Abstract
Introduction
Hip Osteoarthritis (HOA) is a highly heterogeneous disease with potentially different genetic aetiologies. Thus far, few genetic variants have been robustly associated with broad definitions of HOA. We aimed to identify novel HOA susceptibility variants by examining a set of more homogeneous, radiographically-derived endophenotypes relating to anatomic pattern of joint involvement and bone remodelling response in HOA.
Materials, Methods and Results
Individuals with HOA and AP-hip radiographs (n=2,109) were selected from the arcOGEN study, genotyped on 2 platforms. The most arthritic hip per patient was categorised using Bombelli's classification (atrophic/normotrophic/hypertrophic), and for pattern of joint space narrowing (superior/medial-axial/concentric). Following 1000-Genomes-Project-based imputation and stringent quality control over 7 million variants were tested for association with each phenotype under the additive model. Fixed-effects meta-analysis was used to combine the results from the two GWA studies.
In the discovery GWAS of atrophic HOA vs cases with non-atrophic HOA a strong signal (rs16869403, OR[95% CI]=2.47[1.81–3.38], p=1.53×10−8) was detected in the G protein-coupled receptor, GPR98. Polymorphisms in GPR98 have been associated with osteoporotic fracture and low bone mineral density and GPR98 knockout mice have a low bone mass phenotype. The meta-analysis of medial joint space narrowing showed strong association with variants in LRCH1 (rs754106, OR[95% CI]=1.46[1.26–1.68], P=2.85×10−7) previously associated with OA but with conflicting replication evidence, and BMP1 (chr8:22065846, OR[95% CI]=3.36[2.12–5.33], P=2.6×10−7) which induces bone and cartilage development. None of these variants showed significant evidence for association when broadly classified HOA cases were compared to population-based controls from the arcOGEN GWAS.
Conclusion
Through comprehensive examination of radiographically-derived, HOA-related phenotypes we have identified several promising signals that point to novel biologically plausible genes for HOA. Our results indicate that, in a heterogeneous disease like OA the study of narrower phenotype definitions closer to the biology of the disease has the potential to enhance power to detect OA-relevant associations.
