Introduction
CHD is a leading cause of premature mortality and disability-adjusted life years in Western society with 2 million deaths in the European Union and 4·3 million deaths in Europe(Reference Allender, Scarborough and Peto1). Within the European Union, the economic cost of CVD is substantial and has been calculated to be €192 billion annually in direct and indirect healthcare costs(Reference Allender, Scarborough and Peto1).
While the causes of CHD are multifactorial, many of these factors are related to lifestyle such as tobacco smoking, lack of physical activity, and dietary habits and, as such, are modifiable. Other modifiable risk factors include elevated blood pressure, type 2 diabetes and dyslipidaemias(Reference Reiner, Catapano and De Backer2). The importance of raised blood cholesterol as a modifiable risk factor for CHD is well established(3). The WHO estimates that over 60 % of CHD and 40 % of ischaemic stoke in developed countries is due to total blood cholesterol levels in excess of the theoretical minimum 3·8 mmol/l(3). For some time, public health campaigns such as the National Cholesterol Education Program and Joint British Societies' guidelines have emphasised reducing elevated LDL-cholesterol to lower CHD risk(4, 5). Furthermore, it has been suggested that every 1·0 mmol/l (39 mg/dl) reduction in LDL-cholesterol is associated with a 22 % reduction in the risk of CVD mortality and morbidity. Consequently, the reduction of elevated LDL-cholesterol is a significant public health goal.
A central recommendation in guidance given to many Western populations is the need to reduce fat intake, particularly saturated fat and trans-fat(6). The conventional approach to SFA reduction has been to exclude food high in saturates or identify low-fat alternatives such as low-fat dairy products, meat products, etc. While this approach can be effective, recent advice suggests that, ideally, SFA reduction should also be associated with improvements in fat quality achieved by substituting SFA with unsaturated fat(Reference Stanner7, Reference Gillingham, Harris-Janz and Jones8), with some suggesting that the preferred source of these fats is MUFA(Reference Gillingham, Harris-Janz and Jones8). Furthermore, in a recently produced unified theory based on evolutionary, historical, global and modern perspectives relating dietary fat quality and CHD prevention, it was identified that reduction or elimination of trans-fatty acids and replacement of SFA with MUFA may provide cardiovascular benefits(Reference Ramsden, Faurot and Carrera-Bastos9). Moreover, replacement of medium-chain n-6 PUFA (for example, linoleic acid), moderate consumption of medium-chain n-3 PUFA (for example, α-linolenic acid; ALA) and increased consumption of long-chain n-3 PUFA (for example, EPA and DHA) may provide beneficial cardiometabolic effects and are unlikely to do any harm(Reference Ramsden, Faurot and Carrera-Bastos9).
Decreasing SFA intake conventionally was the first therapeutic option recommended to individuals to lower blood cholesterol, however, of increasing importance is the use of effective combinations of cholesterol-lowering dietary components, such as β-glucans, soya protein and plant stanols and sterols as an approach to managing blood cholesterol (National Cholesterol Education Program Panel III guidelines)(4). A dual approach, based on modification of both improved fat quality and use of cholesterol-lowering food components, maximises the beneficial effect on blood lipids, thereby reducing the risk of heart disease and also other metabolic conditions(Reference Stanner7).
A number of foods, many of them the so-called functional foods or nutriceuticals, have been identified that modify blood lipids. These foods have traditionally been considered as a single addition to a lipid-modification diet. A more sophisticated approach to combining foods with enhanced cholesterol-lowering properties is that adopted in the ‘portfolio diet’(Reference Jenkins, Kendall and Axelsen10). This terminology, adopted by Jenkins over a decade ago, refers to a low-fat vegan or vegetarian diet enhanced by the inclusion of plant stanols or sterols, soya protein, viscous fibre and tree nuts(Reference Jenkins, Kendall and Faulkner11). Latterly the ‘portfolio diet’ has been developed by the addition of MUFA(Reference Jenkins, Chiavaroli and Wong12).
It is suggested that as important as either looking at these components in isolation, or as sophisticated combinations, is the role that simple combinations of cholesterol-lowering foods can play in meeting the dual objective of improving the fat quality of the diet and blood cholesterol lowering. In order to establish baseline values for the key food and food components that can modify blood cholesterol and to understand the degree to which they are additive, a search of the literature was conducted to identify the most up-to-date information relating to their individual and combined potential. Information relating to each food or component was extracted and converted into a common format for subsequent tabulation of the data.
Methods
A search of the scientific literature (Medline, December 2011) was used to identify papers, specifically randomised controlled trials (RCT) and meta-analyses, conducted where the effect of a food/food component intervention on blood lipids was assessed. Search terms were ‘cholesterol’ or ‘blood lipids’ and ‘the food name’ and ‘meta’. The designated ‘food names’ included beta-glucan(s), soy(a) protein, nuts, plant stanol and sterol. In addition, the search terms ‘cholesterol’ or ‘blood lipids’ and ‘portfolio’ were used to identify studies conducted using a portfolio approach. These searches were complemented by a hand search of key reviews and meta-analyses. The key findings from identified papers were extracted and the information converted into a common form. For example, all data were converted to mmol/l from mg/dl by use of the factor 0·02 586 for total cholesterol, LDL- and HDL-cholesterol and by use of the factor 0·01 125 for TAG. In order to assess the relative effects on fat quality following dietary modification (in g) and LDL-cholesterol reduction (as a percentage reduction compared with control treatment) as detailed in Tables 2 and 5, baseline LDL-cholesterol concentration has been assumed to be 3·75 mmol/l, if no specific value is given.
Results and discussion
The focus of the present review is on the reduction of total and LDL-cholesterol, as this appears to be where dietary manipulations are most effective. However, this approach is not intended to undermine the valuable contribution that can be made by the maintenance of or an increase in HDL-cholesterol. It has been argued, particularly for those at risk for CHD, that the total cholesterol:HDL-cholesterol ratio is a better assessment of risk than single lipid components(5). Also important to coronary risk is elevated TAG, particularly in diabetics and other sub-sectors of the population, where a combination of measures may be appropriate(Reference Miller, Stone and Ballantyne13–Reference Chapman, Ginsberg and Amarenco15). Reference to the effect of food and food components on all main lipid components will be made where possible.
Macronutrient changes that reduce blood lipids
There are many ways to predict the effect of macronutrient intake on blood lipid changes and CHD risk; see, for example, Givens(Reference Givens16), who used five models based on prediction equations from the original proposed by Keys(Reference Keys17). Traditionally, SFA reduction has been the key tactic employed to reduce blood cholesterol levels to lower the risk of both CVD and diabetes. Generally, LDL-cholesterol increases proportionally to the amount of SFA in the diet. However, advice has been less conclusive with regards to the best replacement for SFA in the diet, particularly as a desirable increase in HDL-cholesterol was reported in a meta-analysis of sixty studies that evaluated the effect of replacing carbohydrate with SFA(Reference Mensink, Zock and Kester18). However, the replacement of large amounts of dietary SFA with carbohydrate should be avoided, as fasting TAG concentrations may increase and dyslipidaemia associated with the insulin resistance syndrome may be induced, which would be counterproductive, particularly in diabetics(Reference Minihane and Harland19).
A role has been identified for the enhanced intake of either cis-MUFA or PUFA as the most effective replacements of dietary SFA to reduce CHD risk(Reference Mensink, Zock and Kester18). However, the authors also reported that the effect on LDL-cholesterol may be markedly different to the effect on the total:HDL-cholesterol ratio, which they indicated was a better indicator of CVD risk, a view shared by the Joint British Societies in their recommendations for CVD assessment(5). It was calculated that the replacement of 1 % carbohydrate energy with cis-MUFA or PUFA improved total:HDL-cholesterol ratio by about 0·03 mmol/l (P < 0·001), whereas LDL-cholesterol was reduced by about 0·1 mmol/l (P < 0·01) when replaced by MUFA and about 0·2 mmol/l (P < 0·001) when replaced by PUFA. The same replacement with SFA increased LDL-cholesterol by about 0·03 mmol/l (P < 0·001).
Oils with the highest MUFA content most frequently consumed in Western society are olive and rapeseed. The predicted change in total cholesterol:HDL-cholesterol ratio when ‘mixed fat’ that constituted 10 % of energy of the average US diet was replaced isoenergetically with either olive or rapeseed oil has been calculated and the improvement was about 0·1 and 0·12 mmol/l, respectively(Reference Mensink, Zock and Kester18). By reference to a meta-analysis of nine studies, where rapeseed oil was either added to, or substituted for SFA in the diet, it was calculated that each 1 g of SFA intake reduction was associated with reductions in both LDL- and total cholesterol of 0·03 mmol/l, while each 1 g increase in either MUFA or PUFA reduced total cholesterol by 0·05 and 0·09 mmol/l, respectively, and LDL-cholesterol by 0·04 and 0·08 mmol/l, respectively(Reference Harland20). It was also calculated that each 10 g increase in MUFA or rapeseed oil intake was associated with a reduction in total and LDL-cholesterol of 9·8 and 5·8 %, respectively.
There are also a number of epidemiological studies that demonstrate a reduction in CHD risk with moderate- to high-fat diets rich in MUFA; see, for example, the Seven Countries Study(Reference Keys, Menotti and Karvonen21), the Nurses' Health Study(Reference Hu, Stampfer and Manson22) and the large-scale intervention studies OmniHeart and KANWU(Reference Swain, McCarron and Hamilton23, Reference Vessby, Unsitupa and Hermansen24).
A further consideration is that while PUFA, especially long-chain n-3 PUFA, reduce CHD risk, there is concern that consumption of n-6 PUFA, particularly linoleic acid, is at an all-time high in Western society(Reference Ramsden, Faurot and Carrera-Bastos9). Furthermore, it has been suggested that the amount of n-6 PUFA should not increase further(25), and the ratio of n-6:n-3 PUFA should reduce(Reference Simopoulos26).
Foods or components that reduce blood lipids
The ‘portfolio diet’ is based on increasing intake of plant stanols and sterols, soluble (viscous) fibre, soya protein and tree nuts to enhance the cholesterol-reducing effect of a plant-based diet. These dietary components are those for which most evidence exists and for which health claim(s) have been permitted in one or more countries. In addition, there are a number of foods where blood lipid modification evidence is accumulating and that may have potential as cholesterol-lowering agents; these include dark chocolate, green tea, flaxseed and garlic. As evidence relating to these various foodstuffs is limited, their role has not been assessed in detail. However, as further clinical research becomes available, they, too, can be incorporated into a cholesterol-lowering regimen. Neither does this list of foods claim to be exhaustive, but provides an insight into the evidence that currently exists for the main cholesterol-lowering food components. Finally, there are numerous foods for which a single human study or only animal data are available; while these foods may offer potential for the future, further evidence is needed to properly assess the contribution that can be made by these foods.
Plant stanols and sterols
There is an extensive literature that has evaluated the effects of plant stanols and sterols on blood lipids.
Plant sterols and stanols are naturally occurring compounds structurally similar to cholesterol with a cellular function similar to human cholesterol. Plant sterols have a higher degree of absorption than plant stanols. There are in excess of 200 clinical studies that have evaluated the effects of these components on blood lipids, primarily in adults with normal or elevated plasma cholesterol at baseline. In 2009, the European Food Safety Authority (EFSA) assessed the scientific evidence available up to this date and a health claim was granted in Europe. The following statement was devised to be reflective of the evidence: ‘Plant stanols/sterols have been shown to lower/reduce blood cholesterol’. On the basis of evidence supplied, it was estimated that 1·5–1·9 g plant sterols per d and 2·0–2·4 g plant stanols per d reduce LDL-cholesterol by an average of 8·5 and 8·9 %, respectively(27). Typically for an intake of 1·5–2·4 g plant sterols or stanols per d, a reduction in LDL-cholesterol of between 7 and 10·5 % can be expected. A similar health claim also exists in the USA(28). Evidence from two early analyses(Reference Katan, Grundy and Jones29, Reference Law30) and two more recent meta-analyses(Reference Demonty, Ras and van der Knaap31, Reference Abumweis, Barake and Jones32) were cited in support of the claim. In brief, the more recent analyses involved pooling of data from a large number of RCT, either eighty-four trials including 141 trial arms (6805 subjects)(Reference Demonty, Ras and van der Knaap31), or fifty-nine RCT (n >4500 subjects)(Reference Abumweis, Barake and Jones32). A non-linear equation comprising two parameters (the maximal LDL-cholesterol lowering and an incremental dose step) was used to describe the dose-response curve, and the overall pooled absolute (mmol/l) and relative (%) LDL-cholesterol-lowering effects of phytosterols were also assessed with a random-effects model(Reference Demonty, Ras and van der Knaap31). The pooled LDL-cholesterol reduction was 0·34 (95 % CI − 0·36, − 0·31) mmol/l, or 8·8 % for a mean daily dose of 2·15 g phytosterols. The impacts of subject baseline characteristics, food formats, type of phytosterols and study quality on the continuous dose-response curve were determined by regression or subgroup analyses. Higher baseline LDL-cholesterol concentrations resulted in greater absolute LDL-cholesterol reductions. No significant differences were found between dose-response curves established for plant sterols v. stanols, fat-based v. non-fat-based food formats and dairy v. non-dairy foods. A larger effect was observed with solid foods than with liquid foods only at high phytosterol doses (>2 g/d). There was a strong tendency (P =0·054) towards a slightly lower efficacy of single v. multiple daily intakes of phytosterols(Reference Demonty, Ras and van der Knaap31). The results of this analysis are not dissimilar to the second meta-analysis in which the weighted mean net difference in LDL-cholesterol was − 0·31 (95 % CI − 0·35, − 0·27) mmol/l (P < 0·0001) compared with control(Reference Abumweis, Barake and Jones32). Reductions in LDL-cholesterol levels were reported to be greater in individuals with high baseline LDL-cholesterol and when plant sterols were incorporated into fat spreads, mayonnaise, salad dressing, milk and yoghurt compared with other food products such as croissants and muffins, orange juice, non-fat beverages, cereal bars and chocolate. There has been significant product development in this area, with a wide range of products, particularly yoghurts, yoghurt-type or breakfast drinks and spreads now containing a substantial quantity of plant sterols and stanols, enabling the regular achievement of the target daily amount identified.
A further meta-analysis identified seventy-six studies, of which twenty were considered of sufficient quality for data pooling, and provides data on TAG, in addition to total and LDL-cholesterol(Reference Wu, Fu and Yang33). The results of this analysis indicated that volunteers consuming foods enriched with 2·0 g plant stanols or sterols could significantly decrease LDL- and total cholesterol and TAG compared with control groups; mean differences were − 0·35 (95 % CI − 0·47, − 0·22) mmol/l (P < 0·00 001), − 0·36 (95 % CI − 0·46, − 0·26) mmol/l (P < 0·00 001) and − 0·1 (95 % CI − 0·16, − 0·03) mmol/l (P = 0·004), respectively.
Recently, two further analyses have been performed that compare the effect of plant sterols v. plant stanols on serum lipid levels in healthy patients and patients with hypercholesterolaemia. In the most recent systematic review and meta-analysis, 113 publications and one unpublished study report containing 182 treatment arms met the pre-defined inclusion and exclusion criteria and were included in the assessment(Reference Musa-Veloso, Poon and Elliot34). The maximal LDL-cholesterol reductions for plant stanols (16·4 %) and plant stanol ester (17·1 %) were significantly greater than the maximal LDL-cholesterol reductions for plant sterols (8·3 %) and plant sterol ester (8·4 %)(Reference Musa-Veloso, Poon and Elliot34); however, intake was far in excess of usual target intake levels of 1·5 to < 2·5 g/d, at which level of intake the extent of LDL-cholesterol reduction is comparable(27). An analysis specifically designed to summarise direct comparisons between the effect on plant sterols and stanols on serum lipid levels demonstrated that in fourteen studies with 531 subjects, there is no statistically or clinically significant difference between plant sterols and plant stanols in their abilities to modify total cholesterol (weighted mean difference (WMD), − 0·03 (95 % CI − 0·11, 0·05) mmol/l (P = 0·47); LDL-cholesterol WMD, − 0·010 (95 % CI − 0·08, 0·06) mmol/l (NS); HDL-cholesterol WMD, 0·001 (95 % CI − 0·03, 0·02) mmol/l (NS); TAG WMD, 0·02 (95 % CI − 0·08, 0·04) mmol/l (NS))(Reference Talati, Sobieraj and Makanji35).
Looking at the database of RCT and meta-analyses as a whole, there is consistent evidence indicating that both plant stanols and sterols reduce total and LDL-cholesterol by a clinically and statistically significant amount. However, whether there is a meaningful difference in the extent of the cholesterol reduction when plant stanols and sterols are included in the diet at the recommended dose and in individuals with similar baseline cholesterol and in the same food matrix is questionable.
Soya protein
The literature relating to soya protein is extensive, with over 100 RCT published and a plethora of meta-analyses. A total of eleven meta-analyses conducted with RCT that relate soya protein with or without soya isoflavones to a measure of blood lipids have been identified(Reference Anderson, Johnstone and Cook-Newell36–Reference Anderson and Bush46), as well as a review and nomogram(Reference Sirtori, Eberini and Arnoldi47). These meta-analyses have recently been reviewed and it was concluded that despite employing widely varying inclusion criteria, all meta-analyses published to date consistently report a reduction in total and LDL-cholesterol. The main area of difference appears to be the extent of the reported reduction in LDL-cholesterol, which was typically reported to be about 0·22 mmol/l and equivalent to a reduction of 4·5–7·0 %. HDL-cholesterol is typically increased by 0·01–0·04 mmol/l and TAG reduced by 0·10 mmol/l. The assessment of the effect of soya protein on blood lipids from studies conducted in the 21st century was achieved by pooling data from eligible RCT conducted with about 25 g soya protein; the standard mean difference in LDL-cholesterol in this subset of twenty-seven studies was − 0·22 (95 % CI − 0·28, − 0·15) mmol/l, which was highly statistically significant (P < 0·0001) and equivalent to a reduction compared with baseline of 5·5 %.
A novel approach to pooling data was adopted in the analysis of Jenkins et al. (Reference Jenkins, Mirrahimi and Srichaikul48). These authors determined the ‘intrinsic’ effect of soya protein (that related to the soya protein per se) on LDL-cholesterol, separately from the ‘extrinsic’ (displacement) potential of soya in replacing SFA and other components in the diet associated with elevated blood lipids. The intrinsic effect of soya to lower plasma cholesterol was derived from a meta-analysis of eleven RCT where macronutrient profiles of both soya protein and protein control diets were balanced. The results from this meta-analysis indicated a mean LDL-cholesterol reduction of 0·17 mmol/l (n 22; P < 0·0001) or 4·3 % for soya, which could be attributable to the soya protein itself and not due to any other dietary factor. They then further explored the proposed ‘extrinsic’ effect of soya protein in displacing foods higher in SFA and cholesterol using the standard predictive equations for LDL-cholesterol, based on the isoenergetic substitution of a range of 13–58 g soya protein-containing foods per d for animal protein foods, for example, cows' milk, yoghurt and meat. The estimated displacement value of soya (intake range, 13–58 g/d) using typical American diets (50th percentile Third National Health and Nutrition Examination Survey (NHANES III) intake data) was a 3·6–6·0 % reduction in LDL-cholesterol, due to displacement of SFA and cholesterol from animal foods. Thus, the total LDL-cholesterol reduction attributable to the combined intrinsic and extrinsic effects of soya protein foods therefore ranged from 7·9 to 10·3 %(Reference Jenkins, Mirrahimi and Srichaikul48).
The extent to which soya protein can reduce total and LDL-cholesterol has been evaluated largely in subjects who are healthy or with modest hypercholesterolaemia, although some of the meta-analyses include subjects with familial heart disease, diabetes and the metabolic syndrome(Reference Anderson, Johnstone and Cook-Newell36, Reference Balk, Chung and Chew40). A further important finding from the meta-analyses is that no dose-response relationship has been identified in the range of 15–40 g soya protein intake and any of the changes in blood lipids, indicating that lower amounts of soya protein intake can be effective(Reference Harland and Haffner43).
Individual studies have compared the efficacy of different sources of soya protein, an aspect further investigated by sub-analyses conducted in recent meta-analyses(Reference Jenkins, Mirrahimi and Srichaikul45, Reference Anderson and Bush46). In a well-controlled study conducted with 25 g soya protein that was presented as soya milk, prepared from either whole bean or isolated soya protein and compared with cows' milk, no difference was reported in the extent of LDL reduction between the two soya milks, and both resulted in LDL-cholesterol levels that were significantly lower than with cows' milk (P = 0·02)(Reference Gardner, Messina and Kiazand49). In the recent Anderson & Bush meta-analysis, a sub-analysis of forty-seven studies evaluated the effects of food form of the soya protein on LDL-cholesterol lowering. It was concluded that different food forms of soya protein did not have significantly differing effects on net reductions of LDL-cholesterol when isolated soya protein was compared with soya protein presented as either soya milk or yoghurt or other food forms(Reference Anderson and Bush46).
At the present time, health claims exist in both Europe and the USA and are based upon an intake of 25 g soya protein being associated with lowered blood cholesterol(50, 51). The claim can be made on foods containing a minimum content of soya protein, which is 5 g in Europe and 6·25 g in the USA. There are a wide range of food that supply >5 g soya protein per serving; these include soya milk, soya yoghurts, tofu, edamame and meat replacer products such as burgers, meatballs, sausages and mince.
Soluble fibre
An interest in the role that soluble fibre could play in reducing blood cholesterol stems from research carried out in the 1970 s and 1980 s. At this time, oats, pectin, guar gum and psyllium were considered to be the main sources of soluble fibre. A meta-analysis of sixty-seven RCT incorporating these fibres in the diet, at a range of inclusion of 2–10 g soluble fibre/d, indicated a reduction in total or LDL-cholesterol of − 0·045 (95 % CI − 0·054, − 0·035) mmol/l per g soluble fibre and − 0·057 (95 % CI − 0·070, − 0·044) mmol/l per g soluble fibre, respectively(Reference Brown, Rosner and Willett52). At this time it was concluded that there was no significant difference between the extent of the cholesterol reduction promoted by the various fibre sources. In recent times, analyses have focused on specific fibre type, indicating that there is a belief that all soluble fibres may not be equal, or that the term ‘soluble’ is not sufficiently specific. The soluble fibre component of importance for cholesterol reduction in oats has been identified as the β-glucans. Sub-analyses of the data, by source of soluble fibre, indicated that a dose-response relationship existed between oat soluble fibre intake and the degree of cholesterol reduction. It could be calculated that 3 g soluble fibre from oats (three servings of oatmeal, 28 g each) could decrease total and LDL-cholesterol by approximately 0·13 mmol/l(Reference Brown, Rosner and Willett52). An earlier analysis of ten studies evaluating the effect of oats inclusion and blood lipid modification indicated a change in total cholesterol level of − 0·13 (95 % CI − 0·19, − 0·017) mmol/l compared with control(Reference Ripsin, Keenan and Jacobs53). In recent times, a further meta-analysis was conducted as part of the health claim application for oats(54). In this meta-analysis, data were pooled from eighteen RCT; the overall effect was a reduction of − 0·34 (95 % CI − 0·42, − 0·25) mmol/l (P < 0·001) for total cholesterol and a reduction of − 0·28 (95 % CI − 0·35, − 0·22) mmol/l (P < 0·001) for LDL-cholesterol. When mean study differences were plotted against dose, estimates for effects of the recommended intake of 3 g oat β-glucan per d were − 0·23 (95 % CI − 0·35, − 0·10) and − 0·21 (95 % CI − 0·31, − 0·11) mmol/l for total and LDL-cholesterol, respectively, with no significant effect on HDL-cholesterol(54).
Barley β-glucan has also been shown to have cholesterol-lowering properties(Reference Talati, Baker and Pabilonia55). In a meta-analysis evaluating the lipid-reducing effects of barley, eight eligible trials (n 391) of 4–12 weeks' duration were identified. The inclusion of 2–10 g barley β-glucan lowered mean difference in total cholesterol by − 0·35 (95 % CI − 0·48, − 0·21) mmol/l, LDL-cholesterol by − 0·26 (95 % CI − 0·36, − 0·16) mmol/l and TAG by − 0·13 (95 % CI − 0·23, − 0·04) mmol/l, but did not significantly alter HDL-cholesterol. A further meta-analysis conducted with eleven eligible studies reached similar conclusions(Reference Abumweis, Jew and Ames56). Unpublished results from a third recently conducted meta-analysis, which included only studies using barley β-glucan of >50 kDa molecular weight in the analysis, reported a reduction in total and LDL-cholesterol of 0·41 (95 % CI − 0·75, − 0·44) and 0·33 (95 % CI − 0·42, − 0·25) mmol/l, respectively (both P < 0·0001) (JI Harland, M Ruffell and L Kolberg, unpublished results). By reference to a sub-analysis it was established that 3 g barley β-glucan was the minimum effective dose and, at this level of inclusion, LDL-cholesterol is reduced by 0·28 mmol/l, a reduction of approximately 7 % compared with baseline, and total cholesterol by 0·34 mmol/l, a reduction of 5·7 % compared with baseline. This is not dissimilar to estimates for calculated effects of 3 g oat β-glucan per d. Barley and oats contain a similar amount of β-glucans (3·5–5·9 % DM); it seems unlikely that there is a meaningful difference between the two cereals in cholesterol modification. However, as yet there are insufficient data to determine a dose-response relationship for barley β-glucan.
Health claims exist in the USA for oat and barley β-glucans and in Europe for β-glucans and maintenance of healthy cholesterol levels(Reference Ariyo and Villablanca57, 58) and for oat β-glucan and barley glucan for cholesterol reduction(54, 59). In order to carry the health claim in Europe, foods must meet the condition that 3 g β-glucans per d from oats, oat bran, barley, barley bran, or from mixtures of non-processed or minimally processed β-glucans, should be contained in one or more servings. To achieve this level of β-glucan intake, a typical serving of cereals required is about 84 g/d. In practice this is difficult to achieve and realistically half of this amount could be consumed in a single meal at, say, breakfast.
Cereals, particularly whole-grain cereals, have been associated with a lower incidence of CHD or coronary death. However, the evidence is largely provided by epidemiological studies and the mechanisms of action uncertain; consequently, whether cholesterol reduction is of significance, over and above the contribution made by viscous fibre components of whole grains, is not clear(Reference Seal60). In a meta-analysis of seven prospective cohort studies with quantitative measures of dietary whole grains and clinical cardiovascular outcomes it was shown that a greater whole-grain intake (pooled average 2·5 servings v. 0·2 servings/d) was associated with a 21 % lower risk of CVD events (OR 0·79; 95 % CI 0·73, 0·85)(Reference Mellen, Walsh and Herrington61).
Viscous fibre in the form of pectins has a recognised blood cholesterol-lowering effect(Reference Brown, Rosner and Willett52, 62). Pectins are linear chains of α-1-4-galacturonic acid units with side chains including galacturonic and glucuronic acids; they are viscous and water-soluble fibres found in fruits and vegetables. By reference to two meta-analyses, the most recent of which included seven studies (n 277 subjects), a statistically significant effect of pectins on total and LDL-cholesterol at intakes of 2·2–9 g/d was reported. There was also a significant dose-response relationship between the intake of soluble fibre (including pectins) and total and LDL-cholesterol lowering, but no significant dose-response relationship for HDL-cholesterol and TAG(Reference Brown, Rosner and Willett52). It was estimated that 1 g pectins/d produced significant changes in total and LDL-cholesterol of − 0·07 (95 % CI − 0·117, − 0·022) and − 0·05 (95 % CI − 0·087, − 0·022) mmol/l, respectively (both P < 0·05). It was suggested that in order to make a claim relating to the cholesterol effect, foods should provide at least 6 g pectins/d in one or more servings(62).
Another polysaccharide component for which an association between the maintenance of LDL-cholesterol and its consumption exists is chitosan(63). The EFSA assessed evidence relating this polymer of β-1-4-linked d-glucosamine and N-acetyl-d-glucosamine, which is a component of the exoskeleton of crustaceans and the cell walls of fungi. They suggested that the evidence indicated a small, but statistically significant effect on the reduction of both total (combining five studies) and LDL-cholesterol (combining two studies) concentrations, with no effect observed on HDL-cholesterol. The Panel suggested that in order to have an effect on blood lipids, 3 g chitosan/d should be consumed. The mechanism by which chitosan is presumed to exert the claimed effect is far from conclusive, but it was suggested that it binds to negatively charged lipids and reduces their gastrointestinal uptake(63).
Tree nuts
Initially data demonstrating an association between tree nut intake and CHD were derived largely from epidemiological evidence. A pooled analysis of four US epidemiological studies showed that subjects in the highest intake group for nut consumption had an approximately 35 % reduced risk of CHD incidence(Reference Kris-Etherton, Hu and Ros64). There was speculation as to the actual mechanism(s) involved and it has been suggested that a lipid-lowering effect, involving total and LDL-cholesterol, may be the key mechanism involved. However, the level of blood lipid reduction tends to be greater than that predicted from the fatty acid composition of nuts. For example, in an analysis of seventeen RCT, it was demonstrated in the results from ten of these studies that nuts had a greater lipid-lowering effect than would have been predicted from their fatty acid composition alone(Reference Kris-Etherton, Hu and Ros64). The average predicted reduction in LDL-cholesterol for the seventeen studies was 0·23 mmol/l and the measured reduction 0·29 mmol/l. It is suggested that nut protein, or some component thereof, may be responsible from this greater than predicted response. In a systematic review of studies that examined the relationship between the intake of nuts and their effect on blood lipids, it was found that in three almond (intake 50–100 g/d), two groundnut (35–68 g/d), one pecan nut (72 g/d) and four walnut (40–84 g/d) studies that the decreases in total cholesterol were between 2 and 16 % and those in LDL-cholesterol were between 2 and 19 %, compared with subjects consuming control diets(Reference Mukuddem-Petersen, Oosthuizen and Jerling65).
A further meta-analysis of five RCT (n 142) conducted specifically with almonds, where intake was 25–168 g/d, demonstrated a reduction in total cholesterol of 0·18 (95 % CI − 0·34, − 0·02) mmol/l (P < 0·05) and LDL-cholesterol of 0·15 (95 % CI − 0·29, 0·00) mmol/l (NS)(Reference Phung, Makanji and White66).
Studies conducted with walnuts were reviewed and random-effects meta-analysis of blood lipid outcomes was conducted in thirteen RCT (n 365)(Reference Banel and Hu67). In these RCT, walnuts provided 10–24 % of total energy intake. Diets supplemented with walnuts resulted in a significant reduction in total and LDL-cholesterol of 0·27 and 0·24 mmol/l, respectively (both P <0·001), while HDL-cholesterol and TAG were not significantly different. A further recent small study also indicated an effect on LDL-cholesterol, over and above that predicted from the composition of nuts, either walnuts or almonds, and virgin olive oil(Reference Damasceno, Pérez-Heras and Serra68).
In 2003 in the USA, a Qualified Health Claim was approved, which identified that the data suggested, but did not prove, that the intake of 1·5 oz (42 g) nuts/d may reduce the risk of heart disease. Usual intakes in both the USA and Europe fall short of this, where intakes are 21 and 31 g/d, respectively(Reference King, Blumberg and Ingwersen69). The important factor with nuts is that they are largely consumed as a snack and, as such, offer a healthy alternative to high-fat micronutrient-poor snacks that furthermore can make a significant contribution to cholesterol reduction and improve the fat quality by replacing about 5–8 g SFA with MUFA per portion.
Other foods or beverages that may modify blood lipids
A masterly analysis of the effects of flavonoids on heart disease risk factors was conducted in 2008 where data from 133 RCT were evaluated(Reference Hooper, Kroon and Rimm44). Within this review, sub-analysis of the effects of either chocolate/cocoa or green tea flavonoids on blood lipids (LDL) was conducted. Four studies were pooled for green tea (n 378) and five (n 214) for chocolate flavonoids. The resulting reduction in LDL-cholesterol was significant for green tea ( − 0·23 (95 % CI − 0·34, − 0·12) mmol/l), but not for chocolate flavonoids ( − 0·04 (95 % CI − 0·21, 0·13) mmol/l). While the RCT data for green tea look promising, further studies are needed to confirm the effect on blood lipids and when data were submitted to the EFSA for a health claim, an association with lipid lowering was not established(70). It seems likely that the beneficial effect that chocolate flavonoids may have on CHD risk is mediated through effects on blood pressure, not cholesterol.
Two meta-analyses, both retaining twenty-eight RCT, studied the effect on blood lipids where the dietary intervention was flaxseed, either whole or its lignan component(Reference Zhang, Wang and Liu71, Reference Pan, Yu and Demark-Wahnefried72). Flaxseed interventions reduced total and LDL-cholesterol by 0·10 (95 % CI − 0·20, 0·00) and 0·08 (95 % CI − 0·16, 0·00) mmol/l, respectively. No significant changes were found in the concentrations of HDL-cholesterol and TAG. The cholesterol-lowering effects were more apparent in females (particularly postmenopausal women) and those with high initial cholesterol concentrations. The authors concluded that while total cholesterol and LDL-cholesterol were reduced, further studies are needed to determine the efficiency of flaxseed on lipid profiles in men and premenopausal women. Flaxseed is a rich source of ALA, itself the subject of an assessment by the EFSA to evaluate its effect on blood lipids. It was concluded that a relationship was established between the intake of ALA and the maintenance of blood cholesterol concentrations. In order to bear the claim a food should contain at least 15 % of the proposed labelling reference intake value of 2 g ALA/d(73). This quantity of ALA can be found in approximately 2–3 g flaxseed.
Other legumes can be effective in cholesterol reduction, as indicated in a systematic review of the literature where 140 relevant reports were identified, of which ten RCT (n 268) were selected as being suitable for meta-analysis(Reference Bazzano, Thompson and Tees74). A non-soya legume diet was compared with control, for a minimum duration of 3 weeks and changes in blood lipid reported. Pooled mean net change in total cholesterol for those consuming a legume diet compared with control was − 0·31 (95 % CI − 0·42, − 0·19) mmol/l and mean net change in LDL-cholesterol was − 0·21 (95 % CI − 0·30, − 0·12) mmol/l.
Garlic has many claimed health effects; in order to establish if there are quantifiable benefits on blood lipids, a meta-analysis was conducted of thirteen RCT (n 1056)(Reference Khoo and Aziz75). In this analysis the inclusion of garlic did not result in any significant differences in all outcome measures examined, when compared with placebo. The effects on total and LDL-cholesterol were − 0·04 (95 % CI − 0·15, 0·07) and − 0·01 (95 % CI − 0·10, 0·11) mmol/l, respectively; there were no marked differences in HDL-cholesterol, TAG and apoB. In a second meta-analysis, twenty-nine trials were pooled and the effect of garlic on blood lipids reported. Garlic was found to significantly reduce total cholesterol by − 0·19 (95 % CI − 0·33, − 0·06) mmol/l and TAG by − 0·11 (95 % CI − 0·19, − 0·06) mmol/l, but had no significant effect on LDL- or HDL-cholesterol. There was a moderate degree of statistical heterogeneity in both meta-analyses, indicating a need for further studies to improve the confidence of this finding. Recently it has been suggested that garlic can affect several cardiovascular risk factors, not just lipoprotein modification, and may have important antioxidant and antiplatelet functions as well as inhibition of LDL uptake(Reference Rana, Pal and Vaiphe76).
Contribution that specific food combinations can make to blood lipid modification
Data relating to the cholesterol-lowering effect of single individual foods and food ingredients have been summarised, but a key question remains to be addressed: to what extent are the reductions in cholesterol additive when these food components are combined in the diet? Evidence provided by the ‘portfolio diet’ studies, where multiple changes to the diet have been made, has been reviewed to gain an understanding of this aspect.
Portfolio diets
To identify such studies, the Medline search undertaken used the specific search term ‘portfolio diet’. From this, and a hand search of identified papers, eighteen ‘portfolio’ studies were identified, of which eleven were retained for further examination(Reference Jenkins, Kendall and Faulkner11, Reference Jenkins, Chiavaroli and Wong12, Reference Jenkins, Kendall and Marchie 77–Reference Jenkins, Kendall and Mehling85). Excluded studies were those that did not measure blood lipids(Reference Robinson and Maheshwari86–Reference Jenkins, Kendall and Marchie91), or were duplicates of other reports(Reference Jenkins, Nguyen and Kendall92). The oldest study identified in this series makes only two dietary changes, the introduction of soya protein and viscous fibre; information related to it has been detailed separately from the main portfolio diet studies, but it is useful in gaining an understanding of the extent to which the changes promoted by these foods are additive(Reference Jenkins, Kendall and Mehling85). A summary of the retained studies and the dietary modifications made is detailed in Table 1.
Intake of specific foods that constitute component parts of ‘portfolio diets’ in trials identified by the present review
SP soya protein; CHO, carbohydrate; NCEP, National Cholesterol Education Program.
* 1000 kcal = 4184 kJ.
† Included study period of 10 weeks without plant sterols. Note that this is an extension to Jenkins et al. (Reference Jenkins, Kendall and Faulkner81).
There are 656 subjects that have participated in ‘portfolio studies’, approximately half of whom are men and half are postmenopausal women. Study duration was from 4–80 weeks. Studies have been conducted in both a metabolic setting (short-term clinical efficacy studies) with virtually all foods supplied and in community-based studies of ≥ 6 months' duration, where free-living subjects are given dietary advice to help them adopt a portfolio diet.
Typically, the portfolio diets contain per 1000 kcal (4·18 MJ): 1 g plant stanols or sterols, 16–23 g soya protein, 8–10 g viscous fibre (generally from oats, barley, psyllium and okra) and 16–23 g nuts, usually almonds or a combination of tree nuts and groundnuts. Typical total energy intake is about 2500 kcal (10·5 MJ)/d except in the weight-loss study, where it is 2000 kcal (8·4 MJ)/d(Reference Jenkins, Wong and Kendall79). A single study has evaluated the effect of adding MUFA to this combination of foods(Reference Jenkins, Chiavaroli and Wong12). The diets are generally low or very low in SFA and the emphasis is on plant foods, with many of the interventions being either vegan or vegetarian.
The predicted reduction in cholesterol for each of these components has been calculated for a ‘reasonable’ daily intake of that food component as proposed by the portfolio diet and, in many cases, this amount consumed is markedly less than the quantity from which the effect on LDL-cholesterol was originally derived. Where this occurs, it has been assumed that a dose-response relationship exists between the amount of food or food ingredient intake and cholesterol lowering and a pro rata response calculated (see Table 2). This assumption is known to be valid and a dose-response relationship exists for β-glucans from oats and cholesterol reduction(58), and similarly for MUFA from a variety of sources(Reference Mensink, Zock and Kester18). However, it is rather more contentious as to whether such a relationship exists for soya protein, where a dose-response relationship was reported in one analysis(Reference Reynolds, Chin and Lees37) and not in others(Reference Weggemans and Trautwein38, Reference Harland and Haffner43). Nevertheless, despite these weaknesses, the data in Table 2 demonstrate that each single dietary intervention can have a modest effect on LDL-cholesterol and that these effects are biologically plausible. When the predicted cholesterol lowering for each component part of the portfolio diet is added together, the overall predicted reduction in LDL-cholesterol is about 33 %.
Relative contribution that specific foods can make to blood lipid modification
NA, not available.
The extent of cholesterol modification associated with each of the ‘portfolio diet’ trials is detailed in Table 3. Where there was a control group, often a low-fat-diet, included in the study, the reported LDL-cholesterol reduction in the treatment group is over and above that reported for the control, i.e. a net effect. Where there was no control group, and in all studies where measures of LDL-cholesterol are taken at intervals, the value recorded for the longest period on trial is reported.
Study characteristics and the effect on blood cholesterol and ratios of total or LDL-cholesterol:HDL-cholesterol where ‘portfolio diets’ adopted
RCP, randomised controlled parallel study; M, male; PF, postmenopausal female; NA, not available; RP, randomised parallel study; F, female; IT, single-phase intervention trial (comparison with baseline); RCX, randomised controlled cross-over trial.
* Standard counselling two visits/6 months.
† Intensive counselling seven visits/6 months.
‡ Difference compared with baseline.
§ End of study high-MUFA portfolio compared with low MUFA.
‖ Plant sterols throughout study period.
¶ No plant sterols for weeks 52–62.
From Table 3, it can be seen that the reported reduction in LDL-cholesterol is close to the predicted value of about 33 % in two studies, both of which were simple interventions of short duration (approximately 4 weeks) with no control group, where the reduction reported was 30 and 29 %, respectively(Reference Jenkins, Kendall and Faulkner11, Reference Jenkins, Kendall and Marchie84). Longer-term intervention studies (>6 months) with a control group generally result in a reduction of LDL-cholesterol, approximately half of the predicted value, at about 15 %(Reference Jenkins, Jones and Lamarche78, Reference Jenkins, Kendall and Nguyen80, Reference Jenkins, Kendall and Faulkner81). One of the main factors responsible for this lower rate appears to be a reduction in compliance. In the short-term metabolic studies compliance is typically >90 %(Reference Jenkins, Chiavaroli and Wong12, Reference Jenkins, Jones and Lamarche78), whereas compliance measured in the 80-week study was 62, 53, 68 and 67 % for soya protein, viscous fibre, almonds and plant sterol and stanols components, respectively(Reference Jenkins, Kendall and Nguyen80). Overall compliance with the ‘portfolio diet’ was typically about 60 %, but during a period when plant sterols or stanols were removed from the diet, compliance dropped to 45 %(Reference Jenkins, Kendall and Nguyen80). Attempts to improve compliance by more intensive dietary counselling, increasing counselling sessions from two to seven, had little overall effect on compliance although dietary adherence was directly associated with LDL-cholesterol reduction(Reference Jenkins, Jones and Lamarche78). From the studies it appears that it is most difficult to comply with the target viscous fibre intake and perhaps easiest to comply with the target plant sterol and stanols and nuts intake.
Perhaps this is not too surprising when considered from a practical perspective, as generally foods providing plant sterols or stanols or soya protein are simple exchanges and easy to include in the diet, for example, replacing a pot of dairy yoghurt with a pot of soya or plant stanol or sterol equivalent, or replacing dairy milk with soya milk. Almonds or tree nuts can be provided to study volunteers as snack packets; these, too, are easy to accommodate in the diet as convenient replacements for higher-fat, -salt and -sugar snacks. However, introducing viscous fibres into the diet may necessitate changes to a number of dietary components to provide a balanced meal and require a fundamentally different approach to the diet. Where MUFA were included in the diet this was achieved by the addition of high-oleic sunflower-seed oil to the diet. Both high-oleic sunflower-seed and rapeseed are attractive salad oils that have the added advantage of being stable during many food processes and, as such, are suitable alternatives to fats high in TFA and SFA for a variety of food manufacturing purposes(Reference Minihane and Harland93).
Mechanisms of action
Clearly, while compliance with the portfolio regimen is a significant factor is assessing whether the component parts of the diet are genuinely additive, an understanding of the mechanisms of action by which each component in the diet is thought to act and the identification of any potential incompatibility between them are also important.
An inhibition of intestinal absorption of both dietary and endogenous cholesterol is the mechanism by which plant sterols and sterols are thought to exert their effect, although the molecular mechanisms involved are not fully resolved. In simple terms, the presence of increased quantities of plant sterols or sterols competes with cholesterol for micellar solubility, therefore lowering the amount of cholesterol available for absorption by intestinal mucosal cells(Reference Deng94, Reference Brown, Hang and Dussault95).
Despite markedly different structure and composition, it has been suggested that viscosity rather than quantity of soluble dietary fibre predicts its cholesterol-lowering effect(Reference Vuksan, Jenkins and Rogovik96). Furthermore, it has been reported that no major differences exist between the four different types of water-soluble fibres (β-glucan, psyllium, pectin and guar gum) in the extent to which they reduce cholesterol(Reference Brown, Rosner and Willett52); thus it is proposed that there is a common underlying mechanism, namely that water-soluble fibre slows the reabsorption of particular bile acids and as a result hepatic conversion of cholesterol into bile acids increases, resulting in turn in an increased LDL uptake by the liver(Reference Theuwissen and Mensink97). Certainly with regard to β-glucan there is good evidence that its mechanism of action is related to viscosity; with viscosity in the small intestine determined by the concentration, molecular weight and solubility of the β-glucans(54). With pectins it is their molecular size and degree of esterification that have been shown to be important determinants of effectiveness(Reference Brouns, Theuwissen and Adam98). Nevertheless, the core cholesterol-lowering mechanisms for soluble fibres seem to involve reabsorption of bile acids, increased synthesis of bile acids from cholesterol and reduced circulating LDL-cholesterol concentrations.
The actual mechanism involved in the cholesterol-lowering process when soya protein is incorporated in the diet is less clear. Some researchers suggest it is the protein itself, in particular the 7S globulin, that up-regulates the activity of the LDL receptors(Reference Manzoni, Duranti and Eberini99), while others also suggest a role for the soya isoflavones associated with the protein(Reference Xiao, Mei and Wood100). However, isolated soya isoflavones have generally not been found to reduce cholesterol(Reference Taku, Umegaki and Ishimi101). In all probability the major mechanism(s) by which cholesterol is lowered is largely related to the soya protein component per se, but as the isoflavones in their natural form are integrally linked to the protein, they too may be important in determining effectiveness. Certainly, soya protein processed in a manner to deplete the isoflavone content generally has a reduced cholesterol-lowering capacity(Reference Taku, Umegaki and Sato42).
As recently concluded by the EFSA, the mechanism by which tree nuts, or more specifically walnuts, affect blood cholesterol was attributed to their content of MUFA and PUFA(102). This conclusion was made despite indications from the data detailed previously; nonetheless, in weighing the evidence it concluded that the consumption of walnuts did not have an effect on blood LDL-concentrations beyond what could be expected from their fatty acid composition.
The added value of including MUFA in the diet and their effect on blood lipid profile was also shown in a single ‘portfolio diet’ study that incorporated a high-oleic sunflower-seed oil(Reference Jenkins, Chiavaroli and Wong12). This study achieved the joint objectives of reducing LDL-cholesterol levels and improving HDL-cholesterol levels, thus demonstrating the value of improving fat quality of the diet, as well as introducing cholesterol-lowering foods into the diet.
As to whether there are any conflicts or synergies between the modus operandi of the various components of the portfolio diet is less well understood. As the reported changes in blood lipids in short-term well-controlled portfolio-diet studies appear to be not too different from the value predicted from the sum of the individual components, there largely appears to be an additive effect. However, there is still a high degree of uncertainty with regard to the detailed mechanisms involved and, because of this, the degree of synergy can best be determined by looking at simple substitutions, made in a stepwise fashion, rather than in studies where multiple changes are made at the same time. Thus, well-conducted studies where only two component changes are made and each accessed independently, as well as together, provide a better insight to the extent of synergy or antagonism between the cholesterol-lowering foods(Reference Theuwissen and Mensink103).
Simple food combinations
Long-term compliance with dietary intervention is crucial to effectiveness. What influences compliance is a huge subject in its own right, a key aspect of which is motivation that can vary widely among individuals. In a review of three studies investigating subjects' motivation to change their diet to match nutritional goals, all participants achieved a change in dietary intake over 1 week and met one nutritional goal(Reference De Looy, Auty and Coates104). However, it was much easier to substitute one item in the diet for another or reduce consumption of certain foods that contained or contributed a significant amount of fat, such as margarine and milk. In the short term dietary change may be relatively easy, but over the longer term availability of foods, especially snack foods, can make adherence difficult. Subjects in the Diet and Reinfarction Trial (DART) found it relatively easy to make one dietary alteration, but where two nutrients were involved it became more difficult(Reference Fehily, Vaughan-Williams and Shiels105). In many respects the macronutrient content of the ‘portfolio diet’ closely resembles DART interventions, particularly the two interventions used in the OmniHeart studies, one of which was a protein-rich diet with about half the protein from plant sources and the second, a diet rich in unsaturated fat, predominantly MUFA(Reference Appel, Sacks and Carey106).
In considering effective food combinations for cholesterol modification and maximum compliance, it may therefore be more effective to restrict the number of changes to the diet and focus on making only two changes. Two pertinent dietary changes with a high level of compliance could result in substantial cholesterol lowering. There is a limited number of studies in the literature where changes have been made to two discrete cholesterol-lowering components, without significant manipulation of the background diet, and where the effect of each has been quantified separately and as a combination, necessary to establish the extent of synergy between the agents; see, for example, Theuwissen & Mensink(Reference Theuwissen and Mensink103) and Yoshida et al. (Reference Yoshida, Vanstone and Parsons107). In a further study the effects of oats and soya protein were assessed independently and together, but after all subjects had complied with the macronutrient targets of an American Heart Association step 1 diet for 3 weeks(Reference Van Horn, Liu and Gerber108). Other studies report the effect of introducing two cholesterol-modification agents in the diet, but do not assess the effect of each independently; see, for example, the evaluation of soya drinks with plant sterol or stanols(Reference Weidner, Krempf and Bard109–Reference Cicero, Fiorito and Panourgia111). A summary of studies that report use of two of the cholesterol-lowering components detailed in the earlier sections is given in Table 4.
Study characteristics and the effect on LDL- and HDL-cholesterol when two cholesterol-modifying components are introduced in the diet
IT, single-phase intervention trial (comparison with baseline); M, male; F, female; RCX, randomised controlled cross-over trial; RCP, randomised controlled parallel study; DB, double-blinded study; PF, postmenopausal female; RP, randomised parallel study; T II, type 2 diabetics.
* Between treatment and control.
† Within treatment.
‡ Within treatment, total cholesterol.
§ Data presented for non-diabetic subjects only.
There is considerable variation in the quality of the studies detailed in Table 4; three studies were double-blinded, six studies included a control and, of these, only three studies included a placebo intervention. Study duration was relatively short and varied between 3 weeks and 3 months. Three studies used plant stanol or sterol–soluble fibre combinations(Reference Theuwissen and Mensink103, Reference Yoshida, Vanstone and Parsons107, Reference Shrestha, Volek and Udani112), of which two determined the effect of the individual components, as well as the combination(Reference Theuwissen and Mensink103, Reference Yoshida, Vanstone and Parsons107).
From these studies it can be seen that the effects of introducing two food components into the diet are to a degree additive, although less than would have been predicted in respect of the plant stanol–oat β-glucan combination(Reference Theuwissen and Mensink103). Specifically, the addition of plant stanol esters to β-glucan-enriched muesli significantly lowered LDL-cholesterol concentrations by 4·4 %, less than the estimated change of approximately 8·5 %. The authors suggested that the lower response was due to the higher viscosity in the intestinal lumen produced by the β-glucan, which interfered with the absorption of plant stanols. As plant stanols reduce intestinal cholesterol absorption through competition for micellar incorporation and promote cholesterol efflux back into the intestinal lumen this may have been compromised, resulting in both reduced serum concentrations and lower efficacy of the plant stanols. Alternatively, higher intestinal viscosity may affect lipid emulsification and the hydrolysis of plant stanol esters to active free plant stanols released later in the intestinal tract and thus affect overall efficacy and absorption. A study in which a glucomannan–plant sterol combination was used also provides an insight to the mechanisms involved. In this study plasma lathosterol, a cholesterol precursor and an index of cholesterol synthesis, was determined(Reference Yoshida, Vanstone and Parsons107). Plasma lathosterol concentrations were increased from baseline with plant sterol treatment, but not with glucomannan supplementation; furthermore the glucomannan–plant sterol combination resulted in decreased plasma lathosterol (P < 0·05), suggesting that glucomannan suppresses the compensatory increase of cholesterol synthesis by plant sterol intake(Reference Yoshida, Vanstone and Parsons107). In a study where mildly hypercholesterolaemic subjects consumed β-glucan with or without plant stanol esters, there was no significant effect on inflammatory parameters, nor were there changes in expression profiles of eighty-four genes involved in atherosclerosis metabolism when β-glucan was included in the diet and the expression of only three genes – ADFP (adipose differentiation-related protein), CDH5 (cadherin 5, type 2) and CSF2 (granulocyte-macrophage colony-stimulating factor 2) – changed after plant stanol ester consumption(Reference Theuwissen, Plat and Mensink113).
It has also been suggested that the food matrix in which plant stanols or sterols are presented can influence their effectiveness(Reference Abumweis, Barake and Jones32). Four studies evaluated the effect of soya protein–plant stanol or sterol combinations and reported a LDL-cholesterol reduction of between 6·8 and 14·6 %(Reference Weidner, Krempf and Bard109–Reference Cicero, Fiorito and Panourgia111, Reference Lukaczer, Liska and Lerman114). However, as only one of these studies included a control group, which was advice to follow an American Heart Association step 1 diet, there is neither the basis to assess the degree to which these changes were additive nor the effect of the food matrix.
Soya protein–soluble fibre combinations demonstrated an effect on cholesterol that is additive to some degree in the early study by Jenkins et al. (Reference Jenkins, Kendall and Mehling85), although there was little synergistic effect on LDL-cholesterol in a further well-controlled study(Reference Van Horn, Liu and Gerber108), but a greater effect in a less controlled study of longer duration(Reference Torres, Guevara-Cruz and Granados115).
The extent to which these reported reductions in LDL-cholesterol is a reflection of the true additive nature of the dietary interventions and that could be predicted from each component, or good compliance, is not clear. The variable quality, relatively small size and limited duration of these studies make interpretation difficult.
Nevertheless, in practical situations, reductions in LDL-cholesterol concentration of about 15 % have been achieved in studies of 4–12 weeks' duration by both soya protein–plant sterol and viscous fibre–soya protein combinations(Reference Cicero, Minardi and Mirembe110, Reference Cicero, Fiorito and Panourgia111, Reference Lukaczer, Liska and Lerman114), representing a significant benefit and reduction in coronary risk to those otherwise healthy subjects with mild hypercholesterolaemia.
When considering effective public health measures, reference to habitual diet could be made to select the interventions most likely to be effective for each individual. Identification of the most appropriate food or food component exchanges for an individual, by reference to a diet diary, improves the likelihood of success due to better compliance. In addition, by careful selection of the components involved, the joint objective of cholesterol improvement and improved fat quality in the diet could be achieved. Two relatively small dietary changes could typically result in a reduction in LDL-cholesterol of 5–15 %, sufficient to reduce coronary risk by 10–30 % and make a meaningful contribution to public health(Reference Baigent, Blackwell and Emberson116). Examples of the impact on both LDL-cholesterol and fat quality of the diet by including some of the previously discussed food or food component exchanges as replacements for less healthy frequently consumed foods and snacks are summarised in Table 5.
Relative contribution that practical intakes of specific foods can make to blood lipid and fat quality modification
* Food compositional data taken from McCance and Widdowson's Composition of Foods(117). The suggested food replacements are based on a typical serving or reasonable daily intake of that food.
† Assumes that baseline LDL-cholesterol is 3·75 mmol/l.
‡ Data converted from mg/dl using a conversion factor of 0·02 586 throughout.
§ Statistically significant in some analyses, but not all.
Conclusions
There are a number of foods and food components for which good evidence, from well-conducted RCT, indicates an ability to significantly reduce LDL-cholesterol; typically the reduction is 3–9 % compared with baseline when a single component is included in the diet. In the main, it appears that the combined effect of individual components is to some degree additive when used in simple combinations, although determination of the extent of this synergy is limited by the quality of some of the studies in the literature. Nevertheless, evidence exists that suggests that the introduction of two or more of these foods or food components into the diet, at their target level of intake and in a highly practical manner, can be achieved and results in a reduction of LDL-cholesterol of 5–15 %. As a public health measure, simple dietary changes can have a significant impact on cholesterol management in those with mild hypercholesterolaemia.
Where a ‘portfolio diet’ is adopted the extent of the LDL-cholesterol reduction is much greater and up to 30 % in short-term well-controlled studies where the foods are provided and 15 % in community-based studies over periods of 6 months or more. Compliance with dietary change and, in particular, the challenge of multiple interventions influence the degree of effectiveness of the regimen. Consequently, careful consideration needs to be given to the appropriate food or food components combination for an individual in order to assist long-term compliance and effectiveness of the intervention. Furthermore, improving the fat quality of the diet by the introduction of unsaturated fat can also enhance blood lipid changes by helping to maintain or improve HDL-cholesterol concentrations. It is suggested that the future focus of cholesterol-lowering regimens needs to place emphasis not only on the effect of food components on LDL-cholesterol, but also the role they can play in improving fat quality.
Acknowledgements
The present review and its interpretation have been produced independently. This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors. The assistance of Katie Kennedy in assessing the macronutrient intake from portfolio diets is acknowledged.
There are no conflicts of interest. J. I. H. provides nutrition consultancy services to a wide cross-section of the food industry and food industry organisations.
