ELSEVIER
Contents lists available at ScienceDirect
Clinica Chimica Acta
journal homepage: www.elsevier.com/locate/clinchim
E
Clinica Chimica Acta
Short communication
Cross reactions elicited by serum 17-OH progesterone and 11-desoxycortisol in cortisol assays
Julie Brossaud a,*, Pascal Barat b, Dominique Gualde a, Jean-Benoît Corcuff ª
a Department of Nuclear Medicine, University Hospital and University of Bordeaux, France
b Department of Paediatrics, University Hospital and University of Bordeaux, France
ARTICLE INFO
Article history: Received 30 March 2009 Received in revised form 22 May 2009 Accepted 22 May 2009 Available online 28 May 2009
Keywords:
Cortisol assay 17-OH progesterone 11-Desoxycortisol
Infant Cross reaction
ABSTRACT
Background: Different pathophysiological situations such as congenital adrenal hyperplasia, adrenocortical carcinoma, metyrapone treatment, etc. elicit specificity problems with serum cortisol assay.
Methods: We assayed cortisol using 2 kits and performed cross reaction studies as well as multiple regression analysis using 2 other steroids: 11-desoxycortisol and 17-OH progesterone.
Results: Analysis showed the existence of an analytical bias. Importantly, significantly different biases were demonstrated in newborns or patients taking metyrapone. Multiple regression analysis and cross reaction studies showed that 11-desoxycortisol level significantly influenced cortisol determination. Moreover, despite using the normal ranges provided by manufacturers discrepant results occurred such as 17% discordance in the diagnosis of hypocorticism in infants.
Conclusion: We wish to raise awareness about the consequences of the (lack of) specificity of cortisol assays with regard to the evaluation of hypocorticism in infants or when “unusual” steroids may be increased.
@ 2009 Elsevier B.V. All rights reserved.
1. Introduction
Cross reaction problems in steroids immunoassay are well de- scribed as these very similar molecules are assayed by competitive techniques that potentially overestimate true concentrations [1-3]. Overestimating cortisol concentrations can have serious consequences as it leads to under diagnosing hypocortisolism. Mass spectrometry provides gold standards but its common use is limited to standardize and compare kits [4,5]. The routine use of these immunoassay com- mercial kits usually rests upon reference ranges obtained in healthy adult (usually obtained in the morning). In such a physiological con- dition, cortisol concentrations are clearly above those of potentially interfering steroids thus minimizing the impact of cross reactions. Hence, a good agreement is obtained between different kits in these normal conditions.
Obviously, in a university hospital, the situations requesting cortisol assay depart from physiology. Indeed, in some pathological situations, the concentration of steroids closely related to cortisol dramatically increases (e.g. cortisol precursors in patients with congenital adrenal hyperplasia (CAH) [6] or with adrenocortical carcinoma [7] or under metyrapone treatment or undergoing a metyrapone test, or in pre- mature infants [8]). Manufacturers usually provide some data about
the cross reactivity of their assays although there is no mandatory list of compounds to be tested.
We investigated 2 cortisol assay kits to assess the clinical impact of possible differences of cross reactivity in real-life conditions in a university hospital, in which some physiological or pathological situ- ations require the assessment of cortisol concentrations.
2. Materials and methods
2.1. Materials
Cortisol concentration was assayed by Gamma Coat CA 1529, kit A (Clinical Assays, Stillwater, USA) and by Spectria cortisol RIA, kit B (Orion Diagnostic, Espoo, Finland). Inter-assay coefficients of variation (CV) were: kit A 7.7% and 3.5% (81 and 571 nmol/L) and kit B 7.4% and 9.7% (73.5 and 446 nmol/L). Normal ranges provided by the manufacturers were: [200-700] and [131-642] nmol/L for kit A and B, respectively. Both kit leaflets contained information about cross reactions: for 17-OH progesterone (17OHP), kit A 1.2 and kit B 0.02%, for 11-desoxycortisol (11DSC) kit A 6.3 and kit B 0.3%.
17OHP concentration was assayed by OHP-CT (Cis-bio international, Gif-sur-yvette Cedex, France) after diethyl ether extraction. Inter-assay CV was 8.3% and 7.8% (2.5 and 7.6 nmol/L). Normal ranges for serum 17OHP concentration (nmol/L) as provided by the manufacturer were: boys and girls 0-1 m [4.6-26.8] and [4.6-26.8], 1-2 m [3.6-8.5] and [1.6-4.6], 6-9 m [0.2-1.5] and [0.5-3.1], 9-12 m [0.2-1.5] and [0.5-2.1], 1 to 7 yr [0.25-1.6] and [0.25-1.6] and adult males [1.5-6.5] females follicular [0.5-2.6], luteal [2.0-10.0], menopausal phases [0.5-3.0]. 11DSC concentration was assayed by 11- DesoxycortisolCT 125I RIA (Biosource). Inter-assay CV was 10.9% and 5.4% (2.6 and 14.4 nmol/L). Normal serum 11DSC concentration as provided by the manufacturer was <20.8 nmol/L.
170HP and 11DSC (Steraloids, Inc. Newport, RI, USA) solutions (40 umol/L) were prepared in ethanol. Dilutions (1000, 500, 250, 125, 62.5, 31.25 nmol/L) were performed in steroid-free serum (Clinical Assays).
* Corresponding author. service de Médecine Nucléaire, Hôpital Haut-Lévêque, 33604 Pessac, France.
E-mail address: julie.brossaud@u-bordeaux2.fr (J. Brossaud).
RELATIVE DIFFERENCE
· Newborns
Metyrapone treatment
160%
Healthy adults
120%
80%
40%
0%
-40%
200
400
600
800
1000
Mean (nmol/L)
-80%
-120%
-160%
2.2. Patients
Sera (n=114) from the left-over of routine 94 patients (50 infants <1 yr, 13 children 1-2 yr, 5 children with 21-hydroxylase deficiency, 12 adults with metyrapone treatment and 14 healthy adults) were investigated. Sera (n = 36) were also collected for a multivariate analysis with cortisol (A and B kit), 11DSC and 17OHP measurements: 22 subjects presented high 11DSC and/or 17OHP concentrations (≥28 nmol/L and ≥10 nmol/L, respectively), 14 presented normal 11DSC and 17OHP levels.
2.3. Statistical analysis
Paired comparisons were performed using the paired t-test and unpaired comparisons were performed by the Mann-Whitney test (StatView, SAS Institute). Relationships be- tween data were assessed by multiple regression analysis.
3. Results
Two RIA kits were used to evaluate the degree of agreement be- tween cortisol concentrations: concentrations obtained with the kit A were higher than those obtained with the kit B (n = 114, p<0.0001).
In a first approach, the ratios of cortisol concentrations obtained in both kits were calculated ([A]/[B]). We found differences of [A]/ [B] ratios between healthy control adults (ratio 1.2) and infants <1 yr (ratio 1.8, p<0.001), patients with CAH (ratio 6.4, p<0.005), patients under metyrapone treatment (ratio 1.5, p<0.05) but not infants of 1 to 2 yr (ratio 1.3, p>0.05). As a consequence, and despite different normal ranges were provided by the manufacturers, these values would have lead to different diagnoses (e.g. eucorticism vs hypocorti- cism). For instance, 16/65 sera from infants (<1 yr with or without CAH) had low cortisol concentrations with the kit A vs 27/65 with the kit B (p<0.05, chi2; 17% discrepancies). Thus, we positioned the data on a Bland-Altman plot to investigate the possibility of a variable bias (Fig. 1). About 75% infants, 14% of adults and 47% adults under metyr- apone treatment presented a relative difference higher than 40%.
This discordance could be explained by the 2 antibodies expressing different specificities against structurally close compounds (i.e. dif- ferent levels of cross reactivity). Such a hypothesis was tested by carrying out a multivariate analysis (dependent variable: cortisol concentrations (kit A) and independent variables: cortisol concentra- tions (kit B), 17OHP and 11DSC concentrations). This analysis showed a significant impact of 11DSC concentration over cortisol determina- tion (kit A) (n=36, p=0.0013) but no clear interference of 17-OH progesterone concentration (p=0.065). Experimental elucidation of this point was undertaken to investigate cross reactivity in immu- noassays: a liaison percentage B/Box 100 was calculated for kits A and
B for each sample of diluted compounds ( Fig. 2) [9]. 17OHP cross reactivity was found minor (1000 nmol/L of 17OHP caused a cortisol displacement of 9.8 and 1.4% using kits A and B, respectively). Con- versely, 11DSC displacement was noticeable (1000 nmol/L of 11DSC caused a cortisol displacement of 33.6 and 28.3% using kits A and B, respectively).
Informal investigations were conducted from some leaflets and manufacturers providing cortisol assay in France (Diasorin, Beckman- Coulter, Orion Siemens Abbott, Roche, Biomerieux). When cross re- activities are mentioned their ranges are [0.02-1.9] and [0.3-21.6] % for 17OHP and 11DSC, respectively.
4. Discussion
Assay accuracy is a major point when a diagnosis rests upon hormonal threshold concentrations. Adrenal status for instance can be established by cortisol concentrations at 8 h. An overestimation of cortisol concentrations leading to a subsequent failure to diagnose hypocorticism may have dramatic consequences. In the current study, we showed an example of 2 kits failing to identically assess adrenal status when an excess of “unusual” steroids is present in the serum of patients despite providing kit-specific normal ranges. This was especially true for infants. This could be related to elevated concen- trations of 11DSC but many other steroids not assayed here may be involved [10].
Obviously, accuracy and cross reactivity are antibody-specific and thus variable amongst commercial kits. Our point is to underline that using kit-specific normal values should not be considered failsafe when departing from “physiological” conditions i.e. in the likely pres- ence of amounts of low-level steroids. This can be as high as an apparent 6.4 fold increase of cortisol levels in individual children suffering from CAH compared to the “expected” 1.2 bias in normal adults. When choosing an assay, great care should thus be paid if clinical situations depart from healthy adulthood. Many manufacturers provide cross reactivity indications in the assay booklets but, as we
A
100
80
60
B%
40
20
0
B
100
80
B%
60
40
20
0
10
100
1000
10000
Concentration (nmol/L)
found here, the data are not always accurate or clinically relevant. Furthermore, a list of all interacting compounds cannot be realistically expected.
In conclusion, one should be aware of the occurrence of cross reactions when investigating cortisol levels in situations of unusual steroids excess and in infants. Ideally, kit-specific normal ranges of cortisol values should be established in normal infant populations.
Acknowledgment
The authors wish to thank Dr A Georges for her help in reading an early version of the manuscript.
References
[1] Jiang NS, Machacek D, Wadel OP. Comparison of clinical assays for serum corticosteroids. Clin Chem 1975;21:387-91.
[2] Andrew R. Clinical measurement of steroid metabolism. Best Pract Res Clin Endocrinol Metab 2001;15:1-16.
[3] Davies C. Concepts. In: Wild D, editor. Concepts. Amsterdam: Elsevier Ltd; 2005. p. 103-35.
[4] Wilke TJ, Hirning DA. The evaluation of five commercial serum cortisol kits regarding precision and accuracy. Clin Biochem 1984;17:311-6.
[5] Gaskell SJ, Sieckmann L. Mass-spectrometric determination of serum cortisol: comparison of data from two independent laboratories. Clin Chem 1986;32:536-8.
[6] Merke DP, Bornstein SR. Congenital adrenal hyperplasia. Lancet 2005;365:2125-36.
[7] Powell-Jackson JD, Calin A, Fraser R, et al. Excess deoxycorticosterone secretion from adrenocortical carcinoma. Br Med J 1974;2:32-3.
[8] Janzen N, Peter M, Sander S, et al. Newborn screening for congenital adrenal hyperplasia: additional steroid profile using liquid chromatography-tandem mass spectrometry. J Clin Endocrinol Metab 2007;92:2581-9.
[9] Abraham GE. Radioimmunoassay of steroids in biological fluids. J Steroid Biochem 1975;6:261-70.
[10] Tonetto-Fernandes V, Lemos-Marini SH, Kuperman H, Ribeiro-Neto LM, Verreschi IT, Kater CE. Serum 21-deoxycortisol, 17-hydroxyprogesterone, and 11-deoxycortisol in classic congenital adrenal hyperplasia: clinical and hormonal correlations and identification of patients with 11beta-hydroxylase deficiency among a large group with alleged 21-hydroxylase deficiency. J Clin Endocrinol Metab 2006;91:2179-84.