Massive hemolysis in paroxysmal nocturnal hemoglobinuria after switching from proximal complement inhibitor to anti-C5 therapy: A lesson not to be forgotten

The treatment of paroxysmal nocturnal hemoglobinuria (PNH) is undergoing a second revolution with the introduction of proximal complement inhibitors.¹ Indeed, recent clinical studies have shown that molecules targeting either the C3, the complement factor B, or the complement factor D (this latter in association with anti-C5 therapy) have been effective in increasing hemoglobin levels in PNH patients with residual anemia despite standard anti-C5 treatment.^2-4 The effectiveness of this class of complement inhibitors is achieved by preventing C3-mediated extravascular hemolysis, which we have identified as the major factor limiting hemoglobin normalization in PNH patients treated with anti-C5.^{1, 5} Indeed, unlike C5 inhibitors, proximal complement inhibitors are able to almost completely normalize the lifespan of PNH erythrocytes, leading to the paradox that a better hematological response is associated with a significant and large increase in both the proportion and the mass of the population of ‘crippled’ PNH erythrocytes. Eventually, the increased size of the population of PNH erythrocytes resulting from the therapeutic effectiveness of proximal complement inhibitors has raised the concern that clinically severe intravascular breakthrough hemolysis (BTH) may occur if the therapeutic blockade is lost, even temporarily. An interesting debate on the risk of BTH is currently ongoing, spurring discussions about whether proximal complement inhibitors should be used as monotherapy or in a combination with anti-C5.⁶ These discussions are based on comparisons of different clinical trials but suffer from the paucity of data. In addition, there is an almost complete lack of knowledge about the clinical course of BTH in patients with very large proportions of PNH erythrocytes (even higher than 90%) because, simply, this condition has never been observed in PNH patients before the introduction of treatments with proximal complement inhibitors. Thus, even index cases provide valuable insights that enhance the understanding of this novel condition within the PNH community.

We report the unique circumstance of three PNH patients treated with the factor D inhibitor vemircopan as monotherapy within the clinical trial NCT04170023.⁷ The trial was unexpectedly terminated by the sponsor's decision, following an interim analysis claiming that vemircopan monotherapy “Did not appropriately control IVH; significantly increased rates of BTH and LDH excursions (LDH excursions is defined by LDH values >x2ULN) compared with ravulizumab” (https://clinicaltrials.gov/study/NCT04170023). Consequently, these patients, despite achieving a very good hematological response associated with a large population of PNH erythrocytes, were compelled to discontinue this effective treatment, and to be switched to standard-of-care anti-C5 treatment (single 2700 mg loading dose, followed 2 weeks later by 3300 mg maintenance, then given every 8 weeks, all intravenously). To minimize the risk of massive intravascular hemolysis, the switch to the standard-of-care ravulizumab was planned 2 weeks before vemircopan discontinuation. Thus, vemircopan discontinuation only occurred after completing the ravulizumab loading phase, guaranteeing that ravulizumab was in its therapeutic range at time of vemircopan discontinuation.

Since the lack of an established strategy to mitigate possible extravascular hemolysis (compassionate add-on danicopan was not yet available, and commercially available pegcetacoplan was labeled only as monotherapy after failure of anti-C5 treatment), we addressed this risk by planning in advance the use of the factor B inhibitor, iptacopan, through a named-patient compassionate use study program. This study program, associated with a strict monitoring of hemolysis biomarkers, was approved by the local IRBs. The patients (once vaccine boosters were performed, if needed) could initiate iptacopan treatment at Day 42 from vemircopan discontinuation. Nevertheless, after a vemircopan washout period of at least 14 days, earlier treatment was possible in cases of severe anemia or breakthrough hemolysis. Breakthrough hemolysis qualifying for early rescue treatment was defined as meeting one or more of the following criteria: (i) Hemoglobin decrease ≥3 g/dL from baseline; (ii) requirement for red blood cell transfusion; (iii) severe signs and/or symptoms of hemolysis.

At baseline (last day of vemircopan treatment), hemoglobin level was remarkable in all three patients, the lowest value was 11 g/dL in patient A because of iron deficiency secondary to hypermenorrhea. All patients had normal or slightly increased LDH, consistent with adequate control of intravascular hemolysis (Figure 1 and Table 1). None of the patients showed any laboratory sign of extravascular hemolysis, with normal reticulocyte count and bilirubin, and negligible (<0.5% of PNH RBC) C3d-opsonized erythrocytes; their proportion of PNH erythrocytes was 75%, 87%, and 44% (paralleling the proportion of PNH granulocytes in all patients, Figure 1 and Table 1), highlighting the large PNH erythrocyte mass eventually at risk of complement-mediated hemolysis. This excellent clinical picture observed at baseline drastically changed once vemircopan was discontinued, despite the anti-C5 treatment with ravulizumab started 2 weeks earlier. Indeed, within 1–2 weeks, all patients developed a significant hemoglobin drop, which was very severe and symptomatic in two patients. Patient A's hemoglobin fell to around 7 g/dL within 2 weeks, with very severe fatigue and dyspnea; unfortunately, she could not be transfused because of her religious beliefs. Patient C did even worse as he experienced a complement-amplifying condition (flu-like viral infection with fever) on the same day of vemircopan discontinuation, leading to immediate symptomatic paroxysm of intravascular hemolysis (demonstrated by gross hemoglobinuria and increased LDH) and severe anemia. Indeed, the patient required red blood cell transfusions on Days 7 and 13, despite the intravascular hemolytic crisis resolved within 72 h. Notably, in these two patients, the dramatic fall in hemoglobin was not associated with massive increase of LDH, which even in patient C at time of the intravascular hemolytic crisis did not exceed two times the upper limit of normal. Interestingly, both patients quickly developed large proportions of C3d-opsonized erythrocytes (40% and 62%, respectively) and increased bilirubin levels (Table 1), suggesting that extravascular hemolysis extensively contributed to hyperacute anemia in these patients. For this reason, already at Day 14, these two patients were switched to compassionate iptacopan in monotherapy, which provided immediate clinical benefit. Indeed, signs and symptoms of hemolysis (both intravascular and extravascular) disappeared within a few hours, with hemoglobin reaching baseline values within a few weeks and C3 opsonization becoming negligible within 4–6 weeks. Patient B had a different clinical behavior; despite an early hemoglobin drop of 2 g/dL, he did not develop clinically meaningful anemia (and thus he started iptacopan later at Day 56). This was likely due to the fact that his PNH hematopoiesis was as low as 60%, with a relevant non-PNH hematopoiesis (the proportion of normal granulocytes was around 40%), which eventually prevented anemia despite a degree of hemolysis similar to that of the other patients. Indeed, we tried to estimate the extent of hemolysis by examining the absolute number of PNH erythrocyte undergoing hemolysis at time of vemircopan discontinuation, together with the loss of Hb carried by them. According to this analysis, all three patients suffered from an acute loss of 1.5–2.5 × 10⁶/μL of PNH erythrocytes, which obviously accounts for the drop in hemoglobin (Figure 1). Comparing the behaviors of the three patients, it appears that switching from vemircopan to ravulizumab led to massive, acute, extravascular hemolysis in patients A and C, while patient B experienced a low-grade residual intravascular hemolysis. One might speculate that inherited variants of complement genes could contribute to these heterogenous biological and clinical findings. This is in keeping with the highest C3 binding and hemoglobin drop seen in patient C who was heterozygous for the rare allele of the HindIII polymorphism of the complement receptor 1 (CR1) gene that has been associated with the risk of extravascular hemolysis in PNH on anti-C5 treatment.⁸

These severe hyperacute hemolytic events were not a surprise; indeed, as investigators of the NCT04170023 vemircopan monotherapy trial, we have expressed our safety concern regarding the risk of severe hemolysis (not necessarily restricted to intravascular, but possibly also extravascular) in the case of vemircopan discontinuation, asking for an appropriate risk-mitigation strategy. These concerns are not different from the broader issues emerging in the context of PNH treatment with proximal complement inhibitors when there is a significant increase in the mass of PNH erythrocytes at risk of hemolysis. Indeed, there are no guidelines for managing such risk, because of the limited knowledge of the complications that might arise when such a large mass of erythrocytes undergoes (hyper)acute hemolysis. Indirect information regarding this risk can be inferred from the PEGASUS trial. In the PEGASUS trial all patients initially underwent a 4-week ramp-up period during which they received concomitant pegcetacoplan and anti-C5 treatment. Subsequently, those patients randomized to anti-C5 monotherapy experienced an average hemoglobin level drop of nearly 4 g/dL over the subsequent 4 weeks.²

Here, we show for the first time, that rebound hemolysis following the discontinuation of effective treatment with proximal complement inhibitors is a major clinical risk that cannot be fully prevented by switching to anti-C5 therapy. Indeed, the large mass of PNH erythrocytes is susceptible not only to acute intravascular hemolysis due to incomplete blockade by anti-C5 during complement-amplifying condition, but also to massive C3-mediated extravascular hemolysis, which may develop acutely as well. In this case, terminal complement inhibitor treatment and supportive care with transfusions remain far from optimal. Therefore, whenever possible, rescue treatment with alternative proximal complement inhibitors should be strenuously pursued.

These findings provide informative guidance for treating physicians in any situations that result in a transient breach of complete complement inhibition, taking also into account that, unlike anti-C5, the currently available proximal complement inhibitors have short half-lives^{3, 4} and may also have suboptimal pharmacokinetics.^{9, 10} Similar management problems can occur in clinical conditions that prevent oral intake or drug absorption, such as gastrointestinal comorbidities or conscience impairments. Moreover, these findings highlight the importance of considering the potential risk associated with treatment discontinuation—including lack of availability, planning or incidental occurrence of pregnancy—before initiating therapy with proximal complement inhibitors.

While not the primary focus, this report highlights the intricacies of modern clinical research, where industrial considerations can sometimes hinder the development of potentially beneficial treatments or compromise patient safety. It also emphasizes the responsibilities of academic investigators in providing strong guidance in the design of clinical trials, avoiding that lack of expertise or industrial interests could compromise science and patient safety.

A.M.R. has received consulting fees and honoraria from Alexion Pharmaceuticals, AstraZeneca Rare Disease, Apellis, Novartis, Omeros, Roche, and Samsung; research funding from Alexion Pharmaceuticals, AstraZeneca Rare Disease, Novartis, and Roche; and served as consultant for Amyndas. R.N. has received lecture fees from Alexion Pharmaceuticals-AstraZeneca Rare Disease, served as member of Investigator Board for Alexion Pharmaceuticals, AstraZeneca Rare Disease, Novartis and SOBI Pharmaceuticals.

All the three patients, once they discontinued the NCT04170023 vemircopan monotherapy study, entered the compassionate program after providing written informed consent.