Comparative Effect of Statins and Omega-3 Supplementation on Cardiovascular Events: Meta-Analysis and Network Meta-Analysis of 63 Randomized Controlled Trials Including 264,516 Participants

Statins and omega-3 supplementation have been recommended for cardiovascular disease prevention, but comparative effects have not been investigated. This study aimed to summarize current evidence of the effect of statins and omega-3 supplementation on cardiovascular events. A meta-analysis and a network meta-analysis of 63 randomized controlled trials were used to calculate pooled relative risks (RRs) and 95% confidence intervals (CIs) for the effects of specific statins and omega-3 supplementation compared with controls. Overall, the statin group showed significant risk reductions in total cardiovascular disease, coronary heart disease, myocardial infarction, and stroke; however, omega-3 supplementation significantly decreased the risks of coronary heart disease and myocardial infarction only, in the comparison with the control group. In comparison with omega-3 supplementation, pravastatin significantly reduced the risks of total cardiovascular disease (RR = 0.81, 95% CI = 0.72–0.91), coronary heart disease (RR = 0.75, 95% CI = 0.60–0.94), and myocardial infarction (RR = 0.71, 95% CI = 0.55–0.94). Risks of total cardiovascular disease, coronary heart disease, myocardial infarction, and stroke in the atorvastatin group were statistically lower than those in the omega-3 group, with RRs (95% CIs) of 0.80 (0.73–0.88), 0.64 (0.50–0.82), 0.75 (0.60–0.93), and 0.81 (0.66–0.99), respectively. The findings of this study suggest that pravastatin and atorvastatin may be more beneficial than omega-3 supplementation in reducing the risk of total cardiovascular disease, coronary heart disease, and myocardial infarction.


Introduction
Cardiovascular diseases (CVDs) are characterized by any disorders relating to the heart or blood vessels [1], including coronary heart disease (CHD), cerebrovascular disease, stroke, peripheral vascular disease, rheumatic and congenital heart diseases, and venous thromboembolism [2]. It has been reported that the burden of CVDs has remained high over the past decades, with 75% of overall CVD deaths occurring in low-to middle-income countries [3]. In general, the prevalence of CVDs is closely related to various risk factors, such as hypertension, overweight and obesity, dyslipidemia, and diabetes [4,5]. According to recent guidelines of the American College of Cardiology/American Heart Association (ACC/AHA) regarding the primary prevention of CVD, individuals aged 40 to 75 years with diabetes mellitus and low-density lipoprotein-cholesterol (LDL-C) ≥70 mg/dL or those with hypercholesterolemia (LDL-C ≥190 mg/dL) regardless of CVD risk are recommended with moderate between cardiovascular event prevention effect and LDL-C reduction in statins and omega-3 supplementation arm separately.
Relative risks (RRs) and 95% confidence intervals (CIs) for the comparative efficacy of statins, omega-3 supplementation, and control were computed in the NMA. As the prior probability was not established in the study hypothesis, we performed the NMA with the frequentist approach to combine direct and indirect evidence [19]. A funnel plot was utilized to assess publication bias [20], and a contribution plot was applied to examine the influence of direct estimates on network summary effects [21]. Stata SE version 14.0 (StataCorp, College Station, TX, USA) and R software (R Foundation for Statistical Computing, Vienna, Austria) (version 3.5.2) were used for all statistical analyses.

Meta-Analysis of Direct Estimates
In comparison with the control group, the pooled direct estimates of statins or omega-3 supplementation and the risk of CVD events are summarized in Table 3. Overall, the levels of significant findings were equivalent in both the fixed-effects and random-effect models, except for the association between atorvastatin or rosuvastatin and the risk of stroke. The total statin and omega-3 groups exhibited reduced risks of total CVD (RR = 0.89, 95% CI = 0.85-0.94), CHD (RR = 0.81, 95% CI = 0.75-0.89), MI (RR = 0.78, 95% CI = 0.78-0.85), and stroke (RR = 0.91, 95% CI = 0.85-0.98).
In subgroup analysis by statins, total statins were associated with decreased risks of total CVD, CHD, MI, and stroke, with RRs (95% CIs) of 0.81 (0.76-0.86), 0.70 (0.62-0.77), 0.69 (0.61-0.78), and 0.85 (0.79-0.82), respectively. Individually, atorvastatin and pravastatin had significant effects on all outcomes. Furthermore, rosuvastatin was associated with significantly reduced risks of CHD and MI, and lovastatin was associated with a significant risk reduction in total CVD. Fewer stroke events occurred in the simvastatin group than in the control group, and fluvastatin was observed to reduce total CVD. Moreover, omega-3 supplementation was associated with a 19% reduced risk of CHD (RR = 0.81, 95% CI = 0.75-0.89) and an 11% reduced risk of MI (RR = 0.89, 95% CI = 0.80-0.99).
The meta-regression coefficients for the association between the lipid-lowering effect and risk reduction of cardiovascular events are presented in Table 4. The size of TC reduction achieved with overall statins and omega-3 supplementation was significantly associated with the size of total CVD and CHD reduction effect, with coefficients (p-values) of 0.0046 (<0.001) and 0.0042 (0.047), respectively. Additionally, total CVD and CHD reduction effect might be driven by per unit change in LDL-C level, with decreases of 0.0034 (p = 0.004) and 0.0044 (p = 0.047) in log (RRs) for total CVD and CHD per 1 mg/dL LDL-C lowering effect of statins and omega-3 supplementation, respectively. The combined meta-analysis of 20 effect sizes showed a significant reduction in LDL-C concentration following administration of statins (WMD, −33.63 mg/dL; 95% CI, −45.77 to −21.49 mg/dL; Figure A1), but not omega-3 supplementation (WMD, 0.12; 95% CI, −0.81 to 1.06 mg/dL; Figure A1). Relationships between cardiovascular event reduction and LDL-C lowering effects by subgroup for statins and omega-3 supplementation are presented in Figures A2 and A3. CHD reduction effect of statins (p = 0.004) and stroke reduction effect of omega-3 supplementation (p = 0.02) were observed to be non-linearly associated with the LDL-C lowering effect, with the explained deviance of over 97%.
The coefficient from the meta-regression model represents the change in the log-transformed relative risk for every increase of one mg/dL in serum lipid concentrations.          The coefficient from the meta-regression model represents the change in the log-transformed relative risk for every increase of one mg/dL in serum lipid concentrations.   Among statins, pravastatin, fluvastatin, lovastatin, and atorvastatin were found to be associated with lower risks of total CVD than simvastatin, with RRs (95% CIs) of 0.82 (0.69-0.96), 0.42 (0.18-0.97), 0.78 (0.62-0.98), and 0.81 (0.74-0.89), respectively. The risk of CHD in the atorvastatin group was also 19% lower than that in the simvastatin group (RR = 0.81, 95% CI = 0.73-0.90). Fluvastatin was associated with a higher MI risk than pravastatin, with RR = 2.21 and 95% CI = 1.04-4.69.

Network Meta-Analysis of Direct and Indirect Estimates
Data are presented as the relative risk and 95% confidence interval of row-defined intervention versus column-defined comparison for the risk of different cardiovascular events and with corresponding network plots of available direct comparison. The size of nodes is proportional to the number of studies including the respective treatments. The thickness of the edges is proportional to the mean control group risk for comparisons between control and active treatment groups.

Publication Bias and Contribution Plot
Funnel plot asymmetry indicating potential publication bias was not observed for different outcomes of CVD events (Figure 3). The contribution to mixed estimates and the influence on indirect estimates of direct estimates are presented in Figures A4-A7. Overall, most of the direct comparisons contributed nearly 100% to the same mixed comparisons in terms of total CVD ( Figure A4) and CHD outcomes ( Figure A5), except for control versus simvastatin. In contrast, approximately half of the direct comparisons did not contribute to the same mixed comparisons in terms of MI ( Figure A6) and stroke outcomes ( Figure A7).
Data are presented as the relative risk and 95% confidence interval of row-defined intervention versus column-defined comparison for the risk of different cardiovascular events and with corresponding network plots of available direct comparison. The size of nodes is proportional to the number of studies including the respective treatments. The thickness of the edges is proportional to the mean control group risk for comparisons between control and active treatment groups.

Publication Bias and Contribution Plot
Funnel plot asymmetry indicating potential publication bias was not observed for different outcomes of CVD events (Figure 3). The contribution to mixed estimates and the influence on indirect estimates of direct estimates are presented in Figures A4-A7. Overall, most of the direct comparisons contributed nearly 100% to the same mixed comparisons in terms of total CVD ( Figure A4) and CHD outcomes ( Figure A5), except for control versus simvastatin. In contrast, approximately half of the direct comparisons did not contribute to the same mixed comparisons in terms of MI ( Figure A6) and stroke outcomes ( Figure A7).

Treatment Ranking
Fluvastatin had the highest probability of being the primary intervention for total CVD (85.4%) and stroke (42.2%), whereas atorvastatin and pitavastatin had the highest probabilities in terms of CHD (78.1%) and MI (72.7), respectively ( Table 5). The surface under the cumulative ranking curve (SUCRA) values of cumulative ranking probability, which accounts for the uncertainty of spuriously high ranks, suggested better ranks for fluvastatin, atorvastatin, pitavastatin, and rosuvastatin in the prevention of total CVD (0.9), CHD (1.0), MI (0.9), and stroke (0.7).

Discussion
This meta-analysis and NMA of 63 RCTs investigated the effect of statins and omega-3 supplementation as well as their comparative efficacy of for the prevention of cardiovascular events. TG and HDL-C independent effects of total statins and omega-3 supplementation on the prevention of total CVD were observed. Furthermore, the effects of total statins and omega-3 supplementation on CHD, MI, and stroke were independent of the lipid-lowering effects of TC, TG, HDL-C, and LDL-C. Although the RCTs examined diverse populations, the findings were generally consistent for the significant risk reduction associated with statins in total CVD, CHD, MI, and stroke; conversely, a beneficial effect on stroke prevention was not observed for omega-3 supplementation. The findings were generally consistent in separate subgroup analyses of statins and omega-3 supplementation or in combined analyses of statins and omega-3 supplementation in terms of total CVD and CHD. Moreover, evidence was robust according to the Dersimonian-Laird methodology of the random-effects model, which accounts for between-study variability when at least two individual RCTs are included in pooled analysis.
The NMA of 264,516 adults showed that most statins and omega-3 supplementation were effective in reducing total CVD, CHD, and MI risks compared to the control group. Fluvastatin, atorvastatin, pitavastatin, and rosuvastatin were found to have the greatest effects on reducing total CVD, CHD, MI, and stroke, respectively.
Several guidelines have recommended statins for the primary and secondary prevention of CVD [89]. Statins can be classified as weak statins (simvastatin and pravastatin, compounds found in nature), strong (atorvastatin, pitavastatin, and rosuvastatin, synthetic molecules), and normal (fluvastatin and lovastatin) based on their ability to reduce cholesterol levels [90]. Yebyo et al. recently investigated the primary prevention effects of specific statins by NMA, though the date of the literature search was restricted to between 2013 and 2018, and the effect on total CVD and CHD was not investigated [31]. Statins were generally found to be associated with risk reductions in nonfatal MI (RR = 0.62, 95% CI = 0.53-0.72) and nonfatal stroke (RR = 0.83, 95% CI = 0.53-0.72) but not fatal MI or fatal stroke. In the current study, we found that statins significantly lowered the risks of combined fatal and nonfatal MI and stroke.
This study had some intrinsic limitations. First, we did not specify the effect of statins and omega-3 supplementation on primary and secondary prevention of event outcomes, and primary prevention may be defined differentially in various studies. For example, Yebyo et al. categorized the population with CVD history was less than 10% of the total sample size as primary prevention [31], Naci et al. accepted a percentage up to 20% [22]. Additionally, the time frame in the identification of CVD history differed among RCTs. Therefore, we assumed that the effects of CVD and MI histories as well as other risk factors, such as hypertension, diabetes, and smoking, might not be significant. Second, arm-based data considering the sample size and number of events were included in the final analysis. Although the time to outcome for survival data is important in evaluating the effect of the intervention, the findings were not based on the contrast-based data of relative estimates because the hazard ratio was not available for all outcomes and in all RCTs. Third, there were only a few head-to-head RCTs of a statin versus a statin [80][81][82][83][84][85] and no RCT of a statin versus omega-3 supplementation. Last, due to the limitation of data availability, meta-regression for each type of statin as well as for omega-3 supplementation could not be performed separately.
Despite the abovementioned limitations, this study has important methodological strengths. To our knowledge, this is the first indirect comparison of statins versus omega-3 supplementation borrowing direct evidence of statins versus controls and omega-3 supplementation versus controls to provide quantitative evidence for the prevention of CVD events. The study was designed to minimize heterogeneity among controls from different RCTs. Although Abdelhamid et al. reported comprehensive evidence from 79 RCTs of omega-3 supplementation for the primary and secondary prevention of CVD [27], several RCTs were excluded from our analysis because of inappropriate control groups. The control group, which was defined as those with a lower intake of omega-3, might inappropriately develop an intermediary association with statins. RCTs in which the control group contained omega-6 were also excluded, as the debated effect of omega-6 intake has been reported [95][96][97][98][99]; thus, the inclusion of omega-6 in the control group might underestimate or overestimate the protective effect of omega-3. Moreover, robust findings were obtained by using different models in the meta-analysis and by comparing direct evidence from the meta-analysis and combined evidence from the NMA.

Conclusions
In summary, this study suggests that pravastatin and atorvastatin are more beneficial than omega-3 supplementation in reducing the risk of total CVD, CHD, and MI. Fluvastatin, atorvastatin, pitavastatin, and rosuvastatin showed the greatest effects on reducing total CVD, CHD, MI, and stroke, respectively.