Cindy Boer

Hand osteoarthritis and Matrix-Gla Protein | 127 3.1 Styrkarsdottir et al [ 19] reported on common genetic variants that associate with severe hand OA, among the replication cohorts were the Leiden and Rotterdam cohorts. [19] Although we observe suggestive signals at the reported locus ( ALDH1A2 gene, 1p31), the respective variants did not meet the genome-wide significance threshold in our analyses ( Supplementary Table 5 ). This difference is likely caused by the marked- ly different phenotypes that were used for either analyses. Where Styrkarsdottir et al studied a dichotomous severe hand OA phenotype, our phenotype was semi-quantita- tively phenotype. To conclude, we here present coding variants in MGP that are associated with radiographic hand OA, and the hand OA risk allele marks lower expression of MGP in ar- ticular cartilage. Our findings suggest that MGP might play an important role in hand OA pathogenesis through pathways related to articular cartilage calcification and vitamin K. Better understanding of MGP gene and protein regulation and its relation to vitamin K intake and OA may reveal novel therapeutic drug targets for hand OA . Materials & Methods Discovery GWAS, replication and meta‐analysis For a detailed description on the GWAS methods, participating studies, quality control procedures for genotyping and imputation, see online Supplementary Text 1 and Sup- plementary Table 1 . Detailed phenotype description of Kellgren and Lawrence sum score We have used a semi-quantitative bilateral measure of OA of the hand based on the ra- diographic Kellgren and Lawrence (KL) score.[20] Using radiographs of both hands, the KL score was determined for each joint in the hand. Using these KL scores we defined the KL sum score: the total KL score, the sum, of the following hand joints for both hands (left and right): all distal interphalangeal joints, all proximial interphalangeal joints, all metacarpophalangeal (MCP) joints, the interphalangeal joint and the first carpometa- carpal joint, which gives the sum of 15 joints on each hand, and in total 30 joints for both hands together, resulting in a minimum score of 0 and a maximum score of 120. In the Leiden Studies (LS) cohort, no KL scoring was done of the MCP joints, resulting in a KL sum score of maximum 88. Individuals lacking KL grading for both hands or one hand and individuals with missing age or gender information were excluded from all analyses in all cohorts. As the KL sum score has a skewed distribution, the top finding of the meta-analysis was repeated in the discovery cohorts using a Poisson regression.