Prognostic patterns and predictors in epilepsy: a multicentre study (PRO-LONG).

Objectives To describe the long-term prognosis of epilepsy and prognostic patterns in a large cohort of newly diagnosed patients and identify prognostic factors. Methods Study participants were 13 Italian epilepsy centres with accessible records dating back to 2005 or earlier, complete data on seizure outcome and treatments, precise epilepsy diagnosis, and follow-up of at least 10 years. Records were examined by trained neurology residents for demographics, seizure characteristics, neurological signs, psychiatric comorbidity, first electroencephalogram (EEG) and MRI/CT, epilepsy type and aetiology, antiepileptic drugs (AEDs), and 1-year, 2-year, 5-year and 10-year seizure remissions. Five predefined prognostic patterns were identified: early remission, late remission, relapsing-remitting course, worsening course and no remission. Prognostic factors were assessed using multinomial logistic regression models. Results 1006 children and adults were followed for 17 892 person-years (median 16 years; range 10-57). During follow-up, 923 patients (91.7%) experienced 1-year remission. 2-year, 5-year and 10-year remissions were present in 89.5%, 77.1% and 44.4% of cases. 5-year remission was associated with one to two seizures at diagnosis, generalised epilepsy, no psychiatric comorbidity, and treatment with one or two AEDs during follow-up. 10-year remission was associated with one or two AEDs. The most common prognostic pattern was relapsing-remitting (52.2%), followed by early remission (24.5%). 8.3% of cases experienced no remission. Predictors of a relapsing-remitting course were <6 seizures at diagnosis, (presumed) genetic aetiology and no psychiatric comorbidity. Conclusions Few seizures at diagnosis, generalised epilepsy and no psychiatric comorbidity predict early or late seizure freedom in epilepsy. Achieving remission at any time after the diagnosis does not exclude further relapses.