Effect of Cocoa Beverage and Dark Chocolate Intake on Lipid Profile in People Living with Normal and Elevated LDL Cholesterol: A Systematic Review and Meta-Analysis

Abstract

Cocoa beverage and dark chocolate are important dietary sources of polyphenol and have been hypothesised to improve the lipid profile. This systematic review and meta-analysis aimed to investigate the effect of cocoa beverage and dark chocolate intake on lipid profile in individuals living with normal and elevated LDL cholesterol. The question on whether cocoa beverage and chocolate differentially modify the lipid profile was also explored. A systematic literature search was conducted on PubMed and Cochrane Library on 26 February 2022 following the PRISMA guideline. Cocoa beverage and chocolate consumption had no significant effect on circulating concentrations of total cholesterol, LDL cholesterol, and triglycerides (p > 0.05, all), but favourably and significantly increased circulating concentration of HDL cholesterol by 0.05 (95% CI [0.02, 0.09]) mmol/L (p = 0.002). Changes in lipid profile were similar when comparing populations with normal vs. elevated LDL cholesterol (p > 0.05, all). When considering the food matrix, cocoa beverage intake significantly increased HDL cholesterol by 0.11 (95% CI [0.06, 0.17]) mmol/L (p < 0.001), but the improvement in HDL cholesterol was not significant when chocolate (p = 0.10) or a combination of cocoa beverage and chocolate (p = 0.19) (subgroup differences, p = 0.03) was administered. Cocoa consumption could be recommended as part of a healthy diet in the general population with normal and elevated LDL cholesterol.

1. Introduction

Dyslipidaemia is a condition associated with abnormal concentrations of lipid profile and is characterised by either elevated triglycerides, low-density lipoprotein (LDL) cholesterol, and total cholesterol, or decreased high-density lipoprotein (HDL) cholesterol concentrations, or their combination in the blood [1]. It remains an important risk factor for the development of cardiovascular diseases, such as ischaemic heart disease and ischaemic stroke [1], which were ranked amongst the top 10 causes of death worldwide in 2019 [2]. Even though the prevalence of elevated total cholesterol concentration (hypercholesterolaemia) has improved generally in developed countries owing to the increased adherence and use of cholesterol-lowering drugs, including statins and a general improvement in lifestyle factors, the situation is different for developing countries, including sub-Saharan Africa countries, such as Ghana [3,4]. For example, recently, a systematic review and meta-analysis of cross-sectional studies involving adult Africans, the authors reported dyslipidaemia prevalence of 17.0% for hypertriglyceridaemia, 25.5% for hypercholesterolaemia, 37.4% for low concentrations of HDL cholesterol, and 28.6% for elevated concentrations of LDL cholesterol [5]. Certain factors, including increased urbanisation with its attendant increased physical inactivity and higher intake of poor-quality diets, have contributed towards the surge in dyslipidaemia prevalence [1].

Improving dyslipidaemia through dietary intervention is considered as a sustainable approach. In developing countries, the higher cost of cholesterol-lowering drugs, further aggravated by poor drug adherence, are some of the common reasons for uncontrolled dyslipidaemias. Several systematic reviews and meta-analysis of prospective studies [6,7] have reported that increased intake of energy-dense and nutrient-poor foods is positively associated with increased risk of dyslipidaemia. Foods that are rich in dietary fibre [8] especially soluble fibre, and some bioactive compounds present in plants are important in improving lipid profile [8]. Excretion of circulating cholesterol is known to be mediated by bile acids. Upon ingesting dietary fibre, the soluble fibre chelates bile acids, which are released during fat digestion, preventing bile acids from being re-absorbed from the intestinal lumen, resulting in bile acid excretion through faeces [9]. This inhibition of bile acid re-absorption stimulates liver to replenish the bile acids pool through its endogenous synthesis from cholesterol, consequently lowering the circulating concentration of total cholesterol. In the case of bioactives, especially polyphenols, some in vitro and animal studies proposed that certain polyphenols may lower endogenous cholesterol and fatty acid synthesis by modulating gene expressions [10].

Cocoa bean is a unique crop that has the potential to improve dyslipidaemia prevalence, as it is a good source of dietary fibre (3.1–19.4% on dry weight basis) [11] and a rich source of polyphenol, especially epicatechin [12]. Dried cocoa bean is the raw material for cocoa powder production and is naturally high in cocoa butter, consisting of fat ranging from 33.0% to 62.9% on dry weight basis [11]. In the final phase of the production process, cocoa liquor produced through the pulverisation of roasted deshelled cocoa beans is pressed to release the fat (butter) and cocoa cake. The cocoa cake is subsequently milled into cocoa powder, which serves as raw material for cocoa-based products, including dark chocolate. In comparison with cocoa powder, chocolate is generally more energy-dense, containing less fibre, proteins, and minerals (

Table 1. Nutritional composition of cocoa powder, cocoa butter, and dark chocolate [13].

Nutrient	Cocoa Powder, Unsweetened	Cocoa Butter	Dark Chocolate
Water (g/100 g)	3	0	1.37
Ash (g/100 g)	5.8	0	2.32
Total lipid (fat) (g/100 g)	13.7	100	42.6
Total dietary fibre (g/100 g)	37	0	10.9
Protein (g/100 g)	19.6	0	7.79
Carbohydrate, by difference (g/100 g)	57.9	0	45.9
Energy (kcal/100 g)	228	884	598
Calcium (mg/100 g)	128	0	73
Magnesium (mg/100 g)	499	0	228
Phosphorus (mg/100 g)	734	0	308
Potassium (mg/100 g)	1520	0	715
Sodium (mg/100 g)	21	0	20

) [13].Chocolate also has more additives, including emulsifiers and sugar, but a lower concentration of polyphenols [14]. Cocoa butter, on the other hand, is energy-dense and nutrient-poor (Table 1) [13].

Another feature that distinguishes chocolate from cocoa powder is that chocolate contains cocoa butter, which is composed of mainly saturated fatty acids (25% palmitic acid and 35% stearic acid), some monosaturated fatty acids (35% oleic acid), and a small amount of polyunsaturated fatty acid (3% linoleic acid) [15]. Whilst standard dietary recommendation generally mention replacing saturated fat intake with mono- and poly-unsaturated fat intake for improving lipid profile outcomes [16], there is a lack in understanding the effect of chocolate on differentially modifying lipid profile when compared to fat-free cocoa powder beverages.

Previous systematic reviews and meta-analyses have produced contradictory results on the effect of cocoa and chocolate intake on lipid profile. This could be attributed to inconsistencies in the participant characteristics, dose of bioactives in the cocoa or dark cholate administered, and the food matrix. In an earlier review, Jia et al. [17] reported that consuming cocoa or dark chocolate at a lower dose (≤260 mg polyphenol/day) in the short term favoured significant reductions in total cholesterol and LDL cholesterol in individuals with cardiovascular risk factors compared to healthy counterparts, whereas medium (260–665 mg polyphenols) and high (≥665 mg polyphenols) doses had no significant effect. Other studies have produced contradictory results regarding the effect of epicatechin and flavanol doses on the serum lipid profile. For example, Hooper et al. [18] showed that ingestion of 50–100 mg epicatechin per day favourably lowered triglycerides but not at higher doses, while no effect on HDL, LDL, or total cholesterol was observed. In contrast, Lin et al. [19] reported that a daily intake of 166–2110 mg flavanol had a significant effect on triglycerides and HDL cholesterol, with no effect on LDL and total cholesterol.

Therefore, the objective of this systematic review and meta-analysis was to investigate and provide an update on (i) the effects of cocoa and chocolate consumption on lipid profile (triglycerides, LDL, HDL, and total cholesterol) in individuals with normal and elevated LDL cholesterol concentrations, (ii) whether cocoa and chocolate differ in their effects on lipid profile, and (iii) to verify the threshold concentration of flavanol and epicatechin necessary to improve lipid profile.

2. Materials and Methods

The present systematic review was conducted in accordance with the guidelines stipulated in the recently updated Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) statement [20]. Published articles that investigated the effect of cocoa beverage and dark chocolate intake on lipid profile were searched systematically on the databases PubMed and Cochrane Library on 26 February 2022. The authors carefully formulated the search term (cocoa OR cacao OR chocolate) AND (lipid profile OR dyslipidaemia OR total cholesterol OR triglyceride OR HDL OR high-density lipoprotein OR LDL OR low-density lipoprotein OR cardiovascular OR vascular), which was subsequently adopted for the article search process.

The use of filters to restrict articles to certain publication periods, dates, languages, and study types was avoided. The obtained records following application of the search terms to the databases were exported to the reference manager ENDNOTE™. Records that appeared as duplicate were removed and the finally selected articles were subjected to meet the inclusion and exclusion criteria set out for this systematic review.

2.1. Eligibility Criteria

The following inclusion criteria were adhered when selecting records for the systematic review:

(i)

Articles must be published as original research.

(ii)

Articles should be written in English and accessible in full-text.

(iii)

Articles must present data for lipid profile at baseline and at post-intervention or the change in lipid profile at post-intervention.

(iv)

The duration of the study should be equal or longer than 2 weeks.

(v)

The study must involve a comparison between a cocoa product used as a treatment, and either a non-cocoa product or a cocoa product with a negligible amount of polyphenol used as a control.

(vi)

The study must focus on an adult population, specifically individuals aged 18 years and above.

2.2. Exclusion Criteria

The exclusion criteria were as follows:

(i)

Studies published as editorials, review articles, conference proceedings, and commentaries.

(ii)

Studies that incorporated cocoa or chocolate intervention alongside other lifestyle interventions, such as exercise or weight-loss intervention.

(iii)

Studies that administered cocoa or chocolate supplemented with other nutrients.

(iv)

Studies with participants presenting significant comorbidities, including cardiovascular diseases and diabetes.

2.3. Data Extraction

Relevant study characteristics were extracted, including country of study, publication year, study design, sample size, age, and body mass index (BMI) of participants, daily doses of polyphenol, flavanol, or epicatechin, as well as study outcomes (lipid profile, including total cholesterol, HDL-cholesterol, LDL-cholesterol, and triglycerides).

This systematic review also focused on evaluating the effects of cocoa consumption on lipid profile in two distinct populations: (i) individuals living with elevated LDL cholesterol and (ii) individuals living with normal LDL cholesterol level. To achieve this, the study population was categorised based on baseline LDL cholesterol concentrations. The population with normal LDL cholesterol concentration is defined as having a mean LDL cholesterol concentration < 3.3 mmol/L, whereby the population with elevated LDL cholesterol concentration is defined as having a mean LDL cholesterol concentration ≥ 3.3 mmol/L [5].

The article search was conducted by four independent authors, including I.A., E.O.O., M.A. and J.C.C. In cases where there was a discrepancy regarding the inclusion or exclusion of identified records, consensus was reached between the authors I.A., E.O.O., M.A., J.C.C. and J.J.L.

2.4. Risk of Bias

A risk of bias assessment was carried out on all studies included the systematic review. The assessment was carried out according to the Cochrane risk of bias tool for randomised controlled trials using the headings: random sequence generation, allocation concealment, blinding of participants or study personnel, blinding of outcome assessment, incomplete outcome data, selective reporting, and other bias [21].

2.5. Meta-Analysis

Data extracted from original reports were input into Review Manager (Version 5.4.1, Cochrane Collaboration, Oxford, UK) for meta-analysis. We followed the methodology recommended in the Cochrane Handbook for Systematic Reviews of Interventions Version 6.2 [21]. Mean difference and standard deviation of the difference (SD_diff) were extracted as the effect size. In instances when the outcome was reported as post-intervention mean, adjusted for baseline value as covariate, this value was extracted as the effect size. When both mean difference and ‘adjusted’ post-intervention mean were missing, we calculated the mean difference by using post-intervention mean minus pre-intervention mean, and imputed SD_diff by using the following equation [21]:

$${\mathit{SD}}_{\mathit{diff}}=\sqrt{{{SD}_{post}}^2+{{SD}_{pre}}^2-(2\times Corr\times {SD}_{post}\times {SD}_{pre})}$$

where SD_post is the SD of post-intervention mean, SD_pre is the SD of pre-intervention mean, and Corr is the correlation between SD_post and SD_pre, which was assumed to be 0.5 based on Follmann et al. [22].

When the unit of variance was available as the standard error or confidence interval, we converted them into SD following the Cochrane guidelines [20]. When the data were only available in the form of graphs, we estimated the values using a PDF measurement tool (Adobe Acrobat Pro DC, Burlington, NJ, USA).

The active treatment required the administration of flavanol-containing cocoa or chocolate, whereas the control treatment required the administration of a non-cocoa product or a cocoa product with a negligible amount of polyphenol. In cases where a report consisted of multiple treatment comparisons, we selected the most appropriate treatment comparisons for this meta-analysis. When a report presented different doses of cocoa and/or chocolate, we compared the highest dose of cocoa and/or chocolate with the control. When a report consisted of multiple unique study population, each unique study population was considered as a unique treatment comparison.

The outcome of the meta-analysis was weighted mean difference (WMD) (95% confidence interval) in circulating concentrations of total cholesterol, LDL cholesterol, HDL cholesterol, and triglycerides, computed using a random effect model, presented as forest plots. Following our previous systematic review and meta-analysis on the effect of cocoa and chocolate on blood pressure outcome [12], we also designed pre-specified subgroup comparisons. These comparisons assessed the modifying effects of study duration (≤4 weeks vs. >4 weeks), food matrix (beverage, chocolate, or a combination of both), daily dose of polyphenol (<500 mg vs. ≥500 mg), daily dose of flavanol (<900 mg vs. ≥900 mg), and daily dose of epicatechin (<100 mg vs. ≥100 mg) on lipid profile. The arbitrary cut-off for flavanol and epicatechin was proposed to have favourable effect on flow mediated dilation [23] and blood pressure [12], but no cut-off or threshold was previously proposed for improving lipid outcome. Hence, we would like to assess if the threshold for improving flow mediated dilation and blood pressure could also improve lipid profile, all of which would contribute to the improvement in cardiovascular health.

To evaluate the magnitude of between-study and between-subgroup heterogeneity, I² statistics was used. I² values of ≤40%, 30–60%, 50–90%, and 75–100% indicate low, moderate, substantial, and considerable heterogeneity, respectively. We also evaluated the robustness of the results by conducting sensitivity analysis using leave-one-out method [24] and removing ambiguous data.

3. Results

A total of 1870 records were retrieved from the databases PubMed and Cochrane Library. Following the removal of 241 duplicate records, a total of 1629 articles were obtained. The 1629 articles were screened for their titles and abstract resulting in the production of 48 records. Of the 48 records, all were retrieved and assessed for eligibility. The eligibility check was carried out using the inclusion and exclusion criteria. A total of 30 reports were identified as eligible, but only 24 reports were eligible for meta-analysis. The detailed description of the search and study selection process is detailed below (Figure 1).

3.1. Study Characteristics

Of the 24 reports eligible for meta-analysis, there were 29 unique treatment comparisons, as some reports [25,26,27,28] had more than one study population or treatment comparison. The study characteristics and outcomes are summarised in

Table 2. Summary of study characteristics and lipid profile outcomes arranged in alphabetical order by the first author’s surname.

Authors	Country of Study	Age (Years)	BMI (kg/m2)	Sample Size	Study Design	Baseline LDL-C a	Intervention Duration (Weeks)	Control	Intervention	Daily Dose of Polyphenol (mg)	Daily Dose of Flavanol (mg)	Daily Dose of Epicatechin (mg)	Outcome
Al-Faris [29]	Saudi Arabia	21 ± 2	19.4 ± 0.4 – 22.0 ± 1.5	89	C, PT	Normal	2	White Chocolate	Dark chocolate	500	—	4	≠∆ TC, HDL-C, LDL-C, and TG.
Baba et al. [30]	Japan	49 ± 9	24.2 ± 3.5	160	DB, P, CT	Elevated	4	Placebo	Cocoa powder	—	—	64.5/96.7/129	↑ HDL-C after cocoa intervention. No significant difference between doses. ≠∆ TC, LDL-C, and TG.
Baba et al. [31]	Japan	38 ± 5	22.1 ± 1.0	25	R, C, PT	Elevated	12	12 g sugar	Cocoa powder + 12 g sugar	—	—	98	↑ HDL-C (cocoa vs. control). ≠∆ TC, LDL-C, TG.
Crews Jr et al. [32]	USA	69 ± 9–69 ± 8	25.2 ± 3.4 – 25.5 ± 3.6	90	R, DB, PC, PT	Elevated	6	Placebo	Dark chocolate bar and cocoa beverage	—	—	—	≠∆ TC, HDL-C, LDL-C, and TG.
Davison et al. [33]	Australia	44 ± 4–46 ± 4	32.8 ± 1.1 – 34.5 ± 1.8	49	R, DB, PC, FPT	Elevated	12	LF (36 mg flavanol) cocoa beverage	HF cocoa beverage	—	902	—	≠∆ TC, HDL-C, LDL-C, and TG.
Engler et al. [34]	USA	32 ± 3–33 ± 3	21.9 ± 1.6 – 23.2 ± 1.7	21	R, DB, PC, PT	Normal	2	LF chocolate (trace amount epicatechin)	Dark chocolate bar	—	—	46	≠∆ TC, HDL-C, LDL-C, and TG.
Fraga et al. [35]	Argentina	18 ± 1	24.1 ± 0.2	28	R, C, CT	Normal	2	Cocoa butter chocolate	Flavanol-containing milk chocolate	—	168	39	↓ TC and LDL-C (flavanol cocoa milk chocolate not cocoa butter chocolate) ≠∆ TG and HDL
Garcia-Yu et al. [36]	Spain	57 ± 4	25.6 ± 3.1 – 25.7 ± 3.8	137	R, DB, C, PT	Normal	26	No product	Chocolate	65.5	—	26.1	≠∆ TC, HDL-C, LDL-C, and TG.
Grassi et al. [37]	Italy	34 ± 8	22.6 ± 3.0	15	R, CT	Normal	2	White chocolate	Dark chocolate	500	—	—	≠∆ TC, HDL-C, LDL-C, and TG.
Grassi et al. [38]	Italy	44 ± 8	25.4 ± 1.7	20	R, CT	Elevated	2	White chocolate	Dark chocolate	—	88	66	↓ TC (Dark chocolate not white chocolate) ↓ LDL-C (Dark chocolate compared with white chocolate). ≠∆ HDL-C and TG.
Koli et al. [39]	Finland	46 ± 8	27.7 ± 3.7	22	R, C, CT	Elevated	8	Reduced snack intake	Replaced snack intake with dark chocolate	—	603	—	≠∆ TC, HDL-C, LDL-C, and TG.
Lee et al. [40]	USA	46 ± 10	29.6 ± 2.8	31	R, SB, C, CT	Elevated	4	No product	Cocoa powder + dark chocolate	—	—	—	≠∆ TC, HDL-C, LDL-C, and TG.
Martinez-Lopez et al. [26]	Spain	26 ± 6 (NC group); 30 ± 10 (HC group)	23 ± 3 (NC group); 24 ± 3 (HC group)	24 (NC group); 20 (HC group)	NR, C, CT	Normal (NC group); Elevated (HC group)	4	Milk	Cocoa powder + milk	—	45.3	18.9	↑ HDL-C (both cocoa and milk groups). No significant difference between groups. ≠∆ TC, LDL-C, and TG.
McFarlin et al. [25]	USA	21 ± 2–22 ± 3	21.6 ± 1.9 (NW group); 27.0 ± 1.4 (OW group); 34.9 ± 3.0 (OB group)	24	R, DB, PC, CT	Normal (all NW, OW, and OB groups)	4	Placebo	Chocolate	—	640	48	↑ HDL-C (chocolate compared to placebo). ≠∆ TC, and TG.
Muniyappa et al. [41]	USA	51 ± 7	33.2 ± 6.3	20	R, DB, PC, CT	Elevated	2	Placebo	Cocoa beverage	902	—	174	≠∆ TC, HDL-C, LDL-C, and TG.
Mursu et al. [42]	Finland	19–49	21.5 ± 2.9 – 24.1 ± 3.5	45	NR, PT	Normal	3	White Chocolate	Dark Chocolate	274/418	—	151.5/227	↑ HDL-C (Dark chocolate compared to white chocolate). ≠∆ TC, LDL-C, and TG.
Neufingerl et al. [43]	Netherlands	53 ± 9–56 ± 8	23.8 ± 2.5 – 24.9 ± 3.1	143	R, DB, PC, FPT	Elevated	4	Placebo	Cocoa beverage	—	325	—	≠∆ TC, HDL-C, LDL-C, and TG.
Njike et al. [28]	USA	53 ± 11–54 ± 10	29.9 ± 4.2 – 30.7 ± 5.0	122	R, DB, C, MLSPT	Normal	8	Placebo	Chocolate and cocoa powder	—	257/514	24/46	≠∆ TC and LDL-C. ↑ HDL-C (High dose cocoa compared with baseline, no significant difference between groups). ↑ TG (Low dose cocoa compared with baseline, no significant difference between groups).
Rull et al. [44]	UK	55 ± 8	26.6 ± 2.8	32	R, DB, PC, CT	Elevated	6	LF dark chocolate (88 mg flavanol)	HF dark chocolate	—	1064	—	≠∆ TC, HDL-C, LDL-C, and TG.
Sansone et al. [45]	Germany	44 ± 9–45 ± 8	25 ± 3 – 26 ± 3	100	R, DB, PC, PT	Normal	4	Placebo beverage + theobromine + caffeine	Fruit-flavored cocoa beverage	—	900	128	↓ TC, LDL-C and ↑ HDL-C (Cocoa beverage compared to placebo.)
Sarria et al. [27]	Spain	26 ± 6 – 38 ± 8 (NC group); 25 ± 7 – 36 ± 11 (HC group)	22.0 ± 2.6 – 24.1 ± 3.6 (NC group); 22.4 ± 2.3 – 26.2 ± 4.2 (HC group)	24 (NC group); 20 (HC group)	R, C, CT	Normal (NC group); Elevated (HC group)	4	Milk	Cocoa + milk	416	—	—	↑ HDL-C (cocoa compared to milk) ≠∆ TC, LDL-C, and TG.
Shiina et al. [46]	Japan	30 ± 4	22.6 ± 2.0	39	R, SB, PT	Normal	2	White Chocolate	Dark chocolate	550	—	—	≠∆ TC, HDL-C, LDL-C, and TG.
Taubert et al. [47]	Germany	64 ± 5	24.0 ± 1.6	44	R, SB, C, PT	Normal	18	White Chocolate	Dark chocolate	30	—	5.1	≠∆ TC, HDL-C, LDL-C, and TG.
West et al. [48]	USA	53 ± 2	27.8 ± 3.9	30	R, DB, PC, CT	Normal	4	LF chocolate (43 mg flavanol) + placebo beverage	Dark chocolate + cocoa beverage	—	814	—	≠∆ TC, HDL-C, LDL-C, and TG.

Age and BMI are reported as mean ± standard deviation and/or range. LDL-C ^a ≥ 3.4 mmol/L is classified as elevated. Symbols and abbreviations: —, not reported; ≠∆, No change; ↓, significant decrease; ↑, significant increase; C, controlled; CT, cross-over trial; DB, double-blind; FPT, factorial parallel trial; HC, hypercholesterolaemic; HDL-C, high-density lipoprotein cholesterol; HF, high-flavanol; LC, low-flavanol; LDL-C, low-density lipoprotein cholesterol; MLSPT, modified Latin-square parallel trial; NC, normocholesterolaemic; NR, non-randomised; NW, normal weight; OB, obese; OW, overweight; PC, placebo controlled; PT, parallel trial; R, randomised; SB, single-blind; TC, total cholesterol; TG, triglycerides.

.

In general, the studies had mean age between 18–69 years old, mean BMI between 19.4–33.2 kg/m². Twelve studies (50%) employed a parallel design, whereas 12 studies (50%) employed a cross-over design. At baseline, 12 studies included participants with normal LDL cholesterol, whereas 10 studies included participants with elevated LDL cholesterol. Of note, two studies [26,27] compared the effect of treatments between participants with normal and elevated LDL cholesterol. The studies had intervention durations between 2–26 weeks, with a median duration of 4 weeks. Dark chocolate and cocoa powder beverages were commonly used products in the active treatment group, except Fraga et al. [35], who used flavanol-containing milk chocolate as an intervention product. A total of 11 studies (14 treatment comparisons) delivered cocoa in the format of chocolate, 8 studies (10 treatment comparisons) delivered cocoa in the format of a beverage, whereas 4 studies (5 treatment comparisons) delivered cocoa in the format of both chocolate and a beverage. Contrary to the active treatment group, products used in the control group were more variable. When the cocoa was delivered as chocolate, the control products were placebo chocolate, white or milk chocolate, low-flavanol chocolate, or non-cocoa snacks. Of note, in two studies [36,39], the control group did not receive any products. When cocoa powder was delivered as a beverage, the control products were a placebo cocoa powder beverage, a low-flavanol (negligible amount) cocoa powder beverage, sugar, and milk. Whilst polyphenols, flavanols, and epicatechin were the bioactive compounds hypothesised to promote improvement in lipid profile, only a small number of studies reported the daily dose of these bioactive compounds delivered by their intervention products. Of the 24 studies eligible for meta-analysis, only 8 studies (9 treatment comparisons) reported the polyphenol content (median: 459 mg, range: 30–902 mg), 11 studies (15 treatment comparisons) reported the flavanol content (median: 603 mg, range: 45.3–1064 mg), and 14 studies (18 treatment comparisons) reported the epicatechin content (median: 47 mg, range: 4–227 mg). When grouped by the pre-specified cut-off, only four studies reported polyphenol content ≥ 500 mg, three studies reported flavanol content ≥ 900 mg, and four studies reported epicatechin content ≥ 100 mg.

3.2. Risk of Bias Assessment

The risk of bias assessment is summarised in Figure 2. Of the 24 studies, only 9 studies clearly reported the random sequence generation, while 11 randomised studies did not describe the random sequence generation. Of note, four studies were regarded as at high risk of selection bias in random sequence generation, including randomisation not being described in the study by Al-Faris [29], whereas the studies of Baba et al. [30], Martinez-Lopez et al. [26], and Mursu et al. [42] were non-randomised trials. Concerning allocation concealment, nine studies had reported allocation concealment using appropriate methods. Double blinding was performed in 12 studies. Double blinding was better maintained when comparing chocolate or cocoa beverage with placebo chocolate or placebo beverage. In contrast, double blinding was challenging when comparing dark chocolate with white or milk chocolate, as participants were aware of their treatment groups. In terms of outcome assessment, detection bias was regarded as low, since lipid profile was analysed in laboratories using standardised methods. Attrition bias was low in 20 studies. Of note, only four studies had a high risk of attrition bias due to high attrition rate (>15%). Neither reporting bias nor any other bias of significance were detected in our analysis. However, it was worth noting that Fraga et al. [35] exclusively recruited active soccer players as participants, and therefore, we conservatively regarded it as having an unclear risk of bias.

3.3. Meta-Analysis: Effect of Flavanol-Containing Cocoa Beverage and Chocolate Consumption on Lipid Profile

The pooled effects of flavanol-containing cocoa beverage and chocolate consumption on lipid profile are summarised in Figure 3, Figure 4, Figure 5 and Figure 6. The outcomes of pre-specified subgroup analyses are summarised in

Table 3. Subgroup analysis of the meta-analysis on the effect of cocoa beverages and chocolate on the lipid profile based on intervention duration, food matrix, daily dose of polyphenols, daily dose of flavanols, and daily dose of epicatechin.

Subgroups	WMD (95% CI) mmol/L	p-Value	I2 (%)	n	Subgroup Differences (p-Value)
Intervention Duration
Total Cholesterol
≤4 weeks	−0.06 [−0.15, 0.03]	0.22	48	20
>4 weeks	0.07 [0.02, 0.11]	0.002 a	0	9
Overall	−0.03 [−0.09, 0.03]	0.33	45	29	0.02 a
HDL Cholesterol
≤4 weeks	0.07 [0.03, 0.12]	0.002	30	20
>4 weeks	0.03 [−0.02, 0.07]	0.21	61	9
Overall	0.05 [0.02, 0.09]	0.002	52	29	0.16
LDL Cholesterol
≤4 weeks	−0.08 [−0.16, 0.00]	0.06	0	17
>4 weeks	0.01 [−0.07, 0.09]	0.83	38	8
Overall	−0.03 [−0.08, 0.03]	0.32	22	25	0.13
Triglycerides
≤4 weeks	−0.02 [−0.07, 0.04]	0.56	0	19
>4 weeks	−0.03 [−0.09, 0.03]	0.33	16	9
Overall	−0.02 [−0.06, 0.01]	0.16	0	28	0.77
Food Matrices
Total Cholesterol
Beverage	−0.05 [−0.17, 0.07]	0.40	0	10
Chocolate	−0.04 [−0.15, 0.06]	0.41	70	14
Beverage+Chocolate	−0.01 [−0.11, 0.08]	0.79	0	5
Overall	−0.03 [−0.09, 0.03]	0.33	45	29	0.86
HDL Cholesterol
Beverage	0.11 [0.06, 0.17]	<0.001	0	10
Chocolate	0.05 [−0.01, 0.11]	0.10	67	14
Beverage+Chocolate	0.02 [−0.01, 0.06]	0.19	0	5
Overall	0.05 [0.02, 0.09]	0.002	52	29	0.03
LDL Cholesterol
Beverage	−0.07 [−0.19, 0.05]	0.27	0	9
Chocolate	−0.04 [−0.14, 0.05]	0.39	42	11
Beverage+Chocolate	−0.03 [−0.13, 0.07]	0.58	12	5
Overall	−0.03 [−0.08, 0.03]	0.32	22	25	0.10
Triglycerides
Beverage	0.01 [−0.07, 0.10]	0.75	0	9
Chocolate	−0.03 [−0.07, 0.01]	0.12	0	14
Beverage+Chocolate	0.01 [−0.23, 0.25]	0.94	57	5
Overall	−0.02 [−0.06, 0.01]	0.16	0	28	0.64
Daily Dose of Polyphenols
Total Cholesterol
<500 mg	0.08 [0.04, 0.12]	<0.001 a	0	5
≥500 mg	0.10 [−0.07, 0.26]	0.27	0	4
Overall	0.08 [0.03, 0.12]	<0.001 a	0	9	0.86
HDL Cholesterol
<500 mg	0.06 [−0.04, 0.15]	0.23	77	5
≥500 mg	0.02 [−0.07, 0.12]	0.69	11	4
Overall	0.04 [−0.02, 0.11]	0.21	62	9	0.93
LDL Cholesterol
<500 mg	0.06 [0.03, 0.10]	<0.001 a	0	5
≥500 mg	0.00 [−0.16, 0.16]	0.99	0	4
Overall	0.06 [0.02, 0.10]	0.001 a	0	9	0.46
Triglycerides
<500 mg	−0.01 [−0.06, 0.03]	0.52	0	5
≥500 mg	−0.04 [−0.18, 0.10]	0.55	0	4
Overall	−0.02 [−0.06, 0.02]	0.42	0	9	0.71
Daily Dose of Flavanols
Total Cholesterol
<900 mg	−0.08 [−0.21, 0.04]	0.17	66	12
≥900 mg	−0.11 [−0.37, 0.15]	0.85	0	3
Overall	−0.08 [−0.19, 0.02]	0.12	58	15	0.85
HDL Cholesterol
<900 mg	0.05 [0.00, 0.10]	0.04	37	12
≥900 mg	0.06 [−0.05, 0.16]	0.27	0	3
Overall	0.05 [0.01, 0.09]	0.02	20	15	0.18
LDL Cholesterol
<900 mg	−0.05 [−0.16, 0.06]	0.36	28	10
≥900 mg	−0.12 [−0.38, 0.14]	0.37	0	2
Overall	−0.06 [−0.15, 0.04]	0.23	16	12	0.64
Triglycerides
<900 mg	−0.03 [−0.12, 0.05]	0.45	26	12
≥900 mg	−0.03 [−0.25, 0.19]	0.80	0	2
Overall	−0.04 [−0.11, 0.04]	0.34	7	14	0.97
Daily Dose of Epicatechin
Total Cholesterol
<100 mg	−0.05 [−0.15, 0.04]	0.26	70	14
≥100 mg	−0.04 [−0.23, 0.15]	0.71	0	4
Overall	−0.05 [−0.13, 0.03]	0.25	63	18	0.86
HDL Cholesterol
<100 mg	0.06 [0.01, 0.11]	0.02	65	14
≥100 mg	0.12 [0.04, 0.19]	0.002	0	4
Overall	0.07 [0.02, 0.12]	0.003	55	18	0.22
LDL Cholesterol
<100 mg	−0.05 [−0.14, 0.05]	0.35	57	11
≥100 mg	−0.15 [−0.33, 0.04]	0.12	0	4
Overall	−0.06 [−0.14, 0.03]	0.17	49	15	0.34
Triglycerides
<100 mg	−0.04 [−0.09, 0.02]	0.20	26	14
≥100 mg	0.03 [−0.21, 0.27]	0.79	0	3
Overall	−0.03 [−0.08, 0.02]	0.19	13	17	0.58

WMD, weighted mean difference; CI, confidence interval; I², heterogeneity; n, number of studies. Significance was presented as p-value. ^a The outcome became statistically non-significant (p > 0.05) after removing the data from Taubert et al. [47] using a leave-one-out method in sensitivity analysis.

.

A total of 29 treatment comparisons were included in the meta-analysis, of which all had reported total cholesterol and HDL cholesterol outcomes. Concerning the LDL cholesterol outcome, three treatment comparisons reported by McFarlin et al. [25] did not report data for LDL cholesterol, and hence, they could not be included in the analysis of LDL cholesterol outcome; conversely, Davison et al. [33] reported an increase in 2.02 ± 0.30 mmol/L in LDL cholesterol in the control group, approximately 100 times greater than the WMD (WMD: −0.03, 95% CI [−0.09, 0.03] mmol/L) identified in the current meta-analysis, and hence, the data were flagged as an ambiguous and excluded from the analysis of the LDL cholesterol outcome. Consequently, 25 treatment comparisons were included in the meta-analysis of the LDL cholesterol outcome. Concerning the triglycerides outcome, Sansone et al. [45] did not report data for triglycerides. Consequently, 28 treatment comparisons were included in the meta-analysis of the triglycerides outcome.

The meta-analysis showed that flavanol-containing cocoa beverage and chocolate consumption had no significant effect on circulating concentrations of total cholesterol, LDL cholesterol, and triglycerides (p > 0.05, all), but had a favourable and significant increase in circulating concentration of HDL cholesterol (WMD = 0.05, 95% CI [0.02, 0.09] mmol/L, p = 0.002). The between-study heterogeneity is small-to-moderate (I² between 0–52%). Therefore, the effect of flavanol-containing cocoa beverage and chocolate consumption on lipid profile was consistent.

We conducted a subgroup analysis to test whether individuals who were more at risk of cardiovascular disease, characterised by having an elevated LDL cholesterol at baseline, may benefit from consuming a flavanol-containing cocoa beverage and chocolate more than their healthier counterparts. The results showed no significant differences in all lipid profile outcomes between the two populations (subgroup differences, p > 0.05, all).

The following subgroup analysis tested whether the intervention duration (≤4 weeks vs. >4 weeks) differentially modify the effect of flavanol-containing cocoa beverage and chocolate consumption on lipid profile. The subgroup analysis showed that studies of a duration of >4 weeks significantly increased total cholesterol (WMD = 0.07, 95% CI [0.02, 0.11] mmol/L, p = 0.002), but not in studies of a duration of ≤4 weeks (WMD = 0.06, 95% CI [−0.15, 0.03] mmol/L, p = 0.22) (subgroup differences, p = 0.02). However, the subgroup difference in total cholesterol outcome was not robust to sensitivity analysis, as the difference became not significant after removing the data from Taubert et al. [47]. Furthermore, the intervention duration did not differentially modify HDL cholesterol, LDL cholesterol, and triglycerides outcomes (subgroup differences, p > 0.05, all).

We also tested the effect of the food matrix (cocoa beverage vs. chocolate vs. combination of cocoa beverage and chocolate) on the lipid profile in subgroup analyses. The results showed that total cholesterol, LDL cholesterol, and triglycerides outcomes were not significantly different between different food matrices. Interestingly, there was a significant subgroup differences in HDL cholesterol outcome (p = 0.03), whereby cocoa beverage significantly increased HDL cholesterol by 0.11 (95% CI [0.06, 0.17]) mmol/L (p < 0.001), but not when cocoa was consumed as chocolate (p = 0.10) or a combination of cocoa beverages and chocolate (p = 0.19).

When grouping studies by daily dose of polyphenols, flavanols, and epicatechin, subgroup analyses did not show significant differences in all lipid profile outcomes between doses (p > 0.05, all). Nevertheless, the outcome should be treated with caution, as not many studies identified for the meta-analysis had reported a daily dose of polyphenols, flavanols, and epicatechin.

4. Discussion

This meta-analysis showed that cocoa beverage and dark chocolate consumption was significantly associated with increased HDL cholesterol concentration (WMD = 0.05, 95% CI [0.02, 0.09] mmol/L, p = 0.002), but had no effect on total cholesterol, LDL cholesterol, and triglyceride concentrations. A large epidemiology meta-analysis study showed that coronary artery disease risk was 66% higher in individuals with every 0.39 mmol/L decrease in HDL cholesterol concentration than the population mean (1.34 mmol/L) [49]. Therefore, we suggest that high-flavanol cocoa products, especially cocoa powder, could be included as part of a healthy diet to promote an increase in HDL cholesterol concentrations, lowering cardiovascular risk.

Cocoa is an antioxidant-rich source of polyphenol particularly epicatechin which is hypothesised to inhibit endogenous cholesterol synthesis, improve HDL cholesterol concentration, and have anti-atherogenic effects [50]. This subsequently results in enhanced vascular function and decreases platelet adhesion, which are risk factors for atherosclerosis. Similar findings regarding the role of polyphenols in modulating dyslipidaemia had been reported in a recent systematic review of clinical studies [51]. The authors reported a positive association between polyphenol-rich food intake and improved dyslipidaemia outcomes. HDL have been reported to provide cardio-protective role in mobilising cholesterol from extrahepatic tissues to the liver for excretion from the digestive tract through bile acid formation [52].

Elevated LDL cholesterol concentration is an important risk factor for cardiovascular disease outcome. The mechanism includes that LDL cholesterol build up in the blood vessel resulting in the reduction of the intima diameter. They subsequently get oxidised, producing oxidised-LDL cholesterol, which is immunogenic and pro-inflammatory in nature. This remains the prominent step in atherogenesis, a key process in the progression of cardiovascular disease. It results in attendant dysfunction of the endothelial vascular system. This increases the permeability of the intima to the plasma lipoproteins, encouraging their retention in the subendothelial space [50]. Consequently, in the present work, the effect of cocoa beverage and dark chocolate intake was evaluated to understand whether it would benefit people living with elevated LDL cholesterol. We found that the effect of cocoa beverage and dark chocolate on lipid profile was similar between people living with elevated and normal LDL cholesterol. Furthermore, we found that the duration of intervention (≤4 weeks vs. >4 weeks) did not differentially modify HDL cholesterol, LDL cholesterol, and triglycerides outcomes (subgroup differences, p > 0.05, all). This could be attributed to the fact that it takes much time for changes in cholesterol to be observed clinically.

In our previous systematic review and meta-analysis, we established that the matrix-type or the format through which cocoa beverage and dark chocolate is consumed impacts on blood pressure outcomes [12]. We reported that although cocoa beverage intake significantly lowered systolic blood pressure, the effect was more prominent when dark chocolate was consumed. In this present work, the delivery of cocoa in the format of a beverage resulted in a significant increase (0.11 (95% CI [0.06, 0.17]) mmol/L) in the HDL cholesterol concentration. In contrast, chocolate intake had a neutral effect on the HDL cholesterol concentration. The plausible rationale could be attributed to the additives used in chocolate making, including the presence of cocoa butter and emulsifiers that may increase the saturated fat composition of the chocolate [53], negating the beneficial properties of cocoa. Studies that directly compare the effect of cocoa powder vs. dark chocolate on blood pressure and lipid profile is warranted to confirm our observations. It should, however, be noted that, even though improvement in HDL cholesterol is an essential outcome, the endothelial function was not reported, and hence, this may not necessarily translate to improved cardiovascular health.

Even though daily dose of polyphenols, flavanols, and epicatechin was varied, subgroup analyses did not show significant differences in all lipid profile outcomes between doses (p > 0.05, all). We have indicated that the outcome should be treated with caution, as not many studies identified for the meta-analysis had reported the daily dose of polyphenols, flavanols, and epicatechin. Future studies are recommended to report the daily dose of epicatechin consumed and match it with the body weight of participants, as it is an important factor for detecting measurable changes in clinical biomarkers. For example, we established in a cross-sectional study that body weight and for that matter BMI is an important determinant of an individual’s carotenoid store [54]. Thus, a higher dose of polyphenol particularly epicatechin may be required by those with higher BMI than those with normal BMI.

Of note, our meta-analysis is different from earlier meta-analyses [17,18,55,56], as we excluded studies which involved individuals with co-morbidities, such as diabetes. While focusing on relatively healthier populations, our updated meta-analysis also included more studies than the prior meta-analyses. These earlier meta-analyses found that cocoa consumption lowered LDL cholesterol, whereby Darand et al. [55] remarkably showed that cocoa consumption lowered LDL cholesterol by 0.40 [95% CI 0.17–0.64] mmol/L in diabetic patients. Of the mentioned meta-analyses, only Hooper et al. [18] reported a marginal increase in HDL cholesterol by 0.03 [95% CI 0.00–0.06] mmol/L. Despite the differences in the health condition of study populations potentially having contributed to the inconsistencies in lipid profile outcome between meta-analyses, we are confident that daily cocoa beverage and chocolate consumption was at least beneficial in increasing HDL cholesterol in a relatively healthy population without increasing the atherogenic LDL cholesterol.

Finally, geographical distribution of the studies revealed that Europe (n = 11), with countries including Spain (n = 3), Italy (n = 2), Finland (n = 2), Germany (n = 2), the Netherlands (n = 1), and the UK (n = 1), was the continent with the highest number of studies, followed by North America (n = 7), specifically the USA. This was followed by Asia/Australasia (n = 5), with countries including Japan (n = 3), and Saudi Arabia (n = 1), and Australia (n = 1). Argentina (n = 1) was the only country in South America (n = 1) that recorded a study that investigated the effect of cocoa intake on dyslipidaemia. Surprisingly, no studies from a sub-Saharan African country, including Ghana or Ivory Coast, which produce more than 60% of the global cocoa beans output, were recorded. One of the possible reasons of this observation is that clinical trials are expensive in developing countries, including Ghana. This may potentially create a lack of interest in such kinds of studies. Additionally, the apparent publication biases that tend to favour the publication of studies from developed countries over those from developing countries due to factors such as lack of funding.

5. Conclusions

The meta-analysis has shown that cocoa beverage and dark chocolate consumption had no significant effect on circulating concentrations of total cholesterol, LDL cholesterol, and triglycerides, but had a favourably and significantly increase circulating concentration of HDL cholesterol by 0.05 mmol/L. When the studies were grouped by food matrix, we demonstrated that cocoa beverage intake significantly increased HDL cholesterol by 0.11 mmol/L, but dark chocolate or a combination of cocoa beverage and dark chocolate only had a neutral effect on the HDL cholesterol. Population with normal and elevated LDL cholesterol at baseline did not respond differently to the cocoa beverage and dark chocolate consumption. Cocoa consumption could be recommended as part of a healthy diet to promote cardiovascular health. In a general population with normal and elevated LDL cholesterol. Future studies are recommended to investigate the effect of weight-adjusted dose of epicatechin and the long-term effect of cocoa intake on lipid profile parameters.