Omega-3 Blood Levels and Stroke Risk: A Pooled and Harmonized Analysis of 183 291 Participants From 29 Prospective Studies.
Dec 22, 2023DOI: 10.1161/strokeaha.123.044281Journal ArticleAHA
Written by: James H O'KeefeNathan L TintleWilliam S HarrisEvan L O'KeefeAleix Sala-VilaJohn AttiaG Manohar GargAlexis HureChristian Sørensen BorkErik Berg SchmidtStine Krogh VenøKuo-Liong ChienYun-Yu Amelia ChenSarah EgertTobias Rudholm FeldreichJohan ÄrnlövLars LindNita G ForouhiJohanna M GeleijnseKamalita PertiwiFumiaki ImamuraVanessa de Mello LaaksonenW Matti UusitupaJaakko TuomilehtoMarkku LaaksoMaria Anneli LankinenDanielle LaurinPierre-Hugues CarmichaelJoan LindsayKarin LeanderFederica LaguzziBrenton R SwensonWilliam T LongstrethJoAnn E MansonSamia MoraNancy R CookMatti MarklundDebora Melo van LentRachel MurphyVilmundur GudnasonToshihara NinomiyaYoichiro HirakawaFrank QianQi SunFrank HuAndres V Ardisson KoratUlf RisérusIolanda LázaroCecilia SamieriMélanie Le GoffCatherine HelmerMarinka SteurTrudy VoortmanM Kamran IkramToshiko TanakaJayanta K DasLuigi FerrucciStefania BandinelliMichael TsaiWeihua GuanParveen GargW M Monique VerschurenJolanda M A BoerAnneke BiokstraJyrki VirtanenMichael WagnerJason WestraLuc AlbuissonKazumasa YamagishiDavid S SiscovickRozenn N LemaitreDariush Mozaffarian
According to a 2021 worldwide analysis, 1 in 4 adults will suffer a stroke in their lifetime, and it is the second-leading cause of death and the third-leading cause of death and disability combined.1 To reduce the risk of ASCVD and ischemic stroke, nutritional approaches have historically focused on high fiber diets that are low in sodium, saturated fat and cholesterol. Consumption of marine omega-3 PUFAs, however, has also shown promise in the prevention of CVD.Docosahexaenoic (DHA) and eicosapentaenoic (EPA) acids have been widely studied since the 1970s when these PUFAs were reported to be inversely associated with risk of acute MI among Greenland Inuits.2 Over the ensuing 4 decades, intensive scientific investigation established the cardioprotective effects of EPA and DHA.3,4 A meta-analysis of 14 prospective studies used food frequency questionnaires to estimate self-reported omega-3 PUFA intake among 514,483 individuals with 9065 strokes during follow-up.5 Compared with the lowest omega-3 intake group in each study, individuals in the highest intake group had a 13% lower risk of total stroke and a 16% lower risk for fatal stroke.5 A widely accepted limitation of studies based on self-reporting, however, is the ability to accurately estimate exposure to omega-3 PUFAs. Recently, a meta-analysis of 38 RCTs that included 149,051 people reported that marine omega-3 PUFA supplementation was associated with statistically significant reductions in CV mortality, major adverse CV events (composite endpoint of nonfatal MI, nonfatal stroke, or CV death), and coronary revascularization.6 In that metanalysis, omega-3 was not associated with reduction in risk of stroke as a separate endpoint, however, only 927 strokes (0.6% of the study participants) occurred during the relatively short median follow-up of 2.0 years. Meta-analyses have also reported pharmacological intervention with marine omega-3 PUFAs may increase risk for AF.6,7 Without question, AF increases risk of stroke,8 but the effects of omega-3 PUFAs on stroke risk remain uncertain.An objective measure of omega-3 PUFA exposure is the level of EPA+DHA in blood or tissues, which is largely determined by habitual dietary intake of these PUFAs. 9–12 While biomarker-based investigations have strong potential to clarify the biological relationship between omega-3 PUFAs and stroke outcomes, reports from single studies can be limited by insufficient power, selection bias, publication bias, and inconsistent definitions of exposure, outcome, and adjustment for potential confounding factors. To address these challenges, the present study pooled de novo, harmonized, individual-level analyses across 29 prospective studies in the Fatty Acid and Outcome Research Consortium (FORCE)13 to investigate the associations between marine omega-3 PUFA biomarker levels and incident stroke—total, ischemic, and hemorrhagic.
Methods
Authors will make data, analytic methods, and study materials available to other researchers upon request. This study was conducted within the FORCE consortium13 − an international collaboration of observational studies with baseline PUFA biomarker data and ascertainment of chronic disease events during follow-up. For the current project, all 60 prospective studies in the consortium as of November 2021 were invited to participate. Of these, 18 did not have the required PUFA and/or stroke data, 11 indicated a lack of funding/analyst time and two had too few incident stroke outcomes for adjusted statistical models to fit robustly. Thus, the current investigation comprised data from 29 studies across 15 countries. The details of each individual study are presented in Supplemental Table 1, including individual cohort IRB approvals. All participating studies followed a prespecified standardized analysis protocol with harmonized inclusion and exclusion criteria, exposures, outcomes, covariates, and analytical methods inclusive of statistical models and assessment of missing covariate data. In each study, new analyses of individual data were performed according to the protocol, and study-specific results were collected using a standardized electronic form. Information regarding registration required by any of the cohorts included herein is shown in Supplemental Table 1. Individual cohorts conducted their studies in accordance with the criteria set by the Declaration of Helsinki, and informed consent was obtained from all participants after IRB approval. Details of other methods including participant criteria, fatty acid analysis, covariates, and statistical analysis and pooling are given in the Supplemental Methods section.
RESULTS
The pooled analyses of 29 cohorts included marine omega-3 PUFA biomarker levels from 183,291 individuals (Supplemental Table 2), 67,165 excluding the UKBB. The median follow-up time was 14 years (range of 5 to 30 years). In total, 10,561 participants were recorded as incident stroke cases: 78% ischemic, 11% hemorrhagic and 11% unspecified. At baseline, the mean age was 65 years, 53% were women; White adults comprised 82% of the study population. Total stroke in relation to each PUFA exposure was analyzed in two models−those with and without pre-existing CVD (Supplementary Table 3). The distribution (10th, 50th and 90th percentiles) of PUFA levels by lipid compartment and by cohort are shown in Supplemental Table 4. In the pooled data, the estimated 10th, 50th and 90th percentiles for RBC EPA+DHA content were 3.4%, 5.2% and 7.9%, respectively as shown in Supplemental Table 5.
When analyzing by quintile, for DHA (including the UKBB cohort) a significant inverse association was observed for total stroke, with the risk in Q4 and Q5 being 12–13% lower than the reference group, Q1 (Figure 1). Similar patterns were seen for EPA and EPA+DHA in analyses without the UKBB with the risk in Q5 17% lower than that in Q1 (Table 1). The quintile analysis of associations between DPA and total stroke (again, excluding the UKBB) suggested a threshold effect, with no associations in Q2-Q4 but an 11% lower risk in Q5 vs. Q1 (Table 1). These analyses showed low levels of heterogeneity (I2<50%) in all cases. When analyzing the relations between DHA, EPA and their sum on a per IQ5R basis, each was associated with 8%−9% lower risk of total stroke (Supplemental Table 6). DHA results were similar with or without inclusion of the UKBB cohort (Supplemental Table 6). DPA was not significantly associated with risk of stroke in this analysis.
Patterns for ischemic stroke followed those of total stroke, with risk in Q5 versus Q1 being 14–18% lower (Table 1). In contrast to ischemic stroke, results for hemorrhagic stroke showed little to no evidence of differential risk for any omega-3 PUFA tested. In IQ5R analyses, risk for ischemic stroke was reduced by 11–13% for DHA and EPA+DHA (Supplemental Table 6). No significant associations with DPA were observed for either stroke subtype.
Regarding DHA, risk for total stroke comparing Q5 with Q1 ranged from 6% in RBCs to 21% for cholesteryl esters (Figure 2). Similar trends were seen for total stroke with EPA and EPA+DHA, but not with DPA (Supplemental Figures 1-3). DHA levels were also inversely related to risk for ischemic stroke (Supplemental Figure 4), but not with hemorrhagic stroke (Supplemental Figure 5).
As shown in Table 1, linear trends across quintiles showed a significant inverse relationship for each of DHA, EPA and DHA+EPA for both total stroke and ischemic stroke (P<0.05), but not hemorrhagic stroke (P>0.05). All tests for non-linearity (for each PUFA and stroke outcome) yielded p-values >0.05. (Supplemental Table 7).
In exploratory subgroup analyses, PUFA by age (< 65 versus >65), sex (male vs. female) and prevalent AF (Yes versus No), no significant effect modification was seen for omega-3 PUFA biomarker relationships with total stroke (using Bonferroni-corrected alpha=0.004, based on 3 subgroups x 4 PUFAs) (Supplemental Table 8). Results were also similar after excluding individuals with prevalent CVD (Supplemental Table 9) and censoring all participants at 10-years of follow-up (Supplemental Table 10). Because data on DPA were available from only 21 of the 29 cohorts, a post-hoc sensitivity analysis was conducted to assess relationships between DHA, EPA and their sum in only those cohorts providing data on DPA in order to compare the findings seen with the full dataset. In these analyses, similar significant, inverse associations with DHA, EPA and EPA+DHA were observed, suggesting that the lack of significant associations with DPA were not simply driven by the smaller sample size (Supplemental Table 11). Because of its larger N (N=116,126 participants; 2478 stroke cases), Supplemental Table 12 shows the results for UKBB alone. These findings are similar to those seen with the combined cohorts, with the exception that DHA levels were significantly and inversely related to risk for hemorrhagic stroke in the UKBB.
Discussion
In this harmonized, pooled, de novo analysis of data from up to 183,291 people in 29 prospective studies from 15 nations, in vivo DHA, EPA and EPA+DHA levels were inversely associated with risk of total and ischemic stroke during an average of 14 years of follow-up. Comparing Q5 with Q1, risk of ischemic stroke was 18% lower for EPA+DHA, 18% lower for EPA, and 16% lower for DHA. Omega-3 levels were not significantly associated with risk of hemorrhagic stroke. In addition, the results were robust when excluding individuals with prevalent CVD, diminishing concerns about reverse causation. DPA is a minor omega-3 PUFA that is not influenced by dietary intake of fish, seafood or fish oil, and it was not significantly associated with stroke risk in the current study.
Scientific plausibility exists to support the hypothesis that marine omega-3 PUFA might reduce susceptibility to stroke. DHA and EPA have anti-inflammatory, antiplatelet and hypotriglyceridemic effects, reduce arterial stiffness, improve endothelial function, provide dose-dependent reductions in blood pressure and resting heart rate via heightened vagal tone.3,4,14 Stroke has an ischemic etiology in about 87% of cases in the US, generally due to atherosclerotic cerebrovascular arteries or embolic events.15 Low blood levels of EPA and DHA have been linked with lipid-rich, inflamed plaques with thin, vulnerable, fibrous caps that are at increased risk of rupturing.16 In a study of 188 patients awaiting carotid endarterectomy, those randomized to fish oil (providing 1.4 grams of DHA + EPA per day vs placebo) prior to surgery had a lower prevalence of vulnerable fibrous caps, an increased thickness of fibrous caps and reduced signs of inflammation within the carotid plaques.17 The authors concluded, “Atherosclerotic plaques readily incorporate omega-3s from fish-oil supplementation, inducing changes that can enhance stability of atherosclerotic plaques”.17
Because ASCVD complications such as acute coronary syndrome and ischemic stroke are triggered by inflammation, plaque erosion/rupture and thrombosis, the disease is best described as atherothrombosis.18 Marine omega-3 PUFAs inhibit platelet aggregation, reduce both whole blood viscosity and von Willebrand factor and improve RBC membrane flexibility/deformability.19 These mechanisms of action might be playing a role in omega-3-mediated protection against ischemic stroke of atherothrombotic or cardioembolic origin.
Since stroke remains the most feared complication of AF, this study was undertaken in part due to concerns that high doses of prescription omega-3 PUFAs may increase risk of AF.7,20 Yet, the relationship of omega-3 and AF is controversial as previous studies examining standard consumption levels of omega-3 have linked higher dietary intakes of DHA + EPA with lower incidence of AF.21 This observation was recently supported by a biomarker-based analysis from the FORCE group.22 The reduced risk of stroke demonstrated in the present study is consistent with the findings of REDUCE-IT, and neither a history of AF nor pre-existing CVD confounded the inverse associations between levels of omega-3 and risk for stroke.23
High doses of omega-3 PUFA can inhibit platelet function and could possibly increase risk for bleeding and hemorrhagic stroke. Greenland Inuits consuming their traditional diet that was very high in marine omega-3 PUFAs (about 14 g/d DHA+EPA) had increased mortality from hemorrhagic stroke compared with Inuits living in Denmark.24,25 In the Nurses’ Health Study, women with relatively high intakes of fish, still much lower than that of the Inuits, had a lower risk of ischemic stroke with no significant association with risk of hemorrhagic stroke.26 Very high doses (>7 g/d) of omega-3 PUFAs, especially EPA, may increase bleeding times and hemorrhagic complications.14,19,22 In REDUCE-IT, serious bleeding events occurred in 2.7% of the people randomized to EPA 4 g/d group and in 2.1% in the placebo group (P=0.06), but the EPA group did not have an increased risk of hemorrhagic stroke.23 Observational cohorts in the US and Japan, using food frequency questionnaire, found no associations of fish or omega-3 intakes with risk of hemorrhagic stroke.26,27 In the current study, individuals in the top quintile of omega-3 PUFA levels showed no signal for increased risk of hemorrhagic stroke. In another observational study, a lower prevalence of cerebral microbleeds was noted among older adults eating large amounts of oily fish (13 servings per week, equivalent to about 2 g/d of EPA+DHA per day).28 The Omega-3 PUFAs for Prevention of Post-operative Atrial Fibrillation trial randomized 1516 patients prior to open-heart surgery to either 2 g/d of EPA+DHA or matching placebo capsules. Omega-3 did not increase the perioperative bleeding risk after cardiac surgery and, surprisingly, reduced the number of blood transfusions needed postoperatively.29 The current study, which stands as the largest omega-3 biomarker study to date, showed no signal for increased risk of hemorrhagic stroke.
Our findings support the American Heart Association’s Science Advisory recommendation that, “1 to 2 seafood meals per week be included to reduce the risk of congestive heart failure, coronary heart disease, ischemic stroke, and sudden cardiac death.” The typical US adult eats less than 1 serving per week of fish/seafood; accordingly, the average intake of EPA+DHA in the US is only about 100 mg/d with a mean omega-3 index of 5.4%.30 The median RBC EPA+DHA (i.e., the omega-3 index) for Q5 in the present study was approximately 8%, suggesting that reaching an omega-3 index of 8% or greater could be a cardioprotective goal−originally proposed in 2004.31 To raise an omega-3 index of 5.4% to 8% would require intake of approximately 1000 mg/d of EPA+DHA, whereas to go from the median of Q1 (about 3.5% omega-3 index) to 8% would require about 1600 mg/d of EPA+DHA.30 These intakes are achievable from dietary fish/seafood and/or omega-3 supplements.
Strengths of the current study include the use of objective omega-3 PUFA biomarkers, rather than dietary questionnaires, which increases the accuracy of exposure assessment and allows for separate statistical analyses of individual omega-3 PUFA-stroke outcomes. The use of prespecified, harmonized, de novo individual-level analyses across multiple cohorts substantively increases generalizability, reduces confounding through consistent adjustment for covariates and limits the potential for publication bias. The pooling of 29 studies including over 10,000 incident stroke cases strengthens the generalizability of the findings, and allowed for the statistical evaluation of stroke subtypes as well as potential heterogeneity across subgroups that may modify the observed associations.
Potential limitations of this study include limited diversity—most individuals were White, which could lower generalizability to other races/ethnicities. Despite extensive efforts to harmonize study-specific methods, moderate heterogeneity remained among studies, which may be due to study designs, unmeasured background population characteristics, differences in laboratory assessment of omega-3 PUFAs, variability in ascertainment of outcomes, chance, or any combination of these. Omega-3 PUFAs and covariates were measured once at baseline and changes over time could lead to misclassification, which could bias the results in unpredictable directions. Even so, omega-3 concentrations have shown reasonably good reproducibility over time.32 Pooling by quintiles instead of absolute PUFA values was necessary because values physiologically differ by lipid compartment. Nevertheless, such an approach was reasonable given the observed correlations among different lipid compartments. For example, correlation coefficients were >0.9 for plasma phospholipid and RBC EPA+DHA levels,31 for EPA and DHA in plasma CE and PL fractions33 and for whole plasma vs RBC DHA+EPA (N=2312, WS Harris, unpublished data). Beyond classification as ischemic or hemorrhagic, we did not have specific information on whether strokes were due to emboli, large artery atherosclerosis, microganiopathy, hypertension, etc. The analysis plan did not request information from cohorts about demographics of AF patients, so interaction analyses could not performed on this subgroup. Although we adjusted for many major demographic and socioeconomic risk factors, physical activity, smoking and prevalent diseases, some covariates were self-reported; so residual and unmeasured confounding could bias our results in unknown directions. Still, the magnitude of the observed relationships between omega-3 PUFA levels and risk of stroke reported herein is consistent with the known associations of EPA+DHA with CHD events, cardiac mortality, all-cause mortality, and sudden cardiac death.6,11,34–36
In summary, this harmonized and pooled analysis of prospective studies showed that long chain omega-3 PUFA levels were inversely associated with risk of total and ischemic stroke but were unrelated to risk of hemorrhagic stroke. Thus, higher dietary intakes of DHA and EPA would be expected to lower risk for stroke.
JO
Dr. James O'Keefe
Internal Medicine · Primary CareExcel Health Primary Carejameso