To study the pathophysiology of hypercortisolism in patients with anorexia nervosa, we examined plasma ACTH and cortisol responses to ovine corticotropin-releasing hormone before and after correction of weight loss. We also studied patients with bulimia whose weight was normal, since this disorder has been suspected to be a variant of anorexia nervosa. Before their weight loss was corrected, the anorexic patients had marked hypercortisolism but normal basal plasma ACTH. The hypercortisolism was associated with a marked reduction in the plasma ACTH response to corticotropin-releasing hormone. When these patients were studied three to four weeks after their body weight had been restored to normal, the hypercortisolism had resolved but the abnormal response to corticotropin-releasing hormone remained unchanged. On the other hand, at least six months after correction of weight loss their responses were normal. The bulimic patients whose weight was normal also had a normal response to corticotropin-releasing hormone. We conclude that in underweight anorexics, the pituitary responds appropriately to corticotropin-releasing hormone, being restrained in its response by the elevated levels of cortisol. This suggests that hypercortisolism in anorexics reflects a defect at or above the hypothalamus. The return to eucortisolism soon after correction of the weight loss indicates resolution of this central defect despite persistence of abnormalities in adrenal function.

We studied the relationship of serum insulin-like growth factor-I (IGF-I), IGF-II, the IGF-binding proteins IGFBP-1, IGFBP-2, and IGFBP-3, and GH-binding protein (GHBP; which is postulated to be derived from the extracellular portion of the GH receptor) in normal volunteers and patients with anorexia nervosa before and after a refeeding program. Serum GHBP, IGF-I, and IGFBP-3 were all significantly decreased in low weight patients with anorexia nervosa and returned to nearly normal levels with refeeding. Fasting serum GH and serum IGFBP-1 and IGFBP-2 were significantly increased in low weight patients with anorexia nervosa and also returned to nearly normal levels with refeeding. Serum IGF-II was 27% lower in the low weight group than in normal subjects, but this difference was not statistically significant. Both serum IGF-I and IGF-II were positively correlated with serum IGFBP-3 and negatively correlated with serum IGFBP-1 and IGFBP-2. These data are consistent with the hypothesis that nutritional deprivation alters the GH-IGF axis by down-regulation of the GH receptor or its postreceptor mechanisms, and that this effect is reversible with refeeding.

Adrenal androgen secretion is decreased in patients with anorexia nervosa. To assess the reversibility of the decreased secretion with recovery of body weight, we measured ACTH-stimulated adrenal androgen levels at different stages of recovery. Basal plasma GH and somatomedin-C levels also were measured, because both have been proposed as potential stimuli for adrenal androgen secretion. When studied at low body weight [58 +/- 3% (+/- SEM) ideal BW], women with anorexia nervosa had decreased ACTH-stimulated levels of dehydroepiandrosterone (DHA), DHA sulfate (DHAS), and androstenedione and decreased DHA to cortisol, DHAS to cortisol, and androstenedione to cortisol ratios compared to normal women. Women who had recently completed a refeeding program (within 2-4 weeks, 81 +/- 2% ideal BW) had an increased somatomedin-C level compared to low weight patients, but similar ACTH-stimulated adrenal androgen levels. Long term weight-recovered women (86 +/- 4% ideal BW, recovery for more than 6 months, with resumption of menses), however, had significant increases in ACTH-stimulated DHA and DHAS levels and DHA to cortisol and DHAS to cortisol ratios, and their hormone levels and ratios were not different from those in normal women. GH levels fell during weight recovery, although the values in the three patient groups did not differ significantly. We conclude that the recovery of adrenal androgen secretion while GH levels were falling provides evidence against a direct effect of GH as a stimulus for adrenal androgen secretion. The recovery of somatomedin-C before the recovery of adrenal androgens, however, and the positive correlation between plasma somatomedin-C and the integrated level of plasma DHAS (r = 0.50; P less than 0.02) are consistent with the hypothesis that somatomedin-C is a stimulus for adrenal androgen secretion.