Functional AGXT2 SNP rs37369 Variant Is a Risk Factor for Diabetes Mellitus: Baseline Data From the Aidai Cohort Study in Japan

multiplicative

Alanine-glyoxylate aminotransferase 2 (AGXT2), first identified in 1969, is an important enzyme that regulates the metabolism of methylarginines, such as ADMA and SDMA (6). AGXT2 is located on chromosome 5p13. In a study in Egypt, serum concentrations of ADMA and SDMA were higher in the TT genotype of AGXT2 singlenucleotide polymorphism (SNP) rs37369 (V140I) compared with the CC genotype, and in the AA genotype of rs16899974 (V498L) compared with the CC genotype (7). In a German study, plasma concentrations of SDMA, but not ADMA, were higher in those heterozygous or homozygous for the minor allele of rs37369, whereas no such positive relationship was found between rs16899974 and plasma concentrations of ADMA or SDMA (8). AGXT2 SNPs have also been shown to be related to the following vascular conditions and diseases and related biological data: coronary artery disease (7); heart rate variability (9); carotid atherosclerosis (10); atrial fibrillation (11); chronic heart failure (12); and blood pressure, blood glucose and blood creatinine (13).
In view of the lack of epidemiologic information regarding the relationship between AGXT2 SNPs and the risk of DM, in this caseecontrol study we investigated this issue in Japanese adults using baseline data from the Aidai Cohort Study (AICOS). In addition, haplotype analyses were performed, and the possibility of interaction between SNPs and smoking history was assessed. Tobacco products acutely increase the endothelial production of ADMA (16), and studies in China have reported a possible interaction between SNP rs37369 and smoking with regard to heart disease (12,17).

Study population
The AICOS is an ongoing population-based prospective study that started in 2015 (18e22). The present caseecontrol study was conducted using baseline data from the AICOS collected in Yawatahama city in 2015, Uchiko town in 2016 and Seiyo city and Ainan town in 2017. Yawatahama city, Uchiko town, Seiyo city and Ainan town, with total populations of nearly 36,000, 17,000, 38,000 and 22,000, respectively, are 4 of the 20 municipalities in Ehime Prefecture on Shikoku Island, which is located south of Japan's main island. In Yawatahama city in 2015, Uchiko town in 2016, Seiyo city in 2017 and Ainan town in 2017, 798, 347, 524 and 755 study participants, respectively, were recruited from among those who had received health checkups conducted by their municipality of residence or through one of several alternative recruiting procedures. In addition, 223 employees of a manufacturing company (PHC Corp) in Toon city also participated in the AICOS. Ultimately, 2,647 participants, aged 33 to 89 years (1,069 men aged 35 to 89 years and 1,578 women aged 33 to 85 years), gave written informed consent and completed the baseline questionnaire. From among these participants, 257 were excluded because of missing or incomplete data on the factors under study, leaving data on 2,390 participants available for analysis (977 men aged 35 to 88 years and 1,413 women aged 34 to 85 years). The AICOS was approved by the ethics committee of the Ehime University Graduate School of Medicine (Approval No. 1504012).

Measurements
A self-administered questionnaire elicited information on age; sex; smoking habits; alcohol consumption habits; physical activity level; education; and current use of antihypertensive, cholesterol-lowering and diabetic medications. Never smoking was defined as having smoked <100 cigarettes over the lifetime. Former smoking was defined as having smoked >100 cigarettes over the lifetime, but having quit smoking by the time of the survey. Current smoking was defined as having smoked >100 cigarettes over the lifetime and still smoking at the time of the survey. Leisure-time physical activity was defined as present if the study participant had engaged in at least 30 minutes of any type of moderate-to-vigourous physical activity, such as brisk walking, golf, gardening, jogging or playing tennis, at least once a week. Body height and weight were measured in light clothing without shoes, and body mass index (BMI) was calculated as body weight (in kilograms) divided by height (in metres squared). An automated sphygmomanometer was used to take 2 blood pressure measurements, each with the study subject in the sitting position after at least 5 minutes of rest; the second measurement was used for the present study. Hypertension was defined as systolic blood pressure !140 mmHg, diastolic blood pressure !90 mmHg or current use of antihypertensive medication. Blood samples were collected from an antecubital vein after overnight fasting. Serum low-density lipoprotein cholesterol, high-density lipoprotein cholesterol, triglyceride concentrations, plasma glucose concentrations and glycated hemoglobin (A1C) levels were measured at an external laboratory (Shikoku Chuken, Ehime, Japan). Dyslipidemia was defined as a serum low-density lipoprotein cholesterol concentration !140 mg/dL, high-density lipoprotein cholesterol concentration <40 mg/dL, triglyceride concentration !150 mg/dL or current use of cholesterol-lowering medication. DM was defined as a fasting plasma glucose level !126 mg/dL, A1C level !6.5% or current use of diabetic medication. Among the 2,390 participants, 190 cases of DM were identified; the 2,290 remaining participants served as control subjects.

Statistical analysis
Calculations of the HardyeWeinberg equilibrium were performed among the controls using a chi-square test. Linkage disequilibrium was examined using Haploview software version 4.2 (Broad Institute, Cambridge, Massachusetts, United States) (23). Using logistic regression analysis, crude odds ratios (ORs) and 95% confidence intervals (CIs) for DM associated with the SNPs under investigation were calculated, with the reference category being the homozygote of the ancestral allele based on the SNP database of the National Center for Biotechnology Information. Multiple logistic regression analysis was performed to adjust for age, sex, smoking status, alcohol consumption, leisuretime physical activity, education, BMI, hypertension and dyslipidemia. The statistical power calculation was performed using QUANTO version 1.2.4 (24). Haplotypes and their frequencies were inferred by means of the expectation maximization algorithm using Stata/SE software version 13.1 (StataCorp, College Station, Texas, United States). For differences in haplotype frequency between the cases and controls, crude ORs and 95% CIs were estimated based on the frequency of each haplotype relative to all other haplotypes combined. The multiplicative interaction was estimated by introducing a multiplicative term into a multiple logistic regression model. All analyses, excluding the calculation of linkage disequilibrium, statistical power and haplotype analysis, were conducted using SAS version 9.4 (SAS Institute, Cary, North Carolina, United States).

Results
Compared with the control subjects, the cases were more likely to be older, male and current alcohol drinkers, to report higher BMI, hypertension and dyslipidemia, and to report never having smoked (Table 1). No differences were found between cases and controls with respect to leisure-time physical activity and education.
Compared with study participants with the CC genotype of AGXT2 SNP rs37369, those with the TT, but not the CT, genotype had a significantly increased risk of DM (Table 2). Adjustment for confounders under study did not materially change the result: the adjusted OR for the TT genotype was 1.83 (95% CI, 1.04 to 3.47). AGXT2 SNPs rs37370 and rs180749 were not significantly associated with the risk of DM. The adjusted OR for SNP rs180749 under the additive model was 0.90 (95% CI, 0.70 to 1.17); the statistical power calculation revealed that, using our sample size, we could detect a geneedisease association for an OR of 0.703 with an accuracy of >80% at a significance level of 0.05 with a 2-sided alternative hypothesis under the log-additive model. Seven haplotypes with a frequency of !1% in both cases and controls were detected ( Table 3). Given that the haplotype order was rs37370, rs37369 and rs180749, the CTA haplotype was significantly positively associated with the risk of DM compared with all other haplotypes combined: the crude OR was 1.25 (95% CI, 1.01 to 1.56). By contrast, the CCA haplotype was significantly inversely related to the risk of DM in comparison to all other haplotypes combined: the crude OR was 0.53 (95% CI, 0.27 to 0.95).
Compared with having never smoked, former and current smoking was not significantly related to the risk of DM (adjusted ORs: 1.12 [95% CI, 0.73 to 1.71] and 1.73 [95% CI, 0.94 to 3.07], respectively). We investigated the potential interaction between AGXT2 SNP rs37369 and smoking status with respect to the risk of DM (Table 4). A positive association was observed between current smoking and risk of DM among participants with at least 1 C allele of AGXT2 SNP rs37369, whereas no association was found between current smoking and risk of DM in participants with the TT genotype; the multiplicative interaction between AGXT2 SNP rs37369 and current smoking was not statistically significant (p¼0.32 for interaction).

Discussion
To our knowledge, this current caseecontrol study is the first to report an association between AGXT2 SNPs and the risk of DM. We have shown that, compared with the CC genotype of AGXT2 SNP rs37369, the TT genotype, occurring in 39.9% of the control subjects, was independently positively related to DM, whereas the CT genotype was not associated with DM. No relationships were observed between AGXT2 SNPs rs37370 or rs180749 and DM. Haplotype analyses identified a significant positive association between the CTA haplotype of rs37370, rs37369 and rs180749, occurring in 36.5% of the control subjects, and DM, and a significant inverse relationship between the CCA haplotype, occurring in 5.8% of the control subjects, and DM.
A caseecontrol study of 100 patients with coronary artery disease and 50 healthy controls conducted in Egypt showed that the rs37369 T allele and the rs16899974 A allele were significantly positively associated with coronary artery disease (7). With respect to the relationship with SNP rs37369, this result is in partial agreement with our findings. The rs37369 GG genotype was significantly associated with an increased risk of chronic heart failure in a Chinese caseecontrol study of 1,000 cases and 1,200 healthy controls (12). The Young Finns Study showed that the rs37369 C and rs16899974 C alleles were associated with an increase in the ratio between the low-and high-frequency spectral components of short-term heart rate variability (9). SNP rs37370, but not rs37369 or rs180749, and the aforementioned CAAA haplotype were significantly related to carotid atherosclerosis in a study of 1,426 Japanese individuals (10). Another study of 750 older Japanese adults showed that SNP rs16899974 and the CAAA haplotype, but not rs37370, rs37369 or rs180749, were significantly associated with diastolic blood pressure, and that rs16899974, but not rs37370, rs37369, rs180749 or the CAAA haplotype, was significantly associated with casual blood glucose (13). In another Chinese caseecontrol study of 942 patients with coronary heart disease and 1,103 controls, rs37369 was not independently related to coronary heart disease (17). SNP rs16899974, but not rs37369, was significantly associated with atrial fibrillation in a study of 1,834 individuals with atrial fibrillation and 7,159 unaffected individuals from Germany and Finland (11). Regarding the presence or direction of the relationship with SNP rs37369, those results are at variance with our findings. On the other hand, the results related to the CAAA haplotype are in partial agreement with our finding regarding the CTA haplotype.
Hepatocyte nuclear factor 4a (HNF4a) directly binds to the AGXT2 promoter and acts as a major regulator of AGXT2 expression (25). Severe inborn HNF4a deficiency is rare and leads to maturityonset diabetes of the young (26), and the mild HNF4A polymorphism is associated with an increased risk of type 2 DM (27,28). HNF4a deficiency and subsequent downregulation of AGXT2 leads to an increase in plasma ADMA and SDMA concentrations and may be related to the risk of DM. Given the positive association between the TT genotype of SNP rs37369 and serum concentrations of ADMA and SDMA (7), and the positive relationship between the CAAA haplotype and plasma ADMA concentrations (13), the TT genotype of SNP rs37369 and the CTA haplotype may be associated with the development of DM via higher serum concentrations of ADMA and/ or SDMA.  In the present study, no significant multiplicative interaction was observed between AGXT2 SNP rs37369 and smoking status with regard to risk of DM. In a previously cited caseecontrol study in China, under the recessive model, a significant positive association was found between the rs37369 GG genotype and coronary heart disease among smokers, but not among nonsmokers, yet the interaction was not assessed (17). Similarly, in another previously cited caseecontrol study in China, although the interaction was not examined, a significant positive association was found between the rs37369 GG genotype and chronic heart failure among smokers, but not among nonsmokers (12).
The present study had methodological advantages. The participants were homogeneous with respect to their residential area, and we controlled for several potential confounding factors.
Several limitations of our study should also be considered. In the baseline survey of the AICOS, the participation rate was low. Thus, our study participants were probably not representative of the Japanese general population. For example, the educational level of our participants was higher than that of the general population. According to a population census conducted in 2010 in Ehime prefecture (29), the proportions of persons aged 60 to 69 years with low, medium, high and unknown educational levels were 28.2%, 48.6%, 19.0% and 4.2%, respectively, in men, and 26.7%, 56.4%, 12.9% and 4.0%, respectively, in women. The corresponding figures in the present study for persons aged 60 to 69 years were 12.2%, 54.8%, 33.0% and 0.0%, respectively, in men, and 16.5%, 52.4%, 31.2% and 0.0%, respectively, in women. On the other hand, the distributions of all 3 SNPs under study were within the HardyeWeinberg equilibrium and any selection bias in relation to genotype distribution was negligible.
Because no attempt was made to ascertain outcome status through reviews of medical records, we could not distinguish type 1 from type 2 DM. Nondifferential outcome misclassification may have biased the magnitude of the observed associations toward the null. In addition, the number of cases was rather small for a valid genetic association study, although a significant association was detected. The lack of significant relationships between AGXT2 SNPs rs37370 or rs180749 and DM may be ascribed to insufficient statistical power. Residual confounding effects could not be ruled out.
In conclusion, in this caseecontrol study we demonstrated that the TT genotype of AGXT2 SNP rs37369 and the CTA haplotype of rs37370, rs37369 and rs180749 were significantly positively associated with the risk of DM, whereas the CCA haplotype was significantly inversely related to DM in Japanese adults. No significant interaction between AGXT2 SNP rs37369 and smoking status was observed with respect to the risk of DM. These findings should be confirmed by further epidemiologic studies with larger numbers of subjects, especially patients with DM, in addition to functional studies.