Introduction

Asymmetrical dimethylarginine (ADMA), an endogenous inhibitor of nitric oxide synthase (NOS), inhibits nitric oxide (NO) biosynthesis1. Therefore, increase in circulating ADMA may impair the beneficial effects of NO on endothelial function, vasodilatation, and arterial stiffness2. Elevation of ADMA has been observed to be associated with hypertension, diabetes, hypertriglyceridemia, and hyperhomocysteinemia. These are risk factors for cardiovascular and cerebrovascular diseases3,4,5. Furthermore, studies have shown that a high circulating ADMA concentration is associated with a higher incidence of cardiovascular disease (CVD) and worse cerebrovascular disease outcomes6,7.

Owing to its involvement in CVD, circulating ADMA as a predictor of risk for mortality in patients with pre-existing CVD has been investigated. The prognostic value of ADMA has been well documented in patients with dilated cardiomyopathy, diabetes mellitus, and acute ischemic stroke8,9. Additionally, in patients with chronic kidney disease, ADMA has also been shown to be an important predictor of progression to end-stage renal disease (ESRD) and all-cause mortality10. Circulating ADMA is also associated with all-cause mortality in patients with diabetes mellitus11.

Even in a healthy population, high levels of circulating ADMA may be associated with higher rates of all-cause death12. However, the predictive value of ADMA for all-cause mortality remains controversial. Some researchers have reported that ADMA is not an independent predictor of all-cause mortality after multivariate adjustment13,14. The complicated interpretation of the predictive value of ADMA might be attributed to the limited number of participants involved in each individual study and reduced mortality. Therefore, it is important to clarify the relationship between ADMA and survival for potential interventions. To investigate the association between circulating ADMA and all-cause mortality, we conducted a systematic review and meta-analysis.

Methods

Literature search

We searched the following databases: Cochrane Library, PubMed, Web of Knowledge, and Elsevier (ScienceDirect OnLine) to retrieve literature investigating the association between circulating ADMA and all-cause mortality. The terms “ADMA or asymmetrical dimethylarginine” and “mortality or survival or outcomes or prognosis or prognostic” were used as search terms. Eligible trials were identified up to June 1, 2016 through electronic searches. Hand searches of the references of the identified trials were also conducted. The meta-analysis was conducted in accordance with the PRISMA guidelines.

Inclusion and exclusion criteria

Studies were considered for inclusion if they met the following criteria: (i) written in English, (ii) reported the all-cause mortality of participants, (iii) investigated the relationship between circulating ADMA with all-cause mortality, (iv) reported hazard ratio (HR), relative risk (RR), or odds ratio (OR) with 95% confidence intervals (95% CI) for ADMA associated with all-cause mortality, and (v) used a multivariate Cox’s proportional-hazards model to analyze the HR, RR, or OR value. Studies were excluded on the basis of the following criteria: (i) not written in English, (ii) used univariate analyses alone for the correlation of plasma ADMA with all-cause mortality, and (iii) the full-text article was not available.

Data extraction

The data were extracted independently by two reviewers (Qianqian Zhu and Xiang Li) and validated by a third reviewer (Shaoli Zhou). The following information was extracted: primary author, year of publication, geographical location, number of participants, sex ratio, and baseline average age (mean, median, or range).

Statistical analysis

Meta-analysis was performed in Review Manager 5 (The Cochrane Collaboration, Oxford, UK). The pooled effect of ADMA as a predictor of all-cause mortality was calculated as RR with 95% CI. A chi-square test was used to assess the heterogeneity. I2 value <25% was defined as no heterogeneity and a random-effects model was used when heterogeneity existed among the studies analyzed. A Begg and Egger test was used to test the publication bias and analyses were performed by using Stata 12.1 (Stata Corp., College Station, TX). Differences were considered significant when the two-tailed p values were <0.05.

Results

Search

The search strategy yielded a total of 1993 non-duplicated entries. After screening the titles, the type of entries, and the abstracts, 43 articles were chosen for full review. We identified 34 eligible studies for the final analysis. The inclusion and exclusion process of the studies is shown in Fig. 1.

Figure 1
figure 1

The flow chart of inclusion and exclusion.

Characteristics of the included studies

A total of 34 studies reporting a total of 39137 participants were included in the final analysis (Table 1)7,9,11,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43. Eleven studies were based on 9319 participants with pre-existing renal diseases including 2533 renal transplant recipients7,15,19,21,22,25,26,28,38,39,43. Nineteen studies included participants at high cardiovascular events risk9,11,13,17,18,20,26,27,29,30,31,32,35,36,37,40,41,42,44. Of these 19 studies, 14 studies involved 15584 participants with pre-existing cardiovascular disease9,13,18,26,27,29,30,32,35,36,37,40,41,42.

Table 1 Characteristics of studies included in meta-analysis.

Circulating ADMA concentration and all-cause mortality for all participants

All of the 34 studies included in the meta-analysis provided a multivariate HR, RR, or OR with a 95% CI for the pooled predictive effect of ADMA for all-cause mortality. Our results demonstrated that ADMA was independently associated with all-cause mortality (RR = 1.27, 95% CI: 1.20–1.34, Fig. 2). Moderate heterogeneity (I2 = 71%) existed for these studies. After excluding the four studies11,21,26,29 with fewer participants that might have come from the same study group used in four other studies9,22,27,30, circulating ADMA was still associated with all-cause mortality (RR = 1.25, 95% CI: 1.18–1.31), and the level of between-study heterogeneity was moderate (I2 = 71%).

Figure 2
figure 2

Circulating ADMA concentration and all-cause mortality for all participants.

Circulating ADMA concentration and all-cause mortality for participants with pre-existing renal diseases

After analyzing the data from 11 studies based on participants with pre-existing renal diseases7,15,19,21,22,25,26,28,38,39,43, we found that circulating ADMA still showed a predictive value for mortality (RR = 1.30, 95% CI: 1.19–1.43, Fig. 3). The value remaining after excluding one of the two studies that used patients from the same study group was almost identical (RR = 1.27, 95% CI: 1.16–1.39)21,22. However, the heterogeneity in both the 11 (I2 = 71%) and the 10 studies was significant (I2 = 67%).

Figure 3
figure 3

Circulating ADMA concentration and all-cause mortality for participants with pre-existing renal diseases.

Circulating ADMA concentration and all-cause mortality for participants with pre-existing cardiovascular disease

Fourteen studies included patients with pre-existing cardiovascular disease9,13,18,26,27,29,30,32,35,36,37,40,41,42. Circulating ADMA was also an independent predictor for all-cause mortality for participants with pre-existing cardiovascular disease (RR = 1.26, 95% CI: 1.16–1.37, Fig. 4). There was moderate heterogeneity among studies (I2 = 72%).

Figure 4
figure 4

Circulating ADMA concentration and all-cause mortality for participants with pre-existing cardiovascular disease.

After excluding one of the two studies using patients from the same study group, circulating ADMA was still associated with all-cause mortality (RR = 1.23, 95% CI: 1.13–1.33)29,30.

Circulating ADMA concentration and all-cause mortality for participants without pre-existing renal diseases or CVD

Consistent with the results of studies involving participants with CVD or renal disease, circulating ADMA was independently associated with all-cause mortality for participants without CVD or renal disease (RR = 1.31, 95% CI: 1.13–1.53, Fig. 5). The heterogeneity among studies was moderate (I2 = 76%).

Figure 5
figure 5

Circulating ADMA concentration and all-cause mortality for participants without pre-existing renal diseases or CVD.

Circulating ADMA concentration and cardiac events for all participants

Statistically significant associations existed between circulating ADMA and major cardiovascular events (RR = 1.18, 95% CI: 1.10–1.27, Fig. 6) and cardiovascular death (RR = 1.19, 95% CI: 1.12–1.25, Fig. 7).

Figure 6
figure 6

Circulating ADMA concentration and major cardiovascular events.

Figure 7
figure 7

Circulating ADMA concentration and cardiovascular death.

Publication bias

The funnel plot showed that there might be publication bias (Fig. 8). A Begg and Egger test was then used to test the publication bias for all studies and the studies enrolling participants with pre-existing diseases (p < 0.001 for both).

Figure 8
figure 8

Funnel plot for all publication.

There was no significant publication bias for the studies that enrolled participants with pre-existing renal diseases (p = 0.306). However, publication bias was observed for studies enrolling participants with pre-existing cardiovascular diseases (p = 0.017).

Discussion

We performed a systematic review and meta-analysis to investigate the association between circulating ADMA and all-cause mortality. Our analysis showed that high circulating ADMA was independently associated with all-cause mortality, and the association remained in patients with or without pre-existing renal disease and pre-existing CVD. In addition, high circulating ADMA was also associated with major cardiovascular events or cardiovascular death.

An elevation of circulating ADMA in patients with CVD and pre-existing renal disease has been reported9,18. A previous study reported that high ADMA was associated with CVD events, consistent with the results of the present subgroup analyses that high circulating ADMA was associated with both major cardiovascular events and cardiovascular death24. Additionally, previous studies also indicated that a high circulating ADMA concentration was inversely related to glomerular filtration rate and positively correlated with progression to dialysis10,26. Furthermore, ADMA could predict mortality in patients with CVD or chronic renal disease18,21.The predictive value of ADMA for CVD outcomes remained significant even in participants without pre-existing CVD or kidney disease at baseline6.

However, studies investigating the relationship between ADMA and all-cause mortality reported inconsistent results. Although some studies reported that ADMA was significantly associated with all-cause mortality11,16,18, other studies did not find that ADMA was associated with all-cause mortality32,33,42. The different results might be because of the limited number of participants involved in each single-center study in addition to the complicated role of ADMA in vivo.

Our present study included data from 34 studies with more than 35000 participants and our results suggest that increased levels of circulating ADMA are an independent predictor for all-cause mortality after multivariate Cox’s proportional-hazards model adjustment. The association was similar in patients with or without pre-existing chronic renal disease and pre-existing CVD.

The main results of our present study are consistent with a recent published meta-analysis by Schlesinger and colleagues that showed that ADMA was an independent risk marker for all-cause mortality and CVD45. However, we searched more databases than they did and included as many studies as possible. We also tested for publication bias.

Furthermore, Review Manager was used to carry out the meta-analysis evaluating the weight of each study to determine which studies influenced the final results more than the others did. In our present study, after excluding the studies ranked in the top three weights, significance remained in every meta-analysis. For example, after excluding the top three weighted studies18,30,46, ADMA was still independently associated with all-cause mortality (RR = 1.31, 95% CI: 1.23–1.40). In addition, although the studies included heterogeneous populations with different conditions at baseline or different interventions, we only included the HR, RR, or OR value from multivariate Cox’s proportional-hazards model analyses, which reduced the bias.

For both CVD and chronic renal disease, the relationship between ADMA and outcomes might involve endothelial dysfunction. The endothelial dysfunction might be because of disturbed NO regulation, which would lead to impaired biological activity of NO2,47. NO deficiency is attributed to two possible causes, substrate (L-arginine) limitations and increased levels of circulating endogenous inhibitors of NOS, particularly ADMA. Therefore, the elevation of circulating ADMA might be partly reflecting the imbalance of arginine and ADMA ratio that affects NO production and has been shown to be related to changes in microcirculation of major organs and increased hospital mortality48,49. In addition to its important role in endothelial regulation, NO is also involved in vascular smooth muscle cell growth, platelet aggregation, and leukocyte adhesion, which play important roles in microcirculation50. The reduction of NO production in glomerular endothelia might cause vascular damage and enhance endothelial adhesion of leukocytes and platelets, which would subsequently increase the mortality of patients with renal disease21.

In addition to its role in endothelial dysfunction and microcirculation, various studies have reported that ADMA is consistently associated with biomarkers of inflammation in chronic conditions including diabetes, renal disease, and hypertension38,51,52. ADMA can induce TNF-α and IL-8 production via oxidative stress due to generation of reactive oxygen species/NF-κB-dependent pathway in vitro53. Therefore, circulating ADMA might be a potential pro-inflammatory factor in addition to inhibiting NOS.

Furthermore, the prognostic role of ADMA in all-cause mortality might involve additional underlying mechanisms. ADMA belongs to a family of amino acid methylation derivatives including N-mono-methylarginine (MMA), the immediate precursor to ADMA, and symmetric dimethylarginine (SDMA), a stereoisomer of ADMA54. Low plasma MMA is the most potent NOS inhibitor and has been found to be inversely related to cardiovascular disease outcomes55. Unlike ADMA, SDMA lacks NOS inhibitory activity. However, SDMA is a weak inhibitor of arginine transporters54. Studies have also reported that SDMA is associated with an increased prevalence of major adverse cardiac events, renal dysfunction, and all-cause mortality55,56. These derivatives mentioned above also show predictive value for mortality in patients with pre-exiting CVD or renal disease55,56. Furthermore, some findings suggest that ADMA may directly promote vascular disease57. Therefore, these findings raised the possibility that the underlying mechanisms of ADMA in predicting all-cause mortality might be partly independent of NOS inhibition and NO production.

Various studies have explored the predictive value of some biomarkers in patients with cardiac failure and those undergoing hemodialysis58,59,60. However, none have had satisfactory high sensitivity and specificity. Since elevated plasma ADMA concentrations independently predict CVD and renal disease outcomes, ADMA may have potential utility as a clinical biomarker. Ideal biomarkers should also exhibit pharmacologic responses to a therapeutic intervention. Some drugs such as statins and angiotensin converting enzyme inhibitors and angiotensin receptor blockers can affect ADMA61. A meta-analysis reported a significant reduction in plasma ADMA concentrations following statin therapy62. However, the cut-off value of ADMA in prediction, whether the absolute increase in ADMA is clinically relevant, and whether earlier or more aggressive intervention can improve clinical outcomes remains unclear. Therefore, studies exploring the above-mentioned unclear problems are needed in the future.

Several limitations of our present study should be considered. First, not all of the studies were prospective, which might lead to biases. Second, the substantial heterogeneity of the included studies might decrease the power of the results. Third, we only included published studies written in English. This might lead to publication bias. However, the large number of participants included in the present meta-analysis provided sufficient data for calculation of the pooled effect of ADMA on all-cause mortality. Furthermore, we only used the OR, RR, or HR with 95% CI from multivariate Cox’s proportional-hazards analysis to calculate the pooled predictive value of ADMA on mortality, which should reduce any potential bias.

In summary, the present study suggests a positive association of circulating ADMA with all-cause mortality. Further studies are needed to investigate the associations in the general population, the effects of interventions on ADMA, and the value of ADMA as a biomarker, especially in patients with CVD and those with renal diseases.

Additional Information

How to cite this article: Zhou, S. et al. Asymmetric dimethylarginine and all-cause mortality: a systematic review and meta-analysis. Sci. Rep. 7, 44692; doi: 10.1038/srep44692 (2017).

Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.