How Opioid Antagonists Reduce The Craving For Alcohol – Part One (of Two)

by Dr David J Sinclair with Dr Roy Eskapa and Michael Sinclair

October 2014

There is a widespread misunderstanding about how and when opioid antagonists such as naltrexone, naloxone and nalmefene suppress the craving for alcohol.  The preclinical and clinical evidence reviewed here show that craving is not reduced simply by the presence of the antagonists in the body.  Instead, they work by the mechanism of extinction.  Alcohol drinking is learned through reinforcement involving the opioid system.  Drinking alcohol while an opioid antagonist blocks the reinforcement starts extinction of the drinking behaviour and craving.

Evidence for this includes the fact that naloxone caused no reduction in the rate of lever pressing for alcohol by rats at the beginning of the first session.  Similarly, prescribing naltrexone to abstinent alcoholics has not delayed significantly the resumption of drinking.  Both rats and humans that are given opioid antagonists show little or no effect at first.  Subsequently, both craving and drinking decrease progressively as a function of the number of sessions when alcohol was consumed while the antagonist was present.  Almost all of the mean variation is explained by the theoretical extinction curve.  Moreover, the motivation for alcohol remains suppressed long after all of the antagonist has been removed from the body.  The results suggest that the primary influence of the antagonists on craving and drinking is not directly from the medicine itself but instead is produced by repeated extinction sessions in which alcohol is drunk while reinforcement is blocked by an opioid antagonist.

One of the strongest effects of using naltrexone in the treatment of alcoholism is the progressive decrease in the craving for alcohol.  For example, the significance of the effect was p=0.00000000001 when we analysed the data from the first 210 patients at ContrAl Clinics (Figure 1 below). Patients themselves often comment after a couple months of treatment that to their surprise, they no longer are particularly interested in drinking.  Previously, alcohol had been the focus of their lives, the main thing they were thinking about.  Somehow the treatment had gotten rid of the obsession.

Perhaps because of these results, some people have gotten the mistaken idea that naltrexone itself reduces the craving.  Alcoholics are writing in response to Roy Eskapa’s book (1), stating that they can only find doctors who will prescribe naltrexone for the purpose of suppressing craving while abstaining.  A recent video on YouTube, although positive about naltrexone, said that it ‘blocks the craving and the high’.  The practice of giving naltrexone in a targeted manner (ie. taking the medicine only on drinking days) is often described as telling patients to take naltrexone only when they need it to suppress a particularly high crabbing.  Another clinician tells patients to take a double dose of naltrexone on days when they are having very high craving.  Most of the theories proposed for how naltrexone works (eg. that alcoholism is caused by too little (2) or too much (3) opioidergic activity), aside from the initial one that it causes extinction (4, 5), predict that the antagonists directly block craving.  Certainly one would expect that if the dose of antagonist was sufficiently high to make the person or animal feel different, ie. changing the stimulus situation from that in which alcohol drinking was learned there would be a direct reduction in craving before the first sip of alcohol was consumed.

As shown in this review, most of the available evidence suggests that naltrexone and other antagonists do not directly suppress craving.  Or course, one cannot prove there is no such effect; perhaps the experiments have not been powerful enough to demonstrate it.  One can conclude, however, that if there is any direct effect on craving, it is too small to be relevant clinically.

On the other hand, the reduction in craving during naltrexone treatment is one of the most powerful and remarkable effects observed.  The craving definitely is suppressed, but it is suppressed after drinking alcohol while on naltrexone; that is, after the mechanism of extinction has had an effect.  Before the first sampling of alcohol, however, the opioid antagonists produce no significant reduction in craving.  The incorrect belief that opioid naltrexone alone blocks craving has probably played a major role in its being prescribed with instructions that make it ineffective, and thus in the low rate of prescribing the medicine.

The material here is from the annual presentations that Dr Sinclair made before the medical students of the University of Helsinki.

Table 1 summarizes the differing predictions from the two hypotheses for how opioid antagonists work.  The evidence from preclinical and clinical research relevant to each of the predictions is then presented.  The analyses are limited to these two hypotheses.  Naturally, it is possible that neither is correct and that some other explanation is correct.  Therefore, evidence that one hypotheses is false does not automatically imply that the other is true.  The possibility that both are correct can also not be excluded a priori.

Figure 1.  The reduction in craving developing as a function of days in treatment with 50mg naltrexone 1 hour before drinking. Craving was reported at each clinical visit, using a 100mm visual analogue scale. 198 of the first 210 patients provided usable date.

Pre-Clinical studies of voluntary alcohol drinking

We have conducted a large number of experiments (at last count, 47 studies) on the effect of opioid antagonists on voluntary alcohol drinking.  Most involved giving placebo or naloxone, nalmefene, or naltrexone to rats just before returning access to alcohol after a period of alcohol deprivation; often this was with ‘limited access’ in which the rats had continual access to food and water but access to the alcohol solution for only one hour per day.  Rats reliably wake up and start drinking the alcohol solution as soon as it is returned, this demonstrating high motivation for alcohol.  The rate of alcohol drinking during the first 10 minutes is orders of magnitude higher than that seen during continual access.  In order to see the direct effect of the opioid antagonists on craving – separate from any effects that develop after drinking while reinforcement is blocked (ie. extinction effects) – we need to look at the rats’ behaviour the very first time the antagonist is administered.

We have never in all these studies seen a decrease in the behaviour of starting to drink alcohol solution after the first administration of antagonist (eg. 6).  In the experiment shown in Figure 2, all of the rats started drinking immediately when the alcohol bottle was put back on the cage during both the pre-treatment period and on the first nalmefene sessions.  On subsequent treatment days, the percentage of rats starting to drink immediately decreases progressively, becoming significantly lower than seen in the rats given placebo, and by the 4th or 5th pairing of drinking with antagonist, almost none of the animals were showing the behaviour.  The same result is seen in the amount of alcohol consumed in the first 10 minutes: no significant effect on the first day of receiving the antagonist, but a significant reduction developing on subsequent days.

Drinking may decrease slightly later in the first extinction session and the rats’ total alcohol intake is often significantly reduced in the first sessions following injection of naloxone, naltrexone or nalmefene.  This can be seen in operant behaviour in Figure 3.  However, when the antagonist was given to rats in a stress-free oral manner, similar to how humans take the medicine, there was no decrease in alcohol drinking during the first hour-long alcohol session, but rather a slight tendency for an increase in drinking (7) (see figure 2).

Figure 2. Extinction of voluntary alcohol drinking to male Wistar rats with one hour daily access to 10% alcohol solution and continual access to food and water.  Prior to alcohol sessions, the rats ate measured amounts of a cocoa-flavoured sucrose paste.  These 7 rats then received 10 mg/kg nalmefene in the paste before each of the next 5 sessions. *p<0.05;**p<0.01 relative to 7 controls given only the vehicle.


Lever pressing for alcohol solution

The lack of reduction in motivation for alcohol before the first drink is shown very clearly when opioid antagonists are administered to rats that have learned operant responding for drops of alcohol solution.  The graph below (Figure 3, unpublished data from Petri Hyytiä’s experiments published in 1993(8)) shows the number of lever presses for 10% alcohol solution, recorded automatically every two minutes, in rats having access to alcohol one hour a day.

The important data for the question of craving is the bar showing the responding during the first two minutes of the FIRST NALOXONE session.  The rats have been injected with 1 mg/kg naloxone 30 minutes earlier.  The antagonist produced no reduction in responding for alcohol during the first 2 minutes.  Naloxone had no direct effect on the motivation for alcohol.

Figure 3.  Lever pressing for alcohol by rats during an hour sessions proceeded by a saline injection (left) and then in three successive sessions proceeded by a naloxone injection.  Responding in general was significantly reduced by naloxone, but responding during the first 2 minutes of the first naloxone session – indicative of any direct effect of naloxone on motivation for alcohol – was not reduced.

Responding during the first 2 minutes of the SECOND NALOXONE and THIRD NALOXONE session, however, reduced progressively.  The motivation for alcohol was reduced after pairing of naloxone with lever pressing and alcohol drinking ie. after triggering extinction.

Published cumulative response patterns from two experiments with different doses of naloxone 30 minutes before access to lever pressing for alcohol also show that naloxone does not directly reduce motivation for alcohol.  The AA rats began responding immediately for oral ethanol with naloxone having no effect on responding in either experiment during the first 2 minutes and no significant effect during the first 5 minutes.  Higher doses of naloxone did, however, suppress responding later in the 1 hour sessions.

Clinical evidence with opiate addicts

The first naltrexone clinical trial was for treatment of opiate addiction (9).  Once the patients began self-administering opiates whilst on naltrexone, the reported craving began decreasing and by the end of the trial, their craving was significantly lower than that reported by the patients receiving the placebo.  Among the majority of patients who never self-administered opiates, however, naltrexone had no effect on craving.  Naltrexone itself did not reduce craving for opiates.

The explanation given for these results was that naltrexone is thought to work through the mechanism of extinction.  Extinction is the mechanism that weakens a learned response after a response is made and then is not followed by reinforcement.  In the best known example, Pavlov’s dogs, conditioned to salivate to the sound of a bell by having a bell ring before getting food, had the conditioned response extinguished by not getting food after salivating to the bell.  Extinction requires that the response first be made – and then not produced any reinforcement.  Consequently, only the patients who made the response of taking opiates while naltrexone blocked the reinforcement had their craving weakened by extinction.

Our clinical results with alcoholics in regular practice

The daily intake of alcohol, reported in the drinking diaries of the first 210 alcoholics treated at the Tapiola and Tampere ControlAl clinics (not previously published) is shown in Figure 4.  As in the pre-clinical results, the critical data is the effect on the first day of treatment with naltrexone.  If naltrexone itself is able to reduce the patients’ craving to the extent that it is clinically relevant for drinking, then there should be a significant reduction in the number of drinks taken on this first day of treatment.

In fact, there was no reduction at all from the mean level of alcohol drinking during the two weeks before treatment.  Indeed, the number of drinks consumed on the first day is slightly, although not significantly, higher than the pre-treatment mean.  Naltrexone alone did not reduce craving and drinking.

Subsequently, after drinking has been paired with naltrexone and extinction could have an influence, alcohol consumption decreased progressively.  The decrease is highly significant.

Figure 4:Daily intake of alcohol, reported in drinking diaries, by the first 210 Finnish patients taking 50mg naltrexone daily,  The theoretical extinction curve is from the Rescorla-Wagner equation (see Extinction Curve section below) with V set at the pre-treatment mean, 6.07 drinks per day; Vsum set at 1.3 drinks per day, the mean level eventually found after about a year for successful patients; S was set at 0.015.  The extinction curve explains 66.4% of the variability in the daily drinking plan.

Other clinical results

The most common protocol used in the clinical trials of treating alcoholism with naltrexone has been first to detoxify the patients, and then instruct them to abstain completely from alcohol while taking either naltrexone or placebo every day.  If naltrexone directly reduced craving, the patients getting the antagonist would be able to abstain longer than those getting placebo.

Our database has 24 publications using this protocol with either naltrexone of nalmefene.  Every one of them found no significant effect of the antagonist on the time to the first drink ie. no significant effect of the antagonist on clinically-relevant craving; (10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33).  Reviews also conclude that the evidence for both naltrexone, (34) and nalmefene, (35), shows the antagonists are effective in preventing patients who are drinking from relapsing to heavy drinking, but do not prolong abstinence.

First drink priming from endorphins

In addition to extinction, there is at least one other mechanism through which naltrexone seems to suppress alcohol drinking: blocking of the ‘first drink effect’.  The evidence for this is rather weak but one can see possible evidence in Figure 3.  The rats in the SALINE session showed a second bout of level pressing for alcohol, approximately 20 to 30 minutes into the hour.  The second bout is missing in each of the naloxone sessions, including the first, when the responding in the first 2 minutes was not reduced.  A likely explanation is that effects from the first bout of drinking were stimuli for the second bout.  One of the effects of the first bout was a release of endorphins, and the effects of the endorphins were among the stimuli for the second bout.  Naltrexone blocks the effects of the endorphins, and this removes a stimulus for the second bout of drinking.

There is also anecdotal evidence from clinical practice.  Patients frequently report, early in naltrexone treatment, that they are surprised to find that they did not finish the bottle, that they are satisfied by only a few drinks and do not bother taking more.  Our explanation has been that the naltrexone is blocking the first drink effect.  The one  thing that always has been present when the second drink is taken is the effect produced by the first drink.  This is generally a stimulatory effect, and it is has been shown in animals that the stimulatory effect of alcohol is blocked by opioid antagonists. (36)

Direct opioidergic stimulation of drinking

In addition to effects relating to learning, there is some evidence that endorphins may have a direct ability to stimulate alcohol drinking prior to any experience with alcohol.  Figure 5 shows results obtained with AA rats given a single injection of morphine; the time was distributed throughout the day in order to preclude any circadian influences.  A high peak in alcohol drinking was observed 4 hours after the morphine injection, regardless o the time of day when the morphine had been administered.  This was followed by a suppression of alcohol drinking, keeping it significantly lower than the rats’ own prior level of intake for several days.

These results were observed in rats that had had prolonged prior experience with voluntary alcohol drinking.  Almost identical results, however, had previously been seen in naïve Sprague Dawley rats ie. ones that had never had alcohol before the large morphine injection (30 mg/kp IP). (37).  As in the alcohol-experiences rats, there was a very high peak in alcohol drinking during the fourth hour after receiving morphine, followed by a significant suppression, relative to controls, for the next 6 days.

These results indicate that opioidergic activation can stimulate alcohol drinking even without prior drinking experience.

Figure 5: Effects of a single injection of morphine on subsequent alcohol drinking.  In both AA rats (here) and Long Evans (Sinclair et al, 1982) morphine blocked alcohol drinking while the rats were impaired by the opiate, followd by a short period when drinking of alcohol, but not water, was greatly increased, after which alcohol drinking was significantly suppressed for many days.

Theory for expecting direct blocking of craving

The learned ability of stimuli related to alcohol to elicit feelings of craving is known to be reduced after drinking alcohol while naltrexone is present.  The theory proposing, however, that naltrexone alone can block craving starts with the assumption that when alcohol is consumed, and thus endorphins released, the stimuli present at the time would be classically conditioned to cause a release of endorphins by themselves.  In analogy with Pavlov’s experiment, if you repeatedly heard a bell being rung before you drank alcohol, eventually the bell itself would cause endorphins to be released.  The binding of these endorphins to their receptors or some effect from it would then be felt as a craving and would help to stimulate more alcohol drinking, just as the morphine in Figure 5 did and as the first drink is believed to work.  Naturally, if the opioid receptors were blocked by naltrexone, the craving related to alcohol cues would not be produced and drinking would not be stimulated, just as naltrexone probably blocks the first drink stimulation of subsequent drinking.

The weakest part of the theory is the assumption that alcohol-related cues develop the ability to release endorphins.  There is no doubt such cues are able to produce craving and that they help elicit drinking, but to my knowledge there has been no proof that they cause endorphins to be released.  If the cues do not cause endorphins to be released, opioid antagonists could not affect whatever effects the cues have: specifically, opioid antagonists could not block craving elicited by alcohol-related stimuli.

Against the hypotheses that the cues release endorphins is the apparent failure of the cues to produce reinforcement.  Does watching a bar or seeing advertisements for alcoholic beverages cause reinforcement?  Probably not.  It does not satisfy the desire for alcohol.  In fact, there are other theories claiming such stimuli do not product pleasant reactions but instead produce unpleasant feelings similar to withdrawal.

Click HERE for part two