Multi-omics signatures of circulating factors associated with cardiorespiratory fitness adaptations in individuals with prediabetes

Abstract Background Patients with insulin resistance exhibit reduced cardiorespiratory fitness (CRF), assessed by peak oxygen consumption (VO₂peak), compared with healthy age-matched individuals. Although high-intensity interval training (HIIT) can substantially improve VO₂peak, there is considerable interindividual variability in this response. Therefore, further research is needed to elucidate the molecular mechanisms underlying the heterogeneous response of VO₂peak to HIIT in individuals with prediabetes. Methods Proteomic analyses of serum samples, along with fecal metagenomic and targeted metabolomic profiling, were conducted in medication-naïve, overweight and obese Chinese men with prediabetes ( n = 35; aged 24–62 years). All participants underwent a 12-week HIIT intervention, and biological samples were collected both before and after the intervention to evaluate exercise-induced alterations in circulating proteins, gut microbial composition, and metabolite profiles. Results After 12 weeks of HIIT, mean VO₂peak increased by 0.47 L/min with individual responses ranging from 0 to 1.7 L/min. Baseline levels of short-chain fatty acid (SCFA)-producing genera, including Prevotella (β = 105.65, P = < 0.001, FDR = 0.034), Coprococcus (β = 50.22, P = 0.01, FDR = 0.39), and Hungatella (β = 40.72, P = 0.025, FDR = 0.50), were positively associated with ΔVO₂ peak. In contrast, baseline levels of the erythropoiesis-stimulating hormone erythropoietin (EPO) (β = −279.03, P = 0.024, FDR = 0.99) were negatively associated with ΔVO₂ peak. Exercise-induced changes in growth hormone 1 (β = 63.97, P = 0.04, FDR = 0.99) were positively associated with ΔVO₂ peak, whereas exercise-induced changes in BTB and CNC Homology 1 (β = −250.82, P = 0.01, FDR = 0.99), a repressor of heme oxygenase-1, were negatively associated with ΔVO₂ peak. In multiple linear regression analysis including clinical variables, percentage lean mass (β = 64.17, P = 0.0005) was the strongest variable associated with Δ VO₂peak. The clinical model explained 27% of the variance which increased to 37% ( P = 0.002) upon inclusion of exercise-associated circulating factors such as EPO. Conclusions Our findings reveal that baseline proteomic and metagenomic signatures are associated with VO₂peak adaptations. These multi-omics signatures may support the clinical implementation of personalized exercise interventions to improve CRF in individuals with prediabetes.