Cyclooxygenase Inhibition for Postoperative Analgesia

The anesthesiologist has a central role in the management of acute postoperative pain, and effective management has been clearly demonstrated to improve clinical outcomes (1,2). Acute pain can rapidly evolve into chronic pain, and the two should not be viewed as separate entities (3). Failure to achieve effective analgesia is a major predictive factor for the conversion of acute postoperative pain to chronic pain after several different types of surgery (4). Traditionally, opioids and local anesthetics have been the pharmacological foundation for postoperative pain management; however, optimization has been difficult because of anxieties regarding deleterious side effects. Opioids may cause respiratory depression, oversedation, and prolongation of postoperative ileus. Regional anesthetic techniques are limited by the requirement for special monitoring and the irreversibility of neuronal blockade once it is instituted. The effectiveness of nonsteroidal antiinflammatory drugs (NSAIDs) in alleviating pain and reducing requirements for opioids in the postoperative period has been well documented (5–7), but concerns regarding gastropathy, renal dysfunction, and hemostatic defects have limited their use. There is further anxiety because fasting and the frequent prevalence of hypovolemia in the postoperative period may potentiate gastric and renal side effects. Do anesthesiologists underuse NSAIDs as perioperative analgesics because of exaggerated anxieties regarding side effects and inadequate information regarding mode of action? This review article describes prostaglandin (PG) biosynthesis and the central role of cyclooxygenase (COX) expression in pain perception, and then it focuses on the potential side effects of COX inhibition. Goals of Postoperative Analgesia The goal of postoperative pain relief is to achieve optimal analgesia, facilitating a speedy return to normal physiological organ function with minimal side effects. Anesthesiologists must also try to reduce the incidence of chronic pain after surgery by effective treatment of acute postoperative pain. Minimization of opioid and local anesthetic side effects is a major consideration, and here NSAIDs have great potential. This process begins at the preoperative assessment, when the anesthesiologist must identify patients at increased risk of NSAID-induced complications. Although renal failure, allergy, and late pregnancy are absolute contraindications, a history of peptic ulcer disease, congestive cardiac failure, or cirrhosis, previous thrombosis, and concomitant corticosteroid therapy require further consideration before being classed as contraindications. A history of NSAID consumption for other conditions, such as arthritis, is reassuring, and a normal creatinine clearance will allay fears regarding postoperative renal failure. PG Biosynthesis In 1971, Vane (8) demonstrated that aspirin exerted its effects by inhibiting PG synthesis via inhibition of the enzyme COX. In 1991 the DNA sequence in human tissue for COX-1 was identified and, for COX-2, in 1992 (9,10). The two different isoforms are now called COX-1 and COX-2 (11). Arachidonic acid is released from cell membrane phospholipid by the activity of phospholipase A2 (PLA2). PLA2 is essential for PG synthesis, and a synchronous regulation of PLA2 and COX has been demonstrated (12,13). The enzymatic activity of COX on arachidonic acid forms PGH2. Newly generated PGH2 is then converted to PGD2, PGE2, and PGF2α, prostacyclin (PGI2), and thromboxane A2 (TXA2) by tissue-specific isomerases (11,14). All prostanoid synthesis is dependent on the action of COX on arachidonic acid, although the effect at the cellular level is cell specific. Endothelial cells synthesize mainly PGI2, which inhibits platelet aggregation, and platelets synthesize mainly TXA2, which has a proaggregational effect. Recent work on rat tissue has also demonstrated different levels of expression of COX-1 and COX-2 messenger RNA (mRNA) in different organs. COX-1 mRNA was expressed in the liver, lung, and kidney, but not the heart, whereas COX-2 mRNA was expressed in the lung, heart, and kidney, but not the liver. This suggests that COX-1 and COX-2 have a regulatory physiological role that is organ specific. This may also indicate that certain organs will be more sensitive to COX inhibition than others (15). COX and Pain Surgical tissue trauma results in the release of a vast number of inflammatory mediators, including prostanoids. These mediators affect nociceptors, altering their firing threshold and sometimes causing direct stimulation (16). COX-2 plays a key role in the central neurological response to inflammation. Immunoreactivity studies have demonstrated the presence of COX-2 in the superficial dorsal horn of the spinal cord (17,18), and PG receptors are found in dorsal root ganglia (19,20). It is at this level that the afferent sensory nerve axons enter the spinal cord and spinal nociceptive processing begins. Central neurons generate both PGE2 and PGI2 in response to peripheral injury (21), and intrathecal application of PGs induces pain behavioral responses (22,23). In a rat model, peripheral injury has resulted in COX-2 induction in the dorsal root neurons (21,24–27). Moreover, intrathecal COX inhibition has been demonstrated to reduce hyperalgesic responses to thermal injury in rats (28). COX-1 and -2 are found in brain tissue, with COX-2 mRNA being demonstrated in the resting condition (29). Electroconvulsive seizures induce COX-2 mRNA in rat brain, and this was found to be N-methyl-d-aspartate (NMDA) dependent, because the glutamate antagonist MK-801 abolished this response. COX-2 was also colocalized with glutamate in the dendrites of excitatory neurons (30). NMDA receptor activation results in increased prostanoid synthesis (31). Neuronal depolarization is mediated by the excitatory amino acids glutamate and aspartate. These excitatory amino acids are present in many types of sensory neurons and exert their effects initially via α-amino-3-hydroxy-5-methyl isoxazolepropionic acid (AMPA) and kainate channels (32). Excitatory amino acid activation of AMPA and kainate channels leads to activation of NMDA receptors, with resultant long-lasting partial depolarization; subsequent low-intensity stimuli, which previously were unable to generate an action potential, can then do so (33). The activation of NMDA receptors is crucial for both acute pain perception and for central sensitization and windup to occur, potentially leading to chronic pain (34,35). Spinal NMDA activation causes acute mechanical nociception (through activation of COX-2 enzymes) and hyperalgesia (as a result of peripheral inflammation via upregulation of COX-2) (36). NMDA receptors have also been linked to COX-2 expression in a focal ischemia model, because the induction of COX-2 by ischemia is inhibited by an NMDA antagonist (37). As with NMDA receptors, AMPA activation seen after global ischemia results in prolonged neuronal COX-2 expression (38). The effects of NMDA and AMPA activation may be COX-2 dependent, because intrathecal injection of a COX-2 inhibitor inhibits the development of hyperalgesia induced by intrathecal injection of either NMDA or AMPA in a dose-dependent manner (39). Thus, both AMPA and NMDA receptor activation are linked to COX-2 expression with resultant nociception (Fig. 1).Figure 1: A diagrammatic representation of the central role played by cyclooxygenase (COX)-2 in nociception at the spinal cord level. Activation of both α-amino-3-hydroxy-5-methyl isoxazolepropionic acid (AMPA) and N-methyl-d-aspartate (NMDA) receptors results in COX-2 expression, whereas the administration of an NMDA antagonist inhibits this surge in COX-2 after initial activation. NMDA activation also results in mitochondrial membrane potential changes, which are COX-2 dependent. Intrathecal administration of a COX-2 inhibitor prevents the hyperalgesia seen after intrathecal administration of both AMPA and NMDA.There is a distinct neurochemical response in acute and persistent pain. It is interesting to note that rat spinal cord sections from an adjuvant-induced arthritis model have demonstrated the induction of COX-2 at 22 days (40). This is further evidence in support of the concept that the neurochemistry of the central spinal cord is altered in a specific fashion after an insult and that this effect may be prolonged (41,42). Thus, COX-2 expression induced by surgery may last longer than previously thought, and this raises the possibility of persistent COX-2 expression being a possible factor in the etiology of chronic pain after surgery. Gastrointestinal Effects of COX Clinically the most common side effects of NSAIDs are seen in the gastrointestinal system (43,44). The risk of NSAID-induced peptic ulcer complications results in approximately 1.25 additional hospitalizations per 100 patient-years on average. However, this may be affected by the presence of Helicobacter pylori and varies among different NSAIDs and dose regimens. Small-dose ibuprofen has been reported to cause fewer gastrointestinal complications than other NSAIDs. The risk of gastropathy is increased in patients who are elderly, have a history of peptic ulcer disease, or take concomitant corticosteroids or anticoagulants (45). There has been great interest in the cytoprotective role of PGs in the gastrointestinal system. In particular, PGE2 and PGI2 have been shown to increase mucous production and inhibit acid secretion in the stomach (46). The healing of peptic ulcers is enhanced by COX-2-dependent PG synthesis (47–49). Thus, it would seem that inhibition of PG synthesis by NSAIDs might provide a possible explanation for NSAID-induced peptic ulceration. It has been postulated that NSAID inhibition of COX-1 biosynthesis in the stomach is the key factor responsible for gastric erosion formation (50). However, it has been demonstrated, in mice with a disrupted COX-1 gene, that the associated gastric PG synthesis reduction was not associated with gastric damage (51). H. pylori infection in the stomach is associated with peptic ulcer disease, and it is interesting that even in the presence of H. pylori colonization, COX-2-dependent PG synthesis is present and may be protective (52). The use of selective COX-2 inhibitors in patients with osteoarthritis and rheumatoid arthritis has resulted in less gastric erosion than that seen when using earlier NSAIDs at dosages that have an antiinflammatory effect (53,54). These observations again support the hypothesis that it is COX-1 inhibition that leads to gastric erosion. However, recent evidence suggests that selectively inhibiting either COX-1 or COX-2 in isolation does not cause gastric erosion. Inhibition of both isoforms is to some extent required for gastric mucosal damage to occur (55). The mechanism of NSAID-induced gastric mucosal ulceration is unclear, but it probably involves an NSAID-induced adherence of leukocytes to the vascular endothelium in the gastric microcirculation (56) in combination with a reduction in gastric blood flow (57). There has always been a concern that patients may be exposed to an increased risk of gastrointestinal and renal side effects from the use of NSAIDs for postoperative analgesia, because of postoperative fasting and the relatively frequent incidence of hypovolemia after major surgery. However, the newer NSAIDs have a greater COX-2 inhibitory effect, and although there are still some COX-1 inhibitory effects, there is growing clinical evidence of the superior tolerability of the more selective COX-2 inhibitors, with the Vioxx Gastrointestinal Outcome Research Group study demonstrating a 60% reduction in perforation, obstruction, and bleeding (58–64). Renal Effects of COX Renal complications of NSAID use include acute renal failure, hyperkalemia, water and sodium retention, nephrotic syndrome, and interstitial nephritis (65,66). It is interesting to note that in the United States, prescription NSAID use results in 25 excess hospitalizations because of renal failure per 10,000 years of use in patients older than 65 yr. This suggests that NSAID-induced complications in older people are uncommon (67). However, it may not be possible to extrapolate these encouraging data to the perioperative period. COX-2 is constitutively expressed in the kidney (68), and the role of PGs in renal homeostasis is complex. PGI2 and PGE2 are the major mediators, with PGI2 found in the renal cortex and glomeruli and PGE2 in the juxtamedullary glomeruli and medullary portion of the collecting ducts. They produce arteriolar dilation, increasing renal perfusion and sodium excretion, while reducing medullary tonicity and antidiuretic hormone effectiveness and stimulating renin secretion. PGI2 attenuates podocyte constriction, thereby maintaining glomerular filtration rate (GFR) in the presence of angiotensin-2 and antidiuretic hormone. TXA2 is found in the glomeruli, where it causes vasoconstriction and podocyte contraction and thus reduces glomerular filtration. PGF2α enhances tubular sodium and water excretion (69). Tubulo-glomerular feedback is an important mechanism of autoregulation for renal perfusion. The macula densa is the sensing part of this mechanism, and its activity requires a COX-2 pathway for signaling (70). The effect of a selective COX-2 inhibitor (nimesulide) was compared with that of a nonselective inhibitor (meclofenamate) in dogs with normal and small sodium intake. The effects on renal hemodynamics, urinary sodium excretion, and fractional lithium excretion were quite different, suggesting that COX-2-derived PGs play an important role in sodium excretion and renal hemodynamics (71). Endogenous PGs have been demonstrated not to affect renal hemodynamics under the pressor effects of prolonged angiotensin-2 exposure. In fact, after significant renal injury, prolonged therapy with selective COX-2 inhibition decreased proteinuria and retarded further renal injury (72). However, renal PGs do modulate the effects of subpressor angiotensin-2 levels by reducing vasoconstriction and enhancing tubular sodium absorption. This suggests that under the threat of hypoperfusion, endogenous renal PGs ensure that the renin-angiotensin system will work to maintain both renal and systemic perfusion. In patients with chronic renal failure, residual function is PG dependent, and here NSAIDs may cause rapid deterioration (69). Sodium retention and edema are found in 5% of the population taking NSAIDs (73). Hyperkalemia is noted in up to 46% of patients because of decreased renin secretion (74). Clinical trials have produced inconsistent results; one study demonstrated no GFR reduction in patients with chronic renal failure taking celecoxib or naproxen, whereas another demonstrated no difference in GFR reduction between patients taking rofecoxib or indomethacin (75–78). Patients with cardiac failure, cirrhosis, or hypertension have a predisposition to prerenal renal failure. The anesthesiologist should view these patients as being at increased risk for NSAID-induced renal failure. A recent study demonstrated that celecoxib caused less edema and destabilization of blood pressure control than rofecoxib in hypertensive patients and suggested that these patients should be monitored for cardiorenal events if they are prescribed selective COX-2 inhibitors (79). It is interesting that both celecoxib and rofecoxib have been reported to cause acute renal failure (80,81). The effect of diclofenac and ketorolac on perioperative renal function has been investigated in a rat model. The administration of the NSAIDs did not induce renal dysfunction. However, if combined with gentamicin, both NSAIDs produced marked deterioration in renal function, with associated pathologic changes in renal structure (82,83). An interaction between ibuprofen and gentamicin resulting in renal failure has also been reported in humans (84). This evidence suggests that the coadministration of aminoglycosides with NSAIDs in the perioperative period may be a significant risk factor for inducing renal failure. PGs and the Cardiovascular System Myocardial ischemia results in ischemic preconditioning, which protects myocardial cell function from further ischemic episodes. Evidence suggests that upregulation of nitric oxide is required for preconditioning (85), and this is associated with a simultaneous induction of COX-2 (86). COX-2-mediated cardioprotection occurs in the later phases of preconditioning and is due to PGE2 and PGΙ2 synthesis (87). These PGs attenuate myocardial stunning (88) and reduce infarct size (89). The cytoprotective effects of PGE2 and PGΙ2 are due to activation of adenosine triphosphate-sensitive potassium channels (89) and a reduction in myocardial neutrophil infiltration (90). Although these animal data suggest that COX-2 inhibition may worsen myocardial performance in the presence of ischemia, there is now evidence that COX-2 inhibition reduces atherosclerosis. Atherosclerosis is now viewed as a chronic inflammatory condition (91) in which COX-2 is upregulated (92). COX-2 inhibition markedly reduces neointimal formation in an animal model of atherosclerosis (93). There is concern regarding the effect of COX inhibition on renal artery tension. PGs have different effects on vascular smooth muscle; TXA2 causes vasoconstriction and platelet aggregation (94), whereas PGI2 has the opposite effect (95). The in vitro vasoconstriction caused by angiotensin-2 in the human internal mammary artery is enhanced by, but not dependent on, eicosanoid biosynthesis produced from the COX pathway (96). Therefore, vascular tension may be affected by COX inhibition, but neither isoform is predominant or is the sole factor affecting tension. Postoperative Nausea and Vomiting Work in conscious piglets has demonstrated that activation of the medullary vomiting center involves PGs. Lipopolysaccharide-induced emesis was abolished by COX inhibition by using both meloxicam and indomethacin, when these were administered before the lipopolysaccharide (97). Hemostatic Complications COX inhibition with NSAIDs may increase the risk of vascular thrombus formation by upsetting the balance of pro- and antiplatelet aggregation effects: TXA2 synthesis is primarily a COX-1-induced effect, and PGI2 is a COX-2 effect. Four patients with Raynaud’s syndrome developed thrombotic complications after receiving COX-2 inhibition as part of their treatment (98). This side effect is with the nonselective COX Although platelet occurs with COX-1 inhibition, there has been no control clinical demonstrating an increase in in patients who have NSAIDs. PG with administration enhances the response. However, with these effects in rats when they were with and the effects were abolished by nitric oxide and These results suggest that is a nitric and pathway PGs and and NSAIDs cause in approximately of although it is interesting that they it in approximately of patients This the of COX interaction in the Recent work has possible for and these will NSAID by anesthesiologists in this difficult COX inhibition arachidonic acid to the with an associated reduction in PGE2 synthesis and an increase in Patients with have increased levels because of enzymatic which can be by more from COX inhibition There is growing evidence that is due to COX-1 inhibitory effects and that selective COX-2 inhibitors are relatively to use in this population Analgesia with NSAIDs The concept of pain as to pain is to and here NSAIDs have a role to Inhibition of COX expression prevents AMPA and NMDA receptor activation and spinal Although support for analgesia in the of the because of interest has developed as a greater of the mechanism of central sensitization has been of the major is on the of It may be as treatment that prevents the of central sensitization caused by and inflammatory both the and initial postoperative it is that treatment must be before surgery to have its effect. Therefore, COX expression must be inhibited before surgery and the initial postoperative because the response to surgery may last longer than Work a role for spinal COX-2 expression in postoperative nociception in humans There is growing support for the concept that of action is at the spinal level Therefore, the administered NSAID must have COX inhibition before the NSAID are in facilitating rapid levels in the perioperative and this will of the effect, which may occur with other of and are for and analgesia has been demonstrated with both drugs analgesia now clearly further is to this should part of of NSAIDs techniques can be to the for different including for TXA2 and PGE2 and inhibition of human gastric PGs. There may be significant in on the although inhibition of human gastric PGs may the The prevalence of side effects from prescription NSAIDs is and in the postoperative pain relief they are of diclofenac analgesia to of The of selective COX-2 inhibitors is to that of the nonselective with less potential for COX-1-induced complications There has been no that that patients who NSAIDs have an increased risk of peptic renal dysfunction, or hemostatic in the postoperative period compared with patients who did not NSAIDs. the risk of there is a of evidence for selective COX-2 inhibitors It is that COX-2 plays a central role in pain The of NSAIDs that selectively inhibit COX-2 may side effects while analgesia, a reduction in opioid and the potential to reduce the incidence of chronic pain after surgery and, also and Therefore, the development of these more selective COX-2 inhibitors, in with preoperative the for their use in the perioperative period. The one of concern to be renal function, and further evidence preoperative evidence of renal should their use. However, because selective COX-2 inhibitors are well by patients with are associated with less gastric injury and do not they may a to nonselective NSAIDs in the treatment of postoperative pain.