Prader–Willi Syndrome (PWS) is a genetically determined neurodevelopmental disorder associated with mild to moderate intellectual disability, growth and sex-hormone deficiencies and a propensity to overeat that leads to severe obesity. The PWS phenotype changes from an early disinterest in food to an increasing pre-occupation with eating and a failure of the normal satiety response to food intake. The prevention of severe obesity is primarily through strict control of access to food and it is this aspect that most limits the independence of those with PWS. This review considers the eating disorder in PWS, specifically how the as yet uncertain genetics of the syndrome and the transition from the early to the later phenotype might account for the later hyperphagia. On the basis of behavioural and imaging studies, a failure of satiety and excessive activation of neural reward pathways have both been suggested. We speculate that the overeating behaviour, consequent upon one or other of the above, could either be due to a direct effect of the PWS genotype on the feeding pathways of the hypothalamus or a consequence of prenatal changes in the regulation of genes responsible for energy balance that sets a high satiation threshold. Understanding the overeating in PWS will lead to more focused and successful management and ultimately, treatment of this life-threatening behaviour.
We review the present state of knowledge on the eating disorder and potential resultant life-threatening obesity that is associated with Prader–Willi Syndrome (PWS), a genetically determined neurodevelopmental disorder affecting 1 in 22–25 000 live births.1, 2, 3 The aims of the review are: first, to consider our understanding of the eating disorder associated with PWS primarily from a behavioural and neuroscience perspective; second, to speculate as to the possible mechanisms that might directly or indirectly link the PWS genotype to the ‘eating’ phenotype; and third, to reflect on the implications for future research and for treatment. We discuss the major theories that could account for eating behaviour and whether any of these adequately explain the complex array of characteristics that exist in PWS and fit with what we know about the pathophysiology and the genetics of the disorder. The main discussion items include considering PWS as a disorder of satiety, as well as theories of hyper-responsive rewards systems and over-sensitivity to food stimuli, as has been proposed in general obesity. Further, we consider the importance of the prenatal environment and examine the possibility that people with PWS have an altered inner physical awareness, as shown by abnormal pain responses, and how this may alter the eating behaviour.
This review has been carried out through a search of relevant articles primarily on the PubMed search engine (www.pubmed.gov) using ‘eating behaviour Prader–Willi syndrome OR Prader Willi syndrome’ as the search term. As the eating behaviour is such a key feature of this rare disorder, almost all articles contain some mention of it. Therefore, articles specifically focused on eating and particularly related to behavioural aspects of eating in PWS were selected. Those papers that did not come up through the search engine but were frequently mentioned in articles were also sought through Google Scholar. Advice was also sought from experts in the field of eating behaviour in PWS and hyperphagia.
In the background we include a brief overview on the full PWS phenotype and the PWS genotype, both of which are necessary when considering neural mechanisms that might account for the eating behaviour. More detailed reviews of these aspects of the syndrome are available.4, 5 We then focus primarily on the nature of the evolving eating abnormalities in PWS: we consider the neural mechanisms that might underpin the abnormal eating behaviour; and the potential pathophysiological basis for why the regulation of eating behaviour, and therefore of energy balance, is abnormal in PWS, consequent upon the presence of the genetic abnormality.
The primary, secondary and supportive diagnostic criteria for PWS were first categorized by Holm and colleagues6 before the genetic basis for the syndrome was fully established. Since then the early and later phenotypes have been characterized and certain core features have been identified that, if absent, are invariably associated with negative genetic testing for PWS.7 At birth these include severe neonatal hypotonia and failure to thrive, and undescended testes in males. In later childhood, PWS is associated with the failure of normal secondary sexual development; a characteristic physical appearance and short stature consequent upon relative growth hormone deficiency; and the development of severe overeating behaviour. The nature and development of the eating behaviour means that if food intake is not controlled by others, obesity arises from around age 1–4 years, exacerbated by low metabolic rate due to reduced physical exertion, and also by a low lean body mass.8 These are the core features.7 In addition, other secondary and supportive features are common but their absence does not necessarily predict a negative methylation test for PWS—the established genetic diagnostic test. These characteristics include evidence of mild or moderate intellectual disability,6, 8 a high pain threshold, temperature instability and a propensity, which may continue into adult life, to certain behaviours, such as temper outbursts, repetitive and ritualistic behaviours similar to those found in autistic spectrum conditions,9, 10, 11 and severe skin picking. PWS also shows a complex psychiatric phenotype with high rates of affective disorder and the development of a psychotic illness in adult life, particularly in those with one genetic subtype of PWS.12, 13 This same sub-group (those with PWS due to uniparental disomy), in addition to repetitive and ritualistic behaviours, may also show relatively more impaired social functioning, similar to that found in people with autism spectrum disorders.14, 15
The PWS genotype
The precise genetic abnormality that, by itself, is sufficient to account for the core features of the syndrome, remains unclear. There are three genetic subtypes of PWS, the common feature being loss of expression of the paternally-expressed imprinted genes in the critical region, 15q11-q13.8 Around 70% of people with PWS have a deletion of the critical region involving the chromosome 15 of paternal origin. Approximately 20–25% have inherited both copies of chromosome 15 from the mother (maternal uniparental disomy), and 2–5% have defects of the imprinting centre (PWS-IC) at 15q11–13.16 Defects in the PWS imprinting centre can result as a failure of one of a number of factors, as discussed by Horsthemke and Wagstaff.5 The relative proportion of genetic subtypes found in a sample depend crucially on the number of older mothers in a population, as the risk of maternal uniparental disomy rises steeply with age, at least from the mid thirties.17 These genetic subtypes are indicative of PWS resulting from the absence of expression of a gene or genes whose alleles have, through evolution, become imprinted (not expressed) depending on the gender of the parent of origin. The gene(s) of relevance to PWS are normally expressed from the chromosome 15 of paternal origin with the allele(s) of maternal origin being silenced. In the case of the two main genetic subtypes (del15q11–13 of paternal origin and maternal uniparental disomy of chromosome 15), PWS arises as these genetic abnormalities result in the absence of the paternal copies of the relevant genes. In contrast, either deletions at a similar locus on chromosome 15 but affecting the chromosome 15 of maternal origin or the presence of a chromosome 15 paternal uniparental disomy, results in the phenotypically very different Angelman syndrome. In this case it is the absence of expression of a specific maternally expressed/paternally imprinted gene (Ube3a) that is causal.18 Candidate genes for PWS, on the basis of their location and being maternally imprinted, include necdin, Magel2, SNRPN, MKRN3 and snoRNA genes. Although it has been generally assumed that the wider phenotype of PWS results from the absence of expression of more than one maternally imprinted gene, a recent study of a child with PWS has shown that many of the major features are due to loss of expression of C/D box HBII-85 snoRNAs.19 Although this genetic abnormality may not account for the full phenotype of PWS, it is associated with several aspects including the abnormal eating behaviour. These snoRNAs are important in the splicing and editing of messenger RNA, but why the absence of their expression in PWS leads to overeating behaviour is uncertain. Several theories have been put forward to explain the evolutionary benefit of imprinted genes, although the precise selective advantage is unknown. In brief, genomic imprinting is thought to increase the rate of evolutionary adaptation, protection from ovarian trophoblastic disease, or occur as a result of parent–offspring conflict, as reviewed by Wilkins and Haig.20 The most studied of these theories is parent–offspring conflict or ‘kinship’ or ‘conflict’ theory. From this perspective it has been argued that any gene whose alleles have become imprinted, depending on the gender of the parent of origin, must have been subject to competing maternal/paternal selection pressures.21 In the case of the relevant gene in PWS the prediction is that the competition is for maternal resources. The allele of paternal origin would drive the use of maternal resources during pregnancy to optimize fetal growth as the paternal interest is in the survival of that particular offspring. Genes that are maternally imprinted/paternally expressed are expressed in the placenta22 and in the central control areas of the brain.23, 24 Comprehensive reviews of genomic imprinting can be found elsewhere.20, 21
Eating behaviour in PWS
Behavioural characteristics and development
The eating disorder is one of the factors, if not the factor that most affects the lives of children and adults with PWS. Controlling the hyperphagia remains a major and chronic problem for people with PWS and their families, and it severely limits independence in adult life because of the risk of life-threatening obesity. This has led to the establishment of PWS-specific homes for adults, where access to food (and usually money) is controlled. In experimental settings people with PWS have been observed to consume around three to six times more than the normal calorific intake at a given meal.25, 26 Overeating has been known to lead to stomach rupture27, 28 and individuals with PWS have been known to steal and hoard food. People with PWS have also been reported to eat inappropriate food, such as uncooked chicken, or to eat non-food items (pica).29, 30 Mortality rates in PWS have been reported as six times higher than in those with other causes for their intellectual disability31 and mortality has been estimated at around 3% per year over the lifetime.1 Obesity-related problems, such as respiratory difficulties and cardiovascular disease are the most common cause of death.32, 33 Deaths related to eating itself have also occurred including choking whilst binging, which was cause of death in 7% of people with PWS in one study.34
The developing ‘eating’ phenotype
The eating behaviour of people with PWS is traditionally characterized by two distinct phenotypic stages. The first stage, which is apparent at birth, is dominated by a lack of suck, severe hypotonia and a disinterest in food. Evidence of higher rates of hydramnios during pregnancy suggests that these problems are also of functional significance prenatally, as will be discussed later.35, 36 A recent study from our group found that 100% of the sample of children under 5 years with PWS had sucking problems from birth and 87% had at some point been tube fed.37 However, this difficulty with food intake is a transient stage and generally disappears in the earliest years of the child's life.
The later stage, which typically appears at around the ages of 2–5 years, is characterized by increasingly obsessive food-related behaviours with evidence of hyperphagia. Recent studies have examined the nature of the transition between the phenotypic stages in PWS, and suggest that there are more discrete phases within the early to late phenotype transition, and that the transition itself is gradual, rather than a distinct switch between food disinterest and food preoccupation. This change involves a period of normalization of attitudes towards eating, reported to occur around the end of the first year as assessed by Eating Code Vignettes, developed specifically for PWS.37 Unexpectedly, there is also evidence to suggest that the propensity to obesity may develop before the onset of hyperphagia with a parentally-reported increase in food interest developing after an initial escalation in body mass index.37, 38, 39
Eating behaviour in PWS varies between individuals, notable from an early age.37 Studies have shown that people with PWS do discriminate and show preferences between foods.40 PWS is associated with a high response on the ‘ritualized eating behaviour item’ from the Yale–Brown Obsessive Compulsive Scale (Y-BOCS41), and people with PWS switch less between food types than mental-age matched controls on a standard meal.42 This study also found that the bite rate correlated negatively with ordering non-food items into patterns.
Despite heterogeneity in some aspects of idiosyncratic eating behaviour, excessive food intake remains the key characteristic of PWS. Limited energy expenditure and high fat mass may contribute to rates of weight increase or difficulty in losing weight but the fault in PWS is, fundamentally, a failure to regulate energy intake. The question, then, is what level in the peripheral and central feeding pathways and in the higher centres of the brain, where conscious experiences of hunger and fullness and control of eating behaviour are represented, does the abnormality lie, and as a consequence of which pathophysiological mechanism has it arisen?
Hormonal control of appetite in PWS
In recent years there has been a drive to understand endogenous mechanisms that control appetite and eating in PWS, including gastrointestinal signals, such as pancreatic polypeptide, cholecystokinin, ghrelin and peptide YY. A full summary of hormonal control of appetite in PWS, which has been very well reviewed elsewhere,43 is not the intention of this paper. In brief, PWS presents a different endocrine picture than would be expected in general obesity. For example, adults with PWS retain insulin sensitivity and show reduced fasting levels of insulin.44, 45 Hypoinsulinemia may partially account for the hyperghrelinemia seen in PWS.46 The role of ghrelin has yet to be determined in PWS. Ghrelin levels have been reported as falling post-prandially in PWS children, suggesting that ghrelin function may be operating in childhood.47 Unpublished findings from our group show that ghrelin decreases as expected with age, although at a slower rate. Failure of circulating ghrelin levels to fail post-prandially have been reported in PWS adults48 with levels of fasting and post-prandial ghrelin elevated relative to obesity.48, 49 In a cross-sectional study in children with PWS, plasma levels of ghrelin were found to be higher than in controls, even in the early years of life, suggesting that this precedes the onset of overeating and obesity.50 Other studies however, have found normal ghrelin levels in non-obese children under 5 years of age,51 and infants52 with PWS compared with matched controls, although a subset of infants with PWS in the latter study were hyperghrelinemic and overall, the PWS group showed more heterogeneity in ghrelin levels. Reducing plasma-ghrelin level with somatostatin does not suppress appetite53 indicating that hyperghrelinemia is not the sole cause of the hyperphagia in PWS. The role of other gut hormones, such as peptide YY3−36, an anorexigenic hormone known to affect satiety,54 is not fully determined. Recently, high postprandial level of peptide YY has been shown in PWS children although fasting levels were similar to that of normal weight and obese controls.55 There is conflicting evidence of its role in adults,56 although plasma levels are thought to be appropriate for obesity (see Goldstone43). Long-term studies on the effects of age on gut hormones will determine the development of atypical regulatory systems more fully and it is likely that this will involve dysregulation in a number of systems.
Models to explain eating behaviour in PWS
Hunger vs satiety impairment
The behavioural depiction of eating in PWS suggests that individuals suffer from a continuing hunger and a consequent desperation to eat. However, an emergent literature suggests that the eating behaviour is more consistent with an abnormal satiety response, rather than an abnormal hunger.8 In line with this view, eating in PWS is more likely a consequence of not wanting to stop eating rather than always wanting to start. In one of the earliest studies to show this, 10 children with PWS took part in a test involving eating sandwich quarters. Compared with healthy weight controls, PWS children had similar initial rates of eating, but ate for a significantly longer time suggesting that there was a dysfunction of satiety signals once eating had begun.57 Subsequent studies have supported this finding.25, 26, 58, 59 Studies from our group25, 58 using an otherwise similar paradigm to Zipf and Bernston, asked 13 participants with PWS to rate their hunger, desire to eat and fullness, using a visual analogue scale. Although there was evidence of a shift in ratings in the expected direction, those with PWS had to eat on average three times more calories for the change in ratings to occur, and ratings returned towards those of the pre-meal state very shortly after food was no longer available. Reductions in ratings of hunger were found to correlate with increases in blood-glucose levels above the accepted normal range. A disturbance in satiety has also been implicated by non-decelerating eating curves seen in a PWS group compared with obese and normal weight controls.59 A study by Lindgren and colleagues, measuring meal consumption on a hidden scale, also showed that the PWS group had a slower initial eating rate than controls, further attenuating the theory that excessive hunger is the driving factor in PWS eating behaviour. Eating rate curves were accelerating or linear in 10% of the healthy weight controls and 30% of obese controls as compared with 56% of the PWS group. Moreover, the results from these behavioural studies show that the satiety response in PWS is delayed, rather than absent. In contrast, one study has shown that 75% of a group with PWS showed a decelerating eating curve.60 However, the findings of this study also show a normal eating rate, a normal reaction to food on a visual analogue scale measuring hunger and satiation and further, a failure to finish the food that they were given. The disparity between these findings and other studies is considerable, and it is likely that this study does not represent the picture in PWS and may be due to anomalies related to the experimental setting.
One of the key features of many brain imaging studies investigating eating in PWS is that most of the abnormalities described are most prominent after, rather than before eating, thus supporting the behavioural findings of irregular satiation. Recent neuroimaging studies show an irregular post-meal response with atypical brain activation post-meal,61 delayed activation of satiety centres following glucose administration62 and an abnormal or absent activation in areas associated with satiety post-meal,63 all providing support for a delayed satiation response. In a further analysis of the data from that sub-sample of people with PWS who reported changes in ratings of hunger and fullness during food intake, Hinton et al.63 found that there were changes in cortical activation from the lateral to the medial orbital frontal cortex. This observation further suggests that, whatever the cause of the abnormal satiety response, it is not absolute, rather it is impaired and delayed.
Abnormal satiety and inner physiological awareness
It is possible that the delayed satiety response is the result of a disturbance in the perception of inner physiological states. This would relate to the suggestion that people with PWS may have a problem with interoception, perception of internal sensation,64 as previously suggested (E Hinton, unpublished thesis). Interoceptive Awareness is a domain on the Eating Disorders Inventory, and is affected in obesity and anorexia nervosa.65, 66 On the Eating Disorders Inventory, Interoceptive Awareness measures the ability to differentiate sensations and feelings of hunger and satiety. The Eating Disorders Inventory has not been very widely used in PWS, as it is thought to be unable to capture the unique range of behaviours in PWS.67 General visceral awareness and interoception can be measured using a heartbeat perception task, as has been used in anorexia nervosa.66 To our knowledge, such a study has yet to be carried out in PWS.
A high pain threshold has long been noted in PWS and has recently been shown in experimental conditions.68 The authors of that study postulate that this may be due to the complex derangement of neurotransmitter balance involving the hypothalamic areas associated with PWS.42 Altered gamma-aminobutyric acidA receptor function in brain areas including the insula, a region associated with awareness of bodily states,69 which with gamma-aminobutyric acid-ergic abnormalities in other regions, may be involved in phenotypical characteristics, including altered pain sensation.70 A delayed signal reduction62 and significantly different responses to food before and after meal in the insula have also been shown.61 Any dysfunction in interoceptive awareness may be the result of the abnormal feeding problems noted in young children with PWS. In this respect, the sensation of normal feeding may never be experienced, thus the normal pattern of development of eating-related physiological awareness may be halted. Recent literature suggests that there is a stage of normalization in the eating behaviour within the transition between the early and later phenotypes.37, 38 However, it is possible that this stage of behavioural normalization is not long enough to compensate or correct for deficient physiological awareness. It is also possible that the eating behaviour during this stage of normalization is in some way qualitatively different to that of healthy peers, despite amounts eaten being perceived as normal.
Hyper-responsive reward systems
An abnormal reward response is one of the most discussed recent behavioural theories of obesity, with evidence that obese people may eat more than healthy peers because they get over and above the normal hedonic value from food. In its most potent form, the theory of reward lends support for considering eating in PWS as an addictive behaviour, with food being the substance of dependence or of abuse. In obesity in the general population, thinking of eating in terms of addiction and behavioural or neurobiological substance dependence is not a new concept; literature in this field is ever-increasing, although it remains a controversial topic.71 Addiction and substance abuse has also been reviewed in PWS in regard to eating.72 James and colleagues argue that redefining obesity as a disorder of substance dependence disorder would reduce stigmatization,73 although food is currently not accepted as a substance of abuse in the Diagnostic and Statistical Manual IV (DSM-IV-TR).74 We briefly consider the DSM-IV-TR criteria of substance dependence (as cited in Davison et al.75) for aspects of eating in PWS below:
tolerance: tolerance can be indicated by the substance's effects being markedly less if the normal amount is ingested. This is clearly shown by the large amount of food required for any degree of satiation.25
withdrawal: irritability and temper tantrums often result from being denied food.
substance taken for a longer time or greater amount than intended: difficulty in controlling food intake in PWS is common and often access has to be severely limited.
negative consequences: overeating continues unless restraints are in place despite negative consequences on health, employment and family life.
putting off or neglecting activities: eating often takes priority over tasks at work or at home.
spending significant time or emotional energy in obtaining: significant efforts in acquiring foods, concealing eating or planning to obtain foods are common.
desire to reduce or control use: many people with PWS, and certainly their families, attempt daily to cut down the overeating behaviour (although it is thought that many people with PWS will talk of their desire to cut down without actually trying to do so themselves).
Neuroimaging studies in PWS have shown a hyper-responsive reward circuit related to food.76 A delayed signal reduction after glucose ingestion has been shown in PWS in the hypothalamus, the ventromedial prefrontal cortex, insula and the nucleus accumbens of the ventral striatum,62 an area long established as a ‘pleasure centre’ and one involved in food preference.77 Further studies show greater reward circuit activation in response to high-calorie foods compared with intelligence quotient-matched controls.78 Recently, it was shown that the food motivation network is activated before and after meal to a larger extent in deletion subtypes than in UPD, suggesting a specific genetic effect on reward systems.79 Further to imaging studies, there is recent suggestion that aberrant mesolimbic dopamine signalling, a key neurotransmitter implicated in addiction, may be involved in PWS (as well as in bulimia nervosa and binge eating disorder).80 The case for addiction theories or at least of an abnormally high reward attachment with food in PWS is compelling. Anecdotal reports show that people with PWS who live in specialized residential homes often ‘swap’ their preoccupation with food for a preoccupation with smoking and cigarettes when food is not available, suggesting an overarching need to stimulate neural reward systems, preferably with food, but if not, with nicotine. Considering over eating in PWS as a disorder of addiction may open the possibility of exploring the efficacy of behavioural or pharmacological treatments currently available for substance abusers. It would also help to underline the importance of avoiding food cues, to which people with PWS are susceptible.81
With the application of functional scanning technologies it has been possible to move from behavioural observations to investigating underlying neural responses to food intake and how these correlate with observed behaviours and with ratings of hunger and satiation. Thus, in the context of these above theories, the question that arises is how can this be explained given what is known about the PWS genotype, early development in PWS and the peripheral and hypothalamic feeding pathways in PWS? We propose two models that we speculate could account for eating behaviour. The first is that there is a direct connection between the absence of expression of the ‘PWS gene’ and a disruption in the feeding pathways of the hypothalamus. PWS should, therefore, be seen as a consequence of abnormal hypothalamic development resulting in the many aspects of the phenotype that are suggestive of hypothalamic dysfunction (feeding irregularities, sex and growth hormone regulation and temperature instability). The second is that abnormal placental function of fetal origin results in abnormal energy transfer in utero and a resultant compensatory mechanism in other systems in fetal tissue that then sets an abnormally high ‘satiety threshold’—this becomes apparent once the infant is required to control his/her own energy balance. We examine each of these in turn.
Eating behaviour as a direct consequence of a genetically determined deficit in a hypothalamic feeding pathway
PWS clearly results in hypothalamic dysfunction. As described above many of the phenotypic characteristics of the syndrome are likely manifestations of such dysfunction. The question is whether the eating disorder in particular can be traced directly back to a genetic abnormality in the hypothalamic feeding pathways. Genetic causes of non-PWS extreme obesity and overeating continue to be identified, showing a genetic basis for direct disruption in the peripheral or hypothalamic feeding pathways. Starting with the periphery and working towards the hypothalamus: leptin deficiency, leptin receptor abnormalities, melanocortin-4 receptor and pro-opiomelanocortin abnormalities are among those that have been described.82 Melanocortin-4 receptor abnormalities are now the single most prevalent obesity syndrome, overtaking that previously claimed by PWS.83 Recently, a form of obesity has also been associated with deletions on chromosome 16p11.2.84 None of the neurotransmitters or their receptors identified so far that are known to be involved in these pathways have been found to be localized at 15q11-13 and, therefore, candidate genes for the eating disorder in PWS. It is possible that other genes involved in the feeding pathways are implicated in PWS, but not other forms of obesity. Histochemical and molecular studies of hypothalamic tissue obtained at post-mortem from people with PWS have been difficult to undertake due to problems obtaining the necessary tissue in the ideal condition. One particular finding, that there are reduced numbers of oxytocin-containing cells in the paraventricular nucleus of the hypothalamus, has been proposed as a possible cause of the eating behaviour.85 Given that the gene for oxytocin is located on chromosome 20, this deficit must, however, be an indirect consequence of the PWS genotype. Within the broader phenotype of PWS, central oxytocin deficiency may be of relevance,86 for example, in explaining anxiety and attachment behaviours, although whether it accounts for the eating disorder is open for question. Levels of neuropeptide Y and agouti-related polypeptide in the paraventricular nucleus have also been studied and found to be normal. At present, therefore, no direct or indirect link has been established between a candidate imprinted gene located at 15q11-13 and the feeding pathways of the hypothalamus.
Prenatal Problems and the ‘Thrifty Phenotype’
A number of fetal abnormalities have been described in PWS, including reduced fetal movement, evidence of abnormal fetal growth, abnormal fetal position at birth and severe hypotonia that is central in origin.8 Infants with PWS subsequently show poor suck and weight gain, with tube feeding often necessary. Two studies examining the early years of PWS indicate that abnormalities seen in PWS may be prenatal in origin.35, 36 Common findings between these studies included, small gestational weight, increased number of Caesarean sections, increased rate of polyhydramnios, suckling deficit and hypotonia. A low birth weight as opposed to length, found in these studies and others,37 has been interpreted as a prenatal deficiency in nutrition rather than in growth.36 This has been hypothesized as being due to placental abnormalities, which are yet to be investigated in full.
One theory, which can explain an early failure to thrive and reduced neonatal weight, is that the genetic defect in PWS leads to fetal starvation and consequently, abnormal brain development, perhaps as a result of inadequate placental function.87 We have suggested that the entire PWS phenotype could, therefore, be the result of a monogenic abnormality, and the failure of expression of the paternal allele of a maternally imprinted gene that is involved in energy homeostasis.87 Much work has been done in other fields to correlate fetal nutritional abnormalities and subsequent adult-onset disorders. The Barker hypothesis, otherwise known as the ‘Thrifty Phenotype hypothesis’,88 has been proposed to explain the relationship between impaired fetal growth and the risk of developing abnormal glucose tolerance89 and metabolic syndrome.90 More recently, nutritional problems in the womb and in early life, leading to under-nourished neonates, have been associated with a higher rate of many diseases, including obesity.91 We speculate that in PWS, the Barker hypothesis could be extended to explain the high threshold and delay in satiety resulting from fetal starvation.
The fetal starvation theory postulated by our group is part of a larger hypothesis that PWS can be conceptualized as a disorder of starvation that presents as obesity in a food-rich environment. As there is no enduring sense of satiety due to failed neurological mechanisms, the body may incorrectly perceive itself to be in a state of starvation and to eat whenever food is available, which coincides with the theories of impaired interoception. In support of PWS as a model for starvation, there are metabolic commonalities between anorexia nervosa and PWS, including reports of elevated levels of ghrelin.92 There is much to suggest that ghrelin levels at birth are normal, as discussed earlier. Other similarities with anorexia include reduced metabolic rate and reduced gonadotropin release.93 Studies with mouse models of PWS also provide support for early starvation theory with neonatal failure to thrive and the later hyperphagic phenotype resultant from a failure of compensatory mechanisms,94, 95 although the latter study did not find problems with placenta of the mice. We have suggested that in PWS patients, fetal starvation may be related to the food intake reducing hormone leptin in the placenta.87 Leptin and its receptors are widely expressed in the placenta and may be related to fetal size, fetal tissue development and short- and long-term energy control.96 Leptin is a key anorexigenic hormone fundamental to the energy homeostasis97 and in the interplay of gut, vagal and brain regions regulating satiation.98 Congenital leptin deficiency results in PWS-like symptoms, including hyperphagia and hypogonadism, which is reversed, often dramatically, by recombinant leptin therapy.99, 100 At first glance, it may be assumed that leptin therapy would be of use in treating PWS. However, in PWS, leptin levels are as expected for fat mass and leptin genes are located on chromosome 7. Functional neuroimaging studies of people with congenital leptin deficiency have shown that leptin modulates neural activity in circuits involving food intake to reduce the perception of food reward and increase response to satiety signals.101 Therefore, if people with PWS are in fact insensitive to leptin, as we predict that they are, then it follows that their reward circuitry in response to food may be hyper-responsive as is shown in imaging studies. Together, the evidence suggests that people with PWS are insensitive to leptin, possibly because of an altered feedback mechanism as a result of early metabolic abnormalities and fetal starvation.
To date, attempts to curb the life-threatening hyperphagia have been unsuccessful. Appetite-reducing drugs used in simple obesity have not shown wide-spread improvements,102 and bariatric procedures have been associated with severe complications.103, 104 Growth hormone treatment has been shown to improve the fat-muscle mass ratio,105 and may help to avoid the obesity for the younger generations of people with PWS, but has no effect on eating behaviour. As parents are now aware of the implications of PWS from very early on in their child's life, eating management advice and discipline may help delay problems in future. However, at present, the only available control for hyperphagia for most people with PWS is close mentoring by carers and life-long restricted access to food, with food cupboards and refrigerators often being locked.29 Understanding the characteristics and development of hyperphagia is key to improving its management and, ultimately, its prevention through effective treatments. However, from a positive perspective, studies of neurodevelopmental disorders in general have increasingly focused on symptomology and how treatment might address one particular aspect of the phenotype (for example as in fragile-X syndrome and tuberous sclerosis). The same is clearly true in PWS. Growth hormone treatment changes the physical phenotype and corrects short stature, treatments are available for the associated psychiatric disorders,106 and more sophisticated models of understanding of the behavioural problems have been developed.107 Considering the eating disorder as a stand-alone aspect of PWS will likely lead to similar advances.
We have reviewed theories that may help to elucidate the eating behaviour in PWS. In summary we have considered overeating in PWS as a problem of satiety as opposed to hunger; the role of inner physiological awareness and its potential effects on feelings of hunger and satiation; hyper-responsive reward systems and food as a substance of abuse; the direct consequence of genetics on the hypothalamic feeding pathway; and the role of the prenatal environment. The eating behaviour in PWS is complex, and we note that these theoretical models do not have to be regarded as mutually exclusive. We propose that all of the models described can in some way be encompassed by conceptualizing PWS as a model for starvation.87 In this case, it is hypothesized that the genetic abnormality could involve a gene regulating metabolism, which may result in an inadequate placenta and pre-setting of thresholds of anorexigenic hormones, such as leptin. The genetic abnormality may be the cause of fetal malnourishment or fetal starvation, leading to neonatal failure to thrive and problems with feeding. Ghrelin may be involved in the instigation of the binging and hyperphagic stage, and later development of atypical reward circuitry in response to food may be the result of altered pathways generated in the early binging stages in childhood, combined with insensitivity to satiation cues, such as leptin levels. Examining reward pathways by neural imaging in response to food in younger children, for example, may elucidate the development of the hyper-responsive circuitry.
Studies in the past years have greatly increased our knowledge about the neural, endocrinological and behavioural aspects of PWS. Reviewing behavioural advances, further strengthened by neural data, will help define and focus how PWS is conceptualized, for example, to think of PWS as a disorder of satiety rather than hunger, and to acknowledge the role of reward systems. Much is yet to be discovered about the behavioural, neural, genetic and endocrinological aspects of the syndrome, and a multi-disciplinary approach will be necessary in considering treatment or management strategies. It is hoped that evaluating the theories to explain PWS will aid the direction of future studies.