An LC/MS/MS method for analyzing the steroid metabolome with high accuracy and from small serum samples

Analyzing global steroid metabolism in humans can shed light on the etiologies of steroid-related diseases. However, existing methods require large amounts of serum and lack the evaluation of accuracy. Here, we developed an LC/MS/MS method for the simultaneous quantification of 12 steroid hormones: testosterone, pregnenolone, progesterone, androstenedione, corticosterone, 11-deoxycortisol, cortisol, 17-hydroxypregnenolone, 17-hydroxyprogesterone, dehydroepiandrosterone, estriol, and estradiol. Steroids and spiked internal standards in 100 μl serum were extracted by protein precipitation and liquid-liquid extraction. The organic phase was dried by evaporation, and isonicotinoyl chloride was added for steroid derivatization, followed by evaporation under nitrogen and redissolution in 50% methanol. Chromatographic separation was performed on a reverse-phase PFP column, and analytes were detected on a triple quadrupole mass spectrometer with ESI. The lower limits of quantification ranged from 0.005 ng/ml for estradiol to 1 ng/ml for cortisol. Apparent recoveries of steroids at high, medium, and low concentrations in quality control samples were between 86.4% and 115.0%. There were limited biases (−10.7% to 10.5%) between the measured values and the authentic values, indicating that the method has excellent reliability. An analysis of the steroid metabolome in pregnant women highlighted the applicability of the method in clinical serum samples. We conclude that the LC/MS/MS method reported here enables steroid metabolome analysis with high accuracy and reduced serum consumption, indicating that it may be a useful tool in both clinical and scientific laboratory research.

In this study, we established a novel LC/MS/MS method based on acylation derivatization with alcoholic hydroxyl groups for the steroid metabolome, including progestogens, androgens, estrogens, mineralocorticoids, and glucocorticoids ( Fig. 1), and this proposed method was successfully applied to detect the steroid metabolome in second-trimester pregnant women. The proposed method shows a number of advantages. First, the method was proven to exhibit high accuracy by certified reference materials covering most steroids except for pregnenolone, 17-hydroxypregnenolone, and estriol. Second, only 100 l serum was required in this method, and the serum consumption was very low among the reported methods analyzing the steroid metabolome (1,12,14,17,18). The minimized serum consumption will increase the applicability of method. Third, the important estrogens (estradiol and estriol) were included in steroid profiling. In this proposed method, similar to other steroids, estrogens could react to generate derivative products that could be detected in the ESI positive mode. Therefore, we achieved the simultaneous quantification for estrogens and other steroids in a single analysis.

Subjects
In this study, we enrolled 30 singleton second-trimester pregnant women and analyzed their steroid metabolome

Sample preparation
We transferred 100 l of calibrators, QC samples, and serum samples to polypropylene tubes. Then, 5 l of IS working solution and 200 l acetonitrile were added, followed by vortexing (30 s) for protein precipitation. Subsequently, 1 ml MTBE was added, and the mixture was vortexed for 5 min for liquid-liquid extraction. After centrifugation (12,000 rpm, 5 min), the upper organic layer was transferred to a clean tube and evaporated under nitrogen at 55°C. The dried residue was redissolved in 100 l dichloromethane and 10 l INC. The final mixture was then evaporated under nitrogen at 55°C and reconstituted in 100 l 50% methanol and transferred to an autosampler vial.

LC/MS/MS instrumentation and analytical conditions
The LC instrumentation was an LC-20AD chromatograph (Shimadzu, Kyoto, Japan), and the tandem mass spectrometer was an AB SCIEX 4500 QTRAP mass spectrometer. We conducted the chromatographic separation of steroids on a Kinetex™ 2.6 m PFP 100 Å column (100 × 3 mm) with a flow rate of 0.3 ml/min. The column temperature was set to 45°C, and the injection volume was 20 l. Mobile phases were purified water (A) and methanol (B). The gradient elution was summarized as follows: 0-2 min, 60% B; 2-13.9 min, 60% to 100% B; 13.9-14.0 min, 60% B; and 14-21 min, 60% B. The total analysis required 21 min, and more polar compounds were diverted to waste for the first 6.8 min to keep the mass spectrometer clean.
The mass spectrometer was operated in positive mode using an ESI source. The analytes were monitored by multiple reaction monitoring, and optimized quantitative mass transitions are listed in Table 1. In addition, positive ESI mass spectra of derivatized steroids are shown in supplemental Figs. S1-S3). The cleavages and fragments of analytes are shown in supplemental Table S2. Data acquisition and processing were conducted using AB SCIEX Analyst version 1.6.2.

Method validation
To validate the LC/MS/MS method, we evaluated the accuracy, linearity, sensitivity, selectivity, precision, matrix effect, and stability in accordance with the US Food and Drug Administration and Clinical and Laboratory Standards document C62-A.
To assess accuracy, we calculated apparent recoveries by dividing the measured steroid concentrations to the nominal spiked values in the blank matrix, and the values of apparent recoveries should be between 85% and 115%. The calibration curves were calculated by measuring the six calibrators in triplicate using a linear least-squares regression model, and the coefficient of determination (R 2 ) was used to evaluate the linearity. The lower limits of quantification (LLOQs) were defined as the lowest concentration in serum that could be quantitatively measured with imprecision (<20%) and accuracy (within ±20%). To evaluate selectivity, we respectively spiked high concentrations (C6) of 12 steroids into low-QC and high-QC samples to fabricate interference-introduced serum, and then we calculated the apparent recoveries of steroids except for the introduced steroid, and the apparent recoveries should be between 85% and 115%. Method imprecision was assessed using QC samples at three concentrations. The measurement was performed six times in one day (intra-assay) and in six consecutive days (interassay). The CV of measurement values was used to represent the imprecision, and the CV should not be greater than 15%. Matrix effects (MEs) were calculated with the equation ME = B/A, where B represents the peak area ratios of steroids to ISs from extracts after liquid-liquid extraction of the matrix, and A refers to the peak area ratios of steroids to ISs from extracts after liquid-liquid extraction of pure water. The ME values should be between 85% and 115%. Stability experiments were performed using QC samples at three levels. First, storage stability was assessed after storage at room temperature (12 h and 72 h), 4°C (72 h and 7 d), and 80°C (30 d and 60 d). Second, freezethaw stability was tested after three cycles of freezing (80°C) and thawing (25°C). Third, posttreatment stability was evaluated at 4°C and 25°C for 4 h and 24 h. Moreover, we evaluated the traceability of this method using certified reference materials. The MassCheck Steroid Panel included testosterone, E2, progesterone, AD, DHEA, 17OHP, COR, CORT, and DOC at three concentrations. The SRM included testosterone, progesterone, and COR at two concentrations, and the BCR included E2 at three concentrations. These reference materials were measured in triplicate using the proposed method, and then the measured values were compared with the certified values.

Optimization of the derivatization reaction conditions
The derivatization reaction is shown in Fig. 3. To achieve the greatest reaction efficiency, several parameters were optimized, including reaction solvents, time, and temperature. We found that dichloromethane was the most conducive to the reaction compared with ACN, methylbenzene, ethyl acetate, hexane, MTBE, and trichloromethane. Moreover, evaporation by nitrogen after adding dichloromethane and INC was an indispensable step for the reaction. The reason may be that evaporation could lead to the volatilization of HCl and then facilitate the reaction. Before evaporation, time and temperature could have little influence on the reaction. Considering the stability of steroids at high temperatures and pretreatment efficiency, immediate evaporation after introducing dichloromethane and INC and a temperature of 55°C were chosen. The entire evaporation required 2 min. Consequently, the reaction was allowed to complete within 2 min. There are two main advantages of isonicotinoyl ester derivatization compared with other derivatization methods. First, this derivatization reaction is simple and fast. It takes just 2 min. Second, sodium and other nonvolatile substances or trifluoroacetic acid would be not introduced. These substances may influence the ionization or damage the chromatographic column and mass spectrometer.

Method validation
Sensitivity and linearity. The LLOQ and linearity for steroids are listed in Table 2. Such sensitivity was sufficient, even for special groups of individuals such as children and postmenopausal women, and the R 2 values

The selection of the extracting solvents
After the protein precipitation, liquid-liquid extraction was conducted. In order to obtain maximum extraction efficiency, we compared the extraction recoveries of several organic solvents, including MTBE, diethyl ether, dichloromethane, and hexane. Extraction recoveries were evaluated on the basis of two groups. For group A, a standard solution mixture of steroids was added to blank serum before extraction, and the IS mixture was added to the organic extract after extraction. For group B, both the standard and IS mixture were added after extraction. Extraction recoveries were calculated by dividing the peak areas ratios of steroids to the IS in group A to the ratios in group B. As shown in Fig. 2, MTBE produced the best extraction recovery for most steroids. Therefore, MTBE was chosen as the extracting solvent for liquid-liquid extraction in this method.  for the linearity of steroids were not less than 0.99. Although AD and progesterone were detected without derivatization due to the lack of alcoholic hydroxyl groups, their sensitivity was not compromised by the process of derivatization.
Selectivity. Due to the similar structures of steroids, we evaluated the specificity by introducing steroids as analog interferences of other steroids. After introducing interferences, we found no significant effect for targets' apparent recoveries (ranging from 86.5% to 102.8%), indicating satisfactory selectivity of this method.
Accuracy and imprecision. Both accuracy and imprecision were evaluated on three QC levels: high, medium, and low. Apparent recoveries of steroids at high-, mediumand low-QC concentrations are listed in Table 3. For all steroids, the apparent recoveries were between 86.4% and 115.0% at different concentrations. After evaluating method precision, we found that intra-assay and interassay precisions were between 5.3% and 12.8% for each steroid ( Table 4). These results demonstrated satisfactory accuracy and precision.
Matrix effect and stability. The ME ranges of steroids for all concentration levels were all between 91.8% and 103.5%, indicating that there was no significant ion suppression or enhancement from the matrix, and the quantification of steroids would not be affected by the serum matrix during steroid analysis. Although the commercial blank matrix is widely used for matrix effect evaluation, we should recognize the possibility that the charcoal-stripped process could remove many other components that themselves may affect matrix effect.
After conducting stability experiments, we found that all steroids could be well preserved in serum for 72 h at 25°C, 7 d at 4°C, and 60 d at 80°C, and the changes in steroid concentrations were within 10% of theoretical concentrations after three freeze-thaw cycles. Posttreatment stability experiments showed that the final products were stable for 24 h at 4°C and 25°C.

Traceability from certified reference materials
After measuring the certified reference materials and comparing the measured values with certified values at different levels, we found that there were limited biases (10.7% to 10.5%) between the measured values of our developed method and the certified values in the SRM and BCR reference materials ( Table 5) and the MassCheck Steroid Panel Serum Control ( Table 6). These results show the high accuracy of our methods.

Clinical application
Pregnancy is a special period for women. During this period, steroids are of particular importance for normal pregnancy, and the concentrations of many steroids undergo great changes. In this study, we enrolled 30 singleton second-trimester pregnant women and analyzed their steroid metabolome using the proposed LC/MS/ MS method. The separation efficiency remained excellent during real sample detection, and chromatograms of steroids in serum samples of pregnant women are shown in Fig. 4. The baseline characteristics and steroid concentrations are listed in Table 7.

CONCLUSION
In summary, this LC/MS/MS method describes a reliable and sensitive assay based on acylation derivatization for steroid metabolome that possesses high accuracy and requires very low serum consumption. This method was successfully applied in a steroid metabolome analysis of pregnant women and is a promising avenue for future research on steroid metabonomics.