Prediction of acute heart failure events in patients with hypertrophic cardiomyopathy using RNA-Sequencing of plasma small non-coding RNAs

Abstract Background Hypertrophic cardiomyopathy (HCM) often causes acute heart failure (HF) events, which lead to worse quality of life and prognosis of patients with HCM. The recent emergence of cardiac myosin inhibitors to treat HCM highlights the clinical significance of identifying patients with HCM at high risk of developing acute HF events. However, no tools are available to predict acute HF events in HCM. Furthermore, the underlying molecular mechanisms responsible for the development of HF in HCM remain uncertain. Plasma RNA-sequencing (RNA-Seq) determines concentrations of thousands of small non-coding RNAs (sncRNAs) and has the potential to predict acute HF events in HCM. Purpose To predict acute HF events in patients with HCM using RNA-Seq of plasma sncRNAs and to determine signaling pathways dysregulated in those who subsequently experience acute HF events. Methods In this prospective, multi-center cohort study of patients with HCM, we conducted RNA-Seq of 3,740 plasma sncRNAs on 390 patients with HCM. The primary outcome was acute HF event defined as unplanned hospitalization due to HF. We developed a sncRNA-Seq-based model with linear discriminant analysis to predict acute HF events using data from one institution (training set, n=271). We also developed a reference model using clinical parameters with significant differences between patients with and without acute HF events in the training set. Then we tested the predictive ability in samples from the other institutions (test set, n=119), comparing the area under the receiver-operating-characteristic curve (AUC) between a combined model using the reference model plus the sncRNA-Seq-based model and the reference model using the Delong’s test. We also performed pathway analysis of microRNAs significantly (i.e., nominal P<0.05) associated with acute HF events. Results During a median follow-up time of 2.5 (interquartile 1.7–4.9) years, a total of 37 patients in the training set (14%) and 15 in the test set (13%) experienced an acute HF event. The model combining the reference model with the sncRNA-Seq-based machine learning model developed in the training set significantly outperformed the reference model (AUC 0.86 [95% confidence interval (CI) 0.74–0.97] vs. 0.68 [0.52–0.85], P=0.01, Figure) in the test set. The pathway analysis exhibited that both known pathways related to fibrosis, inflammation, and cell proliferation (e.g., the TGF-β signaling pathway) and novel pathways (e.g., the mTOR signaling pathway) were dysregulated with false discovery rate <0.01 in patients with HCM who subsequently experienced acute HF events (Table). Conclusions This study serves as the first to demonstrate the ability of plasma sncRNA-Seq to predict acute HF events in patients with HCM. Patients who subsequently experienced acute HF events exhibited dysregulation of both known and novel pathways.FigureTable