REVIEWS

Pros and cons of screening for occult

Cushing syndrome
Antoine Tabarin and Paul Perez
Abstract | Systematic screening studies performed mainly in patients with diabetes mellitus have revealed
an unexpectedly high prevalence of occult Cushing syndrome. Such studies may provide a rationale for
systematically screening obese patients with type2 diabetes mellitus. However, a screening strategy is only
justified if it is supported by enough evidence of its efficacy and if the benefits will outweigh drawbacks.
To date, the natural history of occult Cushing syndrome and its possible effect on long-term morbidity are
unknown. The clinical spectrum of occult Cushing syndrome and its relatively low prevalence may potentially
negatively affect the performance of endocrine tests used to diagnose overt Cushing syndrome and generate
false positives. Whether the cure of occult Cushing syndrome favorably influences clinical outcomes and is
more beneficial than treatment of diabetes mellitus and cardiovascular risk factors with currently available
pharmacological tools remains to be demonstrated. Last, the acceptability of a screening program by
professionals and the health-care system in terms of workload and costs is highly questionable. Thus, an
assessment of the indications for and against screening for occult Cushing syndrome on the basis of currently
available data suggests that, to date, the cons surpass the pros.
Tabarin, A. & Perez, P. Nat. Rev. Endocrinol. 7, 445455 (2011); published online 22 March 2011; doi:10.1038/nrendo.2011.51

Introduction
Cushing syndrome is a clinical entity that reflects
chronic and excessive exposure of the body to gluco
corticoids. It includes numerous clinical features that
reflect the widespread distribution of glucocorticoid
receptors in multiple target organs.1,2 In most instances,
Cushing syndrome develops secondary to the use of
exogenous glucocorticoids, so-called iatrogenic Cushing
syndrome. Spontaneous or endogenous Cushing syn
drome is far less frequent and results from an excessive
production of cortisol by the adrenal cortex. From a
pathophysiological perspective, three conditions can
lead to endogenous Cushing syndrome.2,3 In the major
ity of cases, the excessive production of cortisol is driven
by an inappropriate secretion of adrenocorticotropic
hormone (ACTH) by a pituitary corticotroph adenoma,
so-called Cushing disease. Rarely, ACTH can be pro
duced by a nonpituitary (ectopic) tumor. This entity
is called the ectopic ACTH syndrome. Finally, cortisol
can be autonomously oversecreted by unilateral or, more
rarely, by bilateral adrenal tumors.

Epidemiology of Cushing syndrome

Endogenous Cushing syndrome was first described
80years ago; 4 however, very few data concerning its
epidemiology are currently available. The two main
epidemiologic studies on Cushing syndrome were con
ducted in Europe more than 15years ago. A Danish
population-based study conducted between 1985 and
Competing interests
The authors declare no competing interests.

1995 found an incidence of 2.3 cases per million inhabi

tants per year.5 As found in many series published by
experie nced centers,2,6 Cushing syndrome occurred
mainly in adult women, and the predominant cause
was Cushing disease (60% of cases). Other etiologies,
such as cortisol-secreting adrenal tumors and ectopic
ACTH syndrome, occurred in less than 30% and 10% of
cases, respectively.5 Somewhat similar results concern
ing Cushing disease have been found by Etxabe etal.7 in
a population-based study conducted in Vizcaya, Spain,
between 1975 and 1992. The investigators identified an
average incidence of Cushing disease of 2.4 cases per
million inhabitants per year and a prevalence of 39.1
cases per million inhabitants. Interestingly, a progres
sive increase in the incidence of Cushing disease was
identified over the 18years of the study. Similarly, a
threefold increase in the number of patients newly diag
nosed with Cushing disease after 1987 was found in a
multicenter Italian study that was conducted between
1975 and 1995.6,8 An increased awareness of the disease,
improvements in the performance of diagnostic tools and
the establishment of endocrinology referral centers may
account for these findings. However, despite this appar
ent increasing incidence, it is commonly accepted that
endogenous Cushing syndrome is a rare disorder.
One clear limitation of these studies is that they only
assessed patients referred to hospitals, which usually
follow up patients with the most severe and clinically
obvious diseases. Although full-blown Cushing syndrome,
in other words when the syndrome is overt, is clinically
instantly recognizable, one obstacle for its diagnosis is

NATURE REVIEWS | ENDOCRINOLOGY

Department of
Endocrinology
(A.Tabarin), Clinical
Epidemiology Unit
(USMR) & CIC-EC7,
INSERM, (P. Perez),
Universit Bordeaux 2,
Centre Hospitalier
Universitaire de
Bordeaux, 146Rue Lo
Saignat, F33076
Bordeaux CEDEX,
France.
Correspondence to:
A. Tabarin
antoine.tabarin@
chu-bordeaux.fr

VOLUME 7 | AUGUST 2011 | 445

2011 Macmillan Publishers Limited. All rights reserved

REVIEWS
Key points
Common clinical experience and a few epidemiological studies suggest that
endogenous Cushing syndrome is a rare condition
Systematic screening studies performed in specific populations that exhibit
nonspecific symptoms of Cushing syndrome (mainly patients with diabetes
mellitus) reveal a prevalence of occult Cushing syndrome that reaches 3.3%
A screening strategy is justified when its efficacy is supported by evidence
and if benefits outweigh potential drawbacks
Examination of the indications for and against screening for occult Cushing
syndrome reveals that the cons exceed the pros
Additional studies are needed before we can acknowledge, from an
evidence-based perspective, the usefulness of systematic screening for occult
Cushing syndrome

Cushing disease
Pseudo-Cushing syndrome

Glucose intolerance
Hirsutism
Acne
Hypertension
Myopathy

Purple striae

*
*

Easy bruising
Visceral obesity
25

50
Prevalence of symptoms (%)

75

100

Figure1 | Limitations of physical examination to identify mild Cushing syndrome.

Prevalence of symptoms in patients with mild Cushing disease (red bars) and
pseudo-Cushing (blue bars). The two groups had similar and moderate increases
in 24h urinary free cortisol as determined by Gatta and colleagues.11 *Statistical
significance at P<0.05. Note that specific symptoms of Cushing syndrome are
lacking in roughly half of patients with true Cushing syndrome.

its broad spectrum of presentation. Symptoms such as

obesity, diabetes mellitus and hypertension are highly
prevalent in patients with Cushing syndrome but are also
extremely common in the general population and are thus
poorly discriminant.1,2,9,10 Conversely, so-called specific
symptoms that reflect protein-wasting owing to longstanding and/or intense hypercortisolism, such as purple
striae, proximal myopathy, easy bruising and osteoporosis
in young patients, have a strong discriminant power but
may be subtle or lacking in a number of patients.1113 In a
previous study designed to differentiate ambiguous cases
of Cushing syndrome from other conditions that mimick
this situation (pseudo-Cushing syndrome), we found that
specific symptoms were absent in approximately half of
patients with Cushing disease of mild intensity as reflected
by urinary free cortisol (UFC) excretion (Figure1).11 This
variability of presentation is responsible for diagnostic
446 | AUGUST 2011 | VOLUME 7

difficulties. Moreover, the selection of patients in whom

biochemical screening for Cushing syndrome is relevant
also depends on the experience of the examiner. Thus,
one may hypothesize that this spectrum of clinical pre
sentation may lead to underdiagnosis and that the true
prevalence of Cushing syndrome in the population may
be higher than that quoted in population-based studies.
This hypothesis is also supported by several isolated
case reports of minimally symptomatic cases of Cushing
disease that present as common hypertension and dia
betes mellitus14,15 and, more importantly, by systematic
screening studies conducted in specific populations.

Systematic screening studies

Screening patients with diabetes mellitus
Cortisol plays a key part in glucose metabolism and acts
on critical target tissues, such as liver, adipose tissue, skel
etal muscle and the endocrine pancreas.16,17 Consequently,
chronic hypercortisolism results in the entire spectrum
of the metabolic syndrome (abdominal obesity, insulin
resistance, dyslipidemia and hypertension).3,18,19 Glucose
intolerance or type2 diabetes mellitus secondary to accu
mulation of abdominal adipose tissue and insulin resis
tance may occur in up to 60% of patients with Cushing
syndrome.3,2023 Type2 diabetes mellitus and the meta
bolic syndrome can thus be a clue to the presence of
Cushing syndrome.
In 2000, a prospective study on Cushing syndrome in
200 overweight adults with type2 diabetes mellitus, con
secutively referred to hospital for poor metabolic control
(HbA1c >8%), showed that none of the patients had overt
Cushing syndrome under careful clinical examination.24
The diagnostic algorithm and main results of the study
are summarized in Figure2. We performed a first screen
ing step with the 1mg overnight dexamethasone sup
pression test (DST) using a low threshold for cortisol
suppression to maximize the sensitivity of the procedure.
A second confirmatory step involved multiple biological
investigations of the hypothalamicpituitaryadrenal
(HPA) axis. Patients who displayed at least one abnor
mal test underwent a third step of imaging investiga
tions. Overall, and considering only patients with visible
tumors, an unexpected prevalence of Cushing syndrome
of 5.5% was found in this particular population. Given
that none of the patients were a priori suspected of
having Cushing syndrome on clinical grounds following
examination by experienced endocrinologists, the term
of occult Cushing syndrome was chosen to design this
peculiar presentation.
Several methodologically acceptable studies that use
different designs have prospectively investigated the pres
ence of occult Cushing syndrome in various populations
with diabetes mellitus (Table1). The common feature of
these studies was the absence of specific clinical signs
of Cushing syndrome in study participants. Chiodini
etal.25 used a similar procedure to that mentioned above
in 294 hospitalized patients with type2 diabetes mellitus
aged >30years. The study included a control group of
nondiabetic patients matched for age and BMI, who were
admitted to the hospital during the same period.25 The
www.nature.com/nrendo

2011 Macmillan Publishers Limited. All rights reserved

REVIEWS
prevalence of occult Cushing syndrome in patients with
diabetes mellitus reached 9.4% and was 4.8-fold more
frequent than in the control group. Similarly, Leibowitz
etal.26 performed a screening study using the 1mg DST in
90 overweight individuals with poorly controlled diabetes
mellitus (HbA1c >8%) and identified 3.3% of patients with
occult Cushing syndrome. Another screening study per
formed in 100 patients with newly diagnosed type1 and
type2 diabetes mellitus who were not selected for their
clinical characteristics, such as poor metabolic control
or obesity, identified one case (1%) of occult Cushing
syndrome.27 Two methodologically acceptable negative
studies have also been published. A screening strategy per
formed in 154 elderly men with type2 diabetes mellitus
using the late-night salivary cortisol test did not identify
any case of occult Cushing syndrome.28 Moreover, Mullan
etal.29 did not find any case of occult Cushing syndrome
among 201 consecutive patients with diabetes mellitus
who had at least two of the following criteria: HbA1c >7%,
overweight (BMI >25kg/m2) and hypertension.29 Other
studies with questionable methodologies have yielded
divergent inconclusive results.30,31
Several arguments may account for the discrepancies
between positive and negative series. Sample variability
is important to consider. If the true prevalence of occult
Cushing syndrome is around 1%, the probability of iden
tifying at least one case in a cohort of 200 patients is not
more than 87%. The characteristics of the investigated
population also influence the prevalence of Cushing syn
drome. Intuitively, the presence of this condition is more
probable in patients with central obesity, hypertension
and long-standing, uncontrolled diabetes mellitus24,26 than
in patients with mild impairment of glycemic control29 or
newly diagnosed diabetes mellitus.27 Screening exclusively
or predominantly men for a disease that occurs mainly in
women is also a shortcoming of some studies.27,28 With
drawal of patients from the study who had a positive
screening test but did not undergo confirmation tests
may also hamper the identification of patients with occult
Cushing syndrome.28,29 Obviously, the type and sensitivity
of the screening investigation is of paramount importance
and will be discussed later.

Screening patients with osteoporosis

Glucocorticoids inhibit calcium absorption from the
gut, calcium reabsorption by the renal tubule and
directly modulate the function of osteoblasts and osteo
clasts. In addition, hypercortisolism indirectly impairs
calcium and bone metabolism through the inhibition
of growth hormone secretion and induction of gonadal
dysfunction.32 Consequently, osteoporosis is a common
feature of Cushing syndrome.3 Only one prospective study
has assessed the occurrence of occult Cushing syndrome
in apparently primary osteoporosis.33 Using a screening
strategy based on the 1mg DST as first-line investigation
in 219 outpatients referred for evaluation of osteoporosis
and lacking specific clinical features of hypercortisolism,
Chiodini etal.33 identified occult Cushing syndrome in
10.8% of patients with both asymptomatic vertebral
fractures and low BMD (95% CI 3.2318.31).

100%
(200 patients)

1 Screening step
1 mg DST

Impaired 1 mg DST
(cortisol >60 nmol/l)
26%
Drop-out: 5 patients
2 Confirmation step
ACTH, cortisol,
circadian rhythm, UFC
Midnight cortisol, 4 mg DST

24%

Normal endocrine
evaluation

30 patients
(15% false-positive rate)
Drop-out: 3 patients

3 Imaging step

8%

Adrenal tumor:
4.0% (8 patients)

Cushing disease:
1.5% (3 patients)

No visible tumor:
1.5% (3 patients)

Figure2 | Algorithm used to screen for occult Cushing syndrome in obese patients
with type2 diabetes mellitus. Abbreviations: ACTH, adrenocorticotropic hormone;
DST, dexamethasone suppression test; UFC, urinary free cortisol. Adapted with
permission from The Endocrine Society Catargi, B. etal. J. Clin. Endocrinol.
Metab. 88, 58085813 (2003).

Table 1 | Screening studies for occult Cushing syndrome in diabetes mellitus

Reference

Inclusion criteria

Suspected Cushing
syndrome (%)

Proven Cushing
syndrome* (%)

Leibowitz
etal.26

90

T2DM, BMI >25kg/m2,

HbA1c >9%

3.3

3.3

Catargi
etal.24

200

T2DM, BMI >25kg/m2,

HbA1c >8%

7.0

2.0

Chiodini
etal.33

294

>30years, in-patients

9.4

1.0

Liu etal.28

154

T2DM, veterans, men

0.0

0.0

Reimondo
etal.27

99

Newly diagnosed DM

1.0

1.0

Taniguchi
etal.59

77

In-patients, T2DM

7.8

2.6

Mullan
etal.29

201

Two of: BMI >25kg/m2,

hypertension, HbA1c >7%

0.0

0.0

*Patients with definitive diagnosis of Cushing syndrome obtained after surgical excision of a pituitary
corticotroph adenoma or cortisol-secreting adrenal adenoma. Abbreviations: DM, diabetes mellitus;
n,number of patients screened; T2DM, type2 diabetes mellitus.

Screening patients with hypertension

Chronic hypercortisolism increases blood pressure
through a complex interplay between several mecha
nisms that regulate plasma volume, peripheral vascular
resist ance, endothelial function and cardiac output. 4
Hypertension is a common feature of Cushing syndrome
and is found in approximately two-thirds of adults and
more than half of pediatric patients.3,20,22,23 Two studies
that address the prevalence of different causes of secon
dary hypertension have been conducted in large cohorts
of patients in Japan35 and in the USA36 (1,020 and 4,429
patients, respectively). The prevalence of Cushing syn
drome among patients with hypertension was 0.5% in
the US study and 2.0% in the Japanese study. Although

NATURE REVIEWS | ENDOCRINOLOGY

VOLUME 7 | AUGUST 2011 | 447

2011 Macmillan Publishers Limited. All rights reserved

REVIEWS
Box 1 | Indication criteria for screening
Burden of target disease justifies action
Natural history of the disease is adequately understood
Preclinical or early symptomatic phase is long enough
Diagnostic performance of screening test is good
Accepted treatment or useful intervention for patients
with the disease is available
Early diagnosis improves clinical outcomes
Screening does more good than harm
Screening is acceptable by the screened population
and the health-care system
Based on Sackett etal.54 and WHO criteria for screening.55

hypertension, and not clinical suspicion of Cushing syn

drome, was the referral reason, little information con
cerning the clinical presentation of patients is available.
Thus, the nosologic distinction between undiagnosed
Cushing syndrome owing to clinical unawareness of
physicians and truly occult Cushing syndrome remains
unknown. Moreover, underestimation of the prevalence
of Cushing syndrome in these cohorts may be antici
pated, because the biological tools used to screen for
Cushing syndrome have a poor diagnostic sensitivity
(morning and mid-afternoon plasma cortisol).

Screening patients with obesity

Weight gain is one of the earliest signs of Cushing syn
drome, with a characteristic redistribution of adipose
tissue from limbs to the abdominal region and to the face,
neck and trunk. Although it is well-acknowledged that
glucocorticoid excess has a pivotal role in fat redistribu
tion, the precise pathophysiology of the accumulation
of visceral adipose tissue still remains debated.16,3739 Of
note, this type of weight gain is common and often seen
in patients who are affected by the metabolic syndrome.
Screening studies in patients with obesity have pro
vided controversial results. A retrospective study, with
its inherent methodological biases, involved 86 patients
referred for obesity (associated with hypertension or
diabetes mellitus in half of them) who underwent a
1mg DST screening procedure.40 After complementary
investigation of those with no suppression of cortisol,
occult Cushing syndrome was identified in 3.5% of
patients. Even more impressively, a prospective Turkish
study conducted in 150 obese patients with no associated
diabetes mellitus and without specific clinical features
identified occult Cushing syndrome in 8.7% of cases.41 In
contrast with these results, a wide-scale, cross-sectional,
prospective study conducted in 369 patients with over
weight or obesity who were enrolled in a weight manage
ment program in the USA did not identify a single case
of Cushing syndrome.42
Polycystic ovary syndrome and hirsutism
Polycystic ovary syndrome (PCOS) is a multifactorial
complex disorder with dysregulated ovarian steroido
genesis that is responsible for impaired ovulation,
448 | AUGUST 2011 | VOLUME 7

hyperandrogenism and eventually characteristic morpho

logical changes of the ovaries.43 PCOS affects up to 10%
of women of reproductive age and is often associated with
obesity, insulin resistance and type2 diabetes mellitus.
Conversely, hypercortisolism impairs the hypothalamo
pituitarygonadal axis; whereas major hypercortisolism
usually induces hypogonadotrophic hypogonadism,
moderate hypercortisolism is compatible with main
tenance of gonadotropin stimulation to the ovary,
persistent estradiol secretion and may induce second
ary PCOS.44 Hence, PCOS may be a clue to the presence
of Cushing syndrome. Although isolated case reports of
occult Cushing syndrome presenting as PCOS have been
published,45 only one study has addressed the prevalence
of Cushing syndrome in patients referred for hirsutism.46
A single case of Cushing syndrome was identified in 340
women, but it is of note that biological investigations of
the HPA axis were performed only in patients with clini
cal suspicion of hypercortisolism. Hence, the prevalence
of clinical occult Cushing syndrome in this population
remains unknown.

Adrenal incidentalomas
This Review will not address the peculiar situation of
adrenal tumors incidentally discovered during abdomi
nal imaging. Depending on the referral processes, diag
nostic tests and criteria, biochemical hypercortisolism,
usually of mild intensity, may be identified in a variable
percentage (220%) of patients without clinical features
of Cushing syndrome.4752 Although many issues remain
unsolved concerning the definition and consequences
of this entity, several academic societies advocate the
indication of biochemical screening for occult Cushing
syndrome in this situation.10,51,53

Systematic appraisal of screening criteria

The potential indication for screening is based on a
set of various criteria (Box1). 54,55 Firstly, the burden
of the targeted condition, as evaluated by its frequency
and severity, should justify action. This assumption is
tightly linked to an appropriate knowledge of the natural
history of the disease. Secondly, in a screening perspec
tive, natural history should ideally include a latent or
preclinical stage to allow early diagnosis and interven
tion. Thirdly, the diagnostic accuracy of tests must be
high and the assessment of diagnostic accuracy needs the
existence of a reference standard. One must also bear in
mind that reproducibility is a prerequisite of diagnostic
accuracy. Fourthly, detection of the disease is only useful
if treatment improves clinical outcomes.
Although, screening often seems intuitively a good
action, potential harms of screening must be acknowl
edged. In the short term, screening can expose indivi
duals to investigations that are of no use to most of
them. Also, a substantial amount of evidence supports
the notion that the psychological consequences may not
be trivial.5658 Moreover, patients who are false positive
may be subject to adverse effects without any benefits.
Finally, the screening program must be acceptable by
professionals and the health-care system, in terms of
www.nature.com/nrendo

2011 Macmillan Publishers Limited. All rights reserved

REVIEWS
work load and costs. Given that type2 diabetes mellitus
has been the most extensively and rigorously studied
situation, we will examine the question of screening for
occult Cushing syndrome in patients with this disorder.

Burden of occult Cushing syndrome

The prevalence rates of suspected occult Cushing syn
drome in relevant studies range from 0.0% to 9.4%.2429,59
The pooled estimate of prevalence would be 1.98% (95%
CI 0.005.25), but this estimation is not sound, because
the prevalences are statistically heterogeneous between
studies (P<104).60 Of note, definite proof of Cushing
syndrome requires corticotropic insufficiency after
surgery and pathological examination of the removed
pituitary tumor or adrenal tumor. However, only 24%
of diabetic patients with suspected occult Cushing
syndrome identified in the positive screening studies
underwent surgery.2427,59 Meticulous examination of the
published data raises the possibility that Cushing syn
drome might have been overdiagnosed in some cases.
Similarly, definite diagnosis of Cushing syndrome was
obtained in less than half of the osteoporotic patients
with suspected occult Cushing syndrome,33 resulting in
a 2% prevalence among patients with osteoporosis.
Overall, the prevalence rates of proven occult Cushing
syndrome in individuals with diabetes mellitus in rele
vant studies range from 0.0% to 3.3%. The pooled preva
lence rate is 1.11% (95% CI 0.112.10; Table1), without
evidence of statistical heterogeneity between studies
(P=0.71). Although this number remains large with
regard to the epidemic prevalence of type2 diabetes mel
litus, as mentioned above, the study populations were
recruited from referral hospitals, which usually follow
up the more difficult patients to manage, and might thus
have been enriched for occult Cushing syndrome com
pared with the general community. Such recruitment
bias may potentially account for major inconsistencies
observed in the results of screening studies in patients
with obesity.4042
Several population-based studies and patient series
from expert centers conducted after 1975 have demon
strated that overt Cushing syndrome is a severe condition.
Although these studies are not devoid of methodological
concerns,61 they conclude that Cushing syndrome is
associated with an increase in standard mortality ratio,
with most deaths caused by vascular complications (myo
cardial infarction and cerebrovascular accident). Patients
with persistent hypercortisolism despite treatment have
an increased standard mortality ratio, whereas mortality
becomes similar to that of age-matched populations
after normalization of cortisol levels.5,7,61,62 Is this adverse
outcome transposable to occult Cushing syndrome?
To date, we lack prospective, long-term observational
or interventional studies that would provide evidencebased knowledge of the extent to which occult Cushing
syndrome is really an adverse condition. Two types of
indirect evidence of potential morbidity can be examined.
First, is the severity of the metabolic disease increased
in patients with occult Cushing syndrome? And second,
what is the intensity of biological hypercortisolism?

Chiodini etal.25 suggested that patients with both dia

betes mellitus and occult Cushing syndrome are affected
by a more severe metabolic disease than those with dia
betes mellitus alone, as indicated by a score that associ
ates the presence of hypertension, dyslipidemia and the
need for insulin treatment. However, treatments for these
conditions were not controlled and, in our study, patients
did not differ with regard to these criteria, nor for HbA1c
concentration.24 Thus, to date, it remains unclear whether
individuals with type2 diab etes mellitus and occult
Cushing syndrome have a more severe form of diabetes
mellitus or worsening of cardiovascular risk markers. Of
note, no data about metabolic comorbidities are provided
in the study of patients with osteoporosis.33
From a clinical perspective, the fact that patients were
examined by experienced endocrinologists who did not
observe so-called specific symptoms that reflect the
catabolic effects of glucocorticoid excess suggests that
they had mild Cushing syndrome rather than undiag
nosed overt Cushing syndrome. The pathophysiology
of the clinical consequences of chronic hypercortisolism
remains largely misunderstood. Multiple factors may
also be involved such as age of onset,63 sex,64 behavior
that may limit weight gain (such as diet and exercise),
inherited susceptibility for specific disorders such as dia
betes mellitus,65 individual tissue sensitivity to cortisol66
and duration of the disease.22 Several lines of evidence
also demonstrate a rough correlation between the inten
sity of hypercortisolism, the severity of clinical syndrome
and the presence of catabolic features.64,6772 Examination
of the results of individual biological investigations in
patients identified in screening studies confirms that
Cushing syndrome was of mild intensity in a number of
cases. Indeed, 24h UFC values quoted in five published
studies2427,33 were within the normal range in the major
ity of patients and were above 1.5-fold the upper limit of
normal in only 17% of cases.
Examination of the etiologies of occult Cushing syn
drome identified in screening studies also supports the idea
that it may represent a different entity from overt Cushing
syndrome. Although sex ratios are roughly similar in the
two conditions, with a female predominance, a remarkable
inverse relationship exists in the prevalence of cortisolsecreting adrenal adenomas versus ACTH-secreting
pituitary adenomas between overt and occult Cushing
syndrome.2,6 Indeed, pituitary-dependent Cushing disease,
which usually represents more than 60% of cases of overt
Cushing syndrome, only accounts for 23% of suspected
cases of occult Cushing syndrome in studies of patients
with diabetes mellitus and osteoporosis.2427,33,59 Thus, in a
number of cases, occult Cushing syndrome identified in
screening studies refers to mild adrenal-dependent hyper
cortisolism. This entity is close to that of so-called sub
clinical cortisol-secreting adrenal incidentalomas (SCSAs),
and some studies suggest that SCSAs are responsible for an
adverse metabolic condition compared with age-matched
and BMI-matched controls or patients with nonsecreting
adrenal incidentalomas.51,52,7375 Other studies have shown
that the clinical and metabolic phenotype of patients with
SCSAs is intermediary between that of normal individuals

NATURE REVIEWS | ENDOCRINOLOGY

VOLUME 7 | AUGUST 2011 | 449

2011 Macmillan Publishers Limited. All rights reserved

REVIEWS
and patients with overt Cushing syndrome.50,73 However,
divergent observations on the evolution of the metabolic
condition in the absence of specific intervention on the
adrenal tumors have been published.76,77 Consequently,
the long-term consequences of SCSAs are not precisely
known. The considerable heterogeneity across studies on
the biochemical definition of SCSAs may account for these
discrepancies.52 Be that as it may, whether or not SCSAs
markedly affects patients health is still a matter of debate.
Thus, to date, the possible effect of occult Cushing syn
drome on long-term morbidity in patients with diabetes
mellitus is not evidence-based.

Natural history of occult Cushing syndrome

Another aspect of the natural history of occult Cushing
syndrome that would support a screening strategy is if
occult Cushing syndrome corresponds to a preclinical
phase that precedes evolution towards overt Cushing syn
drome. Apart from reducing the period of time spent in
the hypothetically atherosclerotic and thrombotic milieu
of occult Cushing syndrome, this preclinical phase would
allow early intervention and prevention of the develop
ment of overt Cushing syndrome. However, this aspect
of the natural history of occult Cushing syndrome is
also unknown, and whether or not patients diagnosed
in screening studies have been identified by chance at an
early stage before the evolution towards severe hyper
cortisolism remains hypothetical. Although limited, the
literature concerning the follow-up of SCSAs suggests that
only a minority of patients will develop overt Cushing
syndrome (at least in the mid term) and that in most
patients the intensity of biological hypercortisolism does
not progress over time.76,7880 Alternatively, it has been
reported that some patients with SCSAs may develop
specific features of Cushing syndrome with time.80
Diagnostic performance of screening tests
Two aspects should be discussed: firstly, the most appro
priate primary screening test and, secondly, the confir
mation strategy. As mentioned in numerous reviews and
comprehensive textbooks, these two steps rely exclusively
on biological evaluation and not on imaging studies.13,10
The diagnostic accuracy of tests
The biological diagnosis of Cushing syndrome has gen
erated an impressive body of literature. For our purpose
we will rely on the diagnostic meta-analysis conducted by
Elamin etal.,81 which constituted the foundation for the
latest recommendations of the Endocrine Society.10 One
of the strengths of this meta-analysis was to focus mostly
on studies that recruited patients in whom diagnosis was
uncertain, that is, patients suspected of the target condition
and in whom tests would be applied in real life. Moreover,
this design avoids the acknowledged overestimation of
test accuracy from studies in which investigators evaluate
the test by distinguishing patients with already confirmed
disease from individuals who are disease-free.82
Elamin etal.81 report the accuracy of usual tests of
the HPA axis in terms of pooled diagnostic likelihood
ratios. Likelihood ratio parameters integrate sensitivity
450 | AUGUST 2011 | VOLUME 7

and specificity and express the gain in the likelihood of a

disease that is brought by the result of a diagnostic test.83
The negative likelihood ratio is of particular importance
in the context of screening tests that aim to safely exclude
patients without Cushing syndrome from a further con
firmation strategy. This ratio has to be the lowest, and it
is usually accepted that a value below 0.1 reflects a good
ability to rule out the target condition. Conversely, a
positive likelihood ratio above 10 is considered accept
able for a test to rule in the target condition.83 The main
conclusion of the study by Elamin etal.81 is that the 1mg
overnight DST, the low-dose DST, midnight salivary and
plasma cortisol concentrations and 24h UFC are accu
rate, because they all approximately satisfy these criteria
and have roughly similar performances according to
their likelihood ratios. However, the lack of precision in
estimates, as reflected by the 95% CIs, makes these seem
ingly satisfactory values compatible with lower accuracy,
and the performance of tests may be very different across
studies. The latter is quantified by the I2 statistic, an index
of inconsistency among studies.84 In Elamins work, the I2
value was specifically high for midnight plasma and sali
vary cortisol level (50% and 78%) and was the lowest for
DSTs (0% and 11%), which reflect more homogeneous
results. This point has been observed for measurement of
salivary cortisol in an independent meta-analysis.85 Given
the feasibility of the 1mg DST, these results favor its use
for screening, assuming that patients do not take drugs
that interfere with the metabolism of dexamethasone or
increase corticosteroid-binding globulin (CBG) levels.10
However, Elamins meta-analysis81 is a mix of studies with
different patient characteristics. Although several sys
tematic screening studies were included, most selected
studies included patients with clinical suspicion of Cushing
syndrome, as in a diagnostic situation. The majority of
study patients may, therefore, have shown symptoms
of full-blown Cushing syndrome. Thus, whether or not the
performance of tests is transposable to the specific context
of mild and occult Cushing syndrome in patients with
diabetes mellitus remains debatable. The possibility that
the conclusion of the meta-analysis may not apply fully to
a screening strategy can be discussed through three main
arguments that directly influence the performances of
screening tests: the pathological context, the pathological
spectrum and the prevalence of the disease.
Influence of pathological context
Stress and hypoglycemia are two major stimuli of the HPA
axis. Consequently, poorly controlled diabetes mellitus
(acidoketosis, as well as recurrent and profound hyper
glycemia) will result in subsequent activation of the HPA
axis and a biological picture of Cushing syndrome.1,10,86
This drawback generated numerous false-positive results
when the screening strategy was applied to individuals
with newly diagnosed diabetes mellitus, which were no
longer present when patients were re-evaluated some time
after the initial hospital admission.27
Several studies have also suggested that diabetes mel
litus and its complications (such as hypertension, neuro
pathy, microangiopathy) are associated with functional
www.nature.com/nrendo

2011 Macmillan Publishers Limited. All rights reserved

REVIEWS

Influence of the intensity of hypercortisolism

As discussed previously, abnormalities in the biology
of the HPA axis are mild in most patients with occult
Cushing syndrome and most screening studies used lowstringency thresholds to maximize the sensitivity of the
screening procedure. The major drawback of this strat
egy is an increase in false-positive results. This finding
is illustrated in Figure3, which is based on data from
a screening study that we conducted in patients with
type2 diabetes mellitus.24 Use of the historical 138nmol/l
threshold for cortisol concentrations following a 1mg
DST would have missed half of the patients with sus
pected occult Cushing syndrome. By contrast, the use
of a low 60nmol/l threshold that picked up all cases of
occult Cushing syndrome led to a 15% false-positive
rate. Moreover, although measurement of midnight
salivary cortisol is widely accepted for the diagnosis of
Cushing disease of mild intensity, 9092 several studies
have questioned the sensitivity of this procedure to diag
nose mild Cushing syndrome in the context of adrenal
incidentalomas and SCSAs.93,94

700
600
500

Cortisol (nmol/l)

activation of the HPA axis.28,8789 The normal reference

ranges of different biological tests of the HPA axis may not
apply to this specific population. During the screening of
206 male veterans (including 154 with diabetes mellitus)
without Cushing syndrome, Liu etal.28 demonstrated
that increasing age, diabetes mellitus and hypertension
were independently associated with elevated average latenight salivary cortisol levels. These findings argue against
the use of conservative thresholds for midnight plasma
cortisol in this population and suggest that comorbiditysp ecific thresholds, which to our knowledge are not
available, may be required.

400
300
200
100
0

10

11
12
HbA1c (%)

13

14

15

16

Figure3 | Influence of the spectrum of the disease on the performance of

screening tests. Plasma cortisol levels after the overnight 1mg DST versus HbA1c
in 200 consecutive overweight patients with type2 diabetes mellitus with poor
metabolic control. Horizontal lines indicate thresholds for cortisol suppression at
60nmol/l and 138nmol/l. Occult Cushing syndrome with a visible tumor was
identified in three patients with Cushing disease (closed circles) and seven
patients with an adrenal adenoma (closed squares). Note that the use of a low
threshold is mandatory to identify all cases of occult Cushing syndrome but results
in approximately 15% of false-positive results among patients with diabetes
mellitus but without hypercortisolism (small symbols). Abbreviation: DST,
dexamethasone suppression test. Adapted with permission from The Endocrine
Society Catargi,B. etal. J. Clin. Endocrinol. Metab. 88, 58085813 (2003).

Influence of the intraindividual variability

Reproducibility is a prerequisite of diagnostic accuracy.95
Another point of difficulty is spontaneous fluctuation in
cortisol secretion, a common finding in overt Cushing
syndrome which led to the recommendation to perform
screening tests such as midnight salivary cortisol and 24h
UFC at least twice.10 The potential drawbacks of these
fluctuations may be amplified in patients with mild hyper
cortisolism, where biological endocrine abnormalities
fluctuate around the limits of normal and low-stringency
thresholds are used.96 In a prospective follow-up study
of 51 patients with adrenocortical incidentalomas who
were followed at annual intervals, we found that 20% of
patients fluctuated between the biological definitions
of nonsecreting adenoma or SCSA across evaluations.80
Similar observations of intermittent autonomous cortisol
secretion in SCSAs have been reported in up to 50% of
patients.97 The consequences of these fluctuations will be
the need to repeat screening tests with borderline results
and multiplication of follow-up evaluations to understand
the underlying reason of this variability.

that individuals with a negative screening test will not

be submitted to further evaluation. The relatively low
prevalence of occult Cushing syndrome in patients with
diabetes mellitus will lead to high negative predictive
values, which are desirable in a screening setting. When
the cortisol suppression test is negative in such lowprevalence populations, the post-test probability would
be less than 0.1% (Figure4). Conversely, positive predic
tive values will be limited; in other words, the propor
tion of false-positive results will be high. Calculation or
nomograms allow the post-test probability of a disease to
be established according to its pre-test probability (actual
prevalence of the disease in the target population) and the
likelihood ratios of the diagnostic test used.54 As illus
trated in Figure4, in a population with a 35% prevalence
of Cushing syndrome (equivalent to the mean quoted in
most studies of the meta-analysis by Elamin et al.81), lack
of cortisol suppression with the 1mg DST increases the
probability of a diagnosis of Cushing syndrome to about
86%. In a population with a 12% prevalence of Cushing
syndrome (such as proven cases in screening studies of
patients with diabetes mellitus), the same results will
increase the probability of Cushing syndrome only
modestly, to around 1019%.

Influence of the prevalence of the disease

In a screening setting, the post-test probability of disease
when the screening test is negative must be low, given

Drawbacks of screening
Altogether, the influence of pathological context, disease
spectrum and low prevalence of occult Cushing syndrome

NATURE REVIEWS | ENDOCRINOLOGY

VOLUME 7 | AUGUST 2011 | 451

2011 Macmillan Publishers Limited. All rights reserved

REVIEWS
0.1

99

0.2

0.5

95

1,000
500

200

90
80

100
50

5
10
20

70

20
10

60

40

2
1

30

50

20

30

0.5

40
50

0.2
0.1

60

0.05

70

0.02

80

0.01
0.005

90

0.002
0.001

95

10
5

0.5

0.2
99

0.1

Pre-test
probability (%)

Likelihood
ratio

Post-test
probability (%)

Figure4 | Fagan nomogram showing the influence of the

prevalence of the disease (pre-test probability 1% or 35%)
on the probability of disease after a positive (red arrows) or
a negative (blue arrows) screening test.104

on the results of the screening strategy will be the occur

rence of a large number of false-positive results that would
expose a number of patients to possible psychological
stress and unnecessary complementary investigations.
The assessment of diagnostic accuracy requires the
existence of a reference standard. As emphasized in
the recommendations of the Endocrine Society, no bio
logical gold standard currently exists, and the presence of
two abnormal tests of high diagnostic accuracy is required
for the diagnosis of overt Cushing syndrome.10 However,
tests used for confirmation are the same as those used
for screening; thus, the accuracy of this strategy is highly
questionable in the context of mild hypercortisolism.52,96
In the specific context of adrenal incidentalomas, cortisol
secretion is distributed continuously among patients along
a spectrum from nonfunctioning adenomas to SCSAs,
and it is obviously difficult to identify different categories
within this continuum.
452 | AUGUST 2011 | VOLUME 7

This difficulty to definitely establish a diagnosis of mild

hypercortisolism with biochemical investigations used to
diagnose overt Cushing syndrome has been illustrated in
a study of patients undergoing surgery for adrenal inci
dentalomas.98 Eller-Vainicher etal.98 retrospectively exam
ined the ability of biochemical investigations of the HPA
axis to predict the postoperative occurrence of cortico
tropic insufficiency. In such patients, the occurrence of
corticotropic insufficiency after removal of the adrenal
tumor can be taken as a reliable and definitive criterion
of chronic cortisol excess at the central nervous system.
Indeed, the most accurate combination of two abnor
mal biological parameters of the HPA axis only had 68%
overall accuracy, with a 1.73 positive likelihood ratio and
a 0.44 negative likelihood ratio. Such results emphasize
the limitations of available biological tools to efficiently
diagnose true cortisol excess in tissues in the context of
mild hypercortisolism. Glucocorticoid sensitivity may
differ among tissues,99 and conceivably, in some patients,
subtle hypercortisolism may be enough to participate in
deleterious metabolic effects but not constantly suppress
HPA axis function. The absence of studies that directly
compare the effect of different biochemical testing algo
rithms on therapeutic strategies and important clinical
outcomes has also been emphasized in the context of
overt Cushing syndrome.10

Effect of specific treatment on outcomes

A critical issue is whether or not the cure of occult Cushing
syndrome has a favorable influence on health outcomes
and specifically on surrogate markers of outcomes, such as
the metabolic abnormalities that first prompted explora
tions. Screening studies published to date did not include a
therapeutic component and, as mentioned previously, only
a minority of patients with suspected occult Cushing syn
drome underwent surgery. Overall, these studies acknowl
edge a substantial improvement in glycemic control and
requirement for antidiabetic drugs in most if not all
patients after cure of hypercortisolism.2426,59 However,
the small number of patients operated on, the absence
of data on the selection of those who underwent surgery
(who may have had the more intense hypercortisolism), as
well as the lack of a comparison group treated medically,
hampers any conclusion about the benefit of cure of occult
Cushing syndrome.
A number of studies have reported the consequences of
surgical excision of SCSAs.51 Overall, most studies agree
on the improvement of blood pressure and diabetes mel
litus or glucose intolerance in the majority of patients with
SCSAs, whereas the benefits on BMI are more contro
versial. However, the definition of SCSA was highly variable
across studies, and all studies suffer from their retrospective
design and inherent biases. Above all else, one major issue
is whether surgical treatment does better than pharmaco
logical treatment of diabetes mellitus and other metabolic
abnormalities in these patients. Only four studies, all
marred with methodological shortcomings, have compared
the metabolic outcome of surgery to that of medical treat
ments in patients with SCSAs. A beneficial effect of surgery
has been suggested in most studies,100102 but opposing
www.nature.com/nrendo

2011 Macmillan Publishers Limited. All rights reserved

REVIEWS
conclusions have also been reported.77 In addition, dia
betes mellitus, obesity, hypertension and dyslipidemia may
persist in a number of patients despite biochemical cure
of overt Cushing syndrome.23,103 A prospective, random
ized study that compares the benefits of surgery to that of
the currently available pharmacological tools for diabetes
mellitus is, therefore, necessary to conclude on the benefit
of specific treatment of hypercortisolism in patients with
occult and mild Cushing syndrome.

Acceptability of screening
Whether the screening process is acceptable to the
screened population and to the health-care system is a
matter of debate. The convenience of the 1mg DST or latenight salivary cortisol sampling that can be performed in
ambulatory patients94 is well-acknowledged and would
probably be accepted by patients with diabetes mellitus
who are familiar with repeated, multiple biochemical
investigations for their metabolic disease. This hypothesis
remains to be demonstrated for subsequent confirma
tory steps in patients with positive screening results and
who are not followed up in referral clinics or hospitals.
From this perspective, it should be noted that 1030%
of patients who initially gave informed consent during
the diabetes mellitus, obesity and osteoporosis screening
studies refused further endocrine evaluation after a posi
tive initial investigation.24,25,28,29,33,42 The acceptability of
the workload for the medical system is also questionable,
given the epidemic increase in the prevalence of type2
diabetes mellitus and that 525% of patients with diabetes
mellitus screen positive.24,25,2729
Last, but not least, the cost of a strategy with numer
ous uncertainties as mentioned above has to be assessed.
Epidemiologic data in the French population acknowl
edge a prevalence of type2 diabetes mellitus of around
1.5 million in individuals older than 50years. Thus, in
France, the cost of the first ambulatory screening proce
dure in patients with diabetes mellitus would be around
76 million. The cost of complementary evaluations
including biochemical and imaging studies in 300,000
individuals who had a positive screening test would be
286 million. If we assume that 2% of patients have occult
Cushing syndrome and will undergo surgery, the total
cost will be at least 445 million without including pos
sible adverse events of surgery. The cost of treating true
positive patients would be around 181 million but that
of endocrine and diagnostic imaging investigations for
false positives would amount to at least 264 million.
Such sums have to be contrasted with the number of
scientific uncertainties surrounding screening that we
have discussed above.

Conclusions
In conclusion, systematic screening studies in selected
populations deserve credit for showing that the true
prevalence of Cushing syndrome might be higher than
previously thought. However, occult Cushing syndrome
may potentially be mild in most patients, which raises
important issues with regard to the difficulty to identify
and confirm Cushing syndrome. In addition, we lack the

Cons
Performance of
diagnostic tests
Outcomes and beneficial
impact of treatment?
Acceptability of screening
by patients and health
care system?

Severity?

Pros
Prevalence from
screening studies

Figure5 | Summary of pros and cons of screening for

occult Cushing syndrome.

demonstration that mild hypercortisolism, first, confers

an increased morbidity; second, is responsible, at least
partly, for metabolic abnormalities; and, third, that its
cure has a beneficial effect, superior in terms of morbidity
and cost to the pharmacological treatment of the meta
bolic disease. Owing to the lack of knowledge on efficacy
of interventions, uncertainty on the accuracy and possible
adverse effects of the screening strategy, the evidence to
date does not suggest that screening would do more good
than harm (Figure5).
Consequently, important complementary studies are
needed before we can acknowledge, from an evidencebased perspective, the usefulness of systematic screening
in targeted populations of patients who present with non
specific symptoms of Cushing syndrome. Such studies
may be observational, prospective, prognostic cohorts to
compare the evolution of type2 diabetes mellitus in patients
with different levels of HPA axis abnormalities and pro
spective randomized clinical trials to assess the beneficial
effect of a surgical cure of hypercortisolism compared to an
aggressive lifestyle and pharmacological intervention.
While the results of such studies (that will be difficult to
conduct) are eagerly awaited, we recommend increasing
clinical awareness in physician communities involved in
the treatment of diabetes mellitus and osteoporosis to look
accurately for specific symptoms of Cushing syndrome.
In patients with some of these features, the pre-test prob
ability of Cushing syndrome will be increased and screen
ing tests will display better performances. Whatever test is
used, we suggest, in accordance with the recommendations
of the Endocrine Society, the application of conservative
thresholds to minimize the number of false positives,
unnecessary complementary investigations and select for
specific treatments the more severe forms of Cushing syn
drome in which there is, to date, the best evidence of the
benefits of surgical intervention.10
Review criteria
A search for original articles published between 1975 and
2010 and focusing on Cushing syndrome was performed
in MEDLINE and PubMed. The search terms used were
subclinical Cushing syndrome, adrenal incidentaloma
and subclinical hypercortisolism. All articles identified
were French-language or English-language papers. We
also searched the reference lists of identified articles for
further papers.

NATURE REVIEWS | ENDOCRINOLOGY

VOLUME 7 | AUGUST 2011 | 453

2011 Macmillan Publishers Limited. All rights reserved

REVIEWS
1.

2.

3.

4.

5.

6.

7.

8.

9.

10.

11.

12.

13.

14.

15.

16.

17.

18.

Findling, J.W. & Raff, H. Screening and

diagnosis of Cushings syndrome. Endocrinol.
Metab. Clin. North Am. 34, 385402, ixx
(2005).
Newell-Price, J., Bertagna, X., Grossman, A.B. &
Nieman, L.K. Cushings syndrome. Lancet 367,
16051617 (2006).
Arnaldi, G. et al. Diagnosis and complications of
Cushings syndrome: a consensus statement.
J.Clin. Endocrinol. Metab. 88, 55935602
(2003).
Cushings, H. The basophil adenomas of the
pituitary body and their clinical manifestations
(pituitary basophilism). Obes. Res. 2, 486508
(1994).
Lindholm, J. et al. Incidence and late prognosis
of Cushings syndrome: a population-based
study. J. Clin. Endocrinol. Metab. 86, 117123
(2001).
Invitti, C., Giraldi, F.P., de Martin, M. &
Cavagnini,F. Diagnosis and management of
Cushings syndrome: results of an Italian
multicentre study. Study Group of the Italian
Society of Endocrinology on the Pathophysiology
of the HypothalamicPituitaryAdrenal Axis.
J.Clin. Endocrinol. Metab. 84, 440448 (1999).
Etxabe, J. & Vazquez, J.A. Morbidity
and mortality in Cushings disease:
an epidemiological approach. Clin. Endocrinol.
(Oxf.) 40, 479484 (1994).
Cavagnini, F. & Pecori Giraldi, F. Epidemiology
and follow-up of Cushings disease. Ann.
Endocrinol. (Paris) 62, 168172 (2001).
Newell-Price, J., Trainer, P., Besser, M. &
Grossman, A. The diagnosis and differential
diagnosis of Cushings syndrome and pseudoCushings states. Endocr. Rev. 19, 647672
(1998).
Nieman, L.K. et al. The diagnosis of Cushings
syndrome: an Endocrine Society Clinical Practice
Guideline. J. Clin. Endocrinol. Metab. 93,
15261540 (2008).
Gatta, B. et al. Reevaluation of the combined
dexamethasone suppressioncorticotropin
releasing hormone test for differentiation of mild
Cushings disease from pseudo-Cushings
syndrome. J. Clin. Endocrinol. Metab. 92,
42904293 (2007).
Pecori Giraldi, F. et al. The dexamethasonesuppressed corticotropin-releasing hormone
stimulation test and the desmopressin test to
distinguish Cushings syndrome from pseudoCushings states. Clin. Endocrinol. (Oxf.) 66,
251257 (2007).
Ross, E.J. & Linch, D.C. Cushings syndrome
killing disease: discriminatory value of signs and
symptoms aiding early diagnosis. Lancet 2,
646649 (1982).
Minami, I. et al. Subclinical Cushings disease
with amelioration of metabolic comorbidities
after removal of pituitary tumor. Intern Med. 45,
12311235 (2006).
Nagai, T., Imamura, M., Misumi, S. & Mori, M.
Subclinical Cushings disease accompanied by
malignant hypertension and diabetes mellitus.
Intern Med. 41, 566570 (2002).
Roberge, C. et al. Adrenocortical dysregulation
as a major player in insulin resistance and onset
of obesity. Am. J. Physiol. Endocrinol. Metab. 293,
E1465E1478 (2007).
van Raalte, D.H., Ouwens, D.M. & Diamant, M.
Novel insights into glucocorticoid-mediated
diabetogenic effects: towards expansion of
therapeutic options? Eur. J Clin. Invest. 39,
8193 (2009).
Alberti, K.G. et al. Harmonizing the metabolic
syndrome: a joint interim statement of the
International Diabetes Federation Task Force on

454 | AUGUST 2011 | VOLUME 7

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

31.

32.

33.

34.

35.

36.

Epidemiology and Prevention; National Heart,

Lung, and Blood Institute; American Heart
Association; World Heart Federation;
International Atherosclerosis Society; and
International Association for the Study of
Obesity. Circulation 120, 16401645 (2009).
Qatanani, M. & Lazar, M.A. Mechanisms of
obesity-associated insulin resistance: many
choices on the menu. Genes Dev. 21,
14431455 (2007).
Arnaldi, G., Mancini, T., Polenta, B. & Boscaro, M.
Cardiovascular risk in Cushings syndrome.
Pituitary 7, 253256 (2004).
Faggiano, A. et al. Cardiovascular risk factors
and common carotid artery caliber and stiffness
in patients with Cushings disease during active
disease and 1 year after disease remission. J.
Clin. Endocrinol. Metab. 88,
25272533 (2003).
Mancini, T., Kola, B., Mantero, F., Boscaro, M. &
Arnaldi, G. High cardiovascular risk in patients
with Cushings syndrome according to 1999
WHO/ISH guidelines. Clin. Endocrinol. (Oxf.) 61,
768777 (2004).
Pivonello, R., Faggiano, A., Lombardi, G. &
Colao,A. The metabolic syndrome and
cardiovascular risk in Cushings syndrome.
Endocrinol. Metab. Clin. North Am. 34, 327339,
viii (2005).
Catargi, B. et al. Occult Cushings syndrome in
type-2 diabetes. J. Clin. Endocrinol. Metab. 88,
58085813 (2003).
Chiodini, I. et al. Association of subclinical
hypercortisolism with type 2 diabetes mellitus:
acasecontrol study in hospitalized patients.
Eur. J. Endocrinol. 153, 837844 (2005).
Leibowitz, G. et al. Pre-clinical Cushings
syndrome: an unexpected frequent cause of poor
glycaemic control in obese diabetic patients. Clin.
Endocrinol. (Oxf.) 44, 717722 (1996).
Reimondo, G. et al. Screening of Cushings
syndrome in adult patients with newly diagnosed
diabetes mellitus. Clin. Endocrinol. (Oxf.) 67,
225229 (2007).
Liu, H., Bravata, D.M., Cabaccan, J., Raff, H. &
Ryzen, E. Elevated late-night salivary cortisol
levels in elderly male type 2 diabetic veterans.
Clin. Endocrinol. (Oxf.) 63, 642649 (2005).
Mullan, K. et al. Is there value in routine
screening for Cushings syndrome in patients
with diabetes? J. Clin. Endocrinol. Metab. 95,
22622265 (2010).
Murakami, H. et al. The frequency of type 2
diabetic patients who meet the endocrinological
screening criteria of subclinical Cushings
disease. Endocr. J. 57, 267272 (2010).
Newsome, S. et al. Cushings syndrome in a
clinic population with diabetes. Intern. Med. J.
38, 178182 (2008).
Shaker, J.L. & Lukert, B.P. Osteoporosis
associated with excess glucocorticoids.
Endocrinol. Metab. Clin. North Am. 34, 341356,
viiiix (2005).
Chiodini, I. et al. Subclinical hypercortisolism
among outpatients referred for osteoporosis.
Ann. Intern. Med. 147, 541548 (2007).
Magiakou, M.A., Smyrnaki, P. & Chrousos, G.P.
Hypertension in Cushings syndrome. Best Pract.
Res. Clin. Endocrinol. Metab. 20, 467482
(2006).
Omura, M., Saito, J., Yamaguchi, K., Kakuta, Y. &
Nishikawa, T. Prospective study on the
prevalence of secondary hypertension among
hypertensive patients visiting a general
outpatient clinic in Japan. Hypertens. Res. 27,
193202 (2004).
Anderson, G.H. Jr, Blakeman, N. &
Streeten,D.H. The effect of age on prevalence

37.

38.

39.

40.

41.

42.

43.

44.

45.

46.

47.

48.

49.

50.

51.

52.

53.

54.

of secondary forms of hypertension in 4,429

consecutively referred patients. J. Hypertens. 12,
609615 (1994).
Cooper, M.S. & Stewart, P.M.
11-hydroxysteroid dehydrogenase type 1 and
its role in the hypothalamuspituitaryadrenal
axis, metabolic syndrome, and inflammation.
J.Clin. Endocrinol. Metab. 94, 46454654
(2009).
Kola, B. et al. Changes in adenosine
5'monophosphateactivated protein kinase as a
mechanism of visceral obesity in Cushings
syndrome. J. Clin. Endocrinol. Metab. 93,
49694973 (2008).
Rockall, A.G. et al. Computed tomography
assessment of fat distribution in male and
female patients with Cushings syndrome. Eur. J.
Endocrinol. 149, 561567 (2003).
Ness-Abramof, R. et al. Overnight
dexamethasone suppression test: a reliable
screen for Cushings syndrome in the obese.
Obes. Res. 10, 12171221 (2002).
Tiryakioglu, O. et al. Screening for Cushings
syndrome in obese patients. Clinics (Sao Paulo)
65, 913 (2010).
Baid, S.K. et al. Specificity of screening tests for
Cushings syndrome in an overweight and obese
population. J. Clin. Endocrinol. Metab. 94,
38573864 (2009).
Setji, T.L. & Brown, A.J. Polycystic ovary
syndrome: diagnosis and treatment. Am. J. Med.
120, 128132 (2007).
Kaltsas, G.A. et al. How common are polycystic
ovaries and the polycystic ovarian syndrome in
women with Cushings syndrome? Clin.
Endocrinol. (Oxf.) 53, 493500 (2000).
Fegan, P.G. et al. Cushings syndrome in women
with polycystic ovaries and hyperandrogenism.
Nat. Clin. Pract. Endocrinol. Metab. 3,
778783 (2007).
Glintborg, D. et al. Prevalence of endocrine
diseases and abnormal glucose tolerance tests
in 340 Caucasian premenopausal women with
hirsutism as the referral diagnosis. Fertil. Steril.
82, 15701579 (2004).
Cawood, T.J., Hunt, P.J., OShea, D., Cole, D. &
Soule, S. Recommended evaluation of adrenal
incidentalomas is costly, has high false-positive
rates and confers a risk of fatal cancer that is
similar to the risk of the adrenal lesion becoming
malignant; time for a rethink? Eur. J. Endocrinol.
161, 513527 (2009).
Lib, R. et al. Long-term follow-up study of
patients with adrenal incidentalomas. Eur. J.
Endocrinol. 147, 489494 (2002).
Mantero, F. et al. A survey on adrenal
incidentaloma in Italy. Study Group on Adrenal
Tumors of the Italian Society of Endocrinology.
J.Clin. Endocrinol. Metab. 85, 637644 (2000).
Reincke, M. et al. Preclinical Cushings syndrome
in adrenal incidentalomas: comparison with
adrenal Cushings syndrome. J. Clin. Endocrinol.
Metab. 75, 826832 (1992).
Tabarin, A. et al. Exploration and management of
adrenal incidentalomas. French Society of
Endocrinology Consensus. Ann. Endocrinol.
(Paris) 69, 487500 (2008).
Terzolo, M. et al. Subclinical Cushings syndrome
in adrenal incidentalomas. Endocrinol. Metab.
Clin. North Am. 34, 423439 (2005).
Grumbach, M.M. et al. Management of the
clinically inapparent adrenal mass
(incidentaloma). Ann. Intern. Med. 138,
424429 (2003).
Sackett, D.L., Haynes, R.B., Guyatt, G.H. &
Tugwell, P. Clinical epidemiology. A basic science
for clinical medicine. (Little, Brown and Company,
Boston, 1991).

www.nature.com/nrendo
2011 Macmillan Publishers Limited. All rights reserved

REVIEWS
55. Wilson, J. & Jungner, G. Principles and practice
of screening for disease. Public Health Papers
No. 34. (WHO, Geneva, 1968).
56. Adriaanse, M.C. & Snoek, F.J. The psychological
impact of screening for type 2 diabetes.
Diabetes Metab. Res. Rev. 22, 2025 (2006).
57. Barton, M.B. et al. Decreasing womens
anxieties after abnormal mammograms:
acontrolled trial. J. Natl Cancer Inst. 96,
529538 (2004).
58. Brett, J., Austoker, J. & Ong, G. Do women who
undergo further investigation for breast
screening suffer adverse psychological
consequences? A multi-centre follow-up study
comparing different breast creening result
groups five months after their last breast
screening appointment. J. Public Health Med. 20,
396403 (1998).
59. Taniguchi, T., Hamasaki, A. & Okamoto, M.
Subclinical hypercortisolism in hospitalized
patients with type 2 diabetes mellitus. Endocr. J.
55, 429432 (2008).
60. Gao, S. Combining binomial data using the
logistic normal model. J. Stat. Comput. Simul. 74,
293306 (2004).
61. Clayton, R.N. Mortality in Cushings disease.
Neuroendocrinology 92 (Suppl. 1), 7176
(2010).
62. Hammer, G.D. et al. Transsphenoidal
microsurgery for Cushings disease: initial
outcome and long-term results. J. Clin.
Endocrinol. Metab. 89, 63486357 (2004).
63. Storr, H.L., Chan, L.F., Grossman, A.B. &
Savage, M.O. Paediatric Cushings syndrome:
epidemiology, investigation and therapeutic
advances. Trends Endocrinol. Metab. 18,
167174 (2007).
64. Pecori Giraldi, F., Moro, M. & Cavagnini, F.
Gender-related differences in the presentation
and course of Cushings disease. J. Clin.
Endocrinol. Metab. 88, 15541558 (2003).
65. Wajngot, A., Giacca, A., Grill, V., Vranic, M. &
Efendic, S. The diabetogenic effects of
glucocorticoids are more pronounced in lowthan in high-insulin responders. Proc. Natl Acad.
Sci. USA 89, 60356039 (1992).
66. van Rossum, E.F. & Lamberts, S.W.
Polymorphisms in the glucocorticoid receptor
gene and their associations with metabolic
parameters and body composition. Recent Prog.
Horm. Res. 59, 333357 (2004).
67. Devenport, L., Knehans, A., Sundstrom, A. &
Thomas, T. Corticosterones dual metabolic
actions. Life Sci. 45, 13891396 (1989).
68. Sonino, N., Boscaro, M., Fallo, F. & Fava, G.A.
Aclinical index for rating severity in Cushings
syndrome. Psychother. Psychosom. 69, 216220
(2000).
69. Sarlis, N.J., Chanock, S.J. & Nieman, L.K.
Cortisolemic indices predict severe infections in
Cushings syndrome due to ectopic production of
adrenocorticotropin. J. Clin. Endocrinol. Metab.
85, 4247 (2000).
70. Sonino, N., Fava, G.A., Raffi, A.R., Boscaro, M.
& Fallo, F. Clinical correlates of major depression
in Cushings disease. Psychopathology 31,
302306 (1998).
71. Strack, A.M., Horsley, C.J., Sebastian, R.J.,
Akana, S.F. & Dallman, M.F. Glucocorticoids
and insulin: complex interaction on brown
adipose tissue. Am. J. Physiol. 268,
R1209R1216 (1995).
72. Webb, S.M. et al. Evaluation of health-related
quality of life in patients with Cushings

73.

74.

75.

76.

77.

78.

79.

80.

81.

82.

83.

84.

85.

86.

87.

88.

89.

90.

syndrome with a new questionnaire. Eur. J.

Endocrinol. 158, 623630 (2008).
Garrapa, G.G., Pantanetti, P., Arnaldi, G.,
Mantero, F. & Faloia, E. Body composition and
metabolic features in women with adrenal
incidentaloma or Cushings syndrome. J. Clin.
Endocrinol. Metab. 86, 53015306 (2001).
Tauchmanova, L. et al. Patients with subclinical
Cushings syndrome due to adrenal adenoma
have increased cardiovascular risk. J. Clin.
Endocrinol. Metab. 87, 48724878 (2002).
Terzolo, M. et al. Adrenal incidentaloma: a new
cause of the metabolic syndrome? J. Clin.
Endocrinol. Metab. 87, 9981003 (2002).
Giordano, R. et al. Long-term morphological,
hormonal, and clinical follow-up in a single unit
on 118 patients with adrenal incidentalomas.
Eur. J. Endocrinol. 162, 779785 (2010).
Sereg, M. et al. Atherosclerotic risk factors and
complications in patients with non-functioning
adrenal adenomas treated with or without
adrenalectomy: a long-term follow-up study. Eur.
J. Endocrinol. 160, 647655 (2009).
Barzon, L. et al. Risk factors and long-term
follow-up of adrenal incidentalomas. J. Clin.
Endocrinol. Metab. 84, 520526 (1999).
Bernini, G.P. et al. Long-term morphological and
hormonal follow-up in a single unit on 115
patients with adrenal incidentalomas. Br. J.
Cancer 92, 11041109 (2005).
Fagour, C. et al. Usefulness of adrenal
scintigraphy in the follow-up of adrenocortical
incidentalomas: a prospective multicenter study.
Eur. J. Endocrinol. 160, 257264 (2009).
Elamin, M.B. et al. Accuracy of diagnostic tests
for Cushings syndrome: a systematic review and
metaanalyses. J. Clin. Endocrinol. Metab. 93,
15531562 (2008).
Lijmer, J.G. et al. Empirical evidence of designrelated bias in studies of diagnostic tests. JAMA
282, 10611066 (1999).
Grimes, D.A. & Schulz, K.F. Refining clinical
diagnosis with likelihood ratios. Lancet 365,
15001505 (2005).
Higgins, J.P.T., Thompson, S.G., Deeks, J.J. &
Altman, D.G. Measuring inconsistency in
meta-analyses. BMJ 327, 557560 (2003).
Carroll, T., Raff, H. & Findling, J.W. Late-night
salivary cortisol for the diagnosis of Cushings
syndrome: a meta-analysis. Endocr. Pract. 15,
335342 (2009).
Yanovski, J.A., Cutler, G.B. Jr, Chrousos, G.P. &
Nieman, L.K. Corticotropin-releasing hormone
stimulation following low-dose dexamethasone
administration. A new test to distinguish
Cushings syndrome from pseudo-Cushings
states. JAMA 269, 22322238 (1993).
Oltmanns, K.M. et al. Cortisol correlates with
metabolic disturbances in a population study of
type 2 diabetic patients. Eur. J. Endocrinol. 154,
325331 (2006).
Roy, M.S. et al. The ovine corticotropin-releasing
hormone-stimulation test in typeI diabetic
patients and controls: suggestion of mild chronic
hypercortisolism. Metabolism 42, 696700
(1993).
Tsigos, C., Young, R.J. & White, A. Diabetic
neuropathy is associated with increased activity
of the hypothalamicpituitaryadrenal axis.
J.Clin. Endocrinol. Metab. 76, 554558 (1993).
Papanicolaou, D.A., Yanovski, J.A.,
Cutler,G.B.Jr, Chrousos, G.P. & Nieman, L.K.
Asingle midnight serum cortisol measurement
distinguishes Cushings syndrome from

NATURE REVIEWS | ENDOCRINOLOGY

pseudo-Cushings states. J. Clin. Endocrinol.

Metab. 83, 11631167 (1998).
91. Putignano, P. et al. Midnight salivary cortisol versus
urinary free and midnight serum cortisol as
screening tests for Cushings syndrome. J.Clin.
Endocrinol. Metab. 88, 41534157 (2003).
92. Raff, H., Raff, J.L. & Findling, J.W. Late-night
salivary cortisol as a screening test for
Cushings syndrome. J. Clin. Endocrinol. Metab.
83, 26812686 (1998).
93. Masserini, B. et al. The limited role of midnight
salivary cortisol levels in the diagnosis of
subclinical hypercortisolism in patients with
adrenal incidentaloma. Eur. J. Endocrinol. 160,
8792 (2009).
94. Nunes, M.L. et al. Late-night salivary cortisol for
diagnosis of overt and subclinical Cushings
syndrome in hospitalized and ambulatory
patients. J. Clin. Endocrinol. Metab. 94, 456462
(2009).
95. Streiner, D.L. & Norman, G.R. Health
measurement scales. A practical guide to their
development and use (Oxford Medical
Publications, Oxford, 1995).
96. Stewart, P.M. Is subclinical Cushings syndrome
an entity or a statistical fallout from diagnostic
testing? Consensus surrounding the diagnosis is
required before optimal treatment can be defined.
J. Clin. Endocrinol. Metab. 95, 26182620 (2010).
97. Vassilatou, E. et al. Hormonal activity of adrenal
incidentalomas: results from a long-term
follow-up study. Clin. Endocrinol. (Oxf.) 70,
674679 (2009).
98. Eller-Vainicher, C. et al. Post-surgical
hypocortisolism after removal of an adrenal
incidentaloma: is it predictable by an accurate
endocrinological work-up before surgery? Eur. J.
Endocrinol. 162, 9199 (2010).
99. Gross, K.L. & Cidlowski, J.A. Tissue-specific
glucocorticoid action: a family affair. Trends
Endocrinol. Metab. 19, 331339 (2008).
100. Chiodini, I. et al. Beneficial metabolic effects of
prompt surgical treatment in patients with an
adrenal incidentaloma causing biochemical
hypercortisolism. J. Clin. Endocrinol. Metab. 95,
27362745 (2010).
101. Guerrieri, M. et al. Primary adrenal
hypercortisolism: minimally invasive surgical
treatment or medical therapy? A retrospective
study with long-term follow-up evaluation. Surg.
Endosc. 24, 25422546 (2010).
102. Toniato, A. et al. Surgical versus conservative
management for subclinical Cushings syndrome
in adrenal incidentalomas: a prospective
randomized study. Ann. Surg. 249, 388391
(2009).
103. Barahona, M.J., Resmini, E., Sucunza, N. &
Webb, S.M. Diagnosis of cure in Cushings
syndrome: lessons from long-term follow-up.
Front. Horm. Res. 38, 152157 (2010).
104. Fagan, T.J. Letter: Nomogram for Bayes
theorem. N. Engl. J. Med. 293, 257 (1975).
Acknowledgments
The authors are indebted to V. Bouteloup (INSERM
U897, University of Bordeaux 2, France) for statistical
analysis.
Author contributions
Both authors researched the data for the article,
provided a substantial contribution to discussions
of the content, contributed equally to writing the
article and reviewed and/or edited the manuscript
before submission.

VOLUME 7 | AUGUST 2011 | 455

2011 Macmillan Publishers Limited. All rights reserved