http://www.medscape.com/LEA/AppliedNeuropsych/2001/v08.n03/an0803.01.pan/an0803.01.pan-01.html
Applied Neuropsychology 8(3):155-160,
2001.
J. W. Pan, Medical Department, Brookhaven
National Laboratory, Upton, New York, L. B. Krupp, L. E. Elkins, and P.
K. Coyle, Department of Neurology, State University of New York at Stony
Brook, Stony Brook, New York
Abstract
Recent studies have demonstrated the utility of magnetic resonance (MR) spectroscopic imaging to evaluate axonal integrity in patients with multiple sclerosis (MS). Patient status in MS is frequently assessed by the Expanded Disability Status Scale, which emphasizes ambulation but underestimates the contribution of cognitive factors. Yet, cognitive functions of memory and processing are known to be impaired in MS. We used quantitative MR spectroscopy to determine this relation between cognitive function and N-acetyl aspartate (NAA) levels. We find a significant correlation (r = .63, p < .005) for the left periventricular (PV) NAA concentrations with performance on the verbal Selective Reminding Test. Right PV NAA was significantly (p < .02) correlated with the Tower of Hanoi performance, with r = .58. [Applied Neuropsychology 8(3):155-160, 2001. © 2001 Lawrence Erlbaum Associates, Inc.]
Introduction
Multiple sclerosis (MS) is a central
nervous system disorder characterized by repeated cycles of white matter
damage, recovery, and injury. As the damage is believed to be mediated
by dysfunction of the immune system and the blood-brain barrier, an important
clinical characteristic of MS is the fluctuant nature of the symptoms in
both clinical time course and neuroanatomical involvement. This also makes
the clinical assessment of the disease complex because symptoms can wax
and wane on the order of days to weeks. The most commonly used outcome
measures in MS are the Expanded Disability Status Scale (EDSS; Kurtzke,
1984) and MRI. However, the EDSS is known to be limited through its emphasis
on ambulation (Fischer, Rudick, Cutter, & Reingold, 1999; Provinciali,
Ceravolo, Bartolini, Logullo, & Danni, 1999) and insensitivity to cognitive
function.
More recently, magnetic resonance
(MR) spectroscopy has been used to evaluate axonal integrity through the
measurement of N-acetyl aspartate (NAA). NAA is synthesized in neuronal
mitochondria (Clark, 1998) and has been shown to be a good indicator of
neuronal functionality (Ellis et al., 1998; Pioro et al., 1998). Other
metabolic measures available throughMRspectroscopy include creatine (Cr)
and choline (Ch). Both Cr and Ch are found in high concentrations in glia
(Urenjak, Williams, Gadian,&Noble, 1993). The ratio of NAA/Cr, in which
decreases in NAA are compounded by increases in Cr, has been used frequently
as an evaluation of the gliotic reaction that accompanies many neurodegenerative
diseases (e.g., as reported by Ellis et al., 1998).Howeverthe utility ofNAA/CrinMSmaybe
less, given that the reactionary Cr changesmayoccur late in the disease
process, whereas NAA is more dynamic, most likely reflecting the degree
of demyelination and remyelination (De Stefano et al., 1999). Thus, quantification
of the metabolites in vivo is important, particularly given the pathophysiology
of MS.
Our goal was to evaluate the relation between NAA levels and clinical status as assessed by the EDSS and cognitive function, emphasizing those areas known to be affected in MS. We applied recently developed methods of quantification in spectroscopic imaging (Pan, Twieg, & Hetherington, 1998) to evaluate periventricular (PV) NAA. We chose this area because previous studies have found it to be chronically involved with MS (Narayanan et al., 1997; Pan et al., 1996). The cognitive evaluation was performed using tests focusing on memory function because it is the most common area of impairment in MS (Rao, Leo, & St. Aubin-Faubert, 1989; Thornton & Raz, 1997; Wishart & Shapiro, 1997). We selected a verbal and a visual memory task, including a measure of auditory attention and a visuospatial problem-solving task. In general terms, the verbal memory task and measure of auditory attention were considered to represent "left hemispheric" function, and the visual memory and visuospatial problem-solving task were considered to represent "right hemispheric" function (Lezak, 1995). We then combined the quantified right and left PV NAA levels and compared these to cognitive testing.
Methods
Clinical
Nineteen patients (8 women, 11 men) were recruited from the State University of New York (SUNY) at Stony Brook Comprehensive MS Clinic. Mean age was 44 (SD = 9.3), with a range from 24 to 60 years. They were clinically described as relapsing remitting (RR; n = 7), secondary progressive (SP; n = 5), and primary progressive (PP; n = 7). EDSS scores ranged from 1.0 to 6.5 (M = 3.3, SD = 1.5). All studies were performed with Institutional Review Board approval from SUNY Stony Brook and Brookhaven National Laboratory.
Cognitive Measures
The neuropsychological battery consisted
of verbal and visual memory tests, a test of auditory attention, and a
test of visuospatial problem solving. These four measures were administered
as part of amore extensive battery and selected to represent a range of
functions.
The two memory tests were the Selective
Reminding Test (SRT) and the 10/36 Spatial Recall Test (10/36), both from
the Brief Repeatable Battery (BRB; Rao & the Cognitive Function Study
Group, 1990) developed specifically for use in patients with MS. The SRT
is a six-trial word-list learning task (Buschke & Fuld, 1974) and the
sum of recall across the trials was used as the representative score. The
10/36 is a three-trial test requiring the participant to learn an arrangement
of checkers on a board, with the total recall at the third trial as the
representative score. For auditory attention, we used the Digit Span Test
(Wechsler, 1981), a task requiring the repetition of numbers of increasing
length. The Tower of Hanoi (TOH; Simon, 1975) is a task of visuospatial
problem solving, requiring the participant to arrange disks on three pegs
into a given design in as few moves as possible. Total score was used for
both the Digit Span and TOH. All patients participated in all components
of the BRB except for one patient who did not complete the TOH.
In all cases, the cognitive testing was performed within 3 months of MR spectroscopy study. Analysis of the neuropsychological andMRdata were blinded, with the two evaluations being performed independently.
Magnetic Resonance
The Brookhaven National Laboratory Varian Siemens 4T Inova system and volume head coil was used. Inversion recovery gradient echo scout images (TR 2.5 sec, TE 15 msec) were obtained to determine the midline sagittal plane. Double oblique images were then taken orthogonal to the midline sagittal to define the 1-cm slice of interest, taken through the posterior and anterior aspects of the corpus callosum. To provide tissue-type information for the spectroscopic images, T1-based tissue segmentation was performed, using a rapid inversion recovery sequence (Pan et al., 1998). This allowed decomposition of each spectral voxel into its component tissue types (i.e., gray, white matter, and cerebrospinal fluid [CSF]), taking into account the intrinsic point spread function incurred by the two-dimensional phase encoding.
The spectroscopic imaging data were acquired with a TE 50 msec, TR 2 sec spin echo. Water suppression was achieved with a semiselective refocusing pulse. Two dimensions of phase encoding were applied (24 Å~ 24) on a field of view of 192 mm, giving a nominal spatial resolution of 0.64 cc (effective voxel size = 1.2 cc; acquisition time = 17 min). Metabolite quantification was performed using a rapid water spectroscopic image (24 Å~ 24 encoding, field of view 192 mm) taken through the ventricles (TR 0.85 sec, TE 50 msec; acquisition time = 8 min). Because the acquisition steps are equivalent between the metabolite and water spectroscopic images, the data are directly comparable. The internal reference (i.e., pure ventricular CSF signal) was calculated by regression of the spectroscopic water signal relative to CSF content as provided by the tissue segmentation data. Ventricular CSF content was assumed 110 M, with a T2 500 msec and T1 3.5 sec. A B1 map was also acquired for coil homogeneity correction, which was applied to both the water and metabolite spectroscopic image. Tissue NAA, Cr, and Ch were quantified relative to ventricular CSF, with corrections included for non-CSF tissue volume as determined by tissue segmentation data. The entireMRstudy typically lasted 70 min.
Data analysis was performed using spectral processing for optimal signal-to-noise (250 Hz convolution difference, 5 Hz gaussian broadening). For all patients, approximately 100 voxels were analyzed. The PV data were selected through elimination of those voxels with greater than 30% CSF and by anatomical placement. For each patient, the right and left posterior PV NAA, Cr, and Ch content were defined, as were metabolite ratios. No attempt to avoid lesions was taken in these data. This was done because the presence of lesions represents the status of the patient and in those cases where chronic lesions display significant CSF replacement, the tissue segmentation correction would minimize such effects. Corrections for T2 and T1 relaxation were performed using values previously reported (Hetherington et al., 1994).
Results
Table 1 describes the patient characteristics
of the studied group, indicating duration of disease, treatment status,
and clinical subtypes.
| ID | MS Subtype | Age (Years) | EDSS | Disease Duration (Years) | NAA (Millimolar) | |
|---|---|---|---|---|---|---|
| Right | Left | |||||
| 201 | RR | 34 | 1.0 | 2 | 9.84 | 10.97 |
| 202 | RR | 33 | 2.0 | 1 | 9.05 | 9.79 |
| 207 | RR | 24 | 2.0 | 2 | 8.05 | 8.76 |
| 208 | RR | 41 | 2.0 | 1 | 8.51 | 6.89 |
| 211 | RR | 39 | 2.0 | 1 | 9.54 | 11.07 |
| 240 | RR | 38 | 1.5 | 4 | 10.24 | 10.26 |
| 243 | RR | 42 | 3.5 | 2 | 8.11 | 8.36 |
| 204 | SP | 60 | 2.5 | 20 | 7.37 | 6.24 |
| 206 | SP | 53 | 4.5 | 8 | 7.87 | 9.79 |
| 212 | SP | 43 | 4.0 | 5 | 9.39 | 5.89 |
| 223 | SP | 59 | 6.0 | 12 | 7.26 | 9.12 |
| 226 | SP | 51 | 3.5 | 0 | 8.28 | 8.97 |
| 203 | PP | 44 | 2.0 | 0 | 10.99 | 10.03 |
| 205 | PP | 55 | 5.0 | 4 | 9.65 | 10.01 |
| 210 | PP | 49 | 4.0 | 1 | 6.30 | 9.21 |
| 216 | PP | 43 | 4.5 | 0 | 9.32 | 9.73 |
| 217 | PP | 39 | 3.5 | 3 | 10.58 | 11.05 |
| 222 | PP | 42 | 3.0 | 5 | 8.77 | 9.34 |
| 239 | PP | 53 | 6.5 | 14 | 9.46 | 9.80 |
Table 1. Patient Characteristics
Note: MS = multiple sclerosis; EDSS = Expanded Diability Status Scale; NAA = N-acetyl-asparate; RR = relapsing remitting; SP = secondary progressive; PP = primary progressive.
Figure 1 shows an inversion recovery gradient echo, the segmentation data from the same slice, and selected spectra from a patient with SP MS. The slice demonstrates damaged posterior PV white matter. The range of the PV NAA levels measured was 6 mM-12 mM, and explicitly excludes CSF partial volume effects as previously described.
Figure 1.
Segmented image (a), CSF image (b), white matter image (c), gray matter
image (d), magnitude NAA image (e) and spectra (f) from a patient with
secondary progressive MS. The numbered positions on (a) indicate the location
of the spectra shown in (f). The resonances NAA, Cr, and Ch are indicated
in (f) together with the millimolar concentrations at each location.
We analyzed the data for a relation
between left hemispheric PVNAAand the SRT, ameasure of verbal function
and memory. These data demonstrated a significant correlation (r = .63,
p = .004; see Figure 2). Of note is that the right hemispheric PV NAA did
not correlate with the SRT measure (p > .64). However, the TOH score, a
test of conceptual planning, did significantly correlate with the right
but not left hemispheric PV NAA (p = .013 vs. p = .60), with r = .58 (Figure
3). Neither right nor left PV NAA was significantly correlated with the
10/36 (visual memory) or Digit Span scores. Interestingly, in this group
of patients, no correlation was detected for lateralized ratios of NAA/Cr
or NAA/Ch to cognitive performance.
Figure 2. Regression of Selective Reminding Test (verbal memory) with measurements of left periventricular NAA from all the data. These data are significantly (p < .005) correlated with a Pearson correlation coefficient of r = .63.
Figure 3. Regression
of performance on the Tower of Hanoi test (visuospatial conceptual planning)
with measurements of right periventricular NAA. These data are significantly
(p < .02) correlated with a Pearson correlation coefficient of r = .58.
The spectroscopic data were also
analyzed relative to EDSS. The EDSS is not anticipated to specifically
relate to a given hemisphere, and no correlation was seen between the EDSS
scores and the mean PV NAA measurements (Figure 4).
Discussion
NAA
We have combined cognitive testing
with quantitative MR spectroscopic imaging to find a strong relation between
left and right hemispheric cognitive function with the lateralized concentration
of PV NAA in MS patients. We believe these data are a consequence of two
factors. First, given the prevalence of lesions in PV tissue with MS, we
interpret the PV NAA measurements to be a manifestation of overall chronic
disease activity. It may be argued that given the high likelihood of ventriculomegaly
in advanced cases of MS, the study of PV tissue will bias toward those
patients with more advanced disease. However, because CSF contributions
were factored into the concentration measurements, we believe that such
effects are minimal. Additionally, lesions that are frequently present
periventricularly were also segmented to exclude CSF, thus giving the optimal
tissue concentration of NAA.
Second, we observed that the lateralized
spectroscopic measurements correlated with cognitive function. This is
analogous to the work reported by Lee et al. (2000) who demonstrated lateralization
of motor impairments with asymmetries in NAA levels in the capsular white
matter ofMSpatients. In the datam reported here, this correlation reflects
the overall neuropsychological functionality of these regions. MR and PET
studies have shown the SRT to be more clearly left hemispheric (for review,
see Smith & Jonides, 1999). Although localization of planning functions
that are tested by the TOH are not well defined aside from that of the
frontal lobe (Wishart & Shapiro, 1997), this test may emphasize the
right hemisphere, given its spatial components.
The other two of our four cognitive
measures -- the 10/36 and the Digit Span -- did not significantly correlate
with the NAA values. The nature of these tasks would have also predicted
general lateralization, representing right and left hemispheric function
due to visual versus verbal modalities, respectively. However, both tasks
are relatively easier than the SRT and TOH and have a more limited range
of possible scores. Further, these two measures are arguably less representative
of the typical cognitive impairments that occur in MS (Wishart & Shapiro,
1997). Therefore, these tasks may have been overall less sensitive to the
NAA values. Future studies including a larger battery of both visual and
verbal tests will determine the specificity of NAA to both hemisphere and
region. PAN ET AL.
NAA/Cr and NAA/Ch
We did not detect a significant relation
between cognitive testing and the ratios of NAA/CrorNAA/Ch.Two factors
are likely to be contributing to this finding. First, these data support
the view that the gliotic changes -- believed to be most likely causing
changes in concentration of Cr or Ch, or both -- are much more variable
in MS. This is believed to reflect the changes in astrocyte and oligodendrocyte
number and functionality in the development and recovery of MS lesions.
For example, as described by Prineas et al. (1993), lesion evolution includes
preservation of oligodendrocytes during demyelination, which subsequently
can cycle either through remyelination or oligodendrocyte destruction with
astrogliosis and axonal loss. Thus, decreases in NAA that involve axonal
shrinkage, loss, or both, may occur with or without an increase in Cr.
This relative independence of NAA to Cr, Ch, or both, may also be the reason
for the absence of a clear correlation of the metabolite ratio to frontal
lobe executive function as reported by Foong et al. (1999). In that study,
only those patients with the most severely abnormal metabolite ratios (i.e.,
2 SD below normal) display poor cognitive function. However,thismaybe expected
because a profoundly decreasedNAAcan dominate the NAA/Cr ratio. By direct
quantification ofNAAlevels, we are decreasing the biological variability
introduced by the Cr measurement.
Second, we studied several subtypes
of MS patients, including PP, RR, and SP groups. Thus, as has been suggested,
particularly for the PP group, differences in pathology would result in
varying degrees of gliosis and oligodendrocyte loss in this group of patients.
Therefore, changes in Cr andChwould be variable, and detection of a decrease
in NAA/Crdifficult (in fact, a trend indicating a decrease in the rightPVNAA/Chwasdetected
with the TOH test, p = .057). However, NAA, located in the neuronal compartment,
would be expected to be sensitive to axonal function and would therefore
be commonly affected in all of the MS subtypes.
The inclusion of different MS subtypes
in the patient group may also be contributing to the apparent lack of a
clear relation between the EDSS scores and themeanPV NAA concentrations.
However, we believe this to be an important absent finding because this
is consistent with the known tendency of PP patients to fractionally bear
a larger disease burden in the spinal cord and brain stem. We would anticipate
that this would result in a weaker relation between hemispheric function
and EDSS. Thus, the evaluation of disease status in PP patients may optimally
include several measures of functionality focusing on the hemispheres,
brain stem, and spinal cord.
Conclusion
In summary, we have used quantitative
MR spectroscopic imaging to measure PV NAA content, finding a significant
correlation with lateralized tests of cognitive function. We believe that
this finding further supports the belief that NAA is a measure of neuronal
and axonal integrity. These data also demonstrate that cognitive performance
in MS can be related to a metabolic parameter. Finally, that we observed
a consistent lateralization between the NAA and cognitive performance is
suggestive that the axonal damage resulting in decreased NAA is, itself,
contributing to performance losses.
References
© 2001 Lawrence Erlbaum Associates,
Inc.
Copyright © 1994-2001 by Medscape
Inc