A ¹⁸F-fluorodeoxyglucose MicroPET Imaging Study to Assess Changes in Brain Glucose Metabolism in a Rat Model of Surgery-induced Latent Pain Sensitization

• Surgery and opioids exposure can independently and together cause hypersensitivity to light touch in animals, but their effect on brain activation has not been studied

• In rats, surgical paw incision and remifentanil resulted in enhanced brain activity in many regions associated with pain for up to 21 days

• Administration of naloxone 21 days after surgery and opioid exposure produced hypersensitivity and brain activation

SEVERAL studies show that a previous pain experience can increase pain sensitivity, which is exacerbated by opioid administration.1,2In a mouse model of postsurgical pain, we have shown that the intraoperative administration of short-acting opioids enhances the extent and duration of postoperative pain sensitization.3In addition, a second surgery performed 21 days later, after healing of the wound and when nociceptive behavior returned to baseline, induced greater postoperative hyperalgesia.4In the same model, the administration 21 or more days after surgery of (−)-naloxone, but not (+)-naloxone, induced significant hypersensitivity, suggesting latent pain sensitization.5These and other reports6,7suggest that plastic changes in somatosensory pathways (from primary afferents to multiple brain areas) may participate in the brain-adaptation mechanisms involved in latent pain sensitization.

During the last few years, functional imaging techniques, such as functional magnetic resonance imaging and positron emission tomography (PET), have established those supraspinal areas related to pain perception often referred as pain matrix. The techniques allow great temporal and spatial resolution both in healthy subjects and under pathologic conditions.8These areas portray the lateral pain system, which includes the thalamus and the primary and secondary somatosensory cortices, components involved in the sensory-discriminative nociceptive transmission. The medial pain system, which is mainly involved in the affective emotional/cognitive aspects of pain, contains the anterior cingulate, the insular and the prefrontal cortices, and the hippocampus.9Other structures such as the brainstem, amygdalae, basal ganglia, and cerebellum could also be involved in the processing of nociceptive information and pain perception, although their precise role remains unclear.10In animals, the areas of the brain participating in pain transmission can be investigated in vivo  using microPET.11Systemic administration of a radiotracer such as ¹⁸F-fluorodeoxyglucose (¹⁸F-FDG) allows quantification of regional changes in glucose metabolism that reflect neuronal activity in vivo . Voxel-by-voxel subtraction of images are used to assess the changes in ¹⁸F-FDG uptake for different brain areas.

The present study was conducted to further contribute to the understanding of long-lasting complex events occurring in the brain after surgery, and the mechanisms of latent pain sensitization in rats. We assessed temporal changes in glucose brain metabolism induced by surgery under remifentanil anesthetic and/or the administration of naloxone using a rodent microPET R4 scanner (Concorde Microsystems, Siemens, Knoxville, TN) and ¹⁸F-FDG as a radiotracer.

Forty adult male Sprague-Dawley rats weighing 180–200 g were used. The experiments were performed according to the Ethical Guidelines of the International Association for the Study of Pain, and the Ethical Committee for Animal Welfare of our institution (ComitéÉtico de Experimentación Animal–Parc de Recerca Biomèdica de Barcelona, Barcelona, Spain) approved the protocol. Animal Facilities (located in the Parc de Recerca Biomèdica de Barcelona), has been accredited by the Association for Assessment and Accreditation of Laboratory Animal Care since June 2010. Animals were housed two per cage, with autoclaved poplar soft wood bedding (Souralit S.L., Barcelona, Spain), and had free access to food and water. They were maintained in a controlled temperature (22 ± 1°C and 60% relative humidity) and light (12-hour dark-light cycle with lights on at 8:00 AM). Rats were acclimatized to the facilities at least for 2 days before beginning the experiment. Behavioral testing was performed between 9:00 AM and 2:00 PM, in a quiet room.

We used the incisional postoperative pain model previously described.12Animals were anesthetized with sevoflurane with a nose mask (induction, 3.5% v/v; surgery, 3.0% v/v) plus an infusion of intravenous remifentanil (200 μg/kg in a volume of 2.5 ml/h), administered during a period of 30 min. A KD Scientific pump (KD Scientific Inc., Holliston, MA) was used for the infusion. After disinfection of the skin with povidone, a 1-cm longitudinal incision was made with a No. 20 blade through the skin and fascia of the plantar surface of the right hind paw, starting 0.5 cm from the proximal edge of the heel extending toward the toes. The underlying plantaris muscle was exposed and incised longitudinally, but the muscle origin and insertion remained intact. After hemostasis with slight pressure, the skin was stitched with two sutures of 6.0 silk, and the wound site was covered with povidone-iodine antiseptic ointment. After surgery, the animals were allowed to recover under a heat source in cages with sterile bedding. No cardiovascular side effects or respiratory depression were observed after the animals recovered.

Allodynia to punctate mechanical stimuli, what we refer to in this text as hypersensitivity, was assessed as the frequency of foot withdrawal elicited by a mechanical stimulus, according to the method previously described,13and served as a measure of nociception. Based on previous studies of our group, the final testing point was established at 21 days postsurgery.4,5Animals were placed in methacrylate transparent plexiglass chambers (30 × 30 × 30 cm), with a wide grid bottom through which the von Frey filaments of increasing force (2, 4, 6, 8, and 10 g; North Coast Medical Inc., San Jose, CA) were applied. Mechanical thresholds were tested by sequential series of 10 tactile stimulations applied to the plantar surface of both hind paws, excluding the toes and the heel. Each filament was applied 10 times alternatively to each hind paw for approximately 6 s. We considered that paw withdrawal, vigorous shaking, or biting of the paw were indicative of hypersensitivity; the number of responses for each individual filament were recorded. Stimulation with von Frey was carried out at different time points (table 1). For each hind paw, percent threshold changes were calculated as follows: [(postoperative value × 100)/baseline value)]− 100.

Rats received 37 MBq of ¹⁸F-FDG in 0.5 ml of physiologic saline administered subcutaneously, and 45 min later were placed in the PET scanner under light sevoflurane anesthetic for image acquisition. Data were collected for a period of 30 min in a rodent microPET. The data were corrected for detector nonuniformity, random coincidences, and radionuclide decay, but not for photon scatter or attenuation, and reconstructed into a matrix size of 128 × 128 × 63 and a voxel size of 0.85 × 0.85 × 1.21 mm by a 3D ordered-subset expectation maximization followed by maximum a posteriori reconstruction. This algorithm and process yielded a spatial resolution of approximately 1.5 mm full width at half-maximum.

Before performing the current imaging study, we evaluated in a separate group of 30 animals the pronociceptive effects of surgery (fig. 1A), remifentanil (fig. 1B), and the combination of both (fig. 1C).

The present imaging study was designed to identify changes in brain glucose metabolism in rats that underwent surgery under remifentanil anesthetic, and received 1 mg/kg (−)-naloxone 21 days later (group 1, fig. 2, fig. 3, and fig. 4, n = 8). We also performed two additional groups of experiments (n = 8 each): in group 2, rats underwent surgery performed under sevoflurane anesthetic (table 1and figs. 2A and 5A, n = 8), and in group 3, animals received an infusion of remifentanil under inhalational anesthetic (table 1, figs. 2A and 5B, n = 8). Both groups also received 1 mg/kg of subcutaneously administered (−)-naloxone on postoperative day 21.

In groups 1, 2, and 3 (n = 8 each), we performed a complete longitudinal study, evaluating both mechanical hypersensitivity with the von Frey filaments and brain glucose metabolic activity by means of ¹⁸F-FDG. In these experiments (table 1), rats in group 1 received eight consecutive subcutaneous injections of ¹⁸F-FDG. Animals in groups 2 and 3 also received eight injections (table 1), but only two of them were with radiotracer ¹⁸F-FDG. These were given on days −1 (baseline state) and 21 (naloxone administration), when microPET scans were obtained.

In all instances, the glucose uptake period was of 45 min after the injection. Nociceptive stimulation with von Frey filaments was performed 5 min after ¹⁸F-FDG injection on days −1, 2, 7, and 21, each stimulation lasting 30 min. In group 1 (table 1), images were also acquired the day before, without von Frey stimulation (days −2, 1, 6, and 20). On day 19, animals were also tested with von Frey filaments, but were not scanned. On days 20 and 21, rats subcutaneously received 1 mg/kg (−)-naloxone, mixed in the same syringe with the ¹⁸F-FDG (table 1).

In addition, two other experimental conditions were evaluated: in group 4 (n = 8, effect of surgery plus remifentanil), rats underwent surgery under remifentanil anesthetic, and were stimulated with von Frey filaments at the same time points as rats in groups 1, 2, and 3 (table 1). Mechanical thresholds were also tested on day 19 to assess the recovery to baseline values, and then on day 21 after receiving a subcutaneous injection of the same volume of saline (instead of naloxone) (fig. 2B). Animals received eight injections (table 1), but only two of them were of radiotracer ¹⁸F-FDG. Brain glucose metabolism was assessed on days −1 (baseline state) and 21, when microPET scans were obtained (fig. 5C). Rats in group 5 (n = 8, effect of naloxone) were anesthetized with sevoflurane but had no surgery or remifentanil (sham-operated), and were stimulated with the von Frey filaments at the same time points as rats in groups 1, 2, 3, and 4 (fig. 2B). On day 21, they received 1 mg/kg of subcutaneously administered naloxone together with the radiotracer and were scanned with the microPET (table 1).

Sevoflurane (Sevorane^®; Abbot Laboratories SA, Madrid, Spain), remifentanil (Ultiva^®; GlaxoSmithKline, Madrid, Spain) and (−)-naloxone hydrochloride (Naloxone; Kern Pharma, Barcelona, Spain) were supplied by the Department of Anesthesiology of the Hospital del Mar (Barcelona, Spain). Remifentanil (0.2 mg/kg14) was dissolved in saline (NaCl 0.9%) and intravenously infused for 30 min (rate of 2.5 ml/h), using a Harvard Apparatus pump (Biosis S.L., Biologic Systems, Barcelona, Spain). ¹⁸F-FDG (37 MBq per animal), was subcutaneously injected without anesthetic. Lastly, (−)-naloxone (1 mg/kg) was dissolved in saline and subcutaneously injected, together with the radiotracer.

For each rat and von Frey filament tested, the responses in grams are expressed as the percent changes (% mean ± SD) with respect to the baseline values, represented in the figures by a horizontal line. Negative values indicate a net decrease in mechanical thresholds, or hypersensitivity. Mean values ± SD for the time-effects curves (area above the curves), corresponding to each experimental condition, were calculated. Because the areas above the curves data are reduced to a single value per mouse, only 10 data per group were used. For this reason, the overall comparison was carried out by means of the nonparametric Kruskal-Wallis test for independent samples. For the post hoc  pairwise comparisons, the Wilcoxon test was used together with a Bonferroni correction in order to guarantee the 95% confidence level. To analyze changes in the mechanical thresholds in the different groups of study, one-way ANOVA models for repeated measures, one for each group, were used. For the post hoc  analyses comparing the values at each time with the baseline values, Dunnett's many-to-one test in the framework of these models was applied. In addition, for the analyses of the data at days 2, 7, 19, and 21, a two-way ANOVA model for repeated measures was applied, including as factors the experimental condition, the day, and the interaction of both. Within the framework of this model, pairwise comparisons of the experimental conditions were carried out using the Tukey test. Values of days 2 and 21 were compared for each of the three experimental conditions. Analyses were performed with SPSS version 12.0 (SPSS Inc., Chicago, IL), and R (The R Foundation for Statistical Computing, Vienna, Austria), using its libraries nlme15and multcomp.16 

A synthetic FDG image was created based on the expected uptake of ¹⁸F-FDG in the respective brain areas by the method previously described.17This synthetic template was smoothed with a Gaussian filter with a full width at half-maximum, similar to that of the reconstructed images, which were then coregistered to the template using an affine transformation by the method implemented in the Statistical Parametric Mapping SPM8 software (SPM Welcome Department of Imaging Neuroscience, Institute of Neurology, UCL, London, United Kingdom). Global cerebral ¹⁸F-FDG uptake was calculated as standardized uptake values and statistically compared among experimental groups (P < 0.05). Since no statistical differences were found, images were intensity normalized by proportional scaling, and then voxelwise. After global normalization by proportional scaling, voxelwise regional between-group differences were assessed with the SPM8 software, setting up a one-way repeated measures ANOVA design and a threshold for statistical significance of P < 0.001 (uncorrected for multiple comparisons). To improve localization of the significant activations, SPM maps were overlaid onto an anatomical brain template. Our selection of a statistical threshold of P < 0.001, uncorrected for multiple comparisons (a fairly common practice in prospective studies) is based on the lack of validation with rodent brain data of multiple comparison correction methods, such as the familywise error rate or the false discovery rate.

Before performing the present imaging study, we evaluated in a different group of animals the pronociceptive effects induced by surgery (fig. 1A), remifentanil (fig. 1B), and the combination of both (fig. 1C) (n = 30). Significant postoperative mechanical hypersensitivity in the operated paw was observed in all groups. Hypersensitivity was maximal between 4 h and 2 days after surgery and lasted approximately up to 4 days; differences in the changes in mechanical thresholds between 4 h, 1 day, and 2 days were not significantly different. The mean areas above the time-effect curves (0–21 days) of the operated paw, obtained in the different experimental conditions, showed that the pronociceptive effects induced by the surgical procedure performed under remifentanil anesthetic were of greater magnitude (140 ± 5.8, P < 0.05) than those of surgery or remifentanil individually (80 ± 5.2 and 70 ± 4.3, respectively).

For the mPET study, rats from all experimental groups (table 1) were assessed with von Frey filaments on days 2, 7, 19, and 21 after manipulation (fig. 2). On day 2, rats from the three central groups (group 1 = surgery and remifentanil, group 2 = surgery, group 3 = remifentanil, fig. 2A) showed significant hypersensitivity that was more prominent in group 1 (P < 0.01, P < 0.05 when group 1 compared with groups 2 and 3, respectively); after complete recovery of mechanical thresholds and healing of the surgical wound (days 7 and 19), nociceptive thresholds were comparable to baseline in all groups. The subcutaneous administration of 1 mg/kg (−)-naloxone on day 21 induced a similar degree of hypersensitivity than in day 2, a finding that has been previously reported by our group in mice.5 

Figure 2shows the percent changes in mechanical thresholds in all study groups. Animals in group 1 (surgery and remifentanil) showed a significant larger degree of hypersensitivity on postoperative day 2, compared with groups 2 (surgery alone, P < 0.05) and 3 (remifentanil infusion, P < 0.01); actual changes in mechanical thresholds were −37.0 ± 5.0, −20.7 ± 7.0%, and −17.9 ± 3.7%, respectively, when compared with basal values (P < 0.001 for each group). No differences were observed between the surgery and remifentanil groups (P = 0.717).

Animals in group 4 (surgery and remifentanil) showed significant hypersensitivity (−35.8 ± 3.5%, fig. 2B) on day 2, similar to the magnitude of group 1. No hypersensitivity was present in sham-operated rats (group 5, −2.9 ± 5.4%).

On day 21, the administration of (−)-naloxone to groups 1, 2, and 3 precipitated significant hypersensitivity (−27.2 ± 2.4, −19.0 ± 3.8 and −17.2 ± 5.1, respectively, P < 0.001 when compared with basal values). The magnitude of the effect was not significantly different among themselves (P > 0.3), but it was analogous to day 2 in group 1.

The same dose of naloxone administered to rats in group 5 (sham-operated) did not significantly change nociceptive thresholds (−4.53 ± 4.8%). Similarly, the subcutaneous injection of saline 21 days after surgery performed under remifentanil anesthetic (group 4, fig. 2B) did not induce significant changes in mechanical thresholds (−5.90 ± 3.68%).

Rats in group 1 received eight subcutaneous injections of ¹⁸F-FDG and 5 min later were stimulated with von Frey filaments for a period of 30 min (table 1), also recording the extent of mechanical hypersensitivity (fig. 2A). Brain images were acquired 45 min after ¹⁸F-FDG injection.

In baseline conditions (on days −2 and −1), von Frey stimulation did not induce significant changes in glucose uptake in brain areas related to the pain matrix (fig. 3). Similarly, on day 1 after surgery, no changes were observed in nonstimulated conditions (fig. 3). On day 2, after von Frey stimulation, we observed a slight increase in ¹⁸F-FDG uptake in the right hemisphere of the entorhinal cortex, the hippocampus, both colliculi (superior and inferior), and the cerebellum (fig. 4A). On postoperative days 6 and 7, when nociceptive thresholds had returned to baseline values (fig. 2A), the magnitude of changes in the glucose uptake was further increased in the same areas of the brain, as well as in the hypothalamus, thalamus, and brainstem (fig. 4B); no differences between nonstimulated (day 6) and stimulated conditions were observed (day 7) (fig. 3).

In the same animals (group 1), (−)-naloxone administration in the absence or presence of von Frey stimulation (days 20 and 21, fig. 3) greatly increased ¹⁸F-FDG uptake in the hypothalamus, cerebellum, entorhinal cortex, and brainstem, with substantial decreases in uptake in the ipsi-dorsolateral cortex, hippocampus, thalamus, and anterior cingulate cortex (fig. 4C).

For these experiments, animals were similarly handled as in group 1 (table 1), including sequential von Frey stimulation at the same time points, but only received two injections of radiotracer ¹⁸F-FDG, on days −1 (baseline state) and 21, when microPET scans were obtained (see Materials and Methods). The effects of naloxone administered 21 days after surgery (group 2) or remifentanil (group 3) are represented in figure 5(panels A and B, respectively). In both groups, statistical parametric imaging analysis detected brain areas significantly activated (the basal striatum and entorhinal cortex) and deactivated (the prefrontal and anterior cingulate cortices, and the thalamus).

For these experiments, animals were similarly handled as in group 1, 2, and 3 (table 1), including sequential von Frey stimulation at the same time points, but only received two injections of ¹⁸F-FDG on days −1 (baseline state) and 21, when microPET scans were obtained (see Materials and Methods). A comparison of baseline values of ¹⁸F-FDG uptake obtained on days −1 and 21 in this group is shown in figure 5C. Twenty-one days after the administration of subcutaneous saline to animals that had surgery with remifentanil (group 4), we could observe an increase in ¹⁸F-FDG uptake in the entorhinal cortex and cerebellum, results that are consistent with those obtained in group 1, 7 days after surgery under remifentanil anesthetic (fig. 4B). Moreover, significant decreases in metabolic activity were also detected in the brainstem, which includes structures from mesencephalon, pons, and medulla.

In sham-operated animals (group 5), the administration of naloxone 21 days after manipulation slightly increased radiotracer uptake in small areas of the brain, mainly the striate nucleus and the cerebellum (image data not shown). The changes were of smaller magnitude than those observed in animals that had surgery under remifentanil anesthetic.

Using the ¹⁸F-FDG microPET, we performed a longitudinal study to investigate possible changes in brain glucose metabolism occurring in the postoperative period in rats, likely related to the development of latent pain sensitization.

Surgery performed under remifentanil anesthetic induced slight changes in brain ¹⁸F-FDG uptake in the cerebellum on days 1 and 2 after surgery, when pain and hypersensitivity were more severe; the small magnitude of the changes could be related to the limited intensity of injury (incision) or the time of image acquisition. Other groups using formalin,18,–,20capsaicin,21,22zymosan,23or electrical24and mechanical stimuli,25have shown changes in pain-processing brain areas. In these studies, variations in brain activity were observed a few hours after the nociceptive stimulus, whereas we assessed brain activity 1 and 2 days after surgery, because of technical limitations of ¹⁸F-FDG, which has a half-life of 110 min. On day 7 (fig. 4B), larger metabolic changes were observed in the same areas, further extending over the hypothalamus, thalamus, and ventral area of the brainstem. Smaller changes were still present 21 days postoperatively (fig. 5C). These results suggest that changes in glucose uptake over time could reflect the slow development of surgery-induced neuroplastic adaptative modifications, indicative of latent pain sensitization.

The administration of naloxone and remifentanil 21 days after surgery induced similar nociceptive hypersensitivity to that of day 2, supporting previous studies.5,6,26,– 28Analogous naloxone-induced hypersensitivity was observed in the surgery and the remifentanil groups; the scans obtained after naloxone revealed smaller, but significant, changes than the combination in related areas of the brain.

In the surgery and remifentanil group, naloxone provoked increased glucose metabolism in the hypothalamus, a brain region that regulates affective, endocrine, and sympathetic responses29; since the hypothalamus contributes to the descending pain modulatory pathways,30the results suggest that could be implicated in nociceptive inhibitory modulation. In addition, enhanced ¹⁸F-FDG uptake was observed in the right cortex, which receives inputs from the visual, auditory, and somatosensory systems and also participates in pain perception.31Increases were also observed in both superior and inferior colliculi, which receive visual and auditory sensory inputs and participate in the integration of antinociceptive behavior.32There were increases in the entorhinal cortex involved in sensory memory, a relay between the cortices and hippocampus that receives serotoninergic, dopaminergic, and adrenergic inputs.33Nuclei of the brainstem, such as the medulla oblongata and the pons, which participate in the descending modulation of nociception and in opioid analgesia, were significantly activated.34Increased metabolic activity after naloxone was also observed in the cerebellum, an area that integrates motor responses35but also contains a high density of opioid receptors and could be implicated in nociceptive processing.36Several imaging studies describe the activation of the cerebellum during acute and chronic pain states and its response to aversive stimuli, indicating that may contain specific regions involved in encoding generalized aversive processing.37 

Significant reductions in glucose uptake were observed in the medial pain system, such as the hippocampus, a component of the limbic system participating in memory, temporal orientation and long-term potentiation.38Deactivations of metabolic activity were observed in the anterior cingulate cortex, a region commonly activated in pain studies in humans,39and in the thalamus, where responses correlate with the nociceptive responses in the cingulate cortex.40The thalamus filters sensory information before reaching the somatosensory cortex, and functional imaging has shown that the anterior cingulate cortex modulates nociception. It has also been reported that initial neuronal synaptic changes, or short-term plasticity, in the thalamo-cingulate pathway may facilitate how acute nociceptive responses become persistent.41Our results suggest that modifications in metabolic neuronal activity in the thalamus and cingulate cortex may reflect changes in the integration of nociceptive signals.

The effects of surgery on brain activity were assessed on day 21 after naloxone administration. We did not investigate the immediate or early effects of surgery as elegantly reported in humans by Pogatzki-Zahn et al.  42Actually, even at different times and experimental conditions, activation and deactivation of the different areas of the brain in both studies were similar. The results suggest that the surgery itself induces changes in the pain matrix resulting in long-lasting neuroplastic changes contributing to postoperative latent pain sensitization.

The effects of remifentanil on brain activity have been investigated in animals43and humans44,– 46using PET and functional magnetic resonance imaging. Remifentanil infusions induced dose-related analgesia correlating with changes in activity in the pain matrix, such as decreases in the thalamus, insular cortex, and anterior and posterior cingulate cortex, with increases in the periaqueductal gray and the cingulo-frontal cortex; generally images were acquired during or shortly after opioid administration. These areas are roughly the same as those activated or deactivated in our report; the similarities suggest that a single exposure to remifentanil induces long-lasting neuroplastic changes and may also contribute to postoperative latent pain sensitization.

We compared the areas of the brain activated or deactivated after naloxone in the different groups, observing slight differences in the nucleus involved and in the magnitude. All changes in the surgery or remifentanil groups were also present in the surgery and remifentanil group, except a bilateral hyperactivation of basal striatum area, thus suggesting that the differences are not related to the effects of naloxone. Surgery alone or combined with remifentanil induced an increase in cerebellum activity, which was not observed in the remifentanil group. Since the cerebellum is implicated in aversive adaptations, we suggest that surgery could be more aversive that remifentanil. A hypoactivation of the prefrontal cortex was present in the three groups, but to a different extent; the entorhinal cortex was bilaterally hyperactivated in all groups, but changes were larger in the surgery and remifentanil group, though surgery induced changes slightly broader than remifentanil.

Baseline stimulation with von Frey filaments did not change radiotracer uptake, but these results have limited external validity because we used a single mechanical stimulation modality. Other studies using functional magnetic resonance in rats47reported changes in brain activity after the application of nociceptive punctate mechanical stimulus, suggesting a direct relationship between stimulus intensity and changes in brain activity. Von Frey filaments applied with a force of 15.1 g induced a noxious paw withdrawal in awake rats13and firing of spinal and thalamic sensory neurons in anesthetized animals.48In our study, we applied a maximum force of 10 g to the paw, stimulus that do not activate nociceptive fibers and did not induce changes in brain glucose metabolism; however, when the same stimulus was applied in the early postoperative period (day 2), significant hypersensitivity was observed without changes in brain glucose metabolism. Thus, no correlation between mechanical hypersensitivity and brain image was present on days 2 and 7; however, on postoperative day 21, naloxone significantly increased ¹⁸F-FDG uptake and evoked mechanical hypersensitivity, showing a positive correlation between both variables.

There is controversy regarding the effects of naloxone on neuronal activity identified by brain imaging. In rats, no changes in blood oxygen level-dependent signal intensity were observed after systemic naloxone18; however, in healthy volunteers, functional magnetic resonance imaging showed that naloxone induced significant changes in cortical and subcortical areas of the lateral and medial pain systems49; also, non-nociceptive stimuli failed to alter baseline images, whereas painful heat increased the global signal intensity in the same areas. In our report, naloxone administration to sham-operated rats (data not shown) induced slight changes in the striatum and the cerebellum, consistent with those previously reported,49also supporting that innocuous stimulus with von Frey filaments does not evoke changes in brain activity. The slight effect of naloxone alone on brain activity in our study could also be related to the stress induced by manipulation of the rats.

The pain-related areas where changes in glucose metabolism occur express a high density of opioid receptors, suggesting an important role for the endogenous opioid system in latent pain sensitization after surgery. Studies of opioid-receptor binding, using image techniques in humans, show the involvement of the anterior cingulated, insular cortices and the thalamus in the endogenous opioidergic inhibition of tonic acute pain, induced by heat or capsaicin.50,51Moreover, after topical capsaicin in volunteers, naloxone increased pain perception, suggesting that acute pain is actively suppressed by endogenous opioid-receptor activation.52Thus, our results suggest an up-regulation of the endogenous opioid system as a compensatory mechanism to maintain pain-free conditions in the postoperative period in rats.

In conclusion, the administration of naloxone on postoperative day 21 to animals that had surgery under remifentanil anesthetic and were completely recovered from surgery induced mechanical hypersensitivity and profound changes in glucose metabolism in areas of the brain related to pain pathways. Less significant changes in similar brain areas were observed after a remifentanil infusion or when the surgery was performed under sevoflurane. The results suggest that the surgical injury under remifentanil anesthetic induces long-lasting neuroplastic adaptations in opioid circuits, or their projections, involved in nociceptive processing and latent pain sensitization. The slow reversibility of these neuroplastic changes, or memory of pain, may contribute to the development of chronic postsurgical pain.

The authors thank Marta Pulido, M.D. (Institut Municipal d'Investigació Mèdica, Parc de Salut Mar-Menamiri, Barcelona, Spain), for editing the manuscript and editorial assistance, and Klaus Langohr, Ph.D. (Statistician, Grup de Recerca Clínica en Farmacologia Humana i Neurociències, Programa de Recerca en Neurociències, Institut Municipal d'Investigació Mèdica, Parc de Salut Mar-Menarini, Barcelona, Spain), for his assistance in the statistical analysis.