Over 90% of women with anorexia nervosa demonstrate osteopenia, and almost 40% demonstrate osteoporosis at one or more skeletal sites. In addition to estrogen deficiency causing an increase in bone resorption, nutritional effects on the GH-IGI-I axis may contribute to the severe bone loss in this population by decreasing bone formation. We tested the hypothesis that recombinant human IGF-I (rhIGF-I) would increase bone density in women with anorexia nervosa and furthermore assessed the effects of combined rhIGF-I and oral contraceptive administration (OCP) in this population. Sixty osteopenic women with Diagnosis and Statistical Manual of Mental Disorders IV Revised confirmed anorexia nervosa [age (25.2 +/- 0.7 yr, range 18-38 yr), body mass index (17.8 +/- 0.3 kg/m(2) ), spinal bone mineral density T score (-2.1 +/- 0.1 SD) were randomized to one of four treatment groups [rhIGF-I (30 microg/kg sc twice daily) and a daily oral contraceptive (Ovcon 35, 35 microg ethinyl estradiol and 0.4 mg norethindrone], rhIGF-I alone (30 microg/kg sc twice daily), oral contraceptive alone, or neither treatment for 9 months. All subjects received calcium 1500 mg/d and a standard multivitamin containing 400 IU of vitamin D. Administration of rhIGF-I was placebo controlled and blinded to subjects. The rhIGF-I was titrated to maintain IGF-I levels within the age-adjusted normal range for each patient and was well tolerated. The effects of rhIGF-I and OCP were analyzed simultaneously among all subjects in a factorial analysis and in an analysis of the four individual treatment groups. Anteroposterior spinal bone density increased significantly in response to rhIGF-I (1.1% +/- 0.5% vs. -0.6% +/- 0.8%, P = 0.05, all rhIGF-I vs. all placebo treated, respectively, by analysis of covariance). In contrast, OCP did not result in increased bone density (0.8% +/- 0.6% vs. -0.4% +/- 0.8%, P = 0.21, all OCP vs. all non-OCP treated, respectively, by analysis of covariance). However, bone density increased to the greatest extent in the combined treatment group (rhIGF-I and OCP), compared with control patients receiving no active therapy (1.8% +/- 0.8% vs. 0.3% +/- 0.6% vs. -0.2% +/- 0.8% vs. -1.0% +/- 1.3%, rhIGF-I and OCP vs. rhIGF-I alone vs. OCP alone vs. no active therapy, P < 0.05 for rhIGF-I and OCP vs. no active therapy). These data demonstrate that osteopenic women with anorexia nervosa treated with rhIGF-I showed more beneficial changes in bone density, compared with patients not treated with rhIGF-I. Antiresorptive therapy with OCP is not sufficient to improve bone density in undernourished patients, but such therapy may augment the effects of rhIGF-I in a combined treatment strategy. Further long-term studies are needed to investigate the effects of rhIGF-I and combined anabolic/antiresorptive strategies on bone in women with anorexia nervosa.