Introduction

Diabetes mellitus (DM) is a highly investigated and complex chronic disease that has become increasingly prevalent with economic development. In 2016, DM was reported to be the eighth most prevalent cause of disease-related mortality. If DM is not properly managed, it can result in diabetic macroangiopathy (of the heart, brain, lower limb arteries, etc.), a specific form of accelerated atherosclerosis, as well as diabetic microangiopathy (of the kidney or eye, neuropathy, etc.), which are associated with increased morbidity and mortality. It is reported that 20–30% of diabetic patients have diabetic macroangiopathy [1]. Atherosclerosis in DM is more severe and aggressive than that in non-DM patients, and usually involves multiple arteries. DM also adversely affects the microvasculature in many organs, and remains the leading cause of chronic kidney disease and blindness [2]. Therefore, the ultimate goal of managing DM is to lower the risks of diabetic macroangiopathy (DMMA) and highly morbid microangiopathy (DMMI).

There are many risk factors associated with diabetic macroangiopathy and microangiopathy, including genetic factors, dyslipidemia, smoking, alcohol, obesity, exercise, and oxidative stress, but diabetic macroangiopathy and microangiopathy susceptibility is still not completely understood. Paraoxonase 1 is a 354-aa, 45-kDa glycoprotein that is synthesized in the liver and released into the blood, and which binds to high-density lipoprotein in a calcium-dependent manner [3]. Paraoxonase 1 is the main constituent of high-density lipoprotein (HDL), and contributes to the protective role of HDL against vascular diseases [4]. Paraoxonase 1 activity is reported to affect diabetic macroangiopathy and microangiopathy susceptibility [4,5,6]. There are large ethnic differences in the genetic polymorphisms of paraoxonase 1, according to the 1000 Genomes database [7]. A meta-analysis showed that paraoxonase 1 L55M and Q192R genetic polymorphisms were significantly associated with susceptibility to DM, but there were notable ethnic differences [7]. In 1998, Kao et al. reported that paraoxonase 1 L55M genetic polymorphism was significantly associated with diabetic retinopathy, whereas Q192R was not [8]. In another study, the author found that both (i.e., paraoxonase 1 L55M and Q192R) genetic polymorphisms were significantly related to diabetic complications [9]. Plenty of studies have focused on the associations of paraoxonase 1 Q192R and L55M genetic polymorphisms with diabetic macroangiopathy and microangiopathy susceptibility, but these links remain controversial. Therefore, the meta-analysis reported in the present paper was conducted to demonstrate these relationships.

Methods

Search Strategy

We carried out literature searches for papers published in English or Chinese in the Embase, PubMed, Wanfang Database, and China National Knowledge Internet (CNKI) databases using the following keywords: “PON1” or “paraoxonase 1,” “diabetes” or “diabetes mellitus” or “DM” or “Type 2 diabetes mellitus (T2DM)” or “Type 1 diabetes mellitus (T1DM),” and “variation” or “polymorphism” or “single nucleotide polymorphism” or “SNP.” We also searched through relevant articles manually to find additional reports.

Inclusion and Exclusion Criteria

Articles were included in this meta-analysis if they complied with the following criteria: (1) they were published as original case–control studies; (2) they reported the relationship(s) of paraoxonase 1 L55M and/or Q192R genetic polymorphisms with diabetic macroangiopathy and/or diabetic microangiopathy susceptibility; (3) they used DM without any complication as a control. When the same series of patients were used in more than one article, we used the latest and most complete study.

Data Extraction

WCF and WDL screened all of the articles found in the searches and extracted data from all eligible publications independently. For each study included in our meta-analysis, we carefully extracted the following information: first author, published year, country, ethnicity, type of DM, diagnostic criteria used for diabetic complications, and paraoxonase 1 L55M and/or Q192R genotypes in each group. If there were any discrepancies between the data extracted by WCF and WDL, they attempted to resolve the disagreement through discussion; if no agreement was reached, another author was consulted to resolve the dispute.

Compliance with Ethics Guidelines

This article is based on previously conducted studies and does not contain any studies with human participants or animals performed by any of the authors.

Statistical Analysis

Our meta-analysis followed the recommendations of the PRISMA statement and used the Newcastle–Ottawa Scale (NOS) criteria to assess study quality. Studies that met five or more of the NOS criteria were considered of high quality. Pooled odds ratios (ORs) were used to explore the relationships of paraoxonase 1 L55M and Q192R polymorphisms with the susceptibilities to diabetic macroangiopathy and/or diabetic microangiopathy. Heterogeneity was measured using the chi-square-based Q test. A random effects model was applied when I2 > 50% and P < 0.05. Otherwise, a fixed effects model was used. Publication bias was investigated using Egger’s test and a funnel plot. This meta-analysis was performed using STATA 12.0 software (Stata Corporation, TX, USA). P < 0.05 was taken to imply statistical significance.

Results

Study Characteristics

As shown in Fig. S1 of the Electronic supplementary material (ESM), 332 publications were identified, among which 111 duplicates and 57 irrelevant papers were subsequently excluded. Of the remaining 164 papers, 139 were excluded for the following reasons: the paper was a review, no diabetic macroangiopathy or microangiopathy groups were included, there was no DM without complication group, the paper was a duplicate, and other paraoxonase 1 genetic polymorphisms were studied. Thus, 25 articles were ultimately included in the present meta-analysis. Eleven publications were written in English [8,9,10,11,12,13,14,15,16,17,18] and 14 were written in Chinese [19,20,21,22,23,24,25,26,27,28,29,30,31,32]. Table 1 shows detailed information on those 25 studies. The name of the first author, the year published, country, ethnicity, the type of DM, the numbers and ages of the cases and controls, the diagnostic criteria used for diabetic complications, and the genetic polymorphisms of paraoxonase 1 studied in each work are presented. All studies were found to be of high quality according to the NOS criteria.

Table 1 Characteristics of the studies included in the meta-analysis of paraoxonase 1 L55M and Q192R in relation to diabetes complications
Full size table

Paraoxonase 1 L55M Genetic Polymorphism and Risk of Diabetic Macroangiopathy

As shown in Table 2, six studies probed the association between paraoxonase 1 L55M genetic polymorphism and the risk of diabetic macroangiopathy; these studies included 431 cases and 640 DM patients without complications as the control. Two studies were performed in Europe [10, 17], and four were carried out in Asia [9, 16, 31, 32]. The results indicated that there was no heterogeneity in the dominant model (LM + MM vs LL: I2 = 0.0%, P = 0.447, Fig. 1a, Table 4), homozygous model (MM vs LL: I2 = 30.7%, P = 0.217, Fig. 1b, Table 4), allelic contrast model (M vs L: I2 = 16.8%, P = 0.305, Fig. 1c, Table 4), recessive model (MM vs LL + LM: I2 = 19.9%, P = 0.288, Fig. 1d, Table 4), and heterozygous model (LM vs LL: I2 = 0.0%, P = 0.610, Fig. 1e, Table 4). Meta-analysis suggested that there was no significant association of paraoxonase 1 L55M genetic polymorphism with susceptibility to diabetic macroangiopathy in all five models (LM + MM vs LL: OR 0.98, 95% CI 0.69–1.38, P = 0.996, Fig. 1a; MM vs LL: OR 1.30, 95% CI 0.81–2.08, P = 0.279, Fig. 1b; and M vs L: OR 1.10, 95% CI 0.87–1.39, P = 0.414, Fig. 1c; MM vs LLLM: OR 1.30, 95% CI 0.99–1.90, P = 0.176, Fig. 1d; LM vs LL: OR 0.90, 95% CI 0.62–1.30, P = 0.569, Fig. 1e; Table 4). When the effects models were altered, the significances of these three models did not change statistically significantly (data not shown). Based on the funnel plot and Egger’s test, none of the three models had significant publication bias (LM + MM vs LL: t = − 0.040, P = 0.970; MM vs LL: t = 0.300, P = 0.781; M vs L: t = − 0.340, P = 0.752; MM vs LL + LM: t = 0.69, P = 0.539; LM vs LL: t = − 0.10, P = 0.923; data not shown).

Table 2 Characteristics of the studies included in this meta-analysis of the relationship between paraoxonase 1 L55M polymorphism and risk of diabetes complications
Full size table
Fig. 1a–e
figure 1

Forest plots for the association of paraoxonase 1 L55M genetic polymorphism with diabetic macroangiopathy. a Dominant model, LM + MM vs LL; b homozygous model, MM vs LL; c allelic contrast model, M vs L; d recessive model, MM vs LL + LM; e heterozygous model, LM vs LL

Full size image

Paraoxonase 1 L55M Genetic Polymorphism and Susceptibility to Diabetic Microangiopathy

As shown in Table 2, nine studies focused on the association of paraoxonase 1 L55M genetic polymorphism with the risk of diabetic microangiopathy; these studies included 992 cases and 1212 controls [8, 10, 12, 13, 15, 17,18,19, 30]. The results showed that there was significant heterogeneity in the dominant model (LM + MM vs LL: I2 = 81.0%, P = 0.000, Fig. 2a, Table 4), homozygous model (MM vs LL: I2 = 80.9%, P = 0.000, Fig. 2b, Table 4), allelic contrast model (M vs L: I2 = 84.1%, P = 0.000, Fig. 2c, Table 4), recessive model (MM vs LL + LM: I2 = 68.6%, P = 0.001, Fig. 2d, Table 4), and heterozygous model (LM vs LL: I2 = 71.1%, P = 0.000, Fig. 2e, Table 4). Thus, subgroup analysis of ethnicity, type of DM, and Hardy–Weinberg equilibrium (HWE) was performed, and the results indicated that ethnicity and type of DM can explain the heterogeneity in the recessive model, but not the heterogeneity in the other models (Table 4). Meta-regression analysis was then performed using the covariates published year, sample size, ethnicity, type of DM, and HWE to investigate the sources of the between-study heterogeneity, but it failed to identify those sources (Table 5). Paraoxonase 1 L55M genetic polymorphism was clearly related to diabetic microangiopathy susceptibility in the dominant model (OR 0.53, 95% CI 0.33–0.83, P = 0.006, Fig. 2a, Table 4), the homozygous model (OR 0.37, 95% CI 0.16–0.86, P = 0.021, Fig. 2b, Table 4), the allelic contrast model (OR 0.62, 95% CI 0.43–0.90, P = 0.011, Fig. 2c, Table 4), the recessive model (OR 12.04, 95% CI 8.02–18.06, P = 0.000, Fig. 2d, Table 4), and the heterozygous model (OR 0.57, 95% CI 0.38–0.85, P = 0.006, Fig. 2e, Table 4). When the effects models were changed, the statistical significance did not alter (data not shown). Meanwhile, the funnel plot and Egger’s test indicated that there was no significant publication bias in the models (LM + MM vs LL: t = − 0.640, P = 0.543; MM vs LL: t = − 1.750, P = 0.124; M vs L: t = − 0.710, P = 0.503; LM vs LL: t = − 0.540, P = 0.607; data not shown) except for the recessive model (MM vs LL + LM: t = 2.580, P = 0.037, data not shown).

Fig. 2a–e
figure 2

Forest plots for the association of paraoxonase 1 L55M genetic polymorphism with diabetic microangiopathy. a Dominant model, LM + MM vs LL; b homozygous model, MM vs LL; c allelic contrast model, M vs L; d recessive model, MM vs LL + LM; e heterozygous model, LM vs LL

Full size image

Paraoxonase 1 Q192R Genetic Polymorphism and Susceptibility to Diabetic Macroangiopathy

As shown in Table 3, twelve studies focused on the association of paraoxonase 1 Q192R genetic polymorphism with the risk of diabetic macroangiopathy; these studies included 827 cases and 1069 controls [9, 10, 16, 17, 22, 23, 25,26,27,28,29, 32]. The results showed that there was significant heterogeneity in the dominant model (QR + RR vs QQ: I2 = 63.6%, P = 0.001, Fig. 3a, Table 4), homozygous model (RR vs QQ: I2 = 64.5%, P = 0.001, Fig. 3b, Table 4), allelic contrast model (R vs Q: I2 = 66.5%, P = 0.001, Fig. 3c, Table 4), recessive model (RR vs QQ + QR: I2 = 45.7%, P = 0.042, Fig. 3d, Table 4), and heterozygous model (QR vs QQ: I2 = 52.5%, P = 0.017, Fig. 3e, Table 4). Subgroup analysis of ethnicity, type of DM, and HWE was therefore conducted, but it failed to find the sources of heterogeneity (Table 4). Meta-regression analysis was then performed using the covariates published year, sample size, ethnicity, type of DM, and HWE. The results showed that sample size could explain the heterogeneity in the dominant allelic and heterozygous models, and HWE could explain the heterogeneity in the homozygous and recessive models (Table 5). Paraoxonase 1 Q192R genetic polymorphism was significantly related to diabetic macroangiopathy susceptibility in the homozygous model (OR 1.88, 95% CI 1.06–3.32, P = 0.030, Fig. 3b, Table 4), allelic contrast model (OR 1.31, 95% CI 1.02–1.69, P = 0.038, Fig. 3c, Table 4), and recessive model (OR 1.55, 95% CI 1.11–2.16, P = 0.010, Fig. 3d, Table 4), but not the dominant model (OR 1.35, 95% CI 0.88–2.08, P = 0.163, Fig. 3a, Table 4) or heterozygous model (OR 1.20, 95% CI 0.81–1.78, P = 0.370, Fig. 3e, Table 4). Sensitivity analysis indicated that when the effects models were changed, the statistical significance did not alter (data not shown). Meanwhile, there was no significant publication bias according to the funnel plot and Egger’s test (QR + RR vs QQ: t = 1.44, P = 0.179; RR vs QQ: t = 1.46, P = 0.174; R vs Q: t = 1.00, P = 0.343; RR vs QQ + QR: t = 1.30, P = 0.224; QR vs QQ: t = 1.03, P = 0.326; data not shown).

Table 3 Characteristics of the studies included in this meta-analysis of the relationships between paraoxonase 1 Q192R polymorphism and the risks of diabetic macroangiopathy and microangiopathy
Full size table
Fig. 3a–e
figure 3

Forest plots for the association of paraoxonase 1 Q192R genetic polymorphism with diabetic macroangiopathy. a Dominant model, QR + RR vs QQ; b homozygous model, RR vs QQ; c allelic contrast model, R vs Q; d recessive model, RR vs QQ + QR; e heterozygous model, QR vs QQ

Full size image
Table 4 Main results of the pooled data in this meta-analysis
Full size table
Table 5 The meta-regression results for the associations between paraoxonase 1 genetic polymorphism and diabetes macroangiopathy/microangiopathy susceptibility
Full size table

Paraoxonase 1 Q192R Genetic Polymorphism and Susceptibility to Diabetic Microangiopathy

As shown in Table 3, twelve studies focused on the association of paraoxonase 1 Q192R genetic polymorphism with the risk of diabetic microangiopathy; these studies included 1455 cases and 1353 controls [8, 10,11,12, 14, 17, 19,20,21, 24, 25, 30]. The results showed that there was significant heterogeneity in the dominant model (QR + RR vs QQ: I2 = 53.5%, P = 0.014, Fig. 4a, Table 4), homozygote model (RR vs QQ: I2 = 60.4%, P = 0.004, Fig. 4b, Table 4), and allelic contrast model (R vs Q: I2 = 59.3%, P = 0.005, Fig. 4c, Table 4) but not in the recessive model (RR vs QQ + QR: I2 = 39.8%, P = 0.075, Fig. 4d, Table 4) and heterozygous model (QR vs QQ: I2 = 35.1%, P = 0.109, Fig. 4e, Table 4). Subgroup analysis of ethnicity, type of DM, and HWE and meta-regression analysis using the covariates published year, sample size, ethnicity, type of DM, and HWE were therefore performed to investigate the sources of heterogeneity. Unfortunately, the sources of heterogeneity were not found (Tables 4, 5). There was no marked association between paraoxonase 1 Q192R genetic polymorphism and susceptibility to diabetic microangiopathy in the dominant model (OR 1.21, 95% CI 0.90–1.64, P = 0.209, Fig. 4a, Table 4), the homozygous model (OR 1.33, 95% CI 0.86–2.05, P = 0.119, Fig. 4b, Table 4), the allelic contrast model (OR 1.12, 95% CI 0.93–1.35, P = 0.227, Fig. 4c, Table 4), the recessive model (OR 1.12, 95% CI 0.93–1.33, P = 0.225, Fig. 4d, Table 4), and the heterozygous model (OR 1.12, 95% CI 0.92–1.36, P = 0.272, Fig. 4e, Table 4). Sensitivity analysis indicated that when the effects models were changed, the statistical significances did not alter (data not shown). Meanwhile, according to the funnel plot and Egger’s test, there was no significant publication bias in the models (QR + RR vs QQ: t = 2.19, P = 0.054; RR vs QQ: t = 1.35, P = 0.207; R vs Q: t = 0.69, P = 0.505, RR vs QQ + QR: t = 0.61, P = 0.555; data not shown) except in the heterozygous model (QR vs QQ: t = 2.36, P = 0.040, data not shown).

Fig. 4a–e
figure 4

Forest plots for the association of paraoxonase 1 Q192R genetic polymorphism with diabetic microangiopathy. a Dominant model, QR + RR vs QQ; b homozygous model, RR vs QQ; c allelic contrast model, R vs Q; d recessive model, RR vs QQ + QR; e heterozygous model, QR vs QQ

Full size image

Discussion

This study is the first to demonstrate that paraoxonase-1 L55M and Q192R genetic polymorphisms are associated with susceptibility to diabetic macroangiopathy and microangiopathy. The results showed that paraoxonase 1 L55M genetic polymorphism was significantly related to diabetic microangiopathy susceptibility, but not diabetic macroangiopathy susceptibility, in the dominant, homozygous, and allelic contrast models. There was also a significant association of paraoxonase 1 Q192R genetic polymorphism with diabetic macroangiopathy susceptibility in the homozygous and allelic contrast models, but not in the dominant model. Meanwhile, there was no significant relationship between paraoxonase 1 Q192R genetic polymorphism and susceptibility to diabetic microangiopathy.

Paraoxonase 1 has been shown to have antioxidant properties and to protect low-density lipoprotein from oxidative damage. The 192Q allele of paraoxonase 1 is more effective than 192R at preventing low-density lipoprotein oxidation [33]. The 192R allele was found to be significantly associated with lower serum paraoxonase 1 activity in DM patients in an Egyptian population [16]. A recent meta-analysis also demonstrated that Q192R genetic polymorphism was significantly related to the risk of type 2 diabetes mellitus, although there were some ethnic differences [7]. However, the relationships between paraoxonase 1 Q192R genetic polymorphism and various diabetic complications are considered to be rather unclear. Therefore, the meta-analysis reported in the present paper was performed to examine these relationships, and we found that there was a significant relationship between paraoxonase 1 Q192R genetic polymorphism and susceptibility to diabetic macroangiopathy in the homozygous and allelic contrast models, but not in the dominant model. Meanwhile, there was no significant association between paraoxonase 1 Q192R genetic polymorphism and diabetic microangiopathy susceptibility. However, there were significant heterogeneities in all three of these models when the associations of Q192R with diabetic macroangiopathy and microangiopathy susceptibility were analyzed. Meta-regression analysis was performed using the covariates published year, sample size, ethnicity, and type of DM, but it failed to identify the sources of heterogeneity.

Paraoxonase 1 L55M genetic polymorphism, another common single nucleotide polymorphism of paraoxonase 1, can affect the enzyme activity of paraoxonase 1, with the L allele linked to a higher serum concentration and greater stability to proteolysis [33, 34]. A recent meta-analysis found that L55M polymorphism was significantly associated with many diseases, such as any cancer, breast cancer [35], and diabetes mellitus [7]. Paraoxonase 1 L55M genetic polymorphism was also found to be significantly associated with diabetic retinopathy in a meta-analysis which included just two articles [36]. Thus, before the present meta-analysis was carried out, the relationships of L55M genetic polymorphism with diabetic macroangiopathy and microangiopathy were not very clear. In this meta-analysis, L55M genetic polymorphism was found to be significantly associated with susceptibility to diabetic microangiopathy but not susceptibility to diabetic macroangiopathy in the dominant, homozygous, and allelic contrast models. Thus, subgroup analysis and meta-regression were conducted, and the results showed that type of DM was able to explain the heterogeneity in the recessive model, but not in the other models.

It is important to note the limitations of this meta-analysis. First, only articles written in English and Chinese were included in this study, and there may have been some publication bias, although Egger’s test and a funnel plot did not point to any significant publication bias. Second, we used composites of diabetic complications in the present meta-analysis. Although macroangiopathy and microangiopathy are two kinds of diabetic vascular complications, the pathophysiological mechanisms and the contributions of genetic polymorphisms were probably different for these complications. Third, significant heterogeneity was found in our analysis, but subgroup analysis and meta-regression failed to find any of the sources of the between-study heterogeneity in the associations of paraoxonase 1 L55M/Q192R genetic polymorphisms with the risk of diabetic microangiopathy. Therefore, further well-designed studies with large samples are needed to confirm the results of this meta-analysis.

Conclusions

Paraoxonase 1 L55M and Q192R genetic polymorphisms play important roles in the susceptibility to diabetic macroangiopathy and microangiopathy.