The study population comprised 5835 unrelated Japanese individuals (3309 men, 2526 women) who either visited outpatient clinics or were admitted to one of the participating hospitals (Gifu Prefectural General Medical Center and Gifu Prefectural Tajimi Hospital in Gifu Prefecture, and Hirosaki University Hospital, Reimeikyo Rehabilitation Hospital and Hirosaki Stroke Center in Aomori Prefecture, Japan) between October 2002 and March 2008 due to various symptoms or for an annual health checkup, or were recruited to a population-based prospective cohort study of aging and age-related diseases in Nakanojo, Gunma Prefecture, Japan.

Estimated glomerular filtration rate (eGFR) was calculated with the use of the simplified prediction equation derived from a modified version of that described in the Modification of Diet in Renal Disease (MDRD) Study, as proposed by the Japanese Society of Nephrology (13): eGFR (ml min⁻¹ 1.73 m⁻²) = 194 × [age (years)] – 0.287 × [serum creatinine (mg/dl)] − 1.094 × 0.739 (if female). The National Kidney Foundation’s Kidney Disease Outcomes Quality Initiative guidelines recommend a diagnosis of CKD if the eGFR is <60 ml min⁻¹ 1.73 m⁻² (4). On the basis of this criterion, 1763 subjects (1076 men, 687 women) were diagnosed with CKD. The control subjects comprised 4072 individuals (2233 men, 1839 women) recruited from among community-dwelling healthy individuals or patients who visited outpatient clinics regularly for treatment of various common diseases, with an eGFR ≥60 ml min⁻¹ 1.73 m⁻². Subjects with CKD and controls thus either had or did not have conventional risk factors for CKD, including hypertension (systolic blood pressure of ≥140 mmHg, diastolic blood pressure of ≥90 mmHg, or taking antihypertensive medication), diabetes mellitus (fasting blood glucose of ≥6.93 mmol/l, hemoglobin A_1c of ≥6.5%, or taking antidiabetes medication), or hypercholesterolemia (serum total cholesterol of ≥5.72 mmol/l or taking lipid-lowering medication). On the basis of these criteria, 3434 and 2401 subjects were diagnosed with or without hypertension, respectively, 1710 and 4125 subjects were diagnosed with or without diabetes mellitus, respectively, and 1296 and 1987 subjects were diagnosed with or without hypertension and diabetes mellitus, respectively.

The study protocol complied with the Declaration of Helsinki and was approved by the Committees on the Ethics of Human Research of Mie University Graduate School of Medicine, Hirosaki University Graduate School of Medicine, Gifu International Institute of Biotechnology, Tokyo Metropolitan Institute of Gerontology, and participating hospitals. Written informed consent was obtained from each participant.

A total of 150 polymorphisms (data not shown) were selected by genome-wide association studies of ischemic stroke and myocardial infarction (P-value for allele frequency <1.0×10⁻⁷) with the use of the GeneChip Human Mapping 500K Array Set (Affymetrix, Santa Clara, CA, USA) (14). The relationship of these polymorphisms to CKD was not previously examined in our studies (9–12,15,16).

Venous blood (7 ml) was collected into tubes containing 50 mmol/l ethylenediamine-tetraacetic acid (disodium salt), and genomic DNA was isolated with a kit (Genomix; Talent, Trieste, Italy). Genotypes of the 150 polymorphisms were determined at G&G Science (Fukushima, Japan) by a method that combines polymerase chain reaction and sequence-specific oligonucleotide probes with suspension array technology (Luminex, Austin, TX, USA). Primers, probes and other conditions for the genotyping of polymorphisms significantly associated with CKD are shown in Table I. Detailed genotyping methodology was as described previously (17).

Quantitative data were compared between subjects with CKD and controls by the unpaired Student’s t-test. Categorical data were compared by the χ² test. Allele frequencies were estimated by the gene counting method, and the χ² test was used to identify departures from Hardy-Weinberg equilibrium. In the initial screen, the allele frequencies of each polymorphism were compared between subjects with CKD and controls by the χ² test. Polymorphisms with a P-value for allele frequency of <0.005 were further examined by multivariable logistic regression analysis with adjustment for covariates. Multivariable logistic regression analysis was thus performed with CKD as a dependent variable, and independent variables including age, gender (0, woman; 1, man), body mass index (BMI), smoking status (0, non-smoker; 1, smoker), history of hypertension, diabetes mellitus or hypercholesterolemia (0, no history; 1, positive history), and the genotype of each polymorphism. Subsequenty, the P-value, odds ratio and 95% confidence interval were calculated. Each genotype was assessed according to dominant, recessive and additive genetic models. Additive models included the additive 1 (heterozygotes vs. wild-type homozygotes) and additive 2 (variant homozygotes vs. wild-type homozygotes) models, which were analyzed simultaneously using a single statistical model. We also performed a stepwise forward selection procedure to examine the effects of the genotypes as well as of other covariates on CKD. Each genotype was examined according to a dominant or recessive model on the basis of statistical significance in the multivariable logistic regression analysis. The P-levels for inclusion in and exclusion from the model were 0.25 and 0.1, respectively. Given the multiple comparisons of genotypes with CKD, we adopted the criterion of a P-value of <0.005 for statistical significance of association. For other clinical background data, a P-value of <0.05 was considered statistically significant. Statistical significance was examined by two-sided tests performed with JMP version 6.0 and JMP Genomics version 3.2 software (SAS Institute, Cary, NC, USA).

The characteristics of the subjects with or without hypertension are shown in Table II. For individuals with hypertension, age, systolic blood pressure, serum concentrations of triglycerides and low density lipoprotein (LDL)-cholesterol, blood glycosylated hemoglobin content and the prevalence of diabetes mellitus were greater, whereas BMI, the percentage of smokers, diastolic blood pressure and serum concentration of high density lipoprotein (HDL)-cholesterol were lower, in subjects with CKD than in controls. For individuals without hypertension, age, the frequency of male subjects, serum concentration of triglycerides and the prevalence of diabetes mellitus were greater, whereas diastolic blood pressure and serum concentration of HDL-cholesterol were lower, in subjects with CKD than in controls.

Comparison of allele frequencies with the χ² test revealed that the C→T polymorphism of F10 (rs5962) was significantly (P=0.0014) associated with CKD in individuals without hypertension, while no polymorphism was significantly associated with CKD in individuals with hypertension (data not shown). Multivariable logistic regression analysis with adjustment for age, gender, BMI, smoking status and the prevalence of diabetes mellitus and hypercholesterolemia revealed that no polymorphism was significantly (P<0.005) associated with CKD in individuals without hypertension (data not shown).

A stepwise forward selection procedure was performed to examine the effects of genotypes for the polymorphism associated with CKD by the χ² test, as well as the effects of age, gender, BMI, smoking status and the prevalence of diabetes mellitus and hypercholesterolemia on CKD. For individuals without hypertension, age, BMI, gender and the F10 genotype (dominant model), in descending order of statistical significance, were significant (P<0.005) and independent determinants of CKD (data not shown).

The characteristics of the subjects with or without diabetes mellitus are shown in Table III. For individuals with diabetes mellitus, age, systolic blood pressure and the prevalence of hypertension were greater, whereas the percentage of smokers and serum concentration of HDL-cholesterol were lower in subjects with CKD than in controls. For individuals without diabetes mellitus, age, the frequency of male subjects, systolic blood pressure, serum concentrations of triglycerides and LDL-cholesterol, and the prevalence of hypertension were greater, whereas the percentage of smokers, diastolic blood pressure and serum concentration of HDL-cholesterol were lower in subjects with CKD than in controls.

Comparison of allele frequencies with the χ² test revealed that two or one polymorphisms were significantly (P<0.005) associated with CKD in individuals with or without diabetes mellitus, respectively (Table IV). Multivariable logistic regression analysis with adjustment for age, gender, BMI, smoking status and the prevalence of hypertension and hypercholesterolemia revealed that the A→G polymorphism of PLA2G3 (rs5753472, additive 2 model) and the C→T polymorphism of RUVBL2 (rs1062708, additive 2 model) were significantly (P<0.005) associated with CKD in individuals with diabetes mellitus, and that the A→G polymorphism of CDH4 (rs6142884, recessive model) was significantly associated with CKD in individuals without diabetes mellitus (Table V).

A stepwise forward selection procedure was performed to examine the effects of genotypes for the polymorphism associated with CKD by the χ² test, as well as the effects of age, gender, BMI, smoking status and the prevalence of hypertension and hypercholesterolemia on CKD (Table VI). For individuals with diabetes mellitus, age and hypertension were a significant (P<0.005) and independent determinant of CKD. For individuals without diabetes mellitus, age, hypertension, smoking, gender and the CDH4 genotype (recessive model), in descending order of statistical significance, were significant and independent determinants of CKD.

The characteristics of the subjects with hypertension and diabetes mellitus or without these conditions are shown in Table VII. For individuals with hypertension and diabetes mellitus, age and systolic blood pressure were greater, whereas BMI, the percentage of smokers and serum concentration of HDL-cholesterol were lower in subjects with CKD than in controls. For individuals without hypertension or diabetes mellitus, age, the frequency of male subjects and serum concentration of triglycerides were greater, whereas the serum concentration of HDL-cholesterol was lower in subjects with CKD than in controls.

Comparison of allele frequencies with the χ² test revealed that three or one polymorphisms were significantly (P<0.005) associated with CKD in individuals with hypertension and diabetes mellitus or without these conditions, respectively (Table VIII). Multivariable logistic regression analysis with adjustment for age, gender, BMI, smoking status and the prevalence of hypercholesterolemia revealed that the C→T polymorphism of RUVBL2 (rs1062708, additive 2 model) and the C→T polymorphism of PTPRN2 (rs1638021, dominant model) were significantly (P<0.005) associated with CKD in individuals with hypertension and diabetes mellitus (Table IX). No polymorphism was significantly associated with CKD in individuals without hypertension or diabetes mellitus.

A stepwise forward selection procedure was performed to examine the effects of genotypes for the polymorphism associated with CKD by the χ² test, as well as the effects of age, gender, BMI, smoking status and the prevalence of hypercholesterolemia on CKD (Table X). For individuals with hypertension and diabetes mellitus, age and PTPRN2 genotype (dominant model) were significant (P<0.005) and independent determinants of CKD. For individuals without hypertension or diabetes mellitus, age, BMI, gender and the JPH3 genotype (recessive model), in descending order of statistical significance, were significant (P<0.005) and independent determinants of CKD.

Finally, we examined whether the genotype distributions for the polymorphisms associated with CKD were in Hardy-Weinberg equilibrium. The genotype distributions for the A→G polymorphism of CDH4 (subjects with CKD, P=0.2050; controls, P=0.3414) and the C→T polymorphism of PTPRN2 (subjects with CKD, P=0.0658; controls, P=0.6959) were in Hardy-Weinberg equilibrium both in subjects with CKD and in controls.