Exploiting collider bias to apply two-sample summary data Mendelian randomization methods to one-sample individual level data

Association between the sarcopenia-related traits and Parkinson's disease: A bidirectional two-sample Mendelian randomization study

OBJECTIVE

To explore the causal association between sarcopenia-related traits and Parkinson's disease by Mendelian randomization (MR) approach.

METHODS

A genome-wide association study (GWAS) of sarcopenia-related traits was done at the UK Biobank (UKB). The traits were appendicular lean mass, low hand grip strength (including the European Working Group on Sarcopenia in Older People (EWGSOP) and the Foundation for the National Institutes of Health (FNIH) criteria and usual walking pace. The International Parkinson's Disease Genomics Consortium (IPDGC) gave us GWAS data for Parkinson's disease (PD). We used three different types of MR analyses: including Inverse-variance weighted (IVW), Mendelian randomized Egger regression (MR-Egger), and weighted median methods (both weighted and simple modes).

RESULTS

The MR analysis showed that low hand grip strength was negatively associated with the risk of developing Parkinson's disease, including EWGSOP criterion (odds ratio (OR) = 0.734; 95% confidence interval (CI) = 0.575-0.937, P = 0.013) and FNIH criterion (OR = 0.619; 95% CI = 0.419-0.914, P = 0.016), and usual walking pace was also a risk factor for Parkinson's disease (OR = 3.307, 95% CI = 1.277-8.565, P = 0.014).

CONCLUSIONS

In European population, low hand grip strength is negatively associated with the risk of developing Parkinson's disease, and usual walking pace is also a risk factor for Parkinson's disease. Further exploration of the potential genetic mechanisms underlying hand grip strength and Parkinson's disease and the potential relationship between walking pace, balance, and falls in Parkinson's patients may help to reduce the burden of sarcopenia and Parkinson's disease.

The role of accelerometer-derived sleep traits on glycated haemoglobin and glucose levels: a Mendelian randomization study

Self-reported shorter/longer sleep duration, insomnia, and evening preference are associated with hyperglycaemia in observational analyses, with similar observations in small studies using accelerometer-derived sleep traits. Mendelian randomization (MR) studies support an effect of self-reported insomnia, but not others, on glycated haemoglobin (HbA1c). To explore potential effects, we used MR methods to assess effects of accelerometer-derived sleep traits (duration, mid-point least active 5-h, mid-point most active 10-h, sleep fragmentation, and efficiency) on HbA1c/glucose in European adults from the UK Biobank (UKB) (n = 73,797) and the MAGIC consortium (n = 146,806). Cross-trait linkage disequilibrium score regression was applied to determine genetic correlations across accelerometer-derived, self-reported sleep traits, and HbA1c/glucose. We found no causal effect of any accelerometer-derived sleep trait on HbA1c or glucose. Similar MR results for self-reported sleep traits in the UKB sub-sample with accelerometer-derived measures suggested our results were not explained by selection bias. Phenotypic and genetic correlation analyses suggested complex relationships between self-reported and accelerometer-derived traits indicating that they may reflect different types of exposure. These findings suggested accelerometer-derived sleep traits do not affect HbA1c. Accelerometer-derived measures of sleep duration and quality might not simply be 'objective' measures of self-reported sleep duration and insomnia, but rather captured different sleep characteristics.

Association of sex hormones with non-alcoholic fatty liver disease: An observational and Mendelian randomization study

BACKGROUND AND AIMS

Sex-specific associations of sex hormone-binding globulin (SHBG) and bioavailable testosterone (BAT) with NAFLD remain indeterminate. We aimed to explore observational and genetically determined relationships between each hormone and NAFLD.

METHODS

We included 187 395 men and 170 193 women from the UK Biobank. Linear and nonlinear Cox regression models and Mendelian randomization (MR) analysis were used to test the associations.

RESULTS

During 12.49 years of follow-up, 2209 male and 1886 female NAFLD cases were documented. Elevated SHBG levels were linearly associated with a lower risk of NAFLD in women (HR (95% CI), .71 (.63, .79)), but not in men (a "U" shape, pnon-linear < .001). Higher BAT levels were associated with a lower NAFLD risk in men (HR (95% CI), .81 (.71, .93)) but a higher risk in women (HR (95% CI): 1.25 (1.15, 1.36)). Genetically determined SHBG and BAT levels were linearly associated with NAFLD risk in women (OR (95% CI): .57 (.38, .87) and 2.21 (1.41, 3.26) respectively); in men, an "L-shaped" MR association between SHBG levels and NAFLD risk was found (pnon-linear = .016). The bidirectional MR analysis further revealed the effect of NAFLD on SHBG and BAT levels in both sexes.

CONCLUSIONS

Consistently, linear associations of lower SHBG and higher BAT levels with increased NAFLD risk were both conventionally and genetically found in women, while in men, SHBG acts in a nonlinear manner. In addition, NAFLD may affect SHBG and BAT levels.

Correlations in sleeping patterns and circadian preference between spouses

Spouses may affect each other's sleeping behaviour. In 47,420 spouse-pairs from the UK Biobank, we found a weak positive phenotypic correlation between spouses for self-reported sleep duration (r = 0.11; 95% CI = 0.10, 0.12) and a weak inverse correlation for chronotype (diurnal preference) (r = -0.11; -0.12, -0.10), which replicated in up to 127,035 23andMe spouse-pairs. Using accelerometer data on 3454 UK Biobank spouse-pairs, the correlation for derived sleep duration was similar to self-report (r = 0.12; 0.09, 0.15). Timing of diurnal activity was positively correlated (r = 0.24; 0.21, 0.27) in contrast to the inverse correlation for chronotype. In Mendelian randomization analysis, positive effects of sleep duration (mean difference=0.13; 0.04, 0.23 SD per SD) and diurnal activity (0.49; 0.03, 0.94) were observed, as were inverse effects of chronotype (-0.15; -0.26, -0.04) and snoring (-0.15; -0.27, -0.04). Findings support the notion that an individual's sleep may impact that of their partner, promoting opportunities for sleep interventions at the family-level.

Investigating the impact of metabolic syndrome traits on telomere length: a Mendelian randomization study

OBJECTIVE

Observational studies have reported bidirectional associations between metabolic syndrome (MetS) traits and short leukocyte telomere length (LTL), a TL marker in somatic tissues and a proposed risk factor for age-related degenerative diseases. However, in Mendelian randomization studies, longer LTL has been paradoxically associated with higher MetS risk. This study investigated the hypothesis that shorter LTL might be a consequence of metabolic dysfunction.

METHODS

This study undertook univariable and multivariable Mendelian randomization. As instrumental variables for MetS traits, all of the genome-wide significant independent signals identified in genome-wide association studies for anthropometric, glycemic, lipid, and blood pressure traits conducted in European individuals were used. Summary-level data for LTL were obtained from a genome-wide association study conducted in the UK Biobank.

RESULTS

Higher BMI was associated with shorter LTL (β = -0.039, 95% CI: -0.058 to -0.020, p = 5 × 10-5 ) equivalent to 1.70 years of age-related LTL change. In contrast, higher low-density lipoprotein cholesterol was associated with longer LTL (β = 0.022, 95% CI: 0.007 to 0.037, p = 0.003) equivalent to 0.96 years of age-related LTL change. Mechanistically, increased low-grade systemic inflammation, as measured by circulating C-reactive protein, and lower circulating linoleic acid levels might link higher BMI to shorter LTL.

CONCLUSIONS

Overweight and obesity might promote the development of aging-related degenerative diseases by accelerating telomere shortening.

The impact of obesity on lung function measurements and respiratory disease: A Mendelian randomization study

INTRODUCTION

Observational studies have shown that body mass index (BMI) and waist-to-hip ratio (WHR) are both inversely associated with lung function, as assessed by forced vital capacity (FVC) and forced expiratory volume in 1 s (FEV1). However, observational data are susceptible to confounding and reverse causation.

METHODS

We selected genetic instruments based on their relevant large-scale genome-wide association studies. Summary statistics of lung function and asthma came from the UK Biobank and SpiroMeta Consortium meta-analysis (n = 400,102). After examining pleiotropy and removing outliers, we applied inverse-variance weighting to estimate the causal association of BMI and BMI-adjusted WHR (WHRadjBMI) with FVC, FEV1, FEV1/FVC, and asthma. Sensitivity analyses were performed using weighted median, MR-Egger, and MRlap methods.

RESULTS

We found that BMI was inversely associated with FVC (effect estimate, -0.167; 95% confidence interval (CI), -0.203 to -0.130) and FEV1 (effect estimate, -0.111; 95%CI, -0.149 to -0.074). Higher BMI was associated with higher FEV1/FVC (effect estimate, 0.079; 95%CI, 0.049 to 0.110) but was not significantly associated with asthma. WHRadjBMI was inversely associated with FVC (effect estimate, -0.132; 95%CI, -0.180 to -0.084) but has no significant association with FEV1. Higher WHR was associated with higher FEV1/FVC (effect estimate, 0.181; 95%CI, 0.130 to 0.232) and with increased risk of asthma (effect estimate, 0.027; 95%CI, 0.001 to 0.053).

CONCLUSION

We found significant evidence that increased BMI is suggested to be causally related to decreased FVC and FEV1, and increased BMI-adjusted WHR could lead to lower FVC value and higher risk of asthma. Higher BMI and BMI-adjusted WHR were suggested to be causally associated with higher FEV1/FVC.

Association of educational attainment with esophageal cancer, Barrett's esophagus, and gastroesophageal reflux disease, and the mediating role of modifiable risk factors: A Mendelian randomization study

Background

Observational studies have reported that educational attainment has been related to the risk of esophageal cancer (EC) and its precancerous lesions. However, the causal relationship remains controversial. We aimed to apply the Mendelian randomization (MR) design to determine the causal associations between genetically predicted educational attainment and EC, Barrett's esophagus (BE), and gastroesophageal reflux disease (GERD), and to explore whether modifiable risk factors play a mediating role.

Methods

Using summary statistics from genome-wide association studies (GWASs) based on European ancestry individuals of several years in education (EduYears, primary analysis, n = 293,723), college completion (College, secondary analysis, n = 95,427), EC (n = 420,531), BE (n = 361,194), and GERD (n = 420,531), genetic associations between two education phenotypes and EC, BE, and GERD were tested by two-sample MR analyses. Then, two-step MR mediation analyses were used to assess the proportion of the aforementioned association that might be mediated by body mass index (BMI), major depressive disorder (MDD), smoking, drinking, carbohydrates, fat, and protein intake.

Results

Genetically predicted EduYears was negatively associated with the risk of EC, BE, and GERD {odds ratio (OR), 0.64 [95% confidence interval (CI) 0.44–0.94], 0.86 (95% CI, 0.75–0.99), and 0.62 (95%CI, 0.58–0.67)}. EduYears was negatively associated with BMI, MDD, and smoking (range of OR: 0.76–0.84). There were positive associations between BMI, smoking with EC, BE, and GERD, as well as between MDD with GERD (range of OR: 1.08–1.50). For individual mediating effect, BMI and smoking mediated 15.75 and 14.15% of the EduYears-EC association and 15.46 and 16.85% of the EduYears-BE association. BMI, MDD, and smoking mediated 5.23, 4.98, and 4.49% of the EduYears-GERD association. For combined mediation, the aforementioned mediators explained 26.62, 28.38, and 11.48% of the effect of EduYears on EC, BE, and GERD. The mediating effects of drinking and dietary composition were not significant in the effect of education on EC, BE, and GERD.

Conclusion

Our study supports that genetically predicted higher educational attainment has a protective effect on EC, BE, and GERD, and is partly mediated by reducing adiposity, smoking, and depression.

Weak and pleiotropy robust sex-stratified Mendelian randomization in the one sample and two sample settings

BACKGROUND

Mendelian randomization (MR) leverages genetic data as an instrumental variable to provide estimates for the causal effect of an exposure X on a health outcome Y that is robust to confounding. Unfortunately, horizontal pleiotropy-the direct association of a genetic variant with multiple phenotypes-is highly prevalent and can easily render a genetic variant an invalid instrument.

METHODS

Building on existing work, we propose a simple method for leveraging sex-specific genetic associations to perform weak and pleiotropy-robust MR analysis. This is achieved by constructing an MR estimator in which pleiotropy is perfectly removed by cancellation, while placing it within the powerful machinery of the robust adjusted profile score (MR-RAPS) method. Pleiotropy cancellation has the attractive property that it removes heterogeneity and therefore justifies a statistically efficient fixed effects model. We extend the method from the typical two-sample summary-data MR setting to the one-sample setting by adapting the technique of Collider-Correction. Simulation studies and applied examples are used to assess how the sex-stratified MR-RAPS estimator performs against other common approaches.

RESULTS

The sex-stratified MR-RAPS method is shown to be robust to pleiotropy even in cases where all genetic variants violated the standard Instrument Strength Independent of Direct Effect assumption. In some cases where the strength of the pleiotropic effect additionally varied by sex (and so perfect cancellation was not achieved), over-dispersed MR-RAPS implementations can still consistently estimate the true causal effect. In applied analyses, we investigate the causal effect of waist-hip ratio (WHR), an important marker of central obesity, on a range of downstream traits. While the conventional approaches suggested paradoxical links between WHR and height and body mass index, the sex-stratified approach obtained a more realistic null effect. Nonzero effects were also detected for systolic and diastolic blood pressure as well as high-density and low-density lipoprotein cholesterol.

DISCUSSION

We provide a simple but attractive method for weak and pleiotropy robust causal estimation of sexually dimorphic traits on downstream outcomes, by combining several existing approaches in a novel fashion.

Nongenetic Factors Associated With Psychotic Experiences Among UK Biobank Participants: Exposome-Wide Analysis and Mendelian Randomization Analysis

Importance

Although hypothesis-driven research has identified several factors associated with psychosis, this one-exposure-to-one-outcome approach fails to embrace the multiplicity of exposures. Systematic approaches, similar to agnostic genome-wide analyses, are needed to identify genuine signals.

Objective

To systematically investigate nongenetic correlates of psychotic experiences through data-driven agnostic analyses and genetically informed approaches to evaluate associations.

Design, Setting, Participants

This cohort study analyzed data from the UK Biobank Mental Health Survey from January 1 to June 1, 2021. An exposome-wide association study was performed in 2 equal-sized split discovery and replication data sets. Variables associated with psychotic experiences in the exposome-wide analysis were tested in a multivariable model. For the variables associated with psychotic experiences in the final multivariable model, the single-nucleotide variant-based heritability and genetic overlap with psychotic experiences using linkage disequilibrium score regression were estimated, and mendelian randomization (MR) approaches were applied to test potential causality. The significant associations observed in 1-sample MR analyses were further tested in multiple sensitivity tests, including collider-correction MR, 2-sample MR, and multivariable MR analyses.

Exposures

After quality control based on a priori criteria, 247 environmental, lifestyle, behavioral, and economic variables.

Main Outcomes and Measures

Psychotic experiences.

Results

The study included 155 247 participants (87 896 [57%] female; mean [SD] age, 55.94 [7.74] years). In the discovery data set, 162 variables (66%) were associated with psychotic experiences. Of these, 148 (91%) were replicated. The multivariable analysis identified 36 variables that were associated with psychotic experiences. Of these, 28 had significant genetic overlap with psychotic experiences. One-sample MR analyses revealed forward associations with 3 variables and reverse associations with 3. Forward associations with ever having experienced sexual assault and pleiotropy of risk-taking behavior and reverse associations without pleiotropy of experiencing a physically violent crime as well as cannabis use and the reverse association with pleiotropy of worrying too long after embarrassment were confirmed in sensitivity tests. Thus, associations with psychotic experiences were found with both well-studied and unexplored multiple correlated variables. For several variables, the direction of the association was reversed in the final multivariable and MR analyses.

Conclusions and Relevance

The findings of this study underscore the need for systematic approaches and triangulation of evidence to build a knowledge base from ever-growing observational data to guide population-level prevention strategies for psychosis.

Cross-fitted instrument: A blueprint for one-sample Mendelian randomization

Bias from weak instruments may undermine the ability to estimate causal effects in instrumental variable regression (IVR). We present here a new approach to handling weak instrument bias through the application of a new type of instrumental variable coined 'Cross-Fitted Instrument' (CFI). CFI splits the data at random and estimates the impact of the instrument on the exposure in each partition. These estimates are then used to perform an IVR on each partition. We adapt CFI to the Mendelian randomization (MR) setting and term this adaptation 'Cross-Fitting for Mendelian Randomization' (CFMR). We show that, even when using weak instruments, CFMR is, at worst, biased towards the null, which makes it a conservative one-sample MR approach. In particular, CFMR remains conservative even when the two samples used to perform the MR analysis completely overlap, whereas current state-of-the-art approaches (e.g., MR RAPS) display substantial bias in this setting. Another major advantage of CFMR lies in its use of all of the available data to select genetic instruments, which maximizes statistical power, as opposed to traditional two-sample MR where only part of the data is used to select the instrument. Consequently, CFMR is able to enhance statistical power in consortia-led meta-analyses by enabling a conservative one-sample MR to be performed in each cohort prior to a meta-analysis of the results across all the cohorts. In addition, CFMR enables a cross-ethnic MR analysis by accounting for ethnic heterogeneity, which is particularly important in meta-analyses where the participating cohorts may have different ethnicities. To our knowledge, none of the current MR approaches can account for such heterogeneity. Finally, CFMR enables the application of MR to exposures that are either rare or difficult to measure, which would normally preclude their analysis in the regular two-sample MR setting.

Smoking and infertility: multivariable regression and Mendelian randomization analyses in the Norwegian Mother, Father and Child Cohort Study

OBJECTIVE

To investigate the association between smoking and infertility.

DESIGN

Prospective study.

SETTING

Nationwide cohort.

PATIENTS

28,606 women and 27,096 men with questionnaire and genotype information from the Norwegian Mother, Father, and Child Cohort Study.

INTERVENTION

Self-reported information on smoking (having ever smoked [both sexes], age at initiation [women only], cessation [women only], and cigarettes/week in current smokers [both sexes]) was gathered. Genetically predetermined levels or likelihood of presenting these traits were estimated for Mendelian randomization.

MAIN OUTCOME MEASURE

Infertility (time-to-pregnancy ≥12 months).

RESULTS

Having ever smoked was unrelated to infertility in women or men. Higher smoking intensity in women was associated with greater infertility odds (+1 standard deviation [SD, 48 cigarettes/week]: odds ratio [OR]crude, 1.19; 95% confidence interval [CI] 1.11-1.28; ORadjusted 1.12; 95% CI, 1.03-1.21), also after adjusting for the partner's tobacco use. Later smoking initiation (+1 SD [3.2 years]: ORcrude, 0.94; 95% CI, 0.88-0.99; ORadjusted 0.89; 95% CI, 0.84-0.95) and smoking cessation (vs. not quitting: ORcrude, 0.83; 95% CI, 0.75-0.91; ORadjusted, 0.83; 95% CI, 0.75-0.93) were linked to decreased infertility in women. Nevertheless, Mendelian randomization results were not directionally consistent for smoking intensity and cessation and were estimated imprecisely in the 2-sample approach. In men, greater smoking intensity was not robustly associated with infertility in multivariable regression and Mendelian randomization.

CONCLUSIONS

We did not find robust evidence of an effect of smoking on infertility. This may be due to a true lack of effect, weak genetic instruments, or other kinds of confounding.

A theory-based practical solution to correct for sex-differential participation bias

Most genomic cohorts are retrospective where the exposures and outcomes are predetermined prior to sample collection. Therefore, a spurious association between an exposure and an outcome can arise if both variables affect study participation. Such concerns were raised in previous studies questioning the representativeness of the UK Biobank. Recently, a genome-wide association study (GWAS) on biological sex found many autosomal hits and non-negligible autosomal heritability which the authors attribute to selection bias. In this study, we propose a simple and a practical method that can overcome sex-driven selection bias based on theoretical analysis and simulations.

Associations of sleep and circadian phenotypes with COVID-19 susceptibility and hospitalization: an observational cohort study based on the UK Biobank and a two-sample Mendelian randomization study

Study Objectives

Sleep and circadian phenotypes are associated with several diseases. The present study aimed to investigate whether sleep and circadian phenotypes were causally linked with coronavirus disease 2019 (COVID-19)-related outcomes.

Methods

Habitual sleep duration, insomnia, excessive daytime sleepiness, daytime napping, and chronotype were selected as exposures. Key outcomes included positivity and hospitalization for COVID-19. In the observation cohort study, multivariable risk ratios (RRs) and their 95% confidence intervals (CIs) were calculated. Two-sample Mendelian randomization (MR) analyses were conducted to estimate the causal effects of the significant findings in the observation analyses. Beta values and the corresponding 95% CIs were calculated and compared using the inverse variance weighting, weighted median, and MR-Egger methods.

Results

In the UK Biobank cohort study, both often excessive daytime sleepiness and sometimes daytime napping were associated with hospitalized COVID-19 (excessive daytime sleepiness [often vs. never]: RR=1.24, 95% CI=1.02-1.5; daytime napping [sometimes vs. never]: RR=1.12, 95% CI=1.02-1.22). In addition, sometimes daytime napping was also associated with an increased risk of COVID-19 susceptibility (sometimes vs. never: RR= 1.04, 95% CI=1.01-1.28). In the MR analyses, excessive daytime sleepiness was found to increase the risk of hospitalized COVID-19 (MR IVW method: OR = 4.53, 95% CI = 1.04-19.82), whereas little evidence supported a causal link between daytime napping and COVID-19 outcomes.

Conclusions

Observational and genetic evidence supports a potential causal link between excessive daytime sleepiness and an increased risk of COVID-19 hospitalization, suggesting that interventions targeting excessive daytime sleepiness symptoms might decrease severe COVID-19 rate.

Assessing the Causal Role of Sleep Traits on Glycated Hemoglobin: A Mendelian Randomization Study

OBJECTIVE

To examine the effects of sleep traits on glycated hemoglobin (HbA1c).

RESEARCH DESIGN AND METHODS

This study triangulated evidence across multivariable regression (MVR) and one- (1SMR) and two-sample Mendelian randomization (2SMR) including sensitivity analyses on the effects of five self-reported sleep traits (i.e., insomnia symptoms [difficulty initiating or maintaining sleep], sleep duration, daytime sleepiness, napping, and chronotype) on HbA1c (in SD units) in adults of European ancestry from the UK Biobank (for MVR and 1SMR analyses) (n = 336,999; mean [SD] age 57 [8] years; 54% female) and in the genome-wide association studies from the Meta-Analyses of Glucose and Insulin-Related Traits Consortium (MAGIC) (for 2SMR analysis) (n = 46,368; 53 [11] years; 52% female).

RESULTS

Across MVR, 1SMR, 2SMR, and their sensitivity analyses, we found a higher frequency of insomnia symptoms (usually vs. sometimes or rarely/never) was associated with higher HbA1c (MVR 0.05 SD units [95% CI 0.04-0.06]; 1SMR 0.52 [0.42-0.63]; 2SMR 0.24 [0.11-0.36]). Associations remained, but point estimates were somewhat attenuated after excluding participants with diabetes. For other sleep traits, there was less consistency across methods, with some but not all providing evidence of an effect.

CONCLUSIONS

Our results suggest that frequent insomnia symptoms cause higher HbA1c levels and, by implication, that insomnia has a causal role in type 2 diabetes. These findings could have important implications for developing and evaluating strategies that improve sleep habits to reduce hyperglycemia and prevent diabetes.