Abstracts

The value of proton magnetic resonance spectroscopy imaging (¹H-MRSI) in the pre-surgical evaluation of patients with intractable temporal lobe epilepsy (TLE) remains controversial. Early reports suggested that ¹H-MRSI might serve as a useful lateralizing test. However, later reports have refuted that impression. Some authors have noted variable results in the same subject, including improvements in contralateral hippocampal spectra following ipsilateral epilepsy surgery. Thus, ¹H-MRSI may partly reflect hippocampal physiology or [ldquo]health[rdquo] rather than simply neuronal and glial cell count ratios. Our aim is to prospectively examine the predictive value of ¹H-MRSI in patients with intractable TLE.
Consecutive patients with intractable TLE (n=14) undergoing standard pre-surgical test batteries prospectively underwent ¹H-MRSI. Voxels measuring 2cm x 2cm were placed over bilateral mesial temporal structures with standard protocol. ¹H-MRSI spectral peaks corresponding to n-acetyl aspartate (NAA) were quantified and normalized with respect to creatine and choline. Spectral results were then categorized as normal, bilaterally symmetrically reduced, or asymmetrically reduced. Proton spectroscopy was then compared for concordance to independently acquired data including LTM, MRI, PET, neuropsychologic and Wada testing.
¹H-MRSI showed bilaterally reduced NAA ratios in 10 of 14 patients, with asymmetric reduction in 6 of these 10. One patient had normal ratios bilaterally. ¹H-MRSI results were concordant with ictal LTM results in only 6/12 and with interictal LTM data in 7/13. Comparisons with MRI and PET were similarly mediocre (8/14 and 6/13, respectively). Greatest concordance (11/14) was observed with neuropsychologic testing. Wada memory testing was concordant in 6/10. MRI and PET were highly predictive of the physiologic focus. Only 3 of 14 patients had non-localizing MRI or PET scans. One of these patients had bilateral foci per invasive LTM. Their ¹H-MRSI showed bilaterally, asymmetrically reduced ratios, but indistinguishable from several with unifocal disease including 2 with successful unilateral temporal resections. The only patient with bilaterally normal ratios has had an imperfect surgical outcome.
¹H-MRSI offers mediocre localizing value for TLE patients undergoing pre-surgical evaluation. It appears to be insufficiently specific and possibly overly sensitive toward hippocampal dysfunction to reliably predict lateralization of seizure foci. Still, it most closely parallels results of neuropsychologic testing and may therefore provide an independent measure of hippocampal health. More data will accrue as our cohort expands and others undergo invasive LTM and/or surgery.