| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Departments of Medicine (S.-C.T., C.-H.C., Y.-W.N., F.W.K.C., C.-M.N., A.P.S.K.) and Pathology (C.-C.S.), Queen Elizabeth Hospital, Hong Kong Special Administrative Region, China
Address all correspondence and requests for reprints to: Sau-Cheung Tiu, Department of Medicine, Queen Elizabeth Hospital, 30 Gascoigne Road, Hong Kong Special Administrative Region, China. E-mail: tscz01{at}ha.org.hk.
 |
Abstract |
|---|
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
ARR yielded larger areas under the curve in the receiver-operating-characteristics curve than PRA or aldosterone under all conditions of testing. Our results confirmed the superiority of ARR to either aldosterone or PRA alone as a diagnostic test for primary hyperaldosteronism.
ARR cutoff levels were significantly affected by the condition of testing. Depending on posture and time of day, it varied from 13.1–35.0 ng/dl per ng/ml·h in our study population. When using ARR for screening primary hyperaldosteronism, posture and time of sampling should be standardized both within and between centers to minimize variability in cutoff levels.
|
Introduction |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
In our endocrine unit, patients referred for suspicion of primary hyperaldosteronism were investigated according to a defined protocol that included a combination of saline suppression test, postural test, adrenal computed tomography (CT) and adrenal venous sampling (AVS). This study is a review of our data in an attempt to 1) assess the usefulness of ARR and identify an optimal ARR cutoff value in the Chinese population and 2) study the effect of conditions of blood sampling on the performance characteristics of ARR as a screening test in primary hyperaldosteronism.
|
Subjects and Methods |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Blood samples were obtained under standardized conditions by metabolic nurses. For postural study, patients were advised to have liberalized salt intake (>100 mmol of sodium daily) for at least 3 d, to discontinue diuretics and spironolactone for 6 wk, and to discontinue ß-blockers, angiotensin-converting enzyme inhibitors, and angiotensin receptor blockers for 2 wk. If antihypertensives could not be stopped completely, patients were stabilized with calcium channel blockers, 
-blockers, central-acting antihypertensives, or vasodilator drugs (the choice of drugs being dependent on blood pressure control and drug tolerability in individual patients) for at least 2 months before testing. Medication that might interfere with the renin-aldosterone axis, such as steroids, sex hormones, licorice, or nonsteroidal antiinflammatory drugs were also withheld for at least 2 wk. Serum potassium levels were brought into normal range with oral potassium supplementation before the test in patients with hypokalemia. Patients were admitted 1 d before the test. Blood was sampled at 0900 h after overnight recumbency and again at 1300 h after 4 h of ambulation and assayed for cortisol, aldosterone, and PRA. For the saline suppression test, patients were prepared in the same way before the test. On the day of testing, patients attended the metabolic day ward at approximately 1000 h. Blood was sampled after 30 min in a seated position and again after infusion of 2 liters of normal saline over 4 h in the seated position and assayed for aldosterone and PRA. Suppression was defined as a decrease in postsaline aldosterone level to less than 5.0 ng/dl. In most patients, the saline suppression test and the postural study were performed within 1 wk of each other, before the result of the other test was known to the investigators. Adrenal CT and AVS were performed according to clinical indications.
Plasma samples were kept at –70 C until assay. Serum aldosterone was measured in duplicate by RIA using a commercial kit from Diagnostic Products (Los Angeles, CA). The interassay coefficient of variation was 5.5% at a concentration of 26.4 ng/dl, and the reference range for ambulatory and recumbent subjects were 4.0–31.0 and less than 16.0 ng/dl, respectively. The PRA was measured by RIA of angiotensin I generated after incubation of the plasma sample in standardized conditions (Renin Maia, Adaltis, Italy). The interassay coefficient of variation for PRA was 12.7% at an activity of 1.81 ng/ml·h, and the reference range was 0.97–4.18 and 0.51–2.64 ng/ml·h for ambulatory and recumbent subjects, respectively. The lower limit of PRA determination was 0.1 ng/ml·h.
The area under the curve (AUC) for the receiver-operating-characteristics (ROC) curves, the cutoff values, and the test performance characteristics were obtained from MedCalc. Other statistical analyses were performed with SPSS 11.0. ARR was calculated with serum aldosterone level as the numerator and PRA as the denominator and expressed in ng/dl per ng/ml·h (pmol/liter per µg/liter·h). In samples with PRA less than 0.1 ng/ml·h (most of which belonged to subjects with a final diagnosis of primary hyperaldosteronism), a PRA value of 0.1 ng/ml·h was used for the calculation of ARR. Data were expressed as mean ± SD (range), with differences between groups assessed using the 
2, Fisher’s exact, Mann-Whitney U, or t test, as appropriate. P < 0.05 was considered statistically significant. ROC curves were drawn and the test performance characteristics calculated to assess the utility of the tests and to identify the best cutoff values for ARR.
|
Results |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Eleven patients had two or more sets of postural studies performed. Eight patients in the group with primary hyperaldosteronism had two sets of data each, and two had three sets of data each. One patient in the group with essential hypertension had two sets of data, yielding a total of 75 sets of paired recumbent PRA and aldosterone values and 75 sets of paired ambulatory PRA and aldosterone values for analysis. A repeat analysis of the data using only one arbitrarily chosen set of data for each subject did not significantly alter the findings of the study.
The saline suppression test was performed in 48 patients, yielding a total of 48 sets of pre- and postsaline infusion PRA and aldosterone values for analysis. In the 14 patients in whom the saline suppression test was not performed, the diagnosis of primary hyperaldosteronism was validated by the surgical findings.
Forty-nine patients (79%) were on dihydropyridine calcium channel blockers (nifedipine or amlodipine) at the time of testing. Of these, 12 were also on either prazosin or methyldopa for blood pressure control. Two other patients were taking prazosin or minoxidil alone. In only 11 patients (18%) did the blood pressure allow discontinuation of all antihypertensive medication. Prescription patterns were similar between subjects with essential hypertension and subjects with primary hyperaldosteronism. Adrenal CT was performed in 54 subjects; AVS was also performed in six of them.
Suppression of postsaline aldosterone level to less than 5 ng/dl (range, 1.6–4.9 ng/dl) was observed in 12 subjects. Postsaline aldosterone level ranged from 5.1–48.3 ng/dl in the 36 subjects in whom suppression was considered to be inadequate for excluding primary hyperaldosteronism. Borderline postsaline aldosterone levels between 5.0 and 10.0 ng/dl were seen in four subjects with a final diagnosis of primary hyperaldosteronism (postsaline aldosterone levels of 6.2, 7.0, 8.1, and 9.1 ng/dl) and in three subjects with a final diagnosis of essential hypertension (postsaline aldosterone levels of 5.1, 7.8, and 10.0 ng/dl).
The diagnosis of primary hyperaldosteronism was established in 45 subjects, based on either 1) histology and subsequent clinical improvement (n = 27), 2) classical renin-aldosterone profile as defined by elevated aldosterone and suppressed renin in at least one simultaneous blood sample, plus a nonsuppressible saline suppression test (n = 16), or 3) probable renin-aldosterone profile as defined by either elevated aldosterone and normal renin or normal aldosterone and suppressed renin in a simultaneously taken sample, plus a nonsuppressible saline suppression test and finding of a single nodule on adrenal CT (n = 2). All 16 subjects under criteria 2 also had adrenal CT performed, showing a single adrenal nodule (n = 7), bilateral adrenal nodules (n = 1), hyperplasia of both adrenals (n = 3), or normal adrenals (n = 5). AVS was performed in four of these subjects, confirming the diagnosis of a unilateral aldosterone-producing adenoma (APA) in two patients whose adrenal CT showed either bilateral adrenal masses (n = 1) or an equivocal adenoma (n = 1) and bilateral hyperplasia in two patients whose adrenal CT showed an adenoma. Of these 45 patients, 27 underwent unilateral adrenalectomy. Histological examination of the 27 surgical specimens showed an adrenal adenoma in 26 of these and a hyperplastic adrenal in the remaining specimen. All 27 subjects had resolution of hypokalemia and improvement or cure of hypertension after operation. Four other subjects whose postural and CT adrenal/AVS studies suggested an APA preferred medical therapy. Nine patients had postural and adrenal CT results that were compatible with idiopathic hyperaldosteronism (IHA) due to bilateral zona glomerulosa hyperplasia. In two of them, the diagnosis was confirmed by AVS. In the remaining five patients, including the two patients under criteria 3, differentiation between APA and IHA could not be made because postural and adrenal CT studies yielded discrepant results with respect to the differentiation between these two types of primary hyperaldosteronism, and they refused to undergo AVS.
In 17 subjects, the diagnosis of primary hyperaldosteronism could not be established. These subjects were considered to have essential hypertension. All 17 subjects had either 1) normal renin and aldosterone profiles on repeated testing and suppression of aldosterone level after saline infusion to less than 5.0 ng/dl (n = 6), 2) normal renin and aldosterone profiles on repeated testing and normal CT findings despite nonsuppression in the saline suppression test (n = 5), or 3) probable renin-aldosterone profile as defined in the preceding paragraph and suppression of aldosterone level by saline infusion (n = 6). Of the five patients under criteria 2, three might have secondary hyperaldosteronism. Two of them (postsaline aldosterone levels, 10.0 and 18.9 ng/dl) had both elevated PRA and aldosterone levels, and the remaining one (postsaline aldosterone level, 15.4 ng/dl) had elevated PRA and normal aldosterone levels. All of them were taking calcium-channel blockers. Investigations for renal artery stenosis were negative. The remaining two subjects had postsaline aldosterone levels of 5.0 and 7.8 ng/dl, but repeated testing showed PRA and aldosterone levels within reference ranges. A repeat analysis of the data after exclusion of these five subjects did not significantly alter the findings of the study.
Data on PRA, aldosterone, and ARR values during postural study and saline suppression test are tabulated in Table 1
. Statistically significant differences were seen between presaline (morning seated), postsaline, and 0900 h recumbent and 1300 h ambulatory PRA, aldosterone, and ARR levels between patients with primary hyperaldosteronism and patients with essential hypertension.
|
, and the respective AUC are tabulated in Table 2. ARR determinations yielded higher AUC than either PRA or aldosterone under all conditions of assessment. Cutoff levels for ARR, aldosterone, and PRA that corresponded to the highest accuracy (minimal false positive and false negative results or the point nearest to the left upper corner on the ROC curve) are shown in Table 3. These levels were different under different conditions of testing, ranging from 13.1 ng/dl per ng/ml·h (364 pmol/liter per µg/liter·h) after 4 h of ambulation to 35.0 ng/dl per ng/ml·h (970 pmol/liter per µg/liter·h) after overnight recumbency for ARR. The sensitivity, specificity, and positive and negative likelihood ratios at these cutoff values are also tabulated in Table 3. For sampling performed at 1300 h after 4 h of ambulation, raising the ARR cutoff value to 35.0 ng/dl per ng/ml·h improved the specificity to 94.4% (72.6–99.1%) at the cost of a loss of sensitivity to 80.7% (68.1–89.9%). A further increase in the cutoff value to 37.1 ng/dl per ng/ml·h further improved the specificity to 100.0% (81.3–100.0%) at the cost of a further loss of sensitivity to 78.9% (66.1–88.6%). For sampling performed in the morning after the subject has been seated for 30 min (before the infusion of normal saline in the saline suppression test), raising the cutoff value of ARR to 66.9 ng/dl per ng/ml·h improved the specificity to 100.0% (80.3–100.0%) at the cost of a loss of sensitivity to 64.5% (45.4–80.8%).
|
|
|
|
Discussion |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
). However, ARR may not be so robust as to justify the use of the same cutoff values under all conditions of testing. There is as yet no consensus on the ideal cutoff value for ARR and the conditions under which the test should be performed. Some investigators performed the test with the patient in a supine position for 30 min (7), some after the patient had been seated for 15 min (5, 6), and some after 2 h of ambulation (8, 11). Some recommended stopping medication before the test (12, 13), whereas others believed that it was not necessary (5, 6, 7, 11, 14). Widely different cutoff levels for the ratio have been proposed, ranging from 7.2–100.1 ng/dl per ng/ml·h (4). The results of our study showed that, despite its being more robust than aldosterone and PRA, the cutoff values for ARR were affected by the conditions of testing, such as posture, time of day, and acute salt loading. The variability of cutoff levels reported by other investigators may therefore be partly accounted for by the different conditions of testing.
A major difficulty in studies on the diagnostic utility of ARR is the absence of a gold standard for differentiating primary hyperaldosteronism from essential hypertension. There are no pathognomonic clinical features in primary hyperaldosteronism. Hypokalemia, once believed to be a common and important feature of the condition, is now found to be absent in most patients (15). The diagnosis of primary hyperaldosteronism can be established beyond doubt only in the classical patient who is cured of the hypokalemia and/or hypertension after removal of a histologically proven adrenal adenoma, as in Conn’s original description (16). In patients who are unwilling to undergo surgery or for whom surgery is not the recommended treatment, as in cases of IHA, a definitive diagnosis will often have to rely on a combination of a consistent laboratory profile of high aldosterone and low renin levels, nonsuppressibility of aldosterone with saline (7) or fludrocortisone (17), unresponsiveness of PRA to captopril (18) or diuretics (19), and results of imaging studies. All these tests, however, have their own limitations in sensitivity and specificity (18, 19, 20) and suffer from variations in protocols and cutoff values. The establishment of a diagnosis of essential hypertension is even more difficult, because it is basically a diagnosis by exclusion, and there is as yet no consensus on the number and types of tests that one should perform before secondary causes of hypertension can be confidently excluded. This point is illustrated in a recent paper by Tanabe et al. (21), in which they recommended repeating renin/aldosterone assays at least three times before dismissing the diagnosis of primary hyperaldosteronism because two thirds of their series of 71 patients with histologically proven primary hyperaldosteronism did not have an abnormal renin/aldosterone profile on every occasion. This problem of verification bias undermines the reliability of many prospective and retrospective studies on cutoff values of ARR in primary hyperaldosteronism (4). In this study, we tried to overcome this difficulty by including for analysis only those patients whose diagnosis of primary hyperaldosteronism was established by surgery, or whose combination of investigation results allowed a reasonably confident diagnosis of primary hyperaldosteronism or essential hypertension, relying on consistencies among aldosterone/renin profiling, saline suppression test and imaging studies or, in the few cases with minor inconsistencies among various tests, a major direction indicated by the tests. In the literature, cutoff values for saline suppression test varied from 5–10 ng/dl (3, 6, 22, 23, 24). To minimize the possibility of misdiagnosing primary hyperaldosteronism as essential hypertension, we employed the more stringent cutoff of postsaline aldosterone level of less than 5 ng/dl to indicate suppressibility with saline loading. The inclusion of aldosterone/renin profiling in the gold standard for establishing the diagnoses can theoretically artificially enhance the diagnostic ROCs for these parameters, but we expect its impact to be similar for aldosterone, PRA, and ARR. A neater approach would be to rely on a combination of gold standard tests, such as fludrocortisone suppression test or measurement of 24-h urine aldosterone after 3 d of oral sodium loading, that are independent of the screening parameters themselves.
Cutoff values that give the highest accuracy (minimal false positive and negative results) can be derived from a ROC curve. However, to determine the optimal threshold value or the cutoff point of a test variable, it is necessary to take into account the test characteristics associated with various operating points on a ROC curve, the pretest probability of the disease, the benefits of true positive and true negative results, as well as the harms of false positive and false negative results. The pretest probability or disease prevalence of primary hyperaldosteronism has been reported to be around 10% in a study on a predominantly Chinese hypertensive population seen in a primary care setting (6). Assuming this level of disease prevalence, a test with sensitivity and specificity levels comparable to those of ARR under the various conditions of sampling in our study would yield positive and negative predictive values that ranged from 49.1–100% and from 99.6–100%, respectively, with positive likelihood ratios of 8.68 to infinity and negative likelihood ratios ranging from 0.00–0.04 (Table 3). Although negative likelihood ratios and negative predictive values in these ranges clearly justify the use of ARR as a screening test, a positive predictive value of 49.1% is relatively low. Cases screened positive for primary hyperaldosteronism with ARR will have to undergo more sophisticated and costly tests. Given the high prevalence of hypertension, a lot of patients will be unnecessarily subjected to further investigations if the positive predictive value is only around 50%. Raising the cutoff value improves the specificity and the positive predictive value of the test at the cost of a loss in sensitivity, compromising its utility for screening purposes. Levels of sensitivity and specificity that are considered to be acceptable may differ from center to center, depending on health care resources.
The 0900 h overnight recumbency and the postsaline loading ARR produce the best test characteristics (Tables 2 and 3). However, they are relatively inconvenient to perform, requiring overnight admission for the former test and 4 h of hospital stay for the latter. Acute saline loading may also be relatively contraindicated in patients with other medical conditions that predispose them to fluid overload.
One limitation of our study is the nature of the study population. The prevalence of APA among our cases of primary hyperaldosteronism was at least 57.8% (26 of the 45 cases of primary hyperaldosteronism having histologically proven adenoma). This high prevalence is partly because we included only relatively definitive cases of primary hyperaldosteronism and partly because the tests were performed in patients with high clinical suspicion of the condition, mostly because of the presence of hypokalemia. Whether the ARR cutoffs derived from this population is applicable to a patient population with predominantly normokalemic hypertension, who presumably have less florid forms of the disease and a higher percentage of the IHA subtype (25), would require additional studies. The relationship between PRA and aldosterone is a negative one in Conn’s syndrome, with aldosterone secretion being autonomous and independent of renin, whereas in IHA, the relationship between renin and aldosterone is still partly retained (26). The ARR is therefore expected to exaggerate the difference in aldosterone and PRA levels between patients with APA and patients with essential hypertension, but this may not hold true for differentiating IHA from essential hypertension. However, it is the identification of patients with APA that is more important, because of the possibility of a definitive cure with surgery. Although identification of IHA can also lead to more targeted pharmacotherapy, the therapeutic value of identifying this group of patients is still not without contention (26).
We also did not assess the impact of disparate salt intake and the effects of medication on the ARR ratio. Our population had a high mean salt intake, and the 24-h urinary sodium excretion had been reported to be between 150 and 175 ± 50–60 mmol/d among hypertensive subjects (27, 28). The stimulatory effect of a habitual low dietary salt intake on renin production and secretion may lead to a lower ARR, and sensitivity of ARR is expected to improve if patients are maintained on a liberal dietary salt intake before testing (25). Most of our patients were on dihydropyridine calcium channel blockers, some were also taking -blockers, and a few were taking methyldopa or minoxidil. Calcium channel blockers may lead to false negative results by suppressing aldosterone and stimulating renin secretion. Minoxidil can also stimulate PRA and lower ARR, whereas methyldopa can suppress PRA and raise ARR (29, 30). As a general principle, medication should be stopped before the test, but when this is not practically feasible, other investigators have recommended the use of drugs that have a lesser effect on the ratio, such as verapamil slow-release, hydralazine, and prazosin (25).
In conclusion, this study demonstrates that ARR is better than either aldosterone or PRA alone in screening for primary hyperaldosteronism. The optimal ARR cutoff level is dependent on the conditions of testing as well as the a priori determination of acceptable levels of specificity and sensitivity. For screening a population of hypertensive outpatients, it would probably be most convenient to sample blood in the morning after 30 min in the seated position. Under this condition of testing, using a cutoff value of 23.6 ng/dl per ng/ml·h will have a sensitivity and specificity of 96.8 and 94.1%, respectively, whereas a cutoff value of 66.9 ng/dl per ng/ml·h yields sensitivity and specificity levels of 64.5 and 100.0%, respectively. If the ARR value is less than 23.6 ng/dl per ng/ml·h, clinicians can be assured that the likelihood of primary hyperaldosteronism is extremely low. On the other hand, if it is above 66.9 ng/dl per ng/ml·h, the diagnosis of primary hyperaldosteronism can practically be considered established. For intermediate levels, the decision to perform additional investigations will have to depend on the clinical algorithm adopted by individual centers, based on knowledge of test characteristics as described in this study and availability of resources.
|
Footnotes |
|---|
Abbreviations: APA, Aldosterone-producing adenoma; ARR, aldosterone/renin ratio; AVS, adrenal venous sampling; CT, computed tomography; IHA, idiopathic hyperaldosteronism; PRA, plasma renin activity; ROC, receiver operating characteristics.
Received June 28, 2004.
Accepted October 1, 2004.
|
References |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  B Rayner Primary aldosteronism and aldosterone-associated hypertension J. Clin. Pathol., July 1, 2008; 61(7): 825 - 831. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 N. M. Kaplan and D. A. Calhoun Is There an Unrecognized Epidemic of Primary Aldosteronism? (Con) Hypertension, September 1, 2007; 50(3): 454 - 458. [Full Text] [PDF]
|
|
|
|
 |
|
 S. J. Hood, K. P. Taylor, M. J. Ashby, and M. J. Brown The Spironolactone, Amiloride, Losartan, and Thiazide (SALT) Double-Blind Crossover Trial in Patients With Low-Renin Hypertension and Elevated Aldosterone-Renin Ratio Circulation, July 17, 2007; 116(3): 268 - 275. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. Newton-Cheh, C.-Y. Guo, P. Gona, M. G. Larson, E. J. Benjamin, T. J. Wang, S. Kathiresan, C. J. O'Donnell, S. L. Musone, A. L. Camargo, et al. Clinical and Genetic Correlates of Aldosterone-to-Renin Ratio and Relations to Blood Pressure in a Community Sample Hypertension, April 1, 2007; 49(4): 846 - 856. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. Kazory and I. D. Weiner Primary hyperaldosteronism in a patient with end-stage renal disease Nephrol. Dial. Transplant., March 1, 2007; 22(3): 917 - 919. [Full Text] [PDF]
|
|
|
|
 |
|
 G. P. Rossi, G. Bernini, C. Caliumi, G. Desideri, B. Fabris, C. Ferri, C. Ganzaroli, G. Giacchetti, C. Letizia, M. Maccario, et al. A Prospective Study of the Prevalence of Primary Aldosteronism in 1,125 Hypertensive Patients J. Am. Coll. Cardiol., December 5, 2006; 48(11): 2293 - 2300. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 F. BEUSCHLEIN and M. REINCKE Adrenocortical Tumorigenesis Ann. N.Y. Acad. Sci., November 1, 2006; 1088(1): 319 - 334. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 P. Mulatero, A. Milan, F. Fallo, G. Regolisti, F. Pizzolo, C. Fardella, L. Mosso, L. Marafetti, F. Veglio, and M. Maccario Comparison of Confirmatory Tests for the Diagnosis of Primary Aldosteronism J. Clin. Endocrinol. Metab., July 1, 2006; 91(7): 2618 - 2623. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. Schirpenbach, L. Seiler, C. Maser-Gluth, F. Rudiger, C. Nickel, F. Beuschlein, and M. Reincke Confirmatory testing in normokalaemic primary aldosteronism: the value of the saline infusion test and urinary aldosterone metabolites. Eur. J. Endocrinol., June 1, 2006; 154(6): 865 - 873. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 Aldosterone-Renin Ratio in Primary Hyperaldosteronism Journal Watch (General), February 11, 2005; 2005(211): 1 - 1. [Full Text]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
