Genetic variants near insulin-like growth factor binding protein 3 (IGFBP3) are associated with hip osteoarthritis

Abstract

Background/Purpose: Hip osteoarthritis (HOA) is one of the most common joint disorders and can result in pain and disability.HOA is heritable, but the particular genes contributing to the development of HOA are not well defined.To identify genetic associations with HOA, we conducted a two-stage genome-wide association study (GWAS). Methods: All analyses were restricted to individuals of European ancestry. The discovery phase was performed using radiographically determined HOA cases and controls selected from the Osteoporotic Fractures in Men (MrOS) Study and the Study of Osteoporotic Fractures (SOF) (combined cases = 662).HOA cases were defined as Croft grade > 2 or total hip replacement (THR).HOA controls were defined as Croft grade < 1, maximum joint space narrowing (JSN) < 1, maximum osteophytes < 1, and no THR.SNPs were genotyped using the Illumina Omni1-Quad array and approximately 2.5 million SNPs were imputed using the HapMap reference panel.Logistic regression was performed, and MrOS and SOF results were combined using inverse variance weighted fixed effect meta-analysis.SNP associations with P-values < 5 x 10-8were examined for replication in the Rotterdam cohorts (RSI = original cohort, RSII = second recruitment cycle), Twins UK, and Chingford cohorts.Publicly available eQTL data from HapMap CEU lymphoblastoid cell lines were used. Results: On average, MrOS participants (100% male) were older than SOF participants (100% female) at the clinic visits when HOA was assessed (MrOS: mean+/-SD = 77.5+/-5.4 years, range = 69-97; SOF: mean+/-SD = 70.9+/-5.0 years, range=65-91).In the discovery meta-analysis, the rs788748 A allele and the rs879966 G allele were associated with decreased odds for HOA (Table 1).The two directly genotyped SNPs were 23 kb apart and were in moderate LD (HapMap CEU r2= 0.54).Neither SNP remained nominally significant in conditional analysis, indicating dependence.The SNP rs788748 replicated in the RSI cohort, but not the RSII cohort or the Chingford cohort.In the Twins UK cohort, the association between the rs788748 A allele and HOA was significant and in the opposite direction relative to the discovery samples (Table 1).The SNP rs879966 did not replicate (Table 1). (Table Presented) From eQTL data, rs788748 and rs879966 were marginally associated with IGFBP3 expression (P-value = 0.07 and 0.06, respectively), but not IGFBP1 expression (P-value = 0.74 and 0.34, respectively).SNP associations with circulating IGFBP3 levels and results from IGFBP3 knockdown experiments in the ATDC5 chondrogenesis model system will be presented. Conclusion: Our genetic association and eQTL results provide suggestive evidence for a link between IGFBP3 and HOA, but further replication is required for the association results to be considered robust.IGFBP3 is known to be expressed in human chondrocytes and might be an attractive candidate for follow-up studies.