A genome-wide association and polygenic risk score study on abnormal electrocardiogram in a Chinese population

Abstract

Electrocardiography is a common and widely-performed medical examination based on the measurement and evaluation of electrocardiogram (ECG) to assess the up-to-date cardiac rhythms and thus suggest the health conditions of cardiovascular system and on a larger level the individual's wellness. Abnormal ECG assessment from the detection of abnormal heart rhythms may have clinical implications including blood clots in formation, ongoing heart attack, coronary artery blockage, etc. Past genetic-phenotypic research focused primarily on the physical parameters of ECG but not the medical evaluation. To unbiasedly uncover the underlying links of genetic variants with normal vs. abnormal ECG assessment, a genome-wide association study (GWAS) is carried out in a 1006-participant cohort of Chinese population effectively genotyped for 243487 single nucleotide polymorphisms (SNPs). Both age and sex are influential factors, and six novel SNPs are identified for potential association with abnormal ECG. With the selected SNPs, a polygenic risk score (PRS) differentiates the case-control subgroups, and correlates well with increased risk of abnormal ECG. The findings are reproduced in an independent validation cohort. The derived PRS may function as a potential biomarker for prospectively screening the high-risk subgroup of heart issues in the Chinese population.

Figure 1

Percentage of abnormal ECG assessment by sex (A), age (B), sex and age (C), and subtypes of ECG issues (D) in the cohort.

Figure 2

Manhattan plot of GWAS on the phenotype of normal vs. abnormal ECG. Bonferroni corrected threshold (2 × 10^–7) and less stringent candidate threshold (2 × 10^–5) respectively correspond to levels of 6.7 and 4.7 for − log₁₀(P).

Figure 3

Distribution of PRS by ECG phenotype in the jittered dot plot (A) and the histogram (B).

Figure 4

Percentage of abnormal ECG in subgroups of PRS deciles (A) and PRS scores (B). The ten deciles refer to the sequential 10% segments of the cohort ordered by the PRS values. For example, decile #1 annotates individuals with the lowest 10% PRS, and decile #10 annotates individuals with the highest 10% PRS.

Figure 5

Distribution of PRS by ECG phenotype in the jittered dot plot (A) and the histogram (B) for the validation set.

Figure 6

Percentage of abnormal ECG in subgroups of PRS percentiles (A) and PRS scores (B). Percentile #1, #2, and #3 respectively annotates the low, middle, and high one-third of PRS individuals. The validation set is divided into three percentiles rather than ten deciles due to its limited sample size.

Figure 7

ROC plot of the PRS in the validation set. Cutoff (in the range of [0, 1]) refers to the threshold to dichotomize predicted probability into binary classes. AUC area under the ROC curve. Note: sex and age are included in the predictive model.