The purpose of this study is to determine whether the association between IBD and COPD is causal. The current MR analyses suggest that COPD has no causal effects on IBD and its subtypes, but the reverse MR analysis provided genetic evidence that IBD and two subtypes are negatively correlated with the risk of COPD, and excluded smoking as a confounding factor.
There are several studies on the association between COPD and IBD and its subtypes, but the results are not consistent with MR analysis(18–20). Based on a meta-analysis of four observational studies, COPD and IBD have a strong association, and a significant increase in the number of newly diagnosed IBDs in the COPD population(14, 21, 22). Paul Brassard el assessed the incidences of CD and UC among patients with asthma and COPD based on the data from the Canadian administrative health databases for the 2001–2006 period. According to the study, COPD patients were more likely to develop CD than the general population by 55%, and UC was 30%(13). In a retrospective observational study of 87506 IBD patients, the obstructive respiratory disease was the most common among IBD patients, with COPD accounting for 8.75% of patients, and CD patients had an approximately 30% higher risk of developing COPD than UC patients(23).
For these observational and retrospective studies, there are some confounding variables that bias the results. Smoking is an important risk factor for COPD and IBD, and smokers with COPD have an increased risk of CD(24). Studies have shown that the risk of CD in COPD patients is 2.72 times that of smokers, and the risk of CD in smokers is 2 times that of non-smokers(25, 26). However, the data for several studies about smoking were not detailed and were not analyzed as a major confounding factor. Therefore, the influence of smoking on the increased risk of COPD and IBD cannot be excluded. MR analysis was used in this study. The causal relationship between COPD and IBD was confirmed from the perspective of genetics. And the SNPs related to smoking among the SNPs related to exposure and outcome were excluded, and the sensitivity analysis confirmed that the results were more stable and reliable in the absence of heterogeneity and pleiotropy.
In addition, drug treatment can also affect the relationship between IBD and COPD. The main effects of stable COPD treatment drugs are to relax the bronchi, inhibit the secretion of inflammatory media in the body(27, 28), reduce sputum secretion, and improve the patient's cough, expectoration, and other related symptoms(29, 30). Therefore, it maybe has little effect on IBD inflammation. The therapeutic drugs for IBD mainly include aminosalicylic acid, glucocorticoids, immunosuppressive drugs, and so on(31–33). There are relatively many antibiotic treatments for IBD, which mainly play an anti-inflammatory and immunomodulatory role(34). The occurrence of COPD is caused by the release of inflammatory factors, and anti-inflammatory drugs have a good therapeutic effect on it. Therefore, we speculate that drugs related to IBD treatment can inhibit the occurrence and development of COPD.
Using a two-sample bidirectional MR method, we evaluated the causal effect of IBD overall and two major subtypes (UC and CD) on COPD risk. Compared with the randomized controlled experiment, the MR analysis method can exclude confounding factors such as smoking, and avoid the influence of reverse causality and detection error. At the same time, this result has been strengthened by conducting a sensitivity analysis to ensure that it is consistent with previous results.
There are some limitations to the study. First, the samples included in the study were all European populations. The findings may not be generalizable to other populations as a result. There are some differences in the incidence rate of IBD and COPD between Asian and European populations, and studies of the causal relationship between IBD and COPD in other populations are warranted. Second, the limited number of SNPs in COPD, a low threshold was chosen for MR analysis (P < 5×10− 6), which may introduce weak instrument bias. However, the F statistic of each SNP included in the study is greater than 10, which reduces the possibility of weak tool bias.