8
Applying the new research diagnostic criteria: MRI findings and neuropsychological correlations of prodromal AD Lorena Rami 1 , Cristina Sole ´-Padulle ´s 1 , Juan Fortea 1 , Beatriz Bosch 1 , Albert Llado ´ 1,2 , Anna Antonell 1 , Jaume Olives 1 , Magda Castellvı´ 1 , David Bartres-Faz 3 , Raquel Sa ´ nchez-Valle 1,2 and Jose Luis Molinuevo 1,2 1 Alzheimer’s Disease and Other Cognitive Disorders Unit, Neurology Service, Hospital Clı ´nic, Barcelona, Spain 2 Institut d’Investigacio ´ Biome `dica August Pi i Sunyer (IDIBAPS) 3 Department de Psiquiatria i Psicobiologia Clinica, Universitat de Barcelona, Barcelona, Spain Correspondence to: J. Luis Molinuevo, E-mail: [email protected] Objective: We describe the neuroimaging characteristics of prodromal AD (PrdAD) patients diagnosed using the new research criteria in a clinical setting. In order to further characterize these patients, we also study the relationship between neuropsychology, CSF biomarkers and magnetic resonance imaging (MRI) findings. Methods/Patients: 76 participants—24 controls (CTR), 20 amnesic patients, and 32 Alzheimer’s disease (AD) patients—were included in the study. PrdAD was defined on the basis of an objective episodic memory deficit and an AD CSF profile. Structural MRI was performed in all participants. Results: After FWE correction, voxel-based morphometry (VBM) analysis of PrdAD patients versus CTR showed significant clusters of decreased gray matter (GM) volume in the left hemisphere regions including the parahippocampal gyrus, uncus, precuneus, and middle frontal gyrus. We did not find differences in brain atrophy between PrdAD and mild AD patients. Some significant associations were found between CSF levels and episodic and semantic fluency tests in the PrdAD group. Correlations in the PrdAD group revealed that patients with higher scores on delayed free recall had significantly greater GM volume in the left superior temporal gyrus (t ¼ 6.64, p < 0.0001). Conclusions: PrdAD patients presented mainly medial temporal GM atrophy, which was related with significant episodic memory impairment. The cognitive deficit observed in PrdAD patients was also associated with CSF biomarker levels. Copyright # 2011 John Wiley & Sons, Ltd. Key words: Alzheimer’s disease; magnetic resonance imaging; cognition; mild cognitive impairment; cerebrospinal fluid; amyloid; neuroimage; neuropsychological tests; diagnosis History: Received 8 October 2010; Accepted 4 January 2011; Published online 7 March 2011 in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/gps.2696 Introduction Future modifying treatments for Alzheimer’s disease (AD) will require earlier diagnosis to optimize their potential benefits. Furthermore, in order to show their efficacy, upcoming trials with disease-modifying drugs will have to target a population with a biological signature of the disease together with the presence of mild clinical symptomatology (Winblad et al., 2006). Recently, Dubois and collaborators presented a consensus set of criteria for AD diagnosis in a research setting. The new AD diagnostic research criteria may substantially aid an etiological diagnosis based on a biomarker profile, and may allow this diagnosis to be made earlier (Dubois et al., 2007). The main changes with respect to the well-established NINDS-ADRDA criteria are that only significant impairment in episodic memory (regardless of the degree of functional loss) and the presence of at least one abnormal biomarker are required for the diagnosis of probable AD. The strength of these criteria is that they allow earlier diagnosis of AD: hence the proposal of the term RESEARCH ARTICLE Copyright # 2011 John Wiley & Sons, Ltd. Int J Geriatr Psychiatry 2012; 27: 127–134.

Applying the new research diagnostic criteria: MRI findings and neuropsychological correlations of prodromal AD

Embed Size (px)

Citation preview

Page 1: Applying the new research diagnostic criteria: MRI findings and neuropsychological correlations of prodromal AD

RESEARCH ARTICLE

Applying the new research diagnostic criteria: MRI findings

and neuropsychological correlations of prodromal AD

Lorena Rami1, Cristina Sole-Padulles1, Juan Fortea1, Beatriz Bosch1, Albert Llado1,2, Anna Antonell1,Jaume Olives1, Magda Castellvı1, David Bartres-Faz3, Raquel Sanchez-Valle1,2 and Jose Luis Molinuevo1,2

1Alzheimer’s Disease and Other Cognitive Disorders Unit, Neurology Service, Hospital Clınic, Barcelona, Spain2Institut d’Investigacio Biomedica August Pi i Sunyer (IDIBAPS)3Department de Psiquiatria i Psicobiologia Clinica, Universitat de Barcelona, Barcelona, SpainCorrespondence to: J. Luis Molinuevo, E-mail: [email protected]

Copyr

Objective: We describe the neuroimaging characteristics of prodromal AD (PrdAD) patients diagnosedusing the new research criteria in a clinical setting. In order to further characterize these patients, we alsostudy the relationship between neuropsychology, CSF biomarkers and magnetic resonance imaging(MRI) findings.

Methods/Patients: 76 participants—24 controls (CTR), 20 amnesic patients, and 32 Alzheimer’s disease(AD) patients—were included in the study. PrdAD was defined on the basis of an objective episodicmemory deficit and an AD CSF profile. Structural MRI was performed in all participants.

Results:After FWE correction, voxel-based morphometry (VBM) analysis of PrdAD patients versus CTRshowed significant clusters of decreased gray matter (GM) volume in the left hemisphere regionsincluding the parahippocampal gyrus, uncus, precuneus, and middle frontal gyrus. We did not finddifferences in brain atrophy between PrdAD and mild AD patients. Some significant associations werefound between CSF levels and episodic and semantic fluency tests in the PrdAD group. Correlations inthe PrdAD group revealed that patients with higher scores on delayed free recall had significantly greaterGM volume in the left superior temporal gyrus (t¼ 6.64, p< 0.0001).

Conclusions: PrdAD patients presented mainly medial temporal GM atrophy, which was related withsignificant episodic memory impairment. The cognitive deficit observed in PrdAD patients was alsoassociated with CSF biomarker levels. Copyright # 2011 John Wiley & Sons, Ltd.

Key words: Alzheimer’s disease; magnetic resonance imaging; cognition; mild cognitive impairment; cerebrospinal fluid;amyloid; neuroimage; neuropsychological tests; diagnosis

History: Received 8 October 2010; Accepted 4 January 2011; Published online 7 March 2011 in Wiley Online Library(wileyonlinelibrary.com).DOI: 10.1002/gps.2696

Introduction

Future modifying treatments for Alzheimer’s disease(AD) will require earlier diagnosis to optimize theirpotential benefits. Furthermore, in order to show theirefficacy, upcoming trials with disease-modifying drugswill have to target a population with a biologicalsignature of the disease together with the presence ofmild clinical symptomatology (Winblad et al., 2006).Recently, Dubois and collaborators presented aconsensus set of criteria for AD diagnosis in a research

ight # 2011 John Wiley & Sons, Ltd.

setting. The new AD diagnostic research criteria maysubstantially aid an etiological diagnosis based on abiomarker profile, and may allow this diagnosis to bemade earlier (Dubois et al., 2007). The main changeswith respect to the well-established NINDS-ADRDAcriteria are that only significant impairment in episodicmemory (regardless of the degree of functional loss)and the presence of at least one abnormal biomarkerare required for the diagnosis of probable AD. Thestrength of these criteria is that they allow earlierdiagnosis of AD: hence the proposal of the term

Int J Geriatr Psychiatry 2012; 27: 127–134.

Page 2: Applying the new research diagnostic criteria: MRI findings and neuropsychological correlations of prodromal AD

128 L. Rami et al.

prodromal, with an evolution from mild cognitivesymptoms into a full-blown dementia. Since thedisease will therefore be diagnosed with minimalsymptomatology, the diagnosis should be supported bythe presence of a positive biomarker profile. Thereforeapplying these criteria will help to diagnose AD in itsprodromal stage. However, some concerns have beenraised about their applicability in a clinical setting.

The main pathological lesions in AD are neurofi-brillary tangles and senile plaques formed by neuronalaccumulations of abnormal tau filaments and extra-cellular deposits of amyloid (AB) fibrils. Cerebrospinalfluid (CSF) studies measuring amyloid 42 (Ab1–42) andtau protein levels in AD patients with confirmedpathology have demonstrated that abnormal levels ofboth biomarkers constitute a specific signature of theunderlying AD-related pathology (Shaw et al., 2009).CSF studies show an acceptable sensitivity andspecificity to help in the diagnosis of AD (Blennowand Hampel, 2003; Sunderland et al., 2003; Andreasenand Blennow, 2005). In a recent study, individualsdiagnosed as having ‘mild cognitive impairment’(MCI) who had ‘pathological’ concentrations of CSFtau or Ab1–42 had a 17.7fold increased risk ofprogressing to diagnosed AD during a 5-year period(Hansson et al., 2006). In addition, some studies havefound that biomarker levels can predict the rate ofcognitive decline across time in individuals with verymild dementia (Snider et al., 2009). Recent studies(Fagan et al., 2006, 2007) combining the use of PETamyloid ligands with the determination of CSF Ab1–42

have accurately detected individuals who haveappreciable deposits of neocortical amyloid. Thepresence of a positive CSF profile has been proposedas a criterion for AD diagnosis, and we believe thatapplying these new AD research criteria will probablyallow clinicians to detect very mild AD with highsensitivity and specificity.

Although there were initial concerns about theapplicability of the new research criteria, Bouwmanet al. (2010) recently demonstrated that these criteriapresented good specificity in the clinical setting andallow diagnosis of the disease in the prodromal stage.This non-demented AD population is diagnosedthrough the presence of an isolated memory impair-ment and a positive CSF profile, and is a potentialtarget for upcoming treatments. It is thereforeimportant to characterize these patients clinicallyand via the use of neuroimaging techniques, and tostudy the relations between their neuropsychologicalprofile, CSF biomarker values, and neuroimagingcharacteristics. The aim of this study is to describe theneuroimaging characteristics of prodromal AD

Copyright # 2011 John Wiley & Sons, Ltd.

patients (PrdAD) who meet the new research criteriain a clinical setting. A secondary objective, in order tofurther characterize these patients, is to study therelationship between neuropsychology, CSF biomarkerlevels, and MRI findings.

Methods

Subjects

Seventy-six participants—24 controls, 20 amnesicpatients (14 of them considered PrdAD), and 32 ADpatients—were consecutively recruited at the AD andother cognitive disorders unit at the Hospital Clinic,Barcelona. All subjects underwent clinical and neu-ropsychological assessment and magnetic resonanceimaging (MRI). Lumbar puncture (LP) was onlyperformed in patients presenting with an isolatedmemory deficit without dementia. The study wasapproved by the local ethics committee and allparticipants gave written informed consent to partici-pate in the study.

Normal controls (CTR). Healthy elders who presentedno cognitive complaints and no evidence of cognitiveimpairment (below 1.5 standard deviation) on any ofthe neuropsychological tests administered.

Amnesic patients criteria. (i) Presence of an objectiveepisodic memory loss defined through the free andcued selective reminding test (FCSRT) (total recallsubtest) (Grober and Buschke, 1987): (ii) clinicaldementia rating scales (CDR) (Morris, 1993) of 0.5with memory box of 0.5 or 1; (iii) essentially preservedactivities of daily living, measured by the functionalactivities questionnaire (FAQ) (Pfeffer et al., 1982). Allthe amnesic patients presented a FAQ score <6; (iv)absence of dementia.

PrdAD criteria. (i) Presence of an objective episodicmemory loss defined using the FCSRT (Grober andBuschke, 1987); (ii) CDR of 0.5 with memory box of0.5 or 1; (iii) essentially preserved activities of dailyliving; (iv) absence of dementia, and (v) AD CSF profile.

CSF analysis. Subjects underwent LP between 9 and12 am and 10 mL of CSF was collected. The sampleswere centrifuged and stored within the first hour afterextraction in polypropylene tubes at �80 8C. Levels ofAb1–42, total tau (t-tau) and phosphorylated tau atthreonine-181 (p-tau) were measured by enzyme-linked immunosorbent assay kits (Innogenetics, Ghent,

Int J Geriatr Psychiatry 2012; 27: 127–134.

Page 3: Applying the new research diagnostic criteria: MRI findings and neuropsychological correlations of prodromal AD

Prodromal ad diagnosis: MRI and neuropsychology 129

Belgium). A positive CSF profile was considered whenAb1–42 <495 pg/mL, and t-tau> 356 or p-tau181 > 54 pg/mL (Schoonenboom et al., 2004, 2005).

AD patients. NINCDS-ADRDA criteria were applied.All patients included were in the mild stages of thedisease (Global Deterioration Scale¼ 4). Two atypicalAD variants (posterior cortical atrophy) with non-significant episodic memory impairment wereexcluded from the study.

MRI acquisition and analysis

A high resolution T1-weighted three-dimensionalmagnetization prepared rapid gradient sequence wasobtained with a 3T SIEMENS MAGNETOM Trio Tim(Siemens, Erlangen, Germany) at the Centre for ImageDiagnosis of the Hospital Clinic (CDIC). Thefollowing parameters were used: 240 sagittal slices;repetition time (TR)¼ 2300 ms; echo time (TE)¼ 2.98 m;slice thickness¼ 1 mm; inversion time (TI)¼ 900; FOV(field of view)¼ 256 mm; matrix size¼ 256� 256; flipangle¼ 98. In order to rule out patients with a highload of ischemic damage, a T2-weighted volume (B0)or a 3D flair image sequence, when available, was usedto quantify the rate of cerebrovascular disease. TheFazekas scale was used by a neuroradiologist blind tothe clinical diagnosis and subjects with scores of 3 wereremoved (Fazekas et al., 1987). For the voxel-basedmorphometry (VBM) analysis, 3D images weremanually reoriented along the anterior and posteriorcommissure line, and a unified segmentation was thencarried out using the software SPM5 (statistical para-metric mapping). After segmentation, normalization, andmodulation, gray matter (GM) tissue segments weresmoothed with an 8-mm-full-width at half-maximumGaussian kernel. Initial results obtained at a significance of0.001 uncorrected for voxel level were saved as a mask anda false-discovery rate (FDR) and family-wise error (FWE)correction was applied using this mask in order to controlfor type I error. For the FWE correction, clusters below20 voxels were not considered. In order to examinepossible associations between memory performanceand GM volume in the PrdAD group, we carried outcorrelations between these variables only for this groupof patients using SPM5.

Neuropsychological and functional assessment

Participants were administered a comprehensive 1-hbattery of tests by a trained neuropsychologist. Thebattery included memory, language, praxis, visual

Copyright # 2011 John Wiley & Sons, Ltd.

perception, and frontal functions assessment. Allneuropsychological scores were adjusted for age andeducational level. Neuropsychological normative dataand neuropsychological battery have been describedelsewhere (Rami et al., 2007).

Statistical analysis

To compare GM volume differences in the clinicalgroups, two-sample t-tests were carried out and a GMmask was applied in order to eliminate any voxels thatdid not belong to the GM tissue.

For statistical analysis of demographical, clinical,and neuropsychological variables and relationshipswith CSF biomarkers we used SPSS for Windowsversion 17 (SPSS Inc.). Pearson’s correlations were alsoused to determine possible associations between CSFbiomarker levels and neuropsychological variables.Correlations were considered significant when p< 0.005in order to correct the multiple comparisons effect.

Results

Demographic, clinical, and CSF characteristics

Demographic and clinical characteristics are summar-ized in Table 1. As AD patients were significantly olderthan the other groups, all results were covariated byage. Fourteen of the twenty amnesic patients from ourunit presented a positive CSF profile and were thereforediagnosed as PrdAD. Only six amnesic patientspresented a negative CSF profile: four had normalAb1–42 concentration, while two had normal t-tau andp-tau. The number of patients in these two groups wasvery low, so they could not be statistically comparedwith the PrdAD group. This meant that statistical MRIanalysis in the amnesic group of patients was appliedonly in the group fulfilling PrdAD criteria.

Neuropsychological features (see Table 1)

PrdAD patients presented significantly lower perform-ances in all episodic and semantic memory measuresthan controls (p< 0.005). Learning list and visualmemory from the CERAD battery and M@T scoreswere significantly lower in the AD than in the PrdADpatients (p< 0.001). Semantic verbal fluency(p<<0.05) and the TMT-A scores (p< 0.005) werealso lower in the AD than in the PrdAD group.

Int J Geriatr Psychiatry 2012; 27: 127–134.

Page 4: Applying the new research diagnostic criteria: MRI findings and neuropsychological correlations of prodromal AD

Table 1 Clinical and neuropsychological differences among three groups (Controls, PrdAD and AD)

CTR (n¼24) PrdAD (n¼14) AD (n¼ 32) F p

Mean SD Mean SD Mean SD

Age (Years) 71.4 6.6 71.6 5.3 75.5 5.5 3.8 0.028Education (Years) 9.0 4.7 8.5 5.0 8.3 2.9 0.1 n.sM@T 45.0 2.9 30.4** 3.1 24.5þþ 7.0 97.5 000MMSE 28.1 1.4 24.8** 1.6 22.5þ 3.3 32.6 000CERAD_L 18.2 3.8 12.6** 3.8 7.8þþ 3.9 47.1 000CERAD_R 5.4 1.4 1.2** 1.1 0.5 0.9 97.0 000CERAD_Recognition 18.8 1.0 16.0** 3.0 14.9 2.4 19.3 000FCSRT- L 25.0 6.3 8.8** 4.5 6.3 4.0 94.6 000FCSRT - TL 40.6 6.4 18.8** 8.0 13.8 8.8 70.0 000FCSRT – R 9.1 2.4 1.7** 1.6 0.7 1.9 105.1 000FCSRT -TR 13.7 2.1 4.3** 2.9 2.8 3.4 76.8 000CERAD_vis_mem 8.5 2.1 3.8** 1.9 1.8þ 2.2 56.8 000BNT 50.3 6.3 46.8 4.9 41.9 8.7 5.9 004Semantic Fluency 17.5 5.0 13.6* 1.6 10.0þ 3.4 24.2 000PDT 14.1 0.8 13.7 0.7 12.0 1.7 19.1 000VOSP-letters 19.6 1.5 19.3 1.4 17.3 3.1 7.5 001Poppelreuter 10.0 0.0 10.0 0 9.6þ 0.6 6.1 004CERAD_praxis 9.8 1.4 10.3 1.0 9.2 1.9 3.1 n.sTMT-A 59.9 21.0 55.5 27.4 110.9þþ 47.3 14.3 000FAS 28.6 10.8 27.2 7.8 21.0 9.6 4.4 016Digit symbol 35.7 20.6 28.5 10.6 19.6 10.9 6.8 002

CTR, controls; PrdAD, prodromal AD patients; AD, Alzheimer’s disease; M@T, memory alteration test; MMSE, mini mental state examination; CERAD,

Neuropsychological Battery of the Consortium to Establish a Registry for Alzheimer’s Disease; FCSRT, free and cued selective reminding test; L, learning; TL,

total learning; R, recall; TR, total recall; vis_mem, visual memory; BNT, Boston naming test; PDT, perceptual digital test; VOSP, visual object and space

perception battery; TMT-A, trail making test A; FAS, phonetic verbal fluency.

All post-hoc contrasts between CTR and AD were significant (p< 0.005) except in VOSP-letters, CERAD-Praxis, Poppelreuter and FAS (p< 0.05).

*Significant post-hoc contrast (p< 0.05) between CTR and PrdAD.

**Significant post-hoc contrast (p< 0.005) between CTR and PrdAD.þSignificant post-hoc contrast (p< 0.05) between PrdAD and AD.þþSignificant post-hoc contrast (p< 0.005) between PrdAD and AD.

130 L. Rami et al.

VBM analysis

The VBM comparison of PrdAD patients versus CTRshowed clusters of decreased GM volume in bilateralmedial temporal lobe regions, and in parietal andfrontal areas (Figure 1(A), Table 2). After FWEcorrection, VBM analysis of PrdAD patients versusCTR showed significant clusters of decreased GMvolume in the left hemisphere regions including theparahippocampal gyrus, uncus, precuneus, andmiddle frontal gyrus. Compared with CTR, ADpatients showed several areas of decreased GM volumein bilateral medial temporal lobe regions and avariety of areas encompassing frontal, temporal, andparietal lobes (Figure 1(B), Table 2). When the FWEcorrection was applied, regions that remained signifi-cant were the bilateral parahippocampal gyri, the rightinferior parietal lobule and the left inferior frontalgyrus. Finally, comparison between AD and PrdADpatients did not show significant differences in GMvolume.

Copyright # 2011 John Wiley & Sons, Ltd.

Association between neuropsychological variablesand CSF levels in PrdAD

Some significant correlations were found between CSFlevels and neuropsychological tests. Lower Ab1–42

levels significantly correlated with poorer semanticverbal fluency (r¼ 0.54; p< 0.005). Higher t-tau levelscorrelated with poorer performance on retention listfrom the CERAD battery (r¼�0.487; p< 0.005) andhigher p-tau correlated with poorer performance onthe three memory tests: M@T (r¼�0.521; p< 001),retention list from the CERAD (r¼�0.527;p< 0.001) andtotal free recall from the FCRST (r¼ 0.518; p< 0.005).

Association between neuropsychological variablesand VBM in PrdAD

Correlations between delayed recall measures of theFCSRT and GM volumes in the PrdAD group revealedthat patients with higher scores on delayed free recall

Int J Geriatr Psychiatry 2012; 27: 127–134.

Page 5: Applying the new research diagnostic criteria: MRI findings and neuropsychological correlations of prodromal AD

Figure 1 Brain regions representing increased GM volume in controls as compared to (A) PrdAD patients and (B) mild AD patients (corrected at FDRp< 0.05).

Prodromal ad diagnosis: MRI and neuropsychology 131

had significantly greater GM volume in the leftsuperior temporal gyrus (t¼ 6.64, p< 0.0001). Like-wise, patients with better performance in the cueddelayed recall score had increased GM volume in theleft supramarginal gyrus (t¼ 5.24, p¼ 0.002), the leftsuperior temporal gyrus (t¼ 4.63, p¼ 0.007) and theanterior cerebellum (t¼ 5.63, p< 0.0001). Theseresults are presented uncorrected for voxel level atp< 0.01. No correlations were seen for semanticfluency scores and brain volume.

Discussion

This study applies specific cut-off memory and CSFvalues in order to diagnose AD, before the appearance

Table 2 Anatomic structures showing group differences in GM volume betwe

Anatomic structure Clu

A. Controls>PrdAD Left uncus extending to theparahippocampal gyrusLeft middle frontal gyrusLeft precuneusRight inferior parietal lobeRight parahippocampal gyrusRight middle frontal gyrusLeft superior frontal gyrusRight supramarginal gyrusLeft precuneusRight middle frontal gyrus

B. Controls>mild AD Right parahippocampal gyrusLeft parahippocampal gyrusLeft inferior frontal gyrus,extending to the superiortemporal gyrusRight inferior parietal lobeextending top the superiortemporal gyrus

Copyright # 2011 John Wiley & Sons, Ltd.

of dementia, with the new AD research criteria in aclinical setting. PrdAD patients fulfilling the newcriteria presented significant GM loss in brain areastraditionally associated with the initial pathologicalchanges of AD. By definition, PrdAD presented anisolated memory problem, and the rest of cognitivefunctions were preserved. Furthermore, an interestingassociation between the different CSF biomarkers andmemory performance of PrdAD subjects emergedfrom the correlation analysis.

During the last decade, several efforts have beenmade to characterize a more homogeneous groupwithin amnesic patients, with a higher AD conversionrate, who could be considered as being in theprodromal stage of the disease. Among the conceptsdeveloped in the attempts to promote earlier AD

en controls, PrdAD and mild AD

ster size,voxel

t-value p-value Mni coordinates(x,y,z)

1132 5.44 <0.0001 (�20,4,�32)

157 5.39 <0.0001 (�36,60,�10)328 4.92 <0.0001 (�34,�78,42)151 5.13 <0.0001 (38,�68,48)729 4.88 <0.0001 (18,�12,�20)101 4.81 <0.0001 (52,22,30)63 4.50 <0.0001 (�20,62,20)53 4.36 0.001 (64,�48,30)60 4.25 <0.0001 (�4,�58,56)55 4.20 0.001 (30,44,40)

785 5.38 <0.0001 (18,�36,4)1378 4.73 <0.0001 (�14,�8,�24)476 4.66 <0.0001 (�50,20,24)

1196 4.42 <0.0001 (64,�38,30)

Int J Geriatr Psychiatry 2012; 27: 127–134.

Page 6: Applying the new research diagnostic criteria: MRI findings and neuropsychological correlations of prodromal AD

132 L. Rami et al.

diagnosis, the term MCI was coined to refer to thesyndromal entity that defines a group of patients withisolated memory problems who do not have dementiabut may eventually develop it (Petersen et al., 1999).Other recent efforts, which culminated in the newresearch criteria, aimed to detect AD in its earlierstages. The objective now is to detect AD whenmanifesting as an isolated memory syndrome whichwill eventually evolve into a full-blown dementiasyndrome. The strength of this approach is that it willdetect disease at a much earlier stage—hence the termPrdAD, before the evolution from mild cognitivedysfunction into full dementia. Since the diagnosis willtherefore be made with minimal symptomatology, itneeds to be supported by the presence of a positivebiomarker profile.

Bouwman et al. (2010) found that the combinationof memory impairment plus a positive CSF profileoffered optimal sensitivity and specificity whenapplying the new research criteria. We thereforedecided to use this combination. In order to customizethe new proposed research criteria, we used specificcut-off episodic memory scores and CSF values, whichcan be applied in any clinical setting. Althoughlongitudinal follow up with pathological confirmationwill be the definite proof that our patients have AD,their (by definition) pathological CSF profile togetherwith the characteristic MRI findings of the PrdADgroup may be considered proof of concept of theunderlying AD pathology. In this regard, the GM lossdetected by MRI in the PrdAD group clearly matchesthe affected areas described in the seminal pathologicalstudies by Braak and Braak (1991), who found thatpathological lesions initially affected the limbic brainregions near the hippocampus, then the neurofibrillarytangle damaged neurons of the temporal lobe, and laterthe neocortex and occipital cortex. So the MRI findingsrecorded in PrdAD patients presented the same patternof atrophy as those in AD patients compared tocontrols, though to a lower extent. These results alsocorroborate those of a previous study in a MCIpopulation with a positive or high risk CSF profilewhich was compared with a MCI population with anegative or low risk CSF profile (Fjell et al., 2010a). Inthat study both the high and the low-risk MCI groupshad significantly greater atrophy rates in target areasthan the control group, although atrophy rates weregreater for the high risk CSF profile MCI group. As inour study, they found that MCI with positive CSFprofile showed atrophy in the hippocampus andmiddle temporal cortex, which, as noted above, aretypically involved in early AD. In another study (Fjellet al., 2010b) of 13 MCI patients with pathological tau

Copyright # 2011 John Wiley & Sons, Ltd.

and 9 with pathological Ab1–42 values these authorsfound significant correlations in these patients betweenTau and Ab1–42 levels and morphometric reductions ofthe medial temporo-parietal circuit.

From a cognitive point of view, by definition PrdADpatients were mildly affected, with all neocorticalfunctions being preserved. Previous studies haveshown that frontal dysfunction is the most frequentadditionally affected domain in some MCI patients andhas been proposed as a cognitive risk factor for ADconversion (Daly et al., 2000; Rami et al., 2009). Bycontrast, in this study PrdAD patients showed anisolated episodic memory problem but preservedlanguage, praxis, perceptual, and frontal functions,which suggests that they were in a very early stage of thedisease. This may be the reason for the significantcorrelation observed between the three biomarkers andthe memory performance, but not with any othercognitive function. t-tau levels were inversely associ-ated with retention list scores and p-tau correlated withthree memory tests: CERAD, FCRST, and M@T, whileAb1–42 presented a positive correlation only withsemantic fluency. It is worth noting that CSFbiomarkers present a particular time sequence inrelation to AD pathophysiology: b-amyloid may startto decrease years before the appearance of any clinicalmanifestation, and the increases in tau and p-tau occurlater in the disease course (Moonis et al., 2005;Ringman et al., 2008; Jack et al., 2010). In theprodromal stage of the disease b-amyloid may still bedecreasing, and is probably close to reaching itsplateau, and tau and p-tau are already increasing,moving further away from their plateau levels. Sincethe correlation analysis was performed only in theprodromal stage of the disease, we may consider thatAb1–42 was reaching its plataeu while p-tau, which ismore specific and was still far from its plateau level,may correlate better with memory performance. Inaddition, the fact that p-tau (which has been associatedwith the tau-related phosphorylation process) is amore specific AD biomarker than total tau, mayexplain why cognition presents more correlations withp-tau than with t-tau.

Interestingly, p-tau was the biomarker with thehighest memory performance association. p-tau levelscorrelated with three verbal episodic memory tests.Previous studies have only found significant corre-lations in the AD continuum between memoryperformance and Ab1–42 (Hildebrandt et al., 2009)or between memory performance and t-tau levels (Linet al., 2009) in AD patients, but not with p-tau levels.

Early diagnosis of AD presents numerous medical,psychological, and social advantages. From the medical

Int J Geriatr Psychiatry 2012; 27: 127–134.

Page 7: Applying the new research diagnostic criteria: MRI findings and neuropsychological correlations of prodromal AD

Key points

� PrdAD patients fulfilling the new criteria pre-sented significant GM loss in brain areastraditionally associated with the initial ADpathological changes.

� Compared with controls (CTR), PrdAD (PrdAD)patients showed significant clusters of decreasedGM volume in the left hemisphere regionsincluding the parahippocampal gyrus, uncus,precuneus, and middle frontal gyrus.

� Correlations in the PrdAD group revealed thatpatients with higher scores on delayed free recallhad significantly greater GM volume in the leftsuperior temporal gyrus.

� The cognitive deficit observed in PrdAD patientswas associated with CSF biomarker levels.

Prodromal ad diagnosis: MRI and neuropsychology 133

point of view, as we mentioned above, it is particularlynecessary since the development of new disease-modifying drugs depends on the performance ofclinical trials at a very early stage prior to dementiaonset. In this regard, the group of PrdAD patientsdescribed in this article may be good candidates fortesting upcoming disease-modifying drugs, which, iftheir efficacy is proven, may bring biological andtherapeutic benefits for the patients. In addition, earlydiagnosis of AD enables the patient to make decisionsabout the disease before the onset of dementia andhelps them to prepare themselves psychologically andsocially. Additional benefits for patients includereducing the uncertainty about what is happening tothem and developing their coping skills to avoid crises.The benefits for the caregivers are a better under-standing of the disease and the extra time gained fororganizing support plans for the future. Furthermore,early diagnosis may result in social healthcare benefits,since the number of people developing dementia willfall when an effective disease-modifying drug is appliedin the prodromal stage of the disease. These advanceswill justify the initial investment required to make ADbiomarker-based diagnosis available in the clinicalscenario worldwide (Iliffe et al., 2003; De Lepeleireet al., 2008).

Although our results highlight certain characteristicsof PrdAD patients by applying the new researchcriteria, the study has some methodological limitationsthat should be resolved in upcoming studies. First, alongitudinal follow-up study is necessary to determinethe clinical and cognitive outcome of PrdAD diagnosedthrough these criteria, and ideally clinicopathologicalstudies will confirm definite AD diagnosis. Second,controls, AD patients and PrdAD groups did not allundergo the same test procedure (CSF studies). In thisstudy LP was only performed in amnesic patients, butnot in controls or AD patients. We acknowledge thatthis introduces a certain heterogeneity in the study;however, as we wanted to define the control groupbased on their cognitive profile and not on their CSFresults, for ethical reasons we decided not to perform aLP in controls. Third, the sample studied represents anamnesic population at an AD referral unit, a factreflected in the low number of amnesic patients with anormal CSF profile. Due to their low number, thisgroup of amnesic patients was described but notincluded in MRI analysis. In this respect, a comparativestudy with a non-AD MCI population is lacking andthe results may not be broadly generalizable. Finally,the results are the consequence of the definition ofPrdAD based on the new research criteria which try tobe as specific as possible. We are aware that a positive

Copyright # 2011 John Wiley & Sons, Ltd.

CSF profile could also be defined on the basis of thepresence of low Ab1–42 levels alone without the needfor a second positive CSF marker. As the results were tobe applied in a clinical setting, we decided to use morespecific criteria to define PrdAD. On the other hand,although the results are limited to a very specific groupof PrdAD, they present correlations between the MRIdescription and cognition in a homogeneous popu-lation.

In summary, this study presents the MRI charac-teristics and neuropsychological correlations ofPrdAD. Their clinical, neuropsychological and imagingfindings suggest that these patients are in theprodromal stage of the disease and are good candidatesfor upcoming trials with disease-modifying drugs.

Conflicts of interest

None declared.

Acknowledgements

This project has been supported by the Ministry ofScience, Spain. Dr Lorena Rami is a recipient of aMiguel Servet grant as a senior investigator from theMinistry of Science, Spain (CP08/00147). The authorswould like to thank Alan Nance and Michael Maudsleyof the Language Advisory Service of the University ofBarcelona for revising the English.

Int J Geriatr Psychiatry 2012; 27: 127–134.

Page 8: Applying the new research diagnostic criteria: MRI findings and neuropsychological correlations of prodromal AD

134 L. Rami et al.

References

Andreasen N, Blennow K. 2005. CSF biomarkers for mild cognitive impairment andearly Alzheimer’s disease. Clin Neurol Neurosurg 107: 165–173.

Blennow K, Hampel H. 2003. CSF markers for incipient Alzheimer’s disease. LancetNeurol 2: 605–613.

Bouwman FH, Verwey NA, Klein M, et al. 2010. New research criteria for the diagnosisof Alzheimer’s disease applied in a memory clinic population. Dement Geriatr CognDisord 30: 1–7.

Braak H, Braak E. 1991. Neuropathological staging of Alzheimer-related changes. ActaNeuropathol 82: 239–259.

Daly E, Zaitchik D, Copeland M, et al. 2000. Predicting conversation to Alzheimerdisease using standardized clinical information. Arch Neurol 57: 675–680.

De Lepeleire J, Wind AW, Iliffe S, et al. Interdem Group. 2008. The primary carediagnosis of dementia in Europe: an analysis using multidisciplinary, multinationalexpert groups. Aging Ment Health 12: 568–576.

Dubois B, Feldman HH, Jacova C, et al. 2007. Research criteria for the diagnosis ofAlzheimer’s disease: revising the NINCDS-ADRDA criteria. Lancet 6: 734–746.

Fagan AM, Mintun MA, Mach RH, et al. 2006. Inverse relation between in vivo amyloidimaging load and cerebrospinal fluid A_42 in humans. Ann Neurol 59: 512–519.

Fagan AM, Roe CM, Xiong C, et al. 2007. Cerebrospinal fluid tau/beta-amyloid(42)ratio as a prediction of cognitive decline in nondemented older adults. Arch Neurol64: 343–349.

Fazekas F, Chawluk JB, Alavi A, Hurtig HI, Zimmerman RA. 1987. MR signalabnormalities at 1.5 T in Alzheimer’s dementia and normal aging. AJR Am JRoentgenol 149: 351–356.

Fjell AM, Amlien IK, Westlye LT, et al. 2010. CSF biomarker pathology correlates with amedial temporo-parietal network affected by very mild to moderate Alzheimer’s diseasebut not a fronto-striatal network affected by healthy aging. Neuroimage 49: 1820–1830.

Fjell AM, Walhovd KB, Fennema-Notestine C, et al. Alzheimer’s Disease Neuroima-ging Initiative. 2010. CSF biomarkers in prediction of cerebral and clinical changein mild cognitive impairment and Alzheimer’s disease. J Neurosci 30: 2088–2101.

Grober E, Buschke H. 1987. Genuine memory deficits in dementia. Dev Neuropsychol3: 13–36.

Hansson O, Zetterberg H, Buchhave P, et al. 2006. Association between CSFbiomarkers and incipient Alzheimer’s disease in patients with mild cognitiveimpairment: a follow-up study. Lancet Neurol 5: 228–234.

Hildebrandt H, Haldenwanger A, Eling P. 2009. False recognition correlates withamyloid-1–42 but not with total tau in cerebrospinal fluid of patients with dementiaand mild cognitive impairment. J Alzheimers Dis 16: 157–165.

Copyright # 2011 John Wiley & Sons, Ltd.

Iliffe S, Manthorpe J, Eden A. 2003. Sooner or later? Issues in the early diagnosis ofdementia in general practice: a qualitative study. Fam Pract 20: 376–381.

Jack CR Jr, Knopman DS, Jagust WJ, et al. 2010. Hypothetical model of dynamicbiomarkers of the Alzheimer’s pathological cascade. Lancet Neurol 9: 119–128.

Lin YT, Cheng JT, Yao YC, et al. 2009. Increased total TAU but not amyloid-beta (42)in cerebrospinal fluid correlates with short-term memory impairment in Alzhei-mer’s disease. J Alzheimers Dis 18: 907–918.

Moonis M, Swearer JM, Dayaw MP, et al. 2005. Familial Alzheimer disease: decreasesin CSF Abeta42 levels precede cognitive decline. Neurology 65: 323–325.

Morris JC. 1993. The clinical dementia rating (CDR): current version and scoringrules. Neurology 43: 2412–2414.

Petersen RC, Smith GE, Waring SC, et al. 1999. Mild cognitive impairment: clinicalcharacterization and outcome. Arch Neurol 56: 303–308.

Pfeffer RI, Kurosaki TT, Harrah CH, Chance JM, Filos S. 1982. Measurement offunctional activities in older adults in the community. J Gerontol 37: 323–329.

Rami L, Gomez-Anson B, Sanchez-Valle R, et al. 2007. Longitudinal study of amnesicpatients at high risk for Alzheimer’s disease: clinical, neuropsychological andmagnetic resonance spectroscopy features. Dement Geriatr Cogn Disord 24: 402–410.

Rami L, Gomez-Anson B, Monte GC, et al. 2009. Voxel based morphometry featuresand follow-up of amnestic patients at high risk for Alzheimer’s disease conversion.Int J Geriatr Psychiatry 24: 875–884.

Ringman JM, Younkin SG, Pratico D, et al. 2008. Biochemical markers in persons withpreclinical familial Alzheimer disease. Neurology 71: 85–92.

Schoonenboom NS, Pijnenburg YA, Mulder C, et al. 2004. Amyloid beta(1–42) andphosphorylated tau in CSF as markers for early-onset Alzheimer disease. Neurology62: 1580–1584.

Schoonenboom SN, Visser PJ, Mulder C, et al. 2005. Biomarker profiles and theirrelation to clinical variables in mild cognitive impairment. Neurocase 11: 8–13.

Shaw LM, Vanderstichele H, Knapik-Czajka M, et al. 2009. Alzheimer’s diseaseneuroimaging initiative. Cerebrospinal fluid biomarker signature in Alzheimer’sdisease neuroimaging initiative subjects. Ann Neurol 65: 403–413.

Snider BJ, Fagan AM, Roe C, et al. 2009. Cerebrospinal fluid biomarkers and rate ofcognitive decline in very mild dementia of the Alzheimer type. Arch Neurol 66: 638–645.

Sunderland T, Linker G, Mirza N, et al. 2003. Decreased amyloid 1–42 and increasedtau levels in cerebrospinal fluid of patients with Alzheimer disease. JAMA 289:2094–2103.

Winblad B, Wimo A, Engedal K, et al. 2006. 3-Year study of donepezil therapy inAlzheimer’s disease: effects of early and continuous therapy. Dement Geriatr CognDisord 21: 353–363.

Int J Geriatr Psychiatry 2012; 27: 127–134.