Correlation Between Hippocampal Volume and Autobiographical Memory Depending on Retrieval Frequency in Healthy Individuals and Patients with Alzheimer’s Disease

Abstract

The hippocampus plays an indispensable role in episodic memory, particularly during the consolidation process. However, its precise role in retrieval of episodic memory is still ambiguous. In this study, we investigated the correlation of hippocampal morphometry and the performance in an autobiographical memory task in 27 healthy controls and 24 patients suffering from Alzheimer’s disease (AD). Most importantly, correlations were defined separately and comparatively for memory contents with different retrieval frequency in the past. In healthy subjects, memory performance for seldom retrieved autobiographical events was significantly associated with gray matter density in the bilateral hippocampus, whereas this correlation was not present for events with high retrieval frequency. This pattern of findings confirms that retrieval frequency plays a critical role in the consolidation of episodic autobiographical memories, thereby making them more independent of the hippocampal system. In AD patients, on the other hand, successful memory retrieval appeared to be related to hippocampal morphometry irrespective of the contents’ retrieval frequency, comprising events with high retrieval frequency, too. The observed differences between patients and control subjects suggest that AD-related neurodegeneration not only impairs the function, but also decreases the functional specialization of the hippocampal memory system, which, thus, may be considered as marker for AD.

Keywords

Alzheimer’s disease autobiographical memory cognitive impairment episodic memory gray matter hippocampus historic events magnetic resonance imaging medial temporal lobe voxel-based morphometry

INTRODUCTION

Alzheimer’s disease (AD) is a progressive, neurodegenerative disease of the brain and is the most common cause of dementia [1]. Thereby, memory disturbances are cardinal symptoms resulting from early neurodegenerative processes in the medial temporal lobe (MTL) structures [2]. According to consolidation theories, structures within the MTL, particularly the hippocampus, but also the adjacent parahippocampal, perirhinal, and entorhinal cortices, are crucial for the acquisition of new memories [3]. However, the specific role of the hippocampus within the MTL structures as well as the time course of its involvement in the storage and retrieval of declarative knowledge (i.e., episodic and semantic memory) is ambiguous [4 –6]. AD patients typically develop a retrograde amnesia, which is the inability to retrieve memories acquired before the onset of the disorder [7]. It is characterized by a loss of autobiographical memories [8 –10] and public knowledge [9 , 12]. Autobiographical memory has been found to be impaired in the earliest stages of AD [13 –15], also present in individuals suffering from amnestic mild cognitive impairment (aMCI) and relates to the ability to retrieve personally experienced events from the past and includes general knowledge about the self [16]. Empirical evidence suggests that in retrograde amnesia remote memories are typically better preserved than recent memories, indicating a temporal gradient (TG) of memory decline. A temporally graded retrograde amnesia after damage to the hippocampal complex suggests that remote memories are not dependent on this structure anymore, indicating an important role of the hippocampus for the consolidation process but not for the long-term storage of memory. However, reports of a TG in retrograde amnesia have been inconclusive. Some investigators found higher preservation of remote memories compared to more recently acquired memories in patients suffering from AD [17 –21], whereas others did not [9 , 23].

In a previous study [12], our group examined the recall of memories for past public events in AD patients and healthy controls (HC) as a function of their retrieval frequency and age, by the newly developed Historic Events Test (HET), which assesses knowledge about famous events from the past 60 years. We found impaired memory for past public events in AD patients relative to HC and additionally a strong effect of retrieval frequency across all time segments, which, most importantly, occurred irrespective of the age of the memory. From this finding, we inferred that more frequently retrieved memories become independent of the hippocampus, whereas less frequently recalled knowledge remains dependent on an intact hippocampus and might therefore be more susceptible to early hippocampal damage due to AD. In another study [17], we found impaired autobiographical memory in aMCI and AD patients with better preservation of remote compared to recent memories. As remote memories were recalled more often than recent ones, we concluded that the temporal gradient of memory decline is mediated by retrieval frequency. Consequently, this indicates that more frequently recalled memories become gradually independent of the hippocampal complex, probably integrated into neocortical memory structures and are therefore less affected by early neuropathological changes in the MTL.

In the present study, we used voxel-based morphometry (VBM) and a region of interest (ROI) analysis to investigate the relationship between hippocampal morphometry and memory performance for autobiographical events subdivided according to their retrieval frequency in 24 patients suffering from early dementia of Alzheimer type (eDAT) or aMCI and 27 HC. Based on the above reported results and related interpretations, we hypothesized that autobiographical memory performance correlates with hippocampal volume specifically for seldom retrieved events, whereas no such correlation was expected for frequently retrieved ones.

METHODS

Participants

Our study included 51 participants (23 males and 28 females) with a mean age of 72±5.5 years. No participant had a physical handicap that affected his or her ability to perform the tasks or any indication of other neurological or psychiatric disorders unrelated to his or her diagnosis. To exclude acute symptoms of depression, all participants completed the German 15-item version of the Geriatric Depression Scale (GDS) [24]. The local ethical committee at the University Hospital of Tübingen approved the study. All participants signed an informed consent form after receiving a detailed description of the study. The study population contained HC, and patients with aMCI and eDAT. For the statistical analysis, we pooled aMCI and eDAT patients together into an AD cluster.

Healthy control group

HC individuals (n = 27) had no history of neurological or psychiatric disease or any sign of cognitive decline, which was confirmed by a clinical interview. The MRI of each HC subject was evaluated by trained radiologists to confirm that no signs of AD (mesiotemporal atrophy) were present.

Patients with aMCI or eDAT

Patients with aMCI or eDAT (n = 24) were recruited from the Memory Clinic of the Department of Psychiatry and Psychotherapy at the University Hospital of Tübingen. They underwent physical, neurological, neuropsychological, and psychiatric examinations, as well as brain imaging.

The diagnosis of aMCI was defined by Mayo criteria [25, 26], which include the presence of a memory complaint (corroborated by an informant), objectively impaired memory function, intact activities of daily living, and the absence of dementia. All patients in this group met the MCI-core clinical criteria according to the National Institute on Aging-Alzheimer’s Association (NIA-AA) clinical research criteria for MCI due to AD [27]. Although NIA-AA criteria do not explicitly exclude deficits in cognitive domains other than memory, we focused on memory impairment, as pre-dementia AD is most frequently characterized as an amnestic syndrome [1].

Patients with eDAT met the diagnostic criteria of probable Alzheimer’s dementia according to the National Institute of Neurological and Communicative Disorders and Stroke Alzheimer’s Disease and Related Disorders Association [25, 28]. All of these patients had a score of four on the Global Deterioration Scale [29].

Neuropsychological assessment

All participants completed the Mini-Mental State examination (MMSE) [30] and underwent the HET. The HET was developed to assess the ability to recall famous events of the past 60 years and the autobiographical, contextual information surrounding the events. The test was performed according to a previous study of our group [12], in which we used an updated and extended version of the HET with additional events occurring between 2006 and 2010. The original HET is provided in another paper of our group [9]. The 56-year period was subdivided into four time segments: A (1954–1974), B (1975–1999), C (2000–2009), and D (2010), each consisting of six famous events. The 24 historical events were presented to participants in a random order and for each event, they were asked to perform three tasks: a subjective memory rating task, a dating accuracy task, and a contextual memory task. In this study, we focused on the contextual memory task, which evaluates episodic autobiographical memories.

For the contextual memory task, participants were required to recall their personal circumstances surrounding a historic event, i.e., they had to extensively describe the situation in which they found themselves when they first heard about the incident. The amount of detail provided and the specificity in space and time was rated. Three points were given for a fully comprehensive answer (i.e., answers were spontaneous and reflected a memory that was specific for the time and space of the situation in which participants heard about or experienced the event). Two points were given for a personal but non-specific memory or a specific memory for which time and place were not recalled. For a vague personal memory, participants received one point. Zero points were given if participants did not remember the event, provided an incorrect answer, made general, non-specific statements, or provided a response based on a semantic memory. Informants, such as spouses or life-long companions, verified the accuracy of the responses.

Retrieval frequency was operationalized by a paired comparison analysis. For each of the four time segments, participants matched each event pairwise with each of the other events and indicated which of the two events they remembered more frequently during their lifetime. Fifteen paired comparisons were performed for each time segment. Retrieval frequency of an event was categorized as “high” when it was preferred four to five times in the head-to-head record (i.e., suggesting that the event was remembered more frequently than four or five other events during the participant’s lifetime), categorized as “medium”, when it was preferred two to three times in the head-to-head record, and categorized as “low” when it was preferred zero to one time in the head-to-head record. By doing so, for each participant, there were two events with a high, two events with a medium, and two events with a low retrieval frequency for each of the four time segments. Subsequently, for each subject, mean values from these two scores per time segment were calculated whereby there remained one memory performance score with low, one with medium, and one with high retrieval frequency per person and time segment. Eventually, these scores were averaged across time segments, finally leading to three values for each participant: one memory performance score for seldom, one for medium often, and one for frequently retrieved events. To examine our specific hypothesis, in this analysis we focused on the polar values, meaning, we only used events with low and high retrieval frequency for regression analysis.

Statistical analysis of sociodemographic and behavioral data

Sociodemographic and neuropsychological data were analyzed with the statistical software package SPSS 24.0 (IBM Corp., Somers, NY, USA). For all tests, the level of statistical significance was set to p < 0.05. Levene’s test was used to assess homogeneity of variance. Differences in age, years of education, and MMSE were assessed using independent t-tests. The Pearson chi-square-test was applied to detect group differences in gender distribution and the nonparametric Mann-Whitney U test was used to detect group differences in GDS scores. Total intracranial volume (TIV) was compared between HC and AD subjects applying an independent t-test.

Multivariate (two-factorial) repeated measures analysis of variance (MANOVA) with retrieval frequency (low, medium, high) as within-subject factor and group (HC, AD) as between-subject factor was calculated to examine the main effects of group and retrieval frequency on the contextual memory and dating accuracy task scores. Significant effects of the MANOVA were specified by post-hoc tests, comprising of one-way ANOVA and subsequent paired t-tests, using Bonferroni adjustments to control for alpha error accumulation. Mauchly’s test was used to evaluate violations of the sphericity assumption.

Differences in the three retrieval frequency categories between groups were assessed using one-way ANOVA. To test the effect of retrieval frequency on memory performance paired sample t-tests were applied within each group (retrieval frequency low versus medium, low versus high, medium versus high). For each group, we calculated the correlations between the performance scores for low, medium, and high retrieval frequency. Moreover, we present descriptive statistics of sociodemographic and behavioral data separately and comparatively for the two patient sub-groups (aMCI and eDAT) and also provide related t statics to evaluate the corresponding group differences (see Supplementary Table 1).

Image acquisition

MRI data were acquired on a 3.0 Tesla-Scanner (Siemens, Trio, Erlangen, Germany). A high resolution structural T1-weighted image was obtained using a magnetization prepared rapid acquisition gradient echo (MPRAGE) sequence with a voxel size of 1×1×1 mm³.

Voxel-based morphometry processing

VBM [31] analyses were performed with the CAT12 toolbox (Gaser & Dahnke, 2016; http://www.neuro.uni-jena.de/hbm2016/GaserHBM2016.pdf) implemented in the statistical analysis software Statistical Parametric Mapping version 12 (SPM12; http://www.fil.ion.ucl.ac.uk/spm) and MATLAB (R2016b, The MathWorks, Natick, MA, USA). Initially, pre-processing steps were applied according to the CAT12 manual (http://dbm.neuro.uni-jena.de/cat12/CAT12-Manual.pdf), using default settings. T1 images were first spatially normalized to the Montreal Neurological Institute (MNI) space template using the DARTEL algorithm [32]. Then, the whole brain structural data was segmented into gray matter (GM), white matter (WM), and cerebrospinal fluid (CSF). Finally, the normalized GM images were smoothed using a Gaussian kernel of 8 mm. After pre-processing, all scans passed through an automated quality check protocol.

Voxel-based morphometry analysis

For statistical analysis, we applied a parametric general linear model (GLM) approach implemented in SPM12. To avoid possible edge effects around the border between GM and WM, and to exclusively include voxels with sufficient GM proportion, only voxels with values above an absolute GM threshold of 0.1 were analyzed (absolute threshold masking). TIV was included as nuisance variable in the GLM in order to correct for different brain sizes among subjects and remove the related variance. To identify significant clusters and eliminate spurious small clusters, the minimum cluster size (k) was empirically determined for each contrast separately. We used the number of expected voxels per cluster as cluster extent threshold. This is necessary to avoid arbitrary threshold setting.

We performed voxel-wise independent two sample t-tests to compare GM tissue volume between HC subjects and AD patients. Statistical inferences were made at single voxel level and thresholded at p < 0.05 with whole brain family-wise error (FWE) correction.

Region of interest analysis

In addition, a region of interest (ROI) analysis focusing on the bilateral hippocampus was conducted. The corresponding target regions were defined using the anatomical automatic labeling (AAL) atlas [33] and mean GM values of the included voxels were calculated. Statistical analyses on these values were run in the R environment (R version 3.5.0) and defined the correlational relationship between the hippocampal morphometry and the quality of the mnemonic recall (centered values) in dependence of the experimental group. More specifically, we defined two general linear regression models including the bilateral hippocampal morphometry (i.e., average GM values) as dependent variable. Independent variables were group and the performance scores for seldom retrieved events in the first model—or rather the performance scores for often retrieved events in the second model—as well as the interaction between group and the respective performance score. Subsequently, for each group we calculated the Pearson correlation between memory performance and hippocampal morphometry, separately for seldom and often retrieved memory events. For a pairwise comparison of the corresponding correlation coefficients, we used the Pearson and Filon’s Z test [34] for dependent groups, as implemented in the R package cocor [35]. Tests were thresholded at p < 0.05, one-sided (based on a directional hypothesis).

RESULTS

Sociodemographic data

Sociodemographic characteristics and mean composite scores for both groups are presented in Table 1. AD and HC groups did not significantly differ in terms of age, years of education, and gender distribution. Also, GDS scores were not significantly different between groups and well below the cut-off score of six, indicating that participants did not exhibit signs for depression [24]. As expected, AD subjects had significantly lower MMSE scores compared to HC (t[49] = 7.691, p < 0.001), indicating greater impairment in global cognition. TIV did not significantly differ between patients and healthy subjects (t[49] = 3.53, p = 0.725), indicating equivalent total brain volume between.

Table 1

Descriptive statistics of sociodemographic and clinical variables

	Group
	HC	AD	p
N	27	24
Gender (f/m)	16/11	12/12	0.51
Age (y)	71.5 (6.6)	72.4 (3.9)	0.55
GDS	2.2 (1.8)	2.7 (1.7)	0.24
MMSE	29.0 (0.9)	24.8 (2.8)	<0.00
Education (years)	13.3 (3.8)	12.5 (2.7)	0.39
TIV (mm³)	1541.6 (160.4)	1526.5 (140.8)	0.73
Low retrieval frequency	1.0 (0.3)	0.6 (0.5)	<0.00
Medium retrieval frequency	1.4 (0.42)	0.9 (1.4)	<0.00
High retrieval frequency	1.7 (0.4)	1.1 (0.5)	<0.00

Values are expressed as mean (standard deviation). HC, healthy control subjects; AD, patients with Alzheimer’s disease; N, number (sample size); f/m, female/male; GDS, Geriatric Depression Scale (higher score indicates more severe depressive symptoms; maximum 15; a cut-off of 6 indicates mild depression); MMSE, Mini-Mental State Examination; TIV, total intracranial volume.

Episodic autobiographical memory performance

We found a significant main effect of group on the autobiographical memory task scores (F = 24.986; p < 0.001). More specifically, a one-way ANOVA indicated that patients performed worse in the contextual memory task than HC to a significant extent in and across all frequency conditions (low retrieval frequency: F[1,49] = 12.820; p = 0.001; medium retrieval frequency: F[1,49] = 15.557; p < 0.001; high retrieval frequency: F[1,49] = 23.636; p < 0.001). We also found a significant main effect of retrieval frequency on the autobiographical memory task scores (F[2,98] = 49.7; p < 0.001). Both HC and patient subjects reported significantly less detailed memories when retrieval frequency was low compared to medium (p < 0.001/p = 0.002) or high (p < 0.001/p<0.001) and when retrieval frequency was medium compared to high (p = 0.005/p = 0.002). Correlations between the performance scores in the low, medium, and high retrieval frequency conditions are depicted in Table 2 and Table 3, separately for each group.

Table 2

Descriptive statistics for performance of healthy control subjects in the autobiographical memory test: means, standard deviations, and correlations with confidence intervals

Variable	M	SD	1	2
1. High retrieval frequency	1.68	0.39
2. Medium retrieval frequency	1.43	0.42	0.43^* [0.06, 0.70]
3. Low retrieval frequency	1.00	0.31	0.22 [–0.17, 0.56]	0.42^* [0.05, 0.69]

M and SD are used to represent mean and standard deviation, respectively. Values in square brackets indicate the 95% confidence interval for each correlation. ^*p < 0.05; ^**p < 0.01.

Table 3

Descriptive statistics for performance of Alzheimer’s disease patients in the autobiographical memory test: means, standard deviations, and correlations with confidence intervals

Variable	M	SD	1	2
1. High retrieval frequency	1.09	0.48
2. Medium retrieval frequency	0.92	0.49	0.73^** [0.47, 0.88]
3. Low retrieval frequency	0.61	0.47	0.60^** [0.26, 0.81]	0.59^** [0.25, 0.80]

M and SD are used to represent mean and standard deviation, respectively. Values in square brackets indicate the 95% confidence interval for each correlation. ^*p < 0.05; ^**p < 0.01.

Voxel-based morphometry analysis

Whole-brain analyses

Comparison of HC and AD patients showed several clusters of significantly reduced GM volume. These regions include the bilateral amygdala, the bilateral entorhinal cortex, the bilateral hippocampus, and the right temporal pole. The reported clusters (MNI coordinates and related statistics) are listed in Table 4 and graphically depicted in Fig. 1. HC subjects (compared to patients), on the other hand, exhibited no regions of reduced cortical volume.

Table 4

Regions of gray matter volume loss in the patients with Alzheimer’s disease showed by voxel-based morphometry group comparison analysis

	MNI coordinates
Structure	Hemisphere	x	y	z	k	t-value
Amygdala	left	–22	–3	–14	1907	7.38^**
Entorhinal area	left	–33	2	–21		7.11^*
Hippocampus	left	–20	–18	–22		6.07^*
Amygdala	right	18	–3	–14	1343	6.71^**
Entorhinal area	right	30	4	–21		6.61^*
Hippocampus	right	20	–16	22		6.10^*
Temporal pole	right	39	8	–30	32	5.79^**

MNI, Montreal Neurological Institute; k, cluster size; cluster size in voxels; voxel size = 1×1×1 mm; cluster size threshold of 22 contiguous voxels. Thresholded at p_FWE < 0.05. ^*survives FWE correction at cluster-level, ^**survives FWE correction at peak-level.

Fig.1

Results of group comparison of gray matter volume between patients with Alzheimer’s disease and healthy controls are presented on coronal brain slices (MNI template). Statistical parametric maps were thresholded at p_FWE < 0.05 at peak level with a minimum cluster extent of 22 contiguous voxels.

Region of interest analyses

Prediction of hippocampal morphometry by memory performance and group. First, we defined a multiple regression model (AIC: – 100.70) to explain variance in hippocampal cortical volume by (a) memory performance for seldom recalled events, (b) group (HC versus patient), and (c) their interaction. The estimated regression equation was significant at p < 0.000 (F = 22.43), with an R² of 0.589 and an adjusted R² of 0.563. Thereby, both the regression coefficients for memory performance and the one for group were significant at p < 0.001, while their interaction exhibited no significant effect (regression equation/intercept and coefficients: Y = 0.806 + 0.058 + 0.052 – 0.019; see Fig. 2A).

Fig.2

Illustrations of the association (linear regression) between hippocampal morphometry and recall quality (A) in seldom recalled events and (B) in often recalled events, separated by group. Blue color of data points and corresponding trend lines represent the healthy control group, red color of data points and corresponding trend lines represent the AD patients.

In a second analogous regression model (AIC: – 91.160), hippocampal cortical volume was predicted by (a) the memory performance score for often recalled events, (b) group affiliation and (c) their interaction. Again, the estimated regression equation was found to be significant (F[3,47] = 15.93, p < 0.000), with an R² of 0.504 and an adjusted R² of 0.473, while all three regression coefficients reached the level of statistical significance (memory performance: p < 0.05; group: p < 0.001; interaction term: p < 0.05; regression equation / intercept and coefficients: Y = 0.819 + 0.035 + 0.057 – 0.039; see Fig. 2B).

Relationship between hippocampal morphometry and memory performance, depending on retrieval frequency. The correlational relationship between hippocampal morphometry and memory performance was expected to be different for seldom and frequently recalled memory contents. This expectation was confirmed by the data of the HC subjects, who exhibited a significantly decreased correlation for frequently recalled events (difference: z[27] = 1.666, p < 0.05; seldom recalled events: r[27] = 0.340, n.s.; often recalled events: r[27] = –0.036, n.s.; see Fig. 3A). AD patients, on the other hand, exhibited no differential, i.e., equivalently high correlations for the two frequency conditions (difference: z[26] = 0.877, n.s.; seldom recalled events: r[24] = 0.670, p < 0.001; often recalled events: r[24] = 0.549, p < 0.05; see Fig. 3B).

Fig.3

Illustration of the “recall quality gradient”, i.e., the association (linear regression) between recall quality and hippocampal morphometry in (A) the healthy control group and (B) AD/aMCI patients. Red color of data points and corresponding trend lines represent the association of the hippocampus with recall quality in seldom recalled events, blue color of data points and corresponding trend lines represent the association of the hippocampus with recall quality in often recalled events.

DISCUSSION

In this combined VBM-based and neuropsychological study, we examined whether episodic autobiographical memory performance is associated with hippocampal morphometry (i.e., GM volume) and whether this relationship is modulated by the retrieval frequency of the corresponding stored memory episodes. This research question was addressed comparatively in two study groups: (a) a patient group of subjects diagnosed with aMCI and early AD and (b) a control group of healthy elderly subjects. Patients compared to control subjects showed several regions of significantly reduced GM volume, which lie in the MTL involving the bilateral amygdala, the bilateral entorhinal cortex, the bilateral hippocampus, and the right temporal pole. These are basic (but not trivial) findings that essentially reflect the AD-typical neuropathology (i.e., neurodegeneration) and which therefore are highly consistent with the prior MRI literature reporting volume reductions in temporal lobe structures in both patients with AD and patients with aMCI [2, 36]. Moreover, we found impaired autobiographical memory performance in patients across all retrieval frequency conditions. Again, this is a quite basic finding which is well in line with results from our previous studies [12 , 17] and also empirical evidence from others, which emphasizes that autobiographical memory loss occurs early in the course of the disease and is already present in patients with aMCI [15 , 38]. Such neuropsychological deficits are commonly explained by neurodegeneration or other neuropathological changes within mesiotemporal structures, brain regions that are well known to be crucial for memory formation [8]. Beyond between-group differences, there was a significant effect of retrieval frequency on memory performance across groups. More specifically, both patients and control subjects exhibited a better memory performance (in terms of a richer, more detailed recall of contextual episodic information) for events with a higher retrieval frequency, which moreover occurred across all investigated time segments. This finding is in line with previous studies postulating a preserving effect of retrieval on episodic memories in AD patients [12, 39].

The hippocampal memory system is known to underlie age-related declines occurring even in absence of neurodegenerative disorders, so that also “healthy aging” may be accompanied by marked impairments in autobiographical memory performance [40]. Accordingly, several studies found age-related reductions in the contextual richness of autobiographical memory reports which have been construed to emanate from impaired strategic retrieval processes and poor recruitment of the hippocampus and ventrolateral prefrontal cortex [40]. In line with this reasoning, there are consistent reports of age-related decreases in hippocampal activity during episodic memory performance in non-demented subjects [41 –44]. Such neurofunctional declines in the hippocampal system during normal aging can also explain why in the present work not only patients, but also healthy non-demented subjects exhibited better performance for memory episodes with high retrieval frequency.

Based on the outlined findings, the present study focused on the differential relationship between hippocampal morphometry and episodic memory performance in dependence on the retrieval frequency of the corresponding mnemonic contents (i.e., autobiographical episodes). Thereby, based on theoretical assumptions and prior empirical findings [17 , 45], we expected a strong association between hippocampal GM volume and memory performance for seldom retrieved but no such (or a significantly reduced) relationship between hippocampal morphometry and the memory for frequently retrieved events. The analysis confirmed this expectation, however only for the healthy control subjects, whereas the patient subjects exhibited a significant correlation between hippocampal morphometry and memory performance irrespective of retrieval frequency, i.e., across all frequency conditions. Our finding of a differential relationship of seldom retrieved and frequently retrieved memory episodes with hippocampal morphometry in healthy subjects essentially corroborates that the retrieval of autobiographical memory episodes plays an important role in memory consolidation making memory traces stronger and more stable and, at the same time, becoming gradually independent of the hippocampal system. This consolidation process presumably protects (at least in part) the respective memory entries from decomposition related to neurodegeneration particularly in mesiotemporal regions [2]. Moreover, a differential association of hippocampal morphometry with seldom and frequently retrieved autobiographical memory contents may be considered in the context of the “semanticization” theory. This theory assumes that memory entries undergo a gradual transition over time from episodic to semantic memory by a process of abstraction, coined semanticization [46, 47]. Accordingly, this theory assumes an episodic-semantic continuum (rather than a categorical distinction) of memory entries, which also provides a conceptual explanation for Ribot’s temporal gradient according to which more recent memories are relatively more susceptible to loss than more remote memories in retrograde amnesia [48]. Most importantly, neuroimaging studies have shown that the semanticization of autobiographical memories is accompanied by a gradual disengagement of hippocampal activity related to memory retrieval [49]. The disengagement of the hippocampus for frequently retrieved memory events observed in the present work may be based, at least in part, on the same neurofunctional transition process. In this sense, the repeated retrieval of autobiographical events may be assumed to lead to a cumulative abstraction of the respective entries in memory, making them more factual and disconnected from their original temporal context in terms of less naturalistic or lively.

The increased relation of hippocampal morphometry to the memory of frequently retrieved episodes in the patient group arguably reflects that AD-related neurodegeneration not only impairs neuronal functioning, but also diffuses the neurofunctional specialization of involved brain regions, arguably based on disturbed neuroanatomical connections [50, 51]. In this context, a reduced neurofunctional specialization may be construed as part of the degenerating brain’s efforts to compensate for the increasing functional deficits by using other or additional available neural resources. Corroborating this assumption, exclusively in the performance data of the patient group we found a significant correlation between the memory performances for seldom retrieved events and frequently retrieved events, suggesting that they involve a broader common neurofunctional basis than in non-demented subjects. Based on this reasoning, one may speculate that the ratio of the memory performances for often and seldom recalled events could provide a useful neuropsychological marker for AD pathology.

Possible limitations of the presented work have to be mentioned. First, given the methodological approach of our study, it is important to emphasize that a present (or absent) statistical relationship between brain morphometry in a specific region on the one hand and a cognitive performance parameter on the other does not necessarily reflect the functional neuroanatomy of the respective cognitive process. Therefore, future studies using functional neuroimaging methods may confirm and also amend or refine the conclusions of the present work. Moreover, in the investigation of associations between neural structure and cognitive performance, “cognitive reserve” (CR) represents an important theoretical concept, which, however, was not explicitly considered in the present work. CR is construed as capacity to compensate for a given neural damage by flexibly using alternative brain networks [52 –54]. Accordingly, CR provides an explanatory framework for the fact that individuals could markedly differ in their levels of cognitive or intellectual impairment, despite exhibiting similar levels of neural damage [55, 56]. In this respect, parameters of CR may represent important nuisance variables which could suppress or bias functional associations of brain-structural parameters. Moreover, if in a between-subject design compared groups are not matched for their CR, this may lead to either an over- or underestimation of group differences and, thus, complicate their interpretation [57]. Although the present study did not explicitly control for subjects’ CR, it was confirmed that the two groups (AD and HC subjects) did not deviate significantly in terms of age, years of education, as well as their gender distribution, which are all variables which have been shown to be importantly related to CR [58]. Moreover, age was confirmed to be normally distributed across groups without any outliers. Therefore, taken together, we can assume that CR did not significantly influence the here presented findings. Finally, in the present study, we did not assess CSF-based biomarkers which may be considered a limitation of the between-group study design. More specifically, amyloid-beta 42 and tau proteins could have been used, on the one hand, to confirm a common (i.e., homogeneous) AD-related neuropathology in the patient group and, on the other hand, to confirm the absence of neurodegenerative processes in the healthy control subjects [59]. However, our group assignment can be still considered sufficiently reliable as it is based on highly valid and internationally accepted diagnostic criteria (see method section). Moreover, each control subject‘s MRIs were assessed by trained radiologists who confirmed that the images exhibit no signs of AD (particularly no signs of mesiotemporal atrophy). In this context, the literature clearly shows that neuroimaging biomarkers already have a high predictive value for early onset AD [59, 60]. Concerning the aMCI patient subjects, admittedly we cannot be absolutely sure that really all of them will actually convert to AD. This is due to the purely cross-sectional study design and may be considered another potential limitation of the group classification. However, the aMCI diagnosis was established according to high standard criteria (the Mayo criteria and NIA-AA criteria; see Methods), which provide a strong argument to assume that the vast majority of aMCI subjects of our sample will actually convert to AD.

Conclusions

In the present combined brain-morphometric and neuropsychological study, we were able to define differential associations between hippocampal GM volume and the retrieval of autobiographical memory episodes of different retrieval frequencies. The results suggest that memory retrieval is associated with a neurofunctional transition (conceptually related to consolidation or “semanticization”) of the respective episodic memory entries whereby they gradually become independent of the hippocampal system. Most importantly, this neurofunctional transition process seems to be diffused or reversed in the progression of AD-pathology, so that the recall of frequently retrieved memory episodes is presumably related to hippocampal functional integrity in affected patients, as well. Future studies using functional neuroimaging methods are desirable to amend and refine the outlined conclusions.

Footnotes

ACKNOWLEDGMENTS

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Authors’ disclosures available online ().

The supplementary material is available in the electronic version of this article: .

References

Dubois

, Feldman

, Jacova

, DeKosky

, Barberger-Gateau

, Cummings

, Delacourte

, Galasko

, Gauthier

, Jicha

, Meguro

, O’Brien

, Pasquier

, Robert

, Rossor

, Salloway

, Stern

, Visser

, Scheltens

(2007) Research criteria for the diagnosis of Alzheimer’s disease: Revising the NINCDS-ADRDA criteria. Lancet Neurol 6, 734–746.

Braak

, Braak

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

Squire

, Alvarez

(1995) Retrograde amnesia and memory consolidation: A neurobiological perspective. Curr Opin Neurobiol 5, 169–177.

Squire

, Zola

(1998) Episodic memory, semantic memory, and amnesia. Hippocampus 8, 205–211.

Svoboda

, McKinnon

, Levine

(2006) The functional neuroanatomy of autobiographical memory: A meta-analysis. Neuropsychologia 44, 2189–2208.

Tulving

, Markowitsch

(1998) Episodic and declarative memory: Role of the hippocampus. Hippocampus 8, 198–204.

Kapur

(1999) Syndromes of retrograde amnesia: A conceptual and empirical synthesis. Psychol Bull 125, 800–825.

El Haj

, Antoine

, Nandrino

, Kapogiannis

(2015) Autobiographical memory decline in Alzheimer’s disease, a theoretical and clinical overview. Ageing Res Rev 23, 183–192.

Leyhe

, Müller

, Eschweiler

, Saur

(2010) Deterioration of the memory for historic events in patients with mild cognitive impairment and early Alzheimer’s disease. Neuropsychologia 48, 4093–4101.

10.

Meulenbroek

, Rijpkema

, Kessels

RPC

, Olde Rikkert

MGM

, Fernández

(2010) Autobiographical memory retrieval in patients with Alzheimer’s disease. Neuroimage 53, 331–340.

11.

Barbeau

, Didic

, Joubert

, Guedj

, Koric

, Felician

, Ranjeva

, Cozzone

, Ceccaldi

(2012) Extent and neural basis of semantic memory impairment in mild cognitive impairment. J Alzheimers Dis 28, 823–837.

12.

Müller

, Mychajliw

, Hautzinger

, Fallgatter

, Saur

, Leyhe

(2014) Memory for past public events depends on retrieval frequency but not memory age in Alzheimer’s disease. J Alzheimers Dis 38, 379–390.

13.

Addis

, Sacchetti

, Ally

, Budson

, Schacter

(2009) Episodic simulation of future events is impaired in mild Alzheimer’s disease. Neuropsychologia 47, 2660–2671.

14.

Addis

, Tippett

(2004) Memory of myself: Autobiographical memory and identity in Alzheimer’s disease. Memory 12, 56–74.

15.

Leyhe

, Müller

, Milian

, Eschweiler

, Saur

(2009) Impairment of episodic and semantic autobiographical memory in patients with mild cognitive impairment and early Alzheimer’s disease. Neuropsychologia 47, 2464–2469.

16.

Tulving

(1972) Episodic and semantic memory. Organ Mem 1, 381–403.

17.

Müller

, Mychajliw

, Reichert

, Melcher

, Leyhe

(2016) Autobiographical memory performance in Alzheimer’s disease depends on retrieval frequency. J Alzheimers Dis 52, 1215–1225.

18.

Beatty

, Salmon

(1991) Remote memory for visuospatial information in patients with Alzheimer’s disease. J Geriatr Psychiatry Neurol 4, 14–17.

19.

Brown

(2002) Consolidation theory and retrograde amnesia in humans. Psychon Bull Rev 9, 403–425.

20.

Meeter

, Kollen

, Scheltens

(2005) Retrograde amnesia for semantic information in Alzheimer’s disease. J Int Neuropsychol Soc 11, 40–48.

21.

Tomadesso

, Perrotin

, Mutlu

, Mézenge

, Landeau

, Egret

, De La Sayette

, Jonin

, Eustache

, Desgranges

, Chételat

(2015) Brain structural, functional, and cognitive correlates of recent versus remote autobiographical memories in amnestic mild cognitive impairment. Neuroimage Clin 8, 473–482.

22.

Dall’Ora

, Della Sala

, Spinnler

(1989) Autobiographical memory. Its impairment in amnesic syndromes. Cortex 25, 197–217.

23.

Leplow

, Dierks

, Herrmann

, Pieper

, Annecke

, Ulm

(1997) Remote memory in Parkinson’s disease and senile dementia. Neuropsychologia 35, 547–57.

24.

Brink

, Yesavage

, Lum

, Heersema

, Adey

, Rose

(1982) Screening tests for geriatric depression. Clin Gerontol 1, 37–43.

25.

Petersen

, Aisen

, Beckett

, Donohue

, Gamst

, Harvey

, Jack

, Jagust

, Shaw

, Toga

, Trojanowski

, Weiner

(2010) Alzheimer’s Disease Neuroimaging Initiative (ADNI): Clinical characterization. Neurology 74, 201–209.

26.

Petersen

, Smith

, Waring

, Ivnik

, Tangalos

, Kokmen

(1999) Mild cognitive impairment: Clinical characterization and outcome. Arch Neurol 56, 303–308.

27.

Albert

, DeKosky

, Dickson

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

28.

Rosen

, Mohs

, Davis

(1984) A new rating scale for Alzheimer’s disease. Am J Psychiatry 141, 1356–1364.

29.

Reisberg

, Ferris

, De Leon

, Crook

(1982) The global deterioration scale for assessment of primary degenerative dementia. Am J Psychiatry 139, 1136–1139.

30.

Folstein

, Folstein

, McHugh

(1975) “Mini-mental state”. A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 12, 189–198.

31.

Ashburner

, Friston

(2000) Voxel-based morphometry—the methods. Neuroimage 11, 805–821.

32.

Ashburner

(2007) A fast diffeomorphic image registration algorithm. Neuroimage 38, 95–113.

33.

Tzourio-Mazoyer

, Landeau

, Papathanassiou

, Crivello

, Etard

, Delcroix

, Mazoyer

, Joliot

(2002) Automated anatomical labeling of activations in SPM using a macroscopic anatomical parcellation of the MNI MRI single-subject brain. Neuroimage 15, 273–289.

34.

Pearson

, Filon

LNG

(1898) VII. Mathematical contributions to the theory of evolution. IV. On the probable errors of frequency constants and on the influence of random selection on variation and correlation. Proc R Soc Lond 62, 173–176.

35.

Diedenhofen

, Musch

(2015) Cocor: A comprehensive solution for the statistical comparison of correlations. PLoS One 10, e0121945.

36.

Irish

, Hornberger

, El Wahsh

, Lam

, Lah

, Miller

, Hsieh

, Hodges

, Piguet

(2014) Grey and white matter correlates of recent and remote autobiographical memory retrieval–insights from the dementias. PLoS One 9, e113081.

37.

Irish

, Lawlor

, O’Mara

, Coen

(2010) Exploring the recollective experience during autobiographical memory retrieval in amnestic mild cognitive impairment. J Int Neuropsychol Soc 16, 546–555.

38.

Murphy

, Troyer

, Levine

, Moscovitch

(2008) Episodic, but not semantic, autobiographical memory is reduced in amnestic mild cognitive impairment. Neuropsychologia 46, 3116–3123.

39.

De Simone

, Fadda

, Perri

, Aloisi

, Caltagirone

, Carlesimo

(2016) Does retrieval frequency account for the pattern of autobiographical memory loss in early Alzheimer’s disease patients? Neuropsychologia 80, 194–200.

40.

St. Jacques

, Rubin

, Cabeza

(2012) Age-related effects on the neural correlates of autobiographical memory retrieval. Neurobiol Aging 33, 1298–1310.

41.

Cabeza

, Daselaar

, Dolcos

, Prince

, Budde

, Nyberg

(2004) Task-independent and task-specific age effects on brain activity during working memory, visual attention and episodic retrieval. Cereb Cortex 14, 364–375.

42.

Grady

, McIntosh

, Horwitz

, Maisog

, Ungerleider

, Mentis

, Pietrini

, Schapiro

, Haxby

(1995) Age-related reductions in human recognition memory due to impaired encoding. Science 269, 218–221.

43.

Daselaar

(2003) Neuroanatomical correlates of episodic encoding and retrieval in young and elderly subjects. Brain 126, 43–56.

44.

Morcom

, Good

, Frackowiak

RSJ

, Rugg

(2003) Age effects on the neural correlates of successful memory encoding. Brain 126, 213–229.

45.

Starkstein

, Boller

, Garau

(2005) A two-year follow-up study of remote memory in Alzheimer’s disease. J Neuropsychiatry Clin Neurosci 17, 336–341.

46.

Cermak

(1984) The episodic-semantic distinction in amnesia. In Neuropsychology of Memory, Squire

, Butters

, eds. The Guilford Press, New York, pp. 55–62.

47.

Piolino

, Desgranges

, Eustache

(2009) Episodic autobiographical memories over the course of time: Cognitive, neuropsychological and neuroimaging findings. Neuropsychologia 47, 2314–2329.

48.

Ribot

(1882) Les maladies de la memoire Paris: Alcan. [Diseases of memory (1882), English translation of Les maladies de la memoire, 1881.].

49.

Moscovitch

, Rosenbaum

, Gilboa

, Addis

, Westmacott

, Grady

, McAndrews

, Levine

, Black

, Winocur

, Nadel

(2005) Functional neuroanatomy of remote episodic, semantic and spatial memory: A unified account based on multiple trace theory. J Anat 207, 35–66.

50.

Rose

, Chen

, Chalk

, Zelaya

, Strugnell

, Benson

, Semple

, Doddrell

(2000) Loss of connectivity in Alzheimer’s disease: An evaluation of white matter tract integrity with colour coded MR diffusion tensor imaging. J Neurol Neurosurg Psychiatry 69, 528–530.

51.

Daianu

, Jahanshad

, Nir

, Toga

, Jack

, Weiner

, Thompson

(2013) Breakdown of brain connectivity between normal aging and Alzheimer’s disease: A structural k-Core network analysis. Brain Connect 3, 407–422.

52.

Groot

, Van Loenhoud

, Barkhof

, Van Berckel

BNM

, Koene

, Teunissen

, Scheltens

, Van Der Flier

, Ossenkoppele

(2018) Differential effects of cognitive reserve and brain reserve on cognition in Alzheimer disease. Neurology 90, e149–e156.

53.

Stern

(2002) What is cognitive reserve? Theory and research application of the reserve concept. J Int Neuropsychol Soc 8, 448–460.

54.

Whalley

, Deary

, Appleton

, Starr

(2004) Cognitive reserve and the neurobiology of cognitive aging. Ageing Res Rev 3, 369–382.

55.

Mortimer

(1988) Do psychosocial risk factors contribute to Alzheimer’s disease? In Etiology of Dementia of the Alzheimer’s Type, Henderson

Henderson

, eds. John Wiley & Sons, New York, pp. 39–52.

56.

Katzman

, Terry

, DeTeresa

, Brown

, Davies

, Fuld

, Renbing

, Peck

(1988) Clinical, pathological, and neurochemical changes in dementia: A subgroup with preserved mental status and numerous neocortical plaques. Ann Neurol 23, 138–144.

57.

Frankenmolen

, Fasotti

, Kessels

RPC

, Oosterman

(2018) The influence of cognitive reserve and age on the use of memory strategies. Exp Aging Res 44, 117–134.

58.

Nucci

, Mapelli

, Mondini

(2012) Cognitive Reserve Index questionnaire (CRIq): A new instrument for measuring cognitive reserve. Aging Clin Exp Res 24, 218–226.

59.

Sheikh-Bahaei

, Sajjadi

, Pierce

(2017) Current role for biomarkers in clinical diagnosis of Alzheimer disease and frontotemporal dementia. Curr Treat Options Neurol 19, 46.

60.

Coupé

, Fonov

, Bernard

, Zandifar

, Eskildsen

, Helmer

, Manjón

, Amieva

, Dartigues

, Allard

, Catheline

, Collins

(2015) Detection of Alzheimer’s disease signature in MR images seven years before conversion to dementia: Toward an early individual prognosis. Hum Brain Mapp 36, 4758–4770.