Abstract
Objective: To quantitate the extent of neuronal cell loss in MS via the whole brain’s N-acetylaspartate (NAA) concentration (WBNAA).
Methods: Because NAA is assumed to be present only in neuronal cell bodies and their axons, we measured WBNAA as a marker for viable neurons in 12 patients (9 women and 3 men, 26 to 53 years of age) suffering from relapsing-remitting (RR) MS for at least 5 years and compared them with 13 age- and sex-matched normal controls. Total brain NAA was determined with proton MR spectroscopy, and WBNAA was obtained by dividing it by the total brain volume, calculated from high resolution MRI.
Results: The WBNAA of the RR MS patients was lower than their matched controls (p < 0.005). This difference was greater among older than younger subjects. The linear prediction equations of WBNAA with age indicate a faster, ×10, decline in the patients, ∼0.8% per year of age (p = 0.022).
Conclusion: The age-dependent decrease of whole brain N-acetylaspartate (WBNAA) in the patients suggests that progressive neuronal cell loss is a cardinal feature of this disease. WBNAA offers a quick, highly reproducible measure of disease progression and may be an important marker of treatment efficacy in MS as well as other neurodegenerative diseases.
MS is the most common inflammatory disease of the CNS and is the most frequent cause of nontraumatic neurologic disability in young and middle-aged adults.1 Although T2-weighted MRI (T2WI) is considered to be among the more sensitive diagnostic tools for MS,2 the correlation between the total T2 lesion volume, (commonly used to gauge the burden of the disease) and clinical disability has been inconsistent.3-9 There are several reasons for this discrepancy. First, different pathologic processes (e.g., edema, demyelination, gliosis, and axonal loss), which appear similar on T2WI, may individually, or in concert, affect the neurologic function.10-11 Second, microscopic disease in “normal-appearing white matter” (NAWM) may have clinical significance in addition to the lesions on T2WI.12-14 Third, lesion location was not considered in these correlations, although their effects clearly do vary with position. Fourth, spinal cord disease, although common and a frequent cause of disability, is usually not assessed.15,16
This suggests the need for a more sensitive correlate for the assessment of MS in addition to or instead of the T2 lesion load. One candidate measure could be the amount of neuronal loss,17 assessable from the level of the amino acid derivative, N-acetylaspartate (NAA), which is believed to be present almost exclusively in neurons and their dendritic and axonal processes.17-19 Although the significance NAA levels in MS brain remains unclear, previous studies have reported its decrease in lesions12,13,20-26 and sometimes in NAWM as well.12,13 The methyl group of NAA yields the most prominent peak in the brain’s 1H nuclear magnetic resonance (NMR) spectrum (1H-MRS) and can be easily quantitated.21,27
Unfortunately, current localized 1H-MRS methods restrict the volume(s)-of-interest (VOI) to <500 cm3, located completely within the brain. This must be done to protect the small metabolites’ peaks from being obscured by the intense signals of subcutaneous lipids and the skull’s bone marrow, which resonate at similar chemical shifts.28,29 Much smaller than the brain, such VOIs must be image guided onto the pathology of interest, which poses two problems. First, 1H-MRS is subject to the assumption that relevant metabolic changes occur only at sites of imaging abnormalities. Second, unavoidable repositioning misalignment between measurements can lead to uncertainty in the interpretation of NAA data acquired in longitudinal studies.
Whole brain N-acetylaspartate (WBNAA) can simultaneously address the problems of partial coverage, serial misalignment, and the need for image guidance. A recent study has shown that the inter-individual WBNAA variations in a cohort of normal, middle-aged women (range, 42 ± 5 years of age) are small, <3% (p < 0.01), and the intra-individual temporal variation is smaller still, <2.5%.30 Such a tight distribution is our motivation to compare the WBNAA of relapsing-remitting (RR) MS patients with their age- and sex-matched normal controls as a highly reproducible marker of disease. This article examines whether consistent NAA deficits, indicating the global neuronal/axonal losses sustained throughout the brain of MS patients, can be quantitated.
Methods.
Subjects.
Twelve patients, nine women and three men suffering from RR MS for at least 5 years and enrolled in a long-term follow-up study, were recruited. Although they ranged in age from 26 to 53 years, four of them were clustered at 30 ± 3 years, and another four at 40 ± 4 years, as shown in the table. Five of them were on immunomodulating treatment regimens for 3 years or more, and seven were untreated. Thirteen age-matched normal controls, i.e., not having any known neurologic disease, also were recruited (see table). Based on our previous study indicating statistically constant WBNAA over 42 ± 5 years, we consider an “age match” to be within ±10%.30
Summary of WBNAA from 12 RR MS patients and 13 normal controls
1H-MRS sequence and instrumentation.
A radio-frequency and gradient 1H-MRS pulse sequence to obtain WBNAA should meet the following criteria:
-
1. be non/minimally T1- or T2-weighted because these relaxation times in pathologies are generally unknown31;
-
2. be immune to lipid signals, which are very intense at non/short echo-time (TE) and occur in the same spectral region as the methyl resonance of NAA; and
-
3. have high signal-to-noise-ratio (SNR) to ensure that observed variations are of biological origin and not noise.
A simple sequence to address all these needs was introduced recently and demonstrated both theoretically and in vivo in a cohort of normal women.30
Using that sequence, the reported WBNAA 1H-MRS study was conducted at 1.5 T in a full-body imager with its standard circularly polarized head coil (Magnetom 63SP; Siemens AG, Erlangen, Germany). Our three-dimensional chemical shift imaging–based shimming yielded consistent 9 ± 0.5 Hz full-width at half-maximum water line from the whole head.32 Free induction decays were acquired with 2,048 complex points at 0.5 msec/point. At a recycle-time (TR) of 10 seconds to ensure full equilibrium of all species, the MRS cycle took ∼2.5 minutes. To further improve the precision of the result, up to 5 such back-to-back cycles were acquired. Together with loading, shimming, tuning, and 1H-MRS, the entire procedure took 15 to 20 minutes.
WBNAA quantitation.
After the subject’s 1H-MRS, a 3-L sphere containing 130 mmol NaCl and 5 mM NAA in water was placed at approximately the same position as the head. It was shimmed to a similar water linewidth and subjected to the same protocol. Subject and phantom NAA spectral peak areas, SV and SP, respectively, were integrated using the manufacturer’s software. The amount of brain NAA, QNAA, was obtained similarly to Soher et al.33 
where VV180° and VP180° are the voltages into 50Ω of radio frequency power required for nonselective, 1-msec 180° inversion pulses, and gauge the “receive” sensitivity. Because the acquisition is neither T1- nor T2-weighted and both phantom and head sample similar radio frequency, B1 and magnetic B0 field distributions, no corrections for any of these parameters were made.
The QNAA from was divided by the subject’s brain volume, VB, to convert it into a concentration. VB was obtained from high-resolution MRI: 256 × 256, 220 mm field of view, 3-mm slice thickness, proton density, and T2-weighted fast-spin-echo (TE = 16/80 ms; TR = 2,500 ms). These images were post-processed with the 3-D VIEWNIX software package described by Udupa et al.34,35 Specifically, based on several manually pre-selected intensity points in the CSF, gray and white matter, 3DVIEWNIX uses the fuzzy-connectedness principle to create a brain mask. The mask is then manually corrected to remove the skull or external CSF or to add missing brain parts. Finally, VB is obtained by “counting” the image pixels within the mask. The method has shown better than 99% reproducibility.36
Results.
We determined the WBNAA of the 12 RR MS patients and 13 healthy age- and sex-matched controls in the table. Up to five back-to-back WBNAA observations from each subject, patient and control, are shown in the figure, and their results are summarized in the table. As a preliminary step in their statistical analysis, all subjects were categorized into four groups, according to their disease status (RR MS patient or control) and whether their age was below the median of 33. Mann-Whitney U tests were conducted to compare the WBNAA measurements among these four groups. Using a Bonferroni correction for multiple hypothesis tests, we observed the WBNAA of the older (>33 years) control subjects (median = 11.87 mmol) to be significantly greater (p = 0.0052) than that of older MS patients (median = 9.48 mmol). Although a similar result was observed for the younger patients (median = 11.87 mmol) and controls (median = 12.38 mmol), that difference was not significant at the 5% level.
Figure. The WBNAA concentration versus age data from all subjects (up to 5 measurements from each) in the table. The solid lines are the prediction for Ŷ from equations 3 (above) and 2 (below) with their respective lower and upper 95% confidence intervals (dashed lines).
To further characterize the relationship between WBNAA, age, and disease status, least squares regression and analysis of variance was used. Because multiple measurements were taken on each subject, each could not be analyzed under the assumption of independence. Instead, a mixed model analysis was implemented with the stochastic dependence of the WBNAA measurements modeled through a random classification factor representing subject identity. As a precaution, the results were compared with those obtained by averaging the measurements of each person and analyzing the results via weighted least squares. Because the two methods gave consistent results and the latter did not require assumptions concerning the dependence structure of the data, conclusions were based on the more informative mixed model analysis.
This analysis showed that for the MS patients, WBNAA was correlated with age (p = 0.022) over the 26 to 53 years age span studied. The following linear prediction equation was used: 
where Ŷpatient is the predicted concentration for an RR MS patient AGE years old. In contrast, the following linear prediction equation for the WBNAA concentration of the controls was used: 
Thus, over approximately the same age range, 25 to 52 years, the WBNAA of the RR MS patients, on average, is predicted to decrease almost 10-fold faster than their controls, about 0.8% per year. In contrast, the WBNAA decline with age predicted for the controls was insignificant (p = 0.35). These results are summarized in the figure, which shows the data from all subjects, the prediction equations 2 and 3 for the MS and control subjects, and the lower and upper 95% confidence bands for the predicted mean WBNAA value as a function of age.
A variance components analysis showed that even after accounting for WBNAA variations attributable to age, its person to person variation was greater (p = 0.023) among patients (estimated SD = 1.41) than among controls (SD = 0.76). The variation in WBNAA measurements taken on a single individual was also observed to be higher (p = 0.048) in patients (SD = 1.35) than controls (SD = 0.91). The temporal changes in the patients may be influenced by factors not examined in this study (e.g., disease duration, whether a patient was on medication, and if so of what kind and for how long. Although this information is provided in the table, the subject pool was too small to correlate these parameters.
Note that a small sample size may usually imply low statistical power and small probability of observing significant results. Consequently, in small samples it could be difficult to conclude a lack of significance. Conversely, the consistent observed NAA deficit in our 12 patients must indicate a true effect because of the relatively small sample size.
Discussion.
Although demyelination was first believed to be the main cause of the clinical deficits of MS,10,11,20 recent pathologic studies have supported earlier observations of neuronal loss,21 which may have a significant impact on clinical progression. To assess neuronal cell viability, in vivo MRS regional detection has shown NAA decrease in the lesions12,13,21-26 and in NAWM,12-14 indicating that axonal loss was also occurring, possibly by Wallerian degeneration.12,13,37 This hypothesis was further supported by recent neuropathology showing axonal losses and demyelination in old lesions.17 In contrast, others proposed that the NAA decrease is more likely caused by (reversible) reduced concentration in the axons rather than their death.12,20,38 This could arise, for example, from impaired blood-brain-barrier, cytotoxic edema, or efflux of intracellular NAA subsequent to cellular swelling in acute lesions.20,38 It might also explain the sustained decrease of NAA in new gadolinium-enhancing lesions.23,24
Because only the amount of NAA in the VOI and/or the voxel(s) within it is detected, not its (microscopic) distribution, 1H-MRS cannot directly distinguish between neuronal/axonal death or a reduction in the concentration of NAA in them. In contrast, WBNAA contains the entire brain, not just one (or few) lesion(s). Therefore, because typical lesion burdens in RR MS are <50 cm3 or <3% of the brain volume, 10 to 20% of WBNAA losses (see the table and figure) cannot be explained by changes within the lesions alone. Such extensive deficits can only be accounted for in terms of significant NAWM involvement, lending support to the axonal death hypothesis.12,13,37 The predicted decline of WBNAA with age in and the figure lends additional support to this theory. Furthermore, this study indicates that this decline is a continual one-way process. Therefore, even if the decreased concentration is also correct, in the long run, either mechanism would lead to the same end: total neuronal/axonal loss in the affected regions.
The different levels of NAA cited in the literature, especially in acute lesions, may also be due to the particular 1H-MRS acquisition methodology used: specifically, 1) TEs, ranging from 30 to 270 msec, incurring different T2-weighting; 2) voxel sizes of 1 to 8 cm3, which varied the partial-volume-effect; and 3) different TRs, which vary the T1-weighting of NAA.12,14,21,22
The localized MRS used in these studies made them susceptible to variations in repositioning the small VOI(s). This is especially deleterious in serial studies used to evaluate NAA loss and subsequent recovery.13,22,39 In contrast, WBNAA is not T1- or T2-weighted or restricted to a pre-selected/pre-positioned VOI,30 and, therefore, does not suffer from any of these difficulties. However, it does maximize the partial-volume-effect because the entire brain is averaged into one number.
To our knowledge, this is the first time that NAA is quantitated over the entire brain. In MS, its concentration was found to be significantly lower compared with age-matched controls and lower in older compared with younger patients. The evolution of WBNAA differences between patients and controls predicted from this study reflects the known course of RR MS, usually affecting young adults in their 20s. At that early stage, no significant difference was seen between patients and controls. However, as they age, the patients’ WBNAA is predicted to decline on average 0.8% per year (see , reflecting the degenerative nature of MS. The predicted stability of the controls over that period (equation 3) may reflect that normal age-related VB decline leads to proportional loss of neurons/axons, QNAA, in that parenchyma. Since WBNAA = QNAA/VB, proportional losses are undetectable. Therefore, since NAA decrease is believed to be secondary to neuronal loss, this study supports the hypotheses that neuronal/axonal losses are consequent to MS and that they worsen as the disease progresses over time.
Because our study examined a single time-point for each patient, it did not directly measure the rate of NAA decline over time in any given patient. Therefore, we cannot assess the effects of immunomodulating medication (interferon beta and glatiramer acetate) on the WBNAA of the individual patients in this study on these medications. Evaluation of the effects of such treatment will depend on longitudinal follow-up of our patient cohort. Such studies are currently under way.
Acknowledgments
Supported by a Biomedical Technology Development grant from the Whitaker Foundation, a grant from the French Society of Radiology, and NIH grants NS33385, NS37739, and NS29029.
- Received August 13, 1999.
- Accepted October 2, 1999.
References
Letters: Rapid online correspondence
REQUIREMENTS
You must ensure that your Disclosures have been updated within the previous six months. Please go to our Submission Site to add or update your Disclosure information.
Your co-authors must send a completed Publishing Agreement Form to Neurology Staff (not necessary for the lead/corresponding author as the form below will suffice) before you upload your comment.
If you are responding to a comment that was written about an article you originally authored:
You (and co-authors) do not need to fill out forms or check disclosures as author forms are still valid
and apply to letter.
Submission specifications:
- Submissions must be < 200 words with < 5 references. Reference 1 must be the article on which you are commenting.
- Submissions should not have more than 5 authors. (Exception: original author replies can include all original authors of the article)
- Submit only on articles published within 6 months of issue date.
- Do not be redundant. Read any comments already posted on the article prior to submission.
- Submitted comments are subject to editing and editor review prior to posting.
You May Also be Interested in
Hastening the Diagnosis of Amyotrophic Lateral Sclerosis
Dr. Brian Callaghan and Dr. Kellen Quigg
► Watch
Alert Me
Recommended articles
A type 2 biomarker separates relapsing-remitting from secondary progressive multiple sclerosis
A multicenter 1H-MRS study of the medial temporal lobe in AD and MCI
Localized proton magnetic resonance spectroscopy in relapsing remitting versus secondary progressive multiple sclerosis
Nonconventional MRI biomarkers for in vivo monitoring of pathogenesis in multiple sclerosis