Paroxysmal nocturnal hemoglobinuria (PNH) is an uncommon acquired hematopoietic stem cell disorder whose clinical manifestations include intravascular hemolysis, venous thrombosis, bone marrow failure, and, occasionally, transition to a myelodysplastic syndrome or acute myelogenous leukemia11,39,43. The characteristic biochemical feature in PNH is a lack of glycosylphosphatidylinositol (GPI)-anchored proteins (GPI-APs), due to an incomplete assembly of the GPI anchor14,17. The absence of complement regulatory proteins CD55 and CD59 results in complement-mediated hemolysis and hemoglobinuria. In the affected hematopoietic cells from patients with PNH, the first step in biosynthesis of the GPI anchor is defective. The X-linked gene termed PIG-A, the product of which participates in this reaction step, has an acquired mutation in the abnormal hematopoietic cells of all patients with PNH reported to date18,19,40.
Despite the fact that all patients with PNH have an acquired PIG-A mutation, the clinical course of PNH is highly variable. Some patients have a severe and rapidly fatal course, while others have a chronic illness with few life-threatening complications. Cases of spontaneous remission have even been reported11. The development of thrombosis is generally felt to be a grave prognostic feature, although this complication is allegedly rare in Asian patients with PNH7,15,16. From several clinical studies on patients with PNH, the clinical manifestations of PNH in white and Asian patients may be quite different7,11,15,16,39,43. White patients are reported to have more thrombosis, whereas Asian patients may have more aplastic anemia, but there has never been a direct comparison between these 2 ethnic groups.
To elucidate and begin to understand the clinical differences between PNH as seen in Japan and the United States, 2 large cohorts of patients were identified and analyzed. First, clinical and laboratory characteristics at diagnosis were compared, and significant differences were identified. The basis for these differences in the clinical expression of PNH was then hypothesized to relate to the size of the abnormal PIG-A mutant clone. The 2 cohorts of patients with PNH were studied by quantitative flow cytometry using GPI-linked CD55 and CD5910, to directly compare the proportions of PNH clonal populations and to correlate clinical manifestations of PNH with clone size.
In most patients who have clinical manifestations of PNH, hematopoiesis is dominated by cells of the PIG-A mutant clone that are presumably all descended from a single abnormal hematopoietic stem cell. The growth of the abnormal clone over time to a clinically significant size is not well documented or understood. Several lines of evidence suggest that the PIG-A mutation alone is not sufficient for this clonal expansion20,41. To study clonal expansion of the abnormal cells in PNH over time, results of serial testing by flow cytometry (with an interval of at least 1 year) in these 2 large cohorts were also analyzed.
Finally, to study the natural history of PNH and to compare the clinical course of PNH patients from the United States and Japan directly, diagnostic, complication, long-term follow-up, and survival data were analyzed. Current treatment of PNH is largely supportive, and allogeneic bone marrow transplantation (BMT) is the only available cure45. However, even HLA-matched sibling donor BMT is associated with substantial morbidity and mortality for patients with PNH34. Recent advances in the knowledge of the molecular mechanisms of PNH have raised the possibility of a gene therapy approach26,29. An understanding of the natural history of PNH is essential to develop prognostic criteria and to define indications for intensive therapeutic approaches.
Since 1966, the diagnosis of PNH was established or confirmed for over 400 patients at the Departments of Medicine and Pediatrics, Duke University Medical Center, with the use of the complement lysis sensitivity test and flow cytometry. The majority (267 patients) received some medical care at Duke; complete histories and laboratory results for 176 Duke patients, including 164 with complete flow cytometric records, were available for this analysis. To obtain follow-up data on complications and survival in some patients, questionnaires were sent to referring physicians.
In Japan, a questionnaire was completed by hematologists at 21 large medical institutions throughout the country and 2 children's hospitals that are associated with the Research Committee for the Idiopathic Hematopoietic Disorders, Ministry of Health, Labour and Welfare, Government of Japan. This committee has registered a large number of Japanese patients with PNH since 1976, and follow-up data are available on the registered patients. A questionnaire was additionally sent to 3 large hematology referral centers that are not associated with the committee. A total of 233 patients with PNH in Japan were registered. Complete histories and laboratory results for 209 Japanese patients, including 151 with complete flow cytometric records, were available for this analysis.
To guard against a possible referral bias, Japanese patients who were referred specifically to designated PNH centers (Osaka University, Kinki University, Fukushima Medical University, Kumamoto University School of Medicine, and University of Tsukuba) were defined as the Japanese PNH subset. These patients were analyzed separately from other Japanese patients in some statistical analyses. A total of 125 (59.8%) of the total 209 Japanese patients and 111 (73.5%) of the 151 patients with flow cytometric data were included in this Japanese PNH subset; almost all of these patients were referred specifically for diagnosis and treatment of PNH, similar to the majority of the patients referred to Duke University. White patients seen at Duke were also analyzed separately at Duke in some statistical analyses. A total of 153 (86.9%) of the 176 Duke patients were included in this Duke white-patient subset.
Diagnostic Criteria and Definitions
In all patients, the diagnosis of PNH was established or confirmed by at least 1 of the following accepted diagnostic tests: the Ham (acidified serum) test, sugar water (sucrose lysis) test, complement lysis sensitivity test, or flow cytometry analysis. In many cases, the diagnosis was established and confirmed by more than 1 test. Any cases with discordant results were not included in the study. A positive complement lysis sensitivity test was defined as more than 5% red blood cells (RBC) abnormally sensitive to complement-mediated lysis (PNH II and III cells)31,32. A positive flow cytometry test was defined as ≥3% GPI-deficient RBC or polymorphonuclear cells (PMN), using monoclonal antibodies to detect surface expression of CD59 or CD5510,19. The diagnoses of myelodysplastic syndrome and acute myelogenous leukemia were made by the standardized diagnostic criteria of the French American British (FAB) classification system.
Patients were classified as having anemia if the hemoglobin (Hb) concentration was <13 g/dL in male patients (<12 g/dL in females), or the RBC count was <4.0 × 1012/L in males (<3.5 × 1012/L in females). Leukopenia was defined as a white blood cell (WBC) count <4000 × 106/L, neutropenia as the absolute neutrophil count <2000 × 106/L, and thrombocytopenia as a platelet count <100 × 109/L.
Bone marrow failure was defined using the following scoring system: each cytopenia was given 1 point; an additional point was added for severity if the Hb concentration was <10 g/dL, WBC was <3000 × 106/L, or the platelet count was <60 × 109/L; 2 additional points were added for Hb concentration <6 g/dL, WBC <1000 × 106/L, or the platelet count <20 × 109/L. Patients whose total score at latest evaluation was greater than or equal to 4 points were classified with bone marrow failure, except for those patients whose score recovered 3 points or more from initial evaluation. Patients were also classified for survival analysis as severe leukopenia if WBC was <1000 × 106/L, severe neutropenia if the absolute neutrophil count was <500 × 106/L, and severe thrombocytopenia if the platelet count was <30 × 109/L.
Blood Samples and Flow Cytometric Analysis
With the informed consent of the patients, peripheral blood was obtained directly from each patient or was sent for analysis from the primary institute. Mononuclear cells and PMN were separated by sedimentation in 6% dextran and Ficoll-Hypaque (Pharmacia, Uppsala, Sweden) centrifugation as described previously10,19. RBC and fractionated PMN and mononuclear cells were stained with monoclonal antibodies, including anti-CD55 (IA10 or 1H4) and anti-CD59 (5H8 or 10G10) for the detection of GPI-APs10,19. After staining, cells were analyzed with a FacsCalibur (Becton Dickinson, Bedford, MA), FACScan (Becton Dickinson), or Ortho Cytoron Absolute (Ortho Diagnostic Systems, Raritan, NJ). Duke patients were analyzed exclusively at Duke University, and Japanese patients were analyzed mostly at Osaka University, Fukushima Medical University, Kumamoto University, or University of Tsukuba. Cell types were determined by their forward and side scatter, according to standardized guidelines10.
Data from these 2 patient groups were entered into a common database. Laboratory data were compared by Mann-Whitney test. Incidence rates were compared by chi-square test for independence. Survival rates were estimated by the Kaplan-Meier method. Survival risk factors were compared with the Kaplan-Meier cumulative survival plot by log-rank (Mantel-Cox) test. The percentages of CD59- or CD55-deficient RBC, PMN, and mononuclear cells were compared by either the Student t-test or the Welch t-test. The comparison of cells deficient for CD59 versus CD55 expression, and for RBC versus PMN fractions, was performed by the Pearson correlation coefficient. The distribution of Type II cells in patients who had smaller or larger populations of CD59(-) cells was compared by the Fisher exact probability. The correlation of distribution between Type II RBC and Type II PMN was estimated by Spearman rank correlation. The percentages of GPI-deficient cells and the changes over time with correlation to clinical characteristics were compared by Mann-Whitney test. The changes of the percentages of GPI-deficient cells between the initial analysis and the latest analysis were compared by Wilcoxon signed-rank test. For all analyses, the entire cohort of Japanese patients was compared to the Duke cohort. For critical analyses, the smaller Japanese PNH subset (n = 125) was also analyzed separately.
The 176 Duke patients, living almost exclusively in the United States, had the following ethnic characteristics: 153 (86.9%) were white, 12 (6.8%) were black, 6 (3.4%) were Hispanic, 3 (1.7%) were Arabian, and 1 (0.6%) each was Native American and Asian. The 209 Japanese patients were exclusively Asian. The median age at diagnosis of PNH was younger for all Duke patients (30 yr; range, 4-80 yr) compared to Japanese patients (45 yr; range, 10-86 yr), p < .0001 (Figure 1). The white subset (30 yr; range, 4-80 yr) also was significantly younger than the Japanese patients (p < .0001).
Clinical and Laboratory Characteristics at Diagnosis
There was a larger percentage of female patients (56.3%) in the Duke cohort than in the Japanese cohort (43.5%), p = .01 (Table 1). Although the diagnosis has been established historically mainly by a Ham test and a sugar water test, flow cytometric assays have become more popular in recent years (see Table 1). At the time of diagnosis, Duke patients had a slightly lower prevalence of previous aplastic anemia (29.0% vs. 37.8%, p = .07), but a similar prevalence of previous myelodysplastic syndrome (5.1% vs. 4.8%, p = not significant [NS]). Duke patients had a significantly higher prevalence of classical PNH symptoms at diagnosis compared to the Japanese patients, including hemoglobinuria (50.0% vs. 33.5%, p = .001), infection (13.6% vs. 3.4%, p = .0002), and thrombosis (19.3% vs. 6.2%, p < .0001) (see Table 1). In contrast, Duke patients had a significantly lower prevalence at diagnosis of bone marrow failure including less anemia (88.1% vs. 94.3%, p = .03), leukopenia/neutropenia (45.5% vs. 72.3%, p < .0001), and thrombocytopenia (52.3% vs. 63.2%, p = .03). These same differences were observed when the Japanese PNH subset (patients referred specifically to designated Japanese PNH clinics) was compared to the Duke patients; for example, 35.2% of the Japanese PNH subset had prior aplastic anemia (p = .3) and 5.6% had symptoms of thrombosis at diagnosis (p = .0006) (see Table 1).
At diagnosis, Duke patients had a higher Hb concentration (9.7 g/dL vs. 8.2 g/dL, p < .0001), higher absolute neutrophil count (3005 × 106/L vs. 1782 × 106/L, p < .0001), higher platelet count (140 × 109/L vs. 96 × 109/L, p < .0001), higher reticulocyte count (195 × 109/L vs. 78 × 109/L, p < .0001), and higher lactate dehydrogenase (2337 vs. 1572 IU/L, p = .1) than Japanese patients (see Table 1). Bone marrow analysis revealed a similar prevalence of morphologic dysplasia (14 of 90 Duke patients vs. 32 of 149 Japanese patients, p = NS). Chromosomal abnormalities were found in selected patients (6 of 25 Duke patients vs. 10 of 85 Japanese patients, p = NS). The 6 white Duke patients had the following bone marrow karyotypes: -13/-13q (1 patient), -7 (1 patient), +8 (2 patients), -18 (1 patient) and +21 (1 patient). The 10 Asian patients had the following karyotypes: -13/-13q (2 patients), -7 (1 patient) and -21 (2 patients), and miscellaneous/complex (5 patients).
Proportion of GPI-AP-Deficient Cells at the Initial Analysis
Flow cytometry analysis included 164 Duke patients with PNH who were mostly white and 151 Japanese patients who were exclusively Asian. The average fraction of cells deficient in CD59 or CD55 at the initial analysis (mean ± SE) was first compared within each patient group (Figure 2). In both patient populations, the average proportion of GPI-deficient PMN was higher than GPI-deficient RBC, which in turn was higher than GPI-deficient mononuclear cells. When the 2 cohorts were compared, Duke patients had a higher percentage of GPI-deficient cells than Japanese patients (see Figure 2). The fraction of PMN deficient in CD59 or CD55 at initial analysis was significantly higher in Duke patients than in Japanese patients (p < .0001 and p < .0001, respectively). Similarly, the fraction of CD59(-) RBC was also significantly higher in Duke patients than in Japanese patients (p = .03). These same differences were observed when the smaller Japanese PNH subset was compared to the Duke patients; for example, Duke PNH patients had 68.6 ± 3.3% CD59(-) PMN, compared to 45.3 ± 4.6% for the Japanese PNH subset, p < .0001.
Although the fraction of affected PMN was typically higher than RBC, there was significant correlation between the 2 cell types in each patient cohort. The percentages of affected RBC and PMN were correlated both in CD59(-) expression (correlation 0.52, p < .0001, n = 97 in Duke patients; correlation 0.58, p < .0001, n = 85 in Japanese patients) and CD55(-) expression (correlation 0.49, p < .0001, n = 88 in Duke patients; correlation 0.45, p = .0007, n = 52 in Japanese patients). Figure 3 illustrates the significant correlation of CD59(-) cells in each patient cohort.
Finally, the utility of CD59 vs. CD55 monoclonal antibodies for quantitation of the PNH clones was compared. The percentages of CD59(-) versus CD55(-) cells determined by flow cytometric analysis were highly correlated for both RBC (correlation 0.80, p < .0001, n = 138 in Duke patients; correlation 0.90, p < .0001, n = 65 in Japanese patients) and PMN (correlation 0.95, p < .0001, n = 93 in Duke patients; correlation 0.88, p < .0001, n = 52 in Japanese patients) (data not shown). These results indicate that monoclonal antibodies for either CD59 or CD55 accurately quantitate the affected clone of cells in PNH.
Intermediate Expression of GPI-AP
At Duke, 69 of the 164 patients with flow cytometry results (42.1%) had intermediate expression (Type II) RBC and 19 of 98 patients (18.6%) had Type II PMN, suggesting that many PNH patients have more than 1 PIG-A mutant clone (Table 2). Most patients with Type II RBC also had Type II PMN (p = .01, data not shown). In both RBC and PMN, Type II cells were more frequently detected in patients who had larger fractions of affected cells (p < .0001, and p = .03, respectively). However, relative dominance of Type II over Type III cells was not related to the fraction of affected cells (p = NS for both RBC and PMN) (see Table 2).
Correlation of Flow Cytometry With Clinical and Laboratory Characteristics
The proportions of affected cells were next compared with clinical and laboratory characteristics. In both cohorts, PNH patients who presented with classical symptoms of PNH including hemoglobinuria, infection, thrombosis, and anemia tended to have larger populations of CD59(-) cells (Figure 4 and Table 3). Conversely, patients who presented with documented aplastic anemia or initial laboratory evidence of marrow failure, including leukopenia/neutropenia or thrombocytopenia, had smaller populations of CD59(-) cells (see Figure 4 and Table 3). As shown in Figure 4, all analyses had the same tendency, although only some of the comparisons were statistically significant.
During follow-up, significantly more Duke patients developed a new thrombotic event (31.8% vs. 4.3%, p < .0001) or severe infection (18.2% vs. 9.1%, p = .009) than Japanese patients, but the proportion of patients who developed hematopoietic failure (33.0% vs. 36.4%, p = NS), myelodysplastic syndrome (3.4% vs. 3.8%, p = NS), leukemia (0.6% vs. 2.9%, p = NS), or renal failure (9.1% vs. 10.5%, p = NS) was not significantly different (Table 4). These same differences were observed when the smaller Japanese PNH subset was compared to the Duke patients; for example, only 3.2% of the Japanese PNH subset developed thrombosis (p < .0001), while 36.0% developed hematopoietic failure (p = .6) (see Table 4). The major locations of thrombosis in Duke versus Japanese patients were hepatic vein (Budd-Chiari syndrome) (8 vs. 1, respectively), portal vein (7 vs. 1), vena cava (4 vs. 0), mesenteric vein (4 vs. 0), deep vein (10 vs. 1), central nervous system vein (5 vs. 1), pulmonary vein (5 vs. 1), cerebral artery (0 vs. 2), and coronary artery (0 vs. 1). Duke patients with thrombosis (initial symptom or complication) had significantly higher proportions of GPI-deficient PMN than patients without thrombosis (81.7% ± 3.7% vs. 65.1% ± 4.5%, p < .05) (Figure 5).
During follow-up, a larger proportion of Japanese patients did not require treatment compared to Duke patients (28.2% vs. 5.7%, p < .0001) (see Table 4). In contrast, a larger proportion of Duke patients than Japanese patients received erythrocyte transfusions (61.9% vs. 40.2%, p < .0001), prednisone (82.4% vs. 46.9%, p < .0001), antithymocyte or antilymphocyte globulin (15.3% vs. 2.9%, p < .0001), anti-coagulation (26.7% vs. 4.3%, p < .0001), or BMT (8.0% vs. 1.9%, p = .005) (see Table 4). Six babies were born to 5 Duke patients, 4 of whom had thrombotic complications during pregnancy. The only Duke patient who had an uneventful pregnancy was the single patient of Asian (Vietnamese) ancestry. Fourteen babies were born to 8 Japanese patients, only 1 of whom had complications of thrombosis.
Changes in the Proportion of GPI-AP-Deficient Cells
To assess the changes in the size of the PNH clone over time, flow cytometry results were analyzed for patients who had a second testing after an interval of at least 1 year (Figure 6). Overall, the average fraction of CD59(-) RBC was relatively stable over time in both patient cohorts: 55.3% ± 4.0% initially and 58.3% ± 4.3% at follow-up for Duke patients (n = 52, p = NS); 39.6% ± 3.7% initially and 40.5% ± 4.5% at follow-up for Japanese patients (n = 56, p = NS) (see Figure 6A). The average fraction of CD59(-) PMN was also relatively stable over time in both patient cohorts: 75.2% ± 4.2% initially and 74.1% ± 4.7% at follow-up for Duke patients (n = 42, p = NS); 40.0% ± 8.3% initially and 50.7% ± 8.6% at follow-up for Japanese patients (n = 22, p = NS) (see Figure 6A).
Changes in the size of the PNH clone over time were then compared with laboratory and clinical characteristics. For both Duke and Japanese patients, the development of overall hematopoietic failure was associated with a significant diminution in the fraction of CD59(-) PMN over time (p = .04 for Duke patients, p = .05 for Japanese patients) (see Figure 6B and Table 3). Other Duke patients and Japanese patients had an increasing proportion of CD59(-) PMN, which was not associated with the development of aplasia.
Clinical Outcome and Survival Risk Factors
Survival analysis revealed a similar incidence of deaths in each cohort: 21.7% of Duke patients vs. 21.2% of Japanese patients (Table 5). However, the causes of death were different; for example, 42.1% of Duke patient deaths were from thrombosis, compared to 7.9% of Japanese patient deaths, p = .0006 (see Table 5). The locations of thrombosis in the patients who died from this complication (Duke patients vs. Japanese patients) were hepatic vein (Budd-Chiari syndrome) (7 vs. 0), portal vein (3 vs. 1), vena cava (1 vs. 0), mesenteric vein (1 vs. 0), central nervous system vein (3 vs. 1), pulmonary vein (4 vs. 0), cerebral artery (1 vs. 1), and coronary artery (1 vs. 1). Japanese patients had a longer mean survival time than Duke patients (32.1 yr vs. 19.4 yr), although the Kaplan-Meier survival curves were not significantly different (p = .7) (Figure 7).
To identify prognostic criteria that might help define indications for intensive therapeutic approaches, survival was analyzed according to several variables, including age and prior aplastic anemia or myelodysplastic syndrome. Initial symptoms were also analyzed including hemoglobinuria, mild or severe leukopenia/neutropenia, mild or severe thrombocytopenia, any combination of cytopenia, infection, or thrombosis at diagnosis as well as hematopoietic failure, thrombosis, severe infection, myelodysplastic syndrome or acute myelogenous leukemia, other malignancies, renal failure, and hepatitis (B or C) as a subsequent complication. The development of pregnancy and the use of therapy such as immunosuppression, growth factor support, or anti-coagulants were compared by log-rank (Mantel-Cox) test. As shown in Table 6, the major prognostic risk factors for survival were age over 50 years and severe leukopenia/neutropenia at diagnosis, and severe infection as a complication. It is noteworthy that thrombosis at diagnosis and as a complication were risk factors only for Duke patients, but not for Japanese patients, probably due to the low incidence in the Asian population. In contrast, renal failure as a complication was a significant risk factor for Japanese patients, and remained significant even when the smaller Japanese subset was analyzed. Pregnancy, severe thrombocytopenia at diagnosis, and malignancy as a complication did not affect survival in either group of patients, probably due to the low number of cases.
These analyses provide important diagnostic and long-term data on 2 large cohorts of PNH patients, and identify important differences between white and Asian patients with PNH. Although PNH is mainly a disease of adults, it is known that PNH can also present in childhood and adolescence43. Indeed, the median age at diagnosis was significantly younger for Duke patients (30 yr) compared to Japanese patients (45 yr). Previous reports include a French study with a median age of 33 years39 and a British cohort that averaged 42 years11. Both genders were affected in approximately equivalent numbers, although there was a slight female predominance among Duke patients (56.3%) and a slight male predominance among Japanese patients (56.5%). Previous reports from France39 and Japan7 have documented that males and females are equally affected with PNH. However, studies from other parts of Asia including Thailand and China15,16 have reported a strong male predominance (72.9% and 84.0%, respectively), perhaps explained by cultural biases, biologic differences, or unrecognized environmental influences.
In our cohorts, the diagnostic tests were quite different due to several factors. The complement lysis sensitivity test was used more frequently at Duke, whereas the Ham test, the sugar water test, and flow cytometry were more common in Japan. Duke University is one of the few institutions in the world that accurately and reproducibly performed the complement lysis sensitivity test, which was developed by Rosse and Dacie31,32. Most Japanese patients with PNH were diagnosed by the Ham test or the sugar water test, while flow cytometry was simultaneously used for additional confirmation; many PNH centers in Japan established flow cytometric testing earlier than Duke 1994. Currently the flow cytometric assay is generally accepted to be more sensitive than other classical tests to detect PNH cells, and a positive result in our analysis was defined as ≥3% GPI-deficient cells. Although many patients with aplastic anemia2,9,35,36 and even some normal individuals1 have very small (<1%) subclinical populations of circulating PNH cells, we believe that they are entirely different from clinical PNH patients. Indeed, the availability of extremely sensitive techniques, such as aerolysin-based flow cytometry4,5, can identify <0.01% GPI-deficient cell populations that have no clear clinical consequence. Unfortunately, there is currently no worldwide consensus on diagnostic criteria for PNH3. This is an urgent issue that should be addressed in the near future.
In our analysis, initial symptoms were quite different in the 2 cohorts of patients. Significantly more Duke patients had classical symptoms of PNH, including hemoglobinuria, thrombosis, and infection, while Japanese patients had more evidence of bone marrow aplasia, including anemia, leukopenia/neutropenia, and thrombocytopenia. This important observation was potentially due to a referral bias, since patients in the United States were typically referred to Duke University specifically for PNH, whereas patients from Japan were collected from a variety of large medical institutions including recognized aplastic anemia centers like Tokyo Women's Medical Collage, Kanazawa University, and NTT Kanto Hospital. However, only 12 (5.7%) of the 209 Japanese patients with epidemiologic data and only 9 (6.0%) of the 151 Japanese patients with flow cytometric data were registered from aplastic anemia centers. To exclude this potential referral bias, the Japanese cohort was analyzed both as a single group (n = 209 with epidemiologic data and n = 151 with flow cytometric data) and as a smaller PNH subset (n = 125 with epidemiologic data and n = 111 with flow cytometric data); the latter group included only patients referred to large PNH centers in Japan. Even with these additional analyses, the clinical manifestations remained significantly different between white and Asian patients with PNH, suggesting that our findings identify real differences that do not result from a referral bias.
Laboratory data at diagnosis also reflected these differences, since Duke patients had a higher average reticulocyte count and higher lactate dehydrogenase due to greater hemolysis. In contrast, Japanese patients had lower Hb concentration, absolute neutrophil count, and platelet count, reflecting the higher incidence of bone marrow aplasia. A number of viruses, including hepatitis viruses, Epstein-Barr virus, B19 parvovirus, human herpes virus-6, and human immunodeficiency virus, have been implicated in the pathogenesis of marrow failure, and some of them (or other unknown viruses) could contribute to this difference between the 2 ethnic groups.
Flow cytometric data documented significant differences between Duke patients and Japanese patients with PNH, especially a significantly higher proportion of GPI-deficient PMN in Duke patients at the initial analysis. As previously described42, a larger proportion of PMN was affected compared to RBC and mononuclear cells, probably reflecting differences in the rates of production and destruction. In both cohorts, a larger PNH clone was associated with classical PNH symptoms including hemoglobinuria, infection, thrombosis, and anemia, while a smaller PNH clone was associated with bone marrow aplasia including leukopenia/neutropenia and thrombocytopenia. Taken together, these observations indicate that Duke patients have a higher proportion of GPI-deficient clones that leads to classical PNH symptoms, which results in an earlier diagnosis, while Asian patients have a lower proportion of GPI-deficient cells that is associated with aplasia, which results in a delayed diagnosis of PNH.
Type II (intermediate sensitivity) PNH RBC were originally observed by the complement lysis sensitivity test of Rosse and Dacie31,32, and were later confirmed by flow cytometry as having an intermediate expression of GPI-APs33,37,38. Type II PMN can also be detected by flow cytometry, and it is now believed that this intermediate phenotype is due to partial inactivation of the PIG-A gene by a missense mutation. In this study, we documented an overall prevalence of the Type II phenotype in RBC (42.1%) and in PMN (18.6%), which was similar to our previous study10. In addition, the presence of Type II RBC and Type II PMN was well correlated within individual patients. These data suggest that many PNH patients have more than 1 PIG-A mutant clone, although it is technically difficult to identify the specific PIG-A mutations in patients with small clones.
Thrombosis was perhaps the most striking clinical difference between these 2 patient cohorts; Duke accounted for 46% of the total number of patients but 76% of those with thrombosis (χ2 = 41.2, p < .0001). The reasons for this difference in incidence are not entirely clear; blood pressure, total cholesterol, high density lipoprotein cholesterol, triglycerides, and fasting blood sugar at diagnosis were compared, but there were no significant abnormalities or differences between the 2 groups (data not shown). Several factors associated with thrombosis in PNH have been suggested, including activated platelets8, elevated levels of circulating procoagulant microparticles13,24, excess soluble urokinase-type plasminogen activator receptor25,30, deficient GPI expression on RBC and platelets44,46, and hemolysis23,28. In the current study, we found that Duke patients who developed thrombosis as a complication had larger populations of CD59(-) RBC and PMN (see Figure 5). If thrombosis relates to a higher number of circulating GPI-deficient cells, then Japanese patients may be relatively protected from thrombosis by their greater marrow aplasia and significantly smaller clone size. Additional genetic differences between white and Asian patients may also influence the risk of thrombosis in PNH, although the more common inherited causes of thrombophilia do not predict thrombosis in patients with PNH21,47.
Reflecting their higher prevalence of initial PNH symptoms, significantly more Duke patients also developed complications of classical symptoms of PNH during follow-up, especially thrombosis and infection. The length of follow-up was highly variable among individual patients, but the average length was similar between the 2 cohorts (91.2 mo for Duke patients vs. 101.4 mo for Japanese patients). Since the Duke patients had a slightly shorter follow-up, the observation of more new thrombotic events and severe infections is even more compelling. However, despite a lower prevalence of bone marrow aplasia at diagnosis, fully one-third of Duke patients developed hematopoietic failure during follow-up. Importantly, the number of Japanese patients who developed hematopoietic failure was not significantly higher than Duke patients. This observation suggests that marrow failure is a common terminal event in PNH, regardless of initial symptoms or age at diagnosis. The observed differences in therapy (see Table 4) primarily reflect a lack of consensus regarding treatment guidelines for PNH.
Clonal expansion is known to occur in most patients with PNH, although the pathogenesis and timing of this phenomenon is poorly understood. In our analysis of patients using serial flow cytometry, the average proportion of GPI-deficient cells was relatively stable over time, suggesting that in most patients the abnormal clone has already expanded at the time of diagnosis. For individual patients, however, the size of the GPI-deficient populations varied considerably, illustrating the complexity of clonal expansion in PNH and its marked variability. We documented that some Duke patients and Japanese patients had a diminution in their CD59(-) PMN over time, which was associated with the development of marrow aplasia. In general, a decreasing PNH clone size reflected a decline in hematopoietic capacity by the abnormal PIG-A mutant clone, which indicated impending hematopoietic failure.
In vivo, the life span of a dominant PNH clone is unknown, although additional clones may emerge to take its place22. Recently, we reanalyzed the PIG-A gene mutations in 9 patients 6-10 years after the initial analysis27. The proportions of CD59(-) peripheral blood PMN in individual patients were highly variable; the clone size increased in 2 patients, was stable in 4, and diminished in 3 patients. In all cases the previously predominant clone was still present and dominant, proving that 1 stem cell clone can sustain hematopoiesis for 6-10 years in patients with PNH. Two patients whose CD59(-) PMN decreased due to decline of the predominant PNH clone developed an aplastic condition, further supporting the hypothesis that marrow aplasia is a terminal feature of PNH. Marrow failure with aplasia likely results when the proliferative capacity of the PNH clones is exhausted in the setting of ongoing autoimmune suppression of hematopoiesis. The pathogenesis of PNH remains a mystery, however, since specific autoreactive T lymphocyte clones and target antigens have not yet been identified.
A similar number of deaths occurred in each group, but the causes of death were different. Almost half of the deaths in Duke patients were associated with thrombosis, similar to previous reports for Europeans (30%-40%)11,39, while fewer than 10% of Japanese patients died from thrombosis. The median interval from diagnosis of PNH to onset of thrombosis for 16 Duke and 3 Japanese patients who died from thrombosis was 31 months and 29 months, respectively, and most of these patients (68.4%) had a thrombotic episode within 5 years of diagnosis. More Japanese patients with PNH died from hemorrhage, evolution to myelodysplastic syndrome/acute myelogenous leukemia, or renal failure, although the numbers were small and not statistically different. These outcomes likely reflect the higher incidence of marrow aplasia and older age. The median survival time of Duke patients from diagnosis (23.3 yr) was similar to that of Japanese patients (25.0 yr), both longer than previous reports for French (14.6 yr)39, British (10.0 yr)11, Japanese (16.0 yr)7, and American pediatric (13.5 yr)43 patients with PNH.
Several significant survival risk factors were identified for both cohorts, including age over 50 years, severe leukopenia/neutropenia at diagnosis, and severe infection as a complication. Thrombosis at diagnosis or as a complication was a poor prognostic sign for Duke patients, as was the development of renal failure for Japanese patients. These findings compare with a 1996 French report39 in which the occurrence of thrombosis; evolution to pancytopenia, myelodysplastic syndrome, or leukemia; older age at diagnosis; need for additional treatment; and thrombocytopenia at diagnosis were poor survival factors. Renal failure as a complication in Japanese patients was a new finding as a survival risk factor, which may be related to the older age of this patient population. Thrombosis in Japanese patients did not significantly influence prognosis. The identification of early events as poor risk factors, namely thrombosis or severe leukopenia/neutropenia at diagnosis, can be used to design clinical trials and define indications for intensive therapies.
BMT is currently the only available cure for PNH. Syngeneic or HLA-matched related donor allogeneic BMT with eradication of the PNH clone has been successful in several patients with PNH45. At this time, however, even HLA-matched sibling donor BMT is associated with substantial morbidity and mortality34. Immunoablative high-dose cyclophosphamide without BMT might be another option, although long-term results suggest the PNH clone may not be fully eradicated6. The discovery of PIG-A gene mutations in PNH allows a potential new approach for patients with PNH, namely gene therapy to restore PIG-A gene function. In vitro experiments are encouraging26,29, although further investigation will be necessary before in vivo experimentation. Targeted therapy against hemolysis and thrombosis is currently lacking, but new agents may be forthcoming in the near future12. Since thrombosis and severe leukopenia/neutropenia are significant survival risk factors at diagnosis, we recommend that these patients be considered for early and aggressive therapies. Potentially curative interventions such as HLA-matched sibling BMT should be provided for these high-risk patients.
We thank Toshiyuki Hirota, Yuki Murakami, Yvonne Ellis, Kathleen Greenwell, Keiko Kinoshita, and Keiko Yamamoto for their excellent technical assistance. We also thank Drs. T. Uchiyama, K. Tohyama, A. Urabe, K. Ozawa, M. Tomonaga, H. Hirai, K. Mitani, T. Murate, Y. Niho, T. Ueda, R. Ohno, K. Ohyashiki, M. Karasawa, A. Kimura, M. Teramura, Y. Niitsu, M. Harada, M. Bessyo, T. Hotta, and S. Nakao for providing patient data. Rhonda Laney provided clinical care for many of the Duke patients. Dr. Sharyne Donfield provided useful epidemiologic discussions.
1. Araten DJ, Nafa K, Pakdeesuwan K, Luzzatto L. Clonal populations of hematopoietic cells with paroxysmal nocturnal hemoglobinuria genotype and phenotype are present in normal individuals. Proc Natl Acad Sci U S A
2. Azenishi Y, Ueda E, Machii T, Nishimura J, Hirota T, Shibano M, Nakao S, Kinoshita T, Mizoguchi H, Kitani T. CD59-deficient blood cells and PIG-A gene abnormalities in Japanese patients with aplastic anaemia. Br J Haematol
3. Borowitz MJ. Flow cytometry testing in PNH. How much is enough? Cytometry
4. Brodsky RA, Mukhina GL, Nelson KL, Lawrence TS, Jones RJ, Buckley JT. Resistance of paroxysmal nocturnal hemoglobinuria cells to the glycosylphosphatidylinositol-binding toxin aerolysin. Blood
5. Brodsky RA, Mukhina GL, Li S, Nelson KL, Chiurazzi PL, Buckley JT, Borowitz MJ. Improved detection and characterization of paroxysmal nocturnal hemoglobinuria using fluorescent aerolysin. Am J Clin Pathol
6. Brodsky RA, Sensenbrenner LL, Jones RJ. Complete remission in severe aplastic anemia after high-dose cyclophosphamide without bone marrow transplantation. Blood
7. Fujioka S, Asai T. Prognostic features of paroxysmal nocturnal hemoglobinuria in Japan. Nippon Ketsueki Gakkai Zasshi
8. Gralnick HR, Vail M, McKeown LP, Merryman P, Wilson O, Chu I, Kimball J. Activated platelets in paroxysmal nocturnal hemoglobinuria. Br J Haematol
9. Griscelli-Bennaceur A, Gluckman E, Scrobohaci ML, Jonveaux P, Vu T, Bazarbachi A, Carosella ED, Sigaux F, Socie G. Aplastic anemia and paroxysmal nocturnal hemoglobinuria: Search for a pathogenetic link. Blood
10. Hall SE, Rosse WF. The use of monoclonal antibodies and flow cytometry in the diagnosis of paroxysmal nocturnal hemoglobinuria. Blood
11. Hillmen P, Lewis SM, Bessler M, Luzzatto L, Dacie JV. Natural history of paroxysmal nocturnal hemoglobinuria. N Engl J Med
12. Hillmen P, Hall C, Marsh JC, Elebute M, Bombara MP, Petro BE, Cullen MJ, Richard SJ, Rollins SA, Mojcik CF, Rother RP. Effect of eculizumab on hemolysis and transfusion requirements in patients with paroxysmal noctural hemoglobinuria. N Engl J Med
13. Hugel B, Socie G, Vu T, Toti F, Gluckman E, Freyssinet JM, Scrobohaci ML. Elevated levels of circulating procoagulant microparticles in patients with paroxysmal nocturnal hemoglobinuria and aplastic anemia. Blood
14. Kinoshita T, Inoue N, Takeda J. Defective glycosyl phosphatidylinositol anchor synthesis and paroxysmal nocturnal hemoglobinuria. Adv Immunol
15. Kruatrachue M, Wasi P, Na-Nakorn S. Paroxysmal nocturnal hemoglobinuria in Thailand with special reference to an association with aplastic anemia. Br J Haematol
16. Le X, Yang T, Yang X, Wang X. Characteristics of paroxysmal nocturnal hemoglobinuria in China. Chin Med J
17. Luzzatto L, Bessler M, Rotoli B. Somatic mutations in paroxysmal nocturnal hemoglobinuria: A blessing in disguise? Cell
18. Miyata T, Takeda J, Iida Y, Yamada N, Inoue N, Takahashi M, Maeda K, Kitani T, Kinoshita T. Cloning of PIG-A, a component in the early step of GPI-anchor biosynthesis. Science
19. Miyata T, Yamada N, Iida Y, Nishimura J, Takeda J, Kitani T, Kinoshita T. Abnormalities of PIG-A transcripts in granulocytes from patients with paroxysmal nocturnal hemoglobinuria. N Engl J Med
20. Murakami Y, Kinoshita T, Maeda Y, Nakano T, Kosaka H, Takeda J. Different roles of glycosylphosphatidylinositol in various hematopoietic cells as revealed by a mouse model of paroxysmal nocturnal hemoglobinuria. Blood
21. Nafa K, Bessler M, Mason P, Vulliamy T, Hillmen P, Castro-Malaspina H, Luzzatto L. Factor V Leiden mutation investigated by amplification created restriction enzyme site (ACRES) in PNH patients with and without thrombosis. Haematologica
22. Nafa K, Bessler M, Deeg HJ, Luzzatto L. New somatic mutation in the PIG-A gene emerges at relapse of paroxysmal nocturnal hemoglobinuria. Blood
23. Newcomb TF, Gardner FH. Thrombin generation in PNH. Br J Haematol
24. Ninomiya H, Kawashima Y, Hasegawa Y, Nagasawa T. Complement-induced procoagulant alteration of red blood cell membranes with microvesicle formation in paroxysmal nocturnal hemoglobinuria (PNH): Implications for thrombogenesis in PNH. Br J Haematol
25. Ninomiya H, Hasegawa Y, Nagasawa T, Abe T. Excess soluble urokinase-type plasminogen activator receptor in the plasma of patients with paroxysmal nocturnal hemoglobinuria inhibits cell-associated fibrinolytic activity. Int J Hematol
26. Nishimura J, Phillips KL, Ware RE, Hall S, Wilson L, Gentry TL, Howard TA, Murakami Y, Shibano M, Machii T, Gilboa E, Kanakura Y, Takeda J, Kinoshita T, Rosse WF, Smith CA. Efficient retrovirus-mediated PIG-A gene transfer and stable restoration of GPI-anchored protein expression in cells with the PNH phenotype. Blood
27. Nishimura J, Hirota T, Kanakura Y, Machii T, Kageyama T, Doi S, Wada H, Masaoka T, Kanayama Y, Fujii H, Inoue N, Kuwayama M, Inoue N, Ohishi K, Kinoshita T. Long-term support of hematopoiesis by a single stem cell clone in patients with paroxysmal nocturnal hemoglobinuria. Blood
28. Peytremann R, Rhodes RS, Hartmann RC. Thrombosis in paroxysmal nocturnal hemoglobinuria (PNH) with particular reference to progressive, diffuse hepatic venous thrombosis. Ser Haematol
29. Robert D, Mahon FX, Richard E, Etienne G, de Verneuil H, Moreau-Gaudry F. A SIN lentiviral vector containing piga cdna allows long-term phenotypic correction of CD34(+)-derived cells from patients with paroxysmal nocturnal hemoglobinuria. Mol Ther
30. Ronne E, Pappot H, Grondahl-Hansen J, Hoyer-Hansen G, Plesner T, Hansen NE, Dano K. The receptor for urokinase plasminogen activator is present in plasma from healthy donors and elevated in patients with paroxysmal nocturnal hemoglobinuria. Br J Haematol
31. Rosse WF, Dacie JV. Immune lysis of normal human and paroxysmal nocturnal hemoglobinuria (PNH) red blood cells. I. The sensitivity of PNH red cells to lysis by complement and specific antibody. J Clin Invest
32. Rosse WF, Dacie JV. Immune lysis of normal human and paroxysmal nocturnal hemoglobinuria (PNH) red blood cells. II. The role of complement components in the increased sensitivity of PNH red cells to immune lysis. J Clin Invest
33. Rosse WF, Hoffman S, Campbell M, Borowitz M, Moore JO, Parker CJ. The erythrocytes in paroxysmal nocturnal haemoglobinuria of intermediate sensitivity to complement lysis. Br J Haematol
34. Saso R, Marsh J, Cevreska L, Szer J, Gale RP, Rowlings PA, Passweg JR, Nugent ML, Luzzatto L, Horowitz MM, Gordon-Smith EC. Bone marrow transplants for paroxysmal nocturnal haemoglobinuria. Br J Haematol
35. Schrezenmeier H, Hertenstein B, Wagner B, Raghavachar A, Heimpel H. A pathogenetic link between aplastic anemia and paroxysmal nocturnal hemoglobinuria is suggested by a high frequency of aplastic anemia patients with a deficiency of phosphatidylinositol glycan anchored proteins. Exp Hematol
36. Schubert J, Vogt HG, Zielinska-Skowronek M, Freund M, Kaltwasser JP, Hoelzer D, Schmidt RE. Development of the glycosylphosphatitylinositol-anchoring defect characteristic for paroxysmal nocturnal hemoglobinuria in patients with aplastic anemia. Blood
37. Shichishima T, Terasawa T, Hashimoto C, Ohto H, Uchida T, Maruyama Y. Heterogenous expression of decay accelerating factor and CD59/membrane attack complex inhibition factor and CD59/membrane attack complex inhibition factor on paroxysmal noctural haemoglobinuria (PNH) erythrocytes. Br J Haematol
38. Shichishima T, Terasawa T, Saitoh Y, Hashimoto C, Ohto H, Maruyama Y. Diagnosis of paroxysmal nocturnal haemoglobinuria by phenotypic analysis of erythrocytes using two-colour flow cytometry with monoclonal antibodies to DAF and CD59/MACIF. Br J Haematol
39. Socie G, Mary JY, de Gramont A, Rio B, Leporrier M, Rose C, Heudier P, Rochant H, Cahn JY, Gluckman E. Paroxysmal nocturnal haemoglobinuria: long-term follow-up and prognostic factors. Lancet
40. Takeda J, Miyata T, Kawagoe K, Iida Y, Endo Y, Fujita T, Takahashi M, Kitani T, Kinoshita T. Deficiency of the GPI anchor caused by a somatic mutation of the PIG-A gene in paroxysmal nocturnal hemoglobinuria. Cell
41. Tremml G, Dominguez C, Rosti V, Zhang Z, Pandolfi PP, Keller P, Bessler M. Increased sensitivity to complement and a decreased red blood cell life span in mice mosaic for a nonfunctional Piga gene. Blood
42. Tseng JE, Hall SE, Howard TA, Ware RE. Phenotypic and functional analysis of lymphocytes in paroxysmal nocturnal hemoglobinuria. Am J Hematol
43. Ware RE, Hall SE, Rosse WF. Paroxysmal nocturnal hemoglobinuria with onset in childhood and adolescence. N Engl J Med
44. Wiedmer T, Hall SE, Ortel TL, Kane WH, Rosse WF, Sims PJ. Complement-induced vesiculation and exposure of membrane prothrombinase sites in platelets of paroxysmal nocturnal hemoglobinuria. Blood
45. Woodard P, Wang W, Pitts N, Benaim E, Horwitz E, Cunningham J, Bowman L. Successful unrelated donor bone marrow transplantation for paroxysmal nocturnal hemoglobinuria. Bone Marrow Transplant
46. Yamashina M, Ueda E, Kinoshita T, Takami T, Ojima A, Ono H, Tanaka H, Kondo N, Orii T, Okada N, Okada H, Kitani T. Inherited complete deficiency of 20-kilodalton homologous restriction factor (CD59) as a cause of paroxysmal nocturnal hemoglobinuria. N Engl J Med
© 2004 Lippincott Williams & Wilkins, Inc.
47. Zimmerman SA, Howard TA, Whorton MR, Rosse WF, Ware RE. Analysis of genetic thrombophilic mutations in patients with paroxysmal nocturnal hemoglobinuria and thrombosis. Blood