Effects of Statins on Nonlipid Serum Markers Associated with Cardiovascular Disease

Key Summary Points All statins are equally effective at lowering C-reactive protein (CRP) levels. There is no evidence of a dose effect on the degree of CRP-lowering. There is insufficient evidence to establish the effect of statin dose on CRP level. Strong evidence shows that statins do not affect fibrinogen levels. Limited evidence found no effect of statins on levels of homocysteine, low-density lipoprotein cholesterol oxidation, tissue plasminogen activator, or plasminogen activator inhibitor. The data on platelet aggregation are inconclusive. No study evaluated whether the effect of statins on any marker is related to their effect on cardiovascular outcomes. Over the past decade, 3-hydroxy-3-methylglutaryl coenzyme A (HMG-CoA) reductase inhibitors (commonly referred to as statins) have emerged as one of the most effective means of reducing risk for cardiovascular disease (CVD). Several large randomized, controlled trials have demonstrated that statins statistically significantly reduce risk for CVD in both primary and secondary prevention settings (1-4). Most studies focused on the efficacy of statins in patients with elevated circulating levels of low-density lipoprotein (LDL) cholesterol. The clinical benefit of statins observed in these trials was their ability to lower LDL cholesterol levels. More recently, however, there has been growing interest in the possibility that some of the clinical benefits of statins are due to so-called pleiotropic effects that are not directly related to their lipid-altering effects (5, 6). Support for this hypothesis comes from various sources. First, in vitro studies have clearly demonstrated lipid-independent effects of statins on various signaling pathways that are potentially relevant to the pathogenesis of atherosclerosis (7). The list of reported nonlipid effects of statins is increasing and includes such factors as decreasing expression or activity of inflammatory elements, enhancing endothelial function, and decreasing expression of matrix metaloproteinases (7, 8). Second, recent clinical data suggest that statins may also reduce risk for CVD even in individuals with low LDL cholesterol levels. This suggests that factors other than LDL cholesterol reduction may explain statins' effects on CVD, although an alternative interpretation is that current guidelines have selected LDL cholesterollowering goals that are not stringent enough to reflect maximum benefit with LDL cholesterol lowering itself (9). Third, researchers increasingly recognize the importance of nonlipid factors in the pathogenesis of atherosclerosis. These include modulators of thrombosis, thrombolysis, and inflammation (10, 11). Growing evidence supports important prognostic and pathophysiologic roles for many nonlipid markers in CVD (12). Thus, there is increasing support that nonlipid markers are associated with CVD and that nonlipid effects of statins contribute to their clinical benefit; researchers have identified many nonlipid factors as candidates for explaining some of the benefit of statins that are independent of altering circulating lipid levels. We undertook a systematic review of the literature examining the effects of statins on 7 nonlipid serum markers associated with risk for CVD. We focused on 3 issues: 1) Do statins as a class affect the nonlipid serum markers, and, if so, are there differences in the effects of specific statins on each of the nonlipid serum markers? 2) Does the magnitude of the statin's effect on lipids correlate with the magnitude of its effect on the nonlipid marker? and 3) Do statin-induced alterations in nonlipid serum markers predict changes in cardiovascular outcomes? Methods We conducted a systematic search of the English-language literature on statins in MEDLINE between 1980 and November 2002. We reviewed additional publications found by domain experts and in bibliographies of retrieved articles. We included studies that reported original data on the effect of statins in adults on the outcomes of interest. We excluded studies of patients with organ transplants, studies of cerivastatin or other unmarketed statins, studies of only combination statin and other lipid-lowering agents, studies that did not report outcomes for specific statins, studies that had fewer than 10 participants treated with statins, and studies in which change in outcome measurements could not be calculated. We performed an updated search in June 2003 for randomized trials of at least 30 participants that compared either a statin with placebo or 2 or more statins. We did not include studies of fibrinogen because we had several relevant studies from the original search. We chose 7 serum markers from a list of approximately 90 potential markers of CVD on the basis of their potential clinical relevance (as determined by a cardiologist with expertise in lipid research), how frequently the markers have been analyzed in statin trials, and how many statins have been tested. We compared the effectiveness of different statins by relying on studies that directly compared statins; however, because there are few such studies, we indirectly compared statins across placebo-controlled and cohort studies. For randomized, controlled trials and single-group cohort trials, we defined baseline as the time before the start of statin (or control) use after any washout period. All nonstatin, noncontrol cohorts (for example, those receiving fish oil or warfarin) were ignored. If necessary, for each study, we calculated the change of outcome compared with baseline for people receiving statin. From placebo-controlled trials, we calculated the relative change of the outcome compared with placebo: the net difference between the within-treatment effect and the within-placebo effect. Data from crossover studies were handled in 2 ways, depending on the number of available studies for a given marker. Crossover studies of fibrinogen, for which there are several studies, were eligible for meta-analysis only if data were reported for the first phase of the trial (before the crossover). This excluded the possibility of effects due to incomplete washout between phases. For other outcomes, which had relatively few studies, we used data from combined phases if first-phase data were not reported. To summarize and compare studies, we required data on both the mean change in outcome level and the variance of the change. However, many studies provided the variance for only the baseline and final outcome levels. To include these studies, we used the following equation to estimate the standard error (SE) of the change: SE12 = (SE1 2 + Se 2 2 2 r SE1 Se 2), where SE1, Se 2, and SE12 are the SEs for baseline, final, and change, respectively, and r is the correlation between SE1 and Se 2 (13). We assumed a correlation of 0.50. We performed meta-analyses by using the DerSimonian and Laird random-effects model, which assigns a weight to each study that is based on both the individual study variance and the between-study heterogeneity (14). Statistical significance was defined as a P value less than 0.05. We included all relevant studies in meta-analyses and attempted to find explanations for heterogeneity. After tabulating data, we defined criteria to categorize findings (Table). Table. Summary of Evidence from Randomized, Controlled and Cohort Trials The funding source formulated the initial study questions and provided the preliminary reference list of studies. The funding source did not participate in the literature search, the determination of study eligibility criteria, the data analysis and interpretation, or the decision to submit the manuscript for publication. Data Synthesis The original search yielded 4265 citations; the updated search yielded an additional 396 citations. We retrieved 430 articles for review; of these, 104 reported data on the effect of statins on the serum markers evaluated in this paper. The Appendix Figure shows the study selection process, and Appendix Tables 1, 2, 3, 4, and 5 summarize the placebo-controlled trials. Summaries of other included studies are available from the authors. The Table summarizes the number and duration of studies and the treatment effect of each statin for the evaluated markers, including whether treatment effects were associated with lipid reductions or CVD outcomes. No study addressed whether the effect of the statins on the markers was associated with cardiovascular outcomes. The studies were heterogeneous. They examined 5 different statins in a wide range of doses that were often adjusted mid-study, were of varying durations, and included different study samples. Many studies presented only uncontrolled cohort data in which outcomes could have been affected by variables other than the use of statins. The specific effects of statins relevant to our analysis were often reported as secondary outcomes, which may have been reported selectively on the basis of statistically significant or novel results. While we considered the effect of statin dose, eligibility criteria, and baseline level of factor, our conclusions from these types of analyses can be, at most, hypothesis-generating. C-Reactive Protein C-reactive protein (CRP) is an acute phase reactant produced in the liver that is thought to represent an integrated assessment of the overall state of activation of the inflammatory system. Recently, a high-sensitivity assay for measuring CRP levels was developed that detects levels of CRP below what was previously considered the normal range. A growing body of studies suggests that elevations in CRP levels detected by the high-sensitivity assay predict a poor cardiovascular prognosis (15). The studies that described changes in high-sensitivity CRP levels with statin use showed that statins decrease CRP levels. Studies of similar doses found no difference in the effect on CRP among the different statins. These studies found no consistent associations between statins' effects on CRP