J. Clin. Endocrinol. Metab. 2003 88: 4193-4198, doi: 10.1210/jc.2002-021897  

I. Lopez Schmidt, H. Lahner, K. Mann and S. Petersenn  

Test in Patients with Hypothalamic-Pituitary-Adrenal DiseaseHormone Test and Basal Serum Cortisol in Comparison to the Insulin Tolerance

Diagnosis of Adrenal Insufficiency: Evaluation of the Corticotropin-Releasing

Society please go to: http://jcem.endojournals.org//subscriptions/ or any of the other journals published by The EndocrineJournal of Clinical Endocrinology & Metabolism To subscribe to

Copyright © The Endocrine Society. All rights reserved. Print ISSN: 0021-972X. Online

Diagnosis of Adrenal Insufficiency: Evaluation of theCorticotropin-Releasing Hormone Test and Basal SerumCortisol in Comparison to the Insulin Tolerance Test inPatients with Hypothalamic-Pituitary-Adrenal Disease

I. LOPEZ SCHMIDT, H. LAHNER, K. MANN, AND S. PETERSENN

Division of Endocrinology, Medical Center, University of Essen, 45122 Essen, Germany

The aim of the study was to evaluate the diagnostic value ofthe human CRH test and the basal morning serum cortisol forthe diagnosis of adrenal insufficiency. Putative peak cortisolcut points for the CRH test and basal cortisol cut points weredetermined by receiver operating characteristic (ROC) anal-ysis with the insulin tolerance test as reference test. Fifty-fourpatients with suspected hypothalamic-pituitary-adrenal dis-ease were tested. In 20 healthy controls, CRH led to a meanpeak cortisol of 594.8 � 21.7 nmol/liter. The lower limit of anormal response was calculated as 400 nmol/liter. ROC anal-ysis of peak cortisol levels during CRH testing of patients withsuspected hypothalamic-pituitary-adrenal disease suggestedan optimal peak cortisol cut point of <377 nmol/liter for thediagnosis of adrenal insufficiency and a 96% specificity butpoor sensitivity of 76%. The baseline cortisol in the healthy

control group showed a mean of 439.3 � 24.9 nmol/liter, re-sulting in a lower limit of 267 nmol/liter. ROC analysis ofpatients suggested the highest accuracy for basal cortisol lev-els of 285 nmol/liter or more for the diagnosis of adrenal in-sufficiency (100% sensitivity and 61% specificity). Within thispatient group, a cortisol of more than 98 nmol/liter excludedadrenal insufficiency among those without the disorder,yielding 100% specificity. Using these criteria of upper (285nmol/liter) and lower (98 nmol/liter) cut-off points with highsensitivity and specificity can reduce the number of individ-uals who need provocative tests. Basal cortisol is less expen-sive, and we therefore suggest to use it as a first-line test ofadrenal insufficiency. Because of the low sensitivity of thehuman CRH test, we do not recommend it as a second test.(J Clin Endocrinol Metab 88: 4193–4198, 2003)

ACCURATE ASSESSMENT OF the hypothalamic-pitu-itary-adrenal (HPA) axis is essential in the manage-

ment of patients with suspected pituitary or hypothalamicdisease. The insulin tolerance test (ITT) is considered to bethe gold standard to evaluate adrenal function in such pa-tients (1). Peak cortisol cut points (PCCPs) between 500 and550 nmol/liter are variably used for the diagnosis of adrenalsufficiency (2). However, the ITT is contraindicated in pa-tients with cardio- or cerebrovascular diseases and convul-sive disorders. Furthermore, the test is unpleasant for thepatient and costly with respect to the degree of medicalsupervision required. As a consequence, alternative tests forevaluating HPA axis have been sought, such as glucagon,metyrapone, ACTH, and CRH testing.

The human CRH (hCRH) test may be an alternative test toexamine the HPA axis. CRH stimulates the secretion ofACTH at the level of the pituitary. The administration ofCRH is free of serious side effects (3). After iv injection of 100�g CRH, the plasma ACTH value usually peaks at 15 or 30min, whereas the cortisol value usually peaks between 30and 60 min (4). In patients treated with synthetic glucocor-ticoids, testing with hCRH is nearly as useful as the ITT (5).

Besides dynamic testing of the HPA axis, early-morningserum cortisol measurement has gained interest as a screen-

ing test. Several cut points have been proposed, leading tofurther dynamic testing or proving diagnosis (1, 6–8).

The aim of the present study was to evaluate the diagnosticvalue of the hCRH test and the early-morning serum cortisolmeasurement in comparison with the ITT and to define cutpoints for the diagnosis of adrenal insufficiency using a totalof 54 patients with suspected pituitary or hypothalamic dis-eases. Because there are only limited data for the peak cor-tisol values (hCRH test) in healthy subjects, we establishedreference values in a healthy control group. Furthermore, wedefined lower limits for baseline cortisol in our controlgroup.

Subjects and MethodsNormal volunteers

Twenty volunteers free of any endocrine disorder (Table 1) under-went hCRH administration, and 12 of them also underwent early-morn-ing cortisol sampling (Table 2). The local ethics committee approved thestudy protocol, and all subjects gave their informed written consent toparticipate in the study. None of the eight female normal volunteerswere taking estrogen at the time of the investigation.

Patients

Fifty-four patients were evaluated at our department because of sus-pected disease of the HPA axis from July 2001 until July 2002 (Table 1).Patients underwent testing by ITT and hCRH administration. Forty-sixpatients had a history of tumors in the pituitary area (20 nonfunctioningadenomas, eight somatotropic adenomas, six prolactinomas, two casesof neurosarcoidosis, three cases of hypophysitis, six craniopharyngeo-mas, and one chordoma), three patients had a history of pituitary hor-mone deficiency of unknown origin (two congenital, one diabetes in-

Abbreviations: AI, Adrenal insufficient; AS, adrenal sufficient; AUC,area under the curve; BMI, body mass index; hCRH, human CRH; HPA,hypothalamic-pituitary-adrenal; ITT, insulin tolerance test; PCCP, peakcortisol cut point; ROC, receiver operating characteristic.

0021-972X/03/$15.00/0 The Journal of Clinical Endocrinology & Metabolism 88(9):4193–4198Printed in U.S.A. Copyright © 2003 by The Endocrine Society

doi: 10.1210/jc.2002-021897

4193

sipidus), and in five patients disease of the HPA axis was suspectedbecause of clinical symptoms. Forty-one of the above-mentioned pa-tients were also examined by measurement of the morning serum cor-tisol (Table 2). Thirty-five had a history of tumors in the pituitary area(14 nonfunctioning adenomas, seven somatotropic adenomas, four pro-lactinomas, one case of neurosarcoidosis, two cases of hypophysitis, sixcraniopharyngeomas, and one chordoma), one patient had diabetesinsipidus, and in five cases disease was suspected because of clinicalsymptoms. At the time of investigation, none of the female patients wereon estrogen.

Methods

Patients underwent an ITT between 0900 and 1030 h by injection of0.15 IU/kg regular insulin (Actrapid, Novo Nordisk, Mainz, Germany) toachieve blood glucose levels less than 40 mg/dl and until symptoms ofhypoglycemia developed. Blood samples were taken at 0, 15, 30, 45, 60, 90,and 120 min. CRH testing in the patient and volunteer groups was per-formed between 0900 and 1030 h, using 100 �g hCRH (Ferring GmbH, Kiel,Germany) and an indwelling line. Blood samples were taken at 0, 15, 30, and45 min. Unstimulated serum cortisol values between 0800 and 0900 h wereavailable for comparison with the peak cortisol response to the dynamictesting. If the patients were initially treated with glucocorticoid replacementtherapy, it was stopped 24 h before hormonal evaluation. In addition, theother anterior pituitary axes (thyrotropic, gonadotropic, and somatotropic)were evaluated by hormone baseline levels and provocative testing, asrequired. Function of the posterior pituitary was assessed by determinationof 24-h urine volume and urine-osmolality. Further testing was performed,if required.

Serum cortisol levels (nanomoles per liter) were assayed at each timepoint by competitive immunoassay (ADVIA Centaur System, Bayer,Fernwald, Germany). The lower detection limit was assessed to be 5.5nmol/liter (0.20 �g/dl). Intraassay variations as coefficient of variationfor various cortisol values were 3.69% (107.05 nmol/liter), 3.09% (155.33nmol/liter), 2.89% (390.95 nmol/liter), 3.82% (759.55 nmol/liter), and2.98% (1024.97 nmol/liter). Interassay variations for the above-mentioned cortisol concentrations were 5.45, 3.83, 3.07, 1.86, and 3.99%.

Results (mean � sem) are expressed as absolute values for cortisol.GraphPad Prism 3.0 software for Macintosh (GraphPad Software, SanDiego, CA) was used for statistical analysis. Spearman’s rank correlationanalysis was carried out to determine relationships between variables.For further statistical analysis, a Mann-Whitney U test was performedwhere appropriate. Receiver operating characteristic (ROC) analysis wasobtained using MedCalc 6.16 software (MedCalc software, Mariakerke,Belgium).

ResultshCRH test

Normal volunteers. Twenty healthy volunteers underwenthCRH testing for determination of reference values. Follow-

ing hCRH administration, serum cortisol (mean � sem,range) rose from baseline to a mean peak of 594.8 � 21.7nmol/liter (448.0–843.0 nmol/liter) (Fig. 1A). The lower nor-mal cortisol peak value (mean peak, �2 sd) was calculatedas 400 nmol/liter. Neither age nor body mass index (BMI)had significant influence on the peak serum cortisol levelsduring hCRH testing.

The percentage increment (baseline equal to 100%) of se-rum cortisol during hCRH test showed a mean of 147.9 �8.4% (108.0–264.0%), as illustrated in Fig. 2.

Patients

Fifty-four patients with suspected disease of the HPA axiswere examined by hCRH test and ITT as reference test. Usinga PCCP of 500 nmol/liter in the ITT, 29 patients were con-sidered to be adrenal insufficient (AI), whereas 25 were suf-ficient (AS). The cortisol mean peak of the AI patients(mean � sem, range) was 233.8 � 31.1 nmol/liter (25.00–499.0 nmol/liter), compared with 616.5 � 25.7 nmol/liter(503.0–1071 nmol/liter) in AS patients (Fig. 1A). With regardto sex and BMI, there was no significant difference betweenboth groups (Table 1). However, age was significantly dif-ferent between the AI and AS group (P � 0.05). Deficienciesof other pituitary axes were also examined (Fig. 3). Amongthe AS patients, 15 had no pituitary axis defect at all, nine hadone insufficiency, and one patient had a defect in two pitu-itary axes. In contrast, among the AI patients, three had noinsufficiency of other axes, five had insufficiency of one axis,two patients had defect in two pituitary axes, 13 had insuf-ficiency of three axes, and six patients were insufficient infour other pituitary axes. Deficiencies of other pituitary axescorrelated with the degree of functional impairment inthe ITT.

There was a significant correlation for the cortisol peaklevels during ITT and hCRH testing (r � 0.80, P � 0.0001),as shown in Fig. 4A. The results of the hCRH testing weregrouped according to the classification by ITT peak cortisolvalues (Fig. 1). In AI patients, the cortisol mean peak (mean �sem, range) was 255.9 � 30.8 nmol/liter (25.00–514 nmol/liter). In contrast, AS patients showed a cortisol mean peakof 498.1 � 17.8 nmol/liter (354.0–749.0 nmol/liter). To bal-ance between high specificity and high sensitivity, ROC anal-

TABLE 1. Demographic data (mean � SEM) of volunteers and patients examined by hCRH test

Volunteers (n � 20) All patients (n � 54) AI patients (n � 29) AS patients (n � 25)

Sex M: 12 M: 27 M: 16 M: 11F: 8 F: 27 F: 13 F: 14

Age (yr) 33.3 � 2.7 46.1 � 2.0 49.8 � 2.5 40.2 � 3.0BMI (kg/m2) 24.4 � 0.7 27.8 � 0.7 29.0 � 1.1 26.4 � 0.9Additional pituitary hormone deficiencies

(number)None 1.5 � 0.2 2.5 � 0.2 0.4 � 0.1

TABLE 2. Demographic data (mean � SEM) of volunteers and patients examined by basal cortisol levels

Volunteers (n � 12) All patients (n � 41) AI patients (n � 20) AS patients (n � 21)

Sex M: 8 M: 19 M: 10 M: 8F: 4 F: 22 F: 10 F: 13

Age (yr) 35.8 � 3.9 46.6 � 2.5 50.2 � 3.4 41.3 � 3.3BMI (kg/m2) 24.6 � 0.9 27.2 � 0.8 28.4 � 1.2 27.1 � 0.9Additional pituitary hormone deficiencies

(number)None 1.4 � 0.2 2.5 � 0.3 0.5 � 0.1

4194 J Clin Endocrinol Metab, September 2003, 88(9):4193–4198 Lopez Schmidt et al. • CRH Testing in Comparison with the ITT

ysis was performed (Fig. 5), which suggested an optimalPCCP of 377 nmol/liter, resulting in 96% specificity and 76%sensitivity [area under the curve (AUC) � 0.89, 95% confi-dence interval � 0.78–0.96]. A cortisol cut point of 349 nmol/liter allowed for 100% specificity and 66% sensitivity,whereas a cut point of 514 nmol/liter determined 100% sen-sitivity but low specificity of 32%.

Percentage increment of the baseline cortisol has also beenused for analysis of the results of hCRH testing. In the AIsubjects, the mean cortisol increment (baseline equal to 100%)is 238.8 � 24.7% (100.0–660.0%), compared with 213.4 �18.9% (100.0–530.8%) in AS patients (Fig. 2). ROC analysissuggested an optimal cut-off value of 251.3%, with 41% sen-sitivity and 76% specificity (AUC � 0.54, 95% confidenceinterval 0.40–0.68).

We analyzed the time points of the cortisol peak during thehCRH test. No significant increment from baseline cortisolwas observed in seven subjects (13%). In two subjects (4%),the peak value occurred 15 min after hCRH administration,

whereas in eight subjects (15%), it was observed at time point30 min. In 37 patients (68%), the peak cortisol value was seen45 min after hCRH administration. We also performed ROCanalysis of cortisol values at defined time points during thehCRH test. Regarding a specificity of 96%, the correspondingvalues for sensitivity are 62% for time point 15 min (AUC �0.884), 65% for time point 30 min (AUC � 0.894), and 69% fortime point 45 min (AUC � 0.888).

Basal cortisol

Normal volunteers. Twelve morning cortisol values of the totalof 20 healthy volunteers were available. The mean basalcortisol value was 439.3 � 24.9 nmol/liter (326.0–600.0nmol/liter) (Fig. 1B). The lower limit of a normal basal cor-tisol was calculated as 267 nmol/liter (mean � 2 sd).

Patients. Because there were only 41 baseline cortisol valuesof the total of 54 patients available (Table 2), a separateanalysis of the corresponding results of the ITT was per-formed. Twenty patients were considered to be AI with amean cortisol peak (mean � sem, range) of 228.0 � 39.5nmol/liter (25.00–499.0 nmol/liter), compared with 21 ASpatients with a mean cortisol peak of 620.3 � 30.6 nmol/liter(503.0–1071 nmol/liter) (Fig. 1B). No significant differencesconcerning sex, age, or BMI could be observed between bothgroups (Table 2). Basal cortisol values were plotted againstthe number of pituitary deficiencies (Fig. 3B). On average,patients with intact pituitary axes had higher basal cortisollevels than patients with multiple pituitary hormonedeficiencies.

There was a significant correlation for the basal cortisolvalues and the peak levels during ITT (r � 0.72, P � 0.0001),

FIG. 1. Comparison of peak cortisol levels during hCRH test (A)morning basal cortisol levels (B) to peak cortisol levels during ITT.Individual values of healthy volunteers (N) during hCRH test (A) andfor basal cortisol (B) are shown by circles. Individual values duringITT (A�B, squares) and hCRH test (A, triangles) or basal cortisol (B,triangles) are demonstrated. Patients were defined as adrenal insuf-ficient (I, closed symbols) or sufficient (S, open symbols) based on theircortisol peak response to hypoglycemia of less than 500 nmol/liter ormore than 500 nmol/liter.

FIG. 2. Analysis of percentage increment of serum cortisol (baselinevalues equal to 100%). Individual values of maximum percentagecortisol increment during hCRH test are shown for a healthy controlgroup by circles. Individual values of maximum percentage cortisolincrement during ITT (squares) and hCRH test (triangles) are dem-onstrated. Classification of patients as adrenal insufficient (I, closedsymbols) or sufficient (S, open symbols) was based on their cortisolpeak response to hypoglycemia of less than 500 nmol/liter or morethan 500 nmol/liter.

Lopez Schmidt et al. • CRH Testing in Comparison with the ITT J Clin Endocrinol Metab, September 2003, 88(9):4193–4198 4195

as shown in Fig. 4B. The results of the baseline cortisol valueswere grouped according to the classification by ITT peakcortisol values (Fig. 1B). The mean baseline cortisol level forthe AI patients was 126.5 � 21.8 nmol/liter (25.00–285.0nmol/liter). In contrast, AS patients showed mean morningcortisol levels of 357.1 � 36.9 nmol/liter (119.0–654.0 nmol/

liter). ROC analysis suggested an optimal baseline cortisolcut point of 285 nmol/liter (AUC � 0.88, confidence interval0.74–0.96), resulting in 100% sensitivity and 62% specificity.A cortisol cut point of 98 nmol/liter determined 100% spec-ificity but low sensitivity of 50% (Fig. 5B).

Combination of basal cortisol and hCRH test

The combination of the basal cortisol measurement andsubsequent hCRH testing was analyzed. Using a lower cut-off value of 98 nmol/liter or less for basal cortisol, 10 patientscould be classified as AI, whereas an upper cut point of morethan 285 nmol/liter permitted the diagnosis of AS in 13patients. In 18 patients of the total of 41 patients (44%), thediagnosis still remained unclear. On the basis of the subse-quent hCRH test, five patients were considered to be AI(�349 nmol/liter) and two patients to be AS (�514 nmol/liter). As a consequence, diagnosis obtained by the ITT couldnot be confirmed in 11 patients (27%).

Discussion

The diagnosis of secondary adrenal insufficiency is diffi-cult. Considerable controversy exists about which test is best,especially with regard to the low-dose and high-dose ACTHstimulation test (9). In the present study, we compared thehCRH test and morning basal cortisol levels with the ITTwith regard to their value for diagnosis of adrenal insuffi-ciency in a cohort of patients with suspected impaired HPAaxis. The ITT was used as reference test because it is still beconsidered the gold standard for diagnosis of adrenal insuf-ficiency (1). The PCCP for separation between normal andsubnormal adrenal response to hypoglycemia has been vari-ably defined between 500 and 550 nmol/liter (2). A recentinvestigation of subjects without endocrine disease (10) sug-gested a cut point of 500 nmol/liter, which we relied on.

Our investigation of 20 healthy volunteers with hCRHrevealed mean cortisol peak levels very similar to the resultsobtained by Schlaghecke et al. (5). In contrast, higher peak

FIG. 3. A, Percentage of additional pituitary axis insufficiencies inrelation to adrenal function. Patients were separated as AS (openbars) and AI (closed bars) according to a PCCP of 500 nmol/liter in theITT. Bars indicate the proportion of patients of each group sufferingfrom zero to four pituitary axis deficiencies. B, Basal cortisol valuesplotted against the number of pituitary insufficiencies.

FIG. 4. Individual cortisol peak levels during ITT in 54 patients (A) and 41 patients (B) plotted against cortisol peaks during hCRH test (A)(r � 0.80, P � 0.0001) and morning basal cortisol (B) (r � 0.72, P � 0.0001).

4196 J Clin Endocrinol Metab, September 2003, 88(9):4193–4198 Lopez Schmidt et al. • CRH Testing in Comparison with the ITT

levels in normal subjects have been reported in response to100 �g ovine CRH administration (11). Ovine CRH was pre-viously found to cause greater peak cortisol and ACTH lev-els, compared with hCRH (12). It is known to result in pro-longed elevation of ACTH and cortisol (13), possibly becauseof its prolonged circulating half-life (14). To our knowledge,a lower normal limit for serum cortisol during the hCRH testby calculation of mean peak cortisol levels to 2 sd, corre-sponding to 400 nmol/liter in our study, has not been de-fined so far.

In the hCRH test of examined patients, peak cortisol levelsof 377 nmol/liter or less reflected adrenal insufficiency veryaccurately (96% specificity), whereas cortisol levels above377 nmol/liter did not necessarily determine adrenal suffi-ciency (76% sensitivity). The PCCP is in good agreement withthe above-mentioned lower normal peak value of the controlgroup. However, 24% of patients will be misdiagnosed as ASusing this cut point. In a previous study of the hCRH test,Dullaart et al. (15) studied a cohort of 80 patients with hy-pothalamic-pituitary disorders but without normal volun-teers by ROC analysis. Peak cortisol values of 420 nmol/literreflected 100% specificity (349 nmol/liter in our study). A100% sensitivity was reached by a PCCP of 615 nmol/liter,compared with 514 nmol/liter in our study. Higher cortisolvalues obtained by Dullaart et al. may be explained by earliertime of testing (0800 to 0900 h).

In our study, the highest cortisol levels were observed 45min after hCRH administration in 37 of 47 patients (79%). Inonly two subjects (4%), peak levels occurred at time point 15min. In agreement, other investigators reported cortisol peaklevels between 30 and 60 min after hCRH administration (4,5). ROC analysis of different time points revealed a sensi-tivity of 69% for the 45-min value, compared with 76% sen-sitivity for the peak levels (same specificity of 96%). There-fore, sampling may be limited to the 45-min time point.

Schlaghecke et al. (5) defined an 1.5-fold increase in theplasma cortisol concentration to a value of at least 276 nmol/liter during ITT or hCRH test as a normal response. Fordifferential diagnosis of Cushing’s syndrome, percentagevariations between baseline and stimulated cortisol by CRHadministration has been established (16). However, based onthe percentage cortisol increment during our patient’s hCRHtest, ROC analysis revealed an AUC value of 0.54, indicatingnearly no discrimination between AS and AI patients. There-fore, the percentage increment of baseline cortisol is inap-propriate to analyze the results of the hCRH test with regardto AS. Similarly, Erturk et al. (6) suggested that the use of anincremental rise in plasma cortisol during ITT as a criterionfor HPA normalcy should be abolished.

In our study, morning basal cortisol values were assessedin a total of 41 patients in comparison with the ITT. Theoptimal cut point of 285 nmol/liter (100% sensitivity and 62%specificity) defined by ROC analysis is in good agreementwith the normal value obtained in our volunteer group (267nmol/liter), although the size of the volunteers group issmall. Sensitivity and specificity were clearly different forbasal cortisol and the hCRH test, suggesting specific diag-nostic utility. In contrast, the AUCs of both tests overlappedtremendously, indicating that the tests are the same by ROC.Therefore, a simple comparison of AUC values represents apotential pitfall of ROC determination.

The lower cut point for basal cortisol, proving adrenalinsufficiency, was determined as 98 nmol/liter (100% spec-ificity and 50% sensitivity). That value corresponds to otherproposed lower cut points for cortisol, ranging between 80and 110 nmol/liter (1, 6–8). The upper cortisol cut point toconfirm adrenal sufficiency is less well standardized. Cutpoints of 250 nmol/liter (1) and 300 nmol/liter (7) have beenreported, which are in good agreement with the result of ourstudy (285 nmol/liter). However, two other studies demon-strated significantly higher cut points of 470 nmol/liter (6)and 500 nmol/liter (8). The differences may depend on thetime point of blood withdrawal, the technique (venipunctureor collection from a catheter), and the conditions of bloodsampling (outpatient or inpatient).

The number of pituitary axis deficiencies and basal cortisollevels were inversely correlated. Furthermore, deficiencies ofother pituitary axes correlated clearly with the degree offunctional impairment in the ITT. Such dependency has beenpreviously demonstrated for the severity of GH deficiencyand the degree of hypopituitarism (17).

Our results suggest the use of an algorithm for the diag-nosis of AI. Use of a lower cut point of 98 nmol/liter (100%specificity) and a higher cut point of 285 nmol/liter (100%sensitivity) for basal cortisol levels identified 23 of the pa-tients (56%) correctly, as classified by ITT. The remaining 18

FIG. 5. ROC curve of the hCRH test (A) and the morning basal cor-tisol (B) with the ITT as reference test. Optimal cut points are givenin bold letters. Peak cortisol levels to the ITT more than 500 nmol/literindicate corticotropic sufficiency.

Lopez Schmidt et al. • CRH Testing in Comparison with the ITT J Clin Endocrinol Metab, September 2003, 88(9):4193–4198 4197

patients, who fell between these cut points and could not beidentified as sufficient or insufficient, would require furthertesting. However, the hCRH test did not confirm the diag-nosis obtained by ITT in 11 of these patients (27% of the totalgroup).

In conclusion, the measurement of early morning serumcortisol reduces the numbers of individuals who need pro-vocative tests. Adequate upper and lower cut points withhigh sensitivity and specificity, respectively, need to be ap-plied. Basal cortisol is less expensive, and we therefore sug-gest to use it as a first-line test of adrenal insufficiency.Because of the low sensitivity of the hCRH test, we do notrecommend it as a second test.

Acknowledgments

Received December 3, 2002. Accepted May 28, 2003.Address all correspondence and requests for reprints to: Priv.-Doz.

Dr. med. Stephan Petersenn, Division of Endocrinology, Medical Center,University of Essen, Hufelandstr. 55, 45122 Essen, Germany. E-mail:stephan.petersenn@uni-essen.de.

References

1. Watts NB, Tindall GT 1988 Rapid assessment of corticotropin reserve afterpituitary surgery. JAMA 259:708–711

2. Oelkers W 1996 Adrenal insufficiency. N Engl J Med 335:1206–12123. Orth DN 1992 Corticotropin-releasing hormone in humans. Endocr Rev 13:

164–1914. Trainer PJ, Faria M, Newell Price J, Browne P, Kopelman P, Coy DH, Besser

GM, Grossman AB 1995 A comparison of the effects of human and ovinecorticotropin-releasing hormone on the pituitary-adrenal axis. J Clin Endo-crinol Metab 80:412–417

5. Schlaghecke R, Kornely E, Santen RT, Ridderskamp P 1992 The effect oflong-term glucocorticoid therapy on pituitary-adrenal responses to exogenouscorticotropin-releasing hormone. N Engl J Med 326:226–230

6. Erturk E, Jaffe CA, Barkan AL 1998 Evaluation of the integrity of the hypo-

thalamic-pituitary-adrenal axis by insulin hypoglycemia test. J Clin EndocrinolMetab 83:2350–2354

7. Hagg E, Asplund K, Lithner F 1987 Value of basal plasma cortisol assays inthe assessment of pituitary-adrenal insufficiency. Clin Endocrinol (Oxf) 26:221–226

8. Jones SL, Trainer PJ, Perry L, Wass JA, Bessser GM, Grossman A 1994 Anaudit of the insulin tolerance test in adult subjects in an acute investigation unitover one year. Clin Endocrinol (Oxf) 41:123–128

9. Stewart PM, Clark P 1999 The short Synacthen test: is less best? Clin Endo-crinol (Oxf) 51:151–152

10. Tuchelt H, Dekker K, Bahr V, Oelkers W 2000 Dose-response relationshipbetween plasma ACTH and serum cortisol in the insulin-hypoglycaemia testin 25 healthy subjects and 109 patients with pituitary disease. Clin Endocrinol(Oxf) 53:301–307

11. Lytras N, Grossman A, Perry L, Tomlin S, Wass JA, Coy DH, Schally AV,Rees LH, Besser GM 1984 Corticotrophin releasing factor: responses in normalsubjects and patients with disorders of the hypothalamus and pituitary. ClinEndocrinol (Oxf) 20:71–84

12. Schurmeyer TH, Avgerinos PC, Gold PW, Gallucci WT, Tomai TP, Cutler JrGB, Loriaux DL, Chrousos GP 1984 Human corticotropin-releasing factor inman: pharmacokinetic properties and dose-response of plasma adrenocorti-cotropin and cortisol secretion. J Clin Endocrinol Metab 59:1103–1108

13. DeBold CR, DeCherney GS, Jackson RV, Sheldon WR, Alexander AN,Island DP, Rivier J, Vale W, Orth DN 1983 Effect of synthetic ovine corti-cotropin-releasing factor: prolonged duration of action and biphasic responseof plasma adrenocorticotropin and cortisol. J Clin Endocrinol Metab 57:294–298

14. Nicholson WE, DeCherney GS, Jackson RV, DeBold CR, Uderman H, Al-exander AN, Rivier J, Vale W, Orth DN 1983 Plasma distribution, disappear-ance half-time, metabolic clearance rate, and degradation of synthetic ovinecorticotropin-releasing factor in man. J Clin Endocrinol Metab 57:1263–1269

15. Dullaart RP, Pasterkamp SH, Beentjes JA, Sluiter WJ 1999 Evaluation ofadrenal function in patients with hypothalamic and pituitary disorders: com-parison of serum cortisol, urinary free cortisol and the human-corticotrophinreleasing hormone test with the insulin tolerance test. Clin Endocrinol (Oxf)50:465–471

16. Nieman LK, Oldfield EH, Wesley R, Chrousos GP, Loriaux DL, Cutler Jr GB1993 A simplified morning ovine corticotropin-releasing hormone stimulationtest for the differential diagnosis of adrenocorticotropin-dependent Cushing’ssyndrome. J Clin Endocrinol Metab 77:1308–1312

17. Toogood AA, Beardwell CG, Shalet SM 1994 The severity of growth hormonedeficiency in adults with pituitary disease is related to the degree of hypo-pituitarism. Clin Endocrinol (Oxf) 41:511–516

4198 J Clin Endocrinol Metab, September 2003, 88(9):4193–4198 Lopez Schmidt et al. • CRH Testing in Comparison with the ITT