Integrative Clinical-Molecular Modeling Identifies LRRN4CL as a Determinant of Structural and Functional Myocardial Improvement

Abstract

BackgroundMechanical ventricular unloading and systemic circulatory support with left ventricular assist devices (LVADs) enable myocardial recovery in a subset of advanced heart failure (HF) patients, but predictors and mechanisms of recovery are not well understood. Integrating clinical and molecular data may improve identification of patients most likely to recover and uncover biologically relevant targets in HF. MethodsWe collected and analyzed left ventricular apical myocardial tissue and clinical data from 208 patients undergoing LVAD implantation across five centers. Pre-implant transcriptomic profiles (22,373 mRNA transcripts) were integrated with 59 clinical variables using supervised machine learning with repeated cross-validation to identify and prioritize features associated with myocardial recovery, defined as a binary outcome based on improvement in left ventricular ejection fraction (LVEF [≥]40%) and left ventricular end-diastolic diameter (LVEDD [≤]5.9 cm). We also modeled functional (LVEF) and structural (LVEDD) improvement as a continuous outcome without any predefined LVEF and LVEDD pathological thresholds. Feature prioritization was followed by validation in human myocardial tissue and mechanistic interrogation in human induced pluripotent stem cell-derived cardiomyocytes (iPSC-CMs). ResultsIntegrative models achieved modest discrimination for myocardial recovery as a binary categorical outcome (maximum mean cross-validated area under the curve 0.73{+/-}0.15), identifying clinical features such as HF duration, LVEDD, HF pharmacologic therapy, and device configuration. Leucine-rich repeat neuronal 4C-like (LRRN4CL), measured in human myocardium, consistently emerged as a top transcriptomic predictor across both binary and continuous metric models (functional and structural). Higher pre-LVAD LRRN4CL expression was associated with reduced likelihood of myocardial recovery and localized primarily to cardiomyocytes. In iPSC-CMs, LRRN4CL overexpression localized to the sarcoplasmic reticulum, induced transcriptional remodeling characterized by suppression of contractile pathways and activation of stress programs, impaired calcium handling, impaired contraction-relaxation kinetics, and diminished mitochondrial respiratory reserve capacity. ConclusionsIntegration of clinical and myocardial transcriptomic data identifies LRRN4CL as a novel marker associated with impaired myocardial recovery following LVAD-mediated ventricular unloading and systemic circulatory support. These findings move beyond predictive modeling, linking integrative computational discovery to cardiomyocyte dysfunction and providing a translational framework for biologically informed risk stratification and therapeutic targeting for myocardial recovery. CLINICAL PERSPECTIVEO_ST_ABSWhat Is New?C_ST_ABSO_LIIntegrative clinical and myocardial transcriptomic modeling identifies LRRN4CL as a novel molecular determinant of structural and functional changes after LVAD-mediated ventricular unloading and enhanced systemic circulatory support. C_LIO_LIElevated LRRN4CL expression is associated with adverse remodeling signatures, impaired calcium handling, and stress responses in human iPSC-derived cardiomyocytes. C_LIO_LIExperimental overexpression of LRRN4CL directly disrupts calcium cycling, contractile performance, and mitochondrial respiration linking molecular signature to functional phenotype. C_LI What Are the Clinical Implications?O_LIIdentification of LRRN4CL as a marker associated with impaired myocardial recovery supports future efforts toward biologically informed risk stratification for patients undergoing LVAD therapy. C_LIO_LILRRN4CL as a marker of cardiac improvement potential may extend beyond advanced HF to earlier stage disease patients and inform prognosis, risk stratification, and response to medical therapies. C_LIO_LIThese findings highlight LRRN4CL-associated pathways as potential therapeutic targets and demonstrate how integrative clinical-transcriptomic approaches can move beyond clinical prediction toward identification of new biologically precise therapeutic targets in HF following a bedside to bench and back approach. C_LI