Systematic Review on Polyphenol Intake and Health Outcomes: Is there Sufficient Evidence to Define a Health-Promoting Polyphenol-Rich Dietary Pattern?

Growing evidence support association between polyphenol intake and reduced risk for chronic diseases, even if there is a broad debate about the effective amount of polyphenols able to exert such protective effect. The present systematic review provides an overview of the last 10-year literature on the evaluation of polyphenol intake and its association with specific disease markers and/or endpoints. An estimation of the mean total polyphenol intake has been performed despite the large heterogeneity of data reviewed. In addition, the contribution of dietary sources was considered, suggesting tea, coffee, red wine, fruit and vegetables as the main products providing polyphenols. Total flavonoids and specific subclasses, but not total polyphenols, have been apparently associated with a low risk of diabetes, cardiovascular events and all-cause mortality. However, large variability in terms of methods for the evaluation and quantification of polyphenol intake, markers and endpoints considered, makes it still difficult to establish an evidence-based reference intake for the whole class and subclass of compounds. Nevertheless, the critical mass of data available seem to strongly suggest the protective effect of a polyphenol-rich dietary pattern even if further well targeted and methodologically sound research should be encouraged in order to define specific recommendations.


Introduction
The possibility to develop dietary guidelines for the intake of food bioactives with health promoting effects can be of utmost importance to try to evolve the concept of adequate nutrition to that of optimal nutrition. Clearly, this implies at least 2 levels of knowledge: (1) the availability of reliable data of food composition and food intake to estimate exposure to food bioactives and (2) the capacity to assess the amount needed to exert the protective activity.

Introduction
The possibility to develop dietary guidelines for the intake of food bioactives with health promoting effects can be of utmost importance to try to evolve the concept of adequate nutrition to that of optimal nutrition. Clearly, this implies at least 2 levels of knowledge: (1) the availability of reliable data of food composition and food intake to estimate exposure to food bioactives and (2) the capacity to assess the amount needed to exert the protective activity.
Polyphenols have been suggested to exert a plethora of biological activities including antioxidant, anti-inflammatory, anti-microbial, anti-proliferative, pro-apoptotic activity and hormonal regulation capacity [1]. There is also increasing evidence that long-term intake can have favorable effects on the incidence of several cancers and other chronic diseases, including cardiovascular disease (CVD), type II diabetes, and neurodegenerative diseases [2]. More recently research has been focused on the impact of polyphenols on healthy aging and/or age-related diseases [3].
The emerging evidence, obtained through both animal models and human studies, on the direct and indirect role of polyphenols in the modulation of metabolic and functional features of the host, has enhanced the interest for an estimation of polyphenol intake in the general population or in at risk target groups. In addition, the assessment of specificity in the protective properties of the single polyphenol classes/compounds ( Figure 1) has been increased in the last years favored by the improvement of dedicated food databases (i.e., Phenol-Explorer, USDA database) reporting more accurate and detailed polyphenols composition and considering factors affecting the intake such as the "retention factors" (i.e., the loss or gain of a compound during food processing). Despite the transformation of food intake data into polyphenol intake remains still a critical, even if improved, step of the process, the accuracy of self-reported methods to evaluate dietary patterns is often limited. In particular, it has been suggested that the notion that fruit and vegetables intake represents the main dietary sources of polyphenols could be over-reported [4]. Finally, as far as polyphenols are concerned, the low bioavailability and extensive metabolism demonstrated in numerous studies makes it difficult to clearly state recommendations on intake. Nevertheless, the analysis of polyphenol intake data registered in several target population with different dietary patterns and lifestyle/exposure may help better understanding whether it is possible to identify a range of intake apparently associated to an overall reduced risk.
To this aim a comprehensive updated review on data and tools/methods used for the estimation of polyphenol intake was performed by considering differences in total and subclasses intake depending on factors related to dietary habits. In addition, main results on the association among polyphenol intake and specific endpoints of disease risk have been taken into account, when available, to suggest possible recommendation.

Search Strategy and Study Selection
A literature search of all English language studies published was performed using PubMed (http://www.ncbi.nlm.nih.gov/pubmed), and EMBASE (http://www.embase.com/) databases (updated December 2018) with the addition of other scientific papers of relevance fount in web sources or in previously published reviews. The search terms and strategy used for the study selection were: polyphenols OR flavonoids OR anthocyanins OR flavanols OR flavanones OR flavones OR flavonols OR isoflavones OR proanthocyanidins OR phenolic acids OR hydroxycinnamic acids OR hydroxybenzoic acids OR lignans OR stilbenes AND intake. Human studies were used as further criteria of literature search. The search was limited to the last 10 years of publication. Three independent reviewers (S.B., M.M. and M.T.) conducted the literature search in the scientific databases and assessed and verified the eligibility of the studies based on the title and abstract. Disagreement between reviewers was resolved through consultation with a third reviewer (P.R. or C.D.B.) to reach a consensus. Inclusion criteria: (i) prospective, cohort and case-control studies analysing/estimating dietary total/classes/individual polyphenol intake; (ii) studies reporting association between dietary total/classes/individual polyphenol intake and endpoints of disease risk and mortality; (iii) studies published from January 2008 to December 2018. The exclusion criteria were: (i)-dietary intervention studies; (ii)-studies measuring polyphenols intake through urine excretion; (iii)-studies performed in in-vitro or in animal models; (iv)-studies reporting data on polyphenol intake from supplements (not food related); (v)-studies evaluating the association between polyphenol intake and cancer risk/mortality (numerous systematic reviews and meta-analysis have been recently performed); (vi)-published articles in a language different from English and with no accessible translation.

Data Extraction
For the papers meeting the inclusion criteria, the full text was retrieved, analysed and summarized in Tables. Data extraction was performed by three independent reviewers (S.B., M.M., M.T., P.R. and C.D.B.). The following information was collected: (i) first author name and year of publication; (ii) study design; (iii) number and subjects' characteristics; (iv) country; (v) tools used for estimating dietary polyphenols intake; (vi) polyphenol database source; (vii) overall results. For the studies evaluating the association with disease risk or mortality this information was included in the table. Additional revisions of contents have been performed by other reviewers (N.H.L., B.K. and B.C.).

Study Selection
The study selection process according to PRISMA guidelines is reported in Figure 2. A total of 3004 records were identified from the database search (PubMed and EMBASE) and other sources. After removing 48 duplicate articles, 2956 studies were screened and 2566 were excluded based on title and/or abstract. The full text of eligible studies (n = 390) was read; 299 studies were excluded because not meeting the inclusion criteria (n = 282) or not of interest/pertinent (n = 17). At the end of the selection process, 91 papers were included.

Study Characteristics
The main characteristics of the 91 included studies are reported in Tables 1-4; 45 studies focused only on polyphenol intake in specific target populations, 24 studies assessed the association between polyphenols and cardiovascular/diabetes risk (1 study included also data on CV mortality), 9 studies focused specifically on the association with mortality for cardiovascular and all other events, while 13 studies evaluated the association between polyphenol intake with others outcomes (e.g., frailty, bone fractures).

Study Characteristics
The main characteristics of the 91 included studies are reported in Tables 1-4; 45 studies focused only on polyphenol intake in specific target populations, 24 studies assessed the association between polyphenols and cardiovascular/diabetes risk (1 study included also data on CV mortality), 9 studies focused specifically on the association with mortality for cardiovascular and all other events, while 13 studies evaluated the association between polyphenol intake with others outcomes (e.g., frailty, bone fractures). Table 1 shows reported data from literature focused on polyphenols intake. A total of 45 studies were found and analyzed . Most of the studies were performed in Europe, North America and Asia ( Figure 3A). The researches ( Figure 3B) were carried out in the adult + older population (63%) or only adults (20%), while few studied reported data specifically in older subjects (7%), in children and adolescents (7%); the dietary intake of polyphenols was assessed generally through 24-h dietary records (24-h DR; 56%) and food frequency questionnaire (FFQ; 31%) as reported in Figure 3C. The main scientific databases ( Figure 3D) used for the estimation of polyphenol intake were USDA (22%) and Phenol-Explorer (PE; 20%). However, most of the studies combined USDA with PE and other databases and/or scientific sources (24%). Total polyphenol intake for the overall population was estimated to be about 900 mg/day; this value varied according to differences in target groups of subjects. The main food sources of polyphenols were represented by tea, coffee, red wine, fruit and vegetables.  Table 1 shows reported data from literature focused on polyphenols intake. A total of 45 studies were found and analyzed . Most of the studies were performed in Europe, North America and Asia ( Figure 3A). The researches ( Figure 3B) were carried out in the adult + older population (63%) or only adults (20%), while few studied reported data specifically in older subjects (7%), in children and adolescents (7%); the dietary intake of polyphenols was assessed generally through 24-h dietary records (24-h DR; 56%) and food frequency questionnaire (FFQ; 31%) as reported in Figure 3C. The main scientific databases ( Figure 3D) used for the estimation of polyphenol intake were USDA (22%) and Phenol-Explorer (PE; 20%). However, most of the studies combined USDA with PE and other databases and/or scientific sources (24%). Total polyphenol intake for the overall population was estimated to be about 900 mg/day; this value varied according to differences in target groups of subjects. The main food sources of polyphenols were represented by tea, coffee, red wine, fruit and vegetables.   The total intake of flavonols and flavones was higher in men than in women. Gender and above all age were independent predictors for total flavonols and flavones intake. Main food sources were vegetables (61%) and fruits (36%) while tea was only a minor source

Dietary Intake of Polyphenols
Mean intake of polyphenols was three times higher in men from Denmark than in women from Greece. Stratifying by region, mean of total polyphenols intake was in non-MED countries due to the higher intake of phenolic acids. The study showed a large heterogeneity in both the nature of polyphenols and levels of intake across the countries due to different habits and socio-demographics status Non-alcoholic beverages Fruits Large regional differences, both in the type of flavonoids consumed and the distribution of intake. Intakes of anthocyanidins (in particular cyanidin) and flavanones (in particular hesperetin) were highest in the Northern Region, in particular in Finland. Within the Central Region, there was also a large variability of intake between countries. While overall flavonoid intake in Ireland was the highest in Europe, the intake of anthocyanidins was the lowest overall, and intake of flavanones was also very low. France was included in the Southern Region as dietary intake was more comparable with intake in Italy and Spain. However, there are some important differences, and the intake of flavan-3-ols and anthocyanidins in France is considerably higher than in the other countries of the Southern Region A positive association between flavonoid intake and dietary quality suggest that a diet high in flavonoids is synonymous with greater compliance with national guidance. Individuals with higher flavonoids intake not only consume more fruit and vegetables but also eat more healthfully The present study showed that a population of elderly Japanese (mostly men) consumed higher amounts of polyphenols than previous data in Japanese adults, and coffee and green tea were the largest sources of polyphenols in their daily life Significant differences in mean adjusted total polyphenol intakes were observed between dietary patterns. 34% of the participants reported coffee consumption in the FFQ.
In the group of non-coffee consumers vegans reported the highest intake of total polyphenols followed by pesco-vegetarians, lacto-ovo vegetarians, semivegetarians and non-vegetarians.
In the group of coffee consumers non-vegetarians reporting the highest intakes, followed by vegans, semi-vegetarians, pesco-vegetarians and lacto-ovo-vegetarians  The polyphenol intake was three times lower than the estimated value compared with other countries probably due to sociodemographic differences and food choices. Older subjects (>60 y) consumed more flavonoids and tyrosol than adults (20-59 y) and also more fruits.  Substantial within-individual variation and between individual variation was documented for both total flavonoid intake and intake of flavonoid subclasses. The within-individual variation was in the range 80-140% while the between individual variation was in the range 60-117%. It is speculated that a minimum of 6-day weighed food records is necessary to obtain a reliable estimate of flavonoid intake.

Polyphenol Intake and Cardiovascular Diseases/Diabetes Risk
In Table 2 the results of studies that examined the association between polyphenol intake and cardiovascular diseases risk are reported . Seven out of 24 studies were conducted in United States (US), 2 in South America, 12 in Europe, 3 in Asia ( Figure 4A). Most of the studies were carried out in the adult population-including older subjects (63%) while the remaining studies were performed in adult population (37%) i.e., aged < 65 years ( Figure 4B).   ↓ LDL associated with higher flavanones intake in the group with diabetes ↓ LDL associated with higher flavan-3-ols, and flavanones intake in the group without diabetes ↓ LDL associated with lower polyflavonoids intake in the group without diabetes ↑ HDL associated with higher anthocyanidins and flavan-3-ols intake in the group without diabetes ↓ HDL associated with lower polyflavonoids intake in the group without diabetes There was no relationship between HDL and flavonoids for the group with diabetes.
The main databases used were USDA (42%) and PE (25%). Three studied combined USDA and PE, while the rest of the studies evaluated polyphenol intake with different databases alone or in combination such as Epic Nutrient database, EuroFIR, U.K. Food Standard Agency, Flavonoid Korean Database ( Figure 4D).
The association between polyphenol intake and cardiovascular disease risk and diabetes was evaluated by considering several outcomes such as: HDL-cholesterol, triacylglycerols (TAGs), TAG: HDL-cholesterol ratio, HOMA-IR (Homeostatic Model Assessment of Insuline Resistence), Body Mass Index (BMI), cardiovascular events (CV events), stroke events, hypertension and type 2 diabetes (T2D).
On the whole, 12 studies reported an inverse association between polyphenol intake and CV events. In some studies a significant decreased CV risk was observed at the highest quartile of total polyphenol intake (1170 mg/day for Spain and 2632 mg/day for Poland) [57,69] while no effect was demonstrated in other studies performed in Spain and Iran (1248 mg/day and 2459 mg/day respectively) [72,73]. Ten studies evaluated the association with polyphenol subclasses, mainly total flavonoids but only 3 found a significant inverse association with CV events [52,67,72] with intake ranging from 115 to 944 mg/day.
As regard T2D, 1 study performed in Poland showed an increased protection for total polyphenol intake higher than 2632 mg/day while mixed results were found in the other studies focused on total flavonoids and/or subclasses only in some cases able to demonstrate significant T2D risk reduction [53,55,61,62]. Finally, 1 study [67] reported an inverse association for both CV and T2D with the highest quartile of total flavonoids (585 mg/day).

Polyphenols Intake and all-Cause/Cardiovascular Mortality
In Table 3 the association between polyphenol intake and all-cause mortality is reported with a specific focus on cardiovascular mortality. A total of 10 studies [59,[74][75][76][77][78][79][80][81][82] (Figure 5A) were found; most of the them (50%; 5 out of 10) were performed in Europe (Spain, Italy and The Netherland), 2 in USA, 2 in Australia and 1 was performed including USA, Canada and Australia. Five out of 10 trials (50%) involved older subjects (> 65 years), 3 studies were performed in adults while 2 trails included both adult and older subjects ( Figure 5B). The food intake was assessed mainly by FFQ (60%; 6 out of 10 studies); however, some studies (30%) associated FFQs with other tools for the evaluation of food intake (i.e., computerized dietary history questionnaire). One study combined FFQ with EPIC questionnaire ( Figure 5C). The evaluation of polyphenol intake was estimated by USDA database (30%; 3 out of 10 studies), or a combination of USDA with others database (40%), or USDA with PE (20%; 2 out of 10 studies). When polyphenol content of specific food-products was missing in available databases, data were obtained from the literature. One study estimated polyphenol intake, in particular monomeric flavan-3-ol, by considering their content in 120 commonly consumed plant foods and beverages obtained by combining results from reverse-phase HPLC and data from literature ( Figure 5D).
Overall, one study that investigated the association with total polyphenol intake and all-cause mortality failed to demonstrate a significant effect [75]. Similar findings were also reported by considering the association between total flavonoids and CV mortality [59]. On the contrary, a reduction of mortality risk for cardiovascular events and all-cause mortality was associated with total flavonoid intake in the highest quintiles ranging from 360 mg/day [78] to about 800 mg/day [80]. The impact of the single subclasses has been evaluated in some of the studies, but the effects were conflicting depending on the subject's characteristics (i.e., age, sex) and cause of mortality. Generally, the models adjusted for the age, as confounding factor, reported a protection also for specific flavonoid subclasses such as isoflavones, flavan-3-ols, flavones. The effects in some cases were found both in women and men. However, generally adjustments for the different confounding factors (i.e., BMI, smoking and alcohol habits, energy intake, physical activity, medications, etc. affected the significance of the associations. Overall, one study that investigated the association with total polyphenol intake and all-cause mortality failed to demonstrate a significant effect [75]. Similar findings were also reported by considering the association between total flavonoids and CV mortality [59]. On the contrary, a reduction of mortality risk for cardiovascular events and all-cause mortality was associated with total  Subjects with high total flavonoid consumption (median 512.5 (≥359.7) mg/day) showed a low risk of death (−18%) in both the sex.Inverse association for high anthocyanidin, flavan-3-ol, flavones, flavanol and proanthocyanidin intake by considering women + men. Ischemic heart disease mortality Age-adjusted model: Inverse association for high anthocyanidin and flavone intake in both the sex. Inverse association for high total flavonoid intake in men and women + men; high flavanone intake in women + men; high flavanol intake in women and women + men; high proanthocyanidin intake in women + men; high isoflavone intake in men and women + men. Multivariable-adjusted model 3 : Inverse association for high flavone intake in women and women + men Stroke mortality Age-adjusted model: Inverse association for high total flavonoid intake in men, and high flavones intake in men and women + men. Multivariable-adjusted model 3 : Inverse association for high total flavonoid intake in men   Unadjusted model: Subjects with high intake of total flflavonol (>35 mg/day), flflavan3-ol (>563 mg/day), flavone (>3 mg/day) and flavanone (>61 mg/day) showed a reduced risk of atherosclerotic vascular disease mortality. Age-and energy-adjusted model and multivariate-adjusted model 4 : Subjects with high intake of total flflavonol (>35 mg/day), flflavan3-ol (>563 mg/day) showed a reduced risk of atherosclerotic vascular disease mortality No association was observed for the other flavonoid subclasses Tea contributed 59% of total flavonoid intake; the major contributors were flavonols (65%) and flavan-3-ols (93%). Multivariate-adjusted model 4 : Subjects with high intake of flavonols derived from tea and non-tea sources (≥12 mg/day and ≥27 mg/d, respectively) showed a low risk of atherosclerotic vascular disease mortality. Monomeric flavan-3-ol contents of 120 commonly consumed plant foods and beverages were determined with the use of reverse-phase HPLC with ultraviolet and fluorescence detection. Coronary heart disease mortality Subjects with high epicatechin intake (>18 mg/day) showed a low (−38%) risk of CHD mortality Cardiovascular disease mortality Subjects with high epicatechin intake (>18 mg/day) showed a low (−46%) risk of CVD mortality in men with prevalent CVD but not in men who were free of CVD The major dietary sources of epicatechin intake were tea (7.8 mg/day; 51% of total epicatechin intake), apples (4.3 mg/day; 28% of total epicatechin intake), cocoa (1.1 mg/day; 7% of total epicatechin intake), and other sources (2.0 mg/day; 13% of total epicatechin intake)  Data reported as polyphenol antioxidant content (PAC)-score 6 (−28 to 28) Risk for all-cause mortality Women: low risk for high intake of flavones (>1.12 mg/day), flavanones (>46.5 mg/day), isoflavones (>32.7 mg/day), and lignans (>116.1 mg/day) had a low risk. After adjustments for potential confounders (model 4) 7 : the effects remained significant for Q4  1 BMI, smoking status, menopausal status, family history of diabetes/cancer/myocardial infarction, multivitamin supplement use, aspirin use, race, type 2 diabetes, hypercholesterolaemia, hypertension, physical activity, alcohol consumption and energy intake and the Alternative Health Eating Index (minus alcohol) score. 2 Cox proportional hazards regression models were stratified by center, age (1 year) and sex and adjusted for BMI, education level, physical activity, tobacco smoking, alcohol lifetime, total energy, vitamin C and fiber intake. 3 Age, smoking, beer and liquor intake, history of hypertension, history of cholesterol, family history of myocardial infarction, BMI, physical activity, energy intake, aspirin use, hormone replacement therapy (in women only), and sex (in combined model only) by using Cox proportional hazards regression. 4 age, previous CVD, previous diabetes, energy expended in physical activity and history of smoking. 5 Table 4 shows the associations between polyphenol intake and other outcomes in a total of 13 studies [83][84][85][86][87][88][89][90][91][92][93][94][95]. The associations were evaluated for endothelial function (1 study), kidney function (1 study), bone health (i.e., bone mineral density, frailty and fractures; 3 studies), eyes health (i.e., cataract and macular degeneration; 2 studies), physical performance decline (1 study), dementia (1 study), cognitive decline (1 study) and pubertal development (1 study).

Polyphenols Intake and other Outcomes
Six out of 13 studies (46%) were performed in Europe, 3 in Australia, 2 in the USA and in Asia Figure 6A). Over than a half of the studies (58%) were carried out in the older population while 33% included adult and older subjects. 1 study was performed only in adults and 1 in adolescents ( Figure 6B). The most frequent tools used for the evaluation of the diet were the FFQs (77%; 10 studies), 1 study used a 24-h DR while 2 studies combined FFQs with other tools ( Figure 6C). Half of the studies (50%) used USDA database, or a combination of USDA with PE (3 studies) or USDA with other databases (2 studies). Only one study performed the estimation using PE, while one study used a different specific database for the calculation of polyphenol intake ( Figure 6D). An overall association between high intake of polyphenols and subclasses, and different outcomes was observed. Conversely, in the InCHIANTI study urinary total polyphenols, but not total dietary polyphenols, were associated with a lower probability of frailty or pre-frailty [86] and cognitive decline [95]. Flavonoids have been associated with a higher endothelial function (>640 mg/day) [83], a lower risk of reduced forced vital capacity and spirometric restriction of the lung (≈290 mg/day) [90], a higher bone mineral density (≈490 mg/day). In addition, flavonoids have been inversely associated with bone fractures (≈1500 mg/day) [85,87] and macular degeneration (≈875 mg/day) [91]. Proanthocyanidins (≥229 mg/day) were inversely associated with risk of renal failure events and kidney insufficiency, while isoflavones (>3 mg/day) with a better pubertal development [84,94].      Higher polyphenol intake was associated with better pulmonary function (forced vital capacity, and forced expiratory volume in the first second) in the population under study. A potential anti-inflammatory activity of polyphenols was hypothesized in men where a reduction in C-reactive protein and white blood cells was observed

Discussion
The great interest for the protective role of polyphenols is demonstrated by the rapid increase of publications evaluating the mechanisms of action of these heterogeneous/complex and multi-target compounds, and also by the studies focused on association between polyphenol intake and different diseases or mortality. In particular, the association of both total or polyphenol subclasses with different types of cancer has been largely addressed in recent reviews and meta-analyses even if the effects are often nulls [96][97][98][99][100][101].
The present study analyzed the literature on polyphenol intake assessment per se or in relation to CVD, diabetes, other health outcomes or mortality.
As expected, the review of data obtained from different studies underlines a consistent difference in the estimated polyphenol intake which may be attributed to different methodological issues such as the type of tool administered to assess the intake, the database used for the calculation of polyphenol intake and the type of polyphenols under evaluation.
It is well known that dietary intake is difficult to measure, and single methods (i.e., questionnaires) cannot perfectly estimate dietary exposure. This is particularly critical especially for micronutrients and bioactive compounds. FFQs, and sometimes 24-h DR, represent the main tools used within the epidemiological studies to assess dietary intake. They have different characteristics; for example, FFQs consist in a pre-finite list of foods and beverages (the number of items queried typically ranges from 80 to 120) with response categories to indicate usual frequency of consumption over the time period queried. Conversely, the 24-h DR consists of an open-ended questionnaire administered by a trained interviewer able to collect detailed information about all foods and beverages consumed by the subjects in the previous 24 h. Both questionnaires present several limitations; for example, FFQs lack of detailed information about food preparation, specific food and beverages consumed, as well as different brands. Moreover, the pre-specified food list does not necessarily reflect the eating behavior of the population under study and the presence of systematic errors must be partially mitigated through appropriate statistical modeling that take into consideration the adjustments for cofounding factors such, as an example, age and energy intake. Regarding 24-h DR it requires multiple days to assess usual intake. In addition, multiple administrations are also recommended when 24-h DRs are used to examine diet impact on health outcomes or other parameters. On the other hand, it has been reported that the assessment of total flavonoid intake requires at least 6 days of weighed food records, and between 6 and 10 days to determine intake of specific flavonoid subclasses with an acceptable degree of accuracy [47]. Most of the studies analyzed in the present review did not perform a multiple evaluation of food intake as highly recommended thus, an under or overestimation of total polyphenols and their classes/subclasses intake cannot be excluded.
Another important critical point for the estimation of polyphenol intake is the choice of the databases. The most commonly used are USDA and Phenol-Explorer. USDA database focuses predominantly on flavonoids as aglycones (anthocyanins, flavanols, flavanones, flavones, flavonols and isoflavones), while Phenol-Explorer, in addition to the above mentioned flavonoids (mainly as glycosides), provides data also of the precursors (chalcones, dihydrochalcones and dihydroflavonols) and information on total polyphenols measured by Folin-Ciocalteu [25]. Despite both data sources are systematically extended to reflect most accurately phenolic contents in food, it is clear that they show several limitations. First of all, since they provide information on different classes of polyphenols, the comparison of the results obtained on the basis of the various data sources may differ. For example, some studies reported that the intake of flavonoids are generally higher when calculated using the USDA databases in relation to the Phenol-Explorer database [102]. In addition, despite they provide information on a wide range of foods, the list does not include all food and polyphenol sources; this represents a critical aspect since missing data have to be found by using different databases and/or by consulting the scientific literature with an increase of risk of bias. Moreover, the effect of seasonality, storage and cooking process is not always considered but certainly, it could represent a critical point. Finally, in view of these issues, it should be remarked that all databases allow only an estimation of dietary polyphenols intake. In this regard, it is noteworthy that databases do not consider non-extractable polyphenols thus contributing to an overall under estimation of intake [103]. This is relevant since these compounds seem to have potential protective properties exerted through gut microbiota metabolites production [104].
In the present review, we found that most of the studies used USDA and Phenol-Explorer databases alone, in combination, or together with other databases and/or data sources (i.e., specific scientific publications). An estimation of polyphenol intake data obtained from reviewed studies using FFQs and from those using 24-h DR, seem to provide comparable results in terms of total polyphenol intake (FFQs 910 mg; 24-h DR 890 mg), total flavonoids (FFQs 360 mg; 24-h DR 380 mg) and total phenolic acids (FFQs 410 mg; 24-h DR 450 mg). In addition, it is noticeable that generally data come from single evaluations instead of multiple evaluations of food intake as recommendable, thus an under or overestimation of polyphenols and/or specific subclasses cannot be excluded.
Polyphenol intake is also affected by intrinsic factors such as the geographical area, the population characteristics in term of age, gender and socio-cultural factors and above all the dietary habits. In this regard, we have found that the intake of total polyphenols is higher in Japan (about 1500 mg/day) compared to European countries and North and South America (about 900 mg/day and 800 mg/day respectively). Within Europe, we found a large variability of intake between countries; Poland and France had the highest intake of total polyphenols (above 1000 mg/day), followed by Italy (about 650 mg/day) and Spain (about 300 mg/day). Conversely, within the EPIC study, Denmark showed the highest intake of total polyphenols (1786 mg/day) while Greece the lowest (584 mg/day) [27].
Regarding total flavonoids, Poland and Australia had the highest intake (about 600 mg/day) while USA and South America the lowest (about 200 and 400 mg/day, respectively) followed by Asia (China and Korea, at about 60 mg/day). Finally, regarding total phenolic acids, France, Poland and Brazil had the highest intake (above 600 mg/day), while USA, Italy and Spain the lowest (about 300 mg/day). These data were also in accordance with the results obtained within EPIC study, which showed a high flavonoid and phenolic acid intake in non-Mediterranean countries [15] associated to different dietary habits. For example, in the North and Central Europe, non-alcoholic beverages, in particular tea and coffee, are the main polyphenol contributors, while in South Europe the main contributors are fruits alcoholic beverages (e.g., red wine). In Asia, such as China and Korea, apples and vegetables seem to be the main polyphenol sources, while green tea in the Japanese population. Finally, tea, citrus and legumes seem to be the main polyphenol contributors in the USA.
As far as gender differences in polyphenol intake are concerned, data in literature are not univocal even if more studies suggest a higher intake in females compared to males, above all when standardization for energy intake is taken into account. In addition, differences in polyphenol sources selected seem to be dependent on gender (e.g., higher contribution of fruit and vegetables in females compared to males who are higher consumers of alcoholic beverages and coffee).
Notwithstanding, most of the data available have been assessing polyphenol intake in adults, a large number of studies considered also the intake in older subjects. Nine studies specifically reported results on total polyphenol and/or subclasses in target of older populations (2 Australia, 2 Spain, 1 Brazil, 1 Italy, 1 Poland, 1 UK and 1 Japan). Total polyphenol intake ranged from about 333 mg/day in Spain [44] to 1492 mg/day in Japan [32]. In addition, those considering total flavonoid intake registered values from about 170 mg/day in Spain [44] to about 834 mg/day in Australia [102]. When available the contribution of phenolic acids was approximately 30-40% of the total polyphenol intake. Studies considering different age classes found controversial results, even if generally, all studies reported differences in food habits affecting polyphenol intake. For example, Vitale et al. [48] showed that flavonoid and stilbene increased with age in the TOSCA.IT study, being higher in over 65 years subjects compared to those with age lower than 65 years. Accordingly, Miranda et al. [39] reported that older subjects (>60 years) from a Brazil cohort consumed more flavonoids and tyrosol than adults (20-59 years) and also more fruits. Moreover, Zamora-Ros et al. [27], showed an increased intake of flavonoids, stilbenes, lignans and other polyphenols with age, while no effect on total polyphenol intake in the EPIC cohort. Other studies reported no differences in polyphenol intake depending on age, or a slight increase after energy adjustment [43,49]. Others (Zujko et al. [19]) showed lower levels of flavonoid intake in older Brazilian subjects who generally consumed less beverages and vegetables. Finally, Karam et al. [44] found an increased energy adjusted polyphenol intake by age classes in older adults from Mallorca island showing also the impact of factors such as gender, educational level and lifestyle significantly affecting eating habits. Large differences in food selection depending on region/country have been underlined reflecting a different pattern of polyphenol intake.
Only 3 studies reported data on children and adolescents showing a low polyphenol intake associated to the overall dietary pattern generally poor in fruit and vegetables even if direct comparison among results is difficult due to the lack of energy adjustment of data in the different age subclasses. The main sources of polyphenols identified depending on the country were non-alcoholic beverages (UK, Argentine), fruit (apple, pear), juices, chocolate (in Helena European study [46]).
Extensive research on polyphenols in human studies has shown a potential role of these compounds in the modulation of CVD markers [105]. In the present systematic review, we found an overall inverse association between total polyphenol intake (highest quantile, above 1170 mg/day) and CV risk events and mortality. In addition, an increased protection against T2D events was observed for total polyphenol intake (mean intake of the 4th quartile) higher than 2632 mg/day [69]. However, the results are not univocal and 4 out of 9 papers reported no association at doses of polyphenols higher than 1200 mg/day or above (>2400 mg/day). These conflicting results could be attributed to the high heterogeneity of the studies in term of selected population characteristic, markers/endpoints measured (i.e., marker of CV risk analyzed), dietary habits (very different between countries), and polyphenol food sources (i.e., tea, coffee, fruits, alcoholic beverages).
Recent evidence from systematic reviews and meta-analyses of cross-sectional and prospective cohort studies seem to suggest that the intake of certain polyphenol classes and subclasses, more than total polyphenols, may reduce the incidence of T2D, CVD events and CVD mortality. However, most of the effects were found when comparing the highest quantiles versus the lowest. In fact, we reported a lower risk of CV events for an intake of total flavonoids ranging from 115 to 944 mg/day, an inverse association for T2D with the highest quartile of total flavonoids (585 mg/day), and a low risk of mortality for cardiovascular events and all-cause mortality for the highest quintile of total flavonoid intake (range 360-800 mg/day) [78,80]. These results are in line with observations reported by other authors. For example, McCullough et al. [74], showed that a total flavonoid intake above 512 mg/day was inversely associated with fatal events for CVD in men and women. Feliciano and coworkers [106], reported that high consumers (>788 mg/day of total flavonoids) showed an inverse association with CVD events and CVD mortality. Wang and colleagues [107] found a reduced risk of CVD events for doses of flavonoids (including flavonols, anthocyanidins, proanthocyanidins, flavones, flavanones and flavan-3-ols) between 139 and 604 mg/day. Finally, Grosso et al. [108] showed that increasing by 100-mg/day flavonoid intake led to a linear decreased risk of 6% and 4% of all-cause and CVD mortality.
As regard the diverse subclasses of polyphenols, several studies have reported a positive effect for flavonols, flavones, flavanones, isoflavones, anthocyanidins and proanthocyanidins. For example, Wedick and coworkers [53], have shown that the highest quintile of anthocyanins (about 22.3 mg/day) and anthocyanin-rich fruit intake (≥5 times/week) was associated with a lower risk of T2D. Conversely, limited evidence is available for lignans. One study performed by Rienks and colleagues [109] showed that high levels of plasma enterolactones (lignan precursors) were associated with a 30% and 45% reduction of all-cause and CVD mortality risk.
Interestingly, in the last years, a growing attention has been devoted to the impact of polyphenols on different health outcomes including for instance renal insufficiency, respiratory function, immune function, and vascular activity. For these outcomes, flavonoids and proantocyanidins have shown an apparent promising beneficial effect. Very recently, another research path has focused on the contribution of polyphenols in the older subject health outcomes. Specifically, the effect on retardation/prevention of some age-related complications such as cognitive decline, frailty and bone fractures has been investigated. On the whole, we have found an overall positive association between high intake of polyphenols and classes/subclasses, and a modulation of different outcomes associated with aging. In particular, total flavonoids and subclasses have been apparently associated with a higher bone mineral density, low risk of bone fractures and macular degeneration, while only total urinary polyphenols, but not dietary polyphenols, have been associated with a low risk of pre-frailty and frailty in older subjects. However, this type of investigation is at early stages thus, further studies have to be performed in order to strength the evidence on the associations found. In addition, since the preliminary observations on protective effects have been found mainly for specific compounds, future studies should be focused on the contribution of subclasses or individual polyphenolic compounds, and even metabolites, instead of total polyphenols.

Conclusions
Undoubtedly, polyphenols exert numerous biological activities as reported in a plethora of in vitro and in vivo studies. In addition, several systematic reviews and meta-analyses of observational and intervention studies have found a reduced risk for numerous chronic diseases. We documented an overall inverse association between polyphenol intake and CV risk events and mortality, as well as, between polyphenols and other outcomes of health status. However, most of the associations were found for specific polyphenol classes/subclasses as well as markers/endpoints. At present, few and conflicting results are available for total polyphenols thus, as also reported more than 10 years ago [110], it is still difficult to establish a reference and/or prudent intake of total polyphenols, even if we found an approximate mean intake of about 900 mg/day. Some studies suggest an inverse association between high total flavonoid intake (generally higher 500 mg/day) and CV events and/or mortality. However, this value should be considered as a temptative level due to the elevated heterogeneity of the studies and the numerous limitations associated with the evaluation and estimation of polyphenol intake. It is then fundamental to consider that polyphenol intake correspond to differences in dietary behavior and selection of diverse food sources of the same compounds could affect the overall impact differently. Therefore, it is reasonable to argue in terms of dietary patterns more than focusing on single contributions. In this context, polyphenol-rich dietary pattern seems to exert health benefits and should be considered a valid tool for the prevention of numerous chronic diseases.
At the same time, further investigation is highly recommended in order to address the need for: (1) improved dietary assessment methods; (2) standardized and validated analytical procedures for the analysis of polyphenols and related subclasses in foods; (3) implementation of food databases increasing food items and information available on the different polyphenol subclasses; (4) validation of specific polyphenol intake biomarkers. Nevertheless, despite information from observational studies are necessary to identify potential role of diet-related compounds, the availability of well controlled and specifically targeted dietary intervention studies (addressing also dose-response effects) seems to be mandatory to allow the identification of a reference or prudent intake (e.g., in term of health-promoting properties) for food bioactives such as polyphenols, directed to the general population or specific vulnerable groups (e.g., older subjects).