3.1 The causal effect of IMIDs on AA
A total of 179 independent SNPs associated with IMIDs were collected
following the established screening criteria for the MR relevance
hypothesis. In order to meet MR independence and exclusion assumption,
we conducted a literature review and found smoking(1), high blood
pressure(37), obesity(38), and alcohol consumption(39) as established
risk factors for AA. After excluding palindromic SNPs and SNPs
associated with these confounding variables, 154 independent SNPs
remained (10 SNPs for IBD on AA, 6 SNPs for CD on AA, 7 SNPs for UC on
AA, 77 SNPs for RA on AA, 39 SNPs for SLE on AA, 15 SNPs for PSO on AA).
Most of the F-statistics of these IVs are greater than 10, indicating a
low risk of weak instrument bias.
Complete information about the IVs
is summarized in Supplementary Table S3. In the MR analysis, the result
of IVW with fixed effects model indicated that no causal relationship
was found between IBD and AA (OR = 0.91, 95%CI = 0.81-1.01, p =
0.072), as well as CD and AA (OR = 0.97, 95%CI = 0.88-1.07, p =
0.544), UC and AA (OR = 0.95, 95%CI = 0.84-1.07, p = 0.363), RA
and AA (OR = 0.97, 95%CI = 0.92-1.02, p = 0.204), SLE and AA (OR
= 0.99, 95%CI = 0.96-1.02, p = 0.547), and PSO and AA (OR =
0.96, 95%CI = 0.91-1.03, p = 0.245). Meanwhile, the results of
other MR methods were also similar to IVW-FE mode (Figure 2). In MR
sensitivity analysis, p value of Cochran Q-test revealed no
significant heterogeneity was observed (all p >
0.05, Table 2). The intercept of MR-Egger regression showed there was no
intercept deviated from zero and all p values were greater than
0.05, explaining no impact of horizontal pleiotropy (Table 2). Indeed,
these results were also confirmed in the funnel plot with symmetrical
distribution (Supplementary Figure S2). Further, the leave-one-out
method indicated that no single SNP
differed apparently from the overall
odds ratio (Supplementary Figure
S1).