Purpose: The objectives of this study were to describe the genome wide DNA methylation changes in hip and knee osteoarthritis (OA) and to identify novel genes and pathways involved in OA by comparing the DNA methylome of the hip and knee osteoarthritic cartilage with those of healthy individuals.

Methods: OA cartilage samples were collected from the tibial plateau and femoral head articular surfaces of patients with severe OA undergoing total hip/knee replacement surgery. Healthy cartilage samples were obtained from individuals with femoral neck fractures, but no OA. Chondrocyte DNA methylation was assayed by Illumina Infinium HumanMethylation450 BeadChip array, allowing for the analysis of >480,000 CpG sites. A sub-analysis was also done to compare hip and knee OA. An independent sample T-test was conducted for each CpG site and those sites with at least 10% methylation difference and a P-value of <0.0001 were selected as differentially methylated regions (DMR). DAVID v6.7 was used for the functional annotation clustering of the DMR genes.

Results: The study population comprises 5 patients with hip OA, 6 patients with knee OA and 7 hip samples from healthy individuals. The comparisons of hip, knee and combined hip/knee OA patients with controls resulted in 26, 72, and 103 DMRs, respectively. The comparison between hip and knee OA revealed 67 DMRs. The overall number of the sites after considering the overlaps was 239, among which 151 sites were annotated to 145 genes. 28 of these genes were reported in previous methylation studies of OA. The functional annotation clustering of the identified genes revealed clusters significantly enriched in skeletal system morphogenesis (bonferroni corrected p=8.4e^-9) and development (bonferroni corrected p=9.0e^-6), embryonic organ (bonferroni corrected p=1.4e^-5) and skeletal system morphogenesis (bonferroni corrected p=9.9e^-7), as well as Homeobox genes (bonferroni corrected p=3.1e^-4). A sub-analysis after the removal of the genes differentially methylated between hip and knee OA highlighted the same pathways excluding Homeobox genes.

Conclusions: We demonstrated a number of CpG sites and genes across the genome differentially methylated among OA patients and controls, a remarkable portion of which seem to be involved in potential etiologic mechanisms of OA. Genes involved in skeletal developmental pathways and embryonic organ morphogenesis may be a potential area of concentration for future osteoarthritis genetic/epigenetic research.