Mendelian Randomization in Autoimmune Disease Research

Abstract

Autoimmune diseases (ADs) constitute a diverse array of disorders marked by the disruption of immune tolerance and the aberrant activation of the immune system. Recent research has revealed a progressive annual increase in both the prevalence and incidence of ADs, impacting 5%–10% of the global population [1]. Drawing on Mendelian inheritance laws and instrumental variable estimation techniques, Mendelian Randomization (MR) harnesses genetic variations linked to particular exposures to investigate the causal impacts of modifiable exposures—such as potential risk factors—on health, society, and economy. MR has been widely used in the medical field for a multitude of applications. For example, it can not only confirm and reveal the causal relationships between clinically relevant risk factors, behavioral traits, and environmental elements with associated diseases, but also be employed to simulate drug targets and investigate genetic susceptibilities [2]. As current research struggles to pinpoint the causal risks of ADs, the emerging technique of MR aids in hypothesizing risk factors, pathogenesis, potential drug targets, and the potential causal relationships with other diseases, offering a fresh perspective on understanding these conditions.

Epidemiological research indicated that environmental influences are critical risk factors contributing to the breakdown of immune tolerance [3]. Wen et al. [4] conducted a MR study that identified a causal link between air pollutants and an increased risk of developing hypothyroidism, systemic lupus erythematosus (SLE), rheumatoid arthritis (RA), and ulcerative colitis (UC). Conversely, they found a decreased risk of coeliac disease associated with air pollution. Additionally, another MR study [5] supported the role of environmental pollution in AD, which revealed a causal association between PM2.5 and psoriasis in European populations. Research has established a correlation between Body Mass Index (BMI) and the risk of developing various ADs, with obesity increasing the risk of AD onset [6]. When exploring the relationship between diet and disease, existing observational studies on the association between folate and vitamin B12 with ADs present conflicting findings [6]. In a MR analysis conducted by Yang et al. [7] discovered that genetically elevated circulating folate levels are linked to a lower risk of vitiligo, suggesting that folate supplementation might serve as a preventative strategy against this condition. Moreover, Zhao et al. [8] reported similar findings, which showed that higher levels of 25-hydroxyvitamin are causally associated with a reduced risk of psoriasis and SLE. Additionally, attention must also be paid to the relationship between individual factors (such as mental health, education) and ADs. Emotional instability was considered a modifiable risk factor for hypothyroidism and SLE [9]. Furthermore, higher levels of education-related factors have a protective effect against ADs such as RA, UC, Crohn's disease (CD), and irritable bowel syndrome (IBS) [10]. In summary, MR can provide relatively convincing evidence for studying the potential causal relationships between relevant risk factors and ADs.

In recent years, the role of the gut microbiota in the pathogenesis of ADs has emerged as a focal point in medical research. While cross-sectional studies have indicated associations between dysbiosis of gut microbiota and the development of ADs, they have not yet established a causal link [11]. Xu et al. employed a two-sample MR analysis to investigate the potential causal relationships between the gut microbiome and six prevalent ADs, including SLE, RA, inflammatory bowel disease (IBD), multiple sclerosis (MS), type 1 diabetes (T1D), and coeliac disease. Their findings revealed a causal association between an increased relative abundance of the Bifidobacterium genus and the risk of developing T1D and celiac disease, offering novel perspectives on the mechanisms by which the gut microbiota may mediate the onset of ADs [11].

Moreover, MR is a valuable tool in drug treatment. Genetic variations in gene coding regions can influence target gene expression, acting like drug interventions on these targets [12]. MR integrates data on SNP gene expression and disease associations to establish causality between exposures and outcomes [13], which allows the use of genetic variations to identify new drug targets for diseases. For instance, researchers used MR analysis to integrate multi-omics data including DNA methylation, gene expression, and protein abundance, along with SMR and collocation analyses, to reveal potential targets such as TNFAIP3, BTN3A1, and PLAU for Sjögren's syndrome [14]. Cao et al. [13] identified seven potential drug targets for RA, suggesting that drugs targeting these genes may have a higher chance of success in clinical trials. Moreover, MR can provide evidence for emerging treatment strategies. An increasing amount of evidence highlights the interplay between lipid metabolism and immune regulation, yet the causal relationship between lipids and AD, as well as their potential as drug targets for AD, still lacks substantial evidence. Hu et al. [12] conducted a study on the association between lipid traits (such as cholesterol and triglycerides) and AD, and assessed the possibility of lipid-lowering drug targets for AD treatment. The results showed no evidence of a causal effect of these lipid traits and lipid-lowering drug targets on AD. However, a causal relationship was found between HMGCR-mediated LDL-C reduction and decreased HMGCR expression with a lower risk of RA, indicating that HMGCR could be a promising therapeutic target for RA. Xie et al. [15] also found through MR that lipid-lowering drug PCSK9 inhibition significantly reduced the risk of SLE but increased the risk of asthma and CD. This discovery offered a new perspective on the role of PCSK9 inhibitors in different diseases.

MR has emerged as a valuable causal inference tool, revealing potential causal links between ADs and a range of other health conditions, including respiratory system diseases [16, 17] (bronchiectasis, sinusitis), central nervous system diseases [18] (Alzheimer's disease), urinary system diseases [19] (prostate cancer), gynecological diseases [20, 21] (premature ovarian insufficiency, endometriosis), pain-related conditions [22, 23] (chronic pain in multiple sites, migraine), skin diseases [24] (psoriasis), eye diseases [25] (age-related macular degeneration), kidney tumors [26] and so on. These findings were crucial for clinical practice, highlighting the need for vigilant screening and management of potential risk factors and complications associated with ADs. For instance, SLE patients might require regular ovarian function assessments, while those with POI should be screened for CD [20].

MR is becoming increasingly popular in biomedical research to identify risk factors for various diseases. Unlike correlation analyses like differential expression analysis or weighted gene co-expression network analysis that only show associations, MR provides deeper insights into the causal relationship between gene expression and disease. Currently popular methods like machine learning and deep learning focus on the process of fitting predictive models to data or identifying informative groupings within data, with the interest being in the description and prediction of the data [27, 28]. Many articles pursue the combination of the two, screening feature genes with high predictive performance through machine learning algorithms, obtaining causal genes through MR, and finally confirming key biomarkers [29, 30], contributing to the value of prediction. As methodological and bioinformatics innovations progress, coupled with the development of computational tools and the availability of extensive Genome-Wide Association Studies (GWAS) datasets, the automation of MR analysis has become feasible, which has greatly facilitated the conduct of MR studies.

Although ADs are considered relatively rare, their incidence and mortality rates cannot be ignored. Our comprehension of ADs remains incomplete. With its strengths in elucidating causality, MR has risen as an innovative and indispensable tool for advancing our understanding of ADs. Relying on fixed genetic variations to minimize ascertainment bias and unmeasured confounders, MR has been employed to identify potential factors in disease progression, conduct drug trials, forecast the causal impacts of interventions on long-term clinical outcomes, and elucidate molecular mechanisms [2, 31]. Nevertheless, MR faces certain limitations, such as insufficient sample sizes in GWAS, the lack of diversity in study populations, and the potential for horizontal pleiotropy. Thus, there is a continued need to improve MR studies to ensure their robustness and reproducibility [2, 31]. It is anticipated that MR will further facilitate the translation of observational findings into clinical practice in future research, contributing to the prevention, diagnosis, and treatment of ADs.

Chen Li and Yuanhao Wu conceived and designed the study. Yunuo Wang and Hanjing Huang wrote the paper. Mengjiao Gu participated in the literature search. All authors approved the final manuscript. Yunuo Wang and Hanjing Huang contributed equally to this work.

The authors declare no conflicts of interest.