Exploring GRK2 correlation with ventricular remodeling and functional capacity in HF patients: a data augmentation-empowered approach

Abstract Background G-protein coupled receptor kinase 2 (GRK2) is a serine/threonine kinase that regulates multiple intracellular signaling pathways. It plays a key role in the modulation of the catecholaminergic response mediated by β-adrenergic receptors, triggering processes of desensitization and downregulation of these receptors, which are hallmark features of chronic Heart Failure (HF). A direct correlation has been identified between GRK2 in lymphocytes and cardiac tissue, suggesting blood levels of GRK2 as a potential new biomarker for diagnostic and prognostic assessment of patients with HF. Purpose We aimed to assess the correlation between GRK2 levels, cardiac structural remodeling, and functional capacity in patients with chronic HF. Methods We enrolled 21 HF patients who underwent echocardiography and cardiopulmonary exercise testing (C-PET) to assess left ventricular volume (LVEDVi) and functional capacity (VO2max/kg), respectively. GRK2 levels were evaluated in peripheral blood mononuclear cells (PBMC) by Western Blot (WB), on blood samples collected before and after C-PET. Western blot quantification of GRK2 was normalized to β-actin as a loading control, and results were reported as fold increase (GRK2-FI) compared to baseline/control values. To understand the relationship between these parameters, we conducted a comprehensive linear regression study, considering all potential combinations of the variables of interest. In addition, to expand our limited sample size, we used a Conditional Tabular Generative Adversarial Network-based Data Augmentation technique that generates new, synthesized data samples from an existing dataset with the same features as the original data, thereby improving the performance of the regression model. In this way, we obtained a sample of 100 elements, comprising 75 for training and 25 for testing. Results We investigated 10 different linear regression models to detect the best-fitting one. Bayesian Ridge provided the best results (MAE 0,465). Data augmentation resulted in a 100% structure and validity score and in a 86,63% column shapes/pair trends score. Regression results highlight that VO2max/kg negatively correlates with LVEDVi, exhibiting a clear linear trend. This indicates that, as expected, HF patients with increased ventricular volumes demonstrate reduced exercise capacity. This negative correlation is also evident between GRK2-FI and LVEDVi. Thus, GRK2 levels can predict the degree of ventricular remodeling. Conversely, the relationship between GRK2-FI and VO2max/kg is not entirely clear, although they seem to share a common trend. Conclusions These data suggest that GRK2 might reflect exercise capacity in patients with HF. The assessment of peripheral GRK2 levels is a promising biomarker to predict both the degree of left ventricular remodeling and functional capacity. Further research is needed to better understand the relationship between peripheral GRK2 levels and clinical outcomes.Regression resultsFor image description, please refer to the figure legend and surrounding text.