James L. Willmore MD
Associate Dean and Professor, Department of Pharmacology and Physiology, Saint Louis University School of Medicine; and Attending Neurologist, Department of Neurology, Saint Louis University Hospital, Saint Louis, Missouri
Treatment of patients with epilepsy that is guided by the goals of complete seizure control without intolerable drug side effects is commonly compromised when control cannot be achieved because titration is limited by secondary drug toxicity (1,2). Good patient management requires establishment of a therapeutic alliance with active patient involvement. Toxic effects of drugs serve as one end point, independent of blood level monitoring, to allow clinical titration to efficacy. Although adverse effects from dose-related central nervous system toxicity of antiepileptic drugs (AEDs) are components of the pharmacologic effects of drugs, it is the unpredictable and dangerous idiosyncratic side effects that are of concern in monitoring patients during long-term treatment.
Idiosyncratic effects of drugs are rare but may be life-threatening. Physicians use scheduled monitoring laboratory studies in the hope of protecting patients against such serious problems, with the expectation of detecting dangerous reactions in time to intervene. Some physicians use regularly scheduled monitoring of drug blood levels along with a program of accumulating hematologic data, routine serum chemistry studies, and urinalysis (3). Pharmaceutical companies and published regulatory materials appear to require this monitoring strategy by incorporating standard recommendations for drug use as published in the Physicians' Desk Reference (4) in the United States and the Compendium of Pharmaceuticals and Specialties (5) in Canada. Although these reference sources appear to define the medicolegal standard of practice for many clinicians, in fact, these documents preserve observations about a specific and well-defined group of patients under close scrutiny during drug trials. Some reports in these documents are amended when data show that specific warnings are needed to protect patients. Contrary to clinical practice, and these publications, scientific criteria based on accumulated evidence fail to support routine monitoring because such archival data rarely predict the occurrence of serious drug reactions. Although habits and practice vary both within the United States and in other countries, routine studies should be obtained at baseline, before starting treatment with a new drug, by measuring biochemical function and structural circulating elements in the blood (6) (Table 10.1).
All the established AEDs and some of the newer drugs (7, 8, 9) have caused serious idiosyncratic drug reactions that do not depend on drug dose and are unpredictable in their occurrence. All organs have been affected, but skin involvement tends to be the most common (Table 10.2). Established AEDs, used in millions of patients, are known to cause agranulocytosis, aplastic anemia, blistering skin rash, hepatic necrosis, allergic dermatitis, serum sickness, and pancreatitis. Newly available drugs used in many fewer patients throughout the world have caused allergic dermatitis and serious skin reactions (Table 10.2). Other than with felbamate (FBM), additional numerous serious reactions have yet to be reported with any credibility with these newer medications.
TABLE 10.1. BASELINE SCREENING STUDIES BEFORE BEGINNING ANTIEPILEPTIC DRUG TREATMENT
TABLE 10.2. IDIOSYNCRATIC REACTIONS TO ANTIEPILEPTIC DRUGS
After a drug is selected for use, the physician must review the relative benefits and risks with the patient and must document, in the patient's record, that this discussion took place. This process of review and information forms the basis for informal informed consent. The patient should be taught the criteria for success and should be reminded about the necessary process of trial and error in drug selection and about the methods for changing drugs. Because common dose-related side effects are used to aid in management, but interfere with treatment, this process should be one of negotiation. The patient must know the nature of the side effects, what must be tolerated, and how the physician will use these side effects in the titration process. Serious, life-threatening, idiosyncratic effects of a selected drug must be reviewed in clear terms, but within the context of rarity. Although the patient must participate in this therapeutic alliance and be ready to communicate should symptoms develop, the physician must identify patients who are without advocates or whose ability to communicate is impaired. These special patients may need the design of a monitoring strategy, a plan not needed by most patients with epilepsy. A program of screening may be useful in some high-risk patients (1).
Clinical monitoring is useful, especially when viewed within the context of the incidence of serious adverse reactions. Although routine monitoring of hepatic function revealed elevation of values in 5% to 15% of patients treated with carbamazepine (CBZ), fewer than 20 patients with significant hepatic complications were reported in the United States from 1978 to 1989 (10). Fewer cases of pancreatitis were reported. Transient leukopenia occurs in up to 12% of adults and children treated with CBZ (11,12). Aplastic anemia or agranulocytosis, unrelated to the aforementioned benign leukopenia, occurs in two per 575,000, with a mortality rate of approximately one in 575,000 treated patients per year (10). Only four of the 65 cases of agranulocytosis or aplastic anemia occurred in children.
Of patients developing exfoliative dermatitis alone or as part of systemic hypersensitivity, blood test abnormalities were not found until patients developed clinical symptoms. Presymptomatic blood studies fail to predict disease development. Test abnormalities such as benign leukopenia or transient hepatic enzyme elevations do not predict the occurrence of life-threatening reactions. A genetic abnormality in arene oxide metabolism may occur in those patients at higher risk of some types of adverse responses, such as hepatitis (13). A screening test for such defects is not available. The data show that routine monitoring, as practiced commonly, does not allow anticipation of life-threatening effects associated with CBZ treatment. Findings for phenytoin (PHT) and phenobarbital (PB) are similar (1).
Assessment of patients developing hepatotoxicity from treatment with valproate (VPA) suggests that the highest risk is in children <2 years old who are being treated with several AEDs. Most fatalities occurred in the first 6 months of treatment, but some patients developed hepatotoxicity up to 2 years after VPA initiation. Children <2 years of age who were receiving polytherapy had a one in 500 to 800 chance of developing fatal VPA hepatotoxicity. Negative predictors were documented. Patients at negligible risk were those >10 years old who were treated with VPA alone and who were free of indication of underlying metabolic or neurologic disorders. Children at intermediate risk were those between age 2 and 10 years and who were receiving monotherapy and all patients requiring polytherapy. The risk of fatal VPA hepatotoxicity continues to decline with increasing age, even in polytherapy, after the first decade of life (14).
Additional risk characteristics include patients with presumed metabolic disorders or with severe epilepsy complicating mental retardation and organic brain disease
(15,16,17,18). Although this pattern of incidence provides useful clinical guidelines, most clinicians consider them too restrictive or insufficiently detailed to allow identification of patients at highest risk (19). This lack of more specific guidelines to identify patients at highest risk for development of VPA hepatotoxicity has caused the use of that drug to be restricted in patients with intractable epilepsy. Further complicating management strategies, routine laboratory monitoring does not predict the development of fulminant and irreversible hepatic failure (20). Some patients progressing to fatal hepatotoxicity never develop abnormalities of specific hepatic function tests. Conversely, abnormalities of serum ammonia, carnitine, fibrinogen, and hepatic function tests have been reported to occur without the presence of clinically significant hepatotoxicity. Drug interactions should be considered as well (21,22). Reporting clinical symptoms and identification of patients at greatest risk of fatal hepatotoxicity are more reliable means for monitoring. Vomiting is the most frequently reported initial symptom in fatal cases (15,16). Combined symptoms of nausea, vomiting, and anorexia occurred in 82% of patients with reported VPA-associated hepatotoxicity, whereas lethargy, drowsiness, and come were reported in 40% (23,24). Although some patients may have reversal of hepatotoxicity by early drug discontinuation, fatalities still result after such prompt action (25). No biochemical markers have been identified to differentiate those patients who survive from those with a fatal outcome (25).
Whether children or adults, most patients reported with fatal hepatotoxicity had neurologic abnormalities, including mental retardation, encephalopathy, and decline of neurologic function. In patients >21 years old, two of four had degenerative disease of the nervous system. One report stated that nine of 16 patients with hepatic fatalities were neurologically abnormal (26). In one series, all patients in the 11- to 20-year age group were neurologically abnormal. In one review, only seven of 26 adults reported with fatal hepatic failure related to VPA were considered to be neurologically normal (27).
The specific biochemical disorders associated with VPA hepatotoxicity include urea cycle defects, organic acidurias, multiple carboxylase deficiency, mitochondrial or respiratory chain dysfunction, cytochrome aa3 deficiency in muscle, pyruvate carboxylase deficiency, and hepatic pyruvate dehydrogenase complex deficiency (brain) (19,28,29). Clinical disorders associated with VPA toxicity include GM1 gangliosidosis type 2, spinocerebellar degeneration, Friedreich's ataxia, Lafora's body disease, Alper's disease, and MERRF (myoclonic epilepsy with ragged red fiber myopathy) syndrome (23). Patients with such disorders must be identified because of their higher risk of VPA hepatotoxicity.
High-risk patients usually are identified by clinical assessment. Medical history, health status at the initiation of AED treatment, and both patient and physician awareness of clinically important symptoms and signs are more likely to suggest the need for further evaluation.
LIMITATIONS OF ROUTINE LABORATORY MONITORING
Two prospective studies evaluated the efficacy of routine blood and urine screening in patients being treated on a long-term basis with AEDs. Camfield et al. (30) performed blood and urine testing in 199 children to evaluate liver, blood, and renal function at initiation of therapy and at 1,3, and 6 months. These investigators repeated the screening studies every 6 months. There were no serious clinical reactions in these patients treated with PB, PHT, CBZ, or VPA. Studies were repeated in 6% because of abnormal but clinically insignificant results, and in two children therapy was discontinued unnecessarily. These investigators concluded that routine monitoring provided no useful information and sometimes led to unnecessary responses. A second study of 662 adults treated with CBZ, PHT, PB, or primidone failed to detect significant laboratory abnormalities during 6 months of monitoring (31). The authors concluded that routine screening was not cost effective or of significant value for asymptomatic patients. Treatment of 480 patients with either CBZ or VPA in a double-blind, controlled trial also demonstrated a lack of usefulness of routine laboratory monitoring (32).
Laboratory standards vary. Certain changes are expected and acceptable for patients undergoing long-term AED therapy. In asymptomatic patients, few significant abnormalities occur at three times the upper limit of normal for hepatic functions tests. Leukocyte counts as low as 2 × 109/L are frequently insignificant and do not, in and of themselves, predict bone marrow suppression. Such changes occur in at least 10% of patients treated with CBZ or PHT, are usually transient, and do not predict the occurrence of aplastic anemia or agranulocytosis. Platelet counts > 100,000 also are usually asymptomatic and do not predict the development of thrombocytopenia.
Regular monitoring of hematology, chemistry, and other routine studies may be most helpful only if the patient is immediately presymptomatic and presenting with abnormal symptoms and signs (33). Thus, regularly scheduled laboratory monitoring for all patients treated with AEDs both is wasteful and does not lead to the desired result of identifying patients at risk of development of life-threatening adverse drug reactions. Camfield et al. (33) estimated that if every patient with epilepsy in North America were tested three times each year for complete blood count, serum amino aspartate, and transaminase levels, the cost would be more than $400,000,000 annually. A modification of recommendations regarding routine monitoring has been suggested by the Canadian Association for Child Neurology (33).
Although obtaining routine, scheduled screening studies is the habit of most clinicians, the key to treatment monitoring is patient and parent education and counseling. All concerned about a patient must be aware of potential complications and the symptoms that may herald the occurrence of an adverse event. Furthermore, physicians must be willing to evaluate patients on an urgent basis when changes suggesting the development of significant adverse drug reactions are reported. Such symptoms include bruising, bleeding, rash, abdominal pain, vomiting, jaundice, lethargy, coma, and deterioration in seizure control. Exacerbation of seizures or marked shortening of the seizure-free interval is a cause both for review of treatment and survey for the presence of adverse drug effects. The development of any of these possibly ominous symptoms is the best indication for repeating laboratory evaluations.
Although the data suggest that routine, scheduled monitoring is neither cost effective nor helpful, the physician must obtain baseline studies before initiation of an AED. Baseline studies are listed in Table 10.1. Review and retention of such pretreatment data in the medical record may identify patients with heretofore unidentified illness and may allow comparison should symptoms develop and laboratory studies need to be repeated.
Because prediction of the occurrence of serious adverse effects is not possible by routine laboratory monitoring, one useful strategy may be to identify high-risk patients. Glauser (7) has attempted to construct “at risk” clinical profiles for some of the AEDs. Profiles rely on reports of patients who developed idiopathic drug reactions and then constructing a profile based on common occurrences among such cohorts. Drugs allowing profiling include FBM, lamotrigine (LMT), and VPA. VPA has a risk profile for hepatotoxicity that is too nonspecific to be of much practical help. Patients at risk of developing hepatic failure during treatment with VPA include those <2 years old who are being treated with several AEDs and who have a known metabolic disease associated with developmental delay (10, 11, 12). Patients fitting the at-risk profile need detailed laboratory screening for the presence of metabolic disorders. Studies suggested (6) include serum lactate, serum pyruvate, serum carnitine, urinary organic acids, and routine hematologic and chemical screening. Prothrombin time and partial thromboplastin time along with arterial blood gases and ammonia are useful as well (Table 10.3).
TABLE 10.3. ASSESSMENT FOR HIGH-RISK PATIENTS TREATED WITH VALPROATE
NEW DRUGS: MONITORING STRATEGIES
As new drugs are developed and are added to the regimen available for treating patients with epilepsy, physicians have an obligation to review source documents about those medications and to devise a strategy for treatment and monitoring. Drug development is performed by treating selected patients. Age ranges are defined, epilepsy syndromes and seizure types are identified, and patients with associated illness or need for concomitant medications are excluded. Studies are performed exposing limited numbers of patients to a drug for a finite period. During such studies, patients undergo intense scrutiny to identify treatment-related symptoms or adverse effects of a drug. These restrictions in study design may fail to uncover drug interactions or the occurrence of serious adverse effects of a drug.
Because data are limited, initiation of treatment with a newly available drug requires special caution. Although the process of informed consent remains informal, patients should be given as much information as possible. Industryproduced materials may prove useful, but the physician should also provide copies of package inserts coupled with material they prepare describing how the drug is to be used and any monitoring strategy planned. Although the guiding principle of monitoring of patients who are treated with established drugs is parsimony in terms of obtaining routine chemical and hematologic studies, based on the knowledge that such monitoring is ineffective in detecting the occurrence of serious adverse events, such is not the case with a newly introduced drug. As with the established drugs, baseline data should be obtained. Communication is still key; the patient must be prepared to contact the physician, and the physician must facilitate that communication. Chemical and hematologic monitoring with use of a new drug may be recommended in the materials a company develops in concert with regulatory functions of the Food and Drug Administration in the United States. Although recommendations may seem excessively conservative, it may be wise to follow those guidelines until a larger experience is obtained and data become available. This admonition is best illustrated by the experience reported in the communications about FBM.
TABLE 10.4. GUIDELINES FOR USE OF FELBAMATE
Serious idiosyncratic reactions to FBM, including aplastic anemia, occur with an incidence of approximately one per 4,000 to 8,000, as compared with an incidence of 2 to 2.5 per million persons in the general population (7,34, 35, 36, 37). The rate of death resulting from aplastic anemia from FBM is more than 20 times the rate associated with CBZ (7,35,36). Some features related to patients developing aplastic anemia during FBM treatment include the occurrence of an immunologic disorder, such as lupus erythematosus, a condition found in 33% of affected patients (7,35,36), a history of cytopenia, occurring in 42% of patients, and a history of an allergic reaction or toxic reaction to another AED, observed in 52% of patients (7,35,36).
Clinical risk profiles for FBM suggest a screening strategy. Although some features, such as being a white woman, are not specific, the occurrence a prior AED allergic reaction, cytopenia, a history of an immune disorder, especially lupus erythematosus, and less than 1 year of treatment are more worrisome. Screening studies to measure the excretion of atopaldehyde and the ratio of monocarbamate metabolites may offer a method for screening of patients treated with FBM. Table 10.4 lists screening guidelines for patients to be treated with FBM.
TABLE 10.5. SUGGESTED GUIDELINES FOR USE OF LAMOTRIGINE
Hepatotoxic effects of FBM seem less clearly associated with risk factors. Of the 18 reported patients, an estimated incidence for all patients was 1 per 18,500 to 25,000, a pattern similar to that found during treatment with VPA. However, detailed review suggested that only seven of these 18 patients actually suffered hepatic injury from FBM, numbers that are more than likely too few to construct a risk profile (7). Because atopaldehyde, an electrophylic and cytotoxic substance, can be formed from the monocarbamate metabolite of FBM, a measure of the ratio of excreted mercapturic acid to carbamoyl derivatives indirectly reflects the ability of a patient's liver to conjugate glutathione to the atopaldehyde (38, 39,40). Kits for evaluating patients are available from the manufacturer of FBM and should be used when treating patients with that drug.
LMT is known to cause serious dermatologic reactions (41). Although erythematous rash with a morbilliform pattern or urticaria and patterns with a maculopapular component are most common (34,41, 42, 43, 44, 45), some patients can develop erythema mulitforme and blistering reactions such as the Stevens-Johnson syndrome or toxic epidermal necrolysis. Simple rashes require careful assessment to ensure that a hypersensitivity syndrome is not developing. Sensitivity reactions include fever, lymphadenopathy, elevated liver enzymes, and altered numbers of circulating cellular elements of blood (41).
In drug trials in the United States, rash was observed in about 10% of patients, with 3.8% having to discontinue and 0.3% hospitalized (41). Most serious rashes developed within 6 weeks of beginning treatment with LMT. In children treated in drug trials, rash was observed in 12.9%, with serious rash in 1.1% with half of those with Stevens-Johnson syndrome (41). More than 80% of patients with complete data developing serious rash were being treated with VPA or had been given doses at a rate higher than recommended (41). Overall, children treated with LMT have a threefold increased risk of developing serious rash compared with adults. Apparently, when specific treatment
guidelines are followed, the incidence of serious rash may possibly be reduced (41,45,46). Table 10.5 lists the suggested treatment plan for LMT use.
RECOMMENDATIONS FOR MANAGEMENT