As population ageing advances in many countries, chronic diseases such as CVD are having an enormous impact on society. CVD has become the leading cause of death worldwide, with an estimated 17·9 million deaths (31 % of all deaths), attributed to it in 2016(1). CHD was the number two cause of death among people aged ≤59 years and number one among those aged ≥60 years(2). Accumulating evidence suggests higher fruit consumption is inversely associated with the risk of CHD(Reference Aune, Giovannucci and Boffetta3). Fruits are rich in vitamins, minerals and dietary fibre. Also, flavonoids in fruit may contribute beneficial cardiovascular effects because of their antioxidant properties(Reference Grassi, Desideri and Di Giosia4,Reference Heiss, Keen and Kelm5) ; this probably occurs through suppressing the formation and progress of atherosclerosis, the major pathology of CHD(Reference Hansson6,Reference Lozano, Naghavi and Foreman7) .
Flavonoids are a large class of polyphenols widely present in many common foods, including fruits, tea, chocolate, nuts and red wine(Reference McCullough, Peterson and Patel8). They are classified into multiple subgroups, including anthocyanidins, flavanones, flavanonols, flavans and isoflavones(Reference Manach, Scalbert and Morand9). These non-energy, non-nutrient polyphenols are secondary metabolites that humans cannot synthesise(Reference Manach, Scalbert and Morand9). They are, however, regarded as indispensable dietary components for humans because of their antioxidative property related to prevention of various diseases associated with oxidative stress(Reference Manach, Scalbert and Morand9). A review of prospective cohort studies showed that dietary flavonoid consumption was inversely associated with mortality from CHD(Reference Peterson, Dwyer and Jacques10).
Although the association between total fruits and CHD risk was well studied, few studies have focused on flavonoid-rich fruits (FRF), in particular strawberries and grapes. Understanding the role of FRF in the development of CHD may have implications in the prevention practice. We, therefore, aimed to verify the association of total and specific FRF consumption with CHD risk, through use of a large prospective cohort study.
Materials and methods
Participants
The Japan Public Health Center-based Prospective Study (JPHC Study) aims to provide evidence on the prevention of non-communicable diseases and on health promotion. Details of the study design have been published(Reference Tsugane and Sawada11). Briefly, in 1990, participants aged 40–59 years from five public health centres (cohort I) and, in 1993, participants aged 40–69 years from six public health centres (cohort II) were recruited. At the fifth year after the first survey, a second follow-up survey was carried out. As consumption of pears, grapes and strawberries was not assessed in the first survey, we treated the second survey that assessed 138 food and beverage items as the study baseline for the present analysis (i.e. 1995 for cohort I and 1998 for cohort II). A total of 103 802 men and women answered the questionnaire at study baseline. We excluded 16 625 of them because they had a previous history of CVD or cancer (n 7012), had an extreme BMI (<14 or >40 kg/m2, n 3854) or had an implausible energy intake (<3347 kJ/d (<800 kcal/d) or >15062 kJ/d (>3600 kcal/d), n 5759). Ultimately, 87 177 men and women were eligible for the present analysis on FRF consumption and CHD risk (Fig. 1). The present study was approved by the institutional review boards of the National Cancer Center and Osaka University. Informed consents were obtained from individuals or from community leaders in some areas.
Fig. 1. Flow chart for participant selection.
Dietary assessment
Dietary assessment was based on a validated FFQ(Reference Ishihara, Sobue and Yamamoto12,Reference Sasaki, Matsumura and Ishihara13) . Participants were asked to report their frequencies of food and beverage consumption during the preceding 12 months. The FFQ included fourteen fruit items: apples, pears, oranges, other citruses, strawberries, grapes, papayas, Japanese persimmons, melons, watermelons, peaches, kiwifruits, pineapples and bananas. FRF, defined as fruits with high content of total flavonoids >50 mg/100 g(Reference McCullough, Peterson and Patel8), included apples, pears, oranges, other citruses, strawberries and grapes in the present study. Total FRF consumption was defined as the total intakes of the above fruits. Other fruits were defined as the remaining fruits in FFQ (papayas, Japanese persimmons, melons, watermelons, peaches, kiwifruits, pineapples and bananas). The question about fruit consumption included nine predefined frequency categories ranging from never to ≥7 times per d. Each respondent’s fruit intake was calculated based on the frequency and portion size (85 g for apples, 80 g for pears, 140 g for oranges, 75 g for strawberries, 100 g for grapes and 50–130 g for other fruits). The FFQ was previously found to have acceptable reproducibility and validity(Reference Ishihara, Sobue and Yamamoto12,Reference Sasaki, Matsumura and Ishihara13) .
Covariate assessment
Data on socio-demographic information, self-rated physical and mental health, disease history, medication use, smoking status, drinking status and physical activities were obtained through self-administered questionnaires. BMI was calculated using self-reported height and weight (BMI = weight (kg)/(height (m))2).
Outcome assessment
The outcome of this analysis was the first incidence of CHD. Cases of CHD, including nonfatal myocardial infarction and coronary deaths, were diagnosed in accordance with the criteria of the Monitoring Trends and Determinants of Cardiovascular Disease (MONICA) project(Reference Tunstall-Pedoe, Kuulasmaa and Amouyel14), with the use of electrocardiograms, cardiac enzymes and/or autopsy. Cases were diagnosed in hospitals that were registered within the study areas. A total of 3194 participants (3·7 %) who moved out of the registered study areas were not able to follow and they were treated as loss to follow-up. Physicians, hospital workers or investigators, who were not aware of the baseline data, were responsible for identifying CHD cases by reviewing the medical records.
Statistical analysis
Person-years of follow-up were calculated from the date of the return of the second survey questionnaire until the date of CHD incidence, death or 31 December 2009, for cohort I, and 31 December 2012, for cohort II, whichever came first. We used a Cox proportional hazards regression model to compute hazard ratios (HR) for CHD incidence based on the quintiles of FRF consumption. We combined men and women in the main analysis and conducted a stratified analysis by sex. We treated the group with the lowest consumption as the reference. All HR were age-adjusted in model 1. In model 2, we further adjusted for study areas, sex, occupation (unemployed, white-collar worker, blue-collar worker, other or missing), BMI (<18·5, 18·5–20·9, 21–22·9, 23–24·9, 25–29·9 or ≥30 kg/m2), use of medication for hypertension and hypercholesterolaemia (yes or no), history of diabetes (yes or no), smoking status (never, past, current <1 or ≥1 pack/d or missing), alcohol use (never, 0–22·9, 23–45·9, 46–68·9, ≥69 g/d or missing) and physical exercise (never, 1–3 times/month, 1–2, 3–4, ≥5 times/week or missing). In model 3, we further adjusted for dietary factors including coffee intake (never, 1–2, 3–6 cups/week, 1, ≥2 cups/d or missing), green tea intake (never, 1–2, 3–6 cups/week, 1, 2–3, ≥4 cups/d or missing) and quintile intakes of total energy, seafood, red meat, processed meat, milk, soya foods, vegetables and other fruits. We conducted trend tests by treating FRF consumption as a continuous variable. We also examined the associations between specific FRF intake and CHD risk. Apples and pears were combined as apples/pears and oranges and other citrus were combined as citrus fruits. As strawberry and grape consumptions were relatively lower, participants were divided into quartile groups. A sensitivity analysis was performed to examine the association of energy-adjusted FRF consumption with CHD risk by using the residual method. All analyses were performed using SAS 9.4 software (SAS Institute Inc.). All P values were two-sided, with P < 0·05 considered statistically significant.
Results
Table 1 shows the baseline characteristics of the study population based on the quintiles of FRF consumption. Median FRF intakes were 13·4, 43·0, 85·4, 143·4 and 289·8 g/d for the quintiles. People with higher consumption were slightly older, more likely to be women and less likely to smoke and drink than those in the lowest quintile. Additionally, higher FRF consumption was associated with higher intakes of seafood, vegetables, soya foods, green tea, other fruits and total energy.
Table 1. Baseline characteristics according to quintiles (Q) of flavonoid-rich fruit consumption*
(Numbers and percentages; mean values and standard deviations)
* Values are age-adjusted. Continuous variables are expressed as mean values and standard deviations and categorical variables are expressed as percentages. Flavonoid-rich fruit consumption is calculated as the total consumption of apples, pears, citrus fruits, strawberries and grapes. Other fruits included papayas, Japanese persimmons, melons, watermelons, peaches, kiwifruits, pineapples and bananas. The percentages of missing data were as follows: 3·6 % for occupation, 3·5 % for smoking, 0·1 % for drinking, 3·9 % for physical exercise, 3·6 % for green tea consumption and 4·9 % for coffee consumption.
† To convert energy values from kcal to kJ, multiply by 4·184.
During a mean follow-up of 13·2 years, we identified 1156 incident CHD cases from the 87 177 participants (1·3 %). In the age-adjusted model, participants in the highest quintile of FRF consumption (median 289·8 g/d), compared with those in the lowest (median 13·4 g/d), had a lower risk of developing CHD (age-adjusted HR = 0·58 (95 % CI 0·48, 0·70)) (Table 2). The association was attenuated but remained significant after adjusting for potential risk factors including study area, sex, BMI, occupation, smoking, physical exercise, drinking, hypertension and hypercholesterolaemia medication use, and diabetes history (model 2). Further adjustment for dietary factors yielded similar results to those with model 2; the fully multivariable-adjusted HR were 0·93 (95 % CI 0·77, 1·11), 0·91 (95 % CI 0·75, 1·11), 0·84 (95 % CI 0·67, 1·04) and 0·78 (95 % CI 0·62, 0·99) for the second through fifth quintiles, respectively. In the analyses for specific FRF consumption associated with CHD risk, we observed a significant inverse association for higher citrus fruit consumption and a borderline significantly inverse association for higher strawberry consumption, though no association for higher apple/pear or grape consumption after adjustment for all covariates in model 3. Sensitivity analyses using energy-adjusted data yielded similar results, for example, the fully adjusted HR of the highest group were 0·81 (95 % CI 0·66, 0·995) for total FRF and 0·82 (95 % CI 0·68, 0·98) for citrus fruits.
Table 2. Flavonoid-rich fruit consumption and risk of CHD among the Japanese population
(Hazard ratios (HR) and 95 % confidence intervals)
Q, quintile.
* Model 1: adjusted for age (87 177 participants without missing data). Model 2: model 1 and further adjusted for the study area, sex, BMI, smoking, drinking, physical exercise, occupation, medication use for hypertension and hypercholesterolaemia, and history of diabetes (79 507 participants without missing data). Model 3: model 2 and further adjusted for dietary intakes of seafood, red meat, processed meat, milk, soya food, green tea, coffee, vegetables, other fruits and total energy (75 519 participants without missing data). Flavonoid-rich fruit consumption is calculated as the total consumption of apples, pears, citrus fruits, strawberries and grapes. Other fruits included papayas, Japanese persimmons, melons, watermelons, peaches, kiwifruits, pineapples and bananas.
In the subgroup analysis stratified by sex, significant associations of total FRF consumption and citrus fruit consumption with CHD risk were observed in women but not in men (Fig. 2). However, the differences were not statistically significant (all P for interaction ≥ 0·15).
Fig. 2. Associations between consumptions of total and specific flavonoid-rich fruits (FRF) and risk of CHD stratified by sex. Hazard ratios (HR, highest v. lowest group) were adjusted for age, study area, BMI, smoking, drinking, physical exercise, occupation, medication use for hypertension and hypercholesterolaemia, history of diabetes and dietary intakes of seafood, red meat, processed meat, milk, soya food, green tea, coffee, vegetables, other fruits and total energy. All P for interaction ≥ 0·15.
As for total fruits, the fully adjusted HR of CHD were 1·07 (95 % CI 0·89, 1·27), 1·06 (95 % CI 0·88, 1·29), 1·12 (95 % CI 0·91, 1·38) and 0·98 (95 % CI 0·78, 1·23) for the second, third, fourth and fifth quintiles compared with the lowest quintile.
Discussion
In this large prospective cohort study, we examined the association of total and specific FRF consumption with CHD risk. We observed that higher FRF consumption, particularly consumption of citrus fruits, was associated with a lower risk of developing CHD. Although the associations appeared to be stronger in women, they were not significantly modified by sex.
The potential cardiovascular effects of FRF may be partially attributed to the antioxidative features of flavonoids and their endothelial control of inflammation, vasodilatation, vascular homeostasis and thrombosis(Reference Vita and Keaney15). Flavonoids have been shown to resist oxidative stress by modifying the function of enzymes that modulates superoxide anion production, such as with xanthine oxidase(Reference Hanasaki, Ogawa and Fukui16) and protein kinase C(Reference Mansuri, Parihar and Solanki17). They also work as a free radical scavenger of superoxide(Reference Korkina and Afanas’ev18). The process both inhibits oxidised LDL in vitro (Reference Kerry and Abbey19) and antagonises the effect of nitric oxide inactivation(Reference Shutenko, Henry and Pinard20). Flavonoids may also protect nitric oxide from superoxide-driven inactivation or they can directly scavenge nitric oxide molecules under certain conditions(Reference Duarte, Francisco and Perez-Vizcaino21). Moreover, flavonoids exert an anti-inflammatory effect mediated through signal transduction pathways rather than antioxidant capacity, which has shifted focus from antioxidant effects to enzyme-induced pathways(Reference Leiherer, Mundlein and Drexel22). For example, the flavonols quercetin and myricetin may inhibit NF-κB activity through down-regulating NF-κB pathway signalling(Reference Holmes-McNary and Baldwin23,Reference Ruiz and Haller24) , thus reducing the inflammatory reaction in cells(Reference Yamamoto and Gaynor25).
In addition to flavonoids, other nutrients in fruits, vitamin C in particular, may have contributed to the observed association. Vitamin C has the abilities to prevent LDL from atherogenic modification(Reference Retsky, Freeman and Frei26) and counteract the damage caused by existing highly bio-reactive-oxidised LDL(Reference Siow, Richards and Pedley27). Vitamin C can also decrease the bioavailability of superoxide, relieving superoxide-mediated nitric oxide inactivation(Reference Jackson, Xu and Vita28).
A meta-analysis of prospective studies confirmed the inverse association of apple and pear consumption with CHD risk (n 10, pooled HR = 0·85 (95 % CI 0·79, 0·93))(Reference Aune, Giovannucci and Boffetta3). The absence of such an association in the Japanese population was, however, not surprising. Database from the United States Department of Agriculture(29) has shown the flavonoid composition of fruits, though the specific compositions of flesh and peel have not been shown. Recent research has shown apple peel possesses more flavonoid than apple flesh(Reference Vieira, Borges Gda and Copetti30), thus indicating apple peel’s dietary benefit(Reference Gonzalez, Donoso and Sandoval31,Reference Tian, Wu and Zhang32) . In Japan, unlike in many countries, it is common to peel apples before eating them. The loss of abundant flavonoids and other nutrients in the peel may account for the non-significant association observed in our study.
In line with findings from previous studies, we found higher consumption of citrus fruits was associated with a decreased risk of CHD. The previously mentioned meta-analysis pooled the results of fourteen cohort studies and supported an inverse association between citrus fruit consumption and CHD risk, showing a combined HR of 0·91 (95 % CI 0·86, 0·96)(Reference Aune, Giovannucci and Boffetta3). Moreover, individual study results included in that meta-analysis did not significantly differ (P for heterogeneity = 0·69)(Reference Aune, Giovannucci and Boffetta3). Citrus fruits are rich in flavonoids and vitamin C, and they were the most consumed FRF in this Japanese population (Table 1). In the Iowa Women’s Health Study with 15 years of follow-up, a significant inverse association was observed between flavanones – the major flavonoid class existing in citrus fruits – and CHD mortality for the highest v. the lowest quintile (HR = 0·78 (95 % CI 0·65, 0·94))(Reference Mink, Scrafford and Barraj33).
To date, the evidence linking strawberry and grape consumption to CHD risk is very limited. We observed a borderline significant association between strawberry consumption and lower CHD risk. Strawberries have abundant vitamin C and polyphenols such as anthocyanin, which are suggested to have antioxidant and anti-inflammatory activity associated with beneficial effects on improving lipid profile, blood pressure, endothelial function and insulin resistance(Reference Giampieri, Alvarez-Suarez and Battino34,Reference Giampieri, Forbes-Hernandez and Gasparrini35) . However, two cohort studies found no association of strawberry consumption with myocardial infarction incidence or CHD mortality(Reference Mink, Scrafford and Barraj33,Reference Sesso, Gaziano and Jenkins36) . We observed no association of grape consumption with CHD risk, which was consistent with results from previous studies conducted in Western countries(Reference Mink, Scrafford and Barraj33,Reference Cassidy, Mukamal and Liu37–Reference Lin, Rexrode and Hu39) .
Strengths of the present study included its prospective community-based cohort design and a large sample size. Several limitations, however, should also be noted. First, as an observational study, the risk of confounding bias should be considered when interpreting the results. As shown in Table 1, participants with higher FRF consumption were more likely to have a healthier lifestyle, which includes less smoking and higher intakes of vegetables, seafood and other fruits. It was therefore unsurprising that adjustment for age only (model 1) always led to statistically significant HR while further adjustments for covariates in model 2 and model 3 resulted in attenuated associations between FRF consumption and risk of CHD. Other unmeasured factors associated with FRF consumption, such as income and education level, may at least partly explain the observed associations. Second, because the diet assessment was based on a self-reported questionnaire, the influence of measurement errors cannot be completely ruled out; these were probably non-differential and could attenuate the association towards the null. Third, the questionnaire addressed frequently consumed fruits in Japan but did not include FRF such as blueberries, which have been shown to have potential effects on preventing CVD(Reference Cassidy, Mukamal and Liu37). Finally, an underestimation of CHD occurrence was inevitable due to 3·7 % of loss to follow-up, yet such a proportion of loss to follow-up was not likely to change our findings materially.
In conclusion, the findings from this prospective cohort study suggest that higher FRF consumption, especially citrus fruits, is associated with a lower risk of developing CHD.
Acknowledgements
Members of the JPHC study group are listed in https://epi.ncc.go.jp/en/jphc/781/8233.html
This work was supported by the JSPS KAKENHI (A18H06391, T19K214700) to Dr. Dong. The JPHC study was supported by National Cancer Center Research and Development Fund (23-A-31(toku) and 26-A-2) (since 2011) and a Grant-in-Aid for Cancer Research from the Ministry of Health, Labour and Welfare of Japan (from 1989 to 2010). The funders had no role in the design and conduct of the study; collection, management, analysis and interpretation of the data; preparation, review or approval of the manuscript; or the decision to submit the manuscript for publication.
J.-Y. D. designed the study, analysed the data and edited the manuscript. Y. Y. drafted the manuscript. All authors conducted the technique review and edited the manuscript. J.-Y. D. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. The authors declare that there are no conflicts of interest.
