Strategies to investigate and mitigate collider bias in genetic and Mendelian randomisation studies of disease progression.

Genetic studies of disease progression can be used to identify factors that may influence survival or prognosis, which may differ from factors that influence on disease susceptibility. Studies of disease progression feed directly into therapeutics for disease, whereas studies of incidence inform prevention strategies. However, studies of disease progression are known to be affected by collider (also known as "index event") bias since the disease progression phenotype can only be observed for individuals who have the disease. This applies equally to observational and genetic studies, including genome-wide association studies and Mendelian randomisation (MR) analyses. In this paper, our aim is to review several statistical methods that can be used to detect and adjust for index event bias in studies of disease progression, and how they apply to genetic and MR studies using both individual- and summary-level data. Methods to detect the presence of index event bias include the use of negative controls, a comparison of associations between risk factors for incidence in individuals with and without the disease, and an inspection of Miami plots. Methods to adjust for the bias include inverse probability weighting (with individual-level data), or Slope-Hunter and Dudbridge et al.'s index event bias adjustment (when only summary-level data are available). We also outline two approaches for sensitivity analysis. We then illustrate how three methods to minimise bias can be used in practice with two applied examples. Our first example investigates the effects of blood lipid traits on mortality from coronary heart disease, while our second example investigates genetic associations with breast cancer mortality.

Investigating the effect of sexual behaviour on oropharyngeal cancer risk: a methodological assessment of Mendelian randomization.

BACKGROUND: Human papilloma virus infection is known to influence oropharyngeal cancer (OPC) risk, likely via sexual transmission. However, sexual behaviour has been correlated with other risk factors including smoking and alcohol, meaning independent effects are difficult to establish. We aimed to evaluate the causal effect of sexual behaviour on the risk of OPC using Mendelian randomization (MR). METHODS: Genetic variants robustly associated with age at first sex (AFS) and the number of sexual partners (NSP) were used to perform both univariable and multivariable MR analyses with summary data on 2641 OPC cases and 6585 controls, obtained from the largest available genome-wide association studies (GWAS). Given the potential for genetic pleiotropy, we performed a number of sensitivity analyses: (i) MR methods to account for horizontal pleiotropy, (ii) MR of sexual behaviours on positive (cervical cancer and seropositivity for Chlamydia trachomatis) and negative control outcomes (lung and oral cancer), (iii) Causal Analysis Using Summary Effect estimates (CAUSE), to account for correlated and uncorrelated horizontal pleiotropic effects, (iv) multivariable MR analysis to account for the effects of smoking, alcohol, risk tolerance and educational attainment. RESULTS: In univariable MR, we found evidence supportive of an effect of both later AFS (IVW OR = 0.4, 95%CI (0.3, 0.7), per standard deviation (SD), p = < 0.001) and increasing NSP (IVW OR = 2.2, 95%CI (1.3, 3.8) per SD, p = < 0.001) on OPC risk. These effects were largely robust to sensitivity analyses accounting for horizontal pleiotropy. However, negative control analysis suggested potential violation of the core MR assumptions and subsequent CAUSE analysis implicated pleiotropy of the genetic instruments used to proxy sexual behaviours. Finally, there was some attenuation of the univariable MR results in the multivariable models (AFS IVW OR = 0.7, 95%CI (0.4, 1.2), p = 0.21; NSP IVW OR = 0.9, 95%CI (0.5 1.7), p = 0.76). CONCLUSIONS: Despite using genetic variants strongly related sexual behaviour traits in large-scale GWAS, we found evidence for correlated pleiotropy. This emphasizes a need for multivariable approaches and the triangulation of evidence when performing MR of complex behavioural traits.

Genetic Analyses of Common Infections in the Avon Longitudinal Study of Parents and Children Cohort.

The burden of infections on an individual and public health is profound. Many observational studies have shown a link between infections and the pathogenesis of disease; however a greater understanding of the role of host genetics is essential. Children from the longitudinal birth cohort, the Avon Longitudinal Study of Parents and Children, had 14 antibodies measured in plasma at age 7: Alpha-casein protein, beta-casein protein, cytomegalovirus, Epstein-Barr virus, feline herpes virus, Helicobacter pylori, herpes simplex virus 1, influenza virus subtype H1N1, influenza virus subtype H3N2, measles virus, Saccharomyces cerevisiae, Theiler's virus, Toxoplasma gondii, and SAG1 protein domain, a surface antigen of Toxoplasma gondii measured for greater precision. We performed genome-wide association analyses of antibody levels against these 14 infections (N = 357 - 5010) and identified three genome-wide signals (P < 5×10-8), two associated with measles virus antibodies and one with Toxoplasma gondii antibodies. In an association analysis focused on the human leukocyte antigen (HLA) region of the genome, we further detected 15 HLA alleles at a two-digit resolution and 23 HLA alleles at a four-digit resolution associated with five antibodies, with eight HLA alleles associated with Epstein-Barr virus antibodies showing strong evidence of replication in UK Biobank. We discuss how our findings from antibody levels complement other studies using self-reported phenotypes in understanding the architecture of host genetics related to infections.

What Change in Body Mass Index Is Required to Improve Cardiovascular Outcomes in Childhood and Adolescent Obesity through Lifestyle Interventions: A Meta-Regression.

Background: Using meta-regression, this article aims at establishing the minimum change in BMI-standard deviation score (SDS) needed to improve lipid profiles and blood pressure in children and adolescents with obesity, to aid future trials and guidelines. Methods: Studies with participants involved in lifestyle interventions, aged 4-19 years, with a diagnosis of obesity according to defined BMI thresholds, were considered for inclusion in a large systematic review. Interventions had to report pre- and post-intervention (or mean change in) BMI-SDS, plus either systolic blood pressure (SBP), high-density lipoprotein (HDL) cholesterol, low-density lipoprotein (LDL) cholesterol, and/or triglycerides (TGs). Random-effects meta-regression quantified the relationship between mean change in BMI-SDS and mean change in cardiovascular outcomes. Results: Seventy-one papers reported various cardiovascular measurements and mean change in BMI-SDS. Fifty-four, 59, 46, and 54 studies were analyzed, reporting a change in SBP, HDL, LDL, and TG, respectively. Reduction in mean BMI-SDS was significantly related to improvements in SBP, LDL, TG, and HDL (p < 0.05); BMI-SDS reductions of 1, 1.2, and 0.7 ensured a mean reduction of SBP, LDL, and TG, respectively, although an equivalent value for HDL improvement was indeterminate. Conclusion: Reductions in mean BMI-SDS of >1, >1.2, or >0.7 are likely to reduce SBP, LDL, and TG, respectively. Further studies are needed to clarify the optimal duration, intensity, and setting for interventions. Consistency is required regarding derived BMI values to facilitate future systematic reviews and meta-analyses.