APOE and the Association of Fatty Acids with the Risk of Stroke, CHD and Mortality

Abstract

Background and Purpose:

The role of dietary fat on cardiovascular health and mortality remains under debate. Since the APOE is central to the transport and metabolism of lipids, we examined associations between plasma fatty acids and the risk of stroke, coronary heart disease, and mortality by APOE-ε4 genotype.

Methods:

We included 943 Framingham Heart Study, and 1,406 Three-City Bordeaux Study participants. Plasma docosahexaenoic, linoleic, arachidonic, and palmitic fatty acids were measured at baseline by gas chromatography. All-cause stroke, ischemic stroke, coronary heart disease, and all-cause mortality events were identified prospectively using standardized protocols. Each cohort used Cox models to separately relate fatty acid levels to the risk of developing each event during up to 10 years of follow-up adjusting for potential confounders and stratifying by APOE genotype (ε4 carriers vs. non-carriers). We then meta-analyzed summary statistics using random effects models.

Results:

On average, participants had a mean age of 74 years, 61% were women, and 21% (n=483) were APOE-ε4 carriers. Meta-analysis results showed that, only among APOE-ε4 carriers, every standard deviation unit increase in linoleic acid was associated with a reduced risk of all-cause stroke (HR=0.54 [95%CI, 0.38–0.78]), ischemic stroke (HR=0.48 [95%CI, 0.33–0.71]), and all-cause mortality (HR=0.70 [95%CI, 0.57–0.85]). In contrast, every standard deviation unit increase in palmitic acid was related to an increased risk of all-cause stroke (HR=1.58 [95%CI, 1.16–2.17]), ischemic stroke (HR=1.76 [95%CI, 1.26–2.45]), and coronary heart disease (HR=1.48 [95%CI, 1.09–2.01]), also in APOE-ε4 carriers only. Results for DHA and arachidonic acid were heterogeneous between cohorts.

Conclusions:

These exploratory results suggest that APOE-ε4 carriers may be more susceptible to the beneficial or adverse impact of fatty acids on cardiovascular disease and mortality. In this subgroup, higher linoleic acid was protective for stroke and mortality, whereas palmitic acid was a risk factor for stroke and coronary heart disease. The mechanisms underlying these novel findings warrant further investigation.

Diet is an important modifiable risk factor for cardiovascular disease and mortality, but the impact of dietary fat in these outcomes continues to be an active topic of debate. Although dietary guidelines recommend limiting the consumption of saturated fat,^{1, 2} pooled analyses from observational studies failed to show that lower intake of saturated fatty acids (SFA) translated into fewer cardiovascular events or mortality,³ and randomized clinical trials found no benefit when SFA were replaced with carbohydrates or mono-saturated fatty acids (MUFAs).⁴ Similarly, much debate remains on the recommendation to consume omega-6 polyunsaturated fatty acids (PUFAs) to decrease cardiovascular risk⁵ as some concern has been raised on whether diets rich in omega-6 PUFAs promote pro-inflammatory responses.⁶ Most of these studies have relied on food-frequency questionnaires subject to recall bias⁷ and assessment of clustered fatty acids (i.e. SFA, MUFA, PUFA), but individual fatty acids may have differential effects on health. Thus, measuring individual fatty acids in blood, a more sensitive marker of dietary fat intake and metabolism, could help disentangle the impact of different fatty acids on cardiovascular health.⁸

Another source of heterogeneity may arise from gene-environment interactions. For instance, some studies suggest that APOE genotype modifies the associations between omega-3 PUFAs and dementia risk,⁹ or between omega-3 PUFAs and lipoprotein profiles¹⁰ Because APOE is central to the transport and metabolism of lipids,¹¹ the question arises as to whether APOE genotype would also modulate the impact of fatty acids on cardiovascular disease and mortality.

A few population-based studies have related circulating levels of docosahexaenoic acid (DHA), linoleic acid, arachidonic acid and palmitic acid to the risk of cardiovascular disease and mortality, but results are mixed,^12–21 and thus far, studies assessing effect modification by APOE genotype on these associations are lacking. Therefore, we aimed to investigate the role of APOE genotype in the association between these four commonly investigated plasma fatty acids and the risk of incident stroke, coronary heart disease (CHD) and all-cause mortality in the Framingham Heart Study (FHS) and the Three-City (3C) Study.

Methods

The data, analytical methods, and study materials are not currently available to other researchers for purposes of reproducing the results or replicating the procedure.

We included 943 participants from the FHS Original Cohort and 1,406 3C participants from Bordeaux (3C-Bordeaux) with measures of fatty acids and APOE genotyping. Additional study sample details are provided in Supplemental Methods (please see http://stroke.ahajournals.org).

FHS and 3C participants provided written informed consent. FHS protocols and consent forms were approved by the Institutional Review Board of the Boston University Medical Center. The 3C Study protocol has been approved by the Consultative Committee for the Protection of Persons participating in Biomedical Research of the Kremlin-Bicêtre University Hospital.

Fatty acids and covariates

Non-fasting (FHS) and fasting (3C-Bordeaux) blood samples were collected at baseline (1985–1988 in FHS, 1999–2001 in 3C-Bordeaux) and measured using gas chromatography. Individual fatty acids were identified by comparison with reference fatty acid esters and are expressed as a percentage of total fatty acids (as membrane fatty acid composition has been typically reported). This report investigates the levels of docosahexaenoic acid (DHA), linoleic acid, arachidonic acid and palmitic acid as exposure measures.

We considered the components of the Framingham Stroke Risk Profile²² and other vascular risk factors at baseline as covariates in our analyses, which had similar definitions in both cohorts. Participants were classified as carriers of at least one APOE-ε4 allele or non-carriers. We did not consider APOE-ε4 homozygous as a separate subgroup given the low frequency (15 in FHS; 18 in 3C-Bordeaux).

Further information on the quantification of fatty acids and the definition of covariates is provided in the supplement (please see http://stroke.ahajournals.org).

Cardiovascular and mortality outcomes

Cardiovascular events and vital status is continuously monitored in FHS and tracked at each follow-up visit in 3C. Surveillance and validation of events details are provided in Supplemental Methods (please see http://stroke.ahajournals.org).

In both cohorts, clinical stroke was defined as a sudden onset of a focal neurological disturbance of presumed vascular etiology lasting more than 24 hours. This report considered all-cause stroke and the subtype of ischemic stroke. CHD events were defined as angina pectoris, recognized and unrecognized myocardial infarction, coronary insufficiency, or CHD death in FHS. In 3C-Bordeaux, CHD was defined as a diagnosis of hospitalized angina, hospitalized myocardial infarction, definite revascularization procedure, or definite CHD death.

Statistical analysis

FHS and 3C-Bordeaux followed the same analytical strategy. We used Cox models to estimate the association between plasma fatty acid levels and the risk of four outcomes: all-cause stroke, ischemic stroke, CHD, or all-cause mortality. Models were carried out separately among APOE-ε4 carriers (one or two copies of the ε4 allele) and non-carriers (no copies of the ε4 allele). Each analysis included participants who were free of the outcome of interest at baseline and observed events over up to 10 years of follow-up. We established this period of follow-up to balance two aspects: first, to minimize potential exposure misclassification, such that a longer follow-up may not reflect dietary and metabolic patterns at baseline; and second, to observe enough events. In persons with incident events, the follow-up time was measured in years from the baseline examination to development of the outcome. Follow-up time in persons without incident events was defined as the number of years from baseline to the last date when they were known not to have suffered the event, up to a maximum of 10 years. Fatty acids were modeled as continuous variables (per standard deviation unit, SDU) or in tertiles. DHA was log-transformed to normalize its distribution. To explore threshold effects, we created sex-stratified tertile categories for all fatty acids and compared the risk of events in the top two tertiles versus the lowest tertile (i.e. tertile 2–3 vs. tertile 1). Models were adjusted for age, sex, systolic blood pressure, anti-hypertensive medications, BMI, smoking, diabetes, and atrial fibrillation. The proportional hazards assumption was met for all models in both cohorts, except for two models in 3C-Bordeaux. In this cohort, we found deviations when modeling continuous arachidonic acid on ischemic stroke among APOE-ε4 carriers, and tertiles of palmitic acid on all-cause mortality among APOE-ε4 non-carriers. Those models were excluded from meta-analyses. Finally, we conducted random-effects meta-analyses of FHS and 3C-Bordeaux summary statistics. Associations were considered significant at P<0.05 and were not corrected for multiple testing given the exploratory nature of this investigation. Analyses were performed using SAS version 9.4 (SAS Institute Inc., Cary, NC) and the R package metafor.²³

Results

Population characteristics are presented in Table 1. On average, the mean age of participants was 74 years, 61% were women, and 21% were APOE ε4 carriers. Fatty acid levels were similar in both cohorts except for arachidonic acid, whereby levels in FHS almost doubled those of 3C-Bordeaux. Supplementary Table I (please see http://stroke.ahajournals.org) presents the range of fatty acid levels included in tertile categories, which may include different absolute values in each study. Fatty acid levels were comparable in APOE ε4 carriers and non-carriers in both cohorts (Supplementary Table II, see http://stroke.ahajournals.org).

Table 1.

Baseline characteristics per study sample

	FHS	3C-Bordeaux
	N=943	N=1,406
Clinical characteristics
Age, mean (SD), years	73.9 (5.1)	74.6 (5.0)
Women, No. (%)	589 (62.5)	852 (60.6)
APOE-ε4 carriership, No. (%)	201 (21.3)	282 (20.1)
Systolic blood pressure, mean (SD), mmHg	144 (20)	144 (21)
Diastolic blood pressure, mean (SD), mmHg	77 (10)	81 (11)
Anti-hypertensive medications, No. (%)	401 (42.5)	785 (55.8)
Body mass index, mean (SD), kg/m2	26.6 (4.5)	26.4 (4.2)
Smoking, No. (%)	105 (11.1)	70(5.0)
Diabetes, No. (%)	98 (10.4)	144 (10.2)
Atrial fibrillation, No. (%)	43 (4.6)	76 (5.4)
Plasma fatty acids*
DHA (22:6), median [interquartile range]	3.4 [2.8–4.2]	2.3 [1.8–2.9]
Linoleic acid (18:2), mean (SD)	24.8 (3.1)	24.9 (5.5)
Arachidonic acid (20:4), mean (SD)	11.0 (2.0)	6.7 (1.9)
Palmitic acid (16:0), mean (SD)	26.3 (1.9)	28.2 (5.7)

APOE=apolipoprotein; DHA=Docosahexaenoic acid

^*As % of total fatty acids

FHS experienced a higher burden of cardiovascular and mortality events than 3C-Bordeaux. We observed 81 (8.9%) strokes, of which 75 (8.3%) were ischemic, 130 (16.8%) cases of CHD, and 251 (26.6%) deaths in FHS during a mean follow-up of 9.1±2.0 years. In 3C-Bordeaux, we observed 51 (3.6%) strokes, of which 38 (2.7%) were ischemic, 53(3.8%) cases of CHD, and 251 (17.9%) deaths during a mean follow-up of 8.1±3 years.

Table 2 presents the number of outcome events and associations per APOE genotype and cohort. In APOE-ε4 carriers only, we observed consistent protective associations for linoleic acid with reduced stroke and mortality events. Every SDU increase in linoleic acid was related to a 43% and 48% decrease in the risk of all-cause stroke in FHS and 3C-Bordeaux, respectively. Similar results of slightly stronger magnitude were observed for ischemic stroke. Furthermore, every SDU increase in linoleic acid was associated with 33% and 27% reductions in all-cause mortality for FHS and 3C-Bordeaux, respectively. Greater mortality reductions (52% in FHS and 51% in 3C-Bordeaux) were observed for participants in the top two tertiles of linoleic acid levels, compared to those in the lowest tertile. In contrast, higher levels of palmitic acid were consistently associated with an increased risk of ischemic stroke in APOE-ε4 carriers. Every SDU increase in palmitic acid levels was related to a 70% and 80% increased risk of ischemic stroke events in FHS and 3C-Bordeaux, respectively.

Table 2.

Plasma fatty acids and the 10-year risk of cardiovascular outcomes and mortality stratified by APOE genotype in FHS and 3C-Bordeaux

	FHS				3C-Bordeaux
	APOE-ε4 non-carriers		APOE-ε4 carriers		APOE-ε4 non-carriers		APOE-ε4 carriers
	HR [95% CI]	P	HR [95% CI]	P	HR [95% CI]	P	HR [95% CI]	P
All-cause stroke (Events/N)	(65/712)		(16/194)		(38/1055)		(13/258)
DHA* (22:6)
SDU	0.85[0.66–1.08]	0.17	1.07[0.63–1.83]	0.81	1.21[0.85–1.72]	0.30	0.66[0.40–1.09]	0.10
T2–3 vs. T1	1.33[0.80–2.21]	0.27	1.69[0.59–4.82]	0.33	1.54[0.72–3.28]	0.26	0.40[0.13–1.22]	0.11
Linoleic acid (18:2 n-6)
SDU	1.04[0.80–1.34]	0.78	0.57[0.34–0.93]	0.025	1.22[0.86–1.73]	0.27	0.52[0.31–0.89]	0.016
T2–3 vs. T1	1.14[0.67–1.94]	0.62	0.32[0.11–0.92]	0.034	1.34[0.65–2.80]	0.43	0.39[0.13–1.21]	0.10
Arachidonic acid (20:4 n-6)
SDU	0.96[0.75–1.24]	0.77	1.81[1.02–3.22]	0.042	0.90[0.64–1.26]	0.54	1.19[0.68–2.07]	0.54
T2–3 vs. T1	0.80[0.48–1.33]	0.38	6.04[1.26–29.01]	0.025	0.79[0.41–1.54]	0.49	0.45[0.14–1.40]	0.17
Palmitic acid (16:0)
SDU	1.15[0.90–1.47]	0.27	1.58[0.94–2.63]	0.083	1.05[0.76–1.43]	0.78	1.59[1.07–2.35]	0.021
T2–3 vs. T1	1.85[1.01–3.38]	0.047	1.52[0.46–5.02]	0.49	0.85[0.43–1.70]	0.65	1.41[0.42–4.73]	0.58
Ischemic stroke (Events/N)	(60/712)		(15/194)		(27/1055)		(11/258)
DHA* (22:6)
SDU	0.87[0.68–1.12]	0.29	1.15[0.66–2.00]	0.63	1.12[0.74–1.69]	0.59	0.58[0.34–0.98]	0.041
T2–3 vs. T1	1.24[0.73–2.12]	0.43	1.42[0.47–4.30]	0.54	1.71[0.68–4.29]	0.25	0.24[0.07–0.85]	0.026
Linoleic acid (18:2 n-6)
SDU	1.14[0.87–1.49]	0.34	0.54[0.32–0.92]	0.022	1.23[0.81–1.86]	0.34	0.42[0.24–0.76]	0.004
T2–3 vs. T1	1.29[0.73–2.26]	0.38	0.29[0.10–0.85]	0.025	1.35[0.56–3.24]	0.50	0.39[0.11–1.33]	0.13
Arachidonic acid (20:4 n-6)
SDU	0.92[0.71–1.20]	0.56	2.09[1.13–3.87]	0.019	0.72[0.49–1.06]	0.10	NA†
T2–3 vs. T1	0.77[0.45–1.30]	0.32	14.12[1.70–117.35]	0.014	0.55[0.25–1.21]	0.14	0.41[0.12–1.41]	0.155
Palmitic acid (16:0)
SDU	1.16[0.90–1.50]	0.25	1.70[1.01–2.86]	0.047	1.18[0.86–1.61]	0.31	1.80[1.16–2.77]	0.008
T2–3 vs. T1	1.81[0.97–3.40]	0.064	2.02[0.54–7.57]	0.30	0.89[0.39–2.04]	0.79	1.58[0.40–6.23]	0.51
CHD (Events/N)	(109/606)		(21/166)		(37/937)		(16/237)
DHA* (22:6)
SDU	0.96[0.80–1.16]	0.70	1.07[0.65–1.77]	0.79	0.85[0.62–1.17]	0.32	1.16[0.65–2.08]	0.62
T2–3 vs. T1	1.03[0.69–1.53]	0.90	1.01[0.38–2.64]	0.99	1.13[0.57–2.27]	0.73	1.84[0.53–6.40]	0.34
Linoleic acid (18:2 n-6)
SDU	1.03[0.86–1.25]	0.73	0.95[0.63–1.44]	0.81	1.12[0.79–1.57]	0.53	0.58[0.33–1.02]	0.06
T2–3 vs. T1	1.16[0.77–1.74]	0.49	1.05[0.36–3.06]	0.93	0.97[0.49–1.93]	0.94	0.54[0.18–1.59]	0.27
Arachidonic acid (20:4 n-6)
SDU	0.99[0.81–1.19]	0.87	0.89[0.57–1.41]	0.63	1.18[0.84–1.66]	0.34	0.57[0.34–0.95]	0.032
T2–3 vs. T1	1.35[0.88–2.06]	0.17	0.61[0.24–1.55]	0.29	1.46[0.70–3.04]	0.32	0.38[0.12–1.19]	0.10
Palmitic acid (16:0)
SDU	1.15[0.95–1.39]	0.15	1.57[1.03–2.38]	0.035	0.86[0.56–1.32]	0.49	1.38[0.88–2.16]	0.16
T2–3 vs. T1	1.25[0.82–1.90]	0.30	1.58[0.58–4.34]	0.37	0.60[0.31–1.16]	0.13	2.08[0.54–8.08]	0.29
All-cause mortality (Events/N)	(185/742)		(66/201)		(192/1124)		(59/282)
DHA* (22:6)
SDU	0.89[0.77–1.03]	0.11	1.17[0.89–1.53]	0.26	0.93[0.80–1.07]	0.29	0.94[0.71–1.23]	0.64
T2–3 vs. T1	1.28[0.95–1.72]	0.11	0.87[0.50–1.50]	0.61	0.94[0.70–1.28]	0.71	1.37[0.77–2.44]	0.28
Linoleic acid (18:2 n-6)
SDU	0.95[0.81–1.11]	0.49	0.67[0.51–0.87]	0.003	0.98[0.85–1.13]	0.76	0.73[0.54–0.97]	0.028
T2–3 vs. T1	0.90[0.66–1.23]	0.52	0.48[0.28–0.84]	0.009	0.79[0.58–1.06]	0.11	0.49[0.28–0.83]	0.009
Arachidonic acid (20:4 n-6)
SDU	0.90[0.77–1.04]	0.16	1.55[1.19–2.01]	0.001	1.04[0.89–1.21]	0.65	0.78[0.61–0.99]	0.041
T2–3 vs. T1	0.82[0.61–1.11]	0.20	2.57[1.44–4.59]	0.002	1.15[0.85–1.56]	0.38	0.51[0.29–0.87]	0.014
Palmitic acid (16:0)
SDU	1.06[0.91–1.23]	0.45	0.96[0.73–1.26]	0.75	0.95[0.81–1.11]	0.50	1.25[0.99–1.58]	0.057
T2–3 vs. T1	1.21[0.87–1.68]	0.25	0.72[0.43–1.18]	0.19	NA†		2.24[1.19–4.24]	0.013

SDU=Standard deviation units;

DHA=docosahexaenoic acid;

CHD=coronary heart disease;

T=tertiles Tertiles are sex-specific Models are adjusted for age, sex, systolic blood pressure, antihypertensive medications, body mass index, smoking, diabetes mellitus, and atrial fibrillation^.

^*DHA levels are log-transformed

^†Proportionality of hazards not met

Significant cohort-specific results in APOE-ε4 carriers were confirmed in meta-analysis (Figure 1 and Figure 2). Additionally, meta-analysis showed that, compared to participants in the bottom tertile, participants with linoleic acid levels in the top two tertiles had a 65% reduced risk for all-cause stroke, a 66% reduced risk of ischemic stroke, and a 51% reduced risk of all-cause mortality. Finally, meta-analysis also revealed that every SDU increase in palmitic acid was associated with a 58%, 76% and 48% increased risk of all-cause stroke, ischemic stroke and CHD, respectively. No associations were found between linoleic acid and CHD, or between palmitic acid and all-cause mortality in either APOE-ε4 carriers or non-carriers.

Figure 1. Meta-analysis for associations between fatty acids (per SDU increase) and the 10-year risk of cardiovascular outcomes and mortality stratified by APOE genotype.

Models are adjusted for age, sex, systolic blood pressure, antihypertensive medications, body mass index, smoking, diabetes mellitus, and atrial fibrillation DHA=docosahexaenoic acid; E4+ve (blue)=APOE-ε4 carriers; E4-ve (black)=APOE-ε4 non-carriers; CHD=coronary heart disease, DHA levels are log-transformed, I²=Denotes heterogeneity as the estimated proportion total variance, P_het=P-value from test of heterogeneity, *Meta-analysis was not performed due to a deviation from the proportional hazards assumption when modeling arachidonic acid on ischemic stroke risk among APOE-ε4 carriers from 3C-Bordeaux, †Meta-analysis estimates are not presented due to significant heterogeneity (P_het<0.05)

Figure 2. Meta-analysis for associations between fatty acids (upper two vs. bottom tertile) and the 10-year risk of cardiovascular outcomes and mortality stratified by APOE genotype.

Models are adjusted for age, sex, systolic blood pressure, antihypertensive medications, body mass index, smoking, diabetes mellitus, and atrial fibrillation, DHA=docosahexaenoic acid; E4+ve (blue)=APOE-ε4 carriers; E4-ve (black)=APOE-ε4 non-carriers; CHD=coronary heart disease, Tertiles are sex-specific I²=Denotes heterogeneity as the estimated proportion total variance, P_het=P-value from test of heterogeneity, *Meta-analysis estimates are not presented due to significant heterogeneity (P_het<0.05), †Meta-analysis was not performed due to a deviation from the proportional hazards assumption when modeling palmitic acid on all-cause mortality risk among APOE-ε4 non-carriers from 3C-Bordeaux

Our findings for DHA and arachidonic acid were mixed. Whereas higher levels of DHA were associated with a reduced risk of ischemic stroke in 3C-Bordeaux, we found no significant associations in FHS. Furthermore, every SDU increase in arachidonic acid was related to a 55% higher mortality risk in FHS, whereas in 3C-Bordeaux it was related to a 22% decreased risk. We do not present pooled estimates if there was evidence of high heterogeneity (P_het<0.05).

Discussion

The findings of this exploratory study in two independent cohorts suggest that APOE-ε4 carriers with higher levels of linoleic acid are less likely to experience stroke and death, and that those with higher levels of palmitic acid are at increased risk of stroke and CHD. These results were independent of traditional vascular risk factors. Additional adjustment for lipid-lowering and platelet antiaggregant medication use did not materially change our findings (data not shown).

Higher linoleic acid levels have been related to a decreased risk of all-cause stroke in Swedish men¹² and ischemic stroke in women and men from the ARIC Study.¹⁴ Other studies in Asian populations suggest protection against all-cause and ischemic stroke (notably lacunar infarction).^{24, 25} Higher levels of linoleic acid have also been associated with reduced mortality risk.^{18, 19} However, other studies did not find associations with stroke risk,^{17, 18} or mortality.^{20, 21} Further, previous studies have related circulating palmitic acid to an increased risk of ischemic stroke in post-menopausal women from the WHI¹³ and to all-cause stroke in men.¹² In ARIC, however, although SFA levels were related to a 64% increased risk of stroke, results were not significant for palmitic acid alone;¹⁴ and in CHS there was no association between palmitic acid and CHD.¹⁶

Although previous studies have investigated associations between individual or clustered fatty acids in blood and the risk of cardiovascular disease and mortality, they have not explored the impact of APOE and the novelty of this investigation is the assessment of effect modification by APOE genotype on these associations. Interestingly, our findings were restricted to APOE-ε4 carriers, which could explain in part inconsistent results from previous studies. It is conceivable that the differences observed are partly due to differential metabolic patterns by APOE status. Studies in mice show that ApoE-ε4 carriers have an increased mobilization and utilization of fatty acids as compared to APOE-ε3 carriers.²⁶ A study in Alaskan Natives showed correlations for plasma levels of palmitic acid with higher cholesterol and ApoB concentrations only in APOE-ε4 carriers,²⁷ and another investigation in individuals with the metabolic syndrome from eight European countries found that APOE-ε4 carriers with higher plasma levels of palmitic acid had increased markers of insulin resistance compared to those with lower levels.²⁸ Therefore, APOE-ε4 carriers could be more vulnerable to the beneficial or adverse biological effects of fatty acids in response to dietary changes. However, information is limited in the published literature, and more research is needed to understand these associations. Alternatively, APOE may modify the impact of additional factors influencing the risk of stroke, CHD and mortality risk, such as Lipoprotein(a) or apolipoprotein B.²⁹

Finally, our results for DHA and arachidonic were heterogeneous in APOE-ε4 carriers. Notably, arachidonic acid appeared as a risk factor in FHS but protective in 3C-Bordeaux when assessing all-cause mortality. Although most fatty acids levels were comparable in both cohorts, the levels of arachidonic acid were almost two-fold in FHS than in 3C-Bordeaux. This could reflect differences in dietary patterns or endogenous metabolism between Americans and French.^{30, 31} Previous studies have related higher arachidonic acid levels to a decreased risk of ischemic stroke¹³ or CHD,¹⁵ but others find it associated with greater odds of overall²⁵ and cardio-embolic stroke.³² Therefore, there are potentially other factors in addition to those considered in this study influencing the associations of arachidonic acid with cardiovascular and mortality risk.

Strengths of our study include the prospective evaluation of two independent community-based samples, with rigorous prospective surveillance for cardiovascular events and mortality, consistent diagnostic criteria over time, objective measurement of individual fatty acids, and control for potential confounders. We acknowledge several limitations. First, our samples are composed of European descent, which limits the generalization of our results to other ethnic groups. Second, we were limited to study stroke subtypes such as intracerebral hemorrhage, sub-classifications of ischemic stroke etiology, or APOE-ε4 homozygotes separately, due to small numbers. Third, we used a single measure of fatty acids at baseline to represent long-term dietary and metabolic patterns. However, we tried to minimize exposure misclassification by establishing a follow-up of 10 years. Some studies found moderate correlations between fatty acids measured 3³³ or 15³⁴ years apart, and similar results were obtained using a single measure or the average of two measures of fatty acids to predict heart failure risk.³⁴ Thus, a single measurement of fatty acids seems reliable, and has been the practical approach by epidemiological studies. Forth, fatty acids were derived from non-fasting blood samples in FHS, and thus, their composition may not only reflect dietary intake over the past couple of weeks³⁵ but also some variation linked to individual postprandial responses to metabolize the last meal.³⁶ However, postprandial blood samples have proved useful for the prediction of cardiovascular outcomes,³⁷ and fasting status does not appear to modify the association between dietary and plasma fatty acid composition.³⁸ Any potential misclassification due to the use of a single measure, or variation resulting from the use of fasting and non-fasting blood samples, would likely bias our results towards the null and the true effect size could be stronger than the reported. Fifth, although the occurrence of vascular risk factors was comparable in both cohorts, that of events was higher in FHS. This could be due to different study designs (continuous monitoring in the FHS vs. monitoring at study visits in 3C-Bordeaux), differences in follow-up, or it may reflect differences in unknown factors affecting the US and France not considered in this study, potentially constituting an additional source of heterogeneity. Finally, we acknowledge the exploratory nature of these analyses and the need for replication in additional samples.

In conclusion, our exploratory analyses suggest that linoleic acid is a protective factor for all-cause stroke, ischemic stroke, and all-cause mortality, whereas palmitic acid is a risk factor for all-cause stroke, ischemic stroke and CHD. These associations were observed only in APOE-ε4 carriers. Our study provides novel findings opening new possibilities of research. The mechanisms underlying these findings warrant further investigation.