Br J clin Pharmac 1994; 37: 445-452

Dose-related effects of ibuprofen on pain-related potentials

G. KOBAL', C. HUMMEL2, M. GRUBER', G. GEISSLINGER' & T. HUMMEL''Department of Experimental and Clinical Pharmacology and Toxicology, University of Erlangen-Nurnberg,Universitiitsstrasse 22, 91054 Erlangen and 2Department of Neurology, University of Erlangen-Niirnberg,Schwabachanlage 6, 91054 Erlangen, Germany

1 The aim of this study was to investigate the dose-related effects of ibuprofen(placebo, 400 and 800 mg rac-ibuprofen [Aktren®], administered orally) on experi-mentally-induced tonic and phasic pain.

2 Eighteen volunteers participated in this randomized, double-blind, three-fold cross-

over study. Measurements were obtained before and 90 min after administration ofthe drugs. Phasic pain was produced by CO2 pulses of two concentrations appliedto the right nostril. The left nostril was stimulated with a constant stream of dry airwhich produced a tonic painful sensation described as dull and burning. Subjectsrated the intensity of the painful stimuli by means of visual analogue scales. Inaddition, chemo-somatosensory event-related potentials (CSSERP) were alsorecorded in response to phasic painful CO2 stimuli.

3 While mean intensity estimates of both tonic and phasic painful stimuli showed a

non-significant dose-related decrease, a statistically significant dose-relateddecrease was observed for CSSERP amplitudes.

4 In conclusion, in order to investigate analgesic drug effects, CSSERP appear to bea more sensitive measure compared with psychophysically obtained responses.

Keywords ibuprofen chemo-somatosensory event-related potential EEGpain nociception trigeminal nerve

Introduction

While there is a growing number of studies using event-related potentials (ERP) for assessment of pain [1-4],many questions remain uncovered in this area ofresearch, e.g., ERPs have not yet been applied for theassessment of dose-related analgesic effects of non-steroidal anti-inflammatory drugs (NSAIDs). Thus, thepresent study aimed to close this gap. Among NSAIDsibuprofen was chosen as a potent analgesic the dose-related analgesic effects of which have been studiedintensively in a clinical setting [5,6]. The present inves-tigation included the testing of placebo and two doses ofrac-ibuprofen (400 and 800 mg). The drug's absorptionwas controlled by analysis of rac-ibuprofen plasma con-centrations.

Phasic chemical stimulation of the nasal mucosa withcarbon dioxide (CO2) was employed as an establishedmodel for experimental pain. This technique has beenapplied in the investigation of a number of centrally act-ing analgesics [4, 7, 8]. In addition to the phasic painfulstimuli, dry air was used as a long-lasting, tonic painful

stimulus [9]. The parameters recorded in this study notonly included chemo-somatosensory event-relatedpotentials (CSSERP) in response to phasic painful stim-ulation but also the subjects' ratings of the stimulusintensity. To control for possible non-specific effects ofibuprofen, ERPs to acoustical stimuli, the subjects' per-formance in a tracking task, and the ratings for sedation-related symptoms (e.g., 'tiredness,' 'drowsiness') wereassessed.

Methods

Eighteen healthy volunteers (nine male and nine female,aged between 22-37 years, mean age 25.8 years) partic-ipated in the experiments after they had provided writ-ten informed consent. The study was approved by theEthics Committee of the medical faculty of the Univer-sity of Erlangen-Niimberg.

Correspondence: Dr Thomas Hummel, Department of Experimental and Clinical Pharmacology and Toxicology, University ofErlangen-Nuirnberg, Universitatsstrasse 22, 91054 Erlangen, Germany

445

446 G. Kobal et al.

Experimental protocol Acoustically event-related potentials (AERP)

A double-blind, randomized, placebo-controlled three-fold cross-over design was chosen. Thus, the subjectsparticipated in three experiments separated by at least 7days. In each experiment the medication (i.e., eitherplacebo, 400 or 800 mg rac-ibuprofen [Aktren®]) wasadministered orally in two capsules together with100 ml tap water. Subjects were requested to abstainfrom solid food for at least 8 h before commencementof measurements. The experiments consisted of twosessions. The first session took place before administra-tion of the medication, and the second 90 min after. Thisschedule was chosen because rac-ibuprofen reaches itshighest plasma concentrations approximately 60-90 minafter administration [10]. During the experimentssubjects were comfortably seated in an air conditionedroom. White noise of approximately 50 dB HL wasused to mask switching clicks of the chemical stimula-tor (ERA stimulator, Horniss & Zeissberg, FRG).

In an additional training session prior to the actualexperiments subjects became acquainted with the exper-imental procedures, and, specifically, with a breathingtechnique which avoids respiratory flow inside the nasalcavity during stimulation (velopharyngeal closure [11]).

Painful stimulation of the nasal mucosa

For painful stimulation, tonic and phasic stimuli wereapplied heterotopically to the left and right nasalcavities, i.e., long-lasting pain was induced in the leftnostril while short pulses of painful stimuli weredelivered to the right nostril. Tonic painful stimulation[9] was produced by means of a dry airstream of con-trolled temperature and flow rate (220 C, 145 ml s-1).The airstream was delivered throughout the sessionsstarting 5 min before the beginning of each session.Subjects reported a dull or burning pain whichreached its steady state within a few minutes. As arule, both the slight swelling induced by this proce-dure and the pain from the nasal mucosa decreasedimmediately after termination of the stimulation andwas completely gone within 1 h [9]. In order to elicitphasic painful sensations, CO2 was applied to theright nostril (stimulus duration 200 ms, intervalapproximately 40 s). As described previously [12,13], presentation of the CO2 stimuli did not simulta-neously activate mechano- or thermoreceptors in thenasal mucosa. For each session, 30 painful stimuli oftwo different concentrations (65 and 70% v/v C02)were applied in a randomized order.

CSSERP

EEG was recorded from three positions (Fz, Cz, Pz)of the 10/20 system referenced to linked earlobes (Al+ A2). Possible blink artifacts were monitored froman additional site (Fp2/A1 + A2). The sampling fre-quency of the stimulus linked EEG-segments of 2048ms was 250 Hz (bandpass 0.2-30 Hz, pre-stimulusperiod 500 ms). Pain-related CSSERP were obtainedin response to phasic stimuli. Both amplitudes (PINIand N1P2) and latencies (P1, NI and P2) wereanalysed [4].

During each session, 20 acoustical stimuli (1 kHzbursts, 85 dB HL, duration 50 ms; ERA stimulator)were applied randomly to the right ear in the intervalsbetween painful stimuli. Analysis of the AERP wasperformed in exactly the same manner as with theCSSERP.

Intensity estimates ofpainful stimuli

After presentation of each phasic stimulus, subjectsestimated the perceived intensity in relation to a stan-dard stimulus (65% v/v C02) applied at the beginningof the first session. The painful intensity was rated bymeans of a visual analogue scale (VAS) displayed ona computer monitor [4]. The intensity of the standardwas defined as 100 Estimation Units (100 EU). Theintensity of the tonic painful sensation was reportedin a similar manner. However, because the subjectsdid not receive a standard stimulus prior to themeasurements, a different kind of VAS wasemployed. That is, the scale's left hand end indicated'no pain' (0 EU) and the right hand end 'unbearablepain' (100 EU). For further statistical evaluation esti-mates were averaged separately for each session.

Tracking performance

During intervals between phasic painful stimuli,subjects were requested to perform a tracking task ona video screen [4]. Using a joystick they had to keepa small square inside a larger one which movedaround at random. By checking how often, andmeasuring for how long, the subjects had lost track ofthe independently moving square it was possible todetect changes in the state of vigilance and/or motorcoordination (range between 0-100% successfulperformance in tracking). As with intensity estimatesthe data were averaged for each session.

Adverse reactions

After each session subjects estimated the intensity offour symptoms ('tiredness,' 'headache,' 'drowsiness,'and 'vertigo') by means of VAS. The left pole of thescales was defined as 'no such symptom' (0 EU) andthe right pole as 'symptom experienced at maximum'(100 EU). In addition, all spontaneous reports of thesubjects were recorded verbatim.

Plasma concentrations of rac-ibuprofen

In order to control for rate of absorption of rac-ibuprofen, blood samples were drawn through anintravenous catheter 90 min after administration ofthe medication. Rac-ibuprofen concentrations weremeasured by h.p.l.c. assay as described byGeisslinger et al. [14]. The limit of quantificationconsistent with a precision of 10% or less was 0.1 ugml-'. The coefficient of variation over the calibrationrange of 0.1-50 ,ug ml-' of rac-ibuprofen was lessthan 6%.

Dose-related effects of ibuprofen 447

Statistical analyses

For statistical evaluation SPSS PC+ programs wereemployed. Differences between data obtained afterand before administration of the medication werecomputed and submitted to analyses of variance(ANOVA, repeated measurement design). Intensityestimates of the painful stimuli, tracking perfor-mance, and adverse reactions were analysed bymeans of one-way ANOVAs (within-subject factor'dose' [df 34/2]). CSSERP were analysed by meansof three-way ANOVAs (factors 'dose' [df 22/2],'stimulus concentration' [df 11/1], and 'recordingposition' [df 22/2]). AERP data were submitted to atwo-way ANOVA (factors 'dose' [df 22/2], and'recording position' [22/2]).

Results

Plasma concentrations of rac-ibuprofen

Mean plasma concentrations of rac-ibuprofenobtained 90 min after intake exhibited a dose-relatedincrease indicating regular absorption of the drug(400 mg rac-ibuprofen: mean 28.0 ± 11.7 ,ug ml-';800 mg rac-ibuprofen: mean 41.7 ± 18.9 ,ug ml-').

CSSERP

To demonstrate both the CSSERPs' shape andchanges in relation to administration of the drugs,

c0

E0-

cnCnU

Placebo

Figure 1 shows CSSERP averaged across all subjects.Due to an excessive number of blink artifacts ERPscould not be analysed in six subjects.

Amplitudes PINI were reduced in a dose-relatedmanner (factor 'dose': F = 3.81, P < 0.05). That is,when compared with placebo, 400 and 800 mgibuprofen reduced the mean amplitudes PINI byapproximately 20 and 30%, respectively (Figure 2;Table la). These changes were most pronounced atthe fronto-central recording positions (Fz, Cz). Incontrast, the drug-induced decrease of amplitudeN1P2 did not reach a statistically significant level.

Although CSSERP latencies tended to increase inrelation to the concentration of ibuprofen no signifi-cant main effect of the factor 'dose' could be found.However, the statistical interaction between thefactors 'dose' and 'recording position' becamesignificant for latencies P1 (F = 2.75, P < 0.05), andNi (F = 3.26, P < 0.05) emphasizing that CSSERPlatencies changed not only in relation to the dosageof ibuprofen but also in relation to the recordingsites. In other words, while a dose-related increase oflatencies was observed at position Fz these changeswere less pronounced at Cz and Pz (Figure 3).

AERP

In contrast to the pain-related responses, neither oneof the two factors 'dose' or 'recording position' wasfound to produce a significant effect on AERPs(Table Ib).

0 ms 960 0 -ms 960 I0 m. 9o ms 960

Figure 1 CSSERP at recording position Cz (referenced against Al + A2) averaged across all subjects. Thin and thicklines indicate CSSERP before and after administration of the drugs, respectively. Schematic drawings of both the locationof recording positions on the scalp and CSSERP peaks are shown in the insert.

448 G. Kobal et al.

Amplitude PlNl

Fz Cz Pz Fz Cz Pz

Figure 2 Means and standard errors of CSSERP amplitudes PINI and NlP2 in relation to baseline measures (differencesbetween measures obtained 90 min after administration of the drugs (U placebo, 400 mg ibuprofen, 0 800 mgibuprofen) and baseline) at recording positions Fz, Cz, and Pz. Results are shown for both concentrations of painfulphasic stimuli. The significant dose-related decrease of amplitudes PlNI was most pronounced at positions Fz and Cz.

Latency P1 Latency N1

CDE0

en-

to0.cc

w

cDCL)cnvz

Latency P2

IT lI Ii

r

70% v/vCo2

65% V/vCo2

Fz Cz Pz Fz Cz Pz Fz Cz Pz

Figure 3 Means and standard errors of CSSERP latencies P1, Ni, and P2 in relation to baseline measures (differencesbetween measures obtained 90 min after administration of the drugs (U placebo, El 400 mg ibuprofen, El 800 mgibuprofen) and baseline) at recording positions Fz, Cz, and Pz. Results are shown for the two concentrations of painfulphasic stimuli. For latencies P1 and NI the dose-related prolongation at position Fz was emphasized by the significantinteraction between factors 'dose' and 'recording position' (P < 0.05). These changes are less clear at Cz and Pz.

Intensity estimates ofpainful stimuli Adverse reactions

Mean intensity estimates for both phasic and tonic Ibuprofen did not produce any significant reportedstimuli decreased in relation to the administered dose side effects. That is, there were neither dose-relatedof ibuprofen (Figure 4; Table 2a). However, these changes of the side effects' estimates (Table 2b) nor

changes did not reach a level of statistical spontaneous reports of the subjects which might havesignificance. indicated the presence of adverse reactions related to

Amplitude N1P2

(a

0

-

m

w

C-)cn

70% v/vCo2

65% v/vCO2

..- 11 ..Im

Dose-related effects of ibuprofen 449

Table 1 Means and standard deviations (s.d.) of EEG derived parameters

Amplitudes (,uV) Latencies (ms)PIN] NIP2 P1 NJ P2

Position Mean s.d. Mean s.d. Mean s.d. Mean s.d. Mean s.d.

a) CSSERP at recording sites Fz, Cz, and Pz (70% vlv CO2)Placebo

Before Fz 9.2 5.4 17.3 7.0 248Cz 10.5 4.2 17.2 7.8 225Pz 7.5 3.5 26.1 8.6 243

After Fz 7.8 6.1 14.7 4.6 238Cz 10.8 8.4 13.6 6.0 230Pz 7.7 4.8 20.1 6.4 243

400 mg ibuprofenBefore Fz 10.5 5.7 19.0 7.5 233

Cz 12.1 3.6 18.6 6.6 219Pz 9.6 4.2 27.1 10.1 220

After Fz 6.7 4.0 15.0 6.6 263Cz 9.0 4.3 15.2 6.7 239Pz 6.5 2.7 21.3 7.0 243

800 mg ibuprofenBefore Fz

CzPz

After FzCzPz

12.8 10.9 18.0 5.6 23315.3 10.9 17.7 6.4 22210.8 6.7 25.5 8.5 223

8.6 6.6 14.2 6.4 26210.5 6.0 13.1 7.2 2278.2 3.3 21.1 7.5 251

b) AERP at recording site Cz

PlaceboBeforeAfter

14.2 6.5 22.2 8.9 9111.2 3.4 19.7 13.0 88

400 mg ibuprofenBeforeAfter

800 mg ibuprofenBeforeAfter

12.6 5.8 22.6 10.2 8911.7 5.5 17.3 8.2 91

13.8 5.8 24.1 13.6 8810.4 3.9 19.4 9.5 100

a b

D 10 5

0)

IE 1-W

00 ~~~II

:LI

a0)

0-1 400 800 0

Ibuprofen (mg)

Figure 4 Means and standard errors of intensity estimates(differences between measures obtained 90 min afteradministration of the drugs and baseline) for a) phasicand b) tonic painful stimulation. Although there was nostatistically significant effect of the factor 'dose', meanestimates for both kinds of stimuli exhibited a dose-relateddecrease. EU = Estimation Units; 0 70% v/v CO2.* 65% v/v CO2.

the intake of ibuprofen. Moreover, the trackingperformance was unaffected by ibuprofen suggestingthat the subjects' vigilance was not altered by themedication (Table 2b).

Discussion

Ibuprofen produced a dose-related decrease inCSSERP amplitudes. It is important to note that theacoustical ERPs, the subjects' tracking performance,and the ratings for 'tiredness' were not significantlyaffected by ibuprofen. Therefore, it can be assumedthat the dose-related decrease in CSSERP amplitudeswas likely not due to non-specific effects of ibuprofensuch as sedation, but to its analgesic properties.

384336

485738

343932

574455

308296305

298294294

302290284

321305306

434339

556446

453943

555056

483493489

485486505

454466470

493483480

709396

837664

625538

595248

494045

594460

294289286

330312302

563435

725142

471460474

502498486

877181

858278

2020

144139

2324

246 42243 48

1919

2428

148143

143148

2223

2923

262 48255 51

250 52264 57

450 G. Kobal et al.

Table 2 Means and standard deviations of intensity estimates, tracking performance, and esti-mates of side effects

Phavic pain70% vlv CO2 65% vlv CO2Mean s.d. Mean s.d.

Tonic painMean s.d.

a) Intensity estimates of painful stimuli [in estimation units]

PlaceboBefore 76.0After 78.9

400 mg ibuprofenBefore 80.0After 79.7

800 mg ibuprofenBefore 72.8After 74.4

18.629.8

23.633.3

20.428.5

57.6 13.1 36.9 22.761.0 26.3 32.3 24.1

62.2 21.0 38.7 24.864.4 29.3 30.2 24.0

56.8 18.6 41.9 24.255.5 27.3 32.2 23.2

TrackingPerformance Tiredness Headache Drowsiness VertigoMean s.d. Mean s.d. Mean s.d. Mean s.d. Mean s.d.

b) Tracking performance [in %], estimates of side effects [in estimation units]Placebo

Before 49.1 13.3 7.9 14.9 0.1 0.5 5.2 20.5 0.0 0.0After 52.9 10.6 14.8 18.7 1.8 4.5 7.9 14.5 0.1 0.2

400 mg ibuprofenBefore 46.7 11.7 10.6 16.7 3.3 12.0 5.7 19.4 0.0 0.0After 51.1 10.9 6.9 8.7 5.1 21.2 5.1 7.1 0.0 0.0

800 mg ibuprofenBefore 50.2 12.6 10.3 16.1 0.0 0.0 0.7 2.0 0.7 2.0After 51.5 10.2 13.3 15.1 0.7 2.8 2.1 5.0 2.1 5.0

The results also suggest that CSSERP constitutes amore sensitive measure when compared with psycho-physically obtained responses. While CSSERP ampli-tudes significantly decreased in relation to an increasein the dosage of ibuprofen, ratings for both tonicand phasic painful stimuli only exhibited atendency to decrease in a dose-related manner thatdid not reach statistical significance. In that respect,the current study confirmed previous reports byKobal et al. [4] who found that after a single doseadministration of aspirin there was a decrease ofCSSERP amplitudes accompanied by a meandecrease in intensity estimates of phasic painfulstimuli. However, as with the present data, thedecrease in the subjects' ratings was not statisticallysignificant.

Regarding the specificity of uncovering a drug'sanalgesic properties, similar results on CSSERPshave also been obtained in two other studies on'centrally acting analgesics' [8, 15]. An investigationof flupirtine [8], a novel centrally acting analgesic,showed that CSSERP amplitudes decreased in a dose-related manner for 100 and 200 mg flupirtine while

there was no further decrease for 300 mg. In contrast,intensity estimates of the painful stimuli steadilydecreased in relation to an increase of the dose.However, the decrease of the ratings for 300 mgflupirtine was accompanied by a strong increase ofthe side effect 'elation in mood' indicating that thedecrease in intensity estimates was subject to a non-specific change in the perception of painful stimuli.In addition, Hummel et al. [15] recently published aninvestigation of the drugs anpirtoline, tramadol andimipramine all of which have been reported to pos-sess pain-relieving properties. They demonstrated thatimipramine reduced estimates of painful stimuli whileCSSERP amplitudes remained unchanged. Sinceimipramine also produced a large increase of the sideeffect 'sedation,' a decrease in the tracking per-formance, and an increase in the alpha-band of thespontaneous EEG, it was concluded that the pain-relief was not based on antinociceptive effects butrather on non-specific properties of the tricyclic anti-depressant. Thus, the present study adds to the cumu-lative evidence that CSSERPs are a useful toolin the evaluation of analgesic efficacy in humans

Dose-related effects of ibuprofen 451

producing information different from the sole analy-sis of psychophysics.

It is interesting to note that ibuprofen induced adecrease in CSSERP amplitudes which was mostpronounced at the frontal recording sites. Similarresults have been obtained by Kobal et al. [4] whoobserved that aspirin at a dosage of 1000 mg admin-istered orally produced a decrease of CSSERP ampli-tudes comparable with those produced by 30 mgpentazocine, a partial opioid-receptor agonist/antagonist, administered intravenously. However, incontrast to the opioid, aspirin failed to inducesignificant CSSERP changes at parietal recordingsites. These results have been interpreted in terms ofthe possible differential influence of the two drugs,aspirin and pentazocine, on cortical sources responsi-ble for the generation of CSSERP [16].

Ibuprofen affected the CSSERPs' early amplitudePINI. In light of other studies [17-19], this likelyindicates that ibuprofen affected exogenous stimuluscharacteristics (e.g., stimulus intensity) rather thanthe processsing of endogenous information (e.g.emotional evaluation of the painful stimulus) beingrepresented by later CSSERP components. Thus, thedata suggest that ibuprofen activates antinociceptivemechanisms at relatively early stages of the stimulusprocessing which may include both actions at periph-eral receptor sites [20] and antinociceptive activity inthe dorsal horn of the spinal cord [21].

Since CSSERP are generated in response to short-lasting, phasic stimuli, the findings of the presentstudy question the widely accepted view that NSAIDsdo not interfere with the perception of phasic painfulstimuli which predominantly activate Adelta-fibres[22, 23]. Surprisingly, this view is not based on

single fibre recordings but rather on the interpretation ofvarious NSAIDs' actions in different animal modelsinvolving the evaluation of reflex mechanisms suchas the tail-flick or writhing [24-26]. Although theeffect on Adelta-fibres may not be the predominantpharmacologic action of NSAIDs, the present resultsadd to other findings indicating the presumed actionof NSAIDs on Adelta-fibre mediated nociception [27,28].

In contrast to CSSERP amplitudes, the ratings forboth tonic and phasic pain tended to decrease in adose-related manner but did not exhibit a significanteffect of ibuprofen. This negative result confirmsother research in single-dose studies [4, 28]. That is,although the subjects' ratings decreased after drugadministration, those studies did not demonstrate asignificant effect of so-called antipyretic analgesicson intensity estimates for painful stimuli. However, ithas been shown that NSAIDs produce a significantreduction in ERP amplitudes. In line with this, it islikely that pain-related ERPs are a more sensitivemeasure of analgesia compared with intensityestimates of pain. In other words, compared withERPs, intensity estimates of painful stimuli appear toexhibit a larger degree of both inter- and intra-individual variability. This may be due to factors suchas the subjects' mood, bias, or willingness to respondto painful stimuli.

This research was supported by DFG grant SFB 353 (A7).We would like to thank Dr Elisabeth Pauli, University ofErlangen-Niirnberg, Department of Neurology, for statisti-cal evaluation of the data. We are indebted to Dr SteveMeller, University of Iowa, Department of Pharmacology,Iowa City, USA, for helpful suggestions during preparationof the manuscript.

References

1 Hill HF, Chapman CR, Saeger LS, Bjurstrom R,Walter MH, Schoene RB, Kippes M. Steady-state infu-sions of opioids in human. II. Concentration-effectrelationships and therapeutic margins. Pain 1990; 43:69-79.

2 Arendt-Nielsen L. First pain event related potentials toargon laser stimuli: Recording and quantification. JNeurol Neurosurg Psychiat 1990; 53: 398-404.

3 Bromm B, Meier W, Scharein E. Imipramine reducesexperimental pain. Pain 1986; 25: 245-254.

4 Kobal G, Hummel C, Niirnberg B, Brune K. Effects ofpentazocine and acetylsalicylic acid on pain-rating,pain-related evoked potentials and vigilance in relation-ship to pharmacokinetic parameters. Agents Actions1990; 29: 342-359.

5 Laska EM, Sunshine A, Marrero I, Olson N, Siegel C,McCormick N. The correlation between blood levels ofibuprofen and clinical analgesic response. Clin Phar-mac Ther 1986; 40: 1-7.

6 Seymour RA, Hawkesford JE, Weldon M, Brewster D.An evaluation of different ibuprofen preparations in thecontrol of postoperative pain after third molar surgery.Br J clin Pharmac 1991; 31: 83-87.

7 Kobal G, Hummel T, Hosl M. Pain-related evokedpotentials by chemical stimuli: Effects of analgesics. InAdvances in pain research and therapy, Vol. 12, Lipton

S, Tunks E, Zoppi M, New York: Raven Press, 1989:95-98.

8 Hummel T, Friedmann T, Pauli E, Niebch G, Borbe H-O, Kobal G. Dose-related analgesic effects of flupirtine.Br J clin Pharmac 1991; 32: 69-76.

9 Kobal G, Hube H, Pauli E, Hummel T. Testing theanalgesic activity of dipyrone in human subjects usingan experimental pain model with tonic and phasic painstimuli. In New pharmacological and epidemiologicaldata in analgesics research, ed Brune K, Basel:Birkhiiuser, 1990: 29-36.

10 Geisslinger G, Dietzel K, Bezler H, Nuernberg B,Brune K. Therapeutically relevant differences in thepharmacokinetical and pharmaceutical behaviour ofibuprofen lysinate as compared to ibuprofen acid. Int Jclin Pharmac Ther Tox 1989; 27: 324-328.

11 Kobal G, Hummel T. Human Electro-olfactograms andBrain Responses to Olfactory Stimulation. In Thehuman sense of smell, eds Laing DG, Doty RL,Breipohl W, Berlin: Springer, 1992: 135-151.

12 Kobal G. Elektrophysiologische Untersuchungen desmenschlichen Geruchssinns. Stuttgart, Thieme-Verlag,1981.

13 Kobal G. Pain-related electrical potentials of the humannasal mucosa elicited by chemical stimulation. Pain1985; 22: 151-163.

452 G. Kobal et al.

14 Geisslinger G, Dietzel K, Loew D, Schuster 0, Rau G,Lachmann G, Brune K. High-performance liquid chro-matographic determination of ibuprofen, its metabolitesand enantiomers in biological fluids. J Chromatogr1989; 491: 139-149.

15 Hummel T, Hummel C, Friedel I, Pauli E, Kobal G.Assessment of antinociceptive effects in man: compari-son of imipramine, tramadol, and anpirtoline. Br J clinPharmac 1994; 37: 325-333.

16 Huttunen J, Kobal G, Kaukoronta E, Hari R. Corticalresponses to painful C02-stimulation of nasal mucosa;a magnetoencephalographic study in man. Electro-enceph clin Neurophysiol 1986; 64: 347-349.

17 Chen ACN, Chapman RC, Harkins SW. Brain evokedpotentials are functional correlates of induced pain inman. Pain 1979; 6: 365-374.

18 Hummel T, Livermore A, Hummel C, Kobal G.Chemosensory event-related potentials: relation toolfactory and painful sensations elicited by nicotine.Electroenceph clin Neurophysiol 1992; 84: 192-195.

19 Livermore A, Hummel T, Kobal G. Chemosensoryevoked potentials in the investigation of interactionsbetween the olfactory and the somatosensory (trigemi-nal) systems. Electroenceph clin Neurophysiol 1992;83:201-206.

20 Guzman F, Braun C, Lim RKS, Potter GD, RodgersDW. Narcotic and non-narcotic analgesics which blockvisceral pain evoked by intra-arterial injection ofbradykinin and other analgesic actions. Arch Int Phar-macodyn Ther 1964; 149: 571-588.

21 Malmberg AB, Yaksh TL. Hyperalgesia mediated byspinal glutamate or substance P receptor blocked byspinal cyclooxygenase inhibition. Science 1992; 257:1276-1279.

22 Treede R-D, Kief S, Hoizer T, Bromm B. Latesomatosensory evoked cerebral potentials in responseto cutaneous heat stimuli. Electroenceph clin Neuro-physiol 1988; 70: 429-441.

23 Harkins SW, Price DD, Katz MA. Are cerebral evokedpotentials reliable indices of first or second pain? InAdvances in pain research and therapy, Vol. 5, edsBonica JJ, Lindblom U, Iggo A, New York: RavenPress, 1983: 185-191.

24 Moncada S, Ferreira SH, Vane JR. Pain and inflamma-tory mediators. In Inflammation, Handbook of experi-mental pharmacology, eds Vane JR, Ferreira SH,Berlin: Springer-Verlag, 1979: 588-590.

25. Yaksh TL. Central and peripheral mechanisms for theantialgesic action of acetylsalicyclic acid. In Acetylsali-cylic acid: new uses for an old drug, eds Barett HJM,Hirsh J, Mustard JF, Raven Press, NY, 1982: 137-15 1.

26 Wood PL. Animal models in analgesic testing. In Anal-gesics; neurochemical, behavioral and clinical per-spectives, eds Kuhar MJ, Pasternak GW, New York:Raven Press, 1984: 175-194.

27 Jurna I, Brune K. Central effect of the non-steroid anti-inflammatory agents, indometacin, ibuprofen, anddiclofenac, determined in C fibre-evoked activity insingle neurones of the rat thalamus. Pain 1990; 41:71-80.

28 Bromm B, Forth W, Richter E, Scharein E. Effects ofacetaminophen and antipyrine on non-inflammatorypain and EEG activity. Pain 1992; 50: 213-221.

(Received 16 August 1993,accepted 21 December 1993)