Howard J. Friedman and David Lester
Center of Alcohol Studies, Rutgers University, New Brunswick, New Jersey, 08903.

* Supported in part by a grant (AA01849) from the National Institute on Alcohol Abuse and Alcoholism.

INTRODUCTION

Various sets of criteria for an animal model of alcoholism (1,2,3) have been published. Central to each is the requirement that substantial oral ingestion of alcohol be produced. Two of the sets of criteria (1,3) specify that drinking ought to produce measurable intoxication and not be associated with food deprivation; in addition, they stipulate that the animals should overcome obstacles (ranging from merely performing work to facing aversive consequences) to obtain alcohol, sometimes referred to as psychological dependence (1). The only other criteria, besides oral ingestion, which is common to all three are the consequences of chronic alcohol consumption: physical dependence and withdrawal. Other requirements which are not common among the sets of criteria specify that animals consume alcohol to prevent withdrawal (1) or that subsequent to withdrawal the animals re-initiate consumption so that intoxication and dependence are-again produced (3).

Some experimental work has been conducted concerning the effect of physical dependence on ethanol selection. Deutsch and Koopmans (4) infused approximately 9 g/kg/day of ethanol into the stomachs of rats continuously for six days; alcohol selection was enhanced thereafter for fourteen days. Physical dependence was not reported, but the intake, a mean of 4 g/kg, would not have been enough to sustain dependence. In any case, these results have not been replicable (Lester et a/. unpublished data, 1974, 1975). Myers et a/. (5) were unable to produce any increase in voluntary alcohol intake in monkeys which had been made dependent by intubation. Because the testing took place two days after intubation had ended, the procedure could not evaluate drinking to avoid withdrawal but only the effect of a period of dependence on later consumption. Begleiter (6) presented an alcohol choice to rats after a period of intubation but did not find any significant intake. In this study, it is not clear when the alcohol choice was presented, but since the animals exhibited withdrawal symptoms, we must assume that choices were presented well after the end of the intubation period and too late, therefore, to prevent withdrawal. Hunter et al. (7) found a significant increase in alcohol consumption in rats given choices between alcohol or sucrose liquid diets during withdrawal. The presentation of the choice before the onset of withdrawal plus the long experience with alcohol consumption undoubtedly contributed to these positive results. Unfortunately, the increase lasted only one to two days and some withdrawal symptoms were observed on the second day. Perhaps a more prolonged increase in consumption could be produced if the procedure also actively built up an association for the animals that alcohol consumption would prevent withdrawal illness; rats exhibited an increased consumption for ten days when they associated ethanol consumption with prevention of thiamine deficiency (8).

It is questionable whether maintenance of drinking to prevent a withdrawal syndrome needs to be a general requirement for an animal model. In some instances alcoholics have undergone partial withdrawal during voluntary administration periods (9,10); likewise, voluntary abstinence and withdrawal symptoms have been observed in rats (11) and monkeys (12). On the other hand, sometimes alcoholics have tried to avoid withdrawal symptoms by tapering off their consumption of alcohol (9,10). Until more is known about the processes underlying physical dependence, the question of the role of physical dependence in maintenance of alcohol consumption will probably remain unanswered.

Much of the research conducted before 1970, which was related to an animal model of alcoholism, seems to have concentrated on inducing consumption of alcohol (13). Little progress has been made in producing pharmacologically significant selective consumption of ethanol (cf., 8). Indeed, some procedures for the induction of alcohol consumption which had previously been regarded as promising, i.e., hypothalamic stimulation (14), intraventricular (15) and intragastric infusions (4), can no longer be so regarded (16,17) (Lester et a/., unpublished data, 1974, 1975). Perhaps because of this lack of success the number of articles on this aspect of an animal model has diminished in recent years. In contrast, an increasing amount of research conducted towards the end of contributing to an animal model has concentrated on the aspect of physical dependence and may be due to the development of useful techniques for producing the required condition. It would appear now to be an appropriate time to assess the work of the last few years and thus this review seeks to evaluate the various techniques used to produce dependence and those used to measure it.

CRITERIA FOR METHODS OF INDUCTION OF PHYSICAL DEPENDENCE

One should judge any method by its ability to produce objective signs of withdrawal. As Freund (18, p. 314) proposed, an ideal method should do so:

(a) reproducibly, (b) rapidly, (c) by simple procedures, (d) introducing the fewest variables, (e) by resulting in spontaneous major withdrawal signs and

(f) in all treated animals.

An additional, and important criterion, should be the accurate control of alcohol doses, particularly in such a manner that a scale of doses is utilized. This requirement will be assessed in the section on quantifying withdrawal symptoms. Thus far no method fulfills all the criteria mentioned.

METHODS FOR PRODUCTION OF PHYSICAL DEPENDENCE

A review by Mello (19) provides an excellent survey of this area. This section reviews the additions to the literature since then.

Intubation methods

Slight changes in intubation methods from one study to another result in numerous, somewhat different procedures.

Cannon et a/. (20) started the intubation schedule in rats with a large (6 g/kg) priming dose. Every eight hours thereafter the animals received supplemental doses of 0-3 g/kg, depending on whether the animals were ambulatory, conscious or neither. Administration continued for two days in one group and three days in another. All animals showed symptoms of withdrawal (tremors and excitability) approximately fourteen hours after the last dose and seven of the twelve animals also showed some form of seizure activity in response to auditory stimulation (key shaking and banging a tin plate with a spoon). No food or fluid was avilable throughout the administration period. Consequently, the experimental groups and the isocalorically treated controls lost about 15% of body weight during this period. The three day ethanol group lost significantly more weight than the two day ethanol group and their own pair-fed control group.

Intubations every eight hours were also used by Mucha et al. (21), but the initial doses were 1 g/kg. Doses were increased by 0.2 g/kg if the previous dose did not produce intoxication one hour after intubation, otherwise the previous dose was continued. To compensate for weight losses, if a rat's weight dropped below 90% of its initial weight, it received Metrecal supplements. This regimen was given to different groups for seven, fifteen or thirty days. Withdrawal symptoms, such as hyperreactivity, convulsive activity and audiogenic seizures were found to various degrees among the groups; the interpretation of these differences will be covered in a following section. Seven days of treatment by this procedure did not produce a full range of withdrawal symptoms, for a significant amount of seizure activity was not found. Although care was taken, by means of Metrecal supplements, that body weight was maintained within 90% of initial weight, no compensation was made for the normal growth and weight gain evidenced by the control group. The alcohol animals were, on the whole, 11% below the control animals, and those animals on the regimen the longest, the thirty day group, were 14% below.

Majchrowicz (22) used high doses of ethanol so that dependence was produced in a shorter time. The rats were all begun with a priming dose of 5 g/kg. Over the next four days they received daily doses of 9-15 g/kg divided into 3-5 fractional doses. The actual amount an animals received depended on its degree of intoxication. Unfortunately, due, to the adjustment of doses and fluctuation of the number of doses (the exact intervals between doses and the reason for changes being unreported) every animal was treated somewhat differently, and replication would appear difficult. Hunt (23) also used a variable dose method; the doses ranged from 11-15 g/kg/day for seven days. One group received two fractional doses and another three doses per day. The 3-dose/day group tended to have greater changes in seizure threshold during withdrawal than the 2-dose/day group, but the statistical significance was not tested. In both the Majchrowicz (22) and Hunt (23) studies no mention is made of weight losses during ethanol treatment or of procedures to control for this. However, since substantial weight losses were found with methods which used lower doses of alcohol (20) or tried to limit weight loss (21), it must be assumed that these studies were additionally complicated by severe weight losses.

Hillbom (24) and Noble et al. (25) used intubation methods which avoided the complication of variable doses and intervals. Noble et al. (25) started every rat with a total daily dose of 5 g/kg and increased the dose every other day by 1 g/kg until a dose of 10 g/kg was reached; the daily dose was given in two equal portions every twelve hours. Hillbom (24) began with doses of 4-5 g/kg per day and continued at that level for ten days. Thereafter, the doses were increased every other day by 1 g/kg to a final dose of 10 g/kg; the doses were divided into two portions every twelve hours. In both studies a non-alcoholic liquid diet was available as the sole source of food and fluid. These procedures prevented weight loss during treatment (24,25) and enabled almost normal weight gain (24). The lack of weight loss seemed, in large part, due to the use of the liquid diets. If animals were maintained on solid lab chow, they lost a significant amount of weight (25). The animals in both studies displayed withdrawal seizures after ethanol administration ceased.

Polydipsia

This procedure has been proposed by its developers (2) to be an ideal animal model of human alcoholism and to satisfy the criteria they themselves proposed, criteria not as complete as those outlined by Cicero and Smithloff (1) and Lester and Freed (3). In any event, Deutsch (26) argues that polydipsia does not in fact fulfill the criteria advanced by Falk et al. (2). As with every other method of producing dependence, the animals were not voluntarily ingesting alcohol. Since polydipsia is produced by a process which is not understood (27), it seems hardly possible for the authors to state flatly that the drinking is not dependent on any contingencies. In polydipsia, excessive alcohol intake appears moderated by the calories provided by alcohol, the pharmacological properties being of secondary importance (28). Finally, this procedure did not produce dependence when an attempt to replicate the paradigm was made (29).

Inhalation

The method developed by Goldstein and Pal (30) has been criticized (19) for its use of pyrazole and for weight loss during inhalation. A modification used by Griffiths et al. (31) overcomes these problems. Ethanol was administered to mice for a longer period, ten days, and the ethanol vapor concentration was gradually increased from 10-15 mg/L on day one to 25-35 mg/L on day ten (31) rather than three days with alcohol exposure of 10-16 mg/L (30). This method of gradual increase in vapor concentration resulted in less mortality, 10% vs. 50%, than when no pyrazole was used (32) and in no significant weight loss (31). No data were presented to indicate whether or not the fluctuations in blood alcohol concentrations, which pyrazole was used to prevent, were under control. Presumably, since mortality was decreased, the fluctuations were attenuated. Due to the more rapid rate of ethanol elimination without pyrazole, withdrawal symptoms occurred sooner. One puzzling aspect of this study is that the control animals had appreciable seizure scores themselves: possibly these mice, T.O. strain, are prone to seizures, with alcohol accentuating this factor.

Roach et a/. (33) maintained rats for seven days in an ethanol atmosphere of 15-30 mg/L. No pyrazole was used. Severe withdrawal symptoms were produced, including convulsions when suspended by the tail, and the relative intensity of the withdrawal symptoms was related to the mean daily blood alcohol level, as with mice (32). A drawback of the procedure is that the mean daily blood alcohol levels (and daily alcohol intake) were outside of the control of the experimenter, but were dependent upon each animal's ventilation and metabolic rates. Presumably, individual blood alcohol levels fluctuated considerably; no mention was made of mortality, and weight losses, averaging 20%, were not prevented with this procedure.

French and Morris (34) maintained rats in chambers containing low concentrations of ethanol vapor, 1.4 mg/L for two weeks. Not surprisingly, there were no detectable blood alcohol levels in the rats. Whether this procedure truly produced dependence is questionable. None of the usual behavioral symptoms of withdrawal were evident. The only symptoms were a decrease in weight one day after withdrawal and a slight increase in reactivity only after seventy-two hours; the body weight decrease was not tested for significance.

WITHDRAWAL MEASURES

Withdrawal symptomatology has been described for a wide range of animal species: monkey (12,35), chimpanzee (36), dog (37,38), rat (22), and mouse (30,39). The withdrawal syndrome of the higher animals, i.e., monkeys, chimpanzees and dogs, more closely resembles the human syndrome, with clinical symptoms such as alkalosis and hallucinations observed in these higher species (12,35,37). However, these observations have been at best semi-quantified with the variety of symptoms grouped into three categories judged to be of increasing severity (40) and at worst only qualitative, with the various symptoms present merely listed (12,36,37, 38). One should also note that withdrawal symptoms in humans, such as changes in body temperature, heart rate and blood pressure, should be quantified so that better comparisons to the animal data are available (41).

Scales for ranking withdrawal symptom severity have been developed with rodents. Some of the scales are only semi-quantitative: the symptoms are divided into broad stages of differing severity and the particular stage in which an animal is placed depends on a non-instrumental observation (11,22,39,42,43). These methods have several disadvantages. Perhaps the most obvious is simply the amount of time which must be spent in observing the animals for withdrawal symptoms with no assurance that, despite the large investment of time, every occurrence will be observed. Also, no allowance is made for accurately comparing the severity of the various symptoms between animals: an animal which has one convulsion is rated equal to an animal which has ten. Since these methods regard withdrawal as a continuum, as a series of symptoms of increasing severity, each following the preceding ones, a symptom which occurs early, such as hyperactivity, but is very intense, would be overshadowed in the rating by a later symptom, such as whole-body rigidity, which might be very mild. It would probably be best to rate the severity of each symptom rather than merely note its presence or absence. Replication becomes a problem with those methods which arbitrarily label an occurrence of a symptom as mild, intermediate or severe (22,42). Another researcher would find it formidable to rate intensities similarly. Mucha et al. (21) attempted to assess severity by observing the animals over several short observation periods, counting the number of periods in which each of several withdrawal symptoms appeared. However, this count is still neither a reliable measure of severity nor even of frequency: whereas an animal only scores zero or one for a ten minute observation period, several intense bursts of forelimb clonics will score equally with one instance of head-shake. On the other hand, a positive aspect of this method is that the three symptom categories are counted separately. Although the withdrawal symptoms are displayed in a definite order, it is probable that the various ethanol withdrawal symptoms, just as morphine withdrawal symptoms (44,45), are mediated by different neural systems. Experiments testing modification of withdrawal would, thus, be better served by a rating method which treats the symptoms as separately modifiable. In addition, the impossible task of determining whether intense hyperreactivity is more severe a symptom than intermediate whole-body rigidity (or vice-versa) is avoided.

Separate Symptoms

Convulsions

The most commonly studied symptom has been convulsions, evoked or spontaneous. Generally, convulsions are scored simply on the basis of how many animals have seizures in response to a sound, usually a specified bell (21,24,46,47,48). Some investigators (2,11,20), however, persist in using keys, an "instrument" which can hardly be conceived of as scientific or standardized: the "instrument" is certainly not reproductible. Simply recording the number of animals having seizures does not measure differences in severity.

Since there are several aspects of seizure, the different gradations can be scored. Goldstein (32) devised a five level scale for grading seizures elicited by suspending mice by the tail. After multiple tests over time, a curve is plotted and both the peak height and the area under the curve serve as measures of severity. This method has been found to be reliable and replicable by others (31,49). Similarly, seizure responses have been graded in rats; Ratcliffe (47) and Noble et a/. (25) both used six level scales, but the categories
were not identical. Since the two studies used different eliciting stimuli, a bell by Noble et al. (25) and electroshock by Ratcliffe (47), different seizure patterns may have resulted, thus necessitating the particular categories. Various other scales (50,51) which have been developed to rate audiogenic seizures can also be applied to ethanol withdrawal testing. Ratcliffe (47) also measured the duration of the seizures in response to bell ringing; the reliability of this measure has not been determined because the small number of animals did not permit statistical tests.

Hunter et al. (52) and Walker and Zornetzer (53) compared changes in EEG activity with behavioral changes during withdrawal in rats and mice respectively. EEG events were found not to be correlated with behavioral symptoms of withdrawal. Convulsions could be elicited while EEG events were at pre-convulsive stages indicating that cortical activity is not a reliable indicator of behavioral hyperexcitability.

An alternative to measuring variations in response is, of course, measuring changes in threshold, the changes in level of stimulus needed to produce a criterion response. McQuarrie and Fingl (54) measured changes in threshold for electroshock seizures in mice. A decrease in threshold, a hyperexcitability, reached its peak fully two days after the last dose of ethanol; the threshold returned to normal five days later. A disadvantage of this method is that large numbers of subjects must be used. Since an individual animal's threshold is affected by a sub-threshold stimulus (55), additional animals must be used for successive stimulus levels. Drug stimuli, such as leptazol, strychnine (47) and pentylenetetrazole (23), enable the same animal to be injected with successive increments of convulsant agent until the criterion response is obtained. Thus, a threshold is determined directly for each animal, rather than a calculated value at which 50% of the animals would have responded. Hunt
(23) found, as did McQuarrie and Fingl (54), that seizure threshold decreased on the first day of withdrawal. However, on succeeding days threshold for pentylenetetrazole-induced seizures increased above normal levels (23) whereas McQuarrie and Fingl (54) had found a further drop followed by a gradual return to normal.

Reactivity

Hyperreactivity can also be measured in either of two ways: a change in stimulus threshold needed to produce a criterion response or a change in response magnitude after a constant level stimulus. Gibbons et al. (55) found a decrease in threshold of foot shock needed to produce both a flinch and a jump response in rats after a period of alcohol deprivation. The hyperreactivity measure seems to be a rather sensitive measure of withdrawal since the alcohol administration procedure used, once daily intubations of ethanol which increased from 3 to 7 g/kg (56), produced a relative paucity of other withdrawal symptoms in comparison to more frequent schedules (23). A semi-quantitative variation of a threshold method was used by Hillbom (24). A wire heated to one of three different temperatures was placed on a rat's tail; tail movement within sixty seconds was recorded as a positive response. Inexplicably this test found behavioral hypo-reactivity, which is at odds with other reports of withdrawal symptoms. In any event, the test, as used, gives no indication of any threshold changes since the results of all the heat levels were reported together; there was no grading of the response. Thus, the test merely reported the presence or absence of a reponse to what was essentially a unitary stimulus.

A test of hyperreactivity used by Mucha et al. (21) measured changes in response, but did so only semiquantatively. Rats were judged as being hyperreactive if they resisted handling more than normally; a simple positive or negative response was recorded. The animals were tested thusly every other hour for sixteen hours. Because of the lack of a response magnitude this test seems to be more a measure of duration of withdrawal hyperreactivity than of withdrawal severity. French and Morris (34) utilized a simple apparatus to quantify response magnitude. Rats were given foot shocks of six different intensities, ranging from 0 to 0.5 mA, while suspended in a scale; the deflection in grams of the scale was taken as proportional to the response magnitude. However, with the alcohol dosage used, this method detected no changes in reactivity during the first day of withdrawal, and it was not until seventy-two hours afterwards that any significant hyperreactivity was detected. Given the trivial amounts of alcohol these rats received (not more than 1 g/kg/day), it may be doubted that these results are meaningful or replicable. Pohorecky et a/. (57) obtained changes in reactivity in rats which had a time-course similar to that which has been grossly observed during withdrawal (11). The magnitude of rats' startle responses to a buzzer was measured by means of a stabilimeter which produced a change in electromagnetic field proportional to an animal's movement; the characteristics of the buzzer were, unfortunately, not reported.

Body Weight

Several investigators have found body weight loss to be a reliable symptom of ethanol withdrawal in various species. Essig and Lam (38) found weight loss in dogs throughout the first eight days of withdrawal. Cannon et a/. (20) found a significant absolute weight loss in treated rats thirty-two hours following the last alcohol dose. It might be expected that the more severe the dependence and, hence, the withdrawal, the larger the withdrawal weight loss. However, this study (20) is not in agreement with such an expectation: the animals which had a shorter ethanol treatment period, two versus three days, had a larger withdrawal weight loss. This result may have occurred because the groups were not at the same weight at the start of withdrawal. Those animals treated for three days had lost significantly more weight during the treatment period and, thus, had less available weight to lose during withdrawal. Goldstein and Kakihana (57) found that, when weight loss was presented as proportion of weight loss during withdrawal, weight changes in mice paralleled other indicants of withdrawal severity. DBA mice had both higher seizure scores (32) and had greater weight losses than C57BL mice. Weight loss is quite possibly a very sensitive measure, for even those C57BL mice which had no seizure responses did show weight changes. Indeed, it is surprising that body weight changes have received such little attention in alcohol withdrawal; besides its simplicity, this method has been shown to be an excellent indicant of morphine withdrawal (59,60).

Activity

Spontaneous activity changes during withdrawal have been measured in both rats and mice. Hunter et al. (11) derived hourly composite activity scores from observations of grid crossings, rearings and grooming of rats for the first eight hours of withdrawal. Although there was substantial variability, it appears that there was a period of increased activity before the blood alcohol concentration reached zero; a phase of reduced activity followed. However, it is impossible to determine if the effects are significant since there were no statistics and no control or baseline scores. Cicero et al. (61) reported increased activity in most rats which had been maintained on ethanol. However, the results are not comparable to those of Hunter et al. (11) since the measure was open-field activity, and testing commenced twenty-four hours after the beginning of withdrawal. It should be noted that Cicero et al. (61) found that some of the animals displayed decreased rather than increased activity. This finding is implicit in the variability of activity found by Hunter et al. (11), and the results together indicate that neither activity change is an exclusive symptom of withdrawal in rats. Increased activity, as measured by an electromagnetic activity platform, was found in mice during the first six hours of ethanol withdrawal (31). However, once again no statistics were presented so that the significance of the changes is unknown. No study has yet demonstrated that any particular activity change is correlated with withdrawal nor what the time course of such changes might be.

Tremors

Tremors are probably the most widely cited symptom of withdrawal, but only in a qualitative sense: either they are present or absent. Mello (19) raised the possibility of using certain tremor analysers (62,63) to quantify ethanol withdrawal tremors. Freund (64) described a procedure whereby mice were housed in a jar suspended from a muscle transducer so that the duration, frequency and amptitude of both tremors and seizures could be recorded. However, no data have yet been presented to assess this stratagem.

Temperature

It is surprising that body temperature changes, as with weight changes, have been so rarely investigated. Temperature changes, besides being evident during human ethanol withdrawal (65) are a good indicant of morphine withdrawal (45). Tabakoff (66) found hypothermia in mice undergoing withdrawal which was proportional to the duration of the ethanol administration period; temperatures were measured every several hours. The time-course of temperature changes paralleled the development of other symptoms. The only other instance of temperature changes found in animals during withdrawal was in chimpanzees, but data were not reported (36). Pohorecky (67) was unable to detect any temperature changes in rats undergoing withdrawal; this difference from the Tabakoff (66) findings might have resulted from the temperatures being recorded at about four hours after the peak of withdrawal (twelve hours vs. eight hours) or simply a species difference. In any case, a change in preference of environmental temperature was found. Animals were placed in a T-maze which had the alley at ambient temperature, one arm at 40° C. and the other arm at 4° C.; the amount of time spent in each location during a five minute trial was recorded. Animals withdrawn from alcohol increased their preference for the warm arm; the peak of this behavior was found seventy-two hours after ethanol. The results were interpreted as a change in set-point for temperature, a measure more sensitive than body temperature changes.

Aggression

Enhanced aggression is another measure of withdrawal which appears applicable to ethanol withdrawal. To measure aggression threshold, pairs of rats can be given increasing levels of foot-shock, five shocks at each level, and the levels at which vocalizations, rearings, and biting attacks occur, recorded. A decrease in threshold for shock-induced aggression was thus found seventy-two hours after the beginning of ethanol withdrawal (Lal, personal communication, 1975), but not at fourty-eight hours, a time course identical to that of morphine withdrawal (68).

Up to now we have been labelling the methods which de7 termine numerical scores for withdrawal symptoms as quantitative. Let us now consider just how quantitative they are. What is the basis for believing that a lower score means less severe withdrawal? Exactly how much less severe is a low score compared to a high score? In the numerous articles which report measuring withdrawal symptoms the majority merely compare the withdrawal scores of a control group and a single alcohol treated group. These studies can hardly be said to have determined how reliable their particular withdrawal measures were for distinguishing differences in severity of withdrawal. For any withdrawal measure to reliably determine differences in severity, that measure must be applied to groups of animals which are differentially dependent; in other words, a dose-response function must be determined. Using an inhalation method of alcohol administration, Goldstein (32) has indeed determined such a function. Mice were exposed to several combinations of different ethanol vapor levels and different durations of ethanol inhalation. The different ethanol vapor levels produced different blood alcohol concentrations. Measuring withdrawal by scoring handling-induced seizures detected differences in intensity of withdrawal which were linearly related to the time-integrated dose of alcohol. Thus, an animal which had a blood alcohol concentration (BAC) of 1.14 mg/ml for three days received approximately the same total dose (3.42 mg/ml X days) as an animal which had a BAC of 0.28 mg/ml for thirteen days (3.64 mg/ml X days), and both had the same intensity of withdrawal seizures. Doubling the total dose of alcohol approximately doubled the withdrawal score.

The ability to determine a dose-response function seems to depend on the method of alcohol administration. Those studies which tried to vary the degree of dependence using oral administration procedures, for the most part, were not able to obtain as plausible dose-response functions as found by Goldstein (32). Cannon et al. (20) did not find a difference in incidence of seizure between groups of rats intubated every eight hours for either two days or three days. The other measure used, body weight loss, was the reverse of what would be expected; as mentioned before, the two day group lost more weight during withdrawal than the three day group. Mucha et al. (21) also intubated rats every eight hours but for longer periods: seven, fifteen and thirty days. The semi-quantitative withdrawal measures of hyper-reactivity and convulsive activity detected differences in withdrawal severity only between the seven and thirty day groups; no differences were found between either seven and fifteen day groups or fifteen and thirty day groups. Another measure, incidence of audiogenic seizures, found no differences between either an untreated group and the seven day group or between the fifteen day and the thirty day groups. Of course, it is possible that the measures used were not sufficiently sensitive or accurate to detect differences in severity. However, aspects of these intubation procedures inherently weaken their ability to serve as models for quantification of withdrawal severity. All the animals within a group did not receive the same amount of alcohol because each dose which an animal received was raised or lowered dependent on its degree of intoxication (20,21). This is not to imply that all intubation procedures are unfeasible. It is simple necessary that all animals within a particular treatment group receive the same treatment, e.g., the intubation procedure used by Noble et al. (25).

Another administration procedure, liquid diet, seems generally ill-suited to determinations of a dose-response function. The duration that animals are on the diet is, of course, controlled by the experimenter, but the actual amount the animals consume is not under the experimenter's control. Animals often manage, perversely, to utilize opportunities to obfuscate experiments: thus Hunter et al. (11) maintained rats on a liquid diet for ten, fifteen, twenty or thirty days. Spontaneous abstinence periods were observed. As the duration of the alcohol period increased both the number of animals spontaneously abstaining and the number of times individual animals abstained increased. Thus, a reliable scale of doses became unobtainable.

Determining a dose-response function is, therefore, a sine qua non for determining how accurate withdrawal measures are and for determining how reliable the alcohol administration procedure is. As mentioned above, such a function can be obtained by recording of total alcohol intake; frequent determinations of blood alcohol levels, and the subsequent calculations of a time-integrated function, may also produce an accurate dose-response curve. In either case, it appears important that no discontinuity in the intake of alcohol occur.

SUMMARY

That physical dependence and an abstinence syndrome can result in animals after substantial alcohol intake is beyond dispute. Scientific elegance and analysis would be served if the variety of withdrawal symptoms were each separately quantified; such quantification will make it possible not only to better assess the relation between dose and response, but also to assess such treatments as may modify one or another of the dysfunctions of the withdrawal period. Of the various procedures used for giving animals alcohol, we are convinced that only such procedures which give the experimenter real control of the amounts of alcohol obtained by the animals can be valid; even with such a method (e.g., intubation), it may be essential to establish the equality of the dosing level within a group by measuring their time-integrated blood alcohol levels. Observance of these strictures of experimental detail should lead to a comprehensive description of the dose-response function and, reasonably, to a better understanding of the central effects of alcohol.
Although the various facets of an animal model have not yet been put together, the techniques which may be useful in its assemblage appear at hand. The methods described by which voluntary alcohol consumption can be increased may prove to meet the initial criteria advanced for an animal model in specific mouse or rat strains. Mouse or rat strains which either "prefer" an alcohol solution (C57BL) or have been bred for this purpose [the AA rat strain (69)], the long or short sleep mice bred by McClearn (70) (see also 71) or the rat strains bred for disparate impairment of motor activity in response to alcohol (72) might well be prime candidates for an animal model. The quantification of the withdrawal symptomatology should also make it possible to assay with precision whether the experimental modifications producing increased voluntary consumption of alcohol do in fact lead to physical dependence of any significant degree.

REFERENCES

1. Cicero, T.J. and Smithloff, B.R.: Alcohol oral self-administration in rats: attempts to elicit excessive intake and dependence, Alcohol Intoxication and Withdrawal. Experimental Studies. Edited by Gross, M.M., Plenum Press, New York, p. 213-224 (1973).
2. Falk, J.L., Samson, H.H. and Winger, G.: Behavioral maintenance of high concentration of blood ethanol and physical dependence in the rat. Science 177:811-813 (1972).
3. Lester, D. and Freed, E.X.: Criteria for an animal model of alcoholism. Pharmacol. Biochem. Behay. 1:103-107 (1973).
4. Deutsch, J.A. and Koopmans, H.S.: Preference enhancement for alcohol by passive exposure. Science 179:1242-1243 (1973).
5. Myers, R.D., Stoltman, W.P. and Martin, G.E.: Effects of ethanol dependence induced artificially in the rhesus monkey on the subsequent preference for ethyl alcohol. Physiol. Behay. 9:43-48 (1972).
6. Begleiter, H.: Ethanol consumption subsequent to physical dependence, Alcohol Intoxication and Withdrawal. Experimental Studies. Vol. II. Edited by Gross, M.M., Plenum Press, New York, p. 373-378 (1975).
7. Hunter, B.E., Walker, D.W. and Riley, J.N.: Dissociation between physical dependence and volitional ethanol consumption: Role of multiple withdrawal episodes. Pharmacol. Biochem. Behay. 2:523-530 (1974).
8. Bass, M.B. and Lester, D.: Increased alcohol selection in rats after alcohol drinking paired with recovery from thiamine deficiency. Physiol. Behay., submitted (1976).
9. Mello, N.K. and Mendelson, J.H.: Experimental induced intoxication in alcoholics: A comparison between programmed and spontaneous drinking. J. Pharmacol. exp. Ther. 173:101-116 (1970).
10. Mello, N.K. and Mendelson, J.H.: Drinking patterns during work-contingent and non-contingent alcohol acquisition. Psychosom. Med. 34:139-164 (1972).
11. Hunter, B.E., Riley, J.N., Walker, D.W. and Freund, G.: Ethanol dependence in the rat: A parametric analysis. Pharmacol. Biochem. Behay. 3:619-629 (1975).
12. Deneau, G., Yanagita, T. and Seevers, M.J.: Self-administration of psychoactive substances by the monkey. Psychopharmacologia 16:30-48 (1969).
13. Wallgren, H. and Barry, H. III: Actions of Alcohol. Elsevier Publishing Co., Amsterdam (1970).
14. Amit, Z., Stern, M.H. and Wise, R.A.: Alcohol preference in the laboratory rat induced by hypothalamic stimulation.
16    HOWARD J. FRIEDMAN AND DAVID LESTER
15. Myers, R.D.: Alcohol consumption in rats: Effects of intracranial injections of ethanol. Science 142:240-241 (1963).
16. Martin, G.E. and Myers, R.D.: Ethanol ingestion in the rat induced by rewarding brain stimulation. Physiol. Behay. 8:1151-1160 (1972).
17. Friedman, H.J. and Lester, D.: Intraventricular ethanol and ethanol intake: A behavioral and radiographic study. Pharmacol. Biochem. Behay. 3:393-401 (1975).
18. Freund, G.: Induction of physical dependence on alcohol in rodents, Biochemical Pharmacology of Ethanol. Edited by Majchrowicz, E., Plenum Press, New York, p. 311-325 (1975b).
19. Mello, N.K.: A review of methods to induce alcohol addiction in animals. Pharmacol. Biochem. Behay. 1:89101 (1973).
20. Cannon, D.S., Baker, T.B., Berman, R.F. and Atkinson, C. A.: A rapid technique for producing ethanol dependence in the rat. Pharmacol. Biochem. Behay. 2:831-834 (1974).
21. Mucha, R.F., Pinel, J.P.J. and VanOot, P.H.: Simple method for producing an alcohol withdrawal syndrome in rats. Pharmacol. Biochem. Behay. 3:765-769 (1975).
22. Majchrowicz, E.: Induction of physical dependence upon ethanol and the associated behavioral changes in rats'. Psychopharmacologia 43:245-254 (1975).
23. Hunt, W.A.: Changes in the neuro-excitability of alcohol dependent rats undergoing withdrawal as measured by the pentylenetetrazole seizure threshold. Neuropharmacology 72:1097-1102 (1973).
24. Hillbom, M.E.: The prevention of ethanol withdrawal seizures in rats by dipropylacetate. Neuropharmacology 14:755-761 (1975).
25. Noble, E.P., Gillies, R., Vigran, R. and Mandel, P.: The modification of the ethanol withdrawal syndrome in rats by di-n-propylacetate. Psychopharmacologia 46:127131 (1976).
26. Deutsch, J.A.: Behavioral maintenance of high concentrations of blood ethanol and physical dependence in the rat. Science 180:880 (1973).
27. Falk, J.L.: The nature and determinants of adjunctive behavior. Physiol. Behay. 6:577-588 (1971).
28. Freed, E.X. and Lester, D.: Schedule-induced consumption of ethanol: Calories or chemotherapy? Physiol. Behay. 5:555-560 (1970).
29. Heintzelman, M.E., Best, J. and Senter, R.J.: Polydipsia-induced alcohol dependency in rats: A reexamination. Science 191:482-483 (1976).
30. Goldstein, D.B. and Pal, N.: Alcohol dependence produced in mice by inhalation of ethanol: Grading the withdrawal reaction. Science 172:288-290 (1971).
31. Griffiths, P.J., Littleton, J.M.L. and Ortiz, A.: Changes in monoamine concentrations in mouse brain associated with ethanol dependence and withdrawal. Brit. J. Pharmacol. 50:489-498 (1974).
32. Goldstein, D.B.: Relationship of alcohol dose to intensity of withdrawal signs in mice. J. Pharmacol. exp. Ther. 180:203-215 (1972).
33. Roach, M.K., Khan, M.M., Coffman, R., Pennington, W. and Davis, D.L.: Brain (iat+e) activated adenosine triphosphate activity and neurotransmitter uptake in alcohol dependent rats. Brain Res. 63:323-329 (1973).
34. French, S.W. and Morris, J.R.: Ethanol dependence in the rat induced by non-intoxicating levels of ethanol. Res. Comm. Chem. Path. Pharmacol. 4:221-233 (1972).
35. Ellis, F.W. and Pick, J.R.: Ethanol-induced withdrawal reactions in rhesus monkeys. The Pharmacologist 11:256 (1969).
36. Pieper, W.A., Skeen, M.J., McClure, H.H. and Bourne, P. G.: The chimpanzee as an animal model for investigating alcoholism. Science 176:71-73 (1972).
37. Ellis, F.W. and Pick, J.R.: Evidence of ethanol dependence in dogs. Fed. Proc. 79:649 (1970b).
38. Essig, C.F. and Lam, R.C.: Convulsions and hallucinatory behavior following alcohol withdrawal in the dog. Arch. Neurol. 18:626-632 (1968).
39. Freund, G.: Alcohol withdrawal syndrome in mice. Arch. Neurol. 21:315-320 (1969).
40. Ellis, F.W. and Pick, J.R.: Experimentally induced ethanol dependence in rhesus monkeys. J. Pharmacol. exp. Ther. 175:88-93 (1970a).
41. Woods, J.H. and Winger, G.D.: Alcoholism and animals. Prey. Med. 3:49-60 (1974).
42. Branchey, M., Ranscher, G. and Kissen, B.: Modifications in the response to alcohol following the establishment of physical dependence. Psychopharmacologia 22:314322 (1971).
43. Wallgren, H., Koshnen, A.L. and Ahtee, L.: Technique for producing an alcohol withdrawal syndrome in rats. Israel J. Med. Sci. 9, suppl.:63-71 (1973).
44. Herz, A., Blasig, J. and Papeschi, R.: Role of catecholaminergic mechanisms in the expression of the morphine abstinence syndrome in rats. Psychopharmacologia 39: 121-144 (1974).
45. Schwartz, D.S. and Eidelberg, E.: Role of biogenic amines in morphine dependence. Life Sci. 9 (Part 1): 613-624 (1970).
46. Freund, G. and Walker, D.W.: Sound induced seizures during ethanol withdrawal in mice. Psychopharmacologia 22:45-49 (1971).
47. Ratcliffe, F.: Ethanol dependence in the rat: Its production and characteristics. Arch. Int. Pharmacodyn. 196:146-156 (1972).
48. Sze, P.Y., Yanai, J. and Ginsburg, B.E.: Adrenal glucocorticoids as a required factor in the development of ethanol withdrawal seizures in mice. Brain Res. 80:155159 (1974).
49. Blum, K. and Wallace, J.E.: Effects of catecholamine synthesis inhibition on ethanol-induced withdrawal symptoms in mice. Br. J. Pharmacol. 51:109-111 (1974).
50. Gates, G.R., Chen, C.S. and Bock, G.R.: Effects of monaural and binaural auditory deprivation on audio-genic seizure susceptability in BALB/c mice. Exp. Neurol. 38:488-493 (1973).
51. Jobe, P.C., Picchioni, A.L. and Chin, L.: Role of brain norepinephrine in audiogenic seizure in rat. J. Pharmacol. exp. Ther. 184:1-10 (1973).
52. Hunter, B.E., Boast, C.A., Walker, D.W. and Zornetzer, S.F.: Alcohol withdrawal syndrome in rats: Neural and behavioral correlates. Pharmacol. Biochem. Behay. 1: 719-725 (1973).
53. Walker, D.W. and Zornetzer, S.F.: Alcohol withdrawal in mice: Electroencephelographic and behavioral correlates. EEG c1in. Neurophysiol. 36:233-244 (1974).
54. McQuarries, D.G. and Fingl, E.: Effects of single doses and chronic administration of ethanol on experimental, seizures in mice. J. Pharmacol. exp. Ther. 124:264-271 (1958).
55. Woodbury, L.A. and Davenport, V.D.: Design and use of a new electroshock seizure apparatus, and analysis of factors altering seizure threshold and pattern. Arch. Int. Pharmacodyn. 92:97-107 (1952).
56. Gibbons, R.J., Kalant, H., LeBlanc, A.E. and Clark, J.W.: The effects of chronic administration of ethanol on startle threshold in rats. Psychopharmacologia 19:95-104 (1971).
57. Pohorecky, L.A., Cagan, M., Brick, J. and Jaffe, L.S.: Startle response in rats: Effect of ethanol. Pharmacol. Biochem. Behay. 4:311-316 (1976).
58. Goldstein, D.B. and Kakihana, R.: Alcohol withdrawal reactions and reserpine effects in inbred strains of mice. Life Sci. 15:415-425 (1974).
59. Akera, T. and Brody, T.M.: The addiction cycle to narcotics in the rat and its relation to catecholamines. Biochem. Pharmacol. 17:675-688 (1968).
60. Bhargava, J. and Way, E.L.: Effect of 1-pheny1-3-(2- thiazaly1)-2-thiourea, a dopamine beta-hydroxylase inhibitor on morphine analgesia, tolerance and physical dependence. J. Pharmacol. exp. Ther. 190:165-175 (1974).
61. Cicero, T.J., Snider, S.R., Perez, V.S. and Swanson, L. W.: Physical dependence on and tolerance to alcohol in the rat. Physiol. Behay. 6:191-198 (1971).
62. Otis, L.S., Willis, J.A. and Pryor, G.T.: Tremor detector for the mouse. J. Appl. Physiol. 26:640-643 (1969).
63. Silverman, R.W. and Jenden, D.J.: Tremor analyzer for small laboratory animals. J. appl. Physiol. 28:513-514 (1970).
64. Freund, G.: Animal models of ethanol withdrawal syndromes and their relevance to pharmacology, Biological and Behavioral Approaches to Drug Dependence. Edited by Cappell, H. and LeBlanc, A.E., Addiction Research Foundation, Toronto, Ontario, p. 13-25 (1975a).
65. Gross, M.M., Lewis, E., Best, S., Young, N. and Feuer, L.: Quantitative changes of signs and symptoms associated with acute alcohol withdrawal: Incidence, severity and circadian effects in experimental studies of alcoholics, Alcohol Intoxication and Withdrawal. Experimental Studies. Vol. II. Edited by Gross, M.M., Plenum Press, New York, p. 615-631 (1975).
66. Tabakoff, B.: A biological model of tolerance and physical dependence: Use of body temperature. Paper presented at National Council on Alcoholism, Annual 'Conference, Washington, D.C., May, 1976.
67. Pohorecky, L.A.: Locomotor and thermoregulatory disturbances in rats during ethanol withdrawal. Paper presented at National Council on Alcoholism, Annual Conference, Washington, D.C., May, 1976.
68. Puri, S.K. and Lal, H.: Reduced threshold to pain induced aggression specifically related to morphine dependence. Psychopharmacologia 35:237-241 (1974).
69. Eriksson, K.: Alcohol consumption and blood alcohol in rat strains selected for their behavior towards alcohol, Biological Aspects of Alcohol Consumption. Edited by Forsander, O. and Eriksson, K., Finnish Foundation for Alcohol Studies, Helsinki, 20:121-125 (1972).
70. McClearn, G.E.: The genetic aspects of alcoholism, Alcoholism: Progress In Research and Treatment. Edited by Bourne, P.G. and Fox, R., Academic Press, New York, p. 337-358 (1973).
71. Heston, W.D.W., Erwin, V.G., Anderson, S.M. and Robbins, H.: A comparison of the effects of alcohol on mice selectively bred for differences of ethanol sleep-time. Life Sci. 14:365-370 (1974).
72. Riley, E.P., Freed, E.X. and Lester, D.: Selective breeding of rats for differences in reactivity to alcohol: An approach to an animal model of alcoholism. I. General procedures. J. Stud. Alc. 37:(11):in press (1976).