This study aimed to evaluate whether pre-emptive analgesia modifies the tissue expression of tumour necrosis factor alpha (TNF-α) and interleukin 1 beta (IL-1β), and whether there is an association with postoperative surgical outcomes. A triple-blind, randomized, placebo-controlled study of patients undergoing mandibular third molar removal was performed. Volunteers were allocated randomly to receive etoricoxib 120 mg, ibuprofen 400 mg, or placebo 1 h before surgery. Twenty-four surgical sites per group were required (95% confidence level and 80% statistical power). Pain scores differed significantly between groups ( P < 0.001). Etoricoxib and ibuprofen reduced pain scores compared to placebo ( P < 0.05). Pain scores peaked at 4 h postoperative in the experimental groups, but at 2 h postoperative in the placebo group ( P < 0.05). A significant reduction in TNF-α concentration from time 0′ to time 30′ was seen for ibuprofen ( P = 0.001) and etoricoxib ( P = 0.016). The ibuprofen group showed a significant reduction in IL-1β levels from time 0′ to time 30′ ( P = 0.038). In conclusion, TNF-α and IL-1β levels and the inflammatory events in third molar surgery were inversely associated with the degree of cyclooxygenase 2 selectivity of the non-steroidal anti-inflammatory drugs used pre-emptively. Patients given pre-emptive analgesia showed significant reductions in the clinical parameters pain, trismus, and oedema when compared to the placebo group.
The surgical removal of mandibular third molars is one of the most commonly performed dental procedures in the outpatient setting. The level of dental impaction may be explained by the lack of physical space for eruption. This is usually the result of the size of the maxillary bone being insufficient to properly accommodate all teeth present within the arch. Thus, third molar extraction is a procedure commonly related to local tissue injury, associated with varying degrees of postoperative pain .
From a clinical perspective, the removal of such teeth can affect the patient’s quality of life postoperatively, particularly during the first 3 days, due to the intensity of pain and inflammation arising from the surgical procedure. Approximately 40–60% of these patients experience moderate to severe pain, requiring the use of rescue analgesics . In this context, pre-emptive analgesia is used as a pharmacological strategy for the management, reduction, or even prevention of postoperative pain related to dental procedures. This strategy of pain control has been studied widely in recent decades .
In addition to pain, oedema and limited mouth opening (trismus) are the postoperative complications most commonly associated with the removal of mandibular third molars. These clinical signs and symptoms result from a local inflammatory process generated by the activation of the cyclooxygenase (COX) pathway and the subsequent increase in levels of prostaglandins within the injured site. Both of these play important roles in the release of proinflammatory cytokines (tumour necrosis factor alpha (TNF-α), interleukin 1 (IL-1), and interleukin 6 (IL-6)) related to the pathophysiology of pain and inflammation , .
Clinical and experimental studies evaluating the pathogenesis of inflammation resulting from third molar surgery, as well as the role of the inflammatory mediators implicated in the subsequent processes of pain and inflammation, are extremely relevant and currently warranted. Certain mechanisms have been proposed to explain the inflammation that occurs as a result of surgical procedures, but a complete understanding of how these events are fully triggered remains unclear. Proinflammatory cytokines such as TNF-α and interleukin 1 beta (IL-1β) have often been described as important mediators in this process . Experimental studies involving acute pain models suggest that IL-1β sensitizes nociceptors and causes hyperalgesia, therefore working actively in the pathophysiology of this type of pain . On the other hand, it is recognized that TNF-α exerts remarkable effects, including activating lymphocytes, stimulating the synthesis of other proinflammatory cytokines such as IL-1β and IL-6, and triggering the production of prostaglandins .
Although clinical studies have investigated the effectiveness of pre-emptive analgesia in third molar surgery , it is currently highly relevant to publish translational studies aimed at assessing the influence of the preoperative administration of non-steroidal anti-inflammatory drugs (NSAIDs) on the pathophysiology of inflammatory events established within these clinical situations. Another question that has yet to be answered is whether or not the selectivity of NSAIDs to COX isoforms may render different effects in the expression of proinflammatory cytokines such as TNF-α and IL-1β in these surgical situations.
Therefore, this study aimed to test the hypothesis that pre−emptive analgesia, through the administration of NSAIDs prior to the removal of mandibular third molars, can quantitatively change the tissue expression of TNF−α and IL−1β, and that these changes may be related to clinical effects (pain, oedema, and trismus) postoperatively.
Materials and methods
Study design and sample
This study was approved by the Ethics Committee of Walter Cantídio University Hospital and was performed in accordance with the Helsinki statements. A triple−blind, randomized, crossover, placebo−controlled study of patients undergoing mandibular third molar extractions was performed. These patients were recruited from the Division of Oral and Maxillofacial Surgery of Walter Cantídio University Hospital at the Federal University of Ceará (Brazil). The volunteers were recruited between March 2014 and December 2015 according to the CONSORT protocol .
The sample unit used in this study was the surgical site. Healthy individuals (American Society of Anesthesiologists, ASA 1) of both sexes, aged between 18 and 35 years, requiring the removal of both mandibular third molars, were invited to participate in this study.
The following inclusion criteria were adopted to standardize the level of traumatic injury generated by surgery: (1) patients with third molars requiring ostectomy, with or without associated tooth sectioning; (2) patients with third molars that showed similar patterns of root formation, position, and degree of impaction. Patients were excluded if they met any of the following criteria: smokers, pregnant or breast feeding, users of medications that could interact with the drugs used in this study, patients with orthodontic bands on the mandibular second molars, confirmed history of allergy to NSAIDs, signs of any preoperative inflammatory or infectious condition, systemic chronic disease, use of NSAIDs within the past 21 days, or the presence of periodontal disease, swelling, fever, or trismus prior to surgery. Patients who did not follow the recommendations prescribed, who underwent surgical procedures that exceeded 2 h, who had an intolerance to the pharmacological regimen, who presented a postoperative infection, and those who did not return for postoperative assessment consultations were removed from the study.
All recruited individuals were informed about the objectives and study design, and those who consented to participate signed a written informed consent agreement.
Sample size calculation
The sample size calculation was based on that described in the study conducted by Al-Shukun et al. . These authors observed a negative response to the pharmacological treatment instituted of 58% in the control group and 18% in the experimental groups. Thus, 24 samples of gingival tissue per group were required to conduct this clinical trial and statistically reject the null hypothesis with 80% power and a 95% confidence interval. For this sample calculation, the type 1 error associated with the test was 0.05; the χ 2 test without correction was used to evaluate the null hypothesis.
The following data were collected prior to surgery: sex, age, general health status, periodontal condition, and intraoral and extraoral aspects of the dental impaction. A panoramic radiograph was acquired and the tooth position according to the classifications of Pell and Gregory and Winter , degree of tooth development, and level of impaction were recorded.
The surgical procedures were performed in two distinct clinical sessions (one side at a time). Each patient underwent removal of only one third molar per surgical session (i.e., one surgical site). Patients were instructed to call and schedule their second surgical appointment after a period of 21 days, and surgeries were scheduled 1 week later. Hence, the second surgery was performed 28 days after the first (wash-out period).
In order to exclude a possible confounding factor, it was established that both surgical sites in the same patient could not be allocated to a single experimental group. Each treatment was coded as a different group, and groups were identified with the letters A, B, or C. Based on the nomenclature of the blinded groups previously provided to the statistician, six blocks of combinations were created (AB, BA, BC, CB, AC, and CA). Each block represented the treatment that would be administered prior to the removal of the right and left mandibular third molars, respectively. They were randomized through a computer-generated randomization code in Microsoft Excel, confirming that each patient received two different medications. The drugs used in this study were etoricoxib 120 mg, ibuprofen 400 mg, and placebo (without active drug). The medications were administered 1 h before surgery. No antibiotic prophylaxis was administered to the volunteers.
All patients underwent a standardized surgical technique, performed in an outpatient setting, following a strict biosafety protocol. A surgeon with experience in oral and maxillofacial surgery performed all of the surgical procedures. The same surgical protocol was adopted for both sides of the mouth, with the aim of reducing differences in the level of intraoperative trauma. The third molar removal was performed under local anaesthesia with mepivacaine 2% and epinephrine 1:100,000 (DFL, Rio de Janeiro, Brazil), using a maximum of three 1.8-ml cartridges.
A triangular full-thickness flap was raised, followed by peripheral ostectomy using a high-speed hand piece under irrigation with cooled double-distilled water. One sample of soft tissue was collected from the region distal to the third molar before the surgical flap was raised (time 0′) and a second soft tissue sample was obtained 30 min after the surgical procedure (time 30′) for the laboratory analysis of cytokines. The surgical wound was closed with a 4–0 silk suture.
After surgery, ibuprofen 300 mg was prescribed as rescue analgesic, to be taken at intervals of 8 h. Postoperative instructions were also carefully read and explained to the patients. They were instructed to maintain a liquid and soft diet and to avoid hot liquids and/or foods during the first 24 h, and to perform careful oral hygiene without vigorous mouthwashes in order to prevent post-surgical bleeding. The patients were instructed to contact the surgeon by telephone in the case of persistent bleeding, or if they deemed it necessary. In addition, the patients were also asked to report any physical symptoms experienced during the postoperative period of the study, such as nausea, vomiting, dizziness, headache, insomnia, and signs of infection.
The primary outcome of the study was the occurrence of postoperative inflammatory events (pain, facial oedema, and trismus). The intensity of postoperative pain was measured using a visual analogue scale (VAS) of 10 cm, with 0 representing the absence of pain or discomfort and 10 representing the maximum pain or discomfort. After the surgical procedure, each patient received a standardized form with the VAS to report postoperative pain values. Study participants were asked to mark the intensity of their pain at 0, 2, 4, 6, 8, 10, 12, 24, 48, and 72 h, as well as on days 5 and 7 after surgery. In addition, data were collected on the use of rescue medication, including: (1) the time elapsed between the end of the surgical procedure and the ingestion of the medicine by the patient; (2) number of rescue analgesic consumed.
Postoperative oedema was measured using lines drawn between facial points ( Fig. 1 ). These were the distances from the mandibular angle (MA) to: (1) tragus (MA–Tr distance), (2) external corner of the eye (MA–ECE distance), (3) nasal border (MA–NB distance), (4) labial commissure (MA–LC distance), and (5) soft pogonion (MA–SP distance). The preoperative values and those obtained at 24 h, 72 h, and 7 days after surgery were analyzed.
Furthermore, to estimate the degree of mouth opening, the maximum mouth opening was measured in the pre- and postoperative periods (after 24 h, 72 h, and 7 days) by measuring the distance in millimetres between the upper and lower central incisors using a calibrated ruler.
The secondary outcome of this study was the occurrence of changes in the tissue levels of TNF-α and IL-1β in each study group. The gingival tissue samples were stored at −80 °C in Eppendorf tubes containing Radio-Immunoprecipitation Assay solution (Santa Cruz Biotechnology, Santa Cruz, CA, USA) until required for each assay. The collected tissue was homogenized and followed by centrifugation (10,000 rpm/10 min/4 °C). The supernatant was used to determine the expression levels of TNF-α and IL-1β by ELISA method, using commercial kits (Quantikine Human TNF-α Kit (catalogue DY210) and Quantikine Human IL-1 β/IL-1f2 Kit (catalogue DY201); R&D system, Minneapolis, MN, USA). The assays were performed in accordance with the manufacturer’s instructions. The levels of TNF-α and IL-1β were recorded in picograms per millilitre (pg/ml).
The method used to generate the random allocation sequence was the function ‘randbetween’ in Microsoft Excel, 2010 version. The randomization was based on simple type, without any restriction. The mechanism used to implement the random allocation sequence was envelopes that stated the numbers of randomization on the outside. These envelopes contained information specifying the group to which the patient would belong. A collaborating researcher who did not participate in the surgical procedures was responsible for generating the random allocation sequence, as well as for organizing and distributing the participants in each group.
Through the blinding protocol used in this study, the patient, researcher, and statistician did not know to which group each patient belonged. Before the surgical procedures, a list containing the random distribution of all surgical sites and respective medicines to be administered was kept in a sealed envelope by an external collaborator who remained unaware of the study protocol until the data analysis. The statistical analysis was initially performed with groups encoded with the letters ‘A’, ‘B’, and ‘C’, which represented the different groups studied. These codes were revealed at the end of the study.
The data were expressed as the mean and standard deviation and submitted to the Kolmogorov–Smirnov normality test prior to further analysis by Kruskal–Wallis test followed by the Dunn or Wilcoxon post-hoc test (non-parametric data), or one- or two-way analysis of variance (ANOVA) for repeated (or not) measures followed by the Bonferroni post-hoc test (parametric data). Pearson’s correlation analysis was used to assess the correlation between the sum of the pain scores and the total consumption of rescue medication (parametric data), and Spearman’s correlation analysis was used to assess the correlation with the levels of cytokines (non-parametric data). Categorical data were analyzed by χ 2 test.
All analyses were performed in GraphPad Prism 5.0 (GraphPad Software, San Diego, CA, USA) adopting a 95% confidence interval; P < 0.05 was considered statistically significant.
A total of 540 patients were screened for study eligibility ( Fig. 2 ). From this total, 504 individuals were excluded because they did not meet the study criteria. The final sample was composed of 36 volunteers (16 male, 44.4%; 20 female, 55.6%), for a total of 72 surgical sites, divided into 24 procedures per group.
There were no statistically significant differences in clinical, radiographic, or surgical characteristics between the groups ( P > 0.05); the level of operative difficulty was similar across the surgical procedures. The distribution of surgical sites did not differ between males (9 placebo, 11 ibuprofen, 12 etoricoxib) and females (15 placebo, 13 ibuprofen, 12 etoricoxib) ( P = 0.590). There was no statistically significant difference in age group distribution between the groups ( P = 0.436): <20 years (4 placebo, 1 ibuprofen, 5 etoricoxib), 20–30 years (19 placebo, 21 ibuprofen, 16 etoricoxib), >30 years (1 placebo, 2 ibuprofen, 3 etoricoxib). There was also no statistically significant difference between teeth with total bone inclusion (5 placebo, 7 ibuprofen, 8 etoricoxib), partial bone inclusion (11 placebo, 6 ibuprofen, 7 etoricoxib), and semi inclusion (8 placebo, 11 ibuprofen, 9 etoricoxib) ( P = 0.490).
When considering the Pell and Gregory classification regarding the amount of tooth covered by the anterior border of the ramus, no differences were identified between the groups ( P = 0.817): placebo (12 class I, 11 class II, 1 class III), ibuprofen (15 class I, 9 class II, 0 class III), and etoricoxib (14 class I, 9 class II, 1 class III). In addition, there were no differences between the groups regarding the impaction depth relative to the adjacent tooth: placebo (11 class A, 12 class B, 1 class C), ibuprofen (15 class A, 8 class B, 1 class C), and etoricoxib (10 class A, 13 class B, 1 class C) ( P = 0.787). Tooth position according to the Winter classification did not differ between the placebo (11 vertical, 8 mesioangular, 1 distoangular, 4 horizontal), ibuprofen (13 vertical, 5 mesioangular, 1 distoangular, 5 horizontal), and etoricoxib (9 vertical, 8 mesioangular, 0 distoangular, 7 horizontal) groups ( P = 0.614).
The need for tooth sectioning during surgery did not differ between the groups: placebo (12 presence, 12 absence), ibuprofen (11 presence, 13 absence), and etoricoxib (12 presence, 12 absence) ( P = 0.829). The total number of anaesthetic cartridges used ranged from 1.5 to 2, with no difference between the groups ( P > 0.05). Furthermore, the time required for third molar removal (range 10–40 min) did not differ between the groups ( P = 0.875).
Analysis of pain
The mean time to the need for rescue medication was significantly reduced in the etoricoxib (2.0 ± 1.8 h) and ibuprofen (2.6 ± 1.1 h) groups, when compared to the placebo group (4.5 ± 1.7 h) ( P < 0.001), but the range of mouth opening did not differ between the groups ( P = 0.682).
The placebo group presented a significantly elevated pain peak after 2 h (VAS 6.3 ± 2.9), and the groups treated with ibuprofen and etoricoxib showed significantly elevated pain peaks at 4 h after the surgical procedure (3.9 ± 2.4 and 3.0 ± 2.3, respectively) ( Fig. 3 ). The pain peak reduced significantly in both groups at 6 h after the end of surgery (3.6 ± 1.9 in the placebo group, and 2.6 ± 1.8 and 1.9 ± 1.5 for the ibuprofen and etoricoxib groups, respectively) ( P < 0.001). The group treated with ibuprofen showed significantly less pain than the placebo group at 2 h and 4 h after surgery, and the group treated with etoricoxib showed significantly less pain than the placebo group at 2 h, 4 h, 6 h, 8 h, and 10 h following surgery ( P < 0.001). The cumulative effect of all scores up to 6 h and 12 h postoperatively was significantly lower in the ibuprofen group (2.3 ± 2.3 and 1.5 ± 1.8, respectively) and etoricoxib group (1.6 ± 1.8 and 1.0 ± 1.4, respectively), when compared to the placebo group (3.8 ± 3.2 and 2.5 ± 2.8, respectively) ( P < 0.001). The group treated with etoricoxib showed a lower mean accumulated pain during 12 h when compared to the group treated with ibuprofen ( P < 0.001).
The total consumption of rescue analgesics showed a statistically significant correlation with the sum of the pain scores in all groups (placebo: P = 0.019, r = 0.495; ibuprofen: P = 0.027, r = 0.492; etoricoxib: P = 0.011, r = 0.509).
Analysis of oedema
Measurement of MA–Tr showed a significant increase in the placebo (65.7 ± 9.4 mm), ibuprofen (61.6 ± 7.3 mm), and etoricoxib groups (58.3 ± 5.1 mm) at 24 h after surgery, followed by a significant reduction in the placebo group after 7 days (60.0 ± 65.0 mm), while the ibuprofen and etoricoxib groups showed a significant reduction after 72 h (59.5 ± 5.0 mm and 57.2 ± 4.5 mm, respectively) ( P < 0.001). The mean MA–Tr distance was significantly lower in the group treated with etoricoxib when compared to the placebo at 24 h and 72 h postoperatively ( P < 0.001). The cumulative effect of all the mean MA–Tr distances within the different postoperative times in the placebo group (61.9 ± 8.0 mm) showed a higher value than the ibuprofen group (59.6 ± 5.7 mm) and etoricoxib group (57.2 ± 4.7 mm), while the etoricoxib group had a lower value than the ibuprofen group ( P < 0.001).
The MA–ECE peak measurement was significantly increased at 24 h after the surgical procedure in the placebo (107.8 ± 8.2 mm), ibuprofen (105.7 ± 6.9 mm), and etoricoxib groups (105.4 ± 7.0 mm), and demonstrated a significant reduction in the ibuprofen group after 7 days (103.2 ± 6.2 mm) and in the placebo and etoricoxib groups after 72 h (105.6 ± 6.0 mm and 104.2 ± 6.2 mm, respectively). The etoricoxib group (103.4 ± 6.2 mm) expressed significantly less oedema at 7 days after surgery when compared to the placebo group (104.0 ± 5.6 mm) ( P = 0.004). However, the cumulative effect of the MA–ECE distance did not differ between the three study groups ( P = 0.568).
The peak MA–NB measurement was significantly increased at 24 h after the surgical procedure in the placebo (118.3 ± 8.7 mm), ibuprofen (114.0 ± 6.8 mm), and etoricoxib groups (113.5 ± 8.5 mm), and reduced significantly in the placebo group after 72 h (115.3 ± 7.0 mm) and in the ibuprofen (108.6 ± 7.5 mm) and etoricoxib groups (110.2 ± 7.7 mm) after 7 days. The cumulative effect of all the MA–NB measurements was significantly reduced in the ibuprofen group (110.9 ± 7.3 mm) and etoricoxib group (111.6 ± 7.9 mm) when compared to the placebo group (114.3 ± 7.9 mm) ( P = 0.007).
The peak MA–LC measurement was significantly increased 24 h after surgery in the placebo (101.8 ± 9.3 mm), ibuprofen (94.4 ± 7.1 mm), and etoricoxib groups (94.2 ± 8.5 mm) ( P < 0.001). These peaks were significantly lower in the ibuprofen and etoricoxib groups ( P < 0.001). The oedema peaks reduced significantly in all groups after the first 72 h (95.5 ± 5.3 mm, 92.7 ± 6.7 mm, and 92.9 ± 7.8 mm, respectively). The cumulative effect of all MA–LC measurements was significantly reduced in the ibuprofen (91.1 ± 7.0 mm) and etoricoxib groups (92.0 ± 7.8 mm) when compared to the placebo group (94.3 ± 8.1 mm) ( P = 0.014).
The peak MA–SP distance was significantly increased at 24 h after the surgery in the placebo (122.0 ± 9.1 mm), ibuprofen (116.2 ± 6.8 mm), and etoricoxib groups (116.0 ± 8.6 mm) ( P < 0.001), and the peaks were significantly reduced in the ibuprofen and etoricoxib groups ( P < 0.001). There was a significant reduction in oedema in the placebo (117.7 ± 7.0 mm) and etoricoxib groups (114.4 ± 7.4 mm) after 48 h. The ibuprofen group (111.3 ± 7.3 mm) demonstrated a reduction in oedema after 7 days. The cumulative effect of all the MA–SP measurements was lower in the ibuprofen group (113.2 ± 7.6 mm) compared to the placebo group (116.0 ± 8.1 mm). No difference in the cumulative effect of all oedema measurements was observed when comparing the three study groups ( P = 0.108).
Analysis of mouth opening
The group treated with placebo showed a maximum mouth opening value of 32.1 ± 7.0 mm after 24 h, with a significant improvement after 72 h (37.1 ± 8.3 mm) and at 7 days postoperative (43.1 ± 8.0 mm) ( P < 0.001). The ibuprofen and etoricoxib groups showed maximum mouth opening values after 24 h of 34.4 ± 8.7 mm and 39.2 ± 8.4 mm, respectively, with significant improvements after 7 days (49.2 ± 8.9 mm and 49.9 ± 8.0 mm, respectively) ( P < 0.001). Mouth opening was significantly greater in the group treated with ibuprofen on day 7 postoperative and in the group treated with etoricoxib at 24 h, 72 h, and 7 days after surgery, when compared to the placebo group ( P = 0.040). The group treated with etoricoxib (46.5 ± 9.5 mm) demonstrated greater mean mouth opening measurements than the placebo group (40.8 ± 10.0 mm) ( P = 0.001).
The level of TNF-α in the placebo group showed no statistically significant difference from time 0′ (99.8 ± 123.3 pg/ml) to time 30′ (47.7 ± 42.7 pg/ml, P = 0.127); however, the ibuprofen (137.4 ± 130.2 to 32.1 ± 31.6 pg/ml, P = 0.001) and etoricoxib groups (88.5 ± 1.0 to 31.7 ± 33.4 7 pg/ml, P = 0.016) showed statistically significant reductions in TNF-α from 0 to 30 min after the beginning of the surgical procedure ( Fig. 4 ). There was no significant difference between the three groups at time 0′ ( P = 0.274) or time 30′ ( P = 0.230).
No significant variations (Δ) in the levels of IL-1β were observed in the placebo (244.3 ± 99.7 to 268.5 ± 85.1 pg/ml; P = 0.487) and etoricoxib groups (246.5 ± 134.3 to 217.5 ± 97.9 pg/ml; P = 0.593); however, there was a significant reduction in the levels of IL-1β in the ibuprofen group from time 0′ (322.0 ± 168.8 pg/ml) to time 30′ (253.1 ± 132.9 pg/ml; P = 0.038). There was no significant difference in the levels of IL-1β between the three groups at time 0′ ( P = 0.354) or time 30′ ( P = 0.500).
Analysis of the correlation between clinical parameters and cytokine levels
Levels of TNF-α at time 0′ ( r = 0.603) and at time 30′ ( r = 0.451) were positively correlated with the amount of rescue medication consumed in the placebo group ( Table 1 ). The levels of TNF-α at time 0′ also showed a positive correlation with the VAS at 2 h ( r = 0.461). In the ibuprofen group, the consumption of rescue medications was positively correlated with levels of TNF-α at time 30′ ( r = 0.546) and the levels of IL-1β at time 30′ ( r = 0.568). The level of pain at 7 days after surgery showed a significant negative correlation with the levels of TNF-α at time 30′ ( r = −0.573) and the levels of IL-1β at time 30′ ( r = −0.501), whereas the reduction (Δ) in IL-1β levels showed a significant positive correlation with pain levels after 7 days ( r = 0.434). In the etoricoxib group, the variation (Δ) in TNF-α levels correlated significantly with the consumption of rescue medication ( r = −0.436), difficulty in mouth opening ( r = −0.477), VAS at 6 h ( r = −0.466), and VAS at 8 h ( r = −0.419).