Introduction: Selective COX-2 (Cyclooxygenase-2) inhibitors have been found to impede fracture healing. The effect of selective COX-2 inhibitors on tendon healing in a bone tunnel, however, is unknown.

Methods: The authors performed bilateral anterior cruciate ligament reconstructions in 32 rabbits and used peripheral quantitative computed tomography (pQCT) to compare tendon-to-bone healing between tunnel aperture and midtunnel regarding bone mineral density (BMD) and ingrowth of new bone. Each animal was assigned to one of four groups. Two groups received selective COX-2 inhibitors orally for 3 weeks (Cele-coxib; 10 mg/kg/d), the two other groups received no COX-2 inhibitors (controls). The animals were sacrificed 3 and 6 weeks after surgery. In biomechanical testing maximum load to failure and stiffness of the tendon grafts were calculated from the load displacement curve and failure modes were recorded. To assess indirectly the effect on local COX-2 activity the synovial content of Prostaglandin E2 (PGE2), the major metabolite of arachnidonic acid metabolism and catalyzed by COX-2, was measured by Enzyme-linked Immunosorbent Assay (ELISA).

Results: Animals treated with selective COX-2 inhibitors had significantly lower BMD at the tunnel aperture (P=.02). In all groups the BMD at the tunnel aperture was significantly higher in comparison with the midtunnel (P< .05). In the controls ingrowth of new bone was greater at the tunnel aperture at 3 weeks (P=.028). After 3 weeks of COX-2 inhibitor administration synovial fluid concentrations of PGE2 were significantly lowered (P=.018) and increased more than threefold by 6 weeks after surgery and 3 weeks after last drug administration (P=.022), while in the controls there was a decrease in PGE2 between week 3 and 6. At 6 weeks the controls exhibited a twofold increase in maximum load to failure (3 weeeks: 28.2±20.9 N; 6 weeks: 59.6±53.6 N; P=.394), whereas the COX-2 inhibitor treated specimens decreased 1.9fold (3 weeks: 69.3±50.5 N; 6 weeks: 37.4±16.8 N; P=.24). Maximum load to failure values correlated with PGE2 changes, but not statistically significant (r²= −0,502; p=0,056). Failure modes at 3 and 6 weeks were rupture and degloving, respectively, of the tendon graft.

Discussion: This study revealed decreased bone mineral density at the tunnel aperture at 3 weeks, an increase of the inflammatory mediator PGE2 and decreased graft stability with time after treatment with selective COX-2 inhibitors. Untreated controls appeared to have a more physiological healing course with a continuous decrease in PGE2 and an increase in graft stability. Our results suggest, that selective COX-2 inhibitors may delay tendon healing in a bone tunnel.