Alcohol: Brain Evoked Potentials as Predictors of Risk


Sensory evoked potentials are capable of demonstrating brain sensory and cognitive function. These measures of brain activity can be used to demonstrate genetic influences in alcoholism. Auditory evoked potentials have been used successfully to demonstrate inherited differences in alcohol sensitivity. As in animal models, these inherited differences are limited to particular neuronal mechanisms and are not a general property of all neurons. The P300 wave, which is elicited in particular paradigms in which the subject is required to attend to specific stimuli, is smaller in subjects who are at high risk for alcoholism by virtue of having an alcoholic father. These subjects at risk for alcoholism show lower P300 amplitudes in paradigms in which they are given small doses of alcohol. P300 is also small in younger high-risk subjects who have never been exposed to alcohol. The evoked potential data are in general agreement with earlier electroencephalographic data that suggested the presence of electrophysiological abnormalities in the children of alcoholics.

Sensory evoked potentials have been used extensively over the past decade to characterize abnormalities in brain function. This chapter describes their use in studies of genetic factors of risk for alcoholism. We shall review (1) the rationale for using evoked potential methods to study brain function, (2) their use in determination of the heritability of sensitivity to alcohol, (3) studies of their abnormalities in children of alcoholics, who form a group at high risk for alcoholism, and (4) potential strengths and pitfalls in their use in investigations of genetic factors in alcoholism.

Evoked Potentials and Brain Function

Studies of the familial transmission of alcoholism have suggested the existence of genetic factors in severe alcohol abuse. The existence of genetic factors further implies that specific biological abnormalities are at least partly responsible for the development of alcoholism. The spontaneous electroencephalogram (EEG) has been extensively studied using the technique of frequency analysis in an attempt to uncover such factors. Alcohol has been shown to have differing effects on the EEG in individuals with various genetically determined variants of alpha and beta rhythms. Children of alcoholic fathers, who form a presumptive high-risk group for alcoholism, have been shown to have an excess of high-frequency activity. This abnormality was present only in the male children. Further investigation of male children of alcoholics showed that after alcohol they exhibit greater increases of slow alpha energy than control subjects without a family history of alcoholism. Taken together the EEG data suggest that subjects at risk for alcoholism, particularly males, may have borderline abnormal EEGs which are normalized after alcohol administration. Thus, these subjects are more responsive to alcohol than normal controls and might tend to seek alcohol as a corrective medication for a genetically determined abnormality.

These provocative findings are nonetheless limited in their impact because of a lack of specificity in the EEG measure itself. EEG waves and their rhythms have never been definitively linked to particular neuronal mechanisms in the cerebral cortex or underlying brain structures. They are likely to represent the summated action of many neuronal systems. The EEG therefore may not be the optimal parameter to search for genetic mechanisms, since presumably such mechanisms would be reflected specifically in the action of restricted neuronal groups. For example, the long-sleep and short-sleep mice, which are inbred strains selected for differences in alcohol sensitivity, show difference in sensitivity only in restricted neuronal populations. Although single-neuron specificity is not normally attainable in human studies, study of restricted brain areas may be possible using the evoked-potential technique.

Like the EEG, the evoked-potential technique relies on the recording of electrical activity at the scalp surface. The principal difference is that the evoked potential uses sensory stimulation, such as tones or visual patterns, to activate selectively particular neuronal circuits involved in sensory processing. Because only a small fraction of EEG activity is ascribable to any particular neuronal circuit, a computer averaging technique is used to enhance the response evoked by the stimuli over other brain electrical activity. Generally trains of 30 or more identical stimuli are presented to the subject. The computer adds the electrical activity following each stimulus to its registers, so that activity that is not time locked to the stimulus tends to cancel itself out. The computer-derived average of brain electrical activity following the stimulus is referred to as the evoked potential.

This potential, which is a series of waves, represents sensory processing of the stimulus by various brain centers. Although not every wave can be unambiguously identified as originating in a particular neuronal circuit, identification of neuronal generators from brainstem relay nuclei through the cerebral cortex has been made for a number of the waves. Use of this technique to study altered brain function is still in its infancy; nevertheless, several pathological states have been linked to evoked-potential abnormalities. Evoked potentials have also been used to detect alteration in brain function in normal subjects produced by alcohol. A feature of evoked potentials that has received increasing interest has been their ability to correlate with cognitive features of the stimulus presentation. This feature has been most widely studied with a positive potential that occurs approximately 300 msec following the presentation of a stimulus and is therefore referred to as P300. This potential is increased when subjects are induced to pay more attention to the stimulus, particularly when it occurs infrequently among other stimuli.

Evoked Potentials and Heritability of Alcohol Response

One of the factors determining risk for alcoholism is thought to be responsiveness to alcohol; it has been postulated by Goodwin that alcoholics are relatively unresponsive to low-dose alcohol. Animal studies show that increased or decreased responsiveness to alcohol can be determined by selective breeding.

Our own work has used auditory evoked potentials to examine heritability of the response to alcohol in normal subjects. It is the only evoked-potential study to directly address this question, and therefore we shall briefly review it here. The study selected normal subjects who were monozygotic twins, dizygotic twins, siblings, or adoptees, all same sex and reared together. This scheme allows any genetic influences to be manifest as similarity in response between monozygotic twins with less similarity in dizygotic sibling pairs and no similarity in adoptees. Similarity is expressed statistically as intraclass correlation. This correlation compares the differences between two members of a pair with differences between all the pairs in the sample population. Positive values of this statistic indicate that subjects are more closely similar within a pair than within the population as a whole. Negative values indicate that the similarities between the two members of the pair are less than the similarities between all the pairs in the population. To prevent problems of variance due to previous alcohol experience, only subjects aged 25-45 with no personal history of alcoholism were participants in the study.

The evoked-potential wave chosen for this study was the vertex P50 response to auditory stimuli which were presented in a conditioning testing paradigm. The P50 wave was chosen because its response to alcohol shows high test-retest reliability, in subjects who were tested on two separate occasions. The conditioning testing paradigm presents stimuli in pairs; generally, the response to the second or test stimulus is decremented because of inhibitory mechanisms activated by the first or conditioning stimulus. This paradigm thus is commonly used by neurophysiologists to examine excitatory pathways that give rise to the conditioning response and inhibitory mechanisms that cause suppression of the test response. Alcohol generally diminished the amplitude of the conditioning wave but also diminished the suppression of the test wave. Thus, alcohol was likely to have effects on both excitatory and inhibitory mechanisms in auditory sensory circuits.

Seventy-seven subjects were recorded. The odd number reflects one grouping of three siblings. The percent change in the amplitude of the conditioning P50 wave and the percent change in the test response after alcohol administration were both calculated. As Table Effects of Relationship on Ethanol’s Action on the Auditory Evoked Potential: Intraclass Correlations shows, the intraclass correlations for the effect of alcohol on the initial conditioning wave were negative for all four sibling groups, which indicates no genetic influence on this effect of alcohol. The change in suppression, however, shows more similarity for monozygotic twins than for the other three groups. Dizygotic twins are intermediate in their values. The data imply that the effects of alcohol on inhibitory mechanisms that give rise to the suppression of the test waves in the conditioning testing paradigm are genetically influenced. The lack of positive intraclass correlation for the adoptees reared together as well as for the siblings suggests that environmental factors, at least postbirth, are unlikely to play a significant role in the development of sensitivity to alcohol in adults.

Although this study does not directly indicate that alcohol itself has an inherited brain basis, it strongly suggests that sensitivity to alcohol, one of the putative bases for alcoholism has a significant genetic influence.

Table Effects of Relationship on Ethanol’s Action on the Auditory Evoked Potential: Intraclass Correlations

Conditioning amplitude Suppression of test amplitude
Monozygotic twins (n = 20) -0.18 0.43
Dizygotic twins (n = 12) -0.02 0.22
Siblings (n = 23) -0.18 -0.58
Adoptees (n = 22) -0.54 -0.16

Evoked Potential in High-Risk Subjects

The high-risk strategy has been employed in several evoked-potential studies to determine whether there are differences in brain function as revealed in specific evoked-potential waves in children of alcoholics. The initial study was performed by Elmasian et al. In this study 15 subjects with a biological father who met the criteria for primary alcoholism were compared with 15 male subjects from the same university community without a similar family history. Subjects were recorded after placebo, low-, and high-dose ethanol treatment. Evoked potentials were recorded to a train of tone bursts presented in groups of six. In about 20% of the groups the last tone was shorter than the preceding tones. This shorter tone was the target tone and required a button press. In all subjects the target tone elicited a larger P300 wave, reflecting the special attention given to it in the experimental paradigm. As reported by other investigators, alcohol reduced the size of the P300. This study demonstrated in addition, however, that the reduction was significantly greater in family-history-positive subjects than in family-history-negative subjects. There was also a significant reduction with placebo, which was described as sample cups whose rims had been wetted with alcohol, in the family-history-positive group but not in the family-history-negative group. No significant difference was described in the initial P300 between the two groups prior to the administration of alcohol.

Neville and Schmidt performed a similar study in 1983 in which 10 male subjects between age 20 and 25 years with alcoholic fathers were studied along with 10 controls. In addition, subjects were excluded if their biological mother had ever used alcohol in excess, to prevent in utero exposure from being a significant factor. The study also involved a more extensive questioning of the subjects to assure that both frequency of social use of alcohol and the amount of alcohol used per drinking episode did not differ between its two subject groups. Alcohol was not administered to these subjects as part of the experimental paradigm to prevent the expectancy of alcohol ingestion from altering the evoked-potential recordings. The marked placebo effect seen in the first study had raised as a possibility that expectancy even in the placebo group might have contributed to the some of the marked changes in P300. In addition, a simpler P300 paradigm, which was essentially a two-tone discrimination test, was used. There were no mean differences between the family-history-negative and the family-history-positive subjects in their P300. However, when frequency of alcohol use measured as drinks per occasion was plotted against latency of the P300 wave or amplitude of the wave, there was a significantly longer latency and smaller amplitude in subjects who drank more than two drinks per occasion in family-history-positive subjects than in family-history-negative subjects. The data suggest that alcohol use is differently reflected in family-history-positive subjects than in family-history-negative subjects. Whether the evoked-potential abnormality reflects a brain abnormality that is causative of increased drinking or whether increased drinking has different effects in family-history-positive than in family-history-negative subjects cannot be determined from this study.

In this regard the Begleiter et al. study in 1984 is an important contribution. Younger subjects, aged 7-13 years, who had never been been exposed to alcohol were the subjects, so that alcohol itself as a causative factor was eliminated. Twenty-five sons of alcoholic fathers were matched to 25 control subjects. Excluded were subjects whose mothers had alcohol exposure during pregnancy or a history of alcoholism. A visual rather than an auditory stimulus was presented. The subjects were instructed to press either a right or left switch when the right or left ear was showing on a figure of a head. An easy condition in which the head was upright was contrasted with a difficult condition in which the head was rotated so that the ear on the right side of the figure would actually represent the target left ear. For both the easy and difficult situations, the P300 waves were significantly larger in normal controls than in the high-risk subjects. Since the reaction time did not differ between the groups, it is not likely that these differences are due to problems in the boys’ willingness to participate in the task.

Taken together these studies suggest that the mechanisms that give rise to P300 may be dysfunctional in subjects who have a family history of alcoholism. This deficit may become apparent before any exposure to alcohol, even though alcohol itself can cause further deficits in P300. The differences before alcohol are more striking in children aged 7-13, but young adults show a greater sensitivity to alcohol in terms of the reduction of the size of this wave, which suggests a similar neurophysiological vulnerability. Although the site of the P300 wave has not been definitely determined, it may originate in the hippocampus.

One study has failed to find a P300 difference in family-history-positive and family-history-negative subjects. The Schmidt and Neville study used a very dissimilar task, however, which asked subjects to distinguish whether two letters presented 1.8 sec apart were rhyming and not rhyming. The test elicited a number of waves, one of which was the contingent negative variation, which is an expectancy wave between the first and second letters. The P300 was quite small in this task and did not differentiate family-history-positive from family-history-negative subjects. The subsequent wave, the N430, which is perhaps related to the decision as to whether the letters rhyme or not, was significantly smaller over the right hemisphere in alcoholic subjects.

Strengths and Possible Pitfalls

The evoked-potential studies, like the earlier EEG findings, suggest that the relative sensitivity to alcohol is under genetic control and that alcoholism itself may reflect a brain abnormality that can be observed in electrical brain activity in subjects before exposure to alcohol. Furthermore, the evoked-potential data suggest that the neuronal alterations may be rather specific. For example, effects of alcohol on the inhibitory gating mechanisms of the P50, but not the excitatory mechanisms that give rise to the wave itself, seem to be genetically influenced. Only the later cognitive waves P300 and N430 seem to reflect the familial predisposition for alcoholism. Earlier waves of the evoked potential, including both brainstem and middle-latency waves, seem to be unchanged between family-history-positive and family-history-negative subjects. Although definitive localization of the neuronal mechanism has not yet been achieved, the evoked potential seems to indicate that only certain brain functions differentiate family-history-positive and family-history-negative subjects.

There are numerous pitfalls, however, which remain in the data. The EEG data, for example, show changes only in males and not in females. Only male subjects were tested in these evoked-potential paradigms. There were no significant differences between females and males on the evoked-potential study that looked at alcohol sensitivity. Although male and female alcoholism may well have different genetic bases, the question of such difference has not yet received full experimental treatment. At this stage less than 100 subjects who are family-history-positive have been published, so that the data further suffer from being based on a relatively small sample.

There are more global problems, however, which need to be considered as further experiments are designed. One problem is the assessment of family-history risks, which has been described by Cloninger and Reich. They point out that assessment of family history by means of a single relative does not adequately address the problem of heritability in what may well be a poly-genetic illness. Quantitative measures of the distribution of alcoholism in the family tree of both parents may well be necessary, even if alcoholism in the mother per se is excused because of the in utero exposure. Other issues, such as environmental factors during early childhood, that could affect complex cognitive functions reflective in the evoked-potential waves being analyzed also have to be excluded by adoption paradigms. Monozygotic twin strategies have been employed to look at alcohol sensitivity. Such strategies would be more difficult for a study of risk factors; nevertheless, they may be necessary to eventually separate environmental from genetic factors.

The age at which to record subjects is particularly problematic. The age of onset of alcoholism has become earlier in our society, so that as early as age 14 it may be impossible to get an accurate sample of children of alcoholics who themselves have not begun drinking. Using very young children for the evoked-potential recording has the advantage of assuring that alcoholic exposure is minimal or nonexistent, but it has the disadvantage that the nervous system may not be mature. Cognitive evoked potentials in children are not well studied, but what data are present suggests that the potential has not fully matured. Amplitudes tend to be smaller in children, even in children who are family-history-negative, than they would be in normal adults. Thus, it is possible that what is being measured in these young children is not a defect in the nervous system that is permanent, but rather a delayed maturation. It may be that the early onset of alcoholism in later teenage years may reflect a maturational problem rather than a more permanent brain structural problem, so that the data on this issue might have great treatment importance for a young alcoholic.

The recording of older subjects, even young adults, is problematic in a different way. If many of the first bouts of alcoholism occur in teenage years, young adults who are not alcoholic may represent survivors rather than potential victims of alcoholism. They may well be family-history-positive subjects who are destined not to drink because of their social experience or because of compensatory biological factors. Their brain abnormalities may not be representative of those in subjects who are already alcoholics. In this regard, Neville’s findings of differences in subjects at risk who drink at least two drinks per occasion, compared to subjects at risk who drink less or not at all, may suggest that separation of high-risk drinkers by drinking behavior may be critical. Certainly, a longitudinal perspective may now be warranted, given the promise of the current studies and their limitations.


Selections from the book: “Recent Developments in Alcoholism. Volume 6: Posttraumatic Stress Disorder. The Workplace. Consequences in Women. Markers for Risk.” Edited by Marc Galanter. An Official Publication of the American Medical Society on Alcoholism, the Research Society on Alcoholism, and the National Council on Alcoholism. 1986.