Introduction
Non-alcoholic fatty liver disease (NAFLD) is a common disorder that is characterized by an excessive amount of fat stored in the liver of people who do not consume large amounts of alcohol or have other liver diseases [
1]. NAFLD has become the most common chronic liver disease worldwide, with an estimated prevalence rate of 30% among adults globally [
2]. A recent meta-analysis involving over 1.2 million people showed that the incidence rate of NAFLD was 46.13 per 1000 person years, with considerable disparities in gender, body mass index (BMI), geography and time-period [
3]. An increasing amount of evidence indicates that NAFLD is a multisystem disease, and its clinical and economic impacts are not limited to the progression of liver disease (nonalcoholic steatohepatitis, liver fibrosis, cirrhosis and hepatocellular carcinoma) but are also linked to an increased risk of numerous extrahepatic diseases [
4‐
7].
Gestational diabetes mellitus (GDM) is the most common medical complication in pregnant women, referring to any glucose intolerance that is identified or develops during pregnancy [
8,
9]. It has been observed that the prevalence of GDM varies greatly across the world, with some countries having a rate of 1% while others have a rate of over 30% [
9].
GDM put pregnant women and their newborns at risk in several ways, including higher chances of adverse pregnancy outcomes like hypertension, pre-eclampsia, preterm delivery, macrosomia, respiratory distress syndrome and neonatal jaundice. Furthermore, it can also have a lasting effect on both mother and child, including type 2 diabetes, metabolic syndrome, cardiovascular and cerebrovascular diseases [
9‐
11]. Nevertheless, the cause of GDM is yet to be determined.
In the last few years, the association between NAFLD and GDM has been a subject of great fascination for researchers. Previous observational studies showed that NAFLD was associated with GDM in this rural south Asian community [
12]. Several observational studies also indicated that GDM were at increased risk of developing NAFLD [
13,
14]. A recent cohort study of Korean adults showed that a history of GDM was an independent risk factor for the emergence of NAFLD. The correlation between GDM and NAFLD events was only explained by insulin resistance (IR) measured by homeostasis model assessment of insulin resistance (HOMA-IR) and the development of diabetes to a limited extent (10%) [
15]. However, correlation does not equate to causation; it merely reflects the statistical connection between two variables that can be measured [
16]. To date, the causal relationship between NAFLD and GDM has not yet been fully established.
By employing genetic variants as instrumental variables (IVs), Mendelian randomization (MR) is an analytical approach that can better ascertain the causality of exposure towards an outcome [
17]. Unlike observational studies, MR analyses are not influenced by common confounding factors like postnatal environment, socioeconomic status and behavioural factors, as alleles are randomly and independently segregated during meiosis [
17,
18]. Moreover, since genetic variations are fixed from birth and remain the same throughout life, MR can help prevent reverse causality bias [
19].
Hence, this study aims to investigate the causal relationship between NAFLD and GDM through a bidirectional MR analysis of two samples to examine whether NAFLD is a cause of GDM and if GDM is a risk factor for NAFLD. Given the high prevalence and burden of NAFLD and GDM, we believe that elucidating the causal relationship between NAFLD and GDM is of great clinical significance for the development of future prevention and treatment strategies for both conditions.
Discussion
To the best of our knowledge, the MR study is the first to explore the potential bidirectional causal association between NAFLD and GDM. Currently, there is no sufficient genetic evidence to suggest that NAFLD causes GDM, or that GDM causes NAFLD. The MR causal effect estimates were confirmed to be robust and reliable through multiple sensitivity analyses.
So far, the relationship between NAFLD and GDM has not been confirmed. Previous observational studies or meta-analyses have indicated a mutual relationship between them. In 2011, Forbes et al. [
30] conducted a cross-sectional study and showed that compared to European women without a history of GDM, ultrasound-diagnosed NAFLD was significantly more prevalent in those with a history of GDM. Nevertheless, the study did not consider pre-pregnancy metabolic risk factors and the sample size was inadequate. A multicentre prospective cohort study from pregnant Korean women showed that the risk of GDM was considerably higher in those with NAFLD and was linked to the severity of steatosis. This association between NAFLD and GDM kept its significance even after considering metabolic risk factors, including indicators of insulin resistance [
31]. A recent population-based prospective cohort study from rural south Asian community also showed that NAFLD may be a major risk factor for GDM [
12]. A recent meta-analysis suggested that NAFLD was associated with multiple pregnancy related diabetic complications [
32]. In addition, research suggested that GDM may also be linked to an increased risk of developing NAFLD. A meta-analysis of three cohort studies indicated a significantly higher risk of developing NAFLD after a GDM diagnosis (OR = 2.60, 95% CI:1.90–3.57, I
2 = 0%) [
14]. Similar results were observed in a recent cohort study from Korean [
15].
Despite the fact that our research evidence, based on MR analysis, failed to demonstrate a bidirectional causal relationship between NAFLD and GDM, several potential mechanisms could exist to explain the association between NAFLD and GDM. Firstly, NAFLD and GDM were two distinct metabolic illnesses, both of which could be linked to a shared metabolic abnormality such as IR [
33,
34]. IR served as a key factor in the relationship between the development of GDM and NAFLD, although a recent study [
15] indicated that its mediating impact on this connection was less than 10%. Secondly, the mechanism of how NAFLD leads to impaired glucose tolerance is likely due to a shared pro-inflammatory response of both adipokines and hepatokines [
35]. Low levels of adiponectin and high levels of selenoprotein-P were linked to the sonographic and biochemical severity of NAFLD, as well as being independent predictors of late pregnancy GDM. These inflammatory markers could potentially serve as valuable biomarkers in the future for gauging the severity of NAFLD and the likelihood of developing GDM, regardless of other metabolic factors [
31,
32]. Thirdly, women with a prior history of GDM, who have decreased insulin sensitivity and increased insulin secretion, may be more prone to developing NAFLD due to compensatory hyperinsulinemia, as insulin is known to stimulate hepatic lipogenesis. Insulin sensitivity impairment reduces the ability to suppress hepatic glucose production and insulin-stimulated glucose uptake in skeletal muscle, as well as increases fatty acid production from adipose tissue, ultimately resulting in a higher influx of fatty acids to the liver, consequently causing the emergence of NAFLD [
30,
36‐
38]. Additionally, women with a history of GDM may have lower levels of adiponectin or other adipocytokines, which could be a factor in the pathophysiological pathways connecting GDM and NAFLD [
39]. To sum up, the detailed mechanisms underlying the relationship between NAFLD and GDM are still not completely understood and require additional study in the future.
Considering that NAFLD and GDM may shared common risk factors, in order to explore whether there were intermediate factors influencing the causal relationship between NAFLD and GDM, we further analyzed the effects of obesity traits, lipid traits, HOMA-B, HOMA-IR, and sedentary behaviour on NAFLD or GDM. The MR analyses demonstrated that genetically determined BMI, WHR, TG, and time spent watching TV may be positively associated with both NAFLD and GDM, while HDL-C and Apo A-1 may have a negative association with both NAFLD and GDM. Considering that the previous observational studies did not fully adjust these possible influencing factors, these possible influencing factors may be one of the reasons why previous observational studies have found a correlation between NAFLD and GDM.
While our study concluded that there was no significant causal relationship between NAFLD and GDM based on the MR analysis, it is essential to acknowledge the limitations associated with null findings. Null results may be indicative of insufficient statistical power, genetic instrument validity issues, or complexities in the relationship between the exposures and outcomes, rather than definitive evidence of absence of causality. Given that no single source of evidence can establish a conclusive causal relationship, it is essential to interpret MR research results based on comprehensive evidence. This involves integrating the findings of various studies with different study designs to arrive at more reliable conclusions for a specific causal relationship issue [
40].
Strengths and limitations
Our MR study has several strengths. Firstly, as mentioned earlier in the introduction, in contrast to observational epidemiological studies, MR studies have a notable advantage in that they are less susceptible to bias from confounders or reverse causation. Secondly, the reliable evidence for the bidirectional relationship between NAFLD and GDM is supported by both the consistency of main effect estimation and sensitivity analysis of various methods. Thirdly, considering that the GWASs we relied on for our study only involved individuals of European ancestry and had genomic controls, it is improbable that our MR findings were influenced by population stratification.
Nevertheless, it is important to acknowledge several limitations that must be taken into account in this MR study. First, the GWAS dataset utilized in our MR analysis was obtained from European populations, which may limit the generalizability of the results to other ethnic or geographical populations. More studies are needed to determine whether a causal link exists between NAFLD and GDM in other regions. Second, since the data was limited, we were unable to stratify the analysis based on the severity of NAFLD. Third, despite using different techniques to remove outliers and variants with pleiotropy, and the horizontal pleiotropy test not indicating any problems, we cannot be certain that our results were not affected by unidentified pleiotropic variants.
Future directions
Given the complexity of NAFLD and GDM etiology, our study could provide insights into future research directions, such as investigating gene-environment interactions or conducting larger prospective longitudinal studies to assess the temporal relationship between NAFLD and GDM. Furthermore, our research primarily focuses on the GWAS data of the European population. Once GWAS data from other populations becomes available, future studies should investigate the causal link between NAFLD and GDM in non-European populations. Additionally, if there is available data in the future, it is crucial to explore the causal relationship between the severity of NAFLD (such as non-alcoholic steatohepatitis, liver fibrosis, etc.) and GDM.
In conclusion, our current bidirectional MR study failed to provide sufficient genetic evidence for the causal relationship between NAFLD and GDM. Further updated MR analysis is necessary to confirm our findings once a comprehensive and more detailed GWAS database of NAFLD and GDM patients becomes accessible.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.