J Clin Endocrinol Metab. 2005 May.
Fasting and postprandial hyperghrelinemia in Prader-Willi syndrome is partially explained by hypoinsulinemia, and is not due to peptide YY_3-36 deficiency or seen in hypothalamic obesity due to craniopharyngioma.
Goldstone AP, Patterson M, Kalingag N, Ghatei MA, Brynes AE, Bloom SR, Grossman AB, Korbonits M.
Department of Endocrinology, St. Bartholomew's Hospital, London, United Kingdom.
[ PubMed ] [ Free full text ]

Abstract: The cause of the unique elevation in fasting plasma levels of the orexigenic gastric hormone ghrelin in many patients with Prader-Willi syndrome (PWS) is unclear. We measured fasting and postprandial plasma ghrelin in nonobese (n = 16 fasting and n = 8 postprandial) and obese non-PWS adults (n = 16 and 9), adults with genetically confirmed PWS (n = 26 and 10), and patients with hypothalamic obesity from craniopharyngioma tumors (n = 9 and 6). We show that 1) plasma ghrelin levels decline normally after food consumption in PWS, but there is still fasting and postprandial hyperghrelinemia relative to the patient's obesity (2.0-fold higher fasting ghrelin, 1.8-fold higher postprandial ghrelin, adjusting for percentage of body fat); 2) the fasting and postprandial hyperghrelinemia in PWS appears to be at least partially, but possibly not solely, explained by the concurrent relative hypoinsulinemia and preserved insulin sensitivity for the patient's obesity (residual 1.3- to 1.6-fold higher fasting ghrelin, 1.2- to 1.5-fold higher postprandial ghrelin in PWS, adjusting for insulin levels or homeostasis model assessment of insulin resistance); 3) hyperghrelinemia and hypoinsulinemia are not found in craniopharyngioma patients with hypothalamic obesity, and indeed, these patients have relative hyperinsulinemia for their obesity; and 4) there is no deficiency of the anorexigenic intestinal hormone peptide YY(3-36) in PWS contributing to their hyperghrelinemia.

Excerpts from the full text article

Introduction

Prader-Willi syndrome (PWS) is a genetic cause of hyperphagia and obesity thought to arise from developmental defects in the brain, including the hypothalamus (1). Many PWS subjects have elevated fasting plasma levels of the stomach-derived GH secretagogue ghrelin, especially when assessed relative to their body mass index (BMI) or total adiposity, with obesity itself normally associated with reduced plasma ghrelin (2, 3, 4, 5, 6). Given the orexigenic and metabolic actions of ghrelin, it has been suggested that this chronic elevation in ghrelin levels could contribute to phenotypes such as hyperphagia, GH deficiency, or sleep disturbance in some PWS subjects, although this currently remains unproven (7, 8, 9).

The cause of hyperghrelinemia in PWS relative to their obesity is unknown. In a previous study it appeared partly explicable by relative hypoinsulinemia and preserved insulin sensitivity in PWS, which in itself may reflect reduced visceral adiposity (5, 10, 11), but it does not appear to be related to the concurrent GH deficiency seen in PWS (12, 13). A lack of the normal postprandial suppression of plasma ghrelin has also been reported in PWS adults, which could theoretically contribute to early return of hunger after a meal (3). However, studies in children with PWS have shown normal postmeal ghrelin suppression (14, 15).

Obesity and hyperphagia are common sequelae to intracranial tumors involving the hypothalamus, such as craniopharyngioma (16). Vagally mediated hyperinsulinemia and autonomic imbalance are also thought to contribute to hypothalamic obesity from craniopharyngioma (17). Such obesity may respond to somatostatin analogs, perhaps through reductions in insulin secretion (18). Somatostatin and its analogs also reduce ghrelin secretion in non-PWS and PWS subjects (14, 19, 41). Peptide YY3–36 (PYY) is an anorexigenic hormone secreted postprandially from the distal intestine that reduces plasma ghrelin (20). Reduced PYY secretion in obesity and increased PYY secretion after gastric bypass surgery to treat obesity may play pathogenic roles in alterations in appetite and food intake (20, 21).

We therefore hypothesized that 1) hyperghrelinemia might also be seen in hypothalamic obesity due to craniopharyngioma; 2) hyperghrelinemia in PWS is caused by PYY deficiency, which could also contribute to obesity in PWS; and 3) there is abnormal suppression of plasma ghrelin after meals in PWS adults. We therefore measured fasting and postprandial plasma ghrelin and PYY in control, PWS, and craniopharyngioma adults with and without hypothalamic obesity.

Subjects and Methods

Recruitment

... All subjects were over 18 yr of age and were not known to be diabetic, none had a fasting glucose level greater than 6.0 mmol/liter (108 mg/dl), and in the postprandial study all had a peak glucose level less than 9.8 mmol/liter (176 mg/dl) and a 2-h postprandial (77 g carbohydrate) glucose level less than 8.3 mmol/liter (149 mg/dl). Non-PWS and noncraniopharyngioma females were premenopausal. All PWS subjects had positive genetic testing: fasting study: eight exact molecular class unknown (e.g. only methylation pattern studied), 10 ch15q11-q13 deletion, five maternal uniparental disomy (UPD), two UPD or imprinting center defect, and one unbalanced chromosomal translocation (46,XYt15:Y); postprandial study: three exact molecular class unknown, four ch15q11-q13 deletion, two UPD, and one UPD or imprinting center defect. PWS subjects had not had GH stimulation testing or GH day profiles measured, but IGF-I levels were available for all PWS subjects in the postprandial study. Of these, 10% had IGF-I levels less than 2 SD below the age-related median reference value (<120–126 ng/ml), 40% between –2 and –1 SD (<155–173 ng/ml), 20% between –1 SD and the median (<189–220 ng/ml), and 30% between the median and 1 SD (<260–294 ng/ml).

[...]

Results

Fasting study

Clinical information for those subjects who had only fasting blood sampling is given in Table 1. In view of significant differences between some comparison groups, adjustment was made for age, sex, percentage of body fat, fasting insulin, or HOMA-IR by multiple linear regression analysis when comparing plasma ghrelin between groups.

OB, CRHO, and PWS subjects had significantly greater percentage of body fat than NO subjects, and although percentage of body fat was similar in PWS and CRHO subjects, it was slightly lower in PWS than OB subjects (Table 1 and Fig. 1A). However, although OB and CRHO subjects both had significantly higher fasting insulin and HOMA-IR than NO subjects, values were similar in NO and PWS subjects (Table 1 and Fig. 1B). Fasting insulin and HOMA-IR levels were lower in PWS than in OB or CRHO subjects (Table 1 and Fig. 1B), and this remained significant when adjusting for age, sex, and percentage of body fat (all P < 0.001).

Plasma ghrelin was negatively correlated with percentage of body fat in NO and OB subjects combined (r = –0.68; P < 0.001), and all subjects with craniopharyngioma (CR and CRHO; r = –0.53; P < 0.05), but not in PWS subjects (r = 0.04; P = 0.84; Fig. 1D). Plasma ghrelin was negatively correlated with fasting insulin (r = –0.53, P = 0.002; r = –0.52, P < 0.05; and r = –0.53, P = 0.006, respectively), and HOMA-IR (r = –0.52, P = 0.003; r = –0.47, P = 0.08; and r = –0.49, P = 0.01, respectively) in all three groups (Fig. 1E). In NO and OB subjects, the negative correlation of fasting ghrelin with percentage of body fat remained significant when including fasting insulin as a variable in multiple regression analysis (P = 0.003).

Fasting plasma ghrelin was lower in OB and CRHO subjects compared with NO subjects, but was similar in NO and PWS subjects (Table 1 and Fig. 1C). Fasting plasma ghrelin was higher in PWS than in OB or CRHO subjects (Table 1 and Fig. 1C), and this remained significant (2.2- and 2.9-fold higher, respectively) when correcting for age, sex, and percentage of body fat (Table 2). In expanded datasets, fasting plasma ghrelin was 2.0-fold higher in PWS subjects compared with either NO and OB subjects combined or all non-PWS subjects, adjusting for age, sex, and percentage of body fat (Table 2 and Fig. 1D).

There was no significant correlation of fasting plasma PYY with percentage of body fat in NO and OB (r = 0.05; P = 0.80), PWS (r = 0.23; P = 0.28), CR (r = 0.17; P = 0.75), and CRHO (r = –0.14; P = 0.71) subjects; all craniopharyngioma subjects (r = –0.02; P = 0.95), or all subjects (r = 0.07; P = 0.571). Fasting PYY in PWS was not significantly different from that in OB or CRHO subjects (Table 1) or when adjusted for age, sex, and percentage of body fat (both P = 0.6), fasting insulin (P = 0.8–0.9), or HOMA-IR (both P = 0.7).

Postprandial study

Clinical information for those subjects who had postprandial blood sampling is given in Table 3. There was no significant difference in percentage of body fat between OB, CRHO, and PWS subjects (P = 0.6–0.9). Fasting insulin and FHOMA-IR were lower in PWS than in either OB or CRHO subjects, whereas postprandial AUC insulin or PHOMA-IR in PWS subjects were lower than those in CRHO subjects and tended to be lower than those in OB subjects (P = 0.08; Table 3 and Fig. 2B). Indeed, peak postprandial insulin, postprandial insulin AUC, and PHOMA-IR were higher in CRHO than in OB subjects (Table 3 and Fig. 2B).

In NO and OB subjects combined, fasting, postprandial trough, and AUC plasma ghrelin levels were negatively correlated with percentage of body fat (r = –0.83, P < 0.001; r = –0.77, P < 0.001; and r = –0.79, P < 0.001, respectively). There was a stronger negative correlation of fasting ghrelin with postprandial insulin AUC than with fasting insulin or FHOMA-IR in NO and OB subjects combined (Table 4). Similarly, there was a stronger negative correlation of postprandial ghrelin levels with postprandial insulin AUC levels or PHOMA-IR than with either fasting insulin or FHOMA-IR (Table 4).

Fasting and postprandial plasma ghrelin levels in PWS subjects were higher than those in OB and CRHO subjects, but not significantly different from those in NO subjects (Table 3 and Fig. 2A). PWS subjects had 2.0-fold higher fasting ghrelin and 1.8-fold higher postprandial ghrelin levels compared with NO and OB subjects combined, after adjusting for age, sex, and percentage of body fat (Table 5). When adjusting for insulin or HOMA-IR measurements (instead of percentage of body fat), fasting ghrelin levels were significantly higher (1.6-fold) in PWS when adjusting for postprandial insulin levels or PHOMA-IR (Table 5). However, the increase (1.3- to 1.6-fold) in fasting ghrelin in PWS did not quite reach statistical significance when adjusting for fasting insulin or FHOMA-IR (Table 5). Similarly, postprandial ghrelin levels were significantly higher (1.4- to 1.5-fold) in PWS than NO and OB subjects after adjusting for postprandial insulin levels or PHOMA-IR, but the increase (1.2-fold) was not significant when adjusting for fasting insulin or FHOMA-IR (Table 5). This is consistent with postprandial insulin and PHOMA-IR having stronger correlations than fasting insulin and FHOMA-IR with ghrelin levels in non-PWS subjects, as noted above (Table 4).

PWS subjects also had higher plasma ghrelin compared with CRHO subjects, after being adjusted for age, sex, and percentage of body fat: 2.4-fold higher fasting ghrelin (1.2–4.9; P = 0.04), 2.4-fold higher postprandial trough ghrelin (1.2–5.2; P = 0.04), and 2.1-fold higher postprandial AUC ghrelin (1.0–4.3; P = 0.05). However, fasting and postprandial ghrelin levels were not significantly different between PWS and CRHO subjects after adjusting for age, sex, and fasting or postprandial insulin levels or HOMA-IR (P = 0.2–0.7). Mean ghrelin levels were similar in OB compared with CRHO subjects (Table 3) and after adjusting for age, sex, and percentage of body fat (P = 0.2–0.7).

The postprandial fall in ghrelin was significant in all groups (Table 3, basal vs. trough: NO, P = 0.002; OB, P = 0.001; CRHO, P = 0.02; PWS, P < 0.001; by paired t test). The maximum percent postprandial fall in ghrelin was less in OB than NO subjects (Table 3). In NO and OB subjects combined, the maximum percent postprandial fall in ghrelin was positively correlated to basal ghrelin (r = 0.73; P = 0.001) and tended to be negatively correlated with fasting insulin (r = –0.42; P = 0.10) and FHOMA-IR (r = –0.42; P = 0.07), but not with postprandial insulin AUC (r = –0.19; P = 0.48), PHOMA-IR (r = –0.13; P = 0.63), maximum percent postprandial increase in insulin (r = 0.27; P = 0.30), or maximum percent postprandial increase in PYY (r = –0.15; P = 0.56). Similarly, there was no significant correlation between the maximum absolute postprandial change or postprandial incremental AUC for ghrelin and insulin or PYY (P = 0.3).

In PWS subjects, ghrelin fell postprandially by a similar percentage as that in NO subjects (P = 0.7), but by a greater percentage than that in OB subjects (Table 3). The maximum percent fall in ghrelin in PWS subjects was not significantly different from that in NO and OB subjects combined after adjusting for age, sex, baseline ghrelin, fasting or postprandial insulin, or HOMA-IR (P = 0.5–1.0).

There was no significant difference in fasting, postprandial peak, or AUC plasma PYY levels between PWS and other groups (Table 3) or after adjusting for age, sex, and percentage of body fat (P = 0.1–0.9).

Discussion

Elevated plasma ghrelin after fasting and before meals may play a role in meal initiation (24), and its decline after food intake may act together with increased secretion of anorexigenic gut hormones, such as PYY, to limit subsequent food intake (20). Ghrelin secretion is inhibited by insulin and PYY (20, 25, 26, 27). Obesity is usually associated with reduced fasting ghrelin, probably through chronic hyperinsulinemia (5, 28, 29).

In PWS subjects, there is fasting hyperghrelinemia relative to the degree of obesity, although there is not always significant hyperghrelinemia when comparing absolute levels with nonobese control subjects (2, 3, 4, 5, 6). In this study we show that 1) plasma ghrelin levels decline normally after food intake in PWS subjects, but there fasting and postprandial hyperghrelinemia still exists relative to their obesity; 2) the fasting and postprandial hyperghrelinemia in PWS appear to be at least partially, but possibly not solely, explained by their concurrent relative hypoinsulinemia and preserved insulin sensitivity; 3) fasting or postprandial hyperghrelinemia and hypoinsulinemia are not found in craniopharyngioma patients with hypothalamic obesity, and indeed, they have relative hyperinsulinemia; and 4) there is no PYY deficiency in PWS subjects contributing to the hyperghrelinemia.

Obesity and insulin resistance in PWS and craniopharyngioma

Obesity is usually associated with the metabolic syndrome, consisting of a spectrum of detrimental phenotypes, including insulin resistance and hypertriglyceridemia, with increased risk of diabetes mellitus and cardiovascular disease, particularly mediated by increased visceral adiposity (30). We found lower fasting insulin levels and FHOMA-IR in PWS compared with OB subjects and a tendency for lower postprandial insulin and PHOMA-IR. This confirms the findings of several other studies that the metabolic complications of obesity are surprisingly reduced or absent in PWS adults and children, with preservation of insulin sensitivity (5, 10, 31, 32, 33). Possible explanations include childhood-onset GH deficiency and/or a selective reduction in visceral adiposity in PWS adults (10, 11, 31). This preserved insulin sensitivity was not seen in subjects with CRHO, and indeed, postprandial insulin levels and PHOMA-IR were even higher in CRHO patients than in similarly obese OB and PWS subjects, consistent with previous reports (17).

Hyperghrelinemia and adiposity in PWS

A negative relationship between fasting ghrelin levels and overall adiposity was seen in non-PWS, but not PWS, subjects, in agreement with our earlier study, which also showed a significant negative correlation with magnetic resonance imaging (MRI)-determined visceral adiposity in both non-PWS and PWS subjects (5). This is explicable by the unusual relationship between visceral adiposity and overall adiposity in PWS adults (5, 10, 11). Fasting ghrelin levels in PWS adults were 2.0-fold higher than those in non-PWS adults after correcting for total adiposity, consistent with earlier studies (2, 3, 4, 5), and our study also found postprandial ghrelin levels to be 1.8-fold higher in PWS patients after correcting for total adiposity.

A potential criticism of this analysis is our use of bioimpedance analysis (BIA) to measure total adiposity in PWS subjects, because there may be changes in the compart-mentalization of body water in disease states such as GH deficiency. However, we found that there are excellent and parallel correlations between percentage of body fat measurements determined by BIA and whole body MRI in both non-PWS women (n = 44; r = 0.93; P < 0.001) and PWS women (n = 13; r = 0.93; P < 0.001) (Goldstone, A. P., E. L. Thomas, A. E. Brynes, G. Frost, J. D. Bell, unpublished observations). BIA did, however, slightly underestimate MRI-determined percentage of body fat in PWS compared with non-PWS women by an absolute value of 3.2 ± 1.1% (P < 0.005). This underestimate of overall adiposity by BIA in PWS subjects would therefore have, if anything, tended to underestimate the degree of hyperghrelinemia in PWS subjects relative to overall adiposity. This may have contributed to the finding of a lower degree of hyperghrelinemia in PWS subjects relative to adiposity seen in the current study using BIA compared with our previous study in which fasting ghrelin levels were increased 3.4- to 3.6-fold relative to MRI-determined total adiposity (5). The use of different ghrelin assays in these two studies is a potential additional factor.

Hyperghrelinemia and preserved insulin sensitivity in PWS

Furthermore, our previous study and others have found stronger negative correlations of fasting ghrelin levels with insulin levels or insulin resistance than with overall adiposity in non-PWS subjects (5, 28, 29). However, unlike overall adiposity, there was a significant negative correlation of fasting ghrelin with fasting insulin levels and HOMA-IR in both non-PWS and PWS subjects in this and our previous study (5). Adjustment for differences in insulin levels or insulin resistance levels in comparison of ghrelin levels between groups therefore circumvents any confounding factors introduced by the use of BIA for body composition analysis. Postprandial ghrelin levels were also negatively correlated with insulin levels and HOMA-IR in non-PWS subjects. Interestingly, we found a stronger negative correlation between fasting or postprandial ghrelin levels and postprandial than fasting insulin in non-PWS subjects. This difference may be related to the repeated measurements of postprandial insulin values reducing statistical variability compared with fasting values, or postprandial insulin levels giving a better indication of the prevailing chronic hyperinsulinemic environment. Postprandial hyperinsulinemia is also a better predictor than fasting hyperinsulinemia of the risk for metabolic syndrome and coronary artery disease (34).

We found that at least part, but perhaps not all, of the explanation for both the fasting and postprandial hyperghrelinemia in PWS may be these patients’ relative hypoinsulinemia and preserved insulin sensitivity (5, 10, 11). Thus, fasting ghrelin levels were 1.3- to 1.8-fold higher in PWS, adjusting for fasting insulin or HOMA-IR, although this did not always reach statistical significance, probably as a result of the smaller sample numbers in some datasets. When adjusting for postprandial insulin or HOMA-IR, fasting ghrelin levels were 1.6-fold higher, and postprandial ghrelin levels were 1.4- to 1.5-fold higher in PWS. These results suggest a lower degree of hyperghrelinemia than in our earlier study (5), in which fasting ghrelin levels were 3.0-fold higher in PWS after adjusting for fasting insulin or HOMA-IR, which could reflect the use of different ghrelin and insulin assays in these two studies.

Nevertheless, the available evidence of persistent hyperghrelinemia in PWS even when adjusting for simultaneous differences in insulin levels or sensitivity from these two studies does suggest that an additional cause(s) may be present, although the effect may be smaller than previously considered and before adjustment for the hypoinsulinemia. This conclusion that factors additional to hypoinsulinemia contribute to hyperghrelinemia in PWS is also suggested by other studies in children. Two studies have shown that mean fasting ghrelin levels in PWS children tend to be higher than those in lean non-PWS children despite the PWS children having higher mean fasting insulin levels than these less obese non-PWS subjects, although this interpretation is complicated by the lack of formal covariate analysis and, in one study, genetic confirmation of PWS (4, 15).

Measurement of ghrelin levels in PWS children at different stages of development and in larger numbers of PWS adults after correction for prevailing insulin levels will be needed to confirm that there are factors additional to hypoinsulinemia that cause hyperghrelinemia in PWS.

Additional problems in this interpretation are 1) the use of surrogate markers of total insulin secretion and insulin sensitivity (fasting or postprandial plasma insulin or HOMA-IR) in our study; and 2) the fact that other unidentified circulating factors that are normally associated with insulin resistance, such as adipocytokines, could contribute to low ghrelin concentrations in obesity, with low insulin levels and HOMA-IR merely a marker of another regulatory factor that is abnormal in PWS, resulting in both improved insulin sensitivity and hyperghrelinemia (29, 35). Assessment of the relationship between other measures of insulin sensitivity and adipocytokines with ghrelin in PWS will therefore be of interest.

Intact regulatory influences on ghrelin secretion in PWS

Plasma ghrelin levels decrease postprandially by a smaller amount in OB than NO non-PWS subjects, in agreement with other studies (36, 37). In PWS subjects, plasma ghrelin fell postprandially by 32%, which appeared appropriate for their fasting ghrelin levels. This normal postprandial fall of ghrelin in PWS adults is in agreement with recent studies of PWS children, but contradicts a single study in PWS adults that only examined one postprandial time point (3, 14, 15). It suggests a normal response of ghrelin-secreting cells to hormonal or neural mediators in the postprandial state (38). Interestingly, the mediator does not appear to be postprandial secretion of insulin itself, although insulin may provide a permissive environment for the postprandial fall (39, 40). This is supported by the lack of any significant positive correlation between the postprandial fall in ghrelin and the postprandial increase in insulin in non-PWS subjects in our study.

Combined with the 1) normal negative correlation of plasma ghrelin with visceral adiposity and insulin levels in PWS in this and our earlier study (5), and 2) similar falls in plasma ghrelin after somatostatin or octreotide therapy in PWS (14, 41) as in other studies of non-PWS subjects (19, 42), this suggests that the cause of hyperghrelinemia in PWS is not an intrinsic primary abnormality of ghrelin-secreting cells, but, more likely, the loss of an inhibitory, or excess of a stimulatory, neural or hormonal input. Increased nongastric expression of ghrelin in PWS remains another possibility requiring investigation (43).

Hyperghrelinemia and PYY secretion in PWS

Although the anorexigenic intestinal hormone PYY acutely reduces fasting and postprandial ghrelin levels in non-PWS subjects (20), the normal fasting and postprandial plasma levels of PYY in PWS in our study have excluded PYY deficiency as contributing to hyperghrelinemia in PWS. PYY is secreted from the gut in proportion to calories consumed. Although there was a vigorous postprandial elevation in PYY levels in PWS subjects after eating a much larger meal (mean ± SEM, 1737 ± 538 kcal) in our recent study (41), the absence of any comparison with a control group means that we cannot definitively exclude the possibility that impaired PYY release contributes to delayed satiety and earlier return of hunger in PWS after larger meals (44). The lack of any significant correlation between the postprandial fall in ghrelin and the postprandial increase in PYY in non-PWS subjects in the current study suggests that the release of PYY may not be responsible for the postprandial fall in ghrelin secretion, at least with the size and macronutrient nature of the meal used in our study (20).

Other possible contributions to hyperghrelinemia in PWS

It remains possible that other factors contribute to the residual elevation of ghrelin levels in PWS compared with control subjects in addition to differences in insulin levels. These include 1) changes in other gut hormones that are known to alter ghrelin secretion (26, 27, 41, 45); 2) congenital GH or IGF-I deficiency, which has been associated with hyperghrelinemia in examples other than PWS (6, 46, 47); and 3) defects in neural inputs regulating ghrelin secretion from the stomach, because abnormal cardiac, pupillary, and pancreatic autonomic innervation have been suggested by some, but not all, studies in PWS (48, 49, 50, 51, 52), although the presence and nature of any autonomic control of ghrelin secretion in humans is unclear (53, 54, 55, 56, 57).

Hormonal differences between PWS and craniopharyngioma

The finding that hyperghrelinemia and relative hypoinsulinemia are seen in patients with PWS, but not CRHO, suggests significantly different pathophysiologies. These hormonal differences may be related to CRHO subjects: 1) lacking the factors preserving insulin sensitivity and reducing visceral adiposity in PWS (10, 11, 31); 2) having altered autonomic innervation of pancreatic ί-cells and gastric and other peripheral tissues as a result of different hypothalamic defects or a lack of the other neural defects seen in PWS; 3) having a different balance between parasympathetic and sympathetic nervous activity (1, 17); 4) having damage primarily to the basal hypothalamus, because tumor arises from the suprasellar region, akin to that in ventromedial hypothalamus-lesioned rodents (16, 17), whereas in PWS the basal infundibular nucleus appears normal, but there are abnormalities in the more dorsal paraventricular nucleus with reduced total and reduced oxytocin cell number (1, 58); 5) having nonphysiological cortisol dynamics contributing to postbreakfast hyperinsulinemia given the pharmacokinetics of oral glucocorticoid replacement for ACTH deficiency; and 6) having a later age of onset of GH deficiency than PWS subjects, in whom GH deficiency appears to be present from early infancy (59), because the average age of craniopharyngioma diagnosis or initial treatment was between 22 and 30 yr in our CRHO patients.

The hormonal differences between CRHO and PWS (relative hyperinsulinemia and hypoghrelinemia in CRHO vs. relative hypoinsulinemia and hyperghrelinemia in PWS) also have implications for the potential treatment of hypothalamic obesity with somatostatin analogs (14, 18). Somatostatin analogs may benefit CRHO patients by reducing hyperinsulinemia and, hence, insulin-mediated adipogenesis, but this may be less effective in PWS, because there is already relative hypoinsulinemia.

Interestingly hypoghrelinemia and relative hyperinsulinemia are also seen in subjects with mutations in the melanocortin-4 receptor located in the hypothalamus and other brain regions, distinguishing this monogenic cause of human obesity from PWS (2, 4, 60).

Hyperghelinemia and hyperphagia in PWS

Increasing ghrelin levels by 2- to 4-fold in non-PWS subjects increases acute food intake by about 30% (7, 8). Mean ghrelin levels may be elevated by a similar amount in PWS subjects relative to their obesity (2, 3, 4, 5, 6), which may contribute to inappropriate hyperphagia despite obesity. However, an exclusive or even major role for hyperghrelinemia in causing hyperphagia in PWS is questioned by the 1) lower degree of hyperghrelinemia in PWS when correcting for insulin levels, 2) frequent absence of significant elevations in mean ghrelin levels in PWS vs. non-PWS NO subjects, 3) frequent overlap between ghrelin levels between individual PWS and non-PWS subjects over the range of obesity and insulin sensitivity, and 4) the magnitude and near universal presence of hyperphagia in PWS (61, 62). Furthermore, a preliminary study has failed to show any acute anorexigenic effect of normalizing ghrelin levels with a somatostatin infusion in four PWS male adults, although this was complicated by a simultaneous reduction in PYY secretion (41).

The orexigenic effect of brain, particularly hypothalamic, and/or other hormonal abnormalities in PWS may override changes in ghrelin (1). Although neuropeptide Y and agouti-related protein neurons, vital hypothalamic targets for the orexigenic action of ghrelin, appear normal in PWS, it is unknown if other defects in PWS make brain appetite pathways hypo-, hyper-, or normosensitive to changes in circulating ghrelin (1, 58, 63, 64). Chronic studies of somatostatin analogs in PWS and particularly the development of ghrelin antagonists will be necessary to definitively investigate any role for hyperghrelinemia in the hyperphagia of PWS and other phenotypes, such as GH deficiency and sleep disturbance (1, 9).

Conclusion

Fasting and postprandial hyperghrelinemia relative to the degree of obesity is a feature of PWS adults, but not of patients with hypothalamic obesity due to craniopharyngioma. Ghrelin levels fall postprandially by an amount appropriate for their baseline levels in PWS adults. Relative hypoinsulinemia and preserved insulin sensitivity are also features of PWS, but not of craniopharyngioma (where there is, in fact, relative hyperinsulinemia), and this may explain at least some, but perhaps not all, of the hyperghrelinemia in PWS. There is no evidence that impaired secretion of PYY contributes to the hyperghrelinemia in PWS, or that hyperghrelinemia contributes to obesity resulting from hypothalamic damage in craniopharyngioma.