Non-vitamin K antagonist oral anticoagulants in patients with an increased risk of bleeding

Patients with chronic kidney disease (CKD) are particularly vulnerable to bleeding or thrombotic complications. The influence of kidney function on the coagulation system depends on the CKD stage. In advanced stages there is an increased bleeding risk as well as an increased risk for thrombosis. With progressing CKD, the risk of thrombosis increases more than the bleeding risk; however, in hemodialysis patients the bleeding risk exceeds the thrombotic risk. Therefore, decision-making in dialysis patients with an indication for oral anticoagulation is particularly challenging [10‐17].

The role of CKD as a risk factor for the development of bleeding complications is emphasized by the inclusion of CKD in various scoring systems for the assessment of the bleeding risk, e.g., the HAS-BLED score [18].

Of note, there is a new classification of CKD, based on the eGFR (stages G1–G5) as well as the stage of albuminuria (A1–A3; [8]). Although patients with an eGFR <25 ml/min were excluded from the major NOAC trials, the labels of all NOACs except dabigatran provide dosages for patients with an eGFR between 15 and 29 ml/min. Similarly, the current AF guidelines of the European Society of Cardiology (ESC) state that there are no controlled trials of NOACs in patients with a creatinine clearance (CrCl <25–30 ml/min); however, they do not state clearly whether or not NOACs should be used in patients with an eGFR between 15 and 29 m/min [7]. An Austrian consensus paper argues against the use of NOACs in patients with an eGFR <30 m/min [2]. In our opinion, CKD stage and not eGFR should be used in recommendations and guidelines. For instance, in the RE-LY study, an arbitrary eGFR lower limit of 50 ml/min was used [13]; however, that is not a cut-off between different CKD stages. More importantly, eGFR represents a dynamic parameter depending on many constantly changing factors in patients’ state of health, thus it has to be monitored closely in patients on NOAC therapy. Furthermore, although eGFR depends on the formula used for calculation, in many studies it is not clear which formula was used. Yet, for the same serum creatinine value, eGFR differs significantly, based on the formula applied (see Table 5).

As shown in Table 5, according to the Cockroft-Gault formula, more patients would qualify for NOAC treatment than according to the MDRD and CKD-EPI formulas. According to the KDIGO guidelines, the CKD-EPI formula is recommended to evaluate renal function [8]; however, as already mentioned the large randomized NOAC trials applied the Cockroft-Gault formula to assess CKD and therefore this formula should be used to estimate GFR to decide if the NOAC dosage needs to be reduced. It should be mentioned here that in the case of apixaban, dose adjustment is not based on GFR. A lower dosage of apixaban should be used if serum creatinine is ≥1.5 mg/dl (133 µmol/l) and either one of the following criteria is fulfilled: age ≥80 years or weight ≤60 kg. Important pharmacokinetic differences between the available NOACs must be taken into account (Table 1). Subanalyses of the large randomized NOAC trials which compared the efficacy and safety of the VKA warfarin with NOACs in patients with CKD and atrial fibrillation reported the following:

In summary, compared to warfarin, the net clinical benefit for rivaroxaban and edoxaban remains roughly the same, regardless of renal function [15‐17]. For apixaban, the benefit increases with decreasing renal function [14], whereas for dabigatran it decreases with decreasing renal function [13].

Although kidney function has been the subject of several studies and recommendations, there are currently no data on the use of NOACs in patients with massive proteinuria leading to nephrotic syndrome. It is to date unknown whether and how pharmacokinetics or pharmacodynamics (PK/PD) of NOACs are influenced by proteinuria. Currently, a phase 1/2 trial on PK/PD of apixaban is being conducted in patients with nephrotic syndrome [19]. The considerable differences in the rates of protein binding between the various NOACs could be of relevance in the context of nephrotic syndrome due to hypoproteinemia.

Another open question is how to deal with CKD patients of older age, relevant comorbidities or acute worsening of the health condition. For instance, there are data showing that 23% of patients hospitalized with acute heart failure experience a worsening of renal function which remains permanent in 10% and is transitory in 13% [20]. (A transitory worsening of renal function in this study was defined as an increase of serum creatinine by ≥0.3 mg/dl and/or by ≥25% from baseline at the index hospitalization which did not persist until the last measurement before discharge). Patients with unstable kidney function need close monitoring since there might be a need for dose adjustment or withdrawal of the NOAC. The frequency of evaluating renal function also depends on the clinical situation. In cases of acute clinical problems (e.g., intercurrent infections), renal function should be monitored more closely. Furthermore, interactions with other substances might significantly affect plasma levels of NOACs (e.g., some antibiotics and antimycotics). In such cases, therapeutic drug monitoring for the NOAC should be considered.

In patients with CKD stage 5 who are already on or about to be put on renal replacement therapy, there is a clear contraindication for NOACs; however, new pharmacological data indicate that apixaban [21] as well as rivaroxaban (15 mg/day; [4]), both showing the lowest renal excretion rates among NOACs [1], might be used without causing major problems in hemodialysis patients [4, 21]. Of note, patient numbers in these analyses were extremely low. Thus, the use of apixaban and rivaroxaban in the dialysis population cannot be recommended at this point.

Biological age can vary considerably and does not necessarily match chronological age. Patients ≥80 years of age represent a special population for a number of reasons. First, there is a constant decline in renal function with increasing age (see Section “Chronic kidney disease”) and one should bear in mind that renal function is a very dynamic parameter. Furthermore, impairment of liver function is common in the elderly and can be monitored by synthesis parameters like the prothrombin time. Finally, an increased risk of falls and osteoporosis results in an elevated risk of bone fractures.

Several NOAC labels include specific instructions for dose reductions for patients ≥80 years. For dabigatran, dosage should be reduced to 2 × 110 mg daily in VTE and NVAF, if patients are ≥80 years. In contrast, the dose adjustment of apixaban depends on the indications and further criteria. For NVAF, the apixaban dose should be reduced to 2 × 2.5 mg in patients ≥80 years, if 1 of the following 2 other criteria is met: body weight ≤60 kg or serum creatinine ≥1.5 mg/dl. For VTE, no dose reduction is recommended for apixaban. For rivaroxaban and edoxaban, no dose adjustment is recommended for patients ≥80 years.

A recent study tried to determine whether the rates of oral anticoagulation in frail older adults with AF are appropriate or not and how useful risk prediction scores are in this context [22]. For 225 residents of a nursing home with a clinical frailty scale score ≥5, CHA₂DS₂-VASc [1] and HAS-BLED scores were retrospectively calculated. It turned out that bleeding risk was relatively low whereas stroke risk was high. Nevertheless, only 20% of the patients were orally anticoagulated.

In summary, in patients ≥80 years the decision for oral anticoagulation must be made on an individual basis. Several factors (biological age, frailty, dementia, comorbidities, renal and liver function, bleeding risk etc.) should be considered. Most importantly, if a patient ≥80 years is put on a NOAC, this patient should be monitored closely (renal and liver function, nutritional status, consider use of a frailty score, blood in feces, cognitive impairment).

Patients with hereditary or acquired disorders of the blood coagulation system, with thrombocytopenia or platelet function disorders are often at an increased risk of bleeding. The most prevalent disorders are mentioned here.

As the age of the hemophilia population rises, the prevalence of NVAF in hemophilic patients is likely to rise as well. In a report from 2014 the general prevalence of NVAF in people with hemophilia (PWH) in Europe was 0.8%, rising to 3.4% in patients older than 60 years [23]. A European consensus statement focused on the management of AF in PWH [24]. Stroke risk in PWH seems to be lower than in the general population; however, this may only be true for patients with severe hemophilia [25] as this effect was not found in non-severe hemophilia. Consequently, the lower stroke rate in PWH might increase in patients with factor replacement therapy [24].

There is agreement that NVAF in PWH should be treated by a multidisciplinary team, with a hematologist leading anticoagulation therapy [24]. The CHA₂DS₂Vasc score should be used, but it may overestimate stroke risk in PWH [24]. Conversely, the HAS-BLED score should not be used in PWH as it would substantially underestimate the bleeding risk [24]. Oral anticoagulation with VKAs or NOACs may be considered in PWH with a very high risk of stroke if trough levels of FVIII/IX are adequate (Table 6; [24]). The definition of very high risk has to be determined individually and cannot be determined by a single parameter [24]. Of note, there are no data for the use of NOACs in PWH.

Von Willebrand disease (vWD) is one of the most frequent clotting disorders; however, it is still rare with approximately 0.01% clinically affected cases in the population [26]. Von Willebrand factor (vWF) levels have been shown to be increased in patients with NVAF [27], thus is can be assumed that also in patients with a low vWF the levels increase, not only with age but also when they have NVAF. There is a great variability of severity, with severe forms being very rare, and the risk of stroke in patients with vWD is unknown. The cut-offs for anticoagulation suggested by experts in patients with hemophilia might also be applicable for the vWF levels (Table 6); however, as in other bleeding disorders, a case by case decision, balancing the risk of bleeding with the risk of thrombosis or stroke, has to be made for each individual patient with vWD.

Data on thrombocytopenia and anticoagulation are scarce [28]. In guidelines for treatment of patients with cancer, a platelet count of >50 G/l is considered as appropriate for therapeutic anticoagulation [29]. This cut-off has also been used in recent studies [30]. In patients with platelet counts between 20 and 50 G/l anticoagulation may be considered only in very specific situations on a case by case basis and preferably prophylactic doses of heparins should be used. Below a platelet count of 20 G/l, anticoagulation is contraindicated.

In patients with autoimmune thrombocytopenia, if there is a strong indication for anticoagulation, the risk of a thrombotic event should be evaluated (e.g., CHA₂DS₂-VASc score, if applicable). If there is an indication for temporary anticoagulation (as in some VTE cases), a short-term anticoagulation (e.g., 3 months) should be used; however, if long-term anticoagulation is needed in patients with autoimmune thrombocytopenia, treatment options that are capable of increasing the platelet count should be considered [31]. If a patient with high bleeding risk presents with acute VTE, a vena cava inferior filter may be considered as an alternative to anticoagulation [32].

European guidelines on the management of major bleeding and coagulopathy following trauma recommend that in acute bleeding of patients on NOACs the plasma level of the NOAC should be measured [34]. A position paper about reversal strategies for NOACs by the ESC Working Group on Cardiovascular Pharmacotherapy and the ESC Working Group on Thrombosis is also available [33].

For dabigatran, there is an antidote available. The Fab antibody fragment idarucizumab binds to dabigatran in a 1:1 ratio and is licensed as a specific antidote to dabigatran [35]. Usually, a single dose of 5 g idarucizumab, which is rapidly excreted by the kidney is sufficient as an antagonist to dabigatran. Dabigatran plasma concentrations could increase again within 24 h, possibly due to a shift of dabigatran from the extracellular space to the circulation, demanding a second administration of idarucizumab [36]. In the REVERSE-AD study 1.8% of patients received a second dose of the antibody [37]. Readministration of dabigatran 24 h after administration of idarucizumab is possible. An impairment of renal function does not seem to influence the effect of idarucizumab on dabigatran, although idarucizumab exposure (as well as dabigatran concentration) increases with decreasing renal function. Particularly, older patients and patients with reduced renal function need the full dose of 5 g idarucizumab [38]. After the administration of idarucizumab, dabigatran plasma levels should be checked again in order to measure the efficacy in the individual patient, particularly if there is insufficient clinical hemostasis [39, 40].

The specific antidote for factor Xa inhibitors, i.e. andexanet alpha, has so far only been approved for clinical use in patients in the USA. Until andexanet alpha becomes available in Europe, high-dose prothrombin complex concentrates (PCCs; dose of 50 U/kg) should be used to antagonize factor Xa inhibitors if necessary [34]. In the latter case, one should be aware of an increased thromboembolic risk. If the patient is bleeding, intravenous tranexamic acid (15 mg/kg or 1 g) should additionally be considered.

In emergency situations, specific tests should be used. For dabigatran, the Hemoclot™ test (Hyphen Biomed, Neuville-sur-Oise, France) or the ecarin clotting time should be used; for factor Xa inhibitors, calibrated anti-Xa activity tests are recommended. If those are not available, for dabigatran (diluted) thrombin time, and for factor Xa inhibitors non-calibrated anti-Xa tests may be used to exclude residual thrombin inhibition and anti-Xa activity, respectively; however, these tests are not sufficient to guide therapy.

In patients who survived the acute situation a comprehensive evaluation of a restart of anticoagulation should be performed. Particularly in patients with a reversible bleeding risk factor, anticoagulation should be restarted as the risk of thromboembolism is still high in this vulnerable patient group. In patients with NVAF and an irreversible risk factor and therefore a contraindication for anticoagulation, an occlusion of the left atrial appendage should be considered [7].

In VTE patients with malignancies, LMWH was the gold standard of anticoagulation for a long time. It was shown that in cancer patients LMWH is more effective than VKA in preventing VTE recurrences without increasing the risk of bleeding [45]. A network meta-analysis suggested that NOACs are at least equivalent if not better than VKAs in cancer patients with VTE. The indirect network comparison between NOACs and LMWH indicated comparable efficacy and a non-significant relative risk reduction towards improved safety with NOACs. These results prevailed after adjusting for different risk of recurrent VTE and major bleeding between LMWH vs. VKA and NOAC vs. VKA studies [44].

A recent large randomized study (HOKUSAI VTE Cancer) demonstrated non-inferiority of a combined primary outcome of recurrent VTE and severe bleeding for edoxaban versus dalteparin [46]. Interestingly, on edoxaban recurrent events were less (not significantly), but there was an increased risk of gastrointestinal bleeding, primarily in patients with gastrointestinal tumors. Another randomized study (SELECT-D) with less study participants comparing rivaroxaban to dalteparin in patients with cancer was recently published [47]. In this study the recurrence rate of VTE was lower, but the risk of major and clinically relevant non-major bleeding was higher in patients receiving rivaroxaban. There were no significant differences in mortality in both studies [46, 47].

The available published international guidelines still emphasize that NOACs are not the first choice for VTE treatment or prophylaxis in cancer patients [48]; however, these guidelines were published before the HOKUSAI VTE Cancer study was available. The authors of this consensus agree that NOACs (especially edoxaban and rivaroxaban) are a potential alternative to LMWH. In particular, the increased bleeding rate in cases of gastrointestinal tumors has to be kept in mind, and in these patients the potentially higher bleeding risk has to be balanced against the higher convenience of an oral versus a subcutaneous drug. Moreover, in patients with thrombocytopenia caution with full dose NOACs is recommended. After an initial treatment period of 3–6 months with LMWH, if further anticoagulation is indicated, patients can be switched to VKAs or NOACs on the basis of individual choices, if the active phase of the malignancy is over and the tumor is in a stable condition [48].

Further studies directly comparing NOACs with LMWH ± VKA in secondary prophylaxis are under way [49‐51].

In patients with cancer and NVAF either NOACs or VKAs may be used for stroke prophylaxis. If NOACs are given, interactions with various cancer treatments should be taken into account [2].