Paroxysmal Nocturnal Haemoglobinuria

Acquired clonal disorder of the haematopoietic stem cell caused by a somatic PIGA variant, with loss of glycosylphosphatidylinositol-anchored proteins, complement-mediated intravascular haemolysis, thrombophilia and bone marrow failure; it includes the PIGT-related forms (germline and somatic) and congenital CD59 deficiency.

Literature and regulatory information updated: 27 October 2026.

0. Initial Reference Table

Germline defects of GPI anchor biosynthesis (PIGA-CDG, multiple congenital anomalies–hypotonia–seizures syndromes, hyperphosphatasia with intellectual disability syndrome or Mabry syndrome, CHIME syndrome) are included in this article only as molecular context and differential diagnosis; they do not cause haemolysis and are not treated with complement inhibitors [7, 25]. Hereditary predisposition to myeloid neoplasms and inherited bone marrow failure syndromes, such as Fanconi anaemia and Diamond-Blackfan anaemia, are part of the differential diagnosis.

1. Introduction

1.1. Historical overview

Paroxysmal nocturnal haemoglobinuria (PNH) is an acquired haematological disorder in which the progeny of one or several haematopoietic stem cells carrying a somatic variant of the PIGA gene lose all proteins attached to the membrane by a glycosylphosphatidylinositol (GPI) anchor. The first clinical description is attributed to Paul Strübing in 1882, in a historical citation reproduced by the Onkopedia guideline [18]. The name of the disease reflects the observation of dark urine on waking, but this is an inconstant feature: in a classical series of 80 patients assembled by Dacie and Lewis, only 26% reported haemoglobinuria at presentation, whereas 35% presented with symptoms of anaemia and 13% with aplastic anaemia [1]. For decades diagnosis relied on functional lysis tests, the acidified serum lysis test (Ham test) and the sucrose lysis test, now abandoned because they are less sensitive and less quantitative than flow cytometry [1].

The clonal nature of the disease was suggested by the monoclonal origin of the abnormal erythrocytes, described by Oni and colleagues in 1970, and deficiency of decay-accelerating factor (DAF, CD55) on granulocytes, monocytes and platelets was demonstrated by Nicholson-Weller in 1985, both in historical citations from Onkopedia [18]. In 1990, Yamashina and colleagues described a Japanese patient with complete inherited deficiency of the 20 kDa homologous restriction factor (HRF20, later designated CD59) and a PNH-like picture, demonstrating that isolated absence of this terminal complement regulator is sufficient to reproduce almost all of the clinical features of the disease [3, 18]. In 1993, Kinoshita's group cloned the cDNA of PIG-A (phosphatidylinositol glycan class A), required for the synthesis of N-acetylglucosaminyl-phosphatidylinositol, the first intermediate of GPI anchor biosynthesis [26], and showed that granulocytes and B lymphocytes from patients share the same somatic PIG-A variant, located on the X chromosome, which explains why a single somatic variant is enough to express a recessive phenotype and why the same biosynthetic step is affected in all patients [27].

Follow-up of 80 consecutive patients referred to the Hammersmith Hospital, London, between 1940 and 1970, published by Hillmen and colleagues in 1995, defined the natural history without targeted therapy: median age at diagnosis 42 years, median survival 10 years, 28% surviving at 25 years, venous thrombosis in 39% and spontaneous clinical remission in 12 of the 35 patients who survived 10 years or more [28]. In 2005, the IPIG published the first international guideline on diagnosis and treatment, which introduced the classification into classical PNH, PNH in the setting of another bone marrow disorder and subclinical PNH, recommended flow cytometry as the diagnostic test and launched the International PNH Registry [1]. Approval of the anti-C5 monoclonal antibody eculizumab by the FDA (Food and Drug Administration) in March 2007 and by the European Commission in June 2007 changed the natural history: median survival, previously between 15 and 20 years, approaches, with appropriate treatment, that of age- and sex-matched controls [20, 21, 29, 30].

From 2013 onwards, rare forms with a germline component were described. Krawitz and colleagues identified a patient with PNH without a somatic PIGA variant, carrying a heterozygous germline splicing variant in PIGT and a somatic 8 Mb deletion in 20q that removed the other allele in granulocytes [31]. In 2019, Höchsmann and colleagues gathered four patients with the same mechanism and proposed the entity "PIGT-PNH", characterised by autoinflammatory symptoms dependent on complement and inflammasome activation [2]. In 2013, Nevo and colleagues described the founder CD59 variant p.(Cys89Tyr) in four Jewish families of North African origin with chronic haemolysis and relapsing demyelinating polyneuropathy in childhood [12]. The second decade of complement inhibition brought ravulizumab (2018–2019), pegcetacoplan (2021), iptacopan (2023–2024), danicopan (2024), crovalimab (2024) and two eculizumab biosimilars (2023–2024) [32–38].

1.2. Nomenclature and classification

The 2005 IPIG classification remains the common language and is adopted by the 2021 Brazilian consensus, the 2024 Onkopedia guideline, the 2018 ICCS/ESCCA (International Clinical Cytometry Society/European Society for Clinical Cell Analysis) cytometry guidelines and the 2026 Italian consensus [1, 17, 19, 20]. Classical PNH is defined by clinical and laboratory evidence of intravascular haemolysis (reticulocytosis, raised lactate dehydrogenase [LDH] and indirect bilirubin, low haptoglobin) without another defined bone marrow disorder; the marrow is cellular, with erythroid hyperplasia and normal or near-normal morphology, without non-random karyotypic abnormalities [1]. PNH in the setting of another specified bone marrow disorder (for example, PNH/aplastic anaemia or PNH/refractory anaemia-myelodysplastic syndrome) requires clinical or laboratory haemolysis and a concomitant or previous bone marrow disorder diagnosed by its own criteria [1]. Subclinical PNH (PNH-sc) designates small populations of GPI protein-deficient cells, detected by highly sensitive flow cytometry, without clinical or laboratory evidence of haemolysis, observed mainly in aplastic anaemia and in low-risk myelodysplastic syndrome (MDS) [1]. ICCS/ESCCA emphasises that distinguishing between categories can be difficult, because all patients with PNH have some degree of progenitor deficit, and that the classification, useful at diagnosis, does not by itself dictate therapy [19].

The 2026 Italian consensus adds an operational definition of disease activity: high disease activity (HDA) when LDH is greater than 1.5 times the upper limit of normal (ULN) in the presence of at least one PNH sign or symptom [20]. This definition was used as an inclusion criterion in several trials and in the European indication for ravulizumab and crovalimab ("haemolysis with clinical symptom(s) indicative of high disease activity") [22, 39]. In OMIM, the somatic PIGA form is PNH1 (300818) and the PIGT form is PNH2 (615399); the PNH1 entry describes PNH as an acquired, not inherited, genetic disorder, and the PNH2 entry presents the phenotype–gene relationship as provisional, because at the date of its last edit it was based on a single published patient [8, 10, 11, 14].

1.3. Entities within scope and boundaries

Classical PNH due to somatic PIGA accounts for almost all cases; it is an acquired disorder and is not transmissible to offspring [7, 10]. PIGT-associated PNH is a rare form requiring two events (one germline and one somatic) in an autosomal gene; it shares haemolysis and thrombosis with classical PNH and is distinguished by its autoinflammatory component [2, 31]. Congenital CD59 deficiency is a constitutional autosomal recessive disorder in which all cells of the body lack functional CD59 but retain the other GPI proteins, and which combines chronic haemolysis with immune-mediated neurological disease [9, 12]. Germline deficiency of PIGA and other GPI pathway genes (congenital disorders of GPI anchor glycosylation) shares the biochemical lesion but not the tissue or the timing: the defect is present in all cells from conception, manifests as epileptic encephalopathy and malformations, and a germline PIGA finding in a child with seizures does not mean PNH, does not predict it and is not an indication for complement inhibitors [7]. This distinction runs through the whole article.

2. Molecular Genetics and Pathophysiology

2.1. The GPI anchor and the PIGA gene

The GPI anchor is a glycolipid that attaches more than 150 different proteins to the cell surface, including enzymes, adhesion molecules, receptors and complement regulators; at least 27 genes take part in anchor biosynthesis and in the transport of anchored proteins [40, 41]. Synthesis takes place in the endoplasmic reticulum by sequential addition of sugars to phosphatidylinositol; the complete anchor is then transferred to the carboxyl terminus of nascent proteins by the GPI-transamidase complex, and the anchored proteins proceed to the plasma membrane [2]. PIGA encodes the catalytic subunit of the GPI-N-acetylglucosaminyltransferase complex, one of seven proteins required for the first committed step, the transfer of N-acetylglucosamine from UDP-GlcNAc to phosphatidylinositol [7, 25]. The rest of the pathway depends on this step: without PIGA activity no anchor is synthesised and no GPI protein reaches the surface [7].

PIGA is located at Xp22.2 (MANE Select transcript NM_002641.4) [6]. Because males have a single X chromosome and in females one X is inactivated in each cell, a single somatic loss-of-function variant is sufficient to abolish gene function in a stem cell, which explains why PNH is almost always caused by PIGA and not by the other pathway genes, all autosomal, which would require two events [27, 31]. Somatic variants are private and have no hotspots; identification of the variant is not necessary for diagnosis or treatment [7, 18]. Several clones with different variants may coexist in the same patient, and the progeny of each clone defines cell populations with complete or partial deficiency of GPI proteins [1, 20].

2.2. CD55, CD59 and the mechanisms of haemolysis

Among the proteins lost, two determine the clinical picture. CD55 (DAF) accelerates the decay of the C3 and C5 convertases on the cell surface; CD59 (membrane inhibitor of reactive lysis, MIRL) binds to the beta hairpins of the C8α and C8β chains in the assembling membrane attack complex (MAC, C5b-9) and prevents incorporation of the multiple copies of C9 that form the lytic pore [9, 17]. The alternative complement pathway is permanently active at a low level ("tick-over"); in erythrocytes lacking CD55 and CD59 the C3 convertase amplifies unchecked and the MAC forms freely, causing chronic intravascular haemolysis with exacerbations [17, 29]. Almost all specific symptoms of PNH have been described in the patient with isolated CD59 deficiency, which places this regulator at the centre of intravascular haemolysis [3, 18].

A patient's erythrocytes form a mosaic. Type I cells have normal expression; type II cells have partial expression (usually about 10% of normal), with moderate hypersensitivity to lysis (3 to 5 times normal); type III cells have complete absence, with hypersensitivity 15 to 25 times normal and a half-life of 10 to 15 days, compared with about 120 days for normal erythrocytes [1, 19, 42]. Partial expression of CD55 and CD59 protects type II erythrocytes from spontaneous lysis, so a patient with a predominance of type II cells may have a large clone with mild haemolysis, but suffer intense haemolysis when complement activation increases (infection, trauma, surgery, pregnancy) [1]. GPI-deficient granulocytes and monocytes are not lysed to a significant extent and have a normal half-life, so the percentage of deficient neutrophils is the best estimate of clone size [1, 19].

C5 inhibition blocks MAC formation but not the deposition of C3 fragments, which continues on erythrocytes lacking CD55. PNH erythrocytes surviving under anti-C5 accumulate C3b, iC3b and C3dg and are opsonised and removed by macrophages in the spleen and liver (mainly Kupffer cells); this extravascular haemolysis manifests as persistent anaemia, reticulocytosis, raised unconjugated bilirubin and a direct antiglobulin test (DAT) positive for C3d only [18, 29]. Extravascular haemolysis is clinically significant in about one third of patients on anti-C5 and is a quantitative difference, because C3 deposition is detectable by flow cytometry even in minimally symptomatic patients [18]. Variants in CR1 and C3 modulate its intensity [17, 18]. Proximal inhibitors (of C3, factor B or factor D) act upstream and control both intravascular and extravascular haemolysis [20].

2.3. Clonal expansion

The PIGA variant is not sufficient for the clone to expand. GPI-deficient cells are found in very low numbers in healthy individuals, detectable by high-resolution methods [18, 19]. In healthy controls, the PIG-A mutation rate in colony assays is about 1 in 50,000 (approximately 0.002%), and the frequency of PNH erythrocyte phenotypes in fresh peripheral blood is 2 to 6 per million erythrocytes; the diagnostic threshold of 0.01% lies 5 to 10 times above the maximum value observed in healthy controls [19]. In healthy people and in patients with MDS these cells derive from colony-forming cells, the variants are transient, non-clonal and do not involve T lymphocytes; in aplastic anaemia and PNH they derive from a multipotent stem cell and are clonal in all lineages [19]. The dominant hypothesis is that of a conditional survival advantage: the autoimmune attack that characterises aplastic anaemia preferentially targets normal stem cells, sparing GPI-deficient cells, which then sustain haematopoiesis [19, 29].

The studies collected in OMIM show that the PIGA variant alone is not sufficient. Araten and colleagues (1999) found granulocytes with a PNH phenotype in 9 healthy individuals, at a mean frequency of 22 per million, and erythrocytes with that phenotype at 8 per million; in cells sorted by flow cytometry PIGA variants were identified in 6 cases, and the same Tyr98Ter variant (published designation; corresponds to codon 98, tyrosine, of NM_002641.4, with no nucleotide change stated) previously described in a patient with PNH was found, confirmed in two samples taken 8 weeks apart, in a 61-year-old man without PNH undergoing phlebotomy for haemochromatosis [14]. Hu and colleagues (2005) showed that these variants of normal haematopoiesis arise in differentiated progenitors and not in stem cells [14]. Luzzatto and colleagues formulated the dual pathogenesis hypothesis: the clinical phenotype requires the somatic PIGA variant (factor A) and a particular form of bone marrow failure (factor B); A alone would give clones without clinical significance, and B alone the picture of aplastic anaemia [14]. As an alternative or complementary explanation, Brodsky and colleagues (1997) observed that granulocytes and CD34+ cells from patients with PNH survive longer in serum-free medium and resist apoptosis induced by ionising radiation, and that restoration of normal PIGA reverses this resistance [14]. In cells from patients with PNH, Horikawa and colleagues (2002) found T-lymphocyte colonies resistant to 6-thioguanine, a marker of HPRT variants, in 8 of 12 patients versus 3 of 17 age-matched controls, with a mean frequency of 149 versus 7 per 10⁷ cells, suggesting an environment favourable to diverse somatic variants [14].

Next-generation sequencing (NGS) has revealed complex clonal hierarchies, with additional somatic myeloid variants and immune escape variants in HLA genes [18]. By exome sequencing of cells with and without a PIGA variant from 12 patients and deep targeted sequencing in 36 others, Shen and colleagues (2014) identified in PIGA-mutant cells additional somatic variants in myeloid neoplasm genes (TET2, SUZ12, U2AF1, JAK2), some preceding the PIGA variant itself, pointing to stepwise clonal evolution from a single stem cell that may explain part of the variability in clinical course [14]. Clonal haematopoiesis variants such as JAK2 V617F (NM_004972.4:c.1849G>T p.(Val617Phe)) and CALR have been found in patients with PNH and may explain cases without previous aplastic anaemia [29, 43]. About 20% of patients with PNH have cytogenetic abnormalities at diagnosis, without leukaemic evolution in these patients in the series cited by Onkopedia [18]. When evolution to MDS or acute myeloid leukaemia occurs, monosomy 7 is the most frequent abnormality, and the monosomy 7 clone appears to replace the PNH clone rather than arise within it [44]. PIGA is on the gene list of the next-generation sequencing (NGS) myeloid panels recommended by ESMO, so somatic PIGA variants may appear as incidental findings in panels requested for other reasons [45].

2.4. Thrombosis, nitric oxide and smooth muscle

Free haemoglobin released by intravascular haemolysis reaches concentrations in PNH about 10 times higher than in haemoglobinopathies and scavenges nitric oxide (NO) as it is oxidised to methaemoglobin; arginase released from erythrocytes depletes arginine, the substrate for NO synthesis [18, 29]. NO depletion causes smooth muscle dystonia (oesophageal spasm with dysphagia and odynophagia, abdominal and lumbar pain, erectile dysfunction), vasoconstriction, endothelial dysfunction and platelet activation [1, 29]. These symptoms resolve within hours of starting eculizumab or ravulizumab, and the headache that appears after the first dose is attributed to cerebral vasodilatation from NO recovery [29]. Procoagulant erythrocyte and platelet microvesicles, inhibition of ADAMTS13 and complement activation on platelets lacking CD59 also contribute to thrombophilia; no single mechanism explains the phenomenon [18]. Thrombin activates C3 and C5 and plasmin activates C5, closing a loop between coagulation and complement and supporting immediate interruption of intravascular haemolysis when thrombosis is the first manifestation [18]. The propensity to thrombosis is approximately proportional to the size of the granulocyte clone [1].

2.5. PIGT-associated PNH

PIGT (20q13.12; MANE NM_015937.6) encodes an essential subunit of the GPI-transamidase complex, which cleaves the carboxyl signal of the nascent protein and attaches the preassembled anchor to it [8, 31]. In PIGT deficiency the anchor is synthesised but not transferred, and free GPI, not bound to protein, reaches the membrane; in PIGA deficiency no anchor is produced [2]. In the four patients of the 2019 series (one Japanese, J1, and three German, G1, G2 and G3) there was a germline loss-of-function variant in one PIGT allele, always the maternal allele, and a somatic 8 to 18 Mb deletion in 20q on the paternal allele, encompassing all of PIGT and the myeloid common deleted region (CDR, about 1.9 Mb between PTPRT and OSER1) [2]. This CDR contains the maternally imprinted genes L3MBTL1 and SGK2, expressed only from the paternal allele; the paternal deletion abolishes their expression, a clonal expansion mechanism similar to that of myeloproliferative neoplasms with 20q−, in contrast to the conditional immunological advantage of PIGA-related PNH [2]. There was no documented bone marrow failure in these patients [2]. In the first published case (G1), the somatic 8 Mb deletion spanned 20q11.23-q13.12 in a large proportion of granulocytes, and the germline splicing variant, transfected into CHO (Chinese hamster ovary) cells lacking Pigt, only minimally increased CD55 expression and hardly restored that of CD59, confirming loss of function [11].

Free GPI on the surface of PIGT-deficient cells has an effect of its own. In THP-1 monocytic cells with inactivated PIGT, free GPI activated the lectin complement pathway, C5b-9 formation was much greater than in PIGA-deficient cells, and there was inflammasome-dependent secretion of interleukin 1β (IL-1β) mediated by the MAC and not by C5a [2]. This mechanism explains the autoinflammatory picture (urticaria, arthralgia, fever, aseptic meningitis with neutrophil predominance in the cerebrospinal fluid), comparable to that of the cryopyrinopathies caused by gain-of-function NLRP3 variants, and explains why eculizumab prevents both haemolysis and inflammatory attacks [2]. Interleukin 18 remained 4 to 5 times above the ULN in J1 despite eculizumab, indicating a pathway independent of terminal complement activation [2]. The PIGU gene, another transamidase component, is located about 7.4 Mb centromeric to the CDR and was included in the deletion in J1 and G2, which led the authors to consider the possibility of autoinflammatory PNH caused by PIGU [2].

Germline PIGT variants found in inherited GPI deficiency (IGD; multiple congenital anomalies–hypotonia–seizures syndrome type 3, MCAHS3) are hypomorphic in the homozygous or compound heterozygous state and do not cause haemolysis or inflammation, whereas in PIGT-PNH both the germline and the somatic alleles are null or near-null [2]. The same variant c.250G>T, p.(Glu84Ter), from J1 was found in the heterozygous state in the healthy mothers of two Japanese patients with IGD, without inflammatory symptoms, indicating that haploinsufficiency is not sufficient and that disease begins only after somatic loss of the second allele [2]. In MCAHS3, the same allele (E84X in the publication's notation) was described in compound heterozygosity with hypomorphic missense variants, R488W (published designation; the variant described, c.1342C>T, corresponds in NM_015937.6 to p.(Arg448Trp)) in a Japanese girl [43, 46] and G366W (NM_015937.6:c.1096G>T p.(Gly366Trp)) in a Japanese infant with acetazolamide-responsive epileptic apnoea and very low alkaline phosphatase; in PIGT-deficient CHO cells, E84X had no activity and p.(Arg448Trp) partially restored expression of GPI proteins [47].

2.6. Congenital CD59 deficiency

In primary CD59 deficiency (11p13; MANE NM_000611.6), C5b-9 is deposited unopposed on erythrocytes, causing intravascular haemolysis, and on the nodes of Ranvier and the endothelium, causing demyelinating neuropathy and strokes [9]. Unlike PNH, all cells lack functional CD59 and all other GPI proteins (including CD55) are present [9]. The founder variant p.(Cys89Tyr) produces protein in reduced quantity, with failure of surface localisation [12]. In functional studies of the four variants described, all produced non-functional protein, without MAC inhibition and with increased lysis; the two missense products reached the surface and the two frameshift products were retained inside the cell, and none was recognised by the available anti-CD59 antibodies [9]. The original Japanese patient was homozygous for single-base deletions which, in the mature protein numbering used by the authors, affected amino acids 16 and 96 [48]. In a girl with CD59 deficiency treated with eculizumab, Höchsmann and colleagues (2014) identified a homozygous truncating variant [5]. In contrast to the human phenotype, Cd59-/- mice obtained by homologous recombination were healthy and fertile and had no anaemia, but had reticulocytosis, increased free haemoglobin in plasma and urine and erythrocytes that were complement-sensitive and positive in the acidified serum lysis test, with worsening intravascular haemolysis after complement activation by cobra venom factor [5].

2.7. Variant spectrum and nomenclature

Variant G2 was published as c.761_764delGAAA; the HGVS 3′ alignment rule shifts it to c.766_769del, which is the form to use in reports, citing the original notation [2, 6]. The 2019 paper refers to transcript NM_015937 without a version; the correspondence with NM_015937.6 results from normalisation in VariantValidator [6]. For CD59, the normalised form is p.(Cys89Tyr); GeneGuidance curation uses elsewhere the designation p.Cys64Tyr for the same variant, whereas the primary sources and ClinVar use p.(Cys89Tyr), which is adopted here [9, 49]. Germline PIGA variants causing neurodevelopmental disease, such as the recurrent NM_002641.4:c.1234C>T p.(Arg412Ter) (R412X in OMIM), found in families with lethal MCAHS2, including the family published in 1995 as a severe form of Simpson-Golabi-Behmel syndrome, and the recurrent PIGT allele NM_015937.6:c.1582G>A p.(Val528Met) in MCAHS3 are published without a transcript; the correspondence with MANE transcripts is that of ClinVar [7, 8, 43, 51, 52]. In PIGA-deficient cells, the R412X protein retained some activity and partially restored GPI proteins, so it is not a null allele, consistent with the rule that only hypomorphic germline alleles are compatible with life [52].

2.8. C5 pharmacogenetics and other modifiers

Among 345 Japanese patients treated with eculizumab, 11 had a poor response, all heterozygous for C5 c.2654G>A, p.(Arg885His); the prevalence of the variant was 3.2% in patients and 3.5% in healthy Japanese individuals, and it was also found in Han Chinese; an Argentinian patient of Asian ancestry had the p.(Arg885Cys) variant [50]. The mutant C5 is functional and causes haemolysis in vitro, but does not bind eculizumab; it was blocked by an antibody binding a different epitope [50]. Ravulizumab shares the eculizumab epitope and is equally ineffective in these patients; crovalimab binds a different C5 epitope and is effective in Arg885 variants [18]. The recycling property of crovalimab results from pH-dependent binding to C5 and to the neonatal Fc receptor (FcRn), with a high isoelectric point of the immune complex, and that of ravulizumab from amino acid substitutions in the C5-binding region and in the CH3 domain, which raise the terminal half-life from 11.3 days (eculizumab) to 49.7 days (ravulizumab) [18]. A CR1 polymorphism is associated with lower eculizumab efficacy, and CR1 and C3 variants modulate extravascular haemolysis; genotyping of these variants is not routinely available to predict response [17, 18].

2.9. Germline defects of the GPI anchor: context and allelic disorders

The biochemical lesion of PNH also exists as a germline disease, but with different consequences, because it is present in all cells from conception and only hypomorphic alleles with residual activity are compatible with life [7]. These disorders are summarised here for their allelic relationship with PNH1 and PNH2 and their usefulness in differential diagnosis.

In the assessment of these disorders, plasma alkaline phosphatase may be raised (because it is not anchored to the membrane, as in HPMRS) or low (in most cases of PIGT-CDG), and flow cytometry for CD16, CD24, CD55, CD59 and FLAER on granulocytes, the same technique used for PNH, demonstrates the anchoring defect but does not identify the gene; diagnosis is made by exome or genome sequencing [7, 8, 41, 77]. Restoration of CD59 and CD55 expression and of FLAER binding in transfected deficient cells is used as a functional assay to classify variants in GPI pathway genes, for example PIGN [60].

The nomenclature of these disorders is in transition. OMIM retains clinical names (MCAHS1 to MCAHS4, MCAHS4 being caused by PIGQ, OMIM 618548; HPMRS1 to HPMRS6; numbered developmental and epileptic encephalopathies) and in parallel assigns each entity a number in the GPI biosynthesis defect series (GPIBD, with the discussion of heterogeneity in GPIBD1, OMIM 610293) [62, 65, 75]. In the edition consulted, this series runs from GPIBD1 (PIGM) to GPIBD25 (C18orf32) and includes, in addition to the genes in the table, PGAP1, PIGG, PIGP, GPAA1, PIGC, PIGH, PIGU, PIGK, ARV1 and PIGF [72, 73]. Functional analysis of PGAP2 variants suggested that defects of the early steps of the pathway tend to cause more severe phenotypes, with dysmorphism and malformations, than those of the late remodelling steps, such as PGAP2 and PGAP3, although variability within each gene is wide [68]. The review by Knaus and colleagues (2018) concluded that the separation between MCAHS and HPMRS may be artificial, because it found no clear correlation between the level of alkaline phosphatase or the degree of GPI protein deficiency on flow cytometry and neurological severity, variant class or gene, and proposed classifying all these disorders together as GPI biosynthesis defects; the same review described distinct facial patterns, detectable by computer-assisted comparison, in patients with PIGN, PIGA, PIGT, PIGV and PGAP3 variants [62, 65].

3. Clinical Spectrum and Manifestations by System

3.1. Presentation and natural history without complement inhibitors

PNH presents as haemolytic anaemia, as a form of bone marrow failure, as thrombophilia or as any combination of these three components, and the proportion of each varies from patient to patient and over time in the same patient [18, 29]. The IPIG (International PNH Interest Group) classification organises this heterogeneity into three subcategories: classical PNH, with clinical and laboratory intravascular haemolysis (reticulocytosis, raised LDH and indirect bilirubin, low haptoglobin) and a cellular marrow with erythroid hyperplasia, normal or near-normal morphology and no non-random karyotypic abnormalities; PNH in the setting of another specified bone marrow disorder, such as aplastic anaemia, myelodysplastic syndrome (MDS) or myelofibrosis, with haemolysis; and subclinical PNH, with small populations of cells deficient in GPI-anchored proteins detected by high-sensitivity flow cytometry, without clinical or laboratory evidence of haemolysis, observed mainly in aplastic anaemia and refractory anaemia-MDS [1, 17].

The initial presentation rarely matches the picture the name describes. In a series of 80 patients assembled by Dacie and Lewis and reproduced in the IPIG guideline, the presenting features were distributed as shown in the following table; thrombosis as the first manifestation was rare, about 5%, and gastrointestinal complaints (episodic abdominal pain, dysphagia or odynophagia) heralded the disease in about 10% [1].

Source: IPIG guideline, Table 4, modified from Dacie and Lewis [1].

Over the course of the disease, the range of complaints widens. The series cited by the 2026 Italian consensus record asthenia in 80% of patients, dyspnoea in 64%, haemoglobinuria in 62%, abdominal pain in 44%, erectile dysfunction in 38%, chest pain in 33% and dysphagia in more than 10%, signs which, being non-specific, delay diagnosis and treatment [20]. The Orphanet phenotypic annotation, expressed as frequency classes, considers very frequent (80% to 99%) haemolytic anaemia, haemoglobinuria and asthenia; frequent (30% to 79%) thromboembolism, deep vein thrombosis, reticulocytosis, dyspnoea, headache, episodic abdominal pain, chest pain, haemosiderinuria, unconjugated hyperbilirubinaemia, conjunctival icterus, reduced haptoglobin, raised LDH, decreased serum iron, raised urea and chronic kidney disease; occasional (5% to 29%) acute kidney injury and renal failure, proteinuria, erectile dysfunction, hypertension, stroke, myocardial infarction, pulmonary embolism, Budd-Chiari syndrome, mesenteric venous thrombosis, arterial thrombosis, odynophagia, thrombocytopenia, leucopenia and pancytopenia; and very rare (1% to 4%) dysphagia, oesophageal spasms, glycosuria and renal Fanconi syndrome [10]. The two sources are not directly comparable: the first summarises clinical cohorts and the second curated frequency classes, which explains, for example, dysphagia in more than 10% in one and in the 1% to 4% class in the other.

The natural history without targeted therapy has been described in two European cohorts. At the Hammersmith Hospital, London, 80 consecutive patients referred between 1940 and 1970 and treated only with supportive measures (transfusions and oral anticoagulation after established thrombosis) had a median age at diagnosis of 42 years (range 16 to 75); median survival after diagnosis was 10 years, 22 patients (28%) survived 25 years, 60 died and, of the 48 with a known cause of death, 28 died of venous thrombosis or haemorrhage; 31 patients (39%) had at least one episode of venous thrombosis, 12 of the 35 who survived 10 or more years had spontaneous clinical recovery and none developed leukaemia [28]. In the French Society of Haematology registry, with 220 patients, the largest natural history cohort available at the time of the IPIG guideline, median survival was 12 years, and worse survival was associated with transformation to MDS or acute leukaemia (relative risk, RR, 19.1), thrombosis at any site (RR 10.2), progression to pancytopenia (RR 5.5), age over 55 years (RR 4.0) and thrombocytopenia at diagnosis (RR 2.2) [1]. The pooled data cited by the Brazilian consensus place the median age at diagnosis in the early to mid-30s, with no difference between sexes, and mortality with best supportive care at 35% at 5 years and about 50% at 10 years; median survival before 2007 was between 15 and 20 years, with thrombosis as the main cause of death [17, 29]. These figures come from patients referred to hospital centres in eras without high-sensitivity flow cytometry or complement inhibitors, and do not describe current prognosis.

Clinical severity generally correlates with clone size in untreated patients, but the correlation is imperfect: there are minimally symptomatic patients with large clones [17]. Part of the discrepancy is explained by phenotypic mosaicism. Type III erythrocytes (complete absence of GPI proteins) are lysed spontaneously, whereas partial expression of CD55 and CD59 on type II cells is enough to protect them from spontaneous lysis in vivo; a patient with a predominance of type II cells may have a large clone with mild haemolysis, which becomes intense when complement activation increases (infection, trauma, surgery, pregnancy or another stress situation) [1]. Ancestry also modulates the phenotype: bone marrow failure manifestations predominate in Asian, Pacific Islander and Hispanic patients, and thrombosis and infections in American and European patients, for unknown reasons [1].

3.2. Intravascular haemolysis and anaemia

Coombs-negative haemolytic anaemia is the clinical hallmark of PNH [1]. Absence of CD59 leaves the clone's erythrocytes defenceless against formation of the membrane attack complex, and absence of CD55 accelerates the deposition of C3 fragments; haemolysis is intravascular, complement-mediated and non-immune, with a negative direct antiglobulin test [17, 29]. Consequently, all patients with clinically significant haemolysis have raised serum LDH, so a normal LDH makes a clinical form of PNH unlikely, although it does not exclude the subclinical form, which by definition has no haemolysis [1]. The full laboratory profile includes anaemia, reticulocytosis, raised LDH and indirect bilirubin, decreased haptoglobin, free plasma haemoglobin (pink or red serum), haemoglobinuria with a dipstick positive for haem and a sediment without erythrocytes, and haemosiderinuria [10, 17]. Free haemoglobin in plasma often reaches levels 10 times higher than in haemoglobinopathies [29].

Haemoglobinuria, which gives the disease its name, is inconstant and rarely nocturnal. Only about 26% of patients report dark brown morning urine at the time of diagnosis; many have no apparent haemoglobinuria or have intermittent episodes unrelated to the day-night rhythm, and its occurrence correlates with clone size [1, 18]. Practically all patients with classical PNH report macroscopic haemoglobinuria at some point in the disease, but this sign may be absent in PNH associated with aplastic anaemia or refractory anaemia-MDS, in which the clone is often small [1]. The relative weight of haemolysis in the anaemia depends on the proportion and type of the affected cells: in a hypothetical example from the IPIG guideline, a patient with 15% deficient erythrocytes would have a modest haemolytic contribution, and another with 75% would have a significant haemolytic component [1].

The anaemia of PNH is multifactorial. It results from the combination of erythrocyte destruction, insufficient production due to underlying bone marrow failure, iron deficiency and other marrow insults [1, 17]. Iron deficiency is frequent because chronic intravascular haemolysis causes high urinary iron loss, in the form of haemoglobinuria and haemosiderinuria [1]. Severity ranges from minimal to very severe [17]. The symptoms of chronic haemolytic anaemia are weakness, fatigue and exertional dyspnoea, but the intensity of fatigue is not linearly proportional to the degree of anaemia, correlating instead with haemolytic activity and clone size [18].

Haemolytic crises are superimposed on chronic haemolysis. They are accompanied by fatigue, lumbar and abdominal pain, headache and fever, sometimes by intense myalgia, and are precipitated by infections, surgery, trauma, transfusions, pregnancy or other situations that amplify complement activation [1, 17, 18]. Depletion of nitric oxide (NO) by free haemoglobin, which metabolises it to methaemoglobin, and by arginase released from erythrocytes, which reduces the arginine available for its synthesis, links haemolysis to endothelial dysfunction, platelet activation, smooth muscle dystonia and fatigue [18]. A published example of the florid haemolytic picture is a 44-year-old man with bright red haemoglobinuria, fatigue, chest pain, exertional dyspnoea and difficulty swallowing liquids and solids after mowing the lawn on a hot day, with dark morning urine on most days, dull low back pain and transient impotence; he had haemoglobin 6.7 g/dL, leucocytes 3.8 × 10⁹/L, platelets 98 × 10⁹/L, reticulocytes 287,000/mm³ and LDH 2800 IU/L, with a negative direct antiglobulin test, had received 6 units of red cell concentrate in the previous month and the clone was 18% in erythrocytes, 85% in granulocytes and 87% in monocytes [29]. The discrepancy between the erythrocyte and granulocyte clones is characteristic: PNH granulocytes have normal survival and reflect the true clone size, whereas PNH erythrocytes are destroyed and diluted by transfusions [1].

3.3. Thrombosis

Thromboembolism is the most clinically relevant complication of PNH and the main cause of the increased morbidity and mortality [18]. Before 2007, the year eculizumab was approved, thrombosis was the main cause of death, accounting for 40% to 67% of deaths; the first thrombotic event can be fatal and increases the risk of death 5- to 10-fold [17, 29]. In the Hammersmith cohort, 39% of patients had at least one episode of venous thrombosis, and in the French registry thrombosis at any site was associated with a relative risk of death of 10.2 [1, 28].

Thrombosis in PNH is predominantly venous and frequently occurs at atypical sites. The abdominal veins (hepatic, with Budd-Chiari syndrome, portal, splenic and mesenteric) are the most common sites, followed by the cerebral veins and the dermal veins [1, 29]. Arterial thrombosis is less frequent than venous thrombosis, but has been described in the cerebral, coronary, visceral and retinal territories, with myocardial infarction and stroke; in a Korean cohort arterial occlusions accounted for 39% of all thromboembolic events, a figure Onkopedia considers unusually high compared with other cohorts, and PNH should be considered in an arterial event without previous vascular disease [17, 18]. Mesenteric thrombosis may cause intestinal infarction requiring surgery, and some episodes of colicky abdominal pain are attributed to mesenteric vessel thrombosis and vascular spasm [1]. Abdominal and chest pain, dyspnoea and haemoglobinuria are early indicators of an impending thromboembolic event [18].

The probability of thrombosis is related to clone size measured in granulocytes. In the study by Hall and colleagues, the 10-year risk of thrombosis was 44% in patients with more than 50% GPI protein-deficient granulocytes, compared with 5.8% in patients with less than 50%; in a logistic regression model, Moyo and colleagues calculated an odds ratio of 1.64 for each 10% increase in the percentage of deficient granulocytes, corresponding to an 11.8-fold higher risk with a clone above 70% compared with a 20% clone [1]. Thrombosis is more frequent with large clones but also occurs with smaller clones [29]. Patients with thrombosis generally have large clones and clinically apparent haemolysis, with the exception of patients with a predominance of type II cells, who may have a large clone with mild spontaneous haemolysis [1]. These estimates come from retrospective analyses from the era before complement inhibitors [1].

Ancestry modifies the risk independently of clone size. Even after adjustment for the clone, the prevalence of thromboembolism in patients from Japan, China and Mexico is about 2%, lower than that observed in the United States and Europe, and the risk is higher in white and African American patients than in patients of Asian, Pacific Islander or Hispanic ancestry [1]. This difference contrasts with the high proportion of arterial occlusions in the Korean cohort, showing that the distribution between venous and arterial territories and the overall frequency of thrombosis do not necessarily vary in the same direction [18].

The mechanism is multifactorial and no single factor explains it. The factors identified are endothelial dysfunction due to NO depletion, vasoconstriction with possible tissue factor expression, procoagulant microparticles released by platelets and erythrocytes under complement attack, inhibition of ADAMTS13, deficiency of proteins involved in fibrinolysis and increased pro-inflammatory and prothrombotic cytokines [17, 18]. The observation that C5 inhibition substantially reduces thrombotic events, together with the accumulated laboratory and clinical evidence, establishes complement as the main driver of thrombosis in PNH [17, 29]. A published case illustrates thrombotic presentation without haemoglobinuria: a 45-year-old woman with no previous history, with three weeks of intermittent colicky right upper quadrant abdominal pain, fatigue, dyspnoea and pleuritic pain, initially assessed for suspected acalculous cholecystitis, had right pulmonary embolism and hepatic and splenic vein thrombosis with collaterals; haemoglobin was 7.7 g/dL, leucocytes 3.7 × 10⁹/L, platelets 102 × 10⁹/L, reticulocytes 181,000/mm³ and LDH 647 IU/L (reference 119 to 226), and the clone was 60% in erythrocytes, 88% in monocytes and 94% in granulocytes [29]. Antithrombotic management and the indications for prophylaxis are given in section 7.3.

3.4. Kidney, lung, smooth muscle and fatigue

The kidney is damaged both acutely and chronically. Renal dysfunction is found in two thirds of patients with PNH and, in 21%, is already at an advanced stage [18]. Two components are distinguished: a reversible dysfunction, an expression of afferent arteriolar vasospasm linked to increased NO catabolism in intravascular haemolysis, and a parenchymal lesion with tubular damage, whose correlates are haemosiderin deposition in the proximal tubules and microvascular thrombosis; clinically, tubular dysfunction and a gradual decline in creatinine clearance predominate [18]. Severe acute haemolytic crises may cause acute renal failure through direct haem toxicity, and chronic haemolysis may lead to renal haemosiderosis with proximal tubular dysfunction, interstitial scarring and cortical infarcts [17, 20]. In the Orphanet annotation, chronic kidney disease and raised urea are frequent, acute kidney injury, renal failure and proteinuria occasional, and glycosuria and renal Fanconi syndrome very rare [10].

Dyspnoea is frequent in haemolytic PNH and does not depend on anaemia alone [18]. NO consumption by intravascular haemolysis may lead to pulmonary hypertension, detected by elevation of B-type natriuretic peptide (pro-BNP), used as a marker of right ventricular dysfunction; in a study from the same group, echocardiography or cardiac magnetic resonance imaging showed right ventricular dysfunction in 8 of 10 patients, and perfusion scintigraphy showed thromboembolic-type defects in 4 of these 8, with no evidence of thromboembolic events in the other 4 [18]. Pulmonary hypertension thus results from two pathways, recurrent embolism and the direct effect of NO consumption [18]. The Brazilian consensus emphasises that clinically significant pulmonary hypertension is rare and, when it occurs, is often associated with pulmonary embolism [17].

Smooth muscle dystonia is common and is linked to NO depletion by free haemoglobin [17, 29]. It manifests as abdominal and lumbar pain, oesophageal spasm with dysphagia and odynophagia, especially during haemolytic exacerbations, chest pain, nausea and erectile dysfunction or priapism, and some patients describe symptoms resembling a transient ischaemic attack during severe haemolytic crises, attributed to vasoconstriction of central nervous system vessels [1, 17, 18, 29]. These vascular symptoms often resolve within a few hours of administration of eculizumab or ravulizumab, and the same mechanism explains the frequency of headache after the first dose of these drugs, attributed to cerebral vasodilatation, which does not recur with subsequent doses and is not prominent when complement inhibition is used in other diseases [29]. Recurrent colicky abdominal pain may be disabling and, in some patients, corresponds to mesenteric thrombosis [1].

Fatigue is one of the central manifestations for the patient, reported by more than 80% in several cohorts [18]. Its intensity is not proportional to the degree of anaemia; patients with high absolute reticulocyte counts complain more, which points to haemolytic activity, and complement activation itself is another proposed explanation, although the mechanism has not been clarified [18]. Fatigue manifests as physical and cognitive limitation, with a state patients describe as "switching off", compromises work capacity, family plans and social life, and may set in gradually, with the patient becoming accustomed to reduced capacity; it is therefore recommended to record it from the start of treatment with the FACIT-Fatigue questionnaire (Functional Assessment of Chronic Illness Therapy-Fatigue) and the AA/PNH QLQ-54, the latter not yet validated by the IPIG [18]. The Italian consensus includes in the initial assessment of disease activity haemoglobinuria, abdominal pain, dysphagia, erectile dysfunction and fatigue measured by FACIT, the EORTC (European Organisation for Research and Treatment of Cancer) questionnaire or the PNH/HMPRO scale [20].

3.5. Bone marrow failure, aplastic anaemia-PNH syndrome and clonal evolution

PNH and acquired aplastic anaemia are linked in pathophysiology and clinically, because immune bone marrow failure favours the survival of GPI-deficient stem cells [18, 29]. Small, clinically silent PNH clones are found in up to 70% of adults and 10% of children with acquired aplastic anaemia, but not in inherited aplastic anaemia; Onkopedia states that more than 50% of patients with aplastic anaemia have, already at diagnosis, a small or moderate GPI-deficient population detectable by high-sensitivity flow cytometry, which is considered an indicator of immune-mediated bone marrow failure [18, 19, 29]. Because PNH is almost never associated with inherited bone marrow failure (Fanconi anaemia, dyskeratosis congenita), detection of a clone also helps to rule out these forms [29].

The transition from aplastic anaemia to clinical PNH is frequent and late. The risk of clinical PNH in a patient with acquired aplastic anaemia is 15% to 25%, depending on the size of the initial clone, and 10% to 20% of patients with aplastic anaemia develop overt haemolytic PNH, often many years after immunosuppression has ended [18]. The Italian consensus describes clones mostly of medium size that may grow and evolve to classical PNH in more than 40% of cases, and in a series of 131 patients with aplastic anaemia and a PNH clone there was no difference in haematological response to immunosuppression between expanding and stable clone patterns [20]. From the reverse perspective, up to 40% of PNH cases evolve from aplastic anaemia [19]. Allogeneic haematopoietic stem cell transplantation (HSCT) in aplastic anaemia prevents secondary PNH [18]. According to OMIM, 10% to 57% of patients with severe aplastic anaemia in remission after antithymocyte globulin and ciclosporin develop PNH; in the four patients studied by Nagarajan and colleagues (1995), the Ham test was negative at diagnosis of aplastic anaemia and became positive up to 3 months after immunosuppression in two, at 6 months in one and at 18 months in another, all had a PIGA variant, and all remained transfusion-independent and free of thrombosis over a mean follow-up of 30 months, which led the authors to attribute the apparently indolent course of this secondary PNH mainly to early detection [14]. In the United Kingdom population study, 12 of the 175 patients with aplastic anaemia and a PNH clone (6.8%) had disease evolution: 7 to symptomatic haemolytic PNH, 1 to thrombotic disease associated with an increase in type II cells, 3 to MDS and 1 to acute myeloid leukaemia; evolution was concentrated in patients aged 40 years or more (10 of 120, 8.3%, versus 2 of 55, 3.6%, below 40 years), and none of the patients presenting with classical haemolytic PNH developed a haematological neoplasm [16].

Clone size differentiates the subcategories. In the International PNH Registry, the median granulocyte clone in the first 1600 patients was 68.1%, with 83% in classical PNH versus 35% in aplastic anaemia; among patients with PNH in the setting of bone marrow failure, 34% had clones of 50% or more and 40% clones below 10%, whereas in patients without bone marrow failure 76% had clones above 50% and only 8% below 10% [19]. Patients with small clones, below 10% deficient granulocytes, have little or no clinical evidence of haemolysis and, in most cases, aplastic anaemia or refractory anaemia-MDS [1]. In the United Kingdom population study, 88% of the 197 patients with detectable clones above 0.01% in at least two lineages had aplastic anaemia, 8% classical PNH and 3% MDS [16]. Bone marrow failure manifests as cytopenias and haemorrhage (haemorrhagic signs in 18% of the Dacie and Lewis series), and worsening anaemia with reticulocytopenia or a new cytopenia point to bone marrow failure rather than haemolysis [1, 20].

Evolution to a myeloid neoplasm is possible but is not the rule. In the Hammersmith cohort no patient developed leukaemia, but in the French registry transformation to MDS or acute leukaemia was the worst prognostic factor (RR 19.1), followed by progression to pancytopenia (RR 5.5) and thrombocytopenia at diagnosis (RR 2.2) [1, 28]. About 20% of patients with PNH have cytogenetic abnormalities already at diagnosis, without this leading to leukaemic evolution in these patients, in a retrospective series [18]. The distinction between PNH with limited marrow regeneration and PNH in the setting of MDS is not reliable on morphology alone, so MDS in haemolytic PNH is diagnosed mainly by cytogenetics; abnormalities of chromosomes 5q, 7 and 20q are associated with MDS, and in MDS-associated PNH the Italian consensus requires an NGS panel of myeloid genes to classify and stratify risk [1, 18, 20]. Monosomy 7 is the most frequent somatic cytogenetic abnormality in MDS or acute myeloid leukaemia secondary to acquired aplastic anaemia or PNH, and frequently occurs in patients who started with a predominant PNH clone [44].

Outside aplastic anaemia, PNH clones arise in other bone marrow disorders. Clonal haematopoiesis mutations, such as JAK2 p.(Val617Phe) or CALR, have been found in patients with PNH and explain cases without previous aplastic anaemia, and PNH clones were identified in 4.8% of a small series of 21 patients with chronic myeloproliferative neoplasms [20, 29]. In the elderly patient, MDS is much more frequent than aplastic anaemia, and a clone of a few per cent is unusually large for MDS: in a published case, a 68-year-old woman treated with azacitidine for refractory MDS, with a 5.3% neutrophil clone, trisomy 8 and a consistently hypocellular marrow, almost completely normalised her counts after switching to immunosuppression, and trisomy 8, a non-specific abnormality, did not exclude an immune mechanism [19].

3.6. The disease under complement inhibition

Complement inhibition has changed the natural history, and the natural history series no longer represent the expected course in a treated patient. Terminal inhibition at C5 is highly effective against intravascular haemolysis and practically eliminates the risk of thrombosis, but does not treat bone marrow failure [29]. With adequate inhibition, survival approaches that of age- and sex-matched controls [20, 29]. Smooth muscle dystonia symptoms often resolve within hours of the first administration, LDH and fatigue fall in the first week, quality of life and dyspnoea improve in 2 to 3 weeks and transfusions decrease in 2 to 6 months [17, 29]. Comparative data from the pivotal trials are given in section 7.8.

The treated disease has its own manifestations. Lack of CD55, which acts upstream of C5, allows C3 fragments (C3b, iC3b, C3dg) to accumulate on PNH erythrocytes protected from intravascular lysis, which are opsonised and removed in the spleen and liver; this extravascular haemolysis occurs in practically all patients on eculizumab or ravulizumab, to a variable degree, and manifests as persistent mild to moderate anaemia, reticulocytosis, a rise in unconjugated bilirubin and a direct antiglobulin test positive for C3d only [17, 20, 29]. More than half of patients on C5 inhibitors have persistent mild to moderate symptoms, up to 20% still need occasional transfusions and about a quarter remain transfusion-dependent, extravascular haemolysis being the most common cause of persistent anaemia [17, 29]. Fatigue persists in some patients on C5 inhibitors, especially in those who develop clinically significant extravascular haemolysis, and data with proximal inhibitors show better fatigue scores when haemoglobin normalises [18].

Breakthrough haemolysis is the other characteristic manifestation of the treated disease: return of PNH symptoms (haemoglobinuria, abdominal pain or others) with a rise in LDH to 2 or more times the upper limit of normal, after a previous fall below 1.5 times on therapy [29]. It may be pharmacokinetic, due to insufficient drug concentrations with detectable free C5, which occurs regularly at the end of the dosing interval, or pharmacodynamic, due to conditions that amplify complement (infection, trauma, surgery, advanced pregnancy) despite adequate levels; it ranges from mild, with haemoglobinuria as the only sign, to severe, requiring transfusion and adjustment of the inhibitor [20, 29]. The 44-year-old patient described in section 3.2, maintained on eculizumab 900 mg every 2 weeks, continued to need 1 to 2 units of red cell concentrate every 4 to 6 weeks, had dark urine and return of impotence and fatigue one or two days before each infusion, with LDH of 302 IU/L after induction and 732 IU/L before the last infusion; with ravulizumab, the crises ceased and he became transfusion-independent, but had persistent mild fatigue, haemoglobin 9.4 g/dL, reticulocytes 265,000/mm³ and a direct antiglobulin test positive for C3d, an expression of residual extravascular haemolysis [29]. In patients of East Asian ancestry, a poor response may be due to the C5 variant p.(Arg885His), which prevents binding of eculizumab and ravulizumab [50].

On treatment, the erythrocyte clone grows, because PNH erythrocytes are no longer destroyed intravascularly; growth and stabilisation of the erythrocyte clone are compatible with disease control, and the percentage of PNH neutrophils and monocytes is usually much higher than that of erythrocytes, a difference that tracks the degree of haemolysis and LDH [20]. Normalisation of haemoglobin appears to predict a reduction of the erythrocyte clone [19]. C5 inhibition also introduces a specific infectious risk: even with vaccination, the risk of Neisseria infection is more than 1000 times that of healthy controls, with an absolute risk of up to 0.5% per year [29]. The complement inhibitor does not modify the stem cell defect and is not curative; only allogeneic HSCT eradicates the clone [17, 18].

3.7. Paediatric age and pregnancy

PNH may arise at a young age: about 10% of patients are younger than 21 years, and at this age the disease is often misdiagnosed and mismanaged [1]. In the Hammersmith cohort, the youngest patient was 16 years old at diagnosis [28]. A retrospective analysis of 26 cases with onset in childhood and adolescence, published by Ware and colleagues in 1991 and summarised in the IPIG guideline, showed the similarities with adult disease: the picture was dominated by signs and symptoms of haemolysis, bone marrow failure and thrombosis, although haemoglobinuria could be less frequent; response to immunosuppression was good in 6 of 9 patients, but there were no spontaneous remissions and long-term survival was poor, 80% at 5 years, 60% at 10 years and only 28% at 20 years [1]. A Dutch study of 11 children confirmed the frequent presentation with bone marrow failure; 5 underwent bone marrow transplantation (3 from matched unrelated donors and 2 from matched family donors) and 4 were alive [1]; the median age was 12 years, and PNH was associated with aplastic anaemia in 7 children and with MDS in 4 [14]. The predominance of bone marrow failure in children corresponds to the low frequency of clones in paediatric aplastic anaemia, about 10% versus up to 70% in adults, and to the need to rule out inherited bone marrow failure [29]. These data pre-date complement inhibitors and describe the natural history, not the prognosis of a child treated today; the authorised paediatric indications are given in section 7.7.

Pregnancy is a period of risk because of physiological amplification of complement, which increases especially after 20 weeks [29]. Before complement inhibitors, maternal mortality approached 20%, mainly from thromboembolism, and fetal mortality was 9%, mainly from prematurity, and the IPIG guideline anticipated complications in at least half of mothers and in some newborns, although a favourable outcome was the rule [1, 29]. In pregnancy, pharmacodynamic breakthrough haemolysis is frequent from the second trimester even on treatment, and a published case illustrates the presentation: a 25-year-old woman, second pregnancy, at 33 weeks, with fatigue and an episode of "cola-coloured" haemoglobinuria, with haemoglobin falling from 9.9 to 7.8 g/dL and LDH rising from 820 to 1740 IU/L, haptoglobin below 10 mg/dL and a clone of 56% in granulocytes, 55% in monocytes and 13% in erythrocytes [29]. Outcomes under eculizumab and ravulizumab and management are described in section 7.11.

3.8. PIGT-associated PNH

PIGT-associated PNH shares intravascular haemolysis and thrombosis with PIGA-related PNH, but is distinguished by autoinflammatory manifestations: recurrent urticaria, arthralgia, fever and aseptic meningitis [2, 8]. In the series that defined it as an entity, four patients (one Japanese, J1, and three German, G1, G2 and G3) were diagnosed with PNH at 68, 49, 65 and 66 years; J1, G1 and G3 had inflammatory symptoms from the ages of 30, 26 and 48 years, respectively, that is, 18 to 38 years before the haematological diagnosis [2]. The first case (G1), published in 2013, was a Caucasian woman with direct antiglobulin test-negative haemolytic anaemia diagnosed at 44 years, followed for several years by frequent haemolytic crises, abdominal pain, diarrhoea, headache, arthralgia, dyspnoea and fatigue, and with GPI protein deficiency demonstrated by flow cytometry at 49 years; her past history included cold-induced urticaria in childhood and, later, episodic urticaria without cold exposure, sometimes with systemic symptoms, and eculizumab improved the picture [11]. The age of 26 years given for the onset of G1's inflammatory symptoms in the 2019 series does not match the cold urticaria in childhood described in the original report, and the sources consulted do not clarify the difference [2, 11]. Urticaria was associated with a neutrophilic infiltrate in J1 and a mixed inflammatory infiltrate in G3, and J1 and G3 had recurrent aseptic meningitis with neutrophil predominance in the cerebrospinal fluid: 121 episodes in J1 between 53 and 69 years and 19 episodes in G3 between 62 and 65 years [2]. G2 had severe atherosclerosis, which the authors consider may be related to autoinflammation, but died before autoinflammatory symptoms could be confirmed [2].

The chronology suggests a relationship between clone size and phenotype. The disease always began with isolated autoinflammation, and PNH became apparent only many years later; in DNA from J1 obtained before the onset of recurrent haemolysis, only about 3% of leucocytes were PIGT-deficient, supporting the hypothesis that a small clone suffices for autoinflammation and that haemolysis requires a large clone [2]. The inflammatory picture resembles that of the cryopyrinopathies, caused by gain-of-function variants in NLRP3, in which small mosaic clones are also sufficient to produce symptoms [2]. In J1, serum interleukin 18 was raised before and after eculizumab, and serum amyloid A was raised before and normalised afterwards, indicating one component dependent on and another independent of C5 activation [2]. Unlike PIGA-related PNH, none of the patients had documented bone marrow failure, and clonal expansion is associated with the somatic 20q deletion, by a mechanism similar to that of myeloid neoplasms with 20q− [2].

Eculizumab controlled intravascular haemolysis and prevented the autoinflammatory manifestations: after it was started, 3 to 5 years before publication, J1 and G3 had no new episodes of meningitis, and urticaria and arthralgia improved in the three patients with inflammatory symptoms [2]. The treated disease showed two particular features. As in PIGA-related PNH, extravascular haemolysis appeared, with a positive direct antiglobulin test in G1 and residual haemolysis in G2 [2]. Unlike PIGA-related PNH, the percentage of PNH cells rose rapidly not only in erythrocytes but also in granulocytes in G1, G3 and J1, which the authors attribute to prevention of lysis of PIGT-deficient granulocytes by the membrane attack complex, without excluding a coincidence with an active phase of clonal expansion [2]. The authors recommend considering GPI protein deficiency in the presence of recurrent autoinflammatory symptoms, such as aseptic meningitis, even without PNH symptoms [2]. The case of G3 is described in detail in section 10.1.

3.9. Congenital CD59 deficiency

Congenital CD59 deficiency reproduces the haemolytic component of PNH from infancy and adds a neurological disease that acquired PNH does not have. The first case, described in 1990, was a Japanese patient with complete CD59 deficiency and a PNH-like picture, the son of cousin parents with reduced expression of the protein [3, 48]. The series that defined the current phenotype brought together infants from four unrelated Jewish families of North African origin, all homozygous for p.(Cys89Tyr), with chronic Coombs-negative haemolysis and relapsing polyneuropathy presenting as chronic inflammatory demyelinating polyradiculoneuropathy (CIDP) from early infancy [12]. The Orphanet definition describes chronic Coombs-negative haemolysis, early-onset, relapsing immune-mediated, inflammatory, axonal or demyelinating sensorimotor peripheral polyneuropathy, and isolated or recurrent cerebrovascular events in the anterior or posterior circulation, with neonatal or childhood onset [4].

The original reports, summarised in OMIM, specify the picture. The 1990 patient was a 22-year-old Japanese man with intermittent pallor and haematuria for 9 years; PNH had been diagnosed at 13 years, in an episode of haemolytic anaemia with haemoglobinuria, the Ham and sucrose tests were always positive, and over the following 9 years he had 9 haemolytic episodes, with cerebral infarction in the third and ninth; the parents, cousins without a history of haemolysis, had erythrocytes with partial CD59 deficiency, compatible with heterozygosity, and no polyneuropathy was described in this patient [5]. The 2013 series included 5 children from 4 families (one consanguineous), with onset between 3 and 7 months, usually after a mild viral infection, and symmetrical weakness, hypotonia and hyporeflexia more marked in the lower limbs; intravenous immunoglobulin and corticosteroids restored strength in the upper limbs, but the course was one of relapses and remissions triggered by infections, with progressive atrophy of the hands and feet and persistent paralysis of the lower limbs with areflexia [5]. Some exacerbations required mechanical ventilation, one patient had an episode of haemolytic uraemic syndrome during plasmapheresis, which resolved, cognitive development was normal, during episodes there was acute haemolysis with raised C-reactive protein and cerebrospinal fluid protein, nerve conduction studies showed demyelination with axonal damage in 2 of 3 patients and spinal magnetic resonance imaging showed nerve root enhancement in 2 of 3; immunosuppression lengthened the intervals between relapses and reduced their severity without preventing them, and one patient died of acute respiratory failure at 3.5 years [5].

The neuropathy is usually the first diagnosis, labelled recurrent Guillain-Barré syndrome or CIDP, and relapses are triggered by intercurrent infections, with hypotonia, limb weakness and hyporeflexia [9]. In the treatment study of four paediatric patients homozygous for p.(Cys89Tyr), the neuropathy was characterised as acute recurrent, predominantly motor and demyelinating, with conduction block and secondary axonal damage [23]. In the 13 individuals characterised for the four variants known at the time of curation, the recurrent triad was Guillain-Barré-type polyneuropathy or recurrent CIDP, recurrent strokes and chronic haemolysis; recurrent ischaemic cerebral infarcts with retinal disease, moyamoya-pattern cerebral small vessel changes, optic neuritis and a prothrombotic tendency have also been described [9]. The proposed mechanism is unopposed deposition of C5b-9 on erythrocytes, nodes of Ranvier and endothelium [9].

The 2026 systematic review, with 20 case reports and small series, describes a severe and heterogeneous phenotype with predominant neuromuscular involvement, frequently associated with haematological and cerebrovascular manifestations, often delayed diagnosis because of overlap with more common neurological and haematological disorders, and cases coming mainly from regions with high consanguinity in the Middle East and Asia; the relationship between the predicted functional impact of the variant and clinical severity was only exploratory [13]. Responses to conventional immunomodulators (corticosteroids and intravenous immunoglobulin) are variable and often transient [9, 13].

Disease treated with eculizumab, off-label, has a single prospective study. In the 24-month open-label phase IIa study, the four patients had marked neurological improvement in the upper limbs and trunk and more modest improvement in the lower limbs, corticosteroids and intravenous immunoglobulin were discontinued completely, there were no relapses in 104 weeks despite intercurrent infections and no hospital admissions, and C3bi and C5b-9 deposition on erythrocytes and neutrophils decreased [23]. Outside this study, a girl with a homozygous truncating CD59 variant, who presented at 7 months with bulbar symptoms, severe hypotonia and areflexia and later developed, during febrile illnesses, acute haemolytic anaemia with progressive neurological deterioration, T2-hyperintense lesions on brain magnetic resonance imaging, seizures and visual impairment, had neurological improvement about 6 months after starting eculizumab; at 5.5 years she ate and swallowed normally, walked short distances with support and had improved cognition and language production [5]. In the systematic review, benefit from eculizumab was observed in some patients, but inconsistently [13]. The following table summarises the clinical differences between the three entities that share loss of CD59.

4. Clinical Recognition and Warning Signs

4.1. When to request testing for a PNH clone

Flow cytometry for PNH is the most sensitive and specific test available, but it is not a screening test for anaemia; common causes of anaemia should be excluded first, and the initial assessment includes full blood count, reticulocytes, Coombs test and LDH [19]. The indications of the main guidelines agree on essentials and differ in details of threshold and population.

The Italian consensus recommends a higher degree of suspicion in people younger than 45 years with any of these presentations and points out that a combination of signs increases the likelihood of finding a clinically significant clone [20]. In series of high-risk patients tested in Spain and Brazil, 563 of 3938 cases (14%) had a PNH clone, of whom 34% had bone marrow failure, 24% haemolysis, 8.8% cytopenia and 1.5% thrombosis [20]. In a reference laboratory in the USA, 633 of 10,236 patients tested over 9 years (6.2%) had a clone, and 42% of these clones were below 1% [42]. Thrombosis as the presenting manifestation is rare (about 5%), and abdominal pain or dysphagia is the first manifestation in about 10% of patients [1].

4.2. Warning and emergency signs

In a patient with known or suspected PNH, abdominal pain, chest pain, dyspnoea and haemoglobinuria are early indicators of an impending thrombotic event [18]. Thrombosis of the hepatic veins (Budd-Chiari syndrome), portal vein, mesenteric veins or cerebral venous sinuses is an emergency requiring immediate anticoagulation, consideration of local or systemic thrombolysis in a specialised centre and interruption of intravascular haemolysis with an effective terminal inhibitor [1, 18]. Acute haemolytic exacerbation with a fall in haemoglobin, rise in LDH and haemoglobinuria requires hydration, pain control, transfusion assessment and identification of the complement-amplifying condition that precipitated it [1, 78].

In patients on complement inhibitors, sudden fever, headache with nausea or vomiting, neck stiffness, rash or confusion require immediate medical assessment for suspected meningococcal disease; the patient should carry an emergency card and stand-by medication [18, 21]. Worsening anaemia with reticulocytopenia or the appearance of a new cytopenia calls for bone marrow reassessment, because it reflects bone marrow failure rather than haemolysis [20].

4.3. Recognition of the forms with a germline component

In PIGT-associated PNH, autoinflammatory symptoms precede haemolysis by years or decades: recurrent urticaria, arthralgia and fever from the third to fifth decades and repeated episodes of aseptic meningitis with neutrophil predominance in the cerebrospinal fluid, while the clone is small; haemolysis becomes apparent when the clone grows [2]. An adult with unexplained recurrent aseptic meningitis, urticaria and arthralgia, especially if Coombs-negative haemolysis appears, warrants flow cytometry for a PNH clone and, if this is positive without a somatic PIGA variant, sequencing of GPI pathway genes and testing for a 20q deletion [2, 31].

In congenital CD59 deficiency, recognition occurs in childhood: an infant or young child with chronic Coombs-negative haemolysis and recurrent episodes of weakness with hypotonia and hyporeflexia, triggered by intercurrent infections, usually labelled recurrent Guillain-Barré syndrome or chronic inflammatory demyelinating polyradiculoneuropathy, with or without cerebral infarcts [4, 9, 12]. The triad of relapsing polyneuropathy, recurrent strokes and chronic haemolysis was the dominant presentation in the 13 individuals characterised for the four known variants, and diagnosis is frequently delayed [9, 13]. Family origin in regions with high consanguinity and North African Jewish ancestry increase suspicion [12, 13].

4.4. Newborn screening

PNH is acquired and is not subject to newborn screening in any of the sources consulted. For congenital CD59 deficiency and the germline GPI anchor defects, the sources consulted do not describe newborn screening programmes; in the latter, raised alkaline phosphatase in infants with epilepsy may be the first biochemical sign [41, 79].

5. Diagnosis: Criteria, Tests and Algorithms

5.1. Minimum criteria and classification

The essential minimum IPIG criteria are three: demonstration, by flow cytometry of erythrocytes, granulocytes or preferably both, of a population partially or completely deficient in all GPI proteins, using antibodies against GPI proteins or FLAER; full blood count with reticulocytes, LDH, fractionated bilirubin and haptoglobin; and bone marrow aspirate, biopsy and cytogenetics, required for classification [1]. Recommended ancillary studies include a targeted clinical history (haemoglobinuria, thromboembolic disease, dysphagia, odynophagia, abdominal pain, erectile dysfunction), the proportion of type I, II and III erythrocytes, the proportion of deficient granulocytes, iron studies (iron, total iron-binding capacity, ferritin), creatinine and urea, serum erythropoietin, urinary haemosiderin and HLA typing [1]. The flow cytometry result does not by itself establish the clinical diagnosis of PNH: it is the combination with clinical and laboratory data that defines the disease and its subcategory [42].

The Brazilian consensus summarises the laboratory pattern of non-immune intravascular haemolysis: anaemia, raised reticulocytes, raised LDH, raised bilirubin, low haptoglobin, free haemoglobin in serum (pink serum), haemoglobinuria (dipstick positive for haem with a sediment without erythrocytes) and negative DAT [17]. The initial assessment also includes D-dimers, prothrombin time, activated partial thromboplastin time and international normalised ratio (INR), bone marrow aspirate with cytogenetics and biopsy (especially with moderate to severe cytopenia), upper abdominal Doppler ultrasound, echocardiogram, pulmonary CT angiography if pulmonary hypertension is suspected, folate and vitamin B12, renal function, ferritin and transferrin saturation, natriuretic peptide and HLA typing of the patient and siblings in young transplant candidates [17]. Onkopedia adds exclusion of congenital membranopathies and enzymopathies by history, a search for schistocytes to exclude thrombotic microangiopathy, haemopexin, urinary haemosiderin, abdominal ultrasound with Doppler of the hepatic, portal, splenic and mesenteric veins, and, if ferritin exceeds 1000 ng/mL, assessment of organ damage with echocardiogram, blood glucose, thyroid function and multiparametric hepatic and renal magnetic resonance imaging [18].

5.2. Flow cytometry: method and sensitivity

High-sensitivity flow cytometry of peripheral blood has replaced the lysis tests. The 2018 ICCS/ESCCA guidelines recommend, for erythrocytes, identification of the population by CD235a (glycophorin A) and detection of deficient cells by CD59, with separation and quantification of type I, II and III populations; without a population-gating antibody, assays based only on CD55 and CD59 are subject to pipetting and mixing errors and are neither accurate nor sensitive below clones of 1% to 4% [42, 80]. For leucocytes they recommend FLAER (fluorescent aerolysin, which binds directly to the glycan part of the GPI anchor and cannot be used on erythrocytes), with gating markers CD45 for leucocytes, CD15 for neutrophils and CD64 for monocytes, and GPI markers such as CD24 and CD14 or, alternatively, CD157, which allows neutrophils and monocytes to be assessed in a single five-colour tube [20, 42, 80]. The historical designation "granulocytes" is replaced by "neutrophils", because modern gating by light scatter, CD45 and CD15 excludes basophils and most eosinophils [42].

The sample is anticoagulated peripheral blood, preferably with EDTA (heparin and ACD are acceptable), fresh, less than 48 hours old; stored at 4 °C, stability has been validated beyond 48 hours for erythrocytes and leucocytes [42, 80]. Bone marrow is not used, because immature myelomonocytic cells express low levels of GPI proteins, including CD14, and expression varies with differentiation [18, 80]. Onkopedia describes the practical "2×2" rule: two lineages, each with two reagents, and considers reliable a clone with absent or reduced expression of at least two GPI markers in at least two lineages [18]. The report should be quantitative and separate type III (complete absence) from type II (partial expression) [18, 42].

Sensitivity depends on the number of events acquired. The Italian consensus makes the calculation explicit: assuming 20 to 30 events as the minimum for detection and 50 for quantification, the limit of detection is 20 (or 30) divided by the total events in the population, and the limit of quantification is 50 divided by that total; 5000 events are needed for 1%, 50,000 for 0.1% and 500,000 for 0.01% [20]. The limits of detection and quantification should be stated in the report for each population [20]. In a validation study with one million erythrocytes counted in ten healthy donors, the mean number of type III events was 2 with a standard deviation of 1, giving calculated limits of 4 and 6 events, corresponding to 0.0006% and 0.0009% [81]. In practice, the achievable limit of quantification is 0.01% in erythrocytes and 0.05% to 0.1% in neutrophils [42]; in monocytes, in hypocellular samples or in severe neutropenia it may fall to 0.1% to 0.5%, and that limit should be stated [20]. For the routine diagnosis of classical haemolytic PNH, a sensitivity of 0.1% is generally acceptable; to detect small clones in aplastic anaemia and MDS a sensitivity of 0.01% is needed [18, 20].

Clone size is preferably determined in neutrophils, which are the reference population, because their half-life is normal and is not affected by red cell transfusion [1, 20]. The erythrocyte clone underestimates the disease because of haemolysis and transfusions; erythrocyte analysis should be done before transfusion or after one month without transfusion, when clinically safe [1]. Inclusion of CD71 (transferrin receptor) in the panel makes it possible to distinguish immature erythrocytes, which are neither haemolysed nor diluted by transfusions, and improves estimation of the true type II and III erythrocyte clone [20]. Neutrophil and monocyte clones are usually concordant; discrepancies rarely change management [42]. Participation in interlaboratory exercises and formal analytical validation (accuracy, specificity, sensitivity, imprecision, linearity, carry-over, measurable range, stability and reference interval), with a written operating procedure fixed before validation, are required by the ICCS/ESCCA guidelines and by the ICSH/ICCS standards for laboratory-developed tests [18, 81].

5.3. Interpretation, test limits and pitfalls

A negative result in erythrocytes and leucocytes excludes clinical PNH even with an assay of only 1% sensitivity, but gives no information about aplastic anaemia; for that purpose, only a negative result obtained with a high-sensitivity assay (limit of quantification 0.01%) is informative [19]. Conversely, too low a threshold loses specificity: deficient neutrophil populations as low as 0.003% have been considered abnormal by some authors, but the rate of PIG-A variants in colony-forming cells from healthy people and the frequency of PNH erythrocyte phenotypes in normal blood lie in that order of magnitude, so the 0.01% threshold is the reasonable one for defining a disease-related clone [19]. False results, positive and negative, arise from poor instrument set-up, antibody clones and conjugates with insufficient signal-to-noise ratio, inadequate titration, contaminated wash media and poorly designed gating strategies [80]. Some CD235a conjugates lose fluorescence after washing, so acquisition should be done within a few minutes with these reagents [42]. A simultaneous decrease in CD235a and CD59 usually reflects a staining artefact and not true type II erythrocytes [42].

Quantification of at least two GPI proteins has been recommended since the 2005 guideline precisely to exclude that the picture results from an isolated inherited deficiency of a single GPI protein, such as congenital CD59 deficiency or CD55 deficiency [1]. In the former situation CD59 is absent and CD55 and other GPI markers, such as CD14, are preserved; none of the four known mutant CD59 proteins is detected by the available anti-CD59 antibodies, even the two that reach the membrane, so flow cytometry reports absent CD59 without distinguishing absent protein from protein that is present but non-functional, and confirmation requires CD59 sequencing [9]. The two-protein rule does not exclude germline defects of anchor biosynthesis, in which several GPI proteins are simultaneously reduced on blood cells; in inherited GPI deficiency due to a PIGM promoter variant, with portal and cerebral thrombosis in childhood, CD59 and CD24 were variably decreased on haematopoietic cells, without haemolysis or bone marrow failure, and diagnosis was made by PIGM sequencing [72]. The work-up of congenital CD59 deficiency is completed with a negative DAT, markers of intravascular haemolysis, nerve conduction studies (demyelination with conduction block) and brain magnetic resonance imaging [9].

5.4. Molecular diagnosis and the limits of genetics

In PNH, identification of the somatic PIGA variant would confirm the diagnosis, but has historically been confined to research because of technical difficulties, and Onkopedia explicitly states that genetic analysis of PIGA is not necessary to confirm the diagnosis routinely in typical constellations of findings [1, 18]. In atypical manifestations or flow cytometry, extended molecular diagnosis may be useful: this was the route that led to the discovery of PIGT-associated PNH, by deep sequencing of GPI pathway genes in a patient without a somatic PIGA variant, followed by array comparative genomic hybridisation that demonstrated the 20q deletion [2, 31]. NGS analysis should be performed on granulocytes with a PNH phenotype sorted by flow cytometry, because the somatic variant is confined to the clone [2]. In MDS-associated PNH, the NGS myeloid panel is necessary for classification and risk stratification; in classical PNH and in PNH associated with aplastic anaemia, the usefulness of the myeloid panel remains under investigation, because the prognostic significance of somatic variants in aplastic anaemia has not been clarified [20].

The intrinsic limitations of sequencing technologies apply to the germline forms. In CD59 and PIGT deficiencies and in germline GPI anchor defects, exome or genome sequencing is the diagnostic method, but it does not reliably detect deep intronic variants or balanced rearrangements, and detection of copy number variants requires targeted analysis; in germline PIGA defects, all published cases were identified by exome or genome sequencing, because raised alkaline phosphatase and reduced expression of GPI proteins support the diagnosis but do not identify the gene, since the signature is common to several pathway genes [7, 77]. A negative result does not exclude these disorders. In X-linked forms, skewed X-chromosome inactivation in the mother supports carrier status but its absence does not exclude it; incomplete penetrance has been documented in a family with a germline PIGA variant, with CNTN2 proposed as a modifier [7]. In the particular case of PIGT-associated PNH, demonstration of the mechanism requires showing that the germline variant and the somatic deletion are on different alleles, which in the original study was done by imprinting and methylation analysis of the L3MBTL1 and SGK2 genes [2].

5.5. Report content and variant classification

A PNH flow cytometry report should state the populations analysed, the gating and detection markers used, the clone size in each lineage (neutrophils, monocytes and erythrocytes), the separation between type II and type III erythrocytes, the limits of detection and quantification achieved in each population and, in serial analyses, the relative difference from the previous test [19, 20, 42]. A genetics report should state the gene, the transcript with version, the HGVS notation at DNA and protein level, the genomic position with the reference genome version, the zygosity or, in the somatic case, the allele fraction and the tissue analysed, and the variant classification according to ACMG/AMP (American College of Medical Genetics and Genomics/Association for Molecular Pathology) criteria [6, 49]. Variants of uncertain significance should not guide therapeutic decisions; they require family segregation studies, where applicable, and may benefit from functional assays, such as restoration of CD59 and CD55 expression and FLAER binding in deficient cells, used to classify variants in GPI pathway genes [60].

5.6. Prenatal and preimplantation diagnosis

PNH due to somatic PIGA is acquired and is not subject to prenatal or preimplantation diagnosis [7, 10]. In congenital CD59 deficiency, an autosomal recessive disorder with known familial variants, prenatal diagnosis by amniocentesis or chorionic villus sampling and preimplantation genetic testing are technically possible, as in any monogenic disorder with identified variants, and the same principles described for germline GPI anchor defects with known familial variants apply [60]. In PIGT-associated PNH, the heterozygous germline variant does not cause disease by itself, so prenatal diagnosis is not indicated [2].

5.7. Practical algorithm

In the presence of Coombs-negative haemolysis, thrombosis at an atypical site, unexplained cytopenia or the smooth muscle symptoms described in section 4, intravascular haemolysis is confirmed first (LDH, fractionated bilirubin, haptoglobin, reticulocytes, haemoglobinuria) and immune haemolysis (DAT) and thrombotic microangiopathy (schistocytes) are excluded [17, 18]. High-sensitivity flow cytometry of peripheral blood is then requested, in two lineages with two reagents each [18, 20]. If a clone is present, it is quantified in neutrophils, monocytes and erythrocytes, with separation of types II and III, and the patient is classified into the three IPIG subcategories with the support of the bone marrow study (cytology, cytogenetics, histology and, if indicated, myeloid panel) [1, 20]. Disease activity is then assessed (LDH greater than 1.5×ULN with at least one sign or symptom), along with thromboembolic risk (history and risk factors), organ damage (renal function, pulmonary hypertension) and vascular imaging (Doppler, CT or magnetic resonance) [20]. If flow cytometry is compatible with PNH but the picture is atypical, notably with autoinflammation, sequencing of GPI pathway genes and testing for a 20q deletion are performed; if the deficiency is restricted to CD59, CD59 is sequenced [1, 2].

6. Differential Diagnosis

6.1. Coombs-negative haemolysis and microangiopathies

The differential diagnosis of PNH is organised along three clinical axes: haemolysis, thrombosis and cytopenia [17]. On the haemolysis axis, the decisive step is the DAT, which separates immune from non-immune haemolysis; next, the intravascular pattern (very high LDH, undetectable haptoglobin, haemoglobinuria, haemosiderinuria) separates PNH from predominantly extravascular causes [17, 18]. The presence of schistocytes on the blood film points to thrombotic microangiopathy and requires assessment of ADAMTS13 activity and renal function before attributing haemolysis to PNH [18]. The Italian consensus insists on a practical point: including all markers of haemolysis in the investigation of any anaemia is the step that allows both common and ultra-rare causes, PNH among them, to be diagnosed [20]. In autoimmune haemolytic anaemias, with a positive DAT, rare PNH clones have been described, and screening is justified when the rise in LDH is disproportionate to the degree of anaemia or when there is atypical thrombosis [20].

6.2. Thrombosis at an atypical site

A young patient with thrombosis of the hepatic, portal, mesenteric or splenic veins, cerebral venous sinuses or dermal veins should be assessed for PNH, especially if there is haemolysis, even mild, or cytopenia [1, 19]. The differentials on this axis are myeloproliferative neoplasms, solid tumours with hypercoagulability, extrinsic vascular compression and inherited and acquired thrombophilias [17]. In the GeneReviews chapters on factor V Leiden and prothrombin, PNH is listed among the acquired thrombophilias, alongside antiphospholipid syndrome and myeloproliferative disorders, and the effect of these acquired conditions on the thrombotic risk conferred by factor V Leiden is not well defined [83, 84]. Testing for inherited thrombophilia in PNH is reserved, according to Onkopedia, for patients with a family history of thrombosis [18]. In paediatric patients, inherited GPI deficiency due to PIGM is added to this differential: the children described had portal thrombosis before 2 years or cerebral venous thrombosis, with macrocephaly, dilated superficial veins and atypical absence seizures, without haemolysis; in one of the siblings, prophylactic oral anticoagulation prevented thrombosis but not seizures [72]. Flow cytometry shows reduction of GPI proteins on haematopoietic cells, so the absence of haemolysis, consanguinity, epilepsy and macrocephaly point towards sequencing of PIGM, whose causal variant lies in the promoter and may escape analyses restricted to coding regions [72].

6.3. Cytopenias and bone marrow failure

On the cytopenia axis, the practical question is distinguishing acquired aplastic anaemia from inherited bone marrow failure syndromes and hypoplastic MDS. Detection of a PNH clone is informative in both directions: it occurs in up to 70% of adults and 10% of children with acquired aplastic anaemia, and its presence helps to exclude inherited bone marrow failure, because PNH is almost never associated with Fanconi anaemia or dyskeratosis congenita, which may avoid extensive genetic studies in a young person with an aplastic marrow [19, 29]. In healthy people and in patients with MDS, the GPI-deficient cells detected derive from colony-forming cells and not from a multipotent stem cell, which helps to explain why PNH arises from aplastic anaemia but not from MDS [19]. In older patients, a clone of a few per cent is unusually large for MDS and should raise suspicion of an immune component treatable with immunosuppression [19]. Serial monitoring of the clone is useful in aplastic anaemia, but there is no role for serial monitoring of clone size in patients with MDS [19]. MDS associated with PNH is diagnosed mainly by cytogenetics, because morphology is insufficient in a context of dysplastic haematopoiesis due to haemolysis [18].

6.4. Germline versus somatic

The most important distinction from a genetic standpoint is between the somatic defect restricted to haematopoiesis and the germline defects of the same pathway. In germline defects, the patient is typically an infant or a child with epileptic encephalopathy, hypotonia, dysmorphism and malformations, without haemolysis or thrombosis, and alkaline phosphatase is raised in most hyperphosphatasia syndromes and normal or low in PIGT deficiency [8, 25, 63]. The exception to the absence of thrombosis is inherited GPI deficiency due to PIGM, in which portal and cerebral thrombosis dominates the picture, also without haemolysis [72]. GPI protein flow cytometry, with reduced CD16 and CD24, demonstrates the anchoring defect but does not identify the responsible gene, so diagnosis is molecular [8, 25]. Complement inhibitors have no role in these disorders, because they do not correct an anchor biosynthesis defect expressed in the brain [7]. PIGT-associated PNH is the only entity in which germline and somatic components coexist with a haemolytic phenotype, and its autoinflammation brings it clinically close to the cryopyrinopathies, which are distinguished by the absence of haemolysis and of a GPI-deficient clone [2]. Congenital CD59 deficiency occupies an intermediate position: it is constitutional and inherited, but its haematological phenotype and response to C5 inhibition bring it close to PNH [9, 23].

7. Treatments and Risk Management Strategies

7.1. Whom to treat and with what aim

Detection of a clone is not by itself an indication for complement inhibition; the decision is based on clinically relevant haemolysis, assessed from the clinical and laboratory picture as a whole, without a single fixed parameter, and is typically associated with larger clones [18]. The 2021 Brazilian consensus formulated criteria for starting treatment: symptomatic haemolysis with LDH of 1.5×ULN or more, plus at least one of haemoglobin below 7 g/dL or below 10 g/dL on two measurements with cardiac symptoms, PNH-related thrombosis, complications of haemolysis (renal dysfunction, pulmonary hypertension), abdominal pain, dysphagia or erectile dysfunction, or pregnancy, especially with previous gestational complications, regardless of transfusion requirement and clone size [17]. The same consensus considers treatment not indicated for mild or absent symptoms, a granulocyte clone below 30% without haemolysis and with a normal blood count, or aplastic anaemia and MDS without haemolysis or thrombosis, because the inhibitor does not correct the stem cell defect [17]. By category: in the subclinical form there is no specific therapy, with attention to bone marrow failure; in PNH with bone marrow failure the focus is the marrow disease, although very large clones may benefit; in classical PNH the inhibitor is indicated [17]. Onkopedia adds as indications transfusion requirement due to haemolysis (as distinct from anaemia due to cytopenia), status after thromboembolism, associated renal failure and attacks of abdominal pain or other severe symptoms, and in asymptomatic patients proposes surveillance, with prophylactic anticoagulation an option to consider [18]. The Italian consensus considers high disease activity or PNH-related complications sufficient to consider anti-C5, regardless of transfusions and clone size [20].

7.2. Supportive treatment

Folic acid is recommended to compensate for the increased consumption by accelerated erythropoiesis, at a dose of 5 mg daily in the IPIG guideline and 1 to 5 mg daily in Onkopedia, with vitamin B12 if deficient [1, 17, 18]. Iron deficiency is frequent, owing to urinary loss of haemoglobin and haemosiderin, and may occur even without macroscopic haemoglobinuria; replacement by any route may exacerbate haemolysis, but concern about this exacerbation should not prevent replacement, because iron deficiency limits erythropoiesis and itself worsens haemolysis, and the exacerbation, if it occurs, is controlled with corticosteroid, androgen or suppression of erythropoiesis by transfusion [1]. The oral route is associated with less intense exacerbations, but urinary loss may be so great as to prevent repletion; the parenteral route is generally safe [1]. Under effective complement inhibition, urinary loss ceases, so Onkopedia warns against giving iron without monitoring and advises considering depletion if there is overload, especially in patients with residual transfusions [18].

Red cell concentrate is transfused according to clinical need and also reduces haemolysis by suppressing erythropoiesis; concerns about exacerbation induced by small amounts of donor plasma have not been confirmed, but filtration is recommended to avoid reactions between donor leucocytes and recipient antibodies [1]. Washed concentrates are not necessary [18]. Transfusional iatrogenic haemochromatosis is a late event in these patients because of urinary iron loss, and iron overload in classical PNH is rare, but remains a concern in chronically transfused patients whose anaemia is mainly due to marrow failure [1]. Bacterial infections should be treated early because they amplify complement and precipitate crises, and hydration is recommended during crises [17, 18].

Glucocorticoids are an empirical therapy without randomised trials: the IPIG guideline recorded that some members of the group do not advocate their use under any circumstances, whereas others observe a rapid response, often within the first 24 hours, with prednisone 0.25 to 1.0 mg/kg daily; their main value probably lies in attenuating acute exacerbations with short pulses, and prolonged use is limited by toxicity (myopathy, osteopenia, iatrogenic Cushing syndrome), with antibacterial and osteopenia prophylaxis recommended if it is continued [1]. The Brazilian consensus maintains the same position: empirical therapy, without clear benefit, only in specific situations and for a few days during crises [17]. Androgens, alone or with corticosteroid, have been used successfully in the anaemia of PNH; danazol, an attenuated synthetic androgen, has a more favourable toxicity profile, with a recommended starting dose of 400 mg twice daily and possible maintenance with 200 to 400 mg daily, under mandatory monitoring of liver function, given the risks of cholestatic jaundice and peliosis hepatis, prostatic hypertrophy and virilisation [1]. It is not known whether androgens increase the risk of Budd-Chiari syndrome in PNH [1]. Recombinant erythropoietin may be used if there is absolute or relative deficiency, with monitoring, because it may worsen haemolysis by increasing production of GPI-deficient erythrocytes [1]. The role of splenectomy is undefined, with anecdotal reports of improvement and a risk of postoperative complications, notably thrombosis, and some IPIG members consider that it has no place in the treatment of PNH [1]. Immunosuppression is not indicated for isolated haemolysis; it is indicated in PNH with aplastic anaemia and bone marrow failure [17, 18].

7.3. Anticoagulation

After a thromboembolic event, anticoagulation should be indefinite, with standard-intensity warfarin (INR 2.0 to 3.0) in the 2005 guideline, and reassessed if there is spontaneous remission or if the clone falls below 50% [1]. In acute thrombosis, heparin or low-molecular-weight heparin (LMWH) is started and thrombolysis or radiological intervention is considered in Budd-Chiari syndrome of acute onset; thrombocytopenia is a relative contraindication, and the approach is to transfuse platelets to keep them in a safe range rather than stop anticoagulation [1]. Primary prophylaxis was and remains a matter of disagreement. Hall and colleagues proposed warfarin prophylaxis for patients with more than 50% deficient granulocytes without contraindication, based on a 10-year thrombosis risk of 44% in that group versus 5.8% in patients with less than 50%, and on the absence of events in 39 anticoagulated patients versus a 36.5% 10-year risk in 56 non-anticoagulated patients, at the cost of two major haemorrhages in about 100 patient-years [1]. Moyo and colleagues calculated an odds ratio of 1.64 for each 10% increase in the percentage of deficient granulocytes, with an 11.8-fold higher risk in a patient with more than 70% compared with a 20% clone [1]. The prevalence of thromboembolism in patients from Japan, China and Mexico, about 2%, is lower than in patients from the United States and Europe even after adjustment for clone size, which makes it difficult to justify prophylaxis in these groups [1]. The Brazilian consensus recommends primary prophylaxis in patients not on eculizumab with a granulocyte clone above 50%, raised D-dimers, pregnancy, the perioperative period or other thrombophilic factors, without contraindication and with stable platelets above 100×10⁹/L [17].

With complement inhibition under way, the disagreement concerns discontinuation. The FDA label for eculizumab states that the effect of withdrawing anticoagulation during treatment has not been established and that treatment with eculizumab should not alter anticoagulation management [21]. Onkopedia considers that, under effective inhibition, anticoagulation may be discontinued according to the general rules for thromboembolism, and recommends starting a complement inhibitor in situations of prolonged immobilisation [18]. Brodsky overlaps anticoagulation and complement inhibition for 3 to 6 months and stops anticoagulation if the thrombotic symptoms have resolved and LDH is below 1.5×ULN, noting that other haematologists recommend indefinite anticoagulation [29]. Coumarins and heparins can be used; direct oral anticoagulants have data in very small groups, apparently comparable under inhibitors, and Brodsky considers them probably equivalent but little studied in PNH [18, 29]. Anticoagulation alone is insufficient in the long term after thrombosis in PNH, so the inhibitor is started as early as possible [29].

7.4. Allogeneic haematopoietic stem cell transplantation

Allogeneic transplantation is the only potentially curative therapy [7, 18]. The indications, according to the IPIG guideline, are bone marrow failure (in which case the decision is governed by the standards for the underlying marrow disease, for example aplastic anaemia) and major complications of PNH: recurrent, life-threatening thromboembolic disease and refractory, transfusion-dependent haemolytic anaemia, weighing age, comorbidities, availability of a matched sibling and time since diagnosis [1]. As for conditioning, cyclophosphamide with antithymocyte globulin is recommended in PNH with aplastic anaemia and a more myeloablative regimen (with radiation or busulfan) in classical PNH, and in syngeneic transplants a myeloablative regimen to avoid recurrence; the role of non-myeloablative regimens requires further research [1]. There are no PNH-specific adverse events; severe acute graft-versus-host disease occurs in more than one third of patients and chronic disease in about 35%; overall survival in unselected patients transplanted from a matched sibling is between 50% and 60% [1], a figure matching the long-term survival of about 50% to 60% reported by Onkopedia [18]. The Brazilian consensus reserves it for severe cases: severe aplastic anaemia with a matched donor, high-risk MDS and complications refractory to eculizumab or unavailability of the drug, with the decision shared with a PNH specialist [17]. Onkopedia adds secondary bone marrow failure, evolution to MDS or acute myeloid leukaemia and recurrent thromboembolism despite prophylaxis and inhibition [18]. Allogeneic transplantation in aplastic anaemia prevents secondary PNH [18]. C5 inhibition does not compromise transplantation and may serve as a bridge: in one case described, eculizumab was stopped at conditioning with antithymocyte globulin and cyclophosphamide, with 100% donor chimerism at 30 days and complete remission without graft-versus-host disease at one year; in registry data from 1978 to 2007, thromboembolism in patients with aplastic anaemia and a PNH clone predicted higher transplant-related mortality [29]. In classical PNH, the possibility of spontaneous remission and the heterogeneity of the course require the timing of transplantation to be considered case by case [1, 28].

7.5. Complement inhibitors: drugs, doses and dosing regimens

Eight products are indicated for PNH in the United States and the European Union: three eculizumab products (the originator and two biosimilars), ravulizumab and crovalimab as anti-C5 agents, pegcetacoplan as an anti-C3 agent, iptacopan as a factor B inhibitor and danicopan as an add-on factor D inhibitor [21, 32–36, 85, 86].

Eculizumab (Soliris®; biosimilars Bkemv®/Bekemv®, eculizumab-aeeb, and Epysqli®, eculizumab-aagh) is a humanised anti-C5 monoclonal antibody given intravenously. The approved dose for PNH in adults is 600 mg weekly for the first four weeks, 900 mg as the fifth dose one week later and 900 mg every two weeks for maintenance, at the recommended time points or within two days in the FDA label, and every 14 ± 2 days in the European summary of product characteristics (SmPC) [21, 30]. The infusion is given over 35 minutes in adults (25 to 45 minutes in the European SmPC), with monitoring for at least one hour; if it is slowed, the total should not exceed two hours in adults [18, 21, 30]. The European SmPC allows home infusion by a qualified professional after good tolerance and considers treatment lifelong unless clinically indicated otherwise [30]. With worsening haemolysis, the Brazilian consensus and the European SmPC allow shortening the interval to 12 days or increasing the dose to 1200 mg [17, 30]. The terminal half-life is 11.3 days and plasmapheresis reduces it to 90.2 hours, so the label provides for supplementary doses after plasma exchange procedures in the other indications [18, 21]. In PNH, the FDA states that safety and efficacy in paediatric patients have not been established, and the label has weight-based paediatric doses only for atypical haemolytic uraemic syndrome and generalised myasthenia gravis; the dose of 900 mg weekly for the first four weeks, 1200 mg in the fifth and 1200 mg every two weeks above 40 kg, and the descending regimens down to 300 mg every three weeks at 5 to less than 10 kg, are the declared basis for extrapolation when a child with PNH needs treatment in the United States [21]. In the European Union the indication covers adults and children: above 40 kg the adult dose is used and below that the same weight-based tables apply, with the explicit caveat that the medicine has not been studied in patients with PNH weighing less than 40 kg, so the dosing results from pharmacokinetic and pharmacodynamic extrapolation [30].

Ravulizumab (Ultomiris®) is an anti-C5 agent derived from eculizumab by substitutions that confer pH-dependent dissociation from C5 in the endosome and recycling by FcRn, with a terminal half-life of 49.6 ± 9.08 days in PNH [18, 32]. The dose is intravenous and weight-based, with a loading dose followed by maintenance starting two weeks later: 600 mg followed by 300 mg every four weeks at 5 to less than 10 kg; 600 and 600 mg every four weeks at 10 to less than 20 kg; 900 and 2100 mg every eight weeks at 20 to less than 30 kg; 1200 and 2700 mg every eight weeks at 30 to less than 40 kg; 2400 and 3000 mg at 40 to less than 60 kg; 2700 and 3300 mg at 60 to less than 100 kg; 3000 and 3600 mg from 100 kg [22, 32]. The tolerance window is ±7 days, except for the first maintenance dose; in patients switching from eculizumab, the loading dose is given on the date the next eculizumab dose was due (or two weeks after the last infusion, in children, according to the European SmPC) [22, 32]. Supplementary doses are provided for after plasmapheresis, plasma exchange or intravenous immunoglobulin: for example, in a patient of 60 to less than 100 kg, 1500 mg after each procedure if the last dose was 2700 mg, 1800 mg if it was 3300 mg, and 600 mg after a course of immunoglobulin, given within four hours [32]. Dilution is 1:1 in 0.9% sodium chloride, with a 0.2 µm filter and a final flush with 100 mL, infused over 25 to 45 minutes; the solution is stable for 24 hours in the refrigerator and 5 hours at room temperature [18]. The European SmPC consulted contains only the intravenous formulations of 300 mg/3 mL and 1100 mg/11 mL, and the FDA label text obtained describes only the intravenous route [22, 32].

Crovalimab (PiaSky®) is a recycling anti-C5 agent that binds an epitope distinct from that of eculizumab, with a half-life of 53.1 days (90% confidence interval: 47.7 to 58.6) and subcutaneous bioavailability of 83.0%; the target serum concentration for complete C5 inhibition is 100 µg/mL [18, 36]. The regimen combines an intravenous loading dose on day 1 (1000 mg at 40 to less than 100 kg; 1500 mg from 100 kg) with 340 mg subcutaneously on days 2, 8, 15 and 22 and subcutaneous maintenance every four weeks from day 29 (680 mg at 40 to less than 100 kg; 1020 mg from 100 kg), with a tolerance of ±2 days except on days 1 and 2 and adjustment if weight remains consistently above or below 100 kg [36, 39]. The day 1 intravenous dose is given over 60 minutes, diluted in 100 to 200 mL of sodium chloride, at 500 mg per 30 minutes and with a filter; 680 mg corresponds to two 340 mg vials and 1020 mg to three [18]. When switching from another anti-C5 agent, the first loading dose is not given before the date the next administration of the previous inhibitor was due [36].

Pegcetacoplan (Empaveli® in the United States and Brazil; Aspaveli® in the European Union) is a pegylated cyclic peptide that binds C3 and C3b, with an effective half-life of 8.0 to 8.6 days, steady state in 4 to 6 weeks and bioavailability of 77% [18, 33]. The dose is 1080 mg subcutaneously twice weekly (on days 1 and 4 of each week in the European SmPC), given in about 20 mL by a commercial infusion pump with a reservoir of at least 20 mL or by a disposable on-body injector, over about 30 minutes at two sites or 60 minutes at one, within two hours of preparing the syringe, with self-administration possible after training [18, 33, 87]. If LDH exceeds 2×ULN, the regimen changes to 1080 mg every three days, with LDH twice weekly for at least four weeks [33, 87]. When switching from eculizumab, pegcetacoplan is started while continuing eculizumab at the current dose for four weeks and then stopping it; from ravulizumab, the FDA label states that it should be started no more than four weeks after the last dose, whereas the European SmPC describes four weeks of overlap with the C5 inhibitor in use and Onkopedia recommends starting in the second half of the interval, four weeks before the next dose [18, 33, 87]. The vial contains 820 mg of sorbitol and no adjustment is needed in severe renal impairment, with no data in dialysis [18].

Iptacopan (Fabhalta®) is an oral factor B inhibitor, in 200 mg capsules, at a dose of 200 mg twice daily with or without food, swallowed whole, 12 hours apart [18, 34]. Time to maximum concentration is about 2 hours, steady state is reached in about 5 days and the half-life is about 25 hours [18]. When switching, it is started no later than one week after the last dose of eculizumab and no later than six weeks after the last dose of ravulizumab, preferably at four weeks (mid-interval) [18, 34, 88]. If a dose is missed, one dose is taken as soon as possible and the regimen resumed; adherence is critical, because of the large expansion of the erythrocyte clone under proximal inhibition, and haemolysis should be monitored after several missed doses [18, 34, 88]. It is not recommended in severe hepatic impairment (Child-Pugh class C), with no adjustment needed in classes A and B or in mild to moderate renal impairment, and no data in severe renal impairment or dialysis [34, 88]. CYP2C8 inducers such as rifampicin reduce exposure and strong inhibitors such as gemfibrozil are not recommended; strong inducers of CYP2C8, UGT1A1, P-glycoprotein, BCRP and OATP1B1/3 are not recommended; clopidogrel increases the area under the curve by 36% and ciclosporin increases the maximum concentration by 41% and the area under the curve by 50% [18, 34].

Danicopan (Voydeya®) is an oral factor D inhibitor authorised only as add-on therapy to ravulizumab or eculizumab for extravascular haemolysis in adults with PNH, with the explicit limitation that it has not shown efficacy as monotherapy [35]. It is available as 50 and 100 mg tablets; the dose is 150 mg three times daily, with or without food, about 8 hours apart (±2 hours), increasing to 200 mg three times daily if haemoglobin has not risen by more than 2 g/dL after four weeks, if there was a transfusion in the previous four weeks or on clinical judgement; about 70% of patients in the ALPHA trial required the increase [18, 35, 89]. If more than three hours remain until the next dose, the missed dose may be taken; otherwise, it is skipped [35]. In severe renal impairment (estimated glomerular filtration rate below 30 mL/min/1.73 m²), the European SmPC states that treatment should start with 100 mg three times daily, up to a maximum of 150 mg three times daily; it is not recommended in severe hepatic impairment [35, 89]. The European SmPC provides for discontinuation with gradual tapering over 6 days, because of the risk of alanine aminotransferase elevation after cessation: from 150 mg to 100 mg three times daily for 3 days and then 50 mg three times daily for 3 days; from 200 mg to 100 mg three times daily for 3 days and then 100 mg twice daily for 3 days; from 100 mg to 100 mg twice daily for 3 days and then 100 mg once daily for 3 days — a regimen the FDA label does not describe [35, 89]. Liver enzymes are assessed before and periodically, with interruption if there is a clinically significant elevation, and lipids are monitored [35]. It is an inhibitor of BCRP (rosuvastatin limited to 10 mg daily; sulfasalazine) and of P-glycoprotein (fexofenadine, edoxaban, digoxin, dabigatran) [18, 35]. Because the background anti-C5 agent is maintained, irregular dosing should not cause severe breakthrough, and reticulocytosis and C3d deposition do not normalise completely [18].

In congenital CD59 deficiency, eculizumab is used off-label. In the 24-month open-label phase IIa study (NCT01579838), four children homozygous for p.(Cys89Tyr) received repeated intravenous eculizumab, with assessment every 2 to 4 weeks for 104 weeks [23]. The dosing regimen is not given in the abstract or in the accessible text, so the dose for this indication is not documented in the sources consulted; in practice, the available reference is the weight-based paediatric dosing in the eculizumab label, explicitly applied by extrapolation, and the decision should be taken in an experienced centre [21, 23, 30]. In PIGT-associated PNH, eculizumab was used at the usual PNH dosing and abolished intravascular haemolysis and autoinflammatory attacks [2].

7.6. Infectious risk, vaccination and antibiotic prophylaxis

All complement inhibitors carry a boxed warning for serious infections caused by encapsulated bacteria [21, 32–36]. According to the ACIP (Advisory Committee on Immunization Practices), use of complement inhibitors is associated with an incidence of meningococcal disease about 2000 times higher, and patients remain at risk even when vaccinated, because opsonophagocytic activity is insufficient in the presence of eculizumab: after MenB-4C vaccine, only 17% of individuals with terminal complement deficiency or on an inhibitor had bactericidal activity with endogenous complement [90]. Brodsky estimates the risk of Neisseria infection under terminal inhibition at more than 1000 times that of healthy controls, with an absolute risk of up to 0.5% per year [29].

The ACIP schedule for people with complement deficiency or on an inhibitor is, for MenACWY (conjugate, any formulation) from 2 years of age, two doses at least 8 weeks apart, with boosters 3 years later and every 5 years in children under 7 years and every 5 years in those over 7 years, and age- and product-specific schedules between 2 and 23 months; for MenB from 10 years of age, three doses of MenB-FHbp at 0, 1 to 2 and 6 months or two doses of MenB-4C at least 1 month apart, not interchangeable, with a booster one year after the primary series and then every 2 to 3 years while the risk persists [90]. The 2025 update, published in January 2026, allows use of the pentavalent MenACWY-CRM/MenB-4C vaccine (Penmenvy®) when both vaccines are indicated at the same visit in people aged 10 years or more at increased risk, including use of complement inhibitors, without changing the 2020 indications [91]. The labels require complete or up-to-date vaccination against serogroups A, C, W, Y and B at least two weeks before the first dose, unless the risk of delaying treatment outweighs the risk of infection; if starting is urgent, antibacterial prophylaxis is given and vaccination is done as soon as possible, the label acknowledging that the optimal duration and regimens of prophylaxis and its efficacy have not been studied [21, 30]. With proximal inhibitors (pegcetacoplan, iptacopan, danicopan), vaccination against Streptococcus pneumoniae is also mandatory and vaccination against Haemophilus influenzae type b is recommended [18, 33, 34].

There is a relevant disagreement about the timing of vaccination. The FDA and EMA labels and the ACIP maintain vaccination at least two weeks before starting [21, 30, 90]. Onkopedia recommends, in new patients, vaccinating on the day of the first infusion and not before, because vaccines may increase haemolysis up to haemolytic crisis and severe thrombosis, with phenoxymethylpenicillin 1 million units twice daily as prophylaxis until vaccination [18]. The ACIP documents the basis for this concern: in Canada, increased anaemia and haemolysis were observed with MenB-4C in patients on eculizumab, especially when the eculizumab dose had been given in the previous two weeks, and the Canadian label now recommends vaccinating before or at the same time as starting treatment and, during treatment, only with controlled disease and a high eculizumab concentration [90]. When switching from a terminal to a proximal inhibitor, vaccination may precede the switch, because the ongoing terminal inhibition prevents haemolytic crisis; in treatment-naive patients starting a proximal inhibitor, Onkopedia recommends vaccinating immediately or, preferably, one day after the first administration, and all vaccines except meningococcal ones may be co-administered [18].

Continued antibiotic prophylaxis is a matter of disagreement. The ACIP indicates antimicrobial prophylaxis (for example with penicillin) together with vaccination and for two weeks afterwards in patients who cannot wait, and allows clinicians to consider prophylaxis throughout treatment [90]. The Brazilian consensus takes a broader position: although not formally studied, all patients on eculizumab should receive penicillin prophylaxis; for urgent initiation it recommends prophylactic ciprofloxacin for 2 to 3 weeks until the vaccine takes effect, and revaccination usually every 2.5 to 3 years [17]. Brodsky recommends penicillin 500 mg twice daily, in addition to vaccination, in all patients younger than 45 years, noting that this is expert opinion and differs from other practices [29]. The GeneReviews chapter on atypical haemolytic uraemic syndrome recorded, as a general principle of C5 inhibition, daily prophylactic antibiotics (oral penicillin or a macrolide) for two weeks after vaccination in patients not vaccinated in advance, and continuous prophylaxis throughout treatment in some countries, with additional vaccination against pneumococcus and H. influenzae type b in children, before the serogroup B vaccine appeared in the current labels [82].

All patients should have stand-by medication and an emergency card: Onkopedia specifies co-amoxiclav 1000/250 mg or, in case of allergy, ciprofloxacin 750 mg, to be taken in the event of sudden fever, and immediate medical assessment [18]. The Brazilian consensus specifies empirical ciprofloxacin 750 mg and attendance at the emergency department in case of suspected meningococcal infection [17]. In the United States, eculizumab and ravulizumab are subject to a joint risk evaluation and mitigation strategy (REMS), and pegcetacoplan, iptacopan and danicopan to their own programmes [21, 32–35]. In addition to the meningococcus, the labels warn of other Neisseria species, including disseminated gonococcal disease, S. pneumoniae, H. influenzae and, in immunocompromised or neutropenic patients, Aspergillus [21]. Initiation is contraindicated in the presence of unresolved serious meningococcal infection or without vaccination, unless with prophylactic antibiotic therapy until two weeks after vaccination [21, 30].

7.7. Comparative regulatory status

The most recent revisions of the FDA labels consulted date from 2026 for Soliris (8 July), Empaveli (21 August, with an efficacy supplement approved on 20 August), Fabhalta (16 July), Voydeya (12 May), Bkemv (2 July) and Epysqli (6 July), and from 2025 for Ultomiris (9 October) and PiaSky (11 November) [21, 32–36, 85, 86]. In the European Union, the latest recorded European Commission decisions date from 2026 for Bekemv (28 August), Epysqli (7 August), Aspaveli (20 July), Ultomiris (25 June) and Voydeya (12 January), and from 2025 for Fabhalta (19 December), Piasky (17 October) and Soliris (17 July) [22, 30, 37–39, 87–89]. The scope of each FDA efficacy supplement and of each European Commission decision is not given in the databases consulted, so it is not possible to attribute a specific indication extension to each date without consulting the approval letters and variation reports; in particular, the dates of the paediatric extension of ravulizumab cannot be assigned to a particular supplement [32].

In Portugal, these medicines have a centralised authorisation valid throughout the European Union, and the Portuguese SmPC is the official translation of the European SmPC [30, 93]. In the Infomed search of 27 October 2026, Soliris®, Ultomiris®, Aspaveli®, Piasky®, Bekemv® and Epysqli® returned results; Fabhalta® and Voydeya® did not return results, which does not demonstrate absence of authorisation, given the centralised procedure, and national marketing status and reimbursement conditions remain an access limitation of this review [93]. In Brazil, ANVISA open data of 27 October 2026 show active registrations of Soliris® (eculizumab, AstraZeneca do Brasil, process finalised on 19 August 2024, with the original Alexion registration of 13 March 2017 inactive owing to transfer of ownership), Bekemv® (Amgen, 25 November 2024), Epysqli® (Samsung Bioepis, 2 December 2024), Herculiz® (eculizumab from the Brazilian company Bionovis, active registration of 22 April 2026), Ultomiris® (ravulizumab, AstraZeneca, 15 July 2024), Empaveli® (pegcetacoplan, Pint Pharma, 24 March 2025), Fabhalta® (iptacopan hydrochloride monohydrate, Novartis, 27 January 2025), Voydeya® (danicopan, AstraZeneca, 13 January 2025) and Piasky® (crovalimab, Roche, 10 February 2025) [92]. The open data file gives the registration, holder and status, but not the approved indication or incorporation into the Unified Health System (SUS); the ANVISA electronic package-insert database could not be consulted, so the text of the Brazilian package inserts remains a limitation [92]. The Brazilian consensus places ANVISA approval of eculizumab in March 2017, consistent with the original registration [17].

7.8. Pivotal trials

These trials must be read with caution. The populations differ decisively: TRIUMPH and PRINCE recruited patients not on an inhibitor; studies 302, PEGASUS, APPLY-PNH and ALPHA recruited previously treated patients, with endpoints focused on residual anaemia; study 301 and COMMODORE 2 compared anti-C5 agents with each other in treatment-naive patients. It is not legitimate to compare indirectly percentages of transfusion avoidance or LDH normalisation between trials with different inclusion criteria, comparators and endpoint definitions. The proportions of Asian patients varied widely (57.6% and 47.1% in the arms of study 301; 65.7% and 88.9% in PRINCE; 63.7% and 73.9% in COMMODORE 2), which matters given the pharmacogenetic C5 variant and geographical differences in thrombotic risk [1, 32, 33, 36, 50]. Relative differences should be read alongside absolute rates: in PEGASUS, the 3.84 g/dL difference in haemoglobin started from a mean baseline of 8.7 g/dL in both arms; in APPLY-PNH, the 81.5 percentage point difference results from a 0% rate in the comparator arm, by definition of the composite endpoint [33, 34]. The non-inferiority studies (301, 302 and COMMODORE 2) do not demonstrate superiority and several are open-label, which affects subjective endpoints: in study 301 there was no difference in fatigue measured by the FACIT scale, in an unblinded design [32]. The effect on thrombosis cannot be determined given the size and duration of TRIUMPH, although no thrombotic event was recorded in the eculizumab arm and one in the placebo arm; in the E05-001 extension, with 187 patients and exposure of 10 to 54 months, there were fewer thrombotic events than in the same period before treatment, but most patients were on anticoagulants [21]. In PEGASUS, three patients in the pegcetacoplan arm discontinued because of haemolysis during the randomised period and two needed the every-three-days regimen; between weeks 16 and 48, ten discontinued because of adverse reactions and thirteen had dose adjustments [33]. APPOINT-PNH is single-arm, and PRINCE compared with supportive care and not with an anti-C5 agent, so there is insufficient evidence to choose between terminal and proximal inhibition as first-line therapy [18].

Negative or terminated programmes are part of the evidence. BCX9930, an oral factor D inhibitor, had four trials terminated by sponsor decision; nomacopan (rVA576, formerly Coversin) had its phase 3 trial terminated with the recorded reason of a business decision to close the global phase 3 programme in PNH; two trials, NCT03588026 and NCT03427060, completed with 9 and 1 patients; one transplant trial was terminated after a futility analysis and another was replaced [94]. A phase 3 trial of pozelimab with cemdisiran was terminated with 3 patients, with the recorded reason of a sponsor decision unrelated to efficacy or safety [94].

7.9. Breakthrough haemolysis and extravascular haemolysis

Breakthrough haemolysis is defined, in the label and in Brodsky's practice, as the return of symptoms with a rise in LDH to at least 2×ULN after a previous fall below 1.5×ULN on therapy [29]. The expert consensus definition requires a fall in haemoglobin of at least 1.5 g/dL with a new rise in previously stable LDH above 1.5×ULN and new or worsening symptoms [18, 78]. Two mechanisms are distinguished: pharmacokinetic breakthrough, with free C5 of 0.5 µg/mL or more owing to insufficient drug levels, regular and typically from the tenth day after the dose, and pharmacodynamic breakthrough, with adequate levels but in a complement-amplifying condition (third trimester, infection, major surgery), in which erythrocytes densely coated with C3b cause C5 to adopt a C5b-like conformation that is not inhibited by eculizumab or ravulizumab [29]. In an analysis of 182 patients from Italy and the United Kingdom, breakthrough was described in 75% of patients, with infections as the most frequent cause [18].

The 2024 expert consensus, obtained by modified RAND/UCLA Delphi methodology with 13 experts from the United States and Europe on more than 400 scenarios, with agreement in 77%, proposes a severity classification and corresponding management [78]. Severity is defined by the fall in haemoglobin and by symptoms, in the presence of LDH newly above 1.5×ULN caused by a known amplifying condition: a fall of 1.5 to less than 2.5 g/dL in an asymptomatic patient is mild; with symptoms other than dyspnoea, chest pain or abdominal pain, it is mild if the amplifying condition is stabilising or improving and moderate if it is worsening; with dyspnoea, chest pain or abdominal pain, it is moderate; a fall of 2.5 to 4 g/dL is moderate without these symptoms and severe with them; a fall of 4 g/dL or more is always severe [78]. Transfusion is recommended in severe breakthrough, usually not in mild breakthrough and to be considered in moderate breakthrough; tests (haemoglobin, LDH, reticulocytes, bilirubin, complement levels) are done with increasing frequency, daily if the fall is at least 2.5 g/dL or if the patient is admitted [78].

As for dosing, the consensus distinguishes terminal from proximal inhibitors. On a terminal inhibitor, with a fall of 1.5 to less than 2.5 g/dL, the regimen is maintained if the patient is in the first half of the interval or asymptomatic, and the interval is shortened (the same dose earlier) if the patient is in the second half with dyspnoea, chest or abdominal pain and a worsening amplifying condition; a proximal inhibitor is not added, for lack of data [78]. With larger falls, the interval is shortened, unless the patient is at the start of the interval and the amplifying condition is improving [78]. In patients on ravulizumab, shortening the interval by more than two weeks is hardly feasible for reasons of cost and hospital formulary, so adding a dose of eculizumab is part of the recommendation [78]. On a proximal inhibitor, the first measure is to add a dose of the proximal inhibitor itself and, if ineffective (no improvement in signs or symptoms within 24 to 48 hours), to treat with a terminal inhibitor, acknowledging the lack of safety data for this combination, especially with repeated use [78]. There was no consensus on corticosteroids or on prophylactic anticoagulation, with anticoagulation maintained in those already receiving it [78]. The classification has not been validated with real-world data and is based on clinical opinion, without randomised trials [78].

For mitigation, the consensus recommends scheduling vaccination or elective surgery at the start of the dosing interval for drugs that are not taken daily, or maintaining the regimen for those taken at least daily; in unplanned surgery or trauma, maintaining the regimen if the patient is on a proximal inhibitor or at the start of the interval of a terminal inhibitor, and shortening the interval if in the second half; if the patient is on an oral inhibitor and cannot absorb medication via the gastrointestinal tract for more than 24 hours, switching to a non-oral anti-C5 agent until the oral route is tolerated, with daily tests as an inpatient and every 2 to 3 days as an outpatient [78]. The Brazilian consensus recommends performing elective surgery the day after the last administration of eculizumab, considering an extra dose if there is breakthrough and mandatory perioperative prophylaxis with LMWH [17]. Onkopedia describes management of breakthrough on pegcetacoplan (LDH above 2×ULN with a fall in haemoglobin greater than 2 g/dL): treat the amplifying condition, intensify with 1080 mg on three consecutive days or a single intravenous dose followed by three doses per week, which is outside the terms of the authorisation, transfuse red cell concentrate to dilute the clone, monitor LDH, haemoglobin, bilirubin and reticulocytes, consider therapeutic anticoagulation (mandatory if there is thromboembolism) and, in massive cases, add a terminal inhibitor, also outside the authorisation, in contact with an experienced centre [18]. For iptacopan there are no standardised rules: the amplifying condition is treated, the patient is transfused and hydrated, and a terminal inhibitor may be added outside the authorisation [18]. On danicopan combined with anti-C5, breakthrough episodes are few and self-limiting [18]. The reason breakthrough is more intense under proximal inhibition is threefold: the short half-life amplifies the effect of a missed dose, the erythrocyte clone is much larger (it may exceed 90%) and each uninhibited C5 convertase cleaves several C5 molecules, generating several membrane attack complexes, compared with one complex per C5 molecule escaping anti-C5 [18, 78]. High-dose immunoglobulins increase the catabolism of eculizumab and ravulizumab and may precipitate breakthrough, so they require monitoring or an additional dose [18].

Pharmacokinetic breakthrough on eculizumab responds to increasing the dose, shortening the interval or switching to ravulizumab; in the two largest ravulizumab trials there were no breakthrough episodes associated with a rise in free C5, and there is no difference between the two drugs in pharmacodynamic breakthrough [29]. In extravascular haemolysis, more than half of patients on C5 inhibitors have persistent mild to moderate symptoms and up to 20% still need occasional transfusions, the main cause being extravascular haemolysis [29]. Onkopedia defines clinically relevant extravascular haemolysis as symptomatic anaemia, with or without transfusions, for at least three months on a C5 inhibitor, with significant reticulocytosis and signs of haemolysis, and requires this picture to be distinguished from bone marrow failure and persistent breakthrough [18]. The options are adding danicopan or switching to iptacopan or pegcetacoplan [33–35]. Testing for C3 deposition on PNH erythrocytes by flow cytometry may support estimation of C3-mediated extravascular haemolysis and is under investigation [20].

7.10. Response criteria

Assessment of haematological response under anti-complement therapy follows the criteria of the EBMT (European Society for Blood and Marrow Transplantation) severe aplastic anaemia working party, based on median haemoglobin, LDH and reticulocytes over six months, with categories A (no relevant intravascular haemolysis) and B (with relevant intravascular haemolysis): complete response without transfusions, haemoglobin 12 g/dL or more, LDH up to 1.5×ULN and reticulocytes up to 150/nL; very good response without transfusions, haemoglobin 12 g/dL or more, with LDH above 1.5×ULN or reticulocytes above 150/nL; good response without transfusions with haemoglobin 10 to less than 12 g/dL, bone marrow failure excluded; partial response without transfusions or with occasional transfusions (up to 2 units per six months) and haemoglobin 8 to less than 10 g/dL; minor response without transfusions or with occasional transfusions and haemoglobin below 8 g/dL, or with regular transfusions (3 to 4 units per six months) and haemoglobin below 10 g/dL, or a reduction of at least 50% in transfusions with haemoglobin below 10 g/dL; no response with regular transfusions (more than 6 units per six months) and haemoglobin below 10 g/dL [18]. A bone marrow study is recommended if reticulocytes are below 60/nL, and in patients who refuse transfusions specific thresholds apply (minor response with haemoglobin 6 to less than 8 g/dL and no response below 6 g/dL) [18]. The Brazilian consensus describes the expected chronology of response to eculizumab: reduction in LDH and fatigue in the first week, improvement in quality of life and dyspnoea in 2 to 3 weeks and reduction in transfusions in 2 to 6 months, with a good response in more than two thirds of patients and persistent transfusion dependence in about a quarter [17].

7.11. Pregnancy, lactation and special populations

Complement activation increases in pregnancy, especially after 20 weeks, and PNH in pregnancy is a high-risk situation. Historically, maternal mortality approached 20%, mainly from thromboembolism, and fetal mortality was about 9%, from prematurity; in a retrospective review cited by the IPIG guideline, 5 of 24 women died (all-cause mortality 20.8%; 95% confidence interval: 7.3 to 39.0), three from venous thromboembolism and two from infection, the incidence of clinically apparent venous thromboembolism in pregnancy was about 10%, and in 25% of cases the diagnosis of PNH was made during pregnancy; among 34 births there were 3 infant deaths (perinatal mortality 8.8%; 95% CI 1.9 to 23.7), with normal growth and development in all survivors [1, 29]. In 2005, without experience of complement inhibitors, transfusion was the mainstay of treatment and heparin prophylaxis was recommended from the time pregnancy was diagnosed until at least 6 weeks after delivery, with warfarin contraindicated because of teratogenicity in the first trimester and haemorrhagic risk later, and with platelet monitoring at least every two weeks [1].

The IPIG and International PNH Registry survey changed this practice: 75 pregnancies in 61 women on eculizumab, with no maternal deaths, 3 fetal deaths (4%), 6 first-trimester miscarriages (8%), 22 premature births (29%), an increase in transfusions from 0.14 to 0.92 units per month, platelet transfusion in 16 pregnancies, need to increase the dose or frequency of eculizumab in 54% of pregnancies that progressed beyond the first trimester, LMWH in 88%, 10 haemorrhagic and 2 thrombotic events (both postpartum); eculizumab was detected in 7 of 20 cord blood samples and was not detected in 10 of 10 breast milk samples, with 25 breastfed infants [95]. Onkopedia, citing this series, gives maternal and fetal mortality of 11.6% and 7.2% [18]. A systematic review with meta-analysis of 190 pregnancies in 135 patients, mostly case reports or small series, with eculizumab in 131, recorded a single maternal death, in a woman not on eculizumab, fetal survival of 82% with eculizumab versus 69% without, miscarriages before 24 weeks twice as frequent without eculizumab, and intrauterine death, maternal thrombosis, haemorrhage and pre-eclampsia over-represented but without a significant difference between groups, with prematurity in 32% versus 44% [96].

Eculizumab is the drug of choice in pregnancy, because it does not cross the placenta at concentrations that affect fetal complement and is not detectable in breast milk [29, 30]. The European SmPC for eculizumab recommends adequate contraception during treatment and for at least 5 months after the last dose, recorded fewer than 300 exposed pregnancy outcomes without an increase in malformations, notes that immunoglobulin G crosses the placenta and that the medicine should be given to a pregnant woman only if clearly needed, and states that limited data suggest it is not excreted in breast milk [30]. Pregnant women on eculizumab should not stop it during pregnancy or breastfeeding, and it may be necessary to increase the dose up to 900 mg weekly for breakthrough episodes [17, 18]. The recommended antithrombotic prophylaxis is LMWH: enoxaparin 40 mg subcutaneously daily from the start of the second trimester until at least 8 weeks after delivery, in Brodsky's practice; the Brazilian consensus mentions warfarin as a prophylactic option, which conflicts with the contraindication of warfarin in the first trimester stated by the IPIG guideline and with published practice, so LMWH is the option to follow [1, 17, 29]. The 2024 consensus recommends co-management with a maternal-fetal medicine specialist, tests with increasing frequency (monthly in the first trimester, every two weeks or weekly in the third), a low threshold for increasing the eculizumab dose in the face of worsening haemolysis and an additional dose of eculizumab if delivery occurs at the end of the interval [78].

Data on the newer inhibitors in pregnancy were insufficient in 2024, and Onkopedia recommended avoiding them in pregnancy and when planning one [18]. An international retrospective cohort from five centres in the United States and Europe, published in 2026, compared 19 pregnancies on ravulizumab in 16 patients with 8 earlier pregnancies of the same patients on eculizumab: in the pregnancies on eculizumab there were 3 miscarriages and 1 early preterm delivery for threatened fetal death and severe growth restriction (4 of 8 with miscarriage or early neonatal death); in the 19 pregnancies on ravulizumab there were 19 live births, with a median gestational age of 267 days (interquartile range 259 to 275) and a median weight of 3115 g (interquartile range 2458 to 3349); inhibitor intensification occurred in 10 pregnancies (9 with ravulizumab, by dose increase, shortening of the interval to as little as 7 weeks or an additional dose), anticoagulation in 23 pregnancies (mainly prophylactic LMWH with aspirin), one vena cava thrombosis on ravulizumab in a patient with type 2 PNH, JAK2-positive polycythaemia vera and previous Budd-Chiari syndrome with liver transplantation, during an interruption of anticoagulation for haemoptysis, and infections in 9 pregnancies [97]. Ravulizumab was detected in cord blood at concentrations of 179 and 135 µg/mL (2.3 to 3.8 times lower than maternal levels), with free C5 in the cord of 0.159 and 0.199 µg/mL, and in breast milk at 1.9 µg/mL on the third day after delivery; 16 of 19 mothers (80%) breastfed without apparent adverse effects, and the median follow-up of 16.2 months (interquartile range 4.4 to 40.1) revealed no developmental abnormalities or serious infections [97]. The authors perform tests at least monthly in early pregnancy, intensify at the end of the second and in the third trimester and give prophylactic LMWH throughout pregnancy and for 6 weeks after delivery, and consider it reasonable to continue ravulizumab, a position that differs from the caution of the 2024 Onkopedia guideline and from the label recommendation to stop breastfeeding [18, 32, 97].

Label recommendations on pregnancy and lactation differ between drugs and between jurisdictions. The ravulizumab label states that there are no data in pregnant women, refers to an exposure registry and advises stopping breastfeeding during treatment and for 8 months after the last dose, with effective contraception during treatment and for 8 months afterwards [22, 32]. For pegcetacoplan, the FDA label allows that use may be considered after assessing risks and benefits, whereas the European SmPC states that it is not recommended in pregnancy or in women of childbearing potential not using contraception; effective contraception and a pregnancy test before starting are required, and breastfeeding should be stopped during treatment and for 40 days after the last dose; in monkeys, 28 mg/kg daily (2.9 times human exposure) increased miscarriages and stillbirths, with no increase at 7 mg/kg daily, and fetal exposure was less than 1% of maternal exposure [33, 87]. For iptacopan, use may be considered after assessing risks and benefits, with exposures of 4 to 6 times human exposure in rats and rabbits; the FDA label advises stopping breastfeeding during treatment and for 5 days after the last dose, and the European SmPC leaves the decision between stopping breastfeeding or the drug [34, 88]. For danicopan, breastfeeding should be avoided during treatment and for 3 days after discontinuation [35, 89]. For crovalimab, data are insufficient, immunoglobulin G crosses the placenta with a maximum in the third trimester, breastfeeding is not recommended during treatment and for 9 months after the last dose in the FDA label, and the European SmPC allows that it may be considered if the clinical condition requires it [36, 39].

In paediatric patients, the indications differ: ravulizumab from one month of age in the United States and from 10 kg in the European Union; crovalimab from 13 years and 40 kg in the United States and from 12 years in the European Union, with 12 patients aged 13 to 17 years treated with the adult dose in the COMMODORE trials; eculizumab with a paediatric indication in the European Union and without established efficacy in paediatric PNH in the United States; pegcetacoplan, iptacopan and danicopan in adults only [21, 22, 30, 32–36, 39]. The three eculizumab products available in Europe are also approved in children and adolescents [18]. In the transition from adolescence to adulthood, monitoring of vaccination status, education about infectious risk and adherence, critical with oral inhibitors, are maintained [18, 20].

Agents to avoid or use with caution include iron without monitoring under effective inhibition, prolonged use of corticosteroids, androgens without liver monitoring, warfarin in the first trimester of pregnancy, strong inducers and inhibitors of the metabolic pathways of iptacopan, and substrates of BCRP and P-glycoprotein with danicopan [1, 18, 34, 35]. Pegcetacoplan interferes with the silica reagents used in coagulation panels, artificially prolonging the activated partial thromboplastin time, so these reagents should be avoided [33]. When switching between C5 inhibitors with different epitopes, drug-target-drug complexes arise: type III hypersensitivity reactions occurred in 39 of 201 patients (19%) switching from eculizumab or ravulizumab to crovalimab, 4 without complete recovery at the last visit and 8 serious owing to hospitalisation, and in 2 of 8 patients (25%) switching from crovalimab to eculizumab or ravulizumab, one of them with grade 3 axonal neuropathy; symptoms include arthralgia, rash, fever, myalgia, headache, fatigue, petechiae and abdominal pain, with possible renal changes [36]. Onkopedia describes that these reactions appear at most about one week after starting, regress gradually, manifest mainly as cutaneous vasculitis and more rarely as arthralgia, are treated with topical corticosteroid in mild forms and with prednisolone 20 to 30 mg in more severe ones, and that switching from ravulizumab may give more severe and prolonged reactions [18].

The most frequent adverse reactions differ by class. With eculizumab, in the controlled trial in 196 adults (18 to 85 years, 55% women), headache was recorded in 44% versus 27% with placebo, nasopharyngitis in 23% versus 18%, back pain in 19% versus 9%, nausea in 16% versus 11%, fatigue in 12% versus 2% and herpes simplex in 7% versus 0%, with serious reactions in 9% versus 21% on placebo and no deaths [21]. With pegcetacoplan, the reactions with a frequency of 10% or more were injection-site reactions, infections, diarrhoea, abdominal pain, respiratory tract infection, pain in the extremities, hypokalaemia, fatigue, viral infection, cough, arthralgia, dizziness, headache and rash, with serious reactions in 17% in the APL2-302 trial (infections in 5%) [33]. With iptacopan, they were headache, nasopharyngitis, diarrhoea, abdominal pain, bacterial infection, viral infection, nausea and rash, with grade 1 hypercholesterolaemia in 43% of the 54 patients with normal total cholesterol and thrombocytopenia of any grade in 43% of the 37 with normal platelets in APPLY-PNH, which warrants monitoring of lipids and platelets [34]. With danicopan, the reaction with a frequency of 10% or more was headache, with monitoring of liver enzymes and lipids [35]. Infusion reactions, including anaphylaxis, are listed in the eculizumab label, none of which required treatment discontinuation in the trials [21].

7.12. Developments in the last twelve months

In October 2026, a search of ClinicalTrials.gov returned 79 records of interventional studies in PNH updated since 2025 [94]. The most advanced programme is that of pozelimab (anti-C5) with cemdisiran (RNA interference against C5): the phase 3 trial ACCESS-1 (NCT05133531), with 202 patients, was active and not recruiting, with primary completion expected on 12 October 2026; a phase 3 extension (NCT05744921, 202 patients) and a phase 3 trial in patients with an inadequate response to current treatment (NCT07154745, 35 patients, since June 2026) are recruiting [94]. An exploratory cohort of ACCESS-1 was presented at a scientific meeting in December 2024, with adequate LDH control in 96% versus 80% with ravulizumab, a figure obtained from secondary specialist press sources and not from the original abstract; no marketing application for PNH was found at the FDA or EMA in the search performed, which does not demonstrate that none exists, and the application announced by the sponsor for the first quarter of 2026 concerns cemdisiran in myasthenia gravis and not PNH [94]. The level of evidence is therefore that of a congress presentation and a secondary source, without publication or regulatory decision.

Zaltenibart (OMS906), an antibody against MASP-3 of the lectin pathway, completed a phase 1 trial (NCT05889299, 15 participants, in November 2025) and a phase 2 trial (NCT05972967, 12 patients), with a phase 2 extension active and not recruiting (NCT06298955, 25 patients, expected completion December 2026), without results posted in the registry [94]. Several oral inhibitors in development in China have completed phase 3 trials with eculizumab as comparator: HRS-5965, a factor B inhibitor (NCT06593938, 76 patients, primary completion July 2025; NCT06715943, 39 patients with persistent anaemia on anti-C5, January 2026), with a published result for the phase 2 monotherapy study in treatment-naive patients [94, 98]; and HSK39297, a factor D inhibitor (NCT06799546, 73 patients, November 2025; NCT07052838, 36 anaemic patients on anti-C5, December 2025; phase 2 NCT06561841, 47 patients), with no publication located [94]. Trials of MY008211A, NTQ5082, FXS6837 and XH-S003 are ongoing or about to start [94]. Nucleic acid therapies are at early stages: ALN-CFB (RNA interference against factor B) in a first-in-human phase 1/2 study since February 2026, ADX-038 in phase 1/2, VSA012 in phase 1 with a phase 3 not yet started, SGB-9768 and SLN12140 in phase 2 not yet started [94]. An oral factor B inhibitor, ciprocopan (NXP100), was described as approved by the Chinese regulatory authority in July 2026 for patients not previously treated with complement inhibitors, information obtained from the financial press and not confirmed in a primary regulatory source [94]. In the field of biosimilars, BCD-148 completed a phase 3 extension with 24 patients, and two trials with eculizumab and ravulizumab in a Chinese population have results posted in the registry but not read [94].

On the regulatory front, during the period analysed no new class was authorised in PNH in the United States or the European Union; the changes recorded correspond to label revisions and variation decisions whose exact content is not given in the databases consulted [21, 37]. Proximal inhibition as first-line therapy still lacks direct comparative evidence against terminal inhibition, and germline genotyping of CR1 and C3 to predict extravascular haemolysis is not routinely available, so Onkopedia recommends caution and inclusion of patients in the International PNH Registry [18].

8. Reproductive Options and Family Counselling

8.1. Classical PNH: an acquired disease

PNH due to a somatic PIGA variant is an acquired disorder of a haematopoietic stem cell. Orphanet gives inheritance as not applicable, there is no risk of familial recurrence attributable to the somatic variant and testing of relatives is not indicated [7, 10]. Counselling therefore focuses on other content: explaining that the disease is not hereditary and is not transmitted to children, that the risk for siblings and offspring is not increased, and that the finding of a somatic PIGA variant on a myeloid panel has no family implications [7, 45]. HLA typing of the patient and siblings is done for therapeutic reasons, to assess donor availability in transplant candidates, and not for genetic counselling [1, 17].

In pregnancy, the information to convey is that of acquired risk: increased maternal risk, mainly thromboembolic, and fetal risk of prematurity and loss, with the figures in section 7 [1, 95, 96]. Preconception counselling includes assessment of disease activity, need for an inhibitor, vaccination status, planned anticoagulation and choice of inhibitor, given that eculizumab has the most data in pregnancy and ravulizumab now has a favourable cohort, whereas the proximal inhibitors have insufficient data and restrictive label recommendations [18, 29, 87, 97]. The decision to become pregnant is not contraindicated, but requires joint management by haematology and high-risk obstetrics [1, 17, 78].

8.2. PIGT-associated PNH

PIGT-associated PNH is the most complex counselling situation among the entities described, because it combines a germline allele with a somatic event. In the four patients described, the germline loss-of-function variant was on the maternal allele in the heterozygous state, and the disease arose only when a clone somatically lost the paternal allele through a 20q deletion [2]. Inheritance of the germline variant is likely but has not been formally demonstrated, because no DNA from relatives was available; the allele frequency of 0.0002316 for the c.250G>T variant in the East Asian population, corresponding to an estimated 55,000 carriers in Japan, makes it very likely that the patient inherited it [2]. Known heterozygous carriers are healthy: the same variant was found in the heterozygous state in the mothers of two unrelated Japanese patients with inherited GPI deficiency, without haemolysis or inflammatory symptoms, indicating that PIGT haploinsufficiency does not cause disease [2].

Two distinct implications for the family follow from this picture. The first is the risk of PNH: it requires a rare second somatic event, neither predictable nor measurable with the available sources, so it cannot be quantified and does not justify systematic haematological surveillance of asymptomatic carriers; the sources consulted establish no follow-up protocol for carriers. The second is the risk of inherited GPI deficiency in offspring: if a carrier's partner is also a carrier of a PIGT variant, the risk of MCAHS3 (an autosomal recessive disorder) is 25% in each pregnancy, and the variants causing this disease are hypomorphic rather than null, unlike those found in PIGT-associated PNH [2, 8, 59]. Cascade testing for the germline variant in first-degree relatives is therefore informative mainly for reproductive purposes, and should be offered with an explicit explanation that being a carrier does not mean a predictable risk of PNH. The OMIM entry 615399 (PNH2) classifies inheritance as autosomal dominant with somatic mutation ("AD, SMu"), flags the phenotype–gene relationship as provisional and states that the additional somatic mutation appears necessary for the phenotype; the designation "dominant" reflects the presence of a single germline allele and should not be conveyed to the family as a dominant risk of PNH, given that the disease requires the second somatic event [2, 8, 11].

8.3. Congenital CD59 deficiency

Congenital CD59 deficiency is autosomal recessive. The parents of an affected individual are obligate heterozygous carriers, the risk for each sibling is 25% of being affected, 50% of being a healthy carrier and 25% of not being a carrier, and the offspring of an affected individual will be obligate carriers, with a risk of disease depending on the partner's status [9, 12, 13]. The founder variant p.(Cys89Tyr) has a carrier frequency of 1 in 66 in Jews of North African origin and accounts for practically all pathogenic variants of the gene in this population, with no other variants described in these individuals up to the date of the source [12, 15]. This frequency should not be converted into disease prevalence, because the size and ascertainment of the sample are not given in the source [9]. In a union between two carriers from this population, the risk is 25% per pregnancy; carrier testing is available by targeted sequencing of the variant, and prenatal diagnosis and preimplantation genetic testing are possible when the familial variants are known [12, 15]. The 2026 systematic review notes that most cases come from regions with high consanguinity, in the Middle East and Asia, suggesting additional founder effects and underdiagnosis, and that diagnosis is frequently delayed, which makes counselling extended to the family relevant when a case is identified [13]. In siblings of an affected individual with haemolysis or neurological episodes, testing is diagnostic and immediate, not presymptomatic.

8.4. Germline GPI anchor defects

In germline PIGA defects, with X-linked recessive inheritance, a carrier mother has a 50% risk of transmitting the allele to each child; sons who receive it will be affected and daughters will be carriers [7]. Skewed X-chromosome inactivation in the mother supports carrier status, but its absence does not exclude it, and incomplete penetrance has been documented in one family, with CNTN2 proposed as a modifier, so counselling should include the possibility of variable expression [7]. Pathogenic germline PIGA variants are expected to be hypomorphic, because complete loss of function is incompatible with life [7]. In the autosomal recessive defects of the pathway (PIGN, PIGV, PIGO, PIGL, PIGW, PIGT, PIGS, PIGB, PIGQ, PIGM, PGAP2, PGAP3, PIGY), the risk for siblings is 25%, with carrier testing and prenatal or preimplantation diagnosis possible if the familial variants are known [60, 72, 73]. Several of these disorders cluster in consanguineous families and in founder or recurrent variants, such as PIGL p.(Leu167Pro) in CHIME syndrome, the PGAP3 His284 variants in the Arabian Peninsula and the PIGM promoter variant shared by Middle Eastern, Turkish and Arab families [69, 71, 72]. Germline mosaicism is not documented in the sources consulted for these genes. The essential point of counselling is that these disorders confer no risk of PNH and that PNH confers no risk of these disorders, despite sharing the biochemical pathway [7].

8.5. Counselling of minors and content of the consultation

Genetic testing of an asymptomatic minor is justified when there is clinical benefit in childhood. In congenital CD59 deficiency, early diagnosis of a sibling makes it possible to recognise and treat neurological relapses and haemolysis and to avoid years of non-specific treatment with corticosteroids and immunoglobulins, so testing is appropriate [12, 23]. In PIGT-associated PNH, testing a minor for the germline allele has no immediate clinical benefit, given that being a carrier does not predict disease, and its value is reproductive, so it may be deferred to adulthood. In classical PNH there is no family testing to offer [7]. In any of these situations, the consultation should record the distinction between acquired and inherited disease, the difference between being a carrier and being at risk, the limits of the tests described in section 5, the existence of reference networks and patient associations, and the need to revisit counselling when new data emerge, notably on PIGT-associated PNH, whose genetic basis was established in a very small number of patients [2, 20].

9. Special Considerations: Population Diversity and Epidemiology

9.1. Incidence and prevalence: what each study measures

The epidemiology of PNH is dominated by a problem of case definition: counting clones detected by flow cytometry is not the same as counting patients with clinical PNH. The most robust population-based study is the retrospective analysis of the United Kingdom Haematological Malignancy Research Network (HMRN), which covers a population of 3.8 million in Yorkshire and identified all patients by flow cytometry in a single reference laboratory between 2004 and 2018. A total of 197 patients were identified with detectable clones above 0.01% in at least two cell lineages, of whom 88% had aplastic anaemia, 8% classical PNH and 3% myelodysplastic syndrome; overall incidence was 0.35 cases per 100,000 people per year, corresponding to about 220 new cases per year in the United Kingdom, and overall prevalence was 3.81 per 100,000, equivalent to about 2400 prevalent cases; 5-year overall and relative survival were 72% and 82.7% [16]. The authors' explicit conclusion is that classical haemolytic PNH is rare and represents a small proportion of all patients with detectable PNH cells, most of whom have aplastic anaemia [16]. These figures, included in Orphanet as validated, therefore refer to clones and not to haemolytic disease [10].

The full text of the study specifies the composition of the cohort. Of the 197 patients (median age 59 years, range 5 to 91; 54% men), 175 (88.8%) had aplastic anaemia, 12 (6.1%) haemolytic PNH, 7 (3.6%) MDS and 3 (1.5%) other forms: one pure thrombotic PNH, one haemolytic and thrombotic, and one myeloproliferative neoplasm with JAK2 p.(Val617Phe) and a large clone, who died 4 months after diagnosis [16]. The abstract gives 8% classical PNH, the text refers to 15 patients classifiable as classical PNH and the evolution table counts 14 haemolytic patients by including the haemolytic and thrombotic form, an internal discrepancy in the publication that does not alter the conclusion [16]. The median clone was 0.68% in neutrophils and 0.09% in erythrocytes in aplastic anaemia, versus 63.02% and 29.33% in haemolytic PNH, with mean LDH of 492 versus 1156 IU/L; median age was 59 years in aplastic anaemia, with a bimodal distribution, 42 years in haemolytic PNH and 79 years in MDS [16]. The number of haemolytic cases was insufficient to calculate incidence and prevalence specific to classical PNH, and the authors warn that the results may not apply to Asian countries, where the incidence of aplastic anaemia is higher [16]. Five-year overall survival was 100% in haemolytic PNH, treated with eculizumab when indicated, 97.9% in aplastic anaemia below 40 years, 57.4% in aplastic anaemia from 40 years and 63.6% in MDS; 29.9% of the total had recovery (successful transplantation, sustained loss or reduction of the clone, or sustained improvement in counts), 67.3% below 40 years versus 16.6% from that age, and two patients with classical PNH had a slow, sustained reduction of the clone over 10 years, one of whom no longer needed a complement inhibitor [16].

Estimates of clinical PNH are lower. The Brazilian consensus gives an incidence of about 1 to 2 cases per million and a prevalence of 10 to 20 per million, with incidence in Great Britain and France of about 1.3 per million and 0.13 per 100,000 per year in Yorkshire [17]. Onkopedia estimates prevalence at up to 16 per million and incidence at about 1.3 per million, based on data from the United Kingdom and France, states that there are no reliable data for Germany and acknowledges probable underdiagnosis [18]. Orphanet recorded, for Europe, a point prevalence of 2.0 per 100,000 originating from a 2016 EMA document [10]. The spread between 2 and 16 per million and 3.81 per 100,000 is explained by case definition and test sensitivity, not by biological differences: the more sensitive the screening in patients with aplastic anaemia, the higher the measured prevalence.

The probability of finding a clone depends on the population tested. In series of high-risk patients tested in Spain and Brazil, 563 of 3938 cases (14%) had a clone, distributed across bone marrow failure (34%), haemolysis (24%), cytopenia (8.8%) and thrombosis (1.5%) [20]. In a reference laboratory in the United States, the incidence of clones in risk groups was 6.2% (633 of 10,236 patients tested over nine years), a value that remained consistently between 6% and 7%, with 42% of clones below 1% [42]. In the International PNH Registry, the median granulocyte clone in the first 1600 patients enrolled was 68.1% (range 0.01% to 100%), with 83% in classical PNH versus 35% in aplastic anaemia; 34% of patients with PNH in the setting of bone marrow failure had clones of 50% or more and 40% had clones below 10%, whereas in patients without bone marrow failure 76% had clones above 50% and only 8% below 10% [19]. In patients with aplastic anaemia, the clone may grow and evolve to classical PNH in more than 40% of cases, and 10% to 20% of patients with aplastic anaemia develop overt haemolytic PNH, often years after immunosuppression [18, 20].

9.2. Geography, ancestry and founder effect

The IPIG guideline describes consistent phenotypic differences between regions: bone marrow failure manifestations are more frequent in Asian, Pacific Islander and Hispanic patients, whereas thrombosis and infections are more frequent in American and European patients; the basis of these differences is unknown, but they should be taken into account in the management plan [1]. Even after adjustment for clone size, the prevalence of thromboembolism in patients from Japan, China and Mexico is about 2%, lower than in patients from the United States and Europe, and the risk is higher in white and African American patients, which led the guideline to restrict the proposal of primary warfarin prophylaxis to white and African American patients with more than 50% deficient granulocytes [1]. Conversely, in a Korean cohort arterial occlusions accounted for 39% of thromboembolic events, a figure considered unusually high compared with other cohorts [18]. The proportion of Asian patients in the pivotal trials was high (between 47% and 89% in the different arms), which is relevant to the interpretation of the results [32, 33, 36].

The clearest example of population variation with therapeutic consequences is pharmacogenetic. The C5 variant c.2654G>A, p.(Arg885His), present in 3.5% of healthy Japanese individuals and also in Han Chinese, prevents binding of eculizumab and ravulizumab and accounted for all 11 cases of poor response among 345 Japanese patients treated; an Argentinian patient of Asian ancestry had the p.(Arg885Cys) variant [50]. In patients of East Asian ancestry with an insufficient response to an anti-C5 agent, C5 genotyping and switching to crovalimab, which binds a different epitope, is the evidence-based approach [18, 50].

In the forms with a germline component, the variation reflects a founder effect. The CD59 variant p.(Cys89Tyr) has a carrier frequency of 1 in 66 in Jews of North African origin and accounts for practically all pathogenic variants of the gene in that population [12, 15]. The 2026 systematic review concludes that most published cases come from regions with high consanguinity in the Middle East and Asia, suggesting additional founder effects and underdiagnosis [13]. In PIGT-associated PNH, the c.250G>T variant has an allele frequency of 0.0002316 in the East Asian population, with an estimated 55,000 carriers in Japan, and the same variant was found in two unrelated Japanese families with inherited GPI deficiency [2]. The ClinVar counts for PIGA (184 pathogenic, 33 likely pathogenic and 193 uncertain significance variants) reflect the volume of sequencing in epilepsy panels and not the frequency of the disorders, and the prevalence figures for PNH do not apply to germline PIGA deficiency, which has no established prevalence [7].

9.3. Portugal and Brazil

The sources consulted contain no epidemiological data from Portugal or Brazil for PNH. The national information obtained is limited to medicine registration: in Portugal, the presence in Infomed of Soliris®, Ultomiris®, Aspaveli®, Piasky®, Bekemv® and Epysqli®, with Fabhalta® and Voydeya® returning no result in the search, which does not demonstrate absence of authorisation given the European centralised procedure; in Brazil, nine active registrations at ANVISA, including an eculizumab from a national manufacturer [92, 93]. The 2021 Brazilian consensus, prepared by Brazilian haematologists, is the national reference for diagnosis and treatment and places ANVISA approval of eculizumab in March 2017, but presents no incidence or prevalence data specific to Brazil, relying on international estimates [17]. The absence of data in the sources consulted does not demonstrate absence of published data, and the gap is stated.

9.4. Special populations

About 10% of patients with PNH are younger than 21 years, and the disease is often misdiagnosed and mistreated in this age group [1]. In a retrospective analysis of 26 paediatric cases, the clinical picture was dominated by haemolysis, bone marrow failure and thrombosis, with haemoglobinuria possibly less frequent than in adults, good response to immunosuppression in 6 of 9 patients, but no spontaneous remissions and unfavourable long-term survival (80% at 5 years, 60% at 10 years and only 28% at 20 years), which led to the recommendation of matched sibling transplantation as treatment for childhood PNH at the time; in a Dutch study of 11 children, presentation with bone marrow failure was equally common and 5 patients underwent transplantation, 4 of whom survived [1]. These data pre-date complement inhibitors, which are now approved in paediatric patients in the European Union and, in the case of ravulizumab, from one month of age in the United States [30, 32]. The frequency of PNH clones in acquired aplastic anaemia also differs by age: up to 70% of adults have small clones, compared with about 10% of children [29].

In older people, PNH is less frequent than MDS, and a clone of a few per cent is unusually large for MDS, which may lead to reconsideration of the diagnosis and therapy [19]. Age over 55 years was a factor for worse survival in the French registry [1]. In women of childbearing age, the disease requires the planning described in sections 7 and 8 [95, 97]. The disease mainly affects people of working age, so Onkopedia recommends psychological support, self-help groups and rehabilitation, and notes that in Germany patients are entitled to a severe disability card [18]. Fatigue is reported by more than 80% of patients in several cohorts, including those on C5 inhibitors with extravascular haemolysis, and correlates more with haemolysis and clone size than with the degree of anaemia; disease burden is associated with clones of 10% or more, and the FACIT-Fatigue and AA/PNH QLQ-54 questionnaires are used, the latter not yet validated by the IPIG [18, 20].

9.5. Reference networks, registries and associations

The European reference network for rare haematological diseases is ERN-EuroBloodNet, accessible at eurobloodnet.eu [24]. The IPIG created the International PNH Registry, enrolment in which began in autumn 2004 and in which Onkopedia recommends enrolling patients [1], especially those starting proximal inhibition, given the insufficiency of long-term data; Germany also has a national registry of aplastic anaemia and bone marrow failure and a monthly virtual conference for discussion of difficult cases [18]. The IPIG and International Registry websites were not accessible at the time of this review: the first returns an access error and the second redirects to a parked page without content, so enrolment should be arranged through the reference centre and not via these pages. Among patient associations, the Aplastic Anemia & MDS International Foundation, in the United States, also covers PNH (aamds.org) [99]. In Brazil, ABRALE (Associação Brasileira de Linfoma e Leucemia, the Brazilian Lymphoma and Leukaemia Association) is active in haematological diseases, although its specific scope in PNH was not verified in this review [100]. The domain of the British PNH support association was not responding at the time of this review. Information on the disease for patients and professionals is available in the Orphanet entry ORPHA:447 [10].

10. Illustrative Clinical Case

10.1. PIGT-associated PNH with autoinflammation

The case of the patient identified as G3 in the series by Höchsmann and colleagues illustrates the form with a germline component and the long interval between autoinflammation and haemolysis [2]. He was a German man who began, at 48 years, to have recurrent urticaria, arthralgia and fever; biopsy of the skin lesions showed a mixed inflammatory infiltrate. Between 62 and 65 years he had 19 episodes of aseptic meningitis, with neutrophil-rich cerebrospinal fluid and no identified agent. At 66 years, after an episode of severe haemolysis, paroxysmal nocturnal haemoglobinuria was diagnosed. Flow cytometry showed 60% erythrocytes with a PNH phenotype, and a high fraction of these (87%) was stained by a monoclonal antibody that recognises free GPI not bound to protein, a finding that does not occur in PIGA-related PNH. Molecular study identified a germline PIGT variant, c.197delA, causing a frameshift after residue Tyr66 (normalised as NM_015937.6:c.197del, p.(Tyr66SerfsTer11)), on the maternal allele, and a somatic 20q deletion on the paternal allele that removed all of PIGT and the myeloid common deleted region; methylation analysis showed that the remaining L3MBTL1 allele in the clone was hypermethylated, confirming that the lost allele was the paternal one [2, 6].

Before the diagnosis of PNH, the inflammation had been treated with prednisolone combined with anakinra (an interleukin 1 receptor antagonist) and subsequently with canakinumab (an anti-interleukin 1β antibody); these combinations reduced the urticaria, but not the arthralgia or the episodes of meningitis [2]. Serum amyloid A was raised, up to 10.5 µg/mL (normal value below 5 µg/mL), and soluble interleukin 2 receptor and thymidine kinase were increased before, but not after, the start of eculizumab, supporting the autoinflammatory nature of the picture [2]. With eculizumab, started at 66 years, the patient had no further episodes of meningitis and the urticaria and arthralgia improved; the erythrocyte clone and also the granulocyte clone increased rapidly, a rise in granulocytes not described in PIGA-related PNH that the authors attribute to prevention of membrane attack complex-mediated lysis in these cells [2].

The Japanese patient in the same series, J1, had 121 episodes of meningitis between 53 and 69 years, until he started eculizumab, and PNH was diagnosed at 68 years; in DNA obtained before the onset of recurrent haemolysis, only about 3% of leucocytes were PIGT-deficient, supporting the hypothesis that a small clone produces only autoinflammation and that PNH becomes apparent when the clone grows [2]. Of the four patients in the series, G1 developed a positive direct antiglobulin test after starting eculizumab, indicating extravascular haemolysis, and G2, with severe atherosclerosis, died before it could be confirmed whether he had autoinflammatory symptoms [2].

Figure 1 of the article, consulted in the open-access version, brings together the scheme of GPI anchor biosynthesis in normal cells, in PIGA-defective cells (no GPI synthesis) and in PIGT-defective cells (GPI synthesised but not transferred, with free GPI on the membrane); the graphs of the time course of clone size in monocytes, granulocytes, erythrocytes and reticulocytes in patients G1, G3 and J1, with the arrow marking the start of eculizumab and the rise of the clones after that date; photographs of the urticarial lesions of the chest and left lower limb of G3 and of the dark urine during an episode of haemoglobinuria; and the clinical timelines of G3 and J1, with the episodes of meningitis, haemolysis and severe haemolysis and the periods of effective treatment [2]. Figure 2 shows the map of the 20q region with the germline variant on the maternal allele, the 8 to 18 Mb paternal deletion encompassing PIGT and the myeloid common deleted region, the position of PIGT about 1.2 Mb telomeric and of PIGU about 7.4 Mb centromeric to that region, and the loss of expression of L3MBTL1 and SGK2 in cells with a PNH phenotype [2].

This case does not represent usual practice in PNH: it is an entity described in four patients, in whom sequencing of GPI pathway genes and testing for a 20q deletion were done in a research setting, in the presence of PNH without a somatic PIGA variant and with atypical manifestations [2, 31]. Eculizumab was used within the approved PNH indication, but the effect on autoinflammatory attacks and aseptic meningitis is not an authorised indication and has not been studied in a controlled trial [2, 21]. The approaches described, including the earlier use of anakinra and canakinumab, correspond to therapeutic attempts in a patient without an established diagnosis and not to a recommended algorithm.

10.2. Typical presentation of classical PNH

By way of contrast, the vignette of the first patient described by Brodsky illustrates the thrombotic presentation of classical PNH [29]. A 45-year-old woman presented with right upper quadrant pain, dyspnoea and pleuritic pain; investigation showed right pulmonary thromboembolism and hepatic and splenic vein thrombosis with collateral circulation. The blood count showed leucocytes 3.7×10⁹/L, haemoglobin 7.7 g/dL, platelets 102×10⁹/L, reticulocytes 181,000/mm³ and LDH 647 IU/L (reference range 119 to 226). Flow cytometry showed a clone of 60% in erythrocytes, 88% in monocytes and 94% in granulocytes, and the patient had no macroscopic haemoglobinuria. She was treated with eculizumab, 600 mg intravenously weekly for induction and 900 mg every two weeks thereafter, with a rise in haemoglobin from 7.7 to 9.7 g/dL and a fall in LDH from 647 to 271 IU/L; warfarin was stopped at three months, in line with the author's practice of overlapping anticoagulation and complement inhibition for 3 to 6 months, a practice that differs from the recommendation of indefinite anticoagulation after thrombosis [1, 21, 29]. The case illustrates three points: thrombosis may be the first manifestation, the absence of haemoglobinuria does not exclude the disease and, although thrombosis is more frequent with a granulocyte clone above 50%, it also occurs with smaller clones [29].

11. Prognosis and Long-Term Follow-Up

11.1. Prognosis

Without targeted therapy, PNH shortened life: in the Hammersmith cohort, median survival after diagnosis was 10 years, with 28% surviving at 25 years, and of the 48 patients with a known cause of death, 28 died of venous thrombosis or haemorrhage [28]. In the French registry of 220 patients, median survival was 12 years, and the factors associated with worse outcome were transformation to MDS or acute leukaemia (relative risk 19.1), thrombosis at any site (10.2), progression to pancytopenia (5.5), age over 55 years (4.0) and thrombocytopenia at diagnosis (2.2) [1]. Thrombosis is the main cause of death, accounting for 40% to 67% of deaths; the first thrombotic event may be fatal and increases the risk of death 5- to 10-fold [7, 17]. Mortality with best supportive care was 35% at 5 years and about 50% at 10 years [17]. Median survival before 2007 was between 15 and 20 years [20, 29]. The OMIM summary describes the French series with a mean survival of 15 years and identifies, in addition to the adverse factors already mentioned and multiple therapeutic attempts, a single protective factor: aplastic anaemia preceding the diagnosis of PNH [14].

With adequate complement inhibition, survival approaches that of age- and sex-matched controls, according to real-world data [18, 20, 29]. This gain is not uniform: more than half of patients on C5 inhibitors have persistent mild to moderate symptoms, up to 20% continue to need occasional transfusions and about a quarter remain transfusion-dependent, mainly owing to extravascular haemolysis [17, 29]. Fatigue persists in more than 80% of patients in several cohorts, including those on anti-C5 [18]. In the United Kingdom population study, which mainly includes patients with clones in the setting of aplastic anaemia, 5-year overall survival was 72% and relative survival 82.7%; by group, it was 100% in haemolytic PNH, 97.9% in aplastic anaemia below 40 years and 57.4% in aplastic anaemia from 40 years, with most deaths in the older groups occurring in the first year after diagnosis [16].

Spontaneous remission exists and influences decisions: 12 of 35 patients who survived 10 or more years in the Hammersmith cohort had spontaneous clinical recovery, without detectable PNH cells in erythrocytes and neutrophils, although some affected lymphocytes remained detectable in three of four patients tested [28]. This possibility, together with the heterogeneity of the course, is one of the reasons for considering the timing of transplantation case by case [1]. Clinical improvement or normalisation of haemoglobin should prompt reassessment of the clone, because they may precede its reduction or disappearance, and a fall below 50% deficient granulocytes may change the decision on prophylactic and long-term anticoagulation [1, 19]. None of the patients in the Hammersmith cohort developed leukaemia, but evolution to MDS or acute myeloid leukaemia does occur and is frequently associated with monosomy 7 [28, 44]. For context, monosomy 7 was detected in 3.5% of a North American paediatric series of aplastic anaemia treated with immunosuppression and in up to 5% to 10% of acquired aplastic anaemia in adults [44].

In the forms with a germline component, prognostic data are scarce. In PIGT-associated PNH, of the four patients described, two remained free of meningitis episodes for 3 to 5 years on eculizumab, one developed extravascular haemolysis and one died with severe atherosclerosis [2]. In congenital CD59 deficiency, the natural history includes infection-triggered neurological relapses and recurrent cerebral infarcts, with frequently delayed diagnosis and a variable and transient response to immunomodulators; on eculizumab, the four patients in the phase IIa study had marked neurological improvement in the upper limbs and trunk and more modest improvement in the lower limbs, complete discontinuation of corticosteroids and intravenous immunoglobulin, no relapses for 104 weeks despite intercurrent infections, no hospital admissions, reduced C3bi and C5b-9 deposition on erythrocytes and neutrophils (p<0.0001) and significant improvement on the SF-12 quality of life scale (p<0.003); the 2026 systematic review describes, across the reports as a whole, benefit from eculizumab in some patients but inconsistently [9, 13, 23].

11.2. Surveillance

11.3. Follow-up of the forms with a germline component

In PIGT-associated PNH, haematological follow-up is that of PNH on an inhibitor, with the particular feature that the granulocyte clone may rise rapidly after eculizumab is started, which should not be interpreted as treatment failure [2]. Resolution of inflammatory attacks and meningitis episodes is a clinical outcome to be monitored, and persistence of raised interleukin 18 shows that part of the inflammatory activation is independent of terminal complement inhibition [2]. In congenital CD59 deficiency, surveillance combines haemolysis markers with periodic neurological assessment (motor examination, nerve conduction studies, brain magnetic resonance imaging when indicated) and attention to intercurrent infections as triggers of relapse; with off-label eculizumab, the same vaccination and antibiotic prophylaxis requirements as in PNH apply [9, 23, 90]. The intervals for neurological surveillance are not given in the sources consulted and should be set by the team caring for the patient.

12. Summary and Concluding Remarks

Paroxysmal nocturnal haemoglobinuria is an acquired disorder of a haematopoietic stem cell, caused by a somatic PIGA variant that abolishes the first step of GPI anchor biosynthesis and deprives the progeny of that clone of all GPI-anchored proteins, among them the complement regulators CD55 and CD59. Loss of these two regulators results in complement-mediated intravascular haemolysis, the thrombophilia that is the main cause of death and the smooth muscle symptoms attributable to nitric oxide consumption by free haemoglobin. The somatic variant is not sufficient for clonal expansion, which depends on a conditional survival advantage in an autoimmune context, which explains the clinical continuity with acquired aplastic anaemia and why small clones appear in up to 70% of adults with that disease.

Diagnosis is by flow cytometry, not genetics. It requires demonstration, in fresh peripheral blood, of a population deficient in at least two GPI markers in at least two lineages, with FLAER and appropriate gating markers, and separate quantification of types II and III in erythrocytes, with clone size determined in neutrophils. Sensitivity is a declared property of the assay and not an abstract attribute: 0.1% is sufficient for haemolytic PNH, but 0.01% must be reached to detect the small clones of aplastic anaemia, which requires acquiring half a million events per population. Testing for the PIGA variant is not routinely necessary; it is necessary when the phenotype is atypical, and it was by this route that PIGT-associated PNH was identified, a rare form in which a heterozygous germline variant combines with a somatic 20q deletion, produces free GPI on the membrane and adds to the haemolysis an autoinflammatory picture with recurrent aseptic meningitis. Congenital CD59 deficiency is the third entity in this article: autosomal recessive, constitutional, with absent CD59 and the other GPI proteins preserved, and a phenotype combining chronic haemolysis with relapsing demyelinating polyneuropathy and cerebral infarcts in childhood. Germline defects of the GPI pathway genes share the biochemical lesion but not the disease: they manifest as epileptic encephalopathy and malformations, do not cause haemolysis and are not an indication for complement inhibitors.

Treatment has changed twice in less than twenty years. Terminal complement inhibition with eculizumab, and later with ravulizumab and crovalimab, suppresses intravascular haemolysis, reduces thrombosis and brings survival close to that of the general population, but leaves uncovered extravascular haemolysis mediated by C3 deposition, which is now the main cause of residual anaemia. Proximal inhibition with pegcetacoplan, iptacopan or danicopan addresses this problem, at the cost of strict dependence on adherence, a much larger erythrocyte clone and potentially more intense breakthrough episodes, and a broader infectious spectrum that requires vaccination also against pneumococcus and Haemophilus influenzae type b. There is no direct comparative evidence to choose between terminal and proximal inhibition as first-line therapy. The decision to treat is not based on clone size but on clinically relevant haemolysis and complications; detection of a clone is not by itself an indication for therapy, and in subclinical forms and those associated with bone marrow failure the main target is the marrow disease. Allogeneic transplantation remains the only curative option and is reserved for bone marrow failure that would already indicate it, clonal evolution and refractory complications.

Three practical requirements run through follow-up. The first is infectious risk: the incidence of meningococcal disease on inhibitors is about 2000 times higher and persists despite vaccination, and management combines vaccination against serogroups A, C, W, Y and B, antibiotic prophylaxis, stand-by medication and an emergency card, with a documented disagreement over vaccinating two weeks before starting, as the labels and the ACIP require, or on the day of the first administration, as Onkopedia recommends because of the risk of vaccine-induced haemolytic crisis. The second is recognition and classification of breakthrough haemolysis, with management depending on the inhibitor class, the point in the dosing interval and the underlying amplifying condition. The third is pregnancy, a high-risk situation in which eculizumab is the drug with the most data, in which the dose frequently needs to be intensified in the second and third trimesters, in which prophylaxis with low-molecular-weight heparin is the rule and in which a recent cohort supports continuing ravulizumab, against the caution of earlier guidelines and the restrictions of the labels.

From a genetic standpoint, it is important to keep separate three messages that are easily confused. Classical PNH is acquired, is not transmitted and does not warrant family testing. PIGT-associated PNH has a transmissible germline allele, but being a carrier does not predict PNH, because this requires a rare second somatic event; the value of cascade testing is reproductive, because of the autosomal recessive risk of inherited GPI deficiency. Congenital CD59 deficiency is a classical recessive disorder, with a 25% risk for siblings, a founder variant in Jews of North African origin with a carrier frequency of 1 in 66, and carrier testing, prenatal diagnosis and preimplantation testing available. The C5 variant p.(Arg885His), present in about 3.5% of healthy Japanese individuals, is the clearest example of how population genetics alters therapeutic choice, by making eculizumab and ravulizumab ineffective and indicating crovalimab. Substantial gaps remain: national epidemiology in Portugal and Brazil, direct comparison between inhibitor classes, the long-term prognosis of proximal inhibition, clinical predictors of extravascular haemolysis and the natural history of the forms with a germline component, described in numbers of patients that can be counted in single digits.

13. References

1. Parker C, Omine M, Richards S, Nishimura J, Bessler M, Ware R, et al.; for the International PNH Interest Group. Diagnosis and management of paroxysmal nocturnal hemoglobinuria. Blood. 2005;106(12):3699-709. doi:10.1182/blood-2005-04-1717. PMID: 16051736. PMC1895106.
2. Höchsmann B, Murakami Y, Osato M, Knaus A, Kawamoto M, Inoue N, et al. Complement and inflammasome overactivation mediates paroxysmal nocturnal hemoglobinuria with autoinflammation. J Clin Invest. 2019;129(12):5123-36. doi:10.1172/JCI123501. PMC6877298.
3. Yamashina M, Ueda E, Kinoshita T, Takami T, Ojima A, Ono H, et al. Inherited complete deficiency of 20-kilodalton homologous restriction factor (CD59) as a cause of paroxysmal nocturnal hemoglobinuria. N Engl J Med. 1990;323(17):1184-9. doi:10.1056/NEJM199010253231707.
4. Orphanet. Primary CD59 deficiency (ORPHA:169464) [Internet]. Paris: INSERM; Orphadata data of 2026-06-23 [accessed 2026-10-27]. Available at: https://www.orpha.net/en/disease/detail/169464
5. OMIM® Online Mendelian Inheritance in Man. #612300 HEMOLYTIC ANEMIA, CD59-MEDIATED, WITH OR WITHOUT IMMUNE-MEDIATED POLYNEUROPATHY; HACD59. Johns Hopkins University; last edited 20/06/2017. https://omim.org/entry/612300. Accessed 28 October 2026.
6. VariantValidator (GRCh38, MANE Select). https://variantvalidator.org
7. GeneGuidance. PIGA — clinical gene curation. Version of 26 September 2026.
8. GeneGuidance. PIGT — clinical gene curation. Version of 26 September 2026.
9. GeneGuidance. CD59 — clinical gene curation. Version of 26 September 2026.
10. Orphanet. Paroxysmal nocturnal hemoglobinuria (ORPHA:447) [Internet]. Paris: INSERM; Orphadata data of 2026-06-23 [accessed 2026-10-27]. Available at: https://www.orpha.net/en/disease/detail/447
11. OMIM® Online Mendelian Inheritance in Man. #615399 PAROXYSMAL NOCTURNAL HEMOGLOBINURIA 2; PNH2. Johns Hopkins University; last edited 09/09/2013. https://omim.org/entry/615399. Accessed 28 October 2026.
12. Nevo Y, Ben-Zeev B, Tabib A, Straussberg R, Anikster Y, Shorer Z, et al. CD59 deficiency is associated with chronic hemolysis and childhood relapsing immune-mediated polyneuropathy. Blood. 2013;121(1):129-35. doi:10.1182/blood-2012-07-441857.
13. Alwahiby L, Alsuhim L, Alzahrani M, Albaloshi J, Skrypnyk C, Sridharan K. Congenital/Inherited CD59 Deficiency: A Systematic Review. J Clin Med. 2026;15(19):7751. doi:10.3390/jcm15197751. PMC13666367.
14. OMIM® Online Mendelian Inheritance in Man. #300818 PAROXYSMAL NOCTURNAL HEMOGLOBINURIA 1; PNH1. Johns Hopkins University; last edited 16/04/2024. https://omim.org/entry/300818. Accessed 28 October 2026.
15. Wallace SE, Mirzaa GM. Resources for genetics professionals — genetic disorders associated with founder variants common in the Sephardic Jewish population. 2023 Jan 12. In: Adam MP, Feldman J, Mirzaa GM, et al., editors. GeneReviews® [Internet]. Seattle (WA): University of Washington, Seattle; 1993–2026.
16. Richards SJ, Painter D, Dickinson AJ, Griffin M, Munir T, Arnold L, et al. The incidence and prevalence of patients with paroxysmal nocturnal haemoglobinuria and aplastic anaemia PNH syndrome: a retrospective analysis of the UK's population-based haematological malignancy research network 2004-2018. Eur J Haematol. 2021;107(2):211-8. doi:10.1111/ejh.13640.
17. Cançado RD, Araújo ADS, Sandes AF, Arrais C, Lobo CLC, Figueiredo MS, et al. Consensus statement for diagnosis and treatment of paroxysmal nocturnal haemoglobinuria. Hematol Transfus Cell Ther. 2021;43(3):341-8. doi:10.1016/j.htct.2020.06.006. PMC8446255.
18. Schubert J, Panse J, Bettelheim P, Brümmendorf TH, Burmester PO, Göbel U, et al. Paroxysmal Nocturnal Hemoglobinuria (PNH). Onkopedia Guidelines (DGHO, ÖGHO, SGH-SSH, SGMO); document date: September 2024. Available at: https://www.onkopedia-guidelines.info/en/onkopedia/guidelines/paroxysmal-nocturnal-hemoglobinuria-pnh/@@guideline/html/index.html
19. Dezern AE, Borowitz MJ. ICCS/ESCCA consensus guidelines to detect GPI-deficient cells in paroxysmal nocturnal hemoglobinuria (PNH) and related disorders. Part 1 — clinical utility. Cytometry B Clin Cytom. 2018;94(1):16-22. doi:10.1002/cyto.b.21608. PMID: 29236352.
20. Fattizzo B, Marchetti M, Cannizzo E, Gatti A, Geuna M, Iori AP, et al. Expert consensus on the diagnosis and monitoring of paroxysmal nocturnal hemoglobinuria (PNH): an algorithmic approach in an era of new treatments. Eur J Haematol. 2026 Aug 31 [online ahead of print]. doi:10.1111/ejh.70294. PMID: 42675626.
21. US Food and Drug Administration. SOLIRIS (eculizumab) injection — prescribing information. BLA 125166, Alexion Pharmaceuticals; revision of 2026-07-08 (set id ebcd67fa-b4d1-4a22-b33d-ee8bf6b9c722). Available at: https://api.fda.gov/drug/label.json?search=openfda.brand_name:%22Soliris%22
22. European Medicines Agency. Ultomiris (ravulizumab) — SmPC (EPAR annex I). Alexion Europe SAS; marketing authorisation of 2019-07-02, EC decision of 2026-06-25. https://www.ema.europa.eu/en/medicines/human/EPAR/ultomiris
23. Mevorach D, Reiner I, Grau A, Ilan U, Berkun Y, Ta-Shma A, et al. Therapy with eculizumab for patients with CD59 p.Cys89Tyr mutation. Ann Neurol. 2016;80(5):708-17. doi:10.1002/ana.24770.
24. ERN-EuroBloodNet — European Reference Network on Rare Haematological Diseases. https://eurobloodnet.eu
25. Jaeken J, Morava E. Congenital disorders of glycosylation, dolichol and glycosylphosphatidylinositol metabolism. In: Saudubray JM, Baumgartner MR, García-Cazorla Á, Walter J, editors. Inborn Metabolic Diseases: Diagnosis and Treatment. 7th ed. Berlin: Springer; 2022. Chapter 43, section 43.3.3 PIGA Deficiency (PIGA-CDG). doi:10.1007/978-3-662-63123-2_43
26. Miyata T, Takeda J, Iida Y, Yamada N, Inoue N, Takahashi M, et al. The cloning of PIG-A, a component in the early step of GPI-anchor biosynthesis. Science. 1993;259(5099):1318-20. doi:10.1126/science.7680492.
27. Takeda J, Miyata T, Kawagoe K, Iida Y, Endo Y, Fujita T, et al. Deficiency of the GPI anchor caused by a somatic mutation of the PIG-A gene in paroxysmal nocturnal hemoglobinuria. Cell. 1993;73(4):703-11. doi:10.1016/0092-8674(93)90250-T.
28. Hillmen P, Lewis SM, Bessler M, Luzzatto L, Dacie JV. Natural history of paroxysmal nocturnal hemoglobinuria. N Engl J Med. 1995;333(19):1253-8. doi:10.1056/NEJM199511093331904.
29. Brodsky RA. How I treat paroxysmal nocturnal hemoglobinuria. Blood. 2021;137(10):1304-9. doi:10.1182/blood.2019003812. PMC7955407.
30. European Medicines Agency. Soliris (eculizumab) — Summary of Product Characteristics (EPAR annex I). Alexion Europe SAS; marketing authorisation of 2007-06-20, latest European Commission decision of 2025-07-17. https://www.ema.europa.eu/en/medicines/human/EPAR/soliris
31. Krawitz PM, Höchsmann B, Murakami Y, Teubner B, Krüger U, Klopocki E, et al. A case of paroxysmal nocturnal hemoglobinuria caused by a germline mutation and a somatic mutation in PIGT. Blood. 2013;122(7):1312-5. doi:10.1182/blood-2013-01-481499.
32. US Food and Drug Administration. ULTOMIRIS (ravulizumab-cwvz) injection — prescribing information. BLA 761108, Alexion; revision of 2025-10-09.
33. US Food and Drug Administration. EMPAVELI (pegcetacoplan) injection — prescribing information. NDA 215014, Apellis; revision of 2026-08-21.
34. US Food and Drug Administration. FABHALTA (iptacopan) capsules — prescribing information. NDA 218276, Novartis; revision of 2026-07-16.
35. US Food and Drug Administration. VOYDEYA (danicopan) tablets — prescribing information. NDA 218037, Alexion; revision of 2026-05-12.
36. US Food and Drug Administration. PIASKY (crovalimab-akkz) injection — prescribing information. BLA 761388, Genentech; revision of 2025-11-11.
37. European Medicines Agency. Bekemv (eculizumab) — SmPC (EPAR annex I). Amgen Europe B.V.; marketing authorisation of 2023-04-19 (biosimilar), EC decision of 2026-08-28. https://www.ema.europa.eu/en/medicines/human/EPAR/bekemv
38. European Medicines Agency. Epysqli (eculizumab) — SmPC (EPAR annex I). Samsung Bioepis NL B.V.; marketing authorisation of 2023-05-26 (biosimilar), EC decision of 2026-08-07. https://www.ema.europa.eu/en/medicines/human/EPAR/epysqli
39. European Medicines Agency. Piasky (crovalimab) — SmPC (EPAR annex I). Roche Registration GmbH; marketing authorisation of 2024-08-22, EC decision of 2025-10-17. https://www.ema.europa.eu/en/medicines/human/EPAR/piasky
40. Firth HV, Hurst JA. Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. p. 107–8.
41. Firth HV, Hurst JA. Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. p. 94.
42. Illingworth A, Marinov I, Sutherland DR, Wagner-Ballon O, DelVecchio L. ICCS/ESCCA consensus guidelines to detect GPI-deficient cells in paroxysmal nocturnal hemoglobinuria (PNH) and related disorders. Part 3 — data analysis, reporting and case studies. Cytometry B Clin Cytom. 2018;94(1):49-66. doi:10.1002/cyto.b.21609. PMID: 29236350.
43. ClinVar. National Center for Biotechnology Information. Records VCV000029988 (NM_002641.4(PIGA):c.1234C>T (p.Arg412Ter)), VCV000143194 (NM_015937.6(PIGT):c.1342C>T (p.Arg448Trp)), VCV000619956 (NM_015937.6(PIGT):c.1096G>T (p.Gly366Trp)), VCV000440973 (NM_015937.6(PIGT):c.1582G>A (p.Val528Met)) and VCV000014662 (NM_004972.4(JAK2):c.1849G>T (p.Val617Phe)). https://www.ncbi.nlm.nih.gov/clinvar/. Accessed 1 November 2026.
44. Olson TS, Dickerson KE, Nakano TA, Wlodarski M. Monosomy 7 predisposition syndromes overview. 2021 Jun 10. In: Adam MP, Feldman J, Mirzaa GM, et al., editors. GeneReviews® [Internet]. Seattle (WA): University of Washington, Seattle; 1993–2026. Available from: https://www.ncbi.nlm.nih.gov/books/NBK571103/
45. ESMO Handbook — Cancer Genomics. Lugano: ESMO Press. Chapter 8, Genomics in haematological malignancies. p. 115.
46. Nakashima M, Kashii H, Murakami Y, Kato M, Tsurusaki Y, Miyake N, Kubota M, Kinoshita T, Saitsu H, Matsumoto N. Novel compound heterozygous PIGT mutations caused multiple congenital anomalies-hypotonia-seizures syndrome 3. Neurogenetics. 2014;15(3):193-200. doi:10.1007/s10048-014-0408-y. PMID: 24906948.
47. OMIM® Online Mendelian Inheritance in Man. #615398 MULTIPLE CONGENITAL ANOMALIES-HYPOTONIA-SEIZURES SYNDROME 3; MCAHS3. Johns Hopkins University; last edited 26/05/2023. https://omim.org/entry/615398. Accessed 28 October 2026.
48. Motoyama N, Okada N, Yamashina M, Okada H. Paroxysmal nocturnal hemoglobinuria due to hereditary nucleotide deletion in the HRF20 (CD59) gene. Eur J Immunol. 1992;22(10):2669-73. doi:10.1002/eji.1830221029.
49. ClinVar. NM_000611.6(CD59):c.266G>A (p.Cys89Tyr), VCV 64690: "Pathogenic", criteria provided, multiple submitters, no conflicts; condition "CD59-related disorder". https://www.ncbi.nlm.nih.gov/clinvar/variation/64690/
50. Nishimura J, Yamamoto M, Hayashi S, Ohyashiki K, Ando K, Brodsky AL, et al. Genetic variants in C5 and poor response to eculizumab. N Engl J Med. 2014;370(7):632-9. doi:10.1056/NEJMoa1311084.
51. Jones KL, Jones MC, del Campo M. Smith's Recognizable Patterns of Human Malformation. 8th ed. Philadelphia: Elsevier; 2022. p. 224–5 (Simpson-Golabi-Behmel syndrome).
52. OMIM® Online Mendelian Inheritance in Man. #300868 MULTIPLE CONGENITAL ANOMALIES-HYPOTONIA-SEIZURES SYNDROME 2; MCAHS2. Johns Hopkins University; last edited 16/03/2023. https://omim.org/entry/300868. Accessed 28 October 2026.
53. Orphanet. Multiple congenital anomalies-hypotonia-seizures syndrome type 2 (ORPHA:300496) [Internet]. Paris: INSERM; Orphadata data of 2026-06-23 [accessed 2026-10-27]. Available at: https://www.orpha.net/en/disease/detail/300496
54. Orphanet. Ferro-cerebro-cutaneous syndrome (ORPHA:397922) [Internet]. Paris: INSERM; Orphadata data of 2026-06-23 [accessed 2026-10-27]. Available at: https://www.orpha.net/en/disease/detail/397922
55. Mariani R, Bertola F, Piperno A. HAMP- and HJV-related hemochromatosis. 2005 Feb 17 [updated 2025 Dec 4]. In: Adam MP, Feldman J, Mirzaa GM, et al., editors. GeneReviews® [Internet]. Seattle (WA): University of Washington, Seattle; 1993–2026. Available from: https://www.ncbi.nlm.nih.gov/books/NBK1170/
56. OMIM® Online Mendelian Inheritance in Man. #301072 NEURODEVELOPMENTAL DISORDER WITH EPILEPSY AND HEMOCHROMATOSIS; NEDEPH. Johns Hopkins University; last edited 17/08/2022. https://omim.org/entry/301072. Accessed 28 October 2026.
57. OMIM® Online Mendelian Inheritance in Man. #300209 SIMPSON-GOLABI-BEHMEL SYNDROME, TYPE 2; SGBS2. Johns Hopkins University; last edited 25/10/2024. https://omim.org/entry/300209. Accessed 28 October 2026.
58. Franco B, Bruel AL, Thauvin-Robinet C. Oral-facial-digital syndrome type I. 2002 Jul 24 [updated 2023 May 11]. In: Adam MP, Feldman J, Mirzaa GM, et al., editors. GeneReviews® [Internet]. Seattle (WA): University of Washington, Seattle; 1993–2026. Available from: https://www.ncbi.nlm.nih.gov/books/NBK1188/
59. Orphanet. Intellectual disability-seizures-hypophosphatasia-ophthalmic-skeletal anomalies syndrome (ORPHA:369837) [Internet]. Paris: INSERM; Orphadata data of 2026-06-23 [accessed 2026-10-27]. Available at: https://www.orpha.net/en/disease/detail/369837
60. Slavotinek A. Fryns syndrome. 2007 Apr 18 [updated 2025 Dec 4]. In: Adam MP, Feldman J, Mirzaa GM, et al., editors. GeneReviews® [Internet]. Seattle (WA): University of Washington, Seattle; 1993–2026. Available from: https://www.ncbi.nlm.nih.gov/books/NBK1459/
61. Orphanet. Multiple congenital anomalies-hypotonia-seizures syndrome (ORPHA:280633) [Internet]. Paris: INSERM; Orphadata data of 2026-06-23 [accessed 2026-10-27]. Available at: https://www.orpha.net/en/disease/detail/280633
62. OMIM® Online Mendelian Inheritance in Man. #614080 MULTIPLE CONGENITAL ANOMALIES-HYPOTONIA-SEIZURES SYNDROME 1; MCAHS1. Johns Hopkins University; last edited 03/04/2024. https://omim.org/entry/614080. Accessed 28 October 2026.
63. Plecko B, Mills P. PNPO deficiency. 2022 Jun 23. In: Adam MP, Feldman J, Mirzaa GM, et al., editors. GeneReviews® [Internet]. Seattle (WA): University of Washington, Seattle; 1993–2026. Available from: https://www.ncbi.nlm.nih.gov/books/NBK581452/
64. Orphanet. Hyperphosphatasia-intellectual disability syndrome (ORPHA:247262) [Internet]. Paris: INSERM; Orphadata data of 2026-06-23 [accessed 2026-10-27]. Available at: https://www.orpha.net/en/disease/detail/247262
65. OMIM® Online Mendelian Inheritance in Man. #239300 HYPERPHOSPHATASIA WITH IMPAIRED INTELLECTUAL DEVELOPMENT SYNDROME 1; HPMRS1. Johns Hopkins University; last edited 05/12/2023. https://omim.org/entry/239300. Accessed 28 October 2026.
66. OMIM® Online Mendelian Inheritance in Man. #614749 HYPERPHOSPHATASIA WITH IMPAIRED INTELLECTUAL DEVELOPMENT SYNDROME 2; HPMRS2. Johns Hopkins University; last edited 07/11/2022. https://omim.org/entry/614749. Accessed 28 October 2026.
67. OMIM® Online Mendelian Inheritance in Man. #616025 GLYCOSYLPHOSPHATIDYLINOSITOL BIOSYNTHESIS DEFECT 11; GPIBD11. Johns Hopkins University; last edited 09/09/2022. https://omim.org/entry/616025. Accessed 28 October 2026.
68. OMIM® Online Mendelian Inheritance in Man. #614207 HYPERPHOSPHATASIA WITH IMPAIRED INTELLECTUAL DEVELOPMENT SYNDROME 3; HPMRS3. Johns Hopkins University; last edited 26/09/2025. https://omim.org/entry/614207. Accessed 28 October 2026.
69. OMIM® Online Mendelian Inheritance in Man. #615716 HYPERPHOSPHATASIA WITH IMPAIRED INTELLECTUAL DEVELOPMENT SYNDROME 4; HPMRS4. Johns Hopkins University; last edited 22/12/2023. https://omim.org/entry/615716. Accessed 28 October 2026.
70. OMIM® Online Mendelian Inheritance in Man. #616809 HYPERPHOSPHATASIA WITH IMPAIRED INTELLECTUAL DEVELOPMENT SYNDROME 6; HPMRS6. Johns Hopkins University; last edited 19/12/2022. https://omim.org/entry/616809. Accessed 28 October 2026.
71. OMIM® Online Mendelian Inheritance in Man. #280000 COLOBOMA, CONGENITAL HEART DISEASE, ICHTHYOSIFORM DERMATOSIS, IMPAIRED INTELLECTUAL DEVELOPMENT, AND EAR ANOMALIES SYNDROME; CHIME. Johns Hopkins University; last edited 25/04/2024. https://omim.org/entry/280000. Accessed 28 October 2026.
72. OMIM® Online Mendelian Inheritance in Man. #610293 GLYCOSYLPHOSPHATIDYLINOSITOL BIOSYNTHESIS DEFECT 1; GPIBD1. Johns Hopkins University; last edited 05/08/2022. https://omim.org/entry/610293. Accessed 28 October 2026.
73. OMIM® Online Mendelian Inheritance in Man. #618548 MULTIPLE CONGENITAL ANOMALIES-HYPOTONIA-SEIZURES SYNDROME 4; MCAHS4. Johns Hopkins University; last edited 08/06/2023. https://omim.org/entry/618548. Accessed 28 October 2026.
74. GeneGuidance. PIGB — clinical gene curation. Version of 26 September 2026.
75. OMIM® Online Mendelian Inheritance in Man. #618580 DEVELOPMENTAL AND EPILEPTIC ENCEPHALOPATHY 80; DEE80. Johns Hopkins University; last edited 20/10/2020. https://omim.org/entry/618580. Accessed 28 October 2026.
76. OMIM® Online Mendelian Inheritance in Man. #618143 DEVELOPMENTAL AND EPILEPTIC ENCEPHALOPATHY 95; DEE95. Johns Hopkins University; last edited 12/09/2024. https://omim.org/entry/618143. Accessed 28 October 2026.
77. Firth HV, Hurst JA. Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. p. 309.
78. Dingli D, De Castro C 3rd, Koprivnikar J, Kulasekararaj A, Maciejewski J, Mulherin B, et al. Expert consensus on the management of pharmacodynamic breakthrough-hemolysis in treated paroxysmal nocturnal hemoglobinuria. Hematology. 2024;29(1):2329030. doi:10.1080/16078454.2024.2329030.
79. Nyhan WL, Hoffmann GF, Al-Aqeel AI, Barshop BA. Atlas of Inherited Metabolic Diseases. 4th ed. Boca Raton: CRC Press; 2020. p. 764 (Disorders of vitamin B6 metabolism).
80. Sutherland DR, Illingworth A, Marinov I, Ortiz F, Andreasen J, Payne D, et al. ICCS/ESCCA consensus guidelines to detect GPI-deficient cells in paroxysmal nocturnal hemoglobinuria (PNH) and related disorders. Part 2 — reagent selection and assay optimization for high-sensitivity testing. Cytometry B Clin Cytom. 2018;94(1):23-48. doi:10.1002/cyto.b.21610. PMID: 29236353.
81. Oldaker T, Whitby L, Saber M, Holden J, Wallace PK, Litwin V. ICCS/ESCCA consensus guidelines to detect GPI-deficient cells in paroxysmal nocturnal hemoglobinuria (PNH) and related disorders. Part 4 — assay validation and quality assurance. Cytometry B Clin Cytom. 2018;94(1):67-81. doi:10.1002/cyto.b.21615. PMID: 29251828.
82. Noris M, Bresin E, Mele C, Remuzzi G. Genetic atypical hemolytic-uremic syndrome. 2007 Nov 16 [updated 2021 Sep 23]. In: Adam MP, Feldman J, Mirzaa GM, et al., editors. GeneReviews® [Internet]. Seattle (WA): University of Washington, Seattle; 1993–2026. Available from: https://www.ncbi.nlm.nih.gov/books/NBK1367/
83. Pastori D, Menichelli D, Valeriani E, Pignatelli P. Factor V Leiden thrombophilia. 1999 May 14 [updated 2024 May 16]. In: Adam MP, Feldman J, Mirzaa GM, et al., editors. GeneReviews® [Internet]. Seattle (WA): University of Washington, Seattle; 1993–2026. Available from: https://www.ncbi.nlm.nih.gov/books/NBK1368/
84. Kujovich JL. Prothrombin thrombophilia. 2006 Jul 25 [updated 2021 Feb 4]. In: Adam MP, Feldman J, Mirzaa GM, et al., editors. GeneReviews® [Internet]. Seattle (WA): University of Washington, Seattle; 1993–2026. Available from: https://www.ncbi.nlm.nih.gov/books/NBK1148/
85. US Food and Drug Administration. BKEMV (eculizumab-aeeb) injection — prescribing information. BLA 761333, Amgen; revision of 2026-07-02.
86. US Food and Drug Administration. EPYSQLI (eculizumab-aagh) injection — prescribing information. BLA 761340, Samsung Bioepis; revision of 2026-07-06.
87. European Medicines Agency. Aspaveli (pegcetacoplan) — SmPC (EPAR annex I). Swedish Orphan Biovitrum AB; marketing authorisation of 2021-12-13, EC decision of 2026-07-20. https://www.ema.europa.eu/en/medicines/human/EPAR/aspaveli
88. European Medicines Agency. Fabhalta (iptacopan) — SmPC (EPAR annex I). Novartis Europharm; marketing authorisation of 2024-05-17, EC decision of 2025-12-19. https://www.ema.europa.eu/en/medicines/human/EPAR/fabhalta
89. European Medicines Agency. Voydeya (danicopan) — SmPC (EPAR annex I). Alexion Europe SAS; marketing authorisation of 2024-04-19, EC decision of 2026-01-12. https://www.ema.europa.eu/en/medicines/human/EPAR/voydeya
90. Mbaeyi SA, Bozio CH, Duffy J, Rubin LG, Hariri S, Stephens DS, et al. Meningococcal Vaccination: Recommendations of the Advisory Committee on Immunization Practices, United States, 2020. MMWR Recomm Rep. 2020;69(9):1-41. doi:10.15585/mmwr.rr6909a1. PMC7527029.
91. Amin AB, Collins JP, Dong X, Leidner AJ, Loehr J, Moser CA, et al. Use of the GSK MenACWY-CRM/MenB-4C Pentavalent Meningococcal Vaccine Among Persons Aged ≥10 Years: Recommendations of the Advisory Committee on Immunization Practices — United States, 2025. MMWR Morb Mortal Wkly Rep. 2026;75(1):7-14. doi:10.15585/mmwr.mm7501a2. PMC12782235.
92. Agência Nacional de Vigilância Sanitária (Brazil). Open data on registered medicines (DADOS_ABERTOS_MEDICAMENTOS.csv). https://dados.anvisa.gov.br/dados/DADOS_ABERTOS_MEDICAMENTOS.csv
93. INFARMED. Infomed — Base de dados de medicamentos de uso humano. https://extranet.infarmed.pt/INFOMED-fo/
94. ClinicalTrials.gov. Interventional studies in paroxysmal nocturnal haemoglobinuria updated since 2025-01-01 (API v2). https://clinicaltrials.gov/api/v2/studies
95. Kelly RJ, Höchsmann B, Szer J, Kulasekararaj A, de Guibert S, Röth A, et al. Eculizumab in Pregnant Patients with Paroxysmal Nocturnal Hemoglobinuria. N Engl J Med. 2015;373(11):1032-9. doi:10.1056/NEJMoa1502950.
96. Manning JE, Ciantar E, Griffin M, Kelly RJ. Paroxysmal nocturnal haemoglobinuria in pregnancy — a systematic review with meta analysis. Ann Hematol. 2025;104(4):2517-25. doi:10.1007/s00277-025-06353-7. PMC12052813.
97. Höchsmann B, Gerber GF, Leopold W, Patel BJ, Gavriilaki E, Hefter A, et al. Ravulizumab for treatment of paroxysmal nocturnal hemoglobinuria during pregnancy. Blood Adv. 2026;10(9):3067-74. doi:10.1182/bloodadvances.2025019253. PMC13055122.
98. Zhang L, Liu Z, Zhao X, Hu B, Huang R, Shu C, et al. Efficacy and safety of HRS-5965 monotherapy in complement inhibitor-naïve patients with paroxysmal nocturnal haemoglobinuria. Br J Haematol. 2025. doi:10.1111/bjh.20230. PMC12378936.
99. Aplastic Anemia & MDS International Foundation. https://www.aamds.org
100. ABRALE — Associação Brasileira de Linfoma e Leucemia [Internet]. São Paulo: ABRALE [accessed 2026-10-27]. Available at: https://www.abrale.org.br