The relationship between whole brain volume and disability in multiple sclerosis: A comparison of normalized gray vs. white matter with misclassification correction
Introduction
Multiple sclerosis (MS) is a central nervous system (CNS) disease that classically involves multifocal white matter lesions (Bruck et al., 1997) but is also associated with a myriad of subtle gray matter abnormalities, such as hypometabolism (Bakshi et al., 1998), direct plaque formation (Bakshi et al., 2001, Kidd et al., 1999), decreased neuronal viability (Cifelli et al., 2002, Inglese et al., 2004), T2 hypointensity (Bakshi et al., 2002, Bermel et al., in press), reduced magnetization transfer ratios (Dehmeshki et al., 2003, Ge et al., 2001a), and subcortical volume loss (Bermel et al., 2003a, Cifelli et al., 2002). In addition, whole brain atrophy is common in MS (Miller et al., 2002, Zivadinov and Bakshi, 2004b) and can be progressive early in the disease course (Zivadinov and Bakshi, 2004a).
More recent MS brain atrophy studies have segmented brain parenchyma into gray and white matter, showing, with one exception (Ge et al., 2001b), that generalized gray matter atrophy occurs (Chard et al., 2002a, Dalton et al., 2004, Quarantelli et al., 2003, Sastre-Garriga et al., 2004) and affects both the cerebral cortex (Amato et al., 2004, De Stefano et al., 2003, Sailer et al., 2003) and subcortical nuclei (Bermel et al., 2003a, Cifelli et al., 2002). While whole brain decreases in white matter volume have been found in most studies (Chard et al., 2002a, Ge et al., 2001b, Sastre-Garriga et al., 2004), negative findings also have been reported (Dalton et al., 2004, Quarantelli et al., 2003).
An important challenge in segmenting gray and white matter is addressing how MS-related lesions are handled in volume calculations of whole brain gray and white matter. In most studies (Chard et al., 2002a, Ge et al., 2001b, Quarantelli et al., 2003, Sastre-Garriga et al., 2004), lesion volumes were added to segmented whole white matter volume as a means to correct for MS lesions. This MS lesion correction technique, while preferable to no correction at all, may still be insufficient, because it assumes that all segmentation failures occur only within white matter lesions. However, it is quite possible that MS-related disease pathology also may be responsible for tissue compartment misclassification in gray matter and in white matter areas surrounding lesions. In other words, the use of a simple correction for white matter lesions may underestimate the effects of MS-related brain pathology and does not directly assess the possible shortcomings of segmentation algorithms.
In the present study, we used Statistical Parametric Mapping (SPM99) (Ashburner and Friston, 1997, Ashburner and Friston, 2000), as in prior reports (Chard et al., 2002a, Dalton et al., 2004, Sastre-Garriga et al., 2004), to obtain estimates of gray matter, white matter, and cerebrospinal fluid (CSF). Unlike those studies, however, we also performed a segmentation misclassification analysis to determine if SPM99's segmentation algorithm was potentially vulnerable to MS-related brain lesions (due to alterations in MR signal intensities) and therefore led to misclassification of voxels as gray matter, white matter, or CSF. We quantified the amount of misclassification in each tissue compartment and used these data to correct SPM99-derived volumes of whole gray matter, white matter, and CSF in a sample of MS patients, thereby allowing us to examine the effects of MS-related brain lesions in estimates for each brain compartment. We also explored the sensitivity and validity of SPM99 in detecting whole gray/white atrophy in MS (after misclassification correction) and assessed the association between gray/white matter atrophy and established clinical and MRI lesion parameters.
Section snippets
Subjects
Multiple sclerosis (MS) patients (n = 41) and normal control (NC) (n = 18) subjects gave informed consent to participate in the study, which was approved by the local Institutional Review Board. Patients with MS were seen at a tertiary care university-affiliated comprehensive MS research and treatment center and have been described previously (Benedict et al., 2004b, Sanfilipo et al., 2004, Sharma et al., 2004). Physical disability was rated with the Expanded Disability Status Scale (EDSS) (
Demographic variable analyses
For all subjects (n = 59), partial correlations (adjusting for ICV) revealed that age was inversely related to cGMV (rp[56] = −0.52, P < 0.0001) and cBPV (rp[57] = −0.39, P = 0.003), but not to cWMV (rp[57] = −0.01, P = 0.94). Based on a one-way ANCOVA design (adjusting for ICV and age) for the entire sample, no sex differences were found with regard to cWMV (F[1,55] = 0.16, P = 0.69), cGMV (F[1,55] = 2.40, P = 0.13), and cBPV (F[1,55] = 1.57, P = 0.22).
MS–control differences
The one-way group ANCOVA results
Discussion
In the present study, we used SPM99 to segment T1-weighted 3D gradient echo MR images for the purpose of obtaining normalized volumes of whole gray and white matter in a sample of MS patients and age-/sex-matched normal controls. Our segmented compartments were corrected for tissue misclassification due to MS-related brain pathology (cGMV, cWMV, cCSFV, and cBPV). For MS patients, we also measured conventional lesions (T1LV and FLLV), central atrophy (TVW and BCR), and clinical status (EDSS,
Acknowledgments
This research was supported in part by an Alpha Omega Alpha Student Research Fellowship (M. Sanfilipo), a University at Buffalo School of Medicine and Biological Sciences Summer Research Fellowship (M. Sanfilipo), and by research grants from the National Institutes of Health (NIH-NINDS 1 K23 NS42379-01, R. Bakshi), National Multiple Sclerosis Society (RG 3258A2/1, BWG, R. Bakshi; RG 3574A1, R. Bakshi), and National Science Foundation (DBI-0234895, BWG, R. Bakshi). The authors thank Christopher
References (42)
- et al.
Multimodal image coregistration and partitioning—A unified framework
NeuroImage
(1997) - et al.
Voxel-based morphometry—The methods
NeuroImage
(2000) - et al.
A semiautomated measure of whole-brain atrophy in multiple sclerosis
J. Neurol. Sci.
(2003) - et al.
Statistical mapping analysis of lesion location and neurological disability in multiple sclerosis: application to 452 patient data sets
NeuroImage
(2003) - et al.
Brain tissue volume changes in relapsing–remitting multiple sclerosis: correlation with lesion load
NeuroImage
(2003) - et al.
Correction for intracranial volume in analysis of whole brain atrophy in multiple sclerosis: the proportion vs. residual method
NeuroImage
(2004) - et al.
Grey and white matter atrophy in early clinical stages of primary progressive multiple sclerosis
NeuroImage
(2004) - et al.
Neocortical volume decrease in relapsing–remitting MS patients with mild cognitive impairment
Neurology
(2004) - et al.
High-resolution fluorodeoxyglucose positron emission tomography shows both global and regional cerebral hypometabolism in multiple sclerosis
J. Neuroimaging
(1998) - et al.
Intraventricular CSF pulsation artifact on fast fluid-attenuated inversion-recovery MR images: analysis of 100 consecutive normal studies
Am. J. Neuroradiol.
(2000)
Fluid-attenuated inversion recovery magnetic resonance imaging detects cortical and juxtacortical multiple sclerosis lesions
Arch. Neurol.
T2 hypointensity in the deep gray matter of patients with multiple sclerosis: a quantitative magnetic resonance imaging study
Arch. Neurol.
The relationship of gray, white, and total brain parenchymal fraction with cognitive performance in multiple sclerosis (abstract)
Neurology
Prediction of neuropsychological impairment in multiple sclerosis: comparison of conventional magnetic resonance imaging measures of atrophy and lesion burden
Arch. Neurol.
Bicaudate ratio as a magnetic resonance imaging marker of brain atrophy in multiple sclerosis
Arch. Neurol.
Selective caudate atrophy in multiple sclerosis: a 3D MRI parcellation study
NeuroReport
Inflammatory central nervous system demyelination: correlation of magnetic resonance imaging findings with lesion pathology
Ann. Neurol.
Brain atrophy in clinically early relapsing–remitting multiple sclerosis
Brain
The reproducibility and sensitivity of brain tissue volume measurements derived from an SPM-based segmentation methodology
J. Magn. Reson. Imaging
Cited by (147)
Three Dimensional Brain Parameters of Multiple Sclerosis (MS) Patients
2023, Multiple Sclerosis and Related DisordersExploring the effect of glatiramer acetate on cerebral gray matter atrophy in multiple sclerosis
2023, Journal of the Neurological SciencesRacial differences in retinal neurodegeneration as a surrogate marker for cortical atrophy in multiple sclerosis
2019, Multiple Sclerosis and Related DisordersCitation Excerpt :MS causes focal lesions in white matter with diffuse neurodegeneration, however, GM atrophy is a poorly understood aspect of MS and to date this pathological finding has not been addressed with disease-modifying therapies (DMT) (Ontaneda and Fox, 2017). It has been reported that an association between GM atrophy and clinical disability exists, highlighting its importance as a prognostic imaging biomarker (Rocca et al., 2017; Pirko et al., 2007; Sanfilipo et al., 2005). Furthermore, GM atrophy correlates with various symptoms, such as fatigue, cognitive impairment, depression and eventually quality of life (Sanfilipo et al., 2006; Sbardella et al., 2013; Bakshi et al., 2000; Janardhan and Bakshi., 2000; Bernitsas et al., 2017).
Association of circulating anti-CD64 IgM levels with favourable long-term clinical outcomes in multiple sclerosis patients
2019, Journal of NeuroimmunologyLinear regression
2019, Machine Learning: Methods and Applications to Brain DisordersBrain imaging of locomotion in neurological conditions
2018, Neurophysiologie Clinique
- 1
Current address: Center for Neurological Imaging, Partners MS Center, Departments of Neurology and Radiology, Brigham and Women's Hospital, Harvard Medical School, 77 Avenue Louis Pasteur – HIM 730, Boston, MA 02115, USA.