Enhanced UHPLC-MS/MS screening of selective androgen receptor modulators following urine hydrolysis

Graphical abstract Image, graphical abstract


a b s t r a c t
Selective androgen receptor modulators (SARMs) represent non-steroidal agents commonly abused in human and animal (i.e. equine, canine) sports, with potential for further misuse as growth promoting agents in livestockbased farming. As a direct response to the real and possible implications of illicit application in both sport as well as food production systems, this study incorporated enzymatic hydrolysis ( β-glucuronidase/arylsulfatase) into a previously established protocol while maintaining the minimal volume (200 μL) of urine sample required to detect SARMs encompassing various pharmacophores in urine from a range of species (i.e. equine, bovine, human, canine and rodent). The newly presented semi-quantitative UHPLC-MS/MS-based assay is shown to be fit-for-purpose, being rapid and offering high-throughput, with validation findings fulfilling criteria stipulated within relevant doping and food control legislation.
• CC β values determined at 1 ng mL −1 for majority of analytes.
• Deconjugation step included in the method led to significantly increased relative abundance of ostarine in analysed incurred urine samples demonstrating the requirement for hydrolysis to detect a total form of emerging SARMs. • Assay amenable for use within routine testing to ensure fair play in animal and human sports and that animalderived food is free from contamination with SARM residues.

Background
Selective androgen receptor modulators (SARMs) encompass a class of drugs with diverse nonsteroidal pharmacophores reported to be widely abused in human and animal sports through their oral bioavailability and biological potency which is facilitated by their widespread availability [1] . SARM compounds have potential to find use in livestock-based food production [2] systems seeking growth promoting and feed efficiency benefits. Hormonal acting substances are banned within farming in the EU since 1988 [3] , and assays with capability to detect potential SARMs' abuse are therefore needed to aid the effective enforcement of prohibition [4] . In this regard a range of methods in respect to the LC-MS/MS analysis of SARM residues in urine have been reported (recently reviewed by Ventura et al. [5] ). Additionally investigations into the metabolic fate of selected SARM compounds in various species (e.g. equine, bovine, human) have revealed that intact molecules and/or their respective generated phase I SARM metabolites undergo phase II conjugation (i.e. with glucuronic acid and/or sulphate moieties) [1 , 6-8] . However, variability in the range of different SARM pharmacophores and also in the pattern of interspecies metabolic biotransformation, is compounded by the lack of firm data in the scientific literature arising from drug elimination studies as well as an absence of reference materials and standards for associated biotransformation products. Consequently, implementation into routine urine analysis of procedures employing an enzymatic deconjugation step (cleavage of both glucuronide and sulphate conjugates) using e.g. Helix pomatia digestive juice [7 , 9 , 10] is recommended providing for superior detection windows via the indirect detection of the corresponding aglycones of SARMs and/or their metabolites. Our group reported previously [5] a semi-quantitative method to monitor the misuse of 15 SARM compounds belonging to nine different families, in urine matrices from a range of species (equine, canine, human, bovine and rodent). Briefly, SARM residues were extracted from urine (200 μL) with TBME without further clean-up and analysed by UHPLC-MS/MS. A 12 min gradient separation was carried out on a Luna Omega Polar C18 column, employing water and methanol, both containing 0.1% acetic acid ( v/v ), as mobile phases.

Validation was performed according to the EU Commission Decision 2002/657/EC criteria and European Union
Reference Laboratories for Residues (EU-RLs) guidelines with CC β values determined at 1 ng mL −1 , excluding andarine (2 ng mL −1 ) and BMS-564929 (5 ng mL −1 ), in all species. The current study therefore seeks to incorporate enzymatic hydrolysis into a previously reported screening protocol [5] to deliver a reliable and effective tool to reveal illicit SARM use in urine from animal and human sport animals as well as food-based livestock that can be adopted and implemented in various residue monitoring programmes.

Analysis of SARM residues in urine by UHPLC-MS/MS
Urine samples were stored at -80 °C and centrifuged at 4,500 × g for 10 min at 4 °C prior to analysis. Urinary specific gravity was assessed and pH adjusted as required with acetic acid to 5.5 Table 1 Analytical platform and respective conditions.  [5] . Nevertheless, the current chromatographic separation was employed following some modifications [11 , 12] , with conditions specific to the presented method summarised in Tables 1 and 2 . A typical chromatogram is shown in Fig. 1 with all target SARM compounds separated during the first 9.45 min of chromatographic analysis.

Extracted urine screen positive and recovery control checks
Pooled negative urine ( n = 10) was used for quality control (QC) purposes as described previously [5] . Briefly, extracted matrix screen positive controls were prepared by fortifying negative QC samples   d Internal standard. The response factor was obtained as a ratio between analyte peak area and internal standard peak area, in the case of the other SARMs, peak area was used as the response. e Diagnostic ion. ( n = 3) prior to extraction with 20 μL of quality control standard solution (10 / 20 (andarine) / 50 (BMS-564929) ng mL −1 ) to give a screening target concentration in urine for all analytes of 1 ng mL −1 , with the exception of andarine (2 ng mL −1 ) and BMS-564929 (5 ng mL −1 ). Additional negative samples ( n = 2) were spiked post-extraction with QC standard solution (20 μL) to monitor for analyte loss during extraction. Results from on-going QC samples (i.e. negative, screen positive and recovery controls) are being recorded to verify performance reliability and robustness of the assay.

Method optimization
This study aimed to incorporate an enzymatic urine hydrolysis step into a previously established UHPLC-MS/MS protocol [5 , 11 , 12] whilst maintaining the minimal volume (200 μL) of sample required. Enzymatic methods are commonly used as a hydrolysis approach being generally more specific with procedures performed in milder conditions in comparison to chemical (acid or alkaline) hydrolysis, ensuring the stability of target analytes and/or sample integrity. However, a number of factors impact efficiency of residue deconjugation namely temperature, time of incubation, pH and amount of enzyme [13] . Due to the lack of respective SARM conjugate standards, β-glucuronidase/arylsulfatase from Helix pomatia was used as per manufacturer's instructions. The sole parameter assessed during method development was the applicability of 0.1 mol L −1 carbonate buffer (pH 9.5) or 50 mmol L −1 aqueous NH 4 OH (pH 10.5) to elevate pH from pH 5.5 used during the enzymatic hydrolysis process, with the later chosen providing satisfactory recovery for all SARM compounds of interest (Supplementary data -Fig. S1).

Method validation
The current assay was validated with regard to selectivity, specificity, detection capability (CC β), sensitivity, limit of detection (LOD), absolute recovery and matrix effects, according to respective EU legislation [14 , 15] to demonstrate compliance with required performance criteria. Validation was carried out at the screening target concentration (C val ) of 1 ng mL −1 excluding andarine (2 ng mL −1 ) and BMS-564929 (5 ng mL −1 ) as detailed in Ventura et al. [5] .

Selectivity, specificity, and matrix effect studies
Method specificity has been reported previously highlighting the absence of cross talk between analytes and/or internal standards [11] , whereas selectivity in this modified study was established through analysis of 161 urine samples (collected and previously tested as reported by Ventura et al. [5] ) in the absence of matrix interferences. Injection of blank solvent (MeOH) following the screen positive control during every analysis was performed to monitor for carry-over, with no analyte signal in blank solvent observed. Matrix effects assessed through analysis of blank urine samples ( n = 5 per species) of different origins spiked post-extraction at 2 × C val , and calculated for each analyte as the percentage difference between signals obtained when matrix extracts or a standard solution of equivalent concentration were injected, divided by the signal of the latter [16] , ( Fig.  2 and Supplementary data -Table S1) highlighted signal suppression for the majority of analytes, with BMS-564929 and RAD140 reporting the greatest suppression (exceeding 75% for all target species). Incorporation of affordable isotope-labelled internal standards as they become available into this method is therefore recommended with the aim of compensating for matrix effects (signal suppression/enhancement) and further improvement of accuracy and precision.

Detection capability (CC β)
CC β [14] was determined by assessing threshold value (T) and cut-off factor (Fm) [15] through analysis of equine urine ( n = 26) from different sources, both blank and fortified at C val . CC β of the screening method is validated when Fm > T [15] and then it can be concluded that CC β is truly below the validation level. As recommended urine levels of various SARM compounds have not yet been established [17 , 18] , C val in the presented study was set as previously reported [5] at levels  Table 1 . -----±20% limit.

No
Analyte based on anabolic activity and comparable to that of other exogenous anabolic androgenic steroids and agents [17 , 19] . The developed assay enables detection of 14 SARM compounds (exception been LGD-2226 where T > Fm) in urine of all species with a false-negative rate ≤5% as stipulated in current EU legislation [14 , 15] . A sensitivity ≥95% at C val , expressed as percentage based on the ratio of samples detected as positive in true positive samples (i.e. following fortification) [20] , indicates that the number of false-negative samples is truly ≤5%. Adequate low detection limits, estimated at a signal-to-noise ratio ( S/N ) of at least three measured peak-to-peak, were accomplished for all SARMs of interest excluding BMS-564929 in equine urine (eLOD 1.5 ng mL −1 , Table 3 ). Absolute recoveries measured and recorded for all compounds within each analytical run aimed to verify assay performance during routine analysis (54-97%, Supplementary data - Table S1 and Fig. S2). As reported previously [5] , relative cut-off factor (RFm), expressed as percentage based on the ratio of the Fm and the mean response of fortified samples, was determined for each analyte ( Table 3 ), and during routine analysis should be applied to screen positive controls (QC samples).

Extension of validation to bovine, canine, human and rodent urine
The ruggedness study included animal species as a factor potentially impacting results, thus an extension of the initial validation in equine urine was performed with bovine, canine, human and rodent urine (by testing urine from different sources, n = 5 per species, both blank and fortified at C val as per equine urine), providing sensitivity as highlighted in Table 4 . Accordingly, the method is seen to be applicable to these additional species, with the same CC β values for all analytes as per equine urine. Furthermore, the ruggedness study, executed on a different day and by a different operator [15] , reported correct classification of all analysed urine, with 15 blank samples ( n = 5 per species) all "screen negative" and corresponding fortified (C val ) samples all "screen positive" (i.e. exceeding the cut-off factor).

Application to real samples
Bovine urine collected from a two months old steer calf orally administered 200 mg of ostarine (S-22) as described previously [6] were assayed employing the developed method. Three samples were tested blindly in triplicate and each was assigned correctly, with one sample screened negative (A -collected prior to SARM treatment), and the remaining two samples screened positive (collected B -2 h and C -3 days, respectively, post-ostarine (S-22) administration) - Fig. 3 and Supplementary data - Fig. S3. The current findings are in agreement with ostarine urinary concentration results (following UHPLC-MS/MS in-house validated analysis, CC α 0.25 μg L −1 ) reported by de Rijke et al. [6] .  [5] and total ostarine residues detected in tested bovine urine samples, whereas total form represents the sum of free ostarine and ostarine liberated within enzymatic hydrolysis step from respective conjugates. A deconjugation step included in the method led to significantly increased relative abundance of ostarine, namely 16.2-fold in sample B and 2.9-fold in sample C, respectively. Additionally, equine, bovine, canine and human urine samples ( n = 161) have been screened employing the developed assay with hydrolysis, with no tested samples reporting detectable levels of SARM compounds.

Concluding remarks
The current study describes the simultaneous monitoring of 14 SARMs in hydrolysed urine from equine, canine, human, bovine and rodent via an UHPLC-MS/MS-based semi-quantitative screening developed to incorporate an enzymatic hydrolysis step into a previously established protocol [5] . The method was validated in accordance with criteria stipulated in relevant legislation and demonstrates required sensitivity at ≥95% [14 , 15] with CC β values determined at 1 ng mL −1 , except for andarine (2 ng mL −1 ) and BMS-564929 (5 ng mL −1 ). The analysis of incurred samples highlighted the diagnostic capability of the presented method to detect a total form of emerging SARMs in urine matrix. This modified assay can serve as an effective approach to reveal illicit SARM use in urine from animal and human sport animals as well as food-based livestock.

Declaration of Competing Interest
There are no conflicts of interest to declare.