Dose and Time-Dependent Effects of Cyclooxygenase-2 Inhibition on Fracture-Healing

Simon, Ann Marie BS; O'Connor, J. Patrick PhD

Journal of Bone & Joint Surgery - American Volume:
doi: 10.2106/JBJS.F.00127
Scientific Articles

Background: Fracture-healing is impaired in mice lacking a functional cyclooxygenase-2 (COX-2) gene or in rats continuously treated with COX-2 inhibitors. These observations indicate that COX-2 is a critical regulator of fracture repair. Nonsteroidal anti-inflammatory drugs are commonly used to treat pain associated with musculoskeletal trauma and disease. Nonsteroidal anti-inflammatory drugs inhibit COX-2 function and in so doing can impair fracture-healing. The goal of the present study was to determine how variations in nonsteroidal anti-inflammatory drug therapy ultimately affect fracture-healing.

Methods: Closed femoral fractures were made in female Sprague-Dawley rats. The rats were treated with different doses of celecoxib (a COX-2-selective nonsteroidal anti-inflammatory drug) or were treated for different periods before or after fracture with celecoxib. Eight weeks after the fracture, healing was assessed with radiography and destructive torsional mechanical testing. The effect of celecoxib treatment on fracture callus prostaglandin E2 and F levels was determined as a measure of cyclooxygenase activity.

Results: Celecoxib doses as small as 2 mg/kg/day reduced fracture callus mechanical properties and caused a significant increase in the proportion of nonunions. Similarly, treatment with celecoxib at a dose of 4 mg/kg/day for just five days reduced fracture callus mechanical properties and significantly increased the proportion of nonunions. Conversely, celecoxib therapy prior to fracture or initiated fourteen days after fracture did not significantly increase the proportion of nonunions. Celecoxib treatment at a dose of 4 mg/kg/day reduced fracture callus prostaglandin E2 and F levels by >60%.

Conclusions: COX-2-selective nonsteroidal anti-inflammatory drug therapy during the early stages of fracture repair significantly reduced fracture callus mechanical properties at later stages of healing and increased the proportion of nonunions in this animal model.

Clinical Relevance: Celecoxib therapy in the therapeutic range used by humans significantly impaired fracture-healing if administered during the first two weeks after a fracture in this animal model. However, celecoxib therapy prior to fracture or after the early stages of fracture-healing did not impede healing. These observations indicate that nonsteroidal anti-inflammatory drug therapy during the early stages of fracture-healing probably should be avoided.

Author Information

1 Department of Orthopaedics, UMDNJ-New Jersey Medical School, MSB G580/ORTHO, 185 South Orange Avenue, Newark, NJ 07103. E-mail address for J.P. O'Connor:

Article Outline

The process of normal bone fracture-healing can be divided into four phases: immediate, early, regenerative, and remodeling. The immediate phase is characterized by localized tissue hypoxia and hematoma formation. Hypoxia is thought to be one of the signaling events used to initiate the healing cascade1,2. The early phase is characterized by an inflammation response and recruitment of cells to the fracture site through chemoattraction and increased proliferation3-5. In the regenerative phase, the fracture is bridged with new bone through a combination of intramembranous and endochondral ossification that requires angiogenesis6-10. Cells that have migrated to or proliferated in the fracture callus differentiate into chondrocytes and osteoblasts during endochondral ossification. In the final or remodeling phase, the newly formed woven bone that spans the fracture site is remodeled into mature lamellar bone to restore the mechanical integrity of the bone11. The phases of fracture-healing have been well characterized histologically and mechanically. Much less is known of the molecular events that initiate, maintain, or coordinate the different phases of fracture-healing.

The early phase may be critical for the ultimate success of fracture-healing. Several studies have shown increased levels or increased gene expression of pro-inflammatory cytokines, mitogenic growth factors, and cell-differentiating growth factors within the early fracture callus4,12-18. Given the known effects of these cytokines and growth factors on stem cells, osteoblasts, and chondrocytes, an obvious conclusion is that this milieu of signaling molecules promotes and maintains the healing pathway.

Inflammation, which occurs at the outset of the early phase of fracture-healing, is postulated to provide the initial signaling molecules for sustaining the healing cascade19. Immunohistochemical analysis of human fracture callus has shown transient increases in T-cells and granulocytes, a sustained increase in macrophages, and a paucity of B-cells during normal fracture-healing20. Pro-inflammatory cytokine levels also are increased in the fracture callus4. Loss of signaling capacity for tumor necrosis factor-α (TNF-α), a proinflammatory cytokine, dramatically inhibits fracture-healing, demonstrating a role for inflammation in fracture-healing21.

Cyclooxygenase-2 (COX-2) is an enzyme that is critical for normal inflammation22-24. COX-2 catalyzes the rate-limiting step in the synthesis of prostaglandins from arachidonic acid22. Prostaglandins are secreted bioactive lipids with diverse physiological effects, including pro-inflammatory effects25. Two genes, Cox1 and Cox2, encode cyclooxygenases22,26. Cyclooxygenase-1 (COX-1) is made constitutively in many tissues, including the stomach lining, whereas pro-inflammatory stimuli induce COX-2 synthesis.

Increased levels of E and F prostaglandins have been detected in the early phase of healing in rabbit osteotomy callus and surrounding muscle, suggesting a functional role for prostaglandins in healing27. More recently, it has been shown that fracture-healing is impaired in mice lacking a functional Cox2 gene28,29. These observations again indicate a functional role for inflammation in fracture-healing.

Nonsteroidal anti-inflammatory drugs are commonly used to treat pain and inflammation in patients with chronic musculoskeletal conditions and acute injuries, such as fractures30. Nonsteroidal anti-inflammatory drugs achieve the desired effect by inhibiting cyclooxygenase activity. Traditional nonsteroidal anti-inflammatory drugs, such as aspirin and indomethacin, inhibit COX-1 and COX-2 to effectively reduce inflammation and pain but also can lead to gastrointestinal bleeding, impaired blood clotting, and other side effects31-33. Newer COX-2-selective nonsteroidal anti-inflammatory drugs are effective anti-inflammatory medications and analgesics but produce fewer episodes of gastrointestinal bleeding and do not affect blood clotting34.

On the basis of the hypothesis that inflammation is necessary for normal fracture-healing, nonsteroidal anti-inflammatory drug treatment has been shown to delay fracture-healing in several animal models35-39. More recently, it has been shown that COX-2-selective nonsteroidal anti-inflammatory drug treatment also can impair fracture-healing in rats and can inhibit bone-healing in rabbits, findings that are in accord with observations of impaired fracture-healing in mice lacking a functional Cox2 gene28,40. These experiments support the hypothesis that inflammation is necessary for fracture-healing and that COX-2 is specifically necessary for this process. In previous experiments, COX-2 activity was nominally inhibited or absent throughout most of the time-course of fracture-healing28. Those findings have led to the present study, in which the effects of nonsteroidal anti-inflammatory drug dose and treatment regime were evaluated in order to determine how COX-2 inhibition affects ultimate healing outcomes.

Back to Top | Article Outline

Materials and Methods

Animal Procedures and Disposition

In the present study, female Sprague-Dawley rats received celecoxib once a day by means of oral gavage. Celecoxib has a plasma elimination half-life of fourteen hours in female rats, compared with 3.73 hours in male rats41. In humans, the elimination half-life of celecoxib is approximately eleven hours42. Thus, celecoxib was chosen as the experimental COX-2-selective nonsteroidal anti-inflammatory drug and female rats were chosen as the animal model because the pharmacokinetics of celecoxib in female rats closely mimics those in humans.

Three hundred and seventy-seven female Sprague-Dawley rats weighing an average (and standard deviation) of 272 ± 7 g were used in the study. Ninety-eight of these rats were excluded from the study because of fracture destabilization, anesthesia-related death, poor fracture quality, infection, or experimental error (Tables I and II). Fracture destabilization was considered to have occurred if the intramedullary rod that was used to stabilize the fracture (described below) had dislodged from the proximal part of the femur and moved distally such that adequate fracture reduction was not assured. A number of rats were used only for radiographic analysis because the fracture was too proximal or distal to the midpoint of the femur and thus was not acceptable for mechanical testing but was adequate for radiographic scoring or because an error that occurred during the mechanical testing precluded use of the mechanical data but not the radiographic data.

The rats were divided randomly into three study groups: a dose-dependent study group, a time-dependent study group, and a time-delay study group. The animals in the three study groups were treated with celecoxib or vehicle (1% methylcellulose) by means of oral gavage once a day. In addition, eighteen animals were randomly selected for the determination of the effect of nonsteroidal anti-inflammatory drug treatment on fracture callus prostaglandin levels. Three treatment groups comprising six animals each received either vehicle (control), celecoxib (4 mg/kg), or diclofenac (5 mg/kg) and were analyzed at four days after fracture. One rat that was treated with diclofenac died prior to the end point. The overall success rate for rats reaching their experimental end point and successful data acquisition was 74%. The New Jersey Medical School Institutional Animal Use and Care Committee approved all animal procedures. Diclofenac was purchased from Cayman Chemical (Ann Arbor, Michigan), celecoxib was obtained in the form of Celebrex capsules (Pfizer, New York, NY), and methylcellulose was purchased from Sigma (St. Louis, Missouri).

The dose-dependent study involved 154 female rats. These animals were randomly selected into four treatment groups that received (1) vehicle only (the control group) (forty-eight rats), (2) celecoxib at a dose of 2 mg/kg (thirty-two rats), (3) celecoxib at a dose of 4 mg/kg (thirty-eight rats), and (4) celecoxib at a dose of 8 mg/kg (thirty-six rats). The animals were treated once daily beginning four hours after the fracture and continuing for fifteen days. The time-dependent study involved 139 rats. These animals were selected randomly into four groups that were treated with 4 mg/kg of celecoxib once a day for (1) five days (thirty-seven rats), (2) ten days (forty-four rats), (3) twenty-one days (twenty rats), or (4) twenty-eight days (thirty-eight rats) beginning four hours after the fracture. The time-delay study involved sixty-six rats. These animals were selected randomly into three groups that received (1) a five-day pretreatment (twenty-three rats), (2) a seven to twenty-eight-day treatment (twenty rats), or (3) a fourteen to twenty-eight-day treatment (twenty-three rats). The drug treatment was started either five days before the fracture or on the seventh or fourteenth day after the fracture, depending on the experimental group. In the fourteen to twenty-eight-day treatment group and the seven to twenty-eight-day treatment group, treatment was continued until the twenty-eighth day after the fracture. In the pretreatment group, a femoral fracture was induced in each rat four hours after the fifth and final celecoxib dose.

The rats were anesthetized with an intraperitoneal injection of ketamine (40 mg/kg) and xylazine (5 mg/kg). The right hindlimb was shaved and was washed with a povidoneiodine solution. Under aseptic conditions, a medial parapatellar incision (0.4 to 0.5 cm) was created. The patella was dislocated laterally and the medullary canal was entered through the intercondylar notch and was reamed with a 21-gauge needle. A 0.7-mm stainless steel pin (Small Parts, Miami Lakes, Florida) was then inserted through the medullary canal into the proximal part of the greater trochanter and was tapped into place. The distal portion of the pin was cut flush with the femoral condyles so as not to interfere with knee function. The patellar dislocation was reduced, and the soft tissue and skin were closed with 4-0 Vicryl resorbable sutures in two layers. After the incision was closed, the middle of the diaphysis of the pinned femur was fractured with use of a custom-made three-point bending device (BBC Specialty Automotive Center, Linden, New Jersey) as described previously28,43. The animals were caged in pairs and were allowed to walk freely after surgery.

Back to Top | Article Outline
Radiographic Analysis

All animals were examined with mediolateral radiographs immediately after the fracture and at the time of death. Radiographs were made with use of a Packard Faxitron (model 804; Field Emission, McMinnville, Oregon) and Kodak Min-R 2000 film (Eastman Kodak, Rochester, New York). In addition, radiographs were made of femora that were designated for mechanical testing after resection. Blinded samples of the eight-week post-resection radiographs were independently examined by three observers (including one of the authors [J.P.O'C.]) and were graded by assigning a score ranging from 0 to 4 as described previously44. The grading scheme was based on the bridging of the fracture by callus and the cortical bone at the fracture site. One point was assigned to each aspect of the fracture (that is, the right callus and cortex and the left callus and cortex) that appeared bridged. A score of 0 represented the absence of radiographic bridging among all four aspects of the fracture, indicating a fracture nonunion. A score of 4 represented a fully bridged fracture.

Back to Top | Article Outline
Mechanical Testing

Animals within each treatment group were killed at eight weeks after the fracture by means of CO2 asphyxiation. Animals with an oblique fracture, a comminuted fracture, or an infection involving the femur were excluded from mechanical testing. Femora were resected and cleaned of all soft tissue, with the fracture callus being left intact. Femoral length, maximum and minimum fracture callus diameters, and maximum and minimum mid-diaphyseal diameters of the intact (contralateral) femur were measured with use of digital calipers. The intramedullary rod was left in place because, as a result of the smaller size of the animals, removal of the rod would damage the condyles and also the integrity of the fracture callus. It was postulated that the intramedullary rod would not interfere with the mechanical testing because it lay along the neutral axis of the torsional test. The femoral ends were potted in 1-in (2.54-cm) hexagonal nuts with use of a low-melt-temperature metal (Wood's metal; Alfa Aesar, Ward Hill, Massachusetts). The samples were wrapped in saline solution-soaked gauze to prevent dehydration between steps.

Once the femoral ends had been potted, the gauge length of the sample was measured and torsional testing was conducted with use of a servohydraulic testing machine (MTS, Eden Prairie, Minnesota) with a 20-Nm reaction torque cell (Interface, Scottsdale, Arizona). The testing was carried out to failure at an actuator head displacement rate of 2° per second and a data recording rate of 20 Hz. The fractured and intact femora were tested in internal rotation and proper anatomic orientation. The failure mode of each femur during the mechanical testing procedure was determined by means of visual inspection and was designated as union (indicating that the femur had failed spirally), incomplete union (indicating that the femur had some osseous bridging but had failed principally along the original fracture line), or nonunion (indicating that the femur had failed along the original fracture line and had no evidence of osseous bridging). The results were recorded with an Olympus C-3040 Zoom digital camera (Olympus Imaging America, Center Valley, Pennsylvania).

The peak torque and the angle at the time of failure were obtained from the load-deformation curves. Internal callus dimensions were measured after torsional testing. The femoral dimensions were used to calculate shear stress, shear modulus, and torsional rigidity45,46. The femora were modeled as hollow ellipses, and the polar moment of inertia was calculated45,47.

Back to Top | Article Outline
Fracture Callus Prostaglandin Level

Animals within each treatment group were killed one hour after the final drug dose was administered on the fourth day after the fracture. The femora were resected and the fracture callus was isolated and flash frozen with use of liquid nitrogen. The samples were then weighed and pulverized with a mortar and pestle. The pulverized callus was extracted with five volumes of M-PER reagent (Pierce, Rockford, Illinois) that was supplemented with protease inhibitors (Sigma Aldrich, St. Louis, Missouri). The samples were placed on a mixer at 4°C for thirty minutes. Insoluble material was removed from the extract by means of centrifugation (10,000 RPM for ten minutes). The supernatant (clarified extract) was collected and stored at –80°C. Two milliliters of ethanol was added to 0.5 mL of clarified extract, and the precipitated proteins were removed by centrifugation at 3000 times gravity for ten minutes. The supernatant was dried in a vacuum and then was dissolved in 1-M citrate buffer (pH 4). The samples then were applied to 500-mg C18 columns (Waters Sep-Pak; Waters, Milford, Massachusetts) that had been preactivated by means of methanol and water washes. The columns were washed with water and then hexane. The eicosanoids were eluted with 5 mL of ethyl acetate containing 1% methanol. The eluted eicosanoids were dried in a vacuum and were resuspended in EIA buffer (0.1-M sodium phosphate, pH 7.4; 0.4-M sodium chloride; 0.1% bovine serum albumin; 1-mM ethylenediaminetetraacetic acid; and 0.01% sodium azide). Prostaglandin E2 (PGE2) and prostaglandin F (PGF) concentrations were determined by means of enzyme-linked immunoassay (Cayman Chemical). The assays and analyses were performed according to the manufacturer's instructions. The PGE2 and PGF concentrations were normalized to total protein concentration measured with use of bicinchoninic acid (BCA Protein Assay; Pierce)48.

Back to Top | Article Outline
Statistical Methods

SigmaStat software (version 3.0; SPSS, Chicago, Illinois) was used for all statistical analyses. The radiographic analysis, mechanical testing data, and prostaglandin levels were evaluated for significant difference with use of a one-way analysis of variance and post hoc Holm-Sidak tests. The failure mode was compared between groups with use of Fisher exact tests by comparing the proportion of nonunions in a treatment group with that in the control group. Data from specific treatment groups, such as the control group, are repetitively shown in the figures and tables for clarity. However, each outcome parameter (radiographic scores, peak torque, rigidity, maximum shear stress, and shear modulus) was compared between all of the treatment groups as a single statistical analysis because only one control group was used for these experiments. Differences between treatment groups were considered to be significant when the p value was ≤0.05. Analysis of variance indicated that statistical differences existed between the tested outcome parameters with p values of <0.001 for all tested outcomes and with statistical power of >0.9 (range, 0.94 to 0.999) for each test at an alpha value of 0.05.

Back to Top | Article Outline


Effects of Celecoxib Dose on Fracture-Healing

Inhibition of fracture-healing was detected in association with all doses of celecoxib used (Fig. 1; see Appendix). Rats were treated with celecoxib once per day, for fifteen days, beginning four hours after the fracture. The celecoxib doses were 2, 4, and 8 mg/kg. Healing was measured at eight weeks after the fracture by means of radiography and mechanical testing. Radiographic scoring showed that whereas the 2-mg/kg celecoxib dose had no significant effect on healing, the 4 and 8-mg/kg doses significantly impaired healing (Fig. 1, panel A). Torsional mechanical testing of the healing femora also demonstrated that celecoxib treatment impaired healing (Fig. 1, panels B and C). While a dose-dependent decrease in peak torque was observed, the differences between the control and celecoxib-treated samples were significantly different only in the 4 and 8-mg/kg celecoxib treatment groups (see Appendix). In contrast, the rigidity (Fig. 1, panel B), shear stress (Fig. 1, panel C), and shear modulus (see Appendix) of the healing femora from the celecoxib-treated rats were all significantly reduced relative to the healing femora from control rats. Finally, all femora were examined after torsional mechanical testing to determine if the femur failed through the callus as a spiral fracture (indicating union), principally along the original fracture with some bone-bridging evident (indicating incomplete union), or along the original fracture with no bone-bridging apparent (indicating nonunion). All thirty-seven femora from the control rats failed as unions (twenty-nine femora) or incomplete unions (eight femora). Celecoxib treatment, however, significantly increased the number of nonunions in the 2, 4 and 8-mg/kg celecoxib treatment groups (with nonunion occurring in five of fourteen, thirteen of twenty-three, and four of thirteen femora, respectively) (Fig. 1, panel D).

These data show that, in rats, celecoxib treatment within the therapeutic range used by humans can significantly impair fracture-healing. On the basis of these data, subsequent experiments employed the 4-mg/kg celecoxib dose to determine the effects of treatment regime on fracture-healing.

Back to Top | Article Outline
Effects of Celecoxib Treatment Duration on Fracture-Healing

Continuous pharmacological inhibition of COX-2 or ablation of the COX-2 gene dramatically impairs fracture-healing28. However, this is unlike the common clinical scenario when patients use analgesics for the first one or two weeks following a fracture. Thus, it is possible that inhibition of COX-2 activity during the early phases of fracture-healing may have little or no deleterious effects on ultimate healing outcomes and that it is only sustained inhibition of COX-2 or inhibition of COX-2 during a critical later phase of healing that impairs fracture-healing.

To address this question, rats were treated with a human therapeutic dose of celecoxib (4 mg/kg) for five, ten, fifteen, twenty-one, or twenty-eight days, beginning four hours after the fracture.

The treatment periods spanned phases of fracture-healing from the early inflammatory phase to normal fracture-bridging (approximately twenty-eight days). Fracture-healing was assessed at eight weeks by means of radiography and mechanical testing (Fig. 2; see Appendix). Radiographic scoring at eight weeks after the fracture showed significant impairment of fracture-healing when celecoxib treatment had lasted fifteen days or longer (Fig. 2, panel A). In contrast, torsional mechanical testing demonstrated that celecoxib treatment significantly reduced peak torque (see Appendix), rigidity (Fig. 2, panel B), maximum shear stress (Fig. 2, panel C), and shear modulus at all treatment periods (see Appendix). Post-mechanical testing assessment of the failure mode showed that five days of celecoxib treatment significantly increased the prevalence of nonunion after eight weeks of healing (with nonunion occurring in nine of twenty-five femora) (Fig. 2, panel D). The proportion of nonunions after ten days of celecoxib treatment was lower (with nonunion occurring in four of twenty-six femora), but it increased to >50% after fifteen, twenty-one, or twenty-eight days of celecoxib treatment (with nonunion occurring in thirteen of twenty-three, eight of twelve, and twelve of twenty femora, respectively; Fig. 2, panel D). These data suggest that the duration of COX-2 inhibition therapy correlates with the inhibition of fracture-healing. Interestingly, after ten days of celecoxib treatment, peak torque (see Appendix) and maximum shear stress (Fig. 2, panel C) were not significantly different from control values but rigidity (Fig. 2, panel B) and shear modulus (see Appendix) were significantly less.

Back to Top | Article Outline
Effects of Delayed or Previous Celecoxib Treatment on Fracture-Healing

Additional clinical scenarios of nonsteroidal anti-inflammatory drug therapy were tested for potential negative effects on fracture-healing. Specifically, we sought to determine whether nonsteroidal anti-inflammatory drug use prior to fracture affects healing as might occur in a chronic nonsteroidal anti-inflammatory drug user, such as a patient with arthritis. We also sought to determine when nonsteroidal anti-inflammatory drug use can be resumed following a fracture. To mimic these clinical scenarios, rats were treated with celecoxib (4 mg/kg/day) for five days, receiving the final celecoxib dose four hours prior to the femoral fracture (pre-5 days). In the other groups, celecoxib treatment (4 mg/kg/day) was initiated seven or fourteen days after the fracture and was continued to the twenty-eighth day after the fracture, when fracture bridging normally occurs in the rat.

Prior celecoxib treatment had little negative effect on fracture-healing. Radiographic scoring, peak torque, maximum shear stress, shear modulus, and the prevalence of nonunion (zero of fourteen) were similar to control values (Fig. 3; see Appendix). One exception was that femoral torsional rigidity was significantly lower after eight weeks of healing in the pretreated rats (Fig. 3, panel B).

Delayed celecoxib treatment was associated with improvement in many healing outcomes as compared with celecoxib treatment that began immediately after the fracture, although most values were still less than the control values. Continuous celecoxib treatment for twenty-eight days after the fracture was associated with the lowest radiographic and mechanical scores (Fig. 3; see Appendix). Delaying celecoxib treatment until seven days after the fracture was associated with a higher radiographic score, although it was still significantly less than the control value (Fig. 3, panel A). Rats that were treated with celecoxib from the seventh through the twenty-eight day also had reduced rigidity (Fig. 3, panel B) and shear modulus (see Appendix), but the values for peak torque (see Appendix) and maximum shear stress (Fig. 3, panel C) were similar to control values. The proportion of nonunions in the seven to twenty-eight-day celecoxib treatment group (four of fifteen) was reduced compared with that in the group that received continuous celecoxib treatment (Fig. 3, panel D). Delaying celecoxib treatment until fourteen days after the fracture was associated with normalized radiographic scores (Fig. 3, panel A), peak torque (see Appendix), and maximum shear stress (Fig. 3, panel C), with a lower proportion of nonunions (two of twelve) compared with the group that received continuous celecoxib treatment (Fig. 3, panel D), although the values for torsional rigidity (Fig. 3, panel B) and shear modulus (see Appendix) remained significantly lower than control values.

Back to Top | Article Outline
Effects of Nonsteroidal Anti-Inflammatory Drug Therapy on Fracture Callus Prostaglandin Levels

The straightforward hypothesis indicated by these experimental observations is that celecoxib treatment reduces COX-2 activity and prostaglandin levels that are essential for normal fracture-healing to proceed. To test this hypothesis, rats were treated with carrier (1% methylcellulose), celecoxib (4 mg/kg), or diclofenac (5 mg/kg) once a day for four days after the fracture. One hour after drug dosing on the fourth day, the rats were killed, eicosanoids were extracted from the callus, and PGE2 and PGF were quantified. Celecoxib or diclofenac treatment reduced callus PGE2 levels by >60% (Fig. 4, panel A). Similarly, PGF levels were reduced significantly (by >75%) in the fracture calluses of the rats that had been treated with nonsteroidal anti-inflammatory drug therapy (Fig. 4, panel B). These data indicated that nonsteroidal anti-inflammatory drug therapy reduced prostaglandin levels at the fracture site.

Back to Top | Article Outline


With use of a rodent model, the effects of celecoxib (a COX-2-selective nonsteroidal anti-inflammatory drug) on fracture-healing were investigated in detail. The data suggest that higher doses (Fig. 1) and longer periods of celecoxib treatment (Fig. 2) are more detrimental to fracture-healing than lower doses and shorter periods of celecoxib treatment are. However, rats that were treated with a modest dose of celecoxib (4 mg/kg/day) for five days following a fracture had significantly worse outcomes after eight weeks of healing than control rats did. These experimental observations suggest that nonsteroidal anti-inflammatory drug therapy following a fracture may adversely affect healing in humans. We are unaware of any prospective human studies that support this contention. However, recent human retrospective studies have supported these experimental findings49,50.

In the present study, 43% (sixty-eight) of 160 rats that were treated with celecoxib after the fracture appeared to have union after eight weeks of healing, 19% (thirty-one) showed some evidence of new bone bridging the fracture site, and 38% (sixty-one) appeared to have a nonunion. In contrast, 82% (forty-two) of fifty-one control rats or rats that had been treated with celecoxib prior to fracture had union, 18% (nine) showed some evidence of new bone bridging the fracture site, and 0% were observed to have a nonunion. This dichotomy in the healing pattern of the celecoxib-treated rats creates considerable variation within the experimental results and could lead to discrepancies in how results were interpreted in studies that have used fewer animals or employed a limited number of outcome measures28,51,52. The large number of animals used in the present study shows that COX-2-selective nonsteroidal anti-inflammatory drug therapy impairs fracture-healing.

Five days of treatment with celecoxib at a dose of 4 mg/kg/day significantly impaired fracture-healing, with 36% (nine) of twenty-five rats having nonunion after eight weeks. In female Sprague-Dawley rats, bridging of a fracture site is normally observed by four weeks after a fracture. Since celecoxib is nominally inhibiting COX-2 (Fig. 4), these observations indicate that inhibiting the early inflammation phase of fracture-healing can ultimately impair fracture-healing at later times.

In the group in which celecoxib treatment was delayed for two weeks after the fracture, certain mechanical testing values were less than those in control animals (Fig. 3) and nonunion still developed in two (17%) of twelve rats. These findings suggest that COX-2 may have an additional function during fracture-healing beyond those associated with inducing or enhancing inflammation.

A curious observation was that five or fifteen days of celecoxib therapy was more deleterious than ten days of treatment after the fracture (Fig. 2). The rigidity of the femora of the rats that had been treated with celecoxib for ten days was significantly less than that in the control group, and the proporation of nonunions was signficantly higher than that in the control group. However, other outcome parameters were similar to control values. One potential explanation for this observation may be that a bounce-back effect occurs when COX-2 expression is still high and that discontinuing celecoxib treatment leads to a burst of prostaglandin synthesis that rescues healing. In support of this explanation, it was previously shown that COX-2 expression peaks on the third day after a fracture, declines by the fifth day, peaks again on the tenth day, and declines to background levels by the twenty-first day during fracture-healing in rats51. When celecoxib therapy is stopped on the fifth day, the fifteenth day, or later, the level of COX-2 expression may not be sufficient to provide a bounce-back effect of sufficient magnitude to rescue healing as may occur when celecoxib therapy is stopped on the tenth day.

The mechanistic functions of COX-2 during fracture-healing are not known; thus, the mechanism by which celecoxib inhibits fracture-healing is not understood. There are a number of possible mechanisms, and it is likely that multiple cellular and molecular mechanisms involved in fracture-healing are impaired by COX-2 inhibition. For instance, angiogenesis is essential for successful fracture-healing and endochondral ossification6-9. Prostaglandins are known to promote angiogenesis53,54. Furthermore, inhibition of COX-2 has been shown to inhibit angiogenesis in animal tumor models55-57. Thus, one likely mechanism to account for impaired fracture-healing in celecoxib-treated rats is that celecoxib treatment reduces angiogenesis at the fracture site. It is conceivable that because of a reduction in prostaglandin levels at the fracture site, the initial inflammatory phase is blunted and the local production and release of cytokines and growth factors that recruit mesenchymal cells to the fracture site are reduced. In due course, this leads to failed healing in some animals. An additional possibility is that inhibition of COX-2 alters the normal temporal pattern of cellular and molecular events, leading to an uncoordinated healing response. Ultimately, inhibition of cyclooxygenase activity that reduces PGE2 levels by >60% and PGF levels by >75% is sufficient to impair fracture-healing in a large proportion of animals.

Prostaglandins have direct effects on chondrocytes and osteoblasts. In vitro, PGF increases chondrocyte proliferation and matrix synthesis58,59. PGE2 does not enhance chondrocyte matrix synthesis but inhibits the expression of genes associated with the terminally differentiated phenotype of chondrocytes (type-X collagen, vascular endothelial growth factor [VEGF], and alkaline phosphatase), and only enhances chondrocyte proliferation at very high concentrations58-60. Osteoblasts respond to prostaglandins by increasing cell division, elaborating more osteoid, and enhancing osteoclast activity61-68. Thus, one could expect that chondrocyte as well as osteoblast metabolism would be affected by a decrease in fracture callus prostaglandin levels.

Other experimental evidence also indicates that COX-2 is necessary for fracture repair. For example, the treatment of fractures with PGE2 can enhance fracture-healing, but the PGE2 must be administered locally and continuously to be effective69. Similarly, compounds that stimulate the EP2 or EP4 PGE2 receptors also have been shown to stimulate osteogenesis and fracture repair70-72.

There have been few human studies concerning the function of COX-2 or its metabolites in fracture-healing. Recent retrospective studies have correlated nonsteroidal anti-inflammatory drug use with impaired fracture-healing49,50. In particular, Burd and colleagues performed a retrospective study of the effects of indomethacin therapy on the prevalence of nonunions of long-bone fractures among patients who had also sustained an acetabular fracture50. Data for the analysis by Burd and colleagues came from a prospective study comparing the efficacy of localized radiation therapy with that of indomethacin therapy for the prevention of heterotopic ossification following an acetabular fracture. They found that patients who had received localized radiation therapy had a 7% rate of nonunion (five nonunions among 118 fractures in seventy-four patients), whereas those who had received indomethacin therapy had a 29% rate of nonunion (eleven nonunions among seventy-two fractures in thirty-eight patients). Several studies have indicated that nonsteroidal anti-inflammatory drug use reduces the prevalence and severity of heterotopic ossification in humans, again indicating a role for COX-2 in human osteogenesis73-78.

The present study demonstrates that even short-term treatment with a COX-2 inhibitor can impair ultimate fracture-healing outcomes. In our previous study, we found that long-term administration of celecoxib at a dose of 4 mg/kg/day showed impaired fracture-healing in male rats on the basis of radiographs and histological findings but that no difference was noted between the mechanical properties of the healing femora from the control and celecoxib-treated rats28. We suggest that the difference between the previous experiments and the present study is the difference in pharmacokinetics of celecoxib between male and female rats. In male rats, the half-life is approximately four hours whereas in female rats it is approximately fourteen hours, which is similar to the half-life of celecoxib in humans41,42. Therefore, a 4 mg/kg/day dose of celecoxib in female rats should provide for substantially longer COX-2 inhibition over a twenty-four-hour cycle than the same dose in male rats. We theorize that this accounts for the difference in results between this study, our previous study, and another study that examined the effects of celecoxib on fracture-healing in male rats treated with a 3 mg/kg/day dose of celecoxib52.

The conclusions of the present study also differ from those reported by Gerstenfeld et al.51. In their study, Gerstenfeld et al. found that parecoxib treatment (at a dose of 0.3 or 1.5 mg/kg/day) ultimately did not impair fracture-healing in male rats. Unfortunately, the pharmacokinetics of parecoxib are not well described in the literature. However, in other studies, parecoxib has been used at doses of 6.4 and 10 mg/kg/day to assess the effects on tendon-healing and soft-tissue injury in rats79,80. Additional experiments examining the efficacy of valdecoxib, the active form of parecoxib, on models of pain and inflammation in rats have shown effective doses (ED50) of 5.9 mg/kg for edema and 14 mg/kg for hyperalgesia81. Thus, it is likely that equivalence in fracture-healing between control and parecoxib-treated rats as described was due to insufficient inhibition of COX-2 by the parecoxib dose used.

The data from the present study indicate that nonsteroidal anti-inflammatory drug use probably should be avoided by patients during fracture-healing, especially during the immediate and early inflammatory phases of fracture-healing. Conversely, nonsteroidal anti-inflammatory drug use prior to a fracture does not appear to have a negative effect on healing.

Additional research is needed to confirm these animal findings in humans and to define the mechanistic role of COX-2 in bone regeneration.

Back to Top | Article Outline


Tables showing the effects of celecoxib treatment on fracture-healing are available with the electronic versions of this article, on our web site at (go to the article citation and click on “Supplementary Material”) and on our quarterly CD-ROM (call our subscription department, at 781-449-9780, to order the CD-ROM). ▪

NOTE: The authors would like to dedicate this manuscript to the memory of Professor Fred Behrens, MD, a dear friend, colleague, and mentor. This work was supported by grants from the Orthopaedic Research and Education Foundation and the Arthritis Foundation to J.P.O'C.

Disclosure: In support of their research for or preparation of this work, one or more of the authors received, in any one year, outside funding or grants in excess of $10,000 from the Arthritis Foundation and the Orthopaedic Research and Education Foundation. Neither they nor a member of their immediate families received payments or other benefits or a commitment or agreement to provide such benefits from a commercial entity. No commercial entity paid or directed, or agreed to pay or direct, any benefits to any research fund, foundation, division, center, clinical practice, or other charitable or nonprofit organization with which the authors, or a member of their immediate families, are affiliated or associated.

A commentary is available with the electronic versions of this article, on our web site () and on our quarterly CD-ROM (call our subscription department, at 781-449-9780, to order the CD-ROM).

Investigation performed at the Department of Orthopaedics, UMDNJ-New Jersey Medical School, Newark, New Jersey

1. , Warren SM, Spector JA, Peled ZM, Gerrets RP, Greenwald JA, Longaker MT. Hypoxia and VEGF up-regulate BMP-2 mRNA and protein expression in microvascular endothelial cells: implications for fracture healing. Plast Reconstr Surg. 2002;109: 2384-97.
2. , Edmison JM, Caplan AI. Cultivation of rat marrow-derived mesenchymal stem cells in reduced oxygen tension: effects on in vitro and in vivo osteochondrogenesis. J Cell Physiol. 2001;187: 345-55.
3. , Spiegel HU. Cellular mechanisms of bone repair. J Invest Surg. 1997;10: 77-86.
4. , Majeska RJ, Rush EB, Levine PM, Horowitz MC. The expression of cytokine activity by fracture callus. J Bone Miner Res. 1995;10: 1272-81.
5. , Li X, Tomin E, Doty SB, Lane JM, Carney DH, Ryaby JT. Thrombin peptide (TP508) promotes fracture repair by up-regulating inflammatory mediators, early growth factors, and increasing angiogenesis. J Orthop Res. 2005;23: 671-9.
6. , Vu TH, Ryan AM, Kowalski J, Werb Z, Ferrara N. VEGF couples hypertrophic cartilage remodeling, ossification, and angiogenesis during endochondral bone formation. Nat Med. 1999;5: 623-8.
7. , McLean W, Ng YS, Fukai N, Reginato AM, Lovejoy S, D'Amore PA, Olsen BR. Skeletal defects in VEGF (120/120) mice reveal multiple roles for VEGF in skeletogenesis. Development. 2002;129: 1893-904.
8. , Carmeliet P, Moermans K, Stockmans I, Smets N, Collen D, Bouillon R, Carmeliet G. Impaired angiogenesis and endochondral bone formation in mice lacking the vascular endothelial growth factor isoforms VEGF164 and VEGF188. Mech Dev. 2002;111: 61-73.
9. , Schaffler MB, Majeska RJ. Prevention of fracture healing in rats by an inhibitor of angiogenesis. Bone. 2001;29: 560-4.
10. , Bao M, deGuzman L, Bunting S, Peale FV Jr, Ferrara N, Steinmetz H, Hoeffel J, Cleland JL, Daugherty A, van Bruggen N, Redmond HP, Carano RA, Filvaroff EH. Vascular endothelial growth factor stimulates bone repair by promoting angiogenesis and bone turnover. Proc Natl Acad Sci USA. 2002;99: 9656-61.
11. , Beaupre GS, Giori NJ, Helms JA. Mechanobiology of skeletal regeneration. Clin Orthop Relat Res. 1998;355 Suppl: S41-S55.
12. , Tsiridis E, Giannoudis PV. Current concepts of molecular aspects of bone healing. Injury. 2005;36: 1392-404.
13. , Cho TJ, Kon T, Aizawa T, Cruceta J, Graves BD, Einhorn TA. Impaired intramembranous bone formation during bone repair in the absence of tumor necrosis factor-alpha signaling. Cells Tissues Organs. 2001;169: 285-94.
14. , Gross M, Hall BK. Expression of four growth factors during fracture repair. Int J Dev Biol. 1993;37: 573-9.
15. , Kitazawa R, Maeda S, Mizuno K, Kitazawa S. Expression of platelet-derived growth factor proteins and their receptor alpha and beta mRNAs during fracture healing in the normal mouse. Histochem Cell Biol. 1999;112: 131-8.
16. , Gerstenfeld LC, Einhorn TA. Differential temporal expression of members of the transforming growth factor beta superfamily during murine fracture healing. J Bone Miner Res. 2002;17: 513-20.
17. , Joshi P, Zhou X, Kregor P, Hardy KJ, Devidas M, Scott P, Hughes JL. Production of interleukin-10 in human fracture soft-tissue hematomas. Shock. 1996;6: 3-6.
18. , Meyer MH, Tenholder M, Wondracek S, Wasserman R, Garges P. Gene expression in older rats with delayed union of femoral fractures. J Bone Joint Surg Am. 2003;85: 1243-54.
19. , Takebe J, Offenbacher S, Cooper LF. Macrophage cell lines produce osteoinductive signals that include bone morphogenetic protein-2. Bone. 2002;30: 26-31.
20. , Andrew SM, Freemont AJ, Marsh DR. Inflammatory cells in normal human fracture healing. Acta Orthop Scand. 1994;65: 462-6.
21. , Cho TJ, Kon T, Aizawa T, Tsay A, Fitch J, Barnes GL, Graves DT, Einhorn TA. Impaired fracture healing in the absence of TNF-alpha signaling: the role of TNF-alpha in endochondral cartilage resorption. J Bone Miner Res. 2003;18: 1584-92.
22. , DeWitt DL, Garavito RM. Cyclooxygenases: structural, cellular, and molecular biology. Annu Rev Biochem. 2000;69: 145-82.
23. , Bakhle YS, Botting RM. Cyclooxygenases 1 and 2. Annu Rev Pharmacol Toxicol. 1998;38: 97-120.
24. . Role and regulation of cylooxygenase-2 during inflammation. Am J Med. 1999;106: 37S-42S.
25. . Prostaglandins and leukotrienes: advances in eicosanoid biology. Science. 2001;294: 1871-5.
26. . Distribution and expression of cyclooxygenase (COX) isoenzymes, their physiological roles, and the categorization of nonsteroidal anti-inflammatory drugs (NSAIDs). Am J Med. 1999;107: 11S-7S.
27. , Lenthall G, Francis MJ. Release of prostaglandins from bone and muscle after tibial fracture. An experimental study in rabbits. J Bone Joint Surg Br. 1981;63: 185-9.
28. , Manigrasso MB, O'Connor JP. Cyclo-oxygenase 2 function is essential for bone fracture healing. J Bone Miner Res. 2002;17: 963-76.
29. , Schwarz EM, Young DA, Puzas JE, Rosier RN, O'Keefe RJ. Cyclooxygenase-2 regulates mesenchymal cell differentiation into the osteoblast lineage and is critically involved in bone repair. J Clin Invest. 2002;109: 1405-15. Erratum in: J Clin Invest. 2002;110:1211.
30. . Pharmacological basis for the therapy of pain and inflammation with non-steroidal anti-inflammatory drugs. Arthritis Res. 2000;2: 379-85.
31. . Gastrointestinal effects of nonsteroidal anti-inflammatory therapy. Am J Med. 1999;106: 3S-12S.
32. . Effects of nonsteroidal anti-inflammatory therapy on platelets. Am J Med. 1999;106: 25S-36S.
33. Nephrotoxicity of nonsteroidal anti-inflammatory drugs: physiologic foundations and clinical implications. Am J Med. 1999;106: 13S-24S.
34. , Patrono C. The coxibs, selective inhibitors of cyclooxygenase-2. N Engl J Med. 2001;345: 433-42.
35. , Sudmann E, Marton PF. Effect of indomethacin on fracture healing in rats. Acta Orthop Scand. 1976;47: 588-99.
36. , Dregelid E, Bessesen A, Morland J. Inhibition of fracture healing by indomethacin in rats. Eur J Clin Invest. 1979;9: 333-9.
37. , Wase A, Bear WT. Indomethacin and aspirin: effect of nonsteroidal anti-inflammatory agents on the rate of fracture repair in the rat. Acta Orthop Scand. 1980;51: 595-600.
38. , Troiano NW, Pelker RR, Gundberg CM, Friedlaender GE. The influence of ibuprofen on fracture repair: biomechanical, biochemical, histological, and histomorphometric parameters in rats. J Orthop Res. 1991;9: 383-90.
39. , Latta LL, Keer R, Renfree K, Hornicek FJ, Banovac K. Effect of nonsteroidal antiinflammatory drugs on fracture healing: a laboratory study in rats. J Orthop Trauma. 1995;9: 392-400.
40. , Ma T, Trindade M, Ikenoue T, Matsuura I, Wong N, Fox N, Genovese M, Regula D, Smith RL. COX-2 selective NSAID decreases bone ingrowth in vivo. J Orthop Res. 2002;20: 1164-9.
41. , Zhang JY, Breau AP, Hribar JD, Liu NW, Jessen SM, Lawal YM, Cogburn JN, Gresk CJ, Markos CS, Maziasz TJ, Schoenhard GL, Burton EG. Pharmacokinetics, tissue distribution, metabolism, and excretion of celecoxib in rats. Drug Metab Dispos. 2000;28: 514-21.
42. , McLachlan AJ, Day RO, Williams KM. Clinical pharmacokinetics and pharmacodynamics of celecoxib: a selective cyclo-oxygenase-2 inhibitor. Clin Pharmacokinet. 2000;38: 225-42.
43. , Einhorn TA. Production of a standard closed fracture in laboratory animal bone. J Orthop Res. 1984;2: 97-101.
44. , Min W, Simon AM, Sabatino C, O'Connor JP. A comparison between the effects of acetaminophen and celecoxib on bone fracture healing in rats. J Orthop Trauma. 2005;19: 717-23.
45. , Ekeland A, Langeland N. Methods for testing the mechanical properties of the rat femur. Acta Orthop Scand. 1978;49: 512-8.
46. , Engesaeter LB, Langeland N. Mechanical properties of fractured and intact rat femora evaluated by bending, torsional and tensile tests. Acta Orthop Scand. 1981;52: 605-13.
47. , Engesaeter LB, Langeland N. Torsional properties of rat femora. Eur Surg Res. 1984;16 Suppl 2: 28-33.
48. , Krohn RI, Hermanson GT, Mallia AK, Gartner FH, Provenzano MD, Fujimoto EK, Goeke NM, Olson BJ, Klenk DC. Measurement of protein using bicinchoninic acid. Anal Biochem. 1985;150: 76-85. Erratum in: Anal Biochem. 1987;163:279.
49. , MacDonald DA, Matthews SJ, Smith RM, Furlong AJ, De Boer P. Nonunion of the femoral diaphysis: the influence of reaming and non-steroidal anti-inflammatory drugs. J Bone Joint Surg Br. 2000;82: 655-8.
50. , Hughes MS, Anglen JO. Heterotopic ossification prophylaxis with indomethacin increases the risk of long-bone nonunion. J Bone Joint Surg Br. 2003;85: 700-5.
51. , Thiede M, Seibert K, Mielke C, Phippard D, Svagr B, Cullinane D, Einhorn TA. Differential inhibition of fracture healing by non-selective and cyclooxygenase-2 selective non-steroidal anti-inflammatory drugs. J Orthop Res. 2003;21: 670-5.
52. , Saunders MM, Kirsch T, Donahue HJ, Reid JS. Effect of COX-2-specific inhibition on fracture-healing in the rat femur. J Bone Joint Surg Am. 2004;86: 116-23.
53. , Gutierrez R, Valladares F, Varela H, Perez M. Intense vascular sprouting from rat femoral vein induced by prostaglandins E1 and E2. Anat Rec. 1994;238: 68-76.
54. , Auerbach R. PGE2 and angiogenesis. Proc Soc Exp Biol Med. 1983;172: 214-8.
55. , Wang H, Peskar BM, Levin E, Itani RM, Sarfeh IJ, Tarnawski AS. Inhibition of angiogenesis by nonsteroidal anti-inflammatory drugs: insight into the mechanisms and implications for cancer growth and ulcer healing. Nat Med. 1999;5: 1418-23.
56. , Koki A, Seibert K. COX-2 inhibitors. A new class of antiangiogenic agents. Ann N Y Acad Sci. 1999;889: 84-6.
57. , Ornberg RL, Wang Y, Zweifel BS, Koki AT, Masferrer JL. Cyclooxygenase-2 inhibition by celecoxib reduces proliferation and induces apoptosis in angiogenic endothelial cells in vivo. Cancer Res. 2002;62: 625-31.
58. , Fu YH, McDougall S, Polendo R, Williams A, Benya PD, Hahn TJ. Effects of prostaglandins on deoxyribonucleic acid and aggrecan synthesis in the RCJ 3.1C5.18 chondrocyte cell line: role of second messengers. Endocrinology. 1996;137: 2208-16.
59. , Demarteau O, Suetterlin R, Heberer M, Martin I. Chondrogenesis of expanded adult human articular chondrocytes is enhanced by specific prostaglandins. Rheumatology (Oxford). 2004;43: 852-7.
60. , Zuscik MJ, Ionescu AM, Zhang X, Rosier RN, Schwarz EM, Drissi H, O'Keefe RJ. PGE2 inhibits chondrocyte differentiation through PKA and PKC signaling. Exp Cell Res. 2004;300: 159-69.
61. , Fall PM, Raisz LG. Anabolic effects of prostaglandins in cultured fetal rat calvariae: structure-activity relations and signal transduction pathway. J Bone Miner Res. 1996;11: 1249-55.
62. , Harada S, Matsumoto T, Tezuka K, Higashino K, Kodama H, Hashimoto-Goto T, Ogata E, Kumegawa M. Prostaglandin F2 alpha stimulates proliferation of clonal osteoblastic MC3T3-E1 cells by up-regulation of insulin-like growth factor I receptors. J Biol Chem. 1991;266: 21044-50.
63. , Hotta T, Kurihara N, Ikeda E, Maeda N, Yagyu Y, Kumegawa M. Prostaglandin E1 and F2 alpha stimulate differentiation and proliferation, respectively, of clonal osteoblastic MC3T3-E1 cells by different second messengers in vitro. Endocrinology. 1987;121: 1966-74.
64. , Takasugi I, Fujieda M, Kiriu M, Mizuochi S, Ide H. Prostaglandin E2 stimulates the formation of mineralized bone nodules by a cAMP-independent mechanism in the culture of adult rat calvarial osteoblasts. J Cell Biochem. 1999;73: 36-48.
65. , Casinghino S, McCarthy TL. Differential actions of prostaglandins in separate cell populations from fetal rat bone. Endocrinology. 1994;135: 1611-20.
66. , Di Bon A, van der Plas A, Lowik CW, Nijweide PJ. Effects of exogenous prostanoids on the proliferation of osteoblast-like cells in vitro. Prostaglandins. 1985;30: 827-40.
67. , Pilbeam CC, Pan L, Breyer RM, Raisz LG. Effects of prostaglandin E2 on gene expression in primary osteoblastic cells from prostaglandin receptor knockout mice. Bone. 2002;30: 567-73.
68. , Lorenzo JA, Freeman AM, Tomita M, Morham SG, Raisz LG, Pilbeam CC. Prostaglandin G/H synthase-2 is required for maximal formation of osteoclast-like cells in culture. J Clin Invest. 2000;105: 823-32.
69. , Klamer A, Bak B, Suder P. Effect of local prostaglandin E2 on fracture callus in rabbits. Acta Orthop Scand. 1993;64: 59-63.
70. , Borovecki F, Ke HZ, Cameron KO, Lefker B, Grasser WA, Owen TA, Li M, DaSilva-Jardine P, Zhou M, Dunn RL, Dumont F, Korsmeyer R, Krasney P, Brown TA, Plowchalk D, Vukicevic S, Thompson DD. An EP2 receptor-selective prostaglandin E2 agonist induces bone healing. Proc Natl Acad Sci USA. 2003;100: 6736-40.
71. , Sakai A, Uchida S, Tanaka S, Nagashima M, Katayama T, Yamaguchi K, Nakamura T. Prostaglandin E2 receptor (EP4) selective agonist (ONO-4819.CD) accelerates bone repair of femoral cortex after drill-hole injury associated with local upregulation of bone turnover in mature rats. Bone. 2004;34: 940-8.
72. , Ke HZ, Qi H, Healy DR, Li Y, Crawford DT, Paralkar VM, Owen TA, Cameron KO, Lefker BA, Brown TA, Thompson DD. A novel, non-prostanoid EP2 receptor-selective prostaglandin E2 agonist stimulates local bone formation and enhances fracture healing. J Bone Miner Res. 2003;18: 2033-42.
73. , Lowry KJ, Anglen JO. Indomethacin compared with localized irradiation for the prevention of heterotopic ossification following surgical treatment of acetabular fractures. J Bone Joint Surg Am. 2001;83: 1783-8. Erratum in: J Bone Joint Surg Am. 2002;84:100.
74. , Goss K, Anglen JO. Indomethacin versus radiation therapy for prophylaxis against heterotopic ossification in acetabular fractures: a randomised, prospective study. J Bone Joint Surg Br. 1998;80: 259-63.
75. . Ketorolac prophylaxis against heterotopic ossification after hip replacement. Clin Orthop Relat Res. 1995;314: 162-5.
76. , Wilbek H, Soelberg M. Naproxen for 8 days can prevent heterotopic ossification after hip arthroplasty. Clin Orthop Relat Res. 1995;314: 166-9.
77. , Letournel E. Low-dose irradiation and indomethacin prevent heterotopic ossification after acetabular fracture repair. J Bone Joint Surg Br. 1994;76: 895-900.
78. . Prophylaxis with indomethacin for heterotopic bone. After open reduction of fractures of the acetabulum. J Bone Joint Surg Am. 1990;72: 245-7.
79. , Skoglund B, Aspenberg P. Parecoxib impairs early tendon repair but improves later remodeling. Am J Sports Med. 2004;32: 1743-7.
80. , Mittlmeier T, Bordel R, Schaser KD, Gradl G, Vollmar B. Selective cyclooxygenase-2 inhibition reverses microcirculatory and inflammatory sequelae of closed soft-tissue trauma in an animal model. J Bone Joint Surg Am. 2005;87: 153-60.
81. , Zhang Y, Hood WF, Walker MC, Trigg JS, Maziasz TJ, Koboldt CM, Muhammad JL, Zweifel BS, Masferrer JL, Isakson PC, Seibert K. Valdecoxib: assessment of cyclooxygenase-2 potency and selectivity. J Pharmacol Exp Ther. 2005;312: 1206-12.
Copyright 2007 by The Journal of Bone and Joint Surgery, Incorporated