T cell depletion in paediatric stem cell transplantation


Correspondence: C. Booth, Institute of Child Health, 30 Guilford Street, London WC1N 1EH, UK.

E-mail: c.booth@ucl.ac.uk


Haematopoietic stem cell transplantation (HSCT) can be a curative procedure for a growing number of paediatric diseases, but as the indications for HSCT grow, so does the need to find suitable stem cell donors. When the preferred option of a genoidentical sibling donor is not available alternative donors, including unrelated adult or umbilical cord blood donors, or haploidentical related donors may be considered. Outcome following alternative donor HSCT has improved over the past 20 years but graft-versus-host disease (GvHD) remains a significant obstacle. T cell depletion (TCD) for non-genoidentical grafts aims to reduce the morbidity and mortality associated with GvHD, but this intervention has not led directly to improved survival due to delayed immune reconstitution and increased infections, graft rejection and increased rates of disease relapse. Limited data from the paediatric population, however, suggest some encouraging results for children undergoing haploidentical HSCT: a move from positive selection of CD34+ haematopoietic stem cells towards negative depletion of specific cell subsets in order to retain useful accessory cells within the graft appears to enhance immune reconstitution and improve disease-free survival. Here we review recent paediatric outcome data for T cell-depleted HSCT, explore the role of serotherapy in conditioning regimens and look at future possibilities to improve outcome, including novel allodepletion techniques, suicide gene therapy and pathogen-specific immunotherapy.

Other Articles Published in this Series

Tolerogenic dendritic cell therapy for rheumatoid arthritis: where are we now? Clinical and Experimental Immunology 2013, 172: 148–57.

Promoting transplantation tolerance; adoptive regulatory T cell therapy. Clinical and Experimental Immunology 2013, 172: 158–68.

Whole body imaging of adoptively transferred T cells using magnetic resonance imaging, single photon emission computed tomography and positron emission tomography techniques, with a focus on regulatory T cells. Clinical and Experimental Immunology 2013, 172: 169–77


Haematopoietic stem cell transplantation (HSCT) is a curative procedure for a wide range of paediatric diseases, including haematological malignancies, primary immunodeficiency disorders, inherited metabolic conditions and bone marrow failure. With an increasing number of conditions for which HSCT can be employed, there is the need to find increasing numbers of suitable donors. A genoidentical sibling donor is still the preferred option for most conditions, but outcome is improving after alternative donor transplantation from matched unrelated donors (MUD), unrelated umbilical cord blood transplantation (UCBT) and haploidentical HSCT thanks to reduced toxicity conditioning regimens, improved viral surveillance and better treatment of infectious complications. The advent of sensitive molecular-based tissue typing has improved human leucocyte antigen (HLA) matching and consequently reduced the risk of graft-versus-host disease (GvHD), but at the same time lowers the probability of finding an ‘HLA-matched’ donor defined by more stringent criteria.

The major drawback of non-genoidentical HSCT is an increased incidence of GvHD. T cell depletion (TCD) of grafts from unrelated or haploidentical donors has been used for more than 30 years to reduce the incidence and severity of GvHD, but this has not equated directly to improved survival of patients, as the removal of T cells results in delayed immune reconstitution and reduced engraftment alongside increased rates of disease relapse, lymphoproliferative disease (LPD) and viral reactivation. The pros and cons associated with T cell-depleted grafts are presented in Table 1. Graft manipulation techniques are becoming more sophisticated and our growing understanding of such strategies allows us to decide on an optimal approach for patients based on underlying disease, donor, infectious history and a desired graft-versus-malignancy effect. The challenge remains to improve survival by maintaining low rates of GvHD while hastening immune recovery and, in the case of malignancy, allowing effective anti-tumour activity. Data specific to the paediatric population are lacking, and most paediatric transplant centres draw conclusions from the experience of adult patients with leukaemia or lymphoma. This review will focus upon more recent outcome data relating to TCD in paediatric transplantation and current trends in TCD methods. We will also summarize future perspectives of further improving the survival and reducing morbidity associated with T cell-depleted transplants.

Table 1. Advantages and disadvantages of T cell depletion.
  1. Adapted from Ho and Soiffer [1]. EBV: Epstein–Barr virus; LPD: lymphoproliferative disease; GvHD: graft-versus host disease.
Decreased incidence of GvHDDelayed immune reconstitution
Decreased organ toxicityViral reactivation/infectious complications
Reduced need for immunosuppressionReduced graft versus leukaemia effect
Lower early transplant-related mortalityIncreased incidence of graft failure
 Increased risk of EBV-associated LPD

Methods of TCD

TCD of donor grafts has been utilized for more than 30 years, with various methods employed during that time; methods have generally involved physical separation of cells or antibody-based techniques [2-12]. Different forms of depletion exist and are grouped most easily into in-vivo or ex-vivo TCD and the latter into positive or negative selection methods.

In-vivo TCD relies upon anti-T cell antibodies, most commonly anti-thymocyte globulin (ATG) and anti-CD52 monoclonal antibody (alemtuzumab), to remove donor and recipient T cells involved in immune-mediated graft rejection or GvHD. In-vivo methods are inherently less predictable than ex-vivo techniques, as the half-lives of the antibodies are variable and the pharmacokinetics are still being determined. This is of particular importance in paediatric transplantation, where evidence supporting specific doses is lacking. Polyclonal ATG, raised in rabbits or horses, has immunomodulatory as well as immunosuppressive effects [13] and the dose of ATG affects both GvHD and immune reconstitution, with higher doses affording some protection from acute and chronic GvHD. However, this leaves patients susceptible to fatal infections through delayed immune reconstitution [14]. A recent study linked the development of anti-ATG IgG antibodies in the early post-transplant period to an increased incidence of acute GvHD, leading the authors to suggest that measurement of such antibodies should be considered, particularly in children who have received ATG prior to conditioning in the treatment of their underlying condition [15]. Alemtuzumab is a humanized anti-CD52 monoclonal antibody and has been used in pre-HSCT conditioning regimens for more than 10 years now, replacing the older Campath 1G and 1M antibodies. The few data there are from adult studies suggest huge interpatient variability in pharmacokinetics [16]. Data from our centre confirm a wide range of alemtuzumab levels in children (n = 57): higher levels on day 0, as measured by flow cytometric analysis [17], reduced grades III–IV acute GvHD significantly in the matched donor setting but also increased the risk of adenoviraemia significantly. Lymphocytotoxic levels >100 ng/ml were still present 28 days post-HSCT in almost 20% of patients (personal communication, S. Adams, 2011).

Ex-vivo depletion of T cells or T cell subsets achieved through physical separation (using soybean agglutinin and sheep red blood cells [4] or counterflow centrifugal elutriation [7, 18]) has now been superseded by antibody-based techniques. In contrast to in-vivo techniques, the T cell dose can be quantified accurately allowing optimal graft manipulation based on transplant characteristics such as the degree of donor mismatch. The introduction in the late 1990s of several platforms for cell separation, including the CliniMACS system (Miltenyi Biotec, Bergisch-Gladbach, Germany), allowed for efficient positive selection of CD34+ haematopoietic stem cell progenitors through immunomagnetic separation, leading to a 4–5-log reduction in CD3+ cell numbers in a granulocyte colony-stimulating factor (GCSF)-mobilized peripheral blood stem cell (PBSC) graft [19-21]. Although profound reduction of T cell numbers lowers the risk of GvHD, other immune cells which may be beneficial to engraftment, innate immunity and tumour surveillance such as natural killer (NK) cells, dendritic cells and regulatory T cells are also removed from the graft. Negative selection methods include removal of CD3+/CD19+ lymphocytes or T cell receptor (TCR)-αβ/CD19+ lymphocytes which retains monocytes and natural killer (NK) cells in the graft, and results using these techniques will be discussed later.

TCD in matched unrelated donor HSCT

An HLA-matched sibling donor would be the preferred donor for the majority of transplants conferring the lowest rates of GvHD, especially when bone marrow is the source of stem cells. Results from TCD-matched unrelated donor grafts are improving in terms of acute GvHD, but published data have been lacking in the paediatric population. In-vivo TCD with ATG was compared to GvHD prophylaxis with cyclosporin A and methotrexate (CYA/MTX) in a prospective, randomized study of 109 adults with haematological malignancies, and results demonstrated no difference in overall survival: reduced rates of grades III–IV GvHD observed in patients receiving ATG were counterbalanced by an increased risk of infection in this group [14]. A later prospective, randomized study including some children with leukaemia compared CYA/MTX with partial (×1·5 log) ex-vivo TCD. TCD proved superior in some areas with less acute GvHD, less regimen-related toxicity and accelerated neutrophil recovery but, again, this came at the expense of infectious complications. No difference could be determined between the groups in terms of overall survival, disease-free survival, acute leukaemia relapse, graft failure or chronic GvHD [22].

CD34+ selection has been shown to minimize the risk of GvHD to 10% (grades II–IV) in a MUD PBSCT setting without impacting upon relapse rates or survival and removing the need for post-transplant immunosuppression, and in this particular study all cases of higher-grade GvHD (≥grade II) were associated with human herpesvirus 6 (HHV-6) infection [23]. Lawson and Darbyshire (personal communication, 2012) compared positive CD34+ stem cell selection against CD3+/CD19+ depletion methods in matched related donors (MRD), MUD and haploidentical donors for malignant and non-malignant conditions. They found significantly higher rates of acute grades II–IV GvHD in patients receiving CD3+/CD19+-depleted grafts (55% versus 35%) and in the MUD setting CD3+/CD19+ depletion resulted in inferior relapse-free survival. The median CD3+ cell dose reported by this group with CD3+/CD19+ depletion was 1 × 105/kg versus 0·37 × 105/kg with CD34+ selection. Geyer et al. [24] investigated addback of 1–2·5 × 105 CD3+ cells/kg after CD34+ selection of PBSCs, with the hypothesis that addition of a small T cell dose would reduce the risk of graft failure while maintaining lower GvHD rates associated with CD34+ selection. They were able to demonstrate high rates of engraftment as well as low risk of grades II–IV acute GvHD (15·8%), chronic GvHD and early transplant-related mortality (TRM), suggesting this as a reasonable strategy for children lacking a matched related donor.

To address concerns over a reduced GvL effect after TCD transplants using antibody-based serotherapy, an international study on behalf of the Centre for International Blood and Bone Marrow Transplant Research (CIBMTR) reported on 715 children with acute lymphoblastic leukaemia (ALL) who were grouped depending on receipt of ATG, alemtuzumab or no in-vivo T cell depletion. Conditioning with serotherapy resulted in significantly lower rates of grades II–IV acute GvHD and chronic GvHD, most pronounced with alemtuzumab (Fig. 1), yet no difference was observed between the groups for non-relapse mortality, relapse or survival (including leukaemia-free survival) [25]. There was a trend towards more infectious deaths in the group of patients receiving alemtuzumab compared to ATG or no serotherapy; other single-centre studies in paediatric patients have also shown less GvHD with alemtuzumab but slower immune reconstitution [26] and more adenovirus infections [27] when compared to ATG. The CIBMTR study suggests that in the context of a myeloablative, total body irradiation (TBI)-based conditioning regimen in childhood ALL, there is little additional benefit from a GvL effect. The story is somewhat different in the face of reduced-intensity conditioning (RIC). Another multi-centre study asked the same question of 1676 adult patients with haematological malignancies who received a RIC regimen with or without in-vivo TCD. The risk of grades II–IV aGvHD was significantly lower in patients receiving alemtuzumab compared to those receiving ATG or T replete transplants, but relapse was significantly more frequent with either form or TCD (P > 0·001) [28].

Figure 1.

Incidence of (a) acute graft-versus host disease (aGvHD) grades 2–4 and (b) chronic GvHD following unrelated donor haematopoietic stem cell transplantation (UD HSCT) with no in-vivo T cell depletion versus the use of anti-thymocyte globulin (ATG) or alemtuzumab.

Another important observation from the CIBMTR study was that use of PBSC grafts with or without in-vivo T cell depletion led to significantly higher risks of chronic GvHD, higher non-relapse mortality and lower leukaemia-free and overall survival. This is in contrast to the findings of a large adult study using in-vivo alemtuzumab in the myeloablative setting, where the incidence of chronic GvHD, either overall (bone marrow 47%, PBSC 49%) or extensive (bone marrow 15%, PBSC 13%), was not increased with PBSCs [29]. The data suggest thus far that there are differences in outcome between adult and paediatric recipients of PBSC grafts, and avoiding transplantation of PBSC in children and adolescents undergoing myeloablative HSCT, even in the setting of in-vivo T cell depletion, may be preferable. Again, the RIC or modified conditioning HSCT setting may be different, and HSCT with alemtuzumab/treosulfan/fludarabine and PBSCs from MUD donors led to excellent engraftment with low rates of GvHD in children with non-malignant conditions [30]. It should also be stressed that the impact of TCD on graft failure and long-term survival is dependent, to some extent, upon the underlying condition; for example, in the case of Hurler's syndrome (mucopolysaccharidosis type I) TCD was a significant risk factor for graft failure and was associated with lower survival rates [31].

TCD and unrelated umbilical cord blood transplant (UCBT)

UCBT offers a rapidly available alternative donor source and has been used as a stem cell source since 1993 [32]. Although the quantity of cells available from a cryopreserved unit is variable, for paediatric practice a single donor unit often suffices, with a minimum of 1 × 105 CD34+ cells required for successful engraftment [33] and double-cord transplants can be used if a higher cell dose is required. Immune reconstitution post-UCBT is slow and viral reactivation in recipients can cause severe morbidity or death, due to the prolonged time to engraftment and the naive phenotype of cord lymphocytes. In outcome studies of adult leukaemia patients, survival after 1- to 2-antigen-mismatched UCBT is comparable to MUD BMT [34] or 1A-mismatched PBSCT or BMT [35], with higher cell doses the main predictor of success. Further paediatric studies also demonstrate success after unrelated UCBT with delayed engraftment but disease-free survival comparable to other stem cell and donor sources with less acute GvHD reported [32, 36-39]. After initial concerns that a lack of GvHD would correlate to a reduced GvL effect, studies suggest that this is not the case.

As most UCBT are HLA-mismatched, many investigators utilize in-vivo T cell depletion with ATG. Chiesa et al. [40] recently described a cohort of 30 paediatric patients undergoing unrelated UCBT for either malignancy or immunodeficiency in whom serotherapy was omitted. This led to a rapid peripheral CD4+ T cell expansion, and although viral infection was common (68%), virus-specific lymphocytes were detectable 2 months post-procedure, and viral-related morbidity and mortality was very low. Acute GvHD (grades II–IV) was seen in 50% of patients but responded to steroid therapy and rates of severe GvHD were comparable to previous UBCT studies. This demonstrates that, in the right setting, in-vivo T cell depletion may be omitted in unrelated UCBT to allow accelerated immune recovery, even with a 2-HLA-antigen mismatch between cord and recipient. A recent study comparing UCBT with no serotherapy versus early or late ATG showed, respectively, improving immune reconstitution (Fig. 2), but with increasing rates of acute GvHD (Fig. 3) (personal communication, R. Chiesa, 2012). Rates of chronic GvHD and overall survival were similar between the groups. The study concluded that an individualized approach to the use of serotherapy in UCBT may be the best approach (Fig. 4).

Figure 2.

Effects of early, late or no serotherapy on reconsitution of (a) CD3+ lymphocytes and (b) CD4+ T lymphocytes after umbilical cord blood transplantation (UCBT). Immune reconstitution is significantly faster if in-vivo T cell depletion (TCD) is omitted. Numbers are presented as median counts (± standard error of the mean).

Figure 3.

Comparison of the incidence of acute graft-versus-host disease (aGvHD) after umbilical cord blood transplantation (UCBT) with early, late or no in-vivo T cell depletion. (a) Cases of grades II–IV aGvHD (b) Cases of grades III–IV aGvHD.

Figure 4.

Experience-based suggestions for serotherapy use in umbilical cord blood transplantation dependent on clinical situation.

Development of TCD in the haploidentical setting

Haploidentical transplants offer great promise in terms of donor availability, not only for initial transplant procedures but also for donor-derived cellular therapies post-HSCT, should the need arise. However, it has been challenging over recent decades to overcome the significant risks of severe GvHD and delayed immune reconstitution leading to fatal infections and an increased risk of disease relapse. The development of improved TCD methods, as summarized in Table 2, RIC regimens and understanding of the role of NK alloreactivity in engraftment and GvL have improved outcome after haplo-HSCT. The importance of achieving remission prior to transplant for successful outcome in children with high-risk disease had also been reinforced, as Klingebiel et al. reported 5-year leukaemia-free survival as 30% in CR1, 34% in CR2, 22% in CR3 and 0% for those children who did not achieve remission [41]. This study also highlighted the significance of a good stem cell dose (as demonstrated by the trend towards improved leukaemia-free survival and relapse incidence in children receiving > 12·4 × 106/kg CD34+ cells) and the benefit of performing such high-risk procedures in an experienced centre.

Table 2. T cell depletion (TCD) methods utilized in haploidentical HSCT
Ex vivoIn vivo
  1. DLI: donor lymphocyte infusion; TK: tyrosine kinase; CTL: cytotoxic lymphocytes; TCR: T cell receptor.

CD34+ stem cell enrichment

+Allodepleted DLI



Allodepletion with cyclophosphamide

CD3/19 depletion


TCR α/β depletion 

The use of CD34+ stem cell megadoses derived from GCSF-mobilized PBSCs allowed the HLA barrier in the haploidentical setting to be crossed without intensive myeloablative conditioning regimens and heavy immunosuppression, supporting improved engraftment rates while maintaining the reduced risk of aGvHD associated with ex-vivo TCD [42-47]. In 2007, Handgretinger et al. [48] published data providing evidence that reduced intensity conditioning (fludarabine, melphalan, thiotepa and OKT-3 serotherapy), in combination with a CD3+/CD19+ depletion strategy using PBSCs, was a feasible option for children with haematological malignancy. A 3·5–4-log depletion of T cells was achieved, and although higher rates of treatment-responsive aGvHD were seen in comparison to CD34+ selected grafts, TRM was low and the infusion of large numbers of NK cells afforded immunological anti-leukaemic activity, resulting in improved disease-free survival. Studies in adult patients with leukaemia and lymphoma also confirmed that a RIC approach in the haploidentical setting (with various TCD strategies) could offer a curative management option, with overall survival rates comparable to MUD transplants [49-51].

CD3+/CD19+-depleted PBSCT in the mismatched setting following RIC conditioning is a useful option for children with both malignant and non-malignant diseases. Bader et al. [52] reported very encouraging results from 59 children transplanted with this approach between 2005 and 2011, demonstrating rapid, sustained engraftment and immune recovery with low TRM in both groups. As shown in previous studies, the initial T cell regeneration seen early post-transplant is the result of peripheral T cell expansion. Thymus-dependent T cell production was evident from day+100 [evaluated with T cell receptor excision circles (TRECS) analysis]. Viral reactivation [cytomegalovirus (CMV), adenovirus and Epstein–Barr virus (EBV)] was seen in up to 52% of children, but proved fatal in only one patient (unpublished data). For leukaemic patients a good GvL effect was reported, with a 3-year survival rate of 68%. However, patients who were not in remission at the time of transplant did not survive.

Immunomagnetic depletion of TCR-αβ and CD19 lymphocytes

Selective depletion of TCR-αβ/CD19+ lymphocytes represent a further refinement in TCD methods leaving beneficial effector cells, such as NK cells, dendritic cells, regulatory T cells (Tregs) and γδ T cells within the graft while maintaining TCD of 4–5 logs. The inclusion of γδ T cells can hasten immune reconstitution, and these cells have not been implicated in aGvHD [53]. More than a decade ago, Lamb et al. [54] reported significantly improved disease-free survival in a cohort of patients undergoing haplo-HSCT for leukaemia and moreover, of 43 survivors, those with higher γδ T cell numbers demonstrated improved overall survival and disease-free survival at 30 months. More recent results from a larger patient cohort were also impressive, with significantly increased overall survival and disease-free survival in patients with higher γδ T cell numbers (54·4% versus 15%, P = 0·0003 and 70·8% versus 19·6%, P = 0·0001, respectively). The incidence of GvHD was comparable between the two groups [55]. Early data from children transplanted using this TCD technique are promising (Handgretinger, R. et al., Abstract, 53rd ASH Annual Meeting 2011).

Improving outcome after TCD HSCT

The prolonged period of immunodeficiency associated with T cell-depleted HSCT leaves the recipient susceptible to potentially fatal infections. Many years of research have focused upon ways to infuse T cell doses large enough to promote and rapid immune recovery, thereby reducing infectious complications while minimizing the risk of GvHD and graft failure.

Donor lymphocyte infusions (DLI) and pathogen-specific immunotherapy are two strategies that have been employed to reduce mortality from post-HSCT viral reactivation, but both rely generally upon access to the donor (with adequate anti-viral immune responses), limiting scope at present to haploidentical HSCT. Adoptive transfer of T cells increases the risk of GvHD, as alloreactive cells are also present. Several groups have utilized this approach successfully to treat post-transplant infection with CMV, adenovirus and aspergillus, but the production process is dependent upon cell culture and is time-consuming [56-59]. In addition, EBV-specific CTLs have proved effective as rescue therapy for post-transplant lymphoproliferative disorder [59]. To overcome the problem of donor access, approaches using third-party cytotoxic T cells (CTLs) have been investigated to treat both EBV and adenovirus [60, 61].

Selective ex-vivo depletion of alloreactive lymphocytes from haploidentical grafts allows accelerated immune recovery (with effective pathogen specific responses) without the risk of severe GvHD. Various techniques have been used to remove alloreactive lymphocytes, with the majority based on cell surface expression of activation markers on alloreactive donor cells in the context of a mixed leucocyte reaction (MLR). Methods include immunomagnetic selection [62-65], immunotoxin-based therapy [66], photodynamic purging [67, 68], FasL-mediated killing [69], heat shock protein 90 (HSP 90)-induced apoptosis [70] and induction of anergy through co-stimulatory blockade [71].

RIC T cell-replete haplo-HSCT with in-vivo allodepletion of T cells using cyclophosphamide on days 3 and 4 post-HSCT was compared recently to RIC double-cord blood HSCT in the adult setting. The haploidentical approach gave reduced rates of GvHD and lower transplant-related mortality compared to double-cord blood, but relapse rates were increased, giving comparable overall survival [50].

Initial suicide gene therapy procedures involved insertion of the herpes-simplex thymidine kinase suicide gene into donor T lymphocytes, rendering them susceptible to gancyclovir [72]. Phases I–II studies in 50 adult patients with high-risk leukaemia after haplo-HSCT demonstrated successful control of acute and chronic GvHD (n = 11) after gancyclovir treatment [73]. However, this approach precludes the use of gancyclovir to treat viral infection. Other suicide genes are being explored in children undergoing haplo-HSCT, including inducible caspase 9, which upon activation by a bioinert drug induces apoptosis of the cell [74].

CD4+CD25+forkhead box protein 3 (FoxP3)+ Tregs are known to mediate tolerance and have been implicated in GvHD. Brunstein et al. [75] infused third-party cord blood Tregs which had been expanded ex-vivo into 23 patients undergoing double-UD UBCT with reduced incidence of grades II–IV GvHD compared to historical controls. This form of immunotherapy had no effect on relapse rate. Di Ianni et al. then demonstrated expansion to be an unnecessary step, as infusion of 2 × 106/kg freshly isolated donor CD4+CD25+ cells prior to haplo-HSCT (with megadoses of CD34+ cells and 1 × 106/kg T cells) resulted in rapid immune recovery and prevention of GvHD without post-transplant immunosuppression [76]. Although this may not represent an infusion of pure Tregs it would appear that sufficient regulatory cells were infused in order to control the development of GvHD.


TCD methods have been employed for more than 30 years and multiple approaches have reduced significantly the burden of acute and chronic GvHD in paediatric patients undergoing HSCT. However, due to an increased incidence of graft rejection, disease relapse and infectious complications, TCD has not improved the overall outcome of HSCT, with the exception of haploidentical grafts. Selective TCD approaches, preserving the function of other accessory cells in the graft, the use of TCD with adjunctive cellular therapy and the use of TCD in selected patients, offers significant promise for the future.


The authors have no conflicts of interest to declare.