Comprehensive Steroid Assay with Non-Targeted Analysis Using Liquid Chromatography Ion Mobility Mass Spectrometry

Aldosterone-producing adenomas (APAs) have different steroid profiles in serum, depending on the causative genetic mutation. Ion mobility is a separation technique for gas-phase ions based on their m/z values, shapes, and sizes. Human serum (100 µL) was purified by liquid–liquid extraction using tert-butyl methyl ether/ethyl acetate at 1/1 (v/v) and mixed with deuterium-labeled steroids as the internal standard. The separated supernatant was dried, re-dissolved in water containing 20% methanol, and injected into a liquid chromatography–ion mobility–mass spectrometer (LC/IM/MS). We established a highly sensitive assay system by separating 20 steroids based on their retention time, m/z value, and drift time. Twenty steroids were measured in the serum of patients with primary aldosteronism, essential hypertension, and healthy subjects and were clearly classified using principal component analysis. This method was also able to detect phosphatidylcholine and phosphatidylethanolamine, which were not targeted. LC/IM/MS has a high selectivity for known compounds and has the potential to provide information on unknown compounds. This analytical method has the potential to elucidate the pathogenesis of APA and identify unknown steroids that could serve as biomarkers for APA with different genetic mutations.


LC/IM/MS Measurement of Standard Solution Containing Steroids
For LC/IM/MS analysis, injection of 20 µL of the sample solution prepared with H 2 O/MeOH at 8/2 (v/v) allowed for the sensitive detection of steroids. Highly hydrophilic steroids, such as aldosterone, 18-hydroxycorticosterone, and 19-hydroxy-4-androstene-3,17dione were adequately separated by adopting a gentle slope as a gradient condition for LC. Highly hydrophobic steroids, such as progesterone, testosterone, and androstenedione, can be separated easily with a steeper gradient program. The peak shape of compounds with a long retention time (RT) at a constant flow rate of the mobile phase is broad, generally because of the slow flow speed in LC. A gradient condition with a combination of gentle and steep slopes was adopted to obtain sharp peaks and high sensitivity.
Calibration curves were created for each of the 20 steroids, and they all yielded good linearities, as shown in Supplementary Figure S1. The coefficient of determination (R2) for steroids was ≥0.994.  IM. From the above results, steroids were separated and detected using LC/IM/MS, and spots of these steroids were observed in 2D plots. Calibration curves were created for each of the 20 steroids, and they all yielded good linearities, as shown in Supplementary Figure S1. The coefficient of determination (R2)

LC/IM/MS Measurement of Human Serum
Human serum spiked with a standard steroid mixture and an internal standard was analyzed using LC/IM/MS. The recovery rates of steroids in the serum were determined. A summary of the limit of detection (LOD), the limit of quantitation (LOQ), analytical recoveries, and matrix effect for LC/IM/MS of serum samples are shown in Table 2. The quality control (QC) values of the recoveries were run in replicates (n = 5). Cortisone, cortisol, corticosterone, and testosterone were evaluated using a 100-fold diluted serum. Recovery rates for most steroids were greater than 80%. The LOD was determined at a concentration of ≥3 signal-to-noise (S/N) ratio. The LOQ was determined at concentrations ≥5 S/N ratio and <20% relative standard deviation (RSD). The analysis of a small amount of aldosterone as the final metabolite originating from cortisol required an LOD of 3 ng/dL because of the low aldosterone levels in healthy individuals. LODs were 1.0, 1.0, 1.0, and 0.5 ng/dL for aldosterone, 18-hydroxycorticosterone, testosterone, and androstenedione, respectively. This indicated that quantitative analysis of aldosterone and other steroids with high sensitivity could be performed using LC/IM/MS. Table 2. Internal standards (IS), IS concentration (IS conc.), limit of detection (LOD), limit of quantitation (LOQ), analytical recoveries, and matrix effect of steroids in serum samples.
androstenedione-d3, 17α-hydroxyprogesterone-d8 (17α-OHP-d8), and progesterone-d9 were used as internal standards. The QC values of the recoveries were run in replicates (n = 5). The matrix effect was established using [peak area of steroid obtained from serum spiked with standard solution]/(10 × [peak area of steroid obtained from 10-fold diluted serum spiked with standard solution]). Cortisone, cortisol, corticosterone, and testosterone were evaluated using a 100-fold diluted serum. The concentrations of all steroids are expressed as ng/dL. LOD was obtained at an S/N ratio concentration of ≥3. The LOQ was obtained with a concentration of S/N ratio ≥5 and RSD <20%. The 100-fold water-diluted serum samples were prepared for analysis of cortisone, cortisol, corticosterone, and testosterone. A marginal ion suppression was observed as the matrix effect derived from serum components ( Table 2). Early-eluting compounds from LC have a lower matrix effect than late-eluting compounds do. The matrix effect coefficient of late-eluting compounds was not close to 1.0. However, we decided to use the obtained calibration curves for quantitative analysis because of the small matrix effect and high recovery rate. Androstenedione and progesterone with high matrix effect had low recovery rates.

Compounds
Serum specimens from four individuals were analyzed using LC/IM/MS. These included two healthy subjects: one with EHT and the other with PA. The steroids detected are shown in Table 3. The aldosterone concentrations in the sera of patients with EHT and PA were higher than those in healthy subjects. The activity of CYP11B2, an aldosterone synthase, is expressed as 18-oxocortisol (18-oxoF)/cortisol ratio or aldosterone/corticosterone ratio. The 18-oxoF/cortisol ratio was 0.000361, 0.000187, 0.000426, and 0.000745 for healthy subject 1, healthy subject 2, EHT patients, and PA patients, respectively. The aldosterone/corticosterone ratio was 0.013571, 0.016875, 0.026667, and 0.065385, respectively. Compared with that in the other subjects, CYP11B2 activity was the highest in patients with PA. It is difficult to evaluate the correlation between all steroids. Principal component analysis (PCA) was used to statistically analyze the steroidal differences in each specimen. The results are shown in Figure 3. PCA score plots showed clear separation and grouping of healthy and hypertensive samples, respectively. The PCA loading plots show that steroid ions contribute to clustering. The X-axis and Y-axis describe the first and second principal components (PCs), respectively. The major contributors to the differences between healthy patients and other patients were aldosterone, 11-deoxycortisol, 18-hydroxycortisol, and 18-hydroxycorticosterone. The results suggest that PCA analysis with LC/IM/MS data has the potential to group patients with hypertension and discover novel biomarkers.

Discussion
We established a system for steroids that simultaneously enables targeted measurement and non-targeted analysis using LC/IM/MS. In addition to classical steroid hormones, 20 types of steroids, including hybrid steroids, such as 18-oxocortisol and 18hydroxycortisol, can be measured with high sensitivity. Using this method, steroid profiles in the sera of the patient with PA, patient with EHT, and healthy subjects were measured and clearly classified using PCA. Quantitative analysis of these steroids simultaneously detected PC and PE, which were not targeted lipids. However, our

Discussion
We established a system for steroids that simultaneously enables targeted measurement and non-targeted analysis using LC/IM/MS. In addition to classical steroid hormones, 20 types of steroids, including hybrid steroids, such as 18-oxocortisol and 18-hydroxycortisol, can be measured with high sensitivity. Using this method, steroid profiles in the sera of the patient with PA, patient with EHT, and healthy subjects were measured and clearly classified using PCA. Quantitative analysis of these steroids simultaneously detected PC and PE, which were not targeted lipids. However, our method does not require derivatization and can identify non-targeted steroids and lipid metabolites, regardless of the ease of compound derivatization.
The comprehensive measurement of steroids using MS is a very useful tool in PA research. Adina et al. measured 17 steroids from adrenal venous sampling and peripheral serum and examined their usefulness for PA subtype classification. They showed that 11β-hydroxyandrostenedione and 11-deoxycortisol were superior to cortisol from the interpretation of adrenal venous sampling data and that a multi-steroid panel measured in peripheral blood was useful for the stratified classification of PA [16]. In addition, Eisenhofer et al. reported that comprehensive plasma steroid profiling combined with machine learning could facilitate screening for PA and identify patients with unilateral adenomas due to pathogenic KCNJ5 variants [17]. The vast amount of information obtained from comprehensive steroid profiling in combination with artificial intelligence analysis, such as machine learning, may contribute to the understanding of the pathogenesis of aldosterone overproduction and provide more efficient and less invasive testing.
Ion mobility spectrometry (IMS) is a technique used to separate compounds in a mixture by CCSs. The IMS cell is filled with nitrogen gas (N 2 ), and the electrostatic field applied in the direction of travel causes the ions to move. The bulkier compounds collide more frequently with N 2 molecules, resulting in lower mobility. The difference in the resulting transit time (DT) allows for the separation of compounds. The CCS of an ionized compound depends only on the DT and m/z and is unique to that compound. Therefore, IMS-MS, which combines IMS with MS, enables the separation of components with the same molecular weight and the highly accurate analysis of the steric structure of compounds based on the compositional formula estimated by MS and the CCS information obtained by IMS. Furthermore, when used in combination with LC, IMS-MS is also effective for qualitative analysis of structural isomers that cannot be separated by LC [18][19][20][21]. The differences in the steric steroid ring structure related to structural isomers have markedly different biological effects in the human body [22,23]. Steroids are compounds with several structural isomers. Fourteen (e.g., m/z 361.200; cortisone and aldosterone) of the twenty steroids have structural isomer relationships in this study. These steroidal structural isomers were also clearly separated using three labels: RT, m/z, and DT. The principle of LC/IM/MS is well suited for the quantification of steroids, as structural isomers can be clearly separated. Hernandez-Mesa and Arthur separated steroid isomers by LC/IM/MS [14,15]. To improve the separation of isomers by IM, formed steroid alkali metal adduct ions [M + alkali metal] + were used instead of proton adduct ions [M + H] + . The advantage of our LC/IM/MS method over LC/MS/MS, as a steroid determination method, is that it can simultaneously perform highly sensitive quantitative and non-targeted analyses.
Indeed, the analysis of protein complexes using IM/MS has been reported [24,25], and Baker et al. reported the application of LC/IM/MS for proteomics [26]. LC/IM/MS has also been used for metabolomic analysis [27], improving the S/N ratio of lipids by reducing chemical noise. For urinary steroid analysis, LC/IM/MS has been reported to separate steroids and their isomers for anti-doping analysis [28]. Velosa [29]. The results with multiple ionic species of [M + H] + and [M + Na] + enabled secondary characterization and improved the resolution of stereoisomers. However, [M + Na] + is non-volatile and may not be suitable for continuous quantitative analysis because it is more likely to contaminate a mass spectrometer than [M + H] + . As described above, LC/IM/MS has the potential to improve the reliability of screening for known steroids and provide structural information for the elucidation of unknown compounds, even steroids. This may allow us to classify common endogenous steroids and synthetic/exogenous compounds more accurately or to discover new steroid metabolic pathways in vivo. LC/IM/MS may also be beneficial in the search for biomarkers and drug targets for the diagnosis and prediction of disease severity.
In clinical practice, steroid hormones have been measured using immunoassays (IAs) [30][31][32]. However, IAs are questionable for the analysis of complex biological samples, such as human specimens, because of their cross-reactivity. IAs are inferior to MS-based assays, which can measure multiple components simultaneously, because they can measure only one compound at a time. The ability to measure multiple steroids thoroughly at once, the lack of cross-reactivity, low LOD, and high accuracy are characteristics that make LC/IM/MS superior to IAs.
This study has several limitations. First, it measured only a small number of human serum samples. More cases need to be studied, particularly for evaluation using statistical analysis. Second, regarding the optimization of the pretreatment, there is a trade-off between contamination of the mass spectrometer and the detection sensitivity of non-targeted analysis [33]. For quantitative analysis, exact extraction is necessary to reduce the signal noise. In contrast, for non-targeted analysis, over-extraction results in fewer compounds being detected. Inadequate pretreatment is not suitable for continuous measurements because the sample contaminates the column and instrument and requires frequent cleaning. It is necessary to further investigate the optimal conditions for efficient non-targeted analysis with high sensitivity while minimizing cleaning operations associated with contamination. Finally, the absence of a compound library for LC/IM/MS precludes easy compound identification. The accumulation of RT, m/z, and DT data for compounds is needed in the future.

Sample Preparation
The mixed standard stock solution of 20 steroids and 3 deuterium-labeled steroids were prepared with water, methanol, and acetonitrile from a concentration of 0.1 mg/mL to 1 mg/mL. Thereafter, the non-labeled stock solution was diluted with methanol to 20,000, 6000, 2000, 600, 200, 60, 40, 20, and 10 ng/dL as steroid standard solutions. Seventeen deuterium-labeled steroid mixtures from the AbsoluteIDQ ® Stero17 Kit were dissolved in methanol. The 17 deuterium-labeled steroid solutions were diluted five times with three deuterium-labeled steroid solutions, methanol, and water. The three deuterium-labeled steroids were prepared at 60 ng/mL. The obtained 20 deuterium-labeled steroid solution in methanol/water at 8/2 (v/v) was used as the internal standard solution. These solutions were stored at −24 • C in glass tubes.
The sera were stored at −80 • C in plastic tubes. Serum samples from a healthy person were purchased from Cosmo Bio Co. Ltd. (Tokyo, Japan) as healthy 1. Healthy adult serum samples were obtained from a 34-year-old male volunteer as healthy 2. There were two patients: one was a 40-year-old man with PA who was administered spironolactone, and the other was a 45-year-old man with EHT who was administered angiotensin receptor blockers and calcium channel blockers. Written informed consent was obtained from all three subjects. Sera containing 1, 2, 3, 10, 30, 100, and 300 ng/dL steroids were prepared by the addition of 5 µL methanol solution of steroid standard. The solution was mixed with 5 µL internal standard solution in a 1.5 mL tube. These solutions were used for creation of calibration curves. The 100-fold water-diluted serum samples were prepared for analysis of cortisone, cortisol, corticosterone, and testosterone. Serum steroids were purified using LLE with tert-butyl methyl ether (MTBE)/ethyl acetate at 1/1 (v/v). LLE is a commonly used steroid extraction method [10,34]. We have improved the protocol for non-target LC/IM/MS. MTBE/ethyl acetate solution (400 µL) was added to the serum sample. The obtained solution was vortexed (VTX-3000L, LMS Co. Ltd., Tokyo, Japan), centrifuged (800× g, 1 min, KUBOTA Model 3520, Kubota, Tokyo, Japan), and the supernatant was transferred to a 1.5 mL tube. This LLE procedure was repeated twice to extract steroids from the remaining aqueous phase in high yield. The supernatant was dried by decompression centrifugation at 50 • C, and water with 20% methanol aqueous solution of 100 µL was added to the dried sample. The solution was injected into the LC/IM/MS.
To establish the matrix effect, the serum samples and 10-fold water-diluted serum samples were compared. The matrix effect was evaluated using the following formula: [peak area of steroid obtained by serum spiked with standard solution]/[(peak area of steroid obtained by 10-fold diluted serum spiked with standard solution) × 10].

Measurement Apparatus
LC/MS was performed using Acquity UPLC H-Class PLUS (Waters, Milford, MA, USA) and MALDI Synapt G2-Si HDMS (Waters) coupled with electrospray ionization (ESI) source. Data acquisition and processing were performed using MassLynx 4.2 and DriftScope 2.9 (Waters). The sample injection volume was 20 µL, and the column temperature was 42 • C. An Inertsil ODS-HL column (1.9 µm 1.0 × 100 mm, GL Sciences, Tokyo, Japan) was used at a flow rate of 0.1 mL/min. The mobile phase consisted of A (1 mM ammonium formate aqueous solution), B (acetonitrile), C (methanol), and D (THF and acetonitrile (v/v, 1/1)). Mobile phase D was used to clean the LC column after gradient separation. The gradient program occurred for 40 min as follows: 0 min to 0% B and 0% C, 1 min to 0% B and 0% C, 2.5 min to 18% B and 2% C, 13 min to 22.7% B and 2.3% C, 27 min to 58.5% B and 6.5% C, 31 min to 90% B and 10% C, 39 min to 90% B and 10% C, and 40 min to 100% D.
The ESI capillary voltage was set to 2.0 kV, and the sampling cone voltage was 30 V. The ion source and desolvation temperatures were 125 and 450 • C, respectively. The injection voltages into the trap and transfer cells were 4 and 2 V, respectively. Argon gas flowed into the trap and the transferred cells. The ion mobility cell traveling wave velocities increased from 650 to 1500 m/s with a wave height of 40 V. The trapping release time was 54 µs. N 2 and helium gas were used in the helium and IMS cells, respectively. Sodium formate was used for the m/z calibration. Recalibration of the lock mass was performed using the fragmentation ion of leucine-enkephalin at m/z 397.187.
PCA was performed with [steroid peak area obtained from each subject]/[average of steroid peak area obtained from healthy subject 1 (n = 3)] using SIMCA-P+ ver. 12.0.1 software (Umetrics AB, Umea, Sweden). Each sample was measured in triplicate for PCA.

Conclusions
Twenty steroids in the serum were detected with high sensitivity and selectivity using LC/IM/MS. Although quantitative analysis via MS/MS is generally performed using multiple reaction monitoring, the application of ion mobility instead of the fragmentation method has better selectivity and an improved S/N ratio. In addition, this method can simultaneously perform non-targeted analysis of hydrophobic compounds, such as lipids, which are maintained in the same pretreatment as steroids. This analytical method is useful for establishing pathogenesis and elucidating the mechanism of aldosterone-producing adenomas, which are undergoing remarkable molecular developments, and for advancing medical research on glucocorticoids and sex steroids. For example, the ratio of precursor to product can be used to evaluate the activity of steroid enzymes. In addition, the ability of this method to identify non-targeted compounds, even if they are not commercially available and cannot be purchased as standards, would allow relative comparisons of the abundance of unknown compounds among subjects. This would be useful for finding new unknown biomarkers. Steroids and steroid-like compounds are also present in plants, mammals, and prokaryotes. The detection and quantification of steroids and steroid-like substances in these organisms could lead to a wide range of biological advances.  Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.

Data Availability Statement:
The de-identified participant data will be shared on a reasonable request basis. Data are available from Issey Osaka with permission from Toyama Prefectural University.