Glutamate concentrations in human brain using single voxel proton magnetic resonance spectroscopy at 3 Tesla

Abstract

A method for quantitative determination of the glutamate (Glu) concentration in human brain using PRESS-based single voxel MR spectroscopy (MRS) at 3 T has been developed and validated by repeatedly analyzing voxels comprising the anterior cingulate cortex (acc) and the left hippocampus (hc) in 40 healthy volunteer brains. At an optimum echo time of 80 ms, the C4 resonance of Glu appears well resolved and separated from major interferents, that is, glutamine and N-acetylaspartate. As a complementary method, a multiple quantum coherence filter sequence for Glu was employed. For quantification of Glu and the principal MRS-visible metabolites as well as for an estimate of the glutamine level, analysis of both types of in vivo spectra was carried out by a time domain-frequency domain method involving prior knowledge obtained from phantom spectra. Using PRESS, coefficients of variation (CV) for Glu concentration were of the order of 10%. When the concentrations were corrected by individual cerebrospinal fluid fractions obtained by segmentation using spm, CVs tended to increase and the correlation coefficients for the two MRS sessions tended to decrease, indicating that this type of correction adds uncertainty to the data. The concentrations of Glu in the two voxels studied were found to be significantly different (11.6 mmol/l in acc, 10.9 mmol/l in hc, P = 0.023) and decrease with age (P < 0.04). These concentrations agreed well with those determined using the quantum coherence filter method although the uncertainty of the latter limits reliable analysis.

Introduction

l-glutamate is the major excitatory neurotransmitter in mammalian brain. Besides, it is involved in protein and fatty acid synthesis in brain. As the influx of glutamate across the blood–brain barrier is much lower than its efflux (Palmada and Centelles, 1998), glutamate metabolism itself is decisive for glutamate regulation in the brain. Among the precursors and reaction products of the glutamate system, glutamine plays a major part, besides α-ketoglutarate and γ-aminobutyric acid (GABA). Given the importance of glutamate and the diversity of the enzyme systems involved and of the associated membrane receptors and their regulators Danboldt, 2001, Palmada and Centelles, 1998, Ross, 1991, it is not surprising that the glutamate system plays an emerging role in the study of psychiatric disorders, such as epilepsy Petroff et al., 1995, Petroff et al., 2002, Simister et al., 2003, schizophrenia Bartha et al., 1997, Bartha et al., 1999, Kegeles et al., 2000, Théberge et al., 2002, and affective disorders Auer et al., 2000, Pfleiderer et al., 2003, and elements of this system might be identified as targets for novel drugs against such disorders Goff and Coyle, 2001, Ketter and Wang, 2003, Krystal et al., 2002. Consequently, practical and reliable methods for the in vivo quantification of glutamate in the relevant brain regions are required.

Owing to its molecular structure, glutamate (Glu) gives rise to a complex proton NMR spectrum characterized by the coupled spins of the C2–C4 hydrogen nuclei. At the moderate but prevalent field strength of 1.5 T, the in vivo brain spectrum in the respective spectral ranges exhibits poor resolution and, despite the relatively high Glu concentration of 7–12 mmol/l (see below), low sensitivity due to heavy overlap and rapid dephasing of the multiplets. Primary approaches to the in vivo detection of the coupled spin system of Glu have relied on the acquisition of spin echoes or stimulated echoes using short echo times, targeting at the C4 proton multiplet resonance around 2.35 ppm. Although keeping signal losses due to J-modulation low, such spectra suffer from substantial contributions by glutamine as well as ill-defined background features. In conventional spin echo spectroscopy at 1.5 T, the resonances in this range are therefore mostly assigned to a mixture of Glu and glutamine (and sometimes GABA), designated Glx. Glutamate determination using MRS at higher B₀, such as 4 T Kassem and Bartha, 2003, Théberge et al., 2002 or 7 T (Tkač et al., 2001) may lead to higher selectivity and accuracy but is usually unavailable to a wider range of users. Spectral editing methods for Glu making use of higher magnetic field strengths to decrease the effects of strong coupling have also been devised Lee et al., 1995, Pan et al., 1996, Thompson and Allen, 1998. Though having the potential of good background suppression and fair glutamine separation, the promising multiple quantum coherence filter operated at 3 T devised by Thompson and Allen (1998) is limited by its low efficiency which requires long acquisition times. However, with the increasing availability of 3 T scanners in research and clinical facilities the implementation of an arsenal of methods truly exploiting that field strength becomes highly desirable, including reliable spectroscopy beyond the usual singlets of N-acetylaspartate (NAA), creatine + creatine phosphate (tCr), and choline-containing compounds (tCho). Thus, a promising method for GABA measurement at 3 T has recently been proposed, offering a solution to the problem of the glutamate multiplet interfering with the GABA target resonance around 2.3 ppm (Hanstock et al., 2002).

The present work aims at the development of a method for quantitative determination of the glutamate (Glu) concentration in human brain using single voxel MR spectroscopy at 3 T that would be reasonably selective and at the same time practical and reliable in the clinical setting. To this end, conventional point resolved spectroscopy (PRESS) was optimized at 3 T; complementarily the above-mentioned multiple quantum filter sequence for glutamate (Thompson and Allen, 1998) was modified and implemented, and its performance for glutamate measurement in brain compared with that of the former method. For quantification a newly developed, versatile time domain-frequency domain method of data analysis involving prior knowledge obtained from phantom spectra was employed for both types of in vivo spectra. This method package was applied to the repeat study of two important cortical voxels, namely the anterior cingulate and the left hippocampus, in 40 healthy volunteer brains to create a database of glutamate concentrations for forthcoming studies, for example, of psychiatric disorders. Since the PRESS method also provides the concentrations of the principal MRS-visible metabolites NAA, tCr, and tCho as well as an estimate of the glutamine level, these parameters were included in the analysis.