Devogelaer and associates fail to recognize our main findings: 1) an inverse correlation between hydrocortisone dose per kg of body weight and bone mineral density [BMD] of the lumbar spine in men receiving long-term, conventional glucocorticoid replacement therapy for Addison disease and 2) the decreased BMD in 30% of these patients associated with use of a relatively high dose of hydrocortisone per kg of body weight. The inevitable conclusion is that conventional glucocorticoid replacement therapy induces chronic, mild oversubstitution; the title of our article is therefore correct. The effects of glucocorticoid treatment on statural growth in children with the adrenogenital syndrome are not completely comparable to the effects of glucocorticoid replacement therapy on BMD in adults with Addison disease. That other measures of glucocorticoid replacement therapy have failed to recognize this phenomenon of overreplacement only emphasizes the value of BMD measurements.

In Devogelaer and colleagues' study of 12 men receiving glucocorticoid replacement for Addison disease, a normal BMD was found. We gave some of the reasons for our differing results. The authors suggest that a BMD observed at the distal radius of less than 1 SD of the mean of the control group in three of their male patients might be attributed to tuberculosis in two patients and to diabetes mellitus in one patient. however, the authors did not try to correlate BMD with hydrocortisone dose per kg of body weight in their patients! In our group of 10 men with decreased BMD, we were unable to identify another cause for the decreased BMD; no diabetes was found, only one man had Addison disease of tuberculous origin, and none receiving thyroid replacement therapy for concomitant hypothyroidism had suppressed plasma thyroid-stimulating hormone concentrations.

Devogelaer and colleagues conclude from their findings that glucocorticoid replacement therapy in patients with Addison disease does not affect bone mass when secretion of ovarian estrogens in premenopausal women or testicular androgens are normal. The loss of BMD in postmenopausal women was therefore attributed to the low levels of adrenal androgens. In contrast, our findings suggest that glucorticoids may also be involved in this process.

We agree with Drs. Stoffer and Karkauer that, for optional glucocorticoid replacement therapy for Addison disease, not only the total daily dose but also the distribution over the 24-hour period is critically important. We have no experience with their protocol and with measurement of urinary free cortisol in the follow-up of patients with Addison disease. Our study shows that serum corticotropin levels are not a good measure of conventional glucocorticoid replacement therapy, but this may change when glucorticosteroids are administered in a more physiologic way. Serum sodium levels do not appear useful in "fine-tuning" glucocorticoid replacement therapy. In our 91 patients with Addison disease, a serum sodium level less than 138 mmol/L was found in only two, and hypernatremia was not detected. Although plasma renin levels are useful in monitoring mineralocorticoid replacement in particular, plasma aldosterone levels should not be measured because they are almost invariably low because of adrenocortical destruction.

Dr. Wener raises some important questions. We did not use a Bonferroni or other statistical correction in calculating the P values. The body weights in the men with decreased BMD was indeed lower than that in the men with normal BMD (74.7 ±9.8 kg compared with 80.9 =/- 11.0 kg), but the difference was not statistically different (P = 0.089). In the stepwise linear regression analysis, body mass index did not contribute significantly to the variance of BMD of the lumbar spine (r = 0.15, P = 0.41) or of the femoral neck (r = 0.30, P = 0.09). Therefore, we conclude that the weight-adjusted glucocorticoid dose is more important than body mass in determining BMD in men with Addison disease.

We found no difference in duration of glucocorticoid therapy between the men with normal BMD and those with decreased BMD. The calculated cumulative glucocorticoid dose was not statistically different in the men with decreased BMD than in the men with normal BMD (136 ± 138 g of hydrocortisone) compared with 115 ± 95 (P = 0.62). The range of duration of glucorticoid therapy among the men with decreased BMD was wide (2 to 32 years) and two men experienced bone loss during a relatively short time (2 and 3 years, respectively), but we were unable to identify any other factor that could have contributed to bone loss in any of our patients.