Improving the visualization, interpretation and analysis of two-sample summary data Mendelian randomization via the Radial plot and Radial regression.

Bowden J, Spiller W, Del Greco M F, Sheehan N, Thompson J, Minelli C, Davey Smith G. · 2018

Read Improving the visualization, interpretation and analysis of two-sample summary data Mendelian randomization via the Radial plot and Radial regression. by Bowden J, Spiller W, Del Greco M F, Sheehan N, Thompson J, Minelli C, Davey Smith G. (2018) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/europmc-pmc6124632.txt.

Introduction

Bowden and co-authors present a methods article in International Journal of Epidemiology on Mendelian randomization studies. These studies use gene variants to test cause and effect. Results are often shown on a scatter plot. That plot can mislead when some estimates are far less precise than others.

The authors propose a radial plot instead. It makes outliers and strong data points easier to spot. It also removes the need to recode gene data. The plot leads to a new form of radial regression. They show the method on blood pressure and heart disease data. One outlying variant explained a clash between two common estimates.

Catalog record

CREATORBowden J, Spiller W, Del Greco M F, Sheehan N, Thompson J, Minelli C, Davey Smith G.
YEAR2018
COLLECTIONPAPERS
FORMATTXT
FILE SIZE45 KB
LICENSECC BY 4.0
RIGHTSCC BY
SOURCEEurope PMC

Source: https://europepmc.org/articles/PMC6124632

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Improving the visualization, interpretation and analysis of two-sample summary data Mendelian randomization via the Radial plot and Radial regression By Jack Bowden, Wesley Spiller, Fabiola Del Greco M, Nuala Sheehan, John Thompson, Cosetta Minelli, George Davey Smith Published in: International Journal of Epidemiology DOI: 10.1093/ije/dyy101 ABSTRACT Abstract Background Summary data furnishing a two-sample Mendelian randomization (MR) study are often visualized with the aid of a scatter plot, in which single-nucleotide polymorphism (SNP)–outcome associations are plotted against the SNP–exposure associations to provide an immediate picture of the causal-effect estimate for each individual variant. It is also convenient to overlay the standard inverse-variance weighted (IVW) estimate of causal effect as a fitted slope, to see whether an individual SNP provides evidence that supports, or conflicts with, the overall consensus. Unfortunately, the traditional scatter plot is not the most appropriate means to achieve this aim whenever SNP–outcome associations are estimated with varying degrees of precision and this is reflected in the analysis. Methods We propose instead to use a small modification of the scatter plot—the Galbraith Radial plot—for the presentation of data and results from an MR study, which enjoys many advantages over the original method. On a practical level, it removes the need to recode the genetic data and enables a more straightforward detection of outliers and influential data points. Its use extends beyond the purely aesthetic, however, to suggest a more general modelling framework to operate within when conducting an MR study, including a new form of MR-Egger regression. Results We illustrate the methods using data from a two-sample MR study to probe the causal effect of systolic blood pressure on coronary heart disease risk, allowing for the possible effects of pleiotropy. The Radial plot is shown to aid the detection of a single outlying variant that is responsible for large differences between IVW and MR-Egger regression estimates. Several additional plots are also proposed for informative data visualization. Conclusions The Radial plot should be considered in place of the scatter plot for visualizing, analysing and interpreting data from a two-sample summary data MR study. Software is provided to help facilitate its use. Keywords: Two sample summary data Mendelian randomization, Scatter plot, Heterogeneity statistics, Radial plot, Radial MR-Egger Key MessagesSummary data furnishing a two-sample Mendelian randomization (MR) study are often visualized with the aid of a scatter plot. The scatter plot is also used to interpret the validity of the standard inverse-variance weighted (IVW) estimate and pleiotropy robust methods such as MR-Egger regression.A close relation of the scatter plot—the Radial plot—can instead be used for this purpose.The Radial plot removes the need to pre-process the summary data (a pre-requisite for MR-Egger), improves the detection of outliers and influential data points in either an IVW or MR-Egger analysis, and can incorporate any set of weights desired by the user.A more general form of MR-Egger regression is proposed that flows from, and naturally compliments, the Radial plot.Radial funnel and leave-one-out analysis plots can also be used to aid the visualization and interpretation of MR studies. BACKGROUND Mendelian randomization (MR)1 is a methodological framework for probing questions of causality in observational epidemiology using genetic data—typically in the form of single-nucleotide polymorphisms (SNPs)—to infer whether a modifiable risk factor truly influences a health outcome. A particular MR study design gaining in popularity combines publically available data on SNP–exposure and SNP–outcome associations from separate but homogeneous cohort studies of unrelated individuals for large numbers of uncorrelated SNPs. Each SNP is used to estimate the causal effect under the primary assumption that it is a valid instrumental variable (IV), by dividing its SNP–outcome association by its SNP–exposure association to yield the ratio estimate. Secondary modelling assumptions are also required in order for this estimate to be consistent. Ratio estimates are then combined into an overall estimate of causal effect using an inverse-variance weighted (IVW) fixed-effect meta-analysis. This is referred to as the IVW estimate and the general framework as two-sample summary data MR.2,3 For further details, see Box 1. Box 1: Standard two-sample summary data MR analysis The IV assumptions: The canonical approach to MR assumes that the group of SNPs are valid IVs for the purposes of inferring the causal effect of an exposure, X, on an outcome, Y. That is, they are: associated with X (IV1); not associated with any confounders of X and Y (IV2); and can only be associated with Y through X (IV3). The IV assumptions are represented by the solid lines in the causal diagram below for a SNP Gj, with unobserved confounding represented by U. Dotted lines represent dependencies between G and U, and G and Y that are prohibited by the IV assumptions. The causal effect of a unit increase in X on the outcome Y, denoted by β, is the quantity we are aiming to estimate. The ratio estimate: Assume that exposure X causally affects outcome Y linearly across all values of X, so that a hypothetical intervention that induced a 1-unit increase in X would induce a β increase in Y. Suppose also that all L SNPs predict the exposure via an additive linear model with no interactions. If SNP j is a valid IV, and the two study samples are homogeneous, then the underlying SNP–outcome association from sample 1, Γj, should be a scalar multiple of the underlying SNP–exposure association estimate from sample 2, γj, the scalar multiple being the causal effect β. That is: Γj=βγj.The ratio estimate for the causal effect of X on Y using SNP j (out of L), β^j = Γ^j/γ^j, where Γ^j is the estimate for SNP j’s association with the outcome (with standard error σYj) and γ^j is the estimate for SNP j’s association with the exposure (with standard error σXj). The IVW estimate: The overall IVW estimate for the causal effect obtained across L uncorrelated SNPs is then given by β^IVW=∑j=1Lwjβ^j∑j=1Lwj, where wj is the inverse-variance of β^j. Two popular choices for the inverse-variance weights are 1st order (fixed effect) weights: wj=γ^j2σYj22nd order (fixed effect) weights: wj=(σYj2γ^j2+Γ^j2σXj2γ^j4)−1. When SNP–exposure association estimates are sufficiently precise, so that σXj2 is negligible, or the causal effect β is small, then both weighting schemes are very similar. When this is not the case, both first- and second-order weights can perform poorly in terms of causal estimation and heterogeneity detection (see Box 2). ‘Iterative' and ‘exact’ weighting has recently been proposed by Bowden et al.4 to address this issue. For simplicity, first-order weights will be used throughout this paper. In our example, the IVW estimate obtained using first-order weights is 0.053. This represents the causal effect of a 1-mmHg increase in SBP on the log-odds ratio of CHD (see Table 1). Different formulae for the inverse-variance weights can be employed, the most popular being simple ‘first-order’ weights, which assume the uncertainty in the SNP–exposure association estimates is negligible. Although more sophisticated weighting approaches have recently been proposed,4 for simplicity, we will use first-order weights throughout this paper. The scatter plot Figure 1 shows a traditional scatter plot of summary data estimates for the associations of 26 genetic variants with systolic blood pressure (SBP, the exposure) and coronary heart disease (CHD, the outcome). SNP–SBP association estimates were obtained from the International Consortium for Blood Pressure consortium

Read the complete text