Background
Mixed vascular and neurodegenerative dementia such as patients with Alzheimer’s disease (AD) with concomitant cerebrovascular disease (CeVD) has emerged as the leading cause of age-related cognitive impairment [
1]. Accumulating evidence suggests that AD and CeVD share multiple risk factors and overlap neuropathologically, leading to additive or synergistic effects on cognitive decline [
2]. White matter hyperintensity (WMH), which is associated with increased water content and vascular changes [
3], serves as an important clinical diagnostic criterion by which to identify CeVD status [
4]. WMH may be due to vessel disease causing infarction or a failure in the clearance of interstitial fluid from white matter (WM), which is associated with blood-brain barrier (BBB) permeability modulation [
5]. However, WMH visual rating suffers from interrater variability. More importantly, WMH usually reflects a severe water increase such as edema. It is not sensitive to detection of changes, owing to small vessel damage and inflammation, particularly in normal-appearing WM [
6,
7], which might exacerbate neurological dysfunction and brain damage in dementia.
It is thus important to isolate WM degeneration and vascular changes in mixed vascular and neurodegenerative dementia. Diffusion-weighted magnetic resonance imaging (MRI) has been used to examine WM microstructural alterations in dementia [
8]. However, the conventional diffusion tensor imaging (DTI) indices, such as fractional anisotropy (FA) or axonal diffusivity (DA), provide inconsistent findings [
7,
8], partly because changes in these indices stem from both tissue degeneration (i.e., axonal damage and demyelination) [
9] and excessive extracellular fluid [
10,
11]. Recently, the free water (FW) imaging method for obtaining diffusion-weighted MRI data has been proposed to correct each voxel for contamination from freely diffusing extracellular water molecules [
12]. The resulting fractional volume of the FW compartment estimates the extracellular water content, and the FW-corrected DTI metrics represent microstructural tissue changes [
13]. Compared with WMH, voxel-based FW increases may be more sensitive to mild vascular problems in normal-appearing tissue, including neuroinflammation and BBB permeability modulation [
6].
In the context of mixed vascular and neurodegenerative dementia, it is difficult to define which neuropathological changes contribute to cognitive decline, as well as the degree to which these changes contribute to cognitive decline, because of the heterogeneous localization of lesions and the coexistence of multiple pathologies [
14]. Although the FW imaging method has revealed whole-brain FW increases in AD and mild cognitive impairment (MCI) [
15‐
17], there is a lack of understanding of the differential patterns of FW increases, especially in the normal-appearing WM, as well as microstructural changes in AD with and without CeVD. In the present study, we examined extracellular and tissue-related WM abnormalities separately in a cohort of patients with AD without CeVD (AD), AD with CeVD (AD + CeVD), and vascular dementia (VaD) using the FW imaging method. We hypothesized that patients with AD + CeVD would have higher FW values and more tissue disruption (particularly in the frontal lobe) than patients with AD because of synergistic effects. Patients with AD would have greater FW in the normal-appearing WM than age-matched healthy control subjects (HC), even though their WMH burden is comparable. Last, we assessed whether and how FW increases and tissue compartment alterations contribute to symptom severity and cognitive impairment.
Discussion
Our findings provide new insights into differential WM microstructural and FW alterations in normal-appearing tissue in mixed vascular and neurodegenerative dementia. All patients, including patients with AD without CeVD, had increased FW compared with HC. Importantly, increased FW was detected in normal-appearing WM regions in patients with AD, reflecting mild vascular changes. Widespread FW increases in the whole WM and normal-appearing WM (but not the WMH ratio) were associated with symptom severity. Both lobar and subcortical FW increases were associated with poorer cognitive performance, such as in attention and executive functioning, whereas only left hemispheric FW increases were related to language deficits. In parallel, compared with the original DTI metrics, which might overestimate axonal damage or demyelination, the FW-corrected tissue compartment had better clinical specificity because it showed tissue-specific microstructural differences between the CeVD and non-CeVD groups. In addition, these region-specific tissue changes were associated with poorer cognitive performance. Our findings underscore the value of FW corrections in isolating vascular damage from WM microstructural alterations and help connect both extracellular FW and microstructural abnormalities with cognitive impairment in mixed vascular and neurodegenerative dementia.
As far as we know, this is the first study demonstrating that the FW measure is sensitive to vascular changes because it successfully distinguished the non-CeVD from the CeVD groups, corresponding well with their WMH ratio differences. Moreover, we further demonstrate that patients with AD had widespread FW increases compared with HC, particularly in the normal-appearing WM. Previous studies showed similar FW increases in AD [
15], but they did not reveal whether the FW increase is due to WM lesions or to changes in the normal-appearing WM. Our findings of increased FW in normal-appearing WM in the AD without CeVD group suggest mild vascular damage that may be due to microvascular degeneration [
28] and neuroinflammation-related BBB permeability modulation [
29] in normal-looking WM tissue in AD. Both imaging and postmortem studies have indicated that vascular pathology exists in AD [
2,
29]. Deposition of amyloid-β (Aβ) might lead to vascular problems such as BBB breakdown and neurovascular regulation impairment. The pathogenic chain of these vascular effects, in a vicious cycle, produces further Aβ deposition and neurofibrillary tangle formation. Together, these pathological factors may contribute to the later development of neurodegeneration and cognitive decline in AD [
30]. Other possible causes of FW increases might include abnormally low cell density, low dendrite number and volume, and WM degeneration [
25,
31‐
34]. However, group differences in FW remained after controlling for the total WMV.
Critically, the mean FW value, both across the whole WM and in the normal-appearing WM, was associated with symptom severity, whereas the WMH ratio was not. Although both WMH and FW reflect accumulating water within the WM [
5,
12], WMH reflects only severe water increases such as edema [
6]. Mild increases in FLAIR image contrast may be hard to identify. Further, the segmented WMH is often surrounded by a penumbra of subtly injured tissue with mild water content increases. These changes cannot be accurately quantified by the WMH volume [
35]. In contrast, our findings suggest that the FW metric can detect these mild water content increases in a voxel-based manner, which might contribute to the etiology of AD [
36]. Therefore, the FW compartment provides a more sensitive way of detecting mild vascular changes in normal-appearing WM regions in dementia, which could potentially help in differential diagnosis and neuropathological understanding of mixed vascular and neurodegenerative dementia.
The second important finding is that the FW-corrected DTI metrics revealed region-specific WM microstructural differences between the CeVD and non-CeVD dementia groups. CeVD-related axonal damage and demyelination have been reported in postmortem studies [
37]. However, recent data suggest that DTI-based metrics may not correspond well with the axonal pathology measured in patients with CeVD [
38]. This is in line with our observations that the original DTI metrics identified nonspecific and global WM microstructural differences between the CeVD and non-CeVD groups. Increases in the FW compartment could mask the true alterations in WM fibers [
16] (i.e., loss of oligodendrocytes, myelin sheaths, axons, and neurons) and could even overestimate tissue damage, particularly in the subcortical and brainstem regions [
24]. Therefore, on the basis of FW-corrected DTI metrics, the patients with AD + CeVD had more focal lobar damage (FA
T decrease and DR
T increase) in the occipital and frontal lobes than the patients with AD. Moreover, in previous DTI studies, researchers reported inconsistent DA changes (both increases and decreases) in patients with AD and patients with MCI compared with control subjects [
8,
9]. This inconsistency is partly due to the fact that increased extracellular fluid may lead to increased DA [
10,
11]. Using the FW correction method, we confirmed lower DA
T in patients with dementia than in HC as well as lower DA
T in the frontal and occipital regions in the CeVD group than in the non-CeVD group. This evidence lends further support to the accumulating evidence that CeVD exacerbates neurological dysfunction and brain damage in AD [
30].
The next question is whether and how FW and tissue compartment changes contribute to cognitive impairment. A recent study illustrated that FW was associated with cognitive performance in Parkinson’s disease [
39]. Only two studies showed associations between tissue compartment changes and cognition in predefined regions in MCI: Tissue fraction in the parahippocampal cingulum was associated with memory decline [
17], and tissue compartment measures in the anterior cingulum were associated with cognitive control [
40]. We found a region-specific influence of extracellular and microstructural changes on cognitive decline across patients with AD with and without CeVD using a whole-brain search. Widespread FW increases (both lobar and subcortical/brainstem) appear to be related to poor cognitive performance, particularly executive functioning, attention, visuomotor performance, and visual construction. Moreover, consistent with the literature regarding a dominant role of the left hemisphere in language [
41], only left hemispheric FW increases (but not tissue damage) were related to language deficits. In contrast, only focal lobar FW-corrected WM microstructural damage was associated with each cognitive domain, suggesting distinct default mode and executive control network disruptions in non-CeVD [
42,
43] and CeVD groups [
44]. Such differential associations of vascular and WM microstructure with cognition reveal the possible neural mechanism in patients with mixed vascular and neural degeneration.
Some limitations of the study should be noted. The possible influence of WM degeneration on FW increases could not be completely excluded [
34]. However, to mitigate this concern, we controlled for total WMV in the FW analyses, and the results remained. In addition, we did not examine amyloid deposition in our patients. Diagnoses of AD and VaD based on conventional clinical criteria may lead to heterogeneous groups with overlapping pathologies.