Search

Causal discovery is a powerful tool to disclose underlying structures by analyzing purely observational data. Genetic variants can provide useful complementary information for structure learning. Recently, Mendelian randomization (MR) studies have provided abundant marginal causal relationships of traits. Here, we propose a causal network pruning algorithm MRSL (MR-based structure learning algorithm) based on these marginal causal relationships. MRSL combines the graph theory with multivariable MR to learn the conditional causal structure using only genome-wide association analyses (GWAS) summary statistics. Specifically, MRSL utilizes topological sorting to improve the precision of structure learning. It proposes MR-separation instead of d-separation and three candidates of sufficient separating set for MR-separation. The results of simulations revealed that MRSL had up to 2-fold higher F1 score and 100 times faster computing time than other eight competitive methods. Furthermore, we applied MRSL to 26 biomarkers and 44 International Classification of Diseases 10 (ICD10)-defined diseases using GWAS summary data from UK Biobank. The results cover most of the expected causal links that have biological interpretations and several new links supported by clinical case reports or previous observational literatures. [ABSTRACT FROM AUTHOR]

Causal mediation analysis investigates the mechanism linking exposure and outcome. Dealing with the impact of unobserved confounders among exposure, mediator and outcome is an issue of great concern. Moreover, when multiple mediators exist, this causal pathway intertwines with other causal pathways, rendering it difficult to estimate the path-specific effects. In this study, we propose a method (PSE-MR) to identify and estimate path-specific effects of an exposure (e.g. education) on an outcome (e.g. osteoarthritis risk) through multiple causally ordered and non-ordered mediators (e.g. body mass index and pack-years of smoking) using summarized genetic data, when the sequential ignorability assumption is violated. Specifically, PSE-MR requires a specific rank condition in which the number of instrumental variables is larger than the number of mediators. Furthermore, we illustrate the utility of PSE-MR by providing guidance for practitioners and exploring the mediation effects of body mass index and pack-years of smoking in the causal pathways from education to osteoarthritis risk. Additionally, the results of simulation reveal that the causal estimates of path-specific effects are almost unbiased with good coverage and Type I error properties. Also, we summarize the least number of instrumental variables for the specific number of mediators to achieve 80% power. [ABSTRACT FROM AUTHOR]

Some studies reported that mildly elevated serum bilirubin levels were associated with decreased risk of cardiovascular disease (CVD) and diabetes. Whether these are causal relationships remains unclear. This study aims to examine the causal effects of bilirubin on CVD, diabetes and their subtypes. The data we used in this study includes individual data from the UK Biobank cohort with 331,002 white British participants, and summary data from published genome wide associations studies (GWAS) findings. We used individual data to perform logistic regression for the observational study and two-stage least squares method for the Mendelian randomization (MR) study. We also performed several traditional MR methods and MR-TRYX by summary data. The observational study supported the association relationships between bilirubin and CVD and diabetes and their subtypes. Results of MR showed strong evidence for negative causal associations of log e total bilirubin with CVD [OR 0.92, 95%CI 0.88–0.95, p -value 2.15 × 10−6], coronary heart disease [OR 0.90, 95%CI 0.85–0.96, p -value 1.54 × 10−3] and hypertensive diseases [OR 0.91, 95%CI 0.88–0.95, p -value 5.89 × 10−6], but no evidence for diabetes [OR 0.94, 95%CI 0.86–1.02, p -value 0.14] and its subtypes. We also obtained similar results for direct bilirubin. We found that blood pressure, cholesterol, C-reactive protein, alcohol and white blood cell count played important roles in the causal pathway from bilirubin to CVD. Two sample MR and sensitivity analyses showed consistent results with one sample MR. Genetically determined bilirubin was negatively associated with the risk of CVD but had no evident causal association with diabetes in the UK Biobank cohort of white British. • Genetic determined bilirubin is negatively associated with the risk of cardiovascular disease (CVD). • Bilirubin has no evident causal association with diabetes in white British. • We identified the mediators of the causal pathway from bilirubin to CVD. [ABSTRACT FROM AUTHOR]

Although observational studies have shown an association between sex hormone-binding globulin (SHBG), testosterone (T) and cardiovascular diseases (CVD), controversy remains. In this study, we aim to explore the causal effects of SHBG and T on Coronary heart disease (CHD). We used univariable, network and multivariable mendelian randomization (MR) analysis to investigate the causal effect of SHBG and T on CHD. We performed inverse variance weighted (IVW) MR as the primary analysis, with the robustness of this approach further tested by other methods in sensitivity analysis. The SHBG and T were collected from the UK Biobank data, about 180,000 men aged 40 to 69 years. CHD was collected from CARDIoGRAMplusC4D 1000 Genomes-based GWAS, which was a meta-analysis including 48 studies and involving 60,801 CHD cases and 123,504 controls. Using univariable MR-IVW, the results suggested that a one standard deviation (SD) increase in SHBG, the risk of CHD decreased by approximately 14% (OR (95% CI): 0.86(0.76,0.97)), and that a SD increase in total testosterone (TT), the risk also decreased, approximately 8% (OR (95% CI): 0.92(0.85,0.99)). Multivariable MR showed that both SHBG and TT had no direct causal effect with CHD (a SD increase in SHBG: OR (95% CI):0.75(0.57,1.00), P = 0.053; a SD increase in TT: OR (95% CI): 1.05(0.90,1.22), P = 0.53). In the network MR analysis, the results suggested that TT might act as mediator in the causal pathway from SHBG to CHD and account for 93% of the total effect of SHBG on CHD, and that SHBG might be a mediator in the causal pathway from TT to CHD and account for 67% of the total effect of TT on CHD. Genetically predicted SHBG and TT were negatively correlated with CHD in both univariable and network MR, which may provide a causal explanation behind the observed conclusion. In addition, TT and SHBG had a bidirectional causal effect. Further work is required to disentangle the downstream effects of SHBG/TT on CHD and the molecular pathways involved, as the simultaneous regulation of SHBG and TT may make it a viable strategy for the prevention or treatment of CHD. • Multivariable MR showed that both SHBG and TT had no direct causal effect with CHD (a SD increase in SHBG: OR (95%CI):0.75(0.57,1.00), P = 0.053; a SD increase in TT: OR (95%CI): 1.05(0.90,1.22), P = 0.53). • In the network MR analysis, the results suggested that TT might act as mediator in the causal pathway from SHBG to CHD and account for 93% of the total effect of SHBG on CHD, and that SHBG might be a mediator in the causal pathway from TT to CHD and account for 67% of the total effect of TT on CHD. [ABSTRACT FROM AUTHOR]