Introduction
Type 2 diabetes is associated with a substantial risk of macrovascular disease, including coronary and cerebrovascular diseases; microvascular disease, including kidney disease and retinopathy; and premature death [
1]. Early recognition of diabetes in its progression and initiation of an intervention are therefore needed for preventing such adverse long-term outcomes.
Fatty acids are vital nutrients which play regulatory roles in energy metabolism. The composition of fatty acids in blood can be affected by dietary intake [
2] and has been reported to be involved in pathological mechanisms of various diseases such as insulin resistance, obesity, diabetes and atherosclerosis [
3]. For several decades, clinical trials and population-based studies have attempted to determine the effects of dietary intake of fatty acids on vascular outcomes and mortality; however, the results remain controversial and inconsistent. Several meta-analyses of clinical trials have suggested that dietary intake of
n-3 fatty acids such as docosahexaenoic acid (DHA) and eicosapentaenoic acid (EPA), which are generally known as seafood-derived fatty acids, had no, or at most a weak, protective effect on cardiovascular disease (CVD) and death in people with and without diabetes [
4,
5]. In contrast, a recent meta-analysis reported that
n-3 supplementation was associated with a lower risk of CVD [
6]. In particular, the recent Reduction of Cardiovascular Events with Icosapent Ethyl–Intervention Trial (REDUCE-IT) observed pronounced protective effects of high-dose (4 g/day) supplementation with
n-3 fatty acids against cardiovascular outcomes among individuals with established CVD or with diabetes and other risk factors [
7]. Based upon these results, the recent scientific statement from the American Heart Association (AHA) recommends the prescription of
n-3 fatty acids, whether EPA+DHA or EPA only, at a dose of 4 g/day as an effective and safe treatment for reducing triacylglycerols among individuals with hypertriglyceridaemia [
8].
Blood or tissue
n-3 fatty acids have been reported to be associated with a lower risk of CHD [
9,
10]. These results support the belief that measurements of circulating
n-3 fatty acids may be useful for the prediction and management of cardiovascular risk. However, the benefits of measuring circulating fatty acids in people with type 2 diabetes in predicting the risks of cardiovascular and other vascular diseases are unclear. In addition, most studies report the results for individual fatty acids, and there are limited studies investigating multiple circulating fatty acids together.
To address these questions, we assessed the association of baseline plasma fatty acids with the risk of macrovascular and microvascular disease and death in participants with type 2 diabetes included in the Action in Diabetes and Vascular Disease: Preterax and Diamicron Modified-Release Controlled Evaluation (ADVANCE) study.
Methods
Statistical analyses
Baseline participant characteristics were summarised according to the study outcomes (major macrovascular and microvascular events and death). Categorical data were presented as number (percentage), and continuous data according to the data distribution, mean (SD) for approximately symmetrically distributed data and median (interquartile interval [IQI]) for skewed distributions. The percentage contribution of the fatty acid biomarkers of the total fatty acid and the absolute fatty acid values were summarised as mean (SD) values for each study outcome.
Cox proportional hazards models for case-cohort data were used to model the associations between fatty acids and the study outcomes, adjusting covariates considered as traditional cardiovascular risk factors. Models estimated HR (per SD percentage higher) of the percentage contribution of total fatty acids. Two sets of models were fit for each fatty acid-outcome combination: model 1, adjusted for age, sex, region and the treatments randomly allocated in the RCT; and multiple-adjusted model 2, additionally adjusted for history of macrovascular disease, duration of diabetes, current smoking status, systolic BP, BMI, urinary albumin/creatinine ratio, eGFR (calculated using the Chronic Kidney Disease Epidemiology Collaboration [CKD-EPI] creatinine equation), HbA1c, HDL-cholesterol, triacylglycerols, and use of aspirin or other antiplatelet agents, statins or other lipid-lowering agents, β-blockers and ACE inhibitors or angiotensin receptor blockers. All p values reported are two-sided, with the 5% threshold used to determine statistical significance. Since several statistical tests are included in this article, the reader is recommended to treat marginal levels of significance with caution.
Sensitivity analyses included models estimating HRs for 1 SD higher of the absolute level of each fatty acid. For the fatty acid-outcome combinations that yielded statistically significant results in multiple-adjusted models in the main analysis, models were fitted estimating HRs in quarters of the range of values of the percentage of fatty acids.
Subgroup analyses were performed according to baseline covariates, including age (<65 years or ≥65 years), sex, region of residence (Australia, New Zealand and South East Asia, Canada, Continental Europe and Northern Europe), history of macrovascular disease, history of microvascular disease, eGFR (<60 ml min−1 [1.73 m]−2 or ≥60 ml min−1 [1.73 m]−2), triacylglycerols (<1.7 mmol/l or ≥1.7 mmol/l), and randomised treatments (BP- and glucose-lowering treatments).
The ability of fatty acids to discriminate between those who will and those who will not go on to suffer major macrovascular and microvascular events and death was estimated using C statistics accounting for censoring [
17,
18]. Further, the ability of fatty acids to reclassify participants was estimated using the continuous net reclassification improvement (NRI) [
19,
20]. These statistics were computed, for 5 year risk, for individuals in the random subcohort only for those fatty acids whose percentage contribution demonstrated a statistically significant association with outcomes. 95% CIs for the C statistic (and increments in it when adding fatty acids) and NRI were calculated using bootstrap methods with a normal approximation and 500 bootstrap iterations. All analyses in this study were performed using Stata/MP, version 15 (Stata Corporation, College Station, TX, USA), and R, version 3.5.3 (R Foundation for Statistical Computing, Vienna, Austria); the R code is provided in the ESM Methods.
Discussion
This biomarker study showed inverse associations of baseline plasma n-3 fatty acids and DHA with the risk of macrovascular events and for n-3 fatty acids with the risk of death among individuals with type 2 diabetes. These inverse associations appeared approximately linear, and among the macrovascular events, n-3 fatty acids and DHA demonstrated stronger associations with cardiovascular death and non-fatal stroke. In contrast, no significant associations were observed for the predominant fatty acids such as n-6 fatty acids, LA, MUFA and SFA with the risk of macrovascular events and death, after adjustment for multiple traditional risk factors.
The prospective associations of circulating
n-3 fatty acids and DHA with the risk of CVD are consistent with the prior studies. According to previous pooled analyses, DHA in whole plasma was associated with a lower risk of fatal and non-fatal CHD (RR [95% CI]: 0.78 [0.69, 0.90] and 0.91 [0.84, 0.98], respectively) in people without a history of CVD [
9]. In addition, another study reported that plasma
n-3 fatty acids were associated with a lower risk of non-fatal myocardial infarction [
21]. Similar inverse associations with CHD and stroke were observed in other studies which measured
n-3 fatty acids in plasma phospholipid [
10,
22‐
24], whole blood [
25] and serum [
26], among people without prior CVD. These studies did not assess the C statistic and NRI; however, the present study did not detect significant improvement in predicting macrovascular events by adding
n-3 fatty acids or DHA into the model including traditional cardiovascular risk factors. This indicates that the predictive power of plasma
n-3 fatty acids and DHA may be limited in the presence of a complement of traditional cardiovascular risk factors. On the other hand, cardiovascular benefits of high-dose supplementation with
n-3 fatty acids were recently observed in REDUCE-IT, where there were potentially greater benefits of
n-3 fatty acid supplementation in those with lower plasma levels of
n-3 fatty acids.
In contrast to generally consistent results from observational studies of circulating
n-3 fatty acids and DHA, the effects of
n-3 fatty acid supplementation on cardiovascular outcomes in RCTs have been mixed [
4,
6,
27‐
30]. The recent study, A Study of Cardiovascular Events in Diabetes (ASCEND) trial, of 15,480 individuals with diabetes free of prior CVD, which tested
n-3 fatty acid supplementation (1 g/day) for 7.4 years, did not lower the risk of composite major vascular outcomes, while only vascular deaths were less frequent in the supplementation group than in the placebo group (RR 0.82 [95% CI 0.68, 0.98]) [
31]. On the other hand, the recent AHA science advisory has suggested that the use of
n-3 fatty acid supplementation was probably justified in individuals at high cardiovascular risk [
8,
28]. In REDUCE-IT, which used high-dose (4 g/day)
n-3 fatty acid supplementation in 8179 individuals with established CVD or with diabetes and other risk factors, the risk of composite cardiovascular outcomes was substantially reduced (HR 0.75 [95% CI 0.68, 0.83]) [
7]. In addition, some secondary analyses from large trials have reported the benefit of
n-3 fatty acid supplementation on CVD in diabetic populations with hypercholesterolaemia [
32], chronic heart failure [
33] and history of myocardial infarction [
34]. Further investigation of the benefits of
n-3 fatty acid supplementation will, therefore, be required among people with type 2 diabetes and high CVD risks.
Several mechanisms may explain the favourable associations between
n-3 fatty acids and the risk of CVD. Previous clinical trials looking at intermediate cardiovascular outcomes among people with diabetes have reported that
n-3 fatty acid supplementation could lower triacylglycerol concentrations [
35,
36], improve arterial blood flow and attenuate inflammatory signals [
37,
38]. These effects were supported by clinical trials in the general population and in experimental studies [
39‐
41].
The present study observed the inverse associations between baseline plasma
n-6 fatty acids and LA with the risk of death, but these associations attenuated after adjustment for multiple risk factors. Similar findings have been reported between serum LA and mortality in a cohort of older adults (≥65 years of age) [
42]. We did not, however, detect significant associations between
n-6 fatty acids and LA with the risk of macrovascular events. These results were consistent with those of previous observational studies [
23,
24,
43] and directionally concordant with the recent pooled analyses of 30 cohort studies which reported that higher circulating and tissue levels of LA were associated with a lower risk of major cardiovascular events (HR 0.93 [95% CI 0.88, 0.99]) [
44]. In addition, our study did not detect significant associations with the risk of CVD for circulating MUFA and SFA, which is consistent with previous studies [
24,
43,
45], while some studies have reported that MUFA in blood were associated with a higher risk of CVD [
46].
There were no significant associations between any fatty acids and the risk of microvascular events. A further analysis specifically for renal outcome also demonstrated no significant associations. Although limited studies have assessed the association between circulating fatty acids and renal outcomes, in an Italian cohort of 931 adults, plasma
n-3 fatty acids were inversely associated with the risk of developing renal insufficiency (creatinine clearance rate <60 ml/min) [
47]. However, in a cohort of 2792 individuals, levels of
n-3 and
n-6 fatty acids and SFA in plasma phospholipid were not associated with kidney function [
48].
The strengths of the current study include the use of an efficiently designed case-cohort study from a well-characterised clinical trial to yield a powerful study for a range of outcomes, which were independently adjudicated according to pre-specified criteria. This study included multiple plasma fatty acids and the ability to adjust for multiple covariates including lipids and lipid-lowering drugs, such as statins. We also considered the percentage that individual fatty acids contributed to total fatty acids, as well as the absolute levels of fatty acid, and both measures are important for interpreting fatty acid values since an increased intake of a specific fatty acid could alter the relative percentage of other fatty acids while their absolute levels are unlikely to be altered. The present study, however, has several limitations. First, as the study cohort was derived from a randomised trial of individuals with type 2 diabetes, our results may have limited generalisability to broader populations. Second, as fatty acids were measured in pragmatically collected plasma samples in a randomised trial, we cannot rule out the potential for differential pre-analytical sample handling or sample degradation during storage, which may have biased our results [
49]. Further, as plasma samples were collected from non-fasted participants, the levels of fatty acids might have been affected by the consumption of a recent meal [
50], although, in clinical practice, fasting is rarely required among individuals with type 2 diabetes. Third, fatty acids were only measured in plasma samples collected at study baseline; thus, we were unable to consider how the change in fatty acid values during the study follow-up might have influenced the exposure–outcome association. Finally, our study considered only two individual fatty acids (DHA and LA), since the resolution of the employed high-throughput NMR platform was limited in terms of individual fatty acid types, and only allowed robust quantification of DHA within
n-3, and LA within
n-6. The specific set of measures was determined by their overall concentration in plasma and also on spectroscopic aspects, such as overlapping signals, which makes it challenging to quantify from native plasma where no lipid extraction is used [
14]. LA and DHA were reported since they generate distinct peaks in the spectral data produced by the measurement, and we were able quantify them separately as part of our high-throughput service. However, fatty acid concentrations quantified by the NMR metabolomics platform were highly consistent with the concentrations compared with GC, the latter being challenging with large samples [
46]. Further, NMR, is a novel technology with the potential of offering a cost-effective platform for multiple biomarker testing and has great potential in regard to fatty acid measurement.
In conclusion, we report distinct associations of different plasma fatty acids with the risk of major clinical outcomes in individuals with type 2 diabetes. In particular, plasma n-3 fatty acids were associated with a lower risk of macrovascular disease and death, and DHA was associated with a lower risk of macrovascular disease. These results support the cardioprotective effects of n-3 fatty acids and DHA and further merit testing the role of high-dose n-3 fatty acid supplementation in individuals with type 2 diabetes.
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.