Possible role of shared epitope status in the relationship between matrix metalloproteinase 3 genotype and radiographic progression of rheumatoid arthritis: comment on the article by Constantin et al (letter)