irreversibility of kindled alcohol-withdrawal behaviour … · irreversibility of kindled...

Alcohol & Alcoholism Vol. 33, No. 3, pp. 230-243, 1998

IRREVERSIBILITY OF KINDLED ALCOHOL-WITHDRAWAL BEHAVIOURIN RATS

JAKOB ULRICHSEN*, STEVEN HAUGB0L, CHRISTOPHER F. BRANDT,PETER ALLERUP1 and RALF HEMMINGSEN2

Neuropsychiatric Research Group, Department of Psychiatry — 6214, Rigshospitalet, Blegdamsvej 9, DK-2100 Copenhagen,'The Danish National Institute for Educational Research, Hermodsgade 28, DK-2200 Copenhagen and 2Department of

Psychiatry, Bispebjerg Hospital, Bispebjerg Bakke 23, DK-2400 Copenhagen NV, Denmark

(Received 23 June 1997; in revised form 17 December 1997; accepted 5 January 1998)

Abstract — In order to investigate whether alcohol-withdrawal kindling is an irreversible process, maleWistar rats were exposed to 12 episodes, each consisting of 2 days of severe alcohol intoxication and 5days of alcohol withdrawal. Spontaneous withdrawal seizures were found in 15% of the animals duringepisodes 10-12. After an alcohol-free period of 26 days, the animals were subjected to three moreepisodes of alcohol dependence (i.e. episodes 13-15) in which 12% of the animals developedspontaneous withdrawal seizures. Based on several statistical tests, we concluded that there was no truedifference between the seizure activity in episodes 10-12 and episodes 13-15, indicating that alcohol-withdrawal kindling is a long-lasting and perhaps irreversible process. In a second experiment, analcohol-withdrawal kindled group was first exposed to seven episodes of alcohol dependence. Adiazepam group went through the same alcohol regimen, but each withdrawal reaction was blocked bydiazepam treatment. Finally, a single episode group was included which was fed isocalorically with thekindled animals. After an alcohol-free period of 11 days, all three groups were subjected to 4 days ofsevere alcohol intoxication. During the subsequent withdrawal reaction seizures were observed in22—26% of the animals with no significant differences across the groups. These results call for amodification of the kindling hypothesis of alcohol withdrawal and suggest that kindling-inducedalterations may be overlooked if convulsive behaviour is tested during a relatively strong withdrawal

INTRODUCTION

The 'kindling hypothesis' of alcohol withdrawalput forward by Ballenger and Post (1978) hasrecently gained considerable support from bothclinical (Brown et al., 1988; Lechtenberg andWorner, 1992; Booth and Blow, 1993; Moak andAnton, 1996) and animal (Becker and Hale, 1993;Becker, 1994) studies. These researchers specu-lated that alcohol withdrawal serves as a kindling-like stimulus that, upon repetition, comes toproduce a progressive increase in the severity offuture withdrawal reactions. The term 'kindling'refers to a process in which repeated subthresholdelectrical stimuli to the brain eventually result infull motor seizures (Goddard et al., 1969). In ourlaboratory, we have performed a series of experi-

*Author to whom correspondence should be addressed.

ments in which the results are consistent with thekindling hypothesis of alcohol withdrawal. Thusrepeating withdrawal from 2 days of severealcohol intoxication has been shown to kindlerats such that the convulsive withdrawal behaviouris augmented (Clemmesen and Hemmingsen,1984; Ulrichsen et al, 1992, 1995). In thesestudies, the withdrawal seizure activity was testedfollowing 2 days of alcohol intoxication. In arecent investigation in which we used the samealcohol-withdrawal kindling procedure, the seiz-ure activity was measured during withdrawal from4 days of chronic alcohol intoxication (Ulrichsenet al., 1996). To our surprise, we detected asignificant decrease in spontaneous seizure activ-ity in the alcohol-withdrawal kindled groupcompared to rats which for the first time werewithdrawing from chronic alcohol intoxication. Asthis result was in disagreement with both ourprevious results (Clemmesen and Hemmingsen,

230

© 1998 Medical Council on Alcoholism

at Pennsylvania State University on February 23, 2013

http://alcalc.oxfordjournals.org/D

ownloaded from

http://alcalc.oxfordjournals.org/

ALCOHOL-WITHDRAWAL KINDLING 231

1984; Ulrichsen et al., 1992, 1995) and most of theprevailing literature on alcohol-withdrawal kind-ling, one of the aims of the present study was toreproduce the result of Ulrichsen et al. (1996).

Another object of the present investigation wasto study whether alcohol-withdrawal kindling is areversible process. To our knowledge, this issuehas only been addressed in two studies. In ratsintermittently exposed to alcohol for 60 days, areduction in the seizure threshold to pentylene-tetrazol and a decrease in central y-aminobutyricacid (GABA) function, which lasted at least 40days, have been reported (Kokka et al., 1993;Kang et al., 1996), suggesting that the cerebralalterations induced by alcohol-withdrawal kind-ling are in fact irreversible.

MATERIALS AND METHODS

Animals and intoxication procedure

Male Wistar rats (M0llegaard, K0ge, Denmark)were housed in a room with a 12 h light/12 h darkcycle (lights on at 07:00) and had free access tofood pellets and water. The initial body weight ofrats was 240-270 g in the single episode experi-ment and 145-180 g and 210-260 g in the first andsecond alcohol-withdrawal kindling experimentsrespectively (see below). During each intoxicationperiod, alcohol was administered five times a daybetween 08:00 and 24:00 by the intragastricintubation method (Majchrowicz, 1975). Beforeeach feeding session, the degree of alcoholintoxication was assessed by using the followingrating scale:

(0) neutrality: no signs of intoxication; (1)sedation: reduced muscle tone, dulled appearance,and slow locomotor activity, but no impairment ofgait or coordination; (2) ataxia 1: slight gaitimpairment and slight motor incoordination, butable to elevate abdomen and pelvis; (3) ataxia 2:clearly impaired staggering gait and impairedmotor coordination, some elevation of abdomenand pelvis; (4) ataxia 3: slowed righting reflex,heavily impaired motor coordination, no elevationof abdomen and pelvis; (5) loss of righting reflex(LRR): unable to right itself when placed on itsback, other reflexes still present; (6) coma: nosigns of movement; no response to pain stimuli; noblinking reflex; spontaneous breathing.

The alcohol dose was adjusted individually to

the degree of intoxication. Neutral rats received5-7 g/kg, whereas animals with LRR received0-1 g/kg. The aim was to reach an intoxicationlevel of 3-5 (ataxia 2-LRR) within 8h andthereafter maintaining this level for the rest ofthe intoxication period. The alcohol solutionconsisted of: ethanol (200 g/1), sucrose (300 g/1),and multivitamin mixture (4 ml/1) in Ringer'ssolution, the sucrose being added in order toprevent hypoglycaemia and ketosis (Hemmingsenand Chapman, 1980). Control animals received anisocaloric amount of sucrose instead of alcoholand the same amount of water and food as thealcohol-treated rats.

Blood-alcohol concentration was measured inblood collected from the retro-orbital sinus withheparinized capillary tubes using a computerizedgas chromatography method (Felby and Nielsen,1981).

Alcohol withdrawal

Detection of spontaneous withdrawal seizureactivity was done by two persons who observedthe animals 9-15 h after the last alcohol dose. Thenon-convulsive withdrawal behaviour wasassessed as previously described (Ulrichsen etal., 1986). Briefly, the three individual itemsintentional tremor, rigidity, and hyperactivity/irritability were each scored on a four-level scale(0-3) and the sum of these scores (0-9) was usedas a quantitative measure of the severity of thenon-convulsive withdrawal reaction. Measure-ments of both the convulsive and non-convulsivewithdrawal behaviour were all done under blindconditions, except for the rating of the non-convulsive withdrawal symptoms in the singleepisode experiment (see below). The non-con-vulsive withdrawal reaction was only measured inanimals in which seizures had not been observed,in order to ensure that the results were not affectedby postictal hypoactivity.

Design

In order to demonstrate how spontaneousseizure activity during the alcohol-withdrawalreaction depends on the length of the previousintoxication period, we performed a single episodeexperiment in which three groups of animals wereexposed respectively to 2, 3, and 4 days of alcoholintoxication (n = 30 in each group). The experi-ment was set up such that the groups were



ownloaded from


232 J. ULRICHSEN et al.

withdrawn from intoxication at approximately thesame time, i.e. alcohol administration in thegroups exposed to 2 and 3 days of alcoholintoxication began at the third and second daysof alcohol administration in the 4-day group,respectively. Blind rating of spontaneous seizureactivity during the withdrawal reaction wasperformed by two persons 9-15 h after the lastalcohol dose. Subsequently, the group code wasbroken and the non-convulsive withdrawal reac-tion rated openly 16-18 h into withdrawal on eightto 10 animals from each group, which wererandomly selected from the animals in whichseizures were not observed.

Alcohol-withdrawal kindling was performed bysubjecting the animals to multiple episodes, eachconsisting of 2 days of alcohol intoxicationfollowed by 5 days of alcohol withdrawal(Clemmesen and Hemmingsen, 1984). Two al-cohol-withdrawal kindling experiments were con-ducted. In the first alcohol-withdrawal kindlingexperiment, the animals (n = 80) were firstsubjected to 12 episodes of alcohol dependence.Following an alcohol-free period of 26 days, theanimals were subjected to three more episodes ofalcohol dependence (i.e. episodes 13-15). Theisocalorically fed control animals (n = 60) weredivided into two groups which were exposed toalcohol either during episodes 10—12 (n = 30) orduring episodes 13-15 (n = 30). Spontaneousseizure activity was measured during the with-drawal reactions in episodes 10-12 and 13-15. Bylimiting the number of withdrawal episodes in thecontrol animals to three, the risk of kindlingprocesses occurring in these animals was kept at aminimum (Clemmesen and Hemmingsen, 1984).

Three groups were run in parallel in the secondalcohol-withdrawal kindling experiment. Onegroup was subjected to seven episodes of alcoholdependence (the alcohol-withdrawal kindledgroup, n = 40), while another group (the diazepamgroup, n = 40) was exposed to the same alcoholregimen, but following each alcohol intoxicationepisode, the withdrawal reactions were blocked bydiazepam treatment (see below), which has beenshown recently to prevent alcohol-withdrawalkindling processes (Ulrichsen et al., 1995). Thethird group, i.e. the single episode group (n — 25),was isocalorically fed with the alcohol-withdrawalkindled group during episodes 1-7, with sucrosereplacing alcohol. Following the seventh intoxica-

tion period, all three groups went through analcohol-free recovery period of 11 days. Subse-quently, all groups were exposed to a final episodeof 4 days of severe alcohol intoxication (i.e.episode 8). During the following withdrawalreaction, we measured the spontaneous withdrawalseizure activity in all three groups 9-15 h after thelast alcohol dose. All 'seizure animals' and n = lrandomly selected 'non-seizure animals' weredecapitated for receptor studies (to be presentedin a separate paper). In the remaining animals thenon-convulsive withdrawal reaction was rated16-18 h after the last alcohol dose.

Diazepam treatment

Blocking the withdrawal reactions in the secondalcohol-withdrawal kindling experiment (seeabove) was achieved by administering a diazepamsolution (5 mg/ml) intraperitoneally (i.p.) 8, 11and 15 h after the last alcohol dose, as recentlydescribed (Ulrichsen et al., 1995). The intoxica-tion level of the animals was assessed before eachdiazepam administration by employing the samerating scale (0-6) as that used to evaluate thealcohol intoxication level (see above) (Majchro-wicz, 1975). At various times, intoxicated animalsreceived 0-25 mg/kg diazepam i.p., depending onthe intoxication score. Animals which showed anysigns of alcohol withdrawal received 30 mg/kgdiazepam i.p. An open rating of the non-convulsive withdrawal reaction and level ofalcohol and/or diazepam intoxication in thealcohol-withdrawal kindled group and the diaze-pam group was performed prior to the seconddiazepam administration (i.e. 11 h after the lastalcohol dose) during episode 5.

Statistical analysis

The correlation of the duration of chronicalcohol intoxication with the severity of the non-convulsive withdrawal reaction in the singleepisode experiment was assessed by testing thehypothesis that the Spearman's rank correlationcoefficient rs was not different from zero (Ho'.rs = 0) (Altman, 1991). Simple linear regressionswere used to analyse the relation between thediazepam dose and the alcohol dose and therelation between the diazepam dose and theseverity of the withdrawal reaction in episode 8.In the second alcohol-withdrawal kindling experi-ment, possible group effects on the non-convulsive



ownloaded from



withdrawal reaction were analysed by non-para-metric techniques (Mann-Whitney or Kruskal—Wallis test).

Analysis of data from the seizure observationsin the single episode experiment and the secondalcohol-withdrawal kindling experiment was doneby ordinary yj tests for contingency tables (Alt-man, 1991). In the single episode experiment, thehypothesis to be tested was that the number ofwithdrawal seizures was not affected by theduration of the previous alcohol intoxicationperiod. As the three groups obviously could beordered according to the duration of the chronicalcohol intoxication, the / 2 test for trends wasapplied in this experiment. The scores attributed tothe three groups were equal to the number of daysthe animals were exposed to alcohol, i.e. 2, 3, and4. In the second alcohol-withdrawal kindlingexperiment, the #2 test for a 2 x 3 frequencytable was employed to test the hypothesis thatthere was no difference in the seizure activitybetween the three groups.

The analysis of seizure activity in the firstalcohol-withdrawal kindling experiment was per-formed as previously reported (Ulrichsen et al.,1995) with modifications. Two differentapproaches were applied. First, the numbers ofseizures were analysed as traditional survivalcurves: tests of significant differences in seizureactivity between two groups were based on theKaplan—Meier estimated survival curves using theordinary Mantel-Haenszel test statistic (^2) (logrank test) (Matthews and Farewell, 1988). Second,based on the convulsive behaviour observations,we used the Monte Carlo simulation technique inorder to obtain confidence limits for the prob-ability of spontaneous seizure activity duringepisodes 10-12 and 13-15 in the alcohol-with-drawal kindled animals and the respective controlanimals (see below); the groups were comparedwith respect to this variable. The reason forconducting the Monte Carlo simulations besidesthe ordinary point value Mantel-Haenszel teststatistic was that preliminary calculations showedthat the Mantel-Haenszel test in some situationswas not conclusive, i.e. the significance prob-ability was close to 0.05, and that the statisticalassumptions, e.g. independence between the riskof death and the risk of seizure (which are part ofthe assumptions behind the Mantel-Haenszel test),could be explicitly built into the Monte Carlo

simulations. Finally the Monte Carlo simulationtechnique allowed us to obtain interval estimatesfor the variability of the convulsive withdrawalbehaviour, whereas the outcome value of theMantel-Haenszel test is always a single signifi-cance probability value.

In the following, the calculation of the prob-ability of spontaneous seizure activity is onlydescribed for the alcohol-withdrawal kindledgroup during episodes 10-12, as the sameprocedure was used for calculating this variablein the alcohol-withdrawal kindled group duringepisodes 13-15 and in the two control groups. Thecalculation of the probability of spontaneousseizure activity across all episodes (Pcum) wasbased on a splitting of the simultaneous prob-ability of no seizures during episodes 10-12 [lefthand side of (1)] into successive conditionalprobabilities of observing no seizures in eachindividual episode [right hand side of (1)]:

JXO12.O1 i,O,o) = ^0,210, ,,010) x />(0,, 10,o)xP(0,o) (D

where 0x indicates 'zero seizures at episode x'.The three probabilities on the right-hand side of

(1) were estimated via binomials as the relativefrequencies of observed non-seizures at the givenepisode, where the numbers of animals at risk fordeveloping 'first time seizures' (AO were adjustedaccording to the conditions given in the threeconditional probabilities — i.e. the number ofobserved seizures during earlier episodes weresuccessively taken out of N; likewise the numberof deaths (due to alcohol intoxication) within eachepisode were taken out (subtracted) of N. Hence,only animals going into withdrawal which had notpreviously shown convulsive activity wereincluded in the analysis. The cumulated prob-ability Pcum of (at least one) spontaneous seizureduring episodes 10-12 was then obtained from (1),simply as

^cum=l -^ (0 ,2 ,0 , !,010) (2)

In order to analyse statistically and compare thedifference in Pcum between the groups, 2000Monte Carlo simulations of complete episode10-12 observations were generated with n = 57 asthe initial value. In each episode, two independentbinomials, i.e. the probability of first time seizureand death, were used for the generation of thenumber of seizures and the number of deaths. The



ownloaded from



episode-specific probabilities of seizure and deathwere simply based on the observed seizure anddeath frequencies in each episode, respectively {Nadjusted successively from one episode to the nextas described above). Based on the distribution ofsimulated Pcum values within each group, thedifference in seizure activity was then evaluatedstatistically by means of confidence limits.

All tests were two-sided and the overall level ofsignificance was set everywhere to 5%. Data arepresented as means with SEM or median withrange.

Ethics

The study was approved by the Danish AnimalExperiment Inspectorate, Ministry of Justice.

RESULTS

Single episode experiment

There was a mean (SEM) weight loss of 13.8%(0.3), 16.0% (0.5), and 18.9% (0.5) in the groupsexposed respectively to 2, 3, and 4 days of chronicalcohol intoxication, and the numbers of animalsin these groups were reduced from 30 to 29, 24and 23, respectively. Both the weight loss, whichis most likely due to dehydration secondary to thediuretic effects of ethanol (see Hemmingsen andChapman, 1980), and the deaths which occurredbecause of overdosage of alcohol, are unavoidablecomplications of the present alcohol administra-

tion technique (Majchrowicz, 1975). The alcoholdose, mean intoxication score and withdrawalbehaviour are presented in Table 1. There was astatistically significant effect of duration of al-cohol administration on the severity of both thenon-convulsive and the convulsive withdrawalbehaviour. We did not detect any seizures in theanimals exposed to 2 days of alcohol intoxication,whereas seizures were detected in 21% and 35%of the animals exposed to 3 and 4 days of alcoholintoxication, respectively.

Alcohol-withdrawal kindling experiment 1

The alcohol dose, mean intoxication score andblood-alcohol concentration during alcohol depen-dence episodes 10-12 and 13-15 are shown inTable 2. During the kindling regimen (i.e.episodes 1-9), the number of animals in thealcohol-withdrawal kindled group was reducedfrom 80 to 57. The numbers of animals which diedduring episodes 10-15 are shown in Table 3. Alldeaths were due to overdosage of alcohol. Duringthe 15 episodes of alcohol dependence, a netweight increase of 105% and 101% was detectedin the alcohol-withdrawal kindled animals and thecontrol animals, respectively. In the alcohol-with-drawal kindled group spontaneous seizures wereobserved 14 times during episodes 10-12 and ninetimes during episodes 13-15, whereas no seizureswere observed in the respective control groups.During episodes 10-12, six of the 14 seizuresoccurred in alcohol-withdrawal kindled animals

Table 1. Alcohol dose, intoxication score and the severity of the withdrawal reaction during a single episode of chronicalcohol intoxication

Withdrawal behaviourDurationof alcoholintoxication(days)

Daily alcohol dose(g/kg/day)

Mean SEM

Mean intoxicationscore (0-7)

Mean SEM

Non-convulsive (0-9)

n Median Range

Convulsive

n No. of seizures

2 10.6 0.2 3.4 0.043 10.8 0.3 3.3 0.064 10.1 0.2 3.4 0.04

Statistical analysis:

10

10

2.54.56

(1-6)(4-6)(3-8)

292423

02135

Spearman's rs = 0.757P < 0.002

A e n d = H 0 9 , d f = 2P < 0.01

Rats were exposed to a single episode of 2, 3, and 4 days of severe alcohol intoxication. During the withdrawal reac-tion, the animals were observed blindly 9-15 h after the last alcohol dose in order to detect spontaneous seizure ac-tivity. Subsequently, the code was broken and the non-convulsive withdrawal reaction assessed openly on 8-10animals from each group which were randomly chosen among the animals in which no seizures were observed. Seetext for other details.



ownloaded from



Table 2. Alcohol intoxication variables in alcohol-withdrawal kindled rats

Group

Episodes 10-12Alcohol-withdrawalControl

Episodes 13-15Alcohol-withdrawalControl

kindling

kindling

n

5228

4425


Mean

12.711.2

11.711.0

SEM

0.30.4

0.30.7

Mean intoxicationscore

Mean

2.92.8

2.72.8

(0-6)

SEM

0.030.04

0.080.09

Blood-alcoholconcentration

(g/1)

Mean

4.033.76

3.653.68

SEM

0.190.21

0.200.52

Intoxication scoreat time <

sample

Mean

4.24.2

3.84.2

if blood•(0-6)

SEM

0.30.3

0.40.4

Rats were first subjected to 12 episodes of alcohol dependence. Following an alcohol-free period of 26 days, the ani-mals were subjected to three more episodes of alcohol dependence (i.e. episodes 13-15). Isocalorically fed controlanimals were exposed to alcohol either during episodes 10—12 or 13-15. Spontaneous seizure activity was measuredduring the withdrawal reactions in episodes 10—12 and 13-15. n indicates the number of animals surviving episodes10-12 and 13-15, respectively. Blood-alcohol concentration was measured on the second day of alcohol intoxicationin episodes 10 and 15 in six randomly selected animals from each group.

which had previously had at least one seizure, andhence the number of first time seizures was eight.Similarly, in episodes 13-15, three of the nineseizures were observed in alcohol-withdrawalkindled animals in which convulsive behaviourhad already been detected, yielding a total of sixfirst time seizures. All these six seizures occurredin animals in which seizures had been observed

during episodes 10-12. In the two animals whichshowed seizure activity during episodes 10-12 butnot in episodes 13-15, we observed seizureactivity in only one episode. The occurrence offirst time seizures and deaths during episodes10-12 and 13-15 are presented in Table 3. Basedon the figures in this Table, the probability of firsttime seizure in each specific episode could be

Table 3. Number of first time withdrawal seizures and deaths during alcohol dependence episodes 10-12 and 13-15

Group

Episodes 10-12Alcohol-withdrawal kindling

Control

Episodes 13-15Alcohol-withdrawal kindling

Control

Episode

101112101112

131415131415

No. ofanimals

575654303030

525049302726

Deaths

122002

215311

Previousseizuresu v l b w l WftJ

(cumulated)

027000

025000

Animals atriskW

565245303028

504739272625

Spontaneous seizures(first time)

Yes

251000

231000

No

544744303028

484438272625

Rats were first subjected to 12 episodes of alcohol-dependence. Following an alcohol-free period of 26 days, the ani-mals were subjected to three more episodes of alcohol dependence (i.e. episodes 13-15). Isocalorically fed controlanimals were exposed to alcohol either during episodes 10—12 or 13-15. Spontaneous seizure activity was measuredduring the withdrawal reactions in episodes 10-12 and 13-15. Deaths were caused by ethanol intoxication. Animalsat risk were defined as animals going into withdrawal in which seizure activity had not previously been observed.



ownloaded from



estimated using simple binomials. Thus theprobabilities of first time seizure in episodes10-12 were estimated as 2/56 = 0.04,5/52 = 0.10, and 1/45 = 0.02, respectively, in thealcohol-withdrawal kindled group, and 0/30 = 0,0/30 = 0, and 0/28 = 0 respectively in the controlgroup. When the seizure activity in the animalswas re-tested during episodes 13-15, the prob-abilities of first time seizure in the alcohol-withdrawal kindled group were estimated as2/50 = 0.04, 3/47 = 0.06, and 1/39 = 0.03 duringepisodes 13, 14 and 15 respectively. In the controlgroup, the probabilities of first time seizure wereestimated as 0 /27=0, 0/26 = 0, and 0/25 = 0respectively. The episode-specific probabilities ofdeath were estimated in a similar manner. Forinstance, the probabilities of death in episodes10-12 were 1/57 = 0.02, 2/54 = 0.04, and2/47 = 0.04 respectively in the alcohol-withdrawalkindled group. Subsequently, the episode-specificseizure probability and death probability of thealcohol-withdrawal kindled group during episodes10-12 and 13-15 were included in two indepen-dent binomials to generate Monte Carlo simulationexperiments. In the two control groups, theestimated seizure probability values were notused in the Monte Carlo simulation experiments,because zero values inevitably lead to constantMonte Carlo generation of no seizures. However,the lack of seizures in the two control groups ofn = 30 justifies the assumption that the true Pcum

value (the probability of at least one seizure inepisodes 10-12) lies in the interval 0-1/60(0-0.0167). /'cum w a s therefore arbitrarily set to0.01 in both control groups. Assuming that theseizure probability in the control groups wasconstant across each episode (Pepi)> it can beshown (Ulrichsen et al., 1995) that

Pepi(control) = 1 - (1 - Pcum) I /3 or/^(control) = 1 - (1 - 0.01)1/3 = 0.00334.

The distribution of 2000 simulated Pcum valuesin the alcohol-withdrawal kindled and the controlgroups during episodes 10-12 and 13-15 is shownin Fig. 1. The simulation results are summarized inTable 4 together with the ordinary Mantel-Haenszel tests. The main output from the simula-tions is the 95% confidence limits of Pcum- Theseconfidence limits form the basis for intergroupcomparisons together with the ordinary log ranktests (Mantel-Haenszel tests). During episodes

10-12, the log rank test resulted in a significantdifference in seizure activity between the alcohol-withdrawal kindled group and the control group(X2 = 4.23, d f = l , P<0.05). This finding wassupported by the non-overlapping 95% confidenceintervals [0.056, 0.241] and [0.000, 0.039] in thealcohol-withdrawal kindled group and the controlgroup, respectively. During episodes 13-15, thelog rank test did not result in a significantMantel-Haenszel test statistic (#2 = 3.40, df = 1,P = 0.066 > 0.05). The low P-value, however,indicates that the difference in the seizure activitybetween the two groups is a true finding. Inagreement with the low, but non-significant,P-value of the log rank test (P = 0.066), a smalloverlap was found between the 95% confidenceintervals of the alcohol-withdrawal kindled group([0.040, 0.224]) and the control group ([0.000,0.065]). No significant difference between theseizure activity in the alcohol-withdrawal kindledgroup during episodes 10-12 and 13-15 wasfound in the log rank test (^2 = 0.11, d f = l ,P > 0.05), a finding which coincides with thegreat overlap between the two 95% confidencelimits ([0.056, 0.241] and [0.040, 0.224], respec-tively).

Alcohol-withdrawal kindling experiment 2During the alcohol-withdrawal kindling regi-

men of the second alcohol-withdrawal kindlingexperiment, the number of animals was reducedfrom initially 40 to 25 and 32 in the alcohol-withdrawal kindled group and the diazepam grouprespectively. In the subsequent challenge episode(episode 8), all animals from these two groupssurvived, whereas six of the 25 animals in thesingle episode group died. All deaths were due tooverdosage of alcohol. During episodes 1-7 andthe 11-day alcohol-free period, a net weight in-crease of 22.6%, 19.6%, and 26.9% was detectedin the alcohol-withdrawal kindled group, thediazepam group, and the single episode grouprespectively.

The alcohol dose, mean intoxication score, andblood-alcohol concentration during episode 8 arepresented in Table 5. During the subsequentwithdrawal reaction, seizures were detected infive out of the 19 animals from the single episodegroup (26%), six out of the 25 alcohol-withdrawalkindled animals (24%), and seven out of the 32animals (22%) from the diazepam group. No



ownloaded from



900

800

700

600

500

400

300

200

100

0

(A)

0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45 0.50

Seizure probability (Pcu.)

0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45 0.50

Seizure probability (Pcua)

1600 i (D )

0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45 0.50


0.00 0.05 0.10 0.15 0.20 0.25 0.30 0 .35 0.40 0 .45 0.50


Fig. 1. Distribution of the probability of a first time spontaneous seizure (Pcum) during episodes 10-12 (A, B) and 13-15 (C,D) as a result of 2000 Monte Carlo simulations of seizure events in the alcohol-withdrawal kindled groups (A, C) and the

control groups (B, D).The lower limits of Pen, intervals of 0.05 are shown on the abscissae. The bars represent the frequencies of Pcum in these

intervals. Note the different scaling on the ordinates. The simulation experiments were based on two independentbinomials, i.e. the probabilities of seizure and death in each individual episode, as described in the text. The alcohol-withdrawal kindled rats were first subjected to 12 episodes of alcohol dependence and after an alcohol-free period of 26days they were exposed to three more episodes of alcohol dependence (i.e. episodes 13-15). Isocalorically fed controlanimals were exposed to alcohol either during episodes 10-12 or 13-15. Spontaneous seizure activity was measured during

the wididrawal reactions in episodes 10-12 and 13-15.

significant group effects on the seizure activity single episode group (n = 7), 6 (3-7) in thewere observed. The median (range) of the non- alcohol-withdrawal kindled group (n = 11), andconvulsive withdrawal score was 5 (4-7) in the 6 (4—8) in the diazepam group (n= 18). No



ownloaded from



Table 4. Probability of spontaneous withdrawal seizures evaluated by Monte Carlo simulations

Episodes 10-12 Episodes 13-15

Simulated probabilityMeanStandard deviation

Alcohol-withdrawalkindling

I' cum/0.146370.04769

95% confidence interval [0.056, 0.241]Statistical analysis of

Alcohol-withdrawalAlcohol-withdrawalAlcohol-withdrawal

seizure activity

kindling vs control (episodes 10-12)kindling vs control (episodes 13-15)kindling (episodes 10-12) vs

alcohol-withdrawal kindling (episodes 13-15)

Control

0.009900.00575

[0.000, 0.039]


0.123690.04773

[0.040, 0.224]

Mantel-Haenszel test (log rank

x2--K2 = 4.23, df= 1, P < 0.05

Control

0.009100.00652

[0.000, 0.065]

test)

= 3.40, d f= l ,NS (P = 0.066)72 = 0.11, d f = l , NS

Rats were first subjected to 12 episodes of alcohol dependence. Following an alcohol-free period of 26 days, the ani-mals were subjected to three more episodes of alcohol dependence (i.e. episodes 13-15). Isocalorically fed controlanimals were exposed to alcohol either during episodes 10—12 or 13-15. Spontaneous seizure activity was measuredduring the withdrawal reactions in episodes 10-12 and 13-15. Distribution curves of the cumulated probability(̂ cum) of one or more seizures during episodes 10-12 or 13-15 were created by 2000 Monte Carlo simulations asdescribed in the text. NS denotes non significant.

Table 5. Alcohol intoxication variables in alcohol-withdrawal kindled rats exposed to 4 days of chronic alcoholintoxication


Mean intoxicationscore (0-6)

Blood-alcoholconcentration

(g/1)

Intoxication scoreat time of blood

sample (0-6)

Group


DiazepamSingle episode

No. ofanimals

25

3219

Mean

9.8

10.88.5

SEM

0.2

0.20.2

Mean

3.0

3.12.7

SEM

0.04

0.030.04

Mean

3.87

3.853.63

SEM

0.15

0.260.21

Mean

4.3

4.24.2

SEM

0.2

0.40.3

Rats were subjected to seven episodes consisting of 2 days of alcohol intoxication and 5 days of alcohol withdrawal. Inthe alcohol-withdrawal kindled group, clinical withdrawal symptoms developed following each intoxication episode.In the diazepam group, withdrawal episodes 1-7 were blocked by diazepam administration 8, 11, and 15 h after thelast alcohol dose. Following 11 alcohol-free days, these two groups and the single episode group which was iso-calorically fed with the alcohol-withdrawal kindled group during episodes 1-7 were exposed to 4 days of severechronic alcohol intoxication. Blood-alcohol concentration was measured in six animals from each group at the fourthday of alcohol intoxication. See the text for other details.

significant group effects on this variable werefound.

An open rating performed 11 h after the lastalcohol dose and immediately before the seconddiazepam administration in episode 5 revealed thatthe median (range) of the non-convulsive with-drawal score was 1 (0-5) and 0 (0-2) in thealcohol-withdrawal kindled group (n = 29) andthe diazepam group (n — 36), respectively,whereas the median (range) of the intoxication

score was 1 (0-5) and 3 (1-5), respectively. Thedifferences in both the withdrawal score andintoxication score were statistically significant.Diazepam was administered to the diazepamgroup three times per episode in episodes 1-7yielding a total of 21 diazepam administrations. Ofthe 32 surviving animals in the diazepam group,21 never showed any withdrawal signs and sevenanimals showed withdrawal symptoms at only oneof the 21 ratings. In the remaining four animals,



ownloaded from



withdrawal signs were detected two, four, four,and six times, respectively. No animal from thediazepam group showed withdrawal symptoms inmore than three of the seven episodes in whichthey were treated with diazepam.

Although only a weak negative linear relation-ship between the diazepam dose and the alcoholdose in episodes 1-7 was found (^ = 0.10), thespecific alcohol contribution was significant(f-test, P < 0.0001).

The mean diazepam dose per episode variedfrom 43.5 mg/kg to 68.6 mg/kg in the survivinganimals in the diazepam group. The diazepamdoses did not vary across episodes in anysystematic way. The mean diazepam dose acrossepisodes 1-7 was significantly increased in theseven animals which developed withdrawal seiz-ures in episode 8 compared to the 25 animalswhich did not develop seizures (60.4 mg/kg/episode vs 49.9 mg/kg/episode, respectively). Incontrast, a negative linear relationship between themean diazepam dose across episodes 1-7 and thenon-convulsive withdrawal score was detected(r2 = 0.18). A trend towards a specific diazepameffect was found, although statistical significanceof the diazepam contribution was not reached(Mest, P = 0.07).

DISCUSSION

The results of the current study showed thatalcohol-withdrawal kindling led to an increasedseizure activity during withdrawal after 2 days ofsevere alcohol intoxication. Thus during episodes10-12 of the first alcohol-withdrawal kindlingexperiment, all the statistical tests showed a sig-nificant increase in seizure activity of the alcohol-withdrawal kindled, compared to the control,group. This finding is in absolute agreement withearlier results from our laboratory in which thesame alcohol-withdrawal kindling paradigm wasused (Clemmesen and Hemmingsen, 1984;Ulrichsen et al., 1992, 1995) and with most ofthe results from other alcohol-withdrawal kindlingexperiments (Branchey et al., 1971; Walker andZornetzer, 1974; Baker and Cannon, 1979; Brownet al., 1988; Lechtenberg and Worner, 1992;Booth and Blow, 1993; Becker and Hale, 1993;Kokka et al., 1993; Becker, 1994; Kang et al.,1996; Moak and Anton, 1996). When the with-drawal behaviour of the kindled animals was re-

tested after a 26-day alcohol-free period duringthree subsequent withdrawal episodes (i.e. epi-sodes 13-15), the seizure activity did not differsignificantly from the seizure activity duringepisodes 10-12. During the withdrawal episodes13-15, the seizure activity in the alcohol-with-drawal kindled animals was still increased,compared to the control group, and although theP-value of the log rank test did not reach statisticalsignificance and the 95% confidence limits of thesimulated Pcum values did overlap, the small P-value (0.066) and the small overlap of the 95%confidence limits justify the conclusion that thedifference in seizure activity between the alcohol-withdrawal kindled group and the control group inepisodes 13-15 represents a true finding. The lackof seizure activity in the group subjected to 2 daysof alcohol intoxication in the 'single episodeexperiment' is consistent with this suggestion.Thus, the current results suggest that the cerebralalterations underlying alcohol-withdrawal kind-ling are long-lasting and perhaps irreversible. Thisconclusion is in line with the results of twoprevious animal studies. Kokka et al. (1993)reported a reduction in seizure threshold topentylenetetrazol lasting at least 40 days in ratswhich received a large single daily dose of alcoholfor 60 days. Using the same chronic intermittentalcohol regimen, Kang et al. (1996) found areduction in GABA (muscimol)-stimulated chlor-ide efflux in the hippocampus which persisted forat least 40 days. The issue of reversibility ofalcohol-withdrawal kindling has not been system-atically examined in clinical studies. However,both Brown et al. (1988) and Booth and Blow(1993) reported that patients with a history ofconvulsive withdrawal behaviour were at greatrisk for developing seizures during later alcohol-withdrawal reactions. Further, electrical kindling,which may share mechanisms in common withalcohol-withdrawal kindling (Pinel and Van Oot,1975; McCown and Breese, 1990), seems to be anirreversible phenomenon [reviewed by Racine(1978)]. Presently, it seems justifiable to speculatethat kindling due to alcohol withdrawal is also anirreversible process, but more studies are neededto confirm this hypothesis.

Whereas convulsive withdrawal behaviour inthe first alcohol-withdrawal kindling experimentwas examined during withdrawal from 2 days ofalcohol intoxication, the animals in the second



ownloaded from



alcohol-withdrawal kindling experiment weretested during withdrawal from 4 days of alcoholintoxication. In contrast to the results of the formerexperiment, we found no effect of alcohol-with-drawal kindling on the convulsive withdrawalbehaviour in the second alcohol-withdrawal kind-ling experiment. Using the same design, except foromission of the diazepam group, we recentlyreported that seizure activity in alcohol-with-drawal kindled animals was actually decreasedcompared to the single episode group, but due to ahigher intoxication level in the latter group weconcluded that there was no true difference inconvulsive withdrawal behaviour between the twogroups (Ulrichsen et al., 1996). Therefore theresults from the second alcohol-withdrawal kind-ling experiment are reliable, even though theydisagree with most of the previous alcohol-with-drawal kindling studies (Branchey et al, 1971;Walker and Zornetzer, 1974; Baker and Cannon,1979; Clemmesen and Hemmingsen, 1984; Brownet al, 1988; Lechtenberg and Worner, 1992;Ulrichsen et al., 1992; Booth and Blow, 1993;Becker and Hale, 1993; Kokka et al, 1993;Becker, 1994; Ulrichsen et al, 1995; Kang etal, 1996; Moak and Anton, 1996). The currentstudy clearly demonstrated that in both kindledand non-kindled animals the severity of thewithdrawal reaction was markedly increasedwhen the preceding alcohol intoxication periodwas 4, as opposed to 2, days. Thus, in the singleepisode experiment, no withdrawal seizures weredetected after 2 days of severe alcohol intoxica-tion, whereas withdrawal after 4 days of alcoholintoxication resulted in seizures in 35% of theanimals. In the first alcohol-withdrawal kindlingexperiment, in which the animals were testedduring withdrawal from 2 days of alcoholintoxication, the estimated risk of a first timeseizure during episodes 10-15 varied between 2%and 10%, whereas withdrawal seizures weredetected in 24% of the alcohol-withdrawal kindledanimals following 4 days of alcohol intoxication inthe second alcohol-withdrawal kindling experi-ment. We hypothesize that alcohol-withdrawalkindling in rats only produces subtle cerebralalterations causing a latent hyperactivity of thebrain. This hyperactivity does not produce beha-vioural alterations per se, but alterations whichcan be demonstrated if the alcohol-withdrawalkindled rats are challenged with a relatively weak

stimulus like withdrawal from 2 days of severealcohol intoxication. Exposing the rats to aconsiderably stronger withdrawal reaction, i.e.withdrawal from 4 days of severe alcoholintoxication, results in a cerebral hyperactivity sointense that it far exceeds the kindling-inducedneuroexcitability. In other words, we may theor-etically be dealing with a ceiling effect resulting inlack of detection of the kindled neuroexcitability,a speculation which is consistent with earlierresults showing that in the intragastric intubationparadigm of alcohol dependence in rats (Majchro-wicz, 1975), 4 days of alcohol intoxicationrepresent the maximal obtainable withdrawalreaction (Majchrowicz and Hunt, 1976). Thishypothesis may explain the results of Goldstein(1974), who failed to detect increased convulsivebehaviour in mice intermittently exposed toethanol vapour for 3 days, while other researchershave shown an augmented seizure activity inalcohol-withdrawal kindled mice withdrawn from16 h of ethanol vapour (Becker and Hale, 1993;Becker, 1994).

In the present study, two paradigms of alcoholdependence were employed, i.e. a single-episodemodel and a multiple-episode model. The formerresembles most of the models described in theliterature, as it uses alcohol-naive animals.Obviously, this approach does not mimic theclinical situation sufficiently well. In humans withno or limited experience with alcohol, blood-alcohol concentrations of 3-4 g/1 (300-400 mg/dl),as in the present study, are lethal. However, inextreme cases and under certain experimentalconditions, such blood-alcohol levels may actuallybe reached. This was the case in the classical studyof Isbell et al (1955), in which 10 prisoners wereoffered an unlimited amount of alcohol for 78days. It is noteworthy that one of the subjects withno history of alcohol abuse actually developeddelirium tremens during the subsequent with-drawal reaction. This finding is in disagreementwith the kindling hypothesis of alcohol withdrawaland suggests that humans too, when exposed tosevere alcohol intoxication for an extremely longperiod, may develop maximal severe withdrawalreactions even if they are not previously kindled.In general, however, humans who develop severealcohol-withdrawal reactions have been drinkingfor several years (Ballenger and Post, 1978). Theytypically do not drink continuously, as somatic



ownloaded from



complications of their high alcohol intake and/orlack of money to pay for the alcohol eventuallyforce them to terminate drinking for a while(Victor and Adams, 1953). Based on clinicalobservations in our detoxification units, many ofthese patients have developed a high degree ofalcohol tolerance and it is not unusual that theyachieve blood-alcohol concentrations of 3-5 g/1.These blood levels of alcohol are comparable tothose reported in the current study, and wetherefore find that our multiple episode model ofalcohol dependence in several ways mimics theclinical situation of severe alcohol abuse.

One disadvantage of the intragastric intubationtechnique is that this method does not allow us tocontrol for the amount of alcohol intake and thelevel of alcohol intoxication as alcohol adminis-tration is based on a subjective measure ofintoxication of each individual animal. In thefirst alcohol-withdrawal kindling experiment, wefound that both the alcohol dose and the meanintoxication score in the alcohol-withdrawalkindled group were slightly decreased in episodes13-15, compared to episodes 10-12 (Table 2).Although we did not detect any significant changein seizure activity of the alcohol-withdrawalkindled group between episodes 10-12 and13-15, this small decrease in alcohol exposureand intoxication level may explain why only sixanimals developed seizures in episodes 13—15 asopposed to eight animals in episodes 10-12,especially since the two animals which onlydeveloped seizures in episodes 10-12 only showedseizure activity in a single episode. In comparisonwith the control groups, there was a small increasein the alcohol dose administered to the alcohol-withdrawal kindled group in both episodes 10-12and episodes 13-15. We find it unlikely that thedifference in seizure activity can be attributed tothis difference in the alcohol dose, first becausethe seizure activity was considerably increased inthe alcohol-withdrawal kindled group; second, theintoxication levels were almost identical and thesmall changes were in opposite directions as themean intoxication score of the alcohol-withdrawalkindled group was increased and decreased,respectively, in episodes 10-12 and 13-15 com-pared to control levels; third, the lack of convul-sions in the control group is in completeagreement with the results of the single episodeexperiment, in which no seizures were detected

following 2 days of alcohol intoxication, whereasthe seizure activity found in the alcohol-with-drawal kindled group was similar to previouslypublished results (Ulrichsen et al, 1992, 1995). Itmay be added, that, blockade of the multiplewithdrawal reactions with diazepam (Ulrichsen etal., 1995) led to increased alcohol tolerance and ahigher alcohol intake prior to testing of theconvulsive withdrawal behaviour, compared tountreated alcohol-withdrawal kindled animals, yetthe seizure activity in the latter group wassignificantly increased (Ulrichsen et al., 1995).As for the second alcohol-withdrawal kindlingexperiment, the toxic effects of alcohol wereprobably lower in the single episode groupcompared to both the alcohol-withdrawal kindledgroup and the diazepam group, as both the alcoholdose, and perhaps more importantly the alcoholintoxication level, were decreased. Presumably, anincrease in both the alcohol dose and the meanintoxication score would have resulted in anincrease in the severity of the withdrawal reactionand withdrawal seizure activity in the singleepisode group. Therefore, we cannot totally ruleout that testing the convulsive withdrawal beha-viour following 4 days of severe alcohol intoxica-tion actually leads to a paradoxical decrease inseizure activity in alcohol-withdrawal kindledanimals, although we have no explanation ofpossible mechanisms involved in such putativealterations.

Using a similar experimental design as in thefirst alcohol-withdrawal kindling experiment, wehave recently demonstrated that the increasedwithdrawal seizure activity which occurs in ratsexposed to multiple episodes of 2 days of alcoholintoxication followed by 5 days of alcohol with-drawal can be attributed to the multiple with-drawal reactions, because diazepam treatment ofthe early withdrawal episodes (i.e. episodes 1-9)led to a decrease in the subsequent withdrawalepisodes (i.e. episodes 10-13) (Ulrichsen et al.,1995). In the second alcohol-withdrawal kindlingexperiment of the present investigation, a diaze-pam group was included in order to allow us todiscriminate between possible effects of alcoholintoxication and alcohol withdrawal during therepeated episodes of alcohol dependence. Thus, ifthe seizure activity in the alcohol-withdrawalkindled group had been changed compared to thesingle episode group, we would have attributed



ownloaded from



this change to kindling effects during the with-drawal reactions if the seizure activity in thediazepam group was similar to that in the controlgroup. If, on the other hand, the seizure activity inthe diazepam group was altered in the samemanner as in the alcohol-withdrawal kindledgroup, such a finding would indicate that alcoholtoxicity was the primary cause of the change inconvulsive behaviour. As the present results didnot show any change in seizure activity followingwithdrawal from 4 days of alcohol intoxication inthe alcohol-withdrawal kindled group compared tothe single episode group, we actually do not knowwhether kindling occurred in the second alcohol-withdrawal kindling experiment and therefore theadditional information provided by including thediazepam group was limited.

As to the effect of the diazepam treatment onthe withdrawal reaction per se, the current study isin agreement with that of Ulrichsen et al. (1995) inwhich diazepam was administered in the samemanner as in the present study. In this earlierstudy, the three preceding diazepam administra-tions clearly reduced the withdrawal reaction 16 hafter the last alcohol dose. Although the openrating of the withdrawal reaction in the presentstudy was performed before the severity of thewithdrawal reaction had reached its maximum[which is approximately 16 h after the last alcoholdose (Majchrowicz, 1975)] and at a time whendiazepam had only been administered once, westill detected a significant reduction in the with-drawal score in the diazepam group, comparedwith the alcohol-withdrawal kindled group in thepresent experiment. Moreover the 21 ratingsperformed prior to diazepam administrationsthroughout the second experiment revealed thatin 28 of the 32 animals we either saw no signs ofalcohol withdrawal or detected withdrawal symp-toms at only one rating. The increased diazepamdoses in the seven animals of the diazepam groupwhich later developed seizures points to theimportance of administering diazepam on thebasis of a clinical evaluation of each individualanimal, as the seizure animals probably are moresusceptible to alcohol-withdrawal kindling thanthe non-seizure animals.

The finding in the present study that thediazepam doses needed to block the withdrawalreactions did not vary systematically during thecourse of the alcohol-withdrawal kindling is in

agreement with previous results from our labora-tory showing that in the present alcohol-with-drawal kindling paradigm it is only the convulsivewithdrawal behaviour which is augmented, whilethe non-convulsive withdrawal reaction is unaf-fected by repetition of alcohol withdrawal (Ulrich-sen et al, 1992, 1995).

In conclusion, the present study showed thatalcohol-withdrawal kindled convulsive behaviourin rats is a long-lasting and perhaps irreversibleprocess.

Acknowledgements — This study was supported by thefollowing Danish foundations, the Foundation of IvanNielsen, the Foundation of Karen Elise Jensen, theUniversity of Copenhagen, the Beckett Foundation, theFoundation of Jacob Madsen and wife Eva Madsen, theMemorial Grant of Ove Villiam Buhl Olesen and wifeEdith Buhl Olesen, the Foundation for Research inNeurology, the Danish Hospital Foundation for MedicalResearch, Region of Copenhagen, The Faroe Islands andGreenland, the Foundation of L. F. Foght, the Foundationfor Research of Mental Disorders, and the Foundation of A.P. M0ller for promoting Medical Research. The authorsthank the technician Line Hansen for her skilful work.

REFERENCES

Altman, D. G. (1991) Practical Statistics for MedicalResearch, 4th edn. Chapman & Hall, London.

Baker, T. B. and Cannon, D. S. (1979) Potentiation ofethanol withdrawal by prior dependence. Psycho-pharmacologia (Berlin) 60, 105-110.

Ballenger, J. C. and Post, R. M. (1978) Kindling as amodel for alcohol withdrawal syndromes. BritishJournal of Psychiatry 133, 1-14.

Becker, H. C. (1994) Positive relationship between thenumber of prior ethanol withdrawal episodes and theseverity of subsequent withdrawal seizures. Psycho-pharmacologia (Berlin) 116, 26-32.

Becker, H. C. and Hale, R. L. (1993) Repeated episodesof ethanol withdrawal potentiate the severity ofsubsequent withdrawal seizures: an animal model ofalcohol withdrawal 'kindling'. Alcoholism: Clinicaland Experimental Research 17, 94—98.

Booth, B. M. and Blow, F. C. (1993) The kindlinghypothesis: further evidence from a U.S. nationalstudy of alcoholic men. Alcohol and Alcoholism 28,593-598.

Branchey, M., Rauscher, G. and Kissin, B. (1971)Modifications in the response to alcohol followingthe establishment of physical dependence. Psycho-pharmacologia (Berlin) 22, 314-322.

Brown, M. E., Anton, R. F., Malcolm, R. and Ballenger,J. C. (1988) Alcohol detoxification and withdrawalseizures: clinical support for a kindling hypothesis.Biological Psychiatry 23, 507-514.

Clemmesen, L. and Hemmingsen, R. (1984) Physical



ownloaded from



dependence on ethanol during multiple intoxicationand withdrawal episodes in the rat: evidence of apotentiation. Ada Pharmacologica et Toxicologica(Copenhagen) 55, 345-350.

Felby, S. and Nielsen, E. (1981) Automatized bloodalcohol determination with on-line computerized gaschromatography. Blutalkohol 18, 139-148.

Goddard, G. V., Mclntyre, D. C. and Leech, C. K.(1969) A permanent change in brain functionresulting from daily electrical stimulation. Experi-mental Neurology 25, 295-330.

Goldstein, D. B. (1974) Rates of onset and decay ofalcohol physical dependence in mice. Journal ofPharmacology and Experimental Therapeutics 10,377-383.

Hemmingsen, R. and Chapman, A. G. (1980) Cerebralmetabolic state during the ethanol withdrawalreaction in the rat. Journal of Neurochemistry 34,1561-1566.

Isbell, H., Fraser, H. F., Wilder, A., Belleville, R. E. andEisenman, A. J. (1955) An experimental study of theetiology of 'rum fits' and delirium tremens. Quar-terly Journal of Studies on Alcohol 16, 1-33.

Kang, M., Spigelman, I., Sapp, D. W. and Olsen, R. W.(1996) Persistent reduction of GABAA receptor-mediated inhibition in rat hippocampus after chronicintermittent ethanol treatment. Brain Research 709,221-228.

Kokka, N., Sapp, D. W., Taylor, A. M. and Olsen, R. W.(1993) The kindling model of alcohol dependence:similar persistent reduction in seizure threshold topentylenetetrazol in animals receiving chronic etha-nol or chronic pentylenetetrazol. Alcoholism: Clin-ical and Experimental Research 17, 525-531.

Lechtenberg, R. and Worner, T. M. (1992) Total ethanolconsumption as a seizure risk factor in alcoholics.Ada Neurologica Scandinavica 85, 90-94.

Majchrowicz, E. (1975) Induction of physical depen-dence upon ethanol and the associated behavioralchanges in rats. Psychopharmacologia (Berlin) 43,245-254.

Majchrowicz, E. and Hunt, W. A. (1976) Temporalrelationship of the induction of tolerance andphysical dependence after continuous intoxicationwith maximum tolerable doses of ethanol in rats.

Psychopharmacologia (Berlin) 50, 107-112.Matthews, D. E. and Farewell, V. T. (1988) Using and

Understanding Medical Statistics, 2nd edn. Karger,Basel.

McCown, T. J. and Breese, G. R. (1990) Multiplewithdrawals from chronic ethanol 'kindles' inferiorcollicular seizure activity: evidence for kindling ofseizures associated with alcoholism. Alcoholism:Clinical and Experimental Research 14, 394-399.

Moak, D. H. and Anton, R. F. (1996) Alcohol-relatedseizures and the kindling effect of repeated detoxi-fications: The influence of cocaine. Alcohol andAlcoholism 31, 135-143.

Pinel, J. P. and Van Oot, P. H. (1975) Generality of thekindling phenomenon: some clinical implications.Canadian Journal of Neurological Science 2, 467-475.

Racine, R. (1978) Kindling: The first decade. Neuro-surgery 3, 234-252.

Ulrichsen, J., Clemmesen, L., Flachs, H. and Hemming-sen, R. (1986) The effect of phenobarbital andcarbamazepine on the ethanol withdrawal reaction inthe rat. Psychopharmacologia (Berlin) 89, 162-166.

Ulrichsen, J., Clemmesen, L. and Hemmingsen, R.(1992) Convulsive behaviour during alcohol depen-dence: discrimination between the role of intoxica-tion and withdrawal. Psychopharmacologia (Berlin)107, 97-102.

Ulrichsen, J., Bech, B., Allerup, P. and Hemmingsen, R.(1995) Diazepam prevents progression of kindledalcohol withdrawal behaviour. Psychopharmacolo-gia (Berlin) 121, 451^60.

Ulrichsen, J., Woldbye, D. P., Olsen, C. H., Haugb0l, S.,Bolwg, T. G. and Hemmingsen, R. (1996) No loss ofsomatostatin-immunoreactive neurons in the hippo-campal dentate hilus of alcohol-withdrawal kindledrats. Alcohol and Alcoholism 31, 411—419.

Victor, M. and Adams, R. D. (1953) The effects ofalcohol on the nervous system. Association forResearch in Nervous and Mental Disease 32, 526-573.

Walker, D. W. and Zornetzer, S. F. (1974) Alcoholwithdrawal in mice: electroencephalographic andbehavioral correlates. Eledroencephalography:Clinical Neurophysiology 36, 233-243.



ownloaded from


irreversibility of kindled alcohol-withdrawal behaviour … · irreversibility of kindled...

Documents