Selective and Nonselective Cyclooxygenase-2 Inhibitors and Experimental Fracture-Healing

Abstract

Cyclooxygenase-2-specific anti-inflammatory drugs (coxibs) and nonspecific nonsteroidal anti-inflammatory drugs have been shown to inhibit experimental fracture-healing. The present study tested the hypothesis that these effects are reversible after short-term treatment. With use of a standard model of fracture-healing, identical ED50 dosages of either a nonsteroidal anti-inflammatory drug (ketorolac), a coxib (valdecoxib), or vehicle (control) were orally administered to rats for either seven or twenty-one days and fracture-healing was assessed with biomechanical, histological, and biochemical analyses. When healing was assessed at twenty-one days, the seven-day treatment produced only a trend for a higher rate of nonunion in valdecoxib and ketorolac-treated animals as compared with controls. No differences were observed at thirty-five days. The twenty-one-day treatment produced significantly more nonunions in valdecoxib-treated animals as compared with either ketorolac-treated or control animals (p < 0.05), but these differences disappeared by thirty-five days. The dose-specific inhibition of these drugs on prostaglandin E2 levels and the reversibility of the effects after drug withdrawal were assessed in fracture calluses and showed that ketorolac treatment led to twofold to threefold lower levels of prostaglandin E2 than did valdecoxib. Withdrawal of either drug after six days led to a twofold rebound in these levels by fourteen days. Histological analysis showed delayed remodeling of calcified cartilage and reduced bone formation in association with valdecoxib treatment. Cyclooxygenase-2-specific drugs inhibit fracture-healing more than nonspecific nonsteroidal anti-inflammatory drugs, and the magnitude of the effect is related to the duration of treatment. However, after the discontinuation of treatment, prostaglandin E2 levels are gradually restored and the regain of strength returns to levels similar to control.