We read with great interest the article by Haynes et al, published in a recent issue of Arthritis Care & Research (1). The authors investigated the effect of disease-modifying antirheumatic drug (DMARD) treatment on the expression of RANKL and osteoprotegerin (OPG) in synovial tissue, although we think the clinical design of the study was not ideal. We appreciate the authors' consideration of our previously published data regarding modulation of the RANKL/OPG system in vivo and in vitro following tumor necrosis factor (TNF) blockers (2) and intraarticular glucocorticoid treatment (3). We would like to take this opportunity to clarify some issues raised by the authors with the hope that it will help to interpret the current study.
Haynes et al suggested that we have demonstrated in a previous study that OPG expression was low in rheumatoid arthritis (RA) synovial tissue, whereas RANKL expression was high in patients with active disease. However, we clearly and specifically made a case in the cited study against any correlation between synovial RANKL or OPG expression and disease activity, as evaluated by the Disease Activity Score in 28 joints (DAS28). Moreover, no correlation between RANKL changes and DAS28 scores were observed in the same study. As Haynes et al mentioned, we have indeed suggested that synovial OPG modulation is a direct effect of anti-TNF therapy, and demonstrated this in our in vitro experiments.
Haynes et al suggested that our previous observation concerning the absence of OPG modulation by glucocorticoids might be due to methodology issues such as antibody selection. However, the authors clearly demonstrated that this is not the case, because they obtained similar results with 2 different antibodies (one of which was also used in our study). Another comment by Haynes et al is that the different outcomes of previous studies might reside in inclusion of patients with long disease duration, as compared with short disease duration in their cohort of patients. However, only 16 of 25 total patients investigated by Haynes et al had a disease duration <1 year, with a mean disease duration of 4.3 years. Moreover, the authors reported no influence of the disease duration on changes in OPG and RANKL expression following DMARD treatment.
Haynes et al discussed the issue of patient selection and the choice of patients subjected to various co-medications in our previous studies. We agree that patient as well as medication selection are key to the ability to interpret these studies, and that heterogeneities in patient population as well as in medication use must be acknowledged and controlled for. In our case, we did this by means of assuring that all patients were receiving stable therapy at least 4 weeks before study entry, and thereafter studying effects of 2 distinct therapies (TNF blockade and intraarticular glucocorticoid injections) (2, 3). The study by Haynes et al instead investigated effects of several different DMARDs on the OPG/RANKL expression. This heterogeneity in treatments, however, precludes conclusions concerning effects of a distinct therapeutic agent, and therefore does not permit any direct comparison between any of the various DMARDs used in the study by Haynes et al and the effects of TNF blockade and corticosteroids in our previous investigations.
In conclusion, we believe that further analysis of molecular effects of distinct medications in synovial tissue of treated patients remains a most valuable method for understanding effects of different medications in distinct groups of patients. The study by Haynes et al emphasizes the potential value of our previous approach for understanding the molecular mode of action of distinct frequently used DMARDs in addition to already studied effects of TNF blockade and glucocorticoids.