Background: Genome-wide association studies (GWAS) have reported hundreds of genetic loci for resting heart rate (HR). Importantly, resting HR is not constant but varies physiologically. The objective of this work was to evaluate the effect of different levels of variability in resting HR measurements on the strength of genetic associations.
Methods: We computed the standard deviation (SD_raw) of the difference between two resting HR measurements taken during the same day in ~74,000 subjects free of cardiovascular disease from UK Biobank. Physiological variability was simulated by creating normal distributions with mean 0, standard deviation 0.5, 1, and 2 times SD_raw (3.95, 7.90, and 15.80 bpm, respectively) and a sample size that equalled the number of subjects. These values were added in a random order to the raw heart rate measurements. This process was repeated 10 times. We then conducted a GWAS using BOLT-LMM for the raw and simulated datasets. The effect of physiological variation was evaluated by counting the number of genome-wide significant (p < 5x10-8) variants in the the locus with the strongest reported association for resting HR (MYH6, chromosome 14) and by measuring the p-value of corresponding lead variant (rs422086).
Results: The number of variants with a genome-wide significant p-value was highest for the original resting HR measurements (n=44) and decreased with increasing levels of variability: median values: N=42.5, 35.5, and 7.5 for 0.5, 1, and 2x SD_org respectively. A similar trend was observed for the change in p-value: p-value was lowest for the raw HR measurements 1x10-25, but increased when applying variability: median –log10(p-value): 23.9, 17.4, and 8.6, for 0.5, 1, and 2x SD_org respectively.
Conclusions: Results from the simulation study demonstrate that physiological variability can dramatically affect the discovery of genetic variants. It is therefore important to ensure that measurements are taking under very similar conditions.