We thank Dr. Dieppe for his complimentary remarks (1) about our randomized controlled trial (RCT) of doxycycline in knee osteoarthritis (OA) (2) and concur that a drug that slows progression of radiographic changes of OA will hold little therapeutic interest if this effect is not accompanied by clinically important benefits, such as a reduction in joint pain or an improvement in function. After all, the real problem is not OA; almost all of us will develop OA. The real problem is painful OA, with its medical, social, and economic consequences.Although previous data suggest only a weak association between progression of radiographic changes and progression of knee pain and disability in patients with OA, studies that have examined this relationship have used radiographic protocols that are less sensitive to joint space narrowing (JSN) than that used in the doxycycline trial. Indeed, in the RCT, the rate of JSN at 30 months in the index knee of subjects with a baseline Western Ontario and McMaster Universities Osteoarthritis Index (3) pain score Ն16 was more than twice as rapid as that in subjects with a lower baseline pain score (mean Ϯ SD 0.73 Ϯ 0.75 mm versus 0.33 Ϯ 0.64 mm; P Ͻ 0.001). Because of the low mean level of knee pain in the subjects enrolled in the RCT, the results of this clinical trial do not answer the question of whether doxycycline treatment has a salutary clinical effect in patients with more severe OA pain. That can be tested, however, and it should be tested.We share with Dr. Dieppe the view that OA is a mechanically induced disease that is biochemically mediated (e.g., matrix metalloproteinases play a prominent role in the degradation of articular cartilage). It is indeed reasonable to anticipate that the response to a structure-modifying drug may be suboptimal if the underlying mechanical abnormalities are not corrected. Osteotomy and joint distraction procedures, however, are not optimal therapies for most patients with knee OA. Furthermore, in most of the animal models in which the efficacy of doxycycline, diacerhein, and other putative structure-modifying drugs for OA has been demonstrated, OA has been surgically induced. Prevention or slowing of cartilage damage has occurred despite persistence of the mechanical derangement. If this point is dismissed, incentives to attempt to develop structure-modifying OA drugs may be stifled. Perhaps such drugs will, as Dieppe suggests, be of little value in treatment of OA in humans, but it is too soon to reach that conclusion on the basis of the evidence to date.Of course, pharmacologic therapy for OA must be safe. This is as true for analgesics, nonselective nonsteroidal antiinflammatory drugs, and coxibs as it is for diseasemodifying drugs. In the RCT, no serious adverse events attributable to doxycycline occurred among the 218 subjects randomized to the active treatment group. That number is far too small to provide reassurance that such events would not occur if the drug were used widely, without adequate medical supervision, by older subjects...