Objective. The identified osteoarthritis (OA) susceptibility genes are mainly active in skeletal development and could thus affect joint geometry. Because nonoptimal joint geometry is a risk factor for the development of OA, we investigated if and how the path that leads from nonoptimal joint geometry to OA of the hip is influenced by these genes.Methods. The shape of the hips of subjects in the Genetics, Osteoarthritis and Progression Study, consisting of sibling pairs with symptomatic OA at multiple joint locations, was quantified by applying a statistical shape model to radiographs. Shape aspects (modes) were correlated to OA characteristics. We then tested for the association of shape modes with OA susceptibility single-nucleotide polymorphisms (SNPs) of GDF5, FRZB, and DIO2.Results. Four of 23 shape modes (mode 1, mode 17, mode 18, and mode 21) were strongly associated with OA characteristics. We observed a significant interaction between carrier status of DIO2 rs12885300 and hip OA characteristics for mode 1 (P ؍ 0.005). This indicates that this specific aspect of hip shape correlates with OA characteristics only in carriers of the susceptibility allele.Conclusion. Our results suggest that it is more likely that the rs12885300 SNP of DIO2 increases the vulnerability of cartilage to nonoptimal bone shapes rather than directly influencing the formation of these shapes.The genetic predisposition for osteoarthritis (OA) (1) may reflect the result of multiple interacting genes with small effects (2). The identification of such genes should be considered essential to obtain a better understanding of OA and the underlying biologic events preceding its onset (3). One path leading to OA in which genes might play a role starts with the causal effect of hip morphology on the development of OA and may explain part of the heritability of hip OA. The life-long biomechanical stress caused by a nonoptimal shape of the bones in the joint leads to recurrent damage of cartilage, which eventually triggers the onset of OA (4).What is interesting in this respect is the fact that genes that are mainly active in skeletal development, such as FRZB and GDF5, are identified as OA susceptibility genes (5,6). These genes are involved in the orchestration of growth plate chondrocytes that results in formation of cartilage and eventually bone during endochondral ossification (7). A recent large-scale metaanalysis (6) did not provide evidence for the relatively rare FRZB OA risk alleles, indicating that these variants do not confer susceptibility to common OA. It is, however, generally known that meta-analyses have little power to detect relatively rare variants with a modest effect that confer risk in a specific subset of OA cases. Furthermore, functional studies in Frzb-knockout mice showed increased cortical bone thickness and density, resulting in stiffer bones upon mechanical loading, which may increase OA susceptibility and stresses developmental aspects (8).