Background: Carbon isotope ratio methods are used in doping control to determine whether urinary steroids are endogenous or pharmaceutical.
Methods: Gas chromatography-combustion-isotope ratio mass spectrometry (GC-C-IRMS) was used to determine the δ13C values for 5β-androstane-3α,17β-diyl diacetate (5βA), 5α-androstane-3α,17β-diyl diacetate (5αA), and 5β-pregnane-3α,20α-diyl diacetate (5βP) in a control group of 73 healthy males and 6 athletes with testosterone/epitestosterone ratios (T/E) >6.
Results: The within-assay precision SDs for 5βA, 5αA, and 5βP were ± 0.27‰, ± 0.38‰, and ± 0.28‰, respectively. The between-assay precision SDs ranged from ± 0.40‰ to ± 0.52‰. The system suitability and batch acceptance scheme is based on SDs. For the control group, the mean δ13C (SD) values were −25.69‰ (± 0.92‰), −26.35‰ (± 0.68‰), and −24.26‰ (± 0.70‰), for 5βA, 5αA, and 5βP, respectively. 5βP was greater than 5βA and 5αA (P <0.01), and 5βA was greater than 5αA (P <0.01). The means − 3 SD were −28.46‰, −28.39‰, and −26.37‰ for 5βA, 5αA, and 5βP, respectively. The maximum difference between 5βP and 5βA was 3.2‰, and the maximum 5βA/5βP was 1.13. Three athletes with chronically elevated T/Es had δ13C values consistent with testosterone administration and three did not.
Conclusions: This GC-C-IRMS assay of urine diols has low within- and between-assay SDs; therefore, analysis of one urine sample suffices for doping control. The means, SDs, ±3 SDs, and ranges of δ13C values in a control group are established. In comparison, testosterone users have low 5βA and 5αA, large differences between 5βA or 5αA and 5βP, and high 5βA/5βP and 5αA/5βP ratios.
A critical issue in doping control is establishing the origin of testosterone and other steroids typically found in human urine. If the origin can be shown to be pharmaceutical, as opposed to endogenous, a doping offense has occurred. Androgens for pharmaceutical use are not synthesized de novo; they are obtained by semisynthesis from starting materials such as diosgenin and stigmasterol, which are derived from plants (1). Furthermore, these compounds have less 13C than their endogenous homologs (2)(3); therefore, urinary steroids with a low 13C/12C ratio are likely to have originated from pharmaceutical sources. This thesis has led to the use of gas chromatography-combustion-isotope ratio mass spectrometry (GC-C-IRMS)1 in doping control because the method determines the δ13C value of steroids extracted from urine, based on the equation: where 13C/12C refers to the isotope ratio in the sample or an international standard (a reference gas calibrated relative to Pee Dee Belemnite). Accordingly, for several years, GC-C-IRMS has been applied to detect the use of exogenous steroids such as testosterone (3)(4), dihydrotestosterone (5)(6), and dehydroepiandrosterone (6)(7) by athletes.
Testosterone administration increases the ratio of urine testosterone to epitestosterone (T/E). This finding led to the use of T/E as a screening test for testosterone administration (8). If the ratio exceeds 6, laboratories accredited by the International Olympic Committee report the result to the sport authority (9). It is known, however, that factors other than testosterone administration may also increase the T/E ratios (10)(11)(12). Furthermore, testosterone administration does not always lead to an elevated4 T/E ratio (3)(13). These factors complicate the interpretation of elevated T/E ratios and limit the significance of T/E ratios <6. Determining the T/E time profile of an individual is useful, but it requires consideration of the results of past tests or the collection of additional samples. Our previous work demonstrated that the measurement of δ13C values of testosterone and its metabolites can detect testosterone use (4)(14)(15). Others have reported that δ13C values may be abnormally low even in samples with T/E ratios <6 (3).
With the introduction of new methods, it is essential to fully characterize the assay and to establish the values in the control group. This is particularly true with δ13C measurements when applied to drug control because the method is inherently difficult, the analytical results are likely to be litigated, and it has been suggested (3)(5) that diet or ethnicity may influence the 13C/12C ratio of urine steroids. Accordingly, herein we characterize the δ13C measurements of the diacetates of 5β- and 5α-androstane-3α,17β-diol (5βA and 5αA, respectively), and 5β-pregnane-3α,20α-diol (5βP) with respect to precision, linearity, system suitability, and batch acceptance. In addition, we determined δ13C values for these steroids in a control group of 73 healthy male subjects from four different ethnic groups. Finally, data gathered from the control group were used to interpret δ13C values obtained on athletes with elevated T/E ratios and to discuss diagnostic criteria for doping.
Materials and Methods
steroid calibration solution
A steroid calibration solution was prepared that contained 25 mg/L 5β-androstane-3α,17β-diol, 5α-androstane-3α,17β-diol, and 5β-pregnane-3α,20α-diol, all as diacetates (Steraloids, Inc.).
The control group consisted of 74 male medical students (age range, 19–29 years) from the University of California at Los Angeles. Body weights were 45–110 kg. The students had the option to give a medical and drug history and to describe their ethnicity. No student declared taking steroids or reported any chronic disease. The ethnicities were “Asian” (n = 16), “African American” (n = 9), “Caucasian” (n = 25), “Hispanic” (n = 13), “other” (n = 7), and “unknown” (n = 4). The students collected 24-h urine samples, and 50-mL aliquots were stored at 4 °C. The study was conducted under the guidance of the Human Subject Protection Committee of the University of California at Los Angeles.
athletes with elevated T/E ratios
Of the large number of anonymous urine samples that our laboratory analyzes for sport drug control programs in the US, some have a urine T/E ratio >6. In such cases, the sport organization typically collects additional samples to track the urine T/E over weeks or months. We selected some of these urine samples for GC-C-IRMS analysis based on the availability of at least two samples per person, sufficient volume to perform the analysis, and permission from the sport organization.
quality-control urine samples
Quality-control urines (QC-H and QC-L) were obtained from two subjects known not to be using testosterone or any substance likely to alter the δ13C values of 5β-androstane-3α,17β-diol, 5α-androstane-3α,17β-diol, and 5β-pregnane-3α,20α-diol (measured as diacetates). Each subject donated one 8-h urine, which was aliquoted into 10-mL cryogenic tubes and stored at −20 °C until analysis.
urinary steroid concentrations and sample preparation for GC-C-IRMS
The T/E ratio was estimated by GC-MS as described previously (16). The concentrations of 5β- and 5α-androstane-3α,17β-diol were estimated from a QC sample prepared by adding to steroid-free urine 5β- and 5α-androstane-3α,17β-diol. [16,16,17-2H]-testosterone (CDN Isotopes) was used as an internal standard. The 5β-pregnane-3α,20α-diol was not quantified. The sample preparation methods for GC-C-IRMS analysis and the instrument conditions have been described (15). The volume of urine extracted for GC-C-IRMS was 10 mL.
gc-c-irms determination of 5αA, 5βA, and5βP
The GC-C-IRMS analyses were conducted on a Finnigan Delta Plus Isotope Ratio Mass Spectrometer. The IRMS was connected to a Hewlett Packard Model 6890 gas chromatograph via a Finnigan Combustion III interface. The GC was equipped with an HP-50+ capillary column [30 m × 0.25 mm (i.d.); 0.15 μm film thickness] and a Finnigan A200S autosampler. The combustion oven was oxidized by back-flushing with oxygen for 1 h every 30 samples. The oxidation reactor in the combustion oven was replaced every 500 samples. The injection volume was 1 μL. The recoveries for 5β- and 5α-androstane-3α,17β-diol were 85% and 86%, respectively.
GC-C-IRMS response linearity
The linearity of the GC-C-IRMS response was determined by preparing seven solutions containing all three analytes at 2.5, 5, 10, 25, 50, 100, and 150 mg/L, respectively. One microliter of each solution was injected three times on 1 day, and the three δ13C values for each compound were averaged.
Instrument precision for 5βP, 5βA, and 5αA was determined by extracting one aliquot of each QC urine and injecting each extract four times in succession. The within-day precision was determined by extracting 20 aliquots of QC-H in the same batch and injecting each one once. The between-assay precision was determined by extracting one aliquot of QC-H and QC-L per day for 16 days, spanning 15 months, and injecting each aliquot once.
daily system suitability test
To establish a tolerance range, the steroid calibrator was injected five times on each of 5 different days over 2 weeks (n = 25), and the mean δ13C values and SDs were calculated for each steroid. Each day, before urine sample analysis, the system suitability was assessed by injecting the calibrator three times and calculating the mean δ13C values. The system was considered suitable for batch analyses if the mean δ13C values of at least two of the three steroids were within ± 2 SD of the means described above. If the system suitability test failed, maintenance was performed on the gas chromatograph injection port, the GC column and/or the oxidation reactor were replaced, and the calibrator was re-injected. Data acceptance criteria included absence of peak tailing, retention times of the steroids within ± 1% of established values, and a minimum 0.8 V response at m/z 44 for each steroid.
criteria for batch acceptance
To establish a tolerance range for the QC urines for batch data acceptance, two aliquots of each QC urine were extracted and each injected twice per day for 5 days spanning 2 weeks (n = 20). ANOVA was performed using the factors day (5), injection (2), duplicate (2), and QC (2). The mean δ13C values and SDs were calculated for each of the three steroids. For batch analysis, each unknown urine sample was extracted and injected once. One aliquot of the two QC urines and one aliquot of the steroid calibrator were analyzed with each batch of samples. These three controls were injected at the beginning, in the middle, and at the end of the batch. The batch was accepted if at least six of the nine means for the three steroids were within ± 2 SD of the previously established means. The acceptance criteria were as noted above except that an instrument response of 0.3 V at m/z 44 was accepted.
The Smirnov-Grubbs method was used to test the control group for outliers; otherwise, all statistical tests utilized statistical software (SAS). The linearity of the GC-C-IRMS response was assessed by least-squares linear regression. The response was considered linear if the slope was zero at P = 0.01. The method of Koch and Peters (17) was used to determine the SDs of duplicates. The normality of the distributions of 5αA, 5βA, and 5βP, differences, and ratios was determined by the Anderson-Darling test. The values for 5αA, 5βA, and 5βP were correlated with Pearson’s correlation coefficient, r. Two-sided paired t-tests were used to compare the means of 5αA, 5βA, and 5βP in the control group. The general linear model procedure was used to assess differences between the mean δ13C values of the ethnic groups, and the power of the analysis was assessed by one-way ANOVA.
linearity of GC-C-IRMS response
The IRMS response was linear for 5β- and 5α-androstane-3α,17β-diyl diacetate from 2.5 to 150 mg/L (Fig. 1⇓ ). However, for 5β-pregnane-3α,20α-diyl diacetate, the response was linear only between 5 and 150 mg/L. The concentrations of 5β- and 5α-androstane-3α,17β-diol in the control group were 23–430 and 25–204 μg/L, respectively. Therefore, assuming 85% recovery, the amounts of 5β- and 5α-androstane-3α,17β-diol extracted from the 10-mL urine and reconstituted in 25 μL of cyclohexane were analyzed in the linear range of the IRMS.
precision of the GC-C-IRMS instrument and assay
Instrument precision (SD) for 5βA in QC-H and QC-L was 0.32‰ and 0.41‰, respectively. For 5αA, it was 0.08‰ and 0.59‰, respectively, and for 5βP, it was 0.27‰ and 0.16‰, respectively. The descriptive statistics for the within-assay experiment on QC-H are shown in Table 1⇓ . The SDs were 0.27‰, 0.38‰, and 0.28‰ for 5βA, 5αA, and 5βP, respectively, and the CVs were ≤1.4%. The range of values was 0.9‰ for 5βA, 1.2‰ for 5βP, and 1.8‰ for 5αA. The between-assay SDs for QC-H were 0.40‰, 0.42‰, and 0.44‰ for 5βA, 5αA, and 5βP, respectively, and the CVs were ≤1.8% (Table 2⇓ ). For QC-L, the values were slightly higher. The mean δ13C values for the three steroids in QC-L were significantly lower than the means for QC-H.
system suitability and batch acceptance
In the system suitability test, the SDs of 5βA, 5αA, and 5βP were 0.31‰, 0.59‰, and 0.54‰, respectively. The tolerance range for the batch acceptance data provided SDs for 5βA, 5αA, and 5βP of 0.68‰, 0.67‰, and 0.65‰ for QC-H, and 0.66‰, 0.56‰, and 0.39‰ for QC-L, respectively. Because these data consisted of duplicate analyses, we calculated the SDs of duplicates and performed an ANOVA. The SDs of duplicates were 0.24‰, 0.32‰, and 0.37‰ for 5βA, 5αA, and 5βP, respectively. These values were approximately one-half the SDs obtained for the factor injection. ANOVA revealed that QC-H and QC-L were different, and there was no difference for the factors injection, duplicate, or day.
control subjects: descriptive statistics for the concentrations and δ13C of urinary steroids
The geometric means for urine testosterone, epitestosterone, and 5β- and 5α-androstane-3α,17β-diol concentrations were 20, 29, 98, and 58 μg/L, respectively. The geometric mean for urine T/E was 0.7 (arithmetic range, 0.2–5.3). The concentration of urine 5β-pregnane-3α,20α-diol was not determined. Table 3⇓ and Fig. 2⇓ show the descriptive statistics and the histograms for 5βA, 5αA, and 5βP. The outlier test was applied, and no data points were excluded. The distributions of 5αA and 5βP were gaussian (A2 = 0.28, P >0.25; and A2 = 0.21, P >0.25, respectively). The distribution of 5βA was gaussian at P = 0.01, but not gaussian at P = 0.05, A2 = 0.85.
There were significant (P <0.0001) correlations between 5βA and 5αA (r = 0.70), between 5βA and 5βP (r = 0.67), and between 5βP and 5αA (r = 0.58). The CVs of the δ13C values for the diacetates were 2.6–3.6%. The means − 3 SD for 5βA, 5αA, and 5βP were −28.46‰, −28.39‰, and −26.37‰, respectively. The mean 5αA was lower (−26.35‰) than the mean 5βA (−25.69‰; P <0.001). The Wilcoxon nonparametric two-sample test confirmed that 5αA was lower than 5βA. The means for 5βA and 5αA were lower than the mean for 5βP (both P <0.001).
Table 3⇑ also summarizes the statistics of the differences between the means of 5βP and 5βA (5βP − 5βA) and 5βP and 5αA (5βP − 5αA), and the ratios 5βA/5βP and 5αA/5βP. The distributions of 5βP − 5βA and 5βP − 5αA were gaussian (A2 = 0.23 and 0.19, respectively). The 5βP − 5βA differences ranged from −0.08‰ to 3.17‰, and the mean + 3 SD was 3.47‰. The 5βP − 5αA differences ranged from 0.16‰ to 3.72‰, and the mean + 3 SD was 3.99‰. The distributions of 5βA/5βP and 5αA/5βP were gaussian (A2 = 0.19 and 0.21, respectively). The means of 5βA/5βP and 5αA/5βP were 1.06 and 1.09, respectively, and their corresponding means + 3 SD were 1.14 and 1.17. There were no differences between ethnic groups for the means of 5αA, 5βA, or 5βP (Table 4⇓ ).
δ13C values in athletes with elevatedT/E ratios
The urine concentrations of testosterone and epitestosterone, the T/E ratios, and the δ13C values for the six male athletes with urine T/E ratios >6 are summarized in Table 5⇓ . Of the six, athletes 1 and 2 were known to be taking testosterone for medical reasons. According to the sport organizations ordering the tests, athletes 3–6 denied taking any substance known to influence the T/E ratio. The number of samples per athlete ranged from three to eight. The T/E ratio of all samples was >6 for four of the six athletes. Athlete 4 had one T/E ratio of 5.0, and athlete 5 had one T/E ratio of 5.1; otherwise, all T/E ratios were >6.
Table 5⇑ reveals that for athletes 1–3, all 5βA and 5αA values were lower than the mean − 3 SD values of the controls in Table 3⇑ , whereas the 5βP values were all close to the mean 5βP value of the controls (−24.26‰). The cells in Table 5⇑ that are outside the ± 3 SD values of Table 4⇑ are outlined or shaded to facilitate comparisons. The combination of 5βP values that are near the mean and low 5βA and 5αA values produce large differences in 5βP − 5βA and 5βP − 5αA, and large 5βA/5βP and 5αA/5βP ratios. For athletes 1–3, all differences and ratios were outside the + 3 SD range. The means of athletes 1–3 and 4–6 and their z-scores are presented at the bottom of Table 5⇑ . Another aspect of athletes 1–3 was that all 5αA values were lower than the 5βA values for the same urine: the means were −29.84‰ for 5βA and −31.83‰ for 5αA (P = 0.04). The means of 5βP − 5βA (5.9‰; z-score = 6.5) and of 5βP − 5αA (7.9‰; z-score = 9.1) were substantially greater than the + 3 SD values determined for the control group. Similarly, the mean 5βA/5βP ratio was 1.25 (z-score = 6.6), and the mean 5αA/5βP ratio was 1.33 (z-score = 8.9).
To compare the testosterone concentrations in Table 5⇑ to the general population of athletes that we routinely test, we determined the log mean (3.22 μg/L) and log SD (1.19) of the distribution of the latest 11 938 male urine samples tested in our laboratory. The urine testosterone concentrations in subjects 1 and 2, the two permitted testosterone users, ranged from a low of 2.3 μg/L on the first urine ever collected to 186 μg/L (z-score = 1.5, in log units). For subject 3, the urine testosterone concentrations (mean, 613 μg/L; z-score = 2.6) and T/E ratios (mean = 64) were very high. In addition, like athletes 1 and 2, athlete 3 was characterized by low 5βA and 5αA values, and large differences and ratios outside the ± 3 SD range.
In contrast to athletes 1–3, for athletes 4–6, the values for 5βA, 5αA, and 5βP and their differences and ratios were similar to the averages for the control group, and none of the values were remarkable except for one 5βP − 5αA value of −0.4‰ and one 5αA/5βP ratio of 0.99, which were slightly lower than the mean − 3 SD. For subjects 4–6, 5βP − 5βA and 5βP − 5αA were less than 2.1‰ and 3.7‰, respectively, i.e., within ± 3 SD of the control group values. Similarly, all but one of the ratios to 5βP were inside the ± 3 SD range. In addition, the urine testosterone concentrations for subjects 5 and 6 were within + 1.5 SD of the log mean for 11 938 athletes. Thus, compared with the control group, athletes 5 and 6 were characterized by elevated T/E ratios; whereas the values for 5βA, 5αA, and 5βP, their differences, and their ratios for athletes 5 and 6 were, with two minor exceptions, within ± 3 SD of the means of the controls. Athlete 4 had a relatively high mean urine testosterone (234 μg/L; z-score = 1.9).
The use of testosterone or testosterone precursors is indicated by low 5αA and 5βA values, large 5βP − 5αA and 5βP − 5βA differences, and large 5αA/5βP and 5βA/5βP ratios. Of these six variables (two δ13C values, two differences, and two ratios), for athletes 1–3, the ratio 5βA/5βP (z-score = 8.9) and the difference 5βP − 5βA (z-score = 9.1) were the most sensitive indicators of testosterone administration as judged by z-scores. The value for 5αA was also a good indicator (z-score = 8.0). For differences and ratios that utilized 5βA (5βP − 5βA and 5βA/5βP), the z-scores were 6.5 and 4.5, respectively. The least discriminating variable was 5βA (z-score = 4.5). In contrast, for athletes 4–6 in Table 5⇑ , the z-scores were all low (range, −0.02 to 0.9).
characteristics of the GC-C-IRMS assay
GC-C-IRMS measurement of urinary steroids has become a potent analytical method in doping control, although scrupulous attention to detail is necessary. Successful implementation of the method requires strict adherence to QC, system suitability, and batch acceptance criteria such as those described herein. In the present study, after the tolerance ranges for the system and batches were established, the system suitability test never failed and the batch acceptance criteria were always met over the 14 months of the study.
In the instrument precision study, the mean SDs were less than ± 0.33‰; thus, when the variability attributable to extraction and derivatization was eliminated, all of the SDs were low. In the within-assay study, the SDs were also low: ± 0.27‰, ± 0.38‰, and ± 0.28‰ for 5βA, 5αA, and 5βP, respectively. The between-assay SDs ranged from ± 0.40‰ to ± 0.52‰, which is slightly higher than the within-assay values, but they indicate the excellent repeatability and reproducibility of the entire assay.
We did not find comparable precision studies for urinary steroids in the literature; however, Shackleton et al. (3) provided evidence that δ13C values for diols are reproducible based on duplicate analysis of 13 samples from one subject. We used the data in Table 1⇑ from Shackleton et al. (5) to calculate the SD of duplicates and found values of ± 0.36‰, ± 0.15‰, and ± 0.28‰ for 5βA, 5αA, and 5βP, respectively. These data show that the SD for duplicates in another laboratory was similar to the SD that we obtained for duplicates in our batch acceptance studies. The concentrations of 5β-androstane-3α,17β-diol and 5α-androstane-3α,17β-diol in the control urines were 23–430 μg/L; thus, given the 85% recovery of steroids from the 10-mL sample volume used in the analysis, the concentrations of 5β-androstane-3α,17β-diol and 5α-androstane-3α,17β-diol in the 25 μL of the injection solution were 8–146 mg/L, and the samples were analyzed in the linear range of the system (Fig. 1⇑ ).
healthy control male subjects
The δ13C values in Table 3⇑ represent the largest group of healthy males living in the US measured to date. The ranges for the T/E ratios (0.1–3.6) and urine testosterone concentrations (2–116 μg/L) are consistent with the fact that no student declared taking testosterone or a steroid supplement. In addition, none of the students declared any endocrine disorders. Of the 74 urines, we were able to measure 5βA, 5αA, and 5βP values for all but 1. Sixty-eight of the samples provided satisfactory data on the first extraction of 10 mL of urine. For the other six, three provided an adequate signal on repeat analysis. Two samples required a 20-mL sample volume to obtain an adequate signal. In one sample with low diol concentrations (<25 μg/L), the voltage criteria were not met even with a 20-mL sample. Thus, our signal criteria were met on the first analysis with 10 mL of urine in 94% of the 74 cases, and all but 1 of the remaining samples provided satisfactory data on repeat analysis with 10 or 20 mL of urine. One person with one instrument can perform approximately three assays per week with a batch size of 20 samples plus 5 controls. The software to process the samples is somewhat cumbersome; thus, ∼2 days per week are needed for data processing.
The mean 5αA and 5βA values for our control group were −26.35‰ and −25.69‰, respectively. The mean δ values − 3 SD for 5βA and 5αA were approximately −28.4‰, which is similar to the δ values reported for synthetic testosterone (6)(7); however, this comparison should not be made. This is because the diols measured herein were acetylated, which makes the δ value more negative by approximately two units, whereas the synthetic testosterone (6)(7) was underivatized.
There are no comparable control group values in the literature; however, the baseline values in Fig. 6 of Shackleton et al. (3), which was based on analysis of 20 samples from individuals with mixed nationalities, are very similar to ours. Their lowest value for a diol was −28.2‰, whereas our lowest diol value was −27.89‰. Ueki and Okano (7) also measured 5αA and 5βA; however, their data are not directly comparable to ours because they used a correction factor to convert all measured δ13C values of the acetylated compounds to give values for underivatized steroids, which we could not estimate. The correction factor would be the same for 5αA and 5βA; thus, their difference of 2.6‰ between the means of 5αA (−16.4‰) and 5βA (−19.0‰) for 10 Japanese male samples can be compared to our difference of 0.66‰ (Table 3⇑ ). In addition, the value for 5αA was lower than 5βA in the study by Ueki and Okano (7), which is the opposite of what we found. We can also compare the range of values in our control subjects to the range reported by Ueki and Okano (7) in their Table 2⇑ for 20 healthy Japanese males. Our ranges for 5αA and 5βA were 2.9‰ and 3.9 ‰ (Table 4⇑ ), whereas they found ranges of 10.8‰ and 7.2‰, respectively. The striking difference between the ranges found in the two laboratories could reflect differences in method, calibration, ethnicity, or diet. The explanation that Japanese subjects have different steroid biochemistry and metabolism was deemed unlikely because equally large ranges of 11.0‰ (5βA) and 14.1‰ (5αA) were reported for urine samples from >350 Olympic athletes, and Olympic athletes are multiracial. In addition, Shackleton et al. (3) found similar values for Chinese and other nationalities, and we found no difference among four ethnic groups in our control group. That dietary factors could explain the differences was also considered, but diet is an unlikely explanation for the large differences in Olympic athletes because it is likely to take several weeks on a local diet before a measurable change in urinary steroid δ13C develops. Therefore, it is likely that our analytical method differs from that of Ueki and Okano (7).
The diversity of the control group was an advantage in that four major ethnic groups were present; however, it was a disadvantage because the n for each ethnic group was relatively small and therefore the statistical power of the GLM procedure was relatively low. Thus, we could not attribute any differences to ethnicity. Similarly, Shackleton et al. (5), who compared δ13C values for 20 individuals from 12 nationalities, did not attribute any of the differences to ethnicity or diet.
In the present study, the mean δ13C values for 5αA (−26.35‰) and 5βA (−25.69‰) in the control group were different (P <0.001), whereas in our previous study with only 10 subjects, we did not observe such a difference (15). In addition, it appears from the data in Table 5⇑ of the present study and from Table 1⇑ in our previous study (15) that testosterone administration may decrease 5αA more than it decreases 5βA. This difference occurs despite the fact that the percentage of conversion of a tracer dose of [14C]testosterone to 5β-androstane-3α,17β-diol (3.2%) is greater than the percentage of conversion to 5α-androstane-3α,17β-diol (1.2%) (18). Urinary 5α-androstane-3α,17β-diol is known to arise from both hepatic and peripheral metabolism, whereas urinary 5β-androstane-3α,17β-diol is considered to arise only from hepatic metabolism (19). After pharmaceutical doses of testosterone, 5αA may decrease more than 5βA because peripheral paths to 5α-androstane-3α,17β-diol are favored. However, there is also a path from dehydroepiandrosterone sulfate to 5β-androstane-3α,17β-diol that does not pass through testosterone (20). Utilization of this path could explain why 5βA is higher than 5αA. Finally, the observed differences may be an analytical artifact. Similarly, the higher mean δ13C value for 5βP (−24.26‰) might be related to metabolism or artifacts.
athletes with elevated T/E ratios
To clarify the status of an athlete with an elevated T/E ratio, the International Olympic Committee recommends that additional samples be obtained (9). The relevant sport authority typically charts the course of the urine T/E values over time and interprets the data pattern to decide whether to penalize the athlete. Unless there is a record of three or more past samples, this involves collecting additional samples. There are no guidelines stating how many samples are needed, how often they should be collected, and for how long, and there are no published criteria for determining whether the elevated T/E ratio is natural or is attributable to taking an exogenous substance. The data in Tables 1–3⇑ ⇑ ⇑ and 5 support the argument that by measuring δ13C, the decision to penalize the athlete can be made on a single sample. The six athletes described in Table 5⇑ were undergoing such testing, although at the time of the analysis the laboratory was not aware that subjects 1 and 2 were permitted testosterone users. No additional information was available on the status of athletes 3–6; however, inspection of the data for subject 3 indicates that it is reasonable to classify him as a testosterone or testosterone precursor user.
The six subjects in Table 5⇑ with increased T/E ratios may be characterized by the concentration of testosterone in their urines and by their 5βA and 5αA values and the associated differences and ratios to 5βP. A classification based on δ13C values, differences, and ratios revealed that athletes 1–3 were outside the control range ± 3 SD except for 5βP. For athletes 1 and 2, this fits with our understanding of the basis of the carbon isotope ratio method, specifically, that after testosterone administration only the δ13C values of testosterone and testosterone metabolites will decrease and that those of other urinary steroids will not. Because 5β-pregnane-3α,20α-diol is not a metabolite of testosterone, its δ13C values are not expected to change with testosterone administration (3)(4)(15).
Applying the means ± 3 SD criteria to the δ13C data for athlete 3 requires classifying him as a testosterone or testosterone precursor user. This possibility is further supported by the very high urine T/E ratios and testosterone concentrations. The z-scores for his urine testosterone concentrations range between 2.2 and 2.9, which is a further indication of testosterone or testosterone precursor administration. Once the GC-C-IRMS method is validated to the point of full acceptance by the sport and legal community, we expect that athlete 3 could be classified as a user on one urine showing the pattern of a high T/E ratio, high testosterone concentration, and δ13C values less than − 3 SD.
Athletes with naturally elevated urine T/E ratios would be expected to have persistently elevated T/E ratios, unremarkable urine testosterone concentrations, and values for 5βA and 5αA that are within ± 3 SD of the mean of the control group. The data for athletes 5 and 6 fit this pattern. Athlete 4 also fits, but some of his urine testosterone concentrations were relatively high. The z-score for the values 383 and 394 μg/L was ∼2.3, and all but one of the others were >1.3. Until additional data are available, athlete 4 has been classified as having a naturally elevated T/E ratio, although we cannot completely exclude the possibility that athlete 4 is a testosterone or testosterone precursor user with normal δ13C values.
The data on mean z-scores in Table 5⇑ reveal that both the differences and ratios are better indices of testosterone use than the absolute δ13C values, 5βA and 5αA. This is not unexpected given that 5βP correlates with both 5βA and 5αA. Furthermore, the difference 5βP − 5βA and the ratio 5βA/5βP are more robust indicators of testosterone use than the corresponding differences 5βP − 5αA and the ratio 5αA/5βP. This follows from the finding (Tables 3⇑ and 5⇑ ) that the δ13C values for 5αA were lower than the values for 5βA.
In conclusion, a fundamental issue in doping control is whether GC-C-IRMS techniques will resolve ambiguities in the interpretation of the T/E test. The methods presented herein provide compelling evidence that the test has excellent precision and that when it is coupled with strict system suitability and batch acceptance criteria, it can be used in a routine fashion to obtain valid δ13C data. By comparing the δ13C values of a control group to data obtained from athletes, it is possible to make informed decisions regarding the origin of urinary 5β-androstane-3α,17β-diol, 5α-androstane-3α,17β-diol, and 5β-pregnane-3α,20α-diol.
We are grateful for financial support from the National Collegiate Athletic Association, the National Football League, and the United States Olympic Committee. We thank K. Schramm for technical assistance.
1 Mean significantly different from 5βA.
2 Mean significantly different from 5βA and 5αA.
1 Cauc., Caucasian; A.A., African American.
1 Day of urine collection relative to the first sample (day = 0).
2 [T] and [E], testosterone and epitestosterone.
3 Cells >3 SD, shaded; cells <3 SD, solid outline.
4 Value is an approximation because of large error in measurement of epitestosterone.
5 Unable to measure because of low response (<0.3 V).
↵1 Nonstandard abbreviations: GC-C-IRMS, gas chromatography-combustion-isotope ratio mass spectrometry; T/E, testosterone/epitestosterone ratio; 5βA, δ13C value for 5β-androstane-3α,17β-diyl diacetate; 5αA, δ13C value for 5α-androstane-3α,17β-diyl diacetate; 5βP, δ13C value for 5β-pregnane-3α,20α-diyl diacetate; and QC, quality control.
- © 2001 The American Association for Clinical Chemistry