Non-dilated left ventricular cardiomyopathy carries distinct circulating epigenetic signatures
B Hingst, L Nicin, A Kind, C OeingAbstract
Introduction
The European Society of Cardiology (ESC) recently defined non-dilated left ventricular cardiomyopathy (NDLVC) as a distinct subtype of cardiomyopathies, characterized by the absence of a dilated left ventricle. In contrast to dilated cardiomyopathy (DCM), patients with NDLVC have a higher prevalence of genetic variants in arrhythmogenic genes, but less risk of ventricular arrhythmia and sudden cardiac death. Only a minority of NDLVC patients can develop DCM over time. Little is known about the pathophysiological mechanisms that occur in NDLVC and how they differ or overlap with mechanisms compared to DCM and other cardiomyopathies.
Purpose
The objective of this study was to evaluate whether circulating methylation signatures differentiate NDLVC from DCM as well as ischemic cardiomyopathy (ICM) and healthy controls (CTR).
Methods
We analyzed the DNA-methylation of circulating blood in a discovery cohort of NDLVC (n=9), DCM (n=43), ICM (n=40) and CTR (n=21) and in a matched validation cohort consisting of NDLVC (n=10), DCM (n=4) and CTR (n=6) via the Illumina EPIC array. Quality control and processing was performed in R using the sesame package. Differentially methylated probes (DMPs) were identified with linear regression adjusted for cell composition, age and sex. Fixed effect meta-analyses were used to determine the most reliable DMPs in discovery and validation cohort. Benjamini-Hochberg false discovery rate (FDR) was used to adjust for multiple testing.
Results
We identified 21 DMPs in the discovery cohort and validation cohort that passed the FDR of 0.05 in a fixed-effect meta-analysis. We found the methylation values of these probes to perform better classification in unsupervised learning, when added to clinical data compared to solely clinical data (Fig. 1 A). Interestingly, most of the epigenetic signatures in NDLVC with a nominal p-value of less than 0.05 and a methylation difference of 4% differed from DCM and only a few DMPs were shared between both conditions (Fig. 1 B). By integrating a STRING network analysis on NDLVC specific CpG-Sites, we identified hub genes, of which seven were used to train a combined generalized linear model for classifying patients as NDLVC or not-NDLVC in the discovery cohort, including patients with ICM (Fig. 1 C). When tested on the validation cohort, the model showed good performance with an AUC of 0,824 (Fig. 1 D).
Conclusion
This study shows that the circulating methylome of NDLVC differs compared to CTR and DCM and adds to a growing understanding of NDLVC as a distinct cardiomyopathy. This could potentially be used to classify patients with cardiomyopathies in NDLVC or not-NDLVC. Further validation of the model in large scale studies is necessary to assess its performance and reliability.Figure1