Determination of the origin of urinary norandrosterone traces by gas chromatography combustion isotope ratio mass spectrometry

Abstract

On the one hand, 19-norandrosterone (NA) is the most abundant metabolite of the synthetic anabolic steroid 19-nortestosterone and related prohormones. On the other hand, small amounts are biosynthesized by pregnant women and further evidence exists for physiological origin of this compound. The World Anti-Doping Agency (WADA) formerly introduced threshold concentrations of 2 or 5 ng of NA per ml of urine to discriminate 19-nortestosterone abuse from biosynthetic origin. Recent findings showed however, that formation of NA resulting in concentrations in the range of the threshold levels might be due to demethylation of androsterone in urine, and the WADA 2006 Prohibited List has defined NA as endogenous steroid. To elucidate the endogenous or exogenous origin of NA, ¹³C/¹²C-analysis is the method of choice since synthetic 19-nortestosterone is derived from C₃-plants by partial synthesis and shows δ¹³C_VPDB-values of around −28‰. Endogenous steroids are less depleted in ¹³C due to a dietary mixture of C₃- and C₄-plants. An extensive cleanup based on two high performance liquid chromatography cleanup steps was applied to quality control and doping control samples, which contained NA in concentrations down to 2 ng per ml of urine. ¹³C/¹²C-ratios of NA, androsterone and etiocholanolone were measured by gas chromatography/combustion/isotope ratio mass spectrometry. By comparing δ¹³C_VPDB-values of androsterone as endogenous reference compound with NA, the origin of NA in doping control samples was determined as either endogenous or exogenous.

---

Introduction

Anabolic steroids have been prohibited for use in sport by the International Olympic Committee (IOC) since the Olympic Games in Montréal in 1976. Nandrolone has been known as an anabolic agent since the 1930s¹ and its use by athletes became popular in the late 1950s.² It is mainly metabolized to conjugates of 19-norandrosterone (NA) and 19-noretiocholanolone (NE).^3,4 Prohormones of nandrolone such as norandrostenediol or norandrostenedione likewise can be abused by athletes, but basically exhibit the same metabolism. The application of norethisterone which is present in contraceptives also leads to formation of NA. In these cases the major metabolite tetrahydronorethisterone can be found in contrast to illicit intake of anabolic 19-norsteroids.⁴ The question, whether urinary 19-norsteroids such as NA might be of endogenous origin has been discussed extensively. A review of this issue was published by Bricout and Wright.⁵ Recent findings showed that, besides the possible sources discussed (exercise, contaminated nutritional supplements, intermediates during the biological synthesis of estrogens etc.), a demethylating activity in urine can transform endogenous steroids such as androsterone (AND) or etiocholanolone (ETIO) into the corresponding 19-norsteroids (NA or NE).⁶ Amounts up to 2.2 ng of NA per ml of urine in doping control samples were due to this demethylating activity.⁷ Accordingly, the World Anti-Doping Agency (WADA) has revised the threshold concentration of 2 ng of NA per ml of urine^4,8 and defined it an endogenous steroid.⁹ Thus, it is a pressing task to determine the origin of NA in urine in order to decide if its presence is due to the administration of nandrolone or a related substance or if it is of physiological origin due to e.g. the transformation of AND into NA. To differentiate between endogenous and synthesized naturally occurring steroids, the method of choice is gas chromatography combustion isotope ratio mass spectrometry (GC/C/IRMS), which was introduced to doping control in 1994.¹⁰ Synthesized steroids are normally made from Dioscorea spp. or soy.¹¹ These are C₃-plants, which are depleted in ¹³C in contrast to C₄-plants. In consequence, synthesized steroids and their metabolites are also depleted in ¹³C in contrast to endogenous steroids since endogenous steroids derive from the diet, which is usually a mixture of C₃-plants and C₄-plants. Carbon isotope ratios are expressed as δ¹³C_VPDB-values according to eqn 1 where R represents the ¹³C/¹²C molar ratio of the sample and of the VPDB-standard respectively.

The challenge is to obtain δ¹³C_VPDB-values from NA in urine at very low concentrations. δ¹³C_VPDB-values down to 2 ng ml⁻¹ should be confirmed whereas the lower limit for GC/C/IRMS-systems is about 10 ng of NA per injection to obtain valid ¹³C/¹²C-ratios. Thus at least 10 ml of urine have to be cleaned up efficiently, and the purity of the isolated target analytes is of utmost importance to guarantee baseline separation on the GC for reliable δ¹³C_VPDB-values.¹² Hence, a cleanup-method for 10 ml of urine was developed, which allows reliable ¹³C/¹²C-ratios of urinary NA down to 3 ng ml⁻¹ of urine or even lower to be obtained. The δ¹³C_VPDB-values can be compared with those from ETIO and AND which are isolated at the same time.

Experimental

Chemicals and standards

Methanol (puriss., distilled before use), acetone (GC grade) and methyl-tert-butylether (distilled before use) (MTBE) were purchased from KMF Laborchemie Handels GmbH (St. Augustin, Germany), n-hexane (gradient grade for liquid chromatography), isopropanol (p.a.) (IPA) and sodium phosphate (p.a.) from Merck (Darmstadt, Germany), β-Glucuronidase from E. coli from Roche Diagnostics (Mannheim, Germany) and acetonitrile (gradient grade for liquid chromatography) (AcN) from JT Baker (Deventer, Holland). The steroids AND and ETIO were purchased from Sigma-Aldrich (Steinheim, Germany) and the reference standard 5α-androstane-3α,17β-diacetate from Steraloids (Newport, USA). NA and its glucuronide were synthezised in our laboratory.^13–15

Method

The method comprises the following steps:

1. Reversed phase solid phase extraction (RP SPE)

2. Enzymatic hydrolysis of glucuronides

3. Liquid–liquid extraction (LLE)

4. 1st normal phase high performance liquid chromatography (NP-HPLC) purification on a dimethylaminopropyl column: separation in two fractions

5. 2nd RP-HPLC purification on a C₁₈ column: separation of one fraction in two subfractions

6. GC/C/IRMS

7. Interpretation of results

1. Reversed phase SPE. The RP SPE cartridges from Macherey-Nagel (Düren, Germany) (MN) (Chromabond C₁₈, 500 mg, 6 ml) were conditioned with first 2 ml of methanol, then 2 ml of water. Ten ml of urine was passed through the column and washed with 2 ml of water. Elution was performed with 2 ml of methanol.

2. Enzymatic hydrolysis of glucuronides. The dried eluate was dissolved in 1 ml of sodium phosphate buffer (0.2 M, pH 7). Fifty µl of β-Glucuronidase was added and incubated for 1 h at 50 °C.

3. Liquid–liquid extraction. 250 µl of carbonate buffer (200 g l⁻¹, K₂CO₃/KHCO₃ = 1/1 (w/w)) was added and the aqueous layer was extracted with 5 ml of MTBE. The organic layer was transferred into a conical test tube and evaporated to dryness. This extract was transferred with two washings of 100 µl of methanol each to an HPLC-Vial with micro-insert and dried in a desiccator.

4. 1st NP-HPLC on a dimethylaminopropyl column from MN: EC 250/4.6 Nucleosil 100-5 N(CH₃)₂ + guard column (CC 8/4 Nucleosil 100-5 N(CH₃)₂). The aim of the first HPLC is to separate NA and AND from ETIO and necessary cleanup for GC/C/IRMS. To estimate the retention times and the collection pattern for NA, AND and ETIO, a mixed standard of 5 µg each dissolved in 50 µl of n-hexane/IPA = 90/10 (v/v) was injected into an Agilent 1100 HPLC-system from Agilent (Waldbronn, Germany), equipped with degasser, quaternary pump, autosampler, column-oven and UV-detector. The mobile phase was a mixture of n-hexane and IPA starting with 4% of IPA increasing to 20% within 15 min followed by 9 min column-washing with 80% of IPA and 20% of n-hexane. Reconditioning of the HPLC column was achieved by washing 9 min isocratically with the initial mixture. The temperature was set to 50 °C to reduce the pressure during the column-washing. The flow rate was 1 ml min⁻¹ and detection was performed at 200 nm. Fig. 1 displays a UV-chromatogram of the mixed standard and the collection pattern for the samples. After dissolving the extracts in 50 µl of n-hexane/IPA = 90/10 and injection of 50 µl into HPLC, fractions were collected with an automatic fraction collector (Foxy 200) from Isco (Lincoln, Nebraska, USA). After drying the sample-fractions, the NA/AND-fraction was transferred into another HPLC-vial with micro-insert using twice 100 µl aliquots of methanol and dried again. The ETIO-fraction was clean enough for GC/C/IRMS (conditions see below).

Fig. 1 NP-HPLC/UV chromatogram at 200 nm of 5 µg of NA, AND and ETIO each; collection pattern.

5. 2nd RP-HPLC on a C₁₈ column: LiChroCART 250-4 LiChrospher 100 RP₁₈ EC (5 µm) + pre-column: LiChroCart 25-4 LiChrospher 100 RP₁₈ (5 µm), both from Merck. The aims of the second HPLC are to separate NA from AND and to eliminate further matrix-compounds with regard to low amounts of NA in urine. To estimate the retention times and the collection pattern for AND and NA, a mixed standard of 5 µg each dissolved in 50 µl of methanol was injected into the above mentioned Agilent 1100 HPLC-system. The mobile phase was a mixture of water and AcN starting with 30% of AcN increasing to 100% within 20 min. Reconditioning of the HPLC column was achieved by washing 5 min isocratically with the initial mixture. The temperature was set to 24 °C, the flow rate was 1 ml min⁻¹, detection was performed at 200 nm. Fig. 2 displays a UV-chromatogram of the mixed standard and the collection pattern for the samples. After dissolving the NA/AND-fraction in 50 µl of methanol and injection into HPLC, NA and AND were separated from each other according to the above estimated collection pattern. After drying both fractions, cleanup is completed for GC/C/IRMS.

Fig. 2 RP-HPLC/UV chromatogram at 200 nm of 5 µg of NA and AND each; collection pattern.

6. GC/C/IRMS. Measurement of ¹³C/¹²C-ratios was performed by GC/C/IRMS. A GC 5890 II (Hewlett & Packard, now Agilent Technologies) was coupled to a Delta C gas isotope ratio mass spectrometer from ThermoElectron (Bremen, Germany) by a combustion interface II (ThermoElectron). The first GC column used was an Optima δ3 (MN). Dimensions of the column were 17 m length and 0.25 mm inner diameter. The film thickness was 0.25 µm. The second GC column used was a J&W HP-5MS (Agilent Technologies). Dimensions of this column were 30 m length and 0.25 mm inner diameter. The film thickness was 0.25 µm. A retention gap (deactivated fused silica) was connected to the analytical column. Dimensions of the retention gap for splitless injections were 2 m length and 0.32 mm inner diameter. Cool on-column (COC) injections were performed with a retention gap of 1 m length and 0.53 mm inner diameter. Helium (purity 5.0 ∼99.999%) was the carrier gas at a constant pressure of 30 psi. In splitless mode, up to 3 µl from at least 5 µl of the sample dissolved in methanol were injected (splitless time 1:30 min) at 300 °C injector temperature and 60 °C oven temperature. For COC injections, up to 2 µl from at least 5 µl of the sample dissolved in acetone were injected at a temperature of 50 °C, which was held for 0.5 min. The GC temperature using the Optima δ3 column was increased with 30 °C min⁻¹ to 265 °C. A second ramp of 3 °C min⁻¹ followed during which separation of the relevant compounds was achieved. The final temperature of 295 °C was held for 2 min. The GC temperature using the HP-5MS column was increased with 30 °C min⁻¹ to 240 °C. A second ramp of 2 °C min⁻¹ to 265 °C was applied followed by 15 °C min⁻¹ to 295 °C. The final temperature was held for 2 min. To improve recovery, samples were usually injected manually. In some cases an autosampler (A200S, CTC Analytics) was employed. It was operated at an injection speed of 5.5 µl s⁻¹. An isotopically characterized reference standard (5α-androstane-3α,17β-diol-diacetate) was co-injected (0.5 µl, 100 µg ml⁻¹) to control chromatographic conditions and validity of the calculated isotope ratios. ¹³C/¹²C-ratios are expressed as δ¹³C_VPDB-values, where the working standard (carbon dioxide, δ¹³C_VPDB = −3.0 ‰) was calibrated vs. an n-alkane mixture.¹⁶

Results and discussion

Validity of the method

Specificity. As both GC columns showed very similar chromatographic properties, Fig. 3 and 4 show the cleanup efficiency exemplified on the Optima δ3 column. Fig. 3 shows a GC/C-chromatogram of the relevant time period of the NA-fraction of a purified blank urine. The left panel shows the intensity of m/z 44. Almost no biological components coelute in the relevant time period for NA between the first two peaks. The right panel of Fig. 3 shows the ratio of the masses 45/44 (with an offset of 100 mV each). The ratio is quite constant within the relevant time period which is also promising for NA-containing samples. Fig. 4 is similar to Fig. 3, but a urine sample was analysed, which was spiked with 4 ng of NA per ml as glucuronic acid conjugate. The response (left panel) is sufficient and the ratio 45/44 (right panel) indicates that no relevant coelutions are present which might confound with NA. To further check for coelutions, all NA-containing samples were also injected into a GC/MS, which was equipped similar to the GC/C/IRMS. The method is considered specific for NA as none of the urine samples showed confounding coelutions in the relevant time period.

Fig. 3 Relevant section of a GC/C-chromatogram of the NA-fraction of a blank urine; intensity m/z 44 on the left and the corresponding ratio 45/44 on the right panel.

Fig. 4 Relevant section of a GC/C-chromatogram of the NA-fraction of a urine sample spiked with 4 ng of NA ml⁻¹; intensity m/z 44 on the left and the corresponding ratio 45/44 on the right panel.

Repeatability. Blank urine samples were spiked with NA as glucuronic acid conjugate at concentrations of 2, 4, 6 and 8 ng of NA per ml. Table 1 shows the mean δ¹³C_VPDB-values of NA for the different concentrations, which were measured in each sequence over 12 months. The mean δ¹³C_VPDB-values for the 4 different concentrations were between −26.5 ‰ and −26.8 ‰ whereas the mean of all measurements was −26.7 ‰. The standard deviations indicate that the method is sufficiently precise although the response of the 2 ng ml⁻¹ urine was sometimes close to the limit of the dynamic linear range of the IRMS.

Table 1 Precision of δ¹³C_VPDB-values of NA-glucuronide spiked urine specimens

Concentration of NA/ng ml^−1	Mean CVPDB of NA ± standard deviation (‰)
2	−26.5 ± 0.35 (n = 5)
4	−26.8 ± 0.77 (n = 6)
6	−26.6 ± 0.73 (n = 9)
8	−26.8 ± 0.39 (n = 8)
Total	−26.7 ± 0.59 (n = 28)

---

Reproducibility. The results obtained from repeated analyses of 10 doping control samples are shown in Fig. 5. The x-axis indicates the 10 different urine specimens whereas the y-axis shows the δ¹³C_VPDB-values of NA. Concentrations obtained by GC/MS, corrected for specific gravity if necessary, are also given.^4,17 The first analyses were carried out on the Optima δ3 column with splitless injection (●). The injection mode for the second analyses was changed. Due to more robust hardware conditions concerning the GC/C/IRMS, a COC injector was installed. Four samples were measured on the Optima δ3 column (o) whereas 6 samples were measured on the HP-5MS column (Δ). No significant trend between the δ-values for NA of the first and the second analyses was observed, neither for the different injection methods nor for the different GC-columns. δ-Values are reproducible even at very low concentrations. The largest difference between the repeated analyses of NA is 1.7 ‰ and the mean difference of the 10 data pairs is 0.03 ‰ with a standard deviation of ±0.88 ‰. Concluding the repeated analyses, both columns and both injection methods were well suited for these purposes.

Fig. 5 Repeated analyses of δ¹³C_VPDB-values of NA in doping control samples and its concentration. First analysis under splitless conditions on the Optima δ3 column (●); second analysis COC on the Optima δ3 column (○) or on the HP-5MS column (Δ).

Accuracy of the method. In Fig. 6, δ¹³C_VPDB-values of NA vs. AND of control samples are shown. The theoretical identity of both parameters is indicated by a solid line. The exogenous values derive from urine samples from excretion studies with either norandrostenedione¹⁸ or norandrostenediol.¹⁹ Endogenous values close to the identity line are from urine samples of two pregnant women (4th, 7th and 8th month of pregnancy). The two groups are clearly separated from each other as NA of exogenous origin is more depleted in the ¹³C-amount than NA of endogenous origin caused by pregnancy.

Fig. 6 δ¹³C_VPDB-values of NA vs. AND of control samples; the ellipses indicate endogenous or exogenous origin of NA.

Fig. 7 shows the differences in the δ-values (expressed as Δδ) of AND and NA vs. the urinary concentration of NA of the control samples. Endogenous NA-concentrations caused by pregnancy did not exceed 2 ng ml⁻¹ and the Δδ-values were lower than 2 ‰. NA-concentrations caused by oral administration of norandrostenediol or norandrostenedione of the selected urine samples were between 4 and 8 ng ml⁻¹ and the Δδ-values did not fall below 4 ‰. By means of the Δδ-values, the origin of NA from control samples was classified accurately.

Fig. 7 Δδ of AND and NA vs. the concentration of NA in control samples; the ellipses indicate endogenous or exogenous origin of NA.

Doping control samples. 22 doping control samples containing between 2 ng and 17 ng of NA per ml of urine were analysed. Each sample was analysed under splitless conditions on the Optima δ3 GC-column whereas 10 samples were reanalysed using COC as described above. Fig. 8 shows all δ¹³C_VPDB-values of NA vs. AND obtained from the doping control samples (splitless and COC injection). The theoretical identity of both parameters is indicated again by a solid line. The δ¹³C_VPDB-values of AND reflect the endogenous ¹³C/¹²C-ratios. Again, two groups were identified. In the one group, δ-values were close to the identity line. In these 10 samples (2 in duplicate) NA was obviously of endogenous origin. The other group showed similar δ-values for AND, but the ¹³C/¹²C-ratios of NA were generally more depleted. In these 12 samples (8 in duplicate) NA was obviously of exogenous origin.

Fig. 8 δ¹³C_VPDB-values of NA vs. AND of 25 doping control samples; 12 samples reanalysed under varying conditions; the ellipses indicate the presumable classification into endogenous or exogenous origin of NA.

Fig. 9 shows the Δδ-values of AND and NA of the doping control samples vs. the concentration of NA, corrected for specific gravity if necessary.⁴ Δδ-values of the 12 samples showing exogenous δ¹³C_VPDB-values for NA were larger than 4 ‰ and up to almost 9 ‰. They were identified over the whole concentration range. The differences in the δ¹³C_VPDB-values between NA of endogenous origin and AND were below the WADA criterion of 3 ‰²⁰ and did not exceed 2 ‰. The highest concentration for NA of endogenous origin was 5.6 ng ml⁻¹. A steroids demethylating activity according to Grosse et al.⁶ was not detected in all of the samples with inconspicuous isotope signature. It was not detected in any of the samples showing exogenous origin of NA. No correlation depending on the gender of the athletes was observed. The dashed line in Fig. 9 indicates a possible decision limit to discriminate between endogenous and exogenous origin of NA. Hence, a doping offence with nandrolone or a corresponding 19-norsteroid was detected if the Δδ-value between NA and AND was above this limit.

Fig. 9 Δδ of AND and NA vs. the concentration of NA of 25 doping control samples (12 of them reanalysed under different conditions); Δδ-values above the possible decision limit indicated by the dashed line were deemed to show exogenous origin for NA whereas Δδ-values below this line were deemed to show endogenous origin for NA.

Conclusions

The presented method, based on two HPLC cleanup steps, enables reliable measurements of the carbon isotope ratio of urinary NA traces by GC/C/IRMS under varying conditions. It was developed to elucidate the origin of urinary NA as either endogenous or exogenous to prove an abuse with nandrolone or corresponding 19-norsteroids. By means of the differences in the δ¹³C_VPDB-values of NA and AND, the origin of NA was elucidated correctly as either endogenous or exogenous in control samples containing low amounts of NA of known origin. Accordingly, doping control samples were analysed and the origin of NA was elucidated as either endogenous or exogenous. Data of doping control samples are presented with an NA-concentration of up to 5.6 ng per ml of urine, but the carbon isotope ratios of NA indicate endogenous origin.

Acknowledgements

This project was carried out with support of the WADA. Gratefully acknowledged are also the contributions of Vassilios Gougoulidis for the sample preparation and Yvonne Schrader for providing NA-containing urine samples of excretion studies.

References

1. A. Butenandt and K. Tscherning, Z. Physiol. Chem., 1934, 229, 167–182.
2. H. A. Haupt and G. D. Rovere, Am. J. Sports Med., 1984, 12, 469–484.
3. W. Schänzer, A. Breidbach, H. Geyer, C. von Kuk, E. Nolteernsting and M. Thevis, in Recent Advances in Doping Analysis. Proceedings of the 18th Cologne Workshop on Dope Analysis, 20th to 25th February 2000, ed. W. Schänzer, H. Geyer, A. Gotzmann and U. Mareck-Engelke, Sport und Buch Strauß, Köln, 2001, vol. 8, pp. 155–174.
4. WADA, Reporting Norandrosterone Findings. Technical Document TD2004NA, World Anti-Doping Agency, Montréal, Canada, 2004, http://www.wada-ama.org/rtecontent/document/nandrolone_aug_04.pdf.
5. V. Bricout and F. Wright, Eur. J. Appl. Physiol., 2004, 92, 1–12.
6. J. Grosse, P. Anielski, P. Hemmersbach, H. Lund, R. K. Mueller, C. Rautenberg and D. Thieme, Steroids, 2005, 70, 499–506.
7. D. Thieme, P. Anielski, J. Grosse, P. Hemmersbach, H. Lund and C. Rautenberg, in Recent Advances in Doping Analysis. Proceedings of the Manfred Donike Workshop. 22nd Cologne Workshop on Dope Analysis, 7th to 12th March 2004, ed. U. Mareck, Sport und Buch Strauß, Köln, 2005, vol. 12, pp. 177–188.
8. WADA, The 2005 Prohibited List, World Anti-Doping Agency, Montréal, Canada, 2005, http://www.wada-ama.org/rtecontent/document/list_2005.pdf.
9. WADA, The 2006 Prohibited List, World Anti-Doping Agency, Montréal, Canada, 2006, http://www.wada-ama.org/rtecontent/document/2006_LIST.pdf.
10. M. Becchi, R. Aguilera, Y. Farizon, M.-M. Flament, H. Casabianca and P. James, Rapid Commun. Mass Spectrom., 1994, 8, 304–308.
11. A. Kleemann and H. J. Roth, Arzneistoffgewinnung: Naturstoffe und Derivate, Thieme, Stuttgart, 1983.
12. A. Newman, Anal. Chem., 1996, 68, 373A–377A.
13. M. Thevis, G. Opfermann, H. Schmickler and W. Schanzer, J. Mass Spectrom., 2001, 36, 159–168.
14. W. Schänzer and M. Donike, Anal. Chim. Acta, 1993, 275, 23–48.
15. W. Schänzer and M. Donike, in Recent Advances in Doping Analysis. Proceedings of the 12th Cologne Workshop on Dope Analysis, 10th to 15th April 1994, ed. M. Donike, H. Geyer, A. Gotzmann and U. Mareck-Engelke, Sport und Buch Strauß, Köln, 1995, vol. 2, pp. 93–112.
16. A. Schimmelmann, http://php.indiana.edu/%E2%88%BCaschimme/hc.html, 2004.
17. U. Mareck-Engelke, G. Schultze, H. Geyer and W. Schänzer, in Recent Advances in Doping Analysis. Proceedings of the 18th Cologne Workshop on Dope Analysis, 20th to 25th February 2000, ed. W. Schänzer, H. Geyer, A. Gotzmann and U. Mareck-Engelke, Sport und Buch Strauß, Köln, 2000, vol. 8, pp. 145–150.
18. Y. Schrader and W. Schänzer, in Recent Advances in Doping Analysis. Proceedings of the 22nd Cologne Workshop on Dope Analysis, 7th to 12th March 2004, ed. W. Schänzer, H. Geyer, A. Gotzmann and U. Mareck, Sport und Buch Strauß, Köln, 2004, vol. 12, pp. 109–119.
19. Y. Schrader and W. Schänzer, unpublished work.
20. WADA, Reporting and Evaluation Guidance for Testosterone, Epitestosterone, T/E Ratio and other Endogenous Steroids. Technical Document TD2004EAAS, World Anti-Doping Agency, Montréal, Canada, 2004, http://www.wada-ama.org/rtecontent/document/end_steroids_aug_04.pdf.

---