BMI Affects the Relationship between Long Chain N-3 Polyunsaturated Fatty Acid Intake and Stroke Risk: a Meta-Analysis

We performed a meta-analysis to clarify the relationship between long chain n-3 polyunsaturated fatty acid (PUFA) intake and stroke risk. Relevant studies were identified by searching online databases through May 2015. Log relative risks (RRs) of the highest versus the lowest for cohort studies were weighed by the inverse variance method to obtain pooled RRs. Fourteen prospective cohort studies including 514,483 individuals and 9,065 strokes were included. The pooled RR of overall stroke risk for long chain n-3 PUFA intake was 0.87 [95% confidence interval (CI), 0.79–0.95]. Stratification analysis showed that higher long chain n-3 PUFAs intake was associated with reduced fatal stroke risk (RR = 0.84; 95% CI, 0.73–0.97), reduced stroke risk for BMI < 24 (RR = 0.86; 95% CI, 0.75–0.98) and reduced stroke risk for females (RR = 0.81; 95% CI, 0.71–0.92), but was not associated with stroke risk for either BMI ≥ 24 or men. This meta-analysis reveals that higher long chain n-3 PUFA intake is inversely associated with risk of stroke morbidity and mortality with BMI and sex as key factors influencing this risk. Individuals should be encouraged to manage their body weight while increasing their intake of long chain n-3 PUFAs.


Study characteristics. A summary of the characteristics of the included studies is shown in
.
Sensitivity analysis. Sensitivity analysis demonstrated that the relationship between higher long chain n-3 PUFA intake and reduction of stroke risk remained persistent after applying the leave-one-out method (Fig. 6). Publication bias. The funnel plot was symmetrical by visual inspection (Fig. 7), and no significant publication bias was statistically detected by Egger's test (p = 0.28).

Discussion
This meta-analysis consisting of 14 prospective studies of 514,483 individuals and 9,065 stroke events reveals that higher long chain n-3 PUFA intake is associated with a reduced overall stroke risk, which confirms the findings from a previous meta-analysis 8 . In contrast, four recent prospective studies found inconsistent findings regarding the relationship between long chain n-3 PUFA intake and stroke risk. For instance, one study 9 found no association between long chain n-3 PUFA intake and stroke risk in either men or women, and another study 10 found no association between long chain n-3 PUFA intake and stroke mortality. In contrast, another study 11 demonstrated an inverse association between long chain n-3 PUFA intake and ischemic (but not hemorrhagic) stroke risk, while the most recent prospective study 12 reported a significant reduction in stroke mortality for only the highest quartile of long chain n-3 PUFA intake. The mechanism(s) by which higher long chain n-3 PUFA consumption contributes to decreased stroke risk remain unknown. Some lines of evidence 4,5 show that dietary long chain n-3 PUFA enhances the stability of atherosclerotic plaques, lowers blood pressure, decreased blood triglyceride concentrations, decreases inflammation, and improves vascular function. There is evidence 24 demonstrating that consumption of processed meats is associated with higher incidence of coronary heart disease and diabetes mellitus, while a recent meta-analysis study 25 indicates that consumption of fresh or processed red meat as well as total red meat is positively associated with increased risk of total stroke and ischemic stroke.
Red meat is a source of heme iron. Higher iron mediates damage to tissues by catalyzing the production of reactive oxygen species (ROS), which leads to lipid peroxidation, protein modification, and DNA damage [26][27][28][29] . One meta-analysis 30 showed that higher intake of heme iron is associated with an increased risk of cardiovascular disease. Moreover, red meat is also a source of saturated fat and cholesterol; accordingly, a higher intake of saturated fat from meat or high absorption of cholesterol is associated with greater risk of cardiovascular disease 31,32 . Moreover, higher sodium intake from processed meat can also contribute to elevated blood pressure 33 , reduced arterial compliance, and augmented vascular stiffness 34 . Furthermore, nitrates and their products may facilitate vascular dysfunction and atherogenesis 35 . Thus, based upon the above findings and our results, the reduction of stroke risk may be an effect from higher long chain n-3 PUFA intake through greater ingestion of fish combined with a lower intake of red meat and processed meat intake.
Of these fourteen prospective studies, one study 13 reported that fish oil supplements were not included in the assessment of dietary intake long chain n-3 PUFAs, and the use of fish oil supplements had little effects on the results. Another study 14 conducted from 1980 to 1994 reported a fish soil supplement consumption rate of only 1.6% in 1990. Another study 16 that found no association between fish oil intake and stroke risk reported that approximately 2.7% of participants used fish oil supplements. Two studies 12,19 reported no baseline data on fish oil supplementation, but fish oil supplement use was not common among the participants. The remaining nine studies [9][10][11]15,17,18,[20][21][22] failed to report any information on fish oil supplement use among the participants. However, eleven studies 10-16,18-21 included in this meta-analysis indicated that fish and seafood were the main sources of long chain n-3 PUFAs, while three studies 9,17,22 did not specify the sources of long chain n-3 PUFAs. Thus, based upon the above findings, fish and seafood (as opposed to fish oil supplements) were the primary sources of long chain n-3 PUFAs for the participants in the included studies in this meta-analysis.
The stratification results demonstrated that higher long chain n-3 PUFA intake is inversely associated with fatal stroke risk but failed to demonstrate any relationship between ethnicity, stroke type, follow-up duration, or study quality and reduced stroke risk with higher dietary long chain n-3 PUFAs intake. Sensitivity analysis demonstrated a persistent relationship between higher long chain n-3 PUFA intake and reduced stroke risk. Moreover, there was no publication bias detected between the included studies.
Although there is some evidence 36 suggesting that BMI is a risk factor for stroke, BMI's influence on the relationship between long chain n-3 PUFA intake and stroke risk remains unknown. In the current stratification analysis, there was a lack of canonical standards for defining BMI subgroups, because this meta-analysis included both East Asian and non-East Asian studies. Specifically, the upper limit for normal BMI in East Asian populations should be 23 kg/m 2 , while the 1997 World Health Organization (WHO) guidelines specify an upper normal BMI limit of 25 kg/m 2 . 37 Thus, in the stratification analysis for BMI, we applied the midpoint of 24 kg/m 2 as the cut-off point for BMI (24). Our results showed that individuals with a low BMI (< 24 kg/m 2 ) demonstrated reduced stroke risk through a higher intake of long chain n-3 PUFAs.    Sex has previously been shown as a factor influencing stroke risk 38 ; based on our results, it seems that females benefit more from the increased intake of long chain n-3 PUFAs. As platelets play a pivotal role in development of cardiovascular disease 39 , and platelet aggregation is an early event in the induction of thrombosis and arteriosclerosis 40 , one possible explanation for this sex-based phenomenon may be the differential sex-based effects of anti-platelet aggregation produced by different categories of long chain n-3 PUFAs. There is evidence 41 showing that DHA, DPA, and EPA are all equally effective in platelet aggregation in females, while both DHA and DPA are significantly less effective in reducing platelet aggregation in males as compared with females (EPA is equally effective in reducing platelet aggregation in both sexes).
There are several notable limitations to this study. First, dietary long chain n-3 PUFA intake tends to be associated with other nutrients that may prevent stroke, such as potassium 42 , magnesium 43 , fiber 44 , and protein 45 . However, the association between long chain n-3 PUFA intake and stroke risk was persistent when we confined the analysis to studies that adjusted for these risk factors. Second, there existed heterogeneity between the included studies, although we are unable to determine the sources of heterogeneity. Third, a healthy diet for the primary prevention of cardiovascular and cerebrovascular disease should include adequate intake of vegetables and fruits [46][47][48][49] as well as whole grains 50,51 and olive oil 52 .
In conclusion, this meta-analysis reveals that higher long chain n-3 PUFA intake is inversely associated with risk of stroke morbidity and mortality with BMI and sex as key factors influencing this risk. Individuals should be encouraged to manage their body weight while increasing their intake of long chain n-3 PUFAs.  Methods Data sources and searches. PubMed, Embase, Web of Knowledge, and Google Scholar were searched without language restrictions as follows: ("fat" OR "fatty acids") AND ("stroke" OR "cerebrovascular disease" OR "cerebrovascular disorder" OR "cerebrovascular accident" OR "TIA" OR "transient ischemic attack"). Other potential studies were identified by consulting previous reviews and reference lists of retrieved records.
Inclusion and exclusion criteria. The inclusion criteria were as follows: (i) a prospective cohort design; (ii) reported RRs and their corresponding 95% CIs for long chain n-3 PUFA intake and stroke risk; (iii) multivariates (such as age, smoking, etc.) were controlled; and (iv) only the most recent publication, or the one with the longest follow-up period, was included when duplicate reports based on the same cohort were used.
The exclusion criteria were as follows: (i) case-control or non-prospective cohort study design; (ii) reviews; (iii) experimental studies; and (iv) conference abstracts.
Data extraction and quality assessment. Data were extracted independently by two investigators (P.F.C. and W.H.), and any differences were resolved by discussion with a third investigator (X.F.Z.) We retrieved the following parameters from each included study: first author's name, publication year, country of study population, age range or mean age, sex (%), number of participants, fat intake assessment, follow-up duration, number of stroke events, outcome assessment, RRs of stroke and corresponding 95% CIs for specific fat intake, and covariates adjusted in the statistical analysis. We used the Newcastle-Ottawa Scale (NOS) 23 to assess the study quality in this meta-analysis with a high-quality study defined as a study with > 8 awarded stars.
Statistical analysis. Log RRs of the highest versus the lowest for cohort studies were weighed by the inverse variance method to obtain pooled RRs. For calculating more robust RRs of stroke from long chain n-3 PUFAs intake, we retrieved all supplemental files of the included studies for RRs of specific types of stroke or specific sex of stroke patient if available. However, 7 of 14 of the included studies did not provide results for males and females separately. For these studies, we contacted the authors to ask them to provide these data. Authors of one study 10 provided us with separate results for females and males, authors of another study 11 claimed that they did not have these data, authors of three studies [17][18][19] did not respond us, authors of two studies 13,20 were failed to be contacted because of unsuccessful emails and missing contact information; thus, these six studies 11,13,[17][18][19][20] were not included in the sex subgroup analysis. Stratification analyses were based on BMI (< 24 versus ≥ 24), follow-up duration (< 14 versus ≥ 14 years), ethnicity (non-East Asians versus East Asians), sex (males versus females), stroke type (ischemic versus hemorrhagic), fatal stroke risk, maximum multivariates (pooling RRs of included studies with hypertension, diabetes, and smoking controlled simultaneously), study quality score (≤ 8 versus > 8). For the purpose of obtaining more conservative results, we used a random-effects model for pooling RRs. Smoking, hypertension, and diabetes could not be simultaneously adjusted in one 11 of the included studies; therefore, we did not include this study in the maximum multivariates adjusted analysis. A meta-regression model was used to detect potential heterogeneity between the included studies. A sensitivity analysis was conducted using the leave-one-out method. Furthermore, publication bias was assessed using Egger's test. Data obtained from the included studies were analyzed using Stata, version 12.0 (Stata Corp, College Station, Texas).