Nora Vogg*, Eleanor North, Arne Gessner, Felix Fels, Markus R. Heinrich, Matthias Kroiss, Max Kurlbaum, Martin Fassnacht and Martin F. Fromm
An untargeted metabolomics approach to evaluate enzymatically deconjugated steroids and intact steroid conjugates in urine as diagnostic biomarkers for adrenal tumors
https://doi.org/10.1515/cclm-2024-1337
Received September 17, 2024; accepted December 19, 2024; published online January 7, 2025
Abstract
Objectives: Urinary steroid profiling after hydrolysis of conjugates is an emerging tool to differentiate aggressive adrenocortical carcinomas (ACC) from benign adrenocor- tical adenomas (ACA). However, the shortcomings of deconjugation are the lack of standardized and fully vali- dated hydrolysis protocols and the loss of information about the originally conjugated form of the steroids. This study
*Corresponding author: Nora Vogg, PhD, Institute of Experimental and Clinical Pharmacology and Toxicology, Friedrich-Alexander-Universität Erlangen-Nürnberg, Fahrstr. 17, 91054 Erlangen, Germany,
E-mail: nora.vogg@fau.de. https://orcid.org/0000-0001-6559-2455 Eleanor North, Institute of Experimental and Clinical Pharmacology and Toxicology, Friedrich-Alexander-Universität Erlangen-Nürnberg, Erlangen, Germany
Arne Gessner and Martin F. Fromm, Institute of Experimental and Clinical Pharmacology and Toxicology, Friedrich-Alexander-Universität Erlangen- Nürnberg, Erlangen, Germany; and FAU NeW - Research Center New Bioactive Compounds, Friedrich-Alexander-Universität Erlangen-Nürnberg, Erlangen, Germany. https://orcid.org/0000-0002-1729-4231 (A. Gessner). https://orcid.org/0000-0002-0334-7478 (M.F. Fromm)
Felix Fels, Department of Chemistry and Pharmacy, Pharmaceutical Chemistry, Friedrich-Alexander-Universität Erlangen-Nürnberg, Erlangen, Germany
Markus R. Heinrich, FAU NeW - Research Center New Bioactive Compounds, Friedrich-Alexander-Universität Erlangen-Nürnberg, Erlangen, Germany; and Department of Chemistry and Pharmacy, Pharmaceutical Chemistry, Friedrich-Alexander-Universität Erlangen- Nürnberg, Erlangen, Germany
Matthias Kroiss, Division of Endocrinology and Diabetes, Department of Internal Medicine I, University Hospital, University of Würzburg, Würzburg, Germany; and Department of Internal Medicine IV, University Hospital Munich, Ludwig-Maximilians-Universität München, Munich, Germany Max Kurlbaum and Martin Fassnacht, Division of Endocrinology and Diabetes, Department of Internal Medicine I, University Hospital, University of Würzburg, Würzburg, Germany; and Central Laboratory, Core Unit Clinical Mass Spectrometry, University Hospital Würzburg, Würzburg, Germany. https://orcid.org/0000-0002-8438-2127 (M. Kurlbaum)
aimed to evaluate the quality of the deconjugation process and investigate novel diagnostic biomarkers in urine without enzymatic hydrolysis.
Methods: 24 h urine samples from 40 patients with ACC and 40 patients with ACA were analyzed by untargeted metab- olomics using liquid chromatography-high-resolution mass spectrometry both unmodified and after hydrolysis with arylsulfatase/glucuronidase from Helix pomatia. Both ap- proaches were compared regarding the differentiation of ACC vs. ACA via ROC analyses and to evaluate the hydro- lyzation efficiency of steroid conjugates.
Results: Steroid glucuronides were fully deconjugated, while some disulfates and all monosulfates were still largely detectable after enzymatic hydrolysis, suggesting incom- plete and variable deconjugation. In unhydrolyzed urine, steroid monosulfates showed the best differentiation be- tween ACC and ACA (highest AUC=0.983 for C21H32O6S, fol- lowed by its isomer and two isomers with the molecular formula C21H32O7S). Moreover, several disulfates were highly abundant and increased in ACC compared to ACA.
Conclusions: This work highlights the limitations of hy- drolyzing steroid conjugates before analysis and shows a possible superiority of a direct analysis approach compared to a hydrolysis approach from a methodological point of view and regarding diagnostic accuracy. Several steroid conjugates were found as promising diagnostic biomarkers for differentiation between ACC and ACA.
Keywords: adrenocortical adenoma; adrenocortical carci- noma; enzymatic hydrolysis; LC-MS; mass spectrometry; conjugated steroids
Introduction
Adrenal tumors are among the most frequent neoplasms in humans and are often detected incidentally during imaging for other diagnostic purposes [1]. The diagnosis of inci- dentalomas aims to identify potential autonomous hormone secretion and malignancy of the tumor and is based on
imaging and an endocrine workup [2]. Differentiating benign adrenocortical adenomas (ACA) from adrenocortical carcinomas (ACC) commonly relies on imaging modalities such as unenhanced computed tomography or magnetic resonance imaging with chemical-shift analysis [3]. These imaging characteristics are highly sensitive to identify ACC via the low lipid content of malignant adrenal lesions. However, at least a third of ACA are likewise lipid-poor resulting in a limited clinical specificity [3]. Targeted steroid hormone profiling in plasma and urine has increasingly been applied for the differentiation of ACC and ACA over the last decades yielding promising results [4].
Regarding the analysis of urinary steroids, a preceding hydrolysis step is commonly performed in order to determine the sum of previously unconjugated steroids and deconju- gated steroids after hydrolysis. Several studies identified tet- rahydro-11-deoxycortisol (THS), 5-pregnene-3,17a,20a-triol (5-PT), and 5-pregnene-30,20a-diol (5-PD) after deconjugation as the steroids with the highest discriminative power for ACC vs. ACA [5-11]. However, a considerable limitation of urine hydrolysis is the lack of standardized or universal hydrolysis protocols, i.e., different enzyme types and/or incubation conditions are used in different laboratories, which were not necessarily evaluated for a complete deconjugation. The most commonly used enzyme preparation is arylsulfatase/glucu- ronidase derived from Helix (H.) pomatia, although incom- plete hydrolysis and unwanted transformations of steroids to different steroids during hydrolysis due to side enzyme ac- tivity in the H. pomatia extract were reported [12-14].
A further limitation of the common process of hydro- lysis of steroid conjugates, is that information about the originally conjugated forms is lost and little is known about the excretion and diagnostic utility of the individual steroid conjugates. Even though it is technically possible to detect intact steroid conjugates by liquid chromatography-mass spectrometry (LC-MS), authentic reference standards for most steroid conjugates are not commercially available. This challenges method development and research regarding steroid conjugates as diagnostic biomarkers for ACC. Espe- cially steroid monosulfates and disulfates have recently raised attention as potential biomarkers for adrenal tumors [15, 16] after estrogen sulfates were identified with a poten- tial prognostic relevance in ACC tissue by MS imaging [17].
Untargeted metabolomics by high-resolution LC-MS provides the opportunity to comprehensively evaluate the urinary excretion of metabolites including conjugated ste- roids and is not limited to a predetermined analyte panel as it is the case for targeted assays. Full-scan mode provides insights into the full range of ionizable small-molecule me- tabolites and enables the detection of novel biomarkers within an unbiased screening approach. In recent years,
several studies reported on the successful discovery of novel diagnostic, prognostic, or predictive cancer biomarkers in urine by untargeted metabolomics [18]. To the best of our knowledge, only one study previously investigated the uri- nary metabolome of patients with ACC vs. patients with benign adrenal tumors [19]. Creatine riboside was found to be elevated 2.1-fold in patients with ACC and L-tryptophan, NE,NE,NE-trimethyl-L-lysine and 3-methylhistine to be lower 0.33-fold, 0.56-fold, and 0.33-fold in patients with ACC compared to benign adrenal tumors, respectively. However, steroid metabolites were not mentioned [19].
The main objective of this study was to compare the urine metabolomes of patients with ACC and ACA to screen for novel diagnostic biomarkers for ACC by using untar- geted metabolomics. Therefore, a direct analysis approach and an enzymatic hydrolysis approach were respectively applied to 24 h urine samples from 40 patients with ACC and 40 patients with ACA and the detected compounds with high and significant group differences and promising diagnostic accuracy were compared between both ap- proaches. A secondary aim of the present study was the evaluation of the efficiency of steroid deconjugation with arylsulfatase/glucuronidase from H. pomatia, which was investigated both in the patient samples and within a sub- study analyzing the hydrolysis of steroid reference stan- dards in artificial urine.
Materials and methods
Chemicals and reagents
LC-MS grade water, methanol, and acetonitrile were pur- chased from VWR (Darmstadt, Germany). Steroid reference standards were obtained from Steraloids (Newport, RI, USA), Sigma-Aldrich (Taufkirchen, Germany), and Biomol (Hamburg, Germany). Arylsulfatase/glucuronidase from H. pomatia, artificial urine diluent, ammonium formate, and formic acid were purchased from Sigma-Aldrich. 16a-hydroxy-dehydroepiandrosterone (16OH-DHEA)-3,16- disulfate, 5-PD-3,20-disulfate, and 17a-hydroxy-pregneno- lone-3,17-disulfate (17OH-pregnenolone disulfate) were custom synthesized as described in the Supplementary Material, Method S1.
Urine samples and study design
As an initial sub-study for the evaluation of hydrolysis efficiency of arylsulfatase/glucuronidase derived from H. pomatia, several reference standards of conjugated and
| ACC (n=40) | ACA (n=40) | |
|---|---|---|
| Male/female | 17/23 | 17/23 |
| Age, years, median (range) | 52 (28-73) | 52 (29-73) |
| Tumor diameter, cm, median (range) | 10.0 (3.4-33.0) | 3.3 (2.0-9.3) |
| Biochemical/clinical evidence of hormone excess, nª | ||
| Cushing syndrome | 8 | 5 |
| Mild autonomous cortisol secretion | 22 | 21 |
| Conn syndrome | 3 | 4 |
| Hyperandrogenemia | 21 | 0 |
| Nonfunctioning | 6 | 12 |
| ENSAT stage at diagnosis, n | n/a | |
| I | 4 | |
| II | 13 | |
| III | 8 | |
| IV | 15 |
a19 patients with ACC had combined steroid excess, which was present in only two patients with ACA (Conn syndrome + mild autonomous cortisol secretion).
unconjugated steroids were spiked to artificial urine. This solution was analyzed by untargeted metabolomics both directly and after different incubation conditions. This experiment is further described in the Supplementary
Material, Method S2. To evaluate hydrolysis efficiency in patient samples and diagnostic utility of deconjugated and conjugated steroids, 24 h urine samples were taken from patients treated at the University Hospital Würzburg (Ger- many). The patients were part of the European Network for the Study of Adrenal Tumors (ENSAT) registry that has been approved by the Ethics Committee of the Julius-Maximilians University Würzburg (# 88/11). All patients provided written informed consent. Retrospective metabolomics analyses of these samples were additionally approved by the Ethics Committee of the Friedrich-Alexander Universität Erlangen- Nürnberg (# 23-162-Bp). The study included 40 patients with ACC and 40 sex- and age-matched patients with ACA (Ta- ble 1). Final diagnoses were based on clinical practice guidelines for the management of adrenal incidentalomas and ACC [2, 20, 21] with post-operative histopathology and/or follow-up investigations as gold standards. Samples were from a 24 h urine collection at the time point of first diag- nosis before potential adrenalectomy and/or pharmacolog- ical treatment. Each sample was analyzed both without and with enzymatic hydrolysis (H. pomatia-derived arylsulfa- tase/glucuronidase, 3 h incubation, 55 ℃, pH 4.9) by untar- geted metabolomics for a direct comparison of the diagnostic potential of deconjugated steroids and steroid conjugates (Figure 1).
24 h Urine ACC n = 40
Age- and sex- matched patients
24 h Urine ACA n = 40
Without hydrolysis
Enzymatic hydrolysis
Without hydrolysis
Enzymatic hydrolysis
Untargeted Metabolomics by LC-MS
Identification of differential metabolites (ACC vs. ACA) without hydrolysis
Identification of differential metabolites (ACC vs. ACA) after hydrolysis
Focus on steroid conjugates
Focus on remaining intact steroid conjugates
Focus on deconjugated steroids
Evaluation of hydrolysis efficiency
Comparison of diagnostic utility
Sample preparation
Pooled quality control samples with equal volumes of each study sample were prepared and treated in accordance with the study samples. For the direct analysis approach without enzymatic hydrolysis, 100 uL urine or spiked artificial urine diluent were mixed with 400 uL methanol containing four re- covery standards (tridecanoic acid, DL-2-fluorophenylglycine, cholesterol-d6, and DL-4-chlorophenylalanine) which were used to account for potential sample loss during sample preparation. After centrifugation for 10 min at 16,000 rpm, 2 × 200 uL supernatant were transferred to two HPLC vials with insert for hydrophilic interaction chromatography (HILIC) and reverse phase (RP) analysis, respectively. Solvent was evaporated to dryness under a stream of nitrogen and samples were reconstituted in 50 uL HILIC and RP eluent containing 13 internal standards (imipramine, 1-methyl- nicotineamide, ezetimibe, trimethylamine N-oxide-d9, hex- adecanedioic acid-d28, tetradecanedioic acid-d24, carnitine-d3, creatinine-d3, L-tryptophan-d5, pravastatin-d3, phenoloph- thalein-ß-D-glucuronide, ornithine-d, and citrulline-d7) to monitor potential sample loss during analysis. For enzymatic hydrolysis, 100 uL urine were gently mixed with 200 uL incu- bation buffer (composed of 20 uL arylsulfatase/glucuronidase mix from H. pomatia and 180 uL ammonium acetate buffer, pH 4.9). After incubation at the respective time and temperature conditions, samples were mixed with 400 uL methanolic re- covery standard, centrifuged, 2 × 280 uL supernatant were transferred to HPLC vials with insert and further processed as described above for unhydrolyzed samples.
LC-MS instrumentation and method
Untargeted metabolomics analyses were performed on a Dionex Ultimate 3000 chromatographic system coupled to a Q Exactive Focus mass spectrometer (both from Thermo Fisher Scientific, Dreieich, Germany) fitted with a heated electrospray source. The software interface was TraceFinder 4.1 (Thermo Fisher Scientific, Dreieich, Germany). Each sample was analyzed via HILIC and RP chromatographic separation and each chromatographic separation was analyzed in positive and negative mode resulting in four runs per sample. For HILIC, an Acquity UPLC BEH Amide, 1.7 um, 2.1 x 100 mm column was used and for RP an Acquity UPLC BEH C18, 1.7 um, 2.1 x 100 mm column was installed. Both columns were equipped with a 2.1 x 5 mm guard col- umn (all columns from Waters, Eschborn, Germany). The column temperature was set to 40℃, the flow rate to 0.35 mL/min, and the injection volume to 2 uL. The HILIC eluents were A: water, 0.1 % formic acid, 10 mM ammonium
formate and B: acetonitrile 95 %, 5 % water, 0.1 % formic acid, 10 mM ammonium formate. The RP eluents were A: water, 0.1 % formic acid and B: methanol, 0.1 % formic acid. The HILIC gradient program started with 100 % B (0-2 min), which was decreased to 30 % over 12 min. This concentration was held (14-16.5 min) and returned to the starting condi- tions over 1 min (17.5 min). The system was re-equilibrated for 10 min. The RP gradient started with 0.5 % B. After in- jection, the percentage of B increased from 10 to 98 % over 11 min. From 11 to 15 min, eluent B was maintained at 98 %. Finally, the starting conditions (0.5 % B) were reconditioned from 15 to 15.50 min and maintained until 20 min for column re-equilibration. The mass spectrometer was operated in full MS/dd-MS2 (discovery) mode with a resolution of 70,000 in full scan mode (automatic gain control target of 1e6) and 35,000 in dd-MS2 mode (automatic gain control target 5e4) with the three most intensive ions being fragmented after each survey scan. The scan range was from 66.7-1,000 m/z. Sheath gas, aux gas and sweep gas flow rates were set to 60, 20 and 0, respectively. The capillary voltage was -3 kV in negative and +3 kV in positive ionization mode. Further data processing including spectral alignment and metabolite an- notations was achieved using the Compound Discoverer Software V.3.3 (Thermo Fisher Scientific). Before data anal- ysis in Compound Discoverer, the peak areas of the recovery standards and internal standards were evaluated for an acceptable variability among study samples to exclude po- tential sample loss during sample preparation or LC-MS analysis.
Data processing and identification of metabolites
Processing and evaluation of the metabolomics dataset including spectral alignment, descriptive statistic and metabolite annotations was achieved using the software Compound Discoverer V3.3 (Thermo Fisher Scientific). Peak areas of the metabolites detected in patient samples were normalized to urinary creatinine concentration which was determined using a Cobas® 8000 immunoassay (Roche Di- agnostics GmbH, Mannheim, Germany). The main rationale for the normalization of steroid concentrations to creati- nine instead of collection volume was to be able to compare normalized 24-h excretions to respective spot urine and/or first morning urine samples in a subsequent project. Moreover, it was shown that >30 % of 24-h urine collections might be incomplete [22, 23] and therefore underestimate the true 24-h excretion when normalized to the collection volume. This underestimation is reduced by normalization
of the steroid concentrations to the respective urinary creatinine concentration.
Compounds were annotated via the exact mass and fragmentation patterns from MS2-spectra from data- dependent MS2-mode. Steroid sulfates and disulfates were assigned due to characteristic fragments at m/z 79.96 and/or 96.96 and glucuronides at m/z 75.01, 85.03, and 113.02 in negative ionization mode [24]. Furthermore, metabolite identification relied on the retention behavior, which helped to eliminate artifacts resulting from in-source fragmenta- tion: as monosulfates eluted considerably before disulfates on the HILIC column, several features with a putative m/z of monosulfates were identified as fragments of actual disul- fates. The software Compound Discoverer initially based the annotation suggestion on our in-house reference library and the databases mzVault, mzCloud, and Chemspider. These annotations were manually double-checked and corrected if required. The Human Metabolome Database HMDB and Metabolomics Workbench were used for the manual anno- tation. Additionally, we annotated some steroid conjugates detected in the direct analysis approach by comparison with the results of the hydrolysis approach. By comparing the pattern among the patient cohort, conclusions were drawn from the respective deconjugated steroids with available reference standards to their individual conjugates. Annota- tion levels from 1 to 4 stating the confidence for annotation [25] were given based on the following criteria: Level 1 - unequivocal matching of spectra and retention times to authentic standards (identified), level 2 - annotation verified using additional techniques, e.g., comparison of patterns among patient cohort (steroid conjugate vs. respective un- conjugated steroid) or matching >80 % of MS2 fragments with mzVault or mzCloud database, level 3 - annotated compound class due to spectral (fragments) or chemical (retention time) properties or >60 % matching of MS2 frag- ments with mzVault or mzCloud database, level 4 - un- known, reproducible MS signal.
Statistical analysis
Statistical analyses were performed using GraphPad Prism 5.0 and IBM SPSS 29.0. Groups were compared by Wilcoxon test or ANOVA with post-hoc Tukey test and log2-fold differences were calculated via the ratio of the median of ACC and ACA groups. Statistical significance was defined as p<0.05. ROC analyses were performed for diagnosis of ACC by using the creatinine-normalized peak areas of the indi- vidual metabolites.
Results
Detection of steroid metabolites
Metabolomics data was acquired in the four detection modes HILIC negative, HILIC positive, RP negative, and RP positive. PCA plots of the samples in the four detection modes are shown in the Supplementary Material, Figure S1. After the initial screening of the detected metabolites in the urine samples of 40 patients with ACC and 40 patients with ACA in all four detection modes, HILIC negative was selected as the most suitable mode for detection of steroid conjugates. This choice was made due to a better ionization of the respective compounds in negative mode and sufficient retention time differences between sulfates, disulfates and glucuronides on the HILIC column, while disulfates were also not detectable in all other modes. RP positive was considered as the best mode for detection of unconjugated steroids after hydroly- sis, as most unconjugated steroids were only detectable after RP chromatography and positive ionization.
Evaluation of hydrolysis
The hydrolysis efficiency study with conjugated and uncon- jugated steroid reference standards spiked to artificial urine revealed incomplete hydrolysis, especially of steroid mono- sulfates, and steroid conversions of 3ß-hydroxy-5-ene steroids (Supplementary Material, Figure S2 and Method S2). When looking for remaining steroid conjugates in hydrolyzed pa- tient samples for evaluation of hydrolysis efficiency, steroid glucuronides were not detectable, suggesting their complete hydrolysis. However, several steroid monosulfates and few disulfates were still detectable after enzymatic treatment as shown in Figure 2 depicting the peak areas of four fully identified steroid sulfates and two disulfates in patient sam- ples untreated or treated with hydrolyzing enzyme. The pro- portion of deconjugation was highly variable for different steroid conjugates as well as for urine samples from different patients while the peak area of all evaluated compounds except for 11-ketoetiocholanolone was reduced on average. DHEAS and 11-ketoetiocholanolone sulfate abundance showed non-significant mean changes of -19.0 % and +9.0 % after hydrolysis, in contrast to significantly changed 16OH-DHEAS (-55.9 %), 17OH-pregnenolone sulfate (-17.0 %), 5-PD disulfate (-80.4%), and androstenediol disulfate (-35.2 %). Especially 17OH-pregnenolone sulfate and androstenediol disulfate were associated with high between-subject variabilities regardless of initial abundance, suggesting no enzyme saturation effects.
16OH-DHEAS
DHEAS -19.0 %
11-ketoetio- cholanolone sulfate +9.0 % (SD 49.2 %)
17OH-pregnenolone sulfate
androstenediol disulfate
5-PD disulfate
-55.9 % (SD 21.1 %)
-17.0 %
(SD 56.5 %)
(SD 110 %)
-80.4 % (SD 13.2 %)
-35.2 % (SD 89.2 %)
ns
**
10 10.
*
10 10
109
ns
109,
10 10.
10 10.
**
Peak Area [cps]
10 9
109
108
108
109
109
108
107
107
108
108
108
107
107
107
106
107
106
10 6.
106
10€
105
106.
105
105
105
105
10
unmodified
hydrolyzed
105
unmodified
hydrolyzed
104
unmodified
hydrolyzed
104
unmodified
hydrolyzed
104
unmodified
hydrolyzed
104
unmodified
hydrolyzed
ACC vs. ACA in unhydrolyzed urine
In total, 1,772 features were detected in the patient samples in HILIC negative mode, with 184 features being significantly increased in ACC compared to ACA. By characteristic frag- ments in MS2 spectra, 74 features were annotated as steroid monosulfates (Supplementary Material, Table S1), as well as 18 steroid disulfates (Supplementary Material, Table S2) and 20 steroid glucuronides (Supplementary Material, Table S3). Besides the molecular formula assigned via the exact mass and characteristic conjugate fragments, the different reten- tion times of the conjugate classes on the HILIC column helped to annotate the respective class, with monosulfates eluting first (0.98-5.47 min), followed by disulfates (6.0- 7.78 min) and glucuronides (7.76-9.10 min). Thereof, 8 monosulfates, 4 disulfates, and 4 glucuronides were fully identified (level 1) via reference standards (Supplementary Material, Figure S3A). A volcano plot of the metabolites detected in unhydrolyzed urine in HILIC negative mode is shown in Figure 3A. Several steroid monosulfates turned out to have the highest log2-fold differences between ACC and ACA groups, followed by some disulfates and glucuronides. In our analysis, we also found the four metabolites detected as most differentiating by Patel et al. [19], but creatine riboside (annotation level 2) was elevated with a log2 fold change of 0.56 in ACC compared to ACA (p=0.006), while L-tryptophan (level 1), NE,NE,Ne-trimethyl-L-lysine (level 1) and 3-methylhistidine (level 2) showed no significant dif- ferences between ACC and ACA groups.
ACC vs. ACA in hydrolyzed urine
Analysis of hydrolyzed urine samples revealed both decon- jugated steroids in RP positive mode as well as remaining steroid conjugates in HILIC negative mode to be most increased in urine of patients with ACC (Figure 3B). In fact, several steroid monosulfates and disulfates, which were not completely deconjugated by the arylsulfatase enzyme showed even higher group differences than most of the deconjugated steroids - these remaining steroid conjugates are not relevant for ACC diagnosis, but this finding supports the possible superiority of a direct analysis of steroid conju- gates compared to hydrolyzed steroids. In RP positive mode, from a total of 2,590 features, 163 features were significantly increased in ACC compared to ACA. Thereof, 63 compounds were annotated as sum of previously unconjugated and deconjugated steroids (Supplementary Material, Table S4) and 18 steroids were fully identified with annotation level 1 after comparison to reference standards (Supplementary Material, Figure S3B and Table S4). In RP positive mode, most unconjugated steroids were detected either as sodium adduct ([M+Na]+) or after water loss ([M-H2O+H]+ or [M-2H2O+H]+), which was considered for the calculation of the suspected molecular formula which was characteristic for steroidal compounds. Additionally, comparison of the MS2 spectra with the Human Metabolome Database and comparison of the retention times on the RP column (5.22-7.57) helped to annotate these compounds as unconjugated steroids (anno- tation level 3 if no reference standard was available).
A
PD-G
17OH-Preg-diS
5-PD-diS
12-
· · 17OH-PregS
10-
Etio-G
16OH-DHEAS
8-
· sulfate
DHEAS
disulfate
glucuronide
6
-Log10 p-value (ACC / ACA)
2
An-S
4
6
8
14
12
DHEA-G
Androstenediol-diS
10
11-Ketoetio-S
8.
6-
4.
2-
0
-4
-2
0
2
4
6
8
Log2 fold difference (ACC / ACA)
17OH-Preg
B
PT
THDOC
80
12-
5-PT
8
17HP
00
10-
P
5-PD
16OH-DHEA
8
☒ deconjugated
☐ conjugate
6
-Log10 p-value (ACC / ACA)
14
2
THS
4
6
8
12.
10-
8.
6-
4.
2-
0
-4
-2
0
2
4
6
8
Log2 fold difference (ACC / ACA)
Diagnostic accuracy
Diagnostic accuracy for ACC was evaluated by ROC analyses for the individual steroid conjugates in untreated urine samples (Figure 4A). Four steroid monosulfates had ROC curves with an area under the curve (AUC) above 0.975 that were annotated with the molecular formulas C21H32O6S (two isomers with AUC of 0.983 and 0.981), and C21H32O7S (two isomers with AUC of 0.981 and 0.976). Of the 10 me- tabolites with highest AUC values in untreated urine, nine were annotated as steroid monosulfates and one was a
disulfate with the molecular formula C21H32OgS2, possibly 21OH-pregnenolone-3,21-disulfate (level 3).
When looking at the total steroids after enzymatic hydrolysis that were detected in RP positive mode, 17OH- pregnenolone had the highest AUC of 0.975, followed by tetrahydro-11-deoxycorticosterone (THDOC) with an AUC of 0.961. 5-PT and THS that were included to the decision- tree-based classification model from our previous study [10] had AUC values of 0.959 and 0.856, respectively. In this detection mode, 21 steroids had an AUC of 0.856 or higher (Supplementary Material, Table S4) among
A
1.0
1.0
1.0
1.0
1.0
Sensitivity
HN-52 C21 H32 O6 S AUC 0.983
Sensitivity
HN-817 C21 H32 06 S
Sensitivity
HN-1368 C21 H32 07 S AUC 0.981
Sensitivity
HN-657
Sensitivity
HN-919
0.5
0.5
0.5
0.5
C21 H32 07 S AUC 0.976
0.5
C21 H34 07 S
AUC 0.981
AUC 0.975
0.0
0.0
0.5
1.0
0.0
0.0
0.0
0.5
1.0
0.0
0.5
1.0
0.0
0.0
0.5
0.0
1.0
0.0
0.5
1.0
1 - Specificity
1 - Specificity
1 - Specificity
1 - Specificity
1 - Specificity
1.0
1.0
1.0
1.0
1.0
Sensitivity
HN-45
Sensitivity
HN-117 C21 H32 09 S2
Sensitivity
HN-695
HN-88
HN-607
0.5
C21 H32 O6 S
0.5
0.5
AUC 0.971
C21 H34 07 S AUC 0.970
Sensitivity
Sensitivity
0.5
C21 H34 O6 S
0.5
C21 H36 06 S
AUC 0.973
AUC 0.968
AUC 0.966
0.0
0.5
0.0
0.5
0.0
0.5
0.0
0.5
0.0
0.0
1.0
0.0
1.0
0.0
1.0
0.0
1.0
0.0
0.5
1.0
1 - Specificity
1 - Specificity
1 - Specificity
1 - Specificity
1 - Specificity
B
1.0
1.0
1.0
1.0
1.0
Sensitivity
RP-615 17OH-Preg AUC 0.975
Sensitivity
RP-212 THDOC
Sensitivity
RP-631 5-PT AUC 0.959
Sensitivity
RP-2393
5-PD AUC 0.951
Sensitivity
RP-4348
0.5
0.5
0.5
0.5
0.5
PD
AUC 0.961
AUC 0.946
0.0
1.0
0.0
0.0
0.0
0.0
0.0
0.5
0.0
0.5
1.0
0.0
0.5
1.0
0.0
0.5
1.0
0.0
0.5
1.0
1 - Specificity
1 - Specificity
1 - Specificity
1 - Specificity
1 - Specificity
1.0
1.0
1.0
1.0
Sensitivity
RP-132 PT AUC 0.945
Sensitivity
RP-1289 17-HP AUC 0.924
Sensitivity
RP-1295
Sensitivity
0.5
0.5
0.5
16OH-DHEA
RP-2226
AUC 0.913
0.5
THS
AUC 0.856
0.0
0.0
0.0
0.0
0.0
0.5
1.0
0.0
0.5
1.0
0.0
0.5
1.0
0.0
0.5
1.0
1 - Specificity
1 - Specificity
1 - Specificity
1 - Specificity
which the nine steroids shown in Figure 4B were fully identified.
Discussion
The present study employed an untargeted metabolomics approach to urine samples from patients with ACC or ACA and artificial urine spiked with conjugated and unconju- gated steroids. Each sample was analyzed with and without enzymatic hydrolysis in order to assess the efficiency of enzymatic hydrolysis of urinary steroid conjugates and to compare steroid conjugates and total steroids after decon- jugation as diagnostic biomarkers for ACC. This untargeted discovery approach following several previously published targeted quantitative analyses of deconjugated steroids is opposed to the commonly performed reversed order - untargeted before targeted - because a direct comprehen- sive analysis of urinary steroid conjugates has not been evaluated for the differentiation of ACC and ACA before.
In this investigation we comprehensively showed that a multitude of different steroid conjugates (monosulfates, disulfates, and glucuronides) in unhydrolyzed urine are highly increased in patients with ACC compared to ACA and therefore bear biomarker potential. Furthermore, we could prove that hydrolysis of all steroid monosulfates was extremely incomplete and variable using arylsulfatase/ glucuronidase derived from H. pomatia and steroid con- versions through enzymatic side activities can be an issue.
Due to non-existing standardized hydrolysis protocols for urinary steroids, the incubation conditions (3 h at 55 ℃) were adopted from our previously published diagnostic study based on deconjugated steroids [10, 26], which are similar to long established deconjugation methods from Shackleton [27] and resemble the conditions from further studies using deconjugated steroids for diagnosis of ACC [9, 11]. Even though a higher deconjugation efficiency for ste- roid sulfates was found for longer incubation time and higher pH, we selected the traditionally used conditions to provide good comparability between the direct analysis approach and the enzymatic hydrolysis approach that was previously used in these published studies. The incomplete hydrolysis of steroid monosulfates with this enzyme type, which was clearly shown within this study, leads to a drastic underestimation of most steroid sulfates. Additionally, we showed the previously described conversion of 3B-hy- droxy-5-ene steroids into 3-keto-4-ene steroids caused by H. pomatia digestive juice [28]. This needs to be taken into account when comparing absolute excretions of deconju- gated steroids among different laboratories. Even though reproducibility within a laboratory is commonly confirmed
during method validation, minimally different incubation conditions would produce different results for several deconjugated steroids in an identical urine sample. There- fore, cutoff values or diagnostic algorithms based on the excretion of total steroids are not transferrable to samples analyzed in other laboratories applying slightly different sample preparation procedures.
We found the previously reported [5-10] most discrim- inative deconjugated steroids for ACC vs. ACA such as THS, 5-PT, 5-PD, PT, and PD in the hydrolyzed samples. Addition- ally, further deconjugated steroids with even higher group differences than the above mentioned steroids were detec- ted, such as 17-OH-pregnenolone, which has been previously found to be increased in plasma or serum of patients with ACC [29, 30], but is commonly not considered in the pub- lished targeted urine assays as most steroid panels rather include its urinary metabolite 5-PT. Moreover, in our anal- ysis THDOC and 16OH-DHEA had high and significant group differences. Both analytes were previously determined in the 32-analyte-panel by Arlt et al. [6] and the 26-analyte panel by Hines et al. [9], but not considered anymore by more recent studies investigating 15 [11] and 11 analytes [10].
This untargeted metabolomics analysis of the unhydro- lyzed urine samples revealed the molecular formulas of the individual steroid conjugates, of which these deconjugated steroids originated from. When only considering the group differences via log2-fold differences of the group medians, the metabolites with the highest group differences were several steroid monosulfates. Likewise, quantitatively, monosulfates turned out to be by far the most of the steroid conjugates, followed by glucuronides and disulfates. Regarding non- steroidal compounds, we confirmed urinary creatine ribo- side to be slightly increased in patients with ACC compared to ACA, however, significance and fold-change were lower than previously reported [19]. The other three metabolites previ- ously reported with a significant decrease in patients with ACC, L-tryptophan, NE,NE,Ne-trimethyl-L-lysine and 3-meth- ylhistine, had no significant group differences in our study, which is in line with the fact that they could not be validated in Patel et al. and might therefore have been false-positives in the training set of that study [19]. Generally, non-steroidal compounds with better or equal discriminative power compared to the steroid metabolites were not found in our analysis.
When looking at the discriminative power (ACC vs. ACA) of deconjugated steroids and steroid conjugates, four un- conjugated steroids after hydrolysis have AUC values above 0.95, indicating very high diagnostic accuracy. In untreated urine samples, four steroid monosulfates have even higher diagnostic accuracy with AUC values above 0.975. Never- theless, the hydrolysis approach is certainly justified for the
differential diagnosis of adrenal tumors for two main rea- sons: (1) several studies have previously shown a good diagnostic accuracy for ACC based on the analysis of urinary steroids after enzymatic hydrolysis [6-11]. (2) The quantita- tive analysis of the - possibly more accurate - individual steroid conjugates in urine is currently still hampered due to the lack of commercially available reference standards. Therefore, the indirect analysis via the deconjugated ste- roids represents the currently best-available alternative, as their reference standards are more easily available. To at least partially overcome the aforementioned challenges when measuring deconjugated compounds it could be a practical approach to use pooled quality control samples to monitor reproducible and consistent deconjugation in each analytical run. Yet it is uncertain whether the incomplete hydrolysis already taps the full diagnostic potential of uri- nary steroids.
Our study has several limitations. With the applied method, only a semiquantitative analysis could be per- formed and an absolute quantification of analytes was not possible. This is common for untargeted metabolomics, which is generally the first step in biomarker discovery, in order to generate a first hypothesis that subsequently re- quires validation in a targeted approach. For the evaluation of hydrolysis efficiency, the peak areas of steroid conjugates were compared in hydrolyzed and non-hydrolyzed urine samples which might be affected by different recoveries and/ or matrix effects between both types of extracts and lead to a bias in the comparison of peak areas. Nevertheless, the general problem of incomplete enzymatic hydrolysis, namely that steroid conjugates are still detectable in high abundance after hydrolysis is well visible with this data. Another limitation of our study is the fact, that the untar- geted metabolomics method was not sufficiently sensitive to detect unconjugated steroids in non-hydrolyzed samples. These metabolites are present in very low concentrations in urine and the method was not tailored to the analysis of their low concentrations as it was generally developed to capture a broad spectrum of all kinds of metabolites. Therefore, we cannot draw conclusions regarding their relevance for ACC diagnosis. Another methodological challenge was the chro- matographic separation of the various isomeric steroids which are present in all classes of steroid metabolites. For example, androsterone sulfate and etiocholanolone sulfate could not be separated with the applied gradient elution and were therefore assessed as the sum of the two metabolites. We cannot exclude this to be the case for other metabolites as well, which should be kept in mind when transferring this approach to a different chromatographic method. Moreover, the sample size was limited by the rarity of ACC and the duration of metabolomics analyses. Furthermore, a
complete identification based on reference standards was only possible for some steroid conjugates, nevertheless, the detected compounds serve as a good starting point for deeper investigations, as the molecular formula and conju- gate type were identified for most metabolites with high confidence. The analysis of steroid conjugates remains challenging due to the lack of available reference standards that are essential for a definite identification in untargeted approaches and quantitative analysis. Even though the se- lective synthesis of steroid conjugates is possible [31-34] and within this study, three promising steroid disulfates were synthesized for identification, feasibility is extremely costly and cumbersome for all possible conjugates. This hampers the identification of most discriminative metabolites and the development of targeted quantification methods for verifi- cation and clinical validation in larger patient cohorts.
To conclude, this study gives insights into the multitude of excreted steroid conjugates in urine and their diagnostic utility. Moreover, the incomplete and variable hydrolysis of most steroid sulfates as well as steroid conversions by a H. pomatia-derived enzyme mix are shown. The hydrolysis process in general, therefore, requires further research to reach quantitative deconjugations, e.g., by testing chemical hydrolysis [28], a combination of deglucuronidation by E. coli glucuronidase followed by solvolysis [12], and/or an SPE extraction before enzymatic hydrolysis [35]. The inves- tigation of alternative enzymes and the batch-to-batch variability of wild-type enzymes should be tested as part of further research under consideration of a higher number of steroids covering the various stereochemistries of steroidal compounds. Beyond, we propose a deeper investigation of the field of steroid conjugates by the direct quantification of the most suitable steroid conjugates in urine and plasma. Until now, only few LC-MS methods for the targeted quan- tification of several steroid glucuronides [36, 37], sulfates [38-40], or both [41, 42] have been published. A practical advantage thereof is the fact that the time-consuming and laborious process of hydrolysis could be omitted. This direct approach however presupposes the availability of the respective reference standards. This gap will hopefully be closed soon by simple synthesis strategies that would enable further and more detailed research on this rather unex- plored field of steroid conjugates.
Research ethics: This study was conducted in accordance with the Declaration of Helsinki (as revised in 2013) and has been approved by the Ethics Committee of the Julius- Maximilians University Würzburg and the Ethics Committee of the Friedrich-Alexander Universität Erlangen-Nürnberg. Informed consent: Informed consent was obtained from all individuals included in this study.
Author contributions: The authors have accepted re- sponsibility for the entire content of this manuscript and approved its submission.
Use of Large Language Models, AI and Machine Learning Tools: None declared.
Conflict of interest: The authors state no conflict of interest. Research funding: This study was funded by the Interdis- ciplinary Center for Clinical Research (IZKF) at the Univer- sity Hospital of the University of Erlangen-Nuremberg (ELAN pilot project P134 and junior project J115). It was also supported by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) (CRC/Transregio 205/2). The Orbitrap mass spectrometer was funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Founda- tion) - INST 90/1048-1 FUGG.
Data availability: The raw data can be obtained on reason- able request from the corresponding author.
References
1. Young WF, Jr. The incidentally discovered adrenal mass. N Engl J Med 2007;356:601-10.
2. Fassnacht M, Tsagarakis S, Terzolo M, Tabarin A, Sahdev A, Newell- Price J, et al. European Society of Endocrinology clinical practice guidelines on the management of adrenal incidentalomas, in collaboration with the European Network for the Study of Adrenal Tumors. Eur J Endocrinol 2023;189:G1-42.
3. Dinnes J, Bancos I, Ferrante di Ruffano L, Chortis V, Davenport C, Bayliss S, et al. Management of endocrine disease: imaging for the diagnosis of malignancy in incidentally discovered adrenal masses: a systematic review and meta-analysis. Eur J Endocrinol 2016;175:R51-64.
4. Rossi C, Cicalini I, Verrocchio S, Di Dalmazi G, Federici L, Bucci I. The potential of steroid profiling by mass spectrometry in the management of adrenocortical carcinoma. Biomedicines 2020;8. https://doi.org/10. 3390/biomedicines8090314.
5. Bancos I, Arlt W. Steroid biomarkers and the diagnosis of adrenal cortical carcinoma. Curr Opin Endocrinol Diabetes Obes 2019;8:167-73.
6. Arlt W, Biehl M, Taylor AE, Hahner S, Libe R, Hughes BA, et al. Urine steroid metabolomics as a biomarker tool for detecting malignancy in adrenal tumors. J Clin Endocrinol Metab 2011;96:3775-84.
7. Kerkhofs TM, Kerstens MN, Kema IP, Willems TP, Haak HR. Diagnostic value of urinary steroid profiling in the evaluation of adrenal tumors. Horm Cancer 2015;6:168-75.
8. Velikanova LI, Shafigullina ZR, Lisitsin AA, Vorokhobina NV, Grigoryan K, Kukhianidze EA, et al. Different types of urinary steroid profiling obtained by high-performance liquid chromatography and gas chromatography-mass spectrometry in patients with adrenocortical carcinoma. Horm Cancer 2016;7:327-35.
9. Hines JM, Bancos I, Bancos C, Singh RD, Avula AV, Young WF, et al. High- resolution, accurate-mass (HRAM) mass spectrometry urine steroid profiling in the diagnosis of adrenal disorders. Clin Chem 2017;63: 1824-35.
10. Vogg N, Müller T, Floren A, Dandekar T, Riester A, Dischinger U, et al. Simplified urinary steroid profiling by LC-MS as diagnostic tool for malignancy in adrenocortical tumors. Clin Chim Acta 2023;543:117301.
11. Bancos I, Taylor AE, Chortis V, Sitch AJ, Jenkinson C, Davidge-Pitts CJ, et al. Urine steroid metabolomics for the differential diagnosis of adrenal incidentalomas in the EURINE-ACT study: a prospective test validation study. Lancet Diabetes Endocrinol 2020;8:773-81.
12. Gomes RL, Meredith W, Snape CE, Sephton MA. Analysis of conjugated steroid androgens: deconjugation, derivatisation and associated issues. J Pharm Biomed Anal 2009;49:1133-40.
13. Moon JY, Ha YW, Moon MH, Chung BC, Choi MH. Systematic error in gas chromatography-mass spectrometry-based quantification of hydrolyzed urinary steroids. Cancer Epidemiol Biomarkers Prev 2010;19:388-97.
14. Robles J, Marcos J, Renau N, Garrostas L, Segura J, Ventura R, et al. Quantifying endogenous androgens, estrogens, pregnenolone and progesterone metabolites in human urine by gas chromatography tandem mass spectrometry. Talanta 2017;169:20-9.
15. Mueller JW, Vogg N, Lightning TA, Weigand I, Ronchi CL, Foster PA, et al. Steroid sulfation in adrenal tumors. J Clin Endocrinol Metab 2021;106: 3385-97.
16. Lightning TA, Gesteira TF, Mueller JW. Steroid disulfates - sulfation double trouble. Mol Cell Endocrinol 2021;524:111161.
17. Sun N, Kunzke T, Sbiera S, Kircher S, Feuchtinger A, Aichler M, et al. Prognostic relevance of steroid sulfation in adrenocortical carcinoma revealed by molecular phenotyping using high-resolution mass spectrometry imaging. Clin Chem 2019;65:1276-86.
18. Burton C, Ma Y. Current trends in cancer biomarker discovery using urinary metabolomics: achievements and new challenges. Curr Med Chem 2019;26:5-28.
19. Patel D, Thompson MD, Manna SK, Krausz KW, Zhang L, Nilubol N, et al. Unique and novel urinary metabolomic features in malignant versus benign adrenal neoplasms. Clin Cancer Res 2017;23:5302-10.
20. Fassnacht M, Dekkers OM, Else T, Baudin E, Berruti A, de Krijger R, et al. European society of endocrinology clinical practice guidelines on the management of adrenocortical carcinoma in adults, in collaboration with the European Network for the study of adrenal tumors. Eur J Endocrinol 2018;179:G1-46.
21. Fassnacht M, Assie G, Baudin E, Eisenhofer G, de la Fouchardiere C, Haak HR, et al. Adrenocortical carcinomas and malignant phaeochromocytomas: ESMO-EURACAN Clinical Practice Guidelines for diagnosis, treatment and follow-up. Ann Oncol 2020;31:1476-90.
22. Mann SJ, Gerber LM. Addressing the problem of inaccuracy of measured 24-hour urine collections due to incomplete collection. J Clin Hypertens 2019;21:1626-34.
23. Xiang A, Nourian A, Ghiraldi E, Friedlander JI. Improving compliance with 24-H urine collections: understanding inadequacies in the collection process and risk factors for poor compliance. Curr Urol Rep 2021;22:38.
24. Gomez C, Fabregat A, Pozo ÓJ, Marcos ], Segura J, Ventura R. Analytical strategies based on mass spectrometric techniques for the study of steroid metabolism. TrAC, Trends Anal Chem 2014;53:106-16.
25. Reisdorph NA, Walmsley S, Reisdorph R. A perspective and framework for developing sample type specific databases for LC/MS-based clinical metabolomics. Metabolites 2019;10:8.
26. Vogg N, Müller T, Floren A, Dandekar T, Scherf-Clavel O, Fassnacht M, et al. Targeted metabolic profiling of urinary steroids with a focus on analytical accuracy and sample stability. J Mass Spectrom Adv Clin Lab 2022;25:44-52.
27. Shackleton CH. Mass spectrometry in the diagnosis of steroid-related disorders and in hypertension research. J Steroid Biochem Mol Biol 1993;45:127-40.
28. Iannone M, Botrè F, Martinez-Brito D, Matteucci R, de la Torre X. Development and application of analytical procedures for the GC-MS/
MS analysis of the sulfates metabolites of anabolic androgenic steroids: the pivotal role of chemical hydrolysis. J Chromatogr B 2020;1155:122280.
29. Taylor DR, Ghataore L, Couchman L, Vincent RP, Whitelaw B, Lewis D, et al. A 13-steroid serum panel based on LC-MS/MS: use in detection of adrenocortical carcinoma. Clin Chem 2017;63:1836-46.
30. Suzuki S, Minamidate T, Shiga A, Ruike Y, Ishiwata K, Naito K, et al. Steroid metabolites for diagnosing and predicting clinicopathological features in cortisol-producing adrenocortical carcinoma. BMC Endocr Disord 2020;20:1-9.
31. Okihara R, Mitamura K, Hasegawa M, Mori M, Muto A, Kakiyama G, et al. Potential corticoid metabolites: chemical synthesis of 3- and 21- monosulfates and their double-conjugates of tetrahydrocorticosteroids in the 5a- and 50-series. Chem Pharm Bull 2010;58:344-53.
32. Waller CC, McLeod MD. A simple method for the small scale synthesis and solid-phase extraction purification of steroid sulfates. Steroids 2014;92:74-80.
33. Pranata A, Fitzgerald CC, Khymenets O, Westley E, Anderson NJ, Ma P, et al. Synthesis of steroid bisglucuronide and sulfate glucuronide reference materials: unearthing neglected treasures of steroid metabolism. Steroids 2019;143:25-40.
34. Alshehri JA, Gill DM, Jones AM. A Sulfuryl group transfer strategy to selectively prepare sulfated steroids and isotopically labelled derivatives. Front Mol Biosci 2021;8:776900.
35. Gomes RL, Meredith W, Snape CE, Sephton MA. Conjugated steroids: analytical approaches and applications. Anal Bioanal Chem 2009;393: 453-8.
36. Wang R, Hartmann M, Wudy S. Targeted LC-MS/MS analysis of steroid glucuronides in human urine. J Steroid Biochem Mol Biol 2021;205:105774.
37. Pozo OJ, Van Eenoo P, Van Thuyne W, Deventer K, Delbeke FT. Direct quantification of steroid glucuronides in human urine by liquid chromatography-electrospray tandem mass spectrometry. J Chromatogr A 2008;1183:108-18.
38. Galuska CE, Hartmann MF, Sánchez-Guijo A, Bakhaus K, Geyer J, Schuler G, et al. Profiling intact steroid sulfates and unconjugated steroids in biological fluids by liquid chromatography-tandem mass spectrometry (LC-MS-MS). Analyst 2013;138:3792-801.
39. Mitamura K, Satoh R, Kamibayashi M, Sato K, Iida T, Ikegawa S. Simultaneous determination of 18 tetrahydrocorticosteroid sulfates in human urine by liquid chromatography/electrospray ionization- tandem mass spectrometry. Steroids 2014;85:18-29.
40. Esquivel A, Alechaga É, Monfort N, Ventura R. Direct quantitation of endogenous steroid sulfates in human urine by liquid chromatography-electrospray tandem mass spectrometry. Drug Test Anal 2018;10:1734-43.
41. Badoud F, Grata E, Boccard J, Guillarme D, Veuthey J-L, Rudaz S, et al. Quantification of glucuronidated and sulfated steroids in human urine by ultra-high pressure liquid chromatography quadrupole time-of-flight mass spectrometry. Anal Bioanal Chem 2011;400: 503-16.
42. Strahm E, Kohler I, Rudaz S, Martel S, Carrupt P-A, Veuthey J-L, et al. Isolation and quantification by high-performance liquid chromatography-ion-trap mass spectrometry of androgen sulfoconjugates in human urine. J Chromatogr A 2008;1196:153-60.
Supplementary Material: This article contains supplementary material (https://doi.org/10.1515/cclm-2024-1337).