A 63-year-old man with metastatic squamous cell carcinoma of the tonsils to the brain. Patient had two large metastases, one in the right parieto-occipital area (resected, with recurrence) and one in the left frontal area, status post whole brain radiation and chemotherapy. He was admitted to the NICU for change in mental status: drowsy, able to say only “Eeeh” with minimal stimulation, moving all four extremities, but not following commands. No toxic or metabolic reason was present in the work-up. Patient was on phenytoin with therapeutic levels. (a) EEG: nearly continuous triphasic waves over both frontal regions on a theta/delta background. (b) EEG after 2 mg of midazolam IV were administered: marked attenuation of the background, including the triphasic waves. The patient was placed on lorazepam 1 mg po tid. Two days later, his mental status had improved, and he was able to carry some conversation. Despite that, he was transferred to palliative care where he expired 9 days later
Evaluation of Patients with ICU Seizures
Most of the seizures associated with primary or metastatic CNS tumors are of focal onset (Figs. 12.2 and 12.3) with or without secondary generalization. These patients may progress to convulsive status and permanent neurologic damage. Brain tumors are not intrinsic and can lead to seizures associated with increased blood volume, intracranial pressure, and tissue displacement, resulting in cerebral herniation. Posturing in this case has to be differentiated from a seizure. Seizures due to brain tumors must also be differentiated from intermittent episodes of increased intracranial pressure with plateau waves, which cause headache, diplopia and other visual disturbances, fluctuation of mental status, motor deficits, or dystonic or opisthotonic postures.
A 59-year-old man post left frontal oligodendroglioma resection 7 days earlier, and readmitted to the NICU because of significant edema, presents with intermittent episodes of right upper extremity clonic activity lasting for 15–60 s. (a) EEG revealing left frontocentral epileptiform discharges at a frequency of 2–3 Hz progressing to involve the right occipital head region (right side of the epoque). (b) EEG 30 s later: abrupt cessation of spike and slow-wave activity with subsequent attenuation of the record. The patient responded to IV lorazepam 1 mg and extra phenytoin to correct the low levels
A 48-year-old woman admitted for frequent paroxysmal episodes of staring and found to host a lesion on the CT of the head. (a) EEG showing rhythmic sharp waves maximally over the right frontocentral region. The patient was unresponsive with head and eyes turned to the left during this event (b) Gadolinium-enhanced T1-weighted MRI of the head showing a ring-enhancing lesion on the right frontal lobe. The lesion was resected and found to be a metastasis
A patient who has a sudden change of mental status postoperatively after brain tumor resection will need to be evaluated for hemorrhage, edema, infarction, as well as seizures, clinical or subclinical. In parallel with a head CT, MRI, and the appropriate workup for other common critical care causes of encephalopathy (see above), an EEG will confirm whether the patient is having nonconvulsive seizure activity and may also help in assessing the appropriate response to treatment. In a study of 102 patients with meningioma resection, Rothoerl et al. reported normal preoperative 30-minute EEGs in 49% and normal postoperative EEGs in 33.3%. Thirty-two percent of patients had preoperative and 15% postoperative seizures. Of those with preoperative seizures, 53% had complete seizure resolution postoperatively. Dominant hemispheric localization and pre-or postoperative headache were associated with postoperative seizures. Interestingly, the pre- or postoperative EEG findings were not associated with postoperative seizures in this series . This may be due to the short period of EEG recording, which may have missed significant abnormalities more easily picked up on a longer or continuous EEG. The role of continuous EEG (cEEG) monitoring in this ICU population has not been well established, but there is growing evidence of its utility in diagnosing NCSE in tumor patients. Jordan monitored 124 NICU patients with cEEG and reported that 34% of them had nonconvulsive seizures and 27% were in NCSE. Among the 11 patients with brain tumors, six (54%) had nonconvulsive seizures. Overall, cEEG played a decisive or contributing role in the ICU management in 81% of brain tumor patients in a later report with additional patients by the same author [70, 71]. NCSE was reported in two patients with non-Hodgkin’s lymphoma presenting with mutism and confusional state after ifosfamide (an alkylating agent, structurally an isomer of cyclophosphamide) infusion . Another patient with glioblastoma multiforme was treated with IV tirapazamine and brain irradiation. After CT scan was performed with intravenous contrast medium, the patient became aphasic, and the EEG showed NCSE. IV lorazepam and a loading dose of phenytoin resolved the symptoms . In the aforementioned study by Marcuse, out of 259 patients with brain tumors and an EEG, 24 (9.2%) had NCSE. Because 13 patients had only subclinical seizures and the vast majority of seizures captured in the rest were also subclinical, these patients would have never been diagnosed and treated without an EEG or CEEG .
If an EEG is not considered and an MRI is performed in a patient with cancer and mental status change, abnormalities that may be attributed to seizures can be found. These may alarm the intensivist and after an EEG is performed contribute to the correct diagnosis. This is the case of four patients with primary or metastatic brain tumors from Memorial Sloan Kettering Cancer Center, who had been intermittently confused or unresponsive 1–7 days before the MRI: the MRI showed cortical hyperintensity on FLAIR, T2-weighted, or diffusion-weighted images, with or without leptomeningeal enhancement on T1 with gadolinium. In two of these patients who had 18 F positron emission tomography, hypermetabolism was shown in the abnormal cortical MRI locations. All patients and another eight were in NCSE on the EEG, and all except for one improved clinically after receiving antiepileptic drugs. Repeat MRI 1–4 weeks later showed complete resolution of abnormalities in three patients and improvement in the fourth patient . These results emphasize the need for electroencephalographic emergent evaluation of tumor patients with unexplained change in the neurological examination in the ICU.
Prophylactic Administration of AEDs
The issue of prophylactic treatment of patients with brain tumors is very complex. If a seizure has already occurred, there is little doubt for the value of AEDs  to avoid escalation to more refractory seizures or SE, but when the patient has never exhibited epileptic phenomena, such a treatment becomes more controversial. Efficacy of the treatment has to be balanced with adverse events associated with the chosen drugs. Despite the best efforts, a significant percentage of patients still have breakthrough seizures, and the response to the treatment is very unpredictable. Several reasons have to be considered: lack of AEDs to have an effect on a vast array of physiologic derangements induced by brain tumors, difficulty maintaining appropriate AEDs levels , and tumor progression or recurrence [26, 75]. In fact, in a recent retrospective study of postoperative patients with brain tumors, the odds of seizure for patients on prophylactic AED was 1.62 times higher than those not on AED, although the difference was not significant . Likewise, AED use is not without adverse effects, some of them potentially serious, like severe Stevens-Johnson syndrome . Moreover, there is evidence supporting increased frequency and severity of side effects from these drugs in this specific patient population: in a meta-analysis of studies examining prophylactic AED use in patients with newly diagnosed brain tumors, 23.8% (range 5–38%) of treated patients experienced side effects that were severe enough to lead to change or discontinuation of the medications. This incidence is higher than that in the general population and should make physicians sceptic about the real need for using them . Unfortunately, personal preference and previous training or experience of physicians may be more important in making the decision than clinical evidence for pros and cons. According to a study conducted in Rhode Island, 55% of participating physicians gave AED prophylaxis, but the percentage differed according to the subspecialty: 33% of radiation oncologists, 50% of oncologists, 53% of neurologists, and 81% of neurosurgeons [74, 78].
The effect of surgery on seizures has been studied in numerous trials. The effect of craniotomy per se, with the meningeal or parenchymal injury that ensues, on seizure occurrence cannot be easily separated from the very effect the tumor induces. Most of the available studies were performed in mixed tumor and non-tumor patients; therefore, the conclusions may not be applicable to the former. In the next sections, we will review some of the most important studies regarding craniotomy, all including tumor patients, because these data are pertinent to the decisions that an intensivist has to make. Subsequently, we will present the data regarding prophylactic AED use specifically in patients with brain tumors.
Kvam et al. showed that out of 538 post-craniotomy patients, 23 had postoperative seizures. Out of these 23 patients, only 5 had seizures preoperatively. The authors suggested a preoperative loading dose of 10 mg/kg of phenytoin, followed by a postoperative dose of 5 mg/kg/day . A study more pertinent to the ICU was conducted in Taiwan . Three hundred seventy-four patients post-craniotomy were randomized to receive phenytoin (15 mg/kg IV during surgery, followed by 3–6 mg/kg/day for 3 days) or placebo. The group receiving phenytoin had two early postoperative seizures, and the placebo group had nine, but the difference was not statistically significant. Eighty percent of the seizures occurred within 20 min after surgery. Thus, the authors recommended that prophylactic anticonvulsant medication be given at least 20 min before completion of wound closure. This view was not shared by the authors of a subsequent large prospective study, who did not recommend prophylactic AEDs after supratentorial craniotomy. In this study, 276 post-craniotomy patients were randomized to receive carbamazepine or phenytoin for 6 or 24 months or no treatment . The three treatment groups did not overall differ in the risk of seizures, but there was a nonsignificant 10% reduction of seizures in the two groups which received AEDs. Meningiomas had the highest risk for seizures (75% by 4 years) and pituitary tumors the lowest (21% by 4 years). Longer operations, those associated with dissection of the lesion away from the surface of the brain, and left-sided or bilateral lesions also carried a higher risk. Early seizures (within 1 week) after craniotomy did not increase the likelihood of late epilepsy. In a systematic review of seizures and meningiomas, however, no difference in the rate of new postoperative seizures was observed with or without perioperative prophylactic anticonvulsants . Adding to the debate are the results of a prospective, stratified, randomized, double-blind Dutch study that compared 300 mg phenytoin/day to 1500 mg valproate/day given for 1 year in 100 post-craniotomy patients. Fourteen patients had postoperative seizures, but there was no difference in seizure incidence between the two groups . Finally, a meta-analysis of six controlled studies addressing the issue showed a tendency of prophylactic AEDs to prevent postoperative convulsions in patients without preexisting seizures, but this effect did not reach statistical significance .
Several studies have examined the need for AED use, either prophylactically or after surgery, usually in mixed primary or metastatic brain tumor populations. In a double-blind, randomized study of phenytoin (100 mg tid) vs placebo in 281 post-craniotomy patients, the phenytoin group had significantly fewer seizures (12.9% vs 18.4%), and highest protection was present between days 7 and 72. However, the subgroup analysis of 81 patients with brain tumors and craniotomy showed that 21% of patients treated with phenytoin had seizures versus only 13% of nontreated (odds ratio 1.8, 95% CI 0.6–6.1). Only the meningioma subgroup in this study had slightly lower risk for seizures in the treated versus placebo patients. Therefore, based on these results, the recommendations for phenytoin prophylaxis should not apply to brain tumors .
In a subsequent Italian study, 65/128 (51%) patients with supratentorial brain tumors had preoperative seizures and were treated with AEDs. Those without preoperative seizures were randomized to receive phenobarbital or phenytoin as prophylactic treatment or no treatment. No significant difference in seizure incidence was found between patients treated (7%) and those not treated (18%). The authors suggested short-term preventive antiepileptic treatment after surgery in patients without preoperative seizures and continuation of postoperative treatment in patients with preoperative epilepsy .
Other AEDs have also been used. Glantz et al. conducted a well-designed randomized, double-blind, placebo-controlled study comparing the incidence of first seizures in 74 valproate versus placebo-treated patients with newly diagnosed supratentorial brain tumors. The drug and placebo groups did not differ significantly in the incidence of seizures (35% in the valproate and, surprisingly, 24% in the placebo-treated group). Based on these results, no prophylactic treatment with valproate could be recommended .
Finally, a prospective, randomized, unblinded study from Canada examined the effect of prophylactic phenytoin administration in newly diagnosed patients with primary and metastatic brain tumors without prior seizures. Seizures occurred in 26% of all patients, 24% in the treated, and 28% in the nontreated group (odds ratio 0.82, 95% CI 0.3–2) .
Based on the aforementioned evidence, there is no clear benefit of prophylactic use of AEDs perioperatively . However, a short-term perioperative course even of enzyme inducers (e.g., one dose at the end of the surgery; it takes 1–2 weeks of therapy to develop enzyme induction) or a short-term course of the AEDs (which have a safer profile) may be a reasonable research endeavor for the future.
Similarly, reports on patients exclusively with metastatic brain tumors do not support the use of prophylactic anticonvulsants [10, 87]. In a large retrospective analysis of 195 patients with metastatic brain tumors, Cohen et al. reported that 18% of patients presented with seizures. Of the remaining seizure-free patients, 40% were treated prophylactically with AEDs (phenytoin in >90%). During a follow-up period of up to 59 weeks, 10% of patients developed late seizures. The incidence of seizures did not differ between treated (13.1%) and untreated (11.1%) groups. However, this study is flawed due to the fact that two thirds of patients with seizures had subtherapeutic AED levels. The authors did not advocate AED use, unless the patient has the first seizure . This is in accord to a more recent meta-analysis of adult patients with metastatic tumors without a seizure ever, where prophylactic AED treatment was not recommended .
Likewise, a meta-analysis evaluated five trials with specific inclusion criteria (patients with a neoplasm, either primary glial tumors, cerebral metastases, or meningiomas, but no history of epilepsy) who were randomized to either an AED or placebo. The three AEDs studied were phenobarbital, phenytoin, and valproic acid . This meta-analysis confirmed the lack of antiepileptic benefit at 1 week and at 6 months of follow-up. In addition, the AEDs had no effect on seizure prevention for specific tumor pathology .
Summarizing the above information, the Quality Standards Subcommittee of the American Academy of Neurology published a meta-analysis of 12 studies, which had addressed the issue of prophylactic antiepileptic treatment for newly diagnosed brain tumor patients. Four were randomized and eight were cohorts. Only one study showed significant difference between treated and untreated groups and, actually, favored the untreated. The overall odds ratio from the randomized trials was 1.09, 95% CI 0.63—1.89 (P = 0.8) for seizure incidence and 1.03, 0.74–1.44 (P = 0.9) for seizure-free survival. Therefore, the subcommittee recommended no prophylactic use of AED on patients with newly diagnosed brain tumors. Tapering and discontinuing the AEDs was appropriate after the first postoperative week in those patients without a seizure (who were, nevertheless, treated before). Although not excluding the possibility that some subgroups of brain tumor patients may be at a higher risk for seizures (melanoma, hemorrhagic or multiple metastatic lesions, tumors located near the Rolandic fissure, slow-growing primary brain tumors), the subcommittee did not find any reason for prophylaxis in those patients either . This guideline has been retired by the AAN Board of Directors on June 4, 2012, but until a new guideline is published, one should consider this evidence still as the best available, especially since newer studies have not disputed its recommendations [86, 90]. Likewise, these recommendations extend to secondary brain tumors.
In a systematic review of adult patients with solid metastases, never having experienced a seizure due to their metastatic brain disease, routine prophylactic use of anticonvulsants was not recommended . As with primary brain tumors, however, some subgroups of metastatic tumors may have higher incidence of seizures and may benefit from AEDs. For example, in a retrospective study of 105 patients with brain tumors using susceptibility-weighted MRI to detect hemosiderin deposition preoperatively, Roelcke et al. found a significant correlation between cortical hemosiderin deposition and the presence of seizures in the subgroup of patients with brain metastasis . This finding has not been replicated in longitudinal studies; though, neither the effect of any AED in this subgroup is known.
How often these guidelines are followed is questionable, with some data showing that there is a discordance between the recommendations and the current practice. In a recent study from Brazil, for example, 70.2% of seizure-naïve patients with primary brain tumors had received primary prophylaxis with AEDs .
Treatment of Seizures in the ICU
Treatment of seizures or SE in patients with brain tumors follows the general guidelines that are presented in the chapter Management of Status Epilepticus and Critical Care Seizures. There are, however, several important details regarding these complex patients that the intensivist should master.
Firstly, one should not forget that seizure control may be influenced by the evolution of the brain tumor and its treatment . Secondly, surgery may be a potent treatment modality in patients with refractory epilepsy and brain tumors, because studies have shown that resection of the epileptogenic zone due to brain tumors may lead to seizure freedom or significant control of seizures in 56–90% of patients [35, 93–96].
Thirdly, interactions between the various medications are a major problem and can lead to unforeseen complications. AEDs, especially those affecting the cytochrome P450 system, may affect the metabolism of chemotherapeutic agents used for the treatment of metastatic or primary brain tumors (Table 12.1). These agents have a narrow therapeutic window and real potential for toxicity or lethal side effects, if their level is increased by an additional agent or to lose their anticancer efficacy and reduce the chance for remission, if their level is decreased. Usually, the addition of phenytoin, carbamazepine, phenobarbital, and other inducer AEDs reduces the levels or efficacy of cyclophosphamide, methotrexate, adriamycin, nitrosoureas, paclitaxel, etoposide, topotecan, irinotecan, thiotepa, and corticosteroids [74, 92, 97]. Therefore, when these inducing agents are used, the chemotherapeutic agents’ dosage should be increased. Conversely, when these AEDs are stopped, and since induction is a reversible phenomenon, the anticancer agent dose should be decreased . Oxcarbazepine has lower interaction potential, but can reduce the levels of anticancer drugs, such as imatinib . For the ICU, lamotrigine, topiramate, and zonisamide, lacking parenteral formulations and requiring slow-dose titration remain in a disadvantage . Valproic acid, being an inhibitor, can have the opposite effect and increase the chemotherapeutic agents’ levels and lead to higher toxicity from these agents or myelosuppression . There are also some data showing increased postoperative bleeding with valproic acid and that makes some neurosurgeons reluctant to operate with this drug on board [92, 100]. This negative effect of valproic acid on platelets, however, has not been confirmed in subsequent analyses. In a study of 35 patients with glioblastoma, platelet count <100.000/mm3 was only associated with accumulated temozolomide and not independently with valproic acid . More recently, no hematologic toxicity could be proven in patients with glioblastoma multiforme treated with radiochemotherapy and use of levetiracetam or valproic acid . Interaction between valproic acid and warfarin with resulting elevation of the international normalized ratio and bleeding at the tumor bed has been reported in glioblastoma multiforme and requires caution when both drugs are coadministered . On the other hand, valproic acid may also have beneficial effects in patients with seizures due to brain tumors via direct or indirect antitumor properties (see below) and decreased refractoriness to seizure control via suppression of the multidrug resistance gene MDR1 .
Hepatic cytochrome P system used
CYP3A4, CYOC19, CYP2D6
80–90% renally excreted unchanged
CYP3A4 weak inducer
Fourthly, competition for binding to plasma proteins may be important with several of those medications, especially in states of hypoalbuminemia, not uncommon in the ICU or during chemotherapy. Measuring the free levels of drugs and adjusting the dose can be useful in order to avoid toxicity or subtherapeutic levels.
Lastly, regarding the steroids , either their dose should be increased or the patient should be switched to one of the newer, non-inducing AEDs (lamotrigine, levetiracetam, zonisamide, vigabatrin, and gabapentin). These newer drugs are either renally excreted (levetiracetam, zonisamide, gabapentin, vigabatrin) or, if hepatically metabolized, either non-inducers of the cytochrome P system (lamotrigine) or mild inducers (oxcarbazepine) [30, 105].
Overall, the intensivist should be cautious because, contrary to the aforementioned data, there have also been studies showing improved outcomes in patients with brain tumors exposed to enzyme-inducing AEDs [106–108]. One possible explanation is that patients with brain tumors and seizures are diagnosed earlier and therefore may have a better prognosis because of that reason [43, 92].
On the other hand, enzyme inhibitors, such as valproic acid, may lead to better outcomes in these patients. Weller et al. analyzed the survival data of patients with glioblastoma enrolled in a randomized study and treated with radiotherapy alone versus radiotherapy plus temozolomide. Patients receiving valproic acid alone (97 [16.9%]) appeared to derive more survival benefit from temozolomide/radiotherapy (hazard ratio [HR], 95% confidence intervals 0.39, 0.24–0.63) than patients receiving an enzyme-inducing AED only (HR 0.69, 0.53–0.90) or patients not receiving any AED (HR 0.67, 0.49–0.93). Valproic acid did not confer any survival advantage in the radiotherapy-alone arm and was more likely to induce thrombocytopenia and leukopenia . Potential explanations of this effect are either decreased clearance of temozolomide by valproic acid-induced enzymatic inhibition or direct potentiation by valproic of the temozolomide effect . These positive effects of the combination of temozolomide and valproic acid on survival were also found in a more recent Dutch study of patients with glioblastoma  and in a Brazilian study of children with a variety of brain tumors . In a recent systematic review of valproic acid in patients with glioblastoma, a prolonged survival was confirmed .
Chemotherapy per se may have additional effects on seizure control. Newer chemotherapeutic agents, such as temozolomide, may decrease seizure frequency in 50–60% or lead to seizure freedom in 20–40% of treated patients, but the mechanism is unclear [111, 112]. In a retrospective study, seizure frequency in patients with low-grade gliomas and intractable epilepsy was analyzed before and after treatment with temozolomide in 69 patients. There was a significant difference in >50% seizure reduction frequency between patients receiving temozolomide (59%) and those who did not receive temozolomide (13%). Seven patients (18%) in the temozolomide group displayed this improvement independent of AED adjustment compared with no patient in the control group, a significant difference .
Newer AEDs are believed to have a more favorable safety profile; fewer interactions with other drugs and some are also available in an IV formulation, all useful characteristics for their use in the ICU. Levetiracetam belongs to this group of newer AEDs and has a favorable profile for use in patients with brain tumors .There are data that the addition of levetiracetam (1–4 g/day) to older AEDs in patients with refractory seizures can lead to reduction of seizures by 65–90%. In fact, 44–46% of patients in these small case series (a total 86 patients included) were switched to levetiracetam monotherapy later on [115–117]. In a prospective small study of 17 patients with brain tumors (mostly glioblastoma multiforme), levetiracetam monotherapy achieved ≥50% seizure reduction in 91.7% of them. A total of 92 drug interactions were avoided by using levetiracetam (instead of using phenytoin as a comparison), with dexamethasone, acetaminophen, and fentanyl being the most common interacting drugs . In another older small study of 14 patients with brain tumors, addition of gabapentin (0.3–2.4 g/day) to phenytoin, carbamazepine, or clobazam led to 100% seizure reduction and 57% seizure freedom . More recently, lacosamide , a newer AED with minimal drug interactions and renal excretion, had been used in a retrospective study of patients with primary brain tumors. This drug was an add-on for recurrent seizures in 74% of patients and used due to previous drug toxicity in another 23%. Lacosamide, in daily doses ranging from 100 to 600 mg, led to decreased seizure frequency in 66% and stable seizure frequency in 30% of patients. Seizure frequency also decreased in 4 out of 12 (33.3%) patients where lacosamide was not used concomitantly with any other AED. No toxicity was seen in 77% of patients . In another study of 14 patients with brain tumor-associated seizures, who had already been treated with other AEDs and who had not experienced adequate seizure control, the mean seizure number in the last month prior to the introduction of lacosamide was 15.4. After the introduction of this drug, the mean seizure number was reduced to 1.9/month. Lacosamide mean dosage was of 332.1 mg/day (min 100 max 400 mg/day) and the overall responder rate was 78.6%. Only one patient discontinued the drug because of side effects (dizziness and blurred vision) .
Chemotherapeutic agents or corticosteroids can also affect the metabolism of several AEDs, increasing or decreasing their levels [122–124]. This may explain the subtherapeutic levels of these drugs in the studies that evaluated their prophylactic use [10, 78, 84, 85]. Phenytoin concentrations may become toxic after withdrawal of dexamethasone , probably due to slower hepatic metabolism of the former. Poor seizure control may result from the combinations of phenytoin with cisplatin or corticosteroids and valproic acid with methotrexate. Increased toxicity of AEDs can occur when phenytoin is combined with 5-fluorouracil . In addition, several chemotherapeutic agents may have pro-convulsant activity on their own [38, 72] (see chapter Drugs Used for the Critically Ill and Critical Care Seizures), and the intensivist should be aware that the aforementioned studies of prophylactic AED use did not control for the presence of specific chemotherapeutic agents.
An interesting aspect of the antiepileptic drug use in patients with cancer is their potential for antineoplastic or immunosuppressive effect [92, 99, 127–130]. Valproate is exhibiting inherent antitumor activity through inhibition of histone deacetylase (which leads to cell differentiation, growth arrest, and apoptosis of cancer cells, including gliomas) [42, 96, 110, 126, 131] and tumor angiogenesis .