The concept of ‘behavioural phenotypes in organic genetic diseases’ was first proposed by Nyhan in his Presidential lecture to the American Paediatric Association in 1971. He was referring to the fact that severe self-injurious behaviour was as much a part of the phenotype of the X-linked syndrome that subsequently was named the Lesch—Nyhan syndrome, as was any physical abnormality. In the last few years the association between specific genetically determined developmental syndromes and particular cognitive, linguistic and behavioural profiles has been increasingly recognised. This has partly been due to advances in genetic techniques, particularly in the identification of novel genetic mechanisms that give rise to developmental disorders and can account for phenotypic variation within syndromes. The best example of the latter is the relevance of the CGG triplet repeat sequence length in people with Fragile X syndrome.

In the case of genetic syndromes associated with a ‘behavioural phenotype’, it is proposed that single genes may have specific effects on brain development and function, and thereby give rise to particular patterns of social, emotional and/or cognitive development, and an increased propensity to aberrant behaviours that are recognised as being specifically, but not exclusively, associated with that syndrome. This association between genetic syndromes and ‘behavioural phenotypes’ is in contrast to the usual model of behavioural ‘normality’, which presumes an interaction between the inheritance of allelic variants of several genes and shared or non-shared environmental influences on cognitive development, and on behaviour patterns and personality.

Several psychiatric conditions have been associated with PWS. As noted in Chapter 10, increased risk rates have been proposed for psychotic illness and obsessive—compulsive disorder (OCD). This chapter is concerned with the latter. Obsessive—compulsive behaviour has long been recognised as part of the PWS phenotype and it is generally agreed that people with PWS have ‘obsessive personalities’, but this is not the same as OCD.

While some reports on the obsessive—compulsive behaviour in PWS have raised the possibility of heightened rates of OCD in this population, the difficulties of psychiatric diagnosis in people with learning disabilities (LD) seem to have discouraged attempts at quantification in PWS, although there are several estimates of prevalence in undifferentiated groups of people with LD. Exceptionally, the 1996 paper by Dykens, Leckman & Cassidy does seek to elucidate the nature of the obsessive—compulsive behaviour by people with PWS. We describe this paper in detail since, as well as being the most thorough investigation of obsessive—compulsive behaviour in PWS of the last decade, it raises several interesting issues.

In DSM-IV a diagnosis of OCD presumes the presence of either obsessions (intrusive, recurrent thoughts or images that exceed real-life worries) or compulsions (repetitive behaviours that a person is driven to perform in response to an obsessional thought or set of rigid rules).

The book follows a process of discovery that included getting to know many people with Prader—Willi syndrome (PWS) and their families through the undertaking of a population-based study of children and adults with PWS living in the then Anglia and Oxford Health Region of the UK. The aims of this study were to systematically investigate many aspects of the syndrome free from the selection biases that may have influenced previous clinic and volunteer-based studies. The in-depth knowledge gained from this ‘Cambridge study’ through direct contact with people with PWS, their families and other carers, and the integration of this knowledge with that in the literature, has led in turn to new hypotheses and new interpretations of some aspects of PWS.

Prader—Willi syndrome is a relatively rare, genetically determined complex neurodevelopmental disorder. People with the syndrome have their own individual characteristics but share in common particular physical features, specific cognitive strengths and weaknesses, and a propensity to particular behaviours and mental health problems that change with age and together make up the early and later physical and behavioural ‘phenotypes’ of the syndrome.

The genetics of PWS are still not fully elucidated and are clearly more complex than originally anticipated. Candidate gene(s), located on the long arm of chromosome 15 between bands 11 to 13 (15q11—q13), which are maternally imprinted and only expressed when inherited from the father, have been identified but the precise gene or genes whose absence of expression leads to PWS are still not known.

The areas of research considered in this chapter were not directly included in the Cambridge population-based study. However, these studies are important in that they further identified some of the possible pathophysiological mechanisms that underpin the physical and behavioural characteristics that the diagnostic criteria have identified as being central to the syndrome. For example, the very striking features of obesity, short stature and impaired sexual development, together with the less obvious features of temperature dysregulation, abnormal pain threshold and abnormal body composition, all require an explanation. In essence, it is important to establish whether these apparently diverse abnormalities might be traced back to a common single causal mechanism (and by implication the absence of expression of one gene) and, if so, whether the fundamental biological deficits consequent upon the PWS genotype are central (i.e. an abnormality of brain function) or more peripheral and due to abnormal feedback mechanisms, for example, from the gut and/or fat stores. An alternative view is that PWS should be seen as a ‘contiguous gene disorder’ and that it is the absence of expression of more than one imprinted gene that leads to the full PWS phenotype. If more than one gene is involved, then there may not be a single unifying pathophysiology, but rather, different and discrete pathological mechanisms that lead to particular phenotypic abnormalities.

In Chapter 1, we outlined the events leading up to the adoption of the Consensus Diagnostic Criteria (CDC). A weighted score of eight or more for ages four years and above (five or more for ages less than four years), based on the presence of eight major (score 1) and 11 minor (score 0.5) symptoms, is considered sufficient for a clinical diagnosis of PWS. According to these diagnostic criteria, there is no requirement that any particular one (or more) of the criteria are present, rather that the total number present exceeds a given threshold. This was a relaxation of the implicit criteria of individual physicians. As noted in Chapter 1, some early reports, such as Hall & Smith and Pipes & Holm, did consider that the presence of some criteria were critical in the diagnosis of PWS.

Before PWS was recognised as a genetic syndrome, diagnosis was necessarily based on clinical criteria, and the primary role of such criteria was to ensure as accurate a diagnostic process as possible. When it was recognised that at least half of the people diagnosed with PWS had a deletion on the long arm of chromosome 15, clinical criteria and genetic criteria were used interchangeably.

In the previous chapters we examined different aspects of PWS drawing upon anecdotal information, previously published research and systematically collected data from our Cambridge population-based study, to help clarify and illuminate particular issues. Whilst a considerable amount is now known about the syndrome, some of the most crucial aspects still remain elusive. Most striking is whether it is caused by the absence of expression of one or more genes, and which genes. What is clear is that these genes are normally only expressed when inherited from the father, being imprinted when inherited from the mother. At present the SnoRNAs are the main candidates, but until this is fully elucidated, it is not possible to say what the ‘PWS gene(s)’ do and therefore what does not happen when they are not expressed. The second area that remains a considerable mystery is the mechanisms that provide the link between the abnormal gene expression (genotype) and the physical and behavioural phenotype. The crucial issue here is whether all that is characteristic of people with PWS can be explained by one mechanism (and by implication the absence of expression of one gene) or whether more than one underlying process is necessary to explain the full phenotype (and by implication absence of expression of more than one gene).

Prevalence of psychiatric illness in people with learning disabilities (LD) may be slightly higher than that in the normal population, especially when the difficulties of diagnosis in the learning disabled population is taken into account. Cooper & Collacott, in a review article, suggested that rates of affective disorder among people with a learning disability may be as low as 1–5%. Corbett, in his Camberwell study of people with LD, aged 15 years and older, found rates of present or past significant affective disorders of 5.5%. A population-based study of adults with Down syndrome reported rates of depression of 11%. Psychiatrists working with people with PWS have come to speculate that rates in PWS may be higher than those in the general LD population, and that general depression and anxiety may not be the most common symptoms found in PWS. A study that looked at psychiatric symptoms in 23 people with PWS (genetically confirmed in 20 cases) and a comparison group of 73 people with other LD found higher rates of affective disorder (4/23 v. 3/73), schizophrenia/delusional disorder (1/23 v. 2/73), and obsessive—compulsive disorder (OCD) (1/23 v. 2/73) in PWS compared with other causes for LD. Unlike the LD group, there were no cases of generalised anxiety disorder, agoraphobia, other phobias, alcoholism, dementia or autism in the PWS group. In a group of 35 adolescents with PWS, it was estimated that more than half had definite or probable DSM-III diagnoses, compared with 18.7% of the general population.

The need for a population survey

At the Third Triennial International PWS Conference in Venice in 1998, the scientists present agreed that a large population prevalence study of PWS was needed for two main reasons. First, the two earlier population surveys had not been large enough to produce reliable estimates and, in the case of one of them that surveyed people aged 0–25 years, had not been comprehensive. Moreover, neither of these two studies used genetic criteria for the diagnosis of PWS and therefore the prevalences were very likely to be over-estimates. A larger population study, using established clinical diagnostic criteria together with recently agreed genetic diagnostic techniques, would therefore provide more accurate population prevalence and birth incidence figures. Second, the possibility of sampling bias could not be ruled out where conflicting findings had been obtained in different studies that used groups of people with PWS who were ascertained via specialist clinics or who had volunteered. It was strongly suspected that such a bias was an explanation for at least some of the contradictions. That is, in these conflicting studies, the groups of people with PWS who were included were probably not representative of the PWS population more generally. A new population survey, with genetic confirmation of the diagnosis, would both provide a representative sample of people with PWS to help with future research and more accurate prevalence data on which to determine service requirements.

Previous population studies

In the literature on PWS, prevalence has been variously quoted as ‘about 1 in 25,000 live births’, ‘between one in 25,000 and one in 10,000 live born children’, ‘[estimates] vary 6-fold from 1 in 5,000 to 10,000; 1 in 10,000; 1 in 15,000; 1 in 25,000; to 1 in 10,000 to 30,000’. These estimates are confusing because some authors use the term prevalence where others use birth incidence. We use birth incidence to refer to the number of live births of babies with PWS and prevalence to refer to the number of people with PWS of all ages in the population. Only two estimates appear to be based on epidemiological data, those of Akefeldt et al. and Burd et al. (see Chapter 1). In the North Dakota study (Burd et al.) the authors used similar informants to those contacted in the Cambridge study. The ascertainment methods used in the North Dakota study identified eight males, eight females and one person whose gender was not given, with an age range from 9 to 30 years. At that time the population of North Dakota for that age range was 263 444, giving a prevalence rate of 1:16 062, equivalent to 1:38 395 in the entire population. No figures were given for the number of people with a genetic diagnosis, and the diagnostic assessment was based, in at least some cases, on a one-page questionnaire pictorially demonstrating the signs of PWS to aid identification. Birth incidence was not estimated.