Abstract: BACKGROUND: Diabetes mellitus and hypertrophic cardiomyopathy influence each other, with diabetic patients exhibiting a significantly increased risk of cardiovascular complications. However, the causal relationship and underlying molecular mechanisms between these conditions remain unclear. 
OBJECTIVE: To assess the causal effect of diabetes mellitus on hypertrophic cardiomyopathy using a Mendelian randomization approach combined with data from genome-wide association studies (GWAS) and Gene Expression Omnibus (GEO) databases, and to screen for key regulatory factors by bioinformatics analysis.
METHODS: From the latest GWAS database (a collaboration between the U.S. National Human Genome Research Institute and the European Bioinformatics Institute), we obtained 162 single nucleotide polymorphisms associated with diabetes, which were identified from 24 659 diabetic cases and 459 939 control participants, and harvested data on hypertrophic cardiomyopathy, which were from 507 cases of hypertrophic cardiomyopathy and 489 220 control participants. A two-sample Mendelian randomization analysis was then conducted. The inverse variance weighted method was utilized to estimate the causal relationship between diabetes and hypertrophic cardiomyopathy. The diabetes dataset GSE184050 and the hypertrophic cardiomyopathy dataset GSE160997 profiles were downloaded from the GEO database. Weighted gene co-expression network analysis (WGCNA) was employed to investigate the important modules and core genes associated with diabetes and hypertrophic cardiomyopathy. Gene set enrichment analysis (GSEA) was utilized to explore enriched terms and pathways related to core genes. Core genes were inputted into the comparative toxicogenomics database (CTD) website to identify diseases most relevant to core genes. Various algorithms were applied to explore the role of HSF1 in diabetes and hypertrophic cardiomyopathy.
RESULTS AND CONCLUSION: (1) We used diabetes as the exposure factor and extracted results from 160 single nucleotide polymorphisms. The inverse variance weighted method, MR-Egger, and weighted median regression methods were employed to estimate the causal relationship between genetically predicted diabetes and hypertrophic cardiomyopathy. The MR-Egger method demonstrated a significant causal association between diabetes and hypertrophic cardiomyopathy, with HSF1 identified as a core biomarker for the causal relationship between diabetes and hypertrophic cardiomyopathy (P=0.018 028 98). The weighted median and inverse variance weighted algorithms also provided consistent trends. Furthermore, validation through datasets from the GEO database confirmed HSF1 as a potential core biomarker influencing both diabetes and hypertrophic cardiomyopathy. HSF1 exhibited high expression in diabetes and low expression in hypertrophic cardiomyopathy. According to the analysis results, systematically removing individual single nucleotide polymorphisms and repeating Mendelian randomization analysis did not significantly alter the causal relationship between diabetes and hypertrophic cardiomyopathy. This suggests that no single single nucleotide polymorphisms significantly influence the causal estimate. In this study, we input HSF1 into the CTD website to identify diseases associated with core genes. Core gene HSF1 was found to be associated with diabetes, type 2 diabetes, hypertrophic cardiomyopathy, heart disease, hypertension, metabolic disorders, and inflammation. (2) This study is primarily based on large international databases, including GWAS and GEO expression profiles from European populations, and employs genetic instrumental variables for causal inference, ensuring high scientific rigor. For biomedical research in China, the findings highlight the need to strengthen the collection and integration of local multi-omics data and to apply similar approaches to identify precision prevention and treatment targets suitable for the Chinese population.

Key words: diabetes, hypertrophic cardiomyopathy, Mendelian randomization, HSF1, genetics, differentially expressed genes