St. Georges Hospital Medical School London SW17 0RE, United Kingdom Gerard S. Conway Department of Medicine, University College London SW1 8AA, United Kingdom
Recently, Eckert et al. (1) presented longitudinal data confirming previous cross-sectional reports (2, 3) that leptin concentrations in female anorectics were significantly lower than normal weight controls, increased significantly on refeeding, and correlated linearly with body mass index. They also suggested an uncoupling of leptin and adiposity at lowest body weights. However, they acknowledged that "a major limitation of this study is the lack of data on percent body fat," as body mass index does not necessarily correlate with percent body fat. Similarly the cross-sectional reports (2, 3) estimated percent body fat using bioimpedance. The latter assumes uniform hydration, which may not be true in refeeding anorectics (4). To our knowledge, no previous study has longitudinally correlated leptin against total body fat during weight restoration in anorexia nervosa, except by such crude means of body fat estimation.
Like Eckert et al., we carried out a longitudinal study of the relationship between leptin and total body fat. However, we employed dual-energy X-ray absorpitometry (DXA) to measure body fat directly (5), thus avoiding the pitfalls of body mass index or bioimpedance. We monitored plasma leptin concentrations, body mass index, and percent body fat longitudinally in seven female patients with anorexia nervosa (mean age 20.4 yr, SD 4.76, range 1627 yr) during weight restoration. Patients met DSM-IV criteria for anorexia nervosa. Mean body mass index (BMI) before weight restoration was 15.8 kg/m2 (SD 1.51; range 13.817.2 kg/m2) and after weight restoration was 19.8 kg/m2 (SD 0.71; range 19.220.9 kg/m2). Plasma was obtained from patients between 0900 and 1000, after a standardized meal containing 250 kcal, in the context of standardized daily intake of between 2100 and 2600 kcal. Serum leptin was measured by RIA using recombinant human leptin as standard with intra- and interassay coefficients of variation of no more than 5%.
The results are shown in Table
1. Spearman correlation coefficients, with their associated two-tailed
significance levels, were calculated between percentage changes in
leptin and other variables. The level of statistical significance was
set at 5%. There was a highly significant, positive correlation
between percentage changes in leptin and total body fat (r = 0.89;
P = 0.007) and also between percentage changes in
leptin and BMI (r = 0.86; P = 0.014).
|
Footnotes
Received August 4, 1998. Address correspondence to: John F. Morgan, Department of General Psychiatry, St. Georges Hospital Medical School, London SW17 0RE, United Kingdom (
References