EEGevaluation of reflex testing as assessmentofdepth of pentobarbital anaesthesia in the rat

Z. L. Haberham', W. E. van den Brom2, A. J. Venker-van Haagen2,

V. Baumans1, H. N. M. de Grooe & L. J. Hellebrekers '.2

'Department of Laboratory Animal Science and 2Department of Clinical Sciences of CompanionAnimals, Faculty of Veterinary Medicine, Utrecht University, Utrecht, The Netherlands

SummaryElectroencephalography (EEG)was applied to evaluate the validity of the paw pinch reflex asan indicator of anaesthetic depth in rats which are anaesthetized with a single intraperitonealdose of pentobarbital.

After induction of the anaesthesia, characterized by the rapid loss of the animals' ability tomaintain upright posture, the EEGof 10 out of 11 rats was dominated by paroxysmal (burstsuppression Iactivity, associated with unconsciousness. In seven out of 11 rats, the paw pinchreflex was lost after onset of paroxysmal electroencephalographic activity. However, the pawpinch reflex remained present in four out of 11 animals, demonstrating that the response isindependent of cortical activity.

In five out of seven rats, the EEG still showed paroxysmal activity when the paw pinchreflex was regained. However, in two other rats the EEG returned to a pattern similar to thatshown by awake animals, 4 and 21 min respectively, before the reflex was regained.

These data indicate that in the pentobarbital-anaesthetized rat, presence of the paw pinchreflex is not related to the level of depression of electrical activity in the cerebral cortex, andconsequently is probably not related to the level of consciousness. Based upon these findingsit is concluded that the paw pinch reflex is unreliable as a sole indicator of anaesthetic depth.

Keywords EEG; anaesthesia; rat; pentobarbital; anaesthetic depth

The assessment of reflexes as indicators ofanaesthetic depth is a common practice dur-ing surgical interventions in laboratory rats.Particularly the paw pinch reflex (retractionof the hind paw in response to toe pinching)has been recommended as a reliable indicatorof surgical anaesthetic depth IBertens et a1.1993, Flecknelll9961. However, reflexes arelargely mediated at the level of the spinalcord, and may therefore be inadequate forassessing unconsciousness and, to a certainextent, analgesia.

Correspondence to: Zainal L. Haberham, Department ofLaboratory Animal Science, Faculty of Veterinary Medicine,Utrecht University, PO Box 80.166, NL-3508 TD Utrecht,The Netherlands. B-mail: haberham@las.vet.uu.nl

Accepted 17 April 1998

It has been demonstrated that purposefulmovement in response to a painful stimuluscan occur in decerebrated rats (Rampil et a1.19931and it has been suggested that suchso-called nocifensive responses can be fullyintegrated at the subcortical level (Rampil &.Laster 1992, Antognini &. Schwartz 1993,Rampil et a1. 19931.It is therefore possiblethat the presence of the paw pinch reflex isindependent of activity of higher brainstructures, and that the absence of thereflex is due to anaesthetic depression ofsubcortical sites. Sufficient evidencethat the absence of the reflex occurs onlyduring a state of anaesthetic depressionof the brain is lacking. It has been shown

© Laboratory Animals Ltd. Laboratory Animals (1999) 33, 47-57

48

for isoflurane anaesthesia in goats(Antognini & Schwartz 1993), that the doserequired to abolish nocifensivemovement due to depression of the cere-bral cortex only, is substantially higherthan the dose required to abolish such move-ment when the spinal cord is anaes-thetized as well.

Together with the aforementioned experi-mental data, the increasing number of reportsof awareness combined with occurrence ofimplicit or explicit memory of surgical pro-cedures in anaesthetized human patients (Liuet a1. 1991) raise questions about the possibleoccurrence of these phenomena in animals. Ifperiods of inadequate anaesthesia occur inanimals and remain undetected and thereforeuntreated, this will be detrimental to theirwelfare.

Apart from animal welfare considerations,it can be argued that intraoperative aware-ness in laboratory animals will increase thesympathetic tone and subsequently thevariability of baseline conditions in theseanimals, hence increasing the variabilityof data collected under these conditions.Furthermore, analogous to humans, animalsmay experience intraoperative awareness as astrong aversive stimulus. Recall of aversivememory may lead to alterations in a range ofphysiological parameters (Davis 1992). Thisintroduces the possibility that inadequateanaesthesia may have long-term effectswhich interfere with the outcome of anexperiment.

Changes in EEG pattern have previouslybeen shown to correlate with changes in thelevel of consciousness (Clark & Rosner 1973).Consciousness is usually reflected by thepresence of asynchronous low voltage fastactivity (LVFA)in the EEG, whereas uncon-sciousness is often demonstrated byattenuation of this activity, and the presenceof synchronous high voltage slow activity(HVSA). Long et 01. (1989) demonstratedthat in human patients, awakening from.anaesthesia correlates with an increase in theLVFA/HVSA ratio. Rampil and Matteo (1987) .demonstrated that an increase in special edgefrequency (SEF,indicator of the highest rele-vant frequencies present in the EEG) corre-lates with autonomic responses to a painful

Haberham et a/.

stimulus. Although the current consensusof understanding of the EEG is thatdiscrepancies exist between behaviouralobservations on consciousness and uncon-sciousness on the one hand, and specificLVFA/HVSA patterns on the other hand(Vanderwolf 1992), a number of distinct EEGpatterns exist that can be clearly classified asindicative of an unconscious state (includingiso-electricity and the related 'burst-sup-pression' pattern). Despite its shortcomings,the EEG is still one of the most direct mea-sures of brain activity and consequentlyof mental processing.

In the present study, pentobarbital-inducedchanges in EEG activity were investigated inrelation to the paw pinch reflex. The pawpinch reflex can be considered a fully reliableindicator of anaesthetic depth only whenit can be demonstrated that the absenceof this reflex is invariably accompanied byan EEG pattern reflecting adequate loss ofconsciousness. If, on the other hand,no such relation can be demonstrated,the reliability of the reflex should be ques-tioned.

In order to evaluate the validity of reflex.testing as a tool to assess anaesthetic depthin laboratory rats, reflex evaluation wascombined with EEG recording before, andafter, a single intraperitoneal injection with50 mg/kg sodium pentobarbital. Pento-barbital was chosen for reasons that (1) itacts on both cortical and subcortical sites inthe central nervous system, and (2) itis commonly used as a mono-anaestheticfor general anaesthesia in laboratory rats. Inaddition, the barbiturate-inducedchanges in EEG patterns in relation toconsciousness have been well described inhumans (Tucci et 01. 1949, Shimazono et a1.1953, Clark & Rosner 1973), allowing for acomparative discussion about theinterpretation of EEG data as a measure ofconsciousness.

Materials and methods

The protocol of this study was approved bythe Faculty's Science Committee, the Facul-ty's Ethics Committee and the Ethics Com-

Depth of anaesthesia in the rat

mit tee of the Department of LaboratoryAnimal Science.

AnimalsNine male and two female Wistar rats [U:WU(Cpb)]with body weights ranging between330 g and 490 g (median: 380)were used. Theanimals were obtained from the SPF colonyof the Central Laboratory Animal Facility(GDL,Utrecht University, The Netherlands).Before undergoing the surgical procedure,animals were housed in groups of three, inclear plastic cages (Macrolon™ type 3, Tec-niplast, Italy) with a stainless-steel wire top.

Following the surgical procedure, all ani-mals were individually housed in clear plas-tic cages (Macrolon™ type 2, Tecniplast,Italy) with a purpose-built, clear perspexperforated top, in order to prevent damage toconnectors on the animals' heads. Woodshavings were used as bedding (changedtwice a week), and a sheet of paper tissue wasalways added as cage enrichment. Animalshad access to commercial rat chow (RMH-TM, Hope Farms, Woerden, The Nether-lands) and tap water ad libitum.

Cages were placed inside a mobile climate-controlled isolator cabinet with positivepressure (I££aCredo, Francel, maintainingtemperature at 22°C, and relative humidityat 60%. Animals were maintained under a12: 12 h light regime.

After completion of the experiments, ani-mals were returned to their home cages forfurther use in separate experiments regardingneurophysiological evaluation of anaesthesiaand euthanasia methods.

Surgical proceduresGeneral information Throughout the entireprocedure, rats were kept on an electricalheating pad, maintaining body temperatureat 38°C. Dehydration of the cornea wasprevented by the application of chlor-amphenicol/vitamin A eye ointment. Asep-sis was practised at all times. In none of theanimals was infection of the surgical sitesobserved afterwards.

Anaesthesia and analgesia Rats wereanaesthetized with fentanyl/fluanisone

49

(HypnormTM, Janssen Pharmaceutical,Oxford, UK, containing 0.315 mg fentanylcitrate/m1, and 10 mg fluanisone/ml. Dose:0.5 ml/kg i.m.) and midazolam (Dormi-cum TM, Roche,S mg/kg i.p.), which weregiven with 45-min intervals throughout theprocedure. Adequacy of anaesthesia wasdetermined by traditional reflex testing andabsence of nocifensive movement. Post-operative analgesia was provided by means ofs.c. injections with buprenorphine (Temge-sic™, Schering-Plough, Amstelveen, TheNetherlands, 1 ml/kg), given twice at 6-hintervals after completing the procedure.

Placement of EEG electrodes The animalwas placed in a stereotactic frame. Afterdisinfecting the scalp (70% ethanol), theskull was exposed via a single incision. Theperiosteum was cleared from the bone,following topical application of lidocainespray (Xylocaine 10% spray, Astra Pharma-ceutica BV,Rijswijk, The Netherlands).

For the insertion of electrodes, four holeswere drilled in the skull by means of amanually-operated, purpose-built drill, sus-pended in the stereotactic frame. Corticalelectrodes (ell and e12)were inserted bilat-erally at a distance of 3 mm with respect tothe midline, and 6 mm rostrally from theintra-aural line. A third electrode (el3)wasinserted at A (imaginative intersectionbetween the midline and the extrapolatedintra-aural line), and a fourth electrode (eI4)was placed 2 mm from the midline, 10 mmrostral to the intra-aural line, on the rightside of the head.

Each electrode consisted of a sterilestainless-steel screw (Fabory DIN 84A-A2,1x 3 mm, Borstlap BV,Tilburg, The Nether-lands), pre-soldered to a Teflon-coated,braided stainless steel wire (7 cores, 0.044 emin diameter, length about 2.5 em, ElektronBV,Oss, The Netherlands), by use of a corro-sive flux (Rovita flux, Chemische FabriekenSchiedam BV,Schiedam, The Netherlands)and high-quality solder (LMP silver solder,Multicore Solders, Germany). After place-ment of the electrodes, the pre-soldered freeends of the wires were soldered to a smallreceptacle (Mecap Preci-Dip 917-93-108-41-005). Soldering was performed by melting

50

the pre-soldered connection points on thereceptacle, followed by rapid joining of thepre-soldered end of the steel wire, hencepreventing a temperature increase in theelectrode and subsequent tissue damage. Thereceptacle and electrodes were cemented(ESPE Ourelon Polycarboxylate) to the skull.After hardening of the cement, the skin wassutured around the receptacle.

Experimental procedures

General information Animals wereindividually placed in a purpose-built50 x 40 x 40 cm Plexiglas box, shielded by aFaraday cage when indicated, in order toreduce interference with the EEG signal bynoise from the electrical environment. Theglass bottom of the cage was placed on awarm water heating pad, maintaining normalbody temperature. The receptacle fixed onthe animal's head mated with a connector,attached to a suspended electrical swivelconnector (Organon Diosynth, Oss, TheNetherlands) by a shielded multi-lead cable,feeding the signal to the bioelectric ampli-fiers. This set-up allowed free movement ofthe animal inside the box.

Reflex testing Two reflexes were testedthroughout the procedure: righting reflex(RR, defined as immediate restoring of pos-ture by the hind legs, after placing the animalon its back, used as a behavioural descriptionof anaesthetic effect per sel, and paw pinchreflex (PPR, retraction of a hind paw afterpinching between the toes). Reflexes wererated on a three-level scale: strong response(score=2), weak response (score = 1) or noresponse (score = 0). The frequency of testingis described below.

Electrophysiologicalrecording Corticalactivity was differentially recorded over twochannels. Ell and e12were connected to therespective positive inputs of the amplifiers,el3 to both negative inputs, and el4 served asthe common neutral electrode. The signalwas amplified 1000 times, and band-pass fil-tered between 0.53 and 300 Hz (Bio-electricamplifier AB 601-G, Nihon Kohden, Tokyo,Japan). The analogue signal was fed to a 386

Haberham et a/.

OX-based personal computer, digitized on-line at 1 kHz by an analogue/digital con-verter (PC LabCard, Advantech Systems,Taiwanl. Acquisition was performed withdedicated software, developed in-house(WEvdB).

The acquisition protocol was the follow-ing: each recording consisted of 28 con-secutive epochs of 1024 data points each,immediately stored on disk as ASCII files(internal hard drive or external removablemedia (lOa MB ZIP disk, Iomega, Roy, Utah,USA)!. Two or three control (baseline EEe)recordings were performed, before inductionof anaesthesia by means of a single injectionwith pentobarbital (Nembutal™, SanofiSante, Maassluis, The Netherlands,50 mg/kg). Immediately following injection,recordings were performed every minute,until loss of righting reflexes was observed.Intervals between measurements were thengradually increased to about 5-10 min, untilthe end of the experiment as defined by theanimal walking about spontaneously.

Data analysis Data analysis was performedoff-line by visual inspection and by powerspectrum analysis (dedicated softwaredeveloped in-house).

Power spectra were calculated by fastFourier transforms (FFT) over each one sec-ond epoch of EEG, provided that the activitywithin the epoch was stationary and free ofartefacts overloading the amplifiers (asjudged by visual inspection). For eachrecording, the spectra extracted from all 28consecutive epochs were averaged. Ininstances where rapid overall changes in theEEG were clearly visible in the raw data, the28 one-second recordings were divided intofour blocks of seven epochs. Accordingly, theextracted power spectra over these blockswere the average of seven epochs of onesecond each. Recordings in which the EEGcontained transient irregular activity ofhigh amplitude (paroxysmal EEG, or 'burstsuppression'l, were omitted from powerspectrum analysis, since reliable interpreta-tion of power spectra requires stationarity.Analysis of these periods was limited tovisual detection of the presence of bursts.

Depth of anaesthesiain the rat

From the obtained averaged spectra, thefollowing parameters were extracted: powerin the four classical frequency bands, namely[) (1.0-2.9 Hz), e (3.9-6.8 Hz), a (7.8-11.7 Hz),f3 (12.7-25.4 Hzl, and total power (1.0-25.4 Hz). The ratio f31(a + 0) was calculated asthe 'vigilance index', indicating the amountof 'conscious' high frequency activity relativeto 'unconscious' lower frequencies. In addi-tion, median frequency (MF, defined as thefrequency separating the lower and higher50% of power in the entire passed band) andthe 95% spectral edge frequency (SEF,definedas the frequency under which 95% of allpower was contained) were extracted. In eachanimal, data were normalized by expressingthe activity in each band as a percentage ofthe total power.

Comparisons between parameters wereperformed by means of analysis of variance(ANOVA) for repeated measures, or when thedata were not normally distributed, by meansof Friedman's non-parametric ANOVA forrepeated measures. Differences were con-sidered statistically significant at P < 0.05.Calculations were performed on computerwith the aid of Microsoft Excel 6.0 and JandelSigmas tat.

ResultsEEG stages during anaesthesiaPentobarbital administration produced adistinct sequence of visually detectablepattern changes in the EEG, with concurrentchanges in spectral parameters (Fig 1).

Stage 1: baseline EEG Baseline EEG (FiglA) primarily consisted of a low voltage, fastf3 activity, superimposed on a pattern con-sisting of lower frequencies with slightlyhigher amplitude [Table 1).

Stage 2: induction In 10/11 rats threedistinct EEG patterns were observedimmediately following injection, both byvisual inspection (Fig I) and by spectralanalysis (Fig 2). The initial EEG patternchange was a subtle increase in amplitude ofthe fast f3 waves (substage 2a, Fig IBj. Thispattern was rapidly replaced by slower f3

51

Fig 1 EEGstages during anaesthesia in a represen-tative rat. [A]: Stage 1. Baseline EEG,characterized bylow voltage fast p activity. [B]: Stage 2, substage a.Induction: subtle increase in amplitude of the fast pactivity. Ie]: Stage 2. High-voltage p-spindles (sub-stage 2b, indicated by thin arrows) alternating withhigh-voltage slower (0:) waves (substage 2c, indicatedby bold arrows). The animal loses righting reflex. [D]:Transition from substage 2c to stage 3. Sharp, highvoltage waves occur in between high voltage slowerwaves. [E]: Stage 3. High voltage sharp waves (bursts)alternating with periods of low activity (suppression).[F]: Transition from stage 3 to stage 4. High voltage 0:

waves dominate the EEG,some sharp waves occur.[G]: Stage 4. High voltage 0: waves form the dominantactivity, faster waves of low amplitude are super-imposed on the trace. [H]: 10 min after regain ofrighting reflex. The EEGis still in stage 4, and a Iispindle (arrow) is evident slightly left of the centre ofthe trace. [I]: one hour after regain of righting reflex.Despite the increase in p activity, stage 4 EEGstilldominates

waves of higher amplitude (substage 2b, FigIC), often appearing in spindle-shaped vol-leys ('barbiturate spindles'). Slower waves(predominantly O-al of high voltage (~ 400-600 IlV) then replaced the f3 waves (substage2c, Fig IC, bold arrows).

Between these three substages, transitionstages were observed, appearing as a mixtureof the activities in both adjoining stages. In

52 Haberham et a/.

Table 1 Basicfeatures of baseline electroencephalo-gram

these 10/11 rats, duration of stage 2 was 3-18(median: 5: 5) min, with substage 2a and 2boccurring within the first 3-4 min. In the oneremaining rat (subject 6), stage 2 was notobserved until 47 min after injection.

Stage 4: recovery In 9/11 rats, followingstage 3, the EEG was dominated by slowwaves of high amplitude, similar to substage2c (Fig IF-H). Activity of higher frequencyand lower amplitude, resembling substages2a and 2b, appeared superimposed on theslow waves, but no consistent sequence orintensity of occurrence between individualrats could be detected by visual inspection.The slow waves (substage 2c-type) remainedthe dominant activity up until the end ofeach experiment. In 1/11 rats (subject 6), nodifference could be detected between sub-stage 2c and stage 4. In one other rat (subject8), the presence of stage 4 activity could not

Stage 3: paroxysmal EEG In 10/11 rats,stage 2 was followed by paroxysmal EEG,characterized by the irregular occurrence ofsharp high voltage (about 500-1500 /-LVpeak-to-peak) waves ('bursts'). Bursts were eitherseparated by slow (<5-01) waves resemblingsubstage 2c (Fig ID), or by periods of electro-silence ('suppression', Fig IE). In 9/11 rats,duration of stage 3 was 20-334 (median:59) min. In one rat (subject 8) duration ofstage 3 could not be assessed due to a failureof the acquisition system. In one further rat(again subject 6), no paroxysmal EEG wasobserved. Due to the non-stationary char-acteristics of stage 3, no spectral data arepresented.

Paw pinch reflex PPR was lost in 7/11 ani-mals. Time of loss was 8-34 (median: 10) minafter injection. PPR remained absent for25-74 (median: 42) min. Loss of PPRoccurred 2-23 (median: 7) min after onset ofstage 3 paroxysmal EEG activity. In five rats,PPR re-occurred 7-250 (median: 21.51 minbefore cessation of stage 3 activity. In tworats, PPR reoccurred at 4 and 21 min respec-tively, after onset of stage 4 activity.

EEG observations during reflexassessment

Righting reflex RR was lost in all 11 rats. In10/11 animals time of loss occurred at 2-5(median: 41min after injection. In these 10animals, RR remained absent for 45-350(median: 96) min. In one animal, loss of RRoccurred at 47 min after induction, for aperiod of 43 min. In all rats, loss of RRoccurred during stage 2 EEG activity. Sincethe rats often spontaneously collapsed duringan EEG recording epoch, no reliable relationbetween EEG substage and loss of the reflexcould be made. However, loss of RR coin-cided with clear changes in EEG spectralparameters, predominantly the SEF95 (Fig 2).All rats regained RR, though in one rat thisoccurred during failure of the EEG acquisi-tion system. In the 10 rats in which the EEGcould be recorded, regain of RR occurredduring stage 4 EEG activity. Visual inspec-tion of the EEG elicited no differencesbetween activity before and after regain ofRR. None of the parameters derived fromspectral analysis elicited statistically sig-nificant differences between the EEG beforeand after regain of righting reflex (Fig 3), butthe EEG parameters after regain of rightingreflex statistically differed from baselineEEG.

be assessed due to a failure of the acquisitionsystem. In the nine rats in which stage 4could be detected, duration was 10-73 (med-ian: 251min.

DiscussionIn this study, evidence is presented that (1)changes in the withdrawal response to toe

Range

3.78104-1.96105

21.73-59.714.75-32.597.65-42.819.63-23.603.9-26.449.8-76.20.29--0.50

Median

6.38 104

342421168.858.60.34

*% of 1-25.4 Hz

Parameter

1-25 Hz (J.lV2)

i5 (%)*(J (%)*IX (%)*{3 (%)*mt (Hz)set (Hz)Vigilance index (J.lV2)

Depth of anaesthesiain the rat 53

500 50<>

0450_ Total power a<00-

IIHIHI!m'"

t!IIr1h!h!350-

j 300 3"-

.a 250-

; ZOO- 200_

ISO -

100 - -.I.I "0 ·II ••SO-

DSequence In lime Sequence in time

'.' Vigilance index700_

I!hf!IdII;60'->.'

.n tt",t"1.0500_

.<

N "

....,~ B

'.' ;300

'.' ••••• • 2"•., "0 •••• I0.0

Soquence in ~meSaquenceIn timD

20600

8

hIIII!I!IHSEF95" I60 ! !!!

• 300~S5

J ("; 200 !45 !I"'00 -.I.! III!!!I!!!"

SeQ\lflnce~lirr.a Sequence IIIlme

16500 e 16

" I!t-MF

Iht1Ihhn "j 300 "11 1 I tIlfI11!11II~ 20010

100 • II!I

s.oq''''l'ICIIin time SOqueoce in lime

Fig 2 Spectral characteristics of EEGchanges during induction of anaesthesia (and concurrent loss of rightingreflex). Data are expressed as % of baseline power, except for the SEF,MF and vigilance index. In each animal[n = 10 (One rat (subject 6) in which induction was delayed for 47 min, was omitted from this graph)]. thebaseline value for each parameter was normalized as 100%. Data represent group averages (±SEM) of fourblocks of consecutive 7 s measurements each, taken immediately after injection. After the first fourmeasurements, SEF9S,p, () and total power differed significantly from control

pinching in the pentobarbital-anaesthetizedrat are not uniformly related to changes inactivity of the cerebral cortex, and that (2) theabsence of this reflex does not uniformlycoincide with EEG patterns that indicateunconsciousness.

In five out of seven rats, the cerebral cortexwas in a state of paroxysmal activity, alsotermed 'burst suppression', both during lossand subsequent regain of the paw pinchreflex. Burst suppression is a state of activityin which some brain structures such as neo-

cortex and hippocampus alternate betweenrelatively long periods of electrosilence(absence of electrical activity) and general-ized synchronous discharges. Essentially, thisis the result of alternating periods of low andhigh activity in the thalamus, enforcing itsactivity on the cortex (Mori 1987). Normalmental processes are impossible both duringperiods of electrosilence and during the massdischarges. Therefore the occurrence of burstsuppression (regardless of the amount of timespent in electro silence) excludes both con-

54

500

". !~300 t I f I1 1 !~~200

''''Total power

5eQueflce in llme

'.' --~---------,,---,:.'''''"~,_.! I I\.. ,!!.. •

'.>

Haberham et al.

'00

! t! f !l'

,200

aS'Qu."~lrltlme

13f f f f ! I.. 1

20'

Sl\!q•.•••"".,,,ljm.

SEF95

60 !g[" .!" + • !• • •2C

Sl!q~nc;ointiml!

MF

j:: ! , !! !

,'-,

ISeq~.n••••inlrme

Fig 3 Spectral characteristics of EEGchanges during regain of righting reflex. Data are expressed as % ofbaseline power, except for the SEF,MF and vigilance index, in which the control value is indicated by anasterisk(*). In each animal, the baseline value for each parameter was normalized as 100%. Data, read from leftto right, represent group averages (± SEM) of six consecutive 28-s measurements. The measurement duringwhich righting reflex was regained is indicated by an arrow. For all parameters except D, () and MF, allmeasurements significantly differed from control values (P < 0.05, Friedman's test followed by Dunn's test), butno statistical difference was found between measurements

sciousness and the formation and consolida-tion of memory. Since in the majority of theanimals in the present study burst suppres-sion continued for substantial periods of time(7-250 min) after regain of the PPR, presenceof this reflex cannot be related to one specificstate of cortical activity. Therefore, presenceof the reflex seems to depend on the amount

of suppression of activity in subcorticalstructures. The major site of action may bethe spinal cord, which is suggested by Rampil(1994) who found that the dose of isofluranerequired to suppress purposeful movement inthe rat did not change when the cervicalspinal cord was disrupted, and hence un-coupled of cerebral influence.

Depth of anaesthesia in the rat

In two rats, PPR first reappeared at 4 and21 min respectively, after cessation of burstsuppression. In contrast to the other animals,the EEG of these two rats was dominated byregular slow waves of high amplitude, inter-mingled with faster waves of lower ampli-tude, suggesting reoccurrence of a certaindegree of consciousness.

Furthermore, four rats that did not losePPR, showed periods of burst suppression,associated with a deep level of unconscious-ness. This indicates that burst suppression isnot required for abolishment of the PPR, andit provides further evidence that the absenceof the PPR does not correspond to uniformlevels of cortical depression in pentobarbital-anaesthetized rats.

These findings in pentobarbital-anaes-thetized rats support the observations byAntognini and Schwartz (1993) for the goat.These authors found that the anaestheticdose of isoflurane required for abolishment ofpurposeful nocifensive movement was higherwhen only the brain (and not the spinal cordIof the animals was anaesthetized, comparedto when the entire animal was anaesthetized.The dose required for absence of nocifensivemovement when only the brain was anaes-thetized produced burst suppression, whereasthe required dose in the whole body producedan EEG pattern similar to what is termedstage 4 activity in the present study. Thisagain demonstrates that anaesthetic actionon the spinal cord plays a major role inabolishing nocifensive movement.

It has been proposed that the dose ofbarbiturates (as well as other anaestheticsincluding isoflurane) required to produceunconsciousness and amnesia generally issubstantially lower than the dose required toabolish purposeful movement (Clark & Rosner1973, Long et al. 1989, Dwyer et al. 1992). Inhumans, consciousness is lost at the onset ofspindle [12.-14 Hz) EEG activity. Barbituratespindles are functionally related to spindlingactivity seen during the natural transitionfrom wakefulness to sleep in mammals, as aconsequence of the inhibition of thalamic cellsrelaying sensory information to the cortex.This inhibition is accomplished by rhythmicbursts of GABAergic neurones of the reticularthalamic nucleus. With slightly elevated

55

pentobarbital levels, spindle activity is con-verted into regular delta (1-4 Hz) waves,similar to the EEG pattern of non-REM sleep(Steriade et al. 1993).

The present experiments extend theseobservations to the rat. During induction dis-appearing of righting reflexes took placearound the time spindle activity occurred,which was subsequently followed by briefperiods of slower waves of high amplitude.In both phenomena, frequencies weresubstantially higher than in humans labout30 Hz during spindle activity, 4-9 Hz in theslow waves), which may be accounted for bygeneral species differences. If recovery fromanaesthesia (stage 4 EEG activity) is to be theelectroencephalographic mirror image ofinduction (stage 2 activity), the 8-9 Hz wavesobserved in the two rats lacking burst sup-pression during regain of the PPR might beequivalent to the slow waves that were brieflyobserved during induction, and hence indica-tive of sleep. Combined with the observationthat all other rats showing loss of PPRdemonstrated EEG burst suppression duringthe entire period that PPR was absent, it seemslikely that in the present study absence of PPRexclusively occurred during unconsciousness.

Conflicting with this conclusion is theobservation that during regain of rightingreflex, all rats in this study still showed EEGstage 4 activity, which could not be dis-criminated from the stage 4 activity observedprior to regain of righting reflex. Neither astatistically significant increase in high-frequency EEG activity, nor an increase inspectral edge frequency, was observed duringregain of righting reflex. This contrasts withthe decrease in spectral edge frequencyobserved during loss of righting reflex, as wellas with the observations of other investiga-tors on loss and regain of consciousness (Levy1986, Rampil & Matteo 1987, Long et al.1989). This can largely be explained by thefact that barbiturate spindles occur in thep-range. The high amplitude of the spindlesmay obscure the increase in power of theasynchronous p-activity of higher fre-quencies and much lower amplitude (Fig 1).These EEG parameters may therefore not becompletely reliable as indicators of con-sciousness during anaesthesia with spindle-

56

inducing drugs. The failure to discriminatebetween the stage 4 EEG activity in ratslacking paw pinch reflex} and rats showingrighting reflex, may simply be due to a lackin sensitivity of the EEG method used.

A second explanation for the observationsmay be that in rats, stage 4 EEG activityinvolves a certain level of consciousness. Ifthis is the case, there remains a possibilitythat consciousness reappears in rats beforePPR does. In that light, it needs to be con-sidered that barbiturate anaesthesia mayinvolve an alternative type of partial }neuro-muscular block}. Sodium pentobarbital actsas a potentiator of the inhibitory neuro-transmitter y-aminobutyric acid (GABA)}bybinding to the GABAA receptor. GABAAreceptors are widely distributed throughoutthe rat central nervous system, includingmotor neurones in the spinal cord (McKernan& Whiting 1996). Several anaesthetic agents,including barbiturates, can inhibit reflexes byinhibiting excitatory post-synaptic potentialsin motor neurons, and by raising the firethreshold in these neurons (Somjen & Gill1963}de Jong et a1. 1968}Richens 1969}King& Rampi11994} Collins et a1. 1995). Inhibi-tion of motor neurones in the spinal cordmay therefore lead to an erroneous inter-pretation of the level of anaesthesia in therat. The lack of a relation between suppres-sion of the PPR and the degree of depressionof cortical activity in the studied rats suggestinter-individual differences in the ratio ofcortical and subcortical sensitivity to barbi-turates. The possibility remains that in ani-mals with a high subcortical sensitivitycombined with a low cortical sensitivity}lack of a PPR results from depression ofmotor neurone blockade rather than depres-sion of cortical activity.

A third explanation may be found in thepossibility that rats are not fully consciousafter regain of all reflexes} and during spon-taneous walking behaviour. Decerebratedrats are capable of righting} and a wide rangeof complex (including nocifensive) beha-viours (Woods 1964, Lovick 1972). It istherefore possible that the assumption thatrats that regain righting reflex and showspontaneous walking are conscious is incor-rect. This would also explain the discrepancy

Haberham et a/.

between the EEG observations in the rats inthe present study} and in the human subjectsin the study by Tucci et a1. (19491, in whichHVSA is abolished before the subjects showany signs of consciousness. However} in thatstudy} consciousness was assumed to bepresent only when the subjects showedunequivocal signs of consciousness likeresponse to verbal command, and the possi-bility of periods of a certain degree of unde-tected consciousness occurring duringanaesthesia was not addressed.

At present} it is unclear whether intra-operative awareness under anaesthesia exclu-sively occurs when specific neuro-muscularblocking agents (NMBsl are administered inaddition to compounds with primary hypnoticand analgesic properties. When nocifensivemovements are used as stand-alone indicatorsof anaesthetic depth, paralysis induced byNMBs may prevent detection of regain ofconsciousness. Despite the fact that NMBs,which are routinely used in surgery on humanpatients and are therefore generally consideredthe primary cause of intra-operative aware-ness, the possibility remains that anaestheticaction by pentobarbital and similarly actingagents on motor neurones of the spinal cordcontributes to a state of immobility. Futureresearch should therefore be aimed at findingunequivocal indicators of absence and pre-sence of consciousness during anaesthesia inthe absence of NMBs.

In summary} it is concluded that duringbarbiturate anaesthesia in rats the paw pinchreflex occurs independently of the level ofactivity in the cerebral cortex. Although inthe present study absence of the reflex ingeneral coincided with EEG patterns sug-gesting unconsciousness} the possibilityexists that in a limited number of rats, somedegree of consciousness was present duringabsence of the paw pinch reflex. It is there-fore concluded that the paw pinch reflex isunreliable as a stand-alone indicator ofanaesthetic depth in the pentobarbital-anaesthetized rat.

Acknowledgments This study was supported bygrant PAD 94-30 of the Dutch Platform Alternativesto Animal Experiments (PAD). The authors wish to

Depth of anaesthesia in the rat

thank Mr Kees Brandt (Central Laboratory AnimalFacility, Utrecht University, The Netherlands) forhis assistance in the instrumentation of theanimals.

References

Antognini JF, Schwartz K (1993) Exaggerated anaes-thetic requirements in the preferentially anesthe-tized brain. Anesthesiology 79, 1244-9

Bertens APMG, Booij LHDT, Flecknell PA, Lagerweij E(1993) Anaesthesia, analgesia and euthanasia. In:Principles of Laboratory Animal Science (vanZutphen LFM, Baumans V, Beynen AC, eds).Amsterdam: Elsevier, pp 267-98

Clark DL, Rosner BS (1973) Neurophysiologic effectsof general anesthetics. Anesthesiology 38, 564-82

Collins JG, Kendig IT, Mason P 119951Anestheticactions within the spinal cord: contributions to thestate of general anesthesia. Trends in Neuro-sciences 18, 549-53

Davis M (19921 The role of the amygdala in condi-tioned fear. In: The Amygdala: NeurobiologicalAspects of Emotion, Memory, and Mental Dys-function (Aggleton W, ed). New York: Wiley-LissInc., pp 255-305

De Jong RH, Robles R, Corbin RW, Nace RA (19681Effect of inhalation anesthetics on monosynapticand polysynaptic transmission in the spinal cord.TournaI of Pharmacology and ExperimentalTherapeutics 162, 326-30

Dwyer R, Bennett HL, Eger EI, Heilbron D 11992)Effects of isoflurane and nitrous oxide in subanes-thetic concentrations on memory and responsive-ness in volunteers. Anesthesiology 77, 888-98

Flecknell PA 11996) Laboratory Animal Anaesthesia.London: Academic Press

King BS, Rampil IT(1994) Anesthetic depression ofspinal motor neurons may contribute to lack ofmovement response to noxious stimuli. Anesthe-siology 81, 1484-92

Levy WJ 119861Power spectrum correlates of changesin consciousness during anesthetic induction withenflurane. Anesthesiology 64, 688-93

Liu WHD, Thorp TAS, Graham SG, Aitkenhead AR119911Incidence of awareness with recall duringgeneral anaesthesia. Anaesthesia 46, 435-7

Long CW, Shah NK, Loughlin C, Spydell T,Bedford RF(1989) A comparison of EEG determinants of near-

57

awakening from isoflurane and fentanyl anesthe-sia. Anesthesia and Analgesia 69, 169-73

Lovick TA (1972) The behavioural repertoire ofprecollicular decerebrate rats. TournaI of Physiology226, 4-6

McKernan R, Whiting PJ (19861 Which GABAA'receptor subtypes really occur in the brain? Trendsin Neurosciences 19, 139-43

Mori K (1987) The EEG and awareness duringanesthesia. Anaesthesia 42, 1153-5

Rampil IT119941Anesthetic potency is not alteredafter hypothermic spinal cord transection in rats.Anesthesiology 80, 606-10

Rampil IT,Laster MJ (1992) No correlation betweenquantitative electroencephalographicmeasurements and movement response to noxiousstimuli during isoflurane anesthesia in rats.Anesthesiology 77, 920-5

Rampil IT,Mason P, Singh H (1993) Anestheticpotency (MAC) is independent of forebrain struc-tures in the rat. Anesthesiology 78, 707-12

Rampil IT,Matteo RS 11987) Changes in EEG spectraledge frequency correlate with thc hemodynamicresponse to laryngoscopy and intubation. Anesthe-siology 67, 139-42

Richens A (1969) The action of general anestheticagents on root responses of the frog isolated spinalcord. British TournaI of Pharmacology 36,294-311

Shimazono Y, Okuma T, Fukuda T, Hirai T,Yamamasu E (1953) An electroencephalographicstudy of barbiturate anesthesia. Electroencephalo-graphy and Clinical Neurophysiology 5, 525--32

Somjen GG, Gill M (1963) The mechanism ofblockade of synaptic transmission in the mamma-lian spinal cord by diethyl ether and by thiopental.Journal of Pharmacology and ExperimentalTherapeutics 140, 19-30

Steriade M, McCormick DA, Sejnowski TJ (1993)Thalamocortical oscillations in the sleeping andaroused brain. Science 262, 679-85

Tucci JH, Brazier MAB, Miles HHW, Finesinger JE119491A study of penthothal sodium anesthesiaand a critical investigation of the use of succinateas an antidote. Anesthesiology 10, 25-39

Vanderwolf CH (1992) The electrocorticogram inrelation to physiology and behavior: a new analysis.Electroencephalography and Clinical Neuro-physiology 82, 165-75

Woods JW 119641Behaviour of chronic decerebraterats. TournaI of Neurophysiology 27, 635-44