Objective: To determine the occurrence of magnetic resonance imaging-detected hippocampal atrophy (HA) in patients with partial epilepsy (temporal and extratemporal, cryptogenic, or symptomatic). Magnetic resonance imaging-detected HA has been demonstrated to be both sensitive and specific for hippocampal sclerosis in cryptogenic temporal lobe epilepsy. Design: Patients' hippocampal formations were measured on a computerized system using Tl-weighted, 5-mm contiguous magnetic resonance coronal images made perpendicular to the hippocampus long axis. Hippocampal atrophy was defined on the basis of a normative asymmetry index and correlated with the epileptogenic focus defined by clinical, electroencephalographic, and magnetic resonance imaging (apart from HA) localizing data. Patients and Control Subjects: Seventy patients with intractable complex partial seizures of temporal, extratemporal,or undefined origin and 21 healthy control subjects. Results: Hippocampal atrophy was present in 70% of patients with cryptogenic temporal lobe epilepsy (TLE) (n=40), 44% of patients with symptomatic TLE (n=9), 29% of patients with extratemporal epilepsy (n=14), and 6% of unclassified patients (n=16). In the cryptogenic TLE category, HA was marked and usually concordant (93%) with electroencephalographic lateralization. Hippocampal atrophy was often mild in the extratemporal epilepsy category. With the use of a wider confidence interval (+/- 3.1 SD instead of +/- 2.2 SD), HA specificity for TLE increased to 93%, HA specificity for lateralizing cryptogenic TLE reached 96%, and HA sensitivity for cryptogenic TLE stood almost unchanged (68%). We found a link between early convulsions and HA occurrence. Conclusions: Hippocampal atrophy is a marker for TLE. Dual pathologic findings are detected in 44% of symptomatic TLE cases. Mild HA is rarely associated with extratemporal epilepsy. Magnetic resonance imaging-based hippocampal volumetric analysis is a useful method to localize the origin of partial complex seizures.
