1. Epidemiology and Global Health
  2. Genetics and Genomics

Mendelian randomization suggests a bidirectional, causal relationship between physical inactivity and adiposity

  1. Germán D Carrasquilla  Is a corresponding author
  2. Mario García-Ureña
  3. Tove Fall
  4. Thorkild IA Sørensen
  5. Tuomas O Kilpeläinen
  1. Novo Nordisk Foundation Center for Basic Metabolic Research, Faculty of Health and Medical Sciences, University of Copenhagen, Denmark
  2. Molecular Epidemiology, Department of Medical Sciences, and Science for Life Laboratory, Uppsala University, Sweden
  3. Department of Public Health, Section of Epidemiology, Faculty of Health and Medical Sciences, University of Copenhagen, Denmark
https://doi.org/10.7554/eLife.70386
Share this article
Cite this article
  1. Germán D Carrasquilla
  2. Mario García-Ureña
  3. Tove Fall
  4. Thorkild IA Sørensen
  5. Tuomas O Kilpeläinen
(2022)
Mendelian randomization suggests a bidirectional, causal relationship between physical inactivity and adiposity
eLife 11:e70386.
https://doi.org/10.7554/eLife.70386
  2. Download BibTeX
  3. Download .RIS

Abstract

Physical inactivity and increased sedentary time are associated with excess weight gain in observational studies. However, some longitudinal studies indicate reverse causality where weight gain leads to physical inactivity and increased sedentary time. As observational studies suffer from reverse causality, it is challenging to assess the true causal directions. Here, we assess the bidirectional causality between physical inactivity, sedentary time, and adiposity by bidirectional Mendelian randomization analysis. We used results from genome-wide association studies for accelerometer-based physical activity and sedentary time in 91,105 individuals and for body mass index (BMI) in 806,834 individuals. We implemented Mendelian randomization using CAUSE method that accounts for pleiotropy and sample overlap using full genome-wide data. We also applied inverse variance-weighted, MR-Egger, weighted median, and weighted mode methods using genome-wide significant variants only. We found evidence of bidirectional causality between sedentary time and BMI: longer sedentary time was causal for higher BMI [beta (95% CI) from CAUSE method: 0.11 (0.02, 0.2), p = 0.02], and higher BMI was causal for longer sedentary time (0.13 (0.08, 0.17), p = 6.3 x 10-4). Our analyses suggest that higher moderate and vigorous physical activity are causal for lower BMI (moderate: –0.18 (-0.3,–0.05), p = 0.006; vigorous: –0.16 (-0.24,–0.08), p = 3.8 × 10-4), but indicate that the association between higher BMI and lower levels of physical activity is due to horizontal pleiotropy. The bidirectional, causal relationship between sedentary time and BMI suggests that decreasing sedentary time is beneficial for weight management, but also that targeting adiposity may lead to additional health benefits by reducing sedentary time.

Editor's evaluation

This paper is of potential interest to those researchers and clinicians working in the area of physical activity and obesity. The authors have presented strong evidence of a causal relationship between physical activity and higher BMI. The conclusions that have been made are supported by the data and are translationally relevant.

https://doi.org/10.7554/eLife.70386.sa0

Introduction

Obesity and physical inactivity are major risk factors for a number of chronic diseases, such as type 2 diabetes, cardiovascular diseases and several types of cancer. Today’s epidemic of obesity and sedentary lifestyle is thus a major burden on public health systems worldwide (GBD 2019 Risk Factors Collaborators, 2020).

Many observational studies suggest that physical inactivity and increased sedentary time are associated with a higher risk of obesity (Lee et al., 2010; Du et al., 2013; Silva et al., 2019; Myers et al., 2017). However, other studies have indicated a reverse effect, where obesity leads to physical inactivity and increased sedentary time (Petersen et al., 2004; Mortensen et al., 2006; Bak et al., 2004; Barone Gibbs et al., 2020; Ekelund et al., 2008; Myers et al., 2017). Furthermore, randomized clinical trials of physical activity interventions have indicated that the causal effects of physical activity on body weight are modest (Church et al., 2009; Rosenkilde et al., 2012; Golubic et al., 2015) compared to the strong inverse relationship between physical activity and body weight observed in cross-sectional epidemiological studies. This suggests that the observational results may be affected by bias, such as reverse causality or confounding by other lifestyle or environmental factors (Schnurr et al., 2021). To date, the causal relationships between physical inactivity, sedentary time and adiposity remain unclear and warrant further investigation. It is important to assess the causal relationship with adiposity separately for physical activity and sedentary time, as individuals can be physically active even if they engaged a substantial part of their time to sedentary behaviors, and vice versa (Panahi and Tremblay, 2018).

Mendelian randomization is a powerful method to minimize the influence of reverse causality and confounding on causal estimates derived from observational data. Since genotypes are randomly allocated at conception, genetic alleles associated with physical activity, sedentary behavior, and body mass index (BMI) can be used to assign individuals according to higher or lower mean levels of these exposures in a randomized manner (Figure 1).

Figure 1
Download asset Open asset
Mendelian randomization assumptions and directional associations between physical activity, sedentary time and adiposity.

Panel A shows Mendelian randomization assumptions when estimating the causal association between a given exposure and outcome: (1) The instruments are associated with the exposure, (2) the instruments do not cause the outcome independently from the exposure (horizontal uncorrelated pleiotropy), and (3) the effects of the exposure on the outcome are not influenced by any confounders (horizontal correlated pleiotropy). Panel B indicates a unidirectional causal effect of physical activity or sedentary time on adiposity, whereas panel C indicates a unidirectional causal effect of adiposity on physical activity or sedentary time. Panel D indicates a bidirectional causal relationship where physical activity or sedentary time has a causal effect on adiposity, but at the same time, adiposity has a causal effect on physical activity or sedentary time. Figure icons were created with BioRender.com.

Here, we aimed to assess the causality between the genetic liability of physical inactivity, sedentary time and BMI by applying bidirectional Mendelian randomization analyses on summary results of accelerometer-based physical activity and sedentary time for 91,105 adults and of BMI for 806,834 adults.

Results

We used the Mendelian randomization CAUSE method to take advantage of the full genome-wide summary results (Morrison et al., 2020). We found evidence of causality between higher vigorous and moderate physical activity and lower BMI (p = 3.8 × 10–4 and p = 0.006, respectively), and between more sedentary time and higher BMI (p = 0.02) (Table 1, Figure 2, and Appendix 1—table 1 and Appendix 1—table 2). In absolute units, we approximate that each one hour daily increase in moderate physical activity or decrease in sedentary time causally decreased BMI by 0.27 kg/m2 (~0.8 kg) or 0.14 kg/m2 (~0.4 kg) (Appendix 1—table 3). The median shared effect, which reflects the effect induced by correlated horizontal pleiotropy, ranged from –0.01 to 0 for all trait pairs, indicating that there was no bias induced by horizontal pleiotropy. The low q values under the causal model (q = 0.18–0.20), which reflect the proportion of variants that show correlated horizontal pleiotropy, also suggested that horizontal pleiotropy was limited. We checked for the existence of possible outlier variants that could have a large effect on the exposure but no effect on the outcome, by producing a scatter plot for CAUSE test statistics on the causal association between BMI and sedentary time (Appendix 1—figure 1). While CAUSE is not sensitive to outliers as such, the existence of outlier variants could provide evidence against the causal model. However, we found no evidence of outlier variants. As we found high shared model q values in our CAUSE analyses (q=0.7-0.9), which may suggest that the genetic effects between the two traits are highly correlated and causality will be difficult to establish, we also assessed the genetic correlation between BMI and sedentary time. However, the genetic correlation analysis suggested only a modest genetic correlation between BMI and sedentary time (rg=0.27), which provides further reassurance that our CAUSE results are valid.

Table 1
Results for Mendelian randomization analyses using the CAUSE method.
Causal model better fit for the data
DirectionMedian causal effect (95% CI)Median q (CI)P causal vs sharing
Vigorous PA→ BMI–0.16 (-0.24,–0.08)0.19 (0, 0.86)3.8 × 10–04
Moderate PA→ BMI–0.18 (-0.3,–0.05)0.20 (0.01, 0.86)0.006
Sedentary time → BMI0.11 (0.02, 0.20)0.19 (0, 0.86)0.02
BMI → Sedentary time0.13 (0.08, 0.17)0.18 (0, 0.85)6.3 × 10–4
Sharing model better fit for the data
DirectionMedian shared effect (CI)Median q (CI)P causal vs sharing
BMI → Vigorous PA–0.16 (-0.19,–0.14)0.90 (0.77, 0.98)0.35
BMI → Moderate PA–0.14 (-0.19,–0.11)0.77 (0.55, 0.95)0.31

  1. BMI, Body mass index (BMI); SE, standard error; N SNPs, number of single-nucleotide polymorphism; P, P-value; PA, physical activity; IVW, inverse variance weighted; CI, 95% confidence interval.

  2. The results display the data according to the goodness-of-fit for the causal or the sharing model. The median q value indicates the proportion of variants with correlated pleiotropy.

Figure 2
Download asset Open asset
Causal estimates for Mendelian randomization analyses using the CAUSE, inverse-variance-weighted (IVW), weighted median, and weighted mode methods.

Median causal estimates for Mendelian randomization analyses using the CAUSE method are shown in the left panel and mean causal estimates from the inverse variance weighted (IVW), weighted median (Wmedian) and weighted mode (Wmode) methods are shown in the right panel after outlier removal and accounting for horizontal pleiotropy. A diamond (♢) in the estimate for CAUSE indicates that the sharing model fit the data better than the causal model, that is that the association between the traits was more likely to be explained by horizontal correlated pleiotropy than causality. PA, physical activity; BMI, body mass index. Figure icons were created with BioRender.com.

In the reverse direction, we found no evidence of a causal effect of BMI on vigorous physical activity (p = 0.35) or moderate physical activity (p = 0.31) using CAUSE (Table 1, Figure 2, and Appendix 1—table 1 and Appendix 1—table 2). However, we found evidence of a causal effect of BMI on more sedentary time (p = 6.3 × 10–4), indicating bidirectional causality between the traits. The median shared effect in the causal association between BMI and sedentary time was zero and the q value was 0.18, suggesting that the causal association between sedentary time and BMI was unlikely to be biased by horizontal pleiotropy. In absolute units, we approximate that each kg/m2 (~3 kg) increase in BMI was causally associated with a 3.5 min increase in sedentary time per day (Appendix 1—table 3).

We also estimated the causal effects of moderate physical activity, vigorous physical activity and sedentary time on BMI with four commonly used Mendelian randomization methods, including IVW, Egger, weighted median and weighted mode methods. Due to the low number of independent, genome-wide significant loci for vigorous physical activity, moderate physical activity and sedentary time that were present in the GWAS results for BMI, we used a less stringent threshold of p < 5 × 10–7 to identify genetic instruments for these traits, resulting in 5, 3, and 5 independent loci, respectively. The directions of causal estimates were consistent with the findings from CAUSE, but the evidence for causality was weaker (Table 2, Figure 2, Appendix 1, and Appendix 1—table 4). To estimate the causal effect of BMI on moderate physical activity, vigorous physical activity and sedentary time, we used genome-wide significant BMI loci (p < 5 × 10–8) as instruments (n = 57, n = 55 and n = 57, respectively). Again, the directions of causal estimates were consistent with the CAUSE results, but the associations were weaker (Table 2, Figure 2, Appendix 1, and Appendix 1—table 5).

Table 2
Mendelian randomization results for inverse variance weighted, weighted median, weighted mode, and MR-Egger methods.
DirectionVigorous physical activity → BMIModerate physical activity → BMISedentary time → BMI
MR methodbetaSEp-valueN SNPsbetaSEp-valueN SNPsbetaSEp-valueN SNPs
IVW–0.170.080.045–0.240.120.0530.110.080.165
Weighted median–0.180.100.085–0.180.120.1330.100.100.295
Weighted mode–0.190.120.195–0.140.150.4330.080.140.595
MR-Egger1.332.210.5950.120.450.8430.490.360.265
DirectionBMI → Vigorous physical activityBMI → Moderate physical activityBMI → Sedentary time
IVW–0.070.020.0157–0.060.030.03550.070.030.0157
Weighted mode–0.010.060.83570.020.060.6855–0.060.080.4657
Weighted median–0.040.030.27570.0030.040.93550.030.040.4857
MR-Egger0.130.070.06570.160.070.0455–0.020.080.7657
BMI, Body mass index (BMI); SE, standard error; N SNPs, number of single nucleotide polymorphism; IVW, inverse variance weighted; CI, 95% confidence interval.

  1. The results display the data according to the goodness-of-fit for the causal or the sharing model. The median q value indicates the proportion of variants with correlated pleiotropy. CI, confidence interval; PA, physical activity; BMI, body mass index; P, P-value.

The results from CAUSE analyses for BFP were consistent with those for BMI, that is there was evidence of a causal effect of higher vigorous and moderate physical activity on lower BFP (p = 1.4 × 10–6 and p = 0.004, respectively), and a causal effect of more sedentary time on higher BFP (p = 0.009) (Figure 2, Appendix 1—table 6 and Appendix 1—table 7, Appendix 1—figure 2). We found no evidence of a causal relationship between physical activity or sedentary time and measures of central adiposity, including WCadjBMI and WHRadjBMI (Figure 2, Appendix 1—table 6 and Appendix 1—table 7, Appendix 1—figure 2).

Discussion

The present Mendelian randomization analyses suggest a bidirectional causal relationship between higher sedentary time and higher BMI, implying that decreasing sedentary time is beneficial for weight management, but also that reducing adiposity may lead to additional health benefits by reducing sedentary time. The analyses also suggest there is a causal association between higher levels of physical activity and lower BMI, supporting the view that preventive programs for increasing physical activity and decreasing sedentary time are beneficial for weight management.

Based on the causal effect size in our analysis, we estimated that each 1 hr daily increase in moderate physical activity or 1 hr decrease in sedentary time was associated with a 0.27 kg/m2 (~ 0.8 kg) or 0.14 kg/m2 (~0.4 kg) decrease in BMI or body weight, respectively. Our results also suggest that each 1 kg/m2 (~3 kg) higher BMI increases daily sedentary time by ~3.5 min, but do not suggest a causal effect of BMI on physical activity (Appendix 1 and Appendix 1—table 3). Our results are well in accordance with a previous observational study that aimed to assess the bidirectional relationship between physical activity and weight change during a 10-year period (Barone Gibbs et al., 2020). Examining associations between accelerometer-based activity measures and weight change in 866 men and women, the study suggested a bidirectional relationship where higher sedentary time at baseline increased 10-year weight gain and higher baseline weight was associated with an unfavorable 10-year change in sedentary time. The effect sizes indicated that 1 hr lower sedentary time at baseline was associated with ~0.3 kg decrease in body weight over the 10-year follow up, whereas 3 kg higher baseline body weight was associated with ~2 min increase in sedentary time, closely resembling the causal estimates we observed in the present Mendelian randomization analyses. Our results are also in accordance with randomized clinical trials of physical activity interventions which generally suggest that increasing physical activity leads to a moderate loss of body weight in overweight or obese participants (Church et al., 2009; Rosenkilde et al., 2012; Golubic et al., 2015; Kim et al., 2019; Biddle et al., 2015). Resembling the causal estimates we observed, meta-analyses of randomized clinical trials have suggested weight losses ranging from 0.3 kg to 1.8 kg for various physical activity interventions ranging from 2 to 52 weeks (Twells et al., 2021). However, it is important to note that the causal estimates from Mendelian randomization are not fully comparable to those from randomized clinical trials, because they represent lifelong effects rather than effects lasting a defined length of an intervention, and furthermore, physical activity interventions may operate on body weight through other pathways than those affected by the genotypes. The causal effect of higher adiposity on sedentary time has not been to date assessed in randomized clinical trials, likely due to the ethical and practical limitations of performing such a study.

In a previous Mendelian randomization analysis of adult populations, evidence for a causal, bidirectional relationship between overall activity levels and higher BMI was observed using the maximum likelihood method, but the results showed evidence of horizontal pleiotropy that could not be fully accounted for and the role of activity intensity level remained unclear (Doherty et al., 2018). Here, using a method that takes advantage of full genome-wide summary results and corrects for sample overlap between the exposure and the outcome traits to maximize statistical power and correct for pleiotropy, we showed that the causal bidirectional relationship is particularly evident for the relationship between sedentary time and adiposity. Our results may also be compared with two independent one-sample Mendelian randomization studies performed in children (Richmond et al., 2014; Schnurr et al., 2018). The first study, including 4296 children at 11 years of age from the United Kingdom, indicated a causal association between higher BMI and lower accelerometer-based moderate and moderate-to-vigorous physical activity and more sedentary time (Richmond et al., 2014). The second study, including 679 children at age 3–8 years from Denmark, also indicated that higher BMI is causal for higher accelerometer-based sedentary time, but did not find a causality for moderate or moderate-to-vigorous physical activity (Schnurr et al., 2018). Consistent with the latter study of children, our results indicate a causal effect of BMI on sedentary behavior, but not on physical activity, in adults. The differences between studies could be due to different applied methods, or methodological limitations, such as weak instrument bias when smaller sample sizes are used, which may lead to estimated causal effects towards the observational association. One could also expect differences between children and adults given the distinct patterns by which they engage in physical activity. For example, while physical activity in adults consists of commuting, occupational and structured leisure-time activities, children primarily engage in spontaneous, play-oriented activities. Higher BMI leads to higher perceived exertion during physical activity (Groslambert and Mahon, 2006), which could reduce the natural inclination of children to engage in play-oriented activities, whereas adults exert more conscious control over their daily activities.

The strengths of the present studies include the use of genome-wide summary results for objectively measured physical activity and sedentary time, which avoided misreporting bias evident for self-reported measures, as well as the use of newly developed Mendelian randomization method that utilizes full genome-wide summary results to improve statistical power, correct for sample overlap, and assess horizontal pleiotropy, successfully applied in recent Mendelian randomization studies (Jäger et al., 2020; Mitchell et al., 2020). The limitations are that we cannot exclude other sources of bias in the measurement of physical activity and sedentary time that could influence the observed causal estimates, including the observer effect and the limited 7-day period of the measurement, which may not be representative of long-term activity habits. Furthermore, even if we used the largest available data on objectively measured physical activity, the statistical power was limited, as very few genome-wide significant loci have thus far been identified. When larger sample sizes for accelerometer-based physical activity become available, the results should be replicated. The present study is also limited in the fact that the findings are not generalizable across different age-groups or populations. Moreover, further research is needed to investigate causal relationships at various BMI thresholds.

In conclusion, the present Mendelian randomization analyses indicate a bidirectional causal relationship between higher sedentary time and higher BMI. Thus, decreasing sedentary time is likely to be beneficial for weight management, but reducing adiposity may also lead to additional health benefits by reducing sedentary time. Our analyses also suggest that there is a causal association between higher levels of physical activity and lower BMI, supporting the view that lifelong preventive programs for increasing physical activity and decreasing sedentary time are beneficial for weight management.

Materials and methods

Data sources and populations

Request a detailed protocol

We used summary results from the largest published genome-wide association studies (GWAS) of objectively assessed physical activity, sedentary behavior, and BMI in individuals of European ancestry. The physical activity GWAS included up to 91,105 individuals for accelerometer-based vigorous physical activity, moderate physical activity, or sedentary time from the UK Biobank (Klimentidis et al., 2018; Doherty et al., 2018). In these studies, accelerometer was worn continuously for at least 72 hr and up to 7 days. Vigorous physical activity was defined as the fraction of accelerations > 425 milli-gravities, and moderate physical activity was predicted using a machine-learning method for moderate intensity activity time (Doherty et al., 2018). Sedentary time was defined as the time spent in activities with metabolic equivalent of task (MET) ≤1.5 during sitting, lying, or in reclining posture, except for driving and certain non-desk work instances where MET ≤2.5 was applied (Doherty et al., 2018). For BMI, we utilized GWAS results from a meta-analysis of the Genetic Investigation of Anthropometric Traits (GIANT Consortium) and the UK Biobank data, including altogether 806,834 individuals of European ancestry (Pulit et al., 2019). For Mendelian randomization analyses using the inverse variance-weighted (IVW), weighted median, weighted mode, and MR-Egger regression methods, we used only the GIANT Consortium BMI meta-analysis data of 339,224 individuals without the UK Biobank data to avoid sample overlap between the exposure and outcome traits as these methods are sensitive to bias from overlapping samples (Locke et al., 2015).

In addition to BMI, we assessed causal effects of physical activity and sedentary time on body fat percentage (BFP) and on two measures of central adiposity, including waist circumference and waist-hip ratio adjusted for BMI (WCadjBMI and WHRadjBMI, respectively). GWAS results for BFP were available from an analysis of 454,633 participants of European ancestry in the UK Biobank (Elsworth et al., 2020). GWAS results for WCadjBMI and WHRadjBMI were available from GIANT Consortium meta-analyses of 231,355 and 694,649 individuals of European ancestry, respectively (Justice et al., 2017; Pulit et al., 2019).

Mendelian randomization using full genome-wide summary results for the exposure trait

Request a detailed protocol

Only few genetic loci have been found to be associated with accelerometer-based moderate physical activity (n = 2), vigorous physical activity (n = 1) or sedentary time (n = 4) at genome-wide significance (p < 5 × 10–8) (Klimentidis et al., 2018; Doherty et al., 2018), and the loci thus provide a limited power to study causal associations with BMI using Mendelian randomization. The recently published Causal Analysis Using Summary Effect Estimates (CAUSE) Mendelian randomization method (Morrison et al., 2020) improves statistical power in such cases, by utilizing full genome-wide summary results instead of genome-wide significant loci only. Furthermore, the CAUSE method is able to correct for sample overlap between the exposure and the outcome trait, which allows using the largest sample sizes available for both traits. CAUSE has also been found to be less prone to identify false positive associations compared to other commonly used Mendelian randomization methods (Burgess et al., 2019; Morrison et al., 2020).

The CAUSE method calculates the posterior probabilities of the causal effect and the shared (non-causal) effect, where the causal effect reflects the effect of the variants on the outcome trait through the exposure and the shared effect reflects correlated horizontal pleiotropy (Figure 1), that is the effect of the variants on the outcome through confounders. The distinction between a causal effect and correlated horizontal pleiotropy follows the assumption that a causal effect leads to non-zero genetic correlation between the exposure and the outcome where the correlation is driven by all variants associated with the exposure. If only a subset of variants contributes to the genetic correlation between the exposure and the outcome, it is considered the result of correlated horizontal pleiotropy. The CAUSE method also provides an estimate of the proportion of variants that are likely to show correlated horizontal pleiotropy, the q value.

We used the CAUSE settings and procedures originally recommended by the authors (Morrison et al., 2020), with the exception of q priors that were set to fit the strictest model possible (q_alpha = 1 and q_beta = 2) in order to avoid false positive findings. A thorough explanation of the steps used to perform CAUSE analysis is included in the supplementary text (Appendix 1).

Mendelian randomization using genome-wide significant loci for the exposure trait

Request a detailed protocol

In addition to the CAUSE method that implements Mendelian randomization analyses using full genome-wide summary results for the exposure trait, we implemented four commonly used Mendelian randomization methods that utilize genome-wide significant loci only: the IVW, MR-Egger, weighted median and weighted mode methods (Appendix 1). We performed sensitivity analyses using Steiger filtering to remove variants that showed stronger association with the outcome than the exposure trait and that were thus not considered suitable as instruments for the exposure trait. To create the genetic instrument for the exposure trait, we only included the lead variants that showed genome-wide significant associations with the trait (p < 5 × 10–8) and with a pairwise linkage disequilibrium (LD) r2 <0.001 with their neighboring variants, in a window of 10,000 kb. Variants that were not available in the outcome trait GWAS were substituted by their LD proxies (r2 >0.8). Palindromic variants (A/T, G/C) were excluded. If less than three genetic variants were identified with these parameters, we used a less stringent p-value threshold of p < 5 × 10–7 to identify enough genetic instruments. In order to assess the strength of the genetic instrument, we obtained F-statistics for each trait. The analyses were performed using the TwoSampleMR package in R and are described in detail in the Appendix 1 (Hemani et al., 2018).

We estimated heterogeneity across the causal estimates of the SNPs using the Meta R package (Schwarzer et al., 2015). The causal estimates were considered heterogeneous if the p value for Cochran’s Q test was significantly different from zero (p < 0.05) and I2 was above 0.25. We assessed bias introduced by horizontal pleiotropy by implementing the Egger’s intercept test using the TwoSampleMR package in R (Hemani et al., 2018). An Egger’s intercept that deviated significantly from zero (p < 0.05) was considered as evidence of horizontal pleiotropy. We used the Rucker framework (Bowden et al., 2018) to assess whether Egger regression that accounts for horizontal pleiotropy but limits statistical power should be applied instead of the standard IVW model. To visually assess heterogeneity and horizontal pleiotropy, we observed forest plots and funnel plots (Figure 2, Appendix 1—figure 3). To detect individual pleiotropic variants that might bias the results, we applied the RadialMR package in R using an iterative Cochran’s Q method and setting a strict outlier P value threshold of <0.05 (Bowden et al., 2018). The iterative Cochran’s Q, either IVW’s Q or Egger’s Q was chosen depending on Rucker framework results. After removing outlier variants detected with RadialMR, we re-run the Mendelian randomization and sensitivity tests and plots, to make sure that the variants introducing horizontal pleiotropy (Figure 1) had been removed. The analysis plan for this study is described in the supplementary text (Appendix 2).

The CAUSE method’s median posterior probability of the causal effect cannot be easily transformed to absolute units. To convert the causal estimates to absolute units, we calculated a causal effect with weighted median method using independent variants identified in CAUSE that were not removed by the outlier extraction protocol described above, in order to mimic CAUSE control for correlated and uncorrelated pleiotropy (Appendix 1).

Appendix 1

Mendelian randomization using the CAUSE, IVW, Egger, weighted median, and weighted mode methods

Mendelian randomization using the CAUSE method

The CAUSE method performs Mendelian randomization analyses following six different steps, as described in the CAUSE online tutorial (https://jean997.github.io/cause/ldl_cad.html) and the original publication (Morrison et al., 2020). The steps are the following:

  1. Installing the following versions of these three packages were used: CAUSE v1.0.0, mixsqp v.0.1–97 and ashr v.2.2–32.

  2. Filtering data by including variants with imputation quality score INFO >0.7 and minimum allele frequency (MAF) >0.01.

  3. Excluding variants from the Major Histocompatibility Complex (MHC) present in chromosome six between the base pairs 26 M and 34 M in build 37.

  4. Merging the exposure and outcome GWAS summary level data. Gwas_merge function from CAUSE package was used to identify the variants present in exposure and outcome summary statistics data and to align exposure and the outcome effect sizes to the same allele

  5. Calculating nuisance parameters to correct for sample overlap between exposure and outcome GWAS.

  6. Using 1000 Genomes CEU population data after using LDshrink as reference panel to clump the variants.

  7. Setting the priors for the three model parameters – causal effect, shared effect and q – and calculating their posterior probabilities. For the causal effect and the shared effect, their priors are set automatically to 0, while for q the software allows the user to set the thresholds for the priors. In this case, q priors are set to qalpha = 1 and qbeta = 2.

  8. Calculating two models to fit the posterior probabilities: the sharing model, where the causal effect is set to 0, and the causal model, where the posterior probability for the causal effect is calculated.

  9. Comparing the two models, sharing and causal, against the null and against each other with expected log pointwise posterior density (ELPD) method to identify which model is the most fitting for the data.

Mendelian randomization using the IVW, Egger, weighted median, and weighted mode methods

We used the following parameters to interpret the findings:

  1. Since the version of TwoSampleMR used in the analysis v0.5.4 removes duplicates by excluding the second instance when introducing data locally, only the variant that presented the same alleles as in the outcome data and with the lowest p-value were kept. In none of the combinations of traits, variants in the MHC were found.

  2. To clump variants, the function ld_clump_local from the package ieugwasr v0.1.5 (https://github.com/MRCIEU/ieugwasr) was used using the updated European 1000 Genomes reference panel available in https://github.com/mrcieu/gwasglue.

  3. Only variants with MAF >0.01 and INFO > 0.7 were included in the analysis.

  4. Only variants that passed the Steiger filtering using the function steiger_filtering from the package TwoSampleMR were used.

  5. The no measurement error (NOME) assumption, which assumes no measurement error between the genetic instrument and the exposure, was checked by calculating the mean F-statistic and the variation of the I2 statistic (Bowden et al., 2019). The latter can be calculated with the Isq function from TwoSampleMR package.

  6. The causal estimates were considered heterogeneous if the P value for Cochran’s Q test was <0.05 and I2 was >0.25. Both estimates were calculated using the meta package.

  7. An Egger’s intercept p-value < 0.05 was considered as evidence of horizontal pleiotropy.

  8. To assess whether MR-Egger regression should be applied instead of the standard IVW model, we used the Rucker framework test.

  9. To detect individual pleiotropic variants, we used RadialMR’s iterative Cochran’s Q method following a p value threshold < 0.05. RadialMR presents two functions: ivw_radial or egger_radial, depending on the Cochran’s Q, either IVW’s Q or Egger’s Q’, used. The function used was chosen depending on Rucker framework test’s result. If Rucker framework presented contradictory results, an iterative version of the Rucker framework (i.e. rucker_jackknife from TwoSampleMR package) was used to assess whether IVW was still chosen as the main model.

To visualize the effects of heterogeneity and horizontal pleiotropy on the results, we generated leave-one-out forest plots and funnel plots. After removing outlier variants detected with RadialMR, we re-run the Mendelian randomization methods and sensitivity tests (Appendix 1—table 4 and Appendix 1—table 5) and re-generated the plots (Appendix 1—figure 3), to make sure that the variants introducing horizontal pleiotropy had been removed. The F-statistic ranged between 286–6179, indicating that the instrumental variables are strong (F statistic >10) and weak instrument bias may not be an issue. Furthermore, we tested whether the amount of pleiotropy was independent of instrument strength by calculating a variation of the I2 (Bowden et al., 2019). We found it to be above 0.90 in all cases. Below we describe and interpret the findings for each combination of traits.

Vigorous physical activity → BMI

All MR methods except Egger showed negative causal effects, from which only IVW is significant (p < 0.05) (Appendix 1—table 4). No outlier extraction was performed since Cochran's Q test was not significant (p = 0.80) and low I2 (0.0%) showed no evidence of heterogeneity. Egger’s intercept was not significant (p = 0.30) and Rucker test indicated that IVW model is more fitting for the data (Appendix 1—table 4). Consistent with the sensitivity tests, leave-one-out forest plot (Appendix 1—figure 3, panel a) showed that the IVW causal effect did not strongly change after the removal of any of the five independent variants used to calculate the causal effects. Funnel plot (Appendix 1—figure 3, panel a) showed no signs of asymmetry, which is consistent with Egger’s intercept result. Thus, these findings indicate that the causal effects are not biased and they reflect weak evidence of causality between an increase of vigorous physical activity and a decrease of BMI. While IVW negative causal effect is significantly different from 0, the evidence for causality is considered weak due to (1) IVW’s p-value is still close to nominal threshold (p = 0.04), (2) the other methods with negative causal effects are not significantly different from 0 (p > 0.05) and (3) Egger’s regression presents a positive causal effect. These weak results are most probably due to the low number of variants used (Appendix 1—table 4).

Moderate physical activity → BMI

All MR methods except Egger showed negative effects directions, with only IVW’s being significant (Appendix 1—table 4). No outlier extraction was performed since Cochran's Q test was not significant (p = 0.30) and I2 was below 25% (18.9%). Egger’s intercept was not significantly different from 0 (p = 0.16) indicating that no pleiotropic variants were found among the four independent variants used to calculate the causal effects. Rucker test indicated that IVW was a better fit for the model than Egger, in line with Egger’s intercept results (Appendix 1—table 4). The leave-one-out forest sensitivity plot (Appendix 1—figure 3, panel c) showed that no variant presented heterogenic causal effect since the IVW causal effect did not strongly diverge after removing any of the variants. The funnel sensitivity plot (Appendix 1—figure 3, panel c) showed no asymmetry. All in all, the findings indicate that the causal effects are not biased, but that they suggest causality between increased moderate physical activity and decreased BMI. The reason behind this is that (1) IVW’s negative causal effect present a p-value close to the nominal threshold (p = 0.02), (2) the other methods present negative causal effect that are not significantly different from 0 (p > 0.05) and (3) Egger’s regression presents a positive causal effect, although it is not significantly different from 0 (Appendix 1—table 4).

Sedentary time → BMI

Nine independent variants were used as instruments, and two were removed after outlier extraction as indicated by RadialMR. All methods were not significant, and all reported a positive causal effect (p > 0.05), except weighted mode (Appendix 1—table 4). After outlier extraction, Cochran’s Q test was non-significant (p = 0.15), but the I2 was 36.4%, indicating slight heterogeneity. Egger’s intercept was not significant (p = 0.43) and Rucker test indicated that IVW method is a better fit for the data than Egger, implying that pleiotropic variants were not present (Appendix 1—table 4). In accordance, the forest plot and funnel plot (Appendix 1—figure 3, panel e) showed that the results are not strongly affected by horizontal pleiotropy. The forest plot indicated that the heterogeneity comes from variants with both positive and negative causal effects, since the IVW causal effect becomes either more strongly negative or positive when the variants are removed one at a time. The funnel plot (Appendix 1—figure 3, panel e) showed that the causal estimates are evenly distributed, implying no asymmetry and hence no pleiotropy. To conclude, the findings indicate a non-significant positive causal effect between sedentary time and BMI.

BMI → vigorous physical activity

All MR methods except Egger presented negative causal estimates. The causal estimate from the IVW method reached statistical significance (p < 0.01) (Appendix 1—table 5). Non-significant p value from Cochran’s Q test (p = 0.27) and low I2 of 9.6% indicated no heterogeneity. Egger intercept’s was significant (p = 0.0017) and IVW’s Q and Egger’s Q were significantly different (Q-Q’ = 9.83, p = 0.0017), indicating that horizontal pleiotropy may still affect the causal estimates. Rucker framework selected IVW more fitting for the data than Egger (Appendix 1—table 5). Forest and funnel plots (Appendix 1—figure 3, panel b) indicated that some variants, both with positive and negative effects, may be introducing heterogeneity. The forest plot showed slight deviations from the mean causal effect when extreme variants were extracted one at a time. In the funnel plot, symmetry was disrupted by variants with positive causal effects and small standard errors. Considering both plots, we infer that heterogeneity and horizontal pleiotropy were introduced by specific variants with small standard errors. Leave-one-out plots indicated that two variants (rs13021737 and rs6567160) introduced horizontal pleiotropy, showing positive causal effects and having the smallest standard errors of all variants. To conclude, we found residual pleiotropy that RadialMR could not properly address and deem that the association between BMI and vigorous physical activity is unlikely to be causal.

BMI → moderate physical activity

The IVW method showed a negative causal effect, whereas Egger, weighted median and weighted mode showed a positive causal effect (Appendix 1—table 5). Only the causal effect from Egger was significant (p = 0.04). Non-significant p from Cochran’s Q test (p = 0.91) and I2 of 0% indicated no presence of heterogeneity. Egger intercept’s was significant (P = 0.002) and IVW’s Q and Egger’s Q were significantly different (Q-Q’ = 9.60, p = 1.95 × 10–5), indicating that horizontal pleiotropy may still affect the causal estimates. Rucker framework chose IVW as the most fitting method for the data. Forest and funnel plots (Appendix 1—figure 3, panel d) indicated heterogeneity. The leave-one-out forest plots indicated deviation from the mean causal effect when variants with positive causal effects were excluded, and the funnel plots showed that asymmetry was introduced by variants with positive causal effects and small standard errors. These findings were in agreement with the findings from Rucker framework and Egger’s intercept test results. To conclude, we found evidence of horizontal pleiotropy that RadialMR could not properly remove, and where pleiotropy was introduced by variants with positive causal estimates. The presence of horizontal pleiotropy may explain the negative direction of the causal estimate in the IVW method, whereas other methods showed positive causal estimates (Appendix 1—table 5). We deem that the association between BMI and moderate physical activity is unlikely to be causal.

BMI → sedentary time

The IVW and weighted median methods showed positive causal effects, while Egger and weighted mode presented negative causal effects (Appendix 1—table 5), with only IVW being significant (p = 0.02). The Cochran’s Q test (p = 0.4) and I2 of 2.2% indicated no presence of heterogeneity. Egger intercept test (p = 0.20) and the non-significant difference between IVW’s Q and Egger’s Q (Q-Q’ = 1.62, p = 0.20) indicated no presence of horizontal pleiotropy and that the IVW method was a better fit for the data than Egger (Appendix 1—table 5). Forest and funnel sensitivity plots (Appendix 1—figure 3, panel f) indicated that there was no heterogeneity, except for one variant (rs6567160) with a negative effect. In the forest plot, upper and lower extreme variants induced deviations from the mean causal effect when extracted one at a time. Taken together, sensitivity tests indicate no horizontal pleiotropy and forest and funnel plots indicate the presence of heterogeneity for one variant, while RadialMR could not detect any outliers. To conclude, as the forest and funnel plots indicated heterogeneity, implying that there was residual horizontal pleiotropy, a causal relationship between BMI and sedentary time cannot be deemed, despite the significant p value in the IVW method.

Conversion of the causal estimates to absolute units

To interpret the causal estimates from CAUSE, we applied the weighted median method on the independent instrumental variants selected by the CAUSE method and performed outlier extraction and sensitivity tests, following the steps described below. The results are reported in Appendix 1—table 3.

  1. We selected independent variants associated with each exposure identified with the CAUSE method.

  2. We performed outlier extraction using RadialMR to control for uncorrelated pleiotropy.

  3. We obtained weighted median’s causal estimates, to correct for correlated pleiotropy.

  4. We approximated the causal estimates between outcome and exposure in absolute units. As the genome-wide summary results were reported in standard deviation units, we multiplied the causal estimates by the standard deviation of the corresponding non-transformed trait to derive the estimates in original trait units. The equations used in these calculations are described for each combination of traits in Appendix 1—table 3.

Appendix 1—figure 1
Download asset Open asset
Scatter plot for CAUSE test statistics on the causal association between BMI and sedentary time.

Effect estimates for body mass index (x-axis) are plotted against estimates for sedentary time (y-axis). Error bars indicate 95% confidence intervals and the dotted line indicates the median causal effect of BMI on sedentary time.

Appendix 1—figure 2
Download asset Open asset
Estimates from the CAUSE method for the causal relationship between physical activity or sedentary time and waist-hip ratio, waist circumference, or body fat percentage.

Median causal estimates for Mendelian randomization analyses using the CAUSE method. A diamond (♢) in the estimate for CAUSE indicates that the sharing model fit the data better than the causal model, that is that the association between the traits was more likely to be explained by horizontal correlated pleiotropy than causality. PA, physical activity; WHRadjBMI, waist-to-hip ratio adjusted for body mass index; WCadjBMI, waist circumference adjusted for body mass index; BFP, body fat percentage. Figure icons were created with BioRender.com.

Appendix 1—figure 3
Download asset Open asset
Leave-one-out forest and funnel sensitivity plots after outlier extraction.
Appendix 1—table 1
CAUSE expected log pointwise posterior density (ELPD) results for each combination of traits.
Vigorous physical activity → BMIBMI → vigorous physical activity
Model 1Model 2Delta_ELPDSE_delta_ELPDzDelta_ELPDSE_delta_ELPDZ
NullSharing–17.474.24–4.12–251.8821.54–11.69
NullCausal–18.984.68–4.05–252.2221.65–11.65
SharingCausal–1.510.45–3.36–0.350.90–0.38
Moderate physical activity → BMIBMI → Moderate physical activity
NullSharing–8.502.86–2.97–119.3915.15–7.88
NullCausal–9.833.39–2.90–119.7715.34–7.81
SharingCausal–1.340.53–2.52–0.380.77–0.49
Sedentary time → BMIBMI → Sedentary time
NullSharing–6.512.57–2.53–137.84157.46–8.75
NullCausal–7.673.14–2.44–138.8915.88–8.75
SharingCausal–1.160.57–2.05–1.060.33–3.22

  1. BMI, Body mass index; SE, standard error; NA, not applicable; CI, 95% confidence interval.

Appendix 1—table 2
CAUSE posterior probabilities and q values for the causal effect and the shared effect.
ModelCausal effect (CI)Shared effect (CI)q (CI)Sharing vs causal model p-value
Vigorous physical activity → BMI
SharingNA–0.18 (-0.25,–0.12)0.85 (0.57, 0.98)3.80E-04
Causal–0.16 (-0.24,–0.08)0 (-0.31, 0.31)0.19 (0.00, 0.86)
Moderate physical activity → BMI
SharingNA–0.2 (-0.33,–0.11)0.75 (0.35, 0.96)5.80E-03
Causal–0.18 (-0.3,–0.05)–0.01 (-0.44, 0.39)0.2 (0.01, 0.86)
Sedentary time → BMI
SharingNA0.13 (0.06, 0.23)0.69 (0.25, 0.95)2.00E-02
Causal0.11 (0.02, 0.20)0 (-0.32, 0.32)0.19 (0, 0.86)
BMI → Vigorous physical activity
SharingNA–0.16 (-0.19,–0.14)0.9 (0.77, 0.98)0.35
Causal–0.15 (-0.2,–0.08)0.01 (-0.18, 0.21)0.21 (0.01, 0.83)
BMI → Moderate physical activity
SharingNA–0.14 (-0.19,–0.11)0.77 (0.55, 0.95)0.31
Causal–0.11 (-0.19,–0.03)0 (-0.28, 0.2)0.23 (0.01, 0.84)
BMI → Sedentary time
SharingNA0.14 (0.12, 0.17)0.9 (0.73, 0.98)6.30E-04
Causal0.13 (0.08, 0.17)0 (-0.23, 0.26)0.18 (0.00, 0.85)

  1. BMI, Body mass index; NA, not applicable; CI, 95% confidence interval.

Appendix 1—table 3
Approximation of the causal estimates in absolute units.
Original causal estimates (weighted median)Conversion of causal estimatesConverted causal estimates
AnalysisN SNPsBetaSEpEquationSDexposureSDoutcome 
Vigorous physical activity → BMI1,036–0·09 SD0·006 SD9.28E-57(Beta*SDoutcome)/
SDexposure /
(0,01*24 hours)		NA4·807 kg/m2–0·48 kg/m2 per hour of moderate physical activity per day
Moderate physical activity → BMI914–0·05 SD0·005 SD1.21E-20(Beta*SDoutcome)/
SDexposure /
(0,01*24 hours)		3.76%4·807 kg/m2–0·27 kg/m2 per hour of moderate physical activity per day
Sedentary time → BMI1,0360·06 SD0·005 SD6.33E-30(Beta*SDoutcome)/
SDexposure /
(0,01*24 hours)		7.38%4·807 kg/m20,14 kg/m2 per hour of sedentary time per day
BMI → Sedentary time11,8490·16 SD0·009 SD2.87E-68(Beta*SDoutcome)/
SDexposure *(0·01*24 hours)		4·807 kg/m27.38%3·54 min of sedentary time per kg/m2 per day
Original units of the traits:Moderate physical activity: Probability of moderate physical activity in a 30s-epoch frame (Doherty et al., 2018)
Sedentary time: Probability of sedentary time in a 30s-epoch frame (Doherty et al., 2018)
BMI: kg/m2 (Pulit et al., 2019)
Note: For vigorous physical activity, SD exposure could not be found, thus we use the same as for moderate physical activity, i.e. 3·76%

  1. SNP, single nucleotide polymorphism; BMI, SE, standard error; SD, standard deviation; Body mass index (BMI).

Appendix 1—table 4
Mendelian randomization results using the IVW, Egger, weighted median and weighted mode methods of for the causal effect of moderate or vigorous physical activity or sedentary time on BMI.
 Before outlier extraction
 Vigorous physical activity → BMIModerate physical activity → BMISedentary time → BMI
MR methodSNPbetaSEp-valueSNPbetaSEp-valueSNPbetaSEp-value
Egger51.332.210.5930.120.450.8480.460.640.50
Weighted median5–0.180.100.083–0.180.120.1380.020.090.82
IVW5–0.170.080.043–0.240.120.0580.040.110.72
Weighted mode5–0.190.120.193–0.140.150.438–0.050.180.78
Sensitivity testEstimatep-value/CIEstimatep-value/CIEstimatep-value/CI
Q (p-value)1.650.83.60.1723.081.70E-03
I2 (CI)0.00%(0,0%; 49,5%)44.30%(0,0%; 83,4%)69.70%(36,8%; 85,4%)
Q-Q' (p-value)0.460.51.750.191.960.16
Egger intercept (p-value)–0.040.29–0.010.19–0.010.75
Rucker TestIVWIVWIVW
 After outlier extraction
 Vigorous physical activity → BMIModerate physical activity → BMISedentary time → BMI
MR methodSNPbetaSEp-valueSNPbetaSEp-valueSNPbetaSEp-value
EggerNANANANANANANANA50.490.360.26
Weighted medianNANANANANANANANA50.100.100.29
IVWNANANANANANANANA50.110.080.16
Weighted modeNANANANANANANANA50.080.140.59
Sensitivity testEstimatep-value/CIEstimatep-value/CIEstimatep-value/CI
Q (p-value)NANANANA4.70.32
I2 (CI)NANANANA15(0·0%; 82·3%)
Q-Q' (p-value)NANANANA1.410.24
Egger intercept (p-value)NANANANA–1.00E-020.24
Rucker TestNANAIVW

  1. BMI, Body mass index (BMI); MR, Mendelian randomization; SNP, single nucleotide polymorphism; SE, standard error; NA, not applicable; IVW, inverse variance weighted; CI, 95% confidence interval. NA: not applicable. Note: No outlier extraction was performed for moderate physical activity → BMI and vigorous physical activity → BMI directions, since Q test and I2 did not indicate heterogeneity.

Appendix 1—table 5
Mendelian randomization results using the IVW, Egger, weighted median and weighted mode methods of BMI on moderate PA, vigorous PA, or sedentary time.
Before outlier extraction
 BMI → Vigorous physical activityBMI → Moderate physical activityBMI → Sedentary time
MR methodSNPbetaSEp-valueSNPbetaSEp-valueSNPbetaSEp-value
Egger640.090.080.22650.170.080.0465–0.100.090.28
Weighted median64–0.040.030.21650.0040.040.92650.030.040.44
IVW64–0.080.030.00565–0.050.030.09650.070.030.02
Weighted mode64–0.010.070.86650.020.050.7265–0.050.100.59
Sensitivity testEstimatep-value/CIEstimatep-value/CIEstimatep-value/CI
Q (p-value)92.49.30E-0383.730.0597.54.40E-03
I2 (CI)31.80%6·9%; 50·00%23.60%(0·00%; 44·20%)34.40%10·80%; 51·70%
Q-Q' (p-value)8.363.80E-0310.869.81E-046.591.00E-02
Egger intercept (p-value)–4.99E-039.90E-01–6.40E-039.81E-044.96E-032.00E-02
Rucker testEggerEggerEgger
After outlier extraction
 BMI → Vigorous physical activityBMI → Moderate physical activityBMI → Sedentary time
MR methodSNPbetaSEp-valueSNPbetaSEp-valueSNPbetaSEp-value
Egger570.130.070.06550.160.070.0457–0.020.080.76
Weighted median57–0.040.030.27552.95E-030.040.93570.030.040.48
IVW57–0.070.020.0155–0.060.030.03570.070.030.01
Weighted mode57–0.010.060.83550.020.060.6857–0.060.080.46
Sensitivity testEstimatep-value/CIEstimatep-value/CIEstimatep-value/CI
Q (p-value)61.880.2740.510.9156.290.46
I2 (CI)9.50%0·00%–35·10%0.00%(0·0%; 8·9%)0.50%0·0%; 31·50%
Q-Q' (p-value)9.791.75E-039.61.94E-031.622.00E-01
Egger intercept–5.70E-031.75E-03–6.44E-039.15E-052.60E-032.00E-01
Rucker testIVWIVWIVW

  1. BMI, Body mass index; MR, Mendelian randomization; SNP, single nucleotide polymorphism; SE, standard error; NA, not applicable; IVW, inverse variance weighted; CI, 95% confidence interval.

Appendix 1—table 6
CAUSE expected log pointwise posterior density (ELPD) results for central adiposity and fat percentage.
Vigorous physical activity → BFPBFP → vigorous physical activity
Model 1Model 2Delta_ELPDSE_delta_ELPDzDelta_ELPDSE_delta_ELPDz
NullSharing–23.554.67–5.04–360.4824.46–14.74
NullCausal–25.115.00–5.02–361.7724.49–14.77
SharingCausal–1.560.33–4.68–1.300.17–7.77
  Moderate physical activity → BFPBFP → Moderate physical activity
NullSharing–8.712.60–3.34–170.4617.46–9.76
NullCausal–10.113.13–3.23–171.5017.63–9.73
SharingCausal–1.400.53–2.65–1.030.54–1.90
  Sedentary time → BFPBFP → Sedentary time
NullSharing–9.743.30–2.95–143.0415.86–9.02
NullCausal–11.013.83–2.87–144.2516.13–8.94
SharingCausal–1.270.54–2.38–1.210.36–3.41
  
  Vigorous physical activity → WCadjBMIWCadjBMI → Vigorous physical activity
  Delta_ELPDSE_delta_ELPDzDelta_ELPDSE_delta_ELPDz
NullSharing0.570.077.71–6.373.13–2.03
NullCausal1.360.159.28–7.163.65–1.96
SharingCausal0.790.0710.60–0.790.54–1.46
  Moderate physical activity → WCadjBMIWCadjBMI → Moderate physical activity
NullSharing0.500.560.89–4.832.96–1.63
NullCausal1.010.871.16–5.403.58–1.51
SharingCausal0.510.361.44–0.560.65–0.87
  Sedentary time → WCadjBMIWCadjBMI → Sedentary time
NullSharing0.550.077.56–7.593.34–2.27
NullCausal1.390.168.55–8.613.97–2.17
SharingCausal0.830.099.21–1.020.65–1.56
Vigorous physical activity → WHRadjBMIWHRadjBMI → Vigorous physical activity
Delta_ELPDSE_delta_ELPDzDelta_ELPDSE_delta_ELPDz
NullSharing–0.911.19–0.770.060.840.07
NullCausal–1.682.10–0.800.311.520.21
SharingCausal–0.770.91–0.850.260.680.37
  Moderate physical activity → WHRadjBMIWHRadjBMI → Moderate physical activity
NullSharing0.480.163.000.220.770.28
NullCausal0.920.561.630.441.440.30
SharingCausal0.440.411.090.220.670.32
  Sedentary time → WHRadjBMIWHRadjBMI → Sedentary time
NullSharing0.570.115.29–6.723.44–1.95
NullCausal1.370.284.83–7.314.03–1.81
SharingCausal0.800.184.48–0.590.61–0.97

  1. BFP, body fat percentage; WCadjBMI, waist cirumference adjusted for body mass index; WHRadjBMI, waist-to-hip ratio adjusted for body mass index, SE, standard error; NA, not applicable; CI, 95% confidence interval.

Appendix 1—table 7
CAUSE posterior probabilities and q values for the causal and sharing models for central adiposity and body fat percentage.
ModelCausal Effect (CI)Shared effect (CI)Q (CI)Sharing vs causal model p-value
 Vigorous physical activity → BFP
SharingNA–0.18 (-0.23,–0.13)0.89 (0.67, 0.98)1.4e-06
Causal–0.17 (-0.24,–0.09)0 (-0.28, 0.26)0.19 (0, 0.86)
 Moderate physical activity → BFP
SharingNA–0.17 (-0.26,–0.09)0.78 (0.39, 0.97)0.0041
Causal–0.15 (-0.24,–0.05)0 (-0.32, 0.31)0.2 (0, 0.86)
 Sedentary time → BFP
SharingNA0.11 (0.06, 0.17)0.75 (0.37, 0.96)0.0088
Causal0.09 (0.03, 0.15)0 (-0.26, 0.27)0.19 (0, 0.86)
 BFP → Vigorous physical activity
SharingNA–0.27 (-0.29,–0.25)0.97 (0.92, 1)3.9e-15
Causal–0.26 (-0.31,–0.22)0 (-0.26, 0.22)0.18 (0, 0.85)
 BFP → Moderate physical activity
SharingNA–0.21 (-0.25,–0.18)0.91 (0.77, 0.99)0.029
Causal–0.19 (-0.26,–0.12)0 (-0.3, 0.31)0.19 (0, 0.86)
 BFP → Sedentary time
SharingNA0.2 (0.17, 0.24)0.91 (0.76, 0.99)0.00032
Causal0.19 (0.12, 0.25)0 (-0.33, 0.31)0.18 (0, 0.86)
 
 Vigorous physical activity → WCadjBMI
SharingNA–0.01 (-0.62, 0.59)0.12 (0, 0.72)1
Causal0 (-0.15, 0.14)0 (-0.55, 0.52)0.2 (0.01, 0.86)
 Moderate physical activity → WCadjBMI
SharingNA–0.14 (-0.98, 0.6)0.13 (0.01, 0.69)9.20E-01
Causal–0.03 (-0.36, 0.27)–0.03 (-0.83, 0.7)0.26 (0.01, 0.84)
 Sedentary time → WCadjBMI
SharingNA0.01 (-0.44, 0.57)0.12 (0, 0.71)1
Causal0 (-0.13, 0.14)0.01 (-0.41, 0.52)0.21 (0.01, 0.85)
 WCadjBMI → Vigorous physical activity
SharingNA–0.1 (-0.21,–0.05)0.59 (0.19, 0.92)0.072
Causal–0.07 (-0.14, 0)0 (-0.31, 0.36)0.19 (0, 0.86)
 WCadjBMI → Moderate physical activity
SharingNA–0.11 (-0.27,–0.04)0.47 (0.11, 0.88)0.19
Causal–0.06 (-0.14, 0.03)0 (-0.34, 0.36)0.19 (0, 0.86)
 WCadjBMI → Sedentary time
SharingNA0.1 (0.05, 0.2)0.63 (0.23, 0.93)0.059
Causal0.07 (0.01, 0.14)0.01 (-0.3, 0.48)0.17 (0, 0.85)
 
 Vigorous physical activity → WHRadjBMI
SharingNA–0.1 (-0.39, 0.07)0.36 (0.01, 0.86)0.2
Causal–0.06 (-0.16, 0.04)0 (-0.37, 0.44)0.19 (0, 0.86)
 Moderate physical activity → WHRadjBMI
SharingNA–0.04 (-0.48, 0.39)0.15 (0, 0.76)8.60E-01
Causal–0.03 (-0.14, 0.09)0 (-0.41, 0.4)0.2 (0, 0.86)
 Sedentary time → WHRadjBMI
SharingNA0.01 (-0.26, 0.34)0.12 (0, 0.72)1
Causal0 (-0.06, 0.07)0 (-0.22, 0.28)0.2 (0, 0.86)
 WHRadjBMI → Vigorous physical activity
SharingNA–0.04 (-0.3, 0.12)0.16 (0, 0.76)0.65
Causal–0.01 (-0.06, 0.03)0 (-0.25, 0.21)0.18 (0, 0.85)
 WHRadjBMI → Moderate physical activity
SharingNA–0.04 (-0.31, 0.14)0.16 (0, 0.75)6.30E-01
Causal–0.01 (-0.06, 0.03)0 (-0.25, 0.22)0.18 (0, 0.85)
 WHRadjBMI → Sedentary time
SharingNA0.07 (0.03, 0.18)0.49 (0.13, 0.89)0.17
Causal0.04 (-0.01, 0.09)0 (-0.29, 0.4)0.17 (0, 0.83)

  1. BFP, Body Fat Percentage; WCadjBMI, waist circumference adjusted for BMI; WHRadjBMI,waist-to-hip ratio adjusted for BMI; NA, not applicable; CI, 95% confidence interval.

Appendix 2

Analysis plan for causality between physical activity, sedentary behaviour, and obesity: A Mendelian randomization study

Date: initiated in March 2020

Motivation: Obesity is a global epidemic increasing morbidity and mortality worldwide. Physical inactivity and increased sedentary time are associated with excess weight gain as shown from observational studies. However, observational studies suffer from residual confounding and reverse causality. Mendelian randomization helps to overcome confounding and reverse causality by instrumenting the exposure trait using genetic variants. Here, we aim to assess the causality between physical activity, sedentary behaviour and body mass index in adults by bidirectional Mendelian randomization analyses.

Inclusion criteria for genome wide summary level data:

  • Largest published genome-wide association study summary statistics

  • European ancestry

  • Objectively measured continuous traits

Exposures:

  • Vigorous physical activity

  • Moderate physical activity

  • Sedentary time

  • Body mass index

Outcomes:

  • Vigorous physical activity

  • Moderate physical activity

  • Sedentary time

  • Body mass index

Directions of association for outcome and exposure:

  • Vigorous physical activity → body mass index

  • Moderate physical activity → body mass index

  • Sedentary time → body mass index

  • Body mass index → vigorous physical activity

  • Body mass index → moderate physical activity

  • Body mass index → sedentary time

Mendelian randomization methods:

  • The Causal Analysis Using Summary Effect estimates (CAUSE) method

  • Inverse variance weighted (IVW)

  • Mendelian randomization-Egger (MR-Egger)

  • Weighted median

  • Weighted mode

Sensitivity tests and plots to account for heterogeneity and horizontal pleiotropy in the IVW, MR-Egger, weighted median and weighted mode analyses:

  • Steiger filtering

  • Automated outlier removal with RadialMR

  • Rucker framework

  • Cochran’s Q method

  • Leave-one-out forest plots

  • Funnel plots

Note: The Mendelian randomization analyses, sensitivity tests, and sensitivity plots for IVW, MR-Egger, weighted median and weighted mode methods were repeated after outlier removal.

Post-hoc analysis:

When less than three genetic variants associated with the exposure trait were available for the IVW, MR-Egger, weighted median, and weighted mode methods using genome-wide significant threshold (p < 5 × 10-8), we used a p-value threshold of p < 5 × 10-7 to identify a sufficient number of genetic instruments to produce stable estimates and plots.

R packages used:

Post-review analysis: during the peer-review process we performed further analysis using CAUSE method for body fat percentage (BFP), waist circumference adjusted for BMI, and waist-to-hip ratio adjusted for BMI.

Data availability

Data sharing: All analyses were performed using R statistical package freely available at https://cran.r-project.org/mirrors.html. The CAUSE R package and instructions are available at https://jean997.github.io/cause/. The Two-sample MR package is available at https://mrcieu.github.io/TwoSampleMR/. The RadialMR package is available at https://github.com/WSpiller/RadialMR. The code and curated data for the current analysis are available at https://github.com/MarioGuCBMR/MR_Physical_Activity_BMI, (copy archived at swh:1:rev:b6c3fcd3a97a2d67329ab26b7c061da5941bd4f2).

References

    1. Justice AE
    2. Winkler TW
    3. Feitosa MF
    4. Graff M
    5. Fisher VA
    6. Young K
    7. Barata L
    8. Deng X
    9. Czajkowski J
    10. Hadley D
    11. Ngwa JS
    12. Ahluwalia TS
    13. Chu AY
    14. Heard-Costa NL
    15. Lim E
    16. Perez J
    17. Eicher JD
    18. Kutalik Z
    19. Xue L
    20. Mahajan A
    21. Renström F
    22. Wu J
    23. Qi Q
    24. Ahmad S
    25. Alfred T
    26. Amin N
    27. Bielak LF
    28. Bonnefond A
    29. Bragg J
    30. Cadby G
    31. Chittani M
    32. Coggeshall S
    33. Corre T
    34. Direk N
    35. Eriksson J
    36. Fischer K
    37. Gorski M
    38. Neergaard Harder M
    39. Horikoshi M
    40. Huang T
    41. Huffman JE
    42. Jackson AU
    43. Justesen JM
    44. Kanoni S
    45. Kinnunen L
    46. Kleber ME
    47. Komulainen P
    48. Kumari M
    49. Lim U
    50. Luan J
    51. Lyytikäinen L-P
    52. Mangino M
    53. Manichaikul A
    54. Marten J
    55. Middelberg RPS
    56. Müller-Nurasyid M
    57. Navarro P
    58. Pérusse L
    59. Pervjakova N
    60. Sarti C
    61. Smith AV
    62. Smith JA
    63. Stančáková A
    64. Strawbridge RJ
    65. Stringham HM
    66. Sung YJ
    67. Tanaka T
    68. Teumer A
    69. Trompet S
    70. van der Laan SW
    71. van der Most PJ
    72. Van Vliet-Ostaptchouk JV
    73. Vedantam SL
    74. Verweij N
    75. Vink JM
    76. Vitart V
    77. Wu Y
    78. Yengo L
    79. Zhang W
    80. Hua Zhao J
    81. Zimmermann ME
    82. Zubair N
    83. Abecasis GR
    84. Adair LS
    85. Afaq S
    86. Afzal U
    87. Bakker SJL
    88. Bartz TM
    89. Beilby J
    90. Bergman RN
    91. Bergmann S
    92. Biffar R
    93. Blangero J
    94. Boerwinkle E
    95. Bonnycastle LL
    96. Bottinger E
    97. Braga D
    98. Buckley BM
    99. Buyske S
    100. Campbell H
    101. Chambers JC
    102. Collins FS
    103. Curran JE
    104. de Borst GJ
    105. de Craen AJM
    106. de Geus EJC
    107. Dedoussis G
    108. Delgado GE
    109. den Ruijter HM
    110. Eiriksdottir G
    111. Eriksson AL
    112. Esko T
    113. Faul JD
    114. Ford I
    115. Forrester T
    116. Gertow K
    117. Gigante B
    118. Glorioso N
    119. Gong J
    120. Grallert H
    121. Grammer TB
    122. Grarup N
    123. Haitjema S
    124. Hallmans G
    125. Hamsten A
    126. Hansen T
    127. Harris TB
    128. Hartman CA
    129. Hassinen M
    130. Hastie ND
    131. Heath AC
    132. Hernandez D
    133. Hindorff L
    134. Hocking LJ
    135. Hollensted M
    136. Holmen OL
    137. Homuth G
    138. Jan Hottenga J
    139. Huang J
    140. Hung J
    141. Hutri-Kähönen N
    142. Ingelsson E
    143. James AL
    144. Jansson J-O
    145. Jarvelin M-R
    146. Jhun MA
    147. Jørgensen ME
    148. Juonala M
    149. Kähönen M
    150. Karlsson M
    151. Koistinen HA
    152. Kolcic I
    153. Kolovou G
    154. Kooperberg C
    155. Krämer BK
    156. Kuusisto J
    157. Kvaløy K
    158. Lakka TA
    159. Langenberg C
    160. Launer LJ
    161. Leander K
    162. Lee NR
    163. Lind L
    164. Lindgren CM
    165. Linneberg A
    166. Lobbens S
    167. Loh M
    168. Lorentzon M
    169. Luben R
    170. Lubke G
    171. Ludolph-Donislawski A
    172. Lupoli S
    173. Madden PAF
    174. Männikkö R
    175. Marques-Vidal P
    176. Martin NG
    177. McKenzie CA
    178. McKnight B
    179. Mellström D
    180. Menni C
    181. Montgomery GW
    182. Musk AB
    183. Narisu N
    184. Nauck M
    185. Nolte IM
    186. Oldehinkel AJ
    187. Olden M
    188. Ong KK
    189. Padmanabhan S
    190. Peyser PA
    191. Pisinger C
    192. Porteous DJ
    193. Raitakari OT
    194. Rankinen T
    195. Rao DC
    196. Rasmussen-Torvik LJ
    197. Rawal R
    198. Rice T
    199. Ridker PM
    200. Rose LM
    201. Bien SA
    202. Rudan I
    203. Sanna S
    204. Sarzynski MA
    205. Sattar N
    206. Savonen K
    207. Schlessinger D
    208. Scholtens S
    209. Schurmann C
    210. Scott RA
    211. Sennblad B
    212. Siemelink MA
    213. Silbernagel G
    214. Slagboom PE
    215. Snieder H
    216. Staessen JA
    217. Stott DJ
    218. Swertz MA
    219. Swift AJ
    220. Taylor KD
    221. Tayo BO
    222. Thorand B
    223. Thuillier D
    224. Tuomilehto J
    225. Uitterlinden AG
    226. Vandenput L
    227. Vohl M-C
    228. Völzke H
    229. Vonk JM
    230. Waeber G
    231. Waldenberger M
    232. Westendorp RGJ
    233. Wild S
    234. Willemsen G
    235. Wolffenbuttel BHR
    236. Wong A
    237. Wright AF
    238. Zhao W
    239. Zillikens MC
    240. Baldassarre D
    241. Balkau B
    242. Bandinelli S
    243. Böger CA
    244. Boomsma DI
    245. Bouchard C
    246. Bruinenberg M
    247. Chasman DI
    248. Chen Y-D
    249. Chines PS
    250. Cooper RS
    251. Cucca F
    252. Cusi D
    253. Faire U de
    254. Ferrucci L
    255. Franks PW
    256. Froguel P
    257. Gordon-Larsen P
    258. Grabe H-J
    259. Gudnason V
    260. Haiman CA
    261. Hayward C
    262. Hveem K
    263. Johnson AD
    264. Wouter Jukema J
    265. Kardia SLR
    266. Kivimaki M
    267. Kooner JS
    268. Kuh D
    269. Laakso M
    270. Lehtimäki T
    271. Marchand LL
    272. März W
    273. McCarthy MI
    274. Metspalu A
    275. Morris AP
    276. Ohlsson C
    277. Palmer LJ
    278. Pasterkamp G
    279. Pedersen O
    280. Peters A
    281. Peters U
    282. Polasek O
    283. Psaty BM
    284. Qi L
    285. Rauramaa R
    286. Smith BH
    287. Sørensen TIA
    288. Strauch K
    289. Tiemeier H
    290. Tremoli E
    291. van der Harst P
    292. Vestergaard H
    293. Vollenweider P
    294. Wareham NJ
    295. Weir DR
    296. Whitfield JB
    297. Wilson JF
    298. Tyrrell J
    299. Frayling TM
    300. Barroso I
    301. Boehnke M
    302. Deloukas P
    303. Fox CS
    304. Hirschhorn JN
    305. Hunter DJ
    306. Spector TD
    307. Strachan DP
    308. van Duijn CM
    309. Heid IM
    310. Mohlke KL
    311. Marchini J
    312. Loos RJF
    313. Kilpeläinen TO
    314. Liu C-T
    315. Borecki IB
    316. North KE
    317. Cupples LA
    (2017) Genome-wide meta-analysis of 241,258 adults accounting for smoking behaviour identifies novel loci for obesity traits
    Nature Communications 8:14977.
    https://doi.org/10.1038/ncomms14977
    1. Locke AE
    2. Kahali B
    3. Berndt SI
    4. Justice AE
    5. Pers TH
    6. Day FR
    7. Powell C
    8. Vedantam S
    9. Buchkovich ML
    10. Yang J
    11. Croteau-Chonka DC
    12. Esko T
    13. Fall T
    14. Ferreira T
    15. Gustafsson S
    16. Kutalik Z
    17. Luan J
    18. Mägi R
    19. Randall JC
    20. Winkler TW
    21. Wood AR
    22. Workalemahu T
    23. Faul JD
    24. Smith JA
    25. Zhao JH
    26. Zhao W
    27. Chen J
    28. Fehrmann R
    29. Hedman ÅK
    30. Karjalainen J
    31. Schmidt EM
    32. Absher D
    33. Amin N
    34. Anderson D
    35. Beekman M
    36. Bolton JL
    37. Bragg-Gresham JL
    38. Buyske S
    39. Demirkan A
    40. Deng G
    41. Ehret GB
    42. Feenstra B
    43. Feitosa MF
    44. Fischer K
    45. Goel A
    46. Gong J
    47. Jackson AU
    48. Kanoni S
    49. Kleber ME
    50. Kristiansson K
    51. Lim U
    52. Lotay V
    53. Mangino M
    54. Leach IM
    55. Medina-Gomez C
    56. Medland SE
    57. Nalls MA
    58. Palmer CD
    59. Pasko D
    60. Pechlivanis S
    61. Peters MJ
    62. Prokopenko I
    63. Shungin D
    64. Stančáková A
    65. Strawbridge RJ
    66. Sung YJ
    67. Tanaka T
    68. Teumer A
    69. Trompet S
    70. van der Laan SW
    71. van Setten J
    72. Van Vliet-Ostaptchouk JV
    73. Wang Z
    74. Yengo L
    75. Zhang W
    76. Isaacs A
    77. Albrecht E
    78. Ärnlöv J
    79. Arscott GM
    80. Attwood AP
    81. Bandinelli S
    82. Barrett A
    83. Bas IN
    84. Bellis C
    85. Bennett AJ
    86. Berne C
    87. Blagieva R
    88. Blüher M
    89. Böhringer S
    90. Bonnycastle LL
    91. Böttcher Y
    92. Boyd HA
    93. Bruinenberg M
    94. Caspersen IH
    95. Chen Y-DI
    96. Clarke R
    97. Daw EW
    98. de Craen AJM
    99. Delgado G
    100. Dimitriou M
    101. Doney ASF
    102. Eklund N
    103. Estrada K
    104. Eury E
    105. Folkersen L
    106. Fraser RM
    107. Garcia ME
    108. Geller F
    109. Giedraitis V
    110. Gigante B
    111. Go AS
    112. Golay A
    113. Goodall AH
    114. Gordon SD
    115. Gorski M
    116. Grabe H-J
    117. Grallert H
    118. Grammer TB
    119. Gräßler J
    120. Grönberg H
    121. Groves CJ
    122. Gusto G
    123. Haessler J
    124. Hall P
    125. Haller T
    126. Hallmans G
    127. Hartman CA
    128. Hassinen M
    129. Hayward C
    130. Heard-Costa NL
    131. Helmer Q
    132. Hengstenberg C
    133. Holmen O
    134. Hottenga J-J
    135. James AL
    136. Jeff JM
    137. Johansson Å
    138. Jolley J
    139. Juliusdottir T
    140. Kinnunen L
    141. Koenig W
    142. Koskenvuo M
    143. Kratzer W
    144. Laitinen J
    145. Lamina C
    146. Leander K
    147. Lee NR
    148. Lichtner P
    149. Lind L
    150. Lindström J
    151. Lo KS
    152. Lobbens S
    153. Lorbeer R
    154. Lu Y
    155. Mach F
    156. Magnusson PKE
    157. Mahajan A
    158. McArdle WL
    159. McLachlan S
    160. Menni C
    161. Merger S
    162. Mihailov E
    163. Milani L
    164. Moayyeri A
    165. Monda KL
    166. Morken MA
    167. Mulas A
    168. Müller G
    169. Müller-Nurasyid M
    170. Musk AW
    171. Nagaraja R
    172. Nöthen MM
    173. Nolte IM
    174. Pilz S
    175. Rayner NW
    176. Renstrom F
    177. Rettig R
    178. Ried JS
    179. Ripke S
    180. Robertson NR
    181. Rose LM
    182. Sanna S
    183. Scharnagl H
    184. Scholtens S
    185. Schumacher FR
    186. Scott WR
    187. Seufferlein T
    188. Shi J
    189. Smith AV
    190. Smolonska J
    191. Stanton AV
    192. Steinthorsdottir V
    193. Stirrups K
    194. Stringham HM
    195. Sundström J
    196. Swertz MA
    197. Swift AJ
    198. Syvänen A-C
    199. Tan S-T
    200. Tayo BO
    201. Thorand B
    202. Thorleifsson G
    203. Tyrer JP
    204. Uh H-W
    205. Vandenput L
    206. Verhulst FC
    207. Vermeulen SH
    208. Verweij N
    209. Vonk JM
    210. Waite LL
    211. Warren HR
    212. Waterworth D
    213. Weedon MN
    214. Wilkens LR
    215. Willenborg C
    216. Wilsgaard T
    217. Wojczynski MK
    218. Wong A
    219. Wright AF
    220. Zhang Q
    221. Brennan EP
    222. Choi M
    223. Dastani Z
    224. Drong AW
    225. Eriksson P
    226. Franco-Cereceda A
    227. Gådin JR
    228. Gharavi AG
    229. Goddard ME
    230. Handsaker RE
    231. Huang J
    232. Karpe F
    233. Kathiresan S
    234. Keildson S
    235. Kiryluk K
    236. Kubo M
    237. Lee J-Y
    238. Liang L
    239. Lifton RP
    240. Ma B
    241. McCarroll SA
    242. McKnight AJ
    243. Min JL
    244. Moffatt MF
    245. Montgomery GW
    246. Murabito JM
    247. Nicholson G
    248. Nyholt DR
    249. Okada Y
    250. Perry JRB
    251. Dorajoo R
    252. Reinmaa E
    253. Salem RM
    254. Sandholm N
    255. Scott RA
    256. Stolk L
    257. Takahashi A
    258. Tanaka T
    259. van ’t Hooft FM
    260. Vinkhuyzen AAE
    261. Westra H-J
    262. Zheng W
    263. Zondervan KT
    264. Heath AC
    265. Arveiler D
    266. Bakker SJL
    267. Beilby J
    268. Bergman RN
    269. Blangero J
    270. Bovet P
    271. Campbell H
    272. Caulfield MJ
    273. Cesana G
    274. Chakravarti A
    275. Chasman DI
    276. Chines PS
    277. Collins FS
    278. Crawford DC
    279. Cupples LA
    280. Cusi D
    281. Danesh J
    282. de Faire U
    283. den Ruijter HM
    284. Dominiczak AF
    285. Erbel R
    286. Erdmann J
    287. Eriksson JG
    288. Farrall M
    289. Felix SB
    290. Ferrannini E
    291. Ferrières J
    292. Ford I
    293. Forouhi NG
    294. Forrester T
    295. Franco OH
    296. Gansevoort RT
    297. Gejman PV
    298. Gieger C
    299. Gottesman O
    300. Gudnason V
    301. Gyllensten U
    302. Hall AS
    303. Harris TB
    304. Hattersley AT
    305. Hicks AA
    306. Hindorff LA
    307. Hingorani AD
    308. Hofman A
    309. Homuth G
    310. Hovingh GK
    311. Humphries SE
    312. Hunt SC
    313. Hyppönen E
    314. Illig T
    315. Jacobs KB
    316. Jarvelin M-R
    317. Jöckel K-H
    318. Johansen B
    319. Jousilahti P
    320. Jukema JW
    321. Jula AM
    322. Kaprio J
    323. Kastelein JJP
    324. Keinanen-Kiukaanniemi SM
    325. Kiemeney LA
    326. Knekt P
    327. Kooner JS
    328. Kooperberg C
    329. Kovacs P
    330. Kraja AT
    331. Kumari M
    332. Kuusisto J
    333. Lakka TA
    334. Langenberg C
    335. Marchand LL
    336. Lehtimäki T
    337. Lyssenko V
    338. Männistö S
    339. Marette A
    340. Matise TC
    341. McKenzie CA
    342. McKnight B
    343. Moll FL
    344. Morris AD
    345. Morris AP
    346. Murray JC
    347. Nelis M
    348. Ohlsson C
    349. Oldehinkel AJ
    350. Ong KK
    351. Madden PAF
    352. Pasterkamp G
    353. Peden JF
    354. Peters A
    355. Postma DS
    356. Pramstaller PP
    357. Price JF
    358. Qi L
    359. Raitakari OT
    360. Rankinen T
    361. Rao DC
    362. Rice TK
    363. Ridker PM
    364. Rioux JD
    365. Ritchie MD
    366. Rudan I
    367. Salomaa V
    368. Samani NJ
    369. Saramies J
    370. Sarzynski MA
    371. Schunkert H
    372. Schwarz PEH
    373. Sever P
    374. Shuldiner AR
    375. Sinisalo J
    376. Stolk RP
    377. Strauch K
    378. Tönjes A
    379. Trégouët D-A
    380. Tremblay A
    381. Tremoli E
    382. Virtamo J
    383. Vohl M-C
    384. Völker U
    385. Waeber G
    386. Willemsen G
    387. Witteman JC
    388. Zillikens MC
    389. Adair LS
    390. Amouyel P
    391. Asselbergs FW
    392. Assimes TL
    393. Bochud M
    394. Boehm BO
    395. Boerwinkle E
    396. Bornstein SR
    397. Bottinger EP
    398. Bouchard C
    399. Cauchi S
    400. Chambers JC
    401. Chanock SJ
    402. Cooper RS
    403. de Bakker PIW
    404. Dedoussis G
    405. Ferrucci L
    406. Franks PW
    407. Froguel P
    408. Groop LC
    409. Haiman CA
    410. Hamsten A
    411. Hui J
    412. Hunter DJ
    413. Hveem K
    414. Kaplan RC
    415. Kivimaki M
    416. Kuh D
    417. Laakso M
    418. Liu Y
    419. Martin NG
    420. März W
    421. Melbye M
    422. Metspalu A
    423. Moebus S
    424. Munroe PB
    425. Njølstad I
    426. Oostra BA
    427. Palmer CNA
    428. Pedersen NL
    429. Perola M
    430. Pérusse L
    431. Peters U
    432. Power C
    433. Quertermous T
    434. Rauramaa R
    435. Rivadeneira F
    436. Saaristo TE
    437. Saleheen D
    438. Sattar N
    439. Schadt EE
    440. Schlessinger D
    441. Slagboom PE
    442. Snieder H
    443. Spector TD
    444. Thorsteinsdottir U
    445. Stumvoll M
    446. Tuomilehto J
    447. Uitterlinden AG
    448. Uusitupa M
    449. van der Harst P
    450. Walker M
    451. Wallaschofski H
    452. Wareham NJ
    453. Watkins H
    454. Weir DR
    455. Wichmann H-E
    456. Wilson JF
    457. Zanen P
    458. Borecki IB
    459. Deloukas P
    460. Fox CS
    461. Heid IM
    462. O’Connell JR
    463. Strachan DP
    464. Stefansson K
    465. van Duijn CM
    466. Abecasis GR
    467. Franke L
    468. Frayling TM
    469. McCarthy MI
    470. Visscher PM
    471. Scherag A
    472. Willer CJ
    473. Boehnke M
    474. Mohlke KL
    475. Lindgren CM
    476. Beckmann JS
    477. Barroso I
    478. North KE
    479. Ingelsson E
    480. Hirschhorn JN
    481. Loos RJF
    482. Speliotes EK
    483. LifeLines Cohort Study
    484. ADIPOGen Consortium
    485. AGEN-BMI Working Group
    486. CARDIOGRAMplusC4D Consortium
    487. CKDGen Consortium
    488. GLGC
    489. ICBP
    490. MAGIC Investigators
    491. MuTHER Consortium
    492. MIGen Consortium
    493. PAGE Consortium
    494. ReproGen Consortium
    495. GENIE Consortium
    496. International Endogene Consortium
    (2015) Genetic studies of body mass index yield new insights for obesity biology
    Nature 518:197–206.
    https://doi.org/10.1038/nature14177

Article and author information

Author details

  1. Germán D Carrasquilla

    Novo Nordisk Foundation Center for Basic Metabolic Research, Faculty of Health and Medical Sciences, University of Copenhagen, Copenhagen, Denmark
    Contribution
    Conceptualization, Data curation, Formal analysis, Funding acquisition, Methodology, Project administration, Supervision, Visualization, Writing – original draft, Writing – review and editing
    For correspondence
    german.carrasquilla@sund.ku.dk
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-7147-9421

  2. Mario García-Ureña

    Novo Nordisk Foundation Center for Basic Metabolic Research, Faculty of Health and Medical Sciences, University of Copenhagen, Copenhagen, Denmark
    Contribution
    Data curation, Formal analysis, Methodology, Visualization, Writing – review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-3376-9460

  3. Tove Fall

    Molecular Epidemiology, Department of Medical Sciences, and Science for Life Laboratory, Uppsala University, Uppsala, Sweden
    Contribution
    Supervision, Writing – review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-2071-5866

  4. Thorkild IA Sørensen

    1. Novo Nordisk Foundation Center for Basic Metabolic Research, Faculty of Health and Medical Sciences, University of Copenhagen, Copenhagen, Denmark
    2. Department of Public Health, Section of Epidemiology, Faculty of Health and Medical Sciences, University of Copenhagen, Copenhagen, Denmark
    Contribution
    Supervision, Writing – review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-4821-430X

  5. Tuomas O Kilpeläinen

    Novo Nordisk Foundation Center for Basic Metabolic Research, Faculty of Health and Medical Sciences, University of Copenhagen, Copenhagen, Denmark
    Contribution
    Conceptualization, Data curation, Funding acquisition, Methodology, Resources, Supervision, Writing – review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-8349-3028

Funding

H2020 Marie Skłodowska-Curie Actions (846502)

  • Germán Darío Carrasquilla

Novo Nordisk Foundation Center for Basic Metabolic Research (NNF18CC0034900)

  • Tuomas Kilpeläinen

Danish Diabetes Academy (NNF17SA0031406)

  • Germán Darío Carrasquilla

Novo Nordisk Fonden (NNF17OC0026848)

  • Tuomas Kilpeläinen

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

This project has received funding from the European Union’s Horizon 2020 research and innovation programme under the Marie Sklodowska-Curie grant agreement No 846,502. Novo Nordisk Foundation Center for Basic Metabolic Research is an independent research center at the University of Copenhagen partially funded by an unrestricted donation from the Novo Nordisk Foundation (NNF18CC0034900). Germán D Carrasquilla was supported by a grant from the Danish Diabetes Academy that is funded by the Novo Nordisk Foundation (NNF17SA0031406). Tuomas O Kilpeläinen was supported by a grant from the Novo Nordisk Foundation (NNF17OC0026848). The funding source had no role in study design, data collection, data analysis, data interpretation, or writing of the report. All authors had full access to all of the data in the study and had final responsibility for the decision to submit for publication.

Copyright

© 2022, Carrasquilla et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 3,066
    views
  • 411
    downloads
  • 24
    citations

Views, downloads and citations are aggregated across all versions of this paper published by eLife.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Germán D Carrasquilla
  2. Mario García-Ureña
  3. Tove Fall
  4. Thorkild IA Sørensen
  5. Tuomas O Kilpeläinen
(2022)
Mendelian randomization suggests a bidirectional, causal relationship between physical inactivity and adiposity
eLife 11:e70386.
https://doi.org/10.7554/eLife.70386

Share this article

https://doi.org/10.7554/eLife.70386

Categories and tags

Further reading

    1. Epidemiology and Global Health
    2. Genetics and Genomics

    Systematic evaluation of multifactorial causal associations for Alzheimer’s disease and an interactive platform MRAD developed based on Mendelian randomization analysis

    Tianyu Zhao, Hui Li ... Li Chen
    Research Article

    Alzheimer’s disease (AD) is a complex degenerative disease of the central nervous system, and elucidating its pathogenesis remains challenging. In this study, we used the inverse-variance weighted (IVW) model as the major analysis method to perform hypothesis-free Mendelian randomization (MR) analysis on the data from MRC IEU OpenGWAS (18,097 exposure traits and 16 AD outcome traits), and conducted sensitivity analysis with six models, to assess the robustness of the IVW results, to identify various classes of risk or protective factors for AD, early-onset AD, and late-onset AD. We generated 400,274 data entries in total, among which the major analysis method of the IVW model consists of 73,129 records with 4840 exposure traits, which fall into 10 categories: Disease, Medical laboratory science, Imaging, Anthropometric, Treatment, Molecular trait, Gut microbiota, Past history, Family history, and Lifestyle trait. More importantly, a freely accessed online platform called MRAD (https://gwasmrad.com/mrad/) has been developed using the Shiny package with MR analysis results. Additionally, novel potential AD therapeutic targets (CD33, TBCA, VPS29, GNAI3, PSME1) are identified, among which CD33 was positively associated with the main outcome traits of AD, as well as with both EOAD and LOAD. TBCA and VPS29 were negatively associated with the main outcome traits of AD, as well as with both EOAD and LOAD. GNAI3 and PSME1 were negatively associated with the main outcome traits of AD, as well as with LOAD, but had no significant causal association with EOAD. The findings of our research advance our understanding of the etiology of AD.

    1. Epidemiology and Global Health

    Noncaloric monosaccharides induce excessive sprouting angiogenesis in zebrafish via foxo1a-marcksl1a signal

    Xiaoning Wang, Jinxiang Zhao ... Dong Liu
    Research Article

    Artificially sweetened beverages containing noncaloric monosaccharides were suggested as healthier alternatives to sugar-sweetened beverages. Nevertheless, the potential detrimental effects of these noncaloric monosaccharides on blood vessel function remain inadequately understood. We have established a zebrafish model that exhibits significant excessive angiogenesis induced by high glucose, resembling the hyperangiogenic characteristics observed in proliferative diabetic retinopathy (PDR). Utilizing this model, we observed that glucose and noncaloric monosaccharides could induce excessive formation of blood vessels, especially intersegmental vessels (ISVs). The excessively branched vessels were observed to be formed by ectopic activation of quiescent endothelial cells (ECs) into tip cells. Single-cell transcriptomic sequencing analysis of the ECs in the embryos exposed to high glucose revealed an augmented ratio of capillary ECs, proliferating ECs, and a series of upregulated proangiogenic genes. Further analysis and experiments validated that reduced foxo1a mediated the excessive angiogenesis induced by monosaccharides via upregulating the expression of marcksl1a. This study has provided new evidence showing the negative effects of noncaloric monosaccharides on the vascular system and the underlying mechanisms.

    1. Epidemiology and Global Health
    2. Microbiology and Infectious Disease

    Antigenic drift and subtype interference shape A(H3N2) epidemic dynamics in the United States

    Amanda C Perofsky, John Huddleston ... Cécile Viboud
    Research Article

    Influenza viruses continually evolve new antigenic variants, through mutations in epitopes of their major surface proteins, hemagglutinin (HA) and neuraminidase (NA). Antigenic drift potentiates the reinfection of previously infected individuals, but the contribution of this process to variability in annual epidemics is not well understood. Here, we link influenza A(H3N2) virus evolution to regional epidemic dynamics in the United States during 1997—2019. We integrate phenotypic measures of HA antigenic drift and sequence-based measures of HA and NA fitness to infer antigenic and genetic distances between viruses circulating in successive seasons. We estimate the magnitude, severity, timing, transmission rate, age-specific patterns, and subtype dominance of each regional outbreak and find that genetic distance based on broad sets of epitope sites is the strongest evolutionary predictor of A(H3N2) virus epidemiology. Increased HA and NA epitope distance between seasons correlates with larger, more intense epidemics, higher transmission, greater A(H3N2) subtype dominance, and a greater proportion of cases in adults relative to children, consistent with increased population susceptibility. Based on random forest models, A(H1N1) incidence impacts A(H3N2) epidemics to a greater extent than viral evolution, suggesting that subtype interference is a major driver of influenza A virus infection ynamics, presumably via heterosubtypic cross-immunity.