Introduction

Synthetic cannabinoids bind to CB1 and/or CB2 receptors, often with greater binding affinities than delta-9-tetrahydrocannabinol, and were originally synthesized for researching endocannabinoid pharmacology. Synthetic cannabinoids are sold over the Internet, are smoked or inhaled for their psychoactive effects, and are not detected via traditional cannabinoid testing. These designer drugs are abused to circumvent drug monitoring programs and also because young people like to experiment with new drugs. Synthetic cannabinoids are structurally diverse and are currently categorized into 12 classes based on chemical structure: adamantoylindole, benzoylindoles, cyclohexylphenols, classical dibenzopyrans, indazole based, naphthoylindoles and WIN 55,212-2, naphthoylpyrroles, naphthylmethylindoles, naphthylmethylindenes, phenylacetylindoles, quinolinyl ester (indole based), and tetramethylcyclopropyl ketone indoles [1, 2].

New synthetic cannabinoids are constantly emerging, and their widespread availability makes it difficult for regulatory agencies to stay abreast of this major public health issue. Synthetic cannabinoids emerged in Europe before being identified in the USA; European countries (including Austria, Germany, France, Luxembourg, Sweden, Estonia, Poland, Hungary, and the UK) began banning these compounds in 2009 [2]. Initial US scheduling efforts included five abused synthetic cannabinoids that were classified as Schedule I under a temporary ban by the US Department of Justice in March 2011 [3] and were more recently expanded under US law in July 2012 explicitly banning 15 synthetic cannabinoids under Schedule I of the Controlled Substances Act [4].

Constantly emerging synthetic cannabinoids also pose a challenge for laboratories performing drugs of abuse testing, since new drugs cannot be detected by existing, targeted analytical methods. Laboratories are consequently burdened by continuously developing and validating new methods in response to the next wave of synthetic drugs. Ideally, a nontargeted method is desirable, capable of detecting all possible drugs of abuse, whether known or unknown at the time of testing.

Immunoassays provide an inexpensive, sensitive and rapid screening alternative to hyphenated chromatographic techniques, offering high-throughput testing for targeted synthetic cannabinoids. Definitive mass spectrometry confirmation is required for presumptive positive screening results. Synthetic cannabinoid immunoassays effectively detected early generation synthetic cannabinoids [5–7]. However, antibody cross-reactivities to newly emerging synthetic cannabinoids are unknown and may limit detection of new compounds currently on the market. Several reports suggested liquid chromatography tandem mass spectrometric (LC-MS/MS), quadrupole-linear ion trap [8], or high-resolution quadrupole/time-of-flight MS (QTOF) [9–12] are more useful for screening than immunoassays. LC-MS/MS affords flexibility for including new compounds as reference standards or potentially as new compound structures and spectra become available, while immunoassays are limited by the time delay for developing antibodies for new analytes.

A major challenge for synthetic cannabinoid urine testing is that new analytes constantly emerge on the market with unknown metabolites. Synthetic cannabinoid metabolites predominate in urine with minimal or no unchanged parent compound present in urine such that identifying optimal metabolite targets is critical for identifying intake [13–16]. Safety concerns and lack of acute and chronic toxicity prevent timely human controlled drug administration studies, making identifying metabolite markers difficult. We have identified synthetic cannabinoid metabolites following in vitro incubation of synthetic cannabinoids with human hepatocytes followed by high-resolution mass spectrometry [17–22]. Others utilized human liver microsomes [23–26] or by analyzing urine from preclinical controlled administration studies [27, 16, 28] or specimens from individuals suspected of synthetic cannabinoid intake (overdose cases or when an individual is found in possession of synthetic cannabinoids) [27, 16, 14, 15, 13, 29, 30, 26]. Converging metabolic pathways result in common metabolites from multiple synthetic cannabinoids, further confounding synthetic cannabinoid urine testing [26]. Evaluation of certified metabolite reference standards is required for validating methodology before reporting results. Thus, an up-to-date urine method targeting as many synthetic cannabinoids for which reference standards are currently available is critical for effectively monitoring synthetic cannabinoid intake.

We recently published an LC-MS/MS qualitative urine screening method with an AB Sciex QTRAP instrument targeting 9 synthetic cannabinoids and 20 metabolites [8]. Nominal-mass multiple reaction monitoring (MRM) triggered product ion spectra detection was employed for targeted compounds in our quadrupole-linear ion trap (QTRAP) method [8]. However, high-resolution tandem MS provides accurate mass information on molecular and fragment ions, affording more selectivity and simplifying data interpretation compared to our previous unit resolution QTRAP method. Two LC-QTOF synthetic cannabinoid urine screening methods were reported employing nontargeted data acquisition and validation for 5 or 12 synthetic cannabinoid families [9, 10]. These nontargeted methods allow later data re-interrogation to search for unanticipated compounds. These acquisition methods do not require updating in order to address next-generation synthetic cannabinoids, since the methods are nontargeted in nature. However, some nontargeted approaches are not sufficiently selective for detecting low drug concentrations in biological matrices such as urine, primarily because the methods rely upon non-quadrupole filtered full-scan TOF MS, and not MS/MS, for detection. A truly nontargeted method providing selective MS/MS detection has proven elusive.

We employed a novel, nontargeted analytical technique with a LC-QTOF high-resolution tandem mass spectrometer, providing highly selective MS/MS analysis of all analytes. Our method employs SWATH™ acquisition (Sequential Windowed Acquisition of all Theoretical mass spectra), which sequentially acquires MS/MS of all precursor ions across a specified mass range, by parsing the mass range into small windows. In every scan cycle, the instrument rapidly and sequentially acquires MS/MS of all mass windows across the specified mass range. Since the approach is nontargeted, our method should not require future modifications for incorporating newly emerging compounds, and MS/MS spectra are acquired for unknown compounds that can be identified retrospectively via data re-interrogation. We specified a mass range of 228–408 Da for our SWATH MS/MS acquisition, since synthetic cannabinoid masses fall between 232 and 406 Da. Any emerging synthetic cannabinoid with a mass between 228 and 408 Da also will be detected via this method with a high degree of selectivity and sensitivity.

An up-to-date LC-QTOF synthetic cannabinoid urine screening method is urgently needed for detecting scheduled, currently abused, and newly emerging synthetic cannabinoid metabolites in urine. We present a validated qualitative LC-QTOF urine screening method targeting 47 synthetic cannabinoid metabolites from 21 synthetic cannabinoid families: 5-fluoro AB-PINACA, 5-fluoro-AKB48, 5-fluoro PB-22, AB-PINACA, ADB-PINACA, AKB48, AM2201, JWH-018, JWH-019, JWH-073, JWH-081, JWH-122, JWH-200, JWH-210, JWH-250, JWH-398, MAM2201, PB-22, RCS-4, UR-144, and XLR11.

Methods

Reagents and supplies

All standards and deuterated internal standards were purchased from Cayman Chemical (Ann Arbor, MI). Ammonium acetate, formic acid, acetonitrile, methanol, and ethyl acetate were obtained from Sigma-Aldrich (St. Louis, MO). Water was purified by an ELGA Purelab Ultra Analytic purifier (Siemens Water Technologies, Lowell, MA). All solvents were of high-performance liquid chromatography (HPLC) grade or better. Red Abalone beta-glucuronidase solution of 100,000 units/mL beta-glucuronidase and 8,000 units/mL sulfatase was diluted with distilled water to contain 15,625 units/mL beta-glucuronidase and 1,250 units/mL sulfatase activity for enzymatic hydrolysis (Kura Biotec, Puerto Varas, Chile); 1-mL Isolute SLE+ supported-liquid extraction cartridges were utilized for preparing the samples (Biotage, Inc., Charlotte, NC). A Cerex System 48 positive pressure manifold (SPEware Corp, Baldwin Park, CA) was employed for specimen extraction. Analytical chromatography was performed on an Ultra Biphenyl HPLC column (100× 2.1 mm; 3 μm particle size) combined with a 10 × 2.1 mm guard column of identical phase purchased from Restek (Bellefonte, PA).

Instrumentation

An AB Sciex TripleTOF® 5600+ high-resolution QTOF mass spectrometer with a DuoSpray ion source operated in positive electrospray (ESI+) mode (AB Sciex, Foster City, CA) was coupled with an LC-20ADxr HPLC system (Shimadzu Corp, Columbia, MD). Automatic mass calibration was performed every 2 h via infusion of calibration solution through the atmospheric pressure chemical ionization DuoSpray source probe. Data were acquired with AnalystTF version 1.5.1 and analyzed with MasterView 1.0 operated within PeakView 2.0 software (all software from AB Sciex). A MasterView extracted ion chromatogram (XIC) data processing worksheet was constructed for building XICs for molecular ions and the most intense MS/MS fragment for each target analyte (10-mDa-wide extraction widths for all ions). Four MS/MS fragments were extracted for PB-22 N-hydroxypentyl-3-carboxyindole and PB-22 N-pentanoic-3-carboxyindole (see Table 1 for a list of analytes, molecular ions, and fragments).

Table 1 List of synthetic cannabinoid metabolites, chemical formulas, masses of protonated molecular ion (M-H+), MS/MS fragments, and retention times (RT) for liquid chromatography-high-resolution tandem mass spectrometry screening method
Full size table

Calibrators and internal standards

Blank urine was evaluated to ensure the absence of detectable synthetic cannabinoids or metabolites prior to fortification with working stock solutions for preparing limits of detection (LOD) samples. Individual analyte primary stock solutions for 47 synthetic cannabinoid metabolites at 100 mg/L were prepared in methanol (see analyte list in Table 1). Preliminary experiments with decreasing analyte concentrations fortified into blank urine (0.1–20 μg/L) were conducted in triplicate to approximate the LOD prior to beginning validation; working solutions contained all analytes at equivalent concentrations for these preliminary LOD studies and were prepared from the 100-mg/L individual analyte stock solutions. Subsequently, 0.5, 1, 2, and 20× LOD working standard solutions employed for validation experiments were prepared by diluting 100 mg/L of individual analyte stock solutions. Adding a 25-μL 1× LOD working standard solution to 250-μL blank urine created LOD concentrations for each analyte as listed in Table 2.

Table 2 List of compounds evaluated as potential interferences at 250 μg/L fortified into urine samples containing target analytes at the limits of detection concentrations
Full size table

Deuterated internal standard stock solutions containing JWH-073-d7, JWH-073 5-hydroxyindole-d7, and JWH-073 N-butanoic acid-d5 were diluted in methanol, producing a mixed 500-μg/L internal standard solution; 25 μL 500 μg/L mixed internal standard solution was added to 250-μL blank urine LOD samples and authentic specimens, yielding 50 μg/L of internal standard concentrations.

All primary and working solutions were stored at −20 °C in amber glass vials.

Specimen preparation

For SLE preparation, 250 μL of blank urine fortified to contain 0.5, 1, or 2× LOD standards (or 25 μL methanol for authentic specimens) and internal standard was diluted with 0.6 mL 400 mM ammonium acetate buffer, pH 4.0, prior to addition of 40 μL glucuronidase solution (15,625 units glucuronidase activity/mL). Polypropylene microcentrifuge tubes were capped and incubated at 55 °C for 1 h. Samples were centrifuged at 15,000 g, 4 °C for 5 min after addition of 0.2 mL acetonitrile. Samples were transferred onto SLE columns and gently driven onto the column phase with fine pressure control by slowly increasing pressure up to 1 L/min (achieving 21 mL/min through each column). After equilibration at ambient pressure for 2 min, analytes were eluted with 6 mL ethyl acetate into 16 × 100 mm conical polypropylene tubes. Positive pressure was gradually applied up to 5 L/min (100 mL/min through each column) with fine pressure control until elution was complete. All sample extracts were completely dried at 45 °C under nitrogen in a Zymark TurboVap. Samples were reconstituted in 150 μL mobile phase A/B 80:20 (v/v), vortexed 15 s prior to centrifugation at 4 °C, 4,000g for 5 min, and transferred to autosampler vials containing 300-μL glass inserts.

LC-QTOF

Prepared samples (50 μL) were injected onto the LC-QTOF system, and chromatographic separation was performed on an Ultra Biphenyl column equipped with a guard column containing identical packing material. The column oven and autosampler were maintained at 30 and 4 °C, respectively. Gradient elution was performed with (A) 0.1 % formic acid in water and (B) 0.1 % formic acid in acetonitrile at a flow rate of 0.5 mL/min. Initial gradient conditions were 20 % B, held for 30 s; then increased to 95 % B over 10.5 min, held until 13.0 min; and returned to 20 % B at 13.1 min and held until 15.0 min. HPLC eluent was diverted to waste for the first 2.0 min and after 13.0 min of analysis.

Positive mode electrospray source settings were as follows: gas 1 = 40 psi, gas 2 = 75 psi, temperature = 750 °C, and 4,000 V ion spray voltage with 45 psi curtain gas. Intact protonated molecular ions (M-H+) were detected via TOF MS scan (80 psi declustering potential, 10 V collision energy, 100–1,000 Da TOF MS scan range, and 100 ms accumulation time). SWATH MS/MS experiments acquired spectra in high-sensitivity mode (80 psi declustering potential, 35 V collision energy with ±15 V collision energy spread, 30–1,000 Da TOF MS scan range, and 25 ms accumulation time). Thirty SWATH quadrupole isolation windows were 6 Da wide ranging from 228 to 408 Da. Total cycle time for the TOF MS and 30 SWATH MS/MS scans was 900 ms.

Hydrolysis optimization

Hydrolysis conditions were optimized with blank urine fortified to contain 4,000 μg/L JWH-018 N-5-hydroxypentyl-glucuronide, JWH-018 N-pentanoic-glucuronide, JWH-019 N-(6-hydroxyhexyl)-glucuronide, JWH-073 N-(4-hydroxypentyl)-glucuronide, and UR-144 N-(5-hydroxypentyl)-glucuronide (all of the currently commercially available synthetic glucuronide reference standards). Amount of enzyme, pH, temperature, and duration of incubation were evaluated for optimally hydrolyzing all of these glucuronides. Treatments were evaluated with one-way analysis of variance with Bonferroni post hoc tests (p < 0.05).

Method validation

Specificity, sensitivity, extraction efficiency, matrix effect, stability, and carryover were evaluated during method validation.

Specificity

At least five scans were acquired across each peak, reliably acquiring representative spectra for each analyte. Analyte identification criteria were the following: (1) mass error for the molecular ion <15 ppm, (2) retention time error <2 %, (3) isotopic pattern <80 % difference from theoretical, and (4) library fit >60 %. An overall score was computed based on all four analyte identification criteria with the following weighting: 25, 25, 10, and 40 % for items 1, 2, 3, and 4, respectively. All four criteria must be fulfilled and with visual verification of at least two common fragments in the sample and reference MS/MS spectra. Potential endogenous interferences were assessed by analyzing ten unique blank urine specimens. Potential false positives and false negatives were evaluated for 83 potential interferences from commonly used drugs by fortifying drugs at 250 μg/L into blank urine or LOD samples, respectively (see Table 2 for interferent list). No false-positive interference was noted if no analytes were identified in interference-fortified blank urine; no false-negative interference was noted if all analytes in the LOD sample were identified.

Sensitivity

LOD was evaluated over three runs with ten different urine sources and defined as the lowest concentration, fulfilling our identification criteria as detailed for specificity.

Extraction efficiency and matrix effect

Extraction efficiency and matrix effect were evaluated via three sets of samples as described by Matuszewski et al. [31] (n = 10 for each set). In the first set, urine samples were fortified with analytes and internal standards prior to SLE. In set 2, urine samples were fortified with analytes and internal standards after SLE, and the third set contained analytes and internal standards in mobile phase. Extraction efficiency, expressed as a percentage, was calculated by dividing analyte mean peak areas of set 1 by set 2. Absolute matrix effect was calculated by dividing the mean peak area of the analyte in set 2 by the mean analyte area in set 3, converting to a percentage and subtracting it from 100.

Analyte stability

Analyte stability also was evaluated with blank human urine fortified with analytes of interest at 1 and 20× LOD concentrations (n = 3). Analyte short-term temperature stability was evaluated for fortified human urine stored in the dark in polypropylene microcentrifuge tubes for 24 h at room temperature, 72 h at 4 °C, 72 h on the autosampler (4 °C), and after three freeze-thaw cycles at −20 °C. On the day of analysis, internal standard was added to each sample and analyzed as described along with freshly fortified 1× and 20× LOD samples (see Table 2 for a list of potential interferences). Stability was assessed by comparing peak areas to those of freshly fortified samples. Autosampler stability was assessed by comparing peak areas of samples injected immediately after extraction to those injected after 72 h storage at 4 °C on the autosampler. Peak area decreases of >20 % were considered unstable.

Carryover

Carryover was investigated in triplicate by injecting extracted blank urine samples containing internal standards immediately after samples containing target analytes at 400 μg/L. If blank urine specimens contained an analyte, ≥LOD carryover was demonstrated.

Authentic specimens

Anonymous, randomly collected authentic urine specimens provided from the Department of Defense urine drug testing program were analyzed to assess method utility.

Results

Hydrolysis optimization

Optimized hydrolysis conditions employed 625 U glucuronidase/250 μL urine specimen, with pH 4.0 ammonium acetate buffer and 1 h of incubation at 55 °C, achieving 98.5–98.8 % hydrolysis of 4,000 μg/L JWH-018 N-5-hydroxypentyl-glucuronide, JWH-018 N-pentanoic-glucuronide, JWH-019 N-(6-hydroxyhexyl)-glucuronide, JWH-073 N-(4-hydroxypentyl)-glucuronide, and UR-144 N-(5-hydroxypentyl)-glucuronide (Fig. 1). More enzyme, longer incubation, and higher temperatures did not improve hydrolysis efficiencies. Reducing glucuronidase to 312.5 U/250 μL urine specimen was not significantly different from 625 U/250 μL urine specimen; however, variability was lower with 625 U glucuronidase. A pH 5 ammonium acetate buffer did not improve hydrolysis efficiencies; pH 3 was less effective than pH 4 ammonium acetate buffer for JWH-018 N-5-hydroxypentyl-glucuronide, JWH-019 N-(6-hydroxyhexyl)-glucuronide, JWH-073 N-(4-hydroxypentyl)-glucuronide, and UR-144 N-(5-hydroxypentyl)-glucuronide hydrolysis.

Fig. 1
figure 1

Hydrolysis optimization varying glucuronidase amount, pH, temperature, and incubation duration with JWH-018 N-5-hydroxypentyl-glucuronide, JWH-018 N-pentanoic-glucuronide, JWH-019 N-(6-hydroxyhexyl)-glucuronide, JWH-073 N-(4-hydroxypentyl)-glucuronide, and UR-144 N-(5-hydroxypentyl)-glucuronide standards fortified into blank urine at 4,000 μg/L. *p < 0.05 (significantly different from other hydrolysis conditions)

Full size image

Chromatography

Baseline chromatographic resolution for all analytes was not possible; all isobaric compounds were baseline resolved except for isomeric alkyl hydroxy compounds and JWH-019 hydroxyhexyl and JWH-122 hydroxypentyl. However, JWH-019 hydroxyhexyl and JWH-122 hydroxypentyl could be differentiated because they had unique MS/MS fragment ions. All analytes within SWATH windows were different by >1 Da, eliminating false-positive results.

We added 5-fluoro AB-PINACA N-4-hydroxypentyl, 5-fluoro-AKB48 N-4-hydroxypentyl, AB-PINACA N-5-hydroxypentyl, ADB-PINACA N-5-hydroxypentyl, AKB-48 N-5-hydroxypentyl, AM2201 N-4-hydroxypentyl, JWH-018 N-5-hydroxypentyl, JWH-019 N-6-hydroxyhexyl, JWH-073 N-4-hydroxybutyl, JWH-081 N-5-hydroxypentyl, JWH-122 N-5-hydroxypentyl, JWH-210 N-5-hydroxypentyl, JWH-250 N-5-hydroxypentyl, JWH-398 N-5-hydroxypentyl, MAM2201 N-4-hydroxypentyl, PB-22 N-(5-hydroxypentyl)-3-carboxyindole, PB-22 N-5-hydroxypentyl, RCS-4 N-5-hydroxypentyl, UR-144 N-5-hydroxypentyl, and XLR11 N-4-hydroxypentyl alkyl hydroxy metabolite standards to our working standard solutions, but since isomeric baseline separation was not possible, we can only identify these peaks as alkyl hydroxy metabolites without assigning hydroxy position on the alkyl chain.

Specificity and potential interferences

No interfering peaks in blank urine from ten individuals produced false-positive identification of any analyte (see Fig. 2a for blank urine XIC). None of 83 potential interfering compounds possibly encountered in clinical or forensic specimens fortified at 250 μg/L into blank urine or LOD samples produced false-positive or false-negative results for any analyte.

Fig. 2
figure 2

Extracted ion chromatograms with all analyte MS/MS fragments overlaid for a a blank urine extract and b a urine extract fortified to contain all analytes at their limit of detection (LOD). See Table 1 for an analyte list, MS/MS fragments, and retention times

Full size image

Sensitivity

We evaluated two XICs for each analyte including the molecular ion and the most intense MS/MS fragment. The MS/MS XIC results generally met identification criteria at lower concentrations enabling lower LODs, but sometimes, both molecular ion and fragment XICs achieved similar LODs. PB-22 N-hydroxypentyl-3-carboxyindole and PB-22 N-pentanoic-3-carboxyindole did not ionize efficiently in positive mode yielding poor signal intensities and irreproducible spectra that erratically met our defined >60 % library fit criteria. We employed alternative MS/MS spectra criteria, monitoring four MS/MS fragments specific for these analytes, and required three fragments to be present in the XICs for positive identification (see Table 1). Molecular ion mass error, retention time, and isotopic ratio tolerances were required to meet our standard identification criteria in addition to observing three analyte-specific MS/MS fragments. PB-22 N-hydroxypentyl-3-carboxyindole and PB-22 N-pentanoic-3-carboxyindole had higher LODs (10 and 20 μg/L) than any other analyte (0.25–5 μg/L) (Table 3 and Fig. 2b for XIC of extracted urine fortified to contain each analytes’ LOD concentration). Five to six scans/peak were acquired, yielding deuterated metabolite internal standard peak area coefficients of variation of 4.0–5.4 % for 25 extracted sample injections.

Table 3 Limits of detection, mean extraction efficiencies, and matrix effects for synthetic cannabinoid metabolites extracted from urine by supported-liquid extraction
Full size table

Extraction efficiencies and matrix effects

Extraction efficiencies and matrix effects are presented in Table 3. Mean extraction efficiencies were 55.2–104.4 % (n = 10). Mean matrix effects (% suppressed signal) were −64.8 to 107.0 % (n = 10).

Analyte stability and carryover

All analytes were stable at 1 and 20× LOD concentrations after 24 h at room temperature, 72 h at 4 °C, and after three freeze/thaw cycles (Table 4). All analytes were stable after 72 h at 4 °C on the autosampler (data not shown).

Table 4 Synthetic cannabinoid metabolite stabilities in human urine (% change)
Full size table

No analyte met identification criteria in blank samples injected after samples containing all analytes at 400 μg/L.

Authentic specimen

Figure 3a shows an XIC from an authentic specimen containing JWH-122 N-hydroxypentyl; Fig. 3b shows an XIC for JWH-019 N-hydroxyhexyl from the same specimen. There was no peak for the JWH-019 N-hydroxyhexyl fragment, while there was a peak in the JWH-122 N-hydroxypentyl XIC.

Fig. 3
figure 3

Extracted ion chromatograms (XIC) from an authentic urine specimen containing JWH-122 N-hydroxypentyl metabolite but negative for JWH-019 N-hydroxyhexyl and the ability to distinguish these two co-eluting isobaric analytes via SWATH. a XIC for JWH-122 N-hydroxypentyl (MS/MS fragment, 169.0671 Da) and b XIC for JWH-019 N-hydroxypentyl (MS/MS fragment, 155.0508 Da)

Full size image

Discussion

Despite illicit drug scheduling efforts, the “Monitoring the Future” study reported that 7.9 % of 12th graders smoked synthetic cannabinoids at least once during 2013 [32]. Synthetic cannabinoids effects remain incompletely characterized with no controlled administration clinical studies. Case reports suggest that synthetic cannabinoids produce not only behavioral [33] but also renal [34–36] and cardiovascular toxicities in humans [37–39], but the mechanisms of action are unknown. New compounds continually emerge in response to scheduling efforts presenting significant analytical challenges for clinical and forensic laboratories monitoring synthetic cannabinoid exposure. Testing is complicated since metabolites predominate in urine, and human-controlled administration pharmacokinetic studies identifying optimal metabolite markers are not possible due to required preclinical safety studies prior to obtaining Food and Drug Administration approval or from other governing bodies around the world (Table 5).

Table 5 Synthetic cannabinoids identified in eight anonymous authentic urine specimens
Full size table

Immunoassay screening is typically employed for urine drug testing to identify presumptive positive samples, but confirmatory testing is required. Also, synthetic cannabinoid immunoassay screening may not be effective because existing antibodies frequently lack cross-reactivity and are unable to detect newly marketed compounds [9–12, 40]. LC-MS/MS screening offers an alternative emerging drug screening approach, since new analytes that can be targeted as reference standards become available or possibly as soon as new compound structures and spectra have been identified via human liver microsomes or human hepatocyte incubation and high-resolution LC-QTOF analysis [19, 21, 22, 18, 17, 20]. High-resolution LC-QTOF can identify new compounds based upon their molecular formula and accurate mass. Although less expensive quadrupole-linear ion trap MS can be employed for routine screening [8]; LC-QTOF is also effective with accurate mass assignment of molecular ions and fragment ions providing additional information, simplifying data interpretation [9–12].

Moller et al. were the first to report employing high-resolution LC-MS for detecting JWH-018 and its metabolites in urine in 2011 [12]. There are three reports detailing LC-TOF or LC-QTOF screening panels for multiple synthetic cannabinoids in urine [9–11]. Guale et al. detailed an LC-TOF screening method for 22 synthetic cannabinoid families, but the method targets parent compounds and is most appropriate for blood and oral fluid [11]. Sundstrom et al. presented an ultra-high-performance liquid chromatography (UHPLC)-QTOF method employing broadband collision-induced dissociation, validated for 75 drugs of abuse including 18 synthetic cannabinoid metabolites from five first-generation synthetic cannabinoid families (JWH-018, JWH-073, JWH-122, JWH-200, and RCS-4) in 1 mL of urine, achieving LODs of 0.5–5.5 μg/L [9]. More recently, Kronstrand et al. reported a qualitative UHPLC-QTOF method validated for 38 synthetic cannabinoid metabolites from 12 synthetic cannabinoid families (AM2201, JWH-018, JWH-073, JWH-081, JWH-122, JWH-210, JWH-250, JWH-398, MAM2201, RCS-4, and UR-144) in 0.6 mL of urine, achieving cutoffs of 2 μg/L [10]. Both methods employed nontargeted data acquisition, affording simplified validation as new compounds emerge and retrospective data interrogation for currently unknown compounds [9, 10]. Our presented method employs nontargeted SWATH QTOF MS data acquisition and is validated for 47 metabolites from the most problematic current 21 synthetic cannabinoid families (5-fluoro AB-PINACA, 5-fluoro-AKB48, 5-fluoro PB-22, AB-PINACA, ADB-PINACA, AKB48, AM2201, JWH-018, JWH-019, JWH-073, JWH-081, JWH-122, JWH-200, JWH-210, JWH-250, JWH-398, MAM2201, PB-22, RCS-4, UR-144, and XLR11). This method provides the most comprehensive assay to date with similar or better sensitivity than other published methods for most analytes [9, 10].

SWATH nontargeted acquisition affords some advantages over nontargeted approaches employed previously [9, 10]. SWATH more selectively acquires MS/MS spectra from 6-Da-wide quadrupole filtered mass ranges; previous LC-QTOF methods employed “broadband collision-induced dissociation” that does not incorporate any quadrupole selectivity [9, 10]. Increased SWATH selectivity provides cleaner MS/MS spectra, enabling lower LODs despite co-eluting matrix components. We employed a SWATH scan range of 228–408 Da covering the molecular weights of all compounds included in our validation to maximize scans/peak; newly emerging compounds with masses outside this range would be missed and require updating the SWATH method settings. SWATH acquisition produces fewer scans/peak than previous methods; however, we were able to acquire five to six scans/peak, yielding deuterated metabolite internal standard peak area coefficients of variation of 4.0–5.4 % for 25 extracted sample injections. This suggests that we are reliably acquiring MS/MS data across analyte peaks enabling reproducible analyte detection.

We previously reported a validated LC-MS/MS method quantifying 20 synthetic cannabinoids and 21 metabolites, and semi-quantifying an additional 12 alkyl hydroxy metabolites in human urine via unit resolution QTRAP instrumentation [41]. The current SWATH high-resolution LC-MS/MS qualitative method is validated for 47 synthetic cannabinoid metabolites including recently emerging synthetic cannabinoid metabolites for which standards were unavailable during QTRAP quantitative method development. These new analytes include 5-fluoro AB-PINACA N-hydroxypentyl, 5-fluoro AKB48 N-hydroxypentyl, 5-fluoro PB-22 3-carboxyindole, AB-PINACA N-hydroxypentyl, AB-PINACA pentanoic acid, ADB-PINACA N-hydroxypentyl, AKB48 N-hydroxypentyl, AKB48 pentanoic acid, PB-22 3-carboxyindole, PB-22 N-hydroxypentyl-3-carboxyindole, PB-22 N-pentanoic-3-carboxyindole, PB-22 N-hydroxypentyl, PB-22 N-pentanoic acid, XLR11 6-hydroxyindole, and XLR11 N-hydroxypentyl. The current qualitative SWATH high-resolution LC-MS/MS method achieves similar sensitivity as the previous QTRAP quantitative method that had 0.1–1.0 μg/L lowest limits of linearity [41]. The qualitative SWATH high-resolution LC-MS/MS method affords more flexibility for incorporating new analytes as compounds emerge, since infusion optimization, adding new analyte transitions to the acquisition method and validation of quality control accuracy/imprecision, is not required.

We previously optimized hydrolysis for JWH-018, JWH-073, JWH-122, JWH-210, JWH-250, AM2201, and RCS-4 metabolites with 2,000 U Red abalone β-glucuronidase/100 μL urine achieving >95 % hydrolysis of JWH-018 N-hydroxypentyl-glucuronide [41]. Our goal during method development was to evaluate whether less Red abalone glucuronidase enzyme (from an alternate vendor) could reduce costs and potentially minimize instrument downtime for maintenance. We found that 625 U Red abalone β-glucuronidase/250 μL urine achieved >98 % hydrolysis of JWH-018 N-5-hydroxypentyl-glucuronide, JWH-019 N-(6-hydroxyhexyl)-glucuronide, JWH-073 N-(4-hydroxypentyl)-glucuronide, and UR-144 N-(5-hydroxypentyl)-glucuronide. Our previous method incubated at 55 °C for 2 h; our new method employs a shorter, 1-h incubation duration, similar to a recent report for synthetic cannabinoids [42]. We evaluated all five currently available commercial synthetic cannabinoid glucuronide standards.

Our method was unable to achieve baseline chromatographic resolution for alkyl hydroxy metabolite isomers (i.e., JWH-018 N-2, N-3, N-4, and N-5-hydroxypentyl compounds have similar retention times), simply identifying alkyl hydroxyl metabolites in a specimen. This is a potential limitation for distinguishing the synthetic cannabinoid ingested, as AM2201 metabolically forms JWH-018 N-5-hydroxypentyl with little JWH-018 N-4-hydroxypentyl, and JWH-018 forms less JWH-018 N-5-hydroxypentyl than JWH-018 N-4-hydroxypentyl [26]. Our method includes AM2201 6-hydroxyindole and AM2201 N-hydroxypentyl that will definitively identify AM2201 ingestion, although these analytes are less abundant than JWH-018 N-5-hydroxypentyl. Figure 3 shows the ability of SWATH for distinguishing two co-eluting structural isomers via monitoring specific MS/MS fragments despite equivalent molecular masses.

PB-22 carboxyindole metabolites, PB-22 N-hydroxypentyl-3-carboxyindole and PB-22 N-pentanoic-3-carboxyindole, had the highest ion suppression (−59 and −65 %, respectively) and, hence, the highest LODs of our analytes (10 and 20 μg/L, all other LODs were <5 μg/L). Thus, the method’s ability to identify PB-22 intake may be limited. Typical PB-22 metabolite concentrations are unknown for PB-22 intake because controlled drug administration studies cannot yet be performed due to the lack of preclinical toxicity data.

We demonstrated the utility of this analytical method by detecting synthetic cannabinoids in eight anonymous DOD urine specimens collected from November 2011 to June 2012; however, screening methods must be able to detect the latest novel psychoactive compounds (NPS). Twenty-nine of 81 NPS reported by the European Monitoring Centre for Drugs and Drug Addiction during 2013 were synthetic cannabinoids [43]. Effectively monitoring synthetic cannabinoid abuse for constantly emerging compounds is a major challenge. Our method is validated for detecting 47 synthetic cannabinoid metabolites, the most extensive panel published to date including markers for many of the newest NPS, and the assay is designed to be easily updated for the latest emerging NPS.

We present a nontargeted SWATH LC-QTOF method validated for 47 metabolites from 21 synthetic cannabinoid families achieving LODs of 0.25–20 μg/L with a 0.25-mL urine specimen volume and a 15-min runtime; 250 μL of urine was hydrolyzed with β-glucuronidase before SLE extraction. SLE recoveries were >76.2 % for all metabolites and should accommodate new synthetic cannabinoids as new reference standards become available. Inclusion of new analytes requires less extensive re-validation than for a quantitative method but would include specificity, sensitivity, extraction efficiency, matrix effect, stability, and carryover evaluations for new analytes. SWATH acquisition allows for re-evaluation for newly emerging compounds once their structures and molecular formulas are known; however, reporting results without reference standards should be carefully considered.