Table 2. Estimated strength of SNP IVs and susceptibility to weak instrument bias using the two-sample Mendelian randomization approach for estimating the association between lipids and AAA risk.
| Exposure | # of SNP IVs | SNP-lipid summary estimate source | r2* | F† | Beta‡ (P-value) | Sample overlap | Bias under null† | Type I error rate† |
|---|---|---|---|---|---|---|---|---|
| LDL cholesterol | 78 | GLGC | 0.071 | 184.7 | 4.4% | 0.0001 | 0.050 | |
| ARIC | 0.087 | 10.7 | 0.2391 (<0.001) | 100% | 0.0223 | 0.052 | ||
| HDL cholesterol | 85 | GLGC | 0.050 | 116.7 | 4.4% | <0.0001 | 0.050 | |
| ARIC | 0.057 | 6.2 | -0.1224 (0.13) | 100% | -0.0197 | 0.051 | ||
| Triglycerides | 53 | GLGC | 0.044 | 163.7 | 4.4% | <0.0001 | 0.050 | |
| ARIC | 0.064 | 11.2 | 0.0618 (0.48) | 100% | 0.0055 | 0.050 | ||
| TC | 85 | GLGC | 0.066 | 156.7 | 4.4% | 0.0001 | 0.050 | |
| ARIC | 0.089 | 10.0 | 0.2372 (<0.001) | 100% | 0.0237 | 0.052 |
LDL cholesterol: low-density lipoprotein cholesterol; HDL cholesterol: high-density lipoprotein cholesterol; TC: total cholesterol
*Percentage of variance in the lipids explained by the SNPs
†The F statistic, bias under the null, and the type I error rate values were estimated using equations from Burgess et al.[28]
‡In ARIC, the log-odds ratio for AAA per 1 standard deviation (SD) increment in the lipid fraction using traditional logistic regression, adjusted for age, sex, study center, smoking status, height, white blood count, fibrinogen, hypertension, diabetes, peripheral arterial disease, and the other lipid fractions (for LDL cholesterol, HDL cholesterol, and triglycerides models only)
SDs from GLGC (LDL cholesterol: 38.7 mg/dL; HDL cholesterol: 15.5 mg/dL; triglycerides: 90.7 mg/dL; total cholesterol: 41.8 mg/dL)