Primary aldosteronism (PA), the most common cause of

Primary Aldosteronism Androstenedione and 17-α-Hydroxyprogesterone Are Better Indicators of Adrenal Vein Sampling Selectivity Than Cortisol Giulio Ceolotto, Giorgia Antonelli, Giuseppe Maiolino, Maurizio Cesari, Giacomo Rossitto, Valeria Bisogni, Mario Plebani, Gian Paolo Rossi Abstract For identification of potentially surgically curable primary aldosteronism, guidelines recommend use of adrenal vein sampling (AVS) that requires selective catheterization of both adrenal veins as verified by using the cortisolderived selectivity index. Unfortunately, bilaterally selective studies are not obtained under unstimulated conditions in a proportion of the cases ranging between 15% and 50% depending on the cutoff used. We therefore investigated whether 17-α-hydroxyprogesterone and androstenedione, which showed a higher step-up between adrenal vein and inferior vena cava blood than cortisol, can ascertain selectivity when cortisol failed to do so. We prospectively recruited 32 hypertensive patients with confirmed primary aldosteronism, who underwent bilaterally simultaneous sampling without cosyntropin stimulation and with the same predefined AVS protocol. All were consecutively selected because of a cortisol-based selectivity index <2.00 in at least one of the paired adrenal vein blood samples collected as per protocol. Results showed that the values of the selectivity index based on 17-α-hydroxyprogesterone and androstenedione were higher (P<0.01) on average by 1.6- and 12-fold, respectively, than those based on cortisol. With use of these steroids, we rescued 43% and 73% of the AVS, respectively, from being judged nonselective. Thus, in challenging patients with primary aldosteronism submitted to AVS use of 17-α-hydroxyprogesterone, and even more so of androstenedione, for ascertaining selectivity allows demonstration of correct catheter placement in a proportion of AVS studies better than cortisol. Thus, replacing cortisol measurement with these steroids, and particularly androstenedione, can improve the diagnostic yield of AVS. (Hypertension. 2017;70:342-346. DOI: 10.1161/HYPERTENSIONAHA.117.09415.) Key Words: androstenedione catheterization cosyntropin diagnosis steroids Primary aldosteronism (PA), the most common cause of endocrine arterial hypertension, can be cured at long term if lateralized aldosterone excess is demonstrated. To establish lateralization, all current guidelines consider adrenal vein sampling (AVS) as the gold standard to be offered to all patients with PA seeking surgical cure. 1 3 The diagnostic use of AVS is, however, possible only when bilaterally selective catheterization is achieved. Because the adrenal cortex is the only appreciable site of cortisol production, in clinical practice, detection of a step-up of the level of this hormone between the adrenal vein and a peripheral vein blood or the inferior vena cava blood upstream the adrenal venous drainage is used to ascertain selectivity, which implies calculation of the selectivity index (SI). Bilateral selectivity is considered to be the case when the SI value exceeds a cutoff in both adrenal veins, which is set at 2.00 or 4.00, under unstimulated or cosyntropin-stimulated conditions, respectively. 4 Unfortunately, it is common experience that with these popular cutoffs to define selectivity, a relatively large proportion of AVS studies, ranging between 15% and 50% depending on the cutoff used, is judged to be nonselective under unstimulated conditions. 5 Luckily, recent studies with use of liquid chromatography tandem mass spectometry led to be discovery that the step-up cortisol is smaller than that of several other steroids, 6 including 17-α-hydroxyprogesterone (17α-hydroxypregn-4- ene-3,20-dione, 17α-OHP) and androstenedione (4-androstene-3,17-dione). Hence, it is conceivable that use of these steroids can rescue diagnosis in a proportion of the AVS studies held to be nonselective based on the cortisol-derived SI. This within-patient pairwise comparison study was therefore set up to test this hypothesis in a cohort of patients with PA, who posed special challenges at AVS in that they were judged nonselective on their cortisol-based SI value. Materials and Methods The study used a cohort of referred hypertensive patients with documented PA, who were prospectively recruited for the AVS study following the Endocrine Society guidelines. 1 All patients underwent imaging with computed tomography according to current guidelines. 1 For this protocol, we selected the Received March 21, 2017; first decision April 7, 2017; revision accepted May 5, 2017. From the Clinica dell Ipertensione Arteriosa (G.C., G.M., M.C., G.R., V.B., G.P.R.) and Laboratory Medicine (G.A., M.P.), Department of Medicine, DIMED-University of Padua, Italy. Correspondence to Gian Paolo Rossi, Clinica dell Ipertensione Arteriosa DIMED, University Hospital, via Giustiniani, 235126 Padova, Italy. E-mail gianpaolo.rossi@unipd.it 2017 American Heart Association, Inc. Hypertension is available at http://hyper.ahajournals.org DOI: 10.1161/HYPERTENSIONAHA.117.09415 342

Ceolotto et al Aldosterone Renin Ratio and APA Diagnosis 343 AVS studies that showed a cortisol-based SI <2.00 in at least 1 sample of the paired adrenal vein samples systematically collected at our center. This cutoff value is the most commonly used for unstimulated AVS studies 7 and is recommended by an Expert Consensus Statement. 4 All patients provided informed written consent and underwent AVS with the bilaterally simultaneous catheterization twice under unstimulated conditions at time 15 minutes and again at time 0 as described. 8 The Ethics Committee of the University of Padua approved the protocol; all procedures followed the Helsinki Declaration Principles. The assay used to measure cortisol was previously reported. 9 Plasma 17α-OHP and androstenedione levels were measured by commercial kits: an enzyme immunoassay (DRG Instruments GmbH, Germany) and a competitive chemi-luminescence enzyme immunoassay (Immulite 2000; Siemens), respectively. For plasma 17α-OHP, the imprecision monitored with 3 internal quality controls in our routine practice showed a coefficient of variation of 4.7% at 1.45 nmol/l, 6.8% at 7.08 nmol/l, and 5.5% at 14.06 nmol/l. For plasma androstenedione, a similar analysis demonstrated a coefficient of variation of 7.9%, 6.7%, and 4.8% at 2.59, 9.33, and 21.00 nmol/l, respectively. All these coefficient of variations are lower than those declared by the manufacturers. Definitions The SI was calculated as ratio between the value of each steroid (cortisol, 17α-OHP, or androstenedione) in each adrenal vein and in the infrarenal inferior vena cava. 10 A cutoff of 2.00 was used to establish selectivity, as recommended in an Expert Consensus Statement 4 and by the Endocrine Society guidelines. 1 Statistical Analysis Nonparametric Wilcoxon test was used for within-patient and between-group comparison of quantitative variables because no transformation of skewed variables was used. The distribution of categorical variables was compared by χ 2 analysis. Significance was set at P<0.05. For statistical analysis, we used the SPSS (version 24 for Mac, IBM, Italy) and the GraphPad Prism (version 6.00 for Mac, GraphPad Software, La Jolla, CA) softwares. Results Baseline Characteristics of the Patients The AVS samples that fulfilled these selection criteria pertained to 32 of 160 consecutive patients studied with the same AVS protocol, corresponding to 16% (77 of 480) blood samples. Table 1 shows the baseline characteristics of the patients recruited in this study. In brief, besides high blood pressure, the patients had all of the biochemical features expected for a cohort of patients with PA, including low serum K + and plasma renin activity and high plasma aldosterone concentration and aldosterone renin ratio. As expected by inclusion criterion, no bilateral selectivity was found in at least one of the paired cortisol-based SI values. Therefore, the patients who did not meet the criteria for clinical use of AVS were not submitted to surgery. Those with bilaterally selective results in at least 1 pair of the samples and lateralization of aldosterone excess underwent unilateral adrenalectomy. Accordingly, 2 groups of patients were generated: one comprising 14 patients, who underwent adrenalectomy and received a conclusive diagnosis of aldosterone-producing adenoma (APA), as determined by the demonstration of adenoma at pathology and by cure of PA at post-adrenalectomy follow-up, and one of 18 patients Table 1. Demographic and Clinical Features of the Patients Baseline Age, y 51.8±12.1 Gender, % (F) 50% BMI, kg/m 2 26.1±3.4 Systolic BP, mm Hg 171±32 Diastolic BP, mm Hg 98±15 Serum creatinine, mmol/l 73 (55 90) Serum K +, mmol/l 3.37±0.14 PRA, ng/ml per hour 0.80 (0.35 1.10) PAC, ng/dl 18.9±2.2 ARR, (ng/dl)/(ng/ml per hour) 41.6±6.5 For ARR calculation, PRA the minimum value was set at 0.20. Data presented as mean±sd or median (interquartile range), as appropriate. ARR indicates aldosterone renin ratio; BMI, body mass index; BP, blood pressure; F, female; PAC, plasma aldosterone concentration; and PRA, plasma renin activity. defined as non-apa, because a conclusive subtype diagnosis could not be determined as adrenalectomy was not performed. Comparison of the overall features of the APA and non-apa patients (Table 2) shows that there were no clearcut differences between these groups with the exception of lower serum K + in APA versus non-apa patients. Comparison of SI Based on Cortisol, 17α-OHP, and Androstenedione The pairwise within-patient comparison showed that the SI based on 17α-OHP was higher than that based on cortisol both on the right and the left sides, and that the SI based on Table 2. Demographic and Clinical Features at Baseline of the Patients With and Without a Conclusive Diagnosis of APA APA (n=14) Non-APA (n=18) P Value Age, y 48.4±12.2 54.5±11.8 NS Gender, % (F) 43% 55% NS BMI, kg/m 2 24.9±3.4 27.5±3.0 NS Systolic BP, mm Hg 163±18 181±43 NS Diastolic BP, mm Hg 94±15 101±16 NS Serum creatinine, mmol/l 78 (55 89) 71 (54 90) NS Serum K +, mmol/l 2.89±0.41 3.69±0.26 0.002 Urinary albumin excretion, mg/g creatinine 15.3 (8.9 21) 12.7 (6.7 28) NS PRA, ng/ml per hour 0.55 (0.32 1.09) 0.80 (036 1.11) NS PAC, ng/dl 25.0±4.2 15.9±1.6 NS ARR (ng/dl)/(ng/ml per hour) 53.0±10.9 32.3±7.1 NS Lateralization index 14.2±6.4 3.5±0.9 0.99 For ARR calculation, PRA the minimum value was set at 0.20. Data presented as mean±sd or median and interquartile range, as appropriate. APA indicates aldosterone producing adenoma; and NS, nonsignificant. Other abbreviations are as in Table 1.

344 Hypertension August 2017 Table 3. Comparison of Selectivity Index (SI) Values Based on Cortisol (F), 17-α-Hydroxyprogesterone (17α-OHP), and Androstenedione (A) in the All Cohort F P Value F vs 17α-OHP 17α-OHP P Value 17α- OHP vs A A P Value A vs F Right side 1.12 (1.05 1.42) 0.017 1.61 (1.29 2.82) 0.031 2.87 (2.28 9.78) 0.008 Left side 1.79 (1.57 1.84) 0.0009 2.61 (2.45 4.93) 0.0006 12.78 (9.19 21.01) 0.0005 Data presented as median and 95% confidence interval. androstenedione was even higher on both sides (Table 3). A bar graph of the SI values based on cortisol, 17α-OHP, and androstenedione on the right and the left sides, and bilaterally, is shown in Figure 1. On average, the SI values was higher on the right than on the left side, probably because of anatomic differences, as discussed below. Figure 2 shows a scatter plot of the within-patient relationship SI values based on the different steroids, with a direct linear relationship between cortisol based, 17α- OHP based, and androstenedione based. For both steroids, the slope was significantly different from zero (for 17α-OHP, F=4.83, P=0.03; for androstenedione, F=14.38, P=0.00004). The slope of the androstenedione-based versus the cortisol-based SI values was significantly higher (F=11.72, P=0.0009) than that of the 17α-OHP based SI values because the androstenedione-based SI values were systematically higher than the 17α-OHP based SI values. A Bland Altman plot of the difference between 17α-OHP and cortisol- (Figure 3A) or androstenedione- and cortisolbased SI (Figure 3B) shows a proportional bias for both 17α-OHP and androstenedione; moreover, it unambiguously demonstrated that on average the SI difference of each steroid versus cortisol was much greater for androstenedione than for 17α-OHP. Rate of Selective Studies By inclusion criteria, all AVS studies were nonbilaterally selective when selectivity was assessed by cortisol-based SI value of 2.00. In contrast, 43% and 73% such AVS studies were shown to be selective with 17α-OHP and androstenedione, respectively, with the same cutoff. Of further note, even with androstenedione, the steroid providing the highest SI values of the hormones tested, 27% of the samples were nonselective, suggesting that unfavorable anatomy, as discussed below, cannot be overcome even with use of a steroid released in greater amounts from the adrenal cortex. 11 Discussion This study selected the most challenging patients with PA consecutively submitted to AVS in that they failed to reach a SI cutoff >2.00 under unstimulated conditions. 5,7 It is possible that in this cohort, selective catheterization was not achieved because of unfavorable anatomy, for example, the common drainage of the right adrenal vein with an accessory hepatic vein, which, by diluting adrenal vein blood with that from the liver, decreases the SI value, because the liver clearance of androstenedione is reported to be between 82% and 92% in humans. 12,13 Therefore, from the practical standpoint, it was of interest to verify whether in this cohort of patients, use of adrenocortical steroids other than cortisol could provide a more sensitive assessment of selectivity than cortisol. In fact, the SI based on androstenedione and 17α-OHP was 12- and 1.6-fold higher than that based on cortisol, respectively (Figure 2), a finding that agrees with the larger gradient of these steroids between the inferior vena cava and the adrenal vein blood found in 2 studies that used liquid chromatography tandem mass spectrometry. 6,14 By measuring 17α-OHP and androstenedione in a large data set of PA cases judged nonselective based on cortisol, 43% of and 73%, respectively, of the studies were judged to be selective. Thus, the demonstration that both 17α-OHP and androstenedione allowed assessment of selectivity in such challenging cases, where cortisol failed, is an important novel finding with several clinical implications. The measurement of both 17α-OHP and androstenedione is widely available at most certified laboratories, in contrast with other biomarkers that have been proposed to assess selectivity (eg, plasma metanephrine). Needless to say that use of these steroids can also be exploited at centers where liquid chromatography-tandem mass spectrometry is available. 6 Figure 1. The bar graph shows the value of the selectivity index (SI) on the right and left sides and overall in the patients with SI <2.00 when assessed using cortisol (empty bars). The dashed horizontal line indicates the cutoff value of the SI most commonly used to define selectivity. By definition, the cortisol-based SI was <2.00 in this cohort of patients. Use of 17-α-hydroxyprogesterone (17α- OHP, black bars) and, even more so, androstenedione (A, dashed bars) resulted not only in significantly higher values but also in rescuing many adrenal vein sampling studies from being classified as nonselective.

Ceolotto et al Aldosterone Renin Ratio and APA Diagnosis 345 referral centers and, therefore, is not as widely available as the steroids herein investigated. Conclusions This study extends to clinical use in a challenging PA population the recent knowledge generated with mass spectrometry, 6 that the adrenal plasma levels of several steroids, particularly androstenedione, relative to peripheral vein blood concentrations are higher than those of cortisol. Accordingly, use of androstenedione as biomarker can be advantageous over that of cortisol in those AVS studies judged nonselective with only SI based on cortisol. Currently, these AVS studies are discarded for diagnostic purposes, thus limiting the clinical usefulness of AVS. Of the 2 steroids tested in this study, androstenedione is clearly preferable. Figure 2. The scatter plot of within-patient selectivity index (SI) values shows a direct linear relationship between cortisol and both 17-α-hydroxyprogesterone (17α-OHP) and androstenedione (A). For the purpose of visual clarity, the y axis is on a log 2 scale. The slope was significantly nonzero for both 17α-OHP (F=4.834, P=0.03) and A (F=14.38, P=0.00004). The slope of the A-based vs the cortisol-based SI values was significantly higher (F=11.72, P=0.0009) than that of 17α-OHP based vs the cortisol-based SI values because the A-based SI values were systematically higher than the 17α-OHP based SI values. Some limitations are to be acknowledged in this study. First, because it entailed a single referral center that evaluates mostly white patients, its results need to be replicated in other centers and ethnic groups, who might have a different rate of either 17α-OHP and androstenedione production or unfavorable anatomy, or both. Second, our study was not powered to investigate the question whether there are differences between genders in the adrenal step-up of these steroids. Third, it refers to unstimulated AVS data; therefore, its conclusions cannot be extended to cosyntropin-stimulated AVS studies. Finally, we did not compare the diagnostic performance of these steroids with that of plasma metanephrine, which has also been reported to have a higher SI than cortisol. 15 This was because routine assay of free plasma metanephrine is done only at few Perspectives There is an urgent need to improve the diagnostic accuracy of AVS to open the way for unilateral adrenalectomy in many patients with PA who still do not receive long-term cure because of inconclusive AVS results. The improvement of assessment of selectivity by use of biomarkers better than those currently used can be an important step in the right direction. The wider availability of liquid chromatography coupled to tandem mass spectrometry can further advance this practice by providing more accurate measurement of the steroids, such as androstenedione, that have high within-run coefficient of variation with currently available commercial kits. Sources of Funding This study was supported by the COST BM1301 Aldosterone and Mineralocorticoid Receptor (ADMIRE) EU program (to G.P. Rossi) and by FORICA (The Foundation for advanced Research In hypertension and Cardiovascular disease). None. Disclosures References 1. Funder JW, Carey RM, Mantero F, Murad MH, Reincke M, Shibata H, Stowasser M, Young WF Jr. The management of primary aldosteronism: case detection, diagnosis, and treatment: an Endocrine Society Clinical Practice Guideline. J Clin Endocrinol Metab. 2016;101:1889 1916. doi: 10.1210/jc.2015-4061. 2. Onozawa S, Murata S, Tajima H, Yamaguchi H, Mine T, Ishizaki A, Sugihara H, Oikawa S, Kumita S. Evaluation of right adrenal vein cannulation by computed tomography angiography in 140 consecutive patients Figure 3. The Bland Altman plot of the within-patient differences between the selectivity index (SI) based on 17-αhydroxyprogesterone (17α-OHP) and on cortisol (A) and between the SI based on androstenedione (A) and on cortisol (B). Both plots show a proportional bias in the measurement of the SI because 17α-OHP and androstenedione provided greater SI values than cortisol. Please note that for visualization purposes the x-axis scale was set to be 5-fold larger in the latter than in the former, indicating that the difference between SI values was much larger when determined with androstenedione than with 17α-OHP.

346 Hypertension August 2017 undergoing adrenal venous sampling. Eur J Endocrinol. 2014;170:601 608. doi: 10.1530/EJE-13-0741. 3. Maiolino G, Mareso S, Bisogni V, Rossitto G, Azzolini M, Cesari M, Seccia TM, Calò L, Rossi GP. Assessment of the quantitative value usefulness of the aldosterone-renin ratio (ARR) for primary aldosteronism (AQUARR) study. High Blood Press Cardiovasc Prev. 2016;23:19 23. doi: 10.1007/s40292-015-0125-0. 4. Rossi GP, Auchus RJ, Brown M, Lenders JW, Naruse M, Plouin PF, Satoh F, Young WF Jr. An expert consensus statement on use of adrenal vein sampling for the subtyping of primary aldosteronism. Hypertension. 2014;63:151 160. doi: 10.1161/HYPERTENSIONAHA.113.02097. 5. Rossi GP, Ganzaroli C, Miotto D, De Toni R, Palumbo G, Feltrin GP, Mantero F, Pessina AC. Dynamic testing with high-dose adrenocorticotrophic hormone does not improve lateralization of aldosterone oversecretion in primary aldosteronism patients. J Hypertens. 2006;24:371 379. doi: 10.1097/01.hjh.0000202818.10459.96. 6. Eisenhofer G, Dekkers T, Peitzsch M, Dietz AS, Bidlingmaier M, Treitl M, Williams TA, Bornstein SR, Haase M, Rump LC, Willenberg HS, Beuschlein F, Deinum J, Lenders JW, Reincke M. Mass spectrometrybased adrenal and peripheral venous steroid profiling for subtyping primary aldosteronism. Clin Chem. 2016;62:514 524. doi: 10.1373/ clinchem.2015.251199. 7. Rossi GP, Barisa M, Allolio B, et al. The Adrenal Vein Sampling International Study (AVIS) for identifying the major subtypes of primary aldosteronism. J Clin Endocrinol Metab. 2012;97:1606 1614. doi: 10.1210/jc.2011-2830. 8. Seccia TM, Miotto D, Battistel M, Motta R, Barisa M, Maniero C, Pessina AC, Rossi GP. A stress reaction affects assessment of selectivity of adrenal venous sampling and of lateralization of aldosterone excess in primary aldosteronism. Eur J Endocrinol. 2012;166:869 875. doi: 10.1530/EJE-11-0972. 9. Seccia TM, Miotto D, De Toni R, Pitter G, Mantero F, Pessina AC, Rossi GP. Adrenocorticotropic hormone stimulation during adrenal vein sampling for identifying surgically curable subtypes of primary aldosteronism: comparison of 3 different protocols. Hypertension. 2009;53:761 766. doi: 10.1161/HYPERTENSIONAHA.108.128553. 10. Rossitto G, Miotto D, Battistel M, Barbiero G, Maiolino G, Bisogni V, Sanga V, Rossi GP. Metoclopramide unmasks potentially misleading contralateral suppression in patients undergoing adrenal vein sampling for primary aldosteronism. J Hypertens. 2016;34:2258 2265. doi: 10.1097/ HJH.0000000000001082. 11. Miotto D, De Toni R, Pitter G, Seccia TM, Motta R, Vincenzi M, Feltrin G, Rossi GP. Impact of accessory hepatic veins on adrenal vein sampling for identification of surgically curable primary aldosteronism. Hypertension. 2009;54:885 889. doi: 10.1161/HYPERTENSIONAHA.109.134759. 12. Rivarola MA, Singleton RT, Migeon CJ. Splanchnic extraction and interconversion of testosterone and androstenedione in man. J Clin Invest. 1967;46:2095 2100. doi: 10.1172/JCI105696. 13. Horton R, Tait JF. Androstenedione production and interconversion rates measured in peripheral blood and studies on the possible site of its conversion to testosterone. J Clin Invest. 1966;45:301 313. doi: 10.1172/ JCI105344. 14. Nilubol N, Soldin SJ, Patel D, Rwenji M, Gu J, Masika LS, Chang R, Stratakis CA, Kebebew E. 11-Deoxycortisol may be superior to cortisol in confirming a successful adrenal vein catheterization without cosyntropin: a pilot study. Int J Endo Oncol. 2017;4:75 83. 15. Dekkers T, Deinum J, Schultzekool LJ, Blondin D, Vonend O, Hermus AR, Peitzsch M, Rump LC, Antoch G, Sweep FC, Bornstein SR, Lenders JW, Willenberg HS, Eisenhofer G. Plasma metanephrine for assessing the selectivity of adrenal venous sampling. Hypertension. 2013;62:1152 1157. doi: 10.1161/HYPERTENSIONAHA.113.01601. Novelty and Significance What Is New? The achievement of surgical cure of primary aldosteronism depends on identification of unilateral causes by adrenal vein sampling (AVS). To this aim, guidelines recommend use of AVS, which is based on the achievement of selective catheterization of both adrenal veins, as commonly verified by using the cortisol-derived selectivity index. With a within-patient study design of challenging AVS studies judged to be nonselective using cortisol as a biomarker of selectivity, we discovered that androstenedione and 17-α-hydroxyprogesterone can allow to ascertain selectivity when cortisol failed to do so. What Is Relevant? Use of androstenedione as a biomarker of selectivity can allow rescue for diagnostic purposes up to 73% of AVS studies that would be discarded with use of cortisol. Summary This study demonstrates that the rate of bilaterally selective AVS studies can be markedly increased by using androstenedione as a biomarker of selectivity, over the most widely used cortisol in the most challenging cases. Exploiting this strategy in the subtyping of primary aldosteronism can allow reach a diagnosis in far more cases than by using cortisol as a biomarker for selectivity.