Mild cognitive impairment (MCI) and actual retrieval performance affect cerebral activation in the elderly

Neurobiology of Aging 28 (2007) 404–413

Mild cognitive impairment (MCI) and actual retrieval performance affect cerebral activation in the elderly Reinhard Heun a,b,∗ , Katrin Freymann a , Michael Erb c , Dirk T. Leube d , Frank Jessen a , Tilo T. Kircher d,e , Wolfgang Grodd c a

Department of Psychiatry, University of Bonn, Sigmund-Freud-Strasse 25, 53105 Bonn, Germany Department of Psychiatry, University of Birmingham, Mindelsohn Way, Birmingham B15 2QZ, UK Section on Experimental Magnetic Resonance of CNS, Department of Neuroradiology, University of Tuebingen, Hoppe-Seyler-Strasse 3, D-72076 Tuebingen, Germany d Department of Psychiatry, University of Tuebingen, Osianderstrasse 24, 72076 Tuebingen, Germany e Department of Psychiatry, RWTH Aachen University Hospital, Pauwelstr. 30, D-52074 Aachen, Germany b

c

Received 5 August 2005; received in revised form 16 January 2006; accepted 27 January 2006 Available online 13 March 2006

Abstract Cerebral activation in the elderly may depend on general cognitive decline as well as actual retrieval performance. Consequently, activation between subjects with and without Mild Cognitive Impairment (MCI), and between remembered and non-remembered words was compared. Twenty-one MCI and 29 healthy control subjects learned 180 nouns. During retrieval, subjects had to discriminate these and 180 distractor words. fMRI identified response-related activation. Most retrieval-related activation was comparable in both groups. However, MCI subjects showed more activation in the prefrontal cortex than controls during processing of hits and correct rejections. Hits showed increased activation than misses in the precuneus and left lateral parieto-occipital cortex; misses showed more activation than correct rejections in the precuneus to cuneus. Verbal retrieval activated a large common network in the elderly independently of MCI. Increased activation in MCI subjects in prefrontal cortex depends on response category. Activation differences between response categories might reflect success (hits) and effort (misses). Increased retrieval-related activation may be used as early marker in subjects at risk of Alzheimer’s disease. © 2006 Elsevier Inc. All rights reserved. Keywords: Mild cognitive impairment (MCI); Elderly; Functional imaging; Memory; Retrieval; Retrieval effort; Retrieval success

1. Introduction 1.1. Functional MRI (fMRI) in mild cognitive impairment Mild Cognitive Impairment (MCI) is characterized by significant memory impairment, which is not severe enough to interfere with usual activities of daily living [41]. It is assumed to be an early stage of Alzheimer’s disease [4,27]. It might reduce retrieval success and increase the retrieval effort during memory tasks in affected subjects.

∗

Corresponding author. Tel.: +44 121 678 2360; fax: +44 121 678 2351. E-mail address: [email protected] (R. Heun).

0197-4580/$ – see front matter © 2006 Elsevier Inc. All rights reserved. doi:10.1016/j.neurobiolaging.2006.01.012

Memory related cerebral activation is different in the elderly compared to younger subjects [10]. The Harold model indicates that elderly who still perform adequately compensate by recruiting additional bilateral brain areas when performing memory tasks [8,9]. The need for compensation might be even more pronounced in subjects with Mild Cognitive Impairment (MCI). According to Ref. [17], increased activation in the hippocampal formation and parahippocampal gyrus during visual encoding in subjects with MCI in comparison with healthy controls may represent a compensatory response to accumulating pathology. This compensation is lost in later stages of cognitive decline such as Alzheimer’s disease [AD, 34, 18, 46]. Logan et al. [33] describe under-recruitment and non-selective recruitment of frontal lobe areas during encoding as part of age-associated

R. Heun et al. / Neurobiology of Aging 28 (2007) 404–413

cognitive decline. Rombouts [45] reported reduced deactivations during an encoding and a working memory tasks in several brain areas in subjects with MCI in comparison with controls. In another paradigm this group reported delayed BOLD responses in subjects with early AD [47]. However, only few studies have compared cerebral activation in elderly subjects with and without memory impairment during retrieval [50]. During retrieval of previously presented photographs subjects with MCI showed reduced prefrontal but increased parietal activation in comparison with healthy controls [35]. As far as we are aware direct comparison of activation during verbal retrieval by retrieval success have not yet been performed in elderly with versus without cognitive deficits but might help to assess whether possible activation differences are diagnosis-specific and/or success-related. 1.2. fMRI of retrieval success and retrieval effort According to Ref. [48] retrieval success refers to the processes that are selectively engaged when a retrieval attempt is successful. Functional imaging studies investigating the effect of retrieval success on cerebral activation by comparing high performance with low performance conditions observed that retrieval success was related to increased activation of the right anterior prefrontal cortex [6], the right superior frontal cortex [32], the left medio-temporal lobe [38], and the precuneus [32]. In a reanalysis of four PET studies, Tulving et al. [52] detected the frontal lobe, the medial temporal lobe, the cingulate gyrus, the precuneus, the medial occipital lobe, the cerebellum and the brainstem to be related to recognition success. Event-related fMRI during verbal recognition of previous learned words allows discriminating cerebral activation by success-related response categories: i.e. hits (correctly identified previously presented (old) items), misses (not recognized old items), correct rejections (correctly identified new items) and false alarms (new items falsely assumed to have been seen presented before) [28]. In such studies, activation related to retrieval success has usually been measured by comparing activation during hits versus correct rejections [6,5,31]. We propose that the comparison of hits versus misses might be a more appropriate even though statistically less powerful indicator of successful retrieval than the comparison hits versus correct rejection because novelty detection represents a possibly confounding cognitive process during processing of later correct rejections. Novelty detection may be relevant for correct rejection of new items but may be less relevant for both hits and misses. Retrieval effort refers to the processes engaged by an attempt to retrieve information from memory [48]. The retrieval effort is also likely to be more intense for items, which the subject had failed to remember (misses) than for correctly recognized or correctly rejected items (hits and correct rejections). Consequently, activation which is more intense during misses than hits or correct rejections is likely to be related to retrieval effort.

405

In one study, [51] failed to find activation differences between hits, false alarms, misses and correct rejections in an event-related verbal recognition paradigm. In contrast, [31] observed activation of the left medial and lateral parietal cortex related to successful retrieval of hits compared to correct rejections. In our own study in healthy volunteers, we observed that hits in comparison with misses caused increased activations in the left inferior frontal gyrus, precentral gyrus, medial parietal cortex, both medial parieto-occipital lobes and the brain stem [26]. Consequently, retrieval-related activation in the elderly is likely to depend on the subjects’ general and actual retrieval performance (and/or on their interaction). The former is indicated by the presence of MCI, the latter on the actual retrieval category. Consequently, the following hypotheses were investigated: (1) There are differences in cerebral activation in subjects with versus without MCI indicating differences in retrieval effort. These might depend on the actual retrieval categories. (2) There are differences in cerebral activation in different response categories indicating differences in retrieval success, i.e. hits versus misses and differences in retrieval effort, i.e. misses versus correct rejections. These differences may depend on the presence of MCI.

2. Methods 2.1. Participants A total of 29 elderly controls with unimpaired memory and 21 participants with mild cognitive impairment (MCI) were recruited from the community of Tuebingen and the surrounding area via newspaper advertisement. Subjects were screened for major medical and psychiatric diseases. Those with a history of concurrent cerebrovascular disease, stroke, hydrocephalus, major depression, and relevant alcohol abuse [ICD-10 criteria, 54] were excluded. To evaluate global cognitive functioning and a possible presence of dementia, subjects underwent neuropsychological assessment using the Structured Interview for the Diagnosis of Dementia of the Alzheimer type, Multi-infarct Dementia and Dementia of Other Aetiology according to ICD-10 and DSM-IV [SIDAM, 7], the Mini-Mental State Examination [MMSE, 20], and a German version of the Verbal Learning and Memory Test [VLMT, 24]. ICD-10 diagnoses were made in a consensus conference of psychiatrists and neuropsychologists. MCI was diagnosed according to the criteria of Ref. [41]. All subjects were right-handed according to the Edinburgh Handedness Inventory [39]. The study had been approved by the local ethical committee. Participants gave written informed consent prior to participation.

406

R. Heun et al. / Neurobiology of Aging 28 (2007) 404–413

2.2. Experimental design

2.3. Image acquisition

Two event-related paradigms were performed to analyse encoding (paradigm 1) and retrieval (paradigm 2). Paradigm 1 (encoding) was designed to identify brain regions associated with successful encoding. Subjects were instructed to learn 180 visually presented words (nouns) for later recognition and to press the right-sided of two response buttons, once for every word. The encoding paradigm is not subject of the present publication. (In previous paradigms in healthy controls, we could not show any significant activation differences during encoding between later remembered and non-remembered items, thus the present analyses focus on activation differences during retrieval.) Paradigm 2 (recognition) was intended to study brain regions involved in the successful recognition of words. About 180 learned words from the encoding paradigm (targets) were randomly intermixed with 180 new words (distractors). Targets and distractors were matched with regard to imagery, concreteness and meaningfulness [2,40]. The whole experiment was performed in six consecutive blocks. Each block consisted of 90 words that were presented for 2 s each. To allow monitoring of the complete hemodynamic response, every individual word was followed by a string of eight x’s according to the average word length for 4–12 s (jittering); in addition 20 null events were included per block (eight x’s for another 4–12 s). All words to be learned in the encoding blocks 1 and 4, consisting of 90 new words/block, were again presented in the following two recognition blocks (2, 3 and 5, 6, respectively), each consisting of 45 learned target words and 45 distractors/block). Subjects had to decide whether they had learned the presented word before by button press, the right-sided button for a previously learned word and the left-sided button for new words not studied before. Four different types of responses were possible: hits (i.e. correctly identified target words), misses (i.e. target words incorrectly regarded as new), correct rejections (i.e. distractor words correctly classified as new), and false alarms (i.e. distractor words incorrectly regarded as learned before). The task was explained and practised if necessary for some minutes outside the scanner, until the subjects fully understood the procedures. Due to the length of the paradigm and to avoid unnecessary fatigue we did not use a prolonged practice period. Before every block, the following task was again explained until the subjects fully understood what to do next. Measures of recognition: accuracy d and response bias β were calculated according to signal detection theory [23]. d = 0 indicates no difference between the numbers of yesresponses for old and new items, i.e. a subject is just guessing. The better the subject differentiates between old and new items the higher becomes d . A β of 1 indicated no response bias in either the positive or negative direction; if β is smaller than 1 there is a positive response bias (i.e. the person responds more easily with yes, respectively known in the present study) and vice versa.

Encoding and retrieval were consecutively performed in the MR-scanner. The subjects lay supine in a 1.5 T whole-body scanner (Siemens Sonata; Siemens, Germany). Head movement was limited by foam padding within the head coil. Words were projected on a transparent screen, which could be seen via a mirror attached to the head coil in front of the subject’s head. All words could easily be seen without eye movement. If necessary, participants wore fMRI-compatible glasses to ensure optimal visual acuity. Twenty-five parallel axial slices (thickness = 4.5 mm, gap = 1.0 mm) were obtained across the complete brain volume using an echo planar imaging sequence (64 × 64 matrix, field of view = 192 mm, TE = 40 ms, TR = 2 s, flip angle = 90◦ , voxel size = 3 × 3 × (4.5 + 1) mm3 ). The jittering introduced by variable inter-stimulus intervals caused nonidentical starting slices for every new stimulus and allowed for a complete mapping of the event-related hemodynamic response functions [36]. Each of the six blocks consisted of 467 brain scan lasting 15 min and 34 s. In total, 2802 scans were acquired in 1 h, 33 min and 34 s. High-resolution images obtained with a T1-weighted 3D turbo flash sequence (MPRAGE; 176 sagittal slices, thickness 1.0 mm, 256 × 256 matrix, field of view 256 mm, TE 3.22 ms, TI 660 ms, TR 1300 ms, voxel size = 1 × 1 × 1 mm) served as the anatomical reference for functional images. With resting intervals used for explaining the forthcoming tasks, the total paradigm lasting below 2 h. The scanning was well tolerated and there was no evidence that the diagnosis influenced fatigue levels (for the description of drop-outs and completeness of data collection see Section 3). 2.4. Image pre-processing The functional data were pre-processed and statistically analysed using SPM2 (Wellcome Department of Imaging Neuroscience). After discarding the first five volumes data were slice-time corrected, realigned to the first volume in the time series across all encoding and recognition blocks to correct for head movement; additionally, the residual movement related variance caused by susceptibility-by-movement interactions was modelled to further reduce movement related variance. These images were then co-registered with the individual T1-weighted structural image. Since one cannot suppose that the brain size of an elderly sample corresponds to the standard brain provided by SPM2, an own template image was created using the individual T1-weighted structural images of all participants of the study. Template creation was part of a voxel-based morphometry procedure: each T1 image was normalized to the SPM MNI (Montreal Neurological Institute) template, which approximates the Talairach space. These normalized images were then averaged and smoothed with a Gaussian kernel of 8 mm FWHM [21].

R. Heun et al. / Neurobiology of Aging 28 (2007) 404–413

All functional EPI images were normalized to this template image with a cubic voxel size (3 × 3 × 3 mm3 ) using transformation parameters calculated from the T1 images. The functional images were further smoothed using an isotropic Gaussian filter of 12 mm FWHM. The total search volume consisted of 50490 voxels (corresponding to 1363 cm3 ). 2.5. Statistical analysis Stimuli of the recognition blocks were individually classified into the four main event types (hits, misses, false alarms, correct rejections) according to the subject’s responses. Words that elicited no reaction were modelled separately, but were not considered in later analyses. The hemodynamic response to the onset of each event type was modelled with the canonical hemodynamic response function (HRF) of SPM2 and the first-order time derivative. A high-pass filter with a cut-off period of 128 s was applied to filter out low frequency variations. In addition to the regressors representing the event types hits, misses, false alarms, and correct rejections, six covariates were included for each subject to assess transformation and residual movement related artefacts (the six rigid body translations and rotations parameters estimated at the realignment step). Single-subject contrast maps were determined for the contrasts of interest for the present study according to the general linear model approach of SPM2. These contrast images were entered into random effects analyses using one-sample T-tests to determine within-group effects or two-sample T-tests for betweengroup analyses. All statistical maps were transformed into Z maps. We applied a significance threshold of p < 0.05 or p < 0.001 (corrected for multiple comparisons according to the random field theory) for the assessment of main contrasts. In order to prevent from being too conservative and obtaining too many false negative results, we decided to report activation above a threshold of p < 0.001 (uncorrected) for comparisons between individual diagnostic or individual response categories (in agreement with Daselaar et al. [14,15]). To reduce the likelihood of false positive results in such comparisons only clusters of at least 40 supra-threshold voxels are reported. The following hypotheses/contrasts were examined: there are brain areas which are activated by verbal retrieval in the groups of subject without and with MCI (comparison words versus baseline; this reflects a precondition for the following study hypotheses, results see Fig. 1a and b). (1) Subjects with MCI and healthy controls differ in their cerebral activation caused by hits (or correct rejections or misses or false alarms) (comparison of activation by individual response categories in subjects with versus without MCI, results see Fig. 1c and d). (2) There are brain areas that differ in their activation depending on the retrieval categories (comparison of activation between individual retrieval categories). These effects may be found in the whole group and also

407

in the subgroups (the comparison was thus done for the whole sample and repeated for both subgroups. Results see Fig. 1e and f) Exact BOLD responses for representative activations were calculated by averaging signal intensities of all subthreshold voxels of individual subthreshold clusters corresponding to individual delays and response categories (or their differences), bars in graphs (Fig. 2a–d) correspond to standard errors of means.

3. Results 3.1. Demographical data and neuropsychological results One healthy individual and one MCI patient had to be excluded from the fMRI analyses post-hoc due to extremely negative response bias resulting in very few hits and false alarms (less than 5% of all responses), but many misses and correct rejections. Demographical and neuropsychological data of the remaining individuals are presented in Table 1. There were no significant differences between the two groups regarding age (t = −1.189, d.f. = 46, p = 0.262) and sex (χ2 = 0.157, d.f. = 1, p = 0.692). As expected, Ttests revealed that there were significant group differences in all neuropsychological markers (SIDAM: t = 5.665, d.f. = 46, p < 0.001; MMSE: t = 6.069, d.f. = 46, p < 0.001; verbal learning/immediate recall: t = 3.391, d.f. = 46, p = 0.001; delayed recall: t = 3.601, d.f. = 46, p = 0.001; recognition: t = 2.593, d.f. = 46, p = 0.013). 3.2. Behavioural results Table 1 includes the behavioural performance of the two groups during the fMRI experiment. T-tests revealed no significant differences regarding proportions of hits (t = 1.553, d.f. = 46, p = 0.127) and of correct rejections (t = 1.468, d.f. = 46, p = 0.149). As expected the d index differed between healthy controls and MCI subjects (t = 3.307, d.f. = 46, p = 0.002). Nevertheless, as can be seen from Table 1, the retrieval accuracy in healthy individuals and MCI subjects was well above chance. No differences in the response bias measure β could be found (t = 0.176, d.f. = 46, p = 0.861). A group × response category ANOVA with response category as repeated measure and reaction time as dependent variable revealed a significant main effect of response category (F = 34.586, d.f. = 3, p < 0.001), but not of group (F = 0.223, d.f. = 1, p = 0.639). All subjects responded within the time limit predetermined by the variable interstimulus interval (see Section 2). One healthy individual and one person with MCI performed only two recognition blocks, one healthy subject performed three blocks, while all other subjects completed all four retrieval blocks. This indicates that fatigue is unlikely to have differed between both groups.

408

R. Heun et al. / Neurobiology of Aging 28 (2007) 404–413

Fig. 1. Surface rendered images of cerebral activation during verbal retrieval (paradigm 2), in subjects with and without Mild Cognitive Impairment (MCI) by diagnosis and by response category; from left to right: right lateral, left lateral, left medial, right medial view. (a) (first row) indicates cerebral activation related to verbal retrieval in subjects without MCI (activation by words; words of all categories, i.e. hits, misses, false alarms and correct rejections). (b) (second row) Cerebral activation related to verbal retrieval in subjects with MCI. (c) (third row) Cerebral activation differences between subjects with vs. without MCI during retrieval of hits. (d) (fourth row) Cerebral activation differences between subjects with vs. without MCI during retrieval of correct rejections. (e) (fifth row) Differences of cerebral activation between hits vs. misses in both groups. (f) (sixth row) Differences of cerebral activation between misses vs. correct rejections in both groups.

R. Heun et al. / Neurobiology of Aging 28 (2007) 404–413

409

Fig. 2. BOLD responses in different brain areas. The curves indicate change of signal intensity (%) over time (s). Bars indicate standard errors of mean at different delays. (a) Activation of the right inferior frontal gyrus by hits in subjects with (line) vs. without MCI (dotted line, BOLD response in the superior frontal gyrus was equivalent). (b) Activation of the left middle frontal gyrus by correct rejections in subjects with (line) vs. without MCI (dotted line). (c) Activation of lateral parieto-occipital cortex in the combined sample by hits (line) and by misses (dotted line, BOLD response in the precuneus to cingulate gyrus was equivalent). (d) Activation of the precuneus to cuneus in the combined sample by misses (line) vs. correct rejections (dotted line).

Table 1 Demographic characteristics, neuropsychological results, and performance during fMRI examination (for statistical comparisons see Section 3) Elderly controls (n = 28) Demographic characteristics Female: n (%) Age (years): mean (S.D.; range) Neuropsychological results SIDAM sum score: mean (S.D.; range)a MMSE score: mean (S.D.; range)b Verbal learning test Immediate recall (number of words): mean (S.D.; range)c Delayed recall (number of items): mean (S.D.; range)d Recognition (number of words): mean (S.D.; range)e Retrieval performance during fMRI examination % hits: mean (S.D.) % correct rejections: mean (S.D.) D : mean (S.D.) β: mean (S.D.) Reaction times during hits: mean (S.D.) Reaction time during misses (S.D.) Reaction time during false alarms: mean (S.D.) Reaction time during correct rejections: mean (S.D.)

Subjects with MCI (n = 20)

17 (60.7) 67.5 (5.4; 60–81)

11 (55.0) 69.7 (7.1; 59–82)

52.2 (2.0; 46–55) 28.9 (1.1; 27–30)

48.1 (3.1; 41–54) 26.6 (1.5; 24–30)

48.8 (6.6; 38–60) 9.5 (2.2; 6–13) 12.0 (2.5; 5–15)

42.4 (6.2; 28–59) 7.2 (2.4; 4–13) 9.7 (3.6; 0–14)

73.26 (12.44) 77.72 (14.05) 1.50 (0.44) 1.31 (0.70) 1474.75 (291.41) 1906.03 (337.77) 1797.36 (422.23) 1676.86 (276.72)

67.06 (15.19) 71.19 (16.72) 1.12 (0.35) 1.27 (0.70) 1502.85 (254.36) 1854.40 (388.84) 1681.55 (302.19) 1655.95 (348.23)

a Structured Interview for the Diagnosis of dementia of the Alzheimer type, multi-infarct dementia and dementia of other aetiology according to ICD-10 and DSM-IV, scores may range from 0 to 55 (worst to best). b Mini-Mental State Examination, scores may range from 0 to 30 (worst to best). c Immediate recall/sum of five learning trials of 15 words each, scores may range from 0 to 75 (worst to best). d Delayed recall of the 15-words list after 20 min; scores may range from 0 to 15 (worst to best). e Correct recognitions of the 15 words out of a 50 words list less the false alarms (worst to best).

410

R. Heun et al. / Neurobiology of Aging 28 (2007) 404–413

3.3. Imaging results 3.3.1. Activation differences in MCI subjects and healthy control subjects Fig. 1a and b indicates cerebral activation during verbal retrieval in subjects without and with MCI (words of all response categories versus baseline). There was considerable activation in the inferior prefrontal, temporal, parietal and occipital cortex corresponding to a fifth to a quarter of the total brain volume. Details of the observed brain activation in this comparison and the following comparisons are illustrated in Fig. 1 and are displayed in Table 2. There was increased brain activity during verbal retrieval of hits in right superior and inferior frontal gyrus in subjects with MCI compared with healthy controls (Table 2, Fig. 1c; p < 0.001 at voxel level, at least 40 activated voxels). When correct rejections were compared between subjects with versus without MCI, there was increased activation in the left middle frontal gyrus (Table 2, Fig. 1d). This indicates that the compensatory activation in subjects with MCI depends on the response category. For both categories, activation was stronger in subjects with MCI than in controls (see Fig. 2a and b). We could not identify significantly increased brain activity during retrieval of misses and false alarms in subjects with MCI compared with healthy subjects, there was also no significantly increased activity during retrieval of hits, correct rejections, misses or false alarms in healthy controls compared to those with MCI (data not shown). 3.3.2. Activation differences by response categories We assume activation related to retrieval success is indicated by comparing signal intensities of hits versus misses; both types of words were learned, the latter were not remembered. In the two groups combined, activation by hits in comparison with misses has been found in the inferior parietal to occipital gyrus and precuneus to cingulate gyrus and precuneus (Table 2, Fig. 1e). Hits caused a higher BOLD response that misses (Fig. 2c). Equivalent activation was observed when hits are compared with other item categories (data not shown). Further analysis revealed that misses provide a stronger activation than the correct rejections in the precuneus to cuneus (Table 2, Figs. 1f and 2d). Other comparisons between different response categories as well as equivalent comparisons for the subgroups of subjects with or without MCI independently did not reveal significant activation foci (p > 0.001, >40 voxels).

4. Discussion 4.1. Differences between subjects with versus without MCI The majority of cerebral activation during verbal retrieval is similar in subjects with and without MCI. The activated areas correspond well with those of previous publications as

indicated in Section 1. It is conceivable that a task which needed the subjects’ full attention and consisted of various cognitive processes including concentration, visual perception, letter and word identification, semantic and episodic memory, and adequate response activated large parts of the brain in all subjects. There was some minor, but significant increase of activation in subjects with versus without MCI in the prefrontal cortex during verbal retrieval, but not vice versa. Such retrieval effort related activation in AD patients has been found in different brain areas depending on the memory tasks [3,1,16,19]. In support of such variability, we found that differences between subjects with versus without cognitive deficits depended on the relatively minor differences of cognitive processes as related to different response categories. The right superior and inferior frontal gyrus showed increased activation when signal intensities related to hits were compared between subjects with versus without MCI; the left middle frontal gyrus was activated when correct rejections were compared between the groups. Such response-related differences between subjects with versus without MCI have not yet been reported. The brain areas involved may in the conduction of the motor response, language perception, and memory, but may also reflect non-selective recruitment of frontal lobe areas as described by Logan et al. [33] during encoding. In agreement with our study, [44] described compensatory prefrontal activation during retrieval for elderly versus young subjects. Grady et al. [22] observed a compensatory prefrontal network in AD subjects. Increased prefrontal activation was observed in AD subjects during encoding as well as retrieval [43]. In contrast to Refs. [45,47] during encoding, we observed increased, but neither reduced nor delayed prefrontal activation in subjects with MCI versus healthy controls during retrieval. In partial agreement with our study [35] observed increased as well as reduced BOLD responses depending on the brain area during a picture recognition task. Consequently, there is likely to be much variation of the BOLD response depending on the task, the brain area and subjects’ conditions. This aspects needs a lot of further intensive investigations, the present study may present one of the starting points in this direction. We did not observe significant medio-temporal activation during retrieval. In contrast [14] observed that young subjects had increased medial temporal lobe activity, compared with elderly suffering from reduced memory performance, but not in comparison with those elderly exhibiting normal memory performance. Chetelat et al. [12] reported that reduction in retrieval performance correlated with hippocampal atrophy, but also with reduced resting metabolism in the posterior cingulate gyrus indicating a morphological and functional dissociation. One reason why we did not identify significant medial temporal lobe activity during retrieval in our study might be due to the age-related atrophy and successive dysfunction of the medio-temporal lobe in memory tasks and a related loss of statistical power [15]. The lack of power might

R. Heun et al. / Neurobiology of Aging 28 (2007) 404–413

411

Table 2 Significantly activated regions (peak voxels) for different categories, i.e. hits, misses, false alarms and correct rejections, and for the comparison between correct identifications and other item categories (cluster size >40 voxels) Comparison (corresponding figure)

Activated region

L/R

MNI coordinates of local maxima (mm) x

y

Z

Number of voxels in cluster

Z-value

Inferior to middle frontal gyrus Cingulate and medial frontal gyrus Pre- and post central gyrus Superior temporal gyrus Superior to inferior parietal lobe Inferior and middle occipital gyrus Cuneus, precuneus, and lingual gyrus Thalamus, brainstem, medial cerebellum

L+R L+R L+R L+R L+R L+R L+R L+R

−36 –a –a –a –a –a –a –a

21 –a –a –a –a –a –a –a

−3 –a –a –a –a –a –a –a

15010 –a –a –a –a –a –a –a

Infinity –a –a –a –a –a –a –a

Words in subjects with MCI (Fig. 1b)***

Medial frontal and cingulate gyrus Inferior to middle frontal gyrus Pre- and post central gyrus Cuneus, lingual gyrus Superior to inferior parietal lobe, precuneus Inferior and middle occipital gyrus Thalamus, brainstem, medial cerebellum

L+R L+R L+R L+R L+R L+R L+R

−6 –a –a −15 30 –a –a

9 –a –a −72 −57 –a –a

57 –a –a 9 48 –a –a

10114 –a –a 122 55 –a –a

Infinity –a –a 6.00 5.83 –a –a

Differences between MCI vs. healthy control subjects in activation by hits (Fig. 1c)*

Superior frontal gyrus

R

9

−9

75

48

4.10

Inferior frontal gyrus (Fig. 2a)

R

57

9

15

49

3.54

Differences between MCI vs. healthy control subjects in activation by correct rejections word (Figs. 1d and 2b)* Differences between hits and misses in all subjects (Fig. 1e)b,**

Middle frontal gyrus

L

−33

45

9

101

4.32

Precuneus to cingulate gyrus

R/L

0

−54

30

138

4.56

Inferior parietal to occipital gyrus (Fig. 2c)

L

−39

−78

36

67

4.42

Activation by misses vs. correct rejections (Figs. 1f and 2d)**

Preuneus to cuneus

R/L

−9

−66

18

Words in healthy subjects (Fig.

1a)***

45

3.90

(1)

Both sides were indicated if activated under the described condition, even though the local maxima usually refer to one side only. Voxel information not available since voxels are included in adjacent clusters. b Equivalent foci were also received in comparisons hits vs. false alarms, hits vs. correct rejections (data not given). * p < 0.001 at voxel level, p < 0.05 at cluster level. ** p < 0.05 at voxel level p < 0.001 at cluster level (both corrected for multiple comparisons). *** p < 0.001 at voxel and cluster level (both corrected for multiple comparisons). a

also be responsible for the fact that significant activation differences between subjects with versus without MCI were not observed for the even rarer response categories, i.e. misses and false alarms. 4.2. Differences in cerebral activation by response category Foci of increased activation by hits versus misses were the lateral parieto-occipital cortex and parts of the precuneus and posterior cingulate gyrus. These areas were also reported to be related to retrieval success in healthy subjects by other authors [32,53] and in our own studies in healthy controls subjects [25,26]. An increased BOLD response by remembered versus non-remembered items was also observed in the parietal cortex by Wheeler and Buckner [53]. However, we did not examine the difference between know and remember decisions as these might have been too complicated for the

present sample in our study even though the type of cognitive processes involved during retrieval (remembering, knowing and decision based on familiarity) may alter the type of local BOLD response [53]. The increased activation in the precuneus to cuneus observed in the comparison of misses versus correct rejections might be related to an increased dysfunctional retrieval effort [48]. As mentioned above the observed relationships between activation and memory performance in subjects with cognitive impairment seem complex: Grady et al. [22] found increased activation in the prefrontal cortex related to better memory in AD patients and controls, but activation in the amygdale correlated with memory performance in AD subjects only. Dickerson et al. [17] found medio-temporal activation during encoding to correlate with later retrieval success in subjects with MCI and healthy elderly, even though some of this activation might have been compensatory. Daselaar et al. [13] observed the medial temporal lobe to be

412

R. Heun et al. / Neurobiology of Aging 28 (2007) 404–413

involved in successful recognition of words. A co-occurrence of performance-related as well as compensatory activation was also seen in other cognitive tasks [29,30,49]. Thus, further studies are needed to investigate the extremely complex relationship of cognitive decline and actual memory performance in subjects with neurodegenerative disorders.

in subjects with and without MCI might be more appropriate than matching performance using tasks of variable difficulty and might provide a useful marker to identify subjects at risk for developing AD. To validate this assumption, prospective studies with subjects in different stages of cognitive decline are necessary.

4.3. Limitations and conclusions Even though most of the areas activated in the present study were consistent with those of other studies, it might be that some of the observed regional activation is more related to general attention processes than to memory related processes [11]. Nyberg et al. [37] reported that activation in the prefrontal cortex and the anterior cingulate cortex were found in different types of memory tasks, i.e. working memory, episodic memory and semantic memory. This functional unspecificity of cerebral regions is also supported by others [42]. The relationship between activation, performance, and cognitive decline is additionally complicated by the fact that MCI subjects and AD subjects do not activate the same brain areas for compensating functional deficits [35]. The lack of detecting previously published medial temporal lobe activation during retrieval might be the result of the limited number of individual response categories in several individuals. However, the usual option to increase the statistical power by excluding individual subjects on the basis of low numbers of specific response categories would have led to the exclusion of different subjects in different comparisons and thus would have severely hampered overall comparability of groups. Another reason for the limited group differences might be the considerable overlap of performance in subjects with and without cognitive impairment. We assume that memory performance in the elderly may be better described as a dimension, gradually declining from healthy subjects to subjects with mild cognitive impairment and beyond, than as a category. Consequently, a correlational approach between performance and activation might be more powerful to detect relationships between activation and performance. However, for the present analysis we chose the clinically and epidemiologically more relevant categorical approach by comparing subjects with and without MCI. Our approach was sensitive enough to detect some activation differences between diagnostic groups as well as between different response categories. In summary, we could reasonably demonstrate: (1) that the majority of retrieval-related activation is very similar between subjects with and without MCI as well as between different response categories, (2) but also, that some activation differences on the diagnostic distinction between elderly subjects with versus without MCI exist, and (3) that some other cerebral activation depends on the actual retrieval-response category independent of the subjects’ diagnostic classification. From a methodological perspective, we propose that the present response-related comparison of cerebral activation

Acknowledgements This research was supported by the German Research Council (DFG-HE 2318/5-1; GR 833/7). Mrs. Diana Syzui is greatly appreciated for her secretarial support.

References [1] Backman L, Andersson JL, Nyberg L, Winblad B, Nordberg A, Almkvist O. Brain regions associated with episodic retrieval in normal aging and Alzheimer’s disease. Neurology 1999;52:1861–70. [2] Baschek IL, Bredenkamp J, Oehrle B, Wippich W. Bestimmung der Bildhaftigkeit (I), Konkretheit (C) und der Bedeutungshaltigkeit (m ) von 800 Substantiven. Z Exp Angew Psycho 1977;25:353–96. [3] Becker JT, Mintun MA, Aleva K, Wiseman MB, Nichols T, DeKosky ST. Compensatory reallocation of brain resources supporting verbal episodic memory in Alzheimer’s disease. Neurology 1996;46:692–700. [4] Bennett DA, Wilson RS, Schneider JA, Evans DA, Beckett LA, Aggarwal NT, et al. Natural history of mild cognitive impairment in older persons. Neurology 2002;59:198–205. [5] Buckner RL, Koutstaal W, Schacter DL, Dale AM, Rotte M, Rosen BR. Functional-anatomic study of episodic retrieval. II. Selective averaging of event-related fMRI trials to test the retrieval success hypothesis. Neuroimage 1998;7:163–75. [6] Buckner RL, Koutstaal W, Schacter DL, Wagner AD, Rosen BR. Functional-anatomic study of episodic retrieval using fMRI. I. Retrieval effort versus retrieval success. Neuroimage 1998;7:151–62. [7] Busse A, Aurich C, Zaudig M, Riedel-Heller S, Matschinger H, Angermeyer MC. Alters- und bildungsspezifische Normierung des kognitiven Tests des SIDAM (Strukturiertes Interview zur Diagnose der Demenz vom Alzheimer-Typ, Multiinfarkt-Demenz und Demenzen ¨ anderer Atiologie nach DSM-IV und ICD-10). Z Geronto Geriatrie 2002;35:565–74. [8] Cabeza R. Hemispheric asymmetry reduction in older adults: the HAROLD model. Psychol Aging 2002;17:85–100. [9] Cabeza R, Anderson ND, Locantore JK, McIntosh AR. Aging gracefully: compensatory brain activity in high-performing older adults. Neuroimage 2002;17:1394–402. [10] Cabeza R, Daselaar SM, Dolcos F, Prince SE, Budde M, Nyberg L. Task-independent and task-specific age effects on brain activity during working memory, visual attention and episodic retrieval. Cereb Cortex 2004;14:364–75. [11] Cabeza R, Dolcos F, Prince SE, Rice HJ, Weissman DH, Nyberg L. Attention-related activity during episodic memory retrieval: a crossfunction fMRI study. Neuropsychology 2003;41:390–9. [12] Chetelat G, Desgranges B, de la Sayette V, Viader F, Berkouk K, Landeau B, et al. Dissociating atrophy and hypometabolism impact on episodic memory in mild cognitive impairment. Brain 2003;126:1955–67. [13] Daselaar SM, Rombouts SA, Veltman DJ, Raaijmakers JG, Lazeron RH, Jonker C. Parahippocampal activation during successful recognition of words: a self-paced event-related fMRI study. Neuroimage 2001;13:1113–20.

R. Heun et al. / Neurobiology of Aging 28 (2007) 404–413 [14] Daselaar SM, Veltman DJ, Rombouts SA, Raaijmakers JG, Jonker C. Neuroanatomical correlates of episodic encoding and retrieval in young and elderly subjects. Brain 2003;126:43–56. [15] Daselaar SM, Veltman DJ, Rombouts SA, Raaijmakers JG, Jonker C. Deep processing activates the medial temporal lobe in young but not in old adults. Neurobiol Aging 2003;24:1005–11. [16] Desgranges B, Baron JC, Lalevee C, Giffard B, Viader F, de La Sayette V, et al. The neural substrates of episodic memory impairment in Alzheimer’s disease as revealed by FDG-PET: relationship to degree of deterioration. Brain 2002;125:1116–24. [17] Dickerson BC, Salat DH, Bates JF, Atiya M, Killiany RJ, Greve DN, et al. Medial temporal lobe function and structure in mild cognitive impairment. Ann Neurol 2004;56:27–35. [18] Dickerson BC, Salat DH, Greve DN, Chua EF, Rand-Giovannetti E, Rentz DM, et al. Increased hippocampal activation in mild cognitive impairment compared to normal aging and AD. Neurology 2005;65(3):404–11. [19] Elgh E, Larsson A, Eriksson S, Nyberg L. Altered prefrontal brain activity in persons at risk for Alzheimer’s disease: an fMRI study. Int Psychogeriatr 2003;15:121–33. [20] Folstein MF, Folstein SE, McHugh PR. “Mini-Mental state”. A practical method for grading the cognitive state of patients for the clinician. J Psychiat Res 1975;12:189–98. [21] Good CD, Johnsrude I, Ashburner J, Henson RNA, Friston KJ, Frackowiak RSJ. Cerebral asymmetry and the effects of sex and handedness on brain structure: a voxel-based morphometric analysis of 465 normal adult human brains. Neuroimage 2001;14:685–700. [22] Grady CL, Maura LF, Pietro P, Barry H, Stanley IR. Altered brain functional connectivity and impaired short-term in Alzeihmer’s disease. Brain 2001;124:739–56. [23] Green DM, Swets JA. Signal detection theory and psychophysics. New York: Wiley; 1966. [24] Helmstaedter C, Lendt M, Lux S. VMLT-Verbaler Lern- und Merkf¨ahigkeitstest. Beltz: Weinheim; 2001. [25] Heun R, Freymann N, Granath DO, Stracke CP, Jessen F, Barkow K, et al. Differences of cerebral activation between superior and inferior learners during motor sequence encoding and retrieval. Psychiat Res 2004;132(1):19–32. [26] Heun R, Jessen F, Klose U, Erb M, Granath DO, Grodd W. Responserelated fMRI of veridical and false recognition of words. Eur Psychiat 2004;19:42–52. [27] Heun R, K¨olsch H, Jessen F. Risk factors and early signs of Alzheimer’s disease in a family study sample. Eur Arch Psychiatr Clin Neurosci 2006;256:28–36. [28] Heun R, Jessen F, Klose U, Erb M, Granath D-O, Grodd W. Responserelated fMRI analysis during encoding and retrieval revealed differences in cerebral activation by retrieval success. Psychiat Res, Neuroimaging 2000;99:137–50. [29] Johnson SC, Saykin AJ, Baxter LC, Flashman LA, Santulli RB, McAllister TW, et al. The relationship between fMRI activation and cerebral atrophy: comparison of normal aging and Alzheimer disease. Neuroimage 2000;11:179–87. [30] Kato T, Knopman D, Liu H. Dissociation of regional activation in mild AD during visual encoding: a functional MRI study. Neurology 2001;57:812–6 [59]. [31] Konishi S, Wheeler ME, Donaldson DI, Buckner RL. Neural correlates of episodic retrieval success. Neuroimage 2000;12:276–86. [32] Kopelman MD, Stevens TG, Foli S, Grasby P. PET activation of the medial temporal lobe in learning. Brain 1998;121:875–87. [33] Logan JM, Sanders AL, Snyder AZ, Morris JC, Buckner RL. Underrecruitment and nonselective recruitment: dissociable neural mechanisms associated with aging. Neuron 2002;33(5):827–40. [34] Machulda MM, Ward HA, Borowski B, Gunter JL, Cha RH, O’Brien PC, et al. Comparison of memory fMRI response among normal, MCI, and Alzheimer’s patients. Neurology 2003;61(4):500–6.

413

[35] Mandzia J, Black S, Grady C, McAndrews MP, Graham S. Encoding and retrieval in aging and memory loss, a fMRI study. Brain Cogn 2002;49:225–8. [36] Miezin FM, Maccotta L, Ollinger JM, Petersen SE, Buckner RL. Characterizing the hemodynamic response: effects of presentation rate, sampling procedure, and the possibility of ordering brain activity based on relative timing. Neuroimage 2000;11(6 Pt. 1): 735–59. [37] Nyberg L, Marklund P, Persson J, Cabeza R, Forkstam C, Petersson KM, et al. Common prefrontal activations during working memory, episodic memory, and semantic memory. Neuropsychology 2003;41:371–7. [38] Nyberg L, McIntosh AR, Houle S, Nilsson LG, Tulving E. Activation of medial temporal structures during episodic memory retrieval. Nature 1996;380:715–7. [39] Oldfield RC. The assessment and analysis of handedness: the Edinburgh inventory. Neuropsychology 1971;9:97–113. [40] Oldigs-Kerber J, Adamus WS, Kitzinger M. Zur Beeinflussung von verbalen Lern- und Ged¨achtnisprozessen durch Anticholinergica am Beispiel von Scopolamin. Ein pharmakopsychologischer Beitrag f¨ur die neuropsychologische Praxis. Z NeuroPsychol 1991;2:29–40. [41] Petersen RC, Doody R, Kurz A, Mohs RC, Morris JC, Rabins PV, et al. Current concepts of mild cognitive impairment. Arch Neurol 2001;58:1985–92. [42] Ranganath C, Johnson MK, D’Esposito M. Prefrontal activity associated with working memory and episodic long-term memory. NeuroPsychology 2003;41:378–89. [43] Remy F, Mirrashed F, Campbell B, Richter W. Verbal episodic memory impairment in Alzheimer’s disease: a combined structural and functional MRI study. Neuroimage 2005;25:253–66. [44] Reuter-Lorenz P. New visions of the aging mind and brain. Trends Cogn Sci 2002;6:394–400. [45] Rombouts SA, Barkhof F, Goekoop R, Stam CJ, Scheltens P. Altered resting state networks in mild cognitive impairment and mild Alzheimer’s disease: an fMRI study. Hum Brain Mapp 2005; 26(4):231–9. [46] Rombouts SA, Barkhof F, Veltman DJ, Machielsen WC, Witter MP, Bierlaagh MA, et al. Functional MR imaging in Alzheimer’s disease during memory encoding. Am J Neuroradiol 2000;21(10): 1869–75. [47] Rombouts SA, Goekoop R, Stam CJ, Barkhof F, Scheltens P. Delayed rather than decreased BOLD response as a marker for early Alzheimer’s disease. Neuroimage 2005;26(4):1078–85. [48] Rugg MD, Fletcher PC, Frith CD, Frackowiak RS, Dolan RJ. Differential activation of the prefrontal cortex in successful and unsuccessful memory retrieval. Brain 1996;119:2073–83. [49] Saykin AJ, Flashman LA, Frutiger SA, Johnson SC, Mamourian AC, Moritz CH, et al. Neuroanatomic substrates of semantic memory impairment in Alzheimer’s disease: patterns of functional MRI activation. J Int Neuropsychol Soc 1999;5:377–92. [50] Saykin AJ, Wishart HA. Mild cognitive impairment: conceptual issues and structural and functional brain correlates. Semin Clin Neuropsychiat 2003;8:12–30. [51] Schacter DL, Buckner RL, Koutstaal W, Dale AM, Rosen BR. Late onset of anterior prefrontal activity during true and false recognition: an event-related fMRI study. Neuroimage 1997;6:259–69. [52] Tulving E, Habib R, Nyberg L, Lepage M, McIntosh AR. Positron emission tomography correlations in and beyond medial temporal lobes. Hippocampus 1999;9:71–82. [53] Wheeler ME, Buckner RL. Functional-anatomic correlates of remembering and knowing. Neuroimage 2004;21(4):1337–49. [54] World Health Organization. Tenth Revision of the International Classification of Diseases. Chapter V (F): Mental and Behavioural Disorders (including disorders of psychological development). Clinical Descriptions and Diagnostic Guidelines. WHO; 1991.