Michael R. Trimble MD
Professor of Behavioral Neurology, Department of Neurology, Institute of Neurology, London, United Kingdom
The use of antiepileptic drugs in the management of patients with psychiatric problems but who do not have epilepsy has a long history. In fact, going back to the nineteenth and early twentieth centuries, much more phenobarbital and bromides were prescribed to patients with psychiatric conditions, mainly in the neurotic spectrum, than were ever prescribed for epilepsy. This trend continued in the mid-twentieth century, the nonepileptic indications for phenytoin being broad and encompassing a number of conditions that today would be categorized as anxiety related. In the same vein, drugs such as the benzodiazepines are known to have powerful anxiolytic effects, as well as being anticonvulsants.
Thus, the profile of the antiepileptic drugs is intimately biologically entwined with the underlying neurobiology of psychiatric disorders. This is further reinforced by the fact that benzodiazepine inverse agonists can precipitate not only anxiety and panic attacks, but seizures. There also is an interesting literature on the development of psychiatric syndromes in epileptic patients who have their antiepileptic drugs stopped (e.g., for surgery) and, either with or without the development of withdrawal seizures, can develop a variety of psychopathologic symptoms.
The history of the introduction of carbamazepine for use in psychiatry goes back to the 1970s, shortly after the drug was introduced in the management of epilepsy. In these times (indeed, as of today) in Japan, most patients with epilepsy were treated by psychiatrists. Psychiatrists were thus using the newly discovered anticonvulsant properties of carbamazepine in patients with epilepsy, but at the same time noticed an effect of the drug on mood. These same psychiatrists therefore used carbamazepine in the management of patients with affective disturbances and noted an effect on bipolar disorders. These observations were taken up widely thereafter, both in Europe and the United States, and there now is a considerable body of evidence showing that carbamazepine has an effect on mood regulation.
This review is divided into three sections, the first dealing with the clinical evidence in epilepsy, the second in patients without epilepsy, and the third with some underlying theoretical principles.
CARBAMAZEPINE AS A MOOD-STABILIZING AGENT: STUDIES IN EPILEPSY
Ever since its introduction into the clinical management of epilepsy, carbamazepine has been reported to possess psychotropic properties (1). In an extensive review, Dalby reported that 90% of the published reports of carbamazepine in epilepsy mentioned its psychotropic effects, and that in 40 reports of over 2,000 patients, such effects were seen in approximately 50% of patients. The psychotropic effect was described as follows:
The psychotropic action [of carbamazepine] is uniformly described as an increase in psychic tempo in patients with the so-called epileptic personality. The slowness, sluggishness and stickiness of the perseverations and the apathy and lack of initiative of the patients with longstanding severe epilepsy, usually uncontrolled on large doses of phenobarbital, phenytoin and primidone, diminishes and a quickening of thought and action occurs. The affective changes, such as irritability, aggressive tendencies, impulsivity, dysphoric episodes, and states of depression and anxiety are reduced or abolished giving way to an elevation of mood (1).
A number of controlled investigations in epileptic patients are available. Some of these directly assess affective symptoms; others are of patients with less well defined behavior problems, in some studies patients with learning disability being included. Rajotte et al. (2) noted a significant improvement in behavior with carbamazepine compared with placebo on behavior rating scales in a doubleblind, crossover trial. Similar results were obtained by Marjerrison et al. (3) and Cereghino (4). Rodin et al. (5)
stabilized 45 patients on phenytoin and either primidone or carbamazepine. After 3 months, those on carbamazepine were changed to primidone, and vice versa. Patients were assessed using the Minnesota Multiphasic Personality Inventory (MMPI). Patients scored higher on the psychopathic deviate scale of the MMPI and became more depressed when on primidone, whereas on carbamazepine they became less depressed. In a similarly designed study, Dodrill and Troupin (6) compared carbamazepine with phenytoin over a 4-month period using a double-blind, crossover design, with patients randomly assigned to the drug of study. On the MMPI, every clinical scale favored the carbamazepine treatment period, the results being statistically significant for the F scale, which relates to feelings, attitudes, and emotions. Interestingly, in the aforementioned studies, the psychotropic effects of carbamazepine appeared dissociated from its effect on seizures.
Several studies were carried out evaluating changes of behavior in patients with epilepsy using serum level monitoring of the antiepileptic drugs. In one study of children with epilepsy, Trimble and Corbett (7) analyzed correlations between behavior deviance scores for conduct disorder and serum levels of antiepileptic drugs. The higher deviance scores were associated with higher levels of some anticonvulsants (particularly phenobarbital), whereas for carbamazepine the correlation was negative. A significant difference between the correlations for phenobarbital and carbamazepine was noted.
There are three studies in adult patients showing a similar relationship between serum levels of carbamazepine and scores on behavior rating scales. Rodin and Schmaltz (8) gave the Bear-Fedio Personality Inventory, a scale specifically constructed to assess interictal behavior changes in epilepsy, to 148 patients with epilepsy. They noted significant associations between rating scale scores and carbamazepine, but not for other anticonvulsants. Specifically, they reported that carbamazepine levels were significantly inversely correlated with the total score of psychopathology, and subscores of elation, philosophical interests, a sense of personal destiny, altered sexuality, and hypergraphia.
Andrewes et al. (9) administered to 42 newly referred patients with epilepsy the Mood Adjective Check List and followed the progress of patients receiving monotherapy either with phenytoin or carbamazepine. Again, significant associations were noted between mood scores and serum levels of carbamazepine—the higher the level of carbamazepine, the lower the ratings for anxiety, fatigue, and depression
Robertson et al. (10) provided further data on this relationship. They used the Research Diagnostic Criteria for Major Affective Disorder in patients with epilepsy, and measured serum levels of anticonvulsant drugs. Patients on phenobarbital were more likely to report higher depression scores, whereas patients on carbamazepine rated themselves as less depressed and had lower anxiety scores, and a negative correlation was found between trait anxiety and serum carbamazepine levels.
The association of phenobarbital with affective disorder, and the contrast with carbamazepine, also were seen in the studies of Brent and colleagues (11,12). An overrepresentation of patients with epilepsy was noted in a consecutive series of suicide attempters seen at a children's hospital. They noted that of 131 consecutive suicide attempts made by 126 children seen over a period of 5 years, 9 had epilepsy. This was over 15 times the expected frequency. The attempters with epilepsy made more medically serious attempts, showed more premeditation, and had a higher frequency of suicidal attempts both before and after the index event than did nonepileptic children. Eight of these nine children were on phenobarbital.
The authors then carried out a comparative study of affective disorder in 15 children with epilepsy treated with phenobarbital compared with 24 treated with carbamazepine. The groups were similar over a wide range of demographic and seizure-related variables, but patients on phenobarbital showed a much higher frequency of major affective disorder (40% versus 4%) and of suicidal ideation (47% versus 4%). Depression was particularly noted in those children who had a family history of major affective disorder among first-degree relatives. The contribution of the antiepileptic drugs, however, was supported by the fact that the diagnosis of major depressive disorder was made within 12 months of the initiation of the medication in 33% of the phenobarbital group, and in none of the carbamazepine group. Sixteen patients in the carbamazepine group had been treated previously with phenobarbital. Eight reported having an affective disorder on the phenobarbital, and all eight lost their affective symptoms when switched to carbamazepine.
Other studies have examined the behavioral toxicity profile of antiepileptic drugs, including carbamazepine, looking at the effect of the drugs on mood. Thompson and Trimble (13) followed a group of 20 patients who underwent a reduction in the number of anticonvulsants they were prescribed, and 15 patients who underwent a similar reduction, but who had one of their medications substituted for carbamazepine. There was no consistency in either group regarding which antiepileptic drug was withdrawn. They also had 10 patients who did not undergo any anticonvulsant drug changes. The profile of their mood was studied on three occasions, at baseline, at 3 months, and then at 6 months; in addition to the ratings of mood, a number of cognitive tasks were performed. The overall findings suggested that the drug changes made in both groups of patients resulted in improvements of performance on psychological tests, and that these were not simply the result of practice. Patients changing to carbamazepine, either alone or in combination with existing medication, displayed more widespread improvements in test performance, particularly with regard to measures of memory,
than those that did not have the carbamazepine substitution. Interestingly, these patients were also shown to improve in mood. When patients with initial high rating scales scores for mood disorder were examined, a significant antidepressant affect was noted on switching to carbamazepine.
These studies of people with epilepsy often are difficult to interpret because of the methodologic issues concerned. Thus, patients have epilepsy, and any change of seizure frequency may be expected to alter patients' mood. Further, patients often are on polytherapy, and any reduction of polytherapy toward monotherapy may bring about improvements in mood that perhaps are secondary to resolution of drug side effects or improvement of cognitive function. Third, the monitoring of patients in terms of their mood often has been based on purely clinical grounds and the rating scales used (with few exceptions) have been validated in patients with psychiatric illness, and therefore may not necessarily be of value in patients with epilepsy.
Nevertheless, there is an interesting trend that runs through the literature on carbamazepine since its introduction, namely, that it does appear to have a positive effect on behavior in patients with epilepsy. The most commonly reported behaviors that seem to be improved when the drug is prescribed lie along an affective spectrum, but include improvements in aggressive and irritable behaviors; carbamazepine also has some relationship to elation (8), suggesting possible control of hypomania and mania.
CARBAMAZEPINE AND AFFECTIVE DISORDERS IN NONEPILEPTIC PATIENTS
The terminology used in the following studies varies, particularly since the introduction of the third and fourth editions of the Diagnostic and Statistical Manual of Mental Disorders(DSM-III and DSM-IV), the latter being the standardized statistical and classification manual used by most psychiatrists. The early studies were carried out using basic clinical descriptions.
In the diagnostic manuals, under the affective disorders, several conditions are described. The syndromes most treated with carbamazepine have been bipolar disorders, either for the acute treatment of mania per se, or for the longer-term prophylaxis of bipolar affective disorders.
Thus, DSM-IV (14) distinguishes mood episodes (manic or hypomanic episodes) from mood disorders. Among the mood disorders, a distinction is made between bipolar 1 disorders (mania with or without evidence of a previous depressive episode) and bipolar 2 disorders, which essentially are recurrent major depressive episodes with hypomanic episodes. Until the introduction of carbamazepine for the management of bipolar disorders, the mainstay of treatment was lithium. Lithium was shown not only to have an effect on acute mania, but to prevent the recurrence of episodes of affective disorder in patients with bipolar disorder. However, lithium has many disadvantages. Not only does it have significant side effects, some of which may be profound (e.g., renal problems or hypothyroidism), it is associated with cognitive obtunding, and many patients find it difficult to take. Further, serum level monitoring is carried out because it is quite easy to induce lithium toxicity, which, if unrecognized, can lead to permanent neurologic impairment.
Lithium, in contrast to carbamazepine, is not anticonvulsant (if anything, in overdose it is convulsant) and clearly has a limited spectrum of clinical activity in the sense that it is not used as widely as carbamazepine in neurologic syndromes, such as for paroxysmal pains. However, some of the early studies in psychiatric patients suggested that patients responding to lithium may have a clinical profile different from those responding to carbamazepine.
Carbamazepine in Acute Mania
Many of the observations have been anecdotal, but over time at least 12 double-blind studies have been carried out comparing the effects of carbamazepine in acute mania with placebo, lithium, or neuroleptic medication. These studies reveal that it is equivalent to lithium over a period of up to 8 weeks and that the time course of the antimanic effect is, if anything, a little slower than with neuroleptics, but equivalent to lithium. These studies have been reviewed in detail by others (15,16).
More recent studies include those of Okuma et al. (17), who compared carbamazepine and lithium in a doubleblind trial for 4 weeks. Carbamazepine was noted to be equivalent to lithium in outcome. Small et al. (18) compared carbamazepine with lithium in a double-blind trial that lasted 8 weeks, and again found no difference between the two drugs. Review of some of the clinical data suggested that patients who responded best to carbamazepine were those who were more manic at onset, were more dysphoric, and had a history of rapid cycling in the year before admission. Interestingly, these variables were said to be associated with a poorer response to lithium.
On account of these studies, and clinical impressions, carbamazepine now finds a role in the clinical management of patients with acute mania. There are several advantages to this drug over the use of neuroleptics, not the least being that if patients are kept on a drug prophylactically, the possibility of the later development of extrapyramidal disorders with neuroleptic drugs is avoided. This is particularly relevant for patients who are refractory to lithium and require an alternative to it.
The dose of carbamazepine used in acute mania is somewhat higher initially than in epilepsy because it is important to bring the symptoms under rapid control. This increases the possibility of provoking central nervous system (CNS) side effects. However, in mania, dosages often can be
increased quite rapidly (by 200 mg/day) and tolerated until clinical responses are observed. Dosages up to 1,800 mg/day often are used. There is no relationship between the serum levels of carbamazepine and the antimanic effect; the mean levels of carbamazepine in those treated successfully often are below the upper limits of tolerance. In clinical practice, there are some patients who are treated on a combination of lithium and carbamazepine, which may tend toward greater CNS toxicity. Patients with underlying neurologic disease (secondary manias) are particularly susceptible to these problems.
Carbamazepine as Prophylaxis in Bipolar Disorder
There are at least 14 controlled (or partially controlled) studies of the use of carbamazepine in the prophylaxis of bipolar disorder, and these have been reviewed by Post and colleagues (19). In addition, there have been many uncontrolled studies. Most studies have been in comparison with lithium, although some have used placebo. Another clinical design has been to switch patients from lithium to carbamazepine if they have not responded to lithium.
In reviewing the literature on the use of carbamazepine in psychiatric disorders generally, there have been over 500 reports involving more than 2,000 patients, and most have had bipolar affective disorder. With regard to long-term prophylaxis, Post et al. summarized the 14 controlled or partially controlled studies in this area. Essentially, of 629 patients prescribed carbamazepine, a good response was noted in 390 (62%). Most of these studies were doubleblind, and several of them randomized.
The early studies came from Japan (20,21). Since then, studies have been undertaken by Post and colleagues and by several groups in Europe (22, 23, 24, 25). These studies demonstrated an equivalence between carbamazepine and lithium for prophylaxis, and certain patients appeared to respond preferentially. Patients who are referred to as rapid cyclers, namely, patients with an unstable bipolar disorder with rapid fluctuations (more than four episodes a year), are known to respond poorly to lithium and seem to do better on carbamazepine, or on a combination of carbamazepine and lithium, than on lithium alone. Indeed, there are several reports of lithium nonresponders gaining additional clinical improvement with carbamazepine added to the regime (26).
Serum Levels and Side Effects
The doses of carbamazepine given in prophylaxis of affective disorder are similar to those prescribed for epilepsy, but no study has shown any dose-effect relationship, or indeed serum level-effect relationship. In general, patients are started on 200 mg and titrated slowly. Mean serum levels, where they have been assessed, are within the therapeutic range for epilepsy. Unwanted side effects typical for carbamazepine are seen in a rather higher percentage of patients without epilepsy, and occur in up to a third. Although these are mainly gastrointestinal and neurologic, a rash has been reported in up to 15% of psychiatric patients (27). Thyroid function often is minimally affected, with a lowered plasma level of thyroxine and raised thyroid-stimulating hormone, but hypothyroidism itself is rare (28). However, thyroid function should be monitored, especially when carbamazepine is combined with lithium.
Hyponatremic syndromes are seen and may pose a particular problem in psychotic patients, who may become polydipsic, exacerbating the problems with sodium levels. Interestingly, this effect on sodium is the opposite to that of lithium, which tends to lead to sodium retention.
No clinical information has been gathered on the distinction between slow-release and conventional carbamazepine prescriptions in the treatment of affective disorders, although patients with psychiatric problems, and especially patients with mood disorders, can be remarkably noncompliant with medication and may be prone to abuse their prescriptions. Obviously, the use of slow-release preparations, given on a twice-daily basis, is a clinical advantage when using carbamazepine with psychiatric disorders.
THE USE OF CARBAMAZEPINE IN OTHER PSYCHIATRIC DISORDERS
In addition to its use in bipolar affective disorder and mania, carbamazepine has been used in several other psychiatric conditions, although controlled clinical information is sparse. In uncontrolled studies, moderate or marked clinical effects were reported when carbamazepine was added to existing therapy in 61% of 182 patients with schizoaffective or schizophrenic disorders (29). The carbamazepine usually is given in addition to a neuroleptic drug, and the initial hope was that carbamazepine would either replace the neuroleptic or would lead to a decrease of neuroleptic use, and hence to a diminution of the possible long-term side effects of the traditional antipsychotic drugs. In addition to the clinical reports, there are double-blind case studies and some clinical crossover studies comparing carbamazepine with placebo, which was added to existing neuroleptic treatments. These studies suggest responses in approximately 50% of schizophrenic patients. Although there are some reports of improvement in core schizophrenic symptoms (hallucinations and delusions), the main effects appear to be on other aspects of behavior, particularly the affective symptoms, aggressive disorders, and excitability.
Schizoaffective disorder also has been studied. Greil et al. gave 90 patients with schizoaffective disorder either lithium or carbamazepine, and although no significant difference was noted on treatment outcome, there were fewer side
effects with carbamazepine (30). Patients with schizodepressive syndromes appeared to respond better than patients with schizomanic presentations.
One problem that arises in treating schizophrenic patients with carbamazepine is lowering of the serum levels of any prescribed neuroleptic drugs, particularly haloperidol, due to the liver enzyme induction by carbamazepine. Reports of deterioration of the psychotic state after the administration of carbamazepine may reflect this pharmacokinetic interaction rather than a pharmacodynamic effect. Several investigators have noted exacerbations of psychosis on withdrawal of carbamazepine, which may reflect the therapeutic benefit of this drug, and is in keeping with the carbamazepine withdrawal syndromes reported in epileptic patients (31).
Carbamazepine and Aggressive Disorders
In some of the studies quoted previously, one of the behavioral syndromes that seems to respond to carbamazepine is dysphoric, aggressive behavior. A behavioral syndrome that most often is reported to respond to carbamazepine is episodic dyscontrol. This disorder has an interesting history, with earlier clinical descriptions coming from Munro et al. (32). They examined patients who previously had been investigated by Robert Heath. He monitored the activity of neurons in the limbic system using implanted electrodes, and demonstrated that patterns of sudden onset of aggression correlated with aberrant electrical activity in these areas, particularly in the hippocampus. These episodes tended to occur in patients with probable neurologic diatheses, but similar changes were reported in patients with and without epilepsy. These patients were differentiated from patients with life-long impulsive aggression, usually associated with personality disorders. In episodic dyscontrol, the episodes of aggression seemed to be uncharacteristic; they were of rapid onset after trivial provocation and they subsided rapidly, with the patients often feeling remorse afterward. Electroencephalographic (EEG) abnormalities often were noted on surface electrodes, and a history of brain trauma sometimes was present.
This disorder, referred to earlier as episodic dyscontrol, reemerges in DSM-IV as intermittent explosive disorder. Over time, it has been thought to be related to epilepsy, although the clinical and neurophysiologic evidence does not support that. Nonetheless, early trials of anticonvulsants suggested they may be of value in the treatment of episodic dyscontrol, although no double-blind, controlled studies exist.
Typical of the case reports is that of Stone et al. (33) of a man in whom explosive, aggressive feelings with suicidal ideas and explosive anger suddenly developed. An EEG was carried out that showed bursts of 6-per-second spike-and-wave complexes during hyperventilation, and bursts of 6-and 12-per-second rhythmic sharp waves in the mid-temporal areas. On carbamazepine his symptoms evaporated.
Other situations where carbamazepine has been used are in the management of aggressive disorders associated with the borderline personality disorder, and for the management of drug withdrawal syndromes.
With regard to the former, Cowdry and Gardner (34) compared four psychotropic agents in patients with borderline personality disorders and monitored their behavior, particularly aggressive behavior. The drugs used were alprazolam, carbamazepine, trifluoperazine, and tranylcypromine. In terms of the overall clinical effects, the most interesting finding was that patients given alprazolam showed a deterioration of their aggressive behavior, and those on carbamazepine showed a highly statistically significant improvement.
In alcoholism, the main benefit of carbamazepine in withdrawal states is that it is both psychotropic and anticonvulsant, and may therefore prevent the onset of withdrawal seizures as well as preventing craving (35). A further advantage of carbamazepine over the use of benzodiazepines is that there is less likely to be a withdrawal syndrome on stopping the drug, and much less abuse potential.
Although in some of the aforementioned studies, the presence of EEG abnormalities in patients may suggest a superior response to an anticonvulsant drug such as carbamazepine, the data do not really confirm that.
THEORETICAL BASIS FOR THE USE OF ANTIEPILEPTIC DRUGS IN PSYCHIATRIC DISORDERS
There are clear overlaps in the basic neurochemistry of epilepsy and a number of psychiatric syndromes, in particular revolving around neurotransmitters such as γ-aminobutyric acid (GABA), glutamate, and serotonin. Low levels of serotonin and low serotonin turnover rates lend to depressive, aggressive, and impulsive behavior, and also are proconvulsant (36). GABA agonism is clearly anticonvulsant, but may be associated with the development of psychiatric syndromes, particularly in epilepsy (37). Dopamine is known to be both anticonvulsant and propsychotic, and there is good evidence for an involvement of dopamine in both mood disorders and psychoses.
Carbamazepine is structurally related to the tricyclic antidepressants, and its mode of action is different from some other anticonvulsants such as phenytoin and phenobarbital. Its biochemical actions, apart from its effect at sodium channels, include partial agonism at adenosine receptors, an increase in the firing rate of the locus ceruleus, and a decrease in dopamine turnover (38). Furthermore, it increases the turnover of serotonin, which may imply some influence over affective tone.
In animal models, carbamazepine has been shown to be more effective than other standard antiepileptics in inhibiting the seizures developed from amygdylar kindling (39),
suggesting some limbic system selectivity for the drug. Indeed, it is the model of kindling that has been most put forward to explain the association between severe psychiatric disorders and epilepsy, as well as the effect of anticonvulsant agents on these syndromes. This has been reviewed on many occasions (38). In summary, the suggestion is that a kindling-like process (probably more a sensitization) leads to the development of recurrent manic depressive episodes, which are seen (as with some seizure disorders) to recur more frequently and at shorter intervals as time progresses. The episodes become triggered by less meaningful environmental events, and eventually take on a spontaneity. Such models also have been used to explain the development of other psychopathologic processes such as the addictive behaviors and the symptoms of both alcohol withdrawal and cocaine abuse (40).
Although there is no evidence that kindling per se (which is an animal model for the development of seizures) occurs in patients with psychiatric disorders, the concept of sensitization, the CNS mechanisms that may underlie this, and the effect of certain drugs such as the antiepileptics on biochemical processes related to sensitization and kindling, are being actively explored.
In this chapter, the evidence for the use of carbamazepine in nonepileptic psychiatric disorders is reviewed. Much of the evidence stems from studies conducted over a decade ago, and, since the value of its use in bipolar disorder was noted by the Japanese in the 1960s, carbamazepine has found widespread use in psychiatric settings. It is now approved in over 100 countries for use in affective disorders, and in the United States is the only drug approved by the Food and Drug Administration for the long-term prophylaxis of bipolar disorder, apart from lithium.
It also has found use in a spectrum of other clinical problems. It has been used mainly in bipolar affective disorders, but also has been prescribed for conditions as diverse as the aggressive, dysphoric, and excited behavior of schizophrenia, through episodic dyscontrol, through effects on craving and management of alcohol withdrawal. In the latter conditions, few or no clinical trials have been carried out, and none now are likely to be performed. In spite of this, its use, particularly for the management of aggressive disorders, seems well established, and indeed in a number of countries it remains one of the first-line drugs for treatment of alcohol withdrawal. Its use in bipolar disorder still is being explored because it has become clear that, as with lithium, there are a number of patients who simply do not respond to the drug, either because of side effects, or because it is ineffective. Nonetheless, the introduction of the use of carbamazepine into clinical practice for the management of psychiatric disorders has led to a considerable interest in the use of other anticonvulsant drugs in psychiatry. A number of these are being investigated, either in acute mania or more for the treatment of chronic bipolar disorder. Although the termantiepileptic has been used throughout this chapter for this group of drugs, the fact that carbamazepine is used widely outside epilepsy (and is used in other neurologic conditions) emphasizes the value of retaining the term anticonvulsant.