We carried out a two-sample MR analysis to scrutinize the cause-and-effect relationship of generalized ADs on multiple types of pregnancy loss, which suggested that T1D, autoimmune hypothyroidism and CD are the risk factors of pregnancy loss. Surprisingly, RA was negatively correlated with spontaneous abortion. Sensitivity analyses confirmed that the above findings were credible.
The relationship between ADs and pregnancy loss is complex and multifactorial. Some ADs may cause pregnancy loss directly by affecting the placenta or the fetus, while others may cause pregnancy loss indirectly by increasing the risk of other complications such as infection, bleeding or hypertension. Previous studies of the mechanism have shown that T1D might increase cellular stress by upregulating stress responsive proteins, which may increase the likelihood of pregnancy complications(14). Autoantibodies may be involved in miscarriage due to hypothyroidism(15, 16). CD induced pregnancy loss depends on the degree of disease activity, but the specific mechanism needs further study(17). It is worth noting that pregnancy loss is intertwined with obesity, body mass index (BMI), smoking and excessive alcohol consumption. To ensure the soundness of the hypothesis, MR analysis excluded SNPs that were linked to the confounding factors mentioned above.
Our work suggests that ADs may influence the occurrence of pregnancy loss at a genetic level. A recent review shows that immune-related genes mediate recurrent pregnancy loss (PRL) through immune imbalance(18). Among them, B cells, NK cells, human leukocyte antigen (HLA) related genes and genetic polymorphisms in interleukin genes together constitute the main mechanism of immune imbalance causing PRL. We visualized the main process of regulating PRL(Figure 5).
Some observational studies have focused on the correlation between autoimmune diseases (ADs) and pregnancy loss, but the results are contentious and no definitive conclusion has been reached.
Type 1 diabetes
Preconception diabetes mellitus is a well-known predisposing factor for pregnancy loss, but the role of T1D in pregnancy loss has been a subject of debate. A handful of research has revealed that T2D has an elevated risk of miscarriage than T1D(19), and that T1D is not associated with a substantial elevation in the risk of first-trimester miscarriage in T1D(20). However, these retrospective research was carried out at individual facilities and had comparatively limited numbers of participants, which may have led to biased conclusions. Our findings are consistent with those of a systematic review, which suggested that miscarriages during T1D were more common than T2D and gestational diabetes (GDM)(21).
Autoimmune hypothyroidism
Several observational studies have reported results that contradict our conclusions. In cases of severe hypothyroidism, for instance, an Israeli retrospective study found no significantly higher likelihood of miscarriage compared to normal controls(22). A similar conclusion was drawn from a single-center study conducted in Japan(23). However, these studies may be impacted by limitations such as small sample sizes, ethnic homogeneity, and inadequate information, which can lead to various biases and less accurate inferences. In contrast, our positive causality is consistent with several previous systematic reviews and meta-analyses, which have concluded that pregnancy loss is more likely to occur amongst pregnant women suffering from hypothyroidism(24, 25).
Crohn’s disease
Existing research has indicated that active CD is connected with spontaneous abortion(26), while another study has shown that recurrent CD during pregnancy is not associated with spontaneous abortion(27). Presently, clinical investigations on CD and pregnancy primarily concentrate on assessing medication safety(28). However, there remains a dearth of extensive, multi-center prospective studies on the relationship between CD and pregnancy loss. Our work shows that CD is likely to be the cause of spontaneous abortion when confounding factors and reverse causation are excluded, providing some evidence for further research on mechanism and prevention.
Rheumatoid arthritis
Multiple studies have shown that RA increases the incidence of spontaneous abortion(29, 30). Other studies, however, have come to a different conclusion. A case-control study showed no increase in spontaneous abortion in women diagnosed with RA(7). Another investigation that included 40 women suffering from RA also showed no significant difference in rates of spontaneous abortion between the population of patients with RA and the control group(31). To put it briefly, based on our findings, there was no evidence to suggest a positive causal relationship between RA and spontaneous abortion. These results may inform clinicians' decisions when managing pregnant women with RA. It is essential to point out that our conclusions are limited to the genetic level and may not necessarily reflect real-world observations. For instance, patients with RA may have a higher use of non-steroidal anti-inflammatory drugs (NSAIDs), which can boost the prevalence of spontaneous pregnancy loss(32, 33) and lead to false-positive results in observational studies.
Currently, the bulk of research on the association between ADs and pregnancy loss has mainly focused on the effects of APS(34) and SLE(35), which have led to the strengthening of treatment and comprehensive management for pregnant women with these ADs. However, observational studies have inherent limitations, and the relationship between other types of ADs and pregnancy loss remains unclear(36). This knowledge gap can affect the clinical evidence level and pose challenges for clinical guideline developers, ultimately impacting the decision-making of obstetricians and nurses and potentially affecting pregnancy outcomes.
MR studies have an evidence level second only to RCTs and offer several advantages over traditional observational studies(36), including decreased susceptibility to bias and confounding, a clear temporal relationship between exposure and outcome, and the ability to investigate the effects of long-term exposures. In certain cases, where RCTs on ADs and pregnancy loss are not feasible due to ethical and practical considerations, MR studies can offer valuable understanding of the cause-and-effect relationship between ADs and pregnancy loss, prioritizing clinical research on ADs with positive results. Additionally, by using genetic variants as IVs, MR studies may potentially save time, money, and manpower while identifying new targets for intervention.
Our work has shown that Type 1 diabetes, autoimmune hypothyroidism, and Crohn's disease are all risk factors for pregnancy loss. These findings highlight the need for early intervention and management of these ADs in pregnant women. We can adjust and regulate patients from two angles. First, we can counsel women with these ADs on the importance of controlling their disease before attempting to conceive, from a preconception standpoint. Second, obstetricians and nurses can develop more scientific and comprehensive plans that take into account the specific AD and the severity of the disease during pregnancy. The ultimate goal is to minimize the potential of pregnancy loss and improve maternal and fetal outcomes in women with ADs.
The use of a two-sample MR method is among the distinguishing strengths of our research, which enabled us to investigate the possible cause-and-effect association using extensive GWAS data with substantial sample sizes. Moreover, since SNPs are randomly assigned at conception, this design substantially reduces biases that may arise from potential confounding factors. Notably, we carried out an extensive investigation of the relationship between ADs from various systems and different types of pregnancy loss, which, to our knowledge, has not been previously undertaken. Our findings provide important insights for improving clinical management of maternal pregnancy outcomes in the context of ADs.
Our research has several limitations that are paramount to acknowledge. First, the data used in our research only included participants of European ancestry. By employing a homogenous set of participants, we were able to decrease the impact of demographic confounding bias and improve the precision of our MR findings. However, further validation of the transferability of our results to other populations is necessary through investigations involving individuals from non-European ancestries. Secondly, the outcome data consisted of multiple types of pregnancy loss, which failed to delineate the underlying causes of pregnancy loss and thereby clarify the association between ADs and the direct causes of pregnancy loss. Consequently, additional research is needed to investigate the connection between pregnancy loss and mediating factors (such as metabolites and inflammatory factors, etc.) related to ADs.