Omega-3 intake is associated with attenuated inflammatory response and cardiac remodeling after myocardial infarction

Participants from the Brasília Heart Study (BHS) database were used for this analysis (n = 421). BHS is a an ongoing prospective observational cohort composed of STEMI patients admitted to Hospital de Base do Distrito Federal (Brasília, Brazil) since 2006 [21]. Briefly, BHS inclusion criteria were: (i) < 24 h after onset of MI symptoms, (ii) ST-segment elevation of at least 1-mm (frontal plane) or 2-mm (horizontal plane) in 2 contiguous leads and (iii) increased myocardial necrosis markers above reference limit of creatine kinase-MB (CK-MB) (25 U/L) and troponin I (0.04 ng/mL) followed by a decline of both markers. Participants with new or presumed left bundle branch block, development of pathological Q waves without ST-elevation recording or exclusively with imaging evidence of MI were excluded. Diagram flow is showed in Fig. 1.

Patients enrolled into the study underwent therapies based on current guidelines for treatment of acute coronary disease, as indicated by Table 1 data [22]. Attending physicians, who were blind to the study evaluations, made all therapeutic decisions.

Participants underwent a structured detailed clinical questionnaire, anthropometric measurements, blood collection for biochemical analysis, nutritional, and imagining evaluation. Ex-smoking status, diabetes and hypertension were previously defined elsewhere [23]. Coronary reperfusion therapy was assumed when the participant underwent to percutaneous coronary intervention (PCI) and/or fibrinolytic therapy to restore the blood flow during MI. Anthropometrical measurements obtained were body weight (kg), height (m), and waist circumference (WC). Killip and GRACE scores were evaluated in all enrolled patients. Briefly, the GRACE risk score is a well-validated multivariable algorithm for predicting 10-year death risk following an initial acute coronary syndrome (ACS) [24].

Blood samples were obtained within the first 24 h of MI symptoms (D1) and at day five of hospitalization (D5). High-sensitivity C-reactive protein (hs-CRP), urea, creatinine, triglycerides, total cholesterol, high-density lipoprotein cholesterol (HDL-C), low-density lipoprotein cholesterol (LDL-C), and glomerular filtration rate (GFR) were obtained as described elsewhere [23]. Interleukin-2 and B-type natriuretic peptide levels were quantified by ELISA.

A food frequency intake questionnaire (FFQ) was used to estimate nutritional composition and ω-3 intake [alpha-linolenic acid (ALA), eicosapentaenoic acid (EPA), and docosahexaenoic acid (DHA)] within the previous 3-month period. The questionnaire was applied at admission and at the 3rd month of follow-up. The food intake was clustered in 62 items divided into 10 groups (dairy products, eggs and meats, oils, snacks and canned foods, cereals and vegetables, greens and fruits, desserts and candies, beverages, diet and light products and spices). Approximate portions of usual intake of each item was recalled by patients with the aid of a photographic record for dietary surveys and the portion size was, subsequently, quantified into weights [25]. Total caloric and ω-3 intake were calculated based on a food composition database of the Brazilian Table of Food Composition (TACO) [26]. Briefly, TACO is a 399-item nutritional database with the most consumed foods in Brazil. Sample of foods were collected in 9 cities in all 5 Brazilian regions. Sampling was performed by mixing and packing two units of each principal brand collected in each site. The total units were separated in three different composites of 100–200 g, and then analyzed by certified laboratories. Contents of total lipids, cholesterol, saturated, poly- and monounsaturated fatty acids and trans fats were entered in the database [27]. Quantification of ω-3 from the questionnaires was performed by an experienced nutritionist (A.P.R.C.) and values were omitted from all other study participants.

Participants underwent 3D Echocardiography at the first- and sixth-month following MI (iE 33 system; Philips Medical Systems, Andover, MA) according to current guidelines [28]. Left ventricle end-diastolic diameter (LVEDD), end-systolic diameter (LVESD), septum diastolic thickness (SD), and posterior wall diastolic thickness (PD), left ventricle end-diastolic volume (LVEDV) and end-systolic volume (LVESV) were obtained in 3D mode, and Ejection fraction (EF) was derived from these volumes.

Cardiac Magnetic Resonance imaging (CMRi) and CMRi data analysis were performed according to descriptions shown in a previous study [29].

Normally distributed data were presented as mean ± SD and skewed data as median and interquartile range (IQR). Participants were stratified into two groups according to daily consumption of ω-3 (below or above the median). Chi-square or two-tailed t tests were used for comparison of baseline data. Analyses of covariance (ANCOVA) was used to assess the association between ω-3 consumption and the change (Δ) of hs-CRP, IL-2 and BNP levels between D1 and D5. ANCOVA adjusted by covariates was also used to compare infarcted mass and the change in LVEF and LVEDV between 30 and 180 days. All analyses of the changes between D1 and D5 or between the 30th and the 180th day were additionally adjusted for the baseline levels in order to attenuate the effect of regression toward the mean. Multivariate binary logistic regression was used to evaluate the association between the dichotomous dependent variable ω-3 intake below or above the median and the independent variables Δhs-CPR, ΔIL-2, ΔBNP, ΔLVEF and ΔLVEDV. These dependent variables were categorized into below or above their respective medians to bypass the non-normal distribution and as a strategy to level their effects sizes, allowing direct comparability between variables. Stepwise selection of variables was used to reach the final model using the covariates. The following covariates were used for ANCOVA and regression analyses and were selected using bootstrapping based on t test analysis (Table 1): age, gender, diabetes mellitus, hypertension, coronary reperfusion therapy, and use of ACE inhibitors/ARBs. Restricted cubic spline models were used to assess the relationship between daily consumption of ω-3 and hs-CRP levels. Splines were adjusted by the GRACE score [24] and plasma peak CKMB levels. The GRACE score was chosen because its validity was confirmed as a predictive instrument in STEMI patients and due to the possibility of aggregating into a single index relevant covariate, thus reducing the saturation of the regression models. Statistical analysis was performed using SPSS®, version 21 for Mac (IBM) and STATA, version 15.0 for Mac. A two-sided p-value of 0.05 was considered statistically significant.

Three months after STEMI, the food questionnaire was reapplied in order to evaluate the intake of ω-3 in this interlude. Participants who originally presented with ω-3 intake below the median (< 1.7 g/day) had no changes in their ω-3 intake during follow-up (from 1.0(0.7) to 1.1(0.9) g/day; p = 0.572). However, participants who originally had an intake of ω-3 above the median (≥1.7 g/day) reduced their ω-3 intake after STEMI (from 2.7(1.3) to 1.9(1.5) g/day; p < 0.001). Due to the natural limitation of the FFQ tool, it is not possible to specify the difference of 30% (0.8 g). Regardless, after 3 months, the higher intake group were still consuming significantly higher amounts of ω-3 (Table 1).

Comparative analyses of inflammatory markers are shown in Table 2. Participants who consumed ω-3 above or below the median had equivalent levels of hs-CRP at D1. Participants who consumed ω-3 above the median had higher levels of IL-2 at D1 than their counterparts. The increase in both hs-CRP and IL-2 levels from D1 to D5 were smaller among those who consumed ω-3 above the median than their counterparts. All these comparative results remained significant after adjustment for covariates as indicated above.

Binary logistic regression models are shown in Table 3 and were used to assess the association between the intake of ω-3 and study endpoints. The intake of ω-3 was inversely associated with increase in hs-CRP and IL-2 levels (Δhs-CRP and ΔIL-2 – Model 1). Both associations remained significant after full adjustment for covariates (Model 2). The addition of smoking habit as covariate did not change the results.

Restricted cubic spline models were used to assess the relation between the daily consumption of ω-3 across changes in hs-CPR levels (Fig. 2). The consumption of ω-3 was inversely associated with the change in hs-CRP levels even after fully adjusted analysis (p = 0.01).

Comparative analyses of BNP levels, infarcted mass, LVEF, LVEDV and SI are shown in Table 2. No difference in plasma levels of BNP was observed at D1. However, BNP levels significantly increased in both subgroups from D1 to D5 (p < 0.001 and p = 0.032 for subgroups below and above the median, respectively). In the subgroup analysis, those who consumed ω-3 levels above the median had lower levels of BNP at D5 than their counterparts. Accordingly, participants who consumed ω-3 levels above the median had a smaller change in BNP levels (ΔD5-D1) than their counterparts; this difference remained significant after adjustment.

No differences were observed in the extension of infarcted mass obtained by CMRi neither in the LVEF at first and sixth months after STEMI. However, participants who consumed ω-3 levels above the median had an increase of LVEF from the first to the sixth month after STEMI while their counterparts had a decrease of LVEF in this period of time. Also, LVEDV was not different between the groups at the first and sixth month after STEMI. Nevertheless, participants who consumed ω-3 levels above the median had a decrease in LVEDV from the first to the sixth month while their counterparts had an increase in LVEDV in the same period.

The associations between the intake of ω-3 and the changes in the levels of BNP, LVEF and LVEDV measures were tested by binary logistic regression models as shown in Table 3. The intake of ω-3 below the median was inversely associated with increase in BNP levels. Conversely, the intake of ω-3 below the median was directly associated with reduction in LVEF and decrease in LVEDV. Both associations remained significant after full adjustment for covariates.