Atorvastatin compared with simvastatin

Thirty trials have compared atorvastatin to simvastatin (Evidence Table 1).^{12, 15, 19, 26, 29, 30, 33, 38, 39, 41, 42, 48, 50–53, 55, 57–59, 65, 68, 72, 73, 83, 84, 86–89} One meta-analysis has compared atorvastatin to simvastatin.¹² Thirteen of the trials included patients with coronary heart disease or high risk of coronary heart disease including coronary heart disease equivalents such as diabetes.^{12, 15, 19, 26, 30, 33, 39, 50, 53, 68, 83, 86, 87} At doses below 80 mg, rates of adverse events and withdrawals due to adverse events were similar in patients taking atorvastatin or simvastatin.

Three studies directly compared atorvastatin 80 mg to simvastatin 80 mg daily.^{52, 56, 58} In the first study, atorvastatin 80 mg reduced low-density lipoprotein cholesterol by 53.6% compared with 48.1% for simvastatin 80 mg (P≤0.001).⁵² Compared with the simvastatin 80 mg groups, a greater number of patients in the atorvastatin 80 mg groups reported clinical adverse effects, primarily gastrointestinal diarrhea (23% compared with 11.9%; P<0.001). There was no significant difference between atorvastatin 80 mg and simvastatin 80 mg in withdrawal rates due to adverse effects. Withdrawal from the study due to adverse laboratory events occurred more often in the atorvastatin 80 mg compared with the simvastatin 80 mg daily group (4% compared with 0.8%; P<0.05). Clinically important alanine aminotransferase elevation (greater than 3 times the upper limit of normal) occurred statistically more often in the atorvastatin 80 mg compared with the simvastatin 80 mg group (17 compared with 2 cases, respectively, P=0.002) and was especially pronounced in women (there were statistically more women randomized to atorvastatin than simvastatin). Aminotransferase elevation generally occurred within 6 to 12 weeks after initiation of the 80 mg statin dose.

In the second study,⁵⁸ Karalis and colleagues randomized 1732 patients with hypercholesterolemia to treatment with atorvastatin 10 mg or 80 mg daily or simvastatin 20 mg or 80 mg daily for 6 weeks. This study was unblinded and did not use intention-to-treat statistics. Mean baseline low-density lipoprotein cholesterol in the atorvastatin group was reduced by 53% compared with 47% in the simvastatin group (P<0.0001). With regard to safety at the 80 mg dosage for each statin, atorvastatin was associated with a higher incidence of adverse effects compared to simvastatin (46% compared with 39%) and a higher rate of study discontinuation due to adverse effects (8% compared with 5%). However, neither of these differences was statistically significant.

The STELLAR trial⁵⁶ was a fair- to poor-quality open-label trial designed to compare rosuvastatin to other statins (atorvastatin, simvastatin, and pravastatin). One hundred sixty-seven patients were randomized to atorvastatin 80 mg and 165 to simvastatin 80 mg. Baseline low-density lipoprotein levels were similar in both groups (190 mg/dL). The mean percent change in low-density lipoprotein level after 6 weeks was 51% in the atorvastatin group and 46% in the simvastatin group, a difference (5.3 percentage points) similar to those found in the 2 other studies comparing atorvastatin 80 mg to simvastatin 80 mg. The proportion of patients who withdrew because of adverse events was 3.6% in both groups.

Atorvastatin compared with rosuvastatin

Twenty-nine trials^{14–17, 19–24, 28, 43, 56, 69, 74–76, 78, 79, 86, 90–96} and 3 meta-analyses^{13, 36, 97} have compared rosuvastatin to atorvastatin (see Table 5, below, and Evidence Table 1).

Table 5

Trials comparing atorvastatin to rosuvastatin.

Nine trials concerned patients who had moderate to no risk factors for coronary artery disease^{14, 43, 56, 69, 74, 75, 91, 92, 98} and 19 trials enrolled patients at high risk for cardiovascular disease.^{15–17, 19–24, 28, 76, 78, 79, 86, 87, 93–96} All studies comparing rosuvastatin to atorvastatin that reported low-density lipoprotein cholesterol reductions at 12 weeks ^{36, 43, 69, 86, 87, 91, 93, 94} had similar results, whether or not they included patients at high risk for coronary heart disease. There were 2 studies that provided low-density lipoprotein cholesterol data at 24 weeks^{20, 98} and revealed consistency with the 12-week trial results. One trial continued for 48 weeks²⁴ and had an effect of 30% reduction in low-density lipoprotein with atorvastatin 20 mg compared with 44.3% reduction with rosuvastatin 10 mg. This effect was significantly different at P<0.001.

Most trial designs included a 6-week run-in period during which dietary counseling was provided. After this run-in period, only patients meeting low-density lipoprotein cholesterol requirements were randomized. Eight trials allowed patients to enter the study without a run-in period.^{19, 22, 24, 28, 75, 86, 91, 94} Fifteen trials reported the number screened. The percentage of patients enrolled after screening ranged from 27.1% to 85.9%.

The Strandberg study included patients with hypertension (73%), diabetes (26.9%), other atherosclerotic disease (28%), or coronary heart disease. On average, rosuvastatin 10 mg reduced low-density lipoprotein cholesterol more than atorvastatin 10 mg (46.9% compared with 38%; P<0.05). There was no comparison of rosuvastatin 10 mg to a higher dose of atorvastatin in this trial. At week 12, the 387 patients who had not reached their low-density lipoprotein cholesterol goal (based on the 1998 Second Joint Task Force of European and Other Societies on Coronary Prevention targets) were switched to rosuvastatin from atorvastatin and had their dosage of rosuvastatin increased until their goal was met (only 12 patients titrated up to the maximum daily dose of 40 mg for rosuvastatin). About 3.5 % of the rosuvastatin group (including those occurring during the 36-week extension period) and 3.0% of the atorvastatin group withdrew due to adverse events.

Schwartz et al also enrolled patients who had diabetes or were at high cardiovascular risk.⁹³ Of 383 patients randomized, 3.7% had diabetes alone, 85.4% had atherosclerosis alone (a history of peripheral vascular disease, coronary artery disease, or cerebrovascular disease), and 11% had both diabetes and atherosclerosis. Although the trial was designed to compare rosuvastatin 80 mg to atorvastatin 80 mg over 24 weeks, results at weeks 12 and 18, before patients were titrated to 80 mg, are also available. Rosuvastatin 5 mg daily (39.8%, P<0.01) had a significant difference in reducing low-density lipoprotein cholesterol levels compared to atorvastatin 10 mg (35%) at 12 weeks. The 18-week analysis in this study compared rosuvastatin 20 mg and rosuvastatin 40 mg to atorvastatin 40 mg. Through 12 weeks, similar proportions of patients taking rosuvastatin and atorvastatin withdrew because of adverse events.

A large head-to-head trial that included higher doses of rosuvastatin was a 6-week open label trial (STELLAR) in which about 300 patients took rosuvastatin 40 mg/day or higher.⁵⁶ Rosuvastatin 40 mg, atorvastatin 80 mg, and simvastatin 80 mg had similar rates of withdrawal and of serious adverse events (pravastatin 80 mg was not included). A post hoc subanalysis of 811 patients in the STELLAR trial with metabolic syndrome had results similar to the overall sample.⁹⁹ In this analysis, the low-density lipoprotein cholesterol reductions for rosuvastatin 40 mg and atorvastatin 80 mg were −55.3% and −48.8%, respectively (P=NS).

Many of the trials comparing atorvastatin and rosuvastatin were open-label and were multisite studies that pooled data, including DISCOVERY,¹⁹ STELLAR,¹⁴ MERCURY II,¹⁵ SUBARU,²² SOLAR,⁸⁷ ECLIPSE,²⁰ and STARSHIP.²³ One trial was single-blinded⁹¹ and 1 study was double-blinded.²⁸ Recent open-label trials of atorvastatin compared with rosuvastatin were conducted in African Americans,⁷⁴ patients with type 2 diabetes,^{78, 95} and patients with established cardiovascular disease.⁷⁶ In African Americans, rosuvastatin 10 mg lowered low-density lipoprotein cholesterol more than atorvastatin 10 mg, but not atorvastatin 20 mg. This is similar to results of other studies. In patients with type 2 diabetes and established cardiovascular disease, the percent low-density lipoprotein cholesterol reduction with rosuvastatin and atorvastatin was similar to that found in other studies, and patients taking rosuvastatin had greater low-density lipoprotein cholesterol reductions.

Primary prevention

AFCAPS (lovastatin), WOSCOPS (pravastatin), and JUPITER (rosuvastatin) trials recruited patients without a history of coronary heart disease (primary prevention).^{81, 126, 132} All 3 trials were rated as good quality. One new trial¹⁴³ was rated poor quality due to multiple methodologic weaknesses.

In WOSCOPS,¹³² pravastatin 40 mg reduced coronary events by 31%, or 1 for every 44 patients (men only) treated (absolute risk, 5.5% compared with 7.9%) whereas in AFCAPS/TexCAPS, lovastatin reduced the incidence of new cardiovascular events by 37%, or 1 for every 49 subjects (men and women) treated (absolute risk, 6.8% compared with 10.9%). WOSCOPS used a stricter definition of coronary events, defined as the occurrence of nonfatal myocardial infarction or coronary heart disease death, than AFCAPS, which included incidence of unstable angina in their primary outcome, so the relative risk reductions and numbers-needed-to-treat were not directly comparable.

In WOSCOPS, but not AFCAPS/TexCAPS, pravastatin therapy reduced coronary disease deaths by 33% (95% CI, 1 to 55) and all-cause mortality by 22% (95% CI, 0 to 40), a result that nearly reached statistical significance (P=0.051). The absolute risks of coronary disease death were 1.3% for subjects in the pravastatin group and 1.9% in the placebo group; number needed to treat, 163. In AFCAPS/TexCAPS, the absolute risks of fatal coronary disease events were 3.3 per 1000 subjects in the lovastatin group and 4.5 per 1000 subjects in the placebo group (P=NS). There was no difference in all-cause mortality in AFCAPS/TexCAPS.

The different mortality results should not be taken as evidence that pravastatin and lovastatin would differ if used in subjects at similar risk. Compared with AFCAPS/TexCAPS, WOSCOPS recruited subjects who had about 4 times as high a risk of dying from coronary disease in the first place. The reduction in coronary heart disease deaths was actually comparable in the 2 studies, however in AFCAPS/TexCAPS, it did not reach statistical significance due to the lower number of events.

In JUPITER,⁸¹ a large multicenter, international trial, 17 802 relatively healthy adults with lipid levels below current treatment thresholds who also had elevated C-reactive protein and who had never used lipid lowering therapy, were randomized to rosuvastatin 20 mg or placebo. The trial was initially designed to continue until 520 primary endpoints were documented but was stopped early for benefit. After a median follow-up of 1.9 years, rosuvastatin 20 mg lowered the risk for the occurrence of a first major cardiovascular event by 44% (hazard ratio, 0.56; 95% CI, 0.46 to 0.69; P<0.00001). The absolute risks observed for rosuvastatin was 1.6% compared with 2.8% (number needed to treat, ~83). All-cause mortality was reduced for rosuvastatin-treated patients (hazard ratio, 0.80; 95% CI, 0.67 to 0.97; P=0.02) but the absolute risk difference was small (2.2% compared with 2.8%; number needed to treat, ~167). Most individual components of the primary endpoint showed favorable findings for rosuvastatin in preventing coronary events, except for deaths from cardiovascular causes since these data were not reported. About 41% of patients enrolled had metabolic syndrome, 16% were smokers, and 12% reported family history of coronary disease.

Compared with WOSCOPS and AFCAPS/TexCAPS, the primary endpoint in the JUPITER trial was broader and included incidence of nonfatal myocardial infarction, nonfatal stroke, hospitalizations for unstable angina, need for revascularization, or death from cardiovascular causes. Total withdrawal rates and withdrawals due to adverse events were not reported, though there were no significant differences in the total number of reported serious adverse events between treatment groups (1352 cases with rosuvastatin compared with 1377 placebo; P=0.60). There were 19 cases of myopathy in 10 rosuvastatin-treated and 9 placebo-treated patients (P=0.82). One fatal case of rhabdomyolysis was recorded in a 90-year old patient (rosuvastatin arm) who had febrile influenza, pneumonia, and trauma-induced myopathy. There were no significant differences between rosuvastatin or placebo for elevations in alanine aminotransferase >3 times the upper limit of normal (0.3% compared with 0.2%; P=0.34) but newly diagnosed diabetes, as reported by physicians, was more frequent with rosuvastatin (3.0% compared with 2.4%; P=0.01). These cases were not verified by the endpoint committee and conclusions based on these findings should be considered with caution until further studies are conducted.

Although the risk reductions were significant for rosuvastatin in preventing major cardiovascular events and deaths, the absolute risk differences between treatment groups were small. It is unknown whether these risk reductions will be maintained over longer periods of time for primary prevention since this trial (JUPITER) was stopped early. Truncated trials such as this pose a difficult challenge in determining whether treatment effects are overestimations of the “true” value. It has been shown that truncated trials stopped early for benefit are more likely to show greater treatment effects than trials that were not stopped early.^{175, 176} Therefore, extrapolating results from this trial beyond about 1.9 years (to 4 or 5 years) is not recommended, as was done by the authors of the trial. Further studies longer in duration will need to be conducted to confirm the findings.

Studies enrolling mixed populations or subjects with coronary risk equivalents

Ten trials extended these results to patient populations who were excluded from the earlier trials (Table 9). In the Heart Protection Study, 20 536 men and women aged 40 to 80 years were randomized to simvastatin 40 mg or placebo for an average of 5.5 years.^{123, 174} This study targeted individuals in whom the risk and benefits of cholesterol lowering were uncertain (women, those over 70 years, those with diabetes, those with non-coronary vascular disease, and those with average or below average cholesterol).

The overall low-density lipoprotein reduction was 30%. This figure resulted from a true intention-to-treat analysis, that is, it included patients who never took simvastatin or who quit taking it by the end of the study. In the subset of patients who took simvastatin for the entire study period, the low-density lipoprotein reduction was 40%. Simvastatin reduced all-cause mortality from 14.7% to 12.9% (a 13% reduction).

Simvastatin also reduced the risk of major coronary events (number needed to treat, 32 after 5 years) and of stroke.¹⁷⁷ In subgroups, simvastatin 40 mg was effective in primary prevention of coronary heart disease in patients with diabetes (number needed to treat, 24 to prevent a major event in 5 years)¹⁷⁸ and in patients who had a history of peripheral or carotid atherosclerosis but not coronary heart disease. Simvastatin 40 mg was also effective in patients who had a baseline low-density lipoprotein less than 116 mg/dL (both patients with and without diabetes).

To address concerns about the potential hazards of lowering cholesterol, data from the Heart Protection Study were analyzed to determine the effect of lowering cholesterol on cause-specific mortality, site-specific cancer incidence, and other major morbidity.¹⁷⁹ There was no evidence of any adverse effect of lowering cholesterol for 5 years on non-vascular morbidity or mortality. There was no increased risk of non-vascular mortality (relative risk, 0.95; 95% CI, 0.85 to 1.07) or cancer incidence (relative risk, 1.00; 95% CI, 0.91 to 1.11).

The Anglo-Scandinavian Cardiac Outcomes Trial—Lipid-lowering Arm (ASCOT-LLA) was a randomized, double-blind, placebo-controlled, fair-to-good quality trial of atorvastatin 10 mg in 10 305 patients with well-controlled hypertension, total cholesterol concentrations less than 251 mg/dL, and an average of 3.7 cardiovascular disease risk factors.^{171, 172} The trial was terminated after a median of 3.3 years of follow-up because a statistically significant benefit was shown on the primary endpoint, non-fatal myocardial infarction (including silent myocardial infarction) and fatal coronary heart disease. Treatment with atorvastatin 10 mg per day for 1 year reduced low-density lipoprotein by 35%, from 133 mg/dL to 87 mg/dL. By the end of follow-up (about 3.3 years), low-density lipoprotein was 89 mg/dL in the patients still taking atorvastatin compared with 127 mg/dL in the control group.

There were 100 primary endpoint events in the atorvastatin group (100/5168, or 1.9%) and 150 events in the placebo group (3%). The event rate in the placebo group corresponded to a 10-year coronary event rate of 9.4%. Over 3.3 years, the number needed to treat to prevent 1 nonfatal myocardial infarction or death from coronary heart disease was 94 (P=0.005). Atorvastatin increased the chance of remaining free of myocardial infarction for 3.3 years from 95% to 97%.

For the secondary and tertiary endpoints, strokes were reduced (number needed to treat, 158; P<0.02), as were cardiovascular procedures, total coronary events, and chronic stable angina. All-cause mortality was 3.6% for atorvastatin compared with 4.1% for placebo (P=0.1649). Atorvastatin did not reduce cardiovascular mortality (1.4% compared with 1.6%), development of diabetes, or development of renal impairment, peripheral vascular disease, heart failure (0.8% compared with 0.7%), or unstable angina.

In ALLHAT-LLC (Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack—Lipid-lowering Arm), a fair-to-good quality, open-label randomized trial, 10 355 hypertensive patients, aged 55 and older, were randomized to pravastatin 40 mg or to usual care.¹²¹ Nearly half the subjects were women, 35% had diabetes, 15% had a history of coronary heart disease, and about 35% were African-American. Pravastatin reduced low-density lipoprotein cholesterol from 145.6 mg/dL at baseline to 111 mg/dL after 2 years, a 24% reduction. However, because the control group was usual care instead of placebo, 10% of control patients were taking a lipid-lowering drug by year 2, and, by year 6, 28.5% of control subjects were taking a lipid-lowering drug. Thus the control group had a mean reduction in low-density lipoprotein cholesterol concentration of 11% over the course of the study.

In ALLHAT-LLC, pravastatin did not reduce all-cause mortality or cardiovascular event rates. The reason for the lack of benefit of pravastatin in ALLHAT-LLC was unclear. The high proportion of women and the high rate of use of statins in the control group are possible explanations.

The good-quality PROSPER trial was designed to examine the benefits of statin therapy in women and in the elderly.¹³³ High-risk men and women were randomized to pravastatin 40 mg or to placebo. Before treatment, the mean low-density lipoprotein was 147 mg/dL. Overall, pravastatin reduced the composite primary endpoint (coronary heart disease death, nonfatal myocardial infarction, and fatal/nonfatal stroke) from 16.2% in the placebo group to 14.1% (P=0.014; number needed to treat, 48). There was also a reduction in transient ischemic attacks, but not in strokes, in the pravastatin group. There was no effect on all-cause mortality, which was 10.5% in the placebo group compared with 10.3% in the pravastatin group (hazard ratio, 0.97; 95% CI, 0.83 to 1.14). The reduction in coronary heart disease deaths in the pravastatin group (4.2% compared with 3.3%; P=0.043) was balanced by an increase in cancer deaths (3.1% compared with 4%; P=0.082).

Pravastatin was more effective in men than in women. There were more women (n=3000) than men (n=2804) in the study. The baseline risk in men was higher. In the placebo group, almost 20% of men and 13% of women had an event (coronary heart disease death, nonfatal myocardial infarction, or stroke) over the 3 years of the study. For men, there was a statistically significant reduction in the primary endpoint (hazard ratio, 0.77; 95% CI, 0.65 to 0.92; number needed to treat, 26). For women, there was no apparent effect (hazard ratio, 0.96; 95% CI, 0.79 to 1.18). PROSPER recruited a select group of elderly subjects. Of 23 770 people who were screened, 16 714 were ineligible or refused to participate.

The PREVEND-IT trial¹²⁴ was a population-based (N=864), randomized, placebo-controlled trial with a 2 X 2 factorial design. Residents of 1 city in the Netherlands with persistent microalbuminuria were randomized to fosinopril and pravastatin for the prevention of cardiovascular morbidity and mortality. In the pravastatin 10 mg compared with placebo arm, there was no reduction in urinary albumin excretion and no significant reduction in cardiovascular events after an average 46 months of follow-up (hazard ratio, 0.87; 95% CI, 0.49 to 1.57). In a subgroup analysis of 286 patients with the metabolic syndrome (33% of the total group),¹⁸⁰ the unadjusted hazard ratio was non-significant (hazard ratio, 0.48; 95% CI, 0.21 to 1.07). However, when adjusted for age and sex, there was a significant reduction in cardiovascular events in the pravastatin group (hazard ratio, 0.39; 95% CI, 0.17 to 0.89).

The ALERT trial established the efficacy and safety of fluvastatin in patients who had undergone renal transplant. Fluvastatin was superior to placebo in reducing cardiac deaths or non-fatal myocardial infarction,^{127, 181, 182} but there was no effect on the renal endpoints of graft loss, doubling of serum creatinine, or decline in glomerular filtration rate.¹⁷³

The MEGA study¹⁴⁴ enrolled Japanese adults without known coronary disease who had coronary heart disease risk equivalents or other risk factors (21% diabetes, 42% hypertension, 20% smokers). Patients were randomized to lower doses of pravastatin 10-20 mg (typical doses used in Japan) plus diet or diet alone and found 33% relative reduction in the incidence of coronary events with pravastatin over a mean follow-up of 5.3 years (hazard ratio, 0.67; 95% CI, 0.49 to 0.91; rate, 1.7% pravastatin compared with 2.55% diet alone). The primary endpoint was driven by reductions in nonfatal myocardial infarction and the need for revascularizations. All-cause mortality was lower in pravastatin-treated patients, though statistical significance was not achieved (hazard ratio, 0.72; 95% CI, 0.51 to 1.01; P=0.055).

Patients with diabetes. There were 8 trials^{125, 134, 142, 145, 146, 178, 183, 184} evaluating long-term effectiveness of atorvastatin 10-20 mg, simvastatin 40 mg, and fluvastatin 80 mg in patients with diabetes (Table 10; Evidence Table 2).

Table 10

Placebo-controlled trials in patients with diabetes.

Of the 8 trials, CARDS (Collaborative Atorvastatin Diabetes Study) was the only study designed to assess primary prevention of cardiovascular disease in patients with type 2 diabetes. Two-thousand eight-hundred thirty eight patients without elevated cholesterol levels (mean low-density lipoprotein less than 107 mg/dL), who had no history of cardiovascular disease but at least 1 of the risk factors of retinopathy, albuminuria, current smoking, or hypertension, were randomized to atorvastatin 10 mg or placebo. After 3.9 years of follow-up, there was a significant relative risk reduction of 37% in cardiovascular events but not with all-cause mortality (Table 10). The CARDS trial was stopped 2 years earlier than planned because of significant benefit at the second interim analysis.

In addition to CARDS, 3 placebo-controlled trials (HPS, ASCOT-LLA, ASPEN)^{142, 178, 184} enrolled patients with type 2 diabetes with and without established cardiovascular disease, and subgroup analyses were performed for those classified as primary prevention. Overall, CARDS, HPS, and ASCOT-LLA^{125, 178, 184} found the study statins to be beneficial in reducing coronary events compared with placebo in patients with type 2 diabetes with and without established cardiovascular disease (Table 10; Evidence Table 2). The HPS trial was the largest of these, including 5963 patients with diabetes. There was a 27% reduction in risk of major coronary events (first nonfatal myocardial infarction or coronary death), similar to the reduction in risk in the overall population of high-risk patients with simvastatin 40 mg. Among the 2912 patients with diabetes who did not have known coronary or other occlusive arterial disease at study entry, there was a 33% reduction in first major vascular events (95% CI, 17 to 46; P=0.0003). The reduction in risk for stroke (24%) in patients with diabetes was also similar to the reduction in the overall high-risk group. ASPEN was the only trial that showed a small nonsignificant reduction in the composite primary outcome of cardiovascular deaths or other cardiovascular events with atorvastatin (Table 10; Evidence Table 2). Potential reasons for not finding a significant effect may have been due to a change in study protocol within 2 years of the start of the study, enrollment of “very low risk” patients, and how the primary endpoint was defined.

There were 2 trials^{145, 183} (LIPS, Xu, et al) that studied the effectiveness of fluvastatin 80 mg or atorvastatin 20 mg in patients with diabetes who had undergone percutaneous coronary interventions. Both trials observed a benefit associated with the study statins compared with placebo (Table 10; Evidence Table 2). All-cause mortality reported in 1¹⁴⁵ trial was not significant.

The 4D trial¹³⁴ enrolled patients with type 2 diabetes who had end-stage renal disease and were receiving maintenance hemodialysis (Table 10; Evidence Table 2). After 4 years of follow-up, there was no difference between atorvastatin 20 mg and placebo on the primary endpoint or all-cause mortality despite low-density lipoprotein of 72 mg/dL. There was also an increase in fatal strokes in the atorvastatin group— although this was likely to be a chance finding— and no effect on any individual component of the primary endpoint. Authors of 4D speculated that nonsignificant results for primary outcome may be related to lower baseline low-density lipoprotein levels, sicker population, and a different pathogenesis of events in this population.

One publication¹⁴⁶ was rated poor quality due to unclear randomization, allocation concealment, intention-to-treat analysis, and inadequate blinding.

Secondary prevention

Four placebo-controlled trials recruited patients with documented coronary heart disease while 1¹⁴¹ enrolled patients with recent stroke or transient ischemic attack without history of coronary heart disease. Two trials (LIPID, CARE)^{122, 130} evaluated pravastatin (N=13 173), 1 trial (4S)¹²⁸ evaluated simvastatin (N=4444), 1 trial evaluated fluvastatin,¹²⁹ and 1 trial (SPARCL)¹⁴¹ evaluated atorvastatin.

Pravastatin and simvastatin significantly reduced the incidence of major coronary events, including overall mortality in LIPID and 4S. In 4S, the 8-year probability of survival was 87.6% in the placebo group and 91.3% in the simvastatin group. The risk of stroke was also reduced in CARE and 4S. In a post hoc subanalysis of 2073 patients in the LIPID trial with low low- and high-density lipoprotein cholesterol, pravastatin was associated with a relative risk reduction of 27% (95% CI, 8 to 42), a 4% absolute risk reduction, and a coronary artery disease of 22 to prevent 1 coronary heart disease event over 6 years.¹⁸⁵

In Riegger et al,¹²⁹ patients who had stable angina were randomized to fluvastatin or placebo. The primary endpoint included cardiac death, nonfatal myocardial infarction, and unstable angina pectoris. By 1 year, there were fewer primary events in the fluvastatin group. However, excluding unstable angina, the relative risk of cardiac death and nonfatal myocardial infarction was not significantly reduced with fluvastatin (RR 0.38; 95% CI, 0.09 to 1.68).

In SPARCL, 4731 patients without coronary heart disease who had recent stroke or transient ischemic attack within 6 months were randomized to atorvastatin 80 mg or to placebo. By 4.9 years of follow-up (range: 4 to 6.6 years), atorvastatin significantly reduced the relative risk of fatal or nonfatal stroke by 16% (hazard ratio, 0.84; 95% CI, 0.71 to 0.99) or by a 1.9% absolute risk reduction (number needed to treat, ~53). Post-hoc analyses stratifying by type of stroke found that patients with ischemic or unclassified type benefited the most while those with hemorrhagic type were more likely to experience a harmful event (hazard ratio, 1.66; 95% CI, 1.08 to 2.55).

Even though none of the patients had established coronary disease, atorvastatin reduced the risk of major coronary events and need for revascularization, but not for death from cardiovascular disease or causes (Evidence Table 2). Deaths from any cause were also not reduced with atorvastatin (hazard ratio, 1.00; 95% CI, 0.82 to 1.21; P=0.98). Reductions in stroke and cardiovascular events were consistent in elderly in a post-hoc analysis.¹⁸⁶

Most patients in SPARCL had prior ischemic stroke (~67%) and transient ischemic attack (~30%). About 2% of those with hemorrhagic stroke were considered to be at risk for ischemic events. About 62% of patients had hypertension, 17% had diabetes, and 19% were smokers. Most patients were naive to statin therapy.

Studies in inpatients with acute coronary syndrome

There were 6 placebo-controlled trials in patients with acute myocardial infarction or unstable angina (Table 11).^135–140 No new trials were identified for Update 5. The trials included 3 of pravastatin 20 to 40 mg and 1 each of atorvastatin 80 mg, fluvastatin 80 mg, and simvastatin 20 to 80 mg. One was rated fair-to-poor quality, and the rest were rated fair quality (see Evidence Tables 3 and 4 for details of quality ratings).

Table 11

Inpatient trials of acute myocardial infarction or unstable angina (statins compared with placebo or usual care).

The L-CAD study established that patients with acute coronary syndrome benefit from statin treatment.¹³⁵ In L-CAD, 126 patients were randomized to pravastatin 20 or 40 mg or usual care an average of 6 days after an acute myocardial infarction or emergency percutaneous transluminal coronary angioplasty due to severe or unstable angina. After 2 years of follow-up, there were fewer major coronary events in the pravastatin group (22.9% compared with 52%; P=0.005). There was no difference in all-cause mortality, but each group had only 2 deaths.

An earlier pilot study¹³⁷ of pravastatin 40 mg compared with placebo enrolled patients hospitalized for less than 48 hours with acute myocardial infarction or unstable angina. After 3 months, there was no significant difference on any clinical endpoint, although there was a 25% reduction in low-density lipoprotein cholesterol in the pravastatin group.

PACT¹⁴⁰ assessed outcomes at 30 days in patients with acute myocardial infarction or unstable angina randomly assigned to receive pravastatin 20 to 40 mg or placebo within 24 hours of the onset of chest pain. This study was rated fair-to-poor quality because of some differences in groups at baseline (higher total cholesterol in placebo group, more placebo patients on hormone replacement therapy, and more pravastatin patients on anticoagulants) and no reporting of randomization and allocation concealment methods. The primary endpoint (composite of death, recurrence of myocardial infarction, or readmission to hospital for unstable angina) occurred in 12% of patients. There was no significant reduction in the primary endpoint (relative risk reduction, 6.4%; 95% CI, –1.4 to +3.0), or on any individual component of the primary endpoint.

In MIRACL,¹³⁹ a short-term (16 weeks) placebo-controlled trial of atorvastatin 80 mg in patients with unstable angina or non-Q-wave myocardial infarction, there was a significant reduction in major coronary events (death, nonfatal acute myocardial infarction, cardiac arrest with resuscitation, or recurrent symptomatic myocardial infarction requiring emergency rehospitalization) in the atorvastatin group (17.4% compared with 14.8%). There were no differences between groups on the individual components myocardial infarction or all-cause mortality, although the study was not powered to detect a difference on these endpoints.

FLORIDA¹³⁶ was a placebo-controlled trial of fluvastatin 80 mg in 540 patients with an acute myocardial infarction plus hypercholesterolemia and new or markedly increased chest pain or a new pathological Q wave. At 1 year of follow-up, there was no difference between groups in the occurrence of major coronary events.

The A to Z trial¹³⁸ compared early intensive statin treatment (simvastatin 40 mg for 30 days and then simvastatin 80 mg thereafter) to a less aggressive strategy (placebo for 4 months and then simvastatin 20 mg thereafter) in patients with either non ST elevation acute coronary syndrome or ST elevation myocardial infarction with a total cholesterol level of 250 mg/dL or lower. Patients were followed for up to 24 months. Despite greater lowering of low-density lipoprotein in the early intensive group, there were no differences between the early intensive and less aggressive groups on the primary endpoint (cardiovascular death, myocardial infarction, readmission for acute coronary syndrome, or stroke), or on any individual component of the primary outcome.

Nine patients in the simvastatin only group developed myopathy (creatine kinase level greater than 10 times the upper limit of normal with associated muscle symptoms) while taking 80 mg compared with 1 patient in the placebo first group (P=0.02). Three of the 9 in the simvastatin group had creatine kinase levels higher than 10 000 units/L and met the definition for rhabdomyolysis. The rate of myopathy was high, despite the exclusion of patients at increased risk of myopathy due to renal impairment or concomitant therapy with agents known to enhance myopathy risk, or for having a prior history of nonexercise-related elevations in creatine kinase level or nontraumatic rhabdomyolysis.

The lack of effect of more intensive treatment in this trial may have been due to several factors. The “early intensive” group started with only 40 mg of simvastatin, and did not increase to 80 mg for 30 days. Patients who were taking statin therapy at the time of their myocardial infarction (at randomization) were excluded. The study authors reported that the trial had less statistical power than originally planned due to a lower than expected number of end points and a higher than expected rate of study drug discontinuation.

The large randomized trials summarized above provided strong evidence about the balance of benefits and harms from statin therapy. Because they were analyzed on an intention-to-treat basis, the benefits (reductions in coronary events, strokes, and, in some studies, mortality) in subjects who tolerated and complied with medication were diluted by the lack of benefit in subjects who discontinued medication because of side effects or did not complete the study for other reasons. Moreover, the mortality results of the trials indicated clearly that for the enrolled subjects and the duration of the trials, statins are beneficial. The balance of benefits and harms of statin drugs over a longer time than the trial durations remains unclear.

Studies of the progression of atherosclerosis with secondary or incidental coronary heart disease endpoints

Twelve studies of the effects of statins on progression of atherosclerosis also reported rates of coronary or cardiovascular events.^147–158 A head-to-head trial¹⁸⁷ of the effect of atorvastatin 80 mg compared with pravastatin 40 mg on progression of atherosclerosis did not meet inclusion criteria because it did not report health outcomes. However, this study did meet inclusion criteria for Key Question 1 (see Evidence Table 1). In these studies, the primary endpoint was progression of atherosclerosis, and all of the patients had known coronary heart disease. To answer the question of whether treatment with a statin is associated with a reduction in clinical cardiovascular outcomes in patients with coronary heart disease, these studies were considered fair or fair-to-poor quality. In 6 of the 12 trials clinical outcomes were not a preplanned endpoint (they were “spontaneously reported”), and sample sizes were relatively small.

Table 12 and Evidence Table 5 summarize the results of these studies. The number of trials and patients studied for each statin are as follows: fluvastatin (1 trial; N=429), lovastatin (3 trials; N=1520), pravastatin (5 trials; N=2220), and simvastatin (3 trials; N=1118). The information about fluvastatin was inconclusive and the other 3 statins were already known to be effective from better studies.

Table 12

Studies of atherosclerotic progression that reported coronary heart disease outcomes.

In general, most trials in which coronary heart disease events were not a prespecified endpoint found a trend towards a reduction in clinical events in favor of a statin. In the trials in which coronary heart disease events were a secondary endpoint, there was usually a significant reduction in 1 of the components of coronary heart disease events. While consistent, the results of these studies are difficult to interpret because of possible reporting bias. That is, these trials may have been more likely to report a result if it was statistically significant or indicated a trend favoring treatment. Similar trials of progression of atherosclerosis that found no trend probably did not report coronary events. For this reason, we did not conduct a meta-analysis to pool the results of these studies.

Women and the elderly

Although women and the elderly were under-represented in the early major trials, we found 4 meta-analyses^191–194 suggesting that statins are equally efficacious in men, women, and the elderly.

One meta-analysis¹⁹¹ evaluated the effect of statins on the risk of coronary disease from 5 large, long-term, primary and secondary prevention trials (see Evidence Table 2). Women accounted for an average of 17% of subjects and individuals age 65 and older accounted for an average of 29% of subjects with a range of 21% to 39% (WOSCOPS did not enroll women or anyone 65 years or older). The risk reduction in major coronary events was 29% (95% CI, 13 to 42) in women, 31% (95% CI, 26 to 35) for men, 32% (95% CI, 23 to 39) in those over age 65, and 31% (95% CI, 24 to 36) in those younger than age 65. Similarly, the Heart Protection Study^{123, 178} found that simvastatin reduced cardiovascular events among women generally and particularly in women with diabetes, who benefited dramatically (number needed to treat, 23 to prevent 1 major vascular event).

Unlike the analysis by La Rosa and colleagues¹⁹¹ that reported morbidity results, a meta-analysis by Walsh and colleagues¹⁹² reported on total mortality, coronary heart disease mortality, and other coronary heart disease events in women with and without prior cardiovascular disease. Nine trials of statins that enrolled 16 486 women and 4 additional studies that included 1405 women who used drug therapy other than statins were included in the analysis. For secondary prevention, lipid-lowering therapy reduced risk of coronary heart disease mortality (summary RR 0.74; 95% CI, 0.55 to 1.00), nonfatal myocardial infarction (summary RR 0.73; 95% CI, 0.59 to 0.90), and coronary heart disease events (summary RR 0.80; 95% CI, 0.71 to 0.91), but not total mortality (summary RR 1.00; 95% CI, 0.77 to 1.29). In primary prevention studies, there was insufficient evidence of reduced risk of any clinical outcome in women, because of the small number of events in the trials. Sensitivity analyses including only studies using statins did not significantly affect the summary risk estimates.

Two meta-analyses^{193, 194} specifically evaluating statins in the elderly confirmed prior findings that these drugs are effective in this population. In particular, a hierarchial bayesian meta-analysis¹⁹³ included 9 placebo-controlled trials that enrolled 19 569 elderly patients who had a history of cardiovascular events. The pooled relative risk for all-cause mortality was 0.78 (95% CI, 0.65 to 0.89) with a posterior mean estimate of the number needed to treat of 28 (95% CI, 15 to 56) favoring statins over a mean weighted follow-up period of 4.9 years. Coronary heart disease mortality, nonfatal myocardial infarction, need for revascularization, and stroke were all statistically significantly reduced with statins compared with placebo (Evidence Table 8). Of note, the Heart Protection study (which included primary prevention population) was included in the meta-analysis but a sensitivity analysis with and without this trial showed consistent treatment effects. Statins that were included were simvastatin 20–40 mg, pravastatin 40 mg, and fluvastatin 80 mg.

African American, Hispanic, and other ethnic groups

African Americans had the greatest overall coronary heart disease mortality and the highest out-of-hospital coronary death rates of any other ethnic group in the United States.⁴ Other ethnic and minority groups in the United States included Hispanics, Native Americans, Asian and Pacific Islanders, and South Asians. However, these groups are underrepresented in randomized clinical trials reporting reductions in clinical outcomes. As a result there was no evidence to answer whether or not statins differ in their ability to reduce clinical events in the African American, Hispanic, or other ethnic groups. Significant numbers of African American and Hispanic patients participated in AFCAPS/TexCAPS, but the investigators did not analyze events by racial group. In EXCEL, lovastatin 20 mg, 40 mg, and 80 mg daily reduced low-density lipoprotein cholesterol by similar percentages in blacks and in whites.¹⁹⁵

In short-term head-to-head trials, reductions in low-density lipoprotein cholesterol and frequency of adverse events with rosuvastatin 10 to 20 mg and atorvastatin 10 to 20 mg in Hipanic,²³ South Asian,¹⁹⁶ and African American⁷⁴ patients were similar to those observed in studies conducted in primarily white non-Hispanic populations.

Magnitude of risk

Gaist and colleagues²²² conducted a population-based observational study in which 3 cohorts of patients were identified. The first cohort consisted of patients (n=17 219) who had received at least 1 prescription for lipid-lowering drugs. The second cohort consisted of patients (n=28 974) who had a diagnosis of hyperlipidemia but did not receive lipid-lowering drugs. The third cohort consisted of people (n=50 000) from the general population without a diagnosis of hypercholesterolemia. Using diagnostic visit codes recorded by participants in the U.K. General Practice Research Database, they identified and verified cases of symptomatic myopathic pain. A potential case of myopathy was confirmed with the clinician when the patient presented at least 2 of the following criteria: (1) clinical diagnosis of myopathy confirmed by the general practitioner; (2) muscle weakness, muscle pain, or muscle tenderness (2 of these symptoms); and (3) creatine kinase concentration above the reference limit. By this definition, the incidence of myopathy in the lipid-lowering group was 2.3 per 10 000 person-years (95% CI, 1.2 to 4.4) compared with none per 10 000 person-years in the non treated group (95% CI, 0 to 0.4) and 0.2 per 10 000 person-years (95% CI, 0.1 to 0.4) in the general population. In 17 086 person-years of statin treatment, there were only 2 cases of myopathy. In this study, rates of myotoxicity were not differentiated between statins.

In a systematic review, the incidence of myalgia in clinical trials ranged from 1% to 5% and was not significantly different from placebo. However, a review of 2 databases in the same review found that myalgia (defined as muscle pain without elevated creatine kinase levels) contributed to 19% to 25% and 6% to 14% of all adverse events associated with statin use.²²⁰ In a large meta-analysis of 119 double-blind, placebo-controlled randomized-controlled trials, the odds of myalgia with statin monotherapy were no different than that of placebo (odds ratio, 1.09; 95% CI, 0.97 to 1.23).²⁰⁹ There was an increased risk of myositis with an odds ratio of 2.56 (95% CI, 1.12 to 5.58).

Myotoxicity of different statins

All of the available statins (simvastatin, lovastatin, atorvastatin, fluvastatin, pravastatin, and rosuvastatin), when administered alone, have been associated with infrequent myotoxic adverse effects ranging from myalgia and myopathy to rhabdomyolysis.²⁰⁶ Factors that may increase the risk for myopathy or rhabdomyolysis with statins are higher dosages, drug interactions, other myotoxic drugs (fibrates or niacin), increased age, hypothyroidism, surgery or trauma, heavy exercise, excessive alcohol intake, and renal or liver impairment.^{217, 219, 223, 224}

A retrospective analysis of all domestic and foreign reports of statin-associated rhabdomyolysis has been released by the Food and Drug Administration.²¹⁸ During a 29-month period (November 1997 to March 2000) there were 871 reported cases of rhabdomyolysis. The number of cases (% of total) for each statin were as follows: atorvastatin, 73 (12.2%); fluvastatin, 10 (1.7%); lovastatin, 40 (6.7%); pravastatin, 71 (11.8%); and simvastatin, 215 (35.8%). The report also included cerivastatin with 192 (31.9%) cases of rhabdomyolysis. In the majority of these cases, a drug with the potential for increasing the statin serum level was identified. This report does not provide information about the relative incidence of rhabdomyolysis associated with different statins, because the number of patients taking each statin was not available.

Another review of reports to the US Food and Drug Administration’s MedWatch database limited to events associated with atorvastatin or simvastatin was published in April 2003.²²⁵ The analysis was limited to adverse reactions that affected major organ systems (muscle toxicity, hepatotoxicity, pancreatic toxicity, and bone marrow toxicity). Analyses were adjusted for dose but not low-density lipoprotein cholesterol lowering. Between November 1997 and April 2000, there were 1828 adverse event reports affecting major organ systems associated with the use of atorvastatin, and 1028 reports associated with simvastatin. Muscle-related events were more likely with atorvastatin (dose adjusted odds ratio, 1.7; 95% CI, 1.6 to 1.8; P<0.001). Reports of myalgias were more likely with atorvastatin, but rhabdomyolysis-associated reports were more likely with simvastatin (dose adjusted odds ratio, 2.4; 95% CI, 2.1 to 2.7; P<0.001).

Dale et al, 2007 performed a systematic review of randomized-controlled trials comparing higher with moderate intensity statin therapy. They included 9 trials with primarily high dose of atorvastatin or simvastatin to lower doses of atorvastatin, simvastatin, pravastatin, or lovastatin.²¹⁶ They evaluated hydrophilic (pravastatin) statins separately from the other more lipophilic statins and found an increase risk of significant creatinine kinase elevation but only in the lipophilic statins and not in the hydrophilic statins (relative risk, 6.09; 95% CI, 1.36 to 27.35). They did report that rosuvastatin was considered a hydrophilic statin, however no data on rosuvastatin was included in this review.

From these studies, conclusions regarding the differences in the risk of severe muscle toxicity between statins could not be made since there are significant limitations to voluntary, spontaneous reporting systems. For example, the actual exposure (denominator) of a population to a statin is not known, so the true incidence rates of an adverse effect cannot be determined. Furthermore, the number of reported cases (numerator) may be underestimated.

Another observational study used claims data from 11 United States-managed health care plans to estimate the incidence of rhabdomyolysis leading to hospitalization in patients treated with different statins and fibrates, alone and in combination.²²⁶ Fluvastatin and lovastatin were excluded from the analysis because usage was very low. There were 16 cases of rhabdomyolysis leading to hospitalization with statin monotherapy in 252 460 patients contributing 225 640 person-years of observation. Incidence rates for monotherapy with atorvastatin, pravastatin, and simvastatin were similar.

In our review of 83 head-to-head comparative statin low-density lipoprotein cholesterol-lowering trials, we did not find any differences in rates of muscle toxicity between statins. In the ASTEROID trial, a study of regression of atherosclerosis, there were no cases of rhabdomyolysis in 507 patients taking rosuvastatin 40 mg for 24 months.²²⁷ This trial is not included in our efficacy analysis because health outcomes were not reported.

Patients with diabetes

There are no data to support any special safety concerns in patients with diabetes receiving statins. In short-term head-to-head studies of atorvastatin compared with rosuvastatin in patients with diabetes, the type and frequency of adverse events was similar to those found in studies of patients without diabetes.^{78, 95, 231}

In the Heart Protection Study (HPS, simvastatin), substantial elevations of liver enzymes and creatinine kinase were not significantly higher in patients with diabetes. Moreover, taking simvastatin for 5 years did not adversely affect glycemic control or renal function. It should be noted, however, that the Heart Protection Study had a run-in period in which patients who had liver or muscle enzyme elevations were excluded prior to randomization.

In CARDS,¹²⁵ there was no difference between atorvastatin and placebo in the frequency of adverse events or serious adverse events, including myopathy, myalgia, rise in creatinine phosphokinase, and discontinuation from treatment for muscle-related events. There were no cases of rhabdomyolysis.

A 4-month, head-to-head trial of extended-release fluvastatin 80 mg compared with atorvastatin 20 mg was conducted in 100 patients with type 2 diabetes and low serum high-density lipoprotein levels.²³² The study was designed to measure the metabolic effects of the statins and did not measure clinical endpoints. There were no significant changes in serum creatinine phosphokinase or liver enzymes and no major adverse events after 4 months of treatment.

A 48-week trial assessed efficacy and safety of long-term treatment with fluvastatin in patients with chronic renal disease and hyperlipidemia.²³³ Patients with diabetic nephropathy (N=34) or chronic glomerulonephritis (N=46) were randomized to fluvastatin 20 mg plus dietary therapy, or dietary therapy alone. Over 48 weeks of treatment, there were no significant differences between fluvastatin and placebo groups in serum creatinine concentration, creatinine clearance, or 24-hour urinary albumin excretion rates.

Adverse event rates were similar between atorvastatin and placebo-treated patients enrolled in the ASPEN trial.¹⁴² Abnormal liver function tests occurred in 1.4% using atorvastatin compared with 1.2% in the placebo group. The rate of myalgia was more frequent with atorvastatin (3% compared with 1.6%; P value not reported). Two cases of rhabdomyolysis were reported, 1 in each treatment arm. Neither of the cases were thought to be related to the interventions.

Special populations and statin-drug interactions

To assess whether a particular statin is safer in a special population, a review of potential drug interactions is necessary. We identified 7 non-systematic reviews pertaining to statin drug interactions.^{206, 234–239} Briefly, simvastatin, lovastatin, and atorvastatin are all metabolized in the liver via the cytochrome P450 3A4 isoenzyme system. As a result, all 3 agents are susceptible to drug interactions when administered concomitantly with agents known to inhibit metabolism via CYP 3A4. The use of the agents listed below increases statin concentrations and, theoretically, the possibility for adverse effects and does not include all drugs capable of inhibiting metabolism via the CYP 3A4 isoenzyme system.

The significance of interactions with many drugs that inhibit CYP 3A4 is not known; examples include diltiazem, verapamil, and fluoxetine. Fluvastatin is primarily metabolized via CYP 2C9 and is vulnerable to interactions with drugs known to inhibit CYP 2C9 metabolism. Only about 10% of rosuvastatin is metabolized, primarily through the CYP 2C9 system. Pravastatin is not significantly metabolized via the CYP isoenzyme system and is therefore not affected by drugs inhibiting metabolism via these pathways.

Statin-clopidogrel. Several pharmacokinetic studies have suggested potential drug interaction with atorvastatin (and other CYP 3A4 statins) and clopidogrel. Clopidogrel is a prodrug that requires activation via CYP 3A4/2C19.

We identified 9 publications^240–248 examining the potential drug interaction with regard to clinical outcomes. Of these, 8 studies^{240, 242–248} collectively showed little difference in the risk of cardiovascular events (myocardial infarction, death, revascularization, hospitalization, etc) in patients at high risk for atherothrombotic events (with or without percutaneous coronary intervention) for those receiving statin-clopidogrel combination compared with those using statin or clopidogrel monotherapy. There was also a minimal difference in risk between groups when statins were stratified by whether they were metabolized by 3A4 or non-3A4 pathways.

Study designs were retrospective or post-hoc analyses of larger randomized trials. Each study had its limitations such as small sample size (lack of power), unknown statin doses, unclear duration of statin or clopidogrel combination therapy, potential selection bias in database studies, and unknown adherence to therapy; thus, the results should be interpreted carefully.

Statin-efavirenz. We found 1 small retrospective review (N=13)²⁴⁹ that assessed the potential drug interaction with the combination of simvastatin to an efavirenz-based regimen in HIV-infected and non-infected patients. Efavirenz is a non-nucleoside reverse transcriptase inhibitor that has CYP 3A4 inductive effects and the combination with simvastatin, a 3A4 substrate, could potentially lead to less of a statin treatment effect. This study found small non-significant absolute differences in low-density lipoprotein and total cholesterol lowering effects between those using simvastatin-efavirenz and those using only statin therapy. There were no reports of myopathies or elevated liver transaminase and creatine kinase levels in the chart reviews.

Potent inhibitors of CYP 3A4 are listed below:

• Clarithromycin
• Erythromycin
• Cyclosporine
• Protease inhibitors (indinivir, nelfinavir, ritonavir, saquinavir, amprenavir, lopinavir/ritonavir)
• Delavirdine
• Itraconazole
• Fluconazole
• Ketoconazole
• Nefazodone
• Grapefruit juice

Published reports of rhabdomyolysis exist in patients receiving concomitant statin with Clarithromycin, Erythromycin, Cyclosporine, Itraconazole, and Nefazodone.

Drugs known to inhibit metabolism via CYP 2C9 are listed below:

• Amiodarone
• Azole Antifungals
• Cimetidine
• Fluoxetine
• Fluvoxamine
• Metronidazole
• Omeprazole
• TMP/SMX
• Zafirlukast

Harms in organ transplant recipients. The main concern of statin therapy in organ transplant patients is the potential for increased musculoskeletal and hepato-toxicities from statin-drug interaction, especially for drugs that are substrates (simvastatin, lovastatin, atorvastatin) and inhibitors (cyclosporine) of the CYP 3A4 pathway.

The risk for adverse events with statins in combination with cyclosporine appears to be dose-related. Long-term, single-drug treatment of hyperlipidemia with simvastatin at doses not exceeding 10 mg daily, respectively, has been shown to be well tolerated with minimal harms in cardiac and renal transplant patients receiving cyclosporine.^{250, 251} Fluvastatin 20–80 mg daily and pravastatin at 20–40 mg daily have also been shown to be relatively safe in cyclosporine-managed cardiac and renal transplant recipients.^{127, 252–255} A post hoc analysis of the ALERT trial, one of the largest renal transplant trials evaluating fluvastatin, found little statistical difference between fluvastatin and placebo-treated groups with or without diabetes with regards to changes in serum creatinine, creatinine clearance, proteinuria, serious renal adverse events leading to study withdrawal, or incidence of graft loss.²⁵⁶ There was also little difference in the incidence of transplant rejection within the first post-transplantation year between pravastatin and placebo-treated identified patients in a different retrospective study.²⁵⁷ Rosuvastatin 10 mg (average dose) was studied in a cohort study of 21 cardiac transplant recipients receiving standard immunosuppressive therapy.²⁵⁸ The patients’ lipid levels were above target values on the highest tolerated doses of other statins. After 6 weeks, there were no statistically significant changes in creatine kinase levels or aspartate aminotransferase. There was no clinical evidence of myositis in any patient. One patient had myalgia and 2 patients were withdrawn because of mild elevation of creatine kinase (324 U/liter at 3 weeks and 458 U/liter at 6 weeks). In a premarketing study, cyclosporine had a clinically significant effect on the drug concentrations of rosuvastatin in heart transplant patients. The product label recommends limiting the dose of rosuvastatin to 5 mg in patients taking cyclosporine.

Only 1 case of rhabdomyolysis was identified from a heart transplant registry which included 210 patients managed with a variety of statins for 1 year.²⁵⁹ The patient with rhabdomyolysis was receiving simvastatin 20 mg daily. No rhabdomyolysis was seen in 39 patients receiving simvastatin 10 mg daily. A review of studies involving fluvastatin (up to 80 mg daily) in organ transplant patients receiving cyclosporine identified no cases of rhabdomyolysis.²⁶⁰ One small study²⁶¹ involving atorvastatin (10 mg/day) in 10 renal-transplant recipients taking cyclosporine observed a significant benefit with regard to lipid levels and no cases of myopathy or rhabdomyolysis.

A small prospective, single-center cohort study found that 80% of heart transplant patients who were converted from cyclosporine and high-dose fluvastatin regimen to tacrolimus and atorvastatin 20–40 mg therapy tolerated the switch through 13 months. There were no reports of myalgias, significant elevations in creatine kinase, myopathies, or liver toxicities.²⁶²

Harms in HIV-infected patients: Statins and protease-inhibitors. A significant proportion of HIV-infected patients receiving protease inhibitors developed hyperlipidemia as an adverse effect. As a result, these patients required lipid-lowering treatment. Because of the severity of the lipid elevation, statins are often prescribed to these patients but little is known about the harms observed in this population.

To date, good-quality long-term clinical data evaluating the combination of the protease inhibitors with statins are limited. Pharmacokinetic studies have shown that when simvastatin or atorvastatin (CYP 3A4 substrates) are used in combination with potent CYP 3A4 inhibitors (such as ritonavir and/or saquinavir), increased drug concentrations of statins may lead to greater potential risk for myopathies and rhabdomyolysis.²⁶³

We identified 8 publications^{25, 264–270} that reported harms in HIV-infected patients receiving combination therapy with protease inhibitors and statins or fibrates. Of these, 7^264–270 studied primarily pravastatin while 1²⁵ reported “combined statin” results.

Of the 7 pravastatin studies, 3 randomized trials compared pravastatin 40 mg daily with placebo in HIV-infected patients receiving a protease-inhibitor (45% to 90% were prescribed ritonavir).^{266, 269, 270} Over 8–12 week period, there were no reports of myopathy or rhabdomyolysis and no significant changes in aspartate aminotransferase, alanine aminotransferase, or creatine phosphokinase levels between treatment groups or across trials. Four cases of mild to moderate myalgias were found with pravastatin than with 1 case in the placebo group.^{266, 270} “Severe” muscle aches developed in 2 patients in 1 trial,²⁷⁰ but neither discontinued therapy and their creatine phosphokinase levels were within normal limits. Only 1 pravastatin-treated patient withdrew from a trial because of seizure and hospitalization, which was not related to study treatment.²⁶⁶

Three open-label, randomized trials^{264, 267, 268} and 1 prospective observational study²⁶⁵ also found that HIV-infected patients using combination therapy with a protease-inhibitor and low-dose statin or fibrate tolerated the combination fairly well except for some gastrointestinal complaints such as nausea, dyspepsia, diarrhea, and meteorism (range: 2%–12%). There were no reports of myalgias or myositis during 48–72 weeks of follow-up and no significant elevations in creatine kinase or liver transaminases. All patients were using a protease inhibitor with about 27% to 88% using ritonavir. Totally daily doses of statins and fibrates studied were: pravastatin 10–20 mg, atorvastatin 10 mg, rosuvastatin 10 mg, fluvastatin 20–40 mg, fenofibrate 200 mg, gemfibrozil 1200 mg, and bezafibrate LA 400 mg.

Two groups of experts have made recommendations regarding the use of statins in HIV-infected individuals receiving protease inhibitors, including the Adult AIDS Clinical Trials Research Group (AACTG) Cardiovascular Disease Focus Group and the Centers for Disease Control and Prevention/Department of Health and Human Services/Henry J Kaiser Foundation. Both groups have recommended avoidance of simvastatin and lovastatin in patients receiving protease inhibitors largely based on pharmacokinetic studies and suggest using low-to mid-level doses of atorvastatin, fluvastatin, or pravastatin as alternatives (http://wwwhivatis.org and http://www.aactg.s-3.com/ann.htm).

Statins in HIV-infected patients with comorbidities. One small (N=80) retrospective chart review compared harms in HIV-positive and hepatitis C virus co-infection patients using statins compared with HIV-positive and hepatitis C virus/hepatitis B virus-negative patients using statins.²⁵ The purpose of the study was to evaluate whether statins increased hepatotoxicity between the 2 groups. Most patients were middle-aged men and about 45% were taking antiretroviral therapy with a protease inhibitor. Sixty-four percent of included patients were using atorvastatin, 29% pravastatin, 5% rosuvastatin, and 2.5% simvastatin. Elevated liver enzymes (≥1.5 times the baseline values) were considered significant in this study. Overall, there were no major differences in the number of patients with liver enzymes ≥ 1.5 times baseline values between treatment groups. About 7.9% of co-infected patients observed a ≥ 1.5 time elevation in alanine aminotransferase but this was lower than alanine aminotransferase values found in hepatitis C virus/hepatitis B virus-negative group. No patients discontinued statin therapy because of liver toxicities or modified their antiretroviral therapies due to drug interactions. The results from this study should be considered with caution due to poor internal quality.

Harms of statin-fibrates combination (rhabdomyolysis and myopathy). Myopathy and rhabdomyolysis have also been reported in patients receiving monotherapy with fibrates, especially in patients with impaired renal function. Although the mechanism of the interaction is not completely known, it appears the combination of statins with fibrates, and to a lesser extent niacin, can result in a higher risk for myopathy or rhabdomyolysis. These adverse effects may also be dose-related.^{206, 224, 271} The mechanism for the interaction is unclear but it is hypothesized that gemfibrozil inhibits glucuronidation of statins.

We identified 12 studies^{218, 219, 226, 272–280} reporting harms with statin-fibrate combination. Of these, 8^{218, 219, 226, 272, 275, 276, 278, 280} reported information on rhabdomyolysis, 3^{219, 279, 280} on myopathy, and 4 studies^{219, 273, 274, 277} reported data on other harms such as elevations in liver transaminase or creatine kinase levels.

Of the 8 studies that reported information on rhabdomyolysis, 1 systematic review²¹⁹ of 36 studies (ranging from 2 to 184 weeks in duration) and 2 shorter-term trials^{278, 280} (12 to 22 weeks in duration) that evaluated statin-fibrate combination therapy in the management of hypercholesterolemia, reported no cases of rhabdomyolysis.

In the systematic review by Shek and colleagues,²¹⁹ the majority of included studies used gemfibrozil (total daily dose of 1200 mg; n=20, 63% of patients). Ten studies used bezafibrate, 2 used fenofibrate, 1 used clofibrate, 1 used ciprofibrate, 1 used both bezafibrate and ciprofibrate, 1 used bezafibrate or fenofibrate, and 1 used gemfibrozil or ciprofibrate. No reports of rhabdomyolysis were observed in the 1674 patients receiving statin-fibrate combination. A total of 19 (1.14%) patients withdrew secondary to myalgia or creatine kinase elevation. Two patients (0.12%) developed myopathy (defined as myalgia with creatine kinase >10 times the upper limit of normal) and 33 (1.9%) patients experienced other muscle symptoms including myalgia, musculoskeletal pain or weakness, or myositis. There were 35 reports (2.1%) of subclinical elevation of creatine kinase (<10 times the upper limit of normal) in 16 of the included studies. All but 2 of these studies used gemfibrozil; the others used bezafibrate plus simvastatin 20 mg and fenofibrate plus pravastatin 20 mg or simvastatin 10 mg. Some of the studies did not report whether the creatine kinase elevation was symptomatic or if treatment was discontinued as a result. In 1 of the included studies, a patient tolerated the combination of pravastatin and gemfibrozil for 4 years, and then developed myopathy with clinically important elevation in creatine kinase after being switched to simvastatin.

Shek and colleagues²¹⁹ also found 29 published case reports of rhabdomyolysis secondary to statin-fibrate combination not captured in the above 36 publications. Gemfibrozil was the fibrate used in each case. Statins used were lovastatin in 21 cases, simvastatin in 4 cases, cerivastatin in 3 cases, and atorvastatin in 1 case. Time to developing rhabdomyolysis was rapid (17% within 2 weeks and 93% within 12 weeks) and the onset of symptoms ranged from 36 hours to 36 weeks. No case reports of severe myopathy or rhabdomyolysis in patients receiving pravastatin or fluvastatin combined with a fibrate were found. Similarly, there were no reports of severe myopathy or rhabdomyolysis in a different trial evaluating combination of pravastatin and gemfibrozil.²⁸⁰ However, cases of pravastatin or fluvastatin combined with a fibrate resulting in rhabdomyolysis have been reported.²¹⁸

There were several limitations to this systematic review.²¹⁹ First, included trials tended to exclude patients who had risk factors or comorbidities for developing adverse outcomes. Therefore, data based on these trials likely underestimate rates of adverse events in the broader population. Also, some of the included studies did not report numbers and reasons for study withdrawal and were not of the best quality.

We identified 2 observational studies that found statin-fibrate combination therapy to have higher rates of rhabdomyolysis compared with statin monotherapy.^{226, 272} Data collected in these studies included the time period when cerivastatin was on the market and when serious adverse events were being reported. The inclusion of cerivastatin in both studies could have inflated rates observed, so results should be considered with caution.

A retrospective cohort study of 252 460 patients using claims data from 11 managed health care plans found 24 cases of hospitalized rhabdomyolysis occurring during treatment.²²⁶ The average incidence of rhabdomyolysis requiring hospitalization was 0.44 per 10 000 (95% CI, 0.20 to 0.84) and was similar for atorvastatin, pravastatin, and simvastatin monotherapy. When taken in combination with a fibrate, statins were associated with a higher incidence of hospitalized rhabdomyolysis of 5.98 (95% CI, 0.72 to 216) per 10 000. The study of health plan claims data referred to above reported cases of rhabdomyolysis with the combination of a statin and a fibrate.²²⁶ The cohort represented 7300 person-years of combined therapy with statins and fibrates (gemfibrozil or fenofibrate). There were 8 cases of rhabdomyolysis with combination therapy. Incidence rates per 10 000 person-years were 22.45 (95% CI, 0.57 to 125) for atorvastatin combined with fenofibrate, 18.73 (95% CI, 0.47 to104) for simvastatin combined with gemfibrozil, and 1035 (95% CI, 389 to 2117) for cerivastatin plus gemfibrozil. There were no cases with pravastatin; fluvastatin and lovastatin were excluded from the analysis because usage was very low.

Another retrospective review from the US Food and Drug Administration’s adverse events reporting system found 866 cases of rhabdomyolysis, of which 44% were related to statin-gemfibrozil combination therapy and 56% with statin monotherapy.²⁷² Almost half of the monotherapy cases and about 75% of combination therapy cases were believed to be from cerivastatin. When individual statins were stratified based on mono-or combination therapy, the crude reporting rates for rhabdomyolysis per an estimated 100 000 prescriptions over marketing years (1988-July 2001) was higher with statin-gemfibrozil combinations than statin monotherapy. The crude reporting rates for combination compared with monotherapy were: lovastatin (2.84 compared with 0.12), pravastatin (0.14 compared with 0.02), simvastatin (3.85 compared with 0.08), atorvastatin (0.50 compared with 0.03), fluvastatin (0.00 compared with 0.00), and cerivastatin (1248.66 compared with 1.81).

In addition to the above observational studies, we found 2 retrospective reviews using the US Food and Drug Administration’s adverse event reporting system to compare rates of rhabdomyolysis between statin-fenofibrate and statin-gemfibrozil combination therapies.^{275, 276} Both studies found fewer reports or lower rates of rhabdomyolysis associated with statin-fenofibrate use than statin-gemfibrozil use. The number of cases reported in the Jones study²⁷⁶ for statin-fenofibrate compared with statin-gemfibrozil was 0.58 compared with 8.6 per million prescriptions dispensed, excluding cerivastatin, whereas the odds ratio of rhabdomyolysis was 1.36 (95% CI, 1.12 to 1.71; P=0.002) for statin-fenofibrate compared with an odds ratio of 2.67 (95% CI, 2.11 to 3.30; P<0.001) for statin-gemfibrozil. Since data from the US Food and Drug Administration database are dependent on volunteer reports of adverse events, rates may be an underestimation of “actual” events for either combination therapies and results should be considered carefully.

Of the 12 publications that reported harms associated with statin-fibrate therapy, the remaining publications^{273, 274, 277} showed variable rates of elevated liver transaminase or creatine kinase elevations with combination statin-fibrate usage compared with placebo, statin, or fibrate monotherapies. The evidence base was limited and results should be interpreted carefully.

A pooled analysis evaluated the frequency of creatine kinase elevations in Novartis-funded trials in which fluvastatin was administered in combination with fibrates.²⁷⁴ Of 1017 patients treated with combination therapy, 493 received bezafibrate, 158 fenofibrate, and 366 gemfibrozil. Mean exposure time was 37.6 weeks and ranged from 0.7 to 118.3 weeks. Results were not reported separately by type of fibrate. Five of 1017 patients (0.5%) had creatine kinase elevations greater than or equal to 5 times the upper limit of normal; 2 of these were greater than or equal to 10 times the upper limit of normal. There were no significant differences in the frequency of creatine kinase elevations among the group on combination therapy and patients taking placebo, fibrates only, or fluvastatin only. Similarly, there were no large differences in liver function tests or creatine kinase levels found between the atorvastatin-fenofibrate treatment group and atorvastatin or fenofibrate monotherapy groups in 2 short-term (8–16 week) studies.^{273, 277} There were also no deaths, no increased risk of renal failure, and no liver function tests >3 times the upper limit of normal.²⁷³

A prospective observational cohort study followed 252 patients who were prescribed a statin combined with gemfibrozil for a mean of 2.36 years (range 6 weeks to 8.6 years). Creatine kinase levels, aminotransferase levels, and any reports of muscle soreness or weakness were monitored. One presumed case of myositis occurred in a patient who took simvastatin for 1 year. The patient had previously taken pravastatin combination therapy for 4 years without incident. An asymptomatic 5-fold rise in alanine aminotransferase was observed in 1 patient, and 2 other patients had an alanine aminotransferase elevation between 2 and 3 times the upper limit of normal. The statin involved in these cases is not specified.

Because of the nature of adverse effect reporting and the available evidence, whether one statin is safer than the other with regard to combination therapy with fibrates is still unclear. The US Food and Drug Administration has approved the following recommendations when combining fibric acid derivatives or niacin with a statin:

• Atorvastatin: Weigh the potential benefits and risks and closely monitor patients on combined therapy.
• Fluvastatin: The combination with fibrates should generally be avoided.
• Pravastatin: Avoid the combination with fibrates unless the benefit outweighs the risk of such therapy.
• Simvastatin: Avoid the combination with gemfibrozil unless the benefit outweighs the risk and limit doses to 10 mg if combined with gemfibrozil.
• Lovastatin: Avoid the combination with fibrates unless the benefit outweighs the risk and limit doses to 20 mg if combined with fibrates.
• Rosuvastatin: Avoid the combination with fibrates unless the benefit outweighs the risk and limit doses to 10 mg if combined with gemfibrozil.

Elevation in liver enzymes. In the systematic review by Shek in 2001,²¹⁹ 8 patients in 3 of the 36 included studies discontinued the combination therapy due to significant elevation in liver transaminases (alanine aminotransferase and aspartate aminotransferase). In most of the other studies, there were only reports of subclinical (<3 times the upper limit of normal) elevation in alanine aminotransferase or aspartate aminotransferase. Conclusions regarding the safety of different statins in the liver were not made.

A retrospective database analysis evaluated the risk of elevated liver enzymes in patients who were prescribed a statin.²⁸¹ Changes in liver transaminases at 6 months were compared in 3 cohorts: patients with elevated baseline enzymes (aspartate aminotransferase>40 IU/L or alanine aminotransferase >35 IU/L) who were prescribed a statin (n=342), patients with normal transaminases who were prescribed a statin (n=1437), and patients with elevated liver enzymes who were not prescribed a statin (n=2245). Patients with elevated liver enzymes at baseline had a higher incidence of mild/moderate and severe elevations after 6 months, whether or not they were prescribed a statin. Those with elevated liver enzymes at baseline who were prescribed a statin had a higher incidence of mild-moderate, but not severe, elevations at 6 months than those with normal transaminases who were prescribed a statin. Most patients in this study were prescribed atorvastatin or simvastatin (5 patients were prescribed fluvastatin); there was no difference in results according to the type of statin prescribed.

Harms of statin-thiazolidinediones combination. A recent nested, case-control study²⁸² evaluated the potential association between statin-thiazolidinedione combination and statins, thiazolidinediones, or other antidiabetic medications in patients with type 2 diabetes for muscle-related toxicities such as myopathy, myositis, rhabdomyolysis and myalgias. Of the 25 567 patients included in the analysis, about 5.7% of cases and 4.9% of controls were classified as having been ever exposed to statin-thiazolidinedione combination. Atorvastatin was the most commonly prescribed statin followed by simvastatin; rosiglitazone and pioglitazone were the thiazolidinediones under evaluation.

When compared with patients exposed to statin monotherapy, patients using statin-thiazolidinedione combination did not show an increased risk for muscle-related toxicities (adjusted odds ratio, 1.03; 95% CI, 0.83 to 1.26).

A different retrospective study reviewed the adverse events reported to the US Food and Drug Administration between 1990 and March 2002 in which simvastatin or atorvastatin was listed as a suspect in causing adverse events, and in which antidiabetic medications were listed as co-suspects or concomitant medications. Analysis was limited to adverse events affecting major organ systems (muscles, liver, pancreas, and bone marrow).²⁸³ Atorvastatin-associated adverse event reports were more likely to list concomitant thiazolidinediones compared with simvastatin-associated adverse event reports (3.6% compared with 1.6%, respectively; odds ratio, 2.3; 95% CI, 1.7 to 3.2; P<0.0001). Muscle toxicity was the most common adverse event, followed by liver-related events.

We also found one 24-week, placebo-controlled trial examining the effect of adding simvastatin to patients with type 2 diabetes who were taking a thiazolidinedione (pioglitazone or rosiglitazone).²⁸⁴ There were 2 cases of asymptomatic creatine phosphokinase elevations ≥10 times the upper limit of normal in the simvastatin group (1.7%), no elevations in alanine aminotransferase or aspartate aminotransferase, and no differences in tolerability between patients taking pioglitazone and those taking rosiglitazone.