Hippocampal T2 relaxometry in epilepsy: past, present, and future.

Mesial temporal sclerosis (MTS) is the most common known pathologic substrate of epilepsy. Accurate preoperative identification of unilateral MTS bymagnetic resonance (MR) imaging has had a tremendous impact on the clinical management of medically refractory seizures. It was well known in the pre-MR era that obtaining a temporal lobectomy specimen that contained MTS conferred a higher probability of excellent postoperative seizure control for the patient than if the temporal lobectomy specimen was free of disease. However, that information was available only after the surgery had been performed. MR imaging now provides this information before surgery and the surgical risk-benefit decision-making process is made in a much better informed fashion. Patients with clear-cut unilateral MTS on MR images can be cited a 70% to 90% probability of being free of seizures after a temporal lobectomy (1, 2). In contrast, patients without MTS (or other epileptogenic lesions) on preoperative MR images have less than a 50% probability of being seizure free after lobectomy (1, 2). In turn, the frequency with which prolonged invasive electroencephalographic monitoring is used has declined dramatically. Invasive monitoring is generally considered unnecessary at this time in patients with clear-cut MR-identified unilateral MTS in whom the scalp-recorded electroencephalographic ictal onset is coincident with the side of the MR-identified abnormality. End-stage or mature MTS is characterized histologically by cell loss and astrogliosis throughout medial temporal lobe limbic areas, but particularly in the hippocampal formation (3). Although several MR abnormalities have been described in association with MTS, the two principle MR findings in histologically proved cases of MTS are hippocampal atrophy and signal change indicative of increased tissue-free water (increased signal on T2-weighted images). Both of these principle MR correlates of MTS can be quantified. Tissue hydration is quantified via T2 relaxometry (4) and hippocampal atrophy is quantified via hippocampal volumetry (5). The most common radiologic manifestation of MTS seen in clinical practice is a unilateral atrophic hippocampus with increased signal, with a normal-appearing contralateral hippocampus. The surgical approach to temporal lobe epilepsy (temporal lobectomy) is also driven by the concept that MTS is a unilateral phenomenon. However, autopsy studies and, more recently, quantitative MR studies (volumetry and T2 relaxometry) indicate that MTS is present bilaterally in a substantial percentage of patients with temporal lobe–onset seizures (3, 4, 6). For the sake of illustration, the entire spectrum of MTS can be divided into four possible conceptual categories (6): (a) unilateral hippocampal damage, in which MTS is present unilaterally, and the contralateral hippocampus is completely normal, (b) bilaterally asymmetric damage, in which MTS is present bilaterally, but is more severely represented on one side, (c) bilaterally symmetric damage, in which MTS is present and of equivalent magnitude in both hippocampi, and (d) symmetric normal hippocampi, in which neither hippocampus has changes of MTS. This fourth category is conceptually useful in the context of this discussion, because distinguishing mild MTS from a normal hippocampus is often not straightforward, either with MR imaging or with qualitative pathologic analysis. These four groups represent conceptual points along a continuous distribution of hippocampal damage ranging from normal to severe MTS in one or both hippocampi (6). Most cases of MTS encountered for presurgical evaluation in general clinical practice will have hippocampal atrophy, increased signal, or, more commonly, both (7–11). The accuracy of visual inspection of an appropriately per-