Mendelian Randomization: Methods for Using Genetic Variants in Causal Estimation

by Stephen Burgess, Simon G. Thompson

Overview

Presents the Terminology and Methods of Mendelian Randomization for Epidemiological Studies

Mendelian randomization uses genetic instrumental variables to make inferences about causal effects based on observational data. It, therefore, can be a reliable way of assessing the causal nature of risk factors, such as biomarkers, for a wide range of disease outcomes.

Mendelian Randomization: Methods for Using Genetic Variants in Causal Estimation provides thorough coverage of the methods and practical elements of Mendelian randomization analysis. It brings together diverse aspects of Mendelian randomization spanning epidemiology, statistics, genetics, and econometrics. Although the book mainly focuses on epidemiology, much of the material can be applied to other areas of research.

Through several examples, the first part of the book shows how to perform simple applied Mendelian randomization analyses and interpret their results. The second part addresses specific methodological issues, such as weak instruments, multiple instruments, power calculations, and meta-analysis, relevant to practical applications of Mendelian randomization. In this part, the authors draw on data from the C-reactive protein Coronary heart disease Genetics Collaboration (CCGC) to illustrate the analyses. They present the mathematics in an easy-to-understand way by using nontechnical language and reinforcing key points at the end of each chapter. The last part of the book examines the potential of Mendelian randomization in the future, exploring both methodological and applied developments.

This book gives statisticians, epidemiologists, and geneticists the foundation to understand issues concerning the use of genetic variants as instrumental variables. It will get them up to speed in undertaking and interpreting Mendelian randomization analyses. Chapter summaries, paper summaries, web-based applications, and software code for implementing the statistical techniques are available on a supplementary website.

Product Details

ISBN-13:	9781466573178
Publisher:	Taylor & Francis
Publication date:	03/06/2015
Series:	Chapman & Hall/CRC Interdisciplinary Statistics
Pages:	224
Product dimensions:	5.90(w) x 8.90(h) x 0.60(d)

Table of Contents

Using Genetic Variants as Instrumental Variables to Assess Causal Relationships
Introduction and motivation
Shortcomings of classical epidemiology
The rise of genetic epidemiology
Motivating example: The inflammation hypothesis
Other examples of Mendelian randomization
Overview of book
Summary

What is Mendelian randomization?
What is Mendelian randomization?
Why use Mendelian randomization?
A brief overview of genetics
Summary

Assumptions for causal inference
Observational and causal relationships
Finding a valid instrumental variable
Testing for a causal relationship
Estimating a causal effect
Summary

Methods for instrumental variable analysis
Ratio of coefficients method
Two-stage methods
Likelihood-based methods
Semi-parametric methods
Efficiency and validity of instruments
Computer implementation
Summary

Examples of Mendelian randomization analysis
Fibrinogen and coronary heart disease
Adiposity and blood pressure
Lipoprotein(a) and myocardial infarction
High-density lipoprotein cholesterol and myocardial infarction
Discussion

Generalizability of estimates from Mendelian randomization
Internal and external validity
Comparison of estimates
Discussion
Summary

Statistical Issues in Instrumental Variable Analysis and Mendelian Randomization
Weak instruments and finite-sample bias
Introduction
Demonstrating the bias of IV estimates
Explaining the bias of IV estimates
Properties of IV estimates with weak instruments
Bias of IV estimates with different choices of IV
Minimizing the bias of IV estimates
Discussion
Key points from chapter

Multiple instruments and power
Introduction
Allele scores
Power of IV estimates
Multiple variants and missing data
Discussion
Key points from chapter

Multiple studies and evidence synthesis
Introduction
Assessing the causal relationship
Study-level meta-analysis
Summary-level meta-analysis
Individual-level meta-analysis
Example: C-reactive protein and fibrinogen
Binary outcomes
Discussion
Key points from chapter

Example: The CRP CHD Genetics Collaboration
Overview of the dataset
Single study: Cardiovascular Health Study
Meta-analysis of all studies
Discussion
Key points from chapter

Prospects for Mendelian Randomization
Future directions
Methodological developments
Applied developments
Conclusion

Bibliography

Index

Editorial Reviews

"Mendelian randomization (MR) may be regarded as a modern integration of the concepts of instrumental variables from the field of economics with Mendel’s laws of inheritance in the field of genetics. … Stephen Burgess, a mathematician, and Simon Thompson, a biostatistician, present the fundamental ideas and methods behind MR in an easy-to-read book for researchers from different backgrounds such as statistics, genetics, or clinical research. The book is the first and currently only book published on MR, in which the authors not only summarize MR methodology but also share their experience with real data analysis in the field of MR for cardiovascular diseases. … The book has been very well written, and the authors have done an excellent job of bridging the gap between researchers from various backgrounds. … Overall, the book provides the most comprehensive MR resource to date, and it is a valuable resource for researchers even at the graduate level."
—Sandeep Grover, Universität zu Lübeck, in Biometrical Journal, September 2017

"… The topic of the book is Mendelian Randomization (MR), a form of instrument variable (IV) analysis using genetic factors as instruments. … I enjoyed reading the book very much. The authors give an excellent overview of the assumptions and statistics underlying MR and the book achieves a good balance between methods and theory, case studies and the discussion of practical issues. … I particularly liked the fact that most if not all of the relevant models, methods and analytical derivations on IV analyses are in a single book. The authors are very precise in their definitions and discuss the issue of when ‘causal’ can be used in the context of inference and, importantly, when not. … In summary, MR analyses have an important role to play in making sense of epidemiological observations and we can expect a plethora of applications in the near future. This book is a thorough practical guide to their assumptions, inference and pitfalls."
—Peter M. Visscher, University of Queensland, in Australian & New Zealand Journal of Statistics, June 2017

"The authors have aimed their book at epidemiologists and medical statisticians but anyone with a basic knowledge of regression will understand most of the contents, because the algebra is kept to a minimum and emphasis is on explaining the ideas that underlie MR…will serve as an excellent introduction to Mendelian randomization for anyone who wants to understand the underlying statistical issues…"
—John Thompson, University of Leicester, in Biometrics, March 2017

"Mendelian Randomization, by Stephen Burgess and Simon Thompson, represents a compact and accessible resource for Mendelian randomization, providing exactly what one needs to know in a logical, clear, very thorough, and yet pragmatic way. This book will appeal to applied researchers interested in learning more about Mendelian randomization as well as methodological researchers who work in the area. Those new to the field will find that this book covers everything they need to know, from designing the study (e.g., choosing the "instruments" and considering different options for pooling data across multiple studies) to investigating whether the assumptions may hold, analyzing the data, and interpreting the results… Researchers interested in methods for Mendelian randomization will find this book equally useful in bringing together methodological findings published across different disciplines (genetics, epidemiology, statistics, and econometrics), in an articulate and comprehensive way… Being targeted to epidemiologists and medical statisticians with diverse backgrounds, no prior knowledge of genetics is required, and the book explains in very simple terms the basic concepts needed to understand and apply Mendelian randomization… I really enjoyed reading this book and highly recommend it to anyone with an interest in Mendelian randomization."
—Cosetta Minelli, Imperial College London, in The American Statistician, August 2016

From the Publisher