Circulating Primitive Stem Cells in Paroxysmal Nocturnal Hemoglobinuria (PNH) Are Predominantly Normal in Phenotype But Granulocyte Colony-Stimulating Factor Treatment Mobilizes Mainly PNH Stem Cells

PAROXYSMAL NOCTURNAL hemoglobinuria (PNH) is an acquired, clonal hemopoietic disorder that develops when a somatic mutation of the PIG-A gene occurs in a hemopoietic stem cell in a susceptible individual.1,2 The PIG-A gene product is involved in the first identifiable step in the biosynthesis of glycosylphosphatidylinositol (GPI) anchors by which a range of molecules are attached to the cell membrane.3Thus, PNH cells lack GPI-linked molecules including CD59/membrane inhibitor of reactive lysis (MIRL) and CD55/Decay Accelerating Factor (DAF), which limit the activity of complement. PNH cells are therefore abnormally sensitive to complement, leading to hemolysis of red cells and possibly activation of platelets.4-6 Clinically, this results in hemolytic anemia and a tendency for thrombosis. PNH arises in conditions of marrow hypoplasia and there is evidence for a relative growth advantage of the PNH clone over normal hemopoiesis in such states.7-10 The mechanism of this selection and the level of cell maturity at which it operates is unknown. We have investigated this by using flow cytometry to analyze the proportion of cells that exhibit a PNH or normal phenotype at different stages of maturation. Lack of CD59 expression was taken to define the PNH cells. The combination of CD34 positivity with the level of expression of CD38 was used to identify stem cells of greater or lesser degrees of maturity. The CD34⁺/CD38⁻ cell fraction is known to be enriched for progenitors that can differentiate into myeloid or lymphoid lineages and represents one of the earliest cell types identifiable by flow cytometry.11 We have studied the steady-state peripheral blood stem cells (PBSCs) from four PNH patients in this way. It has been previously reported in PNH patients that the most primitive stem cells circulating in the blood are largely of normal phenotype, even when the marrow CD34⁺ cells are almost all PNH.12,13 This leads to the possibility that these normal cells might be collected from the peripheral blood with a view to possible reinfusion as an autologous transplant. The baseline levels of these CD34⁺ cells are low and therefore an attempt to mobilize normal PBSCs into the peripheral blood by treatment with granulocyte colony-stimulating factor (G-CSF) was made. A regimen similar to that which is used for collection of PBSCs from volunteer donors in allogeneic peripheral stem cell transplantation was used. The change in cell numbers and phenotype after administration of G-CSF was assessed on mature neutrophils and early- and late-stem cells (defined as CD34⁺/CD38⁻ and CD34⁺/CD38⁺, respectively).

Four patients were investigated. They are identified throughout this report by a unique patient number that corresponds to their reference number in the PNH registry maintained in our institution (Table 1). They had all been previously diagnosed in our department and had classical hemolytic PNH. Diagnostic criteria included a positive Ham acid lysis test and the demonstration on red cells and neutrophils of a population of CD59/CD55-deficient cells by flow cytometry. Baseline clinical status and typical hematological values for the patients are shown in Table 1. Patient no. PNH043 was on treatment with Cyclosporin A for aplastic anemia (his original presentation) and all patients were prophylactically warfarinized. The subjects all had large PNH clones as judged by neutrophil surface phenotype (median 87% PNH, range 50 to 97; see Table 2). Before the enrollment into the study, full ethical approval was obtained from the hospital ethical committee according to local practice. Written, informed consent was then obtained from all patients.

PBSCs were mobilized using G-CSF alone (Lenograstim; Chugai Pharma, London, UK) at a dose of 10 μg/kg/d by subcutaneous injection for 4 consecutive days. This was administered on an outpatient basis with daily monitoring of hematological parameters. Phenotypic analysis of neutrophils and CD34 subsets was performed on steady-state peripheral blood and on day 5 (1 day post–G-CSF).

PI-PLC (Oxford Glycosystems, Abingdon, UK) was used at a concentration of 1 U/mL. After washing, 100 μL of PI-PLC was added to 1 × 10⁶ cells in a microtiter well and incubated for 60 minutes before being washed and stained with the appropriate monoclonal antibodies before fluorescence-activated cell sorting (FACS) analysis. Appropriate non–GPI-linked controls were tested in parallel (eg, CD15 for neutrophils).

Leucocytes were prepared by whole blood lysis using a 10-fold excess of ammonium chloride solution (8.6 g/L) at 37°C for 5 minutes, after which they were washed and resuspended in FACSFlow (Becton Dickinson, Oxford, UK). In some cases, CD34⁺ cell subsets were analyzed in the mononuclear cells fraction prepared by density-gradient centrifugation over Lymphoprep (Nycomed, Birmingham, UK).

For antibody staining, 2 to 3 million cells were incubated in microtiter plates for 20 minutes at room temperature with 10 μL of each pretitered antibody conjugate per million cells, then washed twice in FACSFlow. Cells were analyzed on a Becton Dickinson FACSort with CELLQuest software. The antibodies used in this study were (hybridoma name in brackets): CD34 PE (Birma-K3); CD45 fluorescein isothiocyanate [FITC] (9.4); CD38 Cy5 (OKT10); IgG1 control PE/Cy5 (PAP7)-IgG2a control PE/Cy5 (W6/32HK); and CD59 FITC. CD59-FITC was supplied by Cymbus Biosciences UK Ltd (Southampton, UK). All other conjugates were prepared in house.

For CD34 quantitation, cells were incubated with CD34 PE and CD45 FITC, washed, and resuspended in 150 μL FACSFlow with Propidium Iodide (PI). A 50,000 event file was collected. Regions were set around the viable leucocytes (CD45⁺PI⁻) and progenitor cells (CD34⁺SSC^lo), the latter being a wide gate including some contaminating events. A further 200 to 500,000 events that satisfied both these regions were then acquired. A tight CD34 region was then set around the progenitor cells using their CD34 versus CD45 characteristics. FSC, SSC, and PI of these gated cells were assessed to ensure that no apoptotic cells or debris had been included. CD34⁺ cells were calculated as a percentage of viable leucocytes. An equivalent number of leucocytes stained with IgG1/2a control PE, CD45 FITC, and PI were acquired, and the percentage of control events were deducted to give a final CD34⁺ percentage. The absolute numbers of circulating progenitors were then calculated from the total nucleated cell count.

For CD34⁺ cell subset analysis, 50,000 events were acquired and a wide region was set around the progenitor cells (CD34⁺SSC^lo). A 200 to 500,000 event file was then collected using this gate. To allow simultaneous analysis of CD38 and CD59 expression by progenitor cells, only CD34 and light scatter characteristics were available for definition of progenitor cells. To analyze sufficient CD34⁺ cells, large numbers of total leucocytes had to be acquired, resulting in increased contamination of the CD34⁺ region. To overcome this problem, regions were used from the CD34 quantitation assay to improve definition. In each case, the CD34 versus SSC and FSC versus SSC regions were assessed to determine whether they alone could identify progenitor cells with greater than 95% accuracy, as defined by their combined CD34, CD45, PI, and light scatter characteristics. These regions were then applied to allow analysis of CD38 versus CD59 and controls.

NAP stock substrate was prepared by dissolving 30 mg of naphthol AS phosphate in 0.5 mL dimethyl formamide and adding 100 mL 0.2 mol/L tris buffer at pH 9.1. The NAP stain was prepared by adding 2 mg fast blue BB to 10 mL of this stock substrate. Blood films from patients and controls were fixed in buffered cold formol acetate for 30 seconds then rinsed. The slides are overlain with the NAP stain for 20 minutes, rinsed, and counterstained with aqueous neutral red for 1 minute. NAP positivity is indicated by the presence of blue/black granulation in the neutrophil cytoplasm, which is conventionally scored from 0 to 4. The NAP score for a sample is the sum of the scores of 100 neutrophils assessed by light microscopy. The normal range in our laboratory is 40 to 100.

Samples of blood suitable for analysis were obtained from all four patients in the steady state and after the administration of G-CSF. One patient (patient no. PNH046) developed a chest infection after treatment (presumably unrelated to the G-CSF) and this precipitated a hemolytic crisis requiring transfusion. The others had no ill effects apart from mild bone pain related to the growth factor. The peripheral white blood cell count (WBC) rose in all patients during G-CSF treatment from a baseline mean of 3.6 × 10⁹/L to a peak mean of 15.3 × 10⁹/L. The NAP score also increased in all patients (normal range 40 to 100). The mean NAP score on day 1 was 41 (range 0 to 123), whereas the mean on day 4 was 162 (range 60 to 250). Three out of the four patients showed a steady increase that only began to fall again after discontinuing G-CSF, whereas one patient's score rose initially but began to fall back towards baseline before G-CSF finished (see Fig 1).

The absolute number of peripheral blood CD34⁺ cells increased a median of 10-fold (range 4 to 13) from a baseline mean of 0.5 × 10⁶/L (range 0.1 to 0.8) to a peak mean of 4.2 × 10⁶/L (range 1.3 to 9.0). Neither the platelet count nor the reticulocyte count altered significantly. The phenotypic analysis of peripheral blood neutrophils and stem cells is summarized in Table 2. Control samples of peripheral blood from six healthy volunteers were processed in an identical way and confirmed observations that CD59 is expressed uniformly on all identifiable neutrophils and stem cells including the CD34⁺/CD38⁻ subset.12,14 The median absolute CD34 count in normal peripheral blood was 2.5 × 10⁶/L (range 0.9 to 11.0), which is in broad agreement with other published data.15 Three out of the four patients (patients no. PNH043, PNH042, and PNH056) showed considerable differences in the proportion of PNH cells in different cell compartments and a dramatic change after G-CSF. In steady-state peripheral blood, these three patients were found to have almost no detectable PNH cells in the most primitive (CD34⁺/CD38⁻) subset (median 0%; range 0% to 7%), whereas in the mature neutrophils, they had large PNH clones (median 96%; range 50% to 97%). The more mature CD34⁺/CD38⁺ stem cells appeared to have an intermediate number of PNH cells (median 47%; range 29% to 50%). After the administration of G-CSF, the pattern changed dramatically. There was no significant change in the mature neutrophils that remained largely PNH but both the early and intermediate stem cells became phenotypically more PNH (median 87%; range 68% to 90% for the CD34⁺/CD38⁺ cells and median 88%; range 51% to 91% for the CD34⁺/CD38⁻ cells). The changes in the CD38⁻ fraction were statistically significant (P<.05) but those in the CD38⁺fraction were not (P<.1). These changes reverted to steady-state values within a few days of discontinuing G-CSF. One patient (patient no. PNH046) had approximately 75% PNH cells in all subsets examined and this percentage did not change significantly after G-CSF administration. There was no obvious clinical distinction between this patient and the others to explain this observation.

It has previously been reported by Ninomiya et al16 that the GPI-linked molecule CD16 can be induced on PNH neutrophils by G-CSF administration and that this increased expression was sensitive to PI-PLC cleavage, implying that it was GPI-linked CD16. We also observed an increase in CD16 staining after G-CSF but the mean fluorescence intensities of other GPI-linked antibodies on neutrophils (CD55 and CD59) also increased as did the mean fluorescence intensity of non–GPI-linked CD15 and of negative control antibodies (CD3 and IgG1/2). Furthermore, we did not find this apparent increase in expression on PNH cells sensitive to PI-PLC cleavage, whereas the expression on the normal residual cells was abolished by PI-PLC as expected (see Fig 2). We conclude that the previously reported induction of CD16 on PNH cells after treatment with G-CSF is probably an artefactual observation.

The changes in hematological parameters and cell surface phenotype in four PNH patients before and after G-CSF are reported here. The WBC rose in all patients to a peak mean of 15.3 × 10⁹/L. This rise is less than that observed in normal volunteers on G-CSF whose absolute neutrophil count alone reaches mean peak levels of around 28 × 10⁹/L on similar regimens.15These findings presumably reflect the underlying hypoplasia in the PNH patients. There is wide variation in the absolute numbers of peripheral blood CD34 cells in normal individuals but all four PNH patients in this study had baseline values that were 10- to 20-fold less than those generally reported.15 The mean baseline absolute CD34 count in the peripheral blood was 0.5 × 10⁶/L for the PNH patients, compared with 8 × 10⁶/L in a series of normals. After G-CSF, the mean figures were 4.2 × 10⁶/L for the PNH patients and 55 × 10⁶/L for the normals. In both PNH and normal blood this represents a similar proportionate increase (7- to 9-fold).

The rise in NAP score in normal neutrophils on exposure to G-CSF is well documented and represents one of a range of stimulatory or priming effects of this cytokine.17,18 It was interesting to observe this phenomenon in our patient group because this enzyme is known to be GPI-linked and consequently the NAP score in such patients is classically low. Because two of our patients had nearly greater than 95% PNH neutrophils circulating before and after G-CSF, we can be certain that the increase in NAP score occurred in both PNH and normal neutrophils alike because the NAP scoring by light microscopy confirmed that the majority of cells showed increased granulation. Although we did not formally separate normal and PNH cells for scoring it would be a mathematical impossibility for the small normal clones in these patients to account for the increase. As far as we are aware, increase in the NAP score in PNH patients has not been previously reported.

We have been able to study the relative proportions of normal and PNH cells occurring at different stages of cell maturity in steady-state and G-CSF–mobilized peripheral blood in these patients. All four had substantial PNH clones as assessed by neutrophil phenotype, which implies that the marrow precursors are also largely PNH. Previous studies of PNH marrow have supported this view and shown a similar proportion of affected CD34⁺ cells to neutrophils.19 Despite this it has been previously noted that the great majority of the most primitive peripheral blood stem cells are in fact of normal phenotype in sharp contrast to the marrow and the more mature cell types.12,13 Our observations confirm this phenomenon in three out of our four patients. In addition we have shown a gradient of expression with the most primitive cells being normal, the more mature stem cells (CD34⁺/CD38⁺) having an intermediate proportion of PNH cells, and the mature neutrophils being largely PNH. It is not clear why normal pluripotent cells should circulate selectively in the peripheral blood of these patients when they are only a small minority of the marrow precursors. Normal stem cells may be selectively released, perhaps due to a GPI-linkage deficiency involving adherence or homing, or they may have a survival advantage over PNH cells when in the peripheral blood that is not present in the marrow miroenvironment. This phenomenon is marked in the majority of patients and it is intriguing to consider that it may have some relevance to the pathogenesis of PNH.

It has been suggested that normal stem cells could be collected from the blood and used as a source of normal progenitors for autologous transplantation in PNH,12,13 but the absolute numbers of peripheral CD34⁺ cells in PNH patients is generally very small. A logical approach would be to mobilize stem cells from such patients with growth factors, as is done in the management of other hematological malignancies. However, the number of stem cells mobilized by G-CSF is disappointing and the vast majority that are released are of PNH phenotype unlike those that naturally circulate. We found that after G-CSF the absolute number of CD34⁺ cells in the peripheral blood of our patients was a mean of 4.2 × 10⁶/L. (range 1.3 to 9.0) and in the important CD38⁻ fraction only a minority (median 12%) remained of normal phenotype. Contrasting this with recommendations for harvesting in hematological malignancy in which minimum post–G-CSF levels of around 10 to 20 × 10⁶ CD34⁺cells/L of peripheral blood are preferred, it suggests that it would be difficult to mobilize sufficient numbers of normal progenitors into the peripheral blood of PNH patients to support transplantation.

We would like to thank Cymbus Biosciences for provision of some of the monoclonal antibodies used in this work.