Abstract
Objective: To assess normal-appearing white matter (NAWM) characteristics by magnetic resonance spectroscopy (MRS) and diffusion-weighted imaging (DWI) in elderly subjects.
Methods: The authors studied 60 volunteers (mean age 72.6 years; SD 4.7; range 64 to 84 years) without signs of neurologic illness. They used DWI and spectroscopic imaging to investigate whether there were changes in the NAWM that related to the presence of white matter hyperintensities (WMH).
Results: The authors found a correlation (p < 0.001) between the apparent diffusion coefficient in the NAWM and the total volume of WMH. The metabolite ratios N-acetylaspartate/creatine and N-acetylaspartate/choline of the NAWM also correlated significantly with total WMH volume. These correlations were independent of age.
Conclusions: Damage associated with WMH is detectable in NAWM.
In the elderly, deep white matter hyperintensities (WMH) are a common finding on proton density, T2-weighted, and fluid-attenuated inversion recovery (FLAIR) MRI. WMH are associated with increasing age and the presence of vascular risk factors.1 WMH are more common in people with dementia2,3⇓ and correlate with cognitive impairment in nondemented individuals.4 Histopathologically, the lesions coincide with areas of gliosis, loss of myelin, and loss of fibers.5,6⇓ The WMH are associated with lower cognitive functioning, especially in tasks such as attention and executive function.4 A recent histopathologic study report suggested that subtle damage is present outside the WMH.7
Diffusion-weighted imaging (DWI) allows measurement of the apparent diffusion coefficient (ADC) of water in the brain. Water diffusion in the brain is restricted by cellular boundaries, and the ADC increases with any breakdown in cellular structure.8 The ADC is increased in areas of WMH in elderly people,9 and in a small study, increased whole brain ADC associated with increased WMH severity has been reported.10 An increase in the ADC of the normal-appearing white matter (NAWM) has been reported in subjects with ischemic leukoaraiosis compared to controls.11 An association between visual rating of WMH and the ADC of the NAWM has been observed in a group largely consisting of people with carotid stenosis.12
MR spectroscopy (MRS) provides a means of quantifying levels of metabolites in the brain. N-acetylaspartate (NAA) is a metabolite measured by MRS, and is chiefly present in neuronal cells, and damage or dysfunction of neurons and their associated axons is linked with a decrease in NAA. NAA levels are decreased globally in Alzheimer disease,13 and decreased levels of NAA in regions of WMH have been reported.14 We hypothesized that the presence of WMH visualized on FLAIR images represents one extreme of a pathologic process that affects far more of the white matter, and that would be manifest as changes on DWI and MRS in white matter that appeared normal on FLAIR images.
Methods.
The study site and population.
The study was based on a cross-sectional sample of normal volunteers, and patients who attended the outpatient memory clinic of the Newcastle General Hospital, but were found, after clinical and neuropsychological assessment, not to have an organic cause for their cognitive complaints. Subjects were recruited using the following inclusion criteria: age between 64 and 84 years, intact global cognitive functioning, independence in daily activities, and presence of an informant. Exclusionary criteria were history or signs suggestive of dementia or any other organic disorder, stroke,15 leukoencephalopathies of nonvascular origin, severe unrelated neurologic or physical diseases (e.g., epilepsy, carcinoma), acute systemic disease in the previous month, severe current psychiatric disorders, alcohol abuse, use of neuroleptics or benzodiazepines, uncorrected visual or hearing loss, or motor deficiency severe enough to interfere with the tests’ execution. Written, informed consent was obtained from all enrolled subjects. In total, 63 people were enrolled. Two subjects could not tolerate the MRI scan, and one subject was withdrawn owing to an MRI suggestive of a tumor.
MRI procedure.
Subjects were scanned using a Philips 1.5 Tesla Intera MRI scanner (Philips, Eindhoven, the Netherlands). A three-dimensional gradient echo T1-weighted sequence (repetition time [TR] 10 msec, echo time [TE] 4.6 msec, flip angle 20°, 1 mm cubic voxels) was obtained, together with axial fast spin echo T2-weighted (TR 5,676 msec, TE 110 msec) and FLAIR (TR 10,000 msec, TE 100 msec, inversion time [TI] 2,500 msec), to visualize and determine the volume of the WMH. The FLAIR and T2-weighted images were obtained with a pixel size of 0.98 × 0.98 mm, slice thickness 5 mm, interslice gap 0.5 mm.
A diffusion weighted scan was also performed with a spin echo echoplanar image (EPI) sequence with diffusion gradients applied in three orthogonal directions (b value of 1,000 sec/mm2, TR = 4,279 msec, TE = 106 msec, pixel size 1.95 × 1.95 interpolated to 0.98 × 0.98 mm, slice thickness 5 mm, interslice gap 0.5 mm).
The subjects were examined with a fast spin echo spectroscopic imaging sequence (TR = 1,500, TE = 272 and 544 msec, 2 slices, field of view = 250 mm, 20 mm slice thickness, k space sampled on a 24 × 24 grid). The slices were oriented axially, and located superior to and parallel with the anterior-posterior commissure line. The default automatic data processing was carried out on the scanner—the raw data multiplied with a 3 Hz Gaussian, then a 1.5 Hz line narrowing exponential filter. Bo correction was carried out automatically using non-water–suppressed spectra.
The MRI and MRS data were transferred to a personal computer for analysis. Total imaging time was 50 minutes.
The three-dimensional, FLAIR, and T2 sequences were performed on 60 subjects. There was minimal movement artifact present. The diffusion scan was performed on 57 subjects. The spectroscopic examination was performed in all patients, but on reviewing the scans before further analysis, seven scans were rejected due to the line width of all the spectra being broader than 10 Hz.
Data analysis.
The basic steps of the analysis were as follows:
-
Segmentation of the T1-weighted and FLAIR data to produce images showing the distribution of CSF, gray and white matter, and WMH, and calculate the total WMH volume and brain volume.
-
Resampling of the segmented images to the spatial resolution of the spectroscopy data and linear regression between metabolite ratios and tissue content in the spectroscopic voxels to derive values for mean metabolite ratios in gray matter, NAWM, and WMH over the whole spectroscopic image.
-
Use of segmented images to locate NAWM and WMH on the DWI, and thus obtain values for the global median ADC for NAWM and WMH.
The T1-weighted data were segmented using SPM99 (http://www.fil.ion.ucl.ac.uk/spm/)16-18⇓⇓ and regions of WMH were segmented from the FLAIR images using automated image processing software developed in house. We wished to validate the WMH segmentation process, so all the scans were separately segmented using a semiautomated threshold technique by an observer (T.M.) who was blinded to the automated segmentation results.
We quantified the MRS metabolite ratios using fully automated spectral fitting software19 to estimate the peak areas of NAA, choline (Cho), and creatine (Cr). We used routines written in IDL (Research Systems, Boulder, CO) to calculate the regressions between metabolite ratios and tissue content. Appendix E-A-1 (available at www.neurology.org) contains a detailed explanation of the processing steps performed on the data, including a description of the automated WMH segmentation.
Statistical analyses.
All statistical calculations were performed with the SPSS program (SPSS Inc., Chicago, IL). To compare the automated and semiautomated WMH segmentations, we calculated the mean and SD of the difference between the two methods.20
To investigate the relationship between WMH volume and other parameters, we calculated a log transform of WMH volume, because the measured distribution of WMH volumes was heavily skewed toward low values. Differences in brain size were corrected for by dividing the WMH volume by the total brain volume. We calculated the partial correlation coefficient for log(WMH) vs median ADC, controlling for age, because both ADC21 and WMH volume1 have been reported to increase with age. We also calculated the partial correlation coefficients separately for each of the relationships between log(WMH) and ADC vs each of the metabolite ratios in the NAWM, again controlling for the patient’s age in each correlation.
Differences in metabolite ratios between gray and white matter and NAWM and WMH were calculated using paired t-tests, which were also used to investigate differences between median ADC in the NAWM and WMH.
Results.
Of the 60 people examined, 37 were women. The average age was 73 years (SD 5, range 64 to 84 years).
Comparing the automated and semiautomated segmentation techniques, the mean difference in WMH volume between the techniques was 0.53 mL (SD = 3.4 mL). Figure E-F-1 (available at www.neurology.org) shows a comparison of the automatic and manual threshold segmentation techniques as a graph of the difference between them plotted against the mean WMH volume.
Figure 1 shows a typical FLAIR image of a subject with moderate WMH. Figure 2 shows the segmented WMH image obtained from the automated segmentation.
Figure 1. Fluid-attenuated inversion recovery image of a patient with moderate white matter hyperintensities (WMH). Overlaid is a region of interest template generated by the automated WMH segmentation analysis. C_C = corpus callosum; L_Occip = left occipital lobe.
Figure 2. Segmented white matter hyperintensity image. Region of interest template again shown, for comparison. C_C = corpus callosum; L_Occip = left occipital lobe.
Table 1 shows the correlation between the log transformed total lesion volume (as measured by the independent observer) and the median diffusion coefficient in the NAWM.
Table 1. Correlation coefficients between metabolite concentration ratios, white matter volume, and median ADC of the normal-appearing white matter; age has been controlled for in the correlations
As shown in table 1 and figure 3, there is a highly significant positive correlation between the NAWM median ADC value and the total volume of WMH.
Figure 3. Median apparent diffusion coefficient (ADC) in the normal-appearing white matter plotted against the log transform of white matter hyperintensities (WMH) volume as a percent of whole brain volume.
The median ADC over all subjects (n = 57) in the WMH was 1,049 (SD 109) × 10−6 mm2/second, and in the NAWM 788 (SD 35); a paired t-test gave p < 0.0001, t = 22.9.
As can also be seen from table 1, the metabolite ratios NAA/Cre and NAA/Cho show a negative correlation with total WMH volume and with median ADC. The correlation between Cho/Cr and WMH volume was not significant. Table 2 shows the metabolite ratios in WMH and NAWM for those subjects who had spectroscopic voxels with WMH content greater than 20%.
Table 2. Comparison of metabolite ratios in WMH and NAWM for those subjects with voxels containing more than 20% WMH
Discussion.
The main results of this study were that increased ADC and decreased NAA in the NAWM were associated with increased volumes of WMH. The automated segmentation provided a good level of agreement with the user supervised segmentation. Although there was a discrepancy of ∼10 mL for two or three subjects, there was no bias evident. The level of agreement between the two techniques is comparable to that between two expert humans: using a semiautomated contouring technique similar to the one used here, an intraobserver variability (i.e., repeatability) of SD = 2.83 mL has been reported for outlining WMH in multiple sclerosis.22 This compares well with the SD of 3.4 mL we found comparing the fully automated technique with semiautomated contouring.
As expected, the ADC value in the regions of WMH was much higher than in NAWM. However, we also found a highly significant continuous increase of median ADC in the NAWM with total WMH volume. The positive correlation between ADC of the NAWM and the WMH volume indicates that tissue damage associated with WMH extends outside of the region as seen on MRI, and that subtle damage occurs throughout the WM. A comparison of MRI and histopathology of postmortem brains7 found that the abnormal region seen on the pathology samples was greater in extent (by 50%) than that seen on the MRI. The region that appeared abnormal on pathology but not MRI had only subtle changes such as a decrease in axonal density. This is reflected in our study—we see a greater diffusion coefficient in the NAWM of those people with more WMH, but the increase is smaller than the increase in ADC in the WMH themselves.
A potential source of error in this analysis is the inclusion of areas of WMH in the segmented NAWM. This might have come about owing to a number of causes, including differences in resolution between the DWI and the segmented images, poor segmentation of the NAWM, or small WMH not obvious on the FLAIR or T2 scans.
Differences in resolution are unlikely to have been a problem, because for both the DWI and the FLAIR and T2 sequence, the limiting factor in the resolution is the slice thickness, which is 5 mm.
We visually inspected the segmentation of T2 images that was used to define areas of NAWM on the DWI. Abnormal appearing regions seemed to be excluded from the NAWM region. We used the median of the ADC in order to reduce the influence of any regions of WMH remaining in the NAWM area, because the median is relatively insensitive to extreme values. Thus we are confident that the relationship between median ADC and log(WMH volume) represents subtle changes occurring in the bulk of the NAWM associated with WMH.
In our analysis of the spectroscopy data, to compensate for the large spectroscopic voxel size, we have resampled the segmented images to the same resolution as the spectroscopic images, and performed a regression of metabolite ratios to voxel tissue composition in order to determine mean metabolite ratios for the different tissue types. In line with other reports,14,23,24⇓⇓ we found NAA/Cr and Cho/Cr increased and NAA/Cho decreased in white matter relative to gray matter. We also found a decrease in NAA/Cr and NAA/Cho, but not Cho/Cr, in the NAWM associated with increased WMH. NAA is thought to reside primarily in neurons and their axons, and a decrease in NAA concentration reflects a reduced density or metabolic activity of neurons. Our finding of reduced NAA/Cho and NAA/Cr would thus support the data from the ADC images in suggesting a decreased neuronal density, or reduction of myelination in the NAWM associated with the presence of WMH.
Comparing the metabolite ratios in WMH with NAWM for those individuals with spectroscopic voxels containing more than 20% WMH, we found an increase in Cho/Cr, a decrease in NAA/Cho, and no difference in NAA/Cr. These values should be treated with caution, owing to the small number of subjects and the relatively low concentration of WMH in the spectroscopic voxels. However, a possible explanation for the data could be that large areas of WMH reflect a generalized poor neuronal metabolism (reflected in low NAA levels throughout the brain) with demyelination occurring in the WMH, causing a localized increase in Cho, as membrane phosphocholines are broken down into MR visible glycerophosphorocholine. There are not many reports in the literature of metabolite values in WMH, although one25 found change in metabolite ratios suggesting decreased Cho and unchanged NAA in the WMH, and another found decreased NAA but not Cho or Cr in WMH vs NAWM.14 Further MRS studies are needed to clarify which changes occur in WMH.
Measurement of ADC may be more sensitive to pathologic changes in the WM than FLAIR images, and the ADC may be a useful tool for following disease progression or response to treatment—longitudinal changes in ADC over a 2-year period with no change in visual rating of WMH have been reported.26 Data regarding localized changes in ADC and spectroscopic metabolites may be helpful in understanding the contribution of regional WM damage to different pathologies such as depression, and the different types of dementia–localized changes in ADC in a group of people with Alzheimer disease with minimal WMH have been observed.27 Further studies are needed to investigate whether diffusion measurements provide clinically useful information about disease progression.
Acknowledgments
T.M. was supported by grants from Fundação de Amparo a Pesquisa do Estado de São Paulo (FAPESP), grant number 99/04928–9, and Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES), grant number BEX0096/01–6.
Acknowledgment
The authors thank Philip English for his radiographic expertise.
Footnotes
-
Additional material related to this article can be found on the Neurology Web site. Go to www.neurology.org and scroll down the Table of Contents for the October 14 issue to find the title link for this article.
- Received January 14, 2003.
- Accepted June 18, 2003.
References
Disputes & Debates: Rapid online correspondence
REQUIREMENTS
If you are uploading a letter concerning an article:
You must have updated your disclosures within six months: http://submit.neurology.org
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.