The Stanford Asia-Pacific Program for Hypertension and Insulin Resistance (SAPPHIRe) was a collaborative study that was part of the Family Blood Pressure Program of the National Heart, Lung and Blood Institute of the National Institutes of Health and was designed to investigate the genetic determinants of hypertension and insulin resistance in the Chinese and Japanese individuals. The study collected sibling pairs who were either concordant or discordant for high blood pressure. Detailed descriptions of the study cohort were published in our previous work 20. Hypertension was defined as systolic blood pressure ≧140 mmHg or diastolic blood pressure ≧90 mmHg or use of medications for high blood pressure. Individuals with pre-existing chronic illness such as diabetes, cancer, or diseases of the heart, liver, or kidney were excluded. The SAPPHIRe cohort study consisted of 6 field centers at baseline, including 4 field centers in Taiwan, one in Hawaii and one in Stanford University (San Francisco Bay area). A total of 1,143 subjects of Han Chinese descent from 392 families were enrolled at baseline. The follow-up study was conducted in the 4 Taiwan field centers and 753 individuals from 276 families were followed-up through the entire study. Other reasons of lost to follow-up included death (0.21%), lost of contact (5.7%), and rejection (16.56%). The Institutional Review Board of Tri-Service General Hospital in Taiwan, the National Taiwan University Hospital Research Ethics Committee, the Institutional Review Board of Taipei Veterans General Hospital, and the Institutional Review Board of Taichung Veterans General Hospital approved this study. Written informed consent was obtained from each participant.

Blood pressure was measured using a mercury sphygmomanometer in 3 separate intervals at rest. Standardized interview-administered questionnaires were used to obtain information on demographic and lifestyle characteristics as previously described 21. In brief, each participant was asked about smoking status and was categorized as non-smokers, ex-smokers or current smokers. Those who no longer smoked, but had smoked before were categorized as ex-smokers. The monthly alcohol consumption, grams of ethanol per day, was calculated based on the amount of usual ingestion of beer (5% ethanol), wine (12.5% ethanol), sake (16% ethanol) and liquor (40% ethanol) for each subject by a face-to-face interview. Light, modest, and heavy drinking was defined as average daily ethanol consumption 0–5 g, 5–10 g, and more than 10 g respectively 21. Physical activity was assessed by recording the number of hours per day spent at each of five levels of activity 21.

To identify common tag SNPs, we selected SNPs from the HapMap CHB (Chinese Beijing) database (HapMap Genome Browser release #24) (http://www.hapmap.org) using the Tagger program implemented in Haploview version 4 (http://www.broad.mit.edu/mpg/haploview/) with a minor allele frequency threshold of 0.1 and r² of 0.8 2223. Three tag SNPs that captured 100% of SNPs with minor allele frequencies more than 10% with mean maximal r² = 0.962 were selected. The genotyping call rate was listed in Table 1. The concordance rate of this system based on 160 genotyping duplication was 99.38%.

HW P, Hardy –Weinberg P value; MAF, minor allele frequency; % Geno, genotyping call rate.

We additionally recruited 48 morbidly obese adults undergoing weight-loss surgery in Min-Sheng Hospital in Taiwan. Abdominal visceral adipose tissue was biopsied in a fasting state during surgery. Informed consent was obtained from each patient. The study was approved by the Institutional Review Board of Min-Sheng Hospital.

Adipose tissue was placed in liquid nitrogen immediately after resection and stored at −80°C until processed. Total RNA was extracted using REzol (Promega, Madison, WI) according to the manufacturer's instructions. Reverse transcription was performed using an RT kit (Promega, Madison, WI) with 1 μg of total RNA and 0.5 μg random hexamers in a final volume of 25 μl containing 200 U of Maloney murine leukemia virus reverse transcriptase, 20 nM dNTP, and 25 U of rRNasin for 1 h at 37°C. The reaction mixture was diluted to 100 μl with double-distilled water prior to PCR amplification.

A 5-μl sample of diluted cDNA was added to a mixture comprised of 10-μl 2x SYBR Master Mix Buffer to a final volume of 20 μL (Applied Biosystems, Foster City, CA). The pre-developed primers for ALDH2 were PPH17047A-200 (QUIAGEN, Hilden, Germany). The primer sequence for PPIA (cyclophilin A) were:5' AGG TCC CAA AGA CAG CAG AAA AT 3' (forward) and 5' GTG AAA GCA GGA ACC CTT ATA ACC 3'(reverse). Thermocycling was done by 10 min at 95°C followed by 40 cycles of 30 sec at 95°C, 1 min at 60°C and 1 min at 72°C. Real-time quantitative PCR was analyzed by the ABI PRISM 7000 Sequence Detection System (Applied Biosystems, Foster City, CA). The fluorescent signal from each PCR reaction was collected as a peak-normalized value plotted versus the cycle number. Reactions were characterized by comparing the threshold cycle (Ct) values. Relative gene expression in relation to cyclophilin A RNA was calculated using the formula: ΔCt = (Ct of cyclophilin A) – (Ct of ALDH2).

All data are expressed as mean values ± standard deviations (SD) unless otherwise specified. Tests for Hardy-Weinberg equilibrium were performed before marker-trait analysis. Pairwise LD measures D' and r² were estimated to assess linkage disequilibrium (LD) between the SNPs in the ALDH2 gene. The structure of the haplotype block was evaluated using the solid spine LD method implemented in the Haploview program 23. Cox proportional hazard models were used to assess associations between the time to each progression event and the SNPs in the ALDH2 gene. Hazard ratio (HR) for a genotype versus the reference genotype was estimated and 95% confidence intervals (CI) for these hazard ratios were also computed. We accounted for the correlations resulting from the sibship data in the modeling. To analyze the association of ALDH2 genetic variants with quantitative traits, tests for population stratification and total association were performed using the quantitative transmission disequilibrium test (QTDT) program (http://www.sph.umich.edu/csg/abecasis/QTDT), which is based on a variance-components framework 24. A test for total association was performed if there was no significant population stratification. The parameter estimates associated with each additional minor allele were estimated with generalized estimating equation. P-values were adjusted for age, sex, smoking and physical activity, and alcohol consumption. The study-wide significance threshold required for type 1 error less than 5% is estimated using the method proposed by Nyholt et al. which takes into account the LD between markers 25. The population attributable fraction was calculated as follows: 1-{1÷[p² HR_homo + 2p(1-p)HR_hetero + (1-p)²}, where p is the risk-allele frequency, HR_homo is the hazard ratio for homozygotes, and HR_hetero is the hazard ratio for heterozygotes. The association between the ALDH2 genotype and ALDH2 gene expression in adipose tissue was analyzed using non-parametric trend test implemented in STATA 9.0 (StataCorp LP, College Station, Texas).

The basic clinical characteristics of participants at baseline are shown in Table 2. The nucleic acid composition, Hardy-Weinberg equilibrium test, genotyping call rate, and minor allele frequencies of selected SNPs are summarized in Table 1. The LD structure between SNPs is shown in Figure 1. All SNPs were within the same LD block. The allele frequencies of SNPs were similar to those in the HapMap CHB database 22.

Data are presented as mean ± S.D. BMI, body mass index; SBP, systolic blood pressure; DBP, diastolic blood pressure.

Over an average follow-up period of 5.7 year, participants homozygous for the rs2238152 T allele (minor allele) had a higher risk of progression from non-hypertension to hypertension (HR = 2.88; 95% CI: 1.06-7.84; P = 0.03) than did non-carriers (Table 3). The genetic model was best fit with a recessive genetic mode and the corresponding population attributable risk fraction is ~7.1%. Participants homozygous for the mutant rs671 A allele had only a trend of increased hypertension incidence (HR = 2.13; 95% CI: 0.77-5.89; P = 0.15) compared to non-carriers. However, rs2238152 T allele was not significantly associated with changes of systolic/diastolic blood pressure (SBP/DBP) during follow-up. In contrast, the mutant rs671 A allele was significantly associated with increases of SBP (0.865 mmHg increase per year, P = 0.0071) and DBP (0.537 mmHg increase per year, P = 0.0026) during follow-up (Table 4).

A: major allele; a: minor allele.

Estimate: the changes in systolic/diastolic blood pressure associated with each additonal risk allele during follow-up.

* adjusted for age, sex, alcohol intake, smoking, drugs, and physical activity. Corrections of blood pressure for antihypertensive drugs (10/5 mmHg reduction per drug) was made.

We next examined whether there was interaction between the ALDH2 SNP and environmental factors on the risk of hypertension. The rs2238152 T allele was associated with higher incidence of hypertension in participants with moderate or heavy alcohol intake (incidence of hypertension: 4.32, 5.36, and 17.0 cases/person year for rs2238152 TT, TG, and GG genotype, respectively) (Figure 2). However, the genetic effect disappeared in participant who did not drink (incidence of hypertension: 3.94, 2.45, and 1.8 cases/person year for rs2238152 TT, TG, and GG genotype, respectively), indicating an interaction between SNPs and alcohol intake on the risk of hypertension (P for interaction = 0.04) (Figure 2). We did not detect significant interaction between ALDH2 SNPs and other environmental factors including age, sex, smoking or physical activity on hypertension risk (data not shown).

We further explored possible mechanism by which SNP rs2238152 affect clinical phenotypes. We measured ALDH2 mRNA levels in human abdominal adipose tissue sampled from 48 adults. The rs2238152 T allele was associated with a significantly lower ALDH2 gene expression level (P = 0.03) (Figure 3). Other SNPs were not associated with ALDH2 gene expression level (Additional file 1: Figure S1).