Levels of asymmetric dimethylarginine (ADMA), an endogenous nitric oxide synthase inhibitor, and risk of coronary artery disease: A meta-analysis based on 4713 participants

Abstract

Background

Asymmetric dimethylarginine (ADMA) is an endogenous inhibitor of endothelial nitric oxide synthase by competing with L-arginine. As a result, the expression of nitric oxide decreases and endothelial dysfunction occurs. Studies have evaluated the association between the serum ADMA level and risk of coronary artery disease. However, conflicting results have been obtained.

Methods

Pubmed, Web of Science, Embase, Ovid, Cochrane databases were searched to identify eligible studies published in English until December 2014. Association was assessed on the basis of weighted mean differences (WMD) with 95% confidence intervals (CIs). Publication bias was analysed using Begg’s and Egger’s tests. Sensitivity analysis was performed to evaluate result stability.

Results

A total of 16 case–control studies with 2939 patients and 1774 controls were included in the meta-analysis. Pooled result indicated that patients with coronary artery disease yielded a higher ADMA level than healthy controls (WMD: 0.248, 95% CI: 0.156–0.340; p = 1.16 e–7). Sensitivity analysis suggested that our meta-analysis result was stable. Subgroup analysis found a similar pattern in patients with myocardial infarction (WMD: 0.397, 95% CI: 0.112–0.683; p = 0.0106), stable angina pectoris (WMD: 0.197, 95% CI: 0.031–0.364; p = 0.02) and unstable angina pectoris (WMD: 0.857, 95% CI: 0.293–1.420; p = 0.003).

Conclusions

Meta-analysis results indicated that an increased ADMA level is associated with an increased risk of coronary artery disease.

ADMA, coronary artery disease, risk, meta-analysis

Introduction

L-arginine methylation produces three different derivatives of this amino acid: asymmetric isomer monomethyl-L-arginine (L-NMMA), asymmetric dimethylarginine (ADMA) and symmetric dimethylarginine (SDMA). Methylated L-arginines are excreted in human urine;¹ methylated L-arginines have also been detected in neurons of animals² and in human endothelial cells.^3,4

Nitric oxide (NO) regulates cardiovascular functions, and a decrease in NO bioavailability causes endothelial dysfunction.⁵ Endothelial nitric oxide synthase (eNOS) uses L-arginine as a substrate.⁶ Vallance and coworkers first described ADMA as an endogenous inhibitor of NO synthase in 1992.⁷ The endothelium can produce methylated amino acids, such as ADMA, which can compete with L-arginine as a substrate of eNOS; as a result, endothelial dysfunction occurs.⁷

Increased serum ADMA concentration is associated with several conditions, including diabetes mellitus,⁸ atherosclerosis,⁹ hypertension,¹⁰ pre-eclampsia,¹¹ stroke¹² and peripheral vascular disease.¹³ Our recent study also demonstrated that increased ADMA concentrations directly induce endothelial dysfunction by downregulating the protein expression of eNOS and by increasing the production of superoxide anion in human internal thoracic arteries which are discarded from coronary artery bypass grafting.¹⁴

The relationship between ADMA concentration and risk of coronary artery disease (CAD) has been investigated.^15,16 However, inconsistent findings regarding this relationship have been observed partially because of a relatively small sample size in each published study. Therefore, we performed a comprehensive and critical meta-analysis of all relevant studies to generate a clear and evidence-based conclusion on the association between ADMA level and CAD risk.

Materials and methods

Search strategy

A comprehensive literature search in Pubmed, Web of Science, Embase, Ovid, and Cochrane databases was conducted to identify suitable studies published until December 2014. The following keywords were used: ‘coronary artery disease’ OR ‘acute coronary syndrome’ OR ‘myocardial infarction’ OR ‘angina’ OR ‘cardiovascular diseases’ AND ‘ADMA’ OR ‘asymmetric dimethylarginine’. The most complete and most recent results were used when multiple publications from the same study group were found. References of reviews and retrieved articles were also searched simultaneously to determine additional eligible studies.

Inclusion criteria

Two investigators independently reviewed all of the identified studies to determine whether these studies were eligible for meta-analysis. The following selection criteria were considered in this meta-analysis: 1) human study; 2) studies on the association between ADMA level and CAD risk; 3) case–control study, cohort study or randomised clinical trial study; 4) proper CAD diagnosis criteria and 5) original data. The diagnosis of patients with CAD was confirmed by coronary angiography. Studies were excluded if relevant information could not be obtained.

Data extraction

Two investigators independently extracted data; a third investigator reviewed the result. The following information was extracted from each study: first author, year of publication, country, number of patients and healthy controls, age, mean ± SD of ADMA concentration, ADMA array and main types of CAD. If any information essential for the analysis was unavailable from a study, missing data were requested from the corresponding authors.

Statistical analysis

Pooled effects were presented as weighted mean differences (WMDs) with 95% confidence intervals (CI). Heterogeneity across eligible studies was evaluated using Q-test and heterogeneity was considered significant when p < 0.1.^17,18 Heterogeneity was also quantified using I² metric (I²= (Q − df)/Q × 100%): I²< 25%, no heterogeneity; I²= 25% to 50%, moderate heterogeneity; I²= 50% to 75%, large heterogeneity; and I²> 75%, extreme heterogeneity). A fixed-effects model was used when effects were considered homogeneous (p > 0.1, I²< 50%); otherwise, a random-effects model was appropriate.¹⁹

Sensitivity analysis was conducted to investigate the influence of an individual study on overall risk estimate; sensitivity analysis was carried out by sequentially omitting one study in each turn.²⁰ If more than seven studies were included, Begg’s adjusted rank correlation test and Egger’s regression asymmetry test were performed to determine publication bias, which was shown as a funnel plot.²¹  p < 0.05 represented statistically significant publication bias. If publication bias was found, a trim and fill method was used to adjust crude results.²² Subgroup analyses were also conducted to determine associations between ADMA level and main CAD types. Analyses were performed using STATA software version 10.0 (Stata Corporation, College Station, TX, USA) and R statistical software (version 2.15.2, http://www.r-project.org).

Results

Study characteristics

A total of 109 abstracts that satisfied the inclusion criteria were retrieved. Two reviewers independently selected the relevant studies. Twenty relevant studies that described association between ADMA level and CAD were identified. However, after reading the full articles and reaching the corresponding authors, we excluded five studies that were not case–control studies and two studies with unavailable information.^23,24 Subsequently, 16 case–control studies (20 cohorts) with 2939 patients with CAD and 1774 healthy controls were available for analysis; the characteristics of these studies are outlined in Table 1.^25–40  Figure 1 illustrates the process by which studies were selected and excluded; Figure 1 also indicates specific reasons related to study selection and exclusion. Subgroup analysis was also performed in patients with myocardial infraction (MI), stable angina pectoris (SAP), unstable angina pectoris (USAP), occult CAD, acute coronary syndrome (ACS) and slow coronary flow (SCF).

Figure 1.

Flow chart showing the selection and exclusion of studies in the meta-analysis. Specific reasons for exclusion of studies are also shown.

Table 1.

Summary of the studies included in the meta-analysis.

First author	Year of publication	Country	Control group			Patient group			Assay	Main types of CAD
			n	Age	Mean ± SD (µmol/l)	n	Age	Mean ± SD (µmol/l)
Krempl	2005	Germany	40	63.1 ± 7.9	0.59 ± 0.23	81	–	0.76 ± 0.17	ELISA	CAD
			40	63.1 ± 7.9	0.59 ± 0.23	45	63.3 ± 8.7	0.73 ± 0.15		SAP
			40	63.1 ± 7.9	0.59 ± 0.23	36	66.7 ± 9.6	0.82 ± 0.18		USAP
Bae	2005	South Korea	48	54.5 ± 7.1	1.57 ± 0.85	48	54.6 ± 10.3	3.13 ± 0.85	–	ACS
			48	54.5 ± 7.1	1.57 ± 0.85	37	54.0 ± 11.0	3.04 ± 0.90		AMI
			48	54.5 ± 7.1	1.57 ± 0.85	11	58.0 ± 8.0	3.43 ± 0.57		USAP
Selcuk	2007	Turkey	31	50.6 ± 13.7	2.17 ± 1.32	31	54.7 ± 10.1	3.28 ± 2.11	HPLC	SCF
Maas	2007	Germany	254	61.2 ± 8.4	0.79 ± 0.21	88	61.1 ± 8.6	0.80 ± 0.22	ELISA	MI
Iribarren	2007	USA	263	42.2 ± 0.18	0.53 ± 0.345	263	41.2 ± 0.18	0.55 ± 0.505		Occult CAD
Murr	2009	Germany	453	59.2 ± 11.5	0.81 ± 0.14	1577	63.7 ± 9.9	0.82 ± 0.15	HPLC	CAD
Gad	2010	Egypt	100	35–50	0.65 ± 0.18	100	35–50	0.61 ± 0.16	HPLC	CAD
			100	35–50	0.65 ± 0.18	11	35–50	0.75 ± 0.17		AMI
Aktoz	2010	Turkey	22	48.95 ± 8.70	0.50 ± 0.25	29	59.69 ± 9.52	0.50 ± 0.30	ELISA	CAD
Yu	2010	China	64	59.8 ± 10.4	0.36 ± 0.12	54	67.8 ± 9.8	0.51 ± 0.18	HPLC	ACS
			64	59.8 ± 10.4	0.36 ± 0.12	25	62.6 ± 11.0	0.42 ± 0.13		SAP
Bekpinar	2011	Turkey	33	54.8 ± 9.1	0.326 ± 0.099	43	57.8 ± 13.6	0.345 ± 0.178	HPLC	MI
Yucel	2012	Turkey	30	51.1 ± 8.1	0.742 ± 0.31	50	53.0 ± 9.1	0.925 ± 0.38	ELISA	SCF
Djordjevic	2012	Serbia	60	58.7 ± 5.6	0.31 ± 0.175	94	62.3 ± 6.8	0.98 ± 0.383	HPLC	AMI
			60	58.7 ± 5.6	0.31 ± 0.175	74	62.5 ± 7.3	0.94 ± 0.605		USAP
			60	58.7 ± 5.6	0.31 ± 0.175	74	61.6 ± 10.8	0.76 ± 0.698		SAP
Kruszelnicka	2013	Poland	34	56.0 ± 12.0	0.46 ± 0.09	151	57.0 ± 11.0	0.51 ± 0.10	ELISA	CAD
Gürel	2013	Turkey	18	46.0 ± 8.0	0.62 ± 0.16	17	48.0 ± 9.0	0.78 ± 0.14	ELISA	Occult CAD
Jawalekar	2013	India	80	47.29 ± 10.21	0.36 ± 0.30	80	56.21 ± 6.73	0.66 ± 0.76	ELISA	CAD
ShivKaR	2014	India	30	–	0.41 ± 0.09	30	–	0.96 ± 0.08	HPLC	Block > 70% CAD
			30	–	0.41 ± 0.09	30	–	0.64 ± 0.08		Block 40%–70% CAD

HPLC: high performance liquid chromatography; ELISA: enzyme linked immunosorbent assay; CAD: coronary artery disease; SAP: stable angina pectoris; USAP: unstable angina pectoris; ACS: acute coronary syndrome; MI: myocardial infarction; AMI: acute myocardial infarction; SCF: slow coronary flow

ADMA levels and CAD risk

Substantial heterogeneity was observed in the analysis. Thus, random-effects models were used. Overall pooled results showed that ADMA levels were significantly increased in patients with CAD (WMD = 0.248, 95% CI = 0.156–0.340; p = 1.16 e–7; Table 2, Figure 2).

Figure 2.

Forest plots of the associations between serum asymmetric dimethylarginine level and coronary artery disease risk.

WMD: weighted mean difference; CI: confidence interval

Table 2.

Main results of association between ADMA level and risk of CAD.

	Sample size		Test of heterogeneity			Test of association
Subgroup*	Patients	Controls	Q	p	I2 (%)	WMD	95% CI	Z	p
CAD	2939	1774	971.06	0.000	98.0	0.248	0.156–0.340	5.30	1.16 e–7
MI	273	495	221.48	0.000	98.2	0.397	0.112–0.683	2.73	0.006
SAP	144	164	19.52	0.000	89.8	0.197	0.031–0.364	2.32	0.020
USAP	121	148	70.50	0.000	97.2	0.857	0.293–1.420	2.98	0.003
Occult CAD	280	281	4.90	0.027	79.6	0.086	−0.051–0.223	1.23	0.219
ACS	102	112	64.28	0.000	98.4	0.845	−0.537–2.226	1.20	0.231
SCF	81	61	4.17	0.041	76.0	0.542	−0.343–1.427	1.20	0.230

ADMA: asymmetric dimethylarginine; CAD: coronary artery disease; WMD: weighted mean difference; SAP: stable angina pectoris; USAP: unstable angina pectoris; ACS: acute coronary syndrome; MI: myocardial infarction; SCF: slow coronary flow

Different clinical types of CAD were subjected to stratified analysis; the results revealed significant association between ADMA level and MI (WMD = 0.397, 95% CI = 0.112–0.683), SAP (WMD = 0.197, 95% CI = 0.031–0.364) and USAP (WMD = 0.857, 95% CI = 0.293–1.420); by contrast, no significant association was observed between ADMA level and occult CAD (WMD = 0.086, 95% CI = −0.051–0.223), ACS (WMD = 0.845, 95% CI = −0.537–2.226) and SCF (WMD = 0.542, 95% CI = −0.343–1.427). The main results of the meta-analysis are shown in Table 2.

Sensitivity analysis

We conducted sensitivity analysis to evaluate the stability of crude results, which were pooled using a random-effects model to determine the association between ADMA level and CAD risk. The corresponding pooled WMDs were not substantially altered when any single study was deleted (Figure 3), suggesting that the results of this meta-analysis were stable.

Figure 3.

Sensitivity analysis of the association between serum asymmetric dimethylarginine level and coronary artery disease risk. The influence of individual studies on the overall weighted mean difference (WMD) is shown. The middle vertical axis indicates the overall WMD and the two vertical axes indicate 95% confidence interval (CI). Hollow circles represent the pooled WMD when the remaining study is omitted from the meta-analysis. Two ends of each broken line represent 95% CI.

Publication bias

We detected publication biases in association between ADMA level and CAD risk (p_Begg’s test = 0.044; p_{Egger’s test} = 0.025; Figure 4). Publication biases may limit our analysis because studies with null findings, especially those with small sample size, were less likely to be published. Using the trim and fill method, we found that seven studies are necessary to balance the funnel plot if publication bias is the only source of funnel plot asymmetry (Figure 5). The adjusted risk estimate was attenuated. The adjusted WDM was 0.060 (95% CI = −0.045–0.165; p = 0.261).

Figure 4.

Egger’s plot of publication bias test. Each point represents a separate study of the indicated association.

Figure 5.

Begg’s plot of studies after trimming and filling. Dummy studies and genuine studies are represented by enclosed circles and free circles, respectively.

Discussion

The endothelium plays a major role in vascular homeostasis.⁴¹ Endothelial dysfunction contributes to pathological conditions characterised by vasospasm, vasoconstriction, excessive thrombosis and abnormal vascular proliferation.⁴² NO is the primary mediator of endothelium-dependent vasodilation; NO also regulates endothelium-mediated vascular homeostasis.⁴³ In the vasculature, NO is derived from L-arginine oxidation catalysed by the constitutively expressed enzyme eNOS. This endothelial-derived NO diffuses from the vascular endothelium and affects the smooth muscle cell layer; in the smooth muscle cell layer, NO activates guanylate cyclase, leading to smooth muscle cell relaxation.⁴⁴

ADMA is synthesised when arginine residues in proteins are methylated by arginine methyltransferases. Humans generate approximately 300 µmol of ADMA per day.⁴⁵ ADMA is partly eliminated from an organism’s body through renal filtration. In healthy individuals, approximately 10% of methylated arginine is excreted via this mechanism. ADMA is predominantly eliminated when dimethylarginine dimethylamino-hydrolase (DDAH) is enzymatically degraded to citrulline and dimethylamine.⁴⁶ ADMA inhibits the three isoforms of NOs; ADMA is also equipotent with L-NMMA. ADMA can also uncouple the enzyme and generate superoxides; furthermore, ADMA has been associated with endothelial dysfunction.⁴⁷ In the left internal thoracic arteries of humans, increased ADMA levels directly cause endothelial dysfunction by downregulating the protein expression of eNOS and by increasing the production of superoxide.¹⁴

CAD is one of the most important causes of death worldwide. Thus, additional factors should be identified to predict the risk of cardiovascular events related to physiological and pathophysiological mechanisms. An impaired NO pathway is accompanied by endothelial dysfunction, which is an early step of cardiovascular disease development. ADMA interacts with the NO pathway, and ADMA serum levels are increased in diseases related to endothelial dysfunction. Therefore, studies have been conducted to establish the association between serum ADMA level and CAD risk. However, inconsistent findings regarding this relationship have been obtained. A relatively small sample size in each published study may be one of the important reasons.

In our present study, all available data pertaining to the association between ADMA level and CAD risk were pooled; these data included 16 studies (20 cohorts) with a total of 4713 subjects; sensitivity analysis and publication bias test were performed to evaluate result stability. Stratified pooled analyses were also performed on the basis of different clinical types of CAD. Our final meta-analysis results indicated that patients with CAD yielded a higher ADMA level than healthy controls (WMD = 0.248, 95% CI = 0.156–0.340; p = 1.16 e–7; Table 2, Figure 2). Sensitivity analysis results showed that the removal of any study did not affect the overall results and conclusion. Therefore, our meta-analysis results showed a higher degree of certainty; high ADMA level is an important risk factor in patients with CAD.

To the best of our knowledge, this meta-analysis is the first to estimate the association between ADMA level and CAD risk. However, possible limitations of our study must be considered. The major limitation of our study was high heterogeneity, although we failed to identify the exact source of heterogeneity. The use of a random-effects model may be insufficient to adjust for heterogeneity; thus, sensitivity analysis was conducted. The corresponding pooled WMD was not substantially altered when any single study was deleted, suggesting that the results of this meta-analysis are stable. The impact of confounding factors, such as smoking, age and sex, among others, cannot be assessed and incorporated in our meta-analysis. The lack of adjustment for these confounding factors may result in a slight overestimation of risk. Case–control studies may overestimate the correlation. The sample size of studies included in the analysis was relatively small, especially in ACS, SCF and occult CAD subgroups. A large sample size and numerous studies would allow for a more precise size estimation and for a more sophisticated moderator analysis. Publication bias was detected in our meta-analysis. Thus, the trim and fill method was used to adjust WMD. After seven studies were added to balance the funnel plot, the adjusted risk estimate was attenuated. This finding demonstrated that publication biases may affect the stability of positive results.

In conclusion, this meta-analysis suggests that high ADMA levels are a risk factor in patients with CAD. The subgroup analysis result also indicated that increased ADMA levels are found in different clinical types of CAD, including MI, SAP and USAP.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the National Natural Science Foundation of China (grant no. 81301485) and Award Foundation for Outstanding Young Scientists of Shandong Province (grant no. BS2013YY036).