Association of circulating fatty acids with cardiovascular disease risk: analysis of individual-level data in three large prospective cohorts and updated meta-analysis

Abstract

Aims: Associations of saturated and unsaturated fatty acids (FAs) with cardiovascular disease (CVD) remain controversial. We therefore aimed to investigate the prospective associations of objectively measured FAs with CVD, including incident coronary heart disease (CHD) and stroke, as well as CVD mortality.

Methods and results: Circulating FA concentrations expressed as the percentage of total FAs were assayed in 172 891 participants without prior vascular disease at baseline from the European Prospective Investigation into Cancer and Nutrition-CVD (EPIC-CVD) (7343 CHD; 6499 stroke), UK Biobank (1825; 1474), and INTERVAL (285; 209) cohort studies. Hazard ratio (HR) per 1-standard deviation (SD) higher FA concentrations was estimated using Cox regression models and pooled by random-effects meta-analysis. Systematic reviews with meta-analysis published by 6 May 2023 on associations between FAs and CVDs were systematically searched and updated meta-analyses using random-effects model were conducted. Evidence from randomized controlled trials (RCTs) was also summarized. Higher concentrations of total saturated FAs (SFAs) were associated with higher cardiovascular risks in the combined analysis, with differential findings noted for SFA sub-types in further analysis restricted to EPIC-CVD: positive associations for even-chain SFA [HR for CHD 1.24 (95% CI: 1.18-1.32); stroke 1.23 (1.10-1.38)] and negative associations for odd-chain [0.82 (0.76-0.87); 0.73 (0.67-0.78)] and longer-chain [0.95 (0.80-1.12); 0.84 (0.72-0.99)] SFA. In the combined analysis, total n-3 polyunsaturated FA (PUFA) [0.91 (0.85-0.97)], including docosahexaenoic acid (DHA) [0.91 (0.84-0.98)], was negatively associated with incident CHD risk. Similarly, total n-6 PUFA [0.94 (0.91-0.98)], including linoleic acid (LA) [0.89 (0.83-0.95)], was negatively associated with incident stroke risk. In contrast, more detailed analyses in EPIC-CVD revealed that several downstream n-6 PUFAs of LA were positively associated with CHD risk. Updated meta-analyses of 37 FAs including 49 non-overlapping studies, involving between 7787 and 22 802 CHD cases and between 6499 and 14 221 stroke cases, showed broadly similar results as our combined empirical analysis and further suggested significant inverse associations of individual long-chain n-3 PUFAs and LA on both CHD and stroke. The findings of long-chain n-3 PUFAs were consistent with those from published RCTs on CHD despite insufficient evidence in monotherapy, while RCT evidence remained unclear for the rest of the explored FAs.

Conclusion: Our study provides an overview of the most recent evidence on the associations between objectively measured FAs and CVD outcomes. Collectively, the data reveal notable differences in associations by SFA sub-types and call for further studies, especially RCTs, to explore these links.

Keywords: Cardiovascular disease; Cohort study; Coronary heart disease; Fatty acids; Meta-analysis; Stroke.

Graphical Abstract

Figure 1

Circos plot of correlations between major fatty acid sub-types (%) and self-reported food intake (g/day) in EPIC-CVD study, separated into positive (left side) and negative (right side) associations. MUFA, monounsaturated fatty acid; PUFA, polyunsaturated fatty acid; SFA, saturated fatty acid. Analyses were conducted restricted to the sub-cohort of EPIC-CVD study (n = 15 838). Width of curves indicates the strength of statistically significant correlations with P-value <0.05 (i.e. semipartial correlation coefficients adjusted for batch, sex, age, and total energy intake), and correlations with the absolute value of coefficient <0.05 were coloured in grey. For example, correlations of milk intake with even-chain, odd-chain, and longer-chain SFAs were −0.024, 0.118, and 0.066, respectively (highlighted by black borders in the figure). Further detailed correlations between individual plasma phospholipid fatty acids and food intake were illustrated in Supplementary material online, Supplementary FigureS3.

Figure 2

Dose-response curve and hazard ratios for associations of fatty acids with coronary heart disease and stroke, estimated in 172 891 participants from EPIC-CVD, UKB, and INTERVAL studies. CI, confidence interval; DHA, docosahexaenoic acid; HR, hazard ratio; LA, linoleic acid; MUFA, monounsaturated fatty acid; PUFA, polyunsaturated fatty acid; SD, standard deviation; SFA, saturated fatty acid. Results are expressed as HRs according to fatty acid concentrations in SD units using random effects meta-analysis, adjusted for batch (EPIC-CVD only), age, smoking status, history of diabetes, history of hypertension, and physical activity, and stratified by centre (EPIC-CVD only) and sex. The shaded area represents the 95% CI for the dose-response curve. The final estimates are pooled HRs (95% CI) per 1-SD higher fatty acid concentrations with the same adjustment using random-effects meta-analysis. *Restricted to EPIC-CVD (7343 CHD cases; 6499 stroke cases) as concentrations of SFA sub-types were only available from EPIC-CVD (as shown in red curves). Detailed information of estimates for different cardiovascular outcomes in Supplementary material online, Tables S5 and S6.

Figure 3

Updated meta-analysis combining results from EPIC-CVD, UKB, and INTERVAL studies with published evidence for associations of fatty acid biomarkers with coronary heart disease and stroke, separately. RR, relative risk; SD, standard deviation. For the abbreviations of fatty acids, please refer to the Supplementary material online, Table S4. The results of updated meta-analyses using random-effects method were summarized for the fatty acids, of which associations with coronary heart disease or stroke were identified from existing published evidence. Results for ischaemic stroke in Supplementary material online, Figure S7.