Evidence-Based Cardiology in Hemodialysis Patients

Abstract

Cardiac events are the major cause of death in hemodialysis patients. Because of the paucity of randomized clinical trials (RCTs) in hemodialysis patients, most cardiovascular therapies in this population are based on observational studies or results extrapolated from studies that excluded hemodialysis patients. However, associations discovered in observational studies do not prove causality, and these studies often report surrogate outcomes rather than clinical end points. Furthermore, interventions that show effectiveness in the general population may have drastically different outcomes and side effect profiles in hemodialysis patients. This review discusses the results of RCTs undertaken recently to evaluate cardiovascular therapies in hemodialysis patients and emphasizes clinically relevant outcomes. Although some interventions have produced similar outcomes in hemodialysis patients and the general population, others have not, suggesting that the management of cardiovascular disease in hemodialysis patients may require strategies that differ from the best practice guidelines applied to general population.

Survival on hemodialysis is dismal. In a large prospective United States–based study, the 5-year survival was only 40%, with 40% of deaths being cardiac.¹ The annual cardiovascular mortality in the general population increases exponentially with age, rising from 0.01% in 30- to >1% in 80-year-old adults.² In contrast, cardiovascular mortality is extraordinarily high, even in young hemodialysis patients, with only a modest age-related increase. Remarkably, the annual cardiovascular mortality in a 40-year-old hemodialysis patient is comparable with the mortality observed in an 80-year-old adult in the general population. Thus, reducing the high death rate in dialysis patients will require concerted efforts to decrease cardiovascular mortality.

Approximately 80% of hemodialysis patients have one or more types of cardiac disease.³ These diseases include coronary artery disease (22%–39%), congestive heart failure (20%–40%), atrial fibrillation (11%–27%), valvular heart disease (24%), and left ventricular hypertrophy (29%–75%).^3–12 There are numerous high-quality randomized clinical trials (RCTs) supporting therapeutic cardiovascular interventions in the general population.¹³ Unfortunately, the vast majority of these studies specifically excluded patients with kidney disease. In contrast, very few RCTs have evaluated cardiovascular therapy in hemodialysis patients. Cardiologists are well versed in proven therapies for heart disease in the general population (Table 1) and assume that clinical management that has been proven in 98% of cardiac patients with normal kidney function can be extrapolated to 2% of patients on hemodialysis (Figure 1).

Table 1.

Management of heart disease in the general medical population

Type of Heart Disease
AMI
Thrombolytics
Aspirin
Heparin
β-Blockers
Coronary artery disease
β-Blockers
Aspirin/clopidogrel
Statinsa
PCIs
CABG
CHF
Angiotensin-converting enzyme inhibitors
Angiotensin receptor blockers
β-Blockersa
Aldosterone antagonists
Defibrillators
Atrial fibrillation
Anticoagulation

^aRCTs published for this therapy in hemodialysis patients.

Figure 1.

Hemodialysis patients represent a tiny subset of all patients with heart disease. Overlap between heart disease and hemodialysis (HD). In the United States, there are about 27 million patients with cardiac disease and about 400,000 HD patients. Only ∼2% of the patients with heart disease are on HD, whereas ∼80% of HD patients have cardiac disease. Most of the cardiovascular therapies used in HD patients are extrapolated from RCTs that specifically excluded patients with kidney disease.

Why might extrapolating benefits of cardiovascular therapy from RCTs in the general population to hemodialysis patients be inappropriate? First, cardiovascular disease characteristics may differ between the two populations. Second, the response to therapy may differ in hemodialysis patients. Third, the side effect profiles may be different.

Given the paucity of RCTs to evaluate many cardiovascular therapies in hemodialysis patients, nephrologists are frequently forced to rely on observational studies. This approach is problematic for several reasons. First, association does not prove causality. Second, observational studies often report surrogate outcomes rather than hard clinical end points. Third, just because a treatment makes physiologic sense does not mean that it is effective. Although RCTs are regarded as the gold standard for evaluating the efficacy of medical therapy, they typically study a select population, and the results may not generalize to some patients.¹⁴ Moreover, RCTs are not always feasible.

A commentary published in JASN 10 years ago focused on opinion-based management of coronary heart disease in dialysis patients.¹⁵ Fortunately, in the interim, several well designed RCTs have evaluated cardiovascular therapies in this patient population. The primary focus of the current review is to examine the evidence arising from these publications, with an emphasis on clinically relevant cardiovascular outcomes rather than surrogate end points.

Acute Myocardial Infarction and Coronary Interventions

There is abundant evidence from RCTs regarding optimal management of acute myocardial infarction (AMI) in the general population, including administration of thrombolytic agents, aspirin, heparin, β-blockers, percutaneous coronary intervention (PCI), and coronary artery bypass graft surgery (CABG).¹⁶ As noted above, most of these studies excluded patients with kidney disease, and there are no RCTs evaluating the efficacy of such interventions in hemodialysis patients. Observational studies have highlighted important differences between hemodialysis patients and adults with normal kidney function who are hospitalized with AMI (Table 2).¹⁷ The diagnosis is frequently delayed in hemodialysis patients. Dialysis patients are one half as likely as adults in the general population to have ST elevation on the admission electrocardiograms and two times as likely to present with pulmonary edema, which may be mistaken for volume overload. As a consequence, AMI is suspected at the time of admission in only 21.8% of hemodialysis patient compared with 43.8% of patients with normal kidney function.¹⁷ The delay in making the correct diagnosis is accompanied by a lower rate of coronary interventions, including PCI and CABG. Perhaps as a consequence of the delay in diagnosis and the lower rate of coronary and medical interventions, the clinical outcomes of patients with an AMI are inferior in dialysis patients. For example, hemodialysis patients are two times as likely to experience in-hospital sudden cardiac arrest and two times as likely to die during the hospitalization. However, in the absence of RCTs in hemodialysis patients, we do not know whether these interventions would improve outcomes.

Table 2.

Features of AMI in hemodialysis patients versus the general population

Clinical Observations	Dialysis Patients (%)	Nondialysis Patients (%)	P Value
Clinical features on admission
Pulmonary edema	15.2	7.6	<0.001
ST elevation on EKG	19.1	35.9	<0.001
Admission dx of AMI	21.8	43.8	<0.001
Admission dx as other	44.8	21.1	<0.001
Interventions
PCI	8.2	18.6	<0.001
CABG	4.2	9.1	<0.001
In-hospital complications
Sudden cardiac arrest	11.0	5.0	<0.001
Death	21.3	11.7	<0.001

Modified from ref¹⁷, with permission. EKG, electrocardiogram; dx, diagnosis.

Even when dialysis patients undergo successful PCI, their long-term survival is far worse than the survival in patients with normal kidney function undergoing PCI.¹⁸ The 3-year survival after successful PCI is progressively worse with declining kidney function, being ∼90% in patients with a creatinine clearance >70 ml/min but only 50% in hemodialysis patients. Drug-eluting coronary stents may be superior to bare-metal stents in managing coronary artery disease in the general population, markedly reducing the frequency of restenosis,^19,20 although some studies failed to show a difference.²¹ The benefit of drug-eluting stents on clinical end points is less clear cut,²⁰ and one study even reported inferior outcomes of drug-eluting stents versus bare-metal stents in diabetic patients.²¹ There are no RCTs comparing the two types of coronary stents in hemodialysis patients. However, observational studies found no significant differences between drug-eluting stents and bare-metal stents in frequency of revascularization, recurrent AMI, or death.²²

There are no RCTs comparing the outcomes of CABG and PCI in hemodialysis patients. However, a large observational study reported on the outcomes in nearly 22,000 hemodialysis patients undergoing this procedure.²³ The relative hazard ratio for death was 12% higher for the first 6 postoperative months, likely reflecting immediate surgical complications with CABG. However, there was a 12% lower risk of death after the first 6 months with CABG, indicating potential benefit.

Medical Management of Coronary Artery Disease

The Framingham study highlighted several traditional risk factors for coronary artery disease in the general population, including age, men, hypertension, diabetes, hyperlipidemia, and smoking (Table 3).²⁴ These clinical factors are much less predictive of coronary artery disease in hemodialysis patients. As previously described, the effect of age on cardiovascular deaths is much more modest in hemodialysis patients than the general population. Likewise, sex does not seem to affect the frequency of cardiovascular events in hemodialysis patients.² The association between elevated LDL cholesterol and AMI is attenuated in patients with advanced CKD.²⁵ The frequency of coronary artery disease is constant (30%–40%) for hemodialysis patients with predialysis systolic BPs ranging between 100 and 200 mmHg and patients with serum cholesterol values ranging between 100 and 300 mg/dl.³ There is considerable controversy about the role of BP control on cardiovascular outcomes in hemodialysis patients because of the extreme variability of BP readings during dialysis, the poor correlation between ambulatory and dialysis BPs, the contradictory results of observational studies, and the dearth of high-quality interventional trials.²⁶

Table 3.

Traditional and nontraditional risk factors for cardiac disease in hemodialysis patients

Category of Risk Factor
Traditional risk factors
Age
Men
Diabetes
Hypertension
Smoking
Hyperlipidemia
Nontraditional risk factors
Uremia
Anemia
Hyperphosphatemia
Hyperparathyroidism
Coronary calcification
Hyperhomocysteinemia
Inflammation
Oxidative stress
LVH

Hemoglobin A1c is widely used to assess long-term glycemic control in the general population, but it underestimates average blood glucose in diabetic hemodialysis patients caused by the shortened red blood cell survival.²⁷ Glycated albumin is a more accurate measure of glycemic control in diabetic hemodialysis patients, and higher levels have been associated with both mortality and cardiovascular hospitalizations.^28,29 These findings suggest that improving glycemic control may lower deaths and cardiovascular events in diabetic hemodialysis patients, but this hypothesis has not been evaluated in clinical trials. Finally, observational studies have documented a higher cardiovascular mortality and hospitalization rate in hemodialysis patients who were current smokers than patients who never smoked, with intermediate rates seen in former smokers.³⁰

Statins consistently reduce cardiovascular events in the general population in both primary and secondary prevention.³¹ The frequency of cardiac events is far greater in dialysis patients than the general population, which might translate into an enormous benefit of statins.² Three RCTs have evaluated the effect of statins on cardiovascular events in hemodialysis patients (Table 4). The German Diabetes and Dialysis Study (4D) study, a double-blinded German RCT, allocated 1255 diabetic hemodialysis patients to atorvastatin or placebo.¹¹ The primary study outcome (cardiac death, nonfatal AMI, or cerebral vascular accident [CVA]) occurred in ∼50% of the control patients at 5 years. Cardiovascular events were not significantly reduced in the atorvastatin arm, despite achieving significantly lower LDL cholesterol and serum C-reactive protein levels.

Table 4.

RCTs with clinical cardiovascular end points in hemodialysis patients

Study (yr)	Intervention	N of Patients	Primary End Point	HR (95% CI) for Primary End Point
Statins
4D (2005)11	Statins versus placebo	1255	Cardiac death, nonfatal MI, or CVA	0.92 (0.77 to 1.10)
AURORA (2009)32	Statins versus placebo	2776	Cardiac death, nonfatal MI, or CVA	0.96 (0.84 to 1.11)
SHARP (2011)33	Statins versus placebo	3191	Coronary death, MI, ischemic CVA, or any arterial revascularization	0.95 (0.78 to 1.15)
Dialysis dose or flux
HEMO (2002)1	Higher versus standard dialysis dose (Kt/V 1.65 versus 1.25)	1846	All-cause mortality	0.96 (0.84 to 1.10)
HEMO (2002)1	High- versus low-flux dialyzers	1846	All-cause mortality	0.92 (0.81 to 1.05)
MPO (2009)39	High- versus low-flux dialyzers	647	All-cause mortality	N/A (P=0.21)
CONTRAST (2012)43	Online hemodiafiltration versus conventional HD	714	All-cause mortality	0.95 (0.75 to 1.20)
ESHOL (2012)41	Online hemodiafiltration versus conventional HD	906	All-cause mortality	0.70 (0.53 to 0.92)
Turkish OL-HDF Study (2013)42	Online hemodiafiltration versus conventional HD	782	All-cause mortality, nonfatal MI, CVA, coronary revascularization, or unstable angina	0.83 (0.59 to 1.16)
Anemia correction
Normal Hematocrit Trial (1998)45	Target hemoglobin 14 versus 10 g/dl	1233	Death or MI	1.3 (0.9 to 1.9)
Mineral metabolism
RIND (2007)52	Calcium-containing Pi binders versus sevelamer	127	All-cause mortality	3.1 (1.23 to 7.61)
DCOR (2007)55	Sevelamer versus calcium-containing phosphate binders	2103	All-cause mortality	0.93 (0.79 to 1.10)
EVOLVE (2012)12	Cinacalcet versus placebo	3883	Death, MI, unstable angina, CHF, or PVD event	0.93 (0.85 to 1.02)
Homocysteine reduction
Wrone et al. (2004)57	High-dose folic acid	510	Death, coronary intervention, MI, CVA, TIA, carotid endarterectomy, or amputation	N/A (P=0.47)
HOST (2007)58	High-dose folic acid, pyridoxine, and vitamin B12	751	All-cause mortality	1.04 (0.83 to 1.28)
Antioxidants
SPACE (2000)62	Vitamin E versus placebo	196	Acute MI, CVA, PVD, or unstable angina	0.46 (0.27 to 0.78)
Tepel et al. (2003)63	Acetylcysteine versus placebo	134	AMI, CV death, PCI, CABG, CVA, or PVD	0.60 (0.38 to 0.95)
β-Blockers for CHF
Cice et al. (2003)71	Carvediolol versus placebo	114	All-cause mortality	0.51 (0.32 to 0.82)
ACE inhibitor
FOSIDIAL (2006)90	Fosinopril versus placebo	397	CV death, cardiac arrest, CVA, CHF, MI, or revascularization	0.93 (0.59 to 1.1)

Included studies are only RCTs enrolling >100 subjects. HR, hazard ratio; 95% CI, 95% confidence interval; MI, myocardial infarction; MPO, Membrane Permeability Outcome Study; N/A, not available; CONTRAST, Convective Transport Study; HD, hemodialysis; ESHOL, Estudio de Supervivencia de Hemodiafiltracion On-Line; OL-HDF, on-line hemodiafiltration; RIND, Renagel in New Dialysis Study; Pi, phosphate; DCOR, Dialysis Clinical Outcomes Revisited; EVOLVE, Evaluation of Cinacalcet Hydrochloride Therapy to Lower Cardiovascular Events; PVD, peripheral vascular disease; TIA, transient ischemic attack; HOST, Homocysteinemia in Kidney and End Stage Renal Disease; SPACE, Secondary Prevention with Antioxidants of Cardiovascular Disease in Endstage Renal Disease; CV, cardiovascular; ACE, angiotensin-converting enzyme; FOSIDIAL, Fosinopril in Dialysis.

A Study to Evaluate the Use of Rosuvastatin in Subjects on Regular Hemodialysis: An Assessment of Survival and Cardiovascular Events (AURORA) study, an international double-blinded RCT, assigned 2276 hemodialysis patients ages 50–80 years to rosuvastatin or placebo.³² The primary outcome (cardiac death, nonfatal AMI, or CVA) occurred in 35% of the placebo group at 5 years of follow-up. It was not significantly reduced in the statin group, despite having lower LDL cholesterol and C-reactive protein levels.

Finally, the Study of Heart and Renal Protection (SHARP) study, an international double-blinded RCT, enrolled 9438 patients with CKD, including 3191 hemodialysis patients.³³ The patients were allocated to receive simvastatin and ezitimibe or placebo. The primary outcome (coronary death, nonfatal AMI, ischemic CVA, or coronary revascularization) was reduced by 17% in the statin arm when all study patients were included. However, there was a marked discrepancy according to dialysis status. Thus, statins reduced the primary end point by 22% in the CKD patients but only by a nonsignificant 5% in the hemodialysis patients.

Taken together, these three RCTs do not support the use of statins to prevent cardiovascular events in hemodialysis patients. Why might this result be so? First, coronary disease may be so advanced in this population that statins are futile, with no significant benefit of lipid lowering. Second, only 21% of deaths in the 4D trial were attributed to an AMI.¹¹ If the primary benefit of statins is to prevent ischemic cardiac events and they reduce this cardiac end point by ∼25% (which was observed in the CKD patients enrolled in the SHARP study), they would be expected to reduce overall mortality by only ∼5% (25% of 21%). Indeed, this result is approximately the magnitude of mortality reduction observed in the 4D, AURORA, and SHARP studies for hemodialysis patients treated with statins.

Nontraditional Cardiac Risk Factors and Cardiovascular Mortality

Although observational studies have indicated potential benefit of interventions targeting nontraditional cardiac risk factors in hemodialysis patients, RCTs have not always provided support for these interventions (Tables 3 and 4). Observational studies reported an inverse association between dialysis dose and patient mortality.^34–36 Of course, this association does not prove causality, because a lower Kt/V may simply be a surrogate marker for other factors that are deleterious to patient survival, such as noncompliance with the dialysis prescription or use of hemodialysis catheters. The Hemodialysis (HEMO) Study randomized 1846 patients to a standard or high dose of dialysis (Kt/V 1.65 versus 1.25).¹ Outcomes among patients in the two treatment arms were similar in death, a composite end point of death or cardiovascular hospitalization, and death caused by ischemic heart disease, heart failure, or arrhythmias. In a secondary analysis, the higher dose of dialysis was associated with lower mortality only among women.

Likewise, observational studies have revealed an association between high-flux dialyzers and lower mortality.^37,38 Two RCTs evaluated the benefit of high-flux dialysis on cardiovascular outcomes. In the HEMO Study, all-cause mortality did not differ between patients randomized to high- versus low-flux dialyzers.¹ However, a secondary analysis suggested a survival benefit for patients who had been on dialysis for >3.7 years at the time of enrollment. The Membrane Permeability Outcome study randomized 738 incident European hemodialysis patients to dialyze with a high- or low-flux dialyzer.³⁹ There was no significant difference in mortality between the two treatment arms. However, in a subgroup analysis, there was a survival benefit of high-flux dialysis in the patients whose baseline serum albumin was ≤4 g/dl.

Online hemodiafiltration (HDF) is a modification of standard hemodialysis that increases middle-molecule clearance by adding convection to diffusion.⁴⁰ Three RCTs have compared survival and cardiovascular events in patients allocated to HDF versus conventional hemodialysis. One study observed better outcomes with HDF,⁴¹ whereas the other two studies found no difference.^42,43

There is an inverse association between hemoglobin levels and mortality in hemodialysis patients in observational studies.⁴⁴ Anemia correction may improve myocardial oxygenation, thereby reducing mortality, particularly in hemodialysis patients with known heart disease. However, anemia may simply be a marker of high comorbidity and erythropoietin resistance. The Normal Hematocrit Study evaluated whether correction of anemia reduces cardiovascular outcomes in hemodialysis patients with known coronary artery disease or congestive heart failure.⁴⁵ The study randomized 1233 patients to target the hemoglobin to 14 versus 10 g/dl. The primary end point (death or AMI) occurred more frequently in the high-hemoglobin arm, although the difference failed to achieve statistical significance (P=0.07) when the trial was stopped at the third interim analysis. Subsequent data analysis incorporating additional patient follow-up showed a statistically significant higher likelihood of death or AMI in this group (hazard ratio, 1.28; 95% confidence interval, 1.06 to 1.56; P=0.01).⁴⁶ These findings are consistent with three other RCTs in patients with CKD; each RCT showed comparable or worse cardiovascular outcomes in patients randomized to a higher hemoglobin target.^47–49

Several disorders of mineral metabolism (hyperphosphatemia, hypercalcemia, and hyperparathyroidism) have been associated with mortality in hemodialysis patients.⁵⁰ These findings have prompted guidelines advocating phosphate binders to lower serum phosphate and cinacalcet to treat secondary hyperparathyroidism.⁵¹ Vascular calcification, including coronary artery calcification (CAC), is a common finding in hemodialysis patients, and it has been associated with increased mortality.⁵² CAC has been attributed, in part, to the use of calcium-containing phosphate binders. The use of sevelamer (a noncalcium-based phosphate binder) slowed the progression of CAC in hemodialysis patients compared with patients who were treated with calcium-containing binders in one RCT,⁵³ but it had no significant effect in a second trial.⁵⁴ A small RCT (n=127) reported a survival advantage in hemodialysis patients treated with sevelamer.⁵² However, a much larger RCT (2103 patients) showed no difference in mortality between patients allocated to sevelamer and patients receiving calcium-containing phosphate binders.⁵⁵ Finally, a recent RCT allocated 3883 hemodialysis patients to cinacalcet or placebo.¹² Despite lowering parathyroid hormone, cinacalcet did not reduce the primary cardiovascular end point.

In observational studies, homocysteine levels have been associated with mortality in hemodialysis patients.⁵⁶ Two RCTs lowered plasma homocysteine levels and evaluated cardiovascular outcomes. The study by Wrone et al.⁵⁷ allocated 510 dialysis patients (of whom 458 patients were on hemodialysis) to receive 1, 5, or 15 mg folic acid daily. There was a dose-dependent decrease in plasma homocysteine but no difference in the frequency of mortality or cardiovascular outcomes among the three treatment arms. The Homocysteinemia in Kidney and End Stage Renal Disease study enrolled 2056 patients with CKD, of whom 751 patients were on hemodialysis. Patients were assigned to receive a combination of high-dose folic acid (40 mg daily), pyridoxine (100 mg daily), and cyanocobalamin (2 mg daily) or placebo.⁵⁸ The randomization was stratified by dialysis status. The active drug lowered plasma homocysteine by 26% but did not lower all-cause mortality. The lack of benefit of lowering homocysteine on cardiovascular outcomes in hemodialysis patients is consistent with similar findings in the general medical population.⁵⁹

Inflammation and oxidative stress are common in hemodialysis patients and have been associated with mortality. Antioxidants may improve endothelial dysfunction and thereby, reduce cardiovascular events in hemodialysis patients.^60,61 Two small RCTs tested this hypothesis. An Israeli study allocated 196 patients to vitamin E or placebo and found a 54% significant reduction in cardiovascular events with vitamin E.⁶² Likewise, a German study assigned 134 patients to acetylcysteine or placebo and observed a significant 40% reduction in cardiovascular events in the active drug arm.⁶³ A larger RCT is indicated before recommending widespread adoption of antioxidant therapy in hemodialysis patients.

Congestive Heart Failure

RCTs performed in congestive heart failure (CHF) patients with normal kidney function have shown significant benefit of angiotensin-converting enzyme inhibitors,⁶⁴ angiotensin receptor blockers, β-blockers,^65–67 aldosterone antagonists,⁶⁸ and implantable defibrillators.⁶⁹ Of these potential therapies, only β-blockers have been evaluated in hemodialysis patients. A small, single-center, double-blinded Italian RCT allocated 114 hemodialysis patients with systolic heart failure to carvedilol or placebo. The mean left ventricular ejection fraction increased from 26% to 35% in the carvedilol arm but remained unchanged in the placebo group.⁷⁰ Patient survival at 2 years was significantly higher in the carvedilol group (48% versus 27%; P<0.005).⁷¹

Implantable defibrillators prevent sudden cardiac death in CHF patients with normal kidney function,⁶⁹ but their efficacy has not been evaluated by an RCT in hemodialysis patients. An observational study of hemodialysis patients with a history of sudden cardiac death found similar survival in patients treated with medical therapy alone and patients receiving an implantable defibrillator.⁷²

Atrial Fibrillation

Atrial fibrillation is present in 11%–27% of hemodialysis patients,^3,6–8 which is a much higher rate than the 1% prevalence in the general adult population.⁷³ Atrial fibrillation is associated with stroke in the general population, and the magnitude of risk is increased among patients with older age, diabetes, or hypertension.⁷⁴ Anticoagulation with warfarin substantially reduces CVA (by about two thirds) in general medical patients with atrial fibrillation, particularly in high-risk patients.⁷⁵ Finally, the absolute increased risk of major hemorrhage (∼1% annually) with warfarin is relatively modest.⁷⁶ Thus, the benefit of anticoagulation in high-risk patients with atrial fibrillation outweighs the risk of major hemorrhage. Prescribing warfarin in 1000 patients with atrial fibrillation for 1 year would prevent 23 strokes but cause an additional 9 major hemorrhages.⁷⁶

Can these findings be extrapolated to hemodialysis patients with atrial fibrillation? Certainly, these patients are at high risk for CVA given their high frequency of older age, diabetes, and hypertension. Thus, akin to the previous discussion of coronary artery disease in hemodialysis patients, the potential benefit of anticoagulation seems to be huge. However, observations in the general population may not hold true in hemodialysis patients. First, the association between atrial fibrillation and CVA is less apparent in dialysis patients. In a longitudinal study, the 3-year frequency of CVA was similar in hemodialysis patients with and without atrial fibrillation (15.4% versus 12.4%; P=0.40).⁷⁷ Likewise, in a large cross-sectional study of hemodialysis patients, cerebrovascular disease was not associated with atrial fibrillation.⁸ Moreover, CVAs are more likely to be hemorrhagic in hemodialysis patients than the general population (23% versus 13%).^78,79 Second, it is not clear whether warfarin prevents stroke in this population. Four observational studies reported contradictory findings on the association between warfarin use and CVA in hemodialysis patients with atrial fibrillation. Two studies described a higher risk of ischemic stroke,^8,78 a third study found no difference,⁸⁰ and a fourth study observed a reduction.⁸¹ Third, the risk of major hemorrhage with warfarin is substantially higher in hemodialysis patients than the general population. In two observational studies, the adjusted risk of major hemorrhage was 2.4- to 2.7-fold higher in patients with an estimated GFR <30 ml/min compared with adults with normal kidney function.^82,83

In summary, the benefit of anticoagulation in hemodialysis patients with atrial fibrillation is uncertain, and the risk of major hemorrhage is substantial. As a consequence, the risk of warfarin may outweigh the benefit (Figure 2). Until an RCT of anticoagulation in hemodialysis patients with atrial fibrillation is performed, it may be prudent to avoid such therapy in this patient population.^84,85

Figure 2.

The relative risks and benefits of anticoagulation for atrial fibrillation vary between hemodialysis patients and patients with normal kidney function. Risk benefit of anticoagulation in patients with atrial fibrillation. In patients with normal renal function and atrial fibrillation, the benefit of warfarin in preventing strokes exceeds the additional risk of major hemorrhage. In hemodialysis patients, the benefit of warfarin in prevention of strokes is uncertain, whereas there is a substantial risk of major hemorrhage. Thus, it is not clear that the benefit outweighs the risk in hemodialysis patients.

Sudden Cardiac Death

Sudden cardiac death accounts for ∼60% of all cardiac deaths in hemodialysis patients.¹¹ Sudden cardiac death is associated with underlying left ventricular hypertrophy (LVH),⁸⁶ and patients with LVH that regresses have a lower cardiovascular mortality.⁸⁷ Measures that produce LVH regression may reduce sudden cardiac death and overall mortality in hemodialysis patients. However, regression of LVH is difficult to achieve. Anemia correction did not improve LVH in RCTs of hemodialysis patients.^88,89 Likewise, in the largest RCT assessing the effect of BP control in hemodialysis patients (n=397), fosinopril did not improve LVH or reduce cardiovascular events.⁹⁰ Volume overload, which has been implicated in the pathogenesis of LVH, can be alleviated by longer or more frequent dialysis sessions. A small Canadian RCT (n=52) documented LVH regression with more frequent dialysis.⁹¹ The Frequent Hemodialysis Network randomized 245 hemodialysis patients to frequent (6 d/wk) or conventional (3 d/wk) hemodialysis. Frequent hemodialysis lowered interdialytic weight gain, predialysis BP, and left ventricular mass during 1 year of follow-up.⁹ A large RCT is needed to assess whether frequent hemodialysis prevents sudden cardiac death.

Conclusions

Physicians entrusted with the care of hemodialysis patients who are concerned about high mortality in this population must address cardiovascular risk. Unfortunately, the evidence base informing clinical practice is limited, often contradictory, or counterintuitive. The effect of some interventions (homocysteine lowering, coronary stents, and β-blockers for CHF) is similar in hemodialysis patients and the general population. In contrast, other interventions (particularly statins) have markedly different effects in hemodialysis patients. Although coronary artery disease is extremely common in hemodialysis patients, it is a relatively infrequent cause of cardiac death. There is an urgent need for additional RCTs to evaluate interventions to reduce sudden cardiac death in hemodialysis patients. Until such studies are performed, nephrologists will continue to manage cardiovascular disease according to the best practice guidelines recommended for the general population, while recognizing that these guidelines may be imperfect in hemodialysis patients.

Disclosures

None.