Address correspondence to Wolfgang Löscher, Department of Pharmacology, Toxicology and Pharmacy, University of Veterinary Medicine, Bünteweg 17, D-30559 Hannover, Germany. E-mail: email@example.com
In humans, traumatic brain injury (TBI) is one of the most common causes of acquired (symptomatic) epilepsy, but as yet there is no treatment to prevent the development of epilepsy after TBI. Animal models of posttraumatic epilepsy (PTE) are important to characterize epileptogenic mechanisms of TBI and to identify clinically effective antiepileptogenic treatments. The prevalence and phenomenology of naturally occurring canine epilepsy are similar to those in human epilepsy. However, the risk of epilepsy after TBI has not been systemically studied in dogs. We therefore performed a large retrospective study in 1,000 dogs referred to our clinical department over a period of 11.5 years with the aim to determine the incidence of early and late seizures after head trauma in this species.
Two strategies were used: in group I (n = 392), we evaluated whether dogs referred for the treatment of a head trauma (group Ia) or other trauma (group Ib) developed seizures after the trauma, whereas in group II (n = 608) we evaluated whether dogs referred for the treatment of recurrent epileptic seizures had a history of head trauma. Data for this study were obtained from our clinical database, questionnaires sent to the dogs' owners, and owner interviews.
In group Ia, 6.6% of the dogs developed PTE, which was significantly different from group Ib (1.9%), indicating that head trauma increased the risk of developing epilepsy by a factor of 3.4. The risk of PTE increased with severity of TBI; 14.3% of the dogs with skull fracture developed PTE. In group II, 15.5% of the dogs with epilepsy had a history of head injury, which was significantly higher than the incidence of PTE determined for group Ia.
Our study indicates that head trauma in dogs is associated with a significant risk of developing epilepsy. Therefore, dogs with severe TBI are an interesting natural model of PTE that provides a novel translational platform for studies on human PTE.
Traumatic brain injury (TBI) is a major cause of epilepsy (Lowenstein, 2009). Epidemiologic studies have found that posttraumatic epilepsy (PTE) accounts for approximately 20% of symptomatic epilepsy in the general population, and 5% of all patients seen at specialized epilepsy centers (Agrawal et al., 2006). The main, critical determinant of the development of PTE is the severity of the head injury (Lowenstein, 2009). Key risk factors include skull fracture, intracranial hematoma, and a depressed level of consciousness at the time of admission to the emergency department. In addition, the occurrence of seizures within the first week after TBI (i.e., “early” seizures, a type of acute symptomatic seizures) also appears to be a risk factor for the later development of epilepsy (Lowenstein, 2009). In general, the incidence of PTE varies with the population studied and length of follow-up after injury, as well as the spectrum of severity of the inciting injuries, and has been reported to be anywhere from 1.8% to 53% (Frey, 2003; Lowenstein, 2009).
The mechanisms underlying development of PTE are only incompletely understood (D'Ambrosio & Perucca, 2004; Pitkänen et al., 2009). Furthermore, there is a lack of reliable biomarkers that allow predicting which patients develop epilepsy after TBI (Lowenstein, 2009). There is often a delay of months to years in the emergence of epilepsy after the initial injury, which creates an exceptional opportunity for intervention with antiepileptogenesis therapies once they have been developed (Dichter, 2009; Lowenstein, 2009; Löscher & Brandt, 2010; Pitkänen & Lukasiuk, 2011). The development of animal models that closely mimic human PTE may facilitate efforts to characterize relevant epileptogenic mechanisms and to identify clinically effective antiepileptogenic treatments (D'Ambrosio & Perucca, 2004; Kharatishvili & Pitkänen, 2010).
The most commonly used models of PTE are fluid percussion–induced brain injury in rats, the controlled cortical impact brain injury model in rats and mice, and the weight-drop model of TBI in rats (Pitkänen et al., 2009; Kharatishvili & Pitkänen, 2010). However, none of these models recapitulates fully the human syndrome (Pitkänen et al., 2009), and the use of small animals in these models is associated with various problems that complicate translation of experimental findings to humans (Morganti-Kossmann et al., 2010). In the 1980s, several large animal models of TBI have been established, including induction of TBI in cats and dogs (Morganti-Kossmann et al., 2010), but to our knowledge, these models have not been used for studying PTE. More recently, naturally occurring canine epilepsy has been proposed as a translational platform for human therapeutic trials on new epilepsy treatments (Davis et al., 2011; Leppik et al., 2011). This idea is not new but has been proposed by us almost 30 years ago (Löscher & Meldrum, 1984; Löscher et al., 1985). The prevalence and phenomenology of naturally occurring canine epilepsy are similar to those in human epilepsy (Holliday et al., 1970; Löscher et al., 1985; Podell, 1996; Berendt & Gram, 1999; Berendt et al., 1999).
As in humans, epilepsy in dogs may result from various brain insults, including brain tumors, encephalitis, congenital malformations, and vascular disorders (Croft, 1965; Koestner et al., 1989; Pakozdy et al., 2008). Furthermore, it has been repeatedly suggested that epilepsy in dogs may also be induced by head trauma (Holliday et al., 1970; Cunningham, 1971; Fenner, 1981; Koestner, 1989; Pakozdy et al., 2008; Hayes, 2009), but no systematic study on this possibility has been performed. Based on our long-lasting experience with epilepsy in dogs (Schwartz-Porsche et al., 1982, 1985; Löscher et al., 1985, 2004; von Klopmann et al., 2007; Stein et al., 2012), we decided to carry out a large retrospective study in some 1,000 dogs referred to our clinical department over a period of 11.5 years with the aim to determine the incidence of early and late seizures after head trauma in this species. The findings presented in this study suggest that dogs are an interesting natural model of PTE that provides a novel translational platform for studies on human PTE.
Materials and Methods
Study population/Data collection
The clinical database at the department of small animal medicine and surgery at our university for the period between January 1998 and May 2009 was searched for dogs referred for the treatment of a trauma (n = 609; group I) or seizures (n = 734; group II) (Fig. 1). The patient records of these 1,343 dogs of diverse breeds were reviewed. The two main questions of the present study were (1) whether animals with a head trauma (group I) developed seizures later in their life, and (2) if an animal with recurrent seizures (group II) had any kind of head trauma before it was transferred to our clinic because of seizures. For selection of trauma patients (group I), emphasis was on car accidents, falls, head bites, horse boots, and bullet wounds, which were likely associated with head trauma. For seizure patients (group II), animals with recurrent seizures were chosen. For searching trauma and seizure patients in our database, the clinic program Anidata (Comitas Software, Leipzig, Germany) was used. Dogs with seizures before trauma occurred were excluded. Dogs that died or were euthanized (because of the severity of their symptoms) shortly (within 1 week) after head trauma were not excluded but were used for calculating the incidence of early posttraumatic seizures (see 'Results'). Almost all dogs that survived the first week after head trauma also survived the subsequent follow-up period.
In addition to this review of our database, two standardized questionnaires (one for trauma patients, one for seizure patients) with 21 questions each were designed and sent to the owners of the dogs. For group I, the aim of the questionnaire was to gather information about trauma history, trauma type, trauma induced diseases, whether early (within 1 week after injury) or late (>1 week after injury) seizures occurred after trauma, interval between trauma and seizure onset, and causes of death. Dogs that had two or more unprovoked late seizures were considered epileptic. For group II, the aim was to gather information regarding seizure cause, particularly a history of head trauma, seizure onset, seizure frequency, seizure type (partial, generalized, secondarily generalized) and duration, occurrence of the first and the last seizure, and of cluster seizures or status epilepticus, and remission of epilepsy. Furthermore, all dog owners were asked for diagnostic test results, medical treatment, and other diseases of their dogs. The average time span between trauma and sending out the questionnaire was 4.5 years (group I); respective time span between onset of recurrent seizures and sending out the questionnaire was 3.5 years (group II). Owners not responding to the questionnaire were contacted by telephone and interviewed. If owners not responded to the questionnaire and could not be contacted by telephone, we examined whether our clinical database contained enough data to keep the dog in the study. As shown in Fig. 1, overall 1,000 of the initially 1,343 dogs could be finally analyzed.
All dogs included in this study were neurologically examined in our clinical department. In trauma patients (group I), neurologic examination included type, localization, and severity of the head injury, level of consciousness, gait, spinal cord reflexes, proprioceptive and cranial nerve deficits, ophthalmologic examination, and findings from magnetic resonance imaging (MRI). In 156 dogs with trauma of group I, clinical examination and owners' responses indicated that the head was not involved in the trauma but rather the pelvis region or hind legs. These dogs were assigned to group Ib and used as controls for comparison with dogs with head trauma (group Ia). All head trauma patients were hospitalized and received standard treatment to stabilize the patient immediately after the injury, and, if needed, emergency surgery (cf., Guha, 2004).
In patients with recurrent seizures (group II), a similar neurologic examination and a general physical examination, including blood, cerebrospinal fluid, and urine analysis, were performed. MRI scans were performed in 293 of the 608 dogs. An animal was considered to have epilepsy if two or more unprovoked seizures (excluding early posttraumatic seizures) had occurred and no extracranial cause or metabolic disorder could be determined. Epilepsy was categorized into three groups (idiopathic, cryptogenic, and symptomatic) on the basis of the presumed etiologies (cf., Berendt & Gram, 1999; Engel et al., 2008). Because a differentiation between idiopathic and cryptogenic epilepsy was too subjective, both categories were put together for further evaluation. Dogs (n = 196) with extracranial causes of seizures (e.g., hypoglycemia, hepatic encephalopathy, uremia, hypocalcemia, hypothyroidism, intoxication; see Brauer et al., 2011) were excluded from the epilepsy group (group IIa) but assigned to group IIb (“reactive seizures”) and considered when analyzing causes of epileptic seizures in dogs of group II (see 'Results').
The statistical significance of differences between dogs with different types of head injury was calculated by Fisher's exact test. A p ≤ 0.05 was considered statistically significant.
Based on clinical examination and owners' responses, 648 dogs were included in final analysis of data. Two hundred thirty-six of these dogs (group Ia) were initially presented with head trauma to the clinic and were followed up for up to 11 years (mean 4.1 years) after head injury for development of late seizures (Table 1), whereas 412 dogs (group IIa) were presented with recurrent epileptic seizures to the clinic and were clinically and retrospectively investigated for a history of head injury (Table 3).
Table 1. Incidence of early (within the first week) and late seizures following head injury in dogs (group Ia)
Dogs with head injury (n)
Age at head injury (years)
Follow-up period after head injury (years) in dogs that survived the first week after injury (n = 198)
Dogs with seizures after head injury
Latency between head injury and first seizure
Age at onset of epilepsy (late spontaneous recurrent) seizures (years)
Early and/or late seizures (n) (%)
Early (insult- associated) seizures (n) (%)
Late (spontaneous recurrent) seizures (n) (%)
Early (insult-associated) seizures (days)
Late (spontaneous recurrent) seizures (years)
Data are shown as mean and range (plus standard deviation, SD). Dogs (n = 38) that died within the first week after head injury were not used in calculation of the average follow-up period after head injury and percent incidence of late seizures. In the data on latency for early seizures, “0” means that seizures occurred during or immediately after head injury. Data are shown separately for all dogs with head injury, dogs with closed head injury (i.e., without skull fracture or penetrating injury), and dogs with open head injury (i.e., with skull fracture or penetrating injury). Significant differences between dogs with closed or open injury are indicated by asterisk (p ≤ 0.05).
3.76 (0.08–15); SD: 3.45
4.09 (0.03–11); SD: 2.75
0.8 (0–3); SD: 0.84
0.96 (0.02–3); SD: 0.93
4.2 (0.17–10); SD: 2.2
Closed injury: 203/236 (86%)
3.89 (0.08–15); SD: 3.49
4.08 (0.16–11); SD: 2.75
0.65 (0–3); SD: 0.67
1.1 (0.02–3); SD: 0.78
4.7 (0.17–10); SD: 2.0
3.1 (0.15–11); SD: 3.38
4.16 (0.03–10); SD: 2.83
0.96 (0–3); SD: 1.18
0.9 (0.08–1); SD: 1.39
3.32 (1.2–4); SD: 1.3
3.3 (1–9); SD: 2.65
3.4 (0.03–10); SD: 3.63
0.56 (0–0.9); SD: 0.39
Risk of epilepsy following head injury in dogs (group Ia)
Gender distribution in this group of 236 dogs was 104 males, 98 females, 16 spayed, and 18 male-neutered, respectively. The percentages of male versus female dogs did not differ significantly (p = 0.6419). Age at head injury was 3.76 years, with a large range from 0.08 to 15 years (Table 1).
Of the 236 dogs with head injury, 44 (18.6%) exhibited early and/or late seizures (Table 1). Observed seizure types were partial and generalized tonic–clonic; convulsive status epilepticus or cluster seizures were observed in four dogs. Age at head injury in these 44 dogs was 3.69 years (range 0.16–15 years). Twenty-five dogs were female, 13 male, four male-neutered, and two spayed; the proportion of female dogs was significantly higher (p = 0.0174) than the proportion of male animals. The predominant cause of head injury in these 44 dogs was a car accident (n = 20), followed by punches (9), falls (8), horse boots (4), and bite wounds (3), respectively. Thirty-five of the 44 dogs showed a depressed level of consciousness at the time of admission to the hospital; in 14 of these 35 dogs consciousness was absent and in 14 reduced. Abnormal pupillary light responses and other ophthalmologic abnormalities were seen in 38 of the 44 dogs.
Thirty-three (14%) of the 236 dogs with head injury exhibited early seizures, mostly within 24 h following the injury (Table 1). Thirteen (39%) of these 33 dogs died (or were euthanized) within the first week after the injury. Of the whole group of 236 dogs of group Ia, 38 (16.1%) dogs died within the first week after head injury, indicating that mortality was significantly higher (p = 0.05) in dogs with early seizures after head trauma. Of the 198 dogs that did not die immediately after or within 1 week following head injury, 13 dogs (6.6%) developed late recurrent seizures, indicating development of epilepsy. The average latency between head injury and onset of epilepsy was 1 year. Only 2 of the 20 dogs that survived head trauma associated with early seizures developed late seizures, which was not significantly different (p = 0.2332) from the fraction of dogs with PTE in the whole group of animals with head injury.
Most dogs (86%) with head injury had closed injury, but 14% had open injury with broken skull and, in nine dogs, penetrating injury (Table 1). Open injury was associated significantly more often with seizures (early or late) than closed injury. Furthermore, 14.3% of the dogs with open injury developed epilepsy compared to 5.3% of the dogs with closed injury, which, however, did not reach statistical significance (p = 0.0925), most likely as a consequence of the relatively low sample size. The average latency between head injury and onset of epilepsy was similar in dogs with closed and open head injuries. In dogs with penetrating injury, 44% showed early seizures after the injury, but none of the dogs developed epilepsy, which may be secondary to the fact that only seven dogs survived the first week after head injury.
In the MRI, which was performed in 46 dogs of group Ia, intracranial hemorrhage was seen in 9 dogs and intracranial edema in 6 dogs. Six of the nine dogs with intracranial hemorrhage exhibited early seizures, indicating that such hemorrhage was a risk factor for early seizures (p = 0.0086). None of these dogs with intracranial hemorrhage developed late seizures. Additional MRI findings in dogs with early seizures included intracranial edema (n = 1) and hydrocephalus (n = 1). Of the 13 dogs with late seizures, seven exhibited reduced or absent consciousness at time of admission and nine had ophthalmologic abnormalities. Nine of the 13 dogs with PTE were females compared to two males and two neutered dogs, indicating a female preponderance in this group (p = 0.0154).
Risk of epilepsy following head injury (group Ia) versus other types of trauma
Within the 392 dogs with trauma of group I, 156 dogs (group Ib) had no head injury but other types of trauma, so that this group formed an ideal control for comparison with dogs with head trauma (group Ia; see Fig. 1). As shown in Table 2, epilepsy was diagnosed in three of the 156 dogs of group Ib (1.92%), which was significantly different (p = 0.0410) from the 6.6% dogs with epilepsy in group Ia. Age at trauma or age at onset of epilepsy were not significantly different between the two groups (Table 2). Furthermore, the percentage of intact males did not differ between groups, whereas the proportion of intact females was significantly lower in group Ib and the proportions of neutered males and spayed females were significantly higher (Table 2). Table 2 also shows the breed distribution in the two subgroups. Dog breeds with suggested inherited idiopathic epilepsies (e.g., Beagles, German Shepherds, Labrador Retrievers, Golden Retrievers, Dalmatians, Border Collies, Belgian Tervuerens, Poodles, and Vizslas; cf., Chandler, 2006; Ekenstedt et al., 2012) were equally distributed (if present in the cohort) among both subgroups, except that the Dachshund occurred significantly more often in group Ib (Table 2). The only group of dog breeds occurring significantly more often in group Ia was companion dogs (p = 0.0010). With the exception of the Cavalier King Charles Spaniel (n = 1 in each subgroup of group I), for none of the breeds included in this category (e.g., Chihuahua, Pekingese, Maltese, Pug dog, French bulldog) a hereditary basis for epilepsy is known or suggested (Chandler, 2006; Ekenstedt et al., 2012). The significant difference in proportion of companion dogs between group Ia and Ib was due to a relatively high number of Chihuahuas in group Ia (n = 10) versus Ib (n = 1; p = 0.0265). This is easily explained by the fact that these dogs often fall from the arm of their owners, so that the leading health issue in such toy dogs is injury, particularly head injury (Hawthorne et al., 1999; Geary et al., 2008).
Table 2. Comparison of dogs with head trauma (group Ia) and dogs with other types of trauma (group Ib)
Group Ia (head trauma)
Group Ib (other types of trauma)
For group Ia, only dogs that survived the first week after head injury were used (see Table 1). With respect to breeds, for the group of companion dogs only one subgroup (Chihuahuas) is separately shown, because all other breeds in this group did not differ between group Ia and Ib.
Significant differences between groups are indicated by asterisks (p ≤ 0.05).
Number of dogs
Age at trauma (years)
3.68 (range 0.08–14; SD 3.4)
4.03 (range 0.16–14; SD 3.4)
Age at onset of epilepsy
4.2 (range 0.17–10; SD 2.2)
4.68 (range 2–8; SD 2.76)
Other breeds (≤4 dogs per breed)
In addition to comparing the incidence of epilepsy in group Ia and group Ib, we also used a large hospital-based population of dogs that we previously studied for comparison (Löscher et al., 1985). In this study, 126 (0.57%) of 22,110 dogs had epilepsy, which is not statistically different from the incidence of epilepsy (1.92%) in group Ib (p = 0.0911) but significantly different from group Ia (p < 0.0001), thus substantiating that head trauma increases the risk of epilepsy in dogs.
Previous head injury in dogs diagnosed with epilepsy (group IIa)
Gender distribution in the 412 dogs with epilepsy was 188 males, 96 females, 63 spayed, and 65 male-neutered, respectively. The percentages of male versus female dogs differed significantly (p < 0.0001). Age at diagnosis of epilepsy in the 412 dogs was 3.8 years (0.33–13 years), which was similar to group Ia. The predominant seizure types were partial and generalized tonic–clonic.
Of the 412 dogs diagnosed with epilepsy, 64 (15.5%) had a history of head injury (Table 3). Causes of head injury in these 64 dogs were falls (n = 19), punches (16), car accidents (15), bite wounds (12), and horse boots (2), respectively. Average age at head injury was 2.1 years, with a large range from 0.08 to 6 years (Table 3). The average latent period between head injury and onset of epilepsy was 1.7 years, that is, somewhat longer than in group Ia. The majority (95%) of dogs with head injury had a closed injury. Convulsive status epilepticus or cluster seizures were observed in 25 of the 64 dogs with a history of head injury. With respect to sex distribution, 33 dogs were male, 13 female, 9 male-neutered, and 9 spayed, indicating, in obvious contrast to group Ia, a male preponderance in this group (p = 0.05), which was also present in the whole group IIa (see above).
Table 3. Previous head injury in dogs diagnosed with epilepsy (group IIa)
Dogs with epilepsy (n)
Number of epileptic dogs with history of head injury
Age at head injury (years)
Dogs with early (insult-associated) seizures after head injury (n) (%)
Latency between head injury and first seizure
Age at onset of epilepsy (of dogs with head injury) (years)
Early (insult-associated) seizures (days)
Late (spontaneous recurrent) seizures (years)
Data are shown as mean and range (plus standard deviation, SD), Data are shown separately for all dogs with head injury, dogs with closed head injury (i.e., without skull fracture or penetrating injury), and three dogs with open head injury (i.e., with skull fracture and penetrating injury).
All: 64 (15.5%)
2.1 (0.08–6); SD: 1.87
2.6 (0.007–6); SD: 2.72
1.7 (0.04–8); SD: 1.51
3.8 (0.33–13); SD: 2.54
Closed injury: 61/64
2.1 (0.08–6); SD: 1.86
2.6 (0.007–6); SD: 2.72
1.7 (0.04–8); SD: 1.53
3.8 (0.33–13); SD: 2.6
Open injury: 3/64
2.3 (1–5); SD: 2.31
1.4 (0.18–3); SD: 1.45
3.7 (2–5.18); SD: 1.61
In 6 of the 64 epileptic dogs with a previous history of head injury, early seizures had been observed after head trauma. Ophthalmologic abnormalities were seen in 6 of the 64 dogs at time of referral because of recurrent seizures. In the MRI, which was performed in 44 of the 64 dogs with head injury of group IIa at time of referral because of recurrent seizures, brain abnormalities (most likely related to the initial head trauma) were seen in 15 dogs. These included skull fractures, scars, and lesions (hypo- or hyperintensive signals) in cerebellum, lobus frontalis, and pituitary gland, and edema, and dilation or asymmetry of the ventricles. In none of the 64 dogs did neurologic examination, including MRI, and owners' interviews indicate any cause of epilepsy other than head trauma.
Other causes of recurrent epileptic seizures in group IIa
In addition to head trauma (n = 64), several other causes of recurrent epileptic seizures were analyzed in the 412 dogs with epilepsy of group IIa (Fig. 2A). These included brain tumors (n = 33), brain infections (25), cortical dysgenesis (9), and several other probable causes (17), including scars (2), hydrocephalus (10), cysts (2), and vascular malformations (3). Furthermore, six dogs had a dual pathology (e.g., hydrocephalus and cysts) that most likely caused the symptomatic epilepsy. In the majority (258) of the 412 dogs, a cause could not be unequivocally identified, so that these animals were assigned to the idiopathic/cryptogenic group (Fig. 2A).
Apart from epilepsy, seizures may represent a natural response of the normal brain to metabolic disturbances or intoxication (Engel & Starkman, 1994; Podell et al., 1995; Brauer et al., 2011). When all 608 dogs with recurrent epileptic seizures of group II were considered, 196 dogs (32%) exhibited such “reactive” or provoked seizures (group IIb), compared to 154 dogs (25%) with symptomatic epilepsy and 258 dogs (42%) with idiopathic/cryptogenic epilepsy (Fig. 2B).
In humans, PTE is a common and frequently disabling complication of TBI (Annegers et al., 1996; Frey, 2003; Lowenstein, 2009). In a recent population-based study in 2,118 patients hospitalized with TBI, 115 (5.4%) developed epilepsy in the 2–3 years following hospital discharge (Ferguson et al., 2010). Similarly, in the present study in 236 dogs hospitalized with head injury (group Ia), 13 (6.6%) of those dogs that survived head trauma developed epilepsy. As in humans (Lowenstein, 2009), our cohort showed an increased risk of PTE with increased severity of the head injury. When skull fracture was chosen as an indicator of severe TBI (Lowenstein, 2009), incidence of epilepsy was 14.3% in dogs. However, most likely as a consequence of the relatively low sample size, the difference in the proportion of dogs that developed epilepsy after closed (5.3%) versus open head injuries (14.3%) did not reach statistical significance.
Fourteen percent of the dogs with head injury exhibited seizures in the first week following the injury (i.e., early seizures). According to studies in humans, the incidence of such early seizures is correlated with the distribution of head-injury severity within the specific group being studied (Frey, 2003). Accordingly, 44% of the dogs with penetrating injury, the most severe type of TBI (Lowenstein, 2009), displayed early seizures, which was significantly higher than the incidence of early seizures in dogs with closed injury (12.8%).
Only 15% of the dogs with late seizures also exhibited early seizures. Because risk factors for early seizures overlap with those for PTE, the importance of early seizures as an independent risk factor has not been univocally demonstrated in studies in humans conducted to date (D'Ambrosio & Perucca, 2004; Agrawal et al., 2006). In addition to severity of head injury (as reflected by skull fracture and dural penetration by injury), generally accepted risk factors for PTE include brain contusion, intracranial hematoma, nonreactive pupils, and a depressed level of consciousness at the time of admission to the emergency department (Temkin, 2003; D'Ambrosio & Perucca, 2004; Agrawal et al., 2006; Lowenstein, 2009). In the dogs with PTE of the present study, depressed level of consciousness and ophthalmologic abnormalities were the most common clinical signs at time of admission to our hospital. Of interest, in group Ia, female dogs were significantly more affected with early and late posttraumatic seizures than male dogs were, although both genders were equally distributed in the total population of 236 dogs with head injury. Such female preponderance was not observed in dogs with head injury of group IIa. In humans, it has been postulated that female gender may affect the outcome of TBI, but studies in this respect are controversial (Coimbra et al., 2003; Bayir et al., 2004; Berry et al., 2009; Whelan-Goodinson et al., 2010; Leitgeb et al., 2011).
In order to determine whether TBI increases the risk of epilepsy, the incidence of epilepsy following TBI has to be compared with the incidence of epilepsy in the general population. In a cohort of 4,541 children and adults with TBI studied over a 50-year period, 97 (2.1%) of the patients developed epilepsy, resulting in a standardized incidence ratio of 3.1 (based on the expected incidence of epilepsy in the general population), which indicates that patients with TBI had a 3.1-fold increased risk of developing epilepsy compared to the general population (Annegers et al., 1998). Patients with severe TBI had an incidence ratio of 17 (Annegers et al., 1998). In dogs, epilepsy accounts for 0.6–1.0% of all ill dogs referred to veterinary teaching hospitals (Holliday et al., 1970; Löscher et al., 1985; Schwartz-Porsche, 1994). To our knowledge, the largest hospital-based population has been studied by Löscher et al. (1985), in which 126 of 22,110 dogs (0.57%) had epilepsy. Based on the expected incidence of epilepsy in dogs calculated from the study of Löscher et al. (1985), an overall incidence ratio of 11.6 can be calculated for the dogs with head trauma examined in the present study, indicating that dogs with TBI had an 11.6-fold increased risk of developing epilepsy compared to the general hospital population. Incidence ratio increased to 23 in dogs with severe TBI. However, these figures may overestimate the risk of epilepsy in dogs with TBI, because the cohort described in our previous study (Löscher et al., 1985) was not from the same period and hospital as the cohort evaluated in the present study. We therefore compared dogs with head trauma (group Ia) with dogs without head trauma (group Ib) from the same cohort, resulting in a 3.4-fold difference in the incidence of epilepsy in the two subgroups, indicating that dogs with TBI had an 3.4-fold increased risk of developing epilepsy compared to this control group from the same cohort. This figure nicely compares with the 3.1-fold increased risk of developing epilepsy described by Annegers et al. (1998) for human TBI patients.
In humans, the data available suggest that most of the patients develop epilepsy within 2 years after TBI (Agrawal et al., 2006; Diaz-Arrastia et al., 2009). In dogs followed up after head injury (group Ia) for up to 11 years, the average latency between injury and onset of epilepsy was 0.96 years and most of the animals developed epilepsy within 2 years after head injury.
In addition to studying the development of epilepsy in dogs after head injury (group Ia), we examined the history of head injury in dogs (group IIa) that were examined clinically in our hospital because of recurrent epileptic seizures. About 16% of these 412 dogs had a history of head injury, which is significantly higher (p = 0.0016) than the overall incidence of 6.6% determined for group Ia. However, in contrast to group Ia, dogs of group IIa were not clinically examined in our hospital at time of head injury, so that we had to rely on information on head injury provided by the owner and veterinarians who treated the dogs at time of the injury. In addition, we cannot exclude a selection bias in group IIa. Similar differences between prospective and retrospective cohort studies on the risk of epilepsy after brain insults have been reported for febrile seizures in humans, in that the risk of developing epilepsy after febrile seizures is lower than would be estimated from the high percentage of patients with symptomatic epilepsy that have a history of febrile seizures (Scantlebury & Heida, 2010). In the present study, another difference between group IIa and group Ia was the longer latent period between presumed head injury and epilepsy in group IIa. We can, therefore, not exclude that other predisposing causes of epilepsy contributed to the high risk of seizures associated with head injury in group IIa.
To our knowledge, the present study is the first that systematically evaluates the risk of epilepsy after hospitalization of dogs for head injury (group Ia). The second approach used in group IIa, that is, evaluating whether dogs that were referred to the clinic with recurrent seizures had a history of head trauma or other brain insults that may have caused development of spontaneous recurrent seizures, has been used previously in a study on 240 dogs with recurrent seizures referred to a veterinary hospital over a period of 5 years (Pakozdy et al., 2008). When excluding dogs with “reactive seizures” due to extracranial causes, 211 dogs with presumed epilepsy remained, of which only 3 (1.4%) had a history of head trauma (Pakozdy et al., 2008). However, no questionnaires or standardized owner interviews on potential causes of epilepsy were used in the latter study, but data were from clinical examination at time of referral. The most common etiology was brain tumors (18%) and encephalitis (11%). Similar studies in smaller groups of dogs have been reported by Croft (1965), Palmer (1972), and Podell et al. (1995). In our relatively large cohort of 412 dogs with epilepsy (group IIa), brain tumors and infections were also relatively common causes of symptomatic epilepsy, whereas, in contrast to humans, cerebrovascular disease did not play any significant role as a risk factor, because it is rare in dogs (Wessmann et al., 2009). Similar to humans (Lowenstein, 2009), 63% of the dogs of group IIa had idiopathic or cryptogenic epilepsy, which is comparable to a previous study in a much smaller group of dogs (Berendt & Gram, 1999). Also comparable to previous studies (Podell et al., 1995; Pakozdy et al., 2008; Brauer et al., 2011), in addition to epilepsy, metabolic disturbances or intoxications were a relatively common cause of epileptic seizures in dogs (Fig. 2).
Our study has a number of shortcomings. First, although we started with a relatively large cohort of >1,300 dogs, final group size for group Ia (TBI) and IIa (epilepsy) was relatively small, because many dogs could not be finally analyzed or assigned to these groups. Particularly in group Ia, this caused problems in obtaining enough statistical power for comparing further subgroups, that is, dogs with closed and open injuries. Another potential problem is the use of questionnaires or telephone interviews in such a retrospective study, because dog owners may not exactly recapitulate the details of the initial event. However, in case of severe trauma or convulsive seizures, our experience is that most dog owners remember the timing and details of such dramatic events very well. Furthermore, we had details on all dogs in our clinical database, so that we could recheck the owners' responses.
Irrespective of these potential shortcomings, the present study indicates that head trauma in dogs, particularly severe TBI resulting from skull fracture, is associated with a significant risk of developing epilepsy. Therefore, dogs with severe TBI are an interesting natural model of PTE that provides a novel translational platform for studies on human PTE. A dog's head can be injured in many ways, including a car accident, a fall, a blow to the head, or a gunshot wound. The precipitating event, most frequently a car accident, is common and readily recognizable, and usually results in the dog's owner seeking medical attention in a veterinary university hospital. Using dogs with TBI as a natural model, several topics under research in the clinical TBI arena can be addressed. First, identification of reliable biomarkers of epileptogenesis is a high-priority goal for the epilepsy community (Engel, 2011). The identification of clinically relevant biomarkers that precede the onset of epilepsy and correlate with epileptogenesis would have a profound effect of our understanding of the processes that determine whether a patient develops late seizures after TBI and could aid in research aimed at finding a preventive therapy (Mani et al., 2011; Galanopoulou et al., 2012). No validated biomarker is currently available, but recent advances in neuroimaging, electrophysiology, molecular biology, and genetics promise to reveal clinically useful biomarkers for epilepsy in the near future (Engel, 2011). Dogs with TBI may be a valuable tool in this respect. Second, prevention of epileptogenesis after TBI is an unmet clinical need (Löscher & Brandt, 2010; Pitkänen & Lukasiuk, 2011; Galanopoulou et al., 2012). In both humans and dogs, there is typically an interval of months to a few years between the initial injury and onset of epilepsy, offering a therapeutic window of opportunity for antiepileptogenic intervention (Lowenstein, 2009; Kharatishvili & Pitkänen, 2010). Based on our experience with diverse antiepileptogenic strategies in rat models (cf., Löscher & Brandt, 2010), we plan to evaluate the most effective strategy in dogs after TBI, and will include testing of biomarkers that may provide quantitative measures of the process of posttraumatic epileptogenesis and the effects of treatment. Advantages of dogs as a natural model of PTE are that they may reproduce the predominant and relevant epileptogenic mechanisms of human PTE more closely than rodent models of PTE, but they lack many of the confounding factors that complicate the design, conductance, and interpretation of clinical trials on antiepileptogenic therapies in humans (Dichter, 2009; Temkin, 2009; Mani et al., 2011).
In a number of human diseases, the dog already serves as an invaluable large animal model for assessment of novel therapeutic agents and has filled a crucial step in the translation of basic research to new treatment regimens (Nowend et al., 2011). Since we first proposed epileptic dogs as an animal model of human epilepsy (Löscher & Meldrum, 1984; Löscher et al., 1985), various old and new antiepileptic drugs have been evaluated in this species, including placebo-controlled trials (e.g., Schwartz-Porsche et al., 1985; Löscher et al., 2004; Govendir et al., 2005; Volk et al., 2008; Hardy et al., 2012; Munana et al., 2012). If we can further substantiate that canine PTE is a valid translational platform for human PTE, this may facilitate the process of therapeutic development to make antiepileptogenic drugs discovered in the laboratory available for the prophylactic treatment of PTE.
We thank Asla Pitkänen for constructive discussion during preparation of the manuscript and Christina Brauer and Karl Rohn for help during performing the study. The study was supported by grants (Lo 274/11-1; TI 309/4-1) from the Deutsche Forschungsgemeinschaft (DFG; Bonn, Germany) within the DFG Research Unit (FOR) 1103.
None of the authors has any conflict of interest to disclose. We confirm that we have read the Journal's position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.