Abstract

Ageing is associated with reduction of grey matter volume and it is reported that the frontal lobes are preferentially affected. We have applied quantitative magnetic resonance spectroscopy (MRS), incorporating measurement of brain tissue water content and metabolite T2 relaxation times, to determine absolute concentrations of the putative neuronal marker N-acetylaspartate (NAA), creatine (Cr) and choline (Cho) compounds in the frontal lobe of 50 male subjects aged between 20 and 70 years (10 per decade). The fractional brain water content (βMR) did not change significantly as a function of age (r = 0.07, P = 0.65) and had a mean value of 81% (CV = 2%). The concentration (in millimoles per litre brain tissue) of NAA decreased significantly with age (r = –0.42, P = 0.003), with an overall decrease of 12% between the third and seventh decades. The concentrations of Cr and Cho did not change significantly with age. The interpretation of the age-dependent decrease in NAA concentration as reflecting either a reduction in neuronal volume, number or function is discussed.

Introduction

The effect of ageing on the human brain has been studied extensively at the macroscopic (Cowell et al., 1994; Raz, 1996; Coffey et al., 1998) and microscopic levels (Terry et al., 1987; Brody, 1992; Pakkenberg and Gundersen, 1997). By using unbiased modern stereological methods, Pakkenberg and Gundersen observed a decrease in the total number of neurons in the brain of 10% over the range 20–90 years (Pakkenberg and Gundersen, 1997). They also reported that neuronal density remained constant over the same age range, implying that the reduced neuron number relates directly to reduced brain volume. This corroborated an earlier study by Terry et al., which also demonstrated constant neuronal density over a similar age range (Terry et al. 1987). We have used magnetic resonance spectroscopy (MRS) to investigate whether these reported macroscopic and histological findings are associated with changes in the water content and cerebral metabolite concentrations of frontal lobe grey matter in healthy human subjects.

MRS offers the unique possibility of non-invasive measurement of the concentrations of several cerebral metabolites, in particular, N-acetylaspartate (NAA), shown to be present in neurons but not mature glial cells (Urenjak et al., 1993); creatine/phosphocreatine (Cr/PCr), involved in cellular bioenergetics (Miller, 1991); choline-containing compounds (Cho), involved in membrane synthesis and production of acetylcholine (Miller, 1991); glutamine/glutamate (Glx), concerned with neurotransmission (Ross, 1991) and myo-inositol (mI), thought to be present mostly (Brand et al., 1993), though not exclusively (Moore et al., 1999), in glial cells.

MRS experiments are of two kinds: those that measure ratios of metabolite peak areas or signal amplitudes, and those in which absolute concentrations are measured by appropriate calibration. Of the two, the fully quantitative methods are generally more complex, and several different approaches are available (Barker et al., 1993; Danielsen et al., 1995; Kreis, 1997). Full quantification requires ‘compartmentation analysis’ (Ernst et al., 1993), which measures and corrects for the water content of brain tissue (βMR) and the relative fractions of brain water, cerebrospinal fluid (CSF) and solid brain matter in the MRS voxel. This allows concentrations to be expressed in units of millimoles per litre brain tissue or brain water.

Qualitative MRS studies have reported an age-related decrease in the ratio of NAA to total creatine (NAA:Cr), which was attributed to a decrease in NAA concentration (Bruhn et al., 1992; Charles et al., 1994; Fukuzako et al., 1997), assuming the concentration of Cr to be constant. Five studies (Chang et al., 1996; Soher et al., 1996; Pfefferbaum et al., 1999b; Saunders et al., 1999; Leary et al., 2000) have measured metabolite concentrations, and found significantly increased Cr and/or Cho with age, but unchanged NAA. This suggests that the observed decreases in the NAA:Cr ratio are due to increased Cr rather than decreased NAA. However, only one of these studies (Chang et al., 1996) applied compartmentation analysis. In addition, the finding of increased Cr concentration with age is somewhat unexpected given positron emission tomography (PET) evidence of reduced cerebral metabolism with age (Loessner et al., 1995; Murphy et al., 1996), and the known role of Cr/PCr in cellular bioenergetics (Stryer, 1995). Also, recent studies (De Stefano et al., 1999; Jung et al., 1999) have indicated a possible role for NAA as an indicator of neuronal metabolism/function whose concentration might also be expected to fall with age. Interestingly, a recent MRS study of Down's syndrome (Huang et al., 1999) using image segmentation-based CSF correction for absolute concentration estimates from parietal and occipital lobe voxels, reported an age-related decrease in NAA in both patients and age-matched controls.

In the present study we determined absolute metabolite concentrations in 50 male subjects aged between 20 and 70 years, using a recently described method for compartmentation based on the application of standard MR imaging sequences (Brooks et al., 1999b). By acquiring metabolite spectra at several echo times and with long relaxation delay, we have corrected for changes in T2 relaxation time and minimized the effect of T1 relaxation. We have studied frontal lobe grey matter of males, since this is a region where age-related morphological and metabolic changes are reportedly most pronounced (Murphy et al., 1996; Raz et al., 1997).

Materials and Methods

Subjects

Fifty healthy male subjects (10 subjects per decade between 20 and 70 years, mean age = 45.5 years) gave fully informed written consent to this study, which had local Research Ethics Committee approval. Subjects were screened for the presence of neurological, renal, cardiological, psychiatric or endocrinological disease. No subject had abnormally elevated blood pressure (i.e. >160/100 mmHg); two subjects were well-controlled on treatment for hypertension, and the possible effect of this on the measurements is addressed later (see Discussion). Data were acquired using a 1.5 tesla SIGNA whole body MR imaging system (General Electric Medical Systems, Milwaukee, WI, USA) with 5.5 software and standard quadrature head coil. During scanning the head was held comfortably by foam padding. For calibration an external standard containing pure water was attached at a fixed position inside the coil, next to the head.

MRI Acquisition

Images suitable for voxel positioning and estimation of voxel grey and white matter tissue fractions (see Image Processing section below) were acquired in the coronal plane with a fast-inversion recovery prepared gradient echo (F-IRp-GRASS) sequence (TE/flip/TI = 3.5 ms/30°/450 ms, 1.6 mm slice thickness, 124 slices, FOV = 20 cm). From these images a 2 × 2 × 2 cm3 voxel for MRS was prescribed anterior to the genu of the corpus callosum, on the medial aspect of the frontal lobes. The voxel contained mostly grey matter, including cingulate gyrus and prefrontal cortex, as demonstrated by the F-IRp-GRASS images of a young (27 years) and an elderly (68 years) subject shown in Figure 1. Imaging-based compartmentation was performed as described elsewhere (Brooks et al., 1999b). In particular, the total water fraction of the voxel (brain water plus CSF) was determined from proton density (PD) weighted images (Alger et al., 1993; Hittmair et al., 1994) acquired with a fast spoiled gradient echo (F-SPGR) sequence (TE/flip/TR = 3.1 ms/5°/500 ms) for which the prescribed slice thickness of 2 cm coincided with the voxel. The CSF fraction of the voxel was determined from heavily T2-weighted images (Videen et al., 1995) acquired with a fast spin echo (FSE) sequence (TE/TR = 500/5000 ms, echo train length = 8), with four contiguous interleaved 5 mm thick image slices covering the voxel. Both series of images were acquired with a FOV of 20 cm.

MRS Acquisition

Spectra were acquired using the stimulated echo acquisition mode (STEAM) sequence (Frahm et al., 1989). To minimize the influence of susceptibility artefacts arising from the proximity of the voxel to the nasal sinuses, a coronal–sagittal–axial gradient ordering was employed (Ernst and Chang, 1996). Water-suppressed metabolite spectra were recorded from the brain at TEs of 30, 72, 144, 216 and 288 ms, with mixing time (TM) of 13.7 ms, 96 signal averages and with a constant relaxation delay (RD = TR – TE/2 – TM) of 2971.3 ms. For the purpose of eddy current correction and automatic phasing of metabolite spectra (Klose, 1990), unsuppressed water spectra were acquired from the same voxel at each TE. Absolute metabolite concentrations were determined by referencing to the unsuppressed water signal from the external standard, recorded with a single acquisition (TE = 50 ms).

Image Processing

Images were imported into ANALYZE software (Mayo Foundation, Minnesota, USA) for analysis. The fractions of grey and white matter in the voxel were computed from the 3D high-resolution anatomical F-IRp-GRASS images using Statistical Parametric Mapping (SPM 96) software (Ashburner and Friston, 1997). SPM generates four series of segmented images, corresponding to grey matter, white matter, CSF and ‘other’ (i.e. meninges, extra-cerebral tissue, etc.). The intensity scale of the segmented images ranges from 0 to 100, with 100 indicating a pixel comprised entirely of one tissue type (e.g. 100% grey matter), and allows calculation of the relative amounts of the three main tissue types in the voxel (Brooks et al., 1999a) (see Fig. 1).

Compartmentation was performed as described elsewhere (Brooks et al., 1999b). The total water fraction of the voxel was estimated from F-SPGR images by recording the mean signal intensity of the region of interest (ROI) corresponding to the selected voxel. This was divided by the mean signal intensity of an ROI placed in the external standard, to give the fractional MR-visible water content (brain water plus CSF) of the voxel, fTW = fBW + fCSF. The fraction of MR-invisible solid brain matter (fSB) was calculated as 1 – fTW. The fraction of CSF (fCSF) in the voxel was calculated from FSE images by recording the mean signal intensity of the voxel and dividing by the signal intensity from pure CSF, assessed by selecting multiple ROIs containing only CSF. The result was subtracted from fTW to yield fBW. The fractional water content of brain tissue in the voxel (βMR) was calculated from these quantities, using equation (1) below:  

1
\[{\beta}_{\mathit{MR}}\ {=}\ \frac{\mathit{f}_{BW}}{\mathit{f}_{BW}\ {+}\ \mathit{f}_{SB}}\ {=}\ \frac{\mathit{f}_{BW}}{1-\mathit{f}_{CSF}}\]
where fBW + fCSF + fSB = 1.

Spectral Processing

MRS data were processed using the Magnetic Resonance User Interface software (MRUI, version 96.3). Metabolite spectra were line-shape corrected to remove the effect of eddy currents, and zero-order phase adjusted where necessary. Residual water signal in metabolite spectra was removed using the Hankel–Lanczos single value decomposition (HLSVD) technique available in MRUI (Pijnappel et al., 1992). Following eddy current correction the dominant factor affecting line-widths was the local voxel shim, thus all metabolite peaks were constrained to be of the same width (for a given TE) and assumed to be Lorentzian for subsequent time domain fitting using VARPRO (Van der Veen et al., 1988). Measurement of metabolite signal amplitudes was operator-independent, and only required identification of the peaks to be measured and an estimate of the typical line-width (measured from the NAA peak).

Metabolite T2 values and T2-corrected signal amplitudes were computed from signal amplitudes recorded at 72, 144, 216 and 288 ms, using a mono-exponential model and a non-linear least squares fitting routine available in MathCAD software (Cherwell Scientific Publishing, Oxford, UK). Correction was made for the effect of partial saturation using T1 relaxation times for NAA, Cr, Cho and brain water in grey matter of 1170, 1150, 1270 and 1144 ms respectively (Kreis et al., 1993; Cho et al., 1997). The full expression for metabolite signal amplitudes (S0) corrected for T2 relaxation occurring during TE and T1 relaxation during TM and RD, is given by equation (2).  

2
\[\mathit{S}_{0}\ {=}\frac{\mathit{S}}{exp\left({-}\frac{\mathit{T}_{E}}{\mathit{T}_{2}}\right){\cdot}\ exp\left({-}\frac{\mathit{T}_{M}}{\mathit{T}_{1}}\right){\cdot}\left[1{-}exp\left({-}\frac{\mathit{R}_{D}}{\mathit{T}_{1}}\right)\right]}\]
where S is the signal recorded at any TE, TM, RD combination.

The signal amplitude from the voxel placed in the external standard (SEXT) was obtained by integration, using SAGE/IDL software (General Electric Medical Systems) with fixed integration limits of ±200 Hz. SEXT was corrected for relaxation effects using T2,EXT, which was determined to be 1660 ms (SD = 40, n = 3) in a separate experiment.

Metabolite Concentrations

Metabolite concentrations were estimated by scaling the relaxation- corrected metabolite signal amplitudes (S0,met, see equation 2) to the signal measured from the external standard (SEXT), whose water concentration ([cH2O]) is equal to the molar concentration of pure water (55.5 mol l–1), and by correcting for the amount of CSF in the voxel. MRI-based compartmentation analysis allowed metabolite concentrations to be referenced to either the fraction of brain tissue or brain water in the frontal lobe voxel (Kreis et al., 1993). Thus concentrations could be expressed in two ways: either per litre brain tissue (CB)  

3
\[{[}\mathit{C}_{B}{]}_{met}\ {=}\ \frac{\mathit{S}_{0},met}{\mathit{S}_{EXT}}{\cdot}\frac{1}{1\ {-}\ \mathit{f}_{CSF}}{\cdot}{[}\mathit{c}_{H2O}{]}{\cdot}\frac{2}{\mathit{n}}\ mol/1\ of\ brain\ tissue\]
or per litre brain water (CBW)  
4
\[{[}\mathit{C}_{BW}{]}_{met}\ {=}\ \frac{\mathit{S}_{0},met}{\mathit{S}_{EXT}}{\cdot}\frac{1}{\mathit{f}_{BW}}{\cdot}{[}\mathit{c}_{H2O}{]}{\cdot}\frac{2}{\mathit{n}}\ mol/1\ of\ brain\ water\]

The symbol n in the denominator of equations (3) and (4) refers to the number of hydrogen atoms associated with each functional group (NAA 3, Cr 3, Cho 9); the factor 2 in the numerator accounts for the two hydrogen atoms per water molecule. Note that [CB]met and [CBW]met are related to one another by βMR:  

5
\[{[}\mathit{C}_{B}{]}_{met}\ {=}\ {\beta}_{MR}{\cdot}{[}\mathit{C}_{BW}{]}_{met}\]

Results

Image Analysis

All subjects had normal appearance on MRI. The average grey matter content as a fraction of total brain tissue (i.e. grey + white matter) in the voxel in the 50 subjects was 85% (CV = 6%), and did not depend on age (r = –0.2, P = 0.16). However, the actual amount of brain tissue in the voxel tended to decrease and CSF increase with age, in accordance with the anticipated effect of normal age-related brain atrophy — see the results of the SPM segmentation of the MR images of a young (27 years) and an elderly subject (68 years), shown in Figure 1.

Variation in Brain Water Content (βMR) and CSF Fraction (fCSF) with Age

Values of βMR obtained for all 50 subjects by compartmentation analysis are presented in Figure 2. The value of βMR remains constant over the entire age range (r = 0.07, P = 0.65). However, Figure 3 shows that fCSF increased significantly with age (r = 0.50, P < 0.001), as might be expected on the basis of age-related brain atrophy (Lim et al., 1992) (both βMR and metabolite concentrations were corrected for changes in fCSF). This corroborates the SPM analyses, which also show increased CSF fraction with age (see Fig. 1).

Metabolite T2 Relaxation Times and Concentrations

Representative STEAM spectra (TE = 72 ms) acquired for the subjects shown in Figure 1 are presented in Figure 4, scaled so that the peak heights of Cr are equal in the two subjects. Also shown are the corresponding metabolite concentrations; the differences between them are the reduced NAA peak and (in this case) the higher Cho peak in the elderly subject.

The metabolite T2 relaxation times for all 50 subjects of NAA, Cr and Cho, estimated using spectra recorded at 72, 144, 216 and 288 ms, are plotted against age in Figure 5. No significant age dependence was observed. The mean (SD) T2s for NAA, Cr and Cho are 369 (56), 221 (32) and 556 (214) ms, respectively.

The age dependences of the concentrations of Cr and Cho are shown in Figure 6, expressed in units of millimoles per litre brain tissue. The equivalent data expressed per litre brain water are plotted as dashed regression lines. No effect of age on Cr or Cho concentration was observed (see Table 1). Figure 7a shows the age dependence of the concentration of NAA, expressed per litre brain tissue and (as the dashed regression line only) per litre brain water. The regression lines for concentrations estimated either per litre brain tissue or per litre brain water are almost parallel (see Figs 6 and 7), because the concentration measures are related to one another by βMR, which remains essentially constant with age (see Fig. 2). Thus, expressed as millimoles per litre brain tissue or per litre brain water, NAA concentration showed a significant negative correlation with age (see Table 1). The ratio of NAA to Cr is plotted against age in Figure 7b, and also shows a strong negative correlation with age (see Table 1).

Discussion

The main findings of this study are that in a frontal lobe voxel, water content of brain tissue (βMR), relaxation times for NAA, Cr and Cho and concentrations of Cr and Cho remain constant with age, while the concentration of NAA and the ratio NAA/Cr both decrease with age (P < 0.01). This study highlights the need for compartmentation analysis in measuring brain metabolite concentrations, as the fraction of CSF increased significantly (P < 0.001) with age. Possible technical limitations, the relationship of these results to published data and some of their implications are discussed below.

Technical Considerations

We considered the possibility that in estimating the tissue composition of the voxel, error might occur due to mismatch between the cube of data extracted from the segmented 3D volume and the actual shape of the spectroscopy voxel. However, the STEAM pulse sequence generates voxels of well-defined dimensions (Keevil et al., 1995), and the bandwidth of the slice selection pulses is such that the chemical shift artefact is reduced, thus minimizing potential mis-registration between imaging and spectroscopy data. Another potential source of error is the use of an external standard for referencing metabolite concentrations. Whilst such methods are sensitive to inhomogeneity of the B1 field inside the head coil, we minimized errors by consistent positioning of the external standard, and in separate experiments found negligible variation in B1 over the central region (which contains the frontal lobes) of the circularly polarized bird-cage coil (Keevil et al., 1998). Thus, any in-accuracies relating to the use of an external standard would contribute only a systematic error to the measured metabolite concentrations, which could not depend on a subject's age.

One possible limitation of the data analysis technique employed in this study is the use of a linear function to model age-related changes in metabolite concentrations. Previously, Murphy et al. demonstrated that brain atrophy and hypo-metabolism increase in healthy subjects after the age of 60 years (Murphy et al., 1996). The possibility that age-dependent changes in metabolite concentrations were non-linear was investigated by fitting the obtained data to a quadratic function. Statistical comparison of the residual variances following quadratic and linear fitting showed no significant improvement of fit with a non-linear model (F = 1.06). With an increased number of subjects one might be able to discern non-linear changes in the metabolite data. On the basis of results obtained from the 50 controls examined in this study, we did not detect non-linear metabolite changes with age.

Time constraints inherent in single voxel 1H-MRS meant that we have only been able to obtain quantitative MRS data for a single frontal lobe voxel. Beyond possible effects of age and sex differences in the subjects studied, inherent spatial heterogeneity of brain water content and also cerebral metabolite concentrations may explain the differences between our data and values reported for other frontal lobe regions (Chang et al., 1996), as well as regions of occipital lobe (Huang et al., 1999; Saunders et al., 1999), parietal lobe (Huang et al., 1999; Saunders et al., 1999; Leary et al., 2000) and whole-brain average values derived from magnetic resonance spectroscopic imaging (MRSI) studies (Soher et al., 1996; Pfefferbaum et al., 1999b).

Water Content

In an earlier spectroscopy study of frontal lobe grey matter (Chang et al., 1996), βMR was reported to decrease from ~81 to 56% over the age range 19–78 years. This mixed-sex study, which used a spectroscopy-based compartmentation technique, with an unequal distribution between young (<45 years, n = 24) and old (≥45 years, n = 9) subjects, does not agree with our finding of constant βMR with age. Two lines of evidence support our conclusion. First, brain tissue water content may be inferred from in vivo measurements of T1 relaxation time, to which it is linearly related (MacDonald et al., 1986; Paczynski et al., 1997). In a study of 107 subjects aged between 4 and 72 years, T1 of cortical grey matter was 1160 ms at 20 years, decreasing to 1070 ms by 70 years (Cho et al., 1997). Given the established relationship between T1 and water content (MacDonald et al., 1986; Paczynski et al., 1997), this suggests less than 3% reduction in brain water content over the age range of the present study, consistent with our results. Secondly, brain water content can be determined ‘biochemically’ as wet weight minus dry weight. Using this technique, in 139 subjects who died aged 10 weeks to 78 years the water content of forebrain fell from ~95% at gestation to 82% by age 7 years, with only a modest decline to adulthood, where it ranged from 77 to 80% (Dobbing and Sands, 1973). In an autopsy study of grey matter from the medial and superior frontal gyri of 35 previously healthy individuals, brain water content increased by 1.5% over the age range 30–90 years (Wiggins et al., 1988). Thus, consistent with the current study, neither biochemical studies ex vivo nor the study of T1 relaxation times in vivo support a significant reduction in brain water content with age, of the magnitude of 25% between 20 and 70 years as reported by Chang et al. (1996).

Relaxation Times

Measured relaxation times may depend on the technique chosen for estimating peak signal amplitudes (or peak areas) at each TE. Particularly at short TE, signal amplitude may be sensitive to contributions from macromolecules constituting the ‘rolling baseline’ beneath the metabolite peaks (De Beer et al., 1998). An additional complicating factor is the existence of metabolites in two or more pools (Assaf and Cohen, 1998) with different relaxation properties (Kreis et al., 1993), giving rise to multi- exponential relaxation behaviour. By fitting signal amplitudes of the four longest TE spectra (i.e. 72, 144, 216 and 288 ms) to a mono-exponential function, we sought to minimize baseline effects (as TE ≥ 72 ms) and record the relaxation characteristics of the less restricted metabolite pool (i.e. the long T2 component). The measured T2s were in good agreement with published values (Kreis et al., 1993; Danielsen et al., 1995), with the exception of the Cho data, which were 20% larger and associated with a large CV of 39%. The reduced signal-to-noise ratio of the Cho peak compared with NAA and Cr is a well-known problem, exacerbated by the location of the voxel in the frontal lobe, relatively close to the sinuses (Ernst and Chang, 1996).

Metabolite Concentrations: Technical Issues

We have found a significant age-dependent reduction in the concentration of NAA in frontal lobe grey matter of healthy male subjects, but constant Cr and Cho. In contrast, five quantitative MRS studies of ageing have found unchanged NAA concentration but significantly increased Cr and/or Cho with age (Chang et al., 1996; Soher et al., 1996; Pfefferbaum et al., 1999a; Saunders et al., 1999; Leary et al., 2000), suggesting that the decrease in the NAA:Cr ratio reported in several qualitative studies (Bruhn et al., 1992; Charles et al., 1994; Fukuzako et al., 1997) is due to an increase in Cr rather than a decrease in NAA. One possible confounding variable in these published studies is the mixed sex nature of the groups examined; it is possible that females show smaller metabolic changes with age than males, and there is some evidence from PET studies to support this (Murphy et al., 1996). However, as discussed below, several technical issues may contribute to the discrepancies between these studies and the present findings.

Chang et al. (1996) studied the same region of the brain as in the present study. However, as discussed above, their observation of a large decrease in βMR is not consistent either with our compartmentation data or with published biochemical data (Wiggins et al., 1988). As the value of βMR is the result of compartmentation analysis, errors in this suggest technical problems relating to data acquisition or processing, which will also influence estimated metabolite concentrations. Soher et al. (1996) used magnetic resonance spectroscopic imaging (MRSI) and a phantom replacement technique to estimate metabolite concentrations for several voxels distributed throughout the brains of volunteers aged between 5 and 74 years. They reported a significant increase in Cho with age, especially in white matter. While the mean ‘whole brain’ metabolite concentrations obtained were in good agreement with published values, the concentration of NAA recorded from cortical voxels was lower than in white matter [as reported in other MRSI studies, e.g. (Hetherington et al., 1996)]. This work (Soher et al., 1996) did not make any correction for the fraction of CSF in the voxel, which is essential for fully quantitative brain MRS, particularly in studies of the effects of ageing, where CSF fraction may increase substantially (see Fig. 3).

There are several ways of correcting for the amount of CSF in an MRS voxel. Saunders et al. (Saunders et al. 1999) used an internal water referencing scheme (Barker et al., 1993) and CSF correction based on an analysis of T2-weighted images, without full compartmentation analysis. This was used to determine metabolite concentrations in parietal (predominantly white matter) and left occipital (predominately grey matter) lobe voxels, and found that the concentration of Cr increased in parietal white matter with age. Similar results were obtained by Leary et al., who recorded metabolite concentrations in parietal lobe white matter of 44 males and females, and observed significant increases in both Cr and Cho, but constant NAA concentration with age (Leary et al., 2000). Quantification was performed without compartmentation analysis, since voxels were restricted to white matter locations, and as such might reasonably be expected to demonstrate different age-related effects than a voxel place in grey matter. Patterns of brain ageing are heterogeneous (Cowell et al., 1994), and to our knowledge no study has demonstrated a consistent pattern of reduced cortical volume in the parietal or occipital lobe with age, as occurs in the frontal lobe (Raz et al., 1997).

It is interesting that those studies reporting unchanged NAA concentration but significantly increased Cr and/or Cho with age (Pfefferbaum et al., 1999a,b; Saunders et al., 1999) have used an image-based segmentation procedure to correct for CSF that may be vulnerable to partial volume effects in the relatively thick slices used. Pfefferbaum et al. used an MRSI technique to acquire multiple voxel data from entire brain slices (Pfefferbaum et al., 1999a,b), and determined the amount of grey and white matter present in each 1 cm3 MRSI voxel using a dual echo segmentation technique (Lim et al., 1992). Concentrations of NAA, Cr and Cho in pure grey and white matter were determined by expressing recorded metabolite signal in terms of the fractions of grey and white matter in the MRSI voxel, which could be used to obtain the age dependence of pure grey matter concentration. However, this approach relies on ‘harvesting’ voxels from different brain regions and pooling data for subsequent analysis. Possible regional variations in metabolite levels are lost, which may obscure age-related changes in particular brain regions. In addition, this study used both males and females in well-separated age groups (15 males, mean age 25.3 years vs. 9 males/10 females, mean age 73.3 years), which may have contributed variability to the recorded metabolite concentrations.

When investigating possible age-related changes in cerebral metabolite concentrations, it is important to assess the tissue composition of the selected spectroscopy voxel, because the metabolite profiles of grey and white matter have been reported to be different (Kreis et al., 1993; Doyle et al., 1995; Noworolski et al., 1999). By performing SPM analyses of high-resolution 3D data sets we were able to measure the tissue composition of the MRS voxel. The mean fraction of grey matter (as a proportion of total brain tissue, i.e. grey plus white) was 85%, and did not change significantly with age (P = 0.2). Thus our observation of an age-dependent fall in NAA is not due to increased white matter in the voxels of older subjects.

One possible complication in studies of ageing is the presence of other pathology unrelated to ageing per se. The subjects in the present study had no significant history of neurological, renal, cardiological, psychiatric or endocrinological disease. Two subjects in the seventh decade were well-controlled hypertensives: it cannot be assumed a priori that treated hypertension has no effect of brain metabolism (DeCarli et al., 1995). However, omission of these two subjects (aged 62 and 66, NAA 10.1 and 10.7 mmol/l brain tissue, respectively) does not alter the calculated rate of decrease of [NAA] with age.

Metabolite Concentrations: Interpretation

One possible interpretation of our data is that NAA may be maintained at a constant concentration per litre cytosol, and the observed reduction in brain NAA concentration reflects age-dependent reduction in neuronal volume. An alternative interpretation is that neuronal function decreases with age. The possibility that decreased NAA concentration reflects reduced neuronal function has been suggested by several authors (De Stefano et al., 1995, 1999; Jung et al., 1999) and is consistent with PET studies (Loessner et al., 1995; Murphy et al., 1996) reporting reduced metabolic activity with increasing age, particularly in the frontal lobe. However, one must be cautious about the possible influence of age-related atrophy (Lim et al., 1992) on PET measurements.

It has recently been found (Bhakoo and Pearce, 2000) that, contrary to an earlier report (Urenjak et al., 1993), NAA is present in cultures of mature oligodendrocytes. However, given that the age-dependent reduction in frontal lobe volume is associated with loss of neurons rather than glia (Terry et al., 1987), we take our observation of a fall in NAA concentration to reflect a reduction in either neuronal volume or function, rather than a change in the oligodendrocyte population. Reports of unchanged NAA and increased Cr and/or Cho may reflect changes in the neuron:glial ratio occurring in different regions of the brain (Terry et al., 1987). An alternative possibility, as NAA is present in dendrites (Simmons et al., 1991), is that the NAA loss may relate to age-related dendrite dropout.

In summary, using imaging-based compartmentation to determine the effects of ageing on metabolite concentrations recorded from a frontal lobe voxel in 50 male subjects, we found that while CSF fraction increased with age, the water content of brain tissue (βMR) remained constant. Using SPM- based image segmentation, we found that grey matter was a constant fraction of total brain tissue in the voxel. Finally, we measured the relaxation times and compartmentation-corrected concentrations of NAA, Cr and Cho and found that only NAA concentration is significantly correlated (negatively) with age (r = –0.42, P = 0.003). In future studies we will investigate whether region- and sex-specific differences exist in the effect of ageing on NAA concentration. In particular, we will incorporate the imaging-based compartmentation technique described in this paper to the quantification of MRSI data. Two other pertinent applications of the new technique are in quantifying the relationship between cerebral metabolite concentrations and performance on neuropsychological testing (Jung et al., 1999), and employed bilaterally to investigate temporal lobe abnormalities in epilepsy (Connelly et al., 1998).

Notes

We thank Mrs V. McNulty for her assistance with MR image analysis. The MRUI software package was kindly provided by A. van den Boogaart, Katholieke Universiteit Leuven, and is currently funded by the EC project ‘Human Capital and Mobility/Networks’ (HCM-CHRX-CT94-0432).

Address correspondence to Dr J.C.W. Brooks, Magnetic Resonance and Image Analysis Research Centre, University of Liverpool, Pembroke Place, Liverpool L69 3BX, UK. Email: jbrooks@liv.ac.uk.

Table 1

Results from correlation analysis of metabolite concentrations and ratios, with age

Estimation schemea NAA Cr Cho NAA/Cr ratio 
 r b ± SE r b ± SE r b ± SE r b ± SE 
aThe two sets of results correspond to data acquired with the two different metabolite estimation schemes (see Materials and Methods) and summarize the individual data plotted in Figures 5 and 6. Correlation coefficients (r), slope (b) ± SE on the slope and P values for each of the measured metabolites are presented, with significant correlations shown in bold. Note that the ratio data are independent of concentration estimation scheme. 
Per litre brain –0.42 –0.03 ± 0.01 0.15 0.01 ± 0.01 0.05 0.001 ± 0.004 –0.63 –0.005 ± 0.001 
 P= 0.003  P = 0.31  P = 0.72  P< 0.001  
Per litre brain water –0.44 –0.04 ± 0.01 0.14 0.01 ± 0.01 0.04 0.001 ± 0.004 – – 
 P= 0.002  P = 0.35  P = 0.76    
Estimation schemea NAA Cr Cho NAA/Cr ratio 
 r b ± SE r b ± SE r b ± SE r b ± SE 
aThe two sets of results correspond to data acquired with the two different metabolite estimation schemes (see Materials and Methods) and summarize the individual data plotted in Figures 5 and 6. Correlation coefficients (r), slope (b) ± SE on the slope and P values for each of the measured metabolites are presented, with significant correlations shown in bold. Note that the ratio data are independent of concentration estimation scheme. 
Per litre brain –0.42 –0.03 ± 0.01 0.15 0.01 ± 0.01 0.05 0.001 ± 0.004 –0.63 –0.005 ± 0.001 
 P= 0.003  P = 0.31  P = 0.72  P< 0.001  
Per litre brain water –0.44 –0.04 ± 0.01 0.14 0.01 ± 0.01 0.04 0.001 ± 0.004 – – 
 P= 0.002  P = 0.35  P = 0.76    
Figure 1.

 The position of the MRS voxel is illustrated for two subjects (A: aged 27 years, and B: aged 68 years) by drawing its outline on the corresponding F-IRp-GRASS images (furthest left). The voxel was located on the interhemispheric fissure of the medial frontal lobe, such that it included a significant proportion of grey matter and, as far as possible, excluded white matter. The proportions of grey matter, white matter and CSF in the voxel were determined by SPM analysis, and are shown for the two subjects as quantitative tissue maps: these show reduced grey and white matter and correspondingly increased CSF in the elderly subject compared with the young subject.

Figure 1.

 The position of the MRS voxel is illustrated for two subjects (A: aged 27 years, and B: aged 68 years) by drawing its outline on the corresponding F-IRp-GRASS images (furthest left). The voxel was located on the interhemispheric fissure of the medial frontal lobe, such that it included a significant proportion of grey matter and, as far as possible, excluded white matter. The proportions of grey matter, white matter and CSF in the voxel were determined by SPM analysis, and are shown for the two subjects as quantitative tissue maps: these show reduced grey and white matter and correspondingly increased CSF in the elderly subject compared with the young subject.

Figure 2.

 Variation in brain water content (βMR) with age. No significant change with respect to age is observed: P = 0.65, n = 50.

Figure 2.

 Variation in brain water content (βMR) with age. No significant change with respect to age is observed: P = 0.65, n = 50.

Figure 3.

 Variation in the voxel fraction of CSF (fCSF) with age. There is a significant change with respect to age: P < 0.001, n = 50.

Figure 3.

 Variation in the voxel fraction of CSF (fCSF) with age. There is a significant change with respect to age: P < 0.001, n = 50.

Figure 5.

 Linear regression of metabolite T2 relaxation times with age for NAA, Cr and Cho. No significant change with respect to age is observed: P = 0.63, 0.80 and 0.72, respectively, n = 50.

Figure 5.

 Linear regression of metabolite T2 relaxation times with age for NAA, Cr and Cho. No significant change with respect to age is observed: P = 0.63, 0.80 and 0.72, respectively, n = 50.

Figure 6.

 The concentrations of Cr and Cho per litre brain tissue, measured in frontal lobe voxels of 50 male subjects, are plotted with respect to age and the corresponding regression line is shown as a full line. The dashed line is the regression line for concentrations estimated per litre brain water (data points omitted for clarity). No significant effect of age on either Cr or Cho concentration is observed.

Figure 6.

 The concentrations of Cr and Cho per litre brain tissue, measured in frontal lobe voxels of 50 male subjects, are plotted with respect to age and the corresponding regression line is shown as a full line. The dashed line is the regression line for concentrations estimated per litre brain water (data points omitted for clarity). No significant effect of age on either Cr or Cho concentration is observed.

Figure 7.

 The concentration of NAA per litre brain tissue, measured in frontal lobe voxels of 50 male subjects, is plotted with respect to age and the corresponding regression line is shown as a full line (a). The dashed line is the regression line for concentrations estimated per litre brain water (data points have been omitted for clarity). NAA decreased significantly with age according to either concentration estimation scheme (see Table 1 for details). Also plotted is the ratio of NAA:Cr (b) based on T2-corrected signal amplitudes (S0); the observed reduction with age is highly significant (P < 0.001).

 The concentration of NAA per litre brain tissue, measured in frontal lobe voxels of 50 male subjects, is plotted with respect to age and the corresponding regression line is shown as a full line (a). The dashed line is the regression line for concentrations estimated per litre brain water (data points have been omitted for clarity). NAA decreased significantly with age according to either concentration estimation scheme (see Table 1 for details). Also plotted is the ratio of NAA:Cr (b) based on T2-corrected signal amplitudes (S0); the observed reduction with age is highly significant (P < 0.001).

References

Alger JR, Symko SC, Bizzi A, Posse S, DesPres DJ, Armstrong MR (
1993
) Absolute quantitation of short TE brain 1H-MR spectra and spectroscopic imaging data.
J Comput Assist Tomagr
 
17
:
191
–199.
Ashburner J, Friston K (
1997
) Multimodal image coregistration and partitioning – a unified framework.
Neuroimage
 
6
:
209
–217.
Assaf Y, Cohen Y (
1998
) Non-mono-exponential attenuation of water and N-acetyl aspartate signals due to diffusion in brain tissue.
J Magn Reson
 
131
:
69
–85.
Barker PB, Soher BJ, Blackband SJ, Chatham JC, Matthews VP, Nick-Bryan R (
1993
) Quantitation of proton NMR spectra of the human brain using tissue water as an internal concentration reference.
NMR Biomed
 
6
:
89
–94.
Bhakoo K, Pearce D (
2000
) In vitro expression of N-acetyl aspartate by oligodendrocytes: implications for proton magnetic resonance spectroscopy signal in vivo.
J Neurochem
 
74
:
254
–262.
Brand A, Richter-Landsberg C, Leibfritz D (
1993
) Multinuclear NMR studies on the energy metabolism of glial and neuronal cells.
Dev Neurosci
 
15
:
289
–298.
Brody H (
1992
) The aging brain.
Acta Neurol Scand
 
137
:
40
–44.
Brooks, JCW, Roberts N, Kemp G, Gosney M, Spark V, Lye M, Whitehouse G (1999a) Age-related decrease in NAA concentration and NAA/creatine ratio revealed by single voxel 1H-MRS and compartmentation analysis in fifty subjects. Proceedings of the 7th Annual Meeting of ISMRM, Vol 2, Philadelphia, USA, p. 1399.
Brooks JCW, Roberts N, Kemp GJ, Martin PA, Whitehouse GH (
1999
) Magnetic resonance imaging based compartmentation and its application to measuring metabolite concentrations in the frontal lobe.
Magn Reson Med
 
41
:
883
–888.
Bruhn H, Stoppe G, Merboldt KD, Michaelis T, Hanicke W, Frahm J (1992) Cerebral metabolic alterations in normal aging and Alzheimer's dementia detected by proton magnetic resonance spectroscopy. Proceedings of the  11th Annual Meeting of the Society for Magnetic Resonance Medicine, Vol 1, Berlin, Germany, p. 752.
Chang L, Ernst T, Poland RE, Jenden DJ (
1996
) In vivo proton magnetic resonance spectroscopy of the normal aging human brain.
Life Sci
 
58
:
2049
–2056.
Charles HC, Lazeyras F, Krishnan KRR, Boyko OB, Patterson LJ, Doraiswamy PM, McDonald WM (
1994
) Proton spectroscopy of human brain: effects of age and sex.
Prog Neuropsychopharmacol Biol Psychiat
 
18
:
995
–1004.
Cho S, Jones D, Reddick WE, Ogg RJ, Steen RG (
1997
) Establishing norms for age-related changes in proton T1 of normal human brain tissue in vivo.
Magn Reson Imaging
 
15
:
1133
–1143.
Coffey C, Lucke JF, Saxton JA, Ratcliff G, Billig B, Bryan N (
1998
) Sex differences in brain aging: a quantitative magnetic resonance imaging study.
Arch Neurol
 
55
:
169
–179.
Connelly A, Van Paesschen W, Porter DA, Johnson CL, Duncan JS, Gadian DG (
1998
) Proton magnetic resonance spectroscopy in MRI-negative temporal lobe epilepsy.
Neurology
 
51
:
61
–66.
Cowell P, Turetsky B, Gur R, Grossman R, Shtasel D, Gur R (
1994
) Sex differences in aging of the human frontal and temporal lobes.
J Neurosci
 
14
:
4748
–4755.
Danielsen ER, Michaelis T, Ross BD (
1995
) Three methods of calibration in quantitative proton MR spectroscopy.
J Magn Reson B
 
106
:
287
–291.
De Beer R, van den Boogaart A, Cady E, Graveron-Demilly D, Knijn A, Langenberger KW, Lindon JC, Ohlhoff A, Serraî H, Wylezinska-Arridge M (
1998
) Absolute metabolite quantification by in vivo NMR spectroscopy: V. Multicentre quantitative data analysis trial on the overlapping background problem.
Magn Reson Imaging
 
16
:
1127
–1137.
De Stefano N, Matthews PM, Arnold DL (
1995
) Reversible decreases in N-acetylaspartate after acute brain injury.
Magn Reson Med
 
34
:
721
–727.
De Stefano N, Narayanan S, Matthews PM, Francis GS, Antel JP, Arnold DL (
1999
) In vivo evidence for axonal dysfunction remote from focal cerebral demyelination of the type seen in multiple sclerosis.
Brain
 
122
:
1933
–1939.
DeCarli C, Murphy D, Tranh M, Grady C, Haxby J, Gillette J, Salerno J, Gonzales-Aviles A, Horwitz B, Rapoport S, Schapiro M (
1995
) The effect of white matter hyperintensity volume on brain structure, cognitive performance, and cerebral metabolism of glucose in 51 healthy adults.
Neurology
 
45
:
2077
–2084.
Dobbing J, Sands J (
1973
) Quantitative growth and development of human brain.
Arch Dis Child
 
48
:
757
–767.
Doyle TJ, Bedell BJ, Narayana PA (
1995
) Relative concentrations of proton MR visible neurochemicals in gray and white matter in human brain.
Magn Reson Med
 
33
:
755
–759.
Ernst T, Chang L (
1996
) Elimination of artifacts in short echo time 1H MR spectroscopy of the frontal lobe.
Magn Reson Med
 
36
:
462
–468.
Ernst T, Kreis R, Ross BD (
1993
) Absolute quantitation of water and metabolites in the human brain. I. Compartments and water.
J Magn Reson B
 
102
:
1
–8.
Frahm J, Bruhn H, Gyngell L, Merboldt KD, Hanicke W, Sauter R (
1989
) Localised high resolution proton NMR spectroscopy using stimulated echoes: initial applications to human brain in vivo.
Magn Reson Med
 
9
:
79
–93.
Fukuzako H, Hashiguchi T, Sakamoto Y, Okamura H, Doi W, Takenouchi K, Takigawa M (
1997
) Metabolite changes with age measured by proton magnetic resonance spectroscopy in normal subjects.
Psychiatry Clin Neurosci
 
51
:
261
–263.
Hetherington H, Pan J, Mason G, Adams D, Vaughn M, Twieg D, Pohost G (
1996
) Quantitative 1H spectroscopic imaging of human brain at 4.1T using image segmentation.
Magn Reson Med
 
36
:
21
–29.
Hittmair K, Gomiscek G, Langenberger K, Recht M, Imhof H, Kramer J (
1994
) Method for the quantitative assessment of contrast agent uptake in dynamic contrast-enhanced MRI.
Magn Reson Med
 
31
:
567
–571.
Huang W, Alexander GE, Daly EM, Shetty HU, Krasuski JS, Rapoport SI, Schapiro MB (
1999
) High brain myo-inositol levels in the predementia phase of Alzheimer's disease in adults with Down's syndrome: a 1H MRS study.
Am J Psychiatry
 
156
:
1879
–1886.
Jung RE, Brooks WM, Yeo RA, Chiulli SJ, Weers DC, Sibbitt WL (
1999
) Biochemical markers of intelligence: a proton MR spectroscopy study of normal human brain.
Proc R Soc Lond B Biol Sci
 
266
:
1375
–1379.
Keevil SF, Barbiroli B, Collins DJ, Danielsen ER, Hennig J, Henriksen O, Leach MO, Longo R, Lowry M, Moore C, Moser E, Segebarth C, Bovee W, Podo F (
1995
) Quality assessment in in-vivo NMR spectroscopy. 4. A multicenter trial of test objects and protocols for performance assessment in clinical NMR spectroscopy.
Magn Reson Imaging
 
13
:
139
–157.
Keevil SF, Barbiroli B, Brooks JC, Cady EB, Canese R, Carlier P, Collins DJ, Gilligan P, Gobbi G, Hennig J, Kügel H, Leach MO, Metzler D, Mlynárik V, Moser E, Newbold MC, Payne GS, Ring P, Roberts JN, Rowland IJ, Thiel T, Tkác I, Topp S, Wittsack HJ, Podo F et al. (
1998
) Absolute metabolite quantification by in vivo NMR spectroscopy: II. A multicentre trial of protocols for in vivo localised proton studies of human brain.
Magn Reson Imaging
 
16
:
1093
–1106.
Klose U (
1990
) In vivo proton spectroscopy in the presence of eddy currents.
Magn Reson Med
 
14
:
26
–30.
Kreis R (
1997
) Quantitative localized 1H MR spectroscopy for clinical use.
Prog Nucl Magn Reson Spectrosc
 
31
:
155
–195.
Kreis R, Ernst T, Ross BD (
1993
) Absolute quantitation of water and metabolites in the human brain. II. Metabolite concentrations.
J Magn Reson B
 
102
:
9
–19.
Leary S, Brex P, MacManus D, Parker G, Barker G, Miller D, Thompson A (
2000
) A 1H magnetic resonance spectroscopy study of aging in parietal white matter: implications for trials in multiple sclerosis.
Magn Reson Imaging
 
18
:
455
–459.
Lim K, Zipursky R, Watts M, Pfefferbaum A (
1992
) Decreased gray matter in normal aging: an in vivo magnetic resonance study.
J Gerontol
 
47
:
26
–30.
Loessner A, Alavi A, Lewandrowski KU, Mozley D, Souder E, Gur RE (
1995
) Regional cerebral function determined by FDG-PET in healthy volunteers — normal patterns and changes with age.
J Nucl Med
 
36
:
1141
–1149.
MacDonald HL, Bell BA, Smith MA, Kean DM, Tocher JL, Douglas RHB, Miller JD, Best JJK (
1986
) Correlation of human NMR T1 values measured in vivo and brain water content.
Br J Radiol
 
59
:
355
–357.
Miller BL (
1991
) A review of chemical issues in 1H NMR spectroscopy: N-acetyl-l-aspartate, creatine and choline.
NMR Biomed
 
4
:
47
–52.
Moore G, Koch S, Chen G, Santalucia J Jr, Manji H (1999) Myo-inositol is not exclusively glial in human CNS. Proceedings of the 7th Annual Meeting of ISMRM, Vol 1, Philadelphia, USA, p. 33.
Murphy D, DeCarli C, McIntosh A, Daly E, Mentis M, Pietrini P, Szczepanik J, Schapiro M, Grady C, Horwitz B, Rapoport S (
1996
) Sex differences in human brain morphometry and metabolism: an in vivo quantitative magnetic resonance imaging and positron emission tomography study of the effect of aging.
Arch Gen Psychiatry
 
53
:
585
–594.
Noworolski SM, Nelson SJ, Henry RG, Day MR, Wald LL, Star-Lack J, Vigneron DB (
1999
) High spatial resolution 1H-MRSI and segmented MRI of cortical gray matter and subcortical white matter in three regions of the human brain.
Magn Reson Med
 
41
:
21
–29.
Paczynski RP, Lin WL, Venkatesan R, He Y, Hsu C, Haacke EM (
1997
) Accurate quantitation of regional brain water content in vivo.
Stroke
 
28
:
112
.
Pakkenberg B, Gundersen HJG (
1997
) Neocortical neuron number in humans: effect of sex and age.
J Comp Neurol
 
384
:
312
–320.
Pfefferbaum A, Adalsteinsson E, Spielman D, Sullivan EV, Lim KO (
1999
) In vivo brain concentration of N-acetyl compounds, creatine, and choline in Alzheimer's disease.
Arch Gen Psychiatry
 
56
:
185
–192.
Pfefferbaum A, Adalsteinsson E, Spielman S, Sullivan EV, Lim KO (
1999
) In vivo spectroscopic quantification of the N-acetyl moiety, creatine, and choline from large volumes of brain gray and white matter: effects of normal aging.
Magn Reson Med
 
41
:
276
–284.
Pijnappel WWF, van den Boogaart A, de Beer R, van Ormondt D (
1992
) SVD-based quantification of magnetic resonance signals.
J Magn Reson
 
97
:
122
–134.
Raz N (1996) Neuroanatomy of the aging brain observed in vivo: a review of structural MRI findings. In: Neuroimaging II (ED Bigler, ed), pp 153–182. New York: Plenum.
Raz N, Gunning F, Head D, Dupuis J, McQuain J, Briggs S, Loken W, Thornton A, Acker J (
1997
) Selective aging of the human cerebral cortex observed in vivo: differential vulnerability of the prefrontal cortex.
Cereb Cortex
 
7
:
268
–282.
Ross BD (
1991
) Biochemical considerations in 1H spectroscopy. Glutamate and glutamine; myo-inositol and related metabolites.
NMR Biomed
 
4
:
59
–63.
Saunders D, Howe F, Van den Boogaart A, Griffiths J, Brown M (
1999
) Aging of the adult human brain: in vivo quantitation of metabolite content with proton magnetic resonance spectroscopy.
J Magn Reson Imaging
 
9
:
711
–716.
Simmons ML, Frondoza CG, Coyle JT (
1991
) Immunocytochemical localization of N-acetyl-aspartate with monoclonal antibodies.
Neuroscience
 
45
:
37
–45.
Soher BJ, van Zijl PC, Duyn JH, Barker PB (
1996
) Quantitative proton MR spectroscopic imaging of the human brain.
Magn Reson Med
 
35
:
356
–363.
Stryer L (1995) Metabolism: basic concepts and design. In: Biochemistry, Chapter 17, pp. 443–448. New York: Freeman.
Terry RD, DeTeresa R, Hansen LA (
1987
) Neocortical cell counts in normal human adult aging.
Ann Neurol
 
21
:
530
–539.
Urenjak J, Williams SR, Gadian DG, Noble M (
1993
) Proton nuclear magnetic resonance spectroscopy unambiguously identifies different neural cell types.
J Neurosci
 
13
:
981
–989.
Van der Veen JWC, de Beer R, Luyten PR, van Ormondt D (
1988
) Accurate quantification of in vivo 31P NMR signals using the variable projection method and prior knowledge.
Magn Reson Med
 
6
:
92
–98.
Videen JS, Blaettler D, Wong EC, Buxton RB (1995) Quantifying brain volume in a spectroscopic voxel: fast spin echo can duplicate STEAM CSF determination. Proceeding of the  3rd Annual Meeting of the Society for  Magnetic Resonance  Medicine, Vol 3, Nice, France, p. 1965.
Wiggins RC, Gorman A, Rolsten C, Samorajski T, Ballinger WE, Freund G (
1988
) Effects of aging and alcohol on the biochemical composition of histologically normal human brain.
Metab Brain Dis
 
3
:
67
–80.