Introduction
Evidence is now emerging that the incidence of metabolic disorders such as cardiovascular diseases and type 2 diabetes could be affected by dietary fatty acid composition. Among all the dietary lipids, the present review will focus on trans-fatty acids (TFA), since they are the object of a worldwide public health discussion, which requires solid and pertinent metabolic demonstrations in order to promulgate new nutritional recommendations. In fact, food safety authorities need clear, scientifically proven information to further legislate on the subject. TFA are MUFA or PUFA in which the hydrogen atoms are on opposite sides of the fatty acid double bonds, therefore changing their configuration and their chemical properties. TFA represent on average 1 to 2 % of daily energy intake. The main food sources of TFA are partially hydrogenated oil (industrial products) and ruminant-derived foods (dairy products and meat). The isomers of 18 : 1 (oleic acid, 18 : 1Δ9c or 18 : 1 cis-9) have been found to contribute 54 to 82 % of the total TFA intake(1). The major TFA formed by partial hydrogenation of vegetable oils (industrial TFA; I-TFA), derived from oleic acid, is elaidic acid (18 : 1Δ9t or 18 : 1 trans-9), whereas the main TFA resulting from rumen biohydrogenation (ruminant-TFA; R-TFA) is vaccenic acid (18 : 1Δ11t or 18 : 1 trans-11)(Reference Bauman, Mather and Wall2) (Fig. 1). In partial hydrogenation of vegetable oils, the trans-18 : 1 isomers usually vary from 4 to 16 carbons from a Gaussian distribution that centres about the d9 or d10 double bond. The distribution depends first on the starting vegetable oil and second on the extent of hydrogenation. For instance, in mildly hydrogenated soyabean oils, 10t-18 : 1 and 11t-18 : 1 are the major isomers whereas 9t-18 : 1 is the major one in heavily hydrogenated soyabean oil(Reference Wolff and Sebedio3). The distribution is different in ruminant fats. The major isomer is always 11t-18 : 1, reaching about 70 % of the total trans-18 : 1, whereas 10t-18 : 1 and 9t-18 : 1 appear in small amounts(Reference Ratnayake4).
The R-TFA term comprises total TFA (all the geometrical isomers of MUFA and PUFA) having non-conjugated, carbon–carbon double bonds in the trans configuration, except the conjugated linoleic acids, according to the definition of the French Food Safety Agency (L'Agence française de sécurité sanitaire des aliments; AFSSA, now known as ANSES)(Reference Leger and Razanamahefa5). The health effects of both industrial and ruminant sources of TFA are an important issue, since both are widespread ingredients in the food industry. As will be discussed later, the metabolic effects of I-TFA have been widely studied, whereas the impact of R-TFA has thus far been ignored. Indeed, milk fat contains low amounts of TFA (3–7 % by wt) compared with partially hydrogenated oil (10–60 % by wt), making it difficult to compare their respective effects(Reference Jensen6). The study of specific effects of the TFA isomer of ruminant origin has become possible since the discovery that milk fat TFA content can be enhanced by modifying the cows' diet(Reference Bauman and Griinari7). Understanding the specific effect of ruminant-sourced TFA is an important issue for two main reasons. First, in Europe, the contribution of TFA from ruminant fat ranged from about 30 to 80 % of total TFA, corresponding to 0·3–0·8 % of energy(Reference Hulshof, van Erp-Baart and Anttolainen8) and perhaps even more since food safety authorities incite food companies to decrease the amount of industrially sourced TFA in food formula(1, Reference Leger and Razanamahefa5). Second, the average distribution of R-TFA intake is about 85 % from dairy products and 15 % from ruminant meat(Reference Chilliard, Ferlay and Mansbridge9, Reference Chilliard and Ferlay10); consequently, the intake of R-TFA is mainly determined by the intake of high-fat dairy products. In fact, the cows' diet is the major environmental factor regulating milk fat composition and many strategies have been developed to reduce medium-chain SFA, for instance fresh pasture modifies the SFA/MUFA-PUFA ratio, decreasing milk fat SFA and enhancing milk fat conjugated linoleic acids and TFA(1, Reference Jakobsen, Bysted and Andersen11).
In the present review, to identify the most accurate papers, we used both PubMed databases and ISI Web of KnowledgeSM, using ‘trans fatty acid’ and ‘cardiovascular’ as keywords leading to 291 hits for the cardiovascular section, where the only inclusion criterion was human; for the diabetes part, ‘trans fatty acid’ and ‘diabetes’ were used as keywords, there were no inclusion or exclusion criteria, and 208 hits were recovered.
Trans-fatty acids and cardiovascular risk biomarkers
Impact of trans-fatty acids on the risk of CHD
A complete review on the effects of TFA on risks for CVD has been recently published by Micha & Mozaffarian in 2009(Reference Micha and Mozaffarian12, Reference Micha and Mozaffarian13). In the present review, we will only focus on the studies that distinguished the origins of the dietary TFA (Table 1).
R-TFA, ruminant TFA; I-TFA, industrial TFA; TEI, total daily energy intake estimated as 2000 kcal/d (8370 kJ/d); RR, relative risk.
* Values are expressed as RR (95 % CI) except for the Scottish Heart Health Study, where they are expressed as OR (95 % CI), and the study from four Danish cohorts, where they are expressed as number of events.
† Adjusted for age, treatment group, smoking, BMI, blood pressure, intakes of energy, alcohol and fibre, education and physical activities.
‡ Adjusted for age, weight, height, smoking, physical activity, diastolic pressure, TEI, intakes of SFA, linoleic acid and antioxidant vitamins.
Few studies investigated the association between the total intake of TFA and the risk of CHD; moreover, to our knowledge, only five studies looked at the association between intake of R-TFA and CHD. Among them, three prospective studies(Reference Willett, Stampfer and Manson14–Reference Jakobsen, Overvad and Dyerberg16) and one case–control study(Reference Ascherio, Hennekens and Buring15) showed no significant association between intake of R-TFA and risk of CHD and one no differences between the different TFA sources(Reference Oomen, Ocke and Feskens17).
In the Nurses' Health Study, after 20 years of follow-up, total TFA intake was positively associated with CHD risk, as the multivariate relative risk for the highest v. the lowest quintile of intake was 1·33 (P for trend = 0·01). However, when these authors did an analysis excluding women who in 1980 reported that their margarine intake had greatly changed in the previous 10 years, after adjustment for established risk factors and dietary lipids, the relative risk (highest v. lowest quintile of trans-isomer intake) of CHD with the intake of trans-isomers from animals was 0·59 (95 % CI 0·30, 1·17; P for trend = 0·230), which means no significant associations between R-TFA intake and the risk of CHD(Reference Willett, Stampfer and Manson14).
In a case–control study conducted in the Boston area (MA, USA) with both men and women aged about 57 years, Ascherio et al. (Reference Ascherio, Hennekens and Buring15) observed no significant association between the intake of trans-isomers from animal fats (R-TFA) (about 0·7 % of energy in the highest quintile) and myocardial infarction: relative risk adjusted for established risk factors, dietary lipids and energy was 1·02 (95 % CI 0·43, 2·41; P for trend = 0·57).
Recently, results obtained from 3686 men and women from four Danish cohort studies(Reference Jakobsen, Overvad and Dyerberg16) suggest that an intake of R-TFA of about 1·1 % of total energy intake (TEI) (about 2·7 g/d for women and 3·4 g/d for men in the highest quintile) is not associated with a higher risk of CVD. No evidence of a higher risk associated with R-TFA intake within the wide range of intake among both women (90 % central range: 0·5–3·1 g/d) and men (90 % central range: 0·6–4·1 g/d) was found.
Moreover, in the Alpha-Tocopherol, Beta-Carotene Cancer Prevention Study, which was a randomised, double-blind, placebo-controlled primary prevention trial, 21 930 men (aged 50–69 years) followed-up for 5–8 (median 6·1) years were analysed(Reference Pietinen, Ascherio and Korhonen18). Men were grouped into quintiles of energy-adjusted intakes of nutrients as calculated from the dietary questionnaire. No association was reported between coronary death and dietary intake of TFA of animal origin (P for trend = 0·857). The relative risk of coronary death after adjustment for age, smoking, BMI, blood pressure, intakes of energy, alcohol, and fibre, education ( < 7, 7–11, >11 years) and physical activity ( < 1, 1–2, >2 times per week) was 0·83 (95 % CI 0·62, 1·11) (P for trend = 0·035) for animal TFA for the highest (2·5 g, about 33 % of total TFA intake) compared with the lowest (0·6 g/d, about 33 % of total TFA) quintile(Reference Pietinen, Ascherio and Korhonen18).
Only Oomen et al. (Reference Oomen, Ocke and Feskens17) found a similar effect between the different TFA sources. Indeed, in their prospective population-based study, 667 men of the Zutphen Elderly Study aged 64–84 years and free of CHD at baseline were studied. Dietary surveys were used to establish the participants' food consumption patterns(Reference Oomen, Ocke and Feskens17). The relative risks of CHD for an increase of 0·5 % in energy at baseline from TFA from different sources (R-TFA, manufactured 18 : 1 TFA and other manufactured TFA) was calculated. For a mean intake of 0·7 (sd 0·2) % of energy (about 1·22 g/d with TEI at 9·2 MJ) the relative risk was 1·17 (95 % CI 0·69, 1·98) and was similar to the others TFA(Reference Oomen, Ocke and Feskens17). It is important to note that during the follow-up, the food items have varied and the amount of TFA has decreased. For instance, after 10 years of follow-up, the mean contribution of TFA intake to TEI decreased from 1985 to 1990 and 1995 (4·3 (sd 2·2) v. 2·9 (sd 1·5) v. 1·9 (sd 0·6) %).
Impact of trans-fatty acids on markers of CHD
Effects of trans-fatty acids on HDL- and LDL-cholesterol
The story linking ruminant TFA intake or biomarkers of TFA intake with the serum lipid profile began about 15 years ago when Bolton-Smith et al. demonstrated that in between 4490 and 5170, respectively, for men and women, the total cholesterol varied significantly (P < 0·001) by fifths of the R-TFA for men (1·5 to 4·9 g R-TFA/d from the lowest to the highest fifths) and women (1·2 to 4·3 g R-TFA/d from the lowest to the highest fifths) (positive trend) and by fifths of I-TFA for men (1 to 8·8 g I-TFA/d from the lowest to the highest fifths) (negative trend) who participated in the Scottish Heart Health Study(Reference Bolton-Smith, Woodward and Fenton19).
Recently, Chardigny et al. (Reference Chardigny, Destaillats and Malpuech-Brugere20) compared the effects of an 8 weeks consumption of 5·4 % of TEI (about 11–12 g/d) as TFA from hydrogenated sources with the same amount of TFA from ruminant sources on lipoprotein metabolism in forty-six healthy men and women. This study was a monocentric, randomised, double blind, controlled, cross-over design. In women, but not in men, R-TFA beneficially increased HDL-cholesterol concentrations. However, an increase in LDL-cholesterol and TAG concentrations was also observed. For both sexes, the serum total cholesterol:HDL-cholesterol ratio did not change. In a second study published at the same time by Motard-Bélanger et al. (Reference Motard-Bélanger, Charest and Grenier21), the effects of TFA from ruminant sources were studied in a double-blind, randomised, cross-over controlled study according to a Latin square design, in which each man was assigned to four different isoenergetic diets lasting 4 weeks each. In this study, diets were high in R-TFA (10·2 g/2500 kcal (10 460 kJ); 3·7 % of TEI), moderate in R-TFA (4·2 g/2500 kcal; 1·5 % of TEI), high in I-TFA (10·2 g/2500 kcal), or low in TFA from any source (2·2 g/2500 kcal) (control diet). It was found that at high intakes (3·7 % of TEI), TFA from both ruminant and hydrogenated sources adversely affected cholesterol metabolism. At intakes of 1·5 % of TEI, no effects of TFA from ruminant sources on the serum lipoprotein profile were observed.
Relationship between trans-fatty acids and markers of haemostatic function
The story of the relationship between TFA intake and cardiovascular risk factors continues with the study of the effects of TFA on markers for platelet aggregation, coagulation and fibrinolysis, three main determinants of haemostatic function.
Tholstrup et al. (Reference Tholstrup, Raff and Basu22) demonstrated no adverse effect of a daily intake of 3·5 g (about 1 % of TEI) of vaccenic acid (18 : 1Δ11t) from enriched butter during 5 weeks on haemostatic risk markers (plasminogen activator inhibitor 1 and factor FVII coagulant activity) of forty-two healthy young men, compared with a control butter that provided 0·3 g vaccenic acid per d (18 : 1Δ11t)(Reference Tholstrup, Raff and Basu22). However, the authors concluded that these effects may have been partly attributable to the higher MUFA and lower SFA content of the high-vaccenic acid (18 : 1Δ11t) diet, rather than to the effects of vaccenic acid (18 : 1Δ11t) itself.
The consumption of a diet rich in vaccenic acid (18 : 1Δ11t) (about 3·5 g/d; about 1 % TEI) for 5 weeks did not adversely affect blood pressure or isobaric arterial elasticity when measured in healthy young men by a volume-oscillometric method(Reference Raff, Tholstrup and Sejrsen23). These authors compared the effect of a diet based on milk fat high in vaccenic acid (18 : 1Δ11t), and not vaccenic acid (18 : 1Δ11t) per se.
Relationship between trans-fatty acids and markers of inflammation and endothelial dysfunction
Markers of inflammation have been implicated as risk factors for several degenerative diseases. With respect to CVD, the data collected from cross-sectional and prospective studies provide evidence that some markers are independent risk factors, and several of them can be modifiable throughout lifestyle, by physical exercise, smoking and also diet.
No observational studies have reported a correlation between a high content of R-TFA and biomarkers of inflammation in healthy volunteers, and only a few interventional studies have followed the effect of the consumption of R-TFA on inflammatory parameters.
Kuhnt et al. (Reference Kuhnt, Kraft and Vogelsang24) designed a study to investigate the effects of a 6-week dietary supplementation of 3 g of vaccenic acid (18 : 1Δ11t) and 3 g of 18 : 1Δ12t and endogenous conjugated linoleic acid synthesis on several biomarkers (IL-6 and IL-8, TNF-α, C-reactive protein, inter-cellular adhesion molecule-1, leptin and adiponectin) in young men and women. These markers belong to the systemic inflammation which has been reported as an independent risk factor for heart disease(Reference Libby25). No significant differences in the plasma concentrations of any biomarker were observed during the 18 : 1Δ11t and 18 : 1Δ12t intervention; these results were in agreement with the results of Thosltrup et al. (Reference Tholstrup, Raff and Basu22).
In healthy men, Motard-Belanger et al. (Reference Motard-Bélanger, Charest and Grenier21) found that the circulating level of C-reactive protein was not statistically different at the end of 4 weeks of experimental isoenergetic diets high in ruminant or industrial TFA (10·2 g/2500 kcal (10 460 kJ)). In our study published in 2008, we found no effect of consumption for 4 weeks of food items containing TFA (11–12 g/d) from ruminants and from industrial sources on the plasma high-sensitivity C-reactive protein concentration of forty-six healthy subjects(Reference Tardy, Lambert-Porcheron and Malpuech-Brugere26).
These different results on the effects of the intake of R-TFA on the development of cardiovascular pathologies are still contradictory, even though they tend to underline isomer-specific effects. Weggemans et al. (Reference Weggemans, Rudrum and Trautwein27) concluded that epidemiological studies from Europe and the USA show that up to 2·5 g/d, the effects of total, ruminant or industrial TFA are similar. An intake of more than 3 g/d of total and industrial TFA is associated with increased CHD risk. However, there are insufficient data on the health effects of ruminant TFA intakes over 2·5 g/d to allow any comparison in this range. They therefore advised not to discriminate between these sources in legislation or dietary recommendations. However, recently, Jakobsen et al. (Reference Jakobsen, Overvad and Dyerberg16), using data from the four Danish cohort studies, found no evidence of a higher risk associated with R-TFA intake within the wide range of intake among both women (90 % central range: 0·5–3·1 g/d) and men (90 % central range: 0·6–4·1 g/d). The authors concluded that a high intake of TFA from dairy and ruminant meat products is an issue of no concern to public health.
Trans-fatty acids and risks for type 2 diabetes
Type 2 diabetes is characterised by an increased blood glucose level due to impaired insulin sensitivity by peripheral tissues and decreased insulin secretion by pancreatic β-cells.
Effects of trans-fatty acids on glucose tolerance and type 2 diabetes
An important body of evidence relating total TFA intake to risk of type 2 diabetes came from the Nurses' Health Study that prospectively followed, during 14 years, 84 204 healthy women aged between 34 and 59 years(Reference Salmeron, Hu and Manson28). The data suggested that total lipid intake, compared with equivalent energy intake from carbohydrates, was not associated with the risk of type 2 diabetes in women. By contrast, the study reported that high intakes of TFA (3 (sd 0·5) % of total daily energy intake, TEI) correlated positively with the risk of diabetes in age- and BMI-adjusted analyses. In addition, the authors estimated that replacing 2 % of energy from TFA isoenergetically with PUFA would lead to a 40 % lower risk of developing type 2 diabetes. Subgroup analyses revealed that effects were primarily observed in obese women, possibly due to the fact that these women were already more insulin resistant than non-obese women(Reference Salmeron, Hu and Manson28). In addition, the Nurses' Health Study carried certain methodological limitations that may have affected the interpretation of TFA data. Notably, averaged food composition changed substantially over the 20-year experimental follow-up, in particular with regards to trans-fat content. In addition, foods rich in TFA were also rich in refined carbohydrates. Finally, and most tellingly, the techniques used to assess trans-fat content in foods have also improved(Reference Clandinin and Wilke29). In the epidemiological studies, the sources of TFA were not examined, so no conclusions could be brought regarding the respective effects of dairy and hydrogenated fats. In fact, differences in the relative abundance of each TFA isomer in food products could probably lead to different metabolic responses.
Short-term intervention studies in clinical or in animal models have not been very conclusive either. More recently, new data were obtained on the effects of TFA of dairy origin since it is now possible to produce experimental dairy vaccenic acid-enriched fat. A 5-week nutritional double-blind, randomised, parallel intervention study was performed in forty-two healthy young men stratified according to BMI (kg/m2). The volunteers received either a diet high in vaccenic acid (3·5 g/d; about 1 % of TEI; 18 : 1Δ11t) or a control diet providing 0·3 g vaccenic acid per d (18 : 1Δ11t). The authors did not find any effects of vaccenic acid-rich dairy fat on fasting insulin and glucose(Reference Tholstrup, Raff and Basu22).
In rat models, insulin and glucose responses to an intra-peritoneal injection of glucose were not modified after 8 weeks of experimental diets enriched in TFA of either dairy or industrial origin at 4·1 % of total TEI(Reference Tardy, Giraudet and Rousset30). These data were reinforced by recent findings by Wang et al., given that, after 4 weeks of a control diet supplemented with 1·5 % (w/w) vaccenic acid (18 : 1Δ11t), insulin and glucose metabolism of both lean and obese rats was not affected in response to a meal tolerance test(Reference Wang, Lu and Ruth31).
Effects of trans-fatty acids on insulin secretion by pancreatic β-cells
Long-term exposure of β-cells to high levels of fatty acids causes enhanced insulin secretion at low glucose (basal insulin release), while glucose-stimulated insulin secretion is decreased or unchanged. In isolated mouse islets, an acute incubation with the trans isomers trans-vaccenic (18 : 1Δ11t) and elaidic (18 : 1Δ9t) acids elicited a higher maximal insulin output than the cis isomer oleic acid(Reference Alstrup, Gregersen and Jensen32). The same authors compared in cultured INS-1 cells the effect of 3 d incubation with cis- (cis-18 : 1-11) and trans-vaccenic acid (trans-18 : 1-11), as well as oleic (cis-18 : 1-9) and trans-elaidic acid (trans-18 : 1-9), on basal and glucose-stimulated insulin secretion. All fatty acids tested increased basal insulin release; however, a significantly lower basal insulin release was demonstrated for cells cultured with 0·3 to 0·4 mm-trans-vaccenic acid compared with equimolar levels of the cis isomer. Glucose-stimulated insulin secretion was not changed by cis- or trans-vaccenic acid or by oleic acid, whereas it was stimulated by 0·3 to 0·4 mm-trans-elaidic acid(Reference Alstrup, Brock and Hermansen33). Despite these in vitro results, there are no available data on the impact of TFA of dairy origin on insulin secretion in vivo.
Effects of trans-fatty acids on peripheral insulin sensitivity
Little is known on the effects of TFA of both origins on insulin sensitivity in humans. An important step towards glucose intolerance and type 2 diabetes is peripheral resistance to insulin action.
A 4-week nutritional double-blind, randomised, parallel intervention study was performed in sixty-three healthy women with abdominal obesity (waist circumference>88 cm and BMI>28 kg/m2). After a run-in period, the volunteers were randomly assigned to consume for 4 weeks one of the three following diets: 60 g low-TFA lipids/d (0·54 g/d; n 21), R-TFA-rich lipids (4·86 g/d; n 21), or I-TFA-rich lipids (5·58 g/d; n 21). Changes in peripheral insulin sensitivity were assessed by using hyperinsulinaemic–euglycaemic clamps. The results showed that TFA from dairy and industrial sources ingested at a nutritional level (2 % of total TEI) for 4 weeks do not further alter peripheral insulin sensitivity in overweight and obese women with impaired insulin sensitivity(Reference Tardy, Lambert-Porcheron and Malpuech-Brugere26). However, the principle of precaution required that these results are not extended to long-term effects of TFA.
To go further in the comprehension of the mechanisms, Tardy et al. established in C2C12 myotubes that insulin-stimulated Akt phosphorylation was similar to the control group (without fatty acid) after incubation with oleic, elaidic and vaccenic acids(Reference Tardy, Giraudet and Rousset30), suggesting no direct effect of TFA on muscle insulin pathways. Those results were reinforced by a recent study of Hommelberg et al. (Reference Hommelberg, Langen and Schols34) which demonstrated in cultured murine skeletal muscle cells that insulin-stimulated glucose uptake and GLUT4 translocation were similar to those in the control group after incubation with vaccenic and elaidic acids.
In addition, Tardy et al. also demonstrated that oxidative (soleus) and glycolytic (tibialis anterior) muscle mitochondria have the same capacity to oxidise cis-fatty acids and TFA (elaidic and vaccenic acids), suggesting that the geometrical MUFA configuration does not heavily influence their oxidation rate. Thus, any TFA-induced alteration in muscle metabolism is not due to differences in oxidation rates(Reference Tardy, Giraudet and Rousset30).
To conclude, there is a body of evidence showing an absence of TFA effects on skeletal muscle insulin sensitivity. However, additional studies are required, especially on adipose tissue, to further analyse metabolic responses to TFA, including bioenergetic, lipid and carbohydrate pathways.
Conclusion
While TFA from hydrogenated oils have been clearly implicated in the pathogenesis of CVD, those from dairy products may be less deleterious at a level lower than 2 % of total TEI. Furthermore, dietary intakes of TFA from dairy origin are far from being high enough to constitute a serious threat for public health. As concluded by Booker & Mann(Reference Booker and Mann35), the advice would be to remove TFA from industrial sources from the food supply. Therefore, regular moderate consumption of dairy fats could be tolerated with respect to cardiovascular risks. Regarding the role of TFA from both origins on the incidence of type 2 diabetes, studies carried out on the topic are still controversial. Combined results suggest that TFA of dairy (R-TFA) and industrial (I-TFA) origins may not impair glucose tolerance at a physiological dose and during a short-term period. Mechanistic and intervention studies showed no significant effect of TFA on glucose tolerance or insulin resistance, suggesting that these fatty acids may not increase the risk for diabetes. The synthesis of these studies does confirm the food safety authority recommendations that their intake should be limited to 2 % of total TEI. Additional data are required to better understand possible long-term effect of TFA, especially to examine the impact of TFA in individuals at risk for the metabolic syndrome and type 2 diabetes.
Acknowledgements
The present review was supported by the ‘ANR – Agence Nationale de la Recherche – The French National Research Agency’ under the ‘Programme National de Recherche en Alimentation et Nutrition Humaine’, project ‘ANR-05-PNRA-no. 5.E.24’.
All authors have equally contributed to the review; however, C. M.-B. and J.-M. C. wrote together the cardiovascular section and A.-L. T. and B. M. wrote together the diabetes section. A.-L. T. produced Table 1.
There are no conflicts of interest to declare.