Advanced MRI methods

There are some other MRI techniques which are not widely used but they perhaps could contribute new information for the understanding of dementia diseases. These techniques are based on perfusion, diffusion and spectroscopy.

In order to determine the cerebral perfusion by using MRI, one can utilize two perfusion based techniques: MR perfusion weighted imaging (PWI) and functional MRI (fMRI).

PWI can be conducted invasively by injecting a paramagnetic contrast bolus, or non-invasively by applying a magnetic tag. When applying the bolus-tracking paramagnetic contrast, also named dynamic susceptibility contrast (DSC), one measures the signal changes during the bolus passage through the brain, which appear as a decrease in the signal intensity of T2* images (Calamante 2010). DSC illustrates regional alterations in cerebral blood flow (CBF) when comparing AD with MCI and controls (Hauser et al. 2013).

The non-invasive technique is the so-called arterial spin labeling (ASL). ASL produces a flow-sensitized image or “labelled” image and a “control” image in which the static tissue signals are identical, but where there are differences in the extent of magnetization of the inflowing blood (Petersen et al. 2006). One study showed differences in regional CBF in AD patients as compared to controls using ASL with a 3 Tesla scanner (Yoshiura et al. 2009).

Another study reported that frontal and parietal perfusion in ASL could differentiate AD and FTD with an accuracy of 87% (Du et al. 2006). Recently there has been an interest to introduce ASL into clinical practice, particularly in the field of dementia (Golay, Guenther 2012).

Even though there are many techniques which describe the dynamic patterns of function or activity taking place in the brain, when one uses the term fMRI, this is usually referring to blood oxygen-level dependent (BOLD) and resting-state networks (RSN).

In BOLD fMRI the difference on magnetic susceptibility is used as an intrinsic contrast to compare conditions (e.g. a particular response against a stimulus). The technique is based on the differences in the paramagnetic properties between oxygenated hemoglobin and deoxygenated hemoglobin (Buxton 2013). Deoxygenated hemoglobin is paramagnetic whereas oxygenated hemoglobin is not, which causes local dephasing of protons and consequently reduces the signal being emitted from those tissues. CBF increases more than oxygen metabolism when local neural activity increases; oxygenated hemoglobin is introduced into the tissue and deoxygenated hemoglobin is removed, thus tissue experiencing a change in oxygenation is detected.

Several studies have been performed in which AD and FTD patients have been asked to perform a specific task, e.g. performace in tasks targeting social cognitive deficits in bvFTD patients correlate with functional abnormalities in frontal and limbic regions in fMRI, which reflect emotional abnormalities (Virani et al. 2013); memory recall (encoding task) identifies changes in the limbic brain (Filippini et al. 2009).

RSN have gained popularity during recent years. It is based on the fact that the brain never rests completely, and thus it is possible to measure the differences on brain activation from an active state to a state of relative inactivity i.e. the subject is not doing any particular task. In other words one conducts a BOLD fMRI without asking the patient to perform a specific task, simply to be at rest. Resting state fMRI (RSfMRI) also allows the investigation of multiple regions at once, at the same time making it possible to map specific circuits.

In AD, RSN detected a reduced activation in the so-called default-mode network (DMN) (Greicius et al. 2004). DMN is the analogue of consciousness and becomes less active during the performance of a task. DMN involves the posterior cingulated cortex, the hippocampus, the inferior parietal lobule and the prefrontal cortex (Buckner, Andrews-Hanna & Schacter 2008).

The DMN becomes atrophied in AD while in bvFTD it is the salience network (SN) which undergoes atrophy. One study (Zhou et al. 2010) conducted with free-task fMRI by using independent component analysis, and found that in AD the DMN was attenuated, mostly in posterior hippocampus, medial cingulo-parieto-occipital regions and the dorsal raphe nucleus; however the SN connectivity was intensified. In bvFTD, there was reduced SN connectivity, mostly in the frontoinsular, cingulated, striatal, thalamic and brainstem nodes, while DMN connectivity was enhanced.

However, it has also been reported that apart from the SN, the DMN and the fronto-parietal network can be affected in bvFTD (Filippi et al. 2013).

AD patients display a lower functional connectivity within the DMN as compared with controls (Binnewijzend et al. 2012, Wang et al. 2007), while MCI functional connectivity values lie between AD and controls (Binnewijzend et al. 2012, Zhou et al. 2008).

Interestingly, in the posterior DMN there is decreased connectivity while in the anterior DMN there is increased activity in AD versus controls. At follow-up, all functional activity decreased within all default mode networks (Damoiseaux et al. 2012). The initial higher connectivity in the frontal regions may be a compensatory mechanism attempting to overcome the functional loss in posterior regions (Damoiseaux 2012).

Some studies also have found more decreased functional connectivity in the superior parietal lobules and inferior frontal gyrus in MCI as compared to controls (Sorg et al. 2007) and in the connections between the temporal lobe and thalamus and corpus striatum in AD (Supekar et al. 2008).

Diffusion weighted MRI (DWI) is a non-invasive method which allows the mapping of the diffusion of free water molecules in in-vivo biological tissues, working as a contrast.

The diffusion throughout tissues is not random since it is delimited by all the barriers that could be present. The water motion can be described in statistical terms by a displacement distribution, which represents the proportion of molecules that go into a specific direction and till a specific distance (Hagmann et al. 2006). These diffusion patterns may reveal details about the microstructure. In the central nervous system, the water molecules are subjected to less hindrance when parallel to the white matter tracts than when they are perpendicular to these tracts, and this difference can be used to evaluate the integrity of white matter tracts in vivo.

There are different diffusion based techniques; diffusion weighted-imaging, diffusion tensor imaging (DTI) and diffusion spectrum imaging (Hagmann et al. 2006).

DTI is the diffusion method most commonly used in the study of dementia diseases. It measures microstructural changes in WM. These are usually measured by the fractional anisotropy (FA) and the mean diffusivity (MD). A meta-analysis including studies which had compared controls with AD and/or MCI revealed that FA was decreased in AD in all regions with the exception of the parietal WM and internal capsule, while MCI patients had lower FA values in all regions except the parietal and occipital areas. MD was increased in AD in all regions, while in MCI MD was increased in all areas other than the occipital and frontal regions (Sexton et al. 2011). Another study showed that AD patients had lower FA than controls at baseline and 3 months later in the fornix and the anterior portion of the cingulum bundle, and lower FA in these regions and the splenium at baseline and after 3 months compared to MCI (Mielke et al. 2009). If one uses Tract-Based Spatial Statistics, then it is possible to identify a so-called FA-skeleton; this is the major white matter structures, improving the sensitivity, objectivity and interpretability of DTI analysis. This type of technique has revealed a significant FA decrease in the parahippocampal white matter, cingulum, uncinate fasciculus, inferior and superior longitudinal fasciculus, corpus callosum and cerebellar tracts in AD compared to controls, with the MCI values being between the AD and control patients (Liu et al. 2011). DTI has also been used for evaluating FTD. One study including controls, AD and FTD cases, showed that FTD patients had reductions in FA in frontal and temporal regions including the anterior corpus callosum, bilateral anterior and descending cingulum tracts, and uncinate tracts, compared to controls. AD patients exhibited reductions in FA in the parietal, temporal and frontal regions, including the left anterior and posterior cingulum tracts, bilateral descending cingulum tracts and left uncinate tracts, as compared to controls. In the FTD vs. AD comparison, FTD was associated with greater reductions of FA in frontal brain regions, while in AD no region showed a greater reduction of FA as compared to FTD (Zhang et al.

2009).

Another approach is to measure the concentration of particular metabolites in the brain.

This can be done using Proton Magnetic resonance spectroscopy (¹H-MRS); this is a non-invasive method for characterizing the cellular biochemistry within the brain. Specific metabolites are associated with specific locations and functions (e.g. N-acetylaspartate (NAA) is an axonal marker localized in the central and peripheral nervous systems), and their levels may correlate with specific pathology and in this way help to identify a specific disease pattern. There are four metabolites which can be quantified using ¹H-MRS: NAA, a marker of healthy neuronal density; myoinositol (MI), reflecting glial cell proliferation;

creatine/phosphocreatine (Cr) and choline-containing compounds (Cho), which reflect the products of membrane phosphotidylcholine (Kantarci 2007). Neurodegenerative dementia is characterized by elevated myoinositol and decreased NAA levels (Kantarci 2013).

The increase in MI seems to precede the decline in the NAA levels in AD. The NAA/MI ratio in the posterior cingulated gyri decreases with an increasing burden of AD pathology (Kantarci 2013). Glutamate plus glutamine levels are also decreased in AD (Antuono et al.

2001). These changes on metabolites concentrations in AD seem to be widespread (Tedeschi et al. 1996), involving the parietal and frontal lobes (Schuff et al. 1998) and the hippocampus (Schuff et al. 1997). The levels of NAA correlate with dementia severity (Kwo-On-Yuen et al. 1994) and psychotic symptoms (Sweet et al. 2002), suggesting that NAA could represent a marker for AD severity in certain clinical features.

In MCI and pre-symptomatic AD ¹H-MRS shows elevation of MI/Cr ratio and reduction in NAA/Cr ratio (Kantarci 2007). In the pathologic progression of AD, the first difference to appear is the increase in MI/Cr ratio, and later in the disease course there is a reduction in the in NAA/Cr ratio and an increase in the Cho/Cr ratio (Kantarci et al. 2000).

In FTD, ¹H-MRS displays the same pattern as in AD i.e. a decrease in the NAA/Cr ratio and an increase in the MI/Cr ratio (Kantarci 2007). Both diseases display decreased NAA levels in the posterior cingulate cortex however this is lower in the frontal region in the FTD group than in the AD group. In addition, there is a report that the MI/Cr ratio was higher in the frontal region of the FTD group in comparison with the AD group (Mihara et al. 2006).