Case-control study combined with machine learning techniques to identify key genetic variations in GSK3B that affect susceptibility to diabetic kidney diseases.

Abstract

The role of genetic susceptibility in early warning and precise treatment of diabetic kidney disease (DKD) requires further investigation. A case-control study was conducted to evaluate the predictive effect of GSK3B genetic polymorphisms on the susceptibility to DKD, with the aim of providing a theoretical basis and laboratory rationale for the prediction of the risk of developing DKD in patients with type 2 diabetes mellitus (T2DM). The GSK3B genotyping was performed by SNaPshot method based on Genotype-Tissue Expression database and thousand genomes database to screen tag SNPs. The polymorphisms of GSK3B tag SNPs were statistically analyzed for their effects on DKD susceptibility and clinical indicators. Urinary exosomes from DKD patients were extracted, protein expression levels of GSK3β were detected by ELISA kits, and kinase activity of GSK3β was quantified by kinase activity spectrometry to evaluate the correlation between the gene polymorphisms of GSK3B and the expression levels and activities of GSK3β. A machine learning model was constructed for assessing the efficacy of GSK3B polymorphisms in predicting the risk of developing DKD in patients with T2DM. A total of 800 subjects who met the inclusion and exclusion criteria were included in the case-control study, including 200 healthy control subjects, 300 patients with T2DM and 300 patients with DKD. Genetic analysis identified five tag SNPs (rs60393216, rs3732361, rs2199503, rs1488766, and rs59669360) associated with the susceptibility to DKD. The protein level and activity of GSK3β were significantly elevated in DKD patients. On the other hand, the expression levels and kinase activity of GSK3β in exosomes differed significantly between patients with different genotypes of the GSK3B, suggesting that the effect of GSK3B gene polymorphisms on GSK3β expression and activity may be an important mechanism leading to individual differences in susceptibility to DKD. XG Boost algorithm model identified rs60393216 and rs1488766 as important biomarkers for clinical early warning of DKD.