Coffee Consumption and Oxidative Stress: A Review of Human Intervention Studies

Research on the potential protective effects of coffee and its bioactives (caffeine, chlorogenic acids and diterpenes) against oxidative stress and related chronic disease risk has been increasing in the last years. The present review summarizes the main findings on the effect of coffee consumption on protection against lipid, protein and DNA damage, as well as on the modulation of antioxidant capacity and antioxidant enzymes in human studies. Twenty-six dietary intervention studies (involving acute and chronic coffee intake) have been considered. Overall, the results suggest that coffee consumption can increase glutathione levels and improve protection against DNA damage, especially following regular/repeated intake. On the contrary, the effects of coffee on plasma antioxidant capacity and antioxidant enzymes, as well as on protein and lipid damage, are unclear following both acute and chronic exposure. The high heterogeneity in terms of type of coffee, doses and duration of the studies, the lack of information on coffee and/or brew bioactive composition, as well as the choice of biomarkers and the methods used for their evaluation, may partially explain the variability observed among findings. More robust and well-controlled intervention studies are necessary for a thorough understanding of the effect of coffee on oxidative stress markers in humans.


Introduction
Coffee is one of the world's most commonly consumed beverages, just after water and tea, probably thanks to its aromatic bouquet and its stimulating effect on the central nervous system. In 2014, coffee consumption has been estimated to reach over 50 million cups worldwide, with the highest annual consumption registered in Finland, Norway and Denmark (11.4, 8.7 and 8.0 kg per capita, respectively) [1], while per capita consumption for Italy and France was estimated in 5.6 kg and 5.4 kg, respectively [2].
Coffee is made by grinding roasted coffee beans, representing the fruit of the coffee plant, belonging to the Rubiaceae family. The main two species are Coffea arabica L. and Coffea canephora, originated in Ethiopia and in tropical Africa, respectively. These two species are traditionally used for making Arabica and Robusta coffees, with the former representing the most diffused species worldwide [1]. Even if all coffee brews could be prepared with hot water and ground coffee in coffee may vary depending on several factors. For example, genetics has been shown to deeply influence both the occurrence of CGAs, with C. robusta generally displaying a higher CGA content than C. arabica [18,19], and the proportion of individual CGA subgroups. Environmental factors, like rainfall level and different mean temperatures, have been reported to affect coffee CGA content even in the same cultivar grown in the same growing area over different years [20]. As regards the processing of coffee, roasting appears to be a critical process for the evolution of flavor, aroma and color in traditional coffee beverages as well as for their CGA content, with a relevant loss of CQAs (up to 90%) in dark roasted beans compared to unroasted green beans. In addition, CGA levels may be affected by the brewing method so that an over-extraction (~55 mL) has been reported to increase over 85% the total CGAs with respect to the regular extraction (~23 mL) [7]. However, CGAs are generally preserved during coffee brewing, resulting in a final concentration that could be higher than 150 mg per serving [7]. Therefore, in spite of the huge CGA loss during roasting and the differences among preparations, coffee still remains the major dietary source of CGAs, to a level that regular coffee intake by heavy drinkers may lead to a daily intake higher that 1 g.
Despite the high amounts of ingested CGAs, only low concentrations of chlorogenic acids in their native forms have been found in blood, and extensive metabolic transformations were previously described [21]. Actually, if an extremely low amount of these compounds is absorbed in the small intestine, most of CGAs reach the colon, where they can be metabolized by the local microbiota. Metabolites are then absorbed, further conjugated in the liver, and distributed to the tissues [17]. At least ten conjugates, dihydroisoferulic acid 3-O-glucuronide, caffeic acid 3-sulfate, as well as the sulfate and glucuronide derivatives of 3,4-dihydroxyphenylpropionic acid, were identified in human plasma and/or urine after coffee consumption [22]. CGAs are a group of esters formed by hydroxycinnamic acids, such as caffeic acid, ferulic acid and p-coumaric acid, bound to quinic acid in a range of conjugated structures known as caffeoylquinic acids, feruloylquinic acids, and p-coumaroylquinic acids [16]. The isomers of these compounds depend on the site of esterification that can occur at positions 1, 3, 4 or 5 of the quinic acid moiety [17]. The most abundant and most studied CGA is 5-caffeoylquinic acid , accounting for about 50% of the total content in green coffee beans [15]. As reported for caffeine and diterpenes, the content of CGAs in coffee may vary depending on several factors. For example, genetics has been shown to deeply influence both the occurrence of CGAs, with C. robusta generally displaying a higher CGA content than C. arabica [18,19], and the proportion of individual CGA subgroups. Environmental factors, like rainfall level and different mean temperatures, have been reported to affect coffee CGA content even in the same cultivar grown in the same growing area over different years [20].
As regards the processing of coffee, roasting appears to be a critical process for the evolution of flavor, aroma and color in traditional coffee beverages as well as for their CGA content, with a relevant loss of CQAs (up to 90%) in dark roasted beans compared to unroasted green beans. In addition, CGA levels may be affected by the brewing method so that an over-extraction (~55 mL) has been reported to increase over 85% the total CGAs with respect to the regular extraction (~23 mL) [7]. However, CGAs are generally preserved during coffee brewing, resulting in a final concentration that could be higher than 150 mg per serving [7]. Therefore, in spite of the huge CGA loss during roasting and the differences among preparations, coffee still remains the major dietary source of CGAs, to a level that regular coffee intake by heavy drinkers may lead to a daily intake higher that 1 g.
Despite the high amounts of ingested CGAs, only low concentrations of chlorogenic acids in their native forms have been found in blood, and extensive metabolic transformations were previously described [21]. Actually, if an extremely low amount of these compounds is absorbed in the small intestine, most of CGAs reach the colon, where they can be metabolized by the local microbiota. Metabolites are then absorbed, further conjugated in the liver, and distributed to the tissues [17]. At least ten conjugates, dihydroisoferulic acid 3-O-glucuronide, caffeic acid 3-sulfate, as well as the sulfate and glucuronide derivatives of 3,4-dihydroxyphenylpropionic acid, were identified in human plasma and/or urine after coffee consumption [22].

Objective and Literature Search Strategy
The present review aims to summarize the main findings of human intervention studies investigating the effects of coffee consumption on oxidative stress. In detail, the effects of coffee consumption on markers of lipid, protein and DNA damage, as well as on markers related to antioxidant capacity and antioxidant enzymes, are reviewed. Most of these markers are considered sensitive and specific biomarkers for antioxidant status and can be useful for a better comprehension of the role of antioxidant-rich foods, including coffee, against oxidative stress and related conditions. PUBMED, Web of Science and Scopus databases were searched to identify pertinent articles. The searches used the combination of the following terms: "coffee", "antioxidant capacity", "antioxidant activity", "DNA damage", "protein damage", "lipid damage", "oxidation" and "human". Reference lists of the retrieved papers were also searched for additional articles. The search strategy is summarized in Figure 2.

Objective and Literature Search Strategy
The present review aims to summarize the main findings of human intervention studies investigating the effects of coffee consumption on oxidative stress. In detail, the effects of coffee consumption on markers of lipid, protein and DNA damage, as well as on markers related to antioxidant capacity and antioxidant enzymes, are reviewed. Most of these markers are considered sensitive and specific biomarkers for antioxidant status and can be useful for a better comprehension of the role of antioxidant-rich foods, including coffee, against oxidative stress and related conditions. PUBMED, Web of Science and Scopus databases were searched to identify pertinent articles. The searches used the combination of the following terms: ''coffee'', ''antioxidant capacity'', ''antioxidant activity'', ''DNA damage'', ''protein damage'', ''lipid damage'', ''oxidation'' and ''human''. Reference lists of the retrieved papers were also searched for additional articles. The search strategy is summarized in Figure 2. (2) did not concern intervention studies; (3) not including in vivo markers of oxidative stress.
A total of 26 pertinent human intervention studies were identified, published in 18 different peer-reviewed journals and conducted in 12 different countries, mainly Austria, Germany and Italy. More than half of the 26 studies were published during the last 5 years, suggesting a growing interest for this topic. The results obtained are reported in Tables 1 and 2 describing the type of coffee used in the investigation, with details about the way of preparation and the content of bioactive compounds in the final brew, when available. Moreover, the duration of the intervention, the number of subjects and their characteristics, the use of a control/placebo food, the selected markers and the main findings are reported.
Three out of the 26 studies investigated the effects of both acute and chronic interventions with coffee, while 8 were only acute studies (i.e., single dose), and 15 were chronic (medium-long term) dietary intervention. (2) did not concern intervention studies; (3) not including in vivo markers of oxidative stress.
A total of 26 pertinent human intervention studies were identified, published in 18 different peer-reviewed journals and conducted in 12 different countries, mainly Austria, Germany and Italy. More than half of the 26 studies were published during the last 5 years, suggesting a growing interest for this topic. The results obtained are reported in Tables 1 and 2 describing the type of coffee used in the investigation, with details about the way of preparation and the content of bioactive compounds in the final brew, when available. Moreover, the duration of the intervention, the number of subjects and their characteristics, the use of a control/placebo food, the selected markers and the main findings are reported.       Three out of the 26 studies investigated the effects of both acute and chronic interventions with coffee, while 8 were only acute studies (i.e., single dose), and 15 were chronic (medium-long term) dietary intervention.

Total Plasma Antioxidant Capacity and Antioxidant Enzymes
The effect of coffee consumption on the modulation of plasma antioxidant capacity was evaluated in 10 of the selected studies. Two studies reported both acute and chronic interventions [23,24], four were acute studies [25][26][27][28], while four were chronic intervention studies [34][35][36][37]. Three studies were not placebo-controlled [24,34,37] and/or did not provide information about the bioactive composition of coffee [24,26,27]. Total radical trapping antioxidant power (TRAP), Trolox-equivalent antioxidant capacity (TEAC), total antioxidant status (TAS) and oxygen radical absorbance capacity (ORAC) emerged as the most used methods. They differ from each other in terms of reaction mechanisms, oxidant and target/probe species, reaction conditions and expression of the obtained results.
Only four (three acute and one chronic intervention) out of 10 studies reported a significant increase in total plasma antioxidant capacity following coffee consumption. In particular, Moura-Nunes and colleagues documented that a single serving of 200 mL of coffee increased plasma antioxidant capacity, determined through FRAP and TRAP assays, by 2.6% and 7.6%, respectively, in a group of healthy subjects [27]. Natella and coworkers observed an increase in plasma antioxidant capacity (measured as TRAP) and thiol (SH) groups after consumption of 200 mL of coffee, which was even more pronounced than that observed with the same amount of tea (+6% versus +4%, respectively) [28]. Agudelo-Ochoa et al. [23] reported that a single serving of 400 mL of coffee, providing either 420 mg or 780 mg of chlorogenic acid, significantly increased plasma antioxidant capacity (+6% and +5%, respectively) in a group of healthy volunteers, but these effects were lost following a long term intervention. Finally, Corrêa et al. [34] reported that a 4-week intake of 150 mL/day of medium light roast (MLR) or medium roast (MR) paper-filtered coffee increased the levels of TAS by about 21% and 26% respectively, while ORAC increased only after the consumption of medium light roast paper-filtered coffee. The lack of homogeneous results could be due to differences in the duration of the intervention, the type and the amount of coffee provided, the composition in bioactive compounds, and the method used for the determination of antioxidant capacity.
The effect of coffee in the modulation of endogenous antioxidant enzymes has been assessed in seven chronic intervention trials. The duration of the studies varied from 1 to 4 weeks. Superoxide dismutase (SOD), catalase (CAT), glutathione peroxidase (GPx), glutathione reductase (GSR) and glutathione S-transferases (GSTs) were the most widely studied enzymes.
Overall, the results obtained are conflicting and do not provide evidence about the role of coffee in the modulation of antioxidant enzymes. For example, Corrêa et al. [34] reported an increase in erythrocyte antioxidant enzyme activity in healthy subjects following a 4-week intervention with 150 mL/day of MLR and MR paper-filtered coffee. In particular, SOD activity increased by 52% and 75% in MLR and MR group, while GPx activity by 62% and 49%, respectively. Moreover, both the interventions significantly increased CAT activity by about 13%. Kotyczka et al. [38] documented that a 4-week intake of light dark roasted coffee (rich in chlorogenic acid) resulted in increased erythrocyte activities of SOD, GPx and CAT by 12%, 25% and 22%, respectively. On the contrary, the intake of dark roasted coffee (poor in chlorogenic acid) decreased erythrocytes SOD and GPx activity by 5.8% and 15%, respectively. Misik and colleagues showed that a 5-day coffee intake (800 mL/day) did not significantly affect SOD and GPx activity in the cytosolic fractions of the lymphocytes of healthy volunteers [36].
The activity of GST and GSR was evaluated in two studies. Steinkellner and colleagues reported that a 5-day intervention with 1 L unfiltered coffee/day increased GST activity in plasma, but not in saliva, in a group of healthy individuals [39]. Bakuradze et al. [40] showed a significant increase in GSR blood level activity following 4-week consumption of 750 mL/day of freshly brewed coffee.
Eight studies (seven chronic interventions and one acute trial) also investigated the role of coffee in the modulation of blood glutathione (GSH) levels as a substrate of GPx and GST enzymes. Four out of seven chronic intervention studies documented an increase in GSH levels [38,[40][41][42], while two long-term studies [35,36] and one study performing both an acute and a chronic intervention [24] did not show any significant effect. Misik and coworkers attributed the lack of effects to the degradation and metabolic conversion of different coffee constituents in the body [36], while for Teekachunhatean and colleagues the short duration of the intervention could be at least partially responsible for the absence of a relevant effect [24].

Protein Damage
The effect exerted by coffee intake on protein damage has been investigated only in three studies [35,36,43]. Those studies differed for the fed coffee type, which was instant coffee [35], coffee homemade with participants' coffee machines [36] or coffee prepared by paper filtration [43]. In two studies [35,43], partially performed by the same research group, a controlled intervention trial with a cross-over design was scheduled, where subjects were randomized to consume coffee or a control drink (water) for periods of 5 days each, spaced out by a washout phase and dietary restriction. Differently, Kempf and coworkers investigated the effect of two different doses of coffee (4 and 8 cups, corresponding to 600 and 1200 mL respectively) by using a simple experimental design: subjects had to follow 4 weeks of restrictive diet, followed by 4 weeks with 4 cups coffee/day and 4 weeks of 8 cups coffee/day [43]. The studies performed by Hoelzl et al. [35] and Misik et al. [36] were also similar for what concerns the volunteers, being nonsmokers and normal weight young adults in both cases, contrarily to the study by Kempf et al. [43] where subjects with a high risk of type 2 diabetes were recruited. In all the three studies, the marker of protein damage was 3-nytrotirosine (3-NT), a stable post-translational modification in proteins, deriving from the reaction of free or protein-bound tyrosine with reactive nitrogen oxide species including peroxynitrite, nitrogen dioxide and nitrous acid. 3-NT has been suggested to be associated with coronary heart disease (CAD) independently of traditional risk factors [49]. The marker was monitored by LC-MS/MS in two studies [35,36] and by an enzyme immunoassay in the third one [38]. A significant effect of coffee was found only by Hoelzl and colleagues, who observed a significant decrease of 3-NT (16.1%) after 4-week coffee intake [35]. Despite many similarities between the investigations by Hoelzl et al. [35] and Misik et al. [36] (i.e., duration of the intervention, amounts of CGA provided), the latter found no significant effect after coffee intake, in line with the observations of Kempf and colleagues [43]. A possible explanation for such different findings might be linked to the amount of coffee bioactives other than CGA provided with the two different coffee brews. However, Hoelzl et al. [35] provided information only about CGA, making a clear comparison between the two investigations practically impossible.
The small number of investigations about the possible role of coffee consumption on markers of protein damage, together with the heterogeneity of the findings, calls for further studies focusing on this aspect of oxidative damage to biomolecules.
In these studies, isoprostanes (IsoPs) and malondialdehyde (MDA) were the most frequently considered markers of lipid damage. In detail, isoprostanes are a class of end-products of peroxidation of arachidonic acid, which are collectively referred as F2-IsoPS due to their similarity to prostaglandin F2α. Among them, 8-isoprostaglandin F2α (8-Iso PGF2α) is commonly used for evaluating oxidative stress, through both chromatographic techniques and immunoassays. MDA is instead a three carbon, low molecular weight aldehyde representing the main product of polyunsaturated fatty acid peroxidation. It is characterized by a high toxicity due to its ability to react with other molecules like DNA and protein [50]. In all the studies evaluating MDA, the reaction with 2-thiobarbituric acid (TBA) was used, so results were reported as TBA reacting substances (TBARS) instead of MDA. In spite of the risk of overestimation of MDA, the TBARS method represents the most common test for evaluating lipid peroxidation.
Besides 8-IsoPGF2 and MDA, further markers of lipid damage and/or protection considered in the present review were oxidized LDL, resistance to LDL oxidation, serum LDL-conjugated dienes and hydroxyl fatty acids. The analysis of the main findings revealed that most of the interventions failed to demonstrate a significant decrease in markers of lipid damage with exception of results found by Ochiai et al. [30] and Sirota et al. [31]. The former reported a significantly reduced urinary 8-epiPGF2α following consumption of a coffee beverage (providing 600 mg of CGAs) when compared with placebo in healthy men. Results showed that consumption of 200 mL Turkish roasted coffee during a meal based on red-meat cutlets resulted in a significant inhibition of postprandial plasma MDA. No effect between treatments and control/placebo were instead found by other authors [25,29,32]. The investigation by Leelarungrayub et al. [26] deserves a special mention, because it reports a significant higher level of MDA in men consuming caffeinated coffee, when compared to decaffeinated coffee or control, followed by a submaximal exercise test. Authors reported that, similarly to what observed in previous investigations, results demonstrated an increased intramuscular fat oxidation following consumption of caffeine-rich foods.
Among medium-long term intervention studies, a significant decrease in isoprostanes was observed only in three studies [30,35,43], while no significant change was reported by Mursu et al. [32] and Misik et al. [36]. Even if the exact composition of the coffee used has not been always provided, the differences in the findings seem to be at least partially attributable to the different composition of the brews.
For what concerns the other markers of lipid damage, only Yukawa et al. [44] found a modest reduction of LDL oxidation susceptibility and a decrease of MDA levels following consumption of 3 coffees/day for 1 week. No significant effect was instead found by Mursu et al. [32] on serum LDL-conjugated dienes and plasma hydroxyl fatty acids, or by Teekachunhatean et al. [24] on MDA levels [24] and by Hoelzl et al. [35] on both MDA and oxidized LDL.

DNA Damage
The role of coffee on markers of DNA damage has been investigated in nine studies (eight chronic interventions and one acute study), four of which performed by the same research group [33,40,46,47]. Eight out of nine studies measured the levels of DNA damage through the comet assay, a single cell gel electrophoresis technique widely used also in human biomonitoring and dietary intervention studies [51,52]. Three of these studies [33,46,47] investigated the effect of coffee on spontaneous DNA strand breaks (SBs), which directly reflect the background DNA oxidative damage within cells. Background SBs may derive from endogenous and/or exogenous exposure to DNA damaging agents and/or may reflect incomplete DNA repair. The consumption of coffee was associated to reduce DNA SBs in healthy volunteers. Bakuradze et al. [33] reported that the ingestion of 800 mL of coffee (200 mL every 2 h) significantly reduced (up to 30%) DNA SBs in a short-term human intervention study. The same research group documented that coffee consumption (3ˆ250 mL per day) was associated with DNA-protective effects (p < 0.001) in a 4-week, double-blind, randomized, crossover intervention [46]. Finally, the same authors showed that a daily consumption of 750 mL of fresh dark roast coffee decreased by 27% spontaneous DNA SBs in a 4-week, randomized, controlled trial [47].
Five out of eight studies investigated the effects of coffee consumption on oxidized DNA bases through the exploitation of specific enzymes such as formamidopyrimidine-DNA glycosylase

Conclusions
During the last five years, coffee has been the objective of several studies for its potential role in human health, with a specific focus on the prevention of several chronic degenerative diseases. The current review summarized the main findings of 26 studies performed in humans, with the aim of comparing results on the effect of coffee consumption on the main markers of oxidative damage to lipid, DNA and protein, as well as on the modulation of antioxidant capacity and antioxidant enzymes in humans. Studies were performed on healthy subjects with the exception of one study in which patients with chronic hepatitis C were recruited.
Overall, a high heterogeneity among studies was observed, mainly in terms of doses and duration of the interventions, and, in several studies, information concerning the polyphenol content of the coffee used was lacking. Only a few studies provided the content of CGAs, caffeine and other bioactive compounds in the fed coffee brew, and, in general, they did not describe in great detail the way coffee was prepared (i.e., grams of coffee used for each dose). This lack of information about the composition of the brews makes the comparison among studies extremely difficult, with an objective evaluation of a dose-response effect almost impossible. Therefore, the need for more detailed information about the chemical composition of coffees in future studies appears crucial for a more accurate analysis of results.
Despite these complications, the main findings of the reviewed works seem to suggest that consumption of coffee may increase glutathione levels and reduce the levels of DNA damage. These effects are more evident in chronic interventions than in acute studies, letting hypothesize that a long-term exposure to coffee and/or its bioactive compounds is needed to obtain such putative health effects. On the contrary, an extreme heterogeneity of the results has been observed for total plasma and serum antioxidant status, as well as for protein and lipid damage. This could be attributed to the different biomarkers and methods used for their evaluation. Based on the difficulties described above, a comprehensive understanding of the beneficial effects of coffee on oxidative stress markers, through the development of robust and well-controlled intervention studies, is required.