The Early Presentation of Dementia in People with Down Syndrome: a Systematic Review of Longitudinal Studies

In the general population, mild cognitive impairment (MCI) is regarded as the prodromal stage of dementia, but it has been argued that, in the case of people with DS, the exact hierarchy of loss of function is very difficult to establish (Cosgrave et al. 2000). Out of the 15 studies identified here, which attempted to address this question, nine suggested that ‘frontal-like symptoms’ were the earliest sign of AD in this population. Only two studies reported memory as the earliest sign of dementia. Out of the remaining four studies, two reported no decline that could not be accounted for by normal ageing and two did not report the order in which domains of cognitive functioning were affected.

The studies reporting the progression of clinical signs of DSAD are summarized in Table 3 and further discussed in the following sections. It is important to note that the majority of studies included in this review (with the exception of Määttä et al. 2014) have used various screening tools, rather than a diagnosis made by a clinician. Thus, it is possible that discrepancies between results are due to inadequate classification of participants with or without dementia, as well as due to the use of different diagnostic methods in each individual study (Ballard et al. 2016).

DS is known be associated with impairments in working memory, specifically remembering information for short periods of time (Silverman 2007). It has been suggested that tasks that rely on the hippocampus (e.g. associations of items in space and time) are commonly impaired in this population, hence making difficult the measurement of decline in these domains as a marker of AD-related change. By taking into account the level of baseline functioning, researchers interested in memory decline in people with DS have mostly focused on verbal list learning and scene learning (Sabbagh and Edgin 2016).

Compared to the general population where episodic memory impairments characterize prodromal stages of AD in the majority of cases, only two of the longitudinal studies included in this review identified impairments in memory as one of the earliest symptoms of DSAD (Krinsky-McHale et al. 2002; Cosgrave et al. 2000). Krinsky-McHale et al. (2002) found that decline in memory function scores preceded the onset of global changes associated by the authors with the onset of dementia, as reported in informant interviews. Decline in memory function preceded more global changes by more than a full year and in some cases by up to 3 years. More specifically, verbal explicit memory was one of the first domains that were affected, with individuals’ scores on the Selective Reminding Test declining by 20% on two consecutive test sessions, The authors compared the results of participants with dementia to those of age- and IQ-matched non-demented individuals with DS and concluded that the dramatic declines observed in DSAD were distinguishable from normal ageing pattern. Interestingly, even in cases where participants were capable of performing the task, those in the early stages of dementia showed deficits in their capacity to encode and retrieve information from long-term memory.

This view was supported by a previous study in which memory loss was one of the first identifiable hallmarks of AD in an all-female sample of older adults with DS (Cosgrave et al. 2000). In this study, participants had difficulty remembering items that they had seen as part of a previous task (i.e. delayed memory) and difficulty remembering in which hand a paper clip was held by the examiner after both hands were placed behind the examiner’s back (i.e. immediate memory).

However, these findings were only applicable in the case of individuals with mild (IQ 55–70) and moderate ID (IQ 40–55), and memory problems were more difficult to identify in participants with more severe intellectual disabilities (severe = IQ 25–40, or profound = IQ < 25) due to floor effects on established tests. More specifically, participants with IQ under 40 had very low scores on memory tests regardless of dementia diagnosis and thus, these measures may not be suitable as an indicator of decline over time in this population. More recently, in the DSM-V, IQ test scores have been removed from the diagnostic criteria of ID (American Psychiatric Association 2013), emphasizing the need for a more comprehensive assessment in which severity of ID will be based on the level of adaptive functioning, rather than IQ, as the former determines the level of support required (Oakley et al. 2003).

In the study conducted by Devenny et al. (2000), authors suggested that decline does not occur globally but rather as a systematic and progressive loss of cognitive functions, with memory loss being a predominant symptom in all participants who also showed decline on the Block Design and Coding subtests of the WISC-R. These participants were classified as having “questionable decline”. However, the authors emphasize the fact that none of these participants had received an official diagnosis from a physician and thus advise caution when interpreting their results, as it is possible that the observed declines were related to some other undetected condition.

A different view, however, is that decline in memory scores (even in the presence of developmental memory impairment) are observed in all ageing individuals with DS, regardless of dementia status (Ball et al. 2006; Devenny et al. 2000; Carr and Collins 2014). It is possible that by examining alternative measures of decline in other cognitive domains, the generalized pattern of deterioration that accompanies old age can be distinguished from preclinical AD. In the study conducted by Ball et al. (2006), informant reported memory changes were not correlated with subsequent diagnosis of AD and all participants who showed changes in memory also showed BPSD. However, in people with DS, early changes in memory may be more difficult to notice compared to BPSD, the latter having a greater impact on the individual’s daily life (Adams and Oliver 2010). Certain learning and memory tasks such as cued-learning and recall tasks may prove to be of a greater sensitivity and specificity for distinguishing between AD and non-AD in this population (Benejam et al. 2015).

More recently, a review by Sabbagh and Edgin (2016), suggested that the earliest symptoms of dementia in individuals with DS are often subtle but are firstly evident in BPSD rather than episodic memory. They are thus arguing that the late diagnosis of DSAD is often due to the fact that current classification systems have their emphasis on memory impairments, having been modelled after the diagnostic criteria of dementia in the general population and are unfit for use in the DS population (Nieuwenhuis-Mark 2009).

Functional impairments in executive functioning (executive dysfunction), as referring to informant reported difficulties with various goal-directed behaviours, such as planning and attention, and also performance on specific cognitive tests, have been identified in the majority of selected studies as preceding memory problems in adults with DS. However, in the studies reviewed there was not obvious overlap in terms of specific areas of decline identifed by repective authors. In Ball et al. (2006), the most commonly reported changes were impairments in planning, attention and lack of foresight, while in another study (Adams and Oliver 2010), the most affected areas were working memory, agnosia, aphasia and apraxia, with cognitive deterioration not being solely attributable to memory issues. In the Ball et al. (2006) paper, a version of the CAMCOG neuropsychological test battery modified for use with people with DS was completed at baseline and follow up assessments, to provide a measure of decline in global cognitive function. To provide more specific information on the sequence of decline in “frontal lobe associated EF” over the 5 years prior to diagnosis, an additional measure labelled “EF and attention” was designed by combining CAMCOG scores for abstract thinking, attention-calculation, verbal fluency and the clock drawing item. However, impairments in planning and foresight were measured through informant reported changes, rather than a more objective measure. Given the variety of living situations of participants in this study [i.e. residential homes (82%), sheltered accommodation (4%), nursing homes (2%), with a parent orrelative (13%)], we advise caution when interpreting these results. The caution is based on research showing that informant reports differ both qualitatively and quantitatively when recorded from personnel in an institution compared to parents (Nieuwenhuis-Mark 2009). Moreover, it is possible that “lack of foresight” could be affected by intellectual ability and this should also be taken into consideration, as 41 out of 55 participants in this study had moderate or severe learning disability.

In one of the studies, executive dysfunction was investigated by looking at verbal intrusions as an indicator of disruption of inhibition control (Kitler et al. 2006). Responding with an irrelevant word during a task of verbal memory retrieval was found to be predictive of performance on two out of three memory tasks administered within the next three years. Therefore, the authors argued that verbal intrusions are an early sign of Alzheimer-related neuropathology, preceding declines in memory. Interestingly, middle aged participants with DS made more verbal intrusions at baseline compared to participants with unspecified ID (74% vs 44%), giving further support to the idea that executive functioning is more sensitive to decline in DS than in other populations.

Another component of EF was examined by Nelson et al. (2007), who identified the NBAP Pragnosia scale as useful in classifying 80% of all dementia cases. The authors argued that impaired pragmatic language function represents an aspect of cognitive decline, as this measure appears to be strongly correlated with the assessment of dementia status (DMR Cognitive Scale). These impairments appeared to be most common in younger subjects, suggesting decline in functions other than memory, early in the course of dementia. More recently however, researchers have argued that verbal scores do not show significant decrease over the course of 47 years (Carr and Collins 2014), but these results might represent a sample bias, as the stronger representation of verbally able women might have influenced the overall stability of verbal scores in the group.

However, the exact sequence in which subdomains of executive functioning are affected has not yet been established. This could be largely due to the heterogeneity of DS phenotypes. Devenny et al. (2000) attempted to answer this question by following up participants over a 10 year period and noting that scores which showed decline early in the disease process were those that required perception of abstract stimuli and visuo-motor coordination. As participants progressed from prodromal stages to early dementia, deficits in comprehension, measures of visuospatial organization and the working memory component of language were observed, followed by vocabulary, information and digit span tasks in the middle stages of AD. As decline associated with the development of AD seems to involve progressively more areas of cognition and the number of areas in which decline was observed of was correlated to the severity of dementia, the authors conclude that cognitive decline in AD in people with DS is not global, but follows a predictable sequence. Executive functioning seems, according to the evidence presented above, to be affected in pre-clinical or early stages of AD and thus, detecting these changes could prove to be helpful in predicting a later diagnosis of AD.

Behavioural and Psychological Symptoms of Dementia (BPSD) have been defined as a “heterogeneous range of psychological reactions, psychiatrics symptoms and behaviours resulting from the presence of dementia” (Finkel 2001). Recent literature has suggested that BPSD are reported in the prodromal and early stages of DSAD and might predict a more severe diagnosis. Often, BPSD are reported in the absence of informant observed functional memory decline (Ball et al. 2006), in contrast to the presentation of AD in the general population, where impairments in episodic memory are commonly the first reported changes.

The hypothesis that BPSD and executive dysfunction precede memory issues in the pre-clinical and early stages of AD in DS is relatively recent (Ball et al. 2006; Adams and Oliver 2010; Kitler et al. 2006; Nelson et al. 2007; Urv et al. 2008; Määttä et al. 2014; Holland et al. 2000). This systematic review indicates that there is a limited but significant amount of research supporting the hypothesis, indicating that people with DS exhibiting BPSD (i.e. those with a history of the onset of these behaviours as opposed to such behaviours being lifelong) are more likely to decline functionally and on testing have evidence of impaired executive functioning compared to those with no such changes (44% compared to 5.6% of those without BPSD). Later the impaired group are then more likely to exhibit memory problems and meet diagnostic criteria for dementia (Ball et al. 2006). Similarly, in the study conducted by Holland et al. (2000), younger people with DS, who did not yet meet the full criteria for a clinical diagnosis of AD, exhibited more ‘frontal-like symptoms’, while older people had higher rates of clinically diagnosed AD. These results were supported by Nelson et al. (2007), whose study, although not identifying any significant decline in function, due to the short duration between baseline and follow up, argued for the high reliability and validity of measures of “frontal-like symptoms” in the detection of AD early in the disease process, suggesting that EF and BPSD may indeed precede or accompany memory changes in this population

This view is also supported by the results of Holland et al. (2000). At the time of an initial assessment more participants were reportedly exhibiting changes in behaviour (46%) than changes in memory (9%), whereas at follow up. Older participants had the largest proportion of reported deteriorations in memory and personality, whereas younger participants showed high rates of BPSD in absence of any memory issues, suggesting a trend of decline from behaviour and personality to a combination of domains.

This sequential progression of BPSD between various stages of dementia was further examined in studies that reported correlations between the severity of BPSD and a dementia diagnosis (Määttä et al. 2014; Urv et al. 2008). If indeed BPSD are an indicator of dementia development, the question that arises is whether certain types of behaviours are more likely to develop. In Urv et al. (2008), participants who showed cognitive decline insufficient for a diagnosis of AD exhibited more BPSD directed towards others (i.e. destructiveness, aggression) than those with a diagnosis of AD, who exhibited more BPSD directed towards the self (i.e. fearfulness, lack of energy, withdrawal, sadness and self-injury). Interestingly, participants who progressed from no diagnosis to questionable dementia were more likely to experience a worsening of scores on certain measures (sadness, increased dependency, confusion, fearful and regressive behaviour and social inadequacy), while the scores of people who were dementia-free at T2 remained relatively stable on these measures (Urv et al. 2008). Contrary to these findings, Adams and Oliver (2010) and Makary et al. (2014) suggested a more positive pattern of ageing than previously reported.

Similarly to the situation described in the case of executive functioning, a specific pattern of deterioration in BPSD subdomains is difficult to identify. It was proposed that some of the earliest signs of dementia in people with DS are observed in language and socialization difficulties (Margallo-Lana et al. 2007).

Improving our understanding of BPSD and their relationship to AD is essential, given that there are people with DS with very low levels of cognitive functioning who are unable to be diagnosed by measures of cognition alone due to floor levels. This group could, however, be diagnosed on the basis of BPSD and everyday skills (Margallo-Lana et al. (2007). Similar conclusions were reached by Cosgrave et al. (2000), who concluded that even though in their study memory seems to be the earliest symptom, severe intellectual disability conceals the symptoms and thus, BPSD might be easier to detect

Informant reported BPSD such as social withdrawal, apathy (Ball et al. 2006; Holland et al. 2000), inactivity and lack of interest (Adams and Oliver 2010), have been reported in all stages of DSAD (Urv et al. 2008) and it is likely that they significantly affect activities of daily living, with declines having been reported in domestic activity, self-direction and responsibility (Määttä et al. 2014). However, the authors note that depression is one of the factors that correlate most strongly with such changes (Urv et al. 2008) and may often act as a confound due to diagnostic overshadowing (Määttä et al. 2014), alongside hypothyroidism. Because of this potential confound, caution is advised when discussing the role of BPSD in the progression of DSAD and these reported changes must always be analyzed as part of a wider context.

Changes such as restlessness, aggression, repetitive speech and being uncooperative were most often reported in pre-clinical and early-stages of AD and thus might represent some of the earliest observable signs of dementia (Adams and Oliver 2010). Observation of these signs seems to be more likely to lead to a referral for dementia assessment and were shown to emerge alongside cognitive decline. In contrast, Carr and Collins (2014) in her long-term follow-up of a birth cohort did not find any increase in aggression with the development of dementia.

Other BPSD, such as emotional lability, lack of concern for other people, stubbornness, disinhibition and impulsivity, were also reported prior to a diagnosis of dementia (Ball et al. 2006). Stubbornness was the second most reported BPSD after apathy in a sample of aging people with DS (Holland et al. 2000), and increases in BPSD were identified as successful predictors of dementia status. When analyzed alongside scores for pragnosia, BPSD successfully predicted over 70% of dementia cases (Nelson et al. 2007). Urv et al. (2008) suggests that these changes are present before a dementia diagnosis and gradually worsen as dementia progresses, until late stages of AD when the number of reported distinct behaviours is reduced.

This review highlights the importance of using a broad spectrum of assessments when examining the sequence of decline of dementia in people with DS, particularly in the early stages during which the presentation of dementia may differ between individuals. While there is no overall consensus, there are a majority of research studies identifying clinical characteristics, other than the development of functional memory changes that occur early in the course of AD. However, an evaluation of the papers included in this systematic review indicates that assessing the hierarchy of decline that accompanies the development of dementia in DS is a challenge. First, from a methodological perspective, change has been identified in two separate ways, one being a structured informant-based interview asking specifically about functional changes and the other being the use of established cognitive tests that can be repeated over time to established whether there has been a change in test scores or not. Secondly, as the diagnosis of AD represents a threshold effect, it is difficult to pinpoint in time the exact onset of symptoms. Thirdly, heterogeneity of symptoms seems to be characteristic of the early stages, and inferences regarding which subdomains of functioning are preferentially affected are very difficult to make. Fourthly, in people with DS it is especially difficult to determine whether the observed symptoms are due to age-related changes in this population or to preexisting cognitive impairments (Ball et al. 2006). In the longest longitudinal study included in this review, scores on the majority of cognitive tests declined by age 47, regardless of dementia status (Carr and Collins 2014), emphasizing that researchers must be cautious when interpreting results, as for a proportion of participants showing this decline it might suggest pre-clinical symptoms of dementia, it might also reflect normal ageing.

The findings from this systematic review would indicate that BPSD and impairments in executive functioning mark the early course of DSAD. Numerous components of executive function and behavior seem to be affected before memory skills begin to decline, suggesting that current diagnosis procedures of AD in DS may not be effective until later stages. This view is supported by a recent review (Ballard et al. 2016), which suggested that dementia in DS initially manifests as BPSD, only to later be followed by changes in cognition.

There are, however, important differences between study findings that may reflect genuine sample differences or may be accounted for by differences in methodology. These differences make an overall picture difficult to ascertain. BPSD and executive dysfunction were often investigated as part of studies primarily concerned with cognition whereas memory changes were reported as part of studies that were carried out with the intention of confirming the presence of memory decline. Moreover, some studies only analyzed a restricted range of abilities and reported on total scores alone, without specifying the exact area of decline. Given the wide variety of screening and diagnosis tools in use, as well as recent research suggesting decline in cognitive function regardless of whether participants have had a formal diagnosis of dementia, we support the recent suggestion made by Ballard et al. (2016), that measures of prevalence reported in published papers should be interpreted with a degree of caution.

Moreover, as with all neuropsychological testing, there is overlap in measurement. For example, a decline in a test of episodic memory would have to be carefully interpreted if there are poor results on attentional tasks. Due to this confounding factor, we advise caution when discussing the results of these papers, as it often is difficult to establish whether a participant’s decline on one measure is not influenced by decline in another related domain.

Lastly, interpreting results was hindered by the fact that the majority of studies did not report on the level of intellectual disability of participants. This issue is essential, considering that the salience, manifestation and degree of impairment determined by the emergence of AD symptoms can vary according to the level of premorbid intellectual capacity and also the person’s ability to compensate for the newly acquired deficits. Margallo-Lana et al. (2007) demonstrated that significant cognitive decline at follow - up could only be identified for cases where the baseline score had been at or above the median (mild and moderate ID), with little changes being noticeable for people with severe and profound ID. Moreover, the majority of studies (Adams and Oliver 2010, Kitler et al. 2006, Carr and Collins 2014, Cosgrave et al. 2000, Krinsky-McHale et al. 2002, Nelson et al. 2007, Margallo-Lana et al. 2007, McCarron et al. 2014, Makary et al. 2014, Määttä et al. 2014) did not document whether dementia was at an early or advanced stage. As changes in function seem to vary both quantitatively and qualitatively over the course of AD, future research should report the stage of decline, as well as age and IQ.

Given the findings of this review, we believe that establishing a baseline level of cognitive functioning and behaviour in people with DS would be extremely useful in following up on changes in domains known to be affected by AD. We therefore advocate the development of a set of common measures appropriate for individuals with DS that would include in depth evaluation of memory, executive function, language and behaviour and that can be used both in clinical settings as well as research. This would allow the implementation of Burt and Aylward’s model of best practice (Burt and Aylward 2000) which suggests establishing a baseline of premorbid functioning before the age of 35, having annual reassessments and conducting a detailed diagnosis in cases where decline is evident. More research needs to be conducted in order to decide on the exact measures to be included in this tool set, given the intra-individual variability in task performance in people with DS (Krinsky-McHale et al. 2008). At present, standard tools developed for the assessment of cognition in the general population often show floor effects when administered to individuals with DS (Stanton and Coetzee 2004) and the majority of the assessment tools are based on the premise that respondents have an IQ in or close to the normative range and require other skills such as good verbal skills, adequate attention or dexterity (Prasher et al. 2004). Burt et al. (2005) have emphasised the need for an appropriate scoring method for diagnosing dementia in this population and suggested that slope scores are more useful than difference scores. For example, institutionalised individuals with DS are known to score lower on neuropsychological tests than people living in the community (Nieuwenhuis-Mark 2009) and thus having an established baseline and measuring change over time in a longitudinal follow up design would be a better approach than cross-sectional studies.

Given that the majority of the studies included in this review point towards BPDS as an early sign of dementia in DS, we believe that there is an urgent need for the development of an acceptable, validated and standardised evaluation scale for BPSD in DS (Sinai et al. 2016), similar to BEHAVE-AD in the general population (Dekker et al. 2015).

Furthermore, diagnostic and assessment measures developed for use in the general population are often used in studies investigating AD in DS, which often introduces confounding factors (Ballard et al. 2016). This issue was pointed out by Dekker et al. (2015) who stated that although more than 20 scales have been developed to assess BPSD in AD in the general population (Finkel 2001), none of these have been validated in people with DS and therefore are not equally valuable for use in this population as they do not take into consideration factors such as pre-existing behavioural issues associated with ID. In line with the findings of Dekker et al. (2015), such a scale could be implemented in routine practice alongside existing assessment procedures which monitor change over time, as well as being used in longitudinal research and clinical trials.

Having comprehensive assessment tools for outcomes in clinical trials would enable a targeted approach to drug development, which would be essential given recent findings suggesting that people with DS respond differently to treatment when compared to late-onset AD typically developing controls (Ballard et al. 2016). Moreover, a better understanding of BPSD in DS would increase acceptance from the perspective of a carer who would understand that behaviour they perceive as “challenging” is not deliberate but may be a symptom of dementia (Dekker et al. 2015). Imporved recognition of symptoms would facilitate access to non-pharmacological treatments such as behavioural and music therapy and psychosocial interventions (Gauthier et al. 2010).

To account for the fact that BPSD and executive dysfunction seem to be affected earlier than episodic memory in the course of dementia, we suggest Mortimer’s reserve capacity hypothesis (1981) as a viable theoretical explanation. According to this model, the frontal lobes of people with DS are underdeveloped and thus a smaller buden of neuropathology in the brain may be required to reach the threshold at which functioning is impaired. Volume reduction of frontal lobes has been observed during childhood, while hippocampal volume reduction has only been reported later in life (Jacola 2012). These observations would account for the preferential decline in areas of cognition sub-served by these areas. Indirect support for Mortimer’s reserve capacity hypothesis comes from the longitudinal study conducted by Temple et al. (2001), in which progression of dementia appears to be slowed by having high levels of pre-morbid cognitive functioning. The authors suggest that environmental interventions such as improving the level of education and number of years in employment might represent successful strategies to reduce cognitive decline. The suggestion that early clinical presentation of AD in DS matches that of frontal-based symptoms can serve as a theoretical foundation for the findings of neuroimaging studies such as the recent amyloid PET imaging with 18F–FDDNP study that showed increased frontal predominance of amyloid pathology in people with DS who had a diagnosis of AD, compared to typically developing patients with late-onset AD (Nelson et al. 2011).

To conclude, even though there is no overall consensus on what represents the earliest observable sign of dementia in DS, this review highlights the heterogeneity of affected domains. BPSD, developing for the first time in later adult life, predicts the subsequent development of dementia in the majority of the included studies. There is also evidence that emergence of BPSD is underpinned by impairments in executive functioning that may implicate impairments in frontal lobe integrity and in related brain networks (Ball et al. 2008). We recommend that future research should investigate a wider variety of processes, to reflect the heterogeneity of domains that seem to be affected in the prodromal and early stages of dementia in this population. Moreover, considering that in the general population amnestic MCI is widely accepted as the prodromal stage of AD, an alternative should be proposed for people with DS, to reflect this difference in progression. The key to early diagnosis is knowing what observed changes should raise the index of suspicion sufficiently to indicate the need for a full diagnostic assessment and cognitive evaluation. The findings of this review are especially relevant given that current diagnosis procedures (ICD-10 and DSM-V criteria) emphasize declines in memory functions and therefore in this population may fail to facilitate the early detection and diagnosis of dementia when it does occur.

Acknowledgments

This systematic review is part of a larger research study funded by the Baily Thomas Charitable Fund, Addenbrookes Charitable Trust and the Health Foundation. Queens’ College Cambridge contributed via a grant which enabled the dissemination of these findings at BNA 2015 Festival of Neuroscience in Edinburgh.