A practical ID-LC-MS/MS method for the most commonly analyzed steroid hormones in clinical laboratories

Introduction

Steroids; which are derived from cholesterol and mostly synthesized in adrenal cortex, placenta and gonads; are biological molecules functioning in regulation of many metabolic activities in organism [1]. Many endocrinological pathologies emerge because of the defects in steroid biosynthesis resulting in excess or insufficient secretion of steroids [2]. Analysis of steroid hormones in clinical laboratories plays a decisive role in diagnosis and monitoring of these diseases.

Immunoassays are the most commonly used methods for steroid analysis in clinical laboratories. In the past, radio immunoassay (RIA) methods were commonly used, but today automated instruments employing chemiluminescence immunoassay (CLIA) are widely used [3]. The major drawback of this method is the lack of specificity. The antibodies used in this methods may cross-react with other steroids or other similar molecules, and these intereferences result in serious problems [4]. Additionally, some of the commercially available kits and methods have insufficient test performance, considering the precision and trueness. Another method for steroid analysis is GC-MS which is considered as the gold standard [5]. However, this method can not be used in routine clinical laboratories because of laborious sample pretreatment steps and long analysis times. Today, liquid chromatography tandem mass spectrometry (LC-MS/MS) use is getting wider in steroid analysis [6].

The aim of this study was to create a steroid profiling panel which was composed of the most commonly analyzed steroid hormones in clinical laboratories, and had a good test performance considering the precision, trueness, analytical sensitivity and specificity.

Materials and methods

An isotope dilution (ID) LC-MS/MS method was developed, in which 13 steroids (aldosterone, corticosterone, cortisol, cortisone, 11-deoxycortisol, androstenedione, DHEA: dehydroepiandrosterone, DHEAS: dehydroepiandrosterone sulfate, DHT: dihydrotestosterone, E2: estradiol, 17α-OH progesterone, progesterone, testosterone) can be analyzed in the same run. Validated calibrators (6 point calibrators plus blank) for MS and internal standard (IS) mix (Chromsystems, Gräfelfing, Germany) were used in this study. These calibrators are serum-based standards which means they include protein to avoid matrix effect. The content of IS mix (Chromsystems, Gräfelfing, Germany) was aldosterone-d4, corticosterone-d8, cortisol-d4, cortisone-d8, 11-deoxycortisol-d5, androstenedione-d7, DHEA-d5, DHEAS-d6, DHT-d3, E2-d5, 17α-OH progesterone-d8, progesterone-d9, testosterone-d3. In this method; protein precipitation was performed for sample preparation, including the calibrators. The precipitant solution was prepared by mixing 0.3 M zinc sulfate solution (ZnSO₄), methanol and IS. Composition of the precipitant solution was IS:ZnSO₄:Methanol 15:20:65% (v:v:v). One hundred microliters sample and 200 μL precipitant solution were mixed and centrifuged. Supernatant part was transferred to plates and analyzed.

Ultra performance liquid chromatography system (Thermo Scientific Dionex UltiMate 3000 Rapid Separation LC System UHPLC+Focused, San Jose, CA, USA) and an analytical column with C18 selectivity (Thermo Scientific, Hypersil GOLD UHPLC Column, 50×2.1 mm, particle size 1.9 μm) was chosen for chromatographic seperation. Mobile phase A was distilled water (100%) and mobile phase B was methanol (100%). Atmospheric pressure chemical ionization (APCI) ion source was preferred for ionization, and tandem MS with triple quadrupole (QQQ) (Thermo Scientific TSQ Quantum Access MAX, San Jose, CA, USA) was used. MS scan was performed using the SRM (selected reaction monitoring) mode in positive polarity. LC-MS/MS conditions were shown in Table 1.

Method validation procedures were planned according to CLSI guidelines; and precision, trueness, limits of detection and quantitation, recovery and linearity were assessed. Besides, 40 patient samples were analyzed for steroid hormones, using the available immunoassay methods in Gulhane Training and Research Hospital, Medical Biochemistry Laboratories for method comparison study. Patients using exogenous steroid were excluded from the study. This study was approved by local ethics committee (REC number: 08/225), complied with the Declaration of Helsinki and all the subjects were given informed consent. The collected data from ID-LC-MS/MS method was compared with the data from RIA method for aldosterone (Beckman Coulter Immunotech Prague, Czech Republic) and 17α-Hydroxy progesterone (DIAsource, Louvain-la-Neuve, Belgium); CLIA method (Siemens Advia Centaur XP, Tarrytown, NY, USA) for cortisol, DHEAS, estradiol, progesterone, total testosterone; and CLEIA method (Siemens Immulite 2000 XPi, Llanberis, Gwynedd, UK) for androstenedione. MedCalc 9.2.0.1 (Ostend, Belgium) was used for statistical analysis of method comparison data.

Results

A practical LC-MS/MS method was developed for the commonly analyzed 13 steroid hormones by using stable isotope dilution for each steroid hormone (Figure 1). Ionization pattern, retention time (RT), collision energy (CE), precursor and product ions of each steroid hormone were summarized in Table 2. The first steroid eluting from the column was aldosterone (1.11 s) while the last one was corticosterone (5.94 s).

Figure 1:

SRM scan of all steroids and their stable isotope standards.

(A) Chromatograms of the first 6 steroid hormones and their stable isotope dilutions eluting from the system (^aE3 was excluded from validation process because its calibrator was not approved for MS/MS analysis.). (B) Chromatograms of the last 7 steroid hormones and their stable isotope dilutions eluting from the system.

Precision of this new method was evaluated using Biorad Lyphocheck three level control materials. Testosterone and DHEAS had the best precision performance (Table 3). Recovery studies were performed by spiking serum samples. All tests were performed duplicated. Trueness performance was the best for DHEAS and androstenedione considering the recovery data. Limit of detection (LOD) (S/N ratio is at least 3:1) and lower limit of quantitation (LLOQ) (S/N ratio is at least 10:1) were determined by analysis of the appropriate dilutions of level 1 calibrator. Accepted imprecision value was <20% for the experiments performed, during the determination and validation of LLOQ. Analytical sensitivity was worse in this method for aldosterone (20 pg/mL) and E2 (31 pg/mL) with respect to immunoassay based methods (7.64 pg/mL and 11.8 pg/mL respectively) in our core laboratory (Table 4). Linearity was assessed either.

Performance of immunoassay based methods (testosterone, DHEAS, E2, progesterone, testosterone, androstenedione, 17α-hydroxy progesterone, aldosterone) in our core laboratory were evaluated according to EURACHEM CITAC Guide [7] using the internal quality control and external quality assessment data. E2 and 17α-hydoxy progesterone performance were quite good considering the desirable specifications according to Fraser (Table 5) [8]. The results for analysis of aldosterone and estradiol with ID-LC-MS/MS method were suboptimal, considering the RIA and CLIA methods, respectively. Performance of ID-LC-MS/MS method was found to be superior when compared with CLIA method for cortisol, DHEAS, total testosterone, progesterone; chemiluminescence enzyme immunoassay (CLEIA) method for androstenedione and RIA method for 17α-hydroxy progesterone considering the imprecision and trueness performance. Method validation data showed that the test performance of 11-deoxycortisol, cortisone, corticosterone, DHEA and dihydrotestosterone were pretty good.

Forty serum samples were analyzed with both immunoassay and ID-LC-MS/MS methods. Method comparison results revealed that aldosterone results of the two methods showed the worst correlation with a correlation coefficient of r=0.884 (0.791–0.937, 95% confidence interval). DHEAS results of the two methods had the best correlation (r=0.966, 0.937–0.982, 95% confidence interval) (Figure 2). According to Bland Altmann plots of the two methods, testosterone and aldosterone results of the ID-LC-MS/MS method was higher than the immunoassay methods while the others were lower in ID-LC-MS/MS method (Figure 3).

Figure 2:

Correlation of ID-LC-MS/MS and immunoassay methods.

(A) Deming Regression of ID-LC-MS/MS method with CLIA method. Regression line is below “y=x line” for CLIA method which means average results of CLIA method is higher than the ID-LC-MS/MS method except testosterone. The two methods seem to be compatible with each other considering the slope, intercept and correlation coefficients. [Dashed line is y=x line. The other line is regression line. Slope intercept and correlation coefficient (r) were given with 95% confidence interval. Progesteron results were evaluated as <1 ng/mL and >1 ng/mL to increase the resolution of regression line in very high and very low levels since the progesterone levels of the samples varies in a huge range between 0.3 and 40 ng/mL]. (B) Deming Regression of ID-LC-MS/MS method with CLEIA method. Regression line is below “y=x line” for CLEIA method which means average results of CLEIA method is higher than the ID-LC-MS/MS method. However, the two methods seem to be correlated with each other. (C) Deming Regression of ID-LC-MS/MS method with RIA method. Regression line is below “y=x line” for RIA method of 17-OH Progesterone which means average results of RIA method for 17-OH Progesterone is higher than the ID-LC-MS/MS method. Contrarily, regression line is above “y=x line” for RIA method of aldosterone. However, the two methods seem to be correlated with each other.

Figure 3:

Percent differences against mean of the compared methods.

(A) Bland Altman plots of ID-LC-MS/MS method with CLIA method. Mean lines in the graphs are negative except testosterone which means ID-LC-MS/MS method gives lower results than CLIA method except testosterone. Bias seems to be acceptable since most of the samples are between ±1.96 standard deviation (SD). (B) Bland Altman plots of ID-LC-MS/MS method with CLEIA method. Mean line in the graph is negative which means ID-LC-MS/MS method gives lower results than CLEIA method for androstenedione. Bias seems to be acceptable since most of the samples are between ±1.96 standard deviation (SD). (C) Bland Altman plots of ID-LC-MS/MS method with RIA method. Mean line in the graph of 17-OH Progesterone is negative which means average results of ID-LC-MS/MS method for 17-OH Progesterone is lower than the RIA method for 17-OH Progesterone. Contrarily, mean line is positive for aldosterone.

Discussion

An ID-LC-APCI-MS/MS method; having an 11 min of analysis time, with 13 steroids, two of suboptimal; has been validated. A serum sample volume of only 100 μL is required in this method. Protein precipitation was employed for pretreatment steps which made this method simple and practical. SRM mode of the tandem MS provided us high analytical sensitivity and specificity. An IS isotope was used for each steroid so the matrix effect was minimized.

There are some studies which indicate the superiority of APPI to APCI and ESI in steroid analysis. Since APPI is a soft ionization source, it creates just the ions of concern such as steroids. Thus, the suppression due to other excess ionization can be decreased resulting in a better signal-to-noise ratio [9]. Precision of APPI was reported to be better than APCI, considering the peak area/height of IS [10]. In another study, APCI was reported to be the best choice of ion source for steroid analysis except aldosterone and estradiol (APPI was suggested for these steroids) [11]. Since we could not have access to APPI ion source in our laboratory, we used APCI ion source. Many studies recommend MS scan in positive polarity except aldosterone and estradiol which are commonly scanned in negative polarity [11]. As polarity switching was not possible in our instrument, we scanned all steroid hormones in positive polarity. These factors might be the reason for aldosterone and estradiol having suboptimal performance in our method.

This method was found to correlate well with the available validated immunoassay based methods routinely used in our core laboratory. Likely with the previous studies, ID-LC-MS/MS method gave lower results with respect to the CLIA method for cortisol [12], [13], [14], androstenedione, progesterone [14], and 17α-OH progesteron [15]. Contrarily, testosterone results were higher in ID-LC-MS/MS, which was likely with the study of Fanelli et al. [14]. Aldosterone and estradiol performance of immunoassays were better while the other steroids had superior performance in ID-LC-MS/MS method.

The most important challenge in immunoassays is lack of specificity and standardization [16], [17]. Manufacturers use different antibodies for immunoassays and the commonly used calibrators are not traceable to reference materials. This causes incompatible results. This is true for LC-MS/MS studies as well. So, we used the validated calibrators for MS in this study.

Both immunoassays and tandem MS methods have advantages and disadvantages. Automatized immunoassays allow only one steroid analysis for each run. It is simple as it does not require pretreatment steps, but much more sample volume is needed. On the other hand, LC-MS/MS methods allow many steroid analysis on the same run and require less sample volume [18]. Analytical specificity and sensitivity, and test performance of LC-MS/MS method is generally better than immunoassays [16], [17]. Additionally, LC-MS/MS methods are less prone to be affected from interferences with regard to immunoassays. However, LC-MS/MS is more complex and qulified users are needed, and some laborious and unautomatized pretreatment steps increases the duration of analysis [18]. On the other hand, reference ranges and clinical decision levels were defined according to immunoassay based methods in many guidelines. Several studies have been conducted to reevaluate these ranges according to LC-MS/MS based methods [19], [20]. Additionally, diurnal variations of many steroid hormones were assessed using LC-MS/MS methods. Reference range studies should be conducted by same time sampling, considering these diurnal variations [9], [21].

In conclusion, this ID-LC-MS/MS method would be useful in clinical laboratories, especially for the immunoassays having insufficient test performance and when checking for possible interferences in immunoassays available. Accurate and reliable results can be obtained with high throughput with LC-MS/MS assays [22]. LC-MS/MS assays have a key contribution in diagnosis of endocrine disorders [23], and their use in steroid hormone analysis seem to be more common in the future. However, LC-MS/MS assays do not seem to replace immunoassays and they are thought to complement each other [17].