Gregory L. Holmes MD
Professor of Neurology, Department of Neurology, Harvard Medical School; and Director, Center for Research in Pediatric Epilepsy, Children's Hospital, Boston, Massachusetts
Carbamazepine (CBZ) is one of the major antiepileptic drugs (AEDs) used in the treatment of epilepsy in children and adults. One of the advantages of CBZ is the excellent side effect profile; CBZ rarely produces adverse cosmetic effects such as gingival hypertrophy, does not interfere significantly with cognitive function, attention, or behavior, and rarely is associated with life-threatening events.
However, as with all currently available AEDs, CBZ is associated with some adverse events. Most of the side effects are mild and dose related, although severe idiosyncratic adverse events can occur. Although a large number of idiosyncratic adverse effects have been attributed to CBZ therapy, it often is very difficult, in the myriad of anecdotal cases reported, to establish CBZ as conclusively responsible for the side effects. In the case of most severe adverse side effects, the patient is taken off the medication, and if the adverse event subsides, the clinician concludes that CBZ is the culprit. Unless a rechallenge with CBZ is attempted, it is very difficult to prove decisively that CBZ is the responsible agent for the adverse event. However, because with some adverse events rechallenge would be considered dangerous, the relationship between CBZ and the adverse event may remain cloudy.
INCIDENCE OF SIDE EFFECTS
All AEDs have side effects; in a study of 355 primarily adult patients receiving chronic AED therapy, 42% had 1 or more adverse reactions during the study period (1). When the individual AEDs were compared in patients on monotherapy, 15.5% of patients receiving CBZ had side effects at the first visit. This percentage should be compared with a 31% overall rate of side effects with AEDs. Many CBZ-related side effects improve with time; only 7% of patients had side effects at the time of their last visit.
When questioned about possible side effects, many patients note some; however, in most cases these are mild and not of sufficient magnitude to warrant discontinuation of the drug. In a study evaluating side effects of AEDs in children, Herranz et al. (2) found that 43% of 35 children taking CBZ had side effects at some point; however, in only 3% did the side effects lead to withdrawal of the drug. CBZ was the best tolerated of the AEDs evaluated. Pellock (3) found that 70% of children treated with CBZ as monotherapy or as combination therapy had side effects. However, other authors have found a lower rate of side effects in children (4,5). Okuno et al. (4) found that only 20 of 90 (22%) children treated with CBZ had side effects, most of which were mild and transient.
NEUROLOGIC ADVERSE EFFECTS
As with most AEDs, most of the side effects with CBZ involve the central nervous system. Nausea, headache, dizziness, incoordination, vertigo, tiredness, and diplopia are the most frequent symptoms (6, 7, 8, 9, 10, 11, 12). Nystagmus, tremor, and ataxia may be detected with a neurologic examination. These side effects usually are dose related and reversible, and they appear to be more common in the elderly (13). There are significantly fewer side effects when the patient is on monotherapy than when he or she is on polytherapy (11).
Neurologic side effects in children are similar to those in adults (2,5). In a study of 35 children receiving CBZ monotherapy, the most common side effect, drowsiness, occurred in 11% (2). Drowsiness also was the most common side effect, present in 43% of the patients, in a review by Pellock (3). Other commonly reported neurologic side effects in the Pellock series included incoordination in 20% and vertigo in 6%. Although diplopia was reported in only one child by Herranz et al. (2), this side effect likely goes unrecognized in nonverbal children. Other authors have reported a high incidence of this side effect in both children and adults (7,11). It is known that CBZ prolongs the duration of saccadic eye movements (14).
A high incidence of drowsiness and other adverse effects occurs when the drug is first started, especially if therapy is initiated at full dosage (10,15). Although tolerance to some of the side effects occurs rapidly (16), when possible, the dosage should be slowly increased because the rate of change of CBZ escalation is an important determinant of cognitive and motor dysfunction (17).
The short half-life of CBZ results in considerable fluctuation of CBZ levels, even when the drug is given multiple times daily. Many of the neurologic adverse effects occur at peak plasma levels of the drug (18, 19, 20). When peak blood CBZ levels are below 8 µg/mL, neurologic side effects are unusual, whereas levels greater than 12 µg/mL are associated with a higher risk of adverse effects (21). Some authors have reported that the unbound CBZ levels correlate better with neurotoxicity than total CBZ levels in serum (21,22). Because side effects of CBZ tend to occur at the time of a rapid peak concentration, use of sustained- or controlled-release CBZ can significantly reduce the incidence of side effects (16,23, 24,25).
CBZ is metabolized through oxidation to the active metabolite carbamazepine-10,11-epoxide (CBZ epoxide) (26). The epoxide has been implicated as the responsible agent for side effects (11,27), with several studies demonstrating a correlation between plasma CBZ epoxide concentration and side effects (11,27,28). However, Tomson et al. (29) crossed patients over from CBZ to CBZ epoxide. Despite high epoxide levels, no patients reported increased side effects. In fact, the authors noted a trend toward improvement when patients were switched to the epoxide therapy. Likewise, Theodore et al. (30) did not find epoxide levels to correlate with side effects. Riva et al. (22) found that the correlation of epoxide levels with the presence of nystagmus was no better than that with CBZ serum levels alone.
In most patients, neurologic side effects are easily managed (3). Because some side effects occur at peak concentrations, reduction of peak levels and associated toxicity may be achieved by giving the same daily dosage at more frequent intervals (29). Tomson (29) found that CBZ levels were 79% ± 29% higher than trough levels on a twice-daily dosage schedule and 40% ± 13% higher during four-times-daily administration. The appearance and intensity of side effects were related to the fluctuations in CBZ levels and were thus substantially reduced during the four-times-daily regimen.
Controlled-release CBZ reduces fluctuations in CBZ levels (24,31,32). Although the controlled-release preparation is likely to reduce some of the side effects seen at peak concentration (25), many patients tolerate the daily fluctuations of CBZ well. For example, Pieters et al. (24) did not find that changing from conventional CBZ to controlled-release CBZ altered attention or vigilance in children.
Abnormal involuntary movements occasionally have been reported in connection with CBZ treatment (33, 34, 35, 36). These movements have included tics (37,38), asterixis (39), and dystonia (34,40). Movement disorders typically develop in patients on AED polytherapy (34,35), in patients with brain damage (34), or in patients with toxic plasma levels of CBZ (33).
Peripheral Nerves and Muscle
Rarely, AEDs may lead to electromyographic signs of peripheral neuropathy (41). Lühdorf et al. (42) performed electromyography and nerve conduction velocity studies in 12 patients before and 3 months after CBZ treatment. No interval changes were noted. Likewise, Danner et al. (43) found no changes in nerve conduction velocity or the electromyogram in patients receiving CBZ. CBZ does not appear to cause a myopathy or result in weakness (14).
Exacerbation of Seizures
CBZ has been increasingly recognized to exacerbate or precipitate seizures in some patients. Initially described by Snead and Hosey in 1985 (44), the observation that CBZ can precipitate a new type of seizure or increase seizure frequency has been noted by other authors (45, 46, 47). Snead and Hosey (44) described 15 children with seizures in whom one or more seizure type was exacerbated during treatment with CBZ. The most common seizure type exacerbated was generalized atypical absences (11 patients). Four patients had an increase in generalized tonic-clonic or tonic seizures. The authors reported that a bilaterally synchronous spike-and-wave discharge of 2.5 to 3 cycles per second was predictive of increased atypical absence seizures with CBZ, whereas generalized bursts of spikes and slow waves of 1 to 2 cycles per second placed the patient at risk for increased tonic-clonic seizures. CBZ also may exacerbate epileptiform activity on the electroencephalogram (EEG) (48).
The incidence of rash with CBZ varies from 2% to 17% in various series (3,8,12,21,49,50). In most patients the rash is mild and not associated with any other systemic signs. The rash may disappear even when the drug is continued. A variety of cutaneous manifestations may be seen, of which the maculopapular, morbilliform, and urticarial types are the most common. If the rash persists, it is recommended that the drug be discontinued because serious and potentially life-threatening skin reactions (i.e., exfoliative dermatitis, Steven-Johnson syndrome, and Lyell's syndrome), although rare, do occur (3,50,51). CBZ-induced hair loss has been reported (52).
Murphy et al. (53) treated 20 patients with CBZ drug rashes with prednisone and antihistamines. Sixteen patients
were able to continue CBZ, whereas four had to discontinue the drug.
Other Allergic Reactions
In addition to the rashes seen with CBZ, more extensive hypersensitivity reactions with systemic signs such as fever, skin rashes, hepatosplenomegaly, and lymphadenopathy occasionally may occur (54, 55, 56, 57, 58). A variety of various other organ systems may be involved, and vasculitis (55,59), myocarditis (60), interstitial pneumonia (61, 62, 63, 64,65), membranous glomerulopathy (66), pseudolymphoma (58,67), and tubulointerstitial nephritis (68) have been implicated in these CBZ-induced hypersensitivity reactions.
The exact etiology of these multiorgan hypersensitivity reactions is unknown. There is some evidence that they may be type III or IV hypersensitivity reactions (51). Deposition of immune complexes, with CBZ or its metabolites acting as the antigenic stimulus, may be responsible for the multiorgan involvement (54,59). Although corticosteroids and immunosuppressive therapy may be needed (60), discontinuation of CBZ typically causes the symptoms to disappear.
The diagnosis of drug hypersensitivity to CBZ can be established by either in vivo reexposure or in vivo stimulation tests (51). These tests should be performed if there is any doubt about the diagnosis, or if continued treatment with CBZ is crucial.
SYSTEMIC LUPUS ERYTHEMATOSUS
Systemic lupus erythematosus (SLE) may be induced by CBZ (69, 70, 71, 72). As with other rare side effects, it often is difficult to form a definite connection between CBZ and the development of SLE. In contrast with many other skin reactions, the symptoms of SLE usually appear 6 to 12 months after the initiation of CBZ therapy (69,71). Discontinuation of the drug may lead to a gradual improvement and eventual disappearance of the symptoms (72). A positive antinuclear factor titer may persist after the drug is discontinued and clinical symptoms resolve (69,71).
HEPATIC AND GASTROINTESTINAL SIDE EFFECTS
Hepatic enzyme induction is a well known effect of treatment with CBZ, and asymptomatic elevation of liver enzymes commonly is observed during the course of CBZ therapy, occurring in 5% to 10% of patients (3). In one series comprising more than 200 children, 6% had elevation of liver enzymes (3). The modest rise in liver enzymes was of no clinical significance (3). Studies comparing phenobarbital, phenytoin, and CBZ indicate similar enzyme-inducing potentials of these drugs, as measured by antipyrine clearance and urinary excretion of D-glucaric acid (73,74).
Although rare, more severe CBZ-associated hepatotoxicity has been reported (56,75, 76, 77, 78, 79, 80, 81, 82, 83). CBZ-induced hepatotoxicity appears to have two different forms (79,80). The first is a hypersensitivity-induced granulomatous hepatitis with cholestasis (78,79,84). Levy et al. (78) reported three patients with CBZ-induced granulomatous hepatitis. All patients had onset of symptoms approximately 3 to 4 weeks after starting the drug, fever and laboratory evidence of hepatic dysfunction, and granulomatous infiltration on biopsy. All patients improved after discontinuation of the drug. Furthermore, in one patient fever and rash developed on rechallenge with one dose of CBZ. Mitchell et al. (84) described an additional two patients in whom hepatitis developed within 1 month of starting CBZ. Symptoms and signs resolved after discontinuation of the drug.
The second form of hepatotoxicity is a direct toxic effect of CBZ (or its metabolites) on the liver that results in acute hepatitis and hepatocellular necrosis without significant cholestasis (79,80). The hepatitis due to CBZ exposure also is a hypersensitivity reaction, presumably mediated by an immunologic mechanism and typically occurring in the setting of a generalized hypersensitivity response (75).
Most patients who experience hepatic dysfunction have taken CBZ for less than 1 month and have associated fever or rash (78,84). Rechallenge with the drug has resulted in recurrence of the symptoms (78,85). Even with discontinuation of the drug, fatalities can occur (75,86). It has been suggested that the CBZ epoxide metabolite may be the hepatotoxic substance because there has been no correlation between toxicity and CBZ serum concentrations (80,81). Acute hepatitis appears to be rare in children (81,87).
Common gastrointestinal side effects of CBZ consist of vomiting, nausea, and diarrhea (2,88,89). Herranz et al. (2), in a study of 35 children treated with CBZ monotherapy, found that 14% had gastrointestinal disturbances. Pellock (3), in a review of adverse effects in 220 children treated with CBZ as monotherapy or polytherapy, reported that 9% had gastrointestinal side effects. Although pancreatitis has developed in a few patients treated with CBZ (90,91), it is not currently possible to definitively determine a causal link.
Additional gastrointestinal adverse effects reported with CBZ include acute cholangitis (92), eosinophilic colitis (88), constipation (93), and inhibition of intestinal folic acid absorption (94).
Hematologic abnormalities are probably the most feared idiosyncratic reactions associated with CBZ. The package insert for Tegretol (Novartis Pharmaceutical Inc., Summit,
Transient leukopenia commonly occurs in approximately 10% to 20% of patients treated with CBZ (6,51, 97,98), whereas persistent leukopenia has been reported in approximately 2% of patients treated with CBZ (95). Livingston et al. (7) reported moderate leukopenia [3,000 to 3,500 white blood cells (WBCs)/mm3] in 9 of 225 patients and marked leukopenia (2,000 to 3,000 WBCs/mm3) in 4 of 225 patients, usually within the first 3 months of treatment. Despite continuation of CBZ, cell counts returned to normal and in no case did isolated leukopenia precede more widespread marrow suppression.
Some authors have reported that younger patients have a lower prevalence of transient or persistent leukopenia than do older patients (10,15,99). However, in a study of 200 children taking CBZ, leukopenia, defined as a WBC count of less than 4,000/mm3, occurred in 17% of patients younger than 12 years of age and in 8% of children 12 to 17 years of age (100). In a study of 220 children receiving CBZ therapy, Pellock (3) found that 28 (13%) had leukopenia with WBC counts below 4,000/mm3, but only 5 (2%) had counts below 3,000/mm3. None of the leukopenias progressed, and all of the patients remained on CBZ. Most of the children had low WBC counts before starting CBZ, viral infections, or both.
Leukopenia does not appear to be dose related. Transient leukopenia is more common in patients with low WBC counts before treatment (98). The proportion of T lymphocytes and B lymphocytes apparently is not altered by CBZ (97).
In rare cases, CBZ may cause serious hematologic toxicity (i.e., aplastic anemia, persistent leukopenia, and thrombocytopenia). Although CBZ-associated aplastic anemia is rare, the mortality rate is high (50,95,96,101). Hart and Easton (95), based on evidence available in 1982, estimated the prevalence of aplastic anemia to be < 1/50,000. Pellock (3) estimated the risk to be smaller: 5.1 per million for aplastic anemia, 1.4 for agranulocytosis, and 2.2 for deaths associated with these events. A precise estimation of the risk for development of aplastic anemia during CBZ treatment is difficult because patients often are on other medications or have other risk factors for aplastic anemia. Neither the duration of treatment nor the age of the patient appear to be major factors in the development of aplastic anemia. It is uncertain whether aplastic anemia is dose related.
Rarely, isolated thrombocytopenia can occur during treatment with CBZ (9,102, 103, 104). As with the WBC count, there may be a transient decrease in the platelet count at the beginning of CBZ therapy (9,95).
There is a debate as to how frequent complete blood cell counts should be obtained in patients on CBZ (21). Dodson (21) recommends that the WBC count be measured at 6 weeks, 3 months, and 6 months. If the leukocyte count is consistently above 3,500/mm3 and the granulocyte count is above 1,200/mm3, Dodson recommends that the frequency of measurements can be reduced. If the neutrophil count is below 1,200/mm3, the frequency of the observations should be increased. If the neutrophil count is below 900/mm3, it is recommended that the CBZ dosage be reduced. However, other clinicians do not perform routing hematologic monitoring, obtaining studies only when the patient has symptoms of a blood dyscrasia. It does not appear that routine monitoring of the complete blood cell and platelet counts allows early detection and discontinuation of drug before the process becomes irreversible (12,21,105).
If both anemia and neutropenia occur, obtaining a reticulocyte count, serum iron concentration, and iron-binding capacity provides an indirect indication of hematopoietic activity (21). The combination of a normal or elevated reticulocyte count with a normal or low iron level is reassuring that the bone marrow is active. A low reticulocyte count with an increased serum iron concentration indicates that hematopoietic activity is reduced. In this case, CBZ should be stopped.
Antidiuretic Hormone and Water Retention
Hyponatremia and water retention are well known side effects of CBZ (106, 107, 108, 109, 110, 111). The risk increases with age and with serum level of the drug (112, 113, 114,115). Hyponatremia secondary to CBZ is more common among the elderly (46,116), and is quite unusual in children (113,117). Hyponatremia appears to be related to serum drug level: patients whose CBZ levels are ≥6 µg/mL have a 3.5-fold greater prevalence than do those with lower levels (113). Hyponatremia is more frequent in patients on monotherapy, suggesting that other AEDs may prevent the antidiuresis induced by CBZ (114).
There may be several mechanisms by which CBZ has an antidiuretic effect. CBZ has been reported to have both a direct antidiuretic hormone (ADH) effect or ADH-promoting activity and ADH-releasing effect (51,118,119). Evidence favoring a renal effect of CBZ arises from the finding that the antidiuretic effect may be reversed by demeclocycline, a drug used to treat the syndrome of inappropriate ADH through a nephrogenic, dose-dependent inhibition of ADH-sensitive adenylate cyclase activity (119,120).
Evidence supporting a direct hypothalamic effect comes from studies demonstrating that hyponatremia may be normalized
by concomitant treatment with phenytoin (115), an AED known to alter water balance by inhibiting ADH release (121). Some investigators have observed increased arginine vasopressin levels (a measure of ADH activity) during CBZ treatment (118,122), suggesting a hypothalamic effect. However, Stephens et al. (123), in a study of normal volunteers, measured ADH by radioimmunoassay and demonstrated a decrease, not an increase, in ADH levels.
Because several of the symptoms of hyponatremia—dizziness, headache, drowsiness, and nausea—may mimic side effects of CBZ, hyponatremia should be considered when these symptoms occur (51,109). Excessive intake of fluids should be discouraged in patients on CBZ. Finally, caution is in order when elderly patients on a low-salt diet are treated with CBZ.
Thyroid function test results frequently are abnormal in patients taking AEDs, including CBZ (124, 125, 126, 127, 128, 129, 130, 131, 132, 133). In one of the largest studies, Isojärvi et al. (128) measured circulating thyroid hormones, as well as the pituitary function, in 63 male patients with epilepsy receiving either CBZ, phenytoin, or valproate as monotherapy or a combination of CBZ plus phenytoin or CBZ plus valproate. Patients on CBZ, phenytoin, CBZ plus phenytoin, and CBZ plus valproate had lower levels of circulating thyroxine (T4) and free thyroxine (FT4) than control patients. Serum T4 and FT4 concentrations were unaffected by valproate monotherapy. Triiodothyronine (T3) levels were normal in all groups studied. Despite low T4 and FT4 serum levels, the serum thyrotropin [thyroid-stimulating hormone (TSH)] concentration was slightly elevated only in patients treated with a combination of CBZ plus valproate. The TSH responses to thyrotropin-releasing hormone (TRH) after 20 minutes were slightly lower in the patient group receiving phenytoin monotherapy than in the untreated patients, whereas other TRH responses to TSH were unchanged in the other patient groups.
The decrease in serum T4 usually is small to moderate. Rootwelt et al. (131) found that in patients starting on CBZ, serum levels of T4 fell to a stable 70% of the basal level after 1 to 2 weeks, whereas T3 decreased transitorily to 85% of the basal level. TSH showed a complementary, but somewhat delayed transitory increase.
CBZ activates liver microsomal enzymes, thereby increasing peripheral metabolism of thyroid hormones (134). A CBZ-induced conversion and metabolism of T4 to T3 and of T3 itself could explain the changes in thyroid hormone levels during CBZ therapy. In addition, thyroid hormones and CBZ bind competitively to thyroxine-binding globulin (131). Competitive binding to TBG by CBZ results in a reduction of protein-bound T4 and an increase in FT4 (124).
Despite changes in serum levels of T4, patients usually remain clinically euthyroid (131). This observation could be explained by an increased peripheral use of, or sensitivity to, T4. Overt hypothyroidism has been described in only two patients, one treated with CBZ monotherapy and one receiving a combination of CBZ and phenytoin (134). The significance, if any, of low serum T4 and FT4 levels in asymptomatic patients is uncertain. In patients with normal TRH levels and no clinical signs of hypothyroidism replacement, therapy does not appear to be indicated.
Because of the increased peripheral metabolism of thyroid hormones with CBZ, treating T4-supplemented hypothyroid patients with CBZ may require an increase in the dose of T4 to maintain the euthyroid state (135).
Adrenal Cortical Function
CBZ can cause an increase in free cortisol levels (136). In spite of this, cushingoid symptoms have not been observed during treatment with the drug. The significance, if any, of this observation therefore is unclear.
The effect of CBZ and other AEDs on sex hormones has been of considerable interest to many investigators because reproductive dysfunction and hyposexuality frequently are reported in patients with epilepsy (137, 138, 139). As noted by Herzog and Levesque (140) and Isojärvi (141), the cause of these complaints likely is multifactorial, with psychosocial, AED, seizure-related, and hormonal factors. A number of investigators have measured sex hormones in patients on AEDs to determine if changes in sexual function are related to hormonal changes.
Testosterone exists in the serum in three forms: free, albumin bound, and sex hormone-binding globulin (SHBG) bound (140). The SHBG-bound fraction is not biologically active, whereas the free testosterone and albumin-bound testosterone are available to tissues. Because of CBZ-induced hepatic synthesis of SHBG, serum levels of SHBG increase during therapy with CBZ treatment (128,142). However, serum total testosterone concentration does not change or increase, the free androgen index (one indicator of the non-SHBG-bound portion of testosterone) decreases, and the serum free testosterone level remains unchanged or decreases (128,142, 143, 144). The low free testosterone found in some patients may be secondary to increased metabolism (145).
The significance of changes in hormonal levels with CBZ is uncertain. Although Leiderman et al. (146) did not find a correlation between sexual desire and total serum testosterone concentration in men with temporal lobe epilepsy, Toone et al. (147) found that hyposexuality correlated with lower levels of free, rather than of total, testosterone.
Isojärvi et al. (128) did not find serum free testosterone levels or the free androgen index to correlate with sexual dysfunction in their patients. In contrast, Herzog et al. (140,148), found that patients with reproductive or sexual dysfunction had a higher likelihood of abnormally low non-SHBG-bound testosterone than did men with epilepsy without sexual dysfunction.
Because AEDs affect the levels of other hormones, including estradiol, gonadotropin, and thyroid hormone, the relationship between AEDs and sexual function is complex. Seizuresper se, particularly those involving the limbic system, can lead to reproductive dysfunction and hyposexuality, making the assessment of the hormonal status in patients with epilepsy even more complicated (140).
Reports concerning the serum levels of prolactin (PRL) and gonadotropin in patients with epilepsy receiving AEDs are controversial: Some authors have reported elevated levels of these hormones, whereas others have observed unaltered levels (142). Franceschi et al. (149) reported elevated PRL levels in patients with epilepsy receiving CBZ as single-drug therapy. They found no alteration in the basal levels of gonadotropin. Dana-Haeri et al. (150) did not find any changes in basal PRL or follicle-stimulating hormone (FSH) levels in male patients with epilepsy treated with CBZ, but found a significant rise in the mean basal level of luteinizing hormone (LH). Bonuccelli et al. (151) studied the spontaneous secretion of PRL in a small group of male patients with epilepsy receiving CBZ and found no change in PRL. Furthermore, 21-day treatment with CBZ did not cause any change in LH levels in healthy volunteers (145). Isojärvi et al. (142) found no difference in the mean gonadotropin or PRL levels in CBZ-treated patients and control subjects. It is not clear whether alteration of sex hormones plays a role in impotence among men with epilepsy.
To study the effects of CBZ on pituitary responsiveness, Isojärvi et al. (152) studied the effects of LH-releasing hormone, TRH, and metoclopramide on FSH, LH, and PRL in patients taking CBZ. The mean basal concentration of serum LH was significantly lower in the CBZ-treated female patients than in untreated women. In addition, the mean basal concentration of PRL was lower and the response of PRL to TRH was higher in male patients treated with CBZ. These results demonstrate that CBZ has an effect on pituitary responsiveness. The clinical relevance of these findings is not clear.
Effect of Carbamazepine on Oral Contraceptives
Because of the liver enzyme-inducing potential of CBZ, the drug may augment the degradation of estrogen and progesterone. Consequently, the effect of oral contraceptives may be impaired (153). Breakthrough bleeding and a number of unintended pregnancies have occurred in patients treated with CBZ.
Although biochemical changes in the metabolism of vitamin D are observed during treatment with CBZ (154), whether clinically apparent osteomalacia develops during treatment with the drug is controversial (155, 156, 157). In a study of 21 patients with epilepsy receiving CBZ as the only AED, hypocalcemia was detected in 3, hypophosphatemia in 1, and elevated serum alkaline phosphatase in 4 patients (155). Serum 25-hydroxyvitamin D values were significantly lower in the patients than in the control subjects. Although there was no difference in bone mineral density between the patients and control subjects, an increased amount of trabecular resorption surfaces was found in patients treated with CBZ, leading the authors to conclude that CBZ therapy may lead to vitamin D deficiency and osteomalacic skeletal changes. O'Hare et al. (154) also reported hypocalcemia in 10% of patients on CBZ monotherapy. Serum calcium was significantly lower and alkaline phosphatase significantly higher in patients than in matched control subjects. None of the patients was symptomatic. The authors postulated that calcium abnormalities were secondary to induction of hepatic microsomal enzymes.
Although rare, CBZ has been implicated in a number of cardiovascular abnormalities, including congestive heart failure (158), aggravation of the sick sinus syndrome (159), hypertension (6), and conduction defects resulting in bradycardia or Stokes-Adams attacks (107,160, 161, 162, 163, 164). Cardiac side effects, which seem to be especially likely to develop in older patients (in whom the incidence of conduction disturbances is highest), may develop after years of CBZ therapy. Rechallenge with CBZ after insertion of a pacemaker has supported a causal relationship between the CBZ and conduction disturbances (162). The conduction defect may be dose related.
Kasarskis et al. (165), in a review of 40 previous cases of CBZ-induced cardiac dysfunction, suggested that two distinct forms of CBZ-associated cardiac dysfunction occur. In the first form, sinus tachycardia occurs in the setting of a massive CBZ overdose. The 14 patients with this form were characterized by a young age (mean, 25.2 ± 12.6 years; range, 1.5 to 55 years), markedly elevated serum CBZ levels (mean, 227 µmol/L; range, 97.4 to 1,012 µmol/L), the absence of preexisting cardiac dysfunction, and an equal sex distribution. The second group consisted almost exclusively of elderly women who had potentially life-threatening bradyarrhythmias or atrioventricular conduction delay, associated with either therapeutic or modestly elevated serum CBZ levels. The mean age in the second group was 65.0 ± 18.4 years (range, 10 to 87 years). In the 26 patients
with the second form, cardiac dysfunction appeared without warning during routine clinical management of either seizures or trigeminal neuralgia.
In healthy, young individuals, the risk of CBZ-induced cardiac dysfunction appears low (166). Patients with underlying cardiac disease may be at higher risk for conduction defects. Although heart block rarely is reported in children, patients with tuberous sclerosis and cardiac rhabdomyoma may be at increased risk for conduction block (167). Durelli et al. (45) found that in a patient with a myotonic cardiomyopathy, CBZ treatment resulted in a grade 1 heart block that appeared to increase in severity with increasing dosage of the drug.
These clinical observations are supported by experimental work with dogs (168). Parenteral CBZ results in a prolongation of atrioventricular conduction and a decreased rate of phase 4 depolarization of autonomic fibers. It appears that CBZ at low or therapeutic doses can worsen a preexisting atrioventricular block, and that the drug at higher levels can induce conduction delays de novo in otherwise normal elderly people (> 65 years) (165).
RENAL ADVERSE EFFECTS
Isolated renal side effects of CBZ therapy are rare. Cases of acute renal failure due to presumed CBZ-induced acute interstitial nephritis (68,169), and a case of membranous glomerulopathy presumably caused by a type III allergic reaction to CBZ (66) have been reported. CBZ also has been implicated in urinary retention (170).
Several studies indicate that CBZ may influence heme biosynthesis (171,172), and reports of CBZ-induced attacks of nonhereditary porphyria exist (173,174). Conversely, CBZ is recommended by some authors for use in patients with acute intermittent porphyria (175,176).
Levels of the immunoglobulin G (IgG) subclass IgG2 have been reported to be decreased in patients on CBZ (177). The significance, if any, of altered concentrations of IgG subclasses with CBZ is likely to be slight. There are no significant differences in concentrations of serum (178,179) or nasal immunoglobulins (178) or in frequency of respiratory tract infections between patients receiving CBZ and control subjects (178). These findings are in contrast to phenytoin, in which reduced concentrations of serum and nasal IgA have been reported (180).
A reversible retinopathy possibly secondary to CBZ was reported in two adult patients treated with CBZ for more than 7 years (181). Both patients had sudden reduction of visual acuity without other side effects. Lesions in the retinal epithelium were noted in both patients. Discontinuation of the drug led to improvement in visual function and in the morphologic changes. The tricyclic psychotropic drugs are known to have toxic effects on the retinal pigment epithelium. Because CBZ is chemically similar to the tricyclics, it is possible that the development of retinal lesions in these two patients is not coincidental. Color vision can be affected by AEDs. López et al. (182) found that color visual perception was impaired in 82% (30 of 37) of patients taking either CBZ, phenytoin, or valproate. In the CBZ group, abnormalities in the blue-yellow axis occurred in 66.7% (8 of 12) of the patients. None of these patients complained of disturbances in color vision.
CBZ may cause a variety of psychic disturbances, including asthenia, restlessness, insomnia, agitation, and anxiety (50). In addition, CBZ has been implicated in sporadic cases of psychosis (50). Overall, the incidence of psychiatric symptoms with CBZ is low and appears to be lower than with some of the other AEDs (183)
Two patients in whom signs of an encephalopathic process developed while on CBZ have been reported (184,185). The first patient (184) presented with confusion, spasticity, hyperreflexia, bilateral extensor plantar responses, and ataxia, whereas the second (185) became mute and manifested a spastic quadriparesis. Both patients improved dramatically when CBZ was discontinued.
In unpublished cases, the author has cared for three children with partial seizures in whom expressive aphasia developed over several months while on CBZ. EEGs showed bilateral frontotemporal lobe spikes that markedly increased during sleep. After withdrawal of CBZ, speech returned to baseline status and the number of epileptiform discharges during both the awake and sleep states decreased. It is suspected that CBZ, by increasing the epileptiform discharges, caused the aphasia.
The risks of cognitive impairment with nontoxic chronic CBZ appear low (16,23,186, 187, 188). Cognitive improvement may occur when patients are switched from barbiturates or polytherapy to CBZ (10). Most studies examining cognitive effects have used the immediate-release preparations of CBZ. It is likely that both children and adults will perform better with the sustained-release preparations.
Aman et al. (189) evaluated 50 children with well controlled seizures receiving CBZ monotherapy with a battery of cognitive and motor tasks. Testing was performed both after CBZ was administered, presumably at a time of peak concentration, and before CBZ administration, when CBZ concentrations were at trough levels. During testing at peak concentration, response time varied as a function of the test; response time was increased for matching familiar figures and for auditory-visual integration, but decreased for short-term memory tasks. In addition, at peak concentrations the children had fewer extraneous movements while seated; that is, they were less restless, had an improved attention span, and had steadier motor movements. Of note was the finding that concentration of CBZ in the saliva did not correlate with any variables measured.
O'Dougherty et al. (190) studied the relationship between serum CBZ concentrations and cognitive performance in 11 children with partial seizures. Neuropsychological tests demonstrated a mild beneficial effect of CBZ on eye-hand coordination and, with low serum levels, better rapid processing of items in memory. No changes in simple reaction time, sustained attention, behavioral adjustment, or motor performance with the preferred hand were observed at low or moderate CBZ levels. Efficiency of learning new information and memory scanning rate showed a level-dependent relationship with CBZ. Poor performances were significantly associated with higher plasma CBZ levels. As discussed by Trimble (191), the relationship between performance on cognitive tasks and serum levels of CBZ in children appears to vary as a function of the task.
As do children, adults function quite well when serum CBZ levels are within the therapeutic range (187). In a study of 22 adults (mean age, 36 years) studied with a battery of neuropsychological measures involving motor speed, reaction time, attention, and memory, there was no consistent relationship between results on the test battery and serum concentration of CBZ (192). Further support for the lack of significant cognitive impairment with CBZ comes from a large study on the effects of withdrawal of AEDs on cognitive function in children. Aldenkamp et al. (23) found that the impact of AED treatment on cognitive functions appears to be limited in magnitude.
Massive CBZ overdoses have been well described by numerous authors (79,193, 194, 195, 196, 197, 198, 199, 200, 201, 202, 203, 204, 205). Although CBZ overdosage is serious, death rarely occurs with a CBZ overdose (206,207). Fisher and Cysyk (207) reviewed 23 cases of CBZ overdosage reported in the literature. Only two deaths were recorded. The highest serum level report was 65 mg/L (200). A patient who consumed 400 tablets of 200 mg CBZ survived after hemoperfusion (201).
Overdoses of CBZ have been associated with symptoms resembling the central anticholinergic syndrome caused by tricyclic antidepressants (205,207). The most prevalent symptoms of CBZ overdosage are nystagmus, mydriasis, ophthalmoplegia, cerebellar and extrapyramidal signs, and impairment of consciousness progressing to a comatose state, possibly accompanied by seizures, myoclonus, and respiratory depression (208). Cardiac symptoms consist of tachycardia, arrhythmia, conduction disturbances, and low blood pressure. Gastrointestinal and anticholinergic symptoms also may supervene. Coma may develop at serum CBZ levels as low as 80 µmol/L (209,210). Life-threatening cardiac events are rare with CBZ overdose, although cardiorespiratory problems can occur with serum CBZ levels exceeding 35 µg/mL (208).
Weaver et al. (211) described four stages in CBZ overdosage: I—coma, seizures [CBZ levels >25 µg/mol (105 µmol/L)]; II—combativeness, hallucinations, choreiform movements [15 to 25 µg/mL (65 to 105 µmol/L)]; III—drowsiness, ataxia [11 to 15 µg/mL (45 to 65 µmol/L)]; and IV—potentially catastrophic relapse [11 µg/mL (45 µmol/L)].
During CBZ intoxication, there is a prolongation of the CBZ half-life and elevation of the CBZ epoxide/CBZ ratio (211). The concentration of CBZ epoxide may exceed that of the parent compound (195,211). It has been suggested that the evolution of the intoxication correlates more closely with the course of the CBZ epoxide level than with the concentration of CBZ itself, which declines rapidly (79,211).
Because of the anticholinergic-like action of the drug, delayed gastric emptying often is present, leading to impaired absorption of the drug. In addition, when ingested in large quantities, the drug may form a concretion. For that reason, gastric emptying may continue for as long as 12 hours or more in asymptomatic patients and up to 60 hours in symptomatic patients (208). Activated charcoal, metoclopramide, which enhances gut motility, and vigorous cathartics also have been recommended (204,208,212). Relapses can occur owing to delayed absorption of CBZ from the intestine (196,205,211). Because of the high degree of protein binding of CBZ, forced diuresis, peritoneal dialysis, and hemodialysis are not recommended (201,208). Charcoal hemoperfusion, a technique in which blood is pumped through a column containing activated carbon granules coated with acrylic hydrogen polymer, has been used successfully to treat patients with overdosage (194,195,198,199). Charcoal hemoperfusion is associated with thrombocytopenia, coagulopathy, hypothermia, and hypocalcemia. Recently, high-efficiency dialysis, which uses a highly permeable membrane and a high dialysate flow, has been used safely for a patient with a CBZ overdosage (193). This technique appears effective and safe.
It is quite likely that CBZ will remain one of the major AEDs as we enter the twenty-first century. The drug will likely remain popular not only because of its efficacy in controlling a variety of seizures but because of its excellent safety profile. Although CBZ may cause a number of side effects, most are mild and easily managed. Fortunately, severe adverse reactions are rather infrequent.
This work was supported in part by the National Institute of Neurological and Communicative Disorders (NS27984). The author acknowledges the contribution of Drs. Lennart Gram and Peder Klosterskov Jensen, who wrote the chapter on carbamazepine toxicity in the third edition of this book. Their contribution greatly aided the author in the preparation of this review.