Pleiotropic effects of intravascular haemolysis on vascular homeostasis


  • Gregory J. Kato,

    1. Pulmonary and Vascular Medicine Branch, National Heart, Lung and Blood Institute and the Critical Care Medicine Department, Clinical Center, National Institutes of Health, Bethesda, MD, USA
    Search for more papers by this author
  • James G. Taylor VI

    1. Pulmonary and Vascular Medicine Branch, National Heart, Lung and Blood Institute and the Critical Care Medicine Department, Clinical Center, National Institutes of Health, Bethesda, MD, USA
    Search for more papers by this author

Gregory J. Kato, MD, Sickle Cell Vascular Disease Section, Pulmonary and Vascular Medicine Branch, NHLBI, National Institutes of Health, 10 Center Drive, MSC 1476, Building 10-CRC, Room 5-5140, Bethesda, MD 20892-1476, USA.


The breakdown of senescent or defective red blood cells releases red cell contents, especially haemoglobin, which scavenges nitric oxide (NO) and decomposes to haem and free iron. These are potent oxidants, all of which have promoted the evolution of inducible and vasculoprotective compensatory pathways to rapidly clear and detoxify haemoglobin, haem and iron. Chronic haemolytic red cell disorders as diverse as sickle cell disease, thalassaemia, unstable haemoglobinopathy, cytoskeletal defects and enzymopathies have been linked to a clinical constellation of pulmonary hypertension, priapism, leg ulceration and possibly cerebrovascular disease and thrombosis. Besides free haemoglobin, haemolysis has been associated with extracellular arginase that limits substrate availability to NO synthase, endogenous inhibitors of NO synthase activity, and inappropriate activation of haemostatic pathways. This article reviews the haemolytic disorders that have been reported to manifest vascular complications, and explores the speculative possibility that haemolysis mediates some of the vascular complications of inflammation and diabetes.

During the last century, pathological haemolysis in humans was viewed primarily as a cause of anaemia. Haemolytic anaemia induces a programme of physiological compensatory mechanisms, and many of these have been well characterized years ago. Red cell production is increased through a pathway involving the sensing of tissue hypoxia due to decreased oxygen carrying capacity, inducing the prolyl hydroxylase-hypoxia inducible factor-erythropoietin pathway, which then stimulates red cell production in the bone marrow, indicated by an increased number of circulating reticulocytes (Lee, 2008). As long as the hematocrit reached a steady state level by balancing the increased red cell destruction with increased red cell production, haematologists largely considered the haemolytic anaemia compensated, with any symptoms largely attributed to the decreased oxygen carrying capacity from the degree of anaemia.

Efficient and redundant mechanisms have evolved to clear the products of haemolysis, principally haemoglobin and its oxidative components haem and iron, which are sequestered in the circulation by haptoglobin, hemopexin and transferrin, respectively (Schaer et al, 2007; Kato, 2009). These carrier proteins safely escort the relevant molecules primarily to reticuloendothelial cells that intracellularly catabolize their contents and safely store iron. For many years, the only consequence of chronic haemolysis that was widely appreciated is cholelithiasis due to gallstones composed of calcium bilirubinate, a product of haem metabolism. In recent years, the vasculoprotective importance of the red cell clearance pathways has become more evident. There is increasing evidence that the products of haemolysis are vasculotoxic, particularly when the pace of haemolysis exceeds the capacity of this clearance system (Rother et al, 2005). In addition, intravascular haemolysis releases the erythrocyte contents directly into plasma, as distinguished from extravascular haemolysis, in which reticuloendothelial macrophages phagocytose senescent red cells, with little or no release of erythrocyte contents into plasma. This article will review the vascular complications of haemolytic anemias in humans, and the current view of the mechanisms by which haemolysis impairs vascular function.

Haemolytic anaemias associated with pathological vascular complications

A number of haemolytic anaemias have been characterized clinically as associated with vascular complications. Examples of these are highlighted in Table I and discussed in more detail below.

Table I.   Examples of vascular complications described in patients with haemolytic anaemia.
Haemolytic anaemiaPHPriapismLeg ulcersThrombosis
  1. PH, pulmonary hypertension.

Sickle cell disease××××
Pyruvate kinase deficiency× × 
Hereditary spherocytosis/ovalocytosis× × 
Unstable haemoglobins×× ×
Paroxysmal nocturnal haemoglobinuria×× ×
Autoimmune haemolytic anaemia×  ×
Congenital dyserythropoietic anaemia ×× 

Sickle cell disease

The earliest description of this haemolytic anaemia and its characteristic red cell morphology was published almost exactly 100 years ago (Herrick, 1910). Over the next century, this haemoglobinopathy became best known for the intraerythrocytic polymerization of its mutant haemoglobin in the first molecularly defined human disease. The polymer burden reduces the red cell compliance, resulting in rigid erythrocytes that slow or even halt blood flow in the microvasculature and cause tissue ischemia. This poor blood flow may be compounded by adhesion of activated blood cells to activated endothelium, especially in the post-capillary venules. Excessive blood viscosity, tissue ischemia and infarction are clearly part of the vaso-occlusive pain crisis, osteonecrosis and the acute chest syndrome, an acute lung injury that most commonly results from embolization of infarcted bone marrow to lung (Gladwin & Vichinsky, 2008). Pulmonary hypertension (PH), stroke, priapism and leg ulcers were all initially assumed also to be due to sludging of blood flow in the respective organs, but more recent epidemiological data more closely implicate the degree of haemolysis, rather than the degree of blood viscosity, in their development. In fact, sickling is not required for their development, as indicated by their association with other forms of haemolytic anaemia below, although it is likely that sickling also contributes.

Cor pulmonale was long known to be one of the occasional causes of death in patients with SCD (Yater & Hansmann, 1936). As lifespan improved dramatically for patients with SCD over the last 50 years, the diagnosis of symptomatic PH became more common. It is a grave prognostic factor, conferring as low as 50% 2-year survival (Castro et al, 2003). In more recent prospective screening studies in the Unites States using the tricuspid regurgitant velocity (TRV) measured on Doppler echocardiography, one-third of adults with sickle cell disease have a TRV at least two standard deviations above the mean (≥2·5 m/s), and 9% have a TRV approximately three standard deviations above the mean (≥3 m/s), a figure that more closely approximates the degree of pulmonary artery pressure elevation on right heart catheterization procedures classically accepted as proof of PH (Ataga et al, 2004, 2006; Gladwin et al, 2004; Lee et al, 2007; Nelson et al, 2007; Voskaridou et al, 2007; Aliyu et al, 2008; De Castro et al, 2008; Hagar et al, 2008; Onyekwere et al, 2008; Pashankar et al, 2008; Liem et al, 2009; Minniti et al, 2009; Sedrak et al, 2009). Regardless of the choice of semantics, adult SCD patients with a TRV ≥2·5 m/s have an approximately 10-fold relative risk for early mortality, making elevated TRV the strongest risk factor for sickle cell mortality in a proportional hazard model. Although proof is lacking that the patients actually die of PH, sudden death is common, similar to that seen in other forms of PH (Haque et al, 2002). SCD PH patients become symptomatic and die at less elevated pulmonary pressures than non-SCD PH patients, probably explained by the very low oxygen carrying capacity in the very anaemic SCD patients, frequent co-morbid conditions, and episodes of acute chest syndrome, which are associated with very high mortality in the patients with the highest pulmonary pressures.

There is possibly no other genetic disorder that has a higher and earlier risk of stroke than SCD. Median age of first stoke is 7 years of age (Ohene-Frempong et al, 1998). Although microvascular sludging due to sickling is thought to play a role, this might be more likely linked to the more patchy microvascular strokes that are considered silent infarcts of the brain. In the more clinically dramatic strokes, the cerebral infarcts occur as dense, vascular territory lesions associated with proliferative large vessel arterial lesions highly reminiscent histologically of atherosclerosis lesions without atheromatous plaque. The intimal and medial hyperplasia are often associated with irregular, activated and adhesive endothelium with superimposed in situ thrombosis, features that were emphasized many years ago in the definitive histopathology autopsy series of ischemic stroke in SCD (Rothman et al, 1986). These proliferative vasculopathy lesions in large arteries bear many histological, pathobiological and epidemiological similarities to those of PH and to atherosclerosis (Hebbel et al, 2004; Kato et al, 2006; Kato & Gladwin, 2008). The Cooperative Study of Sickle Cell Disease (CSSCD), a large natural history study, identified low haemoglobin values as an independent risk factor for the development of ischemic stroke, the first hint that severity of haemolysis might be associated with stroke (Ohene-Frempong et al, 1998). This was corroborated by additional studies from that group and others indicating that a known genetic modifier of haemolysis (Embury et al, 1982; De Ceulaer et al, 1983), alpha-thalassaemia trait, is protective against abnormal transcranial Doppler blood flow velocity and strokes (Adams et al, 1994; Neonato et al, 2000; Hsu et al, 2003; Bernaudin et al, 2008).

Priapism and leg ulceration in patients with SCD draw less attention than many other complications of SCD. However, they contribute to recurrent and prolonged misery in these patients. They originally were both envisioned to be due to sludging or adhesion of sickled erythrocytes in their respective circulations. However, they both epidemiologically are linked to severity of haemolysis, rather than blood viscosity. Once again, the alpha-thalassaemia trait reduces the severity of haemolysis and is protective against the development of both leg ulcers (Higgs et al, 1982; Steinberg et al, 1984; Koshy et al, 1989) and priapism (Nolan et al, 2005a). They both occur in non-sickling haemolytic anaemia patients, as detailed below, demonstrating that sickling is not required for their development. Finally, the report of priapism occurring in mice genetically deficient in nitric oxide (NO) production implicates vascular dysregulation due to NO deficiency as an important cause (Champion et al, 2005).


This second prevalent and well characterized disease of haemoglobin is usually classified as a disorder of ineffective erythropoiesis, meaning that there is a deficiency of formed erythrocytes released from the bone marrow. Although this concept is useful for modelling certain aspects of its pathophysiology, it may also be useful to understand that the unbalanced globin chain synthesis that defines the disorder results in highly unstable haemoglobin homotetramers that cause lysis of the developing erythroid cells even before they have a chance to exit the bone marrow medulla. From an alternative perspective, then, thalassaemia is a form of intramedullary haemolysis. Beta thalassaemia major patients are commonly treated with scheduled transfusion every 2–4 weeks, keeping their red cell mass relatively normal and intentionally suppressing the intramedullary erythropoiesis that otherwise becomes massive and distorts bone architecture, thereby also reducing the severity of intramedullary haemolysis. However, thalassaemia intermedia patients are only moderately anaemic and usually receive only intermittent transfusions which do not suppress their erythropoiesis, making them more strongly subject to the potential complications of chronic haemolytic anaemia.

Elevated pulmonary pressure by echocardiogram is reported in up to 75% of patients with thalassaemia major and intermedia (Grisaru et al, 1990; Du et al, 1997; Aessopos et al, 2001). Pulmonary thromboses have been reported several times in the literature, reportedly leading to PH (Taher et al, 2002), although it may be difficult clinically to distinguish this diagnosis from patients with initially clinically unsuspected PH who go on to develop in situ pulmonary thrombosis, a frequent scenario in all forms of PH. Patients with thalassaemia have a higher rate of all forms of thromboembolic disease than the general population (Eldor & Rachmilewitz, 2002). Leg ulcers and priapism have also been reported (Stevens et al, 1977; Jackson et al, 1986). Patients with thalassaemia frequently undergo surgical splenectomy to reduce red cell transfusion requirements, and post-splenectomy patients have a higher frequency of all of the above complications.

Hereditary spherocytosis, ovalocytosis and elliptocytosis

These inherited disorders of the red cell cytoskeleton lead to variable degrees of severity of haemolytic anaemia. Patients with these membranopathies have been reported to develop pulmonary thromboemboli (Hayag-Barin et al, 1998), PH, and leg ulcers (Vanscheidt et al, 1990). It has been suggested that these complications are more frequent among patients that have undergone splenectomy (Schilling et al, 2008). As therapeutic splenectomy to reduce red cell destruction is more common in the patients with the most severe haemolytic anaemia, history of splenectomy may simply be a surrogate marker of severity of underlying disease. However, haemolytic anaemia and loss of splenic function are a recurrent combination in the vascular complications being reviewed here.

Pyruvate kinase deficiency

This is one of the most severe congenital haemolytic anaemias, sometimes even treated with chronic scheduled transfusion therapy. Therapeutic splenectomy is almost universally employed. Patients with chronic severe haemolytic anaemia due to pyruvate kinase deficiency have been reported to develop PH (Chou & DeLoughery, 2001) and leg ulceration (Muller-Soyano et al, 1976).

Unstable haemoglobin disorders

Although SCD and thalassaemia could be, in a manner of speaking, considered unstable haemoglobin disorders, this label is usually reserved for other mutations of the globin subunits that produce no other readily observed feature than haemolytic anaemia. PH has been reported in patients with several unstable haemoglobinopathies (Honig et al, 1989; Krishnan et al, 1994; Lode et al, 2007). The same is true for priapism (Thuret et al, 1996; Gyan et al, 2001; Andrieu et al, 2003). Thromboembolic disease has been seen in patients with unstable haemoglobin (Pavlovic et al, 2004; Kim et al, 2005).

Paroxysmal nocturnal haemoglobinuria (PNH)

A chronic haemolytic anaemia due to an acquired somatic mutation in the hematopoietic stem cell, PNH manifests very severe intravascular haemolysis due to a loss of protection from complement-mediated membrane permeabilization of the red cell. Patients have variable degrees of chronic intravascular haemolysis and can develop acute exacerbations of this haemolysis. Well-known for catastrophic thromboembolic complications, this disease has recently been shown to have a high prevalence of elevated pulmonary pressures (Heller et al, 1992; Hill et al, in press). This is all the more remarkable because the haemolytic anaemia often develops only late in life, so that the patients do not have the same lifelong duration of haemolysis as seen in all the congenital haemolytic anaemia patients discussed above. Presumably, the severity of their intravascular haemolysis integrated with the shorter duration results in an equivalent ‘area under the curve’ of risk for haemolysis-associated complications. Priapism has also been reported in males with PNH (Montalban et al, 1986).

Autoimmune haemolytic anaemia

Autoantibodies directed to antigens on the surface of the red cell cause membrane loss and phagocytosis of red cells by reticuloendothelial system macrophages, classically considered a disorder of extravascular haemolysis. However, complement activation can result in a component of intravascular haemolysis. PH has been described in an animal model of immune haemolytic anaemia (Hsu et al, 2007), and in a case report in a patient (Zhang et al, 2007).

Congenital dyserythropoietic anaemia

This group of very rare and poorly understood hereditary anaemias involves defects in red cell production and in many cases, haemolysis. Patients with this diagnosis have been described with leg ulceration (Bordi et al, 2002) and priapism (Edney et al, 2002).


One of the most potent forces of evolutionary pressure in Africa and southern Asia, Plasmodium falciparum malaria produces the most explosive intravascular haemolysis known in clinical haematology. Although the malaria research literature has focused largely on the postulated toxicity of very high levels of inducible NO synthase activity during acute malaria, more recent findings suggest a haemolysis-linked defect in vascular reactivity in humans with malaria (Yeo et al, 2007). Like SCD and thalassaemia, acute haemolysis in malaria can release red cell arginase into plasma and deplete levels of the NO synthase substrate arginine (Lopansri et al, 2003). Furthermore, NO supplementation ameliorates the fatality of cerebral malaria in a mouse model (Gramaglia et al, 2006), and NO synthase polymorphisms that increase NO production appear to be protective in humans (Hobbs et al, 2002). In contrast to the vasculopathic consequences of chronic haemolysis discussed above, these malaria data suggest that acute, very severe haemolysis can have more immediate consequences.

Cerebral vasospasm following subarachnoid haemorrhage

Ischemic stroke is a frequently fatal event about 7 d after a subarachnoid haemorrhage (Pluta et al, 2009). This is approximately the time at which erythrocytes in the subarachnoid clot undergo lysis, which appears to provoke local NO scavenging and oxidative stress that contributes to regional vasospasm severe enough to cause cerebral infarction. An NO donor, such as sodium nitrite, can prevent vasospasm in an animal model of subarachnoid haemorrhage (Pluta et al, 2005). This is an example of localized haemolysis with localized vascular consequences.

Pathophysiology of haemolysis-associated vasculopathy

Nitric oxide scavenging

Intravascular haemolysis unleashes a multi-pronged attack on vascular homeostasis, best characterized for reduced bioavailability of NO (Rother et al, 2005). NO, produced by endothelial, blood and other cells, orchestrates a complex and coordinated programme of vascular regulation, inhibiting vasoconstriction, platelet activation and aggregation, release of procoagulant proteins, inflammatory mediators and proliferative factors (Fig 1A). While encapsulated within the red cell membrane, haemoglobin is sequestered away from endothelium produced NO, protected by biophysical diffusion barriers at the membrane. However, disorders of intravascular haemolysis disrupt these diffusion barriers, releasing haemoglobin into plasma, where it reacts rapidly with NO to oxidize it to nitrate (Figs 1B and 2) (Reiter et al, 2002). Although this is a stoichiometric and not catalytic reaction, the intravascular component of haemolysis in an adult with SCD can release over 3 g of haemoglobin into plasma per day (Bensinger & Gillette, 1974), equivalent to over 125 mg/h, capable of neutralising more than 7 mmol of NO per hour. This is often surpassed in PNH (Hill et al, in press) and severe haemolytic malaria.

Figure 1.

 Nitric oxide-regulated pathways in vascular homeostasis. (A) NO represses an entire programme of pathways involved in inducing hemostasis and a proliferative response. (B) In a state of decreased NO bioavailability, as induced by extracellular haemoglobin during haemolysis, this whole programme is derepressed, favouring hemostasis: vasoconstriction, endothelial adhesiveness, platelet activation and aggregation, clotting, and vessel wall cellular proliferation.VCAM-1, vascular cellular adhesion mlecule-1.

Figure 2.

 Pleiotropic effects of intravascular haemolysis. Haemolysis in the intravascular compartment releases red cell contents into the plasma, including haemoglobin (Hb), which scavenges NO; arginase, which depletes plasma arginine, the substrate for NO synthase (NOS); and lactate dehydrogenase (LDH), which acts as a surrogate marker of Hb and arginase. Linked to haemolysis is a rising plasma level of asymmetric dimethylarginine (ADMA), an endogenous inhibitor of NOS. Also found in SCD are low levels of apoA-I and endothelial progenitor cells (EPC), which have been linked to endothelial dysfunction and decreased NO production. The aggregate state of decreased NO bioavailability has been linked to platelet activation, vascular dysfunction and pulmonary hypertension, as well as leg ulcers, priapism and preliminarily also cerebrovascular disease and stroke.

Ectopic arginase activity in plasma

The consumption of NO is compounded in haemolysis by the simultaneous decompartmentalization of erythrocyte arginase-1 into plasma. This ectopic arginase activity converts arginine to ornithine, reducing plasma arginine, the obligate substrate for NO synthases. Thus, the one event of intravascular haemolysis simultaneously results in accelerated destruction of NO and limits the expected compensatory increase in NO production (Fig 2). In adults with SCD, plasma arginase activity is high and is associated with PH and early mortality (Morris et al, 2005a). Inadequate availability of arginine is believed to contribute to NO synthase uncoupling, which causes NO synthase to produce reactive oxygen species instead of NO. High plasma arginase levels have also been reported in patients with thalassaemia (Morris et al, 2005b). It is likely that this pathobiological mechanism also pertains to other disorders of intravascular haemolysis.

NO synthase inhibitors

High levels of endogenous NO synthase inhibitors have been detected in plasma from patients with SCD, particularly asymmetric dimethylarginine (ADMA)(Schnog et al, 2005; Landburg et al, 2008; Kato et al, 2009). ADMA has been implicated in accelerated development of atherosclerosis and early mortality in patients without haemolytic disease (Bode-Boger et al, 2007). ADMA is known to inhibit the activity of NO synthase and contribute to its uncoupling, and it may serve as yet another factor to limit NO production in patients with SCD (Fig 2), whose plasma ADMA levels are three times that of healthy control subjects. The levels are slightly but significantly higher in SCD patients with elevated pulmonary pressures, suggesting that NO synthase inhibition may contribute to PH in SCD (Kato et al, 2009). ADMA elevation in SCD is linked with the severity of haemolysis, making it a third haemolysis-associated factor contributing to vasculopathy (Schnog et al, 2005; Kato et al, 2009). This pathobiological link may be indirect, since ADMA clearance mechanisms appear to be repressed by erythropoietin, which rises to high levels to provide compensatory erythropoiesis during severe haemolysis (Scalera et al, 2005). However, the potential contribution of ADMA liberated during turnover of red cell proteins during haemolysis has not been investigated.

Oxidative stress

The products of haemolysis are intensely oxidative. As described above, haemoglobin can scavenge NO, an important endogenous antioxidant. In addition, the haem and iron moieties potently induce oxidative stress, against which entire protective pathways have evolved. This pathway includes haptoglobin, which binds to free haemoglobin, preventing its extravasation, and carries it to reticuloendothelial macrophages for endocytosis and clearance via the CD163 receptor (Schaer et al, 2007; Kato, 2009). Haemopexin serves to bind free haem, escorting it to the CD91 receptor on reticuloendothelial cells for clearance. Free iron is bound up by serum transferrin or stored in tissue ferritin, which protect tissues from its highly oxidative potential. Oxidative stress is believed to be a significant contributor to vasculopathies of many types. In SCD, this appears to be potentiated by high expression of enzymes that produce reactive oxygen species, including xanthine oxidase and NADPH oxidase (Aslan et al, 2001; Wood et al, 2005). Depletion of red cell glutathione, its major antioxidant, is also linked to haemolytic rate and to PH in SCD (Morris et al, 2008).

Dyslipidaemia and endothelial dysfunction

Plasma levels of apolipoprotein A-I (apoA-I) levels have been consistently reported to be low in patients with SCD, appear to trend even lower in those SCD patients who have severe haemolysis or elevated pulmonary pressure (Yuditskaya et al, 2009). ApoA-I is the principal protein of high density lipoprotein cholesterol (HDL-C) particles, the ‘good cholesterol’ known in the general population to be associated with good endothelial function and protective against atherosclerosis. Although apoA-I is known to facilitate reverse cholesterol transport, it is also known to promote endothelial production of NO, apparently involving its binding to the scavenger receptor binding protein SRB-I (Fig 2) (Yuhanna et al, 2001). SCD patients with low apoA-I exhibit blunted vascular reactivity to acetylcholine, a hallmark of endothelial dysfunction (Yuditskaya et al, 2009). Serum triglyceride elevation has been observed in parallel with haemolytic severity in SCD, for unclear reasons (Morris et al, 2005a). Hypertriglyceridemia is strongly believed to be a risk factor for cardiovascular disease in the general population, and it is possible that it also contributes in haemolysis-associated vasculopathy.

Endothelial activation

One of the features of activated endothelium includes expression of cell surface adhesion molecules, such as vascular cell adhesion molecule (VCAM-1), intercellular adhesion molecule (ICAM-1), E-selectin and P-selectin. Soluble components of these adhesion molecules are found in plasma in SCD (Duits et al, 1996; Schnog et al, 2003; Setty et al, 2003; Kato et al, 2005), thalassaemia (Aggeli et al, 2005) and haemolytic-uremic syndrome, and in SCD their concentration is correlated to PH (Kato et al, 2005). Circulating endothelial cells in an activated state have been found in SCD (Solovey et al, 1998; Strijbos et al, 2009) and thalassaemia (Butthep et al, 2002) and may mediate effects of vascular inflammation in vasculopathy.

Activated, adhesive endothelium is more likely to interact with abnormal erythrocytes (Kaul et al, 2006; Brittain & Parise, 2008; Cartron & Elion, 2008; Zennadi et al, 2008), activated leucocytes (Frenette, 2004; Kaul et al, 2004; Haynes et al, 2006), and the haemostatic pathway, including activated platelets (Wun et al, 1998; Villagra et al, 2007), to promote thrombosis and vasculopathy (Verhamme & Hoylaerts, 2006; Singer & Ataga, 2008). Thrombosis is commonly observed in several haemolytic disorders (Table I) (Cappellini, 2007), including paroxysmal nocturnal haemoglobinuria (Hill et al, in press), thrombotic thrombocytopenic purpura (TTP) (Sadler, 2008), hereditary spherocytosis (Schilling et al, 2008), thalassaemia (Taher et al, 2008) and sickle cell disease (Singer & Ataga, 2008), with recent preliminary evidence in the latter indicating a haemolysis-associated defect of von Willebrand factor protease activity, suggesting some mechanistic overlap with TTP (Zhou et al, 2009). These conditions all suggest general mechanisms of disordered interaction of endothelium with the haemostatic system in haemolytic anaemias.

Platelet activation

A precursor to platelet aggregation, adhesion and thrombosis, platelet activation can be induced by activation of the coagulation pathway, and is normally repressed by NO (Schafer et al, 2004; Jin et al, 2005). Platelet activation has been known to be prominent in patients with SCD (Westwick et al, 1983; Wun et al, 1998). Cell free haemoglobin can activate platelets ex vivo, and SCD patients with more severe haemolysis and PH have a statistically significant increase in platelet activation (Villagra et al, 2007), both suggesting that the haemolysis-associated defect in NO bioavailability contributes to platelet activation in SCD (Fig 2). Consistent with this model, sildenafil, a drug that amplifies NO signalling, decreases platelet activation in vivo in SCD patients with PH (Villagra et al, 2007). It is conceivable that this platelet activation might play a role in the progression of PH, by releasing factors from platelet granules that may contribute to endothelial activation and dysfunction. Hypothetically, activated platelets might also contribute to in situ pulmonary thrombosis, a known complication of any form of PH. Abnormal platelet activation has been observed also in thalassaemia (Ruf et al, 1997; Singer et al, 2006).

von Willebrand factor (VWF) protease

Dysregulation of this protease has been reported in SCD (Schnog et al, 2006). Specifically, extracellular haemoglobin released during haemolysis is believed to inhibit its protease activity (Zhou et al, 2009). Decreased activity of VWF protease would be predicted to generate high molecular weight multimers of VWF, which contribute to the vascular complications of thrombotic thrombocytopenic purpura. It will be intriguing to see what future study of this mechanism yields in SCD and other haemolytic disorders.

Functional asplenia

Anecdotally, surgical splenectomy in several haemolytic disorders has been associated with an increased prevalence of thromboembolic disease and PH (Crary & Buchanan, 2009). This may be an example of confounding by indication, because typically only the most severely affected patients undergo surgical splenectomy, and splenectomy may therefore be only a marker of more severe disease. However, Westerman et al (2008) found that thalassaemia patients who have undergone splenectomy have higher plasma levels of free haemoglobin and microparticle fragments of cells that might contribute to PH and thromboembolism. Although this is a retrospective and non-randomized study, it is consistent with a model in which the spleen provides a large capacity for extravascular haemolysis, which is far less prone to contribute plasma haemoglobin. Splenectomy removes the largest reservoir of reticuloendothelial macrophages, and this may result in extended red cell survival, but perhaps more likelihood that the eventual demise of the red cell will be via intravascular haemolysis, rather than extravascular. Haemolysis in the intravascular compartment results in a greater elevation of plasma haemoglobin and arginase. Splenectomy also markedly increases the circulating platelet count, and it is unclear if this may mediate any relationship to vasculopathy or thrombosis.

Phosphatidylserine exposure

As red cells age and senesce, they lose the normal asymmetric distribution of phosphatidylserine and other phospholipids, and this is more prominent in SCD (Setty et al, 2002) and thalassaemia (Ruf et al, 1997), and proportional to reticulocyte count. In part, this phosphatidylserine exposure mediates binding and phagocytosis by macrophages. However, it also seems capable of inducing binding to endothelial cells with unknown consequences (Haynes et al, 2006; Setty & Betal, 2008). Microparticles displaying phosphatidylserine derived from erythrocytes (Allan et al, 1982), platelets (Wun et al, 1998) and endothelial cells (Shet et al, 2003) have been observed in sickle cell disease and thalassaemia (Westerman et al, 2008). Such microparticles are believed to contribute to the development of vasculopathy, including reduced NO-dependent vasodilation, and increased arterial stiffness, inflammation and thrombosis (Chironi et al, 2009). The extent of phosphatidylserine exposure in other haemolytic anaemias has not been well characterized. The potential contribution of phosphatidylserine exposure on the development of vasculopathy in haemolytic diseases remains to be determined.

Genetic modifiers of haemolysis

This is an area that has not been well explored. It is clear that in SCD, haemolysis is reduced by co-inherited genetic traits that slow the kinetics of haemoglobin S polymerization, including alpha-thalassaemia trait and hereditary persistence of fetal haemoglobin. There is a suggestion that mild deficiency of glucose-6-phosphate dehydrogenase in SCD may promote cerebral vasculopathy, but an associated defect in vascular resistance to oxidative stress was implicated, rather than accelerated haemolysis (Bernaudin et al, 2008). If the model elaborated above is substantially correct, it is likely that genetic traits in combination that accelerate haemolysis or interfere with detoxifying its oxidative and NO-scavenging effects will similarly accelerate the development of vasculopathy. Evidence supporting this model has been published in SCD with priapism or leg ulceration prevalence associated with the Klotho gene, which regulates NO production (Nolan et al, 2005b, 2006), or in malaria with the NO synthase gene polymorphisms discussed earlier (Hobbs et al, 2002).

In summary, the medical literature clearly supports a link between SCD and vascular complications including PH, leg ulceration, priapism and stroke. However, the occurrence of these complications in non-sickling haemolytic anaemias suggests that common factors related to haemolysis play a significant role, although sickling may augment them. Haemolysis is associated with defects in a wide variety of mechanisms that compromise vascular homeostasis, particularly with regard to NO bioavailability, oxidative stress and haemostatic activation. Although it is commonly assumed that a compensated chronic haemolytic anaemia has little consequence, long-term exposure to its effects appears to be associated with an increased incidence of vasculopathic events.

Does subclinical haemolysis contribute to vascular disease?

Inflammation, diabetes and oxidative stress are three recurring themes in the development of vascular disease in the general population. Intriguingly, inflammation and diabetes each independently have been associated with evidence for low-grade haemolysis that might speculatively contribute to chronic oxidative stress and NO deficiency, and ultimately predispose to vasculopathy, including atherosclerosis.

The anaemia of inflammation, often also called the anaemia of chronic disease, is associated with inflammatory mediators common to acute and chronic infections and to autoimmune disorders. Although the anaemia is well known to involve decreased erythropoiesis, less well appreciated is the contribution of shortened red cell survival, essentially a subtle form of haemolysis (Cartwright, 1966). Interestingly, it now appears that part of the glucocorticoid anti-inflammatory programme includes induction of haptoglobin, the principal protein that binds to haemoglobin released from lysed red cells (Boretti et al, 2009). This pathway suggests that haemolysis evolutionarily has been a significant component of the inflammatory process that has provided a strong selective pressure (Kato, 2009). Chronic inflammation is now believed to contribute to the development of atherosclerosis in the general population, and it is provocative to consider whether inflammation-induced haemolysis and resulting impairment to NO bioavailability is part of this predisposition.

There is also a similar combination of biochemical and teleological evidence that diabetes, a major risk factor for the development of vascular dysfunction and atherosclerosis, also involves a low-grade chronic haemolysis. CD59 is a cell surface protein that functions to protect the cell against complement-mediated lysis, and its dysfunction on erythrocytes causes chronic haemolytic anaemia. In diabetes, hyperglycemia can lead to the glycation of CD59, producing a small, but significant acceleration of physiological haemolysis (Acosta et al, 2000). In addition, glycation of haemoglobin itself decreases its binding by haptoglobin, and subsequent clearance (Asleh et al, 2003). Supporting the speculative idea that this low-grade haemolysis might play a role in vascular disease, a variant allele of haptoglobin that inefficiently clears extracellular haemoglobin increases the risk of atherosclerosis in diabetics (Levy et al, 2002; Asleh et al, 2005).


Long considered a relatively benign process, chronic haemolysis may provide a risk factor for the development of chronic NO deficiency, oxidative stress and vasculopathy. This pathobiology is becoming well characterized in lifelong haemolytic disorders, such as sickle cell disease and thalassaemia, and appears to also play a role in acquired haemolytic disorders, such as paroxysmal haemolytic anaemia, autoimmune haemolytic anaemia and malaria. This pathway is also likely to be active in less chronic haemolytic anaemias, such as thrombotic thrombocytopenic purpura, haemolytic-uremic syndrome, haemolytic transfusion reactions. It is intriguing to speculate that some of these mechanisms may also come into play in the chronic low grade haemolysis of inflammation and hyperglycemia. The contribution of haemolysis of extravasated blood cells in subarachnoid haemorrhage to vasospasm suggests activation of the haemolysis-induced effects in other examples of haematoma formation. It also suggests the potential existence of a beneficial effect of haemolysed blood in physiological haemostasis that is inappropriately activated systemically during haemolytic disorders.