Paroxysmal nocturnal haemoglobinuria

Abstract

Paroxysmal nocturnal haemoglobinuria (PNH) is a clonal haematopoietic stem cell (HSC) disease that presents with haemolytic anaemia, thrombosis and smooth muscle dystonias, as well as bone marrow failure in some cases. PNH is caused by somatic mutations in PIGA (which encodes phosphatidylinositol N-acetylglucosaminyltransferase subunit A) in one or more HSC clones. The gene product of PIGA is required for the biosynthesis of glycosylphosphatidylinositol (GPI) anchors; thus, PIGA mutations lead to a deficiency of GPI-anchored proteins, such as complement decay-accelerating factor (also known as CD55) and CD59 glycoprotein (CD59), which are both complement inhibitors. Clinical manifestations of PNH occur when a HSC clone carrying somatic PIGA mutations acquires a growth advantage and differentiates, generating mature blood cells that are deficient of GPI-anchored proteins. The loss of CD55 and CD59 renders PNH erythrocytes susceptible to intravascular haemolysis, which can lead to thrombosis and to much of the morbidity and mortality of PNH. The accumulation of anaphylatoxins (such as C5a) from complement activation might also have a role. The natural history of PNH is highly variable, ranging from quiescent to life-threatening. Therapeutic strategies include terminal complement blockade and bone marrow transplantation. Eculizumab, a monoclonal antibody complement inhibitor, is highly effective and the only licensed therapy for PNH.

Paroxysmal nocturnal haemoglobinuria (PNH) is a rare haematological disorder characterized by episodes of haemolysis, and has fascinated haematologists for more than a century because of its diverse manifestations and intricate pathophysiology^1,2. The additional clinical manifestations of PNH are thrombosis, bone marrow failure and evolution of the disease to myelodysplastic syndromes. These non-erythroid manifestations indicate that PNH results from the clonal expansion of a mutated haematopoietic stem cell (HSC)³ (FIG. 1).

Figure 1 |. Clonal expansion in paroxysmal nocturnal haemoglobinuria.

Paroxysmal nocturnal haemoglobinuria is caused by somatic mutations (denoted by white stars) in the X-linked PIGA gene in one or more clones of multipotent haematopoietic stem cells (HSCs). For clinical manifestations to develop, the mutated HSC clone must expand, thereby generating many affected peripheral blood cells. PIGA mutations are not sufficient to lead to clonal expansion. Clonal expansion can arise from clonal selection by extrinsic factors (for example, aplastic anaemia) that preferentially target normal HSCs and, therefore, confer a conditional growth advantage to the mutated HSCs. Clonal expansion can also arise from intrinsic clonal evolution, by which HSCs acquire additional mutations that provide an intrinsic survival and growth advantage. Both mechanisms can coexist. NK, natural killer.

The term PNH was introduced by J. Enneking in 1928 (REF. 4). In 1937, Thomas Ham observed that erythrocytes from patients with PNH haemolysed when incubated with acidified serum. This seminal discovery resulted in the first diagnostic test for PNH: the acidified serum or Ham test⁵.

In the 1980s, it was discovered that blood cells from patients with PNH are deficient in proteins that bind to the cell surface through a glycosylphosphatidylinositol (GPI) anchor. Several of these proteins (for example, complement decay-accelerating factor (also known as CD55) and CD59 glycoprotein (CD59)) are complement-regulatory proteins. Indeed, a humanized mono clonal antibody that inhibits terminal complement activation (eculizumab; manufactured with the brand name Soliris by Alexion Pharmaceuticals, New Haven, Connecticut, USA) can ameliorate haemolysis and disease symptoms in patients with PNH⁶. Clonal cells with a deficiency of GPI-anchored proteins can be erythrocytes, platelets, polymorphonuclear leukocytes, monocytes and B cell, T cell and natural killer lymphocytes^7,8. However, because erythrocytes lack a nucleus, they are more susceptible to lysis. Mutations in the phosphatidylinositol glycan anchor biosynthesis class A gene (PIGA), which encodes a protein essential for the synthesis of GPI anchors, are responsible for the GPI-anchored protein deficiency⁹. This Primer covers the epidemiology, pathophysiology, clinical manifestations, diagnosis and management of PNH.

Epidemiology

Incidence and prevalence

PNH incidence is estimated at 1–1.5 cases per million individuals worldwide, but might be higher in certain regions^10,11. The disease occurs more frequently in countries in Asia (for example, Japan, Korea and China) than in western countries (the United States, Spain and the United Kingdom)^10,12. The International PNH Registry was established in 2003 to collect comprehensive data on the natural history of PNH and can provide some epidemiology data¹⁰. Patients of any age with a clinical diagnosis of PNH (by any applicable diagnostic method) or a detectable fraction of PNH-affected blood cells (that is, a PNH clone) of ≥0.01% of all blood cells are eligible for inclusion. As of 30 June 2012, 1,610 patients from 273 centres in 25 countries were enrolled; of these patients, 92.5% were from Europe and North America and 87.5% were of white ethnicity. No definitive biological data exist to fully explain this distribution; furthermore, there may be bias in the registry. The remaining patients were of Asian or Pacific Island descent (5%), of African descent (3.5%), native/ Aboriginal descent (0.2%) or of other or unknown ethnicity (3.9%)¹³. The most prominently represented age range was 30–59 years (54.6% of the registry population). PNH is rare in children; it tends to manifest in the teenage years^14–17. There is a slight female pre-dominance (54.4%) overall¹⁰, although the proportion of women with PNH was significantly lower in Asian countries (44.9%) than in western countries (54.9%) in a meta-analysis¹¹. This finding might correlate with the observation that, in some countries in Asia, men have more access to medical care than women¹¹. A limitation of the data from the International PNH Registry is that information on PNH is not available from all countries. Furthermore, many patients enrolled in the International PNH Registry have aplastic anaemia as their primary diagnosis, as the registry allows inclusion of patients with >0.01% PNH granulocytes^18,19. Thus, these prevalence rates are not only based on patients with haemolytic PNH.

Specific clinical manifestations of PNH might vary in different ethnicities^2,20. PNH-associated thrombotic events occur in up to 30% of patients in western countries compared with <15% of patients in Asian countries^2,10,21. The epidemiology of PNH-associated bone marrow failure is not as well described in these studies. However, given the overall increased prevalence of aplastic anaemia in Asian countries compared with western Europe and the United States^22,23, one could hypothesize that PNH-associated bone marrow failure is also more frequent in Asian patients than is thrombosis associated with PNH.

Mortality

The median survival of patients with PNH was approximately 10 years in the 1990s^2,24,25, but increased to >20 years in the early 2000s²⁵. Since the introduction of eculizumab therapy, patients with PNH can live a relatively normal lifespan^10,26, and patients with haemolytic PNH receiving eculizumab have a more-favourable prognosis than patients with a more profound bone marrow failure component, such as aplastic anaemia^10,27. The reason for this difference is that eculizumab does not treat the underlying production deficit in the bone marrow. For the 122 patients who have died since enrolment into the International PNH Registry, the most frequent cause of death (11.7%) was bone marrow failure in patients who met the diagnostic criteria for PNH and aplastic anaemia¹⁰. Estimations using data from the registry indicate that approximately 75% of patients with PNH are treated with eculizumab¹⁰.

Mechanisms/pathophysiology

GPI biosynthesis

Patients with PNH have clonal blood cells with defective surface expression of various GPI-anchored proteins. These proteins include monocyte differentiation antigen CD14 (CD14), low-affinity immunoglobulin gamma Fc region receptor III-B (CD16b), CD48 antigen (CD48), CD55 and CD59, among many others. Studying B lymphoblastoid and T lymphoblastoid cell lines established from patients with PNH showed that GPI biosynthesis was impaired in these cells^28–31.

GPI is synthesized in the endoplasmic reticulum from phosphatidylinositol through the sequential additions of monosaccharide molecules and other components via 11 reaction steps^32,33 (FIG. 2). Nascent GPI-anchored proteins undergo several remodelling reactions in the endoplasmic reticulum and the Golgi apparatus during transport to the cell surface³³. At the cell surface, the GPI-anchored proteins are primarily localized to microdomains that are rich in glycosphingolipids and cholesterol, termed lipid rafts³⁴. In PNH-affected cells, the first step in GPI biosynthesis is defective^28–30 (FIG. 2); as a result, PNH cells have defective surface expression of various GPI-anchored proteins.

Figure 2 |. Biosynthesis of glycosylphosphatidylinositol-anchored proteins.

Precursor glycosylphosphatidylinositol (GPI)-anchored proteins have a GPI-attachment signal peptide at their carboxyl terminus, where preassembled GPI is attached as a post-translational modification. GPI-transamidase cleaves off the GPI-attachment signal peptide and attaches GPI to the newly generated C terminus by transamidation³³. Paroxysmal nocturnal haemoglobinuria (PNH) cells lack or have severely reduced activity of phosphatidylinositol (PI)-N-acetylglucosamine (GlcNAc) transferase, the enzyme encoded by PIGA that mediates the first step of the GPI biosynthetic pathway: the transfer of GlcNAc from uridine-diphosphate-N-acetylglucosamine (UDP-GlcNAc) to PI that generates PI-GlcNAc³³. Thus, GPI molecules that are competent for attachment to proteins are not generated, the C-terminal GPI-attachment signal peptide in precursor proteins is not cleaved by GPI-transamidase and the precursor proteins are degraded intracellularly, most likely by proteasomes. For each reaction step, the genes involved are shown; a question mark indicates that the corresponding genes have not been identified. From step 5 onwards, diacyl glycerol might be replaced by 1-alkyl,2-acyl glycerol (denoted by the change in colour). ER, endoplasmic reticulum.

Genetic mutations

PNH cells carry a loss-of-function mutation in PIGA^9,35. PNH-linked PIGA mutations are somatic mutations, as patients with PNH can harbour blood cells with normal levels of GPI-anchored proteins⁹. Moreover, several lines of evidence indicate that mutations occur in long-lasting multipotent HSCs. First, the same mutation is found in granulocytes and B lymphocytes⁹. Second, cells defective in GPI-anchored proteins belong to all haematopoietic cell lineages^7,8. Third, the same somatic mutations can be observed in blood samples from the same patients obtained >10 years apart³⁶.

Somatic PIGA mutations in patients with PNH are manifold^37–40. The majority are insertions or deletions (indels) involving a single base or several bases, and single base substitutions. Deletions of the entire gene or a large part of the gene are rare. Single base deletions and single base substitutions account for approximately one-third of known mutations each, and the remaining one-third comprises single base insertions, several base indels and combinations of several base indels. The most frequent outcomes of these somatic mutations are frameshifts³⁹.

PIGA is located on Xp22.2. The X chromosome localization explains why one somatic PIGA mutation can be sufficient to cause GPI deficiency in most patients with PNH, as only one allele is functional in both men and women. However, at least 20% of patients with PNH have ≥2 somatic PIGA mutations, each generating a different PNH clone^41,42. Usually one clone is predominant, although in some cases two clones are comparably large⁴².

In addition to PIGA, >20 genes on various autosomes are involved in GPI biosynthesis and attachment to proteins³² (FIG. 2). Loss of function in many of them causes a GPI-anchored protein-defective phenotype, as indicated by studies of mutant cell lines defective in one of these genes³². In autosomal genes, loss of function generally occurs only if both alleles are mutated, which could be a result of two independent somatic mutations or a germline and a somatic mutation. However, such an event is extremely rare. Only one patient with PNH in whom GPI deficiency was not caused by PIGA mutations has been reported⁴³. This individual had heterozygous mutations in PIGT, which is located on chromosome 20 and encodes an essential subunit of GPI transamidase⁴⁴. One PIGT allele bore a severe loss-of-function germline mutation, and in the GPI-anchored protein-defective cells, the second allele was lost owing to a somatic 8 Mb deletion. Thus, in the affected cells of this patient, GPI transamidase activity was almost completely absent, leading to a GPI-deficient phenotype⁴⁵.

Expansion of GPI-defective HSCs

Somatic PIGA mutations can occur in one or a few HSCs. If the mutant HCSs generate progeny cells at the rate of healthy quiescent HSCs, clinical manifestations do not occur. Clonal expansion of HSCs deficient in GPI-anchored proteins is essential for clinical manifestations of PNH. Two mechanisms of clonal expansion have been demonstrated: clonal selection by extrinsic factors^46–49 and intrinsic clonal evolution⁵⁰ (FIG. 1). These two mechanisms are not mutually exclusive. Clonal selection alone can be responsible for the clonal expansion of PNH HSCs to numbers that are sufficient to cause clinical PNH (20–30% of all circulating erythrocytes) in patients with aplastic anaemia. These patients can be asymptomatic for years if the proportion of their PNH erythrocyte clone stays low (<10% of all circulating erythrocytes). However, when the proportion of the PNH erythrocyte clone increases greatly (>80%) and overt clinical PNH develops, the clonal evolution mechanism might also be operating⁵¹.

CD59-defective cells are clonally expanded in bone marrow cells from patients with PNH. Clonal expansion was demonstrated in cells expressing the haematopoietic progenitor cell antigen CD34 (REF. 52); in patients with PNH, CD34⁺ cells in the bone marrow consist of a small number of CD59⁺ (normal) cells and a large number of CD59⁻ cells. Mechanisms of clonal expansion have been studied actively since somatic PIGA mutations were identified. The consensus is that a PIGA mutation is not sufficient to induce clonal expansion by itself and additional factors are required. Three lines of evidence support this hypothesis. First, ~0.003% of peripheral blood granulocytes in most healthy individuals are deficient in GPI-anchored proteins and possess somatic PIGA mutations, similar to those observed in cells deficient in GPI-anchored proteins from patients with PNH^53,54. The same mutations could also be identified in samples obtained months later, suggesting that somatic PIGA mutations could occur in progenitor HSCs in healthy individuals⁵⁵. Second, when multiple PNH clones bearing different complete loss-of-function PIGA mutations coexist, one is usually predominant, indicating that different clones have different abilities to expand⁴¹. Third, HSCs from Piga-knockout mice do not show clonal expansion in the bone marrow^56–60.

Clonal selection by extrinsic factors.

Several extrinsic factors have been proposed as the underlying mechanisms of clonal selection. A theoretical report suggests that expansion of a PIGA-mutant HSC clone could occur spontaneously by genetic drift (the phenomenon in which the relative frequency of an allele randomly varies) as the overall number of HSCs decrease as a result of conditions such as bone marrow failure⁶¹. Whether this mechanism actually occurs in patients with PNH is yet to be seen. Bone marrow failure can occur independently of PIGA mutations in patients with PNH and can contribute to the clonal expansion of PIGA-mutant HSCs. Bone marrow failure in PNH might be caused by autoimmunity to HSCs, a mechanism similar to that observed in idiopathic aplastic anaemia⁶². One hypothesis is that HSCs expressing GPI-anchored proteins are selectively killed by autoreactive cytotoxic lymphocytes, whereas GPI-anchored protein-deficient HSCs are spared⁶³. Indeed, PNH cells are seldom seen in inherited forms of bone marrow failure, in which autoreactive cytotoxic lymphocytes are not functional⁶⁴.

Whether the self-antigens that induce the auto-immune responses that might cause PNH-associated bone marrow failure involve GPI-anchored proteins is unclear. One hypothesis is that glycolipids from GPI are recognized by a class of cytotoxic CD8⁺ T cells that also express cell markers typically found in natural killer cells⁶⁵. These particular T cells can be observed in both healthy individuals and in patients with PNH. In patients with PNH, these T cells react with complexes of GPI and antigen-presenting glycoprotein CD1d⁶⁶. CD1d is a MHC class I-related protein that presents glycolipid antigens to natural killer cells that express specific CD1d-restricted invariant T cell receptors. Notably, such a repertoire of T cell receptors is highly homologous in patients with PNH, supporting the idea that autoreactive T cells in these patients recognize similar antigens⁶⁷. Another hypothesis proposes that PIGA-mutant HSCs are more resistant to cellular autoimmunity specifically because they lack NKG2D ligand 1, NKG2D ligand 2 and NKG2D ligand 3 (REFS 68,69). These three GPI-anchored proteins activate natural killer cells and CD8⁺ T cells expressing NKG2D (encoded by KLRK1) receptors^68,69.

Intrinsic clonal evolution.

The intrinsic clonal evolution mechanism proposes that one or more additional genetic alterations occurring in a PIGA-mutant HSC confer on this HSC clone a benign growth advantage. This mechanism was demonstrated in some patients with PNH. In two patients, PNH clones in the bone marrow showed aberrant ectopic expression of HMGA2 on chromosome 12 (REF. 70). HMGA2 encodes high-mobility group protein HMGI-C, an architectural transcription factor that is expressed in embryonic stages. Ectopic expression of HMGA2 in adipocytes and uterine smooth muscle cells can cause lipomas and leiomyomas, respectively⁷¹. Furthermore, in an individual with β-thalassaemia who received gene therapy, the insertion of the lentiviral vector activated HMGA2 in a HSC and caused clonal expansion⁷². Transgenic mice expressing HMGA2 showed phenotypes similar to those observed in patients with myeloproliferative neoplasms, and HSCs from these mice had a growth advantage⁷³. Significantly high levels of HMGA2 mRNA were found in granulocytes from 60% of patients with PNH even though these individuals did not have apparent chromosome 12 abnormalities⁷⁴. Together, these findings suggest that HMGA2 activation, through gain-of-function somatic mutations or other putative acquired epigenetic alterations, can induce a benign growth phenotype that can contribute to the clonal expansion of PIGA-mutant HSCs in patients with PNH. In addition, the V617F gain-of-function mutation in the tyrosine protein kinase Janus kinase 2 (JAK2), which is commonly observed in myeloproliferative neoplasms, was identified in PNH clones from patients with PNH who had characteristics of myelofibrosis⁷⁵.

However, dysregulation of HMGA2 and the JAK2 gain-of-function mutation might not be the only mechanisms of clonal expansion by intrinsic clonal evolution. Whole-exome sequencing in 12 patients with PNH identified 21 non-silent somatic mutations (in addition to PIGA mutations), but only 2 loss-of-function mutations were identified, both in TET2 (which encodes a methylcytosine dioxygenase)⁷⁶. In a cohort of 36 patients, targeted deep sequencing of 61 genes, such as TET2, SUZ12 (which encodes a Polycomb protein) and JAK2, which are involved in the pathogenesis of myeloproliferative neoplasms, identified mutations in 15 patients only⁷⁶. Thus, in some patients, somatic PIGA mutations were the only identified mutations and, although additional undetected somatic mutations might coexist, no major driver mutation could account for clonal expansion. Moreover, the average number of somatic mutations identified in patients with PNH is ~2.1, which is much lower than the numbers found in patients with myelodysplastic syndromes or myeloproliferative neoplasms^77,78. Additional whole-genome sequencing studies and epigenomic analyses will be required to clarify the mechanisms of intrinsic clonal evolution.

Pathophysiology

The main consequences of clonal expansion of PIGA-mutant HSCs are intravascular haemolysis and thrombosis; bone marrow failure can develop independently and extravascular haemolysis only manifests under eculizumab therapy.

Intravascular haemolysis.

Complement regulators, such as CD55 and CD59 (REF. 79), are physiologically expressed on cell surfaces to prevent activated complement toxicity (FIG. 3). In patients with PNH, at least 5–10%, and usually much higher fractions, of erythrocytes are PNH cells. Consequently, the affected erythrocytes are more susceptible to complement attack and are simultaneously lysed during infections and other events that trigger complement activation, causing paroxysmal haemolysis (a haemolytic attack). The alternative complement pathway is constantly activated at low levels. Thus, patients with PNH also experience chronic basal haemolysis, which increases during sleep for uncertain reasons (hence the name PNH). Interestingly, even PNH erythrocytes that completely lack CD55 and CD59 can circulate in the blood for up to 60 days in some patients. A possible explanation is that complement factor H, a regulator of the alternative pathway, can also bind to erythrocytes and function as a membrane regulator⁸⁰.

Figure 3 |. Intravascular and extravascular haemolysis in paroxysmal nocturnal haemoglobinuria.

Complement decay-accelerating factor (also known as CD55) and CD59 glycoprotein (CD59) are glycosylphosphatidylinositol (GPI)-anchored self-protective complement regulatory factors⁷⁹. CD55 is a widely expressed membrane protein that accelerates the decay of C3 convertases (C3 con) bound to the cell surface, thereby limiting the formation of C5 convertases. CD59 is also widely expressed: it blocks the generation of the membrane attack complex (MAC) and is, therefore, the major inhibitor of terminal complement^153,154. a | Normal red blood cells (RBCs) are protected from activated complement. b | By contrast, the CD55-defective and CD59-defective paroxysmal nocturnal haemoglobinuria (PNH) RBCs are highly sensitive to complement activation, which causes intravascular haemolysis. c | Upon eculizumab treatment, CD59 deficiency is compensated for and intravascular haemolysis is prevented owing to inhibition of C5 activation and subsequent MAC formation. However, CD55 deficiency remains on unlysed PNH RBCs, which causes inefficient downregulation of C3 convertases and in turn could lead to an accumulation of C3b and its processed forms iC3b and C3dg. iC3b and C3dg are ligands of integrin αM, β2 (CR3), a receptor expressed on macrophages in the spleen and liver^89,155; thus, these macrophages can recognize PNH RBCs, leading to extravascular haemolysis. C5b-8, complex of C5b, C6, C7 and C8; Hb, haemoglobin.

Thrombosis.

Mechanisms of thrombosis in PNH are currently unclear²¹. Intravascular haemolysis and activation of PNH platelets have been proposed as contributing factors. The involvement of signalling pathways that depend on the activation of complement C5 is suggested owing to the observation that eculizumab therapy can substantially reduce the number of thrombotic events^26,81. C5 activation promotes coagulation via various mechanisms⁸²; in turn, coagulation factors can also activate the complement cascade^83–87.

Extravascular haemolysis.

Extravascular haemolysis occurs in the spleen and liver. When PNH erythrocytes accumulate a sufficient amount of the processed forms of complement C3 on their surface, they are recognized and phagocytosed by resident macrophages in these organs^88–90 (FIG. 3c). In fact, C3 fragments can be detected on PNH erythrocytes by flow cytometry and Coombs tests, but not on normal erythrocytes in the same patients^88,91.

Bone marrow failure.

In contrast to intravascular haemolysis and thrombosis, bone marrow failure is not a downstream event of somatic PIGA mutations in HSCs. PNH-associated bone marrow failure is independently caused by cellular autoimmunity to HSCs⁶². In some patients with PNH, usage of the gene encoding T cell receptor is skewed towards generating receptors that are usually expressed by natural killer cells and cytotoxic T cells, and clones of these cells are expanded in some patients^92,93.

Diagnosis, screening and prevention

Classification

PNH has varied clinical presentations, including Coombs-negative haemolytic anaemia, pancytopenia, unexplained abdominal pain, haemoglobinuria (passing urine of a colour ranging from dark tea to black to cherry red owing to high levels of haemoglobin in the urine) and thrombosis. PNH can arise de novo or evolve from acquired aplastic anaemia. No universally accepted classification scheme is available. The International PNH Interest Group classifies PNH into three categories: classical PNH (in which patients have clinical manifestations of haemolysis or thrombosis); PNH in the context of other primary bone marrow disorders (such as aplastic anaemia or myelodysplastic syndromes); and sub clinical PNH, in which patients have low proportions of PNH clones but no clinical or laboratory evidence of haemolysis or thrombosis⁹⁴. The PNH International Registry adopted a classification first used by de Latour et al.²⁵ that includes haemolytic (or classical) PNH, aplastic anaemia-PNH and intermediate PNH¹⁰. Patients with haemolytic PNH tend to have near-physiological neutrophil and platelet counts, lactate dehydrogenase levels of more than two times the upper physiological limit (which points to intravascular haemolysis), a normocellular bone marrow, an increased reticulocyte (an immature red blood cell) count and a relatively large population of PNH granulocytes (usually >50%). Patients with aplastic anaemia-PNH (acquired aplastic anaemia with a low-to-moderate proportion of a PNH clone) are severely pancytopenic, tend to have a hypocellular bone marrow, a relatively low reticulocyte count and a smaller percentage of PNH granulocytes. Patients with a PNH clone identified by flow cytometry who did not fulfil the criteria of either category were classified as intermediate PNH. Both of these classification schemes have limitations, and a patient’s classification can change over time. For example, patients with aplastic anaemia-PNH might experience improved haematopoiesis associ ated with expansion of their PNH clone and later meet the criteria for haemolytic PNH. Less commonly, patients with haemolytic PNH might develop aplastic anaemia-PNH.

Diagnosis

Laboratory tests that should be requested if PNH is suspected (for example, if the patient presents with haemoglobinuria) include a complete blood count with differential, a reticulocyte count, a peripheral blood smear and a lactate dehydrogenase assay. Clinical diagnosis of PNH should be confirmed with peripheral blood flow cytometry to determine the absence or severe deficiency of GPI-anchored proteins on at least two or more lineages of blood cells^95,96. GPI-anchored proteins can be detected after labelling the cells with mono clonal antibodies (for example, anti-CD55 or anti-CD59) or a reagent known as fluorescein-tagged proaerolysin (FLAER)⁹⁷, which binds to the glycan portion of the GPI anchor. FLAER is best used on nucleated cells; it does not stain red blood cells, as red blood cells express high levels of glycophorin, a protein that binds to aerolysin and, therefore, interferes with the assay. Consensus guidelines for detecting GPI-anchored protein-deficient blood cells by using a combination of FLAER and several mono clonal antibodies have been published⁹⁸. Flow cytometry should be requested for patients with Coombs-negative haemolytic anaemia, patients with aplastic anaemia or unexplained pancytopenia and patients with unexplained thrombosis who have other clinical or laboratory evidence of PNH. Notably, small numbers of PNH granulocytes (<0.01%) or red blood cells are present in some healthy controls and in patients with myelodysplastic syndromes. These PIGA mutations often arise from colony-forming cells rather than HSCs and are not clinically or diagnostically significant^53,99.

Bone marrow findings.

Bone marrow biopsy is not required for the diagnosis of PNH but should be performed in patients with severe pancytopenia. In patients with haemolytic PNH, the bone marrow is often normocellular to hypercellular with erythroid hyper plasia. Stainable iron might be absent in patients with a long history of intravascular haemolysis. Erythroid dysplasia is common owing to the rapid turnover of red blood cells; as a result, the bone marrow might resemble the bone marrow observed in patients with myelodysplastic syndromes¹⁰⁰. The myeloid and megakaryocyte lineages are usually morphologically normal. Karyotypic abnormalities might be present in up to 25% of patients¹⁰⁰. Patients with aplastic anaemia-PNH have a hypocellular bone marrow with relative erythroid hyperplasia (red blood cell precursor counts are mildly increased, but this increase is magnified by the absence of many other haematopoietic elements). Stainable iron is usually present. Subtle erythroid dysplasia is common and mega karyocyte counts are markedly decreased. The karyo type is usually normal and the number of CD34⁺ cells (haematopoietic stem and progenitor cells) is markedly reduced¹⁰¹.

Clinical manifestations

Anaemia.

Anaemia in PNH is often multifactorial and can result from a combination of haemolysis and bone marrow failure. Intravascular haemolysis with moderate-to-severe anaemia, an increased reticulocyte count, a normal-to-increased mean corpuscular volume (the average volume of red blood cells) and a markedly increased level of lactate dehydrogenase are common in haemolytic PNH. The peripheral blood smear is often non-descript. In aplastic anaemia-PNH, anaemia is primarily due to bone marrow failure; thus, these patients frequently have hypocellular bone marrow, more-severe thrombocytopenia, a small percentage of PNH granulocytes, lower reticulocyte counts and modest or no increase in lactate dehydrogenase levels¹⁸. Thrombosis might occur in aplastic anaemia-PNH, but is less common than in patients with haemolytic PNH.

Patients with aplastic anaemia-PNH often have a diagnosis of mild aplastic anaemia, and haematopoiesis can recover after treatment of acquired aplastic anaemia¹⁰². Expansion of PNH clones and PNH symptoms might accompany relapse of aplastic anaemia. In some patients, expansion of PNH clones is associated with improved blood counts⁴⁸.

Thrombosis.

Thrombosis is the most common cause of mortality in PNH²¹ (accounting for almost 50% of deaths before complement inhibition therapy was introduced); venous thrombosis tends to be more common than arterial thrombosis. Hepatic vein thrombosis (Budd–Chiari syndrome) is the most common occurrence; other sites frequently affected by thrombosis include abdominal (for example, portal, mesenteric and splenic) and cerebral (sagittal and cavernous sinus) veins. Deep vein thrombosis, pulmonary emboli and dermal thrombosis are also relatively common.

Smooth muscle dystonia.

Abdominal pain, back pain, oesophageal spasm, dysphagia (difficulty swallowing) and erectile dysfunction are common manifestations associated with haemolytic PNH and are often a direct consequence of intravascular haemolysis and the release of free haemoglobin^103,104. Free haemoglobin is normally eliminated by haptoglobin, scavenger receptor cysteine-rich type 1 protein M130 and haemopexin. These clearance mechanisms are overwhelmed in the setting of intravascular haemolysis, and the levels of free haemoglobin in the plasma consequently increase. Free haemoglobin scavenges nitric oxide, which is synthesized by endothelial cells, maintains smooth muscle relaxation and inhibits platelet activation and aggregation. Thus, nitric oxide deficiency at the tissue level contributes to deregulation of smooth muscle tone and platelet activation. Smooth muscle dystonias are more common in patients with large proportions of PNH granulocyte clones and patients with markedly increased levels of lactate dehydrogenase¹⁰³.

Fatigue and haemoglobinuria.

Disabling fatigue is a common feature of PNH and can be disproportionate to the degree of anaemia. Fatigue is often most intense during a haemolytic attack, but is usually present at all times. Episodes of jaundice and haemoglobinuria are reported by almost 50% of patients. These signs and symptoms can be constant or paroxysmal and are often exacerbated by infections, surgery, exercise, pregnancy or excessive alcohol intake.

Other manifestations.

Patients with PNH have an increased risk of chronic kidney disease as a result of long-term intravascular haemolysis¹⁰⁵. Renal tubular damage can occur from microvascular thrombosis, accumulation of iron deposits or both. Mild-to-moderate pulmonary hypertension has also been reported, but the association between chronic kidney disease and clinically significant pulmonary hypertension is still controversial^106–108. Elevated pulmonary pressures and reduced right ventricular function caused by subclinical microthrombi or haemolysis-associated nitric oxide scavenging might also lead to dyspnoea.

Management

The only disease-modifying therapeutic strategies for PNH are complement inhibition therapy (eculizumab) and bone marrow transplantation. Eculizumab is the only licensed therapy for PNH, and its efficacy has relegated bone marrow transplantation to second-line therapy for haemolytic PNH in countries where the drug is available. Bone marrow transplantation might be an option if eculizumab is unavailable, and is a reasonable therapeutic strategy in patients with PNH and severe bone marrow failure. Adjunctive therapies (for example, immunosuppression) could be prescribed to patients with PNH who also have bone marrow failure to ameliorate the latter. However, these adjunctive treatments are not specific for PNH nor do they have consistent effects on the expansion or reduction of PNH clones; thus, they are not discussed further. The only curative strategy for PNH is allogeneic stem cell transplantation, but this procedure continues to carry a considerable risk of mortality^109–111. Given that a small proportion (~2% were observed over 10 years) of patients with PNH could experience spontaneous remission² and the efficacy of eculizumab, stem cell transplantation should only be considered for patients with PNH who also have severe bone marrow failure. Non-myeloablative or reduced intensity conditioning regimens are relatively safe and effective^112,113.

Eculizumab therapy

Eculizumab binds to C5 and prevents cleavage by C5 convertase, thereby inhibiting terminal complement activity^6,114–116 and the formation of the membrane attack complex (FIG. 4). Thus, eculizumab compensates for CD59 deficiency in patients with PNH (FIG. 3). As of 2014, eculizumab for the indication of PNH has been safely administered to >1,000 patients worldwide over 14 years¹⁰. Treatment can lead to resolution of intravascular haemolysis^114–117, reduction in thrombosis rate⁸¹, improvement or stabilization of renal function¹⁰⁵, improvement in pulmonary pressures^106,107, successful pregnancy outcomes¹¹⁸ and improved survival²⁶. The most serious risk of terminal complement blockade is life-threatening neisserial infections (an incidence of ~0.5% per year or 5% after 10 years). Thus, all patients receiving eculizumab should be vaccinated against Neisseria meningitidis with a meningococcal tetravalent polysaccharide vaccine (for example, Menveo (Novartis Vaccines and Diagnostics, Cambridge, Massachusetts, USA)). However, in patients with terminal complement deficiency, vaccination reduces but does not eliminate the risk of infection^119,120. Vaccination against N. meningitidis serotype B should also be administered if available (for example, Bexsero (Novartis Vaccines and Diagnostics). There is a risk of further complement activation with vaccination. Thus, our opinion is that meningococcal vaccination should follow initial eculizumab therapy, as reducing the risk of thrombosis or developing acute renal failure is the clinical priority; however, this protocol has not been validated in clinical trials. Note that our advice differs from the US FDA black box warning to “Immunize patients at least 2 weeks before administering the first dose of Soliris, unless the risks of delaying Soliris therapy outweigh the risks of developing a meningococcal infection.” Patients should receive ciprofloxacin antibiotic prophylaxis for the first 2 weeks after starting eculizumab infusions, and recommendations from many countries opt to continue long-term prophylaxis with penicillin V (if no allergies exist).

Figure 4 |. Complement cascade inhibition.

The lectin, classical and alternative pathways converge at the step of complement component 3 (C3) activation. Haemolysis in paroxysmal nocturnal haemoglobinuria (PNH) is usually chronic because the alternative pathway is always in a state of low-level activation through a process known as tickover. Terminal complement is initiated by C5 convertases, leading to cleavage of C5 to C5a and C5b. C5b oligomerizes with C6, C7, C8 and multiple C9 molecules to form the membrane attack complex (MAC). The complement decay-accelerating factor (CD55) inhibits proximal complement activation by accelerating the decay of C3 convertases; CD59 glycoprotein (CD59) inhibits terminal complement activation by preventing the incorporation of C9 into the MAC. There is a potent amplification loop within the alternative pathway. The absence of CD55 and CD59 on PNH cells leads to haemolysis, inflammation, platelet activation and thrombosis. Eculizumab prevents C5 convertases from cleaving C5 into C5a and C5b. C5 activation promotes coagulation via various mechanisms, including activating thrombin⁸². Thrombin cleaves C3 (REFS 84,85) and also generates C5a in the absence of C3 (REF. 86). The fibrinolytic factors plasmin and kallikrein also directly cleave C3 (REF. 87). MASP, mannose-binding lectin-associated serine protease.

Monitoring during eculizumab therapy

Patients with PNH on eculizumab therapy should be monitored regularly for several reasons: first, the underlying bone marrow failure might progress and require treatment; second, to confirm that intravascular haemolysis is well controlled; and third, to monitor the proportion of PNH cells. In the small percentage of patients who have spontaneous remission of PNH, therapy can be discontinued. However, occasionally what appears to be a spontaneous remission might actually be evolving transformation and careful reassessment is required¹²¹.

Continued anaemia.

Inadequate response to eculizumab has been reported. In true refractory cases, the patients had mutant complement C5 that prevented eculizumab binding¹²². An alternative complement inhibitor is desirable for these patients, with trial enrolment if available. Continued anaemia in patients treated with eculizumab is common and its reason should be explored. Bone marrow failure is a common cause and concomitant immunosuppressive therapy can be safely administered if necessary. Furthermore, up to 20% of patients might require a higher dose of eculizumab to maintain terminal complement blockade during the 14-day interval between infusions. This situation often presents clinically with a return of PNH-related symptoms (abdominal pain, dysphagia and dark urine) 1–2 days before the next infusion is due. However, the patient might also be relatively asymptomatic and the under-dosing can only be uncovered by the need for transfusions or raised lactate dehydrogenase levels. Thus, it is useful to check lactate dehydrogenase levels on the day of the infusion and increase the eculizumab dose accordingly to reduce the risks of PNH-related complications and transfusion requirements¹¹⁴.

Low-level extravascular haemolysis is present in the majority of patients treated with eculizumab^88,90. No intervention is usually necessary, but patients might need occasional red blood cell transfusions^123,124. Importantly, available evidence indicates that extravascular haemolysis does not contribute to reduced survival; for this reason, interventions such as corticosteroid therapy or splenectomy are not recommended. Patients with iron overload also have reduced haematopoiesis, which can limit improvement in the haemoglobin level. Adequate haematinic replacement should be provided: folate supplementation as required and exogenous erythropoietin (the hormone that stimulates erythropoiesis) administration if erythropoietin levels are inappropriately low for the degree of anaemia. If the haemoglobin level temporarily falls during treatment, usually in association with an infection, three scenarios should be considered: suppression of bone marrow haematopoiesis (as seen in the setting of many infections), increased extravascular clearance of red blood cells (which is also often associated with infections) or breakthrough of terminal complement blockade due to increased complement activity. The first two situations require appropriate anti-infective and supportive measures only and do not result in increased PNH-associated complications. The third situation, however, could require attempts to restore terminal complement blockade to avoid complement-induced thrombosis or renal damage. During strong complement activation (for example, in certain infections), intravascular haemolysis might occur owing to a conformational change in C5 that limits the ability of eculizumab to block C5 convertases¹²⁵. We recommend that an extra dose of eculizumab is administered to achieve terminal complement blockade, even in the setting of meningococcal infection (see below); however, this practice is not universally accepted and is not always efficacious. Administering a monoclonal antibody at the time of an infection with a fever often feels counterintuitive to haematologists and nursing staff. However, this reluctance could be eased with education on the unique situation of PNH and eculizumab.

Discontinuing anticoagulation therapy when a patient is on effective doses of eculizumab has been increasingly performed worldwide, and there is little concern about recurrence of thrombotic events as long as the patient is routinely monitored¹²⁶. Plasma products should be avoided in patients on eculizumab therapy as they contain high levels of complement proteins. If the administration of plasma products is required (for example, in case of trauma or substantial haemorrhage), eculizumab will need to be re-administered.

Meningococcal infection.

Individuals treated with eculizumab (for any indication) are more susceptible to N. meningitidis infections. In the past 14 years, 3 confirmed cases of non-fatal meningococcal septicaemia out of >250 patients with PNH who were treated with eculizumab have been identified in the United Kingdom (A.H., unpublished observations). To manage this risk, clinicians must be vigilant and heed several important points. First, patients deficient in terminal complement proteins who develop septicaemia often have attenuated symptoms and reduced early mortality¹²⁷. Thus, patients with PNH receiving eculizumab therapy must be educated to seek immediate medical attention for any acute illness, including fever, chills, myalgia (muscle pain) and headache. Second, a patient can look clinically well but still have infection. Because full complement activation does not occur, cytokine release and inflammatory responses might not be as manifest as they normally are: the fever might not be very high and the levels of inflammatory markers such as C-reactive protein might not be substantially increased in the early phases of infection. Blood cultures should be requested, and prompt empirical treatment at this time could prevent deterioration. We reiterate that delivering an additional dose of eculizumab might be necessary if there are signs of increased complement activation. Although the risk of developing neisserial infections might increase, the risk of succumbing to them might be reduced in patients who are treated with complement inhibitors, as there is less risk of widespread multiple organ failure.

Emergency care

Patients with PNH might require urgent treatment in the event of thrombosis or acute renal failure. Specialist advice should be sought where possible.

Acute thrombosis.

Anticoagulation therapy with heparin or low-molecular-mass heparin is still the first action to take in the setting of a new thrombotic event. Complement inhibition therapy with eculizumab should be commenced within 24 hours of any new thrombotic event, wherever possible, to reduce the risks of propagation of the thrombotic insult, recurrence and subsequent long-term complications²¹. Because eculizumab seems to provide protection against the propagation of thrombosis or the occurrence of further thrombotic events, it is our opinion that the development of a PNH-related thrombosis is one of the primary indications to initiate eculizumab therapy.

In patients with PNH, the management of Budd–Chiari syndrome, which might occur despite anticoagulant prophylaxis, is complex. Immediate commencement of eculizumab is recommended and can reduce mortality and long-term sequelae. Anticoagulation therapy alone does not reliably restore hepatic blood flow. Thrombolytic therapy has also been used, but haemorrhagic complications remain a concern; however, thrombolytic therapy is less likely to be required if eculizumab therapy is started^128,129. There is an increased risk of developing hepatocellular carcinoma after an episode of Budd–Chiari syndrome; thus, patients should receive routine screening, such as regular blood tests for α-fetoprotein or liver ultrasound scans.

Acute renal failure.

Management of acute renal failure involves hydration, supportive care and occasionally haemodialysis. A Doppler ultrasound scan is recommended to rule out renal vein thrombosis. Although acute renal failure can be reversible, end-stage renal failure or ongoing progressive renal damage can ensue. Long-term eculizumab therapy can contribute to improved renal function¹⁰⁵.

Routine care

PNH is typically diagnosed in a non-acute setting. After the initial assessment, patients require regular monitoring to determine whether complement inhibition therapy (or supportive measures if eculizumab is not available) is indicated. If indications for complement inhibition therapy are not present, supportive measures can be used. Folic acid supplementation is necessary to sustain the increased red blood cell production. Chronic haemoglobinuria might lead to severe iron deficiency, and patients could require iron supplementation even if they are transfusion dependent. Iron supplementation should be discontinued once eculizumab therapy is initiated because patients might develop iron overload over the years.

The question of whether to commence prophylactic anticoagulation therapy remains. Before complement inhibition therapy was introduced, anticoagulation therapy could reduce the risk of thrombosis¹³⁰, but complications (such as haemorrhage) were frequent^{2,103,130,131} and a high risk of thrombosis continued despite treatment^{25,81,129,132–134}. Following a thrombotic event, anticoagulation therapy alone as secondary prevention is not sufficient^126,134. In patients with a substantial proportion of PNH cells, platelet counts of >100 × 10⁹ per litre and no known haemorrhagic risk (such as the presence of varices (dilated blood vessels)), if eculizumab is not available, it is reasonable to consider commencing primary prophylaxis therapy with warfarin, provided that the patient receives appropriate counselling regarding the risk of haemorrhage and a continued risk of thrombosis. Patients treated with eculizumab with no prior history of thrombosis do not require primary prophylaxis.

PNH in pregnancy

PNH in pregnancy increases the risk of mortality of both the mother and the fetus. Maternal mortality is estimated to be in the range of 6–20%, with ~12% of patients developing complications related to a thrombotic event^135,136. However, the exact rate is difficult to assess owing to the lack of prospective studies and the possible publication bias of available data. More than 45% of pregnancies in women with PNH result in either spontaneous miscarriage or termination¹³⁶. Of the women who give birth, >50% deliver prematurely^135,137, an event that can have negative implications for the health of the newborn baby. In one report, the mean weight of the infants was 2,800 g and the perinatal mortality rate was 8.8%¹³⁵. Eculizumab is listed as pregnancy category C risk (that is, risk not ruled out). Registry data of 75 pregnancies in 61 women with PNH on eculizumab therapy reported no maternal deaths, 3 fetal deaths and 6 miscarriages during the first trimester¹¹⁸. Most women also received prophylactic enoxaparin (an anticoagulant), but it is unclear whether it had any positive effect. Interestingly, most patients who progressed past the first trimester had their eculizumab dose increased owing to breakthrough of terminal complex blockade and consequent haemolysis. Low levels of eculizumab were detected in the cord blood in 7 of the 20 samples tested; however, the drug was not detected in any of the 10 breast-milk samples that were analysed. On the basis of these data, most haematologists agree that the benefit of eculizumab in pregnant women outweighs the potential risks of the drug, but a thorough discussion with the patient and obstetrician is necessary.

Quality of life

Patients with PNH and their health care providers naturally are concerned about the increased risks of end-organ damage and mortality. However, it is the non-fatal manifestations of PNH, in particular, severe fatigue and transfusion requirements, that truly have a substantial negative effect on patient quality of life. These conditions can even restrict the ability to perform everyday activities^2,138. Multiple retrospective series have described the varied symptoms that patients experience^2,25,103, but appropriate instruments for assessing disability and quality of life in patients with PNH are not well established.

The International PNH Registry gathers clinical symptoms and disease burden information from the patient’s perspective. Specific patient-reported symptoms that affect various measures, including quality of life and employment status, are reviewed at enrolment (baseline)¹³. Specifically, the Functional Assessment of Chronic Illness Therapy Fatigue subscale version 4 (FACIT-Fatigue)¹³⁹ and the European Organization for Research and Treatment of Cancer Quality-of-Life Questionnaire-C30 (REF. 140) assessment tools are collected from the patients. Low scores indicate increased levels of fatigue and poorer quality of life^141–143. A review of the available baseline data from the registry shows that, in the 856 patients considered (53% of the registered patients), 93.3% reported at least one symptom associated with PNH¹³, such as fatigue (80%), dyspnoea (64%), headache (63%) and haemoglobinuria (62%); 38% of men experienced erectile dysfunction. In 91.4% of cases, the patients reported at least one additional symptom besides fatigue, which was the only complaint in 2% of patients¹³. Another study in 29 patients with PNH in the United Kingdom further validated these two tools for the assessment of PNH-associated fatigue and health-related quality of life¹⁴⁴. The overall results demonstrated that the majority of patients regarded the FACIT-Fatigue items as relevant and applicable. As patients with PNH commonly experience fatigue, it is reassuring that the effects of this symptom can be captured well by the FACIT-Fatigue questionnaire¹⁴⁴.

Terminal complement inhibition improves quality of life in patients with PNH. Eculizumab was originally studied in a phase II pilot trial with the goal of treating PNH-associated haemolysis¹¹⁷. This initial trial demonstrated that lactate dehydrogenase levels in trans fusion-dependent patients with anaemia decreased as intravascular haemolysis was blocked with the drug¹¹⁷. These results were confirmed in large multicentre phase III studies (TRIUMPH and SHEPHERD)^115,116. These trials demonstrated a decrease in haemolysis and disappearance of many clinical symptoms of haemolysis, including fatigue, oesophageal spasm and erectile dysfunction in the patients with PNH on eculizumab therapy compared with the patients who received placebo. The beneficial effect of terminal complement inhibition continues to be demonstrated, as it improves patient quality-of-life metrics^10,145. However, one ongoing limitation of eculizumab therapy is the potentially life-long requirement for fortnightly infusions¹⁴⁴. These frequent interactions with the medical system could diminish long-term, health-related quality of life. As with any chronic disease, it is imperative that patients with PNH receive continued assessments of all disease manifestations and their effect on quality of life.

Outlook

Substantial progress in the biology and treatment of PNH has occurred over the past two decades, making PNH a model of what precision medicine should become. Thorough understanding of the molecular and cellular underpinnings of PNH has led to the development of a targeted therapy, eculizumab, which has altered the natural history of the disease. However, there is still room for improvement in the care of patients with PNH. Eculizumab is very expensive, must be administered intravenously every 2 weeks for life and up to 20% of patients still experience symptomatic extravascular haemolysis that requires periodic blood transfusions. The major problems that need to be addressed include mitigating extravascular haemolysis in patients on eculizumab and treating the rare patients who are unresponsive to eculizumab owing to mutant C5. Moreover, managing patients with severe bone marrow failure who are refractory to immunosuppressive therapy, and developing novel complement inhibitors that do not require intravenous administration or intravenous inhibitors with prolonged half-lives need to be addressed. Finally, there are still some unanswered basic and translational research questions about the mechanism (or mechanisms) of thrombosis in PNH and of clonal expansion.

C5 inhibitors

There are more than a dozen novel complement inhibitors in preclinical or clinical development¹⁴⁶. Because terminal complement blockade is safe and effective in patients, several of these compounds target C5. Alexion Pharmaceuticals has developed a new C5-specific monoclonal antibody, ALXN1210, that is virtually identical to eculizumab but with a longer half-life that might only need to be administered (intravenously) every 4–6 weeks. A phase III randomized trial of ALXN1210 is planned for 2017. Another strategy to block C5 cleavage was used by Alnylam pharmaceuticals (Cambridge, Massachusetts, USA) to develop a C5-specific N-acetylgalactosamine (GalNAc)-conjugated small interfering RNA duplex, ALN-CC5. This drug is administered subcutaneously and efficiently silences C5 production in the liver, thereby blocking terminal complement activation. Clinical trials are underway. Coversin, developed by Akari Therapeutics (New York, New York, USA) is derived from a small protein (16 kDa) isolated from the tick Ornithodoros moubata¹⁴⁷. The protein binds to C5 and blocks cleavage by C5 convertases. In vitro, the drug is effective in preventing haemolysis of PNH erythrocytes, even those isolated from patients carrying a poly morphism in C5 that makes eculizumab ineffective. Several other strategies to target C5 are in various stages of preclinical and clinical development. Although the results of these approaches are encouraging and exciting, it is unlikely that targeting C5 will solve the problem of extravascular haemolysis, as these therapies do not prevent the activation of C3 (FIG. 4).

C3 inhibitors

In the complement activation cascade, C3 is upstream of C5; thus, blockade at the level of C3 would theoretically prevent both intravascular and extravascular haemolysis. C3 has a pivotal role in all three complement pathways, which makes it a rational target, but is also the most abundant complement factor in the serum (1.2 mg per ml), which makes it a challenging target. Compstatin is a 13-residue disulfide-bridged peptide that binds to human C3 and its active fragment C3b, thereby preventing C3 cleavage and the incorporation of C3b into C3 and C5 convertases^148,149. A compstatin derivative from Apellis Pharmaceuticals (Crestwood, Kentucky, USA) is in clinical trials to treat PNH. Several other compstatin analogues are also being studied. Because these compounds would disable the classical, lectin and alternative pathways, their safety in the setting of infection and accumulation of immune complexes will need to be carefully tested.

Alternative pathway inhibitors

Complement factor D is a serine protease with only one known substrate, complement factor B. The single Arg–Lys bond of factor B becomes susceptible to the enzymatic activity of factor D only when it forms an Mg²⁺-dependent complex with C3b. Factor D is the only enzyme in the blood that can activate factor B and is, therefore, necessary for the activation of the alternative pathway. The plasma concentration of factor D (1.8 ± 0.4 μg per ml) is the lowest of any complement protein, which makes factor D the limiting factor in the activation cascade and a highly promising drug target. Achillion Pharmaceuticals (New Haven, Connecticut, USA) has synthesized a small molecule, ACH-4471, that inhibits factor D and can block PNH cell haemolysis and mitigate the accumulation of C3 fragments on the surface of PNH cells in vitro¹⁵⁰. Clinical trials evaluating oral ACH-4471 are underway. Other proteins of the alternative pathway (for example, factor B, factor H and properdin) are rational targets for treating PNH, and drug development targeting these proteins is underway.

Clinical trials over the next decade will determine which, if any, of these novel complement inhibitors is superior to eculizumab in safety, efficacy, cost and ease of administration. Trials exclusively recruit patients with haemolytic PNH, as these drugs are not expected to have an effect on bone marrow failure. Thus, bone marrow transplantation and immunosuppressive therapy will be the mainstay of treatment for patients with PNH who also have severe bone marrow failure. Fortunately, there is cause for optimism. Non-myeloablative HLA-haploidentical bone marrow transplantation with post-transplant high-dose cyclophosphamide to mitigate graft-versus-host disease has greatly expanded the pool of bone marrow donors. Consequently, bone marrow transplantation has become increasingly safe and effective for patients with non-malignant diseases, such as aplastic anaemia and PNH^151,152.