Glycosylphosphatidylinositol (GPI) anchoring of proteins is a conserved posttranslational modification in eukaryotes that enables the sorting of GPI-anchored proteins in the secretory and endocytic pathways (1). Constitutional phosphatidylinositol glycan biosynthesis class A (PIGA) protein mutations are rare and have been implicated in severe neurological development impairments, cardiac anomalies, and a high frequency of spontaneous death, especially in childhood (2, 3).

Somatic PIGA mutations, and less commonly, phosphatidylinositol glycan anchor biosynthesis class B (PIGB) and T (PIGT) mutations, are responsible for paroxysmal nocturnal hemoglobinuria (PNH), a complement-driven acquired hemolytic anemia that often manifests as hemoglobinuria, abdominal pain, smooth muscle dystonia, fatigue, and thrombosis (4–6). It has been hypothesized that an autoimmune attack in aplastic anemia (AA) on normal stem cells targets a GPI-linked molecule and therefore preferentially spares PNH stem cells, or that PNH cells are resistant to proapoptotic signals and thus have a growth or survival advantage (7, 8). The risk of clonal malignancy (myelodysplastic syndrome [MDS] and acute myeloid leukemia [AML]) in PNH is relatively low at an estimated 3% to 10% (9–11). Patients with PNH that evolves into MDS/AML usually present with monosomy 7, and mutations in the ASXL1, SETBP1, RUNX1, or RAS pathways (12).

There are two clinically relevant PNH subtypes that modify the therapeutic approach at the time of diagnosis: classic PNH (cPNH), which manifests with hemolytic anemia and thrombosis, and a PNH clone associated with MDS or AA (AA/PNH), with a predominant symptomatology of underlying bone marrow failure (BMF), often without overt hemolysis (13). However, there is a notable population of patients in whom AA/PNH evolves into cPNH after different periods. Bat et al. analyzed the case histories of 319 patients diagnosed with AA at a single center between 2000 and 2017. Of these patients, 117 were diagnosed with AA without a PNH clone, 132 with AA/PNH, 26 with post-AA PNH (secondary PNH), and 44 with primary PNH. Patients were further divided into “expanders” and “stable” groups on the basis of their initial clone size and PNH clone size shifts. Initial clones containing >10% GPI-AP-deficient granulocytes were more likely to expand than those with smaller clones that remained relatively stable. Additionally, patients who received more intense immunosuppressive therapy (IST) containing anti-thymocyte globulin (ATG) were less likely to experience PNH clone expansion than those who were not treated or were treated less intensively with cyclosporin A (CsA) alone. The application of intense IST with ATG treatment has been recommended not only to improve blood counts, but also to reduce the risk of clonal expansion in patients with AA/PNH (14).

The gold standard for the modern diagnosis of PNH is the flow cytometric fluorescein-labeled aerolysin assay (FLAER), which uses proaerolysin with Alexa Fluor 488 to determine the percentage of normal (type I), partially GPI-deficient (type II), and fully GPI-deficient (type III) cells. This assay is based on the ability of aerolysin and proaerolysin to bind selectively and with a high affinity to the GPI anchor (15). Typically, the PIGA mutations responsible for type III defects are frameshift mutations, whereas those responsible for type II defects are missense mutations (16). The presence of type III cells is a hallmark of more severe symptoms resulting from intravascular hemolysis (IVH), especially in the case of a larger clone typical for cPNH. Whereas in the case of partial GPI defects characterized by type II cell occurrence, the hemolytic symptoms might be milder. In most patients, both types of defective cells have been identified at different ratios. However, in patients with AA/PNH with observed expansion of PNH clones, type III subclones are more likely to expand than type II subclones because of their probable survival advantage (14).

Historically, the initial manifestations of hemolytic symptoms and positive tests for non-immune-mediated hemolysis defined cPNH; however, with the development of modern flow cytometry and next-generation sequencing, the presence of PNH clones in different entities has been documented. Notably, PNH clones have been reported in patients with myeloproliferative neoplasms and constitutional BMF syndromes, although they are assumed to be anomalous (17–19). Evidence suggests that small PNH clones can be found in healthy individuals. However, mutations occur in differentiated progenitors rather than in hematopoietic stem cells (20). Fattizzo et al. tested 3085 adult patient samples with AA (531) or MDS (869) using high-sensitivity cytometry and identified PNH clones in 25% of the samples, including small (<10%) and very small (<1%) PNH clones (21). PNH clones, irrespective of size, were good predictors of response to IST or HCT in patients with MDS and AA.

Although PNH clones have been reported in pediatric patients, their clinical significance has not been fully elucidated. Timeus et al. studied 85 children with acquired AA and reported a GPI-negative population in 41% of patients at diagnosis, 48% during IST, and in 45% off therapy (22). In contrast to the findings of an adult study, the therapeutic response was similar between PNH-positive and PNH-negative children. Although PNH clones are strongly associated with AA diagnosis, the management standards of AA acknowledge its presence, but do not involve it in clinical decision-making (23).

Depending on the severity of cytopenia, age, and donor availability, patients with AA/PNH proceed to allogeneic hematopoietic cell transplantation (alloHCT) or are treated with ISTs (such as steroids, CsA, and ATG). Patients in whom pharmacotherapy has failed are referred for delayed alloHCT. Patients with cPNH and associated cytopenia and red cell transfusions can be treated with immunosuppression; however, alloHSCT has historically been the only option to prevent hemolysis-related complications. Currently available therapies for cPNH have dramatically evolved over the past decade, shifting from symptomatic management to targeted therapies aimed at complement system. This revolution began with the introduction of an anti-C5 monoclonal antibody (eculizumab) to reduce terminal complement activation. In response to the need for longer-lasting treatment, ravulizumab was developed with an extended half-life that allows for dosing every 8 weeks and is non-inferior to eculizumab, as validated in clinical studies (24, 25). Another promising C5 inhibitor, crovalimab, is designed for both subcutaneous and intravenous administration (26). Crovalimab binds to a distinct epitope on the beta subunit of C5 and has been shown to be effective in individuals of Asian descent who carry polymorphisms in C5 that affect Arg885, modifying the eculizumab and ravulizumab binding sites and leading to poor IVH control (26, 27).

One study showed that optimal clinical outcomes were not achieved for a considerable proportion of patients with PNH who received eculizumab (28). The majority (79–89%) of patients did not achieve normalized hemoglobin levels, with 76–93% having elevated bilirubin or absolute reticulocyte counts in any 24-week window, which, to a large extent, results from extravascular hemolysis (EVH). Pegcetacoplan, a pegylated peptide targeting the proximal complement protein C3, provides comprehensive control over both IVH and EVH. Clinical trials have shown that pegcetacoplan monotherapy is superior for improving hemoglobin levels and reducing transfusion dependency (29, 30). Iptacopan, an oral factor B inhibitor, targets the alternative complement pathway upstream of C3 and comprehensively inhibits complement-mediated hemolysis in monotherapy (26, 31–33). Another proximal pathway inhibitor, danicopan, is an oral factor D inhibitor designed to prevent the formation of C3 convertase and was approved as an adjuvant treatment to eculizumab or ravulizumab in PNH patients with EVH (34, 35).

Inhibition of the complement activation cascade is associated with a reduction in IVH-related complications, including thromboembolic events. However, in some patients who experience suboptimal control of IVH secondary to incomplete C5 inhibition and episodes of breakthrough intravascular hemolysis (BTH), the risk of thromboembolic complications remains. Although complement-targeted treatment does not seem to modify the mechanical evolution of the underlying bone marrow dysfunction, the data can be ambiguous. A French retrospective comparison study of 123 patients treated with eculizumab after 2005 and 191 historical controls revealed a lower cumulative incidence of AA in the eculizumab group (1% [<1 to 5]) than in the historical control cohort (10% [4 to 8]) (11). Additionally, a similar risk of clonal evolution (MDS or AML) was reported between patients receiving eculizumab (5% [2 to 11]) and historical controls diagnosed after 1985 (5% [2 to 11]). Hill et al. recently revealed that concomitant treatment with eculizumab and IST is safe and effective, regardless of the treatment sequence used (36).

Despite the remarkable advancements in medical therapy, alloHCT remains the only curative option for both cPNH and AA/PNH. This procedure is reserved for patients with cPNH who are unresponsive to complement inhibitors or those with symptomatic BMF. However, transplantation is also considered in countries with limited access to new complement-directed therapies. The potential cure offered by alloHCT must be counterbalanced with the significant risks of the procedure, including graft-versus-host disease (GVHD) and transplant-related mortality (TRM).

Each therapeutic innovation is associated with various levels of effectiveness, quality of life, disease burden, and side effect severity. This review aimed to summarize the knowledge of alloHCT in patients with a PNH clone and identify the optimal direction for future studies.

PNH is one of the oldest hematological entities treated with alloHCT (37). Owing to the non-malignant nature of PNH, bone marrow transplantation (BMT) is preferred over stimulated peripheral blood progenitor cell transplantation (PBPCT) to avoid immune-mediated complications. The outcomes reported in the literature span different eras, with variable patient and donor selections, BMs versus PBPCs as sources of hematopoietic cells, and diverse conditioning regimens (38). Patient characteristics and transplant outcomes are summarized in Table 1 .

Initial attempts at alloHCT have shown a high diversity in preparation regimens, or even a lack thereof. In 1979, Hershko et al. reported the disappearance of PNH after PBPCT in an identical twin donor without pretransplant conditioning (39). In contrast, syngeneic BMT without preparative marrow ablation or IST was unsuccessful in a 10-year-old girl because of the survival advantage of GPI-negative cells over transplanted stem cells and because the PNH clone progressed after BMT (41).

Until fludarabine (Flu) was introduced into clinical practice, patients with BMF syndromes underwent transplantation after receiving various busulfan (Bu) based myeloablative conditioning (MAC) regimens (69, 70). In 1985, Antin et al. reported 4 patients with severe aplastic anemia (SAA)/PNH who underwent BM transplantation from matched sibling donors (MSDs). Three patients were conditioned with cyclophosphamide, procarbazine, and antithymocyte serum (Cy/PCZ/ATS), while the other was conditioned with Bu/Cy/PCZ/ATS (42). There was no reappearance of the PNH clone up to 5 years after transplantation. In 2000, Raiola et al. presented 7 patients with PNH, aged 23–37 years, who received transplants of unmanipulated BM from MSDs (43). The conditioning regimens were Cy (160 mg/kg) and Bu (10–14 mg/kg), and all patients lived with full chimeras, complete hematologic recovery, and no evidence of PNH. In 2001, a Japanese study reported a MSD transplantation after patient consisting of cladribine (6 × 0.11 mg/kg), Bu (2 × 4 mg/kg), and rabbit ATG (2 × 2.5 mg/kg) (44). Donor chimerism of 90–100% was achieved from day 14. Lee et al. published a series of 4 patients with hypercellular marrow who received Bu-Flu-ATG (Bu 4 mg/kg/day for 2 days, Flu 30 mg/m²/day for 6 days, and ATG 20 mg/kg/day for 4 days) as conditioning therapy, and 1 patient with hypocellular marrow was conditioned with Cy-ATG (Cy 50 mg/kg/day for 4 days and ATG 30 mg/kg/day for 3 days) (45). Three patients received hematopoietic grafts from MSDs, and 2 patients received grafts from 1-antigen mismatched unrelated donors (mMUDs), which resulted in the engraftment of three lineages and the disappearance of a PNH clone in all 4 cases.

Since 2000, Bu-based conditioning protocols for BMF syndromes have been less commonly reported, but not completely absent. Chen et al. reported on alloHCT in 4 patients with cPNH and 14 patients with AA/PNH (46). The conditioning regimens included modified Bu (9.6 mg/kg)/Cy (3.6 g/m²)-based regimens for 13 patients and non-myeloablative regimens (Bu 6.4 mg/kg or Cy 100 mg/kg+Flu 150 mg/m²+ATG 10 mg/kg) for 5 patients. The 5-year estimated disease-free survival rate was 80.5±10.2%. A Mexican study reported on 6 patients with PNH who underwent transplantation after conditioning with Bu and Cy (47). The conditioning regimen consisted of 8 mg/kg, Cy 1050 mg/m², and Flu 90 mg/m² (71). An Iranian study presented 13 adult patients with PNH (5 patients with AA/PNH and 8 with cPNH) who underwent alloHCT from MSDs between 2002 and 2014. All patients received Bu-Cy as the conditioning regimen. Survival analysis revealed 5- and 13-year survival rates of 74%, and a significant relationship between a positive history of thrombosis and a high mortality rate (48). A Chinese study included 46 patients who underwent alloHCT (16 patients with PNH and 30 with AA/PNH) between 2017 and 2018 (60). The conditioning regimens were Bu (6.4 mg/kg), Cy (50 mg/kg), and ATG (10 mg/kg) in haploidentical and unrelated donors. The patients with MSDs were treated with Flu (180 mg/m²), Cy (100 mg/kg), or ATG (12.5 mg/kg). The estimated 3-year overall survival (OS) rates in the PNH and AA/PNH groups were 100% (±0.0) and 85.7% (±6.6), respectively (P=0.141); while the rates of GVHD-free and failure-free survival (GFFS) were 100% (±0.0) and 78.7% (±7.7), respectively (P=0.067). Bu-Cy-ATG-based conditioning was administered to 17 cases of unmanipulated haploidentical transplantations, resulting in a 3-year OS probability of 77.8% (±15.2) (50). Another Chinese study reported on 32 patients with PNH who underwent alternative donor alloHCT after Bu (8–16 mg/kg), Cy (2000–3600 mg/m²), or Bu (8–16 mg/kg)-Flu (120–150 mg/m²) (51). Two types of rabbit ATG products were used: ATG-thymoglobulin (Sanofi, Paris, France) at a total dose of 7.5–10 mg/kg in 17 patients, and ATG-fresenius (Astellas Pharma Inc., Tokyo, Japan) at a total dose of 20 mg/kg in 12 patients. Three patients were administered alemtuzumab (1 mg/kg). GVHD prophylaxis consisted of mycophenolate mofetil and methotrexate in all patients, and CsA or tacrolimus. In 31 patients, the PNH clones disappeared within 100 days of alloHCT.

Attempts to reduce conditioning toxicity have resulted in the exploration of various alternative protocols, including low-dose total-body irradiation (TBI) and treosulfan (Treo). In 2001, Hegenbert et al. reported 7 adult patients with high-risk PNH who underwent PBPCT from MSDs (n=2) and MUDs (n=5). Conditioning included Flu 30 mg/m²/day on days -4 to -2 and 2 Gy of TBI on day 0 (52). All 7 patients achieved durable engraftment. After 28 days, a median of 77% (range, 53–96%) T-cell donor chimerism was found in the BM and peripheral blood. T-cell chimerism increased to 91% (range, 76–100%) on day +180 and to 100% in all surviving patients after 12 months. All 7 patients achieved complete disease remission, but 3 patients died of TRM.

Cooper et al. reported 55 patients with PNH who underwent alloHCT with either MAC or reduced-intensity conditioning (RIC) regimens (53). Myeloablative regimens included a Bu dose ranging from 600 to 950 ng/mL in the serum and daily doses for 4–5 days in 18 patients; Treo 42 g/m², Flu 150 mg/kg with or without 2 Gy TBI and rabbit ATG 6 mg/kg in 4 patients; and Cy 120 mg/kg with ATG 60 mg/kg and high-dose TBI (12.0–14.0 Gy) in 4 patients. The RIC regimens comprised 200 mg/kg Cy in 1 patient; 200 mg/kg Cy with 60 to 90 mg/kg ATG in 3 patients; 120 mg/m² Flu, 100 mg/kg Cy, 90 mg/kg ATG, and low-dose TBI (2 Gy) in 2 patients; and 90 to 150 mg/m² Flu with low-dose TBI (2.0–4.0 Gy) in 21 patients. At 5 years, the OS rate was 70%, and neither the choice of conditioning intensity (MAC vs. RIC) nor the PNH subtype (classic vs. having a clone associated with another marrow disorder) affected survival.

In a small, single-center study, Gavrillaki et al. analyzed the outcomes of 24 patients with AA who underwent transplantation between 1992 and 2018. A large PNH clone was found in 3 patients without signs of overt hemolysis or thrombosis. All of these patients achieved complete donor chimerism; however, 2 developed severe acute GVHD, which was associated with thrombotic events during the post-transplant period, leading to multiple organ failure during the early post-transplant period. The presence of PNH clones worsened the outcome of alloHCT in patients with AA (72).

Only one study has described the transplantation of T cell-depleted hematopoietic cells, it involved 5 patients with PNH (54). Conditioning consisted of thoracoabdominal irradiation (4–6 Gy), 100 mg/kg Cy, 150 mg/m² Flu, and ATG or alemtuzumab, whereas graft manipulation involved alpha-beta T cell and CD19 depletion. All patients achieved full donor chimerism and survived beyond 1.7 years after HCT.

Since the 2000s, reduced intensity/toxicity and non-myeloablative protocols for AA/PNH have been explored. In 2012, the European Society for Blood and Marrow Transplantation (EBMT) registry published a retrospective study on 211 patients with PNH, reporting a 5-year OS rate of 68% (±3) in the transplanted group (54%±7 in thromboembolism cases, 69%±5 in AA without thromboembolism, and 86%±6 in hemolytic PNH without thromboembolism or AA) (56). The conditioning protocols comprised 70% Cy-based and 30% Flu-based.

Favorable outcomes of alloHCT have been reported, particularly after Flu-Cy-based conditioning regimens. In 2004, Takahashi et al. reported 5 patients with PNH who were transplanted from MSDs after conditioning with Cy and Flu; all 5 patients survived without evidence of PNH 5 to 39 months after transplantation (57). Non-myeloablative transplant conditioning consisted of Cy 60 mg/kg/day (administered on days -7 and -6), Flu 25 mg/m²/d (administered on days -5 to -1), and ATG 40 mg/kg/day (administered on days -5 to -2). Srinivasan et al. further studied Flu-Cy in heavily transfused patients. Twenty-six patients with transfusion-dependent/ATG-refractory SAA, PNH, or pure red cell aplasia underwent alloHCT from MSD (58). Transplant conditioning consisted of Cy (120 mg/kg) and Flu (125 mg/m²) with or without ATG. The post-transplantation IST consisted of CsA alone or in combination with mycophenolate mofetil or methotrexate. Twenty-four of the 26 patients survived for a median 21 months after transplantation.

Pantin et al. investigated 17 patients with cPNH or AA/PNH (59). Patients received PBPCT from 6/6 MSDs after conditioning with Cy 60 mg/kg/day (days -7 and -6) and Flu 25 mg/m²/day (days -5 to -1). To prevent graft rejection, patients with a significant transfusion history had horse ATG (40 mg/kg/day, days -5 to -2) added to the conditioning regimen (n = 14). Over a median follow-up of nearly 6 years, 15 patients (87.8%) survived without any evidence of PNH, transfusion dependence, or anticoagulation agent use.

Chinese study analyzed the outcomes of 44 patients with PNH who underwent alloHCT from haploidentical donors (25), MSDs (15), and MUDs (4) (49). Patients with MSDs were administered a Flu/Cy-based regimen comprising Flu 30 mg/m²/day intravenously (i.v.) on days -7 to -2, Cy 50 mg/kg/day i.v. on days -4 to -3, and rabbit ATG (Thymoglobulin; Genzyme) 2.5 mg/kg/day i.v. on days -8 to -4. Patients with MUDs or haploidentical donors received a Bu/Cy-based regimen consisting of Bu 3.2 mg/kg/day i.v. on days -7 and -6, Cy 50 mg/kg/day i.v. on days -5 to -2, and ATG 2.5 mg/kg/day i.v. on days -5 to -2. The probabilities of 3-year OS and GFFS were 90.4%±4.6% and 85.6%±5.4%, respectively. The 3-year OS rates of the haplo-donor and MSD groups were 86.5%±7.3% and 93.3%±6, respectively (P=0.520). The 3-year GFFS rates of the haploidentical and MSD groups were 78.3%±8.6 and 92.9%±6.9, respectively (P=0.250).

In a Turkish study, the transplant outcomes of 35 patients with PNH (16 with cPNH and 19 with AA/PNH) after low-intensity conditioning were investigated (62). The most common conditioning protocol for the 24 patients was Flu-Cy. Rabbit ATG was administered in 29 patients, and horse ATG was used in six patients. AlloHCT was performed using MUDs and MSDs in 12 and 23 patients, respectively. The 2-year OS and GFFS rates were 81.2% and 78.1%, respectively. The 2-year OS rates in patients with cPNH and AA/PNH were 81.3% and 79.9%, respectively (P=0.87), and the 2-year GFFS rates in patients with cPNH and AA/PNH were 79% and 76%, respectively (P=0.977).

A Japanese study involving 42 adult patients with PNH who underwent alloHCT reported a 6-year OS rate of 74% (63). The conditioning regimens varied widely; the most commonly used regimens were Flu/Cy-based (N=21) and Cy plus TBI or total lymphoid irradiation (N=10), with ATG administered to 14 patients. The leading cause of death was graft failure, which occurred in 11 patients. Younger age (<30 years) and fewer packed red cell blood transfusions (<20) were associated with improved survival.

One of the largest studies on alloHCT in AA-PNH/PNH was published by the Polish study group (64). This retrospective analysis of patients with PNH who underwent transplantation between 2002 and 2016 included 78 patients, with 27 with cPNH and 51 with AA/PNH. The median age in the group was 29 years old (range, 12 to 65 years). There was a history of thrombosis in 12% and a history of hemolysis in 81%, and 92% of the patients required erythrocyte transfusions before undergoing alloHCT. In 66% of the cases, treosulfan-based conditioning was used, and in 72% of the cases, HCT was performed from a MUD. The 3-year OS rates of patients with cPNH and BMF/PNH were 88.9% and 85.1% (p=ns), respectively. For patients with/without thrombosis, the 3-year OS was 50%/92% (p=ns) in the cPNH group and 83.3%/85.3% (p=ns) in the BMF/PNH group. In the BMF/PNH patients with/without hemolysis, the 3-year OS was 93.9%/62.9% (hazard ratio, 0.13; P = 0.016).

Few studies have included pediatric patients with PNH. An Indian study reported 52 pediatric patients with PNH, 20 with cPNH and 32 with AA/PNH (66). Ten of the 20 (50%) patients received danazol for the treatment of cPNH, and 6 were treated with ATG for AA/PNH. Only two patients with cPNH underwent matched sibling alloHCT after Flu-Cy-horse ATG conditioning. Andolina et al. reported 2 children with PNH who received transplantation after Flu 30 mg/m₂/day on day -5, Cy 60 mg/kg/day on day -3, and horse ATG 30 mg/kg/day on day -4, or Flu 30 mg/m² × 4 days on day -5, Cy 50 mg/kg × 2 days on day -3, low-dose TBI (2 Gy) × 1 on day -1, and alemtuzumab 10 mg × 4 days starting on day -7 (67). Salamonowicz-Bodzioch et al., studied 56 children with SAA, among these, PNH clones were identified in 8, and all patients underwent alloHCT from MUDs or MSDs. The patients received transplantation after a conditioning protocol comprising Flu (120–160 mg/m²), Cy (100–200 mg/kg), and ATG Grafalon (60 mg/kg) or Thymoglobulin (10 mg/kg). The estimated OS was similar between the PNH-positive and PNH-negative subgroups (100% vs. 91%, not significant).