Dr Peter Lang MD, Department of Pediatric Oncology University Children's Hospital Eberhard Karl's University, Tübingen Hoppe Seyler Straße 1 D 72076 Tübingen, Germany. E-mail: firstname.lastname@example.org
Positive selected haematopoietic stem cells are increasingly used for allogeneic transplantation with the CD34 antigen employed in most separation techniques. However, the recently described pentaspan molecule CD133 appears to be a marker of more primitive haematopoietic progenitors. Here we report our experience with a new CD133-based selection method in 10 paediatric patients with matched unrelated (n = 2) or mismatched-related donors (n = 8). These patients received a combination of stem cells (median = 29·3 × 106/kg), selected with either anti-CD34 or anti-CD133 coated microbeads. The proportion of CD133+ selected cells was gradually increased from patient to patient from 10% to 100%. Comparison of CD133+ and CD34+ separation procedures revealed similar purity and recovery of target populations but a lower depletion of T cells by CD133+ selection (3·7 log vs. 4·1 log, P < 0·001). Both separation procedures produced >90% CD34+/CD133+ double positive target cells. Engraftment occurred in all patients (sustained primary, n = 8; after reconditioning, n = 2). No primary acute graft versus host disease (GvHD) ≥ grade II or chronic GvHD was observed. The patients showed a rapid platelet recovery (median time to independence from substitution = 13·5 d), whereas T cell regeneration was variable. Five patients are alive with a median follow-up of 10 months. Our data demonstrates the feasibility of CD133+ selection for transplantation from alternative donors and encourages further trials with total CD133+ separated grafts.
However, recent studies have revealed the existence of CD34− stem cell populations that also have a repopulating capacity and are putative precursors of CD34+ cells (Bhatia et al, 1998; Zanjani et al, 1998). CD133, an important antigen in this context, is a five transmembrane domain glycoprotein that is mainly co-expressed with CD34 (Yin et al, 1997) but also found on CD34−/CD38−/Lin− precursors (Gallacher et al, 2000).
Human CD133+/CD34−/Lin− cells are capable of giving rise to CD34+ cells in vitro and engrafting sublethally irradiated non-obese diabetic severe combined immunodeficient (NOD/SCID) mice (Gallacher et al, 2000). Moreover, several studies indicated that CD133+/CD34+ cells have a higher clonogenic capacity, both in vitro and in vivo, than CD133−/CD34+ cells (de Wynter et al, 1998; Gordon et al, 2003). In megakaryopoiesis, it has been demonstrated that the CD133+ subset contains primitive cells that are able to efficiently produce all categories of megakaryocyte progenitors (Charrier et al, 2002). Finally, after in vitro stimulation, CD133+ selected progenitors have been shown to convert into CD34− precursors with high multilineage engraftment capacity in vivo. In the mouse model, these cells gave rise to cells with T, B and natural killer (NK)-phenotype that, to date, has never been observed for CD34+ selected progenitors (Kuci et al, 2003).
These findings raise the possibility that CD133 may be the more important antigen in terms of multilineage engraftment and that convincing results of transplantations with purified CD34+ progenitors may in part be due to the fact that CD133 is co-expressed on the vast majority of these cells. Thus, our aim was to investigate the clinical feasibility and safety of CD133-based selection and transplantation in a small number of patients by gradually increasing the proportions of CD133 selected progenitors in standard CD34+ selected grafts from unrelated and mismatched-related donors. Furthermore, the efficacy of CD34+ and CD133+ selection procedures was cross-evaluated.
Patients and methods
Nine paediatric patients and one adult received transplants consisting of a combination of CD34+ and CD133+ selected progenitors from matched unrelated or from mismatched-related donors between July 2001 and January 2003. Informed consent was obtained from the legal guardians or patients, as appropriate.
The histocompatibility of each patient and donor was determined by high-resolution molecular [human leucocyte antigen (HLA)-A, -B and-DRB1] typing methods. Two patients had matched unrelated donors and eight patients had one to three loci mismatched, haploidentical parent (n = 7) or sibling (n = 1) donors. Age ranged from 1·2 to 38 years (median = 10 years) and body weight from 9·7 to 84 kg (median = 30 kg). Six patients had acute lymphoblastic leukaemia (ALL; T ALL = 3, B precursor ALL = 3), two had juvenile myelomonocytic leukemia (JMML), one had Wiskott-Aldrich syndrome (WAS) and one had severe aplastic anaemia (SAA) (Table I).
Table I. Diagnoses, donor mismatch, conditioning regimens and graft characteristics.
Patient ID no.
Total number of stem cells/kg
CD34+ selected cells/kg
CD133+ selected cells/kg
Portion of CD133+ added to the graft (%)
Total number of infused stem cells and amount of CD34+ and CD133+ selected cells × 10E6 per kilogram of patient body weight. The portions of CD133+ selected cells added to the graft are shown in ascending order.
Stem cell mobilization and purification of progenitor cells by CD34+ or CD133+ selection
All donors agreed to donate peripheral-blood stem cells (PBSC). Donor PBSC were mobilized by administration of 1 × 10 μg/kg of granulocyte colony-stimulating factor (G-CSF) daily for 5 d and were harvested by 1–3 leukapheresis procedures. We sought to obtain at least 10 × 106 progenitors per kilogram patient body weight (bw). Additional mobilizations were necessary in six mismatched-related donors. Selection of progenitors with anti-CD34 or anti-CD133 coated microbeads was carried out with the automated CLINIMACS device (Miltenyi Biotec, Bergisch Gladbach, Germany). In the first patient, about 10% of the collected progenitor cells were purified with anti-CD133 microbeads. The portion of CD133+ selected cells was successively increased from one patient to the next and reached 100% in the last patient. Each graft was split into two fractions of varying proportions, of which one was processed with anti-CD34 beads and the other with anti-CD133 beads. Thus, a minimum of two column runs was needed per patient.
When the number of residual T-cells was found to be >2·5 × 104/kg bw, a second depletion step was performed as previously described (Lang et al, 2002). Briefly, the selected progenitor cells were adjusted to 20 ml of separation buffer and incubated with 0·5 ml anti-CD3 magnetic beads (Dynal, Hamburg, Germany) for 20 min. The solution was then placed in a weak magnetic field and unbound progenitor cells removed.
Before and after separation, cell populations were stained with anti-CD34, anti-CD133, anti-CD3, anti-CD19 and anti-CD45 monoclonal antibodies (mAbs) and analysed by fluorescence-activated cell sorting (FACS) on FACScalibur instruments (Becton-Dickinson, Munich, Germany) according to the International Society for Hematotherapy and Graft Engineering guidelines (Leuner et al, 1998). Debris, dead cells, cell aggregates and platelets were excluded by gating on forward and side light scatter and subsequently on CD45+ propidium iodide negative cells. A minimum of 50 000 events was used for assessment. Cell viability was consistently >95%.
The myeloablative conditioning regimens were based on either total body irradiation (TBI, six fractions of 2 Gy each) or intravenous busulphan (12·8 mg/kg for age >3 years and 16 mg/kg for age <3 years), with specific modifications according to individual diagnosis and age (Table I). The adult ALL patient (ID no. 10) had rejected unmanipulated bone marrow and peripheral stem cells from a matched unrelated donor after myeloablative conditioning with TBI/ fludarabine (Flud)/cyclophosphamide (Cy). However, successful engraftment was achieved with CD133+ cells from her haploidentical sister after reconditioning with total lymphoid irradiation (TLI, 7 Gy), Flud (30 mg/m2/d ×4) and anti-thymocyte globulin (ATG) Fresenius® (Bad Homburg, Germany; 10 mg/kg/d × 3). G-CSF was routinely administered only in the first three patients. The other patients received G-CSF only in case of severe infections (which occurred only once, in patient no. 10). One patient with JMML received a T cell-replete graft (10 × 106 cells/kg) from a matched unrelated donor, followed by two doses of methotrexate (MTX) and a short course of cyclosporin A (CsA). Prophylactic post-transplant immunosuppression was not required in any of the other patients. Supportive care was carried out as previously described (Lang et al, 2003).
Assessment of engraftment, immune reconstitution and platelet recovery
The day of engraftment was defined as the first of three consecutive days on which the absolute neutrophil count (ANC) was >0·5 × 109/l. Reconstitution of CD3+, CD4+, CD8+, CD19+, and CD56+ lymphocytes was monitored by weekly FACS analysis until T cell recovery began and was subsequently assessed every 3 months.
Platelet recovery was defined as independence from platelet substitution for at least 14 d, with a platelet count routinely used to trigger such a transfusion of ≤20 × 109/l. The date of the last platelet transfusion was taken as the first day of recovery. The Kaplan–Meier method was used to evaluate the recovery probabilities of the CD133+/CD34+ selected group and of a historical cohort of patients from our institution. This cohort comprised paediatric patients who consecutively underwent transplantation between January 1995 and June 2001 with solely CD34+ selected progenitors from matched unrelated and mismatched-related donors and for whom platelet transfusion data were available (n = 76). Furthermore, an additional control group was created out of this cohort: to exclude patients with increased requirement for platelet transfusions, those with haemorrhagic cystitis, venous occlusive disease (VOD), aspergillosis, septicaemia with positive blood cultures, or GvHD ≥grade II were not considered for analysis. To equalize the amount of transplanted stem cells in the CD133+/CD34+ selected group and in the CD34+ selected group, only patients who received >8 × 106 cells/kg were accepted in the CD34+ selected control group. The mean numbers of progenitors were 20·5 ± 9·4 × 106 cells/kg (CD34+ selected group) and 28·2 ± 20·3 × 106 cells/kg (CD133+/CD34+ selected group; difference not significant, P = 0·29). Boosts given after secondary graft failure were excluded. All patients of the CD34+ selected group (n = 32) received G-CSF and the median age and body weight was 6·3 years (0·4–24 years) and 19 kg (4–64 kg) respectively. The diagnoses were acute leukaemias (n = 20), chronic myeloid leukaemia (CML, n = 3) and non-malignant diseases (n = 9).
Probabilities of survival and platelet recovery were evaluated with the method of Kaplan and Meier. Kaplan–Meier curves of platelet recovery were compared by using the log-rank test. The Wilcoxon rank sum test was employed for a cross-evaluation of purity, recovery and T cell depletion between both separation methods. Results are given in medians (range) unless otherwise indicated.
Comparison of CD133+ and CD34+ separation procedures
The purity and phenotype of target cells, contamination by residual T cells and recoveries in both methods were compared (Table II). A total of 20 CD133+ separations and 18 CD34+ separations yielded median purities of 93·4% (62·1–98·4) and 97·5% (38·0–99·2) respectively (P = 0·06). A median of 0·09% (0·05–0·38) contaminating T cells were detectable in the CD133+ selected stem cells, whereas CD34+ selection resulted in significantly less residual T cells [0·06% (<0·01–0·16), P = 0·002]. T cells were thus depleted more efficiently by CD34+ selection (4·1 log) than by CD133+ selection (3·7 log, P < 0·001). B cells were depleted less effectively than T cells and both methods produced similar results. No differences were observed between the recoveries of target cells (80·6% for CD133+ selection and 77·3% for CD34+ selection). Figure 1 shows representative FACS analyses after selection with anti-CD133 coated beads and after selection with anti-CD34 coated beads. The vast majority of stem cells was CD133+/CD34+ double positive for all separations (CD133+ selection: 93·09%; CD34+ selection: 92·22%, Table III). Significant and convincing populations of CD133+/CD34− cells were not detectable. Small CD133−/CD34+ subpopulations were enriched by CD34+ selection (1·5%) but not by CD133+ selection (0·21%). Thus, consideration should be given to the fact that the total number of yielded CD34+ progenitor cells will be slightly lower after CD133+ selection than after CD34+ selection.
Table II. Comparison of separation procedures.
CD133+ selected (n = 20)
CD34+ selected (n = 18)
Target cells (CD133+ or CD34+), as well as CD3+ T cells and CD19+ B cells, were determined by FACS analysis pre- and postenrichment (without second depletion step). Medians, ranges and P-values of the Wilcoxon rank sum test are shown.
nd, not done; FACS, fluorescence-activated cell sorting.
Total cell count
5·5 (2·0–8·7) × 1010
5·7 (3·6–11·8) × 1010
P = 0·45
258·5 (77·8–474·2) × 106
289·8 (85·8–775·4) × 106
P = 0·83
0·64 (0·2–1·01)% CD133+
0·52 (0·27–0·81)% CD34+
P = 0·48
93·4 (62·1–98·4)% CD133+
97·5 (38·0–99·2)% CD34+
P = 0·06
80·6 (42·4–114·2)% CD133+
77·3 (40·5–97·7)% CD34+
P = 0·49
P = 0·44
P = 0·002
3·75 (3·22–4·1) log
4·1 (3·5–5·6) log
P < 0·001
P = 0·67
Table III. Subpopulations of target cells after separation [median (range)], as detected by double staining (anti-CD34/anti-CD133).
ne, not evaluable.
The patients received a combination of CD34+ selected and CD133+ selected stem cells. The proportion of CD133+ selected cells was increased from one patient to the next from about 10% in the first to 100% in the last (Table I). A total of 29·3 (8·2–78·7) × 106 progenitor cells per kg bw were infused. These numbers include stem cell boosts, which were given in six patients 21–130 d after transplantation in order to stabilize the donor-derived granulopoiesis without using G-CSF. Boosting was considered if leucocyte counts fell under 1·0 × 109 cells/l.
The median number of residual T cells was 17 000 cells/kg (6000–29 500). To reach this goal even in small children, a second depletion step was carried out in four of 20 CD133+ separations (20%) and in two of 18 CD34+ separations (10·5%).
Initial engraftment occurred in 10 of 10 patients. The median time to an ANC > 0·5 × 109/l without G-CSF stimulation was 28 d (range 16–36 d; n = 6 patients).
Four patients received G-CSF (5 μg/kg) and had an ANC > 0·5 × 109/l at 11 d (range 9–18). Eight of 10 patients had sustained engraftment after initial transplantation.
Two patients with SAA and JMML, who had received multiple platelet transfusions for more than 1 year prior to transplantation, experienced late graft failure (rejection). However, both patients were successfully regrafted by reconditioning with steroids/OKT 3, ATG and T cell add-backs (SAA) or by reinfusion of stem cells with T cell add-back alone (JMML). Thus, all patients were finally engrafted.
Graft versus host disease
Primary acute GvHD. Seven patients (70%) showed no symptoms of primary acute GvHD. Three patients (30%) experienced grade I GvHD. Primary acute GvHD grade II–IV was absent.
GvHD after infusion of donor T cells. The patient with SAA received 5 × 105 donor T cells/kg bw from his matched unrelated donor because of an increasing mixed chimaerism in order to prevent another graft rejection. After this donor leucocyte infusion he returned to complete donor type but unfortunately experienced grade III GvHD.
Chronic GvHD. Patients were considered evaluable for chronic GvHD if they engrafted and survived for 100 d. None of the eight evaluable patients had chronic GvHD.
Five of the 10 patients are still alive (as at August 2003), with a median follow-up of 10 months (range 9–18 months, Fig 2). Four patients are free of disease. One patient relapsed after transplantation and is currently treated with a mild chemotherapy regimen. The causes of death were relapse (n = 2; JMML patients ID nos 5 and 7), infection (n = 2; adenoviral hepatitis, ID no. 2; systemic adenoviral and fungal infection, ID no. 1), or organ toxicity (n = 1; bronchiolitis obliterans organizing pneumonia in the adult patient ID no. 10). EBV LPD, veno-occlusive disease or haemorrhagic cystitis were not observed.
Platelet recovery and comparison with a historical control group
The median time to platelet recovery of the CD133+/CD34+ selected group was 13·5 d (Fig 3). Patients with (n = 4) and without (n = 6) G-CSF stimulation showed similar recoveries (15 vs. 13·5 d). One patient died from adenoviral hepatitis on day 25 post-transplantation, before he had been independent from platelet transfusion for at least 14 d.
These results from our CD133+/CD34+ selected group were compared with those from a cohort of patients treated at our institution with CD34+ selected stem cells between 1995 and June 2001. The median time to platelet recovery of the whole cohort was 32 d. Furthermore, all patients who had an increased requirement for platelet transfusions or received grafts with less than 8 × 106 CD34+ selected progenitors/kg were excluded, as described in ‘Patients and Methods. The median time to platelet recovery of this control group was 30 d. Thus, a faster recovery was observed in the CD34+/CD133+ group than in the whole cohort (P = 0·0027) or in the selected control group (P = 0·047).
Figure 4 shows the immune recovery of all haploidentical patients with haploidentical donors. CD56+ NK cells recovered quickly, and a clear NK cell peak was observed within 1 month after transplantation. No B-cell deficiency occurred and T-cell recovery was variable among the patients. A subgroup of patients who had received the current standard regimen [TBI, etoposide (VP16), Flud and ATG Fresenius®, but not OKT3, thymoglobulin (Merieux®, Lyon, France) or G-CSF, n = 4] showed a remarkably fast recovery of CD3+ T cells (mean numbers 30, 60 and 90 d post-transplant: 0·038, 0·196 and 0·338 × 109 cells/l respectively).
Our experience has shown CD34+ selection to be a useful tool in producing minimal GvHD in both closely matched unrelated and mismatched-related donors without any post-transplant immunosuppression (Handgretinger et al, 2001; Lang et al, 2003). However, primary engraftment (which has been c. 85%) and recovery of platelets and T cells (delayed in some patients) may be further optimized. We have thus investigated the feasibility and safety of CD133+-based selection and transplantation in a small number of patients by adding increasing proportions of CD133+ selected progenitors to a standard CD34+ selected graft.
After mobilization with G-CSF, a predominantly CD133+/CD34+ double positive donor progenitor population was observed and CD133+ selection as well as CD34+ selection resulted in similar purities and recoveries of these target cells. Although both methods produced a profound depletion of T cells, CD133+ selection was less effective than CD34+ selection. To not exceed our very low T cell threshold of 2·5 × 104 cells/kg, even in small children receiving extremely high stem cell doses from haploidentical donors, we performed a further reduction of T cells in 20% of CD133+ separations. We could thus maintain a very low incidence of primary GvHD in our CD133+/CD34+ patients, in agreement with the incidence in patients who received solely CD34+ selected grafts in the last 7 years. Thus, the selected progenitors are unlikely to induce GvHD themselves and graft manipulations on the basis of anti-CD133 mAbs may also prevent GvHD, provided that a critical threshold of T cells is not exceeded.
Sustained engraftment after initial transplantation was observed in eight of 10 patients, which corresponds to that of our CD34+ patient group (84%). The two patients who rejected their grafts were successfully reconditioned without causing organ toxicity. It is important to note that another patient who experienced graft failure after transplantation of unmanipulated bone marrow from an unrelated donor was successfully engrafted with CD133+ selected stem cells from her haploidentical sister. In several patients, decreasing neutrophil counts were observed 60–90 d post-transplant. Stem cell boosts were capable of stabilizing the donor-derived granulopoiesis without inducing GvHD in this situation. However, further clinical studies must be carried out to determine whether the use of CD133+ selected progenitors will improve the engraftment rate.
We have so far not observed any adverse side-effects attributable to CD133+ selection.
T-cell recovery was variable among our patients, although a favourable T cell recovery was observed in all haploidentical patients who had received ATG Fresenius® but no G-CSF. Thus, the impact of CD133+ selected cells is unclear and the use of G-CSF as well as the type of ATG may act as variables that influence regeneration. G-CSF, in particular, has been reported to interfere with cytokine production and regeneration of lymphocyte subsets (Volpi et al, 2001). Further studies must be carried out to address this issue.
Another interesting finding was that patients with additional CD133+ selected stem cells had a rapid platelet recovery. Moreover, these patients even showed a tendency for faster recovery than our historical cohort of patients transplanted with CD34+ selected grafts. The exclusion of patients with increased requirement for platelet transfusions or with lower CD34+ cell doses did not abolish this tendency. Although in vitro data provide support for this observation (Charrier et al, 2002), some aspects need careful consideration: first, the number of patients in our study is small. Secondly, the use of growth factors has been reported to impair platelet recovery (Keever-Taylor et al, 2001). However, this effect is unclear and remains a subject of controversial discussions (Gisselbrecht et al, 1994; Bernstein et al, 1998). In this report, all patients with CD34+ selected grafts but only four of 10 patients of our CD133+ selected group received G-CSF. Thus, the influence of G-CSF may be unlikely but cannot definitely be ruled out. Thirdly, it has been shown that a high CD34+ content of the graft is associated with faster recovery (Bernstein et al, 1998). This observation is in line with our mega dose concept, with the implication that high stem cell doses contribute to haematopoietic recovery. To eliminate this factor, we tried to adjust the median stem cell dose of the CD34+ patient group to that of the CD133+/CD34+ group. Although the difference between both groups was no longer statistically significant, the CD133+/CD34+ patients still received a slightly higher stem cell dose (28 vs. 20 × 106/kg). Therefore, we cannot rule out that the number of progenitor cells may also be involved in the faster recovery of CD133+ patients.
Immunophenotyping revealed no striking difference between CD133+ selected and CD34+ selected progenitors, as both populations consisted predominantly of CD133+/CD34+ double positive cells. Although rare populations of CD34+/CD133− cells were seen in CD34+ selected grafts, we were not able to detect convincing populations of CD133+/CD34− cells in CD133+ selected progenitors. However, the existence of this subset has previously been demonstrated: low percentages were found in cord blood (Gallacher et al, 2000) and inconsistently in peripheral blood after G-CSF mobilization (Gordon et al, 2003; Handgretinger et al, 2003). It has to be taken into consideration that our standard cytometry may have been insufficient to detect such rare populations. Furthermore, an overlap of fluorescence intensities did not allow differentiating potential low positive CD133+/CD34− cells from the negative fraction in some patients.
Apart from this, an interesting factor may be derived from experimental data suggesting that antibody coated microbeads might, to variable extents, activate intracellular signaling pathways influencing proliferation and differentiation of processed progenitors. Tada et al (1999) have demonstrated that cross-linking of the CD34 antigen on the cell surface induces an increase in tyrosine phosphorylation followed by cap formation and enhanced cytoadhesion. Similar CD133-mediated effects have, to our knowledge, not yet been reported. Thus, interactions between target cells and the antibodies used for their selection may be considered as effectors of outcome, even in phenotypically identical cells.
It has to be mentioned that CD34+/CD133− cells were lost by CD133+ selection. However, experimental data suggest that conversion of double positive cells into CD34+/CD133− cells in vivo may compensate for this.
In summary, we have demonstrated the feasibility of using stem cell progenitors selected with anti-CD133 coated microbeads from alternative donors. The preliminary clinical results presented here provide a basis for further studies in order to evaluate the efficacy of exclusively CD133+ selected grafts.
We thank Shangara Lal for critical reviewing of the manuscript and Olga Bartuli, Christiane Braun, Gabi Hochwelker, and Ulrike Krauter for excellent technical assistance. This work was supported by grants from the Deutsche Forschungsgemeinschaft (SFB 510) and from the Reinhold Beitlich Stiftung, Tuebingen, Germany.