Association Between Triglyceride Lowering and Reduction of Cardiovascular Risk Across Multiple Lipid-Lowering Therapeutic Classes
Randomized trials of therapies that primarily lowered triglycerides have not consistently shown reductions in cardiovascular events.
We performed a systematic review and trial-level meta-regression analysis of 3 classes of lipid-lowering therapies that reduce triglycerides to a greater extent than they do low-density lipoprotein cholesterol (LDL-C): fibrates, niacin, and marine-derived omega-3 fatty acids. Key inclusion criteria were a randomized controlled trial that reported major vascular events. We also incorporated data from a previous meta-regression of 25 statin trials. The main outcome measure was the risk ratio (RR) for major vascular events associated with absolute reductions in lipid parameters.
A total of 197 270 participants from 24 trials of nonstatin therapy with 25 218 major vascular events and 177 088 participants from 25 trials of statin therapy with 20 962 major vascular events were included, for a total of 374 358 patients and 46 180 major cardiovascular events. Starting with non–high-density lipoprotein cholesterol, a surrogate for very-low-density lipoproteins and low-density lipoproteins, the RR per 1-mmol/L reduction in non–high-density lipoprotein cholesterol was 0.79 (95% CI, 0.76–0.82; P<0.0001; 0.78 per 40 mg/dL). In a multivariable meta-regression model that included terms for both LDL-C and triglyceride (surrogates for low-density lipoproteins and very-low-density lipoproteins, respectively), the RR was 0.80 (95% CI, 0.76–0.85; P<0.0001) per 1-mmol/L (0.79 per 40 mg/dL) reduction in LDL-C and 0.84 (95% CI, 0.75–0.94; P=0.0026) per 1-mmol/L (0.92 per 40 mg/dL) reduction in triglycerides. REDUCE-IT (Reduction of Cardiovascular Events With Icosapent Ethyl–Intervention Trial) was a significant outlier and strongly influential trial in the meta-regression. When removed, the RRs became 0.79 (95% CI, 0.76–0.83; P<0.0001) per 1-mmol/L (0.78 per 40 mg/dL) reduction in LDL-C and 0.91 (95% CI, 0.81–1.006; P=0.06) per 1-mmol/L (0.96 per 40 mg/dL) reduction in triglycerides. In regard to omega-3 dose, each 1 g/d eicosapentaenoic acid administered was associated with a 7% relative risk reduction in major vascular events (RR, 0.93 [95% CI, 0.91–0.95]; P<0.0001), whereas there was no significant association between the dose of docosahexaenoic acid and the relative risk reduction in major vascular events (RR 0.96 [95% CI, 0.89–1.03]).
In randomized controlled trials, triglyceride lowering is associated with a lower risk of major vascular events, even after adjustment for LDL-C lowering, although the effect is less than that for LDL-C and attenuated when REDUCE-IT is excluded. Furthermore, the benefits of marine-derived omega-3 fatty acids, particularly high-dose eicosapentaenoic acid, appear to exceed their lipid-lowering effects.
What Is New?
A reduction in non–high-density lipoprotein cholesterol, a measure of atherogenic low-density lipoprotein and very low-density lipoprotein particles, is strongly associated with a lower risk of major vascular events regardless of the lipid-lowering drug class.
Triglyceride lowering is associated with a lower risk of cardiovascular events but to a lesser extent per absolute amount of reduction than with low-density lipoprotein cholesterol.
Nearly all nonstatin trials did not achieve sufficient non–high-density lipoprotein cholesterol lowering to be adequately powered to detect a clinical difference in major vascular events.
The benefits of marine-derived omega-3 fatty acids, particularly high-dose eicosapentaenoic acid, appear to exceed their lipid-lowering effects.
What Are the Clinical Implications?
Developing drugs that achieve large reductions in very-low-density lipoprotein and triglycerides or targeting patients with high baseline levels of triglycerides would likely increase the probability of showing meaningful clinical benefit.
Fibrates could be considered in patients needing further non–high-density lipoprotein cholesterol lowering (being mindful of side effects) because they should offer clinical benefit proportional to the degree of non–high-density lipoprotein cholesterol lowering.
If a disproportionate relationship between lipid lowering and cardiovascular risk reduction is validated in ongoing trials of high-dose omega-3 fatty acids, it will support the hypothesis that they confer unique benefits beyond lipid lowering.
The association between triglycerides and cardiovascular risk is well established from epidemiological studies,1–5 genome-wide analyses,6–8 and mendelian randomization studies.9,10 Furthermore, the American Heart Association has recognized triglycerides as an important biomarker of cardiovascular disease risk.5 However, randomized controlled clinical trials of therapies that primarily lower triglyceride levels have not consistently shown a reduction in major vascular events.11–34 Moreover, triglycerides per se may not be atherogenic but rather may be a marker for the concentration of atherogenic very-low-density lipoprotein (VLDL) particles.35
There are 3 major therapeutic classes that primarily lower triglycerides or VLDL: fibrates,13–21 niacin,13,22,23 and marine-derived omega-3 fatty acids (such as eicosapentaenoic acid [EPA] and docosahexaenoic acid [DHA]).24–34,36,37 Most of the fibrate trials, both of the statin-era niacin trials, and all but 2 of the omega-3 trials have failed to show a statistically significant reduction in major vascular events. However, genetic studies suggest that lower triglyceride levels, when accompanied by lower apolipoprotein B levels, are associated with lower cardiovascular risk but also that, on a milligrams per deciliter basis, the association of triglycerides with major vascular events is less strong than that for low-density lipoprotein (LDL) cholesterol (LDL-C).6,8,35 Thus, prior trials of drugs that primarily lower triglyceride may have been underpowered with regard to the degree of triglyceride or non–high-density lipoprotein cholesterol (HDL-C) lowering to show a statistically significant benefit. Conversely, it has been postulated that some of these drugs, specifically omega-3 fatty acids, reduce the risk of vascular events through nonlipid mechanisms.38,39
In this meta-regression analysis, we examined the association between the magnitude of non–HDL-C, LDL-C, and triglyceride lowering and the reduction in major vascular events across trials of fibrates, niacin, and marine omega-3 fatty acids, as well as statins as an established reference.
We performed a systematic review and trial-level meta-regression analysis of 3 classes of randomized trials of lipid-lowering therapies that reduce triglycerides to a greater extent than they do LDL-C and have clinical outcomes: namely fibrates, niacin, and omega-3 fatty acids. We also incorporated data from a previous meta-regression of 25 statin trials.40 All results are reported in accordance with Preferred Reporting Items for Systematic Reviews and Meta-Analysis guidelines.41 No patients were involved in the conduct of this meta-analysis; thus, no informed consent or institutional review board approval was required. This meta-analysis uses data that have already been published and are available in the medical literature for others to use.
Triglyceride-lowering trials were identified with the following approaches: (1) searching MEDLINE and EMBASE databases with the search terms triglyceride lowering and clinical outcomes, limited to randomized controlled trials and human studies and published between 1968 and March 2019; (2) searching the reference files of 2 authors (N.A.M. and M.S.S); (3) examining the reference lists of original articles, reviews, meta-analyses, and abstracts; and (4) contacting experts in the field. To be eligible for the meta-regression analysis, all of the following inclusion criteria must have been met: randomized clinical trial of 1 of the aforementioned drug classes, single intervention versus placebo, and reported major vascular outcomes. Trials were excluded for either <1 year of follow-up or <400 subjects.
Each trial was reviewed independently by 2 authors (N.A.M. and M.S.S.). Data collected by each of these authors included sample size; intervention; dose; trial duration; baseline and achieved levels of non–HDL-C, triglycerides, and LDL-C in each arm; total major vascular events; and hazard ratios (or risk ratios [RRs] when hazard ratios were not reported) with 95% CIs for treatment effect. All disagreements were discussed and resolved by consensus. Because some older trials have data available only as RRs rather than hazard ratios, all hazard ratios were treated as RRs when the results of studies were pooled and the term RR was used to describe the effect estimate. This allowed the inclusion of all trials and has been done in previous meta-analyses.40
For trial lipid values, time-weighted differences were used when available; otherwise, the reported laboratory values closest to the midpoint of the trial were used. The end point in each trial that most closely approximated major vascular events was included. In most cases, this consisted of cardiovascular death, acute myocardial infarction or acute coronary syndrome, coronary revascularization, and stroke. The specific outcomes used for each trial are listed in the Tables I through IV in the online-only Data Supplement.
For the first analysis, the association between absolute reduction in non–HDL-C (defined as the difference in achieved non–HDL-C level between the 2 treatment arms) and the RR for major vascular events was established. To determine the contribution from LDL-C and triglycerides, a second analysis assessed the association between both LDL-C and triglycerides with the RR for major vascular events. As previously described,40,42–44 meta-regressions were performed using random-effects models with the restricted maximum likelihood estimation estimator for between-study variability, the Knapp and Hartung adjustment for estimation of standard errors of the estimated coefficients. In each analysis, the intercept was set at zero to reflect the clinical assumption that as the dose of a drug goes to zero, the lipid effect goes to zero and the treatment effect goes to zero (ie, the origin). The corresponding RR with 95% CIs and P values per unit reduction in lipid parameter are reported. As a check, models with terms for intercept also were created and showed that the intercept terms were close to the origin and not significant. Outlier and influence diagnostics such as a measure analogous to the Cook’s distance (D, an estimate of influence),44,45 studentized residuals (a measure for outliers), and hat values (a measure of leverage)44,46 were calculated to examine the robustness of the conclusions from these meta-regression analyses. In particular, D is reported to examine the effect of the deletion of the ith study on the fitted values of all studies simultaneously. If the regression line after removing the ith study is not far from the regression line including the ith study, then D will be relatively small. The cutoff value for D is calculated by drawing a quantile value exceeding 50% of the lower tail area of a χ2 distribution with a given degree of freedom. If calculated D exceeds this cutoff, one can consider the ith study to potentially be an influential case. Leverage is a measure of the distance between the predictor variable value (such as the difference in non–HDL-C between treatment arms in a trial) and the average of the predictor variable values (mean of all non–HDL-C difference values) in the entire data set. This distance is scaled according to the SD of each study. A trial with high leverage is identified by a cutoff value that is a function of the number of parameters in the model and number of trials used in the analysis. Two-sample z tests were performed to evaluate deviations of observed RR from the meta-regression line, in logarithm scale, with and without the ith trial (a trial of interest). For omega-3 trials, an additional meta-regression analysis was performed to evaluate the association between EPA and DHA dosage and the RR for major cardiovascular events.
All P values were 2-sided, and statistical significance was assessed at an α level of 0.05. Statistical analyses were performed with the Metafor package version 3.30 R Project for Statistical Computing.43
Risk-of-bias assessments were performed for each trial with the Cochrane Collaboration tool.47 This included evaluation of randomization process, blinding, withdrawal of consent, loss to follow-up, and outcomes assessment and reporting (Table I in the online-only Data Supplement).47 No significant bias was detected across trials, although it was notable that JELIS (Japan EPA Lipid Intervention Study) was the only trial with an open-label design. Publication bias was also assessed by funnel plots and the Egger regression test.48
A total of 259 publications was identified in the screening process and reviewed for eligibility (Figure I in the online-only Data Supplement). Of these, 24 triglyceride-lowering trials were included in the meta-regression analyses. There were 9 fibrate trials (Table I in the online-only Data Supplement), 3 niacin trials (Table II in the online-only Data Supplement), and 13 omega-3 fatty acids trials (Table III in the online-only Data Supplement) (1 trial had both fibrate and niacin experimental arms). These trials included a total of 197 270 patients with an average age of 63 years and a mean baseline triglyceride level of 163 mg/dL (median, 151 mg/dL; interquartile range, 143–165 mg/dL). The average median trial follow-up was 4.8 years, during which there were 25 218 major cardiovascular events (Table). In addition to these trials, we had data from 25 statin trials (Table IV in the online-only Data Supplement) that included an additional 177 088 patients and 20 962 major vascular events, for a grand total of 374 358 patients and 46 180 major vascular events (Table).
|Participants, n||41 520||32 995||125 544||197 270*||177 088||374 358|
|Mean baseline non–HDL-C, mg/dL (mmol/L)||156 (4.0)||87 (2.3)||139 (3.6)||130 (3.4)||167 (4.3)||157 (4.1)|
|Mean baseline triglycerides, mg/dL (mmol/L)||183 (2.1)||210 (2.4)||163 (1.8)||176 (2.0)||149 (1.7)||159 (1.8)|
|Mean baseline LDL-C, mg/dL (mmol/L)||123 (3.2)||65 (1.7)||114 (2.9)||102 (2.6)||130 (3.4)||127 (3.3)|
|Major vascular events, n||4988||5136||15 933||25 218||20 962||46 180|
|Average duration of follow-up, y||5.1||3.9||5.1||4.8||4.5||4.7|
|Mean non–HDL-C response, mg/dL (mmol/L)||19 (0.5)||16 (0.4)||4 (0.1)||11 (0.2)||37 (1.0)||24 (0.6)|
|Mean triglyceride response, mg/dL (mmol/L)||44 (0.5)||37 (0.4)||20 (0.2)||33 (0.4)||22 (0.2)||25 (0.3)|
|Mean LDL-C response, mg/dL (mmol/L)||10 (0.2)||14 (0.4)||1 (0.02)||8 (0.2)||33 (0.9)||22 (0.6)|
Non–HDL-C Lowering and Major Vascular Events
To start, we examined the relationship between lowering non–HDL-C levels, a measure of all apolipoprotein B–containing atherogenic lipoproteins, and risk of major vascular events. When 18 nonstatin trials with non–HDL-C data were pooled (8 fibrate, 3 niacin, and 8 omega-3 fatty acid trials), the RR per 1-mmol/L reduction in non–HDL-C was 0.77 (95% CI, 0.69–0.86; P=0.0001; 0.76 per 40 mg/dL). The relationship was similar in 19 statin trials with non–HDL-C data, with an RR of 0.80 (95% CI, 0.77–0.82; P<0.0001; 0.79 per 40 mg/dL). When the 37 trials were pooled together, the RR was 0.79 (95% CI, 0.76–0.82; P<0.0001; 0.78 per 40 mg/dL; Figure 1). Sensitivity analyses including terms for duration of follow-up and proportion of background statin use (in trials of nonstatin therapy) were performed but showed no significant interactions (P for interaction=0.53 and 0.55, respectively).
In an examination of individual classes, the RR per 1-mmol/L reduction in non–HDL-C was 0.80 (95% CI, 0.77–0.82; 0.79 per 40 mg/dL) for statins, 0.79 (95% CI, 0.71–0.88; 0.78 per 40 mg/dL) for fibrates, 0.84 (95% CI, 0.75–0.95; 0.83 per 40 mg/dL) for niacin, and 0.51 (95% CI, 0.40–0.64; 0.50 per 40 mg/dL) for omega-3 fatty acid trials (Figure II in the online-only Data Supplement). The RRs for fibrate, niacin, and statin trials were similar (P≥0.39 for differences between classes), and when these 3 classes were pooled, the RR was 0.80 (95% CI, 0.78–0.82; 0.79 per 40 mg/dL). In contrast, the RR for the omega-3 fatty acid trials differed significantly from the RRs seen in the fibrate (P=0.008), niacin (P=0.004), and statin (P=0.005) trials. There was no evidence of publication bias (Table VI and Figure III in the online-only Data Supplement).
Three nonstatin trials were outliers and deviated significantly from the regression line: a fibrate trial, VA-HIT (VA HDL Intervention Trial; P=0.02), and 2 omega-3 fatty acid trials, JELIS (P=0.0494) and REDUCE-IT (Reduction of Cardiovascular Events With Icosapent Ethyl–Intervention Trial; P=0.0001). Of these significant outliers, only REDUCE-IT had borderline influence on the meta-regression line (D=0.35, threshold for influence D>0.45; Figure IV in the online-only Data Supplement). With REDUCE-IT removed, the RR for omega-3 fatty acids became similar to that of the other classes (RR, 0.76 [95% CI, 0.27–2.16]; 0.75 per 40 mg/dL), and the overall meta-regression RR became 0.80 (95% CI, 0.78–0.82; P<0.0001; 0.79 per 40 mg/dL).
On the basis of the observed meta-regression relationship between non–HDL-C lowering and reduction in major vascular events and the actual non–HDL-C lowering in nonstatin trials, the power to detect a significant clinical benefit given the number of events in each of the nonstatin trials was quite limited, being ≤20% in 14 trials, 21% to 50% in 2 trials, 51% to 79% in 2 trials, and ≥80% in only 1 trial (Table VII in the online-only Data Supplement).
LDL-C and Triglyceride Lowering and Major Vascular Events
Next, to investigate the individual associations of LDL-C and triglyceride lowering (surrogates for LDL and VLDL, respectively) with risk of major vascular events, a multivariable meta-regression model was constructed that included terms for the difference in both LDL-C and triglyceride between treatment arms. When 44 trials with necessary lipid data were pooled (9 fibrate, 3 niacin, 8 omega-3 fatty acid, and 25 statin trials), the RR was 0.80 (95% CI, 0.76–0.85; P<0.0001) per 1-mmol/L (0.79 per 40 mg/dL) reduction in LDL-C and 0.84 (95% CI, 0.75–0.94; P=0.0026) per 1-mmol/L (0.92 per 40 mg/dL) reduction in triglycerides (Figure 2A and 2B). Once again, REDUCE-IT was a significant outlier (P=0.0007) and had a strong influence on the meta-regression (D=2.03, threshold for influence D>1.39; Figure V in the online-only Data Supplement). With REDUCE-IT removed, the RRs became 0.79 (95% CI, 0.76–0.83; P<0.0001) per 1-mmol/L (0.78 per 40 mg/dL) reduction in LDL-C and 0.91 (95% CI, 0.81–1.006, P=0.06) per 1-mmol/L (0.96 per 40 mg/dL) reduction in triglycerides (Figure 2A and 2B).
Omega-3 Dose and Major Vascular Events
Among the 13 omega-3 fatty acids trials, all but 2 trials used a combination of EPA and DHA. The total omega-3 dose ranged from 376 to 4000 mg daily, with a mean of 1355 mg (median, 1000 mg). The mean EPA dose was 944 mg (median, 500 mg) with a range of 226 to 4000 mg daily. In the 11 trials with DHA, the mean dose was 411 mg (median, 380 mg) with a range of 0 to 950 mg. In a meta-regression by EPA dose, for each 1 g/d EPA administered, there was a 7% relative lower risk in major vascular events (RR, 0.93 [95% CI, 0.91–0.95]; P<0.0001; Figure 3A). In contrast, in a meta-regression by DHA dose, 1 g/d was associated with a nonsignificant 4% lower risk in major cardiovascular events (RR, 0.96 [95% CI, 0.89–1.03]; P=0.27; Figure 3B). The significantly higher dose of 4 g/d EPA in REDUCE-IT gave the trial high leverage but was not influential on the regression line (D=0.27, threshold for influence D>0.45).
In this meta-regression analysis, there are 5 key observations. First, a reduction in non–HDL-C, a measure of atherogenic LDL and VLDL particles, is strongly associated with a lower risk of major vascular events regardless of the lipid-lowering drug class. Second, triglyceride lowering (a surrogate of triglyceride-rich VLDL) is associated with a lower risk of cardiovascular events but to a lesser extent per absolute amount of reduction than with LDL-C (a surrogate for cholesterol-rich LDL). Third, nearly all nonstatin trials focusing on triglyceride lowering have been underpowered with respect to degree of non–HDL-C lowering to detect a clinical difference in major vascular events. Fourth, REDUCE-IT was the most significant outlier and influencer of the meta-regression, and removing it attenuated the effect estimate for triglyceride lowering with an upper bound of the 95% CIs that just crosses 1.0. Fifth, REDUCE-IT may have been an outlier because of the type of omega-3 fatty acid or the dose that was used.
Recent genetic data highlight that lower levels of apolipoprotein B–containing particles, either cholesterol-rich LDL or triglyceride-rich VLDL, are associated with a similar lower risk of major vascular events.35 As a result, interventions that lower VLDL (and therefore triglycerides) but not LDL should still have cardiovascular benefit. We demonstrate that triglyceride lowering is associated with about half the cardiovascular risk reduction per 1 mmol/L and therefore closer to one-quarter the benefit per 1 mg/dL (1 mmol/L=38.67 mg/dL of LDL-C and 88.57 mg/dL of triglycerides), a ratio congruent with our mendelian randomization study data.35 Specifically, a decrease of 40 mg/dL LDL-C should translate into an ≈20% lower cardiovascular risk, but a similar 40-mg/dL reduction in triglycerides is associated with only approximately a 4% to 5% lower cardiovascular risk. Therefore, obtaining cardiovascular benefit with a purely VLDL- or triglyceride-lowering drug will be more difficult to achieve. These data have important implications for the design of cardiovascular outcomes trials for experimental compounds that primarily lower VLDL and triglycerides. Specifically, developing drugs that achieve large reductions in VLDL and triglycerides or targeting patients with high baseline levels of VLDL or triglycerides would likely increase the probability of showing a meaningful clinical benefit.
As is standard in clinical practice, we used circulating non–HDL-C as a surrogate for all atherogenic lipoproteins, which include LDL and intermediate-density lipoprotein (for both LDL and intermediate-density lipoprotein, the major lipid component is cholesterol) and VLDL (which also contains cholesterol, although the major lipid component is triglycerides). We did not add triglycerides to this model because non–HDL-C already accounts for VLDL particles via their cholesterol content. Indeed, there was a high correlation between the change in non–HDL-C and the change in triglycerides (r=0.90, P<0.0001) in nonstatin trials, supporting the decision not to include them in the same model. However, we then further subset atherogenic proteins by using circulating LDL-C and triglyceride levels as surrogates for LDL and intermediate-density lipoprotein and for VLDL, respectively. Nonetheless, it is important to note that our recent mendelian randomization study35 suggests that it is simply the number of atherogenic apolipoprotein B–containing lipoproteins that is the true determinant of cardiovascular risk, not the content of cholesterol and triglyceride per se, which are surrogate biomarkers.
Given the modest reductions in non–HDL-C seen in nonstatin trials (fibrates, niacin, omega-3 fatty acids), virtually all were underpowered to detect a statistical difference in clinical outcomes. Although niacin has notable side effects,23 our data suggest that fibrates should offer clinical benefit proportional to the degree of non–HDL-C lowering (being mindful of side effects, including an increased risk of myopathy when combined with a statin).
Recently presented data for REDUCE-IT have rekindled interest in triglyceride lowering, specifically omega-3 fatty acids. REDUCE-IT is the most striking outlier among the trials in this meta-regression. The observed clinical benefit of 25% relative risk reduction was well beyond the ≈9% risk reduction expected from the 0.41 mmol/L of non–HDL-C lowering and raises the question of what might be driving the difference between expected and observed risk reductions. To put this finding in context, of the 24 nonstatin trials, 5 had relative risk reductions of at least 20% (3 fibrate trials: HHS [Helsinki Heart Study], VA-HIT, DAIS [Diabetes Atherosclerosis Intervention Study]; 2 omega-3 fatty acid trials: JELIS and REDUCE-IT). Three trials (VA-HIT, JELIS, and REDUCE-IT) significantly deviated from the regression line. The other 2 trials (HHS and DAIS), despite having relative risk reductions per 1-mmol/L-lower non–HDL-C that exceeded what was seen in REDUCE-IT, were relatively small and thus did not statistically significantly differ from the regression line.
Whether REDUCE-IT is an outlier as a result of random chance or pleiotropic effects of EPA remains an important question. Proposed non–lipid-lowering mechanisms for icosapent ethyl include anti-inflammatory, antiarrhythmic, and antiplatelet properties.49 Alternatively, concern has been raised that the control arm of REDUCE-IT may have influenced the outcome of the trial. The comparator group received mineral oil, which itself raised high-sensitivity C-reactive protein by 32.3% from baseline, resulting in a between-group difference at 2 years of 0.9 mg/L. Recent data from the CANTOS trial (Canakinumab Antiinflammatory Thrombosis Outcome Study) showed that a 1.6- to 1.7-mg/dL reduction in high-sensitivity C-reactive protein was associated with a 17% reduction in major vascular events.50 Thus, it is possible that the magnitude of clinical benefit observed in REDUCE-IT could reflect a combination of both lower triglycerides and lower high-sensitivity C-reactive protein in the EPA arm versus the comparator arm of mineral oil, but even those 2 factors together seem insufficient. JELIS and DOIT (Diet and Omega-3 Intervention Trial on Atherosclerosis) were the only 2 other trials to test >1 g/d EPA, and both had cardiovascular risk reduction beyond what was expected on the basis of non–HDL-C lowering, supporting the potential pleiotropic benefits of higher-dose EPA.
Ongoing trials will be important to help clarify whether the benefits of omega-3 fatty acids, or EPA in particular, are in keeping with the expected benefit based on non–HDL-C reduction (with some individual trial variability resulting from play of chance) or also deviate from expected values (suggesting pleiotropic benefits of omega-3 fatty acids or EPA in particular). Specifically, the STRENGTH trial (Outcomes Study to Assess Statin Residual Risk Reduction With Epanova in High CV Risk Patients With Hypertriglyceridemia) is testing a dose of 2000 mg EPA plus 2000 mg DHA compared with a corn oil placebo. If the relationship between lipid lowering and risk reduction in STRENGTH is similar to that seen in REDUCE-IT, it will support the hypothesis that omega-3 confers unique benefits beyond lipid lowering (and that DHA may be equally as efficacious as EPA).
The meta-analysis was not performed on patient-level data. However, beyond the logistical complexity of gathering such data from 49 trials published over 43 years, by definition, the analyses were at the trial level, examining the association between lipid lowering and the RR of major vascular events within a trial. Given the large number of trials, there were differences in patient populations, baseline lipid levels, treatment doses or formulations, and cotherapies (eg, statins). There was visual evidence of publication bias (fewer small negative trials than expected), but this was not statistically significant. It should also be noted that very few of these trials included a substantial proportion of patients with very high triglyceride levels (ie, >400 mg/dL), a group in whom triglyceride lowering would be expected to be associated with the largest clinical benefit. All lipid measurements were not available in every trial, and almost no trials had data on apolipoprotein B. Finally, few omega-3 trials used high doses, giving REDUCE-IT greater influence on the meta-regression line.
In randomized controlled trials, triglyceride lowering is associated with a lower risk of major vascular events, even after adjustment for LDL-C lowering, but that relationship is attenuated when REDUCE-IT is excluded. Furthermore, the benefits of marine-derived omega-3 fatty acids, particularly high-dose EPA, appear to exceed their lipid-lowering effects.
N.A.M. contributed to study design, literature search, data collection, statistical analysis, figures, data interpretation, and drafting of the manuscript. R.P.G. contributed to study design, data interpretation, and critical review of the manuscript. K.I. contributed to statistical analysis, figures, and critical review of the manuscript. M.G.S. contributed to data collection, data interpretation, and critical review of the manuscript. M.L.O. andS.D.W. contributed to data interpretation and critical review of the manuscript. B.A.F. contributed to study design, data interpretation, and critical review of the manuscript. M.S.S. contributed to study design, literature search, data collection, statistical analysis, figures, data interpretation, and critical review of the manuscript. M.S.S. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Sources of Funding
Dr Marston reports a research grant from the National Institutes of Health National Research Service Award. Dr Giugliano reports research grants from Amgen; honoraria from Amgen, Daiichi Sankyo, and Merck; and consultant fees from Amgen, Akcea, Amarin, Boehringer-Ingelheim, Bristol-Myers-Squibb, CVS Caremark, Daiichi Sankyo, GlaxoSmithKline, Lexicon, Merck, Portola, and Pfizer. Drs Giugliano and Im report an institutional research grant to the TIMI Study Group at Brigham and Women’s Hospital for research they are not directly involved in from AstraZeneca, Bayer, Eisai, GlaxoSmithKline, Intarcia, Janssen Research and Development, Medicines Company, MedImmune, Novartis, Poxel, Pfizer, Quark Pharmaceuticals, and Takeda. Dr Silverman reports grants from the John S. LaDue Memorial Fellowship from Harvard Medical School and from the ZOLL Foundation. Dr O’Donoghuereports institutional research grants from Amgen, Janssen, The Medicines Company, Eisai, GlaxoSmithKline, and Astra Zeneca. Dr Wiviott reports research grants from Amgen, Arena, AstraZeneca, Bristol-Myers-Squibb, Daiichi Sankyo, Eisai, Eli Lilly, Janssen, Merck, and Sanofi-Aventis, as well as consultant/advisory board fees from Arena, AstraZeneca, Aegerion, Merck, Allergan, Angelmed, Boehringer-Ingelheim, Boston Clinical Research Institute, Bristol-Myers-Squibb, Daiichi Sankyo, Eisai, Eli Lilly, Icon Clinical, Janssen, Lexicon, St Jude Medical, and Xoma. Dr Sabatine reports research grant support from Abbott Laboratories, Amgen, AstraZeneca, Bayer, Critical Diagnostics, Daiichi Sankyo, Eisai, Genzyme, Gilead, GlaxoSmithKline, Intarcia, Janssen Research and Development, The Medicines Company, MedImmune, Merck, Novartis, Poxel, Pfizer, Quark Pharmaceuticals, Roche Diagnostics, and Takeda, as well as consulting fees from Alnylam, Amgen, AstraZeneca, Bristol-Myers-Squibb, CVS Caremark, Dyrnamix, Esperion, IFM Pharmaceuticals, Intarcia, Ionis, Janssen Research and Development, The Medicines Company, MedImmune, Merck, MyoKardia, and Novartis. The other authors report no conflicts.
Nordestgaard BG, Benn M, Schnohr P, Tybjaerg-Hansen A. Nonfasting triglycerides and risk of myocardial infarction, ischemic heart disease, and death in men and women.JAMA. 2007; 298:299–308. doi: 10.1001/jama.298.3.299CrossrefMedlineGoogle Scholar
Sarwar N, Danesh J, Eiriksdottir G, Sigurdsson G, Wareham N, Bingham S, Boekholdt SM, Khaw KT, Gudnason V. Triglycerides and the risk of coronary heart disease: 10,158 incident cases among 262,525 participants in 29 Western prospective studies.Circulation. 2007; 115:450–458. doi: 10.1161/CIRCULATIONAHA.106.637793LinkGoogle Scholar
Miller M, Cannon CP, Murphy SA, Qin J, Ray KK, Braunwald E; PROVE IT-TIMI 22 Investigators. Impact of triglyceride levels beyond low-density lipoprotein cholesterol after acute coronary syndrome in the PROVE IT-TIMI 22 trial.J Am Coll Cardiol. 2008; 51:724–730. doi: 10.1016/j.jacc.2007.10.038CrossrefMedlineGoogle Scholar
- 4. Triglyceride Coronary Disease Genetics Consortium and Emerging Risk Factors Collaboration,
Sarwar N, Sandhu MS, Ricketts SL, Butterworth AS, Di Angelantonio E, Boekholdt SM, Ouwehand W, Watkins H, Samani MJ, Salaheen D,. Triglyceride-mediated pathways and coronary disease: collaborative analysis of 101 studies.Lancet. 2010; 375:1634–1639.CrossrefMedlineGoogle Scholar
Miller M, Stone NJ, Ballantyne C, Bittner V, Criqui MH, Ginsberg HN, Goldberg AC, Howard WJ, Jacobson MS, Kris-Etherton PM,; American Heart Association Clinical Lipidology, Thrombosis, and Prevention Committee of the Council on Nutrition, Physical Activity, and Metabolism; Council on Arteriosclerosis, Thrombosis and Vascular Biology; Council on Cardiovascular Nursing; Council on the Kidney in Cardiovascular Disease. Triglycerides and cardiovascular disease: a scientific statement from the American Heart Association.Circulation. 2011; 123:2292–2333. doi: 10.1161/CIR.0b013e3182160726LinkGoogle Scholar
Do R, Willer CJ, Schmidt EM, Sengupta S, Gao C, Peloso GM, Gustafsson S, Kanoni S, Ganna A, Chen J,. Common variants associated with plasma triglycerides and risk for coronary artery disease.Nat Genet. 2013; 45:1345–1352. doi: 10.1038/ng.2795CrossrefMedlineGoogle Scholar
Teslovich TM, Musunuru K, Smith AV, Edmondson AC, Stylianou IM, Koseki M, Pirruccello JP, Ripatti S, Chasman DI, Willer CJ,. Biological, clinical and population relevance of 95 loci for blood lipids.Nature. 2010; 466:707–713. doi: 10.1038/nature09270CrossrefMedlineGoogle Scholar
Willer CJ, Schmidt EM, Sengupta S, Peloso GM, Gustafsson S, Kanoni S, Ganna A, Chen J, Buchkovich ML, Mora S,; Global Lipids Genetics Consortium. Discovery and refinement of loci associated with lipid levels.Nat Genet. 2013; 45:1274–1283. doi: 10.1038/ng.2797CrossrefMedlineGoogle Scholar
Varbo A, Benn M, Tybjærg-Hansen A, Jørgensen AB, Frikke-Schmidt R, Nordestgaard BG. Remnant cholesterol as a causal risk factor for ischemic heart disease.J Am Coll Cardiol. 2013; 61:427–436. doi: 10.1016/j.jacc.2012.08.1026CrossrefMedlineGoogle Scholar
Varbo A, Benn M, Tybjærg-Hansen A, Nordestgaard BG. Elevated remnant cholesterol causes both low-grade inflammation and ischemic heart disease, whereas elevated low-density lipoprotein cholesterol causes ischemic heart disease without inflammation.Circulation. 2013; 128:1298–1309. doi: 10.1161/CIRCULATIONAHA.113.003008LinkGoogle Scholar
Aung T, Halsey J, Kromhout D, Gerstein HC, Marchioli R, Tavazzi L, Geleijnse JM, Rauch B, Ness A, Galan P,; Omega-3 Treatment Trialists’ Collaboration. Associations of omega-3 fatty acid supplement use with cardiovascular disease risks: meta-analysis of 10 trials involving 77 917 individuals.JAMA Cardiol. 2018; 3:225–234. doi: 10.1001/jamacardio.2017.5205CrossrefMedlineGoogle Scholar
Kotwal S, Jun M, Sullivan D, Perkovic V, Neal B. Omega 3 fatty acids and cardiovascular outcomes: systematic review and meta-analysis.Circ Cardiovasc Qual Outcomes. 2012; 5:808–818. doi: 10.1161/CIRCOUTCOMES.112.966168LinkGoogle Scholar
- 13. Clofibrate and niacin in coronary heart disease.JAMA. 1975; 231:360–381.CrossrefMedlineGoogle Scholar
- 14. A co-operative trial in the primary prevention of ischaemic heart disease using clofibrate: report from the Committee of Principal Investigators.Br Heart J. 1978; 40:1069–1118.CrossrefMedlineGoogle Scholar
Rubins HB, Robins SJ, Collins D, Fye CL, Anderson JW, Elam MB, Faas FH, Linares E, Schaefer EJ, Schectman G,. Gemfibrozil for the secondary prevention of coronary heart disease in men with low levels of high-density lipoprotein cholesterol: Veterans Affairs High-Density Lipoprotein Cholesterol Intervention Trial Study Group.N Engl J Med. 1999; 341:410–418. doi: 10.1056/NEJM199908053410604CrossrefMedlineGoogle Scholar
- 16. Bezafibrate Infarction PreventionStudy.Secondary prevention by raising HDL cholesterol and reducing triglycerides in patients with coronary artery disease.Circulation. 2000; 102:21–27.LinkGoogle Scholar
- 17. Effect of fenofibrate on progression of coronary-artery disease in type 2 diabetes: the Diabetes Atherosclerosis Intervention Study, a randomised study.Lancet. 2001; 357:905–910.CrossrefMedlineGoogle Scholar
Ginsberg HN, Elam MB, Lovato LC, Crouse JR, Leiter LA, Linz P, Friedewald WT, Buse JB, Gerstein HC, Probstfield J,; ACCORD Study Group. Effects of combination lipid therapy in type 2 diabetes mellitus.N Engl J Med. 2010; 362:1563–1574. doi: 10.1056/NEJMoa1001282CrossrefMedlineGoogle Scholar
Frick MH, Elo O, Haapa K, Heinonen OP, Heinsalmi P, Helo P, Huttunen JK, Kaitaniemi P, Koskinen P, Manninen V. Helsinki Heart Study: primary-prevention trial with gemfibrozil in middle-aged men with dyslipidemia: safety of treatment, changes in risk factors, and incidence of coronary heart disease.N Engl J Med. 1987; 317:1237–1245. doi: 10.1056/NEJM198711123172001CrossrefMedlineGoogle Scholar
Meade T, Zuhrie R, Cook C, Cooper J. Bezafibrate in men with lower extremity arterial disease: randomised controlled trial.BMJ. 2002; 325:1139. doi: 10.1136/bmj.325.7373.1139CrossrefMedlineGoogle Scholar
Keech A, Simes RJ, Barter P, Best J, Scott R, Taskinen MR, Forder P, Pillai A, Davis T, Glasziou P,; FIELD Study Investigators. Effects of long-term fenofibrate therapy on cardiovascular events in 9795 people with type 2 diabetes mellitus (the FIELD study): randomised controlled trial.Lancet. 2005; 366:1849–1861. doi: 10.1016/S0140-6736(05)67667-2CrossrefMedlineGoogle Scholar
Boden WE, Probstfield JL, Anderson T, Chaitman BR, Desvignes-Nickens P, Koprowicz K, McBride R, Teo K, Weintraub W; AIM-HIGH Investigators. Niacin in patients with low HDL cholesterol levels receiving intensive statin therapy.N Engl J Med. 2011; 365:2255–2267. doi: 10.1056/NEJMoa1107579CrossrefMedlineGoogle Scholar
Landray MJ, Haynes R, Hopewell JC, Parish S, Aung T, Tomson J, Wallendszus K, Craig M, Jiang L, Collins R,; HPS2-THRIVE Collaborative Group. Effects of extended-release niacin with laropiprant in high-risk patients.N Engl J Med. 2014; 371:203–212. doi: 10.1056/NEJMoa1300955CrossrefMedlineGoogle Scholar
- 24. Dietary supplementation with n-3 polyunsaturated fatty acids and vitamin E after myocardial infarction: results of the GISSI-Prevenzione trial: Gruppo Italiano per lo Studio della Sopravvivenza nell’Infarto miocardico.Lancet. 1999; 354:447–455.CrossrefMedlineGoogle Scholar
Tavazzi L, Maggioni AP, Marchioli R, Barlera S, Franzosi MG, Latini R, Lucci D, Nicolosi GL, Porcu M, Tognoni G; Gissi-HF Investigators. Effect of n-3 polyunsaturated fatty acids in patients with chronic heart failure (the GISSI-HF trial): a randomised, double-blind, placebo-controlled trial.Lancet. 2008; 372:1223–1230. doi: 10.1016/S0140-6736(08)61239-8CrossrefMedlineGoogle Scholar
Kromhout D, Giltay EJ, Geleijnse JM; Alpha Omega Trial Group. n-3 Fatty acids and cardiovascular events after myocardial infarction.N Engl J Med. 2010; 363:2015–2026. doi: 10.1056/NEJMoa1003603CrossrefMedlineGoogle Scholar
Einvik G, Klemsdal TO, Sandvik L, Hjerkinn EM. A randomized clinical trial on n-3 polyunsaturated fatty acids supplementation and all-cause mortality in elderly men at high cardiovascular risk.Eur J Cardiovasc Prev Rehabil. 2010; 17:588–592. doi: 10.1097/HJR.0b013e328339cc70CrossrefMedlineGoogle Scholar
Bosch J, Gerstein HC, Dagenais GR, Díaz R, Dyal L, Jung H, Maggiono AP, Probstfield J, Ramachandran A, Riddle MC,; ORIGIN Trial Investigators. n-3 Fatty acids and cardiovascular outcomes in patients with dysglycemia.N Engl J Med. 2012; 367:309–318. doi: 10.1056/NEJMoa1203859CrossrefMedlineGoogle Scholar
Rauch B, Schiele R, Schneider S, Diller F, Victor N, Gohlke H, Gottwik M, Steinbeck G, Del Castillo U, Sack R,; OMEGA Study Group. OMEGA, a randomized, placebo-controlled trial to test the effect of highly purified omega-3 fatty acids on top of modern guideline-adjusted therapy after myocardial infarction.Circulation. 2010; 122:2152–2159. doi: 10.1161/CIRCULATIONAHA.110.948562LinkGoogle Scholar
Galan P, Kesse-Guyot E, Czernichow S, Briancon S, Blacher J, Hercberg S; SU.FOL.OM3 Collaborative Group. Effects of B vitamins and omega 3 fatty acids on cardiovascular diseases: a randomised placebo controlled trial.BMJ. 2010; 341:c6273. doi: 10.1136/bmj.c6273CrossrefMedlineGoogle Scholar
Roncaglioni MC, Tombesi M, Avanzini F, Barlera S, Caimi V, Longoni P, Marzona I, Milani V, Silletta MG, Tognoni G,; Risk and Prevention Study Collaborative Group. n-3 Fatty acids in patients with multiple cardiovascular risk factors.N Engl J Med. 2013; 368:1800–1808. doi: 10.1056/NEJMoa1205409CrossrefMedlineGoogle Scholar
Yokoyama M, Origasa H, Matsuzaki M, Matsuzawa Y, Saito Y, Ishikawa Y, Oikawa S, Sasaki J, Hishida H, Itakura H,; Japan EPA Lipid Intervention Study (JELIS) Investigators. Effects of eicosapentaenoic acid on major coronary events in hypercholesterolaemic patients (JELIS): a randomised open-label, blinded endpoint analysis.Lancet. 2007; 369:1090–1098. doi: 10.1016/S0140-6736(07)60527-3CrossrefMedlineGoogle Scholar
Bonds DE, Harrington M, Worrall BB, Bertoni AG, Eaton CB, Hsia J, Robinson J, Clemons TE, Fine LJ, Chew EY; Writing Group for the AREDS2 Research Group. Effect of long-chain ω-3 fatty acids and lutein + zeaxanthin supplements on cardiovascular outcomes: results of the Age-Related Eye Disease Study 2 (AREDS2) randomized clinical trial.JAMA Intern Med. 2014; 174:763–771. doi: 10.1001/jamainternmed.2014.328CrossrefMedlineGoogle Scholar
Bowman L, Mafham M, Wallendszus K, Stevens W, Buck G, Barton J, Murphy K, Aung T, Haynes R, Cox J,; ASCEND Study Collaborative Group. Effects of aspirin for primary prevention in persons with diabetes mellitus.N Engl J Med. 2018; 379:1529–1539. doi: 10.1056/NEJMoa1804988CrossrefMedlineGoogle Scholar
Ference BA, Kastelein JJP, Ray KK, Ginsberg HN, Chapman MJ, Packard CJ, Laufs U, Oliver-Williams C, Wood AM, Butterworth AS,. Association of triglyceride-lowering LPL variants and LDL-C-lowering LDLR variants with risk of coronary heart disease.JAMA. 2019; 321:364–373. doi: 10.1001/jama.2018.20045CrossrefMedlineGoogle Scholar
Manson JE, Cook NR, Lee IM, Christen W, Bassuk SS, Mora S, Gibson H, Albert CM, Gordon D, Copeland T,; VITAL Research Group. Marine n-3 fatty acids and prevention of cardiovascular disease and cancer.N Engl J Med. 2019; 380:23–32. doi: 10.1056/NEJMoa1811403CrossrefMedlineGoogle Scholar
Bhatt DL, Steg PG, Miller M, Brinton EA, Jacobson TA, Ketchum SB, Doyle RT, Juliano RA, Jiao L, Granowitz C,; REDUCE-IT Investigators. Cardiovascular risk reduction with icosapent ethyl for hypertriglyceridemia.N Engl J Med. 2019; 380:11–22. doi: 10.1056/NEJMoa1812792CrossrefMedlineGoogle Scholar
Acharya P, Talahalli RR. Aging and hyperglycemia intensify dyslipidemia-induced oxidative stress and inflammation in rats: assessment of restorative potentials of ALA and EPA + DHA.Inflammation. 2019; 42:946–952. doi: 10.1007/s10753-018-0949-6CrossrefMedlineGoogle Scholar
Tribulova N, Szeiffova Bacova B, Egan Benova T, Knezl V, Barancik M, Slezak J. Omega-3 index and anti-arrhythmic potential of omega-3 PUFAs.Nutrients. 2017; 9:1191.CrossrefGoogle Scholar
Silverman MG, Ference BA, Im K, Wiviott SD, Giugliano RP, Grundy SM, Braunwald E, Sabatine MS. Association between lowering LDL-C and cardiovascular risk reduction among different therapeutic interventions: a systematic review and meta-analysis.JAMA. 2016; 316:1289–1297. doi: 10.1001/jama.2016.13985CrossrefMedlineGoogle Scholar
Moher D, Liberati A, Tetzlaff J, Altman DG; PRISMA Group. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement.Ann Intern Med. 2009; 151:264–269, W64. doi: 10.7326/0003-4819-151-4-200908180-00135CrossrefMedlineGoogle Scholar
Knapp G, Hartung J. Improved tests for a random effects meta-regression with a single covariate.Stat Med. 2003; 22:2693–2710. doi: 10.1002/sim.1482CrossrefMedlineGoogle Scholar
Viechtbauer W. Conducting meta-analyses in R with the metafor package.J Stat Softw. 2010; 36:1–48.CrossrefGoogle Scholar
Viechtbauer W, Cheung MW. Outlier and influence diagnostics for meta-analysis.Res Synth Methods. 2010; 1:112–125. doi: 10.1002/jrsm.11CrossrefMedlineGoogle Scholar
Cook RD, Weisberg S. Residuals and Influence in Regression. New York, NY: Chapman and Hall; 1982.Google Scholar
Hoaglin DC, Welsch RE. The hat matrix in regression and ANOVA.Am Statistician. 1978; 32:17–22.Google Scholar
Higgins JP, Altman DG, Gøtzsche PC, Jüni P, Moher D, Oxman AD, Savovic J, Schulz KF, Weeks L, Sterne JA; Cochrane Bias Methods Group; Cochrane Statistical Methods Group. The Cochrane Collaboration’s tool for assessing risk of bias in randomised trials.BMJ. 2011; 343:d5928. doi: 10.1136/bmj.d5928CrossrefMedlineGoogle Scholar
Egger M, Davey Smith G, Schneider M, Minder C. Bias in meta-analysis detected by a simple, graphical test.BMJ. 1997; 315:629–634. doi: 10.1136/bmj.315.7109.629CrossrefMedlineGoogle Scholar
Preston Mason R. New insights into mechanisms of action for omega-3 fatty acids in atherothrombotic cardiovascular disease.Curr Atheroscler Rep. 2019; 21:2. doi: 10.1007/s11883-019-0762-1CrossrefMedlineGoogle Scholar
Ridker PM, Everett BM, Thuren T, MacFadyen JG, Chang WH, Ballantyne C, Fonseca F, Nicolau J, Koenig W, Anker SD,; CANTOS Trial Group. Antiinflammatory therapy with canakinumab for atherosclerotic disease.N Engl J Med. 2017; 377:1119–1131. doi: 10.1056/NEJMoa1707914CrossrefMedlineGoogle Scholar