Abstract
Context
Determining the etiology of adrenocorticotropin (ACTH)-dependent Cushing's syndrome (CS) is often difficult. The gold standard test, inferior petrosal sinus sampling (IPSS), is expensive and not widely available.
Objective
Evaluate the performance of the corticotropin-releasing hormone stimulation test (CRH-ST) and the 8 mg high-dose dexamethasone suppression test (HDDST) in distinguishing Cushing's disease (CD) from ectopic ACTH syndrome (EAS).
Methods
Retrospective review in a tertiary referral center. A total of 323 patients with CD or EAS (n = 78) confirmed by pathology or biochemical cure (n = 15) in 96% underwent CRH-ST and HDDST performed between 1986 and 2019. We calculated test sensitivity (Se), specificity (Sp), positive predictive value (PPV), negative predictive value, and diagnostic accuracy (DA) for the diagnosis of CD, and determined optimal response criteria for each test, alone and in combination.
Results
The CRH-ST performed better than the HDDST (DA 91%, 95% CI 87-94% vs 75%, 95% CI 69-79%). Optimal response criteria were a ≥40% increase of ACTH and/or cortisol during the CRH test and a ≥69% suppression of cortisol during the HDDST. A ≥40% cortisol increase during the CRH test was the most specific measure, PPV 99%. Seventy-four percent of subjects had concordant positive CRH test and HDDST results, yielding Se 93%, Sp 98%, DA 95%, and PPV 99%, with a pretest likelihood of 85%. A proposed algorithm diagnosed 64% of patients with CD with near perfect accuracy (99%), obviating the need for IPSS.
Conclusion
CRH is a valuable tool to correctly diagnose the etiology of ACTH-dependent CS. Its current worldwide unavailability impedes optimal management of these patients.
Keywords: Cushing, corticotropin-releasing hormone, dexamethasone, IPSS, Cushing disease, ectopic ACTH syndrome
Surgical resection of the causal tumor is the optimal treatment of adrenocorticotropin (ACTH)-dependent endogenous Cushing's syndrome (CS) (1). Thus, correct identification of the etiology and subsequent localization of the tumor are critical to achieving remission. Cushing's disease (CD) represents 80% to 90% of ACTH-dependent CS, while ectopic ACTH syndrome (EAS) accounts for the remainder (2, 3). The high pretest probability of CD compared with EAS necessitates highly accurate diagnostic tests to distinguish the two entities and avoid inappropriate surgical procedures. While bilateral inferior petrosal sinus sampling (IPSS) is the gold standard test, it is an invasive and expensive procedure only available at highly specialized centers, and is associated with a small risk of serious complications (4).
Unfortunately, tumor imaging strategies cannot identify all ACTH-secreting tumors. Initial imaging fails to detect up to 50% of CD and EAS tumors (5). Pituitary imaging interpretation is complicated further by identification of incidentalomas up to 6 mm in 10% of the general population in addition to other more subtle abnormalities in over 15% of patients with EAS (3, 5-8). Therefore, biochemical testing is essential to discriminate CD from EAS (3, 9, 10).
The CRH stimulation test and the 8 mg overnight high-dose dexamethasone suppression test (HDDST) are noninvasive tests to distinguish CD from EAS. Most CD tumors release ACTH in response to CRH, while most ectopic sources do not (11). In cases of EAS with cyclic hypercortisolism, periods of normal or low cortisol are associated with a higher frequency of false positive CRH tests, highlighting the importance of testing during an active hypercortisolemic phase (8, 9). The HDDST relies on partially intact feedback inhibition in corticotrope adenomas, causing them to suppress ACTH release in response to dexamethasone (Dex). However, some ectopic tumors also express functional glucocorticoid receptors, leading to false positive results (12).
The reported sensitivity (Se) and specificity (Sp) of the CRH test (Se 76-100%; Sp 81-100%) (3, 5, 8-10, 12-14) overlap with those of the HDDST (Se 64-88%; Sp 60-100%) (2, 3, 5, 8, 10, 12). The performance of both tests depends on the criteria used to define a positive result, with variable cutoff points for the CRH test among series, while most series evaluating HDDST performance have used ≥50% suppression of cortisol to identify CD. The number of patients with EAS also affects the results, as the CI around the point estimate for test specificity is often quite large due to a small number of subjects.
There is no consensus on the minimal amount of data needed for high diagnostic accuracy. For the CRH test, several studies failed to show a diagnostic advantage when using both ACTH and cortisol levels compared with either level alone (5, 8, 9). Similarly, some studies found no additional benefit in combining the results of the CRH test and HDDST (5, 8). One large series found that both low- and high-dose Dex suppression tests performed very well when combined with the CRH test (Se 93-94%, Sp 95-97%), but notably used 48-hour Dex suppression test protocols. These require dosing every 6 hours and have been largely replaced by single-dose overnight tests, both for convenience and to minimize issues with medication compliance (15). Concordant HDDST and CRH test results, reported in 70-76% of CD patients and 64-89% of patients with EAS, have been found to confer high diagnostic accuracy (3, 8, 12).
Unfortunately, neither ovine nor human formulations of CRH are now available. We hypothesized that the HDDST alone would not be an adequate substitute for CRH. We reviewed results of the HDDST and CRH tests at our institution to assess their performance at established cutoffs, identify optimal response criteria, determine whether combining the two tests provided additional diagnostic accuracy, and evaluate the best approach when results are discordant.
Materials and Methods
We performed a retrospective review of the medical records of 323 patients with ACTH-dependent CS who underwent both the CRH test and HDDST at our institution between 1986 and 2019. CD or EAS was confirmed by either surgical pathology or biochemical cure after tumor resection; when this was not possible, the diagnosis was imputed from IPSS results. A subset of these patients has been reported in previous studies evaluating the tests’ performance (11, 16).
The CRH test was conducted as previously reported using 1 µg/kg body weight of ovine CRH (Ferring Pharmaceuticals, Malmo, Sweden) up to a total dose of 100 µg intravenously (11). Baseline ACTH and cortisol levels were determined by calculating the means at timepoints −5 and 0 minutes. Peak values were determined by calculating the mean ACTH at +15 and +30 minutes, and mean cortisol at +30 and +45 minutes. If only 1 ACTH or cortisol value was available, it was used as the mean value. An increase of ≥35% or ≥20% in mean ACTH and/or cortisol, respectively, represented a positive test for the diagnosis of CD. These timepoints and criteria were selected as they provided the highest diagnostic accuracy in an earlier series that included a large number of subjects with EAS (11).
For the HDDST, baseline serum cortisol was measured at 08:30 hours on day 1, with subjects receiving 8 mg of Dex orally at 23:00 hours that evening. On day 2, cortisol was measured at 09:00 hours. A cortisol reduction of ≥50% on day 2 compared with day 1, the most commonly used criterion, represented a positive test for the diagnosis of CD.
ACTH was analyzed using multiple immunoassays, initially at Hazleton Biotechnology Laboratories (Vienna, VA, 1986-2000), followed by NIH Department of Laboratory Medicine (DLM) (Bethesda, MD, 2000-2019), with the latter using the Nichols Advantage immunochemiluminometric assay (2000-2005), Siemens Immulite 2500 chemiluminescence immunoassay (2005-2012, RRID:AB_2877714), and Immulite 200 XPi chemiluminescence immunoassay (2012-2019) (normal ranges: 2000-2005, normal range: 9.0 to 52.0 pg/mL; 2005-2015, normal range: 0.0 to 46.0 pg/mL; 2015-2019, normal range: 5.0 to 46.0 pg/mL). These assays have been described previously and had similar diagnostic performance at baseline, and when used with CRH and IPSS testing (17). Cortisol was analyzed using multiple immunoassays, initially at Hazleton Biotechnology Laboratories (Vienna, VA, 1986-1999), followed by NIH DLM (Bethesda, MD, 1999-2019), with the latter using the Siemens Immulite 2500 Analyzer solid-phase competitive chemiluminescent enzyme immunoassay (normal range 1999-2019: 5.0 to 25.0 μg/dL, RRID:AB_2877715).
Optimal diagnostic criteria for each test were determined by calculating the Youden's index (J) according to the formula (J = [Se% + Sp%] − 100). The cutoff which yielded the highest J (closest to 100) for each test was deemed optimal. For the CRH test, test results were graphed on a scatterplot to visualize the optimal ACTH and cortisol cutoffs when combined. We calculated Se (Se = true positives/[true positives + false negatives]), Sp (Sp = true negatives/[true negatives + false positives]), diagnostic accuracy (DA =[true positives + true negatives]/total subjects) and positive (PPV = true positives/[true positives + false positives]), and negative (NPV = true negatives/[true negatives + false negatives]) predictive values for each diagnostic criterion.
Pituitary imaging was performed using magnetic resonance imaging when possible. Pituitary computed tomography was used instead in 7 patients: 6 due to poor magnetic resonance imaging availability in the period 1986-1989 and 1 in 2000 due to the presence of intracranial coils. IPSS was performed as previously described (4).
Statistical analyses were performed using GraphPad Prism version 9.4.1 (GraphPad, San Diego, CA, USA). Ages, basal ACTH and cortisol levels, and pituitary adenoma sizes are presented as medians with interquartile range as no data set had normal distribution. Comparisons between the CD and EAS cohorts were performed using the Mann–Whitney U-test.
Results
A total of 323 patients were included: 245 with CD and 78 with EAS. The diagnosis was confirmed by tumor histopathology in 91% (n = 295), with the remainder being verified by biochemical cure after tumor resection (n = 15) or imputed by IPSS (n = 13). Patient characteristics are shown in Table 1. As expected, CD patients were predominantly female (n = 193, 79%), while patients with EAS were evenly divided (female n = 40, 51%). CD patients were younger than patients with EAS, with median ages 37 and 43, respectively. Basal ACTH and cortisol levels were higher in the EAS cohort, but with a very large overlap, making basal levels unreliable diagnostic tools. Pituitary imaging was performed in 314 (97%) patients, showing a pituitary adenoma in 147 (61%) CD patients compared with 22 (31%) patients with EAS. When sizes were specified, adenomas were 6 mm or larger in 54 (47%) CD patients, compared with 5 (28%) patients with EAS, making this a poor discriminatory size cutoff. IPSS was performed in 260 (80%) patients with an accuracy of 98% when successful cannulation was achieved. Three patients with EAS had incorrect IPSS results, 1 of whom was a patient with an esthesioneuroblastoma, with the source of ectopic ACTH being adjacent to the pituitary. Another EAS patient with incorrect results was eucortisolemic due to metyrapone treatment at the time of testing, making the results potentially invalid due to lack of suppression of normal corticotropes.
Table 1.
Patient characteristics, performance of pituitary imaging and IPSS
| Variable | CS etiology | P value CD vs EAS | |
|---|---|---|---|
| CD (n = 245, 76%) | EAS (n = 78, 24%) | ||
| Female (%) | 193 (79) | 40 (51) | |
| Age range (years) | 8-77 | 20-80 | |
| Median (IQR) | 37 (28-48) | 43 (31-55) | .0042 |
| Method of diagnosis | |||
| Pathology | 227 | 68 | |
| Biochemical cure | 15 | 0 | |
| IPSS | 3 | 10 | |
| Basal ACTH (pg/mL)a | |||
| Range | 5.5-592.5 | 16.5-3290.0 | |
| Median (IQR) | 42.8 (26.3-67.5) | 96.7 (47.7-181.0) | <.0001 |
| Basal cortisol (µg/dL)a | |||
| Range | 2.5-133.5 | 11.2-135.0 | |
| Median (IQR) | 19.5 (16.1-25.3) | 35.0 (24.0-48.6) | <.0001 |
| Pituitary imaging performed (%) | 242 (99) | 72 (92) | |
| Visible adenoma (%) | 147 (61) | 22 (31) | |
| Size rangeb (mm) | 2.0-23.0 | 3.0-8.0 | |
| Median size (IQR) | 5.0 (4.0-8.0) | 4.0 (3.5-7.0) | .026 |
| Adenoma ≥6 mm (%) | 54 (47) | 5 (28) | |
| IPSS performed (%) | 187 (76) | 73 (94) | |
| Successful samplingc (%) | 183 (98) | 70 (96) | |
| Accuracyd (%) | 178 (99) | 58 (95) | |
Abbreviations: ACTH, adrenocorticotropin; CD, Cushing's disease; CS, Cushing's syndrome; EAS, ectopic ACTH syndrome; IPSS, inferior petrosal sinus sampling; IQR, interquartile range.
a Basal ACTH and cortisol values are the mean of the −5 and 0 minute results during the CRH test.
b Adenoma sizes are specified in millimeters in 115 (78%) of CD patients and 18 (82%) of patients with EAS with visible adenomas.
c Sampling was deemed unsuccessful based on a central to peripheral prolactin ratio <1.8 or documentation of failed cannulation by the performing radiologist.
d Subjects diagnosed based on IPSS were not included in the accuracy calculation. 180 CD patients and 61 patients with EAS had successful sampling and a diagnosis based on pathology or biochemical cure.
Among EAS lesions, 68 were confirmed with surgical pathology, while 10 patients (13%) had occult tumors and were classified as having EAS based on absence of a central to peripheral ACTH gradient on IPSS. Table 2 details the distribution of EAS culprit lesions.
Table 2.
Histopathologically confirmed NETs with ectopic ACTH secretion
| EAS tumor types (n = 68, excluding 10 occult tumors) | |||
|---|---|---|---|
| Tumor type | Number (%) | Tumor type | Number (%) |
| Pulmonary NET | 43 (63%) | Pulmonary tumorlet | 1 (1%) |
| Metastatic NET, unknown primary | 5 (7%) | Mediastinal NET | 1 (1%) |
| Thymic NET | 4 (6%) | Small cell lung carcinoma | 1 (1%) |
| Pancreatic NET | 3 (4%) | Prostate NET | 1 (1%) |
| Appendiceal NET | 3 (4%) | Esthesioneuroblastoma | 1 (1%) |
| Medullary thyroid carcinoma | 2 (3%) | Ovarian teratoma with corticotrope carcinoma | 1 (1%) |
| Pheochromocytoma | 2 (3%) | ||
Most lesions (74%) originated in the chest, with the majority of these (63% of total tumors) being pulmonary NETs.
Abbreviations: ACTH, adrenocorticotropin; EAS, ectopic ACTH syndrome; NET, neuroendocrine tumor.
CRH Test Performance
As shown in Table 3, the established criterion for ACTH, ≥35% increase (J = 75) for the diagnosis of CD, yielded Se of 88%, Sp 87%, and DA 88%. The best performance (J = 80) was seen at ≥40% increase, which yielded Se of 87%, Sp 92%, and DA 89%. For cortisol, the established criterion, ≥20% increase for the diagnosis of CD (J = 78), yielded Se of 89%, Sp 90%, and DA 89%. The optimal cutoff point was determined to be a ≥22% increase (J = 80), with improved specificity: Se of 88%, Sp 92%, DA 89%.
Table 3.
Diagnostic performance of the CRH test using different cutoffs for ACTH (A) and cortisol (F) increase, either alone or in combination
| Hormone | A/F | A | F | ||||
|---|---|---|---|---|---|---|---|
| Criterion | ≥40%/40% | ≥35%/20% (11) | ≥40%/22% | ≥40% | ≥35% | ≥22% | ≥20% |
| Se | 91 (86-94) | 93 (89-96) | 93 (89-96) | 87 (82-91) | 88 (83-91) | 88 (83-91) | 89 (84-92) |
| Sp | 92 (83-97) | 81 (70-88) | 88 (79-94) | 92 (83-97) | 87 (77-93) | 92 (83-97) | 90 (80-95) |
| PPV | 97 (94-99) | 94 (90-96) | 96 (93-98) | 97 (94-99) | 96 (92-98) | 97 (94-99) | 96 (93-98) |
| NPV | 76 (66-84) | 79 (68-87) | 80 (70-88) | 70 (60-78) | 69 (59-78) | 71 (61-79) | 71 (61-80) |
| DA | 91 (87-94) | 90 (86-93) | 92 (88-95) | 89 (84-92) | 88 (83-91) | 89 (85-92) | 89 (85-92) |
Se, Sp, PPV, NPV, and DA are reported as percentages with 95% CI. Our combined optimized criteria (A or F increase ≥40%) performed better than the previously established cutoffs (A ≥ 35% or F ≥ 20%, suggested by Nieman et al (11)) and the optimal individual cutoffs as calculated using Youden's index (A ≥ 40%, F ≥ 22%).
Abbreviations: ACTH, adrenocorticotropin; CRH, corticotropin-releasing hormone; DA, diagnostic accuracy; NPV, negative predictive value; PPV, positive predictive value; Se, sensitivity; Sp, specificity.
When the diagnosis of CD was assigned only if both ACTH and cortisol met response criteria, performance decreased due to a marked drop in sensitivity from 91% to 71% using the optimal response criteria below. Tests were therefore considered positive when either level met the response criteria. Using the established criteria (ACTH ≥35% and/or cortisol ≥20% increase), the combined assessment yielded Se of 93%, Sp 81%, and DA 90%. The optimal response criteria for combined ACTH and cortisol assessment (either/or positive) was a ≥40% increase (Fig. 1), which yielded Se of 91%, Sp 92%, and DA 91%. These revised criteria also worked better than the simplified criteria (15 minutes ACTH ≥43%; 30 minutes ACTH ≥31%; 30 minutes cortisol ≥12%) suggested by Ritzel et al (8) and Detomas et al (5) (Table 4).
Figure 1.
Relative increase of ACTH and cortisol during the CRH test. For clarity, only subjects with increases in both levels between 0 and 100% are shown. The horizontal and vertical dashed lines represent the thresholds (≥40% increase for each) that provided the best discrimination between CD (circles) and EAS (squares).
Table 4.
Diagnostic performance in our cohort of the CRH test using simplified response criteria for ACTH (A) or cortisol (F), suggested by Ritzel et al (8) and Detomas et al (5)
| Hormone | A (8) | A (5) | F (5) |
|---|---|---|---|
| Criterion | ≥43% @15 minutes | ≥31% @ 30 minutes | ≥12% @ 30 minutes |
| Se | 84 (79-88) | 88 (83-91) | 92 (87-95) |
| Sp | 92 (83-97) | 79 (69-87) | 74 (63-83) |
| PPV | 97 (93-99) | 93 (89-96) | 92 (87-95) |
| NPV | 66 (56-74) | 67 (57-77) | 74 (63-83) |
| DA | 86 (82-90) | 86 (81-89) | 88 (83-91) |
Se, Sp, PPV, NPV and DA are reported as percentages with 95% confidence intervals.
Abbreviations: ACTH, adrenocorticotropin; CRH, corticotropin-releasing hormone; DA, diagnostic accuracy; NPV, negative predictive value; PPV, positive predictive value; Se, sensitivity; Sp, specificity.
HDDST Performance
For the HDDST the established cutoff, ≥50% cortisol suppression for the diagnosis of CD (J = 56) yielded Se of 79%, Sp 77%, and DA 79%. The best performance was seen at two different cutoffs with near-identical Youden's indexes, ≥61% suppression (J = 59.7, Se 75%, Sp 85%, and DA 77%), and ≥69% suppression (J = 59.5, Se 70%, Sp 90%, and DA 75%) (Table 5). Placing a higher value on specificity, we opted to use ≥69% suppression as the optimal cutoff, in line with an earlier series (16).
Table 5.
Diagnostic performance of the high-dose dexamethasone suppression test using various criteria for cortisol suppression
| Criterion | −69% | −50% | −61% | −55% (5) |
|---|---|---|---|---|
| Se | 70 (64-75) | 79 (73-84) | 75 (69-80) | 78 (72-83) |
| Sp | 90 (80-95) | 77 (66-85) | 85 (74-91) | 78 (67-86) |
| PPV | 96 (91-98) | 92 (87-95) | 94 (89-97) | 92 (87-95) |
| NPV | 49 (40-57) | 54 (44-63) | 52 (43-61) | 53 (44-62) |
| DA | 75 (69-79) | 79 (74-83) | 77 (72-82) | 78 (73-82) |
Se, Sp, PPV, NPV, and DA are reported as percentages with 95% confidence intervals. Our optimized criterion (cortisol decrease ≥69%) provided a higher Youden's index (i.e., higher combined sensitivity and specificity) than the previously established cutoff point (cortisol decrease ≥50%) and the criterion ≥55% suggested by Detomas et al (5). The Youden's index was nearly identical when using the criterion ≥61%, but ≥69% was ultimately selected due to a higher value being placed on specificity. This criterion had previously been suggested by Dichek et al (16).
Abbreviations: DA, diagnostic accuracy; NPV, negative predictive value; PPV, positive predictive value; Se, sensitivity; Sp, specificity.
Combined CRH Test and HDDST
Finally, we evaluated diagnostic performance when results from both tests were combined. When using the established criteria (an ACTH and/or cortisol increase of ≥35% or ≥20%, respectively, for the CRH test and a ≥50% suppression of cortisol for the HDDST) and interpreting discordant results as negative, this yielded Se 75%, Sp 92%, PPV 97%, NPV 54%, and DA 79%. In total, 248 subjects (77%) had concordant test results, and among these test performance was improved: Se 96%, Sp 89%, PPV 97%, NPV 88%, and DA 95%.
Overall performance depended on whether discordant results were interpreted as positive or negative. If a positive CRH test with a negative HDDST was interpreted as a positive result, and an isolated positive HDDST result was interpreted as negative, this performed like the CRH test alone (Se of 93%, Sp 81%, DA 90%, and PPV 94% with established criteria). Similarly, if a negative CRH test with a positive HDDST was interpreted as a positive result, and an isolated positive CRH test result was interpreted as negative, this performed as well as the HDDST alone (Se of 79%, Sp 77%, DA 79%, and PPV 92% with the established criterion).
As noted above, if all discordant results were instead interpreted as negative and indicative of EAS, this raised Sp, and PPV and was therefore deemed the preferred approach. When interpreting discordant results as negative and instead using the optimized test criteria, an ACTH and/or cortisol increase of ≥40% for the CRH test and a ≥69% suppression of cortisol for the HDDST, this yielded Se 65%, Sp 99%, PPV 99%, NPV 48%, and DA 73%. In total, 238 subjects (74%) had concordant test results, with improved performance in this cohort: Se 93%, Sp 98%, PPV 99%, NPV 84%, and DA 95%. Concordant negative results were seen in 77 subjects, 12 of whom had CD.
Using the optimized criteria above, 85 subjects (26%) had discordant test results, 73 (86%) with CD and 12 (14%) with EAS. Among these patients, the CRH test was more likely to be positive in the CD group compared with EAS (85 vs 42%) while the HDDST was more often incorrect in both, with significant cortisol suppression in 15% of CD and 58% of EAS. The PPVs for the CRH test and HDDST in the group with discordant results were 93% and 61%, respectively. The most reliable measure in the discordant group was a cortisol increase of ≥40% during the CRH test, as this was seen in 44 (60%) of the CD patients, but only 1 (8%) of the patients with EAS (PPV 98%). Thus, by first requiring concordant CRH test and HDDST results, and then assessing the cortisol response during the CRH test in those with discordant results, 204 (83%) of patients with CD had positive results and were correctly diagnosed with high accuracy. This approach yielded Se 83%, Sp 97%, PPV 99%, NPV 65%, and DA 87% (Fig. 2).
Figure 2.
Distribution of patients based on test results. The CRH test performed well when positive, with PPV only increasing slightly when a concordant positive HDDST was taken into account. Concordant negative tests achieved NPV 84%, likely still necessitating IPSS. Among patients with discordant results, assessing the presence of a significant cortisol increase (≥40%) during the CRH test was useful, with only 1 EAS patient showing this pattern.
Discussion
In this large series with an unusually high proportion of patients with EAS due to referral patterns, the CRH stimulation test performed better than the HDDST. The optimal response criteria were a ≥40% increase of ACTH and/or cortisol during the CRH test and a ≥69% suppression of cortisol during the HDDST. Using these criteria and requiring concordant results for both tests improved performance further. In the 74% of subjects with concordant CRH test and HDDST results, these optimized cutoffs yielded Se 93%, Sp 98%, DA 95%, and PPV 99%. Among the 26% of patients with discordant test results, a ≥40% cortisol increase during the CRH test was the most accurate measure, seen in 60% of CD and 8% of patients with EAS (PPV 98%) (Fig. 2).
Both HDDST and CRH stimulation perform well when positive, with PPVs of 96% and 97%, respectively, increasing to 99.4% in cases with concordant positive results. By contrast, negative tests are not as informative, with NPVs of 49% and 76% for the HDDST and CRH tests, respectively, only reaching 84% with concordant negative test results. As mentioned, a cortisol increase ≥40% after CRH was the most accurate measure, with only 2/78 (2.6%) of patients with EAS showing this response, yielding a PPV of 99%.
Based on these data, we suggest the following approach when ovine CRH testing is available (Fig. 3). Consistent hypercortisolism should have been documented for at least 1 month prior to testing, to ensure suppression of normal corticotropes. Then, a CRH test is performed first. If a significant cortisol increase (≥40%) is seen, no further testing is needed as the PPV is 98.9% and will only increase to 99.3% with a positive HDDST, while sensitivity will decrease from 74% to 56%. If the test is negative, the patient will likely need to pursue IPSS as the NPV is 76% and would only be increased to 84% with a concordant negative HDDST. When the CRH test yields an isolated ACTH increase ≥40%, an HDDST should be pursued. A positive HDDST (≥69% cortisol suppression) in this situation yielded a PPV of 100% in our cohort. A negative HDDST will constitute a discordant test, which would trigger a need to pursue IPSS. Using this approach, nearly two-thirds (64%) of patients would be diagnosed with CD with near-perfect accuracy (99%), obviating the need for IPSS. If both tests are performed and have concordant negative results, IPSS could potentially be deferred if imaging rules out an apparent pituitary lesion and identifies a suspected intrathoracic culprit lesion. Any pituitary lesion would warrant IPSS in this setting.
Figure 3.
Suggested approach for noninvasive testing when trying to distinguish CD from EAS. The first step is a CRH test. If a cortisol response ≥40% is seen (± an ACTH response) no further testing is needed, PPV 98.9% and will only increase to 99.3% with a positive HDDST, while sensitivity will decrease from 74% to 56%. If an isolated ACTH response ≥40% is seen, step 2 (HDDST) should be performed. If the HDDST is positive (cortisol suppression ≥69%), no further testing is needed as patient likely has CD. In cases of negative tests in step 1 (CRH) or 2 (HDDST), the patient will likely need to proceed to IPSS as the NPV is not high enough to be reliable. Using this approach, 64% of patients could be diagnosed with CD with near perfect accuracy.
If CRH testing is not available, the HDDST should be pursued since it performs well when positive (PPV 96%), providing a strong argument to proceed directly to pituitary surgery in cases with a positive HDDST in combination with a suspected pituitary adenoma on imaging, especially if IPSS is not readily available.
Frete et al have presented a strategy to reduce the need for IPSS by performing imaging of the pituitary and body during noninvasive testing (9). We opted to not include imaging results in the diagnostic algorithm in our cohort for 3 reasons. First, we have not routinely performed imaging for EAS before IPSS, in part because of the higher radiation doses in the earlier time period of the study. Second, the study included data from 1986 to 2019, during which there was vast improvement in technology and imaging capabilities, so that negative early imaging results would potentially bias the outcome negatively. Third, the referral pattern to our institution brings an unusually high proportion of patients who have been difficult to diagnose, in part because of inconclusive imaging results elsewhere (e.g., a lack of an apparent pituitary or intrathoracic lesion). Thus, we cannot shed light on the utility of the interesting approach presented by Frete et al.
All CRH tests in our cohort were performed using ovine CRH, unlike most previous large series which have used human CRH (5, 8, 9, 13) or both analogs (3, 10, 14). The type of CRH used also affects performance, with ovine CRH inducing a stronger, more prolonged increase in ACTH and, particularly, cortisol compared with human CRH (18). While ovine and human CRH bind the CRH receptor with equal affinity, the increased response seen with ovine CRH is thought to be related to its longer half-life and decreased sequestration by CRH-binding protein (3, 5, 14). Thus, it is possible that outcomes for the human CRH test would be different.
Ovine CRH, the only approved form in the United States, has not been available since July 2020. Human CRH became unavailable in 2023. It is not known whether either agent will be reintroduced. Because of this, the need for alternative means of simple, noninvasive testing is critical. An alternative to the CRH stimulation test is the desmopressin stimulation test, which has not been as thoroughly validated (19). The present data show that the other alternative test, HDDST, does not perform as well as the ovine CRH test. However, we lacked complete data for the day 2 serum Dex levels. We now routinely obtain these since they help to assess the possibility of abnormal absorption or unusually rapid metabolism of Dex when evaluating test results. Only including subjects with appropriate serum Dex levels might have altered test performance. Another potential confounder is concurrent oral estrogen use, which can cause elevations in total cortisol levels by increasing corticosteroid-binding globulin. Between 2005 and 2019, the period during which we have detailed data on concurrent medications and their times of administration, only 4 subjects were taking estrogen formulations at the time of their tests. All 4 had CD and 3 responded as expected with cortisol suppression between 72% and 91%. Two of these 3 subjects had their corticosteroid-binding globulin level measured, both of which returned elevated. The fourth subject had a false negative HDDST, only suppressing cortisol by 37%. While it is feasible that estrogen use affects test results, its use was too limited in our cohort to draw any conclusions.
In addition to loss of an important noninvasive test, the loss of CRH also impacts the performance of IPSS, which had an accuracy of 98% in our cohort. In that invasive test, basal values prior to CRH administration do not perform optimally (4). Desmopressin has been used as a promising substitute for CRH in this setting, achieving similar test sensitivity. However, studies evaluating the diagnostic performance of IPSS using desmopressin have included few patients with EAS, leading to imprecise assessments of test specificity (20). As a result, the IPSS test can no longer be considered a “gold standard”.
We hope that the results from this series will encourage pharmaceutical companies to again provide CRH for diagnostic uses, as it performs better than HDDST and seemingly desmopressin as well (19).
Abbreviations
- ACTH
adrenocorticotropin
- CD
Cushing's disease
- CRH
corticotropin-releasing hormone
- CS
Cushing's syndrome
- DA
diagnostic accuracy
- Dex
dexamethasone
- EAS
ectopic ACTH syndrome
- HDDST
high-dose dexamethasone suppression test
- IPSS
inferior petrosal sinus sampling
- NPV
negative predictive value
- PPV
positive predictive value
- Se
sensitivity
- Sp
specificity
- ST
stimulation test
Contributor Information
Henrik Elenius, Diabetes and Endocrinology and Obesity Branch, National Institute of Diabetes and Digestive and Kidney Disorders, National Institutes of Health, Bethesda, MD 20892, USA.
Raven McGlotten, Diabetes and Endocrinology and Obesity Branch, National Institute of Diabetes and Digestive and Kidney Disorders, National Institutes of Health, Bethesda, MD 20892, USA.
Lynnette K Nieman, Diabetes and Endocrinology and Obesity Branch, National Institute of Diabetes and Digestive and Kidney Disorders, National Institutes of Health, Bethesda, MD 20892, USA.
Funding
This work was funded by the intramural division of NIDDK.
Disclosures
H.E. and R.M. have nothing to disclose. L.K.N. receives royalties from UpToDate.
Data Availability
Some or all datasets generated during and/or analyzed during the current study are not publicly available but are available from the corresponding author on reasonable request.
References
- 1. Nieman LK, Biller BM, Findling JW, et al. Treatment of Cushing’s syndrome: an endocrine society clinical practice guideline. J Clin Endocrinol Metab. 2015;100(8):2807‐2831. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Aron DC, Raff H, Findling JW. Effectiveness versus efficacy: the limited value in clinical practice of high dose dexamethasone suppression testing in the differential diagnosis of adrenocorticotropin-dependent Cushing’s syndrome. J Clin Endocrinol Metab. 1997;82(6):1780‐1785. [DOI] [PubMed] [Google Scholar]
- 3. Barbot M, Trementino L, Zilio M, et al. Second-line tests in the differential diagnosis of ACTH-dependent Cushing’s syndrome. Pituitary. 2016;19(5):488‐495. [DOI] [PubMed] [Google Scholar]
- 4. Oldfield EH, Doppman JL, Nieman LK, et al. Petrosal sinus sampling with and without corticotropin-releasing hormone for the differential diagnosis of Cushing’s syndrome. N Engl J Med. 1991;325(13):897‐905. [DOI] [PubMed] [Google Scholar]
- 5. Detomas M, Ritzel K, Nasi-Kordhishti I, et al. Outcome of CRH stimulation test and overnight 8 mg dexamethasone suppression test in 469 patients with ACTH-dependent Cushing’s syndrome. Front Endocrinol (Lausanne). 2022;13:955945. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Hall WA, Luciano MG, Doppman JL, Patronas NJ, Oldfield EH. Pituitary magnetic resonance imaging in normal human volunteers: occult adenomas in the general population. Ann Intern Med. 1994;120(10):817‐820. [DOI] [PubMed] [Google Scholar]
- 7. Ilias I, Torpy DJ, Pacak K, Mullen N, Wesley RA, Nieman LK. Cushing’s syndrome due to ectopic corticotropin secretion: twenty years’ experience at the national institutes of health. J Clin Endocrinol Metab. 2005;90(8):4955‐4962. [DOI] [PubMed] [Google Scholar]
- 8. Ritzel K, Beuschlein F, Berr C, et al. ACTH after 15 minutes distinguishes between Cushing’s disease and ectopic Cushing’s syndrome: a proposal for a short and simple CRH test. Eur J Endocrinol. 2015;173(2):197‐204. [DOI] [PubMed] [Google Scholar]
- 9. Frete C, Corcuff JB, Kuhn E, et al. Non-invasive diagnostic strategy in ACTH-dependent Cushing’s syndrome. J Clin Endocrinol Metab. 2020;105(10):3273‐3284. [DOI] [PubMed] [Google Scholar]
- 10. Invitti C, Pecori Giraldi F, de Martin M, Cavagnini F. Diagnosis and management of Cushing’s syndrome: results of an Italian multicentre study. Study group of the Italian Society of Endocrinology on the pathophysiology of the hypothalamic-pituitary-adrenal axis. J Clin Endocrinol Metab. 1999;84(2):440‐448. [DOI] [PubMed] [Google Scholar]
- 11. Nieman LK, Oldfield EH, Wesley R, Chrousos GP, Loriaux DL, Cutler GB Jr. A simplified morning ovine corticotropin-releasing hormone stimulation test for the differential diagnosis of adrenocorticotropin-dependent Cushing's syndrome. J Clin Endocrinol Metab 1993;77(5):1308‐1312. [DOI] [PubMed] [Google Scholar]
- 12. Reimondo G, Paccotti P, Minetto M, et al. The corticotrophin-releasing hormone test is the most reliable noninvasive method to differentiate pituitary from ectopic ACTH secretion in Cushing’s syndrome. Clin Endocrinol (Oxf). 2003;58(6):718‐724. [DOI] [PubMed] [Google Scholar]
- 13. Arnaldi G, Tirabassi G, Papa R, et al. Human corticotropin releasing hormone test performance in the differential diagnosis between Cushing's disease and pseudo-Cushing state is enhanced by combined ACTH and cortisol analysis. Eur J Endocrinol. 2009;160(6):891‐898. [DOI] [PubMed] [Google Scholar]
- 14. Pecori Giraldi F, Invitti C, Cavagnini F; Study Group of the Italian Society of Endocrinology on the Pathophysiology of the Hypothalamic-pituitary-adrenal axis . The corticotropin-releasing hormone test in the diagnosis of ACTH-dependent Cushing’s syndrome: a reappraisal. Clin Endocrinol (Oxf). 2001;54(5):601‐607. [DOI] [PubMed] [Google Scholar]
- 15. Isidori AM, Kaltsas GA, Mohammed S, et al. Discriminatory value of the low-dose dexamethasone suppression test in establishing the diagnosis and differential diagnosis of Cushing's syndrome. J Clin Endocrinol Metab. 2003;88(11):5299‐5306. [DOI] [PubMed] [Google Scholar]
- 16. Dichek HL, Nieman LK, Oldfield EH, Pass HI, Malley JD, Cutler GB Jr. A comparison of the standard high dose dexamethasone suppression test and the overnight 8-mg dexamethasone suppression test for the differential diagnosis of adrenocorticotropin-dependent Cushing’s syndrome. J Clin Endocrinol Metab 1994;78(2):418‐422. [DOI] [PubMed] [Google Scholar]
- 17. Lindsay JR, Shanmugam VK, Oldfield EH, Remaley AT, Nieman LK. A comparison of immunometric and radioimmunoassay measurement of ACTH for the differential diagnosis of Cushing’s syndrome. J Endocrinol Invest. 2006;29(11):983‐988. [DOI] [PubMed] [Google Scholar]
- 18. Nieman LK, Cutler GB Jr, Oldfield EH, Loriaux DL, Chrousos GP. The ovine corticotropin-releasing hormone (CRH) stimulation test is superior to the human CRH stimulation test for the diagnosis of Cushing’s disease. J Clin Endocrinol Metab 1989; 69(1):165‐169. [DOI] [PubMed] [Google Scholar]
- 19. Vassiliadi DA, Tsagarakis S. Diagnosis of endocrine disease: the role of the desmopressin test in the diagnosis and follow-up of Cushing’s syndrome. Eur J Endocrinol. 2018;178(5):R201‐RR14. [DOI] [PubMed] [Google Scholar]
- 20. Valizadeh M, Ahmadi AR, Ebadinejad A, Rahmani F, Abiri B. Diagnostic accuracy of bilateral inferior petrosal sinus sampling using desmopressin or corticotropic- releasing hormone in ACTH-dependent Cushing’s syndrome: a systematic review and meta-analysis. Rev Endocr Metab Disord. 2022;23(5):881‐892. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
Some or all datasets generated during and/or analyzed during the current study are not publicly available but are available from the corresponding author on reasonable request.