Bulletin Vol 6, Issue 1, Winter 2000

Target Pituitary

Articles in this issue:

Neuroendocrine Center Bulletin Archives
Of note, there have been significant advances in our understanding and treatment of pituitary disease since the initiation of the Bulletin more than 20 years ago, and early issues of the bulletin should be read wtih this in mind.

Long-Term Mortality After Transsphenoidal Surgery for Cushing's Disease

Pituitary Home

Brooke Swearingen
by Brooke Swearingen, M.D.

NEPTCC Newsletter MGH Neuroendocrine Center Bulletin Vol 6, Issue 1, Winter 2000

Untreated Cushing’s Cyndrome

The clinical course of untreated Cushing’s syndrome is marked by a significant increase in morbidity and mortality. In the original report of this syndrome written by Harvey Cushing in the 1930s, the duration from presentation of illness to death was 4.7 years. In 1952 the 5 year survival rate for patients was approximately 50%. The introduction of early surgical procedures to cure hypercortisolism and the advent of modern glucocorticoid replacement regimens improved the 5 year survival rate after adrenalectomy to 86%. With the advent of modern neurosurgical techniques using transsphenoidal resection, cure rates of hypercortisolism have changed dramatically. Current cure rates for patients with microadenomas, are approximately 90% when performed by experienced pituitary neurosurgeons. An important unanswered question has been the impact of current diagnostic and therapeutic approaches on the long-term mortality rate of patients with Cushing’s disease. A retrospective case series of 161 patients treated for Cushing’s disease at Massachusetts General Hospital between 1978 and 1996 was therefore done to determine long-term mortality rates in patients treated for Cushing’s disease with modern neurosurgical techniques. In this study, records were reviewed for all patients who underwent transsphenoidal surgery for documented Cushing’s disease and all surviving patients were contacted, with deaths confirmed by hospital, physician or family records.

The diagnosis of Cushing’s disease was based on clinical and biochemical evidence. A normal or elevated plasma ACTH level and the results of abnormal suppression testing were all consistent with the diagnosis of Cushing’s disease, and cure was defined as a fasting serum cortisol level of less than 138nm/L and a urine free cortisol of less than 55nm/day. Recurrence was determined in surviving patients by endocrine reevaluation and questionnaire reports. A Kaplan-Meyer product-limit estimation with 95% confidence intervals was used to analyze survival. The 161 patients (32 men and 129 women) had a total of 193 transsphenoidal procedures for Cushing’s disease. The mean age at the time of surgery was 38 years with a range of 8-76 years. Eighty-nine percent of the patients had microadenomas as defined by maximum tumor diameter of less than 1 cm, and 90% of these patients were cured. Of patients with macroadenomas, 65% were surgically cured. The overall cure rate for all patients was 85% and 28 of the patients required multiple procedures. Among the 136 cured patients with long-term follow-up, 7% of patients showed evidence of recurrence with a post-operative interval of 1 to 11 years (median 4 years). Therefore the long-term cure rate for patients with microadenomas was 96% at 5 years and 93% at 10 years, compared to 91% and 55% for macroadenomas. There were no perioperative deaths resulting from the transsphenoidal procedure at our Center. The most common complication was persistent sinus congestion in 9% of patients. Among major complications, there was a 2.6% evidence of cerebrospinal fluid rhinorrhea requiring repair and 1.5% incidence of meningitis. Permanent diabetes insipidus occurred in 6% of cases. In patients cured after one procedure, ACTH, TSH and gonadotropin insufficiency was found in 31, 23 and 14% of patients respectively.


Survival data was obtained in 99% of surgically treated patients with a median follow-up of 8 years. Six patients, 62 to 81 years of age, died at intervals of 4 to 9 years after surgery and the causes of death were cardiovascular in two patients, stroke in two patients, lymphoma in one patient and trauma in one patient. Of importance, the overall survival in the patient group was similar to that in an age- and sex-matched sample from the normal United States population (Figure 1). The overall 5 year survival rate was 99% and the 10 year survival rate was 93%.



The advent of modern diagnostic techniques for Cushing’s disease, improved neurosurgical procedures and refinement in post-operative management and hormone replacement therapy has had a dramatic impact on short-term outcomes and long-term survivals for patients with Cushing’s disease. In contrast to initial reports showing a marked increase in mortality rates among patients with this disease, current data now indicate that transsphenoidal surgical techniques performed in centers with an expertise in pituitary surgery provide cure in the vast majority of patients. Patients who are cured appear to now have survival rates no different from that of the United States population. Of note, other studies that have examined the long-term outcome in Cushing’s disease show a decreased survival despite therapy in a series where a significant minority of patients had ongoing cortisol excess. These data demonstrate that normalization of the hypercortisolemic state can have a significant impact on long-term mortality to the extent that it is indistinguishable from a normal population. In addition, the overall cure rates for patients with microadenomas, representing the vast majority of patients, is approximately 90% and among patients with microadenomas the long-term 10 year cure rate remains high at 93% (Figure 2).

For patients with macroadenomas, the overall cure rate is significantly less at 65% and the 10 year cure rate is 55%. Therefore, even in patients who have short-term surgical cures, vigilance is critical in detecting early recurrence and initiating aggressive therapy to normalize serum cortisol levels.

  1. Cushing H. The basophil adenomas of the pituitary body and their clinical manifestations. Bulletin of the Johns Hopkins Hospital. 1932; 50:137-95.
  2. Meier CA, Biller BM. Clinical and biochemical evaluation of Cushing’s syndrome. Endocrinol Metab Clin North Am. 1997; 26:741-62.
  3. Orth DN, Liddle GW. Results of treatment in 108 patients with Cushing’s syndrome. N Engl J Med. 1971; 285:243-7.
  4. O’Riordain DS, Farley DR, Young WF Jr, Grant CS, van Heerden JA. Long-term outcome of bilateral adrenalectomy in patients with Cushing’s syndrome. Surgery. 1994; 116:1088-93.
  5. Mampalam TJ, Tyrrell JB, Wilson CB. Transsphenoidal microsurgery for Cushing’s disease. A report of 216 cases. Ann Intern Med. 1988; 109:487-93.
  6. Katznelson L, Bogan JS, Trob JR, Schoenfeld DA, Hedley-Whyte ET, Hsu DW, et al. Biochemical assessment of Cushing’s disease in patients with corticotroph macroadenomas. J Clin Endocrinol Metab. 1998; 83:1619-23.
  7. Blevins LS Jr, Christy JH, Khajavi M, Tindall GT. Outcomes of therapy for Cushing’s disease due to adrenocorticotropin-secreting pituitary macroadenomas. J Clin Endocrinol Metab. 1998; 83:63-7.
  8. Swearingen B, Biller BMK, Barker F, Katznelson L, Grinspoon S, Klibanski A, Zervas N. Long-term mortality after transsphenoidal surgery for Cushing disease. Ann Intern Med. 1999; 130(10): 821-4.

Neuroendocrine Complications of radiation Therapy for Non-Pituitary Tumors

Pituitary Home

Anne Klibanski Howard H. Pai
by Howard H. Pai, M.D., FRCPC, and Anne Klibanski, M.D.

NEPTCC Newsletter MGH Neuroendocrine Center Bulletin Vol 6, Issue 1, Winter 2000

Non-Pituitary Tumors

The treatment of benign and malignant neoplasms in the head and neck region and the brain represent a special challenge to oncologists due to the close proximity of these tumors to neurovascular structures. Surgical access is limited and complete tumor resection is often not possible due to the risk of damage to neurovascular structures with aggressive surgery. Extensive surgery can also result in significant disfigurement or loss of function. Some examples include malignant neoplasms of the nasopharynx, nasal cavity and paranasal sinuses, benign and malignant tumors of the brain such as meningiomas, gliomas, and neuroectodermal tumors. Neoplasms located in the base of skull region is another example of where surgical resection is limited by the presence of critical structures such as the pituitary gland, hypothalamus, brainstem, cranial nerves and blood vessels. External beam radiation therapy has the advantage of being non-invasive and has an established role in the management of these tumors. The delivery of radiation therapy to this region must be very carefully planned, as these structures are also susceptible to damage by radiation. One clinically relevant aspect of radiation effects on normal tissue in this region is the neuroendocrine effect of radiation to the pituitary gland and hypothalamus, which is the focus of this article

It is well-established that therapeutic doses of radiation can cause damage to the pituitary gland and hypothalamus. For example, patients with secretory pituitary adenomas who are treated with external beam radiation therapy, typically with fractionated doses of at least 45 Gy, have a gradual decline in excess hormone production by the adenoma. However, because the rest of the normal pituitary gland receives the full dose of radiation, a decline in the other hormones secreted by the pituitary gland is frequently observed. One could debate whether the presence of the adenoma itself compromises pituitary gland function thus contributing to hypopituitarism after external beam radiation therapy. However, patients treated with external beam radiation therapy for non-pituitary gland neoplasms also exhibit hypopituitarism following external beam radiation therapy. Evidence supporting hypothalamus and pituitary gland damage following external beam radiation therapy for non-pituitary gland tumors comes from several sources. Children who receive cranial or craniospinal irradiation for leukemia or primary tumors of the brain often develop hypopituitarism. Hypopituitarism is manifested by anterior pituitary gland hormone deficiencies. Clinically apparent growth hormone insufficiency is particularly prevalent in children. Other anterior pituitary gland hormones can also be affected with deficiencies in TSH, LH, FSH, and ACTH. Patients can also develop hyperprolactinemia after external beam radiation therapy. The mechanism involves decreased availability of dopamine from the hypothalamus to the pituitary gland through radiation damage to the hypothalamus or portal system. The inhibitory effect of dopamine on prolactin secretion is diminished, resulting in a rise in prolactin. More than one hormone can be affected. A less common neuroendocrine effect is precocious puberty observed in prepubertal patients. For reasons not well understood, vasopressin insufficiency (an indicator of posterior pituitary gland function) after external beam radiation therapy is rarely seen.

Radiation induced pituitary gland-hypothalamic dysfunction also occurs in adults with high incidence. The same type of anterior hormone deficiencies can occur as with children. The incidence of each hormone defect varies from study to study. Growth hormone deficiency can range from 60 to 100%, hyperprolactinemia from 15% to 85%, hypothyroidism from 15 to 65%, hypoadrenalism from 14 to 55%, and hypogonadism from 30 to 60%. Clinically overt diabetes insipidus is a very uncommon event in adults, similar to children.

Hypopituitarism after external beam radiation therapy is a late effect typically occurring 2-3 years after radiation therapy but as early as 6 months. Patients continue to be at risk many years after completion of external beam radiation therapy with cases documented 10 to 15 years after treatment. Thus, the importance of lifelong monitoring after external beam radiation therapy cannot be over-emphasized. The patho-physiological mechanism of damage is not well elucidated but may reflect damage to the microvasculature of the pituitary gland or portal system or direct damage to the hormone producing cells of the pituitary gland. The hypothalamus is also susceptible to radiation injury resulting in hypopituitarism.

Monitoring of hypothalamic-pituitary gland function following external beam radiation therapy begins with a complete baseline evaluation prior to completion of external beam radiation therapy to detect any pre-existing endocrine deficits. Baseline neuroendocrine evaluation should assess central and end organ endocrine function including thyroid, adrenal, gonadal, prolactin, vasopressin, and growth hormone secretion when appropriate, using history and physical and blood tests. Provocative blood tests can be used to determine a central or primary origin but certain stimulatory tests such as insulin stress test are relatively contraindicated in patients with brain or parasellar tumors due to the risk of hypoglycemia induced seizures especially if prior neurosurgical procedures have been performed. Patients should be routinely monitored for endocrine changes at 6 months after completion of external beam radiation therapy and then at 1 year and yearly thereafter. Our protocol for neuroendocrine follow-up after external beam radiation therapy to the pituitary gland-hypothalamic region consists of a history and physical, blood levels for prolactin, T4, fT4, 8 am cortisol and/or ACTH stimulation test, FSH, LH, free and total testosterone in males and estradiol in non-menstruating pre-menopausal females. Subclinical adrenal insufficiency is important to diagnosis and treat to avoid precipitating acute adrenal insufficiency during periods of stress such as a surgical procedure. Tests for vasopressin or growth hormone insufficiency are only ordered for adults with symptoms or signs suspicious for diabetes insipidus or in whom GH therapy is contemplated, respectively. Hormone replacement should be instituted when an endocrinopathy is found. Controversy exists as to whether GH replacement should be offered for adult patients with neoplasms. Although there are no data supporting tumor proliferation during exogenous GH administration, it has been our policy to avoid growth hormone replacement for adults after radiation therapy for malignant or aggressive neoplasms.

Table 1. Neoplasms which frequently require radiation therapy to the pituitary / hypothalamus region

  • Primitive neuroectodermal tumor of the CNS (e.g. medulloblastoma)
  • Pineal gland tumors
  • CNS Germ cell tumors
  • Gliomas, ependymomas of the brainstem or thalamus region
  • Pituitary adenomas
  • Craniopharyngiomas
  • Parasellar meningiomas
  • Chordomas, chondrosarcomas, giant cell tumors of the clivus or parasellar region
  • Cancer of the nasopharynx, nasal cavity or paranasal sinuses
  • Leukemia

Not all patients who receive radiation to the pituitary gland-hypothalamic region develop pituitary or hypothalamic insufficiency. Several factors may influence the risk of developing hypopituitarism. These include age, gender, daily dose of radiation and total dose of radiation delivered to the pituitary gland and hypothalamus. With respect to gender, women become hyperprolactinemic after radiation more often than men. Patients who are older (e.g. > 40-50 years) tend to have a higher incidence of endocrinopathies after radiation, noted in 3 separate studies. One possible explanation for this may be decreased pituitary reserve with aging. Limited data have suggested that fraction size or the daily dose of radiation delivered may influence the risk of endocrinopathy with larger fraction sizes causing a higher incidence of hypopituitarism [1]. This observation is consistent with the general axiom that larger fraction size is associated with increased late toxicity for neurovascular tissue. The total dose of radiation absorbed by the pituitary gland and hypothalamus is undoubtedly a risk factor for causing hypothalamic-pituitary gland damage resulting in hypopituitarism. In the early 1980’s, the first indication of a dose response was suggested in a report from Australia, albeit from a small series of patients and with only estimations of dose to the pituitary gland and hypothalamus and using fraction sizes not considered standard in the US. In the late 80’s, a large series of patients (n=268) who received cranial irradiation for various neoplasms were analyzed for dose response effect [1]. It was noted that at very low therapeutic doses of radiation of 12 Gy, the incidence of endocrinopathy was negligible. At doses > 20 Gy, the incidence became clinically relevant and a dose response was seen between 20 Gy and 35 Gy or above. This study was performed in United Kingdom where higher dose per fraction (e.g. 2.5 to 3.75 Gy per fraction) was used compared to US standards. A smaller series of patients (n=32) treated at the University of Rochester in NY were analyzed and found to have a dose response such that patients receiving doses greater than 50 Gy had a higher incidence of hypothyroidism and hypoadrenalism [2]. A dose response from 18 Gy to > 24 Gy to > 35 Gy for GH insufficiency exists for children treated with cranial irradiation for acute leukemia or brain tumors. At Massachusetts General Hospital and the Harvard Cyclotron Laboratory, we have treated over 500 patients with moderate to high dose radiation using proton beam radiation therapy to the parasellar region for non-pituitary gland and non-hypothalamus tumors of the base of skull region, most typically chondrosarcomas and chordomas of the clival region. A significant number of these patients have been followed for neuroendocrine outcome prospectively and recent analysis of 107 such adult patients have shown that doses above 50 centigray equivalent (CGE) to the pituitary gland or hypothalamus significantly increases the risk of hypopituitarism [3]. Patients receiving 70 CGE to any portion of the pituitary gland were also at increased risk. A high incidence of hyperprolactinemia was also observed using proton therapy with a 10-year actuarial incidence of over 85%. What was unique to this study was the ability to determine the dose to the pituitary gland and hypothalamus more precisely using 3 dimensional computerized treatment planning algorithms not available in the past at other institutions.


Radiation therapy has a proven role in the treatment of neoplasms arising from the brain and head and neck region where sensitive neurovascular structures preclude aggressive surgical resection. Radiation can also cause damage to these neurovascular structures such as the pituitary gland and hypothalamus resulting in hypopituitarism. Both children and adults can be affected and the incidence of hypopituitarism is very high. Patients often require hormone replacement therapy and life long monitoring is strongly recommended. Patients with neoplasms listed in Table I are recommended to have longitudinal endocrine follow-up after radiation therapy. However, any patient receiving radiation therapy where the pituitary gland and hypothalamic region is at risk for irradiation should be screened as well. Radiation induced hypopituitarism is generally limited to dysfunction of the anterior pituitary gland. The types of endocrinopathies, timing and risk factors have been described. The radiation dose is an important factor with several studies indicating a dose response for hypopituitarism. In the last decade, advancements in computer technology, diagnostic imaging and radiation treatment techniques have enabled radiation oncologists to deliver radiation more precisely to the intended target. This approach limits the dose to surrounding normal tissue thereby improving the therapeutic ratio. Efforts should continue in this direction to reduce the incidence of radiation induced hypopituitarism.

  1. Littley, M D, Shalet, S M, Beardwell, C G, Robinson, E L, Sutton, M L., Radiation-induced hypopituitarism is dose-dependent. Clinical Endocrinology, 1989. 31: p. 464-373
  2. Constine, Louis S, Woolf, Paul D, Cann, Donald, Mick, Gail, McCormick, Kenneth, Raubertas, Richard F, Rubin, Philip. Hypothalamic-Pituitary dysfunction after radiation for brain tumors. N Engl J Med, 1993. 328: p. 87-94.
  3. Pai, Howard H, Katznelson, Laurence, Klibanski, Anne, Finkelstein, Dianne M, Adams, Judith A, Fullerton, Barbara C, Thornton, Allan, Leibsch, Norbert J., Munzenrider, John E. Hypothalamic/Pituitary Gland Dysfunction following High Dose Conformal Mixed Proton-Photon Beam Radiotherapy to the Base of Skull Region: Demonstration of a Dose Effect Relationship using Dose Volume Histogram Analysis. in 41th ASTRO (American Society for Therapeutic Radiology and Oncology) conference. 1999. San Antonio, TX.

Growth Hormone Replacement in Asults: Cardiovascular Considerations

Pituitary Home

Howard H. Pai
by Gemma Sesmilo, M.D.

NEPTCC Newsletter MGH Neuroendocrine Center Bulletin Vol 6, Issue 1, Winter 2000


The approval of growth hormone (GH) for the treatment of adults with GH deficiency has raised a great deal of interest regarding the long-term benefits of this therapy. A number of clinical studies have demonstrated that GH replacement has beneficial effects on body composition and bone mineral density. A topic of new and important interest is how growth hormone affects the cardiovascular risk profile.

Life expectancy and cardiovascular disease in GH deficient patients

Rosen & Bengtsson in 1990 reported that hypopituitary patients receiving conventional replacement had a decreased life expectancy. They examined the records of 333 patients diagnosed with hypopituitarism between 1956 and 1987 from their endocrine clinic in Goetheburg and found a higher mortality rate in these patients than in the Swedish population. This increased mortality was attributed to cardiovascular disease.

Cross-sectional studies have demonstrated a higher prevalence of atherosclerotic plaques, endothelial dysfunction and increased carotid intimal-medial thickness in hypopituitary patients as compared to controls, even at early stages of the disease. These indicators of atherosclerosis have been shown to correlate with the incidence of coronary events in epidemiological studies. These findings are consistent with the concept of increased cardiovascular risk in hypopituitary patients, however, the role of GH and the effect of GH replacement on this risk is less well known.

Clinical characteristics of the GH deficiency syndrome

The growth hormone deficiency syndrome is characterized by obesity with increased body fat. The excess fat is centrally distributed mostly in the visceral compartment, which is a known cardiovascular risk factor. Other features such as insulin resistance, impaired plasma fibrinolytic activity and dyslipoproteinemia have been described in growth hormone deficient patients in cross-sectional studies and all are thought to contribute to increased cardiovascular risk.

GH effects on body fat distribution

Growth hormone replacement therapy decreases total body fat, including visceral fat and increases lean body mass, resulting in no net change in body weight. Several groups but not all, have demonstrated reduction in central fat with GH treatment. Important differences among studies exist and these are likely due to different doses of GH used as well as duration of therapy. GH is a lipolytic hormone with known dose dependent effects. The first reports regarding GH replacement in adults used high doses, resulting in IGF-I values out of the reference range, with high incidence of adverse events mainly due to fluid retention. More physiological approaches are discordant in the reduction of central fat as assessed by the waist to hip ratio.

GH effects on lipid levels

There are conflicting reports regarding the effect of GH replacement on the lipid profile. One of the most important limitations is the lack of long-term controlled trials, which makes it difficult to ascertain the effects due to GH versus the placebo effect. A randomized placebo-controlled study conducted in our Unit assessed the effect of physiological doses of GH on 32 GH deficient patients treated over 18 months. No long-term changes in the lipid profile were found. Some other studies with a short-term placebo-controlled phase, followed by an open follow-up, have shown reductions in LDL cholesterol and/or increases in HDL cholesterol but others have not. There is agreement in the increase of lipoprotein (a) [Lp(a)] levels with GH replacement, but it is still not clear how Lp (a) contributes to cardiovascular risk.

GH effects on glucose metabolism

GH is known to have anti-insulinic properties, whereas IGF-I is an insulinotropic agent. GH replacement has been reported to impair insulin action in many studies, but studies of insulin sensitivity using clamp techniques have shown reversibility of these findings with maintained GH treatment. Other approaches to the study of glucose metabolism in treated GH deficient patients have not been able to show a complete recovery of the initial decline in insulin sensitivity caused by GH. While it is known that GH treatment initially causes insulin resistance, it is thought that the changes in body composition with decreases in body fat and increases in lean body mass can contribute to the reversal of this effect. It is recommended that glucose levels be monitored in patients who initiate GH treatment. Typically if a patient develops diabetes, the drug is discontinued. Diet and exercise should be reinforced in patients at risk.

Proposed mechanisms of GH effect on atherosclerosis

Based on the reported effects of GH on different cardiovascular risk factors, it is difficult to know how GH administration will affect the process of atherosclerosis. There are two prospective open-labeled studies that have assessed carotid intimal-medial thickness as an indicator of atherosclerosis in GH treated patients. Both of them showed a decrease in this parameter, as early as 6 months after treatment (Figure 1).

Figure 1. Changes of intima media thickness (mean and SE) of the common carotid artery (CCA) during GH treatment. Reprinted with permission from Pfeifer et al. J Clin Endocrinol Metab. 1999; 84(2):453-7.


GH deficient patients have evidence of early atherosclerosis and increased cardiovascular mortality. Ultrasonographic studies have reported a decrease in carotid intimal-medial thickness after GH administration, suggesting a beneficial effect of GH on atherosclerosis. GH decreases total and central body fat, increases Lp (a), but effects on other lipoproteins are more controversial. GH initially causes insulin resistance which may be restored with prolonged treatment. Proposed mechanisms of GH action on atherosclerosis include reduction of nitric oxide production and inflammatory activity modulation. Cardiovascular risk may prove to be an important factor in determining the benefit to risk ratio of GH replacement. However, further studies using clinical cardiovascular end-points are needed to confirm the beneficial effect of GH replacement in the process of atherosclerosis.

  1. Rosen T, Bengtsson BA. Premature mortality due to cardiovascular disease in hypopituitarism. Lancet. 1990; 336(8710):285-8.
  2. Baum HB, Biller BM, Finkelstein JS, et al. Effects of physiologic growth hormone therapy on bone density and body composition in patients with adult-onset growth hormone deficiency. A randomized, placebo-controlled trial. Ann Intern Med. 1996; 125(11):883-90.
  3. Evans LM, Davies JS, Goodfellow J, Rees JA, Scanlon MF. Endothelial dysfunction in hypopituitary adults with growth hormone deficiency. Clin Endocrinol (Oxf). 1999; 50(4):457-64.
  4. Fowelin J, Attvall S, Lager I, Bengtsson BA. Effects of treatment with recombinant human growth hormone on insulin sensitivity and glucose metabolism in adults with growth hormone deficiency. Metabolism. 1993; 42(11):1443-7.
  5. Pfeifer M, Verhovec R, Zizek B, Prezelj J, Poredos P, Clayton RN. Growth hormone (GH) treatment reverses early atherosclerotic changes in GH-deficient adults. J Clin Endocrinol Metab. 1999; 84(2):453-7.
  6. Serri O, St-Jacques P, Sartippour M, Renier G. Alterations of monocyte function in patients with growth hormone (GH) deficiency: effect of substitutive GH therapy. J Clin Endocrinol Metab. 1999; 84(1):58-63.
  7. Figure 1. Changes of intima media thickness (mean and SE) of the common carotid artery (CCA) during GH treatment. Reprinted with permission from Pfeifer et al. J Clin Endocrinol Metab. 1999; 84(2):453-7.