Acta Neuropathologica

, Volume 126, Issue 3, pp 365–384

Non-Alzheimer neurodegenerative pathologies and their combinations are more frequent than commonly believed in the elderly brain: a community-based autopsy series

Authors

    • Institute of NeurologyMedical University Vienna
  • Ivan Milenkovic
    • Institute of NeurologyMedical University Vienna
  • Adelheid Wöhrer
    • Institute of NeurologyMedical University Vienna
  • Romana Höftberger
    • Institute of NeurologyMedical University Vienna
  • Ellen Gelpi
    • Institute of NeurologyMedical University Vienna
    • Neurological Tissue Bank of the Biobanc-Hospital Clinic-Institut d’Investigacions Biomediques August Pi i Sunyer (IDIBAPS)
  • Christine Haberler
    • Institute of NeurologyMedical University Vienna
  • Selma Hönigschnabl
    • Institute of PathologyDanube Hospital
  • Angelika Reiner-Concin
    • Institute of PathologyDanube Hospital
  • Harald Heinzl
    • Center for Medical Statistics, Informatics, and Intelligent SystemsMedical University Vienna
  • Susanne Jungwirth
    • Ludwig Boltzmann Institute of Ageing Research
  • Wolfgang Krampla
    • Department of RadiologyDanube Hospital
  • Peter Fischer
    • Psychiatric DepartmentMedical Research Society Vienna, D.C., Danube Hospital
  • Herbert Budka
    • Institute of NeurologyMedical University Vienna
    • Institute of NeuropathologyUniversity Hospital Zurich
Original Paper

DOI: 10.1007/s00401-013-1157-y

Cite this article as:
Kovacs, G.G., Milenkovic, I., Wöhrer, A. et al. Acta Neuropathol (2013) 126: 365. doi:10.1007/s00401-013-1157-y

Abstract

Neurodegenerative diseases are characterised by neuronal loss and cerebral deposition of proteins with altered physicochemical properties. The major proteins are amyloid-β (Aβ), tau, α-synuclein, and TDP-43. Although neuropathological studies on elderly individuals have emphasised the importance of mixed pathologies, there have been few observations on the full spectrum of proteinopathies in the ageing brain. During a community-based study we performed comprehensive mapping of neurodegeneration-related proteins and vascular pathology in the brains of 233 individuals (age at death 77–87; 73 examined clinically in detail). While all brains (from individuals with and without dementia) showed some degree of neurofibrillary degeneration, Aβ deposits were observed only in 160 (68.7 %). Further pathologies included α-synucleinopathies (24.9 %), non-Alzheimer tauopathies (23.2 %; including novel forms), TDP-43 proteinopathy (13.3 %), vascular lesions (48.9 %), and others (15.1 %; inflammation, metabolic encephalopathy, and tumours). TDP-43 proteinopathy correlated with hippocampal sclerosis (p < 0.001) and Alzheimer-related pathology (CERAD score and Braak and Braak stages, p = 0.001). The presence of one specific variable (cerebral amyloid angiopathy, Aβ parenchymal deposits, TDP-43 proteinopathy, α-synucleinopathy, vascular lesions, non-Alzheimer type tauopathy) did not increase the probability of the co-occurrence of others (p = 0.24). The number of observed pathologies correlated with AD-neuropathologic change (p < 0.0001). In addition to AD-neuropathologic change, tauopathies associated well with dementia, while TDP-43 pathology and α-synucleinopathy showed strong effects but lost significance when evaluated together with AD-neuropathologic change. Non-AD neurodegenerative pathologies and their combinations have been underestimated, but are frequent in reality as demonstrated here. This should be considered in diagnostic evaluation of biomarkers, and for better clinical stratification of patients.

Keywords

Alzheimer’s diseaseAmyloid-βα-SynucleinTauTDP-43Community-basedProtein-coding

Introduction

The increasing number of elderly people has a major impact on the prevalence of age-related neurodegenerative pathologies; an effective therapy is greatly expected. Since the original paper by Alois Alzheimer [7], silver staining methods revealed that in the majority of elderly individuals with dementia, extracellular plaques, later found to be composed of amyloid-beta (Aβ) peptide (for review see [59]), and intracellular neurofibrillary tangles, later found to be composed of tau protein [34, 93], can be detected in the brain. Accordingly, the term Alzheimer’s disease (AD) was regularly used in the context of a single entity defined by plaques and tangles. Meanwhile the original case reported by Alzheimer turned out to be associated with a PSEN1 mutation with aggressive histopathology [67].

AD belongs to the group of neurodegenerative diseases that are characterised by progressive loss of neurons and deposition in the brain of different proteins with altered physicochemical properties [48]. The most frequently involved proteins in sporadic neurodegenerative diseases are Aβ, tau, α-synuclein, and TDP-43. Tauopathies comprise progressive supranuclear palsy (PSP), corticobasal degeneration (CBD), Pick’s disease, argyrophilic grain disease (AGD), tangle-only dementia (TOD), and recently described forms characterised by globular glial inclusions [2, 49, 51]. Moreover, a variety of astrocytic tau immunoreactivities or complex constellations of tau pathologies have been described in ageing brains [12, 24, 52, 56, 77, 82]. Tau pathology in AD is characterised by neurofibrillary tangles that follow stages described by Braak and Braak (BB) [14, 15]. α-Synucleinopathies comprise Lewy-body disease (including a subgroup with Lewy-related pathology predominating in the amygdala) and multiple system atrophy (MSA) [35]. Cerebral TDP-43 pathology can either associate with frontotemporal lobar degeneration (called FTLD-TDP), with or without motor neuron involvement, or may be restricted to the limbic system, as seen mainly associated with other neurodegenerative diseases [48, 58]. This emphasises the need for extensive immunohistochemical studies following harmonised protocols [46], including its application for the staging of neurofibrillary degeneration [3].

Neuropathological studies have emphasised additional pathological alterations (mixed-type dementia), defined mainly as vascular lesions and/or α-synuclein immunoreactive Lewy-bodies, to be seen in the brains of demented elderly individuals [41, 42, 72, 80, 87]. Mixed pathologies were described in differing cohorts with distinct strategies for recruiting and selecting brain samples and not only in AD, but also in other brain disorders [10, 47]. However, many elderly brains show non-AD pathology expanding beyond concomitant vascular or Lewy-body-related pathology [70]. The concept of multiple proteinopathies was further emphasised [48], however, there are few observations on the spectrum of proteinopathies in the ageing human brain, and particularly how these associate with cognitive decline. Moreover, such descriptions used different selection criteria of examinees. In 2010, we presented our neuropathological observations on 169 brains collected in a community-based study (XVIIth International Congress of Neuropathology), and emphasised the significance of multiple protein deposits in individuals with and without dementia [50]. In parallel, the spectrum of evaluated pathologies became expanded and included TDP-43 proteinopathy and additional tau pathology in a large autopsy cohort [68]. Soon “multiprotein deposits” have been reported in a Brain bank collection [23] and TDP-43 proteinopathy was also evaluated in a population-based cohort of 108 individuals aged ≥90 years, in addition to α-synucleinopathies, Aβ and tau burden [74].

The concept of mixed pathologies expanded and a multidimensional approach to diagnosing dementia was proposed [22]. Understanding the spectrum of protein deposits in the brain of demented but also of non-demented is of pivotal importance, since modifications of neurodegeneration-related proteins can be potentially detected in body fluids [48], and their evaluation may improve clinical diagnostic procedures and stratification of patients for therapeutic trials [46]. Here, we present our observations on the spectrum of proteinopathies in the brains of 233 consecutive elderly individuals with and without dementia, collected within a prospective community-based study.

Materials and methods

Acquisition of cases and clinical examinations

Cases were examined within the frame of the Vienna Trans-Danube Aging (VITA) study [26]. Since 2000, the VITA study follows longitudinally a community-based cohort of every resident of the Vienna area on the left shore of the River Danube (districts 21 and 22) born between May 1925 and June 1926. As described by Fischer et al. [27], there was nothing hinting a selection bias concerning participation. Among 1,920 registrants in the official voting registry, 1,505 individuals could be contacted. From these, 697 agreed to participate in the clinical arm of the VITA study. Clinical investigations took 6 h and included blood sampling and neuropsychological testing (CERAD, Fuld Object Memory Test, Trail Making Test, Tokentest, Aachener Aphasia Test); psychiatric examinations (Clinical Dementia Rating, Short Geriatric Depression Scale, Hamilton Depression Scale, State/Trait Anxiety Inventory, Physical Self-Maintenance Scale and Instrumental Activities of Daily Living, Life-Event Scale, DSM-IV criteria for major depression, and for Alzheimer dementia, clinical diagnostic criteria for frontotemporal dementia); and neurological investigations (for details see reference [26]. Cranial MRI examination was performed whenever possible (exceptions were claustrophobia or metal subjects in the body) at each investigation. These clinical examinations were performed at the following timepoints: baseline examination at 75–76 years of age, first follow-up after 30 months (78–79 years), second follow-up after 60 months (80–81 years) and third follow-up after 90 months (82–83 years). Consecutive individuals of the birth cohort (n = 1920) who died in the Danube Hospital, Vienna, underwent neuropathological examination irrespective of the presence of neuropsychiatric symptoms. A subgroup of these patients had participated in the clinical assessment of the VITA study (see below). The local Ethical Committee approved both the clinical and the neuropathological studies.

Neuropathology

Formalin fixed, paraffin-embedded tissue blocks (2.5 × 2.0 cm) were evaluated. Tissue blocks comprised frontal, cingular, temporal, parietal, occipital cortex and white matter, anterior and posterior portion of hippocampus, caudate nucleus, accumbens nucleus, putamen, globus pallidus, thalamus, mesencephalon, pons, medulla oblongata, cerebellar anterior vermis and cerebellar hemisphere and dentate nucleus. In addition to hematoxylin and eosin, Luxol fast blue/nuclear fast red, Bielschowsky and Gallyas stainings, the following monoclonal antibodies were used for immunohistochemistry: anti-tau AT8 (pS202, 1:200; Pierce Biotechnology, Rockford, IL, USA), anti-4R tau (RD4, 1:200, Upstate, Charlottesville, VA, USA), anti-3R tau (RD3, 1:2,000, Upstate), anti-phospho-TDP-43 (pS409/410, 1:2,000, Cosmo Bio, Tokyo, Japan), anti-TDP-43 (1:2,000, Abnova, Taipei, Taiwan), anti-ubiquitin (1:50,000, Millipore, Temecula, CA, USA), anti-α-synuclein (1:10,000, clone 4D6, Signet, Dedham, MA, USA; and clone 5G4, 1:2,000, Roboscreen, Leipzig, Germany [55], anti-Aβ (1:50, clone 6F/3D, Dako, Glostrup, Denmark), anti-p62 (1:500, BD Bisociences, USA). In addition, in selected cases we applied polyclonal anti-FUS (1:1,000, Sigma, St. Louis, MO, USA). The DAKO EnVision© detection kit, peroxidase/DAB, rabbit/mouse (Dako) was used for visualisation of antibody reactions.

Staining strategy was as follows: all tissue blocks were stained for H&E and Luxol fast blue/nuclear fast red. Frontal, parietal, temporal, hippocampus, and occipital block were stained for Bielschowsky, and in selected cases amygdala and hippocampus for Gallyas silver staining. Immunostaining for AT8 (tau) was performed on the frontal, cingular, temporal, hippocampus, occipital, basal ganglia, thalamus, amygdala, brainstem and cerebellum/dentate nucleus tissue blocks; α-synuclein was performed on the amygdala and brainstem tissue blocks and in case of positivity, additionally on the frontal, cingular, temporal, hippocampus, and basal ganglia tissue blocks. Frontal, parietal, temporal, and occipital cortices, hippocampus, basal ganglia and cerebellum were immunostained for anti-Aβ. Phospho-TDP-43 was immunostained in the frontal, hippocampus, amygdala and basal ganglia blocks and in selected cases brainstem, thalamus, and further cortical regions. Further anti-Tau antibodies were applied in the frontal cortex, hippocampus, basal ganglia, and amygdala in selected cases [52]. Anti-ubiquitin and p62 immunoreactivities were evaluated in the frontal, hippocampus, and amygdala tissue blocks and in selected cases further anatomical regions. In selected cases we performed immunostaining for FUS in the hippocampus, cortical areas, basal ganglia, thalamus, and brainstem. All brains collected before 2008 were re-evaluated for TDP-43 pathology and all brains collected before 2010 were screened for α-synuclein pathology using antibody 5G4 that selectively recognises disease-associated α-synuclein [55]. Two or more certified neuropathologists evaluated all cases using a multi-headed microscope.

Examined variables

In addition to gender, the following variables were collected in a database. Age at baseline, at examinations, at death, at the time of dementia diagnosis; type and time of participation, education, clinical dementia diagnosis (possible, probable AD, vascular dementia, Lewy-body dementia), presence of mild cognitive impairment (including type, i.e. amnestic/non-amnestic single/multiple) or depression, cranial MRI examination (time and presence or lack of moderate/severe temporal lobe atrophy at least on one side, see [26, 28]). Furthermore, presence or lack of α-synuclein, TDP-43 pathology, tauopathy (further stratified as AGD, CBD, PSP, TOD, unclassifiable and combined forms, following criteria summarised in a recent review by [49]), Aβ parenchymal deposits, type 1 and 2 cerebral amyloid angiopathy (CAA) [88], neurofibrillary degeneration, vascular lesion, hippocampus sclerosis, hippocampus microinfarct, neoplasia, inflammation, metabolic encephalopathy, and “other” disease were recorded. Grouping comprises types of α-synuclein pathology (stages 1–6, [16]; amygdala-predominant form, atypical distribution of Lewy-related pathology, MSA); TDP-43 pathology (limbic, widespread, typical FTLD-TDP); CERAD criteria (no, possible, probable, definite, score A–C; [62]) neurofibrillary degeneration [3, 14, 15], stages of AGD (I–III, [76]), vascular pathology (single microinfarcts/lacunes, strategic or territorial infarcts, multiinfarct state, status cribrosus, panischemic/hypoxic brain damage). The level of AD-neuropathologic change was evaluated first following the NIA-Reagan criteria [36], but we re-evaluated all cases following the new guidelines for the evaluation of AD-related changes and other pathologies [64]. Combinations of α-synucleinopathy, TDP-43 proteinopathy, tauopathy, Aβ parenchymal deposits, CAA, and vascular lesion entered into analysis as “number of pathologies (1–6)”.

Statistical methods

Chi-square test, Fisher’s exact test and Spearman correlation coefficient were used to evaluate association between variables of interest. Chi-square goodness-of-fit test (Monte Carlo approximation) was used to evaluate whether several pathological variables co-occur or they are independent from each other. SAS version 9.3 and IBM SPSS Statistics Version 20 were used for statistical analysis. Logistic regression models were used to generate odds ratios (OR) and 95 % confidence intervals (CI), where the presence of dementia was the dependent variable and various pathological alterations (see below) were the independent variables. A significance level of 0.05 was used.

Results

Summary of examined individuals

Brains of 233 patients out of the total cohort of 1,920 (12.1 %; age at death, 77–87 years) were evaluated systematically using neuropathological methods. Of these, 73 (31.3 %) had been examined clinically in detail (10.5 % of the 697 involved in the clinical VITA study) at baseline and followed regularly: 58 were examined at 30 months, 42 at 60 months, and 17 at 90 months. Reasons for the lack of repeated examinations were death or refusal (severe disease or lack of interest). Dementia had been clinically diagnosed in 22 out of 73 (30.1 %; 14 possible AD and 8 probable AD). During the follow-up examinations, MCI had been documented in 12, while depression had been diagnosed in 32 out of 73 (from these 17 before the baseline examination). In further 20 patients, who did not belong to the clinical VITA follow-up study, cognitive decline had been reported prior to death (Fig. 1).
https://static-content.springer.com/image/art%3A10.1007%2Fs00401-013-1157-y/MediaObjects/401_2013_1157_Fig1_HTML.gif
Fig. 1

Stratification of the VITA study and examined individuals

Distribution of neuropathological alterations

All 233 brains (from individuals with and without dementia) showed some degree of neurofibrillary degeneration. While the majority (62 %, n = 144) had BB stage I (n = 36) or II (n = 108) (Fig. 2a), only 15 individuals (6.4 %) (7 with BB stage I and 8 with stage II) did not show any other pathology (including lack of Aβ parenchymal deposits). Further 15 individuals, with stage I or II, showed only minor vascular pathology. On the other hand, the brains of 28 individuals showed end-stage neurofibrillary degeneration (BB V or VI) and high CERAD plaque scores. In all 28 except 10 individuals (35.7 %) where CAA was the only additional finding (i.e. represents “pure” AD), there were multiple further pathologies present (Fig. 2a).
https://static-content.springer.com/image/art%3A10.1007%2Fs00401-013-1157-y/MediaObjects/401_2013_1157_Fig2_HTML.gif
Fig. 2

Distribution of pathological alterations. a Distribution of cases (n = 233) with neurofibrillary degeneration (NFD). Only 15 cases did not show any other type of pathological alterations. Thus 218 cases showed NFD “plus” pathology. 160 cases had Aβ plaques; the distribution of CERAD score is indicated in the lower bar. b Distribution of all types of neuropathological alterations. Log scale is used to demonstrate better the distribution of alterations with lower percentage. c–e Distribution of subtypes of α-synucleinopathies (c), tauopathies (dasterisk at PSP indicates that 5 further cases in the “unclassifiable” group were considered as anatomically restricted forms of PSP) and vascular pathology (e). B&B Braak and Braak, CAA cerebral amyloid angiopathy, CBD corticobasal degeneration, MSA multiple system atrophy, PSP progressive supranuclear palsy, TOD tangle-only dementia

Aβ Parenchymal deposits were observed in 160 out of 233 (68.7 %). Brains of 10 individuals from these showed only occasional small parenchymal deposits. CAA was observed in 113 (48.5 %), including 22 cases with capillary CAA (9.4 %). Further pathologies (Fig. 2b) included α-synucleinopathies (Lewy-related pathology Braak stages 1–6 or amygdala variant, or MSA) (Fig. 2c), tauopathies (classified as AGD, CBD, PSP, TOD, mixed forms or unclassifiable forms, see below) (Fig. 2d), TDP-43 proteinopathy (with and without hippocampal sclerosis, limbic predominant or diffuse forms or compatible with FTLD-TDP), vascular lesions (bleeding, panischemic encephalopathy, status cribrosus, multiinfarct state, strategic or territorial infarcts, or single microinfarcts) (Fig. 2e), inflammation, metabolic encephalopathy, and neoplastic disorders (primary and metastatic tumours) (Fig. 2b). A single case showed cerebellar microdysgenesis and another intranuclear inclusion body disease (combined with tauopathy, see below).

Association of multiple pathologies

We evaluated the co-occurrence of specific pathological alterations (Fisher exact test and Spearman correlation) in the cohort of 233 individuals. We observed significant association of higher CERAD score, higher phase of Aβ deposition [89], and presence of CAA (either type 1 or type 2 [88]) with higher BB stages (Fisher exact test, p < 0.02, for all) as well as presence of Aβ parenchymal deposits with CAA (Fisher exact test, p < 0.001, Spearman correlation, R = 0.33). TDP-43 proteinopathy correlated well with the appearance of hippocampal sclerosis (Fisher exact test, p < 0.001, Spearman correlation, R = 0.48) and higher AD-related pathology (CERAD score: Fisher exact test, p = 0.001, Spearman correlation, R = 0.25; or BB stages: Fisher exact test, p = 0.001, Spearman correlation, R = 0.26). Men were more affected by vascular lesions (Fisher exact test, p = 0.035, Spearman correlation, R = 0.14); otherwise, gender did not influence other pathological variables (including CERAD score or BB stages). Only type 1 (capillary) but not type 2 CAA showed association with hippocampal microinfarcts (Fisher exact test, p = 0.018, Spearman correlation, R = 0.18).

In order to evaluate whether six specific pathological alterations (CAA, Aβ parenchymal deposits, tauopathy, TDP-proteinopathy, α-synucleinopathy, vascular lesions) tend to co-occur, we compared the observed proportions of the number of combined pathologies (possible values 0–6) with the expected proportions when stochastic independence between them is assumed. This approach did not confirm that the presence of one specific pathological variable would increase the probability of further ones to combine (p = 0.24). The number of observed pathologies significantly correlated with the BB stages (Spearman correlation, R = 0.426, p < 0.0001), the CERAD score (R = 0.530, p < 0.0001), and presence of AGD (R = 0.25, p < 0.01), but not with the age (at death) of the individuals (R = 0.1, p = 0.096). Thus, the higher the BB stage of neurofibrillary degeneration and the amount of senile plaques, the more probable the presence of further pathological alterations. All together the six pathological variables (other than neurofibrillary degeneration) were represented in high number of variations (Fig. 3).
https://static-content.springer.com/image/art%3A10.1007%2Fs00401-013-1157-y/MediaObjects/401_2013_1157_Fig3_HTML.gif
Fig. 3

Overview of the types and frequency (%) of combinations of pathological variables (indicated with different colours) in the total cohort and in the clinically assessed cases with and without dementia. The aim of this figure is to show the wide spectrum of possible combinations. On the right side the associated Braak & Braak (BB) stages are indicated. BB stages associate with a variable spectrum of other proteinopathies without any predominant “signature” pattern

Clinicopathological and radiological aspects

Next we assessed the association of different pathological variables with dementia in the clinically examined cohort (n = 73). As single variable, presence of TDP-43 (Fisher exact test, p = 0.012, Spearman correlation, R = 0.31) and some kind of α-synuclein pathology (Fisher exact test, p = 0.017, Spearman correlation, R = 0.30), as well as tauopathy (Fisher exact test, p = 0.023, Spearman correlation, R = 0.28), the BB stages of neurofibrillary degeneration, CERAD score, and the level of AD-neuropathologic change [64] (for all Spearman correlation, p < 0.001, R = 0.47–0.50), and the number of pathologies (Spearman correlation, p < 0.001, R = 0.43) associated with dementia. Presence of vascular lesions (p = 0.140) or type 2 CAA alone (p > 0.15) did not associate with dementia. Presence of AGD showed tendency to associate with dementia (p = 0.071), while type 1 (capillary) CAA correlated with cognitive decline (Fisher exact test, p = 0.001, Spearman correlation, R = 0.39).

Logistic regression models were used to assess the univariate and multivariable effect of two variables that are related to the morphological diagnosis of AD (CERAD, BB) and six further pathological variables (presence of tauopathy, TDP-43, α-synucleinopathy, Aβ parenchymal deposits, CAA, vascular lesion) on the presence of dementia (Table 1). As single variables, presence of tauopathy, TDP-43, α-synucleinopathy, Aβ-plaques as well as the CERAD score and stage of neurofibrillary degeneration (BB stages grouped as I–II, III–IV, and V–VI) were associated with dementia (left column of Table 1). From the six pathologies other than AD-type, tauopathy, TDP-43 and α-synucleinopathy pathology were, independently from each other, associated with dementia (middle column of Table 1). Tauopathy remained statistically significant when the AD-related variables (CERAD, BB) were also added to the model (right column of Table 1). Although not significant, TDP-43 and α-synucleinopathy showed strong effects. The non-significant effects of CERAD and BB in this model are probably due to their strong correlation (multicollinearity effect).
Table 1

Logistic regression models for the assessment of the univariate and multivariable effect of two variables that are related to the morphological diagnosis of AD (CERAD, Braak and Braak) and six further pathological variables (presence of tauopathy, TDP-43, α-synucleinopathy, Aβ-plaques, CAA, vascular lesion), on the presence of dementia in the clinically examined group (n = 73)

Variables

Univariate logistic regression models

Multiple logistic regression model (without B&B stage and CERAD score)

Multiple logistic regression model

Odds ratio (95 % CI)

p value

Odds ratio (95 % CI)

p value

Odds ratio (95 % CI)

p value

TDP-43 pathology

7.46 (1.71–32.52)

0.007

9.62 (1.55–59.78)

0.015

5.23 (0.74–37.01)

0.097

α-Syn pathology

3.72 (1.19–11.59)

0.023

4.75 (1.10–20.40)

0.036

3.75 (0.78–17.95)

0.098

Tauopathy

3.5 (1.01–12.06)

0.047

6.16 (1.24–30.40)

0.026

7.92 (1.24–50.20)

0.028

CAA

2.62 (0.94–7.32)

0.065

2.21 (0.55–8.86)

0.263

2.11 (0.44–10.18)

0.349

Aβ parenchymal deposits

2.18 (1.16–4.11)

0.015

1.60 (0.76–3.34)

0.207

0.81 (0.22–3.00)

0.759

Vascular lesions

0.56 (0.20–1.56)

0.275

0.57 (0.14–2.24)

0.428

0.76 (0.16–3.55)

0.733

BB stage

3.70 (1.85–7.40)

0.0001

1.84 (0.35–9.51)

0.467

CERAD score

2.28 (1.43–3.64)

0.001

1.48 (0.29–7.46)

0.631

Bold values indicate statistical significance

BB Braak and Braak, CAA cerebral amyloid angiopathy

Brain MRI scans were reported in 55 individuals. Local atrophy of the medial temporal lobe was assessed qualitatively on a 0–4 scale [28]. We assessed the correlations of variables with moderate or severe medial temporal lobe atrophy. Interval of the last MRI scan and autopsy was also evaluated; for all cases with MRI, median, 2.42; mean ± standard deviation, 3.06 ± 2.45; range, 0.73–11.0; for cases with medial temporal lobe atrophy, 2.09, 2.7 ± 2.01, 1.2–3.61; for cases without medial temporal lobe atrophy, 2.47, 3.2 ± 2.69, 0.73–11.0. High CERAD score (Spearman correlation, R = 0.38, p = 0.003) advanced BB stage (R = 0.45, p < 0.0001), the number of pathologies (R = 0.50, p < 0.0001), as well as the presence of TDP-43 pathology in the medial temporal lobe (R = 0.52, p < 0.0001) correlated well with the presence of moderate or severe medial temporal lobe atrophy. It was more frequent in the group of tauopathies (R = 0.26, p = 0.047; including in AGD or complex tauopathies or combined forms), but we did not observe significance when correlating with the presence of α-synucleinopathies, CAA, vascular lesions, metabolic encephalopathy, or inflammatory disease.

The spectrum of pathologies overlapped between demented and non-demented individuals, however, differences could be noted as well (Fig. 3; Table 2). The non-demented cohort showed low BB stages and CERAD plaque scores. Amygdala-predominant form of Lewy-related α-synucleinopathy was more likely to occur with dementia; furthermore, when tauopathy was present in non-demented cases, this was represented by AGD or mild astrogliopathy in the temporal lobe (see below), in contrast to the demented group of individuals where complex constellations of tau pathology, CBD, or TOD were more frequent.
Table 2

Neuropathological data on the individuals without (n = 51) and with (n = 22) dementia

Pathology

Number of cases

No dementia

Dementia

B&B stages

B&B stages

I–II (n = 39)

III–IV (n = 8)

V–VI (n = 4)

I–II (n = 7)

III–IV (n = 5)

V–VI (n = 10)

TDP-43

 No

36

7

4

6

5

5

 Yes

3

1

0

1

0

5

α-Synuclein

 No

34

7

3

6

2

5

 Yes

5

1

1

1

3

5

  Braak 1–3

2

0

0

0

1

0

  Braak 4–6

3

0

0

1

0

2

  Amygdala variant

0

0

1

0

2

2

  MSA

0

0

0

0

0

1

  Atypical

0

1

0

0

0

0

Tauopathy

 No

35

6

4

4

1

9

 Yes

4

1

0

3

4

1

  AGD

3

0

0

1

2

0

  CBD

0

0

0

1

0

0

  PSP

0

0

0

0

0

0

  TOD

0

0

0

1

0

0

  Unclassifiable

1

1 (+AGD)

0

0

2 (+AGD)

1

CAA

 No

33

4

1

4

4

4

 Yes

6

4

3

3

1

6

Aβ plaques

 No

29

2

0

4

0

0

 Yes

10

6

4

3

5

9

CERAD

 No

31

1

0

4

0

0

 Yes

8

7

4

3

5

10

  Plaques score A

4

1

0

3

1

1

  Plaques score B

4

6

0

0

3

0

  Plaques score C

0

0

4

0

1

9

Vascular pathology

 No

16

3

4

5

4

4

 Yes

23

5

0

2

1

6

  Single microinfarct

10

3

0

1

1

3

  Strategic/territorial

6

1

0

0

0

2

  Multiinfarct

4

1

0

1

0

1

  Status cribrosus

2

0

0

0

0

0

  Panischemic

1

0

0

0

0

0

Mean age at death (years)

81.7

81.8

85.5

82.9

84.0

83.3

Mean time (years) from the last examination or from the date of diagnosis of dementia to death

2.5

3.1

4.7

3.4

5.4

3.6

AGD argyrophilic grain disease, B&B Braak and Braak stages, PSP progressive supranuclear palsy, CBD corticobasal degeneration, MSA multiple system atrophy, TOD tangle-predominant dementia, + with

In the clinically examined study group (n = 73), in four individuals without dementia we observed high CERAD scores and BB stages, and in five further patients reversible clinical symptoms were described. In the four patients without dementia and prominent AD-type pathology, the last clinical assessment before death was at 1.5, 2.2, 6.4 and 8.9 years (Table 3), suggesting that clinical manifestation might have occurred thereafter. Interestingly, frontal lobe symptoms and depression were noted in the patient with 1.5 years examination-free period, and dementia was noted at hospital admission prior to death for the patient with 6.4 years since the last examination. However, for the individual who was examined 2.2 years before death, prominent pathology was evident. Individuals with reversible clinical diagnosis of possible AD showed low BB stages and CERAD scores; however, 3 out of 5 had Lewy-related pathology and vascular lesions. On the other hand, cases with clinically possible AD but with low BB stages and CERAD scores showed other types of pathologies, including vascular lesions, tauopathies, metabolic encephalopathy, or Lewy-related pathology, suggesting that combinations of multiple pathologies are sufficient to reach the threshold for symptomatic cognitive impairment in some individuals.
Table 3

Clinical and pathological data of clinically examined patients with (only cases with CERAD score 0, A or B and stage ≤IV according to Braak and Braak) or without dementia (only cases with high CERAD score or stages V–VI according to Braak and Braak), and with the reversible clinical diagnosis of possible AD

Case

Age at baseline

Age at death

Time from examination or diagnosis (years)

Sex

Examination

TDP-43 Path

Alpha-Synuclein pathology

B&B stage

CERAD score

CAA

Tauopathy

Vascular lesion

Other pathology

Dementia at last VITA examination

Dementia reported prior to death

Depression

No-D-1

75

87

6.4

w

Home

VI

C

+

No

+

+

No-D-2

75

85

2.2

w

Clinics

Amygd

VI

C

No

No-D-3

75

85

8.9

w

Baseline

V

C

+

No

No-D-4

75

85

1.5

m

Clinics

V

C

+

No

a

+

R-Dem-1

75

85

1.98

m

Home

Braak 4

II

0

Singlec

Metabolic

No

+

R-Dem-2

75

86

2.8

m

Clinics

II

B

st-cribr+

No

R-Dem-3

75

83

2.65

w

Clinics

Braak 2

II

0

AGD 3

Multi

MCIb

R-Dem-4

75

83

2.47

w

Clinics

Braak 4

I

0

No

+

R-Dem-5

75

84

0.86

m

Home

I

0

Single

MCIb

Dem-1

75

81

0.64

w

Clinics

I

0

Metabolic

Poss AD

+

+

Dem-2

75

86

5.94

m

Clinics

I

0

CBD

Poss AD

+

Dem-3

75

82

3.72

m

Clinics

I

0

Multi

Poss AD

+

+

Dem-4

75

82

1.2

w

Clinics

+

II

A

+

AGD 3 +

Poss AD

+

Dem-5

75

82

3.74

m

Clinics

II

0

TOD

Poss AD

+

+

Dem-6

76

82

2.87

m

Clinics

Braak 5

II

A

+

Multi

Poss AD

+

Dem-7

76

85

6.06

w

Clinics

II

A

+

Poss AD

+

+

Dem-8

75

83

2.7

w

Clinics

Braak 2

III

B

Single

Prob AD

+

+

Dem-9

75

80

2.12

m

Clinics

Amygd

III

B

+

Poss AD

+

Dem-10

75

86

8.09

w

Clinics

III

B

AGD 2

Poss AD

+

+

Dem-11

75

87

8.72

m

Clinics

IV

A

+

AGD 2

Poss AD

+

Dem-12

75

84

5.32

m

Clinics

Amygd

IV

C

AGD 2 +

Prob AD

+

No-D no dementia, R-Dem reversible clinical diagnosis of possible AD, Dem dementia: AGD argyrophilic grain disease (stages according to [76]); CBD corticobasal degeneration, TOD tangle-only dementia, vascular lesion single infarct, st-cribr status cribrosus in the basal ganglia, multi multiple infarcts. + indicates additional pathology: for AGD this is unclassifiable tauopathy, for st-cribr this means additional myelin loosening in the periventricular white matter

aFrontal lobe symptoms prior to death

bNon-amnestic single mild cognitive impairment

cMicroinfarct in the amygdala region

Morphological characterisation of proteinopathies (α-synuclein, TDP-43, tau)

α-Synucleinopathies comprised Lewy-related pathology Braak stages 1–6, amygdala-predominant form and MSA (Fig. 2c). In the majority of cases Lewy-related pathology was compatible with the Braak stages, except for a case where alpha-synuclein immunoreactive Lewy-related pathology was seen unilaterally and only in the substantia nigra. This was associated with a complex constellation of tau pathology as reported previously [52]. Amygdala-predominant form of Lewy-related pathology was seen in ten individuals. MSA with predominance of glial cytoplasmic inclusions in the cerebellar white matter was associated with definitive AD in one (BB stage VI, CERAD score C, Aβ deposition phase 5) and with PSP and AGD in another individual.

TDP-43 proteinopathy comprised cases of the limbic predominant form (28/31) with or without hippocampal sclerosis, but also cases with widespread anatomical appearance of predominantly neuronal TDP-43 immunopositivity (2/31). In addition to hippocampus sclerosis, TDP-43 immunoreactivity significantly associated with tauopathies that were not clearly classifiable (see below). Widespread TDP-43 proteinopathy associated with a complex constellation of tau pathology as reported previously [52]. It was characterised by neuronal cytoplasmic, thin neuritic and synaptic TDP-43 immunoreactivity and was seen in the cingular, temporal, entorhinal, frontobasal and insular cortex, hippocampus (CA subregions and dentate gyrus), as well as the basal ganglia, septal nuclei, thalamus, substantia nigra, inferior olives. It did not involve the motor cortex or corticospinal tracts. In addition, a single case of FTLD-TDP (type C; [57] associated with amygdala Lewy-bodies but lacking other pathology was also observed.

We observed a wide spectrum of tauopathies (Fig. 4) in 54 individuals in the neuropathology study group (n = 233). Since thorn-shaped astrocytes were seen frequently restricted to the subpial and subependymal region in the mesial temporal lobe [56, 82], the sole presence of these astrocytes was not considered as “tauopathies” in the present series. Interestingly, the majority of cases with tauopathy showed lower BB stages (47 out of 54 showed BB stage III or less). The majority of cases were compatible with AGD (n = 37); however, 10 of these were combined with other tauopathies (Fisher exact test, p = 0.001) and only 27 were “pure” AGD [stage I, 11; II, 12, III (6,418), 4] [76]. In addition to four cases of TOD (one with AGD) and two fully developed forms of PSP (one combined with α-synucleinopathy MSA), and a single CBD, there were further tau pathologies that were not clearly classifiable among recognised entities, including combined phenotypes. From these two major patterns could be recognised (summarised in Fig. 4; online supplemental files 1 and 2):
https://static-content.springer.com/image/art%3A10.1007%2Fs00401-013-1157-y/MediaObjects/401_2013_1157_Fig4_HTML.gif
Fig. 4

Stratification of tau pathologies in the present study. For the details on proposed groups see text. AGD argyrophilic grain disease, TOD tangle-only dementia, PSP progressive supranuclear palsy, CBD corticobasal degeneration. The coloured dots representing each proposed group are placed on the anatomical image to indicate the distribution of pathology. The size of the dots in the anatomical image indicates the severity of pathologies

  1. 1.
    Five cases showed histopathological features of PSP with predominance of 4R tau immunoreactivity, including globose neurofibrillary tangles in the subthalamic nucleus, globus pallidus, caudate nucleus and putamen and variable amount of coiled bodies and tufted astrocytes; however, the anatomical involvement was not as extensive as in classical PSP cases; early stage PSP was diagnosed, and the possibility that these represent PSP-Parkinsonism was discussed. Dementia was reported in one and Parkinson syndrome in another; three individuals did not have clinical follow-up examination. A single case showed additional limbic TDP-43 proteinopathy, while argyrophilic grains were observed in two. In one case many Gallyas positive astrocytes showing p62 immunopositivity were present in all CA subregions of the hippocampus (Fig. 5a–c).
    https://static-content.springer.com/image/art%3A10.1007%2Fs00401-013-1157-y/MediaObjects/401_2013_1157_Fig5_HTML.jpg
    Fig. 5

    Representative histopathological images of tauopathies detected in our study. Tau-positive tufted astrocytes in the CA1 subregion of the hippocampus (aright upper inset anti-RD4 immunostaining), which were Gallyas positive (b) and showed anti-p62 immunopositivity (c). Prominent tau-astrogliopathy restricted to the amygdala (d, f) in a case from Group II (see text), which were Gallyas positive (e) and showed anti-p62 immunopositivity (g). Globular glial inclusions in the white matter of the occipital lobe in a case from Group III (see text) (h). A representative case for Group IV, showing reactive gliosis (i anti-GFAP) accompanied by the accumulation of tau-positive thorn-shaped astrocytes (j) with weaker Gallyas positivity (k) in the dentate gyrus of the hippocampus. Thorn-shaped astrocytes in the temporal white matter (l), and in the midline of the medulla oblongata (m). Extreme amount of tau pathology was seen in the amygdala (n) in Group IV cases with moderate Gallyas positivity (o). The substantia nigra also showed threads and astrocytic tau immunoreactivity (p). Bar in a represents 150 μm in a, 40 μm in b, d, e, h, i, and o, 60 μm in c and p, 25 μm in f and g, 100 μm in j, 30 μm in k, 300 μm in l and n, 50 μm in m

     
  2. 2.
    Sixteen cases showed tau pathology with predominant astrogliopathy, which was different from that seen in PSP or CBD: in the grey matter astrocytic tau immunoreactivity was characterised by diffuse granular tau immunopositivity along astrocytic processes and in some associated with perinuclear accentuation giving a thorn-like appearance [52], and in the white matter they were compatible with thorn-shaped astrocytes. Based on the extent of pathological alterations and additional cytopathologies, four groups could be distinguished (Fig. 4):
    1. (A)

      Group I (medial temporal lobe/granular and thorny astrogliopathy type) consisted of four cases with few astrocytes with granular tau immunoreactivity in the processes, which were Gallyas negative. This was associated with variable amount of thorn-shaped astrocytes in the temporal white matter. Three cases were associated with low BB stages and one showed limbic TDP-43 proteinopathy. The fourth individual who underwent clinical evaluation showed amygdala Lewy-related pathology and AGD (stage 2) and was demented.

       
    2. (B)

      Group II (amygdala/granular, thorny and tufted-like astrogliopathy type) comprised two cases that showed prominent astrogliopathy (granular, thorny and tufted-like) restricted to the amygdala (Fig. 5d). This was associated with prominent neuronal cytoplasmic labelling and argyrophilic grains in one. Some of the tufted-like astrocytes were visible in Gallyas staining and showed immunoreactivity for p62 protein (Fig. 5e–g).

       
    3. (C)
      Group III (limbic-basal ganglia-nigral/granular and thorny astrogliopathy and neuronal tauopathy type) comprised six cases with a peculiar constellation of tau pathology as recently described [52]. One of these cases showed additional tau pathology with globular oligodendroglial inclusions [51] mainly in the occipital (Fig. 5h) and less in the parietal and frontal lobe white matter. This was associated with limbic TDP-43 proteinopathy and Lewy-related pathology (Braak stage 4). A further case displayed eosinophilic intranuclear inclusions in neurons (Fig. 6a). These were ubiquitin and p62 (Fig. 6b), and FUS immunoreactive (Fig. 6c). According to semiquantitative scoring (no, occasional, moderate, and many), the p62 immunoreactive inclusions predominated in all CA subregions of the hippocampus and the subiculum, moderate numbers were seen in the caudate and accumbens nucleus and anterior cingular cortex, and occasionally in the temporal cortex and inferior olives.
      https://static-content.springer.com/image/art%3A10.1007%2Fs00401-013-1157-y/MediaObjects/401_2013_1157_Fig6_HTML.gif
      Fig. 6

      Representative images on neuropathological alterations. Presence of eosinophilic intranuclear inclusion bodies in the CA3 subregion of the hippocampus (a), that are immunoreactive for anti-p62 (b) and anti-FUS (c). Co-immunoreactivity patterns of polyclonal anti-TDP-43 and monoclonal tau (AT8) show predominance of tau in some cells (d), nearly complete overlap in others (e) or dissociation of TDP-43 and tau in the same cell (f). Bar in a represents 50 μm in ac and 1.5 μm in df

       
    4. (D)

      Group IV (hippocampal-dentate gyrus-amygdala predominant/granular and thorny astrogliopathy and neuronal tauopathy type) comprised four cases with similar features as in Group III, however, an additional reactive gliosis accompanied by the accumulation of thorn-shaped astrocytes in the dentate gyrus of the hippocampus (Fig. 5i–k) was recognised as a distinct feature. This was associated with astrogliopathy in the temporal (Fig. 5l), and variably in the frontal, and cingular white matter. In some cases it was noted also in the midline of the medulla oblongata (Fig. 5m). Astrogliopathy was abundant in the amygdala (Fig. 5n), including thorn-shaped astrocytes with moderate Gallyas positivity (Fig. 5o) and fine granular staining of the astrocytic processes, which were Gallyas negative. This distribution could be distinguished from the thorn-shaped astrocytes in the periventricular regions as reported in the elderly [82]. The substantia nigra (Fig. 5p), locus coeruleus, and basal ganglia also showed variable tau pathology. If present, TDP-43 pathology showed considerable overlap with neuronal tau positivity mainly in the amygdala (Fig. 6d–f) (also in Group III).

       
     

Discussion

We performed a comprehensive study of protein depositions in the brains of elderly individuals with and without dementia in population group with a relatively narrow age-range (77–87 years). Age is an important factor in the determination of interventions against dementia; cognitive dysfunction in later life is a life-span issue that differs in younger patients with dementia [78]. The possible number of combinations of proteinopathies and other pathologies is high (see Fig. 3), emphasising the need for a personalised approach in the treatment or diagnostic evaluation of dementia. This expands the concept of mixed-type dementia beyond AD with vascular pathology or Lewy-related pathology and has implications for (1) biomarker development, (2) for the better understanding of the clinicopathologic relevance of proteinopathies, and (3) provides novel aspects on the morphological alterations in the ageing human brain. In addition to traditional diagnostic categories (AD-type pathology with BB stages and CERAD criteria), we used a holistic approach and evaluated pathologies as independent factors in a dichotomised way. Moreover, we were able to find a wide spectrum of disorders, which underpins the importance of non-AD disease forms in the elderly brains [70]. According to the frequent finding of protein depositions, it might be hypothesised that due to the lack of an evolutionary need (i.e. pre-senile death of the large majority of the population until more recent centuries) for the development of effective repair or clearance by the protein processing system of neurons (i.e. ubiquitin–proteasome system, UPS), the brains of the elderly cannot cope with the overloading of cells with neurodegeneration-related proteins. There are “weakest links” (i.e. neuronal subgroups) in the brain, which show the earliest pathological responses of the UPS irrespective of the basic etiological process [1]. Therapeutic support of the UPS system as a general preventive measure against accumulation of various neurodegeneration-related proteins may be considered.

It has been hypothesised that Aβ, tau, and α-synuclein mutually promote their accumulation [17, 33]. This is supported by the observation of multiple neurodegenerative proteins involved in genetic neurodegenerative diseases, even with mutated other genes like the prion protein [54]. The present finding of a positive correlation between higher BB stages of neurofibrillary degeneration and semiquantitative score of senile plaques, and presence of further pathologies (i.e. TDP-proteinopathy or α-synucleinopathy) supports this concept. Indeed, it has been reported [29, 39], and is also our observation that α-synuclein and phosphorylated tau may deposit in the same neuron, as is also true for TDP-43 and tau [32] (see also Fig. 6d–f). Interestingly, this phenomenon is prominently seen in the amygdala, suggesting that yet unidentified regionally different biochemical modifications contribute with higher probability to the interaction of proteins. On the other hand, presence of pathological alterations like CAA, Aβ plaques, non-AD type of tauopathy, TDP-proteinopathy, α-synucleinopathy, or vascular lesions alone or even together will not “generate” further pathologies (i.e. from this group of six variables) with higher probability, suggesting complex interactions depending on the morphological type or biochemical aspects of a lesion (i.e. tau-positive AD-related neurofibrillary tangles or tau-astrogliopathy). This might suggest that ageing itself can initiate different pathogenic processes at the same time in different anatomical regions.

Our observations have relevance for biomarker research as well. Protein biomarkers in combination with neuroimaging have been used to support the clinical diagnosis of AD [83]. Combined evaluation of several proteins in the cerebrospinal fluid was already applied [38, 63]. Moreover, TDP-43 or phosphorylated TDP-43 can be evaluated in body fluids and might correlate with brain pathology [30, 45]. Demonstration of the high number of possible combinations of proteinopathies without any predominant “signature” pattern (Fig. 3) serves as a rationale for evaluating a panel of neurodegeneration-related proteins in body fluids as part of a personalised diagnostic signature [48].

Clinicopathological and radiological aspects

A recent comprehensive review of the literature concluded that AD-related neuropathologic alterations are likely significant and that the extent of cognitive impairment parallels the best with the severity of neocortical NFT pathology [69]. Another recent study found that Aβ and tau measured in the neocortex and hippocampus in the oldest-old associate with dementia [74]. In addition, they found correlation of hippocampal sclerosis with attendant TDP-43 pathology, but not α-synucleinopathy, and dementia. We replicate the findings on the influence of TDP-43 pathology, however, also without hippocampal sclerosis. Although detectable in cognitively normal elderly individuals [8, 32] (6 % in our study), the importance of TDP-43 in AD or in relation to cognitive decline in the elderly has been emphasised by further studies [13, 68]. We found also association of the presence of Aβ parenchymal deposits and capillary (type 1) CAA (but not type 2 CAA) and dementia, although we did not quantify the depositions [74]. Aβ deposits have a wide morphological range with different pathophysiological implications [21]. According to our study, the presence of neuritic plaques should not be interpreted as irrelevant as suggested also by others [70]. Moreover, the level of AD-neuropathologic change [64] is also correlated with dementia. It should be noted that in seven cases, inclusion of the phases of Aβ deposition [89] for the evaluation of the neuropathologic level made the classification straightforward. In particular, where some neocortical areas showed focally high or moderate density of neuritic plaques and would require a higher CERAD score, the distribution of Aβ deposits were in a less advanced phase.

In contrast to others [74], we found that α-synuclein pathology may also have an impact on cognitive decline, as reported by others [81, 87]. We expand the group of non-AD tauopathies, and show that these have impact on the clinical presentation. While AGD is thought to be an age-related disorder that has been variably associated with dementia [25, 68] as seen in our study as well, its presence as concomitant pathology is thought to be important by lowering the threshold for cognitive decline [43]. A recent comprehensive study supported the concept that argyrophilic grains are linked with general ageing processes [75]. Interestingly, AGD-like tauopathy appears in cerebrotendinous xanthomatosis, a disease with premature ageing [44]. Individual thresholds for dementia and cognitive reserve may be decisive whether somebody develops dementia or not with the same extent of AGD-related pathology.

While previous studies have differed in the methods (i.e. acquisition/selection of cases or type of staining or antibodies used), all point to the fact that a multidimensional approach is needed for the evaluation of dementia [22]. However, it has been shown that clinic-based and community cohorts may exhibit considerable differences [79]. Our study demonstrates that the diversity of pathologies is high in the community and could be under-recognised by studies focusing only on restricted anatomical regions. In addition, the number of pathological variables correlates with the progression of AD-type changes, confirming that a simplifying concept of AD fails to appreciate additional personalised components that may be relevant for patient treatment and diagnosis. This has implications for neuroradiology as well. A significant correlation between neurofibrillary tangles and brain atrophy is documented; in addition, the medial temporal lobe is the target region for the evaluation of AD [40, 90]. However, this region may show atrophy in various disorders with overlapping clinical presentations [11, 19], and we also show that TDP-43 pathology or non-AD tauopathies may associate with medial temporal lobe atrophy. Indeed, we also confirm a recent observation that hippocampus sclerosis is significantly associated with TDP-43 immunoreactive structures [71].

Previous studies have suggested that severe AD-related neuropathological alterations may be associated with preserved cognitive functions in some individuals [86]. This has also been shown for non-AD pathologies [24, 61, 73] and was considered as a presymptomatic phase [20]. Moreover, so-called cases with reversible clinical diagnosis of possible AD have been also reported [26]. In our study, we observed five cases clinically showing such a reversible diagnosis of AD. Interestingly, these individuals had low BB stages of neurofibrillary degeneration, but showed further pathologies (see Table 3). Accordingly, cerebrovascular lesions or metabolic changes may underlie cognitive changes that may later show improvement: moreover, the presence of Lewy-related pathology, when not directly thought to associate with AD-type of cognitive decline, may alter the reserve capacity of the brain for cognitive performance. This also emphasises the significant inter-individual variability regarding the amount of pathology and cognitive performance. Indeed, the concept that cognitive dysfunction in later life is a life-span issue has emerged from the aspect that age significantly influences the association of pathological variables and dementia [78]. This is supported by the present study (summarised in supplemental file 3), which evaluates brain alterations in a tissue-based holistic way and does not start from separate diagnostic entities like AD, vascular or mixed dementia. Furthermore, the issue whether severe AD pathology may be present without clinically detectable dementia (four individuals in our study, see Table 3) must be scrutinised; the time period from the last cognitive assessment to death may be the crucial point. On the other hand, 14 individuals with clinically detectable AD (“possible” AD at last examination) exhibited mild or moderate AD-type pathology. However, additional pathologies were always present in these brains; AGD, CBD or TOD, or Lewy-bodies and/or vascular lesions. This confirms the conclusion of a recent extensive review that “no significant subset of patients with severe age-associated cognitive decline exists that lacks any pathological substrate” [69]. While we evaluated different types of vascular lesions, we did not quantify them. The importance of the hippocampal microvasculature and microvascular lesions as structural determinants of cognitive decline in nonagenarians and centenarians or elderly individuals with depression has been emphasised [37, 95]. Detailed evaluation of these types of pathologies is the subject of another study.

Neuropathological aspects

The present study revealed that non-AD type neurodegenerative pathology is considerable in the elderly. Indeed, we found individuals with features of MSA, FTLD-TDP, intranuclear inclusion body disease, as well as various tauopathies and their combinations. Interestingly, in one individual the rare α-synucleinopathy MSA was associated with the tauopathy PSP, which supports earlier observations suggesting a higher prevalence of concomitant PSP and MSA than expected by chance [84]. Recently, frontotemporal dementias were reported as under-recognised in the elderly [9]. Moreover, the rare Pick’s disease can be seen in the elderly (even in nonagenerians) with relatively mild degree of AD-related changes [53]. Intranuclear inclusion body disease is a neurodegenerative disorder thought to be very rare in adults [65]. The characteristic eosinophilic nuclear inclusions are immunoreactive for ubiquitin-related proteins, but also partly for FUS (fused in sarcoma) [66, 94]. In our case, intranuclear inclusion body disease was complicated by the concomitant presence of a tauopathy.

In addition to the detection of PSP, CBD, AGD or TOD, we observed a spectrum of tauopathies that were not clearly classifiable. In comparison to other autopsy cohorts that evaluate restricted anatomical samples and focus on AD-related changes based on histochemical stainings, we used a systematic immunohistochemical approach; this could be one reason why we report more atypical forms of tauopathies. We distinguished two major patterns of these atypical tauopathies: one, which shows the features of PSP but in restricted anatomical locations; and another in which astrogliopathy distinct from tufted astrocytes and astrocytic plaques was the major morphological feature. In addition, globular glial inclusion [51] have been also noted, as mentioned by others [31], further expanding the spectrum of tauopathies in the elderly. AGD and TDP-43 proteinopathy frequently associated with these atypical forms of tauopathies, including two cases with early form of PSP. In individuals with tauopathies and cognitive decline, TDP-43 proteinopathy was characterised by neuronal cytoplasmic and neuritic deposits. Interestingly, in cognitively normal elderly individuals where TDP-43 deposition frequently associates with argyrophilic grains [8], mainly neuritic pattern can be seen. This suggests that the subcellular localisation of pathological TDP-43 may have relevance. Frequent association of cytoplasmic TDP-43 with atypical tauopathies is interesting also in the context of the recent finding that neuronal tau generates TDP-43 pathology in animal models [18].

Individuals showing PSP-type pathology did not present with the typical clinical features of PSP including vertical gaze palsy. Lack of all classical clinical features has been reported in PSP series, and a grading system to evaluate the development of PSP-related pathology has been proposed [91, 92]. These studies also indicated that PSP-parkinsonism should be distinguished from the classical clinical form of PSP (i.e. Richardson syndrome) by the different pathological tau burden and distribution [91, 92]. It is likely that some of our cases, which we named as early forms of PSP due to the restricted amount and distribution of tau pathology, are representatives of PSP-parkinsonism. Moreover, neuropathological findings of PSP, without clinical evidence of PSP, have been reported in the elderly [24]. These observations suggest that PSP-type pathology may be more frequent in elderly brains than reported, and this may have variable influence on clinical presentations, depending on presence of concomitant pathology and the individual’s threshold for the development of clinical symptoms.

Astroglial tau pathology in the elderly brains has been reported in several contexts. In 2004, Schultz et al. [82] reported high prevalence of thorn-shaped astrocytes in the aged human medial temporal lobe. The comprehensive Medical Research Council Cognitive Function and Ageing Study further characterised the frequency of mesial temporal glial tauopathies [56]. They reported high frequency of thorn-shaped astrocytes prominently in the anterior (i.e. amygdala) level. In addition, they noted tufted astrocytes in the occipito-temporal gyrus in 5.2 % [56]. In another study on PSP reported fine tau immunoreactive networks in astrocytes instead of the typical tufted astrocytes [77]. We expand these and our [52] previous findings, based on AT8 immunostaining in several anatomical regions, and suggest at least four groups depending on the distribution of the astrogliopathy. It must be emphasised that this grouping is based on a relatively low number of cases, thus its applicability must be confirmed in a larger multicentric series. Some of these may represent different stages of the same disorder, however, our first attempt of grouping according to cytopathology and anatomical involvement may serve a basis for further classification of tauopathies in the elderly. Both cases with and without cognitive decline were observed (see supplemental file 2); concomitant pathology, including Lewy-bodies and TDP-43 proteinopathy, was variable and included lower stages [15] and phases [89] of AD-related pathology as well as different stages of AGD [76]. These observations further support the concept that the tau-astrogliopathy has a pathogenesis differing from AD [85] and in some cases it may lower the threshold for cognitive impairment. Astrocytic tau pathology distinct from PSP or CBD has been reported in chronic traumatic encephalopathy [60]. However, in our cases astrogliopathy was not accentuated in the depth of sulci and was not associated with irregular cortical distribution of tau immunoreactive neurofibrillary tangles, in particular not preferentially in the superficial layers in the cerebral cortex, as described in chronic traumatic encephalopathy [60].

Conclusions

We describe here a wide spectrum of morphological alterations and proteinopathies in the elderly brain in the community. The main observations can be summarised as follows:
  1. 1.

    All brains showed some degree of neurofibrillary degeneration while Aβ parenchymal deposits were observed in 68.7 %.

     
  2. 2.

    The possible number of combinations of proteinopathies and vascular or other pathologies is high in the elderly.

     
  3. 3.

    The higher the stage of neurofibrillary degeneration and the amount of senile plaques, the more probable the presence of further pathological alterations. Thus, pure AD pathology associated with dementia is infrequent.

     
  4. 4.

    The presence of one specific pathological variable (CAA, Aβ parenchymal deposits, tauopathy, TDP-proteinopathy, α-synucleinopathy, vascular lesions) does not increase the probability of further ones to combine.

     
  5. 5.

    The spectrum of tauopathies in the elderly expands beyond well-known entities and can be grouped according to the predominant cytopathology and involved anatomical regions.

     
  6. 6.

    In addition to AD-neuropathologic change, tauopathies associated well with dementia, while TDP-43 pathology and α-synucleinopathy showed strong effects but lost significance when evaluated together with AD-neuropathologic change.

     
  7. 7.

    Medial temporal lobe atrophy can be detected also in non-AD neuropathologies.

     

These observations suggest three levels of vulnerability: (1) Anatomical vulnerability, which serves as a basis for distinct atrophy patterns related to cognitive decline; (2) cellular vulnerability, which has implications for the understanding of the pathogenesis of distinct disorders and has relevance for biomarker research as well. Finally, (3) protein vulnerability indicating that cognitive decline is associated with the deposition of different proteins in the brain in a yet not fully identified spectrum of biochemical modifications [48]. This may reflect different aetiologies. How these levels of vulnerability associate with different patterns and dynamics of cognitive decline is the essential basis to understand differences in prognosis and therapeutic response of individuals with cognitive decline. Large-scale multicentric studies are needed that combine clinical, neuroimaging, genetic assessment, body-fluid biomarkers, and neuropathological evaluation to provide comprehensive data for better diagnostic stratification of individual patients, for evaluation of prognosis, and personalised therapy trials.

Acknowledgments

The Vienna Trans-Danube Ageing (VITA) study was supported and organised by the Ludwig Boltzmann Institute of Ageing Research. The neuropathology study was supported by the European Commission’s 7th Framework Programme under GA No 278486, “DEVELAGE”.

Conflict of interest

Authors report no conflict of interest.

Supplementary material

401_2013_1157_MOESM1_ESM.tif (21.7 mb)
Supplementary material 1 (TIFF 22178 kb)
401_2013_1157_MOESM2_ESM.pdf (90 kb)
Supplementary material 2 (PDF 91 kb)
401_2013_1157_MOESM3_ESM.tif (4.5 mb)
Supplementary material 3 (TIFF 4640 kb)

Copyright information

© Springer-Verlag Berlin Heidelberg 2013