3.1 Selection of IVs
Supplemental Tables S1-S3 displayed the results of the thorough examination of the IVs for LDL-C, HDL-C, and Lp(a) in the univariate MR study. Clustering was utilized to obtain SNPs for all lipid characteristics used in the MVMR analysis, and 333 SNPs were used in the multivariate MR analysis. The strength of the selected IVs were defined as stronger IVs(Table 2).
3.2 Univariable MR Analysis of Lipid Traits on AA Risks
Table 2 displayed univariate MR estimations of serum lipids on the risk of AA.
According to the IVW1 calculations, a 1-unit increase in genetically predicted LDL-C(OR:1.467, 95%CI:1.184-1.816, P=4.49E-4) and Lp(a) (OR:1.005, 95%CI:1.001-1.010, P=0.009) was responsible for 46.7% and 0.5% of the relative increase in AA risk, respectively. And a 27.2% relative reduction in the risk of AA was causally linked to a 1-unit rise in HDL-C(OR:0.728, 95% CI:0.629-0.842, P = 1.87E-05). The weighted median(WM) and Penalised weighted median(PWM) estimates showed that genetically predicted LDL-C(OR(WM): 1.051, 95% CI: 0.770-1.433, P = 7.55E-01) (OR(PWM): 1.000, 95% CI: 0.728-1.373, P = 9.98E-01) and HDL-C(OR(WM): 0.792, 95% CI: 0.628-1.000, P = 5.05E-02) (OR(PWM): 0.801, 95% CI: 0.634-1.013, P = 6.37E-02)were not significantly associated with the risk of AA. Table 3 displays the MR-PRESSO conclusions. MR-PRESSO obtained P=0.081 for AA, but no outliers were observed; hence, the IVW1 was more reliable.
There was a possibility of heterogeneity in the analysis results, but no evidence of directional pleiotropies (Table 2). The scatter plots and forest plots were depicted in Supplementary Figures S1 and S2, respectively. The Funnel plots were symmetric (Figure 2, 3 and Supplementary Figure S3). The leave-one-out technique revealed that no SNP contributed significantly to the link between lipids characteristics and AA risks (Supplementary Figure S4). The Anderson-Darling normality test showed that LDL-C was not normally distributed (P = 0.041), but the Shapiro Wilk normality test was predominant(Supplemental Table S4).
Table 2. Association between plasma LDL-C, HDL-C and Lp(a) levels and AA in Mendelian Randomization analysis (six different methods corresponding to the results), heterogeneity as well as horizontal multiplicity test analysis
Exposure
|
MRmethods
|
OR(95%CI)
|
PforAssociation
|
PforHeterogeneityTest
|
PforMREggerIntercept
|
F
|
Lp(a)
|
MR Egger
|
1.010(1.002-1.019)
|
0.038
|
0.623
|
0.22
|
159.70
|
|
IVW1
|
1.005(1.001-1.010)
|
0.009
|
0.544
|
|
|
|
WM
|
1.006(1.000-1.011)
|
0.039
|
|
|
|
|
ML
|
1.006(1.001-1.010)
|
0.009
|
|
|
|
|
PWM
|
1.006(1.000-1.011)
|
0.042
|
|
|
|
|
IVW2
|
1.005(1.001-1.010)
|
0.009
|
|
|
|
LDL-C
|
MR Egger
|
1.500(1.072-2.098)
|
1.90E-02
|
5.34E-06
|
0.863
|
50.40
|
|
IVW1
|
1.467(1.184-1.816)
|
4.49E-04
|
6.76E-06
|
|
|
|
WM
|
1.051(0.770-1.433)
|
7.55E-01
|
|
|
|
|
ML
|
1.468(1.237-1.742)
|
1.07E-05
|
|
|
|
|
PWM
|
1.000(0.728-1.373)
|
9.98E-01
|
|
|
|
|
IVW2
|
1.467(1.237-1.738)
|
1.02E-05
|
|
|
|
HDL-C
|
MR Egger
|
0.763(0.612-0.952)
|
1.73E-02
|
0.001
|
0.571
|
40.45
|
|
IVW1
|
0.728(0.629-0.842)
|
1.87E-05
|
0.001
|
|
|
|
WM
|
0.792(0.628-1.000)
|
5.05E-02
|
|
|
|
|
ML
|
0.727(0.638-0.829)
|
1.75E-06
|
|
|
|
|
PWM
|
0.801(0.634-1.013)
|
6.37E-02
|
|
|
|
|
IVW2
|
0.727(0.638-0.829)
|
1.65E-06
|
|
|
|
LDL-C: low-density lipoprotein cholesterol; HDL-C: high-density lipoprotein cholesterol; Lp(a): Lipoprotein(a); IVW1: random-effects inverse variance weighting; WM: Weighted median; ML: Maximum likelihood; IVW2: fixed-effects model inverse variance weighting;PWM: penalized weighted median;OR: odds ratio;CI: confidence interval; AA: Aortic aneurysm
Table 3. Association of plasma LDL-C, HDL-C and Lp(a) levels with AA in a Mendelian randomization analysis (MR-PRESSO)
Lipoprotein
|
Raw
N
|
OR
|
95%CI
|
Estimates
P
|
Outliter
N
|
Corrected
OR
|
95%CI
|
Estimates
P
|
HDL-C
|
319
|
0.723
|
0.627-0.833
|
1.08E-05
|
NA
|
NA
|
NA
|
NA
|
LDL-C
|
152
|
1.457
|
1.179-1.802
|
0.0007
|
151
|
1.489
|
1.212-1.820
|
0.0002
|
Lp(a)
|
12
|
1.005
|
1.000-1.009
|
0.081
|
NA
|
NA
|
NA
|
NA
|
LDL-C: low-density lipoprotein cholesterol; HDL-C: high-density lipoprotein cholesterol; Lp(a): Lipoprotein(a); OR: odds ratio;CI: confidence interval; AA: Aortic aneurysm
3.3 MVMR Analysis
The connection between LDL-C and Lp(a) and the risk of AA was not significant in the model with reciprocal adjustment for LDL-C, HDL-C, and Lp(a). In contrast, genetically predicted HDL-C was still associated with a lower risk of AA (Table 4).
Table 4. MVMR Analysis in the model with reciprocal adjustment for LDL-C, HDL-C, and Lp(a)
Exposures
|
Outcome
|
n.SNPs
|
beta
|
OR
|
95%CI
|
P
|
HDL-C
|
AA
|
333
|
-0.379
|
0.685
|
0.581-0.807
|
6.77E-06
|
LDL-C
|
AA
|
333
|
0.169
|
1.184
|
0.987-1.420
|
6.84E-02
|
Lp(a)
|
AA
|
333
|
0.005
|
1.005
|
0.999-1.011
|
8.43E-02
|
LDL-C: low-density lipoprotein cholesterol; HDL-C: high-density lipoprotein cholesterol; Lp(a): Lipoprotein(a); OR: odds ratio;CI: confidence interval; AA: Aortic aneurysm; SNPs: single nucleotide polymorphisms