AURORA: Is There a Role for Statin Therapy in Dialysis Patients?

What Does This Important Study Show? 

AURORA was a randomized double-blind trial that enrolled 2,776 patients aged 50-80 years who had been treated with maintenance hemodialysis for at least 3 months (median duration, 3.5 years).7 Participants were randomly assigned to treatment with rosuvastatin, 10 mg/d, versus placebo. The primary end point was time to a major CV event (CV death, nonfatal myocardial infarction [MI], or nonfatal stroke), and an appropriate selection of secondary end points also was considered, including changes in serum lipid and CRP levels. The authors predicted a 19.5% decrease in major CV events with active treatment; for α = 0.05, it was estimated that 805 major CV events would yield 87% power to show a significant difference between treatment groups.

Compared with placebo after 3 months of study drug administration, rosuvastatin treatment led to a net decrease in LDL-C levels of 41%, a net decrease in total cholesterol levels of 27%, and a net increase in high-density lipoprotein cholesterol levels of 2%. Rosuvastatin recipients experienced a 14% decrease in CRP levels at 3 months compared with a 4% increase in placebo recipients. After a mean follow-up of 3.2 years, the primary end point occurred in 408 placebo recipients and 396 rosuvastatin recipients (hazard ratio, 0.96; 95% confidence interval [CI], 0.84-1.11; P = 0.59) using an intent-to-treat analysis; results were similar when a per-protocol analysis was performed. Rosuvastatin did not significantly reduce the risk of any individual component of the composite primary end point or any secondary end point. Moreover, rosuvastatin treatment did not significantly reduce the incidence of the primary end point in any prespecified subgroup. No patients were lost to follow-up; however, ∼50% of patients in each treatment arm discontinued treatment before study completion (because of an adverse event, kidney transplant, or other reasons). The risk of adverse events was similar between treatment groups (including muscle symptoms and increase in creatine kinase and alanine aminotransferase levels).

How Does This Study Compare With Prior Studies? 

Observational studies of dialysis patients suggested that statin therapy is associated with decreased mortality. In a cohort of 3,716 patients starting hemodialysis or peritoneal dialysis therapy, statin therapy was associated with decreases in total mortality of 32% and CV-specific mortality of 37%.8 Similar findings were observed in an observational study of 7,365 prevalent hemodialysis patients enrolled in DOPPS (Dialysis Outcome and Practice Patterns Study).9 However, observational studies are prone to bias caused by confounding by indication and other unmeasured characteristics.

4D (Die Deutsche Diabetes Dialyse Studie) was the first adequately powered randomized controlled trial to assess whether statins prevent CV events in dialysis patients.10 This double-blinded study enrolled 1,255 patients in Germany with type 2 diabetes receiving maintenance hemodialysis and compared atorvastatin, 20 mg/d, with placebo for the composite outcome of death from cardiac causes, nonfatal MI, and stroke. Despite a significant decrease in LDL-C levels, atorvastatin did not significantly reduce this primary composite outcome (relative risk, 0.92; 95% CI, 0.77-1.10; P = 0.37). The secondary end point of all cardiac events was significantly decreased by 18% (relative risk, 0.82; 95% CI, 0.68-0.99; P = 0.03), but statin treatment did not decrease the risk of the other 2 secondary outcomes (P = 0.49 for all cerebrovascular events and P = 0.33 for all-cause death).

Treatment of hemodialysis patients with rosuvastatin in AURORA and with atorvastatin in 4D decreased LDL-C levels by 43% (mean decrease, 1.1 mmol/L) and 42% (median decrease, 1.3 mmol/L), respectively. Post hoc analysis of 4D showed that atorvastatin prevented an increase in CRP levels over time that was observed in placebo recipients, but did not reduce the risk of the primary end point in any quartile of baseline CRP level.11 Rosuvastatin treatment in AURORA decreased CRP levels by 11.5% (vs an increase in CRP levels over time in placebo recipients); however, similar to 4D, there was no significant interaction between baseline CRP level and the clinical benefit of statin treatment. Statin use did not prevent the composite primary outcome of combined CV death, MI, and stroke in either study despite these apparently beneficial effects on LDL-C and CRP levels and high CV event rates during the course of the study. These findings substantially differ from those of statin trials in people without kidney failure: a meta-analysis of 14 such trials found that each 1-mmol/L decrease in LDL-C level led to a 21% decrease in risk of major CV event.12 Similarly, data from patients without kidney failure suggest that the CRP level decrease with statin therapy may further decrease mortality independently of LDL-C level decrease.3, 6

What Should Clinicians and Researchers Do? 

Why did 4D and AURORA show no benefit in treating hemodialysis patients with statins? This question deserves consideration before deciding how the findings of these trials should affect clinical practice.13

First, there were significant drop-outs and/or drop-ins between treatment arms in both studies. In 4D, after 2 years, 17% of statin-assigned patients discontinued their drug therapy and 15% of placebo-assigned patients ended up using a nonstudy statin. In AURORA, 50% of patients dropped out of each treatment group, and the proportion of patients who received a nonstudy statin was not reported. By the end of both studies, LDL-C level differences between groups were only ∼0.78 mmol/L (4D) and ∼0.5 mmol/L (AURORA), and this analysis overestimates the true difference between groups because patients who had dropped out were not included.

Second, AURORA excluded patients who clinicians believe would benefit the most from statins (ie, patients who had received a statin within the previous 6 months). Although 4D did not exclude patients receiving statins at baseline (nonstudy statin therapy was discontinued upon enrollment), less than one-third of study participants had coronary artery disease and patients with increased LDL-C levels (>4.9 mmol/L) were excluded. In retrospect, one could speculate that these criteria selected a population of patients who were less likely to benefit from statin therapy.

Finally, although they had adequate statistical power to detect 20%-27% decreases in the relative risk of their primary outcomes, these trials may have been too small to detect a less dramatic benefit that remains clinically important. This hypothesis is supported because 6 of 9 (67%) and 6 of 8 (75%) of the major primary and secondary CV end points in 4D and AURORA, respectively, nonsignificantly favored statin treatment, rather than the ∼50% split that would be expected if statins had no effect on clinical outcomes. In the general population, statins prevent CV events that often occur in association with ruptured atherosclerotic plaque, such as acute MI. Given that many CV deaths in hemodialysis patients are from different causes, such as sudden death (perhaps from electrolyte abnormalities)14, 15 or cardiomyopathy (potentially from chronic extracellular volume overload),16 the effective statistical power of AURORA may have been lower than the number of CV events suggests. Findings of the CORONA (Controlled Rosuvastatin Multinational Trial in Heart Failure; which randomly assigned elderly patients with systolic dysfunction, but without kidney failure, to treatment with rosuvastatin, 10 mg, vs placebo) were broadly similar to those of AURORA: no significant effect on risk of death from CV cause, nonfatal MI, or stroke (hazard ratio, 0.92; 95% CI, 0.83-1.03; P = 0.12) despite an LDL-C level decrease of 45%.17 Statin treatment reduced the risk of coronary events in CORONA participants; however, as in AURORA, these events accounted for the minority of primary outcome and all-cause deaths (10% of CORONA participants had a coronary event and 2% of deaths were due to MI).

These 3 plausible considerations may account for the discrepant findings of AURORA and 4D compared with the undisputable benefits of statins in the general population. However, because these 2 trials failed to show a benefit of statin treatment in hemodialysis patients, the fourth explanation, that statins are truly ineffective when used in patients who are already dialysis dependent, must be assumed to be correct until proved otherwise by future randomized trials.

Although this makes it difficult to justify new prescriptions of statins to patients already receiving hemodialysis, the findings of AURORA and 4D do not directly address the question of whether statin treatment should be discontinued when patients become dialysis dependent. This decision currently must be made after considering the patient's risk of an atherosclerotic (plaque rupture) event in the context of his or her life expectancy. For example, hemodialysis patients with heart failure, those with multiple noncardiac comorbidities, and those at high risk of infection and sepsis would be expected to derive little benefit from statin treatment given that they are unlikely to die of coronary disease regardless of their medical regimen. Unfortunately, the latter group constitutes most patients in contemporary nephrology practice.

SHARP (Study for Heart and Renal Protection) is expected to report in 2010 on the clinical benefits of combination treatment with simvastatin/ezetimibe versus placebo in a large population of patients with advanced kidney disease, and its findings are keenly anticipated. Given the frequency of hypercholesterolemia in peritoneal dialysis patients, the effect of statin monotherapy on CV events in this population also appears worthy of investigation. In the interim, because statin treatment appears beneficial in patients with milder kidney disease, clinicians should focus on identifying and treating such patients with statins and other therapies that reduce CV risk before kidney failure occurs.