Abstract
Background
Hip shape is an important risk factor for hip osteoarthritis (HOA). A previous genome-wide association study (GWAS) identified eight independent single nucleotide polymorphisms (SNPs) associated with hip shape, three of which were associated with hip osteoarthritis (HOA). In this study, we aimed to further characterise the genetic architecture of hip shape, based on a GWAS of hip shape derived from hip dualenergy X-ray absorptiometry (DXA) scans obtained in UK Biobank study (UKB).
Methods
Hip shape was quantified in left hip DXA images from UKB. GWAS was performed on ten orthogonal hip shape modes (HSMs) (which together explained 86.3% of the variability in hip shape) in 38,175 individuals, derived by statistical shape modelling (BoneFinder, University of Manchester). Analyses were adjusted for age, sex, genotyping chip, and 20 ancestry principal components. GCTA-COJO was implemented to identify conditionally independent SNPs at each genome-wide significant HSMassociated loci. LD score regression was used to estimate genetic correlations between HSMs, and between HSMs and other traits based on GWAS summary statistics for alpha angle (AA) (UKB), cam morphology (UKB), HOA (GO consortium), narrowest neck width (NNW) (UKB) and femoral head diameter (fHD) (UKB).
Results
We identified a total of 233 conditionally independent signals associated with the first ten HSMs at genome-wide significance (p<5x10-8) which mapped to 201 loci (based on a 1mb sliding window). SNP heritability (h2) ranged between 0.21-0.36 for all modes (except HSM5 h2 = 0.07). Two HSMs showed evidence of genetic correlation with AA at P<0.05, five with cam morphology, four with HOA, six with other HSMs, six with NNW, and eight with fHD (Figure 1). All loci identified in our previous GWAS were replicated, with those mapping to endochondral bone formation genes (e.g., SOX9, PTHLH, RUNX1, FGFR4, and HHIP) associated with three or more HSMs. Of the 193 novel loci identified, 12 were related to two or more HSMs, 10 were related to HOA, 13 to fHD, and 21 to NNW.
Conclusions
We report the largest hip shape GWAS to date, identifying 233 conditionally independent signals mapping to 201 loci. All previously reported hip shape SNPs were replicated in our study. Loci mapping to HSMs were also related to conventional hip geometry measures, and to HOA, in keeping with the evidence of a genetic correlation between these traits. Further investigations are in progress to determine the genetic mechanisms linking hip shape, hip geometry and HOA.
Hip shape is an important risk factor for hip osteoarthritis (HOA). A previous genome-wide association study (GWAS) identified eight independent single nucleotide polymorphisms (SNPs) associated with hip shape, three of which were associated with hip osteoarthritis (HOA). In this study, we aimed to further characterise the genetic architecture of hip shape, based on a GWAS of hip shape derived from hip dualenergy X-ray absorptiometry (DXA) scans obtained in UK Biobank study (UKB).
Methods
Hip shape was quantified in left hip DXA images from UKB. GWAS was performed on ten orthogonal hip shape modes (HSMs) (which together explained 86.3% of the variability in hip shape) in 38,175 individuals, derived by statistical shape modelling (BoneFinder, University of Manchester). Analyses were adjusted for age, sex, genotyping chip, and 20 ancestry principal components. GCTA-COJO was implemented to identify conditionally independent SNPs at each genome-wide significant HSMassociated loci. LD score regression was used to estimate genetic correlations between HSMs, and between HSMs and other traits based on GWAS summary statistics for alpha angle (AA) (UKB), cam morphology (UKB), HOA (GO consortium), narrowest neck width (NNW) (UKB) and femoral head diameter (fHD) (UKB).
Results
We identified a total of 233 conditionally independent signals associated with the first ten HSMs at genome-wide significance (p<5x10-8) which mapped to 201 loci (based on a 1mb sliding window). SNP heritability (h2) ranged between 0.21-0.36 for all modes (except HSM5 h2 = 0.07). Two HSMs showed evidence of genetic correlation with AA at P<0.05, five with cam morphology, four with HOA, six with other HSMs, six with NNW, and eight with fHD (Figure 1). All loci identified in our previous GWAS were replicated, with those mapping to endochondral bone formation genes (e.g., SOX9, PTHLH, RUNX1, FGFR4, and HHIP) associated with three or more HSMs. Of the 193 novel loci identified, 12 were related to two or more HSMs, 10 were related to HOA, 13 to fHD, and 21 to NNW.
Conclusions
We report the largest hip shape GWAS to date, identifying 233 conditionally independent signals mapping to 201 loci. All previously reported hip shape SNPs were replicated in our study. Loci mapping to HSMs were also related to conventional hip geometry measures, and to HOA, in keeping with the evidence of a genetic correlation between these traits. Further investigations are in progress to determine the genetic mechanisms linking hip shape, hip geometry and HOA.
| Original language | English |
|---|---|
| Article number | e10684 |
| Pages (from-to) | OC1.3 |
| Number of pages | 1 |
| Journal | JBMR Plus |
| Volume | 6 |
| Issue number | S2 |
| Early online date | 25 Oct 2022 |
| DOIs | |
| Publication status | Published - 25 Oct 2022 |
| Event | Bone Research Society Annual Meeting 2022: Bone Research Society - Manchester Duration: 6 Jul 2022 → 8 Jul 2022 https://boneresearchsociety.org/meeting/manchester2022/ |