Most of the available data on AED efficacy and safety in patients with epilepsy and brain tumors originate from uncontrolled studies, and no large-scale well-designed randomized trial has been performed in this population. Therefore, a fully evidence-based approach to AED selection in patients with epilepsy secondary to brain tumors is not possible at the present time. Factors to be considered in making treatment selection include relative efficacy in specific seizure type, tolerability profile, comorbidities, drug-interaction potential, and cost (Perucca & Tomson, 2011). Some of these factors are especially relevant for people with brain tumors, and will be discussed later in more detail.
There is no evidence that the efficacy ranking of currently available AEDs differs for patients with brain tumors compared with patients with focal seizures from other etiologies. Seizure freedom rates in AED-treated patients with brain tumor vary across studies, presumably due to differences in methodology and characteristics of the population assessed. An early retrospective study from the United States found that 21 (72%) of the 29 patients with malignant gliomas who presented with seizures had subsequent (“recurrent”) seizures despite AED therapy (Moots et al., 1995). In a large population of patients with chronic epilepsy from France, however, seizure freedom rate in patients with brain tumors (46%) was similar to that reported for patients with normal magnetic resonance imaging (MRI) findings (42%), and epilepsy secondary to stroke (54%) or vascular malformations (50%), and greater than for patients with posttraumatic gliosis (30%), malformation of cortical development (24%), and hippocampal sclerosis (11%) (Semah et al., 1998). A United Kingdom study that followed mostly newly diagnosed epilepsy patients found that 52% of AED-treated patients with brain tumor were seizure-free, an outcome intermediate between that reported for posttraumatic epilepsy (35%) and that reported for epilepsies associated with cortical malformations (60%), cerebral atrophy (71%), and cerebrovascular disease (70%) (Mohanraj & Brodie, 2005). In a more recent cross-sectional study from Italy, dysembryoplastic neuroepithelial tumours (DNETs)/low-grade gliomas ranked third, after perinatal damage and mesial temporal sclerosis, in terms of probability of being associated with pharmacoresistant seizures (Fig. 1; Gilioli et al., 2012). Most of the estimates reported in the preceding studies were based on relatively small subgroups, and therefore should be regarded only as indicative. In general, secondary generalization seems to be preferentially affected by AED treatment, and can be completely suppressed, even when focal seizures persist (Hildebrand et al., 2005).
Figure 1. Relationship between proportion of patients with pharmacoresistant epilepsy and underlying etiology in a population of 1,115 adults with focal epilepsy who were enrolled consecutively at two epilepsy centers in Italy. Reproduced from Gilioli et al. (2012), with permission.
Download figure to PowerPoint
Adverse effect profile
Limited evidence suggests that patients with brain tumors show increased susceptibility to the adverse effects of AEDs (Glantz et al., 2000; Van Breemen et al., 2007; De Groot et al., 2012; Maschio, 2012). This could be related to the underlying cerebral pathology, or to an interaction with anticancer therapy. In a study of patients with low-grade gliomas, the presence of epilepsy and exposure to AEDs were identified as independent risk factors for cognitive dysfunction, and AED use was associated with a sixfold increase in the prevalence of cognitive side effects compared with side effects of radiation therapy (Klein et al., 2002). The risk of AED-induced skin rashes, including Stevens-Johnson syndrome, is also increased in patients with brain tumors, an effect which, at least for phenytoin, carbamazepine, oxcarbazepine, and phenobarbital, seems to be related at least in part to exposure to cranial radiotherapy (Moots et al., 1995; Micali et al., 1999; Van Breemen et al., 2007; De Groot et al., 2012; Weller et al., 2012).
Interaction potential and outcome studies in patients receiving chemotherapy
The risk of interactions between AEDs and anticancer agents is a major concern. Enzyme-inducing AEDs such as carbamazepine, phenytoin, and barbiturates stimulate the activity of drug-metabolizing enzymes and enhance by this mechanism the metabolic clearance of many concomitantly administered drugs, including corticosteroids and several anticancer medications (Table 1). These interactions may decrease the effectiveness of anticancer therapy, although in the case of chemotherapeutic agents converted to active or toxic metabolites, the possibility of enhanced toxicity should also be considered. Although some anticancer drugs commonly used in primary brain tumors, such as temozolomide, do not seem to be susceptible to enzyme induction, for others, such as the salvage agents irinotecan, etoposide, and tyrosine kinase inhibitors (e.g., erlotinib and imatinib), the increase in clearance caused by enzyme inducers can be considerable, and may need to be compensated by a corresponding increase in the dose of the anticancer agent (Rossetti & Stupp, 2010). The clearance of many cytotoxic agents used to treat brain metastases from solid tumors is also enhanced considerably by enzyme-inducing AEDs (Table 1). Because enzyme induction is a reversible phenomenon, adjustments in the dosage of previously optimized anti-cancer drugs may be required if enzyme-inducing AEDs are discontinued.
Table 1. Anticancer agents
| || |
Nitrosoureas (carmustine, lomustine, fotemustine, nimustine)
Because drug interactions vary depending on the type of antineoplastic regimen used, an adverse impact of enzyme induction on the outcome of cancer therapy has not been shown consistently in all populations. Specifically, although in some types of cancer, decreased survival after exposure to enzyme-inducing AEDs has been clearly shown (Relling et al., 2000), at least for patients with primary brain tumors, evidence is controversial. A retrospective study of 620 patients with glioblastoma enrolled in clinical trials of various (but ineffective) chemotherapeutic agents before the year 2000 found that overall survival and progression-free survival were paradoxically better in patients receiving enzyme-inducing AEDs (71% of the assessed cohort) than in those not exposed to enzyme inducers, the large majority of whom were not on AED therapy (Jaeckle et al., 2009). Although the authors were unable to identify possible bias, this study has been criticized on methodologic grounds (Rossetti & Stupp, 2010). Of interest, two other studies had reported improved outcomes in patients with brain tumor who were exposed to enzyme-inducing AEDs (Reardon et al., 2005; Groves et al., 2006). One possible explanation for these apparently paradoxical findings is that tumors associated with seizures may be diagnosed earlier, and therefore may have a better prognosis for that reason (French, 2006; Maschio, 2012). The difficulties in interpreting retrospective data are well exemplified by an earlier post hoc analysis of outcomes in 168 patients with glioblastoma who were exposed to standard treatments, including surgery, radiation therapy, and chemotherapy, mostly with chloroethylcyclohexylnitrosourea (CCNU) (Oberndorfer et al., 2005). The median survival of patients exposed to enzyme-inducing AEDs (10.8 months, n = 37) was moderately and nonsignificantly reduced compared with patients without seizures not receiving AEDs (11.6 months, n = 75), and was significantly shorter than that reported in patients taking non–enzyme-inducing AEDs, most of whom were on valproic acid (13.7 months, n = 33). Patients taking non–enzyme-inducing AEDs also showed a higher incidence of hematologic toxicity. Whether the lower survival of patients on enzyme inducers compared with patients mostly treated with valproic acid reflected an adverse effect of enzyme induction or a beneficial effect of valproic acid was unclear. In fact, findings from two subsequent studies suggested that valproic acid may indeed favorably influence prognosis in patients with glioblastoma.
Weller et al. (2011) analyzed the survival data of patients with glioblastoma who were enrolled in a randomized trial of radiotherapy alone versus radiotherapy plus temozolomide (Weller et al., 2011). Valproic acid had no apparent effect on survival for patients treated with radiotherapy alone. Conversely, for patients receiving temozolomide chemoradiotherapy, valproic acid appeared to confer greater survival benefits (hazard ratio [HR] 0.39; CI 0.24–0.63, n = 97) compared with treatment with enzyme-inducing AEDs only (HR 0.69, CI 0.53–0.90, n = 257) or no AED treatment (HR 0.67, CI 0.49–0.93, n = 175). Patients receiving valproic acid, however, were more likely to experience grade 3/4 thrombocytopenia and leukopenia than patients in the other groups. The apparent lack of effect of enzyme inducers on survival in this study is consistent with the report that the clearance of temozolomide, a drug cleared by hydrolysis and renal excretion, does not seem to be influenced by enzyme-inducing AEDs (Weller et al., 2012). Conversely, the improved prognosis in patients treated with valproic acid could be explained by valproic acid's ability to cause a minor reduction in temozolomide clearance (Weller et al., 2012) or, more likely, by some pharmacodynamics properties of valproic acid leading to potentiation of temozolomide activity at the site of action (see subsequent text). An apparently beneficial effect of valproic acid on survival in patients with glioblastoma who were treated with temozolomide chemoradiotherapy has been confirmed recently in a Dutch study (Kerkhof et al., 2013) in a population which included, however, some of the patients also evaluated by Weller et al. (2011). After adjusting for age, extent of resection, and O6-methylguanine-DNA methyltransferase (MGMT) promoter methylation status, patients receiving valproic acid with temozolomide for at least 3 months (n = 108) were found to have a median survival of 69 weeks compared with 61 weeks for those receiving other AEDs (mostly levetiracetam, n = 57; HR 0.63; CI 0.43–0.92). Improved survival on valproic acid was also suggested by a study from Brazil that assessed treatment outcomes in children with a variety of malignant brain tumors (Felix et al., 2013). Because all of these studies were retrospective, the possibility of patient selection bias and other bias cannot be excluded, and confirmation in a prospective randomized trial is required.
In laboratory studies, valproic acid has been found to induce apoptosis, autophagy, growth arrest, and cell differentiation of tumor cells through inhibition of histone deacetylase and possibly other mechanisms (De Groot et al., 2012; Weller et al., 2012; You et al., 2012). Inhibition of tumor angiogenesis by valproic acid has also been reported (Osuka et al., 2012). These actions could contribute to inhibit tumor growth, and might explain the apparent prolongation of survival in brain tumor patients treated with valproic acid (Oberndorfer et al., 2005; Weller et al., 2011; Felix et al., 2013; Kerkhof et al., 2013). On the other hand, valproic acid also has intrinsic hematologic toxicity, particularly on platelet function, and inhibits a number of drug-metabolizing enzymes, causing an increase in the serum levels of some anticancer agents. Increased hematologic toxicity has indeed been reported when valproic acid was given in combination with temozolomide (Weller et al., 2011) and with other chemotherapeutic agents, including nitrosoureas, cisplatinum, and etoposide (Bourg et al., 2001; Oberndorfer et al., 2005).
Apart from the effect of AEDs on cancer treatments, a number of interactions have also been reported, whereby anticancer agents can increase or decrease the serum concentrations of AEDs (Van Breemen et al., 2007). To minimize the risk of changes in AED response, serum AED levels should be monitored, if possible, whenever changes are made to concomitant anticancer therapy.