¹Metabolic Research Centre and Department of Biomedical Science, University of Wollongong, Wollongong, NSW, Australia

²St Vincents Clinic, Darlinghurst, NSW, Australia

³Gemini Genomics plc, Cambridge, UK

⁴Twin Research Unit, St Thomas's Hospital, London, UK

⁵Georgia Prevention Institute, Medical College of Georgia, Augusta, Georgia, USA

⁶Diabetes Centre, St Vincents Hospital, Darlinghurst, NSW, Australia

Correspondence to: A B Jenkins, Department of Biomedical Science, University of Wollongong, Wollongong, NSW, 2522 Australia. E-mail: arthur_jenkins@uow.edu.au

OBJECTIVE: To construct a simple physiological model of leptin kinetics, based on measures of body size and composition, which is suitable for investigating the influence of genetic and other influences on circulating leptin levels in humans.

METHODS: Consideration of the kinetics of the secretion and clearance of leptin led to a predicted linear relationship between ln(leptin), ln(fat mass), and a function of non-fat body compartments. Results obtained from this model were compared with those from two published empirical models based on adjustment for fat mass alone or for body mass index. Overnight fasted leptin levels, body composition data (dual-energy X-ray absorptiometry) and questionnaire responses were obtained from 527 twin pairs (127 monozygotic, 400 dizygotic; 37 male (age 18-68 y, BMI 18-32 kg/m²), 489 female (age 18-71, BMI 17-44) drawn from the St Thomas' UK Adult Twin Registry.

RESULTS: In a partial correlation analysis ln(fat mass) and ln(height) (r=0.80, P<0.0001) and r=-0.22, P<0.0001 respectively) were independent predictors of ln(leptin) in females but ln(lean mass) was not (r=-0.01). A regression model incorporating ln(fat mass), ln(height) and a second order polynomial in age provided an adequate fit of the ln(leptin) data in females (r ²=71%). ln(Leptin) values adjusted for body size and composition using the model were not significantly heritable (P=0.11), were significantly related to gender (r ²=2.3%) and to ln(insulin) (r ²=5.7%), but not to menopausal status (r ²=0.7%), hormone replacement therapy (r ²=0.4%), past or current smoking (r ²=1.1%), or percentage trunk fat (r ²=0.5%). Both empirical models found significant heritability (h²=36-42%), overestimated the effect of gender in the data (r ²=14-16%), and produced significant relationships between adjusted ln(leptin) and percentage trunk fat (r ²=4-12%).

CONCLUSIONS: We conclude that our physiologically based model provides an adequate description of the relationship between leptin and body composition and provides a more reliable framework than current empirical approaches for the investigation of other influences on circulating leptin levels. Heritable variations in the control of leptin secretion are unlikely to contribute significantly to variations in leptin levels at the population level.

International Journal of Obesity (2001) 25, 1625-1632

leptin; adiposity; body composition; model; heritability