A comparison of the effects of green tea and cocoa on glycaemic control and insulin sensitivity in patients with type 2 diabetes mellitus: a systematic review and meta-analysis

1.1Diabetes management strategies and dietary flavonoids

Type 2 diabetes mellitus (T2DM) are among the most frequent chronic diseases in virtually all countries, posing a global health threat [1]. As of 2017, T2DM affected more than 8.4% of individuals globally; by 2045, this percentage is expected to climb to 9.9% [2]. Preventive interventions are critical for reducing the enormous burden of diabetes. Although drugs may help postpone the onset of diabetes complications, long-term use can be expensive and have various adverse effects [3]. Therefore, diabetic patients and doctors contemplate alternative and complementary treatments, such as herbal medications with anti-diabetic effects [4]. Indeed, plants have traditionally been a valuable source of pharmaceuticals, and many currently accessible medications are produced directly or indirectly from plants [5]. The herbal remedies may be promising for controlling diabetes.

Lifestyle changes can slow the evolution of T2DM in prediabetic people with impaired glucose tolerance or fasting glucose [6–10]. For example, Flavanols, a subset of plant-derived phytochemicals known as flavonoids, have recently received increased attention after epidemiological studies found a negative relationship between flavanol consumption and the development of diabetes [11, 12] Flavanols may assist people with metabolic disorders in maintaining glucose homeostasis by inhibiting digestive enzymes and glucose transporters, a strategy that assists people with metabolic disorders in maintaining glucose homeostasis [10;13]. In vitro studies have demonstrated that flavanols help inhibit carbohydrate digestive enzyme activity, enhance insulin sensitivity, inhibit hepatic gluconeogenesis by reducing gluconeogenic genes, and improve oxidative stress [14–17].

GT and cocoa have gotten much attention since they’re high in flavanols. Cocoa, GT, grapes, and apples are among the highest sources of flavonols [18]. After the consumption of flavanol-rich food, including cocoa, and tea, various polyphenol-derived phenolic acids, for example, m-hydroxyphenyl acetic acid, m hydrox-yphenylpropionic acid, m-hydroxybenzoic acid, and phenyl-g-valerolactones are generated [19–21]. Furthermore, cocoa may have a favourable effect on insulin signalling, possibly by increasing the incretin response or reducing oxidative stress and inflammation in the insulin-signalling pathways [22]. According to systematic reviews and meta-analyses, chocolate/cocoa improves FBI and insulin resistance in healthy adults [23–29]. The analysis of RCTs included in the review by [10] shows that cocoa consumption lowered FBG.

Catechins, flavonoid-like polyphenols, are primarily responsible for GT’s beneficial effects on human health [3]. These catechins include epigallocatechin-3-gallate (EGCG), as well as quercetin, theaflavins, theanine, caffeine, chlorogenic acid, and gallic acid [30–32]. Epicatechin (EC), epigallocatechin (EGC), epicatechin gallate (ECG), and epigallocatechingallate are the four primary catechins present in green tea extract (GTE) [33;34]. Moreover, epidemiological studies in humans have suggested that long-term use of GT is related to a lower risk of diabetes [35, 36]. Several randomised controlled studies (RCTs) have found that consuming GTE regularly can improve oral glucose tolerance in healthy persons and lower FBG and HbA1c levels in people at risk of diabetes [37–40].

However, systematic studies and meta-analyses on the link between GT intake, insulin resistance, and glycaemic control have produced conflicting results. For example, GT has been shown to enhance glycaemic control in some trials [41], whereas others have found no link between GT consumption and glycaemic management [42, 43]. According to a meta-analysis of seven RCTs by [44], the consumption of GT does not decrease FBG, FBI, HbA1c or HOMA-IR levels in populations at risk of T2DM. Another meta-analysis of 22 RCTs showed that the administration of GT with or without caffeine significantly reduced FBG. Still, the results did not support a positive effect on FBI, HbA1c or HOMA-IR in adults with [45]. In 27 RCTs, short-term GT supplementation significantly reduced FBG but did not significantly affect FBI or HbA1c in T2DM patients [3]. In the results of six RCTs, no evidence supports the suggestion that consuming GT/GTE could reduce HbA1c, HOMA-IR, FBG or FBI levels in adults with prediabetes and T2DM [31]. However, this survey study concluded that the benefits of GT ingestion were due to caffeine and not catechins [46]. The meta-analysis indicated that the supplementary intake of GT had no significant effect on FPG, FBI, HbA1c or HOMA-IR in patients with T2DM [47].

1.2Rationale for the review

Even among researchers that advocate polyphenol consumption, GT and/or cocoa consumption for T2DM is debatable. For example, according to [48, p. 303], “Indeed, tea, red wine and cocoa are good sources of flavanols (a class of polyphenol), but it is not completely established whether these foodstuffs exert a beneficial effect against diabetes.” In addition, [3, p. 2], ‘findings from randomized controlled trials (RCTs) on GT supplementation and glycaemic control or insulin sensitivity in individuals with preclinical diabetes or T2DM are debatable’. The focus was on consuming green tea and cocoa compared to other types reported to contain more abundant flavonoids because tea and cocoa consume more than other foods. For example, Cocoa and its product, chocolate, is a primarily consumed food worldwide [49]. Furthermore, according to the 2013 World Tea Market Report, tea is the most consumed beverage in the world [50]. Cocoa and GT are rich sources of various polyphenol compounds [3, 51]. We want to investigate what the most beneficial for diabetics, which are considered well available. Therefore, this research systematically reviews all existing studies that have evaluated the outcomes of GT and cocoa supplementation in adults with T2DM to assess the effectiveness of polyphenol supplementation in improving glycaemic control and insulin sensitivity. According to [52], FBG measurements do not adequately indicate glycaemic management because they do not reflect what happens after meals and during the day in free-living situations. Therefore, this systematic review and meta-analysis aimed to investigate the effects of polyphenol supplementation (GT and cocoa) on improving glycaemic control and insulin sensitivity in adult patients with T2DM through outcomes FBG, FBI, HbA1c and HOMA-IR.

2.1Types of studies

This study included every pertinent experimental study (including randomized controlled trials with either a parallel or crossover design), regardless of whether a double-blind method was used. Furthermore, reviews, papers and meeting abstracts were examined to obtain more evidence. The inclusion and exclusion criteria were applied as described below.

2.2Participants

Adult patients with T2DM were included in this study regardless of ethnicity.

2.3Intervention/Exposure types

The included studies conducted interventions that were relevant to the effect of GT and/or cocoa on T2DM, and some had information regarding the intensity and/or duration of their studies. The sources of this information were included where available.

2.4Types of outcome measures

In this review, the effectiveness of GT and/or cocoa on glycaemic control and insulin sensitivity in patients with T2DM was assessed using the following: (1) fasting glucose, (2) fasting insulin, (3) HbA1c and/or (4) HOMA-IR.

2.5Search strategy

To identify original research articles, Web of Science, PubMed/MEDLINE, Scopus, and Cochrane were used. Retrieved references and reviews were also examined to select relevant articles. First, the keywords/terms used for green tea were ‘ tea’ OR ‘green tea’ OR ‘green tea extract’ OR ‘tea polyphenols’ OR ‘green tea catechins’ OR ‘catechin’ OR ‘epigallocatechin’ OR ‘epicatechin’ OR ‘tea component(s)’ OR ‘epigallocatechin gallate’ AND ‘glycated protein’ OR ‘insulin sensitivity’ OR ‘hyperglycemia’ OR ‘insulin resistance’ OR ‘diabetes’ OR ‘glucose’ OR ‘glycaemic control’ OR ‘HbA1c’ OR ‘fructosamine’ OR ‘insulin’ OR ‘type 2 diabetes’. Second, the keywords/terms used for cocoa were ‘Cocoa’ OR ‘Cacao’ OR ‘Chocolate’ OR ‘dark chocolate’ OR ‘Cocoa powder’ OR ‘dark chocolate’ OR ‘cocoa flavanols OR ‘cocoa polyphenols’ AND ‘glycated protein’ OR ‘insulin sensitivity’ OR ‘hyperglycemia’ OR ‘insulin resistance’ OR ‘diabetes’ OR ‘glucose’ OR ‘glycaemic control’ OR ‘Hb A1c’ OR ‘fructosamine’ OR ‘insulin’ OR ‘type 2 diabetes’.

2.6Identification of relevant studies

Studies were selected for close examination based on the abstracts of the articles in the studies that were examined to ensure their eligibility for inclusion. Subsequently, these studies were evaluated and verified for inclusion or exclusion based on a thorough review of the full texts. The last search for data was performed on January 3, 2022.

2.7Data extraction

Related findings were derived from the selected studies, and the data extracted from the studies include the name of the author (including the year of publication), country where the study was conducted, study design, study period, population number (total enrolled/completers) and participants’ characteristics, exclusion criteria, intervention type, dose, and duration (intervention/control group), main/other outcomes and p-value.

2.8Risk of Bias assessment

The Risk of Bias 2 (RoB 2) tool is an update to the original risk of bias tool launched in 2008 [53]. RoB 2 tool was used to appraise the risk of bias in the randomized controlled trials included in this review. Bias is assessed as a high, low or some concerns judgment. The following criteria were considered in the evaluation: (1) Randomisation process, (2) Deviations from the intended interventions, (3) Missing outcome data, (4) Measurement of the outcome, and (5) Selection of the reported result. In addition, we used RoB 2 to create a funnel chart to analyse the bias of the studies in this systematic review.

2.9Statistical analysis

To supplement the systematic review, a meta-analysis was performed. We used Review Manager 5.4 software to assess the heterogeneity of indicators between the GT group and the control group, as well as the cocoa group and control group. The mean difference (MD) between the baseline and final measurements of fasting glucose, fasting insulin, HbA1c and HOMA-IR was calculated, and the indexes between the GT, cocoa, and control groups were compared. In this review, heterogeneity was evaluated by the index of inconsistency (I²), and the statistical value of I² reflected the level of heterogeneity. The result was considered heterogeneous if I² was more than 50%. A random-effects model was employed; otherwise, a fixed-effects model was used. In this review, a P-value<0.05 was considered statistically significant. Some studies did not mention the standard deviation of change. In this case, previous systematic reviews were checked, and the data was obtained. The study [38] from [44] as well as [54] A, [55] and [56] from [31]. [54, 57–62] by calculating S.D. [63]:

3.1Study selection

A total of 3329 and 1723 studies that evaluated the effect of GT and cocoa, respectively, on glycaemic control and insulin sensitivity in patients with T2DM were identified by searching the literature, as shown in Fig. 2. A total of fifteen articles on GT [38, 40, 46, 54–56, 61, 62, 64–67] and seven on cocoa [57–60, 68–70], which were published between 2005 and 2021, met the inclusion criteria.

Fig. 1

Flowchart of study selection and exclusion in detail.

Fig. 2

Risk of bias summary: review authors’ judgements about each risk of bias item for each included study in green tea.

3.2Study characteristics

3.3Participants’ characteristics

The studies that focused on GT had 738 participants (348 male/390 female), while the ones that focused on cocoa had 331 participants (83 male/248 female) with T2DM, aged 18–65 years. All the participants were overweight/obese (BMI: 25–34.9 kg/m²), except for the participants in [64] study (which focused on GT) and [68 and 70] studies (which focused on cocoa) (BMI < 25 kg/m²).

Most of the included studies did not report the duration of the diagnosis of T2DM, except for six studies on GT [55, 56, 62, 65–67] and five studies on cocoa [58, 59, 68, 69, 70]. The duration of the diagnosis of diabetes reported by these studies was less than eight years, except for one study on GT that reported a duration > 12 years [66].

3.4Quality assessment and publication bias

Based on the assessment tool, only four studies on GT were of low risk [40, 62, 65, 66]. In contrast, just two studies among studies on cocoa were of low risk [59, 69]. Five studies on GT [46, 54, 56, 61, 67] and only two studies on cocoa [57, 70] had some concerns. Only three studies on GT [38, 55, 64] and three studies on cocoa [58, 60, 68] had high risk, as shown in Figs. 2 and 3.

Fig. 3

Risk of bias summary: review authors’ judgements about each risk of bias item for each included study in cocoa.

3.5Meta-analysis and data synthesis

In total, data from fifteen studies [38, 40, 46, 54–56, 61, 62, 64–67] that compared GT with placebos were included in these meta-analyses. On the other hand, seven studies [57–60, 68–70] that compared cocoa supplements with placebos were included in these meta-analyses. Forest plots depicting the results of the meta-analysis for each assessed outcome are shown in Figs. 4, 5.

Fig. 4

Meta-analysis results of green tea for each assessed outcome.

Fig. 5

Meta-analysis results of cocoa for each assessed outcome.

4.1Other factors that may influence the effectiveness of GT and cocoa on glycaemic control and insulin sensitivity

The changes in outcomes might be related to the different types of the population investigated (disease type), the clinical characteristics of patients (the severity of the disease), the type and dose of supplementation, the duration of intervention, and the sample size at the start of the study. Accordingly, several factors could help to show positive effects, including the duration of a diabetes diagnosis. This disparity might be more evident in diabetic patients who had a mean illness duration of approximately seven years. The majority of studies on cocoa reported that the duration of the disease or the time since diagnosis was less than seven years, while some studies on the effect of GT on T2DM reported the duration to be more than 12 years. For example, [70] found that cocoa led to lower levels of FBS and HbA1c compared to the control. However, while these changes were not high rates, they were considered significant compared to the control group. [40] found that GTE supplementation had a superior FBG-reducing impact in individuals with a diabetes history of fewer than five years than in those with a history of more than five years, while reduced insulin and lower HOMA-IR were only significant in people with illness for less than five years. Though a few studies have focused on the severity of T2DM (in terms of disease duration) in relation to the use of GT, the results have been difficult to ascertain. Accordingly, the clinical features of T2DM patients must be investigated in future research with a larger patient sample.

There may be a relationship between the dose amount and the extent of the effect on glycaemic control. However, the amount and type of polyphenols found in published research vary. [54] suggested that larger dosages may be required for diabetic individuals to detect clear results, such as drinking four cups of GT daily (1000 mg of flavanols). The number of polyphenols consumed, and the amount of insulin was also found to have a favourable association [61]. After 16 weeks of 856 mg EGCG, the HbA1c level and HOMA-IR index in the GTE group decreased significantly compared to the placebo group [65]. The GTE group (1344 mg of catechins) demonstrated substantial within-group alterations in the HOMA-IR index after 16 weeks of therapy, despite no significant changes in FBG or HbA1c [40]. [68] observed a significant reduction in HOMA-IR and an improvement in insulin sensitivity following a significant decrease in FBI concentrations within two months associated with 850 mg of flavanols from cocoa. [59] reported that glucose and lipid metabolism was unaffected by the regular consumption of 2.5 g of cocoa (207.5 mg of flavanols) over three months. [69] indicated that cocoa with 450 mg of polyphenols is ineffective in improving the FBG, FBI, and HbA1c levels in diabetic adults compared to WCG. Moreover, [56] demonstrated no significant differences in the intervention group’s FBG, HbA1c, and FBI levels within two months (with 240 mg of flavanols). From these studies it can be deduced that higher doses are more effective for T2DM, although researchers have reported that there were no restrictions on drinking GT in the control group. This could be the main reason for the results of the previous study.

4.2Limitations of the review

This systematic review and meta-analysis have some limitations that must be addressed. First, T2DM is influenced by several variables, including genetic, metabolic, environmental, behavioural, social, and cultural factors, all of which may diminish the link between polyphenol consumption and T2DM outcomes. The selected studies focused on the impact of GT and cocoa on improving glycaemic control and insulin sensitivity in adults who are overweight/obese with T2DM. Therefore, the effectiveness of other sources of polyphenol in patients newly diagnosed with T2DM might not be determined in this review. Furthermore, since the majority of the RCTs selected included short-term interventions (up to 12 weeks), any long-term effects of the intervention were limited to expectations. In addition, this review’s BMI range was 25–34.9 kg/m², so it was unable to identify the efficacy of GT or chocolate for people with a BMI of≤24 kg/m². Therefore, the difference may be due to the degree of obesity or disease intensity, which merits further investigation. Consequently, most population groups from Asia and Africa may not have similar results compared to their counterparts in United States and Europe. Many participants (67%) were adult women, which limited the popularisation of the results. Moreover, significant differences in the feeding protocols between RCTs, short follow-up periods, small sample sizes, high dropout rates, the characteristics and number of participants, and the differences in the form of intervention make it difficult to determine which polyphenols are the most effective for T2DM. There is a high degree of heterogeneity between the clinical and methodological RCTs included. For example, there was a substantial variation in macronutrient energy percentages. The probable influence of other lifestyle-related factors, which are not obvious in any of the studies included, is one of the biggest methodological challenges in any nutritional intervention. Nonetheless, the results of this meta-analysis may be limited because of the small number of eligible RCTs. Although a thorough search of the literature was conducted using limited search engines, there was bias in selecting RCTs that fit within the search engines’ limited reach.

Some limitations of the included studies should be considered as well. Most of the studies focused on adults with T2DM aged between 18 and 65; hence, the findings of these studies may not hold for those with different ages and levels of health status and are thus limited in their generalisability. Some of the studies did not use a placebo: seven focused on GT [38, 54, 61, 62, 64, 66, 67], and two focused on cocoa [68, 70]. Consequently, it may be difficult to avoid bias. [66] and [58] identified the significant discrepancies in HbA1c levels (≥9), while all the other studies had discrepancies of≤6, which could affect the validity of the findings. Furthermore, since most studies on cocoa were based on small sample sizes and a minority number of male participants, it was difficult to ensure any impact related to gender. In contrast, the sample size for the GT studies was somewhat acceptable, and the ratio of males to females was equal. However, a large number of dropouts was noted in some studies: exceeding 21% for studies by [67] and [68] and 10% for the study by [66]. Despite these limitations, [68] offered the longest intervention to date for assessing the effects of cocoa flavonoids on patients with T2DM.

Some studies measured dietary intake/adherence using food records [38, 46, 54, 59, 61, 67]. Only a few studies have studied dietary intake, and their outcomes might be incorrect. According to [71] and [72], approximately 30% of overweight people underreport their food intake. Most of the participants were overweight and obese in this review, and a lack of substantial differences between the two groups might be due to the control group’s lack of limits on GT consumption in [61]. Although dietary intake was monitored by a weekly intake questionnaire at each visit, this was not restricted in [60]. Some studies have reported that adherence to dietary restrictions was controlled by investigating nutrition status and nutritional intake, although physical activity was not assessed [38, 46, 54, 59, 61, 67]. Therefore, the research may have been underpowered due to confounding effects from the lifestyle changes that may have altered outcome factors. However, in studies by [58–60, 68, 69], the participants were instructed to maintain a normal diet and engage in regular physical activity. In addition to the long duration, this results in ideal and more realistic situations. [38, 58, 64] were crossover trial studies in which there was more/some concern that individuals might be able to distinguish between the two preparations. In its study design, [69] did not have the possibility of double-blinding, and the participants were aware of the intervention grouping type. For [68] and [69], the amounts of serum polyphenols were not assessed at the start and conclusion, which might have influenced the outcomes in both groups. [60] reported that the diet of all patients was not the same 24 h before each blood sampling; moreover, during the trial, the patient’s sleep, stress, and psychological condition were not examined as factors that could influence the findings. Moreover, some studies did not measure HbA1c and insulin resistance [46, 54–57, 59, 60, 62, 64, 66], which might have led to a less accurate evaluation of the changes in glycaemic control.