Heritability of sedentary behavior: Genetic factors explained 56% of the individual differences in objective sedentary behavior & 26% of the individual differences in self‐reported sedentary behavior

Heritability of objectively assessed and self‐reported sedentary behavior. Nienke M. Schutte  Charlotte Huppertz  Stieneke Doornweerd  Meike Bartels  Eco J.C. de Geus  Hidde P. van der Ploeg. Scandinavian Journal of Medicine & Science in Sports, March 18 2020. https://doi.org/10.1111/sms.13658

Abstract: Understanding the sources of the large individual differences in sedentary behavior is of great importance as this behavior is associated with pre‐mature mortality and non‐communicable diseases. Here, we report on the contribution of genetic and environmental factors to the variation in objectively assessed (accelerometer) sedentary behavior and self‐reported sitting and their shared genetic basis. In addition, the overlap of the genetic risk factors influencing sedentary time and moderate‐to‐vigorous physical activity (MVPA) was estimated. A sample of 800 individuals (twins and their siblings) was equipped with an Actigraph accelerometer for 7 days and reported on their sitting time and time spent on MVPA on those days using the IPAQ‐SF. Genetic factors explained 56% (CI: 44%, 65%) of the individual differences in objective sedentary behavior (Actigraph) and 26% (CI: 0%, 51%) of the individual differences in self‐reported sedentary behavior (IPAQ‐SF). A modest correlation (0.33) was found between these measures, which was for 45% accounted for by genetic influences. The genetic correlation was 0.49 reflecting a partly overlapping set of genes that influenced both measurements. A modest correlation (−0.27) between Actigraph‐derived sedentary time and MVPA was found, which was 13% accounted for by genetic effects. The genetic correlation was −0.31, indicating that there are overlapping genetic variants that increase sedentary time and decrease MVPA or vice versa. To conclude, more than half of the individual differences in objective sedentary time could be attributed to genetic differences, while for self‐reported sitting this was much lower. In addition, using objective measurements, this study confirms that sedentary time is not simply the inverse of MVPA. Future studies are needed to understand the pathways translating genomic variation into variation in these behaviors and how this knowledge might feed into the development of health promotion interventions.

4 DISCUSSION

The main aim of the current study was to extend the scarce literature on the heritability of objective sedentary behavior. A relatively large sample of adult male and female twins and their siblings was equipped with an Actigraph for 7 consecutive days and reported on their sitting time and time spent on MPVA activities. We showed that more than half of the individual differences in objectively measured sedentary time could be attributed to genetic differences (56%). This estimate is comparable to the previously reported estimate of 47% for sedentary time measured objectively using a combined heart rate and movement sensor in an older, mostly female population.15

The heritability estimate for self‐reported sitting time in Dutch adults was much lower (26%) than that for the objective measure, but in good keeping with the heritability of 35% for total self‐reported sitting time in a cohort of older (aged 53‐67) Finnish adults.11 The lower heritability of self‐reported sitting time might in part be explained by recall bias, social desirability bias, or other measurement bias, which in the model are part of the unique environmental component (E), thereby inflating the influence of E. The unstandardized environmental variance in self‐reported sitting time was indeed higher compared with the environmental variance component of objective sedentary time. Another explanation for the lower heritability of our self‐report measure compared with objectively measured sedentary time is that the IPAQ‐questionnaire assesses only sitting time, while objectively measured sedentary time will also include time lying down (daytime napping) and standing still.

A few attempts have been made to identify the genetic variants underlying the heritability of self‐reported sedentary behaviors. In the Québec Family Study, a variant of the melanocortin‐4 receptor (MC4R) gene was found to be associated with a combined measure of self‐reported sedentary time and physical inactivity28 and in the Framingham Heart study an association of the fat mass and obesity‐associated (FTO) gene with sitting time was reported.29 These candidate gene studies are now understood to have been underpowered and confirmation through meta‐analysis of genome‐wide association (GWA) studies in very large samples from multiple cohorts are direly needed.21 A GWA study using accelerometer data of ~100 000 participants from the UK Biobank cohort reported 4 loci for sedentary time (rs26579 near MEF2C‐AS2, rs25981 near EFNA5, rs1858242 near LOC105377146; and rs34858520 near CALN1).30

If self‐reported sitting time and objectively measured sedentary time could be safely mixed in meta‐analyses it would become easier to accrue the sample sizes needed to identify the many genetic variants that may play a role in this complex and likely polygenetic behavioral trait. This requires a significant overlap in the genetic variants that influence self‐reported and objectively measured sedentary time. Our results are mildly encouraging for such future endeavors. Previous studies had shown mixed results when comparing accelerometer‐based sedentary time to survey derived sedentary time ranging from poor to reasonably strong agreement.31, 32 At the phenotypic level, we find a significant but modest correlation of r = 0.33 which fits this pattern of results. However, a bivariate genetic decomposition of the phenotypic correlation showed that the genetic variants that are associated with self‐reported sitting are also associated with objectively measured sedentary behavior. Although the phenotypic correlation is modest only, the genetic correlation (rG = 0.49) might support meta‐analysis across both types of measures in genome‐wide gene‐finding studies; at least a part of the genetic variants relevant to sedentary behavior will be associated with both measures.

Recently, the Sedentary Behavior Research Network (SBRN; a network connecting sedentary behavior researchers and health professionals from around the world) updated its definitions on, among others, sedentary behavior and physical inactivity and thereby supported that an insufficient physical activity level is not the same as sedentary behavior.1 This idea is supported by the modest inverse correlations detected between objective sedentary time and MVPA in this study, matching those observed in earlier studies.16-18 This confirms that sedentary time is not simply the inverse of MVPA. Interestingly, the observed association between occupational sedentary time and occupational MVPA is −0.29, whereas the association between non‐occupational sedentary time and non‐occupational MVPA increases to −0.42, suggesting that outside work time, the association between MVPA and sedentary time is stronger. As a large portion of this is leisure time, it might be that when given the free choice, people who do more physical activity, also sit less.

When dividing total accelerometer wear time into occupational and non‐occupational time, the heritability estimate of occupational sedentary time (45%) is higher than the heritability estimate for non‐occupational sedentary time (28%). This was unexpected as sitting time at work was considered to be more under external environmental rather than under the internal control of behavioral disposition. However, sitting time is known to be strongly associated with type of work with white‐collars generally accumulating higher levels of sedentary behavior than blue collars.33 The type of work will be strongly dependent on educational attainment which has shown to be a heritable trait.34 Possibly, the genetic factors that are associated with educational attainment or other traits that co‐determine the employment setting might contribute to the variation in occupational sedentary time.

The heritability of objectively measured MVPA was 46%. This heritability is comparable to the previously reported estimates of 47% for objective MVPA, based on a combined heart rate and movement sensor15 and estimates of 55% and of 47% for MPA and VPA measured with an accelerometer.35 Self‐reported MVPA from the IPAQ showed a low heritability estimate (14%) and this echoes reports of low heritability of self‐reported MVPA in other adult samples.36, 37 Not surprisingly, we detected only a small correlation of r = 0.11 between objective and self‐reported MVPA. The latter may suffer from a larger measurement error because it includes a broad range of commuting, work, and household activities for which both intensity and duration may be hard to recall. Lee et al (2011) conducted a systematic review of the validity of the IPAQ‐SF and displayed that correlations between VPA/MPA/walking and objective standards showed great variability, ranging from −0.18 to 0.76.38 When self‐report is limited to voluntary leisure time activities of moderate‐to‐vigorous intensity, recall seems to be better. Regular exercise behavior, which is arguably easier to recall than all daily MVPA due to the mostly organized nature of exercise, has shown to be heritable in many adult twin and family samples with estimates higher than 40%.39, 40

Some limitations of this study must be noted. The twin sample used in the current study was relatively highly educated and 75% of the sample were female, which limits generalizability to the general population. This could have reduced the variance in our sample, and therefore have influenced our estimates. In addition, because of the small number of male participants we were unable to stratify our analyses by gender. Finally, as some people nowadays have flexible working hours (working part time or from home), it might be difficult to indicate their workday start and finishing times in the diary we used. Estimates of the genetic contribution to the (co)variance in occupational and non‐occupational sedentary time and MVPA might increase when a stricter distinction between working hours and non‐working hours is achieved as measurement error is reduced.

5 PERSPECTIVE

The majority of the variance in objectively measured sedentary time in a large sample of Dutch twins and their siblings could be explained by genetic factors. As opposed to general beliefs regarding the heritability of health behaviors, high heritability estimates do not signal that interventions are wasted efforts. Interventions on behavioral traits which are proven to be hereditary can have a large mean effect. Biological influences on the trait might explain the variation or individual differences in effect. The key is to identify the individuals who benefit the most from the intervention and exploit these biological influences on sedentary behavior in personalized or stratified interventions. Heritability studies serve to remind us that there is a biological component to individual differences in behavior. Sedentary time is not an exception. Interventions ignoring this underlying biology may prove less effective than those that build on furthering our understanding of the pathways from the genomic level to health behaviors.