Epilepsy diagnosis and localization in patients with antecedent childhood febrile convulsions

The association among FC, TLE, and mesial temporal sclerosis (MTS) is well recognized.^1-9 Most reports were of patients with intractable partial epilepsy, specifically TLE referred for presurgical evaluation.^2,4-9 No data clearly indicate a preferential temporal localization of epilepsy after antecedent FC. FC have not been evaluated as an antecedent to extra-temporal partial epilepsy and few reports have identified FC as a risk factor in GE.^10-11 We evaluated the incidence of antecedent FC in a large epilepsy clinic population and related the presence and characteristics of FC to the localization of subsequent epilepsy.

Methods. We prospectively inquired about FC in all patients presenting for evaluation in the epilepsy clinic of one of us (B.A.K.) through a detailed patient questionnaire and interview. We excluded all patients with only nonepileptic attacks (suspected or proven). The population included patients with recent onset as well as chronic refractory epilepsy. Details about FC were asked of parents or other family members who witnessed the seizures by direct interview or by telephone. When feasible, old medical records were retrieved for review. Questions were asked about age at onset, description (including complex features), postictal manifestations, etiology of fever, temperature, duration, total number, maximum number of FC in 24 hours, mode of termination (spontaneous or after medical intervention), and treatment. Family history of epilepsy and FC was also documented.

Seizure and epilepsy diagnosis was made in most of our patients based on a combination of clinical description, standard electroencephalogram (EEG), EEG/closed circuit television (CCTV) (in approximately one-half), magnetic resonance imaging (MRI), and, in 146 patients, a full presurgical evaluation(including positron-emission tomography [PET] in all patients, and ictal single photon emission computed tomography [SPECT] and invasive EEG in some) and surgical outcome. Patients with uncertain epilepsy diagnosis were excluded. Seizure type, epilepsy diagnosis, localization, FC history, and family history were stored in a computerized database. FC were defined as seizures usually occurring in the first 6 years of life in association with fever, without central nervous system (CNS) infection or other definable cause, and not preceded by afebrile seizures. FC were considered complex when one or more of the following characteristics were present: prolonged duration(more than 15 minutes' duration), focal features, or multiple convulsions(more than one in 24 hours or in the same febrile illness). Simple FC were defined as FC with none of the complex features defined above. Patients who had convulsions with fever secondary to CNS infections were studied separately, whereas patients whose initial convulsions were afebrile and preceded the occurrence of febrile seizures were not considered to have had FC.

We reviewed the incidence of antecedent childhood FC in relation to epilepsy, as well as FC description in all patients. We analyzed age at onset of FC, age at onset of epilepsy, interval between onset of FC and onset of epilepsy, the duration of the longest FC, and the presence of each complex feature. We performed the analysis excluding patients whose initial convulsion was in conjunction with a CNS infection, unless specified otherwise. We used the χ² test and Student's t-test to examine differences between groups.

To determine the relationship between antecedent febrile convulsions and pathology, we reviewed the pathology findings in a subgroup of patients with TLE who underwent surgical treatment.

Results. Patients. There were 465 males (46.3%) and 540 females (53.7%). The mean age at presentation was 32.3 years(±14.9). The mean age at onset of epilepsy was 17 years (±15.4). Partial epilepsy was the diagnosis in 805 patients (table 1). Localization could be determined in 526 patients: 310 had TLE, and 216 ETE (123 frontal, 64 parietal, 29 occipital). The remaining 279 patients with partial epilepsy could not have definitive localization of the epileptogenic focus based on available data. We were unable to specifically localize foci within a region of the temporal lobe in all patients with TLE. Hence, the TLE group includes both patients with mesial temporal foci and patients with neocortical temporal foci. GE was the diagnosis in 146 patients, 62 of whom had juvenile myoclonic epilepsy, 38 other primary generalized epilepsies (29 included absence seizures), 41 symptomatic or cryptogenic GE, and five indeterminate generalized epileptic syndromes. Eleven had epilepsy undetermined whether focal or generalized based on available data. Forty-three had special syndromes (isolated seizures or situation-related seizures).

Febrile convulsions. Febrile convulsions were reported by 133 of 1005 (13.2%) patients (see table 1). If convulsions in the setting of a presumed CNS infection are included, the additional 20 patients increased the incidence to 15.2%. The remaining analysis was performed excluding this group, which was studied separately. No difference was found between patients with partial epilepsy and those with GE with respect to the risk of antecedent FC. However, TLE was significantly more likely to be preceded by FC (78/310, 25.2%) than ETE (12/216, 5.6%)(p < 0.000001) or GE (16/146, 11.0%) (p < 0.001). No statistically significant difference was shown between frontal, parietal, and occipital localizations with respect to antecedent FC (frontal: 6/123, 4.9%; parietal: 5/64, 7.8%; occipital: 1/29, 3.4%).

The age at onset of FC tended to be earlier for patients with TLE than for patients with GE, but the difference did not reach statistical significance. No difference existed between patients with TLE and those with ETE with respect to age at first FC.

FC characteristics. Most patients had complex FC (69.2%)(table 2). Patients with GE were significantly more likely than patients with partial epilepsy to have had simple FC (p< 0.00005). The difference was even greater when comparing GE with TLE specifically (p < 0.00001). Eight of 16 patients with GE had juvenile myoclonic epilepsy, all but one with an antecedent simple febrile convulsion, and none in association with a CNS infection. Although complex FC were more likely to be seen in TLE (80.8%) than ETE (66.7%), the difference was not significant. No significant difference was found between the different localizations of ETE with respect to characteristics of FC. The mean maximal duration of FC was significantly longer for patients with subsequent TLE than for patients with subsequent GE (p < 0.005)(table 3). There was no significant difference in FC duration between TLE and ETE patients. Patients with TLE were more likely to have a single FC episode, whereas patients with GE were more likely to have multiple FC episodes (seetable 3). However, the difference did not reach statistical significance.

Patients with antecedent FC. Patients with antecedent FC had a mean age at onset of epilepsy of 9.4 years (±9.5) compared with 18.1 years (±15.0) for those without preceding FC (p < 0.000001). For those patients with antecedent FC who later developed GE, the mean age at seizure onset was 7.7 years (±5.3). For those who subsequently developed TLE, the mean age at seizure onset was 9.9 years(±9.9). The difference was not statistically significant. The mean interval from onset of FC to onset of epilepsy was longer in patients with TLE than those with GE but did not reach statistical significance(p = 0.09) (see table 3).

Twenty-nine patients from the FC group had temporal lobectomy performed at our center. The pathologic diagnosis was mesial temporal sclerosis (MTS) in 23 (79.3%). This is probably an underestimate because all patients but one appeared to have MTS based on MRI criteria (atrophy with or without increased signal in one hippocampus). The patient with no evidence of MTS on MRI had MTS on pathology. MTS may have been missed on pathologic examination because the hippocampal formation was at times not clearly identifiable for diagnosis, and cell count was not performed. FC history was available in 22 of 23 patients with pathologic MTS diagnosis. The FC was complex in 21(95.4%) and specifically prolonged in 17 (81%). In contrast, the incidence of MTS in 64 temporal lobectomy patients without antecedent FC was 37.5%(p < 0.0005).

Patients with convulsions in the setting of CNS infection. Twenty patients had convulsions with high fever in the setting of a CNS infection. Seventeen had meningitis and three encephalitis. The mean age at occurrence of the first convulsion with fever was 20.6 months (±22.1), the mean age at onset of epilepsy was 8.5 years (±7.3) and the mean interval was 6.82 years. These were not significantly different from the group with FC and subsequent partial epilepsy. Details of the FC were available in 18; all were complex, and seven had a combination of two complex features. Thirteen were prolonged, three were focal, and nine were repetitive. The mean maximal duration of FC was 6.2 hours (±13.4).

Seventeen patients developed partial epilepsy (85%). The epileptogenic focus was temporal in 11, extratemporal in four, and unknown in two. Three developed symptomatic GE. Three patients with extratemporal foci, one with unknown localization, and two with GE were mentally retarded, as opposed to none with TLE.

Discussion. Febrile convulsions occur in 3% to 5% of children.¹² Many previous studies have investigated the risk of epilepsy after antecedent FC.^13-16 The overall risk is 2% to 7% in cohort studies of unselected FC, depending on the length of follow-up.^13-16 The risk of epilepsy following FC increases in the presence of certain FC features. For example, Annegers et al. reported a risk of 2.4% after simple FC, 6% to 8% with a single complex feature, 17% to 22% with two complex features, and 49% with all three complex features.¹⁶ In addition, different prognostic factors were present for partial epilepsy and GE after FC. A greater number of FC, a family history of epilepsy, and being older than 3 years at the time of the first FC were associated with GE, whereas complex features of FC were strongly associated with subsequent partial epilepsy.¹⁶ However, prospective epidemiologic studies did not investigate the specific localization (for example, temporal or frontal) of subsequent partial epilepsy, or risk factors for specific localizations.

Several reports have observed an association between TLE and FC, especially complex FC.^1-9 Most were based on series of patients with refractory epilepsy presenting for surgical treatment and patient with this association usually had hippocampal sclerosis as the underlying pathology.^1,2,4-9 It was proposed that TLE preceded by complex FC is a homogenous clinical entity or syndrome^4,7 with excellent response to surgical treatment.^7-9,17 Although it appeared that the association between FC and subsequent partial epilepsy was specific for TLE with hippocampal sclerosis, this hypothesis has not been formally investigated.

Some studies failed to show a specific association between TLE and antecedent FC.^18-19 Rocca et al. found that FC are a risk factor for generalized tonic clonic and absence seizures.^10,11 They found a similar incidence of antecedent FC in patients with generalized tonic clonic (19%), complex partial (18%), and absence seizures (21%).^10,11,19 Camfield et al. investigated the types of seizures and epilepsy preceded by FC in children.²⁰ Fifteen percent of childhood onset epilepsy was preceded by FC. A history of FC was more common in patients with primary generalized seizures than those with partial onset seizures and was not related to the "etiology" of epilepsy. A history of prolonged FC was associated with neurologic impairment but not with a particular seizure type. It was not more common in patients with intractable complex partial seizures.²⁰ Berg et al. found that complex FC had a strong correlation with symptomatic partial epilepsy suggesting an underlying increased susceptibility to seizures related to pre-existing identifiable disturbances in cerebral function.²¹ Limitations of these studies include the absence of epileptogenic focus localization (for example, TLE versus ETE)^18-21 and lack of inclusion of patients with adult onset epilepsy.^20,21

Our study is the largest to examine antecedent FC in both childhood and adult onset epilepsy. We found that 13.2% of epilepsy patients had a history of antecedent FC. This number is comparable with other studies.²⁰ We subdivided patients with partial epilepsy based on the localization of the epileptogenic focus; this helped us demonstrate a significantly higher incidence of antecedent FC in the TLE group (25.1%) when compared with GE (10.9%) and ETE (5.5%) groups. In our study, this association is missed when patients with partial epilepsy are compared with those with GE, showing no significant difference. We feel that this explains why some previous studies found similar incidence of antecedent FC in patients with generalized tonic clonic, absence, and complex partial seizures and no strong association between FC and partial epilepsy.^10,11,19-21 One study that examined patients with mesial temporal epilepsy who became free from seizures after surgery found an even higher incidence of antecedent FC, in 78% of patients.⁸ Our TLE group included intractable as well as easily controlled patients and those with mesial as well as lateral temporal foci. This may have diluted an even stronger association. It is extremely difficult to reliably distinguish patients with lateral temporal foci from those with mesial temporal foci, hence we could not study these two groups separately.

Our study has some limitations. Although we prospectively inquired about FC and gathered data on FC description at the time of first presentation of patients, the ascertainment of FC and its characteristics was retrospective. Validation of retrospective FC description was not always possible, but when old medical records were retrieved and reviewed, information was usually in agreement with that given by living mothers or other informed relatives, particularly with respect to the complex nature of FC. The population studied may not be representative of the whole population with epilepsy as a result of greater representation of intractable epilepsy. However, our patient population is less selective than that from pure surgical series because of the presence of many patients with recent onset or well-controlled epilepsy. Finally, localization of the epileptogenic focus was occasionally based on limited data. The gold standard of localization, being seizure-free after focal resection, is certainly not feasible in this patient group as a whole. The criteria for localization varied between patients, the minimum being a typical aura and typical seizure semiology with signs and symptoms of known localizing value. However, the vast majority of patients had EEG confirmation of the localization of the focus with interictal epileptiform or ictal discharges. When not enough data existed for localization, patients were classified in the "unknown localization" group, rather than forcing a localization with limited data. Patients with less than confident diagnosis of epilepsy were not included in the study.

We found that patients with GE were more likely to have antecedent simple FC, whereas those with partial epilepsy were more likely to have antecedent complex FC. These findings are in accordance with those of Annegers et al.¹⁶ However, we were unable to demonstrate a significant difference between partial epilepsy and GE with respect to total number of FC. Patients with GE preceded by FC tended to have an earlier age at onset of epilepsy and a shorter interval between first FC and first unprovoked seizure, compared with patients with TLE preceded by FC.

In our study, the most common complex feature in the FC preceding TLE was prolonged duration. This is in agreement with several studies of patients with TLE.^{1,2,4,5,7,9,22,23} In addition, one study in families with FC found a prolonged FC duration to be the most important predictor of TLE.²⁴ This would not have been projected by the study of Annegers et al., which found that focal features and repeated episodes were associated with a higher risk of subsequent partial epilepsy than prolonged FC duration.¹⁶ Focal signs were the second most common complex FC feature in our patients.

ETE was preceded by FC in only 5.5% of patients compared with 10.9% in GE and 25.1% in TLE. No statistically significant difference existed between frontal, parietal, and occipital localization with respect to preceding history of FC. The incidence of FC in this group is not very different from the reported incidence in the general population.¹² Therefore, the association between ETE and antecedent FC is at best weak and it is unlikely that FC have an etiologic role in ETE.

The group of patients who had convulsions in the setting of CNS infections and fever deserve special consideration. The convulsions were invariably complex when history was available, and prolonged duration was the most frequent complex feature encountered. Eighty-five percent developed partial epilepsy, with a temporal focus in most. Previous studies have shown that CNS infections are associated with partial epilepsy,²⁵ and, when they occur before age four, with TLE and hippocampal sclerosis.²⁶ This raises the possibility that the severe convulsions that commonly accompany CNS infections are responsible for the association with subsequent TLE. However, CNS infections may be associated with more widespread injury and epileptogenesis as evidenced by the subsequent development of GE in 15% and extratemporal epilepsy in 20% of our patients. When the focus is temporal, some studies have also indicated that hippocampal involvement is more commonly diffuse or bilateral.^27,28

Hippocampal sclerosis or MTS is the most common pathologic finding in adults with intractable TLE. Evidence supports the hypothesis that MTS is the pathologic basis of temporal lobe seizures.^29,30 Our study further supports the association of MTS with complex prolonged FC. However, an etiologic relationship between MTS and complex FC has not been proven. Several possible overlapping explanations exist for the association, among which are the following: (1) complex FC and MTS may both be related to the same etiology, for example, prenatal or perinatal insult; (2) complex FC may be the reflection of a low seizure threshold caused by MTS; (3) MTS may be the result of recurrent seizures over many years; (4) complex FC may be the insult causing MTS and TLE; and (5) complex FC may facilitate the development of MTS in the presence of a predisposing hippocampal or other abnormality (for example, a dysgenesis). Our study cannot resolve the nature of the association between TLE, MTS, and FC. However, because FC were uncommon in patients with extratemporal epilepsy in contrast to TLE, the second explanation is unlikely to be the sole explanation, unless MTS reduces seizure threshold in the presence of fever more than extratemporal lesions do. Other evidence suggests that MTS is unlikely to be the result of recurrent seizures.⁵ In our study, the strong association between TLE and antecedent complex FC, not otherwise found with GE and ETE, can be consistent with the fourth and fifth as well as first possibilities above. A causal relationship should be investigated with a prospective study that includes serial detailed MRI imaging starting at the time of occurrence of complex FC.