Analgesic Efficacy of Low-dose Ketamine: Somatosensory-evoked Responses in Relation to Subjective Pain Ratings

After institutional approval and written informed consent had been obtained, 30 adult volunteers (ASA-physical status 1; 16 females, 14 males; aged 27+/-6 yr) were studied. All subjects were free from neurologic, psychiatric, cardiovascular, renal, or hepatic diseases and had not taken any drug known to interfere with central nervous system function for at least 3 months before the study. Subjects were assigned randomly to one of three groups. Subjects of group 1 (n = 10) received 5 ml saline intravenously as placebo. In groups 2 (n = 10) and 3 (n = 10), intravenous ketamine (0. 25 mg *symbol* kg^-1and 0.50 mg *symbol* kg (^-1), respectively) was administered intravenously for 30 sec. The following variables were recorded at baseline and for 50 min after placebo or ketamine administration: EEG, SEP, HR, (beats per min, RM300, Honeywell, Germany), mean arterial blood pressure (MAP, mmHg; Dinamap, Critikon), end-tidal P^CO2(PET^CO2, mmHg, Datex, Hoyer, Germany).

Electroencephalogram and SEP were recorded continuously with Silver/AgCl electrodes by means of an EEG-amplifier (Biosignalverstaerker, nbn-electronics, Germany) from vertex (Cz) versus linked earlobes (A1-A2). The electrooculogram was recorded from supra- and infraorbital leads. For all recordings, the system bandpass was set at 0.5 and 30 Hz (3 dB cut-off points, 24 dB/octave). The electrode impedances were kept below 5 kOhm (at 12 Hz). The stimulus and the raw EEG signals were monitored on an oscillograph (Hameg) and stored on an analog tape (Store 7DS, Racal, England). Peristimulus EEG segments from 5 s before to 3.5 s after stimulus onset were digitized off-line, with a sampling rate of 200 Hz and a resolution of 12 bit (CED 1401, Cambridge Electronics, Cambridge, England).

Somatosensory stimuli were given to the tip of the left index finger by the intracutaneous technique described by Bromm and Meier. [15]Briefly, the superficial epidermal layers of the glabrous skin are carefully removed, using a small stainless steel drill comparable to a drill used by a dentist (diameter: 1.5 mm). Constant current stimuli (train-of-four 5 msec square waves) with intensities of 2- and 3-fold above the individual pain threshold are passed through an inserted and fixed metal electrode, with a large return electrode placed at the middle finger. This procedure ensures a high current density at the superficial nociceptive nerve terminals. The elicited sensation is described as a stinging pain similar to that evoked by electrical tooth pulp stimulation. With this technique, pain thresholds are one order of magnitude lower than with conventional electrical stimuli. [15]Before medication and before EEG and SEP recordings, individual detection and pain thresholds were determined by four series of increasing and decreasing stimulus intensities. Pain sensation was rated verbally on a numerical scale ranging from “0”(no sensation) to “10”(unbearable pain). The first sensation of any pain was defined as “4”(individual pain threshold). The individually determined intensities ensured pain ratings between “4” and “6.” The subjects were allowed to familiarize themselves with the experimental setup for 30 min before the first two stimulus blocks were recorded. Three seconds after each stimulus, a tone prompted the subject to rate his sensation. Because stimulus randomization and long intervals are essential to minimize effects of habituation and distraction, [11]stimuli were given in blocks that consisted of 40 stimuli with randomized intensities (20 stimuli of 2- and 3-fold pain intensity each) with varying interstimulus intervals between 10 and 20 s, resulting in a block length of 10 min. Seven blocks were applied, from which the first block was not evaluated, to exclude major habituation effects. The second block was taken as baseline. The following five blocks were recorded after drug treatment. The time interval between each block was 1 min.

From the spontaneous EEG segment (epoch length: 5 s) preceding each somatosensory stimulus, the spectral power was calculated for selected frequency bands: delta: 1.0–4.0 Hz; theta: 4.0–8.0 Hz; alpha: 8.0–12.0 Hz; and beta: 12.0–30 Hz after Fast Fourier Transformation (Kaiser window, 40 dB sidelobe depression). [16]In addition, the spontaneous EEG was also quantified by the most commonly used percentile frequencies: f²⁵%, f⁵⁰%, f⁷⁵%, and f⁹⁵%. The percentile frequencies are defined as that frequency f [Hz] below which a certain percentage of power is located; f⁵⁰% is also denoted as median frequency and f⁹⁵% as spectral edge frequency. Electroencephalographic parameters were calculated after averaging single-trial spectra of four successive prestimulus EEG segments, resulting in one value per minute. After this, EEG variables were averaged for all subjects. Trials with electrooculogram artifacts were discarded. SEP amplitudes and latencies were calculated from the respective poststimulus EEG segments of 500 msec duration. Evaluated SEP components consisted of a negativity at approximately 150 msec after stimulus onset (N¹⁵⁰) and a positivity (P²⁵⁰) with a latency of approximately 250 msec. The SEP amplitudes were quantified by the peak-to-peak values between these two components.

Treatment effects for the total period of measurements were analyzed using the respective mean values of all variables for each recording block (10 min recording each; one pre- and 5 posttreatment blocks). At first, differences in the posttreatment values between the groups due to inhomogeneities between groups were eliminated by subtracting the mean pretreatment value from the posttreatment values. Changes in values were then subjected to a two-way analysis of variance, with one within-subjects factor (posttreatment blocks) and one between-subjects factor (treatments) to detect “overall” effects over time, treatment, and their interaction. The Huynh-Feldt correction was used to take into account differences between variances. [17]In case of a significant treatment or interaction effect, differences between groups were evaluated at each time point by two-sample t tests, with Bonferroni corrections. In addition, the stability of changes in values during the five posttreatment blocks was tested for the placebo group by a one-way analysis of variance. The Pearson correlation coefficient was used to quantify the relation between variables, including pain scores. [18]Data are given as mean +/-SD, if not stated otherwise. Significance was assumed at P < 0.05. Statistics were performed by the SPSS package.*

Four subjects of group 2 and all subjects of group 3 became unconscious after ketamine administration and were unresponsive for 3 +/-2 min (group 2) or 6+/-5 min (group 3), respectively. Data for HR, MAP, and PET^CO2are shown in Figure 1. Using analysis of variance, significant treatment effects for HR and MAP were noted in groups 2 and 3 (see Table 1for a summary of the analyses of variance). PET^CO2was not significantly affected by the treatments and did not change over time.

The mean detection threshold for the intracutaneous stimuli was 0.16 +/-0.06 mA, and the mean pain threshold was 0.34+/-0.08 mA. Pain ratings exhibited a significant treatment effect and were changed over time (Table 1), although the pain ratings in the placebo group were stable over the entire observation period (P < 0.05). Four subjects of group 2 and all subjects of group 3 became unresponsive after ketamine administration, and no pain ratings could be obtained over an individually varying time period of 1–11 min. Thereafter, pain ratings from all subjects could be recorded. In both ketamine groups, pain ratings were decreased below the placebo level over the whole observation period. A summary of the Bonferroni corrected t test comparisons is given in Table 2. A ketamine-induced dose-dependent effect on pain relief was seen during the first three posttreatment periods. In the last two experimental blocks, no differences in the analgesic efficacy between both ketamine doses were found.

An example of smoothed EEG single trial spectra of one subject from group 1 (control) is given in Figure 2(left). The drug (placebo) was given after 10 min of baseline recordings. Alpha activity was dominant over time. Corresponding single trial EEG spectra for two subjects given 0.25 mg *symbol* kg^-1and 0.5 mg *symbol* kg^-1ketamine, respectively, are shown in Figure 2(middle, right). The time course of changes in frequency bands and computed EEG variables for all groups is presented in Figure 3.

Ketamine administration resulted in a transient loss of alpha-activity (P < 0.001). As much as a fivefold increase in theta power (P < 0.001, group 3) was observed during the first postmedication stimulus block. Delta and beta activities did not change over time. Changes in theta and alpha activities or other EEG variables were not correlated to reduction in pain perception (Figure 4).

At baseline, SEP consisted of a large biphasic deflection, with a vertex negativity at 150 msec (N¹⁵⁰) and a vertex positivity at 250 msec (P²⁵⁰). In all groups, baseline pain ratings were in the range of 4 to 7. In group 1, SEP and pain ratings did not change over time. Ketamine administration resulted in a dose-dependent attenuation of SEP amplitudes. Even in unresponsive subjects, SEP could be recorded. Pain ratings and averaged SEP components of a subject of group 2 are shown in Figure 5. Data for grand mean SEP amplitudes and subjective pain ratings for all groups are shown in Figure 6. Ketamine administration resulted in decreases in SEP amplitudes, which were different between groups for the first 30 min after bolus injection. At the end of the observation period in both groups, SEP amplitudes were 25–30% less than baseline values (P < 0.05). The correlations between duration of unresponsiveness or reduction in pain relief and EEG/SEP variables are shown in Figure 4. Ketamine-induced changes in the N¹⁵⁰-P²⁵⁰SEP amplitude were significantly correlated with attenuation of pain perception (r = 0.61; P(r) < 0.01) but not with the time of unresponsiveness (r = 0.36; P(r) > 0.05). Conversely, increases in EEG theta power exceeding more than 20% of baseline values were significantly related to the time of unresponsiveness (r = 0.97; P(r) < 0.001).

This study demonstrates a dose-dependent correlation of pain relief and decreases in late cortical SEP components after low-dose ketamine. The dose-dependent decrease in SEP amplitudes seen in the first 30 min after ketamine injection was paralleled by a dose-dependent decrease in pain ratings in all subjects. In contrast, the changes in the spontaneous EEG did not correlate with pain sensation. Within 1 min after ketamine injection, the alpha rhythm disappeared, and a synchronous high-voltage theta-activity was noted. The duration of the dose-dependent increase in theta activity was strongly related to the duration of unconsciousness (r = 0.97, P(r) < 0.001). Hemodynamic responses did not correlate with the neurophysiologic measures or pain relief.

Carefully controlled laboratory studies in healthy volunteers have shown that amplitudes of late SEP elicited by short noxious stimuli are correlated with the intensity of pain sensation. [9–13,19]There is increasing evidence that SEP amplitudes after thermal [9,20,21]intracutaneous [15,22]and tooth pulp [23,24]stimulation decrease as perceived pain is relieved by analgesic treatment. The current study provides electrophysiologic measures on ketamine-induced analgesia that can supplement clinical observations. Most strikingly, after low-dose ketamine, the dose-dependent decrease in subjective pain sensation was paralleled by a dose-dependent attenuation in SEP amplitudes. Our findings are in agreement with previous results showing that 0.44 mg *symbol* kg (^-1) ketamine may produce long-lasting analgesia. [25]Grant et al., [2]using a submaximal tourniquet technique, showed that 0.5 mg *symbol* kg^-1ketamine given by intramuscular and oral route prolonged the period of pain-free ischemic exercise for at least 30 min after drug administration. In addition, it was shown that 0.5 mg *symbol* kg^-1ketamine increases pressure pain detection and tolerance thresholds and attenuates pain sensation induced by electrical stimuli. [5].

Although we did not compare the effect of ketamine to that of other analgesics, our results suggest that the effect of ketamine on late cortical SEP is less depressant than that observed with narcotic analgesics like morphine, fentanyl, or alfentanil. [23,26]From these findings, it may be concluded that the analgesic potency of 0.25–0.5 mg *symbol* kg^-1ketamine with a significant reduction in pain sensation elicited by electrical stimuli is superior to that of aspirin [27]and less than the analgesic effects of centrally acting opioids like tilidine. [28].

Our results are similar to previous findings from Buchsbaum et al., [29]who studied the effects of 5 mg morphine on cortical SEP. In their study, cortical SEP amplitudes were depressed by approximately 15%. This was paralleled by an equivalent decrease in pain sensation. It was pointed out earlier that the N¹⁵⁰-P²⁵⁰amplitude is highly correlated to subjective pain sensation. [9–12,29–31]From this, it may be concluded that late cortical SEP and subjective pain ratings will be at least partially redundant sources of information [32]and, under certain conditions, both measures may be substituted for each other. Consequently, the paradigm used here has successfully been transferred to the clinical setting. [22]In anesthetized patients (0.8% halothane in 66% nitrous oxide), no cortical SEP to intracutaneous electrical stimuli could be recorded. After withdrawal of nitrous oxide, a late SEP similar in waveform to control data was noted. In contrast, changes in EEG spectral power and auditory-evoked responses were not significant. It was concluded that changes in pain-related SEP correlated with analgesic treatment. The applicability of the electrophysiologic technique used here is supported by its intraindividual reliability and its consistency with previous reports on the sensitivity to analgesic treatment effects. [19,29].

Habituation and changes in vigilance are known to affect evoked cortical responses. Randomization of stimulus conditions (intensity and interstimulus intervals) was shown to be of importance to minimize habituation and sensitization effects. [11]No arousal scores were measured in the current study. However, it was shown previously that unchanged SEP and pain ratings in the control group may indicate a fairly constant level of arousal. [10,11,15]From this, it may be concluded that the SEP changes in the treatment group were not due to habituation. In addition, in agreement with previous findings demonstrating unchanged detection and pain thresholds for at least 8 h, no changes in pain threshold were noted in the control group. [15]The fairly constant threshold intensities of electrical pain stimuli are a particular advantage of the intracutaneous stimulus technique, which guarantees a better defined pain sensation than seen with the conventional transcutaneous electrical stimulation of large mixed nerves. [15,33].

The close correlation between verbal pain ratings and SEP amplitudes does not prove, however, that late SEP are pain-specific. The amplitudes depend on a variety of experimental conditions, such as stimulus expectancy and level of arousal. Even when carefully controlled, it is difficult to decide whether the decrease in SEP amplitude is due to a specific analgesic drug effect or due to changes in vigilance. This differentiation between pain relief and decrease in vigilance constitutes a basic problem in pain research. [26]It was pointed out previously that in the EEG, no correlates specific for pain sensation have been found so far. [34]Even with severe pain stimuli, only nonspecific desynchronization patterns may be observed in the EEG. In the current study, vigilance was not assessed directly. However, after an intraindividual different time of unresponsiveness after ketamine administration, all subjects were vigilant enough to respond to the pain stimulus. Consistent with previous findings, changes in EEG parameters were not related to the degree of pain relief. [34]Because the effects of ketamine could clearly be separated from those of the control group, in which EEG activity, SEP, and pain ratings did not change over time, it may be concluded that changes in electrophysiologic parameters reflect ketamine effects. The time of unresponsiveness after ketamine was significantly correlated to increases in theta power, but not to SEP amplitudes. Conversely, SEP but not EEG were highly correlated to pain ratings. It may be concluded that cortical SEP may serve as an indicator of pain relief even in periods when no subjective pain report is available.

The effects of ketamine on early cortical SEP were studied previously. [35,36]Low-dose ketamine (0.5 mg *symbol* kg^-1) did not change subcortical or early cortical (N²⁰-P²⁵) SEP components after median nerve stimulation. Using larger ketamine doses (2 mg *symbol* kg^-1bolus followed by continuous infusion at a rate of 30 micro gram *symbol* kg^-1*symbol* h^-1), Schubert et al. [36]demonstrated increases in the amplitude of the early cortical response (N¹⁹-P²²). The early cortical SEP components are most probably generated in the thalamus or the thalamocortical radiation. [37]They reflect the integrity of afferent somatosensory pathways but may not be specific for nociceptive transmission.

Changes in SEP or EEG seen here cannot be explained by ketamine-induced alterations in HR or MAP, which stayed within the physiologic range during the entire observation period. These changes in hemodynamic variables did not correlate to changes in SEP (r =-0.14; P > 0.05). In addition, it was shown previously that EEG changes after low-dose ketamine are not related to changes in systemic or intracranial hemodynamics. [38]Because PET^CO2was constant over time, an effect of variations of carbon dioxide concentrations on brain electrical activity also can be excluded.

In conclusion, the current study shows that subjective pain sensation is reduced by bolus injection of low-dose ketamine in a dose-dependent manner. Attenuation of late cortical SEP correlated to changes in pain sensation. In the spontaneous EEG, increases in theta activity coincided to the duration of loss of consciousness. Changes in EEG parameters were not related to pain relief. These data indicate that the analgesic effect of low-dose ketamine may be assessed by late cortical SEP.

The authors thank Doris Droese for expert technical assistance.

*Norusis MJ: SPSS for Windows. Release 6, SPSS, Chicago, 1993.