Natural history of paroxysmal nocturnal haemoglobinuria using modern diagnostic assays


  • Victor M. Moyo,

    1. Department of Medicine, Division of Hematology/Oncology, University of Connecticut Health Center, Farmington, CT
    Search for more papers by this author
  • Galina L. Mukhina,

    1. Department of Oncology, Division of Hematologic Malignancies, Sidney Kimmel Comprehensive Cancer Center at Johns Hopkins, Baltimore, MD
    Search for more papers by this author
  • Elizabeth S. Garrett,

    1. Department of Oncology, Division of Biostatistics, Johns Hopkins School of Medicine, Baltimore, MD, USA
    Search for more papers by this author
  • Robert A. Brodsky

    1. Department of Oncology, Division of Hematologic Malignancies, Sidney Kimmel Comprehensive Cancer Center at Johns Hopkins, Baltimore, MD
    Search for more papers by this author
    • *

      RAB is a Clinical Research Scholar of the Leukemia and Lymphoma Society.

Dr Robert A. Brodsky, Division of Hematological Malignancies, Bunting-Blaustein Cancer Research Building, Sidney Kimmel Comprehensive Cancer Center at Johns Hopkins, 1650 Orleans St, Rm. 242, Baltimore, MD 21231, USA.


Paroxysmal nocturnal haemoglobinuria (PNH) is an uncommon, acquired disorder of blood cells caused by mutation of the phosphatidylinositol glycan class A (PIG-A) gene. The disease often manifests with haemoglobinuria, peripheral blood cytopenias, and venous thrombosis. The natural history of PNH has been documented in retrospective series; but there has only been one study that correlated the more sensitive and specific flow cytometric assays that have become available in the last decade with severe symptoms associated with PNH. In a retrospective analysis of 49 consecutive patients with PNH evaluated at Johns Hopkins, large PNH clones were associated with an increased risk for thrombosis as well as haemoglobinuria, abdominal pain, oesophageal spasm, and impotence. Of the 14 (29%) patients that developed thrombosis, nine died; six of these from complications related to thromboses. According to logistic regression modelling, for a 10% change in PNH clone size, the odds ratio for risk of thrombosis was estimated to be 1·64. No patient with <61% PNH granulocytes developed a thrombosis, whereas 12 of 22 patients (54·5%) with ≥61% PNH granulocytes manifested with thrombosis. These data not only confirm that the size of the PNH clone correlates with the risk for thrombosis, but they also suggest a correlation of PNH clone size to more symptomatic PNH.

Paroxysmal nocturnal haemoglobinuria (PNH) is an acquired clonal haematopoietic stem cell disorder caused by mutations in an X-linked gene called phosphatidylinositol glycan class A (PIG-A). The PIG-A gene product is necessary for the first step in the biosynthesis of glycosylphosphatidyinositol anchors, the transfer of GlcNAc from UDP-GlcNAc to form N-acetylglucosaminyl phosphatidylinositol (GlcNAc-PI) (Miyata et al, 1993, 1994; Takeda et al, 1993; Bessler et al, 1994). Hence, PNH cells have a marked deficiency of all proteins that are normally attached to the cell membrane by the GPI anchor. Certain GPI-anchored proteins (e.g. CD55 and CD59) serve as cell surface complement regulators. Consequently, PNH patients frequently manifest with a complement-mediated intravascular haemolysis; other prominent manifestations include peripheral blood cytopenias (anaemia; haemoglobin <12 g/dl, or thrombocytopenia; platelet count <150 × 109/l and neutropenia; neutrophil count of 1·5 × 109/l), and a proclivity for venous thrombosis (Rosse, 1997). PNH may occur de novo (classical PNH) or in the setting of aplastic anaemia (hypoplastic PNH). With the development of more sensitive flow cytometric assays for detecting GPI-anchored protein deficiency, it has been shown that more than 50% of patients with aplastic anaemia possess small PNH populations at diagnosis, thus underscoring the pathophysiological relationship between these disorders (Schrezenmeier et al, 1995; Mukhina et al, 2001).

Before 1990, the diagnosis of PNH was established using complement-based assays, such as the Ham test (Ham, 1937) and the sucrose haemolysis test (Hartmann & Jenkins, 1965). A limitation of these assays is that they are not sensitive enough to detect small PNH populations, and they may be falsely negative in patients requiring regular red cell transfusions. Previous series examining the natural history of PNH were likely to have excluded patients with small PNH populations (hypoplastic PNH). Furthermore, the Ham and sucrose haemolysis tests do not assess the size of the PNH clone. Modern diagnostic assays employ monoclonal antibodies that bind to specific GPI-anchored proteins, and enable the percentage of PNH cells to be measured (Vanderschoot et al, 1990; Hall & Rosse, 1996). However, no single monoclonal antibody is optimal for the detection of GPI-anchored proteins as different haematopoietic lineages display different GPI-anchored proteins. Previously, we described a novel flow cytometric assay for PNH that utilizes a fluorescently labelled inactive variant of the protein aerolysin (FLAER). FLAER binds with high affinity and specificity to the glycan portion of the GPI anchor and serves as a more comprehensive assay to detect PNH cells (Brodsky et al, 1999, 2000). To improve our understanding of the natural history of PNH, we studied 49 consecutive patients with PNH evaluated at Johns Hopkins using these more sensitive and specific diagnostic assays.

Material and methods

Forty-nine consecutive patients with detectable PNH clones, with or without aplastic anaemia, were evaluated at Johns Hopkins Hospital between 1995 and 2002. This sample represents all the patients with PNH that were seen at this institution during the specified period. In some of these patients, the PNH diagnosis was made elsewhere before referral to Johns Hopkins. The initial diagnosis of PNH in some patients was made using a Ham test (Ham, 1937) sucrose haemolysis test (Hartmann & Jenkins, 1965), or antibody-based flow cytometry assays (Vanderschoot et al, 1990; Hall & Rosse, 1996) but was confirmed in 44 cases using FLAER (Brodsky et al, 1999). These 44 patients are the basis of this analysis. Five cases were excluded from the analysis because FLAER data were not available for at least two cell lineages to confirm the diagnosis of PNH. In all patients, a detailed history had been obtained and a physical examination had been performed during the hospital visit. Permission to perform this retrospective review was approved by the Joint Committee on Clinical Investigation of the Johns Hopkins Hospital. Data from each patient's medical record was reviewed to assess for the development of thrombosis, the presence of haemoglobinuria, abdominal pain, oesophageal spasm and erectile dysfunction. The development of clonal sequelae and cytopenias were also noted. The diagnosis of aplastic anaemia was established by bone marrow aspiration and peripheral blood counts as recommended by the International Agranulocytosis and Aplastic Anemia Study (1987). Patients who had demonstrable PNH cells in conjunction with aplastic anaemia were defined as ‘hypoplastic’ PNH. De novo (classical) PNH was defined as two or more lineages displaying deficient GPI-anchored proteins in conjunction with signs and symptoms consistent with haemoglobinuria and/or venous thrombosis, in the absence of a hypocellular bone marrow (Mukhina et al, 2001). Patients were classified as having symptoms consistent with oesophageal spasm if they described three or more episodes of chest tightness and difficulty swallowing. Impotence was defined as erectile failure during attempted intercourse more than 75% of the time (Spark, 2002). Myelodysplastic syndromes were defined using the French-American-British classification (Bennett et al, 1982).

Measurement of PNH cells

Red cell assay.  For detection of PNH erythrocytes, fluorescein isothiocyanate (FITC)-conjugated anti-CD59 was used in conjunction with aerolysin as previously described (Brodsky et al, 2000).

FLAER assay for the detection of GPI anchor protein deficiency.  FLAER has fluorescent properties similar to the FITC antibodies and is available from the web site ( After lysing erythrocytes with ammonium chloride, neutrophils, were identified on the basis of cell size granularity and by staining with a specific antigranulocyte antibody [phycoerythrin (PE)-conjugated anti CD15; Immunotech, Marseille, France]. B- and T-lymphocytes were identified by staining with specific antibodies (PE-conjugated anti-CD3 T cells and CD19 for B cells; Immunotech). Detection of GPI anchor expression on granulocytes and lymphocytes was assessed using FLAER. Gates used to define FLAER-negative granulocytes and lymphocytes were based on normal controls for the respective cell groups analysed on the same day.

Statistical analysis

This was a retrospective analysis of a series of patients who had received care at this institution either on a consultation basis or while receiving treatment for their underlying problem. Logistic regression modelling was used to examine the relationship between thrombosis and the size of the PNH clone as determined by FLAER in granulocytes of patients in whom FLAER data was available. Two sample t-tests allowing for unequal variances were performed to assess differences in PNH clone size in patient subgroups. Odds ratio of symptoms were calculated and tested for statistical significance using Fisher's exact test. Multiple linear regression analysis was used to assess the effects of symptoms on PNH clone size, adjusting for other symptoms. Time to event and cumulative incidence were described graphically using Kaplan–Meier curves. Differences in survival curves were tested using a log-rank test. All analyses were performed using statistical software from spss (SPSS Inc., Chicago, IL, USA) and stata 7.0. (Stata Corp., College Station, TX, USA). P-values <0·05 were considered significant.


Table I shows the patient characteristics. There were 24 male and 25 female patients with a median age at presentation of 34 (range: 6·3–80·7) years. Median disease duration was 2 (range: 0·5–35) years. Twenty patients presented with classical PNH, 25 with aplastic anaemia had asymptomatic PNH clones, and four patients (originally diagnosed as aplastic anaemia) evolved into classical PNH. A history compatible with oesophageal spasm was noted in 10 of 43 patients (23%); seven of 20 males (35%) evaluated had symptoms of impotence and all but one patient had at least one cytopenia. Nine patients died from the complications of PNH. FLAER was used to determine the percentage of PNH granulocytes in 44 of 49 patients. The median PNH clone size in patients with aplastic anaemia was 16·8 (range: 1–96·6)% compared with a median PNH clone size of 91·6 (range: 40–98·7)% for patients with classical PNH. Fourteen patients (29%) developed thrombosis and in 12 of these FLAER data were available. Ten patients had a single thrombotic event, three patients had two thrombotic events, and one patient had three thrombotic events. The location of the first thromboses were intracerebral; sagittal sinus, cerebral vein thrombosis (n = 4), intra-abdominal; hepatic vein thrombosis (Budd-Chiari), portal or systemic vein thrombosis (n = 9), lower extremity deep vein thrombus (n = 1). Nine patients died; six of these from complications related to thromboses. Six of the patients with thromboses had thrombocytopenia <50 × 109/l. The median size of the PNH granulocyte population in patients with a history of thrombosis was 91·6% (range: 61–98·0), compared with 30·35% (range: 0·5–98·7) in patients without thrombosis.

Table I.  Patient characteristics (n = 49).
  1. *43 patients, †20 males, ‡47patients, §44 patients, ¶40 patients.

Sex, n (%)
 Male24 (49)
Age at diagnosis, median (range)34·2 (6·3–80·7)
Presentation, n (%)
 De novo (classical) PNH20 (41)
 Aplastic anaemia with PNH clone25 (51)
 Aplastic anaemia preceding overt PNH with symptoms4 (8)
Associated symptoms during attacks, n (%)
 Oesophageal spasm*10 (23)
 Impotence†7 (35)
 Haemoglobinuria21 (43)
 Fatigue41 (84)
 Headache12 (24)
 Abdominal pain17 (35)
Cytopenias in haematopoetic lineages, n (%)
 01 (2)
 15 (10)
 216 (31)
 329 (57)
Size of PNH clone at first determination (%), median (range)
 Erythrocytes‡7·45 (0–84·2)
 Granulocyte§60·55 (0·5–98·0)
 T-lymphocytes¶0·17 (0–40·7)
 B-lymphocytes¶0·08 (0–43·7)
Anatomic location of first thrombosis, n (%)14 (29)
 Intracerebral4 (8)
 Intra-abdominal8 (16)
 Lower extremity1 (2)
 Dermal1 (2)
Malignancies, n (%)
 Ovarian cancer1 (2)
 Myelodysplastic syndromes3 (6)
Patients who died, n (%)9 (18)

We found a strong association between the size of the PNH clone (as measured in granulocytes) and thrombosis. No patient with <61% PNH granulocytes developed a thrombosis, whereas 12 of 22 patients (54·5%) with ≥61% PNH granulocytes developed a thrombosis (Figs 1 and 2). By logistic regression, we found that, for a 10% change in the size of the PNH granulocyte clone, the odds ratio for thrombosis was 1·64 (P = 0·008).

Figure 1.

Size of granulocyte paroxysmal nocturnal haemoglobinuria (PNH) clone according to the absence/presence of thrombosis (n = 44).

Figure 2.

Proportion of thrombosis over time according to size of paroxysmal nocturnal haemoglobinuria (PNH) clone.

Four of the six symptoms of interest correlated with PNH clone size. There were significant differences in the size of PNH clone in patients with symptoms of abdominal pain (P < 0·0001), haemoglobinuria (P < 0·0001), oesophageal spasm (P < 0·0001) and impotence (P < 0·0004). The mean size of the granulocyte PNH clone in these patients was at least twofold that of patients without these clinical manifestations. Figure 3 shows that patients with abdominal pain (Fig 3A), haemoglobinuria (Fig 3B), oesophageal spasm (Fig 3C) and impotence (Fig 3D) tended to have large PNH clones. The presence of a headache (Fig 3E) and fatigue (Fig 3F) did not appear to differ according to size of PNH granulocyte clone. Of the nine deaths, six were from complications related to thrombosis. Three of these deaths were caused by haemorrhage following anticoagulation while the other three died from complications of liver failure. Two patients died from malignancies, one from ovarian cancer and the other from a myelodysplastic syndrome that evolved into acute leukaemia. The leukaemia blasts from this patient were FLAER-negative, demonstrating that the leukaemia occurred within the PNH clone (data not shown). Another patient died from complications of sepsis. Although there was a trend towards greater survival in patients with PNH clones <60% (Fig 4), this trend did not attain statistical significance (P = 0·39).

Figure 3.

Relationship of presence or absence of symptoms and paroxysmal nocturnal haemoglobinuria (PNH) clone size. The relationship of granulocyte clone size to the absence or presence of individual symptoms. Values in parenthesis indicate the percentage mean PNH clone size ± SD. (A) Abdominal pain (n = 44); yes = 17 (81·6 ± 25·0), no = 27 (35·7 ± 35·6). (B) Haemoglobinuria (n = 44); yes = 21 (81·3 ± 16·3.0), no = 23 (24·5 ± 29·6). (C) Oesophageal spasm (n = 40); yes = 10 (87·8 ± 12·3.0), no = 30 (54·8 ± 38·4). (D) Impotence in males (n = 19); yes = 6 (94·9.3 ± 2·6), no = 13 (46·6 ± 35·5). (E) Headache (n = 44); yes = 12 (61·4 ± 42·1.0), no = 32 (50·4 ± 37·9). (F) Fatigue (n = 44); yes = 37 (54·8 ± 38·4.0), no = 7 (46·3 ± 44·0).

Figure 4.

Overall survival and size of paroxysmal nocturnal haemoglobinuria (PNH) clone (n = 44).


The PNH is a rare and potentially fatal haematological disease. The natural history of the classical form of PNH is well-documented (Hillmen et al, 1995; Socie et al, 1996); however, most of the patients included in these series were diagnosed using complement-based red cell assays. Thus, patients who were transfusion-dependent and those with small PNH populations (i.e. AA/PNH) were probably under-represented. Moreover, complement-based red cell assays are not quantitative, making it impossible to correlate disease manifestations with the size of the PNH clone. Our previous work showed that the FLAER assay for detecting PNH is more accurate in detecting small PNH clones when compared with the more conventional anti-CD59 flow cytometric assays (Brodsky et al, 1999, 2000). The results of the present study further validate the FLAER technique for identifying PNH clone size and suggest that FLAER is useful in identifying individuals at risk for thrombosis and other manifestations of PNH. While validating the work of Hall et al (2003) who recently showed that large PNH clones are predictive of thrombosis, our study further suggests that large PNH granulocyte clones also predict for abdominal pain, haemolysis, oesophageal spasm and erectile dysfunction. The mechanism underlying the impotence and oesophageal spasm is not entirely understood; however, as in sickle cell anaemia, it is likely that free haemoglobin resulting from intravascular haemolysis depletes tissue nitric oxide (Reiter et al, 2002).

Six of the nine patients who died in our series had thromboses, highlighting the significance of thrombosis for mortality in patients with PNH. The mechanism of thrombosis in PNH is also unclear, but is thought to be related to complement-mediated attack on platelets, which in turn results in a phospholipid-rich milieu and the generation of factors Va, Xa and the prothrombin complex (Hall et al, 2002). A recent study also suggests that complement mediated damage to leucocytes leads to increased leucocyte-derived tissue factor levels in plasma, which in turn may trigger thrombosis (Liebman & Feinstein, 2003).

Hall et al (2003) also reported an increased risk for thrombosis in PNH patients with large PNH clones and recommended primary prophylaxis with warfarin for all PNH patients with large clones, a platelet count of ≥100 × 109/l, and no contraindications to anticoagulation. Nevertheless, it remains unclear whether all PNH patients at high risk for thrombosis should receive primary warfarin prophylaxis. Most patients with large PNH clones in our series had concomitant thrombocytopenia (<50 × 109/l). Moreover, maintaining a therapeutic International Normalized Ratio (INR) is often difficult in the setting of PNH attacks that may be associated with anorexia, abdominal pain, nausea and vomiting. Thus, while our study identifies a population of PNH patients that are at high risk for thrombosis, it also helps to highlight the inherent risk for anticoagulation in PNH patients. In three of six patients with thromboses, haemorrhage following anticoagulation contributed to the cause of death. Future randomized studies in this group are required before one can recommend primary prophylactic anticoagulation for PNH.


This work is supported in part by National Institutes of Health grant no. CAA70970 to RAB.