Brain Imaging

Work Package 3

WP3 used novel imaging markers to study pathophysiological mechanisms common to both stroke and Alzheimer's disease, including novel PET ligands and ultra-high field MR imaging. The specific aims were to image the earliest molecular brain changes in stroke and Alzheimer's disease, study brain connectivity and perfusion changes as contributors to cognitive impairment in stroke and Alzheimer's disease and study the relationship of genetic, metabolic, and molecular features with downstream effects in stroke and AD.

The overall design of WP3 consisted of two stages: retrospective combination and analysis of existing patient data and prospective collection of independent samples at opposite ends of the disease spectrum: symptomatic cognitive decline (UNIGE) and patients who present a stoke (KCL).

The initial target population was 90, but ultimately, we were able to include 353 individuals. Of these, 24 individuals were scanned using 7T MRI.

Baseline features of the AD cohort


Cognitively unimpaired (N=70)

Mild Cognitive Impairment










Gender, female

35 (50%)

103 (52%)

40 (47%)












Amyloid positivity, availability

70 (100%)

198 (100%)

85 (100%)


Tau positivity, availability

40 (57%)

117 (59%)

36 (42%)


Structural and functional brain alterations

In this task, retrospective combination and analysis of structural and functional alterations in existing patients’ data was performed.

For participants with tractography automatic atlas-based probabilistic tractography were performed, yielding reconstructions of the 29 tracts. Microstructural measures were extracted for each tract to identify key white matter structures relating to cognitive decline.

In-depth assessment of Alzheimer pathology using amyloid PET

In this task, prospective data acquisition of novel cases is being performed to achieve in-depth assessment of molecular features (amyloid and tau via PET) in clinical samples that reflect contrasting balances of vascular, Alzheimer and combined pathology.

To realize the study the following actions were taken:

  • The neuropsychological assessment was harmonised
  • The MRI research protocol was harmonised and follow the ADNI+ sequence
  • The amyloid PET exam was harmonised. Two sets of images are acquired; a 5 minutes acquisition from 1 to 6 minutes post injection and a 15 minutes acquisition from 50 (+/- 5 min) minutes post injection.
  • Tau PET imaging protocols were set up.
  • A 7T MRI sequence to detect the changes in locus coeruleus (LC) was set up.

Use of a previously published sequence for 7T MRI led to unsatisfactory contrast. We proceeded with the implementation of an alternative sequence, based on results of a different study which was found to provide better.

We received full approval from all relevant ethical committees for the clinical work in Work Package 3.

Study participants were recruited prospectively from the Memory Clinic at the University Hospital of Geneva and from KCL. All participants underwent:

  • A clinical assessment
  • neuropsychological assessment
  • 3T MRI
  • amyloid PET

3T MRI and PET scans are conducted on clinical scanners, following standard operating and safety procedures that are commonly used at clinical setting.  A sub-set of participants undergoes tau PET.

The first patient was scanned in November 2017.

A further sub-set of participants also underwent 7T ultra high field MRI.

For the STRATEGIC sample, the recruitment target was 30 controls and 200 patients. 32 healthy controls and 179 post stroke patients were included. Participants had cognitive evaluation with a standard set of 6 tests at the first screening visit, 30-90 days after stroke. At that visit, patients were also considered for 3T MRI and all patients without contraindications to MRI were invited to have a research MRI. Those agreeing to MRI also underwent extended neuropsychological assessment on subsequent visits. At one year, participants were invited for a single follow up visit at which a subset of tests were repeated.

Based on the data collected in UNIGE, an AT(N) model has been proposed as a research framework to investigate the pathological bases (i.e., amyloidosis, neurofibrillary tangles and neurodegeneration) of Alzheimer’s Disease (AD).

Subjects performed an in-depth clinical/neuropsychological evaluation. PET biomarkers of amyloid and tau pathology and MRI measures of neurodegeneration were acquired. Subjects were classified as amyloid positive or negative (A+/A-) and neurodegeneration positive or negative (N+/N-) according to the presence/absence of amyloid burden and medial temporal atrophy. We compared tau deposition across AN subgroups in regions of interest representing the Braak pathologic stages, and we estimated the Braak staging for each subject.

The subjects who completed the clinical and imaging procedures were characterized by different disease severity. These groups showed significant differences in MTA and amyloid deposition with AD dementia patients showing a higher burden of these pathological phenomena.

As expected, the biomarkers positivity increases across disease stages for amyloid, tau as well as neurodegeneration. Regarding amyloid, the positivity goes from 27% in CU to 81% in the dementia group while CU has a lower proportion of tau positive individuals (10%) and almost the same proportion of positive patients in the dementia group (78%). Neurodegeneration results positive in 14% of CU individuals, 40% of MCI and 66% of dementia.

We confirmed a higher tau deposition in A+ subjects. However, a significant amount of tau pathology characterized a subgroup of A- subjects with and without neurodegeneration. These findings provide new insights in the debate on a possible definition of an amyloid-independent tauopathy.

Moreover, we have assessed the association between age and biomarkers. We observed a positive correlation between age and all biomarkers (amyloid load, SUVr; tau Braak stages; MTA, and WMH).

Association between age and biomarkers

We observed an independent association of amyloid and WMH with, adjusted for age. The performances to tests of speed of processing and visuo-constructive abilities that are sensible and used for screening for cognitive impairment (clock-drawing test and Digit-Symbol-Coding) were specifically associated with WMH but not amyloid burden.

The following figure shows the distribution of vascular lesions in the 3 diagnostic groups assessed using both age related white matter changes (ARWMC) and white matter hyperintensities (WMH) volumes (mm3). There is no difference among the three groups based on the ARWMC but all the groups seem to differ for WMH volumes, with high WMH volumes in dementia patients, lower WMH volumes in CU. This result is coherent with the literature.

Distribution of vascular lesions across syndromic diagnosis.

Investigating the association between AD biomarkers (amyloid and tau) and global cognition (MMSE), we observed a strong association for both biomarkers. On the contrary, the association between vascular markers and MMSE seems to be weak or absent.

Finally, we tested the association between AD biomarkers and vascular markers. Amyloid load is strongly associated with tau deposition, while amyloid and tau seem to be both not associated with vascular markers.

In-depth assessment of vascular pathology using ultra high-field MRI

To achieve this task, in-depth assessment of vascular pathology in clinical samples is performed.

A sub-population underwent 7T MRI to assess vascular disease in more detail, notably microvascular changes and subcortical infracts. The results were correlated with the structural and functional changes assessed previously.

Resting-state fMRI at ultra-high field enables investigation of functional connectivity with high precision in small brain regions. However, given the higher susceptibility to artifacts and spatial distortions at 7T, sequence testing and optimization is a key step prior to data collection.

Test scans were performed in July-August 2018 for sequence optimization and quality control.

Since it is a very innovative technique, there are no standard procedure to adopt for the 7T analyses.

Eight young healthy volunteers and 24 older adults underwent MRI on a 7T.

LC functional connectivity analysis

The LC showed significant connectivity with the brainstem, the left prefrontal cortex, and the left parahippocampal cortex. At a lower threshold, the LC showed positive connectivity also with the orbitofrontal cortex, posterior cingulate cortex, bilateral parietal cortex and bilateral cerebellum. The results were confirmed when using an independent 7T LC atlas (Figure 5, bottom).

LC functional connectivity maps using the study specific LC template (top) and an independent atlas (bottom)

We also found connectivity with the brainstem (Figure 6), while no connectivity was observed between the ECN/SN and the brainstem.

Overlap between LC- and DMN-connectivity maps

Twenty-four subjects (9 CU, 7 with CI due to AD, 8 with CI due to other reasons) were included in the 7T connectivity analysis. The ‘CI due to AD’ group showed greater memory impairment than the ‘CI not due to AD’ group, while the latter group showed greater executive impairment.

Twenty out of 24 subjects were included in the ROI and voxel-wise LC connectivity analysis. There were no differences between groups in mean functional connectivity within the LC or other cognitive networks.

Voxel-wise comparisons showed lower LC connectivity in the ‘CI due to other’ group compared to controls in the mid/posterior cingulate cortex (Figure 7). There were no differences between CI due to AD patients and controls, nor between CI due to AD and CI not due to AD groups.

LC functional connectivity differences between CI due to other and controls

Significant results

At global level, the study population shows to be representative of a memory clinic population. Indeed, most of the dementia patients are amyloid and tau positive.

Vascular markers increase across disease stages but seem to be weakly or not associated with cognition or AD biomarkers. Therefore, we conclude that even if the vascular load is higher in dementia patients compared to prodromal and potentially preclinical stages, it showed to be an independent mechanism from amyloid and tau.

Regarding 7T MRI, the functional connectivity analysis suggests that the LC is functionally connected with the DMN (particularly with frontal and parahippocampal regions) and that this association is specific to this circuit, as the ECN and SN showed no connection with the brainstem. No difference in LC functional connectivity emerged in patients with AD, while patients with CI unlikely due to AD showed lower LC functional connectivity. While the null results in AD patients should be taken with caution given the small sample, these preliminary results indicate that LC assessment in vivo with high-field fMRI might provide unprecedented insight into early pathological changes in neurodegenerative diseases.