As predicted in a visionary review that we published in this journal in 2018, in the last half-decade the field of paroxysmal nocturnal hemoglobinuria (PNH) treatment has been advanced by the approval of novel therapeutic agents targeting different proteins of the complement cascade 1. Indeed, after the prototypical and well-established treatment with anti-C5 monoclonal antibodies (initially eculizumab 2, then followed by ravulizumab 3 and crovalimab) 4, more recently inhibitors of complement component 3 (C3) 5, complement factor D 6, 7 and complement factor B 8, 9 came to the fore resulting in a substantial improvement of clinical outcomes. Complement-mediated hemolysis of PNH erythrocytes is typically intravascular, and it results from the uncontrolled assembly of the lytic membrane attack complex C5b-9 (the very terminal effector mechanism of the complement cascade) due to the lack of the glycosylphosphatidylinositol-linked proteins CD55 and CD59 from the surface of PNH cells 10. While this justifies the therapeutic efficacy of terminal complement inhibitors targeting C5, uncontrolled complement regulation on PNH erythrocytes is quite broader, since CD55 (and possibly CD59) are essential for the regulation of the C3 convertase. C3 convertase is constitutionally enabled by the spontaneous C3 tick-over, and C3 itself may work as an effector mechanism serving with its surface breakdown fragments as opsonin which leads to C3-mediated extravascular hemolysis 11. As a consequence, proximal complement inhibitors (this is the terminology we coined for the novel inhibitors listed above) 12 may better counteract the global impairment of complement regulation in PNH, possibly resulting in better therapeutic efficacy. Despite our deeper biological understanding of target therapies in PNH, data on complement biomarkers remain very limited. Indeed, the only available data are restricted to anti-C5 therapies with CH50 13 (a functional assay measuring residual serum complement activity) and with free-C5/free-eculizumab (pharmacokinetic measurements for anti-C5 antibodies) 13-15. We investigated different serum complement biomarkers in 10 PNH patients followed in our Institution, within a biological study approved by our IRB. After having collected signed informed consent, peripheral blood samples were collected either before or serially during treatment with the investigational proximal complement inhibitors. Two patients enrolled in the phase 2 danicopan trial NCT03053102 7 were tested before any treatment and on danicopan monotherapy; eight patients enrolled in the phase 2 iptacopan trial NCT03439839 9 were tested at baseline while on eculizumab, and then during treatment with iptacopan (initially associated with eculizumab and then in monotherapy). Instead of testing residual complement activity, we investigated actual complement activation looking in patients' sera at breakdown fragments specific for different pathways of the complement cascade, such as soluble C5b-9 (sC5b-9, terminal pathway), Bb (alternative pathway), and C4d (classical pathway). Complement biomarker data were correlated with clinical and laboratory data, and in particular with lactate dehydrogenase (LDH, as measurement of intravascular hemolysis) and C3d deposition on PNH erythrocytes (as surrogate marker of extravascular hemolysis). A total of 278 samples were analyzed; the first observation was that while sC5b-9 and Bb were elevated in a substantial proportion of samples, C4d remained uniformly within the normal range in all tested samples (including those from patients with hemolysis, as demonstrated by highly elevated LDH) (Figure 1). Taking all patients together, we found a statistically significant linear correlation between sC5b-9 and Bb; we also observed a statistically significant (although rough) correlation between Bb (but not sC5b-9) and LDH, but not with the percentage of C3d + PNH erythrocytes (Table 1). Complement breakdown fragments sC5b-9, Bb and C4d in untreated and treated PNH patients. Serum levels in PNH patients untreated or receiving specific complement inhibitors (terminal inhibitors in monotherapy; proximal inhibitors in monotherapy; * combined treatment with proximal and terminal inhibitors). Wiskler plots represent mean value ±1 standard deviation. Dotted lines represent the upper limit of the normal (ULN) for each assay. (A) Complement breakdown fragment sC5b-9. (B) Complement breakdown fragment Bb. (C) Complement breakdown fragment C4d. Complement breakdown fragments were assessed using the Quidel MicroVue sC5b-9 Plus, Bb Plus and C4d fragments EIA diagnostic kits according to manufacturer's instructions. In brief, patients' samples were immediately refrigerated on ice after drawing, and sera were recovered within 3 h before being frozen at −20°C. At time of testing, after thawing sera were incubated in the microplates coated with the specific mAb, according to instruction from the manufacturer. The immunoenzymatic reaction is then generated through a second incubation with a tracer mAb conjugated with horseradish peroxidase, eventually followed by the chromogenic enzymatic reaction with the substrate. The subsequent spectrophotometric detection at 450 nm was performed with a DSX Automated Elisa System (Dynex technologies), and detected value converted in serum level using the reader provided by the manufactures, after calibration. *Since the aim of this study was just to investigate the predictive value of the assays in a clinical setting (i.e., in terms of correlation with hemolysis), in this analysis we merged together results in patients who were receiving either danicopan (NCT03053102) or iptacopan (NCT03439839), without comparative analysis. This is also because split data for danicopan and iptacopan (sC5b-9: 0.965 ± 0.225 ng/mL and 1.852 ± 1.208 μg/mL, respectively; and Bb: 1.264 ± 0.739 and 0.547 ± 0.769 μg/mL, respectively) are not informative due to the dose-finding design of the NCT03053102 7 and NCT03439839 9 phase II trials. A more detailed description for each agent of the pharmacodynamic measurements included in these trials can be found in the original publications 7, 9. Untreated patients: In the four samples available (two patients) we found that both sC5b-9 (2.478 ± 2.151 μg/mL; n.r. 0.020–0.172) and Bb (4.578 ± 0.494 μg/mL; n.r. 0.060–0.566) were very elevated, while C4d was not (0.036 ± 0.023 μg/mL; n.r. 0.033–0.219). These data are consistent with the continuous massive complement activation typical of untreated PNH, and they also demonstrate in vivo that (at least during chronic hemolysis) the effector terminal complement eventually causing intravascular hemolysis is enabled through the alternative pathway (due to the spontaneous C3 tick-over) and not through the classical one. Terminal complement inhibition (eculizumab): patients (n = 8, n samples = 16) exhibited elevated Bb (3.378 ± 2.124 μg/mL) and sC5b-9 (1.273 ± 1.007 μg/mL) values, consistent with the concept that C5 inhibitors may reduce terminal pathway activation without affecting initial complement activation. In this cohort we could not find any correlation between the extent of hemolysis and complement biomarkers. Indeed, LDH did not correlate with either sC5b-9 (even if spikes of LDH were always associated with elevated sC5b-9, but not vice versa) or Bb (which was uniformly elevated, as anticipated by the mechanism of action), and the percentage of C3d+ erythrocytes (that were detectable in all samples from eculizumab treated patients: 32.3% ± 19.2%, n = 56; n.r. < 1%) did not correlate with Bb. Proximal complement inhibition: In patients on a proximal inhibitor monotherapy (either danicopan or iptacopan) both Bb (0.572 ± 0.443 μg/mL) and sC5b-9 (1.151 ± 0.910 μg/mL) were lower as compared with values in untreated patients but still above the normal range in the absence of overt hemolysis. These data are consistent with the pharmacologic effects on upstream complement activation (as demonstrated by C3d+ erythrocytes < 1% in all samples), but exhibited overall an inconsistent correlation with markers of hemolysis like LDH. Elevations of LDH were usually associated with increases in Bb and sC5b-9, but a statistically significant linear correlation with LDH could only be shown between Bb and LDH (please provide values and p = 0.03), but not sC5b-9. Thus, fluctuations of Bb or sC5b-9, which were relatively frequent in these patients, were not reflected in similar changes in hemolysis assessed with LDH. Notably, although substantially lower as compared with untreated patients, both Bb and especially sC5b-9 were not in the normal range, despite no evidence of massive hemolysis (no patient experienced episodes of breakthrough hemolysis) in this setting. Combined terminal and proximal inhibition: Bb and sC5b-9 values were lower compared with untreated patient values (Figure 1). In this setting of subtotal complement inhibition (most samples were collected during the dose-escalation period of the study), large fluctuations of Bb were observed (1.280 ± 0.910 μg/mL), but we could not find a linear correlation with LDH levels. By assessing complement breakdown fragments, we show here that impaired complement regulation in PNH results in in vivo activation of the alternative and of the terminal pathways, but not of the classical pathway, both chronically and at the time of possible exacerbations of hemolysis. Consistently with their mechanism of action, terminal complement inhibitors reduce sC5b-9 values while proximal inhibitors reduce both Bb and sC5b-9 values; these biomarkers remain still elevated irrespective of a remarkable clinical control of hemolysis. We quantitatively correlated sC5b-9 and Bb with the extent of residual hemolysis in vivo, as tracked by LDH; but even if we found a statistical correlation with Bb (but not sC5b-9) in some patient cohort (patients in monotherapy on proximal inhibitors), these biomarkers were useful to track subtle, transient leakages of complement inhibition, but remain quite inaccurate to track or anticipate hemolytic events. This apparent conundrum may be explained by the fact that peaks of complement biomarkers and LDH do not necessarily match due to their different in vivo half-lives (ranging within seconds/min for complement biomarkers 16 and within hours/days for LDH) 17, 18. Moreover, it has to be highlighted that even in conditions where complement biomarkers have a quite established role, such as atypical hemolytic-uremic syndrome 19, inter-laboratory results show large inconsistencies due to the low ex vivo stability and other technical issues which hamper the robustness of the assays 20. In conditions where complement inhibition is complete and sustained, serial analysis of complement breakdown fragments may be informative to track residual in vivo complement activation and to demonstrate possible leakages of therapeutic complement inhibition, but its clinical relevance seems negligible. In this setting, LDH seems more informative to track hemolysis associated with residual complement activation, and it might perform better even in monitoring residual complement activation itself, since it can be still found elevated many hours after the event (i.e., when samples are most likely collected). It is conceivable that if samples are drawn exactly at the time of acute event, complement biomarkers could be elevated, and their quick normalization might anticipate the clinical resolution of the event. In conclusion, our data suggest that in daily practice, LDH remains the most informative blood test in the monitoring of PNH patients on complement inhibitors. Complement breakdown fragments and other complement biomarkers may be helpful to track residual complement activation in vivo, which is a key pharmacodynamic feature of each complement inhibitor. Given that sustained deepness of complement inhibition eventually shapes the safety of any therapeutic complement inhibition, we encourage including complement biomarker investigation (in addition to proper pharmacokinetic assessments) in any clinical trial and clinical research on PNH, and we are eager to see detailed results from all recently approved agents. A.M,R. conceived the study and participated in the collection of samples with C.F. P.R. performed all the experiments and generated the data that were discussed and analyzed by all the authors. A.M.R. drafted the first version of the manuscript that was then developed and finalized with the critical revision of P.R. and C.F. The authors wish to thank Alexion Pharmaceuticals and Novartis for the participation of our center to the NCT03053102 and NCT03439839 trials, respectively, in which the patients described in this study were enrolled. The authors wish to thank the Italian PNH Patient Association (AIEPN), as well as all PNH patients and their families. The authors also wish to thank colleagues of the Severe Aplastic Anemia Working Party of the EBMT for the critical discussion of the data described in this manuscript. A.M.R. received lecture fees from Alexion, Novartis, Pfizer, and SOBI, and served as a member of the advisory/investigator board for Alexion, Roche, Novartis, Apellis, and Omeros. C.F. received lecture fees from SOBI, Novartis, and Alexion, and served as a member of the advisory/investigator board for SOBI, Novartis, and Roche. P.R. has no conflict to disclose. The data that support the findings of this study are available from the corresponding author upon reasonable request.
Risitano et al. (Sun,) studied this question.