The novel findings of this study are as follows. First, a GGT1 variant, rs4820599, the G allele and a low HDL-C level were identified to be risk factors for a high baPWV in Japanese subjects with T2DM in the longitudinal analysis. Interestingly, a significant interactive effect of the GGT1 genotype and low HDL-C level on the risks of both a high baPWV and DR was found. Second, the HDL-C level at baseline was identified to be a significant predictor of a high baPWV only in G allele carriers according to the ROC analysis. This result in the T2DM patients was also noted in the general population. Third, the GGT1 genotype was not associated with the risk of DR, although it affected the principal factors involved in the risk of DR. Fourth, the associations between the levels of GGT and fasting plasma glucose, HDL-C and LDL-C and between dyslipidemia and a high baPWV were significant only in G allele carriers among the general subjects. In this study, the triglyceride levels were positively associated with the serum GGT levels but not the risk of a high baPWV or DR, and the LDL-C levels were positively associated with the serum GGT levels only in the general subjects with the G allele.
The mean serum log GGT level was significantly higher in G allele carriers than in non-carriers among both the T2DM patients and the general population. The expression of GGT increases as an adaptive response to oxidative stress, although the mechanism underlying GGT regulation in response to oxidative stress remains unclear [9,32]. Human GGT is a multigene family consisting of at least seven GGT genes or pseudogenes, in which a major GGT1 and minor GGT5 have been found to encode proteins exhibiting GGT activity . A variant, rs4820599, on the GGT1 gene is predicted to be located at a transcription factor binding site, which can result in differential gene transcription, according to FuncPred (http://snpinfo.niehs.nih.gov/snpinfo/snpfunc.htm, accessed 3/31/2015). This fact may explain the findings of the present study and previous observations obtained using GWAS [27-29]. In addition to GGT1, however, population-based GWAS have revealed evidence of an association between the serum GGT level and a series of SNPs in candidate genes, i.e. HNF1A in various ethnic groups [27-29], C14orf73 and RORA in subjects of European descent , MYL2, C12olf51 and OAS1 in individuals of East Asian descent  and ALDH2 in Japanese . A pathway analysis of the SNP associations showed significant overlap between genes affecting the GGT level and those affecting common metabolic and inflammatory diseases under the control of the Hepatic Nuclear family . Therefore, our results provide only the first glimpse into the association between one of these SNPs and the GGT level, factors affecting the GGT level and the risk of vascular disease resulting from metabolic and inflammatory diseases, as represented by T2DM.
The baPWV is used in epidemiological and clinical research as an index of arterial stiffness and atherosclerosis and is indicated to be an independent predictor of CVD [30,34-36]. A few epidemiological studies have examined the association between the GGT level and baPWV, although the findings are inconsistent [35,36]. In the present study, the GGT1 G allele was identified to be a risk factor for a high baPWV, while the GGT level was found to be a risk factor in non-carriers. GGT is expressed on the cell surface membrane of most cell types, although only liver GGT is detected in the serum, and the relationship between the cellular GGT and serum GGT expressions is unknown [6,37]. The serum GGT level appears to be associated with the risk of CVD from the early stage of arterial stiffness through the onset of CVD events [2,4,6,7,35,36,38]. This is because oxidative stress, vascular inflammation and endothelial dysfunction play a central role in the pathogenesis of atherosclerosis [39,40], and because the serum GGT level is proposed to be an early and sensitive marker of oxidative stress [7,9,32]. A key question is whether the serum GGT level is an simply innocent marker or whether its activity is causally involved in the development of arterial stiffness and atherogenesis, as shown in atheroma plaques [6,7]. Cellular GGT is known to be involved in the generation of reactive oxygen species in the presence of transition metals and serum GGT is speculated to promote the oxidation of circulating lipoproteins [6,7,9,41].
The serum HDL-C level has been reported to be inversely related to the baPWV in a general Chinese population (50 to 90 years of age) and the aortic PWV in other populations . An isolated low HDL-C level is a dyslipidemic phenotype that appears to be more prevalent among Asian populations, in whom low HDL-C levels are strongly associated with an increased CHD risk . Several potential mechanisms may explain the association between the serum HDL-C level and CVD [15,20-22,42-44]. HDL has a role in reverse cholesterol transport and other direct actions on numerous cell types, thereby reducing cardiovascular risks. In endothelial cells and their progenitors, HDL prevents apoptosis and stimulates proliferation and migration. HDL also has diverse anti-inflammatory actions in both endothelial cells and leukocytes. In vascular smooth muscles, HDL attenuates proinflammatory, promigratory and degradative processes. HDL has an antithrombotic effect via its actions on the endothelium and platelets. In addition, HDL protects pancreatic β-cells from apoptosis, decreases the white adipose tissue mass, increases energy expenditures and promotes the production of adiponectin, which possesses its own vascular protective properties. However, the classic HDL (concentration) hypothesis is gradually being replaced by the triglyceride hypothesis and the HDL function hypothesis due to the failure of clinical trials and negative results for human genetics [12-14,18-22]. The classic entity of ‘diabetic dyslipidemia’ is characterized by the so-called atherogenic lipid triad, consisting of an increase in small dense LDL particles and triglyceride-rich lipoproteins and a decrease in HDL-C [12-17,22]. In this study, the triglyceride levels were positively associated with the serum GGT levels, but not any of the risk factors for a high baPWV or DR in both the T2DM patients and general subjects. In order to prove that a low HDL-C level, rather than a high triglyceride level, is a risk factor for a high PWV, we applied a multivariate model including triglycerides on purpose despite the insignificance of this parameter as a covariate. Most serum GGT is bound to lipoproteins, more dominantly to Apo A than to Apo B, although the ratio may change depending on the serum lipid profile [6,8-11,41]. All things considered, the serum GGT level is associated with cardiovascular risks, including dyslipidemia, more greatly in G allele carriers than in non-carriers, and coexisting low HDL-C may lead to increased LDL-associated GGT and GGT-dependent LDL oxidation followed by the possible development of significant arterial stiffness in this population.
The interactive effect of the GGT1 genotype and a low HDL-C level on the risk of DR was also significant in the current study, and a low HDL-C level was found to be an independent risk factor for DR only in G allele carriers. Oxidative stress plays a pivotal role in the development of DR as well as macrovascular diseases [45-49]. The metabolic abnormalities associated with diabetes cause mitochondrial superoxide overproduction in the endothelial cells of both small and large vessels [45,48,49]. This increased superoxide production causes the activation of major pathways involved in the pathogenesis of various complications, such as polyol pathway flux as well as the increased formation of advanced glycation end products, activation of protein kinase C and over-activity of the hexosamine pathway [45,48,49]. The overexpression of superoxide dismutase in transgenic diabetic mice prevents DR, nephropathy and cardiomyopathy [48,49]. In addition to two principal and reversible risk factors for DR, blood glucose and BP, a low HDL-C level is indicated to be a risk factor for DR [16,23,25,46,49]. Many of the pathogenic effects of lipoproteins occur after these particles leak from the circulation [16,25,49]. Du et al.  identified extravasated Apo A1 and Apo B in human diabetic eyes and showed that native HDL completely blocks oxidative stress and the apoptosis of retinal pigment epithelial cells induced by heavily oxidized glycated LDL, thereby suggesting an important new role for extravasated and modified plasma lipoproteins in promoting DR. We speculate that GGT may co-localize with extravasated HDL and LDL and possibly take part in the pathogenesis of DR in G allele carriers with a low HDL-C level. In the present study, among G allele carriers, the diabetes duration and ALDH2*2 allele were other independent risk factors for DR. We previously reported a significant association between the ALDH2*2 allele and DR, as follows: the incidence of DR is significantly higher in ALDH2*2 allele carriers with a high GGT level (>37 IU/L for males and > 26 IU/L for females) than in non-carriers with a high or low GGT level; a high GGT level in non-carriers is most significantly associated with drinking habits, while that in ALDH2*2 allele carriers is significantly associated with multiple cardiovascular risk factors; and ALDH2 therefore may protect the vasculature against reactive aldehydes generated under conditions of sustained oxidative stress . In non-carriers, a female gender, age and the HbA1c level, which is associated with a high GGT level, were found to be independent risk factors for DR.
Diabetes and hypertension promote adverse changes throughout the vascular tree, eliciting both macrovascular and microvascular complications [34,38,45,46]. Increased arterial stiffness and microvascular remodeling are the most prevalent and earliest forms of organ damage in these diseases [34,38,45,46]. Therefore, diabetic subjects with a high PWV and/or MVCs appear to be particularly prone to developing accelerated atherosclerosis and premature death [38,46,47]. A significant positive association between the presence of DR and baPWV has been reported in Japanese T2DM patients without macrovascular complications . The association between a high baPWV and DR is likely to exist in G allele carriers, although this finding did not reach statistical significance in the present study due to the small sample size (data not shown). Although fibrates fail to reduce cardiovascular events in patients treated with statins in general [14,18], fenofibrate has received major attention as a novel medical treatment for DR and other diabetes-induced MCVs . This is because well-designed clinical trials, i.e. the Fenofibrate Intervention and Event Lowering in Diabetes (FIELD) and ACCORD trials, demonstrated large reductions in the progression of DR and the subsequent need for laser intervention, in addition to reductions in adverse renal and neurological outcomes, in patients with T2DM . Fenofibrate regulates the expression of many different genes, with a range of beneficial effects for lipid control, inflammation, angiogenesis and cell apoptosis [24,51]. Interestingly, treatment with 200 mg/day of fenofibrate for 48 weeks in patients with non-alcoholic fatty liver disease was recently shown to improve metabolic syndrome, in addition to glucose and liver parameters, including a reduction in the serum GGT level of 39% . These findings and the current results suggest that the early detection and treatment of low HDL-C levels in populations at high risk for MVC and CVD, such as T2DM patients with arterial stiffness and/or DR, may be an attractive therapeutic target for prevention [6,7,15-17,24,34,38,45,46,51]. The GGT1 G allele is a novel candidate risk factor in this population and we hope to reanalyze the findings of previous clinical trials of agents used to increase the HDL-C levels, stratifying the outcome according to the GGT1 genotype.
The strength of this study is that significant interactive effects between the GGT1 genotype and a low HDL-C level were observed on both a high baPWV and DR in two different populations and different analyses. The crucial limitations of this study include the retrospective study design, small sample size and lack of information on factors such as the effects of drug therapy, the HDL-C fraction and function and specific biomarkers of oxidative stress.
We herein presented for the first time the significant interactive effects of the GGT1 G allele and a low HDL-C level on a high baPWV and DR. These findings may suggest a common pathological mechanism in case of diabetic macro- and micro-angiopathy and prompt repeat-analyses of previous trials according to the GGT1 genotype as well as the development of future clinical trials comparing the efficacy of agents increasing the HDL-C level in improving angiopathy between GGT1 genotypes. However, well-designed studies in larger cohorts are still needed to further confirm our results.