Management of thrombosis in paroxysmal nocturnal hemoglobinuria: a clinician’s guide

Abstract

Paroxysmal nocturnal haemoglobinuria (PNH), an ultra-orphan disease with a prevalence of 15.9 per million in Europe, is a life-threatening disorder, characterized by haemolysis, bone marrow failure and thrombosis. Patients with PNH prior to the availability of eculizumab had a median survival of between 10 and 22 years, with thrombosis accounting for 22–67% of deaths. 29–44% of patients had at least one thrombosis. This paper provides a clinician’s guide to the diagnosis, management and complications of PNH, with an emphasis on thrombosis.

Introduction

Paroxysmal nocturnal haemoglobinuria (PNH), an ultra-orphan disease with a prevalence of 15.9 per million in Europe, is a life-threatening disorder, characterized by haemolysis, bone marrow failure and thrombosis [Hoffbrand et al. 2011; Kelly et al. 2009]. Patients with PNH prior to the availability of eculizumab had a median survival of between 10 and 22 years [Hillmen et al. 1995; de Latour et al. 2008]. Thrombosis risk has been known about for over 30 years, with a significant associated mortality risk [Rosse, 1982]. Prior to the availability of the monoclonal antibody eculizumab, thrombosis accounted for 22–67% of deaths in the PNH case series, with 29–44% of patients with PNH having had at least one thrombosis [Hillmen et al. 1995; Ray et al. 2000; Moyo et al. 2004; Ziakas et al. 2007].

Clinical PNH arises from expansion of the PNH stem cell in the bone marrow often following an immunological ‘insult’, such as preceding aplastic anaemia, although this may be transient and silent [Young et al. 2002]. Somatic mutations in the phosphatidylinositol glycan A gene in bone marrow stem cells result in loss of all glycosylphosphatidylinositol (GPI) anchor proteins on hematopoietic cells [Bessler et al. 1994; Miyata et al. 1994]. CD55 and CD59, the complement regulatory proteins on red blood cells are deficient, rendering the red blood cell susceptible to complement mediated intravascular haemolysis, free haemoglobin release and nitric oxide depletion. Nitric oxide depletion inhibits smooth muscle relaxation, causing PNH symptoms such as abdominal pain, oesophageal spasm, erectile dysfunction and pulmonary hypertension [Hillmen et al. 1995; Rother et al. 2005; Hill et al. 2010]. The complement and coagulation systems are closely interlinked, rendering PNH patients with a significantly increased risk of thrombosis, only partially alleviated by anticoagulation [Hall et al. 2003].

Incidence of thrombosis in paroxysmal nocturnal haemoglobinuria patients

Up to 10% of patients with PNH will present with thrombosis, however throughout the disease course the incidence is significantly higher [de Latour et al. 2008]. Cumulative thrombosis incidence over an 8–10-year period is between 23% and 30% (in the pre-eculizumab era) [Socié et al. 1996; Ray et al. 2000; Hall et al. 2003; de Latour et al. 2008]. Twenty percent of patients have multisite thrombosis, increasing the morbidity risk and complicating patient management [Ziakas et al. 2007].

All patients with PNH are at an increased risk of thrombosis, however the granulocyte clone size correlates with thrombosis risk. Where to delineate a ‘cut off’ value of risk is arbitrary; however those with a clone of over 50% have a cumulative 10-year incidence of thrombosis of 34.5% compared with 5.3% in those with a clone of <50% [Hall et al. 2003]. Similarly, when using a granulocyte clone size of over 61%, 54% of patients develop thrombosis compared with no thrombosis in patients with a clone size <61% [Moyo et al. 2004]. Patients with smaller clones remain at an increased risk of thrombosis, with international registry data demonstrating 7.7% of patients with a clone of 10–49% experiencing thrombotic events; thus clinical education and vigilance is essential [Schrezenmeier et al. 2014].

The incidence of thrombosis has reduced significantly with the introduction of eculizumab. Pooled analysis from the three original clinical trials showed a significant reduction in the thrombotic events from 7.37 to 1.07 per 100 patient-years once patients commenced on eculizumab. This was sustained, at long-term follow up, with an ongoing relative reduction of thrombotic events of 81.8% [Hillmen et al. 2007, 2013; Loschi et al. 2016].

Sites of thrombosis

Thrombosis in PNH can occur at any site; although more commonly venous (80–85%), it can be arterial (15–20%) [Hillmen et al. 2007]. Retrospective trials highlight the most common site of thrombosis as hepatic, accounting for approximately 40% of thrombotic events, with an associated high morbidity and mortality [Hillmen et al. 1995; Socié et al. 1996; Ray et al. 2000; Hall et al. 2003; Ziakas et al. 2007; de Latour et al. 2008; Kelly et al. 2011]. Prior to eculizumab, recurrence risk was high despite anticoagulation and patients often required further procedures such as a transjugular intrahepatic portosystemic shunt [Hoekstra et al. 2009]. PNH was considered a contraindication for a liver transplant, with Budd–Chiari recurrence rates high post orthotic transplantation [Hallf et al. 1990; Bahr et al. 2003]. However, in the eculizumab era, patients with Budd–Chiari syndrome have successfully undergone liver transplantation supported by long-term eculizumab treatment [Singer et al. 2009].

Cerebral thrombosis risk is high, presenting with neurological symptoms such as headache, vomiting, seizures and altered conscious level. Any patient with known PNH presenting with neurological symptoms should have an urgent imaging, such as a magnetic resonance venogram. A French case series of 15 patients highlighted a hormonal contribution to the majority of these events with 6 of 15 having taken the oral contraceptive or being pregnant/postpartum, while one had lupus. Patients have an increased mortality compared to those without PNH (7% versus 2%) [Meppiel et al. 2015]. Patients with PNH should avoid oestrogen based oral contraception, opting for progesterone only or barrier methods instead.

In pooled analysis from the three pivotal PNH trials, the most common thrombosis site is deep vein, accounting for 33% of events, followed by mesenteric events at 18% and hepatic events at 16%. This may be due to the prospective nature of the trials, compared to the majority of previous studies which were retrospective [Hillmen et al. 2007].

Irrespective of the most common site of thrombosis, physicians need increased vigilance as to the risks of thrombosis in PNH, particularly as patients will present to a variety of different medical specialties. Physician and patient education remains an essential component of the haematologists management

Paroxysmal nocturnal haemoglobinuria testing

Current international flow cytometry guidelines recommend PNH testing in patients, as shown in Table 1 [Borowitz et al. 2010]. This should be performed using high-sensitivity flow cytometry on peripheral blood, assessing the red cell clone using CD55 and CD59 markers, the granulocyte clone using CD24, CD66b and CD16 (minimum of two reagents) and the monocyte clone using CD14. Fluorescein-labelled proaerolysin (FLAER), a bacterial lysin-based technique which binds specifically to GPI anchor proteins, is considered by current guidelines and the International PNH Interest Group to be an accurate, sensitive method for detecting granulocyte and monocyte PNH clones, although it is not available in all laboratories [Borowitz et al. 2010; Sutherland et al. 2012]. It is essential to assess the white cell clone, as well as the red cell clone, as patients with haemolytic PNH will often have a low red cell clone due to active haemolysis [Borowitz et al. 2010] and packed red cell transfusion.

Table 1.

Indications for PNH testing.

Clinical indications for PNH testing	Additional features for testing
Intravascular haemolysis	Haemoglobinuria Elevated plasma haemoglobin
Unexplained haemolysis with	Iron deficiency Abdominal pain Oesophageal spasm Cytopenia
Acquired Coombs’-negative haemolytic anaemia
Thrombosis with unusual features	Unusual sites: hepatic, portal, splenic, splanchnic, cerebral, dermal Accompanied with haemolytic anaemia Unexplained cytopenia
Bone marrow failure	Suspected or proven aplastic anaemia Myelodysplasia Cytopenia of unknown aetiology

PNH, paroxysmal nocturnal haemoglobinuria.

Samples should be analysed in an accredited laboratory which participates in an external quality assessment programme, for example UK NEQAS.

Although flow cytometry guidelines exist, a recent UK NEQAS survey raised concerns about varied practice across 105 laboratories worldwide, with marked variability in gating strategy, reagent use for cell identification, and reporting strategies. This affects clinical diagnosis, and consensus guidelines should be adhered to [Fletcher et al. 2016].

Screening all patients with idiopathic thrombosis and no additional signs, as discussed below, is costly, and has a low detection rate, and therefore should be reserved for patients who have thrombosis with atypical features [Lazo-Langner et al. 2015].

Baseline screening investigations in patients with a PNH granulocyte clone of >20% should include an assessment of haemolysis, silent thrombosis and end organ damage, as shown in Table 2.

Table 2.

Baseline investigations of all patients with haemolytic PNH or thrombotic PNH.

Screening modality	Screen	Reasons for screening
Blood tests PNH patients	FBC Renal function Liver function Haemolysis screen Brain Natriuretic Peptide (BNP) D-dimer PNH flow Cytometry	Baseline screening to assess for haemolysis, end organ damage
Ultrasound	Echocardiography Abdominal ultrasound	Right pulmonary artery pressures to assess for pulmonary hypertension Portal blood flow to assess for silent thrombosis
CT	CTPA CT Abdomen	If concerns about silent pulmonary emboli In suspected cases of Budd–Chiari syndrome
MRI	Cardiac Venogram (head)	If concerns about right cardiac failure and pulmonary hypertension If concerns about neurological symptoms and CNS thrombosis

BNP, brain natriuretic peptide; CNS, central nervous system; CT, computerized tomography; CTPA, computerized tomography pulmonary angiography; FBC, full blood count; PNH, paroxysmal nocturnal haemoglobinuria.

This requires a high index of suspicion from the clinician, due to the rarity of PNH and the presentation to a wide variety of different specialties.

Mechanisms for thrombosis: current views

Thrombosis in PNH remains an active area of research, as current knowledge is incomplete. The complement and coagulation systems are inherently linked, accounting for some of the reasons for thrombosis. Intravascular haemolysis is a known risk factor for thrombosis, however the haemolysis independent processes may help to explain why non-haemolytic patients and those with smaller PNH clones have a thrombosis risk [Schrezenmeier et al. 2014]. PNH platelets have increased cell surface P-selectin, increasing C3 surface acquisition, driving the complement system and contributing to haemolysis-independent thrombosis [Conde et al. 2005]. Additional platelet mechanisms include complement mediated intravascular haemolysis, leading to PNH platelet activation, thrombin activation and endothelial dysfunction [Hill et al. 2013; Ortel et al. 1993]. Intravascular haemolysis releases free haemoglobin, resulting in chronic vascular endothelial activation, evidenced by circulating microparticles expressing endothelial markers [Simak et al. 2004]. Administration of free haem also causes thrombophlebitis in healthy volunteers [Simionatto et al. 1988], thus both of the above mechanisms contribute to thrombosis in PNH patients.

For a full discussion of the proposed mechanisms of thrombosis in PNH we direct you to the comprehensive paper by Hill and colleagues [Hill et al. 2013].

How to manage acute thrombosis

Eculizumab (Soliris®, Alexion) has revolutionized patient care in PNH, significantly improving life expectancy and reducing long term complications. A fully humanized immunoglobulin monoclonal antibody to C5, it is currently the only licenced treatment for PNH, binding to human complement protein C5, inhibiting terminal complement activation and membrane attack complex formation that results in intravascular haemolysis. It has been licenced in Europe since 2007, following three pivotal clinical trials which showed marked reduction in intravascular haemolysis, transfusion requirements and thrombosis incidence, with overall survival increasing to 97.9% at 66 months, similar to the general population [Hillmen et al. 2004, 2006, 2013; Brodsky et al. 2008].

It is essential that all patients on eculizumab are vaccinated against meningitis, are maintained on penicillin prophylaxis and remain vigilant for meningitis risk, due to the susceptibility to Neisseria infections, in particular Neisseria meningitides.

Management of acute thrombosis in the eculizumab naïve patient includes immediate commencement of eculizumab and full dose anticoagulation (provided no contraindications). Commencement of eculizumab reduces the risk of thrombosis extension or recurrence [Hillmen et al. 2013]. Patients may also require vascular surgery referral for thrombolysis in cases of severe life-threatening thrombosis, however this is associated with a significant risk of bleeding, and the risks and benefits should be considered. The majority of case reports are prior to the introduction of eculizumab, and this intervention is less commonly required now [McMullin et al. 1994; Kuo et al. 2006; Araten et al. 2012].

Use of direct oral anticoagulants (e.g. an oral factor Xa inhibitor such as Rivaroxiban) in acute thrombosis in the PNH patient population has not been studied, and therefore it is not recommended at present.

Continuing long term anticoagulation in patients once the acute event has passed and eculizumab established remains a difficult clinical decision. There is limited evidence available, and at present continuation of anticoagulation is advised [Kelly et al. 2011]. This decision is also dependent on the extent of thrombotic episode and other factors such as platelet count.

Patients on eculizumab should not receive plasma products except in an emergency, as plasma products contain high levels of complement, resulting in an increased risk of thrombosis and haemolysis due to loss of complement blockade. Patients treated with plasma products will require an immediate additional dose of eculizumab.

Thrombosis while on eculizumab

Thrombosis remains a risk, even in patients who are established on eculizumab, in particular at times of breakthrough haemolysis, triggered for example by infection. Pooled analysis from the three original clinical trials showed ten thrombotic events in seven patients established on eculizumab. Four patients were on concomitant anticoagulation and eculizumab. Of these ten events on eculizumab, three patients had thrombophlebitis/deep vein thrombosis (one patient had three episodes), one had deep vein thrombosis, one had retinal vein thrombosis, one had thrombosis in a fistula and one had portal and splenic vein thrombosis [Hillmen et al. 2007, 2013]. Patients require therapeutic anticoagulation and an immediate additional dose of eculizumab. Ongoing management of these patients requires a careful assessment of the events surrounding the thrombosis. If there are concerns about breakthrough haemolysis, CH50/CH100 can be useful as well to assess complement blockade, and lactate dehydrogenase as a marker of haemolysis. A dose increase of eculizumab by 300 mg is suggested. For patients already therapeutically anticoagulated with warfarin a review of international normalized ratio (INR) levels (or anti-Xa levels for patients on low molecular weight heparin (LMWH)), should be undertaken to assess efficacy and compliance. Dose adjustments or the consideration of alternative anticoagulation will be required if there are issues with attaining therapeutic levels; For example patients failing to achieve therapeutic INR results may require a switch to LMWH.

The use of plasma products increases the risk of haemolysis and thrombosis in PNH patients on eculizumab due to loss of terminal complement blockade, and thus these products should only be used in a clinical emergency, with an additional dose of eculizumab following plasma product administration (unpublished data from Leeds PNH service).

Management of acute thrombosis in countries where eculizumab is not available

In countries where eculizumab is not available, patients with thrombosis should be treated with long-term therapeutic anticoagulation. Patients are at risk of thrombotic extension and death, as anticoagulation does not completely alleviate the risk. Life-threatening events may require thrombolysis [McMullin et al. 1994; Kuo et al. 2006; Araten et al. 2012].

Bone marrow transplant for PNH remains the only curative option, however it has historically been associated with a high treatment related mortality of 40–50%, as evidenced by international bone marrow transplant registry data from 1978 to 1995 [Saso et al. 1999]. Updated 2012 data shows an improved, but still high, treatment related mortality, with an overall survival probability of 68%. PNH associated thrombosis patients have a significantly poorer outcome when compared to a matched cohort of PNH thrombotic patients who were not transplanted, and a lower overall survival probability of 54% [de latour et al. 2012]. Bone marrow transplant remains a curative option, however the patient would need to accept the high risk of treatment related mortality, and therefore in countries where eculizumab is available, bone marrow transplant for PNH alone is not advised.

Participation in clinical trials (depending on clinical trial availability), such as LFG316, a C5 complement inhibitor [ClinicalTrials.gov identifier: NCT02534909]; Coversin, a C5 complement inhibitor [ClinicalTrials.gov identifier: NCT-02591862]; or APL-2, a C3 complement inhibitor [ClinicalTrials.gov identifier: NCT02588833] may offer patients alternative options to bone marrow transplantation.

With the development of new complement inhibitors, the cost of these drugs may also reduce in the future, increasing access to a wider PNH population, and improving the prevention and management of patients with PNH related thrombosis.

Pulmonary hypertension

Pulmonary hypertension associated with intravascular haemolysis has been known about from studies in sickle cell patients [Machado et al. 2006], however it also applies to haemolytic PNH patients. The mechanism behind PNH associated pulmonary hypertension includes smooth muscle constriction due to nitric oxide consumption and increased peripheral resistance contributing to the development of pulmonary hypertension in the PNH patient, as well as chronic silent pulmonary emboli. Brain natriuretic peptide (BNP), a hormone released in response to cardiomyocyte stretch, correlates with the severity of elevated right pulmonary artery pressures and right ventricular function [Machado et al. 2006]. Patients with sickle cell disease who have a pro-BNP level of >160 pg/ml have a 78% positive predictive value for pulmonary hypertension and an increased risk of mortality [Machado et al. 2006].

This is confirmed in haemolytic PNH patients, with 3.9-fold higher BNP levels compared with the general population. Forty-one percent of haemolytic PNH patients (not on eculizumab) have subclinical pulmonary hypertension, as evidenced by echocardiography. Sixty percent of haemolytic PNH patients have subclinical pulmonary emboli and 80% have reduced right ventricle ejection fraction on cardiac magnetic resonance imaging (MRI) investigations [Hill et al. 2012]. Patients with raised BNP levels should be investigated for pulmonary hypertension initially with echocardiography and be considered for cardiac MRI.

Subclinical pulmonary hypertension is an indication to commence eculizumab in the haemolytic PNH patient, resulting in 50% reduction in BNP levels and a 67% reduction in nitric oxide consumption, reducing the risk of developing clinical pulmonary hypertension [Hill et al. 2010].

Patients with haemolytic PNH who do not require eculizumab treatment should undergo annual echocardiography. Those with silent pulmonary emboli should be commenced on anticoagulation and eculizumab.

Prevention

Patient education is essential for all aspects of PNH symptoms, in particular for thrombosis. All patients with a granulocyte clone size of >50% should be commenced on primary prophylaxis providing there are no contraindications [Hall et al. 2003]. Despite of full dose anticoagulation, patients remain at an increased risk of thrombosis as discussed above, and thus patients should remain vigilant. Patients on primary prophylaxis who commence on eculizumab can stop anticoagulation [Kelly et al. 2011]. There are no clinical trials assessing the use of direct oral anticoagulants for primary prophylaxis, and thus warfarin is recommended.

Due to the high incidence of concurrent aplastic anaemia and PNH, there remain difficult management decisions in those patients with thrombocytopenia who have a high granulocyte PNH clone and are not on complement inhibitors.

Pregnancy

Prior to eculizumab, pregnancy in patients with PNH was considered high risk and often not recommended due to the associated high morbidity and mortality for mother and foetus, in particular due to thrombosis. Maternal mortality ranges from 8% to 20%, and foetal mortality of 4%, with over 50% of deliveries preterm [Ray et al. 2000; Ziakas et al. 2007; de Guibert et al. 2011].

Maternal outcome has significantly improved in the eculizumab era, with no reported deaths, although foetal mortality remains at 4%, higher than the general population. Preterm delivery has reduced to 29%. Thrombosis during pregnancy is low, however postpartum thrombosis remains the highest risk period, affecting 3% of patients [Kelly et al. 2015]. Half to two-thirds of patients on eculizumab require either with increased frequency of eculizumab infusions (weekly) or a dose increase by 300 mg, particularly in the third trimester [Kelly et al. 2015; unpublished data UK National PNH service].

Pregnancy in PNH patients is a high-risk scenario for thrombosis and we advise all pregnant patients with a PNH granulocyte clone over 10% to be therapeutically anticoagulated with LMWH throughout pregnancy and for 3 months postpartum (provided there are no contraindications). In addition to therapeutic anticoagulation, eculizumab naïve patients with clones above 20% should be considered for treatment in countries where the treatment is available, during pregnancy and for three months postpartum. Patients are at risk of PNH clone expansion during pregnancy and should have this monitored every two months if not currently on treatment.

It should be noted that practice varies throughout the world with regard to anticoagulation dosing on review of the International PNH Registry data and some physicians will prescribe prophylactic doses of LMWH [Kelly et al. 2015].

Conclusion

Thrombosis is a recognized leading cause of death in PNH patients, both in the undiagnosed and the untreated population. Thrombosis remains a clinical emergency, and requires immediate commencement of eculizumab (in countries where it is available) and therapeutic anticoagulation. Prevention through patient and physician education, anticoagulation, and complement inhibition have reduced the morbidity and mortality associated with it; however as clinicians, the undiagnosed PNH patient who presents with acute thrombosis remains a clinical concern.