Ghrelin is a nutritionally regulated gut peptide that increases with fasting and chronic undernutrition and decreases with food intake. Sex steroid levels change in chronic undernutrition and might signal changes in ghrelin. At the same time, chronic undernutrition is characterized by low IGF-I that might also influence ghrelin, either directly or through changes in the GH axis. Little is known regarding sex steroid regulation of ghrelin and the effects of IGF-I on ghrelin in severe undernutrition. We investigated the effects of sex steroids and IGF-I on ghrelin in 78 female subjects with anorexia nervosa simultaneously randomized to receive estrogen (Ovcon 35, 35 microg ethinyl estradiol, and 0.4 mg norethindrone) as well as recombinant human (rh)IGF-I (30 microg/kg sc twice a day) in a two-by-two factorial model, in which the individual effects of estrogen (E) and rhIGF-I on ghrelin could be determined. Subjects were 24.9 +/- 0.7 (mean +/- sem) yr of age and had low weight (body mass index, 16.7 +/- 0.2 kg/m(2)). At baseline, ghrelin was inversely correlated with body mass index (r = -0.39, P = 0.0005) and IGF-I (r = -0.30, P = 0.01). IGF-I increased significantly more in subjects receiving rhIGF-I alone (Delta 23.0 +/- 5.8 nmol/liter) and rhIGF-I and E (Delta 34.9 +/- 6.3 nmol/liter) compared with subjects receiving E alone (Delta -3.2 +/- 1.9 nmol/liter) or control (C; rhIGF-I placebo and no E) (Delta 0.4 +/- 2.0 nmol/liter) (overall P < 0.0001 by multivariate analysis of variance, P < 0.0001 for rhIGF-I vs. C, P < 0.0001 for rhIGF-I and E vs. C). Ghrelin increased significantly more over 6 months in response to E alone (Delta 150 +/- 86 pg/ml), rhIGF-I alone (Delta 198 +/- 116 pg/ml), and the combination (E and rhIGF-I) (Delta 441 +/- 214 pg/ml) compared with C (Delta -39 +/- 48 pg/ml) (overall P = 0.02 by multivariate analysis of variance, P = 0.01 for E vs. C, P = 0.04 for rhIGF-I vs. C, and P = 0.001 for rhIGF-I and E vs. C). Weight, caloric intake, and morning GH levels did not change significantly between the groups, but the change in ghrelin was inversely related to the change in GH among all subjects (r = -0.27, P = 0.03).Our data demonstrate that, in a model of severe undernutrition, rhIGF-I and E individually increase ghrelin levels. The mechanisms of these effects are unknown and may relate to direct effects on ghrelin or changes in GH. Further studies are needed to determine the mechanisms by which rhIGF-I and E increase ghrelin in human physiology.