Protein biomarkers of cardiac remodeling and inflammation associated with HFpEF and incident events
Jessica A. Regan, L. Truby, Usman A. Tahir, Daniel H. Katz, Maggie Nguyen, Lydia Coulter Kwee, Shuliang Deng, James Wilson, Robert J. Mentz, William E. Kraus, Adrian F. Hernandez, Robert E. Gerszten, Eric D. Peterson, Rury R. Holman, Svati H. Shah
Abstract
Abstract There is increasing evidence that HFpEF is a heterogeneous clinical entity and distinct molecular pathways may contribute to pathophysiology. Leveraging unbiased proteomics to identify novel biomarkers, this study seeks to understand the underlying molecular mechanisms of HFpEF. The discovery cohort consisted of HFpEF cases and non-HF controls from the CATHGEN study (N = 176); the validation cohort consisted of participants from the TECOS trial of patients with diabetes (N = 109). Proteins associated with HFpEF were included in a LASSO model to create a discriminative multi-protein model and assessed in the validation cohort. Survival models and meta-analysis were used to test the association of proteins with incident clinical outcomes, including HF hospitalization, mortality and HFpEF hospitalization in CATHGEN, TECOS and the Jackson Heart Study. In the derivation set, 190 proteins were associated with HFpEF in univariate analysis, of which 65 remained significant in the multivariate model. Twenty (30.8%) of these proteins validated in TECOS, including LCN2, U-PAR, IL-1ra, KIM1, CSTB and Gal-9 (OR 1.93–2.77, p < 0.01). LASSO regression yielded a 13-protein model which, when added to a clinical model inclusive of NT-proBNP, improved the AUC from 0.82 to 0.92 ( p = 1.5 × 10 –4 ). Five proteins were associated with incident HF hospitalization, four with HFpEF hospitalization and eleven with mortality ( p < 0.05). We identified and validated multiple circulating biomarkers associated with HFpEF as well as HF outcomes. These biomarkers added incremental discriminative capabilities beyond clinical factors and NT-proBNP.