Neural correlates of episodic memory in adults with Down syndrome and Alzheimer’s disease

Down syndrome (DS) is the most frequent genetic form of intellectual disability [1] and is associated to an increased risk of developing early Alzheimer’s disease (AD) dementia due to the presence of an extra copy of the amyloid β precursor protein gene, which is coded on chromosome 21, and is thus overexpressed in people with DS [2].

Episodic memory refers to the ability to learn, store, and retrieve information about unique personal experiences that occur in daily life [3]. Episodic memory impairment is the hallmark symptom of typical Alzheimer’s disease (AD) in the general population and has been associated with early atrophy of medial temporal lobe structures [4]. As the disease progresses, AD pathology spreads to other cortical regions, including ventral and lateral temporal cortex, followed by parietal and frontal cortices [4, 5], affecting other cognitive domains. In DS, neurodegenerative changes due to AD follow a similar pattern as those described in sporadic AD, although it occurs at a younger age [6‐8]. Early longitudinal studies suggested executive dysfunction and behavioral and psychological symptoms of dementia as commonly observed symptoms during the early stages of AD in DS [9], although the identification of imaging and fluid biomarkers in the past 5 years has provided new opportunities to characterize the sequence of cognitive decline during the preclinical and prodromal stages of the disease in this population [10]. Recent research now indicates that individuals with DS have a similar clinical presentation to that of people with sporadic or autosomal dominant AD, more specifically, declines in episodic memory and attention measures early in the course of the disease [11, 12].

Although medial temporal lobe regions play a central role in episodic memory, other cortical regions are also necessary to accurately remember new information. More specifically, three different processes or stages have been described in episodic memory that are sustained by different brain networks: encoding, retrieval, and storage [13]. In the general population, these memory processes have been assessed with the Free and Cued Selective Reminding Test (FCSRT) through different subscores and have been associated with different patterns of brain atrophy in patients with mild AD. Specifically, encoding has been associated with atrophy in parietal and temporal regions, storage with atrophy in entorhinal and parahippocampal regions, and retrieval with atrophy in a widespread network encompassing frontal regions [13]. Similarly, Wolk et al. investigated the relationship between performance on different stages of the Rey Auditory Verbal Learning Test (RAVLT), a verbal episodic memory test, and structural brain measures in patients with mild AD from the general population, including immediate and delayed memory recall [14]. This study also found dissociable relationships between regional brain atrophy and different aspects of memory function, indicating that different cognitive processes support different stages of episodic memory performance.

In DS, episodic memory has been evaluated using adapted tests such as the modified Cued Recall Test (mCRT), which has demonstrated to be useful to detect memory decline due to early-stage AD [15] and, more recently, for the diagnosis of prodromal AD [16]. The mCRT has emerged as a promising cognitive indicator of transition to the preclinical and prodromal stage of AD in subjects with DS [11]. Furthermore, poor performance on this episodic memory test has recently been related to white matter degeneration prior to the onset of dementia in adults with DS [17]. The association between regional cortical atrophy and different stages of episodic memory performance in patients with DS and symptomatic AD has not yet been studied.

The aim of this study was to investigate the neural correlates of different episodic memory processes measured with the mCRT in a large cohort of adults with DS, with and without symptomatic AD. Understanding the anatomic basis of the processes underlying episodic memory has potential practical implications, as it may facilitate the integration of clinical and radiological information, helping increase the diagnostic accuracy of AD in this population.

To our knowledge, this is the first study that investigates the neuroanatomical correlates of different processes involved in episodic memory in a cohort of adults with DS. Using structural MRI and an adapted episodic memory test, the mCRT, we found different correlations between the subtests assessing the different episodic memory processes and cortical thickness in symptomatic AD, suggesting that different anatomical regions play unique roles at different stages of episodic memory processing.

Adults with DS and symptomatic AD showed a pattern of cortical atrophy in lateral parieto-temporo-occipital cortices and in the posterior cingulate and precuneus. This pattern of widespread posterior cortical thinning has been previously described in amyloid-positive individuals with DS [25] and is very similar to that observed in sporadic AD [26].

Relationships in symptomatic individuals are detailed hereafter. The correlations between episodic memory measures and cortical thickness extended well beyond the medial temporal lobe structures. The different mCRT subscores were associated with cortical atrophy in different regions. T1 immediate free recall, a score that reflects initial encoding processes [13], mainly involved frontal lobe areas. FIRS, a score used to assess immediate retrieval [13], was associated with a widespread cortical network encompassing lateral frontal, parietal, temporal, and occipital regions, bilaterally, while TIS, which reflects immediate storage [13], was associated with more posterior cortical atrophy, including regions in the lateral parietal, temporal and occipital cortices, precuneus, posterior cingulate cortex, and medial temporal lobe cortices. In contrast to the immediate memory trials, delayed memory performance was related to a less distributed pattern of cortical atrophy, predominantly involving lateral and medial temporal lobe regions. Importantly, these associations were only seen in symptomatic participants and not in asymptomatic, possibly reflecting the smaller dynamic range of the mCRT scores and the milder atrophy in asymptomatic individuals [16]. To our knowledge, no other studies have related cortical thickness to different episodic memory measures in asymptomatic individuals with DS. In the general population, some authors have attempted to correlate FCSRT performance with volumetric measures of the hippocampal formation or entorhinal cortex in healthy aging subjects, without finding significant associations [27, 28].

Episodic memory impairment is a hallmark symptom of sporadic AD and DS-related AD during the early and middle stages of the disease [29‐31] and has been traditionally associated with medial temporal lobe damage [32‐34]. In the last decades, a growing number of neuroimaging studies have demonstrated that episodic memory does not depend solely on medial temporal structures. Several cognitive processes, like language, working memory, semantic processing, attention, and executive functions, which rely upon distinct brain networks, are necessary for successful episodic memory [13, 14, 35, 36]. We found that performance on T1 of the mCRT was associated with cortical thickness in frontal lobe regions, but not in medial temporal lobe structures in subjects with symptomatic AD, thus indicating that initial encoding was more dependent on regions supporting cognitive functions such as attention, working memory, or executive functions. Ventrolateral and dorsolateral prefrontal cortex have been described to implement control processes that support memory encoding (e.g., selection processes that direct attention toward goal-relevant information or organizational processes necessary for optimal memory encoding, respectively) [37, 38]. Our results are in line with those reported by Wolk et al. in sporadic AD, who also found a lack of influence of medial temporal structures during early list learning trials in patients with mild AD from the general population [14].

The temporal pole consistently appeared to be related to immediate and delayed memory measures (with the exception of T1 immediate recall) in our study. In this sense, functional neuroimaging studies have related the anterior temporal lobes to semantic processing (mental representations of the meaning of words, objects, people, and factual information about the world) [39, 40]. The semantic organization of the information that has to be encoded allows for a deeper processing and thus facilitates posterior retrieval processes. Our results are again consistent with previous studies relating the atrophy of the temporal pole with poor immediate recall performance in patients with early-stage AD in the general population [14, 35].

In addition, episodic memory performance (particularly T3, FIRS and TIS immediate recall scores, and TDS) was related to cortical thickness in the lateral posterior parietal cortex (PPC). These areas have been implicated in episodic memory retrieval because of their role in attentional processes (attention to internal representations). The dorsal PPC is implicated in the voluntary orienting of attention to relevant stimuli based on retrieval goals (top-down attentional processes), while the ventral PPC responds to the detection of new behaviorally relevant stimuli which are outside the focus of attention defined by the dorsal PPC (bottom-up attentional processes) [41, 42]. Moreover, inferior parietal regions (together with regions within the inferior frontal gyrus) have been associated with the storage function of the phonological loop and the visuospatial sketchpad supporting speech-based and visuospatial working memory, respectively [43]. Working memory, which temporarily maintains and stores information, is crucial for the transfer of information into a long-term store. Other regions also reported to be supportive of working memory are the posterior superior temporal gyrus, the premotor cortex, and the dorsolateral prefrontal cortex [44], regions found to be strongly associated with immediate memory measures in our study.

Atrophy in medial temporal lobe (MTL) structures, in particular the entorhinal and parahippocampal cortices, was found to be related to TIS and TDS. Low scores on both measures reflect storage deficits (immediate and delayed storage, respectively). Many studies support the role of MTL regions in the acquisition and maintenance of episodic memory [3, 13, 36]. The association between TIS and atrophy of extrahippocampal MTL regions in the right hemisphere in our study is not in line with previous research using the FCSRT or the RAVLT in patients with sporadic AD [35, 36]. One possible explanation for these discrepancies could be the visual nature of the stimuli used in the present study, unlike the FCSRT used in the general population, which is based on the learning of verbal stimuli. Similarly, we found that immediate and delayed recall measures were also associated with atrophy in occipital regions. Ventral occipito-temporal regions have been related to visual object processing [45]. Poor performance on some mCRT measures, which is based on the recall of familiar line drawings, could partly be explained by damage to this visual processing pathway.

Performance on the TIS, and to a lesser extent the TDS, was also associated with posteromedial regions, namely the precuneus and the posterior cingulate cortex. Recent functional imaging studies have suggested that memory function is subserved by a set of distributed networks which include the medial temporal lobe system and a set of cortical regions collectively referred to as the default mode network (DMN). The regions of the DMN include the anteromedial prefrontal cortex, the posterior cingulate cortex, the precuneus, the angular gyrus, and the medial temporal lobe [46]. Some authors have demonstrated that early amyloid deposition in sporadic AD occurs in areas mainly located within the DMN [47] and that this network plays a key role orienting the focus of attention to stored representational knowledge [48]. In DS subjects, recent research found that reductions in DMN connectivity to posterior brain regions were linked to the presence of AD neuropathology in a pattern substantially similar to that seen in sporadic AD [49, 50], even in asymptomatic individuals. Nevertheless, the association between the effects of AD neuropathology on the DMN and performance on episodic memory measures in the DS population has not been described.

Our results have several implications. First, the pattern of cortical atrophy observed in individuals with DS and AD in our study is remarkably similar to that described in the general population with AD [26], indicating that AD in individuals with DS targets the same cortical regions affected in sporadic AD. Second, the cortical areas found to be related to episodic memory in our study are in line with those previously described in sporadic AD. Our results thus reinforce the similarities between AD in DS and sporadic AD. Third, our results also support the use of the mCRT as a potential neuropsychological marker for the diagnosis of AD in people with DS, and particularly the TIS and the TDS (measures reflecting storage deficits), as these scores correlated well with medial temporal lobe regions, areas considered to be the earliest affected by AD neuropathology.

Our study has several strengths, including the large sample size of subjects included in our study with symptomatic AD, as well as the fact that our participants come from a large population-based cohort of adults with DS.

We also acknowledge some limitations. First, this is a single-center study and did not include an age-matched control group of euploid individuals to test the neurodevelopmental abnormalities. Thus, our results need to be replicated in other cohorts to confirm the generalizability of our results. Second, as patients with prodromal AD and AD dementia were analyzed together, it has not been possible to characterize AD in its earliest stages. Future studies should include a more homogeneous sample of subjects with prodromal AD to better characterize the early stages of AD in this population.

Our results provide new evidence about the neuroanatomical correlates of the episodic memory deficits in adults with DS and symptomatic AD that largely corresponds with the findings described in sporadic AD. This study further supports the similarities of the AD pathophysiological process in sporadic AD and in DS.