| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Original Studies |
Departments of Endocrinology and Chemical Pathology (J.A.-Z., J.J., G.C.), Imperial College School of Medicine, London, United Kingdom W6 8RF
Address all correspondence and requests for reprints to: Dr. Donal O’Shea, University College Dublin, Department of Investigative Endocrinology, St. Vincent’s Hospital, Dublin 4, Ireland. E-mail: doshea{at}rpms.ac.uk
 |
Abstract |
|---|
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
|
Introduction |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
No single endocrine test is 100% accurate in the diagnosis of HPA insufficiency; it must be combined with clinical assessment of the individual patient. Failure to recognize secondary adrenal insufficiency can have potentially life-threatening consequences, whereas unnecessary hydrocortisone replacement therapy may cause significant morbidity in terms of glucocorticoid side-effects (4, 5, 6). The ITT is the gold standard for the assessment of the HPA axis and is the only test that has been validated against a clinical end point (7). The ITT is unpleasant and potentially dangerous. It is contraindicated in patients with heart disease, epilepsy, or those over 60 yr. The 250-µg SST is the most commonly used alternative to the ITT. The SST assesses adrenal responsiveness. Its use in the assessment of the HPA axis assumes that in HPA insufficiency the adrenal glands will atrophy in the absence of adequate ACTH. The SST is simpler and safer than the ITT, but it does not assess the entire HPA axis and does not show complete concordance with the ITT (8, 9, 10).
The ITT uses hypoglycemia as a reproducible stimulus to ACTH release. Sleep is established as a reliable and reproducible stimulus to ACTH release. The resulting increase in cortisol release is reflected in a rise in urinary free cortisol levels in the early morning. The ITT and SST rely on serum cortisol responses. It has recently been shown that serum cortisol levels are affected by variations in cortisol-binding globulin (CBG) and do not correlate well with cortisol production rates unless differences in CBG are corrected for (11, 12). Measured serum cortisol is likely to underestimate cortisol production in subjects with lower CBG levels. Urinary free cortisol levels appear to be unaffected by CBG and correlate well with changes in cortisol production (12). The ITT and SST are invasive and require hospital attendance, the stress of which may affect the HPA axis. Collection of spot urine samples for urinary free cortisol estimations is noninvasive and can be performed reliably at home. We hypothesized that a normal range existed for the increment in urinary cortisol levels between midnight and morning (Cort/Cr increment) and that patients with HPA insufficiency would have a below normal Cort/Cr increment. Our study was designed to determine whether the Cort/Cr increment could be used to reliably assess the integrity of the HPA axis.
|
Subjects and Methods |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Forty patients with pituitary disease (see Table 1
) were studied together with 40
controls. The patient group consisted of 21 females and 19 males with a
mean ± SEM age of 49.5 ± 2.6 yr. In the control
group there were 22 females and 18 males with a mean ±
SEM age of 37.0 ± 2.1 yr. Intraindividual variation
was examined in 7 of the controls and 23 of the patients by performing
urine collections on at least two separate occasions (6 weeks to 12
months apart). Postoperative patients performed urine collections
between 5 days and 4 yr after transsphenoidal surgery.
|
All subjects were given verbal and written instructions asking them to provide urine samples (5–20 mL) at midnight and their normal waking time. Urine collections were made of double voided samples to ensure that the urinary free cortisol level reflected cortisol production at the time of sample collection. Subjects were instructed to empty their bladders at 2300 h and to collect a urine sample 1 h later. On waking, subjects were asked to empty their bladders and collect a urine sample 1 h later. Patients taking hydrocortisone replacement stopped hydrocortisone 48 h before collecting urine samples. Samples were returned by post. All samples were collected at home. In addition, six patients provided specimens while in-patients on day 5 postoperatively. The Cort/Cr increment was defined as the morning Cort/Cr minus the midnight Cort/Cr. The study received approval from the local ethical committee (Riverside research ethics committee, no. 1762).
Analytical methods
Urinary free cortisol was measured by a competitive RIA on nonextracted urine (Orion Diagnostica, Turku, Finland). The sensitivity of this assay was 5 nmol/L, and the intra- and interassay variations were 7.8% and 14%, respectively. Urinary cortisol was stable for at least 48 h at room temperature (<5% change). To allow for individual variation in urine flow rate, urinary creatinine was also measured, and the urinary cortisol/creatinine ratio (Cort/Cr) was calculated. Urinary creatinine was measured by a kinetic Jaffé procedure using a Hitachi 747 analyzer (Roche Diagnostica, Lewes, UK) according to the manufacturer’s instructions. The Cort/Cr increment was compared in those patients with HPA insufficiency to the increment in controls and patients with intact HPA axis. Serum cortisol was assayed by RIA using an automated enzyme-linked immunosorbent assay (ES700, Boehringer Mannheim, Mannheim, Germany). CBG assays were performed using a ligand binding assay as described by Hammond and Lahteenmaki (15).
Statistical analysis
A normal range for the Cort/Cr increment in individuals with an intact HPA axis was established from the Cort/Cr increments in the 40 controls. The minimum value for the Cort/Cr increment in the controls was taken as the cut-off value for the diagnosis of an intact HPA axis. The aim of this study was to establish whether the Cort/Cr increment could be used to reliably differentiate between those individuals with an intact HPA axis and those with HPA insufficiency on the basis of this cut-off value. The Cort/Cr increment was compared to the results of conventional HPA testing using the ITT, or the SST when the ITT was contraindicated. The sensitivity, specificity, and positive and negative predictive values of the Cort/Cr increment were calculated on the assumption that the ITT/SST were correct in their assessments of the HPA axis.
|
Results |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
In the 40 controls, the midnight Cort/Cr ratio ranged from 3–20, and the morning Cort/Cr ratio ranged from 14–80. The midnight to morning Cort/Cr increment ranged from 9–75. From this we defined a normal Cort/Cr increment as more than 9.
From conventional endocrine testing (ITT or SST), 15 of the patients had an intact HPA axis (5 ITT and 10 SST), and 25 had HPA insufficiency (13 ITT and 12 SST). Of the 40 patients, 34 (87.5%) had Cort/Cr increments that were concordant with their ITT/SST. The positive predictive of a low Cort/Cr increment for the diagnosis of HPA insufficiency was 95%. For the diagnosis of HPA insufficiency, the Cort/Cr increment had a sensitivity of 80% and a specificity of 93%. The patients with discordant results are discussed in detail below. The use of the morning Cort/Cr ratio alone was less reliable. Taking a cut-off level for the morning Cort/Cr ratio of 14, the concordance with the ITT/SST was 78%.
Analyzing the 34 patients with concordant results, those with an intact
HPA axis had Cort/Cr increments ranging from 9–80. In contrast, the
Cort/Cr increment in patients with HPA insufficiency ranged from -12
to 8 (see Tables 2 and 3). Figure 1 shows the Cort/Cr increment in controls
and patients according to HPA status.
|
|
|
Thirty subjects (23 patients and 7 controls) performed urine collections on 2 separate occasions. Five of these patients performed 3 urine collections. The difference in Cort/Cr increment values on each occasion tested ranged from 2–50 in the 7 controls and from 0–32 in the 23 patients. However, the HPA status, as defined by the Cort/Cr increment, was consistent in all subjects on each occasion tested. This group included 6 patients who performed their initial urine collections on day 5 postoperatively.
Patients with discordant results
In six of the patients there was disagreement between the Cort/Cr
increment and the results of the ITT/SST (see Tables 2
and 3). Ninety
percent of total cortisol is bound to CBG. It is possible that
individuals with an intact HPA axis but a low CBG level might have an
apparently subnormal cortisol response to ITT/SST, as measured by their
serum (i.e. total) cortisol level. The CBG level in the
patients with discordant results was measured (Table 4).
|
Patient 15. This 36-yr-old woman was treated with surgery and radiotherapy 4 yr previously for acromegaly. Her anterior pituitary function remained intact until this year, when an ITT showed a subnormal response (346 nmol/L). She experienced no improvement with hydrocortisone replacement. Her Cort/Cr increments were 58 and 54 on two separate occasions. Her CBG level measured 4 months after starting hydrocortisone replacement was below normal (257 nmol/L; normal range in females, 290–420 nmol/L).
Patient 16. This 38-yr-old woman presented with secondary amennorhea, and a microprolactinoma was treated with bromocriptine. Her PRL level normalized, and her menses returned. Her thyroid hormone levels were normal throughout. An ITT performed showed a subnormal cortisol response (336 nmol/L) despite Cort/Cr increments of 41 and 19. Her CBG level was below normal (249 nmol/L).
Patient 17. This 29-yr-old woman with secondary amennorhea and an 8-mm prolactinoma underwent transsphenoidal surgery. Postoperatively, her PRL level normalized, and her menses resumed. Her thyroid axis was intact. Her Cort/Cr increment was 36, but her peak serum cortisol response to ITT was only 304 nmol/L. Her CBG level was below normal (259 nmol/L).
These three patients have normal urinary Cort/Cr increments, but subnormal ITT results. Their other clinical and endocrine features suggest that they each have an intact HPA axis. All three have low CBG levels. Therefore, it is likely that the measured serum (total) cortisol levels underestimate the true cortisol responses.
Patient 18. This 42-yr-old woman presented with a 2-yr history of secondary amennorhea and underwent transsphenoidal surgery for a nonfunctioning adenoma. Postoperatively her menses resumed, and her thyroid hormone levels were normal. Her peak cortisol response at ITT was only 323 nmol/L, and she was placed on hydrocortisone replacement therapy (10, 5, and 5 mg). She cut down the dosage herself to 5 mg daily and remained well. Her Cort/Cr increments were 9 and 10 on two separate occasions. Her CBG level was low (241 nmol/L). Therefore, her serum cortisol response probably underestimates her true cortisol response to hypoglycemia. Her urinary increment is borderline, and this has been interpreted as partial HPA insufficiency. She remains well on low dose hydrocortisone replacement (5 mg twice daily).
Patients with discordant SST results
Patient 39. This 67-yr-old man was successfully treated with surgery for acromegaly. Anterior pituitary function was intact preoperatively. His testosterone and thyroid hormone levels were also normal postoperatively. He had Cort/Cr increments of 35 and 24, but his peak cortisol response to SST was only 412 nmol/L. His CBG level was normal at 269 nmol/L (male normal range, 200–380 nmol/L). His biochemical and clinical features suggest that his HPA axis is intact, and that he has been misdiagnosed by SST.
Patient 40. This 61-yr-old woman was treated in 1985 for acromegaly with surgery and radiotherapy. She has low gonadotropin levels and secondary hypothyroidism. Her SST is a borderline pass at 556 nmol/L. Her Cort/Cr increments were 1 on two separate occasions. Her CBG level was normal (310 nmol/L). Given her history of radiotherapy and biochemical features, it seems likely that this woman has HPA insufficiency and has been misdiagnosed by SST.
|
Discussion |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
The SST is the most commonly used alternative when the ITT is contraindicated. The 250-µg dose of Synacthen used is several thousand times greater than physiological levels of ACTH, and this may explain some of the nonconcordance between the ITT and the SST. Recent studies suggest that the low dose (1 µg) Synacthen test may show better concordance with the ITT (16, 17), but the timing of the peak response varies from 10–40 min, making it difficult to establish optimal sampling times (18). Both the SST and the low dose Synacthen test are tests of adrenal reserve, and neither assesses the ability of the entire HPA axis to respond physiologically.
Twenty-two of the 40 patients tested were assessed by SST rather than ITT. This is clearly problematic in the analysis of our results, as the SST does not show complete concordance with the ITT. We have used a cut-off of 550 nmol/L for the 30-min cortisol response to 250 µg Synacthen, which gives a concordance of approximately 64% compared with the ITT (8). The concordance can be improved by altering the cortisol cut-off values. If HPA insufficiency is defined as a cortisol value of less than 350 nmol/L and an intact HPA axis as a cortisol level greater than 600 nmol/L, the concordance rises to 96%. Using these cut-off values, patients with cortisol responses between 350–600 nmol/L would be classed as equivocal, i.e. patients 26, 27, and 33–40. Reanalysis of our data, excluding these 10 patients with equivocal SST results, gives the midnight to morning Cort/Cr increment a concordance of 87% with the ITT/unequivocal SST. The positive predictive value of a low Cort/Cr is 100% for the diagnosis of HPA insufficiency, and the positive predictive value of a normal Cort/Cr for the diagnosis of an intact HPA axis is 78%.
The ITT and SST rely on the measurement of serum cortisol levels as an estimation of the individual’s cortisol response to stimulation. Serum cortisol measures total cortisol, of which 80–90% of total cortisol is bound to CBG (11). There is considerable variation in CBG levels between individuals. Therefore, the same free cortisol response in two individuals might give very different serum (i.e. total) cortisol levels if there were significant differences in CBG levels. Our study measures urinary free cortisol, which appears to be unaffected by CBG (12).
The poor correlation between cortisol production and serum cortisol levels was demonstrated by Bright et al. in a study of 36 subjects (12). Cortisol production was simulated by iv cortisol infusions at different rates (6–180 nmol/m2/min) over an 8-h period. This study found that serum cortisol was a poor predictor of cortisol responses (r2 = 0.13). In contrast, there was a strong correlation between urinary free cortisol and cortisol infusion rate (r2 = 0.82). The addition of individual CBG levels as a covariate greatly improved the predictability of plasma cortisol responses, and CBG levels accounted for about 40% of the variability in serum cortisol levels compared to the cortisol infusion rate.
In this study, 34 of the 40 patients (87.5%) had midnight to morning Cort/Cr increments that were concordant with their ITT/SST results. Neither the ITT nor the SST is 100% accurate in assessment of the HPA axis. Individuals with lower CBG have lower measured serum cortisol levels for a given cortisol production rate (11). Patients 15, 16, and 17 had subnormal ITT results but normal Cort/Cr increments. All three patients had low CBG levels. Clinically, these patients have an intact HPA axis, and we would suggest that these three patients have been misclassified by their ITTs. The clinical and biochemical features of the other three patients with discordant results suggest that they have also been incorrectly classified by their ITT/SST results.
Three of the five patients with acromegaly (patients 16, 39, and 49) had Cort/Cr increments that were discordant with their ITT/SST results. Studies of patients with GH deficiency suggest that GH may influence CBG (19, 20), and this may explain the high proportion of discordant results seen in these patients. Further work is required to establish the precise relationship between CBG levels and GH.
Renal impairment will affect the rate of cortisol excretion in urine. Our method uses the midnight to morning increment in Cort/Cr rather than the morning Cort/Cr alone. Any reduction in GFR should affect both the midnight and morning cortisol excretion to approximately the same extent, minimizing the effect of GFR on the Cort/Cr increment. However, further assessment of the Cort/Cr increment method in patients with renal impairment would be needed before it could be recommended in this group of patients.
Unlike the ITT, this method has no medical contraindications and is suitable for the vast majority of patients in whom the HPA needs to be assessed. It is noninvasive and readily repeatable, making it particularly attractive in patients requiring annual pituitary assessment. The ITT and SST require hospital admission and trained clinical staff to perform the tests. For our method the urine collections are performed at home, avoiding the potential stress and cost of hospital admission. The Cort/Cr increment would not be suitable in patients with conditions that might interfere with the normal diurnal pattern of ACTH release, such as patients with depression or alcohol dependence or shift workers. It would be predicted that patients under significant stress or with a disturbed sleep pattern might be identified by abnormally high midnight Cort/Cr ratios.
Six patients performed urine collections in hospital on day 5 postoperatively and then at home 4–6 weeks later. Their HPA status as determined from the urinary increments was the same on both occasions, suggesting that the Cort/Cr increment can be reliably used as early as the fifth postoperative day. Interestingly, in five of the six patients the urine collections performed at home showed greater increments than those performed in a hospital. This would be in keeping with a degree of disturbance to the normal sleep pattern in hospital in-patients and would mean that the test would err on the side of safety. This suggests that the Cort/Cr increment could be used postoperatively as a guide to which patients require hydrocortisone cover on discharge. The test could then be repeated 3–4 weeks later at home when a normal sleep pattern has been resumed and before the standard postoperative endocrine assessment.
The Cort/Cr increment is a simple, noninvasive test that assesses the ability of the entire HPA axis to behave physiologically. It measures changes in urinary free cortisol, which are a more accurate reflection of dynamic cortisol responses than serum cortisol. It is reliable, reproducible, readily repeatable, and suitable for use in the vast majority of patients with pituitary disease. Our data show that the concordance between the Cort/Cr increment and the ITT is better than that between the SST and ITT. We propose that the Cort/Cr increment be used routinely in conjunction with conventional HPA testing to guide decisions on hydrocortisone replacement. With increasing clinical experience, the Cort/Cr increment may replace the ITT as the test of first choice.
Received March 22, 1999.
Revised June 4, 1999.
Accepted June 17, 1999.
|
References |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  C. d. F. Cunha, I. N. Silva, and F. L. Finch Early Adrenocortical Recovery after Glucocorticoid Therapy in Children with Leukemia J. Clin. Endocrinol. Metab., June 1, 2004; 89(6): 2797 - 2802. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. I. Dorin, C. R. Qualls, and L. M. Crapo Diagnosis of Adrenal Insufficiency Ann Intern Med, August 5, 2003; 139(3): 194 - 204. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. Remer, A. Neubert, and C. Maser-Gluth Anthropometry-based reference values for 24-h urinary creatinine excretion during growth and their use in endocrine and nutritional research Am. J. Clinical Nutrition, March 1, 2002; 75(3): 561 - 569. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
