Background
Paroxysmal nocturnal hemoglobinuria (PNH) is an acquired disorder of hematopoiesis characterized by a somatic mutation in the
PIGA gene that prevents or impairs the synthesis of glycosylphosphatidylinositol (GPI) anchors [
1,
2]. The deficiency on red blood cells (RBCs) of GPI-anchored proteins [
3,
4], including the complement regulators CD55 [
5,
6] and CD59 [
7], results in chronic intravascular hemolysis with recurrent exacerbations, anemia, smooth muscle cell dystonia, and high risk of thrombosis [
4,
8‐
10].
The blockade of terminal complement pathway by eculizumab [
11], a monoclonal antibody (moAb) against complement component 5 (C5), abrogates intravascular hemolysis with the consequent normalization of lactate dehydrogenase (LDH) levels in almost all patients suffering from PNH. This treatment has proven to be safe and clinically effective in hemolytic PNH patients [
12‐
14], except those in which bone marrow failure is the major cause of anemia [
15,
16].
The persistence (or the recurrence) of intravascular hemolysis is observed only in few conditions: (i) Japanese patients carrying a rare polymorphism of C5 [
17], (ii) patients with an increased eculizumab turnover requiring extra-dosage (pharmacokinetic breakthrough) [
18], and (iii) patients who occasionally experience transient episodes of intravascular hemolysis because of massive complement activation during infections or inflammatory disorders (pharmacodynamic breakthrough) [
18‐
21]. Despite these small and infrequent limits, the treatment with eculizumab has radically changed the natural history of PNH since in most patients it reduces anemia [
12,
13] and thrombosis [
22], and improves quality of life and survival [
14].
However, the abrogation of intravascular hemolysis is not the only relevant change in PNH pathophysiology associated with eculizumab treatment. In fact, at variance with PNH patients not treated with eculizumab, a population of GPI-negative (PNH) RBCs bound with fragment of complement component 3 (C3) appears in almost all patients on eculizumab [
23] and, in some patients, also the less-sensitive direct antiglobulin test may turn positive [
24]. The PNH RBCs bound with C3 become apparent because PNH RBCs, spared from hemolysis by the blockade of the terminal complement cascade, remain unable to control the early steps of the ongoing complement activation. Eventually, these PNH RBCs, once opsonized with complement, become potential targets of phagocytosis by macrophages, with consequent variable degrees of extravascular hemolysis [
23]. Accordingly, in PNH patients on eculizumab, the extent of C3 binding correlates with reticulocyte count, the in vivo half-life of
51Cr-labeled RBCs is reduced and there is an excess of spleen and liver
51Cr uptake [
23]. This extravascular hemolysis is clinically relevant because it can limit the efficacy of eculizumab to the point that some patients may remain transfusion-dependent [
14,
22‐
25]. The heterogeneity of mechanisms controlling C3 binding and/or removal of C3+ PNH RBCs are likely to account for the variable extent of C3 binding and of the consequent extravascular hemolysis. Part of this variability may result from genetic diversity in genes coding for endogenous regulators of complement: indeed, we have recently shown that polymorphisms of complement receptor 1 (
CR1) gene are associated with the transfusion need of patients on eculizumab [
26].
The most intriguing biological feature of C3 binding is that two distinct populations of PNH RBCs are always present in patients on eculizumab, one with (C3+) and one without (C3−) C3 binding, despite the uniform deficiency of the GPI-anchored proteins on all PNH RBCs [
23]. The understanding of this phenomenon is of more than academic importance because, depending on the answer, one might explore different ways to overcome the consequent clinical problems. In this paper, by reproducing in vitro C3 binding in the presence of C5 blockade, we provide evidence suggesting a stochastic model for the emergence of these two distinct, C3+ and C3−, populations of PNH RBC.
Discussion
The binding of C3 fragments to RBCs, characteristic of various autoimmune hemolytic anemias, had never been found in PNH and it has emerged as a novel phenomenon in PNH patients on eculizumab [
23,
24,
51]. It is possible to visualize that in the absence of eculizumab no C3+ RBCs are seen because, once C3 is activated on the surface of PNH RBCs, they will promptly undergo hemolysis. In contrast, when C5 is blocked by eculizumab, PNH RBCs with bound C3 survive because they are no longer lysed by the terminal complement pathway; however, there will be now a “new” steady state in which part of these C3-opsonized RBCs are removed by macrophages, likely via interaction with the complement receptor 3 [
52] producing the extravascular hemolysis [
23].
This “new” steady state does not detract from the clinical benefit arising from the abrogation of intravascular hemolysis. However, this observation accounts for the fact that only a minority of PNH patients on eculizumab normalizes their hemoglobin level, and in some patients, residual anemia may even require red blood cell transfusions [
14,
15,
20,
23,
24]. For these patients there are not yet standard treatments: steroids have been proven to be ineffective [
51]; splenectomy [
53] or splenic artery embolization [
54,
55] may be effective but raise concerns about infection and thrombosis susceptibility [
15].
We have effectively reproduced in vitro the phenomenon of C3 binding: in fact, in vitro complement activation in the context of C5 blockade generates a distinct population of C3+ PNH RBCs which co-exists with C3− PNH RBCs. It is intriguing that spontaneous complement activation in vitro, a condition that may mimic the chronic low level of complement activation present in vivo, results in the same order of C3 binding we have previously observed in vivo in PNH patients (
n = 41) on eculizumab: 29 vs. 27% [
23].
In addition, this mild spontaneous complement activation is almost unable to lyse PNH RBCs (less than 10% after 5 days), whereas serum acidification results in a non-negligible hemolysis of PNH RBCs (~42% after 24 h) that is not prevented by an excess of eculizumab. This is because serum acidification forces a level of complement activation in excess of what happens in vivo in the steady state [
37,
56] and that might be similar to the massive complement activation present during infection or inflammation. This may explain the hemolytic crisis observed in vivo in PNH patients on eculizumab in these specific clinical circumstances, providing evidence for a “pharmacodynamic breakthrough” [
57]. Our observations [
31,
58] are in keeping with the recent finding that strong complement activation overrides C5 inhibition by eculizumab possibly due to the generation of high density C3 products on the RBC surface [
59].
In any event, the most singular feature of C3 binding in PNH patients on eculizumab is that, despite all PNH RBCs are uniformly lacking GPI-anchored proteins, there are always two distinct PNH RBC populations with or without C3 binding. This is different from other conditions, such as the paradigmatic cold agglutinin disease [
60], in which C3 binding is present on all RBCs. Thus, C3 binding in PNH patients on eculizumab emerges as a unique phenomenon with surprising features.
In principle, these two distinct populations might arise from (a) their different ability to bind C3 or (b) from a stochastic effect. However, when C5 is blocked by eculizumab, there was the prompt appearance of a discrete population of C3+ PNH RBCs whose size increased not only with time but also with the level of complement activation. In fact, in vitro the proportion of C3+ PNH RBCs after mild spontaneous complement activation (Fig.
2b) was smaller than after more intense complement activation by acidification (Fig.
3b). Moreover, in vitro, at variance with in vivo observations, eventually all PNH RBCs become C3+. This holds true also for C3− PNH RBCs from patients on eculizumab that, in vitro, become all C3+, regardless of the size of C3+ PNH RBCs population in vivo (Fig.
4b). Finally, young RBCs are not selectively protected from C3 binding, in fact in vitro a very similar proportion of young (reticulocytes) and mature (non-reticulocytes) PNH RBCs become C3+ (Fig.
5b, c).
Altogether, these results indicate that the C3+ and the C3− PNH RBCs are not intrinsically different: any PNH RBCs, when exposed to complement activation in the context of C5 blockade, may bind C3 to the same extent. Thus, C3 binding is not a prerogative of a discrete subset of PNH RBCs. In addition, the prompt appearance, within 5 min from serum acidification (Fig.
3a), of two distinct populations of PNH RBCs (with and without C3) suggests that C3 binding is an
all-or-nothing phenomenon in which a detectable level of C3 fragments stably bound to PNH RBCs is generated only when complement activation exceed a minimal threshold.
These findings support a stochastic model in which the longer each individual RBC circulates, the higher the probability to be exposed in specific districts of the bloodstream to levels of complement activation that exceeds the threshold able to trigger C3 binding: for this reason in patients on eculizumab in vivo, the percentage of C3+ cells is much lower in reticulocytes (10.5%) than in mature RBCs (27.7%).
On the other hand, this stochastic model may also suggest that complement activation, rather than be always systemic and generalized, may also occur within localized spaces and within a limited time frame. Indeed, different organs might harbor specific mechanisms of activation and regulation of complement. This finely regulated homeostasis could explain why in most complement-mediated diseases the clinical presentations and complications are often organ specific (e.g., abdominal or central nervous vein thrombosis in PNH, renal or ocular involvement in genetically determined hemolytic-uremic syndrome and age-related macular degeneration) [
61].
Acknowledgements
We thank Alessandra Fanelli and Roberto Caporale for helpful technical advises. We thank Luca Boni for statistical analysis advices. We thank Lucio Luzzatto for much support, for the helpful suggestions, and the fruitful discussions. Last but not least, we thank the patients, their families, and the “Associazione Italiana Emoglobinuria Parossistica Notturna” for their warm support.