Causal associations between cardiorespiratory fitness and type 2 diabetes

Abstract

Higher cardiorespiratory fitness is associated with lower risk of type 2 diabetes. However, the causality of this relationship and the biological mechanisms that underlie it are unclear. Here, we examine genetic determinants of cardiorespiratory fitness in 450k European-ancestry individuals in UK Biobank, by leveraging the genetic overlap between fitness measured by an exercise test and resting heart rate. We identified 160 fitness-associated loci which we validated in an independent cohort, the Fenland study. Gene-based analyses prioritised candidate genes, such as CACNA1C, SCN10A, MYH11 and MYH6, that are enriched in biological processes related to cardiac muscle development and muscle contractility. In a Mendelian Randomisation framework, we demonstrate that higher genetically predicted fitness is causally associated with lower risk of type 2 diabetes independent of adiposity. Integration with proteomic data identified N-terminal pro B-type natriuretic peptide, hepatocyte growth factor-like protein and sex hormone-binding globulin as potential mediators of this relationship. Collectively, our findings provide insights into the biological mechanisms underpinning cardiorespiratory fitness and highlight the importance of improving fitness for diabetes prevention.

Fig. 1. Flow chart of the study.

The figure shows the different steps in describing the association between cardiorespiratory fitness and incident type 2 diabetes, the derivation of a genetic score for fitness for the Mendelian randomisation analysis and the investigation of mediation using proteomics.

Fig. 2. Observational association between fitness and risk of type 2 diabetes.

Odds Ratio (and 95% confidence interval) for the association of cardiorespiratory fitness level with incident type 2 diabetes after adjustment for potential confounding variables (Model 3, Supplementary Table 2) and when using 41 ml O₂⋅min⁻¹⋅kg⁻¹ fat-free mass (FFM) as the reference level. Histogram bins are 1.75 ml O₂⋅min⁻¹⋅kg⁻¹ FFM wide and represent the distribution of cardiorespiratory fitness levels in UK Biobank participants (n = 73,574) who completed exercise testing. Source data are provided as a Source Data file.

Fig. 3. Forest plot for Mendelian randomisation analysis results for the genetically predicted effect of cardiorespiratory fitness on type 2 diabetes, and multivariable Mendelian randomisation analyses of cardiorespiratory fitness on type 2 diabetes after adjustment for the effects of glycaemic traits and BMI.

Odds Ratios presented are based on Inverse Variance Weighted MR after using Radial-filtered instrumental variable. Mediation % represents the proportion of effect of fitness on T2D that is mediated by the intermediate traits, i.e. glycaemic traits and BMI. SD: standard deviation; CRF: cardiorespiratory fitness; adjFPG: adjusted for fasting plasma glucose; adjFI, adjusted for fasting insulin; adj2hrPG, adjusted for 2-h plasma glucose after oral glucose tolerance test; adjHbA1c, adjusted for HbA1c; adjBMI, adjusted for BMI; adjAll, adjusted for all the intermediate traits above. Sample sizes for each individual MR analysis are provided as a Source Data file.

Fig. 4. Volcano plot of associations between the enhanced 160-SNP genetic risk score for cardiorespiratory fitness and Protein Targets assessed by the aptamer-based technology (SomaScan©) in 10,707 individuals from the Fenland study.

Each dot represents a Somamer targeting a protein. The horizontal dashed line represents p < 0.0001 and Somamers with p < 0.0001 are annotated. N-terminal pro-BNP, N-terminal pro B-type natriuretic peptide, MSP, Hepatocyte growth factor-like protein, MYL6B Myosin light chain 6B MXRA7 Matrix-remodelling-associated protein 7, CD248 Endosialin, MYPC1 Myosin-binding protein C, slow-type, WISP-2, WNT1-inducible-signalling pathway protein 2, HS3S4 Heparan sulfate glucosamine 3-O-sulfotransferase 4, MMP-2 72 kDa type IV collagenase, Apo B Apolipoprotein B, MYOC Myocilin, VPS29 Vacuolar protein sorting-associated protein 29, SHBG Sex hormone-binding globulin, PAFAH Platelet-activating factor acetylhydrolase. Source data are provided as a Source Data file.

Fig. 5. Tissue and cell-type specific enrichment analysis for cardiorespiratory fitness using LDSC-SEG and DEPICT.

a Expression enrichment in a total 207 tissues and cell type specific expression data from GTEx v7 and Franke lab. Each dot represents a tissue or cell type further categorised into 9 general tissue groups indicated by different colours. b Results from DEPICT using independent fitness-associated variants (p < 10⁻⁵). Each bar represents a tissue or cell which is categorised in 10 physiological systems. Orange bar marks the four significantly enriched tissues (FDR < 0.05).