Metabolic changes in the normal ageing brain: Consistent findings from short and long echo time proton spectroscopy

https://doi.org/10.1016/j.ejrad.2007.08.038Get rights and content

Abstract

Objectives

Sixty three healthy subjects were measured to assess dependence of brain metabolites on age using short- and long echo time spectroscopy in different brain regions.

Material and methods

Younger and elderly humans were measured with long echo time (TE = 135 ms) 3D-MR-spectroscopic imaging (MRSI) (10 subjects) and with ultra-short echo (TE = 11 ms) time 2D-MRSI (7 subjects). In addition, results from single voxel 1H-spectroscopy (TE = 20 ms) of two cohorts of 46 healthy subjects were retrospectively correlated with age.

Results

3D-MR SI revealed reduced NAA/Cr in the older group in the frontal lobe (−22%; p < 0.01), parietal lobe (−28%; p < 0.01) and semiovale (−9%; p < 0.01) compared to the younger group. Cho/Cr was elevated in the semiovale (+35%; p < 0.01) and in the n. lentiformis (+42%; p < 0.01) in the older group. NAA/Cho was reduced in all regions measured, except the thalamus, in the older group compared to the younger group (from −21 to −49%; p < 0.01). 2D-MRSI revealed decreased total NAA (−3.1% per decade; p < 0.01) and NAA/Cr (−3.8% per decade; p < 0.01), increased total Cho (+3.6% per decade; p < 0.01) and Cho/Cr (+4.6% per decade; p < 0.01) and increased total myo-Inositol (mI, +4.7% per decade; p < 0.01) and mI/Cr (+5.4% per decade; p < 0.01) and decreased NAA/Cho (−8% per decade; p < 0.01) in semiovale WM. Results from single voxel spectroscopy revealed a significantly negative correlation of NAA/Cho in frontal (−13% per decade; p < 0.01) and in temporal lobe (−7.4% per decade; p < 0.01) as well as increased total Cr (10% per decade; p < 0.01) in frontal lobe. Other results from single voxel measurements were not significant, but trends were comparable to that from multivoxel spectroscopy.

Conclusion

Age-related changes measured with long echo time and short echo time 1H-MRS were comparable and cannot, therefore, be caused by different T2 relaxation times in young and old subjects, as suggested previously.

Introduction

1H magnetic resonance spectroscopy (MRS) offers the possibility to visualise and quantify different metabolites in the human brain, in vivo and non-invasively.

Potential biochemical changes in the normal ageing brain are an issue of growing interest, increasingly focussing on specific issues such as memory protection in older subjects, and the onset of neurologic or psychiatric disorders [1], [2]. N-acetyl-aspartate (NAA) and choline (Cho) may be of particular interest in the ageing brain. NAA was identified as an amino acid located in neuronal cell bodies, dendrites, and axons [3]. Cho has been regarded as a marker of cellular density [4]. Thus, age-related changes of these metabolites may not only be connected to different neurological diseases, but also to the “normal” development of the (ageing) human brain [5].

A number of spectroscopic methods have been used to allow evaluation of potential correlations between metabolic changes and age [6], [7], [8], [9], [10], [11]. The results were correlated with absolute or relative concentrations of NAA, Cr and Cho. However, the results of different authors or studies have demonstrated a large variability of findings, in part possibly due to the application of different data acquisition and processing methods and due to diverse inclusion criteria for subjects.

If single voxel 1H-MRS (SVS) is employed, one or two relatively large volumes (8–27 cm3) are usually measured in one session. Protocols with long echo times (TE  135 ms) only allow the major metabolites, i.e., NAA, Cho and Cr, to be measured in the healthy brain. Methods using short echo times (TE  20 ms), enable to detect spectral lines of larger number of metabolites (e.g. myo-Inositol [mI], glutamine or glutamate) and are less dependent on potential changes of metabolic ratios due to age-dependent T2 relaxation times, which may influence metabolic ratios in long echo time spectroscopy.

By using multivoxel (2D or 3D) spectroscopic imaging, a larger part or region of the brain can be sampled with better time efficiency (i.e. better signal-to-noise ratio [SNR] per unit time) than with single voxel spectroscopy. Most multivoxel studies have been carried out with spatial resolutions of approximately 1–8 cm3. Commonly, echo times in the range of 135–144 ms have been used, although short echo time spectroscopic imaging with STEAM volume selection (TE = 11 ms) and PRESS (point resolved spectroscopy, TE  30 ms) selection is available [12], [13].

It has recently been shown that 3D-MRSI in the human brain can be performed with nominal voxel sizes <0.5 cm3 using a standard head coil at 3 Tesla [14]. Higher spatial resolution 1H-MRSI is advantageous for selective volumetric averaging of the corresponding spectroscopic data, allowing improved separation between anatomical structures.

In this study, high spatial resolution (a nominal voxel size of 0.33 cm3), long echo time (TE = 135 ms) 3D-1H-MRSI was applied on five young and five older subjects. A newly developed 2D-1H-MRSI with ultra-short echo time (TE = 11 ms) was performed on seven healthy subjects and results were correlated with age. In addition, absolute metabolic values of single voxel short echo time 1H-MRS data (STEAM, TE/TR = 20/6000 or 20/2500 ms, voxel size 8 cm3) of two cohorts of healthy subjects were retrospectively correlated with age.

Section snippets

Human subjects

In total, 63 healthy subjects (39 female, 24 male, age range 18–65 years) were studied with different MRS protocols optimised for 3 Tesla. A demographic overview of all subjects included in this study is given in Table 1. All subjects had no psychiatric or neurologic disease diagnosed at the time of measurement and were free of any medication. MRIs of all subjects showed no pathological findings. Written informed consent, according to the guidelines and approval of the local review board (EK

Experimental

Volume selection for MRS or MRSI was performed based on high resolution, multi-slice turbo spin echo (rapid acquisition relaxation enhanced, RARE) images in the axial orientation with (TE/TR = 80/3180 ms) or on modified driven equilibrium Fourier transform (MDEFT) images.

3D-spectroscopic imaging (3D-MRSI)

The CSI matrices were zero filled to the size of 32 × 32, which provided 1280 voxels within the excited region for each subject, thus giving a nominal spatial resolution of 0.19 cm3. The data were Fourier transformed in the spectroscopic and the two spatial dimensions, Hadamard-transformed in the third spatial dimension and corrected for frequency and zero order phases shifts. The peak areas of NAA, Cr and Cho were calculated using the parametric spectral modelling and least-squares optimisation

3D-spectroscopic imaging

Our robust measurement protocol with PRESS preselection and long TE (135 ms) combined with a modified water suppression technique [14] resulted in high quality data for all measurements despite small voxel sizes. See Fig. 2 for typical examples of extracted spectra.

Discussion

Absolute and/or relative quantification of brain metabolites as a function of age was already addressed in several studies at 1.5T.

Brooks et al. [7] found significantly decreased absolute concentration of NAA (−12% between the third and seventh decade) and no significant changes in Cr and Cho in the interhemispheric tissue of the medial frontal lobe (consisting of GM, cerebrospinal fluid [CSF] and WM) in 50 male subjects (10 subjects per decade between 20 and 70 years), partially in agreement

References (27)

  • E. Angelie et al.

    Regional differences and metabolic changes in normal aging of the human brain: proton MR spectroscopic imaging study

    Ajnr: Am J Neuroradiol

    (2001)
  • P.E. Sijens et al.

    Human brain chemical shift imaging at age 60 to 90: analysis of the causes of the observed sex differences in brain metabolites

    Investigative Radiol

    (2001)
  • D.E. Saunders et al.

    Aging of the adult human brain: in vivo quantitation of metabolite content with proton magnetic resonance spectroscopy

    Stroke

    (1999)
  • Cited by (0)

    View full text