Shinji Nakao, Cellular Transplantation Biology, Division of Cancer Medicine, Kanazawa University Graduate School of Medical Science, 13-1 Takaramachi, Kanazawa, Ishikawa 920-8641, Japan. E-mail: email@example.com
Peripheral blood from 489 recently diagnosed patients with aplastic anaemia (AA) and 316 with refractory anaemia (RA) of myelodysplastic syndrome was evaluated to characterize CD55−CD59− [paroxysmal nocturnal haemoglobinuria (PNH)]-type blood cells associated with bone marrow (BM) failure. PNH-type cells were detected in 57% and 20% of patients with AA and RA, respectively. The percentages of PNH-type granulocytes ranged from 0·003% to 94·2% and the distribution was log-normal with a median of 0·178%. Serial analyses of 75 patients with PNH-type cells over 5 years revealed that the percentage of PNH-type cells constantly increased in 13 (17%), persisted in 44 (59%), disappeared in the remaining 18 (24%) although even in the ‘Disappearance’ group, PNH-type granulocytes persisted for at least 6 months. A scattergram profile of PNH-type cells unique to each patient persisted regardless of the response to immunosuppressive therapy and only single PIGA mutations were detected in PNH-type granulocytes sorted from four patients. These findings suggest that the PNH-type cells in patients with BM failure are derived from single PIGA mutant haematopoietic stem cells even when their percentages are <1% and their fate depends on the proliferation and self-maintenance properties of the individual PIGA mutants.
Classic paroxysmal nocturnal haemoglobinuria (PNH) is an acquired disease characterized by intravascular haemolysis that results from the clonal expansion of the phosphatidylinositol glycan complementation class A gene (PIGA) mutant haematopoietic stem cells (HSCs) producing blood cells deficient in glycosylphosphatidylinositol-anchored proteins (GPI-APs), such as CD55 (decay accelerating factor) and CD59 (membrane inhibitor of reactive lysis) (Lewis & Dacie, 1967; Dacie & Lewis, 1972; Takeda et al, 1993; Bessler et al, 1994; Hillmen et al, 1995). Such PNH-type cells are often detectable in the peripheral blood (PB) of patients with bone marrow (BM) failure syndromes including aplastic anaemia (AA) and refractory anaemia (RA) of myelodysplastic syndrome (MDS), as defined by the French–American–British (FAB) group (Schubert et al, 1994; Griscelli-Bennaceur et al, 1995; Schrezenmeier et al, 1995; Dunn et al, 1999; Wang et al, 2002). Given that many of these patients do not experience any obvious intravascular haemolysis, the BM failure is defined as subclinical PNH (PNH-sc) (Parker et al, 2005).
How PNH-type cells arise and increase in patients with BM failure remains obscure. The PIGA mutation itself does not confer a proliferative advantage to HSCs (Rosti et al, 1997; Ware et al, 1998). The most widely accepted mechanism for clonal expansion of PNH-type cells in patients with BM failure is the ‘escape hypothesis’, which states that the relative number of PIGA mutant HSCs increases by avoiding immunological attacks by T cells or Natural Killer cells (Luzzatto et al, 1997; Young & Maciejewski, 2000). A murine study demonstrated that GPI-AP-deficient haematopoietic cells evaded a T-cell attack due to either the absence of the target peptide recognized by GPI-AP specific T cells or absence of the accessory GPI-AP molecules that are required for HSCs to be attacked by cytotoxic T cells (Murakami et al, 2002). The higher likelihood of responding to immunosuppressive therapy (IST) in AA patients bearing PNH-type cells (PNH+ AA patients) than in AA patients not bearing PNH-type cells (PNH− AA patients) (Dunn et al, 1999; Sugimori et al, 2006) and the high frequency of a particular HLA-DR allele (HLA-DR15) in patients with classic PNH or PNH+ AA (Maciejewski et al, 2001a) support the immune selection theory that states that PIGA mutant HSCs undergo predominant proliferation over normal HSCs. However, the precise mechanisms responsible for the expansion of PIGA mutant HSCs associated with BM failure are completely unknown due to the lack of animal models for BM failure showing increased PNH-type cells. If the escape hypothesis is tenable, one would expect that PNH-type cells in BM failure patients would diminish somewhat after successful IST due to an increase in non-PNH-type cells. However, previous studies of small numbers of PNH+ patients showed that the percentage of PNH-type cells remains stable in most patients responding to IST (Maciejewski et al, 2001b; Araten et al, 2002; Sugimori et al, 2006), suggesting that PNH-type and non-PNH-type HSCs are equally susceptible to an immune system attack once BM failure develops, which is a finding contradictory to the escape theory.
Paroxysmal nocturnal haemoglobinuria-type cells detected in PNH and AA are often oligoclonal (Nishimura et al, 1997; Mortazavi et al, 2003). In view of the expected incidence of PIGA gene abnormalities (10−7 to 10−5) (Araten et al, 2005), the possibility that all the multiple PNH clones in a patient originate from HSCs is therefore highly unlikely (Traulsen et al, 2007). A recent study showed that PNH-type cells in AA patients were polyclonal (Okamoto et al, 2006) just like the CD55−CD59− blood cells detectable in healthy individuals (Araten et al, 1999; Ware et al, 2001; Hu et al, 2005), thus suggesting their origin to be haematopoietic progenitor cells (HPCs) rather than HSCs. However, granulocytes originating from HPCs last only for about 120 d (Dingli et al, 2007). The consistency of the percentage of PNH-type cells in AA patients over many years, as documented by previous studies (Maciejewski et al, 2001b; Araten et al, 2002; Sugimori et al, 2006), therefore contradicts the notion that the PNH-type cells are derived from HPCs. It thus remains unclear whether the PNH-type cells detected in BM failure patients are derived from PIGA mutant HSCs or HPCs.
The presence of PNH-type cells in patients with BM failure may predict the development of haemolytic PNH. Indeed, 10–25% of all patients with AA progress to haemolytic PNH after successful IST (Tichelli et al, 1994; Frickhofen et al, 2003). However, because only a few studies have so far examined the changes in PNH-type cell percentages over long periods, precisely how often PNH+ patients, particularly those bearing a minor (<1%) population of PNH-type cells, undergo a progression to haemolytic PNH thus remains unknown. A recent study suggested HMGA2 gene abnormalities to play a role in the acquisition of proliferative advantage by PIGA mutant HSC clones of two patients with classical PNH (Inoue et al, 2006). However, it is unclear whether HMGA2 gene abnormalities are also involved in the development of classic PNH from PNH-sc.
Hence, the origin and fate of PNH-type cells as well as the mechanisms responsible for their emergence and expansion in patients with BM failure all remain unclear. A detailed follow-up of PNH-type cells over many years may therefore be useful for clarifying these issues. We analysed PNH-type granulocytes and PNH-type erythrocytes from a large number of BM failure patients, and particularly focussed on the change in the PNH-type cell percentage of 489 recently-diagnosed patients. The results of these analyses on the changes in the percentage of PNH-type cells over time as well as on clonality provided some insight into the origin and mechanisms responsible for the clonal expansion of PNH-type cells.
Materials and methods
Paroxysmal nocturnal haemoglobinuria-type cells were analysed in PB samples from 1575 Japanese patients with various haematological diseases [classic PNH, n = 41; AA, n = 749; RA, n = 514; refractory anaemia with excess blasts (RAEB), n = 86; leukaemia, n = 47], other haematological diseases, such as lymphoma and myeloma (n = 138), and collagen vascular diseases, such as rheumatoid arthritis, systemic scleroderma and systemic lupus erythematosus (n = 81). The patients were diagnosed at Kanazawa University Hospital; hospitals participating in a cooperative study led by the Intractable Disease Study Group of Japan and other referring institutions between April 1999 and May 2007. MDS was defined according to the FAB classification (Bennett et al, 1976). The PB of patients with BM failure was routinely examined for the presence of PNH-type cells to rule out PNH. For patients with diseases other than BM failure, residual blood after performing a complete blood count was used for the flow cytometry studies. All patients provided their oral or written informed consent to proceed with the analysis of PNH-type cells. The ethical committee of Kanazawa University Graduate School of Medical Science approved the study.
Horse antithymocyte globulin (ATG, Lymphoglobulin; Genzyme, Cambridge, MA, USA, 15 mg per kg per d, 5 d) and ciclosporine (CsA; Novartis, Basel, Switzerland, 6 mg per kg per d) were used and response to IST was assessed according to the response criteria described by Camitta (2000) as described previously (Sugimori et al, 2006).
Detection of PNH-type cells: high-resolution 2-colour flow cytometry
We analysed the granulocytes and erythrocytes from the patients using a high-resolution 2-colour flow cytometry as described (Sugimori et al, 2006). Briefly, this assay includes fluorescein isothiocyanate (FITC)-conjugated anti CD55 (anti CD55-FITC; clone IA10, mouse IgG2a; Pharmingen, San Diego, CA, USA) and anti CD59-FITC (clone p282, mouse IgG2a; Pharmingen) antibodies (Abs) combined with phycoerythrin (PE)-labelled anti-lineage marker Ab that increases the specificity of detecting small populations of PNH-type cells. Anti-CD11b-PE Ab (mAbs; Becton Dickinson, Franklin Lakes, NJ, USA) and anti-glycophorin-A (GP-A)-PE Ab (clone JC159; Daco, Glostrup, Denmark) served as markers of granulocytes and erythrocytes, respectively. The specificity of flow cytometry was increased as follows; blood samples were stored at 4°C after sampling and analysed within 24 h because CD11b expression levels declined thereafter (C. Sugimori and S. Nakao, unpublished data). Three-step gating excluded the debris and immature granulocytes that are frequently found in samples from patients with leukaemia or MDS. Step 1 involved gating the SSChighFSChigh granulocyte population from FSC-SSC scattergrams (R1), which eliminates small immature cells often associated with MDS. Step 2 involved gating the CD11bbright granulocyte population on the CD11b-SSC scattergram and the GP-Abright erythrocyte population on the GP-A-FSC scattergram. This gating excludes the lineage markerdim cells that are features of old samples or immature cells. Step 3 was gating R1 × R2 and analysing 106 cells on R1 × R2 scattergrams. The presence of ≥0·003% CD55−CD59−CD11b+ granulocytes and of ≥0·005% CD55−CD59−GP-A+ erythrocytes was defined as an abnormal increase (positive) based on the results from 183 healthy individuals (Sugimori et al, 2007). When PNH-type cells were positive only in either granulocytes or erythrocytes, additional samples were tested and patients were judged to be PNH+ when the results of the first and second samplings were identical.
PIGA gene analysis
Peripheral blood mononuclear cells were depleted of CD3+ cells using MACS CD3 Microbeads (Miltenyi Biotec, Auburn, CA, USA). CD55−CD59−CD11b+ and CD55+CD59+CD11b+ granulocytes were separated using a cell sorter (JSAN; Bay Bioscience Co. Ltd., Yokohama, Japan). Over 95% of the sorted cells were CD55−CD59−CD11b+. The coding regions of PIGA were amplified by semi-nested polymerase chain reaction (PCR) or nested PCR from DNA extracted from the sorted PNH-type cells using 12 primer sets and competent Escherichia coli JM109 cells (Nippon Gene, Tokyo, Japan) transformed using six ligation reactions. Five clones were randomly selected from each group of transfectants and sequenced using BigDye Terminator v3.1 Cycle Sequencing kits (Applied Biosystems, Foster City, CA, USA) and an ABI PRISM 3100 Genetic Analyzer (Applied Biosystems) (Mochizuki et al, 2008).
All data were statistically analysed using the jmp version 6.0.3 software programme (SAS Institute, Cary, NC, USA). Whether the log-normal distribution model fit the actual PNH-type cell distribution was verified using the Kolmogorov D test. The direction and strength of the relationship between PNH-type granulocytes and PNH-type erythrocytes were tested using Spearman’s rank correction. The frequency of PNH-type cells was compared in older and younger patients using Fisher’s exact test.
Detection of PNH-type cells in various diseases
Peripheral blood samples were analysed from 1575 patients with various haematological diseases and from 81 patients with collagen vascular disease using flow cytometry. In addition to classic PNH, the number of PNH-type cells significantly increased in 49% (366/749) of the patients with AA and in 17% (87/514) of those with RA, but not in the 86 patients with RAEB, 47 with leukaemia, including 21 with acute myeloid leukaemia (AML), 47 with other haematological diseases including lymphoma and multiple myeloma and 81 with collagen vascular diseases.
Incidence and degree of significant increases in PNH-type cells at diagnosis of BM failure
The percentage of PNH-type cells can change, or the cells can disappear or emerge in during the clinical course. To clarify the incidence and degree of the significant increase in PNH-type cells at diagnosis, the data were analysed from 489 patients who had recently been diagnosed with AA, from 316 patients with RA and from 84 with other types of haematopoietic dyscrasia. The interval between the date of the first analysis and the date of diagnosis was <1 year (median, 15 d; range, −27 to 365). The percentages of PNH-type cells were significantly increased in 57% (278/489) of patients with AA and in 20% (64/316) of patients with RA (Fig 1A). The percentages of PNH-type cells were <1% in about 80% of PNH+ patients. The distribution of the PNH-type granulocyte percentages in PNH+ patients was log-normal (P <0·001; Fig 1A, B). The medians of PNH-type granulocytes and PNH-type erythrocytes in PNH+ patients were 0·178% (range, 0·003–94·2%) and 0·065% (range, 0·005–28·8%). The ratios of PNH-type granulocytes and PNH-type erythrocytes in individual patients correlated (r = 0·8357, ρ = 0·5298, P <0·001; Fig 1B), thus suggesting that these cells were derived from common precursors. The frequency of detecting an increase in PNH-type cells was significantly higher in patients aged ≥21 years than in patients <20 years (59% and 36%, P =0·004).
Fate of PNH-type cells in BM failure patients
To clarify the kinetics of PNH-type cells detected in patients with BM failure, the time course of PNH-type granulocyte percentages was retrospectively analysed in 75 PNH+ patients (72 with AA and three with RA) that met the following three conditions; (i) the first test of PNH-type cells was done within 2000 d after the diagnosis, (ii) the percentage of PNH-type cells in the first test was <10% and (iii) PNH-type cells were serially followed over 1000 d. Within 1000 d, PNH-type cells had either disappeared or expanded to ≥10% in 13 patients. The remaining 62 patients were followed up for a median of 1832 d (range, 1000–3179 d).
Transitional profiles of PNH-type cell percentages could be classified into three groups termed ‘Expansion’, ‘Disappearance’ and ‘Persistent’ according to the terminology of previous report (Schrezenmeier et al, 2000). ‘Expansion’ was defined as ≥10% PNH-type granulocytes and/or PNH-type erythrocytes at the time of the last examination and 13 (17%) met this definition. The rate of the increase in the PNH-type granulocytes in the ‘Expansion’ group remained constant throughout the observation period from the time of diagnosis (Fig 2A). Representative scattergrams of an ‘Expansion’ patient is shown in Fig 2Ai. Eight of the 13 patients progressed to classic PNH which was characterized by signs of intravascular haemolysis, such as haemoglobinuria and LDH levels that increased to >1000 i/u per ml. The expansion rates did not apparently differ between these patients and those who did not progress into classic PNH. These data suggest that the secondary genetic changes in PIGA mutant HSCs, which alter the expansion rates of PNH-type cells, did not occur in these patients during the observation period.
Paroxysmal nocturnal haemoglobinuria-type cells disappeared in 18 (24%) patients that were classified as the ‘Disappearance’ group (Fig 2B). Although the percentages of PNH-type granulocytes in most patients of this group were <1% at the time of diagnosis, their PNH-type granulocytes lasted at least 6 months in all patients and it took more than 3 years for the PNH-type granulocytes in 50% (nine) of the 18 patients to become undetectable. PNH-type granulocytes of this group tended to disappear sooner than PNH-type erythrocytes (Fig 2Bi and ii).
Forty-four (59%) patients who did not meet the definition of either of the other two categories were classified into the ‘Persistent’ group. The transitional curves in this group varied and assumed forms, such as a gentle increase, flat and a gentle decline (Fig 2Ci, ii and iii respectively), and the maximal change in the proportion of PNH-type granulocytes from the first examination was <10% during the observation period. In 80% (35) of the 44 patients, the percentages of PNH-type granulocytes were <1% at the time of diagnosis. The most common type of curve among patients who were followed up for >3 years was parabolic (Fig 2Civ). None of the transitional curves abruptly changed at any point during the follow-up. Notably, the scattergram of each PNH+ patient that was characterized by the ratio (%) of the type II to type III cells, while that of PNH-type granulocytes to PNH-type erythrocytes was unique and remained essentially unchanged over long periods (Fig 2Ci, ii and iii respectively).
Influence of IST on the fate of PNH-type cells
Forty of the 75 patients were treated with ATG and ciclosporine and 35 (88%) of them responded. The respective proportions of the ‘Expansion’, ‘Disappearance’ and ‘Persistent’ group were 17%, 29% and 54% in 35 responders; 0%, 20% and 80% in five non-responders; and 20%, 20% and 60% in 35 untreated patients. The response rate in each patient group was 100% for the ‘Expansion’, 91% for the ‘Disappearance’ and 83% for the ‘Persistent’ group, suggesting no influence of IST on the fate of PNH-type cells in BM failure patients.
Emergence of PNH-type cells in PNH− patients
To clarify whether PNH-type cells can emerge in the course of follow-up of PNH− patients, 114 PNH− patients (101 with AA and 13 with RA) that met the following three conditions were also analysed; (i) the first test of PNH-type cells was done within 2000 d after the diagnosis, (ii) the percentage of PNH-type cells in the first test was negative and (iii) PNH-type cells were serially followed over 1000 d. The median follow-up period was 1240 d (range, 1010–3101 d). Over time, PNH-type cells became detectable in five patients, who were defined as ‘Newly developed’ (Fig 2D). The absence of PNH-type cells before PNH+AA developed in one patient (Fig 2Di) was ascertained: the patient had undergone allogeneic PB stem cell transplantation to treat severe PNH− AA 3 years before developing late graft failure and PNH+ AA affecting donor-derived stem cells was diagnosed because all haematopoietic cells proved to be donor-type. It was the first time that this patient had PNH-type cells detected that were donor-derived. The patient’s donor has been haematologically normal and negative for PNH-type cells. The percentage of the donor-derived PNH-type cells reached 0·147% within several months of diagnosis and gently declined over 2 years (Mochizuki et al, 2008).
Figure 2E summarizes the transitional profiles of PNH-type cells in PNH+ and PNH− patients.
PIGA mutations in PNH-type granulocytes among patients with BM failure
To verify the hypothesis that PNH-type cells detectable in patients with BM failure are derived from one or a few primitive HSCs, the PIGA gene of PNH-type granulocytes sorted from representative patients in the four different groups designated (1) to (4) in Fig 2A–D was analysed. Two samples were obtained from each patient at intervals of at least 6 months and Table I shows the results of the PIGA gene analysis. Only single mutations were detected in the first samples from all four patients at 3/5 to 5/5 frequencies and the same mutations persisted at similar frequencies in the second set of samples, except for Patient 2 in whom the mutation detected in the first sample was not revealed in the second sample. The percentage of PNH-type granulocytes declined to 0·001% when the second sample was obtained.
Table I. PIGA gene mutations of PNH-type granulocytes sorted from four patients.
Age at first analysis (years)
Day after diagnosis (d)
Proportion of PNH-type cells (%)
Frequency of mutation
Transitional curves of patients are shown in Fig 3A–D.
*From the day of late graft failure diagnosis.
276 bp (exon 2)
505 bp (exon 2)
G to A
Gln to His
Splice site (intron 1)
G to A
593 bp (exon 2)
The present study of a large cohort of patients with BM failure confirmed the preliminary findings that an increase in the percentage of PNH-type cells is specific to AA and RA as defined by the FAB criteria (Wang et al, 2002). The absence of the increased PNH-type cells in patients with advanced MDS or AML supports the hypothesis that PNH-type cells represent a benign type of BM failure. The increase in the percentage of PNH-type granulocytes was a specific phenomenon to BM failure because it was undetectable in patients with autoimmune diseases that were not associated with BM.
The median percentage of the increased PNH-type granulocytes in PNH+ AA patients at diagnosis was 0·178%, which is 2-log more than the median percentage of PNH-type granulocytes (0·002%) detected in healthy people by Araten et al (1999) There may be some concern that this number is too low to be considered a significant increase in the percentage of PNH-type cells. However, our high resolution flow cytometry assay revealed discrete scattergram profiles of granulocytes and erythrocytes in individual patients with PNH-type cells as shown in Fig 2A–C. These profiles persisted for >3 years in most patients regardless of changes in the percentage of PNH-type cells even if the percentages at the time of diagnosis were <1%. Moreover, the correlation between PNH-type granulocytes and erythrocytes was positive in individual patients, including those with <1% PNH-type cells (Fig 1B). These findings indicate that our flow cytometry assay quantifies PNH-type cells accurately at percentages of <1%, which has often been used as a cut-off value in previous studies. Because the PNH-type granulocyte percentage in patients with BM failure shows a log-normal distribution with a median of 0·178%, it therefore seems irrational to establish a cut-off point for a significant increase in the percentage of PNH-type cells at either 1% or 0·1%.
The persistence of a scattergram profile unique to individual patients over long periods of time, even in the ‘Disappearance’ group indicates that small populations of PNH-type cells are derived from one (or a few) long lived PIGA mutated HSCs rather than from PIGA mutant HPCs that are short lived. This speculation was substantiated by PIGA gene analyses of four patients that revealed only single mutations in sorted PNH-type granulocytes obtained at various time points (Table I). These results were in contrast to previous reports in which multiple PIGA mutations were detected in PNH-type granulocytes isolated from AA patients (Mortazavi et al, 2003; Okamoto et al, 2006). Although the exact reason for these conflicting results is unclear, multiple PIGA mutant clones detected by their subcloning method might have picked up PIGA mutants, which are thought to be derived from HPCs. Alternatively, our subcloning method may only have detected a predominant clone among several different PIGA mutants because a previous study showed that two or more PIGA mutant clones are detectable in about 15% of classic PNH, usually consisting of one predominant clone and the other minor clones (Nishimura et al, 1999).
The transitional profiles of PNH-type cell percentages present the key clue to understanding the mechanisms of PNH clone expansion. Regardless of the transitional pattern, the curves were essentially linear or parabolic on the semi-logarithmic scale as shown in Fig 2A–C and it looked as though all the transitional profiles started from halfway down the individual curve despite the fact that most patients were examined early after the diagnosis, thus suggesting that the mechanism of the first expansion of PNH-type cells at the initial development of BM failure (first phase), differs from that of PNH-type cell expansion after the development of BM failure (second phase). Figure 3 represents a pattern diagram of the biphasic transitional profiles of PNH-type cells.
The first phase, which occurs before the diagnosis of BM failure, is presumed to represent rapid functional commitment from static to active status of a PIGA mutant HSC for haematopoiesis. In most PNH+ patients, PNH-type cells were detectable from the time of diagnosis, consistent with the preliminary results (Sugimori et al, 2006) and those of others based on much less sensitive flow cytometry (Maciejewski et al, 2001b; Araten et al, 2002); these cells were rarely detectable in the late phase of BM failure in PNH− patients and such cases were deemed to be exceptional (Fig 2D). Such a commitment of PIGA mutant HSCs is probably induced by some immunological mechanisms, due to the fact that the presence of PNH-type cells are associated with good response to IST in AA patients as well as a particular HLA-DR allele (HLA-DR15) (Maciejewski et al, 2001a).
The second phase was a continuous change that occurs from several months to years (sometimes a few dozen years) after the first phase. The vector of the transitional curve in this phase was apparently different from that of the first phase, and it remained essentially constant regardless of the response to IST or patient’s clinical status; the percentages of PNH-type granulocytes constantly increased and decreased from the time of the first examination in the ‘Expansion’ and ‘Disappearance’ groups respectively, thus suggesting the second phase may depend on the proliferation and self-maintenance properties of individual clones that PIGA mutant HSC already have at the beginning of this phase rather than on the continuation of the immune selection, which is strongly considered to participate in the first phase.
Although the bigger clone at the beginning of the second phase tended to expand (Fig 2A) while the smaller clone tended to decline (Fig 2B), the great variety of the transitional curve, longevity and size of the PNH-type granulocytes (Fig 2C) indicate theses cells to be derived from PIGA mutant clones with various proliferation and self-maintenance properties. Some murine studies have shown considerable variation in the proliferative potential of HSCs defined by a common phenotype (McKenzie et al, 2006). Clonal haematopoiesis is detectable in approximately 20% of elderly women (Busque et al, 2009) and patients with typical AA in remission, thus indicating that fewer HSCs can support haematopoiesis for long time. All these findings suggest that extensive increase in the percentage of PNH-type cells leading to classic PNH occurs only when an HSC with high proliferative capacity undergoes PIGA mutation by chance before the development of BM failure and subsequently gets activated by some sort of immune mechanism(s). On the other hand, a recent study suggested HMGA2 abnormalities to play a role in the acquisition of proliferative advantage by PIGA mutant HSC clones of two patients with classical PNH (Inoue et al, 2006). Although the likelihood of HSCs with a PIGA mutation undergoing secondary genetic changes is extremely low and the rates of the increase in the PNH-type cell granulocytes in the ‘Expansion’ group were constant, it is possible that additional gene abnormalities, such as the HGMA2 mutation, may thus take place in PIGA mutant HSCs soon after the development of BM failure.
The results of the present study provide important information for predicting the prognosis of PNH+ patients. Because others have reported that 10–25% of patients with AA treated with IST progress to classical PNH, those with AA or RA bearing a higher proportion of PNH-type cells at the time of diagnosis are therefore considered to be at high-risk (Tichelli et al, 1994; Frickhofen et al, 2003). However, this is not necessarily true. A high proportion of PNH-type cells at the time of diagnosis does not always reflect the proliferative potential of PIGA mutated HSCs. What determines the risk of progression to classic PNH is a vector of the transition of PNH-type cell percentages during the first 1–2 years after diagnosis. PNH+ patients with BM failure may therefore not have to worry about the progression to classical PNH if their PNH-type cell percentages do not increase during the first 1–2 years.
In conclusion, PNH-type cells in patients with BM failure are likely to be derived from single PIGA mutant HSCs rather than HPCs. The fate of PNH-type cells is therefore composed of two phases, namely the expansion phase due to the selective activation of static PIGA mutant HSCs at the onset of immune-mediated BM failure followed by the constant proliferation phase, depending on the self-maintenance and differentiation properties of individual HSC clones. The second phase is not influenced by an ongoing immune attack from which PNH-type cells can escape.
The authors thank Ms Megumi Yoshii and Ms Rie Oumi for their excellent technical assistance. This work was supported by a Grant-in-Aid for Scientific Research from the Ministry of Education, Culture, Sports, Science and Technology of Japan (No.15390298), a grant from the Ministry of Health, Labour and Welfare of Japan and a grant from the Japan Intractable Diseases Research Foundation.
C.S. and K.M. contributed equally to this work and participated in designing and performing the research. Z.Q. performed some experiments. C.S. and S.N. wrote the paper. C.S., K.M., N.S., K.I., Y.K., H.Y., A.T. and H.O. helped manage the samples. All authors approved the final version of the manuscript.
The authors declare no competing financial interests.