Combining MRI and CSF measures for classification of Alzheimer's disease and prediction of mild cognitive impairment conversion☆
Highlights
► Combining MRI and CSF is more powerful than using either measure separately. ► An accuracy of 91.8% was achieved for discriminating AD from controls. ► The combined model showed potential for predicting conversion from MCI to AD.
Introduction
Alzheimer's disease (AD) is one of the most common forms of neurodegenerative disorders characterized by a gradual loss of cognitive functions such as episodic memory. The disease is related to pathological amyloid depositions and hyperphosphorylation of structural proteins which leads to progressive loss of function, metabolic alterations and structural changes in the brain.
A revision of the National Institute of Neurological and Communicative Disorders and Stroke (NINCDS) and the Alzheimer's disease and related Disorders Association (ADRDA) criterion (McKhann et al., 1984) for the diagnosis of AD has been suggested. The new research criterion is still centered on a clinical core of early and significant episodic memory impairment, but also includes at least one abnormal biomarker among magnetic resonance imaging (MRI), positron emission tomography (PET) and cerebrospinal fluid (CSF) (Dubois et al., 2007). The new suggested diagnostic criterion also utilizes biomarkers (McKhann et al., 2011). Since evidence for use of these biomarkers is growing, it strengthens their role in the diagnosis of AD. These biomarkers reflect different yet connected aspects of the disease, with MRI measuring early structural changes in the medial temporal lobe, particularly entorhinal cortex and hippocampus, fluorodeoxyglucose (FDG)-PET measuring glucose metabolism, amyloid PET measuring the build-up of amyloid in tissue and CSF biomarkers reflecting changes in levels of Aβ, tau proteins and ratios of the two. The utilization of the three biomarkers varies from center to center and depends on factors including local availability, cost and historical patterns of usage.
An effective combination of different biomarkers may prove to be more useful than using single biomarkers and could be a potent biomarker in itself for disease diagnosis and prediction of progression from MCI to AD. Therefore, we decided to investigate the impact of combining MRI and CSF measures for the classification of AD and to predict future conversion from the prodromal stage of the disease, mild cognitive impairment (MCI).
The combination of MRI and CSF (Ewers et al., In press, Kohannim et al., 2010, Nettiksimmons et al., 2010) or MRI, CSF and FDG-PET (Kohannim et al., 2010, Walhovd et al., 2010, Zhang et al., 2011) has previously been investigated, but few studies have previously utilized baseline MRI and CSF measures for classification of individual subjects (Ewers et al., In press, Kohannim et al., 2010, Walhovd et al., 2010). No previous studies have systematically and extensively examined the prediction of MCI conversion at multiple future time points.
We have utilized a novel technique (OPLS) with one of the largest sample sizes to date from the ADNI cohort to combine the two measures for individual classification. We aimed to compare the ability of the combination of MRI and CSF, MRI alone and CSF alone, to 1) distinguish subjects with AD from healthy controls at baseline, 2) distinguish subjects with MCI from healthy controls at baseline and 3) use baseline MRI and CSF data to predict conversion from MCI to AD, using the follow-up diagnosis at multiple future time points.
Section snippets
Data
Data was downloaded from the Alzheimer's disease Neuroimaging Initiative (ADNI) database (www.loni.ucla.edu/ADNI, PI Michael M. Weiner). ADNI was launched in 2003 by the National Institute on Aging (NIA), the National Institute of Biomedical Imaging and Bioengineering (NIBIB), the Food and Drug Administration (FDA), private pharmaceutical companies and non-profit organizations, as a $60 million, 5-year public–private partnership. The primary goal of ADNI has been to test whether serial MRI, PET
AD and MCI classification
Table 3 shows the classification accuracy, sensitivity, specificity, positive and negative likelihood ratios (LR + and LR −), positive predictive value (PPV), negative predictive value (NPV), AUC and Q2(Y) of the different models.
For the AD vs. CTL model, combining the MRI measures with the CSF measures resulted in a classification accuracy of 91.8% (sensitivity = 88.5%, specificity = 94.6%, PPV = 93.4%, NPV = 90.5% and AUC = 0.958) compared to 87.0% for MRI only (sensitivity = 83.3%, specificity = 90.1%, PPV =
Discussion
The use of biomarkers for diagnosis and prognosis in AD and MCI is of great importance. At least one abnormal biomarker among MRI, PET and CSF should be included alongside a clinical core of early and significant episodic memory impairment for a diagnosis of AD according to the new criteria (Dubois et al., 2007). A combination of biomarkers may prove to be more useful than single biomarkers for individual classification/prediction of subjects.
MRI and PET generate large volumes of data which can
Conclusion
Different biomarkers provide complementary information, which have been shown to be useful in AD and MCI diagnoses when used together (Apostolova et al., 2010, Fjell et al., 2010, Landau et al., 2010, Zhang et al., 2011). We show that the combination of MRI and CSF using OPLS as a tool more accurately classifies AD, MCI and CTL subjects at baseline compared to using either biomarker separately. At the moment there is no universally agreed gold standard for prediction accuracy, though in the
Acknowledgment
Data collection and sharing for this project was funded by the Alzheimer's disease Neuroimaging Initiative (ADNI) (National Institutes of Health Grant U01 AG024904). ADNI is funded by the National Institute on Aging, the National Institute of Biomedical Imaging and Bioengineering, and through generous contributions from the following: Abbott, AstraZeneca AB, Bayer Schering Pharma AG, Bristol-Myers Squibb, Eisai Global Clinical Development, Elan Corporation, Genentech, GE Healthcare,
References (59)
- et al.
3D PIB and CSF biomarker associations with hippocampal atrophy in ADNI subjects
Neurobiol. Aging
(2010) - et al.
Automatic classification of patients with Alzheimer's disease from structural MRI: a comparison of ten methods using the ADNI database
Neuroimage
(2011) - et al.
Cortical surface-based analysis. I. Segmentation and surface reconstruction
Neuroimage
(1999) - et al.
An automated labeling system for subdividing the human cerebral cortex on MRI scans into gyral based regions of interest
Neuroimage
(2006) - et al.
Research criteria for the diagnosis of Alzheimer's disease: revising the NINCDS–ADRDA criteria
Lancet Neurol.
(2007) - et al.
Structural and functional biomarkers of prodromal Alzheimer's disease: a high-dimensional pattern classification study
Neuroimage
(2008) - et al.
Whole brain segmentation: automated labeling of neuroanatomical structures in the human brain
Neuron
(2002) - et al.
Spatially augmented LPboosting for AD classification with evaluations on the ADNI dataset
Neuroimage
(2009) - et al.
Boosting power for clinical trials using classifiers based on multiple biomarkers
Neurobiol. Aging
(2010) - et al.
Analysis of regional MRI volumes and thicknesses as predictors of conversion from mild cognitive impairment to Alzheimer's disease
Neurobiol. Aging
(2010)
Combination analysis of neuropsychological tests and structural MRI measures in differentiating AD, MCI and control groups — the AddNeuroMed study
Neurobiol. Aging
Partial least squares analysis of neuroimaging data: applications and advances
Neuroimage
The diagnosis of dementia due to Alzheimer's disease: recommendations from the National Institute on Aging-Alzheimer's Association workgroups on diagnostic guidelines for Alzheimer's disease
Alzheimers Dement.
Subtypes based on cerebrospinal fluid and magnetic resonance imaging markers in normal elderly predict cognitive decline
Neurobiol. Aging
How golden is the gold standard of neuropathology in dementia?
Alzheimers Dement.
A hybrid approach to the skull stripping problem in MRI
Neuroimage
Consistent neuroanatomical age-related volume differences across multiple samples
Neurobiol. Aging
AddNeuroMed and ADNI: similar patterns of Alzheimer's atrophy and automated MRI classification accuracy in Europe and North America
Neuroimage
Multivariate analysis of MRI data for Alzheimer's disease, mild cognitive impairment and healthy controls
Neuroimage
PLS-regression: a basic tool of chemometrics
Chemometr. Intell. Lab. Syst.
Some recent developments in PLS modeling
Chemometr. Intell. Lab. Syst.
Multimodal classification of Alzheimer's disease and mild cognitive impairment
Neuroimage
Data integration in plant biology: the O2PLS method for combined modeling of transcript and metabolite data
Plant J.
Identification of conversion from mild cognitive impairment to Alzheimer's disease using multivariate predictors
PLoS One
Improved localization of cortical activity by combining EEG and MEG with MRI cortical surface reconstruction: a linear approach
J. Cogn. Neurosci.
Multi- and Megavariate Data Analysis (Part I — Basics and Principals and Applications)
Measuring the thickness of the human cerebral cortex from magnetic resonance images
Proc. Natl. Acad. Sci. U. S. A.
Cited by (271)
Brain structural indicators of β-amyloid neuropathology
2024, Neurobiology of AgingKNN weighted reduced universum twin SVM for class imbalance learning
2022, Knowledge-Based SystemsCortical folding correlates to aging and Alzheimer’s Disease’s cognitive and CSF biomarkers
2024, Scientific ReportsArtificial Intelligence Techniques for the effective diagnosis of Alzheimer’s Disease: A Review
2024, Multimedia Tools and ApplicationsPlasma Biomarkers as Predictors of Progression to Dementia in Individuals with Mild Cognitive Impairment
2024, Journal of Alzheimer's Disease
- ☆
Data used in preparation of this article were obtained from the Alzheimer's disease Neuroimaging Initiative (ADNI) database (www.loni.ucla.edu/ADNI). As such, the investigators within the ADNI contributed to the design and implementation of ADNI and/or provided data but did not participate in analysis or writing of this report. ADNI investigators include (complete listing available at http://adni.loni.ucla.edu/wp-content/uploads/how_to_apply/ADNI_Authorship_List.pdf).
- 1
For the Alzheimer's Disease Neuroimaging Initiative.