Marlon Cerf

Eduardo Franco

eLife Assessment
The study presents some useful findings on Mendelian randomization-phenome-wide association, with BMI associated with health outcomes, and there is a focus on sex differences. Although there are some solid phenotype and genotype data, some of the data are incomplete and could be better presented, perhaps benefiting from more rigorous approaches. Confirmation and further assessment of the observed sex differences will add further value.

anonymous

Reviewer #1 (Public review):
Summary:
This study uses information from the UK Biobank and aims to investigate the role of BMI on various health outcomes, with a focus on differences by sex. They confirm the relevance of many of the well-known associations between BMI and health outcomes for males and females and suggest that associations for some endpoints may differ by sex. Overall their conclusions appear supported by the data. The significance of the observed sex variations will require confirmation and further assessment.
Strengths:
This is one of the first systematic evaluations of sex differences between BMI and health outcomes.
The hypothesis that BMI may be associated with health differentially based on sex is relevant and even expected. As muscle is heavier than adipose tissue, and as men typically have more muscle than women, as a body composition measure BMI is sometimes prone to classifying even normal weight/muscular men as obese, while this measure is more lenient when used in women.
Confirmation of the many well-known associations is as expected and attests to the validity of their approach.
Demonstration of the possible sex differences is interesting, with this work raising the need for further study.
Weaknesses:
Many of the statistical decisions appeared to target power at the expense of quality/accuracy. For example, they chose to use self-reported information rather than doctor diagnoses for disease outcomes for which both types of data were available.
Despite known problems and bias arising from the use of one sample approach, they chose to use instruments from the UK Biobank instead of those available from the independent GIANT GWAS, despite the difference in sample size being only marginally greater for UKB for the context. With the way the data is presented, it is difficult to assess the extent to which results are compatible across approaches.
The approach to multiple testing correction appears very lenient, although the lack of accuracy in the reporting makes it difficult to know what was done exactly. The way it reads, FDR correction was done separately for men, and then for women (assuming that the duplication in tests following stratification does not affect the number of tests). In the second stage, they compared differences by sex using Z-test, apparently without accounting for multiple testing.
Presentation lacks accuracy in a few places, hence assessment of the accuracy of the statements made by the authors is difficult.
Conclusion "These findings highlight the importance of retaining a healthy BMI" is rather uninformative, especially as they claim that for some attributes the effects of BMI may be opposite depending on sex/gender.

Reviewer #2 (Public review):
Summary:
In this present Mendelian randomization-phenome-wide association study, the authors found BMI to be positively associated with many health-related conditions, such as heart disease, heart failure, and hypertensive heart disease. They also found sex differences in some traits such as cancer, psychological disorders, and ApoB.
Strengths:
The use of the UK-biobank study with detailed phenotype and genotype information.
Weaknesses:
Previous studies have performed this analysis using the same cohort, with in-depth analysis. See this paper: Searching for the causal effects of body mass index in over 300,000 participants in UK Biobank, using Mendelian randomization. <a href="https://journals.plos.org/plosgenetics/article?id=10.1371/journal.pgen.1007951">https://journals.plos.org/plosgenetics/article?id=10.1371/journal.pgen.1007951</a>
I believe that the authors' claim, "To our knowledge, no sex-specific PheWAS has investigated the effects of BMI on health outcomes," is not well supported. They have not cited a relevant paper that conducted both overall and sex-stratified PheWAS using UK Biobank data with a detailed analysis. Given the prior study linked above, I am uncertain about the additional contributions of the present research.

A Sex-specific Mendelian Randomization-Phenome-Wide Association Study of Body Mass Index

Figures and data

Be the first to read new articles from eLife