Keywords
Cushing disease - ACTH-dependent Cushing syndrome - BIPSS - CRH stimulation - non-invasive
Introduction
Endogenous Cushing Syndrome (CS) is a state of glucocorticoid hypersecretion
characterized by the loss of normal diurnal rhythm of cortisol secretion and the
normal negative feedback suppression of hypothalamic-pituitary-adrenal axis.
Adrenocorticotrophic hormone (ACTH)-dependent CS accounts for 80–85%
of endogenous CS. Differentiation between the two subtypes of ACTH-dependent CS,
that is, Cushing disease (CD) and ectopic ACTH syndrome (EAS), is a clinical
challenge [1]. This is especially true when
magnetic resonance imaging (MRI) of pituitary is negative/ambiguous
(e. g., absent lesion or a lesion<6 mm, a common finding
in>60% patients with CD) [2],
or when the results of endocrine dynamic tests [e. g., high-dose
dexamethasone suppression test (HDDST) and peripheral corticotropin-releasing
hormone (CRH) stimulation test] are discordant [3]
[4].
Bilateral inferior petrosal sinus sampling (BIPSS) is considered as the gold standard
in the differential diagnosis between CD and EAS. This procedure involves the
catheterization of bilateral inferior petrosal sinuses and the demonstration of
presence or absence of an ACTH gradient between the central venous effluent and the
peripheral vein [5]. According to a recent
international consensus, all patients with ACTH-dependent CS and no or equivocal
adenoma or microadenoma (size<6–9 mm) on pituitary MRI are
candidates for BIPSS. On the other hand, patients with a macroadenoma (size
≥10 mm) can be presumed to have CD and skip BIPSS [6]. Given that only a small proportion (around
10–15%) of patients with CD present pituitary macroadenoma, BIPSS
may be needed in a majority [7]
[8]. However, BIPSS is invasive, cost-intensive
and is available only at selected centers of expertise. Furthermore, BIPSS can have
false negative and false positive results and the accuracy in clinical practice is
far from 100% [5]
[9]. For instance, in a recent study, using a
post-CRH IPS: P ACTH cut-off≥2.1, Detomas et al., reported 9 false negative
results out of 103 CD cases (sensitivity 91%) and 1 false positive result
out of 14 EAS cases (specificity 93%) [10]. Overall, the sensitivity and specificity of BIPSS varies between
88–100% and 67–100%, respectively [5]. Notably, the availability of human CRH,
employed as a stimulating agent in BIPSS as well as peripheral CRH test, is
presently dwindling globally [11]
[12]. Thus, there is a need for simple, easily
available, cost-effective, and non-invasive alternatives that can be of immense
value to resource-rich and resource-poor settings alike. With this background, the
current study aimed to evaluate the utility of strategies alternative to BIPSS and
peripheral CRH stimulation in the differential diagnosis of ACTH-dependent CS. For
this purpose, we evaluated data from our cohort of 99 patients with treatment
naïve ACTH-dependent CS (EAS, n=23 and CD, n=76), the
clinical and biochemical profile of which have been published previously [13].
Materials and Methods
Settings and study design
The data were collected as a part of an ambispective observational study
performed in the department of Endocrinology and Metabolism, All India Institute
of Medical Sciences, New Delhi, India between 2021 and 2022. Case files for
patients admitted with features of endogenous CS between January 2013 and March
2021 were reviewed retrospectively, while a prospective review was made for
patients admitted between March 2021 and June 2022. Based on this review, a
proforma was filled documenting the relevant demographic, clinical, biochemical,
and radiological details. The protocol was approved by the institutional ethics
committee (IEC No. IECPG 265/24.03.2021) and informed consent was
obtained from all patients enrolled prospectively in the study.
Study objectives
The objectives of our study were: a) to evaluate the performance of various
parameters in the differential diagnosis of ACTH-dependent CS, b) to derive
cohort-specific cut-offs for biochemical parameters and dynamic tests used in
the differential diagnosis, and c) to derive simple and non-invasive
strategies/models that provide 100% specificity and positive
predictive value (PPV) for CD, thus precluding the need for BIPSS alone or both
BIPSS and peripheral CRH stimulation.
Data collection and study protocol and definitions
We collected data for 146 patients with ACTH-dependent CS, of whom 29 patients
with an occult ACTH source and 18 patients with a history of prior surgery or
radiotherapy were excluded. Thus, the current analysis involved 99 patients with
treatment naïve ACTH-dependent CS, including 23 with EAS and 76 with CD.
For evaluating the differential performance of various tests and deriving
cohort-specific cut-offs, we included all 99 patients with ACTH-dependent CS. On
the other hand, for the analysis involving non-invasive models, we excluded
patients with pituitary macroadenoma (n=14), and the sample size was 85
(see section “development of non-invasive models”).
The protocol for diagnosis of ACTH-dependent hypercortisolism and its subsequent
evaluation as well as the study definitions for CD and EAS have been provided in
our previous publication [13]. Briefly, a
plasma 8 AM ACTH level>10 pg/ml was used to subtype
endogenous CS as ACTH-dependent [14]. Such
patients underwent a number of inpatient tests for differential diagnosis,
including contrast-enhanced MRI pituitary (n=92; CD, n=76 and
EAS, n=16), HDDST (n=76; CD, n=59 and EAS,
n=17), peripheral CRH stimulation (n=36; CD, n=30 and
EAS, n=6), contrast-enhanced computed tomography (CT) scan of neck to
pelvis (n=56; CD, n=35 and EAS, n=21),
68-Ga-DOTANOC-PET/CT scan (n=60; CD, n=38 and EAS,
n=22) and CRH-stimulated BIPSS (bilateral and simultaneous approach with
central and peripheral samples collected at baseline and at 3, 5, 10, and
15 minutes after 100 μg human CRH injection;
n=17; CD, n=14 and EAS, n=3). For peripheral CRH
stimulation, venous samples for cortisol and ACTH were collected at baseline and
at 15, 30, 45, 60, 90, and 120 minutes after intravenous human CRH
injection. The percentage ACTH/cortisol response was calculated as the
difference between peak and baseline ACTH/cortisol level divided by the
baseline ACTH/cortisol level, × 100. For HDDST, a single tablet
of 2 mg dexamethasone was administered every 6 hours for
48 hours (9–3–9–3 or
12–6–12–6 regimen) and the venous sample was collected
at 8 AM (2 or 5 hours after the last dose). The percentage cortisol
suppression was calculated as the difference between post-HDDST and basal 8 am
serum cortisol divided by basal 8 AM serum cortisol, × 100. Hypokalemia
was defined as lowest serum potassium level during
admission<3.5 mmol/l.
The diagnosis of CD was based on histopathological evidence of ACTH
immunostaining pituitary adenoma (n=61) and/or biochemical
remission following adenoma resection (early remission, n=51 and delayed
remission, n=10) or evolution of corticotroph tumor progression after
bilateral adrenalectomy/Nelson’s syndrome (CTP-BADX/NS;
n=2). The diagnosis of EAS was based on unambiguous findings on
68Ga-DOTANOC PET/CT (n=21) and/or other
imaging modalities (n=19) and/or resolution of hypercortisolism
following surgery of ectopic tumor source (n=5). Thus, the differential
diagnosis into CD and EAS in this study was not based on the results of BIPSS or
peripheral CRH stimulation tests, which lack 100% accuracy [4]
[5]
[9].
Hormone analysis
Cortisol was measured in blood and saliva samples and ACTH in blood samples using
an electrochemiluminescence immunoassay on Cobas e411 autoanalyzer (Roche
Diagnostics, Germany). The reference ranges provided by the manufacturer are 8
AM serum cortisol (5th–95th percentile):
6.0–18.4 μg/dl, late-night salivary
cortisol:<0.274 μg/dl (95th percentile)
and<0.410 μg/dl (97.5th percentile), and 8 AM
plasma ACTH (5th–95th percentile):
7.2–63.3 pg/ml.
Development of non-invasive models
We developed a total of six non-invasive models combining both qualitative and
quantitative variables; each model was individually associated with 100%
specificity and PPV for CD and thus precluded the need for BIPSS in patients
fulfilling all its components. Because, as per current recommendations, patients
with ACTH-dependent CS and pituitary macroadenoma are considered to have CD and
spared of BIPSS [6], such patients
(n=14) were not included in the analysis pertaining to the non-invasive
models and the final sample on which these models were applied was 85.
For quantitative variables employed in the models, we selected both
cohort-specific thresholds (e. g., plasma ACTH
level<97 pg/ml, HDDST cortisol
suppression≥57%, peripheral CRH stimulation ACTH
rise≥77%, peripheral CRH stimulation cortisol
rise≥11%) as well as higher thresholds described in the
literature that are associated with a greater specificity for CD (e. g.,
HDDST cortisol suppression>80% and peripheral CRH stimulation
ACTH rise>100%).
The first three models included components other than peripheral CRH stimulation
test (model 1: female gender, absence of hypokalemia, plasma
ACTH<97 pg/ml and HDDST cortisol
suppression≥57% ; model 2: absence of hypokalemia, plasma
ACTH<97 pg/ml and HDDST cortisol
suppression>80%; model 3: absence of hypokalemia, plasma
ACTH<97 pg/ml, HDDST cortisol
suppression≥57% and definitive lesion on CEMRI pituitary), while
the next three models (model 4: HDDST cortisol suppression>80%
and peripheral CRH stimulation ACTH rise≥77%; model 5:
peripheral CRH stimulation ACTH rise>100% alone; model 6:
peripheral CRH stimulation cortisol rise≥11% alone) additionally
included this dynamic test. We derived the total number of patients with
ACTH-dependent CS (and no macroadenoma) fulfilling all the components in a given
model as well as the number of individuals exclusive to a given model [for
example, 13 patients fulfilled all the components in model no. 2 (i. e.,
they had no hypokalemia, plasma ACTH was<97 pg/ml and
HDDST cortisol suppression was>80%), however, 10 of them also
satisfied the preceding model (no. 1), and therefore, the number exclusive to
this model was 3].
Non-invasive models: rationale for selected variables
The variables selected for deriving the non-invasive models were the ones that
are relatively simple and more easily available compared to BIPSS and
individually associated with high PPV (>80%) for CD in our
analysis. We chose MRI pituitary and not CT/Ga-DOTANOC PET/CT in
imaging parameters, because MRI pituitary is the first line of investigation for
patients with ACTH-dependent CS and was more uniformly available in study
participants.
Statistical analysis
Statistical analysis was performed using Stata 15.0 (StataCorp LP, Texas, USA).
Qualitative data were represented as number (percentage) and quantitative data
as mean±SD or median (P25–P75). To
evaluate the performance of various tests in differential diagnosis of
ACTH-dependent CS, receiver operating characteristic (ROC) curves were drawn and
area under curve (AUC; 95% CI) values were derived. Using the ROC
analysis, the optimal cut-offs were derived and the corresponding sensitivity
(95% CI), specificity (95% CI), positive predictive value (PPV;
95% CI), negative predictive value (NPV; 95% CI) and likelihood
ratio positive (LR+; 95% CI) were reported. Subsequently, using
a combination of various demographic (female gender), clinical (presence or
absence of hypokalemia), biochemical (plasma ACTH, HDDST cortisol suppression,
peripheral CRH stimulation ACTH and cortisol response) and imaging (MRI
pituitary) parameters, we derived non-invasive models with 100%
specificity and PPV for CD. A p-value<0.05 was considered statistically
significant.
Results
Baseline characteristics of study participants
The detailed clinical, demographic, and biochemical profile of this cohort has
been published previously [13]. The
salient parameters are provided in [Table
1]. Patients with CD (n=76) were more likely to be females and
less likely to manifest hypokalemia compared to those with EAS (n=23).
The median 8 am serum cortisol, late-night salivary cortisol and 8 AM plasma
ACTH levels were significantly higher in EAS group. Among patients with CD, 14
had pituitary macroadenoma.
Table 1 Baseline characteristics of study
participants.
|
Variable
|
EAS (n=23)
|
CD (n=76)
|
p-Value
|
|
Age (years)
|
24.5 (20–40)
|
29.0 (23.1–37.5)
|
0.27
|
|
Female (n, %)
|
8 (34.8%)
|
55 (72.4%)
|
<0.001
|
|
BMI (kg/m2)
|
25.3 (21.9–31.5)
|
27.3 (24.7–32.0)
|
0.051
|
|
Hypertension (n, %)
|
19 (82.6%)
|
64 (84.2%)
|
0.855
|
|
Diabetes (n, %)
|
13 (56.5%)
|
39 (51.3%)
|
0.661
|
|
Hypokalemia (n, %)
|
19 (82.6%)
|
16 (21.0%)
|
0.001
|
|
8 AM serum cortisol (μg/dl)
|
49.7 (29.6–63.4)
|
26.7 (22.6–37.9)
|
<0.001
|
|
11 pm serum cortisol (μg/dl)
|
40.7±18.2
|
26.9±12.1
|
<0.001
|
|
Late-night salivary cortisol (μg/dl)
|
3.7 (2.4–10.3)
|
1.08 (0.68–1.86)
|
<0.001
|
|
ONDST cortisol (μg/dl)
|
25.3 (22.6–82.0)
|
20.1 (14.9–23.7)
|
0.02
|
|
8 AM plasma ACTH (pg/ml)
|
151.4 (94.7–279.0)
|
76.4 (46.2–102.8)
|
<0.001
|
Data are presented as number (%), mean±SD or median
(IQR). ACTH: Adrenocorticotropic hormone; BMI: Body mass index; CD:
Cushing disease; EAS: Ectopic ACTH syndrome; ONDST: Overnight
dexamethasone suppression test.
Differential diagnosis of ACTH-dependent CS
The sensitivity, specificity, PPV, NPV and LR+of different demographic,
clinical and biochemical variables (at standard cut-offs described in the
literature) for differential diagnosis of ACTH-dependent CS have been provided
in [Table 2].
Table 2 Sensitivity, specificity, positive and negative
predictive values of different demographic, clinical and biochemical
parameters for Cushing disease.
|
Parameter
|
Sensitivity (95% CI)
|
Specificity (95% CI)
|
PPV (95% CI)
|
NPV (95% CI)
|
LR+(95% CI)
|
|
Female gender
|
72.4 (61.4–81.2)
|
65.2 (44.9–81.2)
|
87.3 (76.9–93.4)
|
41.7 (27.1–57.8)
|
2.1 (1.6–2.7)
|
|
No hypokalemia
|
79.0 (68.5–86.6)
|
82.6 (62.8–93.0)
|
93.8 (85.0–97.5)
|
54.3 (38.2–69.5)
|
4.5 (2.8–7.5)
|
|
Plasma ACTH<90 pg/ml
|
65.8 (54.6–74.5)
|
82.6 (62.9–93.0)
|
92.6 (82.5–97.1)
|
42.2 (28.9–56.7)
|
3.8 (2.3–6.3)
|
|
HDDST:>50% cortisol
suppressiona
|
69.5 (56.8–80)
|
58.8 (36–78)
|
85.4 (72.8–92.7)
|
35.7 (20.7–54.1)
|
1.7 (1.2–2.3)
|
|
HDDST:>80% cortisol
suppressiona
|
40.7 (29–53.4)
|
82.4 (59.0–93.8)
|
88.9 (72.0–96.1)
|
28.6 (17.8–42.4)
|
2.3 (1.1–5.0)
|
|
Peripheral CRH:>50% ACTH riseb
|
93.3 (78.6–98.1)
|
83.3 (43.6–97.0)
|
96.6 (82.8–99.3)
|
71.4 (35.8–91.7)
|
5.6 (0.78–40)
|
|
Peripheral CRH:>100% ACTH
riseb
|
73.3 (55.5–85.8)
|
100 (60.9–100)
|
100 (85.1–100)
|
42.9 (21.3–67.4)
|
–
|
|
Peripheral CRH:>20% cortisol
riseb
|
86.7 (70.3–94.7)
|
100 (61–100)
|
100 (87.1–100)
|
60.0 (31.2–83.1)
|
–
|
|
Presence of lesion on CEMRI pituitaryc
|
81.6 (71.4–88.7)
|
93.8 (71.6–98.9)
|
98.4 (91.5–99.7)
|
51.7 (34.4–68.6)
|
13.1 (1.8–93.3)
|
|
Absence of lesion on CECT scand
|
97.1 (85.5–99.5)
|
90.5 (71.1–97.3)
|
94.4 (81.9–98.5)
|
95.0 (76.4–99.1)
|
10.2 (3.8–27.2)
|
|
Absence of lesion on 68-Ga-DOTANOC-PET/CT
scane
|
94.7 (82.7–98.5)
|
95.5 (78.2–99.2)
|
97.3 (86.2–99.5)
|
91.3 (73.2–97.6)
|
20.8 (2.9–148.4)
|
ACTH: Adrenocorticotropic hormone; CECT: Contrast enhanced computed
tomography; CEMRI: Contrast enhanced magnetic resonance imaging; CRH:
Corticotropin releasing hormone; HDDST: High dose dexamethasone
suppression test; LR+: Likelihood ratio positive; NPV: Negative
predictive value; PPV: Positive predictive value. a HDDST
performed in 59 patients with Cushing disease and 17 with ectopic ACTH
syndrome. b Peripheral CRH stimulation performed in 30
patients with Cushing disease and 6 with ectopic ACTH syndrome.
c CEMRI pituitary performed in all 76 patients with
Cushing disease and 16 with ectopic ACTH syndrome. d CECT
scan performed in 35 patients with Cushing disease and 21 with ectopic
ACTH syndrome. e 68-Ga-DOTANOC-PET/CT scan performed
in 38 patients with Cushing disease and 22 with ectopic ACTH
syndrome.
Notably, variables with high PPV (> 80%) for CD included:
female gender, absence of hypokalemia, plasma ACTH
levels<90 pg/ml, HDDST cortisol
suppression>50%/80%, peripheral
CRH>50% ACTH increase, presence of a lesion on MRI pituitary,
absence of a peripheral lesion on CT imaging and absence of a peripheral lesion
on 68-Ga-DOTANOC PET/CT imaging. Furthermore, an ACTH
rise>100% and a cortisol rise>20% following CRH
stimulation provided 100% specificity and PPV for CD.
Performance of various biochemical and dynamic tests and optimal
cohort-specific cut-offs
The results of ROC analysis, the area under the curve (AUC) for different
biochemical variables and dynamic tests in differential diagnosis and the
optimal cohort-specific cut-offs (best criteria) with corresponding sensitivity,
specificity, PPV, NPV and LR+are presented in [Table 3].
Table 3 Performance of different biochemical parameters
and dynamic tests and optimal cohort-specific cut-offs for diagnosis
of Cushing disease.
|
Parameter
|
AUC (95% CI)
|
Optimal cut-off
|
Sensitivity (95% CI)
|
Specificity (95% CI)
|
PPV 95% CI)
|
NPV (95% CI)
|
LR+(95% CI)
|
|
Serum 8 AM cortisol (μg/dl)
|
0.730 (0.60–0.85)
|
<34.1
|
67.1 (55.9–76.6)
|
65.2 (44.9–81.2)
|
86.4 (75.5–93.0)
|
37.5 (24.2–53.0)
|
1.9 (1.5–2.5)
|
|
Plasma 8 AM ACTH (pg/ml)
|
0.800 (0.68–0.91)
|
<97.3
|
71.1 (60.0–80.0)
|
69.6 (49.1–84.4)
|
88.5 (78.2–94.3)
|
42.1 (27.9–57.8)
|
2.3 (1.7–3.1)
|
|
HDDST: % cortisol suppression
|
0.702 (0.56–0.84)
|
≥57%
|
61.0 (48.3–72.4)
|
58.8 (36.0–78.4)
|
83.7 (70.0–91.9)
|
30.3 (17.4–47.3)
|
1.5 (1.1–2.0)
|
|
Peripheral CRH: % ACTH rise
|
0.933 (0.84–1.0)
|
≥77%
|
86.7 (70.3–94.7)
|
83.3 (43.7–97.0)
|
96.3 (81.7–99.3)
|
55.6 (26.7–81.1)
|
5.2 (0.72–37.4)
|
|
Peripheral CRH: % cortisol rise
|
0.975 (0.92–1.0)
|
≥11%
|
96.7 (83.3–99.4)
|
100 (61–100)
|
100 (88.3–100)
|
85.7 (48.7–97.4)
|
–
|
ACTH: Adrenocorticotropic hormone; AUC: Area under curve; CRH:
Corticotropin releasing hormone; HDDST: High dose dexamethasone
suppression test; LR+: Likelihood ratio positive; NPV: Negative
predictive value; PPV: Positive predictive value.
Serum 8 AM cortisol and relative percent cortisol suppression on HDDST showed
poor accuracy, plasma 8 AM ACTH showed good accuracy and relative percent ACTH
and cortisol rise on peripheral CRH stimulation showed excellent accuracy in
discriminating between CD and EAS. The optimal cut-offs for diagnosis of CD in
our cohort were plasma ACTH<97.3 pg/ml, HDDST relative
cortisol suppression≥57%, peripheral CRH relative ACTH
increase≥77% and relative cortisol
increase≥11%.
Simple and non-invasive alternative models/strategies in differential
diagnosis
We derived a total of six non-invasive models that provided 100%
specificity and PPV for CD ([Table 4]).
These models precluded the need for BIPPS in 35/85 (41.2%)
patients with ACTH-dependent CS in whom the procedure would have otherwise been
recommended (on account of absence of pituitary macroadenoma). Of these, the
first three models included basic parameters other than peripheral CRH
stimulation and precluded the need for both peripheral CRH stimulation and BIPSS
in 19 (22.4%) participants. On the other hand, the last three models
included peripheral CRH stimulation and precluded the need for BIPSS in another
16 (18.8%) participants.
Table 4 Non-invasive strategies/models for
diagnosis of Cushing disease in study cohort (n=85, after
excluding patients with pituitary macroadenoma).
|
Entry
|
Parameter used
|
Model/Strategy
|
PPV
|
N*
|
n*
|
|
1
|
Gender Serum Potassium Plasma ACTH HDDST
|
Female gender Absence of hypokalemia Plasma
ACTH<97 pg/ml, and
HDDST≥57% cortisol suppression
|
100%
|
15
|
15
|
|
2
|
Serum Potassium Plasma ACTH HDDST
|
Absence of hypokalemia Plasma
ACTH<97 pg/ml, and
HDDST>80% cortisol suppression
|
100%
|
13
|
3
|
|
3
|
Serum Potassium Plasma ACTH HDDST CEMRI pituitary
|
Absence of hypokalemia Plasma
ACTH<97 pg/ml
HDDST≥57% cortisol suppression, and
Definitive lesion on CEMRI pituitary
|
100%
|
16
|
1
|
|
4
|
HDDST Peripheral CRH stimulation
|
HDDST:>80% cortisol suppression and
Peripheral CRH stimulation:≥77% ACTH
rise
|
100%
|
8
|
2
|
|
5
|
Peripheral CRH stimulation
|
Peripheral CRH stimulation:>100% ACTH rise
|
100%
|
19
|
11
|
|
6
|
Peripheral CRH stimulation
|
Peripheral CRH stimulation:≥11% cortisol
rise
|
100%
|
25
|
3
|
|
Total
|
–
|
–
|
–
|
–
|
35/85 (41.2%)
|
ACTH: Adrenocorticotropic hormone; CEMRI: Contrast enhanced magnetic
resonance imaging; CRH: Corticotropin releasing hormone; HDDST: High
dose dexamethasone suppression test; PPV: Positive predictive value.
*“N” is the total number of
individuals with ACTH-dependent CS fulfilling all the variables in a
given model and “n” is the number of individuals
exclusive to that model. For example, 13 individuals (N) fulfilled all
the variables in model no. 2, however, 10 of them also fulfilled all the
variables in the preceding model (no. 1), and therefore, the number
exclusive to this model was 3 (n). Similarly, 19 individuals (N)
satisfied model no. 5, however, 8 of them also satisfied the preceding
models (no. 1 to 4), and therefore, the number exclusive to this model
was 11 (n).
Discussion
The differential diagnosis of ACTH-dependent CS is challenging, especially given that
patients with benign EAS often have an indolent course, at times clinically and
biochemically indistinguishable from CD. Pituitary imaging and endocrine dynamic
tests are useful but have their own limitations and cannot be completely relied upon
to differentiate between an ectopic and pituitary source [3]
[4]
[5]. Thus, BIPSS remains the most
accurate procedure in the differential diagnosis and recommended in most patients
except for those harboring a pituitary macroadenoma [6]. In this study, we evaluated the diagnostic performance of various
tests in the differential diagnosis of ACTH-dependent CS and proposed simple and
non-invasive strategies that preclude the need for more invasive and
resource-intensive procedures such as BIPSS in a large proportion of patients. We
found that peripheral CRH stimulation test (both relative percent ACTH and cortisol
rise) provided excellent discrimination between EAS and CD and that the cortisol
rise≥11% and the ACTH rise>100% individually
excluded the possibility of EAS. We derived six non-invasive models that combined
various demographic, clinical, biochemical, and imaging parameters and provided
100% specificity and PPV for CD, thus precluding the need for BIPSS in
41% participants in whom this procedure would have otherwise been
recommended (on account of absence of a pituitary macroadenoma) [6]. Importantly, the first three models
included basic clinical and biochemical parameters other than peripheral CRH
stimulation test and precluded the need for BIPSS as well as CRH test in 22%
participants.
Recently, Frete et al. in their cohort of 194 patients with ACTH-dependent CS (167
with CD and 27 with EAS) evaluated a strategy of performing intravenous human CRH
and desmopressin stimulation tests (consecutively on different days and in a random
order) alongside pituitary MRI in all patients, followed by CT scan of neck to
pelvis where diagnosis of CD is not clear [15]. The authors reported 100% PPV for CD in patients with positive
response to CRH (37% ACTH and 17% cortisol increase) and
desmopressin (33% ACTH and 18% cortisol increase) tests with a
negative pituitary MRI and a negative CT scan and 100% NPV for CD in
patients with a negative response to CRH and desmopressin tests with a negative
pituitary MRI and a positive CT scan. They concluded that using this approach, BIPSS
could be potentially avoided in 53/112 (47%) patients where it would
have been recommended. Similarly, in our cohort, using simple and non-invasive
strategies that achieved 100% specificity and PPV for CD, BIPSS could be
spared in 35/85 (41.2%) patients where it would have been
recommended according to the current international consensus [6]. Furthermore, including 14 patients with
pituitary macroadenoma who are presumed to be CD and do not require BIPSS,
49/99 patients (49.5%, i. e., nearly 1 in every 2 patients
with ACTH-dependent CS) can be directly referred for a pituitary surgery. BIPSS is
invasive, expensive, and only available at selected centers of excellence; the
non-invasive strategy outlined here simplifies the diagnostic algorithm and has the
potential to limit BIPSS to only a selected proportion of patients (nearly 1 in
every 2 patients with ACTH-dependent CS). In a retrospective analysis of 264
patients with CD and 47 patients with EAS, Lyu et al., similarly reported a
non-invasive scoring model (score ranging from –14 to 14) comprising of
simple variables such as gender, plasma ACTH, MRI pituitary, HDDST and hypokalemia
to minimize the need for BIPSS. The non-invasive model yielded a higher diagnostic
accuracy than HDDST (AUC: 0.915 vs. 0.756) and scores of≥−10
and≥13 provided 100% sensitivity and specificity, respectively, for
diagnosis of CD. The authors concluded that BIPSS may only be performed in patients
with scores between –10 and 12, as scores below and above these were
associated with a high probability of EAS and CD, respectively [16].
Recently, the availability of human CRH has also been a cause of major concern
globally [11]
[12]. Desmopressin has been suggested as an effective and less expensive
alternative to CRH during BIPSS [17]
[18]; however, peripheral desmopressin test has
lower accuracy compared to CRH test in differentiating between CD and EAS [19]
[20]
[21]. To add, intravenous
desmopressin is not available in India and other countries. We found that using a
combination of basic parameters (other than peripheral CRH test; models
1–3), CD could be diagnosed with 100% specificity and PPV in
19/85 (22.4%) patients; such patients could not only be spared of
BIPSS, but also of a peripheral CRH stimulation test. Furthermore, including 14
patients with pituitary macroadenoma who are presumed to be CD, 33/99
patients (33.3%, i. e., nearly 1 in every 3 patients with
ACTH-dependent CS) can be directly referred for a pituitary surgery without a BIPSS
and peripheral CRH test.
The value of HDDST in clinical practice for the differential diagnosis of
ACTH-dependent CS has been debated [22]
[23]. Nearly 20–30% patients
with EAS and a similar proportion of patients with CD can present false positive and
false negative cortisol suppression (>50%) on HDDST. Thus, the
overall diagnostic accuracy of this test is 70–80%, lower than the
pretest probability of 85–90% for CD in a patient with
ACTH-dependent CS [4]
[22]. We also noted that HDDST yielded poor
accuracy (AUC: 0.702) in discriminating between EAS and CD. As expected, the higher
cut-off>80% cortisol suppression (vs.>50%) yielded
higher specificity (82% vs. 59%) for CD, but at the cost of lower
sensitivity (41% vs. 70%) and the most optimal cut-off was derived
at≥57% cortisol suppression (sensitivity 61%, specificity
59%). While the performance of HDDST as a standalone test was suboptimal,
when used in certain models (models 1–4) in combination with various
clinical, biochemical, and radiological variables, the test yielded 100% PPV
for CD. In a recent study, Gupta et al. also highlighted the importance of using
HDDST in their limited invasive protocols alongside MRI pituitary and CT imaging to
avoid the need for BIPSS in 36–62% of patients with ACTH-dependent
CS [24]. In our study, hypokalemia also served
as an important discriminator between CD and EAS, especially in combination with
other parameters (models 1–3). Owing to the higher severity of
hypercortisolism, hypokalemia is more commonly reported in EAS (up to 85% of
patients) than CD (up to 20% of patients) and is related to overwhelming of
the 11-β-hydroxysteroid dehydrogenase type 2 enzyme by excess cortisol [13]
[25].
This results in inadequate conversion of active cortisol into inactive cortisone and
an exaggerated action of the former at mineralocorticoid receptor.
Peripheral CRH stimulation is useful in discriminating between the ectopic and
pituitary forms, however, the results may be unreliable in 7–15%
patients, most often due to false negative results in CD [4]. The sensitivity and specificity depend on
the stimulating agent (ovine vs. human CRH) used and the cut-off criteria chosen; a
range of 35–105% for the increase of ACTH and 14–50%
for the increase of cortisol above basal levels has been used, resulting in a
sensitivity of 70–93% and specificity of 95–100% for
ACTH response, and 50–91% and 88–100% for cortisol
response [4]
[26]. Thus, several thresholds have been published for CRH test, but none
is universal. In our study, peripheral CRH stimulation (using human CRH) was
performed in 36 patients (30 CD and 6 EAS) and both ACTH (AUC: 0.933) and cortisol
(AUC: 0.975) response yielded excellent discrimination between EAS and CD. The most
optimal cut-off was derived at≥77% for ACTH (sensitivity
87%, specificity 83%) and≥11% for cortisol
(sensitivity 97%, specificity 100%). The loss of specificity for
ACTH response at this cut-off was accounted by a single patient in the EAS group,
who demonstrated 97% ACTH rise following CRH stimulation. It is well known
that some neuroendocrine tumors causing EAS express functional CRH receptors,
resulting in false positive response to CRH stimulation [15]
[27].
Our findings are consistent with a previous study by Newell-Price et al. that
reported sensitivity and specificity estimates of 70% and 100%,
respectively for>105% ACTH response, and 85% and
100%, respectively for>14% cortisol response following human
CRH stimulation [28]. Recent studies,
including the ones by Detomas et al. (≥31% ACTH rise, sensitivity,
and specificity: 83% and 85%, respectively;≥12%
cortisol rise: 82% and 89%, respectively) and Ceccato et al.
(≥31% ACTH rise: 91% and 80%,
respectively;≥20% cortisol rise: 86% and 80%,
respectively) have demonstrated similar performance, albeit at different cut-offs
[29]
[30].
The strengths of our study include its relevance to both resource-rich and
resource-limited settings in limiting the use of more invasive and
resource-intensive procedures such as BIPSS and peripheral CRH stimulation. We
acknowledge certain limitations. First, a major component of data collection was
retrospective, with its inherent limitations. Second, we excluded patients with
occult ACTH source from the present analysis; it will be of interest to re-evaluate
the current strategies once the final diagnosis is available for these patients.
Third, the results of peripheral CRH stimulation were available in only 36 patients.
Considering that this test demonstrated excellent performance in differentiating
between CD and EAS and precluded the need for BIPSS in a large proportion of
patients (16/36) who took the test, a more uniform prescription could have
potentially added to the numbers fulfilling the non-invasive strategy. Finally, the
sample size was relatively small, and the study patients were derived from a single
center, with implications for generalizability to other cohorts. We recognize that
there is no one-size-fits-all approach to differential diagnosis of ACTH-dependent
hypercortisolism and the diagnostic protocol may vary from one center to another
depending upon several factors including the expertise and experience of the
treating team, the availability of resources and the nature of patient presentation.
Thus, similar non-invasive protocols should be evaluated and validated in other
settings using these and other emerging modalities such as desmopressin stimulation
test, and CRH-receptor targeted molecular imaging (68Ga-CRH-PET-CT) [31]
[32].
Recent studies have also shown subtype-specific differences in hematological
parameters between CD and EAS [33]
[34]; further validation of these findings and
incorporation in future non-invasive models would add value.
To conclude, using simple and non-invasive alternative strategies, BIPSS could be
avoided in 41% and peripheral CRH stimulation in 22% of patients
with ACTH-dependent CS and no macroadenoma, in whom these procedures would have
otherwise been recommended. These strategies should be validated in other settings
and the outcomes of treatment with the non-invasive vis-à-vis BIPSS-based
invasive strategy should be evaluated in future prospective studies.
