Combined Structural MR and Diffusion Tensor Imaging Classify the Presence of Alzheimer’s Disease With the Same Performance as MR Combined With Amyloid Positron Emission Tomography: A Data Integration Approach

Background: In recent years, classification frameworks using imaging data have shown that multimodal classification methods perform favorably over the use of a single imaging modality for the diagnosis of Alzheimer’s Disease. The currently used clinical approach often emphasizes the use of qualitative MRI and/or PET data for clinical diagnosis. Based on the hypothesis that classification of isolated imaging modalities is not predictive of their respective value in combined approaches, we investigate whether the combination of T1 Weighted MRI and diffusion tensor imaging (DTI) can yield an equivalent performance as the combination of quantitative structural MRI (sMRI) with amyloid-PET.Methods: We parcellated the brain into regions of interest (ROI) following different anatomical labeling atlases. For each region of interest different metrics were extracted from the different imaging modalities (sMRI, PiB-PET, and DTI) to be used as features. Thereafter, the feature sets were reduced using an embedded-based feature selection method. The final reduced sets were then used as input in support vector machine (SVM) classifiers. Three different base classifiers were created, one for each imaging modality, and validated using internal (n = 41) and external data from the ADNI initiative (n = 330 for sMRI, n = 148 for DTI and n = 55 for PiB-PET) sources. Finally, the classifiers were ensembled using a weighted method in order to evaluate the performance of different combinations.Results: For the base classifiers the following performance levels were found: sMRI-based classifier (accuracy, 92%; specificity, 97% and sensitivity, 87%), PiB-PET (accuracy, 91%; specificity, 89%; and sensitivity, 92%) and the lowest performance was attained with DTI (accuracy, 80%; specificity, 76%; and sensitivity, 82%). From the multimodal approaches, when integrating two modalities, the following results were observed: sMRI+PiB-PET (accuracy, 98%; specificity, 98%; and sensitivity, 99%), sMRI+DTI (accuracy, 97%; specificity, 99%; and sensitivity, 94%) and PiB-PET+DTI (accuracy, 91%; specificity, 90%; and sensitivity, 93%). Finally, the combination of all imaging modalities yielded an accuracy of 98%, specificity of 97% and sensitivity of 99%.Conclusion: Although DTI in isolation shows relatively poor performance, when combined with structural MR, it showed a surprising classification performance which was comparable to MR combined with amyloid PET. These results are consistent with the notion that white matter changes are also important in Alzheimer’s Disease.

Classification of 18F-Flutemetamol Scans in Cognitively Normal Older Adults Using Machine Learning Trained with Neuropathology as Groundtruth

PURPOSE: End-of-life studies have validated the binary visual reads of 18F-labeled amyloid PET tracers as an accurate tool for the presence or absence of increased neuritic amyloid plaque density. In this study, the performance of a support vector machine (SVM) based classifier will be tested against pathological groundtruths and its performance determined in cognitively healthy older adults.METHODS: We applied SVM with a linear kernel to an 18F-Flutemetamol end-of-life dataset to determine the regions with the highest feature weights in a data-driven manner and to compare between two different pathological groundtruths: based on neuritic amyloid plaque density or on amyloid phases, respectively. We also trained and tested classifiers based on the 10% voxels with the highest feature weights for each of the two neuropathological groundtruths. Next, we tested the classifiers’ diagnostic performance in the asymptomatic Alzheimer’s disease (AD) phase, a phase of interest for future drug development, in an independent dataset of cognitively intact older adults, the Flemish Prevent AD Cohort-KU Leuven (F-PACK). A regression analysis was conducted between the Centiloid (CL) value in a composite volume of interest (VOI), as index for amyloid load, and the distance to the hyperplane for each of the two classifiers, based on the two pathological groundtruths. A Receiver-Operating-Characteristic analysis was also performed to determine the CL threshold that optimally discriminates between neuritic amyloid plaque positivity versus negativity, or amyloid phase positivity versus negativity, within F-PACK.RESULTS: The classifiers yielded adequate sensitivity and specificity within the end-of-life dataset (neuritic amyloid plaque density classifier: specificity of 90.2% and sensitivity of 83.7%; amyloid phase classifier: specificity of 98.4% and sensitivity of 84.0%). The regions with the highest feature weights corresponded to precuneus, caudate, anteromedial prefrontal, and also posterior inferior temporal and inferior parietal cortex. In the cognitively normal cohort, the correlation coefficient between CL and distance to the hyperplane was -0.66 for the classifier trained with neuritic amyloid plaque density, and -0.88 for the classifier trained with amyloid phases. This difference was significant. The optimal CL cut-off for discriminating positive versus negative scans was CL = 48-51 for the different classifiers (area under the curve (AUC) = 99.9%), except for the classifier trained with amyloid phases and based on the 10% voxels with highest feature weights. There the cut-off was CL = 26 (AUC = 99.5%).DISCUSSION: A neuropathologically validated classifier applied in cognitively normal older adults reveals that amyloid PET values (Centiloids) correlate best with amyloid phases. A CL cut-off of 26 reliably discriminated between amyloid phase 0-2 and 3-5 while only a CL around 50 discriminated between no or sparse and moderate to severe neuritic amyloid plaque density.

Vascular burden and cognition: Mediating roles of neurodegeneration and amyloid-PET

INTRODUCTION: It remains unclear to which extent vascular burden promotes neurodegeneration and cognitive dysfunction in a cohort spanning low-to-severe small vessel disease (SVD) and amyloid-beta pathology. METHODS: In 120 subjects, we investigated 1) whether vascular burden, quantified as total or lobar white matter hyperintensity (WMH) volumes, is associated with different cognitive domains; and 2) whether the total WMH effect on cognition is mediated by amyloid (18F-AV45-PET), glucose metabolism (18F-FDG-PET), and/or cortical atrophy. RESULTS: Increased total WMH volume was associated with poorer performance in all cognitive domains tested, with the strongest effects observed for semantic fluency. These relationships were mediated mainly through cortical atrophy, particularly in the temporal lobe, and to a lesser extent through amyloid and metabolism. WMH volumes differentially impacted cognition depending on lobar location and amyloid status. DISCUSSION: Our study suggests mainly an amyloid-dependent pathway in which vascular burden affects cognitive impairment through temporal lobe atrophy.

Partial Volume Correction Increases the Sensitivity of 18F-Florbetapir-Positron Emission Tomography for the Detection of Early Stage Amyloidosis

Purpose: To test whether correcting for unspecific signal from the cerebral white matter increases the sensitivity of amyloid-PET for early stages of cerebral amyloidosis.Methods: We analyzed 18F-Florbetapir-PET and cerebrospinal fluid (CSF) Aβ42 data from 600 older individuals enrolled in the Alzheimer’s Disease Neuroimaging Initiative (ADNI), including people with normal cognition, mild cognitive impairment (MCI), and Alzheimer’s disease (AD) dementia. We determined whether three compartmental partial volume correction (PVC-3), explicitly modeling signal spill-in from white matter, significantly improved the association of CSF Aβ42 levels with global 18F-Florbetapir-PET values compared with standard processing without PVC (non-PVC) and a widely used two-compartmental PVC method (PVC-2). In additional voxel-wise analyses, we determined the sensitivity of PVC-3 compared with non-PVC and PVC-2 for detecting early regional amyloid build-up as modeled by decreasing CSF Aβ42 levels. For replication, we included an independent sample of 43 older individuals with subjective memory complaints from the INveStIGation of AlzHeimer’s PredicTors cohort (INSIGHT-preAD study).Results: In the ADNI sample, PVC-3 18F-Florbetapir-PET values normalized to whole cerebellum signal showed significantly stronger associations with CSF Aβ42 levels than non-PVC or PVC-2, particularly in the lower range of amyloid levels. These effects were replicated in the INSIGHT-preAD sample. PVC-3 18F-Florbetapir-PET data detected regional amyloid build-up already at higher (less abnormal) CSF Aβ42 levels than non-PVC or PVC-2 data.Conclusion: A PVC approach that explicitly models unspecific white matter binding improves the sensitivity of amyloid-PET for identifying the earliest stages of cerebral amyloid pathology which has implications for future primary prevention trials.

Lower Aβ-PET signal in white matter lesions relates to higher extracellular free water in mixed small vessel disease and Alzheimer's pathology

Background and Objectives: Positron emission tomography (PET) with amyloid-ligands recently gained interest for evaluating white matter (WM) pathology. Here, we investigated the neurobiological underpinnings of the amyloid-PET signal in the WM using free water (FW) diffusion MRI in a mixed cohort of small vessel disease (SVD) and Alzheimer's disease (AD) pathology. Methods: We included data from two large cohort imaging studies, including MITNEC-C6 with moderate-to-severe white matter hyperintensity (WMH) burden and ADNI-2 with mild-to-moderate WMH burden, covering the spectrum of cognitively normal to dementia. Subjects underwent diffusion MRI, 18F-AV45 amyloid-PET, and cognitive testing. We calculated WM diffusion metrics including fractional anisotropy (FA) and mean diffusivity (MD). In addition, we applied a bi-tensor diffusion MRI model that differentiates between extracellular (FW fraction) and tissue-specific (FW-adjusted FA) compartments of the WM. We tested associations of all diffusion metrics with 18F-AV45 SUVR in regions of WMH vs. normal-appearing WM (NAWM), and with cognition. We further performed partial-least-square analysis to investigate how the diffusion metrics as well as age, sex, education, WMH volume, and cortical 18F-AV45 SUVR covary with WM 18F-AV45 SUVR. Results: 18F-AV45 SUVR was significantly lower in regions of WMH compared to NAWM (t=25.08, P<0.0001, n=115 subjects). Within WMH, lower 18F-AV45 SUVR was associated with higher FW (β=-0.36±0.13, P=0.005) and lower FA (β=+0.24±0.12, P=0.046), but not with the tissue-specific metric FW-adjusted FA. Partial-least-square analysis further confirmed that FW had the most influence on 18F-AV45 SUVR in regions of WMH. In contrast, FW-adjusted FA had more influence on 18F-AV45 SUVR in NAWM. Last, correlation with cognitive impairment was higher for FW than FW-adjusted FA, both in regions of WMH (MMSE: βFW=-0.40±0.13, P=0.003; βFW-adjustedFA=0.14±0.09, P=0.11) and NAWM (MMSE: βFW=-0.30±0.11, P=0.01; βFW-adjustedFA=0.21±0.09, P=0.02. Conclusion: In a cohort representative of the more common AD population, reduced amyloid-PET uptake in WM lesions may largely reflect the appearance of extracellular FW, while changes in NAWM may associate with tissue-specific damage. Ultimately, our findings may support FW and amyloid-PET as markers of WM abnormalities in SVD and AD.

The heritability of amyloid burden in older adults: the Older Australian Twins Study

ObjectiveTo determine the proportional genetic contribution to the variability of cerebral β-amyloid load in older adults using the classic twin design.MethodsParticipants (n=206) comprising 61 monozygotic (MZ) twin pairs (68 (55.74%) females; mean age (SD): 71.98 (6.43) years), and 42 dizygotic (DZ) twin pairs (56 (66.67%) females; mean age: 71.14 (5.15) years) were drawn from the Older Australian Twins Study. Participants underwent detailed clinical and neuropsychological evaluations, as well as MRI, diffusion tensor imaging (DTI) and amyloid PET scans. Fifty-eight participants (17 MZ pairs, 12 DZ pairs) had PET scans with 11Carbon-Pittsburgh Compound B, and 148 participants (44 MZ pairs, 30 DZ pairs) with 18Fluorine-NAV4694. Cortical amyloid burden was quantified using the centiloid scale globally, as well as the standardised uptake value ratio (SUVR) globally and in specific brain regions. Small vessel disease (SVD) was quantified using total white matter hyperintensity volume on MRI, and peak width of skeletonised mean diffusivity on DTI. Heritability (h2) and genetic correlations were measured with structural equation modelling under the best fit model, controlling for age, sex, tracer and scanner.ResultsThe heritability of global amyloid burden was moderate (0.41 using SUVR; 0.52 using the centiloid scale) and ranged from 0.20 to 0.54 across different brain regions. There were no significant genetic or environmental correlations between global amyloid burden and markers of SVD.ConclusionAmyloid deposition, the hallmark early feature of Alzheimer’s disease, is under moderate genetic influence, suggesting a major environmental contribution that may be amenable to intervention.

Validation of Plasma Amyloid-β 42/40 for Detecting Alzheimer Disease Amyloid Plaques

Objective:To determine the diagnostic accuracy of a plasma Aβ42/Aβ40 assay in classifying amyloid PET status across global research studies using samples collected by multiple centers that utilize different blood collection and processing protocols.Methods:Plasma samples (n=465) were obtained from three large Alzheimer’s Disease (AD) research cohorts in the US (n=182), Australia (n=183), and Sweden (n=100). Plasma Aβ42/Aβ40 was measured by a high precision immunoprecipitation mass spectrometry (IPMS) assay and compared to the reference standards of amyloid PET and CSF Aβ42/Aβ40.Results:In the combined cohort of 465 participants, plasma Aβ42/Aβ40 had good concordance with amyloid PET status (Receiver Operating Characteristic Area Under the Curve [AUC] of 0.84, 95% confidence interval [CI] 0.80-0.87); concordance improved with the inclusion of APOE ε4 status (AUC 0.88, 95% CI 0.85-0.91). The AUC of plasma Aβ42/Aβ40 with CSF amyloid status was 0.85 (95% CI 0.78-0.91) and improved to 0.93 (95% CI 0.89-0.97) with APOE ε4 status. These findings were consistent across the three cohorts, despite differences in protocols. Further, the assay performed similarly in both cognitively unimpaired and impaired individuals.Conclusions:Plasma Aβ42/Aβ40 is a robust measure for detecting amyloid plaques and can be utilized to aid in the diagnosis of AD, identify those at risk for future dementia due to AD, and improve the diversity of populations enrolled in AD research and clinical trials.Classification of Evidence:This study provides Class II evidence that plasma Aβ42/Aβ40, as measured by a high precision IPMS assay, accurately diagnoses brain amyloidosis in both cognitively unimpaired and impaired research participants.

Unravelling the Association Between Amyloid-PET and Cerebrospinal Fluid Biomarkers in the Alzheimer’s Disease Spectrum: Who Really Deserves an A+?

Background: Association between cerebrospinal fluid (CSF)-amyloid-β (Aβ)42 and amyloid-PET measures is inconstant across the Alzheimer’s disease (AD) spectrum. However, they are considered interchangeable, along with Aβ 42/40 ratio, for defining ‘Alzheimer’s Disease pathologic change’ (A+). Objective: Herein, we further characterized the association between amyloid-PET and CSF biomarkers and tested their agreement in a cohort of AD spectrum patients. Methods: We include ed 23 patients who underwent amyloid-PET, MRI, and CSF analysis showing reduced levels of Aβ 42 within a 365-days interval. Thresholds used for dichotomization were: Aβ 42 < 640 pg/mL (Aβ 42+); pTau > 61 pg/mL (pTau+); and Aβ 42/40 < 0.069 (ADratio+). Amyloid-PET scans were visually assessed and processed by four pipelines (SPMCL, SPMAAL, FSGM, FSWC). Results: Different pipelines gave highly inter-correlated standardized uptake value ratios (SUVRs) (rho = 0.93–0.99). The most significant findings were: pTau positive correlation with SPMCL SUVR (rho = 0.56, p = 0.0063) and Aβ 42/40 negative correlation with SPMCL and SPMAAL SUVRs (rho = –0.56, p = 0.0058; rho = –0.52, p = 0.0117 respectively). No correlations between CSF-Aβ 42 and global SUVRs were observed. In subregion analysis, both pTau and Aβ 42/40 values significantly correlated with cingulate SUVRs from any pipeline (R2 = 0.55–0.59, p < 0.0083), with the strongest associations observed for the posterior/isthmus cingulate areas. However, only associations observed for Aβ 42/40 ratio were still significant in linear regression models. Moreover, combining pTau with Aβ 42 or using Aβ 42/40, instead of Aβ 42 alone, increased concordance with amyloid-PET status from 74% to 91% based on visual reads and from 78% to 96% based on Centiloids. Conclusion: We confirmed that, in the AD spectrum, amyloid-PET measures show a stronger association and a better agreement with CSF-Aβ 42/40 and secondarily pTau rather than Aβ 42 levels.