Brain Atrophy in MS
Brain volume measurements are not used in routine clinical practice, but it is a very attractive MRI metric in MS because it is clinically relevant, robust and could offer an in vivo measure of neurodegeneration. In general, brain volume in MS patients decreases by 0.7–1.0% per year,[15] which is approximately three-times the rate of normal controls. It can be seen as early as CIS.[16,17] It has been shown to correlate with cognitive impairment,[18] EDSS[19] and quality of life.[20]
Overall, brain atrophy (pathological loss of tissue) represents the net effect of tissue damage in MS. The exact mechanism by which brain atrophy occurs remains incompletely understood. Axonal loss within white matter lesions probably contributes to brain atrophy by two mechanisms: the amount of tissue lost within the lesions themselves, and Wallerian degeneration in pathways related to the lesions.[15,21] Some studies have found a tight correlation between white matter lesion burden and brain volume;[22] however, other studies suggest that this correlation is only modest, suggesting that separate pathologic processes are also contributing to the loss of brain volume.[21] Gray matter lesions occur but may not be directly associated with cortical thinning; one pathologic study of 22 MS patients and 17 controls failed to find a significant correlation between gray matter lesions and cortical thinning.[23] Iron deposition may also potentially contribute to brain atrophy.[24]
It is important to realize that both disease activity and current therapies can influence brain volume. Brain volume can transiently increase with the inflammation and edema that are associated with new lesion formation, whereas corticosteroids or newly initiated immunomodulatory therapy can transiently decrease brain volume. This transient decrease in brain volume has been called 'pseudoatrophy'. For example, disease-modifying treatments are associated with rapid reduction of whole brain volume through unclear mechanisms;[25] this is thought to be due to the suppression of inflammation in the CNS and resolution of associated edema.
Gray versus White Matter Atrophy
Brain atrophy seems to be a widespread process that is present in all stages of MS. Although both gray and white matter undergo atrophy, it has been suggested that loss of gray matter is a more sensitive marker of the neurodegenerative process in MS than whole brain atrophy.[26] Futhermore, some in vivo data have suggested that the rate of gray matter volume loss accelerates as the disease progresses.[26,27] For example, in a prospective 4-year study of 87 patients, Fisher et al. found that CIS patients who converted to RRMS had a 3.4-fold increase in the rate of atrophy compared with normal controls, whereas RRMS patients converting to SPMS showed a 14-fold increase.[26]
However, it still remains unclear whether gray matter atrophy accelerates compared with white matter as MS progresses. It may be that, in later stages of MS when the white matter has sustained extensive damage, the percentage change in white matter volume per year is lower than that of gray matter because there is not as much white matter left to lose. Nonetheless, gray matter volume appears to be a very clinically meaningful MRI metric; a recent multicenter study of 977 patients at all stages of MS found gray matter volume to be a better predictor of disability and cognitive impairment, as measured by the EDSS and paced auditory serial addition test, respectively, than white matter volume or white matter lesions.[28] Similarly, gray matter volume correlated better with quality-of-life measures than white matter or whole brain volume.[20]
Regional Brain Atrophy
Different regions of the brain are affected at different phases of the disease. One study using voxel-based morphometry found early loss of gray matter volume involving the thalami and hypothalami in patients with CIS at presentation.[17] A novel use of tensor-based morphometry in MS also found atrophy of deep gray matter structures at modest levels of EDSS.[29] Another study measured cortical thickness using FreeSurfer and found that, of the CIS patients who converted to RRMS during the 4-year study, cortical thinning could be seen in the precentral gyrus, superior frontal gyrus, thalamus and putamen.[30] In RRMS, there is atrophy in the frontotemporal neocortex, whereas in SPMS, atrophy is present in virtually all cortical regions, as well as deep gray matter structures, the cerebellum and brainstem.[31] Attempts have been made to correlate regional atrophy with specific symptoms. Atrophy of the hippocampus has been associated with deficits in memory encoding and retrieval,[32] as well as depression.[33] Atrophy of the parietal lobe has been associated with fatigue.[34] Table 2 summarizes several recent studies showing relationships between brain atrophy and clinical outcomes.
Methods to Measure Brain Volume Loss in vivo
Reviewing technical aspects of brain volume is beyond the scope of this work but several methods are worth mentioning briefly for the benefit of the practicing clinican. Although no brain atrophy measures have been accepted in clinical practice yet, several methods to measure whole brain atrophy do exist, including manual, semiautomated and automated methods. Manual methods include measurements of third ventricular width, lateral ventricular width, brain width, corpus callosum area and bicaudate ratio.[21] These methods are relatively simple and do not require sophisticated software, but they are limited by lack of efficiency, poor reproducibility and lack of precision.[15]
The advantage of automated systems is their increased efficiency, sensitivity, precision and reproducibility. The automated methods for measuring whole brain volume can generally be placed into two categories depending on their reliance on registration or segmentation. Registration refers to a method of measurement that requires two scans separated in time or from a template that are positionally matched, or registered, so that a measure of volume loss or percentage loss can be derived. Registration-based methods include the brain boundary shift integral, structural image evaluation using normalization of atrophy, statistical parametric mapping and voxel-based morphometry.[35]
Segmentation-based methods include the commonly used brain parenchymal fraction (BPF) and its variants, index of brain atrophy and whole brain ratio. The BPF uses an algorithm to detect the outer surface of the brain and the amount of brain tissue within that outer surface, essentially subtracting out (segmenting out) the ventricular cerebrospinal fluid volumes. BPF has been shown to correlate moderately well with disability.[36] Other segmentation-based methods include fuzzy connectedness,[37] the Alfano method,[38] structural image evaluation using normalization of atrophy for cross-sectional measurement[39] and semiautomatic brain region extraction.[40]
Another more recent method using both registration and segmentation is FreeSurfer, which segments out the cortex from the brain by inflating the folded cortical surface. This allows for accurate measures of cortical thickness.[41,42] Methods measuring cortical thinning are likely to dominate the MS MRI research on tissue loss over the next few years. Such studies are likely to provide insights into disease pathogenesis, progression and possible therapeutic targets.
The ideal method for measuring brain volume loss in MS has not been determined. The issue is complicated by the lack of a gold standard, and by lack of standardization of methods for image acquisition, analysis and postprocessing.[35] Well-designed comparative studies will be necessary to help answer some of these questions.
Expert Rev Neurother. 2012;12(3):323-333. © 2012 Expert Reviews Ltd.
Comments