Abstract
Background:
Prospective memory (PM) refers to a future-oriented form of memory in which the individual must remember to execute an intended action either at a future point in time (Time-based) or in response to a specific event (Event-based). Lapses in PM are commonly exhibited in neurodegenerative disorders including Alzheimer’s disease (AD) and frontotemporal dementia (FTD), however, the neurocognitive mechanisms driving these deficits remain unknown.
Objective:
To investigate the clinical and neural correlates of Time- and Event-based PM disruption in AD and the behavioral-variant FTD (bvFTD).
Methods:
Twelve AD, 12 bvFTD, and 12 healthy older Control participants completed a modified version of the Cambridge Prospective Memory test, which examines Time- and Event-based aspects of PM. All participants completed a standard neuropsychological assessment and underwent whole-brain structural MRI.
Results:
AD and bvFTD patients displayed striking impairments across Time- and Event-based PM relative to Controls, however, Time-based PM was disproportionately affected in the AD group. Episodic memory dysfunction and hippocampal atrophy were found to correlate strongly with PM integrity in both patient groups, however, dissociable neural substrates were also evident for PM performance across dementia syndromes.
Conclusion:
Our study reveals the multifaceted nature of PM dysfunction in neurodegenerative disorders, and suggests common and dissociable neurocognitive mechanisms, which subtend these deficits in each patient group. Future studies of PM disturbance in dementia syndromes will be crucial for the development of successful interventions to improve functional independence in the patient’s daily life.
Keywords
INTRODUCTION
Episodic memory dysfunction represents one of the most prominent and characteristic features of Alzheimer’s disease (AD) [1], widely interpreted as reflecting the degeneration of the medial temporal lobes, most notably the hippocampus [2]. Episodic memory impairments in AD are cross-modal extending across standard neuropsychological tests of visual and verbal recall [3] and include the retrieval of personally relevant autobiographical memories from the past [4, 5]. While the hippocampus has been ascribed a central role in the origin of memory deficits in AD, the importance of regions beyond the medial temporal lobes in supporting successful episodic memory functioning is recognized. Notably, atrophy in parietal regions, such as the posterior cingulate cortex, and prefrontal cortices also contributes to episodic memory dysfunction in AD [6, 7].
The behavioral-variant of frontotemporal dementia (bvFTD) is a form of younger-onset dementia characterized clinically by progressive deterioration in interpersonal conduct and personality, executive dysfunction, and emotion dysregulation [8]. These features are attributable to the degeneration of orbitomesial frontal and anterior temporal lobe regions in the brain, particularly the anterior cingulate cortex and the frontoinsular cortex [9, 10]. Mounting evidence, however, reveals marked episodic memory impairments in bvFTD, at a level comparable to that observed in disease-matched cases of AD [6, 11]. These memory deficits are attributable to the degeneration of predominantly medial and lateral prefrontal, and medial temporal lobe structures including the hippocampus [6]. Thus while episodic memory appears comparably affected in bvFTD and AD, the underlying neural substrates of these deficits diverge contingent on dementia subtype [6, 7].
Prospective memory (PM) refers to a sub-branch of episodic memory whereby an individual must remember to perform a planned action or intention at some point in the future [12] and can be fractionated into Time- versus Event-based components depending on the type of cue which prompts recall of the planned activity. Time-based PM involves remembering to do something at a particular time or following a specific period of time whereas Event-based PM involves remembering to do something in response to an external cue. The ability to remember to do something at a later point in time allows us to think and plan beyond the present moment and is essential for successful everyday adaptive functioning. Accordingly, PMdysfunction considerably limits the capacity to function independently and is associated with a decline in independent activities of daily living, difficulties with financial management, and poor medication adherence [13, 14].
Successful PM performance draws upon multiple cognitive processes, including retrospective memory, planning, inhibition, task switching, and sustained attention or monitoring [15, 16]. This complexity is reflected at the neural level. Prefrontal regions, including the rostral prefrontal cortex (BA10), consistently activate during PM performance in healthy individuals potentially reflecting intention maintenance [17, 18]. Lesions to prefrontal regions have been shown to significantly compromise PM function [19, 20], corroborating the neuroimaging findings. Regions beyond the frontal lobes are also implicated in successful PM function. For example, parietal activation may support allocation of attention toward external stimuli and the PM intention, as well as maintenance and retrieval of the intention itself [18]. Moreover, medial temporal lobe regions, including the hippocampus, are implicated in the retrieval of an action or recognition of a cue during PM tasks [21, 22]. Finally human lesion studies [19] and functional neuroimaging studies [21, 23] point to occipital involvement, with the suggestion that representations of the PM cue may be stored in occipital sites [23]. As such, the evidence to date reveals a distributed set of brain regions which support the capacity for PM performance.
Age-related PM decline is observed across both Time- and Event-based components [24]. Notably, however, Time-based PM appears disproportionately disrupted relative to Event-based tasks in healthy older adults [24], suggesting that Time-based PM is a more cognitively difficult endeavor, potentially due to its greater dependence on self-initiated retrieval processes [24]. Compromised PM performance represents one of the most pervasive forms of memory disturbance in dementia and is viewed by caregivers as more disruptive than retrospective memory failures [25]. PM dysfunction is consistently demonstrated irrespective of dementia syndrome, with difficulties reported in Parkinson’s disease patients with cognitive impairment [26] and in amnestic and dysexecutive subtypes of mild cognitive impairment [27]. In addition, while impairments of retrospective episodic memory are considered to be a characteristic feature of AD [1], a number of studies suggest that PM failures occur at least as frequently in this population and are as severe [25]. Given that PM difficulties are evident inpreclinical AD and in very mild AD [28], substantial PM decline may be an early indicator of dementia [29].
To date the vast majority of studies of PM have focused on its disruption in MCI and AD [reviewed by 30], however, a recent study demonstrated marked PM dysfunction in bvFTD. Striking deficits were observed across Time- and Event-based forms of PM, with bvFTD patients scoring in line with disease-matched cases of AD [31]. These PM impairments were suggested to reflect potentially divergent underlying cognitive processes in each patient group; however, the underlying neural substrates of these deficits were not explored.
The present study represents the first investigation of the neural correlates of PM disruption in AD and bvFTD syndromes. Given converging evidence pointing to dissociable neural substrates of episodic memory disruption in AD versus bvFTD [6, 7], we predicted that unique neuroanatomical correlates of PM dysfunction would emerge in each patient group. Specifically, we hypothesized that a distributed network of frontal, medial temporal, and parietal regions would be implicated in AD, whereas predominantly anteromedial temporal and prefrontal regions would be involved in bvFTD. Importantly, given mounting evidence pointing to the involvement of the hippocampus in the genesis of episodic memory dysfunction in bvFTD [6, 32], we predicted that the hippocampus would represent a common neural correlate across the two disease syndromes.
METHODS AND MATERIALS
Participants
A total of 36 subjects participated in this study: 12 with a clinical diagnosis of AD, 12 with a diagnosis of clinically probable bvFTD, and 12 healthy older Control participants, all recruited through FRONTIER at Neuroscience Research Australia (NeuRA), Sydney. Dementia patients met current clinical diagnostic criteria for AD [1] or bvFTD [8]. Diagnosis was established by consensus among a senior neurologist (JRH), neuropsychologist, and occupational therapist based on extensive clinical investigations, detailed cognitive assessment, carer interviews, and evidence of atrophy on structural neuroimaging. Briefly, AD cases displayed significant episodic memory loss in the context of preserved socioemotional functioning and comportment. In contrast, bvFTD patients presented with an insidious decline in personality and socioemotional functioning accompanied by loss of insight and motivation. In addition, all bvFTD patients had normal spoken language; cases with mixed bvFTD/primary progressive aphasia presentation were excluded. Only dementia patients with evidence of definitive progression over time as revealed by atrophy on MRI scans, and information from carer reports, were included. Patients were assessed as part of their routine clinical visit, with PM testing typically taking place on the second day of a two-day visit.
Healthy Controls were recruited through the NeuRA research volunteer panel and local community groups. All Controls scored 0 on the Clinical Dementia Rating Scale [CDR; 33] and 88 or above on the Addenbrooke’s Cognitive Examination Revised [ACE-R; 34]. Exclusion criteria for all participants included prior history of mental illness, movement disorders, significant head injury, cerebrovascular disease, alcohol or other drug abuse, and limited English proficiency. Ethical approval for this study was obtained from the South Eastern Sydney Local Health District and the University of New South Wales ethics committees. All participants, or their person responsible, provided informed consent in accordance with the Declaration of Helsinki.
General cognitive screening
Participants were assessed across the following neuropsychological tests: ACE-R to establish overall level of cognitive functioning [34]; verbal letter fluency [F,A,S; 35]; the Rey Auditory Verbal Learning Task to assess verbal episodic encoding and retrieval [RAVLT; 36]; the Rey Complex Figure as an index of non-verbal episodic delayed recall [37]; the Trail Making test [Parts A and B; 38] to measure set-switching and divided attention; and Digit Span Backwards, from the Wechsler Adult Intelligence Scale, to assess attention/short-term working memory [39]. Verbal semantic performance was assessed using the Naming and Comprehension subscales of the SydBat [40], and the Hayling test [41] was used as an index of response inhibition. Carers rated memory and motivation changes of patients on the Cambridge Behavioural Inventory [CBI; 42]. Finally, the functional status of patients was determined using the Frontotemporal Dementia Functional Rating Scale [FRS; 43], which is a dementia staging tool sensitive to changes in functional abilities and presence of neuropsychiatric symptomatology.
Assessment of prospective memory
The procedure for this study has been described in detail elsewhere [31]. Briefly, a modified version of the Cambridge Prospective Memory Test was used to examine Time- and Event-based components of PM. This shorter version of the task instructed participants to complete three Time-based and three Event-based tasks while engaged in a filler task (viewing and describing humorous cartoons). The Time-based tasks required participants to execute an intended action after a specified amount of time had elapsed (e.g., request a pencil from the experimenter after 10 minutes) while the Event-based tasks required participants to execute an intended action following the occurrence of specific event (e.g., put the notebook on the floor when the alarm sounds). Instructions were read aloud verbatim from a standard script at the beginning of the session. These instructions were provided in the same order for all participants, with the Time-based PM instructions read first, as per Kamminga et al. [31]. If necessary, an instruction was repeated, but no instruction was provided more than twice. Participants were encouraged to use a pencil and paper to assist them with remembering these instructions in order to reflect the access to memory assistance tools which exist outside of an experimental context and to reduce the working memory demands of the task. All the requisite materials for the PM task, including materials for noting the task instructions and a clock to monitor the time, were in full view of the participant for the duration of the task. Full test instructions are provided in the Supplementary Material. The entire task took approximately20 minutes to complete.
Scoring of prospective memory task
PM performance was measured via two outcome scores: Time-based and Event-based PM. As outlined above, each PM subscale comprised three items, each of which contained two components: (i) the action to be executed, (ii) the cue, following which the action should be completed. For Time-based items, the cue represented a set amount of time (e.g., 5 minutes), while for Event-based items, the cue was the occurrence of a specific event (e.g., the alarm ringing). Each item was awarded a maximum of 2 points: participants were given one point for successful execution of the correct action, and one point for performing any action within 3 minutes of the specified cue. In addition, a further point was awarded to each subscale score if the participant used the provided external aids (pencil and paper) to take note of the task instructions. Thus, performance scores for each subscale ranged from 0–7. For each outcome measure, higher scores represented better PM performance.
Statistical analyses
Cognitive data were analyzed using IBM SPSS Statistics (Version 22.0). Multivariate analyses of variance (MANOVA) with Sidak post hoc tests were used to explore main effects of Group (Controls, AD, bvFTD) for all general cognitive tests. PM scores were expressed as a percentage of the maximum score for each PM subscale to increase the overall variation in scores. One overall MANOVA was used to analyze differences in Time- and Event-based PM task performance across the two groups. Paired-samplet-tests were run to investigate within-group differences between Time- and Event-based PM performance. A series of MANCOVAs were subsequently run to determine the contribution of discrete cognitive processes of interest (e.g., general cognitive function, language, motivation, episodic memory). Pearson correlations between PM performance and background neuropsychological variables were also investigated. Chi-squared tests (χ 2 ), based on the frequency patterns of dichotomous variables (e.g., gender), were used where appropriate.
Image acquisition
Participants underwent whole-brain T1-weighted images using a 3T Philips MRI scanner with standard quadrature head coil (8 channels). The 3D T1-weighted images were acquired using the following sequences: coronal orientation, matrix 256×256, 200 slices, 1×1 mm in-plane resolution, slice thickness 1 mm, echo time/repetition time = 2.6/5.8 ms, flip anglea = 19°.
Voxel-based morphometry analysis
Three-dimensional T1-weighted sequences were analyzed with FSL-VBM, a voxel-based morphometry analysis [44, 45] using the FSL-VBM toolbox from the FMRIB software package (http://fsl.fmrib.ox.ac.uk/fsl/fslwiki/FSLVBM/UserGuide) [46]. Briefly, brain-extracted structural images underwent tissue segmentation, and non-linear registration [47, 48] to align gray matter partial volumes to the Montreal Neurological Institute standard space (MNI152) using a b-spline representation of the registration warp field [49]. A study-specific template was created using the resulting images, to which the native gray matter images were re-registered nonlinearly. The registered partial volume maps were then modulated by dividing by the Jacobian of the warp field to correct for local expansion or contraction. Finally, the modulated segmented images were smoothed with an isotropic Gaussian kernel with a sigma of 3 mm.
An unbiased whole-brain general linear model was employed to investigate gray matter intensity differences via permutation-based non-parametric testing [50] with 5000 permutations per contrast. Differences in cortical gray matter intensities between patients and Controls were assessed using regression models with separate directional contrasts (i.e., t-tests). Clusters were extracted using the threshold-free cluster enhancement method (tfce) and corrected for Family Wise Error (FWE) at p < 0.05.
Covariate analyses
Correlations between the two scores of interest on the PM task (Time-based PM, Event-based PM) and regions of gray matter intensity were investigated in AD and bvFTD patients combined with Controls. An unbiased voxel-wise whole-brain approach was used across all covariate VBM analyses. For additional statistical power, a covariate-only general linear statistical model was employed. A positive t-contrast was used in the covariate model, providing an index of association between gray matter volume and PM scores. Two separate models were created to investigate the neural substrates of Time- and Event-based PM performance, with age included as a nuisance variable. Anatomical locations of significant results were overlaid on the MNI standard brain, with maximum coordinates provided in MNI stereotaxic space. Anatomical labels were determined with reference to the Harvard-Oxford probabilistic cortical atlas.
Overlap analyses were then conducted to identify common regions of gray matter intensity decrease implicated in (i) Time-based and (ii) Event-based PM disruption irrespective of patient diagnosis [see also 51]. The statistical maps generated from the Time- and Event-based covariate analyses were scaled using a threshold of p < 0.001, following which the scaled contrasts were multiplied to create an inclusive, or overlap, mask across groups for Time- and Event-based PM performance. To determine the unique contributions to PM performance in each patient group, an exclusive masking procedure was adopted, whereby each scaled image was subsequently multiplied by the inverse of the other scaled image to create an exclusive mask for each patient contrast (e.g. ADTimePM × inverse of BVTimePM). For the overlap and exclusive masking analyses, clusters were extracted at p < 0.001 uncorrected, using a strict cluster threshold of 100 contiguous voxels.
RESULTS
Demographics
The groups did not different significantly with regard to age (F(2, 33) = 2.539, p = 0.094), years of education completed (F(2, 33) = 1.476, p = 0.243), or gender distribution (χ2 = 0.225, p = 0.894). Further, the patient groups did not significantly differ with respect to disease duration (i.e., months elapsed since symptom onset) (F(1, 22) = 0.095, p = 0.761) (see Table 1).
General cognitive functioning
Neuropsychological test results are displayed in Table 1. In brief, group differences were evident for global cognitive functioning (F(2, 33) = 18.434, p < 0.0001), with both patient groups demonstrating significantly poorer performance on the ACE-R screening task relative to Controls (AD, p < 0.0001; bvFTD, p = 0.004). In addition, there was the suggestion that AD patients were more cognitively impaired than bvFTD patients, (p = 0.050). In contrast, bvFTD displayed significantly more behavioral disturbances relative to AD patients, as rated by caregivers on the CBI, in terms of abnormal behavior (t = –2.66, p = 0.014), stereotypical behavior (t = –2.50, p = 0.020) and apathy (t = –3.33, p = 0.003), however, ratings of changes in memory were not found to differ between the patient groups (t = 0.815, p = 0.424). Disease severity reflected in changes in one’s ability to complete everyday tasks and presence of behavioral symptoms (e.g., lack of interest in tasks, impulsivity), as measured by the FRS, was also significantly worse in bvFTD patients compared to AD patients (F(1, 20) = 10.501, p = 0.004).
Neuropsychological testing revealed cognitive profiles characteristic of AD and bvFTD (Table 1). Briefly, both patient groups displayed significant verbal episodic memory impairments relative to Controls on the RAVLT (F(2, 27) = 16.099, p < 0.0001; AD p < 0.0001; bvFTD, p < 0.0001) with no significant difference between the patient groups (p = 0.999). Similarly, visual episodic memory on the RCF was significantly compromised relative to Controls (F(2, 28) = 18.733, p < 0.0001; AD & bvFTD p < 0.0001) with comparable performance in the patient groups (p = 0.959). In contrast, visual episodic recognition memory on the Doors and People Test Part A was found to be relatively intact in AD (p = 0.071) and bvFTD (p = 0.135). A significant group effect was evident on the Trail Making Test Part A (F(2, 29) = 3.366 p = 0.048) driven by reduced speed of processing in AD (p = 0.057) but not in bvFTD (p = 0.896). Difficulties in set-shifting on the Trail Making Test Part B-A were also evident (F(2, 24) = 6.483 p = 0.006) again driven by the AD group (p = 0.004; bvFTD, p = 0.438). Working memory difficulties on the Digit Span Backwards task were found (F(2, 30) = 12.411, p < 0.0001) with deficits present in both AD (p < 0.0001) and bvFTD (p = 0.004) relative to Controls, but no significant difference between the patient groups (p = 0.646). Letter fluency was also compromised in the patient groups relative to Controls (F(2, 31) = 9.824, p < 0.0001; AD, p = 0.019; bvFTD p < 0.0001). Semantic processing impairments emerged on the Naming (F(2, 30) = 8.821, p = 0.001) and Comprehension (F(2, 30) = 8.145, p = 0.001) subscales of the SydBAT. While Naming was affected in both patient groups (AD, p = 0.001; bvFTD, p = 0.015), only AD patients displayed significant Comprehension impairments relative to Controls (AD, p = 0.001; bvFTD, p = 0.490). Finally, both patient groups displayed significant deficits in terms of response inhibition on the Hayling task (F(2, 27) = 14.379, p < 0.0001; AD, p = 0.005; bvFTD p < 0.0001).
Prospective memory performance
Time- and Event-based PM
Figure 1 displays the percentage correct scores for Time- and Event-based PM performance in the participant groups. A multivariate analysis of variance (MANOVA) revealed a significant main effect of group for Time- (F(2, 33) = 34.476, p < 0.0001) and Event-based (F(2,33) = 25.730, p < 0.0001) PM performance. Sidak post hoc tests revealed significant Time-based impairments in AD and bvFTD patients relative to Controls (all p values <0.0001), with AD patients disproportionately affected in comparison with their bvFTD counterparts (p = 0.002). For Event-based PM, both patient groups were impaired relative to Controls (all p values <0.0001) with comparable performance evident between the patient groups (p = 0.322).
Within-group comparisons revealed comparable performance across Time- and Event-based PM tasks in the AD (p = 0.293) and bvFTD (p = 0.420) patient groups, however, a difference on the threshold of significance was observed in Controls with the suggestion that Event-based performance was higher relative to Time-based PM (p = 0.054).
To ensure that PM performance was not predominantly driven by overall level of cognitive functioning, we repeated the above analyses controlling for ACE-R total cognitive scores. The same pattern of results was evident with marked deficits for Time-based PM in both AD (p < 0.0001) and bvFTD (p = 0.015) relative to Controls, with AD patients continuing to score significantly lower than the bvFTD group (p = 0.030). Similarly, both patient groups showed marked Event-based PM impairments relative to Controls (AD: p = 0.004; bvFTD: p = 0.003) with no differences between the patient groups (p = 0.843). Finally, comparable PM performance was observed on the Time- and Event-based subscales across the participant groups (all p values >0.2). As such, PM dysfunction in AD and bvFTD does not appear to be mediated by a general decline in cognitive functioning.
Correlations between PM performance and background neuropsychological tests
Table 2 displays the correlations between Time- and Event-based PM performance and neuropsychological variables in the participant groups. Pearson R correlations revealed significant associations between verbal episodic memory integrity and Time-based PM performance (r = 0.741, p = 0.006) in bvFTD. In contrast, Event-based PM performance was found to strongly correlate with episodic memory integrity in AD (RCF: r = 0.868, p = 0.002; RAVLT: r = 0.849, p = 0.016) and in Controls (RCF: r = 0.653, p = 0.029). No other significant relationships were evident across any of the other neuropsychological tests of global cognitive functioning, attention, executive function, language, or motivation (all p values >0.07).
Analyses of covariance
To determine whether distinct cognitive processes play a modulating role in PM performance in dementia, we conducted a series of analyses of covariance (ANCOVA) using the following variables: SydBAT Comprehension to assess language and semantic processing, delayed recall on the RAVLT and delayed recall on the RCF as indices of verbal and visual episodic memory processes, respectively, and CBI Motivation to measure levels of apathy.
Despite controlling for general language and comprehension performance on the SydBAT, group impairments continued to persist across both Time- (F(2, 29) = 21.014 p < 0.0001; AD: p < 0.0001; bvFTD: p = 0.003) and Event-based (F(2, 29) = 13.917, p < 0.0001; AD: p < 0.0001; bvFTD: p = 0.001) PM subscales. Notably, however, when we controlled for verbal episodic delayed recall on the RAVLT, overall group differences persisted for Time-based PM (F(2, 26) = 7.230, p = 0.003), however, these differences were driven exclusively by the AD group (p = 0.005; bvFTD: p = 0.423). In contrast, for Event-based PM, controlling for verbal episodic memory processes served to ameliorate the overall group effect (F(2, 26) = 2.882, p = 0.074) bringing both patient groups in line with Control performance (AD: p = 0.072; bvFTD: p = 0.207). Controlling for visual delayed episodic recall on the RCF failed to negate the overall group effect for either Time-based (F(2, 27) = 8.827, p = 0.001) or Event-based (F(2, 27) = 4.071, p = 0.028) PM, as AD patients continued to show impairments irrespective of subscale relative to Controls (Time: p = 0.002; Event: p = 0.034). In contrast, bvFTD patients were now found to score in line with Controls (Time: p = 0.151; Event: p = 0.321).
Finally, we investigated the role of apathy in mediating PM performance and found that controlling for motivation levels on the CBI failed to negate the group effect for either Time- (F(2, 31) = 27.865, p < 0.0001) or Event-based PM (F(2, 31) = 15.871, p < 0.0001), however Sidak post hoc tests revealed a differential effect contingent on patient group. For Time-based PM, group differences continued to persist in both AD (p < 0.0001) and bvFTD (p = 0.022), however, controlling for apathy negated the Event-based PM deficit in bvFTD (p = 0.213), but not in AD (p < 0.0001).
As such, these analyses suggest that decline in episodic memory processes may, in part, mediate Time- and Event-based PM dysfunction in bvFTD, with an additional role evident for levels of motivation in modulating Event-based PM. In AD, episodic memory disruption represents a candidate mechanism for Event-based PM impairments; however, the neurocognitive processes mediating Time-based PM deficits in AD remain unclear.
Voxel-based morphometry analyses
Gray matter atrophy in AD and bvFTD
Figure 2 displays the patterns of brain atrophy displayed by (A) AD and (B) bvFTD patients relative to Controls using the threshold free cluster enhancement method (TFCE) and corrected for Family-Wise Error (FWE) at p < 0.05. Briefly, AD patients displayed widespread neural atrophy involving the prefrontal, lateral and medial temporal regions including the bilateral hippocampi, as well as extending posteriorly to include significant parietal atrophy involving the bilateral supramarginal and angular gyri, bilateral occipital cortices, and the left precuneus. BvFTD patients showed pronounced changes in bilateral medial and lateral prefrontal regions including the orbitofrontal cortex and anterior cingulate cortex, extending into the insular cortices bilaterally, as well as lateral and medial temporal regions including the bilateral hippocampi. Posterior regions were significantly affected including the bilateral occipital cortices, and the right angular and supramarginal gyrus (Table 3). Direct comparisons of the patient groups failed to reveal significant clusters at the p < 0.05 FWE corrected threshold. These patterns of atrophy are consistent with previous reports in AD [52] and bvFTD [53].
Neural substrates of Time-based PM
Figure 3A displays the common and unique neural correlates of Time-based PM performance in AD and bvFTD. Regions commonly implicated in Time-based PM disruption, irrespective of group, included the bilateral hippocampi, bilateral amygdalae, as well as lateral temporal regions including the left temporal pole and left temporal fusiform cortex (see Table 4). Exclusive masking analyses revealed a distributed network of regions implicated in Time-based PM disruption in AD, including the bilateral orbitofrontal cortices and frontal poles, bilateral temporal poles, bilateral hippocampi, bilateral occipital cortices, and the left angular gyrus. In contrast, regions exclusively implicated for Time-based PM disruption in bvFTD were centered on the left frontal and left anteromedial temporal lobes including the temporal fusiform cortex, temporal pole, hippocampus, and orbitofrontal cortex (see Table 4). Neural substrates of Control Time-based PM performance are presented in the Supplementary Material.
Neural substrates of Event-based PM
Figure 3B displays the common and unique neural correlates of Event-based PM performance in AD and bvFTD patients. Our overlap analyses failed to reveal any significant overlap in the neural substrates of Event-based PM disruption across the patient groups. Regions exclusively implicated in AD included the right orbitofrontal cortex, left hippocampus, bilateral middle temporal gyrus and right temporal pole, extending posteriorly to include the left superior parietal lobule and bilateral angular gyrus, as well as the right precuneus, and left occipital pole. In contrast, for bvFTD patients, regions exclusively in the temporal lobes including notably the bilateral hippocampi and left lateral temporal cortices were implicated for Event-based PM disruption (see Table 5). Neural substrates of Control Event-based PM performance are presented in the Supplementary Material.
DISCUSSION
This study represents the first investigation of the neural correlates of PM dysfunction in the neurodegenerative disorders of AD and bvFTD. On a behavioral level, our findings reveal marked PM disruption across the patient groups, with Time-based PM disproportionately affected in AD relative to bvFTD participants. Notably, the integrity of episodic memory emerged as a key modulator of performance irrespective of PM subtype in bvFTD, and for Event-based PM disruption in AD. This observation was corroborated on the neural level with the hippocampus emerging as a common region implicated irrespective of group or PM task. Interestingly, levels of motivation were also found to modulate Event-based PM performance in bvFTD, while the mechanisms underpinning Time-based disruption in AD remain unclear. Our findings demonstrate that while commonalities exist in the neural substrates of PM dysfunction in AD and bvFTD, discrete neural systems underlie PM dysfunction in each group. Here, we discuss the implications of our findings for understanding PM dysfunction in dementia syndromes.
The present findings converge with a growing body of evidence pointing to significant episodic memory dysfunction in bvFTD extending across laboratory [6, 11], autobiographical memory [5] and sourcememory [54, 55] tasks. Notably our findings corroborate a previous study in which marked PM impairments were observed in bvFTD across both Time- and Event-based PM subscales [31]. Importantly, here we demonstrate that episodic memory processes significantly modulate the capacity for PM performance, irrespective of PM domain, as controlling for delayed episodic recall significantly ameliorated PM deficits in the bvFTD group. General levels of motivation, as measured on the CBI, were also found to significantly influence Event-based PM performance in bvFTD suggesting that the origin of Event-based disruption is multifactorial. As such, difficulties in retrieving the to-be-remembered information, coupled with increased levels of apathy characteristically seen in this syndrome, represent the chief candidate cognitive mechanisms driving PM impairments in bvFTD. The interplay between apathy and episodic memory processes in bvFTD remains underexplored and it will be important to determine how loss of motivation impacts the capacity to encode, store and retrieve new information in this syndrome.
Our proposal of a central role for episodic memory processes in PM disruption in bvFTD was confirmed on the neuroanatomical level with the hippocampus consistently implicated irrespective of PM subtask in this syndrome. Hippocampal atrophy is well documented in bvFTD [32] and has been shown to correlate robustly with episodic memory dysfunction in this group [6]. Accordingly, compromised PM performance in bvFTD may reflect disruption of the retrospective memory search processes required to retrieve the intended action, a process which is mediated by the hippocampus [56]. Notably, for Time-based PM performance, predominantly left-lateralized regions including the temporal fusiform cortex, parahippocampal gyrus, inferior temporal gyrus, and temporal pole were implicated, reflecting a predominantly anteromedial system, largely implicated in episodic memory performance, that is compromised in bvFTD [6].
PM performance is typically conceptualized as a frontally-mediated function [16], and impairments in this domain in bvFTD would ostensibly be viewed as relating to frontal lobe degeneration. Interestingly, we found significant involvement of the left orbitofrontal cortex albeit exclusively on the Time-based task in bvFTD. Lesions to the orbitofrontal cortex are associated with impulsivity and alterations in time perception [57] leading to a profound myopia for the future [58]. As such, Time-based PM deficits in bvFTD may, in part, stem from an impaired capacity to accurately perceive time, and indeed temporal processing difficulties are present in this disorder [59]. Time perception therefore represents a particularly interesting line of enquiry in light of recent findings revealing marked impairments in the ability to successfully project forwards to envisage the future in bvFTD [reviewed by 60]. Notably, future thinking deficits in bvFTD are attributable to atrophy in the right hippocampus and right frontopolar regions [61], underscoring the role of prefrontal and medial temporal lobe interactions in supporting future-oriented memory processes.
While the neural substrates of PM dysfunction in bvFTD resided predominantly in anteromedial brain regions, a more widespread network was implicated in the AD group. Unsurprisingly, AD patients displayed marked deficits irrespective of PM subscale, however, Time-based PM appeared disproportionately disrupted relative to the bvFTD group. Notably, this profile of PM deficits in AD did not merely reflect a general decline in cognitive functioning, as the same overall pattern of results persisted despite controlling for level of cognitive function using the ACE-R. Interestingly, Event-based PM deficits were ameliorated when we controlled for delayed verbal episodic retrieval processes, replicating our finding in the bvFTD group and pointing towards a common underlying cognitive mechanism driving Event-based PM disruption in both dementia subtypes.
The genesis of Time-based disruption in AD, however, remains unclear, as our covariate analyses failed to reveal a distinct modulator of this form of PM disruption. Disorientation to time is a hallmark clinical feature of AD [1] and it has been suggested that such temporal confusion relates to the degeneration of pathways connecting the hippocampus with posterior parietal structures [62]. Our voxel-based morphometry analyses support the involvement of the hippocampus and parietal regions, such as the angular gyrus, within a distributed network of other frontal, lateral temporal, and occipital regions. A number of studies point to the potential role of the frontal poles in supporting the capacity for time estimation [22, 63] and it is interesting in this regard that the bilateral frontal poles were also implicated in the AD group. Given that Time-based PM tasks require the individual to predict the future time point at which a response should be enacted, it has been proposed that frontopolar involvement might reflect the capacity to envisage one’s future behavior [22]. Notably, simulation of future events and information is grossly compromised in AD [64] with frontopolar regions strongly implicated in these deficits [65]. Accordingly, atrophy in frontopolar regions in AD may disrupt a time perception mechanism that is central to all forms of future-oriented thinking, including PM and future simulation. While we did not explore the contribution of time perception to PM performance in this study, systematic investigation of the interplay between these processes in the dementias is warranted. Further, we did not determine the capacity for patients to accurately read, or indeed remember, the specified times during the Time-based PM trials. Given that deficits in clock reading are prominent in AD [66], use of a clock as an external aid on PM tasks may not prove as effective as Event-based cues such as an alarm, culminating in the differential PM profiles observed here.
In keeping with previous studies of episodic memory dysfunction in AD [6, 7], we also found significant parietal involvement underpinning PM deficits in AD. Activation of parietal regions such as the angular gyrus and supramarginal gyrus, in concert with the prefrontal cortices, is consistently reported in functional neuroimaging studies of PM performance in healthy individuals [18]. These sites form part of a broader cognitive control network, which supports sustained attention and goal-directed cognition, as well as a range of cognitively demanding tasks such as planning and problem-solving [67]. Recent studies indicate that disruption of this frontoparietal cognitive control network in AD adversely impacts episodic memory performance [68]. Our finding of bilateral angular gyri involvement irrespective of PM subtype in AD is therefore notable. While the precise functions of the angular gyrus remain unclear, degeneration of this region in AD may disrupt the integration of information and imparting of meaning towards an intended action [69]. Our findings resonate with the conceptualization of PM as lying at the interface between the domains of memory and attention, with successful PM performance requiring a shift between external stimuli (e.g., cues) and representations of intentions stored in memory [18]. While the precise neurocognitive mechanisms driving Time-based PM deficits in AD remain unclear, our behavioral and neuroimaging analyses underscore the multifactorial nature of PM dysfunction in AD, likely reflecting the deterioration of episodic memory, compromised time perception mechanisms, and disruption of core cognitive control processes.
Finally, our neuroimaging analyses revealed the involvement of the lateral occipital cortices and occipital poles across Time- and Event-based forms of PM in AD. The exact role of occipital regions in PM, and indeed in general episodic memory, remains to be elucidated, however, it is notable that lesions involving occipital regions are associated with gross impairments in retrospective and prospective memory [70]. It has further been suggested that occipital cortical regions may support cue encoding during PM tasks [18, 23]. It will be important for future studies to delineate the contribution of these regions in order to understand how occipital atrophy in AD impacts PM performance.
The clinical implications of our findings are important when we consider the pervasive nature of PM disruption across dementia syndromes. PM is crucial for the successful execution of many everyday tasks essential for functional independence, for example, remembering to take medication, or to turn off the stove after cooking. In the context of dementia patients who are characterized by marked functional impairments, deficits in PM are particularly disquieting and may further contribute to decline in independent activities of daily living, difficulties with financial management, and poor medication adherence [13, reviewed by 14]. Our covariate analyses point to a central role for episodic memory processes in Time- and Event-based PM dysfunction in bvFTD, and Event-based PM in AD. Interestingly, during testing it was observed that the patients did typically respond to the occurrence of the Event-based cues, suggesting that our findings do not reflect a detection problem per se. For example, patients would note and respond to an alarm (one of the Event-based cues signaling that the planned action was to be executed) by pointing out that the alarm had gone off. Nevertheless, they did not give an indication that the alarm signified anything of importance. Thus, it appears that the significance of the cue is lost in dementia suggesting that Event-based impairments reflect, in part, a breakdown in the association between cue and action. Alternatively, increased levels of apathy in bvFTD may render the patient indifferent to such external cues.
A number of methodological limitations warrant consideration. Firstly, our sample size is relatively small, and it will be important to replicate our findings in a larger sample. The PM task we used in this context, although sensitive to PM deficits in AD and bvFTD, arguably lacks ecological validity in terms of the type of difficulties that dementia patients might experience in their everyday lives. Future studies of PM in dementia using more ecologically valid measures will be essential to clarify how the deficits we report here manifest in the everyday environment of the individual. In addition, while the PM task is highly sensitive to memory dysfunction in AD and bvFTD, it is important to note that the task affords low specificity in terms of diagnostic utility. The disproportionate disruption of Time-based PM in AD relative to bvFTD warrants further investigation, as the mechanisms driving this impairment in AD remain unclear. While the AD patients were more cognitively impaired relative to the bvFTD group on a number of cognitive screening tasks, covarying for overall levels of cognitive functioning did not alter our main findings. It may be, however, that the Time-based PM task is more cognitively demanding than the Event-based subscale, as healthy Controls tended to perform higher on the Event-based task. In addition, we did not counterbalance the order of Time- and Event-based conditions in this study and it remains unclear how the use of our distraction task (identifying humor from cartoon scenes) could potentially influence PM performance. These methodological issues speak to the complexity of PM assessment using traditional pencil and paper tasks, and we suggest that future studies exploring PM in dementia should ensure that Time- and Event-based PM subscales are matched in terms of task difficulty and scoring. Finally, we did not ask patients to describe the approach they took to complete the task, particularly whether they relied on any compensatory techniques, and they were not asked to rate their experience of PM disruption on this task or in their daily lives. It would be interesting to ascertain whether the striking PM deficits observed on the experimental task used here correlate with subjective reports of PM dysfunction, or whether loss of insight which emerges in dementia syndromes alters awareness of PM deficits.
In conclusion, this study represents the first investigation of the neural substrates of PM dysfunction in AD and bvFTD. Our findings, while preliminary, suggest that episodic memory dysfunction and loss of motivation contribute to PM deficits in bvFTD, with the hippocampus emerging as a key region in this process. In AD, episodic memory dysfunction underpins Event-based PM deficits however the processes driving Time-based PM disruption remain elusive and reflect the degeneration of a distributed network of regions in the brain. Our findings highlight the multifaceted nature of PM disturbance in dementia syndromes and point to the need for further work in this area to develop targeted remediation strategies to improve functional independence in the patient’s daily life.
Footnotes
ACKNOWLEDGMENTS
The authors are grateful to the patients and their families for their continued support of our research. This work was supported by an Alzheimer’s Australia Dementia Research Foundation Grant, and funding to ForeFront, a collaborative research group dedicated to the study of frontotemporal dementia and motor neuron disease, from the National Health and Medical Research Council (NHMRC) (APP1037746) and the Australian Research Council (ARC) Centre of Excellence in Cognition and its Disorders Memory Node (CE11000102); an NHMRC Career Development Fellowship (APP1022684 to OP); and an ARC Discovery Early Career Researcher Award (DE130100463 to MI).
