Nociception alters sensory processing via peripheral and central mechanisms [1,2]. Animal models of central sensory change after nociception demonstrate excitation as well as inhibition [3,4]. Spinal excitation depends on activation of dorsal horn N-methyl-D-aspartate (NMDA) receptors by excitatory amino acid transmitters . Opioids depress spinal excitation by inhibiting the initial wide dynamic range (WDR) dorsal horn neuron response to incoming nociceptive C-fiber volleys, without directly affecting neuron hyperexcitability or wind-up . NMDA receptor blockers directly inhibit wind-up and hyperexcitability in WDR dorsal horn neurons without affecting the initial WDR response to incoming C-fiber volleys .
Central inhibition has been described in at least two variants, stress-induced analgesia (SIA) or diffuse noxious inhibitory controls (DNIC) [4,6-8]. Central inhibition is generalized, opposes spinal hyperexcitability, and involves descending control originating from supraspinal structures, such as the midline periaqueductal grey and the locus coereolus, via either the spinal dorsolateral funiculus or more diffuse propriospinal connections [4,6-8]. Longer-lasting SIA comprises neural and humoral opioid, monoaminergic, and nonopioid mechanisms, with the latter including NMDA receptors . In animals, opioid agonists augment opioid SIA induced by moderate stressors (e.g., warm water swim). NMDA receptor antagonists decrease nonopioid SIA associated with severe stressors (e.g., cold water swim) [4,6]. DNIC acts on convergent neurons, fades with the conditioning stimulus, and is thus shorter-lasting [7,8]. It is inhibited by morphine; interactions with NMDA antagonists have not been described [7,8].
According to animal studies, opioid agonists and NMDA antagonists may thus exhibit similar effects on spinal sensory excitation but opposing effects on aspects of central sensory inhibition [3-8] after nociception. In a recent human study  of sensory changes after surgery, adjuvant opioids suppressed spinal excitation and augmented central inhibition compared with isoflurane/nitrous oxide/oxygen anesthesia alone, as predicted by animal models. The effects on postsurgical sensory changes of substances acting on the NMDA as opposed to opioid receptor systems have not been studied.
We aimed to study the effect of a micro-opioid agonist (fentanyl) or two different NMDA antagonists (ketamine, magnesium) as adjuvants to general anesthesia on the state of sensory processing after surgical nociception. A major goal was to investigate whether these substances have similar or opposing effects on postsurgical central inhibition compared with spinal excitation.
After local institutional review board/ethics committee approval and informed patient consent, 45 ASA physical status I or II patients undergoing elective abdominal hysterectomy via Pfannenstiel incision were prospectively randomized to receive either fentanyl, ketamine, or magnesium (n = 15 per group) anesthetic supplementation. Statistical power calculations (alpha = 5%, beta = 10%) based on a previous study  suggested that a group size of 15 should detect differences in pain tolerance thresholds of one-third, and in 24-h morphine consumption of one-half. Exclusion criteria included systemic hypertension, epilepsy, chronic magnesium, hypnotic or analgesic use, and diseases predisposing to altered sensation (e.g., diabetes mellitus, neuropathies).
During the anesthesia interview, patients were instructed about threshold measurement, pain intensity verbal rating scores (VRS; 0 = no pain, 10 = worst pain imaginable) and use of a patient-controlled analgesia (PCA) pump. No premedication was given on the morning before the operation. In the induction room, before the insertion of venous catheters, baseline thresholds were measured by an observer who, like the patient, was blinded to the adjuvant drug used. The same observer performed all measures. Anesthesia was conducted unblinded by an anesthetist not involved in the study or postoperative patient care.
Thresholds were measured using electric constant current skin stimulation (Digistim[registered sign]; Biometer A/S, Copenhagen, Denmark; tetanic stimulation at 100 Hz, 0.2-ms square waves, self-adhesive electrodes 3 cm apart) on the dominant upper arm (C7 dermatome), the lateral breast fold (T4), 10 cm lateral to the Pfannenstiel incision (T12), and just above the patella (L3). We avoided stimulating major nerves or muscle bundles. The three thresholds-sensation (stimulation just felt), pain detection (stimulation just becomes painful; "first pain" via A delta-fibers), and pain tolerance (painfulness of stimulation just becomes intolerable; "second pain" via C-fibers) -were measured as the average of three serial assessments within 30 min, separated by at least 5 min.
Three minutes before anesthesia induction, patients received either 1.5 micro g/kg fentanyl, 0.5 mg/kg ketamine, or 20 mg/kg magnesium sulfate as a slow (60 s) intravenous (IV) injection. Anesthesia was induced with 5 mg/kg of thiopental, followed by 0.1 mg/kg vecuronium IV. After tracheal intubation, anesthesia was maintained with isoflurane in oxygen/nitrous oxide (1:2). Five minutes before skin incision, either 0.75 micro g/kg fentanyl, 0.25 mg/kg ketamine, or 10 mg/kg magnesium sulfate was injected and subsequently repeated at 30-min intervals. The final dose was given approximately 45 min before the end of surgery. Dropout was for operations lasting longer than 2 h or for unsatisfactory anesthesia (hemodynamic values >20% of baseline for >5 min).
Morphine PCA was started 30 min postextubation in the recovery room (loading bolus 40 micro g/kg, PCA bolus 25 micro g/kg; lockout 5 min, background infusion 15 micro g [center dot] kg-1 [center dot] h-1). Threshold measures, pain VRS, cumulative morphine consumption, and an observer sedation rating score (1 = unrousable, 2 = deeply sedated, 3 = moderate sedation, 4 = minor sedation, 5 = wide awake) were obtained at 1, 4, and 24 h postextubation. PCA morphine was discontinued 24 h postoperatively, and analgesia on the ward continued with per os diclofenac. Threshold and pain VRS values were reassessed 5 days postoperatively.
Apart from absolute sensory thresholds, which predominantly reflect generalized sensory inhibition, we also analyzed derived relative (or normalized) thresholds to unmask the weaker segmental effects expected from spinal excitation, as described previously . Relative thresholds were calculated by dividing thoracic, incision, or leg threshold values by respective arm threshold values. We chose the arm (C7 dermatome), far from the operation site, and thus was unlikely to be affected by lower thoracic to sacral stimulation with hysterectomy, as the reference site predominantly reflecting generalized sensory changes.
Statistical analysis was performed using Statistica for Windows (version 4.5; Statsoft Inc., Tulsa, OK). Demographic data, cumulative morphine consumption, and thresholds were analyzed by using one-way, repeated-measures two-way, and repeated-measures four-way analysis of variance (fixed effects, three factors: drug [fentanyl, magnesium, ketamine], site [arm, thorax, incision, leg], test [sensation, pain detection, pain tolerance]), respectively. Post hoc testing was performed by using Tukey's honest significant difference test. Pain and sedation scores were analyzed by using Kruskal-Wallis analysis of variance and Bonferronicorrected Mann-Whitney U-testing. Statistical significance was set at P < 0.05.
All patients completed the study without problems. The fentanyl, magnesium, and ketamine groups were similar for age (48 +/- 8, 47 +/- 6, and 47 +/- 8 yr, respectively [means +/- SD]), height (161 +/- 6, 162 +/- 7, and 158 +/- 6 cm, respectively), weight (62 +/- 10, 70 +/- 9, and 63 +/- 9 kg, respectively), and baseline thresholds (Figure 1). Pain intensity VRS, cumulative PCA morphine consumption, and observer sedation scores never differed among groups (Table 1), with pain VRS differences on Day 5 just failing to reach significance (P = 0.054).
The overall courses of the absolute thresholds were similar for the drug groups (Figure 1), differing with test types and time (P < 0.000001). Thresholds were increased compared with baseline 1-24 h postoperatively, taken together (time), for drug groups (drug x time; P < 0.000001), test types (test x time; P = 0.000003), or measurement sites (site x time; P = 0.03). Thresholds were highest at 4 h, returning to baseline on Day 5. Thoracic and incision thresholds were always similar. Arm thresholds were lower than thoracic thresholds at 1 h and lower than incision thresholds at 4 and 24 h. Leg thresholds were lower than incision thresholds at 24 h (site x time). At 1 h, fentanyl thresholds were lower, and at 5 days they were higher, than for ketamine or magnesium (drug x time). At 24 h, ketamine thresholds were higher than those for magnesium.
In all three groups (drug x test x time), pain detection and tolerance thresholds remain increased compared with baseline (P < 0.001) from 1 to 24 h, with the exception of pain detection in the fentanyl group, which was unchanged at 1 h. Sensation thresholds were increased for ketamine at 24 h, for fentanyl at 4-24 h, and for magnesium at 1-4 h. Magnesium pain tolerance thresholds were lower than fentanyl at 5 days and lower than ketamine at 4-24 h. Considering measurement site (site x test x time), pain detection and tolerance thresholds were increased compared with baseline from 1 to 24 h at the thorax and incision sites. Sensation thresholds never changed in the arm and thorax (incision increased at 4-24 h, leg increased at 1-4 h). Leg pain tolerance thresholds were increased at 1-24 h (detection 1-4 h) and were similarly increased for the arm, except at 1 h. At 24 h, incision dermatome pain tolerance thresholds were higher than those for the arm or leg. For all drug groups (drug x site x time), thoracic and incision thresholds were increased versus control at 1-24 h, with the exception of fentanyl, which was unchanged at 1 h, and magnesium/thorax, which was unchanged at 24 h. Arm thresholds remained unchanged with magnesium but were increased at 4-24 h in the fentanyl and ketamine groups. Leg thresholds were increased in all groups at 4 h (ketamine 1-24 h). For fentanyl at 24 h, leg thresholds were lower than incision thresholds. Significant differences for all three factors and time (i.e., drug x site x test x time) are shown in Figure 1.
For the overall course of the relative thresholds (Table 2), there were no differences among drug groups, tests, or sites, with the difference between fentanyl and ketamine just failing to reach significance (P = 0.051). At 1 and 4 h postoperatively, fentanyl relative thresholds were increased compared with baseline and higher than those for ketamine and magnesium (drug x time; P = 0.00006). At no time were relative thresholds decreased compared with baseline. There were no significant differences for all three factors and time (drug x site x test x time) (Figure 2).
Neither patients receiving fentanyl nor those receiving ketamine or magnesium revealed evidence of segmental (spinal) hyperexcitability in this study. Regardless of the type of anesthetic supplementation, all patients show generalized (central) inhibition greatest at the site of surgery as the predominant change in sensory processing up to 24 hours postoperatively. Fentanyl patients had the least generalized inhibition just after surgery, accompanied by significant segmental inhibition not present in the other groups. From four hours postsurgery onward, magnesium-supplemented patients exhibited less generalized sensory inhibition than those in the fentanyl or ketamine groups. Five days postsurgery, patients in the ketamine and magnesium groups had lower thresholds than those in the fentanyl group, long after pharmacological actions of the drugs had worn off. These sensory differences were not reflected in clinical pain measures.
The results emphasize the complexity of postsurgical sensory changes and their interactions with analgesic and anesthetic drugs in the intact human. They demonstrate the difficulty of extrapolating results from (frequently nonintact) animal experiments to the clinical situation. All three drugs suppress spinal hypersensitivity. Animal data suggesting interference with SIA by drugs antagonizing the NMDA receptor system are supported by the greater generalized sensory inhibition up to five days postsurgery in the fentanyl group. The differences between ketamine and magnesium may stem from differing effects on non-NMDA systems. Just postoperatively, there is evidence of multisegmental spinal sensory inhibition accompanied by less generalized inhibition in the fentanyl group. The lesser generalized inhibition might be due to inhibition of DNIC-type mechanisms by fentanyl , with the spinal inhibition involving direct spinal effects of fentanyl, which are well described in the literature but were not observed in this form in our previous study . Conclusive investigation of DNIC depends on measurement of nociceptive flexion reflexes , which was not performed in this study. In the absence of direct measures of intraoperative (and postoperative) nociception, it cannot be determined whether the postoperative sensory differences resulted from differing direct perioperative antinociception or from modulation of reactive sensory changes. Neither the relationships among sensory inhibition and nociception and its sequelae, nor their effects on clinical outcomes, are known.
The lack of clinical effects may be the result of a relatively small sample size or too large a background infusion of morphine. Post hoc power analysis (alpha = 5%, beta = 10%) shows that the study sample size could detect clinically relevant differences of 20% in 24-hour morphine consumption. The background morphine infusion rate was low (0.9 mg/h for a 60-kg patient), providing only approximately one third of total morphine at 24 hours. Virtually all patients demanded a minimum of one bolus per hour in the first 12 hours, which suggests that sufficient pain remained for treatment by PCA boli. We did not measure pain during movement, which might be more sensitive to altered sensory processing, although this is speculative. However, it should be noted that the many studies performed to investigate preemptive analgesia have shown the difficulty of demonstrating clinically relevant postoperative effects after analgesic supplementation of anesthesia. This applies not only to fentanyl and ketamine (e.g., ) but also to magnesium supplementation [12,13].
Having previously demonstrated spinal excitation after unsupplemented isoflurane/nitrous oxide anesthesia and its suppression by fentanyl supplementation , we did not include a placebo group in the present study, using fentanyl supplementation as the standard comparison group, in accordance with typical clinical practice. Fentanyl (micro-opioid receptor agonist), ketamine [noncompetitive NMDA receptor blocker binding at the phencyclidine site of the NMDA ionophore ], and magnesium [physiological blocker of NMDA calcium ionophore ] all reduce postnociceptive spinal excitation in animal models [3,5,16]. Neither ketamine (sigma receptor agonism) nor magnesium (generalized calcium antagonism) can be considered pure NMDA antagonists, and both significantly affect the central nervous system (and thus anesthesia) by mechanisms not involving NMDA receptors. Isoflurane and nitrous oxide produce less depression of spinal excitation than opioids or ketamine in animal models , not sufficient to suppress sensitization in the clinical surgical context .
Dosing schemes for fentanyl and ketamine were chosen to correspond to clinical practice. The schemes are comparably analgesic clinically (trough plasma concentrations) based on pharmacokinetic modeling performed by using IVA-SIM [J. Schuttler, S. Kloos, Department of Anaesthesiology, University of Bonn, Bonn, Germany]: fentanyl approximately 0.8 ng/mL, ketamine approximately 0.4 micro g/mL) [18,19]. Magnesium doses were based on clinical practice in gynecology and anesthesia  and significantly increase cerebrospinal fluid concentrations  to orders of magnitude depressing electrophysiological NMDA receptor activation in vivo .
The interpatient threshold variability of our study corresponds with that of other studies . Threshold variability was reduced by standardizing instructions, avoiding sensitization (only three well spaced measures, stopping on reaching pain tolerance, no difference between first and last measures), and minimizing reaction time effects (slowly ramped current [approximately 0.1 mA/s]; the similar sedation scores returned to baseline by 4 h). Like surgery, transcutaneous electrical stimulation, simple and frequently used in pain research, results in mixed nerve fiber population activation. Because non-nociceptive surgical inputs also contribute to spinal sensitization , this offsets the disadvantage of electrical stimulation not being purely nociceptive.
Anesthetic drug hangover could have influenced thresholds immediately postoperatively. However, increasing thresholds from one to four hours postoperatively make this unlikely. Pain detection thresholds are unaffected by subanesthetic isoflurane concentrations; they may remain increased up to 30 minutes after nitrous oxide . Ketamine increases pain tolerance thresholds, particularly for temporal summation, as do opioids [25,26]. Opioids have little effect on pain detection and no effect on sensation thresholds [26,27]. We found no studies of the effects of magnesium on sensory thresholds. Threshold measures may also have been affected by morphine, another opioid, for analgesia-obligatory for obvious ethical reasons. However, because generalized sensory inhibition affected all thresholds significantly, including sensation and pain detection generally unaffected by opioids, sensory inhibition during the first 24 hours is unlikely to be explained by morphine alone, which suggests the involvement of central mechanisms, such as SIA or DNIC.
Morphine analgesia might have suppressed spinal sensitization. However, a previous study involving intervertebral disc surgery  reported spinal sensitization under comparable morphine analgesia. Similar degrees of sensitization should therefore have been detectable in the present study involving abdominal surgery, which is more painful than back surgery (morphine use 56.9 +/- 12.8 vs 38.1 +/- 25.5 mg/d; P < 0.05). Moreover, placebo group segmental excitation in the previous study increased as morphine PCA continued, and in the present study, initial segmental inhibition in the fentanyl group decreased with PCA. Finally, it is unlikely that PCA morphine explains the threshold differences among groups because initial doses were identical and subsequent 24-hour use was similar. This conclusion is supported by the significant differences present five days postoperatively, long after morphine analgesia had ended.
In another study of hysterectomies and isoflurane anesthesia, ketamine- or fentanyl-supplemented patients had similar wound pressure pain thresholds, higher than those for placebo, 24-48 hours postoperatively . Meperidine consumption with fentanyl and ketamine, alike throughout, was similar to placebo from three hours postsurgery onward. Spontaneous incisional pain did not differ between groups. Although wound hyperalgesia reflects both central and peripheral excitation, the results agree with ours in suggesting similar depression of spinal excitation by ketamine and fentanyl. Three further studies of isoflurane anesthesia and hysterectomy have investigated sensory change after surgery [28-30]. Two demonstrated spinal excitation [28,30] depressed by morphine preemption . One study showed generalized sensory inhibition  for a single postoperative measure, in the other , absent absolute thresholds preclude conclusions about central sensory inhibition.
The present study confirms the ability of fentanyl to inhibit spinal excitation for abdominal surgery involving both visceral and somatic nociception and suggests that ketamine or magnesium supplementation is also effective for this purpose. For generalized sensory inhibition after surgery, NMDA antagonism may interfere with SIA, and opioid agonism may interfere with DNIC, in agreement with experimental findings [6,8]. The effects of all three drugs on the various forms of central sensory inhibition in the surgical context require further investigation. Our results demonstrate the importance of considering inhibition as well as excitation in studying postsurgical changes in sensory processing and their pharmacological modulation. Further studies are required to explore these complex interactions and their relationship to pain and other clinical outcomes after surgery.
1. Coderre TJ, Katz J, Vaccariono AL, Melzack R. Contribution of central neuroplasticity to pathological pain: review of clinical and experimental evidence. Pain 1993;52:259-85.
2. Raja SN, Meyer RA, Campbell JN. Peripheral mechanisms of somatic pain. Anesthesiology 1988;68:571-90.
3. Woolf CJ, Thompson WN. The induction and maintenance of central sensitization is dependent on N-methyl-D-aspartic acid receptor activation: implications for the treatment of post-injury pain hypersensitivity states. Pain 1991;44:293-9.
4. Jayaram A, Singh P, Carp HM. An enkephalinase inhibitor, SC 32615, augments analgesia induced by surgery in mice. Anesthesiology 1995;82:1283-7.
5. Dickenson AH, Sullivan AF. Electrophysiological studies on the effect of intrathecal morphine on nociceptive neurons in the rat dorsal horn. Pain 1986;24:211-22.
6. Marek P, Mogil JS, Sternberg WF, et al. N-methyl-D-aspartate acid (NMDA) receptor antagonist MK-801 blocks non-opioid stress-induced analgesia. II. Comparison across three swim stress paradigms in selectively bred mice. Brain Res 1992;578:197-203.
7. Bouhassira D, Bing Z, Le Bars D. Studies of the brain structures involved in diffuse noxious inhibitory controls in the rat: the rostral ventromedial medulla. J Physiol (Lond) 1993;463:667-87.
8. Le Bars D, Willer JC, De Broucker T. Morphine blocks descending pain inhibitory controls in humans. Pain 1992;48:13-20.
9. Wilder-Smith OHG, Tassonyi E, Senly C, et al. Surgical pain is followed not only by spinal sensitisation but also by supraspinal antinociception. Br J Anaesth 1996;76:816-21.
10. Rollman GB, Harris H. The detectability, discriminability, and perceived magnitude of painful electric shock. Percept Psychophys 1987;42:247-68.
11. Tverskoy M, Oz Y, Isakson A, et al. Preemptive effect of fentanyl and ketamine on postoperative pain and wound hyperalgesia. Anesth Analg 1994;78:205-9.
12. Tramer MR, Schneider J, Marti RA, Rifat KR. Role of magnesium sulfate in postoperative analgesia. Anesthesiology 1996;84:340-7.
13. Wilder-Smith C, Knopfli R, Wilder-Smith OHG. Perioperative magnesium infusion and postoperative pain. Acta Anaesthesiol Scand 1997;41:1023-7.
14. Brockmeyer DM, Kendig JJ. Selective effects of ketamine on amino acid-mediated pathways in the neonatal rat spinal cord. Br J Anaesth 1995;74:79-84.
15. Garthwaite G, Hajos G, Garthwaite J. Ionic requirements for neurotoxic effects of excitatory amino acid analogs in rat cerebellar slices. Neuroscience 1986;18:437-47.
16. Feria M, Abad F, Sanchez A, Abreu P. Magnesium sulphate injected subcutaneously suppresses autotomy in peripherally deafferented rats. Pain 1993;53:287-93.
17. Wilder-Smith OHG. Effect of intravenous anesthesia on outcome. In: White PF, ed. Textbook of intravenous anesthesia. Baltimore: Williams & Wilkins, 1997:583-99.
18. Schuttler J, Zsigmond EK, White PF. Ketamine and its isomers. In: White PF, ed. Textbook of intravenous anesthesia. Baltimore: Williams & Wilkins, 1997:171-88.
19. Gan TJ, Glass PSA. Balanced anesthesia. In: White PF, ed. Textbook of intravenous anesthesia. Baltimore: Williams & Wilkins, 1997:347-74.
20. James MFM Clinical uses of magnesium infusions in anesthesia. Anesth Analg 1992;74:129-36.
21. Thurnau GR, Kemp DB, Jarvis A. Cerebrospinal fluid levels of magnesium in patients with preeclampsia after treatment with intravenous magnesium sulphate: a preliminary report. Am J Obstet Gynecol 1987;157:1435-8.
22. Jahr CE, Jessell TM. Synaptic transmission between dorsal root ganglion and dorsal horn neurons in culture: antagonism of monosynaptic excitatory postsynaptic potentials and glutamate excitation by kynurenate. J Neurosci 1985;5:2281-9.
23. Arendt-Nielsen L, Brennum J, Sindrup S, Bak P. Electrophysiological and psychophysical quantification of temporal summation in the human nociceptive system. Eur J Appl Physiol 1994;68:266-73.
24. Tomi K, Mashimoto T, Tashiro C, et al. Alterations in pain threshold and psychomotor response associated with subanaesthetic concentrations of inhalation anaesthetics in humans. Br J Anaesth 1993;70:684-6.
25. Arendt-Nielsen L, Petersen-Felix S, Fischer M, et al. The effect of NMDA-antagonist (ketamine) on single and repeated nociceptive stimuli: a placebo-controlled experimental human study. Anesth Analg 1995;81:63-8.
26. Hill HF, Chapman CR, Saeger LS, et al. Steady-state infusions of opioids in human. II. Concentration-effect relationships and therapeutic margins. Pain 1990;43:69-79.
27. van der Burght M, Rasmussen SE, Arendt-Nielsen L, Bjerring P. Morphine does not affect laser induced warmth and pin prick thresholds. Acta Anaesthesiol Scand 1994;38:161-4.
28. Dahl JB, Erichsen CJ, Fuglsang-Frederiksen A, Kehlet H. Pain sensation and nociceptive reflex excitability in surgical patients and volunteers. Br J Anaesth 1992;69:117-21.
29. Lund C, Hansen OB, Kehlet H. Effect of surgery on sensory threshold and somatosensory evoked potentials after skin stimulation. Br J Anaesth 1990;65:173-6.
© 1998 International Anesthesia Research Society
30. Richmond CE, Bromley LM, Woolf CJ. Preoperative morphine pre-empts postoperative pain. Lancet 1993;342:73-5.