Increased Haemopoietic Cell Survival in Vitro Induced by a Human Marrow Fibroblast Factor

S ummary. Human marrow fibroblasts were grown in vitro and examined for effects on human and mouse haemopoietic cells. When human marrow cells were incubated with fibroblasts or with fibroblast-conditioned medium for 1 week and then assayed for committed granulocyte/monocyte (CFU-c) and erythroid (BFU-e) progenitor cells, the numbers of CFU-c and BFU-e were considerably increased compared with controls. In contrast, human marrow-fibroblast-conditioned medium, when added directly to CFU-c or BFU-e assay cultures, had no effect on colony formation by these progenitor cells. As these results suggest that the fibroblast-derived factor may be acting on a relatively primitive progenitor cell, possibly a pluripotent haemopoietic stem cell, the effect of this factor on mouse pluripotent haemopoietic stem cells (CFU-s) was examined. Human marrow-fibroblast-conditioned medium considerably enhanced CFU-s survival after a 24 h incubation and increased the proportion of CFU-s in cell-cycle. The increase in CFU-s survival depended on the concentration of the fibroblast-conditioned medium but not on the age of the fibroblast culture. The evidence suggests therefore that human marrow fibroblasts produce a factor that acts on a human myeloid progenitor cell more primitive than BFU-e and CFU-c, possibly the pluripotent haemopoietic stem cell.     

Alkaline Phosphatase Activity of Chronic Granulocytic Leukaemia Neutrophils in Agar Culture

We measured the concentration of granulocyte-committed progenitor cells (CFU-c) and the alkaline phosphatase activity of neutrophils harvested from colonies cultured from the peripheral blood of 30 patients with high-count chronic granulocytic leukaemia (CGL) in the chronic phase. Neutrophils in colonies cultured from marrow of normal donors or patients with acute myeloid leukaemia in remission served as controls. CFU-c numbers in CGL peripheral blood were on average 3 times higher than in normal marrow (130.1 + 126.2 (SD) versus 43.0 + 25.2 per 1 × 10⁵ cells plated, respectively). The mean NAP scores in cultured CGL neutrophils were substantially higher than control values (89.9 + 48.0 versus 55.9 + 33.0 units, respectively). There was a tendency for peak values of NAP in culture to be reached at earlier points in CGL cultures with high CFU-c numbers (i.e. high proliferative rates) than in those with lower CFU-c numbers (lower proliferative rates). Our data accord with the concept that low NAP levels in CGL in vivo result from modulating influences external to the neutrophil.     

Primary polycythaemia, essential thrombocythaemia and myelofibrosis--three facets of a single disease process?

Primary polycythaemia (PP), idiopathic myelofibrosis (MF), essential thrombocythaemia (ET) and chronic granulocytic or myeloid leukaemia (CGL) are clonal disorders of the pluripotent haemopoietic stem cells. We have studied granulocyte, megakaryocyte and erythroid progenitors from the peripheral blood of 7 patients with PP, 9 with ET, 19 with MF and 6 with CGL in order to characterise similarities and differences at the committed progenitor cell level. Spontaneous megakaryocytic and erythrocytic growth was characteristic of MF, PP and ET but was not seen in CGL. Circulating erythroid (BFU-E) and granulocyte/macrophage (CFU-GM) progenitors were markedly increased in MF and CGL, less raised in ET and closest to normal in PP. Erythropoietin-independent erythroid bursts (EIBFU-E) grew from the blood of patients with MF, PP and ET but spontaneous growth of megakaryocytes occurred in only MF and ET. These results suggest a progression of increasing abnormality from PP, where EIBFU-E occurred with relatively normal numbers of circulating progenitors, to ET where both EIBFU-E and megakaryocyte precursors regularly occur with elevated numbers of progenitors, to MF where spontaneous BFU-E, CFU-Mk and CFU-GM occur at high levels.     

Erythropoiesis in vitro. III. The role of potassium ions in erythroid colony formation.

The role of potassium as an essential promotor of erythroid progenitor growth (BFU-e & CFU-e) from normal murine hematopoietic tissues was studied. Dialyzed fetal calf serum, over a wide range of concentrations, was shown to reduce the numbers of BFU-e and CFU-e that could be cultured from normal murine bone marrow. A dose-dependent addition of 1 M KC1 restored erythroid progenitor colony growth to the levels generally seen when normal, non-dialyzed fetal calf serum was used. Furthermore, when [K] was increased in some human urinary and sheep plasma erythropoietin preparations, the number of erythroid progenitor cells cultured also increased. This influence is crucial to the differentiation of committed stem cells into the erythroid pathway and must therefore be considered in the development of serum-free growth media.     

Erythroid progenitor cell kinetics in chronic haemodialysis patients responding to treatment with recombinant human erythropoietin

The response of bone marrow and peripheral blood erythroid progenitors to human recombinant erythropoietin (rHuEPO) was studied in nine haemodialysed renal failure patients receiving this hormone for the correction of their anaemia. The haematocrit rose in all patients in response to thrice weekly injections of escalating rHuEPO doses (12-192 IU/kg). Both the numbers of CUF-e and BFU-e and their proliferative state in the bone marrow as well as BFU-e numbers in the peripheral blood were estimated before treatment and again after correction of the anaemia, at 16 h following an intravenous dose of rHuEPO. Following treatment bone marrow BFU-e numbers fell to a mean of 24.5% (P less than 0.01) of the pre-treatment values although there was no significant change in CFU-e or circulating BFU-e numbers. The mitotic rate (percentage S-phase cells) estimated by tritiated thymidine suicide rose from 45.2% to 68.4% (P less than 0.05) in the case of CFU-e and from 16.4% to 45.1% (P less than 0.05) for BFU-e following treatment with rHuEPO thus indicating in-vivo sensitivity of both the primitive as well as the mature erythroid progenitors to the hormone. The fall in BFU-e numbers in the bone marrow after several months of treatment may be due to a loss of cells from this progenitor pool by maturation that is uncompensated by replacement from the pluripotential stem cell compartment.     

Granulocyte Function in Chronic Granulocytic Leukaemia

The ingestion, bactericidal activity and metabolism of isolated mature neutrophil leucocytes during phagocytosis was studied in 17 patients with chronic granulocytic leukaemia (CGL) with the simultaneous use of normal controls. Seven patients had received no treatment and the others had been treated previously with Busulphan. The phagocytic indices for killed yeast cells did not differ from those of the controls. A diminished bactericidal activity against E. coli was found in nine CGL cases. The bactericidal capacity closely correlated with the degree of leucocytosis since patients with a WBC count of 90 000/mul or higher with one exception showed decreased bactericidal activities while patients with WBC counts below 90 000/mul with two exceptions showed normal bactericidal activities. The [I-14C]-glucose oxidation during phagocytosis was increased in four patients and decreased in three patients. Some correlation was found between abnormally high or low [I-14C]glucose oxidation and diminished bactericidal activity. The intracellular iodination reaction during phagocytosis was decreased in 10 cases while the extracellular iodination was increased in six cases and decreased in one case. The data for granulocyte iodination did not correlate with WBC count, bactericidal capacity or [I-14C]glucose oxidation. The time course for the bactericidal activity and granulocyte iodination seemed to deviate from the controls indicating a slow initial ingestion and/or degranulation phase. The CGL granulocyte content of myeloperoxidase was normal or increased, the lysozyme content was decreased in half of the cases while the amount of antibacterial cationic proteins was increased, normal or low. The present findings indicate a variety of abnormalities in the mature CGL granulocyte, which are not closely interrelated.     

T cells and erythroid burst forming units in chronic lymphocytic leukemia

Substantial evidence exists indicating T cell abnormalities in chronic lymphocytic leukemia (CLL). There is also evidence that the T cell is an important source of burst promoting activity (BPA) for the peripheral blood (PB) erythroid burst forming unit (BFU-e). We studied the BPA of T cells and response of BFU-e in normals and untreated early stage B cell CLL patients in a methylcellulose colony assay. Normal null cell cultures grew significantly more BFU-e than CLL null cell cultures. Addition of autologous T cells to normal or CLL null cells significantly increased BFU-e only in normals. Allogeneic coculture of T cells from CLL patients with null cells from normals yielded normal responses of BFU-e in five of six cases. In contrast, allogeneic coculture of normal T cells with CLL null cells yielded a normal response in only one of six studies. Furthermore, adding increasing quantities of autologous or normal allogeneic T cells to CLL null cells did not augment the BFU-e response. Accounting for the expanded lymphocyte pool in CLL, BFU-e are decreased in concentration but the absolute number is normal or increased. The decrease in concentration could be secondary to expansion of the null cell fraction in CLL by pre-B cells. CLL T cells appeared to augment normal allogeneic PB BFU-e in a normal fashion, whereas, in several cases, CLL BFU-e were hyporesponsive to autologous or normal allogeneic T cells. It is therefore apparent that in untreated early stage B cell CLL, erythroid progenitor cells are present in the peripheral blood but are diluted in an expanded null cell compartment and may, in some cases, be hyporesponsive to T cell BPA. T cell BPA of CLL T cells in this early stage of disease is preserved.     

Continuous human bone marrow culture: Ia antigen characterization of probable pluripotential stem cells

The presence of Ia-like antigens on human CFU-C and BFU-e is confirmed and a cell type that lacked immediate capacity for granulocytic colony formation but generated CFU-c after brief incubation in simple suspension culture is identified. This pre-CFU-c, and its immediate progeny, was extremely sensitive to killing by anti-Ia serum with complement. In contrast, anti-Ia serum plus complement treatment of human bone marrow, while eliminating 93%-97% of all CFU-c and BFU-e, did not prevent the rapid regeneration of these progenitor cells and their production for some weeks under the conditions of continuous marrow culture. These studies suggest that the human equivalent of the pluripotential stem cell can replicate for some weeks in culture and generate committed progenitors, such as CFU-c and BFU-e. Furthermore, it would appear that Ia-like antigen is absent on the pluripotential stem cell, is rapidly gained as commitment to the various progenitor cell types occur, and is subsequently lost as these latter undergo differentiation within the marrow.     

Hematopoietic progenitors in cyclic neutropenia: effect of granulocyte colony-stimulating factor in vivo.

The number and growth factor requirements of committed progenitor cells (colony-forming units-granulocyte/macrophage and burst-forming units-erythroid) in three patients with cyclic neutropenia (two congenital, one acquired) were studied before and during therapy with recombinant human granulocyte colony-stimulating factor (G-CSF; 3 to 10 micrograms/kg/d). When the patients with congenital disease were treated with G-CSF, the cycling of blood cells persisted, but the cycle length was shortened from 21 days to 14 days, and the amplitude of variations in blood counts increased. There was a parallel shortening of the cycle and increase of the amplitude of variations (from two- to three-fold to 10- to 100-fold) in the number of both types of circulating progenitor cells in these two patients. In the patient with acquired cyclic neutropenia, cycling of both blood cells and progenitors could not be seen. In cultures deprived of fetal bovine serum, erythroid and myeloid bone marrow progenitor cells from untreated patients and from normals differed in growth factor responsiveness. As examples, maximal growth of granulocyte/macrophage (GM) colonies was induced by granulocyte/macrophage (GM)-CSF plus G-CSF in the patients, whereas a combination of GM-CSF, G-CSF and interleukin-3 (IL-3) was required in the normals, and erythropoietin alone induced fourfold more erythroid bursts from cyclic neutropenic patients than from normal donors (46% versus 11% of the maximal colony number, respectively). The growth factor responsiveness of marrow progenitor cells slightly changed during the treatment toward the values observed with normal progenitors. These results indicate that treatment with G-CSF not only ameliorated the neutropenia, but also increased the amplitude and the frequency of oscillation of circulating progenitor cell numbers. These data are consistent with the hypothesis that G-CSF therapy affects the proliferation of the hematopoietic stem cell.     

In vitro erythropoiesis. II. Cytochemical enumeration of erythroid stem cells (CFU-e and BFU-e) from normal mouse and human hematopoietic tissues.

A procedure which provides more accuracy in the identification and scoring of in vitro derived erythroid colonies (i.e. BFU-e and CFU-e) from normal mouse and human marrow and spleen cells has been developed. Up to now identification of erythroid colonies has been based on staining with acidified benzidine which consistently displays "background staining" of granulocyte/macrophage colonies. The staining reaction with these reagents is such that identification of erythroid colonies is less than precise. The modified benzidine procedure described herein eliminates the staining reactivity of granulocyte/macrophage colonies and specifically reacts with only those cells that contain hemoglobin. Furthermore, individual colonies and/or cells may readily be counterstained and examined by high resolution light microscopy.     

Antigenic expression and proliferative status of multilineage myeloid progenitor cells (CFU-GEMM) in normal individuals and patients with chronic granulocytic leukaemia

We assayed the number of multilineage myeloid progenitor cells (CFU-GEMM) in the blood and marrow of patients with newly diagnosed chronic granulocytic leukaemia (CGL). The mean number of CFU-GEMM in the blood was increased 600-fold and CFU-GEMM in the marrow was doubled in the CGL patients compared with normal. A complement-fixing monoclonal antibody with HLA-DR specificity inhibited the proliferation of CFU-GEMM from CGL blood to a greater extent than that of comparable cells in normal marrow. Using a hydroxyurea 'suicide' method we found that the proportion of CFU-GEMM in proliferative cycle was higher in CGL blood than in normal marrow. We conclude that (1) CFU-GEMM numbers are greatly increased in the blood of patients with CGL, (2) CFU-GEMM express HLA-DR antigens on their surface, and (3) the apparently increased expression of the antigen on CFU-GEMM from CGL blood in comparison with CFU-GEMM from normal marrow may parallel the relatively higher proportion of CGL CFU-GEMM in cell cycle.     

Transplantation potential of hematopoietic cells released into the circulation during routine chemotherapy for non-Hodgkin's lymphoma [see comments]

Primitive hematopoietic cells released into the peripheral blood (PB) were studied in 50 patients with high-grade non-Hodgkin's lymphoma enrolled in a phase III trial of intensive weekly chemotherapy (VAPEC- B) alone or with granulocyte colony-stimulating factor (G-CSF). Mononuclear cells numbers were monitored and their in vitro growth potential assessed in clonogenic progenitor cell assays and in long- term culture. Total colony-forming cells (granulocyte-macrophage [GM], burst-forming unit, erythroid [BFU-E], Mix-CFC) were increased 40-fold (median) over baseline with chemotherapy alone and 106-fold with chemotherapy and G-CSF after the final dose. CD34+ cells were increased to a median of 4%, equivalent to that in normal bone marrow (BM) controls. Circulating colony-forming cell levels were maximal when the recovering total white blood cell (WBC) count reached 5 to 10 x 10(9)/L. The timing of the maximum was reproducible in individual patients. Therefore the WBC count can be used as a guide to the timing of leukapheresis. PB cells from normal controls' and patients' prechemotherapy were unable to sustain hemopoiesis in two-stage long- term cultures. In contrast, PB cells collected from patients primed with chemotherapy alone or chemotherapy with G-CSF at the time of predicted maximal colony-forming cell release were able to generate and sustain hematopoiesis in long-term cultures at a level comparable or superior to normal BM. These findings indicate that the use of G-CSF after routine outpatient chemotherapy stimulates maximal release of primitive hemopoietic cells into the circulation, including colony- forming cells and long-term culture-initiating cells. Their numbers are comparable with those in normal BM and are such that a single leukapheresis will usually yield enough cells for hemopoietic reconstitution after myeloablative chemotherapy.     

Changes of the phenotype of erythroid progenitor cells (BFU-E, CFU-E) and their progenies during early steps of differentiation

Early differentiation processes of human erythroid progenitor cells (BFU-e, CFU-e) have been studied during in vitro proliferation using a panel of monoclonal antibodies with known reactivity on different levels of the erythroid cell line. Two antibodies recognizing structures on BFU-e (VIP-2b, BMA 021), two antibodies reactive with CFU-e and nucleated red cells (5F1, CLB-Ery-3) and one antibody directed against glycophorin A (VIE-G4) were used for this study. Normal human bone marrow cells were induced to proliferation in an erythroid progenitor cell assay and, after different periods of incubation, agar cultures were treated with these antibodies and complement. Thereafter, the remaining erythroid cells were incubated again to continue their proliferation with the same stimulators as before. The changes of the phenotype of BFU-e and CFU-e progenies during in vitro proliferation were determined by the reduction of colony formation in comparison with untreated control cultures. Our results indicate that the loss of HLA-DR antigens and the p45 structure is accompanied by the acquisition of structures recognized by the antibodies 5F1 and CLB-Ery-3. After 5-7 d of incubation BFU-e derived progenies exhibit the same antigenic structure as has been found for CFU-e. Glycophorin A expression could only be demonstrated at a late differentiation stage of the erythroid cell lineage.     

HLA-DR monoclonal antibodies inhibit the proliferation of normal and chronic granulocytic leukaemia myeloid progenitor cells

S ummary . We studied the capacity of murine monoclonal antibodies with HLA-DR specificity to inhibit the proliferation in vitro of erythroid (BFU-E and CFU-E) and granulocyte-macrophage (CFU-GM) progenitor cells in normal bone marrow and the blood of patients with chronic granulocytic leukaemia (CGL). Two IgG₂ antibodies (CA 2.06 and L243) inhibited the proliferation of normal BFU-E and CFU-GM at relatively high dilution; a third antibody, DA2, had no effect on either progenitor cell. A complement-king monoclonal antibody with T-cell activity (OKT3) produced only minor reduction in progenitor cell proliferation. Further studies with L243 showed that BFU-E and CFU-GM from the blood of patients with CGL were inhibited to the same degree as normal marrow progenitor cells. The inhibition of progenitor
cell proliferation by a given antibody was always complement dependent and was therefore presumed to be due to a direct cytotoxic effect. The inhibitory effect of monoclonal antibodies is a valuable approach to the characterization of antigenic determinants on myeloid progenitor cells and the differential cytotoxicity of selected monoclonal antibodies might be exploitable for therapy.     

GM-CFC growth in chronic granulocytic leukaemia is not affected by a soluble inhibitor released by aplastic anaemia T-cells or mitogen-primed normal T-lymphocytes

Peripheral blood (PB) and bone marrow (BM) cells were obtained from 12 patients with chronic granulocytic leukaemia (CGL) and cultured in agar for granulocyte macrophage colony formation (GM-CFC). In addition PB and BM CGL cells were co-cultured with T-lymphocytes from patients with immune severe aplastic anaemia (SAA), or with T-cells from healthy donors primed with pokeweed mitogen (PWM). The supernatants of SAA and PWM primed T-cells were also added to CGL cells grown in agar. Normal marrow cells obtained from eight healthy donors were used to set up control cultures and co-cultures. The results of this study indicate that, firstly, T-cells derived from SAA patients and their supernatants suppress GM-CFC growth of normal marrow cells but not of PB nor BM CGL cells, and, secondly, normal T-cells primed with PWM and their supernatants suppress normal marrow GM-CFC but not colony formation of BM nor PB CGL cells. These results provide further evidence that mediators which are effective in regulating normal myeloid progenitor cells fail to inhibit the in vitro growth of GM-CFC from CGL patients.     

Mobilization of primitive haemopoietic progenitor cells and stem cells with long-term repopulating ability into peripheral blood in mice by pegylated recombinant human megakaryocyte growth and development factor

We investigated in detail the effect of pegylated recombinant human megakaryocyte growth and development factor (PEG-rHuMGDF) on peripheral blood progenitor cell (PBPC) mobilization in male BDF₁ mice. Treatment with PEG-rHuMGDF for 5 d stimulated a striking expansion of the circulating levels of multiple types of colony-forming units in culture (CFU-c), including CFU-granulocyte-macrophage, CFU-megakaryocyte, burst-forming units-erythroid, and multipotent CFU-c, and primitive day-12 CFU-spleen. All of these progenitors were mobilized into the peripheral blood (PB) with similar kinetics; their numbers peaked after the cessation of treatment and then declined earlier than platelet numbers peaked. The maximal increase in any of the four CFU-c in the PB was attained with at least 300 μg/kg/d of PEG-rHuMGDF, whereas peripheral platelet counts plateaued at 30 μg/kg/d. Adoptive transfer with PB from PEG-rHuMGDF-treated donor mice resulted in greater survival of lethally irradiated recipients. The majority of the recipients that survived at 187 d after transplantation with PEG-rHuMGDF-mobilized PB showed significant donor engraftment at the progenitor cell level. The combined administration of appropriate doses of PEG-rHuMGDF and recombinant human granulocyte colony-stimulating factor induced a synergistic increase in the circulating levels of the four CFU-c compared to either factor alone. These results indicate that PEG-rHuMGDF as a single agent can mobilize a full spectrum of PBPCs in mice.     

Effects of an aplastic anemia urinary extract on mouse erythroid progenitor cells in vivo

We investigated the in vivo effects of a crude extract from the urine of aplastic anemia patients (AA urinary extract) on erythroid precursor cells in the femoral bone marrow and spleens of normal adult mice. A single intraperitoneal injection of AA urinary extract induced a significant increase in the number of splenic erythroid burst-forming units (BFU-e) and erythroid colony-forming units (CFU-e) within 24 h after injection. We then injected pure recombinant erythropoietin (Epo) equivalent to the amount present in the urinary extract. This addition increased the number of splenic CFU-e by almost the same degree as the amount induced by the AA urinary extract 24 h after injection, but failed to elicit any change in the number of splenic BFU-e. In other studies, mice were injected with the same amount of lipopolysaccharide (LPS) and/or pure Epo as that present in the AA urinary extract. Experiments with Limulus amebocyte lysate-adsorbed (endotoxin-depleted) or nonadsorbed (endotoxin-containing) AA urinary extracts showed that endotoxin contamination interfered with the increase in numbers of marrow CFU-e and enhanced the increase in splenic CFU-e numbers induced by pure Epo or Epo activity in the AA urinary extract. The number of splenic BFU-e, however, was not affected by administration of LPS and/or Epo or by adsorbed endotoxin. These data suggest that AA urinary extract contains a stimulating activity for mouse splenic BFU-e, and that this activity is not attributable to the Epo activity or endotoxin contamination within the urinary extract.     

Hydroxyurea promotes the reduction of spontaneous BFU-e to normal levels in SS and S/beta thalassemic patients

We have studied the effects of hydroxyurea on growth and differentiation of early erythroid progenitor cells (BFU-e) from peripheral blood of sickle cell disease patients (five SS and two Hb S/beta-thalassemia) in the presence or absence of exogenous stimulating factors. When the mononuclear cells from the sickle cell disease patients were cultured at diagnosis (before hydroxyurea treatment), there was an increased number of BFU-e in relation to controls (p < 0.05, Wilcoxon test) when cells were grown in the presence or absence of 5637 conditioned medium and erythropoietin. Colonies that developed in the absence of added growth factors were considered "spontaneous". A significant difference was observed after hydroxyurea treatment in the number of BFU-e obtained in the presence and absence of stimulus, with a higher reduction in the spontaneous BFU-e number. As expected, there was an increased Hb F level in these patients when compared with their pretreatment levels. There was no correlation between spontaneous BFU-e and hemoglobin levels in all patients studied.     

Drug-indiced agranilocytosis: Evidence for the commitment of bone marrow haematopoiesis

Clinical and haematological features of 61 patients with drug-induced agranulocytosis (63 episodes) are presented. Multiple drug consumption was a common observation and complicated the attempt to incriminate a particular drug as being aetiologically involved. Bone marrow analysis shortly after the diagnosis revealed evidence for an impairment of proliferative granulopoiesis in the majority of cases. This observation was confirmed by in vitro culturing of granuloid precursor cells (CFU-c). Moreover, the data clearly demonstrated that drug-induced agranulocytosis may not be restricted to the granulocytic series. Thrombocytosis and reticulocytosis during the recovery phase are taken as an indication for the commitment of all haemopoietic cell lineages in agranulocytosis. These observations were in accordance with cytomorphological studies and in vitro culture data of erythroid precursor cells (CFU-e, BFU-e) of bone marrow aspirates taken in the initial phase of agranulocytosis. More than 25% of the patients showed a marked erythroid depression in the marrow.     

Drug-induced agranulocytosis: evidence for the commitment of bone marrow haematopoiesis

Clinical and haematological features of 61 patients with drug-induced agranulocytosis (63 episodes) are presented. Multiple drug consumption was a common observation and complicated the attempt to incriminate a particular drug as being aetiologically involved. Bone marrow analysis shortly after the diagnosis revealed evidence for an impairment of proliferative granulopoiesis in the majority of cases. This observation was confirmed by in vitro culturing of granuloid precursor cells (CFU-c). Moreover, the data clearly demonstrated that drug-induced agranulocytosis may not be restricted to the granulocytic series. Thrombocytosis and reticulocytosis during the recovery phase are taken as an indication for the commitment of all haemopoietic cell lineages in agranulocytosis. These observations were in accordance with cytomorphological studies and in vitro culture data of erythroid precursor cells (CFU-e, BFU-e) of bone marrow aspirates taken in the initial phase of agranulocytosis. More than 25% of the patients showed a marked erythroid depression in the marrow.