The contents of this publication are solely the responsibility of the authors and do not necessarily reflect the views of the Canine Health Foundation.
Corresponding author: Brian Hardy, DVM, 1365 Gortner Avenue, St. Paul, MN 55108; e-mail: email@example.com.
Status epilepticus (SE) and acute repetitive seizures (ARS) are common canine neurologic emergencies. No evidence-based studies are available to guide treatment in veterinary patients. Parenteral levetiracetam (LEV) has many favorable properties for the emergency treatment of seizures, but its safety and efficacy in dogs for SE and ARS are unknown.
Intravenous LEV is superior to placebo in controlling seizures in dogs with SE or ARS after treatment with IV diazepam.
Nineteen client-owned dogs admitted for SE or ARS.
Randomized, placebo-controlled, double-masked study. Dogs with SE or ARS were randomized to receive IV LEV (30 or 60 mg/kg using an adaptive dose-escalation approach) or placebo, in addition to standard of care treatment. They were monitored for at least 24 hours after admission for additional seizures.
The responder rate (defined as dogs with no additional seizures after administration of the study medication) after LEV was 56% compared with 10% for placebo (P = .06). Dogs in the placebo group required significantly more boluses of diazepam compared with the LEV group (P < .03). Seizure etiologies identified were idiopathic epilepsy (n = 10), inflammatory central nervous system disease (n = 4), intracranial neoplasia (n = 2), hepatic encephalopathy (n = 1), and 2 dogs had no cause determined. No serious adverse effects were attributable to LEV administration.
Conclusions and Clinical Importance
LEV was safe and potentially effective for the treatment of SE and ARS in these client-owned dogs. Larger, controlled clinical trials are needed to confirm this preliminary observation.
Seizure emergencies such as status epilepticus (SE) and acute repetitive seizures (ARS) are common reasons for presentation to veterinary emergency clinics.[1, 2] The definitions of SE and ARS have evolved over the last several decades. The duration of seizure activity for a patient to be considered in SE most frequently is defined as seizure activity lasting 5 minutes or longer. ARS is less well defined, but has been described as 2 or more seizures in a 5–12 hour period, differing from the patient's usual seizure behavior.[3-5] ARS also is commonly referred to as cluster seizures in veterinary medicine. Morbidity and mortality from SE and ARS are high in both humans and dogs. In the largest studies to date in veterinary medicine, the mortality rate was 25.3–38.5%.[6, 7] In humans, SE-associated mortality estimates typically range from 3 to 40%, depending on the population evaluated, but overall mortality is approximately 22%, similar to that seen in dogs. The optimal treatment for SE in humans and dogs is unknown. Despite the large number of cases, few well-controlled studies have been performed evaluating the efficacy of available medications in humans.[10-12] To our knowledge, no published studies exist in dogs.
An injectable formulation of levetiracetam1 (LEV) was approved in 2006 for use as bridge therapy in patients unable to take oral medications. Since then, its off-label use in humans has been reported[13, 14] as has its use in dogs for the treatment of refractory SE, but no prospective, blinded studies have been conducted. The mechanism of action of LEV is not completely understood, but it is thought to act by binding to the synaptic vesicle 2a protein on the presynaptic terminal, and modulating synaptic vesicle fusion and neurotransmitter release.[15, 16] There are a number of other purported mechanisms that may inhibit epileptic activity, including indirect effects on GABAergic neurotransmission, inhibiting the Na+-dependent Cl−/HCO3− exchanger, and modulation of K+ and high-voltage Ca2+ channels. LEV has a favorable pharmacokinetic profile. It is relatively rapidly and extensively absorbed via PO and IM routes; readily crosses the blood-brain barrier; is minimally protein-bound (<10%); is primarily eliminated by renal excretion (66% unchanged in the urine, 24% is metabolized by enzymatic hydrolysis); has no known drug interactions; and follows linear kinetics. A number of studies have shown that LEV is safe in healthy dogs at dosages up to 60 mg/kg IV, but no reports of parenteral LEV use in clinical cases have been published to date. The major limitation to chronic LEV treatment in veterinary patients is that it has a relatively short plasma half-life of 3–4 hours.[18, 19] However, in rats, LEV half-life in the brain and cerebrospinal fluid (CSF) is 1.5–2 times longer than the plasma half-life, and it is thought to maintain higher concentrations in the brain in humans.[20, 21]
The primary aim of this pilot study was to evaluate the efficacy and safety of IV LEV compared with placebo in client-owned dogs with SE or ARS. We hypothesized that LEV treatment would be superior to placebo in stopping seizure activity in dogs with SE or ARS after treatment with IV diazepam.
Materials and Methods
Nineteen client-owned dogs with SE or ARS presented to the Veterinary Medical Center (VMC) of the University of Minnesota (UMN) between October 2007 and March 2010 were included in the study. This study was approved by the UMN Institutional Animal Care and Use Committee (#0905A65361). Informed consent was obtained from all owners before enrollment in the study. Dogs were eligible for inclusion if they had SE or ARS. SE was defined as a single seizure lasting longer than 5 minutes or 2 or more seizures without completely regaining consciousness between seizures. Acute repetitive seizures were defined as 3 or more seizures in a 12-hour period in the 24 hours before presentation. Dogs were excluded if they were azotemic with a urine specific gravity <1.030, hypoglycemic (blood glucose < 60 mg/dL), or hypocalcemic (total Ca < 9.3 mg/dL or ionized Ca < 5.1 mg/dL).
This was a randomized, placebo-controlled, double-masked study. Dogs that presented to the UMN VMC for SE or ARS were eligible for inclusion in the study. Animals were enrolled only if they had another seizure while hospitalized or were actively seizuring on presentation. At the time of an additional seizure in the hospital, dogs were treated with IV diazepam2 (0.5–1 mg/kg) as soon as possible after the in-hospital seizure, and either LEV or placebo also was administered IV no longer than 2 hours later. LEV and placebo were stored in sequentially numbered vials with no other labeling.
The LEV treatment was based on adaptive dose-escalation approach. Dogs were randomized to receive LEV or placebo in permuted blocks of 4. By protocol, the first 10 dogs received 30 mg/kg of LEV (n = 5) or an equivalent volume of 0.9% saline3 (n = 5). Because no adverse effects were seen at the planned interim analysis of the first 10 patients, and a higher dosage (60 mg/kg) was found to be well tolerated in healthy dogs, the dosage was increased to 60 mg/kg LEV (n = 4) or an equivalent volume of 0.9% saline (n = 5) for the last 9 dogs. The solution was administered undiluted as a slow bolus over 5 minutes. Blood samples were collected in EDTA tubes at 15, 45, and 180 minutes after administration of the study solution. Subsequently, plasma was harvested and stored at −20°C until analysis for plasma LEV concentrations. If seizures continued or recurred anytime after administration of the study solution, dogs were treated at the clinicians’ discretion, according to hospital guidelines for the emergency treatment of seizures, which included recommended treatment with a diazepam constant rate infusion (CRI), followed by IV phenobarbital4 (PB) if needed, and later with IV propofol or IV pentobarbital if seizures continued.
The primary endpoint was whether dogs were “responders” or not. A responder was defined as a dog that had no additional seizures after administration of the study solution for the next 24 hours. Secondary endpoints were the number of seizures until 24 hours seizure-free, the number of hours until 24 hours seizure-free, the number of episodes of SE in hospital, the percent all-cause mortality while hospitalized, duration of hospitalization, number of bolus injections of diazepam given, percentage of dogs receiving a CRI of diazepam, duration of diazepam CRI, percentage of dogs receiving IV PB, dose of IV PB, percentage of dogs receiving either propofol or pentobarbital treatment, hours of propofol or pentobarbital treatment, and proportion of dogs that experienced vomiting, diarrhea, ataxia, or decreased alertness.
Before this study, power analysis based on estimated variances indicated that 46 patients would be necessary to detect a 40% difference in responder rate with 80% power. Twenty dogs were planned to be included in this pilot study.
Drug and Pharmacokinetic Analyses
Plasma LEV concentration was measured by high-performance liquid chromatography using a previously described method.[18, 19] Pharmacokinetic parameters were determined by noncompartmental analysis using commercially available software.5 Clearance, distribution volume, and elimination half-life calculations were done by standard computation procedures assuming first-order elimination.
Statistical analysis was performed using a commercially available software program.6 Data were analyzed for normality with the Kolmogorov-Smirnov test, and all data were found to be nonparametric. Fisher's exact test was used to compare the number of responders versus nonresponders, the percentage of all-cause mortality, the percentage of dogs receiving diazepam CRI, the percentage of dogs receiving propofol or pentobarbital treatment, and the percentage of dogs that experienced vomiting, diarrhea, ataxia or decreased alertness. The Wilcoxon rank sum test was used to compare the median number of seizures until 24 hours seizure-free, the number of hours until 24 hours seizure-free, the number of episodes of SE, the duration of hospitalization, the number of bolus injections of diazepam given, the duration of diazepam CRI, and the duration of propofol or pentobarbital treatment between the LEV and placebo groups.
Nineteen dogs were included in this study. Twenty cases were planned, but because of budgetary reasons, the 20th case was not enrolled. Twelve additional dogs initially were eligible and owner consent was obtained, but failed to have subsequent seizure activity in the hospital, and therefore were not enrolled. For the 19 enrolled cases, there were 10 female (7 spayed and 3 intact) and 9 male (7 castrated and 2 intact) dogs. The most common breeds were Boston Terrier and Golden Retriever, with 3 patients of each breed. There were 2 Labrador Retrievers, and 1 of each of the following breeds: American Staffordshire Terrier, Basset Hound, Boxer, English Setter, Great Pyrenees, Jack Russell Terrier, Maltese, Miniature Schnauzer, Old English Sheepdog, Standard Poodle, and mixed breed.
There was no significant difference between the LEV and placebo groups in median age, body weight, number of seizures in the 12 hours before presentation, historical seizure onset, at-home diazepam treatment before presentation or previous PO antiepileptic drug (AED) treatment (Table 1).
Table 1. Comparison of selected pretreatment characteristics between the 2 treatment groups. Data presented as median (range).
LEV, levetiracetam; AED, antiepileptic drug; rDVM, referring veterinarian.
Seizure onset (months prior)
# Seizures in 12 hours before presentation
Previous oral AED treatment
At-home/rDVM diazepam treatment before presentation
Idiopathic epilepsy was the most common diagnosis in both groups (LEV n = 6; placebo n = 4). Idiopathic epilepsy was defined as seizure onset from 1 to 6 years of age, normal neurological status between seizures, normal biochemistry results, and no abnormalities detected on imaging of the brain and in CSF analysis (when available). Two dogs in the placebo group had brain tumors confirmed later (confirmed glioma in 1 dog based on postmortem examination, presumed glioma based on magnetic resonance imaging [MRI] findings in the other dog). Three dogs in the placebo group and 1 dog in the LEV group had inflammatory central nervous system (CNS) disease (1 granulomatous meningoencephalitis, 2 lymphoplasmacytic meningitis, and 1 neuronal necrosis). One dog in the LEV group had hepatic cirrhosis and hepatic encephalopathy. No etiology was determined in 2 dogs, 1 in each group. The dog in the LEV group was a 9-year-old, spayed female Labrador Retriever that had acute onset of seizures and a history of pleural effusion of unknown etiology. No additional diagnostic tests were performed, and the dog had no more seizures and was discharged from the hospital. No follow-up information was available. The dog in the placebo group was a 6-year-old spayed female standard poodle that had no previous seizure history. Additional diagnostic tests were declined by the owners, who elected humane euthanasia after 21 hours of hospitalization. The dog had no additional motor seizures, but continued to vocalize and have abnormal mentation, and necropsy was declined after euthanasia.
Primary and Secondary Outcomes
In this study, 5/9 (56%) dogs in the LEV group responded to treatment, compared with only 1/10 (10%) in the placebo group (P = .06). There was no obvious difference when comparing the number of responders for the 30 and 60 mg/kg LEV doses: 3/5 (60%) and 2/4 (50%), respectively. In all 19 cases, LEV or placebo was administered within 30 minutes after the seizure that completed eligibility for final enrollment in this study. Administration of LEV or placebo occurred a median of 3 hours (range, 0.75–8.25 hours) after admission to the hospital. Survival to discharge was not significantly different between the 2 treatment groups. In the LEV group, 2/9 (22%) dogs died (n = 1) or were euthanized (n = 1), and in the placebo group, 4/10 (40%) dogs were euthanized (P = .6). Dogs in the placebo group received significantly more IV boluses of diazepam (median, 2; range 1–6) than the dogs in the LEV group (median, 0; range, 0–2; P < .03). Of 19 dogs, 18 received PB while hospitalized, but there was no difference in the median dose between the 2 treatment groups. No dogs received pentobarbital or propofol treatment.
All dogs that were euthanized were considered stable but continued to have seizure activity or were moderately sedated because of administration of more than one AED or both, and usually were euthanized for financial reasons. There were no significant differences between groups for any of the other secondary endpoints (Table 2).
Table 2. Comparison of primary and secondary endpoints between treatment groups. Data presented as median (range) or proportion.
CRI, continuous rate infusion; PB, phenobarbital.
# Seizures until 24 hours seizure free
# Hours until 24 hours seizure free
# Episodes of SE
Duration of hospitalization (hours)
Duration of hospitalization (hours) survivors only
28 (22–72; n = 7)
40 (31–48; n = 6)
Initial diazepam dose (mg/kg)
# Diazepam boluses
Received diazepam CRI
Duration of CRI (hours)
10 (4–21; n = 3)
16 (16–16; n = 1)
# Seizures 1st 24 hours
Received IV PB
PB dose 1st 24hours (mg/kg)
Regardless of treatment group, 5/6 (83.3%) dogs that were classified as responders were diagnosed with idiopathic epilepsy. No diagnosis was determined in the 6th dog because of lack of diagnostic evaluation permitted by the owners. In contrast, 5/13 (38.5%) dogs in the nonresponder group were diagnosed with idiopathic epilepsy. There were no significant differences between the median age, weight, number of seizures in the 12 hours before presentation, proportion of dogs receiving previous AED treatment, initial diazepam dose or serum PB or serum KBr concentrations of responders compared with nonresponders (Table 3).
Table 3. Comparison of selected parameters between dogs classified as responders or nonresponders. Data presented as median (range) or proportion.
# Seizures in 12 hours before presentation
Previous AED treatment
Initial diazepam dose (mg/kg)
PB level (μg/mL)
22.0 (17–22.6; n = 3)
29.7 (25.1–31; n = 5)
KBr level (mg/mL)
0.6 (0.5–0.8; n = 3)
1.35 (1.0–1.7; n = 2)
Ataxia was seen in 3/9 LEV-treated and 1/10 placebo-treated dogs, and decreased alertness was seen in 4/9 LEV-treated and 4/10 placebo-treated dogs. All dogs with ataxia also had decreased alertness. All but one of the dogs with ataxia, decreased alertness or both also received PB IV after the study injection. The 1 dog that was not treated with PB was in the LEV group, and had been treated with diazepam CRI, and was in continuous SE for 5 hours despite treatment, until euthanasia. Of the 8 dogs that had decreased alertness, 2 had been treated with long-term PO PB before presentation, 3 had inflammatory CNS disease, and 1 had hepatic encephalopathy.
One dog in each group had both vomiting and diarrhea. The dog in the LEV group was in liver failure and had received a lactulose and neomycin-containing enema before developing diarrhea. The 1 dog in the placebo group was diagnosed with a presumed glioma based on MRI findings, and the cause for its vomiting and diarrhea was not apparent.
The only dog to die in this study was in the LEV group. Death occurred approximately 46 hours after administration of 30 mg/kg LEV. After LEV treatment, this dog also received PB (4 mg/kg/d IV) and diazepam CRI (1.2 mg/kg/h) for 21 hours. The cause of death based on postmortem examination was lymphoplasmacytic meningitis, with vascular and cerebrocortical necrosis.
Mean plasma LEV concentration-time data are shown in Figure 1. At all time points, LEV concentrations after both the 30 and 60 mg/kg doses were within or exceeded the proposed human therapeutic range of 5–45 μg/mL.[23, 24] For the dogs that received the 30 mg/kg dose, the mean ± standard deviation plasma LEV concentrations were 57.0 ± 17.8, 39.0 ± 14.1, and 21.4 ± 5.0 μg/mL at 15, 45, and 180 minutes, respectively. For the dogs that received the 60 mg/kg dose, the mean plasma LEV concentrations were 141.4 ± 56.3, 118.6 ± 61.6, and 61.7 ± 68.0 μg/mL at 15, 45, and 180 minutes, respectively. Plasma LEV concentrations were dose proportional and mean CL, VD, and half-life estimates for the 30 and 60 mg/kg doses were similar: Cl = 2.75 versus 2 mL/min/kg, Vd = 0.5 versus 0.4 L/kg, and half-life = 2.2 versus 2.3 hours.
The treatment of seizure emergencies in dogs is based on the results of clinical experience and uncontrolled case series and trials. Our study represents the 1st randomized, double-masked, placebo-controlled trial for the treatment of SE or ARS in a clinical population. Despite the small sample size, IV LEV in addition to IV diazepam treatment showed a trend toward superiority over placebo and IV diazepam for the treatment of SE or ARS in dogs.
Levetiracetam was well tolerated by the dogs in this study. It was not possible to separate ataxia and decreased alertness as a result of seizure activity and the administration of other drugs commonly known to cause these effects from the potential effects of LEV; therefore, some contribution from LEV is possible. The only dog that died was in the LEV group. However, death was unlikely to have been related to LEV administration, because it occurred 46 hours after administration (10–12 half-lives had elapsed by that point). Furthermore, this dog was diagnosed with inflammatory and necrotizing CNS disease, a seizure etiology commonly associated with death.
There were no serious adverse effects attributable to LEV treatment, and rapid infusion of large doses of undiluted LEV was well tolerated. Based on these results in clinical patients and the very wide therapeutic index of LEV, clinicians may consider IV LEV as adjunctive treatment for the treatment of SE or ARS in dogs, because it appears unlikely to lead to or result in any clinically relevant adverse events, and it is potentially effective.
The pharmacokinetics of parenteral LEV have been evaluated in healthy dogs. LEV doses used in these studies were approximately 20 and 60 mg/kg with maximum plasma concentrations of 30.3 and 254 μg/mL, respectively. These concentrations are difficult to compare directly, however, because LEV infusion rates and initial blood sampling times were different in the 2 studies (LEV infused over 2 minutes, and 2-minute postinjection sampling time, and LEV infused over 5 minutes, and 15-minute postinjection sampling time). The pharmacokinetic parameters obtained in our study are consistent with the results from the studies in healthy animals, except for the half-life of 2.2–2.3 hours in clinical patients, compared with 3–4 hours in healthy dogs. Concurrent, chronic PB administration appears to decrease the plasma half-life of LEV administered PO from 3.43 to 1.73 hours in experimental dogs. Five of the 9 patients in the LEV group had been receiving long-term PB treatment PO before presentation for SE or ARS. Despite this shortened half-life, plasma LEV concentrations are predicted to be within the proposed therapeutic range for approximately 9 hours when a 60 mg/kg dose is administered.
An evidenced-based standard of care for SE and ARS has not been established in veterinary medicine. Benzodiazepines are most commonly used as a bolus for first-line treatment or, for refractory cases, as a CRI. PB typically is the second drug of choice, followed by PO or rectal KBR or IV NaBr. Other treatments that have been recommended for treatment of seizure emergencies in dogs that do not respond to initial treatment include pentobarbital, propofol, phenytoin, isoflurane, and ketamine, but there is not any published data to help differentiate which is most effective.
Guidelines for the treatment of SE or ARS in humans are based on only a few well-controlled studies. The best evidence exists for the use of lorazepam as initial treatment. It has been shown to be superior to phenytoin and placebo, and comparable with diazepam, PB, and combination treatment of phenytoin and diazepam.[10-12] If seizures continue despite benzodiazepine treatment, administration of IV phenytoin or fosphenytoin is commonly recommended. If necessary, IV valproic acid or LEV are recommended as third-line treatment, but no studies have proven their efficacy. If a patients fails to respond to these drugs, treatment with PB, or induction of anesthesia with pentobarbital, propofol, or midazolam, the last text given as a CRI, is recommended. Ketamine also is used in some refractory cases.[27, 29]
Definitions of SE and ARS are not uniformly accepted. SE is commonly defined as a single seizure lasting from 5 to 30 minutes, depending on the publication. The definition of ARS is even more variable than that of SE. Most commonly, a range of increased seizure activity compared to normal for the patient over 5–12 hours has been used in human studies to define ARS.[3-5] However, a duration of increased seizure activity of 24 hours has been used previously. In the current study, ARS was defined as 3 or more seizures in a 12-hour period in the 24 hours before presentation. The goals for the definition used were to maximize the number of eligible cases, without including dogs that were unlikely to continue to seizure. We also acknowledge that in veterinary medicine, owners may live several hours away from a referral institution and that dogs may have spent time at the referring veterinarian's clinic before presentation, prolonging the time from seizure activity to presentation to the University of Minnesota.
The outcome of dogs with SE has been evaluated in 2 large studies. Both studies found that symptomatic epilepsy (ie, seizures caused by a lesion in the brain, most commonly inflammatory CNS disease or brain tumor) is associated with higher mortality.[6, 7] In our study, the placebo group contained 5 dogs with symptomatic epilepsy, compared with only 1 dog in the LEV group (P = .14), which may have confounded the statistical results for this limited number of patients. Three of the 5 dogs in the placebo group were euthanized, and the 1 dog in the LEV group died.
No significant differences were detected between the dogs that were classified as responders, compared with those that did not respond (Table 3). Dogs in the nonresponder group had higher PB concentrations, but this difference was not significant (P = .07). Additionally, dogs appeared less likely to respond if they presented with SE. Only 1/6 dogs in the responder group presented for SE, compared with 6/9 dogs in the nonresponder group presented for SE (P = .33). PB concentrations were not always determined at the time of presentation, and dogs in the nonresponder group may have received additional injectable or PO PB administration before blood collection, because of persistent seizure activity. Partial SE has been correlated with poor outcome in 1 previous study, and with larger patient numbers, the presence of SE may have been associated with a lower response rate. Dogs with idiopathic epilepsy were more often classified as responders (5/6 responders [83.3%] were diagnosed with idiopathic epilepsy versus 5/13 [38.5%] of nonresponders). Dogs with idiopathic epilepsy were shown to have better survival than dogs with symptomatic epilepsy in a large retrospective analysis of SE, and our results further support these findings.
A secondary goal of this study was to further establish the dog as a viable clinical model of human SE and ARS. The canine model is gaining support for many reasons. It is a naturally occurring condition, not an artificially induced model, and the body size of dogs is closer to that of humans, compared with rodent models, making pharmacokinetic data potentially more relevant for developing dosing regimens. Another benefit of the dog model is the ability to obtain informed consent from pet owners. In human SE, it is impossible to obtain consent from a person during a seizure, although patient representatives can provide consent, if available. Exceptions from informed consent for emergency research are possible, but must meet strict criteria. One of these criteria is evidence of a benefit of the therapeutic intervention in animal studies (FDA §50.24). Evidence from studies in dogs could provide a vital intermediate step in transitioning from research in rodent models to applications in humans.
Despite the prospective nature of this study, it had some limitations. Not all cases were managed by the same clinician and therefore there was some variation in treatments received (other than LEV); there was a relatively small number of cases with resulting low power; despite randomization, there was likely a clinically important difference in the seizure etiology between groups; in 2 cases, a final diagnosis was not determined due to client or patient limitations; and finally, EEG data could not be collected, making it impossible to know if there was true cessation of abnormal CNS electrical activity or simply cessation of motor activity (so-called nonconvulsive SE, a common sequela to convulsive SE in humans). Similarly, the small sample size prevented adequate exploration of dose (or concentration)-response relationships.
LEV doses below 100 mg/kg IV have been shown to be well tolerated in healthy dogs (data on file, UCB Pharma). Future areas for research include determining the maximum safe dosage of IV LEV for clinically affected dogs, evaluating timing of treatment relative to use of benzodiazepines or other AEDs, determining efficacy of single versus multiple doses of LEV, and the most effective dosing interval.
The results of this study suggest that LEV is a safe, potentially effective drug for the treatment of SE or ARS in dogs. Additional studies involving a larger number of dogs and use of various LEV dosing regimens are indicated.
We thank Usha Mirshra, MS, for assaying levetiracetam in plasma samples. This study was supported by a grant from the American Kennel Club Canine Health Foundation.
Keppra, UCB Pharma, Brussels, Belgium
Diazepam, Hospira Inc, Lake Forest, IL
0.9% NaCl, Hospira Inc
Phenobarbital, Baxter Healthcare Inc, Deerfield, IL
WinNonlin, version 5.2, Pharsight Corporation, Mountain View, CA