Multivariable regression modeling identifies relationships between DUSP-4, MMP-2, MMP-14, and TIMP-4 mRNA with left ventricular total collagen protein variation, independent of heart failure phenotype, in aortic banded swine

Abstract

Heart failure with preserved ejection fraction accounts for half of heart failure patients. Maladaptive interstitial cardiac fibrotic remodeling of the extra cellular matrix (ECM) in both the right (RV) and left ventricles (LV) is a contributing factor to the development of ventricular stiffness and diastolic dysfunction. Thus, the purpose of this study was to use an mRNA dataset of variables influencing the regulation (i.e., synthesis and degradation) of the ECM in two preclinical swine models of pressure overload-induced heart failure to identify biventricular molecular markers highly associated with LV collagen levels. Our objective was to build a model which would explain at least 75% variation in total collagen protein in the LV. Data from 28 sexually mature female Yucatan and Ossabaw swine was pooled per four experimental interventions: regular diet control (N=11), aortic banded (N=7), aortic banded + ovariectomy (N=6), and aortic banded + Western diet (N=5). 30 mRNA targets were measured in both the RV and LV, including collagen I and III, fibronectin, matrix metalloproteinases (MMP-1, 2, 3, 9, 13, 14) and their tissue-inhibitors (TIMP-1, 2, 3, 4), mitogen-activated protein kinases (MAPK-1, 3, 8, 9), dual specificity phosphatases (DUSP-1, 4, 6, 9, 10), estrogen receptor isoforms 1 and 2, and progesterone receptor and its membrane component for a total of 60 input variables. Missing data were mean imputed and linear regression models were used to identify variables with the strongest correlation to LV total collagen protein (r > 0.6). Identified variables were then used to develop a multivariable regression model, optimized with the largest adjusted R2. DUSP-4, MMP-2, MMP-14, and TIMP-4 mRNA in the RV, and DUSP-4 mRNA in the LV, were highly correlated to total collagen protein in the LV (r > 0.67, P < 0.0001). Together, these 5 variables explain 80% of the variation in total collagen protein in the LV (R2 = 0.80, adjusted R2 = 0.75, P = 4.721e-07). A novel multivariable regression model identified one chamber-independent (LV and RV) phosphatase (DUSP-4) involved in the synthesis of collagen and three chamber-dependent (RV) regulators of ECM degradation as critical to understanding variation in total LV collagen in a preclinical setting of pressure overload-induced heart failure. These molecular targets have associations independent of heart failure phenotype i.e., hemodynamic, sex hormone, and dietary combinations. Modeling biological relationships typically averages approximately 60% or less of variability explained, highlighting the potential for the model developed in this study to identify mechanistic targets of disease. Department of Defense W81XWH-18-1-0179 This is the full abstract presented at the American Physiology Summit 2023 meeting and is only available in HTML format. There are no additional versions or additional content available for this abstract. Physiology was not involved in the peer review process.