In this study, we conducted a bidirectional Mendelian randomization analysis to assess the relationships between NAFLD and CKD. Our results revealed that genetically predicted NAFLD was associated with a higher risk of CKD. However, in the reverse direction, there was no significant evidence supporting a causal association between CKD and NAFLD.
CKD, or chronic kidney disease, is a clinical condition that arises as a result of persistent and definitive alterations in kidney function and/or structure. It is characterized by its slow and progressive nature and is known for its irreversible nature (28). NAFLD represents a histopathological spectrum of metabolic liver disorders that includes various conditions. It encompasses the presence of simple steatosis alone, referred to as non-alcoholic fatty liver (NAFL), characterized by the accumulation of fat in the liver. Additionally, it involves steatosis accompanied by hepatocyte inflammation and ballooning, with or without liver fibrosis, known as non-alcoholic steatohepatitis (NASH). In more advanced cases, NAFLD can progress to cirrhosis, which signifies the severe stage of liver fibrosis (29).
In recent years, multiple epidemiological studies and meta-analyses have consistently demonstrated that NAFLD, identified through a range of methods including blood biomarkers, imaging techniques, International Classification of Diseases codes, and liver biopsy, is linked to an elevated risk of incident CKD (30, 31). Importantly, this association persists even after controlling for established CKD risk factors, diabetes-related variables, and other potential confounding factors (32–34). Recent experimental and clinical research suggests that NAFLD/MAFLD and its advanced forms can worsen systemic insulin resistance, partly due to the secretion of hepatokines like fibroblast growth factor-21 (35). Additionally, NAFLD/MAFLD is linked to atherogenic dyslipidemia and hypertension, as well as the activation of the renin-angiotensin system (RAS), which is associated with endothelial dysfunction (36, 37). Furthermore, various mediators released into the bloodstream as a result of NAFLD/MAFLD can promote a chronic pro-inflammatory and pro-coagulant state (38). All these factors may play a role in the pathophysiology of CKD. Besides, numerous hospital-based and community-based studies have consistently shown a significant association between NAFLD, as determined by imaging techniques or liver biopsy, and an elevated prevalence of CKD (defined as eGFR < 60 ml/min/1.73 m2, abnormal albuminuria or overt proteinuria) (39–41). However, the presence of common risk factors for both CKD and NAFLD, such as hypertension, obesity, dyslipidemia, and insulin resistance, poses challenges in establishing a causal relationship between NAFLD and CKD, particularly when accounting for the overlapping "hepatorenal" and "cardiorenal" features (42).
Inferring the causal direction between correlated variables is a pervasive issue in biology that simple regression analysis cannot answer. Determining the causal direction between correlated variables is a common challenge in biology that cannot be adequately addressed by simple regression analysis. While an association between two variables may exist, establishing the direction of causality is often unclear. Mendelian randomization offers a unique advantage in addressing this issue as many traits of interest are influenced by genetic factors, and these genetic effects can serve as robust instrumental variables for various reasons (43).
However, the genetic effect may not be particularly large, resulting in a weak instrument and the requirement for very large sample sizes (44). In general, employing fewer degrees of freedom to model the genetic association (F > 10), that is using parsimonious models, will increase the F-statistic and reduce weak instrument bias (23). This approach was adopted in our study in an attempt to minimize bias given the weak instrument.