Extensive phenotypic analysis of CD34 subsets in successive collections of mobilized peripheral blood progenitors


Dr Serge Leyvraz Centre Pluridisciplinaire d'Oncologie, CHUV - BH 10-505, 1011 Lausanne, Switzerland.


The transplantation of mobilized progenitor cells after high-dose chemotherapy shortens haemopoietic engraftment. CD34 cell subsets were examined in 20 consecutive mobilized progenitor cell collections obtained from patients with solid tumours that had not been previously treated. The analysis of CD34 cells was based on the expression of intracellular antigens, surface antigens including CD38, and cell size using multi-dimensional flow cytometry. We also correlated the numbers of stem cell subsets reinfused to haemopoietic recovery.

The majority of CD34+ cells expressed CD13 and CD33. A significant proportion was cytoplasmic myeloperoxidase (cMPO) positive. CD34+ MPO+ cells increased significantly in late collections. MPO expression was related to cell size. Cells expressing CD13 also increased in late collections in parallel to CFU-GM count. Small subpopulations of CD34+ CD38+ were committed to B cells, T cells and erythroid cell lineages. A small population expressing the megakaryocytic antigen had a small size and were predominantly CD38. A minor subpopulation expressed stem cells antigens. These were significantly higher in late collections (CD34+ Thy-1+ and CD34+ CD33). After mobilization, patients received three cycles of intensive chemotherapy followed by reinfusion of mobilized progenitors (5.45 × 106/kg CD34+ cells, range 3.4–11.88). The numbers of reinfused CD34 cells or the individual subsets did not influence recovery of leucocytes (9 d) or platelets (9 d).

In conclusion, the numbers of stem cells and their subsets differed between collections and, in unpretreated patients receiving intensive chemotherapy, there was no delayed engraftment when sufficient numbers of stem cells were reinfused. The recovery period was short and not correlated to any stem cell subsets.

The administration of either granulocyte colony stimulating factor (G-CSF) or granulocyte macrophage colony stimulating factor (GM-CSF) after myelosuppressive chemotherapy increases the number of circulating progenitor cells in peripheral blood (PB) ( Dührsen et al, 1988 ; Siena et al, 1989 ; Molineux et al, 1990 ; Kessinger & Armitage, 1991). The reinfusion of these cells after intensive chemotherapy results in shortening the recovery time for neutrophils and platelets ( To et al, 1992 ; Leyvraz et al, 1993 ).

The progenitor cells are phenotypically characterized by the expression of CD34 antigen and by their capacity for haemopoietic reconstitution. These cells are made up of two populations: (1) the primitive progenitors which have low rhodamine 123 uptake, high Thy-1 expression, lack the expression of maturation-linked surface antigens (Lin) and CD38 ( Andrews et al, 1989 ; Papayannopoulou et al, 1991 ; Terstappen et al, 1991 ; Bernstein et al, 1991 ; Craig et al, 1993 ), and (2) the committed progenitors which express antigens associated with myeloid (CD13, CD33), erythroid (CD71), B lymphoid (CD19), T lymphoid (CD7) and megakaryocytic lineages (CD61 and CD41) ( Vainchenker et al, 1982 ; Kurtzberg et al, 1989 ; Loken et al, 1987a , b; Pierelliet al, 1993 ).

Despite detailed studies on the characterization of CD34+ cells using multi-colour flow cytometry, insights into haemopoietic cell development and differentiation have been hampered by the limitation of cell surface studies alone ( Knapp et al, 1994 ).

Little is known about the expression of intracellular antigens by haemopoietic stem cells, in particular those mobilized into peripheral blood. These antigens are lineage specific and are used to define the precise sequence of events during cellular differentiation ( Knapp et al, 1994 ). Their expression was first detected in the cytoplasm and later on the membrane ( Campana et al, 1987 ; Mason et al, 1987 ; Janossy et al, 1980 ; Strobl et al, 1993 ). The expression of, e.g., cytoplasmic CD3 (cCD3) in T lymphocytes, cCD22 in B lymphocytes and c-myeloperoxidase in myeloid cells have helped in establishing the differentiation features and the commitment of the early precursors in normal bone marrow and peripheral blood.

Combination analysis of cell surface and intracellular antigens have been used intensively in the diagnosis of acute leukaemia, particularly in those with aberrant expression of different antigens or with mixed lineage ( Mirro et al, 1985 ; Janossy et al, 1989 ; Knapp et al, 1994 ). This approach has led to better classification of leukaemia and helped in decisions regarding patient prognosis and therapy ( Knapp et al, 1994 ). In PBPC such analysis may provide a more direct approach and better understanding of the specificity of a particular subpopulation of committed or uncommitted CD34 cells to be reinfused. The classification of progenitor cells infused may assist in the more precise prediction of the engraftment period than indirect methods such as colony-forming assays. Traditionally, measurements are made using mononuclear cell count, CFU-GM assay and phenotypic evaluation of CD34+ cells ( To et al, 1986 ; Kessinger et al, 1988 ; Fruehauf et al, 1995 ; Haas et al, 1995b ; Millar et al, 1996 ). None of these tests is a direct measure of stem cell content. The CFU-GM are committed myelomonocytic precursor cells and stem cells are a small minority of the cell population expressing the CD34 antigen ( Gordon & Blackett, 1995). As yet the role of various CD34 subpopulations in achieving rapid haemopoietic recovery is not well documented. As suggested in a recent report, the numbers of CD34+/CD33 and CD34+/CD41+ subpopulations may help in predicting short-term repopulating capability of PBPC collections ( Dercksen et al, 1995 ). The present study analysed the CD34 subpopulations by using a wide range of monoclonal antibodies and evaluating by flow cytometry the expression of lineage-specific intracellular antigens as well as membrane antigens. We based the analysis on cell size in order to follow-up the maturation pathway of CD34+ cells. We then phenotypically characterized and compared the expression of CD34 cell subpopulations in successive collections of mobilized peripheral blood progenitor cells and obtained a more accurate assessment of haemopoietic recovery by correlation with the number of the different subpopulations of progenitor cells.


Patients and collection of peripheral blood progenitor cells (PBPC)

20 leukaphereses were collected from eight patients for PBPC reinfusion after 24 cycles of high-dose chemotherapy. None of the patients had received prior cytotoxic treatment for malignancy. Haemopoietic progenitor cells were mobilized by 4-Epirubicine (FarmorubicineR, Pharmacia, Milan), administered intravenously at a dose of 75 mg/m2/d, for 2 d, followed by daily subcutaneous 5 μg/kg granulocyte-colony stimulating factor (G-CSF) (Filgrastim, Roche, Basel, Switzerland). When the leucocyte counts recovered >1.5 × 109/l, peripheral blood mononuclear cells were harvested with a continuous flow blood cell separator (Fenwal CS 3000, Baxter, Dietlikon, Switzerland). 8 litres of blood were processed during each leukapheresis. Patients had one leukapheresis per day until the targeted number of CD34+ cells was achieved (geqslant R: gt-or-equal, slanted6 × 106/kg). This target was chosen following the European Group for Blood and Bone Marrow Transplantation recommended criteria for obtaining a minimum number of 2 × 106/kg CD34+ cells or 2 × 108/kg of mononuclear cells for each reinfusion ( Marmont et al, 1995 ).

Mononuclear cells (MNCs) from leukapheresis products were collected after centrifugation on a density gradient (Lymphoprep, Nycomed, Oslo, Norway) and washed twice in phosphate-buffered saline (PBS). Aliquots containing 20–30 × 106 MNCs were frozen in RPMI containing 40% fetal calf serum (FCS) and 10% DMSO (dimethyl sulphoxide, BDH, Merck AG, Dietikon, Switzerland) for subpopulation analysis, see below.

The percentage of CD34+ cells was determined immediately on fresh samples using double-colour immunofluorescence staining (IF) with anti-CD45 monoclonal antibody (leucocyte common antigen, Becton Dickinson, Basel, Switzerland) (CD45-FITC/CD34-PE). Residual erythrocytes were depleted by lysis with isomolar ammonium chloride buffer for 10 min at room temperature (RT) and were subsequently washed in PBS. The CD34+ cells were identified by flow cytometry on the basis of high CD34 antigen expression and low CD45 antigen density.

Immunofluorescence staining, flow cytometry and colony assay

Stored cells from successive leukaphereses (day 1 to day 3) were analysed at the same time. Cells were rapidly thawed out at 37°C, then diluted in RPMI and washed. Viability of cells exceeded 80%.

In most of the three-colour IF combinations ( 1 Table I), intracellular antigens were identified on the FITC channel. These were anti-terminal deoxynucleotidyl transferase (TdT), anti-CD3, anti-CD22, anti-MPO and anti-glycophorin A (Gly-A). All the other specific marker antibodies reported below are used for the staining of cell-surface antigens. The CD34 antigen was detected on the channel for the third colour. Staining of the latter was achieved with tri-colour conjugated anti-CD34 antibody.

Table 1. Table I. Immunofluorescence staining combination for three-colour analysis of mobilized peripheral blood progenitor cells. CD34 antigen staining was achieved using the tri-colour-conjugated anti-CD34 antibody (Serotech, Dottikon, Switzerland). Reagents were directly labelled with the fluorochromes except for tri-colour and TdT (see Methods): (a) Serotech, Dottikon, Switzerland, (b) Becton Dickinson AG, Basel, Switzerland, (c) Dako Diagnostic AG, Zug, Switzerland, (d) Supertechs, Bad Zursach, Switzerland.Thumbnail image of

The discriminating reagents were identified on the PE channel. These were anti-CD38, anti-CD33, anti-CD13, anti-CD10, anti-CD19, anti-CD7, anti-CD71 and anti-CD41. In some experiments anti-Thy-1 (CDw90), anti-CD45RO, anti-CD61, anti-HLA-DR, anti-CD33, anti-CD10 and anti-CD7 were identified on FITC channel. 18 combinations were used to identify the differentiation pathway and the commitment of haemopoietic cells ( 1 Table I).

The expression of the platelet antigens CD41 and CD61 was measured after removal of platelets by density centrifugation (200 g for 20 min) and washing in PBS/0.2% bovine serum albumin (BSA) that contained 5 mmol/l EDTA to prevent platelet adhesion on CD34+ cells.

Cell suspensions (0.5–1 × 106) were incubated at RT with anti-CD34 conjugated to biotin. After incubation and washing, a mixture of antibodies conjugated to PE, FITC and tri-colour-conjugated streptavidin (Caltag Laboratories Inc., SA) were added and cells were incubated and washed to identify the expression of cell surface antigens ( Table I). For the detection of intracellular antigens, FITC-conjugated antibodies were used for staining after fixation and permeabilization of the prelabelled cells with Ortho PermeafixTM (OPF, Ortho Diagnostic Systems Inc., Raritan, N.J., U.S.A.) ( Pizzolo et al, 1994 ; Francis & Connelly, 1996). Of this fixative, 1 ml diluted in 1:1 ratio with distilled water was incubated for 40 min at RT and cells were washed.

For direct staining, cytoplasmic antigens (c) were stained with anti-CD3 (UCHT-1, 1:10), anti-CD22 (RFB4, 1:5), anti-MPO (MPO-7, 1:50) and anti-Gly-A (JC 159, 1:10) conjugated to FITC for 30 min at RT and washed. For indirect staining, the cells were incubated with rabbit anti-TdT (1 :100) for 30 min at RT followed by washing and further incubation for 30 min with affinity purified goat anti-rabbit Ig (F(ab′)2-FITC, 1:100) (Ready System AG, Zurich, Switzerland). Isotype-matched irrelevant antibodies served as controls.

Flow cytometric analysis was performed on the FACScan equipped with an argon laser tuned to 488 nm. Forward light scattering, orthogonal light scattering and the three-colour immunofluorescence signals were determined for each cell and stored in the listmode data files. Paint-a-Gate and Lysis II softwares (Becton Dickinson AG, Basel, Switzerland) were used for five-parameter analysis which utilized two scatter and three IF gates ( Table I). A minimum of 50 000 events were acquired in listmode data file. The phenotype of CD34+ cells was determined by analysing 5000 cells inside the window gate of the CD34+ cells which was established according to the low side-scattering properties and high fluorescence intensity for anti-CD34 monoclonal antibody. The percentage of CD34+ cells that co-expressed a specific antigen was assessed after correction for the percentage of cells reactive with an isotype control. Further analysis of cellular expression of different antigens was performed on the total CD34 cell population by gating on the lymphoid cell population (region R1) and dividing the CD34 cells by their forward light scatter properties into small (region R2) intermediate (region R3) and large size (region R4) (Fig 1). Once determined, the regions were constant for all samples. The instrument settings and channel compensations were controlled by calibrated fluorescence reference beads (FACS CaliburTM, Becton Dickinson AG, Immunocytometry Systems, Basel, Switzerland) and by staining of normal lymphocytes with three-colour IF (CD3-FITC, CD4-PE and CD8-tri-colour).

Figure 1.

and then were divided by further gating into small (R2), intermediate (R3) and large (R4).

For in vitro colony assay, MNC obtained from fresh samples were plated in four replicate 1 ml cultures for BFU-E, CFU-GM and CFU-Mix as previously described ( Beck et al, 1991 ; Healy et al, 1994 ). The plates were incubated at 37°C in a humidified atmosphere containing 5% CO2 in air. Colonies were counted on day 14.

Sequential high-dose chemotherapy

24 reinfusions were performed in a group of eight patients including five patients with breast cancer, two with small cell lung carcinoma and one with germ cell tumour. The patients' ages ranged from 39 to 58.5 years (median 46 years). The sequential high-dose chemotherapy treatment ( Leyvraz et al, 1997 ) consisted of ifosfamide 10 g/m2, given by continuous infusion for 4 d. Carboplatinum 1200 mg/m2 and etoposide 1200 mg/m2 were administered by 3 h infusions for 4 d. This intensive ICE regimen was repeated every 28 d for three cycles. Patients received their PBPCs 48 h after chemotherapy had been stopped, followed by administration of G-CSF (filgrastim) 5 μg/kg until the leucocyte count reached geqslant R: gt-or-equal, slanted5 × 109/l.

Statistical methods

The hypothesis was to find a trend of increase or decrease in the absolute number of CD34 cells expressing a given antigen over successive leukaphereses. All patients had two collections and four patients had three. The comparison was therefore between the absolute number of specific CD34 subsets in first and in second leukaphereses. In patients with three collections the average of the second and the third leukapheresis was also calculated and compared to the absolute number from the first. The t-test for paired observation was applied on non-transformed data.

Two time intervals describing haematological recovery were of interest: (i) from reinfusion to leucocyte count geqslant R: gt-or-equal, slanted1.0 × 109/l, and (ii) from reinfusion to platelet count geqslant R: gt-or-equal, slanted 20 × 109/l. Time intervals were transformed to natural logarithms in order to linearize their potential association with the number of reinfused cells expressing any given antigen.

Simple statistical summaries were used to describe the absolute numbers and percentages of subpopulations of CD34 cells. All data analyses were performed by means of the statistical package Stata (StataCorp. 1995, Stata Statistical Software: Release 4.0, College Station, Texas: Stata Corporation). The analyses reported were exploratory and conducted over a large number of endpoints: significance levels of the tests performed have therefore to be considered as indicators of possible tendencies. Results are reported in terms of two-sided P-value for the coefficient of the linear term for the antigen of interest. Time to haematological recovery was compared by means of log-rank test.


Peripheral blood progenitor cell collections

Twenty leukaphereses were performed between day 11 or 12 after starting the mobilization procedure; in five patients the first leukapheresis was collected on day 11, and on day 12 in the three remaining patients. There was no difference in peripheral blood leucocyte count between the days of collection (data not shown). Table II shows the peripheral blood leucocyte count and the content of each leukapheresis product with regard to MNC count and number of CD34+ cells in successive collections performed on days 1, 2 and 3. As shown in Table II, the absolute number of CD34+ cells × 106/kg showed a trend of increase on day 2 and 3 as compared to day 1.

Table 2. Table II. The number of cells present in successive collections of peripheral blood progenitor cells (PBPC). * Day 1 versus day 2 and 3.Thumbnail image of

Phenotype of the CD34+ subpopulations in a CD34/SSC-dot plot

The phenotypes of the CD34+ subpopulations that are linked to early multipotent stem cells, as well as the differentiation of haemopoietic cells into different lineages are shown in Fig 2(A–D). Among cells expressing the CD34 antigen, considerable proportions expressed the stem-cell-associated antigens, Thy-1 (median 24.8% ; range 12–45.5%) and CD45R0 (median 22.1% ; range 10.8–40.7%). A small population of CD34+ cells lacked the expression of CD38 (median 3.15%; range 1–7%), HLA-DR (median 5% ; range 1–17%) and CD33 antigens (median 11% ; range 2–38%) (Fig 2A), representing a similar phenotype to that of multipotent stem cells. The great majority of CD34+ cells strongly expressed the CD38 and HLA-DR antigens (data not shown).

Figure 2.

, CD10 and CD19 antigens, demonstrating the commitment to the T- and B-lymphoid lineages.

For the expression of myeloid-lineage associated antigens, the majority of CD34+ cells strongly expressed the pan-myeloid antigen CD13 (median 92% ; range 87–98%). Similarly, a high proportion of cells expressed the CD33 antigen (median 89% ; range 62–98%) but with moderate intensity (see below); only a subpopulation of these cells showed a wide range of cMPO expression, the myeloid-granulocytic lineage-specific marker (median 18.3% ; range 5.7–54%) (Fig 2B).

The intensity of CD33 antigen expression seemed to vary with the fluorochrome used. It was considerably weaker on cells stained with CD33-FITC mAb than CD33-PE (10 samples tested) resulting in difficult discrimination between CD33+ and CD33 cells (data not shown).

The expression of megakaryocytic lineage-associated antigens including CD41 and CD61 was detected in a small subpopulation of CD34+ cells. A median of 7% of the CD34+ cells expressed the CD41 antigen (range 4.3–11.2%), and CD61 was detected on a median of 3% (range 0.9–7.1%) (Fig 2C).

Finally, small subpopulations of CD34+ cells expressed lymphoid- and erythroid-associated antigens. A fraction of CD34+ cells expressed the T-lineage-associated antigens CD7 (median 8.7% ; range 3.2–19.7%) and cCD3 (median 5.3% ; range 1.3–18.2%). The B-lineage-associated antigens were CD10 (median 6.4%; range 2–19.5%), cCD22 (median 2.7% ; range 0.5–8.7%) and CD19 (median 2.3% ; range 1–4%). A considerable proportion of CD34+ cells expressed either B- or T-lineage-associated antigens and were TdT positive (median 8.1% ; range 4–18.4%) (Fig 2D).

The erythroid lineage-associated antigen cGly-A was detected in an even smaller percentage of CD34+ cells (median 1.8% ; range 0.5–3%) (Fig 2C).

Cellular composition of the different subpopulations of CD34+ cells

The CD34+ cells were analysed for their antigen expression according to cell size. Based on their scatter features (forward and orthogonal side-scatter), the cells were divided into small, intermediate and large (regions R2–R4) within the lymphoid cell population (region R1, Fig 1).

The average frequency of cells expressing the CD34 antigen from 20 leukapheresis samples was 2.95% (range 1.3–9.9%). The average frequency of cells expressing the CD34 antigen increased with increasing cell size. Only 7.1 ± 4.06% CD34+ cells were small, 26.9 ± 11.6% intermediate and 66 ± 12.9% were large cells.

To assess the phenotypic composition of CD34+ subpopulations during lineage commitment, we analysed simultaneously the intracellular and surface antigen expression and extended our analysis to the scatter of cells (small, intermediate and large CD34+ cells) (Fig 1). We also studied the expression of CD38 antigen as the first sign of cellular differentiation ( Terstappen et al, 1991 ) and correlated it with the lineage commitment of CD34+ progenitor cells. Forward light-scattering, orthogonal light-scattering and three immunofluorescence parameters were determined in each region.

Region R2 contained small CD34+ cells. Amongst these cells the pluripotent stem cells [CD34s+ (strong positive)/CD38] (15–45%) were Thy-1+ (60–70%) and CD45RO+ (15–50%) and lacked the expression of intracellular antigens (<2%). Unexpectedly, 20–25% were strongly CD61+ (the megakaryocytic lineage-associated antigen).

Region R3 contained intermediate CD34+. The most observed finding was the acquisition of the differentiation antigen CD38 by the majority of CD34+ cells accompanied by the appearance of granulocyte-associated antigen cMPO (5–15% positive). 20–30% of CD34s+/CD38+ also expressed CD45RO and Thy-1. 5–25% of CD34+ cells expressed the B-lineage-associated antigens (cCD22+, TdT+, CD10+ and CD19+).

Region R4 contained large CD34+ cells of which a substantial proportion were cMPO+ (20–60%). The expression of cMPO increased with increasing cell size accompanied by a concordant decrease in CD34 antigen density (Fig 3). Amongst the CD34s+, 10–20% expressed CD45RO of which 20–50% were CD71+ and 2–15% were cGlyA+ showing indication of erythroid-lineage commitment; 5–12% of CD34s+ cells expressed the T-lineage-associated antigens (cCD3, CD7 and TdT).

Figure 3.

8 antibodies, cells were permeabilized with Ortho PermeafixTM and stained for anti-MPO. MPO expression increased with increasing cell size and between days of collection.

These results demonstrated the presence of different subpopulations of CD34 cells that varied in maturation according to cell size. The larger cells were the more differentiated. The multipotent stem cells were mainly small. The expression of cMPO was the most predominant intracellular antigen. Its expression was detected in intermediate cells expressing the CD38 antigen and increased in large cells and in cells with low-density CD34. The expression of cMPO also increased during successive PBPC collections and is directly related to cell size and differentiation.

Absolute counts of CD34 subpopulations in successive PBPC collections

We studied whether the timing of PBPC collection may influence the count of a particular CD34 subpopulation. The absolute count of mononuclear cells, CD34+ cells and various CD34 subpopulations defined on the basis of specific antigen expression from day 1 and day 2 of collections were compared. The results show a significant increase in the absolute count of mononuclear cells collected on day 2 compared to day 1 (P = 0.007). The percentage of CD34 cells in successive collections remained the same, 2.7% (range 1.36–4.5%) for day 1 and 2.7% (range 1.35–7.0%) for day 2, but a trend of increase in the absolute numbers was found (P = 0.07). Within the stem cell compartment, significantly higher counts of both the CD34+/Thy-1+ (P = 0.008) and the CD34+/CD33 (P = 0.02) subpopulations were found. Amongst the myeloid cell compartment, a significant increase for CD34+/MPO+ subpopulation (P = 0.02) and a trend for increase in CD34+/CD13+ subpopulation (P = 0.08) were found. No significant differences in the counts of CD34 cells committed to megakaryocytic lineage, erythroid lineage and B- or T-lymphoid lineages were detected. For patients who had three collections an average of the absolute counts for various CD34 subpopulations collected on days 2 and 3 were compared to day 1. Similar results were obtained confirming that higher numbers of primitive haemopoietic cells and more committed myeloid progenitors were obtained in late collections than in early collections. An evaluation of various CD34 subpopulations was carried out using clonogenic assays. There was a significant increase in CFU-GM colonies (21.82–29.32 × 104/kg; P = 0.008) between days 1 and 2 of collection, in agreement with the increase observed in myeloid cell commitment determined by flow cytometry.

Correlation between reinfused CD34 subpopulations and recovery of leucocytes and platelets

The correlation between the numbers of reinfused total CD34+ cells as well as CD34+ subpopulations and time to leucocyte and platelet recovery was assessed after PBPC transplantation following 24 cycles of high-dose chemotherapy ( Table III). The numbers of reinfused CD34+ cells were not statistically different between the successive cycles (P = 0.644). Medians of 5.29 × 106/kg (range 3.4–8.24), 5.45 × 106/kg (range 4–10.92) and 5.45 × 106/kg (range 4.25–11.88) were reinfused after the first, second and third cycles. The time from PBPC reinfusion to leucocyte recovery geqslant R: gt-or-equal, slanted1 × 109/l was 9 d (range 7–10 d) and to platelet recovery geqslant R: gt-or-equal, slanted20 × 109/l, 9 d (range 7–12 d). No statistical differences were found amongst the three cycles with regard to leucocyte (P = 0.68) and platelet recovery (P = 0.13).

Table 3. Table III. Correlation between number of reinfused CD34+ cells and time to leucocyte and platelet recovery. * Median time to platelet recovery 9 d (range 7–12 d).† Median time to leucocyte recovery 9 d (range 7–10 d).Thumbnail image of

Looking specifically at the impact of reinfusing individual subsets on haemopoietic recovery, we used a regression analysis to calculate the time intervals from reinfusion to achieving a leucocyte count of geqslant R: gt-or-equal, slanted1.0 × 109/l and a platelet count of geqslant R: gt-or-equal, slanted20 × 109/l. There was no correlation between the numbers of infused CD34+/CD33 primitive haemopoietic precursors and leucocyte recovery. The time from reinfusion to leucocyte recovery ranged from 7 to 10 d (median 9 d) when a range of 0.25–4.37 × 106/kg (median 0.58 × 106/kg) CD34+/CD33 were reinfused. The numbers of reinfused CD34+/CD41+ (megakaryocytes) (range 0.25–2.69 × 106/kg, median 0.45 × 106/kg) did not influence the length of platelet recovery (range 7–12 d, median 9 d).

None of the CD34 subsets (CD34+/CD38, CD34+/HLA-DR, CD34+/Thy-1+, CD34+/CD33+, CD34+/cMPO+, CD34+/CD13+ and CD34+/CD61+) were better correlated with time to either leucocyte or platelet recovery than the total number of CD34+ cells ( 3 Table III).

There were no significant correlations when reinfused cells were assessed by colony-forming assays such as CFU-Mix and BFU-E (data not shown).


In this study we showed that mobilized peripheral progenitors form a heterogenous mixture of cells. Phenotypic analysis of CD34 subsets demonstrated the presence of several cell types belonging to different lineages.

Only a small percentage of CD34+ cells expressed the multipotent stem-cell-associated antigens Thy-1 and CD45RO. In addition, small subpopulations of lineage-negative cells (CD34+/CD38, CD34+/CD33 and CD34+/HLA-DR) were also identified. The majority of cells (>62%) expressed myeloid lineage-associated antigens. Small but distinct subpopulations of CD34+ cells expressed the megakaryocytic lineage-associated antigens and a limited number of cells expressed erythroid and T- and B-lymphoid antigens.

These results are consistent with the previously reported data ( Steen et al, 1994 ; Inaba et al, 1994 ; Bender et al, 1992 ; Fukuda et al, 1994 ). New subpopulations of CD34+ cells emerged in the present analysis based on CD34+ cell size, intracellular antigen expression and the acquisition of the differentiation antigen CD38 ( Terstappen et al, 1991 ). We have identified for the first time three new subpopulations of mobilized peripheral blood progenitors.

Small size CD34s+ cells (20–25%) lacked the expression of the CD38 antigen but expressed CD61 (the megakaryocytic lineage-associated antigen) suggesting an early commitment to the megakaryocytic lineage. This is in agreement with a previous report ( Buscemi et al, 1995 ) showing a strong correlation between time to platelet recovery and reinfusion of CD34+/CD38 subpopulation and a recent report demonstrating that the negative depletion of lineage-positive CD34 cells produced a significantly larger quantity of megakaryocytic progenitors ( Bertolini et al, 1997 ). Another support for this finding comes from our data on the ex-vivo expansion of megakaryocytic lineage. We obtained a substantial increase in CD61+ cells between days 12 and 14 despite extremely low numbers of colony-forming units (CFU-mega), suggesting that the majority of early megakaryocytic progenitors cannot be detected by means of colony assay (unpublished observations).

Intermediate size CD34s+ cells (20–30%) expressed the differentiation antigen CD38 as well as Thy-1 and CD45RO (the ‘stem-cell-associated’ antigens) but lacked the expression of lineage-associated antigens (CD7, CD10, CD33) and TdT, suggesting an early stage of committed haemopoietic progenitor cell differentiation. The expression of Thy-1 by CD34+ cells as previously shown is not homogenous and contains not only primitive progenitor cells but also progenitors producing colony-forming unit granulocyte macrophages ( Humeau et al, 1996 ).

Intermediate to large size CD34+ cells (5–60%) expressed intracellular myeloperoxidase (cMPO). These cells were CD33+ and CD13+. The expression of MPO increased amongst the large cells accompanied by a concordant decrease in CD34 antigen density demonstrating the commitment to myeloid-granulocytic lineage.

MPO was the most predominant intracellular antigen amongst the mobilized CD34 progenitors. This is in agreement with previous reports showing that G-CSF induces myeloid cell differentiation of stem cells ( Fukuda et al, 1994 ). Similarly, cMPO expression was reported in normal bone marrow mainly in large CD34 progenitors with low-density CD34 antigen ( Strobl et al, 1993 ). Other intracellular antigens, e.g. cCD22 and cCD3 (B- and T-lymphoid), TdT (immaturity) and cGly-A (erythroid), were also expressed but only in small subpopulations of CD34+ cells ranging in size from small to large accompanied by the expression of one or more surface lineage-associated antigens. The identification of new subpopulations at early stages of cellular differentiation or cells expressing lineage-associated molecules gives new insight into the process of progenitor cell differentiation and the generation of the cellular elements of the blood. This could further refine the process of progenitor cell transplantation. The expression of, e.g., CD61 by CD34s+/CD38 cells can be used in an experimental approach to select or increase the megakaryocytic precursors, most of which are likely to represent the progenitor cells responsible for early platelet rescue. Since we identified CD34 subsets in PBPC collections, we analysed these subpopulations in each of the leukapheresis collections (days 1, 2 and 3). There were no differences between the percentage of any of the subsets amongst the leukapheresis collections; this is in agreement with an earlier report ( Inaba et al, 1994 ). However, differences were found when the absolute numbers rather than percentages of these cells were compared. We obtained significantly higher yields of lineage-negative multipotent stem cells and more cells committed to the myeloid/granulocytic lineage in later collections (days 2 and 3) compared to day 1. We obtained median absolute numbers of 0.62 × 106/kg CD34+/CD33 cells and 1.6 × 106/kg CD34+/Thy-1+ when the leucocyte count reached 15 × 109/l (day 2) compared to 0.23 × 106/kg CD34+/CD33 and 1.26 × 106/kg CD34+/Thy-1+ when the leucocyte count was <6.5 × 109/l (day 1), P = 0.02 and P = 0.08, respectively.

Similarly, when the leucocyte count in peripheral blood reached 15 × 109/l we obtained a significant increase in CD34+/MPO+ cells (1.53 × 106/kg) in late collections compared to early collections (0.88 × 106/kg) (P = 0.02). We also observed a trend of increase in CD34+/CD13+ (6.74 × 106/kg) in late, compared to 4.87 × 106/kg in early, collections (P = 0.08). In line with the phenotypic results, we obtained a significant increase in CFU-GM colony numbers in late (29.32 × 104/kg) compared to early leukapheresis (21.82 × 104/kg).

Our results demonstrate that late leukaphereses contain sufficient numbers of primitive progenitor cells for engraftment as confirmed by CD34 subset analysis ( Haas et al, 1995a ). The presence of higher numbers of primitive haemopoietic cells in a PBPC transplant may contribute to a better quality long-term reconstitution ( To et al, 1992 ). The CFU-GM colony assay has been used to predict haemopoietic reconstitution by correlating it with numbers of mononuclear cells and CD34+ cells ( Kessinger et al, 1988 ; To et al, 1986 , 1997; Faucher et al, 1996 ). In some studies a strong correlation between CFU-GM colony count and CD34+ cells was reported ( Haas et al, 1995b ; Millar et al, 1996 ), but in others no correlation was found ( Serke et al, 1996 ). These differences may be related to the sensitivity of the assay and the phenotypic characterization of CD34 cells ( Fruehauf et al, 1995 ).

In the present study we combined the phenotypic analysis of CD34+ subsets and the clonogenic assays. The data shows that the increase in CFU-GM colony numbers in late collections paralleled the increase of those cells expressing the myeloid/granulocytic phenotype (CD34+/MPO+). No such increase was found in CFU-Mix colony count (data not shown) as compared to percentage of CD34+/Thy-1+ or CD34+/CD33 primitive progenitors. Probably the LTC-IC assay would have been the best candidate for such analysis. We believe that phenotypic determination of CD34+ subsets is reliable and facilitates a rapid quantitation of all the different subpopulations of CD34+ cells ( Dercksen et al, 1995 ). This type of analysis can be used routinely in clinical laboratories.

An important clinical issue is whether a relationship exists between CD34 cell number and the rate of haemopoietic recovery after PBPC transplantation. It is known that a correlation exists between reinfused CD34+ cell number and haemopoietic recovery with a minimal threshold of progenitor cells around 2–2.5 × 106/kg ( Haas et al, 1995b ; Millar et al, 1996 ). However, with less cells being reinfused, the time to platelet recovery was prolonged ( Millar et al, 1996 ; Ketterer et al, 1998 ). Moreover, the value of an optimal threshold is not well known, and recent data suggests a potential benefit of higher CD34+ cell doses for reinfusion ( Ketterer et al, 1998 ). In the present study the numbers of reinfused CD34+ cells (5.45 × 106/kg, range 3.4–11.88) did not influence recovery of leucocytes (9 d, range 7–10 d) or platelets (9 d, range 7–12 d). This was in contrast to the data reported by Dercksen et al (1995 ) where a correlation with neutrophil recovery was found when geqslant R: gt-or-equal, slanted6.0 × 106 CD34+ cells/kg (10 d, range 8–21 d) or <6.0 × 106/kg CD34+ cells were reinfused (11 d, range 9–8 d).

Studying individual subsets rather than the total CD34 population might shed more light on the mechanism of haemopoietic recovery. In one study a significant correlation was found between the number of CD34+/CD33 cells with neutrophil recovery ( Dercksen et al, 1995). Reinfusion of geqslant R: gt-or-equal, slanted2.79  × 106/kg CD34+/CD33 resulted in shorter recovery time (10 d, range 8–17 d) than when <2.79 × 106/kg cells were reinfused (12 d, range 10–28 d). In our patients the numbers of reinfused CD34+/CD33 cells (0.58 × 106/kg, range 0.25–4.37) did not influence leucocyte cell recovery. The time from reinfusion to recovery was 9 d (range 7–10 d). Reinfusion of cells committed to the myeloid lineage (CD34+/MPO+ or CD34+/CD13+) did not influence leucocyte recovery either.

The median megakaryocyte recovery was 9 d when 0.45 × 106/kg CD34+/CD41+ (range 0.25–0.69 × 106/kg) was reinfused. All the patients achieved platelet recovery (geqslant R: gt-or-equal, slanted20 × 109/l) in less than 12 d, unlike the previous report ( Dercksen et al, 1995 ) where a longer recovery time (11 d, range 7–16 d) was achieved when geqslant R: gt-or-equal, slanted0.54 × 106/kg of CD34+/CD41+ cells/kg were reinfused. However, when <0.54 × 106/kg CD34+/CD41+ cells were reinfused, a delay in platelet recovery was seen (19 d, range 9–37 d). Only 16.1% of patients from the latter group recovered within 14 d.

The data presented in the present report did not reveal a significant correlation between the subpopulations of stem cells and haemopoietic recovery. This may be related to the small number of patients studied, the use of partially ablative chemotherapy, and the chemonaivity of the patients. These factors probably contributed to the rapid engraftment. Thus, the discrepancy between the results reported in the present study and previous reports may be related to differences in patient characteristics, such as the degree of pretreatment, different chemotherapy regimens, and different protocols for collections and transplantation. The analysis of CD34+ subsets might still provide an important tool in monitoring heavily pretreated patients undergoing stem cell transplantation, or in patients receiving ex-vivo processed or gene modified progenitor cells.


The work was supported by grants from La Recherche Suisse contre le Cancer (KFS 170-9-1995), La Ligue Vaudoise contre le Cancer and La Société de la Loterie de la Suisse Romande.

We thank Dr Philippe Jaunin, Mrs Eveline Faes, Mrs Katia Balmers and Mrs Christine Beretta for superb technical assistance. Special thanks to Mrs Martine van Overloop for typing the manuscript.