Real-world breakthrough hemolysis patterns across 1,723 patient-years of complement inhibition in paroxysmal nocturnal hemoglobinuria

Abstract Despite the introduction of specific treatment, some patients with paroxysmal nocturnal hemoglobinuria (PNH) will still experience various degrees of hemolysis. Specifically, breakthrough hemolysis (BTH) is the most common cause of treatment failure in PNH, yet contemporary incidence data, especially concerning proximal complement inhibitors (CI), are limited. Herein, we reviewed 1,723 patient-years (PY) on CI therapy from our center from 2004 (with the first FDA approval of a CI) to 2025 across 208 patients to understand features of BTH in a real word PNH patient cohort. BTH was defined as an acute hemoglobin fall ≥ 1.5 g/dL or LDH ≥ 1.5× upper limit of normal (ULN). “Continuous” BTH required LDH elevation for ≥ 2 weeks. Incidence was expressed per 100 PY. For the purpose of this study, 2 levels of granularity were applied: i) all available patients with available data (208 patients, 1723 PY); ii) granular, formal analysis with continuous longitudinal data from 117 patients with 721 PY total. Complete responders were defined of stable Hgb levels ≥10 g/dl. Across 721 PY we documented 153 BTH episodes with 16 continuous events. Exposure was 445 PY to eculizumab, 169 PY to ravulizumab, 46 PY to pegcetacoplan, and 62 PY to iptacopan, yielding BTH rates of 22.3, 24.8, 30.4, and 6.5 events/100 PY, respectively. Iptacopan had a 59% lower BTH rate than the three alternate CI comparators combined (rate ratio 0.41, p=0.046). Notably, there was a ubiquitous decrease in BTH in complete responders to CI (defined as Hgb ≥ 10 g/dL) compared to non-responders (8.1 vs. 26.7 events/100 PY; p=.04). Functionally, all 138 BTH episodes on terminal-CI were intravascular (IVH). Among the 15 proximal-CI events, 2 presented mixed extravascular features; neither patient carried complement-gene mutations on WES. There were 98/153 “clinical” BTH episodes with symptoms, transfusions, or medication changes needed. Clinical BTH episodes were more common in non-responders than responders in eculizumab (OR 2.32; p = .003), ravulizumab (OR 2.32; p = .a044), iptacopan (OR 10; p = 0.18) and pegcetacoplan (OR 2.5; p = 0.21). Six were “massive” (≥ 3 pRBC units): two in responders and four in non-responders. Among 103 therapy switches, ravulizumab-to-iptacopan (25 patients, 19.6 PY) reduced BTH from 34.7 to 19.6 events/100 PY (43%, p=0.39). Pegcetacoplan-to-iptacopan (11 patients, 4.6 PY) dropped rates from 21.6 to 0 events/100 PY (p=0.16). Eculizumab-to-iptacopan (5 patients, 2.6 PY) rose from 61.7 to 79.4 events/100 PY, driven by one early event (p = 0.72). Among terminal-to-terminal moves, eculizumab-to-ravulizumab (34 patients) increased rates from 13.7 to 28.5 events/100 PY (p = 0.12), whereas ravulizumab-to-pegcetacoplan (13 patients) decreased them from 50.8 to 28.8 events/100 PY (p = 0.21). Pooled by mechanism, proximal inhibition (pegcetacoplan + iptacopan; 107 PY) showed 20.4 events/100 PY versus 23.0 events/100 PY with terminal C5 blockade (eculizumab + ravulizumab; 614 PY, p=0.48). Overall, terminal-to-proximal transitions, especially to iptacopan, consistently reduced BTH burden.This two-decade, real-world analysis shows that proximal factor B blockade with iptacopan provided the lowest BTH incidence, outperforming terminal C5 inhibitors and C3 blockade. Importantly, in BTH and clinical BTH, non-responders were more prone for episodes and with no significant difference in massive BTH episodes. This suggests that maintained Hgb does not predispose to more BTH episodes or massive BTH episodes. Additionally, although switch cohorts were limited, terminal-to-proximal transitions consistently lowered BTH rates. These findings support prospective multicenter validation and that iptacopan could maintain long-term disease stability. Current complement sequencing analysis is ongoing to classify patients with hemolysis.