Multiple randomized controlled trials (RCTs) have established the efficacy of statins for the prevention of cardiovascular disease. The benefits observed are often framed in terms of percentage reductions in low-density lipoprotein (LDL) cholesterol from baseline or percentage reductions between control and treatment groups, although epidemiologic data suggest that the absolute intergroup difference in LDL cholesterol (ΔLDL Control-Rx ) is the more informative measure. A systematic review of large-scale trials of statins versus placebo, usual care, or active (lower dose statin) control was conducted to calculate updated summary estimates of risk reduction in coronary artery disease and all-cause mortality. Meta–regression analysis was used to ascertain the relations of different LDL cholesterol metrics to outcomes. In 20 eligible RCTs, there were significant overall reductions for coronary artery disease (odds ratio 0.72, 95% confidence interval 0.67 to 0.78) and mortality (odds ratio 0.89, 95% confidence interval 0.84 to 0.94), but with substantial variability in trial results. ΔLDL Control-Rx was the strongest determinant of coronary artery disease risk reduction, particularly after excluding active-comparator studies, and was independent of baseline LDL cholesterol. In contrast, baseline LDL cholesterol edged out ΔLDL Control-Rx as the strongest determinant of mortality, but neither was significant after the exclusion of active-comparator studies. The exclusion of 3 RCTs involving distinct populations, however, rendered ΔLDL Control-Rx the predominant determinant of mortality reduction. In conclusion, these findings underscore the primacy of absolute reductions in LDL cholesterol in the design and interpretation of RCTs of lipid-lowering therapies and in framing treatment recommendations on the basis of the proved coronary benefits of these drugs.
A multitude of randomized controlled trials (RCT) have documented the efficacy of statins in reducing coronary events in diverse populations. These benefits are well established in RCTs comparing statins against placebo or usual care, with accumulating data suggesting that this is also true for high- versus low-intensity statin treatment. Demonstration of statin-related benefits versus control has not been uniform, however, even in such megatrials. A potential explanation cited for null results is that despite major percentage reductions in low-density lipoprotein (LDL) cholesterol from baseline in the active groups, the percentage differences in LDL cholesterol achieved between study arms have been more modest. Yet assessments of the benefits of lipid-lowering agents, whether in RCTs or in meta-analyses, have often been framed in terms of percentage cholesterol reductions associated with active treatment. The construct of percentage cholesterol lowering from baseline has similarly been used in national treatment guidelines. Focus on percentage, rather than absolute, reductions is itself problematic, however, because available epidemiologic evidence shows that the relation between cholesterol level and cardiovascular risk is curvilinear. Hence, for any absolute change in serum cholesterol irrespective of its starting clinical value, the corresponding change in the relative risk for coronary artery disease (CAD) is constant. Consequently, percentage cholesterol reductions, whether longitudinally in the intervention arm or between study arms, may not be the most apt metric for describing statin-related benefits. We sought to define in a comparative fashion the relation between various measures of LDL cholesterol lowering and outcomes using meta-regression in an updated systematic review of large-scale statin trials.
Methods
We conducted a systematic search of the English-language published research using PubMed, Embase, BIOSIS, the Web of Science, Cochrane Systematic Reviews, the Database of Abstracts of Reviews of Effects, the Central Register of Controlled Trials, and ClinicalTrials.gov from January 1994 to December 2008 to identify pertinent studies. Nineteen relevant keywords were entered, and clinical trials, qualitative and quantitative reviews, and editorials thus identified were reviewed to select relevant studies. Reference sections of reports served to identify additional studies, as did knowledge of experts in the field (CJV, AMG, and RCP).
We focused our systematic review on large-scale RCTs of statins versus placebo, usual care, or active comparators having clinical cardiovascular events or all-cause mortality as their primary end points. Eligible RCTs required a minimum enrollment of 1,000 participants at risk for, or with stable, CAD and follow-up duration >1 year. We excluded a priori clinical trials that evaluated statins in combination with other therapies, that were designed to assess impacts on intermediate primary end points, or that focused on distinct clinical populations or acute settings (which could introduce additional heterogeneity), namely, patients with acute coronary syndromes or advanced renal disease or those who underwent cardiac catheterization.
After the selection of RCTs meeting the inclusion criteria, a single investigator (JRK) extracted data from individual studies. The 2 end points of interest were the primary or secondary outcomes of major CAD events and all-cause mortality. Major CAD events included nonfatal myocardial infarction and fatal CAD, as defined in individual RCTs. When resuscitated cardiac arrest formed part of an RCT’s major CAD end point and could not be clearly excluded from the composite CAD outcome on the basis of the information presented, it was included in the analysis. If fatal CAD was not reported, fatal myocardial infarction was used instead. Mean or median LDL cholesterol at baseline and after allocation to treatment or control was abstracted directly or computed from available information. If the mean in-trial follow-up LDL cholesterol in the placebo group was not provided, it was imputed from the baseline value.
Five measures of LDL cholesterol were examined: (1) baseline LDL cholesterol in the treatment group (baseline LDL Rx ), (2) absolute change in LDL cholesterol in the treatment group (ΔLDL Baseline-Final ), (3) percentage change in LDL cholesterol in the treatment group (%ΔLDL Baseline-Final ), (4) absolute difference between achieved in-trial LDL cholesterol in the control versus treatment group (ΔLDL Control-Rx ), and (5) percentage difference between achieved LDL cholesterol in the control versus treatment group (% ΔLDL Control-Rx ).
Stata version 10.0 (StataCorp LP, College Station, Texas), was used in all analyses. Separate analysis of RCTs not having active comparators was planned a priori. For individual trials, the numbers of participants with and without events in the treatment and control arms at the conclusions of the studies were recorded. When ascertainment of follow-up LDL cholesterol was performed before the conclusion of a trial, events reported as of the time of such measurements were used in the analyses. In view of the presence of significant heterogeneity (Cochran’s Q test), a random-effects approach (DerSimonian and Laird) was used to calculate pooled odds ratios (ORs). The quantity I 2 , the component of heterogeneity not attributable to chance, was also computed. Because only mega-trials were included, an approach that avoids potential biases associated with smaller studies, assessment for publication bias was not undertaken.
To investigate the observed study heterogeneity, random-effects meta-regression of RCT results was performed against different measures of LDL cholesterol or trial duration. Sensitivity analysis of ORs, I 2 values, and meta-regression coefficients entailed the serial exclusion of studies or groups of studies from consideration.
Results
The search yielded 31 candidate RCTs meeting the sample size and duration criteria, of which 11 did not meet additional inclusion criteria (4 for acute coronary syndromes or enrollment after cardiac catheterization, 2 for assessing drug combinations, 2 focusing on end-stage renal disease or renal transplantation, 1 with an intermediate primary end point, 1 for failed randomization, and 1 for incompleteness of end point ascertainment). The characteristics of the 20 large-scale trials selected for inclusion are listed in Table 1 . There were a total of 155,255 participants, of whom 11,508 experienced major CAD events, and 13,687 had all-cause fatal events. With the exclusion of the 3 active-comparator trials, there were 124,302 patients, with 8,332 and 10,448 developing major CAD and fatal events, respectively.
| Trial | Dates | n | Treatment Arm (mg/d) | Control Arm (mg/d) | Double Blinded | Baseline LDL-C (mg/dl) ⁎ | ΔLDL Baseline-Final (mg/dl) | ΔLDL Control-Rx (mg/dl) | Cumulative CAD Incidence: Controls | Follow-Up (mo) |
|---|---|---|---|---|---|---|---|---|---|---|
| 4S | 1988–1994 | 4,444 | Simvastatin 20 † | Placebo | + | 188 | 66 (35%) | 68 (36%) | 27.3% (5.1%/yr) | 65 |
| WOSCOPS | 1989–1995 | 6,595 | Pravastatin 40 | Placebo | + | 192 | 50 (26%) | 50 (26%) | 7.9% (1.6%/yr) | 59 |
| CARE | 1989–1996 | 4,159 | Pravastatin 40 | Placebo | + | 139 | 42 (30%) | 42 (30%) | 13.2% (2.4%/yr) | 60 |
| LIPID | 1990–1997 | 9,014 | Pravastatin 40 | Placebo | + | 150 | 38 (25%) | 38 (25%) | 15.9% (2.6%/yr) | 73 |
| AFCAPS/TexCAPS | 1990–1997 | 6,605 | Lovastatin 20 † | Placebo | + | 150 | 35 (23%) | 41 (26%) | 3.1% (0.6%/yr) | 62 |
| GISSI-P | 1993–1996 | 4,271 | Pravastatin 20 † | Usual care | 0 | 152 | 23 (15%) | 18 (12%) | 3.9% (2.0%/yr) | 23 |
| HPS | 1994–2001 | 20,536 | Simvastatin 40 | Placebo | + | 131 | 54 (41%) | 38 (32%) | 11.8% (2.4%/yr) | 60 |
| ALLHAT-LLT | 1994–2002 | 10,355 | Pravastatin 40 | Usual care | 0 | 146 | 42 (29%) | 17 (14%) | 10.4% (2.2%/yr) | 58 |
| PROSPER | 1997–2001 | 5,804 | Pravastatin 40 | Placebo | + | 147 | 50 (34%) | 50 (34%) | 12.2% (3.8%/yr) | 38 |
| ASCOT-LLA | 1998–2002 | 10,305 | Atorvastatin 10 | Placebo | + | 132 | 42 (32%) | 37 (29%) | 3.0% (0.9%/yr) | 40 |
| GREACE | 1998–2001 | 1,600 | Atorvastatin10 † | Usual care | 0 | 179 | 83 (46%) | 72 (43%) | 11.2% (3.7%/yr) | 36 |
| CARDS | 1997–2003 | 2,838 | Atorvastatin10 | Placebo | + | 117 | 36 (31%) | 39 (32%) | 5.5% (1.4%/yr) | 47 |
| TNT | 1998–2005 | 10,001 | Atorvastatin 80 | Atorvastatin 10 | + | 152 | 75 (49%) | 24 (24%) | 8.3% (1.7%/yr) | 59 |
| IDEAL | 1999–2005 | 8,888 | Atorvastatin 80 | Simvastatin 20–40 | 0 | 122 | 42 (34%) | 20 (20%) | 23.8% (5.0%/yr) | 58 |
| SPARCL | 1998–2005 | 4,731 | Atorvastatin 80 | Placebo | + | 133 | 60 (45%) | 56 (43%) | 5.1% (1.0%/yr) | 59 |
| MEGA | 1994–2004 | 7,832 | Pravastatin 10 † + diet | Diet/usual care | 0 | 157 | 29 (18%) | 23 (15%) | 1.1% (0.2%/yr) | 64 |
| ASPEN | 1996–2003 | 2,410 | Atorvastatin 10 | Placebo | + | 114 | 34 (30%) | 34 (30%) | 4.0% (1.0%/yr) | 48 |
| CORONA | 2003–2007 | 5,001 | Rosuvastatin 10 | Placebo | + | 137 | 61 (45%) | 62 (45%) | 6.0% (2.2%/yr) | 33 |
| JUPITER | 2003–2008 | 17,802 | Rosuvastatin 20 | Placebo | + | 108 | 53 (49%) | 54 (50%) | 0.8% (0.4%/yr) | 23 |
| SEARCH | NA | 12,064 | Simvastatin 80 | Simvastatin 20 | + | 97 | 11 (11%) | 11 (11%) | 13.4% (2.0%/yr) | 80 |
⁎ All data are expressed as means except for LIPID, for which medians are reported.
Analysis of all 20 RCTs showed significant overall reductions in the risk for CAD (OR 0.72, 95% confidence interval [CI] 0.67 to 0.78) and mortality (OR 0.89, 95% CI 0.84 to 0.94), but this was associated with significant underlying heterogeneity for the 2 outcomes (p ≤0.005). The lack of consistency in study findings was not attributable to chance, with moderate to high (I 2 = 69.7%) or moderate (I 2 = 51.1%) variability in effect estimates across RCTs. Figure 1 shows the corresponding findings for the analysis focused on the 17 RCTs of statins versus placebo or usual care. Significant risk reductions were again achieved for CAD and mortality, with less, though still significant, heterogeneity for the 2 outcomes.
The results of meta-regression modeling of risk reduction as a function of various measures of LDL cholesterol or trial duration are listed in Table 2 . Taking all trials into account, ΔLDL Control-Rx exhibited the strongest inverse association (highest negative standardized regression coefficient, β) with relative reduction in CAD events, whereas that for the percentage difference (%ΔLDL Control-Rx ) was minimally weaker. According to the random-effects model, every 39 mg/dl (1 mmol/L) intergroup difference in achieved LDL cholesterol was associated with a 25% relative reduction in CAD risk (OR 0.75, 95% CI 0.68 to 0.82). There were significant but weaker associations for baseline LDL Rx , and absolute (although not percentage) reduction in LDL cholesterol in the treatment arm (ΔLDL Baseline-Final ), whereas trial duration bore no significant relation to CAD risk reduction.
| Variable | RCTs Without Active Comparators | All RCTs | ||
|---|---|---|---|---|
| β per SD † Increase | p Value | β per SD ‡ Increase | p Value | |
| Coronary artery disease | ||||
| lnOR CAD | ||||
| Baseline LDL Rx | −0.045 | 0.320 | −0.083 | 0.014 |
| ΔLDL Baseline-Final | −0.091 | 0.052 | −0.082 | 0.029 |
| %ΔLDL Baseline-Final | −0.063 | 0.221 | −0.062 | 0.143 |
| ΔLDL Control-Rx | −0.108 | <0.001 | −0.129 | <0.001 |
| %ΔLDL Control-Rx | −0.095 | 0.030 | −0.127 | <0.001 |
| Trial duration | 0.011 | 0.813 | 0.052 | 0.237 |
| All-cause mortality | ||||
| lnOR Mortality | ||||
| Baseline LDL Rx | −0.065 | 0.081 | −0.067 | 0.022 |
| ΔLDL Baseline-Final | −0.018 | 0.686 | −0.021 | 0.511 |
| %ΔLDL Baseline-Final | 0.020 | 0.581 | 0.002 | 0.941 |
| ΔLDL Control-Rx | −0.030 | 0.397 | −0.055 | 0.041 |
| %ΔLDL Control-Rx | 0 | 0.993 | −0.030 | 0.297 |
| Trial duration | −0.018 | 0.585 | 0.012 | 0.702 |
⁎ Models are univariate (i.e., the β coefficients refer to the relation with each measure as a single independent variable).
† SDs (RCTs without active comparators): baseline LDL Rx = 24 mg/dl, ΔLDL Baseline-Final = 15 mg/dl, %ΔLDL Baseline-Final = 10%, = ΔLDL Control-Rx = 16 mg/dl, %ΔLDL Control-Rx = 11%, trial duration = 15 months.
‡ SDs (all RCTs): baseline LDL Rx = 25 mg/dl, ΔLDL Baseline-Final = 17 mg/dl, %ΔLDL Baseline-Final = 11%, ΔLDL Control-Rx = 17 mg/dl, %ΔLDL Control-Rx = 11%, trial duration = 23 months.
When active-comparator trials were excluded, the absolute intergroup difference in LDL cholesterol (ΔLDL Control-Rx ) again showed the strongest association to CAD risk reduction, followed by %ΔLDL Control-Rx ( Table 2 ). The associations between these intergroup differences, whether in absolute or percentage terms, and CAD risk decreased with the exclusion of active-comparator trials. Figure 2 shows that on the basis of the model, there was a 23% relative reduction in CAD risk for every 39 mg/dl absolute intergroup decrease in LDL cholesterol (OR 0.77, 95% CI 0.66 to 0.89). Compared to ΔLDL Control-Rx , ΔLDL Baseline-Final was more modest and fell just short of significance. No other variables were significantly related to CAD risk reduction.
Regarding all-cause mortality, analysis of all 20 RCTs showed significant relations only for ΔLDL Control-Rx and baseline LDL Rx , but it was baseline LDL Rx that had the higher negative regression coefficient. Accordingly, for every 39 mg/dl higher value for baseline LDL cholesterol, there was an associated 10% reduction in the relative risk for death (OR 0.90, 95% CI 0.83 to 0.99). The corresponding reduction for every 39 mg/dl of ΔLDL Control-Rx was 12% (OR 0.88, 95% CI 0.78 to 0.99). Notably, analyses of the 17-RCT subset revealed no significant associations for any of the independent measures ( Figure 2 , Table 2 ).
The association between ΔLDL Control-Rx and the relative risk for CAD was not meaningfully altered by adjustment for baseline LDL Rx , whether or not active-comparator trials were taken into account (data not shown). With respect to mortality, the inclusion of ΔLDL Control-Rx and baseline LDL Rx in the full meta-regression model rendered the 2 variables nonsignificant (p ≥0.116).
Because the exclusion of active-comparator RCTs was found to reduce heterogeneity, sensitivity analyses focused on the 17 trials of statin versus placebo or usual care. The exclusion of each of these RCTs individually did not materially influence estimates of risk reduction for CAD or mortality (data not shown). Selected individual and multiple exclusions based in part on outliers in meta–regression analysis ( Figure 2 ) are listed in Table 3 .
| RCTs | Summary OR (95% CI) | I 2 (%) | Standardized β, ΔLDL Control-Rx | p Value | Standardized β, ΔLDL Baseline-Final | p Value |
|---|---|---|---|---|---|---|
| Coronary artery disease | ||||||
| 17-randomized controlled trials meta-analysis | 0.69 (0.64–0.75) | 60 | −0.108 | <0.001 | −0.091 | 0.052 |
| Excluding MEGA | 0.70 (0.64–0.76) | 61 | −0.116 | <0.001 | −0.106 | 0.023 |
| Excluding PROSPER | 0.68 (0.63–0.75) | 60 | −0.121 | <0.001 | −0.094 | 0.042 |
| Excluding CORONA | 0.68 (0.63–0.75) | 62 | −0.128 | <0.001 | −0.110 | 0.019 |
| Excluding PROSPER and CORONA | 0.67 (0.62–0.74) | 61 | −0.142 | <0.001 | −0.116 | 0.012 |
| Excluding MEGA and CORONA | 0.69 (0.63–0.75) | 63 | −0.134 | <0.001 | −0.126 | 0.007 |
| Excluding MEGA, PROSPER, and CORONA | 0.68 (0.62–0.74) | 63 | −0.148 | <0.001 | −0.131 | 0.004 |
| Excluding JUPITER | 0.70 (0.65–0.76) | 58 | −0.104 | <0.001 | −0.087 | 0.054 |
| Excluding GREACE | 0.71 (0.65–0.76) | 55 | −0.094 | 0.003 | −0.052 | 0.325 |
| Excluding ALLHAT | 0.68 (0.63–0.73) | 46 | −0.081 | 0.041 | −0.078 | 0.056 |
| Excluding GISSI-P | 0.69 (0.63–0.75) | 62 | −0.110 | 0.001 | −0.091 | 0.081 |
| Excluding 4S | 0.71 (0.66–0.77) | 50 | −0.092 | 0.019 | −0.070 | 0.180 |
| Excluding SPARCL | 0.69 (0.64–0.76) | 63 | −0.108 | 0.001 | −0.091 | 0.063 |
| Excluding high-risk CAD ⁎ | 0.65 (0.59–0.72) | 0 | −0.057 | 0.437 | −0.046 | 0.522 |
| Excluding low-risk CAD † | 0.72 (0.66–0.78) | 63 | −0.115 | <0.001 | −0.125 | 0.004 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
