Fat Quality Index and Risk of Cardiovascular Disease in the Sun Project

Abstract

Objectives

To examine the association between a dietary fat quality index (FQI), and the risk of incident cardiovascular events or deaths in the Seguimiento Universidad de Navarra (SUN) cohort.

Design

Longitudinal analysis during 10.1 years of median follow-up. Cox models were used to estimate adjusted hazard ratios (HR) of incident cardiovascular diseases (CVD) according to tertiles of FQI and of different fat subtypes.

Setting

University of Navarra, Spain.

Participants

19,341 middle-aged adults.

Measurements

Fat intake was measured with a validated food-frequency questionnaire. The FQI was calculated according to the ratio: (monounsaturated+polyunsaturated) / (saturated+trans fatty acids).

Results

We observed 140 incident cases of CVD. No association was found for FQI (HR=0.94, 95 %CI 0.61–1.47 for the highest vs the lowest tertile, p for trend=0.884). No significant associations were found for different dietary fat subtypes on CVD risk. The results suggest no clear association between a higher FQI and a higher amount of energy from fat and incidence of CVD (p for interaction: 0.259 and p for trend only among participants with a percentage of energy from fat ≥35% of total energy: 0.272).

Conclusion

In this Mediterranean cohort, the FQI was not associated with cardiovascular events. A “heart-healthy diet” should focus its attention on dietary fat sources and should use an overall dietary pattern approach, rather than limiting the focus on fat subtypes. More research is needed to validate dietary advice on specific fatty acids intake or saturated fatty acids replacements for reducing CVD risk.

Abreviations

BMI

body mass index

CI

confidence interval

CVD

cardiovascular disease

FFQ

food frequency questionnaire

FQI

fat quality index

HR

hazard ratio

MET

metabolic equivalent index

MUFA

monounsaturated fatty acids

PUFA

polyunsaturated fatty acids

SD

standard deviation

SFA

saturated fatty acids

SUN

Seguimiento Universidad de Navarra Follow-up University of Navarra

TFA

trans fatty acids

Introduction

Lifestyle behaviour, such as eating habits, physical activity and smoking, is strongly associated with health outcomes (1). Cardiovascular disease (CVD) is a major health concern and globally, is responsible of the largest proportion (37%) of non-communicable diseases deaths under the age of 70 years, according to the World Health Organization (2).

After 50 years of research, a clear link among diet, atherosclerosis and cardiovascular events has been established (3). Particularly, the Mediterranean diet is an effective and feasible tool to prevent CVD (44–6) by different mechanisms including reductions in blood pressure, body mass index (BMI), waist circumference, oxidative stress, low-grade inflammatory status and improvements in lipid profiles and vascular functions (7).

With respect to fat quantity, both prospective cohorts and randomized trials have confirmed that low-fat diets have no benefits for major chronic diseases, even more the replacement of fat with refined carbohydrates does not lower CVD risk (8).

The American Heart Association and the American College of Cardiology Guidelines (9), concluded that additional research is needed on the relative effects of saturated fatty acids (SFA), monounsaturated fatty acids (MUFA), polyunsaturated fatty acids (PUFA) and trans fatty acids (TFA), and that there are still controversies about the role of dietary fat quality on CVD (10).

A direct association between total fat intake and CVD outcomes is unlikely, but, the ideal subtype of unsaturated fat to be recommended remains unclear. Although some evidence suggests that the replacement of SFA and TFA with PUFA(11) or MUFA(12) may decrease CVD risk (13, 14)recent investigations showed certain doubts (15, 16). Ratio PUFA/SFA (17), index of each fatty acid expressed as a percentage of total fat intake (18), are previous explored indices of dietary fat quality.

On the other hand, during the last 2 decades, different dietary indexes have been developed to summarize overall dietary quality assessing only a small number of nutrients or foods. Thus, these brief tools could be applied to provide immediate feedback to patients in intervention studies and also to negociate potential changes with them in order to improve and monitor their dietary habits, always in the context of setting the goals in terms of easily understandable units (food servings). Besides, there is a great interest in nutritional epidemiology to know if indexes o scores of diet quality could predict chronic disease morbidity and mortality.

Thus, our aim was to prospectively investigate the association between a fat quality index (FQI) and the risk of CVD in a large Mediterranean cohort of middle-aged adults.

Methods

Subjects

The SUN Project is an ongoing, multipurpose, prospective, and dynamic cohort composed of Spanish university graduates. The design, recruitment strategy, follow-up and objectives of this project have been described previously (19). Briefly, recruitment started in December 1999 and it is permanently open. Participants are followed-up every two years by mailed or e-mailed questionnaires with questions to evaluate nutritional and lifestyle factors.

To ensure that all participants had the opportunity to answer the 2-y follow-up questionnaire, 21,678 participants who had completed a baseline questionnaire before March 2013 were considered eligible. For this current longitudinal analysis we excluded 198 participants with prevalent CVD at baseline (myocardial infarction or stroke), 428 participants with extreme values for total energy intake (under the 1th percentile and over the 99th percentile), and 1,711 subjects without any follow-up questionnaire, who were considered lost to follow-up (91.9% retention rate). Finally, 19,341 participants were included in our main longitudinal analysis.

The study was conducted according to the Declaration of Helsinki, and the Institutional Review Board of the University of Navarra approved it (ClinicalTrials.gov Identifier: NCT02669602). We considered a response to the initial questionnaire as informed consent to participate in the study.

Dietary assessment

The dietary exposure of this study was the FQI. This specific dietary score was assessed at baseline with a semi-quantitative food frequency questionnaire (FFQ) repeatedly validated in Spain with 136 food items (20, 21). Each item in the FFQ included a typical portion size, and consumption frequencies were measured in 9 categories that ranged from “never or almost never” to “6 times/day. Daily food consumption was estimated by a trained team of dietitians using a computer program based on available information on food composition tables for Spain (22, 23).

The FQI was calculated using the ratio (MUFA+PUFA)/(SFA+TFA) as a continuous variable, ranging from 0.62 to 5.93 among our participants. All these variables had the same weight in the formula. Participants were categorized into tertiles according this score.

This overall dietary score to assess fat intake quality has been previously used to investigate the association with nutritional inadequacy in this cohort (24) and also in the PREvención con DIeta MEDiterránea (PREDIMED) study (25).

Covariates section

The baseline questionnaire also collected information on the following covariate sections: socio-demographic variables, anthropometric characteristics (such as weight, height, and weight change during the past 5 years), medical history, family history of major diseases, obstetric history for women, as well as lifestyle and health related habits (including alcohol intake, smoking status, physical activity and time of television watching). Physical activity was evaluated with the use of a validated 17-item questionnaire (26). Finally, the reproducibility and validity of self-reported anthropometric data have been assessed in a subsample of the cohort, and described elsewhere (27). Adherence to the Mediterranean dietary pattern was evaluated using the 0 to 9 score proposed by Prof. Trichopoulou (28).

Response variable

The response variable for the present analysis was the composite of incident non-fatal acute coronary syndromes (myocardial infarction with or without ST elevation), non-fatal stroke or cardiovascular death. In all follow-up questionnaires, we asked for the medical records if participants reported any of these diagnoses, which were reviewed by an expert panel of physicians who categorized the events by the following criteria: a) non-fatal stroke was defined as a focal-neurological deficit of sudden onset and vascular mechanism that lasted more than 24 h; b) a non-fatal myocardial infarction was defined by the third universal definition for myocardial infarction and c) cardiovascular deaths were confirmed, according to the International Classification of Diseases. Deaths were reported to our research team by the subjects' next of kin, work associates and postal authorities. For participants lost to followup, the National Death Index of Spain was checked once a year to identify any deceased cohort members.

The follow-up time was defined as the interval between the date of recruitment to the date of cardiovascular death or event, or to the date of returning the last follow-up questionnaire.

Statistical analyses

Baseline characteristics of participants were described as mean and standard deviation for numerical variables and proportions for categorical variables.

To assess the relationship between baseline FQI and the risk of incident cardiovascular event or deaths, we used Cox proportional hazards models to estimate HRs and their 95% CIs with the use of the lowest category as the reference, after adjusting for potential confounders. For each participant we computed person-years of follow-up from the date of returning the baseline questionnaire to the date of CVD, the date of death or the date of returning the last follow-up questionnaire, whichever came first.

For FQI and different fat subtypes, we fitted a first multivariable-adjusted model only adjusted for age at recruitment (10 groups), sex and total energy intake (continuous); a second model, additionally adjusted for BMI (continuous), smoking habit (non-smoker, former smoker, current <15 cig/d, current ≥15 cig/d), leisure-time physical activity (metabolic equivalent-h/week), sitting (hours/week), following special diets at baseline, weight gain in the past 5 years previous to entering the cohort (≥3 kg), educational level (years of higher education), marital status (four categories); and a third model, additionally adjusted for family history of CVD, prevalent or previous angina pectoris or revascularization procedures, prevalent diabetes, prevalent hypertension, prevalent hypercholesterolemia, prevalent hypertriglyceridemia and prevalent depression. Finally, only when the FQI was the exposure variable, we performed a fourth model additionally adjusted for snacking between meals, protein intake, total carbohydrates intake and adherence to Mediterranean diet (three categories).

Tests of linear trend across increasing tertiles of FQI and different fat subtypes were conducted by assigning medians of each category and treating them as continuous variables in the respective Cox regression model.

To assess the robustness of our findings, we performed several sensitivity analyses by rerunning all the multivariable-adjusted Cox regression models under different assumptions as follows: including only men or only women; including only health professionals or only non-health professionals; adopting different allowed limits for total energy intake: under the 5th percentile and over the 95th percentile of total energy intake or <800 or >4,000 kcal/day for men and <500 or >3,500 kcal/day for women; including subjects with prevalent CVD; excluding participants who died early (in the first 2 years of follow-up); excluding late cases (after 10 years of follow-up); excluding participants with special diets at baseline or without baseline fat intake; and finally excluding participants who did not answer ≥9 items in the baseline FFQ.

In order to assess the effects of the amount of energy from fat across the tertiles of baseline FQI, we created a new variable combining the exposure to the baseline tertiles of the FQI and the baseline percentage of energy from fat (<35% and ≥35% of total energy). The participants with lowest tertile of FQI and lowest percentage of energy from fat were used as the reference.

Finally, we fitted several isocaloric substitution models to estimate the effect of replacing certain quantities of one different fat subtype for another type or for other macronutrients (e.g., 5% of energy from SFA by 5% of energy from PUFA).

All P values were 2-tailed, and P<0.05 was considered significant. Statistical analyses were performed with the use of STATA version 12.0 software (StataCorp LP).

Results

The main characteristics of the 19,341 participants categorized according to their baseline tertiles of FQI are presented in Table 1. Compared with subjects in the first tertile, those in the upper tertile of fat intake quality were older, women, former smokers, had a higher prevalence of hypertension, diabetes or dyslipidemia at baseline and better adherence to a Mediterranean dietary pattern. On average, subjects in the third tertile had higher total fiber, PUFA, MUFA, n-3, n-6 fatty acids and alcohol intakes. Besides, FQI was inversely associated with protein, SFA and TFA intakes. Finally, the FQI was directly associated with a higher fruit, vegetables, nuts, legumes, fish and olive oil consumption and with lower dairy products, meats and fast-food consumption, and those participants in the highest tertile of FQI were more likely to follow a special diet at baseline. During a median follow-up of 10.1 years, we documented 140 incident cases of CVD.

Table 1.

Baseline characteristics of the SUN Study participants without Cardiovascular Disease at baseline according to Fat Quality Index (N=19,341)

	T1	FQIT2	T3
N	6447	6447	6447
Age (years)	36.7 (11.8)	37.6 (11.8)	39.9 (12.5)
Women (%)	28.9	59.3	65.2
BMI (kg/m2)	23.3 (3.4)	23.5 (3.6)	23.5 (3.5)
Physical activity during leisure time (METs-h/week)	22.0 (23.8)	21.2 (21.9)	22.5 (23.3)
Sitting (hours/week)	5.3 (2.1)	5.3 (2.1)	5.2 (2.0)
Smoking status:
Former smokers (%)	25.3	26.6	32.2
Current smokers (%)	21.8	21.8	20.9
Hypertension (%)	7.2	7.7	9.4
Diabetes (%)	1.5	1.7	2.1
Dyslipidemia (%)	5.7	6.7	7.3
Special diet (%)	6.2	6.4	10.8
Snacking (%)	35.5	35.6	32.2
Educational level (years of higher education)	5.1 (1.5)	5.1 (1.5)	5.0 (1.5)
Mediterranean diet score	3.1 (1.5)	3.6 (1.5)	4.2 (1.5)
Glycemic index	51.6 (4.6)	52.5 (4.4)	52.6 (4.9)
Carbohydrate intake (% total energy)	43.1 (7.5)	43.7 (7.0)	44.2 (7.8)
Fiber intake (g/d)	26.6 (12.6)	29.0 (13.1)	33.0 (15.4)
Energy intake (kcal/d)	2483 (799)	2496 (760)	2500 (780)
Protein intake (% total energy)	18.8 (3.3)	18.3 (3.1)	17.2 (3.3)
Fat intake (% total energy)	36.8 (6.2)	36.6 (6.1)	37.1 (7.3)
PUFA intake (% total energy)*	4.5 (1.2)	5.3 (1.3)	5.9 (1.8)
MUFA intake (% total energy)	14.9 (3.0)	15.5 (3.2)	17.0 (4.4)
SFA intake (% total energy)	14.8 (3.2)	12.5 (2.4)	10.4 (2.5)
n-3 fatty acids (g/d)	2.6 (1.2)	2.8 (1.4)	3.0 (1.7)
n-6 fatty acids (g/d)	18.8 (18.5)	19.7 (13.2)	19.7 (14.7)
TFA intake (g/d)	1.3 (0.8)	1.0 (0.5)	0.7 (0.5)
Cholesterol (mg/d)	473.0 (181.5)	446.8 (161.3)	389.6 (157.6)
Alcohol intake (g/d)	4.4 (8.2)	4.9 (8.0)	5.0 (9.6)
Fruits (g/d)	326.6 (316.9)	351.4 (320.1)	419.0 (363.4)
Vegetables (g/d)	484.0 (343.6)	538.6 (349.5)	616.8 (402.5)
Nuts (g/d)	4.7 (6.5)	7.1 (10.1)	12.5 (20.9)
Legumes (g/d)	23.0 (19.7)	24.0 (18.9)	23.7 (21.1)
Dairy products (g /d)	311.4 (267.0)	198.4 (179.3)	125.6 (138.9)
Meats (g/d)	190.2 (92.5)	191.2 (87.1)	161.1 (79.3)
Fish (g/d)	93.2 (60.8)	102.2 (67.1)	108.6 (72.2)
Olive oil (g/d)	11.7 (9.4)	17.8 (12.3)	28.8 (18.8)
Fast-food (g/d)	26.1 (30.2)	24.4 (22.2)	17.5 (17.8)

^*Values are expressed as mean (standard deviation) unless otherwise stated. T: tertile. PUFA: Polyunsaturated Fatty Acids; SFA: Saturated Fatty Acids; MUFA: Monounsaturated Fatty Acids; TFA: Trans Fatty Acids.

Participants with the highest baseline FQI did not present a significant lower risk of developing CVD in comparison with those in the lowest tertile of FQI (adjusted HR: 0.80; 95% CI 0.53–1.20, p for trend: 0.366) in model 3 (Table 2). When additionally adjusted for total protein and carbohydrates intakes, adherence to Mediterranean diet and snacking between meals, the results were even further attenuated (HR: 0.94; 95% CI 0.61–1.47, p for trend: 0.884).

Table 2.

Hazard Ratios (HRs) and 95% confidence intervals (CI) of incident Cardiovascular Disease by tertiles of Fat Quality Index (FQI).

		Tertiles of FQI
	T1	T2	T3	p for trend
N	6447	6447	6447
FQI value	0.62-1.47	1.47-1.79	1.79-5.93
Incident cases of CVD	55	36	49
Persons-year	61,828	61,891	60,700
Model 1	1 (ref.)	0.69 (0.45-1.06)	0.83 (0.55-1.23)	0.439
Model 2	1 (ref.)	0.71 (0.46-1.10)	0.84 (0.56-1.26)	0.489
Model 3	1 (ref.)	0.67 (0.43-1.05)	0.80 (0.53-1.20)	0.366
Model 4	1 (ref.)	0.73 (0.47-1.15)	0.94 (0.61-1.47)	0.884

Model 1: Adjusted for age at recruitment (10 groups), sex and total energy intake (continuous); Model 2: Additionally adjusted for BMI (continuous), smoking habit (non-smoker, former smoker, current <15 cig/d, current ≥15 cig/d), leisure-time physical activity (metabolic equivalent-h/week), sitting (hours/week), following special diet at baseline, weight gain in the past 5 years previous to entering the cohort (≥3 kg), educational level (years of higher education), marital status (four categories); Model 3: Additionally adjusted for family history of CVD, prevalent or previous angina pectoris or revascularization procedures, prevalent diabetes, prevalent hypertension, prevalent hypercholesterolemia, prevalent hypertriglyceridemia and prevalent depression; Model 4: Additionally adjusted for snacking between meals, protein intake, total carbohydrates intake and adherence to Mediterranean diet (three categories).

Table 3 shows the association between the different fat subtypes and CVD. None of the associations studied were statistically significant, however point-estimates for the second and third tertiles pointed in a different direction for MUFA and PUFA (non-significant inverse association) than for SFA and TFA (non-significant direct association). In model 3, for the third tertile we found a HR (95% CI) of 0.95 (0.62–1.45) for MUFA, 0.72 (0.47–1.10) for PUFA, 1.22 (0.80–1.87) for SFA, and 1.10 (0.71–1.17) for TFA.

Table 3.

Hazard Ratios (HRs) and 95 % confidence intervals (CI) of incident Cardiovascular Disease by tertiles of MUFA, PUFA, SFA and TFA (% of total energy)

MUFA	T1	T2	T3	p for trend
N	6447	6447	6447
Incident cases of CVD	60	41	39
Median % MUFA	12.5	15.5	19.0
Model 1	1 (ref.)	1.02 (0.68-1.54)	1.03 (0.68-1.55)	0.826
Model 2	1 (ref.)	1.00 (0.67-1.51)	0.95 (0.63-1.44)	0.791
Model 3	1 (ref.)	0.95 (0.62-1.44)	0.95 (0.62-1.45)	0.805
PUFA	T1	T2	T3	p for trend
Incident cases of CVD	60	42	38
Median % PUFA	3.8	5.0	6.6
Model 1	1 (ref.)	0.83 (0.55-1.23)	0.75 (0.50-1.14)	0.177
Model 2	1 (ref.)	0.81 (0.54-1.21)	0.71 (0.48-1.11)	0.139
Model 3	1 (ref.)	0.80 (0.52-1.21)	0.72 (0.47-1.10)	0.130
SFA	T1	T2	T3	p for trend
Incident cases of CVD	55	42	43
Median % SFA	9.6	12.4	15.4
Model 1	1 (ref.)	1.13 (0.75-1.70)	1.27 (0.84-1.91)	0.253
Model 2	1 (ref.)	1.10 (0.73-1.66)	1.18 (0.78-1.78)	0.443
Model 3	1 (ref.)	1.12 (0.74-1.71)	1.22 (0.80-1.87)	0.356
TFA	T1	T2	T3	p for trend
Incident cases of CVD	52	45	43
Median % TFA	0.2	0.3	0.5
Model 1	1 (ref.)	1.06 (0.71-1.59)	1.10 (0.72-1.62)	0.661
Model 2	1 (ref.)	1.06 (0.70-1.59)	1.07 (0.70-1.64)	0.754
Model 3	1 (ref.)	1.11 (0.73-1.69)	1.10 (0.71-1.71)	0.666

Model 1: Adjusted for age at recruitment (10 groups), sex and total energy intake (continuous). Model 2: Additionally adjusted for BMI (continuous), smoking habit (non-smoker, former smoker, current <15 cig/d, current >=15 cig/d), leisure-time physical activity (metabolic equivalent-h/week), sitting (hours/week), following special diet at baseline, weight gain in the past 5 years previous to entering the cohort (≥3 kg/<3 kg), educational level (years of higher education), marital status (four categories). Model 3: Additionally adjusted for family history of CVD, prevalent or previous angina pectoris or revascularization procedures, prevalent diabetes, prevalent hypertension, prevalent hypercholesterolemia, prevalent hypertriglyceridemia and prevalent depression.

Finally, we conducted several sensitivity analyses to verify the robustness of these results. Table 4 shows the HR and 95% CI of CVD for the comparisons between the highest vs the lowest tertile of FQI. Results did not substantially change in any of these scenarios. Thus, we observed no association between FQI and the incidence of CVD in all sensitivity analyses. The point estimates were in the same direction as the main HR observed, except to when we included only women (n=11,826 with only 24 incident cases) or only health professionals (n=10,488 with 58 incident cases). No significant interaction was observed between sex and the FQI (p for interaction: 0.229).

Table 4.

Sensitivity analyses. Adjusted^* Hazard Ratios (HRs) and 95 % confidence intervals (CI) of incident Cardiovascular Disease by tertiles of FQI

	n	CVD	FQIHR (95% CI)*	P for trend
Overall	19341	140	0.94 (0.61-1.47)	0.884
Including only men	7515	116	0.73 (0.45-1.19)	0.237
Including only women**	11826	24	3.16 (0.79-12.63)	0.099
Including only health professionals	10488	58	1.40 (0.68-2.91)	0.301
Including only non-health professionals	8893	82	0.69 (0.38-1.25)	0.239
Energy limits: percentiles 5 to 95	17775	122	0.82 (0.50-1.34)	0.452
Energy limits: Men (<500 or >3,500 kcal/d). Women (<800 or >4,000 kcal/d)	17874	135	0.95 (0.60-1.51)	0.908
Including prevalent CVD	19514	156	0.82 (0.56-1.21)	0.440
Excluding early cases (2 first y)	19317	116	0.92 (0.57-1.49)	0.811
Excluding late cases (≥10 first y)	19321	120	0.86 (0.54-1.38)	0.660
Excluding participants with special diet at baseline or without baseline fat intake	17830	121	0.91 (0.57-1.47)	0.781
Excluding participants with≥ 9 missing values in the FFQ	15587	87	0.95 (0.53-1.71)	0.663

^*Multivariable model: Adjusted for age at recruitment (10 groups), sex, total energy intake (continuous), BMI (continuous), smoking habit (non-smoker, former smoker, current <15 cig/d, current >=15 cig/d), leisure-time physical activity (metabolic equivalent-h/week), sitting (hours/week), following special diet at baseline, weight gain in the past 5 years previous to entering the cohort (≥3 kg/<3 kg), educational level (years of higher education), marital status (four categories), adherence to Mediterranean diet (three categories), family history of CVD, prevalent or previous angina pectoris or revascularization procedures, prevalent diabetes, prevalent hypertension, prevalent hypercholesterolemia, prevalent hypertriglyceridemia, prevalent depression, snacking between meals, protein intake, total carbohydrates intake, MUFA intake, PUFA intake, SFA intake, and trans fat intake;

^**p for interaction: 0.229

Figure 1 represents the HR for CVD incidence according to the joint classification by both the percentage of energy from fat (<35% and ≥35% of total energy) and the baseline tertiles of FQI. Unexpectedly, the results suggest no clear association between a higher FQI and a higher amount of energy from fat and incidence of CVD (p for interaction: 0.259 and p for trend only among participants with a percentage of energy from fat ≥35% of total energy: 0.272).

Figure 1.

Hazard Ratios of incident Cardiovascular Disease according to the joint classification by tertiles of FQI and the baseline percentage of energy from fat (<35% and ≥35% of total energy)

Finally, none of the several isocaloric substitution models assessing exchanges between different subtypes of fat showed statistically significant results, though some point-estimates suggested associations in the expected directions: replace 5% SFA by 5% MUFA, HR=1.40 (0.75–2.61); replace 5% SFA by 5% PUFA, HR=1.30 (0.64–2.63); replace 1% TFA by 1% MUFA HR=0.60 (0.17–2.10); replace 2% TFA by 1% PUFA HR=0.33 (0.02–4.17); replace 5 % SFA by 5% carbohydrates HR=1.23 (0.84–1.97) and finally, replace 5% SFA by 5% refined carbohydrates HR=1.14 (0.74–1.78).

Discussion

In this study we investigated the longitudinal association between an overall dietary fat quality index built according to the intake of the different subtypes of fat (FQI=[MUFA+PUFA]/[SFA+TFA]) and the risk of CVD. We have not found any association between a better quality of fat (as assessed by the FQI) and the risk of incident CVD. Regarding the specific role of dietary fat subtypes on CVD risk, no significant associations were observed.

In our sample participants in the best categories of the FQI had a higher energy intake and a higher percentage of total energy from fat. As expected, those individuals in the best category of the FQI had higher intake of MUFA, PUFA, n-3 and n-6 fatty acids, whereas SFA and TFA were lower. Besides, participants in the best category of FQI consumed higher amounts of nuts, olive oil, fish and less meats and dairy products. Notwithstanding these differences in quality of dietary fats, we were not able to find significant associations between dietary fat subtypes on CVD risk, neither a clear inverse association for PUFA or MUFA nor a positive association for SFA or TFA intakes.

In our sample, no clear association between a higher FQI and a higher amount of energy from fat and incidence of CVD was found. The traditional diet-heart advice has consistently recommended reducing total fat intake to 30% of energy or less, but results from trials were disappointing (3, 29). In fact, the 2015 American Dietary Guidelines (30), proposed for the first time that dietary advice should put the emphasis on optimizing the subtypes of dietary fat and not reducing total fat intake.

However, it is not clear whether a low fat diet, a modified fat diet, or a combination of both(31) is better for cardiovascular prevention and also whether MUFA (13) or PUFA (32) are more beneficial.

Nevertheless, previous results supporting a detrimental effect of dietary TFA on CVD risk are consistent (33). In this cohort, TFA intake was extremely low, its median ranged from 0.2 to 0.5 g/d across tertiles. In Spain, the TFA content is still incomplete in several food composition tables and the content in processed products such as margarines, factory-baked goods, among others, has declined over recent years and it is less than 1% in more than 90% of all these products (3434).

CVD risk can be modestly reduced by decreasing SFA and replacing it by a combination of MUFA and PUFA (35) or even more effective, by PUFA (29, 33), especially in men (11). In this sense, we observed an inverse association only between the FQI and incidence of CVD when we included only men in the analysis. Other investigations, have also found that sex may modify the association between total fat or fat subtypes intake and CVD risk (11, 37).

We did not find any significant association for several isocaloric substitutions for SFA and TFA. Results from the European Prospective Investigation into Cancer and Nutrition-Netherlands cohort, showed that lower ischemic heart disease risk did not depend on the substituting macronutrient but appeared to be driven mainly by SFAs from dairy sources (38). In this sense, Michas et al (35), proposed that the total matrix of a food is more important than just its fatty acid content in predicting the effect of a food on CVD risk, and they concluded that a healthy dietary pattern should be promoted (4, 8, 9, 39). In this context, the International Expert Movement to improve dietary quality, has also proposed that dietary recommendations should shift their focus away from single nutrients toward more healthy foods and dietary patterns, such as the Mediterranean diet (15, 40).

It should be noted, that our sample was a middle-aged cohort of Mediterranean adults and that the primary advantage of the Mediterranean diet appears to lie on its synergy among various cardioprotective nutrients, foods and bioactive compounds (41). An illustrative example is the case of virgin olive oil, that apart from MUFa provides numerous phenolic compounds (7, 42).

Furthermore, food processing pertinent to the Mediterranean diet (including certain culinary techniques, or the use of herbs and spices) enhance the nutritional quality of foods (43). The myth that frying foods is generally associated with a higher risk of CVD is not supported by the available evidence (44), even more, Mediterranean sofrito (a special sauce made with olive oil, garlic, tomato and onion) could also reduce CVD risk (45). Olive oil is the major source of fat in the Mediterranean diet.

We acknowledge that the present study has several limitations. First, we used a self-reported FFQ, which has measurement errors and may underestimate true associations. However, in large epidemiological studies, the FFQ is the most practical and feasible tool to assess dietary patterns. Reliability and validity of the FFQ used in SUN cohort has been extensively evaluated, showing good correlation for nutrient and food intake (21). Besides, the recent fat reformulation in food industry (34) is not fully reflected in Spanish food composition tables. Second, residual confounding may have affected our results although we attempted to adjust for the major CVD risk factors. Third, our participants had high educational level and low prevalence of cardiovascular risk factors and the small number of observed cardiovascular events, especially among women, may be associated with a low statistical power. Fourth, we do not have repeated information on diet as to allow us to assess changes in tertiles of the FQI during the follow-up, because in the SUN project the full-length FFQ is self-administered only twice along the study, at baseline and after 10 years of follow-up, and the small number of cases occurring after 10 years, does not allow us to use the 10 year FFQ as the second point in time to assess changes in the FQI (46, 47). Finally, the homogeneity of participants in the SUN cohort (all were university graduates) may limit the generalizability of our findings to the general population.

The strengths of the present study are: use of data obtained from large, adult cohort with a long-term follow-up and high response rate, the use of validated questionnaires and medical confirmation of cardiovascular events and the prospective design of the research which includes multiple and different variables as potential confounders.

In conclusion, in our cohort, the FQI did not predict the risk of cardiovascular events. The role of fat quality intake on CVD prevention in the context of a Mediterranean diet needs to consider that the effect of dietary fat sources and overall dietary pattern are likely to be greater than the isolated effect of dietary fat. Thus, the “heart-healthy diet” should focus its attention on an overall food pattern approach, rather than to focus on limiting total fat intake or some fat subtypes (48). More research is needed to identify the most sensible dietary advice on specific fatty acid intake or SFA replacements for reducing CVD risk.

Funding sources

The Seguimiento Universidad de Navarra (SUN) Project has received funding from the Spanish Ministry of Health -Instituto de Salud Carlos III and European Regional Development Fund (FEDER) (RD06/0045, CIBER-OBN, Grants PI14/01668, PI14/01798, PI14/01764 and G03/140) , the Navarra Regional Government (45/2011, 122/2014),and the University of Navarra.

Conflict of Interest

The authors declare no conflict of interest.

Ethical standard

The study was conducted according to the Declaration of Helsinki, and the Institutional Review Board of the University of Navarra approved it (ClinicalTrials.gov Identifier: NCT02669602).