Sage Journals: Discover world-class research

Abstract

Background

The analysis of steroids in biological matrices is challenging. One can apply immunoassay as well as gas and liquid chromatography with various types of detection, depending on the available equipment and the experience of the analyst. The question is how the methods are interchangeable between themselves. Doubts were reported having compared immunoassays and chromatography-mass spectrometry, but there are scarce data on chromatographic methods with detection types other than mass spectrometry.

Methods

Here, we present the detailed comparison of two liquid chromatographic methods for the determination of free urinary cortisol and cortisone: one with fluorescence detection (high-performance liquid chromatography [HPLC-FLD]) and the other with tandem mass spectrometry (HPLC-MS/MS). The comparison was made with 199 human urine samples. The data analysis included Passing–Bablok and Deming regression, Bland–Altman test, Wilcoxon test, mountain plot and Lin’s concordance correlation coefficient.

Results

The validation data indicated that both methods met the requirements of the European Medicines Agency. However, the statistical analysis revealed the systematic bias between the two assays. The Passing–Bablok and the Deming tests showed that the HPLC-FLD method overestimated results for cortisol and underestimated measurements for cortisone. The Bland–Altman analysis estimated the mean differences between the methods: 18.8 nmol/L for cortisol and −16.9 nmol/L for cortisone measurement.

Conclusions

Both methods’ results led to the same conclusion in observational studies, but the techniques are not interchangeable. The literature data, the observations from the clinical setting and our experience clearly indicate that the future of steroid measurements will belong to chromatography coupled with mass spectrometry.

Keywords

Chromatography mass spectrometry statistics

Introduction

The analysis of steroids in biological fluids is quite challenging. On the one hand, we need the commercially available, simple methods which could be used in everyday clinical practice. On the other hand, the scientists require highly specific and sensitive methods to be applied in research focusing on the analysis of complex hormonal profiles in various biological matrices. Presently, the immunoassays (IAs) still dominate in clinical laboratories, despite their recognized disadvantages: potentially low specificity, cross-reactivity, hard-to-control matrix effects, ability to test only one compound at a time. This is because of IAs’ relatively low price, technical ease and no need for highly specialized equipment. The research laboratories focus rather on chromatographic methods, which allow separating the compounds of interest from the other biological components with resembling structure. The result is the ability to determine multiple substances in one analytical run.¹

Chromatographic methods have been widely used in the analysis of steroids for many years. Initial methods were based on paper and thin-layer systems, then the gas chromatography (GC) and the high-performance liquid chromatography (HPLC) were introduced.² Continuously, the application of the ultra high performance liquid chromatography enabled the better resolution and provided with advantage in terms of time-saving and increased laboratory throughput. The pivot of chromatographic bioanalytical methods is the system of detection. Less specific detectors (flame ionization detector,^3–5 ultraviolet – UV,^6–8 fluorescence detection – FLD⁹) require very scrupulous sample preparation and complete chromatographic resolution of analytes to fulfil the validation criteria and ensure the reliable results. More specific ones, like mass spectrometry (MS), enable one to distinguish and simultaneously determine related compounds. Moreover, one has to realize that some methods require the process of derivatization. GC needs the transformation of steroid compounds into derivatives (most often via silylation) to increase their volatility.^{3–5,10–12} HPLC-FLD requires the chemical modification of steroids to obtain the fluorescent compounds.^9,13

With such a wide range of analytical possibilities, the question is how comparable the different methods are. Here, we present the statistical comparison of two methods for cortisol (F) and cortisone (E) determination, both based on HPLC, but with different systems of detection: FLD and MS/MS. The analysis was performed on 199 urine samples, which provides a sufficient number of data for the convincing and reliable statistical comparison.

Materials and methods

Chemicals and reagents

Standards of F and E, prednisolone and hydrocortisone-9,11,12,12-D₄ (F-D₄), as well as ⩾95% formic acid, were purchased in Sigma-Aldrich (St. Louis, USA). The same company also provided quinuclidine and 9-anthroyl nitrile (9-AN). Organic solvents: acetone, acetonitrile (ACN), dichloromethane (DCM), n-hexane, methanol, all at least HPLC grade, were purchased from Merck (Darmstadt, Germany). Potassium dihydrogen phosphate (Xenon, Łódź, Poland), disodium hydrogen phosphate anhydrous and ≥85% phosphoric acid (both from Fluka Chemie, Buchs, Switzerland) were used to prepare buffers and mobile phase for the HPLC-FLD method. Ultrapure water was obtained using the Simplicity UV system (Merck-Millipore, Darmstadt, Germany), while the mobile phase for HPLC-MS/MS analysis was prepared with HPLC-MS-grade water from Merck (Darmstadt, Germany) or J.T.Baker (Deventer, The Netherlands). The details concerning the preparation of all reagents and solutions were presented previously.^9,14,15

Biological samples

The urine samples were obtained from 211 women in the third trimester of pregnancy. The pregnant women were admitted to Gynecology and Obstetrics Clinical Hospital of Poznan University of Medical Sciences and recruited to participate in a research project concerning the activity of 11β-hydroxysteroid dehydrogenase 2 in hypertensive disorders of pregnancy.¹⁵ The study was approved by the local Bioethical Committee, and the informed consent was obtained from all study participants. Each participant collected the urine for 24 h, then the total volume was noted, and the aliquot of 5 mL was taken for the analysis.

HPLC-FLD

Equipment and conditions

The HPLC-FLD method was based on the method previously developed by our research group⁹ with some additional modifications. Those included the use of Kinetex XB-C18 column (100 × 4.6 mm; 5 μm) guarded by a pre-column SecurityGuard ULTRA C18 (4.6 mm; 5 μm), both from Phenomenex (Torrance, USA). The column temperature was held at 60°C. The analyses were performed on HPLC apparatus HP 1100 (Hewlett-Packard, USA) using the fluorescence detector (G1321A) with the wavelengths of excitation and emission set at 360 and 460 nm, respectively. The mobile phase consisted of ACN and 0.3 mM orthophosphoric acid (470: 530; v/v), and its flow rate was fixed at 2 mL/min. The injection volume was 50 μL.

Preparation of samples

The preparation of the biological samples comprised three steps: liquid–liquid extraction, derivatization with the use of 9-AN and the purification of the samples by the solid-phase extraction (SPE) with C18 cartridges, as described in details previously.⁹ Shortly, 0.5 mL of urine was mixed with 1 mL of phosphate buffer (pH 7.5); prednisolone solution (internal standard, IS); ACN (replacing standard solutions in quality control samples) and 4 mL of DCM and shaken for 10 min. The mixture was then cooled and centrifuged. The organic layer was collected and dried under the gentle stream of nitrogen at 40°C. The dried residue was dissolved in 9-AN (150 μL) and then a mixture of triethylamine and quinuclidine (100 μL; 1:1, v/v). Next, it was stored for 30 min in water-free, dark conditions at room temperature to obtain fluorescent derivatives of F and E. The liquid solvents were again evaporated (at 30°C), and the dry residue was dissolved in ACN/water mixture (1:4, v/v) and transferred to SPE cartridges. The analytes were purified by rinsing the cartridges with water and a mixture of n-hexane and DCM, and then eluted with acetone. The eluate was evaporated to dryness at 40°C. The residue was dissolved in ACN and mobile phase, and the solution was injected into HPLC system.

HPLC-MS/MS

Equipment and conditions

The details were presented previously.¹⁴ Briefly, the analysis was carried out in the Shimadzu system (Kyoto, Japan). It comprised the Nexera chromatograph interfaced to a triple quadrupole mass spectrometer LCMS-8030, equipped with electrospray ionization (ESI). The separation was accomplished using the Kinetex XB-C18 column (2.1 × 100 mm; 2.6 μm) from Phenomenex (Torrance, USA), and isocratic elution with the mobile phase (0.1% formic acid in H₂O: 0.1% formic acid in ACN; 762:238, v/v) pumped with a flow rate of 0.30 mL/min. The desolvation line and the heat block were maintained at 250°C and 400°C, respectively. Nitrogen was used as both the nebulizing gas (2 L/min) and the drying gas (15 L/min). Argon served as the collision gas (pressure 230 kPa). The electrospray needle voltage was 4.5 kV. The MS/MS detection processed in multiple reaction monitoring mode, as presented in detail elsewhere.¹⁴

Preparation of samples

The urine (1 mL) was mixed with 2 mL of phosphate buffer (pH 7.5), spiked with F-D₄ (IS) and pure ACN (analytes’ solution in ACN was used in calibration samples), and extracted with 4 mL of DCM. The organic layer was then collected and evaporated to dryness at 40°C. The residue was dissolved in 80 μL of methanol and injected into the chromatographic column (please see Kosicka et al.¹⁴ for details).

Statistics

Statistical analysis was carried out using the MedCalc software v17.9.7 (Ostend, Belgium). Whenever applicable, a P-value of <0.05 was considered significant. The distribution of data was always assessed with Shapiro-Wilk test. Before the analyses, cases with the results below the lower limit of quantitation (LLOQ) of any method were excluded. Such a procedure allowed to avoid a potential bias associated with different limits of quantification. Finally, comparison of the methods was performed on the results obtained from the analysis of 199 urine samples.

The Bland–Altman analysis¹⁶ was performed to study the mean bias between the HPLC-FLD and HPLC-MS/MS measurements and to construct the limits of agreement (mean bias ± 1.96 of standard deviation; SD) with their respective confidence intervals (CIs). This technique involved constructing a graph with the absolute difference between the two paired measurements plotted against their mean value.

Subsequently, the Passing–Bablok regression¹⁷ and the Deming regression¹⁸ enabled estimating if there was a constant bias between the two assays. Their plots expressed the relationship between the concentrations of compounds determined by HPLC-FLD and HPLC-MS/MS method. The regression lines’ slopes and intercepts with their 95% CIs were calculated. If 95% CI for the slope contained 1 and the 95% CI for the intercept contained 0, neither proportional nor systematic differences between the two methods were recognized. The CV of the measurements, required for the Deming regression, was set at 15%, which is the highest accepted value in the European Medicines Agency (EMA) validation criteria.¹⁹ The Passing–Bablok regression required no specific assumptions regarding the measurement errors or distribution of the samples. Therefore, it was more suitable concerning the lack of normal distribution. However, the Passing–Bablok test demands the linear relationship between the paired observations, which was confirmed with the Cusum test.

The next step was the comparison of measurements obtained by both methods using a test for paired samples. As the variables did not follow the normal distribution, the Wilcoxon test was applied.

Additionally, the agreement between the continuous variables was assessed by calculating the Lin’s concordance correlation coefficient.²⁰ This parameter evaluates the degree to which the paired measurements conform to a line of equality (45° line through the origin). The estimated coefficient is based on the Pearson’s correlation coefficient (a measure of precision) and a bias correction factor (a measure of accuracy). The value of the concordance correlation coefficient indicates the strength of agreement as: almost perfect (>0.99), substantial (0.95–0.99), moderate (0.90–0.95) or poor (<0.90). Additionally, the mountain plots²¹ were constructed to visualize the distribution of differences better. If two assays are unbiased with respect to each other, the scatter plots should be centered over zero.

Results and discussion

Validation of methods

The most important validation parameters of the methods are presented in Table 1. Both HPLC-FLD and HPLC-MS/MS methods were validated and met the requirements contained in the EMA guidelines.¹⁹ Briefly, the HPLC-MS/MS method is characterized by much shorter retention time of the analytes and is more sensitive than the HPLC-FLD. Using the two-fold higher volume of urine than the HPLC-FLD (1.0 vs. 0.5 mL), the HPLC-MS/MS provided a 10-fold lower value of LLOQ.

Table 1.

Comparison of the validation parameters.

	Cortisol (F)		Cortisone (E)
	HPLC-FLD	HPLC-MS/MS	HPLC-FLD	HPLC-MS/MS
Retention time (min)	11.8	5.3	13.9	5.7
LLOQ (nmol/L)	27.6	2.8	27.7	2.8
Linearity (nmol/L)	27.6–2758.9	2.8–2758.9	27.7–2774.4	2.8–2774.4
Precision (CV) (%)
Within-run	1.7–12.89.6 for LLOQ	2.0–5.315.2 for LLOQ	2.6–15.114.9 for LLOQ	2.7–9.79.8 for LLOQ
Between run	4.9–9.317.3 for LLOQ	2.3–10.96.3 for LLOQ	8.0–13.215.6 for LLOQ	2.6–14.65.6 for LLOQ
Accuracy (%)
Within-run	99.2–107.5105.8 for LLOQ	92.8–106.295.7 for LLOQ	84.6–103.194.6 for LLOQ	88.0–105.190.6 for LLOQ
Between-run	98.6–105.5106.9 for LLOQ	94.6–106.8100.1 for LLOQ	94.0–101.6100.1 for LLOQ	95.4–100.1103.3 for LLOQ

LLOQ: lower limit of quantitation; HPLC-FLD: high-performance liquid chromatography-fluorescence detection; HPLC-MS/MS: high-performance liquid chromatography-tandem mass spectrometry.

Comparison of methods

The Shapiro-Wilk test did not confirm that the differences between the measurements nor their log-transformations follow the normal distribution. Therefore, we decided to use the original data for further analyses.

The Bland–Altman analysis demonstrated a positive systematic error for F and a negative one for E (Figure 1(a) and (b)) between the HPLC-FLD measurements referenced to the HPLC-MS/MS ones, but no signs of a proportional error. The mean differences between the methods were 18.8 nmol/L and −16.9 nmol/L, respectively, and their 95% CIs did not cover the line of equality indicating significant biases between the two assays.

Figure 1.

Results from the Bland–Altman analysis (a and b), the Passing–Bablok regression (c and d) and the mountain plots (e and f) for cortisol and cortisone; a and b: mean (horizontal solid bold line), LoA (horizontal dashed lines) and line of equality (horizontal solid thin line); c and d: the Passing–Bablok regression (solid bold line), its 95% CI (dashed bold lines) and identity line (dotted thin line).HPLC-FLD: high-performance liquid chromatography-fluorescence detection; HPLC-MS/MS: high-performance liquid chromatography-tandem mass spectrometry.

The Passing–Bablok and the Deming models of regression are applicable when a linear model fits the data. The Cusum test for linearity did not show significant deviation from linearity either for F (P = 0.90) or for E (P = 0.56). Both the Passing–Bablok and the Deming models showed similar results. No proportional differences were observed because the 95% CIs for the slopes contained the value 1 for all analyses (Table 2). A systematic error was found between the assays (Figure 1(c) and (d)) which confirmed the results of the Bland–Altman analysis. The HPLC-FLD method overestimated results for F (the calculated intercepts and their 95% CI from both studies were above 0) and underestimated measurements for E (the intercepts each time were lower than 0) when compared to the HPLC-MS/MS method.

Table 2.

Results from the analyses comparing HPLC-FLD and HPLC-MS/MS method for determination of cortisol and cortisone in human urine.

	Cortisol (F)	Cortisone (E)
Bland–Altman analysis
Mean difference (95% CI) (nmol/L)	18.8 (15.6–21.9)	−16.9 (−24.5 to −9.2)
Lower LoA (95% CI) (nmol/L)	−25.0 (−30.3 to −19.6)	−123.6 (−136.6 to −110.6)
Upper LoA (95% CI) (nmol/L)	62.5 (57.2–67.9)	89.9 (76.9–102.9)
Passing–Bablok regression
Slope (95% CI)	1.060 (0.993–1.134)	1.007 (0.916–1.112)
Intercept (95% CI)	10.870 (5.340–15.844)	−24.646 (−52.200 to −0.252)
Deming regression
Slope (95% CI) ± Standard error	1.049 (0.960–1.138) ± 0.045	1.077 (0.953–1.201) ± 0.063
Intercept (95% CI) ± Standard error	14.878 (8.366–21.389) ± 3.302	−39.109 (−72.222 to −5.995) ± 16.792
Wilcoxon test	P <0.0001	P <0.0001
Medians (95% Cl) (nmol/L)
HPLC-FLD HPLC-MS/MS	89.0 (79.6–94.8) 68.9 (62.9–76.3)	264.6 (248.1–274.0) 277.7 (263.8–299.4)
Interquartile range (25–75 percentile) (nmol/L)
HPLC-FLD HPLC-MS/MS	62.8–121.444.3–100.8	206.2–324.2231.8–339.8
Concordance correlation coefficient	0.841 (0.800–0.875)	0.800 (0.745–0.844)
Pearson’s coefficient	0.902	0.815
Bias correction	0.932	0.981

LoA: limit of agreement; CI: confidence interval; HPLC-FLD: high-performance liquid chromatography-fluorescence detection; HPLC-MS/MS: high-performance liquid chromatography-tandem mass spectrometry.

The assessment of the mountain plots (Figure 1(e) and (f)) revealed that the mean difference between HPLC-FLD and HPLC-MS/MS was about 15.4 nmol/L for F and −22.5 nmol/L for E.

The existence of a systematic error in the analyses mentioned above explains the differences revealed by the Wilcoxon test (Table 2). It showed that the concentrations of F determined by HPLC-FLD were significantly higher than those by HPLC-MS/MS (89.0 vs. 68.9 nmol/L, P < 0.0001), and the concentrations of E were significantly lower (264.6 vs. 277.7 nmol/L, P < 0.0001).

The calculated Lin’s concordance correlation coefficients were 0.8412 for F and 0.7999 for E with respective Pearson’s coefficients 0.9025 and 0.8152. These results indicate that despite powerful positive correlations between the measurements (Pearson’s coefficient >0.80), the agreement between those two assays was relatively weak (Lin’s concordance correlation coefficient <0.90). The phenomenon could be explained by the fact that the Pearson’s coefficient explains only a linear association between variables and not the actual compatibility between assays. Therefore, it is not advised as an independent tool in method comparison studies.²²

Summary

The literature data on the compatibility of bioanalytical methods are equivocal.^23–28 Usually, a trend toward higher F concentrations in biological matrices was observed in IAs as compared to MS/MS assays,^27,28 especially in patients undergoing metyrapone treatment.²⁶ This phenomenon is explained by the cross-reactivity of antibodies against F with structurally related compounds. Moreover, when IAs methods were involved, the F assay’s bias was reported to be affected by patient gender, and different biological matrix components, which resulted from various clinical conditions of the patients.²⁹ Such interferences are more natural to avoid in HPLC and GC methods owing to extraction, the chromatographic resolution itself and a system of detection. On the other hand, an excellent agreement in urinary F measurements by GC-MS and HPLC-MS/MS was previously reported.²⁸ Such high compatibility of the chromatographic methods is hardly achievable between MS and other types of detection. The presented results indicate that the compared HPLC-FLD and HPLC-MS/MS methods for determination of F and E in urine are not interchangeable. They both fulfilled the criteria for analytical methods established by EMA,¹⁹ but the systematic biases exist between the assays. They resulted in overestimating concentrations for F and underestimating values for E in HPLC-FLD as compared with HPLC-MS/MS. The possible hydrolysis of the steroid conjugates in the analytical process needs to be addressed. It has to be emphasized though that in the analysis of glucocorticoid balance in pregnant women with hypertension, we came to the same conclusions when applying HPLC-FLD¹⁵ and when repeating and extending the analyses using the HPLC-MS/MS.³⁰ The value of urinary free cortisol (UFF) should be lower than the urinary free cortisone (UFE) in healthy humans. The UFF/UFE ratio usually is close to 0.5 and should not be higher than 0.8.^31,32 Higher values indicate dysfunction of the 11β-hydroxysteroid dehydrogenase 2 which converts cortisol to cortisone in the kidneys and in placenta.³³ Lower values were observed in our project involving women in the third trimester of pregnancy. In the cohort, the metabolism of cortisol was increased and the levels of UFF were from 3 (in normotensive controls) to 5 (hypertensive patients) times higher than those of UFE.³⁰ Therefore, our experience shows that both HPLC-FLD⁹ and HPLC-MS/MS¹⁴ methods are suitable for the observational study of glucocorticoid balance in hypertensive pregnant women. However, the assays cannot be compared between themselves because they are not in total agreement. There are doubtless advantages of the HPLC-MS/MS methods (specificity, lower LLOQ, ease of samples preparation), which make it the method of choice for our further research. In conclusion, all of the above makes it quite obvious that the MS measurements will defeat the opponents and be the future method of choice in the steroid analysis.

Footnotes

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The research was supported by the National Science Center (Narodowe Centrum Nauki) in Poland (grant number 2012/05/B/NZ7/02532).

Ethical approval

The study was approved by the Bioethical Committee (approval nos 954/11 and 991/15) at Poznan University of Medical Sciences.

Guarantor

KK, AS, FKG.

Contributorship

AS and KK have equally contributed to the paper. KK conceived the study, performed the analyses, analysed the data and wrote the first draft of the manuscript; AS performed the analyses, analysed the data and wrote the first draft of the manuscript; ASG was involved in protocol development, patient recruitment and data analysis; MK was involved in protocol development, patient recruitment and data analysis; GB was involved in protocol development and supervised the clinical aspects of the study; FK was involved in protocol development and supervised the analytical aspects of the study. All authors reviewed and edited the manuscript and approved the final version of the manuscript.

References

Taylor

Keevil

Huhtaniemi

IT.

Mass spectrometry and immunoassay: how to measure steroid hormones today and tomorrow. Eur J Endocrinol 2015; 173: D1–D12.

Honour

JW.

Steroid profiling. Ann Clin Biochem 1997; 34: 32–44.

Shackleton

CHL

Honour

JW.

Simultaneous estimation of urinary steroids by semiautomated gas chromatography. investigation of neo-natal infants and children with abnormal steroid synthesis. Clin Chim Acta 1976; 69: 267–283.

Kotłowska

Maliński

Sworczak

et al . The urinary steroid profile in patients diagnosed with adrenal incidentaloma. Clin Biochem 2009; 42: 448–454.

Poór

Bíró

Bufa

et al . Urinary steroids in young women with eating disorders. J Biochem Biophys Methods 2004; 61: 199–205.

Lin

C-L

T-J

Machacek

et al . Urinary free cortisol and cortisone determined by high performance liquid chromatography in the diagnosis of Cushing’s syndrome. J Clin Endocrinol Metab 1997; 82: 151–155.

Rouits

Boisdron-Celle

Morel

et al . Simple and sensitive high-performance liquid chromatography method for simultaneous determination of urinary free cortisol and 6β-hydroxycortisol in routine practice: for CYP 3A4 activity evaluation in basal conditions and after grapefruit juice intake. J Chromatogr B 2003; 793: 357–366.

Furuta

Mori

Suzuki

et al . Simultaneous determination of 6β-hydroxycortisol and cortisol in human urine by liquid chromatography with ultraviolet absorbance detection for phenotyping the CYP3A activity determined by the cortisol 6β-hydroxylation clearance. J Chromatogr B 2004; 801: 165–171.

Główka

Kosicka

Karaźniewicz-Łada

HPLC method for determination of fluorescence derivatives of cortisol, cortisone and their tetrahydro- and allo-tetrahydro-metabolites in biological fluids. J Chromatogr B 2010; 878: 283–289.

10.

Bowden

Colosi

Mora-Montero

et al . Enhancement of chemical derivatization of steroids by gas chromatography/mass spectrometry (GC/MS). J Chromatogr B Analyt Technol Biomed Life Sci 2009; 877: 3237–3242.

11.

Matysik

Schmitz

Determination of steroid hormones in human plasma by GC–triple quadrupole MS. Steroids 2015; 99: 151–154.

12.

Krone

Hughes

Lavery

et al . Gas chromatography/mass spectrometry (GC/MS) remains a pre-eminent discovery tool in clinical steroid investigations even in the era of fast liquid chromatography tandem mass spectrometry (LC/MS/MS). J Steroid Biochem Mol Biol 2010; 121: 496–504.

13.

Bayliss

MAJ

Homer

Shepherd

MJ.

Anthracene-9-carbonyl chloride as a fluorescence and ultraviolet derivatising reagent for the high-performance liquid chromatographic analysis of hydroxy compounds. J Chromatogr A 1988; 445: 393–402.

14.

Kosicka

Siemiątkowska

Pałka

et al . Detailed analysis of cortisol, cortisone and their tetrahydro- and allo-tetrahydrometabolites in human urine by LC–MS/MS. J Pharm Biomed Anal 2017; 140: 174–181.

15.

Kosicka

Siemiątkowska

Krzyścin

et al . Glucocorticoid metabolism in hypertensive disorders of pregnancy: analysis of plasma and urinary cortisol and cortisone. Plos One 2015; 10: e0144343.

16.

Bland

Altman

DG.

Measuring agreement in method comparison studies. Stat Methods Med Res 1999; 8: 135–160.

17.

Passing

Bablok

A new biometrical procedure for testing the equality of measurements from two different analytical methods. Application of linear regression procedures for method comparison studies in clinical chemistry, part I. J Clin Chem Clin Biochem 1983; 21: 709–720.

18.

Cornbleet

Gochman

Incorrect least-squares regression coefficients in method-comparison analysis. Clin Chem 1979; 25: 432–438.

19.

European Medicines Agency. Guideline on bioanalytical method validation, www.ema.europa.eu/docs/en_GB/document_library/Scientific_guideline/2011/08/WC500109686.pdf (2011, accessed 15 March 2018).

20.

Lin

LI.

A concordance correlation coefficient to evaluate reproducibility. Biometrics 1989; 45: 255–268.

21.

Krouwer

Monti

KL.

A simple, graphical method to evaluate laboratory assays. Eur J Clin Chem Clin Biochem J Biochem 1995; 33: 525–527.

22.

Watson

Petrie

Method agreement analysis: a review of correct methodology. Theriogenology 2010; 73: 1167–1179.

23.

Fanelli

Belluomo

Di Lallo

et al . Serum steroid profiling by isotopic dilution-liquid chromatography–mass spectrometry: comparison with current immunoassays and reference intervals in healthy adults. Steroids 2011; 76: 244–253.

24.

Koal

Schmiederer

Pham-Tuan

et al . Standardized LC-MS/MS based steroid hormone profile-analysis. J Steroid Biochem Mol Biol 2012; 129: 129–138.

25.

Monaghan

Owen

Trainer

et al . Comparison of serum cortisol measurement by immunoassay and liquid chromatography-tandem mass spectrometry in patients receiving the 11β-hydroxylase inhibitor metyrapone. Ann Clin Biochem 2011; 48: 441–446.

26.

Monaghan

Owen

Trainer

et al . Response to ‘Comparison of serum cortisol measurement by immunoassay and liquid chromatography-tandem mass spectrometry in patients receiving the 11β-hydroxylase inhibitor metyrapone’ by Halsall et al. Ann Clin Biochem 2012; 49: 204–205.

27.

Miller

Plessow

Rauh

et al . Comparison of salivary cortisol as measured by different immunoassays and tandem mass spectrometry. Psychoneuroendocrinology 2013; 38: 50–57.

28.

Wood

Ducroq

Fraser

et al . Measurement of urinary free cortisol by tandem mass spectrometry and comparison with results obtained by gas chromatography-mass spectrometry and two commercial immunoassays. Ann Clin Biochem 2008; 45: 380–388.

29.

Dodd

Ducroq

Neale

et al . The effect of serum matrix and gender on cortisol measurement by commonly used immunoassays. Ann Clin Biochem 2014; 51: 379–385.

30.

Kosicka

Siemiątkowska

Szpera-Goździewicz

et al.

Increased cortisol metabolism in women with pregnancy-related hypertension. Endocrine . Epub ahead of print 2018. DOI: 10.1007/s12020-018-1586-4.

31.

Draper

Stewart

PM.

11β-Hydroxysteroid dehydrogenase and the pre-receptor regulation of corticosteroid hormone action. J Endocrinol 2005; 186: 251–271.

32.

Palermo

Delitala

Mantero

et al . Congenital deficiency of 11β-hydroxysteroid dehydrogenase (apparent mineralocorticoid excess syndrome): Diagnostic value of urinary free cortisol and cortisone. J Endocrinol Invest 2001; 24: 17–23.

33.

Chapman

Holmes

Seckl

11β-hydroxysteroid dehydrogenases: intracellular gate-keepers of tissue glucocorticoid action. Physiol Rev 2013; 93: 1139–1206.

High-performance liquid chromatography methods for the analysis of endogenous cortisol and cortisone in human urine: comparison of mass spectrometry and fluorescence detection

Abstract

Background

Methods

Results

Conclusions

Keywords

Introduction

Materials and methods

Chemicals and reagents

Biological samples

HPLC-FLD

Equipment and conditions

Preparation of samples

HPLC-MS/MS

Equipment and conditions

Preparation of samples

Statistics

Results and discussion

Validation of methods

Comparison of methods

Summary

Footnotes

Declaration of conflicting interests

Funding

Ethical approval

Guarantor

Contributorship

References