Voxel MR Spectroscopy of Hippocampal Structures in Healthy Adults at 1.5 Tesla – How Stable are the Results?

Purpose: Although ¹H-MRS is a non-invasive diagnostic tool for measuring metabolite alterations in patients with temporal lobe epilepsy, there are many sources of error, especially with regard to temporomesial structures. In order to meet the requirements of diagnostic tools, we evaluated the inter-subject variability and the test-retest stability of hippocampal single voxel ¹H-MRS. Methods: We used an optimized, standardized method of short echo time ¹H-MRS to study the hippocampal structures of 30 healthy adults. Spectral analysis and metabolite quantitation of N-acetylaspartate (tNAA), choline (Cho), creatine (Cr), total glutamate plus glutamine (Glu+Gln) and myo-inosidol (Ins) were carried out using an LC-Model, using water as an internal reference. After 4–6 weeks, 15 subjects were scanned again to determine long-term reproducibility. Results: Mean metabolite quantitation (mean value±SD) was 6.9±0.38 mM for tNAA, 5.3±0.52 mM for Cr, 1.2±0.15 mM for Cho, 5.0±0.63 mM for Ins and 10.1±1.48 mM for Glu+Gln. In terms of absolute quantitation, tNAA showed the smallest coefficient of variation (CV) of 5.6%, followed by Cr (9.8%), Cho (12.2%) and Ins (12.5%). Less stable was Glu+Gln (15%). Long-term reproducibility was tested by comparing the means of the metabolites of both sessions by paired t-test (tNAA: p=0.789, T=-0.273, df=14; Cho: p=0.429, T=-2.236, df=14; Ins: p=0.734, T=0.346, df=14; Glu+Gln: p=0.121, T-1,653, df=14; Cr: p=0,042, T=-2.236, df=14). Conclusion: Our results show that ¹H-MRS of hippocampal structures can be performed in a stable and reproducible way in routine clinical diagnostics.