J Pediatr Endocrinol Metab 2017; aop
Review
Kate C. Verbeeten and Alexandra H. Ahmet*
The role of corticosteroid-binding globulin in the
evaluation of adrenal insufficiency
https://doi.org/10.1515/jpem-2017-0270
Received July 13, 2017; accepted November 1, 2017
Abstract: Cortisol is a hydrophobic molecule that is largely
bound to corticosteroid-binding globulin (CBG) in the cir-
culation. In the assessment of adrenal insufficiency, many
clinicians measure a total serum cortisol level, which
assumes that CBG is present in normal concentrations and
with a normal binding affinity for cortisol. CBG concen-
tration and affinity are affected by a number of common
factors including oral contraceptive pills (OCPs), fever and
infection, as well as rare mutations in the serine protease
inhibitor A6 (SERPINA6) gene, and as such, total cortisol
levels might not be the ideal way to assess adrenal function
in all clinical circumstances. This paper reviews the limi-
tations of immunoassay and liquid chromatography-tan-
dem mass spectrometry (LC-MS/MS) in the measurement
of total cortisol, the challenges of measuring free serum
cortisol directly as well as the difficulties in calculating an
estimated free cortisol from total cortisol, CBG and albu-
min concentrations. Newer approaches to the evaluation
of adrenal insufficiency, including the measurement of
cortisol and cortisone in the saliva, are discussed and a
possible future role for these tests is proposed.
Keywords: adrenal insufficiency; corticosteroid-binding
globulin; free cortisol; salivary cortisol; salivary cortisone;
SERPINA6.
Introduction
In pediatric endocrinology, clinical evaluation of the
hypothalamic-pituitary-adrenal (HPA) axis relies heavily
on the measurement of a total cortisol level [1]. In this paper
*Corresponding author: Dr. Alexandra H. Ahmet, Assistant Professor
of Pediatrics, Children’s Hospital of Eastern Ontario, Division of
Endocrinology and Metabolism, 401 Smyth Road, Ottawa, Ontario,
K1H 8L1, Canada, Phone: +613-737-7600, ext. 3939,
Fax: +613-738-4236, E-mail: aahmet@cheo.on.ca
Kate C. Verbeeten: Children’s Hospital of Eastern Ontario, Ottawa,
Canada
we describe the challenges of evaluating the HPA axis,
including the common use of total serum cortisol rather
than the measurement or calculation of the biologically
active free cortisol. We argue that total serum cortisol may
not accurately reflect the complex balance between the
different elements of the HPA axis and, as such, might not
be the ideal way to assess adrenal function in all clinical
circumstances.
Cortisol physiology
Only a small fraction of total serum cortisol is unbound
and free to enter cells, where it interacts with glucocor-
ticoid receptors, provides feedback inhibition in the
hypothalamus and pituitary gland and is ultimately
responsible for crucial functions such as controlling
inflammation and ensuring normotension and euglyce-
mia [2]. This delicate system is influenced by the intrin-
sic pulsatility of corticotropin-releasing hormone (CRH)
and adrenocorticotropic hormone (ACTH)-secreting cells,
diurnal variation, physical and psychological stress,
inflammatory cytokines, medications and genetic factors
affecting receptors, binding proteins and adrenal enzymes
[2, 3]. Intricate local fluctuations in cortisol concentration
are essential for maintaining homeostasis in the body,
through mechanisms that are still not entirely understood.
Cortisol physiology in the systemic
circulation
Cortisol is a hydrophobic molecule that binds to protein
transport molecules in the circulation [1]. Eighty to 90%
of circulating cortisol is bound to corticosteroid-binding
globulin (CBG), a highly conserved 50–60  kDa glyco-
protein encoded on the serine protease inhibitor A6
(SERPINA6) gene on chromosome 14q32.1 [4]. CBG binds
cortisol with high affinity and low capacity; there is only
one cortisol binding site on each CBG molecule [3]. CBG is
saturated when the total cortisol concentration in serum
is 400–500  nmol/L [3]. Ten to 15% of cortisol is bound,
Brought to you by | UCL - University College London
Authenticated
Download Date | 12/31/17 11:36 AM
2      Verbeeten and Ahmet: CBG and the evaluation of adrenal insufficiency
with low affinity, to albumin which is abundant in the
circulation and essentially non-saturable. In the steady
state, approximately 5% of cortisol is unbound [3]. Recent
work has emphasized the “dual role” of CBG as both a res-
ervoir for cortisol and a modulator of cortisol release [5].
CBG binds cortisol when it is synthesized by the adrenal
glands and acts as a buffer to prevent large fluctuations in
free cortisol levels [1]. It directs local free cortisol concen-
tration through changes in CBG-cortisol binding affinity
[6]. The presence of neutrophil elastase, produced by neu-
trophils and macrophages at sites of inflammation, causes
CBG to undergo a conformational change resulting in a
much lower affinity for cortisol [3, 4]. In the steady state,
there is a balance between the two forms, with 30%–35%
of CBG in the low-affinity form [7].
Table 1 lists many of the biological factors that affect
CBG levels in the human body. The half-life of CBG in the
blood is approximately 5  days [4]; however this varies
widely depending on factors such as temperature and gly-
cosylation. The relatively long half-life means that factors
that affect the synthesis and secretion of CBG are only sig-
nificant if they are prolonged [1], while factors affecting
cortisol affinity have an immediate effect. In addition to the
presence of inflammation, increases in temperature result
in a significant decrease in CBG’s affinity for cortisol [3],
leading to an increased proportion of free cortisol. Glyco-
sylation, a form of post-translational modification in which
carbohydrate molecules are attached to specific sites on
protein molecules in an enzyme-dependent fashion, has
been shown, in vitro, to be influenced by the presence of
certain hormones [8]. Several glycoforms of CBG have been
identified [8, 13]. Glycosylation has been demonstrated to
affect CBG secretion, its affinity for cortisol, its half-life in
Table 1: Biological factors that affect CBG levels and affinity for cortisol [1, 6, 8–12].
Factors that increase   Factors that increase  
CBG concentration CBG’s affinity for cortisol
Oral estrogen/OCP   Glycosylation  
Pregnancy    
SERMs    
Mitotane    
Evening    
   
   
   
   
   
   
   
SERM, selective estrogen receptor modulator; IGF-1, insulin-like growth factor 1.
the circulation and direction to specific tissues [7]. The
degree of glycosylation of CBG has been found to interact
with the variables of temperature and conformation to
determine CBG’s affinity for cortisol in a given situation [6].
The free hormone hypothesis [14] states that the biolog-
ical activity of a hormone is affected by its unbound (free)
rather than protein-bound concentration in the plasma.
There is more recent evidence that CBG itself has a biologi-
cally active role by binding to CBG receptors, resulting in
both targeted delivery of cortisol to specific tissues and pos-
sibly second-messenger effects as well [13, 15]. It is hypoth-
esized that some CBG receptors might bind only specific
glycoforms of CBG [8, 13, 16]. The exact role of CBG recep-
tors is not completely clear; the most recent literature sug-
gests that free cortisol is the main determinant of cortisol
effects, with CBG itself playing a more minor role [7]. The
fact that null mutations in the SERPINA6 gene are not lethal
demonstrates that these interactions are not as important
as the effects of free cortisol, however they represent a
poorly understood aspect of glucocorticoid physiology [15].
Mutations in SERPINA6
Nine SERPINA6 mutations have been identified in humans
[7, 17], with varying effects depending on the exact genetic
locus involved. Some mutations result in low or absent
circulating CBG, and some mutations affect only the cor-
tisol-binding site, resulting in normal CBG concentrations
but poor cortisol carrying capacity [1]. Some SERPINA6
polymorphisms have no clinical significance but can
complicate the biochemical picture; Simard et  al. [17]
reported a CBG variant which was not identified by the
Factors that decrease CBG   Factors that decrease
CBG’s affinity for cortisol
Cirrhosis   Inflammation
Ethanol   High temperature
IL-6   Ethanol
Insulin   SERPINA6 mutations
IGF-1  
Hyperthyroidism  
Exogenous glucocorticoids  
Cushing syndrome  
Nephrotic syndrome  
SERPINA6 mutations  
Neutrophil elastase  
Morning  
Brought to you by | UCL - University College London
Authenticated
Download Date | 12/31/17 11:36 AM
 Verbeeten and Ahmet: CBG and the evaluation of adrenal insufficiency      3
CBG immunoassay, but was otherwise functional. Approxi-
mately 1 in 35 individuals in the Han Chinese population
have a SERPINA6 variant leading to a 50% reduction in
CBG levels [7]. These individuals have normal free cortisol
levels and normal ACTH, but a low total cortisol concentra-
tion. In other ethnic groups, these mutations are rare [17].
Free cortisol fractions reported in unstressed adults
are typically close to 5% [1, 7]. Lewis et al. [9] reported a
free cortisol fraction of 25% in an unstressed individual
with a null SERPINA6 mutation; this was the same free
cortisol fraction noted in heat-inactivated plasma of
normal volunteers [9]. Animal studies suggest that indi-
viduals with SERPINA6 mutations may also respond dif-
ferently to critical illness. CBG knockout mice that are
injected with lipopolysaccharide [5] (an animal model of
septic shock) or tumor necrosis factor-alpha (TNF-α) [17]
have considerably lower survival than wild-type mice. It
seems that the larger free fraction of cortisol does not fully
compensate for the lower cortisol pool.
Some individuals with SERPINA6 mutations present
with hypotension or nonspecific symptoms such as
fatigue, nausea or chronic pain, but symptomatology
varies widely, even between family members with
identical mutations [10, 15]. Proposed explanations for
the symptoms experienced by some of these individuals
include dysregulation of cortisol pulsatility leading to
changes in transcription of glucocorticoid-responsive
genes [1] or epigenetic changes [17], and loss of the tissue-
specific targeting and second-messenger effects that occur
when CBG binds its receptors [15].
Methods for measuring cortisol
The two main methods for measuring cortisol, both total
and free, are immunoassays and liquid chromatography-
tandem mass spectrometry (LC-MS/MS). Immunoassay is
currently the most popular method, given its speed, lower
cost and ease of use [18]. There are important limitations
of this technology, however, that must be understood to
ensure appropriate interpretation of results [1, 19–21]. The
main issue, affecting the measurement of both total and
free cortisol, is imperfect antibody specificity. The anti-
bodies used in these assays may cross-react with cortisol
precursors or metabolites, as well as synthetic steroids.
Because of this cross-reactivity, cortisol levels by immuno­
assay are usually higher than those obtained by other
methods [1]. These molecules are easily distinguished by
mass spectrometry [22]. The second issue, which affects
the measurement of total cortisol, involves the step in
which the cortisol molecule is removed from CBG. This
process can involve a change in temperature or pH or the
addition of a reagent [21]. If a reagent is used, then it must
be added in sufficient quantities to allow for high-CBG
states, such as pregnancy or oral contraceptive use, or cor-
tisol levels will be underestimated [21, 23]. The third and
very significant issue is that both the antibodies and the
methods for removing cortisol from CBG in immunoassays
are proprietary and can change over time [24], so even a
patient followed at the same institution may have multi-
ple results that are not directly comparable. These incon-
sistencies complicate the application of the literature to
clinical practice [18, 22, 25] as the methods used to derive
clinical guidelines may be different than those available
at the institution attempting to implement the guidelines.
Many authors [1, 19, 22, 24, 26] have argued that LC-MS/
MS should be the gold standard for quantifying all steroid
hormones. If proper quality control measures are followed,
mass spectrometry allows steroid hormones to be meas-
ured accurately and reproducibly, even in tiny concentra-
tions, without the issue of antibody specificity [27]. A clear
advantage of mass spectrometry is the ability to measure
a concurrent steroid profile, which is useful in conditions
such as polycystic ovary syndrome (PCOS) and congenital
adrenal hyperplasia (CAH) which involve aberrations in
multiple steroids [22]. LC-MS/MS can screen for exogenous
steroids (prescribed or not) [22], which can sometimes
confuse the clinical picture. Hawley et al. [21] compared the
performance of five commercially available cortisol immu-
noassays with an LC-MS/MS candidate reference manage-
ment procedure [27] in five patient populations. They found
considerable variation between immunoassay results and
LC-MS/MS results with a greater discrepancy in both preg-
nant and non-pregnant women compared to men [21]. They
found that all immunoassays they tested (though less on
the newer assays) overestimated cortisol levels in patients
taking metyrapone and prednisolone [21]. While LC-MS/
MS is a superior technique to immunoassay in many ways,
a clear limitation is the expense and the special train-
ing required for laboratory personnel to maintain quality
control [20]. Certainly in the short term, it is realistic that
many clinics will continue to use immunoassays as the
main method for measuring serum cortisol levels. It is
therefore crucial that individuals using this method under-
stand the assumptions and limitations of their assay.
Methods for measuring serum free cortisol
directly
The direct measurement of serum free cortisol involves
first separating the cortisol that is bound to CBG or
Brought to you by | UCL - University College London
Authenticated
Download Date | 12/31/17 11:36 AM
4      Verbeeten and Ahmet: CBG and the evaluation of adrenal insufficiency
albumin from the plasma containing free cortisol, fol-
lowed by quantification of free cortisol by immunoassay
or LC-MS/MS as already described. The most common
techniques used to separate bound cortisol from free
cortisol in the serum are ultrafiltration and equilibrium
dialysis [28]. Equilibrium dialysis involves incubating the
plasma in a cell with a membrane that CBG is not able to
cross, often overnight, while ultrafiltration involves cen-
trifugation of plasma samples in a tube fitted with a filter
with pores smaller than CBG molecules. In all of the tech-
niques, controlling temperature is extremely important
as small fluctuations cause the total:free cortisol ratio
to change dramatically. Several authors have noted that
equilibrium dialysis provides slightly higher results than
ultrafiltration, with inferior reproducibility and a longer
incubation time [28–30].
Measuring serum free cortisol indirectly
Several formulae for estimating free cortisol concentra-
tions from the more easily measured values of total cor-
tisol, CBG and albumin have been published. The most
straightforward is the free cortisol index, a ratio of total
cortisol to CBG [31–33]. While simple and somewhat useful,
this method does not consider the saturability of CBG in
high-cortisol states [11] and the changes in binding affin-
ity due to fever, infection, pregnancy and some SERPINA6
mutations [17]. Concerns with this approach led Coolens
et al. [11] to develop a more complicated formula to derive
free cortisol concentration when the concentrations of
total cortisol and CBG are known. Compared to the free
cortisol index, Coolens’ formula is a more accurate repre-
sentation of free cortisol levels when CBG is saturated [11],
but it does not account for changes in albumin concentra-
tion or CBG affinity [8] and has been demonstrated to have
low precision in patients with sepsis [30].
Dorin et  al. [34] attempted to improve the accuracy
of Coolens’ method by including albumin concentration
and the equilibrium dissociation constant for cortisol
binding to albumin as independent variables, rather than
incorporating them into a constant. They tested their
formula on three populations: healthy controls, patients
with sepsis and patients with septic shock. They found
that their formula was less biased than Coolens’ formula,
and it was better able to model the interaction effect
observed when both CBG and albumin were low [34],
as observed in profound sepsis. Dorin et  al. [34] noted
that there was no correlation between albumin and CBG
levels, indicating that these proteins change in response
to different factors. The Dorin formula oversimplifies
some aspects of cortisol dynamics, including the inaccu-
rate assumption that an albumin molecule can only bind
one cortisol molecule, and the use of a single dissocia-
tion constant for CBG, while physiologically the affinity
of CBG to cortisol is significantly lower in sepsis [34].
Nguyen et  al. [35] developed a formula that addressed
some of the limitations of the Dorin formula; however,
the tests required (separate immunoassays for the high-
and low-affinity forms of CBG) were less practical than
measuring free cortisol directly.
The limitations of indirect methods in general are the
oversimplification of a highly complex system as well as
the incorporation of multiple sources of error, both from
the methods used to measure cortisol, CBG and albumin
in the patient but also the methods used to derive the for-
mulae [29]. For a calculated free cortisol value to be useful
clinically, it must be accurate, reproducible, and easier
and less expensive than measuring free cortisol directly.
To maximize precision, the formula and its constants
should be derived from free cortisol values measured by
ultrafiltration followed by LC-MS/MS. For accuracy in
sepsis, a measure of albumin should be included. Con-
sideration of different dissociation constants for different
clinical situations is worth investigating.
Measuring free cortisol and cortisone in the
saliva
Given the technical challenges of separating bound and
unbound cortisol in serum, many researchers have con-
sidered measuring free cortisol in parts of the body where
CBG and albumin are not present – in the saliva and urine.
Salivary cortisol is widely used in psychological research
[36, 37]. As with serum cortisol, salivary cortisol can be
measured by immunoassay or LC-MS/MS [38]. The anti-
body specificity of immunoassays may be problematic in
saliva due to the presence of other steroids and steroid
metabolites [38]. As with serum measurements, LC-MS/
MS reference ranges for cortisol in saliva are lower than
for immunoassays, reflecting the superior specificity
of LC-MS/MS [36, 39]. Measuring free cortisol in saliva
instead of in serum offers many potential advantages,
particularly in children. The stress associated with phle-
botomy is removed, and samples can be collected at home
at any time of day, allowing the early-morning cortisol
peak and late-night nadir to be captured. Samples are
stable at room temperature for 1–2 days [38] and longer if
refrigerated. The main challenge is ensuring standardiza-
tion of collection, particularly if the test is done at home.
The kit must be used as directed and avoidance of eating,
Brought to you by | UCL - University College London
Authenticated
Download Date | 12/31/17 11:36 AM
 Verbeeten and Ahmet: CBG and the evaluation of adrenal insufficiency      5
drinking, smoking and tooth brushing prior to the test is
key. Patients with xerostomia, intraoral bleeding or recent
dental work will have inaccurate results [37]. However,
similar to serum cortisol levels, non-stimulated salivary
cortisol levels may only be useful as a screening tool [40]
and are less accurate in individuals without an estab-
lished circadian rhythm [23], such as infants and shift
workers, and during intercurrent illness.
Lipophilic unbound cortisol in the serum diffuses
passively into the parotid glands and across the epithelial
membrane into saliva [1]. Salivary cortisol reflects levels of
serum free cortisol and shows the same circadian rhythm
as serum free cortisol [38]; cortisol in saliva increases
within minutes of a rise in serum free cortisol [23]. The
parotid glands produce 11β-hydroxysteroid dehydrogenase
type 2 (11β-HSD2), which converts cortisol to cortisone such
that levels of cortisone in the saliva are 4 to 6 times higher
than in salivary cortisol [1, 41]. Salivary cortisone measured
by LC-MS/MS may be a more accurate proxy for serum free
cortisol than salivary cortisol as it is detectable even when
serum cortisol levels are low and avoids potential contami-
nation with oral hydrocortisone [42–44].
Measuring free cortisol in urine
Urinary free cortisol (UFC) has long been used in the diagno-
sis of Cushing syndrome in adults. UFC correlates with free
serum cortisol and, like salivary cortisol and cortisone, is
more closely associated with cortisol production than total
serum cortisol levels [38]. While CBG is not a confounding
factor in the measurement of UFC, there are several limi-
tations of this test including day-to-day variation, several
common conditions that affect UFC (obesity, pregnancy
and stress) and higher levels of cross-reactive metabolites
in urine relative to serum [38, 41, 45]. Despite the limitations
of the test, UFC continues to be a useful tool in the evalua-
tion of cortisol excess; however, it has not been established
as a test for the evaluation of adrenal insufficiency.
Free cortisol in the diagnosis of
adrenal insufficiency
Free cortisol in the diagnosis of primary,
secondary and tertiary adrenal insufficiency
The limitations of measurement of total cortisol are typi-
cally less important in the evaluation of primary adrenal
insufficiency when ACTH levels are elevated. However,
the diagnosis of secondary or tertiary adrenal insuffi-
ciency depends on a low stimulated cortisol measurement
in the context of a low or inappropriately normal ACTH
level. Low total cortisol and normal ACTH are also seen in
individuals with low CBG levels, due to SERPINA6 poly-
morphisms or concurrent medical conditions that lower
CBG (Table 1). If free cortisol levels are not measured, sec-
ondary or tertiary adrenal insufficiency may be overdiag-
nosed when factors that decrease CBG concentrations are
present (i.e. critical illness or SERPINA6 polymorphisms),
and may be missed if factors that elevate CBG are present
(i.e. oral contraceptive pills [OCPs]). Symptoms of adrenal
insufficiency are nonspecific; therefore, if there is poor
clinical response to glucocorticoid therapy in an individ-
ual with presumed central adrenal insufficiency, the pos-
sibility of a false-positive test secondary to low CBG levels
should be considered.
Clinical guidelines on the diagnosis of Cushing syn-
drome indicate that salivary cortisol levels measured
during the late-night nadir may be a useful diagnostic
tool in the evaluation of hypercortisolism [23, 45, 46], and
recent research has suggested that stimulated salivary cor-
tisol and cortisone levels may be used as an alternative to
serum cortisol in the evaluation of adrenal insufficiency
as well [42, 47]. Similar to serum cortisol, early-morning
basal salivary cortisol levels may only be useful as a
screening tool for adrenal insufficiency in patients with
a diurnal rhythm [40]. While there is emerging research
evaluating the use of basal and stimulated salivary corti-
sol levels as diagnostic tools, to date, there are no widely
supported reference ranges for these tests and more recent
evidence suggests that stimulated salivary cortisone may
be a superior measure [40]. For use in an endocrinol-
ogy clinic, reference ranges following ACTH stimulation
testing would be of greater use. Table 2 summarizes the
advantages and disadvantages of dynamic tests that could
be used in the evaluation of adrenal insufficiency.
Establishing a pediatric reference range for
free cortisol
Establishing reference ranges for cortisol is more compli-
cated than for other hormones, due to diurnal variation
and the association with stress, including the stress of
phlebotomy. To overcome this variability, the diagnosis
of adrenal insufficiency relies on dynamic testing. Eyal
et al. [48] provide an important first step in establishing
reference ranges for stimulated free cortisol levels in chil-
dren. They measured free cortisol levels by equilibrium
Brought to you by | UCL - University College London
Authenticated
Download Date | 12/31/17 11:36 AM
6      Verbeeten and Ahmet: CBG and the evaluation of adrenal insufficiency
Table 2: Summary of dynamic tests that could be used for the evaluation of adrenal insufficiency.
Testa   Laboratory  
requirements
Stimulated total serum   LC-MS/MS or  
cortisol immunoassay
Stimulated free   Ultrafiltration or  
serum cortisol (direct equilibrium dialysis
measurement) followed by LC-MS/MS
or immunoassay
Stimulated free serum   Measurement of total  
cortisol (calculated) cortisol, CBG and
albumin
Stimulated salivary free   LC-MS/MS  
cortisol
Stimulated salivary free   LC-MS/MS  
cortisone
a
Of note, sex, age and Tanner-stage-specific reference ranges based on LC-MS/MS values have not been established for these tests and this
remains a major limitation for all.
dialysis, followed by chemiluminescence assay, in 28
girls and 57 boys aged 0.6–17.7  years before and after a
250  µg/m2 ACTH stimulation test [48]. The mean basal
free cortisol level was 11.1 (±8.3) nmol/L and the mean
peak free cortisol level was 50.0 (±16.7) nmol/L [48]. At
baseline, they noted a mean free fraction of 3.95% (95%
confidence interval [CI] 3.57%–4.33%) and after stimula-
tion testing they found a mean free fraction of 6.69% (95%
CI 6.23%–7.13%). The increase in free cortisol levels after
stimulation testing was greater than the increase in total
cortisol levels, consistent with saturation of CBG [48]. The
authors concluded that an appropriate cutoff for peak free
cortisol following a standard-dose ACTH stimulation test
in children is >25  nmol/L [48]. This was the same cutoff
that was determined by similar studies in adults [49, 50].
Eyal et  al. [48] noted significant differences in peak free
cortisol levels by Tanner stage for both genders, but no
difference between genders. Although the sample size
in each age/Tanner stage group was small and the study
did not use LC-MS/MS to measure cortisol, this study pro-
vides a useful starting point and is, to our knowledge, the
Indications and   Contraindications and disadvantages
advantages
– Most commonly used   –M
 isleading if CBG levels are abnormal
– Guidelines and –C
 osts of equipment and training laboratory
reference ranges personnel (LC-MS/MS)
available for –P
 oor specificity of antibodies (immunoassay)
immunoassay
–U
 seful if CBG levels   – Time-consuming
are abnormal – Special training required
– Expensive
– Useful if CBG levels   –C  BG assay may not be available
are abnormal – L ess accurate in critical illness
–C
 an be used in   – May be contaminated by oral hydrocortisone
patients with poor – Strict rules must be followed about eating,
venous access drinking, smoking and dental care
–U
 seful if CBG levels – Xerostomia and oral bleeding are
are abnormal contraindications
– Abnormalities in 11β-HSD2 may give
inaccurate results
– Can be used in   – Strict rules must be followed about eating,
patients with poor drinking, smoking and dental care
venous access – Xerostomia and oral bleeding are
– Useful if CBG levels contraindications
are abnormal –A  bnormalities in 11β-HSD2 may give
inaccurate results
only study that attempts to determine reference ranges for
serum free cortisol in children.
The use of stimulated salivary cortisone levels by
LC-MS/MS in the evaluation of cortisol deficiency, espe-
cially in states of altered cortisol binding, may address
many of the challenges with testing of the HPA axis;
however, further study is needed to establish pediat-
ric reference ranges. Several studies have attempted to
develop age, sex and time of day-specific reference ranges
for salivary free cortisol levels, notably the CIRCORT
database [36], a meta-dataset made up of 15  studies
that included 104,623  saliva samples from 18,698 par-
ticipants. Though salivary cortisol was measured on two
types of immunoassay in this study, the results were cali-
brated to LC-MS/MS in order to increase generalizability
[36]. These results did not take into account the conver-
sion to cortisone. Two recent studies [47, 51] published
stimulated salivary cortisol and cortisone thresholds for
the diagnosis of adrenal insufficiency in adults using
LC-MS/MS after low- [51] and standard-dose [47] ACTH
stimulation tests. Attempts to establish pediatric norms
Brought to you by | UCL - University College London
Authenticated
Download Date | 12/31/17 11:36 AM
 Verbeeten and Ahmet: CBG and the evaluation of adrenal insufficiency      7
for stimulated salivary cortisol levels have also been
made in smaller studies [52, 53].
Free cortisol in the diagnosis of critical
illness-related corticosteroid insufficiency
(CIRCI)
Levels of free cortisol are significantly affected by criti-
cal illness through two main mechanisms: activation of
the HPA axis and changes in CBG binding affinity due to
elevated temperature and cleavage by neutrophil elastase.
Free serum cortisol assays performed at 37 °C may under-
estimate levels in febrile patients [1]. Interleukin-6 (IL-6)
inhibits transcription of SERPINA6 and CBG secretion
by hepatocytes [1], leading to higher free cortisol in situ-
ations of extreme inflammation. While albumin is less
affected by temperature changes compared to CBG, it is
highly sensitive to acidosis [54]; therefore in the context of
septic shock, cortisol is released from both of its binding
proteins. The low CBG and resultant free cortisol levels in
sepsis mimic a SERPINA6 polymorphism, with high base-
line free cortisol levels and increased pulsatility due to
lack of buffering [1].
It is an unresolved debate in critical care medicine as to
how and when corticosteroids should be given to patients
in septic shock [55, 56]. This discussion is beyond the scope
of this paper, other than to highlight the importance of
considering (and ideally measuring) free cortisol in studies
evaluating the role of glucocorticoids in critical illness, as
nearly all aspects of cortisol physiology, including ACTH,
CBG cleavage, CBG transcription, CBG binding affinity and
albumin binding affinity, are profoundly affected by sepsis.
This is the patient group in whom indirect measures of free
cortisol are most likely to be inaccurate; however, direct
measurements of free cortisol would have to be available
very quickly to be useful clinically. Measuring salivary cor-
tisol in critically ill patients has been explored in the litera-
ture and research on this subject is ongoing [57].
Conclusions
In this paper, we argue that direct measurement of serum
free cortisol, by ultrafiltration followed by LC-MS/MS, is
the most accurate way to quantify the biologically active
free fraction of cortisol and address many of the chal-
lenges that we face in the evaluation of the HPA axis.
Although significantly more research would be required,
the measurement of salivary cortisol and/or cortisone
could also become a useful and convenient approach.
We argue that reference ranges should be established
using LC-MS/MS, and should be post-stimulation test
and specific for sex and age (early childhood) or Tanner
stage (from middle childhood to maturity) [48, 53, 58].
Currently, many of us rely on immunoassays, total serum
cortisol levels and non-standardized reference ranges for
the purpose of clinical decision-making. We hope to raise
awareness of the limitations of our current approaches
and to highlight areas where further research would be
helpful.
Total cortisol might still be the most cost-effective
measurement for evaluating the HPA axis in many clinical
situations, however, consideration of free cortisol meas-
urement, either directly in serum or saliva, or indirectly
using calculated values, is useful in more complicated sit-
uations when cortisol levels do not explain the patient’s
clinical presentation. Further investigation may also be
helpful when faced with a patient with borderline results
after ACTH stimulation testing or in those with underlying
conditions or medications that affect binding proteins,
including critical illness or use of oral contraceptives. The
possibility of abnormal CBG binding must also be consid-
ered if results are unexpected, as calculated free cortisol
formulae do not allow for changes in CBG’s affinity for
cortisol. As Vogeser et al. [29] suggest, perhaps providing
context for the cortisol level, such as a concurrent C-reac-
tive protein level to diagnose an acute-phase response,
could help decide if cortisol-CBG binding is likely to be
altered. While the current literature suggests that free cor-
tisol is the main determinant of cortisol effects, the role
of CBG on cortisol action and the degree of correlation
between serum free cortisol levels and clinical outcomes
require further study [7, 9]. Unfortunately, the currently
available reference ranges for measured and calculated
free cortisol levels are not sufficient to make specific rec-
ommendations, but at minimum an understanding of the
importance of considering free vs. total cortisol is impor-
tant for the clinician.
Author contributions: All the authors have accepted
responsibility for the entire content of this submitted
manuscript and approved submission.
Research funding: KCV was supported by the Canadian
Pediatric Endocrine Group fellowship program and the
Children’s Hospital Academic Medical Organization.
Honorarium: Neither author received an honorarium for
the preparation of this manuscript.
Competing interests: The funding organization played
no role in the writing of this report or in the decision to
submit the report for publication.
Brought to you by | UCL - University College London
Authenticated
Download Date | 12/31/17 11:36 AM
8      Verbeeten and Ahmet: CBG and the evaluation of adrenal insufficiency
References
1. Perogamvros I, Ray D, Trainer P. Regulation of cortisol bioavail-
ability – effects on hormone measurement and action. Nat Rev
Endocrinol 2012;8:717–27.
2. Sarafoglou K, Harrington KD, Bockting WO. Congenital adrenal
hyperplasia. In: Sarafoglou K, Hoffmann GF, Roth KS, editors.
Pediatric endocrinology and inborn errors of metabolism.
­New York: McGraw Hill Medical, 2009:385–409.
3. Henley DE, Lightman SL. New insights into corticosteroid-bind-
ing globulin and glucocorticoid delivery. Neuroscience Elsevier
Inc 2011;180:1–8.
4. Lin HY, Muller YA, Hammond GL. Molecular and structural basis
of steroid hormone binding and release from corticosteroid-
binding globulin. Mol Cell Endocrinol 2010;316:3–12.
5. Moisan MP, Minni AM, Dominguez G, Helbling JC, Foury A, et al.
Role of corticosteroid binding globulin in the fast actions of glu-
cocorticoids on the brain. Steroids Elsevier Inc 2014;81:109–15.
6. Chan WL, Carrell RW, Zhou A, Read RJ. How changes in affinity of
corticosteroid-binding globulin modulate free cortisol concen-
tration. J Clin Endocrinol Metab 2013;98:3315–22.
7. Meyer E, Nenke M, Rankin W, Lewis J, Torpy D. Corticosteroid-
binding globulin: a review of basic and clinical advances. Horm
Metab Res 2016;48:359–71.
8. Mihrshahi R, Lewis J, Ali S. Hormonal effects on the secretion
and glycoform profile of corticosteroid-binding globulin.
J Steroid Biochem Mol Biol 2006;101:275–85.
9. Lewis J, Bagley C, Elder P, Bachmann A, Torpy D. Plasma free cor-
tisol fraction reflects levels of functioning corticosteroid-binding
globulin. Clin Chim Acta 2005;359:189–94.
10. Cizza G, Rother KI. Cortisol binding globulin: more than just a
carrier? J Clin Endocrinol Metab 2012;97:77–80.
11. Coolens J, Van Baelen H, Heyns W. Clinical use of unbound
plasma cortisol as calculated from total cortisol and corticoster-
oid-binding globulin. J Steroid Biochem 1987;26:197–202.
12. Tschop M, Lahner H, Feldmeier H, Grasberger H, Morrison K,
et al. Effects of growth hormone replacement therapy on levels
of cortisol and cortisol-binding globulin in hypopituitary adults.
Eur J Endocrinol 2000;143:769–73.
13. Sumer-Bayraktar Z, Kolarich D, Campbell M, Ali S, Packer N,
et al. N-glycans modulate the function of human corticosteroid-
binding globulin. Mol Cell Proteomics 2011;10:1–14.
14. Mendel C. The free hormone hypothesis: a physiologically based
mathematical model. Endocr Rev 1989;10:232–74.
15. Gagliardi L, Ho J, Torpy D. Corticosteroid-binding globulin: the
clinical significance of altered levels and heritable mutations.
Mol Cell Endocrinol 2010;316:24–34.
16. Strel’chyonok O, Avvakumov G. Interaction of human CBG with
cell membranes. J Steroid Biochem Mol Biol 1991;40:795–803.
17. Simard M, Hill L, Lewis J, Hammond G. Naturally occurring muta-
tions of human corticosteroid-binding globulin. J Clin Endocrinol
Metab 2015;100:E129–39.
18. Burt MG, Mangelsdorf BL, Rogers A, Ho JT, Lewis JG, et al. Free
and total plasma cortisol measured by immunoassay and mass
spectrometry following ACTH 1-24 stimulation in the assessment
of pituitary patients. J Clin Endocrinol Metab 2013;98:1883–90.
19. Monaghan P, Keevil B, Trainer P. Mass spectrometry for the
endocrine clinic – much to digest. Clin Endocrinol (Oxf)
2013;78:344–6.
20. Hoofnagle A, Wener M. The fundamental flaws of immuno-
assays and potential solutions using tandem mass spectrometry.
J Immunol Methods 2009;347:3–11.
21. Hawley J, Owen L, Lockhart S, Monaghan P, Armston A, et al.
Serum cortisol: an up-to-date assessment of routine assay
performance. Clin Chem 2016;62:1220–9.
22. Taylor A, Keevil B, Huhtaniemi I. Mass spectrometry and
immuno­assay: how to measure steroid hormones today and
tomorrow. Eur J Endocrinol 2015;173:D1–12.
23. Nieman L, Biller B, Findling J, Newell-Price J, Savage M, et al.
The diagnosis of Cushing’s syndrome: an Endocrine Society
Clinical Practice Guideline. J Clin Endocrinol Metab 2008;93:
1526–40.
24. Handelsman D, Wartofsky L. Requirement for mass spectrometry
sex steroid assays in the Journal of Clinical Endocrinology and
Metabolism. J Clin Endocrinol Metab 2013;98:3971–3.
25. Kazlauskaite R, Maghnie M. Pitfalls in the diagnosis of central
adrenal insufficiency in children. Endocr Dev 2010;17:96–107.
26. Kyriakopoulou L, Yazdanpanah M, Colantonio DA, Chan MK,
Daly CH, et al. A sensitive and rapid mass spectrometric method
for the simultaneous measurement of eight steroid hormones
and CALIPER pediatric reference intervals. Clin Biochem
2013;46:642–51.
27. Hawley J, Owen L, MacKenzie F, Mussell C, Cowen S, et al. Candi-
date reference measurement procedure for the quantification of
total serum cortisol with LC-MS/MS. Clin Chem 2016;62:262–9.
28. Pretorius CJ, Galligan JP, McWhinney BC, Briscoe SE, Ungerer
JP. Free cortisol method comparison: ultrafiltration, equilibrium
dialysis, tracer dilution, tandem mass spectrometry and calcu-
lated free cortisol. Clin Chim Acta 2011;412:1043–7.
29. Vogeser M, Mohnle P, Briegel J. Free serum cortisol: quanti-
fication applying equilibrium dialysis or ultrafiltration and
an automated immunoassay system. Clin Chem Lab Med
2007;45:521–5.
30. Molenaar N, Groeneveld A, de Jong M. Three calculations of free
cortisol versus measured values in the critically ill. Clin Biochem
2015;48:1053–8.
31. Vincent R, Etogo-Asse F, Dew T, Bernal W, Alaghband-Zadeh J,
et al. Serum total cortisol and free cortisol index give different
information regarding the hypothalamus-pituitary-adrenal axis
reserve in patients with liver impairment. Ann Clin Biochem
2009;46:505–7.
32. le Roux C, Chapman G, Kong W, Dhillo W, Jones J. Free corti-
sol index is better than serum total cortisol in determining
hypothalamic-pituitary-adrenal status in patients undergoing
surgery. J Clin Endocrinol Metab 2003;88:2045–8.
33. Dhillo W, Kong W, Le Roux C, Alaghband-Zadeh J, Jones J, et al.
Cortisol-binding globulin is important in the interpretation of
dynamic tests of the hypothalamic-pituitary-adrenal axis. Eur
J Endocrinol 2002;146:231–5.
34. Dorin R, Pai H, Ho J, Lewis J, Torpy D, et al. Validation of simple
method of estimating plasma free cortisol: role of cortisol bind-
ing to albumin. Clin Biochem 2009;42:64–71.
35. Nguyen P, Lewis J, Sneyd J, Lee R, Torpy D, et al. Development of
a formula for estimating plasma free cortisol concentration from
a measured total cortisol concentration when elastase-cleaved
and intact corticosteroid binding globulin coexist. J Steroid
Biochem Mol Biol 2014;141:16–25.
36. Miller R, Stalder T, Jarczok M, Almeida D, Badrick E, et al. The
CIRCORT database: reference ranges and seasonal changes in
Brought to you by | UCL - University College London
Authenticated
Download Date | 12/31/17 11:36 AM
 Verbeeten and Ahmet: CBG and the evaluation of adrenal insufficiency      9
diurnal salivary cortisol derived from a meta-dataset comprised
of 15 field studies. Psychoneuroendocrinology 2016;73:16–23.
37. Inder W, Dimeski G, Russell A. Measurement of salivary cortisol
in 2012 – laboratory techniques and clinical indications. Clin
Endocrinol (Oxf) 2012;77:645–51.
38. Turpeinen U, Hamalainen E. Determination of cortisol in
serum, saliva, and urine. Best Pract Res Clin Endocrinol Metab
2013;27:795–801.
39. Ju Bae Y, Gaudl A, Jaeger S, Stadelmann S, Hiemisch A, et al.
Immunoassay or LC-MS/MS for the measurement of salivary
cortisol in children? Clin Chem Lab Med 2016;54:811–22.
40. Blair J, Lancaster G, Titman A, Peak M, Newlands P, et al. Early
morning salivary cortisol and cortisone, and adrenal responses
to a simplified low-dose short Synacthen test in children with
asthma. Clin Endocrinol (Oxf) 2014;80:376–83.
41. Giraldi F, Ambrogio A. Variability in laboratory parameters
used for management of Cushing’s syndrome. Endocrine
2015;50:580–9.
42. Perogamvros I, Keevil B, Ray D, Trainer P. Salivary cortisone is
a potential biomarker for serum free cortisol. J Clin Endocrinol
Metab 2010;95:4951–8.
43. Mezzullo M, Fanelli F, Fazzini A, Gambineri A, Vicennati V, et al.
Validation of an LC-MS/MS salivary assay for glucocorticoid
status assessment: evaluation of the diurnal fluctuation of cor-
tisol and cortisone and of their association within and between
serum and saliva. J Steroid Biochem Mol Biol 2016;163:103–12.
44. Debono M, Harrison R, Whitaker M, Eckland D, Arlt W, et al.
Salivary cortisone reflects cortisol exposure under physiologi-
cal conditions and after hydrocortisone. J Clin Endocrinol Metab
2016;101:1469–77.
45. Stratakis C. Diagnosis and clinical genetics of Cushing
syndrome in pediatrics. Endocrinol Metab Clin North Am
2016;45:311–28.
46. Deutschbein T, Broecker-Preuss M, Flitsch J, Jaeger A, Althoff
R, et al. Salivary cortisol as a diagnostic tool for Cushing’s
syndrome and adrenal insufficiency: improved screening by an
automatic immunoassay. Eur J Endocrinol 2012;166:613–8.
47. Cornes M, Ashby H, Khalid Y, Buch H, Ford C, et al. Salivary cor-
tisol and cortisone responses to tetracosactrin (Synacthen). Ann
Clin Biochem 2015;52:606–10.
48. Eyal O, Limor R, Oren A, Schachter-Davidov A, Stern N, et al.
Establishing normal ranges of basal and ACTH-stimulated
serum free cortisol in children. Horm Res Paediatr 2016;
86:94–9.
49. Limor R, Tordjman K, Marcus Y, Greenman Y, Osher E, et al.
Serum free cortisol as an ancillary tool in the interpretation of
the low-dose 1-ug ACTH test. Clin Endocrinol (Oxf) 2011;75:
294–300.
50. Rauschecker M, Abraham SB, Abel BS, Wesley R, Saverino E,
et al. Cosyntropin-stimulated serum free cortisol in healthy,
adrenally insufficient, and mildly cirrhotic populations. J Clin
Endocrinol Metab 2016;101:1075–81.
51. Mak I, Yeung B, Ng Y, Choi C, Iu H, et al. Salivary cortisol and
­cortisone after low-dose corticotropin stimulation in the
­diagnosis of adrenal insufficiency. J Endocr Soc 2017;1:96–108.
52. Cetinkaya S, Ozon A, Yordam N. Diagnostic value of salivary
­cortisol in children with abnormal adrenal cortex functions.
Horm Res 2007;67:301–6.
53. Tornhage C. Reference values for morning salivary cortisol con-
centrations in healthy school-aged children. J Pediatr Endocrinol
Metab 2002;15:197–204.
54. Cameron A, Henley D, Carrell R, Zhou A, Clarke A, et al.
­Temperature-responsive release of cortisol from its binding
globulin: a protein thermocouple. J Clin Endocrinol Metab
2010;95:4689–95.
55. Rhodes A, Evans LE, Alhazzani W, Levy MM, Antonelli M,
et al. Surviving sepsis campaign: international guidelines for
management of sepsis and septic shock: 2016. Crit Care Med
2017;45:486–552.
56. Dellinger RP, Levy MM, Rhodes A, Annane D, Gerlach H, et al.
Surviving sepsis campaign: international guidelines for man-
agement of severe sepsis and septic shock: 2012. Crit Care Med
2013;41:580–637.
57. Gunnala V, Guo R, Minutti C, Durazo-Arvizu R, Laporte C, et al.
Measurement of salivary cortisol level for the diagnosis of
critical illness-related corticosteroid insufficiency in children.
Pediatr Crit Care Med 2015;16:e101–6.
58. Tsai S, Seiler K, Jacobson J. Morning cortisol levels affected by
sex and pubertal status in children and young adults. J Clin Res
Pediatr Endocrinol 2013;5:85–9.
Brought to you by | UCL - University College London
Authenticated
Download Date | 12/31/17 11:36 AM