Genome-wide association study of accelerometer derived physical activity: the HUNT study, Norway

Background: Physical inactivity is a global health concern and causes more than five million premature deaths annually. Although heritability estimates suggest a strong genetic influence on physical activity, genome-wide association studies have reported weak associations involving few genetic variants. Partly because physical activity is a complex phenotypic trait that is difficult to accurately capture.

Purpose: To identify genetic variation associated with different aspects of device-measured moderate to vigorous physical activity, such as average minutes of brisk walking or running per day and accumulated bouts of at least 10 or 30 consecutive minutes.

Methods: In HUNT4, 26,159 participants had at least one valid day of device-measured physical activity and genotyping of good quality. Machine-learning was used to classify every five seconds of activity into moderate walking, brisk walking, running, or cycling. This was accumulated into average minutes per day of ‘moderate to vigorous activity’ or ‘brisk walking or running’, and in consecutive bouts of at least 10 or 30 minutes. REGINIE was used to analyze genome-wide association with significance threshold of 5×10^-8. Follow-up analyzes included LD-score regression, ANNOVAR and GTEx to annotate plausible gene and tissue-expressions.

Results: One genetic locus was associated with 10-minute bouts of moderate to vigorous activity, whereas two loci were associated with brisk walking or running in 30-minute bouts. For the latter phenotype, the lead genetic variant rs111388372 on chromosome 4 is located within the GATB gene that are involved in oxidative phosphorylation and is generally expressed across tissues.

Conclusion: Three genetic loci was associated with bouts of continuous moderate to vigorous physical activity, with GATB as a candidate gene.

Practical implications: Genetic variation associated with physical activity provides biological insight into possible mechanisms and may contribute to genetic instruments applicable in Mendelian randomization studies.

Funding: NTNU Health, Norwegian University of Science and Technology (grant no. 81771516)