Abstract 15379: A Novel Acoustic Biomarker of Quality of Life in Left Ventricular Assist Device Recipients

Introduction: Left ventricular assist device (LVAD) recipients experience variable improvements in quality of life (QoL). Methods for optimizing LVAD support to improve QoL are needed. Hypothesis: We hypothesized that acoustic signatures obtained from digital stethoscopes can predict patient-centered outcomes in LVAD recipients. Methods: Subjects were recruited from the LVAD clinic at Duke University Medical Center. Subjects were trained to use a digital stethoscope and recorded their precordial sounds weekly for 6 months. Subjects also completed a Kansas City Cardiomyopathy Questionnaire (KCCQ) weekly. The primary outcome for this study was QoL, as measured by the KCCQ summary score. Results: Twenty-four subjects, 8 supported by a HeartWare LVAD and 16 supported by a HeartMate3 (HM3) LVAD, were enrolled. We developed a novel algorithm to filter LVAD sounds from recordings. Unsupervised clustering of LVAD-mitigated sounds revealed discrete groups of acoustic features. Six out of 16 [38%] HM3 recipients made up a unique cluster and shared a common acoustic feature that we have termed the pulse synchronized sound (PSS) based on its temporal association to the artificial pulse of the HM3. Subjects with the PSS had significantly better KCCQ scores at baseline (median, 89.1 [IQR, 86.2-90.4] vs 66.1 [IQR, 31.1-73.7], P = 0.03) and over the 6 months of the study (mean = 79.67 [95% CI, 74.55 - 84.78] vs 60.57 [95% CI, 55.58 - 65.55], P < .001). Mechanistically, the PSS shares acoustic features with patient-derived intrinsic sounds. Because the PSS cannot be detected by routine auscultation, we developed a robust machine learning algorithm to detect the PSS in precordial sound recordings (AUC = 0.95 by leave-one-subject out cross-validation). Conclusions: We have identified a novel acoustic feature associated with better QoL in HM3 recipients. This novel biomarker is the first of its kind in LVAD recipients and may provide a method for assaying optimized LVAD support.