Journal of Alzheimers Disease 26 (2011) 337348 DOI 10.

3233/JAD-2011-0025 IOS Press

337

Effects of a 6-Month Cognitive Intervention on Brain Metabolism in Patients with Amnestic MCI and Mild Alzheimers Disease
Stefan F rstera,i, , Verena C. Buschertb , Stefan J. Teipelc,d , Uwe Friesec,g , Hans-Georg Buchholzf , o Alexander Drzezgai , Harald Hampele , Peter Bartensteina and Katharina Buergerb,h
a Department b Dementia

of Nuclear Medicine, Ludwig-Maximilian University, Munich, Germany Research Section and Memory Clinic, Alzheimer Memorial Center and Geriatric Psychiatry Branch, Department of Psychiatry Ludwig-Maximilian University, Munich, Germany c Department of Psychiatry, University Rostock, Rostock, Germany d DZNE, German Center for Neurodegenerative Disorders, Rostock, Germany e Discipline of Psychiatry, School of Medicine & Trinity College Institute of Neuroscience (TCIN), Laboratory of Neuroimaging and Biomarker Research, Trinity College, University of Dublin, Dublin; The Adelaide and Meath Hospital Incorporating the National Childrens Hospital (AMiNCH), Ireland f Department of Nuclear Medicine, University Mainz, Mainz, Germany g Institute of Psychology, University of Osnabrueck, Osnabrueck, Germany h Institute for Stroke and Dementia Research, Klinikum Grohadern, Ludwig-Maximilian University, Munich, Germany i Department of Nuclear Medicine, Technische Universit t M nchen, Munich, Germany a u

Abstract. The effect of cognitive intervention on brain metabolism in AD is largely unexplored. Therefore, we aimed to investigate cognitive parameters and 18 FDG PET to test for effects of a cognitive intervention in patients with aMCI or mild AD. Patients with aMCI (N = 24) or mild AD (N = 15) were randomly assigned either to cognitive intervention groups (IGs), receiving weekly sessions of group-based multicomponent cognitive intervention, or active control groups (CGs), receiving pencil-paper exercises for self-study. We obtained resting-state FDG-PET scans and neuropsychological testing at baseline and after six-months. Normalized FDG-PET images were analyzed using voxel-based SPM5 approaches to determine longitudinal changes, group-by-time interactions and correlations with neuropsychological outcome parameters. Primary global cognitive outcome was determined by analyses of covariance with MMSE and ADAS-cog scores as dependent measures. Both, aMCI and AD subgroups of CGs showed widespread bilateral cortical declines in FDG uptake, while the AD subgroup of IGs showed discrete decline or rather no decline in case of the aMCI subgroup. Group by time analyses revealed strongest attenuation of metabolic decline in the aMCI subgroup of the IGs, involving left anterior temporal pole and anterior cingulate gyrus. However, correlation analyses revealed only weak non-signicant associations between increased FDG uptake and improvement in primary or secondary outcome parameters. Concurrently, there was signicant improvement in global cognitive status in the aMCI subgroup of the IGs.

Correspondence to: Stefan F rster, MD; Klinik und Poliko linik f r Nuklearmedizin, TU M nchen; Ismaninger Str. 22, u u 81675 M nchen, Germany. Tel.: +49 89 4140 2965; Fax: +49 89 u 4140 4950; E-mail: stefan-foerster@gmx.de.

ISSN 1387-2877/11/$27.50 2011 IOS Press and the authors. All rights reserved

338

S. F rster et al. / Effects of a 6-Month Cognitive Intervention on Brain Metabolism o

A six-month cognitive intervention imparted cognitive benets in patients with aMCI, which were concurrent with an attenuated decline of glucose metabolism in cortical regions affected by neurodegenerative AD. Keywords: FDG PET, cognitive intervention, cognitive training, cognitive stimulation, Alzheimers disease, mild cognitive impairment

INTRODUCTION Cognitive intervention for patients with Alzheimers Disease (AD) is considered an important contribution to the treatment of AD [1]. Specically designed cognitive interventions can improve memory performance and can even attenuate the risk of future cognitive decline in non-demented elderly subjects [2, 3]. Benecial effects of interventions on cognitive decline are reported in subjects with preclinical cognitive impairment (see review [4]) or mild-to-moderate stages of dementia (see meta-analyses [5, 6]). Effective cognitive interventions might conceivably impart their effects through altered cerebral metabolism. The rate of cerebral glucose consumption can be assessed in positron emission tomography (PET) studies recording the uptake of the stable glucose analogue [18 F]uorodeoxyglucose (FDG); the PET-FDG technique has been extensively used for studying the pathophysiology of neuropsychiatric and neurodegenerative disorders, including AD (see review [7]). FDG-PET recordings obtained at rest are a sensitive indicator of perturbed brain metabolism, and may reveal very early cerebrometabolic changes preceding the onset of clinical AD symptoms [8]. Voxel-wise mapping of FDG uptake is amenable for the longitudinal assessment of brain functional changes in dementia, and may more sensitively detect treatmentresponse than do commonly-used instruments for cognitive testing, such as the Mini Mental State Examination (MMSE) and the Alzheimers Disease Assessment Scale - cognitive subscale, total score (ADAS-cog) [9]. However, the effect of cognitive intervention on cerebral metabolism in AD is largely unexplored. There are only two published studies investigating the effects of cognitive intervention on brain metabolism. In an early study, FDG-PET was obtained in a visual stimulus activation paradigm with AD patients, and the effects of an unspecied cognitive training of six months duration and a pharmacological intervention were tested [10]; the combined treatment proved to be superior to cognitive training alone. More recently, participation in 14-day long healthy lifestyle program consisting of a combination of mental and

physical exercise, stress reduction, and healthy diet was associated with signicant short-term benets in cognitive function and cerebral FDG uptake in elderly non-demented subjects [11]. In the present study, we tested the hypothesis that a newly-developed multicomponent cognitive intervention [12] applied in a randomized controlled trial would lead to cognitive and non-cognitive benets and mitigate against declining brain metabolism in patients with amnestic MCI or mild AD. To test this hypothesis, we examined global cognitive (MMSE- and ADAS-cog. scores) and additionally non-cognitive parameters, such as mood (Montgomery Asberg Depression Rating Scale, MADRS), which are deemed to be important for cognitive performance [13]. We used FDG-PET to map the pattern of declining cerebral glucose metabolism during six months (between baseline and follow-up scans) in a wellcharacterized sample of aMCI and mild AD patients, participating in a parallel group randomized controlled trial with two treatment arms, (i) specic cognitive intervention and (ii) an active control condition. Prior to the start of the study, a multicomponent cognitive intervention for patients with aMCI and patients with mild AD was conceptualized aiming at capabilities and needs of patients in different stages of AD [12]. In order to gain more direct insights into the functional-anatomical substrate of neuronal function along with cognitive improvement, voxel-based bidirectional SPM patient-group-by-time interaction analyses (later referred to as difference of differences analyses), measuring baseline to follow-up PET metabolic differences between the patient groups and treatment arms, were applied. MATERIALS AND METHODS Subjects We screened 43 patients at the Dementia Research Section and Memory Clinic of the Alzheimer Memorial Center and Geriatric Psychiatry Branch, Department of Psychiatry, Ludwig-Maximilian University, Munich, from March to August 2007. In the

S. F rster et al. / Effects of a 6-Month Cognitive Intervention on Brain Metabolism o

339

screening, we conducted a comprehensive clinical and neuropsychological assessment, in order to support the research diagnosis of either aMCI or mild AD. Participants with aMCI had a memory complaint and performed at least 1.5 SD below the average level of persons of a similar age and education on at least one of three memory tests of the Consortium to Establish a Registry for Alzheimers Disease (CERAD) neuropsychological test-battery (Morris, et al., 1989) (immediate and/or delayed recall and/or recognition). Their cognitive difculties had no signicant repercussions on their functional independence, as assessed through clinical interviews with the patients and caregivers. Hence, they did not full criteria for mild stages of clinically probable AD [14]. Neither of the patient (groups) had major physical illness, other mental disorder (i.e. major depression) or disability which could have affected participation. Thirty-nine patients, meeting stringent inclusion criteria, were included. For details please see [12]. Routine laboratory testing of all included patients consisted of full blood cell count, blood glucose, thyroid function tests, serum Vitamin B12 and folic acid levels as well as cerebrospinal uid protein and Apo-E genotype testing; there were no major abnormalities

at baseline. Each patient underwent structural magnetic resonance imaging (MRI) for detection of brain anatomical abnormalities; according to the Scheltens scale [15] patients with structural lesions visible in the T2-weighted MR scans exceeding 10 mm in diameter, including white matter hyperintensities, were excluded from the study. According to the declaration of Helsinki approval of the local ethics commission (of the Ludwig-Maximilian University, Munich), and the German Radiation Safety Committee (BfS) was obtained prior to starting the study. Design Effects of a six-month cognitive intervention program on brain glucose metabolism were assessed in 36 patients with either aMCI (N = 21) or mild AD (N = 15) syndrome. The trial design is shown in Fig. 1. Briey, patients were randomly assigned to either a cognitive intervention treatment arm (IG) or active control condition treatment arm (CG), resulting in 9 aMCI patients and 8 AD patients for the mixed intervention groups (IGs), as well as 12 aMCI patients and 7 AD patients for the mixed control groups (CGs). During a six month period members of the IGs received

Fig. 1. Trial design. aMCI = amnestic mild cognitive impairment; AD = Alzheimers Disease; IG = intervention treatment arm; CG = control condition treatment arm.

340

S. F rster et al. / Effects of a 6-Month Cognitive Intervention on Brain Metabolism o

weekly sessions of 120-minutes group-based cognitive intervention, whereas the CGs participants met monthly and received pencil and paper exercises for self-study. The cognitive intervention was based on the theory of cognitive reserve [16] and was tailored to the cognitive and functional requirements of aMCI and mild AD patients according to the theory of Retrogenesis, as a theoretical basis for the selection of specic training tasks [17]. It focussed particularly on global cognitive functioning but addressed also noncognitive domains, e.g. mood and quality of life, that are thought to be impaired in aMCI and frank AD [13]. On the supposition that MCI resides within the cognitive continuum from normal aging to AD [18], differences between the intervention for MCI and for mild AD corresponded rather to a quantitative than a qualitative distinction [12]. In contrast to the tailored cognitive intervention, the CGs sessions focussed on exercises of isolated, sustained attention, which is supposed to be largely unimpaired, at least in mild AD [19]. The colleagues who conducted neuropsychological testing sessions and PET-scanning were blinded to subject classication and treatment plan; an instructor uninvolved in these steps administered the IGs and CGs sessions. PET imaging Baseline and six-months follow-up FDG-PET scans of 36 patients were performed successfully. All patients had fasted for at least six hours prior to FDGPET. Recordings were made using a Philips Allegro PET scanner with a 128 128 pixel matrix (pixel size 2 mm), an axial eld of view of 180 mm and a full width at half maximum (FWHM) resolution of 5.5 mm. Patients were asked to recline on the scanner bed, resting quietly with their eyes covered and their ears occluded. Each patients head was positioned within the aperture of the tomograph, and comfortably immobilized using a foam cushion. At 20 minutes after injection of FDG (mean 200 MBq, i.v.), an attenuation scan was obtained with a rotating [137 Cs] point source. A dynamic emission recording consisting of ve six-minute frames was obtained in the interval 30 to 60 minutes following the FDG injection. After visual inspection to exclude frames with unacceptable head motion, attenuation-corrected frames were summed into a single frame, and nal images were reconstructed iteratively using a three-dimensional row action maximum likelihood algorithm (3D-RAMLA).

Cognitive data analysis Prior to the screening-phase, the study was registered in a public trials registry (www.clinicaltrials.gov; ID: NCT00544856). Primary outcome was evaluated separately for aMCI and AD patients by comparing change in the global cognitive parameters (MMSE score [20] and Alzheimers Disease Assessment Scale total score of the cognitive subscale (ADAS-cog) [21]) in the cognitive intervention treatment arm versus change in the control condition treatment arm tested by the time (progression) by arm interaction term in a two-factor ANOVA. Further information regarding secondary outcome measures, not relevant for the present study, and the neuropsychological data analyses is described elsewhere [12]. PET data analysis SPM5 routines (Wellcome Department of Cognitive Neurology, London, UK) implemented in MATLAB (version 7.1) were used to perform basic image processing and voxel-based statistical analysis. All PET scans were spatially normalized using default transformation parameters to the SPM5 standard PET brain template in the Montreal Neurological Institute (MNI) space. Normalized images were composed of a 79 95 69 matrix, with 2 2 2 mm voxel size. These images were then smoothed using an isotropic Gaussian kernel (12 mm FWHM), which accommodates inter-individual anatomical variability, and thus improves the sensitivity of the statistical analysis [22]. FDG uptake scaling was performed using the reference cluster approach as described in detail by Yakushev et al. [23]. Briey, with this approach regional FDG uptake is normalized to that measured in brain regions dened a posteriori to be unaffected, relative to the results in a non-demented control group. This a posteriori method was shown to be more sensitive than the global mean approach in detecting disease-related metabolic disturbances in mild-to-moderate stages of neurodegenerative disorders [24, 25], especially in studies on dementia [23, 26, 27], [28]. For these comparisons we made use of previously obtained FDG-PET data from 11 healthy elderly subjects (5 female; mean age 59.0 10.9 years), who were noncomplainers, had an MMSE score above 28, had no signs of microangiopathy on MRI, no history of neurological- or major psychiatric diseases and did not receive any psychoactive medication [27]. The latter elderly control subjects had been examined in a resting-

S. F rster et al. / Effects of a 6-Month Cognitive Intervention on Brain Metabolism o

341

state on the same scanner and with image processing according to the identical protocol as for the patients. First of all, in order to characterize our patients neurobiological status at baseline, we compared baseline regional FDG uptake between all patients and nondemented controls, using an unpaired t-test. Then for evaluation of longitudinal changes in FDG uptake we performed within-group comparisons (in both the AD and aMCI subgroups of the IGs and the CGs) between baseline and end-of-study PET scans using paired t-tests. For a more specic evaluation of neurobiological intervention effects we elected to use a voxel-based bidirectional patient-group-by-time interactions design, referred to as difference of differences analysis, which assessed the differences in extent of metabolic changes between both the AD and aMCI subgroups of the IGs and the CGs between the initial scan and the end-of-study scan. We considered a statistical threshold of p < 0.001 uncorrected and a threshold for minimum spatial extent of 30 contiguous voxels for anatomical reporting of signicant changes in FDG uptake. Foci of signicant changes were automatically assigned by the SPM software to the coordinate system of the MNI space, and then converted to Talairach and Tournoux coordinates [29] for identication of the Brodmann areas and anatomic designations. In order to link the clinical neuropsychological outcome with the neurobiological PET ndings we performed additional correlation analyses between changes in FDG uptake in brain regions, which had showed the highest metabolic effects, and changes in primary and secondary outcome parameters in aMCI patients. Using a ROI-based approach (MARSBAR toolbox implemented in SPM5), we extracted normalized baseline and follow-up FDG uptake values and performed correlation analyses between FDG uptake changes and the respective changes in neuropsychological test scores.

Table 1 Demographic data of IGs and CGs subjects (n = 36) Igs Number AD aMCI Age (SD) in yrs. minmax Gender f / m Apo E4 allele (at least one) MMSE minmax Education (SD) in yrs. minmax 17 8 9 74.5 (8.6) 53.288.6 7/10 12 CGs 19 7 12 72.0 (7.1) 62.483.7 9/10 11 t-Test

n.s.

26.4 (2.4) 2230 12.2 (3.4) 817

26.1 (1.6) 2329 13.4 (4.1) 1121.5

n.s. n.s.

aMCI = amnestic mild cognitive impairment; AD = mild Alzheimers Disease; IGs = cognitive intervention groups; CGs = active control groups; minmax = ranges of scores; SD = standard deviation. t-Test results refer to post-hoc comparisons testing the differences of means in the adjacent columns

ance. Thirty-seven participants completed the study, with elimination of one aMCI patient from the IG arm due to poor PET image quality. Concurrent with the randomization process there was good matching between IG- and CG patient groups (IGs and CGs) for degree of cognitive impairment, education, age, gender, and for Apo-E genotype. All mild AD patients and one aMCI patient in the IG arm were on antidementia medication (acetylcholinesterase inhibitor and/or Memantine) at stable doses since at least since three months prior to study start and throughout the intervention/control condition until study end. Table 1 shows baseline demographic and clinical characteristics of all patients included in the PET data analysis. Neuropsychological cognitive outcome To detect intervention-related effects of the stagespecic cognitive intervention in the respective patient groups, we conducted a series of ANCOVAs with treatment (intervention treatment arm, IG, versus control treatment arm, CG), and progression (baseline versus end-of-study) as independent variables separately for AD and aMCI patients. The primary outcome measures MMSE and ADAS-cog scores served as dependent measures. Furthermore, we entered educational level (years of schooling) and patients age as covariates into the analysis.

RESULTS Of the 43 patients meeting initial inclusion criteria, 39 were randomly assigned to IG- or CG treatment arms, with exclusion of one patient due to concurrent physical illness, and three patients due to lack of compliance (less than 50% presence at the intervention sessions). Of these 39 patients, two aMCI patients dropped out from the IG arm because of new concurrent physical illness, or lack of compli-

342

S. F rster et al. / Effects of a 6-Month Cognitive Intervention on Brain Metabolism o

Table 2 Change from baseline in measures of efcacy in the cognitive intervention subgroups (IGAD ; IGMCI ) and in the active control subgroups (CGAD ; CGMCI ) regarding MMSE and ADAS-cog; standard deviation in () Target-variate Timepoint IGMCI ADAS-cog Baseline minmax 8.8 (3.1) 313 CGMCI 9.8 (4.3) 519 aMCI ANCOVA** IGAD CGAD AD ANCOVA**

t-Test* MMSE

t-Test*

Interaction Treatment 12.1 (5.3) 16.4 (4.8) Treatment and Progression 621 1323 (F1,11 = 4.4, p = .06, 2 = .282) (F1,17 = 4.7, p = .045, 2 = .22) Study end 7.4 (3.2) 11.7 (5.6) 11.4 (6.0) 16.4 (4.9) n.s. t(11) = 2.8, p = .02 n.s. n.s. Baseline 28.1 (1.6) 26.8 (1.5) Interaction Treatment 24.5 (1.6) 25.3 (1.5) n.s. minmax 2530 2429 and Progression 2227 2327 (F1,17 = 4.3, p = .05, 2 = .21) Main Effect Treatment (F1,17 = 6.8, p = 0.02, 2 = .29) Study end 28.3 (1.2) 26.0 (1.3) 25.0 (2.7) 24.4 (2.4) n.s t(11) = 2.8, p = .02 n.s. n.s.

aMCI = amnestic mild cognitive impairment; AD = mild Alzheimers Disease; ADAS-cog = Alzheimers Disease Assessment Scale; MMSE = Mini Mental State Examination; IGs = cognitive intervention groups; CGs = active control groups.; t-Test results refer to post-hoc comparisons testing the differences of means in the adjacent rows and columns.; **ANCOVA: univariate analysis of covariance; reported effects refer to the interaction between treatment and progression which reects whether a change of the dependent variables from baseline to follow-up is quantitatively different in the two treatment arms.

For AD patients, only the treatment factor approached signicance for ADAS-cog scores (F1,11 = 4.4, p = .06, 2 = .282). For aMCI patients, we determined a marginally signicant interaction effect between treatment and progression for ADAS-cog (F1,17 = 4.7, p = .045) and for MMSE (F1,17 = 4.3, p = .05). Furthermore, the analysis revealed a main effect for treatment regarding MMSE (F1,17 = 6.8, p < .01). Post-hoc t-tests suggest that interaction effects between treatment and progression occurred mainly due to performance decline in the CGMCI . See Table 2 for details.

FDG-PET data All patients relative to elderly non-demented controls showed reduced brain FDG uptake at baseline, mainly in the bilateral temporo-occipital association cortex (left worse than right), left temporal cortex, bilateral posterior cingulate cortices and precuneus, as well as in left prefrontal cortex (Fig. 2). MCI subgroups of the CGs and of the IGs (CGMCI and IGMCI ) showed similar baseline patterns of reduced brain FDG uptake, involving parieto-temporal cortex, posterior cingulate cortex, precuneus and prefrontal cortex (Fig. 3). We omit reporting anatomical coordinates, since this nding matches the typical anatomical distribution of AD-associated changes [3032].

Fig. 2. Baseline pattern of reduced FDG uptake in all patients (n = 36) as compared to non-demented elderly controls (n = 11); (P < 0.005 uncorrected).

S. F rster et al. / Effects of a 6-Month Cognitive Intervention on Brain Metabolism o

343

Fig. 3. Baseline patterns of reduced FDG uptake in a) IGMCI (n = 9) and in b) CGMCI (n = 12), as compared to non-demented elderly controls (n = 11); (P < 0.005 uncorrected).

There were no signicant differences in baseline brain FDG uptake between the respective CGs- and IGs. After six months, the AD subgroup of CGs (CGAD ) showed widespread bilateral declines in FDG uptake in parieto-temporal and parieto-occipital cortices as well as in left prefrontal cortex, while the AD subgroup of IGs (IGAD ) showed minor decline in FDG uptake in two single clusters located in the lingual gyrus and the left inferior temporal gyrus. The CGMCI showed widespread bilateral occipitotemporal (right more then left), parietal and prefrontal decline in FDG uptake, while the IGMCI showed no decline in FDG uptake (Fig. 4). Difference of differences analyses revealed strongest attenuated decline in FDG uptake in the subgroup of aMCI patients in the IGs (IGMCI) in bihemispheric cortical areas, including bilateral temporal, prefrontal- and anterior cingulate cortex (Fig. 5 a). Clusters with main peaks surviving the p < 0.001 (unc.) threshold were located in the left anterior temporal pole and the left anterior cingulate gyrus (Table 4, Fig. 6). The subgroup of AD patients in the IGs (IGAD ) revealed a more restricted attenuated decline in FDG uptake in the right temporal- and posterior cingulate cortex (Fig. 5 b), while none of these clusters survived the p < 0.001 (unc.) threshold. Correlation analyses between changes in normalized FDG uptake (in those brain regions which had showed the highest metabolic attenuation) and

changes in neuropsychological outcome parameters, revealed a single signicant correlation for total MADRS score in the IGMCI (Pearsons correlation coefcient: r = 0.61; p = 0.039 one-tailed), such that an increase in FDG uptake in the left anterior cingulateand anterior temporal pole was associated with a decrease in total MADRS score. However, after eliminating the cognition-related MADRS item (concentration), which had primarily contributed to the total MADRS score, only a weak non-signicant correlation remained (Pearsons correlation coefcient r = 0.398; p = 0.144 one-tailed). Detailed longitudinal outcome of secondary neuropsychological outcome parameters (i.e. MADRS, Trail Making Test, etc.) is described elsewhere [12].

DISCUSSION We used FDG-PET to map effects of a newly developed multicomponent cognitive intervention on brain energy metabolism in patients with aMCI and mild AD. Upon entering the study, participants were randomly assigned to intervention (IGs) or active control groups (CGs). In both CGs, signicant decline of FDG uptake in AD-typical brain areas during a relatively brief period of six months, conrmed that resting-state FDG-PET is a sensitive marker of disease progression [33]. Participation in the cognitive intervention program imparted cognitive benets in the aMCI sub-

344

S. F rster et al. / Effects of a 6-Month Cognitive Intervention on Brain Metabolism o

Fig. 4. Relative decline of brain FDG uptake in aMCI and AD- CGs and IGs after six months; (p < 0.005 uncorrected).

Fig. 5. Attenuated decline of brain FDG uptake after six months a) in the IGMCI relative to the CGMCI , b) in the IGAD relative to the CGAD; (p < 0.005 uncorrected).

group, which were reected by an attenuated decline in cerebral FDG uptake relative to that seen in the respective active control group.

Upon entry in the study, the whole sample of patients had signicantly reduced FDG uptake relative to a group of non-demented elderly control subjects.

S. F rster et al. / Effects of a 6-Month Cognitive Intervention on Brain Metabolism o Table 3 Regions of signicantly attenuated decline in brain FDG uptake after six months in the IGMCI relative to the CGMCI Location BA Talairach and Tournoux x y z 3.39 0.000 152 10 42 Peak P Cluster z-value uncorr. extension

345

L, Ant. 38 32 Temporal Pole L, Ant. 32 10 Cingulate Gyrus

28

20

3.37

0.000

1036

Cluster extension represents the number of contiguous voxels surpassing the threshold of p < 0.001. Bold markings delineate a cluster and the peak z-value within the cluster. Associated anatomic structures are indicated, along with designations of Brodman area (BA); R = right, L = left.

Main differences were seen in brain regions typically impaired in AD, including the bilateral temporooccipital association cortices, left temporal cortex, bilateral posterior cingulate cortices and precuneus, as well as in left prefrontal cortex (Fig. 2). This pattern conrms previous FDG-PET results in aMCI or mild to moderately diseased AD patients [31, 34, 35], suggesting that our stringent clinical inclusion criteria were effective in selecting a representative patient sample. Demographics indicated that the patient groups had similar education background, Apolipoprotein E4 allele status and similar baseline cognitive function, as assessed by MMSE and ADAS-cog scores at onset of the intervention. Longitudinal FDG-PET evaluation in both CGs (AD and aMCI) after six-months revealed widespread bilateral decline in metabolism throughout AD-typical

cortical areas (Fig. 4), consistent with the progression of the clinical scores of cognitive function (MMSE and ADAS-cog). Peaks in the declining FDG uptake overlapped with the pattern of hypometabolism at baseline relative to FDG uptake in the non-demented control group, consistent with an on-going disease process during only six months, as reported in previous PET studies with one-year follow-up [9, 35]. In those oneyear follow-up studies, FDG uptake was normalized to the global mean value, which we have shown to result in spurious detection of metabolic changes, which arise from bias due to undetected but real declines in metabolism [36]. Instead, we normalized regional FDG uptake to that measured in brain regions dened a posteriori to be unaffected, relative to the results in the non-demented elderly control group [23]. The a posteriori method is more sensitive than the global mean approach in detecting disease-related metabolic disturbances in mild-to-moderate stages of neurodegenerative disorders [24, 25], especially in studies on dementia [23, 2628]. In contrast to the widespread six month declines in FDG uptake seen in both CGAD and CDMCI (Fig. 4), we saw in the IGAD discrete decline with two single signicant clusters located in the lingual gyrus and the left inferior temporal gyrus, while the IGMCI showed no decline at all (Fig. 4). The latter nding strongly suggests positive effects of the cognitive intervention program on brain energy metabolism in the MCI subgroup. The relative preservation of normalized FDG uptake during six months especially in the IGMCI concurs with our clinical ndings showing a signicant change in global cognitive status, which seemed to be

Fig. 6. Overlay image on transaxial slices of an averaged MRI data set of 152 healthy subjects, showing peak clusters of attenuated decline of brain FDG uptake after six months in the left anterior cingulate cortex and the left anterior temporal pole in the IGMCI relative to the CGMCI . Radiological convention (Left is Right); set at >30 contiguous voxels passing the signicance threshold p < 0.005 uncorrected.

346

S. F rster et al. / Effects of a 6-Month Cognitive Intervention on Brain Metabolism o

driven by performance stabilisation in the IGMCI , contrasted with declining performance in the CGMCI (see Table 2). Given the progressive nature of AD and its prodromal stage aMCI, the present suggestion of cognitive stabilization may emerge as a benecial effect of the cognitive intervention. In order to test more objectively for timedependent effects of the cognitive intervention on brain metabolism, we made bidirectionally contrasted difference of differences analyses of the FDG-PET results in the aMCI and AD subgroups of the IGs and CGs. These SPM-based approaches revealed attenuated metabolic decline during six months of cognitive intervention mainly in the IGMCI subgroup. Instead, at the same statistical threshold the opposite contrasts revealed no attenuation of declining metabolism in the CGMCI relative to the IGMCI (or the respective AD subgroups), supporting the real presence of an attenuating effect on FDG decline in aMCI patients which had received the cognitive interventions. The effect of attenuated metabolic decline was more widespread and pronounced in the IGMCI than in the IGAD (Fig. 5 a,b), which is in line with the results from the clinical cognitive parameters (MMSE, ADAScog.), showing signicant changes only in the aMCI subgroup (Table 2). As mentioned before, changes in clinical cognitive parameters seemed to be attributed primarily to cognitive decline in the CGMCI , raising the question about whether there were a greater percentage of aMCI subjects with prodromal AD in the control group who may have deteriorated over the 6 month follow-up. This possibility can be denied. Group comparisons of IGMCI and CGMCI relative to elderly non-demented control subjects at baseline, revealed for both MCI subgroups a similar distributional pattern of FDG hypometabolism, involving AD-typical brain regions (see Fig. 3). The latter nding strongly suggests an inclusion of aMCI patients in a preclinical AD stage and is in line with our stringent MCI-diagnosis inclusion criteria. Furthermore, a number-needed-to treat (NNT) analysis in the aMCI subpopulation indicated effectiveness of the cognitive intervention, as improvement by means of 4-point change on ADAScog occurred only in the IGMCI , whereas cognitive decline appeared in both IG- and CGMCI [12]. In the aMCI subgroup we were able to map the peak areas of metabolic attenuation to the left anterior temporal pole and anterior cingulate cortex (ACC). The latter structure belongs to the fronto-limbic network and is well-known for its role in emotional and motivational control [37] as well as in attentional pro-

cessing [38], [39], [40]. To explore if mood-related effects might have lead to the observed PET effects, we performed an additional correlation analysis between the changes in normalized FDG uptake in the left anterior cingulate- as well as left anterior temporal pole, and the changes in MADRS score in the IGMCI . This analysis revealed a single signicant correlation for total MADRS score in the IGMCI , such that an increase in FDG uptake in the left anterior cingulate- and superior temporal gyrus was associated with a decrease in total MADRS score. However, after eliminating the cognition-related MADRS item (concentration), which had primarily contributed to the total MADRS score [12], only a weak, non-signicant correlation remained. Therefore, mood-specic effects might not have led to the observed attenuating effects on brain metabolism. However, we cannot rule out that cognition-related mood stimulating effects (i.e. increased concentration) of the intervention might have to some extent contributed to the observed PET effects. Probably due to the limited patient number and the restricted variance of clinical scores in this relatively short time period of six months, correlation analyses between FDG uptake changes and performance changes in the global cognitive domains (MMSE, ADAS cog) did not show any signicant correlation (data not shown). We feel that the specic linkage between neuropsychological and neurobiological outcome remains to be established in future studies in larger patient populations. Overall, our ndings strongly suggest a positive short-term effect of the cognitive intervention program on brain energy metabolism in aMCI patients, which may manifest in cognitive benets. We modelled our study on an earlier FDG-PET follow-up study investigating effects of a 14-day healthy longevity lifestyle program on cognition and cerebral energy metabolism in a group of non-demented subjects with mild agerelated memory complaints [11]. In that study, the intervention group had after two weeks a 5% relative decrease in normalized FDG uptake in the left dorsolateral prefrontal cortex, which the authors interpreted to reveal an effect of training on cognitive efciency of a brain region involved in working memory. Unlike in that study, we selected patients with objective cognitive impairment, suggesting that our ndings of attenuated declines relate to moderation of a pathological process. In the only other FDG-PET study of cognitive training in AD patients [10], there was a six-month decline in resting-state cerebral glucose metabolism,

S. F rster et al. / Effects of a 6-Month Cognitive Intervention on Brain Metabolism o

347

most notably in the temporo-parietal region, in a group of AD patients who had only received social support. In that study, FDG-PET changes evoked by a visual recognition task served as an indicator of therapeutic efcacy in AD. The AD patients who had received cognitive training with and without pharmacological treatment had a more distinct pattern of FDG activations after six months, whereas the subgroup with both interventions showed some improvement in cognitive performance, as measured by the MMSE. Despite diverse conceptual and methodological differences between that and the present studies, there is agreement that FDG-PET can detect effects of intervention on cerebral metabolism during six months progression of impaired cognition. LIMITATIONS The relatively small number of participants in this study limits its statistical power. Furthermore, this study design reveals only group differences in FDG uptake and cognitive scores, which may however be offset by the benets of the prospective setting, in which the groups were well-matched for demographics and baseline metabolism. Direct result comparisons between the AD and MCI subgroups might be confounded by differences in the intervention design and should be interpreted with caution. We allowed a more permissive statistical threshold of p < 0.005 (uncorrected) for the visualisation of PET results. However, application of the same permissive threshold for the respective negative SPM t-contrasts did not bear any signicant results, which would have indicated an increased risk of false positive results. ACKNOWLEDGMENTS

[3] [4] [5]

[6]

[7] [8]

[9]

[10]

[11]

[12]

[13]

[14]

The authors thank Dr. Paul Cumming for critical review of the manuscript. FINANCIAL DISCLOSURES
[15]

The authors report no biomedical nancial interests or potential conicts of interest.

[16]

REFERENCES
[1] [2] Buschert V, Bokde AL, Hampel H (2010) Cognitive intervention in Alzheimer disease. Nat Rev Neurol 6, 508-517. Valenzuela M, Sachdev P (2009) Can cognitive exercise prevent the onset of dementia? Systematic review of randomized

[17]

clinical trials with longitudinal follow-up. Am J Geriatr Psychiatry 17, 179-187. Valenzuela MJ, Sachdev P (2006) Brain reserve, dementia: a systematic review. Psychol Med 36, 441-454. Belleville S (2008) Cognitive training for persons with mild cognitive impairment. Int Psychogeriatr 20, 57-66. Sitzer DI, Twamley EW, Jeste DV (2006) Cognitive training in Alzheimers disease: a meta-analysis of the literature. Acta Psychiatr Scand 114, 75-90. Olazaran J, Reisberg B, Clare L, Cruz I, Pena-Casanova J, Del Ser T, Woods B, Beck C, Auer S, Lai C, Spector A, Fazio S, Bond J, Kivipelto M, Brodaty H, Rojo JM, Collins H, Teri L, Mittelman M, Orrell M, Feldman HH, Muniz R (2010) Nonpharmacological therapies in Alzheimers disease: A systematic review of efcacy. Dement Geriatr Cogn Disord 30, 161-178. Herholz K (2003) PET studies in dementia. Ann Nucl Med 17, 79-89. Reiman EM, Caselli RJ, Yun LS, Chen K, Bandy D, Minoshima S, Thibodeau SN, Osborne D (1996) Preclinical evidence of Alzheimers disease in persons homozygous for the epsilon 4 allele for apolipoprotein E. N Engl J Med 334, 752-758. Alexander GE, Chen K, Pietrini P, Rapoport SI, Reiman EM (2002) Longitudinal PET Evaluation of Cerebral Metabolic Decline in Dementia: A Potential Outcome Measure in Alzheimers Disease Treatment Studies. Am J Psychiatry 159, 738-745. Heiss WD, Kessler J, Slansky I, Mielke R, Szelies B, Herholz K (1993) Activation PET as an instrument to determine therapeutic efcacy in Alzheimers disease. Ann N Y Acad Sci 695, 327-331. Small GW, Silverman DH, Siddarth P, Ercoli LM, Miller KJ, Lavretsky H, Wright BC, Bookheimer SY, Barrio JR, Phelps ME (2006) Effects of a 14-day healthy longevity lifestyle program on cognition and brain function. Am J Geriatr Psychiatry 14, 538-545. Buschert VC, Friese U, Teipel SJ, Schneider P, Merensky W, Rujescu D, Moller HJ, Hampel H, Buerger K (2011) Effects of a newly developed cognitive intervention in amnestic mild cognitive impairment and mild Alzheimers disease:. A Pilot Study. J Alzheimers Dis, doi10.3233/JAD-2011-100999. Lachman ME, Weaver SL, Bandura M, Elliott E, Lewkowicz CJ (1992) Improving memory and control belief through cognitive restructuring and self-generated strategies. J Gerontol 47, P293-P299. McKhann G, Drachman D, Folstein M, Katzman R, Price D, Stadlan EM (1984) Clinical diagnosis of Alzheimers disease: report of the NINCDS-ADRDA Work Group under the auspices of Department of Health and Human Services Task Force on Alzheimers Disease. Neurology 34, 939944. Scheltens P, Barkhof F, Leys D, Pruvo JP, Nauta JJ, Vermersch P, Steinling M, Valk J (1993) A semiquantative rating scale for the assessment of signal hyperintensities on magnetic resonance imaging. J Neurol Sci 114, 7-12. Stern Y (2006) Cognitive reserve and Alzheimer disease. Alzheimer Dis Assoc Disord 20, 112-117. Reisberg B, Franssen EH, Hasan SM, Monteiro I, Boksay I, Souren LE, Kenowsky S, Auer SR, Elahi S, Kluger A (1999) Retrogenesis: clinical, physiologic, and pathologic mechanisms in brain aging, Alzheimers and other dementing processes. Eur Arch Psychiatry Clin Neurosci 249(3), 28-36.

348 [18]

S. F rster et al. / Effects of a 6-Month Cognitive Intervention on Brain Metabolism o Petersen RC, Smith GE, Waring SC, Ivnik RJ, Tangalos EG, Kokmen E (1999) Mild cognitive impairment: clinical characterization and outcome. Arch Neurol 56, 303-308. Perry RJ, Hodges JR (1999) Attention and executive decits in Alzheimers disease. A critical review. Brain 122(3), 383-404. Folstein M, Folstein S, McHugh P (1975) Mini Mental State A practical method for grading the cognitive state of patients for the clinician. J Psychiatric Res 12, 189-198. Rosen WG, Mohs RC, Davis KL (1984) A new rating scale for Alzheimers disease. Am J Psychiatry 141, 1356-1364. Friston KJ, Frith CD, Liddle PF, Frackowiak RS (1991) Comparing functional (PET) images: the assessment of signicant change. J Cereb Blood Flow Metab 11, 690-699. Yakushev I, Hammers A, Fellgiebel A, Schmidtmann I, Scheurich A, Buchholz HG, Peters J, Bartenstein P, Lieb K, Schreckenberger M (2009) SPM-based count normalization provides excellent discrimination of mild Alzheimers disease and amnestic mild cognitive impairment from healthy aging. Neuroimage 44, 43-50. Borghammer P, Aanerud J, Gjedde A (2009) Data-driven intensity normalization of PET group comparison studies is superior to global mean normalization. Neuroimage 46, 981988. Borghammer P, Cumming P, Aanerud J, Forster S, Gjedde A (2009) Subcortical elevation of metabolism in Parkinsons disease - A critical reappraisal in the context of global mean normalization. Neuroimage 47, 1514-1521. Yakushev I, Landvogt C, Buchholz HG, Fellgiebel A, Hammers A, Scheurich A, Schmidtmann I, Gerhard A, Schreckenberger M, Bartenstein P (2008) Choice of reference area in studies of Alzheimers disease using positron emission tomography with uorodeoxyglucose-F18 . Psychiatry Res 164, 143-153. Forster S, Teipel S, Zach C, Rominger A, Cumming P, Fougere CL, Yakushev I, Haslbeck M, Hampel H, Bartenstein P, Burger K (2010) FDG-PET mapping the brain substrates of visuo-constructive processing in Alzheimer s disease. J Psychiatr Res 44, 462-469. Forster S, Vaitl A, Teipel SJ, Yakushev I, Mustafa M, la Fougere C, Rominger A, Cumming P, Bartenstein P, Hampel H, Hummel T, Buerger K, Hundt W, Steinbach S (2010) Functional representation of olfactory impairment in early Alzheimers disease. J Alzheimers Dis 22, 581-591. Lancaster JL, Woldorff MG, Parsons LM, Liotti M, Freitas CS, Rainey L, Kochunov PV, Nickerson D, Mikiten SA, Fox PT (2000) Automated Talairach atlas labels for functional brain mapping. Hum Brain Mapp 10, 120-131. [30] Ishii K, Willoch F, Minoshima S, Drzezga A, Ficaro EP, Cross DJ, Kuhl DE, Schwaiger M (2001) Statistical brain mapping of 18F-FDG PET in Alzheimers disease: validation of anatomic standardization for atrophied brains. J Nucl Med 42, 548-557. Minoshima S, Giordani B, Berent S, Frey KA, Foster NL, Kuhl DE (1997) Metabolic reduction in the posterior cingulate cortex in very early Alzheimers disease. Ann Neurol 42, 8594. Drzezga A, Riemenschneider M, Strassner B, Grimmer T, Peller M, Knoll A, Wagenpfeil S, Minoshima S, Schwaiger M, Kurz A (2005) Cerebral glucose metabolism in patients with AD and different APOE genotypes. Neurology 64, 102-107. F rster S, Grimmer T, Miederer I, Henriksen G, Youse o BH, Graner P, Wester HJ, F rstl H, Kurz A, Dickerson o BC, Bartenstein P, Drzezga A (2011) Regional expansion of hypometabolism in AD follows amyloid-deposition with temporal delay. Biological Psychiatry (in press). Minoshima S, Frey KA, Koeppe RA, Foster NL, Kuhl DE (1995) A diagnostic approach in Alzheimers disease using three-dimensional stereotactic surface projections of uorine18-FDG PET. J Nucl Med 36, 1238-1248. Drzezga A, Lautenschlager N, Siebner H, Riemenschneider M, Willoch F, Minoshima S, Schwaiger M, Kurz A (2003) Cerebral metabolic changes accompanying conversion of mild cognitive impairment into Alzheimers disease: a PET follow-up study. Eur J Nucl Med Mol Imaging 30, 1104-1113. Borghammer P, Jonsdottir KY, Cumming P, Ostergaard K, Vang K, Ashkanian M, Vafaee M, Iversen P, Gjedde A (2008) Normalization in PET group comparison studies the importance of a valid reference region. Neuroimage 40, 529-540. Wagner G, Koch K, Schachtzabel C, Schultz CC, Sauer H, Schlosser RG (2011) Structural brain alterations in patients with major depressive disorder and high risk for suicide: Evidence for a distinct neurobiological entity? Neuroimage 54, 1607-1614. Haupt S, Axmacher N, Cohen MX, Elger CE, Fell J (2009) Activation of the caudal anterior cingulate cortex due to taskrelated interference in an auditory Stroop paradigm. Hum Brain Mapp 30, 3043-3056. Crottaz-Herbette S, Menon V (2006) Where and when the anterior cingulate cortex modulates attentional response: combined fMRI and ERP evidence. J Cogn Neurosci 18, 766-780. Margulies DS, Kelly AM, Uddin LQ, Biswal BB, Castellanos FX, Milham MP (2007) Mapping the functional connectivity of anterior cingulate cortex. Neuroimage 37, 579-588.

[19] [20]

[31]

[21] [22]

[32]

[23]

[33]

[34]

[24]

[35]

[25]

[26]

[36]

[37]

[27]

[38]

[28]

[39]

[29]

[40]