The Effect of Early, Intensive Statin Therapy on Acute Coronary Syndrome a Meta-Analysis of Randomized Controlled Trials

Home , Colestyramine

REVIEW ARTICLE The Effect of Early, Intensive Statin Therapy on Acute Coronary Syndrome A Meta-analysis of Randomized Controlled Trials

Eddie Hulten, MD, MPH; Jeffrey L. Jackson, MD, MPH; Kevin Douglas, MD, MPH; Susan George, MD; Todd C. Villines, MD

Background: In addition to well-established second- stracted study quality, characteristics, and outcomes. ary prevention benefits for atherosclerotic coronary artery disease, 3-hydroxy-3-methylglutaryl coenzyme Data Synthesis: Thirteen randomized controlled trials A (HMG-CoA) reductase inhibitors (statins) are published before May 2006 were available, involving hypothesized to have short-term benefit in acute coro- 17 963 adults (median number of patients, 135; median nary syndrome (ACS), yet the data are inconsistent, with follow-up, 6 months). Early, intensive statin therapy for some trials underpowered to demonstrate therapeutic ACS decreased the rate of death and cardiovascular events benefit. Our objective was to determine the effects of over 2 years of follow-up (hazard ratio, 0.81 [95% con- Ͻ 2 early, intensive statin therapy for ACS. fidence interval, 0.77-0.87]) (Q3=58.54; P .001; I =95%). Survival curves revealed that this benefit begins to oc- Data Sources: Studies found in the PubMed, MEDLINE, cur between 4 and 12 months, achieving statistical sig- EMBASE, BIOSIS, SciSearch, PASCAL, and Interna- nificance by 12 months. There was no evidence of pub- tional Pharmaceutical Abstracts (IPA) databases and the lication bias, and sensitivity analyses did not identify a Cochrane Controlled Trials Register published between dominating study or study characteristic. January 1974 and May 2006. Conclusions: Early, intensive statin therapy reduces death and cardiovascular events after 4 months of treatment. Study Selection: Randomized controlled trials of stat- The validity of this finding would be strengthened by an ins begun within 14 days of hospitalization for ACS were analysis of individual patient data. included.

Data Extraction: Two investigators independently ab- Arch Intern Med. 2006;166:1814-1821

-HYDROXY-3-METHYLGLU- inflammation, antithrombogenicity, taryl coenzyme A (HMG- enhanced arterial compliance, blood pres- CoA) reductase inhibitors sure reduction, and modulation of endo- (statins) have been clearly thelial function.3,5-8 demonstrated to reduce mor- Our purpose was to systematically re- tality3 and morbidity among patients with view the literature to determine if early, in- cardiovascular disease. Less clear is whether tensive statin therapy reduces the risk of statins provide short-term benefit when death and adverse cardiovascular out- therapy is begun immediately during hos- comes for patients with ACS. pitalization among patients with acute coronary syndrome (ACS). Statins reduce car- METHODS diovascular morbidity and mortality to a Author Affiliations: degree beyond what is expected from re- Departments of Internal duction in low-density lipoprotein choles- LITERATURE SEARCH Medicine (Drs Hulten and terol (LDL-C) level alone.1-4 In addition to George), General Internal the benefits of improved cholesterol pa- We searched the PubMed, MEDLINE, Medicine (Drs Jackson and rameters (reduction in LDL-C level and in- EMBASE, BIOSIS, SciSearch, PASCAL, and In- Douglas), and Cardiology ternational Pharmaceutical Abstracts (IPA) da- (Dr Villines), Walter Reed Army crease in high-density lipoprotein choles- tabases and the Cochrane Controlled Trials Medical Center, Washington, terol level), statins may possess so-called Register for relevant articles published in any DC; and Uniformed Services pleiotropic effects that could affect out- language from January 1974 through Novem- University of the Health comes for patients with cardiovascular dis- ber 2005. We used the Medical Subject Head- Sciences, Bethesda, Md. ease, including plaque stabilization, anti- ings and text words simvastatin, atorvastatin,

(REPRINTED) ARCH INTERN MED/ VOL 166, SEP 25, 2006 WWW.ARCHINTERNMED.COM 1814

Downloaded From: https://jamanetwork.com/ on 10/01/2021 pravastatin, cerivastatin, fluvastatin, ro- were censored for other reasons for each suvastatin, pitavastatin, mevastatin, lo- arm in each trial was abstracted at the Simvastatin, Atorvastatin, Pravastatin, vastatin, acute coronary syndrome, and nonoverlapping time points: 1 month, Cerivastatin, Fluvastatin, Rosuvastatin, myocardial infarction. Search results were 4 months, 1 year, and 2 years. These time Pitavastatin, Mevastatin, Lovastatin, Acute Coronary Syndrome, limited to randomized controlled trials points were selected to maximize the Myocardial Infarction in adults (age Ͼ18 years). References of amount of data that was directly re- reviewed articles were also searched for ported, minimizing the need to calcu- relevant titles. late data from survival curves. From 524 Articles Identified these data, for each trial, an overall HR STUDY SELECTION was calculated by pooling the natural log of the HR using the inverse of the vari- 511 Articles Were Excluded ance as weights. Variance was calcu- 162 Duplicate Two reviewers independently con- 306 Not RCT ducted the literature search and extrac- lated based on the number of events and 5 Intervention Not Statin tion of relevant articles. The title and ab- patients at risk in each arm, separately 30 Wrong End Point stract of potentially relevant studies and for each time interval, as suggested by 3 Wrong Population 11 5 Statin Therapy Not Begun review articles were screened for appro- Parmar et al. The overall HR for each <14 Days priateness before retrieval of the full study was then pooled using the ran- articles. We included randomized con- dom effects of DerSimonian and Laird.12 13 Articles Accepted trolled trials in adults involving compari- Survival curves were created by calcu- son of early, intensive statin therapy with lating the event rate, for both arms, based some control arm within 14 days of hos- on the number of patients still eligible Figure 1. Flow diagram for exclusions of trials pitalization for ACS. We defined early to have the outcome during that period identified. RCT indicates randomized controlled statin therapy as being initiated within 14 and the number of events. Variance of trial. days of hospitalization for ACS. We de- these proportions was based on exact bi- fined intensive therapy as a medication nomial methods. These proportions were study at a time and then rerunning the regimen begun at a higher than usual dose pooled, at each time point, using the ran- analysis to assess the change in effect than recommended by routine treat- dom effects model of DerSimonian and size. All analyses were performed with ment following National Cholesterol Edu- Laird.12 Pooled survival curves were com- 9 Stata version 9.2 StataCorp (StataCorp, cation Panel (NCEP) guidelines. pared using log-rank methods. Assess- College Station, Tex). All P values were ment for publication bias was done us- 2 sided with an ␣ level of .05. We fol- ing the modified method of Macaskill et VALIDITY ASSESSMENT 13 lowed the QUORUM (the Quality of Re- al, based on linear regression using the porting Meta-analyses) guidelines for re- Two reviewers independently rated natural log of the HR as the dependent porting and discussing these meta- study quality using the Jadad instru- variable and the inverse of the total analytic results.17 sample size as the independent variable, ment for the assessment of the quality 14 of trial reports.10 The Jadad instrument as suggested by Peters et al. Heterogeneity was assessed by us- RESULTS is a point scale ranging from 0 to 8, with 2 15 2 points derived from the description of ing the I statistic. The I statistic pro- randomization, blinding, inclusion and vides an estimate of the amount of vari- LITERATURE SEARCH exclusion criteria, withdrawals, and ance due to heterogeneity rather than chance and is based on the traditional method of assessing adverse events. This meta-analysis included 17 963 measure of variance, the Cochrane Q sta- 18-30 tistic. We conducted a stratified analy- subjects from 13 trials (Figure 1). DATA ABSTRACTION sis to assess for potential confounders’ Trials that measured inflammatory contribution to heterogeneity, includ- markers or angiogram data with- We abstracted characteristics of the study ing study duration (greater than or less out clinical outcomes were ex- (author, year, country, design, dura- than or equal to the median duration); cluded because they did not meet tion, statin name and dosage, compara- change in LDL-C level (greater than or tor drug or placebo, time to initiation of our a priori search criteria. For ex- less than or equal to the median change ample, the REVERSAL (Reversal of statin therapy, duration of follow-up, and in LDL-C level); time to initiation of sample size); and patients (age, sex, and Atherosclerosis with Aggressive statin therapy; and study quality. Meta- Lipid Lowering) trial was excluded baseline and follow-up LDL-C levels). regression analysis was performed us- Outcomes abstracted included the com- ing the same independent variables.16 because it did not evaluate clinical bined primary end point from each trial Meta-regression (restricted maximum end points but instead measured ath- 31 in addition to the individual outcomes likelihood method) was used to calcu- erosclerotic plaque volumes. The of death, myocardial infarction (MI), and late the estimated between-study vari- Prevention of Ischemic Events by hospitalization for recurrent ischemia. ance (τ2 statistic) as a measure of the re- Early Treatment with Cerivastatin Two reviewers independently ab- sidual heterogeneity, having adjusted for (PRINCESS) trial was terminated stracted data, and disagreements were re- the covariates. The sensitivity analysis solved by consensus. early; we excluded it because it has of the effect of quality was based on a only been published in abstract component’s analysis in which each 32 QUANTITATIVE DATA form. The IDEAL (Incremental De- quality domain in the Jadad scale was se- crease in Endpoints through Aggres- SYNTHESIS rially tested to see if it had an effect on our study’s reported HRs. To exclude the sive Lipid lowering) trial was ex- Hazard ratios (HRs) were abstracted and possibility that any one study was excluded because patients were not pooled using the methods of Parmar et erting excessive influence on the re- enrolled within 14 days of hospital- al.11 In brief, the number of subjects at sults, we conducted a sensitivity analy- ization for ACS (average of 22 risk who had the event of interest or who sis by systematically excluding each months after MI).33

(REPRINTED) ARCH INTERN MED/ VOL 166, SEP 25, 2006 WWW.ARCHINTERNMED.COM 1815

Downloaded From: https://jamanetwork.com/ on 10/01/2021 Table 1. Characteristics of 13 Trials Meeting Criteria for Inclusion in the Meta-analysis

No. of Mean Prescription Follow-up Jadad Quality Source Country Patients Age, y Intervention Design Begun* Days Score Problem Cannon et al,19 2004 United 4162 58 Atorvastatin, 80 mg, Parallel 10 30, 90, 120, 730 7 Randomization, (PROVE Kingdom, vs pravastatin, blinding IT-TIMI22) United States, 40 mg Australia, Italy, France, Germany, Spain, and Canada Den Hartog et al,22 The Netherlands 99 63 Pravastatin, 40 mg, Parallel 2 90 7 Randomization 2001 (PAIS) vs placebo de Lemos et al,21 International 4497 61 Simvastatin, 40 mg Parallel 5 30, 120, 240, 730 7 Blinding 2004 (A to Z) for 1 mo, 80 mg thereafter, vs placebo for 4 mo, then simvastatin, 20 mg Liem et al,26 2002 The Netherlands 540 61 Fluvastatin, 80 mg, Parallel 14 42, 365 5 Blinding, (FLORIDA) vs placebo randomization Schwartz et al,28 Europe, 3086 65 Atorvastatin, 80 mg, Parallel 4 112 6 Adverse effects, 2001 (MIRACL) Australia, and vs placebo randomization, North/South blinding America Thompson et al, Australia 3408 61 Pravastatin, Parallel 1 30 8 None 200430 (PACT) 20/40 mg, vs placebo Arntz et al,18 2000 Germany 135 56 Pravastatin, Parallel 1 730 4 Randomization, (L-CAD) 20/40 mg, vs blinding, placebo with or adverse effects without niacin or cholestyramine Kesteloot et al,25 Belgium 69 NA Pravastatin, Parallel 2 180 3 Randomization, 1997 (LAMIL) 10/20 mg, vs blinding, placebo statistics, inclusion criteria Okazaki et al,27 2004 Japan 70 62 PCI ϩ atorvastatin, Parallel 1 180 6 Blinding (ESTABLISH) 20 mg, vs usual care Kayikcioglu et al,24 Turkey 77 51 Pravastatin, 40 mg, Parallel 1 180 4 Randomization, 2002 (PTT) vs placebo blinding Colivicchi et al,20 Italy 83 68 Atorvastatin, 80 mg, Parallel 12 360 5 Blinding 2002 vs usual care Serruys et al,29 2002 Europe, Canada, 1677 60 Fluvastatin, 80 mg, Parallel 2 1460 8 None (LIPS) and Brazil vs placebo Dupuis et al,23 2005 Canada 60 55 Pravastatin, 40 mg, Parallel 10 42 5 Randomization, (RECIFE) vs placebo blinding

Abbreviations: NA, not available; PCI, percutaneous coronary intervention. *Days following the hospitalization for acute coronary syndrome.

Eligible studies compared treat- atorvastatin, 80 mg (3 studies)19,20,28; ␬=0.84; PϽ.001). The median Jadad ment with a statin with placebo (9 atorvastatin, 20 mg (1 study)27; prava- score was 6 (range, 3-8). We achieved trials),18,22-26,28-30 placebo for 4 months statin, 40 mg (6 studies)9,18,22-25,30; flu- high agreement for Jadad score (qua- followed by a lower dose of statin (1 vastatin, 80 mg (2 studies)26,29; and dratic ␬=0.75; P=.02). All differ- trial),21 a lower dose of statin (1 trial),19 simvastatin, 80 mg (1 study).21 Time ences in agreement were resolved by or usual care per the discretion of the to initiation of statin therapy ranged consensus. Table 1 gives detailed inpatient’s physician (2 trials).20,27 Four from 1 to 14 days, with a median du- formation from individual studies. studies were international, multi- ration of 4 days. All studies mea- center trials; 1 was a multicentered sured death, recurrent MI, and hos- QUANTITATIVE DATA trial in the Netherlands, and the re- pital readmission for ACS as an a priori SYNTHESIS maining were single-center trials in primary outcome. The median dura- Australia, Germany, Belgium, Canada, tion of follow-up was 6 months The combined primary end point for Italy, Japan, Turkey, and the Nether- (range, 1-48). The mean age of par- all trials included death, recurrent lands. The median number of partici- ticipants was 60 years, and 76% of the ischemia, and recurrent MI. Six stud- pants in each study was 135 (range, participants were male. Agreement for ies included revascularization by per- 60-4497). Statins studied included study selection was good (Cohen cutaneous coronary intervention or

(REPRINTED) ARCH INTERN MED/ VOL 166, SEP 25, 2006 WWW.ARCHINTERNMED.COM 1816

Downloaded From: https://jamanetwork.com/ on 10/01/2021 Table 2. Pooled Hazard Ratios 1 Month (HRs) for Any Cardiovascular Event Thompson et al (PACT)30 Over 24 Months Den Hartog et al (PAIS)22 Schwartz et al (MIRACL)28 Liem et al (FLORIDA)26 Time, de Lemos et al (A-Z)21 mo HR (95% CI) Heterogeneity Arntz et al (LCAD)18 1 1.02 (0.95-1.09) Q = 64.8; Cannon et al (PROVE-IT)19 12 I 2 = 86.1% Dupuis et al (RECIFE)23 4 0.84 (0.72-1.02) Q = 591.6; Serruys et al (LIPS)29 10 I 2 = 98.5% Colvicchi et al20 Kayikcioglu et al (PTT)24 6 0.76 (0.70-0.84) Q6 = 79.4; 2 Okazaki et al (ESTABLISH)27 I = 92.4% Kesteloot et al (LAMIL)25 12 0.80 (0.76-0.84) Q5 = 78.7; 2 Subtotal 1.02 (0.95-1.09) I = 93.6% 24 0.81 (0.77-0.87) Q = 58.5; 4 Months 3 I 2 = 94.9% Den Hartog et al (PAIS)22 Pooled 0.84 (0.76-0.94) Q = 39.5; Schwartz et al (MIRACL)28 4 2 Liem et al (FLORIDA)26 HR I = 89.9% de Lemos et al (A-Z)21 Arntz et al (LCAD)18 Abbreviation: CI, confidence interval. Cannon et al (PROVE-IT)19 Serruys et al (LIPS)29 20 2 Colvicchi et al 0.87; Q4=0.55; I =0.0%) and ische- Kayikcioglu et al (PTT)24 mia (HR, 0.68; 95% CI, 0.50-0.92; 27 Okazaki et al (ESTABLISH) 2 Kesteloot et al (LAMIL)25 Q4=22.8; I =82.5%) but not in MIs Subtotal 0.84 (0.72-1.02) (HR, 0.89; 95% CI, 0.60-1.33; 2 6 Months Q4=15.92; I =74.9%). For both car- de Lemos et al (A-Z)21 diovascular deaths and ischemia, the Arntz et al (LCAD)18 benefit began to accrue after 4 months Cannon et al (PROVE-IT)19 of treatment, with relatively stable Serruys et al (LIPS)29 Colvicchi et al20 HRs thereafter (Figure 3). Kayikcioglu et al (PTT)24 The groups treated with early, in- Okazaki et al (ESTABLISH)27 tensive statin therapy (mean reduc- Subtotal 0.76 (0.70-0.84) tion in LDL-C level of 34±9 mg/dL 12 Months [0.88±0.23 mmol/L]) experienced 26 Liem et al (FLORIDA) significantly greater reduction in deLemos et al (A-Z)21 Arntz et al (LCAD)18 LDL-C level compared with controls Cannon et al (PROVE-IT)19 (6±12 mg/dL [0.16±0.31 mmol/L]; Serruys et al (LIPS)29 PϽ.001). 20 Colvicchi et al While there was evidence of het- Subtotal 0.80 (0.76-0.84) erogeneity, our analyses using both 24 Months de Lemos et al (A-Z)21 stratified and meta-regression meth- Arntz et al (LCAD)18 ods were unable to explain the Cannon et al (PROVE-IT)19 source of heterogeneity. Variables Serruys et al (LIPS)29 explored included percentage of Subtotal 0.81 ( 0.77-0.87) LDL-C level reduction, time to ini- 0 1 5 tiation of statin therapy, specific HR (95% CI) statin used, and study duration. In addition, we found no evidence of Figure 2. Forest plot of any cardiovascular event by duration of treatment. HR indicates hazard ratio; CI, any effect of study quality on out- confidence interval. comes, in particular whether the studies included allocation conceal- coronary artery bypass graft within 0.76; 95% CI, 0.70-0.84; Q6=79.4; ment and had adequate blinding. the primary end point,18,19,26-29 3 stud- I2=92.4%) (Table 2), a benefit that There was no evidence of publi- ies defined revascularization as a sec- persisted through 24 months (HR, cation bias either visually or statis- 21,22,30 ␹2 ondary end point, and 4 stud- 0.81; 95% CI, 0.77-0.87; Q6=58.54; tically (Peter’s test, P=.22). Ex- ies did not evaluate revascularization I2 =95%) (Figure 2). The overall clusion of any single study did not outcomes.20,23-25 pooled HR for the entire 24 months significantly alter our results. There was no reduction in over- was 0.84 (95% CI, 0.76-0.94) all cardiovascular events during (Table 2; Figure 3). SAFETY the first 4 months of treatment In a subgroup analysis (Table 3), (Figure 2). By the sixth month, over the 24 months of follow-up, Safety data showed comparable tol- there was a significant reduction in there was a reduction in cardiovas- erability for intensive statins and the overall cardiovascular events (HR, cular deaths (HR, 0.76; 95% CI, 0.66- control arm. Among the 17 963 pa-

(REPRINTED) ARCH INTERN MED/ VOL 166, SEP 25, 2006 WWW.ARCHINTERNMED.COM 1817

Downloaded From: https://jamanetwork.com/ on 10/01/2021 ploring the effects of statin use in 34 A High-Dose Statin B ACS. Similar to our findings, Briel 34 Usual Care 1.0 et al found no evidence of benefit through 4 months. This may not be 0.8 a surprising finding since 1 poten- 0.6 tial benefit of statins may be plaque 0.4 stabilization, which may take sev- 0.2 Event Free, % eral months to occur. Moreover, in 0 most studies there were relatively few events in the first few months after the initiation of treatment. Our study C D extends their analysis by using meta-

1.0 analytic survival methods to ex- 0.8 plore the effects of statins over a 0.6 longer period. Although there is evidence of sta- 0.4 tistical heterogeneity, we took mea- 0.2 Event Free, % sures to mitigate the impact of this 0 heterogeneity on our results. We used 0 5 10 15 20 25 0 5 10 15 20 25 a random effects model to pool HRs Duration, mo Duration, mo and conducted both stratified analyses and meta-regression to investi- Figure 3. Pooled survival curves by outcome. This figure demonstrates the pooled survival curves for any gate the heterogeneity, although this cardiovascular event (hazard ratio [HR], 0.84; 95% confidence interval [CI], 0.76-0.94) (A); myocardial infarction (HR, 0.89; 95% CI, 0.60-1.33) (B); ischemia (unstable angina or revascularization) (HR, 0.68; sensitivity analysis may be under- 95% CI, 0.50-0.92) (C); and cardiovascular death (HR, 0.76; 95% CI, 0.66-0.87) (D). powered with only 13 trials. The PROVE IT–TIMI 22 trial used 40-mg pravastatin as the control arm, and Table 3. Pooled Hazard Ratios (HRs) for Subgroups Over 24 Months the PACT trial used a relatively low dose of statin (20-mg pravastatin, HR (95% CI) then 40 mg), which may explain in Time, part why these studies did not have mo Myocardial Infarction Ischemia Cardiovascular Death as robust a hazard reduction as the 1 1.07 (0.94-1.21) 0.81 (0.69-0.94) 0.83 (0.58-1.18) other studies, which used a more in- 4 0.90 (0.56-1.44) 0.92 (0.49-1.70) 0.82 (0.49-1.36) 6 0.53 (0.22-1.26) 0.50 (0.44-0.58) 0.69 (0.07-6.80) tensive statin therapy than the PACT 12 1.79 (1.08-2.96) 0.52 (0.12-2.31) 0.64 (0.24-1.72) trial and a less intense control than 24 0.43 (0.24-0.78) 0.70 (0.51-0.97) 0.74 (0.63-0.86) the PROVE IT–TIMI 22 trial. Pooled 0.89 (0.60-1.33) 0.68 (0.50-0.92) 0.76 (0.66-0.87) There are limitations to conduct- HR ing a meta-analysis of these data. First, as described previously, there Abbreviation: CI, confidence interval. was evidence of significant statistical heterogeneity. It may not be ap- tients, only 3 cases of rhabdomyo- COMMENT propriate to pool results of interven- lysis were noted (all patients on tions with different doses or different high-dose simvastatin). There were This systematic review provides evi- types of statins (since statins are slightly higher rates of hepatitis in dence that early, intensive therapy known to have differing potency and the PROVE IT–TIMI 22 (Prava- with statins is associated with a re- effects). It is notable that 4 trials used statin or Atorvastatin Evaluation duction of adverse cardiovascular usual care or a low to moderate dose and Infection Therapy-Thromboly- outcomes, particularly cardiovascu- of statins for the control arm (be- sis in Myocardial Infarction 22)19 lar death, unstable angina, and re- cause of ethical considerations), (3.3% intensive group vs 1.1% con- vascularization when prescribed which may have reduced their power trol) and MIRACL (Myocardial within 14 days of hospitalization for to detect a benefit for early, inten- Ischemia Reduction with Aggres- ACS. These benefits took more than sive statin therapy compared with the sive Cholesterol Lowering)28 stud- 4 months to begin to accrue and other 9 trials that used a placebo. ies (2.5% vs 0.6%) but no differ- were sustained for 2 years. During Second, there are limited trials ence in the FLORIDA (Fluvastatin these 2 years, there was slightly less available. Having fewer trials re- on Risk Diminishment After Acute than a 20% reduction in the risk of duces the power to detect publica- Myocardial Infarction),26 PACT experiencing an adverse coronary tion bias and conduct stratified (Pravastatin in Acute Coronary event. There was no significant evi- analyses, which may affect some of Treatment),30 or PAIS (Pravastatin dence that reduction in LDL-C level our conclusions with regard to bias in Acute Ischaemic Syndromes)22 influenced these results. but not necessarily the pooled HRs. trials. Table 4 gives the rates of ad- Our findings are similar to and ex- Third, our results noted an over- verse effects from all 13 trials. tend on a recent meta-analysis ex- all reduction in all cardiovascular

(REPRINTED) ARCH INTERN MED/ VOL 166, SEP 25, 2006 WWW.ARCHINTERNMED.COM 1818

Downloaded From: https://jamanetwork.com/ on 10/01/2021 Table 4. Adverse Events Among Intensive Statin and Control Groups

Patients, No. (%)

No. of AST/ALT Myalgia or Discontinuation Source/Group Patients Ͼ3 ؋ Normal Myositis Rhabdomyolysis of Therapy Severe Adverse Events Schwartz et al,28 2001 (MIRACL) Treatment 1538 38 (2.5) 0 0 173 (11.3) 3 Treatment group patients Control 1548 9 (0.6) 0 0 160 (10.3) hospitalized for hepatitis Thompson et al, 200430 (PACT) Treatment 1710 7 (0.4) 0 0 NA None Control 1698 5 (0.3) 0 0 NA Cannon et al,19 2004 (PROVE IT- TIMI22) Treatment 2099 69 (3.3) 69 (3.3) 0 638 (30.4) None Control 2063 23 (1.1) 56 (2.7) 0 681 (33.0) Liem et al,26 2002 (FLORIDA) Treatment 265 NA NA NA 30 (11.3) None Control 275 NA NA NA 37 (13.5) de Lemos et al,21 2004 (A to Z) Treatment 2243 19 (0.8) 9 (0.4) 3 (0.1) 41 (1.8) Rhabdomyolysis in 3 Control 2210 8 (0.4) 1 (0.05) 0 34 (1.5) treatment group patients Den Hartog et al,22 2001 (PAIS) Treatment 50 0 2 (4) 0 3 (6) None Control 47 0 0 0 0 Arntz et al,18 2000 (L-CAD) Treatment 70 NA NA NA 22 (31.4) None Control 56 NA NA NA NA Kayikcioglu et al,24 2002 (PTT) Treatment 40 NA NA NA NA None Control 37 NA NA NA NA Okazaki et al,27 2004 (ESTABLISH) Treatment 35 NA NA NA 2 (5.7) None Control 35 NA NA NA 1 (2.9) Serruys et al,29 2002 (LIPS) Treatment 844 10 (1.2) 0 0 174 (20.6) None Control 833 3 (0.4) 3 (0.4) 0 196 (23.5) Kesteloot et al,25 1997 (LAMIL) Treatment 36 NA NA NA NA None Control 33 NA NA NA NA Dupuis et al,23 2005 (RECIFE) Treatment 30 NA NA NA NA None Control 30 NA NA NA NA Colivicchi et al20 Treatment 41 0 1 (2.4) 0 1 (2.4) None Control 40 0 0 0 0

Abbreviations: ALT, alanine aminotransferase; AST aspartate aminotransferase; NA, not applicable.

events and recurrent ischemia but not duction in MI due to random error scribed during the initial inpatient recurrent MI or cardiovascular death. or to the small number of trials avail- stay. In one study of medication com- This inconsistency was not antici- able for pooling. pliance, 77% of patients who began pated. We would have expected that Fourth, it would be better to con- treatment with statins as inpatients for statins to have benefits with re- duct a pooled analysis of the pa- were compliant with their dosage gard to all-cause mortality, the inter- tient data from each of these trials.35 regimens compared with only 40% vention should likewise have re- A meta-analysis based on data ab- of those who began treatment with duced the rate of recurrent MI. stracted from the literature has much a statin after hospitalization.36 Fur- Although the trials recorded event less ability to explore and explain the thermore, improved drug compli- rates for death, cardiovascular death, possible sources of heterogeneity ance is associated with improved strokes, and numerous other out- compared with one based on indi- cholesterol parameters and re- comes such as reperfusion and by- vidual patient data. We strongly rec- duced mortality among cardiovas- pass surgery, it is possible that stat- ommend that one be conducted. cular patients.37 Although there were ins may have had some effect not It is worth noting that studies insufficient data to investigate com- tested for in this or other studies that have demonstrated that patients are pliance in this analysis, this is an im- benefited mortality. It is also pos- more likely to continue taking medi- portant consideration for any treat- sible that there was no significant re- cations after hospital discharge if pre- ment regimen.

(REPRINTED) ARCH INTERN MED/ VOL 166, SEP 25, 2006 WWW.ARCHINTERNMED.COM 1819

Downloaded From: https://jamanetwork.com/ on 10/01/2021 Use of intensive statin therapy is months to begin to accrue. Addi- 4. Scandinavian Simvastatin Survival Study Group. often avoided by clinicians owing to tional trials are ongoing, including Randomized trial of cholesterol lowering in 4444 patients with coronary heart disease: The Scan- fear of increased adverse events due the FACS (Fluvastatin for Acute dinavian Survival Study (4S). Lancet. 1994;344: 42 to the higher statin dose or, in pa- Coronary Syndrome), LUNAR 1383-1389. tients with only mild elevations in (Limiting Undertreatment of lipids 5. Vaughan CJ, Gotto AJ Jr, Basson CT. The evolv- LDL-C level, of driving LDL-C level in ACS with Rosuvastatin),43 and ing role of statins in the management of athero- below a theoretical safe value. Our JAPAN-ACS (Japan Assessment of sclerosis. J Am Coll Cardiol. 2000;35:1-10. 6. Maron DJ, Fazio S, Linton MF. Current perspec- study showed that intensive statin Pitavastatin and Atorvastatin in tives on statins. Circulation. 2000;101:207-213. 44 therapy and controls experienced Acute Coronary Syndrome) trials. 7. Glorioso N, Troffa C, Filigheddu F, et al. Effect of comparable rates of hepatitis, myo- In summary, available evidence the HMG-CoA reductase inhibitors on blood pres- sitis, and rhabdomyolysis. Serious ad- supports early, intensive statin sure in patients with essential hypertension and primary hypercholesterolemia. Hypertension. verse events were rare (Table 4). A re- therapy for patients with ACS. While 1999;34:1281-1286. cent analysis of 49 completed trials the benefit may take up to 6 months 8. Ferrier KE, Muhlmann MH, Baguet JP, et al. In- using atorvastatin demonstrated that to begin to accrue, our analysis sug- tensive cholesterol reduction lowers blood pres- 80-mg atorvastatin, when compared gests that there may be a stable, 20% sure and large artery stiffness in isolated systolic with 10 mg, is safe and well toler- reduction in the rate of cardiovascu- hypertension. J Am Coll Cardiol. 2002;39:1020- 38 1025. ated. In addition, in an analysis of lar events over at least 2 years. There 9. Expert Panel on Detection, Evaluation, and Treat- the PROVE IT–TIMI 22 trial, there is no significant evidence that reduc- ment of High Blood Cholesterol in Adults. was no adverse effect on safety with tion in LDL-C level explains these Executive Summary of the Third Report of the Na- lower achieved LDL-C level (Ͻ40 beneficial effects. Our finding of ben- tional Cholesterol Education Program (NCEP) Ex- mg/dL [Ͻ1.04 mmol/L]). In fact, pa- efits beyond reduction of LDL-C level pert Panel on Detection. Evaluation, and Treat- ment of High Blood Cholesterol in Adults (Adult tients who achieved LDL-C values of are also consistent with the findings Treatment Panel III). JAMA. 2001;285:2486- 33 40 to 60 mg/dL (1.04-1.55 mmol/L) from the IDEAL trial. The dosing 2497. and lower than 40 mg/dL (Ͻ1.04 regimen with the most evidence for 10. Jadad AR, Moore RA, Carroll D, et al. Assessing mmol/L) had fewer major cardiac beneficial effects to date is 80-mg ator- the quality of reports of randomized clinical trials: events compared with all other vastatin, begun within 14 days of hos- is blinding necessary? Control Clin Trials. 1996; 39 17:1-12. groups. The IDEAL trial studied in- pitalization for ACS. Our analysis was 11. Parmar MK, Torri V, Stewart L. Extracting sum- tensive statin therapy in 8888 outpa- limited in its ability to explore the mary statistics to perform meta-analyses of the tients with a history of MI and noted sources of heterogeneity in these data. published literature for survival endpoints. Stat no difference in adverse effect rates for We recommend that a pooled analy- Med. 1998;17:2815-2834. 33 12. DerSimonian R, Laird N. Meta-analysis in clini- intensive vs low-dose statin therapy. sis using patient-level data be per- cal trials. Control Clin Trials. 1986;7:177-188. These safety data, combined with re- formed as soon as possible. 13. Macaskill P, Walter SD, Irwig L. A comparison of cent biological data highlighting re- methods to detect publication bias in meta- duction in atherosclerotic progres- Accepted for Publication: May 31, analysis. Stat Med. 2001;20:641-654. sion and highly sensitive C-reactive 14. Peters JL, Sutton AJ, Jones DR, Abrams KR, Rush- 2006. ton L. Comparison of two methods to detect pub- protein level in patients receiving Correspondence: Eddie Hulten, lication bias in meta-analysis. JAMA. 2006;295: high-dose statins, make intensive MD, MPH, 6900 Georgia Ave NW, 676-680. statin therapy compelling and reas- Washington, DC 20307 (edward 15. Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in meta-analyses. BMJ. suring to health care providers car- [email protected]). ing for patients with ACS.40 2003;327:557-560. Financial Disclosure: None re- 16. Thompson SG, Sharp SJ. Explaining heterogene- Current American College of Car- ported. ity in meta-analysis: a comparison of methods. Stat diology guidelines recommend the Disclaimer: The views expressed Med. 1999;18:2693-2708. following: herein are those of the authors only 17. Moher D, Cook D, Eastwood S, et al. Improving the quality of reports of meta-analyses of ran- v Class I statins for patients with and are not to be construed as those domised controlled trials: the QUOROM statement. ACS with an LDL-C level greater of the Department of the Army or the Lancet. 1999;354:1896-1900. than 130 mg/dL (Ͼ3.37 mmol/L); Department of Defense. 18. Arntz HR, Agrawal R, Wunderlich W, et al. Ben- ϩ v Class IIa statins for patients eficial effects of pravastatin ( /− colestyramine/ niacin) initiated immediately after a coronary event with ACS with an LDL-C level REFERENCES (the randomized Lipid-Coronary Artery Disease greater than 100 mg/dL (Ͼ2.59 [L-CAD] Study). Am J Cardiol. 2000;86:1293- mmol/L). 1. Sacks FM, Gibson CM, Rosner B, Pasternak RC, 1298. Stone PH; Harvard Atherosclerosis Reversibility 19. Cannon CP, Braunwald E, McCabe CH, et al; Prava- Project Research Group. The influence of pre- statin or Atorvastatin Evaluation and Infection The guidelines make note of the ben- treatment low density lipoprotein cholesterol Therapy-Thrombolysis in Myocardial Infarction 22 eficial outcomes of the MIRACL28 concentrations on the effect of hypercholesterol- Investigators. Intensive versus moderate lipid low- emic therapy on coronary atherosclerosis in ering with statins after acute coronary syndromes. and L-CAD (Lipid-Coronary Ar- N Engl J Med. 2004;350:1495-1504. 18 angiographic trials. Am J Cardiol. 1995;76:78C- tery Disease) trials and the good 85C. 20. Colivicchi F, Guido V, Tubaro M, et al. Effects of evidence for safety but make no uni- 2. Gaw A. Can the clinical efficacy of the HMG CoA atorvastatin 80 mg daily early after onset of versal recommendation for early, in- reductase inhibitors be explained solely by their unstable angina pectoris or non-Q-wave myo- tensive statin therapy.41 Our meta- effects on LDL-cholesterol? Atherosclerosis. 1996; cardial infarction. Am J Cardiol. 2002;90:872- 125:267-269. 874. analysis suggests that there might be 3. Davignon J. Beneficial cardiovascular pleiotropic 21. de Lemos JA, Blazing MA, Wiviott SD, et al; A to a role for early intensive therapy, effects of statins. Circulation. 2004;109(23) Z Investigators. Early intensive vs a delayed con- though the benefit will take several (suppl 1):III39-III43. servative simvastatin strategy in patients with acute

(REPRINTED) ARCH INTERN MED/ VOL 166, SEP 25, 2006 WWW.ARCHINTERNMED.COM 1820

Downloaded From: https://jamanetwork.com/ on 10/01/2021 coronary syndromes: phase Z of the A to Z trial. tors. Fluvastatin for prevention of cardiac events 37. Lappe JM, Muhlestein JB, Lappe DL, et al. Im- JAMA. 2004;292:1307-1316. following successful first percutaneous coro- provements in 1-year cardiovascular clinical out- 22. Den Hartog FR, Van Kalmthout PM, Van Loenhout nary intervention: a randomized controlled trial. comes associated with a hospital-based dis- TT, Schaafsma HJ, Rila H, Verheugt FW. Prava- JAMA. 2002;287:3215-3222. charge medication program. Ann Intern Med. 2004; statin in acute ischaemic syndromes: results of a 30. Thompson PL, Meredith I, Amerena J, Campbell 141:446-453. randomised placebo-controlled trial. Int J Clin Pract. TJ, Sloman JG, Harris PJ; Pravastatin in Acute 38. Newman C, Tsai J, Szarek M, Luo D, Gibson E. 2001;55:300-304. Coronary Treatment (PACT) Investigators. Comparative safety of atorvastatin 80 mg versus 10 23. Dupuis J, Tardif JC, Rouleau JL, et al. Intensity Effect of pravastatin compared with placebo ini- mg derived from analysis of 49 completed trials in of lipid lowering with statins and brachial artery tiated within 24 hours of onset of acute myocar- 14,236 patients. Am J Cardiol. 2006;97:61-67. vascular endothelium reactivity after acute coro- dial infarction or unstable angina: the Pravasta- 39. Wiviott SD, Cannon CP, Morrow DA, Ray KK, Pfef- nary syndromes (from the BRAVER trial). Am J tin in Acute Coronary Treatment (PACT) trial. Am fer MA, Braunwald E. Can low-density lipopro- Cardiol. 2005;96:1207-1213. Heart J. 2004;148:e2. tein be too low? the safety and efficacy of achiev- 24. Kayikcioglu M, Can L, Kultursay H, Payzin S, Turko- 31. Nissen SETE, Schoenhagen P, Brown BG, et al; glu C. Early use of pravastatin in patients with acute REVERSAL Investigators. Effect of intensive com- ing very low low-density lipoprotein with intensive myocardial infarction undergoing coronary pared with moderate lipid-lowering therapy on pro- statin therapy: a PROVE IT–TIMI 22 substudy. angioplasty. Acta Cardiol. 2002;57:295-302. gression of coronary atherosclerosis: a random- J Am Coll Cardiol. 2005;46:1411-1416. 25. Kesteloot H, Claeys G, Blanckaert N, Lesaffre E. ized controlled trial. JAMA. 2004;291:1071-1080. 40. Nissen SETE, Schoenhagen P, Crowe T, et al; Re- Time course of serum lipids and apolipoproteins 32. Wright RS. The prevention of ischemic events by versal of Atherosclerosis with Aggressive Lipid after acute myocardial infarction: modification by early treatment with cerivastatin (PRINCESS) Lowering (REVERSAL) Investigators. Statin pravastatin. Acta Cardiol. 1997;52:107-116. study. Paper presented at: the Annual Congress therapy, LDL cholesterol, C-reactive protein, and 26. Liem AH, van Boven AJ, Veeger NJ, et al; Fluva- of the European Society of Cardiology; August 28– coronary artery disease. N Engl J Med. 2005; statin on Risk Diminishment After Acute Myocar- September 1, 2004; Munich, Germany. 352:29-38. dial Infarction Study Group. Effect of fluvastatin 33. Pedersen TR, Faergeman O, Kastelein JJ, et al; In- 41. Braunwald E. Management of patients with un- on ischaemia following acute myocardial infarc- cremental Decrease in End Points Through Ag- stable angina and non-ST segment elevation myo- tion: a randomized trial. Eur Heart J. 2002;23: gressive Lipid Lowering (IDEAL) Study Group. cardial infarction update. http://www.acc.org 1931-1937. High-dose atorvastatin vs usual-dose simvasta- /clinical/guidelines/unstable/incorporated/index 27. Okazaki S, Yokoyama T, Miyauchi K, et al. Early tin for secondary prevention after myocardial in- .htm. Accessed February 10, 2006. statin treatment in patients with acute coronary farction: the IDEAL study: a randomized con- 42. Ostadal P, Alan D, Hajek P, et al. Fluvastatin in the syndrome: demonstration of the beneficial effect trolled trial. JAMA. 2005;294:2437-2445. therapy of acute coronary syndrome: rationale and on atherosclerotic lesions by serial volumetric in- 34. Briel M, Schwartz GG, Thompson PL, et al. Ef- design of a multicenter, randomized, double- travascular ultrasound analysis during half a year fects of early treatment with statins on short- blind, placebo-controlled trial (The FACS Trial) after coronary event: the ESTABLISH Study. term clinical outcomes in acute coronary syn- [ISRCTN81331696]. Curr Control Trials Cardio- Circulation. 2004;110:1061-1068. dromes: a meta-analysis of randomized controlled 28. Schwartz GG, Olsson AG, Ezekowitz MD, et al; Myo- trials. JAMA. 2006;295:2046-2056. vasc Med. 2005;6:4. doi:10.1186/1468-6708-6-4. cardial Ischemia Reduction with Aggressive Cho- 35. Stewart LA, Parmar MK. Meta-analysis of the lit- 43. LUNAR IIIb Study Comparing Rosuvastatin and lesterol Lowering (MIRACL) Study Investiga- erature or of individual patient data: is there a Atorvastatin in Subjects With Acute Coronary tors. Effects of atorvastatin on early recurrent difference? Lancet. 1993;341:418-422. Syndromes. http://www.clinicaltrials.gov/ct/show ischemic events in acute coronary syndromes: the 36. Muhlestein JB, Horne BD, Bair TL, et al. Usefulness /NCT00214630. Accessed January 31, 2006. MIRACL study: a randomized controlled trial. of in-hospital prescription of statin agents after an- 44. JAPAN-ACS ( Japan Assessment of Pitavastatin and JAMA. 2001;285:1711-1718. giographic diagnosis of coronary artery disease in Atorvastatin in Acute Coronary Syndrome). http: 29. Serruys PW, de Feyter P, Macaya C, et al; Lescol improving continued compliance and reduced //www.clinicaltrials.gov/ct/show/NCT00242944 Intervention Prevention Study (LIPS) Investiga- mortality. Am J Cardiol. 2001;87:257-261. ?order=1. Accessed January 31, 2006.

(REPRINTED) ARCH INTERN MED/ VOL 166, SEP 25, 2006 WWW.ARCHINTERNMED.COM 1821

Downloaded From: https://jamanetwork.com/ on 10/01/2021