Anticoagulation for Stroke Prevention in Older Adults with Atrial Fibrillation and Comorbidity: Current Evidence and Treatment Challenges

The overall prevalence of atrial fibrillation (AF) is increasing due to population aging and an increase in prevalence of other conditions that predispose to AF. It is projected that by 2030 there will be 12.1 million AF cases in the US alone.1), 2) Advanced age is a critical determinant of the AF-related stroke and systemic embolic risks mandating oral anticoagulant (OAC) therapy.3) However, advanced age is associated with an increase in bleeding risk, which compounds the risk associated with anticoagulation.4)

This risk is further exacerbated by the fact that advanced age is often accompanied a higher comorbidity burden, polypharmacy, and frailty that further increase the risk of bleeding creating a therapeutic challenge for clinicians. Chronic kidney disease (CKD) is of particular interest since up to one-third of outpatients with AF have CKD,5) and 15–20% of CKD patients have AF.6), 7) As the incidence of CKD is increasing,8) an ever-growing patient population with both AF and CKD will require our attention. Also, current approaches of estimating the thromboembolic or bleeding risks with scores such as Congestive Heart Failure, Hypertension, Age ≥75 (Doubled), Diabetes Mellitus, Prior Stroke or Transient Ischemic Attack (Doubled), Vascular Disease, Age 65–74, Female (CHA₂DS₂-VASc) and Hypertension, Abnormal Renal/Liver Function, Stroke, Bleeding History or Predisposition, Labile INR, Elderly, Drugs/Alcohol Concomitantly (HAS-BLED) may be insufficient in fully appreciating the net clinical benefit of OAC use in elderly patients in whom frailty and multi-comorbidity may play more important roles than traditional clinical risk factors.

A patient's risk of adverse outcomes, particularly major bleeding and more importantly the net clinical benefit of OAC may not be fully approached by simply calculating the CHA₂DS₂-VASc and HAS-BLAD scores. A comprehensive analysis of patient's correct clinical status and treatment regimen is needed. The aim of this review is to appraise the current evidence, review the challenges, and provide recommendations on the decision-making for stroke prevention in older adults with AF and comorbid conditions including kidney disease, frailty, recurrent falls, and polypharmacy (Table 1).

Table 1
Factors to consider in the management of elderly patients with AF

• Click for larger image
• Click for full table
• Download as Excel file

Advanced age is one of the most important patient-related factors one needs to consider in the management of AF particularly when aiming to prevent thromboembolic events. The elderly generally have a higher burden of comorbidities, polypharmacy9) and are more often frail. This combination complicates the decision-making of anticoagulation use due to the more significant competing risks as compared to younger populations. Increased age is one of the strongest independent predictors of thromboembolic events in patients with AF, accounting for a 1.5-fold increase in the relative risk of stroke for every additional decade.3) However, the risk of stroke is paralleled by an increase in treatment-related bleeding, and particularly intracranial bleeding,4) giving clinicians pause before initiating OAC therapy. This duality is well represented by the inclusion of age as a risk variable in both CHA₂DS₂-VASc and HAS-BLED scores.10), 11) Some evidence suggests that advanced age has a greater impact on the risk of bleeding than the risk of stroke, particularly above the age of 75 years.12), 13)

While aging is inevitably associated with some physiological decline, the extent this decline varies widely between individuals. Frailty may be viewed as an exaggeration of the aging process23) resulting in loss of reserve and a diminished ability to withstand stress.24) Frailty is very common with an estimated prevalence of 9.9% among individuals aged over 60 according to a large British study.25) The prevalence is higher among women, the elderly and among racial and ethnic minorities.25), 26), 27) It is associated with increased risk of adverse outcomes including falls, delirium, disability and mortality.27), 28), 29), 30) Identifying and quantifying frailty is not straightforward. Therefore, the use of well validated scales is important. The most widely used and validated models31) are the phenotype model28) and the Canadian Study of Health and Aging (CSHA) frailty index.24)

AF is common among frail patients; moreover, AF has been shown to be associated with frailty independently of age, sex and other common comorbidities.32) While frailty correlates with higher CHA₂DS₂-VASc and HAS-BLED scores in AF patients, it may be independently predictive of both cardiovascular and all-cause mortality.33) It is, therefore, unsurprising that frailty is a common reason for discontinuing or not starting OAC.34), 35) There is very little information about the risk and benefits of long-term OAC treatment in this population. There are no randomized control trials (RCTs) comparing OAC to placebo and observational studies have the potential to be influenced by selection bias.

A large observational study based on US claims data compared propensity matched frail patients with AF treated with novel oral anticoagulants (NOACs) to patients treated with warfarin.36) Frailty was defined by the Johns Hopkins Claims-based Frailty Indicator,37) an algorithm design to identify patients meeting the Fried's Frailty Phenotype.28) A total of 2,700, 2,784, and 5,270 patients on apixaban, dabigatran, and rivaroxaban respectively, were compared separately to an equal number of patients treated with warfarin. Apixaban and dabigatran resulted in similar rates of thromboembolism as warfarin. However, rivaroxaban was associated with reduced thromboembolic risk at 2 years without a significant difference in bleeding. However, these findings should be interpreted cautiously due to potential limitations of low statistical power and residual confounding.37)

CKD is defined as a glomerular filtration rate (GFR) <60 mL/min/1.73 m² for 3 months or longer or with the presence of albuminuria.38) It is independently associated with a 3.7-fold increase in the risk of stroke. CKD has a profound impact on platelet function, coagulation factors and the endothelium resulting in a pro-thrombotic state. This leads to increased activity of the coagulation system and platelet aggregation39) which have additive effects in addition to AF-related thrombogenesis. In patients with both AF and CKD there is a 35% increase in relative risk of death40) and more than a 2-fold increase in major cardiovascular adverse effects.41)

The combination of AF and CKD is increasingly common, partly due to population aging and the increase in the prevalence of shared risk factors such as hypertension, diabetes and advanced age but also due to sheared pathophysiological processes that create a bidirectional relationship.42) Thus, subjects with CKD are at a higher risk of developing AF and the presence of AF increases the risk of developing renal dysfunction.42) Moreover, AF is associated with more rapid progression of CKD and a 67% higher risk of developing end-stage renal disease (ESRD),43) contributing to adverse outcomes and mandating close follow up with dose adjustments or even discontinuation of antiarrhythmic drugs, anticoagulants and other medications over time.

Patients with AF and CKD present a particular therapeutic challenge due to the higher bleeding risk.44) Several large registries have demonstrated an increased risk of both intracranial and gastrointestinal bleeding that parallels the degree of renal dysfunction. Furthermore, renal dysfunction is associated with larger intracranial hematoma volumes, worst outcome after intracerebral hemorrhage45) and an increased risk of hemorrhagic transformation after ischemic stroke.46) This is the result of a multifactorial paradoxical decrease in platelet adhesion and aggregation. Possible pathophysiological mechanisms include decreased storage and secretion of platelet-activating mediators, altered intra-platelet calcium mobilization, disturbances in platelet aggregation, increased formation of vascular prostaglandin I2, impaired platelet glycoprotein IIb-IIIa receptor activation and its binding to fibrinogen and von Willebrand factor, anemia, and the presence of uremic toxins.47), 48), 49)

Current guidelines50), 51) recommend using the CHA₂DS₂-VASc score as the default risk assessment tool for thromboembolic events, which, however, does not incorporate CKD as one of its component risk variables. Despite evidence of higher thromboembolic risk, CKD is not routinely considered as part of the thromboembolic risk stratification process. Attempts to incorporate CKD into the CHA₂DS₂-VASc score failed to show consistent improvement in its predictive value.52), 53) This is likely the result of strong associations with the other co-variants that comprise the score.54) In light of these results, it is unclear if there is a subpopulation of AF patients in whom a certain degree of CKD may justify OAC despite an otherwise low or borderline CHA₂DS₂-VASc score.

Nevertheless, CKD is a component of all major bleeding risk scores, namely, the HAS-BLED, Anticoagulation and Risk Factors in Atrial Fibrillation (ATRIA), and Outcomes Registry for Better Informed Treatment (ORBIT) scores, with varying definitions of CKD between scores. All 3 scores have been validated for both warfarin10), 55) and the non-vitamin K antagonists (VKA) in current use and have been found to have a modest predictive value for bleeding. According the European Society of Cardiology (ESC) guidelines, a high score in one of those scales should not be considered a reason to withhold anticoagulation but rather a reminder to better control all the reversible factors such as labile international normalized ratios, excess alcohol, or concomitant use of nonsteroidal anti-inflammatory drugs and aspirin. Furthermore, in general the net clinical benefit of ischemic stroke prevention versus serious bleeding may even be greater in patients with a high HAS-BLED score.56)

The net benefit of OAC treatment in patients with ESRD undergoing dialysis remains a matter of debate since it has never been addressed in a placebo controlled randomized controlled trial. The risk of thromboembolism is particularly high in this population,44), 66), 67) but it is accompanied by a significant increase in bleeding risk.68) Furthermore, AF is associated with a significant increase in mortality risk in ESRD patients.67) There is abundant evidence that warfarin lowers the risk of stroke in patients with severe CKD including those treated with dialysis, but it is equally well established that it also results in a significant increase in bleeding risk.44), 69) The question of net clinical benefit among dialysis patients was directly addressed in an analysis of a large national Danish registry.58) The authors found no decrease in the combined endpoint of stroke of bleeding, nor a decrease in all-cause mortality or cardiovascular death. Furthermore, use of warfarin in ESRD is associated with a higher risk of calciphylaxis70) possibly due to the inhibition of Matrix Gla protein, which is a vitamin K-dependent inhibitor of vascular calcification.71) The prevalence of calciphylaxis in chronic dialysis patients is estimated at 4% and warfarin may increase this risk more than 3-fold.72)

Patients with ESRD were excluded from all NOAC RCTs and therefore all the available data regarding their use stems from observational studies. The largest cohort to date included 281 and 244 patients treated with dabigatran and rivaroxaban, respectively, and showed that both were associated with an increase in major bleeding compared to warfarin.73) Dabigatran was even associated with a higher risk of hemorrhagic death. Importantly, the bleeding risk was significantly lower when rivaroxaban 15 mg o.d. or dabigatran 75 mg b.i.d. were used. The rates of stroke were too low to allow for any comparison. These results are unsurprising considering the pharmacokinetic studies that demonstrated significant exposure to the drugs in ERSD.74), 75)

Similar to dabigatran and rivaroxaban, ESRD patients were not included in the pivotal trial that led to the FDA approval of apixaban. However, the FDA approved an updated labeling for use of apixaban in ESRD based on a pharmacokinetic study.76) The FDA-approved dose is 5 mg b.i.d. (unless the usual criteria for dose reduction are met) although other studies suggest that this may result in supra-therapeutic levels.77) Data on clinical outcomes with use of apixaban in dialysis-dependent ESRD patients is very limited. A meta-analysis of 5 small observational studies concluded that apixaban was associated with a reduced bleeding risk and a similar stroke rate compared to warfarin.78) Similar conclusions were reached by a larger administrative claims-based analysis of Medicare beneficiaries in the United States that included 2,351 patients on apixaban and 23,172 patients on warfarin.79) In prognostic score-matched cohorts, there was no difference in the rates of thromboembolism between the apixaban and warfarin groups, but apixaban was associated with significantly lower risk of major bleeding (HR, 0.72; 95% CI, 0.59–0.87; p<0.001).These data support the use of apixaban over warfarin for stroke prevention in ESRD patients on dialysis who do not have a prohibitive bleeding risk and in whom anticoagulation is to be used. The larger question of which patients on dialysis would benefit anticoagulation and which would do better without this therapy remains largely unanswered.

The efficacy and long-term safety of edoxaban in ESRD patients was never tested in a clinical study. The drug levels, short-term safety and biomarkers of blood coagulation and fibrinolysis of a reduced dose (15 mg once a day [q.d.]) were tested in a small group of patients severe CKD (30≥ CrCl ≥15 mL/min). Plasma drug levels were comparable patients with normal real function receiving 60mg qD as was the bleeding rate.80) Similar results have been reported regarding reduced dose of apixban (2.5 mg b.i.d.) in dialysis77) patients and rivaroxaban (10 mg q.d.) in ESRD.81) However these are not enough to insure long term safety and provide no information about efficacy. Therefore, the European Heart Rhythm Association issued no recommendation for routine use of any anticoagulation for ESRD patients with AF50), 82) and the American Heart Association and American College of Cardiology referend from recommending any of the NOACs in this setting.83)

Renal function is affected by multiple factors, including medications and comorbidities. As described previously, AF is associated with a higher risk of new CKD and more rapid progression of existing CKD.42), 43) An important consideration in this regard is anticoagulation-related nephropathy (ARN), a common but often underdiagnosed syndrome that is estimated to affect 20% of patients treated with warfarin.84) It is defined as acute kidney injury (AKI) without obvious etiology in the setting of an INR >3.0. ARN can lead to new-onset or progression of CKD, and is associated with significant increases in short-term and long-term mortality.84), 85), 86) Renal biopsies of these patients reveal dysmorphic red blood cells in the glomerulus under electron microscopy, a uniform presence of hemorrhage and the absence of active glomerulonephritis or other inflammatory process.86) A high index of suspicion is warranted in high-risk populations that include patients with CKD, diabetes mellitus, hypertension and the elderly.84) The diagnosis should be suspected in warfarin-treated patients with unexplained hematuria in whom other causes of AKI have been ruled out. Treatment is supportive and is focused on controlling the INR, preventing hypertension and discontinuing concomitant treatment with anti-aggregants.87)

There is evidence that NOACs may be more nephron-protective. A post hoc analysis of the RE-LY cohort showed a more accelerated decline in renal function in patients treated with warfarin compared to both dosages of dabigatran.88) Importantly, poor control of INR (time in therapeutic range <65%) and diabetes were both associated with faster decline in renal function. A post hoc analysis of the ARISTOTLE trial found a similar proportion of decline in renal function in patients on warfarin and apixaban.89) However, apixaban was superior to warfarin in both safety and efficacy regardless of the changes in renal function over time. A nationwide observational study form Taiwan found that all NOACs were associated with a lower risk of acute kidney injury compared to warfarin.90) An analysis of a large U.S. administrative database suggested that NOACs, particularly dabigatran and rivaroxaban, may be associated with lower risks of adverse renal outcomes than warfarin.91)

Like other markers of deconditioning and frailty, history of falls has been shown to independently predict a higher risk for all-cause mortality and thromboembolism.92) Robust data regarding the management of patients at high risk of falls that require OAC have demonstrated that the benefit of stroke prevention generally outweighs the possible increased risk of intracranial bleeding.93), 94), 95), 96) Therefore, current ESC guidelines recommend OAC for this patient population with a few exceptions (e.g. epilepsy or advanced multisystem atrophy with backwards falls).50) These recommendations are based mainly on experiences with warfarin use. A recent subgroup analysis of the ENGAGE AF-TIMI 48 trial showed consistent reductions in all-cause mortality and severe bleeding with edoxaban among patients at risk of falling.97) Another analysis of the ARISTOTLE population showed that the superiority of apixaban with respect to safety and efficacy was consistent with or without a history of falls.98) There are no similar data describing the net benefit of dabigatran or rivaroxaban in patients at high risk of falls.

Polypharmacy, the concomitant use of multiple different drugs, is common in AF patients, and particularly in the elderly.9) Estimations vary widely depending on the population studied and the definition used. A large systematic review comprising of 10,455 individuals aged ≥65 years found that about 1 in 3 patients were prescribed ≥6 drugs.99) Rates of polypharmacy among AF patients may be even higher, ranging from 40% to 95%.100), 101) Polypharmacy is associated with an increased risk of mortality, adverse drug reactions, longer hospitalizations, higher re-hospitalization rates and higher risk for both thromboembolism and major bleeding.102), 103), 104), 105)

The extent of possible drug interactions with warfarin is well-established, as is the mechanism of those interactions. Warfarin is metabolized by the cytochrome P450 system with Cytochrome P450 2C9 isozyme (CYP2C9) as the most dominant component. Potent inhibitors of this system like amiodarone, gemfibrozil and antifungal agents thereby increase the bioavailability of warfarin and the risk of bleeding.106) Antibiotics that reduce vitamin K production like quinolones and macrolides can also increase the risk of bleeding. Nonselective nonsteroidal anti-inflammatory drugs also potentiate the bleeding effects of OAC via the indirect inhibition of platelet activation.107)

All 4 NOACs are excreted by the P-glycoprotein (P-gp) transporter. Several common cardiac drugs such as verapamil, dronedarone, amiodarone, and quinidine inhibit P-gp activity resulting in a significant increase in the plasma levels of all NOACs.82) Cytochrome P450 3A4 has a more minor role in the elimination of apixaban and rivaroxaban, therefore, drugs that inhibit its function will increase drugs levels to a smaller degree. A recent analysis of 91,330 NOAC users from Taiwan, showed a significant increase in major bleeding when amiodarone, fluconazole, rifampin or phenytoin were concomitantly used with a NOAC.108) The net benefit of apixaban and rivaroxaban over warfarin in patients treated with ≥5 drugs was specifically tested in post hoc analyses of the ARISTOTLE and ROCKET AF trials, respectively.105), 109) Apixaban showed a consistent reduction in the risk stroke or systemic embolism (SSE) regardless of number of concomitant drugs used, but the reduction in bleeding risk diminished with the increase in the number of concomitant drugs.105) The extent of polypharmacy or use of ≥1 combined cytochrome P450 3A4 and P-gp inhibitors (like verapamil, diltiazem or amiodarone) did not affect the primary efficacy or safety outcomes of rivaroxaban in comparison to warfarin.109)

The optimal use of OAC for stroke prevention in older patients with AF and comorbidities including kidney disease, frailty, recurrent falls, and polypharmacy remains a challenge for clinicians. Consistent with the RCT evidence in the general AF population, the NOACs appear to preserve their overall superior safety profile over warfarin in this population. However, currently available risk scores may be insufficient to characterize the attendant stroke and bleeding risks in this patient population. Therefore, individualized and thoughtful shared decision making that acknowledges current knowledge gaps is crucial.