Application of flow cytometry to the diagnosis of paroxysmal nocturnal hemoglobinuria


  • Stephen J. Richards,

    Corresponding author
    1. Hematological Malignancy Diagnostic Service, Department of Hematology, Leeds General Infirmary, Leeds, United Kingdom
    • Hematological Malignancy Diagnostic Service, The Algernon Firth Building, Leeds General Infirmary, Leeds, LS1 3EX, United Kingdom
    Search for more papers by this author
  • Andrew C. Rawstron,

    1. Hematological Malignancy Diagnostic Service, Department of Hematology, Leeds General Infirmary, Leeds, United Kingdom
    Search for more papers by this author
  • Peter Hillmen

    1. Hematological Malignancy Diagnostic Service, Department of Hematology, Leeds General Infirmary, Leeds, United Kingdom
    Search for more papers by this author


Within the contemporary multitude of complex methods used in clinical flow cytometry, very few techniques exist which can be described as disease-specific diagnostic tests. Detection of glycophosphatidylinositol (GPI)-linked antigens on hematopoietic cells using monoclonal antibodies and flow cytometry forms the basis of a specific diagnostic test for paroxysmal nocturnal hemoglobinuria (PNH). Absent or markedly diminished expression of GPI-linked antigens is, in the appropriate clinical setting, specific for all patients with PNH. Clinically, PNH is a syndrome characterized by bone marrow failure, acquired hemolytic anemia, and a thrombotic tendency. The molecular genetic lesion responsible for this condition is a somatic mutation of the X-linked pig-a gene within a multipotent hematopoietic stem cell. Due to its rarity, delay in diagnosis is not uncommon for patients with PNH. Once a definitive diagnosis is established, this can make a considerable impact on patient management and prognosis. In this article, we review the complimentary roles that molecular biology and flow cytometry have played in unraveling the genotypic and phenotypic aspects of this unique condition. Cytometry (Comm. Clin. Cytometry) 42:223–233, 2000. © 2000 Wiley-Liss, Inc.


Paroxysmal nocturnal hemoglobinuria (PNH) is an acquired hematopoietic stem cell disorder characterized clinically by bone marrow failure, thrombosis, and chronic hemolytic anemia with acute episodes. PNH is a rare condition. Its true incidence is unknown, although it is likely to be similar to that of aplastic anemia (2–6 per million people) with which it shares a close relationship (1, 2). The wide spectrum of clinical presentation and variable disease course provides significant challenges for both clinician and scientist in terms of managing patients, establishing a diagnosis, and explaining the pathobiological features of PNH. In the last 15 years, great advances have been made in understanding the molecular and cellular biology of PNH, culminating in defining the molecular lesion responsible for the PNH abnormality (3–5). Before reviewing the role of immunophenotyping and flow cytometry in the study and diagnosis of PNH, it is worth summarizing the historical landmarks of this unique disease.


The first report of PNH was made over a century ago by Strubing (6), who described hemolytic anemia with accompanying hemoglobinuria occurring during the night. Fifty years later, Ham and Dingle (7) demonstrated that PNH red cells had an increased susceptibility to lysis in acidified serum. This observation led to the development of the Ham test for the diagnosis of PNH, which even today remains widely used as a diagnostic test. It was soon discovered that hemolysis in PNH was due to an increased sensitivity of red cells to activated complement. Furthermore, it was found that neutrophils and platelets were also abnormal in PNH (8, 9). In view of these findings, Dacie, as early as 1963 (10), hypothesized that PNH was an acquired clonal disorder resulting from a somatic mutation within a stem cell. This assertion was subsequently confirmed by isoenzyme studies of two females with PNH who were heterozygous for the enzyme glucose 6 phosphate dehydrogenase. The PNH red cells contained only one isoform of the enzyme, whereas residual normal red cells in these patients contained both isoenzymes (11).

Biochemical Defect

In 1983, biochemical studies of PNH red cells demonstrated that they were deficient in a complement regulatory protein known as decay acceleration factor (DAF; 12). This protein inhibits the formation of complement C3 convertases and is attached to the cell membrane by a glycophosphatidylinositol (GPI) anchor as a posttranslational processing step. Further studies of PNH cells showed they were deficient in all proteins normally linked to the cell membrane by the GPI anchor (13). This strongly implicated a failure in synthesis of the GPI anchor as the cause of PNH and initiated intense activity to identify the underlying biochemical and molecular defect. The biochemical pathway for the synthesis of GPI was examined in PNH cell lines and found to be abnormal (14, 15). The specific abnormality in all cases studied was at the first stage of the pathway involving transfer of N-acetyl glucosamine to phosphatidylinositol (16).

Molecular Lesion

The underlying genetic defect was identified in 1993 by Miyata et al. (17). Through an elegant series of complementation and transfection studies, they demonstrated that transfection of the phosphatidylinositol glycan complementation class A (pig-a) gene into PNH cell lines restored expression of GPI-linked antigens. Sequence analysis of the pig-a gene, initially by Takeda and subsequently by other groups, has revealed mutations within the pig-a gene in all patients thus far reported with PNH (18–21). Because these mutations are somatic, they are all different and occur throughout the entire pig-a coding region. They comprise deletions, insertions, or point mutations. Most of the deletions and insertions are small (one or two base pairs), resulting in frameshift mutations and a nonfunctional product. Frameshift mutations result in the total absence of GPI-anchored proteins from cells (i.e., type III cells). A minor proportion of mutations are missense, point mutations that, in some instances, have residual activity that allows synthesis of small amounts of GPI anchor. This type of mutation produces cells with partial expression of GPI-anchored proteins (i.e., type II cells). As the pig-a gene is located on the X chromosome, there is only one functional pig-a gene per cell (including females due to X-inactivation). Thus, a single mutation will result in a GPI-deficient phenotype. This is in contrast to the remaining pig genes, which are autosomal and therefore have two active copies per cell. Mutations of both alleles of these autosomal genes would be required to disrupt production of the GPI anchor and produce the PNH phenotype.

Flow Cytometry

Although molecular biology techniques have successfully uncovered the genotypic lesion in PNH, the use of monoclonal antibodies and flow cytometry has made an equally significant and complimentary contribution to defining phenotypic abnormalities in PNH. In 1985, two independent groups used flow cytometry and an antibody to DAF (CD55) to investigate patients with PNH (22, 23). Their studies found DAF-deficient populations of red cells, granulocytes, monocytes, lymphocytes, and platelets. This confirmed the multilineage nature of PNH and provided further support for the concept that PNH was a clonal stem disorder. Furthermore, analysis of hematopoietic progenitor cells isolated from bone marrow aspirates of patients with PNH also failed to express DAF (24). As new monoclonal antibodies to GPI-anchored antigens became available through successive leukocyte typing workshops, these were subsequently applied to the study of PNH cells. Van der Schoot et al. (25) showed that CD16, CD24, and CD67 (later reclassified as CD66b) were absent from PNH neutrophils. They also showed for the first time the superiority of flow cytometry over the Ham test for the diagnosis of PNH. As part of their cohort, they studied 16 patients with aplastic anemia and found small granulocyte PNH clones in three cases, even though Ham tests were negative. This point was further emphasized when it was discovered that some patients with severe aplastic anemia had only GPI-deficient granulocytes and monocytes with apparently normal red cells (26). Flow cytometry rapidly became established as a reliable diagnostic procedure for PNH and for measuring the extent of the PNH clone within the various hematopoietic cell lineages. As a consequence, flow cytometry has now replaced the Ham test as the “gold standard” technique for the diagnosis of PNH, even though there is no widespread consensus on flow cytometry methodology and antibody selection.

The realization that small clones of PNH cells could be detected by flow cytometry in aplastic anemia patients who had negative Ham tests led to classifying PNH into two broad groups: (1) Hemolytic PNH characterized by overt episodes of hemolytic anemia with hemoglobinuria. Importantly, venous thrombosis occurs in up to 50% of this group and is the major cause of death in one third (27, 28). (2) Hypoplastic PNH in which clones of GPI-deficient hematopoietic cells are detectable with no overt hemolysis but may have other symptoms such as aplastic anemia, leukopenia, or thrombocytopenia. Up to 10% of these patients die due to aplastic anemia.

Many patients do not fall into a single category and can often show features of both groups. A proportion of patients with hypoplastic PNH subsequently progress to classical hemolytic PNH. An additional minor, but clinically significant group of patients present primarily with thrombosis and no clinical evidence of hemolysis. In our current series of 117 patients, 7 presented with this clinical picture. Furthermore, three of these patients had a previous undescribed congenital form of GPI deficiency not due to a pig-a mutation (29). Other groups have also reported a high incidence of PNH among patients with Budd-Chiari syndrome (hepatic venous thrombosis; 30). Clearly, patients presenting with unexplained thrombosis in classical sites, such as hepatic or cerebral veins, should be screened for PNH.


Tests Dependent on Complement Sensitivity of Red Cells

Prior to the availability of flow cytometry, the majority of patients screened for PNH were those with a predominantly hemolytic picture. The Ham test, sucrose lysis test, and modified Ham tests rely on the differential sensitivity of PNH red cells to hemolysis (31, 32). These tests, although suitable for hemolytic PNH, cannot reliably detect small populations of PNH red cells nor differentiate partially and completely deficient cells (33, 34). In addition, due to the effect of transfusion and the shortened life span of PNH red cells, these tests do not give an accurate assessment of the true size of the PNH clone. It is clear from several studies that the proportion of GPI-deficient neutrophils, as measured by flow cytometry, gives a more accurate representation of the size of the PNH clone (34–36). The complement lysis sensitivity (CLS) test, which was developed by Rosse and colleagues in the 1960s (33, 37), can provide a more accurate assessment of the proportion and type of PNH red cells. Unfortunately, the CLS test is extremely laborious and difficult to perform and is not applicable to routine diagnostic use. A more recently developed screening test for PNH, designed primarily to overcome the technical complexities of the above methods, is the DiaMed monoclonal antibody gel technique (38). This commercially available test kit can be readily performed by routine hematology laboratories, although its sensitivity and specificity in comparison with flow cytometry has not yet been established on a large number of PNH cases.

Flow Cytometry: General Considerations

For flow cytometric analysis of GPI-anchored proteins on hematopoietic cells, there are a number of important considerations in terms of gating strategies and antibody selection. In addition, interpretation of results in PNH is dependent on a detailed knowledge of the cellular distribution of GPI-linked antigens and their expression at the different stages of hematopoietic cell differentiation (Table 1). Flow cytometry for PNH is unusual as all the diagnostic antigens are missing from PNH cells. Therefore, it is essential to include at least two GPI antigens on any cell type to exclude congenital deficiency of single antigens and to exclude technical problems. It is also imperative to include a transmembrane antigen as a positive control.

Table 1. Expression of GPI-Anchored Proteins by Hematopoietic Cells*
AntigenFunctionCellular distribution
  • *

    Summarized from leukocyte typing VI.

CD14Endotoxin receptor (lipopolysaccharide receptor)Strong expression on monocytes; weakly expressed on granulocytes
CD16Low-affinity Fc receptor for IgG immune complexes FcγRIIINeutrophils (GPI linked with polymorphisms)
NK cells and T-cell subset (transmembrane forms)
CD24UnknownB cells and granulocytes
CD48Possible receptor/ligand binding with CD2Lymphocytes and monocytes
CD52 (CAMPATH)UnknownLymphocytes and monocytes
CD55 (DAF)Complement regulation; limits the formation of C3′ convertasesAll hematopoietic cells
Widely expressed on nonhematopoietic cells
CD58 (LFA-3)Adhesion; costimulatory signal in immune responsesAll hematopoietic cells; occurs as transmembrane and GPI-anchored forms
CD59 (membrane inhibitor of reactive lysis [MIRL])Inhibits formation of membrane attack complex and protects cells from complement-mediated lysisAll hematopoietic cells; widely expressed on nonhematopoietic cells
Cell signaling in T-cell activation
CD66b (formerly CD67)Homophilic/heterophilic adhesion; cellular activationGranulocytes; epithelial cells
CD66e (carcinoembryonic antigen)
CD73 (ecto-5-nucleotidase)ImmunoregulatorySubsets of B and T cells
CD87 (uPAR)Converts plasminogen to plasminT cells, NK cells, monocytes, neutrophils, and nonhematopoietic cells; constitutive expression is weak and increases on activation
CD90 (Thy-1)UnknownStem cell subset; small T-cell subset
CD108 (JMH blood group antigen)Possible role in adhesionErythrocytes; low levels on lymphocytes
CD109UnknownActivated T cells, platelets, megakaryocytes, and subsets of CD34+ stem cell progenitors
CD157Ecto-enzyme; ADP ribose hydrase cyclaseBone marrow stromal cells, monocytes, and granulocytes.

Analysis of Red Cells.

A number of early publications focused on examining red cell expression of GPI-linked antigens (22, 23). This approach is probably historical in nature for two reasons. First, the main clinical manifestation of the disease is hemolysis. Second, previous PNH screening tests, i.e., the Ham and sucrose lysis tests, are red cell based. The disease itself is still considered and classified by many as acquired hemolytic anemia, even though the stem cell nature of the disorder is now well established. Although there is no consensus gating strategy for red cells in the literature, we and others use forward scatter (FSC) and sideways scatter (SSC) amplification in log mode and establish an acquisition/analysis gate based on physical characteristics of red cells. The purity of this physical gate is checked by determining the percentage positivity of a non–GPI-linked antigen such as the glycophorin molecules, known to be expressed on all erythrocytes. This approach is not ideal when compared to the elegant Boolean gating strategies used for identifying lymphocytes, leukemic cells, or stem cells (39–41). The significant technical problem of red cell agglutination is a major consideration when attempting multicolor staining of red cells. The presence of low levels of protein support in cell washing solutions and monoclonal reagents can enhance agglutination of red cells if coated by antibody. Therefore, we and the majority of authors undertake only single-antibody staining of red cells.

For routine screening of red cells for PNH, it is recommended that two GPI-linked antigens are examined (CD55 and CD59) using directly conjugated monoclonal antibodies. If negative or partially deficient cells are detected, then this must be for both GPI-linked antigens in order to establish a diagnosis of PNH. This is because rare hereditary deficiencies of the individual complement regulatory molecules CD55 and CD59 can occur. In contrast to the majority of PNH cases, all cells in these patients are either CD55 or CD59 deficient (42, 43).

Analysis of red cells in an untransfused PNH patient provides the clearest definition of type III (complete deficiency), type II (partial deficiency), and type I (normal expression) populations (Fig. 1). A marked variation in the distributions of these subpopulations is seen from patient to patient and the separation between the types of cells is not always clear cut. The relative frequencies with which these types occur in PNH from our current series of over 100 patients are summarized in Table 2.

Figure 1.

Expression of the GPI-linked antigen CD59 on normal red cells and a series of patients with PNH. a: Histogram is a representative profile of CD59 expression by normal red cells (type I) and an isotype-matched negative control (clear histogram) that defines type III cells. Histograms b–k are representative profiles from patients with PNH. b: Histogram shows a bimodal distribution with type I (normal) and III (complete deficiency) populations. c: Histogram shows a bimodal distribution of type II (partially deficient) and I (normal) cells. d: Case shows distinct type I, II, and III populations. e–k: Histograms highlight the diversity of CD59 expression frequently seen in PNH. The distinction between type II and III cells is not always clearly defined (j,k). When residual normal cells are few in number (plot e), they may merge with type II cells. l: Histogram shows CD59 staining from a congenital form of GPI deficiency not due to a pig-a mutation. In contrast to classical PNH, the majority of cells show a spectrum of deficiency from partial to complete with no clear delineation of type I, II, and III components.

Table 2. Frequency and Distribution of Red Cell Populations in 97 Patients with PNH
CategorybNo. cases (%)Red cell populationsa
I Median % (observed range)II Median % (observed range)III Median % (observed range)
  • a

    Type I population, normal red cells; type II, partially deficient cells; type III cells, complete deficiency of GPI-linked antigens (Fig. 1).

  • b

    Four distinct categories are recognized based on the distributions of type I, II, and III red cell populations.

  • c

    Type II or III components comprise <5% of the total PNH clone.

  • d

    Type II cells comprise >5% of the total PNH clone.

I and III (type II <5%)c33 (34)84 (17–99.9)0 (0–3.2)16 (0.2–79.8)
I and II (type III <5%)c4 (4.1)83.5 (6–93.6)16.5 (5.6–94)0 (0–0.8)
I, II, and III (type II >5%)d59 (60.8)73.82 (11–99.52)5.6 (0.11–64.1)16.0 (0.22–77)
III only1 (1.1)100

In the majority of patients with PNH, the granulocyte clone is proportionately larger than the red cell clone. This difference is almost certainly due to the shortened in vivo life span of PNH red cells when compared to normal red cells and can be further complicated by the effect of blood transfusions. Recent studies have shown that type III red cells survive for about 17–60 days. Type II cells, when expressing at least 15% of the normal amount of CD59, are protected against complement-mediated lysis (44). Interestingly, by examining reticulocytes in PNH using thiazole orange and CD59, the percentage of PNH reticulocytes appears to be similar to that of the neutrophil clone (45, 46). This illustrates that both erythroid and neutrophil bone marrow precursors have a similar proliferative capacity in PNH and that assessing reticulocytes in PNH may provide significant information regarding rates of erythropoiesis, particularly in multitransfused patients with PNH.

A clear aim for the future is to establish a more robust gating strategy for red cells in line with contemporary flow cytometry approaches based on physical and fluorescence characteristics. We are currently exploring the possibility of using antibody fragment variable (Fv) reagents generated by phage display systems (47). These small and highly specific reagents may overcome the problems of red cell agglutination and allow successful multicolor phenotyping of red cells.

Analysis of Granulocytes.

Previous studies of granulocytes in PNH were predominantly single or two-color investigations based on identifying granulocytes by FSC/SSC characteristics (25, 26, 34–36, 48). Although this approach of analyzing single antigens is acceptable, if limited, the gating approach to identifying granulocytes based on physical characteristics alone has potential pitfalls. The assumption made in many cases of PNH is that the majority of cells within the granulocyte gate are neutrophils. This may be true as a generalization, although significant numbers of eosinophils may also be present. This becomes relevant when attempting to detect small granulocyte PNH clones using single antibodies such as CD16. Although suitable for neutrophil study, CD16 is not expressed by normal eosinophils (49). A further drawback of the FSC/SSC gating approach is that in cases of myelodysplastic syndrome where significant numbers of agranular neutrophils can be seen, identifying a population of granulocytes by physical characteristics alone is not always possible and a combined lineage marker/SSC gating approach is required.

An optimal three-color staining procedure for detecting PNH granulocytes should involve sample manipulation procedures that minimize cell loss and use SSC versus a non–GPI-linked lineage antigen, e.g., CD15, CD33 or CD45, to identify granulocytes (Fig. 2). This leaves a further two fluorescence channels with which to examine expression of GPI-linked antigens. The method of staining, i.e., prestain followed by lysis or lysis followed by staining, is a matter of individual preference, each having its individual merits. For whole blood prestaining methods using CD55 and CD59, an optimal antibody titer may be difficult to establish as effectively all cells will be stained including red cells and platelets. By prelysing blood samples with ammonium chloride or other commercially available reagents to isolate leukocytes, CD55 and CD59 antibodies can be optimally titered to label a standard number of cells. Antibodies such as CD16, CD24, and CD66 work equally well by either method, and a combination of CD16 and CD66 has the added advantage of clearly separating neutrophils from eosinophils.

Figure 2.

Granulocyte analysis: Gating strategies and analysis of GPI-linked antigen expression on peripheral blood granulocytes. a: Dot plot shows a gating region (R1) that identifies granulocytes based on CD15 positivity and high SSC characteristics. b: Plot shows a granulocyte gate based solely on light scatter characteristics of leukocytes. Plots (c–e) are representative profiles from three cases of PNH. c: Plot shows a granulocyte PNH clone of 67% defined by weak CD16 expression and absence of CD55. Residual normal cells comprise 30% of total granulocytes. d: Plot shows a CD55 and CD59-negative PNH granulocyte clone of 82%. e: Plot shows a small PNH clone (0.3%) in a patient with untreated aplastic anemia (50,000 event file). The CD59+CD16 events in this plot are residual normal eosinophils.

As with red cell analysis, it is important to demonstrate deficiency of at least two GPI-linked antigens in order to establish a diagnosis of PNH. This can be achieved simultaneously with multicolor combinations. CD16 is worthy of a special mention as used in isolation can cause confusion and potential misdiagnosis. An infrequent polymorphism in the CD16 determinant renders the antigen undetectable to some monoclonal antibodies and not others (49). In addition, samples with proportionately high eosinophils will have a large CD16-negative component as this antigen is not constitutively expressed on eosinophils.

Unlike red cells, which can remain suitable for testing up to 25 days if kept at 4°C, granulocyte analysis is most reliably carried out within the first few hours following collection (34). Reliable separation of type I, II, and III populations is best achieved with anti-CD59. However, it is evident from a number of studies that the separation is inferior to that seen with red cells (34, 35, 48). CD16 appears unsuitable for this purpose as many PNH neutrophils show weak CD16 expression (34). As granulocytes age in vitro, their nonspecific binding of antibodies and autofluorescence levels increase as cell viability declines. Although negative isotype controls in the classical sense are not strictly necessary, they do provide valuable information regarding nonspecific uptake of antibodies by granulocytes, particularly in aged samples. However, in the majority of PNH cases, a residual population of normal neutrophils provides an internal control or reference population with which to compare any GPI-deficient PNH clone and to confirm the reactivity of monoclonal antibodies. What is also evident in a number of patients is a discrepancy between granulocyte and red cell GPI phenotype in that granulocytes show partial deficiency and red cells show complete deficiency or a mixture of type II and III populations (34–36).

Analysis of Monocytes.

Analysis of peripheral blood monocytes in patients with PNH has been undertaken by a number of groups. They assessed the extent of the PNH clone and the suitability of a variety of monoclonal antibodies to GPI-linked antigens known to be expressed on normal monocytes (23, 36, 48). In all studies, monocytes were identified by light scatter characteristics and PNH monocytes within this gate by lack of either CD14, CD55, CD59, or CD48 determinants. No clear consensus was established on the most suitable antibodies, although CD14 and CD55 were favored by the majority of authors. Not unexpectedly, good concordance has been reported between the size of the monocyte PNH clone and the granulocyte PNH clone as the cells are derived from a common myeloid progenitor. However, Alfinito et al. (36) reported a higher percentage of CD14-negative monocytes in patients compared to PNH granulocytes. Moreover, they found that the percentage of CD14-positive (normal) monocytes directly correlated with hemoglobin and platelet count levels and that patients with <10% CD14-positive cells had active disease. Detecting monocytes with partial deficiency of GPI-linked antigens has proven difficult, with only CD48 showing intermediate levels of staining in some patients (48). As absolute numbers of monocytes are generally low in patients with PNH, it can be difficult to obtain sufficient numbers of “monocyte events” to clearly define type II and III populations on fluorescence histograms.

A significant improvement in analyzing monocytes in PNH could be made by adopting a combined lineage marker/SSC gating strategy. As CD14 is GPI anchored, it is unsuitable for identifying monocytes; CD64 (FcγRI), CD33, and CD4 are acceptable alternatives. In multicolor combinations, we used either a CD4dim/SSClow or a CD33bright/SSClow gate to identify monocytes and then examined CD14, CD55, CD59, and CD52 expression (Fig. 3). Using this approach, the monocyte PNH clone is most clearly separated from normal monocytes by lack of CD14 expression. In our current series of patients, the size of the monocyte PNH clone is significantly similar to that of the granulocyte PNH clone (Spearman's R 0.847, P < 0.001, n = 31).

Figure 3.

Analysis of GPI-linked antigens on monocytes in a patient with PNH. Peripheral blood leukocytes are stained with anti-CD4 and anti-CD14. a: A monocyte gate (R1) is drawn around cells with intermediate CD4 staining and low/intermediate SSC characteristics. b: Analysis of the GPI-linked antigen CD14 for gate R1 shows a PNH monocyte clone (CD14) of 74% and normal monocytes (CD14+) of 26%.

Analysis of Lymphocytes.

Using multicolor flow cytometry, GPI-deficient lymphocytes can be demonstrated in a high proportion of patients with PNH, particularly those with a long duration of disease (22, 25, 35, 48, 50–52). When compared with the size of the granulocyte PNH clone, the proportion of PNH B, T, and NK cells is usually much smaller. This is evident even in patients whose hematopoiesis is maintained predominantly from the PNH stem cell as shown by a large (>95%) granulocyte PNH clone. This difference can be explained by the long life span of residual normal T and B lymphocytes generated prior to the onset of PNH. In terms of establishing a diagnosis of PNH, examining expression of GPI-linked proteins on lymphocytes alone is not recommended. The GPI-linked complement regulatory molecules CD55 and CD59 can show variable levels of expression on normal T cells (53). CD48 (function unknown) provides the clearest separation between normal and PNH T, B, and NK cells (51).

The presence of GPI-deficient lymphocytes in patients with PNH has recently been used as a model with which to study the kinetics of T-cell production in vivo (54). Deficiency of GPI-linked antigens from T lymphocytes appears not to effect either in vitro or in vivo function as they undergo conversion from a naive phenotype (CD45RA+) to a memory phenotype (CD45R0+) following stimulation (54, 55). Interestingly, flow cytometry studies of lymphocytes in patients who have undergone spontaneous remission of PNH have shown that GPI-deficient B and T cells can persist for many years following normalization of red cells and granulocytes (27, 56).

Analysis of Platelets.

Thrombosis is a frequent clinical feature of PNH and is a major cause of morbidity and mortality. Although undoubtedly related to deficiency of GPI-anchored proteins and possibly to intravascular hemolysis, the exact mechanism by which this occurs remains ill defined (57, 58). Moreover, it is also unclear why thrombotic events occur in unusual anatomical locations such as intra-abdominal and/or liver veins. Current research activity is centered around deficiency of the GPI-anchored urokinase plasminogen activator receptor (uPAR or CD87) and other GPI-linked complement regulatory antigens (CD55 or CD59) from platelets as a possible cause. The biochemical nature of this hemostatic/thrombolytic pathway has not been fully elucidated (59).

As with red cells and granulocytes, there appears to be no clear consensus in the literature on the immunophenotypic analysis of platelets in PNH. Studies are generally undertaken on platelet-rich plasma (PRP) with or without the addition of agents to prevent activation. Expression of CD55 and CD59 has been examined in normal controls and patients with PNH by a few investigators (60–62). Identification of platelets in most instances was based on FSC/SSC characteristics with inconsistent use of non–GPI-linked positive controls (CD42b or CD61) to establish gating efficacy. The fluorescent intensity of CD59 and CD55 expression on normal platelets appears to be weak, with up to 10% of normal platelets not expressing these two antigens. Whether this is genuine or a reflection of an insensitive technique remains unclear. As a consequence, populations of GPI-deficient platelets in PNH patients are not easily distinguished and identification of type II and III populations is difficult to achieve. Removal of erythroid cells from PRP is claimed to improve the sensitivity as antibodies preferentially bind to red cells (62). In a two-color analysis, Maciejewski et al. (60) used a combination of CD41a and CD42b to identify platelets and examined CD55 and CD59 expression on normal and PNH platelets. On storage, expression levels of both GPI-linked antigens were found to decline over a 5-day period. Substantial proportions of GPI-deficient platelets were detected in most patients with PNH.

The diagnostic utility and clinical relevance of examining platelets in PNH patients have not been established. As might be predicted for myeloid-derived cells, the percentage of PNH platelets is highly correlated with the percentage of PNH granulocytes (62). The understanding of the thrombotic tendency in patients with PNH is ill defined and is a priority for future research.

Detection of Hematopoietic Cells with A PNH Phenotype in Normals and in Patients Treated with CAMPATH-1H Monoclonal Antibody.

In a recent publication, Araten et al. (63) reported the presence of rare GPI-deficient neutrophils in normal individuals. These cells have a frequency of 10–51 per million cells. When collected by flow sorting, they were shown to contain mutations of the pig-a gene. A similar analysis of red cells also identified GPI-deficient clones at a frequency of 0.0001%. These findings demonstrated that pig-a mutations exist frequently among normal individuals but this alone is not sufficient for development of PNH. A second factor is required for clinical PNH to develop. In most instances, this appears to be a failure of normal hematopoiesis in the form of aplastic anemia. This forms the basis of the dual pathogenesis theory for the development of PNH originally proposed by Dacie in 1980 (64) and further developed by Rotoli and Luzzatto in 1989 (65).

Further evidence to support the dual pathogenesis theory of PNH and that cells with a PNH phenotype and genotype can be found in normal individuals comes from recent studies using the therapeutic monoclonal antibody CAMPATH-1H (66). CAMPATH-1H is a humanized therapeutic monoclonal antibody directed against the GPI-linked antigen CD52 (67). It has been successfully used in the treatment of B chronic lymphocytic leukemia (CLL) and T prolymphocytic leukemia and has resulted in remissions in patients refractory to previous conventional drug therapies (68, 69). The antibody is cytotoxic to all lymphocytes and causes a profound depletion of peripheral blood and bone marrow lymphocytes. In approximately two thirds of CLL patients treated with CAMPATH-1H, the majority of T cells that regenerate during or immediately following cessation of therapy are deficient in all GPI-linked antigens, i.e., they have a PNH phenotype (68). Molecular analysis of these cells showed the presence of a pig-a gene mutation. Using an extremely sensitive mutation-specific PCR-based technique, the same mutation was also detectable in the patient's mononuclear cells prior to therapy, even though GPI-deficient T cells were not detectable by flow cytometry at that time (66). In terms of the dual pathogenesis theory of PNH, the CAMPATH-1H antibody is providing the selective pressure allowing preexisting PNH T cells the opportunity to expand to levels detectable by flow cytometry (Fig. 4).

Figure 4.

Flow cytometric analysis of peripheral blood T cells in a patient with B-cell CLL pre (a) and posttreatment (b) with the therapeutic monoclonal antibody CAMPATH-1H (anti-CD52). Prior to treatment, all T cells express the GPI-linked antigens CD52 and CD59. Following treatment, T cells are predominantly CD52 and CD59 negative. These cells with a PNH phenotype contain a pig-a gene mutation.

Flow Cytometry: Recommendations for Testing

Peripheral blood is the most suitable specimen for immunophenotyping PNH. A recent blood transfusion history should be provided with all requests. It is recommended that protocols are available for testing red cells and granulocytes and, whenever possible, both cell types should be screened. This is important for a number of reasons:

  • 1A small proportion (5%) of patients have only granulocyte PNH clones.
  • 2After a severe episode of hemolysis, the frequency of GPI-deficient red cells may be below the detection limit of the assay. Furthermore, if patients have received multiple transfusions prior to testing, the PNH screen is effectively performed on transfused blood and this may lead to erroneous results.
  • 3In patients with severe hypoplastic anemia, there may be insufficient numbers of granulocytes for analysis.

Deficiency of at least two GPI-linked antigens (CD55 and CD59) should be demonstrated from red cells as single-antigen deficiencies have been reported as rare inherited conditions (42, 43). Red cell analysis provides the clearest discrimination between types I, II, and III cells and may predict clinical phenotype (70). For the analysis of GPI-linked antigens on peripheral blood granulocytes, a combined lineage marker (non–GPI-anchored)/SSC gating approach is desirable, although FSC/SSC gating is acceptable in the majority of cases. As for red cells, deficiency of at least two GPI-linked antigens (CD16, CD24, CD55, CD59, or CD66) should be demonstrated in order to establish a diagnosis of PNH (34).


PNH is characterized by a triad of intravascular hemolysis, cytopenias secondary to underlying bone marrow failure, and a predisposition to venous thrombosis. PNH is a chronic condition with a median survival of between 10 and 15 years. The most common causes of morbidity and mortality are venous thrombosis, particularly affecting the hepatic veins, and progressive bone marrow failure (27, 28). The only potentially curative therapy for PNH is allogeneic bone marrow transplantation (BMT). However, only a minority of patients are suitable candidates for BMT (are young with a potential donor). For most of these patients, the risks of BMT are too high to justify its routine use in such an indolent disorder. Optimal conservative management of PNH can be expected to ameliorate many of the symptoms of the disease and to prevent the complications, which are the most frequent causes of death in PNH.

In view of the continuous intravascular hemolysis in PNH, all patients should receive folic acid and those who become iron deficient will need iron supplementation. Blood transfusions should be avoided when possible as commencing a transfusion program for an individual patient may lead to several years of transfusions. However, a minority of patients are transfusion dependent. In contrast to almost all other chronic anemias, iron overload is usually not a problem due to the excessive urinary iron loss seen in PNH. The major complication of PNH is venous thrombosis, which occurs in up to 50% of patients and results in the death of approximately one third of patients with PNH large clones. Prophylactic anticoagulation with warfarin appears to significantly reduce the incidence of thrombosis and should be considered in all patients without a contraindication, such as severe thrombocytopenia. Close observation of blood count parameters allows the early identification of the onset of bone marrow failure, which is seen in approximately 10% of patients. Early intervention with immunosuppressive therapy such as antilymphocyte globulin and/or cyclosporin A may reverse the progressive development of cytopenias.

Despite the fact that PNH is a serious condition, the majority of patients remain alive and live reasonably well with their disease. Historical series of patients identify that even without modern supportive care, approximately one fourth of patients survive 25 years and a significant proportion (>15%) will spontaneously remit and make a full recovery from PNH (27). There is currently no reliable method to identify those patients who may remit. However, there is preliminary evidence that serial flow cytometric analyses may predict which patients will improve over time (70).


Analysis of GPI-linked proteins on red cells and granulocytes by flow cytometry provides a specific and sensitive technique for screening and diagnosing PNH. Due to the low incidence of PNH, there is often a significant delay before a diagnosis is made. Retrospective studies of the clinical histories of individual PNH patients have shown that this delay can be as long as 15 years in rare instances, although up to year is quite frequent (71). Screening can be performed rapidly and although requiring expertise in flow cytometry, has significant advantages over other methods such as the Ham test. For newly diagnosed patients with PNH, it is highly desirable to be able to predict the clinical course and identify patients at increased risk of developing thrombosis and those who may progress to hemolytic or hypoplastic types. The ability to determine whether patients will undergo spontaneous remission has major implications for treatment decisions, particularly for those patients considered for BMT.

The clinical and prognostic significance of monitoring PNH clone size by serial flow cytometry is unknown and we are currently undertaking prospective studies to evaluate this. From an analysis of 120 cases, it is apparent that patients with hemolytic PNH have predominantly completely deficient (type III) red cells, with all but 2 of 49 patients having >10% type III cells. Ten hypoplastic patients had only partially deficient (type II) red cells. As the percentage of PNH granulocytes reflects most accurately the size of the PNH clone, serial measurements of PNH granulocytes in peripheral blood samples should theoretically give an accurate indicator of disease activity. Preliminary data on 25 patients with a follow-up of longer than 3 years are already providing encouraging results (70). In one third of patients, there was a significant decrease in the size of the granulocyte clone, with almost complete disappearance in three patients. Thus, it appears that the size of the PNH clone is an important determinant of the clinical phenotype of individual patients and serial studies allow prediction of prognosis. Consolidation of these preliminary findings is a critical aspect of our future research. It highlights, firstly, the importance of establishing registries for patients with rare conditions and, second, the significant role of flow cytometry in contemporary clinical diagnostics.


We are grateful to Dr. Julian Miller (Cymbus Bioscences, UK) for continued support of PNH research at HMDS.