Abstract
Vitamin K antagonists are advised in pulmonary arterial hypertension patients despite a lack of safety data.
We reviewed major bleeding in three classes of pulmonary hypertension patients, all receiving vitamin K antagonists.
Bleeding event rates were 5.4 per 100 patient-years for patients with idiopathic pulmonary arterial hypertension, 19 per 100 patient-years for connective tissue disease related pulmonary arterial hypertension patients and 2.4 per 100 patient-years for chronic thromboembolic pulmonary hypertension patients. Life tables analysis showed that event-free survival was worse in patients with connective tissue disease related pulmonary hypertension than in patients with idiopathic pulmonary arterial hypertension (Wilcoxon=12.8; p<0.001), and patients with chronic thromboembolic pulmonary hypertension (Wilcoxon=23.2; p<0.001). Patients with idiopathic pulmonary arterial hypertension suffered more events than patients with chronic thromboembolic pulmonary hypertension (Wilcoxon=7.2; p<0.01). Major bleeding was independent of age, sex, target international normalised ratio (INR) range, documented INR, vitamin K antagonist type, or right atrial pressure, but was associated with use of prostacyclin analogues.
Major bleeding risk during vitamin K antagonist therapy differs among groups of patients with pulmonary hypertension. Further research regarding optimal anticoagulant therapy is needed, as well as risk–benefit analyses for pulmonary hypertension patients with a higher bleeding propensity.
Vitamin K antagonists (VKAs) are part of conventional drug treatment for pulmonary hypertension [1]. The observed hypercoagulability in idiopathic pulmonary arterial hypertension (IPAH) patients [2–5], and reported survival-benefit of VKAs in IPAH patients has led to recommendation of administration of VKAs in PAH patients [6], albeit with a class IIa, level of evidence C [6, 7]. This limited level of evidence is due to the lack of prospective comparative trials regarding administration of anticoagulants in PAH patients [7, 8]. Furthermore, clinical studies of systematic anticoagulation in PAH were methodologically limited because of the retrospective nature, lack of randomisation between placebo and anticoagulant drugs, or because of a diverse range of pulmonary hypertension patients [9, 10]. In connective tissue disease related pulmonary arterial hypertension (CTD-PAH) patients, there are no randomised controlled trials on the possible role of VKAs, for which reason the European Society of Cardiology/European Respiratory Society did not make any recommendation on their use in their guidelines [6]. The American College of Chest Physicians guidelines support the use of VKAs in IPAH patients, and advise consideration of their use in CTD-PAH patients, especially when patients have an indwelling catheter for long-term prostanoid treatment [7]. In chronic thromboembolic pulmonary hypertension (CTEPH) patients, a class I, level of evidence C, recommendation is given for VKA administration [6]. Hence, these guidelines suggest that the underlying disease, predisposing to pulmonary hypertension, may influence the outcome of routine administration of VKA [6]. The platelet-inhibiting effect of prostacyclin analogues is widely acknowledged, yet its clinical relevance is still unclear, with respect to concomitant use of VKA [8, 11]. In contrast, sitaxentan interferes with VKA metabolism [12], and has been taken off the market due to increased risk of liver toxicity.
We aimed to investigate the safety of routine administration of VKAs by comparing the incidence of major bleeding associated with the use of VKAs in our population of IPAH patients, CTD-PAH and inoperable CTEPH. We chose IPAH and CTD-PAH patients, since these well-defined patient groups constitute the majority of PAH.
MATERIALS AND METHODS
Study design
We performed a retrospective analysis of pulmonary hypertension patients treated in the VU Medical Center hospital, Amsterdam, the Netherlands, from 1997 until 2009, and evaluated the safety of routine administration of VKAs (Acenocoumarol and Fenprocoumon). More explicitly, we analysed patients with IPAH, CTD-PAH and CTEPH. The IPAH patients included both sporadic and familiar cases. We combined data from the VU Medical Center and from the specialised anticoagulation clinics, where patients were monitored. We obtained informed consent from patients or their relatives (if the patient was deceased).
Patients
We documented the occurrence of major bleeding, as well as nosebleeding. We did not include patients who refused anticoagulant drug treatment (n=2) in this study. We did not include patients with liver disease (n=2) because of spontaneously prolonged international normalised ratio (INR) values, who were therefore not treated with VKA. We compared groups of patients with IPAH, CTD-PAH and CTEPH for the rate of occurrence of major bleeding.
Patients were prescribed VKA (predominantly acenocoumarol; to a lesser extent fenprocoumon). Drugs prescribed primarily for attenuation of pulmonary vascular resistance were administered according to guidelines and local reimbursement, and included prostacyclin analogues, endothelin-receptor antagonists, and phosphodiesterase-inhibitors.
Data collection and definitions of events
The following information was extracted from the VU Medical Center computerised patients records: sex, age, type of anticoagulation, major bleeding, prothrombin-time measurements (INR) at the time of major bleeding, hospital admissions data and, where applicable, time of death. We documented use of non-steroidal anti-inflammatory drugs (NSAIDs), platelet-inhibiting drugs (aspirin and clopidogrel) and platelet count. We considered bleeding to be major if it was: clinically overt and associated with a fall in haemoglobin of at least 20 g·L−1 (≥1.2 mmol·L−1); resulted in the need for transfusion of at least two units of red cells; involved a critical site; or if it was fatal. Critical sites included the central nervous system, the eye and respiratory tract. We also documented clinically relevant bleeding and subcategorised these accordingly. INR target range was determined by the anticoagulation clinics.
Statistical analysis
We used the SPSS for Windows Software package (Version 16: SPSS Inc, Chicago, IL, USA). We describe the data as mean±sd or median (interquartile range (IQR)). We plotted one-minus-survival curves based on life tables analysis. We calculated bleeding-free survival time as the time from initiation of VKA treatment until the occurrence of major bleeding. We counted major bleeding only once for each patient for life tables analysis. We investigated the possible influence of age, type of VKA used, and the INR target range on the occurrence of major bleeding, by performing a Cox proportional hazards analysis with a backwards stepwise regression analysis. The relatively small number of patients precluded a direct analysis of influence on bleeding of all possible pulmonary hypertension attenuating drug therapies, and combinations thereof. We therefore chose to limit ourselves to the evaluation of an association between prostacyclin analogues and sitaxentan on bleeding, using cross tabulations with Pearson's Chi-squared test. We compared the last documented INR value before bleeding with the target range INR. We considered a p-value of <0.05 statistically significant.
RESULTS
Patients
A total of 218 PAH patients were available in the database, of whom 198 patients were included in the study: patients diagnosed with IPAH (n=99), CTD-PAH (n=39) and CTEPH (n=60). A total of 20 patients were never prescribed any VKA drug, and were not included in further analyses: IPAH (n=7), CTD-PAH (n=12) and CTEPH (n=1). Four of the latter patients experienced major bleeding events, before consideration of VKA drug prescription. One IPAH patient had recurrent haemoptysis, one CTD-PAH patient had arteriovenous malformation related bleeding and two CTD-PAH patients had major gastrointestinal bleeding, requiring transfusion; one of these during use of acetylsalicylic acid 100 mg once daily. Patient characteristics are described in table 1. Patients with IPAH were generally younger than CTD-PAH patients and CTEPH patients (table 1, p<0.001 for both). There was no significant difference between the age of CTD-PAH patients and CTEPH patients (p=0.982). Right atrial pressure was not different between the three patient groups. Mean pulmonary artery pressure was higher in IPAH patients than in CTD-PAH patients (p=0.02). INR target ranges were 2.0–3.0 in 29 patients, 2.5–3.5 in 107 patients and 3.0–4.0 in 53 patients; the choice of a range was irrespective of pulmonary hypertension aetiology. The INR target range was intractable in nine patients. There were no significant differences in INR target range between the different patient groups.
Bleeding events
A total of 24 major bleeding events occurred in 99 IPAH patients during 448 patient-exposure-years (bleeding rate 5.4 per 100 patient-years), 15 major bleeding events occurred in 39 CTD-PAH patients during 81 patient-exposure-years (bleeding rate 19 per 100 patient-years), and seven events occurred in 60 CTEPH patients during 291 patient-exposure-years (bleeding rate 2.4 per 100 patient-years; table 2). Fatal bleeding occurred in one patient, who experienced a massive sub-pleural haemorrhage 3 months after combined heart/lung transplantation. The other sites of major bleeding events were intracranial (subdural bleeding; one patient), intraocular (three patients), pulmonary (nine patients), gastrointestinal (25 patients) and urogenital (four patients). One patient developed a critical anaemia due to diffuse, large haematomas, and was diagnosed with hairy cell leukaemia. Two patients suffered bleeding as a consequence of an intervention (one groin bleeding after diagnostic heart catheterisation; one bleeding after portacath placement). The subdural bleeding was non-traumatic in a patient with the antiphospholipid syndrome (target INR range 3.0–4.0) and was treated conservatively.
We counted major bleeding only once for each patient, but for clarity we also documented any bleeding event, including non-major bleeding (table 2). Nosebleeding occurred in a considerable number of patients, but there was no patient with nosebleeding requiring blood transfusion (table 2).
According to life tables analysis (fig. 1) CTD-PAH patients had worse event-free survival than IPAH patients (Wilcoxon=12.0; p<0.001) and CTEPH patients (Wilcoxon=21.5; p<0.001). Event-free survival was less in IPAH patients than in CTEPH patients (Wilcoxon=7.2; p<0.01). Cox proportional hazards analysis found no significant relationship between major bleeding and age, sex, target INR range, type of VKA drug used, or right atrial pressure (table 3). Target INR range did not influence the occurrence of nosebleeding either. The mean last documented INR before major bleeding was not different from the mean INR target range (3.4±1.6 versus 3.1±0.6; p=0.19). In 13 patients the INR at the time of the adverse event was unavailable. For these patients the median number of days between the last documented INR and the occurrence of an adverse event was 7 days (IQR 3–16 days).
Cumulative vitamin K antagonist related major bleeding in pulmonary arterial hypertension patients. Patients with connective tissue disease related pulmonary arterial hypertension (CTD-PAH) experienced worse event-free survival than idiopathic pulmonary arterial hypertension (IPAH) patients (Wilcoxon=12.0; p<0.001) and chronic thromboembolic pulmonary hypertension (CTEPH) patients (Wilcoxon=21.5; p<0.001). Event-free survival was also less in IPAH patients than in CTEPH patients (Wilcoxon=7.2; p<0.01). *: p<0.05.
The small number of patients using sitaxentan (n=7) precluded a conclusion on safety regarding concomitant VKA drug use, but there was a significant association between the use of prostacyclin analogues as pulmonary hypertension therapy and major bleeding (table 4; Pearson's Chi-squared=15.1; p<0.01). We counted just one major bleeding event, a large haematoma after placement of a permanent subcutaneously situated venous catheter used for infusion of epoprostenol, which was treated conservatively. All other bleeding events were unrelated to the prostacyclin administration site or related procedures. NSAID use was documented in five patients, and aspirin use in nine patients. There was no relationship with bleeding. Platelet count was available in 185 patients (93.4%) and was not different between groups of IPAH, CTD-PAH and CTEPH patients, but was related to bleeding (table 5). There were no patients with end-stage renal disease. Several patients had an indication for VKA apart from having pulmonary hypertension (table 6). After the occurrence of major bleeding, physicians discontinued VKA permanently in 13 patients (seven IPAH patients, six CTD-PAH patients). Despite this, subsequent transfusion demanding bleeding occurred in two of these patients. Physicians discontinued VKA temporarily in five patients (five IPAH patients). In these patients three transfusion demanding bleeding events occurred after resumption of VKA. Physicians continued VKA in the remaining 29 patients. In the latter group there were seven subsequent major bleeding events, one of which was gastrointestinal bleeding, and three were haemoptysis. Nosebleeding was documented in four patients. In patients with haemoptysis, embolisation of bronchial arteries was performed. All patients subsequently resumed VKA treatment.
DISCUSSION
The key finding of this study is that routine VKA administration to patients with different types of pulmonary hypertension was associated with an unequal incidence of major bleeding. CTD-PAH patients had a markedly higher rate of major bleeding complications than IPAH or CTEPH patients. IPAH patients in turn demonstrated a higher incidence rate of major bleeding than CTEPH patients.
The clinical efficacy and safety of oral anticoagulant therapy has been well established in patients with a high risk of thrombotic complications, such as those with mechanical heart valves or atrial fibrillation [13, 14]. In general, optimal oral anticoagulant therapy intensity (i.e. the level at which thromboembolic complications are effectively prevented without excessive bleeding) has been well documented by studying large numbers of patients [15]. In pulmonary hypertension patients the latter has not yet been feasible. However, retrospective reports suggest a better outcome in IPAH patients treated with VKAs [9, 10]. Nevertheless, despite the fact that pulmonary hypertension is a common denominator, IPAH patients, CTD-PAH patients and CTEPH patients constitute very different groups of patients. Hypothetically, CTEPH patients should benefit from life-long anticoagulant therapy, similar to patients with recurrent deep-vein thrombosis or pulmonary embolism, since hypercoagulability as well as impaired clot resolution are considered the root of the disease [16, 17]. In our cohort, major bleeding in CTEPH patients was restricted to two patients suffering from gastrointestinal bleeding, four patients with a transfusion demanding fall in haemoglobin level, without an overt focus of bleeding, and one case of haemoptysis. Overall, the incidence of major bleeding was well within limits reported in large studies with VKAs and lower than in IPAH [18]. A representative overview of bleeding in large groups of patients on VKA for atrial fibrillation or venous thromboembolism is given in table 7. The pulmonary arterial hypertension patients we describe are generally younger, and there is a female to male predominance. There was no difference in mean right atrial pressure between the different groups of patients, hence no hint of a predominance of liver congestion in any of these groups, which could otherwise have explained a higher risk of gastrointestinal bleeding. The occurrence of haemoptysis in patients with pulmonary hypertension is well known, but is hardly ever reported in patients using VKAs for atrial fibrillation or venous thromboembolism (table 7) [26]. Haemoptysis occurred predominantly in IPAH patients. IPAH patients may have an inherently higher risk of haemoptysis due to bronchial artery bleeding, which is most likely unrelated to VKA use [27]. CTD-PAH patients experienced an even larger number of major bleeding events than the two other groups. Patients with CTD-PAH tend to be older than IPAH patients, but in our study age did not explain this high rate of major bleeding. Aspirin and NSAID use was very limited, and did not explain bleeding. Documented platelet counts showed a relationship with bleeding, but the retrospective analysis precludes any statement regarding cause and effect. The high rate of gastrointestinal bleeding in CTD-PAH is likely due to gastrointestinal angiodysplasia [28], increasing the risk for unwanted side-effects of anticoagulant therapy.
With respect to concomitant pulmonary hypertension attenuating therapy used, we cannot conclude whether sitaxentan had any influence on bleeding, due the low number of patients exposed to this drug. The platelet-inhibiting effect of prostacyclin analogues has long been recognised [29]. The incidence of alveolar bleeding in PAH patients with concomitant use of epoprostenol and VKA has also been reported [11]. Since the use of prostacyclin analogues was lower in our group of patients without major bleeding, we looked for confounders that might explain these findings, other than prostacyclin analogues themselves. As such, we only identified one event related to a chronic intravenous catheter. Patients using epoprostenol may, on average, have had more advanced disease, yet no association with right atrial pressure was found. Major bleeding may, therefore, be associated with prostacyclin analogues through their platelet-inhibiting effect on top of the VKA [29]. The vasodilatory effect of prostacyclin analogues, most apparent in skin and intestines, may both predispose to bleeding, and enhance bleeding once vascular integrity is lost [30, 31].
Preliminary studies evaluating the benefit of pure platelet function inhibitors have so far not led to prospective trials evaluating optimal anticoagulant treatment in PAH patients [32]. Nevertheless, the major bleeding rate in our cohort of IPAH patients, and in CTD-PAH patients especially, not only supersedes that of CTEPH patients, but more importantly, the rate is much higher than reported bleeding rates in larger, prospective studies [13, 15, 18]. Although we did not find the observed INR target range to be of influence in this respect, one would expect that higher INR target ranges would eventually result in increased bleeding when complications do occur. So far, there has been no consensus on the optimal INR target range for pulmonary hypertension patients [4]. In the Netherlands, patients are dosed with VKA according to local policy in the anticoagulation clinics. Determination of optimal INR intensity will require prospective studies in large numbers of patients.
Whereas there have been many randomised trials evaluating the clinical benefit of new drugs attenuating pulmonary hypertension, there has been little attention to a closer evaluation of more conservative therapies. There have been no randomised controlled prospective studies determining the real benefits of treating different classes of pulmonary hypertension patients with VKA in the presence of other PAH specific therapy. The favourable results of VKA in early observational studies along with the overall poor prognosis of pulmonary hypertension patients [9, 10], however, have resulted in widespread use of VKA as a cornerstone of background therapy in pulmonary hypertension patients [33, 34]. Major bleeding events can nevertheless have important deleterious haemodynamic consequences for these patients and, especially, haemoptysis is often life-threatening [11]. The reported high incidence of VKA drug-related major bleeding in IPAH patients, but especially in CTD-PAH patients, warrants a closer look at the safety of routine administration of anticoagulant therapy to PAH patients.
Limitations of this study are its retrospective nature, as well as the patient data being from a single centre. The lack of a direct control group precludes any direct comparison with the observed bleeding indices. Given a relatively small patient number, the number of end-points was also small with inherent confidence limits. Although the poor survival for the CTD-PAH group is common, it precludes an adequate conclusion regarding long-term bleeding risk. Platelet function tests were not performed, therefore the reported platelet count is of limited value. Nevertheless, all data were derived directly from clinical practice, and documentation of all patients included in the study was also carefully recorded by the Dutch anticoagulation clinics.
In conclusion, oral vitamin K antagonist anticoagulation therapy is associated with different rates of major bleeding in different groups of pulmonary hypertension patients. IPAH patients, but CTD-PAH patients especially, appear to be at a markedly higher risk of major bleeding complications than would generally be accepted. Further studies are needed to determine the role of anticoagulant treatment in the standard of care for different groups of patients with PAH.
Acknowledgments
We conducted this study without external funding and we are very grateful for the help we received from the Dutch anticoagulation clinics, as well as for the patient care provided by them.
Dutch anticoagulation clinics who contributed: Stichting Trombosedienst Alkmaar, Alkmaar; Stichting Trombosedienst Eemvallei, Amersfoort; ATAL Medisch Diagnostisch Centrum, Amsterdam; Trombosedienst Rode Kruis Ziekenhuis, Beverwijk; Stichting Trombosedienst Delft en omstreken, Delft; Stichting Artsenlaboratorium en Trombosedienst, Den Helder; Stichting Regionale Trombosedienst Drechtsteden, Dordrecht; Stichting Trombosedienst Drachten en Omstreken, Drachten; Trombosedienst Neder-Veluwe, Ede Gld; Trombosedienst Scheper Ziekenhuis, Emmen; AntiStollingsCentrum Oost-Nederland ASCON, Enschede; Stichting Regionale Trombosedienst Breda, Etten-Leur; St. Annaziekenhuis, Geldrop Stichting Trombosedienst Gouda, Gouda; Stichting Medische Diagnostiek (Medial), Haarlem; Trombosedienst Ziekenhuis "De Tjongerschans", Heerenveen; Stichting Trombosedienst Oostelijk Zuid-Limburg, Heerlen; Trombosedienst van het Elkerliek Ziekenhuis, Helmond; Stichting Trombosedienst Hengelo en omstreken, Hengelo Stichting Trombosedienst voor het Gooi, Hilversum; Trombosedienst van het Ziekenhuis Bethesda, Hoogeveen; STAT Westfriesland, Hoorn NH; Trombosedienst Friesland Noord, Leeuwarden; Stichting Trombosedienst Leiden e.o., Leiden; Trombosedienst Flevoland, Lelystad; AntiStollingsCentrum Oost-Nederland ASCON, Lichtenvoorde; Stichting Trombosedienst Zeeland, Middelburg; Stichting I.N.R. Trombosedienst, Nijmegen; Trombosedienst NW Veluwe, Nunspeet; Trombosedienst Oldenzaal, Medisch Spectrum Twente, Oldenzaal; Trombosedienst Ziekenhuis Bernhoven, locatie Oss, Oss; Stichting Regionale Trombosedienst 's-Gravenhage e.o., Rijswijk ZH; Trombosedienst Roermond, Roermond; Trombosedienst St. Franciscus Ziekenhuis, Roosendaal; Stichting Trombosedienst & Artsenlaboratorium, Rotterdam; Stichting Trombosedienst 's-Hertogenbosch e.o., 's Hertogenbosch; Stichting Trombosedienst Schiedam e.o., Schiedam; Stichting Trombosedienst Zuid-West Friesland, Sneek; Trombosedienst West-Betuwe/Bommelerwaard, Tiel; Stichting Trombosedienst Midden Brabant, Tilburg; SALTRO, Utrecht; Stichting Trombosedienst Regio Eindhoven, Veldhoven; Trombosedienst Hofpoort Ziekenhuis, Woerden; Stichting Zeister Trombosedienst, Zeist Stichting Trombosedienst Zevenaar e.o., Zevenaar; Gelre Ziekenhuizen, Zutphen Trombosedienst Isala Klinieken, Zwolle.
Footnotes
For editorial comments see page 775.
Statement of Interest
None declared.
- Received March 4, 2012.
- Accepted August 1, 2012.
- ©ERS 2013