB injected directly into the bone marrow (IBCB) of the iliac crest, IBCB patients (n = 50) were matched to 88 IVT recipients. The CI of neutrophil recovery was 70 ± 5% in IVCB recipients vs. 80 ± 6% in the IBCB group (P = 0·27). However, patients receiving CB IBCB had higher CI of platelet recovery at day 60 (82 ± 5%) compared to IVCB group (40 ± 5%; P < 0·0001). Strikingly, the incidence of acute GVHD grade II–IV was 12% in IBCB group compared to 38% in IVCB group (P = 0·0001) and grade III–IV was 2% compared to 18% (P < 0·001). Overall survival at 1 year was 67 ± 7% in the ICCB group compared to 43 ± 5% in the IVCB group (P = 0·07). In conclusion, injection of CB cells via the bone marrow seems to be able to overcome the problem of delayed platelet recovery observed after intravenous tranplantation. The reduced incidence/severity of aGVHD observed in IBCB patients is intriguing and promising. A longer follow-up will disclose whether the IBCB route is also associated with different outcome in term of relapse and survival (Frassoni et al, 2009).

In a recent Eurocord study, use of fludarabine in the myeloablative conditioning regimen was associated with neutrophil and platelet recovery in UCBT adult recipients receiving a lower TNC dose (Nabhan et al, 2008). In this study, it was not possible to evaluate the role of antithymocyte globulin/antilymphocyte globulin because it was used in almost all UCBT performed. Use of fludarabine in the preparative regimen has also been associated with improved engraftment rate independent of cell dose and HLA in UCBT for patients with Fanconi anaemia (Gluckman et al, 2007). The use of methotrexate (MTX)-containing regimens for GVHD prophylaxis have been associated with delayed engraftment and risk of graft failure in patients with haemoglobinopathies undergoing an HLA-identical sibling UCBT (Locatelli et al, 2003). However, its use in UCBT is not established. In Europe and in USA, the most common combination is calcineurin coupled or not with steroids or MMF. However, Japanese transplant centres have shown interesting results with calcineurin combined with low dose MTX (Takahashi et al, 2004, 2007; Atsuta et al 2009). Only prospective studies may establish the role of MTX in GVHD prophylaxis for UCBT.

Back up autologous reinfusion and second transplants

Indication of harvesting an autologous back-up bone marrow or peripheral haematopoietic stem cell is controversial for patients with high risk of graft failure. In 130 unrelated HSCT recipients who had backup harvests, 15 (11%) had their back-up harvests re-infused, all for graft failure (Stotler et al, 2009). Five patients are alive, four died of relapse and six of infection. It seems that harvest back up it is not necessary. Moreover, given that many of the patients have high risk leukaemia, reinfusion of an autologous rescue may increase the risk of relapse. There is no study comparing outcomes of patients with graft failure after UCBT that have been given back up rescue or second transplants. In a recent Eurocord survey for primary graft failure, from May 1995 to July 2007, 1115 UCBT for haematological diseases were reported to the Eurocord registry, 113 of these transplants were diagnosed as primary graft failures (failure to achieve a neutrophil count >0·5 × 109/l until 60 d after UCBT or having received a treatment for graft failure in this period). Patients having relapsed in the first 100 d after UCBT were excluded. From this group, 34 patients did not receive treatment, 25 patients received an autologous rescue from pre-cryopreserved stem cells and 54 patients received a second HSCT. At 2 years after transplant: 3/25 patients given an autologous rescue, 3/5 patients receiving mismatched unrelated bone marrow, 10/13 patients receiving haploidentical related grafts, 5/10 patients receiving double UCBT and 2/26 receiving single UCBT were alive and well (Fernandes et al, 2008). Therefore, in this retrospective analysis, allogeneic second transplants may be considered as a salvage therapy for primary graft failure after UCBT. In a small series of patients double cord blood transplantation was described to rescue all four patients with early graft failure after previous allogeneic HSCT (Fernandes et al, 2007). The choice of the best conditioning regimen and the best cell source are still open questions. The use of autologous rescue was not associated with better survival in this small series of patients.


Unrelated UCBT is considered for patients in the absence of HLA-matched donors. Outcomes after UCBT have improved in recent years, mainly due better donor choice (cell dose and HLA matching), early patient referral for transplantation, improvement in supportive care and greater centre experience. High cell dose (TNC or CD34+ cells) and better HLA matching have been associated with better engraftment and survival. Other considerations for a CB unit choice and transplantation should also include patient’s diagnosis. High quality CB units stored in accredited CBB should be an advantage. Other factors, such as ABO compatibility, use of fludarabine in the myeloablative conditioning regimen and avoidance of MTX in the GVHD prophylaxis regimens may also improve outcomes after UCBT (Fig 3).

Figure 3.

 Algorithm of cord blood choice according to number of HLA disparities, cell dose and diagnosis.