Pathological Correlates of Cognitive Impairment in The University of Manchester Longitudinal Study of Cognition in Normal Healthy Old Age

Abstract

The neuropathological changes responsible for cognitive impairment and dementia remain incompletely understood. Longitudinal studies with a brain donation end point allow the opportunity to examine relationships between cognitive status and neuropathology. We report on the first 97 participants coming to autopsy with sufficient clinical information from The University of Manchester Longitudinal Study of Cognition in Normal Healthy Old Age. This study began in 1983 and recruited 6,542 healthy individuals between 1983 and 1994, 312 of whom consented to brain donation. Alzheimer-type pathology was common throughout the cohort and generally correlated well with cognitive status. However, there was some overlap between cognitive status and measures of Alzheimer pathology with 26% of cognitively intact participants reaching either CERAD B or C, 11% reaching Thal phase 4 or 5, and 29% reaching Braak stage III– VI. Cerebral amyloid angiopathy(CAA), α-synuclein, and TDP-43 pathology was less common, but when present correlated well with cognitive status. Possession of APOE ɛ4 allele(s) was associated with more severe Alzheimer-type and CAA pathology and earlier death, whereas possession of APOE ɛ2 allele(s) had no effect on pathology but was more common in cognitively intact individuals. The University of Manchester Longitudinal Study of Cognition in Normal Healthy Old Age cohort is pathologically representative when compared with similar studies. Cognitive impairment in life correlates strongly with all pathologies examined and the APOE status of an individual can affect pathology severity and longevity.

Keywords

Alpha-synuclein Alzheimer’s disease cognitive dysfunction cohort studies dementia longitudinal studies neuropathology

INTRODUCTION

Longitudinal studies that are community or population-based and have brain donation end points offer an opportunity to examine correlations between pathology and cognitive impairment and are therefore fundamentally important for the field of dementia research. However, true unbiased community-based longitudinal studies are rare as it is common practice for epidemiological studies of dementia to use cohorts selected according to their cognitive status, age, gender, or ethnicity. Nonetheless, there have been several longitudinal studies of brain aging and dementia that include brain autopsy end points: The Medical Research Council Cognitive Function and Ageing Studies (CFAS) [1, 2], the Nun Study [3], the Religious Orders Study [4], the Baltimore Longitudinal Study of Ageing (BLSA) [5], the Cambridge City over 75 Cohort Study (CC75C) [6], the Honolulu-Asia Aging Study (HAAS) [7], the Oxford Project to Investigate Memory and Aging samples (OPTIMA) [8], the Vantaa 85+ Study [9], and the Adult Changes in Thought Study (ACT) [10]. Demographic characteristics of these studies are shown in Table 1. Most commenced in the late 1980 s to early 1990 s and were based either on healthy volunteers of all ages from local communities, or selected cohorts based either on age specification or particular lifestyle which included both cognitively normal and cognitively impaired individuals. Cohort size has ranged from a few hundred to many thousand individuals with brain donations ranging from 180–500 (at time of last publication). Most are ongoing, though the Honolulu-Asia and Optima studies have now closed.

Table 1

Overview of demographic characteristics from longitudinal studies of brain aging and dementia that include a brain donation end point. CI, cognitive impairment

Study	Commenced	Cohort size	Recruitment age	Cognitive status	Demographics	Donations To date
MRC CFAS	1989	18,000	65+	Normal/CI	Population-based	500+
Nun study	1986	678	75+	Random	Selected (women)	221+
Religious Orders Study	1994	1,000+	Aged	Normal	Selected (Roman Catholic clergy)	180+
BLSA	1958	1,400	20–90	Normal	Community	335+
CC75C	1985	2,600	≤75	Normal/CI	Population-based	213+
HAAS	1991–2012	3,734	71–93	Random	Population-based (Japanese-American men)	443+
OPTIMA	1988–2008	1,100+	Aged (65+)	Normal/CI	Community	300
Vantaa 85+	1991	601	85+	Random	Population-based (South Finland)	304
ACT	1994	2,581	65+	Random	Community (King’s County, WA)	232+
University of Manchester	1983	6,542	42–92	Normal	Community (Manchester and Newcastle)	107+

The present report is based on The University of Manchester Longitudinal Study of Cognition in Normal Healthy Old Age [11]. This study began in 1983 and recruited, via local advertisement, 6,542 healthy individuals aged between 42 and 92 years. People with evidence of cognitive decline/dementia at the time of recruitment were not eligible for the study. Hence, this study represents one of the longest running studies in which cognitively healthy individuals at the outset have been followed up for periods of 30 years or more. From 2003, 312 of the surviving individuals consented to brain donation of which 100+ have subsequently died and their brains have become available for investigation. Here, we report neuropathological findings on the first 97 cases coming to donation.

MATERIALS AND METHODS

Participants and study design

Participants from The University of Manchester Longitudinal Study of Cognition in Normal Healthy Old Age [11] were approached (in 2003) for consent to brain donation. From the original recruited total of 6,542 healthy individuals (aged between 42 and 92 years), 312 individuals consented to brain donation. Participants had demographic, education, lifestyle, and health information collected through study-specific self-report questionnaires. Information regarding educational level were standardized using the International Standard Classification of Education (ISCED) guidelines [12].

Over five waves between 2004 and 2017, surviving participants underwent assessment by the modified Telephone Instrument for Cognitive Status (TICSm) which contains 13 questions testing orientation, concentration, immediate and delayed memory, naming, calculation, comprehension, and reasoning. The TICSm test had a maximum score of 39 [13] and the cut-off point, which was used to define cognitive impairment in the present study, was a score below 21 [14].

Cognitive status at death was ascertained using a combination of last TICSm score, patient notes obtained via participants’ general practitioner and cause of death as recorded on the death certificate. The first 97 brains donated were accessioned into the present study.

Research Ethics Committee approval

The study was approved by Manchester Brain Bank Management Committee (REC reference 09/H0906/52+5). Under conditions agreed with the Research Ethics Committee, The Manchester Brain Bank can supply tissue or data to researchers, without requirement for researchers to apply individually to the Research Ethics Committee for approval.

Pathological methods

One fresh hemi-brain was fixed in 10% neutral buffered formalin for 3-4 weeks with the other hemi-brain frozen at – 80°C. Standard blocks of frontal, cingulate, temporal (including superior and middle temporal gyrus), hippocampus, parietal and occipital cortex, amygdala, corpus striatum, thalamus, midbrain, brainstem, and cerebellum were cut from the fixed tissue and processed into wax blocks. Paraffin sections (6μm) were immunostained for Aβ (Cambridge Bioscience, clone 4G8, 1:3000), tau proteins phosphorylated at Ser202 and Thr205 (P-tau) (Innogenetics, clone AT8, 1:750), phosphorylated α-synuclein (#1175) [15] (1:1000) and TDP-43 (Proteintech, 1:1000). For antigen retrieval, sections were either immersed in 70% formic acid for 20 min (for Aβ only) or microwaved in 0.1M citrate buffer, pH 6.0 (all other antibodies) prior to incubation with primary antibody.

The presence and severity of Aβ and tau was assessed in all regions examined. A five-point semi-quantitative grading scale was adopted: 0 – No Aβ/tau pathology present; 1 – Rare Aβ/tau pathology present; 2 – Mild Aβ/tau pathology present; 3 – Moderate Aβ/tau pathology present; 4 – Severe Aβ/tau pathology present.

A CERAD score [16], Thal phase [17], and Braak stage [18] were also assigned and an identical five-point semi-quantitative grading scale was used to assess the presence and severity of cerebral amyloid angiopathy (CAA) throughout the brain.

Vascular pathology was assessed using the Vascular Cognitive Impairment Neuropathology Guidelines (VCING). The presence of at least one large infarct, moderate to severe small vessel disease and moderate to severe CAA was used to assign low, moderate or high risk of vascular pathology contributing to cognitive impairment [19].

The presence/absence of TDP-43 pathology was assessed in the temporal cortex and hippocampus whereas the presence/absence of phosphorylated α-synuclein pathology was assessed in the cingulate gyrus and the substantia nigra region of the midbrain.

The clinical and neuropathological diagnosis was not known to the examiner (AR), who rated all cases. Preliminary scoring of the sections was also hidden during subsequent slide re-examination.

DNA was extracted from frozen brain tissue using REDExtract-N-Amp™ Tissue PCR Kit (Sigma) or from blood (3 cases). The APOE genotype was determined using routine polymerase chain reaction (PCR) methods [20]. APOE could not be determined for two participants because of lack of blood or frozen brain tissue.

Neuropathological diagnoses

After neuropathological assessment was completed, neuropathological diagnosis was assigned by Professor David Mann (Professor of Neuropathology).

Cases were assigned to a neuropathological category that accurately described the principal pathology found. Those considered normal for age included cases that were essentially free from pathology (pathology score 0) and those with a burden of pathology expected for age (pathology scores 1 and 2). The category of limited Aβ/tau pathology was assigned to cases that had some degree of AD pathology (pathology score 3) but insufficient for a full neuropathological diagnosis of AD (pathology score 4).

Statistical analyses

The cohort was split according to severity of pathology with one group representing low severity (pathology scores 0, 1, and 2) and the other high severity (pathology scores 3 and 4). Similar groups were established for CERAD, Thal phase, CAA severity, Braak stage, VCING, presence/absence of TDP-43 pathology, and presence/absence of α-synuclein pathology.

Chi-squared test was used to analyze whether there were differences in severity of pathology according to cognitive status. Mann-Whitney test assessed differences in educational level between cognitive status groups. A p value of <0.05 was considered significant.

Similarly, when assessing age at death, the cohort was split according to age at death into two groups: Under 90 years and 90 years and over.

Chi-squared test was used to analyze whether there were differences in frequencies of APOE ɛ2 or ɛ4 alleles according to age group at death. A p value of <0.05 was considered significant.

The odds ratio of age group at death according to presence of APOE ɛ2 or ɛ4 alleles was ascertained using a binary logistic regression model. Age group at death was the dependent variable and presence of APOE ɛ2 allele(s), APOE ɛ4 allele(s) and cognitive impairment were the covariates. A p value of <0.05 was considered significant.

RESULTS

The first brain donation accrued into the study occurred on 11/03/2005 with the last brain donation for the study taking place on 21/11/2016. The mean time between final TICSm test and death was 42 months (±30) with a range of 134 months. The ratio of women to men in the study was approximately 2:1. The median age at death for the 97 participants was 89 (range 72 to 104 years) and there were no differences in age group at death between men and women (χ² = 2.04; p = 0.15). The proportion of participants with cognitive impairment was 41% with no differences in cognitive status when comparing sex (χ² = 0.62; p = 0.43), age group (χ² = 0.28; p = 0.60) or education level (U = 1006.5; p = 0.38). APOE ɛ4 allele(s) were more common in cognitively impaired individuals (40%) than cognitive intact individuals (25%), this difference being marginally significant (χ² = 3.25; p = 0.07). However, cognitively normal individuals were more likely to carry APOE ɛ2 alleles(s) (χ² = 3.80; p = 0.05) when compared with cognitively impaired participants (Table 2).

Table 2

Characteristics of The University of Manchester Longitudinal Study of Cognition in Normal Healthy Old Age cohort stratified by cognitive status

Characteristic	No cognitive impairment (n = 57)		Cognitive impairment (n = 40)		Total (n = 97)
	n	%	n	%	n	%
Sex
Male	20	35	11	27	31	32
Female	37	65	29	73	66	68
Age at death
Under 90 years	33	58	21	53	54	56
90 years and over	24	42	19	47	43	44
Median age at death (range)	89 (26)		89 (31)		89 (32)
Education (ISCED y equiv)
Mean (±s.d)	15.1 (±3.4)		15.7 (±3.8)		15.4 (±3.6)
Genetics^a
Presence of APOEɛ4 allele(s)	14	25	16	40	30	31
Presence of APOEɛ2 allele(s)	11	19	2	5	13	13

^aGenetic data not available for two cases.

The principal neuropathological diagnosis varied throughout the cohort (Table 3). Those considered pathologically normal for age were more likely to be cognitively intact (χ² = 10.87; p < 0.001) as were those with only limited Aβ/tau pathology (χ² = 5.29; p = 0.02). However, a proportion of cognitively intact individuals exhibited sufficient pathology to meet current neuropathological criteria for AD (8%) or dementia with Lewy bodies (DLB) (3%). The proportion of individuals with incipient AD pathology was similar between the cognitive groups (χ² = 0.21; p = 0.65). Individuals with pathologically confirmed AD (χ² = 12.18; p < 0.001) and DLB/Parkinson’s disease (χ² = 6.91; p = 0.01) were more likely to be cognitively impaired. Conversely, a proportion (10%) of cognitively impaired individuals had only limited Aβ/tau pathology or had pathology considered normal for age. Individuals exhibiting more than one pathology were more likely to be considered cognitively impaired than those with only one principal pathology (χ² = 12.89; p < 0.001) and those considered pathologically normal for age (χ² = 33.23; p < 0.001).

Table 3

Distributions of principal neuropathological diagnosis in The University of Manchester Longitudinal Study of Cognition in Normal Healthy Old Age cohort stratified by cognitive status

Principal neuropathological diagnosis	No cognitive impairment (n = 57)		Cognitive impairment (n = 40)		Total (n = 97)
	n	%	n	%	n	%
Normal for age	21	37	3	8	24	25
Limited Aβ/tau pathology	10	18	1	2	11	11
Incipient AD	9	16	5	13	14	15
AD	4	8	14	36	18	19
DLB/PD	2	3	8	20	10	10
Cerebrovascular disease	2	3	5	13	7	7
CAA	3	5	1	2	4	4
Hemorrhage	2	3	1	2	3	3
CBD	1	2	1	2	2	2
AGD	2	3	0	0	2	2
ARTAG	1	2	0	0	1	1
PART	0	0	1	1	1	1

Four cognitively intact individuals over the age of 90 years at death (mean 93.5±1.7 years) were noted to have very little to no pathological changes, being CERAD score 0, Thal phase 0, Braak stage I or less and low VCING stage, with no TDP-43 or α-synuclein pathology. No APOE ɛ4 alleles were present in this subset of individuals.

Aβ pathology was common across all participants (Table 4). Within the neocortex, cognitively impaired individuals were more severely affected in frontal (χ² = 3.89; p = 0.05), occipital (χ² = 5.38; p = 0.02) and parietal (χ² = 5.30; p = 0.02) regions. However, Aβ pathology was of comparable severity in the temporal region (χ² = 2.47; p = 0.12) when comparing cognitive status groups. In the limbic region, cognitively impaired individuals were more likely to exhibit moderate to severe Aβ pathology in the cingulate (χ² = 4.26; p = 0.04) but not the amygdala (χ² = 2.76; p = 0.10) or hippocampus (χ² = 2.43; p = 0.12). In addition, cognitively impaired individuals showed greater severity of Aβ pathology in the midbrain (χ² = 8.07; p = 0.004) and corpus striatum (χ² = 8.25; p = 0.004).

Table 4

Presence, distribution, and severity of Aβ plaques in selected brain regions from individuals in The University of Manchester Longitudinal Study of Cognition in Normal Healthy Old Age cohort stratified by cognitive status

Brain region	Severity	No cognitive impairment (n = 57)		Cognitive impairment (n = 40)		Total (n = 97)
		n	%	n	%	n	%
Frontal	None to mild	27	47	11	27	38	39
	Moderate to severe	30	53	29	73	59	61
	Missing	0	0	0	0	0	0
Occipital	None to mild	35	61	15	37	50	51
	Moderate to severe	22	39	25	63	47	49
	Missing	0	0	0	0	0	0
Parietal	None to mild	29	51	11	27	40	41
	Moderate to severe	28	49	29	73	57	59
	Missing	0	0	0	0	0	0
Temporal	None to mild	23	40	10	25	33	34
	Moderate to severe	34	60	30	75	64	66
	Missing	0	0	0	0	0	0
Hippocampus	None to mild	47	82	29	72	76	78
	Moderate to severe	8	14	11	28	19	20
	Missing	2	4	0	0	2	2
Amygdala	None to mild	29	51	13	32	42	43
	Moderate to severe	26	46	24	60	50	52
	Missing	2	4	3	8	5	5
Cingulate	None to mild	29	51	11	27	40	41
	Moderate to severe	25	44	24	60	49	51
	Missing	3	5	5	13	8	8
Corpus striatum	None to mild	34	60	12	30	46	47
	Moderate to severe	22	38	27	68	49	51
	Missing	1	1	1	2	2	2
Midbrain	None to mild	52	91	26	65	78	80
	Moderate to severe	5	9	12	30	17	18
	Missing	0	0	2	5	2	2
Medulla	None to mild	52	91	33	82	85	88
	Moderate to severe	0	0	0	0	0	0
	Missing	5	9	7	18	12	12
Cerebellum	None to mild	56	98	37	92	93	96
	Moderate to severe	1	2	2	5	3	3
	Missing	0	0	1	3	1	1

Moderate to severe tau pathology was less common than Aβ pathology but still correlated well with cognitive status (Table 5). Within all regions of the neocortex, individuals considered cognitively impaired were more severely affected by tau pathology: frontal (χ² = 13.28; p < 0.001), occipital (χ² = 11.00; p = 0.001), parietal (χ² = 8.53; p = 0.004), temporal (χ² = 7.86; p = 0.005). The limbic regions showed differences in severity of tau pathology between cognitive groups in the cingulate (χ² = 10.56; p = 0.001) and amygdala (χ² = 5.20; p = 0.02) but not the hippocampus (χ² = 3.06; p = 0.08). Analysis of other brain areas showed cognitively impaired individuals had greater severity of tau pathology in the midbrain (χ² = 6.63; p = 0.01) and corpus striatum (χ² = 7.01; p = 0.01).

Table 5

Presence, distribution and severity of tau pathology in selected brain regions from individuals in The University of Manchester Longitudinal Study of Cognition in Normal Healthy Old Age cohort stratified by cognitive status

Brain region	Severity	No cognitive impairment (n = 57)		Cognitive impairment (n = 40)		Total (n = 97)
		n	%	n	%	n	%
Frontal	None to mild	51	90	23	57	74	76
	Moderate to severe	6	10	17	43	23	24
	Missing	0	0	0	0	0	0
Occipital	None to mild	54	28	95	70	82	85
	Moderate to severe	3	12	5	30	15	15
	Missing	0	0	0	0	0	0
Parietal	None to mild	50	25	88	62	75	77
	Moderate to severe	7	15	12	38	22	23
	Missing	0	0	0	0	0	0
Temporal	None to mild	35	61	13	32	48	49
	Moderate to severe	22	39	27	67	49	51
	Missing	0	0	0	0	0	0
Hippocampus	None to mild	32	56	16	40	48	50
	Moderate to severe	23	40	24	60	47	48
	Missing	2	4	0	0	2	2
Amygdala	None to mild	28	49	10	25	38	39
	Moderate to severe	27	47	27	68	54	56
	Missing	2	4	3	7	5	5
Cingulate	None to mild	48	84	23	58	71	73
	Moderate to severe	5	9	14	35	19	20
	Missing	4	7	3	7	7	7
Corpus striatum	None to mild	54	95	31	78	85	88
	Moderate to severe	2	4	8	20	10	10
	Missing	1	1	1	2	2	8
Midbrain	None to mild	52	91	27	67	79	81
	Moderate to severe	5	9	11	28	16	17
	Missing	0	0	2	5	2	2
Medulla	None to mild	49	86	30	75	79	81
	Moderate to severe	3	5	3	8	6	6
	Missing	5	9	7	17	12	12
Cerebellum	None to mild	57	100	38	96	95	98
	Moderate to severe	0	0	1	2	1	1
	Missing	0	0	1	2	1	1

Only low levels of vascular pathology (as measured by VCING) were found in 74% individuals. Moderate to high levels of vascular pathology were equally likely in both cognitively impaired and cognitively intact individuals (χ² = 1.61; p = 0.20) (Table 6). Cognitively intact individuals exhibiting significant AD pathology were more frequently found to have low levels of vascular pathology than cognitively impaired individuals with significant AD pathology when applying Thal phase (83% versus 60%) or Braak stage (69% versus 64%) criteria. However, neither of these proportional differences reached statistical significance.

Table 6

Presence and severity of other pathologies in individuals from The University of Manchester Longitudinal Study of Cognition in Normal Healthy Old Age cohort stratified by cognitive status

Pathology	Severity/Presence	No cognitive impairment (n = 57)		Cognitive impairment (n = 40)		Total (n = 97)
		n	%	n	%	n	%
VCING	None to Low	45	79	27	67	72	74
	Moderate to High	12	21	13	33	25	26
	Missing	0	0	0	0	0	0
CAA	None to mild	48	84	23	57	71	73
	Moderate to severe	9	16	17	43	26	27
	Missing	0	0	0	0	0	0
TDP-43 in temporal lobe	Absent	53	93	27	67	80	82
	Present	4	7	13	33	17	18
	Missing	0	0	0	0	0	0
α-synuclein in cingulate	Absent	54	95	31	77	85	88
	Present	3	5	9	23	12	12
	Missing	0	0	0	0	0	0
α-synuclein in midbrain	Absent	53	93	31	77	84	87
	Present	4	7	9	23	13	13
	Missing	0	0	0	0	0	0

Other pathologies were less common (Table 6). Moderate to severe CAA was found in 27% of participants and was more likely to occur in cognitively impaired individuals (χ² = 8.55; p = 0.003). Similarly, the presence of α-synuclein pathology in the form of Lewy bodies and Lewy neurites was more common in the cingulate (χ² = 6.44; p = 0.01) and midbrain (χ² = 4.86; p = 0.03) for those considered cognitively impaired. In addition, cognitively impaired individuals were more likely to present with TDP-43 pathology in the temporal lobe than those considered cognitively intact (χ² = 10.56; p = 0.001).

Established protocols for grading AD pathology correlated well with cognitive status (Fig. 1). There were strong positive correlations between CERAD score (r_s = 0.34, p = 0.001), Thal phase (r_s = 0.34, p = 0.001), and Braak stage (r_s = 0.37, p < 0.001) and cognitive impairment. However, there was a degree of overlap between cognitive status and these measures with 26% of cognitively intact participants reaching either CERAD B or C, 11% reaching Thal phase 4 or 5 and 29% reaching Braak stage III–VI.

Fig.1

Distribution of CERAD, Thal and Braak staging in individuals from The University of Manchester Longitudinal Study of Cognition in Normal Healthy Old Age without cognitive impairment (black) and with (white) cognitive impairment.

The presence of APOE ɛ2 allele(s) was more common in cognitively normal participants (χ² = 3.80; p = 0.05). However, there was a comparable distribution of APOE ɛ4 allele(s) between the cognitive groups. Pathologically, there were no differences in the proportions of APOE ɛ2 allele(s) between any of the measures analyzed. Conversely, APOE ɛ4 allele(s) were more commonly found in individuals exhibiting CERAD B– C (χ² = 3.81; p = 0.05), Thal phase 4–5 (χ² = 8.15; p = 0.004), Braak stage III– VI (χ² = 4.65; p = 0.03), and moderate to severe CAA (χ² = 8.22; p = 0.004) (Table 7). When considering age at death, there were no differences in the distribution of APOE ɛ2 allele(s) between the age groups, but APOE ɛ4 allele(s) were more likely to be found in those individuals whose age at death was before they reached 90 years of age (χ² = 5.47; p = 0.02) (Fig. 2). Specifically, the odds of an individual living beyond 90 years of age significantly decreased if they carried one or more APOE ɛ4 allele(s) (OR = 0.30, 95% CI: 0.11–0.80) even when controlling for cognitive status and presence of APOE ɛ2 allele(s).

Fig.2

Distribution of APOE ɛ4 (A) and APOE ɛ2 (B) alleles in individuals from The University of Manchester Longitudinal Study of Cognition in Normal Healthy Old Age stratified by age group. Those without any APOE ɛ4 (A) or APOE ɛ2 (B) alleles are shown in black and those with the relevant allele(s) are shown in white.

Table 7

Presence of APOE ɛ2 and ɛ4 alleles in individuals from The University of Manchester Longitudinal Study of Cognition in Normal Healthy Old Age cohort stratified by pathology type. Two cases were not genotyped for APOE due to lack of fresh frozen tissue

Pathology	Severity/Presence	APOEɛ2 allele(s) absent (n = 82)		APOEɛ2 allele(s) present (n = 13)		APOEɛ4 allele(s) absent (n = 65)		APOEɛ4 allele(s) present (n = 30)
		n	%	n	%	n	%	n	%
CERAD score	0 – A	45	55	10	77	42	65	13	43
	B – C	37	45	3	23	23	35	17	57
	Missing	0	0	0	0	0	0	0	0
Thal phase	0–3	63	77	11	85	56	86	18	60
	4–5	19	23	2	15	9	14	12	40
	Missing	0	0	0	0	0	0	0	0
Braak stage	0– II	48	59	10	77	44	68	14	47
	III– VI	32	39	3	23	19	29	16	53
	Missing	2	2	0	0	2	3	0	0
VCING	None – Low	61	74	10	77	50	77	21	70
	Moderate – High	21	26	3	23	15	23	9	30
	Missing	0	0	0	0	0	0	0	0
CAA	None to mild	60	73	9	69	53	81	16	53
	Moderate to severe	22	27	4	31	12	19	14	47
	Missing	0	0	0	0	0	0	0	0
TDP-43 in temporal lobe	Absent	68	83	12	92	57	88	23	77
	Present	14	17	1	8	8	12	7	23
	Missing	0	0	0	0	0	0	0	0
α-synuclein in cingulate	Absent	70	85	13	100	58	89	25	83
	Present	12	15	0	0	7	11	5	17
	Missing	0	0	0	0	0	0	0	0
α-synuclein in midbrain	Absent	69	84	13	100	57	88	25	83
	Present	13	16	0	0	8	12	5	17
	Missing	0	0	0	0	0	0	0	0

DISCUSSION

The present study reports the neuropathological findings of the first 97 brains collected from The University of Manchester Longitudinal Study of Cognition in Normal Healthy Old Age and relates these findings to cognitive impairment. Although community and population-based studies with brain donation end-points are becoming more common, the present study avoids many of the selection criteria maintained by other studies such as cognitive status, age, gender or ethnicity. With this in mind, there are several limitations to the study. Brain donation was not in the original scope of the longitudinal study and was only introduced in 2004 and consequently a number of potential donors were lost due to early death or withdrawal from the study. Our protocol for assessing cognitive status uses the TICSm test; a telephone administered task for general cognition. Particular limitations of this test are that it consists of a single task rather than battery assessment and that there is a lack of physical clinical assessment. However, TICS is becoming widely used in longitudinal aging research, including the Health and Retirement Survey family of international studies [21] and has been validated against comprehensive cognitive assessment [22, 23]. In addition, the geographical areas covered (Greater Manchester and Newcastle) may not reflect society as a whole and the fact that the cohort was self-selected may indicate that the study sample may not be representative of the general population.

Aβ plaque pathology

In the neocortex and limbic regions, Aβ plaque pathology was common in the cohort as a whole, mirroring findings from previous cohort studies [1 , 25] but contrary to other, smaller, studies [26]. Approximately half of the cohort also had significant Aβ plaque pathology in the corpus striatum, which corroborates findings from other studies; namely that diffuse Aβ plaques can be located in subcortical regions (such as the striatum) of cognitively intact individuals as well as those with AD [27].

Similar to previous work [28], the majority of the cohort had CERAD scores indicative of low risk of AD. This was also the case for Thal staging. However, a proportion of cognitively intact individuals exhibited levels of Aβ plaque pathology which would warrant a diagnosis of possible or probable AD; a common finding in other similar studies [28, 29]. Despite this overlap, Aβ plaque pathology was still related strongly to cognitive function [30 –38] as was CERAD score and Thal phase [29].

Tau pathology

Tau pathology was much less common than Aβ plaque pathology in the cohort as a whole. The majority of brain regions showed only a low burden of tau pathology and moderate to severe tau pathology was only prevalent in the temporal cortex (51% of cases), amygdala (56%) and, to a lesser extent, the hippocampus (48%). The involvement of moderate to severe tau pathology in the limbic region and hippocampus has been described previously [1 , 26] and correlates well with the present study. It has previously been shown that significant tau burden is more likely to be found in cognitive impaired individuals than cognitively normal individuals [39 –43] and the present study shows similar findings.

The majority of individuals in the cohort exhibited Braak stages indicative of low risk of AD. Braak stages of 0– II were considered to be a low risk of AD with Braak II being the most frequently found. A previous population-based study of individuals over the age of 75 showed that Braak stage III is most commonly found [28]. However, the comparative differences between Braak stage II and III are minor. There was a considerable degree of overlap when comparing Braak stage and cognitive status with 29% of cognitively normal individuals reaching Braak stage III– VI; a finding similar to one previously reported [29]. However, it is of note that the majority of cognitively normal individuals in the Braak III– VI group were found to be at Braak III which is not sufficient for a probable AD diagnosis. Despite this overlap, there was a strong relationship between Braak stage and cognitive impairment with those impaired more likely to be at Braak III– VI; a conclusion also found by a number of other studies [29, 44].

VCING

While there are no pathological criteria for vascular dementia that are fully agreed upon by all neuropathologists, the recent study by Skrobot et al. [19] attempted to use consensus guidelines to create an acceptable protocol. Their most successful model incorporated infarct(s) of more than 10 mm, moderate to severe CAA in the occipital lobe and moderate to severe arteriosclerosis to predict whether cognitive impairment was due to vascular problems. Previous studies have shown that the presence of vascular disease may exacerbate the effects of dementia attributed to Aβ plaques and tangles [45]. This finding may help to explain why a proportion of individuals in the present cohort remained cognitively intact despite exhibiting high levels of Aβ plaque and tau pathology, since the majority of these individuals scored ‘low’ on VCING criteria, such that the combined effects of vascular disease, and the amount of Alzheimer or other neurodegenerative pathology present were insufficient to overcome a threshold to cognitive impairment.

CAA pathology

In previous studies, the presence of CAA has been found to be highly prevalent in the elderly [46] and more likely to be found in cognitively impaired individuals, possibly by interacting with other pathologies that may be present [47]. The present study also found CAA to be fairly common with 27% of cases showing moderate to severe CAA in the regions examined. There was a strong relationship between severity of CAA and cognitive status. Similar to previous findings [29], there was a proportion of cognitively normal individuals (16%) who showed moderate to severe CAA. It is possible that these individuals have sufficient compensatory mechanisms and are therefore currently pre-symptomatic.

α-synuclein pathology

The incidence of cortical and subcortical α-synuclein pathology in clinico-pathological or population-based cohorts of aged and very aged individuals has been estimated between 16–25% [25 , 48] although has been described as a rare occurrence in some studies [49]. When examining specific regions, such as the amygdala, the incidence of α-synuclein pathology in individuals over the age of 85 increases to 33% [50]. These findings compare well with the 13% incidence of α-synuclein in the areas examined in the present study. Further examination of α-synuclein in the amygdala of individuals in the present study may a reveal similar incidence of that previously described. Presence of α-synuclein pathology was more likely to occur in cognitively impaired individuals, when compared with cognitively normal individuals. This finding is confirmed by a number of previous studies [48, 51].

TDP-43 pathology

In cohorts of individuals that are very aged, TDP-43 pathology has been found to occur very rarely [25] but in less aged cohorts, TDP-43 has been identified in 27–46% of individuals [52, 53]. TDP-43 pathology was found in the temporal cortex and hippocampus in 18% of cases which reflects the age range of the cohort. Where present, TDP-43 pathology was more likely to occur in individuals exhibiting cognitive impairment [52, 53].

APOE genotype

Those individuals with APOE ɛ2 allele(s) were more likely to be cognitively intact. However, there were no differences in the severity of any of the pathological measures when comparing APOE ɛ2 groups suggesting that carrying APOE ɛ2 allele(s) has no protective effect against amyloid deposition or tau tangle formation. Although this contradicts some previous studies [54, 55] it supports findings from other studies [56]. The relatively small number of APOE ɛ2 carriers in the present study may under-represent this group and mask possible changes in pathology. Therefore, more work and larger sample sizes are needed to clarify these findings.

In agreement with previous studies [57, 58], those individuals carrying APOE ɛ4 allele(s) were more likely to exhibit moderate to severe AD and CAA pathology. Although APOE ɛ4 allele(s) were more likely to occur in those cognitively impaired, this was not significantly so and, statistically, there was a comparable distribution of APOE ɛ4 allele(s) across the cognitive groups. It is possible that a more extensive battery of cognitive tests may have uncovered cognitive impairment in a number of individuals who were considered cognitively normal but carrying APOE ɛ4 allele(s). Another explanation may lie with greater cognitive reserve in those considered cognitively intact and this may explain why they were able to carry APOE ɛ4 allele(s) without succumbing to cognitive impairment.

Participants who were 90 years of age or older at death were much less likely to carry APOE ɛ4 allele(s), probably due to the fact that those who did carry APOE ɛ4 allele(s) succumbed to the increase pathological burden at an earlier age. However, contrary to previous studies [59], those carrying APOE ɛ2 allele(s) did not exhibit increased longevity. Again, the relatively small numbers of APOE ɛ2 carriers in the present study may provide a simple explanation for this result. Alternatively, the discrepancy may lie with the location of the present cohort (Manchester and Newcastle). The main, large studies showing correlations between APOE ɛ2 and longevity were based in Spain, Italy, and Japan where cognitive activity, lifestyle, diet, and environment may have enhanced longevity via gene-environment interactions. Similar interactions have previously been shown to occur with APOE ɛ4 and lifetime cognitive activity [60].

The subset of four individuals that were over 90 years old at death but remained cognitively intact and essentially pathology-free are of particular interest. It has generally been considered that the prevalence of moderate or severe AD-type pathology increases with age in people without dementia, and that the strength of the association between AD-type pathology and dementia is at its weakest in the oldest-old [61], although this has been contested in other studies [25]. Therefore, very aged individuals may show many Aβ neuritic plaques and tau tangles but they do not necessarily show the cognitive impairment expected from such a pathology load. In the present study, we highlight a number of cognitively intact individuals who reached very old age but remained virtually free from neurodegenerative or vascular pathology. Further study of the genetic profile and lifestyle of these individuals may shed light on possible factors that could promote the chances of pathology-free, healthy cognitive aging.

Within the study, there were a number of findings which indicated conflicts between cognitive status in life and pathological lesions found after death. Namely, that there were a number of cognitive intact individuals who exhibited more severe AD-like pathology than would be expected for age. This could be explained in a number of ways. It is possible that these cognitively normal individuals may have had a greater degree of cognitive impairment than first thought which more rigorous cognitive testing may have uncovered. Our methods for concluding cognitive impairment, while robust, are not as extensive as those in the CFAS [62] and Framingham [63] studies where extensive cognitive testing decreased the likelihood of errors in classification. Another possibility may relate to the lack of vascular lesions which are known to exacerbate cognitive impairment in those meeting the pathological criteria for AD [64]. The majority of cognitively intact individuals with significant levels of AD pathology had low levels of vascular pathology, which collectively were insufficient to generate the clinical symptoms of dementia.

In conclusion, the characterization of the clinical and neuropathological findings from the first 97 donated brains of The University of Manchester Longitudinal Study of Cognition in Normal Healthy Old Age correlated well with the established literature and affirmed that the cohort is typical and representative when compared with community– based, population-based and clinico-pathological cohorts. Although, in general, cognitive impairment in life correlated strongly with most pathologies found at postmortem, it was notable that individuals exhibiting more than one pathology were more likely to be cognitively impaired than those with only one principal pathology and those considered pathologically normal for age. Such data suggest that the presence of dementia in very old subjects is represented by the cumulative tissue effects of several ‘lower grade’ pathologies. In themselves, each would be insufficient to bring about significant cognitive impairment, but when present in combination can overcome threshold levels of pathology necessary to bring about clinical change. While the presence of APOE ɛ4 allele was associated with increased severity of all pathologies examined and seemingly reduced the chances of living past 90 years of age, possession of APOE ɛ2 allele was not associated with a reduction in the severity of any of the pathologies measured nor did it increase longevity.

Footnotes

ACKNOWLEDGMENTS

Longitudinal Cognitive studies were funded by Medical Research Council, Economic and Social Research Council, The Wellcome Trust and Unilever PLC. The work of Manchester Brain Bank is supported by Alzheimer’s Research UK and Alzheimer’s Society through the Brains For Dementia Research (BDR) Programme, which also kindly allowed AR to undertake a PhD studentship.

Authors’ disclosures available online ().

References

Neuropathology Group of the Medical Research Council CognitiveFunction and Aging Study (2001) Pathological correlates oflate-onset dementia in a multicentre, community-based population inEngland and Wales. Neuropathology Group of the Medical ResearchCouncil Cognitive Function and Ageing Study (MRC CFAS)., Lancet 357, 169–175.

Matthews

, Brayne

, Lowe

, McKeith

, Wharton

, Ince

(2009) Epidemiological pathology of dementia: Attributable-risks at deathin the Medical Research Council Cognitive Function and Ageing Study. PLoS Med 6, e1000180.

Snowden

, Nun Study (2003) Healthy aging and dementia: Findingsfrom the Nun Study.. Ann Intern Med 139, 450–454.

Bennett

, Schneider

, Arvanitakis

, Wilson

(2012) Overviewand findings from the religious orders study.. Curr AlzheimerRes 9, 628–645.

National Institute on Ageing (2013) Baltimore Longitudinal Study of Ageing. https://www.blsa.nih.gov Accessed January 19 2018.

Cambridge City over 75 Cohort Study (2005) Cambridge City over 75 Cohort Study, http://www.cc75c.group.cam.ac.ukhttp://www.cc75c.group.cam.ac.uk].Accessed January 19, 2018.

Gelber

, Launer

, White

(2012) The Honolulu-Asia Aging Study: Epidemiologic and neuropathologic research on cognitive impairment.. Curr Alzheimer Res 9, 664–672.

Oxford Project to Investigate Memory and Ageing (2016) OPTIMA. http://www.ndcn.ox.ac.uk/research/centre-prevention-stroke-dementia/resources/optima-oxford-project-to-investigate-memory-and-ageing. Accessed January 19, 2018.

University

of Helsinki

(2006) The Vantaa 85+ study. http://www.hi.helsinki.fi/english/research/groups/pathology/vantaa85study.html Accessed January 19 2018.

10.

Kukull

, Higdon

, Bowen

, McCormick

, Teri

, Schellenberg

, van Belle

, Jolley

, Larson

(2002) Dementia and Alzheimerdisease incidence: A prospective cohort study. Arch Neurol 59, 1737–1746.

11.

Rabbitt

PMA

, McInnes

, Diggle

, Holland

, Bent

, Abson

, Pendleton

, Horan

(2004) The University of ManchesterLongitudinal Study of Cognition in Normal Healthy Old Age, 1983 through 2003. Aging Neuropsychol Cogn 11, 245–279.

12.

UNESCO (2006) International Standard Classification of Educationguidelines.http://www.uis.unesco.org/Library/Documents/isced97-en.pdf. Accessed March 21, 2018.

13.

Plassman

, Welsh

, Helms

, Brandt

, Page

, Breitner

(1995) Intelligence and education as predictors of cognitive statein late life: A 50 year follow-up. Neurology 45, 1446–1450.

14.

Evans

, Singleton

, Meltzer

, Stewart

, Prince

(2003) Themental health of older people: Report based on the analysis of theONS survey of psychiatric morbidity among adults in Great Britain Office of National Statistics (IBSN 0116216603).

15.

Obi

, Akiyama

, Kondo

, Shimomura

, Hasegawa

, Iwatsubo

, Mizuno

, Mochizuki

(2008) Relationship of phosphorylated alpha-synuclein and tau accumulation to Abetadeposition in the cerebral cortex of dementia with Lewy bodies.. Exp Neurol 210, 409–420.

16.

Mirra

, Heyman

, McKeel

, Sumi

, Crain

, Brownlee

, Vogel

, Hughes

, van Belle

, Berg

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

17.

Thal

, Rüb

, Orantes

, Braak

(2002) Phases of Abeta-deposition in the human brain and its relevance for thedevelopment of AD. Neurology 58, 1791–1800.

18.

Braak

, Braak

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

19.

Skrobot

, Attems

, Esiri

, Hortobágyi

, Ironside

, Kalaria

, King

, Lammie

, Mann

, Neal

, Ben-Shlomo

, Kehoe

, Love

(2016) Vascular cognitive impairment neuropathologyguidelines (VCING): The contribution of cerebrovascular pathology tocognitive impairment. Brain 139, 2957–2969.

20.

Wenham

, Price

, Blandell

(1991) Apolipoprotein E genotypingby one-stage PCR. Lancet 337, 1158–1159.

21.

Chan

, Gross

, Pezzin

, Brandt

, Kasper

(2015) Harmonizingmeasures of cognitive performance across international surveys ofaging using item response theory. J Aging Health 27, 1392–1414.

22.

Langa

, Larson

, Crimmins

, Faul

, Levine

, Kabeto

, Weir

(2017) A comparison of the prevalence of dementia in theUnited States in 2000 and 2012. JAMA Intern Med 177, 51–58.

23.

Plassman

, Langa

, Fisher

, Heeringa

, Wier

, Ofstedal

, Burke

, Hurd

, Potter

, Rodgers

, Steffend

, Willis

, Wallace

(2007) Prevalence of dementia in the United States: TheAging, Demographics, and Memory Study. Neuroepidemiology 29, 125–132.

24.

Miller

, Hicks

, D’Amato

, Landis

(1984) A descriptivestudy of neuritic plaques and neurofibrillary tangles in an autopsypopulation. Am J Epidemiol 120, 331–341.

25.

Neltner

, Abner

, Jicha

, Schmitt

, Patel

, Poon

, Marla

, Green

, Davey

, Johnson

, Jazwinski

, Kim

, Davis

, Woodard

, Kryscio

, Van Eldik

, Nelson

(2016) Brain pathologies in extreme old age. Neurobiol Aging 37, 1–11.

26.

Troncoso

, Martin

, Dal Fornao

, Kawas

(1996) Neuropathologyin controls and demented subjects from the Baltimore LongitudinalStudy of Ageing. Neurobiol Aging 17, 365–371.

27.

Wolf

, Gearing

, Snowdon

, Mori

, Markesbery

, Mirra

(1999) Progression of regional neuropathology in Alzheimer diseaseand normal elderly: Findings from the Nun study. Alzheimer DisAssoc Disord 13, 226–231.

28.

Xuereb

, Brayne

, Dufouil

, Gertz

, Wischik

, Harrington

, Mukaetova-Ladinska

, McGee

, O’Sullivan

, O’Connor

, Paykel

, Huppert

(2000) Neuropathological findings in the very old.Results from the first 101 brains of a population-based longitudinalstudy of dementing disorders. Ann N Y Acad Sci 903, 490–496.

29.

Sonnen

, Larson

, Crane

, Haneuse

, Li

, Schellenberg

, Craft

, Leverenz

, Montine

(2007) Pathological correlates ofdementia in a longitudinal, population-based sample of aging. Ann Neurol 62, 406–413.

30.

Bancher

, Jellinger

, Lassmann

, Fischer

, Leblhuber

(1996) Correlations between mental state and quantitative neuropathology inthe Longitudinal Study on Dementia. Eur Arch PsychiatryClin Neurosci 246, 137–146 Vienna.

31.

Cummings

, Pike

, Shankle

, Cotman

(1996) Beta-amyloiddeposition and other measures of neuropathology predict cognitivestatus in Alzheimer’s disease. Neurobiol Aging 17, 921–933.

32.

Guillozet

, Weintraub

, Mash

, Mesulam

(2003) Neurofibrillary tangles, amyloid, and memory in aging and mildcognitive impairment. Arch Neurol 60, 729–736.

33.

Koepsell

, Kurland

, Harel

, Johnson

, Zhou

, Kukull

(2008) Education, cognitive function, and severity of neuropathologyin Alzheimer disease. Neurology 70, 1732–1739.

34.

Kraybill

, Larson

, Tsuang

, Teri

, McCormick

, Bowen

, Kukull

, Leverenz

, Cherrier

(2005) Cognitive differences indementia patients with autopsy-verified AD, Lewy body pathology, orboth. Neurology 64, 2069–2073.

35.

Nelson

, Jicha

, Schmitt

, Liu

, Davis

, Mendiondo

, Abner

, Markesbery

(2007) Clinicopathologic correlations in alarge Alzheimer disease center autopsy cohort: Neuritic plaques andneurofibrillary tangles “do count” when staging disease severity. J Neuropathol Exp Neurol 66, 1136–1146.

36.

Nelson

, Abner

, Schmitt

, Kryscio

, Jicha

, Smith

, Davis

, Poduska

, Patel

, Mendiondo

, Markesbery

(2010) Modeling the association between 43 different clinical andpathological variables and the severity of cognitive impairment in alarge autopsy cohort of elderly persons. Brain Pathol 20, 66–79.

37.

Sabbagh

, Cooper

, DeLange

, Stoehr

, Thind

, Lahti

, Reisberg

, Sue

, Vedders

, Fleming

, Beach

(2010) Functional, global and cognitive decline correlates to accumulationof Alzheimer’s pathology in MCI and AD. Curr Alzheimer Res 7, 280–286.

38.

Silver

, Newell

, Brady

, Hedley-White

, Perls

(2002) Distinguishing between neurodegenerative disease and disease-freeaging: Correlating neuropsychological evaluations andneuropathological studies in centenarians. Psychosom Med 64, 493–501.

39.

Bierer

, Hof

, Purohit

, Carlin

, Schmeidler

, Davis

, Perl

(1995) Neocortical neurofibrillary tangles correlate withdementia severity in Alzheimer’s disease. Arch Neurol 52, 81–88.

40.

Crystal

, Dickson

, Sliwinski

, Lipton

, Grober

, Marks-Nelson

, Antis

(1993) Pathological markers associated withnormal aging and dementia in the elderly. Ann Neurol 34, 566–573.

41.

Duyckaerts

, Bennecib

, Grignon

, Uchihara

, He

, Piette

, Hauw

(1997) Modeling the relation between neurofibrillary tanglesand intellectual status. Neurobiol Aging 18, 267–273.

42.

Grober

, Dickson

, Sliwinski

, Buschke

, Katz

, Crystal

, Lipton

(1999) Memory and mental status correlates of modifiedBraak staging. Neurobiol Aging 20, 573–579.

43.

Haroutunian

, Purohit

, Perl

, Marin

, Khan

, Lantz

, Davis

, Mohs

(1999) Neurofibrillary tangles in nondemented elderlysubjects and mild Alzheimer disease. Arch Neurol 56, 713–718.

44.

Riley

, Snowdon

, Markesbery

(2002) Alzheimer’sneurofibrillary pathology and the spectrum of cognitive function:Findings from the Nun Study. Ann Neurol 51, 567–577.

45.

Esiri

, Wilcock

, Morris

(1997) Neuropathological assessmentof the lesions of significance in vascular dementia. J NeurolNeurosurg Psychiatry 63, 749–753.

46.

Tanskanen

, Mäkelä

, Myllykangas

, Notkola

, Polvikoski

, Sulkava

, Kalimo

, Paetau

(2012) Prevalence and severity ofcerebral amyloid angiopathy: A population-based study on veryelderly Finns (Vantaa 85+). Neuropathol Appl Neurobiol 38, 329–336.

47.

Pfeifer

, White

, Ross

, Petrovitch

, Launer

(2002) Cerebral amyloid angiopathy and cognitive function: The HAAS autopsystudy. Neurology 58, 1629–1634.

48.

Mikolaenko

, Pletnikova

, Kawas

, O’Brien

, Resnick

, Crain

, Troncoso

(2005) Alpha-synuclein lesions in normal aging,Parkinson’s disease and Alzheimer’s disease: Evidence from theBaltimore Longitudinal Study of Aging (BLSA). J Neuropathol ExpNeurol 64, 156–162.

49.

Byford

, Brayne

, McKeith

, Chatfield

, Ince

, Matthews

, MRCCFAS Neuropathology group (2009) Lewy bodies and neuronal loss insubcortical areas and disability in non-demented older people: Apopulation based neuropathological cohort study. BMC Geriatr 9, 22.

50.

Raunio

, Myllykangas

, Kero

, Polvikoski

, Paetau

, Oinas

(2017) Amygdala α-synuclein pathology in thepopulation-based Vantaa 85+ study. J Alzheimers Dis 58, 669–674.

51.

Keage

, Ince

, Matthews

, Wharton

, McKeith

, Brayne

, MRC CFAS and CC75C (2012) Impact of less common and “disregarded”neurodegenerative pathologies on dementia burden in apopulation-based cohort. J Alzheimers Dis 28, 485–493.

52.

Wilson

, Yu

, Trojanowski

, Chen

, Boyle

, Bennett

, Schneider

(2013) TDP-43 pathology, cognitive decline, anddementia in old age. Neurol 70, 1418–1424.

53.

Keage

, Hunter

, Matthews

, Ince

, Hodges

, Hokkanen

, Highley

, Dening

, Brayne

(2014) TDP-43 pathology in thepopulation: Prevalence and associations with dementia and age. J Alzheimers Dis 42, 641–650.

54.

Nagy

, Esiri

, Jobst

, Johnson

, Litchfield

, Sim

, Smith

(1995) Influence of the apolipoprotein E genotype on amyloiddeposition and neurofibrillary tangle formation in Alzheimer’sdisease. Neuroscience 69, 757–761.

55.

Morris

, Benjamin

, Leake

, McArthur

, Candy

, Ince

, Torvik

, Bjertness

, Edwardson

(1995) Effect of apolipoprotein E genotype on Alzheimer’s disease neuropathology in a cohort ofelderly Norwegians. Neurosci Lett 201, 45–47.

56.

Berlau

, Corrada

, Head

, Kawas

(2009) epsilon 2 is associated with intact cognition but increased Alzheimerpathology in the oldest old. Neurology 72, 829–834.

57.

Polvikoski

, Sulkava

, Haltia

, Kainulainen

, Vuorio

, Verkkoniemi

, Niinistö

, Halonen

, Kontula

(1995) Apolipoprotein E, dementia, and cortical deposition of beta-amyloidprotein. N Engl J Med 333, 1242–1247.

58.

Nicoll

, Savva

, Stewart

, Matthews

, Brayne

, Ince

, MRCCFAS (2011) Association betweengenotype, neuropathologyand dementia in the older population of England and Wales. Neuropathol Appl Neurobiol 37, 285–294.

59.

Garatachea

, Emanuele

, Calero

, Fuku

, Arai

, Abe

, Murakami

, Miyachi

, Yvert

, Verde

, Zea

, Venturini

, Santiago

, Santos-Lozano

, Rodríguez-Romo

, Ricevuti

, Hirose

, Rábano

, Lucia

(2014) gene and exceptional longevity:Insights from three independent cohorts. Exp Gerontol 53, 16–23.

60.

Wirth

, Villeneuve

, La Joie

, Marks

, Jaqust

(2014) Gene-environment interactions: Lifetime cognitive activity,genotype and β-amyloid burden. J Neurosci 34, 8612–8617.

61.

Savva

, Wharton

, Ince

, Forster

, Matthews

, Brayne C;Medical Research Council Cognitive Function and Ageing Study (2009) Age, neuropathology and dementia. N Eng J Med 360, 2302–2309.

62.

Matthews

, Arthur

, Barnes

, Bond

, Jagger

, Robinson

, Brayne

, Medical Research Council Cognitive Function and Ageing Collaboration (2013) A two-decade comparison of prevalence of dementia in individuals aged 65 yearsand older from three geographical areas of England: Results of the Cognitive Function and Ageing Study I and II. Lancet 382, 1405–1412.

63.

Satizabal

, Beiser

, Chouraki

, Chêne

, Dufouil

, Seshadri

(2016) Incidence of dementia over three decades in theFramingham Heart Study. N Engl J Med 374, 523–532.

64.

Snowden

, Greiner

, Mortimer

, Riley

, Greiner

, Markesbery

(1997) Brain infarction and the clinical expression of Alzheimer’s disease: The Nun Study. JAMA 277, 813–817.