In the present large-scale and comprehensive MR study, we identified 21 immunoinflammatory factors, including 16 immune traits and 5 inflammatory proteins, having a potential causality in relation to the risk of AITD. Our findings indicated a suggestive genetic correlation between IL-1 alpha and AITD. Co-localization analysis excluded horizontal pleiotropy further supported the results. Furthermore, STAMBP and IL-10 were indicated to play a mediating role in 20.6% and 7.99%, respectively, of the causal impact associated with the CD33+ HLA DR+ CD14− absolute count on AITD.
Previous observational studies have investigated the immune cells-inflammatory proteins-thyroid cell interaction network in patients with AITD. They discovered that specific cytokines promote further differentiation of CD4 + T lymphocytes and induce activation of cytotoxic CD8+ T cells [28]. The latter damage thyroid cells, leading to excessive release of pro-inflammatory cytokines [29, 30]. This situation creates an amplifying feedback loop, initiating and sustaining the autoimmune process. However, these studies have not delved into specific subsets of immune cells, and the polymorphism in susceptible gene expression among AITD patients across different studies has made it even harder to figure out what causes AITD. The research indicated elevated levels of IL-1b, IL-6, TGF-b, TNF-a, CXCL10, and other factors, but their precise roles in the development of AITD remain undetermined.
In our study, HLA DR expression on CD33− HLA DR+ in the myeloid cell panel was most prominently associated with increased AITD risk. Notably, the expression of HLA-DR indicates T cell activation, and cells expressing HLA-DR can serve as antigen-presenting cells, playing a crucial role in CD4+ T cell-mediated autoimmune thyroiditis [31, 32]. CD4+ T lymphocytes can be categorized into different subsets based on cellular differentiation and phenotype, namely Th1, Th2, Th17, and regulatory T cells (Treg) [33, 34]. The skewed ratio of Th1/Th2 cells, excessive activation of Th17 cells, and abnormal suppressive function of Treg cells contribute to elucidating the pathogenesis of AITD. The effects of smoking status and vitamin D levels on thyroid function and immune regulation have been well-established, with potential implications in the progression of AITD. In multivariable MR analysis, after incorporating smoking status and 25-hydroxyvitamin D levels, some immunological traits showed non-significant effects on AITD, albeit with minimal changes in effect sizes. It is noteworthy that we found smoking can reduce the risk of AITD using the IVW method, consistent with prior research [35]; however, we did not observe an association between 25-hydroxyvitamin D levels and AITD risk, contrary to previous studies [36].
Our results added to the evidence that several specific inflammatory proteins were causally
associated with AITD. Inflammatory proteins were correlated, in order to detect true causal risk factors even in the presence of high correlation among candidate risk factors, we employed the MR-BMA method. Unfortunately, we did not identify the primary causal determinants of AITD risk. The balanced influence of these five inflammatory proteins reflects the complexity of their regulation in the immune process. We also additionally revealed a suggestive genetic correlation and a causal association indicating that higher levels of IL-1 alpha are associated with an increased risk of AITD. The IL-1 family member IL-1α is a pivotal proinflammatory cytokine with ubiquitous expression. It binds to IL-1R1, which it shares with IL-1β, and induces similar proinflammatory effects [37]. While the inflammatory properties of IL-1β are well-established in the field of AITD, the role of IL-1α has been relatively overlooked. This could be attributed to the rarity of IL-1α in the circulation of patients with inflammatory diseases [38].
We identified STAMBP and IL-10 as mediators in the causal link between the absolute count of CD33+ HLA DR+ CD14− cells and AITD. CD33 + HLA-DR + CD14- cells belong to the myeloid cell panel. IL-10 effectively inhibited the activation of myeloid cells, leading to reduced expression of pro-inflammatory cytokines, inflammatory enzymes, chemokines, and eosinophil chemotactic factors, thereby limiting the occurrence of the inflammatory response [39]. Previous studies have shown that in the monocyte cell line, knockout of the STAMBP gene increased IL-1β secretion and upregulated the expression of cytokine and chemokine genes subsequently driven by IL-1β [40]. While numerous studies have identified many new subtypes of immune cells and inflammatory proteins, the precise mechanisms by which immune cell subtypes affect inflammatory proteins remained unclear. Further research is needed to translate current research findings into clinical practice.
The strengths of our study lie in its large sample size of genetic summary data for immune traits, inflammatory proteins and AITD, and the verification of the results by complementary genetic methods makes it more reliable. Our research notably mitigated the influence of inevitable confounding factors, reverse causation, and other variables. This may pave the way for researchers to further explore the potential biological mechanisms of AITD and promote research on relevant therapeutic targets.
Limitations of our study should also be acknowledged. Firstly, while MR offers an essential alternative for verifying effects, it's important to note that MR effect estimates reflect lifelong genetic exposures. They may not precisely mirror the magnitude of shorter-term changes in immune cells. Nonetheless, understanding the anticipated causal direction of effects can offer insights into potential efficacy, which warrants further formal exploration through animal experiments and clinical trials. Secondly, invalid instruments are pleiotropic variants that influence the outcome through a pathway distinct from the one involving the exposure. As genetic variants originate from the host, instruments for the microbiome as an exposure are prone to pleiotropy. Generally, pleiotropy is more likely to bias estimates away from the null [41]. Consequently, our null results are generally more robust to horizontal pleiotropy. We mitigated weak instrument biases by exclusively incorporating exposures with genetic instruments possessing an average F-statistic exceeding 10. Additionally, we reduced the potential bias from pleiotropic variants by systematically conducting sensitivity analyses. Third, the study relied on European databases, which restricts the generalizability of our conclusions to other ethnic groups, thereby constraining the broader applicability of our findings.