The results of the assay characterization for Stan and LGD-4033 are summarized in Table 1. No interfering signals occurred either in the sf samples from ten different boars or in the bp from ten different humans, and the method was considered selective. For the purpose of the study, qualitative detection and calibration curves supporting the estimation of concentration levels of selected analytes within the defined working range was considered sufficient with the assay offering LODs between 10 and 40 pg/mL and demonstrating linear responses over the concentration range of the analytical procedure for sf and human bp. Carryover and precision had acceptable values and were considered appropriate for this application. Recovery and matrix effects were determined, in accordance with WADA guidelines [20], at mid-range concentration levels using spiked blank matrix samples. Porcine/boar blood is often described as similar in quality to human blood and used in an analogous way, although there are differences in composition [21]. For the validation of the method, the qualitative commonality is considered sufficient; a limitation has been though that the validation results only apply to Stan and LGD-4033, and assay characteristics might be transferable to the metabolites only to a limited extent.

Reference materials of Stan and metabolites were spiked in blank sf and bp to identify retention time and diagnostic product ions (Table 2). The precursor ion of Stan remains intact to a substantial level even at high collision energies and dissociates into specific product ions such as m/z 81.0447 that is characteristic for the pyrazol moiety. Consequently, the product ion at m/z 97.0396 is indicative for an oxygenation in that residue as observed in case of 3′-OH-stan. The product ion at m/z 145.0760 is typically observed with a hydroxylation of Stan at position C-4 and only detected at lower NCEs (Fig. 1E, F) [22]. The reference standard of 16α-OH-Stan elutes considerably earlier than 16β-OH-Stan. The elution order agrees with the sequences described in the literature, although the therein reported retention time delta was smaller. [23, 24] A possible explanation is the fact that a comparably low LC flow rate (200 µL/min) was chosen in order to optimize the separation of target analytes.

ESI-product ion mass spectra ([M + H].+  = 345) of (A) unknown hydroxy-Stan metabolite (boar sf sample), (B) 3′OH-Stan (reference standard in boar sf), (C) 16β-OH-Stan (boar sf sample), (D) 17-Epi-16α-OH-Stan (reference standard in boar sf), (E) 4β-OH-Stan (reference standard in boar sf), (F) 4α-OH-Stan (reference standard in boar sf), (G) 16α-OH-Stan (boar sf sample) recorded with a CE of 72% for A–D and G, CE of 45% for E and F, on an Orbitrap Exploris 480. (H) Structure of Stan with arrows indicating metabolic reactions (adapted and modified from Scarth et al. [24])

In all analyzed in vivo post-dose bp and sf samples, non-metabolized Stan was detected. Concentration levels of the drug in both matrices were nearly identical. The semi-quantitatively estimated concentration of Stan after 24 h were 0.4 ng/mL in bp and 0.25 ng/mL in sf, after 48 h 0.04 ng/mL (bp) and 0.05 ng/mL (sf), and after 72 h 0.02 ng/mL (bp) and 0.01–0.02 ng/mL (sf).

Around twenty phase I and II metabolites from seven different metabolic pathways were detected (Table 2). Of these, four metabolites (3′-OH-Stan, 16α-OH-Stan, 16β-OH-Stan, 3′-OH-Stan-Gluc) could be assigned using reference substances. The signal intensities in the extracted ion chromatograms of most metabolites decreased continuously from the first (24 h) to the last (72 h) post-administration sample and were generally higher in bp than in sf for all metabolites, except S-M4-b (Fig. 2). While the intensity of almost all metabolites was higher in bp than in sf, the concentration and intensity of Stan was similar in both matrices at all three sampling time points.

Abundance ratio of Stan and its metabolites normalized to the ISTD after 24 h (blue), 48 h (orange), and 72 h (green) in sf and bp. Metabolites are specified in Table 2

Three hydroxylated metabolites (m/z 345.2537) were detected at 4.98, 6.45, and 7.12 min, which is shown in Fig. 3 for sf and bp after 24 h. The peak at 5.50 min (Fig. 3) is attributed to in-source dissociation of 3′-OH-Stan-gluc. The metabolite designated as S-M3-a at 4.98 min was assigned to 16α-OH-Stan by reference material and was detected in bp up to 72 h, whereby in sf the analyte was observed only in the 24-h specimen. The metabolite of highest abundance exhibiting a protonated molecule at m/z 345.2537 after 24 h eluted at 6.45 min (S-M3-b), which could not be assigned to any reference substance. However, hydroxylation at C-3′ or C-4 was excluded since the typical product ions (m/z 97.0396 or m/z 145.0760) were not detected in the HRMS/MS spectrum (Fig. 1A). Hydroxylation at C-15, C-6, or another yet unknown site would also be possible as suspected in equine urine samples but without further identification [23, 24]. In greyhound urine, 6α-OH-Stan was determined as the major metabolite after intramuscular injection of Stan [25]. Another peak with product ions consistent with mono-hydroxylation was detected at 7.12 min (S-M3-d, Fig. 3) and was identified as 16β-OH-Stan using a reference standard. This peak had the highest intensity of the hydroxylated metabolites of Stan in bp samples after 24 h and was detectable in all bp samples, whereas in sf such a peak could only be detected after 24 h (Fig. 2).

Full-MS extracted ion chromatogram (left) and MS/MS extracted ion chromatogram (right) of sample before administration (combined bp and sf, top) and post-administration sample collected 24 h after application of anabolic agents, indicating the presence of several stanozolol hydroxy metabolites (left) and ISTD (right) in bp (middle) and sf (bottom)

The metabolites referred to as S-M4 were observed with protonated molecules at m/z 361.2486, which corresponds to bishydroxylated species of Stan. Two of such metabolites were detected in bp after 24 h (2.24 and 4.34 min), with the latter also being detected in sf. There are several reports where bishydroxy-Stan was detected in human and equine urine [16, 24, 26], but since the signal intensities were low and the product ion mass spectra not sufficiently informative, these metabolites were not further characterized.

A metabolite exhibiting a protonated molecule at m/z 343.2380, indicating the introduction of an oxo group into Stan, was observed in all post-dose sf and bp samples (S-M5, 24–72 h). The MS/MS spectrum ([M + H]+ = 343.2380) shows a product ion at m/z 257.2012, which can be attributed to a cleavage of the precursor ion between C-14/C-15 and C-13/C-17 (Fig. 4A), which is consistent with metabolites found and confirmed in human and horse urine that were identified as 16-oxo-Stan with reference standards and in vitro experiments, respectively [24, 27].

Product ion mass spectra and possible structure of molecules (A) S-M5 as 16-oxo-Stan ([M + H]+  = 343) and (B) S-M6 ([M + H]+  = 359)

One metabolite (S-M6) was determined with an accurate mass and elemental composition corresponding to the introduction of an oxo function and a hydroxyl group. The protonated molecule was found at m/z 359.2329 in bp and sf samples collected after 24 h only. In consideration of the rationale presented for S-M5, the product ion at m/z 273.1971 can also be attributed to the cleavage between C-14/C-15 and C-13/C-17, in combination with a hydroxylation at the steroidal B/C-ring scaffold (Fig. 4B), given the abundant product ion at m/z 81.0447 and the lack of a distinct product ion at m/z 145.0760. Three different structures are proposed in the literature that could fit a metabolite with these characteristics, i.e., a carboxylation at C-18 or C-20 [26], hydroxy-oxo reaction at C-20 and C-16 [28], and two hydroxylations plus a reduction [29].

The employed LC-HRMS/MS-based method also allowed for the detection of known phase II metabolites [30]. Such phase II metabolites of steroids commonly exhibit an ionization behavior similar to that of the unconjugated analytes [31], which facilitated the detection of three different hydroxylated and glucurono-conjugated metabolites (at m/z 521.2857) and one hydroxylated and sulfo-conjugated species (at m/z 425.2105). Two major peaks (4.50 min (S-M2-a), 5.45 min (S-M2-b)) with protonated molecules at m/z 521.2857 were observed in bp and sf. The S-M2-b was determined by means of reference material to represent 3′-OH-stan-gluc and, based on literature data, S-M2-a was assigned to 16β-OH-Stan glucuronide [27]. The intensity of signals in extracted ion chromatograms was higher in bp than in sf for all glucuronides, and differences in the relative distribution of peak intensities in bp and sf were observed as exemplified with samples collected 24 h post dosing (Fig. 2).

The hydroxylated and sulfo-conjugated metabolite (at m/z 425.2105, 3.35 min, S-M7-b) was present in all post-administration samples of bp and sf. In bp, two additional signals immediately before (3.07 min) and after (3.78 min) S-M7-b (S-M7-a,-c) with the same accurate mass and similar product ions were detected in the 24 h-specimen, suggesting the formation of further hydroxylated and sulfonated stanozolol metabolites in vivo. In consideration of all available mass spectrometric characteristics, S-M7-b is proposed to exhibit a hydroxyl group at C-16, which is in accordance with observations of Balcells et al., who demonstrated the superior stability of 16-OH-Stan sulfate compared to sulfates of 3′-OH- and 4-OH-stanozolol [32].

Fourteen phase I and phase II metabolites from seven different metabolic pathways, which have been thoroughly investigated previously, were detected for LGD-4033 (Table 3) [18, 33, 34]. The intensities of most metabolites of LGD-4033 are higher in bp than in sf, and abundances decrease from 24 to 72 h post-administration for all but one metabolite, which is discussed in more detail (see below, Fig. 5).

Abundance ratio of LGD-4033 and its metabolites normalized to the ISTD after 24 h (blue), 48 h (orange), and 72 h (green) in sf and bp. Due to the considerable difference in the intensity of the metabolite L-M6 compared to the other metabolites, a secondary axis (right) has been added. Metabolites are specified in Table 3

The most abundant product ion of LGD-4033 under the chosen conditions was m/z 267.0751, which was therefore used as quantifier ion. In addition to the identification of LGD-4033 in bf and sf, a second peak (L-M1) eluting shortly after the intact and unmetabolized compound was observed only in the first sample after 24 h and attributed to the LGD-4033 epimer as described by Wagener et al. [18]. The concentration of LGD-4033 was approximately 0.68 ng/mL in sf and 2.0 ng/mL in bp after 24 h, 0.15 ng/mL in sf and 0.63 ng/mL in bp after 48 h, and 0.04 ng/mL in sf and 0.21 ng/mL in bp after 72 h. The intact drug was detected in all post-administration bp and sf samples although, unlike Stan, concentrations were considerably lower in sf than in bp, which supports the assumption that anabolic agents (and their metabolites) as well as other drugs of specific physicochemical nature can partition differently between blood and ejaculate.

Two metabolites referred to as L-M2-a and L-M2-b were identified as glucuronic acid conjugates of LGD-4033 (and its epimer), featuring a deprotonated molecule at m/z 513.1019 and the characteristic loss of 176 Da [34]. Unlike for most other metabolites of LGD-4033 in this study, the intensity of L-M2-a and L-M2-b was higher in sf than in bp after 48 h and 72 h (Fig. 5). In a recently published study, the metabolic signatures of the three matrices urine, serum and sf were investigated and compared [35]. Sf was found to resemble the signatures of urine (polar compounds such as glucuronides) and serum (non-polar compounds) which could explain why higher abundances of glucuronides are present in seminal fluid than in blood. Alternatively, minute contaminations of sf with urine in the 48-h sf collection could be considered as glucuronidated metabolites of LGD-4033 predominate in urine. A recent study has shown that human sf can contain small volumes of urine, which is why it cannot be excluded that a fraction of the observed analytes originates from this source [36]; however, considering the difference in average volumes of human und porcine ejaculate, the percentage of urine potentially introducing metabolites into ejaculate in the present study is presumably low.

The two metabolites L-M3-a and L-M3-b are related to a deprotonated species at m/z 351.0574, which indicates hydroxylation and dehydrogenation (i.e., formation of an oxo-metabolite), which was confirmed by in-house synthesized reference material [18]. A third signal in the extracted ion chromatogram with m/z 351.0574 is observed at the RT of bishydroxylated LGD-4033 and has been attributed to an in-source fragmentation rather than a factual metabolite (Fig. 6). The decrease in the intensities of L-M3-a from 24 to 72 h is slow compared to the other metabolites of LGD-4033 and, consequently, its intensities are relatively high after 72 h (Fig. 5). Hence, this metabolite might be suitable to detect LGD-4033 also for a prolonged period of time, which however necessitates further studies including more animals and longer sampling times.

Extracted ion chromatograms of a sample before LGD-4033 administration (combined bp and sf, top) and post-administration samples collected 24 h after application, indicating the presence of three LGD-4033 metabolites (L-M3) in bp (middle) and sf (bottom)

The deprotonated molecule [M-H] − of the metabolites termed L-M4-a and L-M4-b was found at m/z 353.0730, indicating mono-hydroxylation of LGD-4033. These metabolites were detected in all post-administration bp and sf samples, and the diagnostic product ion at m/z 185.0332 localizes the hydroxylation at the pyrrolidine moiety, in agreement with earlier publications [18, 33, 34, 37].

Three metabolites (L-M5-a, -b, -c) exhibiting an [M-H]− at m/z 355.0887 matched previously reported metabolic products obtained via hydroxylation and opening of the pyrrolidine ring [18, 33, 37]. The retention time of L-M5-a (Table 3, 8.68 min) is, under the employed chromatographic conditions, identical to that of the bishydroxylated metabolite L-M6 (Table 3, 8.68 min), but here a potential in-source dissociation mimicking a metabolite formation is unlikely.

The metabolite L-M6, assigned to bishydroxylated LGD-4033 with the determined m/z 369.0679, was detected at highest relative abundance of all metabolites in post-administration bp and sf samples. While the intensity of most metabolites decreased continuously from the first to the last sample in both matrices, the intensity of L-M6 peaked with the second sample (48 h) before decreasing at 72 h (Fig. 5). Analyses of human urine of LGD-4033 administration studies were re-analyzed with this method, i.e., without hydrolysis, demonstrating that unconjugated trihydroxylated and bishydroxylated metabolites of LGD-4033 are present in urine (data not shown) [18]. Therefore, contamination with urine (even if at lowest amounts) might contribute to analyte abundances as mentioned above, and thus participate in the increase of L-M6 in the 48-h sample. Hence, in future studies, the collection of corresponding urine samples in addition to bp and sf is advisable for a more complete to better understand metabolite distribution.

The metabolites L-M7-a,-b,-c (m/z 385.0625) are assigned to threefold hydroxylated LGD-4033. The intensities of the L-M7 metabolites in bp and sf were low compared to other metabolic products. In the 24-h samples, the intensities of L-M7-a,-b,-c were higher in bp than in sf, but the 48-h samples exhibited higher intensities of L-M7-a,-b,-c in sf.