Causal associations between liver enzymes and diabetic microvascular complications: A univariable and multivariable Mendelian randomization

Abstract

Background: Observational studies show that liver enzymes are diabetes risk factors. However, previous observational investigations on the relationship between liver enzymes and diabetic microvascular complications produced contradictory results. The purpose of this research is to examine the independent causal effects of liver enzymes on diabetic microvascular complications.

Methods: Univariable Mendelian randomization (UVMR) and multivariable Mendelian randomization (MVMR) were utilized to disentangle the causal effects. The genome-wide association study (GWAS) summary-level statistics were collected from the UK biobank and the FinnGen consortium. Single nucleotide polymorphisms (SNPs) were selected as genetic instruments with genome-wide significance (p < 5 ×10-8). Five UVMR approaches, including inverse variance weighted (IVW), Bayesian weighted Mendelian randomization, MR-Pleiotropy Residual Sum and Outlier (MR-PRESSO), weighted median, and MR-Egger, and three MVMR approaches, including the extended versions of IVW, MR-Egger, and the Q-minimization methods, were performed to evaluate the causal effects. The robustness of the MR results was further confirmed using several sensitivity analyses.

Results: UVMR revealed that a genetically predisposed per standard deviation increase in serum alanine aminotransferase (ALT) level increased the risk of diabetic retinopathy (DR) in type 2 diabetes mellitus (T2DM) (IVW OR = 1.489, 95% CI = 1.206-1.772, p = 0.006). Likewise, serum aspartate aminotransferase (AST) levels showed similar results (IVW OR = 1.376, 95% CI = 1.115-1.638, p = 0.017). Furthermore, these effects were consistent after controlling for glycemia and blood pressure using MVMR analysis. Additionally, sensitivity analyses further strengthened the causality. However, no significant associations were found between alkaline phosphatase (ALP), gamma-glutamyl transferase (GGT), and diabetic microvascular complications.

Conclusions: Robust evidence was demonstrated for an independent causal effect of serum ALT or AST concentration on the risk of DR in T2DM. Further investigations are necessary to elucidate the potential biological mechanisms and confirm their clinical significance for early prevention and intervention.

Fig 1. Study work flow of the UVMR and MVMR analyses revealed the causal associations between liver enzymes and diabetic microvascular complications.

Fig 2. Summary of UVMR causal effects for serum ALT and AST levels on the risk of DR and DN.

(A) Forest plot showed the UVMR estimations of serum ALT levels on the risk of DR and DN. (B) Forest plot showed the UVMR estimations of serum AST levels on the risk of DR and DN.

Fig 3. Summary of MVMR causal effects for serum ALT and AST levels on the risk of DR in T2DM patients when controlling for risk factors.

(A) Forest plot showed the direct causal effect of serum ALT levels on the risk of DR in T2DM patients. (B) Forest plot showed the direct causal effect of serum AST levels on the risk of DR in T2DM patients.