Epidemiological studies

Four large prospective studies conducted in the USA examining associations of dietary components with health outcomes reported a beneficial effect of nut consumption on fatal and non-fatal CHD after follow-up ranging from 6 to 18 years⁽Reference Fraser, Sabaté and Beeson^2, Reference Hu, Stampfer and Manson^23–Reference Blomhoff, Carlsen and Andersen²⁵⁾. These studies were reviewed in 2006 by Kelly & Sabaté⁽Reference Kelly and Sabaté²⁶⁾, who estimated an 8·3 % reduction in CHD risk for each weekly serving of nuts. Protection against CHD mortality with increasing nut consumption was also observed in the PREDIMED study population when assessing the cohort as a longitudinal study⁽Reference Guasch-Ferré, Bulló and Martínez-González²⁷⁾. A recent report of two large cohorts of women from the Nurses' Health Study (NHS) and men from the Health Professionals Follow-up Study with follow-up of 30 and 24 years, respectively, again showed reduced CHD mortality for both sexes with increasing nut consumption⁽Reference Bao, Han and Hu²⁸⁾ (Fig. 1). Remarkably, nut consumption was also associated with a reduction in all-cause and cancer mortality in these two reports⁽Reference Guasch-Ferré, Bulló and Martínez-González^27, Reference Bao, Han and Hu²⁸⁾. An additional report from the NHS concerning women with type 2 diabetes further evaluated the relationship between nut consumption and CVD (including CHD and stroke) and observed a multivariable adjusted 54 % lower risk of CVD⁽Reference Li, Brennan and Wedick²⁹⁾. Although considering nuts and fruits together, findings from the prospective Greek European Prospective Investigation into Cancer and Nutrition cohort also suggest a protective effect of higher consumption against CHD, with 23 and 34 % risk reductions for men and women, respectively⁽Reference Dilis, Katsoulis and Lagiou³⁰⁾. A dose–response relationship between nut consumption and reduced CHD mortality was observed in all studies. Consistent findings in all prospective studies strongly suggest a causal association between nut consumption and reduced CHD rates.

Fig. 1

Incidence of fatal CHD by frequency of nut intake. Shown are the results of six large prospective studies where nut consumption was the exposure of interest and the outcome was death from CHD. , Adventist Health Study, 1992; , Nurses' Health Study, 1998; , Physicians' Health Study, 2002; , Iowa Women's Health Study, 2006; , PREvención con DIeta MEDiterránea, 2013; , Nurses' Health Study and Health Professionals Follow-up Study, 2013. A colour version of this figure can be found online at http://www.journals.cambridge.org/bjn

The findings of prospective studies are less consistent regarding nut consumption and stroke risk. Thus, a non-significant decreased risk of fatal stroke with increasing nut consumption was reported for postmenopausal women participating in the Iowa Women's Health Study⁽Reference Yochum, Folsom and Kushi³¹⁾; no association between frequency of nut consumption and stroke risk was observed in men from the Physicians' Health Study⁽Reference Djoussé, Gaziano and Kase³²⁾, and an additional report of the large NHS and Health Professionals Follow-up Study cohorts showed that regular nut consumption related to a weak but significant lower risk of stroke in women, but not in men⁽Reference Bernstein, Pan and Rexrode³³⁾.

The association between nut consumption and type 2 diabetes risk has also been investigated in prospective studies. Findings from the large NHS cohort suggested that nut and peanut butter consumption was associated with a reduced incidence of diabetes in women⁽Reference Jiang, Manson and Stampfer³⁴⁾. No such relationship was observed in postmenopausal women participating in the Iowa Women's Health Study⁽Reference Parker, Harnack and Folsom³⁵⁾. A subsequent report from a Chinese cohort of nearly 64 000 women also showed that nut consumption protected from diabetes risk⁽Reference Villegas, Gao and Yang³⁶⁾. These findings, however, were not confirmed in male participants in the Physicians' Health Study⁽Reference Kochar, Gaziano and Djoussé³⁷⁾. A more recent report from a large cohort of women combining data from NHS and NHS II after follow-up of 10 years showed that consumption of walnuts, but not other nuts, was associated with reduced rates of incident diabetes⁽Reference Pan, Sun and Manson³⁸⁾. Finally, a cross-sectional assessment of the PREDIMED cohort of older persons at high cardiovascular risk showed a reduced prevalence of diabetes with increasing nut consumption⁽Reference Ibarrola-Jurado, Bulló and Guasch-Ferré³⁹⁾. In this study, a lower prevalence of the metabolic syndrome, a high-cardiovascular risk cluster of risk factors related to insulin resistance, was also observed in frequent nut eaters⁽Reference Ibarrola-Jurado, Bulló and Guasch-Ferré³⁹⁾. Similarly, nut consumption is related to lower incident metabolic syndrome in the prospective Seguimiento Universidad de Navarra cohort of Spanish university graduates⁽Reference Fernández-Montero, Bes-Rastrollo and Beunza⁴⁰⁾ and in a large cross-sectional report from the US National Health and Nutrition Examination Survey for 2001–4⁽Reference O'Neil, Keast and Nicklas⁴¹⁾.

Four prospective studies have evaluated the frequency of nut consumption in relation to incident hypertension⁽Reference Steffen, Kroenke and Yu^42–Reference Weng, Steffen and Szklo⁴⁵⁾. In the young biracial cohort of the Coronary Artery Risk Development in Young Adults study, subjects in the upper category of nut consumption had a 15 % decreased risk of hypertension compared with the lowest category after 15 years of follow-up⁽Reference Steffen, Kroenke and Yu⁴²⁾. In the large Physicians' Health Study men cohort⁽Reference Djoussé, Rudich and Gaziano⁴³⁾, the risk of hypertension was reduced by 18 % in participants who consumed nuts daily compared with non-consumers, although the benefit was restricted to participants who were not overweight or obese at baseline. Another report from the Seguimiento Universidad de Navarra cohort found a non-significant 23 % lower risk of incident hypertension for the highest v. lowest nut consumption category after 4·3 years of follow-up⁽Reference Martínez-Lapiscina, Pimenta and Beunza⁴⁴⁾. Moreover, a recent report from the Atherosclerosis Risk in Communities study in middle-aged adults showed a 13 % lower risk of incident hypertension in the top quintile of nut consumption compared with the bottom quintile after follow-up for 9 years⁽Reference Weng, Steffen and Szklo⁴⁵⁾. The US National Health and Nutrition Examination Survey 2001–4 cross-sectional study also reported lower prevalent hypertension for nut consumers compared with non-consumers⁽Reference O'Neil, Keast and Nicklas⁴¹⁾.

As reviewed⁽Reference Kris-Etherton, Hu and Ros^7, Reference Ros, Tapsell and Sabaté⁸⁾, prospective studies have also reported an association between increasing nut consumption and reduced circulating levels of inflammatory biomarkers.

In summary, the results of several prospective studies strongly suggest that regular consumption of nuts is beneficial for CHD risk. The risk of stroke in relation to nut consumption has been less investigated compared with CHD risk, and data support a weak protective effect in women, but not in men. Confirmation of a protective role on stroke as well as diabetes risk must await further studies. Data from epidemiological studies also suggest a protective effect of nut consumption on the metabolic syndrome and hypertension, one of its components. Finally, there is also suggestive epidemiological evidence that frequent nut consumption is associated with reduced systemic inflammation.

Randomised clinical trials with outcomes on intermediate biomarkers

RCT to investigate the effects on cardiovascular risk factors and an attempt to understand the underlying mechanisms for protection from CHD followed the first observational studies. Most RCT have compared diets supplemented with nuts with control diets for outcomes on blood lipid changes in individuals with normal or moderately elevated blood cholesterol levels. Some studies have also evaluated the effects of nuts on glycaemic control in patients with type 2 diabetes. Other studies have focused on the relevant question of whether unrestricted nut intake influences body weight. Further RCT have assessed the effects of nut diets on intermediate risk markers, such as blood pressure, insulin sensitivity, endothelial function and inflammatory status. Almonds and walnuts have been the most studied nuts, followed by pistachios. The long-term PREDIMED trial has targeted both the effects of nut consumption on intermediate cardiometabolic markers and clinical outcomes, such as the metabolic syndrome, diabetes and cardiovascular events, among others. Given the relevance of PREDIMED for the issue of nuts and cardiovascular health, its results will be dealt within a separate section.

Since the first RCT by Sabaté et al. ⁽Reference Sabaté, Fraser and Burke³⁾ examining the lipid effects of nuts showed that a walnut diet reduced blood cholesterol levels in healthy individuals, the effects of nut consumption on blood lipids and lipoproteins have been investigated in many studies. A pooled analysis of twenty-five RCT conducted in seven countries examining nut-enriched diets v. control diets during periods of 3–8 weeks for outcomes on blood lipids clearly demonstrated that nuts had a cholesterol-lowering effect that was related to both dose and baseline cholesterol level, stratified by sex and across all age groups, and independent of the type of nut tested⁽Reference Sabaté, Oda and Ros⁴⁶⁾. Specifically, daily consumption of 67 g (2·4 oz) of nuts was associated with mean reduction of 10·9 mg/dl (5·1 %) in total cholesterol, 10·2 mg/dl (7·4 %) in LDL-cholesterol and 0·22 (8·3 %) in LDL:HDL ratios (P< 0·001, all). Nuts had no significant effect on HDL-cholesterol or TAG, except in participants with serum TAG >150 mg/dl, in whom a significant mean 10·2 mg/dl reduction was observed. A separate meta-analysis of thirteen RCT examined the effect of walnut-enriched diets on blood lipid levels⁽Reference Banel and Hu⁴⁷⁾. Compared with control diets, diets enriched with walnuts in amounts ranging from 30 to 108 g/d (10–24 % of energy) were associated with weighed mean reductions of total cholesterol and LDL-cholesterol of 10·3 and 9·2 mg/dl (P< 0·001, both), respectively, which are similar to those described in the pooled analysis for different nut types⁽Reference Sabaté, Oda and Ros⁴⁶⁾. RCT testing the lipid effects of nuts have usually been short term, but efficacy is maintained in the long term. Thus, inclusion of walnuts (12 % of energy) to the usual diet of free-living individuals to whom no dietary advice was given during a 6-month period reduced total cholesterol by 5 mg/dl (P= 0·02) and TAG by 8 mg/dl (P= 0·03), with a nearly significant decrease of 3·5 mg/dl in LDL cholesterol (P= 0·06)⁽Reference Torabian, Haddad and Cordero-MacIntyre⁴⁸⁾. Also, a walnut diet against a control diet for up to 1 year in obese patients with type 2 diabetes showed maintained, albeit small, LDL-cholesterol reductions and HDL-cholesterol increases with the walnut diet⁽Reference Tapsell, Batterham and Teuss⁴⁹⁾.

Acute studies using test meals with a high glycaemic index with or without nuts have shown reduced postprandial glucose responses with nut meals, suggesting that nuts may be useful in controlling diabetes⁽Reference Kendall, Josse and Esfahani⁵⁰⁾. Improved glycaemic control has been recently reported in patients with type 2 diabetes fed diets enriched with 2 oz mixed nuts⁽Reference Jenkins, Kendall and Banach⁵¹⁾ or almonds⁽Reference Li, Liu and Liu⁵²⁾v. control diets for 12 weeks. It may seem counterintuitive to recommend energy-dense nuts to diabetic patients for the fear of gaining weight. Indeed, there is a widespread perception that all fatty foods provide excess energy and promote obesity, which has had a negative effect on the image of nuts. However, there is no epidemiological or RCT evidence that their frequent consumption promotes weight gain; on the contrary, reports from large prospective cohorts show inverse associations between the frequency of nut consumption and BMI or weight gain over time after adjustment for confounders⁽Reference Fraser, Sabaté and Beeson^2, Reference Hu, Stampfer and Manson^23, Reference Bes-Rastrollo, Sabaté and Gómez-Gracia^53–Reference Mozaffarian, Hao and Rimm⁵⁵⁾. Similar findings regarding BMI together with reduced waist circumference have been shown for increasing nut consumption in cross-sectional studies⁽Reference O'Neil, Keast and Nicklas^41, Reference Casas-Agustench, Bulló and Ros^56, Reference Martínez-González, García-Arellano and Toledo⁵⁷⁾ and in the PREDIMED cohort considered as a longitudinal study⁽Reference Ibarrola-Jurado, Bulló and Guasch-Ferré³⁹⁾. Finally, a recent meta-analysis of RCT that tested nut diets v. control diets showed small non-significant associations of nut intake with reduced adiposity measures⁽Reference Flores-Mateo, Rojas-Rueda and Basora⁵⁸⁾, as had been previously described for walnut trials⁽Reference Banel and Hu⁴⁷⁾. Mechanistically, the lack of weight gain after consuming nuts is largely due to their prominent satiating effect, which results in food compensation accounting for up to 75 % of the energy they provide⁽Reference Mattes, Kris-Etherton and Foster⁵⁹⁾. An additional factor contributing to offset energy acquisition after eating nuts is fat malabsorption, documented as increased faecal fat excretion, which is due in part to incomplete digestion of their matrices because the fat of nuts is enclosed within cell membranes, which are not readily available to digestive enzymes even after thorough mastication⁽Reference Ellis, Kendall and Ren⁶⁰⁾. In conclusion, a substantial body of evidence permits to allay fear of undue weight gain when eating these high-fat foods.

The unique composition of nuts in bioactive nutrients and phytochemicals suggests that their consumption can beneficially affect cardiometabolic markers other than blood lipids and glycaemic control. A few studies have examined the effects of nuts on oxidative stress, inflammation and vascular reactivity, as reviewed⁽Reference Kris-Etherton, Hu and Ros^7, Reference Ros, Tapsell and Sabaté^8, Reference Ros^61, Reference Casas-Agustench, López-Uriarte and Ros⁶²⁾. A growing body of evidence from small RCT indicates that nuts, particularly walnuts, have beneficial effects on endothelial function, as ascertained by flow-mediated dilatation studies in the brachial artery⁽Reference Ros, Núñez and Pérez-Heras^63–Reference West, Krick and Klein⁶⁶⁾, reduced circulating levels of endothelial activation molecules⁽Reference Zhao, Etherton and Martin⁶⁷⁾ or inhibition of pro-inflammatory cytokine production by blood mononuclear cells⁽Reference Zhao, Etherton and Martin⁶⁸⁾. While an effect of nut diets on resting blood pressure has been difficult to detect in clinical studies, testing haemodynamic responses to stress has uncovered a beneficial effect of walnuts⁽Reference West, Krick and Klein⁶⁶⁾ and pistachios⁽Reference West, Gebauer and Kay⁶⁹⁾ on blood pressure and total peripheral resistance, suggesting a novel mechanism for the cardioprotective effects of nuts. Improved vascular reactivity in studies with walnuts may be ascribed to their richness in l-arginine, PUFA, particularly α-linolenic acid, and polyphenols⁽Reference Casas-Agustench, López-Uriarte and Ros⁶²⁾.

Nuts being a rich source of antioxidants, it is not surprising that their consumption has been related to improved oxidative status⁽Reference López-Uriarte, Bulló and Casas-Agustench⁷⁰⁾. This has been reported for MUFA-rich nuts, such as almonds, both acutely⁽Reference Jenkins, Kendall and Marchie⁷¹⁾ and after 1 month feeding⁽Reference Jenkins, Kebdall and Marchie⁷²⁾, pistachios⁽Reference Kay, Gebauer and West⁷³⁾ and hazelnuts⁽Reference Orem, Yucesan and Orem⁷⁴⁾, but not for PUFA-rich nuts such as walnuts, albeit no deleterious effects on oxidation have been described⁽Reference Banel and Hu^47, Reference López-Uriarte, Bulló and Casas-Agustench⁷⁰⁾. Studies on nut-feeding have documented reduced circulating concentrations of inflammatory cytokines but no consistent changes of C-reactive protein⁽Reference Banel and Hu^47, Reference Ros⁶¹⁾. A novel mechanism by which walnut consumption may reduce cardiovascular risk has recently been reported: increased cholesterol efflux from cholesterol-laden macrophages exposed to human serum collected after a walnut meal⁽Reference Berryman, Grieger and West⁷⁵⁾. Cholesterol efflux from macrophages is an important step in the anti-atherogenic reverse cholesterol transport pathway. This is a provocative finding that needs to be confirmed in further studies; it suggests that nut (walnut) consumption improves the functionality of HDL, the main lipoprotein particles involved in reverse cholesterol transport. In this sense, another recent report of a PREDIMED substudy showed that the Mediterranean diet enriched with mixed nuts shifted both LDL and HDL subfractions to a less atherogenic pattern, also suggesting improved cholesterol transport capacity after nut diets⁽Reference Damasceno, Sala-Vila and Cofán⁷⁶⁾. In summary, the emerging picture is that nut consumption has positive effects on cardiovascular risk factors well beyond established cholesterol lowering without incurring increases in adiposity.

Cardiovascular effects of nuts in the PREvención con DIeta MEDiterránea trial

The PREDIMED study is a large, parallel group, multi-centre, 5-year RCT of nutrition intervention conducted in Spain. Eligible participants were men, 55–80 years of age, and women, 60–80 years of age, at high risk but no CVD at enrolment. Participants were randomised to three diet groups, two Mediterranean diets, supplemented with either extra-virgin olive oil or mixed nuts, or control diet (advice on a low-fat diet). They received quarterly individual and group educational sessions and, depending on group assignment, free provision of extra-virgin olive oil (one litre/week), mixed Mediterranean nuts (30 g/d: 15 g walnuts, 7·5 g almonds and 7·5 g hazelnuts) or small non-food gifts. Of note, the diets were non-energy restricted and increased physical activity was not promoted; thus, the full effect of dietary intervention alone could be ascertained. The primary end point was an aggregate of major cardiovascular events (myocardial infarction, stroke and cardiovascular death).

Since its inception in October 2003, the results of the trial have provided much evidence on the beneficial health effects of Mediterranean diets enriched with both supplemental foods, culminating with the findings concerning the primary end point⁽Reference Estruch, Ros and Salas-Salvadó²²⁾. The efficacy of the PREDIMED nutritional intervention has generally been similar for both the extra-virgin olive oil and nut-supplemented Mediterranean diets in comparison with the control diet. The main results regarding the Mediterranean diet enriched with mixed nuts are summarised below.

In a study involving the first 772 participants completing intervention for 3 months⁽Reference Estruch, Martínez-González and Corella⁷⁷⁾, results showed that, compared with the control diet, the two Mediterranean diets resulted in significant reductions of both systolic and diastolic blood pressure and total and LDL-cholesterol; increased HDL-cholesterol; decreased fasting glucose levels in diabetic participants and increased insulin sensitivity in those without diabetes; and reduction in circulating levels of inflammatory cytokines. Both intervention diets were also associated with reduced oxidised LDL levels⁽Reference Fitó, Guxens and Corella⁷⁸⁾. The nut-enriched diet reduced fasting TAG levels as well⁽Reference Estruch, Martínez-González and Corella⁷⁷⁾. While results on lipids, glycaemic control and oxidation and inflammation biomarkers were confirmatory of previous data, the blood pressure-lowering effect of the nut-enriched diet was novel. Of note, reduced blood pressure by both PREDIMED Mediterranean diets after intervention for 1 year has recently been confirmed in a 24 h ambulatory blood pressure monitoring substudy⁽Reference Doménech, Roman and Lapetra⁷⁹⁾.

Another substudy investigated both 3-month changes in circulating inflammatory biomarkers and the expression of ligands for inflammatory molecules in circulating monocytes after the intervention⁽Reference Mena, Sacanella and Vázquez-Agell⁸⁰⁾. The findings indicate reductions in both circulating inflammatory mediators and, importantly, reduced monocyte expression of pro-inflammatory ligands after the two Mediterranean diets, thus beginning to unravel the molecular bases for their anti-inflammatory effects. Concurring with these findings, a recent 2-year vascular imaging study in a PREDIMED subcohort detected an anti-atherosclerotic effect of the Mediterranean diets v. the control diet, namely delayed progression of carotid intima-media thickness and plaque, which was more pronounced with the nut-supplemented diet⁽Reference Sala-Vila, Romero-Mamani and Gilabert⁸¹⁾.

The effects of the interventions on the metabolic syndrome status and incident diabetes were secondary end points of the PREDIMED study. In the first 1224 trial participants completing intervention for 1 year, the nut-enriched Mediterranean diet was associated with a 14 % reduction in the prevalence of the metabolic syndrome, which was mainly due to reduced visceral adiposity⁽Reference Salas-Salvadó, Fernández-Ballart and Ros⁸²⁾. Because there were no weight changes, this suggests fat redistribution away from the abdominal compartment. In one of the eleven PREDIMED recruiting centres, annual oral glucose tolerance tests were performed in non-diabetic participants to accurately detect incident diabetes. Results in 418 subjects followed for a mean of 4 years showed that both Mediterranean diets protected against development of diabetes, and the reduction of incident diabetes in the nuts arm was 52 % compared with the control diet⁽Reference Salas-Salvadó, Bulló and Babio⁸³⁾. Again, this beneficial effect took place in the absence of weight loss or increased energy expenditure in physical activity. More recently, the results on diabetes incidence for the full PREDIMED cohort have become available and this time the nut-enriched diet only showed a non-significant 18 % reduction⁽Reference Salas-Salvadó, Bulló and Estruch⁸⁴⁾.

Recently, the results of the PREDIMED trial on the primary end point have been reported, showing for the first time a reduction of incident CVD after long-term consumption of the two Mediterranean diets⁽Reference Estruch, Ros and Salas-Salvadó²²⁾. After a median follow-up of 4·8 years in 7447 participants, those assigned the two Mediterranean diets showed a 30 % reduction in CVD events (myocardial infarction, stroke or cardiovascular death) compared with the control diet. As seen in Fig. 2, the curves of the two Mediterranean diet groups and the control group diverged soon after the beginning of the study. Such an early (and sustained) reduction in cardiovascular events fits with the improvement of intermediate markers of CVD detected after follow-up for only 3 months⁽Reference Estruch, Martínez-González and Corella^77, Reference Fitó, Guxens and Corella^78, Reference Mena, Sacanella and Vázquez-Agell⁸⁰⁾. The nut diet was also associated with a significant 49 % reduction in stroke risk⁽Reference Estruch, Ros and Salas-Salvadó²²⁾. The two intervention diets were associated with non-significant reductions in myocardial infarction and no effect could be discerned on total mortality, although one must keep in mind that the trial was not powered for these outcomes. Of note, the interventions were intended to improve the overall dietary pattern, but the major between-group differences in food intake were for the supplemental items. Thus, nuts were probably responsible for most of the observed benefit in the corresponding Mediterranean diet group. A protective effect of both intervention diets on peripheral arterial disease, another cardiovascular outcome, has also been reported recently⁽Reference Ruiz-Canela, Estruch and Corella⁸⁵⁾.

Fig. 2

Incidence of CVD by intervention group in the PREvención con DIeta MEDiterránea study. Med Diet with extra-virgin olive oil (EVOO): hazard ratios (HR) 0·70 (95 % CI 0·53, 0·91), P= 0·009; Med Diet with nuts: HR 0·70 (95 % CI 0·53, 0·94), P= 0·016. Med Diet, Mediterranean diet. Estruch et al. ⁽Reference Estruch, Ros and Salas-Salvadó²²⁾ copyright© (2013) Massachusetts Medical Society. Reprinted with permission. A colour version of this figure can be found online at http://www.journals.cambridge.org/bjn

The beneficial effects shown in various PREDIMED reports for the Mediterranean diet with nuts on intermediate biomarkers of cardiovascular risk include blood pressure, the lipid profile, insulin resistance, lipoprotein oxidation and systemic inflammation. This diet was also associated with reduction in visceral adiposity, metabolic syndrome and diabetes, as well as delayed atherosclerosis progression. Taken together, these multivariate effects provide a mechanistic explanation for the observed protection of the nut-supplemented diet against CVD. The results of the PREDIMED trial show the full potential of nuts and other healthy foods such as extra-virgin olive oil to improve cardiovascular health.