An association between hypothalamic-pituitary dysfunction and peripheral endocrine function in extreme obesity
Post on 26-Sep-2016
Clinical Endocrinology (1991) 35, 97-102 AWNIS 0300066491001042
An association between hypothalamic-pituitary dysfunction and peripheral endocrine function in extreme obesity
J. U. Weaver, K. Noonan and P. G. Kopelman Medical Unit, The Royal London Hospital, London, UK
(Received 13 December 1990; returned for revision 15 January 1991; finally revised 4 February 7991; accepted 27 February 1991)
OBJECTIVE The alm was to investigate a posslble reiatlon- ship between measures of lnsulln secretion and glucose disposal and hypothalamic-pltuitary functlon In extreme obeslty. DESIGN A cross-sectlonal analysis of obese subjects attendlng the Obesity Clinic at the Royal London Hospital and normal weight volunteers was undertaken. Investlga- tlons were performed on separate occasions and In random order. PATIENTS The subjects were 34 extremely obese women, menstruating and with normal glucose tolerance (mean Body Mass Index, BMI =42) and 15 normal welght female controls (mean BMI = 22) MEASUREMENTS The followlng were measured: fastlng Insulln, relatlve insulin resistance calculated uslng fasting Insulin and plasma glucose by the homoeostatlc model of assessment, Insulin release during a 75-g oral glucose tolerance test (Insulin area under the curve), steady-state plasma glucose level achieved during a simultaneous Intravenous Infusion of dextrose, insulin and somatosta- tin, and the prolactin and growth hormone (OH) responses to Insulin-Induced hypoglycaemia. RESULTS In the obese group an Impaired prolactin response to hypogiycaemla (mean area under the curve obese 54 U/I min, controls 155 UII min; P=O.OOOl) was Inversely correlated to fasting Insulin, P= 0.142, P = 0.03; relatlve Insulin reslstance, P= 0.134, P= 0.03 and steady- state plasma glucose level, P= 0.345, P = 0.0004 whereas the impaired GH response (mean GH area under the curve obese 1.9 U/I min, controls 65.7 U/I min; P=O.OWl) was inversely correlated to steady-state plasma glucose level, P = 0.196, P = 0.01. Backward procedure for stepwlse regression analysis conflrmed the steady-state plasma glucose level to be the most Important varlable assoclated
Correspondence: P. G. Kopelman, Medical Unit, The Royal London Hospital, Whitechapel, London El IBB, UK.
wlth the prolactin and growth hormone response among the remaining indices of insulin secretlon/reslstance. CONCLUSION We conclude from these flndlngs that hyper- lnsullnaemla in obeslty Is an Important assoclatlon with altered hypothalamic-pituitary function Indicated by Impaired prolactin and growth hormone secretlon to Insulln-Induced hypogiycaemla.
The underlying mechanisms which result in hyperinsulinae- mia and impaired prolactin and growth hormone secretion in extreme obesity are not fully explained. Increasing deposi- tion of adipose tissue alters peripheral endocrine function, insulin secretion and glucose disposal and such changes are also related to fat distribution in the upper abdomen, central obesity (Kalkhoff et al., 1983; Campbell & Gerich, 1990). Moreover, central neuroendocrine function is also altered in obesity as shown by reduced prolactin and growth hormone responses to insulin-induced hypoglycaemia (Kopelman, 1988). These changes in neuroendocrine function may largely be reversed by weight reduction but a group of obese women has been characterized in which the prolactin response to insulin-induced hypoglycaemia remains impaired despite substantial weight loss, suggesting the possibility of a primary disorder of hypothalamic function (Kopelman et a[., 1980; Jung et af., 1982).
In laboratory-bred genetically obese rodents a relation- ship has been described between hyperinsulinaemia and altered hypothalamic-pituitary function which presents before the onset of obesity. It has been proposed that abnormal hypothalamic autonomic regulation results in parasympathetic overstimulation of pancreatic cells via vagal innervation leading to elevated plasma insulin levels concomitant with a depression of anterior pituitary function (Jeanrenaud, 1985). We have examined the possibility of a similar relationship existing in human obesity by investigat- ing an association between measures of insulin secretion and resistance and the prolactin and growth hormone response to insulin-induced hypoglycaemia in a group of extremely obese women.
Subjects for the study were recruited from the Obesity Clinic at the Royal London Hospital. Thirty-four severely obese,
BE J. U. Weaver et al . Clinical Endocrinology (1991) 35
unrelated, premenopausal Caucasoid women were studied. They were non-diabetic as documented by a normal oral glucose tolerance test (WHO critera) with a mean body mass index BMI, weight in kg/height in metres2 of 42 (range 34-59) and mean age of 31 years (range 2143). The majority of women included in the study had a family history of obesity but this was not confirmed by photographs.
The control group consisted of 15 normal weight, healthy female volunteers, BMI mean 22 (range 20-25), age 25 years (range 23-30) who were drawn from the medical staff.
The obese subjects ate an isocaloric diet for at least 6 weeks prior to investigations.
The study protocol was approved by the Hospital Ethics Committee; all subjects gave their informed, written consent to participate in the study.
Insulin resistance and insufin secretion Four methods for the assessment of insulin secretion/resistance were used in random order within a period of 10 days.
Fasting insulin (FI) was measured at 0900 hours after an overnight fast. Relative insulin resistance was calculated using ambula- tory fasting insulin and plasma glucose after an over- night fast by the homoeostatic model of assessment (HOMA) method (Matthews et al., 1985). The insulin secretion in response to 75 g oral glucose tolerance test was measured by calculating the area under the curve using trapezoid method. Insulin-induced glucose disposal was measured after a Idhours fast by determining the steady-state of plasma glucose (SSPG) during a simultaneous intravenous infu- sion of dextrose (420 mg/min) and insulin (0.77 mU/kg body weight/min) using somatostatin (500 U/h) to suppress endogenous insulin release. Measurements of plasma insulin and glucose were performed at the beginning of the infusion and then every 30 minutes for a further 150 minutes. Steady states of plasma insulin and glucose were achieved after 90 minutes of the infusion. The arithmetical mean of the measurements at 90, 120 and 150 minutes of the infusion was taken to express a qualitative index of insulin resistance, SSPG (Nagules- paran et al., 1979).
Hypothalamic-pituitary function The growth hormone and prolactin response to insulin-induced hypoglycaemia was measured. The investigation was performed during the follicular phase of the menstrual cycle at 0900 hours after an overnight fast. Obese subjects were given 0.2 units and controls 0.15 units of Actrapid insulin/kg body weight. All
the subjects studied experienced symptomatic neuroglyco- paenia; the time of the maximal hormone release varied between the individuals and the results have been expressed as the area under the response curve, calculated using a trapezoid method.
Insulin Serum immunoreactive insulin was determined by a double antibody RIA, using Guildhay antisera. The inter and intra-assay coefficients of variation (CV), were 10 and 7% respectively, with a minimum detectable limit of 3 mU/l.
Prolactin Prolactin was measured using double antibody precipitating assay, NETRIA Prolactin RIA. The inter and intra-assay CVs were 4.9 and 4.7% respectively.
Growth hormone Growth hormone was measured using NETRIA IRMA assay. The inter and intra-assay CVs were 7 and c 5% respectively.
The type of fat distribution was assessed by a single observer using a flexible tape measure with the subject standing and breathing shallowly. The standardized measurements (in centimetres) were taken at the level half-way between the lower rib margin and the iliac crest for the waist and over the widest hip circumference for the hip and expressed as a ratio (WHO, 1988).
All data were logarithmically transformed prior to analysis. The unpaired Students t-test was used to determine differ- ences in hormone response between obese and controls. The Spearman correlation coefficient test was used to assess the association between the studied variables. Univariate and multivariate regression (backward stepwise regression analy- sis) were used to analyse any association between the studied variables in the obese group (Kleinbaum et al., 1988). Confidence intervals were calculated for the differences between the sample means of prolactin and growth hormone responses calculated as the area under the response curve (Gardner et al., 1986).
The results obtained in this study are summarized in Table I . The obese women showed marked fasting hyperinsulinaemia (mean value SEM, 25 f 2 mU/I; controls mean 7 f 1) with
Clinical Endocrinology (1991) 35
Hypothalamic dysfunction and hyperinsulinaemia in obesity 99
0 - . - .
I l l l l l l l ~ l l l ~ l
fable 1 Results obtained for 34 obese women and 15 normal weight female controls.
Obese Controls (mean f SEM) (mean f SEM)
Fasting insulin (mU/I) 26.0 3.0 7.0 f 0.4 HOMA 6.4f 0.7 1 .O f 0.3 Insulin secretion during OGTT (U/I x minutes) 10.2 fO.8 4.4 f 0.6
Area under prolactin response curve (U/I x minutes) 54.0 f 6.3 Area under growth hormone response curve (U/1 x minutes) 1.9 + 0.2 Steady state plasma glucose (mmol/l) 8.0 & 0.4 -
155.4 f 42.0 65.7 f 2.4
HOMA Homoeostatic model of assessment of relative insulin resistance. No value was obtained for the controls due to inadequate steady plasma insulin concentrations attained (see text).
significant resistance as calculated by the HOMA method (6.4f I .O controls, mean I .Ok 0.3). Their mean insulin AUC during an OGTT was 10.2 f 0.8 U/I min (controls 4.4 f 0.6) and mean SSPG 8.0i0.4 mmol/l. It was not possible to obtain a SSPG value for the control subjects because the degree of insulinaemia achieved (mean 350f 1.0 mU/l) during the insulin infusion was considered insufficient to suppress hepatic glucose production. DeFronzo (1987) has previously described complete suppression of hepatic glu- cose production by plasma insulin concentrations of 50 U/l or greater.
The fat distribution in the obese group as judged by waist to hip ratio, varied between 0-73 and 1.03 confirming that the subjects represented a wide spectrum from femoral-gluteal to central obesity.
Prolactin (using log, of PRL-AUC) and growth hormone (log, of hGH-AUC) responses to insulin-hypoglycaemia were impaired in the obese group as compared to the controls; the difference between the sample means for prolactin response area under the curve was 0.95 (95% confidence interval 0.5 1 1.38); P = 0.000 1. The difference between the sample means for growth hormone response area under the curve was 0.8 (95% confidence interval 0.45- 1-14); P=OWO1. Taking logarithms, the prolactin response showed an inverse correlation with FI, P=0.03, 6=0.142; HOMA, P=0.03, 6=0-134 and SSPG, P=0.0004, f i = 0.354. Backwards procedure for stepwise regression analysis was used because the parameters of insulin secretion were inter-related (Appendix Table A l , A2). This confirmed SSPG to be the most important variable associated with the prolactin response among the remaining indices of insulin secretion/resistance (Fig. 1). A similar, albeit less strong, association was found between log, hGH-AUC and log, SSPG, P=O.OI, r2=0.196 with SSPG being once again the most important variable associated with growth hormone response.
No significant relation was found between regional fat distribution as assessed by W/H ratio and parameters of insulin secretion, insulin resistance or, on this occasion, the prolactin response in the extremely obese women studied.
Obesity may be characterized by an elevated fasting insulin, an exaggerated insulin response to a glucose load, and impaired glucose disposal which, in turn, may be correlated to the degree of obesity. The distribution of fat tissue is important in determining the degree of hyperinsulinaemia, with the highest plasma insulin concentrations generally being associated with upper body fatness central obesity, (Kalkhoff el al., 1983).
In this study we have measured both central and periph- eral endocrine function in a group of extremely obese, non- diabetic women and we report a significant relation between the indices of insulin secretion/insulin resistance and hypo-
100 J. U. Weaver et al. Clinical Endocrinology (1991) 35
thalamic dysfunction, as judged by impaired prolactin and growth hormone responses to symptomatic insulin induced hypoglycaemia. Multiple regression analysis suggests the steady-state plasma glucose level (SSPG), a qualitative measure of insulin resistance and glucose disposal, to be the most important factor associated with an impaired prolactin response to hypoglycaemia. Further, a similar relation, albeit less strong, is seen between the growth hormone response and SSPG. We were unable to repeat the previous observa- tion of a significant relation between an impaired prolactin response and increasing central obesity in this expanded group of obese women (Weaver et al., 1990b). The likely explanation is the difficulty encountered in assessing fat distribution in themost severely obese women included in the present study by using tape measurements (Weaver et al., 1990). The more accurate method for the evaluation of abdominal visceral obesity using CT scanning was not feasible in this study.
It appears that the circulating plasma insulin concen- tration is an important factor in determining peripheral endocrine function. Alterations of sex steroid secretion and binding are a feature of extremely obese premenopausal women and a linear relation exists between decreased sex hormone binding globulin (SHBG) and increasing insulin concentrations, the latter variable being closely associated with increasing central obesity (Weaver et al., 1990a). Moreover, insulin appears to be an important modulator of a low molecular weight insulin-like growth factor binding globulin (IGFBP-1) whose concentrations determine free IGF-I activity; IGFBP-1 concentration in obese women is inversely correlated to fasting insulin (Weaver et ul., 1990a). Thus, the hyperinsulinaemia found in obesity is associated with a complex series of events influencing peripheral endocrine function. A reduced SHBG leads to increased free testosterone and altered oestrogen metabolism which, in turn, is associated with abnormal menstrual function (Kopelman et al., 1980). Decreased IGFBP-1 enhances the biological activity of IGF-I at the tissue level (Ooi & Herington 1988), which may also depress pituitary somato- trophin function by a negative fee...