Before the evidence for EEG, EP, or ERP anomalies in patients with histories of alcohol or drug dependence is reviewed, it would be valuable to highlight some of the methodological and interpretive problems that are inherent in studying patients with drug use histories as well as several problems that are created by researchers themselves. These problems are not specific to EEG/EP/ERP research. Rather, they generalize across all of the neuroimaging modalities.
1. It is a logical fallacy to unquestionably attribute an EEG/EP/ERP difference between one group of substance dependent patients and a group of healthy, non-drug-abusing volunteers to an effect of substance abuse. Collectively, substance-dependent patients are known to exhibit higher-than-normal rates of comorbid psychopathology, polydrug abuse, medical disorders (e.g., cirrhosis, hepatitis, cardiovascular disorders, diabetes, HIV/AIDS, malnutrition), licit and illicit medication use, head injuries, seizures, and epilepsy (Lowinson et al., 1992). In addition, substance dependent patients exhibit low levels of compliance/motivation, below average IQ scores, and higher-than-normal rates of childhood psychopathology. Other complicating factors include dysfunctional environmental or genetic histories, and educational, social, or cultural factors (e.g., reading disability or illiteracy, language barriers, a below-average level of educational achievement that is overestimated by educational attainment)—all of which predate their substance abuse careers. The simple demonstration of a difference in EEG/EP/ERP (or other neuroimag-ing) indices between a group of patients and a group of healthy volunteers is therefore insufficient for inferring causation, unless the researchers have either matched the groups on all of these potential confounds, excluded all patients with these confounds, included additional control groups, or explicitly examined the role of these confounds in planned or post hoc analyses. It is unfortunate that many neuroimaging researchers have failed to take these complexities into account.
2. It can be misleading to report findings and derive hypotheses from studies of small numbers of patients, because of the high risk of both type II (from inadequate statistical power) and type I (from outliers, biased sampling, unrecognized confounds) errors. Many errors of inference may also derive from a failure to consider the impact of a small N on the assumptions and validity of the most commonly employed parametric and nonparametric inferential statistics. Furthermore, studies of small numbers of patients preclude the researcher from conducting the types of subgroup analyses which are best suited to revealing the effects of mediating and moderating variables.
3. One cannot assert that a study is examining the effects of acute withdrawal based upon the point in time at which data were collected. That is, a study examining patients several hours after their last use of a drug is not necessarily a study of withdrawal. The differences can just as reasonably be attributed to premorbid factors, a persistent chronic effect, or a protracted withdrawal phenomenon associated with the patient's prior abuse of another substance.
4. It is problematic when neuroimaging studies fail to adopt appropriate and state-of-the-art methods for psychiatric assessment and relapse monitoring. In any standard clinical trial, a failure to distinguish between abuse and dependence diagnoses, or to fully assess psychiatric and medical comorbidity in a structured interview, would be considered naive and "fatally flawed". Similarly, failing to conduct frequent urine toxicology screens or to work with collateral informants to verify the patient's self-reported duration of abstinence would also be considered unacceptable. Yet, a number of neuroimaging studies do not meet these minimal standards.
A number of early, i.e., 1960-1980, studies (reviewed in Porjesz and Begleiter, 1985) indicated that the abrupt cessation of chronic alcohol dependence is accompanied by marked changes in sensory evoked potentials. These changes include an enhance-
ment of the P100 component of the flash-evoked visual potential, and an enhancement of the amplitudes and reduction of the latencies of both auditory brainstem responses and median nerve somatosensory evoked responses. Similar changes have also been reported in the amplitudes of later evoked potential components, such as P300. These EP and ERP findings suggest that the early phase of withdrawal is marked by enhanced central nervous system arousal and excitability. The severity and duration of the acute hyperexcitability phase is determined by a number of factors, including drinking (Begleiter and Porjesz, 1979) and seizure (Brown et al., 1988; Noldy and Carlen, 1990) histories.
Consistent with the results of EP and ERP studies, the vast majority of studies of the spontaneous EEG find evidence for CNS hyperexcitability via the demonstration of enhanced power in the high frequency beta band (Coger et al., 1978; Kaplan et al., 1985; Spehr and Stemmler, 1985; Frank et al., 1986; Krauss and Niedermeyer, 1991; Bauer, 1994a; Costa and Bauer, 1997; Winterer et al., 1998) and decreased power in the slower alpha band (Zilm et al., 1980; Kaplan et al., 1985; Spehr and Stemmler, 1985). There are also scattered reports of enhanced delta or theta activity and qualitative EEG abnormalities (e.g., spiking, epileptiformlike discharges; Holmes and Korteling, 1993). The latter findings appear to be limited to the subset of alcoholic patients with histories of cerebellar or liver disease, withdrawal seizures, severe vitamin deficiency, or head injuries. Qualitative EEG abnormalities are rare in alcoholic patients without these medical complications (Begleiter and Platz, 1972).
As central nervous system hyperexcitability subsides, after approx 1-2 wk, it is replaced by a more protracted period, characterized by attenuated or delayed evoked potentials (Chan et al., 1986; Cadaveira et al., 1991, 1992), but a persistent elevation in EEG beta power (Bauer, 1994a; Costa and Bauer, 1997; Winterer et al., 1998). The attenuated sensory EPs typically normalize (Chan et al., 1986; Bauer and Easton, 1996) unless the patient is complicated by other disorders (e.g., liver disease, vitamin deficiency, Wernicke-Korsakoff's syndrome, head injury). The later, endogenous ERP components (e.g., P300) can remain attenuated, however, after as long as 1 yr of abstinence. The elevation in EEG beta power, and the attenuation of P300 amplitude, persisting beyond the initial months of abstinence appear to reflect a premorbid deficit (Elmasian et al., 1982; Pfefferbaum et al., 1991; Bauer and Hesselbrock, 1993, 1999a,b; Hesselbrock et al., 1993a,b, Bauer, 1994a; Bauer et al., 1994a,b; O'Connor et al., 1994; Polich et al., 1994) and not a persistent effect of alcohol.
To our knowledge, only one study has examined EEG activity in benzodiazepine dependent patients without the complication of a seizure history (Hallstrom and Lader, 1981). Under controlled conditions, 10 patients, who reported an inability to discontinue benzodiazepine abuse, were prescribed either 20 or 135 mg diazepam per day. During a subsequent period in which these doses were tapered downward, patients reported heightened levels of anxiety and perceptual disturbances. Changes in EEG beta power, that were opposite to those associated with the initiation of diazepam treatment, also were found.
It has been difficult for researchers to identify EEG, EP, or ERP differences associated with chronic marijuana/THC use. In part, the difficulty arises from the substantial variability in the intensity and frequency of use even among subjects who uniformly meet diagnostic criteria for cannabis dependence (Wert and Raulin, 1986). In addition, subjects who meet criteria for cannabis dependence often meet criteria for dependence upon other substances or possess premorbid or comorbid psychopathology. Many of the initial neuropsychological and neurophysiological findings in chronic marijuana/THC users have not been replicated for these and other reasons.
Struve and colleagues have discussed these problems and noted that many of the extant EEG studies are deficient for various reasons (Struve et al., 1989, 1994, 1998). They note, for example, that early reports of a high prevalence of qualitative EEG abnormalities among chronic marijuana users were misleading because the prevalence is in fact no greater than the base rate in the normal population. Other qualitative EEG studies reporting a high prevalence of abnormalities have been criticized for imprecise or biased definitions of abnormality, or other problems.
A recent quantitative analysis of the EEG in chronic marijuana/THC users is noteworthy for its many strengths. Struve and colleagues (1998) examined 15 subjects who reported using THC on a daily basis for 15-24 consecutive years. None of the subjects reported significant levels of use of other psychoactive drugs and all were free of serious psychopathology. A group of moderate duration (3-6 yr), daily THC abusers was recruited as a comparison group, as was a larger group of normal, non-drug-abusing volunteers. The results of the study revealed a THC duration-related increase in EEG theta power at all scalp locations. Because the design of this study included subjects with a regular and daily pattern of THC use, the power for detecting an EEG difference between users and nonusers was enhanced. By comparing long and moderate duration THC abusers, and by excluding subjects with psychiatric complications, these investigators were also able to discount alternative explanations for their results.
Studies of short latency evoked potentials have not revealed evidence for a deficit in peripheral or central processing of visual, auditory, or somatosensory stimuli (Patrick and Struve, 1994; Patrick et al., 1997) among chronic THC users. Studies of the P300 ERP have also revealed inconsistent results (Solowij et al., 1991, 1995; Patrick et al., 1995).
An early study examined EEG activity in 63 heroin-dependent patients (Volavka et al., 1970). EEG irregularities, identified by clinical raters, were present in approx 50% of the patients. Unfortunately, no consideration was given in this study to the complicating effects of other drug use, head injury, or psychopathology.
In a 1997 study which excluded individuals with these complications (Fig. 1), we found no evidence of quantitative EEG abnormalities in a group of 19 heroin dependent patients, abstinent for 1-5 mo, compared to healthy controls (Costa and Bauer, 1997). Importantly, the patients in our study failed to exhibit the same enhancement of EEG beta power demonstrated in cocaine or alcohol dependent patients via the same
FAST BETA POWER AGE - CORRECTED MEAN (+1 SE)
P<0.05, RE: NORMAL CONTROL T
Was this article helpful?