ABSTRACT
Breakthrough hemolysis (BTH) is defined as a hemolytic exacerbation in a patient with paroxysmal nocturnal hemoglobinuria (PNH) treated with complement inhibitors (CIs). In the current era of several terminal and proximal inhibitors, there are no guidelines for defining BTH and its severity, and clinical management is not standardized. This retrospective, observational, and multicentric study evaluated BTH frequency and severity in PNH patients treated with complement inhibitors from 2007 until February 2025 at 10 centers across Italy (9 centers) and the United Kingdom (1 center). A total of 271 BTH events occurred in 102 out of 198 patients (51%), with 36/198 (18%) patients experiencing 3 or more BTH episodes over the 18‐year follow‐up. Most events (55%) were prompted by infections, and 24 (10%) were linked to poor compliance with oral alternative pathway inhibitors. BTH was clinically severe in about half of patients, requiring transfusion in 46%, CI dosing adjustment in 17%, and anticoagulant prophylaxis in 16%. Overall, 5 (2%) breakthrough thrombotic events occurred in patients not on prophylaxis, all during an infection. BTH incidence was 0.19 events per patient‐year, maximal with proximal inhibitors (0.4 events per patient‐year), and lower for anti‐C5 (0.18 per patient‐year). By logistic regression analysis, the main predictors of BTH were more severe disease at diagnosis (increased LDH and a shorter time to first complement inhibitor), treatment with proximal inhibitors, and poorer EBMT response category.
Keywords: breakthrough hemolysis, danicopan, eculizumab, iptacopan, paroxysmal nocturnal hemoglobinuria, pegcetacoplan, ravulizumab, thrombosis
1. Introduction
Paroxysmal nocturnal hemoglobinuria (PNH) is a rare and heterogeneous disorder that has greatly benefited from the advent of complement inhibitors (CIs) [1]. These innovative therapies have significantly controlled intravascular hemolysis and reduced PNH‐related thrombotic risk, improved hemoglobin (Hb), decreased transfusion dependence, and nearly normalized patient survival [2, 3, 4]. However, certain challenges persist with their chronic use, including the risk of infections, which is managed through vaccination protocols and patient education, as well as the occurrence of intercurrent hemolytic episodes known as breakthrough hemolysis (BTH).
BTH is highly variable, ranging from mild alteration of hemolytic markers without clinical manifestations (subclinical BTH) to massive increase in LDH associated with rapid declines in Hb levels, leading to severe anemia, transfusion need, and the re‐emergence of thrombotic risk [5].
Several factors can precipitate BTH, such as insufficient drug plasma levels (pharmacokinetic BTH) or conditions that enhance complement activation (pharmacodynamic BTH), including infections, trauma, and surgery, despite therapeutic drug concentration [5, 6]. Early identification and management of these complement activating conditions (CACs), along with supportive measures to mitigate the clinical severity of BTH, such as transfusions, anticoagulant prophylaxis, and intravenous hydration, with additional doses of background CI and/or giving a rescue dose of eculizumab, is of paramount importance [5].
Currently, there are no guidelines for defining BTH and its severity; clinical management is not standardized. Additionally, several terminal and proximal CIs are now licensed, potentially exposing patients to different risks of BTH with diverse clinical presentations.
This study aimed to analyze a large cohort of PNH patients treated with either terminal or proximal complement inhibitors, focusing on the frequency of BTH episodes, their triggers, clinical severity, management strategies, and the occurrence of thrombotic complications.
2. Methods
2.1. Patient Selection
This was a retrospective, observational, and multicentric study including patients from 10 centers across Italy (9 centers) and the United Kingdom (1 center). Inclusion criteria required a diagnosis of PNH, treatment with a CI, and availability of data on clone size and hematologic parameters both at diagnosis and throughout follow‐up. The study was conducted per the Declaration of Helsinki, and all patients provided informed consent. Data collected included patients treated from 2007 until February 2025.
2.2. Hematologic Parameters and PNH Classification
For each patient, clinical and laboratory data at diagnosis and at the time of BTH event were collected, including age, gender, Hb, LDH, reticulocyte counts, PNH clone percentages on granulocytes, monocytes, and erythrocytes, history of thrombosis, and anticoagulant or antiplatelet therapy. Patients were classified by PNH type, according to the International PNH Interest Group criteria [6], into two categories: classic hemolytic PNH or PNH in the context of another hematologic disorder, as aplastic anemia (AA) or myelodysplastic syndrome (MDS).
2.3. Complement Inhibitors
The type of complement inhibitor used, the initiation and discontinuation dates, and laboratory details (Hb, LDH, and clone size in granulocytes, monocytes, and erythrocytes) during treatment were systematically recorded. CIs administered were classified into broader groups: (a) terminal C5 inhibitors, further divided into eculizumab and newer agents (e.g., ravulizumab, crovalimab, pozelimab with cemdisiran, zilucoplan), and (b) proximal inhibitors, subdivided into the C3 inhibitor pegcetacoplan and other alternative pathway (AP) inhibitors (iptacopan, danicopan plus anti‐C5). The sequence of treatment from the first complement inhibitor to subsequent therapies was recorded for each patient. Response to complement inhibitors was evaluated by the response criteria established by the European Bone Marrow Transplant group (EBMT) [6], and categorized into complete, major, good, partial, minor, and no response.
2.4. BTH Events: Frequency, Severity, Triggers, and Clinical Management
BTH events were defined as any hemolytic episodes characterized by a decrease in Hb and/or an increase in LDH, with at least a clinical sign of hemolysis, such as dark urine, abdominal pain, fatigue, or jaundice, and any instances that required specific treatment. For each BTH event, detailed data were collected, including dates, the type of CI in use at the time, identified triggers, Hb and LDH levels, the presence of ongoing anticoagulant therapy, thrombotic events, or transfusions, including the number of RBC units administered, and any other specific treatments, such as antibiotic therapies or adjustments to the complement inhibitor regimen, either an additional dose or the early administration of the next scheduled dose. The frequency and incidence of BTH events were assessed, including timing from diagnosis and complement inhibitor commencement. Moreover, an analysis of changes (deltas) in Hb and LDH levels compared to baseline values was performed. Finally, we considered BTH outcome as the response to treatment after 1 month from the BTH episode, which was evaluated and stratified according to the criteria established by the EBMT group [6]. The outcome was compared to the baseline response and categorized as worse, equal, or better recovery. Outcomes of BTH were compared among different CIs. Demographic data and diagnostic parameters were evaluated to determine their potential role as predictors for the likelihood of BTH events. Furthermore, possible predictors for clinically significant BTH were explored. Clinically significant BTH was defined as the requirement of a specific clinical action, including transfusions, anticoagulant therapy after thrombotic events, or modifications to the complement inhibitor regimen.
2.5. Statistical Analysis
Data were collected employing standard case report forms (CRF) across all the involved centers. Data were elaborated using Microsoft Excel, RStudio, and JupyterLab (Python).
A descriptive statistics analysis was performed for demographic, clinical‐laboratory, and therapy data. Continuous variables were summarized as median and range and compared with Student's t‐test and Mann–Whitney U test for parametric and nonparametric‐distributed data, respectively. Categorical variables were summarized using rates and compared with Fisher's or χ 2 test, depending on the sample size.
The analysis of BTH predictors, including general event, frequency, and clinical significance risk, was conducted using logistic regression modeling to examine associations across demographic, clinical, and laboratory features. Feature selection was performed using LASSO regression with cross‐validation to identify key predictors. Confounding and multicollinearity were minimized through correlation analysis and variance inflation factor assessment. The final logistic regression model was validated with 10‐fold cross‐validation, reporting the receiver operating characteristic (ROC) area under the curve (AUC), sensitivity, and specificity. In the case of near‐separation, Firth's penalized logistic regression was used to mitigate the separation. Regression coefficients in terms of log‐odds and odds ratios (ORs) were derived to interpret the associations between predictors and the likelihood of BTH events. Individual predictors were also assessed using ROC analysis, and optimal cut‐off values were determined using Youden's index. Further information is reported in the Supporting Information.
3. Results
3.1. Demographics and Hematologic Parameters of PNH Patients at Diagnosis
One hundred ninety‐eight PNH patients undergoing treatment with CIs were enrolled. As shown in Table 1, median age at diagnosis was 39 years (range 16–85), 58% were females, 55% had classic hemolytic PNH, while 31% had an associated AA, 10% MDS, and 4% another hematologic disorder. Prior to initiation of CI, median hemoglobin levels were 9.1 g/dL (range 3.6–15.4), LDH values 4.3 times the ULN (range 0.64–18.5), and reticulocyte values 130 × 109/L (14.5–541). Median PNH clone size was 80% (range 10–100%), 79% (10%–99.6%), and 23% (10%–100%) on neutrophils, monocytes, and erythrocytes respectively. Thrombotic events were reported in 42 patients (21%) at/after PNH diagnosis, with a total of 74 thromboses (88% venous and 12% arterial events), more than half (40%–54%) in splanchnic territory; 54 patients were on anticoagulants, either for primary (20, 37%) or secondary prophylaxis (34, 63%; 8 patients with deep venous thrombosis of lower limbs had not been given secondary prophylaxis), and 4 were on anti‐platelet agents for reasons other than PNH (atrial fibrillation in 2, arterial hypertension and diabetes in 2).
TABLE 1.
Baseline clinical and laboratory characteristics (N = 198).
| Median age at diagnosis, years, n (range) | 39 (16–85) |
| < 18 years, n (%) | 14 (7) |
| 18–65 years, n (%) | 151 (76) |
| > 65 years, n (%) | 33 (17) |
| Gender | |
| Males, n (%) | 84 (42) |
| Females, n (%) | 114 (58) |
| PNH type | |
| Classic hemolytic, n (%) | 108 (55) |
| MDS, n (%) | 20 (10) |
| AA, n (%) | 62 (31) |
| Other, n (%) | 8 (4) |
| Laboratory data at diagnosis | |
| Hb, g/dL, median (range) | 9.1 (3.6–15.4) |
| LDH × ULN, median (range) | 4.3 (0.64–18.5) |
| Reticulocytes, ×109/L, median (range) | 130 (14.5–541) |
| PNH clone size | |
| Neutrophil clone, % (range) | 80 (10–100) |
| 20%–50%, n (%) | 22/151 (15) |
| > 50%, n (%) | 110/151 (73) |
| Monocyte clone, % (range) | 79 (10–99.6) |
| Erythrocyte clone, % (range) | 23 (10–100) |
| Thrombotic history and anticoagulant therapy | |
| Patients with thrombotic history, N (%) | 42 (21) |
| Total thromboses, N | 74 |
| Arterial, N (%) | 9 (12) |
| Venous, N (%) | 65 (88) |
| Patients on anticoagulant therapy, N | 54 |
| Prophylaxis, N (%) | 20 (37) |
| Secondary prophylaxis, N (%) | 34 (63) |
| Anticoagulant discontinuation, N (%) | 24 (44) |
Abbreviations: AA aplastic anemia; Hb hemoglobin; LDH lactate dehydrogenase; MDS myelodysplastic syndrome; PNH paroxysmal nocturnal hemoglobinuria.
3.2. Therapy With Complement Inhibitors
Table 2 provides an overview of the complement inhibitors used and the response to therapy, with 198 patients treated over an observation period of 18 years. Eculizumab was the most frequently used in 175/198 (88%) patients, followed by ravulizumab and other C5 inhibitors in 108 (55%), AP inhibitors in 28 (14%), and pegcetacoplan in 29 (15%) subjects (details in Figure S1). The duration of therapy varied by the type of complement inhibitor, with a median of 51 months for eculizumab and 10 months for pegcetacoplan. The median number of complement inhibitors received by each patient was 2 (1–4), resulting in 382 treatments in 198 patients. Response data were available for 327/382 treatments. By evaluating the best response obtained, an optimal response (complete, major, and good) was observed in 220/327 CI courses (68%), mostly with pegcetacoplan (25, 92%), followed by AP inhibitors (19, 76%), other C5 inhibitors (86, 75%), and eculizumab (90, 56%). The sequence of the various complement inhibitors used and the relative responses are detailed in Tables S1 and S2: eculizumab was the most common first inhibitor, ravulizumab the second frequent CI, followed by proximal inhibitors; the rate of optimal response increased from the first to the third inhibitor. Regarding the dynamic changes of PNH clone size, neutrophil clone size increased to a median of 91% (10–100), monocyte to 90% (10–100), and erythrocyte to 43% (range 10–100) after eculizumab (vs 23%, p ≤ 0.0001 pre‐treatment). During treatment with proximal inhibitors, median clone size on different cell populations further increased as compared to C5 inhibitors, particularly on erythrocytes (96% vs. 44%, p < 0.00001; Table S3).
TABLE 2.
Complement inhibition therapy.
| Terminal CIs | Proximal CIs | Alternative pathway inhibitors | |||
|---|---|---|---|---|---|
| Eculizumab | Other C5 inhibitors | Pegcetacoplan | Total | ||
| Complement inhibitors used, N (%) | 187 (49) | 134 (35) a | 29 (8) | 32 (8) b | 382 (100) |
| Patients treated, N (%) | 175 (88) | 108 (55) | 29 (15) | 28 (14) | 198 |
| Time from diagnosis, months, median (range) | 12 (0–45) | 81 (0–103) | 48 (8–81) | 80 (10–84) | 41 (0–103) |
| Duration of therapy, months, median (range) | 51 (6–249) | 24 (6–84) | 10 (6–31) | 23 (6–61) | 30 (6–249) |
| Response rates | |||||
| Available data, n (%) | 161/187 (86) | 114/134 (85) | 27/29 (93) | 25/32 (78) | 327/382 (86) |
| Complete, n (%) | 29 (18) | 33 (29) | 13 (48) | 9 (36) | 84 (26) |
| Major, n (%) | 3 (2) | 6 (5) | 1 (4) | 1 (4) | 11 (3) |
| Good, n (%) | 58 (36) | 47 (41) | 11 (40) | 9 (36) | 125 (39) |
| Partial, n (%) | 52 (32) | 22 (19) | 1 (4) | 5 (20) | 80 (24) |
| Minor, n (%) | 13 (8) | 5 (5) | 1 (4) | — | 19 (6) |
| No response, n (%) | 6 (4) | 1 (1) | — | 1 (4) | 8 (2) |
Abbreviation: CIs, complement inhibitors.
120 ravulizumab, 8 crovalimab, 4 zilucoplan, 2 pozelimab/cemdesiran.
14 iptacopan, 18 danicopan plus anti‐C5.
3.3. Breakthrough Hemolysis
A total of 271 BTH events occurred in 102 out of 198 patients (51%), with 36/198 (18%) patients experiencing 3 or more BTH episodes over the 18‐year follow‐up. As shown in Table 3, most events (133/238 with available data on trigger, 55%) were prompted by infections, 52 (22%) could not be attributed to any known cause, and 24 (10%) were linked to poor compliance with oral AP inhibitors. Less common triggers included acute inflammatory state (9, 4%, such as cholelithiasis, pancreatitis, renal colic, or surgical interventions), vaccinations (8, 3%), and pregnancy (6, 3%, one spontaneous pregnancy termination). The median Hb level during BTH was 7.9 g/dL (3–13.6 g/dL), with a median drop from baseline of 1.65 g/dL (0 to 6.8 g/dL). The median LDH value was 2.1 times the ULN (1.5–20.8 × ULN), with a median delta increase of 0.6 × ULN (0.3 to +11.7 × ULN) from baseline. More specifically, among the 202 events with available Hb values pre‐ and post‐BTH, Hb decreased by > 2 g/dL in 87 (43%). Among the 169 events with available LDH levels pre and post‐BTH, the delta increase was > 3 × ULN in 25 (15%). For the 152 events with complete data (Hb, LDH, and transfusion needs) after 1 month from BTH (without switch to another CI), the hematological response was re‐categorized according to EBMT criteria [98]. Interestingly, in 37 (26%) BTH events, the response worsened compared to pre‐BTH status without association with the type of PNH (i.e., PNH associated with AA/MDS, nor with the severity of BTH), whilst in 86 (59%) remained unchanged, and in 22 (15%) improved.
TABLE 3.
Breakthrough hemolysis (BTH).
| BTH events, n | 271 |
| Under terminal CI, n (%) | 241 (89) |
| Under proximal/AP CI, n (%) | 30 (11) |
| Median BTH events per patient, n (range) | 1 (0–17) |
| Patients experiencing BTH events, n (%) | 102/198 (51%) |
| ≥ 3 BTH, n (%) | 36/198 (18%) |
| Under terminal CI, n | 109/241 (45) |
| Under proximal/AP CI, n | 19/30 (63) |
| Time from CI initiation to BTH event, months, median (range) | 21 (0–168) |
| Triggers of BTH events | |
| Infections, n (%) a | 133 (55) |
| Unknown, n (%) | 52 (22) |
| Compliance, n (%) | 24 (10) |
| Acute inflammatory state, n (%) | 9 (4) |
| Vaccination, n (%) | 8 (3) |
| Pregnancy, n (%) | 6 (3) |
| Others, n (%) | 6 (3) |
| Laboratory details of BTH events | |
| Hb, g/dL, median (range) | 7.9 (3–13.6) |
| Hb drop, g/dL, median (range) | 1.65 (−2–6.8) |
| 0–2, n (%) | 88/202 (44) |
| ≥ 2, n (%) | 87/202 (43) |
| LDH × ULN, median (range) | 2.1 (1.5–20.8) |
| LDH delta increase × ULN, median (range) | 0.6 (0.3–11.7) |
| ≥ 3, n (%) | 25/169 (15) |
| BTH outcome | |
| Available data, n (%) | 152/271 (56) |
| Complete, n (%) | 20 (13) |
| Major, n (%) | 8 (5) |
| Good, n (%) | 48 (32) |
| Partial, n (%) | 41 (27) |
| Minor, n (%) | 10 (7) |
| No response, n (%) | 25 (16) |
| Unavailable data, n (%) | 119/271 (44) |
| Response recovery following BTH event | |
| Worse, n (%) | 37/145 (26) |
| Equal, n (%) | 86/145 (59) |
| Better, n (%) | 22/145 (15) |
Abbreviations: AP, alternative pathway; CI, complement inhibitor; Hb, hemoglobin; LDH, lactate dehydrogenase.
This category included a variety of etiologies: 16 out of 46 with available data (35%) had a bacterial origin, of which 12 (75%) urinary tract infections (UTI), and 30 cases (65%) a viral origin, including 10 (33%) SARS‐CoV‐2 infections.
3.4. BTH Clinical Management
As highlighted in Table 4, 124 out of 271 (46%) BTH events required transfusions, with a total of 308 RBC units transfused (median of 1 unit per patient, 0–45). The second most frequent intervention was anticoagulant prophylaxis in 43/271 (16%) events, of which 21 (49%) with low molecular weight heparin. Notably, 5 (2%) breakthrough thrombotic events occurred in patients not on prophylaxis: 4 deep venous thrombosis (DVT), 1 complicated by pulmonary embolism, and 1 internal jugular vein thrombosis not catheter related. As detailed in Table S4, all events occurred during an infection (2 sepsis, 1 pneumonia, 1 urinary tract infection, and 1 gastroenteritis), and only one patient had a prior history of DVT; two were managed with low molecular weight heparin, one with unfractionated heparin, one with warfarin, and one with a direct oral anticoagulant. Other treatments administered during BTH events were antibiotics, employed in 30 events (11%, amoxicillin/clavulanic acid in 11, 37%). Adjustments to the CI schedule were performed in 47 BTH episodes (17%), including either the administration of an additional inhibitor dose or the early administration of the next scheduled dose (as detailed below). Importantly, complement inhibitor therapy was never discontinued during a BTH event. Notably, 4 patients switched CI within 1 month from the BTH start date (2 to pegcetacoplan and 2 to AP inhibitors). Two patients switched CI within 2 weeks (1 to pegcetacoplan and 1 to AP inhibitors).
TABLE 4.
Clinical management of BTH events, N = 271.
| Transfused RBC units, n | 308 |
| Transfused RBC units per patient, median (range) | 1 (0–45) |
| BTH events requiring transfusions, n (%) | 124/271 (46) |
| Transfused patients, n (%) | 59/102 (58) |
| Antibiotic therapies, n (%) | 30 (11) |
| Anticoagulant therapies, n (%) | 43/271 (16) |
| Additional or early CI administration, n (%) | 47 (17) |
| Thromboses, n (%) | 5/271 (2) |
Abbreviations: BTH, breakthrough hemolysis; RBC, red blood cell.
3.5. BTH Characterization Per Complement Inhibitor Category
Table 5 summarizes the distribution of the BTH events according to the ongoing CI. Most of them (181/271, 67%) occurred on eculizumab; 60 events (22%) occurred with other C5 inhibitors (48 with ravulizumab, and 12 with other C5 inhibitors), 11 (4%) with pegcetacoplan and 19 (7%) with AP inhibitors. After correcting for the time of exposure, the overall BTH incidence was 0.19 events per patient‐year of treatment: 0.18 for eculizumab and other C5 inhibitors, 0.43 for pegcetacoplan (p = 0.004 versus eculizumab), and 0.3 for AP inhibitors [0.45 for iptacopan (p = 0.03 versus eculizumab), and 0.20 for danicopan plus anti‐C5] (Figure 1, Table S5). Median time from CI treatment to the first BTH event was shorter for pegcetacoplan (7 months, 0–18) compared to eculizumab (34, 1–168), other anti‐C5 drugs (13, 0–72), and AP inhibitors (11, 2–55).
TABLE 5.
BTH characterization per complement inhibitor category.
| Eculizumab | Other C5i | Pegcetacoplan | Alternative pathway inhibitors | Total | |
|---|---|---|---|---|---|
| BTH events, n (%) | 181 (67) | 60 (22) | 11 (4) | 19 (7) | 271 (100) |
| BTH incidence rate (events/patients/year of treatment) | 0.18 | 0.18 | 0.43 | 0.3 a | 0.19 |
| Laboratory details | |||||
| Median Hb, g/dL (range) | 7.8 (3–12.9) | 8.3 (4.1–13.6) | 7.3 (5.5–11.7) | 9.3 (6.8–13.2) | 7.9 (3–13.6) |
| Median Hb drop, g/dL (range) | 1.65 (−2–6.8) | 1.5 (−1.2–6.2) | 3.4 (0.1–6) | 1.8 (−1.3–4.9) | 1.7 (−2–6.8) |
| 0.1–2, n BTH (%) | 53/124 (43) | 21/44 (47) | 3/11 (27) | 5/16 (31) | 82/195 (42) |
| ≥ 2, n BTH (%) | 52/124 (42) | 18/44 (41) | 8/11 (73) | 8/16 (50) | 86/195 (44) |
| Median LDH × ULN, (range) | 1.92 (1.5–14) | 2.1 (1.6–10.6) | 7.35 (2.6–20.8) | 3.7 (1.5–8.34) | 2.11 (1.5–20.8) |
| Median LDH × ULN increase, (range) | 0.4 (0.3–11.7) | 0.35 (0.7–3.1) | 5.31 (0.5–20.1) | 2.2 (0.4–5.8) | 0.6 (0.3–20.1) |
| ≥ 3, n (%) | 9/92 (9) | 1/38 (3) | 7/10 (70) | 5/13 (38) | 22/153 (14) |
| Clinical impact and management of BTH events | |||||
| BTH requiring transfusions, n (%) | 83/162 (51) | 26/51 (51) | 9/11 (82) | 6/15 (40) | 124/239 (52) |
| Patients requiring transfusions, n (%) | 40/76 (53) | 16/33 (48) | 5/7 (71) | 6/12 (50) | 59/102 (58) |
| Thromboses, n (%) | 2/141 (1) | 3/56 (5) | 1/10 (10) | 0/18 (0) | 5/225 (2) |
| Therapy modification (excluding switch), n (%) | 39/181 (22) | 4/60 (7) | 3/11 (27) | 1/19 (5) | 47/271 (17) |
| Clinically significant BTH events, n (%) | 104/181 (57) | 30/60 (50) | 10/11 (91) | 7/19 (37) | 151/271 (56) |
Abbreviations: BTH, breakthrough hemolysis; Hb, hemoglobin; LDH, lactate dehydrogenase.
0.4 with iptacopan, 0.19 with danicopan plus anti‐C5.
FIGURE 1.
Incidence of breakthrough hemolysis (BTH) according to the complement inhibitor utilized, by Kaplan –Meier method. [Color figure can be viewed at wileyonlinelibrary.com]
The median Hb drop observed during BTH events was higher in the pegcetacoplan group (3.4 g/dL) compared to anti‐C5 (1.65 g/dL for eculizumab and 1.5 g/dL for other C5 inhibitors) and AP inhibitors (1.8 g/dL). In detail, an Hb drop ≥ 2 g/dL occurred in 8 out of 11 (73%) BTH events on pegcetacoplan, versus 52/124 (42%) with eculizumab, 18/44 (41%) with other C5 inhibitors, and 8/16 (50%) with AP inhibitors (p = 0.001). Accordingly, the median LDH level during BTH events was the highest in patients on pegcetacoplan (7.35, 2.6–20.8 × ULN) and AP inhibitors (3.7, 1.3–8.34 × ULN) versus eculizumab and other C5 inhibitors (1.92 and 2.1, respectively).
Concerning management, transfusions were most frequently required upon pegcetacoplan (9/11, 82%) versus eculizumab (83/162, 51%), other C5 inhibitors (26/51, 51%), and AP inhibitors (6/15, 40%). Thrombotic events associated with BTH were only observed in patients on eculizumab (2/141, 1%) and ravulizumab (3/56, 5%) all hospitalized because of infections. Modifications of the CI schedule were more common in the pegcetacoplan (3/11, 27%, 1 receiving 3 consecutive days of pegcetacoplan and 2 adding eculizumab) and eculizumab (39/181, 22%, all anticipating the next eculizumab dose) groups. BTH with clinical impact (i.e., those associated with thrombotic events, need for transfusion, or requiring a modification of the complement inhibitor schedule) were 151/271 (56%), with a higher prevalence (10/11, 91%) with pegcetacoplan, followed by eculizumab (104/181, 57%), other C5 inhibitors (30/60, 50%), and AP inhibitors (7/19, 37%).
Concerning triggers (Table S6), infections were confirmed as the most common cause of BTH among all CI categories (32%–49%); whereas 6 out of 19 patients (32%) on oral AP inhibitors had compliance issues, such as delays or interruptions in therapy. The poor compliance did occur both with iptacopan and danicopan. The number of missed doses is unknown.
Focusing on recovery after 1 month from BTH (Tables S7 and S8), the worst response was observed with pegcetacoplan (4/9, 44.5%), followed by AP inhibitors (4/11, 36%), eculizumab (25/88, 28%), and other C5 inhibitors (4/37, 11%).
3.6. Features Associated With BTH Event Risk
To identify characteristics associated with BTH events, a cross‐validated logistic regression model was developed, employing features selected through LASSO regression. Variables with a statistically significant association (p < 0.05) with BTH occurrence included female gender (OR = 2.75, p < 0.001), increased LDH levels at diagnosis (OR = 1.14, p = 0.005), a shorter time from diagnosis to the first CI (OR = 1.003, p = 0.001), the administration of proximal inhibitors (OR = 4.85 compared to eculizumab, p = 0.0004), and poorer responses to CI (linear trend with OR = 4.62 and p = 0.0005 for worse responses compared to complete and major responses). Conversely, PNH/MDS was less commonly associated with BTH events compared to classic hemolytic and PNH/AA (OR = 0.26, p = 0.004). The model demonstrated moderate predictive performance, with a cross‐validated AUC of approximately 0.82 (Figure 2). By focusing on the frequency of BTH, we observed that higher monocyte PNH clone size at diagnosis (OR = 23, p = 0.0003) and treatment with proximal inhibitors (OR = 2.51, p = 0.01) were associated with the occurrence of 3 or more BTH events. Concerning BTH severity, the only significant associations were male gender (OR = 4.39, p = 0.0003) and pregnancies (OR = 8.4, p = 0.007).
FIGURE 2.
Predictors of breakthrough hemolysis by Receiver Operating Characteristics (ROC) analysis. Variables with a statistically significant association (p < 0.05) with BTH occurrence included female gender (OR = 2.75, p < 0.001), increased LDH levels at diagnosis (OR = 1.14, p = 0.005), a shorter time from diagnosis to the first CI (OR = 1.003, p = 0.001), the administration of proximal inhibitors (OR = 4.85 compared to eculizumab, p = 0.0004), and poorer responses to CI (linear trend with OR = 4.62 and p = 0.0005 for worse responses compared to complete and major responses). Conversely, PNH/MDS was less commonly associated with BTH events compared to classic hemolytic and PNH/AA (OR = 0.26, p = 0.004). These variables resulted in an AUC = 0.82, sensitivity = 0.79, specificity = 0.67.
Finally, looking at the thresholds of laboratory values associated with clinical BTH events, an Hb drop of 2.85 g/dL and an LDH increase of 0.79 × ULN from pre‐BTH resulted in optimal cut‐offs, although with low Youden's indexes (0.3 and 0.2, respectively).
4. Discussion
This is the first real‐world analysis of BTH frequency among a large cohort of PNH patients treated with different complement inhibitors, outside the setting of clinical trials. BTH occurred with all CIs and in approximately 50% of patients, with a calculated incidence of 0.19 events per patient‐year.
Despite the varying duration of exposure, the exposure‐adjusted incidence was maximal with pegcetacoplan (0.4 events per patient‐year), lower for eculizumab and other anti‐C5 (0.18 per patient‐year), and intermediate for AP inhibitors (0.3 events per patient‐year). This is in line with the data of registration trials reporting an incidence of 0.1–0.2 per patient‐year with eculizumab [7, 8] and < 0.1 per patient‐year with ravulizumab [9], and of 0.33 per patient‐year with pegcetacoplan [10, 11].
Regarding clinical severity, the median Hb drop was 1.6 g/dL overall, and approximately 40% had a decrease of more than 2 g/dL, while the remaining patients had only minor fluctuations in Hb levels. Severity was almost comparable among eculizumab‐ and ravulizumab‐treated patients with the exception of a more profound drop of Hb with the former (which could be easily related to PK BTH with standard dose eculizumab in 20%); though transfusions were required in 40% of subjects in both groups. This compares favorably with reported data of eculizumab and ravulizumab, were Hb drops ≥ 2 g/dL occurred in ~20%–40% of BTH episodes [12, 13]. Among patients treated with pegcetacoplan and AP inhibitors, the majority experienced Hb drops ≥ 2 g/dL, which was accompanied by an LDH increase > 3 × ULN in most cases. However, the transfusion need reached 86% with the former and 31% with the latter, likely in keeping with higher baseline Hb levels in patients treated with AP inhibitors. This higher drop of Hb in patients treated with proximal inhibitors has been related to the higher PNH red cell clone, protected from hemolysis, observed with these treatments [14]. Pegcetacoplan and AP inhibitors were those with the highest PNH red blood cell clone as well as with a higher rate of complete response pre‐BTH in our series. These data are in line with the PEGASUS trial, where factors such as high baseline LDH levels and frequent transfusions were identified as predictors for BTH risk [10, 15].
Notably, clinically significant BTH was associated with an Hb delta ≥ 2.8 g/dL or with an increase of LDH of 0.79 × ULN from baseline (i.e., Hb decrease from 13 to 10.2 and LDH increase from 1 to 1.79 × ULN). These cutoffs may aid in assessing disease severity at presentation and in harnessing the “aggressiveness” of monitoring and treatment of BTH, including anticoagulant prophylaxis.
Consistently, by logistic regression analysis, the main predictors of BTH occurrence were markers of a more severe disease at diagnosis (increased LDH levels, classic hemolytic form, and a shorter time from diagnosis to first complement inhibitor initiation), as well as treatment with proximal inhibitors and poorer EBMT response category. These data are in line with pegcetacoplan and iptacopan trials, where factors such as high baseline LDH levels and frequent transfusions were identified as predictors for BTH risk [16, 17].
Notably, half of the patients had multiple episodes, with 18% of patients experiencing three or more, being associated with higher PNH clone size at diagnosis, as described for eculizumab‐treated patients [7, 8]; but also with treatment with proximal inhibitors, again possibly due to the above‐mentioned intrinsic risk of these compounds, as well as to the selection of a more severe series of suboptimal responders to complement inhibitors.
When comparing outcomes following each BTH event with the initial response achieved with the ongoing complement inhibitor, a deterioration of response was experienced by 29% of subjects, highlighting the potential impact of BTH on overall therapeutic outcomes. Six patients switched to an alternative complement inhibitor within 1 month from the BTH event. The frequency of BTH and worsening of response has been reported as a detrimental factor triggering a switch of treatment in other studies; although this might not resolve this issue [18].
Considering the triggers for BTH events, infections were the most common (54%), reinforcing the importance of mitigation strategies for infectious risk in this patient population [19]. Defective adherence to treatment accounted for 10% of episodes, particularly with oral agents, highlighting the critical need for patient education on compliance. The latter was > 95% in clinical trials of iptacopan and danicopan [20, 21, 22] but may be more challenging in the real world and in the long term. Other less frequent triggers included acute inflammatory states, vaccinations, and pregnancy, as already reported [1, 5, 23, 24], raising patient and clinicians' awareness under such circumstances. Notably, a cause could not be found for 27% of cases, further pinpointing the unpredictability of BTH.
An important point is the practical management of BTH episodes since there is no standardization. Transfusions remained the primary treatment in half of patients [23]. Anticoagulant prophylaxis was the second most frequent intervention (22% of cases) and likely minimized the risk of breakthrough thrombosis, occurring in only 5 cases (2%) in this series, all during infections and on anti‐C5. Notably, the occurrence of thrombosis in anti‐C5 treated patients only is likely due to the longer follow up as compared to proximal inhibitors. In fact, a recent real‐world analysis from the French group reported a case of breakthrough thrombosis among pegcetacoplan‐treated patients, suggesting caution even with proximal inhibitors [25]. An adjustment of the dosing regimen of complement inhibitors was required in 17% of patients in our series, either by anticipating an additional dose of anti‐C5 or by increasing the dose of pegcetacoplan. In the case of eculizumab, ravulizumab, and crovalimab, it has been suggested to anticipate dosing by 1 or 2 weeks depending on BTH severity, timing of the previous dose of CI, and drug half‐life [5]. For pegcetacoplan, it has been proposed to increase the dose or frequency of pegcetacoplan (i.e., 3 consecutive subcutaneous doses or intravenous infusion and then switch to pegcetacoplan thrice a week) [18], with addition of anti‐C5 if hemolysis worsens in the subsequent 48–72 h [11].
This study carries several limitations, particularly the retrospective design and the limited number of patients treated with proximal inhibitors with a shorter follow up. However, the inclusion of all recently approved complement inhibitors, a large series, and the availability of granular data at the time of BTH for the majority of patients strengthen the reliability of our results. Finally, this study adds to the real‐world evidence reports on PNH therapy, improving the knowledge on the efficacy and safety of complement inhibitors in this patient population.
In conclusion, in this real‐world study BTH events occurred in 50% of PNH patients, with higher incidence and severity in those treated with proximal inhibitors, particularly with pegcetacoplan. BTH events were predicted by a more severe disease at presentation and by a suboptimal response to the prior complement inhibitor. Infections remain the most reported trigger for BTH, though the importance of treatment adherence with oral drugs can not be overemphasized. The identification of clinically significant BTH in patients with an Hb delta ≥ 2.8 g/dL or increase of LDH by 0.79 xULN from baseline is crucial to harness treatment and monitoring, and to prompt anticoagulant prophylaxis to minimize the risk of breakthrough thrombosis.
Author Contributions
B.F., G.L.P., E.M., S.S., W.B., and A.K. designed the study. B.F. and G.L.P. analyzed data. All authors followed patients and contributed to data collection. All Authors revised the paper and approved the final version.
Ethics Statement
The study was conducted as a substudy of the CYTOPAN protocol approved by the local ethical committee. The study was conducted according to the Helsinki Declaration and patients gave informed consent.
Conflicts of Interest
The authors declare no conflicts of interest related to the present publication. B.F., Sobi: Speakers Bureau; Samsung: Speakers Bureau; Novartis: Consultancy; Roche: Consultancy, Other: travel to congress; Alexion: Consultancy; Janssen: Consultancy; Agios: Research Funding; Zenas BioPharma: Research Funding. W.B., Alexion, AstraZeneca Rare Disease: Consultancy, Membership on an entity's Board of Directors or advisory committees, Research Funding; Sobi: Consultancy; Sanofi: Consultancy, Honoraria, Speakers Bureau; Novartis: Consultancy, Honoraria, Speakers Bureau.
Supporting information
Figure S1: Disposition of complement inhibitors treatments in the study population.
Data S1: ajh70032‐sup‐0002‐Tables.docx.
Acknowledgment
Open access funding provided by BIBLIOSAN.
Fattizzo B., Pedone G. L., Metafuni E., et al., “Characterization of Breakthrough Hemolysis in Patients With Paroxysmal Nocturnal Hemoglobinuria: An International Multicenter Experience,” American Journal of Hematology 100, no. 11 (2025): 1963–1971, 10.1002/ajh.70032.
Funding: This work was supported by Ministero della Salute, Current Research Grant.
Data Availability Statement
All data were included in the article manuscript and Supporting Information. Additional information may be obtained upon reasonable request to the Corresponding Author.
References
- 1. Brodsky R. A., “How I Treat Paroxysmal Nocturnal Hemoglobinuria,” Blood 137, no. 10 (2021): 1304–1309. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Kulasekararaj A. G., Brodsky R. A., Nishimura J. I., Patriquin C. J., and Schrezenmeier H., “The Importance of Terminal Complement Inhibition in Paroxysmal Nocturnal Hemoglobinuria,” Therapeutic Advances in Hematology 13 (2022): 20406207221091046. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Lee J. W., Brodsky R. A., Nishimura J. I., and Kulasekararaj A. G., “The Role of the Alternative Pathway in Paroxysmal Nocturnal Hemoglobinuria and Emerging Treatments,” Expert Review of Clinical Pharmacology 15, no. 7 (2022): 851–861. [DOI] [PubMed] [Google Scholar]
- 4. Kulasekararaj A., Brodsky R., Schrezenmeier H., et al., “Ravulizumab Demonstrates Long‐Term Efficacy, Safety and Favorable Patient Survival in Patients With Paroxysmal Nocturnal Hemoglobinuria,” Annals of Hematology 104, no. 1 (2025): 81–94. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Fattizzo B., Versino F., and Barcellini W., “Breakthrough Hemolysis in Paroxysmal Nocturnal Hemoglobinuria Throughout Clinical Trials: From Definition to Clinical Practice,” Blood 146, no. 4 (2025): 411–421, 10.1182/blood.2024027574. [DOI] [PubMed] [Google Scholar]
- 6. Risitano A. M., Marotta S., Ricci P., et al., “Anti‐Complement Treatment for Paroxysmal Nocturnal Hemoglobinuria: Time for Proximal Complement Inhibition? A Position Paper From the SAAWP of the EBMT,” Frontiers in Immunology 10 (2019): 1157. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Schaap C. C. M., Heubel‐Moenen F. C. J. I., Nur E., et al., “Nationwide Study of Eculizumab in Paroxysmal Nocturnal Hemoglobinuria: Evaluation of Treatment Indications and Outcomes,” European Journal of Haematology 110, no. 6 (2023): 648–658. [DOI] [PubMed] [Google Scholar]
- 8. Versmold K., Alashkar F., Raiser C., et al., “Long‐Term Outcomes of Patients With Paroxysmal Nocturnal Hemoglobinuria Treated With Eculizumab in a Real‐World Setting,” European Journal of Haematology 111, no. 1 (2023): 84–95. [DOI] [PubMed] [Google Scholar]
- 9. Brodsky R. A., Peffault de Latour R., Rottinghaus S. T., et al., “Characterization of Breakthrough Hemolysis Events Observed in the Phase 3 Randomized Studies of Ravulizumab Versus Eculizumab in Adults With Paroxysmal Nocturnal Hemoglobinuria,” Haematologica 106, no. 1 (2021): 230–237. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Patriquin C. J., Bogdanovic A., Griffin M., et al., “Safety and Efficacy of Pegcetacoplan in Adult Patients With Paroxysmal Nocturnal Hemoglobinuria Over 48 Weeks: 307 Open‐Label ex Tension Study,” Advanced Therapeutics 41, no. 5 (2024): 2050–2069. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Griffin M., Kelly R., Brindel I., et al., “Real‐World Experience of Pegcetacoplan in Paroxysmal Nocturnal Hemoglobinuria,” American Journal of Hematology 99, no. 5 (2024): 816–823, 10.1002/ajh.27242. [DOI] [PubMed] [Google Scholar]
- 12. Kulasekararaj A. G., Hill A., Rottinghaus S. T., et al., “Ravulizumab (ALXN1210) vs Eculizumab in C5‐Inhibitor‐Experienced Adult Patients With PNH: The 302 Study,” Blood 133, no. 6 (2019): 540–549. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Lee J. W., Sicre de Fontbrune F., Wong Lee Lee L., et al., “Ravulizumab (ALXN1210) vs Eculizumab in Adult Patients With PNH Naive to Complement Inhibitors: The 301 Study,” Blood 133, no. 6 (2019): 530–539. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Notaro R. and Luzzatto L., “Breakthrough Hemolysis in PNH With Proximal or Terminal Complement Inhibition,” New England Journal of Medicine 387, no. 2 (2022): 160–166. [DOI] [PubMed] [Google Scholar]
- 15. Hillmen P., Szer J., Weitz I., et al., “Pegcetacoplan Versus Eculizumab in Paroxysmal Nocturnal Hemoglobinuria,” New England Journal of Medicine 384, no. 11 (2021): 1028–1037, 10.1056/NEJMoa2029073. [DOI] [PubMed] [Google Scholar]
- 16. Peffault DeLatour R., De Castro C., Mulherin B., et al., “Characterization of Clinically Significant Breakthrough Hemolysis in Patients With Paroxysmal Nocturnal Hemoglobinuria Treated With Pegcetacoplan,” EHA Library 2024 (2024): P819. [Google Scholar]
- 17. Peffault De Latour R., Kulasekararaj A. G., Scheinberg P., et al., “Clinical Breakthrough Hemolysis (BTH) During Monotherapy With the Oral Factor B Inhibitor Iptacopan Is Generally Not Severe and Managed Without Treatment Discontinuation: 48‐Week Data From the Phase III Apply‐PNH and Appoint‐PNH Trials in Paroxysmal Nocturnal Hemoglobinuria (PNH),” Blood 142 (2023): 1338. [Google Scholar]
- 18. Griffin M., Kelly R. J., Panse J., et al., “Management of Acute Breakthrough Hemolysis With Intensive Pegcetacoplan Dosing in Patients With PNH,” Blood Advances 8, no. 7 (2024): 1776–1786. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Girmenia C., Barcellini W., Bianchi P., et al., “Management of Infection in PNH Patients Treated With Eculizumab or Other Complement Inhibitors: Unmet Clinical Needs,” Blood Reviews 58 (2023): 101013. [DOI] [PubMed] [Google Scholar]
- 20. Peffault de Latour R., Röth A., Kulasekararaj A. G., et al., “Oral Iptacopan Monotherapy in Paroxysmal Nocturnal Hemoglobinuria,” New England Journal of Medicine 390, no. 11 (2024): 994–1008. [DOI] [PubMed] [Google Scholar]
- 21. Lee J. W., Griffin M., Kim J. S., et al., “Addition of Danicopan to Ravulizumab or Eculizumab in Patients With Paroxysmal Nocturnal Haemoglobinuria and Clinically Significant Extravascular Haemolysis (ALPHA): A Double‐Blind, Randomised, Phase 3 Trial,” Lancet Haematology 10, no. 12 (2023): e955–e965. [DOI] [PubMed] [Google Scholar]
- 22. Kulasekararaj A., Griffin M., Piatek C., et al., “Long‐Term Efficacy and Safety of Danicopan as Add‐On Therapy to Ravulizumab or Eculizumab in PNH With Significant EVH,” Blood 145, no. 8 (2025): 811–822. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Dingli D., De CastroIii C., Koprivnikar J., et al., “Expert Consensus on the Management of Pharmacodynamic Breakthrough‐Hemolysis in Treated Paroxysmal Nocturnal Hemoglobinuria,” Hematology 29, no. 1 (2024): 2329030. [DOI] [PubMed] [Google Scholar]
- 24. Portuguese A. J., Sunga C., Kruse‐Jarres R., Gernsheimer T., and Abkowitz J., “Autoimmune‐ and Complement‐Mediated Hematologic Condition Recrudescence Following SARS‐CoV‐2 Vaccination,” Blood Advances 5, no. 13 (2021): 2794–2798. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Sicre de Fontbrune F., Forcade E., Terriou L., et al., “Efficacy and Safety of Pegcetacoplan in PNH Patients With Extra‐Vascular Hemolysis Under Anti‐C5: The Real‐Life Experience of the National French Reference Center,” Blood 144 (2024): 1321. [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Figure S1: Disposition of complement inhibitors treatments in the study population.
Data S1: ajh70032‐sup‐0002‐Tables.docx.
Data Availability Statement
All data were included in the article manuscript and Supporting Information. Additional information may be obtained upon reasonable request to the Corresponding Author.