Visual Rating of Posterior Atrophy as a Marker of Progression to Dementia in Mild Cognitive Impairment Patients

Abstract

Background: Although medial temporal atrophy (MTA) is a useful imaging marker of the progression to dementia in mild cognitive impairment (MCI), substantial numbers of MCI patients without MTA still progress to dementia.

Objective: We investigated whether visual ratings of posterior atrophy (PA) on magnetic resonance imaging show independent predictive value for the progression to dementia in MCI patients.

Methods: This was a retrospective cohort study of elderly patients who visited Seoul National University Bundang Hospital between 2004 and 2012. A total of 148 patients who were initially diagnosed with MCI were followed for up to 3 years (median 22 months) to determine whether they progressed to dementia. We used 4-point and 5-point visual rating scales to assess PA and MTA, respectively. PA and MTA scores were dichotomized into normal (no atrophy) or abnormal (atrophy) in each patient. We performed a Cox regression analysis to examine the hazard ratios (HRs) of PA and MTA for the progression to dementia with adjustment for age, APOE ɛ4 allele status, and baseline Mini-Mental State Examination score.

Results: Among the study population, 47 patients progressed to dementia. Visual assessment of the MRI scans revealed that 67 patients (45.3%) showed PA, whereas 85 patients (57.3%) showed MTA. The HRs with 95% confidence intervals for PA and MTA were 2.516 (1.244–5.091) and 4.238 (1.680–10.687), respectively. The predictive values of visually assessed PA and MTA remained significant, independent of the covariates.

Conclusion: Visual assessment of PA has independent predictive value for progression to dementia in MCI patients.

Keywords

Atrophy biomarker disease progression mild cognitive impairment

INTRODUCTION

Alzheimer’s disease (AD), which is the main cause of dementia in the elderly, is recognized as one of the most burdensome conditions in the elderly [1]. Estimates suggest that it will affect 106 million people worldwide by 2050 [2]. There is a consensus regarding the existence of symptomatic pre-dementia phases in which individuals experience a gradual progressive cognitive decline that results from the accumulation of AD pathologies in the brain [3]. One of these phases is mild cognitive impairment (MCI), which is characterized by the objective evidence of cognitive impairments that are not severe enough to interfere with functional abilities [3].

MCI patients with positive biomarkers for AD are known to have a higher risk of progression to dementia [4] and that approximately 5 to 10% of MCI patients progress to dementia within a year [5]. Many studies have shown that visually rated medial temporal atrophy (MTA) based on magnetic resonance imaging (MRI) [6] is a useful imaging marker for the prediction of the progression from MCI to dementia, and these ratings are widely applied in clinical practice [7]. However, substantial numbers of MCI patients without MTA still progress to dementia [7, 8].

Previous neuroimaging studies have reported that the parietal lobes are involved early in the conversion from MCI to AD [9]. Recently, a visual rating scale for posterior atrophy (PA) has been developed to discriminate AD patients from controls [10]. Because 30% of pathologically confirmed AD patients showed PA in the absence of MTA, adding the PA scale to the MTA scale improved the discrimination of AD patients from healthy controls [11]. With regard to the prediction of conversion to AD in MCI patients, there was a strong association of the MTA scale with conversion to AD, with a borderline significant association with the PA scale [12].

The aim of this study was to evaluate the predictive value of the presence of PA in addition to MTA for the progression to dementia in patients with MCI using Cox regression analysis. We also assessed the discriminability of the prognostic model using PA and MTA.

MATERIALS AND METHODS

Patient data

This was a retrospective cohort study of MCI patients who visited the Clinical Neuroscience Center of Seoul National University Bundang Hospital between October 2004 and October 2012. A consecutive series of patients aged between 60 and 84 years who met the criteria for MCI [13] were identified using our neuropsychological database and by reviewing the electronic medical records. MCI was diagnosed when patients had objective cognitive decline below 1.5 standard deviations of the mean for age- and education-matched peers on at least one neuropsychological test with preserved basic activities of daily living. The neuropsychological battery consisted of five domain-specific tests: attention (digit span forward/backward [14]), executive function (animal naming [15], Controlled Oral Word Association Test [15], Stroop color-naming task [16]), visuospatial ability (Rey Complex Figure Copy [14]), language (Boston Naming Test [17]), and memory (Seoul Verbal Learning test: delayed recall [14]).

Patients who met the following criteria were included: (i) brain MRI performed at the diagnosis of MCI or within 1 year before the diagnosis of MCI, considering that some patients underwent brain MRI for their cognitive complaints before referral to our clinic, (ii) vascular, traumatic, medical, and psychiatric causes ruled out based on historical information and ancillary testing, such as brain MRI, laboratory studies, and neuropsychological tests, and (iii) clinical interviews and neuropsychological tests undertaken between 1 and 3 years after the initial MCI diagnosis to evaluate their progression to dementia. We excluded patients who did not have brain MRI scans in the axial, sagittal, and coronal planes because these three views on the MRI scans were required for visual rating. During the follow-up period, patients were diagnosed with dementia if their cognitive impairment caused difficulties performing activities of daily living and represented a significant decline from the previous level of functioning [18]. The patients who progressed to dementia were designated as having progressive MCI (PMCI), and the patients who did not progress to dementia were designated as having stable MCI (SMCI). We also studied 46 subjects, aged between 60 and 84 years, who demonstrated normal cognition (NC) on neuropsychological testing as a control group [19].

MRI

MTA was assessed using a visual rating scale developed by Scheltens et al. [6], which has been commonly used for clinical and research purposes. It rates atrophy on a 5-point scale from 0 (no atrophy) to 4 (severe atrophy) based on the height of the hippocampal formation and the width of the choroidal fissure and the temporal horn (Supplementary Table 1).

PA was rated according to a recently developed scale [10] that has been shown to have good intra- and inter-rater reliability [11]. This scale rates atrophy on a 4-point scale from 0 (no atrophy) to 3 (severe atrophy) (Fig. 1). In the sagittal orientation, widening of the posterior cingulate and parieto-occipital sulcus and atrophy of the precuneus were evaluated. In the axial orientation, the widening of the posterior cingulate sulcus and sulcal dilatation in the parietal lobes were evaluated. In the coronal orientation, widening of the posterior cingulate sulcus and sulcal dilatation in the parietal lobes were evaluated (Supplementary Table 2). When differences existed between scores for the different planes, the highest score was assigned. Scores for the right and left hemispheres were summed for each patient.

We also measured the atrophy of the frontal lobe as a control region to demonstrate the regional specificity of PA as an imaging marker. Frontal atrophy (FA) was rated using the modified Victoroff’s method, which rates overall frontal lobe atrophy for both hemispheres [19] (Supplementary Figure 1, Supplementary Table 3). All MRI scans were assessed by a single rater (a board-certified neurologist with 7 years of experience in dementia) who was blinded to the clinical information. For MTA and PA, the visual rating scores were dichotomized as normal (no atrophy) and abnormal (atrophy), with a total score of 3 or more considered abnormal [11, 20]. For FA, the visual rating scores were dichotomized with a score of 2 or more considered abnormal (atrophy) [19].

Statistical analyses

The baseline demographics and neuropsychological profiles of patients who were included in and excluded from this study were compared using Student’s t-test or the Mann-Whitney test for continuous variables and Pearson’s chi-squared test for categorical variables. To calculate the power of our sample size, we assumed that the hazard ratio (HR) of the PA for progression to dementia in MCI patients would be 1.52 based on the previous study by Lehmann et al. [12]. Because some of the patients underwent brain MRI before the diagnosis of MCI, we compared the presence of MTA and PA between the patients who underwent brain MRI at and before MCI diagnosis using Pearson’s chi-squared test.

The 25 randomly selected patients were scored twice by the single rater to evaluate intra-rater reliability and were also scored by a different rater to evaluate inter-rater reliability. Reliability was measured by calculating the intraclass correlation coefficient.

We compared the baseline demographics and neuropsychological profiles according to the progression to dementia (SMCI versus PMCI) and according to the presence of MTA and/or PA (no atrophy, MTA only, PA only, MTA and PA) using one-way ANOVA, the Kruskal-Wallis test or the Pearson chi-squared test as appropriate. Furthermore, we compared the presence of MTA, PA, and FA among the NC, SMCI, and PMCI patients using Pearson’s chi-squared test.

To examine the HR of the presence of MTA and PA for the progression to dementia, we performed a Cox regression analysis with follow-up time as a time variable and progression to dementia as a status variable. The proportional assumption was examined by log-log survival plots. Because the patients with brain atrophy may have had more frequent clinic visits, a possibility of detection bias existed. Therefore, we compared the follow-up intervals between patients with and without brain atrophy. We examined the HRs of MTA, PA, baseline demographics, and neuropsychological profiles using univariate Cox regression analysis. Then, we performed bivariate Cox regression analysis with MTA and PA included in the same model and also examined the interaction between MTA and PA.

The multivariate model was performed with adjustment for covariates that were clinically or statistically relevant in the univariate Cox regression analysis (p < 0.2). Because the APOE allele status was unavailable for some patients, we conducted a sensitivity analysis [21]. Multicollinearity between the covariates was tested by calculating the variance inflation factor [22]. As the control region, the HR of the presence of FA for the progression to dementia was examined using univariate Cox regression analysis. To assess the predictive accuracy of MTA and PA, we calculated Harrell’s c-index for the multivariate Cox regression models. Harrell’s c-index is a generalization of the area under the receiver operating characteristic curve for survival data. It ranges from 0.5 (no predictive value) to 1 (perfect concordance between predicted and observed numbers) [23].

We used STATA 13 (STATA Corp., College Station, Texas, USA) to calculate Harrell’s c-index and PASS 11 (NCSS LLC, Kaysville, Utah, USA) to calculate the statistical power. For the remaining analyses, we used SPSS 18 (SPSS Inc., Chicago, Illinois, USA). The study protocol was approved by the Institutional Review Board of Seoul National University Bundang Hospital.

RESULTS

After excluding 42 patients without T1-weighted coronal images, the study population consisted of 148 patients (Table 1). No significant demographic differences were observed between the included and excluded patients (Supplementary Table 4). We calculated a statistical power of 71.17% to detect the predictive value of PA based on a Cox regression model with α= 0.05. Among the 148 patients, 16 patients underwent brain MRI before the diagnosis of MCI; their median time interval between brain MRI and the diagnosis of MCI was 2.5 months with an interquartile range from 1 to 6.2 months. The presence of MTA and PA were not significantly different from those who underwent brain MRI at the diagnosis of MCI (p = 0.332 for MTA and 0.897 for PA).

The intra-rater agreement and 95% confidence intervals (CIs) were 0.979 (0.947–0.992) for the MTA scores and 0.967 (0.915–0.987) for the PA scores. The inter-rater agreement and 95% CIs were 0.894 (0.731–0.958) for the MTA scores and 0.903 (0.750–0.963) for the PA scores.

During the follow-up period (median, 22 months), 47 patients had progressed to dementia, whereas 101 patients had not progressed at their last assessment. Among the PMCI patients, 45 met the criteria for AD [24] and 2 met the criteria for mixed AD with cerebrovascular disease [25]. In the comparison with SMCI, the PMCI patients had a longer follow-up period and lower performance on the memory and executive function domains of neuropsychological testing (Table 2). In the comparison according to the presence of MTA and/or PA, the MCI patients with both MTA and PA had more years of education than those without brain atrophy (p = 0.001). Otherwise, no significant differences were observed in the demographic and neuropsychological profiles among the 4 groups (Supplementary Table 5). The visual rating scores for NC, SMCI, and PMCI are shown in Fig. 2 with significantly different proportions of MTA, PA, and FA demonstrated among the 3 groups (p < 0.001 for MTA, PA, and FA).

With regard to the possibility of detection bias, no significant differences were observed in the follow-up intervals between patients with brain atrophy and those without brain atrophy (Supplementary Table 6). In the univariate Cox regression analysis, the presence of MTA and PA was associated with the hazard of progression to dementia; however, no other variables including FA were statistically significant (Table 1). The proportional assumption was satisfied for both MTA and PA based on log-log survival plots (Supplementary Figure 2). In the bivariate analysis of MTA and PA, the HRs and 95% CIs of MTA and PA were 4.753 (1.97–11.025) and 1.966 (1.044–3.701), respectively. No significant interaction was observed between MTA and PA (p = 0.082), suggesting that the two variables provided independent predictive value.

The multivariate Cox regression analysis using a statistically relevant p < 0.2 (APOE ɛ4 allele, baseline Mini-Mental State Examination (MMSE) score) and clinically relevant variables (age [26]) showed that the adjusted covariates did not alter the significance of the HRs of MTA and PA. The HRs and 95% CIs of MTA, PA, age, baseline MMSE, and APOE ɛ4 were 4.238 (1.680–10.687), 2.516 (1.244–5.091), 0.982 (0.918–1.051), 0.901 (0.802–1.012), and 1.125 (0.598–2.116), respectively (Fig. 3). Variance inflation factors were less than 1.042 for all variables, indicating a low degree of collinearity. Although 12 patients were not included in the multivariate Cox regression analysis because the data for their APOE genotypes were unavailable, the sensitivity analysis showed that the results were consistent with both of the following two extreme scenarios: all the patients with missing data had at least one ɛ4 allele (scenario 1) or they had no ɛ4 allele (scenario 2) (Supplementary Table 7). Harrell’sc-index of MTA in the multivariate model was 0.743 (95% CI, 0.654–0.831). Although adding PA to the model increased the accuracy to 0.756 (95% CI, 0.676–0.835), the difference was not significant (p = 0.27).

DISCUSSION

This study demonstrated that the presence of PA was associated with the progression to dementia in patients with MCI. The predictive value of PA was independent of MTA, as well as age, APOE ɛ4 status, and baseline MMSE.

As in this study, MTA has provided predictive value for the progression to dementia in patients with MCI in other studies [20, 27]. This predictive value may be explained by the accumulation of neurofibrillary tangles (NFTs), which is well correlated with neuronal loss and brain atrophy in AD [28]. NFT depositions develop earlier than the clinical symptoms of AD, with characteristic patterns of progression from the medial temporal lobe to the isocortices [29, 30]. Considering that MCI commonly arises as a result of AD pathology in the elderly population [4], MTA could be a strong predictor of progression to dementia in MCI patients.

PA has also been associated with the progression to dementia in MCI patients based on voxel- and surface-based morphometry. In addition to the medial temporal areas, baseline MRI scans from PMCI patients show more cortical atrophy and decreased cortical thickness in parietal areas, such as the precuneus, posterior cingulate, and retrosplenial cortex, compared with SMCI patients [31 –34]. However, previous quantitative methods have been problematic, not only regarding their use in routine clinical practice due to additional analytical procedures but also regarding the determination of whether brain MRI shows atrophy in each patient. In contrast, the assessment of PA by a standardized visual rating scale is easy to apply in a clinical setting to evaluate whether an individual MCI patient is likely to progress to dementia based on PA [10].

Previously Lehmann et al. investigated the prognostic value of PA for the progression to dementia using a visual rating scale in MCI patients [12]. They reported that PA ratings were marginally significantly associated with the hazard of conversion to dementia, with a 1-point increase associated with a 23% increase in hazard. In this study, we assessed the HR for the presence of PA using a dichotomized atrophy score to determine whether the patient had PA [11]. Furthermore, to assess the regional specificity of PA, we measured FA as a control region, which was not associated with the progression to dementia.

Parietal areas, such as the precuneus, posterior cingulate, angular gyrus, and retrosplenial cortex, as well as the hippocampal formation and parahippocampal cortex, are core components of the posterior medial (PM) network [35]. It has been demonstrated that the PM network is disrupted early in the pathogenesis of AD, with evidence of atrophy in the core components on MRI [36]. As discussed above, MTA could be explained by the accumulation of NFTs, and the tau pathology may be propagated to neurons downstream in the synaptic circuit [37, 38]. PA may result from the transneuronal spread of NFTs from medial temporal to parietal regions within the PMnetwork.

In this study, however, PA predicted the progression of MCI independently of MTA, and a subset of MCI patients showed PA without MTA. In the previous studies, AD patients were classified into distinct subtypes according to their distribution of NFTs [39] as well as their patterns of atrophy on MRI [40]. It has been reported that approximately 20–30% of AD patients, particularly early-onset AD (EOAD) patients, have parietal-dominant patterns of brain atrophy [11, 41]. These results and our findings suggest that neurodegeneration spreading from the medial temporal lobe to the parietal lobe may not be the only pattern of neurodegeneration in AD. In the Cox regression model, adding the PA scale to the MTA scale did not improve the discriminative ability. In the previous study, it was demonstrated that the difference in PA scores was more pronounced between EOAD patients and younger controls than between late-onset AD (LOAD) patients and older controls [11]. The relatively older age of the current study population (mean, 72.29 years) may explain the weak discriminative value of the PA scale when it was added to the Cox model with MTA. Furthermore, because MCI is considered a prodromal state of AD, the difference in PA scores between SMCI and PMCI may be less prominent than that between AD and normal controls.

In this study, although MCI patients with both MTA and PA had more years of education than those without brain atrophy, they showed similar cognitive performance. This finding is in accordance with a previous study in which the pathologic changes of AD were more advanced in patients with higher education; however, their cognitive functions were similar to those of patients with lower education and less pathology [42]. This observation might be explained by the cognitive reserve that compensates for the neuropathological changes and preserves the cognitive functions.

This study has some potential limitations. First, we did not include patients less than 60 years of age. Because PA is suggested to be more of a feature of EOAD than LOAD, it is likely that the predictive value of PA for the progression to dementia would be higher in younger patients with MCI. Therefore, we recommend further study to assess the predictive value of PA for the progression to dementia in younger patients with MCI. A second limitation is that we did not assess biomarkers for amyloid-β(Aβ) or tau proteins. In a previous study, it was demonstrated that MTA was associated with lower levels of cerebrospinal fluid (CSF) Aβ, whereas PA was associated with elevated levels of CSF tau in MCI patients [12]. Considering that CSF examinations cannot demonstrate the spatial distribution of biomarkers in the brain, studies utilizing amyloid and tau positron emission tomography may be needed to assess the regional differences in Aβ, tau, and atrophy and to understand the effects of Aβ and tau on PA. The third limitation is that in the analysis, we used only visual rating scores that were rated by a single rater. However, the visual rating scales of MTA and PA showed good intra- and inter-raterreliabilities.

Parietal areas are involved early as core components of the PM network in the pathogenesis of AD. In conclusion, the present study demonstrates that the visual rating of PA is a useful marker of progression to dementia in MCI patients, and this predictor is independent of MTA.

DISCLOSURE STATEMENT

Authors’ disclosures available online (http://j-alz.com/manuscript-disclosures/16-0339r2).

Footnotes

Appendix

References

Ferri

, Prince

, Brayne

, Brodaty

, Fratiglioni

, Ganguli

, Hall

, Hasegawa

, Hendrie

, Huang

, Jorm

, Mathers

, Menezes

, Rimmer

, Scazufca

, Alzheimer’s

Disease International

(2005) Global prevalence of dementia: A Delphi consensus study. Lancet 366, 2112–2117.

Brookmeyer

, Johnson

, Ziegler-Graham

, Arrighi

(2007) Forecasting the global burden of Alzheimer’s disease. Alzheimers Dement 3, 186–191.

Albert

, DeKosky

, Dickson

, Dubois

, Feldman

, Fox

, Gamst

, Holtzman

, Jagust

, Petersen

, Snyder

, Carrillo

, Thies

, Phelps

(2011) The diagnosis of mild cognitive impairment due to Alzheimer’s disease: Recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimers Dement 7, 270–279.

Dickerson

, Wolk

(2013) Biomarker-based prediction of progression in MCI: Comparison of AD signature and hippocampal volume with spinal fluid amyloid-β and tau. Front Aging Neurosci 5, 55.

Mitchell

, Shiri-Feshki

(2009) Rate of progression of mild cognitive impairment to dementia–meta-analysis of 41 robust inception cohort studies. Acta Psychiatr Scand 119, 252–265.

Scheltens

, Leys

, Barkhof

, Huglo

, Weinstein

, Vermersch

, Kuiper

, Steinling

, Wolters

, Valk

(1992) Atrophy of medial temporal lobes on MRI in “probable” Alzheimer’s disease and normal ageing: Diagnostic value and neuropsychological correlates. J Neurol Neurosurg Psychiatry 55, 967–972.

Frisoni

, Bocchetta

, Chételat

, Rabinovici

, de Leon

, Kaye

, Reiman

, Scheltens

, Barkhof

, Black

, Brooks

, Carrillo

, Fox

, Herholz

, Nordberg

, Jack

, Jagust

, Johnson

, Rowe

, Sperling

(2013) Imaging markers for Alzheimer disease: Which vs how. Neurology 81, 487–500.

DeCarli

, Frisoni

, Clark

, Harvey

, Grundman

, Petersen

, Thal

, Jin

, Jack

, Scheltens

, Alzheimer’s Disease Cooperative Study Group (2007) Qualitative estimates of medial temporal atrophy as a predictor of progression from mild cognitive impairment to dementia. Arch Neurol 64, 108–115.

Jacobs

, Van Boxtel

, Jolles

, Verhey

, Uylings

(2012) Parietal cortex matters in Alzheimer’s disease: An overview of structural, functional and metabolic findings. Neurosci Biobehav Rev 36, 297–309.

10.

Koedam

, Lehmann

, van der Flier

, Scheltens

, Pijnenburg

, Fox

, Barkhof

, Wattjes

(2011) Visual assessment of posterior atrophy development of a MRI rating scale. Euro Radiol 21, 2618–2625.

11.

Lehmann

, Koedam

, Barnes

, Bartlett

, Ryan

, Pijnenburg

, Barkhof

, Wattjes

, Scheltens

, Fox

(2012) Posterior cerebral atrophy in the absence of medial temporal lobe atrophy in pathologically-confirmed Alzheimer’s disease. Neurobiol Aging 33, e621–e627.e12.

12.

Lehmann

, Koedam

, Barnes

, Bartlett

, Barkhof

, Wattjes

, Schott

, Scheltens

, Fox

, Alzheimer’s Disease Neuroimaging Initiative (2013) Visual ratings of atrophy in MCI: Prediction of conversion and relationship with CSF biomarkers. Neurobiol Aging 34, 73–82.

13.

Winblad

, Palmer

, Kivipelto

, Jelic

, Fratiglioni

, Wahlund

, Nordberg

, Bäckman

, Albert

, Almkvist

, Arai

, Basun

, Blennow

, de Leon

, DeCarli

, Erkinjuntti

, Giacobini

, Graff

, Hardy

, Jack

(2004) Mild cognitive impairment–beyond controversies, towards a consensus: Report of the International Working Group on Mild cognitive impairment. J Intern Med 256, 240–246.

14.

Ahn

H-J

, Chin

, Park

, Lee

, Suh

, Seo

, Na

(2010) Seoul neuropsychological screening battery-dementia version (SNSB-D): A useful tool for assessing and monitoring cognitive impairments in dementia patients. J Korean Med Sci 25, 1071–1076.

15.

Kang

, Chin

J-H

, Na

, Lee

, Park

(2000) A normative study of the Korean version of controlled oral Word Association Test (COWAT) in the elderly. Korean J Clin Psychol 19, 385–392.

16.

Kim

, Kim

, Sohn

, Lee

, Yoo

, Lee

, Hong

, Kim

(2004) Development of the Korean Stroop test and study of the validity and the reliability. J Korean Geriatr Soc 8, 233–240.

17.

Kang

, Kim

, Na

(1999) A short form of the Korean-Boston naming test (K-BNT) for using in dementia patients. Korean J Clin Psychol 18, 125–138.

18.

Reisberg

, Ferris

, De Leon

, Crook

(1987) Global deterioration scale (GDS). Psychopharmacol Bull 24, 661–663.

19.

Jang

J-W

, Park

, Baek

, Lim

J-S

, Youn

, Kim

(2015) A comprehensive visual rating scale of brain magnetic resonance imaging: Application in elderly subjects with Alzheimer’s disease, Mild cognitive impairment, and normal cognition. J Alzheimers Dis 44, 1023–1034.

20.

Korf

, Wahlund

L-O

, Visser

, Scheltens

(2004) Medial temporal lobe atrophy on MRI predicts dementia in patients with mild cognitive impairment. Neurology 63, 94–100.

21.

Von Elm

, Altman

, Egger

, Pocock

, Gøtzsche

, Vandenbroucke

, Initiative

(2007) The Strengthening the Reporting of Observational Studies in Epidemiology. (STROBE) statement: Guidelines for reporting observational studies. Prev Med 45, 247–251.

22.

Fox

, Monette

(1992) Generalized collinearity diagnostics. J Am Stat Assoc 87, 178–183.

23.

Harrell

, Lee

, Califf

, Pryor

, Rosati

(1984) Regression modelling strategies for improved prognostic prediction. Stat Med 3, 143–152.

24.

McKhann

, Drachman

, Folstein

, Katzman

, Price

, Stadlan

(1984) Clinical diagnosis of Alzheimer’s disease: Report of the NINCDS-ADRDA Work Group under the auspices of Department of Health and Human Services Task Force on Alzheimer’s Disease. Neurology 34, 939–939.

25.

Dubois

, Feldman

, Jacova

, Cummings

, DeKosky

, Barberger-Gateau

, Delacourte

, Frisoni

, Fox

, Galasko

, Gauthier

, Hampel

, Jicha

, Meguro

, O’Brien

, Pasquier

, Robert

, Rossor

, Salloway

, Sarazin

(2010) Revising the definition of Alzheimer’s disease: A new lexicon. Lancet Neurol 9, 1118–1127.

26.

Tifratene

, Robert

, Metelkina

, Pradier

, Dartigues

(2015) Progression of mild cognitive impairment to dementia due to AD in clinical settings. Neurology 85, 331–338.

27.

Devanand

, Pradhaban

, Liu

, Khandji

, De Santi

, Segal

, Rusinek

, Pelton

, Honig

, Mayeux

, Stern

, Tabert

, de Leon

(2007) Hippocampal and entorhinal atrophy in mild cognitive impairment: Prediction of Alzheimer disease. Neurology 68, 828–836.

28.

Whitwell

, Josephs

, Murray

, Kantarci

, Przybelski

, Weigand

, Vemuri

, Senjem

, Parisi

, Knopman

, Boeve

, Petersen

, Dickson

, Jack

(2008) MRI correlates of neurofibrillary tangle pathology at autopsy: A voxel-based morphometry study. Neurology 71, 743–749.

29.

Jack

, Knopman

, Jagust

, Petersen

, Weiner

, Aisen

, Shaw

, Vemuri

, Wiste

, Weigand

, Lesnick

, Pankratz

, Donohue

, Trojanowski

(2013) Tracking pathophysiological processes in Alzheimer’s disease: An updated hypothetical model of dynamic biomarkers. Lancet Neurol 12, 207–216.

30.

Braak

, Braak

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

31.

Whitwell

, Shiung

, Przybelski

, Weigand

, Knopman

, Boeve

, Petersen

, Jack

(2008) MRI patterns of atrophy associated with progression to AD in amnestic mild cognitive impairment. Neurology 70, 512–520.

32.

Karas

, Sluimer

, Goekoop

, Van Der Flier

, Rombouts

, Vrenken

, Scheltens

, Fox

, Barkhof

(2008) Amnestic mild cognitive impairment: Structural MR imaging findings predictive of conversion to Alzheimer disease. Am J Neuroradiol 29, 944–949.

33.

Hämäläinen

, Tervo

, Grau-Olivares

, Niskanen

, Pennanen

, Huuskonen

, Kivipelto

, Hänninen

, Tapiola

, Vanhanen

, Hallikainen

, Helkala

, Nissinen

, Vanninen

, Soininen

(2007) Voxel-based morphometry to detect brain atrophy in progressive mild cognitive impairment. Neuroimage 37, 1122–1131.

34.

Julkunen

, Niskanen

, Muehlboeck

, Pihlajamäki

, Könönen

, Hallikainen

, Kivipelto

, Tervo

, Vanninen

, Evans

, Soininen

(2009) Cortical thickness analysis to detect progressive mild cognitive impairment: A reference to Alzheimer’s disease. Dement Geriatr Cogn Disord 28, 404–412.

35.

Ranganath

, Ritchey

(2012) Two cortical systems for memory-guided behaviour. Nat Rev Neurosci 13, 713–726.

36.

La Joie

, Landeau

, Perrotin

, Bejanin

, Egret

, Pélerin

, Mézenge

, Belliard

, de La Sayette

, Eustache

, Desgranges

, Chételat

(2014) Intrinsic connectivity identifies the hippocampus as a main crossroad between Alzheimer’s and semantic dementia-targeted networks. Neuron 81, 1417–1428.

37.

de Calignon

, Polydoro

, Suérez-Calvet

, William

, Adamowicz

, Kopeikina

, Pitstick

, Sahara

, Ashe

, Carlson

, Spires-Jones

, Hyman

(2012) Propagation of tau pathology in a model of early Alzheimer’s disease. Neuron 73, 685–697.

38.

Zhou

, Gennatas

, Kramer

, Miller

, Seeley

(2012) Predicting regional neurodegeneration from the healthy brain functional connectome. Neuron 73, 1216–1227.

39.

Murray

, Graff-Radford

, Ross

, Petersen

, Duara

, Dickson

(2011) Neuropathologically defined subtypes of Alzheimer’s disease with distinct clinical characteristics: A retrospective study. Lancet Neurol 10, 785–796.

40.

Whitwell

, Dickson

, Murray

, Weigand

, Tosakulwong

, Senjem

, Knopman

, Boeve

, Parisi

, Petersen

, Jack

, Josephs

(2012) Neuroimaging correlates of pathologically defined subtypes of Alzheimer’s disease: A case-control study. Lancet Neurol 11, 868–877.

41.

Noh

, Jeon

, Lee

, Seo

, Kim

, Cho

, Ye

, Yoon

, Kim

, Chin

, Park

, Heilman

, Na

(2014) Anatomical heterogeneity of Alzheimer disease: Based on cortical thickness on MRIs. Neurology 83, 1936–1944.

42.

Stern

, Alexander

, Prohovnik

, Mayeux

(1992) Inverse relationship between education and parietotemporal perfusion deficit in Alzheimer’s disease. Ann Neurol 32, 371–375.