Neuropathology and cognitive performance in centenarians

With aging, the incidence of neuropathological hallmarks of neurodegenerative diseases increases in the brains of cognitively healthy individuals. It is currently unclear to what extent these hallmarks associate with symptoms of disease at extreme ages. Forty centenarians from the 100-plus Study cohort agreed to post-mortem brain donation. Centenarians self-reported to be cognitive healthy at baseline, which was confirmed by a proxy. Objective ante-mortem measurements of cognitive performance were associated with the prevalence, distribution and quantity of age- and AD-related neuropathological hallmarks. Despite self-reported cognitive health, objective neuropsychological testing suggested varying levels of ante-mortem cognitive functioning. Post-mortem, we found that most neuropathological hallmarks related to age and neurodegenerative diseases such as Aβ and Tau pathology, as well as atherosclerosis, were abundantly present in most or all centenarians, whereas Lewy body and pTDP-43 pathology were scarce. We observed that increased pathology loads correlated across pathology subtypes, and an overall trend of higher pathology loads to associate with a lower cognitive test performance. This trend was carried especially by the presence of neurofibrillary tangles (NFTs) and granulovacuolar degeneration (GVD) and to a lesser extent by Aβ-associated pathologies. Cerebral Amyloid Angiopathy (CAA) specifically associated with lower executive functioning in the centenarians. In conclusion, we find that while the centenarians in this cohort escaped or delayed cognitive impairment until extreme ages, their brains reveal varying levels of disease-associated neuropathological hallmarks, some of which associate with cognitive performance.


Introduction:
The increase in average life expectancy warrants gaining insight in healthy aging. In particular, the study of centenarians might reveal new leads towards healthy aging and potential clues to overcome age-related diseases. Alzheimer's disease (AD) is the most common age-related neurodegenerative disease for which increasing age is the most important risk factor. The prevalence of AD is estimated to double every 5 years after the age of 65, resulting in a population prevalence of roughly 40% for western Europeans aged 90 years and older [22,28,32]. The two most prominent neuropathological hallmarks of AD are (1) the intracellular accumulation of hyperphosphorylated Tau protein (pTau) in the form of Neurofibrillary tangles (NFTs) and (2) the accumulation of extracellular Amyloid-β (Aβ) fibrils into senile plaques [24,30]. Additional AD-associated neuropathological changes are: neuritic plaques (NPs), where Aβ plaques and dystrophic neurites occur in the same structures, cerebral amyloid angiopathy (CAA) illustrated by the deposition of Aβ in the endothelial cells of blood vessels, and granulovacuolar degeneration (GVD), which is characterized by the accumulation of stress proteins in intracellular granules. Whereas Lewy body (LB) and phosphorylated TDP-43  proteinopathies are usually associated with other neurodegenerative diseases, they are also observed in late-onset AD cases [38].
Clinicopathologic correlation studies revealed that the severity of cognitive impairment in AD correlates best with the burden of neocortical NFTs [28]. The presence of Aβ is a prerequisite for AD and may catalyze the pathological process that ultimately leads to clinical dementia [38]. However, the concurrent accumulation of individual pathologies, like CAA, GVD, LB, and TDP43, may have an additive effect on the rate of cognitive decline, and contribute to the apparent decrease of the association between NFTs, Aβ and dementia symptoms in older ages [38]. Brains of older dementia patients commonly show less NFT and Aβ pathology compared to younger AD patients [8,14,15,34]. This suggests that, while disease-associated proteinopathies distinguish well between brains from young AD cases and age-matched nondemented controls, this is more complicated for brains from older individuals. So far, postmortem brain assessments of cognitively healthy individuals who reached >110 years revealed various results: some cognitively healthy centenarians did not accumulate significant pathology, while others appeared to have relatively high levels of pathology [10,29,39].
Previous studies on brain tissues from centenarians commonly lack information on cognitive well-being shortly before death, which is essential for studying the association with neuropathological hallmarks in post mortem brain tissue.
In this study, we report on the occurrence of neuropathological changes in a unique, post mortem brain collection of self-reported cognitively healthy centenarians from the 100-plus Study cohort [16]. The aim of this study is to investigate how neuropathological changes relate with performance on neuropsychological tests in centenarians. Prospective neuropsychological testing of these individuals allowed us to correlate ante-mortem cognitive functioning with the occurrence of pathological hallmarks of neurodegenerative diseases in the post-mortem brain. For this, we focused on inclusions of pTau as NFTs [4] and ARTAG [21]; the distribution of Aβ protein as diffuse [41], classical or NPs [23], and CAA [40]; the distribution of casein kinase 1 delta as a marker of GVD [43], and the distribution Lewy bodies [6] and pTDP-43 accumulations [25]. Furthermore, for an individual case, we report on the in-vivo assessment of pathology using an MRI and amyloid PET scan, and the occurrence of postmortem neuropathological changes.

100-plus Study cohort:
The 100-plus Study is a prospective centenarian cohort study. Inclusion criteria for the study are, next to being able to provide proof of being 100 years or older, that centenarians "selfreport to be cognitively healthy, which is confirmed by a proxy" [16]. Although it is not a requirement for study participation, almost 30% of all participants agree to post-mortem brain donation. Here we compare the ante-and post-mortem status of the first 40 participants of the 100-plus Study that came to autopsy.
The 100-Plus Study was approved by the Medical Ethical Committee of the VU University Medical Center. Informed consent was obtained from all study participants. The detailed study protocol is described elsewhere [16].

Post-mortem brain autopsy procedures:
Autopsies were performed in collaboration with the Netherlands Brain Bank (NBB, Amsterdam, The Netherlands). Brain weight was recorded, and we macroscopically determined levels of atrophy and atherosclerosis of the intracranial vessels. Atrophy was subjectively staged by an experienced neuropathologist according to severity (none (0), slight (1), moderate (2) or strong (3)). Atherosclerosis was scored as mild (1), if only some parts of the circle of Willis and the basal arteries were affected; moderate (2), if the circle of Willis and the basal arteries were severely affected and the arteria cerebri media was not affected beyond the first cm; and severe (3), if also the deeper part of the arteria cerebri media was affected. The right hemisphere of the brain was formalin fixed, and the left hemisphere was dissected and the pieces were snap frozen in liquid nitrogen.

Neuropathology: immunohistochemistry:
Neuropathological characterization of centenarian brains was performed by Haematoxylin and Eosin (H&E) stain, Gallyas silver stain [31,44] and immunohistochemistry (IHC). For IHC, brain tissues were formalin fixed and paraffin embedded (FFPE), then sectioned into 6 μm slices and mounted on microscope slides (Leica Xtra adhesive slides, Leica Microsystems, Rijswijk, The Netherlands). After deparaffinization and rehydration, endogenous peroxidase activity was blocked in 0.3% H 2O2 in methanol for 30 minutes. Antigen retrieval was performed with heated sodium-citrate buffer (10 mM/L, pH 6.0). For the staining of α-synuclein, pretreatment included only heated sodium citrate without endogenous peroxidase blocking. All primary antibodies were diluted in normal antibody diluent (Immunologic, VWR company, Duiven, The Netherlands) (Table 1). Primary antibody incubation was performed overnight at 4°C.
Subsequently, sections were incubated with the EnVision detection system (goat antimouse/rabbit horseradish peroxidase (HRP), DAKO, Heverlee, Belgium) for 40 minutes at room temperature. Between incubation steps, sections were rinsed with phosphate buffered saline, pH 7.0 (PBS). Sections were incubated for 5 minutes with the chromogen 3,3′diaminobenzidine (DAB, EnVision Detection system/HRP, DAKO, Heverlee, Belgium) to visualize immunoreactivity. Nuclei were counterstained with haematoxylin. Hereafter, slides were dehydrated and mounted using the non-aqueous mounting medium Quick-D (Klinipath, Duiven, The Netherlands). Negative controls were obtained by performing the same procedures while omitting the primary antibody. Neuropathological evaluation: Distribution of Aβ and hyperphosphorylated tau (pTau) in NFTs and NPs was determined by immunohistochemical staining (IHC) with respectively IC-16 and AT-8 antibodies on the frontal cortex (F2), temporal pole cortex, parietal cortex (superior and inferior lobule), occipital pole cortex and the hippocampus (CA1 and entorhinal area of the parahippocampal gyrus). NFTs and NPs were identified using a combination of Gallyas silver staining and IHC for pTau (AT-8). Aβ distribution was evaluated according to the Thal staging ranging from 0 to 5 [42]. The temporal-spatial distribution of NFTs was evaluated according to the Braak stages ranging from 0 to VI [3][4][5]. NPs were evaluated according to CERAD scores ranging from 0 to 3 [23].

Neuropsychological tests and educational attainment:
All study participants were visited at home by a trained researcher at baseline inclusion, and when available and willing, at yearly follow-up visits. During these visits, we objectively assessed cognitive functioning using an elaborate test battery described elsewhere [16]. Here we report the performance on the Mini-Mental State Examination (MMSE), Visual Association Test (VAT) and the Clock Drawing Test (CDT). The MMSE is an 11-item test evaluating overall cognitive functioning with a scoring range of 0-30 (bad-good) [13]; missing items were imputed as described before [16], provided no more than 6 items were missing. The VAT is a test for episodic memory with a 0-12 (trial 1+2) range (bad-good) [17], and the CDT is a test for visuospatial and executive functioning, with a 0-5 range (bad-good) [ Correlation of cognitive performance with post mortem brain pathology: To determine whether performance on cognitive tests of centenarians correlated with postmortem brain pathologies, we correlated pre-mortem neuropsychological test scores and demographic descriptives with levels of post-mortem neuropathological hallmarks of aging and disease For this, we used MMSE, VAT and CDT test scores, both at baseline and last visit. A secondary measure of cognitive performance was "the ability to complete all tests" as a proxy of fatigue or non-willingness; scores were 2 (all tests completed, or tests not applied due to sensory problems), 1 (MMSE completed, which was administered first, but not VAT and/or CDT), and 0 (no tests completed). After each study visit, the impression of cognitive health of the participant was subjectively estimated by a trained researcher as 0 (symptoms of impairment), 1 (possible symptoms of impairment), or 2 (no symptoms of impairment) (see [16] for protocol). The scored impression of cognitive health closest to death was used for analysis.

Statistical analyses:
Correlations between neuropsychological test performance and neuropathology were determined using Pearson correlation. We calculated p-values and false discovery rates (fdr) for all correlations (Supplemental Tables 3b and 3c). In Figure 3, strength of the Pearson correlation coefficient is indicated by the size and color of each dot, the fdr is indicated as asterisks (*<20%, **<10%, ***<5%). All calculations were performed in R (version 3.3) [33].

PET/MRI scan in case report:
Scanning was performed 1 month after study inclusion for case 2015-060. The participant was injected with 383 MBq 11C-PIB. Sixty minutes after injection, dynamic images of the brain were captured on the Philips TF PET-MRI scanner in 6 frames of 5 minutes duration. Due to extensive movement, only the activity in the first frame could be used. The PET scan was visually assessed by an experienced nuclear medicine physician.
At autopsy the median brain weight was 1195 g (IQR 1060 g-1355 g) in males and 1115g (IQR  Overall, the centenarians in our cohort showed moderate levels of AD related pathology.
Centenarians rarely showed the highest score for AD pathology, with the exception of GVD.
Overall scores for Lewy bodies and TDP-43 appeared relatively low in centenarians.

Performance on neuropsychological tests of 40 centenarian brain donors:
The aim of this study is to investigate how neuropathological changes are associated with performance on neuropsychological tests in centenarians. The last visit occurred on average 10±7 months prior to brain donation. 18 centenarians were available for at least one follow-up visit; follow-up for two additional centenarians was performed by questionnaire. Centenarians had a variable educational attainment (median years of education: 8, range 6 -20y). MMSE, VAT and CDT scores were available for respectively 95%, 67.5% and 65% of centenarians at baseline (

Correlations between different neuropathological hallmarks:
The levels of neuropathological hallmarks of AD are inter-correlated ( Figure 3 and Supplementary

Correlations between test performance, demographic characteristics and pathology:
We observed an overall trend that better performance correlates among cognitive tests. For the centenarians who completed cognitive tests at baseline and at last visit, we find a strong correlation between performances ( Figure 3). The ability to complete testing was associated with lower ages at death, researcher impression of cognitive health and higher baseline MMSE scores. Likewise, we observed that increased pathology loads correlate among pathology subtypes. We also observe an overall trend that lower pathology loads correlate with better performance on tests (Figure 3). This trend is mainly carried by the association of Braak stages    (f). pTau immuno-staining (g-i) is shown as (pre)tangles and neuritic plaque like structures in the hippocampal CA1 region (g) and temporal pole cortex (h1 and h2), as well as astroglial pTau in ARTAG in the temporal pole cortex (i1 and i2). GVD (CK1δ granules) (j-k) is shown in the hippocampus (j) and temporal pole cortex (k). Scale bar 25µm.

Discussion:
In a unique cohort of 40 centenarians, in which individual cognitive functioning was assessed shortly before death, we investigated the post-mortem brain for disease-and aging-associated neuropathological changes. At baseline, cognitive performance varied: neuropsychological test performance suggested that some centenarians had no symptoms of cognitive impairment, while others showed considerable impairment in several cognitive domains.
Furthermore, a small subset of centenarians with multiple follow-up visits showed some decline after study inclusion, while others remained cognitively stable until death. We also observe an overall trend that higher pathology loads associate amongst each other, but also with a lower performance on cognitive tests: this trend is carried especially by the NFT and GVD loads and to a lesser extent by Aβ-associated pathologies. Whereas previous findings indicated that many elderly individuals with CAA remain asymptomatic [12], our results suggested that the severity of amyloid accumulation in the vessel walls (CAA), but not Aβ accumulation in the form of plaques, significantly associated with lower executive functioning, as measured by the Clock Drawing Test (CDT). Together, our findings suggest that cognitive performance of centenarians is, at least in part, sensitive to the increase of neuropathological hallmarks of disease.

Accumulation of neuropathologies is common in centenarians:
Aβ pathology was present in most centenarians at variable levels, and Braak stages for NFT were found in all centenarians and distribution occurred up to stage IV, and we did not observe Braak stages 0, V or VI. Similarly, we detected significant accumulations of NPs, which did not exceed CERAD score 2. In fact, several centenarians have exceptionally good brain function with a high NFT load (Braak stage IV) and NP accumulations (CERAD scores 1 or 2), and high Aβ load (Thal stages [3][4][5]. This is intriguing, since most younger individuals with similar amounts of pathology have at least some symptoms of AD [28,38]. Overall, the cognitive function of these centenarians was conserved despite the presence of neuropathological hallmarks. We also find that several centenarians have no or only low amounts of Aβ pathology and intermediate levels of NFT loads, corresponding with the neuropathological description of primary age-related tauopathy (PART), [9]. The relation between PART and the occurrence of dementia is still elusive and its implications for cognition are still debated [11,18] which we find only weak evidence in centenarians. We find that higher pTDP-43 stages occurred in all cases with hippocampal sclerosis, which is in agreement with previous observations [25]. Hippocampal sclerosis has previously been associated with lower performance on neuropsychological tests [27], and here we find a significant association with the inability to complete tests, a possible proxy for fatigue. Likewise, the accumulation of αsynuclein (Lewy bodies) was also limited in centenarians, despite the reported increased occurrence of this pathology with age[38]. We did not observe a correlation between the presence of Lewy bodies and cognitive function. The overall low occurrence of TDP-43 and Lewy body pathologies in this group of centenarians implies that these pathologies are more likely disease-related and not aging-related. In addition, we found one odd case with a Braak stage 6 for Lewy bodies, who was diagnosed with Parkinson's disease many years before death, but performed well on the MMSE.
In general, in our cohort of centenarians increased levels of pathology correlate with lower cognitive performance. This observation is in agreement with the idea that simultaneous accumulation of different neuropathological hallmarks of AD could collectively account for the cognitive decline that is associated with aging [18,19,28,38]. It remains elusive whether these centenarians retained their cognitive health to extreme ages because they delayed the accumulation of pathologies to much later ages, or because build-up of pathology was slower.
Thus far, we present the findings in a sample of 40 centenarians. With the continuation of this prospective study, we will be able to validate these findings in a larger sample in the near future. Also, as of yet we did not observe a cognitively healthy centenarian who was free of atherosclerosis or AD-associated pathologies, as reported previously for a 115 year old woman [10]. In line with reports that cognitive performance is highly correlated with survival in the oldest old [37], we find that the centenarians who score highest on cognitive tests in the ongoing prospective study are underrepresented among the first centenarians in the 100-plus Study who came to autopsy (data no shown). We therefore speculate that individuals with a low burden of pathology might survive until extreme ages, and will come to autopsy at a later time after inclusion.
In conclusion, we find that AD associated neuropathology is common in centenarians, albeit within limits for specific neuropathological hallmarks. Furthermore, we find that although the centenarians in this cohort escaped or delayed cognitive impairment until extreme ages, the accumulation of overall pathologies relate with lower cognitive performance. It remains unclear whether pathology accumulation was slower compared to others or whether accumulation started later. Overall, the registration of ante-mortem cognitive performance combined with thorough post-mortem brain analysis provides a unique window of opportunity for the association of cognitive function with the occurrence of neuropathological changes in centenarians.