Impairment of Episodic-Specific Autobiographical Memory in Individuals with Subjective Cognitive Decline and in Patients with Prodromal or Mild Alzheimer’s Disease

Abstract

Background:

Autobiographical memory (AM) is a personal form of memory that becomes impaired in the early, clinical stages of Alzheimer’s disease (AD). In the “preclinical” phase of AD, neuropathological hallmarks are present (especially in a brain network underpinning AM), but performance on standardized neuropsychological tests is normal. Even so, some patients have subjective cognitive decline (SCD).

Objective:

The aim was to 1) investigate AM performance on two tests with different approaches in SCD, and in prodromal and mild AD, and 2) examine the association between the AM tests.

Methods:

We included 17 SCD patients with heightened risk of AD, 17 amnestic mild cognitive impairment (aMCI) patients, 17 patients with mild dementia due to AD, and 30 healthy controls. Patients were diagnosed according to international criteria, and all participants had MMSE scores≥24. AM was assessed using the Columbia Autobiographical Memory Interview-Short Form (CAMI-SF) and the Three Events Test. These tests measure the production of contextual details.

Results:

Significant group effects were found for the Three Events Test and the CAMI-SF. All patient groups produced significantly fewer contextual details than the controls on the Three Events Test. On CAMI-SF, the aMCI and mild AD groups were able to answer fewer questions or gave significantly less detailed answers than the other groups. The SCD patients performed below the controls on CAMI-SF, but the difference was not significant.

Conclusion:

AM may be impaired in very early AD, even in the phases where standardized episodic memory tests show no decline.

Keywords

Alzheimer’s disease autobiographical memory dementia memory disorder mild cognitive impairment subjective cognitive decline

INTRODUCTION

The pathophysiological process of Alzheimer’s disease (AD) begins many years, or even decades, before AD is clinically diagnosed [1]. In the earliest phases of AD, where biomarkers may be abnormal but no cognitive symptoms have yet been objectively identified, some patients feel that their cognitive abilities are declining—they experience subjective cognitive decline (SCD) [2]. Persons with SCD have an increased risk for future cognitive decline or dementia due to AD [3]. However, SCD can be caused by various medical conditions, and in most individuals SCD does not progress to objective cognitive impairment [4]. Studies have found that the risk of progression is highest in persons who have severe cognitive complaints or have SCD-plus, which is SCD with the presence of certain characteristics associated with an increased risk of cognitive decline [2 , 6]. Recently, the list of SCD-plus features, originally presented by the SCD-Initiative working group in 2014 [2], has been updated [4]. The list currently includes the following features: subjective impairment of memory, onset of SCD within the past five years, onset after the age of 60, worries associated with SCD, SCD persistent over time, help-seeking behavior, and confirmation of impairment by an informant (and, if available, biomarker evidence of preclinical AD) [2]. By definition, SCD patients have normal performances on traditional neuropsychological tests assessing for example learning and recall of word lists and pictures. Whether patients with SCD-plus have objective autobiographical memory (AM) deficits has never been investigated, but results from previous studies have led us to believe that impairments in AM could be a very early cognitive marker of AD [7 –16].

Within the field of neuropsychology, AM is a construct that refers to the episodic memories and semantic knowledge related to an individual’s personal experiences. It is impaired in amnestic mild cognitive impairment (aMCI) (e.g., [7 –9]) and in dementia due to AD [17]. Autobiographical memory tasks evoke activity in a core AM network that overlaps considerably with the default mode network (DMN) [10 –12]. Functional connectivity changes as well as accumulation of Aβ₄₂ are found in the DMN in the earliest (“preclinical” and prodromal) stages of AD [13 –15]. This association between the AM network and the pattern of neuropathological changes in early AD is one reason why it seems likely that AM impairment could be present in patients with SCD-plus. Additionally, two studies have suggested that the production of personal semantic or episodic event-related details is compromised in otherwise cognitively normal individuals with positive AD biomarkers [16, 18].

Studies investigating AM impairment in aMCI have not reported information on participants’ AD biomarkers (e.g., [8 , 19]), or they have combined AM results for aMCI and dementia patients (e.g., [20, 21]). Further, different methods for assessing AM have been applied, ranging from structured, often lengthy, interviews to cued or free recall tests, and most studies have only used one AM test. Thus, it is unknown to which degree performance on different types of AM tests may be impaired in the earliest stages of AD.

Executive dysfunction has been found to significantly affect AM retrieval in patients with AD [17], which is not surprising given that AM tasks activate multiple brain areas including frontal regions. However, the association between executive functioning and potential AM deficits in patients with very early AD (SCD or aMCI) is understudied.

This study aimed to investigate performance on two short AM tests with different approaches (but a shared focus on the provision of contextual details) in 1) patients with SCD-plus, patients with prodromal and mild AD, and 2) a group of healthy control participants. A secondary aim was to examine the degree to which the two AM tests were associated with each other and with results from neuropsychological tests assessing attentional and executive processes. Since the chosen AM tests have not been formally validated in patients with AD, the nature of the study was exploratory.

MATERIALS AND METHODS

Participants

A total of 51 patients with SCD (n = 17), aMCI (n = 17), or mild dementia due to AD (n = 17) were recruited from the Memory Clinic at the Danish Dementia Research Centre. The clinic offers diagnostic evaluation and treatment of a wide spectrum of cognitive disorders to patients referred from general practitioners, private consultants in neurology or psychiatry, or other hospitals. At the time of inclusion, all patients had undergone a thorough, clinical assessment, which included a physical and neurological examination, routine laboratory tests, an electrocardiogram, a cerebral magnetic resonance imaging (or in some cases a computed tomography) scan, cognitive testing with Danish versions of the Mini-Mental State Examination (MMSE) and Addenbrooke’s Cognitive Examination (ACE) [22], and a full neuropsychological evaluation to establish a cognitive profile. The neuropsychological test battery included: the Rey Auditory Verbal Learning Task (RAVLT), the Category Cued Memory Test (CCMT-48) [23], the Rey Complex Figure Test (RCFT), the Famous Faces test [24], the Symbol Digit Modalities Test (SDMT), the Trail Making Test (TMT, part A and B), the Boston Naming Test [25], categorical (animals) and phonological (S, F, and A words) verbal fluency tasks, the Five-Point Test [26] as well as a Danish modification of the Block Design test. For all tests except the TMT-A and TMT-B, the variables used for interpretation of the results were the number of correct responses (for TMT, time to complete was used for the analyses). If clinically relevant (as decided by a senior neurologist), further examinations—such as a fluorodeoxyglucose positron emission tomography (FDG PET) scan, a Pittsburgh compound B (PiB) PET scan, an examination of Aβ₄₂, tau, neurofilament, and total protein levels in the cerebrospinal fluid (CSF), an electroencephalography, or a psychiatric evaluation—were performed. All patients were examined for one or more AD biomarkers (CSF Aβ₄₂, CSF p-tau, PiB-PET standardized uptake value ratio, FDG-PET glucose metabolism pattern, and/or medial temporal atrophy). For all patients, a syndromic and etiological diagnosis was established in accordance with international diagnostic guidelines on a multidisciplinary consensus conference (the specific criteria are described in the following section).

In addition to the patient groups, we included a control group with 30 healthy participants recruited on a voluntary basis. The controls were assessed with the same cognitive and psychiatric test battery as the patients. They were not examined for AD biomarkers.

Inclusion and exclusion criteria

To be eligible for inclusion, participants had to be at least 45 years and have Danish as their native language. Exclusion criteria for all groups were: 1) presence of (other) neurological, somatic, or psychiatric disorders that could significantly affect cognition; 2) alcohol or substance abuse; and 3) use of medication known to affect cognitive functioning. For screening of depression, we used the Geriatric Depression Scale with 15 items (GDS-15) [27] with a cut-off score of≥5. Individual symptoms of anxiety or depression, which did not reach the threshold of a disorder, did not lead to exclusion. Neither did a limited benzodiazepine intake at bedtime prescribed to aid sleeping.

Group specific criteria. In accordance with the guidelines published by the National Institute of Aging and the Alzheimer’s Association (NIA-AA) [28], MCI was in this study operationalized as: 1) significant impairments in one or more cognitive domains; 2) concerns about change in cognitive functioning expressed by the patient or an informant; and 3) no impairment of activities of daily living (as measured by the Functional Activities Questionnaire; [29]). A score was classified as impaired when it was below –1.5 SD on age-, gender-, and education-adjusted normative data (for tests with skewed distribution of data in healthy controls, e.g., the Famous Faces Test, scores below the expected 10th percentile were considered impaired). Only MCI patients with amnestic cognitive profiles (aMCI) were included, and 14 out of the 17 patients had at least one abnormal AD biomarker.

Since patients with mild dementia due to AD were also diagnosed based on NIA-AA criteria [30], they had: 1) significant impairments in at least two cognitive domains, which represented a change from an earlier level of functioning; and 2) impaired activities of daily living. To be included, AD patients with mild dementia had to present primarily with impaired episodic memory, have MMSE scores≥24, and have at least one abnormal AD biomarker.

The SCD patients fulfilled the criteria presented in Jessen, et al. [2]. Thus, they all had: 1) SCD unrelated to an acute event; and 2) normal age-, gender-, and education-adjusted performances on standardized cognitive tests (one performance below expectation was accepted). The SCD patients had all been referred to the Memory Clinic for a diagnostic evaluation and had therefore spontaneously expressed concerns related to SCD as well as help-seeking behavior. To further increase the likelihood that they had preclinical AD, they were required to have subjective impairment of memory (this was evaluated based on their main complaints reported in the initial phase of their clinical assessment). Since the included SCD patients had worries associated with SCD, help-seeking behavior, and specific memory complaints, they had at least three of the SCD-plus features listed in Jessen et al. [4]. No additional features of SCD-plus were needed for inclusion in the study, but during the clinical assessment we found that they all had onset within the past five years, that 11 patients had onset after the age of 60, and that five patients had at least one abnormal AD biomarker. An application of a Danish translation of the Cognitive Change Index-self report (CCI-S) [31] showed that 13 SCD patients felt that their memory was impaired compared to others of the same age. A total of 14 SCD patients had an informant who reported on the patient’s memory function on the informant version of the CCI (CCI-I) [31], and this led to the confirmation of memory deficits in nine SCD patients. Thus, even though the SCD patients were only required to have three SCD-plus features to be included in the study, most had more. Even though the coding system for essential features of SCD that has been proposed for research purposes [2] was not systematically used in this study, most of the information was registered and can be derived from the information described above. We did not examine APOE genotype or anxiety, and the patients were not directly asked whether they had subjective impairment of other cognitive domains than memory.

The healthy controls did not have SCD as operationalized in this study, since they had not shown help-seeking behavior and were excluded if they expressed concerns associated with SCD. They were also excluded if they reported that their memory was impaired compared to others of the same age (as reflected on the 12th item on the CCI-S). If no impairment was reported on item 12 on the CCI-S, a total score just above cut-off on this scale (a total score from the first 12 items > 16; [32]) did not lead to exclusion. As in the SCD group, controls were excluded if they had more than one test performance below expectations based on age-, gender-, and education-adjusted norms.

Autobiographical memory tests

For all patients, the AM tests were administered in a separate visit. During this visit, we also applied four questionnaires about patients’ subjective memory functioning (including the CCI scales), five scales investigating their sense of identity or their self-worth, and the GDS-15. Most participants were able to complete this session within two and a half hours, and they were offered a short break midway. The tests were administered in a fixed order by the same neuropsychologist in all participants, and this neuropsychologist was also responsible for coding and analyzing the test results. Since the neuropsychologist had also recruited the study participants, she was not blind to the diagnostic group of each patient.

Three Events Test

Three Events Test is a name that we have chosen for the procedure used to assess AM in El Haj and Antoine [33]. The test consists of three parts with three different themes: Family Event, Professional Event, and Holiday Event. These three cues tend to elicit autobiographical memories in AD patients [34]. Within each theme, participants were asked to recount in detail an event from their lives. To ensure specificity, the participants were informed that the event had to have lasted no more than one day. They were also told that they had a maximum of three minutes to tell about the chosen event, and they were encouraged to provide as many specific details of the event as possible (e.g. time and place). The time limit served the purpose of avoiding redundancy and distractibility, and previous studies have found that three minutes is sufficient for AD patients to recall autobiographical memories [35, 36]. The participants were allowed time to think of an event before timing was started. If no memory could be recalled, the test administrator proceeded to the next theme.

For each part of the Three Events Test, we calculated a Specificity Score and a Contextual Details Score. We then calculated a mean score for both types of scores: a Mean Specificity Score and a Mean Contextual Details Score.

Specificity. The Specificity Score reflects the extent to which a recalled autobiographical memory is specific as opposed to general. To assess the specificity of a recalled memory, we applied a Danish adaptation of the procedure used by El Haj and Antoine [33]. This procedure is derived from the TEMPau-scale [37], which is based upon the Autobiographical Memory Interview (AMI) by Kopelman et al. [38]. Scores were from zero to four points. Zero points were given if no memory was recalled or if only very general information on a theme was provided (e.g., “in my childhood . . . ”). A score of one point was used if the participant told of a repeated or extended event (e.g., “my best friend and I went to the grocery store every day after school”). A score of two points was used if the event was situated in time and/or in space (e.g., “my best and I went to the grocery store, which was on the way to my house, every day after school”). A specific event lasting less than 24 hours and situated in time and/or space elicited three points (e.g., “one day we played around in the store and I overturned a rack of candy”). If phenomenological details, such as feelings, perceptions, or thoughts, were also provided, the participant was given four points (e.g., “it was an accident, and I apologized to the store clerk, who was very angry”).

Contextual details. Since ceiling effects is a known issue with the TEMPau scale when using it on healthy or generally well-functioning individuals (including SCD and aMCI patients), we also recorded the number of contextual details provided during every autobiographical recall [39]. Previous studies have found this to be a sensitive measure of autobiographical specificity [36 , 40]. A contextual detail was defined as any meaningful piece of new information about who, where, what, or when. Specific, phenomenological details (e.g., “it made me very sad” or “I remember feeling very happy”) were also counted, but general statements, such as “really, it was just a lovely day”, were not. In the following fictive example of part of a coded Family Event, the recorded contextual details have been circled:

In the example above, the words “summer”, “2016”, “10”, and “13” are details concerning when the event took place. Details of where the event took place are provided with the words “Zoo” and “Copenhagen”. “My daughter”, “her husband”, and “their children” reveal the persons who were present besides the “I”. The word “I” was not counted as a detail in any reported event unless it was important that the “I”, and not somebody else, did something. “We saw a sea lion show”, “fed fish”, “different tricks”, and “clapped their fore flippers” are details about what happened at the event. Finally, the words “looked funny” is a specific, phenomenological detail.

When the Specificity Score was≤2, no Contextual Details Score was registered (participants with a Specificity Score of≤2 in one theme were likely to have been able to provide some amount of contextual details if they had thought of a more specific memory).

Columbia Autobiographical Memory Interview–Short Form (CAMI-SF) [41]

The CAMI-SF is a relatively short, structured autobiographical memory interview initially developed to measure AM loss due to electro-convulsive therapy (ECT). When the interview has been used to measure memory loss in ECT patients, participants have been assessed both before and after treatment. In this study, testing was only performed once. The scale contains six themes: Family Member, Last Major Trip, Last New Year’s Eve, Last Birthday, Last Employment, and Last Physical Illness. Within each theme, five questions that required detailed, specific answers were asked. Whereas the Family Member and Last Employment themes only required recall of semantic information (e.g., names, addresses, and dates), both episodic and semantic autobiographical details had to be recalled within the remaining four themes. Episodic details were probed by questions such as “What did you enjoy most about that trip?” or “What did you do at midnight?”. Participants were encouraged to be as specific as possible when answering the questions.

We applied the scoring-system developed by Semkovska et al. (2012) [42], which allows a measuring of specificity-richness of retrieved autobiographical memories. Each specific memory provided by a participant elicited one point, and thus the Total Score was unlimited. For example, when participants were asked what they had enjoyed most about their last trip, an answer such as “Swimming in the ocean” would be assigned one point, while the answer “Seeing my grandmother and going to the market” would be assigned two points. Some semantic questions could not elicit more than one point (e.g., “For how many days were you away on your last trip?”).

In this study, no more than 12 full names were recorded when participants had celebrated New Year’s Eve or their birthday with many people. If a participant reported having travelled alone on her/his last trip or having spent New Year’s Eve alone, such a response was recorded as a specific memory. One point was assigned for the full name of a family member or friend (one point was also assigned if a participant reported the first name of a person whose last name they would not be expected to know, e.g., a son’s new girlfriend or the neighbor of a family member). Within the Last Physical Illness and Last Employment themes, two points were assigned if both first and last name of a Doctor or a supervisor was reported. Besides the Total Score, a Semantic Subscore (total score for Family Member and Last Employment), an Episodic-Extended Subscore (total score for Last Major Trip and Last Physical Illness), and an Episodic-Specific Subscore (total score for Last New Year’s Eve and Last Birthday) were also calculated.

Statistical analysis

Group comparisons were performed using one-way analysis of variance (ANOVA) followed by Dunnett’s post hoc test. When the assumptions of homogeneity of variance and/or normality were not met, the Kruskal-Wallis test was applied. For nonparametric comparison of each patient group against the control group, we used the Mann-Whitney U test. To compare all groups one-by-one, Dunn’s post hoc test was applied. All nonparametric post hoc analyses were followed by the Bonferroni-Holm procedure to correct for multiple comparisons (adjusted p values in this article are noted as p_adj). For nonparametric analyses, Cohen’s r was calculated as an estimate of effect size [43] (mean differences and confidence intervals were used for parametric analyses). Pearson’s Chi-Square test was used to investigate the significance of differences in gender distribution. Associations between the main scores or subscores on the AM tests and other variables (age and scores from TMT-A, TMT-B, SDMT, S-word fluency, and animal fluency) were examined using Pearson’s or Spearman’s Correlation Analysis depending on normality and heterogeneity of variance. Differences were considered statistically significant at an alpha level below 0.05.

Ethical approval

The project has been reviewed by the local Ethic Committee of the Capital Region of Denmark (registry number: H-18048977).

RESULTS

Background data for the participants are presented in Table 1 along with results from the cognitive and neuropsychiatric screening tests. Group comparisons showed no significant differences in gender, χ²(3) = 3.02, p = 0.388. There were significant effects of group in age, F(3, 77) = 6.30, p < 0.01, and in the scores from the MMSE, χ²(3) = 31.49, p < 0.001, the ACE, χ²(3) = 41.27, p < 0.001, and the GDS-15, χ²(3) = 13.40, p < 0.01. Post hoc testing showed that the SCD patients were significantly younger than the patients with aMCI, MD = –8.82, CI 95%[–14.40, –3.25], p < 0.01, the patients with mild dementia, MD = –8.71, CI 95%[–14.28, –3.13], p < 0.01, and the controls, MD = –5.42, CI 95%[–10.35, –0.48], p < 0.05. The SCD and aMCI patients had significantly higher scores on the GDS-15 compared to the control subjects, z = 3.17, r = 0.46, p_adj <0.05, and z = 2.84, r = 0.41, p_adj <0.05, respectively. Only one participant had a score above the cut-off of≥5 (an aMCI patient with a score of 6, who was included, since clinical depression was not diagnosed). The slightly higher scores on the GDS-15 in the SCD and aMCI groups may stem from one item addressing subjective memory dysfunction.

Table 1

Background data and results of the cognitive and neuropsychiatric screening tests for the patient and control groups. Age is presented as mean (SD), while the results from GDS-15, MMSE, and ACE are presented as median [range]

	SCD	aMCI due to AD	Mild dementia due to AD	Controls
n	17	17	17	30
Age	63.1 (7.0)^*†‡	71.9 (5.2)	71.8 (6.6)	68.5 (7.6)
Sex (M/F)	5/12	7/10	9/8	16/14
GDS-15	2^* [0–4]	1^* [0–6]	1 [0–4]	0 [0–4]
MMSE	30 [28–30]	28 [25–30]	25^*#† [24–30]	30 [26–30]
ACE	95 [88–98]	90^*# [74–96]	79^*#† [70–95]	96 [84–99]

^*Significant difference from the control group (p < 0.05); ^#Significant difference from the SCD group (p < 0.05); ^†Significant difference from the aMCI group (p < 0.05); ^‡Significant difference from the mild dementia group (p < 0.05). All p values from nonparametric tests are Bonferroni-Holm-adjusted.

Table 2 shows the performances of each group on the Three Events Test and the CAMI-SF. One patient with mild dementia was unable to complete the CAMI-SF due to tiredness and was excluded from the statistical analyses on the CAMI-SF data. On the Three Events Test, four aMCI patients, six mild dementia patients, and two controls had one Specificity Score≤2 (and therefore one missing Contextual Details Score). A single participant, who had mild dementia, had two Specificity Scores≤2 and two missing Contextual Details Scores. When only one or two Contextual Details Scores were available for a participant, these scores were used to calculate the Mean Contextual Details Score. Thus, all participants had a Mean Contextual Details Score. Participants with a missing Contextual Details Score on a specific Three Events Test theme were excluded from any analyses concerning that theme.

Table 2

Performance on CAMI-SF and Three Events Test for the patient and the control groups. The results from CAMI-SF and Three Events Test are presented as mean (SD) and median [range], respectively

	SCD	aMCI due to AD	Mild dementia due to AD	Controls
CAMI-SF	n = 17	n = 17	n = 16	n = 30
Total Score	52.2 (9.3)	47.2 (12.3)^*	41.3 (10.7)^*#	58.3 (10.6)
Three Events Test	n = 17	n = 17	n = 17	n = 30
Mean Contextual Details Score	17.3^* [12.0–30.1]	17.3^* [9.3–32.3]	14.0^* [8.7–29.0]	25.7 [14.3–34.0]
Mean Specificity Score	4.0 [3.7–4.0]	4.0 [2.7–4.0]	3.7^* [2.3–4.0]	4.0 [3.3–4.0]

^*Significant difference from the control group (p < 0.05); ^#Significant difference from the SCD group (p < 0.05); ^†Significant difference from the aMCI group (p < 0.05). Scores from the Three Events Test were calculated as mean scores from the three themes but are presented as median [range] because nonparametric analyses were applied. All p values from nonparametric tests are Bonferroni-Holm-adjusted.

On the Three Events Test, there was a significant effect of group in the Mean Specificity Score, χ²(3) = 12.08, p < 0.01, and in the Mean Contextual Details Score, χ²(3) = 41.27, p < 0.001. The mild dementia group was the only group that had significantly lower Mean Specificity Scores than the control group, U = 351.50, r = 0.37, p_adj < 0.05. Patients with SCD produced significantly fewer contextual details compared to the controls, U = 136.50, r = –0.38, p_adj <0.01. The same was found for the patients with aMCI, U = 123.50, r = –0.42, p_adj < 0.01, and the patients with mild dementia, U = 63.50, r = –0.62, p_adj < 0.001 (see Fig. 1). To assess whether the difference between the SCD and the control group was driven by group differences in performance on one or more themes on the Three Events Test, separate Mann Whitney U tests for the three themes were performed. It was only on the Professional Event theme that the SCD patients (n = 17) had significantly impaired Contextual Details Scores compared to the controls (n = 29), U = 110.50, r = –0.46, p < 0.01. On the Holiday Event theme, there was a trend towards a difference between the SCD (n = 17) and the control group (n = 29), U = 162.00, r = –0.28, p = 0.054. Since the SCD patients were younger than the controls, we used correlation analyses to determine if age had a significant impact on the Mean Contextual Details Score or on the Contextual Details Score from the Professional Event theme. No significant associations were found in the control group, rho(28) = –0.24, p = 0.207 and rho(27) = –0.18, p = 0.338, respectively. Negative results were also found in the SCD group, rho(15) = –0.29, p = 0.265 and rho(15) = –0.25, p = 0.331.

Fig. 1

Distribution of Age by Three Events Test Mean Contextual Details Score in the four participant groups.

Possible associations between the Three Events Test Mean Contextual Details Score and scores on neuropsychological tests assessing executive or attentional processes (verbal fluency [S-words and animals], SDMT, TMT-A, and TMT-B) were investigated with correlation analyses in the three patient groups combined. Significant associations with the SDMT score, rho(48) = 0.35, p < 0.05, and with the TMT-B score, rho(47) = –0.32, p < 0.05, were found, but these did not remain significant after a Bonferroni-Holm correction (p_adj = 0.12 and p_adj = 0.16, respectively).

To investigate the frequency of impairment on the Three Events Test, we examined the number of patients who had a Mean Contextual Details Score below the 15^th percentile. In the SCD group, this was 23.5%, and when only considering the Professional Event theme, the frequency increased to 29.4%. In the control group, these frequencies were 13.3%and 6.9%. The frequencies of aMCI and mild dementia patients with abnormal Mean Contextual Details Scores were 35.3%and 70.6%, respectively.

On the CAMI-SF, there was a significant effect of group in the Total Score, F(3, 76) = 9.77, p < 0.001. The aMCI and mild dementia patients had significantly lower Total Scores than the controls, MD = –11.12, CI 95%[–19.02, –3.23], p < 0.01, and MD = –17.05, CI 95%[–25.10, –9.00], p < 0.001, respectively. Mild dementia patients also had significantly lower Total Scores than the SCD group, MD = –10.93, CI 95%[–19.87, –1.98], p < 0.05. In the SCD group, Total Scores were not significantly lower compared to controls, MD = –6.12, CI 95%[–14.02, 1.77], p = 0.168. To investigate if group differences between the patient groups and the control group on CAMI-SF were mediated by selective impairments in semantic, episodic-extended, or episodic-specific AM, we performed ANOVAs and Dunnett’s tests for each subscore. There were significant group effects for all three subscores, F(3, 76) = 9.20, p < 0.001, F(3, 76) = 7.27, p < 0.001, and F(3, 76) = 4.17, p < 0.01, respectively. The aMCI patients only performed significantly below the controls on the episodic-specific parts of the CAMI-SF, MD = –7.91, CI 95%[–14.61, –1.21], p < 0.05. Compared to the controls, patients with mild dementia were impaired on the Semantic Subscore, MD = –4.31, CI 95%[–6.32, –2.30], p < 0.001, the Episodic-Specific Subscore, MD = –8.15, CI 95%[–14.98, –1.32], p < 0.05, and on the Episodic-Extended Subscore, MD = –4.59, CI 95%[–7.24, –1.94], p < 0.001. The SCD patients differed the most, although not significantly, from the control group on the Episodic-Specific Subscore, MD = –5.61, CI 95%[–12.31, 1.08], p = 0.12.

In a combined aMCI and mild dementia group, we investigated whether the CAMI-SF Total Score was significantly associated with scores from two of the fluency tests (S-words and animals), the SDMT, the TMT-A, or the TMT-B. We did not find any significant correlations.

The association between the CAMI-SF and the Three Events Test was investigated using a correlation analysis with data from the three patient groups. This showed a significant, although relatively weak, correlation between the CAMI-SF Total Score and the Three Events Test Mean Contextual Details Score, rho(48) = 0.319, p < 0.05.

DISCUSSION

Results from previous studies show that AM is impaired in the earliest stages of AD [7–9 , 17], but it has never been investigated in persons with SCD-plus. In this study, we applied two different tests that tap AM. On the Three Events Test, we found that persons with SCD-plus had significant impairments in the ability to recall contextual details of lived events compared to healthy people. We found a moderate effect size (–0.38) [43]. On the CAMI-SF, only the aMCI and mild dementia groups had performances that were significantly impaired compared to the control group. Thus, our results to some extent indicate (and do not contradict) that AM impairment may be an early cognitive marker of AD as suggested in two previous studies [16, 18].

To investigate how tests assessing AM were impaired at different early stages of (probable) AD, our study also included patients with aMCI and mild dementia due to AD. In accordance with previous studies (e.g., [9 , 44]), we found that such tests were impaired in these patient groups. The effects can be interpreted as moderate to large, but relatively broad confidence intervals do suggest that the effect sizes were associated with some uncertainty. By almost exclusively including aMCI and mild dementia patients with one or more abnormal AD biomarkers, we sought to increase the probability that AD was in fact the cause of cognitive decline in these patients.

Even though the Three Events Test and the CAMI-SF are methodologically different, we found that they were significantly associated and therefore, to some extent, measure the same cognitive function. Some uncertainty about the exact nature of this underlying function remains, since the chosen tests can only be expected to tap other cognitive functions such as retrieval and attentional processes in addition to AM. However, we did not find any significant correlations between the Three Events Test Mean Contextual Details Score or the CAMI-SF Total Score and scores from tests assessing other cognitive functions. While this does not the exclude the possibility that subscores from the two AM tests could be significantly associated with other cognitive test scores, these subscores were not included in the correlation analyses in order to avoid a further increase of Type 1 error risk. Our results suggest that the Three Events Test and the CAMI-SF do not simply capture verbal fluency or other executive or attention-related functions.

Several factors could explain why a significant difference between the SCD and the control group was only found on the Three Events Test despite the significant correlation between the AM tests. First, the sample sizes in each group were relatively small, and the performances in the SCD and control groups on the CAMI-SF could possibly have been significantly different in larger samples. Secondly, even though the value of the Total Score on the CAMI-SF is unlimited, the participants often gave brief answers to the open questions that generally followed closed questions (e.g., people were asked when their last birthday was, who they were with and where they were before they were asked what they did). Perhaps this questionnaire format “primed” participants to provide less detailed responses which then led to smaller group differences. Thirdly, the Three Events Test could be measuring aspects of AM likely to be impaired in very early AD in a more focused way than the CAMI-SF. While the Three Events Test only measures information related to specific episodes, the CAMI-SF measures both semantic autobiographical knowledge, episodic-extended, and episodic-specific autobiographical memories (recall of extended events require activation of more general or semanticized information than recall of specific events do) [33]. Semantic memory is impaired in the early stages of AD [45] but the impairment becomes more pronounced with disease progression [8, 46]. Thus, it is not surprising that the patients with mild dementia were impaired on all parts of the CAMI-SF. Even though aMCI patients typically have semantic memory deficits [45], this patient group only performed significantly worse than controls on the episodic-specific part of the CAMI-SF in our study. Of course, there is a possibility that deficits in non-personal aspects of semantic memory in some way contributed to the impaired performances of the three patient groups on the episodic-specific part of the CAMI-SF and on the Three Events Test (though semantic deficits were not found for any SCD patients on the traditional neuropsychological tests). It does seem, however, that the aMCI patients in our study were not yet significantly impaired in the autobiographical aspects of semantic memory assessed by the CAMI-SF. In accordance with this observation, a few other studies have previously found that semantic autobiographical knowledge deteriorates more slowly in early AD than episodic autobiographical memories [47, 48].

The two AM tests in this study were always applied within the same test session and in the same order. Obviously, this could induce the risk of practice effects. A deliberate choice was made to try and minimize the risk of these effects by always administering the Three Events Test before the CAMI-SF. Since autobiographical knowledge often triggers the recall of specific episodic memories [49], the specific questions asked on the CAMI-SF could possibly activate episodic memories that could then be reactivated during the Three Events Test. The Three Events Test, on the other hand, is a free recall test that does not require the recall of specific semantic details relevant for answering many of the questions on the CAMI-SF. Also, unlike on the CAMI-SF, it is not required that the participants focus on their most recent experiences. Although practice effects may have occurred, it was not our impression that the reported events on the Three Events Test overlapped much with the information provided on the CAMI-SF.

The retrieval process initiated in the recall of a personal memory is more complex than the process involved in episodic recall of non-personal test materials such as word lists or pictures [50]. This complexity has made it difficult to develop sensitive, valid AM tests that are not very lengthy and time consuming [51, 52]. For this study, we prioritized the use of relatively short tests that had the potential to uncover any deficits in AM in our patient groups. As is typical for most neuropsychological tests, the CAMI-SF and the Three Events Test have possible weaknesses in addition to their strengths. Like classic AM tests, such as the AMI [38], neither the CAMI-SF nor the Three Events Test (as used in this study) investigate the phenomenological aspects of autobiographical memory retrieval processes. Though an assessment of the participants’ subjective recollective experiences could potentially have contributed with interesting information, it was beyond the scope of this study.

The CAMI-SF has three items that require the participants to list the names of people that they celebrated last New Year’s Eve and their last birthday with as well as the names of those who brought presents for their birthday. These items may have given participants with a large social network an unfair advantage. On the other hand, omitting data from these three items in our analyses could have served to ameliorate deficits in participants who were unable to remember the names of people. Since calculating an adjusted CAMI-SF total score that omitted data from the three items made no significant difference to the results, we decided not to use an adjusted score.

The Three Events Test does not specify time periods from which the participants should retrieve their memories. This may have had at least three possible negative effects: strategic memory search may have been more difficult for the participants; participants were free to choose the memories that were most accessible to them; and comparing of memory ability across participants may have been more difficult, since controlling for the age of the memories has not been not possible. Nonetheless, modifying the test by adding a specific time period would likely not have been unproblematic either. More data could have been missing, since a larger portion of patients with dementia would not have been able to recall an event or they might not have remembered the exact age of a specific memory. Additionally, if the participants had been asked to recall for example a recent memory from within the past five years, some participants would probably still have chosen very recent memories while others would have chosen memories that were several years old.

Finally, there are a variety of factors besides memory (and other cognitive functions) that may affect performance on AM tests in general. Some of these are personality traits (especially openness and extraversion) [53], other psychological factors such as depression [54], and gender [55]. Thus, some individuals may, by nature, give more complex answers than others. Since there were no group differences in gender and no cases of current clinical depression in our sample, we see no reason that these factors would have affected our finding of significant group differences in AM. We cannot, however, exclude the possibility that a history of depression in some of the participants may have had some influence on performance. It would be interesting to investigate to which extent personality factors affect AM (and whether traits are distributed differently) in groups of patients with SCD-plus and aMCI compared to a group of cognitively intact people.

The nature of the present study was exploratory. The CAMI-SF was not previously used for the investigation of AM in AD, and, to our knowledge, the Three Events Test has not been formally validated in this patient group. This study was not hypothesis-driven and should only be seen as an exploration of possible AM deficits in patients with SCD-plus and aMCI. The use of non-validated measures and the lack of reliability measures puts a natural limit to the certainty with which we can claim validity of our results. Thus, a formal validation of the CAMI-SF and the Three Events Test in patients with very early AD is necessary to evaluate the validity of the findings presented in this paper.

In conclusion, the results of the present study suggest that memory loss in SCD may not just be a subjective phenomenon. We applied two tests tapping AM, and we found impairments on a test that measures contextual details related to a specific event. This may support the notion that SCD could represent an even earlier clinical stage of AD than aMCI. Future studies including larger samples, validated measures of AM, measures of personality traits, and investigation of the correlation between AM performance and AD biomarkers (particularly in SCD-plus patients) are needed to gain further insight into AM in the earliest stages of AD and to confirm whether AM decline is a very early cognitive marker of AD.

Footnotes

ACKNOWLEDGMENTS

This study was supported by the Danish Alzheimer Association Research Fund. The Danish Dementia Research Centre is supported by the Danish Ministry of Health.

Authors’ disclosures available online ().

References

Sperling

, Aisen

, Beckett

, Bennett

, Craft

, Fagan

, Iwatsubo

, Jack

, Kaye

, Montine

, Park

, Reiman

, Rowe

, Siemers

, Stern

, Yaffe

, Carrillo

, Thies

, Morrison-Bogorad

, Wagster

, Phelps

(2011) Toward defining the preclinical stages of Alzheimer’s disease: Recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimers Dement 7, 280.

Jessen

, Amariglio

, van Boxtel

, Breteler

, Ceccaldi

, Chetelat

, Dubois

, Dufouil

, Ellis

, van der Flier

, Glodzik

, van Harten

, de Leon

, McHugh

, Mielke

, Molinuevo

, Mosconi

, Osorio

, Perrotin

, Petersen

, Rabin

, Rami

, Reisberg

, Rentz

, Sachdev

, de la Sayette

, Saykin

, Scheltens

, Shulman

, Slavin

, Sperling

, Stewart

, Uspenskaya

, Vellas

, Visser

, Wagner

(2014) A conceptual framework for research on subjective cognitive decline in preclinical Alzheimer’s disease. Alzheimers Dement 10, 844–852.

Reisberg

, Shulman

, Torossian

, Leng

, Zhu

(2010) Outcome over seven years of healthy adults with and without subjective cognitive impairment. Alzheimers Dement 6, 11–24.

Jessen

, Amariglio

, Buckley

, van der Flier

, Han

, Molinuevo

, Rabin

, Rentz

, Rodriguez-Gomez

, Saykin

, Sikkes

SAM

, Smart

, Wolfsgruber

, Wagner

(2020) The characterisation of subjective cognitive decline. Lancet Neurol 19, 271–278.

Pereiro

, Valladares-Rodriguez

, Felpete

, Lojo-Seoane

, Campos-Magdaleno

, Mallo

, Facal

, Anido-Rifón

, Belleville

, Juncos-Rabadán

(2021) Relevance of complaint severity in predicting the progression of subjective cognitive decline and mild cognitive impairment: A machine learning approach. J Alzheimers Dis 82, 1229–1242.

Pavisic

, Lu

, Keuss

, James

, Lane

, Parker

, Keshavan

, Buchanan

, Murray-Smith

, Cash

, Coath

, Wong

, Fox

, Crutch

, Richards

, Schott

(2021) Subjective cognitive complaints at age 70: Associations with amyloid and mental health. J Neurol Neurosurg Psychiatry 92, 1215–1221.

Buckley

, Saling

, Irish

, Ames

, Rowe

, Lautenschlager

, Maruff

, Macaulay

, Martins

, Masters

, Rainey-Smith

, Rembach

, Savage

, Szoeke

, Ellis

(2014) Personal memory function in mild cognitive impairment and subjective memory complaints: Results from the Australian Imaging, Biomarkers, and Lifestyle (AIBL) Study of Ageing. J Alzheimers Dis 40, 551–561.

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.

Donix

, Brons

, Jurjanz

, Poettrich

, Winiecki

, Holthoff

(2010) Overgenerality of autobiographical memory in people with amnestic mild cognitive impairment and early Alzheimer’s disease. Arch Clin Neuropsychol 25, 22–27.

10.

Svoboda

, McKinnon

, Levine

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

11.

Warren

, Hermiller

, Nilakantan

, O’Neil

, Palumbo

, Voss

(2018) Increased fMRI activity correlations in autobiographical memory versus resting states. Hum Brain Mapp 39, 4312–4321.

12.

Philippi

, Tranel

, Duff

, Rudrauf

(2015) Damage to the default mode network disrupts autobiographical memory retrieval. Soc Cogn Affect Neurosci 10, 318–326.

13.

Hampton

, Buckley

, Manning

, Scott

, Properzi

, Peña-Gómez

, Jacobs

HIL

, Chhatwal

, Johnson

, Sperling

, Schultz

(2020) Resting-state functional connectivity and amyloid burden influence longitudinal cortical thinning in the default mode network in preclinical Alzheimer’s disease. Neuroimage Clin 28, 102407.

14.

Buckner

, Snyder

, Shannon

, Larossa

, Sachs

, Fotenos

, Sheline

, Klunk

, Mathis

, Morris

, Mintun

(2005) Molecular, structural, and functional characterization of Alzheimer’s disease: Evidence for a relationship between default activity, amyloid, and memory. J Neurosci 25, 7709.

15.

Xue

, Yuan

, Yue

, Xu

, Wang

, Wu

, Ji

, Zhou

, Zhao

, Rao

, Yang

, Xiao

, Chen

(2019) Distinct disruptive patterns of default mode subnetwork connectivity across the of spectrum preclinical Alzheimer’s disease. Front Aging Neurosci 11, 307.

16.

Grilli

, Wank

, Bercel

, Ryan

(2018) Evidence for reduced autobiographical memory episodic specificity in cognitively normal middle-aged and older individuals at increased risk for Alzheimer’s disease dementia. J Int Neuropsychol Soc 24, 1073–1083.

17.

El Haj

, Antoine

, Nandrino

, Kapogiannis

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

18.

Buckley

, Saling

, Irish

, Ames

, Rowe

, Villemagne

, Lautenschlager

, Maruff

, Macaulay

, Martins

, Szoeke

, Masters

, Rainey-Smith

, Rembach

, Savage

, Ellis

(2014) Autobiographical narratives relate to Alzheimer’s disease biomarkers in older adults. Int Psychogeriatr 26, 1737–1746.

19.

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.

20.

Philippi

, Noblet

, Botzung

, Despres

, Renard

, Sfikas

, Cretin

, Kremer

, Manning

, Blanc

(2012) MRI-based volumetry correlates of autobiographical memory in Alzheimer’s disease. PLoS One 7, e46200.

21.

Meulenbroek

, Rijpkema

, Kessels

RPC

, Rikkert

MGMO

, Fernández

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

22.

Stokholm

, Vogel

, Johannsen

, Waldemar

(2009) Validation of the Danish Addenbrooke’s Cognitive Examination as a screening test in a memory clinic. Dement Geriatr Cogn Disord 27, 361–365.

23.

Vogel

, Stokholm

, Andreasen

, Henriksen

, Brønniche

, Madsen

, Gustafsson

, Overgaard

, Guldberg

, Jørgensen

(2018) Psychometric properties and reference data for Danish versions of Free and Cued Selective Reminding Test, Category Cued Memory Test and Logical Memory. Scand J Psychol 59, 496–502.

24.

Vogel

, Johannsen

, Stokholm

, Jørgensen

(2014) Frequency and severity of semantic deficits in a consecutive memory clinic cohort. Dement Geriatr Cogn Disord 38, 214–223.

25.

Kaplan

, Goodglass

, Weintraub

(1983) The Boston Naming Test, Lea & Febiger, Philadelphia.

26.

Regard

, Strauss

, Knapp

(1982) Children’s production on verbal and non-verbal fluency tasks. Percept Mot Skills 55, 839–844.

27.

Sheikh

, Yesavage

(1986) Geriatric Depression Scale (GDS): Recent evidence and development of a shorter version. Clin Gerontol 5, 165–173.

28.

Albert

, DeKosky

, Dickson

, Dubois

, Feldman

, Fox

, Gamst

, Holtzman

, Jagust

, Petersen

, Snyder

, Carrillo

, Thies

, Phelps

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

29.

Pfeffer

, Kurosaki

, Harrah

Jr. , Chance

, Filos

(1982) Measurement of functional activities in older adults in the community. J Gerontol 37, 323–329.

30.

McKhann

, Knopman

, Chertkow

, Hyman

, Jack

Jr , Kawas

, Klunk

, Koroshetz

, Manly

, Mayeux

, Mohs

, Morris

, Rossor

, Scheltens

, Carrillo

, Thies

, Weintraub

, Phelps

(2011) The diagnosis of dementia due to Alzheimer’s disease: Recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimers Dement 7, 263.

31.

Saykin

, Wishart

, Rabin

, Santulli

, Flashman

, West

, McHugh

, Mamourian

(2006) Older adults with cognitive complaints show brain atrophy similar to that of amnestic MCI. Neurology 67, 834–842.

32.

Risacher

, Kim

, Nho

, Foroud

, Shen

, Petersen

, Jack

Jr. , Beckett

, Aisen

, Koeppe

, Jagust

, Shaw

, Trojanowski

, Weiner

, Saykin

, Initiative

AsDN

(2015) APOE effect on Alzheimer’s disease biomarkers in older adults with significant memory concern. Alzheimers Dement 11, 1417–1429.

33.

El Haj

, Antoine

(2017) Discrepancy between subjective autobiographical reliving and objective recall: The past as seen by Alzheimer’s disease patients. Conscious Cogn 49, 110–116.

34.

Martinelli

, Anssens

, Sperduti

, Piolino

(2013) The influence of normal aging and Alzheimer’s disease in autobiographical memory highly related to the self. Neuropsychology 27, 69–78.

35.

El Haj

, Antoine

, Kapogiannis

(2015) Flexibility decline contributes to similarity of past and future thinking in Alzheimer’s disease. Hippocampus 25, 1447–1455.

36.

El Haj

, Antoine

, Kapogiannis

(2015) Similarity between remembering the past and imagining the future in Alzheimer’s disease: Implication of episodic memory. Neuropsychologia 66, 119–125.

37.

Piolino

, Desgranges

, Benali

, Eustache

(2002) Episodic and semantic remote autobiographical memory in ageing. Memory 10, 239–257.

38.

Kopelman

, Wilson

, Baddeley

(1989) The autobiographical memory interview: A new assessment of autobiographical and personal semantic memory in amnesic patients. J Clin Exp Neuropsychol 11, 724–744.

39.

El Haj

, Antoine

(2017) Describe yourself to improve your autobiographical memory: A study in Alzheimer’s disease. Cortex 88, 165–172.

40.

Ivanoiu

, Cooper

, Shanks

, Venneri

(2006) Patterns of impairment in autobiographical memory in the degenerative dementias constrain models of memory. Neuropsychologia 44, 1936–1955.

41.

McElhiney

, Moody

, Sackheim

(2001) The Autobiographical Memory Interview - Short Form, Department of Biological Psychiatry, New York State Psychiatric Institute, & Department of Psychiatry, College of Physicians and Surgeons, Columbia University, New York.

42.

Semkovska

, Noone

, Carton

, McLoughlin

(2012) Measuring consistency of autobiographical memory recall in depression. Psychiatry Res 197, 41–48.

43.

Fritz

, Morris

, Richler

(2011) Effect size estimates: Current use, calculations, and interpretation. J Exp Psychol Gen 141, 2–18.

44.

Irish

, Hornberger

, Lah

, Miller

, Pengas

, Nestor

, Hodges

, Piguet

(2011) Profiles of recent autobiographical memory retrieval in semantic dementia, behavioural-variant frontotemporal dementia, and Alzheimer’s disease. Neuropsychologia 49, 2694–2702.

45.

Joubert

, Gardy

, Didic

, Rouleau

, Barbeau

(2021) A meta-analysis of semantic memory in mild cognitive impairment. Neuropsychol Rev 31, 221–232.

46.

Vogel

, Gade

, Stokholm

, Waldemar

(2005) Semantic memory impairment in the earliest phases of Alzheimer’s disease. Dement Geriatr Cogn Disord 19, 75–81.

47.

Murphy

, Troyer

, Levine

, Moscovitch

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

48.

Greene

, Hodges

, Baddeley

(1995) Autobiographical memory and executive function in early dementia of Alzheimer type. Neuropsychologia 33, 1647–1670.

49.

Conway

(2005) Memory and the self. J Mem Lang 53, 594–628.

50.

Conway

(2001) Sensory–perceptual episodic memory and its context: Autobiographical memory. Philos Trans R Soc Lond B Biol Sci 356, 1375–1384.

51.

Palombo

, Williams

, Abdi

, Levine

(2013) The survey of autobiographical memory (SAM): A novel measure of trait mnemonics in everyday life. Cortex 49, 1526–1540.

52.

Maguire

(2001) Neuroimaging studies of autobiographical event memory. Philos Trans R Soc Lond B Biol Sci 356, 1441–1451.

53.

Rubin

, Siegler

(2004) Facets of personality and the phenomenology of autobiographical memory. Appl Cogn Psychol 18, 913–930.

54.

Williams

JMG

, Barnhofer

, Crane

, Hermans

, Raes

, Watkins

, Dalgleish

(2007) Autobiographical memory specificity and emotional disorder. Psychol Bull 133, 122–148.

55.

Compère

, Charron

, Gallarda

, Rari

, Lion

, Nys

, Anssens

, Coussinoux

, Machefaux

, Oppenheim

, Piolino

(2021) Gender identity better than sex explains individual differences in episodic and semantic components of autobiographical memory: An fMRI study. Neuroimage 225, 117507.