Our data demonstrate lower mean total testosterone and free testosterone, but not DHEAS, levels in women with anorexia nervosa compared with healthy controls and normal-weight women with hypothalamic amenorrhea. Women with anorexia nervosa who were receiving oral contraceptives had the lowest levels of free testosterone and DHEAS, much lower even than women with anorexia nervosa not receiving oral contraceptives. DHEAS levels were not reduced in women with anorexia nervosa, except for those who were receiving oral contraceptives. In addition, normal-weight women with hypothalamic amenorrhea did not have lower mean levels of total testosterone, free testosterone, or DHEAS than healthy controls. Of note, healthy, eumenorrheic controls were all studied during the early follicular phase of the menstrual cycle. Because testosterone and free testosterone levels increase 20–30% at midcycle, the differences between the women with anorexia nervosa and the controls would have been even greater if we had measured androgen levels at that time. Also of note, endogenous total testosterone, free testosterone, and DHEAS levels were all significant determinants of bone density in this cross-sectional study, raising the question of whether hypoandrogenemia may contribute to low bone density in women with anorexia nervosa and whether the lack of effectiveness of oral contraceptives as a therapy for bone loss may in part be attributed to the resultant reduction in free testosterone levels. Interventional studies would be necessary to investigate this possible mechanism further.
Previous reports in small numbers of women have shown variably low (11
), normal (6
), or elevated (15
) androgens or DHEAS levels in women with anorexia nervosa. Some of these studies compared androgen or preandrogen levels with published normal ranges, instead of with those of recruited healthy control groups. Similarly, investigations of androgen levels in normal-weight women with hypothalamic amenorrhea have yielded conflicting results. There have been several reports of elevated testosterone levels in amenorrheic and oligoamenorrheic athletes (20
), whereas other reports of normal-weight women with hypothalamic amenorrhea have demonstrated decreased levels (19
). In contrast, the data regarding the effects of oral contraceptives on androgen levels in healthy women of reproductive age are congruent in that they clearly demonstrate decreases in free testosterone and DHEAS in healthy women of reproductive age (27
). To our knowledge, ours is the first report of similar effects in women with anorexia nervosa, in whom the reduction of endogenous anabolic hormones could prove particularly relevant in that it may contribute to the known loss of bone mass and lean body mass in this disease.
Bone density is reduced in men with hypogonadism, and testosterone replacement results in reversal of the bone loss incurred. Moreover, we have recently shown that low-dose testosterone replacement therapy increases bone density at the hip and radius in women with severe androgen deficiency due to hypopituitarism (43
). Three other randomized placebo-controlled studies have demonstrated increases in bone density with androgen administration in surgically or naturally menopausal women (44
), whereas one randomized placebo-controlled study showed no difference with methyltestosterone plus esterified estrogens compared with esterified estrogens alone in postmenopausal women (47
). In a study in which DHEA was administered to girls with anorexia nervosa for 1 yr, an increase in hip bone density compared with baseline was found in the initial analysis but was not detectable after controlling for weight gain; moreover, no increase in spine bone density was observed (13
). Cross-sectional studies in healthy women have shown strong positive correlations between both free testosterone and percent free testosterone and bone density (48
). In a longitudinal study, Slemenda et al.
) showed that lower androgen levels predicted subsequent bone loss in premenopausal, perimenopausal, and postmenopausal women. Our group has previously reported strong associations between change in free testosterone and change in surrogate markers of bone formation in adolescent girls with anorexia nervosa (50
). However, to our knowledge, this is the first report of positive associations between bone mineral density and androgens in adults with anorexia nervosa. This finding could be clinically significant in a disease that is complicated by severe bone loss. The fact that associations did not remain significant after controlling for BMI reflects the relationship reported in this manuscript between undernutrition and hypoandrogenemia. Prospective, interventional studies will be necessary to determine whether there is an independent effect of circulating androgens on BMD in anorexia nervosa.
Limitations of this study include its cross-sectional design, assay limitations for androgen levels in women, and the fact that all samples were not drawn at 0800 h. The positive associations observed between both weight and body fat and androgen levels suggest that low weight and/or abnormalities in body composition are mechanisms underlying the hypoandrogenemia in women with anorexia nervosa. However, the reverse is also possible, and causality can be established only in prospective trials. Another limitation of the study is the assay for testosterone used. This is particularly true at the low levels observed in women (42
). Tandem mass spectrometry has been recently introduced and is increasingly accepted as the most accurate testosterone assay, but its use is limited by lack of wide availability and high cost. The direct RIA used in this study, along with one other assay, were shown to correlate most closely with tandem mass spectrometry of 11 manual and automated immunoassays examined in 122 men in a study by Wang et al.
). In the current study, we calculated free testosterone levels using total testosterone and SHBG levels and the mass action equation. We have validated this method as yielding similar results to equilibrium dialysis (42
), but it is dependent on the validity of the total testosterone and SHBG level used.
Our data demonstrate that total and free testosterone, but not DHEAS, are reduced in anorexia nervosa. Moreover, marked reductions in both free testosterone and DHEAS occur in women with anorexia nervosa who take oral contraceptives. In contrast, normal-weight women with hypothalamic amenorrhea appear to have normal androgen and DHEAS levels. Free testosterone, total testosterone, and DHEAS levels predict bone density at most skeletal sites tested in this cross-sectional study. It is not known whether the reduction in free testosterone and DHEAS levels in women with anorexia nervosa using oral contraceptives is harmful to skeletal health or has other deleterious effects. Interventional studies are needed to further investigate these issues.