Effects of Perioperative Antiinflammatory and Immunomodulating Therapy on Surgical Wound Healing

Anthony J. Busti, PharmD; Justin S. Hooper, PharmD; Christopher J. Amaya, PharmD; Salahuddin Kazi, MBBS


Pharmacotherapy. 2005;25(11):1566-1591. 

In This Article

Drugs Used in Inflammatory Conditions and Their Effects on Wound Healing

Nonselective NSAIDs are reversible inhibitors of both COX-1 and COX-2, with each NSAID inhibiting the COX enzymes at varying degrees.[25] Inhibition of both COX-1 and COX-2 results in decreased production of eicosanoids such as prostaglandins and leukotrienes. Deficiency of prostaglandins results in decreased permeability of endothelial cells[26] and inhibition of hyaluronic acid production, which is needed in the proliferative phase of healing.[27,28] These drugs also inhibit the production of thromboxane A2, which decreases platelet aggregation, thus predisposing the patient to hematoma formation and persistent bleeding. Animal studies have shown that the interruption in the balance between prostaglandin and thromboxane A2 can impair angiogenesis and wound healing.[29,30]

The influence of NSAIDs on various aspects of soft tissue wound healing have been evaluated ( Table 1 ).[31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55] In a study performed in 53 rats, tensile strength of a wound made through the skin and subcutaneous muscles was measured using a tensiometer.[31] Thirty-two of the rats were divided into 3 treatment groups: 12 rats received aspirin 75 mg/kg/day and 13 rats received aspirin 150 mg/kg/day starting 1 day before wounding; 7 rats received 150 mg/kg/day starting 2 days after wounding. In comparison with untreated controls, the rats that received 150 mg/kg starting 1 day before wounding had a 40% decrease in dermal tensile strength at 7 days after incision (438 B1 17.7 vs 262 B1 8.7 g, p<0.001). In the group of rats that received aspirin 75 mg/kg and 150 mg/kg before and after incision, respectively, the tensile strength was approximately 22% less than that observed among controls. The authors concluded that aspirin retards the process of wound healing. However, the doses of aspirin used in this study are not likely to be used clinically in human subjects.

Another group evaluated the tensile and breaking strength as well as collagen content after ketorolac treatment of wounds.[32] Twenty rats were assigned to either a control or treatment group. The rats were treated with ketorolac 4 mg/kg/day, as this dosage provides potent antiinflammatory activity. Treatment was begun 7 days before wounding and continued for a total of 3 weeks. Two weeks after wounding, rats were sacrificed, tissue samples were excised, and breaking strength was measured with a tensiometer. A significant decrease in mean breaking strength and mean collagen concentration was found in ketorolac-treated rats. Mean tensile strength was lower in treated rats compared with controls, but this difference was not statistically significant.

In another study, the effect of indomethacin on collagen synthesis using [3]H-hydroxyproline was evaluated.[33] Rabbits were given either placebo or indomethacin 10 mg/kg once/day. The rabbits were treated for 4 weeks and then sacrificed. Tendons were excised, minced, and analyzed for 4-hydroxyproline and proline content. The authors concluded that indomethacin significantly decreased collagen synthesis in healing tendons.

Wound strength was studied in vivo in three groups of rats receiving either ibuprofen 200 mg/kg/day or diclofenac 25 mg/kg/day (both of which are approximately 10 times the human clinical dose), or placebo.[34] A 3-cm ventral abdominal incision was made and sutured in all rats. Abdominal strips were obtained from the wound and tested using a tensiometer. At several time points after wounding, tensile strength was found to be 16-36% less in rats that received an NSAID. Histologic signs of inflammation were reduced in all treatment groups. The investigators concluded that their findings suggest antiinflammatory drugs should not be used in humans after surgery without sufficient reason.

Mortality was assessed in 45 rats assigned to receive ibuprofen 5 mg/kg for either 7 or 14 days, or placebo after burn exposure, which included exposing the rats to 95°C water for 10 seconds in order to scald 30% of the total body surface area.[35] Ten days after burn exposure, a celiotomy was performed with the cecum punctured twice with an 19-gauge needle. Slightly more than 73% in the placebo group survived, whereas survival was reported to be 40% and 26.7% in the short-term (7 days of ibuprofen use) and long-term (14 days of ibuprofen use) treatment groups, respectively. Immunologic assays of sponges implanted on the nonburned side of the rats also revealed treatment groups had decreased neutrophil migration, lymphocyte blastogenesis, and helper or inducer T lymphocyte infiltration on the implant. Results from this study indicate a clear correlation between NSAID use and decreased immune response and increased mortality in rats after trauma.

The effect of NSAIDs on human tendon cell proliferation also has been evaluated. One group studied the effect of NSAIDs (diclofenac, aceclofenac, indomethacin, naproxen) in vitro on tendon cell proliferation and production of proteoglycans in 28 tendons (14 digital flexor tendons, 14 patella tendons).[36] The concentrations of NSAIDs used were similar to those that would be achieved in vivo with doses normally used in clinical practice. The concentration of prostaglandins was also evaluated by radioimmunoassay utilizing dextracoated charcoal to separate bound from free 3H-prostaglandins. Cell proliferation and glycosaminoglycan synthesis in patellar tendon cultures were significantly inhibited by naproxen and indomethacin when compared with media controls. Conversely, no significant difference was found on tendon cell proliferation and glycosaminoglycan synthesis with the use of diclofenac and aceclofenac. The authors concluded that the use of some NSAIDs may inhibit tendon repair after injury. These findings may be due to differences in COX 2:COX 1 specificity, as diclofenac has a greater selectivity for COX-2 inhibition when compared with the other NSAIDs in the study.[25]

The safety of NSAID use was examined in a retrospective analysis of patients with osteoarthritis (mean age 60 yrs) who underwent surgery for primary total hip arthroplasty.[37] Patients were grouped as either taking NSAIDs before surgery (but not within 24 hrs) or not taking NSAIDs. The NSAIDs administered during this study were indomethacin, sulindac, tolmetin, ibuprofen, ketoprofen, naproxen, fenoprofen, meclofenamate, aspirin, and piroxicam. Patients were compared based on intraoperative fluid administration requirements, estimated blood loss, number of transfusions needed, postoperative complications, and length of hospital stay. The authors found that 12% of patients taking NSAIDs had hypotensive or minor bleeding episodes postoperatively, whereas only 2% of patients not taking NSAIDs experienced either of these complications. Both groups had the same length of hospital stay (10 days), the same number of transfusions (2.2 units), and similar postoperative wound drainage (NSAID 483 ml vs no NSAID 548 ml, p=NS). The increased bleeding seen in patients who received NSAIDs may have been due to the inhibition of thromboxane A2. The authors concluded that NSAIDs should be discontinued for a sufficient amount of time to allow for total drug elimination before surgery.

All studies reported are very small, and caution must be exercised when applying these data in the perioperative setting. Clinically significant differences may exist in soft tissue and tendon cell healing when nonselective NSAIDs are given during the postoperative period. We recommend withholding nonselective NSAIDs for at least 3-4 half-lives before surgery involving tendons and soft tissue to ensure removal from the systemic circulation.

Cyclooxygenase-2 inhibitors such as celecoxib, rofecoxib, valdecoxib, and parecoxib are primarily reversible inhibitors of COX-2, yet retain some inhibition of COX-1. Parecoxib, a parenteral COX-2 inhibitor undergoing investigation, will likely have utility in the postoperative setting because of its opioid-sparing effects and its ability to be administered parenterally.[38,39] It will likely be used in the postoperative patient for pain management because of its positive contribution to recovery from perioperative anesthesia.

The COX-2 inhibitors differ from nonselective NSAIDs in their COX-2:COX-1 inhibition ratios.[56] One group used human whole blood and modified whole blood assays to investigate the relative inhibitory potencies of several nonselective NSAIDs and COX-2 inhibitors.[25] Doses used would produce an 80% inhibition of COX-2, which correlates with doses that are commonly used in the clinical setting. Results showed that compounds traditionally associated with the greatest gastrointestinal toxicity had the highest selectivity for COX-1.

Since COX-2 is primarily produced in areas with tissue inflammation and influences the production of prostaglandins and other proinflammatory mediators, inhibiting COX-2 may lead to alterations in the normal processes involved in the first phase of wound healing. Because of previous data showing delayed healing of gastric ulcers after NSAID administration,[57,58,59] another group[40] studied the effects of NSAID and COX-2 inhibitor administration in rats with induced colitis. In this study, rats were given oral diclofenac 10 mg/kg, naproxen 5 mg/kg, nabumetone 25 or 75 mg/kg, etodolac 10 or 50 mg/kg, L745,337 (a highly selective COX-2 inhibitor) 1 or 5 mg/kg, or a placebo vehicle of 0.5% carboxymethylcellulose. Treatment began 3 hours before induction of colitis and continued every 12 hours for up to 7 days. Colonic prostaglandin synthesis, COX-2 messenger-RNA expression, and COX-1 and COX-2 immuno-histochemistry analyses were performed. Investigators found that administration of the selective COX-2 inhibitor L745,337 5 mg/kg resulted in a 53% reduction in colonic prostaglandin synthesis, and the mortality rate in this subpopulation was almost 100% at 14 days. The cause of death was almost invariably perforation of the distal colon. These findings led the investigators to conclude that moderate to highly selective COX-2 inhibitors have detrimental effects in the setting of chronic inflammation. Although this study was conducted in animals, it presented an important finding that may be applied in other wound-healing processes. However, the effects of prostaglandins in wound healing at other anatomic sites need further research.

Recent studies investigating the effects of COX-2 inhibition on soft tissue wound healing revealed mixed results.[41,42,43] In one study, the investigators induced gastric ulceration with acetic acid in mice and analyzed the damaged tissue for ulcer diameter.[41] Mice were given a selective COX-2 inhibitor, N-methane sulfonamide (NS-398) 1 g/100 g of body weight intraperitoneally daily, at various time points after ulcer development. The investigators found that administration of NS-398 significantly increased ulcer indexes at days 10, 20, and 30 (day 10: before 3.6 B1 0.3 mm, after 4.7 B1 0.7 mm, p<0.05; day 20: before 0.8 B1 0.5 mm, after 2.4 B1 0.4 mm, p<0.05; and day 30: before 0 mm, after 0.9 B1 0.05 mm, p<0.05) when begun on the day of ulcer development. Investigators found a significant accumulation of COX-2 mRNA, and COX-2 protein expression detected by anti-COX-2 antibody. These data indicate that COX-2 inhibition impairs the gastric healing process, and COX-2 may be an important component needed in the wound healing process.

However, other studies involving currently available COX-2 inhibitors failed to replicate these findings. One group compared the effects of a more selective COX-2 inhibitor diclofenac with nonselective COX inhibition of dexamethasone on the effects of wound healing in mice.[42] Each drug was given every 12 hours orally by gavage until mice were sacrificed. A tensiometer was used to evaluate wound strength on days 7, 14, 21, and 28 after wounding. Diclofenac did not have any significant effect on the morphology, whereas dexamethasone significantly affected macroscopic skin morphology and healing. Atrophy of the epidermis, dermis, and epidermal appendages were seen in a majority of mice at all time points in the dexamethasone group. Dexamethasone also led to a prolonged expression of mouse keratin 6 at day 14, indicating a delay in reepithelialization, compared with the diclofenac and control groups at day 7. Tensile strength was not affected with the use of selective or nonselective COX inhibition. However, dexamethasone significantly decreased wound strength at days 14, 21, and 28 (p<0.05).

The effects of naproxen 10 mg/kg, celecoxib 10 mg/kg, and a nitric oxide-releasing NSAID (HCT-3012) 14.5 mg/kg on wound healing in 11 rats were studied by another group.[43] Wounds were induced by creating subcutaneous pockets on both sides of a 3-cm incision and implanting polyvinyl sponge disks, which was followed by suturing. The naproxen dose was comparable to adult doses; however, the celecoxib dose was much higher than doses used clinically. The levels of prostaglandins were analyzed by measuring the exudates collected from the wound with use of an enzyme-linked immuno-sorbent assay. When giving a dose sufficient to produce suppression of prostaglandin synthesis, celecoxib did not significantly affect collagen deposition. This suggests that COX-1 plays a bigger role in prostaglandin synthesis at the wound site, and the use of celecoxib may not adversely affect wound healing postoperatively.

Although the implications of COX-2 inhibitors on wound healing have not been specifically investigated in humans, their usage has been analyzed in various perioperative settings. In a recent study, the effects of pain relief with rofecoxib 25 mg/day were compared with placebo after elective total-knee arthroplasty.[44] All NSAIDs were discontinued 10 days before surgery, and patients were randomly assigned to receive placebo or rofecoxib 25 mg to begin 3 days before surgery and continue until 48 hours after surgery. Demographic variables, as well as preoperative visual analog scale pain scores, were similar between the two groups. Although this trial was not designed to assess the safety of rofecoxib in the perioperative period, the authors concluded that the administration of rofecoxib caused no significant increase in perioperative bleeding or international normalized ratio, and they suggested that patients do not need to stop taking rofecoxib before surgery.

To our knowledge, no large human clinical trials have investigated primary outcomes regarding selective COX-2 inhibitors and their effect on soft tissue wound healing. Most of the current data on soft tissue wound healing involves animal studies, which does not reveal any trend toward impairing wound healing when begun before or after surgery. The safety and influence of COX-2 inhibitors on wound healing in postoperative patients remain unknown.

The use of nonselective NSAIDs during procedures involving bone remodeling also has been studied. A study involving the use of indomethacin during healing of nonimmobilized femur fractures was performed in 129 rats.[45] Two matched groups were given indomethacin 2 mg/kg/day orally or placebo immediately after fracture induction and continued until the animal was sacrificed. Investigators noted only small remnants of hematomas and well-developed callus formation in the placebo group after 6 days. However, the treatment group had larger hematoma formation and less-stable fractures 12 days after trauma. At 24 days, the placebo group had significantly less ventral angulation, indicating a better fracture repair than that of the treatment group (25° vs 40°, p<0.0002). Tensile strength and elastic stiffness were significantly higher in the placebo group than in the treatment group throughout the 24-day study. The investigators concluded that indomethacin seriously impairs the normal fracture healing process.

A study involving indomethacin's effects on bone remodeling after osteotomy in rabbits was performed to determine if similar effects were seen in nontraumatic conditions.[46] Thirty-two rabbits had a tibial osteotomy on the hind limb. Seventeen rabbits were given indomethacin 10 mg/kg/day orally 4 days before surgery and continued until sacrificed. Fifteen rabbits served as controls throughout the 6-week study. After 2 and 6 weeks, the number of resorptive and formative foci were reduced in the indomethacin group (46.8 vs 85.6, p<0.05, and 130 vs 236, p<0.05, respectively), but the rate of bone formation did not differ between the groups. Investigators concluded that the inhibition of prostaglandins likely led to decreased stimulation of bone formation.

The effect of prostaglandin synthesis inhibition on bone repair of demineralized bone matrix grafts in 27 rabbits was investigated by using ketorolac and methylprednisolone.[47] Bone particles from sacrificed rabbits were demineralized with chloroform and methanol and implanted into bone of experimental rabbits. The rabbits were divided into the following groups: control, demineralized bone plus ketorolac 2 mg/kg, demineralized bone plus ketorolac 4 mg/kg, and demineralized bone plus methylprednisolone 1 mg/kg. Drugs were given intramuscularly throughout the 6-week period. Torsional stiffness of intact ulnae was significantly lower in the treatment groups compared with controls, and larger differences were seen with higher doses of ketorolac. Methylprednisolone and higher doses of ketorolac were also found to decrease the amount of well-differentiated osteocytes within the demineralized bone.

A common clinical dosage of ibuprofen (16 mg/kg/day) was given to rats in order to study its effect on the chemical composition of fractured and nonfractured tibias.[48] Rats were divided into four groups: fractured tibia plus placebo or ibuprofen, and unfractured tibia plus placebo or ibuprofen. Drug administration began 1 week before fracture and continued throughout the study for up to 9 weeks after injury. Compared with placebo, the ibuprofen group had significantly lower calcium levels (p<0.05). Hydroxyproline content also diminished significantly in the treatment group of fractured tibias (p<0.001), but not significantly different from the unfractured treatment group.

The inhibitory effects of NSAIDs on bone remodeling have been well documented in several animal models. However, many patients rely on these drugs for control of arthritic conditions, and the time in which the NSAIDs could be resumed postoperatively remained unanswered. One group studied 70 rabbits that were given indomethacin 10 mg/kg/day after an L5-L6 bilateral intertransverse process posterior arthrodesis.[49] Group 1 began indomethacin 2 weeks postoperatively and continued for 6 weeks, group 2 began treatment 4 weeks postoperatively, and group 3 served as the control. At 6 weeks, the fusion rate was 21% (5/24) in group 1, 48% (11/23) in group 2, and 65% (15/23) in the control group. Both treatment groups had significantly lower fusion rates compared with controls (p<0.02 and p=0.05, respectively). One limitation to the study is the assessment of fusion by manual palpitation rather than by using biomechanical or radiographic studies. However, investigators concluded that NSAID therapy should be delayed as long as possible after a spinal arthrodesis.

In another study, the medical records of 288 patients were examined retrospectively for at least 2 years after posterior spinal fusion from the L4 vertebra to the sacrum.[50] Patients in the treatment group were administered ketorolac 60 mg intramuscularly for the first dose, followed by 30-mg doses every 6-8 hours as needed for postoperative pain. Seven patients required 2-20-mg doses orally every 6 hours as needed for pain. One hundred twenty-one patients in the study received no NSAID postoperatively. Both groups had similar baseline demographic characteristics except body weight was higher in the ketorolac group (182 vs 173 lbs, p<0.05). Nonunion was significantly higher in the ketorolac group (4% vs 17%, p<0.001) and also led to a 5 times greater chance of developing pseudoarthritis (95% CI 1.8-16.6). The investigators concluded that ketorolac significantly inhibits spinal fusion when used for postoperative pain management.

These studies all indicate that nonselective NSAIDs have inhibitory effects on the bone healing process. Clinical doses of nonselective NSAIDs reduce calcium levels, lower hydroxyproline levels, cause increased rates of nonunion in fractures, and inhibit osteoblast formation. Practitioners should withhold the use of these drugs for 3-4 half-lives preoperatively to ensure systemic clearance in the patient undergoing orthopedic surgery. These drugs should also be withheld for as long as clinically possible in postoperative orthopedic patients since studies indicate the inhibitory effects could last up to 6 weeks after surgery.

Recent evidence from animal data indicate that certain prostaglandins, and thus COX-2 activity, are required for normal fracture healing, normal endochondral ossification, and maximal induction of osteogenesis.[51,60] Thus, COX-2 inhibitors could adversely influence and delay the healing process in orthopedic-related injuries or surgeries.

The use of COX-2 inhibitors in animal ligament and osteopathic injuries has not revealed favorable results. The effects on medial collateral ligament healing were tested in 50 rats with surgically transected medial collateral ligaments.[52] Ligament strength was measured by using a servohydraulic testing machine and loaded to failure. Celecoxib was given orally at 14-25 mg/kg/day, which is similar to human dosages of 5-30 mg/kg/day. The authors found a 32% decrease in strength in the injured ligaments of the rats receiving celecoxib compared with the injured ligaments of the controls (p=0.004). They suggested that celecoxib be used cautiously for the treatment of ligament injuries.

Fracture healing also appears to be affected by COX-2 inhibitors. A study was conducted in 253 rats who were given oral indomethacin, celecoxib, rofecoxib—at doses similar to those in clinical practice—or placebo 2 days before fracture.[51] Bone strength was measured by using a servohydraulic testing machine with a 20-N reaction torque cell. Indomethacin was found to delay bridging of the fracture gap by 1 week compared with celecoxib, rofecoxib, and placebo, but did not prevent healing. Celecoxib- and rofecoxib-treated rats still had fracture evidence at week 8, whereas normal healing had occurred in all the rats given placebo, indicating the importance of COX-2 in fracture healing. Rofecoxib was also found to lead to a significantly reduced normalized peak torque and torsional rigidity at 4 and 8 weeks after fracture compared with placebo (18,702 B1 5772 vs 42,229 B1 2640, p<0.05). Normalized shear stress was significantly reduced in the indomethacin and rofecoxib groups at 4 and 6 weeks after fracture and remained significantly lower in the rofecoxib group at 8 weeks (16% vs 44%, p<0.05). The authors concluded that selective COX-2 inhibition of prostaglandin synthesis at doses similar to those of arthritis treatment may retard the fracture healing process.

Another group also demonstrated the negative effects of selective COX-2 inhibition on fracture healing.[53] Investigators compared the effects of etodolac 20 mg/kg/day, the maximum recommended daily dose in humans, with placebo on fracture healing in eight rat femurs. Bone strength was measured by using a mechanical testing system that loaded the bone until failure at a speed of 10 mm/minute. Etodolac was found to delay fracture union and callus formation in rats 3 weeks after fracture (p<0.05). Strength and stiffness were also significantly lower in the etodolac treatment group than in the placebo group (41.2 B1 6.9 vs 95.1 B1 13.7 N/mm, p<0.05, and 30.4 B1 3.9 vs 118.6 B1 22.5 N/mm, p<0.05, respectively).

In another study, the effect of ketorolac 4 mg/kg, parecoxib 0.3 mg/kg, and parecoxib 1.5 mg/kg daily beginning 24 hours after injury on fracture healing was compared with nontreated controls in 48 rat femurs.[54] Three weeks after femur fracture, fractures failed to unite radiographically in 25% of the ketorolac group and 8% of the high-dose parecoxib group. Shear modulus at 21 days after fracture was significantly lower in the ketorolac and the low-dose parecoxib groups (p<0.05). However, only the ketorolac group had a lower mean ultimate strength compared with controls (p<0.05). Mean ultimate strength was also significantly lower at 35 days after fracture in the ketorolac group and low-dose parecoxib group compared with controls and high-dose parecoxib group (p<0.05).

Another group also found that rats had impaired fracture healing with the administration of COX-2 inhibitors.[55] Rats treated with indomethacin 1 mg/kg/day and celecoxib 3 mg/kg/day were compared with nontreated rats for fracture healing times (57 rats). At 4 and 8 weeks, both treatment groups had significantly more fibrous tissue and less woven bone formation compared with the nontreatment group under microscopic tissue examination. At 4 weeks, both treatment groups had decreased strength and stiffness; however, the difference between the celecoxib group and nontreatment group was not statistically significant.

Literature regarding the effects of selective COX-2 inhibitors on bone healing cite only small animal model studies. However, the doses used in these animal studies are similar to those used in humans clinically. In humans, these agents could impair and delay the normal physiologic bone healing process. Clinicians should consider the risk associated with this class of drugs concerning proper healing during the postoperative period since the use of selective COX-2 inhibitors can cause a delay in fracture union. This may also lead to weaker unions up to 8 weeks after incision. Until further studies determine the safety of these products on a larger scale, this class of drugs should be held at least 3-4 half-lives or 2-3 days before orthopedic surgery and resumed only after complete bone healing has occurred.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as: