Do Inhibitors of Cyclooxygenase-2 Impair Bone Healing?

NON-STEROIDAL ANTI-INFLAMMATORY drugs (NSAIDs) are among the most commonly prescribed drugs worldwide. Indicated in the treatment of several forms of arthritis, menstrual pain, and headache, they act to decrease inflammation by inhibiting cyclooxygenases and reducing prostaglandin synthesis. Yet, whereas their performance as anti-inflammatory drugs has been favorable, their use has been limited by gastrointestinal side effects such as dyspepsia, abdominal pain, and in some instances, gastric or duodenal perforation or bleeding. The development of cyclooxygenase-2-specific inhibitors (coxibs) is a response to the need for drugs that inhibit prostaglandin production without side effects.1 As a class, prostaglandins are known to be involved in numerous body functions, but their role as mediators of inflammation has led to the development of both nonspecific (NSAIDs), and now specific (coxibs), inhibitors. Both groups of drugs inhibit prostaglandin G/H synthase, the enzyme that catalyzes the conversion of arachidonic acid to prostaglandins and thromboxanes. The two recognized forms of this enzyme, cyclooxgenase-1 (cox-1) and cyclooxygenase-2 (Cox-2), are encoded by two separate genes.2, 3 cox-1 is constitutively expressed by many tissues and functions as a so-called “housekeeping” enzyme, which maintains homeostatic levels of prostaglandins for normal function of several organs, in particular, the stomach. On the other hand, cox-2 is induced by an array of stimuli, including proinflammatory cytokines, growth factors, mechanical stress, and notably, injury and inflammation.4-6 Because most NSAIDs inhibit cox-1 and cox-2 with near equal potency, it was hoped that the development of cox-2 selective drugs would be better tolerated and equally efficacious in managing inflammation. However, whereas the selectivity of this group of compounds may allow inflammation to be inhibited with minimal effects on certain homeostatic mechanisms, their role in bone metabolism remains unclear. Prostaglandins are synthesized by osteoblasts and are known to stimulate both bone formation and resorption.7 For over one-quarter of a century, reports have appeared showing that NSAIDs inhibit experimental fracture healing and heterotopic ossification in patients.8-14 While only a few studies have addressed the important question of NSAID inhibition of human bone repair, those that have, have been confounded by the effects of other competing factors14 or have collected data in a retrospective fashion.13 However, the notion that NSAIDs have serious inhibitory effects on bone healing in patients seems hard to accept because millions of people have used these drugs; many have used them to ease pain during fracture healing or after dental, craniofacial, or orthopedic surgery, and an association between nonunion or impaired bone healing has not been recognized as a clinical problem. On the other hand, if these drugs impair the healing process so that it is delayed by only a few weeks and not completely inhibited, then perhaps these effects have been overlooked and should be taken into consideration in the management of patients. In this issue of JBMR, Simon et al.15 report that cox-2 function is essential for fracture healing. These investigators treated rats with the nonselective NSAID, indomethacin, and the two most widely prescribed coxibs, celecoxib and rofecoxib. They showed that all three drugs inhibited fracture healing, but the effects were more profound when the coxibs were used. They also demonstrated impaired fracture healing in mice homozygous for a null mutation in the cox-2 gene. Several aspects of this report, however, require attention. First, whereas the doses of indomethacin and celecoxib used in the rats were roughly equivalent to those used in patients, the dose of rofecoxib was nearly eight times that used to manage inflammation and four times that used to manage acute pain. Moreover, whereas the use of these drugs in the management of acute pain is typically short-term (a few days to 1 week), their continuous use in these experiments does not simulate current clinical usage. Examination of the mechanical testing data also suggests that, for the most part, the inhibitory effects on fracture healing occurred earlier in the process than later, a finding consistent with other studies in which a variety of drugs have been shown to delay fracture healing. This may be explained by the fact that fracture healing is a biologically optimized process, and whereas some substances enhance and others impair it, there is a “catch up” phenomenon whereby differences in control and experimental groups eventually disappear. The finding of impaired fracture healing in the cox-2-null mouse is interesting and supports the concept that cox-2 function is essential for normal fracture healing. Despite these issues, the data reported by Simon et al.15 are compelling and consistent with those communicated by other investigators in studies that have not yet been published. Most notably, at the October 2001 meeting of the American Society for Bone and Mineral Research, Zhang et al.16 reported the critical role of cox-2 in mesenchymal cell differentiation during skeletal repair. These investigators also demonstrated impaired fracture healing and reduced expression of osteoblast and chondrocyte differentiation markers in in vitro cultures of cells derived from cox-2-null mice. These concerns may also extend to patients receiving dental implants and orthopedic patients who have undergone total joint arthroplasty with prostheses that require bone ingrowth and are using an NSAID or coxib to manage arthritis in another joint. At the February 2002 meeting of the Orthopaedic Research Society, Goodman et al.,17 using a validated drug test chamber in rabbits as a model of osteointegration, demonstrated that bone formation is suppressed by the oral administration of either an NSAID (naproxen) or coxib (rofecoxib). Thus, although there are some discrepancies among the effects of different drugs in the various studies reported (e.g., a report from our laboratory shows that parecoxib only mildly inhibits fracture repair),18 the overall body of data suggests that, at least in animal models, inhibition of cox-2 impairs bone healing. Because safety and ethical issues preclude a prospective clinical study from addressing this question, we may have to make clinical decisions based on our knowledge of basic mechanisms and the data derived from animal studies. Therefore, based on our current understanding, it would seem that a prudent approach toward the management of patients is to temporarily avoid or discontinue the use of both NSAIDs and coxibs during a period of bone healing.