Carbamazepine

General adverse effects and adverse reactions

Cerebellovestibular and oculomotor dysfunction (with dizziness, diplopia, nystagmus, ataxia), fatigue, and sedation (usually transient) are relatively common. Nausea, vomiting, cognitive dysfunction, headache, myoclonus (including non-epileptic myoclonus), exacerbation of seizures, movement disorders, behavioral or psychiatric disturbances, hyponatremia, and altered cardiac conduction are less common. Hypersusceptibility reactions to carbamazepine are relatively common and range from cutaneous reactions (including Stevens–Johnson syndrome and toxic epidermal necrolysis, severe forms of erythema multiforme) to systemic reactions with fever, lymphadenopathy, and/or involvement of the bone marrow, the liver, the heart, the gastrointestinal system, the lungs, and other organs.

Observational studies

In 10 children with chorea (eight girls and two boys; aged 7–16 years), nine with rheumatic fever, carbamazepine (4–10 mg/kg/day; plasma concentrations 12–34 μmol/l) produced improvement within 2–14 days [ ]. The chorea disappeared within 2–12 weeks. There were no adverse reactions.

Comparative studies

In a meta-analysis of responses to lamotrigine or carbamazepine monotherapy using data on individual patients from five randomized trials in 1384 patients, the time to treatment withdrawal (an outcome reflecting both seizure control and tolerability) was significantly better with lamotrigine than carbamazepine (HR = 0.55; 95% CI = 0.35, 0.84) [ ].

The cognitive effects of carbamazepine and topiramate have been compared in patients with epilepsy in an open, observer-blinded, parallel-group, randomized trial in children with benign rolandic epilepsy [ ]. Topiramate was introduced at a dose of 12.5 mg/day, with a minimum target dose of 50 mg/day over 4 weeks. Carbamazepine was started at a dose of 10 mg/kg/day, with a minimum target dose of 20 mg/kg/day. Changes in neuropsychological tests performed at baseline and after 28 weeks of treatment showed that topiramate had slightly worse effects than carbamazepine.

There were no differences in behavioral adverse effects between carbamazepine and phenobarbital in a randomized, controlled trial in 108 children with generalized tonic–clonic or partial and secondarily generalized seizures in Bangladesh [ ].

Placebo-controlled studies

The efficacy of carbamazepine in 89 subjects with unipolar depression who had never received antidepressants has been evaluated in a double-blind, randomized, placebo-controlled study [ ]. The patients had had at least two major depressive episodes but had never had mania or hypomania. They were randomized to immediate-release carbamazepine 300–800 mg/day (n = 51) or placebo (n = 38) for 12 weeks. There was mild leukopenia in 14/46 of those taking the active drug and 2/37 of those taking placebo. There were rashes in 11% of those taking carbamazepine and in none of those taking placebo. Nausea, headache, tremor, blurred vision, and somnolence almost exclusively occurred in those taking carbamazepine. Dizziness occurred in similar percentages (13% and 16% respectively).

Cardiovascular

Carbamazepine-induced dysrhythmias have two distinct forms: sinus tachycardia in the setting of massive carbamazepine overdose (mainly seen in young adults) and bradydysrhythmias or atrioventricular block, via depression of the cardiac conducting system [ ], which are almost exclusively seen in elderly women and at carbamazepine concentrations in the usual target range.

There have been a few reports of reversible atrioventricular block [ ], and asystole has been described in a patient with Guillain–Barré syndrome [ ].

Four cases of carbamazepine-induced sinus node dysfunction (n = 3) and atrioventricular block (n = 1) were described in elderly Japanese women taking 200–600 mg/day. In two of the three patients rechallenged, sinus arrest recurred within 48 hours [ ].

Three cases of carbamazepine-induced Stokes–Adams attacks caused by intermittent total atrioventricular block, sinoatrial block with functional escape rhythm, and intermittent asystole have been described; it was suggested that cardiac conduction should be assessed if syncope or changes in seizure type occur in patients taking carbamazepine [ ].

In 12 patients in whom carbamazepine was withdrawn, power-spectrum analysis of RR interval variability was used to investigate changes in sympathetic/parasympathetic autonomic equilibrium [ ]. Abrupt withdrawal of carbamazepine altered the sympathovagal balance during non-REM sleep, shifting the sympathovagal balance toward sympathetic predominance. However, analysis of the before and after withdrawal cardiac Holter recordings showed no serious cardiac dysrhythmias in any patient.

In one patient carbamazepine caused pacemaker failure until the pacemaker was adjusted, no doubt because of its effects on cardiac conduction [ ].

Carbamazepine-induced hypertension has been reported [ ].

• A 33-year-old man with complex partial seizures was switched from phenytoin to carbamazepine. His blood pressure was 118/70, but 4 months later, while he was taking carbamazepine 600 mg bd, his blood pressure was 150/112 and 1 month later 142/110. He was taking no medications beside carbamazepine. Secondary causes of hypertension were ruled out. Carbamazepine was withdrawn and gabapentin prescribed. One month later his blood pressure was normal, and it remained so at subsequent follow-up appointments over the next 2 years.

Congestive heart failure possibly caused by carbamazepine has been described [ ].

Potentially fatal eosinophilic myocarditis may be a manifestation of carbamazepine hypersensitivity [ ]. Carbamazepine overdose has also been associated with focal myocarditis (see below).

Respiratory

Acute hypersensitivity reactions involving the lung or the bronchi have been reported, but are rare [ , ].

Nervous system

Most central nervous system adverse reactions to carbamazepine are mild, transient, and reversible on dosage adjustment. Those reported more commonly (in 10–50% of patients, depending on dosage and assessment method) include somnolence, which is usually transient, diplopia, nystagmus, dizziness, fatigue, headache, ataxia, and cognitive dysfunction. These effects may be intermittent and related to high peak concentrations of the drug. Dividing the total daily dose or switching to a modified-release formulation can minimize these effects [ ]. Patients with MRI evidence of cerebellar atrophy may be more prone to nystagmus, dizziness, and ataxia when they take carbamazepine [ ].

Carbamazepine can precipitate or aggravate myoclonic, atonic, and absence seizures, especially in children and adolescents with a history of bilaterally synchronous spike-and-wave discharges in the electroencephalogram; both generalized and partial seizures can be worsened [ ]. In children, the new appearance of generalized paroxysmal discharges after treatment may be predictive of seizure exacerbation or suboptimal control [ ]. Tonic seizures are occasionally aggravated by carbamazepine [ ]. Non-epileptic myoclonus occurs rarely.

A mild reduction in motor and sensory conduction velocities, possibly related to folate deficiency, has been found in patients taking long-term treatment [ ], but symptoms of peripheral neuropathy are uncommon.

Asterixis, dystonias, and dyskinesias (including motor tics, orofacial and lingual dyskinesias, and oculogyric crises) are uncommon, as are auditory disturbances [ ].

Neuroleptic malignant syndrome occurred in a middle-aged man who had previously experienced the same reaction to tiotixene [ ]. A 54-year-old man who had been taking neuroleptic drugs for about 30 years developed neuroleptic malignant syndrome within 3 days of taking add-on carbamazepine (400 mg/day) [ ]. It was speculated that the pathogenesis could have involved rebound cholinergic activity after a reduction in plasma neuroleptic drug concentrations by carbamazepine.

Aseptic meningitis reversible on drug withdrawal has been described [ ].

Retrospective studies have suggested that antiepileptic drugs can be associated with peripheral nerve dysfunction. This has been prospectively studied in 81 patients (aged 13–67 years) without polyneuropathy who took sodium valproate (n = 44) or carbamazepine (n = 37) as monotherapy in standard daily doses [ ]. After 2 years one patient had clinical signs of polyneuropathy and six had symptoms of polyneuropathy, but electrophysiology did not show significant changes or trends. Only one patient had abnormal electrophysiological findings, which were only subclinical, and eight patients had abnormal values at two subsequent visits. There were no consistent patterns, and the data were unaffected when the drugs were examined separately or when patients were grouped according to whether or not they had symptoms of polyneuropathy. The authors concluded that previously untreated young to middle-aged patients who take valproic acid or carbamazepine for 2 years are not at risk of polyneuropathy.

Combined phonic and motor tics occurred in a 7-year-old boy with Down’s syndrome when he took carbamazepine 19 mg/kg for suspected focal epilepsy [ ]. Carbamazepine concentrations were within the usual target range. The symptoms resolved completely after withdrawal.

Carbamazepine can cause altered visual evoked potentials and brainstem evoked potentials [ ]. In 100 epileptic patients aged 8–18 years taking carbamazepine in a modified-release formulation, interpeak latencies of I–III and III–V of brainstem evoked potentials were significantly delayed and N75/P100 and P100/N145 amplitudes in the visual evoked potentials were reduced.

Relatively low doses of carbamazepine (300–600 mg/day) have been reported to cause serious worsening of disability in five patients with multiple sclerosis [ ]. The authors suggested that this effect was due to blockade of sodium channels by carbamazepine.

Inappropriate use of antiepileptic drugs can aggravate idiopathic generalized epilepsy syndromes. In a retrospective study in 14 adults with idiopathic generalized epilepsy taking at least one potentially aggravating antiepileptic drug, video-EEG showed typical absence status epilepticus in five, atypical absence status epilepticus in five, atypical myoclonic status epilepticus in three and typical myoclonic status epilepticus in one [ ]. Epilepsy had been misclassified as cryptogenic partial in eight cases and cryptogenic generalized in four. The correct diagnosis was juvenile absence epilepsy in six patients, juvenile myoclonic epilepsy in four, epilepsy with grand mal on awakening in two, and childhood absence epilepsy in two. All patients had been taking carbamazepine and had had seizure aggravation or new seizure types. Seven were taking polytherapy with phenytoin, vigabatrin, or gabapentin. Potential precipitating factors included increases in the dose of carbamazepine or carbamazepine + phenytoin; introduction of carbamazepine, vigabatrin, or gabapentin; and a reduced dose of phenobarbital. Withdrawal of the aggravating agents and adjustment of medication resulted in full seizure control.

Sensory systems

Psychological, psychiatric

Behavioral and psychiatric disturbances are less common with carbamazepine than with other anticonvulsants. The absolute effect of antiepileptic drugs on cognition can only be observed in healthy volunteers. The neuropsychological and neurophysiological effects of carbamazepine and levetiracetam have been compared in a double-blind, two-period, crossover, randomized study in 28 healthy adults [ ]. Evaluations were conducted at times of screening, baseline pre-drug treatment, the end of each maintenance phase (4 weeks), and the end of each washout period. The dosage of carbamazepine was adjusted to achieve mid-target range concentrations. The dosage of levetiracetam was titrated to 2000 mg/day. An overall composite score showed significantly worse effects of carbamazepine compared with levetiracetam and for both drugs compared with non-drug treatment. Furthermore, carbamazepine was worse than levetiracetam in 15 of 34 variables and none favored carbamazepine. These results suggest that carbamazepine has more negative cognitive effects than levetiracetam at commonly used dosages and that these differences may be clinically significant in some individuals.

The cognitive effects of carbamazepine and remacemide hydrochloride have been compared in a double-blind, parallel-group study in patients with newly diagnosed epilepsy, who were randomized to either carbamazepine 200 mg/day (n = 282) or remacemide 200 mg/day (n = 288) for 2 weeks and slowly titrated to a target dose of 600 mg/day of both drugs [ ]. After the 6-week titration period, the patients entered a maintenance phase, with possible upward or downward dosage adjustments of 200 mg/day. Repeated assessments of neuropsychological function and mood were carried out using computerized and conventional measures. The trial ended after 20 months, following the second interim analysis, which showed inferiority of remacemide compared with carbamazepine in preventing seizure recurrence. The patients who took carbamazepine had consistently worse performances than those who took remacemide on measures of information processing, psychomotor speed, and attention. The authors suggested that the observed differences may have been due to direct pharmacological effects of the treatments themselves and did not seem to be influenced by aspects such as seizure control and adverse events profiles.

• A 9-year-old boy with seizures developed intermittent complex visual hallucinations during therapy with fosphenytoin and, on a separate occasion, carbamazepine [ ].

The cognitive effects of carbamazepine and tiagabine in adults with newly diagnosed epilepsy have been compared using pooled data from two randomized studies with similar populations, dosing, and cognitive assessments. Tiagabine monotherapy 20–30 mg/day had a cognitive profile similar to that of long-term carbamazepine monotherapy 400–800 mg/day. There was no deterioration in any cognitive functions between tests performed at baseline and after 52 weeks of treatment (for more information see also tiagabine ) [ ].

In 10 children with rolandic epilepsy, carbamazepine impaired memory and possibly visual search tasks [ ]. Evaluation of individual data suggested that some children were especially vulnerable to the adverse effects of carbamazepine on cognition. The authors did not comment on the fact that rolandic epilepsy is regarded as a syndrome for which treatment in most cases is not indicated.

The cognitive effects of carbamazepine and gabapentin have been compared in a double-blind, crossover, randomized study in 34 healthy elderly adults, of whom 19 withdrew (15 while taking carbamazepine, probably because of excessively rapid dosage titration) [ ]. The primary outcome measures were standardized neuropsychological and mood state tests, yielding 17 variables. Each subject had cognitive testing at baseline (before drug treatment), at the end of the first drug phase, the end of the second drug phase, and 4 weeks after completion of the second drug phase. Adverse events were frequently reported with both anticonvulsants, although they were more common with carbamazepine. There were significant differences between carbamazepine and gabapentin for only one of 11 cognitive variables, with better attention/vigilance for gabapentin, although the effect was modest. Both carbamazepine and gabapentin can cause mild cognitive deficits in elderly subjects, and gabapentin has a slightly better profile.

Endocrine

Carbamazepine therapy in eight women was associated with increased serum concentrations of sex hormone binding globulin (SHBG) and a reduced serum ratio of 17-α-estradiol and estradiol to SHBG. Of 56 women who had been taking carbamazepine for over 5 years, 14 had menstrual disturbances, which tended to be associated with raised SHBG and reduced serum concentrations of 17-α-estradiol [ ]. Carbamazepine may also reduce serum LH, progesterone, dehydroepiandrosterone, and the free androgen index [ , ], and some of these changes may be associated with anovulatory cycles and menstrual irregularities.

Men taking carbamazepine had mean low serum concentrations of dehydroepiandrosterone and a high SHBG concentration. Moreover, 18% of men taking carbamazepine for epilepsy reported reduced libido, impaired potency, or both.

Carbamazepine affects the serum concentrations of thyroid hormones, typically reducing concentrations of circulating thyroxine, both total and free, with variable effects on triiodothyronine. In patients with no underlying thyroid pathology this is only exceptionally clinically important. However, in hypothyroid patients such changes may be significant. Thyrotropin, total thyroxine, and free thyroxine serum concentrations have been studied in 29 hypothyroid patients taking levothyroxine and in 19 patients with no thyroid disorders who were taking levothyroxine for other reasons; measurements were made before carbamazepine therapy and then weekly for 7 weeks [ ]. In the first group, total thyroxine fell significantly by 15–25%, starting from the first week; the fall in free thyroxine was smaller (10–15%) and delayed. In contrast, thyrotropin concentrations increased only slightly and stayed within the reference range. In the second group there were similar falls in total and free thyroxine, but followed by a significant increase in thyrotropin. Hence, in patients with no thyroid pathology there was a compensatory reaction, leading to a new steady state. In patients who were taking levothyroxine for hypothyroidism, increased metabolic clearance of thyroid hormones may lead to worsening of hypothyroidism.

Metabolism

Changes in body weight have been evaluated in 349 patients taking carbamazepine, phenytoin, or tiagabine. Carbamazepine add-on therapy caused significant mean weight gain of 1.5% [ ]. Tiagabide add-on therapy caused no significant weight change when added to either phenytoin or carbamazepine.

Some antiepileptic drugs have been associated with low serum and erythrocyte folate concentrations and high total plasma homocysteine concentrations in some patients. The concentrations of folate and homocysteine have been measured in 42 patients taking carbamazepine and 42 matched healthy controls [ ]. Patients taking carbamazepine had significantly lower serum and erythrocyte folate concentrations. There was hyperhomocystinemia (over 15 μmol/l) in 24% of the patients and 5% of the controls.

The concentrations of serum lipids and lipoprotein (a) were measured in 20 children taking carbamazepine, 25 taking valproic acid, and five taking phenobarbital [ ]. In those taking carbamazepine, total, HDL, and LDL cholesterol were increased while they were eating a normal diet and normalized with a low-fat diet and 3 months after the end of treatment.

The possible association between the effects of carbamazepine on thyroid function and lipid profile has been studied in 18 children with epilepsy [ ]. During carbamazepine monotherapy there was a significant association between serum low-density lipoprotein (LDL) cholesterol and thyrotropin concentrations at 6 and 12 months. These results suggest that the increase in LDL cholesterol concentrations that occur in patients taking carbamazepine may be the consequence of altered thyroid function.

Nutrition

Concentrations of plasma homocysteine, plasma pyridoxal 5′-phosphate (active vitamin B ₆ ), serum folate, erythrocyte folate, and serum vitamin B ₁₂ have been measured both during fasting and after methionine in 60 epileptic patients (aged 14–18 years) and 63 sex- and age-matched controls before therapy and after 1 year of therapy with valproate or carbamazepine [ ]. After 1 year the patients who took valproate and carbamazepine had significantly increased plasma homocysteine concentrations compared with both baseline and control values and there was a significant fall in serum folate and plasma pyridoxal 5′-phosphate. Serum vitamin B ₁₂ and erythrocyte folate were unchanged.

The genetic determinants of this effect on homocysteine have been determined in 136 epileptic children taking carbamazepine or valproate as monotherapy [ ]. Nutritional determinants (folate and vitamins B ₆ and B ₁₂ ) and genetic determinants (MTHFR 677CT) of plasma homocysteine were studied in a random sample of 59 of those children. Total homocysteine concentrations were significantly increased and folate and vitamin B ₆ concentrations were significantly reduced. Carbamazepine lowered folate concentrations in association with hyperhomocysteinemia, which seemed to be related to the homozygous MTHFR 677CT mutation. Valproate, although also associated with hyperhomocysteinemia, only reduced vitamin B ₆ concentrations, independent of the MTHFR genotype.

Electrolyte balance

Carbamazepine can cause hyponatremia due to stimulation of antidiuretic hormone secretion [ , ]. In a boy who developed hyponatremia it also seems to have caused fluid retention and cardiomegaly [ ].

Symptomatic cases of hyponatremia are relatively rare, but two cases have been reported [ ].

• A 72-year-old woman developed somnolence and continuous sharp waves over the left hemisphere when carbamazepine was added to primidone and a diuretic. Her serum sodium concentration was initially 130 mmol/l and fell further to 100 mmol/l.

• In a 65-year-woman, carbamazepine-induced hyponatremia (127–130 mmol/l) led to serial tonic–clonic seizures, somnolence, confusion, and an electroencephalographic misdiagnosis of non-convulsive status.

Both patients recovered with normalization of the serum sodium when the carbamazepine dose was tapered.

In 117 patients with chronic epilepsy taking carbamazepine in residential care the retrospective prevalence of hyponatremia was 42% compared with 9.4% in controls [ ]. Higher doses and serum concentrations of carbamazepine were associated with a higher risk of hyponatremia.

Hematologic

In view of the very low incidence of serious blood dyscrasias, such as pancytopenia and agranulocytosis, indiscriminate continuous hematological monitoring is of little value [ ]. Leukopenia occurs in up to 21% of patients, usually during the first 3 months of treatment, with a higher risk in those with a low or low-normal pretreatment white blood cell count. Although it may reverse during continuation of treatment [ ], high-risk cases should be monitored. Bone-marrow aspirates need not be performed routinely in patients with chronic leukopenia and continuation of treatment is probably safe, although caution is required if the neutrophil count is consistently below 1.0 × 10 ⁹ /l [ ]. Other authors have recommended a reduction in dose or withdrawal of treatment if white cell counts are below 3.0 × 10 ⁹ /l or if neutrophil counts are below 1.0 × 10 ⁹ /l.

Some 15 cases of thrombocytopenia, reversible after withdrawal of carbamazepine, have been published [ , ]. There have also been single case reports of reticulocytosis [ ], leukopenia with thrombocytopenia with Henoch–Schönlein purpura [ ], hemolytic anemia, and pure red cell aplasia [ ].

Serious blood dyscrasias from carbamazepine can be accompanied or preceded by a rash.

• A 66-year-old man had severe leukopenia a few days after a generalized rash appeared on the 36th day after he started to take carbamazepine [ ].

• A 69-year-old woman had severe leukopenia and thrombocytopenia about a month after the onset of a rash and 2 months after she started to take carbamazepine [ ].

In both patients, the abnormalities resolved after withdrawal, except for the platelet count, which increased but did not fully normalize over 6 weeks. This report suggests that patients developing a rash on carbamazepine should be monitored for the possible risk of associated blood dyscrasias.

Increased concentrations of PIVKA-II (prothrombin-induced by vitamin K absence for factor II) were found in four of six samples of cord blood collected at parturition from the placenta in 12 mothers exposed to carbamazepine during pregnancy [ ]. Prothrombin concentrations were also reduced in the whole group. High PIVKA-II concentrations were also recorded in cord blood from a newborn exposed prenatally to phenytoin and vigabatrin, but not in two exposed to phenytoin alone and one exposed to valproate alone. These results are consistent with evidence that enzyme-inducing anticonvulsants, particularly carbamazepine, interfere with vitamin K metabolism during pregnancy and can result in bleeding disorders in the newborn. Administration of vitamin K is recommended in these newborns at time of birth or in the mother toward the end of pregnancy. The authors also discussed the possibility that vitamin K supplementation in early pregnancy might reduce the risk of fetal facial dysmorphisms caused by enzyme-inducing anticonvulsants.

In a case–control study in 173 patients with aplastic anemia and 497 controls, of the 16 patients who had used antiepileptic drugs during the year before the index date, eight had used carbamazepine, six valproate, and four phenytoin [ ]. There was a nine-fold increased risk of aplastic anemia in patients taking antiepileptic drugs, especially carbamazepine and valproate. Polytherapy with antiepileptic drugs was more strongly associated with aplastic anemia than monotherapy.

A young patient with a bipolar disorder started carbamazepine and developed pure red cell aplasia [ ].

Carbamazepine can cause pseudolymphoma, and there have also been isolated reports of true lymphomas.

• A 73-year-old patient with erythroderma had an anaplastic large cell lymphoma; he had started to take carbamazepine for diabetic neuropathy 3 months before the development of skin lesions [ ].

However, the time course in this case was very short, suggesting that carbamazepine was not implicated.

In a 5-year-old boy intravascular hemolytic anemia with erythroblastopenia was considered probably due to carbamazepine [ ]. Subsequently reduced erythrocyte activity of the enzyme glutathione peroxidase was found in the boy and his mother. The authors speculated that reduced activity of this enzyme confers a high risk of carbamazepine-induced hemolysis.

Gastrointestinal

Nausea, vomiting, and gastric intolerance are relatively uncommon and mild. However, there have been reports of severe diarrhea requiring drug withdrawal. Eosinophilic colitis [ ] and lymphocytic colitis [ ] are rare reactions.

A report of two cases has suggested that carbamazepine can cause colitis as part of the anticonvulsant hypersensitivity syndrome [ ].

• A 47-year-old man developed fever, lymphadenopathy, flu-like symptoms, facial edema, a rash, and diarrhea after taking carbamazepine 200 mg/day for 3 weeks. Laparotomy because of severe abdominal pain 2 weeks later showed severe colitis with perforations.

• A 41-year-old woman had diarrhea, fever, and a rash after taking carbamazepine 300 mg/day for 4 weeks. Colitis was confirmed at colonic biopsy.

Both patients also had raised liver enzymes, peripheral eosinophilia, and eosinophils in the colonic infiltrate. Persistent recovery occurred after carbamazepine was withdrawn.

Lymphocytic colitis has been attributed to carbamazepine in a 77-year-old man, who had taken it for 6 months [ ].

Two cases of carbamazepine-induced eosinophilic esophagitis in the contest of DRESS (see below) have been reported [ ].

Liver

There have been several reports of fatal hepatic failure [ ], granulomatous hepatitis [ ], cholestatic hepatitis [ ], and cholangitis ascribed to carbamazepine. Most acute hepatotoxic reactions caused by carbamazepine are accompanied by fever, rash, eosinophilia, and other signs of hypersensitivity, although they have also occurred without rash and eosinophilia [ ].