Roberto Michelucci MD, PhD*
Carlo Alberto Tassinari MD**
*Deputy Chief, Department of Neurological Sciences, Bellaria Hospital, Bologna, Italy
**Neurological Chief, Department of Neurological Sciences, University of Bologna, Bologna, Italy
The benzodiazepines (BZDs) represent a large and popular class of drugs with similar mechanisms of action, clinical indications, and efficacy and toxicity profiles. Although all the BZDs have anticonvulsant activity in a variety of animal models, only a few have been developed for the management of epilepsy. Some of these, such as diazepam (DZP) and lorazepam (LZP), are mainly used for epilepsy emergencies (e.g., status epilepticus) whereas other compounds, such as clonazepam (CZP) and clobazam (CLB), are usually employed for long-term treatment. Depending on the route of administration—oral, rectal, intravenous (i.v.), or intramuscular (i.m.)—and the clinical application (acute or chronic), the spectrum of adverse effects and toxicity issues vary widely. Therefore, in the present chapter, the adverse effects of BZDs are reviewed separately according to the route of administration and clinical application. A section concerning the manifestations and management of overdose follows.
ADVERSE EFFECTS OF BENZODIAZEPINES IN THE MAINTENANCE TREATMENT OF EPILEPSY
CZP, CLB, and, to a lesser extent, DZP, LZP, nitrazepam (NZP), and clorazepate (CLP) may be administered orally as add-on drugs in epileptic patients refractory to previous medications. The fact that the BZDs are mostly used as adjunctive therapy may lead to an overestimation of their potential of toxicity, resulting from pharmacokinetic and pharmacodynamic interactions. The adverse effects of BZDs as long-term therapy include dose-related toxicity, hypersensitivity reactions, paradoxical effects, withdrawal effects, side effects from drug interactions, teratogenicity, and effects on neonates.
Common dose-related effects involve the central nervous system (CNS), occur mostly at the beginning of treatment, and disappear with dose reduction. They also tend to lessen with the duration of treatment because of the development of tolerance.
For most BZDs, there seems to be some correlation between plasma drug concentrations after a single dose and CNS-related effects (1, 2, 3, 4, 5). At about 200 ng/mL of DZP, subjects tended to become slightly tired and drowsy (2). After i.m. administration of 10 mg of DZP, coordinative and reactive skills were impaired for as long as 5 hours at plasma concentrations exceeding 180 ng/mL (3). Using a sensitive neurophysiologic test, Bittencourt et al. (6) demonstrated a clear correlation between NZP and DZP concentrations and their effect on the peak velocity of saccadic eye movements (which is a measure of brainstem reticular formation function). With prolonged treatment, however, the relationship between plasma concentration and the effect of BZD drugs tends to become blurred because of the development of tolerance. Although dose dependency of side effects during maintenance treatment has been advocated for most BZDs, no relationship was found between the adverse effects and the plasma concentration of CZP (7,8). With DZP, disturbing side effects such as marked drowsiness, vertigo, ataxia, and impaired performance were associated with plasma concentrations >900 to 1,000 ng/mL (4). However, interindividual variability in response at any given concentration is considerable, and measurement of drug levels has no important role in optimizing BZD treatment.
Sedation or Drowsiness.
This symptom is the most frequent complaint and is reported to occur in 10% to 85% (median, 62%) of patients receiving CZP (9) and in >40% of those taking DZP and NZP (10). Conversely, CLB has been associated with less sedative effects. In a review of 70 double-blind clinical trials comparing the effects of DZP, CLB, and placebo for anxiety and other psychiatric indications, Koeppen (11) found that the incidence of sedation was 26% with CLB, 46% with DZP, and 10% with
placebo. These results were obtained, however, with high doses of CLB (>30 mg a day) and DZP (>15 mg/day). When adverse reactions to CLB were analyzed in epileptic patients, drowsiness was found to occur less frequently, probably because of lower doses employed (12). In a double-blind placebo controlled study of 10 healthy volunteers, CLB (20 to 30 mg/day) had fewer side effects than LZP (2 to 3 mg/day) (13). In healthy volunteers, CLB (10 and 20 mg) was also significantly less sedative than CZP (0.5 and 1.5 mg), although neither drug affected the ventilatory response to carbon dioxide (14).
Incoordination or ataxia is probably the second most common side effect of long-term BZD use. It is sometimes associated with dizziness and dysarthria. NZP, DZP, and CZP are more frequently observed to cause incoordination or ataxia, but this adverse effect has been reported for all the marketed BZDs, including CLB and CLP. Ataxia was found in 7% to 43% of patients treated with CZP (9) and in 5% to 50% of patients taking DZP and NZP as long-term therapy for seizure control (10). Ataxia was also mentioned in four of 23 open-label studies of CLB in epilepsy reviewed by Koeppen (12).
Significant behavioral and personality changes such as hyperactivity, restlessness, short attention span, irritability, and aggression may occur during long-term therapy with BZDs. Children are particularly likely to experience alterations in behavior and emotional state, but the elderly may be also susceptible. Behavioral effects have been reported in 2% to 51% (median, 12%) of patients taking CZP (9). NZP has been shown to induce symptoms of CNS stimulation, such as nightmares, insomnia, and agitation (15,16). Adverse personality changes, especially irritability, can occur during CLP therapy, particularly in association with primidone (17). Although CLB seems to induce behavioral effects less commonly (18), increased irritability, depression, and disinhibition may occur with CLB in chronically institutionalized patients (19). Sheth et al. (20) described seven developmentally disabled children treated with CLB for refractory epilepsy who developed a severe behavior disorder (aggressive agitation, self-injurious behavior, insomnia, incessant motor activity) 10 to 55 days after initiation of drug therapy. These disturbances resolved promptly on drug discontinuation.
Cognitive and Psychomotor Impairment.
Several studies demonstrated that the administration of BZDs may impair human performances and psychomotor skills, mostly after single doses or short-term administration (21,24,25). When BZDs are administered repeatedly, these effects tend to be less prominent. Although some reviews state that no clear differentiation existed between the types of performances affected by the BZDs (22), other reviews have indicated that the performances to which mental speed is essential may be particularly affected (23). Of particular interest are the data concerning the residual effects of NZP on skills related to driving or other daily activities (21). Although some studies failed to detect negative effects of CLB on psychomotor performance (26), CLB may cause a slight impairment in some cognitive tests, mainly on retrieval processes and mental rapidity (27).
Studies of presurgical patients strongly suggest that BZDs can impair recall of events that follow the administration of these drugs. Studies in normal volunteers have also demonstrated clearly detrimental effects of BZDs on the ability to recall stimuli presented after drug administration. In clinical practice, anterograde amnesia has been reported with LZP at usual doses (28).
Other Clinically Relevant Side Effects.
Muscle weakness, fatigue, and hypotonia are sometimes reported during BZD treatment; CZP, CLB, and NZP are particularly prone to cause this effect (9,16,18). BZDs should not be given to myasthenic patients. Visual disturbances, including nystagmus, blurred vision, and diplopia, may occur with any BZD, but they are not common (10). Although BZDs apparently have varying effects on appetite and weight, increased appetite and weight gain are more frequently reported (10). A significant weight gain was observed in nine of 81 patients taking CZP for more than 2 weeks (29), and it was also mentioned in five of 23 open studies with CLB reviewed by Koeppen (12). Acute psychotic reactions have been reported with BZD treatment but are exceptionally rare (9,10).
Hypersecretion and Drooling.
Increased salivation and hypersecretion of the tracheobronchial tree have been noted with all the BZDs in children, but these findings are absent or negligible in adults (10). CZP and NZP are particularly prone to induce these effects (7,9,30,31). Although these symptoms are usually attributed to salivary and bronchial hypersecretion (7,31), Wyllie et al. (32) performed manometric examinations in two children with NZP-induced drooling and aspiration and demonstrated a delay of the cricopharyngeal relaxation during swallowing. These investigators suggested that NZP-induced drooling and aspiration may be caused by impaired swallowing.
Nitrazepam-Induced Death in Children.
In 1987, Murphy et al. (33) observed an association between NZP and death in children with epilepsy who were receiving NZP at dosages ranging from 0.9 to 2.5 mg/kg/day. Consequently, a dosage of NZP <0.8 mg/kg/day was recommended in children (33). In a prospective study of esophageal manometry in 14 children receiving NZP for myoclonic epilepsy, Lim et al. (34) demonstrated swallowing incoordination with delayed cricopharyngeal relaxation in four of 14 patients. One of these patients developed respiratory distress and bronchospasm, which improved dramatically together with normalization
of manometric findings after NZP was discontinued. In light of these results, a disturbance of swallowing was postulated as a possible cause of unexplained sudden death in children, and esophageal manometry was suggested as a helpful technique for defining patients at risk of sudden death (34). Rintahaka et al. (35) reported 21 deaths in 302 children who entered into a NZP study for intractable epilepsy. NZP therapy (at dosages ranging from 0.15 to 2.55 mg/kg/day) was associated with a statistically increased risk of mortality in young children (<3.4 years) with refractory epilepsy. Dysphagia, recurrent respiratory tract infections, gastroesophageal reflux, and aspiration apparently increased the likelihood of death in patients taking NZP. It is not known whether the risk of increased mortality in young children with intractable epilepsy is a NZP-specific problem or is common to other BZDs.
BZDs rarely produce allergic reactions (10), and these drugs are commonly recommended in the management of seizures in patients with hypersensitivity reactions to antiepileptic drugs (36).
Rashes have been attributed to BZDs by several authors (31,37, 38, 39). Leukopenia has been reported to be caused by DZP (40), CZP (41), and NZP (42), and it is apparently reversible. Thrombocytopenic purpura has been described in association with CZP (43), whereas systemic lupus erythematosus has been associated with CLB (44). Hepatic necrosis has been associated with CLP (45) and DZP (46). Esophageal burn was reported in one patient taking CLP (47), and allergic intestinal nephritis occurred in one patient receiving oral DZP (48).
Long-term oral treatment with BZDs, particularly NZP and CZP, has been reported to cause a paradoxical increase in seizure frequency (42,49,50) or the appearance of new seizure types (51,52). Peterson (42) reported that six of 108 epileptic patients treated with oral NZP had an exacerbation of seizures. Similarly, increased frequency and severity of grand mal seizures were observed in 14 patients after NZP oral administration (50). In a survey of a large patient population treated with various oral BZDs, Alvarez et al. (53) reported that 5.8% of patients may have suffered from seizure aggravation. Worsening of seizures appeared to be more common in patients treated with CZP and particularly concerned tonic seizures and absences. In a retrospective study analyzing approximately 2,500 phases of add-on therapy in a total of 1,006 patients suffering from focal epilepsy, CLB and CZP were found to increase seizure frequency in 2% of cases (54). Oral BZDs, particularly CZP, have been correlated with precipitation of tonic-like microseizures in infants with West's syndrome (51,52).
In persons who do not have epilepsy, abrupt withdrawal of BZDs has been associated with convulsions, worsening of insomnia, psychosis, and delirium tremens (10,55, 56, 57). Particular attention has been devoted to the occurrence of de novo nonconvulsive status epilepticus after BZD withdrawal in nonepileptic patients (58, 59, 60). The clinical picture consists of acute mental confusion with an electroencephalographic (EEG) correlate of diffuse spike-and-wave discharges occurring in middle-aged or elderly patients with a history of long-term BZD use. These episodes may be misdiagnosed as acute BZD intoxication, and the diagnosis may be difficult if the EEG is delayed or is not performed; however, treatment with the specific BZD antagonist flumazenil aggravates the confusional state, whereas DZP administration leads to rapid recovery (58, 59, 60). Instances of grand mal seizures or convulsive status epilepticus have been also reported, usually after LZP has been discontinued (61).
In epileptic patients, abrupt withdrawal of BZDs commonly produces an increase of seizure frequency and severity, sometimes leading to convulsive status epilepticus. A “negative myoclonus” state has been reported after discontinuation of CLB (62). Although slow tapering is associated with a lower risk of precipitation of seizures, no optimal rate of withdrawal has been calculated for any BZD. Safe discontinuation rates for CZP were estimated to be <0.04 mg/kg/day (63) and 0.2 mg/day (64). Paradoxically, discontinuation of a BZD may result in amelioration of seizure activity in some patients (65).
Side Effects as a Result of Interaction with Other Drugs
BZDs may interact with other drugs, usually through pharmacodynamic mechanisms. It is well known that BZDs potentiate the action of CNS depressant drugs, such as ethanol or barbiturates, and BZDs may produce CNS depression and respiratory irregularities when they are given in association with amphetamines and methylphenidate (66). Feldman (17) reported the deleterious effects on behavior of the combination of CLP and primidone. The ability of BZDs to interact with other drugs through pharmacokinetic mechanisms is minimal. However, an increase in phenytoin levels, sometimes leading to intoxication, has been reported after CZP (67) or CLB (68) introduction. CLB may also induce an increase in the levels of carbamazepine (69) and carbamazepine-10,11-epoxide (70). A case of carbamazepine intoxication with negative myoclonus after addition of CLB has been reported (71). Cocks et al. (72) found that CLB may elevate valproate levels, and these investigators suggested that a combination of valproate and CLB, particularly at high doses, should be avoided because toxicity may be common. Conversely, valproate may induce a clinically significant potentiation of the effects of NZP (73), LZP (74), and DZP (75).
Teratogenicity and Effects on Postnatal Development
The possible teratogenic effects of BZDs are difficult to assess but appear to be mild. No definite causal relationship has been established, even though an association between first trimester intake of BZDs and oral clefts (especially cleft palate) was found in some retrospective studies (76,77), but not in others (78,79). Oral clefts have been reported with DZP, NZP, oxazepam, and chlordiazepoxide assessed as a group (77), as well as in individual case reports. The actual risk of having a child with cleft lip and cleft palate has been estimated to be 0.4% with DZP (80). Multiple limb anomalies were reported in a newborn child whose mother had taken CLP during the first trimester (81). Intake of BZDs in early pregnancy, as verified by case records and controls (82) and detection of high serum levels (83), has also been associated with an increased risk of CNS malformations and dysmorphism. An amplifying action of BZDs on valproate teratogenicity has been suggested (84).
Newborns, especially premature newborns, exposed to BZDs in utero during the late third trimester or at the time of delivery may present with the floppy infant syndrome consisting of floppy movement, poor sucking, hypotonia, hypothermia, poor reflexes, low Apgar scores, and apnea (66,85,86). A BZD withdrawal syndrome has also been observed in neonates (86).
ADVERSE EFFECTS OF BENZODIAZEPINES GIVEN AS ACUTE THERAPY IN EPILEPSY
BZDs play a prominent role in the emergency management of epilepsy. DZP, CZP, LZP, and midazolam (MDL) are commonly given i.v. for the treatment of status epilepticus (10,87,88); however, alternative routes of administration are feasible for the management of ongoing or serial seizures at home or when i.v. injection is inconvenient because of thrombophlebitis or technical difficulties during status epilepticus. These alternative routes include rectal (for DZP, MDL, LZP, CZP) or i.m. (for MDL) administration. Nasal administration of MDL for the treatment of acute seizures has also been described (89).
Adverse effects of i.v. BZDs include systemic and CNS toxicity, local tissue irritation, and paradoxical effects.
Systemic and Central Nervous System Toxicity
Acute adverse effects consist mostly of hypotension and CNS-related events, specifically respiratory depression and profound sedation.
Browne and Penry (10) reviewed 401 patients with status epilepticus who were treated with DZP and found 16 cases of severe respiratory depression, along with 10 of marked hypotension. In a consecutive series of 33 patients, respiratory depression occurred in only one patient who received 25 mg, and there was mild to moderate respiratory depression in three other patients at doses of 10 mg (90). In another series of 98 patients, apnea appeared after i.v. injection of 5 and 10 mg in two patients with aminophylline-induced and lidocaine-induced seizures (91). Mild to severe hypotension and temporary respiratory depression were reported in 5.2% of 246 patients receiving multiple therapy (92). One of these patients died. Appleton et al. (93) reported a 15% incidence of respiratory depression in 53 patients treated with i.v. DZP, with half of these patients requiring repeated multiple doses. Overall, a review of the pertinent literature demonstrates that respiratory depression and hypotension occur more frequently when DZP is used in combination with other agents, such as barbiturates (94), lidocaine and epinephrine (95), methaqualone (96), chlordiazepoxide, and amobarbital (97). Additional risk factors include severe brain damage as a cause of status epilepticus (98), older age (99), and decompensated liver disease (100). Although the literature is reassuring about the risks, the rate of bolus injection is a critical factor. It has been claimed that the rate of bolus injection should not exceed 2 to 5 mg/min, to avoid serious respiratory depression (101). The propylene glycol solvent may contribute to the cardiorespiratory effects attributed to DZP (66).
The acute side effect profile of i.v. CZP is similar to that of DZP. As with DZP, the risk of respiratory depression and hypotension is greatest after acute brain injury, in patients who have already received barbiturates, and in the elderly (98). CZP sedates and depresses levels of consciousness and is more potent than DZP in this regard (92).
Like DZP and CZP, LZP has the potential for producing respiratory depression and hypotension, but prior medication with other antiepileptic drugs does not seem to increase the risk or severity of adverse reactions (102). Moreover, the risk of respiratory depression seems to be greater after the first injection of LZP than on subsequent injections (103). The incidence of respiratory depression with LZP therapy in status epilepticus ranges between 3% (93) and 10% (90), a rate substantially similar to that reported with DZP, although some studies suggested a lower risk with LZP (93). In the Veterans Administration comparative study of four treatments for generalized convulsive status epilepticus, LZP (0.1 mg/kg) was found to have the same incidence of adverse reactions as DZP (0.15 mg/kg) and phenytoin (18 mg/kg) (104).
In early studies of preoperative anesthetic practice, apnea was reported in 10% to 77% of patients who received i.v. MDL (0.2 to 0.36 mg/kg) (105). In status
epilepticus, the i.v. dose used is lower than that previously used for anesthesia, and apnea has not been reported but is a potential risk. Toxicity is more likely to occur in elderly patients (88).
Local Tissue Irritation
Venous thrombosis, phlebitis, and pain may occur at the site of infusion of DZP or CZP. Thrombophlebitis occurred in 3.5% of >1,500 i.v. injections given during gastroscopy and may have been the result of drug precipitation caused by rapid injection (106,107). However, in a review provoked by a startling article about the loss of a limb after i.v. DZP administration, no significant local vascular complications were found among 15,813 injections (108). Overall, the dangers of local complications of i.v. BZDs appear to be negligible.
FIGURE. 19.1. Paradoxical reaction to an intravenous benzodiazepine injection in a 2.5-year-old child with Lennox-Gastaut syndrome. The patient has “atypical absence status” with diffuse slow spike-and-wave (SW) discharges. A few seconds after receiving a 7-mg injection of clonazepam (Ro-4023), the patient had a tonic seizure, characterized by initial flattening of the tracing and a progressively increasing-amplitude diffuse polyspike discharge. The child opened the eyes, and the eyeballs rose upward. Additional tonic seizures, with the same electroclinical manifestations, occurred over the next few minutes, with persistence of diffuse slow SW discharges between the seizures. (Courtesy of Drs. C.H. Dravet and J. Roger, Centre St. Paul, University of Marseille, France, unpublished data, 1971.)
Paradoxical effects, such as the induction of tonic status epilepticus, have been described with the i.v. use of DZP, CZP, NTZ, and LZP, especially in children prone to this seizure type (109, 110, 111) (Figure 19.1). Tassinari et al. (109) gave the first account of this complication: in five children with Lennox-Gastaut syndrome, the i.v. injection of DZP (10 mg) triggered very frequent tonic seizures, amounting to tonic status, in 5 seconds to 11 minutes. The sleepinducing properties of BZDs did not seem to be a critical factor in the appearance of tonic seizures in these patients.
Numerous studies have stressed the good safety profile of rectally administered BZDs, usually DZP (112, 113, 114).
Overall, these studies have shown that the rectal administration of DZP induces sedation in 17% to 33% of patients, with no respiratory depression. At variance with these favorable results, a 9% incidence of respiratory depression, after the use of rectal DZP as monotherapy, was observed in a prospective study including 97 children with 130 episodes of acute seizures (115). Moreover, Brodtkorb et al. (116) emphasized the risks of long-term or excessive administration of rectal DZP, which may cause a cyclic reappearance of seizures or the combination of toxic and withdrawal effects in some patients.
MDL is the only BZD used in acute epilepsy that can be given with benefit by i.m. injection. Reports of respiratory depression after i.m. MDL injection for status epilepticus have not been published. There is, however, one report of a patient who developed apnea 20 minutes after receiving 10 mg i.m. MDL (88). Ghilain et al. (117) reported one patient with bradycardia, and three patients with a slight decrease in blood pressure, among 14 patients treated with i.m. MDL.
MANIFESTATIONS AND MANAGEMENT OF OVERDOSE
An extensive survey of hospital admissions for drug overdose in the United States showed that 13% involved BZD ingestion, usually in combination with other drugs (e.g., barbiturates, sedative-hypnotics, ethanol, or miscellaneous drug combinations) (118).
Signs of BZD overdose vary with the particular drug, the doses taken, and the age of the patient. Generally, when BZDs are taken at relatively low doses, patients are somewhat somnolent and exhibit ataxia (119). At higher doses, patients are often comatose and areflexic; when awake, these patients exhibit nystagmus, ataxia, dysarthria, and occasionally hypotension (120). Both very young and very old patients may develop these manifestations at lower doses (21). Respiratory depression, which is an infrequent sign in patients with exclusive BZD overdosage, has been reported with higher frequency in cases of multiple drug ingestion, sometimes leading to death (21). By contrast, death after the ingestion of BZDs alone is exceedingly rare (121). Out of 102 patients with NZP overdoses, only six were deeply comatose, and these patients recovered uneventfully in 12 hours (122). Nonetheless, death has been reported after an overdose with an undetermined amount of NZP (123). After massive DZP overdoses, patients with plasma concentrations <20,000 ng/mL DZP and 5,000 ng/mL N-desmethyldiazepam have survived (66). Levels of N-desmethyldiazepam of 10,000 ng/mL have been observed in patients who ingested large amounts of CLP but who remained conscious with ataxia (124). Rapid clinical recovery from BZD overdose does not result from rapid elimination of metabolites—which have a long half-life—but is more likely to be related to the development of tolerance to the depressant effects of the drug (66).
A few case studies provide suggestive evidence that the effects of high doses of LZP may differ from those of other BZDs. Overdose with LZP has resulted in hallucinations (125), delirium (126), and transient global amnesia (127). Bullous skin lesions have been reported in patients with coma induced by DZP (128) and NZP (129). Exocrine sweat gland necrosis may also occur after DZP overdose (128). Comatose patients with NZP (130) or LZP (131) intoxication may show the peculiar EEG pattern of alpha coma. A case of nonfatal cardiac arrest was reported after DZP overdose (70 mg) in a 2-year-old child (132).
Treatment of BZD overdose consists of general supportive care, close observation, and, in most serious cases, admission to intensive care units. Apart from standard intensive care treatment, the BZD receptor antagonist flumazenil is usually given by i.v. repeated single injections or continuous infusion (133) to reverse BZD-induced CNS depression. Activated charcoal (134), exchange transfusion (135), and physostigmine (136) have been also employed in DZP intoxication. In an early study (137), the sedative effects of LZP were reversed with an infusion of 1 mg/kg of aminophylline.
BZDs are widely used for the treatment of epileptic disorders. When administered orally as long-term therapy, BZDs produce dose-related CNS effects, particularly sedation, ataxia, and behavioral and cognitive changes. Many of these effects, which occur in 20% to 60% of patients, tend to decrease over a few weeks because of the development of tolerance. Less frequent, but clinically significant, side effects include fatigue, blurred vision, and memory disturbances. Drooling can be caused by impaired swallowing or hypersecretion in NZP-treated children; moreover, impaired swallowing has been claimed to be responsible for the increased risk of death among young children given NZP for intractable epilepsy. Abrupt BZD withdrawal has been associated with a risk of seizures, psychosis, and status epilepticus, both in epileptic and nonepileptic patients. When administered i.v. for the acute management of status epilepticus, BZDs may cause respiratory depression and hypotension. These events are more frequent when BZDs are used in combination with other sedative agents or in patients with severe brain damage, old age, and liver impairment. Additional effects include damage at the site of injection (which is negligible for LZP and MDL) and paradoxical induction of tonic status epilepticus in patients with Lennox-Gastaut syndrome. BZDs are relatively safe in overdose,
and fatalities are exceedingly rare. Signs of acute intoxication range from somnolence and ataxia to areflexic coma. Hallucinations, delirium, and transient global amnesia have been reported with LZP. Apart from supportive treatment, the BZD receptor antagonist flumazenil can be given by i.v. injection to reverse BZD-induced CNS depression.