`
`A R T I C L E
`
`ACTH Stimulation Tests for the Diagnosis of Adrenal
`Insufficiency: Systematic Review and Meta-Analysis
`
`Naykky Singh Ospina,* Alaa Al Nofal,* Irina Bancos, Asma Javed,
`Khalid Benkhadra, Ekta Kapoor, Aida N. Lteif, Neena Natt,
`and M. Hassan Murad
`
`Evidence-Based Practice Research Program (N.S.O., A.A.N., K.B., M.H.M.), Mayo Clinic, Rochester,
`Minnesota; Knowledge and Evaluation Research Unit (N.S.O., K.B., M.H.M.), Mayo Clinic, Rochester,
`Minnesota; Division of Endocrinology, Diabetes, Metabolism, and Nutrition (N.S.O., N.N., I.B.), Mayo
`Clinic, Rochester, Minnesota; Division of Pediatric Endocrinology and Metabolism (A.A.N., A.J., A.N.L.),
`Mayo Clinic, Rochester, Minnesota; Division of General Internal Medicine (E.K.), Mayo Clinic, Rochester,
`Minnesota 55905
`
`Context: The diagnosis of adrenal insufficiency is clinically challenging and often requires ACTH
`stimulation tests.
`
`Objective: To determine the diagnostic accuracy of the high- (250 mcg) and low- (1 mcg) dose ACTH
`stimulation tests in the diagnosis of adrenal insufficiency.
`
`Methods: We searched six databases through February 2014. Pairs of independent reviewers se-
`lected studies and appraised the risk of bias. Diagnostic association measures were pooled across
`studies using a bivariate model.
`
`Data Synthesis: For secondary adrenal insufficiency, we included 30 studies enrolling 1209 adults
`and 228 children. High- and low-dose ACTH stimulation tests had similar diagnostic accuracy in
`adults and children using different peak serum cortisol cutoffs. In general, both tests had low
`sensitivity and high specificity resulting in reasonable likelihood ratios for a positive test (adults:
`high dose, 9.1; low dose, 5.9; children: high dose, 43.5; low dose, 7.7), but a fairly suboptimal
`likelihood ratio for a negative test (adults: high dose, 0.39; low dose, 0.19; children: high dose, 0.65;
`low dose, 0.34). For primary adrenal insufficiency, we included five studies enrolling 100 patients.
`Data were only available to estimate the sensitivity of high dose ACTH stimulation test (92%; 95%
`confidence interval, 81–97%).
`
`Conclusion: Both high- and low-dose ACTH stimulation tests had similar diagnostic accuracy. Both
`tests are adequate to rule in, but not rule out, secondary adrenal insufficiency. Our confidence in
`these estimates is low to moderate because of the likely risk of bias, heterogeneity, and imprecision.
`(J Clin Endocrinol Metab 101: 427– 434, 2016)
`
`Adrenal
`
`insufficiency is a life-threatening disorder
`characterized by failure of adrenal cortisol produc-
`tion either from adrenal disease (primary adrenal insuffi-
`ciency, PAI) or deficiency of ACTH (secondary adrenal
`insufficiency, SAI) (1, 2). Prompt diagnosis is important
`because adequate hormonal replacement therapy is life-
`saving (1, 3–5). Even with early diagnosis and institution
`
`of therapy, patients with the diagnosis of adrenal insuffi-
`ciency have higher mortality (6, 7), decreased quality of
`life (8, 9), and increased risk of adrenal crisis (10, 11).
`Adrenal insufficiency may present with nonspecific
`symptoms (eg, fatigue, weight loss, nausea, loss of appe-
`tite), resulting in a potential delay in diagnosis. In a cross-
`sectional study of 216 patients with both primary and
`
`ISSN Print 0021-972X ISSN Online 1945-7197
`Printed in USA
`Copyright © 2016 by the Endocrine Society
`Received March 18, 2015. Accepted December 2, 2015.
`First Published Online December 9, 2015
`
`* N.S.O. and A.A.N. contributed equally to this study.
`Abbreviations: CI, confidence interval; LR, likelihood ratio; OR, odds ratio; PAI, primary
`adrenal insufficiency; SAI, secondary adrenal insufficiency.
`
`Amerigen Exhibit 1181
`Amerigen v. Janssen IPR2016-00286
`
`doi: 10.1210/jc.2015-1700
`
`J Clin Endocrinol Metab, February 2016, 101(2):427– 434
`
`press.endocrine.org/journal/jcem 427
`
`
`
`428
`
`Ospina et al
`
`Diagnosis of Adrenal Insufficiency
`
`J Clin Endocrinol Metab, February 2016, 101(2):427– 434
`
`secondary adrenal insufficiency, 47% had symptoms for
`more than 1 year before diagnosis and 20% had symptoms
`for more than 5 years before diagnosis. The correct diag-
`nosis was established during the initial medical encounter
`in only 15% of patients (12).
`Once adrenal insufficiency is suspected, biochemical
`testing is required to confirm the diagnosis (1). The initial
`step in evaluation is the measurement of baseline morning
`serum cortisol and an ACTH stimulation test. The insulin
`hypoglycemia test (insulin tolerance test) is considered the
`gold standard for the diagnosis of SAI. This test may not
`be possible in all situations because it requires medical
`supervision and can be unsafe in patients with history of
`seizures, cardiac disease, or the elderly (1, 13). The single-
`dose overnight metyrapone stimulation test is another
`confirmatory dynamic test that has been used in the past
`for the diagnosis of adrenal insufficiency. Through its in-
`hibition of 11--hydroxylase, metyrapone results in de-
`creased cortisol levels with subsequent feedback stimula-
`tion of ACTH and accumulation of the pre-enzyme block
`substrate 11-deoxycortisol. This test has a similar diag-
`nostic performance to the insulin hypoglycemia test and is
`a potential alternative when there is a contraindication to
`the insulin hypoglycemia test (13, 14).
`The insulin-induced hypoglycemia test and the single-
`dose overnight metyrapone tests are expensive, cumber-
`some, and have potential significant side effects compared
`to the ACTH stimulation tests. These latter tests assess the
`serum cortisol response to acute ACTH stimulation with
`either a 250-g dose (high or standard dose) or 1-g dose
`(low dose) (1, 13).
`The objective of this systematic review and meta-anal-
`ysis was to compare the diagnostic accuracy of the high-
`and low-dose ACTH stimulation tests in patients with ei-
`ther primary or secondary adrenal insufficiency.
`
`Materials and Methods
`
`Eligibility criteria
`Inclusion criteria for eligible studies were predefined in a
`study protocol. We included observational and randomized
`studies that assessed the diagnostic accuracy of high- and low-
`dose ACTH stimulation tests for the diagnosis of PAI or SAI
`when compared to a gold standard. In cases of PAI the gold
`standard included clinical features, serum cortisol, serum ACTH
`levels, and follow-up. In SAI, both the insulin tolerance test and
`metyrapone test were considered gold standards. Exclusion cri-
`teria included case series (uncontrolled studies), review studies,
`and studies that evaluated patients with critical illness; patients
`with expected secondary adrenal insufficiency because of exog-
`enous steroid use (eg, patients with autoimmune diseases treated
`with steroids, patients with asthma) or steroid therapy not dis-
`continued before adrenal insufficiency testing (with no restric-
`tion regarding time of discontinuation).
`
`Search strategy
`We conducted a comprehensive search of several databases
`without language restriction from each database’s earliest incep-
`tion to February 28, 2014. The databases included Ovid Medline
`In-Process & Other Non-Indexed Citations, Ovid MEDLINE,
`Ovid EMBASE, Ovid Cochrane Central Register of Controlled
`Trials, Ovid Cochrane Database of Systematic Reviews, and Sco-
`pus. The search strategy was designed and conducted by an ex-
`perienced librarian with input from the study’s principal inves-
`tigator (M.H.M.). Controlled vocabulary supplemented with
`keywords was used to search for adrenal insufficiency. The de-
`tails of the search are available in the supplemental material.
`Cross-referencing with previously published systematic reviews
`and contacting content experts were also performed to supple-
`ment the electronic search.
`Working independently and in duplicate, the reviewers
`screened the available abstracts (N.SO., A.A., I.B., A.J., K.B.,
`E.K.). Articles in full text were then retrieved and were reviewed
`independently and in duplicate for eligibility. Disagreements be-
`tween reviewers were resolved by consensus.
`
`Data extraction for systematic review
`Working independently and in duplicate, data from the in-
`cluded studies were extracted using a standardized data extrac-
`tion sheet, including baseline information about included studies
`and the number of patients with true-positive, true-negative,
`false-positive, and false-negative results. In cases where the re-
`quired data were not present in the published manuscript, au-
`thors were contacted for additional information (four authors
`were contacted with response obtained from one author).
`
`Quality of the studies
`Critical appraisal of the included studies was performed in-
`dependently and in duplicate following the Quality Assessment
`of Diagnostic Accuracy Studies instrument. This includes the
`assessment of the risk of bias and applicability in the following
`domains: patient selection, index test, reference standard, and
`flow and timing. This tool includes signaling questions to help
`reviewers assess the risk of bias. One domain of the tool evaluates
`patient selection and the methods used for enrolling patients (eg,
`consecutive or random sample) and the appropriateness of ex-
`clusion criteria. Another domain evaluates the index test and
`whether it was interpreted without knowledge of the standard
`reference. A domain about the reference standard evaluates
`whether the interpretation of the reference standard was per-
`formed without knowing the results of the index text. Finally, the
`domain of flow and timing focuses on knowing when the refer-
`ence standard was performed and in how many patients (15, 16).
`Cases in which the reviewers’ assessment of the risk of bias dif-
`fered were resolved by consensus.
`
`Statistical analysis
`Diagnostic estimates from included studies were pooled by
`fitting a two-level mixed logistic regression model with indepen-
`dent binomial distributions for the true positives and true neg-
`atives. These distributions were conditional on the sensitivity
`and specificity in each study. We also used a bivariate normal
`model for the logit transforms of sensitivity and specificity be-
`tween studies (17, 18). The analysis was done using STATA,
`version 13 (StataCorp, College Station, TX). Heterogeneity be-
`tween the studies was assessed using the I2 statistic. We report
`
`
`
`doi: 10.1210/jc.2015-1700
`
`press.endocrine.org/journal/jcem
`
`429
`
`sensitivity, specificity, likelihood ratios, and diagnostic odds ra-
`tios (ORs), with 95% confidence intervals (CIs).
`
`vant references of which 35 studies were included (30 in
`SAI, five in PAI).
`
`Results
`
`Search results
`The results of the systematic search are shown in Figure
`1. The systematic search identified 1284 potentially rele-
`
`Risk of bias
`Using the Quality Assessment of Diagnostic Accuracy
`Studies-2 instrument, all included studied had moderate
`risk of bias as shown in Supplemental Figure 1. This con-
`clusion is mainly driven by unclear or inappropriate pa-
`tient selection and referral bias leading to high prevalence.
`
`Unique abstracts idenƟfied
`
`through database searches
`
`(1284)
`
`Abstracts IdenƟfied - 1390
`
`Ovid Medline In-Process & Other
`Non-Indexed CitaƟons, Ovid
`MEDLINE 713
`
`Ovid EMBASE 466
`
`Ovid Cochrane Central Register
`of Controlled Trials 14
`
`Ovid Cochrane Database of
`SystemaƟc Reviews 27
`
`Abstracts screened
`
`(1284)
`
`Abstracts excluded
`
`(not meeƟng inclusion criteria)
`
`(866)
`
`Full-text arƟcles
`
`assessed for eligibility
`
`(418)
`
`Full-text arƟcles excluded
`
`(not meeƟng inclusion criteria)
`
`(395)
`
`Manual review of references
`
`(12)
`
`Studies included in
`quanƟtaƟve synthesis
`(meta-analysis)
`(35)
`
`30 Secondary Adrenal
`insufficiency
`
`5 Primary Adrenal
`insufficiency
`
`
`
`Figure 1. Study selection.
`
`
`
`IdenƟficaƟon
`
`Screening
`
`Eligibility
`
`Included
`
`
`
`430
`
`Ospina et al
`
`Diagnosis of Adrenal Insufficiency
`
`J Clin Endocrinol Metab, February 2016, 101(2):427– 434
`
`Otherwise, the studies had low risk of bias in the domains
`of index test, reference standard, and flow and timing.
`
`Secondary adrenal insufficiency
`We identified 30 studies (19 – 48) assessing the diag-
`nostic performance of the ACTH stimulation test in pa-
`tients with suspected SAI. Supplemental Tables 1 and 2
`summarize the characteristics of these studies that enrolled
`adults and children, respectively. These studies enrolled a
`total of 1437 patients with a prevalence of SAI of 36%
`(35% in adults and 38% in children). Most studies ad-
`ministered ACTH IV.
`We included studies that defined whether the test was
`positive or negative based on predefined cutoffs that the
`serum cortisol level had to exceed at any time after ACTH
`administration, “peak cortisol level.” Other studies used
`a specific time (30 or 60 minutes) to assess for this pre-
`defined serum cortisol value to determine whether the test
`was positive or negative. The distribution of the included
`studies in terms of test used and cutoff is as follows:
`
`• The overall analysis for the accuracy of high-dose
`ACTH stimulation test in adults included 29 datasets
`(19, 21–23, 25–29, 31, 33– 40, 42, 44 – 46, 48). Six
`studies were included in the analysis of high-dose
`ACTH in adults using 500 nmol/L at 30 minutes as a
`cutoff (22, 25, 28, 34, 39, 44), 14 studies used a 500
`nmol/L peak serum cortisol value as a cutoff (19, 21,
`23, 26, 29, 33, 36 –38, 40, 42, 45, 46, 48), and eight
`a serum cortisol cutoff of 550 nmol/L (21, 23, 31, 33,
`35, 38, 45, 48).
`• The overall analysis for the accuracy of the low-dose
`ACTH stimulation test in adults included 19 datasets
`(19, 20, 23–25, 29, 35, 37, 38, 40, 43, 45, 46, 48).
`Eleven studies used a 500-nmol/L peak serum cortisol
`value for the low-dose ACTH stimulation test in adults
`(19, 20, 23, 29, 37, 38, 40, 43, 45, 46, 48); six used a
`peak serum cortisol level of 550 nmol/L as the cutoff
`value (23, 35, 38, 43, 45, 48).
`• The overall analysis of the low-dose ACTH stimulation
`test in children included five datasets (30, 32, 41, 47).
`Three studies evaluated the low-dose ACTH stimula-
`tion test in children with a peak cortisol of 500 nmol/L
`(32, 41, 47) and two a peak cortisol level of 550 nmol/
`liter as the cutoff (30, 41). The overall analysis of the
`high-dose ACTH stimulation test in children included
`four datasets (30, 41, 47). Two studies evaluated the
`high-dose ACTH in children using a peak of 500 nmol/
`liter (41, 47) and two studies with a peak cortisol of 550
`nmol/liter (30, 41).
`
`Diagnostic performance in SAI
`The diagnostic performance for the high- and low-dose
`ACTH stimulation test in adults and children according to
`three different test cutoffs are summarized in Table 1 and
`2. Summary receiving operator characteristics curves are
`in Figures 2 and 3 for low and high dose, respectively.
`Studies were excluded if patients on long-acting steroid
`were included or, because of the lack of a predefined gold
`standard, reported equivocal results for the gold standard
`or used a gold standard that was not compatible with the
`inclusion criteria (14, 49 – 60).
`In general, both tests had low and high specificity re-
`sulting in reasonable likelihood ratios for a positive test
`(adults: high dose, 9.1; low dose, 5.9; children: high dose,
`43.5; low dose, 7.7), but a fairly suboptimal likelihood
`ratio (LR) for a negative test (adults: high dose, 0.39; low
`dose, 0.19; children: high dose, 0.65; low dose, 0.34). Both
`high- and low-dose tests had moderate accuracy overall
`(diagnostic ORs ranging from 23 to 67) primarily because
`of the low sensitivity. However, there was no statistically
`significant difference between accuracy of the high- and
`the low-dose tests when comparing diagnostic ORs. The
`analysis was associated with significant heterogeneity,
`which is common in diagnostic meta-analysis. A summary
`
`Table 1. Meta-Analysis Results: ACTH Stimulation
`Tests for the Diagnosis of Secondary Adrenal
`Insufficiency
`
`Estimate
`
`95% CI
`
`Adult High-Dose ACTH
`Stimulation Test
`Sensitivity
`Specificity
`Likelihood ratio for positive test
`Likelihood ratio for negative test
`Diagnostic odds ratio
`Adult Low-Dose ACTH
`Stimulation Test
`Sensitivity
`Specificity
`Likelihood ratio for positive test
`Likelihood ratio for negative test
`Diagnostic odds ratio
`Children High-Dose ACTH
`Stimulation Test
`Sensitivity
`Specificity
`Likelihood ratio for positive test
`Likelihood ratio for negative test
`Diagnostic odds ratio
`Children Low-Dose ACTH
`Stimulation Test
`Sensitivity
`Specificity
`Likelihood ratio for positive test
`Likelihood ratio for negative test
`Diagnostic odds ratio
`
`0.64
`0.93
`9.1
`0.39
`23
`
`0.83
`0.86
`5.9
`0.19
`30
`
`0.36
`0.99
`43.5
`0.65
`67
`
`0.69
`0.91
`7.7
`0.34
`23
`
`0.52– 0.73
`0.89 – 0.96
`5.7–14.6
`0.30 – 0.52
`13– 42
`
`0.75– 0.89
`0.78 – 0.91
`3.8 – 8.9
`0.13– 0.29
`18 –50
`
`0.10 – 0.73
`0.81– 0.99
`1–1891.2
`0.36 –1.15
`1– 4152
`
`0.28 – 0.93
`0.63– 0.98
`1.3– 44.8
`0.10 –1.18
`2–313
`
`
`
`doi: 10.1210/jc.2015-1700
`
`press.endocrine.org/journal/jcem
`
`431
`
`Table 2. ACTH Stimulation Tests for the Diagnosis of Secondary Adrenal Insufficiency Based on Cortisol Cutoff
`
`Adults
`
`High-Dose ACTH Test
`
`Cortisol Cutoff
`(nmol/liter)
`
`LRⴙ
`
`LRⴚ
`
`Diagnostic OR
`
`No. of
`Studies
`
`LRⴙ
`
`LRⴚ
`
`Diagnostic OR
`
`No. r of
`Studies
`
`P Value
`(for Difference)
`
`Low-Dose ACTH Test
`
`500 –30 minutes
`500 –peak
`550 –peak
`
`6.3 (2.5–16)
`12.4 (6.7–23.0)
`6.4 (3.4 –12)
`
`0.32 (0.20 – 0.51)
`0.48 (0.32– 0.72)
`0.36 (0.21– 0.61)
`
`20 (5–75)
`26 (11– 60)
`18 (8 – 43)
`
`6
`14
`8
`
`NR
`7.1 (4.3–11.6)
`3.8 (1.5–9.4)
`
`NR
`0.21 (0.13– 0.33)
`0.23 (0.11– 0.49)
`
`NR
`34 (17– 68)
`16 (6 – 40)
`
`NR
`11
`6
`
`NA
`.631
`.855
`
`Children
`
`High-Dose ACTH Test
`
`Low-Dose ACTH Test
`
`500 –peak
`550 –peak
`
`15.96 (2.12–120.04)
`6.1 (1.09 –34.17)
`
`0.37 (0.01–12.95)
`0.78 (0.58 –1.06)
`
`40.67 (1.1–1424.1)
`7.96 (1.2–51.4)
`
`2
`2
`
`18.3 (2.04 –164.73)
`4.3 (2.65–7.06)
`
`0.31 (0.5–1.9)
`0.2 (0.02–1.92)
`
`93.63 (14.6 – 620.1)
`24.8 (1.73–356.9)
`
`3
`2
`
`.686
`.494
`
`Abbreviations: LR⫹, likelihood ratio of a positive test; LR⫺, likelihood ratio of a negative test; NA, not applicable; NR, not reported.
`Heterogeneity values (I2)–adults: high-dose 30-minute cutoff, 32%; high-dose 500 peak cut off, 90%; high-dose 550 peak cutoff: 81% low-dose
`500 peak cut off: 88%; low-dose 550 peak cut off, 93%. Children: high-dose 500 peak cutoff, 60%; high-dose 550 peak cutoff, 0%; low-dose
`500 peak cutoff, 0%; low-dose 550 peak cutoff. 66%.
`
`of the meta-analysis results is shown in Tables 1 and 2. The
`receiver operator characteristic (61) curve for the high-
`and low-dose ACTH stimulation test in adults are found
`in Figures 2 and 3, respectively.
`
`Primary adrenal insufficiency
`We identified five studies (62– 66) investigating the di-
`agnostic performance of the high-dose ACTH stimulation
`test for the diagnosis of PAI. The characteristics of these
`studies are summarized in Supplemental Table 3.
`
`Diagnostic performance in PAI
`Data were insufficient to estimate specificity, likeli-
`hood, and diagnostic ORs. Only the sensitivity (the rate of
`a positive test among patients with confirmed PAI) was
`estimable and was 92% (95% CI, 81–97%).
`
`Discussion
`
`This systematic review and meta-analysis aimed at iden-
`tifying the diagnostic accuracy of ACTH stimulation test
`
`Figure 2. Receiver operator characteristic curve– high-dose ACTH
`stimulation test for secondary adrenal insufficiency. HSROC,
`hierarchical summary receiver operating characteristic.
`
`Figure 3. Receiver operator characteristic curve–low-dose ACTH
`simulation test for secondary adrenal insufficiency. HSROC, hierarchical
`summary receiver operating characteristic.
`
`
`
`432
`
`Ospina et al
`
`Diagnosis of Adrenal Insufficiency
`
`J Clin Endocrinol Metab, February 2016, 101(2):427– 434
`
`in patients with PAI and SAI. We demonstrated that both
`high- and low-dose stimulation tests had similar diagnos-
`tic accuracy in SAI. Both tests in general had moderate
`accuracy because of low sensitivity. Therefore, they are
`more helpful in ruling in the condition when positive.
`However, they are not as reliable in ruling out the condi-
`tion when negative. We demonstrated overall consistency
`of accuracy measures across different peak cortisol cutoffs
`and in children and adults. Data in PAI are insufficient to
`estimate diagnostic accuracy, and one can only conclude
`that the high-dose test had high sensitivity of 92%. Many
`of these PAI patients may have had congenital adrenal
`hyperplasia; however, the available studies did not pro-
`vide data to distinguish these patients and allow estima-
`tion of diagnostic accuracy measures specific to them. The
`quality of evidence (confidence in estimates) generated
`from this analysis is moderate in PAI (because of hetero-
`geneity) and low to moderate in SAI (because of hetero-
`geneity and increased risk of bias).
`Two previous systematic reviews attempted to evaluate
`the diagnostic accuracy of ACTH stimulation tests (67,
`68). Dorin and colleagues reported high sensitivity
`(97.5%) and specificity (96.5%) for the high-dose ACTH
`stimulation test in the diagnosis of primary adrenal insuf-
`ficiency. However, they included studies in which healthy
`volunteers and persons without endocrine disease were used
`as a reference. We did not find any studies that assessed the
`performance accuracy of the high-dose ACTH test in pa-
`tients with suspected PAI and, therefore, are only able to
`report the sensitivity based on studies that included patients
`with known disease. Data from such cohorts exaggerate di-
`agnostic accuracy measures (compared to the optimal study
`design that includes patients with suspected disease).
`Dorin and colleagues noted a positive LR of 11.5 and
`a negative LR of 0.45 for the high-dose ACTH stimulation
`test (at a set specificity of 95%) for evaluating SAI, which
`is comparable with our results. We found no statistically
`significant difference between the diagnostic performance
`of the high- vs low-dose ACTH simulation test for the
`diagnosis of SAI, which is consistent with previous reports
`(67). Our results are in contrast to the findings of Kazlauz-
`kaite and colleagues (68), who performed a systematic re-
`view based on patient level data and reported better perfor-
`mance of the 30-minute cortisol values obtained during low
`dose ACTH stimulation test when compared to the high-
`dose ACTH stimulation test, even when excluding patients
`with steroid use from the analysis. Differences in methods
`(patient level data) and number of included studies (13)
`should be taken into consideration when comparing the re-
`sults of this meta-analysis to prior reports.
`The limitations of the current available literature are
`mostly related to significant variability in 1) the pretest prob-
`
`ability of the diagnosis of adrenal insufficiency in the in-
`cluded populations, 2) the use of different cortisol assays
`(mostly radioimmunoassays in the included studies), and 3)
`different cutoff values for the interpretation of the test results
`(time of measurement and value) in both the index test
`(ACTH stimulation test) and the gold standard (insulin tol-
`erance test and/or metyrapone test). In addition, technical
`differences should also be considered in future studies in
`which the diagnostic performance of the different doses of
`ACTH stimulation tests are evaluated such as the prepara-
`tion of the 1-mcg dose of ACTH and the length of tubing used
`for administration (69). These differences are reflected in the
`significant level of heterogeneity that we encountered be-
`tween studies and the wide CIs for some of the estimates.
`In addition, the quality assessment of the included stud-
`ies showed a moderate risk for bias due to patient selection
`and concern of applicability of the results due to the per-
`formance and interpretation of the index test.
`Despite these limitations, we believe the results of our
`study provide interesting insights for the diagnostic perfor-
`mance of ACTH stimulation studies in diagnosing adrenal
`insufficiency. First, when considering the diagnosis, physi-
`cians should have an understanding of the pretest probability
`of disease. This is important because the presented likelihood
`ratios of both the high- and low-dose ACTH stimulation tests
`suggest that, although helpful, these tests are not perfect and
`can be misleading in some cases. Second, knowledge of the
`limitations of the test and possible responsible factors (cor-
`tisol assay used, time, and cut off used for interpretation)
`should be considered during the medical decision-making
`process. The use of gold standard tests might be needed when
`the results of the ACTH stimulation tests are equivocal or
`when the test is negative in the setting of high clinical suspi-
`cion. For example, in a patient with history of pituitary dis-
`ease who presents with fatigue and deficiency of other pitu-
`itary hormones, most clinicians would be highly suspicious
`of SAI (high risk for SAI). As shown, in Supplemental Figure
`2A, a negative test in that patient would not decrease the
`likelihood of disease to a level at which most physicians
`would be comfortable excluding SAI.
`On the other hand, in a patient with fatigue without any
`signs or risk factors for SAI and an equivocal serum morning
`cortisol (3–18 mcg/dL) (low risk for SAI), a negative result
`will significantly decrease the probability of disease (Supple-
`mental Figure 2B). Unfortunately, there are no validated
`tools to establish a reliable pretest probability for adrenal
`insufficiency and this only depends on clinical experience.
`A taskforce from the Endocrine Society will provide the
`clinical context and interpretation to our findings.
`
`
`
`doi: 10.1210/jc.2015-1700
`
`press.endocrine.org/journal/jcem
`
`433
`
`Conclusion
`
`Both high- and low-dose ACTH stimulation tests have sim-
`ilar diagnostic accuracy. Both tests are adequate to rule in,
`but not rule out, secondary adrenal insufficiency. Our con-
`fidence in these estimates is low-moderate because of the risk
`of bias of
`the included studies, heterogeneity, and
`imprecision.
`
`Acknowledgments
`
`We thank Larry J. Prokop for his help in designing and executing
`the search strategy.
`
`Address all correspondence and requests for reprints to:
`Mohammad Hassan Murad, MD, MPH, Division of Preven-
`tive, Occupational and Aerospace Medicine, Mayo Clinic,
`200 First Street SW, Rochester, MN 55905. E-mail: Murad.
`Mohammad@mayo.edu.
`Funding for this study was provided by The Endocrine
`Society.
`Disclosure Summary: the authors have nothing to disclose.
`
`References
`
`1. Bancos I, Hahner S, Tomlinson J, Arlt W. 2014 diagnosis and man-
`agement of adrenal insufficiency. Lancet Diabetes Endocrinol.
`2015;3:216 –226.
`2. Charmandari E, Nicolaides NC, Chrousos GP. Adrenal insuffi-
`ciency. Lancet. 2014;383:2152–2167.
`3. Lovas K, Husebye ES. High prevalence and increasing incidence of
`Addison’s disease in western Norway. Clin Endocrinol (Oxf). 2002;
`56:787–791.
`4. Regal M, Paramo C, Sierra SM, Garcia-Mayor RV. Prevalence and
`incidence of hypopituitarism in an adult Caucasian population in
`northwestern Spain. Clin Endocrinol (Oxf). 2001;55:735–740.
`5. Arlt W, Allolio B. Adrenal insufficiency. Lancet. 2003;361:1881–1893.
`6. Bensing S, Brandt L, Tabaroj F, et al. Increased death risk and altered
`cancer incidence pattern in patients with isolated or combined au-
`toimmune primary adrenocortical insufficiency. Clin Endocrinol
`(Oxf). 2008;69:697–704.
`7. Burman D, Morrison GD. Too many infant deaths. Br Med J. 1963;
`2:1419 –1420.
`8. Benson S, Neumann P, Unger N, et al. Effects of standard gluco-
`corticoid replacement therapies on subjective well-being: a random-
`ized, double-blind, crossover study in patients with secondary ad-
`renal insufficiency. Eur J Endocrinol. 2012;167:679 – 685.
`9. Hahner S, Loeffler M, Fassnacht M, et al. Impaired subjective health
`status in 256 patients with adrenal insufficiency on standard therapy
`based on cross-sectional analysis. J Clin Endocrinol Metab. 2007;
`92:3912–3922.
`10. Hahner S, Loeffler M, Bleicken B, et al. Epidemiology of adrenal
`crisis in chronic adrenal insufficiency: the need for new prevention
`strategies. Eur J Endocrinol. 2010;162:597– 602.
`11. White K, Arlt W. Adrenal crisis in treated Addison’s disease: a pre-
`dictable but under-managed event. Eur J Endocrinol. 2010;162:
`115–120.
`12. Bleicken B, Hahner S, Ventz M, Quinkler M. Delayed diagnosis of
`adrenal insufficiency is common: a cross-sectional study in 216 pa-
`tients. Am J Med Sci. 2010;339:525–531.
`13. de Miguel Novoa P, Vela ET, Garcia NP, et al. Guidelines for the
`diagnosis and treatment of adrenal insufficiency in the adult. En-
`docrinol Nutr. 2014;1(61 Suppl):1–35.
`
`14. Fiad TM, Kirby JM, Cunningham SK, McKenna TJ. The overnight
`single-dose metyrapone test is a simple and reliable index of the
`hypothalamic-pituitary-adrenal axis. Clin Endocrinol (Oxf). 1994;
`40:603– 609.
`15. Whiting PF, Rutjes AW, Westwood ME, et al. QUADAS-2: a revised
`tool for the quality assessment of diagnostic accuracy studies. Ann
`Intern Med. 2011;155:529 –536.
`16. Swiglo BA, Murad MH, Schunemann HJ, et al. A case for clarity,
`consistency, and helpfulness: state-of-the-art clinical practice guide-
`lines in endocrinology using the grading of recommendations, as-
`sessment, development, and evaluation system. J Clin Endocrinol
`Metab. 2008;93:666 – 673.
`17. Harbord R. 2008 Stata module for meta-analysis of diagnostic ac-
`curacy. In: Statistical Software Components. April 2008; Boston
`College Department of Economics.
`18. Harbord RM, Deeks JJ, Egger M, Whiting P, Sterne JA. A unification
`of models for meta-analysis of diagnostic accuracy studies. Biosta-
`tistics. 2007;8:239 –251.
`19. Abdu TA, Elhadd TA, Neary R, Clayton RN. Comparison of the
`low dose short Synacthen test (1 microg), the conventional dose
`short Synacthen test (250 microg), and the insulin tolerance test for
`assessment of the hypothalamo-pituitary-adrenal axis in patients
`with pituitary disease. J Clin Endocrinol Metab. 1999;84:838 – 843.
`20. Ambrosi B, Barbetta L, Re T, Passini E, Faglia G. The one micro-
`gram adrenocorticotropin test in the assessment of hypothalamic-
`pituitary-adrenal function. Eur J Endocrinol. 1998;139:575–579.
`21. AmmariF,IssaBG,MillwardE,ScanionMF.Acomparisonbetweenshort
`ACTHandinsulinstresstestsforassessinghypothalamo-pituitary-adrenal
`function. Clin Endocrinol (Oxf). 1996;44:473–476.
`22. Bangar V, Clayton RN. How reliable is the short Synacthen test for
`the investigation of the hypothalamic-pituitary-adrenal axis? Eur J
`Endocrinol. 1998;139:580 –583.
`23. Cho HY, Kim JH, Kim SW, et al. Different cut-off values of the
`insulin tolerance test, the high-dose short Synacthen test (250 mug)
`and the low-dose short Synacthen test (1 mug) in assessing central
`adrenal insufficiency. Clin Endocrinol (Oxf). 2014;81:77– 84.
`24. Choi CH, Tiu SC, Shek CC, Choi KL, Chan FK, Kong PS. Use of the
`low-dose corticotropin stimulation test for the diagnosis of secondary
`adrenocortical insufficiency. Hong Kong Med J. 2002;8:427– 434.
`25. Courtney CH, McAllister AS, Bell PM, et al. Low- and standard-
`dose corticotropin and insulin hypoglycemia testing in the assess-
`ment of hypothalamic-pituitary-adrenal function after pituitary
`surgery. J Clin Endocrinol Metab. 2004;89:1712–1717.
`26. Cunningham SK, Moore A, McKenna TJ. Normal cortisol response
`to corticotropin in patients with secondary adrenal failure. Arch
`Intern Med. 1983;143:2276 –2279.
`27. Deutschbein T, Unger N, Mann K, Petersenn S. Diagnosis of sec-
`ondary adrenal insufficiency: unstimulated early morning cortisol in
`saliva and serum in comparison with the insulin tolerance test. Horm
`Metab Res. 2009;41:834 – 839.
`28. Ferrante E, Morelli V, Giavoli C, et al. Is the 250 mug ACTH test a
`useful tool for the diagnosis of central hypoadrenalism in adult patients
`with pituitary disorders? Hormones (Athens). 2012;11:428 – 435.
`29. Giordano R, Picu A, Bonelli L, et al. Hypothalamus-pituitary-ad-
`renal axis evaluation in patients with hypothalamo-pituitary disor-
`ders: comparison of different provocative tests. Clin Endocrinol
`(Oxf). 2008;68:935–941.
`30. Gonc EN, Kandemir N, Kinik ST. Significance of low-dose and
`standard-dose ACTH tests compared to overnight metyrapone test
`in the diagnosis of adrenal insufficiency in childhood. Horm Res.
`2003;60:191–197.
`31. Jackson RS, Carter GD, Wise PH, Alaghband-Zadeh J. 1994 com-
`parison of paired short Synacthen and insulin tolerance tests soon
`after pituitary surgery. Ann Clin Biochem. 31( Pt 1):46 – 49.
`32. Kamrath C, Boehles H. The low-dose ACTH test does not identify
`mild insufficiency of the hypothalamnic-pituitary-adrenal axis in
`children with inadequate stress response. J Pediatr Endocrinol
`Metab. 2010;23:1097–1104.
`
`
`
`434
`
`Ospina et al
`
`Diagnosis of Adrenal Insufficiency
`
`J Clin Endocrinol Metab, February 2016, 101(2):427– 434
`
`33. Kehlet H, Blichert-Toft M, Lindholm J, Rasmussen P. Short ACTH
`test in assessing hypothalamic-pituitary-adrenocortical function. Br
`Med J. 1976;1