Deep learning and genome-wide association meta-analyses of bone marrow adiposity in the UK Biobank

Wei Xu, Ines Mesa-Eguiagaray, David M. Morris, Chengjia Wang, Calum D. Gray, Samuel Sjöström, Giorgos Papanastasiou, Sammy Badr, Julien Paccou, Xue Li, Paul R. H. J. Timmers, Maria Timofeeva, Susan M. Farrington, Malcolm G. Dunlop, Scott I. K. Semple, Tom J. MacGillivray, Evropi Theodoratou*, William P. Cawthorn*

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

7 Downloads (Pure)

Abstract

Bone marrow adipose tissue is a distinct adipose subtype comprising more than 10% of fat mass in healthy humans. However, the functions and pathophysiological correlates of this tissue are unclear, and its genetic determinants remain unknown. Here, we use deep learning to measure bone marrow adiposity in the femoral head, total hip, femoral diaphysis, and spine from MRI scans of approximately 47,000 UK Biobank participants, including over 41,000 white and over 6300 non-white participants. We then establish the heritability and genome-wide significant associations for bone marrow adiposity at each site. Our meta-GWAS in the white population finds 67, 147, 134, and 174 independent significant single nucleotide polymorphisms, which map to 54, 90, 43, and 100 genes for the femoral head, total hip, femoral diaphysis, and spine, respectively. Transcriptome-wide association studies, colocalization analyses, and sex-stratified meta-GWASes in the white participants further resolve functional and sex-specific genes associated with bone marrow adiposity at each site. Finally, we perform a multi-ancestry meta-GWAS to identify genes associated with bone marrow adiposity across the different bone regions and across ancestry groups. Our findings provide insights into BMAT formation and function and provide a basis to study the impact of BMAT on human health and disease.
Original languageEnglish
Article number99
Pages (from-to)99
JournalNature Communications
Volume16
Issue number1
DOIs
Publication statusPublished - 2 Jan 2025

Keywords

  • Bone
  • Genome-wide association studies
  • machine learning

ASJC Scopus subject areas

  • General Chemistry
  • General Biochemistry,Genetics and Molecular Biology
  • General Physics and Astronomy

Fingerprint

Dive into the research topics of 'Deep learning and genome-wide association meta-analyses of bone marrow adiposity in the UK Biobank'. Together they form a unique fingerprint.

Cite this