Abstracts

RATIONALE: The goal of this study was to construct a clinical prediction model for the early identification of children destined to develop refractory temporal lobe epilepsy (TLE) at 2-years after epilepsy onset. The ability to predict refractory TLE at an early point is essential for the timely use of effective surgical therapy.
METHODS: This retrospective cohort study sought to identify every patient between 1-18 years of age with TLE seen as an outpatient in the Division of Neurology at The Children[ssquote]s Hospital of Philadelphia (CHOP) during 1999. Clinical settings included a university-based practice and 6 suburban satellite practices. Patients were identified through centralized billing records, using ICD-9 codes 345.4x (complex partial epilepsy). Eligible patients had a history of at least 2 complex partial seizures with semiology suggesting temporal lobe onset, such as staring, unresponsiveness, automatisms, and/or auras. Five candidate predictor variables for refractory TLE were determined before data extraction began. The 5 predictor variables were: age at epilepsy onset, an early risk factor for epilepsy (history of febrile seizures, CNS infection, significant head trauma, and/or neonatal seizures), family history of epilepsy in a first-degree relative, temporal lobe abnormality on MRI scan, and outcome of the first AED trial. Outcome at 2-years after onset was classified as seizure-free or refractory TLE.
RESULTS: Overall, 1,846 outpatient records were reviewed. Of these, 120 patients met inclusion criteria and had at least 2 years of follow-up data. Forty-five of 120 patients (37.5%) had refractory TLE at 2-years after onset, and 75/120 (62.5%) were seizure-free. Three significant predictors of refractory TLE were found based on bivariate analysis -- an early risk-factor for epilepsy [risk ratio = 3.5 (2.2, 5.6)], temporal lobe abnormality on MRI scan [2.9 (1.9, 4.6)], and failure of the 1st AED trial [16.5 (6.3, 43.9)]. Logistic regression indicated that the best model to predict refractory TLE contained only the variable [italic]failure of 1st AED trial[/italic]. Including only this variable resulted in a positive predictive value (PPV) = 0.89 [0.76, 0.96] and negative predictive value (NPV) = 0.95 [0.87, 0.99] to predict refractory TLE at 2 years after epilepsy onset.
CONCLUSIONS: A clinical prediction model containing the single variable [italic]failure of 1st AED trial[/italic] can accurately predict the outcome [italic]refractory TLE[/italic] at 2-years after epilepsy onset, based on retrospective cohort data. If this syndrome-specific model is verified prospectively, then initial counseling regarding the possibility of highly effective (but irreversible) surgical therapy could begin after 1st AED failure. Earlier surgery in appropriate patients could avoid years of physical and psychosocial dysfunction and maximize long-term outcome.
Support: Univ. of PA McCabe Fellows Award; National EpiFellows Foundation.