Deleterious Impact of a γ-Aminobutyric Acid Type A Receptor Preferring General Anesthetic When Used in the Presence of Persistent Inflammation

• Some studies suggest that the incidence of chronic pain after surgery differs with the type of intraoperative anesthetic.

• In rats, recovery of hypersensitivity after incision was not affected by anesthesia with pentobarbital, isoflurane, or ketamine/xylazine. Recovery from persistent inflammation of the paw, either alone or with incision, was much slower after pentobarbital anesthesia.

AN estimated 234 million people worldwide undergo at least one surgical procedure each year,1a significant fraction of which are performed on patients with persistent inflammation.2Despite great effort, consistently adequate management of postoperative pain remains elusive. Even more concerning is recent evidence suggesting that the magnitude of postoperative pain has a significant impact on the outcome of the surgical intervention, with more pain associated with a greater risk for complications, hospital admission, longer stays in the postoperative anesthesia care unit, and slower recovery.3For these reasons, any factor exacerbating postoperative pain has the potential to have a major negative impact on public health.

Recent experimental data suggest that persistent inflammation is associated with a shift in ionotropic (type A) γ-aminobutyric acid receptor (GABA^A) signaling in the spinal cord. In the absence of inflammation, spinal administration of GABA^Aagonists is analgesic, whereas administration of antagonists results in hypersensitivity.4,–,7However, in the presence of persistent inflammation, this normally inhibitory neurotransmitter system becomes excitatory: GABA^Aagonists facilitate nociceptive transmission, and antagonists become analgesic.4Because, with few exceptions, general anesthetics commonly used for surgical interventions act at the GABA^A, a critical prediction of this preclinical observation is that GABA^A-preferring general anesthetics may increase postanesthesia pain in patients with persistent inflammation.

The current study was designed to begin to test this prediction. Three general anesthetics with different mechanisms of action were selected to evaluate the impact of anesthetic type on the magnitude and duration of postoperative pain and hypersensitivity. A follow-up study with the GABA^A-preferring general anesthetic pentobarbital was used to determine whether hind paw incision was necessary for the manifestation of a deleterious impact of persistent inflammation on ongoing hypersensitivity. Persistent inflammation was induced with a subcutaneous injection of complete Freund adjuvant, and hind paw incision8was used to drive postoperative pain.

All experiments were reviewed and approved by the University of Pittsburgh Institutional Animal Care and Use Committee (Pittsburgh, Pennsylvania) and were performed in accordance with guidelines established by the International Association for the Study of Pain. Adult male Sprague-Dawley rats (125–150 g at the time of purchase; Harlan Sprague-Dawley, Indianapolis, IN) were housed in pairs before surgery and housed singly after surgery. They were maintained on a 12-h light/12-h dark schedule (lights on at 7:00 AM); food and water were provided ad libitum . At the end of the study, all animals were killed with carbon dioxide gas followed by cervical dislocation.

Rats were divided into nine groups: naïve; inflammation only; surgery with pentobarbital; surgery with ketamine/xylazine; surgery with isoflurane; inflammation + surgery with pentobarbital; inflammation + surgery with ketamine/xylazine; inflammation + surgery with isoflurane; and inflammation + general anesthesia with pentobarbital.

Batches of rats were received in three groups of 6, two groups of 12, and three groups of 18, for a total of 96 rats. Manipulations were varied across and within all groups. The timing of the overall study design is illustrated in figure 1A.

After 3 days of baseline behavioral testing, inflammation was induced with a 50-μl injection of complete Freund adjuvant (CFA; Sigma–Aldrich, St. Louis, MO) diluted 1:1 in sterile saline. The injection was made subcutaneously into the left plantar hind paw, 1 cm from the proximal edge of the heel along the midline, during isoflurane anesthesia (5% induction and 2% maintenance). The entire procedure was completed in ∼45 s. To control for the impact of this initial anesthetic exposure, all animals (both inflamed and uninflamed) received the same isoflurane treatment.

Two days after the inflammation/initial isoflurane exposure, control and inflamed rats were further divided into groups defined by the implementation of a surgical intervention, which consisted of the plantar incision originally described by Brennan and colleagues.8,9However, two important modifications to the original procedure were implemented. The first was the general anesthetic used to maintain rats in a surgical plane of anesthesia during the procedure. The second was the duration of anesthesia, which was extended to 3 h. This 3-h period was chosen to reflect the duration of a “typical” surgical procedure.10The three anesthetics used in this second round of anesthesia included pentobarbital (50 mg/kg intraperitoneal), ketamine (70 mg/kg intramuscular) + xylazine (20 mg/kg intramuscular), and isoflurane (2% in 100% oxygen). Rats receiving pentobarbital or ketamine/xylazine were supplemented with subcutaneous injections as needed to maintain areflexia for the duration of the anesthetic period, whereas the isoflurane concentration was adjusted for the same purpose as needed. Rats were randomly assigned to treatment groups. In a follow-up study, another group of rats, which was inflamed and anesthetized but not surgically manipulated, was added to these original groups. Behavioral data were collected in a manner in which the experimenter was blinded to the delivery of pentobarbital and ketamine/xylazine. However, although animals were not marked to indicate the anesthetic they received, because one of the experimenters who collected the behavioral data also participated in the management of the rats during the period of anesthesia, he could not be blinded to the administration of isoflurane as he was for the administration of the other anesthetics.

Isoflurane was used for induction of anesthesia before the initiation of the surgical procedure, and rats were maintained on isoflurane (isoflurane group) or subsequently received pentobarbital or ketamine/xylazine for maintenance of anesthesia. All rats received 100 μl antisialagogue atropine sulfate (0.1 mg/ml subcutaneous).

After a surgical plane of anesthesia under the different anesthetics was established, the left plantar hind paw was sterilized with 10% povidone-iodine. A number 11 blade was used to make a 1-cm midline longitudinal incision through skin and fascia, starting 0.5 cm from the proximal edge of the heel and extending distally. Hemostasis was accomplished by applying pressure with 2-inch × 2-inch gauze pads. Using a curved hemostat for blunt dissection, the plantaris muscle was exposed and lifted. A 0.25-cm longitudinal incision was made through the plantaris muscle with a number 11 blade. A curved hemostat was inserted between the incision to further separate the muscle and verify a complete cut. The muscle origin and insertion remained attached. The skin was apposed with two subcutaneous horizontal mattress sutures using 5-0 nylon. This entire surgical procedure was completed in ∼5 min. After surgery, anesthetized animals were placed on a warm water blanket for the remainder of the 3-h anesthesia period. After 3 h, the rats were monitored until they awoke and then were returned to their individual cages.

Changes in mechanical sensitivity were assessed with the von Frey test, and the presence of ongoing pain was assessed by monitoring guarding behavior.8,9 

Rats were habituated to the testing procedure and experimenter for 3 days before the collection of baseline data. Habituation consisted of moving the rats from their home room to the testing room, handling the rats, and placing them in the testing apparatus for 30 min. The testing apparatus consisted of a row of six clear acrylic enclosures (4 inches [∼10.2 cm] wide, 7–5/8 inches [19.4 cm] long, and 5 inches [12.7 cm] high) atop anodized aluminum mesh (1/4”[6.4 mm] waffle hole). The enclosures were separated by opaque dividers. On testing days, groups of six rats were loaded into enclosures 30 min before the initiation of data collection. An electronic von Frey anesthesiometer (IITC Plantar Test Analgesia Meter 2390; IITC Life Sciences Inc., Woodland Hills, CA) fitted with a rigid tip (1.0-mm tip diameter) was used to assess changes in mechanical threshold. The tip was applied to the ventral surface of the hind paw adjacent to the surgical incision site with steady vertical pressure until the paw was lifted off the mesh floor (at ∼60g ) or until the rat withdrew from the stimulus. The greatest force generated before withdrawal was recorded. Withdrawal data were collected for the right hind paw for each rat, followed by the left hind paw. This procedure was repeated three times, and the average of the three measures for each paw was considered the withdrawal threshold. Although readings above 60g  were obtained on several occasions, 60g  was considered the cutoff. Total testing time (not including habituation) for each group of six rats was 25 to 35 min each day. Testing for each group of rats was initiated at the same time of day with no more than six groups (36 rats) being tested on any given day. Thus, the window of testing was constrained to a period of ∼6 h starting at 7:30 AM. Rats from each experimental and control group were randomized between testing sessions. Most rats were tested for 12 consecutive days consisting of 3 days of baseline, 2 days after inflammation, and 7 days after surgery. The last group of 18 rats was tested for only 5 days after anesthesia.

Guarding behavior was recorded during the second 15 min of habituation in the testing chambers before the collection of mechanical threshold data and again for a 15-min period starting at the end of the collection of mechanical threshold data. Behavior was scored as previously described.8Briefly, the left hind paw was examined through the enclosure during a 10-s period once every minute for 15 min. A score of 0, 1, or 2 was recorded based on the position of the left hind paw during the majority of the 10-s interval. Zero was given if the rat was able to place full weight on the incised foot; 1 was given if the rat's foot touched the mesh but did not bear full weight; 2 was given if the rat held its foot completely off the ground or spent most of the time licking the wound or inflamed paw. The 30 scores were summed. Guarding scores were recorded for 7 days after surgery for every rat except those in the last group tested, including those not undergoing an operation.

We had two a priori  hypotheses. First, that the use of a GABA^A-preferring general anesthetic in the presence of persistent inflammation increases the magnitude of inflammatory hypersensitivity. Second, the use of a GABA^A-preferring general anesthetic slows the recovery from inflammatory hypersensitivity, whether or not the inflamed tissue was surgically manipulated. One- and two-way analyses of variance (ANOVA) were used to test the first hypothesis, with a mechanical threshold at 12 h after surgery/anesthesia serving as the dependent variable. A mixed-design, two-way ANOVA was used to assess the second hypothesis. If there was a significant interaction between factors, the Holm-Sidak test was used for post hoc  comparisons between groups. Given that guarding scores represent a discontinuous variable, these data were analyzed with a Kruskal-Wallis test followed by post hoc  analysis with a Dunn test. To facilitate graphic representation of data, the results of a mixed-design, two-way ANOVA are provided for recovery data. Differences between groups with P < 0.05 were considered statistically significant. All analyses were performed with SigmaStat software (V3.5; Systat Software, Chicago, IL).

Ninety-six rats were randomized to five major groups (naïve, n = 18; inflammation only, n = 16; incision only, n = 24; inflammation + anesthesia only, n = 8; and inflammation + incision during anesthesia, n = 30), for which the incision only and inflammation + incision groups were further divided into three subgroups defined by general anesthetic. The inflammation + anesthesia only group was anesthetized with pentobarbital. Three naïve rats and one inflammation + incision rat anesthetized with pentobarbital were removed from testing on day six after learning to terminate the nociceptive testing procedure by biting the tip off of the electronic von Frey device. One inflamed + incision rat that received isoflurane died during anesthesia. All data from these rats were excluded from additional analysis. Thus, data from 91 rats were used for subsequent analysis.

As expected, the mechanical threshold was relatively stable over the 12 days of testing in naïve rats. There also was no detectable influence of hind paw manipulations (i.e. , CFA or incision) on mechanical threshold in the hind paw contralateral to the site of inflammation (fig. 1B). CFA-induced inflammation was associated with a significant decrease in mechanical threshold that was maximal by 24 h after injection and returned to baseline over the subsequent test days. Hind paw incision was also associated with a significant decrease in mechanical threshold that recovered over a time course comparable with that associated with CFA-induced inflammation. There was a significant (P = 0.03, two-way ANOVA) influence of inflammation on mechanical threshold ∼12 h after surgery: the threshold was lower in the inflamed group (3.5 ± 0.6, n = 28, mean ± SEM) than in the noninflamed group (5.2 ± 0.6, n = 24). There was also a significant (P < 0.01, two-way ANOVA) influence of inflammation on guarding behavior observed ∼12 h after surgery: scores were greater in the inflamed group (38.0; 26.5 and 50.3; median; 25th and 75th percentiles) than in the noninflamed group (29.0; 20.5 and 37.0).

Because the primary purpose of this study was to determine whether the type of anesthetic used during a surgical procedure influenced postoperative measures of pain and sensitivity, we next analyzed the surgery groups as a function of the anesthetic used during the surgical procedure. Based on the suggestion that guarding behavior is a measure of ongoing pain, we first assessed the impact of anesthetic on guarding behavior after surgery in inflamed and uninflamed rats. There was no significant influence of anesthetic on either the peak (at ∼12 h after surgery) or decay of the guarding behavior over time in inflamed (fig. 2A) or uninflamed (fig. 2B) rats.

We next assessed the impact of anesthetic on mechanical sensitivity. There was a small, but significant, influence of anesthetic on the decrease in mechanical threshold observed ∼12 h after the hind paw incision in inflamed rats (fig. 3A). This group effect (P = 0.02, one-way ANOVA) was attributable to the mechanical threshold being significantly (P < 0.05, Holm-Sidak post hoc  test) lower in rats anesthetized with pentobarbital than in those anesthetized with isoflurane. There was no detectable influence of anesthetic on mechanical threshold ∼12 h after incision performed in the absence of persistent inflammation (fig. 3B). Far more striking than the relatively small influence of anesthetic on the magnitude of mechanical hypersensitivity was the observation that the recovery of mechanical threshold was significantly delayed in inflamed rats with plantar incision anesthetized with pentobarbital than in either the ketamine/xylazine- or isoflurane-anesthetized groups (fig. 4A). Analysis of these recovery data with a mixed-design, two-way ANOVA revealed significant main effects associated with day of testing (P < 0.01) and anesthetic (P < 0.01), as well as a significant interaction (P = 0.01) between the two. Post hoc  analysis (Holm-Sidak) indicated that the mechanical threshold in the pentobarbital group was significantly lower than those in the other two groups by 3 days after surgery, and this difference persisted for the remainder of the testing period. When mechanical threshold data were analyzed as a function of the days necessary for recovery to 50% of baseline, the pentobarbital group took almost twice as long to achieve the same level of recovery as that observed in the other two groups; this difference was statistically significant (P = 0.02, Kruskal-Wallis one-way ANOVA, fig. 4B). In contrast, in the absence of inflammation, although there was still an influence of test day on the changes in mechanical threshold associated with plantar incision (P < 0.01, two-way ANOVA), there was no detectable influence of anesthetic (P > 0.05) on the rate of recovery of mechanical threshold when analyzed across testing days (fig. 4C), and there was no significant interaction between the two (P > 0.05). The absence of a difference between anesthetic groups was even more clearly illustrated when data were analyzed as time to 50% of baseline (fig. 4D, P > 0.05, Kruskal-Wallis one-way ANOVA).

The hind paw incision in the inflamed tissue may have been necessary for the manifestation of the deleterious impact of pentobarbital on the resolution of mechanical hypersensitivity. That is, the 3-h period of pentobarbital anesthesia may have slowed recovery of postoperative hypersensitivity. However, it is also possible that the presence of persistent inflammation was sufficient, and the deleterious impact of pentobarbital was on the time course of recovery of the inflammatory hypersensitivity. To test this possibility, we studied a group of rats with a protocol identical to that used with the inflammation + incision groups, except that there was no manipulation of the hind paw during the 3-h period of anesthesia. To enable data collection with the experimenter blinded to the experimental manipulation, this follow-up experiment included four naïve, six inflammation only, and eight inflammation + pentobarbital anesthesia rats. There were no statistically significant differences between the results from the 6 inflammation-only rats studied in this follow-up experiment and the 10 inflammation-only ones studied in the first experiment. Thus, data from these two groups of inflammation-only rats were pooled. Statistical analysis (mixed-design, two-way ANOVA) of the results obtained with the inflammation-only versus  inflammation + anesthesia groups revealed a significant (P < 0.01) influence of day and a significant influence of anesthetic (P = 0.02). However, there was no significant interaction between these two factors. Pooled data are plotted in figure 5A. When data were analyzed as a time to 50% recovery, recovery was significantly delayed in the inflammation + anesthesia group (P < 0.01, Mann–Whitney U test, fig. 5B).

The purpose of this study was to begin to test the hypothesis that in the presence of inflammation, GABA^A-preferring general anesthetics exacerbate the magnitude of and/or delay the recovery from pain observed after general anesthetic administration. The current experiments yield three primary observations directly relevant to this underlying hypothesis. First, in the absence of persistent inflammation, there was no detectable influence of anesthetic type on the magnitude of pain behavior and hypersensitivity associated with the hind paw incision or the time course of recovery. Second, in the presence of inflammation, although there was no influence of anesthetic type on the magnitude of postanesthesia pain behavior and only a small influence on hypersensitivity, the resolution of mechanical hypersensitivity was significantly delayed in rats anesthetized with pentobarbital compared with those anesthetized with either the combination of ketamine and xylazine or with isoflurane. Third, the deleterious impact of pentobarbital in the presence of persistent inflammation on the time course of recovery from hypersensitivity did not depend on the presence of the surgical manipulation.

Our study originally was designed to assess the interaction between the presence of persistent inflammation and general anesthetic type on postoperative pain. Our focus was on postoperative pain because general anesthesia is used clinically only to facilitate execution of an invasive procedure. Because our results indicate that an invasive procedure is not necessary for the manifestation of a deleterious impact of general anesthesia, we use the term “postanesthesia pain” to highlight this distinction. However, implicit in our original experimental design was the prediction that there would be no influence of anesthetic type on mechanical hypersensitivity associated with hind paw incision in the absence of persistent inflammation. We made this prediction being cognizant of the fact that hind paw incision initiates an inflammatory response and drives afferent activity.11Importantly, this prediction was based on the assumption that activation of “normal” inhibitory GABA^Aby pentobarbital before the hind paw incision would be sufficient to attenuate the emergence of excitatory GABA^Asignaling driven by the incision-induced activation of nociceptive afferents and associated inflammation. However, we are also aware that this a priori  assumption was an oversimplification because of evidence that excitatory GABA^A-dependent signaling in the spinal cord develops rapidly (minutes to hours) after nociceptor activation and/or acute inflammation even in the presence of GABA^A-preferring general anesthetics.12Our previous behavioral results indicating that, with spinal administration, the benzodiazepine midazolam retains its analgesic efficacy in the presence of persistent inflammation, despite the presence of GABA^Aantagonist-induced analgesia,4prompted us to hypothesize that a simple depolarizing shift in the chloride equilibrium potential was insufficient to explain everything associated with the emergence of excitatory GABA^Asignaling observed in the presence of persistent inflammation. Results of the current study suggest that these additional mechanisms take time to develop. Additional experiments will be needed to identify these additional mechanisms.

The three anesthetics used in the current study were chosen based on their putative mechanism of action. Although not commonly used for surgical procedures, pentobarbital was used based on evidence that its primary mechanism of action is the activation of GABA^A.13Furthermore, it is possible to obtain a surgical plane of anesthesia with pentobarbital without the need for artificial ventilation, which is required with more commonly used GABA^A-preferring general anesthetics, such as propofol or etomidate. The results of the current study suggest that these other anesthetics may also have a deleterious impact on postanesthesia pain, so it will be important to formally assess this possibility. A ketamine and xylazine combination was used as a second anesthetic based on evidence that these compounds do not directly engage GABA^Aas a primary mechanism of action. Ketamine is thought to act primarily at the N -methyl-D-aspartate subtype of the ionotropic glutamate receptor,13whereas xylazine is thought to act primarily at α²-adrenergic receptors.14The two were used in combination because it is not possible to obtain a surgical plane of anesthesia with ketamine alone in rodents.15,16Finally, isoflurane was used both because it is still widely used clinically and because it is thought to produce anesthesia via  actions at multiple sites, including the GABA^A, glycine receptor, and voltage-gated ion channels.13 

There are several possible explanations for our failure to detect an influence of general anesthetic type on the magnitude of postanesthetic pain behavior and the minimal influence of general anesthetic type on the magnitude of mechanical hypersensitivity. The most obvious interpretation is that there is minimal influence of anesthetic type on these two parameters. However, it is also possible that we failed to detect an influence of anesthetic type on these parameters because of the timing of our assessment relative to the surgery and/or the sensitivity of the tests used. That is, we used a delay of approximately 12 h between anesthesia and assessment to minimize a potential confounding influence of differences in the clearance time for the different anesthetics. Although we cannot rule out a confounding influence of residual anesthetic, we suggest that this confound is unlikely given that pentobarbital has the longest half-life17and thus should have influenced hypersensitivity in the incision-only group if residual drug was still influencing nociceptive behavior at the 12-h time point. On the other hand, if postoperative changes in behavior peak at earlier time points, they would have been missed. Guarding behavior as a measure of ongoing pain was sensitive enough to enable us to distinguish between inflamed and uninflamed groups when animals in all three anesthetic treatments were collapsed in the two-way ANOVA, but it is still a relatively crude measure. A more sensitive test, or at least one validated for use in the detection of ongoing pain, such as the recently described adaptation of the conditioned place preference test,18may be necessary to detect an influence on anesthetic type. Similarly, given that it is difficult to reliably detect a decrease in mechanical threshold below the level observed in the postoperative groups with the electronic von Frey device, a “floor effect” may have precluded the ability to detect a greater influence of anesthetic type on mechanical threshold. In addition, the surgical manipulation used was relatively minor and lasted for a very small fraction (∼5 min) of the time animals were anesthetized. Thus, although the results from the inflammation + anesthesia only group suggested that the relative contribution of hypersensitivity associated with the incision was relatively small in the presence of the CFA-induced hypersensitivity, a surgical intervention more comparable with that normally associated with the operating room, where tissue is manipulated for a considerably larger fraction of the total duration of anesthesia, may have yielded more robust changes in postoperative pain behavior and sensitivity. Thus, it is also possible we have underestimated the deleterious impact of general anesthetic type on the magnitude of postoperative pain per se , as opposed to postanesthetic pain when GABA^A-preferring general anesthetics are used in the presence of persistent inflammation.

As with the peak change in postanesthesia guarding behavior, the failure to detect an influence of anesthetic type on the resolution of this measure of ongoing pain may have been attributable to the sensitivity of the assay or there being no influence of anesthetic type on the time course of the resolution of ongoing pain observed after surgery. The latter possibility would be interesting in light of the significant influence of anesthetic type on the time course of the resolution of the mechanical hypersensitivity because it would suggest the underlying mechanisms influencing ongoing pain and hypersensitivity are distinct, at least in the presence of persistent inflammation.

The delay in the recovery of incision-induced mechanical hypersensitivity observed in the previously inflamed pentobarbital-anesthetized rats was consistent with our major hypothesis, and as discussed below, has significant clinical implications. However, there were at least two additional features of this change that warrant additional consideration. One of these was the observation that, at least for the isoflurane- and ketamine/xylazine-anesthetized animals, there did not appear to be an influence of persistent inflammation on the time course of recovery of mechanical hypersensitivity. In the face of evidence that there are differences in the underlying mechanisms of inflammatory and postoperative pain,11,19we were surprised that there was no evidence of an interaction in underlying processes that influenced the time course of recovery. That is, we initially predicted that all animals undergoing surgery in the presence of persistent inflammation would differ in the rate of recovery from animals that underwent the identical surgery in the absence of persistent inflammation, yet this was not observed. A second notable feature was the change in the slope of the recovery curve in the inflamed pentobarbital anesthetized animals. Recovery was not simply offset (i.e. , delayed by 3–4 days), as one might predict would result if the magnitude of the decrease in threshold was significantly greater (but masked because of a floor effect), in pentobarbital-anesthetized animals. The difference in the rate of recovery suggests that processes initiated with a 3-h period of anesthesia with pentobarbital are distinct from those initiated with the other anesthetics. One intriguing possibility is that the facilitated nociceptive signaling associated with the presence of persistent inflammation enables the GABA^A-preferring general anesthetics to engage a circuit comparable with that of the “hyperalgesic priming” described by Aley et al. : in animals “primed” with an inflammatory insult,20subsequent challenge results in a hypersensitivity that is dramatically prolonged in the absence of an increase in magnitude.

Although not addressed in the current study, it is worth considering, if only briefly, the underlying mechanisms that may contribute to the impact of persistent inflammation on the results obtained with different general anesthetics. If the differential anesthetic effects were attributable to preferential activation of pronociceptive GABA^Aby pentobarbital, the site of action could have been supraspinal, spinal, and/or in the periphery. Although γ-aminobutyric acid–mediated signaling plays a role in descending modulatory circuitry,21the bilateral nature of descending projections22combined with our failure to detect changes in the contralateral hind paw argues against a supraspinal site of action. We suggest a more likely site of action is the spinal cord, particularly in light of the well-described role of spinal GABA^Asignaling in the modulation nociceptive threshold. Furthermore, there are several lines of evidence suggesting that injury-induced alterations in GABA^Asignaling on the central terminals of nociceptive afferents are responsible for the deleterious impact of pentobarbital that was observed only in the presence of inflammation: (1) tissue injury and inflammation result in an increase in GABA^A-dependent action potentials generated at the central terminals of primary afferents (underlying the so-called dorsal root reflex)23; (2) tissue injury and inflammation result in the translocation of the chloride cotransporter NKCC1,24thought to be largely present in primary afferent, to the plasma membrane. This change is thought to underlie a depolarizing shift in the chloride equilibrium potential, which would account for the shift in GABA^Asignaling from inhibition to excitation; and (3) the relatively selective block of NKCC1 attenuates inflammatory hyperalgesia.25,26Thus, perhaps one way to mitigate the deleterious consequences of GABA^A-preferring general anesthetics would be to coadminister selective NKCC1 antagonists. The fact that no deleterious consequences were detected with isoflurane suggests it may also be possible to mitigate the impact of spinal GABA^Aactivation by coactivating other inhibitory receptors and/or blocking other excitatory ion channels. Finally, because there is evidence that GABA^Ais transported to the periphery in primary afferents,27a pronociceptive action of GABA signaling in the periphery cannot be discounted.

This study was designed to test a hypothesis focused on inflammation-induced changes in GABA^Asignaling and has been interpreted within this context. However, there are at least three other factors that may have contributed to our results. First, it is possible that ketamine/xylazine and isoflurane had persistent antinociceptive effects, rather than our hypothesized pronociceptive actions of pentobarbital. In contrast to pentobarbital, both ketamine/xylazine and isoflurane can block N -methyl-D-aspartate receptors,13and the contribution of this receptor subtype to both the initiation and maintenance of injury-induced hyperalgesia has been well documented.28Thus, with an increase in N -methyl-D-aspartate receptor activation in the presence of persistent inflammation, the antinociceptive actions of these anesthetics could have been magnified. However, we suggest that such an antinociceptive effect is unlikely both because of the distinct pharmacologic profile of ketamine/xylazine and isoflurane and because of recent evidence suggesting that at least two of the additional targets for isoflurane beyond its relatively weak activity at N -methyl-D-aspartate receptors may underlie pronociceptive effects of this compound, at least in the short term.29The comparable time course of recovery of the mechanical threshold observed in all groups except those with persistent inflammation anesthetized with pentobarbital, which includes a group of animals not exposed to any anesthetic, also argues against an antinociceptive effect of ketamine/xylazine and isoflurane. Second, it is possible that there is a differential influence of general anesthetics on the inflammatory response per se  (i.e. , immune cell activation, changes in blood flow, plasma extravasation and such), which would have been manifest as changes in the time course of recovery of mechanical threshold. Third, it is also possible that because each anesthetic has relatively distinct mechanisms of action, a differential impact on physiologic parameters, such as body temperature, heart rate, and vessel diameter, could have had a secondary impact on the ongoing inflammation and associated mechanical hypersensitivity. Although we did not detect any gross differences among the anesthetics in their impact on respiratory rate and heart rate, and body temperature was maintained during the period of anesthesia with a heating blanket, a differential influence on the inflammatory response and/or physiologic parameters cannot be discounted. However, neither of these latter two possibilities diminishes the clinical implications of our observations.

Because a significant fraction of surgeries performed under general anesthesia are to address problems associated with persistent inflammation, our observations suggest that decisions made in the operating room may have profound implications for patient health and well-being after surgery. Furthermore, any increase in postoperative pain can have profound repercussions on the cost of healthcare delivery. Because surgeries are also performed on patients to address problems not directly associated with sites of persistent inflammation, such as a cholecystectomy on a patient with rheumatoid arthritis, the observation that GABA^A-preferring general anesthetics may exacerbate ongoing inflammatory pain and hypersensitivity in the absence of additional surgical manipulations suggests the patient population affected by an inflammation-induced change in GABA^Asignaling may be quite large. This may be true even if the deleterious impact of GABA^A-preferring general anesthetics is restricted to recovery from mechanical hypersensitivity, as suggested by the growing list of situations in which therapeutic efficacy depends on activity (e.g. , ambulation, physical therapy, coughing). Persistent mechanical hypersensitivity may be particularly problematic for older patients and those with risk factors for stroke, patients for whom inactivity makes them especially susceptible to unfavorable outcomes. Ongoing studies are designed to assess the extent to which the implications of our preclinical studies are manifest in clinical settings.