An atlas of genetic determinants of forearm fracture

Abstract

Osteoporotic fracture is among the most common and costly of diseases. While reasonably heritable, its genetic determinants have remained elusive. Forearm fractures are the most common clinically recognized osteoporotic fractures with a relatively high heritability. To establish an atlas of the genetic determinants of forearm fractures, we performed genome-wide association analyses including 100,026 forearm fracture cases. We identified 43 loci, including 26 new fracture loci. Although most fracture loci associated with bone mineral density, we also identified loci that primarily regulate bone quality parameters. Functional studies of one such locus, at TAC4, revealed that Tac4^-/- mice have reduced mechanical bone strength. The strongest forearm fracture signal, at WNT16, displayed remarkable bone-site-specificity with no association with hip fractures. Tall stature and low body mass index were identified as new causal risk factors for fractures. The insights from this atlas may improve fracture prediction and enable therapeutic development to prevent fractures.

Fig. 1. Associations with forearm fracture (50,471 cases and 969,623 controls), hip fracture and any fracture risk for the identified top forearm fracture signal at the SOST, ESR1, WNT16 and SALL1 loci.

Data are presented as OR for fracture per effect allele, with 95% CIs (for SOST, rs80107551-C; ESR1, rs2941741-G; WNT16, rs2908007-A; SALL1, rs62028332-G). OR for hip fractures are from Nethander et al. (11,516 cases and 723,838 controls), while OR for any fractures are from Morris et al. (53,184 cases and 373,611 controls). Statistically significant different associations for forearm fracture compared with the corresponding associations with hip fracture or any fracture are indicated with P values. Two-sided z test was used to test differences and the statistical significance limit was set to 0.0005 (Bonferroni adjustment considering 50 SNPs and two traits).

Fig. 2. Bone microstructure and bone strength but not aBMD is affected in Tac4^–/– mice.

a, Two-dimensional aBMD as measured by DXA in tibia from Tac4^–/– (female, n = 11; male, n = 11) mice compared with control (female, n = 9; male, n = 11) mice. b–e, µCT measurements of vertebra L5. Representative three-dimensional images of trabecular bone in transversal plane of vertebra L5 from control and Tac4^–/– male mice (b); the distance between tics in the scale grid in x and y axes in front and back of the three-dimensional image is 200 µm. Trabecular bone volume over total volume (BV/TV) (c), trabecular thickness (Tb. Th) (d) and trabecular number (Tb. N) (e) in vertebra L5 from Tac4^–/– (female, n = 11 male; n = 11) mice compared with control (female, n = 9; male, n = 11) mice. f–i, Cortical bone area (Ct. Ar) (f), periosteal circumference (Peri C) (g), endosteal circumference (Endo C) (h) and cortical moment of inertia (i) in tibia from Tac4^–/– (female, n = 11; male, n = 11) mice compared with control (female, n = 9; male, n = 11) mice. j–l, Maximal load at failure (Fmax) of vertebra L5 (j) as measured by compression test in Tac4^–/– (female, n = 11; male, n = 11) mice compared with control (female, n = 9; male, n = 10) mice. Maximal load at failure (k) and stiffness (l) of tibia as measured by three-point bending in Tac4^–/– (female, n = 11; male, n = 11) mice compared with control (female, n = 9; male, n = 9) mice. All results refer to 12-month-old mice. Individual values are presented with the mean as horizontal lines and ± s.e.m. as vertical lines. A two-way analysis of variance was used to assess the effects of genotype (Tac4^–/– or control (Tac4^+/+)), sex (female or male), as well as their interaction. A difference was considered statistically significant when P < 0.05.

Fig. 3. Multiplex FISH on mouse tibiae.

Representative maximum intensity projection (×40 magnification) images of mouse tibiae labeled with fluorescence RNAscope for Tac4 (red) plus Runx2, Sost or Ctsk (green) mRNA and DAPI (blue). Tac4 and Runx2 or Sost double-positive osteocytes (solid arrows) can be detected both in the trabecular bone (TB) and cortical bone (CB). Tac4 mRNA also can be found in Runx2-expressing osteoblasts (open arrows) and Ctsk-expressing osteoclasts (solid arrowheads). Representative maximum intensity projection images are shown of mouse tibiae; n = 3 biologically independent mice.

Fig. 4. MR to estimate the causal associations for 17 genetically determined risk factors on forearm fracture risk.

Data are given as OR with 95% CIs estimated using IVW MR. OR for the risk of fracture are given per s.d. change in the risk factor for continuous trait or risk of fracture per doubling of odds (obtained by multiplying the causal estimate of log odds by ln₂ ≈ 0.693) (ref. ) of disease susceptibility for binary factors. For menopause and puberty, we used the estimated s.d. from the largest cohorts in the published GWAS (early menopause s.d. = 3.81 in Breast Cancer Association Consortium; late puberty s.d. = 1.40 in Women’s Genome Health Study) to translate the effect unit from year to s.d. For ever smoked regularly, ORs are expressed per unit increase in log odds of ever smoking regularly with a 1 s.d. increase in genetically predicted smoking initiation corresponding to a 10% increased risk of smoking^,. Grip strength is given as grip strength per BW and s.d. for grip strength is given for kg grip strength per kg in BW and was estimated in the UK Biobank to be 0.127. P < 0.0029 (Bonferroni correction accounting for 17 tests) was considered statistically significant. Findings with a nonadjusted P value (based on a two-sided z test) below 0.0029 have their corresponding P value represented in the figure. For risk factors including UK Biobank in the GWAS and displaying statistically significant causal effects with forearm fractures, we performed sensitivity analyses excluding UK Biobank in the forearm fracture meta-analysis used for the MR, revealing essentially similar effect estimates (results excluding UK Biobank in the forearm fracture GWAS; decreased eBMD OR = 2.07; 95% CI, 1.96–2.20; P = 1.0 × 10⁻¹³⁷; early menopause OR = 1.10; 95% CI, 1.02–1.18; P = 1.5 × 10⁻²).