Notice: Wiley Online Library will be unavailable on Saturday 30th July 2016 from 08:00-11:00 BST / 03:00-06:00 EST / 15:00-18:00 SGT for essential maintenance. Apologies for the inconvenience.
The copyright line for this articlewas changed on 26 February 2015 after original online publication.
In sports drug testing, comprehensive studies on the metabolism of therapeutic agents with misuse potential are necessary to identify metabolites that provide utmost retrospectivity and specificity. By commonly employed approaches minor and/or long-term metabolites in urine might remain undetected. Hence, an alternative strategy to unambiguously identify the majority of urinary metabolites including low-abundance representatives is desirable.
Urine samples were collected for 20 days during an elimination study with an oral dose of 5 mg of 17α-C2H3-metandienone. The specimens were processed according to established sample preparation procedures (including fractionation and deconjugation) and subjected to gas chromatography/hydrogen isotope ratio mass spectrometry (GC/IRMS) analysis. Due to the deuteration of the administered drug, urinary metabolites bearing the deuterium label yield abundant and specific signals on the GC/IRMS instrument resulting from the substantially altered 2H/1H ratio. The sample aliquots were measured by gas chromatography/time-of-flight (GC/Q-TOF) mass spectrometry using identical GC conditions, allowing high-resolution/high-accuracy mass data to be obtained on all urinary metabolites previously identified by IRMS.
Within the IRMS chromatograms, labeled metabolites were identified up to 20 days after administration at urinary concentration down to 0.25 ng/mL. More than 50 metabolites were observed with the earlier described long-term metabolite of metandienone, 18-nor-17β-hyroxymethyl,17α-methyl-androst-1,4,13-trien-3-one, being the most prominent glucuronidated metabolite in the studied time window. In the sulfoconjugated steroids fraction, a yet unknown metabolite was observed at m/z 283.1997 comprising the experimentally determined elemental composition of C20H212H3O.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Most substances relevant for sports drug testing undergo extensive phase 1 and phase 2 metabolism before being renally excreted.[1, 2] Consequently, only small amounts (if any) of the intact therapeutic agent are recovered from urine, and metabolites are often excreted for a longer period of time than the administered substance. Hence, comprehensive studies on the metabolism of therapeutic agents with misuse potential are necessary to identify those metabolites that enable the detection of drug abuse with utmost retrospectivity and specificity.
Early approaches in metabolism studies used radioactively labeled compounds (3H or 14C) and scintillation counters to detect the compounds of interest. Due to the considerable health risks associated with the radioactive decay of drugs administered to the study subjects, the alternative of using stable isotopes was discussed as early as the 1960s. Commonly employed approaches of drug metabolite detection and identification rely on gas or liquid chromatography/mass spectrometry and the comparison of measured excretion patterns in urine samples collected before and after the administration of a (labeled) therapeutic compound.[4-8] Such methodologies allow the detection of a great number of main metabolic products; however, they are also time-consuming and minor metabolites might remain undetected. In particular these minor metabolites (usually observed at low ng/mL concentrations) have received much attention recently particularly in doping controls as they frequently represent valuable target analytes for the long-term detection of drug abuse.
Over 20 years ago, isotope ratio mass spectrometry (IRMS) or chemical reaction interface mass spectrometry was tested for metabolite identification studies.[9-13] The detection limits for metabolites at that time were approximately 100 ng/mL in urine and, thus, sufficient for selected clinical applications but not for sports drug testing. With the introduction of a commercially available continuous-flow IRMS system able to measure hydrogen isotope ratios (HIR) at natural abundance in 1999, research activities increased and systems were constantly improved. At present, the required amount of an analyte for the determination of natural HIR is approximately 50 ng 'on-column'. As the natural occurrence of 2H is only 1 out of 6500 1H atoms (0.015%), the use of deuterated compounds for metabolism studies would substantially increase the signal intensities of analytes bearing approximately 10% of deuterium atoms. Consequently, concentrations of a labeled metabolite in the sub-ng/mL range will result in a detectable signal above the naturally present 2H-background in IRMS analyses.
This hypothesis was tested with an elimination study performed with trideuterated metandienone (17α-C2H3-androsta-1,4-dien-17β-ol-3-one, 2H3MD). This particular steroid was chosen as its metabolism has been described in detail and the verified metabolites should cover the majority of signals expected in the current project.[7, 16-18] In order to allow for utmost coverage of potential metabolites, glucuronidated, sulfoconjugated and unconjugated steroids were investigated in this study. For gas chromatography/thermal conversion isotope ratio mass spectrometry (GC/TC-IRMS) measurements all analytes are reduced to hydrogen gas and, consequently, all structural information is lost. However, after locating labeled metabolites in the IRMS chromatogram, aliquots were subjected to high-resolution/high-accuracy gas chromatography/quadrupole/time-of-flight mass spectrometry (GC/Q-TOF MS) operated under identical GC conditions to the IRMS system. Matching of the two sets of measurements was additionally ensured by applying retention time markers.
Chemicals and steroids
Chromabond® C18 cartridges (500 mg, 6 mL) were obtained from Macherey & Nagel (Düren, Germany). Pyridine, acetic anhydride (distilled before use), glacial acetic acid, ethyl acetate, and cyclohexane were purchased from Merck (Darmstadt, Germany). tert-Butyl methyl ether (TBME, distilled before use) was from KMF Laborchemie (St.Augustin, Germany), and β-glucuronidase from Escherichia coli from Roche Diagnostics GmbH (Mannheim, Germany). All solvents and reagents were of analytical grade.
Steroid reference materials [5α-androst-16-en-3α-ol (16EN), 4-hydroxy-androst-4-ene-3,17-dione (F), 3α-hydroxy-5α-androstan-17-one (ANDRO), 3α-hydroxy-5β-androstan-17-one (ETIO), 3α,11β-dihydroxy-5α-androstan-17-one (11OHA), and 5β-pregnane-3α,20α-diol (PD)] were supplied by Sigma (Steinheim, Germany) and metandienone (17α-methyl-androst-1,4-dien-17β-ol-3-one, MD) was purchased from Fluka (Buchs, Switzerland). All other reference steroids were synthesized in-house and characterized as described elsewhere.[7, 16-18] These include 6β-hydroxymetandienone (6-OH-MD), 18-nor-17β-hydroxymethyl,17α-methylandrost-1,4,13-triene-3-one (NW), 17β-methyl-androst-1,4-dien-17α-ol-3-one (EPI-MD), 17β-methyl-5β-androst-1-ene-3α,17α-diol (EPI-DIOL), 18-nor-17,17-dimethylandrost-1,4,13-triene-3-one (18NOR), and 2H3MD (for chemical structures, see Fig. 1).
A slightly modified sample preparation method for routine doping control analyses of steroids was used in this study.[19, 20] In brief, 20 mL of urine were applied to two conditioned solid-phase extraction (SPE) cartridges, washed with water, and eluted with methanol. After evaporation to dryness the residue was reconstituted in 2 mL of sodium phosphate buffer and extracted with 5 mL of TBME. Following shaking and centrifugation, the organic phase was collected and evaporated to dryness, containing the fraction of unconjugated steroids. To the aqueous residue of the extracted sample, β-glucuronidase was added to enzymatically hydrolyze glucuronic acid conjugates by incubating the mixture for 1 h at 50°C. The liberated steroids were extracted with 5 mL of TBME, the organic phase was separated and evaporated and contained the fraction of formerly glucuronidated steroids. The remaining aqueous residue was subjected to another SPE step, and the retained components were eluted with methanol/ethyl acetate (30:70) and incubated for 1 h at 50°C after adding ethyl acetate/sulfuric acid (100 mL/200 ng). After adding 0.5 mL of 1 M sodium hydroxide in methanol the sample was evaporated and reconstituted with 5 mL of water, and extracted with 5 mL of TBME. The organic layer now containing the formerly sulfoconjugated steroids was collected and evaporated. All extracts with deconjugated steroids were subsequently subjected to cleanup by high-performance liquid chromatography (HPLC).
In order to further purify the steroidal extracts, fractionation was performed on a 1100 HPLC system (Agilent, Waldbronn, Germany) equipped with a LiChrospher® 100 RP18 analytical column (250 × 4 mm, 5 µm particle size; Merck). The dried residues were reconstituted in 50 μL of acetonitrile/water (50:50 v/v) and the entire volume was injected into the HPLC instrument. The gradient used was linear from 25:75 acetonitrile/water to 100% acetonitrile in 20 min, isocratic for 10 min, followed by 5 min of re-equilibration with a flow rate of 1 mL/min at starting conditions. Fraction collection times were defined by a two-fold injection of a standard mixture containing 11OHA, ANDRO, ETIO, PD and 16EN (endogenous reference steroids) and 6-OH-MD, NW, EPI-MD, EPI-DIOL, and 18NOR as metabolites of interest. Both peak shape and retention time were monitored using UV detection at 195 nm. Manual fraction collection was performed as listed in Table 1. Highly abundant endogenous steroids such as ETIO and ANDRO or 11OHA were collected separately from the expected (low-abundance) compounds. This allowed for the use of adequate volumes of reconstitution solvents for each HPLC fraction and subsequent GC/TC-IRMS measurements, i.e. increased sensitivity was accomplished as the amount of reconstitution solvent could be kept to a minimum. All the collected fractions were evaporated under a stream of nitrogen and acetylated by adding 50 μL of acetic anhydride and pyridine and incubation for 45 min at 70°C. The samples were dried, reconstituted with cyclohexane and transferred to autosampler vials.
Table 1. Fraction collection times for HPLC cleanup and expected analytes
G – glucuronidated, S – sulfoconjugated.
Detection of deuterium-labeled metabolites was conducted on a Delta V Plus isotope ratio mass spectrometer operated in positive electron ionization (EI) mode and coupled to a Trace GC 1310 gas chromatograph equipped with a TriPlus RSG Autosampler (all from Thermo Fisher, Bremen, Germany). Both systems were connected via the GC IsoLink CNH operated at 1450°C and the ConFlo IV interface (both Thermo Fisher). Injections were performed in splitless mode at 280°C with 2 μL of cyclohexane. The GC column was a J&W Scientific DB-17MS (length 30 m, i.d. 0.25 mm, film thickness 0.25 µm) from Agilent (Waldbronn, Germany). The initial oven temperature was 60°C, maintained at 60°C for 2 min, increased at 40°C/min to 250°C, 5°C/min to 300°C, and finally held at 300°C for 4 min before cool-down to initial conditions. The carrier gas was He (purity grade 4.9) with a constant flow rate of 1.2 mL/min. Isodat 3.0 sotware (Thermo Fisher) was used for data acquisition and evaluation.
GC/Q-TOF instrument setup
For structural identification, each aliquot was re-injected into an Agilent 7200 Accurate-Mass Q-TOF system coupled to an Agilent 7890A gas chromatograph (Santa Barbara, CA, USA). Injections were performed in splitless mode at 280°C with an injection volume of 2 μL. The same GC column and an identical temperature program to that mentioned above were used, as were the same flow rate and carrier gas. The EI ion source was operated at 40 eV and a temperature of 250°C. The scan range of m/z 50–500 was covered at a scan speed of 200 ms/spectrum. For data acquisition and evaluation Mass Hunter software (version B.06; Agilent) was used.
Matching of analyses
In order to account for inevitable (minor) differences in retention times on the two GC systems, i.e. GC/TC-IRMS and GC/Q-TOF, retention time markers were employed. As an early eluting marker compound, 16EN was chosen, whereas F was selected as the late eluting species. Samples were reconstituted in a cyclohexane solution containing 10 µg/mL of each steroid. As the endogenous HIR values were not of interest in this study, the contribution of exogenous 16EN to the endogenous counterpart was not of concern. To match retention times and the chromatograms obtained from the two GC/MS systems, a linear relationship was assumed.
Administration study urine samples
A detailed description of the ethically approved administration study was published earlier. In brief, one healthy male volunteer (54 years, 80 kg) administered 5 mg of 2H3MD orally and collected 28 urine samples over a period of 20 days. All samples were stabilized with sodium azide and stored initially at +4°C and then frozen at −18°C.
RESULTS AND DISCUSSION
Metabolite identification with GC/TC-IRMS
Of the 28 samples collected after 2H3MD administration, a subset of 10 samples was chosen for this study covering the entire period of 20 days. In Fig. 2, a chromatogram of the blank urine is depicted showing the LC fraction V, which contained mainly the endogenous steroid PD. For HIR measurements at natural abundance, the ion chromatogram at m/z 2 shows consistently higher intensities than the corresponding chromatogram at m/z 3, although the abundance of m/z 3 is constantly amplified by a factor of 1000, facilitating the interpretation and illustration of both readouts. The quotient of both mass-to-charge ratios (i.e. the 'quotient ion chromatograms' generated by plotting m/z 3/m/z 2; Fig. 2, top) at natural abundance yields a sigmoid shape for peaks due to the isotopic fractionation occurring on the GC column.[14, 21] Depending on the chosen offset in the Isodat software (here 100/440 for m/z 3/m/z 2) the absolute ratio is found in the range of 0.2 to 0.4. The measured δ2HVSMOW value for the endogenous PD was −247 ‰ and this fits perfectly into the distribution of naturally occurring values for steroids. Following the application of 5 mg of 2H3MD, the obtained chromatograms of the LC fraction V (collected 22 h post-administration) are substantially different from the corresponding blank sample, as illustrated in Fig. 3. Regarding m/z 2, PD remains the most intense signal at 1095 s; however, the entire chromatogram is now dominated by peaks visible in the ion chromatogram of m/z 3. Here, urinary metabolites of the labeled metandienone result in extremely intense signals due to their exceptional deuterium content compared with natural values. This, of course, is directly reflected by the quotient ion chromatogram (Fig. 3, top) showing values of 25 and higher. In addition to the two main metabolites within this LC fraction at 682 s and 900 s, other metabolites of lower abundance are immediately observed by visual inspection of the chromatograms.
Combining GC/TC-IRMS and GC/Q-TOF MS
Each aliquot injected into the IRMS system was re-injected on a GC/Q-TOF MS instrument in order to perform structural elucidation and/or provide information on the elemental composition of the detected metabolite of the labeled metandienone. In Fig. 4, the total ion chromatograms (TICs) of the same LC fraction V sampled before and after administration are depicted. In contrast to the IRMS chromatograms shown in Fig. 3, only the period from 660 s to 930 s is illustrated for clarity purposes. Signal a) (retention time = 682 s in the GC/TC-IRMS run) represents the major metabolite of this fraction and time point of sample collection, readily identified in the IRMS chromatogram and the TIC data. Peaks b) and c), however, are easily misinterpreted as metabolites when comparing pre- and post-administration TIC data as the corresponding IRMS data do not support the presence of deuterium atoms in these compounds. In contrast, peak d), that corresponds to the GC/TC-IRMS peak at 900 s, yielded a clear metabolite signal due to its deuterium content in the IRMS chromatograms but might not have been recognized using the TIC data only. Compound (d) was confirmed as 2H3-EPI-DIOL mono-acetate by means of available reference material.
From the variety of signals visible in the IRMS chromatogram, seven substances were verified as deuterated steroid metabolites by means of high-accuracy/high-resolution mass spectrometric data. The employed derivatization strategy (i.e. acetylation) commonly resulted in modest abundances of the molecular ions; hence, structure confirmation was accomplished by means of prominent fragment ions bearing the 2H3-label with experimentally determined elemental compositions as listed in Table 2. For metabolites found at sufficient concentrations or with very low interferences from other urinary matrix components, full scan mass spectra were obtained, one of which is shown in Fig. 5, representing the prominent peak at 682 s. Its elemental composition was determined as C22H292H3O2 and its physicochemical properties such as retention times on reversed-phase LC (late eluting compound) and GC (early eluting compound) suggest the presence of a mono-acetylated 18-nor-metabolite. In total, more than 50 urinary metabolites of 2H3MD were detected in the fractions of glucuronidated steroids using the reported approach of combined GC/TC-IRMS and /Q-TOF analyses. Structural elucidation of this extensive amount of mostly unknown metabolites necessitates an improved sample cleanup and the use of different derivatization techniques (e.g. trimethylsilylation) to obtain additional structural information on the molecular ion or prominent fragments, and eventually chemical synthesis of the postulated substance. These goals are, however, beyond the scope of the present proof-of-concept study and are the subject of future work.
Table 2. Elemental compositions and m/z values of three-fold deuterium-labeled fragment ions used for verification of metandienone metabolites following acetylation and GC/EI-Q-TOF analysis
Fragment ion (m/z)
Specificity of the approach
As mentioned above, matrix interference (i.e. co-eluting compounds) can confound the identification of labeled metabolites using conventional mass spectrometry-based methodologies. In some cases highly abundant co-eluting species can render the detection of a metabolite impossible. These limitations do not exist for the IRMS approach, as illustrated in Fig. 6 for LC fraction VI containing mainly the endogenous species 16EN. Immediately after 16EN (and before another endogenous steroid with the very common composition of C19H28O) the metandienone metabolite 2H3- 18NOR is eluted. Employing only the TIC of the GC/Q-TOF chromatogram, there is no indication for the presence of a metabolite due to the massive overload of the MS TIC signal caused by 16EN, preventing the detection of any of the diagnostic fragment ions summarized in Table 2. In contrast, the IRMS data readily allows detection of this metabolite up to 7 days post-administration as a long-term metabolite of metandienone. Thus, the 'sniffer dog' character of the GC/TC-IRMS system for detecting isotope-labeled metabolites in a highly complex matrix is clearly demonstrated.
Sensitivity of the GC/TC-IRMS measurements
In sports drug testing the detection window of a compound is of major interest as many substances such as steroidal agents are prone to misuse in out-of-competition periods and their administration is commonly discontinued weeks before competition and scheduled drug tests. Enhancing the retrospectivity of a method for a distinct doping agent immediately results in increased numbers of adverse analytical findings.[23, 24] Consequently, the identification of long-term metabolites is of great relevance in sports drug testing and the approach presented herein enables the rapid and sensitive detection of such long-term metabolites as demonstrated for NM, the long-term marker for metandienone administration. One of the main benefits of this metabolite is its continuous presence in urine samples from 8 h up to 20 days after drug administration. As depicted in Fig. 7, it is excreted into urine at a concentration of approximately 20 ng/mL 6 days after ingestion of metandienone and is readily detected by GC/TC-IRMS analysis. Even 20 days after drug application, a urinary concentration of only 0.25 ng/mL is sufficient to generate an unambiguous signal in the quotient ion chromatogram (Fig. 7, right). Taking into account that 20 mL of urine were processed, that the overall recovery of this sample preparation procedure is approximately 60%, and that only one-sixth of the sample was transferred onto the GC column for IRMS determination, it is concluded that less than 1 ng of labeled substance on-column can be detected with this novel approach in metabolism studies. Modern continuous-flow IRMS is thus definitely capable of detecting metabolites at concentrations relevant for sports drug testing laboratories.
Unconjugated and sulfated steroids
In the fraction of unconjugated steroids, 12 labeled metabolites were detected (Fig. 8). As expected, these metabolites were mainly found at the beginning of the administration study, and after 6 days only one was still present and this was identified as EPI-MD (number 11 in Fig. 8). This finding is in agreement with reported excretion patterns and the origin of the compound was attributed to the urinary degradation of MD excreted as a sulfoconjugate rather than being an authentic metabolite.
Another unconjugated steroid was 6-OH-MD represented by peak number 12, which is considered an authentic urinary metabolite of metandienone. All other signals were attributed to metandienone metabolites as identified by accurate masses of diagnostic (fragment) ions as listed in Table 2: Peaks 1 and 2 contain ions with an elemental composition of C20H252H3O, peaks 3, 4, 5 and 10 with C20H232H3O, and peaks 6 and 8 with C20H212H3O corresponding to reduction reactions. Moreover, peak 10 was found to generate an ion with the composition C20H252H3O2, thus arguably resulting from the hydroxylation of metandienone. It is noteworthy that these fragment ions originate from acetylated species of metandienone and its metabolic products. As the hydroxyl function at C-17 is not derivatized under the chosen reaction conditions, eliminations of acetic acid (60 Da), as observed in most of the above-reported analytes referred to as peaks 1–10, is attributed to reduction of the 3-oxo group (peaks 1–5, 6, and 8) or hydroxylation (peak 10). A closer examination of Fig. 8 reveals several additional minor signals between or co-eluting with the highlighted ones. These probably represent deuterated metabolites but as unambiguous identification via GC/Q-TOF MS was not accomplished these metabolites were not taken further into account.
In the fraction of sulfated steroids another 15 metabolites were observed, some of which were present also as glucuronidated analogs. A high number of hydroxylated steroids were excreted as sulfates, among which the long-term metabolite NW (Fig. 1) was also detected with a comparable detection window of 20 days and at similar concentrations. In addition, another as yet unknown metabolite was found in this fraction; it eluted shortly before NW and was also detectable for 20 days after administration. The structure of the 17-epimer of NW was disproved by the experimental mass of m/z 283.1997 and an elemental composition of C20H212H3O with a mass error of 5.67 ppm and further research is ongoing to elucidate the nature of this metabolite.
In this study the value of GC/TC-IRMS in metabolism studies was demonstrated, the utility of which is not limited to sports drug testing applications. The major issue of this approach as identified 20 years ago was the limited sensitivity; however, modern state-of-the-art instrumentation allows the detection of labeled metabolites injected onto the IRMS system with less than 1 ng on-column as shown in this proof-of-concept project. The combination of this sensitive IRMS technique with comprehensive sample preparation enables the detection of main and minor metabolites as long as the deuterium label of the administered substance is retained. As no structural information is needed for peak detection, monitoring the amplitude of the ratio of m/z 3 divided by m/z 2 serves as a highly sensitive 'sniffer dog' for compounds of interest. This method also operates reliably in the presence of high-abundance co-eluting matrix components. Conversely, structural elucidation of the identified metabolites is not possible with IRMS but necessitates reanalyses with mass spectrometers with mass scanning capability. The GC/Q-TOF mass spectrometer employed in this study supported the identification of all deuterium-labeled fragment ions with high-resolution/high-accuracy mass spectra of compounds observed after matching of measurements conducted on GC/TC-IRMS and GC/Q-TOF systems. Future studies might benefit from a direct coupling of IRMS and a mass analyzer with scanning capabilities (e.g. quadrupole mass spectrometer) to one GC system as well as from further increasing the sensitivity of the IRMS instrument by changing the amplification resistor.
The study was conducted with support of Antidoping Switzerland (Berne, Switzerland) and the Manfred-Donike-Institute for Doping Analysis (Cologne, Germany).