Genetic risk scores to predict the prognosis of chronic heart failure patients in Chinese Han

Abstract

Chronic heart failure (CHF) has poor prognosis and polygenic heritability, and the genetic risk score (GRS) to predict CHF outcome has not yet been researched comprehensively. In this study, we sought to establish GRS to predict the outcomes of CHF. We re-analysed the proteomics data of failing human heart and combined them to filter the data of high-throughput sequencing in 1000 Chinese CHF cohort. Cox hazards models were used based on single nucleotide polymorphisms (SNPs) to estimate the association of GRS with the prognosis of CHF, and to analyse the difference between individual SNPs and tertiles of genetic risk. In the cohort study, GRS encompassing eight SNPs harboured in seven genes were significantly associated with the prognosis of CHF (P = 2.19 × 10-10 after adjustment). GRS was used in stratifying individuals into significantly different CHF risk, with those in the top tertiles of GRS distribution having HR of 3.68 (95% CI: 2.40-5.65 P = 2.47 × 10-10 ) compared with those in the bottom. We developed GRS and demonstrated its association with first event of heart failure endpoint. GRS might be used to stratify individuals for CHF prognostic risk and to predict the outcomes of genomic screening as a complement to conventional risk and NT-proBNP.

Trial registration: ClinicalTrials.gov NCT03754101.

Keywords: chronic heart failure; exome sequencing; genetic risk score; inheritance.

Conflict of interest statement

The authors declare no competing financial interests.

© 2019 The Authors. Journal of Cellular and Molecular Medicine published by Foundation for Cellular and Molecular Medicine and John Wiley & Sons Ltd.

Figures

Figure 1

Combining mass spectrometry and whole exome sequencing to screen for target genes. A, Venn diagram to analyse the overlapping protein of using mass spectrometry of human myocardial tissue (Normal N = 7, ICM N = 6, HCMpEF N = 4, HCMrEF N = 5, DCM N = 6), 319 proteins were detected in more than three groups. B, Principal component analysis (PCA) of whole exome sequencing was performed to populations construction defined by Eigenstrata, the result demonstrated the consistency of the study (loadings of intermediates in PC1 and PC2 are shown in blue). C‐F, Volcano plot of genes used to build GRS in different phenotype of Mass spectrometry

Figure 2

The distribution of GRS and combined effects of risk alleles on the prognosis of HF in prospective cohort study (A). For each subject, the number of risk alleles of eight replicated loci was summed to represent an individual's genetic risk score (range, −3 to 7). Individuals in each risk allele category are shown along the X‐axis, and Y‐axis on left represents the frequency of each genetic risk score category. (B, C) Cox proportional hazards model analysis after adjusted for gender, age, hypertension, hyperlipemia, diabetes mellitus, current smoking and β‐blocker treatment, showed the association of GRS with cardiovascular deaths or cardiac transplantation in the tertiles of genetic risk score (B, HR = 3.68, 95% CI 2.40‐5.65, P = 2.47 × 10−10) and quartiles (C, HR = 6.76, 95% CI 3.21‐14.28, P = 5.27 × 10−7)

Figure 3

Predictive outcomes of HF using the GRS and conventional risk factors. A and B, Receiver operating characteristic (ROC) analyses were performed to individual of traditional risk factors, and compound factors. C, C‐index for Cox regression of incident HF with individually traditional risk and in combination factors was illuminated that the model, traditional risk combining NT‐proBNP and GRS, has better predictive ability compared with others