Automated Chemiluminescence-Immunoassay for Aldosterone during Dynamic Testing: Comparison to Radioimmunoassays with and without Extraction Steps

Transcription

1 Clinical Chemistry 52: (2006) Automation and Analytical Techniques Automated Chemiluminescence-Immunoassay for Aldosterone during Dynamic Testing: Comparison to Radioimmunoassays with and without Extraction Steps Caroline Schirpenbach, 1 Lysann Seiler, 2,3 Christiane Maser-Gluth, 4 Felix Beuschlein, 3 Martin Reincke, 1* and Martin Bidlingmaier 1 Background: Measurements of aldosterone have become more common since the recognition that primary aldosteronism is a more frequent cause of hypertension than previously believed. Our aim was to compare concentrations reported by 4 assays for samples obtained after saline infusion during dynamic testing. Methods: We tested 104 participants (27 with primary aldosteronism, 30 with essential hypertension, and 47 healthy controls) with the intravenous saline infusion test (2.0 L isotonic saline over 4 h), with repetitive sampling. In all blood samples, aldosterone concentration was measured by an in-house RIA after extraction and chromatography, by 2 commercially available RIAs without extraction (Aldosterone Maia, Adaltis; Active Aldosterone, Diagnostics Systems Laboratories) and by an automated CLIA (Advantage, Nichols Institute Diagnostics). Results: Correlation coefficients for results of pairs of assays ranged from 0.74 to Agreement between commercial assays and in-house RIA was best at the low to intermediate concentrations after saline infusion. Mean (SD) Adaltis and DSL RIA results were 2- to 3-times higher [healthy participants: 78 (25) ng/l and 56 (18) ng/l, respectively] than those obtained by Nichols 1 Ludwig-Maximilians-University, Medizinische Klinik Innenstadt, Munich, Germany. 2 Centre Hospitalier Universitaire Vaudois, Département de médecine interne Lausanne, Lausanne, Switzerland. 3 Albert-Ludwigs-University, Medical Department 2, Freiburg, Germany. 4 Ruprecht-Karls-University, Institute of Pharmacology, Heidelberg, Germany. * Address correspondence to this author at: Medizinische Klinik Innenstadt, Klinikum der LMU München, Ziemssenstr. 1, D München, Germany. Fax ; martin.reincke@med.uni-muenchen.de. Received February 8, 2006; accepted June 5, Previously published online at DOI: /clinchem CLIA [17 (8) ng/l] and in-house RIA [23 (18) ng/l]. Aldosterone concentrations measured by the Nichols CLIA were below the limit of detection (limit of the blank) in 27 of 47 healthy participants. Conclusions: Aldosterone concentrations reported by the Adaltis and DSL nonextraction RIAs were consistently higher than those produced by the Nichols CLIA and the in-house RIA. The convenient Nichols CLIA showed better agreement with the in-house RIA, but the concentrations in healthy participants were frequently undetectable by this method. Uncritical application of cutoff values from the literature must be avoided American Association for Clinical Chemistry Accurate measurements of renin and aldosterone concentrations are essential for a correct diagnosis of conditions affecting the renin-angiotensin-aldosterone axis such as primary aldosteronism (PA), 5 renal artery stenosis, Bartter syndrome, Gitelman syndrome, congenital adrenal hyperplasia, renin-secreting tumors, Liddle syndrome, and Gordon syndrome (1 3). Bioassays and double isotope derivative assays are ponderous and complex methods, whereas RIAs for aldosterone (4) allow laboratorians to assay many samples and are frequently used. Only recently, an automated CLIA for measurement of aldosterone has been reported (5, 6). Currently, most immunoassays use rabbit polyclonal antisera with various affinities and specificities (5). Furthermore, most of the assays in use do not involve chromatography or extraction steps (7, 8). Inadequate standardization, poor interlaboratory reproducibility, and the limited comparability of different immunoassays re- 5 Nonstandard abbreviation: PA, primary aldosteronism. 1749

2 1750 Schirpenbach et al.: Comparison of Commercial Aldosterone Assays main problematic and produce difficulties in defining cutoff values for PA (9, 10), essentially requiring each laboratory to establish its own reference intervals. In this study, we compared a newly introduced, commercial, automated CLIA and 2 established commercial aldosterone RIAs with an in-house RIA after extraction and chromatography as comparison method to measure aldosterone concentrations during dynamic testing for PA. Materials and Methods study participants From a total of 104 individuals, 27 patients with PA [mean age (SD) 55.0 (10.0) years; 10 women], 30 patients with essential hypertension [50.2 (13.0) years; 16 women], and 47 normotensive individuals [24.1 (2.6) years; 19 women] underwent infusions of 2 L of 9 g NaCl/L intravenously over 4 h, beginning between 0800 and 0930 in the morning. The participants remained recumbent during the whole testing procedure. Before infusion and after 2, 3, and 4 h, blood specimens were drawn from a forearm vein (serum and EDTA plasma; Monovettes, Sarstedt). After centrifugation, all samples were stored at 20 C until measured. We based the diagnosis of PA on the following biochemical criteria (11): repeatedly increased aldosterone/ renin ratios (according to cutoff values of the local laboratory using the Adaltis RIA for measurement of aldosterone), increased urinary aldosterone excretion [ 15 g/day, or a previous pathological saline infusion test (serum aldosterone at 240 min 80 ng/l or 223 nmol/l)]. Differential diagnosis of PA was based on an algorithm (combining the results of computed tomography/magnetic resonance imaging and a posture test) (12) and/or selective vein catheterization. The latter was performed in 15 of 25 patients. We confirmed the diagnosis of PA in 25 patients. In 2 patients with suspected PA, diagnostic procedures had not been finished by the end of this study. Twelve of the 25 patients had bilateral adrenal hyperplasia, 4 had an aldosterone-producing adenoma and underwent surgery. For the remainder, subtype evaluation had not been completed at the end of this study. With the exception of spironolactone, which was withdrawn 6 weeks before testing, patients took their regular antihypertensive medication while being studied. So- Table 1. Characteristics of the assays used according to directional insert provided by the manufacturer. a Nichols Advantage Assay Adaltis Aldosterone Maia DSL Active Aldosterone Aldosterone In-house assay Tracer 125 I 125 I (chemiluminscence) 3 H Antibody Polyclonal Polyclonal Monoclonal Polyclonal Sample serum or plasma, L Working range, ng/l Intraassay variability, % Interassay variability, % Cross-reactivity, % Aldosterone 100 Aldosterone 100 Aldosterone 100 Negligible because cross-reacting steroids are removed by chromatographic purification Progesterone Progesterone Corticosterone 0.03 Testosterone Testosterone OH corticosterone 0.01 Corticosterone Corticosterone Cortisol 0.01 Cortisol Corticosterone Cortisol 0.01 Estrone Estrone Desoxycorticosterone 0.05 Estradiol Estradiol Dexamethasone 0.05 Estriol Estriol undetectable 5-Dihydroaldosterone hydroxycorticosterone ,5 -Tetrahydroaldosterone 1.1 Dexamethasone Dexamethasone Reference interval (standing) ng/l Serum: not given Serum: Serum: Serum and plasma: standing Plasma: Plasma: Plasma: values not given, but lower recumbent a Conversion: ng/l pmol/l. Cross-reacting steroids are removed by chromatographic purification.

3 Clinical Chemistry 52, No. 9, dium intake was unrestricted. None of the subjects had chronic renal failure (creatinine mol/l), a condition that previously had been associated with falsely increased results when direct methods had been used (7). Mean (SD) serum creatinine was 70.8 (8.9) mol/l for healthy individuals, 70.8 (17.7) mol/l for patients with essential hypertension, and 79.6 (26.5) mol/l for patients with PA. Participants did not take contraceptives or other potentially cross-reactive medication. The study protocol was approved by the ethics committee of the University of Freiburg, according to the requirements of the Declaration of Helsinki, and written informed consent was obtained from all participants. aldosterone assay methods In all blood samples, we determined aldosterone concentration by 4 different methods. As a comparison method, we used the specific in-house RIA (plasma samples) established at the Steroid Laboratory of the University of Heidelberg, using tritiated aldosterone ([1,2,4,6, 3 H]aldosterone; Amersham Biosciences) and a rabbit antialdosterone antiserum, raised and characterized in the steroid laboratory, as described elsewhere (13). Before RIA, we performed recovery-corrected extraction and chromatographic purification, thereby efficiently removing crossreacting serum components. In more detail, we carried out the chromatographic separation of aldosterone, as previously described (14), with minor modifications. Celite (Celite 545 AW; Sigma Aldrich) was used as an inert support for partition chromatography. The stationary phase consisted of 30% formamide in water; while the mobile phase consisted of a mixture of ethyl acetate in n-hexane with increasing polarity, eluting aldosterone with 50% ethyl acetate in n-hexane. The limit of blank [determined as mean (2SD) of blank] of the method was 1.5 pg per tube (7.5 ng/l). The recovery of a known amount of aldosterone determined repeatedly in quality control samples was (8.2)%. Furthermore, aldosterone concentration was determined by 3 commercially available assays according to the protocol given by the manufacturers. We used 2 RIAs without extraction in serum (Aldosterone Maia) and plasma samples (Active Aldosterone, DSL). Furthermore, serum aldosterone concentration was measured on a fully automated immunochemiluminescence analyzer (Aldosterone, Nichols Institute Diagnostics) (15). Additional characteristics of the assays are given in Table 1. statistical analysis Results are displayed as mean (SD), except for Fig. 1 where, for reasons of clarity, SE bars are shown. Statistical analysis of the data was performed using Microcal Origin 6.0 and SAS Pearson coefficient of correlation was calculated to describe the linear component of correlation. For the Nichols assay, values below the limit of detection (as given by the manufacturer, 15 ng/l) were set to 15 ng/l. This was the case in 57% of samples from Fig. 1. Mean (SE) aldosterone concentration during saline infusion in healthy individuals, patients with essential hypertension, and patients with PA. a a Measured by Adaltis Aldosterone Maia RIA, DSL Active Aldosterone RIA, Nichols Advantage Aldosterone CLIA, and in-house RIA after extraction and chromatography; linear regression healthy participants, 3% from patients with PA, and 46% from patients with essential hypertension. To compare the 3 commercial assays with the in-house RIA (reference method), we plotted the results of the assays as described by Bland and Altman (16), showing the difference (given as percent of the mean) against the mean of reference

4 1752 Schirpenbach et al.: Comparison of Commercial Aldosterone Assays Table 2. Correlation among assays (for samples of all subjects). Time after saline infusion In-house Adaltis DSL Nichols 0 min In-house Adaltis DSL Nichols min In-house Adaltis DSL Nichols min In-house Adaltis DSL Nichols min In-house Adaltis DSL Nichols Given in Pearson coefficient of correlation for the aldosterone concentration measured by 4 different assays in samples taken at 0 min, 120 min, 180 min, and 240 min after saline infusion. P was in all cases. method and each commercial assay, respectively. Furthermore, to compare the commercial assays with the reference method, we performed Deming regression analysis with Analyze-It Clinical Laboratory (V 1.67, Analyze-It Ltd.). Results During saline infusion, mean aldosterone concentrations progressively decreased in all 3 groups of individuals studied (Fig. 1). Although absolute values of the assays differed substantially, the course of mean aldosterone concentration during saline infusion was nearly parallel for the results obtained by different methods. Overall correlation between assays ranged from 0.74 to 0.98 (Pearson coefficient of correlation) and increased during the course of saline infusion: Correlation between assays before saline infusion was from 0.74 to 0.95, whereas at the end of saline infusion, correlation was between 0.93 and 0.98 (Table 2). At lower concentrations, aldosterone measured by Adaltis RIA and DSL RIA showed a better correlation than each of these RIAs with Nichols CLIA. In the range of ng/l (0 416 nmol/l), aldosterone (as determined by Adaltis or DSL) measurements by Nichols CLIA produced values below the limit of detection in a substantial part of the samples. Despite good overall correlation, absolute aldosterone concentrations differed dramatically (Table 3 and Fig. 1). Mean aldosterone concentrations measured by Adaltis RIA and DSL RIA were 2 to 3 times higher than those obtained by the reference method (in-house RIA). In contrast, mean values produced by Nichols CLIA better corresponded to those of the reference method (Table 3). In the Deming regression analysis, using the in-house RIA as the reference method, we determined that a comparison with the Nichols CLIA revealed a slope of 1.2 (r 0.91), whereas we found larger deviations for the Adaltis RIA (slope 1.5, r 0.73) and the DSL RIA (slope 1.4, r 0.85). These findings were confirmed when analyzing the data by means of the Bland-Altman plot (Fig. 2). Whereas the results obtained by the Nichols CLIA are scattered symmetrically on both sides of the x-axis (difference 0%), values obtained by the DSL and Adaltis RIA show a distribution clearly above the x-axis, with a mean deviation of 75% to 100% from the reference method. For all 3 commercial assays, absolute deviation from the reference method and the scatter around the mean deviation was most pronounced in the very low range of aldosterone concentrations [ 80 ng/l (220 nmol/l)] (Fig. 2). Discussion Accurate aldosterone measurement is necessary to differentiate among various disorders of the renin-angiotensinaldosterone axis (1, 2), first and foremost in the investigation of suspected PA. The aldosterone-to-renin ratio is currently advocated as the most sensitive screening test for this disorder (9, 10, 17, 18). Because the ratio depends on 2 variables, it is particularly susceptible to interlabo- Table 3. Aldosterone concentration (mean (SD) before, (0 min) and after (240 min), saline infusion after extraction and chromatography. Assay Time (min) Healthy participants (n 47) Essential hypertensives (n 30) Patients with PA (n 27) Adaltis Aldosterone (131) 216 (116) 392 (131) Maia, ng/l (25) 89 (25) 261 (200) DSL (101) 189 (109) 368 (174) Active Aldosterone, ng/l (18) 64 (25) 218 (178) Nichols Advantage (100) 97 (78) 246 (119) Aldosterone, ng/l (8) 18 (10) 147 (190) In-house RIA (83) 107 (93) 195 (108) After extraction, ng/l (18) 22 (19) 140 (144)

5 Clinical Chemistry 52, No. 9, Fig. 2. Bland-Altman plots of the difference (commercial assay minus reference method, expressed as a percentage of the mean) against the mean of in-house RIA (reference method) and each commercial assay (A, Adaltis RIA; B, DSL RIA; C, Nichols CLIA). The correlations by Deming regression were: Adaltis RIA 1.48 in-house RIA 55.7 (r 0.73; s y.x 73 ng/l); DSL RIA 1.42 in-house RIA 32.5 (r 0.85; s y.x 59 ng/l); Nichols RIA 1.20 in-house RIA 18.4 (r 0.91; s y.x 47 ng/l). ratory differences in choice and performance of the assays. Although renin is thought to have an unequally higher impact on the ratio than aldosterone (19, 20), differences in aldosterone concentrations because of different assay procedures can also lead to relevant alterations. Because of the recent advent of new immunoassays allowing direct measurement of renin concentration rather than renin activity, attention has focused primarily on renin rather than aldosterone measurements (15, 21, 22). The automated Nichols CLIA for measurement of aldosterone introduced along with the assay for renin concentration and allowing simultaneous determination in plasma samples has been less thoroughly studied, with only one study evaluating this assay to date (15): The authors report a coefficient of correlation between Nichols CLIA and a not-specified commercial RIA of 0.96 (slope 1.04). This agrees with our results on the overall correlation at different timepoints during saline infusion test, revealing coefficients of correlation between Nichols CLIA and the 2 commercial nonextraction RIAs between 0.91 and However, in our study comparing the Nichols CLIA with 2 commercial RIAs and, as reference method, an in-house aldosterone assay with extraction and chromatography, we found major differences in absolute mean values between these assays despite fairly high interassay correlations. The discrepancies among assay results could be a consequence of different factors: First, the specificity of the antisera or antibodies varies between assays. As a consequence, reported aldosterone concentrations might be falsely high because of cross-reacting steroids or other interfering substances, especially if assays without extraction and chromatography are used. Although crossreactivity of the antibodies reported by the manufacturers does not show major differences (Table 1), this does not exclude the influence of other potentially interfering substances. Second, variability could derive from the fact that aldosterone is measured in either plasma or serum. Although not explicitly investigated in our study, the choice of sample material seems to play a minor role. We found no better correlations between methods using the same or different sample material. Finally, and probably most important, calibration of the assays seems to be different, which, of course, has a strong impact on the results obtained. This might explain the systematically higher values for Adaltis and DSL RIAs when compared with the Nichols CLIA assay. An independent reference method based on mass spectrometry would be a way to further investigate this aspect. Best correlation among the assays was found in the lower to middle range of values observed at the end of the saline infusion test, when Pearson coefficient of correlation ranged from 0.93 to In contrast, at the upper and lower end of the concentration range discrepancies among assays increased. The increasing scatter of data points in the Bland-Altman plots for values below mean aldosterone concentrations of 80 ng/l clearly points to a significant lack of sensitivity of all commercial assays, making results in this concentration range less reliable.

6 1754 Schirpenbach et al.: Comparison of Commercial Aldosterone Assays Although an ideal assay for routine clinical application should be practicable, fast, and cost-effective, as well as accurate across the expected physiologic and pathologic ranges, these requirements are sometimes conflicting. It is important to keep in mind that a fast and simple assay providing wrong results may cause more harm than good. All three commercially available methods investigated in our study skipped the time-consuming and tedious steps of extraction and chromatographic separation of the samples. Major advantages of the recent Nichols CLIA assay are the automated system, time effectiveness, the option for simultaneous determination of renin concentration from the same specimen, and the lack of radioactive hazards. However, according to our results, the sensitivity of the Nichols CLIA assay might be an issue especially in healthy participants, because in this group a high percentage of samples fall below the limit of detection. The inadequate correlation among methods has important clinical implications. Dynamic test results, depending on the assay used, might be interpreted as abnormal if cutoff values from the literature are applied. For the saline infusion test, for example, different cutoff values have been recommended: Ganguly (23) refers to 85 ng/l (240 nmol/l), whereas in a recent review, Mulatero et al. (10) recommend 50 ng/l (140 nmol/l). The definition of assay-specific cutoff levels and a comparison of the assays in terms of accuracy in detecting PA were beyond the scope of our study. This would certainly have required not only a larger population but also control of potentially interfering preanalytic factors, such as antihypertensive medication, before saline infusion test. Whatever cutoff is used, however, their application will result in a higher number of false positive tests if aldosterone is measured by Adaltis RIA or DSL RIA. In contrast, measuring aldosterone with Nichols CLIA, despite an overall acceptable correlation of values, would lead to a number of false-negative test results. Many questions related to biochemical testing in diagnosis and differentiation of the subtypes of primary aldosteronism are, of course, related to difficulties in standardization of preanalytic conditions. Medication, salt intake, local protocol for performing dynamic tests, and many other potentially confounding factors vary among studies. However, our study emphasizes the need for harmonization and standardization of aldosterone assays as a prerequisite to compare results from different studies. The impact of the variability on the aldosterone/ renin ratio, which is being used increasingly as a screening variable, can be expected to be in the same magnitude as the differences in absolute aldosterone values. This will enormously affect the result of the ratio, and thereby also the prevalence of positive test leads to a distinct study. Uncritical application of published cutoff values may lead to serious consequences for the patients affected. Standardized assays, calibrated against a mass spectrometric reference method, would improve patient management and would allow more confidence in comparing results obtained in different laboratories. We acknowledge the generous gift of the Nichols Aldosterone CLIA assays for this study from Nichols Institute Diagnostics, San Clemente, CA. References 1. McKenna T, Sequeira S, Heffernan A, Chambers J, Cunningham S. Diagnosis under random conditions of all disorders of the reninangiotensin-aldosterone axis, including primary aldosteronism. J Clin Endocrinol Metab 1991;73: Mantero F, Bedendo O, Opocher G. Does dynamic testing have a place in the modern assessment of endocrine hypertension? J Endocrinol Invest 2003;26: Cartledge S, Lawson N. Aldosterone and renin measurements. Ann Clin Biochem 2000;37: Bayard F, Beitins I, Kowarski A, Migeaon C. Measurement of plasma aldosterone by radioimmunoassay. J Clin Endocrinol Metab 1970;31: Segre G, Brown E. Measurement of Hormones. In: Wilson J, Foster D, Kronenberg H, Larsen P, eds. Williams Textbook of Endocrinology, Vol. 9 ed. Philadelphia: WB Saunders, 1998: Stabler T, Siegel A. Chemiluminescence immunoassay of aldosterone in serum. Clin Chem 1991;37: Miller M, Sagnella G, MacGregor G. Extraction method and nonextracted kit method compared for measuring plasma aldosterone. Clin Chem 1997;43: Nussberger J, Waeber B, Brunner H, Burris J, Vetter W. Highly sensitive microassay for aldosterone in unextracted plasma: comparison with two other methods. J Lab Clin Med 1984;104: Reincke M, Seiler L, Rump LC. Normokaliämischer primärer hyperaldosteronismus. Dtsch Ärztebl 2003;100:A Mulatero P, Dluhy R, Giacchetti G, Boscaro M, Veglio F, Stewart P. Diagnosis of primary aldosteronism: from screening to subtype differentiation. Trends Endocrinol Metab 2005;16: Schirpenbach C, Seiler L, Maser-Gluth C, Rüdiger F, Nickel C, Beuschlein F, et al. Confirmatory testing in normokalaemic primary aldosteronism: the value of saline infusion test and urinary aldosterone metabolites. Eur J Endocrinol 2006;154: Fontes R, Kater C, Biglieri E, Irony I. Reassessment of the predictive value of the postural stimulation test in primary aldosteronism. Am J Hypertens 1991;4: Vecsei P, Abdelhamid S, Mittelstädt G, Lichtwald K, Haack D, Lewicka S. Aldosterone metabolites and possible aldosterone precursors in hypertension. J Steroid Biochem Mol Biol 1983;19: Abraham G, Buster J, Lucas L, Corrales P, Teller R. Chromatographic separation of steroid hormones for use in radioimmunoassays. Anal Lett 1972;5: Perschel F, Schemer R, Seiler L, Reincke M, Deinum J, Maser- Gluth M, et al. Rapid screening test for primary aldosteronism: ratio of plasma aldosterone to renin concentration determined by fully automated chemiluminescence immunoassays. Clin Chem 2004;50: Bland J, Altman D. Measuring agreement in method comparison studies. Stat Methods Med Res 1999;8: Fardella C, Mosso L, Gómez-Sánchez C, Cortés P, Soto J, Gómez L, et al. Primary aldosteronism in essential hypertensives: prevalence, biochemical profile, and molecular biology. J Clin Endocrinol Metab 2000;85:

7 Clinical Chemistry 52, No. 9, Hamlet S, Tunny T, Woodland E, Gordon R. Hamlet SM, Tunny TJ, et al. Is aldosterone/renin ratio useful to screen a hypertensive population for primary aldosteronism? Clin Exp Pharmacol Physiol 1985;12: Montori V, Schwartz G, Chapman A, Boerwinkle E, Turner S. Validity of the aldosterone-renin ratio used to screen for primary aldosteronism. Mayo Clin Proc 2001;76: Sealey J, Gordon R, Mantero F. Plasma renin and aldosterone measurements in low renin hypertensive states. Trends Endocrinol Metab 2005;16: Ferrari P, Shaw S, Nicod J, Saner E, Nussberger J. Active renin versus plasma renin activity to define aldosterone-to-renin ratio for primary aldosteronism. J Hypertens 2004;22: Trenkel S, Seifarth C, Schobel H, Hahn E, Hensen J. Ratio of serum aldosterone to plasma renin concentration in essential hypertension and primary aldosteronism. Exp Clin Endocrinol Diabetes 2002;110: Ganguly A. Primary aldosteronism. N Engl J Med 1998;25: Holland O, Brown H, von Kuhnert L, Fairchild C, Risk M, Gomez-Sanchez C. Further evaluation of saline infusion for the diagnosis of primary aldosteronism. Hypertension 1984;6: