Radiological-Pathological Correlation in Alzheimer’s Disease: Systematic Review of Antemortem Magnetic Resonance Imaging Findings

Abstract

Background: The standard method of ascertaining Alzheimer’s disease (AD) remains postmortem assessment of amyloid plaques and neurofibrillary degeneration. Vascular pathology, Lewy bodies, TDP-43, and hippocampal sclerosis are frequent comorbidities. There is therefore a need for biomarkers that can assess these etiologies and provide a diagnosis in vivo.

Objective: We conducted a systematic review of published radiological-pathological correlation studies to determine the relationship between antemortem magnetic resonance imaging (MRI) and neuropathological findings in AD.

Methods: We explored PubMed in June-July 2015 using “Alzheimer’s disease” and combinations of radiological and pathological terms. After exclusion following screening and full-text assessment of the 552 extracted manuscripts, three others were added from their reference list. In the end, we report results based on 27 articles.

Results: Independently of normal age-related brain atrophy, AD pathology is associated with whole-brain and hippocampal atrophy and ventricular expansion as observed on T1-weighted images. Moreover, cerebral amyloid angiopathy and cortical microinfarcts are also related to brain volume loss in AD. Hippocampal sclerosis and TDP-43 are associated with hippocampal and medial temporal lobe atrophy, respectively. Brain volume loss correlates more strongly with tangles than with any other pathological finding. White matter hyperintensities observed on proton density, T2-weighted and FLAIR images are strongly related to vascular pathologies, but are also associated with other histological changes such as gliosis or demyelination.

Conclusion: Cerebral atrophy and white matter changes in the living brain reflect underlying neuropathology and may be detectable using antemortem MRI. In vivo MRI may therefore be an avenue for AD pathological staging.

Keywords

Alzheimer’s disease antemortem diagnosis dementia magnetic resonance imaging neuroimaging neuropathology

INTRODUCTION

Alzheimer’s disease (AD), the most prevalent cause of dementia among the elderly, has significant economic and societal consequences for these patients, their caregivers, and the whole community [1]. A definitive diagnosis of AD can only be established following postmortem evaluation; an accepted standard follows Montine’s “ABC staging” [2], which requires the pathological assessment of diffuse amyloid-β (Aβ) plaques, neurofibrillary tangles (NFTs), and neuritic plaque (NP) burden [3]. Beyond AD-specific pathology, however, evidence from autopsy studies indicates frequent comorbid conditions in subjects with a clinical diagnosis of AD [3, 4]. These include cerebrovascular disease [4], cerebral amyloid angiopathy [4, 5], and region-specific findings (e.g., hippocampal sclerosis [6]), as well as pathologies associated with other dementias (e.g., Lewy bodies, TDP-43 inclusions) [7].

Due to this heterogeneity in disease presentation, the clinical diagnosis of AD can be problematic for clinicians, especially in its early phase, with unsatisfactory concordance with postmortem pathology. Additional tools and methodologies are therefore needed to improve this concordance. Non-invasive imaging techniques such as anatomical magnetic resonance imaging (MRI) have the potential to address this issue by measuring etiologies in vivo. While the MRI-clinical relationship in AD has been well-researched [8 –11], in this report we provide an overview of the less abundant literature on MRI-pathological correlation, with the ambition of highlighting the relevance of clinical MRI as a potential biomarker of pathological processes in the continuum of AD.

Operational definitions

For the sake of clarity, we elected to provide operational definitions of the pathological features investigated in this review.

Diffuse plaques are typical extracellular Aβ deposits evenly dispersed in the brains. Found in patients with AD [12], they are also very commonly observed in those of cognitively intact elderly people [13]. Diffuse plaques usually present as amorphous irregular configurations with no evidence of neuritic reaction [14]. Neuritic plaques are focal collections, within the cortical gray matter, of dystrophic neurites surrounding a central extracellular amyloid core, composed of several abnormal proteins, of which the Aβ peptide is the major component [15]. The neuritic corona of NPs is an area of neuronal accumulation of fibrillar hyperphosphorylated tau [16]. The terminology “senile plaque (SP)” is an older and more general term that only refers to focal extracellular Aβ deposit. Both diffuse plaques and NPs are encompassed in the concept of senile plaque [14, 17].

Neurofibrillary tangles are dense intraneuronal aggregates of characteristic paired helical filaments, predominantly composed of hyperphosphorylated microtubule-associated tau protein. Neuropil threads in the adjacent cerebral parenchyma normally accompany these lesions. Paired helical filaments are also found in the outer portions of NPs, within the dystrophic neurites [15]. NFTs are commonly found in AD, as well as in other neurodegenerative diseases (tauopathies) [12], and form an essential part of its pathological diagnosis.

Cerebrovascular disease is commonly found in association with AD pathology [18, 19]. It may manifest in the brain as arteriosclerosis, atherosclerosis, arteriolosclerosis, infarcts, or bleeds. Arteriosclerosis consists of the thickening, hardening and loss of elasticity of arterial walls. It is a general term which refers to both medium and large arteries and encompasses both atherosclerosis and arteriolosclerosis [15]. Atherosclerosis refers to intimal lesions of arteries, characterized by the formation of fibrofatty (or atheromatous) plaques leading to a progressive luminal occlusion, a limited blood flow and eventual ischemic lesions [15, 20]. Arteriolosclerosis is the thickening of small vessels and arteriole walls in association with narrowing of the vascular lumen, which may cause ischemic injury. Arteriolosclerosis is mostly seen in association with diabetes and hypertension [15, 21]. Lacunar infarcts are caused by the occlusion of penetrating vessels that supply deep brain structures, and large-vessel infarcts are caused by the occlusion of major blood vessels on the surface of the brain [22]. These two types of infarcts can be radiologically detected; microinfarcts, on the other hand, are generally invisible lesions seen only during microscopic examination [23]. Lastly, cerebral microbleeds are small foci of chronic brain hemorrhages caused by small vessel structural abnormalities [24]. Although the articles selected for this review did not specifically select for this pathological feature, it is important to note that microbleeds have been suspected to contribute to cognitive impairment. Some interesting articles addressing this topic can be found in the literature [25 –27].

Cerebral amyloid angiopathy (CAA) is a neuropathological lesion characteristic of AD, in which Aβ and other amyloidogenic peptide deposits form in the walls of small- and medium-caliber meningeal and cortical blood vessels, typically in the interstitium between smooth muscle cells of the media [15]. These vascular lesions can lead to hematomas, microbleeds and microinfarcts, contributing to cognitive impairment [12]. CAA is most often found in parieto-occipital brain regions, and least often found in the hippocampus and subcortical white matter [28].

Hippocampal sclerosis (HS) refers to important hippocampal neuronal loss and glial proliferation, with a higher susceptibility in the cornu ammonis 1 (CA1), the cornu ammonis 4 (CA4) and the subiculum subdivisions of the hippocampus [29, 30]. This condition can be unilateral or bilateral, complete or focal, pure (e.g., related to temporal lobe epilepsy [31]) or in association with a dementia like AD, and typically causes severe memory loss [6]. We include HS in this review because it is often found in association with cerebrovascular disease [29, 32], and is believed to result from ischemia in the brain [32] (although this has been disputed; see [33]). Primarily age-related, HS occurs with high frequency in the oldest-old [33] and may contribute to misdiagnosis of AD by mimicking its clinical presentation [32].

Lewy bodies are dense eosinophilic cytoplasmic inclusions, mainly containing α-synuclein, but also parkin and ubiquitin proteins, with a lighter halo. They are found in neurons of individuals with neurodegenerative conditions like Parkinson’s disease and dementia with Lewy bodies (DLB) [15]. Lewy bodies are relatively common pathological lesions in patients with AD as well [34, 35].

TAR DNA-binding protein of 43 kDa (TDP-43) is an intracellular phosphorylated and proteolytically cleaved protein that typically accumulates in the brain of patients with the ubiquitin-positive subtype of frontotemporal lobar degeneration and in those with amyotrophic lateral sclerosis [36, 37]. However, studies have also revealed a high prevalence of these inclusions in other neurodegenerative diseases, with up to 57% of brains otherwise pathologically categorized as AD [38]. The amygdala appears to be the first and most common structure presenting with pathological accumulation of TDP-43 in AD [36 , 38–40].

MATERIALS AND METHODS

Eligibility criteria

Studies were included in this review if they collected MRI and postmortem pathological data in individuals with clinically- or pathologically-diagnosed AD. Since cognitively-silent lesions are often found in elderly brains [41], cognitive performance was not an inclusion criterion. This is relevant given that the purpose of this research was to report on the correlation between AD-associated neuropathological features and MRI findings alongside the neuropathological continuum. Studies had to meet the following criteria:

be an original study (i.e., narrative or systematic reviews or meta-analyses were not included);

use clinical (1.0T–3.0T) MRI scanning, in order to maintain consistency with current practice standards; and

be an in vivo MRI study, as MRI scanning ex vivo may be affected by formalin fixation, brain shrinkage, and normal changes following death, regardless of the time between death and scanning [42]. Further, postmortem MRI is less sensitive than in vivo MRI for detecting smaller white matter hyperintensities (WMH) [25, 43] and perivascular atrophic demyelination [44].

Search strategy and study selection

We explored the PubMed database from June to July 2015 using “Alzheimer’s disease” combined with the following terms: “postmortem pathology in vivo MRI correlation”, “autopsy MRI”, “neuropathology MRI correlation”, “postmortem histopathology MRI association”, “quantitative MRI pathology correlation”, “volumetric MRI histopathology correlation”, “antemortem MRI autopsy”, “structural neuroimaging postmortem pathology”, “MRI cerebral amyloid angiopathy postmortem pathology”, “MRI vascular postmortem pathology”, “MRI Lewy bodies postmortem pathology”, “MRI hippocampal sclerosis postmortem pathology” and “MRI TDP-43 pathology” as keywords and MeSH headings. No limit was applied to the date of publication, MRI being a relatively recent technology and the discoveries already made still applicable to date. The first author (C.D.-T.) screened the articles and selected studies for inclusion first by examining the title and abstract of each article, then the full text version. The remaining articles were then reviewed by other authors (B.L.C. and S.D.) to confirm their relevance. This present systematic review was done following Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [45] (Fig. 1).

Data extraction

We extracted necessary data for description of the collected articles. Study design, population characteristics (source, number of subjects, selection criteria, age, and gender distribution), imaging characteristics (modality, resolution(s) and sequence(s)), time between MRI and death, types of dementia, radiologic and histopathologic variables examined, and main study findings were noted in Table 1. We further assessed the articles alongside three main categories: AD-related findings, cerebrovascular disease-related findings, and incidental findings.

We extracted relevant significant (p < 0.05) coefficients from studies included in the review (Table 2) and performed a general analysis based on significance of correlations between neuropathological and MRI findings (Fig. 2). The interpretation of p-values was based on recommendations from B. Rosner [46].

RESULTS

Studies description

Our PubMed search yielded 552 articles, from which we excluded 481 at initial screening and 47 at full-text assessment. The relevance of the remaining 24 articles was confirmed by the second evaluation. Three articles were added by hand after searching the reference lists of the 24 saved articles. Our review is based on a total of 27 studies published between 1995 and 2015 (Table 1 and Fig. 1).

Twenty-four were longitudinal studies [7 , 47–50] or used subjects from longitudinal cohort studies but analyzed the data cross-sectionally [6 , 51–66]; the remaining three articles were descriptive cross-sectional studies [67 –69]. All studies were published in English; most of them were conducted in the United States [6 , 65–68], one in Canada and the United States [61], two in the United Kingdom [56, 64], and one in the Netherlands [69]. Subjects in all studies were between 43–108 years old from enrollment to death, and all included both men and women. MRI field strengths of 1.0T to 3.0T were used in every study, except for five in which field strength was not explicitly reported [40 , 60]. One study also compared the sensitivity to neuropathological changes between 7.0T and 3.0T MRI [69]. In total, T1-weighted images were evaluated in 21 studies [7 , 66–69], T2-weighted images in five studies [50 , 69], proton density images in three studies [50 , 67], FLAIR images in four studies [48 , 69], and type of images was unspecified in three of the reviewed articles [6 , 52]. Time interval between last MRI and death varied from 0 to 10.5 years, but information was not available in two studies [38, 40].

Neuropathological findings

Figures 2 and 3 give a brief overview of results from studies included in the present review.

AD-related findings

Diffuse and neuritic plaques

Radiological-pathological evidence suggests that NPs are associated with brain volumes. Of the 27 studies reviewed, 10 specifically investigated correlations between brain volumes and diffuse or neuritic plaques. Eight of these studies reported specifically on NPs [48 , 64], and two reported on SPs more generally (i.e., made no distinction between neuritic and diffuse plaques) [47, 68]. The majority of these studies agree that total brain volume [62], medial temporal lobe volume [56], hippocampal volume [52, 58], ventricular size [62], or ventricular expansion over time [49] are significantly associated with NPs in postmortem brain tissue. NFT and senile plaque burden have also been associated with these measures [47, 68], and some authors have reported the rate of ventricular volume enlargement as their strongest correlate [47]. A moderate correlation (ρ= 0.52) has also been reported between CERAD [70] NP count and the adjusted STructural Abnormality iNDex (aSTAND) score, which evaluates the degree and pattern of cerebral atrophy adjusted for age and gender [54]. One study [68] found a non-significant correlation between left and right HVs and hippocampal senile plaque density, but a moderate correlation between total cerebral volume and hippocampal senile plaque density (ρ= –0.68). Only one study [48] suggested a lack of association between total Aβ burden (i.e., compact and diffuse Aβ) and rates of whole-brain atrophy and ventricular expansion in patients with low, intermediate or high probability of AD according to the National Institute on Aging and Reagan Institute Working Group on Diagnostic Criteria for the Neuropathological Assessment of Alzheimer’s disease [71].

Neurofibrillary tangles and Braak stage

Sixteen studies selected for this review focused on NFTs as a pathological variable of interest [47–52 , 68] and nearly all reported moderate to strong positive associations between NFT pathology and brain atrophy. In fact, several authors have suggested that MRI brain volumes correlate more strongly with NFTs than with NPs [48 , 62] or SPs [47]. Selected transversal studies addressing NFT pathology found significant correlations between whole-brain atrophy on MRI and NFT staging (according to the Braak and Braak scale [72]) [48 , 62], as well as between Braak staging and ventricular size [48, 62]. Moreover, a moderate correlation (ρ= 0.62) was found between Braak NFT staging and aSTAND scores [54].

Evidence suggests that Braak stage is also significantly associated with hippocampal atrophy [51 , 68], although some authors note that these correlations are weaker than those involving whole-brain atrophy or aSTAND scores [54]. Importantly, total Braak NFT score, and not percent area of tangles, appeared to be the only significant predictor of MRI-measured medial temporal lobe atrophy [56]. These associations may, to some extent, be influenced by regional distribution of pathology. For example, Burton and colleagues [56] reported that total (left and right) medial temporal atrophy rating was correlated (ρ= 0.49) with percent area of NFTs in the hippocampus. Further, MRI-detected hippocampal atrophy has been specifically linked to the accumulation of NFTs in the CA1 subfield of the hippocampus [68]. Medial temporal atrophy rating based on a 5-point scale (from absent to severe; based on the height of hippocampus and the width of cerebrospinal fluid (CSF)-filled spaces) shows high sensitivity (91%) for autopsy-confirmed AD [56].

According to Whitwell and colleagues [55], the severity of gray matter loss may be more closely related to quantitative NFT load than to Braak NFT stage. Their study was conducted using voxel-based morphometry with group-wise comparisons between each Braak stage (III to VI) and the control group (Braak stages 0 to II). Because no regions of grey matter atrophy were found at Braak stages III or IV when compared to the control group, group analyses were also performed between subjects with low tau burden (<10%) and high tau burden (≥10%) in subjects with Braak stages 0 to IV. Those with high tau burden showed greater gray matter loss in medial and lateral temporal lobes than those with low tau burden. In addition, results from this study showed a graded pattern of increasing brain atrophy across Braak stages V and VI. In subjects with Braak stage V, atrophy was found in temporal, parietal and frontal lobes only. In subjects with Braak stage VI, basal forebrain, posterior cingulate, precuneus, anterior cingulate, insula, subcortical nuclei, and occipital lobe including the primary visual cortex were also involved. In other words, the presence of NFTs in the brain tissue is associated with patterns of gray matter atrophy, with NFT burden playing a more important role in determining the pattern of gray matter loss at Braak stages III and IV, and Braak stage itself (i.e., regional distribution of NFTs) having a greater correlation with patterns of brain atrophy at Braak stages V and VI. Longitudinal data also support associations between repeated structural MRI measures and AD pathology: ventricular expansion [47 –49] and rates of cerebral atrophy [47, 48], but not hippocampal atrophy [47, 49], have been significantly associated with the amount of NFTs at autopsy. In fact, the rate of total ventricular CSF volume increase (an indicator of ventricular expansion) was identified by one study as the best radiological biomarker of AD pathology (R² = 0.422 for cortical NFTs and R² = 0.397 for cortical SPs), when compared to the rate of cerebral atrophy (R² = 0.419 for cortical NFTs, and R² = 0.22 for cortical SPs) [47]. Postmortem pathological diagnosis of AD has also been associated with whole-brain atrophy rate as measured using repeated MRI during life [7].

In addition to the substantial evidence linking NFTs to cerebral atrophy, one study reported that more extensive NFT pathology, as rated by Braak staging, was associated with MRI-evidence of increased longitudinal WMH accumulation [50], while another study reported an inverse correlation when assessed cross-sectionally [65]. The relationship between Braak stage and WMH remains to be confirmed by other studies, as no other work included in this literature review assessed similar associations with WMH.

Cerebrovascular disease-related findings

Arterio-, arteriolo-, and atherosclerosis, infarcts and bleeds

Based on results identified in mixed cohorts of both demented and non-demented participants (see Table 1 for more details), WMH on proton density, T2-weighted and FLAIR MRI seem to be a good in vivo marker of vascular pathology. In cross-sectional studies, WMH have been associated with subcortical infarcts [53], cerebral hemorrhages [65], arteriosclerosis [65, 67], demyelination [53 , 67], breakdown of the ventricular lining [65], reduced periventricular vessels [65], and gliosis [67] in autopsied brains. Microinfarcts, particularly in the deep nuclei, may also be related to WMH burden on antemortem MRI [62, 65] (but see [63] for other results). The only longitudinal study found that, of all vascular lesions, arteriolosclerosis was the most important correlate of accumulating WMH over time [50].

Imaging evidence of vascular disease (e.g., infarcts, hemorrhage, stroke) is also a strong indicator of underlying vascular pathology at autopsy. Raman and colleagues [63] have shown that the presence and number of cortical and subcortical infarcts detected on antemortem MRI are associated with the presence and number of microinfarcts seen on postmortem pathological examination. One study aiming to detect cortical microinfarcts using in vivo MRI, followed by postmortem pathological validation, found that higher-powered 7.0T MRI is more sensitive than 3.0T MRI, with an ability to detect cortical microinfarcts of ≥1 mm as hyperintensities on FLAIR and T2-weighted scans [69]. It has been reported that detectable lacunes on MRI (well-circumscribed foci >2 mm in diameter and hyperintense when compared to CSF on proton density MRI) are highly reflective of subcortical vascular pathology, representing small-vessel infarcts and arteriosclerosis [53].

Brain volumetrics have been less extensively studied as correlates of vascular pathology. Only four of the selected articles cross-sectionally investigated radio-pathological associations between cerebrovascular disease and total brain volume [57], cortical gray matter volume [53], ventricular volume [57], and HV [52 , 62]. Jagust’s group reported that lower cortical gray matter volume was associated with gross subcortical infarcts, but not cortical infarcts [53]. Kaur and colleagues [62] also reported that lower hippocampal volume was associated with higher number of noncavitated and subcortical microinfarcts, while WMH were associated with more deep nuclei microinfarcts. In contrast, the two other groups found no significant association between whole-brain, ventricular and hippocampal volumes and ischemic, hemorrhagic or vascular pathology [52, 57].

Similarly, only two studies longitudinally assessed vascular correlates of ventricular expansion [49] and rates of whole-brain and regional atrophy [63]. Gross infarcts at autopsy have been associated with ventricular volume trajectory, with more ventricular enlargement over time observed in individuals with this type of vascular lesion [49]. Moreover, Raman and colleagues [63] found that whole-brain atrophy rate was the only MRI variable that differentiated subjects with and without microinfarcts after controlling for AD-related pathology, compared to imaging measures of hippocampal atrophy or WMH accumulation.

Cerebral amyloid angiopathy

Despite its vascular involvement, CAA was not significantly associated with WMH in demented or non-demented individuals in any of the studies reviewed [53 , 65]. However, CAA has been correlated with increased hippocampal atrophy over time with a – 0.0008% volume change as proportion of intracranial volume (p = 0.009) and also showed a trend for association with total brain volume atrophy over time with a – 0.13% volume change as proportion of intracranial volume (p = 0.030; the significance threshold was adjusted for multiple comparisons to p < 0.017) [49].

Incidental findings

Lewy bodies

Results from many cross-sectional studies examining both cognitively intact and cognitively impaired subjects have shown no significant associations between Lewy bodies and occipital gray matter [61], medial temporal lobe [56], or hippocampal [51 , 64] volumes. Overall, within the selected publications, no MRI-detected changes have been related to this neuropathological feature, except for a positive correlation between percent area of Lewy bodies in the hippocampus and total hippocampal volume [64].

Hippocampal sclerosis

Hippocampal atrophy is the most prominent radiological correlate of HS. Several cross-sectional studies have reported that the presence of HS in AD and non-AD brains is strongly inversely associated with HV [52 , 62] with greater pathologic severity related to greater atrophy on MRI [51]. Moreover, Zarow and colleagues [52] found a moderate negative correlation (ρ= –0.63) between number of neurons in CA1 subfield of the hippocampus and HS. The same authors also reported a strong association between MRI HV and number of neurons in CA1 (ρ= 0.72) as well as volume of CA1 (ρ= 0.54). Jagust and colleagues [53] have suggested that hippocampal volume is independently related to both typical AD lesions and HS, as hippocampal atrophy attributed to HS seems to be more severe in subjects with low levels of AD pathology. Interestingly, hippocampal atrophy on anatomical MRI was associated with neuronal loss (i.e., reduction in number of neurons) and not shrinkage (i.e., change in neuron size), regardless of etiology [52]. MRI appearance has also been linked to HS in the context of epilepsy [30].

TDP-43

Given the growing literature reporting a high prevalence of TDP-43 in AD brains [36 , 73–76], radiological features associated with this pathological lesion have been examined in many transversal studies captured in the present review [6 , 66]. When comparing AD subjects with and without abnormal TDP-43 immunoreactivity using voxel-based morphometry, those with TDP-43 pathology showed greater volume loss in the medial temporal lobes [38, 66], with greatest loss in the hippocampus [66]. Consistent with these findings, additional group comparisons using MRI segmentation techniques showed similar results: TDP-43 pathology was associated with lower amygdala [38], hippocampal [38 , 61], and entorhinal cortex [38, 40] volumes. These atrophy sites have been correlated to both increased TDP-43 burden [38] and higher TDP-43 stage (i.e., regional distribution) [40]. In one study [6], while HS was found in 61% of AD cases, it was found in as many as 92% of HS cases. This finding suggests a possible association between TDP-43 and HS pathology. However, no differences in hippocampal volume were found between AD TDP-positive subjects with and without HS [6], and hippocampal volumes remain significantly associated with the presence of TDP-43 even after accounting for HS [38].

DISCUSSION

Sensitivity of MRI for AD identification

Overall, the results from this systematic review suggest that in vivo clinical-resolution MRI is sensitive to most neuropathological lesions observed in AD subjects. It has been shown that AD-related anatomical changes (i.e., SPs and NFTs) are associated with volumetric measurements of several brain structures on MRI. Whole-brain, medial temporal lobe and hippocampal atrophy, as well as ventricular enlargement, can predict underlying Aβ and tau pathologies accumulating in the brain tissue with high sensitivity, with ventricular enlargement clearly identified by one study as the best radiological biomarker for AD pathology [47]. Interestingly, the association between brain volumes and plaques or tangles may be limited to individuals with cognitive impairment, and absent in those who are cognitively intact [47, 51]. This might be explained by an existing threshold after which AD neuropathology burden leads to brain volume loss, similar to the onset of cognitive impairment occurring after many years of accumulation of these lesions [77]. Unfortunately, two of the studies investigating the relationship between structural atrophy and amyloid neuropathology [47, 51] did not distinguish between neuritic and diffuse amyloid plaques, and potential differences between them may warrant further investigation.

Importantly, neuronal and synaptic losses have primarily been associated in extant literature with the accumulation of NFTs, rather than with plaque formation [13]. Only one study [62] examined radio-pathologic relationships separately for tau (NFTs) and amyloid (NPs), and found that although whole-brain volume was associated with both pathological variables, only NFTs remained a significant correlate when both were entered into linear regression models. This suggests that the associations between structural atrophy and amyloid plaques reported in the present review may have been confounded to some extent by the degree of tau pathology, and that MRI may be more sensitive for detecting tau than amyloid pathology.

The association between vascular pathology and cerebral atrophy is less clear-cut. Vascular lesions (e.g., arteriosclerotic patterns and infarcts) can generally be easily detected as WMH on antemortem MRI, except for microinfarcts, which are macroscopically invisible by definition. Lacunar infarcts may manifest differently, but are still radiologically detectable as lacunes in the white matter [78].

Underlying HS can also be identified on brain MRI, the most reliable indicator being significant hippocampal volume loss, particularly in the CA1 and CA2 subfields of the hippocampus [52]. Studies included in this review did not show any MRI sensitivity for comorbid Lewy body pathology in individuals with AD, although other authors have identified a specific pattern of relatively focused atrophy, separate from that exemplified in AD subjects, in the midbrain and hypothalamus with relative sparing of the hippocampus and temporoparietal cortices, in individuals with a clinical diagnosis of DLB [79]. There is also increasing evidence showing that concomitant TDP-43 can be visualized on MRI as medial temporal lobe atrophy, independently of HS. Overall, MRI has a good sensitivity for AD pathology and most other frequentcomorbidities.

Specificity of MRI for AD identification

AD-related pathology

Our results suggest moderate specificity of MRI-defined changes in AD subjects. First, cerebral atrophy, decidedly the best-known radiological feature of AD, is also common in other types of neurodegenerative diseases. Increased rates of whole-brain atrophy and ventricular expansion, two usually associated features, are found in AD as well as other dementias such as pathologically confirmed cases of mixed AD/DLB, frontotemporal degeneration, corticobasal degeneration, and progressive supranuclear palsy, when compared to normal aging subjects [7] and therefore cannot, on their own, serve as reliable or specific biomarkers for AD.

However, Whitwell and colleagues [7] have suggested that longitudinal rates of cerebral atrophy differ according to the underlying neuropathological process, which may improve specificity. In their study, the lowest rates were observed in pure DLB pathology, while the highest rates were observed in corticobasal degeneration, followed by frontotemporal dementia. Nonetheless, AD, mixed AD/DLB and progressive supranuclear palsy showed similar intermediary rates of atrophy, and conventional MRI volumetry could not help differentiate those three. Relative to whole–brain atrophy, medial temporal lobe atrophy rating based on a 5-point scale has been shown to be a more accurate diagnostic marker for differentiating AD from DLB and vascular dementia, with good discriminatory power and 94% specificity [56]. Furthermore, the capacity of MRI-derived hippocampal volume to differentiate between AD and vascular dementia has been touted, as neuronal loss in CA1 is more consistently found in AD subjects than in those with vascular dementia [52]. Additional studies using atrophy maps specific to each neurodegenerative dementia [80] or more developed techniques such as image appearance [81] or grading [82], especially focused on medial temporal areas, are needed to confirm these findings, since other authors have suggested that cumulative ischemia without amyloid pathology (i.e., pure subcortical vascular dementia) might lead to atrophy and shape changes in the CA1 and subiculum regions of thehippocampus [83].

Structural MRI may also be used to differentiate pathologically-defined subtypes of AD. For example, the greatest cortical atrophy is seen in hippocampal-sparing AD, followed by typical AD and then limbic-predominant AD [60]. Conversely, the greatest medial temporal lobe atrophy is observed with limbic-predominant AD, followed by typical AD, and then hippocampal-sparing AD [60]. However, the best discriminator of the three subtypes of AD is the ratio of hippocampal to corticalvolumes [60].

One group of researchers found that the estimated number of neurons in both CA1 and CA2 regions of the hippocampus is decreased in patients with pure AD relative to those with ischemic vascular dementia or normal controls [52]. This group also reported an evident correlation between CA1 neuron counts and MRI-derived hippocampal volume. The severity of CA1 neuron loss (i.e., hippocampal atrophy) might even help to distinguish between different stages of AD: the authors reported 48% reduction in CA1 neurons in subjects with isocortical stage AD, and 8% reduction in subjects with limbic stage AD. Unfortunately, hippocampal atrophy is not specific to AD neurodegeneration. In fact, subjects with a pathologically confirmed diagnosis of AD, HS, frontotemporal dementia or NFT-only degeneration usually have considerable hippocampal atrophy compared to those with normal age-related changes [51].

Concomitant cerebrovascular disease

Regarding radiologically-detectable white matter changes, many studies showed that there is no relationship between WMH and AD pathology or CAA [42 , 53]. As suggested by myriad radiological-pathological correlation studies, WMH best reflect vascular changes in the brain [42–44 , 62]. In fact, WMH are related to age and vascular risk factors, and are strongly reflective of underlying cerebrovascular disease and complete and incomplete infarction [53]. Takao and colleagues [42] evaluated the pathologic significance of different types of WMH on T2-weighted MR images of non-demented subjects. They found that small patchy WMH (dot-like or confluent lesions not attached to the lateral ventricles) corresponded to demyelination, gliosis and arteriosclerosis. Neither infarction nor amyloid deposits were found to be associated with these types of lesions. ‘Caps’ (WMH around the anterior or posterior poles of the lateral ventricles) represented demyelination, dilated perivascular spaces and arteriosclerosis. No infarction was noted. ‘Rims’ (thin WMH along the body of the lateral ventricles) accounted for subependymal gliosis and partial disruption of ependymal lining. Lastly, most histologically-discovered silent white matter lesions on MRI were found to reflect partial demyelination and arteriosclerosis. Only two studies reported a positive [59] and a negative [65] relationship between cross-sectional MRI-detected WMH and AD-related deposits (NPs and NFTs) as assessed by CERAD score, Braak stage and a composite AD pathology score [59], but these conflicting findings have yet to be clarified by other research groups. Increased accumulation of WMH over time was also correlated to NFT pathology, when comparing low (0, I or II) versus high (V or VI) Braak stage [50]. The authors from these papers hypothesized that this association may be explained by the fact that subjects were much older at death than those in most other cohorts[50, 59].

The impact of aging on atrophy

The moderate specificity of antemortem MRI may at least help differentiate normal aging from pathological processes, although age at the time of scan seems to influence the rates of brain volume change. The rates of both brain volume loss and ventricular expansion decrease as the age at scan increases, which means that greater atrophic effects of AD pathology are seen in younger individuals [7, 48]. This measure may therefore be particularly useful in detecting cases ofearly-onset AD.

Interestingly, cerebral atrophy over time differs in subjects with and without cognitive impairment even after accounting for the severity of SPs [47], NPs [49], and NFT [47, 49], suggesting that there may be underlying protective factors or other mechanisms that can lead to increased brain reserve and resistance to atrophy. Similarly, one study reported that larger total brain and hippocampal volumes were associated with preserved cognitive function during life despite a high number of AD-relatedlesions [57].

Limitations

Several sources of potential bias in this review must be acknowledged. The major limitation of this research is the publication bias, which may have resulted in incomplete or inexact conclusions. Since this review is based on published literature, studies that have not been published because of negative results (no association found or opposite to expectations) are not included in the analysis. A second factor is related to the populations under study, the vast majority (23 out of 27 studies) being from the United States. Geographic, ethnic, or cultural differences, such as lifestyle, risk factor exposure, and genetics, may influence findings, and thus the generalizability of these results is limited. Another important limitation is that several studies have small samples, which reduces the statistical power of observations. Moreover, studies mainly looked at only one type of structural measure in T1-weighted scans (i.e., volume), and focused in only one region apart from the whole brain and ventricles (i.e., medial temporal lobe, with specific attention to hippocampal volume). More studies on other cortical and subcortical regions of the brain and using other techniques such as additional morphological measures (surfaces, thicknesses) or diffusion are needed in order to draw adequate conclusions.

Conclusions

In summary, the reviewed studies have demonstrated the capacity of clinical-resolution MRI (1.0T–3.0T) to detect most pathological alterations due to AD before death with high sensitivity and moderate specificity. MR-defined cerebral atrophy in the antemortem brain appear to be the most strongly reflective of underlying AD neuropathology and other incidental neurodegenerative findings such as HS and TDP-43, while WMH correlate better with concomitant cerebrovascular disease as seen at autopsy. New protocols and especially post-acquisition image processing techniques should focus on improving the specificity of MRI to detect pathological lesions related to AD. Should this be achieved, then in vivo pathological staging of AD will be possible noninvasively, raising the hope of identifying affected patients well before they develop full-blown dementia.

Footnotes

ACKNOWLEDGMENTS

We gratefully acknowledge financial support from the Alzheimer’s Society of Canada (#13–32). B.L.C. receives financial support from the Alzheimer’s Association (AACF-16-443540). S.D. is a Research Scholar from the Fonds de recherche du Québec – Santé (#30801).

Authors’ disclosures available online ().

References

Alzheimer’s Association (2010) 2010 Alzheimer’s disease facts and figures. Alzheimers Dement 6, 158–194.

Montine

, Phelps

, Beach

, Bigio

, Cairns

, Dickson

, Duyckaerts

, Frosch

, Masliah

, Mirra

, Nelson

, Schneider

, Thal

, Trojanowski

, Vinters

, Hyman

, National Institute on Aging, Alzheimer’s Association (2012) National Institute on Aging-Alzheimer’s Association guidelines for the neuropathologic assessment of Alzheimer’s disease: A practical approach. Acta Neuropathol 123, 1–11.

Braak

, Braak

(1991) Neuropathological stageing of Alzheimer-related changes. Acta Neuropathol 82, 239–259.

Brenowitz

, Nelson

, Besser

, Heller

, Kukull

(2015) Cerebral amyloid angiopathy and its co-occurrence with Alzheimer’s disease and other cerebrovascular neuropathologic changes. Neurobiol Aging 36, 2702–2708.

Jellinger

(2010) Prevalence and impact of cerebrovascular lesions in Alzheimer and lewy body diseases. Neurodegener Dis 7, 112–115.

Zarow

, Weiner

, Ellis

, Chui

(2012) Prevalence, laterality, and comorbidity of hippocampal sclerosis in an autopsy sample. Brain Behav 2, 435–442.

Whitwell

, Jack

Jr , Parisi

, Knopman

, Boeve

, Petersen

, Ferman

, Dickson

, Josephs

(2007) Rates of cerebral atrophy differ in different degenerative pathologies. Brain 130, 1148–1158.

Belleville

, Fouquet

, Duchesne

, Collins

, Hudon

(2014) Detecting early preclinical Alzheimer’s disease via cognition, neuropsychiatry, and neuroimaging: Qualitative review and recommendations for testing. J Alzheimers Dis 42(Suppl 4), S375–S382.

Burggren

, Brown

(2014) Imaging markers of structural and functional brain changes that precede cognitive symptoms in risk for Alzheimer’s disease. Brain Imaging Behav 8, 251–261.

10.

Braskie

, Thompson

(2014) A focus on structural brain imaging in the Alzheimer’s disease neuroimaging initiative. Biol Psychiatry 75, 527–533.

11.

Weiner

, Veitch

, Aisen

, Beckett

, Cairns

, Cedarbaum

, Green

, Harvey

, Jack

, Jagust

, Luthman

, Morris

, Petersen

, Saykin

, Shaw

, Shen

, Schwarz

, Toga

, Trojanowski

, Alzheimer’s Disease Neuroimaging Initiative (2015) 2014 Update of the Alzheimer’s Disease Neuroimaging Initiative: A review of papers published since its inception. Alzheimers Dement 11, e1–120.

12.

Thompson

, Vinters

(2012) Pathologic lesions in neurodegenerative diseases. Prog Mol Biol Transl Sci 107, 1–40.

13.

Serrano-Pozo

, Frosch

, Masliah

, Hyman

(2011) Neuropathological alterations in Alzheimer disease. Cold Spring Harb Perspect Med 1, a006189.

14.

Morris

, Nagy

(2004) Alzheimer’s disease. In The Neuropathology of Dementia, 2nd edition, Esiri

, Lee

VM-Y

, Trojanowski

, eds. Cambridge University Press, UK, pp. 161–206.

15.

Kumar

, Abul

, Fausto

(2005) Robbins and Cotran Pathologic Basis of Disease, Elsevier Saunders, Philadelphia.

16.

Duyckaerts

, Dickson

(2011) Neuropathology of Alzheimer’s disease and its variants. In Neurodegeneration: The Molecular Pathology of Dementia and Movement Disorders, Dickson

, Weller

, eds. Wiley-Blackwell, Oxford, UK.

17.

Dickson

(1997) The pathogenesis of senile plaques. J Neuropathol Exp Neurol 56, 321–339.

18.

Chui

, Ramirez-Gomez

(2015) Clinical and imaging features of mixed Alzheimer and vascular pathologies. Alzheimers Res Ther 7, 21.

19.

Osborn

, Salzman

, Jhaveri

, Barkovich

(2016) Vascular dementia. In Diagnostic Imaging: Brain, 3rd Edition, Osborn

, Salzman

, Jhaveri

, Barkovich

eds. Elsevier, Philadelphia, PA.

20.

Osborn

, Salzman

, Jhaveri

, Barkovich

(2016) Intracranial atherosclerosis. In Diagnostic Imaging: Brain, 3rd Edition, Osborn

, Salzman

, Jhaveri

, Barkovich

eds. Elsevier, Philadelphia, PA.

21.

Osborn

, Salzman

, Jhaveri

, Barkovich

(2016) Arteriolosclerosis. In Diagnostic Imaging: Brain, 3rd Edition, Osborn

, Salzman

, Jhaveri

, Barkovich

eds. Elsevier, Philadelphia, PA.

22.

Blumenfeld

(2010) Neuroanatomy through Clinical Cases, Sinauer Associates, Inc., Sunderland, MA.

23.

Smith

, Schneider

, Wardlaw

, Greenberg

(2012) Cerebral microinfarcts: The invisible lesions. Lancet Neurol 11, 272–282.

24.

Martinez-Ramirez

, Greenberg

, Viswanathan

(2014) Cerebral microbleeds: Overview and implications in cognitive impairment. Alzheimers Res Ther 6, 33.

25.

Gouw

, Seewann

, van der Flier

, Barkhof

, Rozemuller

, Scheltens

, Geurts

(2011) Heterogeneity of small vessel disease: A systematic review of MRI and histopathology correlations. J Neurol Neurosurg Psychiatry 82, 126–135.

26.

Werring

(2007) Cerebral microbleeds: Clinical and pathophysiological significance. J Neuroimaging 17, 193–203.

27.

Schrag

, McAuley

, Pomakian

, Jiffry

, Tung

, Mueller

, Vinters

, Haacke

, Holshouser

, Kido

, Kirsch

(2010) Correlation of hypointensities in susceptibility-weighted images to tissue histology in dementia patients with cerebral amyloid angiopathy: A postmortem MRI study. Acta Neuropathol 119, 291–302.

28.

Vinters

, Gilbert

(1983) Cerebral amyloid angiopathy: Incidence and complications in the aging brain. II. The distribution of amyloid vascular changes. Stroke 14, 924–928.

29.

Dickson

, Davies

, Bevona

, Van Hoeven

, Factor

, Grober

, Aronson

, Crystal

(1994) Hippocampal sclerosis: A common pathological feature of dementia in very old (>or =80 years of age) humans. Acta Neuropathol 88, 212–221.

30.

Blumcke

, Thom

, Aronica

, Armstrong

, Bartolomei

, Bernasconi

, Bien

, Cendes

, Coras

, Cross

, Jacques

, Kahane

, Mathern

, Miyata

, Moshe

, Oz

, Ozkara

, Perucca

, Sisodiya

, Wiebe

, Spreafico

(2013) International consensus classification of hippocampal sclerosis in temporal lobe epilepsy: A Task Force report from the ILAE Commission on Diagnostic Methods. Epilepsia 54, 1315–1329.

31.

Bernasconi

, Duchesne

, Janke

, Lerch

, Collins

, Bernasconi

(2004) Whole-brain voxel-based statistical analysis of gray matter and white matter in temporal lobe epilepsy. Neuroimage 23, 717–723.

32.

Zarow

, Sitzer

, Chui

(2008) Understanding hippocampal sclerosis in the elderly: Epidemiology, characterization, and diagnostic issues. Curr Neurol Neurosci Rep 8, 363–370.

33.

Nelson

, Schmitt

, Lin

, Abner

, Jicha

, Patel

, Thomason

, Neltner

, Smith

, Santacruz

, Sonnen

, Poon

, Gearing

, Green

, Woodard

, Van Eldik

, Kryscio

(2011) Hippocampal sclerosis in advanced age: Clinical and pathological features. Brain 134, 1506–1518.

34.

Hamilton

(2000) Lewy bodies in Alzheimer’s disease: A neuropathological review of 145 cases using alpha-synuclein immunohistochemistry. Brain Pathol 10, 378–384.

35.

Atiya

, Hyman

, Albert

, Killiany

(2003) Structural magnetic resonance imaging in established and prodromal Alzheimer disease: A review. Alzheimer Dis Assoc Disord 17, 177–195.

36.

Arai

, Mackenzie

, Hasegawa

, Nonoka

, Niizato

, Tsuchiya

, Iritani

, Onaya

, Akiyama

(2009) Phosphorylated TDP-43 in Alzheimer’s disease and dementia with Lewy bodies. Acta Neuropathol 117, 125–136.

37.

Neumann

, Sampathu

, Kwong

, Truax

, Micsenyi

, Chou

, Bruce

, Schuck

, Grossman

, Clark

, McCluskey

, Miller

, Masliah

, Mackenzie

, Feldman

, Feiden

, Kretzschmar

, Trojanowski

, Lee

(2006) Ubiquitinated TDP-43 in frontotemporal lobar degeneration and amyotrophic lateral sclerosis. Science 314, 130–133.

38.

Josephs

, Whitwell

, Weigand

, Murray

, Tosakulwong

, Liesinger

, Petrucelli

, Senjem

, Knopman

, Boeve

, Ivnik

, Smith

, Jack

Jr , Parisi

, Petersen

, Dickson

(2014) TDP-43 is a key player in the clinical features associated with Alzheimer’s disease. Acta Neuropathol 127, 811–824.

39.

, Josephs

, Knopman

, Boeve

, Dickson

, Petersen

, Parisi

(2008) Temporal lobar predominance of TDP-43 neuronal cytoplasmic inclusions in Alzheimer disease. Acta Neuropathol 116, 215–220.

40.

Josephs

, Murray

, Whitwell

, Parisi

, Petrucelli

, Jack

, Petersen

, Dickson

(2014) Staging TDP-43 pathology in Alzheimer’s disease. Acta Neuropathol 127, 441–450.

41.

Sonnen

, Santa Cruz

, Hemmy

, Woltjer

, Leverenz

, Montine

, Jack

, Kaye

, Lim

, Larson

, White

, Montine

(2011) Ecology of the aging human brain. Arch Neurol 68, 1049–1056.

42.

Takao

, Koto

, Tanahashi

, Fukuuchi

, Takagi

, Morinaga

(1999) Pathologic findings of silent hyperintense white matter lesions on MRI. J Neurol Sci 167, 127–131.

43.

Fazekas

, Kleinert

, Offenbacher

, Payer

, Schmidt

, Kleinert

, Radner

, Lechner

(1991) The morphologic correlate of incidental punctate white matter hyperintensities on MR images. AJNR Am J Neuroradiol 12, 915–921.

44.

Braffman

, Zimmerman

, Trojanowski

, Gonatas

, Hickey

, Schlaepfer

(1988) Brain MR: Pathologic correlation with gross and histopathology. 2. Hyperintense white-matter foci in the elderly. AJR Am J Roentgenol 151, 559–566.

45.

Moher

, Liberati

, Tetzlaff

, Altman

, PRISMA Group (2009) Preferred reporting items for systematic reviews and meta-analyses: The PRISMA Statement. Open Med 3, e123–e130.

46.

Rosner

(2016) Fundamentals of Biostatistics. USA.

47.

Silbert

, Quinn

, Moore

, Corbridge

, Ball

, Murdoch

, Sexton

, Kaye

(2003) Changes in premorbid brain volume predict Alzheimer’s disease pathology. Neurology 61, 487–492.

48.

Josephs

, Whitwell

, Ahmed

, Shiung

, Weigand

, Knopman

, Boeve

, Parisi

, Petersen

, Dickson

, Jack

Jr (2008) Beta-amyloid burden is not associated with rates of brain atrophy. Ann Neurol 63, 204–212.

49.

Erten-Lyons

, Dodge

, Woltjer

, Silbert

, Howieson

, Kramer

, Kaye

(2013) Neuropathologic basis of age-associated brain atrophy. JAMA Neurol 70, 616–622.

50.

Erten-Lyons

, Woltjer

, Kaye

, Mattek

, Dodge

, Green

, Tran

, Howieson

, Wild

, Silbert

(2013) Neuropathologic basis of white matter hyperintensity accumulation with advanced age. Neurology 81, 977–983.

51.

Jack

Jr , Dickson

, Parisi

, Xu

, Cha

, O’Brien

, Edland

, Smith

, Boeve

, Tangalos

, Kokmen

, Petersen

(2002) Antemortem MRI findings correlate with hippocampal neuropathology in typical aging and dementia. Neurology 58, 750–757.

52.

Zarow

, Vinters

, Ellis

, Weiner

, Mungas

, White

, Chui

(2005) Correlates of hippocampal neuron number in Alzheimer’s disease and ischemic vascular dementia. Ann Neurol 57, 896–903.

53.

Jagust

, Zheng

, Harvey

, Mack

, Vinters

, Weiner

, Ellis

, Zarow

, Mungas

, Reed

, Kramer

, Schuff

, DeCarli

, Chui

(2008) Neuropathological basis of magnetic resonance images in aging and dementia. Ann Neurol 63, 72–80.

54.

Vemuri

, Whitwell

, Kantarci

, Josephs

, Parisi

, Shiung

, Knopman

, Boeve

, Petersen

, Dickson

, Jack

Jr (2008) Antemortem MRI based STructural Abnormality iNDex (STAND)-scores correlate with postmortem Braak neurofibrillary tangle stage. Neuroimage 42, 559–567.

55.

Whitwell

, Josephs

, Murray

, Kantarci

, Przybelski

, Weigand

, Vemuri

, Senjem

, Parisi

, Knopman

, Boeve

, Petersen

, Dickson

, Jack

Jr (2008) MRI correlates of neurofibrillary tangle pathology at autopsy: A voxel-based morphometry study. Neurology 71, 743–749.

56.

Burton

, Barber

, Mukaetova-Ladinska

, Robson

, Perry

, Jaros

, Kalaria

, O’Brien

(2009) Medial temporal lobe atrophy on MRI differentiates Alzheimer’s disease from dementia with Lewy bodies and vascular cognitive impairment: A prospective study with pathological verification of diagnosis. Brain 132, 195–203.

57.

Erten-Lyons

, Woltjer

, Dodge

, Nixon

, Vorobik

, Calvert

, Leahy

, Montine

, Kaye

(2009) Factors associated with resistance to dementia despite high Alzheimer disease pathology. Neurology 72, 354–360.

58.

Kantarci

, Lowe

, Boeve

, Weigand

, Senjem

, Przybelski

, Dickson

, Parisi

, Knopman

, Smith

, Ferman

, Petersen

, Jack

Jr (2012) Multimodality imaging characteristics of dementia with Lewy bodies. Neurobiol Aging 33, 2091–2105.

59.

Moghekar

, Kraut

, Elkins

, Troncoso

, Zonderman

, Resnick

, O’Brien

(2012) Cerebral white matter disease is associated with Alzheimer pathology in a prospective cohort. Alzheimers Dement 8, S71–S77.

60.

Whitwell

, Dickson

, Murray

, Weigand

, Tosakulwong

, Senjem

, Knopman

, Boeve

, Parisi

, Petersen

, Jack

Jr , Josephs

(2012) Neuroimaging correlates of pathologically defined subtypes of Alzheimer’s disease: A case-control study. Lancet Neurol 11, 868–877.

61.

Toledo

, Cairns

, Da

, Chen

, Carter

, Fleisher

, Householder

, Ayutyanont

, Roontiva

, Bauer

, Eisen

, Shaw

, Davatzikos

, Weiner

, Reiman

, Morris

, Trojanowski

, Alzheimer’s Disease Neuroimaging Initiative (2013) Clinical and multimodal biomarker correlates of ADNI neuropathological findings. Acta Neuropathol Commun 1, 65.

62.

Kaur

, Himali

, Seshadri

, Beiser

, Au

, McKee

, Auerbach

, Wolf

, DeCarli

(2014) Association between neuropathology and brain volume in the Framingham Heart Study. Alzheimer Dis Assoc Disord 28, 219–225.

63.

Raman

, Preboske

, Przybelski

, Gunter

, Senjem

, Vemuri

, Murphy

, Murray

, Boeve

, Knopman

, Petersen

, Parisi

, Dickson

, Jack

Jr , Kantarci

(2014) Antemortem MRI findings associated with microinfarcts at autopsy. Neurology 82, 1951–1958.

64.

Burton

, Mukaetova-Ladinska

, Perry

, Jaros

, Barber

, O’Brien

(2012) Quantitative neurodegenerative pathology does not explain the degree of hippocampal atrophy on MRI in degenerative dementia. Int J Geriatr Psychiatry 27, 1267–1274.

65.

Shim

, Yang

, Roe

, Coats

, Benzinger

, Xiong

, Galvin

, Cairns

, Morris

(2015) Pathological correlates of white matter hyperintensities on magnetic resonance imaging. Dement Geriatr Cogn Disord 39, 92–104.

66.

Josephs

, Whitwell

, Knopman

, Hu

, Stroh

, Baker

, Rademakers

, Boeve

, Parisi

, Smith

, Ivnik

, Petersen

, Jack

Jr , Dickson

(2008) Abnormal TDP-43 immunoreactivity in AD modifies clinicopathologic and radiologic phenotype. Neurology 70, 1850–1857.

67.

Brilliant

, Hughes

, Anderson

, Ghobrial

, Elble

(1995) Rarefied white matter in patients with Alzheimer disease. Alzheimer Dis Assoc Disord 9, 39–46.

68.

Csernansky

, Hamstra

, Wang

, McKeel

, Price

, Gado

, Morris

(2004) Correlations between antemortem hippocampal volume and postmortem neuropathology in AD subjects. Alzheimer Dis Assoc Disord 18, 190–195.

69.

van Veluw

, Zwanenburg

, Engelen-Lee

, Spliet

, Hendrikse

, Luijten

, Biessels

(2013) In vivo detection of cerebral cortical microinfarcts with high-resolution 7T MRI. J Cereb Blood Flow Metab 33, 322–329.

70.

Mirra

, Heyman

, McKeel

, Sumi

, Crain

, Brownlee

, Vogel

, Hughes

, van Belle

, Berg

(1991) The Consortium to Establish a Registry for Alzheimer’s Disease (CERAD). Part II. Standardization of the neuropathologic assessment of Alzheimer’s disease. Neurology 41, 479–486.

71.

Hyman

, Trojanowski

(1997) Consensus recommendations for the postmortem diagnosis of Alzheimer disease from the National Institute on Aging and the Reagan Institute Working Group on diagnostic criteria for the neuropathological assessment of Alzheimer disease. J Neuropathol Exp Neurol 56, 1095–1097.

72.

Braak

, Braak

(1995) Staging of Alzheimer’s disease-related neurofibrillary changes. Neurobiol Aging 16, 271-278; discussion 278–284.

73.

Amador-Ortiz

, Lin

, Ahmed

, Personett

, Davies

, Duara

, Graff-Radford

, Hutton

, Dickson

(2007) TDP-43 immunoreactivity in hippocampal sclerosis and Alzheimer’s disease. Ann Neurol 61, 435–445.

74.

Bigio

, Mishra

, Hatanpaa

, White

3rd , Johnson

, Rademaker

, Weitner

, Deng

, Dubner

, Weintraub

, Mesulam

(2010) TDP-43 pathology in primary progressive aphasia and frontotemporal dementia with pathologic Alzheimer disease. Acta Neuropathol 120, 43–54.

75.

Davidson

, Raby

, Foulds

, Robinson

, Thompson

, Sikkink

, Yusuf

, Amin

, DuPlessis

, Troakes

, Al-Sarraj

, Sloan

, Esiri

, Prasher

, Allsop

, Neary

, Pickering-Brown

, Snowden

, Mann

(2011) TDP-43 pathological changes in early onset familial and sporadic Alzheimer’s disease, late onset Alzheimer’s disease and Down’s syndrome: Association with age, hippocampal sclerosis and clinical phenotype. Acta Neuropathol 122, 703–713.

76.

Uryu

, Nakashima-Yasuda

, Forman

, Kwong

, Clark

, Grossman

, Miller

, Kretzschmar

, Lee

, Trojanowski

, Neumann

(2008) Concomitant TAR-DNA-binding protein 43 pathology is present in Alzheimer disease and corticobasal degeneration but not in other tauopathies. J Neuropathol Exp Neurol 67, 555–564.

77.

Jack

Jr , Knopman

, Jagust

, Petersen

, Weiner

, Aisen

, Shaw

, Vemuri

, Wiste

, Weigand

, Lesnick

, Pankratz

, Donohue

, Trojanowski

(2013) Tracking pathophysiological processes in Alzheimer’s disease: An updated hypothetical model of dynamic biomarkers. Lancet Neurol 12, 207–216.

78.

Osborn

, Salzman

, Jhaveri

, Barkovich

(2016) Lacunar infarction. In Diagnostic Imaging: Brain, 3rd Edition, Osborn

, Salzman

, Jhaveri

, Barkovich

eds. Elsevier, Philadelphia, PA.

79.

Whitwell

, Weigand

, Shiung

, Boeve

, Ferman

, Smith

, Knopman

, Petersen

, Benarroch

, Josephs

, Jack

Jr (2007) Focal atrophy in dementia with Lewy bodies on MRI: A distinct pattern from Alzheimer’s disease. Brain 130, 708–719.

80.

Vemuri

, Simon

, Kantarci

, Whitwell

, Senjem

, Przybelski

, Gunter

, Josephs

, Knopman

, Boeve

, Ferman

, Dickson

, Parisi

, Petersen

, Jack

Jr (2011) Antemortem differential diagnosis of dementia pathology using structural MRI: Differential-STAND. Neuroimage 55, 522–531.

81.

Duchesne

, Valdivia

, Mouiha

, Robitaille

, Alzheimer’s Disease Neuroimaging I (2015) Single time point high-dimensional morphometry in Alzheimer’s disease: Group statistics on longitudinally acquired data. Neurobiol Aging 36(Suppl 1), S11–S22.

82.

Coupe

, Fonov

, Bernard

, Zandifar

, Eskildsen

, Helmer

, Manjon

, Amieva

, Dartigues

, Allard

, Catheline

, Collins

, Alzheimer’s Disease Neuroimaging I (2015) Detection of Alzheimer’s disease signature in MR images seven years before conversion to dementia: Toward an early individual prognosis. Hum Brain Mapp 36, 4758–4770.

83.

Kim

, Lee

, Seo

, Kim

, Seong

, Ye

, Cho

, Noh

, Kim

, Yoon

, Oh

, Kim

, Choe

, Lee

, Kim

, Hwang

, Jeong

, Na

(2015) Hippocampal volume and shape in pure subcortical vascular dementia. Neurobiol Aging 36, 485–491.

84.

Yun

, White

(2010) MRI volumetric analysis techniques, including hippocampus extraction, based on data from the Honolulu Asia Aging Study (HAAS). Ethn Dis 20 (Suppl 1), S1-104–6.