Children often present with infantile spasms after herpetic encephalitis




To determine what epilepsy types occur after herpetic encephalitis and what are the determinant factors for subsequent infantile spasms.


We analyzed retrospectively the clinical history of 22 patients, referred to Necker and Saint Vincent de Paul Hospitals (Paris) through the French pediatric epilepsy network from March 1986 to April 2010 and who developed epilepsy some months after herpetic encephalitis. We focused on seizure semiology with video–electroencephalography (EEG) recording, and on neuroradiology and epilepsy follow-up.

Key Findings

Fourteen patients developed pharmacoresistant spasms, and eight developed focal epilepsy, but none had both. The patients who developed spasms were more frequently younger than 30 months at age of onset of epilepsy and had herpetic encephalitis earlier (mean 10.6 months of age) than those who developed focal epilepsy (mean 59.7 and 39.6 months, respectively). Epilepsy follow-up was similar in both groups (8.5 and 11 years, respectively). We found 26 affected cerebral areas; none alone was related to the development of epileptic spasms.


Risk factors to develop epileptic spasms were to have had herpetic encephalitis early (mean 10 months); to be significantly younger at onset of epilepsy (mean 22.1 months); and to have cerebral lesions involving the insula, the hippocampus, and the temporal pole.

Herpes simplex encephalitis (HSE), the most common sporadic encephalitis in European countries, is a devastating illness. Early detection and diagnosis of herpes simplex virus type 1 (HSV-1) encephalitis is crucial, as prompt administration of acyclovir can dramatically reduce morbidity and mortality (Delong et al., 1981; Tyler, 2004a; Baskin & Hedlund, 2007; Hsieh et al., 2007). However, poor outcome remains a possibility (McGrath et al., 1997; Hjalmarsson et al., 2007; Basak et al., 2011). Brain lesions observed following HSE in children typically involve the temporal lobe (Shoji et al., 2002; Wasaya et al., 2005; Gümüş et al., 2007; De Tiège et al., 2008). Little is known regarding epilepsy following herpetic encephalitis, and most studies involve adult patients (Sellner & Trinka, 2012). The most frequent type seems to be focal epilepsy, but lesions are usually multiple so that only few patients benefit from surgery, mostly when epilepsy involves the mesial temporal structures (Sellner & Trinka, 2012). Of interest, two cases of infantile spasms were also reported (Ohtaki et al., 1987; Riikonen, 1993). We hypothesized that this epilepsy type could have been underestimated due to the absence of several ictal electroencephalography (Video-EEG) recordings.

Infantile spasms are a type of epilepsy frequently associated with brain lesions and developmental delay. They are characterized by sudden flexion or extension of the axial and/or proximal limb musculature, with a wide range of intensity. Spasms occur in clusters of different durations. They appear mostly between 3 and 7 months of age, and they are associated with typical EEG pattern (hypsarrhythmia) named West syndrome normally caused by prenatal insults (Cusmai et al., 1993); 3–6% can have a late onset, after 12 months of age, known as late-onset infantile spasms (Bednarek et al., 1998; Eisermann et al., 2006). The etiology of spasms is variable. Brain lesions are frequent, and are certainly related to the generation of spasms, although it remains difficult to determine any specific location. Although basal ganglia are suspected to contribute, given the semiology of the seizures, it is the resection of cortical areas that permits control of the spasms, suggested by the importance of cortical-subcortical interactions in this seizure type (Chugani et al., 1992). Therefore, a loop involving both the cortex and the basal ganglia could be involved (Desguerre et al., 2003). Late-onset infantile spasms usually have postnatal etiology, including central nervous system infection (Eisermann et al., 2006).

We report a group of 22 patients with extensive EEG recordings, who developed epilepsy several months following herpetic encephalitis. Our objective was to determine what epilepsy types occur after herpes encephalitis and what are the determinant factors for subsequent infantile spasms that 14 patients developed in this series.

Patients and Methods

We performed a retrospective study of the electroclinical history, including cognitive functions and imaging findings associated with epilepsy following herpetic encephalitis.

The study was performed with patients of the French pediatric epilepsy network, in the Hospital Necker Enfants Malades in Paris, which coordinates the network. Search for the subjects was performed in the neurophysiology department of Necker hospital, since we selected the patients with recorded seizures. Patients with all diagnostic codes for HSE were included (HSE, HSE + epilepsy, HSE + infantile spasms). From the 47 files collected, we excluded those of patients without epilepsy following encephalitis (12 files) and those with incomplete data (lack of proven HSE and infantile spasms reported but not recorded on EEG; 13 files). Twenty-two patients (13 girls and 9 boys) aged 1–140 months (mean 21 months) at the onset of herpetic encephalitis, between March 1986 and April 2010 were enrolled. All developed epilepsy some months after herpetic encephalitis.

Herpetic encephalitis was confirmed by laboratory results including viral genome using polymerase chain reaction (PCR) (HSV-1) on evaluation of cerebrospinal fluid (CSF) and/or the synthesis of antibodies in the CSF. All patients were treated with acyclovir.

We classified cognitive functions as: “normal,” “schooling difficulties,” and “no speech,” based on the global assessment of the pediatric neurologist at last visit.

We analyzed interictal EEG at onset of seizures. Interictal EEG at onset of spasms (14 patients) revealed monofocal spikes in five cases, multifocal spikes in three, and hypsarrhythmia in four, whereas two patients had no spikes, one with the most extensive lesions and the other with late-onset spasms (42 months). Interictal EEG at onset of focal epilepsy revealed monofocal spikes in seven patients and multifocal spikes in one patient. Seizure semiology was analyzed based on clinical history, contact with the families or the neuropediatrician, and video-EEG recordings. All patients had video-EEG recordings during the initial phase and from the beginning of epilepsy. We then identified the risk factors for either type of seizures. All patients were followed for more than 100 months after the onset of epilepsy.

Neuroimaging was analyzed in the neuroradiology department. All patients had at least two neuroimaging investigations within the first week of encephalitis and the other at the beginning of epilepsy, 1–19 years after HSE. For five patients, magnetic resonance imaging (MRI) was repeated in the course of the last 2 years to better clarify topography of the lesions. We analyzed the last MRI in order to benefit from the best imaging quality and to identify the precise extent of brain damage in the chronic condition. The median time lag from HSE to the MRI analyzed for this study was 69.1 months (range 2–234, standard deviation [SD] 75) for the patients with infantile spasms and 71.2 months (3–165, SD 58) for those with focal epilepsy.

The MRI studies were analyzed by the two neuroradiologists (NB and DG). For each imaging we identified damaged areas and defined them as regions of interest (ROIs), searching whether similar damaged areas affected the other children. All together there were 13 affected areas on each side.

To compare patients with spasms (N = 14) and patients with focal epilepsy (N = 8) we used tests for small samples, namely Fisher exact test, two-sided for proportions and Student t-test or Wilcoxon rank test (in case of unequal variances) for means.


Of the 22 who had herpetic encephalitis, fourteen developed epileptic spasms and eight partial seizures. We first ensured with repeat video-EEG recordings that no patients had exhibited both spasms and focal seizures. On ictal EEG, spasms recording demonstrated in most instances (10/14) generalized diffused or asymmetrical slow waves of high amplitude followed by electrodecrement and/or fast activity. Only four infants with spasms exhibited hypsarrhythmia. Topography of lesions in these cases was similar whether hypsarrhythmia was present or not. Schooling and cognitive abilities are given in Table 1.

Table 1. Cognitive abilities
 SpasmsPartial seizuresp-Value
  1. a

    Fisher exact.

Schooling delay29%100%0.002a
No speech71%00.002a

The patients who exhibited spasms tended to be younger at the onset of encephalitis (mean 10.6 [range 1–26 months]) than those who had focal seizures (mean 39.6 months [range 3–140 months]) (p = 0.11, Wilcoxon rank test). Following herpetic encephalitis, the mean age of occurrence of spasms was significantly lower (22.1 [range 5–50 months]) than that of patients with focal seizures (59.7 [13–174 months]) (p = 0.02). Seventy-five percent of the patients who developed focal epilepsy were older than 30 months at onset of the epilepsy.

In most instances, infantile spasms following herpetic encephalitis begin after the age of 1 year. Although the time lag from the onset of herpetic encephalitis to that of epilepsy did not differ significantly between both groups (2–30 and 3–41), follow-up of the patients who developed focal epilepsy was in all cases over 34 months; therefore, over the maximum interval to the appearance of epileptic spasms following herpetic encephalitis (30 months) (Table 2).

Table 2. Age at onset of encephalitis and epilepsy and follow-up
 SpasmsFocal epilepsyp-Value
  1. a


  2. b

    Fisher exact.

  3. c

    Wilcoxon rank test (two-sided).

N 14 8 
Age (months): m (SD) [range]122.9 (77.9) [10–263]193.2 (77.1) [50–2920.05a
Age at encephalitis10.6 (8.7) [1–26]39.6 (48.2) [3–140]0.11b
Age encephalitis <30 months (%)10062.50.03c
Age at onset epilepsy22.1 (14.9) 5–5059.6 (51.7) 13–1740.02b
Age at epilepsy <30 months (%)64.3 250.18b
Time from encephalitis to onset of epilepsy11.6 (9.4) 2–30 20 (14.6) 3–410.11a
Follow-up from epilepsy onset100.8 (68.9) 5–232133.6 (82.2) 10–2510.33a
Follow-up from encephalitis112.4 (72.2) 7–239153.6 (78.7) 34–2570.23a

Indeed, all patients who exhibited spasms, but only 62% of those who exhibited focal epilepsy, had encephalitis before the age of 30 months (p = 0.03) (Table 2).

Regarding the topography of brain lesions (Table 3) there were 26 ROIs. The insular region was the most affected, concerning all 22 patients. Regarding basal ganglia, only 1 patient had putaminal involvement, whereas in 13 the thalamus was affected.

Table 3. Topography of brain lesions (right and/or left)
 SpasmsFocal epilepsyp-Valuea
  1. a

    Fisher exact test or t-test (two-sided).

Hippocampus (H) (%)86500.14
Temporal pole (TP)71370.19
Insula (I)100750.12
Cortical convexity (CC)   
Temporal lobe0250.12
Frontal lobe36620.38
Parietal lobe50750.38
Occipital lobe21121
Other areas   
Number of affected areas   
H + PT + I: m (SD) [range]3.9 (1.5) [2–6]2.1 (1.2) [0–4]0.01
CC: m (SD) [range]1.5 (1.5) [0–4]2.9 (1.8) [0–6]0.07
Total: m (SD) [range]7.7 (3.2) [2–13]6.5 (3.2) [2–11]0.40

The number of ROIs for patients with spasms (2–13 affected areas) did not significantly differ from that of the patients with focal seizures (2–11 affected areas). The bilateral affectation was similar for the two groups (40.5% in focal epilepsy and 44.6% in spasms). None of these regions was by itself associated with the development of focal epilepsy or spasms, but the patients who developed spasms and therefore had encephalitis before 26 months of age, had significantly more mesial areas involved, (orbitofrontal, temporal pole, hippocampus, cingular gyrus, and insula) (p = 0.05).

The combination of temporal pole, insular, and hippocampal lesions was the most significantly linked to the occurrence of spasms (p = 0.01). There was a significant inverse correlation between the age at onset of encephalitis and the involvement of this area (R = −0.49, p = 0.02, Spearman correlation), thus the younger the patient, the higher the probability of hippocampal, insular, and temporopolar involvement and the higher the probability of developing spasms (Figs. 1 and 2).

Figure 1.

Brain MRI of subject with spasms. (A) Axial inversion recovery T1-weighted image demonstrating the involvement of left insula (red arrow). (B) Axial inversion recovery T1-weighted image showing left hippocampus (green arrow) and temporal lobe (blue arrows) affected.

Figure 2.

Brain MRI of another subject with spasms. (A) Axial T2-weighted image demonstrating the involvement of superior edge of the left insula (red arrow). (B) Axial T2-weighted image showing left hippocampus (green arrow) and temporal lobe (blue arrows) affected.

In contrast, involvement of the convexity tended to be linked to the occurrence of focal seizures (frontal, parietal, temporal, and occipital cortices) (p = 0.07) (Fig. 3).

Figure 3.

Brain MRI of a patient with focal epilepsy. (A) Coronal T2-weighted image demonstrating the involvement of right parietal (red arrow) and occipital (green arrow) lobes. (B) Coronal T2-weighted image showing right temporal lobe (blue arrow) and insula (yellow arrow) affected.

All infants with spasms and 63% of those with focal seizures were pharmacoresistant until the end of follow-up (101 and 133 months, respectively).


This study shows the epilepsy following herpetic encephalitis often consists of infantile spasms and that the main risk factors for the occurrence of spasms in this context are the early age of occurrence of encephalitis and the involvement of insula and mesial temporal structures. The patients who developed encephalitis the earliest were also those who had the highest risk of extensive brain lesions, including the insula (combined with the temporal pole and the hippocampus) and of developing spasms. On the other hand, subcortical structures are mildly involved in the chronic stage, in contrast with reports concerning the acute stage (Panisset et al., 1999). We found a significant relation between the involvement of insula, temporal pole, and hippocampus, and the occurrence of spasms. Because early occurrence of herpetic encephalitis increased the risk for infantile spasms, both factors—the age of occurrence of encephalitis and structures involved—are not independent.

The patients with spasms were properly classified and it is unlikely that they would later develop epileptic spasms, since spasms were the first seizure type to appear, and follow-up was well beyond the usual age of appearance of spasms (>100 months). On the other hand, patients with spasms were properly classified using video-EEG. None of them later developed focal seizures, with a follow-up of >100 months.

We could find only two single case reports mentioning infantile spasms following herpetic encephalitis and, in both these cases, spasms were intractable (Ohtaki et al., 1987; Riikonen, 1993). Spasms began at 16 months, some months following herpetic encephalitis. In both instances, brain lesions involved the temporal lobe.

In our series, the insula was the most affected area, whatever the type of epilepsy following encephalitis. This suggests that the entry of the virus into the brain is likely to include this area, as shown in the rat by Tomlinson. Indeed, following olfactorial or intracorneal inoculation, the virus diffuses along the olfactory nerve to the anterior olfactory nucleus, lateral olfactory tract, and then the septal nuclei and the temporal lobe, hippocampus, and cingular gyrus (Tomlinson & Esiri, 1983; Tyler, 2004b).

The age-dependent topography of brain lesions following herpetic encephalitis does not seem to have been reported. It is consistent with the sequence of maturation of brain structures. According to functional imaging studies, cerebral metabolic rate for glucose and cerebral blood flow increase and reach maximal values earlier in insula than in the convexity of the brain, especially in frontal areas (Chugani et al., 1987; Chiron et al., 1992). This period of early postnatal maturation is known to be of increased vulnerability to various insults (Chiron et al., 1997; Sankar et al., 2002; Vannucci & Hagberg, 2004). This could explain the predominant involvement of insula in case of early encephalitis followed by epilepsy, whereas the convexity is affected mainly in cases of later encephalitis.

Late onset spasms that begin after the age of 1 year can be cryptogenic, or related to postnatal brain lesions in which the temporal lobe is most often involved (Eisermann et al., 2006). The insula is anatomically near the basal ganglia, and any damage to the insula could therefore contribute to the activation of epileptic phenomena within the basal ganglia, provided brain immaturity has retained the ability to develop spasms (Dulac et al., 2007; Mondal et al., 2009). It is therefore the combination of both features that is likely to determine the occurrence of spasms (Chugani et al., 1992).

Spasms following herpetic encephalitis are usually particularly pharmacoresistant. Since this is not the case for focal epilepsy in this series, it is unlikely that etiology is the main factor of pharmacoresistance (Chen et al., 2006). Topography of brain lesions could be the main issue, namely the involvement of the hippocampus, as in many epilepsies beginning in infancy (Semah et al., 1998). It is noteworthy that insular epilepsy produces epileptic spasms when it begins early (Afif et al., 2010).

Considering surgery is difficult to consider because of frequent bilaterally of epileptogenic areas (Lellouch-Tubiana et al., 2000). These children often develop very poor cognitive abilities, whereas the impact of the spasms on quality of life is absolutely dreadful. It therefore seems to us reasonable not to eliminate the surgery option, including removal of the insula, temporal pole, and hippocampus, when the benefit for the control of spasms overcomes the risk for the memory.


I acknowledge that all co-authors have been involved in the study and/or the preparation of this manuscript. No undisclosed groups or persons have had a primary role in the study and/or in manuscript preparation. All co-authors have seen and approved the submitted version of the paper and accept responsibility for its content.

None of the authors has any conflict of interest. We confirm that we have read the Journal's position on issues involved in ethical publication and affirm this report is consistent with those guidelines.