In our large case–control sample for CAD/MI, the incidence of established cardiovascular risk factors, such as male gender, type 2 diabetes, hypercholesterolemia, hypertension, and smoking, as well as increased BMI, was higher in CAD/MI cases (n = 1,460) as compared to controls (n = 1,215) (Table 1).
Association analysis of INSIG2 SNP rs7566605 with CAD/MI. Genotype distributions and allele frequencies of rs7566605 in the CAD/MI case–control cohort are shown in Table 2. There was no significant deviation from Hardy–Weinberg equilibrium in both CAD-free controls (P = 0.845) and CAD/MI cases (P = 0.236).
Table 2. rs7566605 association analysis results in CAD/MI case–control sample
No association of rs7566605 with CAD/MI was observed (P = 0.824; OR = 1.01; 95% confidence interval 0.90–1.14) (Table 2). Multivariate logistic regression analysis adjusting for cardiovascular risk factors (age at first manifestation, male gender, type 2 diabetes, hypercholesterolemia, hypertension, and smoking) did not significantly influence the result (P = 0.144). In addition to allele frequency and genotype distribution, we analyzed additive, dominant, and recessive genetic models with no substantial increase in significance (Table 2).
Separate analyses in men and women revealed similar ORs in allele frequency (Table 2). There was no significant deviation from Hardy–Weinberg equilibrium, both in male and female CAD-free controls (P = 0.702 and P = 1.000, respectively) as well as in male and female CAD/MI cases (P = 0.596 and P = 0.115, respectively).
Our power calculation for comparison of allele frequencies indicated that at a significance level of 0.05 and with a two-sided alternative hypothesis, our CAD/MI sample with n = 1,460 cases and n = 1,215 CAD-free controls (Table 1) had >97% power to detect a significant association with an assumed OR of 1.4 between the tested polymorphism (allele frequency 33%) and the phenotype MI. For smaller effect sizes (OR = 1.2), the power was still 60%. The number of individuals in the subgroups is limited under the hypothesis of a recessive model, proposed by Herbert et al. for the trait obesity (5). This working group estimated in a meta-analysis an OR of 1.22 [95% confidence interval from 1.05 to 1.42] for this SNP and obesity under a recessive model. Accordingly, we have estimated the power of our study on this SNP and obesity using G-Power in a Fisher's exact test (i) comparing allele frequencies and (ii) under a recessive model. To detect an effect with an OR of 1.22, we had a respective power of 47.1% and 26.6%, for an OR of 1.42 a respective power of 92.1% and 67.4%, and for an OR of 1.05 a respective power of 7.2% and 5.9%.
To assess the possibility that other markers located within or close to the INSIG2 gene could exert influence on susceptibility to CAD or MI, we analyzed the data from two recently conducted genome-wide association studies (18). Both from British (WTCCC; http:www.wtccc.org.ukinfosummary_stats.shtml) and German (Cardiogenics; http:www.cardiogenics.imbs-luebeck.de) case–control samples, P values were obtained for association with CAD/MI (WTCCC) and MI (Cardiogenics). A total of 38 SNP markers located in the expanded INSIG2 gene region (215.470 kb) were analyzed. None of the SNPs showed association at a P < 0.05 with CAD/MI both in the WTCCC and in Cardiogenics cohorts (data not shown).
Association analysis of INSIG2 SNP rs7566605 with cardiovascular risk factors. To assess the impact of rs7566605 on classical cardiovascular risk factors, we tested for association with male gender, type 2 diabetes, hypercholesterolemia, hypertension, and smoking. No significant association could be detected (Table 3).
Table 3. rs7566605 genotype distribution according to affection status on dichotomized cardiovascular risk factors in CAD/MI case–control background
Association analysis of INSIG2 SNP rs7566605 with BMI. Within the study population of MI patients, no association between the INSIG2 polymorphism rs7566605 and BMI could be observed in a sex-stratified analysis. The genotype frequencies in 305 female MI cases were 46.2% for GG carriers (mean BMI 27.0 ± 4.0 kg/m2), 40.7% for GC heterozygous individuals (mean BMI 27.7 ± 4.2 kg/m2), and 12.8% for CC homozygous carriers (mean BMI of 26.9 ± 4.2 kg/m2). Thus, no significant effect of the rs7566605 genotype on BMI could be observed (P = 0.244), and adjustment for age did not alter the result (P adjusted for age = 0.393) (Table 4).
Table 4. rs7566605 genotype distribution and BMI in MI cases
In the male MI patients, similar results were observed. The genotype frequencies in the 1,155 male patients were 44.6% for GG (mean BMI of 27.0 ± 3.1 kg/m2), 43.6% for GC (mean BMI of 27.3 ± 3.3 kg/m2), and 11.8% for CC (mean BMI of 27.4 ± 3.2 kg/m2) genotype, respectively. As in female MI patients, no significant association between rs7566605 genotype and BMI could be observed (P = 0.204) and did not change after adjustment for age (P = 0.300) (Table 4).
We also tested for association between the polymorphism and BMI as a continuous trait independent of MI in the whole case–control sample (n = 2,675). No significant results where observed, and adjustment for sex and age did not significantly alter the results. The mean BMI was nearly the same for all genotypes. Carriers of the GG genotype (n = 1,197) had a mean BMI of 27.0 ± 3.8 kg/m2, carriers of GC (n = 1,169) had a mean BMI of 26.9 ± 3.8, and those carrying CC (n = 308) had a mean BMI of 26.7 ± 3.8 kg/m2.
Family-based association analysis of INSIG2 SNP rs7566605 with BMI. To use the full power of all investigated families, a family-based association analysis was performed using the QFAM option implemented in PLINK to examine the effect of rs7566605 on BMI. Due to missing BMI data from family members, 98 families were excluded prior to analysis. Accordingly, testing was performed in 5,257 family members from 1,362 families. Family-based association analysis revealed no significant association of rs7566605 genotype with BMI (P = 0.5). In PLINK, the QFAM procedure allows family-based tests of association with quantitative phenotypes. PLINK performs a simple linear regression of phenotype on genotype but then uses a special permutation procedure to correct for family structure. The QFAM procedure was performed with 100,000 permutations (20).