Introduction
Ageing in people with Prader–Willi syndrome (PWS) has only recently become an important issue. Research has traditionally differentiated only between children and adults as there were too few older adults (aged ⩾40 years) in most research studies to form a separate group. For example, a population study (Whittington et al. Reference Whittington, Holland, Webb, Butler, Clarke and Boer2001) carried out in a geographic region of five million people identified only four people with PWS aged 40 or over. This might be explained by the fact that, in the past, life expectancy for people with intellectual disabilities was much less than it is now (Coppus, Reference Coppus2013). In addition, at the time when those who are now 40 years or older were children, the ascertainment of those with specific syndromes was not so reliable, and little management advice was available (Dykens, Reference Dykens2013). In the population study cited above it was estimated that the mortality rate of people with PWS was 3% per year across all ages, rising to 7% in those aged over 30 (Whittington et al. Reference Whittington, Holland, Webb, Butler, Clarke and Boer2001). Einfeld et al. (Reference Einfeld, Kavanagh and Smith2006) estimated that the mortality rate in PWS was six times that in the general population of people with intellectual disabilities. Smith et al. (Reference Smith, Loughnan and Steinbeck2003) recorded 10 deaths in a cohort of 36 adults over a 10-year period; a rate of 2.8% per year.
However, with the more recent trends of early diagnosis (typically within 2 weeks of birth), with guidelines for parents on management and diet, and interventions for older people with PWS such as group homes with food management designed to prevent obesity, increased longevity is to be expected, so that ageing in people with PWS is likely to become an increasingly important consideration. The aim of our first investigation was to estimate mortality rates in PWS in the 9 years following the population cohort study to see if we could find evidence of a drop in mortality rate from our previous estimate.
Two notable papers that indicate new health and management issues likely to be encountered as people with PWS live longer are the report of a 58-year-old woman with PWS and Alzheimer-type dementia (Sinnema et al. Reference Sinnema, Schrander-Stumpel and Verheij2011) and a report on 12 people with PWS aged ⩾50 showing high levels of what is described as ‘premature ageing’ or of dementia (Sinnema et al. Reference Sinnema, Schrander-Stumpel, Maaskant, Boer and Curfs2012). Similar issues have also arisen in people with Down's syndrome as life expectation has risen from 9 years in the 1920s to 56 years in the 2000s (Coppus, Reference Coppus2013).
At a meeting of the European PWS Research Group in December 2007, and again at the 7th International PWS Conference, Taipei, Taiwan 2010, Swaab reported on pathological findings in the brains of three people with PWS who had died aged 49, 51 and 64 respectively. On examination, these brains showed evidence of the typical plaques and tangles associated with Alzheimer's disease (AD) and the extent of this pathology increased linearly with age. This was the first report of the neuropathological signs of AD in people with PWS and an indication that people with PWS, like those with Down's syndrome, may be at risk for early onset AD (Holland et al. Reference Holland, Hon, Huppert and Stevens2000). This would not be surprising given that some of the risk factors for AD in the general population include diabetes mellitus and obesity (Luchsinger & Mayeux, Reference Luchsinger and Mayeux2007), low levels of sex hormones (Ravaglia et al. Reference Ravaglia, Forti, Maioli, Bastagli, Montesi and Pisacane2007) and low levels of cognitive ability (Temple et al. Reference Temple, Jozsvai, Konstantareas and Hewitt2001). These are all common to people with PWS. Unfortunately, no other information was available on Swaab's three cases, so pre-morbid behavioural and functional changes, which might have accompanied the pathology, are not known.
Given these limited observations and the presence of known risk factors for AD in people with PWS, it is important to investigate whether behavioural and functional changes are observed as people with PWS age and, in particular, to investigate the prevalence rates of AD and other dementias (for example, are they comparable with those in people with Down's syndrome; starting as early as the thirties and reaching 50% over age 60). In addition, if changes suggestive of dementia are observed in later life, it is important to know the early signs so that a diagnosis can be made and appropriate support given to try and prolong the non-dependent stages and to inform longer-term care.
The aim therefore of our second study was to identify people with PWS aged 40 or over in the UK to investigate evidence of age-related decline and particularly the onset or not of dementia.
Method
Participants and procedures
Study 1
This part of our investigation involved tracing all of those people with PWS who participated in the population-based phenotypic study (1998–2000 inclusive) in the UK (Whittington et al. Reference Whittington, Holland, Webb, Butler, Clarke and Boer2001). The Prader–Willi Syndrome Association UK (PWSA-UK) and residential homes were asked to verify whether those participants known to them were still alive. Where people had moved or lost contact with the PWSA, we tried to trace them using their last known address.
Study 2
All 43 PWS members of the PWSA-UK, identified as being aged ⩾40 years, and an informant (usually parent or carer) were invited, by personalized letters sent through the PWSA, to participate in this research. Those not responding within 4 months to this initial mailing were contacted again in the same way. In all, 26 members meeting clinical criteria for PWS and their parents/carers agreed to participate, eight refused and nine did not respond. Of these 26 participants, 11 had the deletion subtype (five male, six female), five had the uniparental disomy (UPD) subtype (one male, four female), one (male) had a translocation and one (male) was genetically PWS but was not subtyped. Eight did not have a genetic diagnosis and had refused blood tests (six male, two female) but were included in the study as they met clinical criteria.
Informant-based semi-structured interviews were undertaken using the Cambridge Examination for Mental Disorders of Older People with Down's Syndrome and Others with Intellectual Disabilities (CAMDEX-DS; Ball et al. Reference Ball, Holland, Huppert, Treppner, Watson and Hon2004, Reference Ball, Holland, Huppert, Treppner and Dodd2006a , Reference Ball, Holland, Hon, Huppert, Treppner and Watson b ) designed to assess signs of dementia, especially AD, in people with intellectual disabilities. The CAMDEX-DS informant interview is a clinical instrument consisting of a structured series of questions that are used to enquire from informants, who know the person well, any evidence of change in specific cognitive and functional domains that characteristically deteriorate with the onset of dementia (e.g. memory, general mental functioning, skills, personality). The diagnosis of dementia using the CAMDEX-DS has been shown to be reliable and to have predictive validity (Ball et al. Reference Ball, Holland, Huppert, Treppner, Watson and Hon2004). Information obtained from the CAMDEX-DS informant interview by J.E.W. was presented to A.J.H. and the diagnosis or not of dementia determined blind to information about their age or genetic subtype. Fifteen of the informant interviews were conducted face-to-face and 11 were conducted over the telephone by the same researcher. In both cases supplementary questions were used to clarify responses and anecdotal evidence was encouraged.
Ethical considerations
The study had a favourable ethical opinion from the National Hospital Research Ethics Committee.
Results
Study 1
Table 1 shows the results of trying to trace the people who participated in the population-based study (1998–2001). Of those 15 in this cohort born before 1 July 1969 (who would therefore now have been over 40 years old), four of those 10 traced from this 15 (all with the deletion subtype; three male, one female) were known to have died in the intervening period (aged 59, 38, and two in their early forties). This gives an estimated death rate of 4.4% per annum. As a third of this cohort was untraced, this proportion could well be higher (6.7% if all those untraced had died) or lower (3.0% if all those untraced were still alive). However, whereas in 2000 there were only four people with PWS within the population covered who were ⩾40 years, in 2009 there were at least six; this is tentative support for some improvement in life expectancy.
Table 1. Follow-up of older people from UK population study (undertaken from October 1998 to May 2001)
Other known deaths from the original lifespan cohort included a 13-year-old girl (not overweight), a young man in his early twenties who gained weight rapidly after being transferred from foster care to a local authority home, and a man in his early thirties who was overweight and had diabetes mellitus. Thus, an estimate of the death rate over all ages in this population cohort in the intervening 9 years is 7/42 or 1.9% per annum but could be as low as 7/62 or 1.25% per annum (if all those untraced were still alive) or as high as 27/62 or 4.8% per annum (if all those untraced had died). This estimate is lower than that made in 2001, although it still seems to be the case that early deaths are fairly common in people with PWS.
Study 2
Of the eight people without a confirmed PWS diagnosis, six showed no deterioration in any area, one in only one area and one in two areas. None met criteria for dementia. These people are not included in the following analyses as they may not have had PWS.
Figure 1 shows the age structure of all 18 confirmed PWS participants aged ⩾40 years and Table 2 shows the total number of domains on the CAMDEX-DS for which deterioration was reported. The CAMDEX-DS is not scored but requires interpretation by a trained psychiatrist. Eight people showed no deterioration in any area, five in only one area and one in two areas. In none of these participants were the full criteria for dementia or age-related cognitive decline met. The findings in the remaining four of these 18 participants are not easy to interpret. Each is described in more detail in the following text.
Fig. 1. Ages of Prader–Willi syndrome (PWS) sample at time of interview.
Participant A is a 48-year-old woman, with reported deterioration across three domains using the CAMDEX-DS informant interview. Her long-term carer of 15 years, however, attributed the apparent lack of interest in her former hobbies of reading and word-search puzzles, her slower physical movements, and her need for help with some aspects of self-care (e.g. putting on tights) to what she described as ‘premature ageing’ and to the effects of increased weight. This interpretation is supported by the observation that, on the Wechsler tests, this woman had the same Digit Span raw score (5) as 10 years ago, although her raw score on Information had fallen from 10 to 8. She was friendly and cheerful and maintained a fluent conversation, although her speech was judged as being slower than 10 years ago. She did not report having a depressed mood. This woman genetically has a UPD and has had psychotic episodes in the past. Dementia was considered unlikely.
With the other three participants aged 55, 45 and 41, parents/carers were concerned about changes in their mental and physical functioning. The assessment of one of these (participant B aged 41) was complicated by the fact that she was in a psychiatric hospital at the time with a psychotic illness so that many changes in her behaviour and functioning could be attributed to this illness. However, carers’ concerns arose because this episode seemed to them to be different from previous episodes. A further complication in this case was that changes in behaviour and functioning were first noticed following a change in her psychiatric medication 10 months before the psychotic episode. On the CAMDEX she seems to fulfil criteria for (at least) mild dementia but the psychosis leaves open another interpretation. Although this woman is a deletion subtype, she has extra genetic material from her mother in the region of the putative ‘psychosis gene’ (thus she is disomic in part of the region that lies between D15S975 and D15S661, towards Break Point 3). Previous research has shown that this additional chromosomal abnormality has been found in all deletion subtypes tested who have a history of psychosis but not in those who do not (Webb et al. Reference Webb, Maina, Soni, Whittington, Boer, Clarke and Holland2008).
Participant C, now aged 55, was administered the Wechsler IQ battery in 1999. Digit span (short-term memory) was poor (she could only repeat back two digits reliably) and Information (general knowledge, long-term memory) was also poor; however, she knew the shape of a ball and what a thermometer is. In 2009 the two memory tests from the IQ battery were repeated. She could still repeat back two digits but could no longer tell the shape of a ball or what a thermometer is. She did not know her age and was not sure whether Sunday or Monday follows Saturday, eventually opting for Monday. The informant, who has known her for 11 years, thought that this deterioration had begun about 2 years ago, first affecting her everyday skills and self-care. She now needs assistance with most aspects of self-care, she is doubly incontinent, and her behaviour in public is more frequently embarrassing. Following a conversation, making decisions and remembering recent events or where things have been left were also reported to have become more difficult. She has PWS due to a UPD subtype and has had intermittent psychotic illness but was not currently mentally unwell. In this case dementia is a distinct possibility.
Participant D, aged 45, has been seen three times over many years as part of our research. In 1999 she was depressed and anxious and assessment of her Digit span was unreliable (some longer sequences were correct and some shorter sequences incorrect, possibly due to concentration difficulties) and her Information test score was eight. On a visit in 2004 she was too depressed to participate in testing. In August 2008 she appeared happy and confident and Digit span was internally consistent, with up to four digits recalled accurately and her score on the Information test was seven. The informant thought that deterioration had begun about 5 years ago, and had first affected her everyday skills. The decline has been very gradual, mainly over the past 2 years, and is still not very marked. However, she does now sometimes have difficulty remembering recent events, where she has left something, what has been said, she does not always know where she is, what day it is, or what time of day it is. She is more easily distracted, her thinking is sometimes muddled and she has more difficulty keeping up with conversation and with finding the right words. She has some difficulty making decisions and with problem solving. Her behaviour sometimes leads to social difficulties and urinary incontinence has increased in frequency. She has PWS due to UPD and has suffered from bipolar affective disorder for many years. Again the picture is somewhat obscure but her carer has described deterioration over and above the phases of her bipolar illness. This was considered to be a possible case of dementia.
Risk of dementia or premature ageing
In total, four of the 18 participants had some evidence of decline and in three of the 18 (17%) aged 40 years or over this was sufficient to meet criteria for dementia, although for participant B the diagnosis was complicated by a relapse in her psychotic illness at the time seen. These three cases in which parents/carers reported deterioration and in which the diagnosis of dementia was possible had commonalities. All were female, all had a history of recurrent psychotic episodes, two had PWS due to UPD and the third had a deletion but with maternal disomy in the region of the putative ‘psychosis gene’ (Webb et al. Reference Webb, Maina, Soni, Whittington, Boer, Clarke and Holland2008). The main changes in behaviour and functioning included loss of interest in usual activities and social withdrawal (three), new or increasing incontinence (three), slowing of physical movement (three), and loss of interest in food (two).
It is clear from these results that there was a difference between the genetic subtypes. One person with an atypical deletion showed deterioration in most areas; otherwise deterioration observed in people with PWS due to a deletion and in those with unknown subtypes was usually isolated and had an alternative explanation other than that due to dementia, for example weight gain, loss of reading and writing skills through loss of interest and therefore disuse, or increasing behaviour problems (Table 2).
Table 2. Areas of decline by gender and history of psychosis
There are also clear effects of gender (none of eight males and four of 10 females) and history of psychosis (none of eight without and four of seven with) (Table 3). The latter is associated with the disomy subtype and it is not clear which of these might be the primary risk factor. However, female gender does seem to be an independent risk factor.
Table 3. Number of areas of deterioration by gender and history of psychosis
Conclusions
Study 1
There does seem to be a slight tendency for older people with PWS to survive longer, based on the people included in our population study. The number of people with PWS in the population study region who are known to have reached the age of 40 years or more has risen from four to at least six in the intervening 9 years. However, the estimated mortality rate over all ages in this cohort in the intervening 9 years is 1.9% per annum based on those traced) and in this older age group 4.4% per annum based on those traced). This compares with an estimated 3% per annum over all ages and 7% per annum over age 30 prior to the population study.
Study 2
In this study our findings must be treated with caution as the numbers were small and ethical restrictions meant that more extensive assessments were not possible. In the case of those who had probably deteriorated, referral to their local community services for further investigation has been recommended. In addition, the age of the sample was skewed and did not cover the full range of Swaab's three cases. The findings, however, are potentially very important. First, although we do not know the genetics of the cases reported by Swaab, his predictions were that people with PWS in general (regardless of genetic subtype) would be at risk for dementia of the Alzheimer type. Contrary to this we found that no one with the deletion subtype or undetermined subtype in the age group seen showed significant signs of deterioration except for the person with PWS who was atypical in that she had psychotic episodes and was disomic (extra genetic material from her mother) in the region between D15S975 and D15S661. Second, the participants where there did seem to be significant deterioration were all female, all had UPD or had a disomic region, and all had had psychotic episodes; their ages were 41, 48 and 55. We note that the one reported case of AD in a person with PWS was female, maternal UPD (mUPD) and had a history of psychosis (Sinnema et al. Reference Sinnema, Schrander-Stumpel and Verheij2011).
How can we begin to explain these findings? At first sight, any one simple explanation (psychosis alone, the mUPD subtype alone, or gender alone) seems unlikely. Psychosis in PWS occurs in both genders, as does the mUPD subtype, and gender itself is not an explanation because females with the deletion subtype were not similarly affected. In the case of gender, there is some slight evidence that females may be more susceptible to AD (Tola-Arribas et al. Reference Tola-Arribas, Yugueros and Garea2013) but this has not been found in all samples. It is also possible that the fall in sex hormones associated with the menopause in females is a risk factor for an earlier onset of ageing or dementia, irrespective of other risk factors. In the case of psychosis, although there are many reports in the literature of the prevalence of psychotic symptoms in AD, we have been unable to find a single estimate of the prevalence of AD or other dementias in people with long-standing psychotic symptoms. If a history of psychosis is associated with a higher risk of dementia or ‘premature ageing’, this would also explain why those with the mUPD subtype are more at risk for dementia. If PWS itself, regardless of subtype, increases the risk of dementia it is possible that those with the deletion subtype become affected by ageing or dementia later in life than those with mUPD and in this study the presence of dementia in this group has not yet become apparent.
We do know that the mUPD subtype is associated with a very high risk of psychosis. In the cases described here the psychosis was of long standing and preceded any signs of the recent deterioration. Psychotic symptoms in AD have been associated with the genetic markers 5-HTTLPR L-allele (Quaranta et al. Reference Quaranta, Bizzarro, Marra, Vita, Seripa and Pilotto2009) and AD7c-NTP (Monte et al. Reference Monte, Ghanbari, Frey, Beheshti, Averback and Hauser1997). In the region of the ‘psychosis gene’ associated with the UPD subtype of PWS, the mRNA AK090651 codes for the theoretical protein FLJ33332, which has similarity to AD7c-NTP. It is possible that maternal disomy in this region may explain the link between psychosis and the apparent deterioration found only in people with PWS with this genetic configuration.
Future research
Because of the small numbers, the present study can only be regarded as a preliminary to further research. Clearly, larger cohorts are required to look at the overall risk of dementia in people with PWS, and in particular the roles of genetic subtype and gender, and these must come from cross-national longitudinal studies of older people. In addition, more extensive investigations would be required to exclude with a greater degree of certainty other possible causes of decline. Moreover, newer cohorts will also have to include subgroups on/off growth hormone and sex hormones to assess their impact on risk of early onset dementia and/or premature ageing. If there is a link between psychosis and the risk of subsequent dementia, a key question will be whether early and effective treatment of dementia reduces that risk. Larger cohorts are also needed to address the question of the relationship between ageing and the risk of dementia and to establish: What is the relationship between age and the risk of dementia in people with PWS and what are the mechanisms that underpin such risk?
Acknowledgements
This research study was supported by a grant from the Foundation for Prader–Willi Research. A.J.H. is supported by the Health Foundation and the National Institute for Health Research (NIHR) Collaborations in Leadership for Applied Health Research and Care (CLAHRC) for the East of England. We thank the participants and their parents and support workers for taking part in the research, and also the PWSA-UK for help with recruitment.
Declaration of Interest
None.
