Fragment-Resolved Carbon and Hydrogen Isotope Analysis of Urinary Steroids by Orbitrap Mass Spectrometry

Abstract

Compound-specific stable isotope analysis (CSIA) is widely used to distinguish sources and transformation pathways of organic compounds. In steroid CSIA each analyte is represented by a single molecular-average isotope value, which can limit discrimination when sources overlap in isotopic composition. This limitation is especially important in anti-doping analysis, where molecular-average δ¹³C measurements can be inconclusive for compounds whose C-isotope composition falls within the endogenous range. Here, we present an Orbitrap-based method for fragment-resolved intramolecular isotope analysis of steroids. Using urinary androsterone as a model analyte, we measured δ¹³C values for four diagnostic ions and δD for one fragment following controlled administration of androstenedione and testosterone undecanoate. The resulting multidimensional isotope fingerprints reveal responses not evident from bulk δ¹³C measurements alone, including cases in which administered androgen lies near the endogenous δ¹³C range. Time-dependent differences among fragment ions further suggest sensitivity to changes in metabolic routing and pool mixing during steroid transformation. These results establish fragment-resolved isotope analysis by Orbitrap mass spectrometry as a promising extension of conventional steroid CSIA and a potential new tool for forensic steroid source discrimination, with possible applications to environmental steroids and geological sterols and steranes.

Introduction

Compound-specific stable isotope analysis (CSIA) is a powerful tool for distinguishing sources and transformation pathways of organic compounds (e.g., refs , ). In steroid analysis, CSIA is widely used in antidoping laboratories to test whether urinary steroids reflect endogenous biosynthesis or administration of synthetic androgens. This approach compares the molecular-average carbon isotope composition of target steroids with endogenous reference compounds, exploiting the fact that many synthetic steroids are derived from plant precursors and can be depleted in ¹³C relative to naturally produced human steroids. ,− CSIA has therefore become an important confirmatory method in steroid doping control, particularly when urinary steroid profiles suggest possible testosterone or androgen administration. ,,,

Despite its utility, conventional steroid CSIA reduces each analyte to a single molecular-average isotope ratio, potentially obscuring diagnostically useful intramolecular isotope information. This can limit discrimination when endogenous and administered steroids overlap in δ¹³C composition or when metabolic heterogeneity complicates interpretation. ,−

A natural extension of steroid CSIA is to perform intramolecular (site-specific isotope analysis, SSIA) or fragment-resolved isotope analysis. Measuring isotope ratios in multiple diagnostic fragments of the same steroid yields a multidimensional isotope fingerprint rather than a single molecular-average value. Such fingerprints could improve source discrimination and reveal whether different structural regions of a molecule respond differently to metabolic routing, enzymatic transformation, or pool mixing. The analytical challenge is that these measurements must be made at natural isotope abundance with high precision on low-abundance compounds purified from complex biological matrices. The resulting isotope values must also be comparable across laboratories through normalization to internationally accepted reference scales and have a sufficiently low uncertainty to resolve meaningful differences among samples.

High-resolution Orbitrap mass spectrometry provides a promising route to this type of measurement. Orbitrap-based isotope methods have been used to measure isotope ratios in specific atomic sites or structurally informative fragments of organic molecules, extending isotope analysis beyond conventional compound-specific measurements. Intramolecular variations in δ¹³C or δD can reflect biosynthetic mechanisms, reaction kinetics, or environmental influences that are averaged out in bulk or conventional CSIA. Previous studies have shown that such approaches can reveal source- and pathway-dependent isotope patterns in amino acids, acetate, meteoritic organic compounds, and other small molecules that are not apparent from molecule-average compositions alone. − Compared with NMR or Infrared spectroscopy site-specific isotope methods, Orbitrap-based approaches offer practical advantages in sensitivity, acquisition speed, generalizability across diverse compound types, and applicability to low-abundance analytes, although they require careful control of fragmentation, spectral purity, signal matching, instrumental drift, and isotope-ratio calibration. ,

Steroids are a demanding target for fragment-resolved isotope analysis. They are structurally complex, occur at low concentrations in biological samples, and require purification from matrices that can introduce spectral interferences and isotope-ratio bias. In addition, structurally informative steroid fragments must be generated reproducibly and assigned with sufficient confidence to support interpretation. Demonstrating fragment-resolved isotope analysis of urinary steroids therefore requires an analytical workflow that integrates sample purification, high-resolution mass spectrometry, fragment assignment, standard–sample bracketing, drift correction, and assessment of measurement precision.

Here, we present an Orbitrap-based method for fragment-resolved carbon and hydrogen isotope analysis of urinary steroids, using androsterone analyzed as its acetate derivative (C₂₁H₃₂O₃) as a model analyte. Androsterone (C₁₉H₃₀O₂) is a major urinary metabolite of testosterone and is routinely relevant to steroid profiling and anti-doping analysis. Rather than measuring only a single whole-molecule δ¹³C value, we resolve isotope ratios for multiple diagnostic androsterone fragments that sample different regions of the steroid carbon skeleton. In controlled oral administration experiments involving androstenedione and testosterone undecanoate, we measure δ¹³C values for four fragment ions and δD for one fragment in serial urine samples, yielding multiple isotope observables from purified urinary androsterone at low concentration (1 μM).

We use these data to ask two analytical questions: first, whether fragment-resolved isotope measurements provide discriminatory information beyond conventional whole-molecule CSIA; and second, whether different steroid fragments record distinct isotope responses to source mixing, metabolic routing, and transformation. The resulting fragment-specific isotope fingerprints reveal temporal patterns not evident from whole-molecule δ¹³C measurements alone, including cases in which the administered steroid lies near the endogenous δ¹³C range. These results establish Orbitrap-based fragment-resolved isotope analysis as an extension of conventional steroid CSIA and provide a framework for applying intramolecular isotope measurements to other complex biomolecules, including environmental steroids and geological sterols and steranes.

Methods

Experimental Procedure

Two in vivo administration experiments were conducted with steroids of known molecular-average δ¹³C_VPDB values. A healthy male volunteer received 80 mg androstenedione (δ¹³C_VPDB = −31.8‰), and a second volunteer received 40 mg testosterone undecanoate (TU; δ¹³C_VPDB = −27.0‰), both orally in ethanol:water. Urine was collected before dosing and at 6-, 10-, 22-, 28-, and 38-h post-dose for androstenedione and 9-, 13-, 24-, 30-, and 37-h post-dose for TU.

Urine Sample Preparation and Extraction

To obtain sufficient amounts of androsterone for downstream analyses, larger urine volumes were processed by pooling multiple aliquots, typically six 20 mL aliquots, and applying the same extraction and clean-up procedure, detailed in , , to all samples (for more details see the Methods section in Supporting Information).

Semi-Preparative HPLC Cleanup and Purification of Androsterone

Steroid fractions were purified by reversed-phase HPLC as described by Piper et al. Separation was performed on an RP18 column (XBridge Shield RP18, 4.6 × 250 mm, 5 μm) at a flow rate of 1.0 mL/min, with UV detection at 195 nm. A linear gradient from acetonitrile/water (40/60, v/v) to 60% acetonitrile over 18 min was followed by an increase to 98% acetonitrile, a hold period, and column re-equilibration. Fraction collection windows were established using authentic steroid standards. Androsterone was collected as a discrete fraction (17.4–19.0 min), evaporated to dryness, and prepared for derivatization.

Derivatization

Dried HPLC fractions were derivatized to their acetate esters by adding pyridine and acetic anhydride and incubating at 70 °C for 60 min following , . After derivatization, samples were evaporated to dryness under a gentle stream of air and reconstituted in cyclohexane for GC-based analyses (GC-MS profiling, preparative GC, and GC-C-IRMS) or in acetonitrile (with 1% formic acid) for direct infusion Orbitrap MS.

A commercially available androsterone standard was derivatized to the acetate using the same procedure, as applied to urine samples, and used as the in-house androsterone acetate reference material. This standard was calibrated for its bulk carbon (δ¹³C_VPDB = −33‰) and hydrogen (δD_VSMOW = −360.5‰) isotope values using GC-IRMS.

Androgen Identification and Quantification by GC-MS

Urinary androgens were identified and quantified by GC-MS using an Agilent 7890 GC (Waldbronn, Germany) equipped with an Agilent J&W Scientific DB-17MS column (30 m length, 0.25 mm inner diameter, 0.25 μm film thickness) and coupled to an Agilent 7000 triple quadrupole mass spectrometer (for more details see Supporting Information).

GC-C-IRMS for Bulk δ¹³C and δD Analysis

Bulk δ¹³C_VPDB and δD_VSMOW values of androgens and the in-house androsterone acetate standard were determined by GC-IRMS following established protocols described in , (for more details see Methods section in the Supporting Information). Molecular-average δ¹³C values were calibrated using the certified steroid isotope reference materials described by Zhang et al., which are traceable to the VPDB scale.

Orbitrap Fusion Lumos Mass Spectrometry

The androsterone acetate standard and derivatized samples were introduced into the Orbitrap Fusion Lumos mass spectrometer (Thermo Fisher Scientific, Bremen, Germany) using a dual-inlet sample introduction system described by Hilkert et al. Purified androsterone acetate fractions and the derivatized standard were dissolved at approximately 1 μM in acetonitrile containing 1% formic acid and loaded into the two syringes of the dual-inlet system. These solutions were infused at a flow rate of 30 μL/min. Briefly, 10 mL Hamilton syringes (part number 1157219) were filled and mounted on a syringe pump (part number F100T2) positioned atop the Orbitrap housing. Samples were delivered through PEEK tubing joined by Luer locks (part number Z227307) and PEEK connectors (part numbers 220314), routed through a diverter valve (part number MXT715-000). This configuration enabled switching between paired analytes (e.g., sample and standard, or sample and blank). The valve outlet was connected by PEEK tubing to the heated APCI source (part number OPTON-32101), which was operated in positive ion mode.

Source conditions were as follows: ion transfer tube temperature, 320 °C; RF lens, 70%; and sweep gas = 0 au. Sheath gas, auxiliary gas, and spray voltage were optimized daily to maintain a stable signal with total ion current (TIC) variation of less than 5% over a measurement block and a typical TIC of approximately 3 × 10⁷ counts.

Under heated APCI conditions, androsterone acetate was protonated and underwent reproducible in-source fragmentation to produce two diagnostic fragment ions (Scheme ): (i) the m/z 255.21127 ion generated by the loss of the acetate group and the carbonyl oxygen, representing the androsterone carbon skeleton (C₁₉H₂₇ ⁺, hereafter 255 Da); (ii) the m/z 124.08882 Da fragment (124 Da), generated by cleavage of the C9–C11 and C8–C14 bonds (Scheme ). The 255 Da ion was monitored over the mass window of m/z 254–258, and the 124 Da over m/z 123.8–126.4. For MS measurements, ions were accumulated in the C-trap using automatic gain control (AGC), with a normalized AGC target of 2% (∼2 × 10⁵ charges) and a maximum injection time of 8000 ms, before Orbitrap analysis.

1. Fragmentation Scheme and Structural Assignments for Androsterone Acetate Ions Measured by Orbitrap MS .

^a (A) Under heated APCI conditions, androsterone acetate undergoes in-source fragmentation, yielding the C₁₉ skeleton ion at m/z 255.21127 (255 Da) following loss of the acetate group and elimination of the ketone oxygen as H₂O. A second diagnostic fragment at m/z 124.08882 (124 Da), formed by cleavage of the C9–C11 and C8–C14 bonds and sampling ring D, the ketone group, and C11–C12 from ring C; (B) Subsequent CID (MS/MS) of the 255 ion produces two diagnostic product ions used for fragment-specific isotope analysis: m/z 199.14868 (199 Da), formed by cleavage of the C1–C10 and C4–C5 bonds and sampling rings B–D; m/z 173.13303 (173 Da), produced by cleavage of the C8–C14 and C12–C13 bonds and sampling rings A–B plus C11–C12 of ring C. Ring labels (A–D) indicate the steroid backbone regions represented by each fragment. Fragment assignments are based on exact m/z values and plausible steroid backbone cleavages consistent with APCI­(+)/CID behavior. ,

For MS² (MS/MS) measurements, precursor ions within m/z 256 ± 3 (m/z 253–259) were isolated in the quadrupole to include the 255 Da ion and its ¹³C-substituted isotopologues and subjected to collision-induced dissociation (CID, 40% normalized collision energy) in the ion-routing multipole. CID generated two additional diagnostic fragment ions: (i) the m/z of 199.14868 (199 Da) fragment ion is formed by cleavage between C1–C10 and C4–C5, (see structure in Scheme ), and (ii) 173.13303 (173 Da) that formed by cleavage between C8–C14 and C12–C13 (see structure in Scheme ). Product ions were then accumulated in the C-trap (AGC target 150%; maximum injection time 8000 ms) and analyzed in the Orbitrap. Product-ion spectra were acquired for the fragments at m/z 199 and 173 over m/z ranges 198–202 and 172.5–175.5, respectively.

The Orbitrap analyzer was operated at resolving powers of 60,000 for the 255, 199, and 173 Da ions and 120,000 for the 124 Da ion. Each spectrum represents the sum of 10 microscans. At a resolving power of 60,000, the D contribution to the M + 1 peak used for ¹³C/¹²C determination is not fully baseline-resolved from the ¹³C isotopologue for the 255, 199, and 173 Da ions. Because the D contribution is minor relative to ¹³C at natural abundance and measurements were performed using intensity-matched standard-sample bracketing, any residual coalescence is expected to have a negligible effect on reported δ¹³C values.

Overall, four diagnostic fragment ions ( m/z 255, 199, 173, and 124 Da) were analyzed for the ¹³C/¹²C ratios. The 124 Da fragment was additionally analyzed for the D/H ratios. Each of these fragment ions corresponds to a specific structural component of the androsterone molecule. Their structures and fragmentation pathways are summarized in Scheme . Product-ion assignments were made using the measured exact m/z values, elemental compositions, and plausible steroid backbone cleavages consistent with reported APCI fragmentation behavior of steroids and androsterone-related analytes (e.g., dehydration/water-loss and ring-cleavage pathways , ).

Data Analysis

Raw Orbitrap transients were processed with Thermo Fisher’s IsoX software to generate mass spectra. Extracted spectra were then processed using IsoXL, a graphical interface built on the isoorbi R package with occasional manual inspections in Microsoft Excel, following previously described procedures ,, to calculate molecular- and fragment-specific isotope ratios from the measured isotopologue abundances. IsoXL provides tools for quality assessment, outlier filtering, isotope-ratio calculation, and aggregation of scan-level measurements into summary isotope ratios.

Each sample was analyzed in blocks of 5–7 replicate measurements using a standard-sample-standard sequence with the synthetic androsterone acetate reference material. The measured isotope ratios were converted to δ values by comparison with the corresponding fragment ion of the synthetic androsterone acetate standard analyzed immediately before and after each sample (standard–sample–standard bracketing), according to

$$\delta =[(R_{\mathrm{sample}}/R_{\mathrm{standard}})-1]\times 1000$$ 1

where R is the measured isotope ratio of the molecular or fragment ion. This normalization corrects for short-term instrumental drift and any constant isotope effects associated with ion transmission or fragmentation.

Carbon and hydrogen isotopic compositions are reported as δ¹³C and δD values, in per mil (‰), relative to the corresponding fragment ion of the synthetic androsterone acetate standard analyzed in the same sequence. For carbon, we report δ¹³C for the molecular ion (255 Da) and for three fragment ions (199, 173, and 124 Da). For hydrogen, we report δD only for the 124 Da fragment because the mass-resolution and sensitivity requirements for accurate δD measurements of the heavier fragments could not be met with the available analyte amounts and acquisition times.

The synthetic androsterone acetate standard was calibrated for its molecular-average δ¹³C value on the VPDB scale, determined by GC-C-IRMS. However, fragment-specific carbon isotope values of this reference material are presently unknown. Therefore, fragment-specific δ¹³C values reported here, including those of 255 Da fragment, are expressed relative to the corresponding fragments of the in-house acetate standard rather than an international isotope scale (namely VPDB or VPDB-LSVEC scale).

Potential bias arising from ion intensity, unresolved isotopic coalescence, fragmentation efficiency, fragment-specific response, or other sources of instrumental nonlinearity was minimized by matching sample and standard signal intensities within a factor of 1.2 and analyzing all measurements using standard–sample–standard bracketing. Intensity matching reduces isotope-ratio bias caused by coalescence, including minor unresolved contribution of D-substituted isotopologues to the M + 1 signals used for ¹³C/¹²C determination, whereas standard-sample-standard normalization corrects for short-term instrumental drift. Because the sample and standard are chemically identical and undergo identical fragmentation pathways, any constant fragment-specific isotope effects are common to both and cancel during normalization. Internal precision was assessed from replicate measurements within each analytical block, and uncertainties are expressed as relative standard error (RSE), calculated as

$$\mathrm{RSE}=\mathrm{SD}/\mathrm{SQRT}(\mathrm{n})$$ 2

where SD is the standard deviation of these replicate isotope values and n is the number of replicates. Overall methodological reproducibility, for each fragment, was assessed by repeated analysis of the standard in a zero-enrichment (“self-comparison”) configuration.

The measurement protocol used here yielded precise abundances of doubly ¹³C-substituted (“clumped”) isotopologues of the same fragment ions. However, we have not yet performed the additional calibration and validation experiments needed to establish the accuracy of those measurements, and therefore we do not report or interpret clumped-isotope results in this study.

To test the internal consistency of our data set and aid modeling of the isotopic evolution of androsterone following androstenedione administration, we used first-order mass balance. Specifically, we estimated the maximum contribution of changes in the ¹³C content of the m/z 124 fragment ion to variation in the whole-skeleton carbon isotope composition measured by the m/z 255 ion according to

$$\mathrm{\Delta }\delta {}^{13}\mathrm{C}_{255}\approx (8/19)\times \mathrm{\Delta }\delta {}^{13}\mathrm{C}_{124}$$ 3

This calculated value was then compared with the difference between the observed m/z 255 value at the final point in our sampling time series (38 h) and the value predicted by an exponential fit to the m/z 255 time series. This fit serves as an empirical null model in which the whole-skeleton isotope signal is assumed to reflect only pulse-chase-style mixing and clearance of exogenous-derived androsterone toward the endogenous baseline, without additional localized isotope effects. Any residual difference between the exponential fit and the measured m/z 255 value should be consistent with any carbon isotope changes affecting the m/z 124 fragment, which is subject to additional isotope effects that may arise through KIEs in chemical reactions (see Note S1 in the Supporting Information for details).

Results

Androgen Concentrations following Androstenedione and Testosterone Administration

Urinary concentrations of androgens and downstream metabolites were quantified by GC–MS to provide metabolic context for the isotope results reported below.

Androstenedione Administration

Pre-administration baseline samples (18 h pre-administration) showed androsterone (A) concentration of 418 ng/mL and etiocholanolone (Etio) concentration of 698.8 ng/mL, with A/Etio ratios of 0.6 (Figure a and Table ). Following oral administration of 80 mg androstenedione, both metabolites increased sharply, reaching maximum concentrations at the first post-administration sampling time, 6 h after (A = 35,272.4 ng/mL; Etio = 51108.4 ng/mL). At this time point, the A/Etio ratio increased to 0.69, indicating a relative enrichment of 5α-reduced metabolites in the immediate post-dose window (Figure a and Table ).

1.

Time evolution of metabolic ratios following androstenedione and testosterone undecanoate administration. Panels (a and b) show the ratios A/Etio and 5α/5β, respectively, following androstenedione administration; panels (c and d) show the same ratios following testosterone undecanoate administration. Horizontal lines indicate pre-administration baseline values, while symbols represent post-administration measurements at the indicated times.

1. Urinary Concentrations (ng/mL) of Androsterone (A), Etiocholanolone (Etio), 5α-Androstanediol (5α), and 5β-Androstanediol (5β), and Corresponding Metabolite Ratios (A/Etio and 5α/5β) following Oral Administration of Androstenedione (top) and Testosterone Undecanoate (bottom) .

androstenedione administration
sampling time (h)	A (ng/mL)	Etio (ng/mL)	A/Etio	5α (ng/mL)	5β (ng/mL)	5α/5β
–18	418.0	698.8	0.60	67.1	509.2	0.13
6	35272.4	51108.4	0.69	402.9	1584.9	0.25
10	2307.7	6744.9	0.34	191.2	1081.8	0.18
22	2082.5	4890.1	0.43	105.7	640.3	0.17
28	1368.8	2630.0	0.52	83.1	649.6	0.13
38	277.0	523.2	0.53	118.3	726.2	0.16

testosterone undecanoate administration
sampling time (h)	A (ng/mL)	Etio (ng/mL)	A/Etio	5α (ng/mL)	5β(ng/mL)	5α/5β
–3	750.0	1340.7	0.56	46.7	85.1	0.55
9	1196.7	2053.7	0.58	117.1	310.2	0.38
13	1541.1	4216.7	0.37	111.6	273.0	0.41
24	2242.2	5962.0	0.38	271.2	721.3	0.38
30	941.5	2133.6	0.44	20.5	68.3	0.30
37	448.2	850.4	0.53	10.9	35.1	0.31

^aSampling times are given in hours (h) relative to administration.

After the peak at 6 h, androsterone and etiocholanolone concentrations declined rapidly. By 10 h, androsterone concentrations had decreased to 2307.7 ng/mL and etiocholanolone to 6744.9 ng/mL, with A/Etio ratios reaching a minimum of 0.34. Concentrations continued to decrease thereafter, approaching baseline levels by 38 h postdose (A = 277.0 ng/mL; Etio = 523.2 ng/mL), while A/Etio ratios rose toward the baseline value, reaching 0.53 (Figure a and Table ).

Measurements of androstanediols in this experiment show that 5β-androstanediol consistently exceeded 5α-androstanediol at all measured time points (Figure b and Table ). At baseline (18 h pre-administration), concentrations were 67.1 ng/mL (5α) and 509.2 ng/mL (5β), yielding a 5α/5β ratio of 0.13. Following administration, both diols increased markedly, with maxima at 6 h (5α = 402.9 ng/mL; 5β = 1584.9 ng/mL), at which time the 5α/5β ratio increased to 0.25 (Figure b and Table ; indicating a transient relative increase in 5 α-reduced products). Although absolute concentrations declined thereafter, the 5α/5β ratio remained elevated (0.13–0.18) throughout the time series, indicating sustained predominance of 5β-reduced downstream metabolism during clearance of the androstenedione-derived androgen load.

Testosterone Undecanoate Administration

Pre-administration baseline samples (3 h pre-administration) showed androsterone (A) concentration of 750.0 ng/mL and etiocholanolone (Etio) concentration of 1340.7 ng/mL, with A/Etio ratios of 0.56 (Figure c and Table ). Following administration of 40 mg TU, androsterone and etiocholanolone concentrations were elevated across much of the post-dose sampling period. At the first sampled time point (9 h), androsterone reached 1196.7 ng/mL and etiocholanolone 2053.7 ng/mL, with an A/Etio ratio of 0.58 (Figure c and Table ). Concentrations increased further over the following time points: at 13 and 24 h, androsterone was observed to be 1541.1 and 2242.2 ng/mL, respectively, and etiocholanolone 4216.7 and 5962.0 ng/mL, while the A/Etio ratio decreased to 0.37–0.38, indicating a relative enrichment of 5β-reduced metabolites in the immediate post-dose window. By 37 h concentrations had decreased further to 448.2 ng/mL for androsterone and 850.4 ng/mL for Etio, and the A/Etio ratio increased to 0.53, approaching baseline values (Figure c and Table ).

The TU administration time series also included comprehensive profiling of androstanediols and additional steroids. The baseline sample (3 h pre-administration) showed 5α/5β androstanediol ratios of 0.55, indicating modest predominance of 5β-reduction under endogenous conditions (Figure d and Table ). Following TU administration, both androstanediols increased substantially, 9 h post-dose (5α = 117.1 ng/mL; 5β = 310.2 ng/mL), while the 5α/5β ratio decreased to 0.38 (Figure d and Table ). This change implies enhanced formation of 5β-reduced metabolites relative to baseline. At later times (>9 h), androstanediol concentrations decreased while the 5α/5β ratio declined slightly to 0.30–0.31 by 30–37 h.

δ¹³C Composition of Androsterone Fragments following Androstenedione Administration

Isotopic analysis of androsterone fragment ions revealed distinct temporal trends in δ¹³C_{in‑house STD} values over a 38 h period following administration of 80 mg androstenedione (δ¹³C_VPDB = −31.8‰). These data are presented in Figure and Table for androsterone fragments of m/z 255, 199, 173, and 124 Da.

2.

Time evolution of fragment-specific carbon isotope compositions of urinary androsterone following oral androstenedione administration (80 mg; bulk δ¹³C_VPDB = −31.8‰). Panels (a–d) show δ¹³C_{in‑house STD} (‰; referenced to the in-house androsterone acetate standard) for the diagnostic fragments (a) m/z 255 (whole carbon skeleton), (b) m/z 199 (rings B–D), (c) m/z 173 (rings A–B + C11–C12), and (d) m/z 124 (ring D + C11–C12). Gray symbols represent post-dose mean values with black error bars (±RSE). The horizontal red line denotes the pre-administration (baseline) value for each fragment, and the pale red band (“halo”) indicates its analytical uncertainty (±RSE).

2. Fragment-Specific Isotopes of Androsterone Pre- and Post- Single Oral Dosing,

androstenedione administration
fragment	255 Da		199 Da		173 Da		124 Da
post-administration sampling time (h)	δ13C (‰)	RSE (‰)	δ13C (‰)	RSE (‰)	δ13C (‰)	RSE (‰)	δ13C (‰)	RSE (‰)	δD (‰)	RSE (‰)
–18	6.8	0.1	5.2	0.4	5.4	0.3	–0.4	0.2	–0.4	1.9
6	0.3	0.2	0.6	0.2	0.3	0.3	–1.7	1.1	–4.7	3.9
10	2.1	0.3	2.1	0.4	2.2	0.2	–0.4	0.7	12.6	4.9
22	5.4	0.2	4.5	0.6	4.1	0.2	1.2	0.6	–29.2	1.7
28	6.2	0.3	5.6	0.4	4.5	0.4	–3.7	0.4	–34.9	8.7
38	5.6	0.3	5.2	0.6	4.3	0.6	–4.1	0.4	–62.8	12.2

testosterone undecanoate administration
fragment	255 Da		199 Da		173 Da		124 Da
post-administration sampling time (h)	δ13C (‰)	RSE (‰)	δ13C (‰)	RSE (‰)	δ13C (‰)	RSE (‰)	δ13C (‰)	RSE (‰)	δD (‰)	RSE (‰)
–3	6.8	0.1	5.2	0.4	5.4	0.3	–0.4	0.2	–0.4	1.9
9	3.8	0.2	4.4	0.5	4.0	0.3	–1.2	0.3	9.0	1.8
13	5.4	0.1	4.8	0.3	4.1	0.9	0.4	0.2	n.d.	n.d.
24	7.7	0.2	7.7	0.8	7.4	0.2	–0.3	0.3	–1.6	3.1
30	7.4	0.4	6.8	0.2	6.2	0.3	–3.2	0.5	–65.9	8.7
37	7.1	0.1	7.2	0.6	n.d.	n.d.	–3.7	0.3	–6.5	1.0

^aReported are mean δ¹³C_{in‑house STD} (‰) for fragments m/z 255 (full C skeleton), 199 (rings B–D), 173 (rings A–B + C11–C12), and 124 (ring D + C11–C12), and δD_{in‑house‑STD} (‰) for m/z 124 only, at the indicated hours post-administration of androstenedione (80 mg; δ¹³C_VPDB = −31.8‰) or testosterone undecanoate (40 mg; δ¹³C_VPDB = −27.0‰). RSE = relative standard error (SD/SQRT­(n); n = 5–7 replicates). Values are referenced to the synthetic androsterone working standard.

^bn.d., not determined due to insufficient analyte abundance.

The δ¹³C_{in‑house STD} value of the m/z 255 Da fragment decreased from +6.8 ± 0.1‰, prior to administration, to +0.3 ± 0.2‰ at 6 h post-administration, then steadily increased, stabilizing at +5.6 ± 0.3‰ by 38 h (Figure a). The m/z 199 fragment showed a similar trend, decreasing from +5.2 ± 0.4‰ before administration to +0.6 ± 0.2‰ at 6 h, before returning to +5.2 ± 0.6‰ by 38 h (Figure b).

The m/z 173 fragment also exhibited an initial depletion followed by recovery, with δ¹³C_{in‑house STD} values decreasing from +5.4 ± 0.3‰ before administration to +0.3 ± 0.3‰ at 6 h and increasing to +4.3 ± 0.6‰ by 38 h (Figure c).

In contrast, m/z 124 showed more complex behavior. Its δ¹³C_{in‑house STD} value was −0.4 ± 0.2‰ before administration, decreased to −1.7 ± 1.1‰ at 6 h, increased to +1.2 ± 0.6‰ by 22 h, and then declined to −4.1 ± 0.4‰ at 38 h (Figure d).

δ¹³C Composition of Androsterone Fragments Following Testosterone Undecanoate Administration

Following administration of 40 mg TU (δ¹³C_VPDB = −27.0‰), isotopic analysis of androsterone fragments revealed temporal patterns over the subsequent 37 h period (Figure and Table ).

3.

Time evolution of fragment-specific carbon isotope compositions of urinary androsterone following testosterone administration as testosterone undecanoate (40 mg; bulk δ¹³C_VPDB = −27.0‰). Panels (a–d) show δ ¹³C_{in‑house STD} (‰; referenced to the in-house androsterone acetate standard) for the diagnostic fragments (a) m/z 255 (whole carbon skeleton), (b) m/z 199 (rings B–D), (c) m/z 173 (rings A–B + C11–C12), and (d) m/z 124 (ring D + C11–C12). Gray symbols represent post-dose mean values with black error bars (±RSE). The horizontal red line denotes the pre-administration (baseline) value for each fragment, and the pale red band (“halo”) indicates its analytical uncertainty (±RSE).

The δ¹³C_{in‑house STD} value of the 255 Da fragment decreased from a value of +6.8 ± 0.1‰ before administration to +3.8 ± 0.2‰ at 9 h, then increased steadily, reaching +7.7 ± 0.2‰ at 24 h, then stabilizing at 7.1 ± 0.1‰ by 37 h (Figure a and Table ). The 199 Da fragment showed a similar trajectory, though differing in an important detail: From a pre-administration δ¹³C_{in‑house STD} value of +5.2 ± 0.4‰ it then decreased only slightly to +4.4 ± 0.5‰, and then rose to +7.2 ± 0.6‰ by the end of the time seriesa value higher than the starting composition and opposite the direction expected (and observed for the 255 Da fragment ion) for simply transferring the low δ¹³C of the TU carbon skeleton to androsterone (Figure b and Table ).

Likewise, the δ¹³C_{in‑house STD} value of the 173 Da fragment declined from a pre-administration value of +5.4 ± 0.3‰ to +4.0 ± 0.3‰ at 9 h, after which it rose to +6.2 ± 0.3‰ by 30 h. Due to limited sample volume, this fragment was not measured at the final 37 h time point (Figure c and Table ).

The 124 Da fragment exhibited even more dramatically non-monotonic and counterintuitive behavior, starting with a pre-administration δ¹³C_{in‑house STD} value of −0.4 ± 0.2‰ which decreased only slightly to −1.2 ± 0.3‰ at 9 h post-administration. The value then rose to +0.4 ± 0.2‰ by 13 h, dipped slightly to −0.3 ± 0.3‰ at 24 h, and ultimately declined to −3.7± 0.3‰ at 37 h (Figure d and Table ).

Hydrogen Isotope Composition of the m/z 124 Da Androsterone Fragment

Hydrogen isotope analysis was performed only on the m/z 124 fragment due to limitations in analyte quantity and the mass resolution requirements for larger fragments. Longer acquisition times and enhanced sensitivity would be needed to accurately quantify δD in heavier fragments. ,

For the androstenedione administration experiment, δD_{in‑house STD} value of the 124 Da fragment was −0.4 ± 1.9‰ prior to administration (Figure a and Table ). At 6 h post-dose, values decreased negligibly to −4.7 ± 3.9‰. However, this was followed by an increase to +12.6 ± 4.9‰ at 10 h, and then, over the subsequent 22 h, δD_{in‑house STD} values progressively declined, reaching −62.8 ± 12.2‰ at 38 h (Figure a and Table ).

4.

Time evolution of hydrogen isotope compositions for the m/z 124 fragment of urinary androsterone following steroid dosing. Panel (a) shows δD_{in‑house STD} (‰; referenced to the m/z 124 fragment of our in-house androsterone acetate standard) for androstenedione administration, and panel (b) shows δD_{in‑house STD} (‰; referenced to the m/z 124 fragment of our in-house androsterone acetate standard) for testosterone administration as testosterone undecanoate. Gray symbols represent post-dose mean values with black error bars (±RSE). The horizontal red line denotes the pre-administration (baseline) value for the 124 fragments, and the pale red band (“halo”) indicates its analytical uncertainty (±RSE). Sampling times were 6, 10, 22, 28, and 38 h for panel (a) and 9, 13, 24, 30, and 37 h for panel (b) (13-h time point post testosterone undecanoate administration was not determined due to analyte limitations).

In the case of the TU administration experiment, δD_{in‑house STD} values of the same androsterone fragment also began at −0.4 ± 1.9‰ and increased to +9.0 ± 1.8‰ at 9 h (Figure b and Table ). A consistent depletion trend followed, with δD falling to −65.9 ± 8.7‰ by 30 h before recovering and stabilizing at −6.5 ± 1.0‰ by 37 h (Figure b and Table ). Due to analyte limitations, δD_{in‑house STD} measurements could not be obtained for the 13-h time point in this series.

Discussion

The results demonstrate that Orbitrap-based fragment-resolved isotope analysis can distinguish whole-skeleton and fragment-specific isotope behavior at natural abundance in a purified steroid from complex biological matrices. Across the two administration experiments, some fragment ions tracked the pulse–chase mixing and clearance behavior expected from conventional CSIA, whereas others showed fragment-specific deviations not evident in the whole-skeleton signal. These observations indicate that fragment-resolved measurements provide analytical information beyond a single molecular-average isotope value and can help distinguish mixing-dominated behavior from localized metabolic isotope effects. This capability opens a path to analogous applications in biomedical, environmental, and geological organic-matter studies.

Fragment-Specific Isotope Responses to Androgen Administration

Across the two administration experiments, distinct fragment-specific isotope responses were observed. The whole-skeleton fragment ( m/z 255) follows the expected pulse–chase pattern in both experiments, consistent with mixing of exogenous-derived androsterone followed by clearance and return toward baseline (Figures a and a). This response was expected to be stronger after androstenedione administration because the administered androstenedione was more ¹³C-depleted than endogenous urinary androgens, whereas the TU had a δ¹³C value closer to the endogenous range. After androstenedione administration, fragments m/z 199 and 173 follow the same pulse–chase trajectory as m/z 255 (Figures b,c and a,b), whereas after TU administration, fragments m/z 199 and 173 exhibit non-monotonic evolution, over-shooting of the pre-administration composition and failing to return to baseline values by the end of the sampling period (Figures b,c and a,b). In both administration experiments, the D-ring fragment ( m/z 124) exhibits even stronger non-monotonic behavior. These anomalous findings indicate that in addition to “pulse–chase” source mixing, pathway-dependent isotopic fractionation contributes to fragment-specific isotope values, particularly in regions undergoing metabolic transformation (Figures d, d, c and c). Interpretation of these fragment-specific isotope signatures demands a deeper consideration of the fates of administered androgens and their impacts on the broader steroidal metabolic network.

5.

Cross-plots of fragment-specific carbon isotope compositions of urinary androsterone following androstenedione administration (80 mg), expressed as δ¹³C_{in‑house STD} (‰) relative to the in-house androsterone acetate standard. Each panel compares a diagnostic fragment to the whole-skeleton ion (m/z 255): (a) m/z 199 (rings B–D), (b) m/z 173 (rings A–B + C11–C12), and (c) m/z 124 (ring D + C11–C12). The red symbol marks the pre-administration (baseline) composition, with its associated analytical uncertainty. Gray symbols represent post-dose samples, shown as mean values with ±RSE in both x and y and annotated by sampling time (h).

6.

Cross-plots of fragment-specific carbon isotope compositions of urinary androsterone following testosterone undecanoate administration (40 mg), expressed as δ¹³C_{in‑house STD} (‰) relative to the in-house androsterone acetate standard. Each panel compares a diagnostic fragment to the whole-skeleton ion (m/z 255): (a) m/z 199 (rings B–D), (b) m/z 173 (rings A–B + C11–C12), and (c) m/z 124 (ring D + C11–C12). The red symbol marks the pre-administration (baseline) composition, with error bars indicating analytical uncertainty. Gray symbols represent post-dose samples, shown as mean values with ±RSE in both x and y and annotated by sampling time (h).

Metabolic-Network Context and Candidate Fractionating Steps

Orally administered unmodified androgens such as androstenedione, and modified androgens such as TU are expected to perturb androgen metabolism differently. Androstenedione undergoes substantial first-pass hepatic metabolismincluding reduction, oxidation, and conjugation reactions that shape the urinary steroid poolbefore systemic release. , In contrast, TU can be absorbed partly through intestinal lymphatics and enter systemic circulation before hepatic processing, producing a different balance between peripheral and hepatic metabolism. −

Androstenedione can be converted to androsterone through two routes: (1) C-17 ketone reduction to a 17β-hydroxyl (testosterone) via 17β-hydroxysteroid dehydrogenase (17β-HSD) type III (HSD17B3) (blue route in Scheme ); or (2) ring A reduction via 5α-reductase type I (SRD5A1), followed by C-3 ketone reduction via 3α-hydroxysteroid dehydrogenase (3α-HSD) (brown route in Scheme ).

2. Simplified Network of Androgen Metabolism Leading to Major Urinary 5α-Androsterone (“andro”) and 5β-Androsterone (“etio-”) Metabolites, Modified following .

^a Arrow color indicates the relative dominance of pathway flux under our dosing conditions: following androstenedione administration, metabolism is routed primarily through the 5α (A-ring reduction) branch (brown arrows), with comparatively reduced 5β flux (light brown arrows); following testosterone undecanoate administration, routing shifts such that the testosterone-driven branch dominates (blue arrows) and the 5β (AKR1D1) branch becomes more prominent (light blue arrows). Key steps are catalyzed by 5α-reductases (SRD5A1 = 5α-reductase type I; SRD5A2 = 5α-reductase type II), 5β-reductase (AKR1D1), 17β-hydroxysteroid dehydrogenases (HSD17B3 and HSD17B2; 17β-HSD types III and II, respectively), and 3α-hydroxysteroid dehydrogenases (3α-HSD; AKR1C isoforms, e.g., AKR1C1–AKR1C4).

During the early post-dose period, rapid hepatic delivery of androstenedione is expected to favor the SRD5A1-mediated 5α route, because SRD5A1 is relatively abundant in the liver, , whereas HSD17B3 is mainly testicular with minimal hepatic expression. − This is further supported by androstenedione being a preferred SRD5A1 substrate, − whose activity increases at higher hepatic substrate concentrations. ,

In parallel, hepatic clearance attenuates the excess androgen pool through phase-II conjugation, including glucuronidation, , and 5β-reduction by AKR1D1, which produces etio-series metabolites. However, because AKR1D1 can show substrate inhibition at elevated steroid concentrations, early high hepatic androstenedione levels may transiently constrain 5β flux. Consequently, SRD5A1-mediated 5α “andro-series” routing becomes relatively more prominent while conjugation promotes clearance and return toward baseline. This shift is reflected in the urinary ratios, with A/Etio and 5α/5β androstanediol peaking early, consistent with a transient increase in 5α routing during the initial post-dose window (Table and Figure a,b).

Following esterase-mediated hydrolysis of TU to free testosterone, local enzyme expression in peripheral tissues may bias initial metabolic routing. TU-derived testosterone can contribute to androsterone formation either through: (1) 5α-reduction of testosterone to dihydrotestosterone (DHT), mediated predominantly by 5α-reductase type II (SRD5A2), followed by 3α-reduction of the 3-keto group by 3α-HSD and subsequent oxidation of the hydroxyl group at C-17 by 17β-HSD (blue route in Scheme ), and/or (2) oxidation of testosterone to androstenedione by 17β-HSD type II (HSD17B2), followed by A-ring reduction (e.g., 5α-reductase for androsterone formation) and 3α-HSD mediated conversion (brown route in Scheme ).

During the early post-dose period, partial lymphatic absorption of TU is expected to increase the relative contribution of peripheral metabolism before complete hepatic processing. This favors the SRD5A2-mediated (5α-reductase type II) route, because SRD5A2 has high affinity for steroid substrates and is prominent in androgen-sensitive tissues, , supporting efficient local conversion of TU-derived testosterone to DHT (route 1; blue route in Scheme ). Consistent with this interpretation, we observed elevated levels of downstream metabolites, particularly androstanediol, which is produced from DHT through 3α-HSD activity (Table ).

In parallel, testosterone can be oxidized to androstenedione through HSD17B2, which is strongly expressed in several epithelia. , This route is therefore consistent with local testosterone inactivation and subsequent downstream processing after recirculation. As steroids return to the liver, high-capacity hepatic enzymes, including AKR1C 3α-HSD activities and AKR1D1, continue to support 5β-reduced etio-series metabolite formation and clearance. , Thus, unlike androstenedione administration, where an early hepatic androstenedione load may transiently constrain AKR1D1-mediated 5β flux via substrate inhibition, TU dosing supports parallel α- and β-routing: enhanced early peripheral 5α metabolism, together with maintained hepatic 5β clearance. This is reflected by increased androstanediol alongside declining α/β androstanediol and A/Etio ratios (Table and Figure c,d).

Because androstenedione and TU metabolism modify different steroid regionsA-ring 5α/5β reduction with C-3 interconversion versus C-17 oxidation/reductionthese pathways can produce fragment-specific δ¹³C values that deviate from simple two-source mixing and pulse–chase behavior. This network view yields two expectations: (i) if isotope composition mainly reflects source mixing and clearance, fragment-ion δ¹³C values should follow the whole-skeleton pulse–chase pattern; and (ii) if pathway routing or reaction-specific isotope effects contribute substantially, fragments representing different regions, such as the A-ring versus C-17/D-ring, should diverge. Such divergence may appear as non-monotonic post-dose evolution or overshoot beyond the range defined by pre-administration values and the earliest post-dose value, when the administered steroid is most abundant. Thus, time-varying 5α/5β routing and isotope effects during reduction/oxidation steps provide plausible mechanisms for departures from simple pulse–chase behavior and frame the following interpretations of the androstenedione and TU experiments:

Interpretation of the Results of Androstenedione Administration in Light of the Metabolic Context

Following androstenedione administration, the whole-skeleton ion ( m/z 255), and the two larger carbon-skeleton fragments (m/z 199, and 173) exhibit the expected pulse–chase behavior, tracking the rise and subsequent clearance of exogenous-derived androsterone (Figures a–c and a,b). This response is consistent with previous GC-IRMS studies of exogenous androgen administration. − The close agreement among these three fragments (Figure a,b) indicates that source mixing dominates their isotope evolution and that no resolvable fragment-specific δ¹³C offset arises from reaction-specific isotope effects.

Early metabolism in the androstenedione experiment is expected to be dominated by hepatic 5α-reduction of Δ⁴-3-keto steroids, reflecting first-pass delivery to the liver and the combination of substantial SRD5A1 activity with minimal hepatic contribution from the HSD17B3-mediated androstenedione to testosterone route. ,,, In this early window, metabolism is dominated by A-ring chemistry. These steps include reduction of the C-3 carbonyl (sp²) to a 3α-hydroxyl (sp³) and hydrogenation of the Δ⁴ double bond, transformations that change local bonding/hybridization and could therefore impart small secondary ¹³C isotope effects. However, the A-ring–containing fragment ( m/z 173; Scheme ) does not diverge detectably from the whole-skeleton signal ( m/z 255) (Figure b), implying that any site-localized effects at C-3/C-4/C-5 are either too small to resolve after intramolecular averaging and dilution or are effectively muted under high-capacity hepatic processing. , Consistent with dominant hepatic 5α routing and a comparatively limited contribution from 5β reduction during the early first-pass window, no relative enrichment of the measured androsterone A-ring-containing fragments is observed. Partial cancellation across sequential steps (e.g., 5α-reduction followed by 3α-HSD) could further reduce net expression at the fragment scale. Definitive tests would require position-specific ¹³C KIE measurements at C-3, C-4, and C-5 to determine whether (and where) A-ring reduction introduces measurable secondary effectsat present, such highly specific isotopic analysis would appear to require some method or technology beyond what we developed here.

The evolution of the δ¹³C value of the m/z 124 fragment departs significantly from this pulse-chase behavior; i.e., it does not exhibit a smooth, monotonic return toward baseline (Figures d and c). This deviation indicates an additional control on the isotope composition of the D-ring/C-17 region (Scheme ). Because m/z 124 isolates the D-ring/C-17 region and averages over fewer carbons than the larger fragments, it is intrinsically more sensitive to small, site-localized fractionations that would be masked in whole-skeleton ions by intramolecular averaging. ,, The simplest interpretation is therefore that m/z 124 reflects mixing plus an additional, time-dependent control specific to transformations in the C-17 neighborhood.

A plausible source of that additional control is increased routing through pathways that involve C-17 redox interconversion. Early androstenedione metabolism is dominated by hepatic 5α-reduction of Δ⁴-3-keto steroids. ,, If later metabolism includes a larger contribution from 17β-HSD interconversions, then C-17 redox chemistry is able to manifest an isotopic signal because it toggles C-17 between the sp² carbonyl of androstenedione and the sp³ alcohol of testosterone, and later between the sp³ C-17 alcohol in testosterone and the sp² C-17 carbonyl in androsterone (Scheme ). Such rehybridization is expected to generate secondary kinetic isotope effects on atoms adjacent to the reaction center through changes in vibrational force constants and zero-point energies. , These effects may be too diluted to alter m/z 255/199/173 detectably, but they can remain visible in m/z 124 precisely because that fragment concentrates signal from the D-ring/C-17 region rather than averaging across the full skeleton (see structure in Scheme ).

A simple mass-balance calculation illustrates how a carbon isotope excursion localized to the D-ring/C-17 region would be expressed in the whole-skeleton isotope time series as a smaller amplitude departure from a monotonic pulse-chase behavior. If the additional effect is confined to the subset of carbons represented by the m/z 124 fragment (8 of 19 carbons in the androsterone skeleton), then the maximum shift imparted to the whole-skeleton composition can be estimated using eq (3), where Δδ¹³C₂₅₅ ≈ (8/19) × Δδ¹³C₁₂₄ ≈ 0.42 × Δδ¹³C₁₂₄. For example, the late-time deviation from baseline observed in m/z 124 at 38 h following androstenedione administration is 3.7 ± 0.5‰ (see note S1 in the Supporting Information for more details). This would correspond to a maximum shift of 1.56 ± 0.21‰ (∼1.6 ± 0.2‰) in m/z 255. This is broadly consistent with the late-time residual of m/z 255 relative to the exponential recovery model (∼0.8–0.9 ± 0.3‰; see note S1 in the Supporting Information for more details), supporting a first-order upper-bound estimate of the whole-skeleton expression of the localized m/z 124 anomaly.

This comparison supports the interpretation that the anomalous behavior is localized to the D-ring/C-17 domain and substantially diluted in whole-skeleton averages. Thus, even several per mil of fractionation in the D-ring/C-17 domain would be expected to contribute only ∼1‰ scale changes to the whole-skeleton δ¹³C, explaining why the m/z 255 ion deviates only slightly from strict pulse–chase behavior. Accordingly, the larger fragments (m/z 255, 199, 173) in the androstenedione administration experiment primarily reflect whole-skeleton mixing and clearance, with only second-order deviations from simple exponential returns to baseline. In contrast, m/z 124 isolates the D-ring/C-17 domain and represents fewer carbons, allowing it to more strongly express additional region-specific isotope effects that are muted by intramolecular averaging in whole-skeleton signals (see note S1 in the Supporting Information for more details).

Hydrogen isotopes provide an independent consistency check on a localized D-ring/C-17 control (Figure a and Table ). Previous studies have shown that whole-molecule hydrogen isotope ratios of urinary steroids can track exogenous testosterone administration in a way that complements bulk carbon isotope ratios, consistent with a pulse–chase-style source-and-clearance signal. , In the present study, however, δD was measurable only for m/z 124, and its time evolution indicates that the D-ring/C-17 region is influenced by processes beyond simple carbon-source mixing that would produce a pulse–chase-style signal. Because androsterone carries a C-17 carbonyl, δD in m/z 124 is expected to be especially sensitive to hydrogens adjacent to C-17, where 17β-HSD redox steps interconvert the local environment between sp³ (alcohol) and sp² (carbonyl) bonding and can express α-secondary H/D effects. , Consistent with this mechanism, δD of m/z 124 shows little change during the early post-dose interval, when conversion of androstenedione to androsterone is expected to proceed largely without introducing a new C-17 rehybridization event, but exhibits a later depletion that is consistent with increased routing through the androstenedione to testosterone and eventually to the androsterone branch, which does include C-17 redox interconversion. Even if intrinsic α-secondary isotope effects are substantial at the step level, the net δD shift in m/z 124 is expected to be only a few per mil after dilution, branching, fragment averaging, and mixing. ,, Taken together, the δ¹³C and δD behavior of m/z 124 supports the interpretation that SSIA captures pathway structure that is largely invisible to whole-skeleton CSIA signals.

Interpretation of the Results of Testosterone Undecanoate Administration in Light of the Metabolic Context

The TU administration experiment differs from the androstenedione administration case in a way that highlights the value of fragment-specific isotope measurements. Although whole-skeleton carbon isotope signals for androsterone are broadly similar (and would likely be indistinguishable if the administered agents had more similar δ¹³C), the smaller fragment ions evolve differently in the two experiments in ways that can be linked to the distinct effects of TU and androstenedione on overall androgen metabolism. The m/z 255 ion shows the expected pulse–chase-style response, reflecting incorporation of exogenous-derived material and recovery toward baseline. , However, in contrast to androstenedione, the A-ring-containing fragments ( m/z 199 and 173) diverge from that whole-skeleton behavior and evolve over time to exhibit enrichment relative to baseline (i.e., the phenomenon we refer to above as “overshoot”; Figures a–c and a–b). This cannot be explained by two-endmember mixing alone and instead points to a changing balance among the metabolic pathways controlling androsterone sources and sinks that differentially affect the isotope composition of structural regions sampled by these fragments.

TU’s lymphatic absorption reduces hepatic first-pass dominance and increases early systemic exposure before recirculation. − Consequently, early androsterone formation can draw on competing peripheral and hepatic routes (Scheme ), so TU perturbs time-dependent 5α/5β partitioning rather than simply adding a low-δ¹³C pool to a static network.

After TU dosing, the urinary A/Etio and 5α/5β-androstanediol ratios decline from baseline and then partially recover (Figure c,d), consistent with a transient increase in relative 5β routing during TU metabolism before partial re-equilibration. This behavior provides an independent, time-resolved indicator that the α/β balance is not constant during the same interval in which carbon isotope contents of the m/z 199 and 173 fragment ions decouple from that of the m/z 255 ion (Figure a,b). This shift in metabolic routing offers a mechanistic explanation for the fragment-specific isotope pattern.

Previous population studies and anti-doping applications have revealed systematic differences in isotopic composition among urinary steroid metabolites associated with different branches of androgen metabolism (e.g., andro- vs etio-series differences), emphasizing that metabolic heterogeneity can complicate interpretation of molecule-average CSIA. ,, Notably, etio-series (5β) products are often observed to be ¹³C-depleted relative to andro-series (5α) products, , so increasing 5β routing introduces a relatively low-¹³C component into compounds that have been subject to reactions affecting the A/B-rings, which can leave the concurrently expressed 5α-associated signal relatively ¹³C-enriched. In the TU administration experiment conducted here, a time-varying contribution of 5β-routed intermediates plausibly contributed to the isotopic evolution of the A/B-ring region where C-4/C-5 stereochemistry and C-3 interconversion occur. Because m/z 199 and 173 fragment ions disproportionately sample this A/B-ring chemistry, they can express enrichment (or “overshoot”) relative to the whole-skeleton average when pathway partitioning, as reflected in the urinary A/Etio and 5α/5β-androstanediol ratios (Figure c,d), and associated isotopic fractionations evolve through time (Figures b,c and a,b).

The time-evolution in δ¹³C of the m/z 124 fragment ion of androsterone in the TU administration experiment (Figure b) suggests additional localized controls near the D-ring/C-17 site (as was discussed above for the androstenedione administration experiment). Unlike the larger fragment ions, m/z 124 isolates a region influenced by 17β-HSD oxidation/reduction steps and associated hybridization changes at C-17, which can generate secondary isotope effects whose visibility is maximized when the fragment averages over fewer carbons. , In this experiment, peripheral processing prior to hepatic equilibration makes it plausible that the timing of C-17 interconversion differs from the timing of A-ring reductions and from the “pulse–chase” signal that dominates the variations in δ¹³C of the m/z 255 fragment ion. Thus, the simplest network-consistent explanation for the TU administration data is that (i) whole-skeleton δ¹³C largely tracks net mixing/clearance, while (ii) A-ring fragments and the D-ring fragment record time-dependent pathway partitioning plus small reaction-specific fractionations that are muted in the whole-skeleton average.

Finally, the TU administration experiment provides a clear demonstration of why fragment-resolved observables may be useful in anti-doping applications: TU’s molecular-average δ¹³C can overlap endogenous ranges, making conventional CSIA less decisive in some cases. ,, In such cases, a multi-fragment isotope fingerprint offers additional constraintssome tracking net mixing and others tracking pathway perturbationthat collectively increase discriminatory power even when whole-molecule carbon isotope contents alone are ambiguous.

Integrating Mixing–Clearance Dynamics and Pathway Routing in Fragment-Resolved SSIA

Across both administration experiments presented here, the isotope systematics indicate that two processes act together to control the profile of fragment-ion-specific carbon isotope changes. First, mixing between endogenous and exogenous-derived steroid pools, coupled to rapid conjugation and clearance, produces a pulse–chase signature that is expressed clearly in the whole-skeleton ion and, in the androstenedione experiment, also in the major carbon-skeleton fragmentsconsistent with decades of application of GC–IRMS CSIA measurements to anti-doping. ,,, Second, time-varying routing through competing metabolic branches and reaction-specific isotope fractionations generate departures from pulse–chase behavior that are most apparent in fragments sampling portions of target molecules that are subject to chemical change in the steroid metabolic network (i.e., the A/B-ring fragments in TU administration; D-ring fragment in both experiments). In practical terms, this means fragment-resolved SSIA can both (i) replicate the mixing/clearance information captured by conventional CSIA and (ii) add a new capability to recognize and interpret perturbations in the steroid reaction network that would be invisible in a single molecular-average value. ,,

Future Applications to Other Compounds and Problems

Generalizability of Methods and Extensions of the Isotopic Observables

The analytical strategy demonstrated here involves isolation and purification of biomolecules from complex matrices, dual-inlet introduction of analyte and standard to an Orbitrap mass spectrometer, and high-resolution isotope measurement of multiple diagnostic fragments using specialized techniques designed to maximize precision and structural specificity of isotopic data. These capabilities were previously demonstrated for small, simple biomolecules (e.g., amino acids), but this study is the first to show their applicability to compounds with the size and structural complexity of steroids. This workflow is transferable to other steroids not examined here and, more generally, to other moderately large biomolecules, so long as the analyte produces reproducible, structurally diagnostic fragment ions and the Orbitrap can resolve those ions from matrix-derived interferences at the required mass resolving power. The practical work of “porting” the method to a new target compound primarily involves (i) selecting a set of fragment ions that together provide complementary structural coverage, (ii) verifying spectral purity and assessing isotopic linearity across the relevant intensity range, and (iii) establishing an appropriate in-house reference material and bracketing strategy to correct instrumental drift and minimize nonlinearity.

Several extensions of the methods presented here would materially increase information content in the study of steroidal compounds. First, expanding the fragment set to target additional regions of the steroid skeleton would help localize isotopic fractionations associated with metabolic steps more precisely and reduce reliance on any single fragment ion and its associated fragmentation pathway. A broader fragment set would also help distinguish true absence of fractionation from effects masked by fragment averaging or cancellation of positive and negative fractionations across sequential metabolic steps. Second, broader coverage of the hydrogen-isotope structures of steroids (additional fragments, and/or whole-skeleton δD) would strengthen mechanistic constraints on steroid sources and sinks, because hydrogen isotopes often exhibit larger and more pathway-sensitive fractionations than carbon during enzymatic reduction/oxidation and H-exchange processes. − Third, the ability to quantify multiply substituted (“clumped”) isotopologues in these same fragments represents a major prospective advance. Although clumped-isotope observables require careful calibration and validation before interpretation, they provide constraints on intramolecular isotope pairing and formation mechanisms that are complementary to single-substitution δ¹³C and δD values. ,, Together, these extensions could transform the approach from “multi-fragment fingerprinting” into a broadly general platform for intramolecular isotope structure in biomolecules. Csernica et al. present an example of extension of Orbitrap methods for isotopic analysis to a large number of independent constraints on a single molecule.

Biomedical and Forensic Applications

From an anti-doping perspective, the immediate advantage of fragment-resolved SSIA is that it yields multiple, partly independent isotope readouts rather than a single whole-molecule average. This is particularly important when the molecular-average δ¹³C falls within the endogenous range, or when formulations are selected to minimize isotopic contrast, reducing the discriminating power of conventional CSIA workflows. ,,, In such cases, a “vector” of fragment-specific isotope values (i.e., in a Cartesian coordinate system with axes comprised of the independent fragment-specific measurements) can supply additional discriminatory power by encoding both source mixing and pathway perturbation. More broadly, fragment-resolved isotope patterns could complement steroid-profile approaches by adding mechanistic specificity, linking atypical profiles not only to an exogenous source but also to altered androgen metabolism and associated shifts in metabolite patterns. ,−

The same analytical principle may have broader applications in pharmaceutical authentication and forensic chemistry. Stable-isotope measurements have been used to distinguish pharmaceutical manufacturers and production batches, investigate falsified medicines, and support process patent protection using multidimensional molecular-average isotope fingerprints. − Fragment-resolved isotope analysis extends this concept by generating a higher-dimensional vector within a single molecule. The present study demonstrates that localized isotope effects associated with chemical transformations can produce fragment-specific isotope signatures that are not resolved by whole-molecule measurements. Similar effects may arise during pharmaceutical manufacture through precursor selection, synthetic transformations, bond formation and cleavage, and downstream processing. Fragment-specific measurements could therefore distinguish manufacturing routes that produce similar molecular-average isotope compositions. Such intramolecular isotope fingerprints could improve pharmaceutical provenance assessment, counterfeit detection, and process patent protection, although validation using products of known manufacturing origin and alternative synthetic routes will be required.

Beyond forensics, the ability to measure fragment-/position-resolved isotope patterns at natural abundance in low-concentration metabolites could be used to address metabolomic questions in which pathway routing and dysregulation are central. Steroid hormone pathways are good targets because the body makes and breaks down steroids through a series of enzymes that work together. These pathways are often disrupted in endocrine disorders, by drug–drug interactions that affect steroid-metabolizing enzymes, and in diseases that alter enzyme expression or activity. In principle, fragment-resolved isotope patterns could provide mechanistically interpretable signatures of such metabolic shifts. Such applications would be analogous to how position-specific isotope analysis has been used to probe changes in metabolic routing and flux in other biochemical systems involving small biomolecules. ,

Geological and Environmental Biomarker Studies

This study demonstrates that intramolecular isotope measurements can tell apart two effects that often look the same in bulk isotope data; mixing of different sources and isotope changes caused by chemical or biological processing. This “source vs process” distinction is a long-standing challenge in organic geochemistry.

Sterols and their diagenetic products (steranes) are widely used biomarkers for reconstructing organic matter sources, depositional environments, and aspects of biological evolution. However, similarities among source organisms and overprinting during early diagenesis can complicate sterol/sterane source attribution based on distributions of molecular proportions alone. ,− Compound-specific isotope analysis has added an important dimension of specificity to biomarker studies. However, their isotope values can still overlap among sources and may be modestly altered in some settings by diagenetic or microbial processing and environmental conditions, limiting interpretive uniqueness. ,−

Fragment- or position-specific isotope measurements could improve this situation in two ways. First, they could help deconvolve mixtures of biologically distinct but structurally similar sterols by adding intramolecular isotope structure as an additional “fingerprint,” analogous to how Orbitrap-based SSIA has distinguished overlapping sources in other classes of small molecules. − Second, they could provide a mechanistic constraint on early diagenetic transformations (e.g., reductions, rearrangements, aromatization) by localizing isotope effects to the sites of chemical change rather than averaging them across an entire skeleton where they may be too small to resolve. , In this sense, the steroid dosing experiments serve as a controlled model for future applications to geological equivalents with less well-known histories.

Conclusions

This study shows that Orbitrap-based fragment-resolved stable isotope analysis can quantify fragment-specific δ¹³C values and, for one fragment, δD values in a purified urinary steroid recovered from low-concentration urine samples. Across androstenedione and TU administration experiments, the whole-skeleton carbon isotope signal and several fragment ions exhibited pulse–chase behavior consistent with mixing of exogenous-derived material into an endogenous metabolite pool followed by clearance, paralleling the logic of conventional steroid CSIA.

Other fragment ions, however, deviate from this mixing-dominated behavior. These fragment-specific differences were most evident in the D-ring fragment in both experiments and in A-ring-containing fragments following the TU administration. The observed divergences are consistent with additional influences from pathway-dependent metabolic transformations and localized isotope effects that are not resolved in whole-skeleton measurements.

Taken together, these results establish fragment-resolved isotope analysis as a practical extension of conventional steroid CSIA and show that it can provide analytical information beyond a single molecular-average isotope value. Fragment-resolved isotope fingerprints may therefore improve steroid source discrimination when bulk δ¹³C values alone are inconclusive. More broadly, this framework motivates future work expanding fragment coverage, hydrogen-isotope constraints, and calibrated multiply substituted isotopologue measurements for applications in biomedical metabolomics and in geological and environmental biomarker studies.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.analchem.6c03178.

• Additional experimental methods for urine sample preparation and extraction, GC-MS, and GC-C-IRMS analysis; first-order analysis of the contribution of the m/z 124 isotope anomaly to the whole-skeleton δ¹³C signal, and exponential recovery modeling including one Supporting Information figure (PDF)

The authors declare no competing financial interest.