Anal Bioanal Chem DOI 10.1007/s00216-013-7505-x
RESEARCH PAPER
Determination of free cortisol and free cortisone in human urine by on-line turbulent flow chromatography coupled to fused-core chromatography–tandem mass spectrometry (TFC–HPLC–MS/MS) Alberto Sánchez-Guijo & Michaela F. Hartmann & Lijie Shi & Thomas Remer & Stefan A. Wudy
Received: 13 September 2013 / Revised: 8 November 2013 / Accepted: 11 November 2013 # Springer-Verlag Berlin Heidelberg 2013
Abstract Urinary free cortisol and urinary free cortisone are decisive markers for the diagnosis of syndromes related to the dysfunction of the adrenal gland or to evaluate certain enzymatic disorders. Here, we present a new method, designed for routine laboratory use, which enables quick determination of these analytes with minor sample workup. Turbulent flow chromatography shortens sample preparation, and connection to a fused-core particle-packed column (rugged amideembedded C18 phase) permits a rapid and effective separation of the analytes, as well as additional separation from other related and isobaric compounds present in urine. Urinary isobaric compounds were successfully identified. The method requires only 100 μl of urine supernatant per sample. The total time between injections is 9.5 min. The solvents used for both turbulent and analytical chromatography are water and methanol, and the relatively low flows needed during the method resulted in an extended life of the columns. Linearity showed a R 2 >0.994. Limit of detection and limit of quantification are
0.5 and 1.0 ng/ml for cortisone and 1.0 and 2.0 ng/ml for cortisol. Recoveries ranged from 99.7 to 109.1 % for cortisone and from 98.7 to 102.9 % for cortisol. Accuracy values (relative errors) for intra- and inter-assay experiments were always below 8 %, whereas precision (percent CV) ranged from 3.7 to 10.7 %. No matrix effects were detected during the validation process. The reproducibility for each analyte’s retention time was excellent, with a coefficient of variation always below 0.2 %. The final validation step included the study of urine samples from healthy children and from children previously diagnosed with corticoidal disorders. The high selectivity achieved enables quick data handling. Keywords HPLC–MS/MS . Turbulent flow chromatography . Fused core technology . Urinary free cortisol . Urinary free cortisone
Introduction Electronic supplementary material The online version of this article (doi:10.1007/s00216-013-7505-x) contains supplementary material, which is available to authorized users. A. Sánchez-Guijo (*) : M. F. Hartmann : S. A. Wudy Steroid Research & Mass Spectrometry Unit, Division of Pediatric Endocrinology & Diabetology, Center of Child and Adolescent Medicine, Justus-Liebig-University, Feulgenstrasse 12, 35392 Giessen, Germany e-mail: [email protected] L. Shi : T. Remer DONALD Study Center at the Research Institute of Child Nutrition Dortmund, IEL-Nutritional Epidemiology, University of Bonn, Heinstück 11, 44225 Dortmund, Germany
Urinary free cortisone (UFE) and urinary free cortisol (UFF), the unconjugated forms of cortisone and cortisol in urine, are widely used as clinical markers for different steroidogenic illnesses, such as Cushing’s syndrome, apparent mineralocorticoid excess syndrome (AME), and cortisone reductase deficiency. Additionally, the ratio UFE/UFF has been used to evaluate stress or depression [1]. Stable isotope dilution liquid chromatography–tandem mass spectrometry (ID–LC–MS/MS) is an extraordinary selective technique by which the analysis of intact metabolites can be performed, avoiding chemical modifications and
A. Sánchez-Guijo et al.
shortening the analysis time when compared with GC–MS [2, 3]. Therefore, it has been progressively incorporated into the endocrinology laboratories [4–6]. Starting from the early 2000s, a number of LC–MS/MS methods have been developed for the specific determination of UFE and UFF [7–11]. Since then, LC–MS/MS has become an alternative to the classical methods for the detection of corticoids in clinical laboratories [12–15]. Turbulent flow technology is capable of completing sample workup in an automated fashion, reducing the total time required for sample preparation [16]. This technology applies turbulent flows to separate big molecules, mainly proteins, from the smaller ones within the matrix. Additional selectivity can be obtained by choosing among the different available column packings. Hence, small molecules are ideally more retained and they can be later eluted with a modification in solvent composition. The resolution of TurboflowTM columns is still very limited, and therefore, an analytical column is normally necessary in order to obtain a good separation of the analytes. In this work, we describe a fast, simple, and semiautomated method for high-throughput determination of UFF and UFE, in which sample handling is minimized. Turbulent flow chromatography was chosen to ease sample workup, and special care was taken to accomplish baseline separation of cortisol and cortisone by selecting the adequate analytical column. With this configuration, the analytes were effectively separated from other corticoid compounds sharing the same MS/MS transitions. This resulted in an improved and quicker data handling.
Experimental Reference steroids and chemicals Cortisol, aldosterone, corticosterone, 20α-dihydrocortisol, and cortisol internal standard, d4F (cortisol-9,11,12,12-2H4), were purchased from Sigma-Aldrich (Taufkirchen, Germany). Cortisone was obtained from Merck (Darmstadt, Germany), and its internal standard EIS (cortisone-1,2,4,19-13C4,1,1,19, 19,19-2H5) was synthesized by Dr. Yasuji Kasuya [17]. Cortisol internal standard, d3F (cortisol-9,12,12-2H4), was obtained from Cambridge Isotope Laboratories (Andover, USA). Prednisolone was purchased from Koch-Light Laboratories (Colnbrook, UK), and the remaining reference corticoids (20α-dihydrocortisone, 20β-dihydrocortisone, and 20βdihydrocortisol) were obtained from Steraloids, Inc. (Newport, RI, USA). Formic acid and LC–MS grade water were from Fluka (Taufkirchen, Germany), whereas analytical grade methanol was from Merck (Darmstadt, Germany).
Mass spectrometry Tandem mass spectrometry experiments were performed on a triple quadrupole mass spectrometer (TSQ, Quantum Ultra, Thermo Fisher Scientific, Dreieich, Germany). Atmospheric pressure chemical ionization (APCI) was chosen in order to minimize the matrix effects [18]. Furthermore, APCI enables the entrance of higher flows to the mass spectrometer from the HPLC system when compared to ESI. The vaporizer temperature was set to 500 °C, the capillary temperature to 280 °C, and the sheath gas and auxiliary gas to 40 and 20 arbitrary units. The discharge current was 6.0 μA, and the collision gas pressure was 1.5 mTorr. Cortisol and cortisone were detected using multiple reaction monitoring (MRM) in the positive-ion mode. Specific quantification and qualification transitions were obtained for each analyte. Transitions, with their collision energies and tube lens voltages for cortisol, cortisone, and internal standards d4F and EIS, can be found in Table 1. MS parameters were optimized by using standard solutions prepared in water/methanol (50 % v/v) with 0.1 % of formic acid, to a final concentration of 1,000 ng/ml. The solutions were then directly infused into the APCI source.
Turbulent liquid chromatography and high-performance liquid chromatography The solvents used for turbulent flow chromatography and analytical chromatography were pure water (A) and pure methanol (B). A C18 HTLC column (1.0×50 mm, 50 μm) was adapted to the Turboflow system for turbulent flow chromatography, with a 100-μl loop. The analytical column was Accucore Polar Premium (50×4.6 mm, 2.6 μm) that enabled fast liquid chromatography. Both columns were from Thermo Fisher Scientific, Dreieich, Germany. Flows and composition of solvents were optimized along the run to improve the connection between the turbulent flow Table 1 MRM transitions and mass spectrometer parameters
Cortisone quantifier Cortisone qualifier Cortisol quantifier Cortisol qualifier Cortisone internal standard Cortisol internal standard
Precursor ion (m/z)
Product ion (m/z)
Collision energy (eV)
Tube lens voltages (V)
361.2
163.1
15
85
361.2
121.1
18
85
363.2
121.1
35
95
363.2
91.0
34
109
370.2
172.0
25
80
367.1
121.0
34
81
Cortisol and cortisone determination in urine by TFC–HPLC–MS/MS
column and the analytical column in the HPLC system (Agilent 1200SL, Waldbronn, Germany). A tee-piece was used to connect the two columns. The analytes were eluted from the analytical column with a gradient from 65 % B to 95 % B and a flow of 0.75 ml/min during 90 s. Flows, gradients, and configurations are shown in Table 2. Sample preparation To remove any urinary sediment, 250 μl of each urine sample was first centrifuged (14,600 rpm, 3 min). Subsequently, 20 μl of internal standards (IS) was mixed with 100 μl of urine supernatant in a conical glass vial to a final concentration of 17 ng/ml for d4F and 15 ng/ml for EIS. Samples from patients with Cushing’s syndrome were diluted with water (mixture of 50 μl urine, 50 μl water, and 20 μl of IS) and prepared accordingly. Thereafter, the vials were vortexed for 30 s and finally shaked for 30 min to allow equilibration with the internal standards. Finally, 20 μl of the sample was injected into the Turboflow–HPLC system in triplicate. Method validation Most of the validation experiments were performed using urine from patients who underwent high-dose dexamethasone treatment, leading to complete adrenal suppression (urine from dexamethasone-treated patients, UDTP). The absence of endogenous corticoids was proven by urinary GC–MS steroid profiling [19]. In some experiments, stripped urine (charcoal-treated urine) was preferred. Linearity and calibration curves Linearity data were obtained using the program Xcalibur 2.1. The area ratio of each analyte to its internal standard was
plotted against the corresponding concentration in nanograms per milliliter. A 1/x weighting regression was chosen to favor higher accuracy and precision at the low concentration end of the curve. Calibration curves were obtained by spiking increasing concentrations of cortisol and cortisone standards into 100 μl of UDTP, plus a constant concentration of internal standards. The final volume of each spiked preparation was 120 μl, and the concentrations of the calibrators were 1, 2, 5, 10, 20, 50, 100, 200, and 250 ng/ml. The concentrations for the internal standards were 17 ng/ml for d4F and 15 ng/ml for EIS. Detection and quantification limits The limits of detection (LODs) and the limits of quantification (LOQs) for the analytes were determined at signal-to-noise (S/ N) ratio higher than 3 and higher than 10, respectively. Selection of concentration levels for method validation Three different quality controls were chosen to estimate the validation of the method: 10, 50, and 100 ng/ml. These were prepared by spiking the standards to UDTP or to stripped urine. Recovery To address the recovery of our method, 10, 50, and 100 ng/ml of cortisol and cortisone were spiked in stripped urine and injected directly onto the analytical column to omit the extraction process. The same preparations were injected into the Turboflow–HPLC system. The area ratios for the injections without Turboflow column were divided by the ones obtained with the on-line extraction system.
Table 2 Flows and solvent compositions for the analytical and turbulent flow columns Step
Minute
Turboflow™ column
Analytical column
Flow (ml/min)
Grad
%A
%B
Tee
Loop
Flow (ml/min)
Grad
%A
%B
1 2 3 4
0.00 0.17 0.67 1.00
1.00 1.50 1.00 0.20
Step Step Step Step
80.0 80.0 80.0 30.0
20.0 20.0 20.0 70.0
No No No Yes
Out Out Out In
0.40 0.40 0.40 0.40
Step Step Step Step
60 60 60 70
40.0 40.0 40.0 30.0
5 6 7 8 9 10 11
2.50 3.50 4.25 5.00 6.50 7.25 7.75
0.65 0.60 0.60 0.60 0.60 0.60 1.00
Step Step Step Step Step Step Step
80.0 30.0 20.0 55.0 0.0 0.0 80.0
20.0 70.0 80.0 45.0 100.0 100.0 20.0
No No No No No No No
In Out In In Out Out Out
0.45 0.50 0.65 0.75 0.60 0.50 0.40
Ramp Ramp Ramp Ramp Ramp Ramp Step
60 45 35 5 35 60 60
40.0 55.0 65.0 95.0 65.0 40.0 40.0
A. Sánchez-Guijo et al.
Matrix effects Evaluation of matrix effects was performed by plotting response ratios of spike experiments prepared in UDTP (Y-axis) against those from experiments with the same concentrations in water (X -axis), as described by Nguyen et al. [20]. A line slope of 1 is indicative of no matrix effects, whereas slopes above 1 correspond to an ion enhancement, and slopes below 1 are found when ion suppression is present. Carryover Carryover was evaluated by injecting standards spiked in urine (400 ng/ml) and by injection of samples from patients suspected with Cushing’s syndrome, followed by blank water samples. Intra-assay and inter-assay precision and accuracy Precision and accuracy were calculated by analysis of cortisol and cortisone at the three quality control samples during six occasions on the same day and on four consecutive days. Reproducibility of the retention times The reproducibility of the retention times was determined by calculating the coefficients of variation for the retention times of 45 runs (15 spiked experiments analyzed in triplicate). Analysis of patient samples The final step of the validation process consisted in the study of real samples previously analyzed by GC–MS using the method described in reference [19]. When available, 24-h urine was preferred as it overcomes corticoidal variations due to circadian rhythms. Interferences Potentially interfering compounds, such as aldosterone, prednisolone, corticosterone, 20α-dihydrocortisone, 20βdihydrocortisone, 20α-dihydrocortisol, 20β-dihydrocortisol, or 11-deoxycortisol, were spiked into UDTP to investigate coelutions with the analytes of interest.
Results
to a real sample from a healthy child with a concentration of 36.5 ng/ml of UFE and 18.3 ng/ml of UFF. In Fig. 1b, the merged chromatograms for cortisol and cortisone (quantifier transitions) are depicted. The group of compounds interfering with the UFF transitions appears at minute 5.30 in Fig. 1a, b. The flow eluting from the analytical column was only diverted from waste to the mass spectrometer for the analysis of the analytes during minutes 5.00–6.20, avoiding other urinary compounds to enter the MS. Addition of a small amount of acid (0.1–0.01 % of formic acid) to the sample or to the mobile phase did not increase the response in the mass spectrometer detector (data not shown). Potentially interfering compounds did not affect the analysis. The signal for aldosterone, traced with a standard solution for the transition 343.2→255/297 (collision energy 18 eV), appears at 5.37 min whereas corticosterone eluted at 5.96 min (347.2→121, collision energy 13 eV). Both peaks were baseline-separated from cortisol and cortisone when spiked in the same solution. Prednisolone coeluted with cortisol (5.72 min), but only affected the signal of cortisone. The volume of solvents needed for each analysis, including the Turboflow consumption, was 4.8 ml of water and 5 ml of methanol. Pressure in the analytical column never exceeded 175 bars during the run, and the total time between injections (including injection, extraction, and analytical separation) was about 9.5 min. Validation process LODs for cortisone and cortisol were first calculated with stripped urine for the spike experiments. The use of stripped urine permits to exclude interferences from other matrix constituents that may coelute with the analytes, as it contains no corticoids or any other chemically related compound. Therefore, a clear difference between instrumental noise and matrix components can be accomplished. Afterwards, these values were compared with the experiments spiked at the same levels in UDTP, to ensure that the peaks could be clearly distinguished from other isobaric compounds in urine. LOD was 0.5 ng/ml for cortisone and 1.0 ng/ ml for cortisol, and LOQ values were 1.0 and 2.0 ng/ml, respectively. The calculated coefficients of variation for the LOQs values were 19.4 % for cortisone and 19.1 % for cortisol (data obtained from five different spiked preparations measured in triplicate). The whole set of results for the validation study is shown in Table 3. Stripped urine was selected for recovery studies, instead of UDTP, to avoid the entrance of real urine into the system without previous cleanup.
Turboflow–HPLC–MS/MS analysis Analysis of real samples The method permitted baseline resolution of UFE and UFF, with retention times of 5.51 and 5.71 min, respectively. Figure 1a shows the different chromatograms corresponding
Thirty-nine 24-h urine samples from healthy children (DONALD study) were analyzed as a part of the validation
Cortisol and cortisone determination in urine by TFC–HPLC–MS/MS
a RT: 0.00 - 8.24
100
RT: 5.49
NL: 7.96E3 TIC F: + c APCI SRM ms2 [email protected] [171.999-172.001]
50 0 100
Cortisone IS RT: 5.51
NL: 4.57E3 TIC F: + c APCI SRM ms2 [email protected] [163.099-163.101]
50
Relative Abundance
0 100
Cortisone quantifier RT: 5.51
NL: 1.66E3 TIC F: + c APCI SRM ms2 [email protected] [121.099-121.101]
50 0 100
Cortisone qualifier
RT: 5.70
NL: 5.35E3 TIC F: + c APCI SRM ms2 [email protected] [120.999-121.001]
50
Cortisol IS 5.49
0 100
5.30
NL: 9.51E3 TIC F: + c APCI SRM ms2 [email protected] [121.099-121.101]
50 0 100
RT: 5.71
5.30
NL: 4.06E3 TIC F: + c APCI SRM ms2 [email protected] [90.999-91.001]
50
Cortisol quantifier
Cortisol qualifier RT: 5.71
0 0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
4.5
5.0
5.5
6.0
6.5
7.0
7.5
8.0
Time (min)
b RT: 3.00 - 8.00 5.29
100 NL: 4.78E3 TIC F: + c APCI SRM ms2 [email protected] [163.099-163.101]
95 90 85 80
NL: 1.03E4 TIC F: + c APCI SRM ms2 [email protected] [121.099-121.101]
75
Relative Abundance
70
5.52
Cortisone quantifier
Cortisol quantifier
5.71
65 60 55 50 45 40 35 30 25 20 15 10 5 0 3.0
3.5
4.0
4.5
5.0
5.5
6.0
6.5
7.0
7.5
8.0
Time (min)
Fig. 1 Chromatograms obtained from a healthy child real sample. a Chromatograms for all the transitions detailed in Table 1. b Merged chromatogram for F and E quantifiers, where baseline separation can be seen
A. Sánchez-Guijo et al. Table 3 Results for the validation process Cortisone
Intra-assay n =6 (spiked in UDTP) Inter-assay n =20 (spiked in UDTP)
% Recovery (in stripped urine)
Cortisol
ng/ml
Mean (concentration SD measured)
Precision (%CV)
Accuracy (%RE)
Mean (concentration SD measured)
Precision (%CV)
Accuracy (%RE)
10
10.8
0.4
3.7
7.7
10.4
0.6
5.7
3.9
50 100 10 50 100
51.6 101.3 10.2 49.5 96.6
3.9 6.0 1.1 3.8 7.1
7.6 5.9 10.7 7.6 7.3
3.1 1.3 1.9 −1.0 −3.5
50.3 95.7 10,7 50,6 98,6
2.5 7.1 1.0 2.7 7.1
5.1 7.5 8.9 5.4 7.3
0.6 −4.5 6.4 1.1 −1.4
10 50 100
99.7 109.1 108.0
102.9 98.7 99.7
Linearity, LOQs, LOD (ng/ml)
Linearity 1–250 Matrix effect study regression lines Y =0.993X +0.068 Reproducibility of retention times Mean (min) (n =45) 5.53
LOD LLOQ ULOQ 0.5 1.0 250.0 R 2 =0.997 (1–250 ng/ml) %CV 0.2
Linearity 2–200 Y =1.017X +0.158 Mean (min) 5.72
LOD LLOQ ULOQ 1.0 2.0 200.0 R 2 =0.997 (2–200 ng/ml) %CV 0.19
SD standard deviation, CV coeficient of variation, RE relative error
process. Two 24-h urine samples from patients with Cushing’s syndrome and two spot urine samples from children previously diagnosed with AME syndrome were studied as well. Daily excretion of UFF and UFE (in micrograms per day) obtained for healthy children and Cushing’s syndrome patients are shown in Table 4. Concentration of UFF and UFE (in nanograms per milliliter) in spot urine from two female siblings with AME syndrome was analyzed and compared with the concentrations from 24-h urines from healthy children (Table 5).
Discussion Method As mentioned before, the use of a C18 column with amideembedded groups made possible not only the effective separation of UFF and UFE but also additional separation from
other isobaric molecules sharing the same transitions with cortisol. To our knowledge, this is the first time this type of chromatography is used for the analysis of cortisol and cortisone. It has been previously shown how corticoids strongly interact through hydrogen bonds in solution [21]. Therefore, most probable is that an additional interaction takes place in the column through hydrogen bonding between the amide group of the stationary phase and hydroxyl groups of the corticoids. As a consequence, the overall result is an increase of their retention to the stationary phase when compared with traditional C18 reverse-phase chromatography. Thus, the elution of the analytes was only achieved when the proportion of methanol reached approximately an 80 %. The use of this chromatography opens new possibilities in the field of the analysis of corticoids. The use of a fused-core column reduces the time needed for analytical separation and seems a good matching for turbulent
Table 4 Values of daily excretion of UFF and UFE in micrograms per day for healthy control children and Cushing’s syndrome patients Healthy control children Female (n =20)
Age (years) UFF (μg/day) UFE (μg/day) Urine volume (ml/day)
Children with Cushing’s syndrome Male (n =19)
Child A (male)
Child B (male)
Mean (SD)
Range
Mean (SD)
Range
Mean (SD)
Mean (SD)
7.9 (0.7) 9.3 (3.1) 26.0 (5.9) 803 (266)
7.0–9.1 3.8–14.4 14.1–34.8 249–1194
7.7 (0.8) 11.8 (4.5) 26.8 (8.8) 645 (188)
6.9–9.4 5.3–20.5 9.0–42.2 296–983
8.7 279.8 (6.9) 270.1 (14.3) 1,350
15.9 722.2 (45.4) 335.7 (18.0) 1,300
Cortisol and cortisone determination in urine by TFC–HPLC–MS/MS Table 5 Values of concentrations in nanograms per milliliter for healthy control children and AME syndrome patients
RESEARCH PAPER
Determination of free cortisol and free cortisone in human urine by on-line turbulent flow chromatography coupled to fused-core chromatography–tandem mass spectrometry (TFC–HPLC–MS/MS) Alberto Sánchez-Guijo & Michaela F. Hartmann & Lijie Shi & Thomas Remer & Stefan A. Wudy
Received: 13 September 2013 / Revised: 8 November 2013 / Accepted: 11 November 2013 # Springer-Verlag Berlin Heidelberg 2013
Abstract Urinary free cortisol and urinary free cortisone are decisive markers for the diagnosis of syndromes related to the dysfunction of the adrenal gland or to evaluate certain enzymatic disorders. Here, we present a new method, designed for routine laboratory use, which enables quick determination of these analytes with minor sample workup. Turbulent flow chromatography shortens sample preparation, and connection to a fused-core particle-packed column (rugged amideembedded C18 phase) permits a rapid and effective separation of the analytes, as well as additional separation from other related and isobaric compounds present in urine. Urinary isobaric compounds were successfully identified. The method requires only 100 μl of urine supernatant per sample. The total time between injections is 9.5 min. The solvents used for both turbulent and analytical chromatography are water and methanol, and the relatively low flows needed during the method resulted in an extended life of the columns. Linearity showed a R 2 >0.994. Limit of detection and limit of quantification are
0.5 and 1.0 ng/ml for cortisone and 1.0 and 2.0 ng/ml for cortisol. Recoveries ranged from 99.7 to 109.1 % for cortisone and from 98.7 to 102.9 % for cortisol. Accuracy values (relative errors) for intra- and inter-assay experiments were always below 8 %, whereas precision (percent CV) ranged from 3.7 to 10.7 %. No matrix effects were detected during the validation process. The reproducibility for each analyte’s retention time was excellent, with a coefficient of variation always below 0.2 %. The final validation step included the study of urine samples from healthy children and from children previously diagnosed with corticoidal disorders. The high selectivity achieved enables quick data handling. Keywords HPLC–MS/MS . Turbulent flow chromatography . Fused core technology . Urinary free cortisol . Urinary free cortisone
Introduction Electronic supplementary material The online version of this article (doi:10.1007/s00216-013-7505-x) contains supplementary material, which is available to authorized users. A. Sánchez-Guijo (*) : M. F. Hartmann : S. A. Wudy Steroid Research & Mass Spectrometry Unit, Division of Pediatric Endocrinology & Diabetology, Center of Child and Adolescent Medicine, Justus-Liebig-University, Feulgenstrasse 12, 35392 Giessen, Germany e-mail: [email protected] L. Shi : T. Remer DONALD Study Center at the Research Institute of Child Nutrition Dortmund, IEL-Nutritional Epidemiology, University of Bonn, Heinstück 11, 44225 Dortmund, Germany
Urinary free cortisone (UFE) and urinary free cortisol (UFF), the unconjugated forms of cortisone and cortisol in urine, are widely used as clinical markers for different steroidogenic illnesses, such as Cushing’s syndrome, apparent mineralocorticoid excess syndrome (AME), and cortisone reductase deficiency. Additionally, the ratio UFE/UFF has been used to evaluate stress or depression [1]. Stable isotope dilution liquid chromatography–tandem mass spectrometry (ID–LC–MS/MS) is an extraordinary selective technique by which the analysis of intact metabolites can be performed, avoiding chemical modifications and
A. Sánchez-Guijo et al.
shortening the analysis time when compared with GC–MS [2, 3]. Therefore, it has been progressively incorporated into the endocrinology laboratories [4–6]. Starting from the early 2000s, a number of LC–MS/MS methods have been developed for the specific determination of UFE and UFF [7–11]. Since then, LC–MS/MS has become an alternative to the classical methods for the detection of corticoids in clinical laboratories [12–15]. Turbulent flow technology is capable of completing sample workup in an automated fashion, reducing the total time required for sample preparation [16]. This technology applies turbulent flows to separate big molecules, mainly proteins, from the smaller ones within the matrix. Additional selectivity can be obtained by choosing among the different available column packings. Hence, small molecules are ideally more retained and they can be later eluted with a modification in solvent composition. The resolution of TurboflowTM columns is still very limited, and therefore, an analytical column is normally necessary in order to obtain a good separation of the analytes. In this work, we describe a fast, simple, and semiautomated method for high-throughput determination of UFF and UFE, in which sample handling is minimized. Turbulent flow chromatography was chosen to ease sample workup, and special care was taken to accomplish baseline separation of cortisol and cortisone by selecting the adequate analytical column. With this configuration, the analytes were effectively separated from other corticoid compounds sharing the same MS/MS transitions. This resulted in an improved and quicker data handling.
Experimental Reference steroids and chemicals Cortisol, aldosterone, corticosterone, 20α-dihydrocortisol, and cortisol internal standard, d4F (cortisol-9,11,12,12-2H4), were purchased from Sigma-Aldrich (Taufkirchen, Germany). Cortisone was obtained from Merck (Darmstadt, Germany), and its internal standard EIS (cortisone-1,2,4,19-13C4,1,1,19, 19,19-2H5) was synthesized by Dr. Yasuji Kasuya [17]. Cortisol internal standard, d3F (cortisol-9,12,12-2H4), was obtained from Cambridge Isotope Laboratories (Andover, USA). Prednisolone was purchased from Koch-Light Laboratories (Colnbrook, UK), and the remaining reference corticoids (20α-dihydrocortisone, 20β-dihydrocortisone, and 20βdihydrocortisol) were obtained from Steraloids, Inc. (Newport, RI, USA). Formic acid and LC–MS grade water were from Fluka (Taufkirchen, Germany), whereas analytical grade methanol was from Merck (Darmstadt, Germany).
Mass spectrometry Tandem mass spectrometry experiments were performed on a triple quadrupole mass spectrometer (TSQ, Quantum Ultra, Thermo Fisher Scientific, Dreieich, Germany). Atmospheric pressure chemical ionization (APCI) was chosen in order to minimize the matrix effects [18]. Furthermore, APCI enables the entrance of higher flows to the mass spectrometer from the HPLC system when compared to ESI. The vaporizer temperature was set to 500 °C, the capillary temperature to 280 °C, and the sheath gas and auxiliary gas to 40 and 20 arbitrary units. The discharge current was 6.0 μA, and the collision gas pressure was 1.5 mTorr. Cortisol and cortisone were detected using multiple reaction monitoring (MRM) in the positive-ion mode. Specific quantification and qualification transitions were obtained for each analyte. Transitions, with their collision energies and tube lens voltages for cortisol, cortisone, and internal standards d4F and EIS, can be found in Table 1. MS parameters were optimized by using standard solutions prepared in water/methanol (50 % v/v) with 0.1 % of formic acid, to a final concentration of 1,000 ng/ml. The solutions were then directly infused into the APCI source.
Turbulent liquid chromatography and high-performance liquid chromatography The solvents used for turbulent flow chromatography and analytical chromatography were pure water (A) and pure methanol (B). A C18 HTLC column (1.0×50 mm, 50 μm) was adapted to the Turboflow system for turbulent flow chromatography, with a 100-μl loop. The analytical column was Accucore Polar Premium (50×4.6 mm, 2.6 μm) that enabled fast liquid chromatography. Both columns were from Thermo Fisher Scientific, Dreieich, Germany. Flows and composition of solvents were optimized along the run to improve the connection between the turbulent flow Table 1 MRM transitions and mass spectrometer parameters
Cortisone quantifier Cortisone qualifier Cortisol quantifier Cortisol qualifier Cortisone internal standard Cortisol internal standard
Precursor ion (m/z)
Product ion (m/z)
Collision energy (eV)
Tube lens voltages (V)
361.2
163.1
15
85
361.2
121.1
18
85
363.2
121.1
35
95
363.2
91.0
34
109
370.2
172.0
25
80
367.1
121.0
34
81
Cortisol and cortisone determination in urine by TFC–HPLC–MS/MS
column and the analytical column in the HPLC system (Agilent 1200SL, Waldbronn, Germany). A tee-piece was used to connect the two columns. The analytes were eluted from the analytical column with a gradient from 65 % B to 95 % B and a flow of 0.75 ml/min during 90 s. Flows, gradients, and configurations are shown in Table 2. Sample preparation To remove any urinary sediment, 250 μl of each urine sample was first centrifuged (14,600 rpm, 3 min). Subsequently, 20 μl of internal standards (IS) was mixed with 100 μl of urine supernatant in a conical glass vial to a final concentration of 17 ng/ml for d4F and 15 ng/ml for EIS. Samples from patients with Cushing’s syndrome were diluted with water (mixture of 50 μl urine, 50 μl water, and 20 μl of IS) and prepared accordingly. Thereafter, the vials were vortexed for 30 s and finally shaked for 30 min to allow equilibration with the internal standards. Finally, 20 μl of the sample was injected into the Turboflow–HPLC system in triplicate. Method validation Most of the validation experiments were performed using urine from patients who underwent high-dose dexamethasone treatment, leading to complete adrenal suppression (urine from dexamethasone-treated patients, UDTP). The absence of endogenous corticoids was proven by urinary GC–MS steroid profiling [19]. In some experiments, stripped urine (charcoal-treated urine) was preferred. Linearity and calibration curves Linearity data were obtained using the program Xcalibur 2.1. The area ratio of each analyte to its internal standard was
plotted against the corresponding concentration in nanograms per milliliter. A 1/x weighting regression was chosen to favor higher accuracy and precision at the low concentration end of the curve. Calibration curves were obtained by spiking increasing concentrations of cortisol and cortisone standards into 100 μl of UDTP, plus a constant concentration of internal standards. The final volume of each spiked preparation was 120 μl, and the concentrations of the calibrators were 1, 2, 5, 10, 20, 50, 100, 200, and 250 ng/ml. The concentrations for the internal standards were 17 ng/ml for d4F and 15 ng/ml for EIS. Detection and quantification limits The limits of detection (LODs) and the limits of quantification (LOQs) for the analytes were determined at signal-to-noise (S/ N) ratio higher than 3 and higher than 10, respectively. Selection of concentration levels for method validation Three different quality controls were chosen to estimate the validation of the method: 10, 50, and 100 ng/ml. These were prepared by spiking the standards to UDTP or to stripped urine. Recovery To address the recovery of our method, 10, 50, and 100 ng/ml of cortisol and cortisone were spiked in stripped urine and injected directly onto the analytical column to omit the extraction process. The same preparations were injected into the Turboflow–HPLC system. The area ratios for the injections without Turboflow column were divided by the ones obtained with the on-line extraction system.
Table 2 Flows and solvent compositions for the analytical and turbulent flow columns Step
Minute
Turboflow™ column
Analytical column
Flow (ml/min)
Grad
%A
%B
Tee
Loop
Flow (ml/min)
Grad
%A
%B
1 2 3 4
0.00 0.17 0.67 1.00
1.00 1.50 1.00 0.20
Step Step Step Step
80.0 80.0 80.0 30.0
20.0 20.0 20.0 70.0
No No No Yes
Out Out Out In
0.40 0.40 0.40 0.40
Step Step Step Step
60 60 60 70
40.0 40.0 40.0 30.0
5 6 7 8 9 10 11
2.50 3.50 4.25 5.00 6.50 7.25 7.75
0.65 0.60 0.60 0.60 0.60 0.60 1.00
Step Step Step Step Step Step Step
80.0 30.0 20.0 55.0 0.0 0.0 80.0
20.0 70.0 80.0 45.0 100.0 100.0 20.0
No No No No No No No
In Out In In Out Out Out
0.45 0.50 0.65 0.75 0.60 0.50 0.40
Ramp Ramp Ramp Ramp Ramp Ramp Step
60 45 35 5 35 60 60
40.0 55.0 65.0 95.0 65.0 40.0 40.0
A. Sánchez-Guijo et al.
Matrix effects Evaluation of matrix effects was performed by plotting response ratios of spike experiments prepared in UDTP (Y-axis) against those from experiments with the same concentrations in water (X -axis), as described by Nguyen et al. [20]. A line slope of 1 is indicative of no matrix effects, whereas slopes above 1 correspond to an ion enhancement, and slopes below 1 are found when ion suppression is present. Carryover Carryover was evaluated by injecting standards spiked in urine (400 ng/ml) and by injection of samples from patients suspected with Cushing’s syndrome, followed by blank water samples. Intra-assay and inter-assay precision and accuracy Precision and accuracy were calculated by analysis of cortisol and cortisone at the three quality control samples during six occasions on the same day and on four consecutive days. Reproducibility of the retention times The reproducibility of the retention times was determined by calculating the coefficients of variation for the retention times of 45 runs (15 spiked experiments analyzed in triplicate). Analysis of patient samples The final step of the validation process consisted in the study of real samples previously analyzed by GC–MS using the method described in reference [19]. When available, 24-h urine was preferred as it overcomes corticoidal variations due to circadian rhythms. Interferences Potentially interfering compounds, such as aldosterone, prednisolone, corticosterone, 20α-dihydrocortisone, 20βdihydrocortisone, 20α-dihydrocortisol, 20β-dihydrocortisol, or 11-deoxycortisol, were spiked into UDTP to investigate coelutions with the analytes of interest.
Results
to a real sample from a healthy child with a concentration of 36.5 ng/ml of UFE and 18.3 ng/ml of UFF. In Fig. 1b, the merged chromatograms for cortisol and cortisone (quantifier transitions) are depicted. The group of compounds interfering with the UFF transitions appears at minute 5.30 in Fig. 1a, b. The flow eluting from the analytical column was only diverted from waste to the mass spectrometer for the analysis of the analytes during minutes 5.00–6.20, avoiding other urinary compounds to enter the MS. Addition of a small amount of acid (0.1–0.01 % of formic acid) to the sample or to the mobile phase did not increase the response in the mass spectrometer detector (data not shown). Potentially interfering compounds did not affect the analysis. The signal for aldosterone, traced with a standard solution for the transition 343.2→255/297 (collision energy 18 eV), appears at 5.37 min whereas corticosterone eluted at 5.96 min (347.2→121, collision energy 13 eV). Both peaks were baseline-separated from cortisol and cortisone when spiked in the same solution. Prednisolone coeluted with cortisol (5.72 min), but only affected the signal of cortisone. The volume of solvents needed for each analysis, including the Turboflow consumption, was 4.8 ml of water and 5 ml of methanol. Pressure in the analytical column never exceeded 175 bars during the run, and the total time between injections (including injection, extraction, and analytical separation) was about 9.5 min. Validation process LODs for cortisone and cortisol were first calculated with stripped urine for the spike experiments. The use of stripped urine permits to exclude interferences from other matrix constituents that may coelute with the analytes, as it contains no corticoids or any other chemically related compound. Therefore, a clear difference between instrumental noise and matrix components can be accomplished. Afterwards, these values were compared with the experiments spiked at the same levels in UDTP, to ensure that the peaks could be clearly distinguished from other isobaric compounds in urine. LOD was 0.5 ng/ml for cortisone and 1.0 ng/ ml for cortisol, and LOQ values were 1.0 and 2.0 ng/ml, respectively. The calculated coefficients of variation for the LOQs values were 19.4 % for cortisone and 19.1 % for cortisol (data obtained from five different spiked preparations measured in triplicate). The whole set of results for the validation study is shown in Table 3. Stripped urine was selected for recovery studies, instead of UDTP, to avoid the entrance of real urine into the system without previous cleanup.
Turboflow–HPLC–MS/MS analysis Analysis of real samples The method permitted baseline resolution of UFE and UFF, with retention times of 5.51 and 5.71 min, respectively. Figure 1a shows the different chromatograms corresponding
Thirty-nine 24-h urine samples from healthy children (DONALD study) were analyzed as a part of the validation
Cortisol and cortisone determination in urine by TFC–HPLC–MS/MS
a RT: 0.00 - 8.24
100
RT: 5.49
NL: 7.96E3 TIC F: + c APCI SRM ms2 [email protected] [171.999-172.001]
50 0 100
Cortisone IS RT: 5.51
NL: 4.57E3 TIC F: + c APCI SRM ms2 [email protected] [163.099-163.101]
50
Relative Abundance
0 100
Cortisone quantifier RT: 5.51
NL: 1.66E3 TIC F: + c APCI SRM ms2 [email protected] [121.099-121.101]
50 0 100
Cortisone qualifier
RT: 5.70
NL: 5.35E3 TIC F: + c APCI SRM ms2 [email protected] [120.999-121.001]
50
Cortisol IS 5.49
0 100
5.30
NL: 9.51E3 TIC F: + c APCI SRM ms2 [email protected] [121.099-121.101]
50 0 100
RT: 5.71
5.30
NL: 4.06E3 TIC F: + c APCI SRM ms2 [email protected] [90.999-91.001]
50
Cortisol quantifier
Cortisol qualifier RT: 5.71
0 0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
4.5
5.0
5.5
6.0
6.5
7.0
7.5
8.0
Time (min)
b RT: 3.00 - 8.00 5.29
100 NL: 4.78E3 TIC F: + c APCI SRM ms2 [email protected] [163.099-163.101]
95 90 85 80
NL: 1.03E4 TIC F: + c APCI SRM ms2 [email protected] [121.099-121.101]
75
Relative Abundance
70
5.52
Cortisone quantifier
Cortisol quantifier
5.71
65 60 55 50 45 40 35 30 25 20 15 10 5 0 3.0
3.5
4.0
4.5
5.0
5.5
6.0
6.5
7.0
7.5
8.0
Time (min)
Fig. 1 Chromatograms obtained from a healthy child real sample. a Chromatograms for all the transitions detailed in Table 1. b Merged chromatogram for F and E quantifiers, where baseline separation can be seen
A. Sánchez-Guijo et al. Table 3 Results for the validation process Cortisone
Intra-assay n =6 (spiked in UDTP) Inter-assay n =20 (spiked in UDTP)
% Recovery (in stripped urine)
Cortisol
ng/ml
Mean (concentration SD measured)
Precision (%CV)
Accuracy (%RE)
Mean (concentration SD measured)
Precision (%CV)
Accuracy (%RE)
10
10.8
0.4
3.7
7.7
10.4
0.6
5.7
3.9
50 100 10 50 100
51.6 101.3 10.2 49.5 96.6
3.9 6.0 1.1 3.8 7.1
7.6 5.9 10.7 7.6 7.3
3.1 1.3 1.9 −1.0 −3.5
50.3 95.7 10,7 50,6 98,6
2.5 7.1 1.0 2.7 7.1
5.1 7.5 8.9 5.4 7.3
0.6 −4.5 6.4 1.1 −1.4
10 50 100
99.7 109.1 108.0
102.9 98.7 99.7
Linearity, LOQs, LOD (ng/ml)
Linearity 1–250 Matrix effect study regression lines Y =0.993X +0.068 Reproducibility of retention times Mean (min) (n =45) 5.53
LOD LLOQ ULOQ 0.5 1.0 250.0 R 2 =0.997 (1–250 ng/ml) %CV 0.2
Linearity 2–200 Y =1.017X +0.158 Mean (min) 5.72
LOD LLOQ ULOQ 1.0 2.0 200.0 R 2 =0.997 (2–200 ng/ml) %CV 0.19
SD standard deviation, CV coeficient of variation, RE relative error
process. Two 24-h urine samples from patients with Cushing’s syndrome and two spot urine samples from children previously diagnosed with AME syndrome were studied as well. Daily excretion of UFF and UFE (in micrograms per day) obtained for healthy children and Cushing’s syndrome patients are shown in Table 4. Concentration of UFF and UFE (in nanograms per milliliter) in spot urine from two female siblings with AME syndrome was analyzed and compared with the concentrations from 24-h urines from healthy children (Table 5).
Discussion Method As mentioned before, the use of a C18 column with amideembedded groups made possible not only the effective separation of UFF and UFE but also additional separation from
other isobaric molecules sharing the same transitions with cortisol. To our knowledge, this is the first time this type of chromatography is used for the analysis of cortisol and cortisone. It has been previously shown how corticoids strongly interact through hydrogen bonds in solution [21]. Therefore, most probable is that an additional interaction takes place in the column through hydrogen bonding between the amide group of the stationary phase and hydroxyl groups of the corticoids. As a consequence, the overall result is an increase of their retention to the stationary phase when compared with traditional C18 reverse-phase chromatography. Thus, the elution of the analytes was only achieved when the proportion of methanol reached approximately an 80 %. The use of this chromatography opens new possibilities in the field of the analysis of corticoids. The use of a fused-core column reduces the time needed for analytical separation and seems a good matching for turbulent
Table 4 Values of daily excretion of UFF and UFE in micrograms per day for healthy control children and Cushing’s syndrome patients Healthy control children Female (n =20)
Age (years) UFF (μg/day) UFE (μg/day) Urine volume (ml/day)
Children with Cushing’s syndrome Male (n =19)
Child A (male)
Child B (male)
Mean (SD)
Range
Mean (SD)
Range
Mean (SD)
Mean (SD)
7.9 (0.7) 9.3 (3.1) 26.0 (5.9) 803 (266)
7.0–9.1 3.8–14.4 14.1–34.8 249–1194
7.7 (0.8) 11.8 (4.5) 26.8 (8.8) 645 (188)
6.9–9.4 5.3–20.5 9.0–42.2 296–983
8.7 279.8 (6.9) 270.1 (14.3) 1,350
15.9 722.2 (45.4) 335.7 (18.0) 1,300
Cortisol and cortisone determination in urine by TFC–HPLC–MS/MS Table 5 Values of concentrations in nanograms per milliliter for healthy control children and AME syndrome patients
