Brain

Brain Advance Access originally published online on October 17, 2005
Brain 2005 128(11):2483-2506; doi:10.1093/brain/awh640

This Article Abstract FREE Full Text (PDF) Supplementary Data All Versions of this Article: 128/11/2483    most recent awh640v1 Alert me when this article is cited Alert me if a correction is posted Services Email this article to a friend Similar articles in this journal Similar articles in ISI Web of Science Similar articles in PubMed Alert me to new issues of the journal Add to My Personal Archive Download to citation manager Search for citing articles in: ISI Web of Science (15) Request Permissions Disclaimer Google Scholar Articles by Caramanos, Z. Articles by Arnold, D. L. Search for Related Content PubMed PubMed Citation Articles by Caramanos, Z. Articles by Arnold, D. L. Social Bookmarking          What's this?

© The Author (2005). Published by Oxford University Press on behalf of the Guarantors of Brain. All rights reserved. For Permissions, please e-mail: journals.permissions@oxfordjournals.org

Review Article

¹H-MRS quantification of tNA and tCr in patients with multiple sclerosis: a meta-analytic review

Zografos Caramanos, Sridar Narayanan and Douglas L. Arnold

Magnetic Resonance Spectroscopy Unit, Montreal Neurological Institute, McGill University, Montreal, Canada

Correspondence to: Dr Douglas L. Arnold, MD, Magnetic Resonance Spectroscopy Unit, Montreal Neurological Institute, McGill University, 3801 University Street, Montreal, Quebec, Canada H3A 2B4 E-mail: doug{at}mrs.mni.mcgill.ca

	Summary

Top Summary Introduction The present study Materials and methods Results Discussion Supplementary material References

 
Meta-analysis was performed on the results of 75 comparisons from the 30 peer-reviewed publications that used proton magnetic resonance spectroscopy (¹H-MRS) or spectroscopic imaging to (i) quantify the mean concentrations of total creatine (tCr, found in neurons, astrocytes and oligodendrocytes), and/or total N-acetyl groups (tNA, found only in neurons), in the lesional and/or non-lesional white matter (WM) and/or the grey matter (GM) of patients with multiple sclerosis (MS) and (ii) compare these values with those in the homologous tissues of normal controls (NC). For mean [tNA] values, there was (i) a large-effect-sized overall decrease in patients' lesional WM relative to NC WM (25 comparisons), (ii) a medium-effect-sized overall decrease in patients' non-lesional WM relative to NC WM (36 comparisons) and (iii) a medium-effect-sized overall decrease in patients' GM relative to NC GM (14 comparisons). Patients' mean [tNA] values were sometimes statistically normal but were never statistically increased. For mean [tCr] values, there was (i) no statistically significant overall change in the patients' lesional WM relative to NC WM (24 comparisons), although statistically significant increases and decreases were sometimes found, (ii) a medium-effect-sized overall increase in patients' non-lesional WM relative to NC WM (33 comparisons) and (iii) no statistically significant overall change in patients' GM relative to NC GM (12 comparisons), although a significant decrease was found in one comparison. Of 41 comparisons with statistically significant changes, 38 combined in a way that would probably result in decreased mean [tNA]/[tCr] ratios such that (i) 66% had statistically decreased mean [tNA] and statistically unchanged mean [tCr] values, (ii) 13% had statistically decreased mean [tNA] and statistically increased mean [tCr] values and (iii) 21% had statistically unchanged mean [tNA] values and statistically increased mean [tCr] values. Of the 25 comparisons that came from studies that also analysed [tNA]/[tCr] ratios, the direction of change in mean [tNA] values and mean [tNA]/[tCr] ratios was concordant in 84%. In comparisons that quantified both [tNA] and [tCr], there was a similar amount of variability in both measures in each of the different tissue types studied, both in patients and NCs. Together, these results suggest that within-voxel tNA/tCr ratios can be interpreted as valid and accurate surrogate measures of ‘cerebral tissue integrity’—with decreased tNA/tCr ratios indicating some combination of neuroaxonal disturbance, oligodendroglial disturbance, and astrocytic proliferation. These results also suggest that, although within-voxel tNA/tCr ratios are not perfect indicators of [tNA] content, they do represent a practical compromise to acquiring surrogate measures of within-voxel neuroaxonal integrity.

Key Words: multiple sclerosis; magnetic resonance spectroscopy; N-acetylaspartate; creatine; meta-analysis

Abbreviations: COV = coefficient of variation; Cr = creatine; CSF = cerebrospinal fluid; CTI = cerebral tissue integrity; GM = grey matter; MRI = magnetic resonance imaging; MS = multiple sclerosis; NAA = N-acetylaspartate; NAAG = N-acetylaspartylglutamate; NAGM = normal appearing grey matter; NAWM = normal appearing white matter; NC = normal control; PCr = Phosphocreatine; ¹H-MRS = proton magnetic resonance spectroscopy; ¹H-MRSI = proton magnetic resonance spectroscopic imaging; ¹H-MRS(I) = ¹H-MRS and ¹H-MRSI; TEs = echo times; TRs = repetition times; tCr = total methyl resonance of creatine and phosphocreatine; tNa = total methyl resonance of N-Acetyl-containing compounds

Received March 30, 2005. Revised July 12, 2005. Accepted August 17, 2005.

	Introduction

Top Summary Introduction The present study Materials and methods Results Discussion Supplementary material References

 
Multiple sclerosis (MS), one of the most common causes of neurological disability in young adults, is an inflammatory, demyelinating and degenerative autoimmune disorder of the central nervous system (CNS) (Compston and Coles, 2002). Its pathological hallmark is the presence of multi-focal, demyelinating, white matter (WM) lesions that are disseminated in space and time (McDonald et al., 2001). Each of these WM lesions are focal regions of inflammation, demyelination, oligodendroglial loss, reactive gliosis (i.e. astrocytic proliferation) and axonal degeneration (Peterson and Trapp, 2005). Importantly, these lesions are visible not only on gross pathological inspection, but also on conventional magnetic resonance imaging (MRI) techniques that allow us to acquire, with high spatial-resolution, information regarding the density of protons associated with free water. For example, (i) WM lesions are almost always visible as hyperintensities on T₂-weighted MRI (Li et al., 2000); (ii) WM lesions that are associated with greater amounts of tissue destruction are also visible as chronic hypointensities on T₁-weighted MRI (Barkhof et al., 2000); and (iii) WM lesions that are ‘active’ (i.e. undergoing acute inflammation) are also ‘enhancing’ (i.e. visible as hyperintensities) on gadolinium-enhanced T₁-weighted MRI (Rovaris and Filippi, 2000).

It is now accepted that these patients' non-lesional, ‘normal-appearing white matter’ (NAWM)—which appears normal both macroscopically and on conventional MRI—is actually not normal. Indeed, NAWM has been shown to demonstrate histopathological abnormalities such as axonal loss (Ganter et al., 1999; Evangelou et al., 2000, 2001; Lovas et al., 2000; Bjartmar et al., 2001) and astrocytic proliferation (Allen and McKeown, 1979; Allen et al., 1981). Furthermore, even though they are not typically visible on conventional MRI (Kidd et al., 1999; Geurts et al., 2005a, b; Peterson and Trapp, 2005), cortical lesions are actually quite common when looked for histopathologically, and substantial demyelination and neuroaxonal loss has been shown to occur in the lesional grey matter (GM) of patients with MS (Peterson et al., 2001; Bo et al., 2003; Geurts et al., 2005a; Peterson and Trapp, 2005).

Findings from a variety of non-conventional MRI techniques have supported the notion that the NAWM and normal-appearing GM (NAGM) of patients with MS is, in fact, abnormal (Filippi et al., 1999; Wolinsky and Narayana, 2002; Caramanos et al., 2003). For example, studies using proton magnetic resonance spectroscopy (¹H-MRS) and proton magnetic resonance spectroscopic imaging (¹H-MRSI) have provided evidence of disease-related changes in these patients' lesional and normal-appearing WM and NAGM as compared to the WM and GM of normal control (NC) subjects—changes that have typically been interpreted as reflecting a neuroaxonal disturbance in these tissues (Matthews et al., 1998; Arnold, 1999; Wolinsky and Narayana, 2002).

¹H-MRS(I)
¹H-MRS and ¹H-MRSI [¹H-MRS(I)] allow us to acquire information about protons from molecules other than water (Ross and Bluml, 2001). Two important metabolite peaks that are routinely quantified on the ¹H-MR spectrum of tissue from the human CNS and the ones that we will focus on in this paper are (i) the total methyl resonance of N-acetyl-containing compounds (tNA), often abbreviated as NA or NAA, that resonate at 2.01 p.p.m. relative to the standard tetramethylsilane and (ii) the total methyl resonance of creatine and phosphocreatine (tCr), often abbreviated as Cr, that resonates at 3.03 p.p.m.

The tNA peak
In the adult mammalian CNS, the tNA peak primarily reflects the presence of N-acetylaspartate (NAA) and, to a lesser extent, N-acetylaspartylglutamate (NAAG). NAA is synthesized by neuronal mitochondria and is found at very high concentrations in the mammalian brain, second only to glutamate in terms of free amino-acid concentrations (Birken and Oldendorf, 1989). NAA has been posited to have multiple roles, serving as (i) a source of acetyl groups necessary for the synthesis of myelin lipids (Chakraborty et al., 2001); (ii) a molecular water pump in myelinated neurons (Baslow, 2002); and (iii) a neuronal precursor for NAAG, which is the most prevalent and widely distributed neuropeptide in the mammalian nervous system (Neale et al., 2000). NAAG is also thought to play multiple roles, serving as (i) a neurotransmitter (Blakely and Coyle, 1988); (ii) a modulator of the effects of other neurotransmitters (Neale et al., 2000); (iii) a source of extracellular glutamate (Neale et al., 2000); and—in conjunction with NAA—a role in cell-specific signalling between neurons, astrocytes and oligodendrocytes (Baslow, 2000). Importantly, these two molecules seem to both be localized exclusively within neurons and neuronal processes (Moffett et al., 1991; Simmons et al., 1991). Because of the presence of its constituent-metabolites in neurons and its prominence in the ¹H-MR spectrum, decreases in the tNA peak have widely been used as an indicator of brain pathology and disease progression in a variety of CNS diseases, including MS (Arnold et al., 1998; Matthews et al., 1998; De Stefano et al., 2005).

It should be noted that, whereas NAA has also been found in cell cultures of oligodendroglial cell lineage (Urenjak et al., 1992, 1993; Bhakoo and Pearce, 2000), this seems to be a phenomenon that is largely limited to in vitro cell cultures; indeed, NAA is not present in significant concentrations in astrocytes or mature oligodendrocytes that are harvested in vivo (Urenjak et al., 1992, 1993). Evidence for the specificity of NAA as an axon-specific marker of mature WM in vivo, even in the presence of injury and a significant density of oligodendroglial cell precursors, has recently been provided in a biochemical and immunohistochemical study of rat optic nerve transection (Bjartmar et al., 2002). Furthermore, evidence for the validity of NAA as a surrogate measure of axonal density in patients with MS has recently been provided in a pair of studies that found strong correlations between (i) findings from in vivo ¹H-MRS and histopathological analysis of cerebral biopsy specimens (Bitsch et al., 1999) and (ii) findings from high-performance liquid chromatography and histopathological analysis of spinal-cord biopsy specimens (Bjartmar et al., 2000).

The tCr peak
In the adult mammalian CNS, the tCr peak reflects the presence of creatine (Cr) and phosphocreatine (PCr)—two molecules that are known to play an important role in energy metabolism, with PCr representing reserves of high-energy phosphates that provide for homeostasis and energy needs (Miller, 1991; Wyss and Kaddurah-Daouk, 2000; Ross and Bluml, 2001). Interestingly, while the tCr peak is present in both neurons and glial cells, its concentration has been shown to be the highest in astrocytes and oligodendrocytes [at least when expressed in terms of (nM/mg protein)] (Urenjak et al., 1993).

Quantification of ¹H-MRS(I) metabolite concentrations
Although the non-invasively acquired biochemical information that can be obtained using ¹H-MRS(I) has shown tremendous utility in furthering our understanding of both the normal and the pathological states of the brain (Ross and Bluml, 2001), in vivo quantification of metabolite concentrations is still quite challenging.

Within-voxel normalization of ¹H-MRS(I) metabolites
Because tCr concentration is both relatively constant throughout the brain and relatively resistant to change, one common approach to quantifying ¹H-MRS(I) data is to use the within-voxel resonance intensity of tCr as an internal standard for that of other metabolites (De Stefano et al., 2000). For example, the use of within-voxel tNA/tCr ratios as a surrogate measure of neuroaxonal integrity is a common practice in the study of patients with MS (Matthews et al., 1991; Arnold et al., 1998, 2000) and seems to have a high degree of convergent validity—relating strongly both to measures of such patients' clinical disability (De Stefano et al., 2001; Mainero et al., 2001; Tartaglia et al., 2004) as well as to their changes on other surrogate measures of cerebral integrity [e.g. measures of cerebral atrophy (De Stefano et al., 2002) and cerebral cortico-functional reorganization (Reddy et al., 2000)].

Some of the strengths that are associated in general with the use of such a within-voxel normalization are that, intrinsically, it partially corrects for some of the issues associated with (i) the tissue composition of the voxel [e.g. partial-volume effects due to within-voxel cerebrospinal fluid (CSF)] and (ii) the dependency of the voxel's position on coil-related electromagnetic properties [e.g. (a) changes in signal intensity related to the position of the voxel in the inhomogeneous radio-frequency (RF) field of the coil, and (b) coupling of the sample voxel to the coil]. Furthermore, some of the strengths that are associated specifically with the use of within-voxel tNA/tCr normalization are that it generally compensates for (i) pathology-related changes in T₁ and T₂ relaxation times (because the constituents of the tNA and the tCr peaks have similar relaxation mechanisms that would be expected to be affected similarly by pathology, although some differential change in relaxation time is possible), and (ii) dilution and partial-volume effects associated with within-voxel oedematous water or CSF (because neither of these contain any of the constituents of the tNA or tCr peaks).

There is, however, an important caveat regarding the use of within-voxel tNA/tCr ratios as a measure of neuroaxonal integrity in the CNS of patients with MS: namely, using tCr as an internal standard is valid only in situations where the tCr signal is relatively insensitive to the within-voxel pathology (Filippi et al., 2005).

Absolute quantification of ¹H-MRS(I) metabolites
The aforementioned requirement of the normalization-to-tCr approach (i.e. that within-voxel tCr values must be unaffected by the disease) is avoided by performing some kind of ‘absolute’ (or ‘semi-absolute’) quantification of ¹H-MRS(I) metabolites: that is, quantifying the metabolites of interest within the ¹H-MRS(I) spectra with reference to either (i) a known internal standard (e.g. CSF or tissue water), or (ii) a known external standard that is scanned at the same time (e.g. a phantom that contains a known concentration of a molecule, or molecules, of interest). A metabolite concentration can then be expressed in some form of molar units in an absolute quantification study (or in some form of arbitrary units in a semi-absolute quantification study).

Unfortunately, with regards to the use of an internal standard, the tissue water signal is not really a stable reference—as evidenced by the exquisite ability of MRI to detect pathology—and it may vary between individuals and in pathology; moreover, whereas the CSF water concentration is more stable, measurement of its signal intensity is error-prone (e.g. because of the pulsatile nature of CSF flow). Furthermore, with regards to the absolute quantification of ¹H-MRS(I) metabolites in general, it is necessary to account for many factors that are dealt with intrinsically by normalizing to within-voxel tCr. As a result, the associated acquisition protocols are more complex and time-consuming to carry out. For example, in order to obtain accurate absolute quantifications of ¹H-MRS metabolites, one must account for the varying effects of (i) partial-volume contamination of a voxel with CSF or oedematous water, (ii) the voxel's location and its relationship to the electromagnetic properties of the coil [e.g. (a) its location in the coil's inhomogeneous RF field map, and (b) the coupling of the sample to the coil] and (iii) the T₁ and T₂ relaxation times of the spectral metabolites and the reference standard. Furthermore, errors in quantification can creep in due to (i) the instability of the scanner across the course of examinations; (ii) the accuracy of the estimate of the actual (as opposed to nominal) voxel size, which may affect some methods of absolute quantification more than they would a within-voxel ratio approach; and (iii) the accuracy of the calibration to the external reference (Sarchielli et al., 2002). All of these factors must either be (i) measured or accounted for, which is generally not feasible due to the associated scanning-time increases [e.g. the effects of T₁ and T₂ relaxation times may be minimized by adapting the pulse sequence to obtain spectra at long repetition times (TRs) and short echo times (TEs), but this approach compromises flexibility and efficiency in the acquisition], or (ii) assumed, which is most often the case. Importantly, such assumptions are associated with errors that are then propagated throughout the quantification process (Bottomley, 1992), and the increase in variance due to such errors can easily overwhelm the ability of ¹H-MRS to detect significant differences in studies that typically include only small numbers of patients.

Potential problems associated with the within-voxel normalization-to-tCr approach
Whereas the strengths associated with the use of within-voxel tCr values as an internal standard for tNA values make it very attractive for studies in MS, the validity of this approach has recently been brought into question (Pan et al., 2002; Schubert et al., 2002; Adalsteinsson et al., 2003; Casanova et al., 2003; Fernando et al., 2004; He et al., 2005; Srinivasan et al., 2005; Vrenken et al., 2005). Indeed, as we will now review, a number of early reports found statistically significant, but often-times conflicting, tCr changes in the lesional and normal-appearing WM of patients with MS relative to the WM of NCs. Furthermore, as we will review below, it has also recently been suggested that mean [tCr] values are associated with too much variance relative to mean [tNA] values for within-voxel-tCr to be used as an appropriate internal-normalization factor for tNA (Li et al., 2003).

Conflicting reports of tCr changes in the brains of patients with MS
The first evidence for an abnormality in the ¹H-MRS(I) tCr signal came from Bruhn et al. (1992) who presented some qualitative in vivo ¹H-MRS evidence of decreased tCr in the chronic T₂-weighted lesions of children with MS (n = 8). However, tCr was not affected in either these children's acute lesions or NAWM.

Two years later, an ¹H-MRSI study by Husted et al. (1994) compared NAWM and T₂-weighted lesions in the centrum semiovale of 13 patients with MS to homologous WM in six NCs. Mean [tCr] values were significantly increased by 14% in both NAWM and lesions (both P < 0.05). Whereas the mean [tNA] value decreased by 23% in lesions (P < 0.05), it did not change significantly in NAWM (which was associated with only a slight reduction of 2%). Importantly, however, the mean [tNA]/[tCr] ratio was reduced in both the lesions (by 31%, P < 0.05) and the NAWM (by 13%, P < 0.05). Thus, this latter finding questioned the use of tNA/tCr ratios as a measure of tNA-related neuroaxonal integrity because, despite their significantly reduced mean tNA/tCr ratio, there was virtually no measurable decrease in the mean tNA value of these patients' NAWM.

In a more direct examination, Davies et al. (1995) performed a high-resolution, absolute-quantification, ex vivo ¹H-MRS study of tissue from the unfixed, post-mortem brains of eight patients with MS and six NCs with no known neurological disease. [Such in vitro measurements avoid the aforementioned technical difficulties associated with in vivo ¹H-MRS(I) and are, therefore, much more reliable. Nevertheless, certain post-mortem changes occur that affect the concentration of tNA (but not that of tCr) and must be accounted for.] In the patients, they examined small blocks of brain tissue that contained either (i) macroscopically visible WM lesions (n = 9), (ii) WM immediately adjacent to such a lesion (n = 7) or (iii) NAWM distal to any macroscopically visible WM lesions (n = 5). Mean [tNA] and [tCr] values from these tissue samples were compared to those from homologous samples in the NCs; in this study, however, [tNA] was calculated as the total of [NAA], [NAAG] and [acetate]—the latter being included in order to account for any post-mortem autolysis. The authors found that, when expressed as mM per kg of wet weight (i) both mean [tNA] and [tCr] values decreased significantly in the lesional tissue (by 34 and 29%, respectively, both P < 0.05) and (ii) both mean [tNA] and [tCr] values also decreased significantly in the WM adjacent to lesions (by 33% and 29% respectively, both P < 0.05). But, in NAWM that was distal to lesions, they found a decrease of only 6% in the mean [tNA] value—a change that did not reach statistical significance. Moreover, there was virtually no effect on the mean [tCr] value in this distal-to-lesion NAWM: a non-significant decrease of only 0.3% relative to NC WM. It should also be noted that Davies et al. (1995) found a very strong correlation between the [tNA] and [tCr] values in their tissue samples (r = 0.89, P < 0.01); and, as might be expected, when they expressed their [tNA] findings relative to those of within-sample [tCr], they did not find any significant change relative to NCs in the [tNA]/[tCr] ratios of the patients' lesional, lesion-adjacent or lesion-distal WM tissue. Thus, in this case, significant reductions in patients' lesional [tNA] values were masked by normalizing to their also-reduced lesional [tCr] values. This finding further called into question the use of tNA/tCr ratios as a measure of tNA-related neuroaxonal integrity.

Soon after, Pan et al. (1996) used a 4.1-T ¹H-MRSI approach to study eight patients with mild RR-MS. They measured tNA and tCr values in 1.1 cc voxels filled with either (i) cortical NAGM, (ii) WM lesions, (iii) NAWM within 1.5 cm of a WM lesion, or (iv) NAWM greater than 1.5 cm from a WM lesion. These values were compared to the WM and GM values from eight NCs. They found that tNA/tCr ratios were reduced in the patients' NAGM as well as in their lesional and normal-appearing WM. Furthermore, based on WM/GM ratios for tNA values and tCr values, they concluded that (i) NAWM tNA values (which start off similar to NC WM) decrease as a lesion is approached and (ii) tNA is lowest in lesions. On the other hand, they found that NAWM tCr values are increased relative to WM values in NCs—tCr being highest farther away from lesions and decreasing slightly (though still elevated) as WM lesions are approached and WM becomes lesional. The authors suggested that, in the brains of patients with MS (i) WM lesions display axonal dysfunction or loss (reflected in the decreased tNA values) and mild gliotic replacement (reflected in the increased tCr values), whereas (ii) NAWM is associated with less axonal disturbance (reflected in an attenuated decrease of tNA) in the face of astrocytic proliferation (reflected in the increased tCr values).

In contrast to the above findings of significant changes in lesional mean [tCr] values is a study by Bitsch et al. (1999) in which in vivo ¹H-MRSI-measured [tNA] and [tCr] values in the WM of 40 NCs were compared with those in the tissue of three patients with short-term MS that had undergone stereotactic biopsy of their WM lesions. In those ¹H-MRSI voxels that contained the regions of T₁-weighted hypointensity from which the lesion samples had previously been biopsied (i.e. between 6 and 21 weeks earlier), Bitsch et al. (1999) found significantly decreased values of [tNA] (i.e. >2 SD away from the mean value of the NCs) that were related to the amount of co-localized reduction in axonal density (as measured relative to the axonal density in peri-lesional WM). On the other hand, they found normal within-lesion-voxel [tCr] levels, which, like Pan et al. (1996), they suggested resulted from a counterbalancing effect of glial proliferation occurring simultaneous to axonal loss.

Normalizing to tCr may introduce more variability than it prevents
In addition to the inconsistencies regarding tCr concentrations in the lesional and normal-appearing WM of patients with MS, a recent study by Li et al. (2003) suggested another possible drawback associated with the normalization-to-tCr approach. These investigators measured [tCr] and [tNA] values in the brains of eight NCs and—as indicated by the associated coefficients of variation (COVs)—they found evidence for (i) increased intra-individual variability in WM-[tCr] values versus WM-[tNA] values (median COVs = 14.6 and 8.4%), and (ii) increased inter-individual variability in [tCr] values versus [tNA] values that were measured in a combination of WM and GM (respective median COVs = 23.3 and 15.6%). Based on these findings, they concluded that the greater variability of [tCr] values relative to those of [tNA] is another reason against the use of within-voxel tCr values to normalize tNA values.

	The present study

Top Summary Introduction The present study Materials and methods Results Discussion Supplementary material References

 
As we have seen, a number of reports have found evidence for statistically significant changes in the tCr concentrations of lesional and normal-appearing WM of patients with MS, thereby calling into question the use of the normalization-to-Cr approach in such patients. Importantly, however, these studies have been quite inconsistent in their findings. For example, (i) tCr values in the NAWM of patients with MS have been found to be decreased (Davies et al., 1995), increased (Husted et al., 1994; Pan et al., 1996; Bitsch et al., 1999) or unaffected (Bruhn et al., 1992; Davies et al., 1995); similarly, (ii) tCr values in the lesional WM of patients with MS have also been found to be decreased (Bruhn et al., 1992; Davies et al., 1995), increased (Husted et al., 1994; Pan et al., 1996) or unaffected (Bruhn et al., 1992; Bitsch et al., 1999). Furthermore, these studies have all been based on relatively small sample sizes and the finding of increased variability in [tCr] values versus [tNA] values has thus far been demonstrated in just one study of only eight NC subjects (Li et al., 2003).

As reviewed by Borenstein and Rothstein (1999), single studies often do not have the sample size required to precisely estimate the impact of a treatment, nor even to properly test the null hypothesis of no effect; meta-analysis of many such small studies allows, however, for the statistical synthesis of data in order to yield more-definitive results. The present study is a meta-analysis of all known peer-reviewed publications that measured and compared [tNA] and/or [tCr] values in the brains of NCs and in the brains of patients with MS. This is done in the hopes of gaining a better understanding of (i) the overall consistency and magnitude of these findings, and (ii) what they have to say about the use of tNA/tCr ratios in the study of patients with MS.

	Materials and methods

Top Summary Introduction The present study Materials and methods Results Discussion Supplementary material References

 
Studies included
To our knowledge, the present analysis includes the relevant results from all pertinent, peer-reviewed publications that were available as at the time of final analysis (i.e. February 18, 2005). Potential studies were first identified in PubMed using the following query: ‘[‘multiple sclerosis’ (MeSH Terms) or multiple sclerosis (Text Word) or ‘MS’ (Title) or ‘Mult Scler’ (Journal)]’. The titles and abstracts of the results of this search were then carefully examined in order to select those articles that would potentially meet our entry criteria. These articles were subsequently retrieved in order to be verified for acceptance and inclusion into the present analysis. The text and the reference sections of these articles were also carefully searched in order to capture any additional pertinent references.

Inclusion criteria
The studies included in the present analysis had to meet the following criteria: (i) include an absolute or semi-absolute in vivo quantification of ¹H-MRS(I)-measured [tNA] and/or [tCr] values from the lesional and/or non-lesional WM, and/or the NAGM of patients with MS; (ii) include similar data from the homologous tissues of NC subjects; and (iii) include the means and SDs of these values for both groups of subjects, which were required to generate measures of effect size and compare findings across studies. These in vivo findings were also compared to the ex vivo findings of Davies et al. (1995) that were described above.

Information extracted from the included studies
From each of the studies included in the present analysis, we recorded the following information when it was present.

Publication data. This included (i) the title of the article; (ii) the authors and their contact information; and (iii) the journal, and the volume, issue, page and PubMed identification numbers of the publication.

Clinical and demographic data. For the NC subjects, this included (i) the number studied and (ii) their age. For the MS patients, this included data regarding (i) the number studied, (ii) their age, (iii) their clinical status on Kurtzke's (1983) Expanded Disability Status Scale (EDSS) and (iv) the duration of their disease. For each of these measures (excluding sample size), an ‘average’ value was calculated as follows: (i) when only a mean value was presented, it was used as the average value; (ii) when only a median value was presented, it was used; (iii) when both a mean and a median value were presented, the mean of these two values was used. When possible, these measures were recorded separately for each of the clinical subgroups [e.g. patients with RR, secondary-progressive (SP), or primary-progressive (PP) MS] that were studied.

Technical information. This included (i) the scanner make, model, and field strength; and (ii) the type of ¹H-MRS(I) acquisition, voxel size and quantification used, as well as the TR, TE and mixing times.

Tissues sampled. This included information regarding the type and location of the tissue that was sampled in the patients with MS and in the NC subjects. It was also noted when lesional WM was considered ‘chronic’ or ‘enhancing’, and when NAGM was ‘cortical’ or ‘sub-cortical’.

Metabolite values. Means and SDs for [tNA] and [tCr] values were extracted from information given in the text, tables or figures of the original articles. Median, minimum and maximum values were also recorded when present. The specifics of the statistical analyses and the results of each comparison were also recorded.

Sample sizes. The sample size for each of the comparisons included in the present analysis was considered to be the number of subjects in each of the NC and MS subgroups studied (and not necessarily the number of lesional or normal-appearing voxels that were studied in each of these subgroups). This was chosen because of differences in the way that metabolite values were summarized in the different studies. For example, whereas some studies collected data from only one ¹H-MRS voxel within an individual, other studies collected data from multiple ¹H-MRS or ¹H-MRSI voxels within an individual; furthermore, whereas some studies that collected multi-voxel data from each subject presented their values on a voxel-by-voxel basis, others presented these values averaged across an individual's multiple voxels. Unless otherwise stated in the original publication, when there were fewer samples than subjects it was assumed that each sample came from a different subject.

Statistical analysis
Effect of MS on mean [tCr] and [tNA] values: measure of effect size
Our measure of effect size was Hedges' adjusted g (Hedges and Olkin, 1985), hereafter referred to as ‘g’. This is a commonly used estimator of effect size that is calculated based on the standardized mean difference between the two groups being compared. For each comparison, g was calculated as ‘[(the mean in patients with MS) – (the mean in NC subjects)]/(the pooled sample standard deviation)’ and then adjusted to remove associated small-sample bias (Hedges and Olkin, 1985; Borenstein and Rothstein, 1999). For the critical-alpha value of 0.05 that was considered to be significant in the present set of analyses, a significant mean difference between the two groups is suggested if g ± its 95% confidence interval (CI) does not include g = 0. Importantly, in addition to providing a measure of the ‘statistical significance’ of a given finding, the g value is also a measure of the ‘practical significance’ (Stevens, 1992) of this finding. By convention, an absolute g-value of 0.20 reflects a ‘small-sized’ effect; of 0.50, a ‘medium-sized’ effect; and of 0.80, a ‘large-sized’ effect (Cohen, 1992).

Effect of MS on mean [tCr] and [tNA] values: meta-analysis statistics
After effect sizes were computed for each comparison, a combined-effect size (i.e. the weighted mean of the effects in each of the comparisons being combined) was computed for each of six separate sets of meta-analyses that examined (i) mean [tNA] values in patients' lesional WM versus those in NC WM, (ii) mean [tCr] values in the same tissues, (iii) mean [tNA] values in patients' non-lesional WM versus those in NC WM, (iv) mean [tCr] values in the same tissues, (v) mean [tNA] values in the patients' NAGM versus those in NC GM, and (vi) mean [tCr] values in the same tissues.

Given the heterogeneous nature of (i) the techniques used, (ii) the clinical groups tested and (iii) the tissue types and locations that were studied in the various studies and comparisons that were included in the present set of analyses, combined-effect sizes were computed using random-effects models. This was considered appropriate because such models are based on the assumption that samples being analysed may be drawn from populations with different effect sizes and, thus, both within-study sampling error and between-study variation are included in the assessment of the uncertainty (i.e. the CIs) of the results of the analyses (Alderson et al., 2003).

In addition to calculating the overall combined effects, the impacts of various moderator variables on these combined-effect sizes were also examined using a random-effects approach. These included (i) MS type (RR, SP or PP), (ii) MS severity [Mild (average EDSS <3.5), Moderate (average EDSS of 3.5–5) or Severe (average EDSS of 6)], (iii) Lesion type (chronic or enhancing), and (iv) NAGM type (cortical or sub-cortical).

Effect of MS on mean [tCr] and [tNA] values: graphical presentation of results
The results of these six sets of analyses are presented graphically in a series of Forest plots. At the top of these, each comparison is summarized in descending order of effect size. For each comparison, circles and extending lines represent g± its 95% CI—the circle being filled when the associated P-value is <0.05. To the right of each line is summarized (i) the comparison that is being described and an indication of the P-value associated with its effect-size (^*** indicates P < 0.001, ^** indicates P < 0.01 and ^* indicates P < 0.05); (ii) the number of NC subjects studied; (iii) the number and type of MS patients studied; (iv) the average MS-Severity of the patients studied; (v) the type of tissue studied in the MS patients; (vi) the location of the tissue studied in the MS patients; and (vii) any additional comments related to that comparison. Below these individual-comparison results are presented the results from the combined-effects analyses: both overall, as well as grouped by moderator level. These are presented similarly to the individual comparisons, except that to the right of each line is presented (i) the total number of NC subjects, (ii) the total number of MS patients, (iii) the number of comparisons that were included in the calculation of that combined effect and (iv) any additional comments related to that comparison.

These Forest plots are presented as black and white images in the printed journal article, but can be seen in colour in the supplementary material available online. In the colour Forest plots, comparisons involving patients with RR-MS are always shown in red, SP-MS in blue and PP-MS in green. Data from any other subgroup or combination of subgroups is presented in black.

Effect of MS on mean [tCr] and [tNA] values: tabular presentation of results
Additional tables are available in the online Supplementary material that summarize each comparison included in the present analysis with regards to (i) the study it came from; (ii) what kind of patients were studied; (iii) the number of NC subjects studied, as well as the means, SDs and COVs of their [tCr] and/or [tNA] values; (iv) similar descriptive statistics for the patient groups; (v) the percent-difference between these two groups' mean [tCr] and/or [tNA] values; (vi) the resulting value of g for this comparison, along with its associated 95% CIs, the weight of its contribution to the overall combined-effect size and its P-value; (vii) the statistical results obtained in the present analysis and in the original study, and whether or not these agreed. [Please note that these may have disagreed due to differences in, for example: (i) power resulting from the use of a different sample size (i.e. the original study may been voxel-based as opposed to the present subject-based analysis), or (ii) the use of some kind of correction for multiple comparisons in the original study. Furthermore, not all of the comparisons that are included in the current analysis were tested in their original studies.] These tables also include the results of each of the combined-effects analyses that were examined.

Effect of MS on mean [tCr] and [tNA] values: correspondence of [tNA] values and [tNA]/[tCr] ratios
In addition to calculating effect-sizes, we also examined (i) the correspondence between any statistically significant changes in the patients' mean [tNA] and mean [tCr] values and (ii) how these changes might most likely affect the resulting [tNA]/[tCr] ratios. Moreover, for those studies that explicitly tested for significant changes in patients' mean [tNA]/[tCr] ratios, we also examined the correspondence between these findings and any changes in the patients' mean [tNA] values.

Variability in [tCr] and [tNA] values
Finally, we also examined the data from the studies included in the present analysis for potential differences in the COVs associated with the [tNA] and [tCr] values. This was done separately for mean COV values that were obtained in vivo in the cerebral WM and GM of the NCs, as well as for those that were obtained in vivo in the patients' lesional WM, cerebral NAWM and NAGM. Statistical significance in these comparisons was assessed using Bonferroni-corrected, paired-sample t-tests.

Software
Calculations of effect sizes for each of the comparisons and combined effects studied, as well as the CIs, P-values and weights associated with each of these under a random-effects model, were generated using Comprehensive Meta-Analysis, version 1.0.23 (Borenstein and Rothstein, 1999). All of the other statistical and graphical analyses included in the present study were performed using SYSTAT For Windows, version 10.2.

	Results

Top Summary Introduction The present study Materials and methods Results Discussion Supplementary material References

 
Studies included in the present analysis
A total of 30 studies met our criteria for inclusion in the present analysis (Davies et al., 1995; Rooney et al., 1997; Schiepers et al., 1997; Narayana et al., 1998; Sarchielli et al., 1998, 1999, 2002; Van Walderveen et al., 1999; Cucurella et al., 2000; Helms et al., 2000; Lee et al., 2000; Mader et al., 2000, 2001; Suhy et al., 2000; Helms, 2001; Kapeller et al., 2001, 2002; Sharma et al., 2001; Chard et al., 2002; Cifelli et al., 2002; Pan et al., 2002; Schubert et al., 2002; Adalsteinsson et al., 2003; Casanova et al., 2003; Inglese et al., 2003, 2004; Wylezinska et al., 2003; Fernando et al., 2004; He et al., 2005; Vrenken et al., 2005). Table 1 summarizes all of the abbreviations that are used in Table 2 and in all subsequent tables and figures. Table 2 summarizes information describing each of these studies.

View this table: [in this window] [in a new window]   Table 1 Abbreviations used in all other tables and figures

 

View this table: [in this window] [in a new window]   Table 2 Information describing the 30 studies included in the present analysis

 
Studies excluded from the present analysis
A recent study by Seeger et al. (2003) also met our inclusion criteria but its pertinent data had already been included in the Mader et al. (2001) study (U. Seeger, personal communication). Similarly, a study by Pendlebury et al. (2000) also met our inclusion criteria but its pertinent data was presented by Lee et al. (2000) in a format that was more appropriate for the present analysis. The results of three other pertinent studies were not included in the present analysis because they were presented only in the form of medians and ranges (Davie et al., 1997; Brex et al., 1999; Leary et al., 1999), and the results of another pertinent study were available only after the date of final analysis (Srinivasan et al., 2005). Nevertheless, as reviewed in the Appendix that is available as an online supplement to this article, the findings from these four studies agree almost entirely with those that were included in the present analysis.

Comparisons included in the present set of analyses
The 30 studies included in the present set of analyses contributed a total of 75 separate comparisons: (i) 15 studies contributed 25 comparisons between patients' lesional WM and NC WM, (ii) 22 studies contributed 36 comparisons between patients' non-lesional WM and NC WM and (iii) 9 studies contributed 14 comparisons between patients' NAGM and NC GM. Tables 3, 4 and 5, respectively, summarize each of these comparisons.

View this table: [in this window] [in a new window]   Table 3 The 25 comparisons of [tCr] and/or [tNA] values in the lesional WM of the patients with MS and in the WM of the NC subjects that were included in the present analysis

 

View this table: [in this window] [in a new window]   Table 4 The 36 comparisons of [tCr] and/or [tNA] values in the non-lesional WM of the patients with MS and in the WM of the NC subjects that were included in the present analysis

 

View this table: [in this window] [in a new window]   Table 5 The 14 comparisons of [tCr] and/or [tNA] values in the NAGM of the patients with MS in the GM of NC subjects and that were included in the present analysis

 
Individual comparisons are referred to by a combination of a ‘study code’ (i.e. the first three letters of the first author's surname and the last two digits of the year of publication) and a ‘comparison number.’ For example, ‘Dav-95_1’ in Table 3 refers to the first of three comparisons in the Davies et al. (1995) study.

It should be noted that four of the 75 comparisons included data that were partly or completely included and re-analysed in a somewhat different fashion in four subsequent comparisons (i.e. data from Roo-97_1 overlaps with Suh-00_3, Ing-03_1 with He-05_3, Kap-01_3 with Cha-02_1, and Kap-01_4 with Cha-02_2). In order to ensure that findings from the meta-analyses that contained all of the possible data sets did not differ from those that included only unique data, a series of additional ‘unique data’ meta-analyses were run that excluded data from the Roo-97_1, Ing-03_1, Kap-01_3, and Kap-01_4 comparisons. In each case, findings from these unique-data analyses were similar to those from the analyses that included data from all of the possible comparisons.

Findings in the lesional WM of patients with MS
Mean [tNA] values
Findings from the 25 comparisons of mean [tNA] values in the patients' lesional WM versus those in NC WM are presented online in Table 6 and summarized in Fig. 1A. A significant decrease was found in 19 of these comparisons (76.0%), and no significant difference was found in the other 6 comparisons (24.0%). There were no instances of significantly increased values. The combined effect across these 25 comparisons showed a large-effect-sized overall decrease in the mean lesional-WM-[tNA] values of patients with MS (g = –1.29, P < 0.0001), with a mean (SD) decrease of 18.8% (9.7).

View larger version (65K): [in this window] [in a new window]   Fig. 1 Forest plots summarizing (A) the results of comparisons between the mean [tNA] values in the cerebral WM of NC subjects versus those in the lesional cerebral WM of patients with MS, and (B) similar results for mean [tCr] values in these tissues. See Materials and methods for further information.

 
Large-effect-sized decreases in mean lesional-WM-[tNA] values were also found in the combined effects for each of the MS-Type subgroups [RR-MS (n = 10, g = –1.09, P < 0.0001); SP-MS (n = 1, g = –2.14, P = 0.0002); PP-MS (n = 2, g = –1.72, P = 0.0108)] and for each of the MS-Severity subgroups [Mild-MS (n = 10, g = –1.11, P < 0.0001); Mod-MS (n = 7, g = –1.56, P < 0.0001); Sev-MS (n = 1, g = –2.45, P = 0.0007)]. There was some evidence for increasingly negative values as the clinical severity of the patients being studied increased, but there was too much overlap to be definitive.

Large-effect-sized decreases were also found in the mean [tNA] values of the combined effects for both the chronic (n = 11, g = –1.35, P < 0.0001) and the enhancing (n = 4, g = –1.01, P = 0.0001) lesions that were studied, but there was not much difference between these two lesion types.

Mean [tCr] values
Findings from the 24 comparisons of mean [tCr] values in the patients' lesional WM versus NC WM are presented online in Table 7 and summarized in Fig. 1B. A significant decrease was found in 2 of these comparisons (8.3%), a significant increase was found in 5 of these comparisons (20.8%) and no significant difference was found in the other 17 comparisons (70.8%). Nevertheless, the combined effect across these 24 comparisons showed no significant overall change in the mean lesional-WM-[tCr] values of patients with MS (g = 0.17, P = 0.2912), with a mean (SD) increase of 1.0% (13.9).

Evidence for a significant increase in mean lesional-WM-[tCr] values was found only in the PP-MS combined effect (n = 2, g = 0.89, P = 0.0074). No other evidence for statistically significant change in mean lesional-WM-[tCr] values was found in the combined effects of the other MS-Type subgroups [RR-MS (n = 9, g = 0.33, P = 0.1625); SP-MS (n = 1, g = –0.05, P = 0.9011)] or any of the MS-Severity subgroups [Mild-MS (n = 9, g = 0.46, P = 0.0595); Mod-MS (n = 7, g = –0.12, P = 0.6707); Sev-MS (n = 1, g = 0.59, P = 0.2382)]. Furthermore, there was no evidence for a relationship between mean lesional-WM-[tCr] values and clinical severity in the patients studied.

No overall changes were found in the mean [tCr] values of the combined effects for either the chronic (n = 10, g = 0.33, P = 0.2738) or the enhancing (n = 4, g = 0.20, P = 0.3683) lesions that were studied. Interestingly, none of the comparisons involving enhancing WM lesions were significantly different from NC WM, but significant decreases were found in three of the comparisons involving chronic WM lesions and a significant increase was found in yet another (Fig. 1B). It should be noted, however, that this pattern was not maintained by the recent Srinivasan et al. (2005) study, which found an increase of 14% (P = 0.02) in the mean [tCr] value of gadolinium-enhancing T1-hypointense WM lesions, but only a non-significant mean increase of 3% (P = 0.56) in chronic T1-hypointense WM lesions.

Findings in the non-lesional WM of patients with MS
Mean [tNA] values
Findings from the 36 comparisons of mean [tNA] values in the patients' non-lesional WM versus those in NC WM are presented online in Table 8 and summarized in Fig. 2A. A significant decrease was found in 15 of these comparisons (41.7%), and no significant difference was found in the other 21 comparisons (58.3%). There were no instances of significantly increased values. The combined effect across these 36 comparisons showed a medium-effect-sized overall decrease in the mean WM-[tNA] values of patients with MS (g = –0.53, P < 0.0001), with a mean (SD) decrease of 6.6% (8.1).

View larger version (63K): [in this window] [in a new window]   Fig. 2 Forest plots summarizing (A) the results of comparisons between the mean [tNA] values in the cerebral WM of NC subjects versus those in the non-lesional WM of patients with MS, and (B) similar results for mean [tCr] values in these tissues. See Materials and methods for further information.

 
Of these 36 comparisons, patient tissue that was not strictly cerebral NAWM was examined in 4: (i) Dav-95_2 examined ex vivo samples of peri-lesional WM (i.e. WM that was immediately adjacent to lesions) whose mean [NAA], [NAAG] and [tCr] values were almost identical to those of their ex vivo samples of lesional WM and very different from those of their distal-to-lesion NAWM samples, (ii) Cas-03_1 and Cas-03_2 examined NAWM in the pons and in the cerebeller peduncles, and (iii) Ing-03_1 examined a large volume of non-segmented cerebral tissue that included GM and lesions. In order to avoid any possible confound associated with including these comparisons, a separate meta-analysis was run to determine the combined effect across only those 32 comparisons that examined cerebral NAWM. These showed a very similar medium-effect-sized decrease in the mean cerebral-NAWM-[tNA] values of patients with MS (g = –0.50, P < 0.0001), with a mean (SD) decrease of 5.3% (6.5).

Medium- to large-effect-sized decreases in mean cerebral-NAWM-[tNA] values were also found in the combined effects for each of the MS-Type subgroups [RR-MS (n = 13, g = –0.44, P = 0.0126); SP-MS (n = 7, g = –0.96, P = 0.039); PP-MS (n = 4, g = –0.47, P = 0.0969)] and for each of the MS-Severity subgroups [Mild-MS (n = 14, g = –0.43, P = 0.0026); Mod-MS (n = 11, g = –0.80, P = 0.0001); Sev-MS (n = 3, g = –0.71, P = 0.2046)], but these were not statistically significant in the PP-MS or the Sev-MS subgroups. There was some evidence for increasingly negative values as the clinical severity of the patients being studied increased, but, again, there was too much overlap to be definitive.

Interestingly, in the 31 comparisons of mean [tNA] values in the patients' non-lesional WM for which there was clear information about the study TE (Table 2), a statistically significant relationship was seen between TE length and a significant decrease relative to NC WM (Fisher's exact test, P = 0.003). Namely, a significantly decreased mean [tNA] value was seen in 10 of the 15 comparisons from ‘long’ TE studies (i.e. with a TE of 135 ms) but in only 2 of the 16 comparisons from ‘short’ TE studies (i.e. with a TE of 90 ms). No such relationship was found for any of the other combinations of tissue-type and metabolite that were studied in the present meta-analysis.

Mean [tCr] values
Findings from the 33 comparisons of mean [tCr] values in the patients' non-lesional WM versus those in NC WM are presented online in Table 9 and summarized in Fig. 2B. A significant decrease was found in 2 of these comparisons (6.1%), no significant difference was found in another 22 (66.7%) and a significant increase was found in the remaining 9 comparisons (27.3%). The combined effect across these 33 comparisons showed a medium-effect-sized overall increase in the mean NAWM-[tCr] values of patients (g = 0.44, P = 0.0001), with a mean (SD) increase of 4.9% (11.1).

As can be seen in Fig. 2, of the four comparisons that did not involve cerebral NAWM, two found the greatest mean [tCr] decreases (Dav-95_02 and Cas-03_2) and another found the greatest mean [tCr] increase (Ing-03_1). When a separate meta-analysis was run to determine the combined effect across only those 29 comparisons that examined mean cerebral-NAWM-[tCr] values, a similar medium-effect-sized increase was found (g = 0.50, P < 0.0001), with a mean (SD) increase of 6.9% (7.1).

Medium-effect-sized increases in mean cerebral-NAWM-[tCr] values were also found in the combined effects for each of the MS-Type subgroups [RR-MS (n = 12, g = 0.52, P = 0.0035); SP-MS (n = 6, g = 0.61, P = 0.059); PP-MS (n = 4, g = 0.68, P = 0.0191)] and for each of the MS-Severity subgroups [Mild-MS (n = 13, g = 0.43, P = 0.0026); Mod-MS (n = 10, g = 0.51, P = 0.0004); Sev-MS (n = 2, g = 0.56, P = 0.5063)], but these were not statistically significant in the SP-MS or the Sev-MS subgroups. There was no evidence for a simple relationship between mean cerebral-NAWM-[tCr] values and clinical severity in the patients studied.

Findings in the NAGM of patients with MS
Mean [tNA] values
Findings from the 14 comparisons of mean [tNA] values in the patients' NAGM versus those in NC GM are presented online in Table 10 and summarized in Fig. 3A. A significant decrease was found in six of these comparisons (42.9%), and no significant difference was found in the other eight comparisons (57.1%). There were no instances of significantly increased values. The combined effect across these 14 comparisons showed a medium-effect-sized overall decrease in the mean NAGM-[tNA] values of patients with MS (g = –0.56, P = 0.0026), with a mean (SD) decrease of 7.1% (9.2).

View larger version (25K): [in this window] [in a new window]   Fig. 3 Forest plots summarizing (A) the results of comparisons between the mean [tNA] values in the cerebral GM of NC subjects versus those in the cerebral NAGM of patients with MS, and (B) similar results for mean [tCr] values in these tissues. See Materials and methods for further information.

 
Medium- to large-effect-sized decreases in mean NAGM-[tNA] values were also found in the combined effects of the RR- and SP-MS subgroups [RR-MS (n = 9, g = –0.39, P = 0.0079); SP-MS (n = 4, g = –1.29, P = 0.0522); PP-MS (n = 1, g = –0.08, P = 0.882)] and for each of the MS-Severity subgroups [Mild-MS (n = 9, g = –0.39, P = 0.0079); Mod-MS (n = 3, g = –0.92, P = 0.3461); Sev-MS (n = 2, g = –1.24, P = 0.0012)], but this was not statistically significant in the SP-MS, the PP-MS or the Mod-MS subgroups. There was again some evidence for increasingly negative values as the clinical severity of the MS patients being studied increased, but, again, there was too much overlap to be definitive.

A medium-effect-sized decrease was found in the mean [tNA] values of the cortical NAGM (n = 11, g = –0.50, P = 0.0242) that was studied, and a large-effect-sized decrease was found in the sub-cortical NAGM (n = 3, g = –0.79, P = 0.0031) that was studied, but there was too much overlap to be definitive about this difference.

Mean [tCr] values
Findings from the 12 comparisons of mean [tCr] values in the patients' NAGM versus those in NC GM are presented online in Table 11 and summarized in Fig. 3B. A significant decrease was found in 1 of these comparisons (8.3%), and no significant difference was found in the other 11 comparisons (91.7%). There were no instances of significantly increased values. The combined effect across these 12 comparisons showed no significant overall change in the mean NAGM-[tCr] values of patients with MS (g = –0.09, P = 0.6069), with a mean (SD) decrease of 0.2% (7.9).

No evidence for a significant change in mean NAGM-[tCr] values was found in the combined effects of any of the MS-Type subgroups [RR-MS (n = 8, g = –0.05, P = 0.7172); SP-MS (n = 3, g = –0.49, P = 0.5012); PP-MS (n = 1, g = 0.60, P = 0.2664)] or any of the MS-Severity subgroups [Mild-MS (n = 8, g = –0.05, P = 0.7172); Mod-MS (n = 3, g = –0.18, P = 0.8374); Sev-MS (n = 1, g = –0.35, P = 0.3871)]. There was no evidence for any relationship between mean NAGM-[tCr] values and clinical severity in the patients studied.

No evidence for a significant change in mean NAGM-[tCr] values was found in the combined effects of either NAGM-Type subgroup [Cortical (n = 9, g = –0.10, P = 0.6776); Sub-Cortical (n = 3, g = –0.06, P = 0.7838)].

Correspondence between significant changes in mean [tNA] and [tCr] values
The correspondence between changes in mean [tNA] and mean [tCr] values in patients with MS (as indicated by the statistically significant g-values described above) is shown online in Table 12 and summarized below.

Correspondence in lesional-WM comparisons
In the 24 lesional-WM comparisons, the majority (66.7%) had results that would probably lead to decreased mean [tNA]/[tCr] values in concordance with decreased mean [tNA] values: these include (i) 12 comparisons (50.0%) with statistically decreased mean [tNA] values but no statistical changes in their mean [tCr] values and (ii) 4 comparisons (16.7%) with statistically decreased mean [tNA] values and statistically increased mean [tCr] values. On the other hand, one comparison (4.2%) had a statistically increased mean [tCr] value despite a statistically normal mean [tNA], which could also lead to a decreased [tNA]/[tCr] ratio. There were also (i) two comparisons (8.3%) with statistically decreased mean [tNA] and mean [tCr] values, which could lead to statistically normal [tNA]/[tCr] ratios; and (ii) five comparisons (20.8%) with statistically normal mean [tNA] and mean [tCr] values.

There were eight comparisons from six studies that tested explicitly for changes in lesional-WM [tNA]/[tCr] ratios. These findings concurred with the [tNA] findings in six of these comparisons (75%): all of which had statistically decreased mean [tNA]/[tCr] ratios and statistically decreased mean [tNA] values (i.e. Cuc-00_1; Cuc-00_2; Sar-98_1; Suh-00_1; Suh-00_2; and the combined MS groups from van-99_1, 2, and 3 comparisons). On the other hand, the [tNA]/[tCr] findings did not concur with the [tNA] findings in two of these comparisons (25%): (i) one had a statistically decreased mean [tNA]/[tCr] ratio despite a statistically normal mean [tNA] value (Mad-00_1), and (ii) one had a statistically normal mean [tNA]/[tCr] ratio despite a statistically decreased mean [tNA] value (Dav-95_1). Nevertheless, when only the direction of numerical change in lesional-WM [tNA] values and [tNA]/[tCr] ratios was considered (i.e. not requiring a statistically significant change, which may not have been found due to the limited power associated with the relatively small sample sizes in each of these comparisons), decreases in both were found in all eight comparisons (100%).

Correspondence in cerebral-NAWM comparisons
In the 29 cerebral-NAWM comparisons, twelve comparisons (41.4%) had statistically normal mean [tNA] and mean [tCr] values, and there were no comparisons with statistically significant decreases in both mean [tNA] and [tCr] values. Importantly, ten comparisons (34.5%) had results that would probably lead to decreased [tNA]/[tCr] values in concordance with decreased mean [tNA] values: these include (i) nine comparisons (31.0%) with statistically decreased mean [tNA] values but statistically normal mean [tCr] values, and (ii) one comparison (3.4%) with a statistically decreased mean [tNA] value and a statistically increased mean [tCr] value. On the other hand, seven comparisons (24.1%) had statistically normal mean [tNA] values and statistically increased mean [tCr] values, which could lead to significantly decreased [tNA]/[tCr] ratios despite statistically normal [tNA] values.

There were 14 comparisons from nine studies that explicitly tested for changes in cerebral-NAWM [tNA]/[tCr] ratios. These findings concurred with the [tNA] findings in six of these comparisons (43%): (i) five had statistically decreased mean [tNA]/[tCr] ratios when the mean [tNA] values were also statistically decreased (i.e. Cuc-00_4, Pan-02_2, Suh-00_3, Suh-00_4, van-99_4), and (ii) one had a statistically normal mean [tNA]/[tCr] ratio when the mean [tNA] value was also statistically normal (Dav-95_3). On the other hand, the [tNA]/[tCr] findings did not concur with the [tNA] findings in eight of these comparisons (57%): (i) seven (50%) had statistically decreased mean [tNA]/[tCr] ratios despite statistically normal mean [tNA] values (Fer-04_1, Pan-02_1, the combined MS groups from the Roo-97_1 and 2 comparisons, Sar-98_2, Vre-05_1, Vre-05_2, and Vre-05_3), and (ii) one (7%) had a statistically normal mean [tNA]/[tCr] ratio despite a statistically decreased mean [tNA] value (Cuc-00_3). Nevertheless, when only the direction of numerical change in cerebral-NAWM [tNA] values and [tNA]/[tCr] ratios was considered, decreases in both were found in 10 (71%) of the 14 comparisons (Cuc-00_3, Cuc-00_4, Dav-95_3, Fer-04_1, Pan-02_2, Sar-98_2, Suh-00_3, Suh-00_4, van-99_4, Vre-05_3).

Correspondence in NAGM comparisons
In the 12 NAGM comparisons, 7 (58.3%) had statistically normal mean [tNA] and mean [tCr] values. On the other hand, four comparisons (33.3%) had statistically decreased mean [tNA] values and statistically normal mean [tCr] values, a combination that would probably lead to decreased [tNA]/[tCr] values. There was also one comparison (8.3%) with statistically decreased mean [tNA] and mean [tCr] values, which could lead to a statistically normal [tNA]/[tCr] ratio despite a significantly decreased mean [tNA] value.

In the three comparisons from the studies that explicitly tested for changes in NAGM [tNA]/[tCr] ratios, these findings concurred with the [tNA] findings in all three (100%): (i) two with statistically decreased mean [tNA]/[tCr] ratios when the mean [tNA] values were also statistically decreased (i.e. Sar-02_2 and Wyl-03_1), and (ii) one with a statistically normal mean [tNA]/[tCr] ratio when the mean [tNA] value was also statistically normal (Sar-02_1).

Variability in [tNA] versus [tCr]
Figure 4 shows the COVs associated with the paired cerebral-[tNA] and -[tCr] values that were measured in vivo in the studies that were included in the present analysis: (i) 22 pairs from comparisons of WM in NCs, (ii) 8 from GM in NCs, (iii) 22 from lesional WM in patients, (iv) 28 from NAWM in patients, and (v) 11 from NAGM in patients. Please note that this analysis was possible only in those studies that measured both [tNA] and [tCr] values in the same individuals—thereby excluding the findings from the Lee-00 and the Ada-03 studies, which only measured [tNA].

View larger version (33K): [in this window] [in a new window]   Fig. 4 Box-and-whisker plots showing the values, medians and quartiles of the COV in cerebral-[tNA] and cerebral-[tCr] values obtained in vivo from (A) the WM and (B) the GM of NC subjects, as well as from (C) the WM lesions, (D) the NAWM, and (E) the NAGM of patients with MS. The lines connect the COV values from the same comparison. The filled dots represent outliers. On top of each plot are the Bonferroni-corrected results of a paired t-test comparing the mean [tNA] and [tCr] COVs in each comparison. Below each plot are summary statistics describing each set of COVs.

 
As can be seen in Fig. 4, the [tNA] and [tCr] COVs were very similar to one another for all tissues in both the NC and the MS groups. There was, however, a statistically significant paired-difference between the mean [tCr]-COV and the mean [tNA]-COV in the WM of NCs (12.6 versus 9.7%, P = 0.010); nevertheless, this discrepancy was much less than that which was found by Li et al. (2003), and none of the other [tCr]-[tNA] COV-pairs differed significantly.

	Discussion

Top Summary Introduction The present study Materials and methods Results Discussion Supplementary material References

 
[tNA] findings
In the present study, we found (i) a large-effect-sized overall decrease of the mean [tNA] values in the lesional WM of patients with MS; (ii) a medium-effect-sized overall decrease in their cerebral NAWM; and (iii) a medium-effect-sized overall decrease in their NAGM. Importantly, whereas patients' mean [tNA] values were statistically normal in some comparisons, they were never statistically increased. Furthermore, the decreases in mean [tNA] values in each of these tissue types seemed to be related to the average clinical severity in the patients that were studied.

As reviewed in the Introduction, the [tNA] signal has been shown to arise primarily from NAA, a molecule that is known to be neuroaxonal marker in the mature CNS (Moffett et al., 1991; Simmons et al., 1991; Bjartmar et al., 2002) and that has been shown to be a good surrogate measure of axonal density in patients with MS (Bitsch et al., 1999; Bjartmar et al., 2000). Thus, the present pattern of decreased ¹H-MRS(I)-obtained [tNA] values is in accordance with the histopathological findings that there is axonal transection and loss in the NAWM and the lesional GM of patients with MS (Ganter et al., 1999; Evangelou et al., 2000a, b, 2001; Lovas et al., 2000; Bjartmar et al., 2001; Peterson and Trapp, 2005), and seemingly even more so in their lesional WM (Trapp et al., 1998). Indeed, the present findings reinforce the idea that ¹H-MRS(I)-measured [tNA] values are a valid surrogate measure of within-voxel neuroaxonal integrity in the brains of patients with MS.

In the present study, we also found preliminary evidence for a relationship between the length of the acquisition TE and the likelihood of finding a relative decrease in the mean [tNA] value of the patients' non-lesional WM. Namely, a significantly decreased mean [tNA] value was seen in 66.7% of the comparisons with a ‘long’ TE acquisition (i.e. 135 ms) but in only 12.5% of the comparisons with a ‘short’ TE acquisition (i.e. 90 ms). (No such relationship was found for any of the other combinations of tissue-type and metabolite that were studied.) Because long-TE acquisitions are potentially more sensitive to metabolite T₂ changes, such changes within the non-lesional WM may partly explain the difference seen between the short- and the long-TE studies. Importantly, however, subtle non-lesional pathology in the WM would be expected to increase metabolite T₂ values, which would tend to counteract decreases in apparent [tNA]. Thus, in order to fully explain the relationship that was seen between acquisition TE and decreased [tNA] in non-lesional WM, a more-detailed analysis is needed of the effects of the different methods used to acquire and to quantify the ¹H-MRS(I) data included in the present meta-analysis. Although this would surely be of interest, such an analysis was beyond the scope of the present study (which was aimed at understanding (i) the consistency and magnitude of any tissue-specific changes in the [tNA] and [tCr] values of patients with MS and (ii) the implications of these findings for the use of tNA/tCr ratios in the study of such patients).

[tCr] findings
In the present study, we found (i) no statistically significant overall change of the mean [tCr] values in the lesional WM of patients with MS, although statistically significant increases and decreases were sometimes found in individual comparisons; (ii) a medium-effect-sized overall increase in their cerebral NAWM; and (iii) no statistically significant overall change in their NAGM, although a significant decrease was found in one comparison. The mean [tCr] values in each of the tissue types tested did not seem to be related to the average clinical severity in the patients that were studied.

As reviewed in the Introduction, the [tCr] signal is present in neurons, astrocytes and oligodendrocytes, and there is some ex vivo evidence that the concentration of tCr is higher in glia than neurons (Urenjak et al., 1993). As a result, the ¹H-MRS(I)-obtained [tCr] findings seem to be consistent with what is found histopathologically in patients with MS; namely, that there are (i) varying degrees of axonal transection and loss in patients' NAWM and lesional GM, that are even greater in their lesional WM (Trapp et al., 1998; Ganter et al., 1999; Evangelou et al., 2000a, b; 2001; Bjartmar et al., 2001; Peterson and Trapp, 2005); (ii) varying degrees of reactive gliosis and astrocytic proliferation that are found in their lesional and normal-appearing WM (Allen and McKeown, 1979; Allen et al., 1981, 2001; Ludwin, 2000); and (iii) varying degrees of demyelination and oligodendroglial loss that are found within their lesional WM (Ludwin, 2000), lesional GM (Geurts et al., 2005a; Peterson and Trapp, 2005) and NAWM (Allen et al., 1981, 2001). Based on these known histopathological findings, it is to be expected that within-voxel [tCr] values might somehow be affected in the brains of patients with MS as a result of varying amounts of neuroaxonal loss, oligodendroglial loss and astrocytic proliferation.

Correspondence of mean [tNA] values and mean [tCr] values
In the present study, we found that when mean [tNA] and [tCr] values are statistically changed in patients with MS, the change—or lack thereof—in these values will most often combine in such a way as to probably result in a decreased mean [tNA]/[tCr] ratio. Indeed, out of the 41 comparisons with some sort of statistical changes in the patients' [tNA] and/or [tCr] values, 38 (93%) had some such combination. Most of these (79%) were associated with a statistically decreased mean [tNA] value, either in conjunction with (i) a statistically unchanged mean [tCr] value (66%) or (ii) a statistically increased mean [tCr] value (13%); far fewer (21%) were associated with a statistically unchanged mean [tNA] value and a statistically increased mean [tCr] value. Furthermore, the direction of change in mean [tNA] values and mean [tNA]/[tCr] ratios was concordant in most of the 25 comparisons (84%) that came from those studies that presented findings regarding their subjects' [tNA]/[tCr] ratios. Together, these results suggest that even though within-voxel tNA/tCr ratios might not be perfect indicators of the [tNA] content of any particular ¹H-MRS(I) voxel, they do represent a practical compromise to the acquisition of a surrogate measure of within-voxel neuroaxonal integrity.

tNA/tCr ratios as a surrogate measure of cerebral tissue integrity
Although the above results suggest that decreases in within-voxel tNA/tCr ratios may not be perfect surrogate measures of disturbed neuroaxonal integrity, they can be interpreted as valid and accurate surrogate measures of an overall disturbance in the ‘cerebral tissue integrity’ (CTI) of that voxel. That is, a decreased tNA/tCr ratio is indicative of some combination of (i) neuroaxonal disturbance (which is likely be associated with decreases in both tNA and tCr concentrations), (ii) oligodendroglial disturbance (which is likely be associated with decreases in tCr concentrations) and (iii) astrocytic proliferation (which is likely be associated with increases in tCr concentration)—each of which can co-occur to varying degrees within a voxel but are always indicative of some sort of disturbance in the CTI of that voxel.

Variability in [tNA] versus [tCr] values
Finally, the present findings suggest that, overall, there is a relatively equal amount of variability in the [tNA] and [tCr] values of the tissues examined in the present study. (It should be noted that the highest [tCr] COV in any of the samples of WM from NCs came from the He et al. (2005) study, which obtained a [tCr] COV of 30.7% versus a [tNA] COV of 17.3%; interestingly, this is a study that came from the same laboratory as the Li et al. (2003) study and they both used the same scanner and the same acquisition and quantification techniques.)

These findings also suggest that, in most cases, the differences in the variability of [tCr] and [tNA] values that are measured simultaneously within the same voxels do not seem to be so large as to warrant the conclusion made by Li et al. (2003) that ‘basing metabolite quantification on ratios and assuming stable [tCr] introduces more variability into ¹H-MRS than it prevents.’ Nevertheless, the warnings raised by these authors with regards to the use of within-voxel tCr-normalization still apply in those specific instances when tCr values are, in fact, much more variable than the metabolites that are to be normalized to them. Thus, it might be wise to quantify the within-voxel variability of tNA and tCr and, if there is a large discrepancy between the individual variability of these two metabolites, be cautious of the validity of the ratios obtained. If this is not the case (i.e. if there is relatively similar variability in the obtained tNA and tCr values), then there seems to be no reason to believe that the benefits associated with the within-voxel-normalization-to-Cr approach (reviewed in the Introduction) are exceeded by any variability-induced costs.

	Supplementary material

Top Summary Introduction The present study Materials and methods Results Discussion Supplementary material References

 
The Supplementary material cited in this article is available at Brain online.

	Acknowledgements

 
This study was supported by grants from the Canadian Institutes of Health Research and the Multiple Sclerosis Society of Canada. The authors gratefully acknowledge Drs Bonaventura Casanova, Declan Chard, Gunther Helms, Belinda Li, David Miller, Ponnada Narayana, Uwe Seeger and Marianne van Walderveen for their kind help in answering our questions regarding their original articles in a quick and clear manner, as well as Mr Stanley Hum for his help in verifying some of the data that were included in the present analysis. We also thank the two referees who reviewed our original submission for their insightful and constructive criticisms.

	References

Top Summary Introduction The present study Materials and methods Results Discussion Supplementary material References

 
Adalsteinsson E, Langer-Gould A, Homer RJ, Rao A, Sullivan EV, Lima CA, et al. Gray matter N-acetyl aspartate deficits in secondary progressive but not relapsing-remitting multiple sclerosis. AJNR Am J Neuroradiol 2003; 24: 1941–5.[Abstract/Free Full Text]

Alderson P, Green S, Higgins J, editors. The Cochrane Reviewers' Handbook Glossary, Version 4.1.5 [updated December 2003]. http://www.cochrane.org/resources/handbook/glossary.pdf (accessed March 6, 2005).

Allen IV, McKeown SR. A histological, histochemical and biochemical study of the macroscopically normal white matter in multiple sclerosis. J Neurol Sci 1979; 41: 81–91.[CrossRef][Web of Science][Medline]

Allen IV, Glover G, Anderson R. Abnormalities in the macroscopically normal white matter in cases of mild or spinal multiple sclerosis (MS). Acta Neuropathol Suppl (Berl) 1981; 7: 176–8.

Allen IV, McQuaid S, Mirakhur M, Nevin G. Pathological abnormalities in the normal-appearing white matter in multiple sclerosis. Neurol Sci 2001; 22: 141–4.[CrossRef][Web of Science][Medline]

Arnold DL. Magnetic resonance spectroscopy: imaging axonal damage in MS. J Neuroimmunol 1999; 98: 2–6.[CrossRef][Web of Science][Medline]

Arnold DL, Wolinsky JS, Matthews PM, Falini A. The use of magnetic resonance spectroscopy in the evaluation of the natural history of multiple sclerosis. J Neurol Neurosurg Psychiatry 1998; 64 Suppl 1: S94–101.

Arnold DL, De Stefano N, Narayanan S, Matthews PM. Proton MR spectroscopy in multiple sclerosis. Neuroimaging Clin N Am 2000; 10: 789–98, ix–x.

Barkhof F, Karas GB, van Walderveen MA. T1 hypointensities and axonal loss. Neuroimaging Clin N Am 2000; 10: 739–752.[Web of Science][Medline]

Baslow MH. Functions of N-acetyl-L-aspartate and N-acetyl-L-aspartylglutamate in the vertebrate brain: role in glial cell-specific signaling. J Neurochem 2000; 75: 453–9.[CrossRef][Web of Science][Medline]

Baslow MH. Evidence supporting a role for N-acetyl-L-aspartate as a molecular water pump in myelinated neurons in the central nervous system. An analytical review. Neurochem Int 2002; 40: 295–300.[CrossRef][Web of Science][Medline]

Bhakoo KK, Pearce D. In vitro expression of N-acetyl aspartate by oligodendrocytes: implications for proton magnetic resonance spectroscopy signal in vivo. J Neurochem 2000; 74: 254–62.[CrossRef][Web of Science][Medline]

Birken DL, Oldendorf WH. N-acetyl-L-aspartic acid: a literature review of a compound prominent in 1H-NMR spectroscopic studies of brain. Neurosci Biobehav Rev 1989; 13: 23–31.[CrossRef][Web of Science][Medline]

Bitsch A, Bruhn H, Vougioukas V, Stringaris A, Lassmann H, Frahm J, et al. Inflammatory CNS demyelination: histopathologic correlation with in vivo quantitative proton MR spectroscopy. AJNR Am J Neuroradiol 1999; 20: 1619–27.[Abstract/Free Full Text]

Bjartmar C, Kidd G, Mork S, Rudick R, Trapp BD. Neurological disability correlates with spinal cord axonal loss and reduced N-acetyl aspartate in chronic multiple sclerosis patients. Ann Neurol 2000; 48: 893–901.[CrossRef][Web of Science][Medline]

Bjartmar C, Kinkel RP, Kidd G, Rudick RA, Trapp BD. Axonal loss in normal-appearing white matter in a patient with acute MS. Neurology 2001; 57: 1248–52.[Abstract/Free Full Text]

Bjartmar C, Battistuta J, Terada N, Dupree E, Trapp BD. N-acetylaspartate is an axon-specific marker of mature white matter in vivo: a biochemical and immunohistochemical study on the rat optic nerve. Ann Neurol 2002; 51: 51–8.[CrossRef][Web of Science][Medline]

Blakely RD, Coyle JT. The neurobiology of N-acetylaspartylglutamate. Int Rev Neurobiol 1988; 30: 39–100.[Web of Science][Medline]

Bo L, Vedeler CA, Nyland HI, Trapp BD, Mork SJ. Subpial demyelination in the cerebral cortex of multiple sclerosis patients. J Neuropathol Exp Neurol 2003; 62: 723–32.[Web of Science][Medline]

Borenstein M, Rothstein H. Comprehensive meta-analysis. Englewood, New Jersey: Biostat; 1999.

Bottomley PA. The trouble with spectroscopy papers. J Magn Reson Imaging 1992; 2: 1–8.[Web of Science][Medline]

Brex PA, Gomez-Anson B, Parker GJ, Molyneux PD, Miszkiel KA, Barker GJ, et al. Proton MR spectroscopy in clinically isolated syndromes suggestive of multiple sclerosis. J Neurol Sci 1999; 166: 16–22.[CrossRef][Web of Science][Medline]

Bruhn H, Frahm J, Merboldt KD, Hanicke W, Hanefeld F, Christen HJ, et al. Multiple sclerosis in children: cerebral metabolic alterations monitored by localized proton magnetic resonance spectroscopy in vivo. Ann Neurol 1992; 32: 140–50.[CrossRef][Web of Science][Medline]

Caramanos Z, Santos AC, Arnold DL. Magnetic resonance imaging and spectroscopy: insights into the pathology and pathophysiology of multiple sclerosis. In: McDonald WI and Noseworthy JH, editors. Multiple sclerosis 2, Vol. 27, 2nd ed. Blue Books of Practical Neurology Series. USA: Butterworth-Heinemann; 2003. p. 139–67.

Casanova B, Martinez-Bisbal MC, Valero C, Celda B, Marti-Bonmati L, Pascual A, et al. Evidence of Wallerian degeneration in normal appearing white matter in the early stages of relapsing-remitting multiple sclerosis: a HMRS study. J Neurol 2003; 250: 22–8.[CrossRef][Web of Science][Medline]

Chakraborty G, Mekala P, Yahya D, Wu G, Ledeen RW. Intraneuronal N-acetylaspartate supplies acetyl groups for myelin lipid synthesis: evidence for myelin-associated aspartoacylase. J Neurochem 2001; 78: 736–45.[CrossRef][Web of Science][Medline]

Chard DT, Griffin CM, McLean MA, Kapeller P, Kapoor R, Thompson AJ, et al. Brain metabolite changes in cortical grey and normal-appearing white matter in clinically early relapsing-remitting multiple sclerosis. Brain 2002; 125: 2342–52.[Abstract/Free Full Text]

Cifelli A, Arridge M, Jezzard P, Esiri MM, Palace J, Matthews PM. Thalamic neurodegeneration in multiple sclerosis. Ann Neurol 2002; 52: 650–3.[CrossRef][Web of Science][Medline]

Cohen J. A power primer. Psychol Bull 1992; 112: 155–9.[CrossRef][Web of Science]

Compston A, Coles A. Multiple sclerosis. Lancet 2002; 359: 1221–31.[CrossRef][Web of Science][Medline]

Cucurella MG, Rovira A, Rio J, Pedraza S, Tintore MM, Montalban X, et al. Proton magnetic resonance spectroscopy in primary and secondary progressive multiple sclerosis. NMR Biomed 2000; 13: 57–63.[CrossRef][Web of Science][Medline]

Davie CA, Barker GJ, Thompson AJ, Tofts PS, McDonald WI, Miller DH. 1H magnetic resonance spectroscopy of chronic cerebral white matter lesions and normal appearing white matter in multiple sclerosis. J Neurol Neurosurg Psychiatry 1997; 63: 736–42.[Abstract/Free Full Text]

Davies SE, Newcombe J, Williams SR, McDonald WI, Clark JB. High resolution proton NMR spectroscopy of multiple sclerosis lesions. J Neurochem 1995; 64: 742–8.[Web of Science][Medline]

De Stefano N, Narayanan S, Mortilla M, Guidi L, Bartolozzi ML, Federico A, et al. Imaging axonal damage in multiple sclerosis by means of MR spectroscopy. Neurol Sci 2000; 21: S883–7.[CrossRef][Medline]

De Stefano N, Narayanan S, Francis GS, Arnaoutelis R, Tartaglia MC, Antel JP, et al. Evidence of axonal damage in the early stages of multiple sclerosis and its relevance to disability. Arch Neurol 2001; 58: 65–70.[Abstract/Free Full Text]

De Stefano N, Iannucci G, Sormani MP, Guidi L, Bartolozzi ML, Comi G, et al. MR correlates of cerebral atrophy in patients with multiple sclerosis. J Neurol 2002; 249: 1072–7.[CrossRef][Web of Science][Medline]

De Stefano N, Bartolozzi ML, Guidi L, Stromillo ML, Federico A. Magnetic resonance spectroscopy as a measure of brain damage in multiple sclerosis. J Neurol Sci 2005; 233: 203–8.[CrossRef][Web of Science][Medline]

Evangelou N, Esiri MM, Smith S, Palace J, Matthews PM. Quantitative pathological evidence for axonal loss in normal appearing white matter in multiple sclerosis. Ann Neurol 2000a; 47: 391–5.[CrossRef][Web of Science][Medline]

Evangelou N, Konz D, Esiri MM, Smith S, Palace J, Matthews PM. Regional axonal loss in the corpus callosum correlates with cerebral white matter lesion volume and distribution in multiple sclerosis. Brain 2000b; 123 (Pt 9): 1845-9.

Evangelou N, Konz D, Esiri MM, Smith S, Palace J, Matthews PM. Size-selective neuronal changes in the anterior optic pathways suggest a differential susceptibility to injury in multiple sclerosis. Brain 2001; 124: 1813–20.[Abstract/Free Full Text]

Fernando KT, McLean MA, Chard DT, MacManus DG, Dalton CM, Miszkiel KA, et al. Elevated white matter myo-inositol in clinically isolated syndromes suggestive of multiple sclerosis. Brain 2004; 127: 1361–9.[Abstract/Free Full Text]

Filippi M, Tortorella C, Bozzali M. Normal-appearing white matter changes in multiple sclerosis: the contribution of magnetic resonance techniques. Mult Scler 1999; 5: 273–82.[Abstract/Free Full Text]

Filippi M, Falini A, Arnold DL, Fazekas F, Gonen O, Simon JH, et al. Magnetic resonance techniques for the in vivo assessment of multiple sclerosis pathology: consensus report of the white matter study group. J Magn Reson Imaging 2005; 21: 669–75.[CrossRef][Web of Science][Medline]

Ganter P, Prince C, Esiri MM. Spinal cord axonal loss in multiple sclerosis: a post-mortem study. Neuropathol Appl Neurobiol 1999; 25: 459–67.[CrossRef][Web of Science][Medline]

Geurts JJ, Bo L, Pouwels PJ, Castelijns JA, Polman CH, Barkhof F. Cortical lesions in multiple sclerosis: combined postmortem MR imaging and histopathology. AJNR Am J Neuroradiol 2005a; 26: 572–7.[Abstract/Free Full Text]

Geurts JJ, Pouwels PJ, Uitdehaag BM, Polman CH, Barkhof F, Castelijns JA. Intracortical lesions in multiple sclerosis: improved detection with 3D double inversion-recovery MR imaging. Radiology 2005b; 236: 254–60.[Abstract/Free Full Text]

He J, Inglese M, Li BS, Babb JS, Grossman RI, Gonen O. Relapsing-remitting multiple sclerosis: metabolic abnormality in nonenhancing lesions and normal-appearing white matter at MR imaging: initial experience. Radiology 2005; 234: 211–7.[Abstract/Free Full Text]

Hedges LV, Olkin I. Statistical methods for meta-analysis. Orlando: Academic Press; 1985.

Helms G, Stawiarz L, Kivisakk P, Link H. Regression analysis of metabolite concentrations estimated from localized proton MR spectra of active and chronic multiple sclerosis lesions. Magn Reson Med 2000; 43: 102–10.[CrossRef][Web of Science][Medline]

Helms G. Volume correction for edema in single-volume proton MR spectroscopy of contrast-enhancing multiple sclerosis lesions. Magn Reson Med 2001; 46: 256–63.[CrossRef][Web of Science][Medline]

Husted CA, Goodin DS, Hugg JW, Maudsley AA, Tsuruda JS, de Bie SH, et al. Biochemical alterations in multiple sclerosis lesions and normal-appearing white matter detected by in vivo 31P and 1H spectroscopic imaging. Ann Neurol 1994; 36: 157–65.[CrossRef][Web of Science][Medline]

Inglese M, Li BS, Rusinek H, Babb JS, Grossman RI, Gonen O. Diffusely elevated cerebral choline and creatine in relapsing-remitting multiple sclerosis. Magn Reson Med 2003; 50: 190–5.[CrossRef][Web of Science][Medline]

Inglese M, Liu S, Babb JS, Mannon LJ, Grossman RI, Gonen O. Three-dimensional proton spectroscopy of deep gray matter nuclei in relapsing-remitting MS. Neurology 2004; 63: 170–2.[Abstract/Free Full Text]

Kapeller P, McLean MA, Griffin CM, Chard D, Parker GJ, Barker GJ, et al. Preliminary evidence for neuronal damage in cortical grey matter and normal appearing white matter in short duration relapsing-remitting multiple sclerosis: a quantitative MR spectroscopic imaging study. J Neurol 2001; 248: 131–8.[CrossRef][Web of Science][Medline]

Kapeller P, Brex PA, Chard D, Dalton C, Griffin CM, McLean MA, et al. Quantitative 1H MRS imaging 14 years after presenting with a clinically isolated syndrome suggestive of multiple sclerosis. Mult Scler 2002; 8: 207–10.[Abstract/Free Full Text]

Kidd D, Barkhof F, McConnell R, Algra PR, Allen IV, Revesz T. Cortical lesions in multiple sclerosis. Brain 1999; 122: 17–26.[Abstract/Free Full Text]

Kurtzke JF. Rating neurologic impairment in multiple sclerosis: an expanded disability status scale (EDSS). Neurology 1983; 33: 1444–52.[Abstract/Free Full Text]

Leary SM, Davie CA, Parker GJ, Stevenson VL, Wang L, Barker GJ, et al. 1H magnetic resonance spectroscopy of normal appearing white matter in primary progressive multiple sclerosis. J Neurol 1999; 246: 1023–6.[CrossRef][Web of Science][Medline]

Lee MA, Blamire AM, Pendlebury S, Ho KH, Mills KR, Styles P, et al. Axonal injury or loss in the internal capsule and motor impairment in multiple sclerosis. Arch Neurol 2000; 57: 65–70.[Abstract/Free Full Text]

Li DK, Zhao G, Paty DW. T2 hyperintensities: findings and significance. Neuroimaging Clin N Am 2000; 10: 717–38, ix.[Web of Science][Medline]

Li BS, Wang H, Gonen O. Metabolite ratios to assumed stable creatine level may confound the quantification of proton brain MR spectroscopy. Magn Reson Imaging 2003; 21: 923–8.[CrossRef][Web of Science][Medline]

Lovas G, Szilagyi N, Majtenyi K, Palkovits M, Komoly S. Axonal changes in chronic demyelinated cervical spinal cord plaques. Brain 2000; 123: 308–17.[Abstract/Free Full Text]

Ludwin SK. The neuropathology of multiple sclerosis. Neuroimaging Clin N Am 2000; 10: 625–48, vii.[Web of Science][Medline]

Mader I, Roser W, Kappos L, Hagberg G, Seelig J, Radue EW, et al. Serial proton MR spectroscopy of contrast-enhancing multiple sclerosis plaques: absolute metabolic values over 2 years during a clinical pharmacological study. AJNR Am J Neuroradiol 2000; 21: 1220–7.[Abstract/Free Full Text]

Mader I, Seeger U, Weissert R, Klose U, Naegele T, Melms A, et al. Proton MR spectroscopy with metabolite-nulling reveals elevated macromolecules in acute multiple sclerosis. Brain 2001; 124: 953–61.[Abstract/Free Full Text]

Mainero C, De Stefano N, Iannucci G, Sormani MP, Guidi L, Federico A, et al. Correlates of MS disability assessed in vivo using aggregates of MR quantities. Neurology 2001; 56: 1331–4.[Abstract/Free Full Text]

Matthews PM, Francis G, Antel J, Arnold DL. Proton magnetic resonance spectroscopy for metabolic characterization of plaques in multiple sclerosis. Neurology 1991; 41: 1251–6.[Abstract/Free Full Text]

Matthews PM, De Stefano N, Narayanan S, Francis GS, Wolinsky JS, Antel JP, et al. Putting magnetic resonance spectroscopy studies in context: axonal damage and disability in multiple sclerosis. Semin Neurol 1998; 18: 327–36.[Web of Science][Medline]

McDonald WI, Compston A, Edan G, Goodkin D, Hartung HP, Lublin FD, et al. Recommended diagnostic criteria for multiple sclerosis: guidelines from the International Panel on the diagnosis of multiple sclerosis. Ann Neurol 2001; 50: 121–7.[CrossRef][Web of Science][Medline]

Miller BL. A review of chemical issues in 1H NMR spectroscopy: N-acetyl-L-aspartate, creatine and choline. NMR Biomed 1991; 4: 47–52.[Web of Science][Medline]

Moffett JR, Namboodiri MA, Cangro CB, Neale JH. Immunohistochemical localization of N-acetylaspartate in rat brain. Neuroreport 1991; 2: 131–4.[Web of Science][Medline]

Narayana PA, Doyle TJ, Lai D, Wolinsky JS. Serial proton magnetic resonance spectroscopic imaging, contrast-enhanced magnetic resonance imaging, and quantitative lesion volumetry in multiple sclerosis. Ann Neurol 1998; 43: 56–71.[CrossRef][Web of Science][Medline]

Neale JH, Bzdega T, Wroblewska B. N-Acetylaspartylglutamate: the most abundant peptide neurotransmitter in the mammalian central nervous system. J Neurochem 2000; 75: 443–52.[CrossRef][Web of Science][Medline]

Pan JW, Hetherington HP, Vaughan JT, Mitchell G, Pohost GM, Whitaker JN. Evaluation of multiple sclerosis by 1H spectroscopic imaging at 4.1 T. Magn Reson Med 1996; 36: 72–7.[Web of Science][Medline]

Pan JW, Coyle PK, Bashir K, Whitaker JN, Krupp LB, Hetherington HP. Metabolic differences between multiple sclerosis subtypes measured by quantitative MR spectroscopy. Mult Scler 2002; 8: 200–6.[Abstract/Free Full Text]

Pendlebury ST, Lee MA, Blamire AM, Styles P, Matthews PM. Correlating magnetic resonance imaging markers of axonal injury and demyelination in motor impairment secondary to stroke and multiple sclerosis. Magn Reson Imaging 2000; 18: 369–78.[CrossRef][Web of Science][Medline]

Peterson JW, Trapp BD. Neuropathobiology of multiple sclerosis. Neurol Clin 2005; 23: 107–29.[CrossRef][Web of Science][Medline]

Peterson JW, Bo L, Mork S, Chang A, Trapp BD. Transected neurites, apoptotic neurons, and reduced inflammation in cortical multiple sclerosis lesions. Ann Neurol 2001; 50: 389–400.[CrossRef][Web of Science][Medline]

Reddy H, Narayanan S, Arnoutelis R, Jenkinson M, Antel J, Matthews PM, et al. Evidence for adaptive functional changes in the cerebral cortex with axonal injury from multiple sclerosis. Brain 2000; 123: 2314–20.[Abstract/Free Full Text]

Rooney WD, Goodkin DE, Schuff N, Meyerhoff DJ, Norman D, Weiner MW. 1H MRSI of normal appearing white matter in multiple sclerosis. Mult Scler 1997; 3: 231–7.[Web of Science][Medline]

Ross B, Bluml S. Magnetic resonance spectroscopy of the human brain. Anat Rec 2001; 265: 54–84.[CrossRef][Medline]

Rovaris M, Filippi M. Contrast enhancement and the acute lesion in multiple sclerosis. Neuroimaging Clin N Am 2000; 10: 705–16, viii–ix.[Web of Science][Medline]

Sarchielli P, Presciutti O, Tarducci R, Gobbi G, Alberti A, Pelliccioli GP, et al. (1)H-MRS in patients with multiple sclerosis undergoing treatment with interferon beta-1a: results of a preliminary study. J Neurol Neurosurg Psychiatry 1998; 64: 204–12.[Abstract/Free Full Text]

Sarchielli P, Presciutti O, Pelliccioli GP, Tarducci R, Gobbi G, Chiarini P, et al. Absolute quantification of brain metabolites by proton magnetic resonance spectroscopy in normal-appearing white matter of multiple sclerosis patients. Brain 1999; 122: 513–21.[Abstract/Free Full Text]

Sarchielli P, Presciutti O, Tarducci R, Gobbi G, Alberti A, Pelliccioli GP, et al. Localized (1)H magnetic resonance spectroscopy in mainly cortical gray matter of patients with multiple sclerosis. J Neurol 2002; 249: 902–10.[CrossRef][Web of Science][Medline]

Schiepers C, Van Hecke P, Vandenberghe R, Van Oostende S, Dupont P, Demaerel P, et al. Positron emission tomography, magnetic resonance imaging and proton NMR spectroscopy of white matter in multiple sclerosis. Mult Scler 1997; 3: 8–17.[Web of Science][Medline]

Schubert F, Seifert F, Elster C, Link A, Walzel M, Mientus S, et al. Serial 1H-MRS in relapsing-remitting multiple sclerosis: effects of interferon-beta therapy on absolute metabolite concentrations. Magma 2002; 14: 213–22.[CrossRef][Medline]

Seeger U, Klose U, Mader I, Grodd W, Nagele T. Parameterized evaluation of macromolecules and lipids in proton MR spectroscopy of brain diseases. Magn Reson Med 2003; 49: 19–28.[CrossRef][Web of Science][Medline]

Sharma R, Narayana PA, Wolinsky JS. Grey matter abnormalities in multiple sclerosis: proton magnetic resonance spectroscopic imaging. Mult Scler 2001; 7: 221–6.[Abstract/Free Full Text]

Simmons ML, Frondoza CG, Coyle JT. Immunocytochemical localization of N-acetyl-aspartate with monoclonal antibodies. Neuroscience 1991; 45: 37–45.[CrossRef][Web of Science][Medline]

Srinivasan R, Sailasuta N, Hurd R, Nelson S, Pelletier D. Evidence of elevated glutamate in multiple sclerosis using magnetic resonance spectroscopy at 3 T. Brain 2005; 128: 1016–25.[Abstract/Free Full Text]

Stevens J. Applied multivariate statistics for the social sciences. Hillsdale: Lawrence Erlbaum Associates; 1992.

Suhy J, Rooney WD, Goodkin DE, Capizzano AA, Soher BJ, Maudsley AA, et al. 1H MRSI comparison of white matter and lesions in primary progressive and relapsing-remitting MS. Mult Scler 2000; 6: 148–55.[Abstract/Free Full Text]

Tartaglia MC, Narayanan S, Francis SJ, Santos AC, De Stefano N, Lapierre Y, et al. The relationship between diffuse axonal damage and fatigue in multiple sclerosis. Arch Neurol 2004; 61: 201–7.[Abstract/Free Full Text]

Trapp BD, Peterson J, Ransohoff RM, Rudick R, Mork S, Bo L. Axonal transection in the lesions of multiple sclerosis. N Engl J Med 1998; 338: 278–85.[Abstract/Free Full Text]

Urenjak J, Williams SR, Gadian DG, Noble M. Specific expression of N-acetylaspartate in neurons, oligodendrocyte-type-2 astrocyte progenitors, and immature oligodendrocytes in vitro. J Neurochem 1992; 59: 55–61.[Web of Science][Medline]

Urenjak J, Williams SR, Gadian DG, Noble M. Proton nuclear magnetic resonance spectroscopy unambiguously identifies different neural cell types. J Neurosci 1993; 13: 981–9.[Abstract]

van Walderveen MA, Barkhof F, Pouwels PJ, van Schijndel RA, Polman CH, Castelijns JA. Neuronal damage in T1-hypointense multiple sclerosis lesions demonstrated in vivo using proton magnetic resonance spectroscopy. Ann Neurol 1999; 46: 79–87.[CrossRef][Web of Science][Medline]

Vrenken H, Barkhof F, Uitdehaag BM, Castelijns JA, Polman CH, Pouwels PJ. MR spectroscopic evidence for glial increase but not for neuro-axonal damage in MS normal-appearing white matter. Magn Reson Med 2005; 53: 256–66.[CrossRef][Web of Science][Medline]

Wolinsky JS, Narayana PA. Magnetic resonance spectroscopy in multiple sclerosis: window into the diseased brain. Curr Opin Neurol 2002; 15: 247–51.[CrossRef][Web of Science][Medline]

Wylezinska M, Cifelli A, Jezzard P, Palace J, Alecci M, Matthews PM. Thalamic neurodegeneration in relapsing-remitting multiple sclerosis. Neurology 2003; 60: 1949–54.[Abstract/Free Full Text]

Wyss M, Kaddurah-Daouk R. Creatine and creatinine metabolism. Physiol Rev 2000; 80: 1107–213.[Abstract/Free Full Text]

CiteULike    Connotea    Del.icio.us    What's this?

This article has been cited by other articles:

				M L Stromillo, M T Dotti, M Battaglini, M Mortilla, S Bianchi, K Plewnia, L Pantoni, D Inzitari, A Federico, and N De Stefano Structural and metabolic brain abnormalities in preclinical cerebral autosomal dominant arteriopathy with subcortical infarcts and leucoencephalopathy J. Neurol. Neurosurg. Psychiatry, January 1, 2009; 80(1): 41 - 47. [Abstract] [Full Text] [PDF]

				A M BLAMIRE The technology of MRI -- the next 10 years? Br. J. Radiol., August 1, 2008; 81(968): 601 - 617. [Abstract] [Full Text] [PDF]

				E. E. Benarroch N-Acetylaspartate and N-acetylaspartylglutamate: Neurobiology and clinical significance Neurology, April 15, 2008; 70(16): 1353 - 1357. [Full Text] [PDF]

				N. De Stefano, M. Filippi, D. Miller, P. J. Pouwels, A. Rovira, A. Gass, C. Enzinger, P. M. Matthews, and D. L. Arnold Guidelines for using proton MR spectroscopy in multicenter clinical MS studies Neurology, November 13, 2007; 69(20): 1942 - 1952. [Abstract] [Full Text] [PDF]

				A. Pulizzi, M. Rovaris, E. Judica, M. P. Sormani, V. Martinelli, G. Comi, and M. Filippi Determinants of Disability in Multiple Sclerosis at Various Disease Stages: A Multiparametric Magnetic Resonance Study Arch Neurol, August 1, 2007; 64(8): 1163 - 1168. [Abstract] [Full Text] [PDF]

				A. M. Pascual, I. Bosca, B. Casanova, M. C. Martinez-Bisbal, B. Celda, F. Coret, and L. Marti-Bonmati An Observation of Interest Relative to the Practice of Spectroscopic Measurements in Multiple Sclerosis. AJNR Am. J. Neuroradiol., November 1, 2006; 27(10): 2022a - 22023. [Full Text] [PDF]

This Article Abstract FREE Full Text (PDF) Supplementary Data All Versions of this Article: 128/11/2483    most recent awh640v1 Alert me when this article is cited Alert me if a correction is posted Services Email this article to a friend Similar articles in this journal Similar articles in ISI Web of Science Similar articles in PubMed Alert me to new issues of the journal Add to My Personal Archive Download to citation manager Search for citing articles in: ISI Web of Science (15) Request Permissions Disclaimer Google Scholar Articles by Caramanos, Z. Articles by Arnold, D. L. Search for Related Content PubMed PubMed Citation Articles by Caramanos, Z. Articles by Arnold, D. L. Social Bookmarking          What's this?

Online ISSN 1460-2156 - Print ISSN 0006-8950

Copyright © 2009 Guarantors of Brain

Oxford Journals Oxford University Press

• Site Map
• Privacy Policy
• Frequently Asked Questions

Other Oxford University Press sites: Oxford University Press Oxford Journals China Oxford Journals Japan American National Biography Booksellers' Information Service Children's Fiction and Poetry Children's Reference Corporate & Special Sales Dictionaries Dictionary of National Biography Digital Reference English Language Teaching Higher Education Textbooks Humanities International Education Unit Law Medicine Music Online Products Oxford English Dictionary Reference Rights and Permissions Science School Books Social Sciences Very Short Introductions World's Classics