The Effect of Age on Seizure Semiology in Childhood Temporal Lobe Epilepsy


Address correspondence and reprint requests to Dr. Tuxhorn at Epilepsie-Zentrum Bethel, Klinik Mara I, D-33617 Bielefeld, Maraweg 21, Germany. E-mail:


Summary:  Purpose: Complex partial seizure is the characteristic seizure type observed in epilepsy arising from temporal lobe structures. The seizure evolution in adult patients is quite stereotyped and well characterized, manifesting initially with an aura, behavioral arrest, and oroalimentary and gestural automatism. A greater variability of semiology including motor features with tonic or myoclonic components, as well as a paucity of automatism, has been reported in young children with temporal lobe epilepsy. The aim of our study was to examine in more detail the effects of age on individual ictal features to be able to determine the critical age when lesional temporal lobe seizure semiology undergoes transition from the pediatric to the more adult-type clinical pattern.

Methods: We performed a video analysis of 83 seizures from 15 children (aged 11–70 months) selected by post–temporal lobectomy seizure-free outcome, looking specifically at the motor and behavioral (nonmotor) manifestations in relation to age of the children.

Results: All of the children younger than 42 months had seizures with early and marked motor features, which included tonic and myoclonic components and epileptic spasms. Parallel with age, the frequency of these motor components decreased, and in five of 11 children older than 3 years, motor features were totally absent. Analyzed quantitatively, we saw a linear and inverse correlation of the ratio of motor components with age at monitoring.

Conclusions: These findings support the hypothesis that events in brain maturation significantly affect clinical seizure semiology and may override the more typical localizing features seen in adult-type temporal lobe epilepsy. These findings are important to consider in the early diagnosis of childhood temporal lobe epilepsy.

Temporal lobe (TL) structures are frequently involved in the genesis of partial epilepsy. In adults, TL epilepsy is characterized by a somewhat stereotypic and well-described semiology consisting of epigastric auras, arrest of activity, staring, altered consciousness, and oroalimentary and hand automatisms, reflecting activation of limbic structures (1–4).

In contrast, the semiology of TL seizures in young children is not that homogeneous, and various age-dependent motor phenomena, including tonic, clonic, hypermotor components, and epileptic spasms (ESs) have been reported (5–16).

The aim of this cross-sectional study was to investigate and describe seizures of young children with “pure” temporal lobe origin, and also to determine the likely age when the transition of lesional TL seizure semiology from the pediatric to the adult-type clinical pattern occurs.



Fifteen consecutive patients younger than 6 years (nine girls and six boys) with refractory lesional TL epilepsy who underwent long-term video-EEG monitoring and became seizure free after temporal lobectomy were selected [postoperative follow-up was between 22 and 84 (mean, 46) months]. Age at onset ranged between 2 days and 46 (mean, 14) months, and age at video-EEG monitoring was 11 to 70 (mean, 44) months. Epilepsy substrates defined by magnetic resonance imaging (MRI) as the etiologies included indolent tumors (nine), focal cortical dysplasia (five), and hippocampal sclerosis in one patient. Five patients had mesial; five patients, lateral lesions; another five children had more extensive lesions involving both the mesial and lateral parts of the temporal lobe. Eleven cases were left-sided, and four right-sided.

Seizure recording and evaluation

Time-labeled video recordings of 83 seizures were reviewed by three independent investigators blinded to the ictal EEG. We excluded those seizures (∼5%) in which the recording quality did not permit analysis of the complete seizure from onset to end (i.e., patient was out of camera sight, record started or stopped within the ictal period, or serious tape errors). Each patient had at least three seizures (mean, 5.6; range, 3–10 seizures per child). Patients were examined during the seizures by especially trained EEG technicians to assess the level of consciousness with response to verbal or nonverbal external stimuli. One patient was evaluated twice: with scalp electrodes at 11 (1a) and with subdural grids at 42 months (1b).

Each seizure was analyzed independently by the authors with regard to the motor, sensory, consciousness, and autonomic spheres of the seizures, as well as manifestations in the postictal period. Data were documented on a specially designed data sheet allowing qualitative and quantitative analysis. The events of each sphere were classified by using a time scale as onset, very early (<10 s), early (10–20 s), or late (>20 s after clinical onset) events.

As has been reported by other authors it was difficult to determine the level of consciousness in a number of the very young—mostly preverbal—children (17–20). Therefore, instead of the classification proposed by the International League Against Epilepsy (21), we used a classification based on seizure semiology (19,22,23). Data sheets included the following seizure components: tonic (sustained muscle contraction of the body or limbs, lasting a minimum 3), myoclonic (sudden, nonrhythmic muscle jerks), clonic (series of rhythmic contractions of the body, face, or limbs), epileptic spasm (brief and abrupt axial posturing, usually with flexion in the neck and extension in the extremities, with a duration <3), hypermotor (stereotypically repeated, purposeless, and violent movements of the limbs and trunk), hypomotor (sudden arrest from preictal activity), and automotor (behavioral arrest with different automatisms) seizures. After recording, the independent blinded investigators (A.F. and E.F.) classified each seizure's components in order of appearance as

initial seizure component → evolutional seizure component(s) (see in Table 1)

Although there was agreement among the investigators in most of the cases, all cases were reviewed and classified together with the senior investigator (I.T.). The little disagreement (<10% of the attacks) was caused by the difficult observation of automatisms in young children. As other authors described, in infants, it is difficult to distinguish subtle apparently voluntary arm or leg movements or oroalimentary activity (possible automatisms) from background behavior (14). After classification we broadly divided all observed seizure components into two groups, depending on the presence of motor manifestation. Tonic, clonic, myoclonic, hypermotor components, and ESs were categorized into a group of motor seizure components. The group of nonmotor seizure components included hypomotor and Automotor (with oral, manual, or pedal automatism) attacks.

Besides seizure classification; we reviewed the patients' medical charts and collected their most important clinical data (Table 1).

Table 1.  Clinical and semiology features of 15 patients younger than years with temporal lobe epilepsy
Patient no.Age at monitoring (mo).Age at epilepsy onset (mo)LocalizationEtiologyAutomatismPossible lateralizing signsSeizure evolutionRatio of MSC compared with all seizure component
  • MSC, motor seizure components; R, right; L, left; FCD, focal cortical dysplasia; HS, hippocampal sclerosis; ES, epileptic spasm; H, hand; nosew. RH, postictal nosewiping with right hand; sgtcs, secondarily generalized tonic–clonic seizure; AMS, automotor seizure; hypom., hypomotor seizure; hyperm., hypermotor seizure.

  • a

     Patient was monitored at two different ages

a1a114R mesialFCDNoES L > R2× ES series100%
2192 daysR mesial
 + lateral
FCD2/5 late oralDystonia LH
Nosew. RH
3× myoclonic (−>2× sgtcs)
2× AMS
33215L mesialHS4/5 early manualDystonia RH3× myoclonic −>2× AMS
1× myoclonic −> 1× AMS
1× AMC
43222L mesial
 + lateral
FCD2/6 early pedalNosew. LH5× tonic
1× AMC −> 1× tonic
53312R lateralTumorNoNosew. LH, RH6× tonic −> 2× hypom.
4× hypom.
63718L mesial
 + lateral
Tumor3/4 late pedal
1/4 late oral
No4× tonic −> 2× AMS67%
74229L mesial
 + lateral
Tumor5/5 late manualES L > R5× AMS −> 4× ES series44%
a1b424R mesialFCD2/5 late oralNo3× hypom.
2× AMS
8487L lateralTumorNoNo3× hypom. −> 1× hyperm. −>
 −> 1× tonic −> 1× clonic
1× hyperm. −> 1× tonic −>
−> 1× clonic
9515L mesialTumor3/5 early manual
2/5 early oral
Dystonia RH
Nosew. LH
5× AMS0%
105218L lateralTumor8/10 late oral
5/10 late manual
No10× AMS0%
11574L mesial
 + lateral
FCD4/6 late oral
2/6 late manual
Dystonia LH
Dystonia RH
4× AMS
2× tonic −> 2× AMS
12575L mesialTumorNoNo2× hypom.0%
136346L mesialTumor5/5 late manualNosew. RH, LH5× AMS0%
146515L lateralTumor4/5 late oral
1/5 late manual
Dystonia RH5× tonic −> 5× AMS50%
157013R lateralFCD4/4 late manual
1/4 late oral
Dystonia LH
Nosew. RH
4× hypom. −>
−> 4× tonic −> 4× AMS

Statistical analysis

Spearman rank correlation was used to test the hypothesis of an age-dependent change in the seizure semiology grouped as motor and nonmotor components of lesional TL epilepsy in young children.


We observed and analyzed 115 seizure components in a total of 83 attacks from 15 patients. Mean duration of seizures was 62 (range, 9–177) s. Most of the patients had very homogeneous attacks with only one or two different types of seizures.

Forty percent of the components were in the defined motor group and consisted of tonic, myoclonic, clonic, hypermotor features, and ES. Automotor and hypomotor seizure components that were categorized into the nonmotor seizure components group were observed in 60% of the seizures (see Table 2). Most patients had a mixture of the components of the two groups, five elder children had only pure nonmotor seizures, whereas the youngest patient showed only motor attacks.

Table 2.  Seizure component types and frequency identified among patients and seizure components
Seizure component typesFrequency among patientsFrequency among seizure components
Motor seizure components  
 Tonic7 (47%)29 (25%)
 Epileptic spasm2 (13%)6 (5%)
 Clonic2 (13%)5 (4%)
 Myoclonic2 (13%)4 (3%)
 Hypermotor1 (7%)2 (2%)
Nonmotor seizure components  
 Automotor12 (80%)51 (44%)
 Hypomotor5 (33%)18 (16%)
Total15 patients
115 seizure

To determine whether the appearance of the semiology grouped as motor and that grouped as nonmotor was a function of age and also better to define the age-specific transition, we further calculated the ratio of motors-seizure components in relation to the total number of seizure components in each patient and represented this as a function of patient's age (see Fig. 1). The results showed that there was a linear and inverse relation of the ratio of motor seizure to total seizure components with the age at monitoring (r = −0.64; p < 0.01). Younger children had significantly more tonic, clonic, and spasm activity than older ones. Five of the 10 children older than 42 months had none of the motor features observed in the younger group.

Figure 1.

Ratio of motor seizure components in 15 patients younger than 6 years with temporal lobe epilepsy.

Longitudinal follow-up of patient 1 also showed this age dependence of semiology, in that at 11 months (1a), only ESs were noted, and at 42 months (1b), hypomotor and automotor seizures were observed exclusively. Monitoring may cause a selection bias for age at epilepsy onset (i.e., there can be a higher ratio of early-onset epilepsies among younger patients). However, we did not find a correlation between the motor-seizure components ratio and either age at epilepsy onset (r = −0.01; p = 0.96), localization, lateralization, or etiology.

Two of the 15 patients reported somatosensory aura (5 and 13), and a suspected aura was observed in an additional 10 children.

We also observed automatisms in 12 patients. In 56 (67%) of 83 seizures analyzed, automatisms were seen. Eighty-two percent of the automatisms did not appear at onset but ≥20 s after clinical onset, which we defined as late. The most frequent automatisms were manual (53%) or oral (39%) and manifested in less complex formats than those seen in adults (11). Patient 1, who was monitored twice showed no automatism at the age of 11 months; however, 31 months later, she had prominent oral automatisms evolving late during the automotor seizures.

We recorded three different lateralizing signs in our group of infants and young children. Ictal dystonic posturing of an arm were produced by six patients; in five cases, it was contralateral to the seizure focus; in one case, it happened in both arms. Postictal nosewiping were recorded also in six children; in four cases, it was ipsilateral to the seizure focus, and twice it was observed in both hands, respectively. Two children showed asymmetric ESs: it was contralateral in one, and ipsilateral in another case.

Secondarily generalized tonic–clonic seizures were recorded in one and had been reported in the history of another two patients.


TL seizure semiology appears to be significantly influenced by age-related mechanisms so that ictal features in young children may not give much clue about the presence of this type of localization-related epilepsy (7,8,11). The diagnostic referral and evaluation of these patients, particularly for epilepsy surgery (which is frequently curative in these children who may have a high seizure frequency and are at risk for secondary developmental deficits) may therefore be unduly delayed. Knowledge about the age-related temporal evolution, what may be considered as the immature ictal manifestations, transform into a typically mature adult-like semiology, is very relevant (1–4).

Our study investigating infants and young children with well-localized lesional TL epilepsy as determined by video-EEG and MRI showed that this transformation occurred in a linear fashion as a function of age during preschool years, so that in the fourth year of life, the nonmotor components of automotor seizures as the hallmark of limbic epilepsy as seen in adults was the dominant seizure manifestation.

By contrast, all patients younger than 42 months had a high ratio of motor features including tonic, clonic, myoclonic components, and ESs compared with the overall observed seizure components. Beyond age 42 month, the rate of complex partial seizure semiology with behavioral arrest and automatisms increased and became the predominant feature in half of the children. A number of studies have demonstrated that the seizures of young children manifesting with focal epilepsy consist mostly of bilateral motor signs, which may be asymmetric, more typically seen in generalized epilepsy (5,6,14,18,24,25). However, although the children had localization-related epilepsy in these studies, the patient population was not lobe specific. A few other studies have examined TL seizures in childhood, but in an older population than ours (8,9,12). One study investigated a group of patients younger than 16 years with seizure-free outcome after temporal lobectomy and showed that children older than 6 years had TL seizures with features seen in adults (11). However, compared with our study, this work included a small group of young patients (only six children were younger than 6). A recent article found that the clinical features of TL epilepsy caused by hippocampal sclerosis in children as young as 4 years were similar to those in adults; however, it was an etiology-based study without involving very young patients [the youngest child was 4 years old (26)]. An earlier study specified that this semiology transformation happens between the second and sixth years of life; however, the patient group selection was based on ictal EEG data and not on postoperative seizure-free outcome (7).

Animal studies investigating the ontogenetic expression of drug-induced limbic epilepsy in immature young rats showed comparable age-dependent ictal behavior. Investigating kainic acid–and pilocarpine-induced seizures in young rats during the first 2 postnatal weeks corresponding to a maturational age of the human infants, these rat pups developed hyperactivity, scratching, hyperextension of the limbs, tremor, head bobbing, and myoclonic movements (13,27–29). More mature rats older than 2 weeks, in addition to prominent motor signs, produced limbic seizures consisting of rearing, akinesia, and masticatory movements. Further studies in hippocampal-kindled rat pups demonstrate that the afterdischarge thresholds (i.e., the lowest current intensity necessary to elicit an afterdischarge) are highest during the second to third postnatal week, suggesting resistance of the limbic system to synchronization (30). These findings from animal studies appear to offer a reasonable explanation why TL seizures in immature humans manifest more clearly with typical automotor features only once the limbic system has matured from the fourth year of life.

In spite of their circumscribed seizure focus, two of the 15 children (1 and 7) showed also generalized ESs series among their seizures. There are earlier studies describing ESs in children with focal lesions (31–34). In a long-term follow-up of 192 children with ESs, it was found that 60% of them developed new focal seizures, mostly from the TL (35). We think that ES is an age-specific seizure manifestation in our cases, too. An earlier study analyzing 8,680 ESs found that most of the asymmetric and asynchronous spasms were associated with a seizure focus contralateral to the behaviorally more involved side (34). In our group of young patients with TL epilepsy, both children showed asymmetric ESs; however, it was contralateral in one, and ipsilateral in another case.

We also observed two different lateralizing signs earlier observed in adulthood TL epilepsy (36,37). Ictal dystonic posturing of an arm was produced by six patients; in five cases, it was contralateral to the seizure focus, and in one case, it happened in both arms. Postictal nosewiping—an ipsilateral lateralizing phenomena in TL epilepsy—were recorded also in six children; in four cases, it was ipsilateral to the seizure focus, and twice it was observed in both hands. These results of our small group of patients are promising; however, an expansion of this series would give more reliable data.

The age-related motor component ratio was independent of the age at epilepsy onset. This is supported not only by our cross-sectional semiology study of the 15 patients but also by the longitudinal follow-up of one child, who showed a definitive change of seizure semiology between the first and fourth year of her life. The ratio of motor seizure components depended on neither the mesial nor lateral localization, the lateralization nor the etiology (tumor, focal cortical dysplasia, hippocampal sclerosis) in our patients. This corresponds to the results of a study on adult patients with mesial and neocortical TL epilepsy, which demonstrated no differences in the seizure semiology reflected involvement of the limbic system (3). Conversely, we can hypothesize that during the first 3 years of life, the immature limbic structures synchronize poorly and remain clinically silent at this age.

In summary, our study supports the evidence that the seizure semiology of lesional TL epilepsy in young children is an expression of late limbic system maturation as well as rapid and extensive subcortical extratemporal activation (13,38). These findings are important to consider to facilitate the early diagnosis and effective management of TL epilepsies in infants and young children. Nevertheless, a multiinstitutional expansion of this small series—particularly a greater number of infants younger than 2 years—would likely provide additional useful information not only to the lateralizing value of several ictal features but also in a more detailed distribution of different seizure types over time.