The current work was conducted within framework of the CORDIOPREV study (https://www.Clinical Trials.gov Identifier: NCT00924937), an ongoing prospective, randomized, open, controlled trial in 1,002 patients with CHD who had their last coronary event over six months before enrolling, and who followed two different dietary models (a low-fat diet and the Mediterranean diet) over a period of seven years, in addition to conventional treatment for CHD. CORDIOPREV inclusion and exclusion criteria can be summarized as follows: patients were eligible if they were over 20 years old but under 75, had established CHD without clinical events in the last 6 months, were thought to be capable of following a long-term dietary intervention, and did not have severe diseases or an estimated life expectancy of less than seven years (23). At the beginning of the CORDIOPREV study, 350 patients out of 1,002 had T2DM (diab group, 280 men and 70 women). In the patients without established diabetes, an oral glucose tolerance test was performed during recruitment. T2DM was defined by levels of HbA1c <6.5%, fasting plasma glucose <126 mg/dl and 2 h plasma glucose in the 75 gr OGTT <200 mg/dl, according to the American Diabetes Association (ADA) diagnosis criteria (24). The results confirmed that 462 patients out of 1,002 did not have diabetes (non-diab group, 389 men and 73 women); however, 190 out of 1,002 were newly-diagnosed patients who had been diagnosed with short-duration diabetes and were not receiving glucose-lowering treatment (new-diab group, 158 men and 32 women). We followed the criteria of the Third Report of the National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III) to assess the presence of MetS (25). The CORDIOPREV study has been approved by the Reina Sofia (Cordoba) University Hospital Ethics and Research Committees. All the participants agreed to their inclusion in these studies. The trial protocol and all its amendments were approved by the Reina Sofia University Hospital Clinical Research Ethics Committee, following the Helsinki Declaration and good clinical practice.

The study design has been previously described (23). Briefly, participants were randomized to receive two diets: a Mediterranean (Med) diet or a low-fat (LF) diet. The LF diet consisted of <30% total fat (<10% saturated fat, 12%-14% MUFA fat, and 6%–8% PUFA fat), 15% protein and a minimum of 55% carbohydrates. The Med diet comprised a minimum of 35% of calories as fat (22% MUFA fat, 6% PUFA fat and <10% saturated fat), 15% proteins and a maximum of 50% carbohydrates. In both diets, the cholesterol content was adjusted to <300 mg/d. The outcomes are ascertained on a yearly basis by a Clinical Events Committee whose members are blinded to the intervention group. During the programmed visits, patient compliance is checked based on the scores of adherence to the Med and LF dietary patterns and by measuring selected biological plasma variables, such as the fatty acids profile. Dietary adherence was assessed with the 14-point Mediterranean Diet Adherence Screener and 9-point low-fat diet adherence (26). No energy restriction was implemented and the study team explicitly did not promote physical activity. The dietary intervention in the CORDIOPREV study included individual face-to-face visits every 6 months, group sessions every 3 months, and telephone calls every 2 months, all of which aimed at guaranteeing frequent contact between the patients and dietitians (at least 12 interactions per year). During the study, the following interventions were done: regular contacts, group sessions, monitoring of adherence, goal setting, social support, and the provision of foods.

Blood was collected in tubes containing EDTA to give a final concentration of 0.1% EDTA. The plasma was separated from the red blood cells by centrifugation at 1,500 × g for 15 min at 4°C. Analytes in the frozen samples, blinded to the team members, were analyzed centrally by members of the laboratory research team at the Lipid and Atherosclerosis Unit at Reina Sofia University Hospital. The clinical plasma parameters were measured as previously described (27). Briefly, the different parameters were measured using spectrophotometric techniques (enzymatic colorimetric methods): hexokinase method for glucose, chemiluminescence for insulin, glycated haemoglobin by high performance liquid chromatography and oxidation-peroxidation for total cholesterol, high-density lipoprotein cholesterol (HDL-C) and triglycerides, immunoturbidimetry for C-reactive protein (28, 29). The low-density lipoprotein cholesterol (LDL-C) was calculated using the Friedewald formula (provided the triglyceride level was <300 mg/dl) (30).

The liposaccharide (LPS) was measured using the limulus amebocyte lysate test (QCL-1000 Chromogenic LAL (LonzaIberica S.A., Spain), as previously (31). LPS binding protein (LBP) levels were determined using a human LBP ELISA kit (HycultBiotech, Netherlands).

TMAO plasma levels were measured using a 1,200 series LC system from Agilent Technologies coupled with an Agilent 6,460 triple quadrupole mass spectrometer from Agilent Technologies (Palo Alto, CA, USA). The MRM transitions used in this research are listed in Supplementary Table S1. The Agilent MassHunter Workstation software (version B.03.01, Agilent Technologies, Santa Clara, CA, USA) was used for data acquisition, and Agilent MassHunter Workstation Quantitative Analysis software was used for the quantitative analysis (version B.07.00).

We assessed the cardiovascular risk was assessed the intima-media thickness intima-media thickness of both common carotid arteries (IMT-CC), considered an early marker of atherosclerosis (32), and more recently an indicator of arterial injury. Moreover, the carotid study also included the determination of the number and height carotid plaques. This analysis was performed by technicians blinded to the clinical information and previous imaging. The carotid study was ultrasonically assessed bilaterally by quantification of intima-media thickness of both common carotid arteries (IMT-CC) and carotid plaques (number and height) as previously (33).

The PASW statistical software package, version 20.0 (IBM Inc., Chicago, IL, USA), was used for univariate statistical analyses of the data. Multiple logistic regression analysis was carried out including the baseline values for age, BMI, waist circumference, insulin, total cholesterol (TC), HDL-c, LDL-c, triglycerides, C-reactive protein, and blood pressure as dependent variables and T2DM (yes/no) as independent variable. The variables corresponding to plasma levels of glucose and HbA1c were removed from the multiple logistic regression analysis as these variables are diagnostic criteria for T2DM. We used One-way ANOVA to test the differences between groups at baseline. The chi-square test was applied to establish differences in T2DM and MetS prevalence, and MetS criteria analysis. Baseline values and values after 3 years of dietary intervention were analyzed together by ANOVA for repeated measures with time as intra-subject factor, and sex as the inter-subject factor, in order to test sex-specific changes induced by the consumption of healthy diets (Figure 1), low-fat diet (Supplementary Figure S2) and Mediterranean diet (Supplementary Figure S3). We used time as intra-subject factor, and diet as the inter-subject factor, in order to test diet-specific changes induced by the consumption of low-fat diet and Mediterranean diet in CHD men (Supplementary Figure S4) and CHD women (Supplementary Figure S5). The variables for age, body mass index, total cholesterol, low-density lipoproteins, and blood pressure were used as co-variables in the adjusted models. The post hoc statistical analysis was completed using Bonferroni's comparison test. P-values <0.05 were considered statistically significant.

We observed a higher age, BMI, plasma levels of TC, HDL-c, LDL-c, C-reactive protein, and diastolic blood pressure, but lower waist circumference in CHD women than in CHD men (all, p < 0.05) (Table 1). These sex differences were found between CHD men and CHD women, independently of the T2DM status.

We also observed that age, BMI, waist circumference, plasma levels of glucose, insulin, HbA1c, total cholesterol (TC), HDL-c, LDL-c, triglycerides, C-reactive protein, and systolic blood pressure all increased from the non-diab to the new-diab group and to the diab group in both CHD men and CHD women (all, p < 0.05) (Table 1). These differences between groups according to their T2DM status were found in CHD men and CHD women separately.

We found statistically significant sex differences in waist circumference, plasma levels of TC, and HDL-c, in the non-diab, new-diab and diab groups separately (Table 1). However, sex-differences in age, C-reactive protein, and systolic blood pressure were only observed between CHD diab men and CHD diab women. In fact, age, C-reactive protein, and systolic blood pressure were higher in CHD diab women than in CHD diab men (p = 0.011, p = 0.003, and p = 0.042, respectively), whereas no differences were observed between sexes in the non-diab group (Table 1).

We observed that the percentage of women with MetS in the CORDIOPREV population was higher than the percentage of men (P = 0.010). Indeed, the percentage of women with a small number of MetS criteria was lower than men, but the percentage of women with a greater number of MetS criteria was higher than men (P = 0.037). However, we did not observe statistically differences in the percentage of men and women according to the T2DM status, while a trend for lower percentage of non-diab and higher percentage of diab women than men was noticeable (Table 2). A multiple logistic regression analysis was conducted to assess the relationship of the anthropometric and biochemical variables and the prevalence of T2DM (new-diab and diab groups together vs. non-diab group) (Supplementary Figure S1). Among the MetS criteria (Supplementary Table S2), we found a higher percentage of women following the criteria of waist circumference and HDL plasma levels than men (P < 0.001 and P = 0.001, respectively), whereas no differences were found in the criteria based on glucose, triacylglycerides and hypertension.

We found that IMT-CC and the number of plaques were higher in CHD men than in CHD women (both, p < 0.05), whereas no sex differences were found in the percentage of carotid plaques presence and plaques height. The sex differences found between CHD men and CHD women, are independent of the T2DM status.

We found that IMT-CC increased from the non-diab to the new-diab group and to the diab group in both, CHD men and CHD women (both, p < 0.001). We observed that the number of plaques and the plaques height increased from the non-diab to the new-diab group and to the diab group in CHD men (p = 0.039 and p = 0.025, respectively), which was not observed in CHD women. However, a trend for the increase of the number of plaques from the non-diab to the new-diab group and to the diab group in CHD women was noticeable. These differences between groups according to their T2DM status were found in CHD men and CHD women separately. Finally, no differences in the percentage of carotid plaques presence were observed according to T2DM status in CHD men nor CHD women.

We observed that IMT-CC and the number of plaques were lower in CHD non-diab women than in CHD non-diab men, whereas no differences were observed between sexes in CHD new-diab or CHD diab groups (Table 3). Thus, our results showed sex differences in IMT-CC and the number of plaques in non-diab CHD patients, while no sex differences we found in CHD diab patients.

We explored the modulation of the LPS, LBP and TMAO plasma levels by the consumption of a LF and the Med diet. These analyses were carried out taken together both LF and Med diets and separately for LF and Med diets. The consumption of healthy diets administered in our study for 3 years reduced the LPS plasma levels in new-diab group (p < 0.001), whereas no diet effect was observed in LPS plasma levels in non-diab and diab groups (Figure 1). We also found a reduction on the LPS plasma levels in new-diab group by the consumption of the LF and by the consumption of the Med diet (Supplementary Figures S2, S3, respectively). In addition, we observed lower LPS plasma levels in CHD non-diab women than in CHD non-diab men (P = 0.044) taken together baseline and y3 plasma levels, whereas no sex differences were found in the CHD new-diab and CHD diab groups (Figure 1). Moreover, we found no modulation by the consumption of healthy diets nor sex differences in the LBP plasma levels. When we analyze the effect of LF and MED diets LPS plasma levels by sex separately (Supplementary Figures S4, S5), we observed a reduction of the LPS plasma levels in new-diab group in men and women irrespective of the diet consumed (p < 0.001 and p = 0.031, respectively), whereas no diets effect was observed in LPS plasma levels in non-diab and diab groups.

By other hand, we did not observe changes on the TMAO plasma levels by the consumption of LF and Med diets (Figure 1, Supplementary Figures S2, S3). However, we observed lower TMAO plasma levels in CHD non-diab women than in CHD non-diab men (P = 0.049) (Figure 1). We found comparable results on TMAO plasma levels when we analyzed LF and Med diets separately (Supplementary Figures S1, S2).