Polyunsaturated Fatty Acids and Risk of Ischemic Stroke

Ischemic stroke is a major cause of death and morbidity worldwide. It has been suggested that polyunsaturated fatty acids (PUFAs) may be associated with a lower risk ischemic stroke, but this has been far less studied than their role for coronary heart disease. In this paper, we summarize the main findings from previous follow-up studies investigating associations between intake or biomarkers of the major PUFAs including alpha-linolenic acid (ALA), marine n-3 PUFAs and linoleic acid (LA) and the development of ischemic stroke. Several follow-up studies have suggested that marine n-3 PUFAs may be associated with a lower risk of ischemic stroke although results have not been consistent and limited knowledge exist on the individual marine n-3 PUFAs and ischemic stroke and its subtypes. The role of ALA is less clear, but most studies have not supported that ALA is appreciably associated with ischemic stroke risk. Some studies have supported that LA might be associated with a lower risk of total ischemic stroke, while limited evidence exist on PUFAs and ischemic stroke subtypes. The associations may depend on the macronutrients that PUFAs replace and this substitution aspect together with focus on dietary patterns represent interesting areas for future research.


Introduction
Ischemic stroke is characterized by an episode of neurological dysfunction attributed to cerebral infarction in the perfusion territory of an stenosed or occluded artery [1]. Ischemic stroke is a major cause of death and disability worldwide, but some studies have indicated that adherence to a healthy diet and lifestyle may be of importance for reduction of ischemic stroke risk [2,3].
An important part of a healthy diet is believed to be an increased intake of polyunsaturated fatty acids (PUFAs). PUFAs are organic acids that naturally contains two or more double bonds in their carbon chain and are named according to the number, configuration and position of these double bonds [4]. PUFAs are upon ingestion incorporated into cell membranes and storage pools, utilised in energy production or converted into longer and more unsaturated fatty acids, which may give rise to biological compounds that may affect a variety of biological processes [5]. PUFAs have traditionally been divided into n-3 PUFAs and n-6 PUFAs and these two families each consist of fatty acids with varying carbon chain length and degree of unsaturation, which are important for their physiological and biological properties [4]. The major n-3 PUFAs include the short-chain plant-derived n-3 PUFA, alpha-linolenic acid (ALA, 18:3n-3), and the longer-chain marine n-3 PUFAs eicosapentaenoic acid (EPA, 20:5n-3), docosapentaenoic acid (DPA, 22:5n-3) and docosahexaenoic acid (DHA, 22:6n-3).
The major dietary n-6 PUFA is linoleic acid (LA, 18:2n-6) and together with ALA are the PUFAs consumed in largest amounts. The typical intake of LA in Western populations range between 10 to

ALA and Ischemic Stroke
Few studies have investigated the association between ALA intake and ischemic stroke [25][26][27][28][29] ( Table 1). In the Monitoring Project on Risk Factors for Chronic Diseases Study (MORGEN Study), dietary intake of ALA was inversely associated with the rate of total ischemic stroke among Dutch men and women [25]. In the Cardiovascular Health Study (CHS), indications of a lower rate of total ischemic stroke were observed when comparing the highest quintile of ALA intake with the lowest, but no consistent pattern of association was observed across quintiles of ALA in this cohort of US men and women [26]. Further, no clear association was observed between ALA intake and the rate of total ischemic stroke among women enrolled into the Women's Health Study [27] or the Swedish Mammography Cohort [28] and among men and women enrolled into the Diet, Cancer and Health (DCH) cohort [29]. Limited knowledge is available regarding the associations between ALA intake and the risk of ischemic stroke subtypes but no appreciable pattern of associations were observed between intake of ALA and the rate of ischemic stroke subtypes caused by large artery atherosclerosis, small vessel-occlusion or cardioembolism in the DCH cohort [29] (Table 2).  Bork et al. 2018 [29] No. of cases 316 835 102 Abbreviations: NS, Not statistically significant; AT, Adipose tissue. 1 Univariate analysis. 2 99% Confidence interval. The presented measures of association was selected for models that included adjustment for lifestyle factors without including dietary covariates whenever possible. See Table 1 and Tables 3-7 for study characteristics.
The findings from previous follow-up and nested-case control studies using biomarkers of ALA to investigate associations with ischemic stroke have not been conclusive [26,[31][32][33][38][39][40] (Table 3). In a large case-cohort study based on data from the DCH cohort we reported indications of a U-shaped pattern of association between adipose tissue content of ALA and the rate of total ischemic stroke although not statistically significant [33]. A follow-up study based on the Finish Kuopio Ischaemic Heart Disease Risk Factors (KIHD) study also reported indications of a U-shaped pattern of association between quartiles of ALA content in serum and total ischemic stroke, while a nested case-control study reported a modest non-significant inverse association between ALA in serum and total ischemic stroke [32,40]. In contrast, several other biomarker studies did not find ALA content in serum, plasma phospholipids or cholesterol esters associated with the risk of total ischemic stroke [26,31,38,39]. Limited data are available on biomarkers of ALA and ischemic stroke subtypes (Table 2). However, data from the DCH cohort suggested a statistically significant U-shaped association between ALA content in adipose tissue and the rate of ischemic stroke due to large artery atherosclerosis with the lowest rate observed around the median content of ALA in adipose tissue [33]. In contrast, a positive statistically non-significant association was observed between ALA content in adipose tissue and the rate of ischemic stroke due to cardioembolism, whereas no appreciable association was found between ALA in adipose tissue and ischemic stroke due to small-vessel occlusion [33].
In summary, previous studies evaluating either intake of ALA or content of ALA in blood components or adipose tissue have shown conflicting results, but most studies does not support that ALA exposure is inversely associated with ischemic stroke risk.

Marine LC n-3 PUFAs and Ischemic Stroke
Several studies have investigated the association between intake of LC n-3 PUFAs and total ischemic stroke [27,28,30,35,[41][42][43][44][45][46] (Table 4) and has recently been reviewed [47]. In brief, findings from the Nurses' Health Study (NHS) indicated a statistically non-significant inverse pattern of associations between EPA+DHA intake and the rate of total ischemic stroke [30]. Analyses of ischemic stroke subtypes showed an inverse pattern of associations between EPA+DHA intake and lacunar infarctions, while no clear associations were observed for large artery occlusive infarctions [30] (Table 2). Lower rates of total ischemic stroke across quintiles of EPA+DHA intake were observed in the Swedish Mammography cohort [28] and in the Health Professionals Follow Study (HPFS) although results were not statistically significant in the latter when comparing the highest quintile of intake with the lowest [41]. In the MORGEN study, gender-specific analyses indicated lower rates of total ischemic stroke in subjects with a high EPA+DHA intake, but the confidence intervals were wide and the point estimates were not statistically significant [42]. A recent study from our groups by Venø et al. [35] investigated associations between dietary intake of total marine LC n-3 PUFAs and of individual marine n-3 PUFAs and the rate of total ischemic stroke and ischemic stroke subtypes using data from the DCH cohort. In dietary analyses, no appreciable associations were observed between intake of total and individual n-3 PUFAs, EPA or DHA and the rate of total ischemic stroke [35]. Also, in five other follow-up studies, no consistent pattern of associations between EPA+DHA intake and rate of total ischemic strokes were reported [27,[43][44][45][46]. However, in the study by Venø et al. a high intake of total n-3 PUFAs, EPA and DHA was statistically significantly associated with lower rates of ischemic stroke due to large artery atherosclerosis when comparing the highest quartile of intake with the lowest [35] (Table 2). In contrast, indications of higher rates of ischemic stroke due to cardioembolism were observed with intake of LC n-3 PUFAs, whereas no consistent pattern of association was observed between LC n-3 PUFAs and the rate of ischemic stroke due small-vessel occlusions [35] (Table 2).  Some biomarker studies have investigated the associations between levels of LC n-3 PUFAs in blood compartments or adipose tissue and the risk of ischemic stroke [31,32,34,35,[38][39][40]48] (Table 5). Thus, a the content of EPA, DPA and DHA in plasma phospholipids was analysed in relation to ischemic stroke and its subtypes using pooled data from the CHS, NHS and HPFS [34]. Lower rates of total ischemic stroke was observed when comparing the highest quintile of content of DPA and DHA with the lowest, respectively, while no appreciable association was found for EPA [34]. However, the associations seemed to depend on the ischemic stroke subtype in question ( Table 2). In analyses of ischemic stroke subtypes, DHA was associated with the rate of atherothrombotic strokes (large artery and small-vessel infarctions), whereas no association was found for EPA and DPA. However, DPA was associated with lower rates of cardioembolic strokes, whereas no associations were observed between EPA or DHA and the rate of cardioembolic stroke [34]. Few other studies have supported that circulating biomarkers of LC n-3 PUFAs might be associated with a lower risk of ischemic stroke although results have not been consistent [31,32,38,40,48] (Table 5).  The associations between adipose tissue content of LC n-3 PUFAs and the risk of ischemic stroke and ischemic stroke subtypes were investigated based on data from the DCH cohort and no appreciable association was observed between total adipose tissue of LC n-3 PUFAs and the rate of total ischemic stroke [35]. However, the EPA content in adipose tissue was inversely associated with the rate of total ischemic stroke when comparing the highest with the lowest quintile [35]. In contrast, adipose tissue content of DPA seemed to be associated with a higher rate of total ischemic stroke, whereas no association was found for DHA [35]. In analyses of ischemic stroke subtypes, the content of EPA in adipose tissue was inversely associated with the rate of ischemic stroke due to large artery atherosclerosis and small-vessel occlusions, while a higher rate of ischemic stroke due to cardioembolism was noted when comparing the highest quartile with the lowest although not statistically significant [35] (Table 2). Adipose tissue content of DPA and DHA was associated with higher rates of ischemic stroke caused by cardioembolism whereas no consistent pattern of association was found for ischemic stroke caused by large artery atherosclerosis or small-vessel occlusions [35] ( Table 2).
In summary, several large prospective studies evaluating EPA and/or DHA intake or their content in blood components or adipose tissue have suggested an inverse association although results have not been consistent. The associations between LC n-3 PUFAs and ischemic stroke subtypes may differ with the most beneficial results observed in subjects with strokes of atherosclerotic etiology.

LA and Ischemic Stroke
Studies of the association between dietary intake of LA and the rate of ischemic stroke are sparse ( Table 6). In analyses based on data from the Swedish Mammography Cohort, no association was observed between intake of LA+AA and the rate of total ischemic stroke [28]. Also, no clear association was observed between LA+AA intake and the rate of total ischemic stroke in another Swedish cohort [45]. However, the intake of LA and AA was not analysed separately and the macronutrient(s) to be replaced for n-6 PUFAs was not specified in these studies. In contrast to n-3 PUFAs, LA constitute a significant proportion of total energy intake and a higher intake of LA must necessarily be accompanied by a lower intake of other macronutrients in an isocaloric setting. In a follow-up study using statistical substitution models we therefore investigated the risk of total ischemic stroke and ischemic stroke subtypes with a 5% higher intake of LA and a concomitant lower intake from saturated fatty acids (SFAs), monounsaturated fatty acids (MUFAs) or glycemic carbohydrates using data from the DCH cohort [36]. This study suggested that replacing MUFAs or glycemic carbohydrates with LA might be associated with a lower risk of total ischemic stroke although not statistically significantly [36]. Further, in analyses of ischemic stroke subtypes, replacement of SFA with LA indicated a lower rate of ischemic stroke due to large artery atherosclerosis and replacement of glycemic carbohydrates with LA indicated a lower rate of ischemic stroke due to small-vessel occlusions, although the observed associations were not statistically significant [36] (Table 2). A statistically significant lower rate of ischemic stroke due to small-vessel occlusions was observed when LA replaced MUFAs [36] (Table 2).  Several prospective biomarker studies have supported that circulating or adipose tissue content of LA may be associated with a lower risk of ischemic stroke although results have not been consistent [23,31,32,[37][38][39][40]49] (Table 7). A previous study by Iso et al. suggested a lower odds of total ischemic stroke and lacunar infarctions per 1 SD increase in LA content in serum [31]. Also, in analyses based on data from the ARIC Study, a statistically significant inverse association between LA content in plasma cholesterol esters and the rate of total ischemic stroke and indications of an inverse association was observed between LA content in plasma phospholipids and the rate of total ischemic stroke [39]. Also, several other studies have indicated a lower risk of total ischemic stroke with the content of LA in serum, plasma phospholipids and cholesterol esters, but these findings were not statistically significant and not entirely consistent [32,38,40,49]. However, a large case-cohort study based on data from the DCH study found a dose-dependent inverse association between adipose tissue content of LA and the rate of total ischemic stroke [37]. In analyses of ischemic stroke subtypes, adipose tissue content of LA was inversely associated with the rate of ischemic stroke due to large artery atherosclerosis and indications of lower rates of ischemic stroke due to small-vessel occlusions was observed, whereas no clear association was found between LA in adipose and the rate of ischemic stroke due to cardioembolism [37]. Recently, a harmonized individual-level analysis based on prospective data from the Fatty Acid and Outcome Research Consortium (FORCE) found an overall lower rate of total ischemic stroke with higher levels of LA in blood compartments and adipose tissue in analyses including 3705 incident ischemic stroke cases [23]. In summary, limited evidence regarding LA intake and ischemic stroke exist, but LA intake might be associated with a lower risk of total ischemic stroke although the association may depend on macronutrient used to replace LA and the ischemic stroke subtype in question. Several biomarker studies have supported that a high content of LA in blood components or adipose tissue may be associated with a lower risk of total ischemic stroke.

Discussion
Several prospective follow-up studies have suggested that LC n-3 PUFAs perhaps in particular EPA and DHA may be associated with a lower risk of total ischemic stroke although results have not been consistent. A few studies have suggested that the associations between individual LC n-3 PUFAs may differ amongst subtypes of ischemic stroke with the most beneficial findings observed in ischemic strokes of presumed atherosclerotic origin. Regarding the role of ALA for development of ischemic strokes, most prospective studies have not supported that ALA may be appreciably associated with a lower risk of ischemic stroke. The associations between LA and ischemic stroke have been less studied than the role of n-3 PUFAs, but some biomarker studies have supported that LA might be associated with a lower risk of ischemic stroke. Limited evidence exist regarding LA intake and ischemic stroke, but the association may depend on the macronutrient used to replace LA.
Biomarkers of PUFA exposure may represent useful complementary measures of exposure to investigate the associations with chronic diseases such as ischemic stroke [50]. Dietary studies using estimated PUFA intakes as exposure are limited by measurement error, which may result in loss of statistical power and attenuation of associations toward null. In contrast, the content of PUFAs in blood components or adipose tissue can be assessed more precise and are considered objective biomarkers of exposure [50,51]. Long-term measures of exposure is likely to be of greater etiological relevance in the development of ischemic stroke compared to short-term measures of exposure. Adipose tissue is considered the gold standard due to a slow turnover time possibly reflecting intake of PUFAs during the previous 1-2 years, whereas shorter-term biomarkers such as PUFA content in blood components may represent the exposure up to few months. However, the content of PUFAs in blood components or adipose tissue reflects both intake and metabolism [50]. Also, direct comparison of measures of associations obtained in previous biomarker of PUFAs from different population should be done with caution because differences in PUFA content in human tissue may be largely influenced by differences in the underlying dietary pattern [50,51].
Finally, ischemic stroke is a heterogeneous condition and it is likely that the biological effects of individual LC n-3 PUFAs and LA may differ across subtypes of ischemic stroke of different etiology. Ischemic stroke due to large artery atherosclerosis is considered to be of atherosclerotic origin, whereas ischemic stroke due to small-vessel occlusion may develop as a result of atherosclerosis or lipohyalinosis affecting the smaller penetrating arteries in the brain [52,53]. In contrast, ischemic cardio-embolic strokes are mainly caused by emboli arising from the heart due to arrhythmias, particularly atrial fibrillation or flutter. Furthermore, ischemic stroke may in rare cases develop as a result of nonatherosclerotic vasculopathies or prothrombotic disorders. Thus, these rather distinct potential causes of ischemic stroke underline that separate analyses of ischemic stroke subtypes may contribute with a better understanding of the underlying biology of exposure to PUFAs. Recent results based on data from the DCH cohort [35] and the pooled analyses by Saber et al. [34] indicated that the associations between individual LC n-3 PUFAs and ischemic stroke subtypes may differ. Interestingly, results from the DCH cohort indicated stronger associations for EPA and LA on ischemic stroke due to large artery atherosclerosis than for ischemic stroke due to small-vessel occlusion. This might indicate that potential anti-atherosclerotic properties of marine n-3 PUFAs may be of importance. Anti-atherosclerotic properties of LC n-3 PUFAs may include lowering of triglycerides and perhaps also of atherogenic small-dense LDL particles, a reduction in inflammation and blood pressure, while LA might inhibit the atherosclerotic process by lowering of LDL-cholesterol and blood pressure. EPA seemed to be stronger associated ischemic stroke due to large artery atherosclerosis than DHA in the DCH cohort, but whether these differences can be attributed to different biological effects remain unclear, but is not unlikely as many data support that a major effect of DHA on heart disease may be an antiarrhythmic effect.
In addition, the associations may depend on the macronutrients used to replace PUFAs and such substitution aspects together-with focus on dietary patterns-represent interesting areas for future research.
Author Contributions: All authors contributed to the conceptualizations of this manuscript. C.S.B. wrote the first draft of the manuscript and S.K.V. and E.B.S. critically revised the manuscript and contributed with intellectual content. All authors approved the final manuscript.

Conflicts of Interest:
The authors declare no conflicts of interest.