Impact of Cocoa Consumption on Inflammation Processes—A Critical Review of Randomized Controlled Trials

Background: Cocoa flavanols have strong anti-inflammatory properties in vitro. If these also occur in vivo, cocoa consumption may contribute to the prevention or treatment of diseases mediated by chronic inflammation. This critical review judged the evidence for such effects occurring after cocoa consumption. Methods: A literature search in Medline was performed for randomized controlled trials (RCTs) that investigated the effects of cocoa consumption on inflammatory biomarkers. Results: Thirty-three RCTs were included, along with 9 bolus and 24 regular consumption studies. Acute cocoa consumption decreased adhesion molecules and 4-series leukotrienes in serum, nuclear factor κB activation in leukocytes, and the expression of CD62P and CD11b on monocytes and neutrophils. In healthy subjects and in patients with cardiovascular diseases, most regular consumption trials did not find any changes except for a decreased number of endothelial microparticles, but several cellular and humoral inflammation markers decreased in patients suffering from type 2 diabetes and impaired fasting glucose. Conclusions: Little evidence exists that consumption of cocoa-rich food may reduce inflammation, probably by lowering the activation of monocytes and neutrophils. The efficacy seems to depend on the extent of the basal inflammatory burden. Further well-designed RCTs with inflammation as the primary outcome are needed, focusing on specific markers of leukocyte activation and considering endothelial microparticles as marker of vascular inflammation.


Introduction
Chronic inflammation has been observed in several chronic degenerative diseases beyond rheumatoid arthritis, e.g., in coronary artery disease (CAD) [1], diabetes mellitus [2], age-related macular degeneration [3], Parkinson's, and Alzheimer's disease [4]. Chronic inflammation is a decisive risk factor for the development of vascular diseases as it plays a key role in all stages of the formation of vascular lesions and leads to endothelial dysfunction [1]. Biomarkers in serum and plasma like proinflammatory cytokines, adhesion molecules, and acute-phase reactants such as C-reactive protein (CRP) are related to increased risk of cardiovascular events [5] and diabetes-associated vascular impairment [6].
This explains the increasing interest in nutrients with anti-inflammatory properties. A growing body of evidence from in vitro studies suggests that nutritive flavanols can modulate the synthesis of pro-and anti-inflammatory metabolites. Flavanols have been shown to reduce the transcription and secretion of adhesion molecules and proinflammatory cytokines (interleukin-1β (IL-1β), tumor necrosis factor-α (TNF-α)) [7][8][9][10]. Moreover, flavanols increased the production of anti-inflammatory cytokines such as interleukin-4 and interleukin-5 in vitro [11]. Since epicatechin, catechin, and dimeric 1.
Type of studies: randomized controlled trials 2.
Type of interventions: supplementing foods rich in cocoa without restrictions to the kind of food used (e.g., dark chocolate, cocoa-rich drinks, cocoa-enriched foods) or the frequency of supplementation (e.g., single or repeated, daily consumption) 3.
Type of controls: supplementing comparable foods lacking cocoa (e.g., white chocolate, milk, placebo drink, foods not enriched with cocoa), being low in cocoa or not providing cocoa-rich foods 4.
Type of inflammatory biomarkers: humoral parameters determined in serum or plasma (e.g., CRP, soluble adhesion molecules, eicosanoids, cytokines) and cellular parameters determined in leukocytes obtained from whole blood (e.g., cellular adhesion molecules, immunological functions) All records identified were checked for duplicates. Duplicates were removed and the remaining records were screened by title and/or abstract to exclude records that did not contribute to answering the question asked in this review. The full-text articles of the remaining records were assessed for eligibility on the basis of the criteria mentioned above.

Criteria for Judging the Quality of the Selected Trials
The quality of the selected RCTs was judged according to the GRADE criteria. Methodological limitations, e.g., lack of allocation concealment, lack of blinding (participants, researchers), lack of reasons for dropout, no data on compliance, industrial funding, and carryover effects in case of crossover trials were considered. This approach is helpful to assess the quality of evidence more differently by rating down the evidence in case of serious limitations [41,42]. Recording the dietary intake and imposing dietary restrictions on the consumption of other flavanol-rich foods than cocoa may exclude potential side effects and were considered as additional criteria to assess the studies' quality. Figure 1 provides an overview of the selection of relevant studies. In total, 338 records could be identified. After removing 225 duplicates, the remaining records (n = 113) were screened by title and/or by abstract. This led to the exclusion of 59 records that were considered to be irrelevant to the question addressed by this review. The remaining 54 records were checked for eligibility by the full-text article, leading to the exclusion of 21 articles. Finally, 33 RCTs were included in this review.

Results
These 33 trials were published in 32 different articles. Nine trials investigated the effect of acute cocoa consumption [43][44][45][46][47][48][49][50] (Table 1) and 24 trials addressed the impact of regular, i.e., daily, cocoa consumption on markers of inflammation. The regular consumption trials were further subdivided according to the subjects investigated: trials performed with healthy subjects [51][52][53][54][55][56][57][58][59] ( Table 2), pre-/hypertensive subjects [51,[60][61][62] (Table 3), and trials investigating patients suffering from type 2 diabetes, impaired glucose tolerance [61,[63][64][65][66][67][68] (Table 4), and from CAD [48,69,70] ( Table 5). For treatment, most trials used dark chocolate [43,45,[47][48][49][50][51][52][53]58,59,61,68,69,71] or cocoa drinks rich in flavanols [44,46,50,54,55,57,59,60,[62][63][64]66,67,70,72]. Chocolate or drinks low [43,44,49,55,58,60,62,63,66,[70][71][72] or even free [45][46][47][48]50,51,[53][54][55][56][57]59,61,[64][65][66][67][68][69] from cocoa mostly served as controls. A few RCTs addressed the effect of flavanol-rich cocoa on postprandial [43,50] or hyperglycemia-induced inflammation [49,66].  Table 1 provides on overview on bolus studies (n = 9) included in this review. In healthy subjects, the consumption of cocoa-rich food decreased the concentration of leukotrienes C4, D4, and E4 in plasma [43], the expression of CD62P (P-selectin) on collagen-activated monocytes and neutrophils [44], and the serum concentration of intercellular adhesion molecule-1 (ICAM-1) and E-selectin [46]. A reduction in E-selectin was only observed if cocoa-rich beverages were prepared with water. A reduction in ICAM-1 was induced after ingestion of water and milk based cocoa-rich beverages, but concentrations after 6 h were comparably lower after ingestion of the water-based drink. Interestingly, the ratio of P-p65/β-actin in peripheral blood mononuclear cells (PBMC) decreased after consuming the water-based cocoa drink, but not after the milk-based cocoa drink, and even increased after consumption of pure milk. In contrast to ICAM-1, vascular cell adhesion molecule-1 (VCAM-1) did not change after any kind of drink [46]. In healthy non-smokers with prolonged exercise, consumption of 100 g of dark chocolate 2 h prior to prolonged exercise did not change IL-6 in plasma and neutrophils' respiratory burst and degranulation when compared to the cocoa-free chocolate and fasting conditions [47]. In patients with type 2 diabetes, ingestion of 13.5 g chocolate rich in flavanols 60 min after an oral 75-g-glucose load did not change ICAM-1, E-selectin, P-selectin, and P-selectin-glycoprotein ligand 1 in serum, whereas ICAM-1 increased after ingestion of the same amount of flavanol-poor chocolate [49]. It is important to mention that the differences between pre-and post-consumption values of these four parameters were significantly lower for the flavanol-rich compared to the flavanol-poor chocolate. In type 2 diabetics, a placebo-controlled trial investigating whether a cocoa-rich drink consumed together with a high-fat fast-food style meal might affect CRP, but CRP levels did not change after either intervention, and differences between pre-and post-consumption values could not be observed [50]. In patients with peripheral artery disease, ingestion of 40 g dark chocolate reduced the concentration of NOX2-derived peptide, a marker of NADPH oxidase activity, in serum 2 h after consumption, whereas no changes occurred after consumption of milk chocolate [71]. In patients with congestive heart failure [48] and in heart transplant recipients [45], bolus consumption of 40 g dark or cocoa-free chocolate did not change CRP.  Table 1 provides on overview on bolus studies (n = 9) included in this review. In healthy subjects, the consumption of cocoa-rich food decreased the concentration of leukotrienes C 4 , D 4 , and E 4 in plasma [43], the expression of CD62P (P-selectin) on collagen-activated monocytes and neutrophils [44], and the serum concentration of intercellular adhesion molecule-1 (ICAM-1) and E-selectin [46]. A reduction in E-selectin was only observed if cocoa-rich beverages were prepared with water. A reduction in ICAM-1 was induced after ingestion of water and milk based cocoa-rich beverages, but concentrations after 6 h were comparably lower after ingestion of the water-based drink. Interestingly, the ratio of P-p65/β-actin in peripheral blood mononuclear cells (PBMC) decreased after consuming the water-based cocoa drink, but not after the milk-based cocoa drink, and even increased after consumption of pure milk. In contrast to ICAM-1, vascular cell adhesion molecule-1 (VCAM-1) did not change after any kind of drink [46]. In healthy non-smokers with prolonged exercise, consumption of 100 g of dark chocolate 2 h prior to prolonged exercise did not change IL-6 in plasma and neutrophils' respiratory burst and degranulation when compared to the cocoa-free chocolate and fasting conditions [47]. In patients with type 2 diabetes, ingestion of 13.5 g chocolate rich in flavanols 60 min after an oral 75-g-glucose load did not change ICAM-1, E-selectin, P-selectin, and P-selectin-glycoprotein ligand 1 in serum, whereas ICAM-1 increased after ingestion of the same amount of flavanol-poor chocolate [49]. It is important to mention that the differences between pre-and post-consumption values of these four parameters were significantly lower for the flavanol-rich compared to the flavanol-poor chocolate. In type 2 diabetics, a placebo-controlled trial investigating whether a cocoa-rich drink consumed together with a high-fat fast-food style meal might affect CRP, but CRP levels did not change after either intervention, and differences between pre-and post-consumption values could not be observed [50]. In patients with peripheral artery disease, ingestion of 40 g dark chocolate reduced the concentration of NOX2-derived peptide, a marker of NADPH oxidase activity, in serum 2 h after consumption, whereas no changes occurred after consumption of milk chocolate [71]. In patients with congestive heart failure [48] and in heart transplant recipients [45], bolus consumption of 40 g dark or cocoa-free chocolate did not change CRP.              Table 2 provides on overview of studies (n = 9) that investigated the impact of regular consumption of cocoa-rich products on markers of inflammation in serum or plasma in healthy subjects after an overnight fast. ICAM-1 decreased after daily consumption of dark chocolate (41 g/day) for 42 weeks [52], but not after providing 100 g/day chocolate for 15 days [51] or 400 mL/day of a cocoa-rich beverage for 28 days [57]. In the study of Ibero-Baraibar et al. [56], ICAM-1 decreased by consumption of ready-to-eat-meals irrespective if these were enriched with cocoa extract or not. However, these meals were part of a hypocaloric diet with an energy restriction about 15% of total energy expenditure [56]. Regular cocoa consumption did not change VCAM-1 in any study [52,56,57]. Changes in proinflammatory cytokines (IL-1β, IL-6, TNF-α) were not observed [57,58]. The concentration of chemokines in plasma (IL-8, monocyte chemoattractant protein-1) was not modulated by 4-week consumption of a cocoa drink [57]. However, interleukin-10 decreased after intervention with cocoa-rich and cocoa-free drinks, but the changes induced by the cocoa drink were significantly lower than those of the cocoa-free control [57]. In most trials that determined serum CRP concentrations, changes did not occur due to regular cocoa consumption [51][52][53][54]57,58]. Only Tzounis et al. found a decrease in CRP after 4-week consumption of a flavanol-rich cocoa drink, which was accompanied by significantly lower post-consumption values compared to the low-flavanol cocoa drink [55]. In three studies, results on CRP were not reported [57][58][59]. In another study [59], reduced concentrations of haptoglobin and endothelial microparticles in serum and plasma, respectively, and a lower expression of CD62L (L-selectin) on monocytes after 4-week consumption of cocoa-rich bars were observed in obese subjects, but not in similar aged subjects who were normal weight or overweight. In obese subjects, a trend towards reduced levels of endocan-1 (also called endothelial cell specific molecule-1) and E-selectin in serum after cocoa consumption could be observed, which, however, did not reach significance. At baseline, the concentrations of haptoglobin, E-selectin, and endothelial microparticles were significantly higher in obese subjects compared to normal-weight and overweight subjects. Moreover, obese participants initially had a higher expression of CD62L on monocytes than participants who were normal weight or overweight.

Pre-/Hypertensive Subjects
The results of RCTs performed with pre-/hypertensive subjects are shown in Table 3. In patients with untreated stage I hypertension, daily consumption of 100 g dark chocolate for 15 days did not affect ICAM-1 [51] and CRP [51,61]. Consumption of flavanol-rich cocoa drinks for two [62] or six weeks [60], respectively, did not change E-selectin [60,62], P-selectin [60], ICAM-1, VCAM-1, monocyte chemoattractant protein-1 (MCP-1), IL-6, or TNF-α [62] in serum or plasma. The only significant change observed concerned VCAM-1 in the study of Wang-Polagruto et al. [60]. Here, VCAM-1 decreased after six weeks' ingestion of a cocoa-rich drink and the differences between pre-and post-consumption values were higher after consuming the cocoa-rich compared to the control drink.

Patients with Type 2 Diabetes or Impaired Glucose Tolerance
Studies with patients suffering from type 2 diabetes or impaired glucose tolerance are shown in Table 4. Daily administration of flavanol-rich chocolate (25 g [68], 45 g [65], 100 g [61]) for 5-56 days compared to white chocolate did not change CRP. Most studies providing cocoa-rich beverages observed anti-inflammatory effects: Parsaeyan et al. found a decrease in CRP, IL-6, and in TNF-α after six weeks [67]. Stote et al. [66] provided water-based beverages with different amounts of flavanols for five days to a group of obese patients who suffered from impaired glucose tolerance. Before and after intervention, an oral 75-g-glucose load was given and a significant decrease in ICAM-1, IL-6, and in CRP by ingestion of the cocoa drink was observed, dependent on the dose of flavanols ingested. Balzer et al. [63] also provided flavanol-rich cocoa, but did not find any changes in CRP. However, this was the only inflammatory biomarker determined in this trial. Monagas et al. [64], who investigated a mixed group of patients (diabetics and patients with at least three cardiovascular disease risk factors such as smoking, hypertension, hypercholesterolemia, obesity, and a family history of premature coronary disease), provided 500 mL/day of a flavanol-rich cocoa drink, based on milk, and pure milk as a control for 28 days. Cocoa consumption reduced the concentration of several adhesion molecules in plasma (ICAM-1, P-selectin) and on monocytes (very late antigen-4; VLA-4; CD49d), and the expression of CD36 and CD40 on monocytes. For these parameters, differences between pre-and post-consumption values were significantly higher after consumption of cocoa-rich cocoa compared to milk. E-selectin, VCAM-1, MCP-1, IL-6, and CRP in serum or plasma and lymphocyte function-associated antigen-1 (LFA-1; CD11a), Mac-1 (CD11b/CD18), Sialil Lewis X (SLe x ; CD15s) on monocytes did not change. The expression of LFA-1, VLA-4, SLe x , and CD40 on T cells was not modulated by cocoa intervention either.

Patients with Coronary Heart Disease
In patients suffering from CAD [69,70,72] and congestive heart failure [48] (Table 5), daily ingestion of flavanol-rich chocolate [48,69] or cocoa [70,72] for 28-42 days did not change CRP [48,69,70], P-selectin, E-selectin, ICAM-1, or VCAM-1 [69]. However, VCAM-1 was significantly higher before the consumption of flavanol-rich chocolate compared to isocaloric placebo. This difference was no longer significant after six weeks of intervention [69]. Chemotaxis of PBMC was not modulated by consumption of flavanol-rich cocoa [70]. The number of endothelial microparticles (CD144+ as well as CD31+/CD41-subpopulation) decreased significantly after ingestion of flavanol-rich cocoa drinks and reached lower concentrations after 30 days of intervention compared to the placebo drinks. At baseline, CAD patients had more CD31+/CD41-endothelial microparticles in blood than age-matched and younger healthy controls, whereas the number of CD144+ endothelial microparticles differed only between young and old subjects, but not between CAD patients and age-matched healthy controls. In contrast to endothelial microparticles, the number of platelet-derived microparticles (CD41+) was not affected by cocoa treatment [72]. Table 6 provides an overview on qualitative criteria of the selected trials. Seventy percent of them had crossover and 30% had a parallel group design. Masking of researchers and participants was ensured in 82% and 61% of all RCTs, respectively. Information on allocation concealment was lacking in half of all studies. Dropout was clearly reported in most studies with regular cocoa consumption and also in bolus studies if patients were lost, but this did rarely occur. Forty percent of all studies were funded by industry and most trials received the cocoa products as gifts from industry (82%). In bolus studies, compliance was not reported as cocoa products were always ingested in the study center. Only 50% of regular consumption studies determined adherence to dietary treatment. Carry-over effects could be excluded in most trials with crossover design by washout periods. About 45% of all studies recorded dietary intake and 76% provided dietary restrictions to prevent confounding effects.  [46] ? Grassi et al., 2005 [51] ?
a concerns studies with crossover design: In case of washout periods were reported, carry-over effects could be excluded, but not if washout periods were missing or not reported. All criteria except for "dietary intake documented" and "dietary restrictions" are GRADE criteria. N: no, Y: yes; ?: unclear.

Discussion
To the best of our knowledge, this is the first review that critically evaluates the impact of cocoa consumption on inflammatory markers on the basis of RCTs.
The different results from bolus and regular consumption studies in healthy subjects raises the question of why anti-inflammatory effects were detectable after acute cocoa consumption, but not in most studies after regular cocoa consumption. First of all, we have to bear in mind that epicatechin, which is claimed to be responsible for anti-inflammatory effects after cocoa intake [31], has an elimination half-time about 2 h [75]. For this reason, anti-inflammatory effects are rather unlikely to occur in blood samples obtained after an overnight fast in regular consumption studies. On the other hand, epicatechin metabolites were found in lymphoid tissues of rats after daily cocoa feeding for three weeks [76]. Consequently, an accumulation of epicatechin in human leukocytes and also in endothelial cells by regular cocoa consumption may occur. Second, if we consider that a balanced release of vasodilating/-constricting factors and pro-/antithrombotic substances prevents vascular inflammation in healthy subjects [77], the lack of anti-inflammatory effects after regular cocoa consumption in fasting blood may be simply explained by the lack of basal inflammation. This is stressed by the results of McFarlin et al., who found anti-inflammatory effects after regular cocoa consumption only in obese participants with initially elevated inflammation markers compared to normal-weight and overweight subjects, but not in the later subgroups [59]. A range of proinflammatory adipokines are released from adipose tissue in obesity [78], but most RCTs on regular cocoa consumption did not investigate healthy subjects with obesity (Table 2).
In contrast to healthy subjects, most regular consumption trials with patients suffering from type 2 diabetes or impaired glucose tolerance found lower concentrations of ICAM-1 [64,66] and P-selectin in serum or plasma [64] and a reduced expression of VLA-4, CD36, and CD40 on monocytes [64]. Results on proinflammatory cytokines and on CRP are divergent (Table 4), but a reduction could be observed in two studies for proinflammatory cytokines [66,67] and in three studies for CRP [66][67][68] (Table 4). In pre-/hypertension (Table 3) and in CAD (Table 5), two studies only observed a reduction in VCAM-1 by regular cocoa consumption [60,69]. The activity of NADPH-oxidase [71] and the number of endothelial microparticles [72] decreased after acute [71] and regular consumption [72] of flavanol-rich cocoa products, respectively. However, further trials did not find any changes after acute [45,48] and regular consumption [48,51,61,62,70]. The results from different groups of participants suggest that the reaction towards cocoa consumption is different. This may be explained by the different stages of vascular inflammation. In patients with type 2 diabetes, vascular endothelial dysfunction can be partly restored by reducing inflammation [79]. In these patients, including those with impaired glucose tolerance, anti-inflammatory effects could be observed in several studies after cocoa consumption at the basal state, but also after an oral glucose challenge ( Table 4). As vascular inflammation in patients with existing cardiovascular diseases is pronounced and associated with proatherogenic modifications [77], the anti-inflammatory properties of cocoa flavanols that were ingested from cocoa-rich foods might have been too weak to reduce inflammation in pre-/hypertensive subjects and in patients with existing CAD.
The anti-inflammatory effect of cocoa depends on the dose of cocoa flavanols ingested [66], but data on total flavanol intake based on HPLC analysis of individual flavanols are only provided by single studies. As the epicatechin content in cocoa products correlates strongly with the sum of catechin, epicatechin, dimer B2, dimer B5, trimer C1, and tetramer D (r = 0.99) [80], epicatechin intake with cocoa reflects total flavanol intake. Therefore, the dose of epicatechin, which is believed to be responsible for the anti-inflammatory effects of cocoa consumption [31], has to be considered for the interpretation of the results. However, data on epicatechin intake were not available from all studies. Apart from the dose, efficacy on inflammatory biomarkers may depend on the food matrix. A higher bioavailability can be achieved by using a liquid matrix (cocoa drink) instead of solid chocolate, by co-administration of carbohydrates (e.g., by sugar or a carbohydrate-rich meal), and by the lack of milk protein [23]. Most studies providing flavanol-rich chocolate did not find anti-inflammatory effects [45,47,48,51,58,61,65,68] except for four studies [43,49,52,69] in which, however, only single parameters changed. On the other hand, nine out of 14 RCTs providing a flavanol-rich cocoa drink observed anti-inflammatory effects in one [44,55,57,60] or even several parameters [46,64,66,67]. However, methodological differences concerning participants (age, BMI, health status, or stage of vascular inflammation), the kind of intervention (food matrix, dose of epicatechin ingested), and the biomarkers used make a comparison between studies difficult. In regular consumption trials, the duration of cocoa treatment might also be relevant if epicatechin would accumulate in vivo. A reduction in ICAM-1 and ICAM-3 as well as a decreased expression of several adhesion molecules on leukocytes was observed after daily consumption of 70 g dark chocolate by overweight men, irrespective of whether the chocolate was enriched with flavanols or not [81]. This suggests that an epicatechin intake >97 mg through dark chocolate does not lead to stronger anti-inflammatory effects in overweight men. If other patients or foods were used, the result might have been different. For these reasons, a minimum dose that is needed to reduce inflammation cannot be ruled out.
Furthermore, the choice of parameters has to be considered when interpreting the results. CRP did not change in most studies despite the use of highly-sensitive CRP kits (Tables 1-5), which are demanded for subjects with cardiovascular risk or established cardiovascular diseases [82]. In bolus studies, the lack of changes in CRP 2 h [45,48] and 6 h [50] after bolus consumption of cocoa is not surprising as the CRP's reaction in vivo to stimuli such as IL-1β, IL-6, and TNF-α takes about 6 h [83]. In regular consumption trials, the lack of changes in CRP indicates that cocoa consumption does not affect systemic inflammation, and the main stimuli for CRP synthesis, IL-1β, IL-6, and TNF-α, did not change in most trials. However, due to their short half-life time in vivo (for TNF-α only 5 min) and their low stability in serum or plasma samples, the concentrations in vivo may be underestimated [84]. This is likely as several adhesion molecules whose gene expression is induced by TNF-α and IL-1β [85] decreased in serum/plasma in several studies. As shown in Tables 1-5 nearly 50% of all trials found a reduction in ICAM-1 (five studies [46,49,52,64,66] out of 11 [46,49,51,52,57,60,62,64,66,69]) and in P-selectin (three studies [44,62,64] out of six [44,49,60,62,64,69]) after cocoa treatment. In contrast, VCAM-1 decreased in a single trial [60] (out of nine [46,49,52,56,57,60,62,64,69]) and E-selectin did not change in any of six RCTs [49,59,60,62,64,69]. For the interpretation of the results, it is important to consider that E-selectin and VCAM-1 are only expressed by activated endothelial cells, and P-selectin additionally by platelets, whereas ICAM-1 is expressed by a variety of cells: activated endothelial cells, but especially by unstimulated and stimulated leukocytes [85,86]. With regard to the origin of adhesion molecules in serum/plasma, the reduction of ICAM-1 and the lack of changes in VCAM-1 may reflect a reduced activation of leukocytes by cocoa. Nevertheless, caution is necessary when interpreting the results. Changes in serum/plasma may not only reflect changes in the expression of the cell surface, but also changes in the degree of shedding and/or internalization. Moreover, shed molecules may interact with counter receptors or soluble ligands, and thus may not be detected [87]. However, the reduced expression of VLA-4, CD36 (scavenger receptor of oxidized LDL) [64], and CD62 L [59] on monocytes and the decrease in 4-series leukotrienes in plasma after cocoa consumption indicates lowered monocyte activation. This is plausible as leukotriene B 4 and oxidized LDL, which are known stimuli for monocytes [88], have been shown to decrease through cocoa consumption [43,89]. T cells are involved in atherogenesis [90], but their expression of LFA-1, VLA-4, and CD40 was not modulated by cocoa consumption [64], suggesting that T cells were not modulated. Thus, our results suggest that anti-inflammatory changes after cocoa consumption may be detected by cellular biomarkers (especially on monocytes) rather than by parameters in the serum/plasma. Microparticles derived from endothelium, platelets, and leukocytes induce the transcription and secretion of proinflammatory cytokines and the expression of adhesion molecules and have been considered as biomarkers of vascular injury and inflammation [91]. If we also bear in mind that both studies investigating endothelial microparticles found reduced concentrations in plasma after consumption of flavanol-rich cocoa [59,72], endothelial microparticles are promising parameters for future studies with a focus on vascular inflammation.
With regard to the study quality (Table 6), most trials considered were crossover studies using washout periods, suggesting a well-performed study design. However, allocation concealment and compliance with intervention were not always reported and participants not always blinded. Masking of participants is difficult due to the characteristic color and taste of cocoa, but it is possible, especially through the use of beverages. If we also bear in mind that inflammation was not the primary aim in most studies except for Monagas et al. [64], which is the only study with sample size calculation on the basis of inflammatory biomarkers, evidence for anti-inflammatory effects achieved by cocoa consumption is rather low.

Conclusions
In conclusion, consumption of flavanol-rich cocoa may reduce inflammation, probably by a reduced activation of monocytes and neutrophils. This may prevent or even reduce vascular inflammation. The efficacy of this measure probably depends on the extent of vascular inflammation, but also on the kind of cocoa product used. However, the evidence for anti-inflammatory effects of cocoa consumption is currently low. Further RCTs with inflammation as primary outcome marker are needed. These should investigate specific markers of leukocyte activation not only in the serum or plasma, but also in leukocytes. Endothelial microparticles should also be determined. Subjects suffering from low basal inflammation are of great interest as well as situations accompanied by an increased inflammatory burden (e.g., postprandial state, oral glucose tolerance test). In future trials, researchers and participants should be blinded and compliance with cocoa treatment should be determined to improve the studies' quality.