Abstract

Alzheimer’s disease, which is defined pathologically by abundant amyloid plaques and neurofibrillary tangles concurrent with synaptic and neuronal loss, is the most common underlying cause of dementia in the elderly. Among the oldest-old, those aged 90 and older, other ageing-related brain pathologies are prevalent in addition to Alzheimer’s disease, including cerebrovascular disease and hippocampal sclerosis. Although definite Alzheimer’s disease pathology can distinguish dementia from normal individuals, the pathologies underlying cognitive impairment, especially in the oldest-old, remain poorly understood. We therefore conducted studies to determine the relative contributions of Alzheimer’s disease pathology, cerebrovascular disease, hippocampal sclerosis and the altered expression of three synaptic proteins to cognitive status and global cognitive function. Relative immunohistochemistry intensity measures were obtained for synaptophysin, Synaptic vesicle transporter Sv2 (now known as SV2A) and Vesicular glutamate transporter 1 in the outer molecular layer of the hippocampal dentate gyrus on the first 157 participants of ‘The 90+ Study’ who came to autopsy, including participants with dementia (n = 84), those with cognitive impairment but no dementia (n = 37) and those with normal cognition (n = 36). Thal phase, Braak stage, cerebrovascular disease, hippocampal sclerosis and Pathological 43-kDa transactive response sequence DNA-binding protein (TDP-43) were also analysed. All measures were obtained blind to cognitive diagnosis. Global cognition was tested by the Mini-Mental State Examinaton. Logistic regression analysis explored the association between the pathological measures and the odds of being in the different cognitive groups whereas multiple regression analyses explored the association between pathological measures and global cognition scores. No measure clearly distinguished the control and cognitive impairment groups. Comparing the cognitive impairment and dementia groups, synaptophysin and SV2 were reduced, whereas Braak stage, TDP-43 and hippocampal sclerosis frequency increased. Thal phase and VGLUT1 did not distinguish the cognitive impairment and dementia groups. All measures distinguished the dementia and control groups and all markers associated with the cognitive test scores. When all markers were analysed simultaneously, a reduction in synaptophysin, a high Braak stage and the presence of TDP-43 and hippocampal sclerosis associated with global cognitive function. These findings suggest that tangle pathology, hippocampal sclerosis, TDP-43 and perforant pathway synaptic loss are the major contributors to dementia in the oldest-old. Although an increase in plaque pathology and glutamatergic synaptic loss may be early events associated with cognitive impairment, we conclude that those with cognitive impairment, but no dementia, are indistinguishable from cognitively normal subjects based on the measures reported here.

Introduction

Along with tangles and plaques, Alzheimer’s disease is characterized by the loss of neurons and their synapses. Indeed, significant synapse loss has been documented in all brain regions affected by Alzheimer’s disease pathology (DeKosky et al., 1996) and reductions in the number of presynaptic and synaptic proteins may be the most consistent progression marker of Alzheimer’s disease as the burden of plaques and tangles grows and neurons continue to degenerate (Beeri et al., 2012). For example, significant reductions in synaptophysin and VGLUT1 were found in both the parietal and occipital cortices of Alzheimer’s disease brains in one study (Kirvell et al., 2006) and our group previously reported a decrease in synaptophysin in the frontal cortex of dementia patients (Head et al., 2009). Nonetheless, the relationship between anatomy, synaptic proteins and Alzheimer’s disease pathology is complex (Honer, 2003). In our previous study, individuals with cognitive impairment, but no dementia (CIND) had increases in synaptophysin whereas others have reported a significant decrease in the hippocampal levels of synaptophysin, even in CIND individuals (Sze et al., 2000).

Because it has long been known that changes in the perforant pathway are associated with memory impairment (Hyman et al., 1986; Senut et al., 1991), we hypothesized that perforant pathway synaptic changes may be one of the major correlates of cognitive impairment in Alzheimer’s disease. The perforant pathway arises from neurons in layers 2 and 3 of the entorhinal cortex and projects into the outer molecular layer of the hippocampal dentate gyrus. Importantly, neurons and synapses in the inner molecular layer remain unaffected by perforant pathway loss in the outer molecular layer. There are several markers of synaptic health that are reliably detected by immunohistochemistry in sections of CNS tissues including synaptophysin, SV2 and VGLUT1.

Synaptophysin is involved in multiple, important aspects of synaptic vesicle exo- and endocytosis (Valtorta et al., 2004). SV2, besides being a ubiquitous presynaptic protein, is the binding site for the antiepileptic drug levetiracetam, a drug shown to improve cognition in CIND individuals (Bakker et al., 2012). VGLUT1 mediates the accumulation of glutamate—the major excitatory neurotransmitter—into secretory vesicles in the neocortex and hippocampus (El Mestikawy et al., 2011).

Individuals in the ‘The 90+ Study’, a population-based cohort of nonagenarians, have a high prevalence of cognitive impairment (Peltz et al., 2012) allowing for the assessment at autopsy of a significant number of CIND individuals. This investigation is a cross-sectional clinical-pathological correlation study involving the first 157 individuals to come to autopsy. For our analysis, we compared measures of perforant pathway synaptic health across each cognitive group. Since the prevalence of severe Alzheimer’s disease pathology is reduced in nonagenarians, whereas the proportion with hippocampal sclerosis and cerebrovascular disease pathology increases (Nelson et al., 2011b), we also measured the underlying pathologies affecting these individuals, specifically their Braak stage, Thal phase, and the presence of hippocampal sclerosis and cerebrovascular disease. Our aim was to understand both the synaptic and pathological changes underlying cognitive impairment and dementia.

Materials and methods

Study participants were the first 157 individuals to come to autopsy from the 90+ Study, a longitudinal population-based study of ageing and dementia in people aged 90 and older who are survivors of the Leisure World Cohort Study (Corrada et al., 2012). Briefly, individuals live at home as well as in institutions, and represent the full spectrum of health and cognitive abilities. All 90+ Study participants had evaluations every 6 months including a neurological examination by a trained physician or nurse practitioner and a full neuropsychological battery that included the Mini-Mental State Examination (MMSE). Relevant medical history, medication use, and demographic information were obtained from the participants or their informants. Medical records, including brain imaging evaluations were obtained from the participant’s physicians. Information about cognitive (Clark and Ewbank, 1996) and functional abilities (Pfeffer et al., 1982) were obtained from informants in frequent contact with the participants. To inquire about the onset of cognitive problems, the Dementia Questionnaire (Silverman et al., 1986; Kawas et al., 1994) interview was conducted over the phone with informants of participants with evidence of cognitive impairment. Shortly after death, the Dementia Questionnaire was done with the decedent’s informant to inquire about the participant’s condition since the last evaluation. The Institutional Review Board of the University of California, Irvine, approved all procedures and all participants or their surrogates gave written informed consent.

Determination of cognitive status

After a participant’s death, all available information was reviewed and discussed during a multidisciplinary consensus diagnostic conference led by ‘The 90+ Study’ principal investigator (C.K.). Participants were classified as normal, CIND, or as having dementia. Dementia diagnosis was established using Diagnostic and Statistical Manual of Mental Disorders 4th Edition criteria (American Psychiatric Association et al., 1994). CIND is defined by initial cognitive impairments such as deficits in episodic memory (Albert et al., 2001), executive dysfunction (Chen et al., 2001), naming difficulties or other aphasias (Saxton et al., 2004). Participants were classified as CIND if they showed cognitive or functional deficits that were not severe enough to meet criteria for dementia. All cognitive diagnoses were made blinded to pathological evaluations.

Neuropathology

All autopsies were performed at the University of California, Irvine. After weighing the whole brain and gross inspection, one hemisphere was dissected as previously described (Berlau et al., 2009). Six-micrometre thick, coronal sections of mid-frontal cortex superior temporal cortex, anterior hippocampus, amygdala, substantia nigra and medulla oblongata were cut. All histological staining, immunohistochemistry and microscopic analyses were performed in the Centre for Neurodegenerative Disease Research (CNDR) at the University of Pennsylvania as described (Robinson et al., 2011). Briefly, sections were subjected to immunohistochemistry using the avidin-biotin complex detection method (VECTASTAIN® ABC kit; Vector Laboratories) with ImmPACT™ diaminobenzidine peroxidase substrate (Vector Laboratories) as the chromogen using monoclonal antibodies to phosphorylated tau (mouse PHF1; 1:1K, gift of Dr Peter Davies, Manhasset, NY), β-amyloid (mouse NAB228; 1:15K; generated in CNDR), TARDBP (rat 409/410; 1:500; gift of Dr Manuela Neumann, Zurich, Switzerland), SV2 (mouse SV2; 1:20K; DSH Iowa), synaptophysin (mouse MAB368; 1:1K; Millipore) and VGLUT1 (Guinea pig VGLUT1; 1:7.5K; SYSY).

Topographical Braak staging (stages I–VI) was assigned from PHF1 stained slides (n = 157) (Braak et al., 2006). Thal phases were determined from NAB228 stained hippocampal slides: phase 0–1, 2, 3 and 4 (n = 150) (Thal et al., 2006). TARDBP inclusions and neurites were determined from 409/410 stained hippocampal slides: presence/absence.

The assessment of cerebrovascular disease pathology (n = 108) and hippocampal sclerosis (n = 155) was determined from Harris haematoxylin and eosin stained mid-frontal cortex, superior temporal cortex, hippocampus, amygdala, substantia nigra and medulla sections. Hippocampal sclerosis was assessed as follows: 0 for no gliosis or neuronal loss; 1+ for mild gliosis or neuronal loss in the CA1 or subiculum; 2+ for moderate or severe gliosis and neuronal loss consistent with definite hippocampal sclerosis. Cerebrovascular lesions such as infarcts, micro-infarcts or micro-bleeds along with cerebral amyloid angiopathy and hippocampal sclerosis were used to generate cerebrovascular disease pathology scores following a simplified staging of Jellinger and Attems as follows: 0 for cases without major infarcts, hippocampal sclerosis, cerebral amyloid angiopathy or other lesions; 1+ for minimal vascular pathology cases with mild to moderate cerebral amyloid angiopathy and/or 1–2 small lacunes; 2+ for moderate vascular pathology cases with severe cerebral amyloid angiopathy and/or hippocampal sclerosis and/or major infarcts (Jellinger and Attems, 2003).

Synaptic relative immunointensity ratios (RIRs) were obtained by the following process. A preliminary study of synaptic and presynaptic proteins was done including antibodies to synaptophysin, SV2, VGLUT1, synataxin 1, VAMP2, synapsin 1, synaptotagmin 1, dynamin, VGAT, PSD95 and GLUR1; obvious outer molecular layer synaptic loss was observed on slides obtained from patients with Alzheimer’s disease when stained with antibodies to synaptophysin, SV2 and VGLUT1. For this study, hippocampal slides were stained for synaptophysin (n = 149), SV2 (n = 151) or VGLUT1 (n = 152). Sections were scanned (Nikon DS-Fi2 camera, gain 1.2, exposure 15 ms) at ×10 on a Nikon Eclipse TE2000 microscope using NIS Elements software (Nikon Instruments, Inc.). ImageJ 1.47t (National Institutes of Health) was used for analysis. Mean pixel intensities of the inner- and outer molecular layer of the dentate gyrus just under CA1 were captured (Fig. 1). Raw values on an 8-bit scale (range of 45–185) were normalized to a blank area for each slide (median value 41). Synaptic RIRs of the outer/inner molecular layer were calculated from (outer molecular layer − blank) / (inner molecular layer − blank) for each case. To verify the repeatability of our measurements, 10 random cases were stained twice by the synaptic antibodies and synaptic RIRs were obtained independently by two researchers (L.M.P., J.L.R.). A K of 0.987 was obtained (Type C intraclass correlation coefficient).

Figure 1

Representative images of synaptic protein immunoreactivity. Synaptic marker immunohistochemistry from two cases shows distinct staining of the inner molecular layer (highlighted in yellow) and the outer molecular (highlighted in purple) of the hippocampus. (A–C) The relatively healthy outer molecular layer: (A) synaptophysin, (B) SV2 and (C) VGLUT1 with synaptic ratios 1.19, 1.13 and 0.78, respectively (note that VGLUT1 is consistently lower than the other two). In contrast, (D–F) represents an individual with severely reduced outer molecular layer, whose inner molecular layer is preserved: (D) synaptophysin, (E) SV2 and (F) VGLUT1 with synaptic ratios 0.53, 0.74 and 0.56, respectively.

Figure 1

Representative images of synaptic protein immunoreactivity. Synaptic marker immunohistochemistry from two cases shows distinct staining of the inner molecular layer (highlighted in yellow) and the outer molecular (highlighted in purple) of the hippocampus. (A–C) The relatively healthy outer molecular layer: (A) synaptophysin, (B) SV2 and (C) VGLUT1 with synaptic ratios 1.19, 1.13 and 0.78, respectively (note that VGLUT1 is consistently lower than the other two). In contrast, (D–F) represents an individual with severely reduced outer molecular layer, whose inner molecular layer is preserved: (D) synaptophysin, (E) SV2 and (F) VGLUT1 with synaptic ratios 0.53, 0.74 and 0.56, respectively.

All pathological diagnoses were done blinded to clinical diagnosis.

Statistical analysis

We compared characteristics of the three cognitive groups: normal, CIND and dementia using χ2 tests for categorical variables and ANOVA for continuous variables. We used multinomial logistic regression models to determine the association between each individual neuropathological or synaptic protein measure and cognitive diagnosis. We report the odds of being in the CIND group compared to the normal group, the odds of being in the dementia group compared to the CIND group, and the odds of being in the dementia group compared to the normal group. Neuropathological and synaptic protein measures were analysed as continuous variables in the logistic regression analyses. Separate regression models were used for each neuropathological and synaptic protein measure. We also explored the association between the individual neuropathological or synaptic measures and the MMSE, a measure of global cognition, using multiple linear regression analyses. Finally, we analysed the relative contribution of the different pathological measures to global cognitive scores, by including all synaptic and pathological measures in a multiple regression model. All regression models were adjusted for age at death and gender and all analyses were performed using SAS version 9.3 (SAS Institute Inc.).

Results

Subject characteristics

Of the 157 participants of this study, 36 had normal cognition, 37 were diagnosed with CIND and the majority (n = 84) had dementia (Table 1). The most frequent clinical diagnosis of the participants with dementia was Alzheimer’s disease alone (65%) or in combination with other dementias (21%) followed by vascular dementia (7%) and dementia with Lewy bodies (6%).

Table 1

Characteristics of The 90+ Study participants

Characteristic All subjects Normal CIND Dementia 
(n = 157) (n = 36) (n = 37) (n = 84) 
 Mean (SD) 
Age at death 98.0 (3.6) 97.8 (2.5) 98.4 (4.0) 97.7 (3.7) 
Last MMSE scorea 18.1 (10.1) 27.8 (1.6) 23.7 (5.4) 11.2 (8.6) 
MMSE interval to death (months) 7.5 (7.7) 5.9 (3.3) 6.5 (4.5) 8.6 (9.8) 
Brain weight (g) 1127 (121) 1179 (104) 1123 (125) 1104 (121) 
 n (%) 
APOE E4b 
    0 alleles 119 (78) 27 (79) 33 (89) 59 (73) 
    ≥ 1 alleles 33 (22) 7 (21) 4 (11) 22 (27) 
APOE E2 
    0 alleles 131 (86) 29 (85) 31 (84) 71 (88) 
    ≥ 1 alleles 21 (14) 5 (15) 6 (16) 10 (12) 
Gender 
    Male 45 (29) 15 (42) 10 (27) 20 (24) 
    Female 112 (71) 21 (58) 27 (73) 64 (76) 
Educationc 
    ≤ High school 45 (29) 7 (19) 8 (22) 30 (36) 
    Any college 73 (47) 16 (44) 20 (54) 37 (45) 
    Any graduate school 38 (24) 13 (36) 9 (24) 16 (19) 
    Normal cognition 36 (23) 36 (100)   
    CIND 37 (24)    
Memory impairment   13 (35)  
Executive impairment   12 (32)  
Other impairmentd   12 (32)  
Dementia 84 (54)  
Alzheimer’s disease only    55 (65) 
Alzheimer’s disease pluse    18 (21) 
Vascular dementia    6 (7) 
Other dementia    5 (6) 
Characteristic All subjects Normal CIND Dementia 
(n = 157) (n = 36) (n = 37) (n = 84) 
 Mean (SD) 
Age at death 98.0 (3.6) 97.8 (2.5) 98.4 (4.0) 97.7 (3.7) 
Last MMSE scorea 18.1 (10.1) 27.8 (1.6) 23.7 (5.4) 11.2 (8.6) 
MMSE interval to death (months) 7.5 (7.7) 5.9 (3.3) 6.5 (4.5) 8.6 (9.8) 
Brain weight (g) 1127 (121) 1179 (104) 1123 (125) 1104 (121) 
 n (%) 
APOE E4b 
    0 alleles 119 (78) 27 (79) 33 (89) 59 (73) 
    ≥ 1 alleles 33 (22) 7 (21) 4 (11) 22 (27) 
APOE E2 
    0 alleles 131 (86) 29 (85) 31 (84) 71 (88) 
    ≥ 1 alleles 21 (14) 5 (15) 6 (16) 10 (12) 
Gender 
    Male 45 (29) 15 (42) 10 (27) 20 (24) 
    Female 112 (71) 21 (58) 27 (73) 64 (76) 
Educationc 
    ≤ High school 45 (29) 7 (19) 8 (22) 30 (36) 
    Any college 73 (47) 16 (44) 20 (54) 37 (45) 
    Any graduate school 38 (24) 13 (36) 9 (24) 16 (19) 
    Normal cognition 36 (23) 36 (100)   
    CIND 37 (24)    
Memory impairment   13 (35)  
Executive impairment   12 (32)  
Other impairmentd   12 (32)  
Dementia 84 (54)  
Alzheimer’s disease only    55 (65) 
Alzheimer’s disease pluse    18 (21) 
Vascular dementia    6 (7) 
Other dementia    5 (6) 

aExcludes four participants with missing MMSE score.

bExcludes five participants with unknown ApoE allele status.

cExcludes one participant with unknown degree of education.

dIncludes impairment in domains other than memory or executive function.

eIncludes mixed Alzheimer’s disease/vascular dementia, Alzheimer’s disease/other and Alzheimer’s disease with dementia with Lewy bodies.

Participants had an average age at death of 98 years (range: 94–101 years), were mostly female (71%), and highly educated (71% had at least a college education). MMSE scores were available for the majority of individuals within 1 year before death (Table 1). The overall frequency of the APOE alleles were 14% for the e2 allele and 22% for the e4 allele. While there was an increase in the frequency of the e4 allele in the dementia group (27%) compared to the normal group (21%), this non-significant increase is consistent with previous work where we found that the e4 allele no longer plays a role in dementia and mortality at very old ages (Corrada et al., 2013). Brain weight was non-significantly lower in the dementia group.

Alzheimer’s disease pathology

In the normal group, Braak stage (median III; mean 3.3) and Thal phase (median 1; mean 1.9) scores indicate that Alzheimer’s disease pathology was present in at least mild to moderate amounts (Fig. 2D and E). The CIND group presented with similar Braak stage (median III; mean 3.4) and Thal phase (median 3; mean 2.3) scores, while Braak stage (median V; mean 4.3) and Thal phase (median 3; mean 2.6) were more moderate to severe in the dementia group. Higher Braak stages were significantly associated with higher odds of being in the dementia versus normal group [odds ratio (OR) = 1.68, P = 0.001] and of being in the dementia versus CIND group (OR = 1.52, P = 0.02) (Table 2). Higher Thal phases were associated with higher odds of being in the dementia versus the normal group (OR = 1.64, P = 0.011) but did not distinguish between the dementia and CIND groups (P > 0.15). Neither Alzheimer’s disease marker distinguished between the normal and CIND groups.

Figure 2

Synaptic intensity and neuropathology measures by cognitive group. Synaptic ratios (A–C) and neuropathology measures (D–F) in the oldest-old. All three synaptic RIRs were reduced in dementia compared to normal with (A) synaptophysin and (B) SV2 and significantly different between the three cognitive groups, however, (C) VGLUT1 did not distinguish between CIND and dementia. No synaptic RIR distinguished normal from CIND. (D) Thal phase and scores were all increased in dementia compared to normal with (D) similar Thal phases reported for CIND and dementia, whereas (E) Braak stage and (F) cerebrovascular disease (CVD) pathologies, including hippocampal sclerosis, were significantly different between the three cognitive groups. Each measure is graphed with the first and third quartiles boxed with the median band dividing the box into darker upper and lighter lower ranges; a whisker line extends to the minimums and maximums; shaded diamonds represent the means.

Figure 2

Synaptic intensity and neuropathology measures by cognitive group. Synaptic ratios (A–C) and neuropathology measures (D–F) in the oldest-old. All three synaptic RIRs were reduced in dementia compared to normal with (A) synaptophysin and (B) SV2 and significantly different between the three cognitive groups, however, (C) VGLUT1 did not distinguish between CIND and dementia. No synaptic RIR distinguished normal from CIND. (D) Thal phase and scores were all increased in dementia compared to normal with (D) similar Thal phases reported for CIND and dementia, whereas (E) Braak stage and (F) cerebrovascular disease (CVD) pathologies, including hippocampal sclerosis, were significantly different between the three cognitive groups. Each measure is graphed with the first and third quartiles boxed with the median band dividing the box into darker upper and lighter lower ranges; a whisker line extends to the minimums and maximums; shaded diamonds represent the means.

Table 2

Association between markers and cognitive groups

Measure Wald test CIND versus Normal Dementia versus CIND Dementia versus Normal 
P-value OR (95% CI) OR (95% CI) OR (95% CI) 
P-value P-value P-value 
Synaptophysin 0.002 5.59 (0.20–159.45) 0.01 (<0.01–0.11) 0.03 (<0.01–0.59) 
0.314 0.001 0.021 
SV2 0.017 2.85 (0.10–84.75) 0.02 (<0.01–0.35) 0.05 (<0.01–0.93) 
0.546 0.008 0.045 
VGLUT1 0.099 0.04 (<0.01–2.41) 0.56 (0.02–18.84) 0.02 (<0.01–0.76) 
0.124 0.747 0.034 
Braak stage 0.001 1.06 (0.75–1.50) 1.52 (1.07–2.16) 1.68 (1.24–2.30) 
0.740 0.020 0.001 
Thal phase 0.037 1.37 (0.89–2.11) 1.20 (0.84–1.71) 1.64 (1.12–2.38) 
0.158 0.318 0.011 
Hippocampal sclerosis <0.001 0.94 (0.36–2.49) 3.32 (1.56–7.05) 3.13 (1.48–6.64) 
0.906 0.002 0.003 
Cerebrovascular disease 0.004 0.94 (0.35–2.55) 2.99 (1.31–6.83) 2.81 (1.29–6.13) 
0.900 0.009 0.009 
TDP-43 0.008 0.90 (0.20–4.05) 4.32 (1.34–13.95) 3.89 (1.23–12.36) 
0.893 0.014 0.021 
Measure Wald test CIND versus Normal Dementia versus CIND Dementia versus Normal 
P-value OR (95% CI) OR (95% CI) OR (95% CI) 
P-value P-value P-value 
Synaptophysin 0.002 5.59 (0.20–159.45) 0.01 (<0.01–0.11) 0.03 (<0.01–0.59) 
0.314 0.001 0.021 
SV2 0.017 2.85 (0.10–84.75) 0.02 (<0.01–0.35) 0.05 (<0.01–0.93) 
0.546 0.008 0.045 
VGLUT1 0.099 0.04 (<0.01–2.41) 0.56 (0.02–18.84) 0.02 (<0.01–0.76) 
0.124 0.747 0.034 
Braak stage 0.001 1.06 (0.75–1.50) 1.52 (1.07–2.16) 1.68 (1.24–2.30) 
0.740 0.020 0.001 
Thal phase 0.037 1.37 (0.89–2.11) 1.20 (0.84–1.71) 1.64 (1.12–2.38) 
0.158 0.318 0.011 
Hippocampal sclerosis <0.001 0.94 (0.36–2.49) 3.32 (1.56–7.05) 3.13 (1.48–6.64) 
0.906 0.002 0.003 
Cerebrovascular disease 0.004 0.94 (0.35–2.55) 2.99 (1.31–6.83) 2.81 (1.29–6.13) 
0.900 0.009 0.009 
TDP-43 0.008 0.90 (0.20–4.05) 4.32 (1.34–13.95) 3.89 (1.23–12.36) 
0.893 0.014 0.021 

Odds ratios and 95% CIs were generated from multinomial logistic regression models where the dependent (outcome) variable was cognitive group (normal, CIND, or dementia) and the independent variables were the neuropathological markers as continuous variables, except TARDBP which was a binary variable. P-values <0.05 are in bold. Age at death and gender were included as covariates and each neuropathological measure was analysed in a separate model. The overall P-value corresponds to the type 3 analyses and tests whether the neuropathological marker (independent variable) is significantly associated with the cognitive group (outcome variable). Odds ratios are per unit. n for each group = 149 (synaptophysin), 151 (SV2), 152 (VGLUT1), 157 (Braak stage), 150 (Thal phase), 155 (hippocampal sclerosis, 108 (cerebrovascular disease) and 146 (TARDBP).

Higher levels of Alzheimer’s disease plaque and tangle burdens were significantly associated with lower MMSE scores (Thal phase, P = 0.02; Braak stage, P < 0.001). The greatest decreases on the cognitive tests occurred with the highest plaque and tangle burdens. The mean MMSE score was significantly lower in Thal phase 4 compared to Thal phases 2 and earlier (P < 0.01) and even to Thal phase 3 (P < 0.05). For Braak stage, the mean MMSE score was significantly lower in stage VI compared to all the other Braak stages (P < 0.001), and in stage V compared to stages III and IV (P < 0.05).

Hippocampal sclerosis and cerebrovascular disease

As the prevalence of hippocampal sclerosis and cerebrovascular disease pathology increases with age, we investigated what proportion of individuals had these lesions in our cohort. Hippocampal sclerosis was found in 15% of all individuals, but in only one individual without dementia. The presence of hippocampal sclerosis in 28% of those with dementia (23/82) allowed the measure to significantly distinguish the dementia group from both the normal (P = 0.003) and CIND (P = 0.002) groups (Table 2).

Cerebrovascular lesions and cerebral amyloid angiopathy were present in modest amounts in The 90+ Study. Cerebral amyloid angiopathy was found in 52% (56/108) of the cases whereas cerebrovascular disease lesions such as infarcts, micro-infarcts or micro-bleeds were rarer (10%; 11/108). Althogether, 63% of all individuals had some level of cerebrovascular disease pathology, including 45% of both the normal and CIND groups and 74% of the dementia group. Although the prevalence of cerebrovascular disease was not different between the normal and CIND groups (Fig. 2F), people with higher levels were more likely to be in the dementia group compared to both the normal and CIND groups (both P = 0.009) (Table 2).

TDP-43

TDP-43 is a common co-morbidity and may associate with a more rapid cognitive decline in the ageing brain (Wilson et al., 2013). In our cohort, TDP-43 pathology was present in the hippocampus of 24% (35/146) of the cases. While a small number of both normal (11%, 4/35) and CIND (12%, 4/33) individuals had pathology, TARDBP inclusions affected a more substantial portion of the dementia group (35%, 27/78). As in our earlier study (Robinson et al., 2011), TARDBP was almost exclusively in individuals with hippocampal sclerosis. 77% (17/22) of the cases with hippocampal sclerosis also had TARDBP inclusions. While 15% (18/124) of individuals without hippocampal sclerosis also had TARDBP pathology, the numbers were too low to analyse the significance of the TARDBP pathology after controlling for hippocampal sclerosis (data not shown).

Synaptic health

As described above, to measure the synaptic levels of the perforant pathway, synaptic RIRs were generated by measuring the intensity of synaptophysin, SV2, and VGLUT1 by immunohistochemistry in the outer molecular layer of the hippocampus relative to the inner molecular layer (Fig. 1). Higher RIRs represent a relatively healthy outer layer in relation to the inner layer, whereas lower RIRs correspond to synaptic loss in the outer layer (Fig, 2A–C). In the normal group, synaptophysin (mean 0.94) and SV2 (mean 0.94) had RIRs close to 1.00, whereas the glutamate-specific VGLUT1 RIR was lower (mean 0.67). Similar values were obtained in the CIND group: synaptophysin (mean 0.98), SV2 (mean 0.96) and VGLUT1 (mean 0.64). In the dementia group, these values were mildly lower: synaptophysin (mean 0.87), SV2 (mean 0.89) and VGLUT1 (mean 0.63).

All three synaptic RIRs distinguished between the dementia and normal groups with higher RIRs associated with lower odds of being in the dementia group (P < 0.05) (Table 2). The synaptophysin and SV2 RIRs also distinguished between the dementia and CIND groups (synaptophysin, P < 0.001; SV2, P = 0.008), whereas the VGLUT1 RIR did not (P = 0.747). No RIR distinguished between the CIND and normal groups (all P > 0.12). We also examined the association between synaptic RIRs and global cognitive scores and found that for all three synaptic markers, higher RIR values were significantly associated with higher MMSE scores (synaptophysin, P < 0.001; SV2, P < 0.001, VGLUT1, P < 0.05).

Multiple pathologies

Pearson correlation coefficients showed significant correlations between the three synaptic RIRs (VGLUT1 and SV2, corr = 0.37, P < 0.001; VGLUT1 and synaptophysin, corr = 0.45, P < 0.001; SV2 and synaptophysin, corr = 0.50, P < 0.001). Given the correlation between synaptic markers, we wanted to explore the independent contribution of each with respect to global cognitive scores. In a multiple regression model that included all three synaptic markers, we found that higher RIRs for both synaptophysin (beta = 20.3, P = 0.002) and SV2 (beta = 13.4, P = 0.03) remained significantly associated with better MMSE scores.

Post hoc analysis revealed that the SV2 RIR was significantly lower at Braak stage VI (P < 0.001) and the VGLUT1 RIR was significantly lower in individuals with hippocampal sclerosis (P = 0.024). To estimate the independent contribution of all the different measures to global cognitive scores, we included all three synaptic RIRs, the Alzheimer’s disease pathology scores, and hippocampal sclerosis in a multiple regression analysis. In this model, the synaptophysin ratio, Braak stage, TARDBP and hippocampal sclerosis all remained significantly associated with MMSE scores (P-values: < 0.05, 0.001, <0.01 and 0.001, respectively).

Memory impairment only subjects

The primary impairment in 13 of the CIND individuals was memory (Table 1). As those with executive impairment may have dysfunction related more to a frontotemporal lobar degeneration rather than Alzheimer’s disease and those with ‘other’ impairment cannot be well defined, it’s possible that the 13 with memory impairment are more likely to be on the continuum between normal cognition and Alzheimer’s disease. To test this hypothesis, we compared the RIRs and our pathological markers in the ‘memory impairment only’ CIND subjects to the normal and dementia subjects to see if our results had more significance. No new associations were found (data not shown), suggesting that the memory impairment subjects weren’t more likely to lead to the intermediate level of Alzheimer’s disease apparent in the dementia group than the other CIND individuals. Similar subgroup analysis with the executive impairment and ‘other’ impairment CIND individuals also did not reveal any new associations that the comparisons with the group as a whole hadn’t already shown (data not shown).

Discussion

We hypothesized that perforant pathway synaptic loss may be one of the early correlates of cognitive impairment in The 90+ Study, but this was not the case. None of the measures distinguished the normal and CIND groups and almost all the measures distinguished the CIND and dementia groups from the cognitively normal individuals. This implies that synaptic loss is tightly linked to pathology associated with dementia in the oldest-old.

This is not the case in younger age groups where it is hypothesized that plaques and neurofibrillary tangles accumulate years, even decades, before cognitive impairment and where pathology accumulates progressively (Sperling et al., 2011). Perhaps nonagenarians represent individuals with significant cognitive reserve? Although their brains may be exposed to the same pathological insults during progressive cognitive impairment as younger cohorts, they are able to minimize any synaptic loss and accumulations of pathology. At the very least, our data are consistent with an acceleration of the pathological cascade implied by other studies.

In this autopsy study of a large cohort of CIND individuals, we sought to elucidate which pathologies among those examined here associate with normal ageing, which associate with dementia and which, if any, associate with cognitive impairment. We review all of our findings below.

First, cognitively normal nonagenarians in The 90+ Study have a median Braak stage of III and a Thal phase of 1 consistent with predictable, age-dependent deposition of Alzheimer’s disease pathology (Braak et al., 2011). Of course, many pathologies besides Alzheimer’s disease affect the ageing brain including other neurodegenerative diseases, cerebrovascular disease and hippocampal sclerosis among others. In this study, we did not examine the role of multiple neurodegenerative diseases, but we previously reported that α-synuclein and TARDBP pathology affect <20% of cognitively normal individuals in this age category and then primarily as co-morbidities (Robinson et al., 2011). Cerebrovascular disease, on the other hand, was common, affecting 45% of the group while hippocampal sclerosis was almost non-existent (n = 1).

Against this background level of multiple age-related pathologies in cognitively normal individuals, we can analyse the increases in Alzheimer’s disease burden, cerebrovascular disease and hippocampal sclerosis associated with dementia (Fig. 3A). As 90+ individuals have a moderate burden of age-related tangles and plaques and a particularly high prevalence of dementia (Corrada et al., 2008), it is not surprising that Alzheimer’s disease pathology is the most common underlying contributor to dementia in this group (Hebert et al., 2013). For individuals with dementia, the median Braak stage is V and the Thal phase of 3. Compared with the normal group, these are substantial increases consistent with an intermediate burden of Alzheimer’s disease neuropathological change (Montine et al., 2012). Along with Alzheimer’s disease, the prevalence of cerebrovascular disease—not including hippocampal sclerosis—increased to 74% of individuals in the dementia group. Additionally, dementia was associated with the appearance of hippocampal sclerosis in a substantial minority (28%). Similarly, and primarily comorbidly, TARDBP also affected a large minority (35%). Both the cerebrovascular disease and hippocampal sclerosis percentages are consistent with reported frequencies of these pathologies in this age-group (Giannakopoulos et al., 1997; Nelson et al., 2011a).

Figure 3

A model of changes associated with cognitive impairment and dementia in the oldest-old. Neuropathology and synaptic measures are associated with cognitive impairment in ‘The 90+ Study’. (A) Mild to moderate Alzheimer’s disease pathology already exists in the oldest-old for cognitively normal individuals and the Thal phase may increase with cognitive impairment. With dementia, all markers increase significantly, especially Braak stage, cerebrovascular disease (CVD) pathology and hippocampal sclerosis (HS). (B) Glutamatergic specific synaptic loss (VGLUT1) in the perforant pathology may occur during cognitive impairment, while significant synaptic loss occurs with dementia. All values are relative to the normal group’s means normalized to 1.0 (see Fig. 1), except cerebrovascular disease which factors out the contribution of hippocampal sclerosis before normalizing; the hippocampal sclerosis percentages are graphed separately on the right axis.

Figure 3

A model of changes associated with cognitive impairment and dementia in the oldest-old. Neuropathology and synaptic measures are associated with cognitive impairment in ‘The 90+ Study’. (A) Mild to moderate Alzheimer’s disease pathology already exists in the oldest-old for cognitively normal individuals and the Thal phase may increase with cognitive impairment. With dementia, all markers increase significantly, especially Braak stage, cerebrovascular disease (CVD) pathology and hippocampal sclerosis (HS). (B) Glutamatergic specific synaptic loss (VGLUT1) in the perforant pathology may occur during cognitive impairment, while significant synaptic loss occurs with dementia. All values are relative to the normal group’s means normalized to 1.0 (see Fig. 1), except cerebrovascular disease which factors out the contribution of hippocampal sclerosis before normalizing; the hippocampal sclerosis percentages are graphed separately on the right axis.

The changes in all three of the synaptic markers studied here are consistent with the interpretation that significant perforant pathway loss occurs with dementia in the 90+ subjects (Fig. 3B). This is consistent with our previous work involving 32 individuals in The 90+ Study where neocortical synaptophysin protein levels were significantly decreased with dementia (Head et al., 2009). In our multiple logistic regression analysis, both the synaptophysin RIR and Braak stage tightly associated with cognitive scores. Concurrently, post hoc analysis revealed that the SV2 RIR was significantly lower at Braak stage VI, raising the possibility that synaptic protein loss is tightly correlated with tau burden. This is supported by studies in tau transgenic mice (Yoshiyama et al., 2007) and in human cohorts where synaptic protein loss only occurs at the highest Braak stages (Mukaetova-Ladinska et al., 2000). On the other hand, as the synaptophysin RIR, Braak stage, hippocampal sclerosis and the TDP-43 pathology measures all remained significantly associated with MMSE scores in our multiple logistic regression analysis, perhaps something more than pathological Alzheimer’s disease is taking place and that synaptic protein loss in the perforant pathway and hippocampal sclerosis are independent factors that contribute to dementia in the oldest-old.

If cognitively normal individuals have age-related tangles and plaques, and these pathologies increase along with synaptic loss in dementia, what changes occur during CIND? CIND in the oldest-old may be associated with a greater substantial distribution of plaques as measured by Thal phase (Fig. 3A). Cognitively normal nonagenarians at Thal phase 2 have plaque deposition that has spread into layers 2/3 of the entorhinal cortex (Thal et al., 2006), suggesting that plaques are already present in the perforant pathway. In the CIND group, the median Thal phase was 3 indicating a greater plaque burden affecting the perforant pathway. As there is no concurrent increase in the Braak stage (median III), this may signify that plaques are having an earlier effect than tangles, which is consistent with numerous biomarker studies that show amyloid-β becomes abnormal before tau in cognitively normal individuals (Aisen, 2010; Jack et al., 2013).

CIND may also be associated with a decrease in glutamatergic synapses in the perforant pathway as measured by the VGLUT1 RIR (Fig. 3B). Although the VGLUT1 RIR can clearly distinguish the normal and dementia groups (P = 0.034), this measure does not distinguish the CIND from the dementia groups (P = 0.747), whereas the synaptophysin (P < 0.001) and SV2 RIRs (P = 0.008) still do (Table 2). The most parsimonious explanation is that non-glutamatergic neurons remain relatively unaffected during CIND whereas glutamatergic neurons experience synaptic loss. If true, this is consistent with previous work showing that observed presynaptic proteins, such as synaptophysin, remain constant or increase in earlier phases of illness before decreasing with dementia onset (Honer, 2003; Head et al., 2009). That the decrease in the VGLUT1 RIR is coincident with the increase in Thal phase may imply that plaques damage glutermatergic synapses. Of course, there are many possibilities including that the reduction in VGLUT1 may simply signify a loss of transporters, rather than the loss of glutamatergic afferents (Tannenberg et al., 2004).

Importantly, our study suggests that individuals with CIND are for the most part pathologically and synaptically indistinguishable from individuals with normal cognition. This being the case, pharmaceutical interventions that target tau and amyloid-β or protect synapses are all viable therapeutic strategies for a population at risk of dementia due to Alzheimer’s disease. In addition, interventions designed to maintain a healthy synaptic system and a strict control of vascular risk factors may be good strategies to preserve cognitive function in nonagenarians and centenarians.

Acknowledgements

We wish to thank the assistance of Terry Schuck, Felix Geser and Laura Volpicelli-Daley.

Funding

This work was supported in part by grants from the National Institutes of Health (AG10124, AG16573, AG17586, AG21055, P50AG16573 and MH64045).

Abbreviations

    Abbreviations
  • CIND

    cognitively impaired no dementia

  • MMSE

    Mini-Mental State Examination

  • RIR

    relative immunointensity ratio

References

Aisen
PS
Pre-dementia Alzheimer’s trials: overview
J Nutr Health Aging
 , 
2010
, vol. 
14
 pg. 
294
 
Albert
MS
Moss
MB
Tanzi
R
Jones
K
Preclinical prediction of AD using neuropsychological tests
J Int Neuropsychol Soc
 , 
2001
, vol. 
7
 (pg. 
631
-
639
)
 
American Psychiatric Association, American Psychiatric Association, and Task Force on DSM-IV. Diagnostic and statistical manual of mental disorders: DSM-IV. Washington, DC: American Psychiatric Association; 1994
Bakker
A
Krauss
GL
Albert
MS
Speck
CL
Jones
LR
Stark
CE
, et al.  . 
Reduction of hippocampal hyperactivity improves cognition in amnestic mild cognitive impairment
Neuron
 , 
2012
, vol. 
74
 (pg. 
467
-
474
)
Beeri
MS
Haroutunian
V
Schmeidler
J
Sano
M
Fam
P
Kavanaugh
A
, et al.  . 
Synaptic protein deficits are associated with dementia irrespective of extreme old age
Neurobiol Aging
 , 
2012
, vol. 
33
 pg. 
1125.e1–8
 
Berlau
DJ
Corrada
MM
Head
E
Kawas
CH
APOE epsilon2 is associated with intact cognition but increased Alzheimer pathology in the oldest old
Neurology
 , 
2009
, vol. 
72
 (pg. 
829
-
834
)
Braak
H
Alafuzoff
I
Arzberger
T
Kretzschmar
H
Del Tredici
K
Staging of Alzheimer disease-associated neurofibrillary pathology using paraffin sections and immunocytochemistry
Acta Neuropathol
 , 
2006
, vol. 
112
 (pg. 
389
-
404
)
Braak
H
Thal
DR
Ghebremedhin
E
Del Tredici
K
Stages of the pathologic process in Alzheimer disease: age categories from 1 to 100 years
J Neuropathol Exp Neurol
 , 
2011
, vol. 
70
 (pg. 
960
-
969
)
Chen
P
Ratcliff
G
Belle
SH
Cauley
JA
DeKosky
ST
Ganguli
M
Patterns of cognitive decline in presymptomatic Alzheimer disease: a prospective community study
Arch Gen Psychiatry
 , 
2001
, vol. 
58
 (pg. 
853
-
858
)
Clark
CM
Ewbank
DC
Performance of the dementia severity rating scale: a caregiver questionnaire for rating severity in Alzheimer disease
Alzheimer Dis Assoc Disord
 , 
1996
, vol. 
10
 (pg. 
31
-
39
)
Corrada
MM
Berlau
DJ
Kawas
CH
A population-based clinicopathological study in the oldest-old: the 90+ study
Curr Alzheimer Res
 , 
2012
, vol. 
9
 (pg. 
709
-
717
)
Corrada
MM
Brookmeyer
R
Berlau
D
Paganini-Hill
A
Kawas
CH
Prevalence of dementia after age 90: results from the 90+ study
Neurology
 , 
2008
, vol. 
71
 (pg. 
337
-
343
)
Corrada
MM
Paganini-Hill
A
Berlau
DJ
Kawas
CH
Apolipoprotein E genotype, dementia, and mortality in the oldest old: the 90+ Study
Alzheimers Dement
 , 
2013
, vol. 
9
 (pg. 
12
-
18
)
DeKosky
ST
Scheff
SW
Styren
SD
Structural correlates of cognition in dementia: quantification and assessment of synapse change
Neurodegener J Neurodegener Disord Neuroprotection Neuroregeneration
 , 
1996
, vol. 
5
 (pg. 
417
-
421
)
El Mestikawy
S
Wallén-Mackenzie
A
Fortin
GM
Descarries
L
Trudeau
LE
From glutamate co-release to vesicular synergy: vesicular glutamate transporters
Nat Rev Neurosci
 , 
2011
, vol. 
12
 (pg. 
204
-
216
)
Giannakopoulos
P
Hof
PR
Michel
JP
Guimon
J
Bouras
C
Cerebral cortex pathology in aging and Alzheimer’s disease: a quantitative survey of large hospital-based geriatric and psychiatric cohorts
Brain Res Brain Res Rev
 , 
1997
, vol. 
25
 (pg. 
217
-
245
)
Head
E
Corrada
MM
Kahle-Wrobleski
K
Kim
RC
Sarsoza
F
Goodus
M
, et al.  . 
Synaptic proteins, neuropathology and cognitive status in the oldest-old
Neurobiol Aging
 , 
2009
, vol. 
30
 (pg. 
1125
-
1134
)
Hebert
LE
Weuve
J
Scherr
PA
Evans
DA
Alzheimer disease in the United States (2010–2050) estimated using the 2010 census
Neurology
 , 
2013
, vol. 
80
 (pg. 
1778
-
1783
)
Honer
WG
Pathology of presynaptic proteins in Alzheimer’s disease: more than simple loss of terminals
Neurobiol Aging
 , 
2003
, vol. 
24
 (pg. 
1047
-
1062
)
Hyman
BT
Van Hoesen
GW
Kromer
LJ
Damasio
AR
Perforant pathway changes and the memory impairment of Alzheimer’s disease
Ann Neurol
 , 
1986
, vol. 
20
 (pg. 
472
-
481
)
Jack
CR
Jr
Knopman
DS
Jagust
WJ
Petersen
RC
Weiner
MW
Aisen
PS
, et al.  . 
Tracking pathophysiological processes in Alzheimer’s disease: an updated hypothetical model of dynamic biomarkers
Lancet Neurol
 , 
2013
, vol. 
12
 (pg. 
207
-
216
)
Jellinger
KA
Attems
J
Incidence of cerebrovascular lesions in Alzheimer’s disease: a postmortem study
Acta Neuropathol
 , 
2003
, vol. 
105
 (pg. 
14
-
17
)
Kawas
C
Segal
J
Stewart
WF
Corrada
M
Thal
LJ
A validation study of the Dementia Questionnaire
Arch Neurol
 , 
1994
, vol. 
51
 (pg. 
901
-
906
)
Kirvell
SL
Esiri
M
Francis
PT
Down-regulation of vesicular glutamate transporters precedes cell loss and pathology in Alzheimer’s disease
J Neurochem
 , 
2006
, vol. 
98
 (pg. 
939
-
950
)
Montine
TJ
Phelps
CH
Beach
TG
Bigio
EH
Cairns
NJ
Dickson
DW
, et al.  . 
National Institute on Aging-Alzheimer’s Association guidelines for the neuropathologic assessment of Alzheimer’s disease: a practical approach
Acta Neuropathol
 , 
2012
, vol. 
123
 (pg. 
1
-
11
)
Mukaetova-Ladinska
EB
Garcia-Siera
F
Hurt
J
Gertz
HJ
Xuereb
JH
Hills
R
, et al.  . 
Staging of cytoskeletal and beta-amyloid changes in human isocortex reveals biphasic synaptic protein response during progression of Alzheimer’s disease
Am J Pathol
 , 
2000
, vol. 
157
 (pg. 
623
-
636
)
Nelson
PT
Head
E
Schmitt
FA
Davis
PR
Neltner
JH
Jicha
GA
, et al.  . 
Alzheimer’s disease is not “brain aging”: neuropathological, genetic, and epidemiological human studies
Acta Neuropathol
 , 
2011a
, vol. 
121
 (pg. 
571
-
587
)
Nelson
PT
Schmitt
FA
Lin
Y
Abner
EL
Jicha
GA
Patel
E
, et al.  . 
Hippocampal sclerosis in advanced age: clinical and pathological features
Brain J Neurol
 , 
2011b
, vol. 
134
 (pg. 
1506
-
1518
)
Peltz
CB
Corrada
MM
Berlau
DJ
Kawas
CH
Cognitive impairment in nondemented oldest-old: prevalence and relationship to cardiovascular risk factors
Alzheimers Dement
 , 
2012
, vol. 
8
 (pg. 
87
-
94
)
Pfeffer
RI
Kurosaki
TT
Harrah
CH
Jr
Chance
JM
Filos
S
Measurement of functional activities in older adults in the community
J Gerontol
 , 
1982
, vol. 
37
 (pg. 
323
-
329
)
Robinson
JL
Geser
F
Corrada
MM
Berlau
DJ
Arnold
SE
Lee
VMY
, et al.  . 
Neocortical and hippocampal amyloid-β and tau measures associate with dementia in the oldest-old
Brain J Neurol
 , 
2011
, vol. 
134
 (pg. 
3708
-
3715
)
Saxton
J
Lopez
OL
Ratcliff
G
Dulberg
C
Fried
LP
Carlson
MC
, et al.  . 
Preclinical Alzheimer disease: neuropsychological test performance 1.5 to 8 years prior to onset
Neurology
 , 
2004
, vol. 
63
 (pg. 
2341
-
2347
)
Senut
MC
Roudier
M
Davous
P
Fallet-Bianco
C
Lamour
Y
Senile dementia of the Alzheimer type: is there a correlation between entorhinal cortex and dentate gyrus lesions?
Acta Neuropathol
 , 
1991
, vol. 
82
 (pg. 
306
-
315
)
Silverman
JM
Breitner
JC
Mohs
RC
Davis
KL
Reliability of the family history method in genetic studies of Alzheimer’s disease and related dementias
Am J Psychiatry
 , 
1986
, vol. 
143
 (pg. 
1279
-
1282
)
Sperling
RA
Aisen
PS
Beckett
LA
Bennett
DA
Craft
S
Fagan
AM
, et al.  . 
Toward defining the preclinical stages of Alzheimer’s disease: recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease
Alzheimers Dement
 , 
2011
, vol. 
7
 (pg. 
280
-
292
)
Sze
CI
Bi
H
Kleinschmidt-DeMasters
BK
Filley
CM
Martin
LJ
Selective regional loss of exocytotic presynaptic vesicle proteins in Alzheimer’s disease brains
J. Neurol Sci
 , 
2000
, vol. 
175
 (pg. 
81
-
90
)
Tannenberg
RK
Scott
HL
Westphalen
RI
Dodd
PR
The identification and characterization of excitotoxic nerve-endings in Alzheimer disease
Curr Alzheimer Res
 , 
2004
, vol. 
1
 (pg. 
11
-
25
)
Thal
DR
Capetillo-Zarate
E
Del Tredici
K
Braak
H
The development of amyloid beta protein deposits in the aged brain
Sci. Aging Knowl Environ
 , 
2006
, vol. 
2006
 pg. 
re1
 
Valtorta
F
Pennuto
M
Bonanomi
D
Benfenati
F
Synaptophysin: leading actor or walk-on role in synaptic vesicle exocytosis?
BioEssays News Rev Mol Cell Dev Biol
 , 
2004
, vol. 
26
 (pg. 
445
-
453
)
Wilson
RS
Yu
L
Trojanowski
JQ
Chen
EY
Boyle
PA
Bennett
DA
, et al.  . 
TDP-43 Pathology, cognitive decline, and dementia in old age
JAMA Neurol
 , 
2013
, vol. 
70
 (pg. 
1418
-
1424
)
Yoshiyama
Y
Higuchi
M
Zhang
B
Huang
SM
Iwata
N
Saido
TC
, et al.  . 
Synapse loss and microglial activation precede tangles in a P301S tauopathy mouse model
Neuron
 , 
2007
, vol. 
53
 (pg. 
337
-
351
)