Cardiovascular and cardiometabolic risk factors

There is unequivocal evidence that reduction of LDL-cholesterol (LDL-C) significantly reduces the incidence of myocardial infarction and death from cardiovascular causes, without adversely affecting the risk of death from all causes in primary and secondary prevention studies^{(Reference Ference, Ginsberg and Graham5)}. The European Atherosclerosis Society Consensus Panel reviewed evidence for the effects of high LDL-C on the development of CVD, including CHD and stroke and showed a clear linear causal relationship as illustrated in Fig. 1^{(Reference Ference, Ginsberg and Graham5)}. A consensus was reached that serum LDL-C increased the progression of atherosclerosis in a dose-dependent manner, with greater detriment arising from longer exposure of the vascular endothelium to LDL-C^{(Reference Ference, Ginsberg and Graham5)}. Evidence also clearly demonstrates that small dense LDL particles, which are more likely to move into the vascular intima, undergo oxidation and contribute to the atherosclerotic plaque are more atherogenic and confer a greater risk for CVD^{(Reference Mikhailidis, Elisaf and Rizzo6)}. In contrast, a low concentration of serum HDL-cholesterol (HDL-C) is related to an increased risk of CHD^{(Reference Assmann, Schulte and von Eckardstein7)}, is a key feature of the metabolic syndrome and is highly prevalent in type-2 diabetes and obesity⁽⁸⁾. HDL particles are involved in a process of reverse cholesterol transport, in which cholesterol is removed from tissues and organs and returned to the liver for metabolism^{(Reference Assmann, Schulte and von Eckardstein7)}. However, recent evidence has shown that increasing serum HDL-C, by use of drugs, may not result in the anticipated reduction in CVD risk, which is more closely related to the functionality, rather than the cholesterol content of HDL particles^{(Reference Asztalos, Tani and Schaefer9)}. However, the total cholesterol (TC):HDL-C ratio is considered a more sensitive and specific CHD risk predictor than individual cholesterol measures; at all ages in women and the only lipid predictor independently related to CHD in men 65–80 years old^{(Reference Assmann, Schulte and von Eckardstein7,Reference Castelli10)} .

Fig. 1. Log-linear association per unit change in LDL-cholesterol (LDL-C) and the risk of CVD as reported in meta-analyses of Mendelian randomisation studies, prospective epidemiologic cohort studies, and randomised trials. The increasingly steeper slope of the log-linear association with increasing length of follow-up time implies that LDL-C has both a causal and a cumulative effect on the risk of CVD. Taken from^{(Reference Ference, Ginsberg and Graham5)}.

Hypertension is the greatest contributor to death globally and a key CVD and metabolic risk factor that is modifiable by diet^{(Reference Mancia, Fagard, Narkiewicz and Redon11)}. While the importance of lowering salt intake to reduce blood pressure is well founded⁽¹²⁾, evidence for the impact of dietary fats on blood pressure and vascular function is lacking^{(Reference Vafeiadou, Weech and Sharma13)}. The health of the vasculature and endothelial function is important for CVD risk reduction and inextricably linked to blood pressure. Endothelial dysfunction occurs when the balance between endothelial injury and repair is disrupted. Circulating bone marrow-derived endothelial progenitor cells play an important role in preserving the structural and functional integrity of the endothelium by inducing neovascularisation at the site of vascular injury^{(Reference Asahara, Murohara and Sullivan14)}. Reduced endothelial progenitor cell number and function have been associated with CVD risk factors, including hypertension and hypercholesterolaemia, and their potential role as prognostic and/or diagnostic markers of CVD is of considerable value^{(Reference Asahara, Murohara and Sullivan14)}. Microparticles are small vesicles released from the surface of many cell types, including endothelial cells and platelets, during activation or apoptosis, which often occurs during endothelial injury. Microparticle numbers are elevated in individuals with CVD and associated risk factors^{(Reference Barteneva, Maltsev and Vorobjev15)}, and the addition of endothelial microparticle numbers to the Framingham risk score has been shown to improve its predictive power of future CVD events^{(Reference Nozaki, Sugiyama and Koga16)}. However, the importance of these novel vascular risk markers needs further confirmation.

Central obesity and insulin resistance are defining characteristics of the metabolic syndrome, the other two of which can include raised plasma TAG, reduced HDL-C concentrations and hypertension (Table 1)⁽⁸⁾. Those with the metabolic syndrome are estimated to have an increased risk of CVD and particularly type-2 diabetes with many shared metabolic risk factors, often presenting with relatively normal TC and LDL-C concentrations⁽⁸⁾. There is evidence to suggest that diet and lifestyle interventions may be more effective at preventing the development of the metabolic syndrome than pharmacological agents, and dietary fats may play a key role in this respect^{(Reference Borkman, Campbell and Chisholm17)}. Evidence for the impact of dietary fat on cardiovascular and cardiometabolic risk, with particular reference to SFA, will be reviewed and presented in an attempt to resolve the perceived inconsistencies and confusion.

Table 1. Definition of the metabolic syndrome according to the International Diabetes Federation⁽⁸⁾

SFA as a strategy to reduce CVD and cardiometabolic risk factors

SFA reduction has been the mainstay of dietary fat recommendations for CHD risk reduction for many decades. UK public health advice on SFA was officially introduced in 1983 in the National Advisory Committee for Nutrition Education report⁽¹⁸⁾, which recommended reducing SFA to no more than 10 % total energy. The Committee of Medical Aspects re-evaluated evidence in 1991 and 1994 and in these reports the advice to reduce SFA intake to no more than about 10 % total energy was based on evidence that increasing or decreasing the contribution of SFA to dietary energy is followed by a rise or fall in LDL-cholesterol and in the commensurate risk of CHD^(19,20) . Since the 1990s evidence for the effects of SFA on a range of health outcomes has increased considerably. This has been reviewed by numerous international organisations with most proposing similar recommendations to limit SFA. Currently, the Australian Government Department of Health and New Zealand Ministry of Health⁽²¹⁾ recommend SFA should contribute between 8 and 10 % energy; the Food and Agriculture Organization/World Health Organization⁽²²⁾, Nordic Council of Ministers⁽²³⁾ and US Dietary Guidelines Advisory Committee⁽²⁴⁾ recommend no more than 10 % energy as SFA and the European Food Safety Authority⁽²⁵⁾ recommend consuming as little as possible. All advise replacement of SFA with PUFA. In contrast, the French Food Safety Agency⁽²⁶⁾ recommended a total SFA intake of no more than 12 % energy, but specify a maximum intake of 8 % energy from specific SFA due to their atherogenic potential, namely lauric, myristic and palmitic acids. In 2015, a novel strategy for dietary advice was proposed by the Health Council of the Netherlands^{(Reference Kromhourt, Spaaij and de Goede27)} in which recommendations were designed around foods and dietary patterns rather than specific nutrients. In these recommendations, advice that related to SFA included: (i) replace butter, hard margarines, and cooking fats by soft margarines, liquid cooking fats, and vegetable oils; (ii) limit the consumption of red meat, particularly processed meat and (iii) a few portions of dairy produce daily, including milk or yoghurt. Evidence for SFA and health outcomes was recently reviewed by the Saturated Fats Working Group of the UK Scientific Advisory Committee on Nutrition. The report entitled Saturated Fats and Health was published on 1st August 2019 with recommendations that the dietary reference values for SFA remain unchanged at population average of no more than 10 % energy from SFA, with recommendations for SFA substitution with unsaturated fats⁽²⁸⁾.

Population intake data

Despite long standing dietary recommendations to limit SFA intake, very few populations comply with this advice. A study which included fatty acid intake data from forty countries throughout the world reported that only eleven met the SFA (<10 % energy) and twenty met the PUFA (6–11 % energy) recommendations. Furthermore, in eighteen of twenty-seven countries examined, more than 50 % of the population had SFA intakes >10 % energy, whereas in thirteen of twenty-seven countries, the majority of the population had PUFA intakes <6 %^{(Reference Harika, Eilander and Alssema29)}. The current SFA intake from the latest data from the UK National Diet and Nutrition Survey (years 7–8) supports these data, with the mean consumption of SFA above recommendations in all age groups with SFA intakes of 11·9, 12·5 and 14·3 % of total dietary energy in adults aged 19–64, 65–74 and 75+ years, respectively. The mean population intakes of different fatty acid classes and the UK Reference Nutrients Intakes are shown in Tables 2 and 3, respectively. The main contributors to SFA intake in adults of all ages were meat and meat products, milk and milk products, and cereals and cereal products (half from pizza, biscuits, buns, cakes, pastries, fruit pies and puddings) with each food group contributing between 20 and 27 % of total SFA intake. Fat spreads contributed 9, 13 and 16 % total dietary energy in those of 19–64, 65–74 and 75+ years, respectively. Interestingly, intakes of total SFA increased with household available income, although generally these differences were small.

Table 2. Mean daily intake of SFA, MUFA and PUFA (% total energy) intake for UK children and adults by age

%total eng, % total energy.

National Diet and Nutrition Survey RP survey years 7–8 (2014/15–2015/16) bases unweighted.

Table 3. UK Dietary Reference Nutrient Intakes for fats for adults as a percentage of total energy intake

LC, long chain. Taken from⁽¹⁹⁾.

Assessment of risk and quality of evidence

The quality of evidence is important to consider when assessing risk. A hierarchy of evidence as represented by a pyramid, is generally accepted, as shown in Fig. 2. Data from ecological studies, although helpful for hypothesis generation, are of limited quality and represents associations which are often linked with considerable potential confounding. Data from cohort studies, particularly longitudinal prospective cohort studies, can offer valuable insight into associations between dietary factors and key outcome measures, such as CVD mortality, but do not prove cause or effect. Furthermore, these studies are often associated with confounding including: dietary change over the follow-up period; reformulation of foods throughout the follow-up period (such as removal of trans fatty acids from the food chain which has occurred over the past decades); lifestyle factors including weight change, smoking status, amount of activity which are not fully accounted for; influence of other dietary components; no consideration of the replacing macronutrient or of the quality of macronutrient (i.e. wholegrain v. refined carbohydrates or n-3 PUFA v. n-6 PUFA) and reverse causality.

Fig. 2. Pyramid depicting hierarchy of evidence.

In contrast, evidence from randomised controlled trials (RCT) are considered to be of higher quality, with data demonstrating the effect of controlled dietary intervention, such as substitution of SFA with PUFA, on hard clinical outcomes (e.g. CVD morality) or validated risk markers (e.g. LDL-C). However, all studies investigating dietary fats can be limited by the sample size; duration of follow-up/intervention; study design; confounding by the presence of dietary trans fatty acids in some intervention foods (known to have a significant detrimental effect on CVD) in studies published before 1990s; and residual confounding. Systematic reviews and meta-analyses of particularly RCT, can offer high-quality data, which represents the totality of evidence available. However, there are potential limitations in meta-analyses, such as the quality of the individual studies, criteria for study inclusion, differences in study design, participant inclusion, type and methods of intervention, which can result in inability or inappropriate study comparison and inconsistent findings between meta-analyses addressing the same question. It is therefore apparent that the type of evidence is of paramount importance and wherever possible, rigorous, current and comprehensive systematic reviews and meta-analyses will be used in this review, although individual studies will also be included where appropriate.

Challenges to the SFA recommendations

As discussed above, there are consistent global dietary recommendations to limit SFA intake for disease risk reduction, which are based on rigorous assessment of the totality of evidence from RCT and prospective cohort studies, yet within the past 5 years the validity of SFA reduction has been questioned. This recent challenge to the SFA recommendations has been in response to a number of systematic reviews and meta-analyses which indicate that there is limited evidence for the significant effects of SFA reduction on CVD mortality^{(Reference Chowdhury, Warnakula and Kunutsor30–Reference Ramsden, Zamora and Leelarthaepin34)}. These data will be discussed in the context of the quality and relevance of evidence.

SFA and CVD risk

There is consistent evidence from systematic reviews and meta-analyses of RCT^{(Reference Hooper, Martin and Abdelhamid35,Reference Harcombe, DiNicolantonio and Grace36)} and prospective cohort studies^{(Reference Chowdhury, Warnakula and Kunutsor30,Reference Harcombe, Baker and Davies32,Reference Siri-Tarino, Sun and Hu33,Reference de Souza, Mente and Maroleanu37,Reference Schwab, Lauritzen and Tholstrup38)} for the lack of a significant relationship between SFA intake and CVD, CHD and stroke mortality, which has fuelled the recent challenges to SFA recommendations. However, a significant 17 % reduction in CVD events in those who reduced their SFA intake compared with usual diet (using a random-effects statistical model) was reported in the most comprehensive, up-to-date and rigorous systematic review and meta-analysis of RCT^{(Reference Hooper, Martin and Abdelhamid35)}. This analysis included eleven studies with 53 300 participants and 4377 CVD events and used the gold-standard Cochrane protocol for systematic review. Furthermore, a significant 7 or 8 % reduction was also observed after using two fixed-effect statistical models (Mantel–Haenszel and Peto, respectively), suggesting that reducing SFA intake to approximately 10 % energy significantly reduces CVD events by between 7 and 17 %^{(Reference Hooper, Martin and Abdelhamid35)}.

Moreover, Hooper found a significant 7–8 % reduction in CHD events when reduced intakes of SFA were compared with usual intakes after fixed effects analysis and a non-significant trend for a 13 % reduction after random-effects analysis (P = 0·07) using twelve RCT, that included 53 199 participants and 3307 cases. In contrast^{(Reference Chowdhury, Warnakula and Kunutsor30)}, Chowdhury and colleagues, in their high-profile systematic review and meta-analysis of twenty prospective cohort studies (including 283 963 participants and 10 518 CHD cases), concluded there was no association between SFA intake and CHD outcomes, when the top v. the bottom tertiles of SFA intakes were compared using a random-effects model. However, the authors also performed a fixed-effect statistical model and found a significant 4 % increased risk of CHD outcomes when higher v. lower saturated fat intakes were compared, although this finding was not commented upon in their paper. The reporting of both random and fixed effects models is becoming increasingly popular as recommended in the Cochrane Handbook for Systematic Reviews of Interventions (http://training.cochrane.org/handbook). However, within the scientific community there are inconsistencies in the application and relevance of these models to different datasets, with differences in the underlying assumptions and statistical considerations. Fixed-effect models give weight in direct proportion to the size of the primary studies, whereas random-effects models generally give similar weight to all studies, irrespective of size. Although random-effects models are used more commonly, fixed-effect models may offer a number of advantages over random-effects models, such as proportionate study weighting, and it would seem prudent to consider both models when reviewing evidence. The increase in CHD outcomes from higher SFA intake from prospective cohort studies^{(Reference Chowdhury, Warnakula and Kunutsor30)} supports the analysis of RCT using fixed-effects analysis^{(Reference Hooper, Martin and Abdelhamid35)}, and suggests reduction of dietary SFA would be of benefit.

Reducing SFA was found to have no effect on the mortality from stroke in a meta-analysis of RCT^{(Reference Hooper, Martin and Abdelhamid35)} and also on ischaemic strokes from the most comprehensive systematic review with meta-analysis of twelve prospective cohort studies with fifteen comparisons including 339 090 participants and 6226 ischaemic stroke deaths^{(Reference de Souza, Mente and Maroleanu37)}. In contrast, a systematic review and meta-analysis of fifteen prospective cohort studies (n 476 569 including 11 074 strokes) reported a significant 11 % reduced overall stroke risk and 25 % fatal stroke risk with higher SFA intake^{(Reference Cheng, Wang and Shao39)}. Interestingly, after subgroup analysis there was no association in non-East Asian populations, but a significant association in East Asian populations (21 % lower risk)^{(Reference Cheng, Wang and Shao39)}. In another meta-analysis of prospective cohort studies, a significant association was identified between lower SFA intake and higher intracerebral haemorrhagic strokes in Japanese populations only^{(Reference Muto40)}. These associations between higher SFA and reduced stroke seem to be isolated to East Asian populations living in East-Asia, who typically consume very low dietary SFA, have distinct differences in dietary patterns, other lifestyle factors and genetic background, in comparison with Western populations in Europe and America.

These studies provide vital evidence for the benefits of reducing intake of SFA on CVD and CHD risk, and to address the recent challenges to these recommendations. However, these studies are limited by the lack of consideration of which macronutrient replaced SFA in the diet, and could not distinguish between, or determine whether, there were any differential effects on CVD risk that were dependent on the substitute macronutrient. This is of paramount importance for the development of valid public health advice and guidance on practical strategies of SFA reduction and replacement.

Impact of the macronutrient replacement of SFA on CVD risk

Unlike pharmaceutical or supplemental studies, while a drug or supplement can be simply added to a participant's regimen and compared to a placebo, dietary interventions involving macronutrients require careful consideration in terms of the replacement macronutrient, particularly in an iso-energetic study design. This adds complexity to the implementation of the study, data analysis and interpretation of the results of a study. In reality, the intervention outcomes could be the result of reduction of one macronutrient, increase in the replacing macronutrient, or a combination of both.

SFA and cardiometabolic risk

Fats, cardiovascular and cardiometabolic risk markers

Dietary lipids

Dietary fats are key modulators of circulating lipids, with the reduction of serum LDL-C through SFA reduction and higher PUFA, particularly n-6 PUFA (linoleic acid) and shorter chain n-3 PUFA (α-linoleic acid), and the serum TAG-lowering effects of long chain n-3 PUFA from fish, fish oil or supplements, being central aspects of these dietary fat recommendations (Table 3).

The most comprehensive analysis investigating the impact of dietary fats, predominantly SFA and replacement macronutrient on serum lipoprotein concentrations was conducted by Mensink for the WHO and published in 2016^{(Reference Mensink47)}. Mensink initially performed a systematic review, which identified eighty-four relevant studies, 211 diet data points and 2353 participants (65 % men and 34 % women) who had a mean age 38 years (21–72 years), BMI 24·2 kg/m² (20·0–28·6 kg/m²), TC 5·1 mmol/l (3·7–6·7 mmol/l); LDL-C 3·3 mmol/l (2·3–4·8 mmol/l); HDL-C 1·2 mmol/l (0·9–1·8 mmol/l) and TAG 1·2 mmol/l (0·7–2·2 mmol/l). After performing a number of multiple regression analyses it was shown that reducing SFA and replacing with a mixture of cis-PUFA (predominantly linoleic acid and α-linolenic acid) or cis-MUFA (predominantly oleic acid) was more effective than replacing SFA with a mixture of carbohydrates on the lipoprotein profile (Table 4). More specifically it was estimated that serum TAG increased by a mean 0·0011 mmol/l for every 1 % energy SFA replacement with mixed carbohydrates, compared to a significant decrease in serum TAG of 0·004 mmol/l and 0·010 mmol/l for 1 % energy replacement by cis-MUFA and cis-PUFA respectively. Furthermore, replacement of 1 % energy from SFA with carbohydrate had no effect on the serum TC:HDL-C ratio compared to a significant reduction of 0·027 and 0·034 after substitution with cis-MUFA and cis-PUFA, respectively (Table 4). The results were consistent across a wide range of SFA intakes including less than 10 % of total energy, consistent for both men and women and not affected by baseline lipid concentrations or type of intervention. Further analysis showed that there were differential lipid responses according to the type of SFA. In comparison with a mixture of carbohydrates, an increased intake of lauric, myristic or palmitic acid raised serum TC, LDL-C and HDL-C and lowered TAG concentrations, while an increased intake of stearic acid had no significant effect on these or other serum lipid values. Lauric acid alone reduced the TC:HDL-C and LDL-C:HDL-C ratios compared with a mixture of carbohydrates^{(Reference Mensink47)}. These data are supported by metabolic ward studies, which provide high-quality data from carefully controlled study which involve provision of total dietary intake, with specific exchange of SFA for other macronutrients^{(Reference Clarke, Frost and Collins48)}.

Table 4. Estimated multiple regression equations for the mean changes in serum lipids when 1 % of dietary energy from SFA is isoenergetically replaced by carbohydrates (CHO), cis-MUFA or cis-PUFA

HDL-C, high-density lipoprotein-cholesterol; LDL-C, low-density lipoprotein-cholesterol; TC, total cholesterol.

* Number of diets/number of studies.

† The 95 % CI refer to the regression coefficients on the line directly above.

Adapted from^{(Reference Mensink47)}.