3.2 The causal effect of AA on IMIDs
The initial screening identified 138 potentially independent SNPs that meet the inclusion criteria. Genetic variables associated with depression, anxiety, smoking, glucocorticoids and non-steroidal anti-inflammatory drugs use, obesity or malnutrition(12, 40, 41), as known to have causal associations with IMIDs, were removed. After harmonizing the exposure and outcome datasets, a total of 112 SNPs were obtained with most F-statistics greater than 10 (21 SNPs for AA on IBD, 21 SNPs for AA on CD, 21 SNPs for AA on UC, 15 SNPs for AA on RA, 13 SNPs for AA on SLE, 21 SNPs for AA on PSO, Supplementary Table S2). 112 independent SNPs associated with AA were selected as IVs to explore the effect of AA on IMIDs. As shown in Figure 2 and the scatter plot (Supplementary Figure S3), the IVW method indicated that patients with AA had a 6% increased risk of RA (OR = 1.06, 95%CI = 1.01-1.12,p = 0.029). MR-PRESSO method also presented unanimous evidence of causality between AA and RA (OR = 1.06, 95% CI = 1.01-1.11, p = 0.019). Cochran’s Q test showed there was no heterogeneity in IVs related to RA (Q = 11.82, p = 0.621). Moreover, both results of MR-PRESSO (no outlier was found, Table 1) and MR-Egger regression (intercept = -0.01, p = 0.570, Figure 2) provided no evidence of horizontal pleiotropy. However, the effect of genetically predicted AA on IBD, CD, UC, SLE, PSO did not reach statistical significance in either IVW or MR-PRESSO method (Figure 2 and Table 1). Sensitivity analysis of the above MR results suggested that there was no underlying heterogeneity and horizontal pleiotropy (all p value of Cochran’s Q and Egger-intercept > 0.05, Table 2).
Discussion
We conducted a two-sample bidirectional MR study to investigate the causal relationship between IMIDs (mainly including IBD, CD, UC, RA, SLE, and PSO) and AA. We found evidence of a causal effect of AA on RA, but not of RA or other IMIDs on AA. Although evidence from previous observational studies supported a causal relationship between some certain types of IMIDs and AA(12-20), causal inference cannot be made with great certainty in these studies for the possibility of residual confounding factors and various bias.
A retrospective case-control study with age- and gender-matched cohort from Israel, which adjust for different factors including socioeconomic and smoking, found that AA is more prevalent in patients with RA compared with controls (OR: 1.406, 95%CI: 1.094-1.789, p = 0.006). It is worth mentioning that the information regarding the treatment of RA patients and assessment of RA severity had not been formally investigated, as well as congenital diseases predisposing for AA(12). Contrary to above findings, a cohort study in England showed no evidence of a link between RA and AA, even after adjusting for age, gender, cardiovascular disease risk factors or anti-inflammatory drugs(13). Nonetheless, these studies lacked detailed information on daily tobacco use and alcohol consumption, family history of AA, as well as the imaging data, which all could bias causal inference towards a positive or negative direction. In addition, selection bias is also a problem that cannot be ignored. The same problems existed in studies of the association between SLE and AA(14, 15). People with SLE may experience a variety of symptoms and receive examination more frequently than people without SLE, therefore, they are more likely to be diagnosed with AA. Similarly, the two cohort studies on psoriasis failed to account for confounders such as body weight, smoking status, detail imaging findings, and assessment of psoriasis severity(16, 17). Moreover, the estimates from these observational studies are subject to substantial uncertainty with a wide confidence interval, indicating causal effects and their magnitude should be viewed with caution. Several isolated case reports have claimed that CD and UC were frequently observed comorbidities in AA patients(18-20), thus, we also conducted an investigation to explore the causal association of these diseases in AA individuals. However, either by forward or backward analysis, we did not fnd causal association between IBD, CD, UC and AA.
Genetic predisposition to AA is present from birth, therefore the MR results of our study would not be affected by the acquired drug treatment or other known or unknown exposure factors of AA patients. However, the exact mechanism by which genetically predisposed AA increases the risk of RA is still not completely understood. A study investigated the incidence of autoimmune diseases in patients with inflammatory aortic aneurysm (IAA). The serologic tests and biopsy findings were analyzed for the presence of autoimmune diseases before surgery in all cases. Autoimmune diseases were shown to be statistically significant in the inflammatory aortic group compared with the non-inflammatory aortic aneurysm group(21). Previous studies have provided evidence that a genetic predisposition to IAA exists and suggested that it may be related to human leukocyte antigen (HLA) molecule in IAA. HLA molecules play a critical role in the immune recognition process of specific antigen-presenting cells, which can be expressed by macrophages and dendritic cells. Amino acid polymorphism of HLA-DR molecule mediates immune responses to different antigens in different individuals and have also been found to be associated with the development of many diseases. A study investigating the genetic risk associated with HLA-DR gene polymorphism in patients with IAA revealed that HLA-DR B1*15 and HLA-DR B1*0404 allele were significantly enriched in these patients compared to control subjects(42). Moreover, observation with B-cell-enriched lymphoid follicles and CD3-positive T cells from the aneurysm wall of IAA patient suggested that a potential shared pathogenesis with IgG4-related diseases(43). The etiology of RA remains elusive, and several studies have consistently demonstrated its association with genetic factors. As early as 1976, Stastny(44) reported a correlation between RA and the HLA-DR4 allele of the human leukocyte antigen class II. Many studies have confirmed the association between RA and HLA-DR4 in a various of different ethnic groups, but the subtypes of HLA-DR4 and other types of HLA-DR vary widely among ethnic groups (45-48). Given the limited sample size and racial difference in previous studies, more large-scale research should focus on investigating the genetic contribution of HLA risk loci in IAA patients worldwide. Furthermore, the investigation of the genetic determinants and their role in the pathogenesis of RA necessitates further explore. Hopefully, the efficacy of autoantigen immunoregulation in treating RA has been demonstrated in animal studies. Clinical trials of autologous modified dendritic cells exposed to citrulline peptide by percutaneous injection are currently underway (49).
Unlike previous observational studies examining the association between IMIDs and AA, we used MR method to deal with known and unknown confounding factors such as smoking, alcohol or fallacies caused by reverse causation. Thus, our results overcome the limitations of observational studies to reflect causal associations and effects of IMIDs and AA. Our work also had some limitations. While we concluded that exposure to genetically predisposed AA increased the lifetime risk of RA, this does not mean that this conclusion can be used directly to infer the possible impact of interventions targeting genetically predisposed AA patients on RA risk. Besides, owing to constraints within the GWAS database, we did not distinguish the anatomic site of the aortic aneurysm. Although the thoracic aorta and the abdominal aorta share certain anatomical features, these two disease subtypes exhibit marked differences with respect to their pathophysiological mechanisms. Finally, our study was dependent on data from European population, which limits generalizability to other racial populations.
Conclusion
In summary, our study has showed genetically predicted AA may increase the risk of RA, while there was no evidence that RA had an increased risk of AA. Furthermore, we found no evidence of association between IBD, CD, UC, SLE, PSO and AA. This finding is in line with other research that demonstrated the genetic risk determinants are linked to HLA molecules in IAA patients. IAA may be a separate entity with similar pathogenesis with IgG4-related disease. Our study provides directions for future research on genetic susceptibility to IAA. The antigen binding role of IAA risk genes in the disease origin of RA needs additional investigation.