Background: Dilated cardiomyopathy (DCM) has been associated with variants in >50 genes. Understanding which associations are important is a key challenge.
Purpose: 1) Define the genetic architecture of DCM in the context of background population genetic variation, 2) Evaluate genotype-phenotype correlations for the commonest genetic cause of DCM.
Methods: Prospectively recruited unrelated DCM patients (n=977) underwent comprehensive clinical evaluation and targeted sequencing (57 DCM genes). Burden testing evaluated the excess of rare variants (minor allele frequency <0.0001) compared to 60,000 reference samples (Exome Aggregation Consortium, ExAC, dataset).
Detailed phenotypic characterisation was performed by cardiac MRI in 733 patients. Those recruited prior to 31st December 2015 (n=605) were evaluated using Cox proportional hazard modelling for the primary composite endpoint of cardiovascular death, major arrhythmic events (aborted sudden cardiac death, appropriate ICD activation, sustained ventricular tachycardia and ventricular fibrillation) and major heart failure events (heart transplant, left ventricular assist device insertion, unplanned heart failure hospitalisation).
Results: Compared to ExAC, there was a significant excess of truncating variants in the genes encoding titin (TTNtv, 11.4%), desmoplakin (1.5%), BCL2 Associated Athanogene 3 (0.4%) and lamin (0.3%), and of non-truncating variants in the genes encoding beta myosin heavy chain (2.5%), troponin T2 (1.0%), titin (6.74%), troponin C (0.5%), and desmoplakin (2.2%) (all p<0.0009, corrected for multiple testing). There was no significant excess variation in the genes TPM1, SCN5A, MYBPC3, ZBTB17, or PRDM16.
Of these genes, we were sufficiently powered to evaluate the phenotype of TTNtv (n=82/733, 11.1%). There was no difference in biventricular function between TTNtv+/− groups (mean ±sd, left ventricular ejection fraction, LVEF (%), 37.7±13.7 vs 39.1±12.3, p=0.42; right ventricular ejection fraction (%); 50.2±14.4 vs 51.9±13.8, p=0.54). However, in the presence of a history of alcohol excess, LVEF was reduced by 15.4% (95% confidence interval -21.9 to -8.8%) compared to a patient without either risk factor (p<0.0001), independently of other predictors of LVEF.
Of 605 patients (TTNtv =71, 11.7%) followed for a median of 3.9 years (IQR 2.0–5.8 years), 9 patients with TTNtv (12.7%) reached the primary end point compared to 71 patients (13.3%) without TTNtv (hazard ratio 0.81 [95% confidence interval 0.41–1.63], p=0.56) (Figure). There were no significant gender differences in outcome amongst TTNtv patients (Figure).
Conclusion(s): TTNtv represent the largest genetic burden in DCM. Some newer DCM genes such as PRDM16 and ZBTB17 showed no significant enrichment. We demonstrate environmental modification of the TTNtv DCM phenotype and show the prognosis of TTNtv DCM is similar to non TTNtv DCM. These data help redefine our understanding of the genetic basis of DCM.
Acknowledgement/Funding: Medical Research Council, Jansons Foundation, Wellcome Trust and NIHR Cardiovascular BRU of Royal Brompton Hospital/Imperial College London
