LDM ' vs ' Normal WW' vs ' BW' vs ' DBW' vs ' DWW

Transcription

1 International Journal of Obesity (1999) 23, 909±917 ß 1999 Stockton Press All rights reserved 0307±0565/99 $ Free fatty acids and insulin levelsðrelationship to leptin levels and body composition in various patient groups from South Africa M-T van der Merwe 1, *, VR Panz 1, NJ Crowther 2, GP Schlaphoff 3, IP Gray 2, P Froguel 4, BI Joffe 1 and PN LoÈnnroth 5 1 Carbohydrate and Lipid Metabolism Research Group, University of Witwatersrand Medical School, Johannesburg, South Africa; 2 Chemical Pathology, University of Witwatersrand Medical School, Johannesburg, South Africa; 3 Radiology, University of Witwatersrand Medical School, Johannesburg, South Africa; 4 Department of Human Genetics, Pasteur de Lille, Lille Cedex, France; and 5 Department of Internal Medicine, University of GoÈteborg, Sweden OBJECTIVE: To investigate the relationship between leptin concentrations, various metabolic indices and body composition in six different groups. DESIGN AND MEASUREMENTS: Anthropometric measurements, fasting plasma glucose, serum insulin, C-peptide, FFA and leptin levels were performed. In the obese and diabetic subjects, body composition was analysed with bioimpedance equipment and as a 5 level CT scan. SUBJECTS: Five lipoatrophic diabetes mellitus (LDM) patients, ve normal subjects (N), nine white and nine black obese women (WW, BW), and nine white and nine black diabetic women (DWW, DBW) were investigated after an overnight fast. RESULTS: In both ethnic groups there was a positive correlation between leptin and BMI (black group: r ˆ 0.8; P < , white group: r ˆ 0.7, P < 0.002) and leptin and SC fat mass (black group: r ˆ 0.6; P < 0.005, white group: r ˆ 0.6; P < 0.004). LDM ' vs ' Normal WW' vs ' BW' vs ' DBW' vs ' DWW BMI (kg/m 2 ) (ns) (ns) 40.1 (ns) 39.4 (ns) Fat mass (kg) (ns) 50.2** 49.4 (ns) SC fat (cm 2 ) (ns) 700** 677 (ns) Visceral fat (cm 2 ) * 138*** 173** Glucose (mmol/l) *** (ns) 9.2*** 9.0 (ns) Insulin (pmol/l) *** ** 124*** 182** C-peptide (pmol/l) (ns) *** 761* 1555* FFA (mmol/l) ** *** 1271** 375*** Leptin (ng/ml) *** * 42.5* 28.2* *P < 0.05; **P < 0.01; ***P < Comparisons: LDM vs Normal; WW vs BW, BW vs DBW, DBW vs DWW. CONCLUSIONS: Across the groups, there were positive linear correlations between leptin concentrations, BMI, SC fat mass and FFA levels. Leptin and FFA concentrations are higher and insulin levels lower in both groups of black women compared to the two groups of white women, despite a similar BMI and body fat mass. In the DBW the large increase in visceral fat mass may be indicative of a more complex relationship between compensatory insulin resistance, elevated FFA levels and leptin secretion. Keywords: FFA; insulin; leptin and body composition Introduction The physiological role of leptin in health and disease has not yet been fully elucidated. Leptin levels have *Correspondence: M-T van der Merwe, MD, Department of Endocrinology, University of the Witwatersrand, Medical School, 7 York Rd, Parktown 2193, Johannesburg, South Africa jhp@chiron.wits.ac.za Revised 2 December 1998; accepted 17 March 1999 Abbreviations: LDM ˆ Lipoatrophic diabetes; N ˆ Normal control subjects; WW ˆ White obese women; BW ˆ Black obese women; DWW ˆ Diabetic white obese women; DBW ˆ Diabetic black obese women; FFA ˆ Free fatty acids; SC ˆ Subcutaneous been consistently associated with body adiposity and body mass index (BMI). 1,2 Leptin is also known to regulate acute and chronic energy balance as well as thermogenesis. 3±5 In order to understand the full range of leptin's biological activities, clari cation is still needed on the impact that leptin may have on insulin regulated responses, 6 and vice versa. 7 Ethnic differences in leptin levels and leptin's role in the regulation of resting energy expenditure have previously been reported by Nicklas et al. 8 We have previously studied black (BW) and white urban obese women (WW) from South Africa. Those urban obese BW who are as westernised in diet and

2 910 lifestyle as obese WW, more commonly present with type 2 diabetes (7% vs 3.6%), 9, 10 but with a much lower incidence 11 and mortality due to ischaemic heart disease (8= vs 55= ). 12 The BW were relatively insulinopenic, with higher FFA and glucose concentrations, but with a more favourable 13 ± 15 lipid pro le. In addition, BW released more glycerol and FFA from their subcutaneous (SC) fat tissue as a result of their higher lipolytic rate (submitted data). In view of the high incidence of obesity amongst urban BW (59.4%), 15 as well as previous observations of differences in the intermediary and fat metabolism between these two ethnic groups, we analysed leptin levels to evaluate the relative contribution of body fat mass and fasting insulin concentrations to leptin levels in our black and white women. We included a small group of patients with lipoatrophic diabetes (LDM). Although this is a rare disease, these patients are an interesting phenotypic group to study because they are virtually devoid of subcutaneous fat. 17 The contribution that factors such as a high FFA or low visceral fat mass may have in these disease pro les, has not been studied in South African diabetic women before. In addition, we outlined the relationship between leptin levels, various fasting metabolic indices and body composition pro les as documented in both obese and diabetic women. Research design and methods Nine black and nine white obese women, and nine black diabetic and white diabetic women were matched for age, BMI and WHR. The other clinical data are shown in Table 1. The obese women were also matched for diet, duration of obesity (8 ± 10 y), level of physical and habitual activity, number of offspring (no more than three) and socio-economic background (verbal questionnaire). None of the women were postmenopausal or on oral contraceptives. The diabetic women had a duration of diabetes between 3 and 5 y, and none of the diabetic patients were treated with insulin (oral sulphonylureas only). The exclusion criteria included diseases of all the major organs, excessive smoking and alcohol consumption. Five normal female subjects were recruited as controls for the ve LDM patients. The LDM patients were a group made up of mixed ethnicity (two black, one coloured and two white patients). Their treatment varied according to the severity of their associated complications (two on insulin, two on oral anti-diabetic agents and one on lipid lowering agents). As a group they were unique, and studied purely within this context. Due to logistical problems they could not have all their anthropometric data, CT scans or body composition analyses performed. All the patients gave informed consent, and the study was approved by the Committee for Research on Human Subjects of the University of the Witwatersrand. After an overnight fast of 10 h, venous blood was taken for analysis of glucose, insulin, C-peptide, FFA and leptin levels. Biochemical analysis Glucose was measured with an enzymatic colorimetric method (GOD-PAP, Boehringer Mannheim, Germany). Radioimmunoassay kits were used to measure C-peptide (Medgenix Diagnostics, Belgium; normal range 110 ± 1270 pmol=l), and leptin levels (Linco Research Inc., St. Louis, Missouri, USA; normal range: 3.7 ± 11.1 ng=ml). Insulin levels were determined with an enzyme ampli ed sensitivity immunoassay (EASIA ± Medgenix Diagnostics, Belgium; normal range 13.8 ± pmol=l). FFA concentrations were measured with an enzymatic method (FFA Half-micro test Boehringer Mannheim, Germany; normal range 300 ± 800 mmol=l). The specimens were stored at 7 70 C until analysed in a single assay to avoid inter-assay variation. Body composition analyses This was the rst test to be performed in the morning with bio-electrical impedance analysis. 13 The Bodystat machine (Bodytrach Pty. Ltd., South Africa) was used to assess the body composition for fat and muscle tissue (kg; %). The bio-impedance equations are based on the South African population as a whole. Table 1 Clinical and anthropometric measurements of the six groups of women studied LDM N WW BW DBW DWW Age (y) Waist (cm) (P < 0.001) b (P < 0.01) c WHR Body weight (kg) (P < 0.02) b BMI (kg=m 2 ) Parasag Diameter (cm) (P < 0.05) a (P < 0.01) b (P < 0.05) c Height (cm) Mean values sem; n ˆ 5 in LDM and normal groups; n ˆ 9 in other groups. P ˆ unpaired Student's t-test. a WW vs BW; b DBW vs BW; c DBW vs DWW

3 (Personal communication: Prof. A.E. Bunn, Medical Research Council, Health Technology Division). CT scan measurements In the week prior to their investigation a ve-level CT scan was done (Phillips SR 7000; The Netherlands) on the women in the fasted state, as previously described by Van der Merwe et al. 13 The scan parameters were: (1) 10 mm slice thickness; (2) 120 kv; (3) 250 ma; (4) 2 s and (5) 340 ± 480 mm FOV, depending on the size of the subject. They were in the supine position, with the photographic images taken in resting expiration. Measurements of visceral and SC fat (truncal and gluteofemoral) were done. The ve levels were derived from two scanograms and included a CT slice at the level of the diaphragm, umbilicus, L4 ± 5 lumbar disc, the widest diameter of the pelvis and mid-thigh (the total distance from the iliac crest to acetabulae and from acetabulae to the knee joint, divided by two). The rst scanogram included the diaphragm, umbilicus (marked with a metallic coin), iliac crest and symphysis pubis, with a maximum length of 500 mm from the umbilicus. The second scanogram extended from the symphysis pubis to the knees. The widest parasagittal diameter was measured at the level of L4 ± 5. The fat areas were calculated with a Region of Interest seeding programme on a Philips Gyroview workstation (fat values were chosen between 7 30 HU and HU). The areas of SC and visceral fat were calculated separately and the anatomical boundaries were as described by Kvist et al. 18,19,20,21 Visceral fat was measured at the top 3 levels. The parasagittal diameter was measured at the level of L4 ± 5. Statistical analysis Statistical signi cance of parametric data was assessed by Student's paired and unpaired t-test. Linear regression analysis was done using the least squares method and linear correlations were tested with the Pearson's correlation coef cient (r). A value of P < 0.05 was considered signi cant. Data are expressed as mean sem. Results Clinical and anthropometric measurements There were no differences in age or BMI between the WW and BW, or the DWW and DBW (Table 1). However, the diabetic obese women were all substantially heavier than the non-diabetic women. BMI and body weight were much lower in the LDM patients and normal subjects. The WHR of the WW and BW showed no signi cant difference ( in both groups), and although it tended to be higher in the DBW and DWW ( and respectively), it was not signi cant. Measurement of the parasaggital diameter at the level of L4 ± 5 of the CT scan was lower in the BW compared with WW ( cm vs cm; P < 0.05), as well as in the DBW compared with DWW ( cm vs cm; P < 0.05). The parasaggital diameter was higher in the diabetic women compared to the non-diabetic women. Waist circumference followed a similar trend to the parasaggital diameter within all four groups. Body composition analyses Body composition analyses showed that the diabetic women had substantially more total body fat than the obese women, both in kilograms and expressed as a percentage, with a corresponding decrease in the percentage lean mass in the diabetic group (Table 2). There was no difference in the body composition between the black and white women (BW vs WW: kg vs kg fat mass and kg vs kg lean mass) (DBW vs DWW: vs kg fat mass and kg vs kg lean mass). SC fat mass was reported as the mean recorded value of the 5 CT scan cuts, ranging from the diaphragm down to the mid-femur. No ethnic difference was found between the black and white women (BW vs WW: cm 2 vs cm 2 ) and (DBW vs DWW: cm 2 vs cm 2 ), however diabetic women had more SC fat compared to the non-diabetic women. The mean amount of visceral fat was recorded from the top 3 CT scan cuts and indicated a larger amount of visceral fat in the obese WW compared to the obese BW ( cm 2 vs cm 2 ; P < 0.01). There was a marked increase in the amount of visceral fat mass in the DBW compared to the BW ( cm 2 vs cm 2 ; P < ), but the DBW still had less visceral fat compared to the DWW ( cm 2 vs cm 2 ; P < 0.01). Table 2 Body composition analyses WW BW DBW DWW Fat mass (kg) P ˆ NS a P < b P ˆ NS c Fat mass (%) P ˆ NS a P < b P ˆ NS c Lean mass (kg) P ˆ NS a P ˆ NS b P ˆ NS c Lean mass (%) P ˆ NS a P < 0.01 b P ˆ NS c SC fat (cm 2 ) P ˆ NS a P < 0.01 b P ˆ NS c Visceral fat (cm 2 ) P < 0.01 a P < b P < 0.01 c Mean values sem; P ˆ unpaired Student's t-test; SC, subcutaneous. n ˆ 9 in each group. a WW vs BW; b BW vs DBW; c DBW vs DWW 911

4 912 Biochemical analyses Mean fasting glucose concentrations showed a U- shaped distribution, with elevated levels recorded in the LDM patients ( mmol=l) and the two groups of diabetic women (DBW mmol=l and DWW mmol=l; P ˆ ns). Normal subjects and the two groups of non-diabetic obese women had fasting glucose levels in the normal range (Figure 1). Mean fasting insulin levels were signi cantly elevated in the LDM patients (LDM vs N: vs pmol=l; P < ), and C-peptide levels were not signi cantly higher when compared with normal subjects (Table 3). In the BW, insulin (P < 0.01) and C-peptide levels (P < ) were signi cantly lower than those of the WW. The DBW had signi cantly higher fasting insulin (P < ) and C-peptide levels (P < 0.05) compared to the non-diabetic BW, but in comparison to the DWW, both the insulin (P < 0.01) and C-peptide levels (P < 0.05) were signi cantly lower. No signi cant increases were observed in the insulin and C- peptide levels between the WW and the DWW. Mean fasting FFA levels increased from normal, to obese, to diabetic subjects. Signi cantly lower levels were recorded in the LDM patients compared to the normal subjects ( vs mmol=l; P < ), and levels were signi cantly higher in both black groups of women compared to the two groups of white women (BW vs WW: mmol=l vs mmol=l; P < and DBW vs DWW: mmol=l vs mmol=l; P < ) (Figure 2). Mean fasting leptin concentrations showed a progressive increase across ve of the groups, with the lowest level recorded in the LDM patients ( ng=ml) and the highest level recorded in DBW ( ng=ml) (Figure 3). Levels were signi cantly higher in both groups of black women compared with the white women despite no signi cant differences in weight, BMI or total body fat mass within the obese or diabetic groups. There was a strong linear correlation between leptin concentrations and BMI when combining the six groups (r ˆ 0.74; P < 0.001) (Figure 4a). When combining the two groups of black women and the two groups of white women, the following correlations were obtained: BW and DBW: r ˆ 0.8; P < , WW and DWW: r ˆ 0.7; P < (Figure 4b,c.) Similarly, leptin Figure 2 Fasting FFA values for the six groups. P ˆ unpaired Student's t-test. Data are means sem; LDM vs N: P < ; WW vs BW: P < ; BW vs DBW: P < 0.01; DBW vs DWW: P < Figure 1 Fasting glucose values for the six groups. P ˆ unpaired Student's t-test; Data are means sem; LDM vs N: P < ; WW vs BW: NS; BW vs DBW: P < ; DBW vs DWW: NS. Figure 3 Fasting leptin values for the six groups. P ˆ unpaired Student's t-test Data are means sem; LDM vs N: P < ; WW vs BW: P < 0.05; BW vs DBW: P < 0.05; DBW vs DWW: P < Table 3 Fasting insulin and C-peptide values of six groups of women studied LDM Normal subjects WW BW DBW DWW Insulin (pmol/l) P < a P < 0.01 b P < P < 0.01 d C-peptide (pmol/l) P ˆ NS a P < b P < 0.04 c P < 0.05 d Mean values sem; n ˆ 5 in LDM and normal groups; n ˆ 9 in other groups. P ˆ unpaired Student's t-test. a LDM vs N; b WW vs BW; c DBW vs BW; d DBW vs DWW

5 correlated positively to FFA levels r ˆ 0.61; P < ) (Figure 5). Combining the obese and diabetic women in the two ethnic groups did not alter statistical signi cance (black women: r ˆ 0.40; white women: r ˆ 0.05). Mean SC fat mass was measured by CT scan, and was available in four of the groups ± the two groups of obese women and two groups of diabetic women. For the study group as a whole, there was a signi cant correlation between leptin concentrations and SC fat mass (r ˆ 0.76; P < ) (Figure 6a). Combining the women in the two ethnic groups showed positive correlations for the black women (r ˆ 0.60; P < 0.005) and the white women (r ˆ 0.60; P < 0.004) (Figure 6b,c). Correlations done within the individual groups showed that in the LDM patients leptin levels correlated negatively with weight (r ˆ 7 0.9; P < 0.02). In the WW, leptin levels correlated positively with weight (r ˆ 0.8; P < 0.009), both kilograms (r ˆ 0.7; P < 0.03) and percentage of fat mass (r ˆ 0.7; P < 0.05), as well as fasting insulin levels (r ˆ 0.8; P < 0.008). A positive correlation was found between leptin levels and weight (r ˆ 0.7; P < 0.04), BMI (r ˆ 0.8; P < 0.008) and mean SC fat mass (r ˆ 0.8; P < 0.008) in DBW. In the DWW, leptin levels correlated positively with weight (r ˆ 0.6; P < 0.01), BMI (r ˆ 0.7; P < 0.01), both kilograms and percentage fat mass (r ˆ 0.7; P < 0.01 and r ˆ 0.5; P < 0.01) and mean SC fat mass (r ˆ 0.7; P < 0.01). 913 Discussion Human obesity is a complex consequence of many factors. We initiated this study to evaluate the possibility that plasma leptin concentrations may be related to the differences found in intermediary metabolism and body composition in various ethnic groups. These included a relative insulinopenia in the black population with more glycerol and FFA released from their adipose tissue. Although the pathophysiology of leptin in humans is not as simple as it seemed to be in rodent models of obesity, its role as an adipocyte-derived signalling molecule seems well established. Figure 4 Linear regression: leptin to BMI, (A) All six groups. r ˆ tailed P < ; (B) BW and DBW. r ˆ tailed P < ; (C) WW and DWW. r ˆ tailed P < Figure 5 Linear regression: leptin to FFA. All six groups included. r ˆ 0.61; 2-tailed P <

6 914 Figure 6 Linear regression: leptin to subcutaneous fat mass. (A) Including WW, BW, DBW, DWW, r ˆ 0.76, 2-tailed P < ; (B) BW and DBW. r ˆ 0.6, 2-tailed P < 0.005; (C) WW and DWW. r ˆ 0.6, 2-tailed P < Adipose tissue leptin mrna concentrations and plasma leptin levels have been found to be closely correlated with the size of the adipose tissue depot. 1,2,22 This was also the case in our patients, in whom BMI ranged from 21.5 to The more speci c in uence of total body fat mass was however highlighted by the LDM patients, who presented with lower leptin levels compared to the normal control subjects despite a similar BMI. Likewise the black ethnic group of women presented with higher leptin levels compared to the white ethnic group of women, despite comparable BMI's. It remains remarkable that the LDM patients are still able to produce leptin despite their virtual total lipo-atrophic state (Dr S O'Rahilly ± personal communication) and perhaps re ects a degree of insulin resistance. 23 Thus, although the BMI may show a correlation with leptin levels when performed across a broad spectrum of body weights, other factors such as peripheral fat mass and distribution, as well as insulin resistance, may be far more important in determining the leptin concentration. 24 In the four groups of obese women, both diabetic and non-diabetic, leptin correlated closely with the mean amount of subcutaneous tissue as measured by CT scanning. This would support the view of LoÈnnqvist et al that leptin mrna, leptin secretion, and circulating leptin levels are all more closely related to the stored amount of lipids in the fat cells of adipose tissue than they are to an arbitrary division into obese vs non-obese. 25 The six groups of patients had FFA levels ranging over a wide spectrum. We have previously documented higher FFA levels and a more brisk lipolysis in the black obese women, when compared to the white obese women. 13 This was thought to be due to the relative insulinopenia documented in the black women, resulting in a diminished anti-lipolytic effect on adipose tissue. This relationship between increased adipocyte lipolysis and excess release of fatty acids has been outlined before. 26 The lack of lipolytic control appears to be present in the obese black diabetic women as well, because we have documented particularly high levels of FFA in this group, despite moderate diabetic control. The very high fasting plasma FFA in the DBW coincided with both an increase in insulin levels and visceral fat mass. The high FFA levels could be the result of an increasing insulin-resistance pattern with portal hyperinsulinemia and subsequent reduction of the antilipolytic effect of insulin. 27 In addition, it is known that high physiological concentrations of FFA reduce hepatic insulin binding and degradation and therefore insulin clearance. 27,28 This effect will be ampli ed by obesity, particularly when localised to the abdominal=visceral regions. 29,30 Furthermore increased FFA availability can induce skeletal muscle insulin resistance by increasing the ux of fructuse-6-phosphate into the hexosamine pathway, 31 as well as alter both insulin secretion32 ± 34 and hepatic degradation. 35 There is also evidence that an increase in leptin levels, characteristic of obese individuals, can lead to insulin resistance, as well as compromise the ability of the b-cell to compensate for the insulin resistance. 36 ± 40 It therefore seems feasible to conclude that the higher leptin and FFA

7 concentrations in our black population are linked, and may act in concert to induce an acquired insulin resistance. This argument is strengthened by the fact that we have found that black obese women are more insulin resistant during the hyperinsulinemic euglycemic clamp procedures (submitted data), and that there is no noticeable difference in the correlation between leptin concentrations and BMI, or leptin levels and subcutaneous fat mass between the two ethnic groups. The relationship between leptin and insulin appears to be very complex. Both hormones participate in the neuro-endocrine obesity axis and the satiety control pathway, but both hormones also modulate each other's functional activities, as well as insulin resistance. 41 ± 44 Berti et al provided data with evidence for a positive crosstalk between the signalling chain of the insulin receptor and the leptin receptor. 45,46 This effect of leptin will occur independently of insulin-receptor-substrate-1 activation, and most likely through activation of phosphatidylinositol-3-kinase. 46 The resulting dual regulating action of leptin on glucose transport and glycogen synthesis, as well as on insulin signalling appears to be concentration-dependent: thus the higher leptin levels in the black population may be compensating for the diminished insulin signalling in this relatively insulinopenic population. Alternatively, the higher leptin levels in the black population may represent a form of leptin resistance due to defective leptin signalling, resulting in hyperglycemia and a possible over-expression of hypothalamic NPY. 47 This will need future investigation. In contrast, leptin has also been described to suppress second phase insulin secretion by reducing the activity of calcium-dependent protein kinase C isoforms. 48 Although the relative insulinopenia of the black population is thought to be a consequence of reduction in the b-cell mass due to transient episodes of protein-energy malnutrition during infancy or in utero, 49 ± 51 the possibility of other compounding or aggravating factors cannot be excluded. The difference in the amount of visceral fat between the black and white women is unlikely to be responsible for the difference in the leptin levels between these groups, as the contribution of visceral fat to the overall body fat mass in women still remains small ± approximately 10%. 52 Effectively the difference in the visceral fat mass between the two groups would then account for less than 3% of the overall contribution to total body fat mass. However, this does not detract from the fact that visceral fat accumulation may be one of the main determinants of obesity-related diseases. 53 In addition, van Harmelen et al has recently shown the SC fat depot to be the major source of leptin in women, owing to the combination of a mass effect and higher secretion rate. 54 Differences in the amount of visceral fat between the black and white groups of women have been shown before. 13 This study con rms the lower amount of visceral fat in the black ethnic group of women when compared to the white women, but both ethnic groups virtually double their amount of visceral fat once they have diabetes. Visceral obesity is particularly associated with hyperinsulinemia, increased portal FFA concentrations and hepatic gluconeogenesis. 26 It is unlikely that the characteristic dyslipidaemia associated with visceral adiposity in the Caucasian population will be a part of the metabolic pro le of the black diabetic population, 55 as this may well be one of the protective factors against the black diabetic patients developing a higher incidence of coronary heart disease. 11,12,56 In conclusion, leptin concentrations are a good marker of BMI and subcutaneous body fat in both black and white obese women, with or without diabetes. However, despite a similar body fat mass, leptin levels in the black ethnic group of women are set at a higher level compared to the white women, implicating additional contributing factors such as insulin and FFA levels, as well as insulin resistance. The lower visceral fat mass in black obese and black diabetic women when compared to white women may be related to their relative insulinopenia, but both ethnic groups show a marked increase in visceral fat mass during the diabetic disease state. Acknowledgements We would like to thank Jenny Pieters for typing the manuscript and Ann Fennel, Karine Clement and Dr Najiba Lahlau for outstanding technical assistance. This work was partly sponsored by the South African Medical Research Council, the South African Institute for Medical Research and the Pasteur de Lille Institute. We are grateful to Dr. Laura Blacking for her clinical assistance, and to Darren Boyd for performing part of the statistical analyses. References 1 Maffei M, Halaas J, Ravussin E. Leptin levels in human and rodent: measurement of plasma leptin and ob mrna in obese and weight reduced subjects. Nat Med 1995; 1: 1155 ± Considine R, Sinha MK, Herman ML. Serum immunoreactive-leptin concentrations in normal-weight and obese humans. N Engl J Med 1996; 334: 292 ± Flier JS. Leptin expression and action: New experimental paradigms. Proc Natl Acad Sci USA 1997; 94: 4242 ± Camp eld LA, Smith FJ, Guisez Y, Devos R, Burn P. Recombinant mouse ob protein: evidence for peripheral signal linking adiposity and central neural networks. Science 1995; 269: 546 ± Kolaczynski JW, Ohannesian JP, Considine RV, Marco CC, Caro JF. Response of leptin to short-term and prolonged overfeeding in humans. J Clin Endocrinol Metab 1996; 81: 4162 ± Cohen B, Novick D, Rubinstein M. Modulation of insulin activities by leptin. Science 1996; 274: 1185 ± Kolaczynski JW, Nyce MR, Considine RV, Boden G, Ndan JJ, Henry R, Mudaliar SR, Olefsky J, Caro JF. Acute and chronic effect of insulin on leptin production in humans. Diabetes 1996; 45: 699 ±

8 916 8 Nicklas BU, Toth MJ, Goldberg AP, Poehlman ET. Racial differences in plasma leptin concentrations in obese postmenopausal women. J Clin Endocrinol Metab 1997; 82: 315 ± Joffe BI, Seftel HC. Diabetes mellitus in the black communities of Southern Africa. J Int Med 1994; 235: 137 ± Omar MAK, Seedat MA, Motlala AA, Dyer RB, Becker P. The prevalence of diabetes mellitus and impaired glucose tolerance in a group of urban South African blacks. S Afr Med J 1993; 83: 641 ± Seftel HC. The rarity of coronary heart disease in South African blacks. S Afr Med J 1978; 54: 99 ± Isles CG, Milne FJ. Low mortality from ischaemic heart disease among urban blacks in South Africa. J Clin Hypertension 1987; 3: 749 ± Van der Merwe M-T, Wing JR, Celgow LH, Gray IP, LoÈnn L, Joffe BI, LoÈnnroth PN. Metabolic indices in relation to body composition, changes during weight loss on Dexfen uramine in obese women from two South African ethnic groups. Int J Obesity 1996; 20: 768 ± Steyn K, Jooste PL, Bourne L, et al. Risk factors for coronary heart disease in the black population of the Cape Peninsula. The Brisk Study. S Afr Med J 1995; 85: 90 ± Walker ARP, Bhanjee D, Walker BF, Martin AP. Serum high density lipoprotein cholesterol, glucose tolerance and other variables in obese Black adolescent girls. S Afr Med J 1979; 56: 221 ± Walker ARP. Epidemiology and health implications of obesity in Southern Africa. MRC technical report on Chronic diseases of lifestyle in South Africa 1995: 73 ± Braunwald E, Isselbacher KJ, Petersdorf RG, Wilson JD, Martin JB, Fauci AS (eds). Harrison's Principles of Internal Medicine. 4th edn. McGraw-Hill: New York, 1987; pp 1677 ± Kvist H, SjoÈstrom L, TyleÂn U. Adipose tissue volume determinations in women by computed tomography. Technical considerations. Int J Obesity 1986; 10: 53 ± SjoÈstrom L, Kvist H, Cederblad A, TyleÂn U. Determinations of total adipose tissue volume and body fat in women by computed tomography, 40K and tritium. Am J Physiol 1986; 250: E736 ± E Kvist H, Chowdbury B, Grangard U, TyleÂn U, SjoÈstrom L. Total and visceral adipose tissue volumes derived from measurements with computed tomography in adult men and women: predictive equations. Am J Clin Nutr 1988; 48: 1351 ± Chowdbury B, Kvist H, Andersson B, BjoÈrntorp P, SjoÈstrom L. CT determined changes in adipose tissue distribution during a small weight reduction in obese males. Int J Obesity 1993; 17: 685 ± Considine RV, Considine EL, Williams CJ. Evidence against either a premature stop codon or the absence of obese gene mrna in human obesity. J Clin Invest 1995; 95: 2986 ± Pardini VC, VictoÂria IMN, Rocha SM, Andrade DG, Rocha AM, Pieroni FB, Milagres G, Purisch S, Velho G. Leptin levels, b-cell function, and insulin sensitivity in families with congenital and acquired generalized lipoatrophic diabetes. J Clin Endocrinol Metab 1998; 83: 503 ± LoÈnnqvist F, Wennlund A, Arner P. Relationship between circulating leptin and peripheral fat distribution in obese subjects. Int J Obes 1997; 21: 255 ± LoÈnnqvist F, Nordfors L, Jansson M, Thorne A, Schalling M, Arner P. Leptin secretion from adipose tissue in women ± Relationship to plasma levels and gene expression. J Clin Invest 1997; 99: 2398 ± Bjorntorp P. `Portal' adipose tissue as a generator of risk factors for cardiovascular disease and diabetes. Atherosclerosis 1990; 10: 493 ± Svedberg J, BjoÈrntorp P, Smith U, LoÈnnroth P. Free fatty acid inhibition of insulin binding, degradation and action in isolated hepatocytes. Diabetes 1990; 39: 570 ± Svedberg J, StroÈmblad G, Wirth A, Smith U, BjoÈrntorp P. Fatty acids in the portal vein of the rat regulate hepatic insulin clearance. J Clin Invest 1991; 88: 2054 ± Peiris AN, Mueller RA, Smith GA, Struve MF, Kissebah AH. Splanchnic insulin metabolism in obesity. In uence of body fat distribution. J Clin Invest 1986; 78: 1648 ± Karakash CF, Jeannet A, Jeanrenaud B. An anomaly of insulin removal in perfused livers of obese-hyperglycemic (ob=ob) mice. J Clin Invest 1976; 57: 1117 ± Hawkins M, Barzilai N, Liu R, Hu M, Chen W, Rosetti L. Role of the glucosamine pathway in fat-induced insulin resistance. J Clin Invest 1997; 99: 2173 ± Lee Y, Hiroshi H, Ohneda M, Johnson JH, McGarry JD, Unger RH. b-cell lipotoxicity in the pathogenesis of noninsulin-dependant diabetes mellitus of obese rats: impairment of adipocyte b-cell relationships. Proc Natl Acad Sci 1994; 91: ± Zou YP, Grill VE. Long term exposure to fatty acids and ketones inhibits b-cell function in human pancreatic islets of Langerhans. J Clin Endocrinol Metab 1995; 80: 1584 ± Padisso G, Gambardella A, Amato L, Tortoriello R, D'Amore A, Vancohio M, D'Onofrio F. Opposite effects of short- and long-term fatty acid infusions on insulin secretion on healthy subjects. Diabetologia 1995; 30: 1295 ± Svedberg J, StroÈmblad G, Wirth A, Smith U, BjoÈrntorp P. Fatty acids in the portal vein of the rat regulate hepatic insulin clearance. J Clin Invest 1991; 88: 2054 ± Segal KR, Landt M, Klein S. Relationship between insulin sensitivity and plasma leptin concentrations in lean and obese men. Diabetes 1996; 45: 988 ± Cohen B, Novick D, Rubinstein M. Modulation of insulin activities by leptin. Science 1996; 274: 1185 ± Widjaja A, Stratton IM, Horn R, Holman RR, Turner R, Brabant G. UKPDS 20: Plasma leptin, obesity and plasma insulin in type 2 diabetic subjects. J Clin Endocrinol Metab 1997; 82: 654 ± Emilsson V, Liu Y-L, Cawthorne MA, Morton NM, Davenport M. Expression of the functional leptin receptor m-rna in pancreatic islets and direct inhibitory action on insulin secretion. Diabetes 1997; 46: 313 ± Kieffer TJ, Heller RS, Leech CA, Holtz GG, Habener JF. Leptin suppression of insulin secretion by the activation of ATP-sensitive K channels in pancreatic b-cells. Diabetes 1997; 46: 1087 ± Schwartz MW, Seeley RJ. Neuro-endocrine responses to starvation and weight loss. N Engl J Med 1997; 336: 1802 ± Cohen B, Novick D, Rubenstein M. Modulation of insulin activities by leptin. Science 1996; 274: 1185 ± MuÈller G, Ertl J, Gerl M, Preibisch G. Leptin impairs metabolic actions of insulin in isolated rat adipocytes. J Biol Chem 1997; 272: ± Kolaczynski JW, Nyce MR, Considine RV, Boden G, Nolan JJ, Henry R, Mualiar SR, Olefsky J, Caro JF. Acute and chronic effect of insulin of leptin production in humans. Studies in vivo and in vitro. Diabetes 1996; 45: 699 ± Kellerer M, Koch M, Metzinger E, Mushack J, Capp E, HaÈring HU. Leptin activates PI-3 kinase in C 2 C 12 myotubes via janus kinase-2 (JAK-2) and insulin receptor substrate-2 (IRS-2) dependent pathways. Diabetologia 1997; 40: 1358 ± Berti L, Kellerer M, Capp E, HaÈring HU. Leptin stimulates glucose transport and glycogen synthesis in C 2 C 12 myotues: evidence for a PI3-kinase mediated effect. Diabetologia 1997; 40: 606 ± Schwartz MW, Baskin DG, Bukowski TR, Kuijper JL, Foster D, Lasser G, Prunkard DE, Porte D, Jr., Woods SC, Seeley RJ, Weigle DS. Speci city of leptin action on elevated blood glucose levels and hypothalamic neuropeptide Y gene expression in ob=ob mice. Diabetes 1996; 45: 531 ± Ookuma M, Kazuyoshi O, York D. Effects of leptin on insulin secretion from isolated rat pancreatic islets. Diabetes 1998; 47: 219 ± 223.

9 49 Swenne I, Borg LAH, Grace CJ, Schnell-Lands. Persistent reduction of pancreatic b-cell mass after a limited period of protein-energy malnutrition in the young rat. Diabetologia 1992; 35: 595 ± Hales CN, Barker DJP. Type 2 (non-insulin-dependent) diabetes mellitus: the thrifty phenotype hypothesis. Diabetologia. 1992; 35: 595 ± Hales CN, Barker DJP, Clarke PMS, Cox LJ, Fall C, Osmons C. Fetal and infant growth and impaired glucose tolerance at age 64. Br Med J 1991; 303: 1019 ± SjoÈstrom L, Kvist H, Cederblad A, TyleÂn U. Determination of total adipose tissue volume and body fat in women by computed tomography, 40 K and tritium. Am J Physiol 1986; 250: E736 ± E Seidell JC, Bouchard C. Visceral fat in relation to health: is it a major culprit or simply an innocent bystander. Int J Obes 1997; 21: 626 ± Van Harmelen V, Reynisdothr S, Eriksson P, ThoÈrne A, Hoffstedt J, LoÈnnqvist F, Arner P. Leptin secretion tissue in women. Diabetes 1998; 47: 913 ± BjoÈrntorp P. Abdominal fat distribution and disease: An overview of epidemiological data. Ann Med 1992; 24: 15 ± Seftel HC, Raal FJ, Joffe BI. Dyslipidaemia in South Africa. MRC technical report on chronic diseases of lifestyle in South Africa. 1995: 61 ±