0021-972X/03/$15.00/0
Printed in U.S.A.
The Journal of Clinical Endocrinology & Metabolism 88(11):5293–5298
Copyright © 2003 by The Endocrine Society
doi: 10.1210/jc.2003-030449
The Value of Dehydroepiandrosterone Sulfate
Measurements in the Assessment of Adrenal Function
MONA P. NASRALLAH
AND
BAHA M. ARAFAH
Division of Clinical and Molecular Endocrinology, University Hospitals of Cleveland, and Case Western Reserve University,
Cleveland, Ohio 44106
Dehydroepiandrosterone (DHEA) and its sulfated ester
(DHEA-S) are corticotropin-dependent adrenal androgen precursors that are uniformly low in treated patients with corticotropin deficiency. There are no data investigating the diagnostic value of DHEA-S measurements in the prospective
assessment of adrenal function. This study examined serum
DHEA-S levels as possible markers for hypothalamicpituitary-adrenal (HPA) function in patients with large pituitary adenomas.
Patients were characterized to have normal HPA (n ⴝ 47) or
abnormal HPA (ABN-HPA, n ⴝ 35) function based on their
respective responses to insulin-induced hypoglycemia. Patients also underwent low-dose Cortrosyn (1 ␮g, LDC) and
standard-dose Cortrosyn stimulation testing.
All patients with ABN-HPA had very low age- and gender-
T
HE INTEGRITY OF the hypothalamic-pituitary-adrenal
(HPA) axis is important in maintaining many physiological functions, and a critical component in the response to
stress. Whereas the secretion of cortisol and adrenal androgen precursors is corticotropin-dependent, the dominant
regulatory mechanism for aldosterone is the renin-angiotensin system. Intrinsic adrenal diseases lead to loss of all three
classes of adrenal steroids (glucocorticoids, mineralocorticoids, and androgens) and result in the clinical entity of
primary adrenal insufficiency. In contrast, central adrenal
insufficiency is caused by impairment of CRH or corticotropin secretion and leads to diminished production of their
adrenal-dependent steroid hormones, namely glucocorticoids and adrenal androgens. In this instance, mineralocorticoid secretion is relatively spared.
The diagnosis of central adrenal insufficiency continues to
be difficult, particularly when the deficiency is partial.
Though insulin-induced hypoglycemia (IIH) is considered to
be the gold standard test, it is commonly associated with
discomfort, requires close monitoring, and has limited use in
children, the elderly, and patients with heart disease (1, 2).
The standard-dose Cortrosyn (SDC) test was introduced as
an alternative test for establishing the diagnosis of primary
(and subsequently, that of central) adrenal insufficiency.
However, several reports demonstrating normal cortisol responses to the SDC, in patients with documented central
adrenal insufficiency, have raised concerns about the test’s
Abbreviations: ABN-HPA, Abnormal HPA; CI, confidence interval;
DHEA, dehydroepiandrosterone; DHEA-S, DHEA sulfate; HPA, hypothalamic-pituitary-adrenal; IIH, insulin-induced hypoglycemia; LDC,
low-dose Cortrosyn; NL-HPA, normal HPA; PRL, prolactin; ROC, receiver operating characteristic; SDC, standard-dose Cortrosyn.
matched serum DHEA-S levels. When the normal response to
LDC was set at a cortisol level of at least 18.1 ␮g/dl, 10 of 31
patients with ABN-HPA exhibited normal responses. Receiver
operating characteristic curves for baseline DHEA-S and for
maximal cortisol responses to LDC had areas of 0.984 (confidence interval, 0.962–1.000) and 0.893 (confidence interval,
0.817– 0.969), respectively.
LDC- or SDC-stimulated serum cortisol levels have significant limitations in defining HPA function. A normal age- and
gender-specific serum DHEA-S level makes the diagnosis of
corticotropin deficiency extremely unlikely. However, when
serum DHEA-S levels are low, further testing is necessary to
define HPA function. (J Clin Endocrinol Metab 88: 5293–5298,
2003)
validity in this condition (3, 4). This led to the introduction
of the low-dose Cortrosyn test (LDC), where 1 ␮g of Cortrosyn (instead of 250 ␮g) is given as the test dose (4 – 8). The
latter dose of Cortrosyn mimics more closely the maximum
endogenous corticotropin stress response generated by the
activation of the hypothalamic-pituitary unit. The sensitivity
and specificity of the LDC test depend on the cortisol cutoff
value used to define a normal response (5, 6), although they
remain less than that of IIH in establishing the diagnosis of
partial central adrenal insufficiency (1, 2, 6 – 8).
In the present investigation, we sought an alternative approach to assess the integrity of the HPA axis. Dehydroepiandrosterone (DHEA) and its sulfated ester (DHEA-S) are
adrenal androgen precursors secreted by the zona reticularis,
under the dominant regulation of corticotropin (9). Serum
levels of DHEA-S are affected by several factors, the most
important of which are age, gender, chronic illness, and the
prior use of glucocorticoids (9). Earlier reports indicate that,
in treated hypopituitary patients, serum DHEA-S levels are
uniformly very low (10 –13). However, data on the use of
DHEA-S in the assessment of adrenal function are limited.
We have conducted a prospective study to investigate the
value of serum DHEA-S measurements in assessing HPA
function. The study involved patients at risk for having central adrenal insufficiency; namely those with large pituitary
tumors. We postulated that a normal DHEA-S serum level is
a reliable indicator of an intact HPA axis. The study included
patients with pituitary tumors who had never previously
received glucocorticoids, as well as normal healthy volunteers. Our data are consistent with the hypothesis that a
normal DHEA-S serum level makes the diagnosis of central
adrenal insufficiency extremely unlikely. However, because
5293
5294
J Clin Endocrinol Metab, November 2003, 88(11):5293–5298
serum DHEA-S levels can be lowered by other factors, a
low level does not necessarily indicate impaired adrenal
function.
Subjects and Methods
Subjects
Consecutive adult patients with newly diagnosed large (⬎1.5 cm)
pituitary adenomas were included in the study. All patients gave informed consent and underwent testing of pituitary function as previously described (14, 15). Only data pertinent to the assessment of HPA
axis function are presented here. Patients with Cushing’s disease were
excluded. None of the patients had received glucocorticoids, and none
of the women were receiving estrogen therapy. Tumors were classified
on the basis of the dominant cell type on immunohistochemical staining
of multiple sections of the resected adenoma. Patients with prolactin
(PRL)-secreting adenomas who did not have surgery were classified as
such on the basis of very high (⬎300 ␮g/liter) serum PRL levels.
The study population comprised the following three groups: 1)
Group I consisted of 35 patients who had abnormally low HPA (ABNHPA) function as defined by either a serum cortisol response of less than
18.5 ␮g/dl (510 nm) during IIH and/or a morning cortisol less than 3
␮g/dl (83 nm). 2) Group II consisted of 47 patients with pituitary macroadenomas, whose HPA function was documented to be normal (NLHPA), as defined by a serum cortisol level of more than 18.5 ␮g/dl (510
nm) in the basal state or in response to IIH. 3) Group III consisted of 21
healthy volunteers, comparable in age and gender with the two groups.
After testing, patients with ABN-HPA were started on hydrocortisone replacement therapy that was withdrawn immediately after surgical removal of the tumor as described previously (14, 15). Patients with
NL-HPA underwent surgical removal of the adenoma without any
glucocorticoid therapy before, during, or after the procedure as described previously (14, 15). The safety of this approach has been demonstrated repeatedly since the original report was published (16). After
surgery, patients with normal preoperative HPA function are closely
monitored, and normal adrenal function is documented before
discharge.
Nasrallah and Arafah • DHEA-S and Adrenal Function
SDC. Serum cortisol and DHEA-S levels were measured before and 30,
45, and 60 min after the iv injection of 250 ␮g Cortrosyn.
LDC. One milliliter of normal saline was drawn from a 250-ml bag,
mixed with 250 ␮g Cortrosyn, and reinjected into the 249-ml bag. After
mixing, 1 ml (equivalent to 1 ␮g) was drawn from the bag and injected
iv into patients of groups I and II. Measurements were obtained in a
manner similar to that of the SDC.
We assessed the sensitivity of each test at two different cutoff levels
of cortisol that are frequently reported in the literature (⬎18.1 and 19.9
␮g/dl, corresponding to 500 and 550 nm, respectively), as indicative of
adequate adrenal function for either LDC or SDC (6 –9).
Laboratory methods
Serum DHEA-S levels were measured by RIA, using a Coat-A-Count
kit (Diagnostic Products Corp. DPC, Los Angeles, CA). The lower limit
of detectability was 1.1 ␮g/dl, and the inter- and the intraassay coefficients of variation were 9.7% and 4.4%, respectively. Serum cortisol
levels were measured by immunoassay using a chemiluminescence technique (Centaur/Bayer Diagnostics, Tarrytown, NY). The lower limit of
detectability was 0.2 ␮g/dl, and the inter- and the intraassay coefficients
of variation were 8.9% and 4.3%, respectively.
Statistical analysis
Baseline demographic and clinical variables were compared between
groups, using Kruskal-Wallis followed by Mann-Whitney U tests for
continuous outcomes and ␹2/exact Fisher tests for categorical outcomes.
Statistical differences in serum cortisol and DHEA-S levels among all
groups were determined using Kruskal-Wallis, followed by Mann-Whitney U test in between two groups. In addition, receiver operating characteristic (ROC) curves were plotted for both cortisol and DHEA-S
serum levels, using IIH as the gold standard to distinguish between
NL-HPA and ABN-HPA groups. Because serum DHEA-S levels are ageand gender-dependent, the respective values in the groups were further
analyzed according to subgroups determined by gender and age intervals of 20 yr. The data are presented as mean ⫾ sd unless stated
otherwise.
Results
Study protocol
Testing was performed preoperatively, in an ambulatory setting, on
different days, in a random order for the same subject, mostly in the
morning. Healthy volunteers were tested mostly in the early afternoon.
IIH. Variable doses of insulin (0.05– 0.3 U/kg) were injected iv to induce
hypoglycemia (glucose ⬍ 40 mg/dl or 2.22 mm). Plasma corticotropin
and cortisol measurements were obtained before and every 15 min, for
120 min, as described previously (12, 13). The highest serum cortisol
level during the test was considered the maximal adrenal response. The
test was performed in both groups of patients with pituitary tumors. A
cortisol response of more than 18.5 ␮g/dl (510 nm) was considered
normal (15, 17).
Whereas there were no differences among the three groups
with respect to gender (Table 1), mean age was higher in
group I patients when compared with group II but not when
compared with the volunteers. Tumor type distribution was
different for gonadotrophs and GH-secreting tumors but not
for prolactinomas or nonsecreting adenomas. All patients in
group I had hypoadrenalism (by definition) and hypogonadism, and 25 of 35 had hypothyroidism. In contrast, only
28 of 47 and five of 47 patients of group II had hypogonadism
and hypothyroidism, respectively.
TABLE 1. Characteristics of the three groups
Pa
Characteristics
Gender (F/M)
Age (yr) (mean ⫾
Tumor type
FSH/LH
GH
PRL
NS
SD)
Pb
ABN-HPA (n ⫽ 35)
NL-HPA (n ⫽ 47)
VOL (n ⫽ 21)
ALL
ABN-HPA
vs. NL-HPA
ABN-HPA
vs. VOL
NL-HPA
vs. VOL
15/20
49 ⫾ 15.8
29/18
41 ⫾ 14.2
14/7
43 ⫾ 6.7
0.172
0.040
0.12
0.019
N/A
0.148
N/A
0.205
10
1
6
18
2
10
9
26
N/A
N/A
N/A
N/A
0.003
0.02
0.97
0.63
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
Groups were compared using ␹2 or exact Fisher test for gender and tumor type. FSH/LH, Gonadotroph; NS, nonsecreting; VOL, healthy
volunteers; N/A, not applicable.
a
Kruskal-Wallis.
b
Mann-Whitney U test for between two groups.
Nasrallah and Arafah • DHEA-S and Adrenal Function
J Clin Endocrinol Metab, November 2003, 88(11):5293–5298 5295
Serum cortisol levels
Maximal serum cortisol levels during IIH were higher (P ⬍
0.001) in patients with NL-HPA (26.9 ⫾ 6.9 ␮g/dl, or 742.2 ⫾
190.4 nm) than in those with ABN-HPA (10.3 ⫾ 6.1 ␮g/dl, or
284.2 ⫾ 168.3 nm). As shown in Table 2, patients with ABNHPA had lower mean basal and Cortrosyn-stimulated (SDC
and LDC) serum cortisol concentrations, compared with the
respective values in those with NL-HPA. Maximal serum
cortisol responses to SDC in patients with NL-HPA were
similar to those of healthy volunteers. Baseline serum cortisol
levels in healthy volunteers were lower than those of patients
with NL-HPA, likely because the former group was tested in
the early afternoon (Table 2). Although baseline serum cortisol is known to vary with time of the day, the Cortrosynstimulated levels do not vary with changes in the time of
testing (17).
The extent of the overlap in serum cortisol responses to
LDC between the two groups can be appreciated in Fig. 1.
Using two different cutoff values to define normal serum
cortisol response to either the LDC or SDC tests (18.1 or 19.9
␮g/dl corresponding to 500 and 550 nm, respectively), the
number of individual responses within each group is presented in Table 3. The data indicate that, even when one uses
a higher cutoff value of 19.9 ␮g/dl (550 nm) for either the SDC
or LDC test, a significant portion of patients (41% and 10%,
respectively) with documented central adrenal insufficiency,
as defined herein, would have been incorrectly characterized
as having NL-HPA.
ABN-HPA, as defined herein, but whose maximal cortisol
level during LDC was above that level. There were 10 such
patients (Table 3), all of whom had low gender- and agematched serum DHEA-S levels. Similarly, there were four of
39 patients with NL-HPA whose LDC stimulated serum cortisol levels were no more than 18.1 ␮g/dl (500 nm), and all
had normal serum DHEA-S concentrations.
The impact of gonadal dysfunction on serum DHEA-S
levels was examined in the group of patients with NL-HPA.
The data indicate that six of 27 patients with hypogonadism,
compared with three of 20 with normal gonadal function,
had low serum DHEA-S levels (P ⫽ 0.605). As a group (n ⫽
47), serum DHEA-S levels in hypogonadal patients (120 ⫾ 40
␮g/dl, or 3.26 ⫾ 1.1 ␮m, n ⫽ 27) were similar to those with
normal gonadal function (133 ⫾ 55 ␮g/dl, or 3.61 ⫾ 1.49 ␮m,
n ⫽ 20). Similar findings were noted when the data on the
18 men with NL-HPA were examined separately. Specifically, four of 12 men with hypogonadism and two of six men
with normal gonadal function had low serum DHEA-S levels
(P ⫽ 1.00). Mean serum DHEA-S levels in the latter two
subgroups of men were similar (121.8 ⫾ 54 ␮g/dl, or 3.31 ⫾
1.47 ␮m vs. 115.6 ⫾ 66 ␮g/dl, or 3.14 ⫾ 1.79 ␮m, respectively).
Serum DHEA-S levels
Patients with ABN-HPA had drastically reduced basal and
stimulated serum DHEA-S concentrations (Table 2). Patients
with NL-HPA function had mean basal and stimulated serum DHEA-S levels that were similar to those of healthy
volunteers (Table 2). Individual and age- and gendermatched serum DHEA-S levels are presented in Fig. 2, in
relation to the 5th percentile of our laboratory reference data.
Whereas all 33 patients with ABN-HPA had clearly low ageand gender-matched serum DHEA-S concentrations, nine of
47 patients with NL-HPA had levels below the lower limits
of their respective normal ranges (Table 4). Whereas two of
those values were only marginally decreased, the remaining
seven values were moderately low (Fig. 2).
An important point of the study was to address the subgroup of patients whose maximum cortisol levels after LDC
would be difficult to interpret clinically, i.e. close to the cutoff
point. Based on literature reports, we arbitrarily chose this
cutoff to be 18.1 ␮g/dl (500 nm) and evaluated patients with
FIG. 1. Individual LDC-stimulated serum cortisol levels are presented in patients with NL-HPA (circles), and in those with ABN-HPA
(squares). The shaded area represents the overlap in the results in the
two groups at two specific cutoff points reported in the literature to
represent normal responses (18.1 and 19.9 ␮g/dl). To convert levels
from ␮g/dl to nM, multiply the value by 27.59.
TABLE 2. Basal and maximal cortisol and DHEA-S serum concentrations (mean ⫾
Cortisol (␮g/dl)
DHEA-S (␮g/dl)
Basal
MAX LDC
MAX SDC
Basal level
SD)
ABN-HPA
NL-HPA
VOL
4.1 ⫾ 3.5
14.1 ⫾ 6.2
17.3 ⫾ 8.5
17.5 ⫾ 13.7
13.4 ⫾ 6.0
23.0 ⫾ 5.5
30.8 ⫾ 6.6
124.8 ⫾ 63.3
8.1 ⫾ 4.9
N/A
27.5 ⫾ 5.6
123.3 ⫾ 78.8
during SDC and LDC in the three groups
Pa
ABN vs. NL
ABN vs. VOL
NL vs. VOL
⬍0.001
⬍0.001
⬍0.001
⬍0.001
⬍0.001
N/A
⬍0.001
⬍0.001
0.001
N/A
0.052
0.490
MAX, Maximal response. To convert serum cortisol levels from ␮g/dl to nM, multiply the value by 27.59. To convert serum DHEA-S from
␮g/dl to ␮M, multiply the value by 0.02714.
a
Mann-Whitney U test.
5296
J Clin Endocrinol Metab, November 2003, 88(11):5293–5298
Nasrallah and Arafah • DHEA-S and Adrenal Function
TABLE 3. Number of patients with maximal cortisol responses of more than 18.1 or more than 19.9 ␮g/dl in the SDC and LDC tests
LDC
Max cortisol ␮g/dl (nM)
ABN-HPA (%)
NL-HPA (%)
VOL (%)
⬎18.1 (500)
10/31 (32)
35/39 (90)
N/A
N/A
SDC
⬎19.9 (550)
3/31 (10)
28/39 (72)
N/A
N/A
⬎18.1 (500)
19/34 (56)
43/43 (100)
20/21 (95)
⬎19.9 (550)
14/34 (41)
41/43 (95)
19/21 (90)
FIG. 2. Individual baseline serum DHEA-S levels are presented in patients with NL-HPA (circles), and in those with
ABN-HPA (triangles), divided into eight subgroups based
on age and gender. The horizontal bars indicate the 5th
percentile of our laboratory reference range for the respective age- and gender-adjusted serum DHEA-S concentrations. To convert levels from ␮g/dl to ␮M, multiply the value
by 0.02714.
ROC curve for DHEA-S and cortisol
Using results from LDC, maximal cortisol and baseline
DHEA-S levels were plotted in an ROC curve (Fig. 3). The
area under the curve for maximal cortisol was 0.893 [confidence interval (CI), 0.817– 0.969], reaching 100% sensitivity
when the cortisol level is above 24.4 ␮g/dl, and 100% specificity when the level is less than 16.9 ␮g/dl, as defined by
IIH. Using DHEA-S values, even unadjusted for age and
gender, the area under the curve was 0.984 (CI, 0.962–1.000),
with near-100% sensitivity when the absolute DHEA-S value
is above 53.5 ␮g/dl, and 100% specificity when the value is
less than 14.0 ␮g/dl, as defined by IIH (Fig. 3). The area
under the curve for the serum DHEA-S levels was significantly different (P ⫽ 0.031) from that of LDC-stimulated
serum cortisol concentrations. Similarly, the baseline
DHEA-S and maximal cortisol, using the SDC test, were
plotted for all three groups. The area under the curve for
cortisol was 0.894 (CI, 0.820 – 0.968) and for DHEA-S was
0.986 (CI, 0.969 –1.000).
Discussion
The data demonstrate that serum DHEA-S levels are uniformly low in patients with newly diagnosed corticotropin
deficiency. Moreover, only a small percentage of patients
with normal corticotropin secretion had borderline low serum DHEA-S. Therefore, the presence of normal levels
would argue strongly against adrenal insufficiency, and
would provide useful support especially in clinical situations
where the Cortrosyn stimulation test results are borderline.
Physiologically, the use of Cortrosyn stimulation in estab-
lishing the diagnosis of central adrenal insufficiency was
based on the assumption that corticotropin acts as a trophic
factor for maintenance of the adrenal cortex (1). Without
corticotropin, the adrenal cortex atrophies and would not be
expected to respond promptly to exogenously administered
Cortrosyn. However, because corticotropin deficiency is
most often partial rather than complete, some patients may
respond well to exogenous Cortrosyn, even though their
corticotropin secretion is subnormal. The normal cortisol
response seen with the high-dose Cortrosyn stimulation test
in 50% of adrenally insufficient patients is, in part, caused by
the use of supraphysiologic doses of corticotropin. Peak
plasma corticotropin levels, after iv administration of 250 ␮g
Cortrosyn (i.e. SDC), are reported to reach 60 ␮g/liter, or 13.2
nm (17). In contrast, plasma corticotropin levels reach approximately 100 ng/liter (22 pm during the IIH. It is for this
reason that the LDC test is less likely to give falsely normal
values than the SDC test in patients with documented central
adrenal insufficiency. However, our data demonstrate that,
even when the dose of Cortrosyn was lowered with the use
of LDC, some patients, albeit fewer in number, continue to
have normal Cortrosyn-stimulated levels despite being adrenally insufficient.
It is unclear as to why DHEA-S values would be that low
even when corticotropin deficiency was only partial. It is
suggested that, in cases of early corticotropin deficiency, the
adrenal would preferentially divert more of its resources
toward cortisol production. Such instances of cortisol/
DHEA-S dissociation have been described in a variety of
clinical situations, even when corticotropin levels are abun-
Nasrallah and Arafah • DHEA-S and Adrenal Function
J Clin Endocrinol Metab, November 2003, 88(11):5293–5298 5297
TABLE 4. Clinical and biochemical characteristics of nine
patients with NL-HPA and low DHEA-S
DHEA-S (␮g/dl)
Age/
gender
Hormone
deficit
Base line
Range
19/F
33/F
41/F
17/M
25/M
29/M
43/M
52/M
64/M
0
G
G
0
G
G,T
G,T
0
G
61
28
13
142
120
160
83
68
27
65–380
45–270
32–120
280 – 640
280 – 640
280 – 640
95–530
70 –310
42–290
The above nine patients were selected from group 2 (n ⫽ 47), all of
whom had pituitary adenomas and NL, HPA function as defined in
the text but had low gender- and age-adjusted serum DHEA-S levels.
G, Gonadotrophs; T, thyroid, 0, none, Range, normal DHEA-S
levels matched for gender and age for 20-yr age-group intervals. To
convert serum DHEA-S from ␮g/dl to ␮M, multiply the value by
0.02714.
dant (9). Mechanisms regulating this dissociation are not well
defined but may involve factors such as preferential zoning
of blood flow away from the reticularis and toward the
fasiculata layer and perhaps altered adrenal enzyme activity.
DHEA-S, the most abundant steroid in the circulation, is
practically all secreted by the adrenal glands, with minor
contribution by the testis (18, 19). DHEA-S is fully derived
from the sulfation of DHEA, and levels of both of these
steroids correlate under most clinical circumstances. In longitudinal studies, levels of DHEA-S have shown tracking in
the same individual (19). Furthermore, DHEA-S has a halflife of 10 –20 h and does not follow a circadian rhythm. Basal
levels correlate closely and are approximately 10 –15% lower
than Cortrosyn-stimulated levels. All of these factors make
a single measurement of DHEA-S practical and reliable.
We do not believe that the concomitant pituitary hormone
abnormalities in some of the patients had a significant impact
on our findings or conclusions. Our data indicate that serum
DHEA-S levels in men and women with NL-HPA function
were not affected by their respective gonadal function. Published studies by other investigators are supportive of this
view (19 –21). Even though some of the women in our study
were menopausal, it is unlikely that DHEA-S data were compromised. In a prospective study of 172 women in whom
DHEA-S levels were followed over a period of 7 yr, through
the menopause, there was no correlation between DHEA-S
and biochemical menopausal changes (20). Likewise,
DHEA-S levels seem independent of states of gonadal dysfunction (21). PRL is reported to increase serum DHEA-S by
mechanisms that are unclear but may be attributable to increased adrenal enzyme expression and/or decreased
DHEA-S clearance (22). This was not found to be a significant
contributor in the presence of corticotropin deficiency (10).
Furthermore, the number of patients with PRL-secreting tumors in the two groups, studied herein, was similar.
One limitation of the present study is the relatively small
sample size, especially when the population is divided into
subgroups. Nevertheless, the uniformity of the response supports our hypothesis that normal DHEA-S levels indicate an
intact HPA function. Because serum DHEA-S levels nor-
FIG. 3. ROC curves for the LDC-stimulated serum cortisol levels
(solid lines) and for basal DHEA-S serum levels (dashed lines). The
curves were drawn based on the IIH test as the gold standard to define
NL-HPA. The area under the curve for LDC-stimulated serum cortisol
levels is 0.893 (CI, 0.817– 0.969) and for serum DHEA-S is 0.984 (CI,
0.962–1.000).
mally decline progressively with advancing age, a low serum
level in the elderly population could be difficult to interpret.
Based on our study and the ROC curve generated from our
data, we believe that the serum DHEA-S level that gives
100% sensitivity should be considered as the minimum required to define normal, in any group or gender. We believe
that, in addition to being in the age- and gender-matched
range, serum DHEA-S levels have to be more than 54.5 ␮g/dl
(1.48 ␮m) to be considered normal.
In view of the cortisol/adrenal androgen dissociation discussed above and the fact that multiple factors such as age,
gender, and previous steroid use modulate DHEA-S production, one should judge the clinical significance of low
serum DHEA-S levels with caution. One major group of
patients in whom interpretation of low serum DHEA-S levels
would not be reliable are those who received exogenous
glucocorticoids during the year before testing and those who
have chronic illnesses. In the former group of patients,
DHEA-S levels can remain low for some time despite normal
cortisol secretion. Thus, even though partial or complete
deficiency of corticotropin results in a low serum DHEA-S
level, it would be imprudent to assume that the converse is
always true. Specifically, because many factors (such as advancing age, chronic illnesses, and prior and current use of
glucocorticoids) can independently result in low serum
DHEA-S levels, one should interpret such low levels with
clinical scrutiny. One should keep the clinical context of the
patient in focus in the evaluation of low serum DHEA-S
levels.
Our data clearly demonstrate that, when serum DHEA-S
levels are normal, the diagnosis of corticotropin deficiency is
extremely unlikely. Although our study included only patients with pituitary adenomas, there is no apparent reason
to limit its applicability to such patients. We believe that the
conclusions and recommendations can be extended to other
patients investigated for the possibility of central adrenal
insufficiency. In view of our findings, we believe that the
5298
J Clin Endocrinol Metab, November 2003, 88(11):5293–5298
biochemical assessment of adrenal function should include
measurements of cortisol and DHEA-S serum levels. In the
absence of corticosteroid-binding globulin excess (e.g. with
estrogen therapy), a random unstimulated serum cortisol
level of more than 18.5 ␮g/dl (510 nm) and/or a normal
age-and gender-adjusted serum DHEA-S level would practically and effectively rule out the diagnosis of adrenal insufficiency. As stated herein, serum DHEA-S levels are extremely helpful only when they are normal. Thus, when they
are low, assessment of adrenal function should rely on clinical evaluation and measurements of serum cortisol level and
its response to stimulation. The LDC provides a simple,
noninvasive, though imperfect test. An LDC-stimulated serum cortisol of less than 16.9 ␮g/dl (466 nm) confirms the
diagnosis of corticotropin deficiency, whereas a level of more
than 24.4 ␮g/dl (673 nm) rules it out. When the LDC-stimulated serum cortisol levels are intermediate, further testing
would be necessary, as clinically indicated. It is in those
instances where the value of serum DHEA-S is most valuable. It is important to point out that DHEA is currently
available as an over-the-counter therapeutic agent or a supplement and could potentially be used by some patients
being investigated for the possibility of adrenal insufficiency.
Thus, it would be prudent to entertain the latter possibility
as a pitfall in the interpretation of a normal random serum
DHEA-S measurement.
In summary, a normal DHEA-S level makes the diagnosis
of corticotropin deficiency extremely unlikely, especially
when the Cortrosyn stimulation test is normal. We recommend measurements of serum DHEA-S as an important component in the assessment of HPA function. The data should
be interpreted with age, gender, and other factors modulating serum DHEA-S levels in mind.
Acknowledgments
The authors are indebted to Dr. Mark Schluchter for his assistance in
the statistical analysis. We also thank our patients’ referring physicians,
the General Clinical Research Center personnel, and Barbara Vaughn
and Nancy Ballou of the ambulatory facility nursing staff. Finally, we
extend a special word of appreciation to our colleagues, Dr. Amir Hamrahian, Dr. Juan Ybarra, and Dr. Dina Serhal, for their help in the study
and to Dr. Saul Genuth and Dr. Faramarz Ismail-Beigi for reviewing the
manuscript.
Received March 13, 2003. Accepted August 6, 2003.
Address all correspondence and requests for reprints to: Baha M.
Arafah, M.D., Division of Clinical and Molecular Endocrinology, University Hospitals of Cleveland, 11100 Euclid Avenue, Cleveland, Ohio
44106. E-mail: [email protected].
This work was conducted and supported by a grant to the General
Clinical Research Center of Case Western Reserve University.
A preliminary report of the data was presented at the 83rd Annual
Meeting of The Endocrine Society, Denver, CO, 2001.
Nasrallah and Arafah • DHEA-S and Adrenal Function
References
1. Oelkers W 1996 Adrenal insufficiency. N Engl J Med 335:1206 –1212
2. Bangar V, Clayton RN 1998 How reliable is the short syncorticotropin test for
the investigation of the hypothalamic-pituitary-adrenal axis? Eur J Endocrinol
139:580 –583
3. Streeten D, Anderson Jr GH, Bonaventura MM 1996 The potential for serious
consequences from misinterpreting normal responses to the rapid adrenocorticotropin test. J Clin Endocrinol Metab 81:285–290
4. Oelkers W 1998 The role of high- and low-dose corticotropin tests in the
diagnosis of secondary adrenal insufficiency. Eur J Endocrinol 139:567–570
5. Abdu TA, Elhadd TA, Neary R, Clayton RN 1999 Comparison of the low dose
short Syncorticotropin test (1 ␮g), the conventional dose short Syncorticotropin
test (250 ␮g), and the insulin tolerance test for assessment of the hypothalamopituitary-adrenal axis in patients with pituitary disease. J Clin Endocrinol
Metab 84:838 – 843
6. Thaler LM, Blevins LS 1998 The low dose (1 ␮g) adrenocorticotropin stimulation test in the evaluation of patients with suspected central adrenal insufficiency. J Clin Endocrinol Metab 83:2726 –2729
7. Mayenknecht J, Diederich S, Bahr V, Plockinger U, Oelkers W 1998 Comparison of low and high dose corticotropin stimulation tests in patients with
pituitary disease. J Clin Endocrinol Metab 83:1558 –1562
8. Dickstein G 2001 Hypothalamo-pituitary-adrenal axis testing: nothing is sacred and caution in interpretation is needed. Clin Endocrinol (Oxf) 54:15–16
9. Parker LN 1991 Control of adrenal androgen secretion. Endocrinol Metab Clin
North Am 20:401– 421
10. Yamaji T, Ishibashi M, Takaku F, Itabashi A, Katayama S, Ishii J 1987 Serum
dehydroepiandrosterone sulfate concentrations in secondary adrenal insufficiency. J Clin Endocrinol Metab 65:448 – 451
11. Young J, Couzinet B, Nahoul K, Brailly S, Chanson P, Bauieu E, Schaison
G 1997 Panhypopituitarism as a model to study the metabolism of dehydroepiandrosterone (DHEA) in humans. J Clin Endocrinol Metab 82:2578 –2585
12. Pang S, Legido A, Levine L, Temeck J, New M 1987 Adrenal androgen
response to metyrapone, adrenocorticotropin, and corticotropin-releasing hormone stimulation in children with hypopituitarism. J Clin Endocrinol Metab
65:282–289
13. Miller KK, Sesmilo G, Schiller A, Schoenfeld D, Burton S, Klibanski A 2001
Androgen deficiency in women with hypopituitarism. J Clin Endocrinol Metab
86:561–567
14. Arafah BM 1986 Reversible hypopituitarism in patients with large nonfunctioning pituitary adenomas. J Clin Endocrinol Metab 62:1173–1179.
15. Arafah BM, Kailani SH, Nekl KE, Gold RS, Selman WR 1994 Immediate
recovery of pituitary function after transsphenoidal resection of pituitary macroadenomas. J Clin Endocrinol Metab 79:348 –354
16. Hout WM, Arafah BM, Salazar R, Selman WR 1988 Evaluation of the hypothalamic-pituitary-adrenal axis immediately after pituitary adenomectomy:
is perioperative steroid therapy necessary? J Clin Endocrinol Metab 66:1208 –
1212
17. Dickstein G, Shechner C, Nicholson WE, Rosner I, Shen-Orr Z, Adawi F,
Lahav M 1991 Adrenocorticotropin stimulation test: effects of basal cortisol
level, time of day, and suggested new sensitive low dose test. J Clin Endocrinol
Metab 72:773–778
18. Kroboth PD, Salek FS, Pittenger AL, Fabian TJ, Frye RF 1999 DHEA and
DHEA-S: a review. J Clin Pharmacol 39:327–348
19. Nafziger AN, Bowlin SJ, Jenkins PL, Pearson TA 1998 Longitudinal changes
in dehydroepiandrosterone concentrations in men and women. J Lab Clin Med
131:316 –323
20. Burger HG, Dudley EC, Cui J, Dennerstein L, Hopper J 2000 A prospective
longitudinal study of serum testosterone, dehydroepiandrosterone sulfate,
and sex hormone-binding globulin levels through the menopause transition.
J Clin Endocrinol Metab 85:2832–2838
21. Counts DR, Pescovitz OH, Barnes KM, Hench KD, Chrousos GP, Sherins RJ,
Comite F, Loriaux DL, Cutler Jr GB 1987 Dissociation of adrenarche and
gonadarche in precocious puberty and in isolated hypogonadotropic hypogonadism. J Clin Endocrinol Metab 64:1174 –1178
22. Schiebinger RJ, Chrousos GP, Cutler Jr GB, Loriaux DL 1986 The effect of
serum PRL on plasma adrenal androgens and the production and metabolic
clearance of dehydroepiandrosterone sulfate in normal and hyperprolactinemic subjects. J Clin Endocrinol Metab 62:202–209