The Effect of Anethum Graveolens L. (Dill) on Lipid Profile in Adults with Cardiovascular Risk Factors: A Systematic Review and Meta-Analysis of Controlled Clinical Trials

Introduction

Dyslipidemia, a well-founded modifiable cardiovascular risk factor, is characterized by reduced levels of high-density lipoprotein cholesterol (HDL-C) and increased levels of triglycerides, total cholesterol (TC), and low-density lipoprotein cholesterol (LDL-C). Several factors, including genetics, lifestyle, and dietary factors, are shown to affect lipid profile [1]. Diet therapy and pharmacological interventions are approaches for controlling dyslipidemia [2, 3]. However, taking lipid‐lowering medications has some limitations due to their adverse effects and attenuated efficacy after prolonged use[4-7]. Considering these limitations, a great interest has been expressed in nutraceuticals as a complement for drug treatments. According to the literature, nutraceuticals and dietary supplementations can improve lipid profile by several mechanisms such as lipid absorption and metabolism [8-10]. It is proposed that Anethum graveolens L. (AG) can improve lipid profile by several mechanisms, including 1) binding to bile acids and reducing cholesterol absorption, 2) inhibiting acetyl-CoA carboxylase and HMG-CoA reductase, which consequently reduces fatty acid and cholesterol synthesis and leads to cholesterol clearance and up-regulation of LDL-C receptors [11].

Anethum graveolens L. known as dill, is one of the most common herbs used as spice in foods [12, 13]. It is cultivated in different regions of the world, including southeastern Iran [13]. The leaves of AG contain 24 types of minerals, proteins, and fibers [14]. Pharmacological effects of AG include: antibacterial [15, 16], antimicrobial [17-19], anti-inflammatory [20], anti-oxidative, anti-hyperglycemia [19, 21-23], anti-hyperlipidemia, and anti-hypercholesterolemia [21, 24, 25]. Several studies showed that the intake of AG could improve serum levels of TC, LDL-C, TG, and HDL-C. However, other trials failed to observe an improvement in lipid parameters measured in blood after AG intake [26-30].

Because of the the mentioned controversies between the available documents, we carried out a systematic review and meta-analysis on the available randomized controlled trials (RCTs). The aim was to assess AG supplementation's efficacy on lipid profile in adults with cardiovascular risk factors.

Materials and Methods

This systematic review and meta-analysis was registered in the international prospective register of systematic reviews (PROSPERO) website (https://www. crd.york.ac.uk/prospero/) with the registration number of CRD42020115000. The Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guideline was used to conduct the systematic review process.

Search strategy

PubMed, Embase, Cochrane's database, Ovid, Web of Science, ProQuest, Scopus, and Google Scholar were searched to find related articles investigating AG's impact on the lipid profile of adults with cardiovascular disease risk factors up to December 2020. Moreover, we investigated the reference lists of the included studies manually to find other relevant studies. The search was not limited based on language, but the search limitation was applied to find human clinical trials.

Three authors conducted the search to find relevant clinical trials evaluating the relationship between AG and lipid profile based on medical subject headings (MeSH terms) as follows: [Anethum Graveolens OR  Dill] AND [ lipid profile OR Lipid metabolism disorders OR Low-density lipoprotein OR High-density lipoprotein OR Cholesterol OR Triglycerides OR  LDL OR HDL OR TC OR TG OR Cardiomyopathies OR  Coronary heart disease OR  CHD OR Heart failure OR Myocardial Ischemia OR Hyperlipidaemia OR  Hypercholesterolemia OR Dyslipidaemia OR hypertriglyceridemia OR  Hypercholesterolemia OR Hypertriglyceridemia OR  Obesity OR Obese subject OR Obesity OR Metabolic syndrome OR MetS OR  Metabolic syndrome X OR Type 2 diabetes OR  hypothyroidism OR non-alcoholic steatohepatitis OR NASH OR  Non-alcoholic fatty liver disease OR NAFLD  OR Polycystic ovarian syndrome OR PCOS ] AND [Randomized clinical trial OR Clinical trial OR Clinical study OR Double-blind randomized clinical trial]. 

Inclusion and exclusion criteria

After removing the duplicates by using EndNote software, the selected trials' titles and abstracts were reviewed to assess their systematic review eligibility. The irrelevant articles were removed, and full texts of the eligible trials were reviewed.

Eligibility criteria were defined based on the following inclusion criteria:

1. Study design: randomized clinical trials
2. Participants: adults with cardiovascular risk factors
3. Comparison: intervention group supplemented with powder or extract of AG versus the control group
4. Outcomes: serum level of at least one of the lipid parameters including TC, TG, LDL-C, and HDL-C

Exclusion criteria were animal trials and studies with the participation of children and adolescents.

Data extraction

Data extraction was performed based on the 'authors' name, publication date, trial design and location, 'participants' mean age, body mass index (BMI), AG dose, supplement type (leaf powder or extract), follow-up duration, sample size, and measurements. Extracted data were recorded on a Microsoft Excel sheet for further use. 

Quality assessment

The Cochrane risk of bias tool was applied by three reviewers to evaluate the quality of eligible trials methodologically[31]. Quality assessment was applied based on several items, including random sequence generation, allocation concealment, blinding of participants and personnel, incomplete outcome data, and other sources of bias. "Yes," "no," and "unclear" were defined as the low, high, and unknown risk of bias according to the Cochrane guideline. NutriGrade scoring system was applied to assess and score the overall quality of meta-analysis using the following domains: precision; study quality, risk of bias, limitations; directness; heterogeneity; publication bias, study design, and funding bias [32, 33].

Data analysis

We used Stata SE software (version 12) to conduct the meta-analysis. The Cochrane guideline was applied to calculate the standard deviations (SDs) of the mean changes using the following formula:

Correlation coefficients (R) of the trials were calculated. If R was not mentioned in a study, R = 0.5 was assumed. The effect size was defined as weighted mean difference (WMD) with a 95% confidence interval (CI). A random-effects model was used to conduct  the meta-analysis. I-squared statistic and Cochran Q test were used to assess the heterogeneity. Heterogeneity significance level was considered at I² value > 50% or p ≤ 0.05 for Cochran Q test . The effect of each trial on the overall effect size was evaluated using sensitivity analysis. The subgroup analysis was used to detect the heterogeneity sources.   Publication bias was checked using funnel plots and statistical asymmetry tests (Egger’s test and Begg’s test).

Results

A total of 121 articles were identified after database research, of which the full-text of eight studies was reviewed for eligibility assessment. Six studies met the inclusion criteria and were included in the meta-analysis [11, 26-30], and two studies were excluded because anti-hyperlipidemic drugs were only taken by the control groups [34, 35]. Figure 1. shows a flowchart of the study process.

 Figure 1. The Study Process of the Systematic Review and Meta-Analysis

In this meta-analysis, six eligible trials were included with seven treatment arms and a total of 291participants (n = 146 in the control group and n = 145 in the melatonin group). All trials were conducted in Iran. Studies with the largest and lowest sample sizes had 100 [30] and 20 [26] participants, respectively. Supplementation was done using AG extracts and AG powder in two [27, 28] and four studies [11, 26, 29, 30], respectively. Treatments included a wide range of AG supplementation; from 2 mg/kg/day to 600 mg/day AG extract and from 3.3 gr/day to 900 mg/kg/day AG powder. The supplementation duration ranged between four weeks [30] and eight weeks[11, 27-30]. Participants were affected by type two diabetes mellitus [11, 26, 29], hyperlipidemia [30], obesity [28], and metabolic syndrome [27]. The study characteristics are represented in Table 1.

Quality assessment and quality of meta-evidence

Quality assessment of eligible studies was done based on Cochrane Collaboration's tool [31] (Table 2). Based on the findings, two studies had a good quality[11, 29], two trials had n acceptable quality [27, 28], and two research had poor quality [26, 30]. This meta-analysis's total quality score was calculated as 7 using the NutriGrade scoring system, which shows a moderate quality of meta-evidence. However, further research may affect overall estimates found in the present study.

Meta-analysis

The effect of dill on serum triglycerides

A significant decrease after AG supplementation was observed after pooling the seven intervention arms comparing with the control group [WMD = -29.20, 95% confidence interval (CI): -34.73,-23.68 mg/dL, p < 0.001; Cochran’s Q test, p < 0.001, I² =92.9%) (Figure 2). Subgroup analysis was performed to find the possible sources of heterogeneity. Subgroup analysis based on quality of trials revealed a reduced heterogeneity in trials with a good quality (WMD = -37.30, 95%CI: -62.69,-11.92 mg/dL, p = 0.004; Cochran’s Q test, p = 0.32, I² = 0.0%) (Table3). However, subgroup analysis based on supplement type and supplementation duration revealed no information with regard to the heterogeneity source. However, significant results were observed after supplementation with AG extract (WMD = -42.66, 95%CI: -49.24,-36.08 mg/dL,
p < 0.001; Cochran’s Q test, p = 0.002, I² = 84.40%) and more than 6 weeks of supplementation
(WMD = -42.32, 95%CI: -48.69,-35.96 mg/dL,
p < 0.001; Cochran’s Q test,p = 0.007, I² =71.3%) compared with the control group. According to the leave-one-out sensitivity analysis, removing a trial by Ostadrahimi et al. [28] showed a significant impact
as the overall effect became non-significant
(WMD= -3.15, 95% CI: -11.64, 5.34).

The effect of dill on serum total cholesterol

Based on the findings, the supplementation had a significant effect on serum TC and the heterogeneity was high (WMD = -16.46, 95%CI: -21.54,-11.39 mg/dL, p < 0.001; Cochran’s Q test, p < 0.001, I² =90.5%) (Figure 3). Subgroup analysis based on quality and duration showed that the source of heterogeneity was reduced in studies with poor quality (WMD = 6.45, 95%CI: -3.28,16.1 mg/dL, p = 0.19; Cochran’s Q test,   p = 0.82, I² =0.0%), good quality (WMD = -14.42, 95%CI:-24.06,-4.79 mg/dL, p = 0.003; Cochran’s Q test,  p = 0.84, I² =0.0%), and supplementation duration ≤ 6 weeks (WMD = 6.45, 95%CI: -3.28, 16.19 mg/dL, p = 0.19; Cochran’s Q test,  p = 0.82, I² =0.0%). Although heterogeneity was not reduced based on the type of supplement, significant reduction was found in the pooled effect size of trials with AG extract supplementation (WMD = -31.57, 95%CI:-39.13,-24.00 mg/dL, p < 0.001; Cochran’s Q test,  p < 0.001, I² =90.5%). Sensitivity analysis showed that removing the study by Ostadrahimi et al. [28]^.had a significant effect and turned the overall estimate to non-significant (WMD= -1.71, 95% CI: -7.62, 4.18).

Figure 2.Forest Plot of Trials Experimenting the Effect of AG on TG

Figure 3. Forest Plot of Trials Experimenting the Effect of AG on TC

The effect of dill on serum low-density lipoprotein cholesterol

Meta-analysis of seven intervention arms showed a significant effect of oral AG on LDL-C (WMD = -13.90, 95%CI: -16.08, -11.72 mg/dL, p < 0.001; Cochran’s Q test, p = 0.01, I² = 63.8%) (Figure 4). In accordance with the subgroup analysis based on quality, reduced heterogeneity was observed in trials with good quality (WMD = -10.45, 95%CI: -16.79, - 4.11 mg/dL, p = 0.001; Cochran’s Q test,
p = 0.55, I² = 0.0%). Trials with AG extract supplementation had a lower heterogeneity in their results (WMD = -15.04, 95%CI: -17.43,-12.65 mg/dL, p <0.001; Cochran’s Q test,   p = 0.22, I² = 33.7%). A reduced heterogeneity had was also found in trials with > 6 weeks of duration (WMD = -14.47, 95%CI: -16.71, -12.23 mg/dL, p < 0.001; Cochran’s Q test, p = 0.27, I² = 22.1%). Sensitivity analysis displayed no significant effect on the pooled effect size after omitting the studies by Ostadrahimi et al. [28] (WMD= -13.53, 95% CI: -15.85,-11.20) and Kojuri et al. (WMD=-14.46, 95% CI:-16.66,-12.25).

Figure 4. Forest Plot of Trials Experimenting the Effect of AG on LDL-C

  The effect of dill on serum high-density lipoprotein cholesterol

 The meta-analysis showed a significant increase in HDL-C with high heterogeneity between studies (WMD = 4.01, 95%CI: 3.48, 4.54 mg/dL, p < 0.001; Cochran’s Q test,  p < 0.001, I² = 94.2%) (Figure 5).  Although heterogeneity was reduced in trials with poor quality (WMD = -0.66, 95%CI: -2.05, 0.73 mg/dL, p = 0.93; Cochran’s Q test,  p = 0.97, I² = 0.0%), only studies with fair quality had a significant effect on serum HDL-C levels (WMD = 5.34,  95%CI: 4.72, 5.95 mg/dL, p < 0.001; Cochran’s Q test,  p < 0.001, I² = 91%). The significant effect of AG was also observed just in trials with extract supplementation (WMD = 5.34, 95%CI: 4.72, 5.95, p < 0.001; Cochran’s Q test,  p < 0.001, I² = 91%). Heterogeneity was reduced in study duration of ≤ 6 weeks (WMD = -0.66, 95%CI:-2.05,0.73 mg/dL, p = 0.35; Cochran’s Q test,   p = 0.97, I² = 0.0%). Moreover, a significant increase was observed in HDL-C in studies with duration of > 6 weeks (WMD = 4.81, 95%CI:4.23,5.39 mg/dL, p < 0.001; Cochran’s Q test,  p < 0.001, I² = 92.5%). Sensitivity analysis revealed no significant effect of removing the studies one by one from the analysis.

Figure 5. Forest Plot of Trials Experimenting the Effect of AG on HDL-C

Publication bias

No asymmetry was observed in funnel plots for meta-analysis of TG, TC, LDL-C, and HDL-C (Figures 6-9). Furthermore, no significant publication bias was found for TG, TC, LDL-C, and HDL-C by 'Egger's regression analysis in the trials assessing AG's effect on lipid profile (p for bias 0.42, 0.15, 0.08, and 0.32, respectively).

 Figure 6. Funnel Plot of Trials Experimenting the Effect of AG on TG

Figure 7. Funnel Plot of Trials Experimenting the Effect of AG on TC

Figure 8. Funnel Plot of Trials Experimenting the Effect of AG on LDL-C

Figure 9. Funnel Plot of trials Experimenting the Effect of AG on HDL-C

Table1. Characteristics of studies included in the meta-analysis.

Values are as mean ± SD.

Abbreviations: AG: Anethum Graveolens; BMI: body mass index.

Table2. Quality assessment of studies included in the meta-analysis.

Abbreviations: ^αLow risk of bias, ^βHigh risk of bias, ^γUn clear risk of bias; G: good quality, F: fair quality, P: poor quality.

Table 3. Overall effect as well as Subgroup and Heterogeneity analysis of trials investigating the effect of oral AG on lipid profile.

^* Random effects  model.^ᶲP value § Weight mean difference.