Association of magnetic resonance imaging identification of mesial temporal sclerosis with pathological diagnosis and surgical outcomes in children following epilepsy surgery

Abstract

Object Mesial temporal sclerosis (MTS) is widely recognized as a significant underlying cause of temporal lobe epilepsy. Magnetic resonance imaging is routinely used in the preoperative evaluation of children with epilepsy. The purpose of this study was to evaluate the prevalence, reliability, and prognostic value of MRI identification of MTS and MRI findings indicative of MTS in a series of patients who underwent resection of the medial temporal lobe for medically refractory epilepsy. Methods The authors reviewed the medical records and preoperative MRI reports of 25 patients who had undergone medial temporal resections (anterior temporal lobectomy or functional hemispherotomy) for medically intractable epilepsy. The preoperative MRI studies were presented for blinded review by 2 neuroradiologists who independently evaluated the radiographs for selected MTS features and provided a final interpretation. To quantify interrater agreement and accuracy, the findings of the 2 blinded neuroradiologists, the nonblinded clinical preoperative radiology report, and the final pathology interpretation were compared. Results The preoperative MRI studies revealed MTS in 6 patients (24%), and histopathological analysis verified MTS in 8 (32%) of 25 specimens. Six MRI features of MTS were specifically evaluated: 1) increased hippocampal signal intensity, 2) reduced hippocampal size, 3) atrophy of the ipsilateral hippocampal collateral white matter, 4) enlarged ipsilateral temporal horn, 5) reduced gray-white matter demarcation in the temporal lobe, and 6) decreased temporal lobe size. The most prevalent feature of MTS identified on MRI was a reduced hippocampal size, found in 11 of the MRI studies (44%). Analysis revealed moderate interrater agreement for MRI identification of MTS between the 2 blinded neuroradiologists and the nonblinded preoperative report (Cohen κ 0.40–0.59). Interrater agreement was highly variable for different MTS features indicative of MTS, ranging from poor to near perfect. Agreement was highest for increased hippocampal signal and decreased temporal lobe size and was consistently poor for reduced gray-white matter demarcation. The sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and proportion perfect agreement were highest for increased hippocampal signal and reduced hippocampal size. An MRI finding of MTS was not predictive of seizure outcome in this small series. Conclusions Mesial temporal sclerosis identification on brain MRI in children evaluated for medial temporal resections has a PPV of 55%–67% and an NPV of 79%–87%. Increased hippocampal signal and reduced hippocampal size were associated with high predictive values, while gray-white differentiation and an enlarged temporal horn were not predictive of MTS. Seizure outcome following medial temporal resections was not associated with MRI findings of MTS or MRI abnormalities indicative of MTS in this small sample size.