Purpose To study prognosis and prognostic predictors of sporadic non-lesional temporal lobe epilepsy (TLE).

Method 474 patients with TLE were consecutively seen from April 1987 to April 2004. 190 had a sporadic non-lesional TLE and a follow-up longer than 2 years. 284 patients were excluded because of family history for TLE, incomplete history, poor compliance with treatment, psychogenic seizures, no brain MRI study, presence of intracranial lesions except for scattered T2 hyperintense spots on hemispheric white matter or mesial temporal sclerosis (MTS). The following prognostic predictors were considered: age at onset of epilepsy, gender, family history of non-TLE or febrile seizures, perinatal factors, history of febrile seizures, ictal phenomena, MTS and interictal EEG. The end point was time to 24 month seizure freedom after treatment onset. The χ² test, Student's t test, Kaplan–Meier survival curves with log rank test (univariate analysis) and Cox proportional hazards regression models (multivariate analysis) were used to assess seizure prognosis and prognostic predictors.

Results At univariate analysis, patients achieving 24 month seizure freedom had a significantly older age at onset of epilepsy (33.5 ±19.9 vs 17.2 ±14.4 years), and lower occurrence of febrile seizures (11.0% vs 24.4%) and MTS (19.0% vs 35.6%). The chance of remission was directly correlated to age at onset of seizures and inversely correlated to a history of febrile seizures and to the presence of MTS. At multivariate analysis, age at onset of epilepsy was the only significant prognostic predictor.

Conclusion Older age at onset predicts better prognosis in sporadic non-lesional TLE.