key: cord-0279537-t4s9zumw
authors: Yoshiji, S.; Tanaka, D.; Minamino, H.; Murakami, T.; Fujita, Y.; Richards, J. B.; Inagaki, N.
title: Causal associations between body fat accumulation and COVID-19 severity: A Mendelian randomization study
date: 2022-01-21
journal: nan
DOI: 10.1101/2022.01.20.22269593
sha: a48e12402004914695a3b6aff86a042e604b7196
doc_id: 279537
cord_uid: t4s9zumw

Purpose: The causal effects of body fat mass and body fat-free mass on coronavirus disease 2019 (COVID-19) severity remain unclear. Here, we used Mendelian randomization (MR) to evaluate the causal relationships between body fat-related traits and COVID-19 severity. Material and Methods: We identified single nucleotide polymorphisms associated with body mass index (BMI) and direct measures of body fat (i.e., body fat percentage, body fat mass, and body fat-free mass) in 461,460, 454,633, 454,137, and 454,850 individuals of European ancestry from the UK Biobank, respectively. We then performed two-sample MR to ascertain their effects on severe COVID-19 (cases: 4,792; controls: 1,054,664) from the COVID-19 Host Genetics Initiative. Results: We found that an increase in BMI, body fat percentage, and body fat mass by one standard deviation were each associated with severe COVID-19 (odds ratio (OR)BMI = 1.49, 95%CI: 1.19-1.87, P = 5.57x10-4; ORbody fat percentage = 1.94, 95%CI: 1.41-2.67, P = 5.07x10-5; and ORbody fat mass = 1.61, 95%CI: 1.28-2.04, P = 5.51x10-5). Further, we evaluated independent causal effects of body fat mass and body fat-free mass using multivariable MR and revealed that only body fat mass was independently associated with severe COVID-19 (ORbody fat mass = 2.91, 95%CI: 1.71-4.96, P = 8.85x10-5 and ORbody fat-free mass = 1.02, 95%CI: 0.61-1.67, P = 0.945). Conclusions: This study demonstrates the causal effects of body fat accumulation on COVID-19 severity and indicates that the biological pathways influencing the relationship between COVID-19 and obesity are likely mediated through body fat mass.

analysis, or interpretation of data; the writing of the report; or the decision to submit this paper 57 for publication. The other authors declare no conflict of interests. 58 59 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted January 21, 2022. 

A major risk factor for COVID-19 appears to be obesity. A community-based cohort 87 study involving 6.9 million individuals in England showed a positive association between body 88 mass index (BMI) and COVID-19 severity (2). In that study, BMI was significantly associated 89 with hospital admission, admission to an intensive care unit, and death due to COVID-19, with 90 these findings being replicated in other independent observational studies (3, 4). However, 91 these results do not support causation; in fact, interpreting these observations as a causal 92 relationship relies on untestable and usually implausible assumptions, including the absence 93 of unmeasured confounders and reverse causation (5). Given these limitations inherent to 94 traditional observational epidemiology studies, Mendelian randomization (MR) has emerged 95 as a way to mitigate against such shortcomings through its use of genetic variants as 96 instrumental variables in order to infer a causal relationship between exposures and outcomes 97 (6, 7). Using MR, we can estimate the causal effects of genetically predicted levels of 98 adiposity-related exposures on COVID-19 outcomes, in contrast to typical observational 99 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted January 21, 2022. ; https://doi.org/10.1101/2022.01.20.22269593 doi: medRxiv preprint 8 studies that evaluate only associations. Because genetic alleles are randomly assigned at 100 conception, which is generally well before the onset of the disease, the risk of reverse 101 causation is substantially decreased.

The selection of proxy measures of adiposity plays a vital role in evaluating the 103 association between obesity and COVID-19 outcomes. BMI can be measured easily and is 104 therefore a common measurement of obesity in epidemiological studies. However, the key 105 limitation of BMI is that it is an indirect measure of obesity because it is calculated only with 106 height and weight and does not consider body composition (i.e., body fat mass, body fat-free 107 mass, and their ratio) (8). Therefore, direct measures of body fat accumulation (i.e., body fat 108 percentage and body fat mass) might better elucidate the association of body fat with 109 COVID-19 outcomes. Taking this into consideration, we used a dataset of the UK Biobank, 110 which is a prospective cohort in the UK involving ~500,000 individuals (463,844 of which are of 111 European ancestry) with detailed genetic and phenotypic information. Data from the UK 112 Biobank includes measurements of not only BMI but also body fat mass and body fat-free 113 mass acquired through the bioelectrical impedance analysis and dual-energy X-ray 114 absorptiometry along with the results from the genome-wide association studies (GWAS) for 115 these traits (9). Recent MR studies utilized the results from the UK Biobank to obtain single 116 nucleotide polymorphisms (SNPs) associated with body fat accumulation and showed 117 estimated causal effects of body fat on various traits ranging from cardiovascular diseases to 118 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted January 21, 2022. ; https://doi.org/10.1101/2022.01.20.22269593 doi: medRxiv preprint 9 depression (10-13). In light of these promising findings, body fat mass has emerged as a 119 valuable indicator of the deleterious effects of fat accumulation; however, it remains unclear 120 whether fat mass and fat percentage are causally related to severe outcomes of COVID-19.

In this study, we therefore conducted a two-sample MR to assess whether BMI and 

Integrative Epidemiology Unit at the University of Bristol (14). The fat mass and fat-free mass 137 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted January 21, 2022. ; https://doi.org/10.1101/2022.01.20.22269593 doi: medRxiv preprint of the UK Biobank participants were evaluated by performing bioelectrical impedance analysis 138 using the Tanita BC418MA body composition analyzer (Tanita, Tokyo, Japan). We restricted 139 the analyses to individuals of European ancestry in order to maximize the statistical power, 

. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted January 21, 2022. ; https://doi.org/10.1101/2022.01.20.22269593 doi: medRxiv preprint the severe COVID-19 group was defined as individuals whose death was due to COVID-19, 157 those requiring respiratory support, or those requiring hospitalization due to symptoms related 158 to laboratory-confirmed SARS-CoV-2 infection. The COVID-19 hospitalization group was 159 defined as individuals requiring hospitalization due to symptoms associated with population-based cohorts. Controls included individuals whose status of exposure to . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted January 21, 2022. 

The 

. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted January 21, 2022. ; https://doi.org/10.1101/2022.01.20.22269593 doi: medRxiv preprint

The causal associations were evaluated using odds ratios (ORs), which are 195 expressed according to a standard deviation (SD) increase in genetically predicted BMI 196 (kg/m 2 ), body fat percentage (%), body fat mass (kg), or body fat-free mass (kg).

Results with a P < 0.0125 were considered statistically significant (P = 0.05/4;

Bonferroni-corrected significance threshold according to the number of exposures). We note 199 that such a correction is likely overly conservative, given that the exposures are 

. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted January 21, 2022. Table 1 . The mean ± SD BMI was 27.4 ± 4.8 kg/m 2 , body fat 238 percentage was 31.4 ± 8.5%, body fat mass was 24.9 ± 9.6 kg, and body fat-free mass was 239 53.2 ± 11.5 kg ( 

To the best of our knowledge, this is the first MR study to evaluate the causal association 285 between directly measured body compositions (i.e., body fat percentage, body fat mass, and 286 body fat-free mass) with COVID-19 severity outcomes. In this study, we found that an increase 287 in BMI, body fat percentage, and body fat mass were associated with an increased risk of 288 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted January 21, 2022. ; https://doi.org/10.1101/2022.01.20.22269593 doi: medRxiv preprint severe COVID-19 and COVID-19 hospitalization. We further evaluated the independent 289 causal effects of body fat mass and body fat-free mass on these outcomes and revealed that 290 only body fat mass was independently associated with the outcomes.

During the COVID-19 pandemic, obesity has emerged as a major risk factor for 

. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted January 21, 2022. ; https://doi.org/10.1101/2022.01.20.22269593 doi: medRxiv preprint We used multivariable MR since most instrumental variables of adiposity effect both 307 fat mass and fat-free mass, although some variants more strongly and proportionally influence 308 fat mass, whereas others influence fat-free mass more strongly. Therefore, multivariable MR 309 can test the differential causal effects of fat mass and fat-free mass. Using this approach, 310 recent MR studies showed differential associations between body fat mass and body fat-free 311 mass with various disorders (10-13). The present findings extend this knowledge to COVID-19.

Results from multivariable MR showed that body fat mass but not body fat-free mass was 313 independently associated with severe COVID-19 and COVID-19 hospitalization. The 314 association between body fat mass and COVID-19 severity was strengthened in multivariable 315 MR relative to findings using univariable MR, whereas the effects of body fat-free mass on 316 COVID-19 severity was markedly attenuated in multivariable MR, thereby illustrating the 317 independent associations between body fat mass and COVID-19 severity.

The underlying mechanism of these associations remains to be clarified. Obesity is a 319 metabolic disease characterized by systemic changes in metabolism, including insulin 320 resistance, glucose intolerance, dyslipidemia, changes in adipokines (e.g., increased leptin 321 and decreased adiponectin levels), chronic inflammation, and altered immune response, all of 322 which could collectively increase the risk of COVID-19 severity (4, 24, 25). Moreover, obesity 323 causes respiratory dysfunction, including impaired respiratory physiology, increased airway 324 resistance, impaired gas exchange, low lung volume, and low muscle strength, which can also 325 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted January 21, 2022. 

. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. 

. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted January 21, 2022. 

. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted January 21, 2022. ; https://doi.org/10.1101/2022.01.20.22269593 doi: medRxiv preprint . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted January 21, 2022. 

. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted January 21, 2022. 

. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted January 21, 2022. 

. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted January 21, 2022. 

. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted January 21, 2022. ; https://doi.org/10.1101/2022.01.20.22269593 doi: medRxiv preprint 

. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted January 21, 2022. ; https://doi.org/10.1101/2022.01.20.22269593 doi: medRxiv preprint

World Health Organization. COVID-19 clinical management: living guidance

Associations between body-mass index and COVID-19 severity in 6·9 million people in 398

England: a prospective, community-based, cohort study

Body mass index and risk of COVID-19

diagnosis, hospitalisation, and death: a cohort study of 2 524 926 Catalans

without obesity in the United States, Spain, and the United Kingdom

Bias and causal associations in observational research

Reporting of Observational Studies in Epidemiology Using Mendelian Randomization

Strengthening the reporting of observational studies 422 in epidemiology using mendelian randomisation (STROBE-MR): explanation and elaboration

BMI-related errors in the measurement of obesity

Body 430 composition and atrial fibrillation: a Mendelian randomization study

Body mass index and body 433 composition in relation to 14 cardiovascular conditions in UK Biobank: a Mendelian 434 randomization study

Genetically predicted body composition in 436 relation to cardiometabolic traits: a Mendelian randomization study

Investigating the 439 association between body fat and depression via Mendelian randomization

The MR-Base platform 16. COVID-19 Host Genetics Initiative. Mapping the human genetic architecture of 448 COVID-19

Selecting likely causal risk factors from 450 high-throughput experiments using multivariable Mendelian randomization

Supplementary material for "Causal association between body fat accumulation

COVID-19 severity: A Mendelian randomization study

Mendelian randomization with invalid 457 instruments: effect estimation and bias detection through Egger regression

Detection of widespread horizontal 460 pleiotropy in causal relationships inferred from Mendelian randomization between complex 461 traits and diseases

Lyons and hospital admission in adults after covid-19 vaccination: national prospective cohort study

COVID-19 and metabolic disease: mechanisms and clinical management

Causal inference for genetic obesity, 475 cardiometabolic profile and COVID-19 susceptibility: a Mendelian randomization study

Review: obesity and COVID-19: a detrimental 478 intersection. Front Endocrinol (Lausanne)

SNPs associated with a higher or lower exposure

Genome-wide association studies Odds Ratio