Cyclooxygenase-2 Induction after Oral Surgery Does Not Entirely Account for Analgesia after Selective Blockade of Cyclooxygenase 2 in the Preoperative Period

We aimed to recruit a total of at least 60 patients among new referrals to the oral surgery clinic of Pisa University Hospital, Pisa, Italy, from January to July 2004. Men and women aged at least 18 yr and scheduled to undergo removal of one lower impacted third molar were enrolled. A thorough medical history and radiographic examination were performed to confirm the need for tooth extraction and verify the absence of inflammation or infection at the extraction site. Patients had no history of drug allergy, and none of them had taken drugs within 2 weeks before the study. Patients were also excluded if they had infectious diseases, history of asthma, peptic ulceration, or other disorders of the upper gastrointestinal tract. Patients who had any concomitant major medical problem, known allergy or intolerance to study medications, or any ongoing pharmacologic treatment were also excluded, as were women who were pregnant or lactating or not using contraceptives. Informed consent was obtained from each patient before entering the trial, and the investigation was approved by the local University Hospital Ethics Committee.

A double-blind, randomized design was used. After screening, patients were randomly assigned to receive one of the following treatments as a single dose by oral route 1 h before surgery: placebo, 50 mg rofecoxib (Dolostop®, Gentili, Pisa, Italy), or 550 mg naproxen (Synflex Forte®, Recordati, Milano, Italy). The random sequence was generated by computer. The subjects chose a sequentially numbered, opaque, sealed envelope, which enclosed the code for the treatment protocol they were to receive. The number of envelopes was equal to the number of subjects. Treatment groups were coded so that the operator, the examiner, and the patients remained blinded throughout the study.

Surgical extractions of one lower third molar per patient were performed by an experienced oral surgeon by a standardized technique and local anesthesia (3% lidocaine for inferior alveolar nerve block and 2% lidocaine with 1:100,000 epinephrine in the surgical area). Briefly, a mucoperiosteal flap was raised and retracted, bone tissue was removed with a bur, and the tooth was elevated. A specimen of mucosal tissue was then dissected from the surgical site, immediately snap-frozen in liquid nitrogen, and stored at −80°C for subsequent reverse-transcription polymerase chain reaction (RT-PCR) analysis of mRNAs coding for cyclooxygenase isoforms. After the extraction, a microdialysis probe (CMA/70 Microdialysis Bolt Catheter; CMA/Microdialysis, Solna, Sweden) was placed beneath the mucoperiosteal flap into the surgical site. The probe fiber consisted of a 10-mm, flexible, nonmetallic, semipermeable dialysis membrane with a molecular cutoff of approximatively 20,000 Da. Silk sutures were then used to secure the probe and to suture the surgical flap. Subsequently, a saline solution (Perfusion Fluid T1; CMA/Microdialysis) was pumped at a rate of 5 μl/min. The solution had the following composition: 147 mm Na⁺, 4 mm K⁺, 2.3 mm Ca²⁺, and 156 mm Cl⁻. The dialytic perfusate collected during the initial 60 min from probe implantation was discarded, and dialytic samples were then collected in vials placed on ice. Vials were changed every 30 min for the subsequent 180 min, and the collected samples were then stored at −80°C until PGE²assay.

Concomitantly with vial changes, pain intensity was assessed every 30 min by means of questionnaires, completed by every patient, using a categorical scale of none (0), mild (1), moderate (2), or severe (3) pain as well as a 100-mm visual analog scale (VAS). Patients were also requested to provide information about inferior alveolar nerve conductivity by rating feelings in their lips as “numb,”“tingling,” or “normal.” If patients rated their pain as 40 or greater, based on the VAS scale, and requested additional analgesic medication, they were allowed to take codeine at the oral dose of 30–60 mg. At the end of the observation period, during local anesthesia, the restraining sutures were cut, and the probe was removed. At this time, an additional specimen of oral mucosa was dissected from the surgical area and stored at −80°C for subsequent RT-PCR assay.

Levels of PGE²in samples of dialytic perfusate were determined by competitive enzyme-linked immunoassay (Cayman, Ann Arbor, MI) in accordance with manufacturer instructions. Each sample was assayed in duplicate. PGE²levels were then expressed as concentrations at various time points (pg/ml).

Specimens of oral mucosa were disrupted with cold glass pestles, and total RNA was isolated by Trizol® (Life Technologies, Carlsbad, CA) and chloroform. RT-PCR was performed by specific primers based on cloned COX-1 and COX-2 human genes15and carried out as previously reported.16PCR consisted of 30 cycles of denaturation at 94°C (1 min), annealing at 50°C (1.5 min), extension at 72°C (2 min), and final extension at 72°C for 10 min. Amplified products were separated by 1.5% agarose gel electrophoresis and stained with ethidium bromide. Complementary DNA bands were visualized by ultraviolet light, quantitated by densitometric analysis with NIH Image program (Scion Corporation, Frederick, MD), and normalized to β-actin.

Results are given as mean ± SD. The significance of differences was evaluated by one-way analysis of variance for unpaired data followed by post hoc  analysis with Bonferroni test (pain intensity and PGE²assay), or the Student t  test for paired data (RT-PCR analysis). P  values lower than 0.05 were considered significant.

A total of 63 patients (28 men and 35 women; age ranging from 20 to 32 yr; mean age, 24.5 ± 4.7 yr) were included in the study. However, 60 patients completed the trial, because 3 patients were withdrawn because of probe failure. Therefore, the study groups consisted of 20 patients each (placebo: 8 men and 12 women; mean age, 25.3 ± 5.1 yr; naproxen: 11 men and 9 women; mean age, 24.7 ± 4.1 yr; rofecoxib: 7 men and 13 women; mean age, 23.8 ± 5.3 yr).

No adverse events were recorded in any treatment group throughout the study period. No alterations of the inferior alveolar nerve conductivity were noted, because all patients returned to their normal lip sensitivity within 180 min from the intervention.

Pain intensity in the placebo group, as assessed by VAS scale, increased progressively throughout the postoperative period, reaching a peak level of 44 ± 5.9 at 180 min from surgery (fig. 1A). Pain intensity values recorded for patients treated with rofecoxib or naproxen were significantly lower than those included in the placebo group, whereas no significant differences were observed when comparing the rofecoxib group with the naproxen group (fig. 1A). Reports of pain intensity evaluated by means of categorical rating scale yielded data similar to those obtained with the VAS scale. Figure 1Bdisplays pain intensity values recorded with the categorical scale at three different times of the study period (120, 180, and 240 min). Three patients in the placebo group required additional analgesic medication because they approached a VAS value of 40 at 180 min.

The collection of dialytic samples, 60 min after probe implantation into the surgical site, allowed measurement of a PGE²concentration of 426.8 ± 61.1 pg/ml at 90 min in patients treated with placebo. Such a value increased progressively throughout the postoperative period, reaching a level of 631.5 ± 36.2 pg/ml at 240 min (fig. 2). PGE²concentrations in the naproxen group were significantly lower than those recorded in placebo patients at all time points examined. After naproxen administration, PGE²levels accounted for 113.8 ± 41.1 pg/ml at 90 min, and this value did not change significantly throughout the evaluation period (fig. 2). PGE²concentrations in dialytic samples collected from patients of rofecoxib group displayed a different trend in comparison with naproxen group. After rofecoxib administration, PGE²levels were 370.6 ± 27.5 pg/ml at 90 min of the postoperative period, and this value declined progressively up to the end of the experimental period. When compared with placebo, PGE²values in rofecoxib-treated patients were significantly lower from 150 min onward, whereas significant differences between naproxen and rofecoxib groups were noted at both 90 and 120 min (fig. 2).

The mRNA expression of cyclooxygenase isoforms was assessed by RT-PCR analysis of gingival specimens obtained from patients immediately before probe implantation (T = 0) as well as at the end of dialytic perfusion and probe removal (T = 240 min). When examining tissue samples retrieved from placebo-treated patients at T = 0, RT-PCR revealed the presence of COX-1 mRNA, whereas COX-2 expression was undetectable (or very faint in few cases). At the end of the observation period, COX-1 mRNA expression was barely unchanged, according to semiquantitative densitometric analysis, whereas a significant induction of gene expression could be detected for COX-2 mRNA (fig. 3). Similar expression patterns were found for COX-1 and COX-2 mRNA, at both T = 0 and T = 240, when performing RT-PCR analysis on gingival samples obtained from patients subjected to treatment with rofecoxib or naproxen (data not shown).

A number of studies suggest that preoperative treatments with coxibs ensure good postoperative analgesia.10,12,17However, the anatomical locations and molecular mechanisms underlying coxib-induced preoperative analgesia remain undetermined. In the current investigation, patients undergoing third molar extraction were preoperatively treated with a selective COX-2 inhibitor or a conventional NSAID and, by a combined assessment of pain intensity with local PGE²production and mRNA COX isoform expression in the postoperative period, consistent evidence was obtained that both drugs exerted significant analgesic effects while displaying different inhibitory profiles on prostaglandin release.

The evaluation of postsurgical pain in patients treated preoperatively with placebo showed that pain intensity increased progressively after surgery and that maximal levels could be recorded by 180 min. Concomitantly, the analysis of dialytic perfusate revealed an active release of PGE²at the surgical site by 90 min after tooth extraction, with progressive increments in PGE²concentrations throughout the observation period. The mRNA coding for cyclooxygenase isoforms was also measured by RT-PCR in tissue samples and, at variance with the stable expression of COX-1, COX-2 mRNA was undetectable in perialveolar gingiva at the time of tooth removal, and a significant increment of its expression could be observed 4 h later. Considering that the increase in COX-2 mRNA could be due to its induction in resident cells or to recruitment of COX-2 expressing cells triggered by surgical injury, our RT-PCR results support the view that COX-2 induction in the operative site may contribute to the increase in PGE²production at a late phase of postsurgical period. Of note, the above lines of evidence, in keeping with previous reports based on similar methodologies,8,18might be taken to suggest that postoperative pain is driven mainly by an enhancement of cyclooxygenase activity and prostaglandin production in oral tissues subjected to surgical injury. However, our data from patients treated with cyclooxygenase inhibitors argue against this hypothesis.

A major finding of the current study was that rofecoxib and naproxen were equieffective against postoperative pain, while acting with different time patterns to modulate PGE²release at the surgical site. In particular, naproxen exerted a significant analgesic action that was closely paralleled by a marked inhibition of tissue PGE²production, whereas rofecoxib ensured a full prevention of postoperative pain that was associated with a significant reduction of local PGE²release only in a late phase of the observation period. In the light of data provided by RT-PCR analysis of COX-1 and COX-2 mRNA expression, the results obtained with naproxen and rofecoxib suggest that (1) COX-1 is mainly responsible for PGE²production at the surgical site in the early postoperative phase, (2) a local induction of COX-2 can account for the enhanced PGE²release in the late observation period, and (3) the analgesic effect observed after preoperative rofecoxib administration is likely to result from a pharmacologic modulation of COX-2 constitutively active at distinct locations from the surgical area. Most importantly, our findings in the setting of drug treatment do not support the hypothesis that local PGE²production plays a significant role in driving postoperative pain, as one might infer by measurements performed in the placebo group.

Cyclooxygenase inhibitors are thought to act both on peripheral tissues and central nervous pathways to counteract pain sensitization.3,19It is also acknowledged that, after acute noxious stimuli, the production of prostanoids responsible for peripheral hyperalgesia requires COX-2 induction.2However, whereas this mechanism has been assumed as the main rationale for clinical development of coxibs, it can hardly account for the analgesic effects of selective COX-2 inhibitors in the early stage of postoperative period, because few hours are required to achieve maximal levels of COX-2 induction.20In this respect, our results emphasize from a clinical perspective the importance of a constitutive COX-2 isoform in preemptive analgesia achievable with cyclooxygenase inhibitors.

Studies on preclinical models have shown that cyclooxygenase isoforms are constitutively expressed in dorsal horns of spinal cord and that central hyperalgesic responses to peripheral stimuli are highly sensitive to COX-2 blockade, indicating that spinal nerves implicated in pain transmission are under the control of COX-2–derived prostaglandins.21Other authors have suggested that spinal COX-1 might be important also in sensitization to postoperative pain.22,23In addition, spinal COX-2 expression can be further enhanced by peripheral inflammation to subserve a late phase of central hyperalgesia, which requires some hours to develop fully.20Therefore, current preclinical evidence supports the notion that the primary site of NSAID action is in the spinal cord, where acute stimuli can activate PGE²production by constitutive COX-2, to initiate facilitated release of primary afferent nociceptive transmitters and block intrinsic dorsal horn inhibition, and where acutely induced hyperalgesia can be blocked by systemic or spinally administered COX-2 inhibitors.24–26Consistent with these preclinical findings, our data, showing that preoperative rofecoxib administration ensured analgesia in the early stage of postsurgical period, without a concomitant reduction of local PGE²production, can be interpreted as resulting from the inhibition of a constitutive COX-2 responsible for the activation of a rapid central hyperalgesia elicited by peripheral surgical injury. In support of this proposal, it has been recently reported that, after oral administration to healthy volunteers, at a single dose of 50 mg, rofecoxib penetrates rapidly into the central nervous system and reaches cerebral spinal fluid concentrations sufficient to inhibit COX-2 activity.27 

In conclusion, the current study provides clinical evidence that the preoperative administration of a selective COX-2 inhibitor determines an effective control of postoperative oral pain. It is also suggested that the selective blockade of inducible COX-2 at the peripheral level does not entirely account for the analgesic action occurring in the postoperative period.