Brain MRI Biomarkers to Predict Cognitive Decline in Older People with Alzheimer’s Disease

Abstract

Background:

Structural magnetic resonance imaging markers predicting symptomatic progression at the individual level can be highly beneficial for early intervention and treatment planning for Alzheimer’s disease (AD). However, the correlation between baseline MRI findings and AD progression has not been fully established.

Objective:

To explore the correlation between baseline MRI findings and AD progression.

Methods:

Brain volumetric measures were applied to differentiate the patients at risk of fast deterioration in AD. We included 194 AD patients with a 24-month follow-up: 65 slow decliners, 63 normal decliners, and 66 fast decliners categorized by changes in Alzheimer’s Disease Assessment Scale-Cognitive Subscale (ADAS-Cog). ANOVA analyses were used to identify baseline brain atrophy between groups. Logistic regressions were further performed to explore the relative merits of AD resemblance structural atrophy index (AD-RAI) and individual regional volumetric measures in prediction of disease progression.

Results:

Atrophy in the temporal and insular lobes was associated with fast cognitive decline over 24 months. Smaller volumes of temporal and insular lobes in the left but not the right brain were associated with fast cognitive decline. Baseline AD-RAI predicted fast versus slow progression of cognitive decline (odds ratio 3.025 (95% CI: 1.064–8.600), high versus low, AUC 0.771). Moreover, AD-RAI was significantly lower among slow decliners when compared with normal decliners (p = 0.039).

Conclusion:

AD-RAI on MRI showed potential in identifying clinical AD patients at risk of accelerated cognitive decline.

Keywords

Alzheimer’s disease atrophy index automated brain volumetry progression

INTRODUCTION

Worldwide, around 55 million people have dementia, and this number is expected to rise to 78 million in 2030 and 139 million in 2050. Alzheimer’s disease (AD) is the most common form and may contribute to 60–70% of cases. There is currently no treatment available to cure dementia. Anti-dementia medicines and disease-modifying therapies developed to date have limited efficacy. Early diagnosis and detection of fast cognitive decliners would promote early and optimal management to AD. However, promising prognostic factors for AD are still lacking.

Structural magnetic resonance imaging (MRI) has been used to examine longitudinal changes in brain morphology in various neurodegenerative diseases, including AD. The characteristic MRI features of AD include progressive cerebral atrophy in the medial temporal lobe [1], entorhinal cortex, followed by hippocampus, amygdala, parahippocampus, and the structures within the limbic lobe such as the posterior cingulate [2 –4].

Pathologically increased in cerebral atrophy starts early in AD even at the presymptomatic stage. Its correlation with clinical decline has led to atrophy on structural MRI being suggested as a marker of disease progression and a potential outcome measure in trials. The amount, distribution, and rate of cerebral atrophy are closely correlated with cognitive deficits [5, 6]. In the absence of an intervention, cerebral volume loss has clear, direct, and profound negative clinical consequences in AD. To date, the rates of hippocampal and whole brain atrophy on MRI have been the most widely included imaging measures assessing cognitive decline in trials [7]. Other MRI measures including cortical thickness or composites of change are also promising measurements of cognitive decline. MRI measures of atrophy reflect cumulative neuronal damage which in turn is directly responsible for clinical state. When compared with other imaging markers (and other biomarkers), cerebral atrophy has demonstrated to have a stronger correlation with cognitive decline.

AD Resemblance Atrophy Index (AD-RAI) derived from the recognition of atrophy pattern by AI has been shown to have high diagnostic value for AD [8]. But its value in predicting cognitive decline in AD patients has not yet been examined.

We therefore analyzed data of an ongoing cohort study of AD patients. The objectives are to predict value of regional brain volumes and AD-RAI in predicting cognitive decline in older people with AD.

MATERIALS AND METHODS

AD registry (Hong Kong) is an ongoing prospective cohort study to determine the biological and clinical markers for the progression of AD. Consecutive outpatients with clinical diagnosis of mild to moderate AD were recruited from specialist geriatric outpatient clinics of Prince of Wales Hospital (PWH) the teaching hospital of the Chinese University of Hong Kong, from 2016 onwards. All subjects fulfilled the NINCDS-ADRDA criteria for AD. The study was approved by the CUHK-NTEC clinical research ethics committee. Data collected till December 2021 was used in the analysis of this study.

After obtaining informed written consent from the patient and or the family caregiver at the research memory clinic at PWH, demographic and clinical information was ascertained by standardized questionnaires administered by trained research assistants. Neuropsychological tests included Alzheimer’s Disease Assessment Scale-Cognitive Subscale (ADAS-Cog), Montreal Cognitive Assessment (MoCA), and Verbal Fluency (VF). Family caregiver provided information on Chinese version of Disability Assessment of Dementia Scale (CDAD) and Functional Assessment Staging Test (FAST).

3D T1-Weighted MRI was performed at Prince of Wales Hospital using a 3.0 Tesla scanner (Achieva TX; Philips Medical Systems, Best, Netherlands). The spatial resolution is 1.04 mm×1.04 mm×1.2 mm. All the MRI brain images were processed using AccuBrain^® v1.2 (BrainNow Medical Technology Limited), a cloud-based tool of automated brain volumetry that performs brain structure and tissue segmentation and quantification in a fully automatic mode. In this study, we focused on the cognitive-relevant regions which included the brain parenchyma, typical subcortical structures (bilateral hippocampus and amygdala), ventricular regions (ventricular system, lateral ventricle, third ventricle, and forth ventricle) and lobar regions (frontal lobe, occipital lobe, temporal lobe, parietal lobe, cingulate lobe, and insular lobe) for quantification of brain volumetry. In detail, the subcortical regions and ventricle structures were measured with relative volume (% of intracranial volume (ICV)), and the lobe atrophy in a particular lobe was measured as the ratio of the volume of cerebrospinal fluid (CSF) to parenchyma volume in that lobe. We also quantified the total volume of white matter hyperintensities (WMH) for each subject using AccuBrain^® to investigate the influence of vascular factor on the outcomes of the study cohort. The WMH volumes to be compared between groups were also normalized by ICV as volume ratios (% of ICV).

In addition to the brain structural volumetry, a machine-learning derived AD-RAI was also calculated for each individual patient by AccuBrain^® to indicate the comprehensive level of AD-like brain atrophy pattern. A detailed description of AD-RAI can be found in previous publication [8]. The AD-RAI ranges from 0 to 1, where higher value indicates a higher degree of similarity to AD brain structure, and 0.5 was the cut-off for differentiating AD-like brain atrophy vs. normal aging. AD-RAI is computed by AccuBrain^® based on the automatically quantified brain structural volume and lobe atrophy indexes.

All subjects were followed up at month 12 and month 24. At month 12, only MoCA was repeated. At month 24, all neuropsychological tests, FAST and CDAD were repeated.

According to the tertile changes of ADAS-Cog at 24 months follow up, subjects were divided to three groups: 1) slow decliner, 2) normal decliner, and 3) fast decliner. The cutoff values for lower and upper tertiles were 2.33 and 10.00, respectively.

Statistical analyses

Three groups: 1) slow decliner, 2) normal decliner, and 3) fast decliner were compared by using ANOVA or Chi square test. The MRI markers which showed significant associations between the decliner groups, were further assessed by multiple logistic regression comparing fast and slow decliners, adjusting for age, gender, body mass index (BMI), years of education, and score of baseline ADAS-Cog, MoCA, Verbal Fluency, and CDAD. For the most significant MRI markers of accelerated cognitive decline, the area under the ROC curve (AUC) was calculated with the same adjustment in previous regression model. AD-RAI was categorized into 3 groups: low (<0.5), medium (0.5–8.99), and high (>0.9). All analyses were carried out using the Windows-based SPSS Statistical Package (version 24.0; IBM Corp, Armonk, NY, USA), and p-values less than 0.05 would be considered statistically significant. p-values were adjusted with Bonferroni correction for multiple comparisons.

RESULTS

This study was based on 194 AD subjects with valid MRI and ADAS-Cog data at month 24 until December 2021. The dropout rate was 105 out of 333 (31.53%). 34 subjects who joined follow up at month 24 failed to complete the ADAS-Cog, because the subjects were uncooperative or unable to complete the test. The reasons for dropout were death, unwilling to join, poor health status, and loss of contact. Subjects who dropped out were older and shared similar baseline characteristics with the fast decliner in cognitive tests and brain volumes. The characteristics of the subjects categorized by the rate of cognitive decline were shown in Table 1. There were no significant differences in age, gender, education, use of AD drug, living status, FAST score, and chronic disease profiles. Fast decliners had lower cognitive status at baseline.

Table 1

Baseline characteristics of subjects categorized by tertiles of change in ADAS-Cog

	Slow decliner (n = 65)		Normal decliner (n = 63)		Fast decliner (n = 66)
	Mean	SD	Mean	SD	Mean	SD	p
Age	79.83	6.20	78.79	5.76	78.83	6.45	0.555
Years of education	4.97	5.00	4.90	4.45	5.24	4.67	0.912
Body Mass Index	24.51	3.98	23.09	3.86	22.36	2.60	0.003
ADAS-Cog	32.41	9.91	31.41	9.11	35.90	12.21	0.036
Verbal fluency	9.57	3.61	8.92	4.01	7.76	3.65	0.022
MoCA	14.09	5.18	13.76	4.58	11.88	5.89	0.036
CDAD (%)	83.07	14.89	79.09	16.40	75.28	18.29	0.029
	N	%	N	%	N	%	p
Female	49	75.4%	44	69.8%	44	66.7%	0.542
On AD drug	54	83.1%	56	88.8%	52	78.8%	0.465
FAST score							0.563
Early dementia	22	33.8%	17	27.0%	12	18.2%
Mild dementia	24	36.9%	26	41.3%	30	45.5%
Moderate dementia	11	16.9%	14	22.2%	11	16.7%
Moderately severe dementia	6	9.2%	5	7.9%	10	15.1%
Living status							0.287
Alone	8	12.5%	4	6.3%	9	13.8%
With family	55	85.9%	55	87.3%	55	84.6%
Chronic disease
Hypertension	39	60.0%	41	65.1%	34	51.5%	0.285
Heart disease	6	9.2%	11	17.5%	5	7.6%	0.168
Diabetes mellitus	16	24.6%	19	30.2%	17	25.8%	0.757
Hyperlipidemia	27	41.5%	21	33.3%	24	36.4%	0.623
Depression	10	15.4%	2	3.2%	4	6.1%	0.031

Cognitive decline is defined as the change of ADAS-Cog score (Month 24 minus baseline). Slow decliner is defined as the change of ADAS-Cog score <2.33, normal decliner is defined as the change of ADAS-Cog score 2.33-10.00, fast decliner is defined as the change of ADAS-Cog score >10.00. ADAS-Cog, Alzheimer’s Disease Assessment Scale-Cognitive Subscale; MoCA, Montreal Cognitive Assessment; CDAD, Chinese version of Disability Assessment of Dementia Scale; FAST, The Functional Assessment Staging Test.

Fast decliners had worst performance in neuropsychological measures was found among at baseline, 12-month, and 24-month follow (Table 2).

Table 2

Repeated neuropsychological tests in slow, normal, and fast decliners at month 12 and month 24

	Slow decliner		Normal decliner		Fast decliner
	Mean	SD	Mean	SD	Mean	SD	p	p1	p2	p3
Month 12
MoCA	14.00	4.64	12.11	4.82	8.74	5.33	<0.001	0.147	<0.001	0.002
Month 24
ADAS-Cog	31.02	9.95	37.40	9.63	52.29	14.35	<0.001	0.007	<0.001	<0.001
Verbal fluency	9.63	4.20	7.60	3.36	3.98	3.67	<0.001	0.008	<0.001	<0.001
MoCA	13.02	5.19	11.05	4.67	7.58	4.84	<0.001	0.078	<0.001	<0.001
CDAD (%)	71.67	18.82	61.45	22.67	47.07	23.29	<0.001	0.027	<0.001	<0.001

ADAS-Cog, Alzheimer’s Disease Assessment Scale-Cognitive Subscale; MoCA, Montreal Cognitive Assessment; CDAD, Chinese version of Disability Assessment of Dementia Scale; p, p value between three groups; p1, p value between slow and normal decliner; p2, p value between slow and fast decliner; p3, p value between normal and fast decliner, with Bonferroni adjustment for multiple comparisons. Only 56 slow decliners, 52 normal decliners, and 51 decliners completed MoCA at month 12.

Table 3 showed the average regional brain volumes at baseline in the three decliner groups. Only those with significant group differences were shown. No group difference was found in amygdala, ventricular system, caudate, putamen, pons, superior cerebellar peduncle, cerebellum, WMH, and white matter of the whole brain. No group difference was found among amygdala, ventricle, caudate, putamen, hypothalamus, and atrophy in frontal lobe, occipital lobe, and cingulate lobe of both left and right brain.

Table 3

Comparison of baseline brain ratios in slow, normal, and fast decliners

	Slow decliner (n = 65)		Normal decliner (n = 63)		Fast decliner (n = 66)
	Mean	SD	Mean	SD	Mean	SD	p1	p2	p3
Whole brain
AD-RAI	0.557	0.377	0.683	0.354	0.783	0.295	0.117	0.001	0.309
Brain parenchyma	73.764	2.038	73.393	2.424	72.445	2.038	1.000	0.002	0.042
White matter	33.401	1.718	33.398	2.056	32.653	1.526	1.000	0.0498	0.054
Left brain
Temporal lobe atrophy	33.104	6.793	35.106	8.099	38.883	9.807	0.528	<0.001	0.033
Insular lobe atrophy	30.177	11.358	30.568	12.216	35.969	13.224	1.000	0.024	0.039
Temporal lobe	3.328	0.283	3.247	0.253	3.132	0.323	0.348	<0.001	0.074
Parietal lobe	2.931	0.225	2.921	0.228	2.829	0.222	1.000	0.030	0.068
Insular lobe	0.390	0.052	0.387	0.044	0.362	0.053	1.000	0.005	0.015
Right brain
Temporal lobe atrophy	24.689	5.417	25.526	6.978	28.553	7.508	1.000	0.003	0.033
Insular lobe atrophy	17.488	6.261	17.531	7.810	20.629	8.048	1.000	0.049	0.056
Temporal lobe	3.181	0.252	3.079	0.298	3.032	0.345	0.168	0.015	1.000
Insular lobe	0.390	0.047	0.383	0.046	0.369	0.047	1.000	0.029	0.297

All the data was provided by AccuBrain^®. Brain ratio of each region is relative values of its whole brain volume. p1, p value between slow and normal decliner; p2, p value between slow and fast decliner; p3, p value between normal and fast decliner. AD-RAI (AD resemblance atrophy index), with Bonferroni adjustment for multiple comparisons. Only regions with significant group difference (p < 0.05) were shown in this table.

When compared with normal and fast decliners, the slow decliners had significantly lower AD-RAI. When compared with slow decliners, fast decliners had greater AD-RAI (p = 0.001) and less brain parenchyma (p = 0.002) and white matter (p = 0.0498).

In logistic regression for fast decliner using slow decliner as reference, AD-RAI and brain parenchyma remained significant after adjustment for age, gender, BMI, years of education, and score of baseline cognitive tests including ADAS-Cog, MoCA, Verbal Fluency, and CDAD. More atrophy in the temporal and insular lobes in both sides of brain were also found to be related to cognitive deterioration. Volume ratios of temporal and insular lobes in left brain, but not in right brain were associated with fast cognitive decline. The corresponding odds ratio and AUC of ROC curves were also shown in Table 4.

Table 4

Differentiation in slow decliner and fast decliner using AD atrophy index and single regional measures

	B	S.E.	p	unit	OR^*	95% CI for OR		AUC	95% CI for AUC
Whole brain
AD-RAI								0.771	0.690	0.852
<0.5 (low)	–	–	–	–	1	–	–
0.5–0.899 (medium)	0.661	0.563	0.240	vs.<0.5	1.937	0.643	5.841
>0.9 (high)	1.107	0.533	0.038	vs.<0.5	3.025	1.064	8.600
Brain parenchyma	–0.315	0.124	0.011	–2.330	2.083	1.180	3.670	0.780	0.700	0.861
Left brain
Temporal lobe atrophy	0.079	0.030	0.009	8.230	1.916	1.181	3.108	0.790	0.711	0.868
Insular lobe atrophy	0.043	0.020	0.033	12.266	1.695	1.048	2.741	0.770	0.688	0.851
Temporal lobe	–2.141	0.813	0.008	–0.301	1.901	1.179	3.066	0.788	0.707	0.869
Insular lobe	–9.675	4.901	0.048	–0.050	1.621	1.003	2.620	0.762	0.679	0.84
Right brain
Temporal lobe atrophy	0.067	0.034	0.0499	6.634	1.560	1.002	2.427	0.773	0.693	0.853
Insular lobe atrophy	0.073	0.032	0.023	7.374	1.713	1.079	2.720	0.769	0.689	0.849

The AD resemblance atrophy index and the single regional ratio measures were evaluated in differentiating slow decliner and fast decliner using logistic regression. Age, gender, BMI, years of education and score of baseline ADAS-Cog, MoCA, Verbal Fluency, and CDAD were covariates. Only the measures that were significantly different between two target groups were tested (as labeled in Table 2) and only the measures with a p value that achieved p < 0.05 and with Odds Ratio (OR) >1 in logistic regression were shown. ^*OR (95% CI) presented as per standard deviation increase/decrease in the predictor.

In the same logistic regression for normal decliner using slow decliner as reference, AD-RAI was a significant factor after adjustment (p = 0.045). There were no significant MRI features between normal and fast decliner after adjustment, except a small but significant increase in parietal lobe volume in fast decliner (p = 0.025).

DISCUSSION

This study showed that AD-RAI was significantly different between fast and slow cognitive decliner among AD patients. Slower cognitive decliner also had significantly lower AD-RAI than normal decliner. These supported the use of AD-RAI for predicting the risk of accelerated cognitive decline in AD patients.

Most of the subjects were in the early and mild stages of AD. The education level was generally low, which was compatible with the current cohort of older population in Hong Kong. Most of the subjects were taking cholinesterase inhibitors or memantine.

It is interesting to note that demographic, clinical, physical functioning, and neuropsychological measures could predict cognitive progression from mild cognitive impairment to AD but they may not predict the progression from early AD to severe stages [9, 10]. A number of potential cognitive and behavioral predictors of progression rate in AD have been studied. But the prognostic significance of most of them is still controversial [11, 12].

In our study, AD-RAI, the atrophy of temporal lobe and insular lobe showed significant predictive value for accelerated cognitive decline even after adjusted for demographics and neuropsychological measures. This suggested that neurodegeneration from AD was the primary driver of cognitive decline in this group of patients. It is interesting to note that the volume ratios of left temporal and insular lobes were associated with cognitive decline more than those in the right brain. The relative importance of left cerebral atrophy as compared to that in right brain in predicting cognitive decline in AD patients has been reported [13, 14]. It has been shown that cortical atrophy occurred earlier and progressed faster in the left hemisphere than in the right in AD patients, whether this applies to the temporal lobe and insular lobe is not certain. As the Wernicke’s area is located in the left temporal lobe, it is plausible that impairment in language ability may have more negative effects on social interactions which would result in more cognitive decline [15]. The assessment of comprehension should therefore be an important part of cognitive assessment of older people with early AD, and caregiver training in communication may be helpful in slowing cognitive decline. In addition, cognitive stimulating activities in small groups has been shown to improve language ability in older people with early dementia [16].

AD-RAI has been shown to be useful in the diagnosis of AD [8]. This study further affirmed its role in predicting cognitive decline in AD patients. The AUC was similar to that of similar composite scores in predicting progression of mild cognitive impairment to AD (AUC 0.77). The predictive value of AD-RAI was comparable to that of CSF amyloid-β (Aβ)₄₂, and CSF total tau [17].

Lower AD-RAI predicted slower cognitive decline as compared with the norm. There is increasing recognition that some patients with typical clinical features of AD do not have prominent AD pathologies on postmortem [18]. The pathological diagnosis of limbic-predominate age-related TDP-43 encephalopathy (LATE), a newly identified type of non-Alzheimer’s dementia, has recently been established as an alternative cause of AD type dementia. Hippocampal atrophy is greater in cases with LATE than in those with pure AD in MR findings [19]. The volume and shape of the amygdala is also associated with underlying LATE [20]. However, the MRI features of LATE has not been characterized.

This study has several limitations. Firstly, the sample size of the study cohort was relatively small, which makes it difficult to determine and validate the optimal cutoff value of AD-RAI and temporal lobe volume ratios in predicting cognitive decline. Secondly, AD diagnosis was based on clinical features only. Amyloid PET scan or CSF Aβ₄₂ and total tau were not available to confirm the AD diagnosis.

In conclusion, higher AD-RAI on MRI was the strongest predictor of cognitive decline in older people with clinical AD.

Footnotes

ACKNOWLEDGMENTS

This study is supported by grants from Therese Pei Fong Chow Research Centre for Prevention of Dementia, The Chinese University of Hong Kong.

Authors’ disclosures available online ().

References

Scahill

, Schott

, Stevens

, Rossor

, Fox

(2002) Mapping the evolution of regional atrophy in Alzheimer’s disease: Unbiased analysis of fluid-registered serial MRI. Proc Natl Acad Sci U S A 99, 4703–4707.

Matsuda

(2016) MRI morphometry in Alzheimer’s disease. Ageing Res Rev 30, 17–24.

Wolk

, Das

, Mueller

, Weiner

, Yushkevich

, Alzheimer’s Disease Neuroimaging Initiative (2017) Medial temporal lobe subregional morphometry using high resolution MRI in Alzheimer’s disease. Neurobiol Aging 49, 204–213.

Chan

, Fox

, Scahill

, Crum

, Whitwell

, Leschziner

, Rossor

, Stevens

, Cipolotti

, Rossor

(2001) Patterns of temporal lobe atrophy in semantic dementia and Alzheimer’s disease. Ann Neurol 49, 433–442.

Ridha

, Anderson

, Barnes

, Boyes

, Price

, Rossor

, Whitwell

, Jenkins

, Black

, Grundman

, Fox

(2008) Volumetric MRI and cognitive measures in Alzheimer disease: Comparison of markers of progression. J Neurol 255, 567–574.

Cardenas

, Chao

, Studholme

, Yaffe

, Miller

, Madison

, Buckley

, Mungas

, Schuff

, Weiner

(2011) Brain atrophy associated with baseline and longitudinal measures of cognition. Neurobiol Aging 32, 572–580.

Johnson

, Fox

, Sperling

, Klunk

(2012) Brain imaging in Alzheimer disease. Cold Spring Harb Perspect Med 2, a006213.

Zhao

, Luo

, Lew

, Liu

, Au

, Mok

, Shi

, Alzheimer’s Disease Neuroimaging Initiative (2019) Risk estimation before progression to mild cognitive impairment and Alzheimer’s disease: An AD resemblance atrophy index. Aging (Albany NY) 11, 6217–6236.

Belleville

, Fouquet

, Hudon

, Zomahoun

HTV

, Croteau

, Identification

(2017) Neuropsychological measures that predict progression from mild cognitive impairment to Alzheimer’s type dementia in older adults: A systematic review and meta-analysis. Neuropsychol Rev 27, 328–353.

10.

Liu-Seifert

, Siemers

, Sundell

, Mynderse

, Cummings

, Mohs

, Aisen

(2018) Analysis of the relationship of cognitive impairment and functional impairment in mild Alzheimer’s disease in EXPEDITION 3. J Prev Alzheimers Dis 5, 184–187.

11.

Buccione

, Perri

, Carlesimo

, Fadda

, Serra

, Scalmana

, Caltagirone

(2007) Cognitive and behavioural predictors of progression rates in Alzheimer’s disease. Eur J Neurol 14, 440–446.

12.

Thalhauser

, Komarova

(2012) Alzheimer’s disease: Rapid and slow progression. J R Soc Interface 9, 119–126.

13.

Vijayakumar

, Vijayakumar

(2013) Comparison of hippocampal volume in dementia subtypes. ISRN Radiol 2013, 174524.

14.

Thompson

, Hayashi

, de Zubicaray

, Janke

, Rose

, Semple

, Herman

, Hong

, Dittmer

, Doddrell

, Toga

(2003) Dynamics of gray matter loss in Alzheimer’s disease. J Neurosci 23, 994–1005.

15.

Gutierrez-Rexach

, Schatz

(2016) Cognitive impairment and pragmatics. Springerplus 5, 127.

16.

Woods

, Aguirre

, Spector

, Orrell

(2012) Cognitive stimulation to improve cognitive functioning in people with dementia. Cochrane Database Syst Rev, CD005562.

17.

Counts

, Ikonomovic

, Mercado

, Vega

, Mufson

(2017) Biomarkers for the early detection and progression of Alzheimer’s disease. Neurotherapeutics 14, 35–53.

18.

Nelson

, Dickson

, Trojanowski

, Jack

, Boyle

, Arfanakis

, Rademakers

, Alafuzoff

, Attems

, Brayne

, Coyle-Gilchrist

ITS

, Chui

, Fardo

, Flanagan

, Halliday

, Hokkanen

SRK

, Hunter

, Jicha

, Katsumata

, Kawas

, Keene

, Kovacs

, Kukull

, Levey

, Makkinejad

, Montine

, Murayama

, Murray

, Nag

, Rissman

, Seeley

, Sperling

, White Iii

, Yu

, Schneider

(2019) Limbic-predominant age-related TDP-43 encephalopathy (LATE): Consensus working group report. Brain 142, 1503–1527.

19.

Hanko

, Apple

, Alpert

, Warren

, Schneider

, Arfanakis

, Bennett

, Wang

(2019) In vivo hippocampal subfield shape related to TDP-43, amyloid beta, and tau pathologies. Neurobiol Aging 74, 171–181.

20.

Makkinejad

, Schneider

, Yu

, Leurgans

, Kotrotsou

, Evia

, Bennett

, Arfanakis

(2019) Associations of amygdala volume and shape with transactive response DNA-binding protein 43 (TDP-43) pathology in a community cohort of older adults. Neurobiol Aging 77, 104–111.